Add files using upload-large-folder tool
Browse filesThis view is limited to 50 files because it contains too many changes. See raw diff
- out/pretrain/ppl_metrics.jsonl +20 -0
- out/pretrain/qwen2_7b_question_focus_lr_plus/teelog.txt +229 -0
- out/pretrain/tinyllama/teelogs/2407.txt +321 -0
- out/pretrain/tinyllama/teelogs/2407_lr4e-5.txt +334 -0
- out/pretrain/tinyllama/teelogs/2408.txt +288 -0
- out/pretrain/tinyllama/teelogs/2408_full.txt +291 -0
- out/pretrain/tinyllama/teelogs/2408_lr4e-5.txt +298 -0
- out/pretrain/tinyllama/teelogs/2409.txt +0 -0
- out/pretrain/tinyllama/teelogs/2409_full.txt +354 -0
- out/pretrain/tinyllama/teelogs/2409_lr4e-5.txt +361 -0
- out/pretrain/tinyllama/teelogs/2410.txt +368 -0
- out/pretrain/tinyllama/teelogs/2410_full.txt +380 -0
- out/pretrain/tinyllama/teelogs/2410_lr4e-5.txt +379 -0
- out/pretrain/tinyllama/teelogs/2411.txt +360 -0
- out/pretrain/tinyllama/teelogs/2411_full.txt +372 -0
- out/pretrain/tinyllama/teelogs/2411_lr4e-5.txt +371 -0
- out/pretrain/tinyllama/teelogs/2412.txt +414 -0
- out/pretrain/tinyllama/teelogs/2412_full.txt +425 -0
- out/pretrain/tinyllama/teelogs/2412_lr4e-5.txt +424 -0
- out/pretrain/tinyllama/teelogs/2501.txt +378 -0
- out/pretrain/tinyllama/teelogs/2501_full.txt +389 -0
- out/pretrain/tinyllama_3_epoch/2407/final/config.json +24 -0
- out/pretrain/tinyllama_3_epoch/2407/final/generation_config.json +7 -0
- out/pretrain/tinyllama_3_epoch/2407/final/hyperparameters.yaml +44 -0
- out/pretrain/tinyllama_3_epoch/2407/final/model_config.yaml +44 -0
- out/pretrain/tinyllama_3_epoch/2407/final/tokenizer.json +0 -0
- out/pretrain/tinyllama_3_epoch/2407/final/tokenizer_config.json +35 -0
- out/pretrain/tinyllama_3_epoch/2408/final/config.json +24 -0
- out/pretrain/tinyllama_3_epoch/2408/final/generation_config.json +7 -0
- out/pretrain/tinyllama_3_epoch/2408/final/hyperparameters.yaml +44 -0
- out/pretrain/tinyllama_3_epoch/2408/final/model_config.yaml +44 -0
- out/pretrain/tinyllama_3_epoch/2408/final/tokenizer.json +0 -0
- out/pretrain/tinyllama_3_epoch/2408/final/tokenizer_config.json +35 -0
- out/pretrain/tinyllama_3_epoch/2409/final/tokenizer.json +0 -0
- out/pretrain/tinyllama_3_epoch/2409/final/tokenizer_config.json +35 -0
- out/pretrain/tinyllama_lr_plus/2501/final/config.json +24 -0
- out/pretrain/tinyllama_lr_plus/2502/final/config.json +24 -0
- out/pretrain/tinyllama_lr_plus/2502/final/generation_config.json +7 -0
- out/pretrain/tinyllama_lr_plus/2502/final/hyperparameters.yaml +44 -0
- out/pretrain/tinyllama_lr_plus/2502/final/model_config.yaml +44 -0
- out/pretrain/tinyllama_lr_plus/2502/final/tokenizer.json +0 -0
- out/pretrain/tinyllama_lr_plus/2502/final/tokenizer_config.json +35 -0
- out/pretrain/tinyllama_lr_plus/2503/final/config.json +24 -0
- out/pretrain/tinyllama_lr_plus/2503/final/generation_config.json +7 -0
- out/pretrain/tinyllama_lr_plus/2503/final/hyperparameters.yaml +44 -0
- out/pretrain/tinyllama_lr_plus/2503/final/model_config.yaml +44 -0
- out/pretrain/tinyllama_lr_plus/2503/final/tokenizer.json +0 -0
- out/pretrain/tinyllama_lr_plus/2503/final/tokenizer_config.json +35 -0
- out/pretrain/tinyllama_lr_plus/2504/final/config.json +24 -0
- out/pretrain/tinyllama_lr_plus/2504/final/generation_config.json +7 -0
out/pretrain/ppl_metrics.jsonl
ADDED
|
@@ -0,0 +1,20 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{"val_loss": 1.2241098880767822, "val_ppl": 3.4011373421309727}
|
| 2 |
+
{"val_loss": 1.4328051805496216, "val_ppl": 4.190437654174611}
|
| 3 |
+
{"val_loss": 1.2608414888381958, "val_ppl": 3.528389338728705}
|
| 4 |
+
{"val_loss": 1.3117088079452515, "val_ppl": 3.7125122654387708}
|
| 5 |
+
{"val_loss": 1.3475964069366455, "val_ppl": 3.848164983201228}
|
| 6 |
+
{"val_loss": 1.3134140968322754, "val_ppl": 3.718848572429408}
|
| 7 |
+
{"val_loss": 1.249189853668213, "val_ppl": 3.4875164140296686}
|
| 8 |
+
{"val_loss": 1.334436058998108, "val_ppl": 3.797853577039805}
|
| 9 |
+
{"val_loss": 1.3168798685073853, "val_ppl": 3.731759612911466}
|
| 10 |
+
{"val_loss": 1.3145214319229126, "val_ppl": 3.722968864801566}
|
| 11 |
+
{"val_loss": 1.2239998579025269, "val_ppl": 3.400763134983968}
|
| 12 |
+
{"val_loss": 1.334436058998108, "val_ppl": 3.797853577039805}
|
| 13 |
+
{"val_loss": 1.334533929824829, "val_ppl": 3.798225294298997}
|
| 14 |
+
{"val_loss": 1.4382433891296387, "val_ppl": 4.213288204894694}
|
| 15 |
+
{"val_loss": 1.3327378034591675, "val_ppl": 3.7914093247089276}
|
| 16 |
+
{"val_loss": 0.47974008321762085, "val_ppl": 1.615654411917815}
|
| 17 |
+
{"val_loss": 1.4382433891296387, "val_ppl": 4.213288204894694}
|
| 18 |
+
{"val_loss": 1.3327378034591675, "val_ppl": 3.7914093247089276}
|
| 19 |
+
{"val_loss": 1.334533929824829, "val_ppl": 3.798225294298997}
|
| 20 |
+
{"val_loss": 0.47974008321762085, "val_ppl": 1.615654411917815}
|
out/pretrain/qwen2_7b_question_focus_lr_plus/teelog.txt
ADDED
|
@@ -0,0 +1,229 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
Initializing distributed: GLOBAL_RANK: 0, MEMBER: 1/4
|
| 2 |
+
Initializing distributed: GLOBAL_RANK: 1, MEMBER: 2/4
|
| 3 |
+
[rank: 1] Seed set to 42
|
| 4 |
+
Initializing distributed: GLOBAL_RANK: 3, MEMBER: 4/4
|
| 5 |
+
[rank: 3] Seed set to 42
|
| 6 |
+
Initializing distributed: GLOBAL_RANK: 2, MEMBER: 3/4
|
| 7 |
+
[rank: 2] Seed set to 42
|
| 8 |
+
----------------------------------------------------------------------------------------------------
|
| 9 |
+
distributed_backend=nccl
|
| 10 |
+
All distributed processes registered. Starting with 4 processes
|
| 11 |
+
----------------------------------------------------------------------------------------------------
|
| 12 |
+
|
| 13 |
+
[rank: 0] Seed set to 42
|
| 14 |
+
/mnt/data/litgpt/litgpt/pretrain.py:495: UserWarning: A newer version of litdata is available (0.2.58). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 15 |
+
train_dataloader = data.train_dataloader()
|
| 16 |
+
/mnt/data/litgpt/litgpt/pretrain.py:495: UserWarning: A newer version of litdata is available (0.2.58). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 17 |
+
train_dataloader = data.train_dataloader()
|
| 18 |
+
/mnt/data/litgpt/litgpt/pretrain.py:495: UserWarning: A newer version of litdata is available (0.2.58). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 19 |
+
train_dataloader = data.train_dataloader()
|
| 20 |
+
All GPUs are fully connected via NVLink.
|
| 21 |
+
{'data': {'batch_size': 1,
|
| 22 |
+
'data_path': PosixPath('litgpt/data/arxiv'),
|
| 23 |
+
'num_workers': 0,
|
| 24 |
+
'ppl': False,
|
| 25 |
+
'seed': 42,
|
| 26 |
+
'seq_length': 1024,
|
| 27 |
+
'use_starcoder': True},
|
| 28 |
+
'data_dir': PosixPath('litgpt/data/arxiv_test_qwen2_tokenized'),
|
| 29 |
+
'devices': 'auto',
|
| 30 |
+
'eval': {'evaluate_example': 'first',
|
| 31 |
+
'final_validation': False,
|
| 32 |
+
'initial_validation': False,
|
| 33 |
+
'interval': 9999,
|
| 34 |
+
'max_iters': 100,
|
| 35 |
+
'max_new_tokens': None},
|
| 36 |
+
'initial_checkpoint_dir': PosixPath('checkpoints/Qwen/Qwen2-7B'),
|
| 37 |
+
'log': {'group': None, 'project': None, 'run': None},
|
| 38 |
+
'logger_name': 'tensorboard',
|
| 39 |
+
'model_config': {'attention_logit_softcapping': None,
|
| 40 |
+
'attention_scores_scalar': None,
|
| 41 |
+
'attn_bias': True,
|
| 42 |
+
'bias': False,
|
| 43 |
+
'block_size': 131072,
|
| 44 |
+
'final_logit_softcapping': None,
|
| 45 |
+
'gelu_approximate': 'none',
|
| 46 |
+
'head_size': 128,
|
| 47 |
+
'hf_config': {'name': 'Qwen2-7B', 'org': 'Qwen'},
|
| 48 |
+
'intermediate_size': 18944,
|
| 49 |
+
'lm_head_bias': False,
|
| 50 |
+
'mlp_class_name': 'LLaMAMLP',
|
| 51 |
+
'moe_intermediate_size': None,
|
| 52 |
+
'n_embd': 3584,
|
| 53 |
+
'n_expert': 0,
|
| 54 |
+
'n_expert_per_token': 0,
|
| 55 |
+
'n_head': 28,
|
| 56 |
+
'n_layer': 28,
|
| 57 |
+
'n_query_groups': 4,
|
| 58 |
+
'name': 'Qwen2-7B',
|
| 59 |
+
'norm_1': True,
|
| 60 |
+
'norm_2': True,
|
| 61 |
+
'norm_class_name': 'RMSNorm',
|
| 62 |
+
'norm_eps': 1e-06,
|
| 63 |
+
'norm_qk': False,
|
| 64 |
+
'norm_qk_type': 'default',
|
| 65 |
+
'padded_vocab_size': 152064,
|
| 66 |
+
'padding_multiple': 512,
|
| 67 |
+
'parallel_residual': False,
|
| 68 |
+
'post_attention_norm': False,
|
| 69 |
+
'post_mlp_norm': False,
|
| 70 |
+
'rope_adjustments': None,
|
| 71 |
+
'rope_base': 1000000,
|
| 72 |
+
'rope_condense_ratio': 1,
|
| 73 |
+
'rope_indices': None,
|
| 74 |
+
'rope_local_base_freq': None,
|
| 75 |
+
'rotary_percentage': 1.0,
|
| 76 |
+
'scale_embeddings': False,
|
| 77 |
+
'shared_attention_norm': False,
|
| 78 |
+
'sliding_window_indices': None,
|
| 79 |
+
'sliding_window_size': None,
|
| 80 |
+
'vocab_size': 151643},
|
| 81 |
+
'model_name': 'Qwen2-7B',
|
| 82 |
+
'num_nodes': 1,
|
| 83 |
+
'optimizer': "{'class_path': 'torch.optim.AdamW', 'init_args': {'lr': 8e-05, "
|
| 84 |
+
"'weight_decay': 0.1, 'betas': [0.9, 0.95]}}",
|
| 85 |
+
'out_dir': PosixPath('out/pretrain/qwen2_7b_question_focus_lr_plus'),
|
| 86 |
+
'precision': 'bf16-true',
|
| 87 |
+
'resume': False,
|
| 88 |
+
'seed': 42,
|
| 89 |
+
'tokenizer_dir': PosixPath('checkpoints/Qwen/Qwen2-7B'),
|
| 90 |
+
'train': {'epochs': None,
|
| 91 |
+
'global_batch_size': 512,
|
| 92 |
+
'log_interval': 1,
|
| 93 |
+
'lr_warmup_fraction': None,
|
| 94 |
+
'lr_warmup_steps': 0,
|
| 95 |
+
'max_norm': 1.0,
|
| 96 |
+
'max_seq_length': 1024,
|
| 97 |
+
'max_steps': None,
|
| 98 |
+
'max_tokens': 52428800,
|
| 99 |
+
'micro_batch_size': 4,
|
| 100 |
+
'min_lr': 8e-05,
|
| 101 |
+
'save_interval': 10,
|
| 102 |
+
'tie_embeddings': None}}
|
| 103 |
+
Time to instantiate model: 0.06 seconds.
|
| 104 |
+
Total parameters: 7,615,616,512
|
| 105 |
+
[ok] checkpoints/Qwen/Qwen2-7B/lit_model.pth 已是纯权重
|
| 106 |
+
/mnt/data/litgpt/litgpt/pretrain.py:495: UserWarning: A newer version of litdata is available (0.2.58). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 107 |
+
train_dataloader = data.train_dataloader()
|
| 108 |
+
Verifying settings ...
|
| 109 |
+
Measured TFLOPs: 419979.36
|
| 110 |
+
Epoch 2 | iter 32 step 1 | loss train: 1.784, val: n/a | iter time: 1130.43 ms (step) remaining time: 0:49:35
|
| 111 |
+
Epoch 4 | iter 64 step 2 | loss train: 3.834, val: n/a | iter time: 990.00 ms (step) remaining time: 0:48:32
|
| 112 |
+
Epoch 6 | iter 96 step 3 | loss train: 2.444, val: n/a | iter time: 987.93 ms (step) remaining time: 0:47:53
|
| 113 |
+
Epoch 7 | iter 128 step 4 | loss train: 1.904, val: n/a | iter time: 992.12 ms (step) remaining time: 0:47:20
|
| 114 |
+
Epoch 9 | iter 160 step 5 | loss train: 1.595, val: n/a | iter time: 991.60 ms (step) remaining time: 0:46:49
|
| 115 |
+
Epoch 11 | iter 192 step 6 | loss train: 1.370, val: n/a | iter time: 991.67 ms (step) remaining time: 0:46:19
|
| 116 |
+
Epoch 12 | iter 224 step 7 | loss train: 1.167, val: n/a | iter time: 989.51 ms (step) remaining time: 0:45:48
|
| 117 |
+
Epoch 14 | iter 256 step 8 | loss train: 0.951, val: n/a | iter time: 992.01 ms (step) remaining time: 0:45:18
|
| 118 |
+
Epoch 16 | iter 288 step 9 | loss train: 0.720, val: n/a | iter time: 993.83 ms (step) remaining time: 0:44:48
|
| 119 |
+
Epoch 17 | iter 320 step 10 | loss train: 0.543, val: n/a | iter time: 992.87 ms (step) remaining time: 0:44:19
|
| 120 |
+
Saving checkpoint to 'out/pretrain/qwen2_7b_question_focus_lr_plus/step-00000010/lit_model.pth'
|
| 121 |
+
Epoch 19 | iter 352 step 11 | loss train: 0.338, val: n/a | iter time: 986.02 ms (step) remaining time: 1:20:01
|
| 122 |
+
Epoch 21 | iter 384 step 12 | loss train: 0.221, val: n/a | iter time: 988.10 ms (step) remaining time: 1:16:07
|
| 123 |
+
Epoch 22 | iter 416 step 13 | loss train: 0.146, val: n/a | iter time: 991.41 ms (step) remaining time: 1:12:45
|
| 124 |
+
Epoch 24 | iter 448 step 14 | loss train: 0.107, val: n/a | iter time: 992.48 ms (step) remaining time: 1:09:48
|
| 125 |
+
Epoch 26 | iter 480 step 15 | loss train: 0.100, val: n/a | iter time: 991.17 ms (step) remaining time: 1:07:10
|
| 126 |
+
Epoch 27 | iter 512 step 16 | loss train: 0.068, val: n/a | iter time: 992.15 ms (step) remaining time: 1:04:49
|
| 127 |
+
Epoch 29 | iter 544 step 17 | loss train: 0.048, val: n/a | iter time: 991.18 ms (step) remaining time: 1:02:40
|
| 128 |
+
Epoch 31 | iter 576 step 18 | loss train: 0.036, val: n/a | iter time: 988.18 ms (step) remaining time: 1:00:46
|
| 129 |
+
Epoch 32 | iter 608 step 19 | loss train: 0.027, val: n/a | iter time: 989.82 ms (step) remaining time: 0:58:58
|
| 130 |
+
Epoch 34 | iter 640 step 20 | loss train: 0.020, val: n/a | iter time: 989.76 ms (step) remaining time: 0:57:18
|
| 131 |
+
Saving checkpoint to 'out/pretrain/qwen2_7b_question_focus_lr_plus/step-00000020/lit_model.pth'
|
| 132 |
+
Epoch 36 | iter 672 step 21 | loss train: 0.013, val: n/a | iter time: 988.12 ms (step) remaining time: 0:59:14
|
| 133 |
+
Epoch 38 | iter 704 step 22 | loss train: 0.008, val: n/a | iter time: 988.99 ms (step) remaining time: 0:57:33
|
| 134 |
+
Epoch 39 | iter 736 step 23 | loss train: 0.007, val: n/a | iter time: 989.45 ms (step) remaining time: 0:56:00
|
| 135 |
+
Epoch 41 | iter 768 step 24 | loss train: 0.006, val: n/a | iter time: 992.48 ms (step) remaining time: 0:54:31
|
| 136 |
+
Epoch 43 | iter 800 step 25 | loss train: 0.006, val: n/a | iter time: 989.03 ms (step) remaining time: 0:53:07
|
| 137 |
+
Epoch 44 | iter 832 step 26 | loss train: 0.005, val: n/a | iter time: 990.26 ms (step) remaining time: 0:51:47
|
| 138 |
+
Epoch 46 | iter 864 step 27 | loss train: 0.005, val: n/a | iter time: 990.10 ms (step) remaining time: 0:50:32
|
| 139 |
+
Epoch 48 | iter 896 step 28 | loss train: 0.005, val: n/a | iter time: 990.45 ms (step) remaining time: 0:49:19
|
| 140 |
+
Epoch 49 | iter 928 step 29 | loss train: 0.005, val: n/a | iter time: 989.48 ms (step) remaining time: 0:48:09
|
| 141 |
+
Epoch 51 | iter 960 step 30 | loss train: 0.004, val: n/a | iter time: 989.24 ms (step) remaining time: 0:47:03
|
| 142 |
+
Saving checkpoint to 'out/pretrain/qwen2_7b_question_focus_lr_plus/step-00000030/lit_model.pth'
|
| 143 |
+
Epoch 53 | iter 992 step 31 | loss train: 0.005, val: n/a | iter time: 986.57 ms (step) remaining time: 0:48:01
|
| 144 |
+
Epoch 54 | iter 1024 step 32 | loss train: 0.004, val: n/a | iter time: 990.88 ms (step) remaining time: 0:46:53
|
| 145 |
+
Epoch 56 | iter 1056 step 33 | loss train: 0.004, val: n/a | iter time: 990.30 ms (step) remaining time: 0:45:48
|
| 146 |
+
Epoch 58 | iter 1088 step 34 | loss train: 0.005, val: n/a | iter time: 990.78 ms (step) remaining time: 0:44:44
|
| 147 |
+
Epoch 59 | iter 1120 step 35 | loss train: 0.004, val: n/a | iter time: 988.25 ms (step) remaining time: 0:43:43
|
| 148 |
+
Epoch 61 | iter 1152 step 36 | loss train: 0.003, val: n/a | iter time: 994.09 ms (step) remaining time: 0:42:43
|
| 149 |
+
Epoch 63 | iter 1184 step 37 | loss train: 0.003, val: n/a | iter time: 990.38 ms (step) remaining time: 0:41:45
|
| 150 |
+
Epoch 64 | iter 1216 step 38 | loss train: 0.003, val: n/a | iter time: 991.09 ms (step) remaining time: 0:40:48
|
| 151 |
+
Epoch 66 | iter 1248 step 39 | loss train: 0.003, val: n/a | iter time: 990.35 ms (step) remaining time: 0:39:53
|
| 152 |
+
Epoch 68 | iter 1280 step 40 | loss train: 0.003, val: n/a | iter time: 990.50 ms (step) remaining time: 0:38:59
|
| 153 |
+
Saving checkpoint to 'out/pretrain/qwen2_7b_question_focus_lr_plus/step-00000040/lit_model.pth'
|
| 154 |
+
Epoch 70 | iter 1312 step 41 | loss train: 0.002, val: n/a | iter time: 985.04 ms (step) remaining time: 0:39:28
|
| 155 |
+
Epoch 71 | iter 1344 step 42 | loss train: 0.002, val: n/a | iter time: 988.08 ms (step) remaining time: 0:38:33
|
| 156 |
+
Epoch 73 | iter 1376 step 43 | loss train: 0.002, val: n/a | iter time: 989.22 ms (step) remaining time: 0:37:39
|
| 157 |
+
Epoch 75 | iter 1408 step 44 | loss train: 0.002, val: n/a | iter time: 988.61 ms (step) remaining time: 0:36:47
|
| 158 |
+
Epoch 76 | iter 1440 step 45 | loss train: 0.002, val: n/a | iter time: 988.63 ms (step) remaining time: 0:35:56
|
| 159 |
+
Epoch 78 | iter 1472 step 46 | loss train: 0.002, val: n/a | iter time: 991.07 ms (step) remaining time: 0:35:05
|
| 160 |
+
Epoch 80 | iter 1504 step 47 | loss train: 0.002, val: n/a | iter time: 993.71 ms (step) remaining time: 0:34:16
|
| 161 |
+
Epoch 81 | iter 1536 step 48 | loss train: 0.002, val: n/a | iter time: 994.17 ms (step) remaining time: 0:33:27
|
| 162 |
+
Epoch 83 | iter 1568 step 49 | loss train: 0.002, val: n/a | iter time: 992.13 ms (step) remaining time: 0:32:39
|
| 163 |
+
Epoch 85 | iter 1600 step 50 | loss train: 0.001, val: n/a | iter time: 989.11 ms (step) remaining time: 0:31:52
|
| 164 |
+
Saving checkpoint to 'out/pretrain/qwen2_7b_question_focus_lr_plus/step-00000050/lit_model.pth'
|
| 165 |
+
Epoch 86 | iter 1632 step 51 | loss train: 0.001, val: n/a | iter time: 987.12 ms (step) remaining time: 0:31:57
|
| 166 |
+
Epoch 88 | iter 1664 step 52 | loss train: 0.001, val: n/a | iter time: 988.33 ms (step) remaining time: 0:31:09
|
| 167 |
+
Epoch 90 | iter 1696 step 53 | loss train: 0.001, val: n/a | iter time: 988.65 ms (step) remaining time: 0:30:21
|
| 168 |
+
Epoch 91 | iter 1728 step 54 | loss train: 0.001, val: n/a | iter time: 990.42 ms (step) remaining time: 0:29:35
|
| 169 |
+
Epoch 93 | iter 1760 step 55 | loss train: 0.001, val: n/a | iter time: 991.74 ms (step) remaining time: 0:28:49
|
| 170 |
+
Epoch 95 | iter 1792 step 56 | loss train: 0.001, val: n/a | iter time: 992.71 ms (step) remaining time: 0:28:03
|
| 171 |
+
Epoch 96 | iter 1824 step 57 | loss train: 0.001, val: n/a | iter time: 990.15 ms (step) remaining time: 0:27:18
|
| 172 |
+
Epoch 98 | iter 1856 step 58 | loss train: 0.001, val: n/a | iter time: 988.80 ms (step) remaining time: 0:26:34
|
| 173 |
+
Epoch 100 | iter 1888 step 59 | loss train: 0.001, val: n/a | iter time: 992.54 ms (step) remaining time: 0:25:50
|
| 174 |
+
Epoch 102 | iter 1920 step 60 | loss train: 0.001, val: n/a | iter time: 987.75 ms (step) remaining time: 0:25:07
|
| 175 |
+
Saving checkpoint to 'out/pretrain/qwen2_7b_question_focus_lr_plus/step-00000060/lit_model.pth'
|
| 176 |
+
Epoch 103 | iter 1952 step 61 | loss train: 0.001, val: n/a | iter time: 990.43 ms (step) remaining time: 0:25:00
|
| 177 |
+
Epoch 105 | iter 1984 step 62 | loss train: 0.001, val: n/a | iter time: 987.48 ms (step) remaining time: 0:24:16
|
| 178 |
+
Epoch 107 | iter 2016 step 63 | loss train: 0.001, val: n/a | iter time: 989.32 ms (step) remaining time: 0:23:32
|
| 179 |
+
Epoch 108 | iter 2048 step 64 | loss train: 0.001, val: n/a | iter time: 988.99 ms (step) remaining time: 0:22:49
|
| 180 |
+
Epoch 110 | iter 2080 step 65 | loss train: 0.001, val: n/a | iter time: 994.41 ms (step) remaining time: 0:22:07
|
| 181 |
+
Epoch 112 | iter 2112 step 66 | loss train: 0.001, val: n/a | iter time: 992.36 ms (step) remaining time: 0:21:24
|
| 182 |
+
Epoch 113 | iter 2144 step 67 | loss train: 0.001, val: n/a | iter time: 990.94 ms (step) remaining time: 0:20:43
|
| 183 |
+
Epoch 115 | iter 2176 step 68 | loss train: 0.001, val: n/a | iter time: 990.64 ms (step) remaining time: 0:20:01
|
| 184 |
+
Epoch 117 | iter 2208 step 69 | loss train: 0.001, val: n/a | iter time: 993.15 ms (step) remaining time: 0:19:20
|
| 185 |
+
Epoch 118 | iter 2240 step 70 | loss train: 0.001, val: n/a | iter time: 992.78 ms (step) remaining time: 0:18:39
|
| 186 |
+
Saving checkpoint to 'out/pretrain/qwen2_7b_question_focus_lr_plus/step-00000070/lit_model.pth'
|
| 187 |
+
Epoch 120 | iter 2272 step 71 | loss train: 0.001, val: n/a | iter time: 987.65 ms (step) remaining time: 0:18:22
|
| 188 |
+
Epoch 122 | iter 2304 step 72 | loss train: 0.001, val: n/a | iter time: 989.30 ms (step) remaining time: 0:17:41
|
| 189 |
+
Epoch 123 | iter 2336 step 73 | loss train: 0.001, val: n/a | iter time: 988.89 ms (step) remaining time: 0:17:00
|
| 190 |
+
Epoch 125 | iter 2368 step 74 | loss train: 0.001, val: n/a | iter time: 995.48 ms (step) remaining time: 0:16:19
|
| 191 |
+
Epoch 127 | iter 2400 step 75 | loss train: 0.001, val: n/a | iter time: 992.03 ms (step) remaining time: 0:15:39
|
| 192 |
+
Epoch 128 | iter 2432 step 76 | loss train: 0.001, val: n/a | iter time: 990.07 ms (step) remaining time: 0:14:59
|
| 193 |
+
Epoch 130 | iter 2464 step 77 | loss train: 0.001, val: n/a | iter time: 988.70 ms (step) remaining time: 0:14:19
|
| 194 |
+
Epoch 132 | iter 2496 step 78 | loss train: 0.001, val: n/a | iter time: 994.75 ms (step) remaining time: 0:13:39
|
| 195 |
+
Epoch 134 | iter 2528 step 79 | loss train: 0.001, val: n/a | iter time: 991.65 ms (step) remaining time: 0:13:00
|
| 196 |
+
Epoch 135 | iter 2560 step 80 | loss train: 0.001, val: n/a | iter time: 994.05 ms (step) remaining time: 0:12:21
|
| 197 |
+
Saving checkpoint to 'out/pretrain/qwen2_7b_question_focus_lr_plus/step-00000080/lit_model.pth'
|
| 198 |
+
Epoch 137 | iter 2592 step 81 | loss train: 0.001, val: n/a | iter time: 989.12 ms (step) remaining time: 0:11:55
|
| 199 |
+
Epoch 139 | iter 2624 step 82 | loss train: 0.001, val: n/a | iter time: 992.55 ms (step) remaining time: 0:11:15
|
| 200 |
+
Epoch 140 | iter 2656 step 83 | loss train: 0.001, val: n/a | iter time: 993.71 ms (step) remaining time: 0:10:36
|
| 201 |
+
Epoch 142 | iter 2688 step 84 | loss train: 0.001, val: n/a | iter time: 987.33 ms (step) remaining time: 0:09:57
|
| 202 |
+
Epoch 144 | iter 2720 step 85 | loss train: 0.001, val: n/a | iter time: 992.72 ms (step) remaining time: 0:09:18
|
| 203 |
+
Epoch 145 | iter 2752 step 86 | loss train: 0.001, val: n/a | iter time: 992.75 ms (step) remaining time: 0:08:40
|
| 204 |
+
Epoch 147 | iter 2784 step 87 | loss train: 0.001, val: n/a | iter time: 990.86 ms (step) remaining time: 0:08:02
|
| 205 |
+
Epoch 149 | iter 2816 step 88 | loss train: 0.001, val: n/a | iter time: 992.41 ms (step) remaining time: 0:07:23
|
| 206 |
+
Epoch 150 | iter 2848 step 89 | loss train: 0.001, val: n/a | iter time: 996.79 ms (step) remaining time: 0:06:46
|
| 207 |
+
Epoch 152 | iter 2880 step 90 | loss train: 0.001, val: n/a | iter time: 987.57 ms (step) remaining time: 0:06:08
|
| 208 |
+
Saving checkpoint to 'out/pretrain/qwen2_7b_question_focus_lr_plus/step-00000090/lit_model.pth'
|
| 209 |
+
Epoch 154 | iter 2912 step 91 | loss train: 0.001, val: n/a | iter time: 988.50 ms (step) remaining time: 0:05:36
|
| 210 |
+
Epoch 155 | iter 2944 step 92 | loss train: 0.001, val: n/a | iter time: 989.06 ms (step) remaining time: 0:04:57
|
| 211 |
+
Epoch 157 | iter 2976 step 93 | loss train: 0.001, val: n/a | iter time: 994.86 ms (step) remaining time: 0:04:20
|
| 212 |
+
Epoch 159 | iter 3008 step 94 | loss train: 0.001, val: n/a | iter time: 990.57 ms (step) remaining time: 0:03:42
|
| 213 |
+
Epoch 160 | iter 3040 step 95 | loss train: 0.001, val: n/a | iter time: 993.62 ms (step) remaining time: 0:03:05
|
| 214 |
+
Epoch 162 | iter 3072 step 96 | loss train: 0.001, val: n/a | iter time: 990.85 ms (step) remaining time: 0:02:27
|
| 215 |
+
Epoch 164 | iter 3104 step 97 | loss train: 0.001, val: n/a | iter time: 992.47 ms (step) remaining time: 0:01:50
|
| 216 |
+
Epoch 166 | iter 3136 step 98 | loss train: 0.001, val: n/a | iter time: 989.93 ms (step) remaining time: 0:01:13
|
| 217 |
+
Epoch 167 | iter 3168 step 99 | loss train: 0.001, val: n/a | iter time: 990.92 ms (step) remaining time: 0:00:36
|
| 218 |
+
Epoch 169 | iter 3200 step 100 | loss train: 0.001, val: n/a | iter time: 992.26 ms (step) remaining time: 0:00:00
|
| 219 |
+
Saving checkpoint to 'out/pretrain/qwen2_7b_question_focus_lr_plus/step-00000100/lit_model.pth'
|
| 220 |
+
Saving checkpoint to 'out/pretrain/qwen2_7b_question_focus_lr_plus/final/lit_model.pth'
|
| 221 |
+
----------------------------------------
|
| 222 |
+
| Performance
|
| 223 |
+
| - Total tokens : 52,428,800
|
| 224 |
+
| - Training Time : 4129.90 s
|
| 225 |
+
| - Tok/sec : 6.05 tok/s
|
| 226 |
+
| ----------------------------------------
|
| 227 |
+
| Memory Usage
|
| 228 |
+
| - Memory Used : 56.04 GB
|
| 229 |
+
----------------------------------------
|
out/pretrain/tinyllama/teelogs/2407.txt
ADDED
|
@@ -0,0 +1,321 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
Initializing distributed: GLOBAL_RANK: 0, MEMBER: 1/4
|
| 2 |
+
Initializing distributed: GLOBAL_RANK: 2, MEMBER: 3/4
|
| 3 |
+
[rank: 2] Seed set to 42
|
| 4 |
+
Initializing distributed: GLOBAL_RANK: 1, MEMBER: 2/4
|
| 5 |
+
Initializing distributed: GLOBAL_RANK: 3, MEMBER: 4/4
|
| 6 |
+
[rank: 3] Seed set to 42
|
| 7 |
+
----------------------------------------------------------------------------------------------------
|
| 8 |
+
distributed_backend=nccl
|
| 9 |
+
All distributed processes registered. Starting with 4 processes
|
| 10 |
+
----------------------------------------------------------------------------------------------------
|
| 11 |
+
|
| 12 |
+
[rank: 1] Seed set to 42
|
| 13 |
+
[rank: 0] Seed set to 42
|
| 14 |
+
All GPUs are fully connected via NVLink.
|
| 15 |
+
{'data': {'batch_size': 1,
|
| 16 |
+
'data_path': PosixPath('litgpt/data/arxiv'),
|
| 17 |
+
'num_workers': 8,
|
| 18 |
+
'seed': 42,
|
| 19 |
+
'seq_length': 2048,
|
| 20 |
+
'use_starcoder': True},
|
| 21 |
+
'devices': 'auto',
|
| 22 |
+
'eval': {'evaluate_example': 'first',
|
| 23 |
+
'final_validation': True,
|
| 24 |
+
'initial_validation': True,
|
| 25 |
+
'interval': 50,
|
| 26 |
+
'max_iters': 100,
|
| 27 |
+
'max_new_tokens': None},
|
| 28 |
+
'initial_checkpoint_dir': PosixPath('checkpoints/TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T'),
|
| 29 |
+
'log': {'group': None, 'project': None, 'run': None},
|
| 30 |
+
'logger_name': 'tensorboard',
|
| 31 |
+
'model_config': {'attention_logit_softcapping': None,
|
| 32 |
+
'attention_scores_scalar': None,
|
| 33 |
+
'attn_bias': False,
|
| 34 |
+
'bias': False,
|
| 35 |
+
'block_size': 2048,
|
| 36 |
+
'final_logit_softcapping': None,
|
| 37 |
+
'gelu_approximate': 'none',
|
| 38 |
+
'head_size': 64,
|
| 39 |
+
'hf_config': {'name': 'TinyLlama-1.1B-intermediate-step-1431k-3T',
|
| 40 |
+
'org': 'TinyLlama'},
|
| 41 |
+
'intermediate_size': 5632,
|
| 42 |
+
'lm_head_bias': False,
|
| 43 |
+
'mlp_class_name': 'LLaMAMLP',
|
| 44 |
+
'moe_intermediate_size': None,
|
| 45 |
+
'n_embd': 2048,
|
| 46 |
+
'n_expert': 0,
|
| 47 |
+
'n_expert_per_token': 0,
|
| 48 |
+
'n_head': 32,
|
| 49 |
+
'n_layer': 22,
|
| 50 |
+
'n_query_groups': 4,
|
| 51 |
+
'name': 'tiny-llama-1.1b',
|
| 52 |
+
'norm_1': True,
|
| 53 |
+
'norm_2': True,
|
| 54 |
+
'norm_class_name': 'RMSNorm',
|
| 55 |
+
'norm_eps': 1e-05,
|
| 56 |
+
'norm_qk': False,
|
| 57 |
+
'norm_qk_type': 'default',
|
| 58 |
+
'padded_vocab_size': 32000,
|
| 59 |
+
'padding_multiple': 64,
|
| 60 |
+
'parallel_residual': False,
|
| 61 |
+
'post_attention_norm': False,
|
| 62 |
+
'post_mlp_norm': False,
|
| 63 |
+
'rope_adjustments': None,
|
| 64 |
+
'rope_base': 10000,
|
| 65 |
+
'rope_condense_ratio': 1,
|
| 66 |
+
'rope_indices': None,
|
| 67 |
+
'rope_local_base_freq': None,
|
| 68 |
+
'rotary_percentage': 1.0,
|
| 69 |
+
'scale_embeddings': False,
|
| 70 |
+
'shared_attention_norm': False,
|
| 71 |
+
'sliding_window_indices': None,
|
| 72 |
+
'sliding_window_size': None,
|
| 73 |
+
'vocab_size': 32000},
|
| 74 |
+
'model_name': 'tiny-llama-1.1b',
|
| 75 |
+
'num_nodes': 1,
|
| 76 |
+
'optimizer': "{'class_path': 'torch.optim.AdamW', 'init_args': {'lr': 0.0004, "
|
| 77 |
+
"'weight_decay': 0.1, 'betas': [0.9, 0.95]}}",
|
| 78 |
+
'out_dir': PosixPath('out/pretrain/tiny-llama-cl-2407'),
|
| 79 |
+
'precision': 'bf16-mixed',
|
| 80 |
+
'resume': False,
|
| 81 |
+
'seed': 42,
|
| 82 |
+
'tokenizer_dir': PosixPath('checkpoints/TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T'),
|
| 83 |
+
'train': {'epochs': None,
|
| 84 |
+
'global_batch_size': 512,
|
| 85 |
+
'log_interval': 1,
|
| 86 |
+
'lr_warmup_fraction': None,
|
| 87 |
+
'lr_warmup_steps': 20,
|
| 88 |
+
'max_norm': 1.0,
|
| 89 |
+
'max_seq_length': 2048,
|
| 90 |
+
'max_steps': None,
|
| 91 |
+
'max_tokens': 209715200,
|
| 92 |
+
'micro_batch_size': 4,
|
| 93 |
+
'min_lr': 4e-05,
|
| 94 |
+
'save_interval': 100,
|
| 95 |
+
'tie_embeddings': None}}
|
| 96 |
+
Time to instantiate model: 0.02 seconds.
|
| 97 |
+
Total parameters: 1,100,048,384
|
| 98 |
+
Validating ...
|
| 99 |
+
Measured TFLOPs: 239.66
|
| 100 |
+
Epoch 1 | iter 32 step 1 | loss train: 1.482, val: 1.516 | iter time: 539.29 ms (step) remaining time: 0:39:18
|
| 101 |
+
Epoch 1 | iter 64 step 2 | loss train: 1.503, val: 1.516 | iter time: 357.93 ms (step) remaining time: 0:37:16
|
| 102 |
+
Epoch 1 | iter 96 step 3 | loss train: 1.500, val: 1.516 | iter time: 357.13 ms (step) remaining time: 0:36:30
|
| 103 |
+
Epoch 1 | iter 128 step 4 | loss train: 1.515, val: 1.516 | iter time: 357.97 ms (step) remaining time: 0:36:03
|
| 104 |
+
Epoch 1 | iter 160 step 5 | loss train: 1.564, val: 1.516 | iter time: 359.36 ms (step) remaining time: 0:35:43
|
| 105 |
+
Epoch 1 | iter 192 step 6 | loss train: 1.819, val: 1.516 | iter time: 359.46 ms (step) remaining time: 0:35:27
|
| 106 |
+
Epoch 1 | iter 224 step 7 | loss train: 1.808, val: 1.516 | iter time: 357.37 ms (step) remaining time: 0:35:12
|
| 107 |
+
Epoch 1 | iter 256 step 8 | loss train: 1.692, val: 1.516 | iter time: 358.38 ms (step) remaining time: 0:34:58
|
| 108 |
+
Epoch 1 | iter 288 step 9 | loss train: 1.620, val: 1.516 | iter time: 359.95 ms (step) remaining time: 0:34:45
|
| 109 |
+
Epoch 1 | iter 320 step 10 | loss train: 1.598, val: 1.516 | iter time: 359.83 ms (step) remaining time: 0:34:33
|
| 110 |
+
Epoch 1 | iter 352 step 11 | loss train: 1.594, val: 1.516 | iter time: 359.85 ms (step) remaining time: 0:34:21
|
| 111 |
+
Epoch 1 | iter 384 step 12 | loss train: 1.536, val: 1.516 | iter time: 359.22 ms (step) remaining time: 0:34:09
|
| 112 |
+
Epoch 1 | iter 416 step 13 | loss train: 1.553, val: 1.516 | iter time: 359.10 ms (step) remaining time: 0:33:57
|
| 113 |
+
Epoch 1 | iter 448 step 14 | loss train: 1.647, val: 1.516 | iter time: 359.76 ms (step) remaining time: 0:33:46
|
| 114 |
+
Epoch 1 | iter 480 step 15 | loss train: 1.564, val: 1.516 | iter time: 357.45 ms (step) remaining time: 0:33:34
|
| 115 |
+
Epoch 1 | iter 512 step 16 | loss train: 1.564, val: 1.516 | iter time: 358.71 ms (step) remaining time: 0:33:23
|
| 116 |
+
Epoch 1 | iter 544 step 17 | loss train: 1.575, val: 1.516 | iter time: 359.15 ms (step) remaining time: 0:33:12
|
| 117 |
+
Epoch 1 | iter 576 step 18 | loss train: 1.584, val: 1.516 | iter time: 358.31 ms (step) remaining time: 0:33:00
|
| 118 |
+
Epoch 1 | iter 608 step 19 | loss train: 1.653, val: 1.516 | iter time: 359.31 ms (step) remaining time: 0:32:49
|
| 119 |
+
Epoch 1 | iter 640 step 20 | loss train: 1.619, val: 1.516 | iter time: 359.90 ms (step) remaining time: 0:32:38
|
| 120 |
+
Epoch 1 | iter 672 step 21 | loss train: 1.645, val: 1.516 | iter time: 361.40 ms (step) remaining time: 0:32:27
|
| 121 |
+
Epoch 1 | iter 704 step 22 | loss train: 1.527, val: 1.516 | iter time: 359.42 ms (step) remaining time: 0:32:16
|
| 122 |
+
Epoch 1 | iter 736 step 23 | loss train: 1.632, val: 1.516 | iter time: 359.27 ms (step) remaining time: 0:32:05
|
| 123 |
+
Epoch 1 | iter 768 step 24 | loss train: 1.552, val: 1.516 | iter time: 359.93 ms (step) remaining time: 0:31:54
|
| 124 |
+
Epoch 1 | iter 800 step 25 | loss train: 1.586, val: 1.516 | iter time: 360.68 ms (step) remaining time: 0:31:43
|
| 125 |
+
Epoch 1 | iter 832 step 26 | loss train: 1.574, val: 1.516 | iter time: 359.89 ms (step) remaining time: 0:31:32
|
| 126 |
+
Epoch 1 | iter 864 step 27 | loss train: 1.554, val: 1.516 | iter time: 358.49 ms (step) remaining time: 0:31:21
|
| 127 |
+
Epoch 1 | iter 896 step 28 | loss train: 1.662, val: 1.516 | iter time: 360.49 ms (step) remaining time: 0:31:10
|
| 128 |
+
Epoch 1 | iter 928 step 29 | loss train: 1.539, val: 1.516 | iter time: 357.63 ms (step) remaining time: 0:30:59
|
| 129 |
+
Epoch 1 | iter 960 step 30 | loss train: 1.608, val: 1.516 | iter time: 359.51 ms (step) remaining time: 0:30:48
|
| 130 |
+
Epoch 1 | iter 992 step 31 | loss train: 1.547, val: 1.516 | iter time: 360.57 ms (step) remaining time: 0:30:37
|
| 131 |
+
Epoch 1 | iter 1024 step 32 | loss train: 1.619, val: 1.516 | iter time: 359.55 ms (step) remaining time: 0:30:26
|
| 132 |
+
Epoch 1 | iter 1056 step 33 | loss train: 1.577, val: 1.516 | iter time: 361.56 ms (step) remaining time: 0:30:15
|
| 133 |
+
Epoch 1 | iter 1088 step 34 | loss train: 1.599, val: 1.516 | iter time: 358.77 ms (step) remaining time: 0:30:05
|
| 134 |
+
Epoch 1 | iter 1120 step 35 | loss train: 1.550, val: 1.516 | iter time: 360.67 ms (step) remaining time: 0:29:54
|
| 135 |
+
Epoch 1 | iter 1152 step 36 | loss train: 1.533, val: 1.516 | iter time: 360.06 ms (step) remaining time: 0:29:43
|
| 136 |
+
Epoch 1 | iter 1184 step 37 | loss train: 1.513, val: 1.516 | iter time: 359.84 ms (step) remaining time: 0:29:33
|
| 137 |
+
Epoch 1 | iter 1216 step 38 | loss train: 1.606, val: 1.516 | iter time: 360.69 ms (step) remaining time: 0:29:22
|
| 138 |
+
Epoch 1 | iter 1248 step 39 | loss train: 1.558, val: 1.516 | iter time: 359.09 ms (step) remaining time: 0:29:12
|
| 139 |
+
Epoch 1 | iter 1280 step 40 | loss train: 1.489, val: 1.516 | iter time: 361.79 ms (step) remaining time: 0:29:01
|
| 140 |
+
Epoch 1 | iter 1312 step 41 | loss train: 1.533, val: 1.516 | iter time: 360.68 ms (step) remaining time: 0:28:50
|
| 141 |
+
Epoch 1 | iter 1344 step 42 | loss train: 1.495, val: 1.516 | iter time: 361.09 ms (step) remaining time: 0:28:39
|
| 142 |
+
Epoch 1 | iter 1376 step 43 | loss train: 1.547, val: 1.516 | iter time: 360.76 ms (step) remaining time: 0:28:28
|
| 143 |
+
Epoch 1 | iter 1408 step 44 | loss train: 1.558, val: 1.516 | iter time: 359.65 ms (step) remaining time: 0:28:17
|
| 144 |
+
Epoch 1 | iter 1440 step 45 | loss train: 1.586, val: 1.516 | iter time: 359.14 ms (step) remaining time: 0:28:06
|
| 145 |
+
Epoch 1 | iter 1472 step 46 | loss train: 1.533, val: 1.516 | iter time: 360.92 ms (step) remaining time: 0:27:55
|
| 146 |
+
Epoch 1 | iter 1504 step 47 | loss train: 1.532, val: 1.516 | iter time: 358.97 ms (step) remaining time: 0:27:44
|
| 147 |
+
Epoch 1 | iter 1536 step 48 | loss train: 1.502, val: 1.516 | iter time: 359.92 ms (step) remaining time: 0:27:33
|
| 148 |
+
Epoch 1 | iter 1568 step 49 | loss train: 1.510, val: 1.516 | iter time: 358.66 ms (step) remaining time: 0:27:23
|
| 149 |
+
Epoch 1 | iter 1600 step 50 | loss train: 1.517, val: 1.516 | iter time: 359.42 ms (step) remaining time: 0:27:12
|
| 150 |
+
Validating ...
|
| 151 |
+
iter 1600: val loss 1.5616, val time: 6628.66 ms
|
| 152 |
+
Epoch 1 | iter 1632 step 51 | loss train: 1.507, val: 1.562 | iter time: 358.98 ms (step) remaining time: 0:27:20
|
| 153 |
+
Epoch 1 | iter 1664 step 52 | loss train: 1.527, val: 1.562 | iter time: 359.82 ms (step) remaining time: 0:27:09
|
| 154 |
+
Epoch 1 | iter 1696 step 53 | loss train: 1.489, val: 1.562 | iter time: 361.73 ms (step) remaining time: 0:26:57
|
| 155 |
+
Epoch 1 | iter 1728 step 54 | loss train: 1.456, val: 1.562 | iter time: 359.39 ms (step) remaining time: 0:26:46
|
| 156 |
+
Epoch 1 | iter 1760 step 55 | loss train: 1.488, val: 1.562 | iter time: 359.70 ms (step) remaining time: 0:26:35
|
| 157 |
+
Epoch 1 | iter 1792 step 56 | loss train: 1.533, val: 1.562 | iter time: 359.97 ms (step) remaining time: 0:26:23
|
| 158 |
+
Epoch 1 | iter 1824 step 57 | loss train: 1.526, val: 1.562 | iter time: 359.95 ms (step) remaining time: 0:26:12
|
| 159 |
+
Epoch 1 | iter 1856 step 58 | loss train: 1.454, val: 1.562 | iter time: 359.66 ms (step) remaining time: 0:26:01
|
| 160 |
+
Epoch 1 | iter 1888 step 59 | loss train: 1.540, val: 1.562 | iter time: 358.72 ms (step) remaining time: 0:25:49
|
| 161 |
+
Epoch 1 | iter 1920 step 60 | loss train: 1.457, val: 1.562 | iter time: 358.39 ms (step) remaining time: 0:25:38
|
| 162 |
+
Epoch 1 | iter 1952 step 61 | loss train: 1.477, val: 1.562 | iter time: 362.21 ms (step) remaining time: 0:25:27
|
| 163 |
+
Epoch 1 | iter 1984 step 62 | loss train: 1.504, val: 1.562 | iter time: 360.92 ms (step) remaining time: 0:25:16
|
| 164 |
+
Epoch 1 | iter 2016 step 63 | loss train: 1.550, val: 1.562 | iter time: 360.32 ms (step) remaining time: 0:25:04
|
| 165 |
+
Epoch 1 | iter 2048 step 64 | loss train: 1.455, val: 1.562 | iter time: 358.84 ms (step) remaining time: 0:24:53
|
| 166 |
+
Epoch 1 | iter 2080 step 65 | loss train: 1.513, val: 1.562 | iter time: 360.02 ms (step) remaining time: 0:24:42
|
| 167 |
+
Epoch 1 | iter 2112 step 66 | loss train: 1.529, val: 1.562 | iter time: 360.58 ms (step) remaining time: 0:24:31
|
| 168 |
+
Epoch 1 | iter 2144 step 67 | loss train: 1.479, val: 1.562 | iter time: 360.08 ms (step) remaining time: 0:24:20
|
| 169 |
+
Epoch 1 | iter 2176 step 68 | loss train: 1.486, val: 1.562 | iter time: 359.46 ms (step) remaining time: 0:24:08
|
| 170 |
+
Epoch 1 | iter 2208 step 69 | loss train: 1.503, val: 1.562 | iter time: 360.66 ms (step) remaining time: 0:23:57
|
| 171 |
+
Epoch 1 | iter 2240 step 70 | loss train: 1.510, val: 1.562 | iter time: 361.29 ms (step) remaining time: 0:23:46
|
| 172 |
+
Epoch 1 | iter 2272 step 71 | loss train: 1.531, val: 1.562 | iter time: 359.82 ms (step) remaining time: 0:23:35
|
| 173 |
+
Epoch 1 | iter 2304 step 72 | loss train: 1.534, val: 1.562 | iter time: 359.55 ms (step) remaining time: 0:23:24
|
| 174 |
+
Epoch 1 | iter 2336 step 73 | loss train: 1.541, val: 1.562 | iter time: 360.04 ms (step) remaining time: 0:23:13
|
| 175 |
+
Epoch 1 | iter 2368 step 74 | loss train: 1.502, val: 1.562 | iter time: 360.40 ms (step) remaining time: 0:23:02
|
| 176 |
+
Epoch 1 | iter 2400 step 75 | loss train: 1.452, val: 1.562 | iter time: 358.72 ms (step) remaining time: 0:22:50
|
| 177 |
+
Epoch 1 | iter 2432 step 76 | loss train: 1.507, val: 1.562 | iter time: 360.41 ms (step) remaining time: 0:22:39
|
| 178 |
+
Epoch 1 | iter 2464 step 77 | loss train: 1.470, val: 1.562 | iter time: 358.43 ms (step) remaining time: 0:22:28
|
| 179 |
+
Epoch 1 | iter 2496 step 78 | loss train: 1.470, val: 1.562 | iter time: 359.45 ms (step) remaining time: 0:22:17
|
| 180 |
+
Epoch 1 | iter 2528 step 79 | loss train: 1.500, val: 1.562 | iter time: 360.39 ms (step) remaining time: 0:22:06
|
| 181 |
+
Epoch 1 | iter 2560 step 80 | loss train: 1.505, val: 1.562 | iter time: 360.25 ms (step) remaining time: 0:21:55
|
| 182 |
+
Epoch 1 | iter 2592 step 81 | loss train: 1.536, val: 1.562 | iter time: 360.15 ms (step) remaining time: 0:21:44
|
| 183 |
+
Epoch 1 | iter 2624 step 82 | loss train: 1.445, val: 1.562 | iter time: 360.00 ms (step) remaining time: 0:21:33
|
| 184 |
+
Epoch 1 | iter 2656 step 83 | loss train: 1.493, val: 1.562 | iter time: 360.27 ms (step) remaining time: 0:21:22
|
| 185 |
+
Epoch 1 | iter 2688 step 84 | loss train: 1.455, val: 1.562 | iter time: 360.84 ms (step) remaining time: 0:21:11
|
| 186 |
+
Epoch 1 | iter 2720 step 85 | loss train: 1.536, val: 1.562 | iter time: 360.31 ms (step) remaining time: 0:21:00
|
| 187 |
+
Epoch 1 | iter 2752 step 86 | loss train: 1.533, val: 1.562 | iter time: 359.74 ms (step) remaining time: 0:20:49
|
| 188 |
+
Epoch 1 | iter 2784 step 87 | loss train: 1.469, val: 1.562 | iter time: 359.46 ms (step) remaining time: 0:20:38
|
| 189 |
+
Epoch 1 | iter 2816 step 88 | loss train: 1.495, val: 1.562 | iter time: 359.86 ms (step) remaining time: 0:20:26
|
| 190 |
+
Epoch 1 | iter 2848 step 89 | loss train: 1.477, val: 1.562 | iter time: 360.08 ms (step) remaining time: 0:20:15
|
| 191 |
+
Epoch 1 | iter 2880 step 90 | loss train: 1.643, val: 1.562 | iter time: 362.43 ms (step) remaining time: 0:20:05
|
| 192 |
+
Epoch 1 | iter 2912 step 91 | loss train: 1.529, val: 1.562 | iter time: 360.04 ms (step) remaining time: 0:19:54
|
| 193 |
+
Epoch 1 | iter 2944 step 92 | loss train: 1.460, val: 1.562 | iter time: 360.48 ms (step) remaining time: 0:19:43
|
| 194 |
+
Epoch 1 | iter 2976 step 93 | loss train: 1.487, val: 1.562 | iter time: 359.14 ms (step) remaining time: 0:19:32
|
| 195 |
+
Epoch 1 | iter 3008 step 94 | loss train: 1.463, val: 1.562 | iter time: 357.89 ms (step) remaining time: 0:19:20
|
| 196 |
+
Epoch 1 | iter 3040 step 95 | loss train: 1.511, val: 1.562 | iter time: 359.13 ms (step) remaining time: 0:19:09
|
| 197 |
+
Epoch 1 | iter 3072 step 96 | loss train: 1.442, val: 1.562 | iter time: 358.94 ms (step) remaining time: 0:18:58
|
| 198 |
+
Epoch 1 | iter 3104 step 97 | loss train: 1.454, val: 1.562 | iter time: 358.38 ms (step) remaining time: 0:18:47
|
| 199 |
+
Epoch 1 | iter 3136 step 98 | loss train: 1.474, val: 1.562 | iter time: 360.59 ms (step) remaining time: 0:18:36
|
| 200 |
+
Epoch 1 | iter 3168 step 99 | loss train: 1.445, val: 1.562 | iter time: 357.61 ms (step) remaining time: 0:18:25
|
| 201 |
+
Epoch 1 | iter 3200 step 100 | loss train: 1.512, val: 1.562 | iter time: 362.55 ms (step) remaining time: 0:18:14
|
| 202 |
+
Validating ...
|
| 203 |
+
iter 3200: val loss 1.5233, val time: 6640.06 ms
|
| 204 |
+
Saving checkpoint to 'out/pretrain/tiny-llama-cl-2407/step-00000100/lit_model.pth'
|
| 205 |
+
Epoch 1 | iter 3232 step 101 | loss train: 1.448, val: 1.523 | iter time: 356.20 ms (step) remaining time: 0:18:27
|
| 206 |
+
Epoch 1 | iter 3264 step 102 | loss train: 1.466, val: 1.523 | iter time: 356.62 ms (step) remaining time: 0:18:15
|
| 207 |
+
Epoch 1 | iter 3296 step 103 | loss train: 1.437, val: 1.523 | iter time: 359.01 ms (step) remaining time: 0:18:04
|
| 208 |
+
Epoch 1 | iter 3328 step 104 | loss train: 1.412, val: 1.523 | iter time: 360.12 ms (step) remaining time: 0:17:52
|
| 209 |
+
Epoch 1 | iter 3360 step 105 | loss train: 1.415, val: 1.523 | iter time: 358.81 ms (step) remaining time: 0:17:41
|
| 210 |
+
Epoch 1 | iter 3392 step 106 | loss train: 1.438, val: 1.523 | iter time: 360.69 ms (step) remaining time: 0:17:29
|
| 211 |
+
Epoch 1 | iter 3424 step 107 | loss train: 1.398, val: 1.523 | iter time: 360.57 ms (step) remaining time: 0:17:18
|
| 212 |
+
Epoch 1 | iter 3456 step 108 | loss train: 1.464, val: 1.523 | iter time: 359.88 ms (step) remaining time: 0:17:06
|
| 213 |
+
Epoch 1 | iter 3488 step 109 | loss train: 1.482, val: 1.523 | iter time: 360.54 ms (step) remaining time: 0:16:55
|
| 214 |
+
Epoch 1 | iter 3520 step 110 | loss train: 1.419, val: 1.523 | iter time: 359.31 ms (step) remaining time: 0:16:44
|
| 215 |
+
Epoch 1 | iter 3552 step 111 | loss train: 1.473, val: 1.523 | iter time: 359.97 ms (step) remaining time: 0:16:32
|
| 216 |
+
Epoch 1 | iter 3584 step 112 | loss train: 1.395, val: 1.523 | iter time: 360.46 ms (step) remaining time: 0:16:21
|
| 217 |
+
Epoch 1 | iter 3616 step 113 | loss train: 1.497, val: 1.523 | iter time: 358.72 ms (step) remaining time: 0:16:09
|
| 218 |
+
Epoch 1 | iter 3648 step 114 | loss train: 1.438, val: 1.523 | iter time: 360.58 ms (step) remaining time: 0:15:58
|
| 219 |
+
Epoch 1 | iter 3680 step 115 | loss train: 1.437, val: 1.523 | iter time: 359.75 ms (step) remaining time: 0:15:47
|
| 220 |
+
Epoch 1 | iter 3712 step 116 | loss train: 1.425, val: 1.523 | iter time: 359.15 ms (step) remaining time: 0:15:35
|
| 221 |
+
Epoch 1 | iter 3744 step 117 | loss train: 1.457, val: 1.523 | iter time: 360.72 ms (step) remaining time: 0:15:24
|
| 222 |
+
Epoch 1 | iter 3776 step 118 | loss train: 1.492, val: 1.523 | iter time: 359.21 ms (step) remaining time: 0:15:13
|
| 223 |
+
Epoch 1 | iter 3808 step 119 | loss train: 1.471, val: 1.523 | iter time: 358.41 ms (step) remaining time: 0:15:01
|
| 224 |
+
Epoch 1 | iter 3840 step 120 | loss train: 1.430, val: 1.523 | iter time: 359.80 ms (step) remaining time: 0:14:50
|
| 225 |
+
Epoch 1 | iter 3872 step 121 | loss train: 1.463, val: 1.523 | iter time: 359.17 ms (step) remaining time: 0:14:39
|
| 226 |
+
Epoch 1 | iter 3904 step 122 | loss train: 1.557, val: 1.523 | iter time: 358.69 ms (step) remaining time: 0:14:28
|
| 227 |
+
Epoch 1 | iter 3936 step 123 | loss train: 1.398, val: 1.523 | iter time: 357.78 ms (step) remaining time: 0:14:16
|
| 228 |
+
Epoch 1 | iter 3968 step 124 | loss train: 1.395, val: 1.523 | iter time: 359.69 ms (step) remaining time: 0:14:05
|
| 229 |
+
Epoch 1 | iter 4000 step 125 | loss train: 1.423, val: 1.523 | iter time: 361.51 ms (step) remaining time: 0:13:54
|
| 230 |
+
Epoch 1 | iter 4032 step 126 | loss train: 1.393, val: 1.523 | iter time: 359.88 ms (step) remaining time: 0:13:42
|
| 231 |
+
Epoch 1 | iter 4064 step 127 | loss train: 1.474, val: 1.523 | iter time: 359.89 ms (step) remaining time: 0:13:31
|
| 232 |
+
Epoch 1 | iter 4096 step 128 | loss train: 1.410, val: 1.523 | iter time: 360.42 ms (step) remaining time: 0:13:20
|
| 233 |
+
Epoch 1 | iter 4128 step 129 | loss train: 1.413, val: 1.523 | iter time: 358.71 ms (step) remaining time: 0:13:09
|
| 234 |
+
Epoch 1 | iter 4160 step 130 | loss train: 1.391, val: 1.523 | iter time: 360.61 ms (step) remaining time: 0:12:57
|
| 235 |
+
Epoch 1 | iter 4192 step 131 | loss train: 1.536, val: 1.523 | iter time: 359.01 ms (step) remaining time: 0:12:46
|
| 236 |
+
Epoch 1 | iter 4224 step 132 | loss train: 1.413, val: 1.523 | iter time: 358.95 ms (step) remaining time: 0:12:35
|
| 237 |
+
Epoch 1 | iter 4256 step 133 | loss train: 1.446, val: 1.523 | iter time: 359.39 ms (step) remaining time: 0:12:24
|
| 238 |
+
Epoch 1 | iter 4288 step 134 | loss train: 1.382, val: 1.523 | iter time: 360.98 ms (step) remaining time: 0:12:12
|
| 239 |
+
Epoch 1 | iter 4320 step 135 | loss train: 1.471, val: 1.523 | iter time: 359.84 ms (step) remaining time: 0:12:01
|
| 240 |
+
Epoch 1 | iter 4352 step 136 | loss train: 1.497, val: 1.523 | iter time: 359.45 ms (step) remaining time: 0:11:50
|
| 241 |
+
Epoch 1 | iter 4384 step 137 | loss train: 1.422, val: 1.523 | iter time: 359.18 ms (step) remaining time: 0:11:39
|
| 242 |
+
Epoch 1 | iter 4416 step 138 | loss train: 1.464, val: 1.523 | iter time: 358.30 ms (step) remaining time: 0:11:28
|
| 243 |
+
Epoch 1 | iter 4448 step 139 | loss train: 1.430, val: 1.523 | iter time: 358.24 ms (step) remaining time: 0:11:16
|
| 244 |
+
Epoch 1 | iter 4480 step 140 | loss train: 1.441, val: 1.523 | iter time: 361.83 ms (step) remaining time: 0:11:05
|
| 245 |
+
Epoch 1 | iter 4512 step 141 | loss train: 1.416, val: 1.523 | iter time: 359.28 ms (step) remaining time: 0:10:54
|
| 246 |
+
Epoch 1 | iter 4544 step 142 | loss train: 1.491, val: 1.523 | iter time: 360.77 ms (step) remaining time: 0:10:43
|
| 247 |
+
Epoch 1 | iter 4576 step 143 | loss train: 1.399, val: 1.523 | iter time: 360.71 ms (step) remaining time: 0:10:32
|
| 248 |
+
Epoch 1 | iter 4608 step 144 | loss train: 1.367, val: 1.523 | iter time: 359.67 ms (step) remaining time: 0:10:21
|
| 249 |
+
Epoch 1 | iter 4640 step 145 | loss train: 1.403, val: 1.523 | iter time: 359.76 ms (step) remaining time: 0:10:10
|
| 250 |
+
Epoch 1 | iter 4672 step 146 | loss train: 1.433, val: 1.523 | iter time: 360.78 ms (step) remaining time: 0:09:58
|
| 251 |
+
Epoch 1 | iter 4704 step 147 | loss train: 1.380, val: 1.523 | iter time: 358.94 ms (step) remaining time: 0:09:47
|
| 252 |
+
Epoch 1 | iter 4736 step 148 | loss train: 1.496, val: 1.523 | iter time: 358.98 ms (step) remaining time: 0:09:36
|
| 253 |
+
Epoch 1 | iter 4768 step 149 | loss train: 1.386, val: 1.523 | iter time: 359.20 ms (step) remaining time: 0:09:25
|
| 254 |
+
Epoch 1 | iter 4800 step 150 | loss train: 1.393, val: 1.523 | iter time: 358.29 ms (step) remaining time: 0:09:14
|
| 255 |
+
Validating ...
|
| 256 |
+
iter 4800: val loss 1.4899, val time: 6636.92 ms
|
| 257 |
+
Epoch 1 | iter 4832 step 151 | loss train: 1.341, val: 1.490 | iter time: 360.17 ms (step) remaining time: 0:09:05
|
| 258 |
+
Epoch 1 | iter 4864 step 152 | loss train: 1.394, val: 1.490 | iter time: 357.93 ms (step) remaining time: 0:08:53
|
| 259 |
+
Epoch 1 | iter 4896 step 153 | loss train: 1.427, val: 1.490 | iter time: 360.60 ms (step) remaining time: 0:08:42
|
| 260 |
+
Epoch 1 | iter 4928 step 154 | loss train: 1.407, val: 1.490 | iter time: 359.68 ms (step) remaining time: 0:08:31
|
| 261 |
+
Epoch 1 | iter 4960 step 155 | loss train: 1.346, val: 1.490 | iter time: 358.64 ms (step) remaining time: 0:08:20
|
| 262 |
+
Epoch 1 | iter 4992 step 156 | loss train: 1.481, val: 1.490 | iter time: 359.33 ms (step) remaining time: 0:08:09
|
| 263 |
+
Epoch 1 | iter 5024 step 157 | loss train: 1.417, val: 1.490 | iter time: 360.26 ms (step) remaining time: 0:07:57
|
| 264 |
+
Epoch 1 | iter 5056 step 158 | loss train: 1.428, val: 1.490 | iter time: 359.38 ms (step) remaining time: 0:07:46
|
| 265 |
+
Epoch 1 | iter 5088 step 159 | loss train: 1.464, val: 1.490 | iter time: 360.43 ms (step) remaining time: 0:07:35
|
| 266 |
+
Epoch 1 | iter 5120 step 160 | loss train: 1.417, val: 1.490 | iter time: 359.95 ms (step) remaining time: 0:07:24
|
| 267 |
+
Epoch 1 | iter 5152 step 161 | loss train: 1.494, val: 1.490 | iter time: 359.53 ms (step) remaining time: 0:07:13
|
| 268 |
+
Epoch 1 | iter 5184 step 162 | loss train: 1.436, val: 1.490 | iter time: 360.53 ms (step) remaining time: 0:07:02
|
| 269 |
+
Epoch 1 | iter 5216 step 163 | loss train: 1.371, val: 1.490 | iter time: 360.98 ms (step) remaining time: 0:06:50
|
| 270 |
+
Epoch 1 | iter 5248 step 164 | loss train: 1.407, val: 1.490 | iter time: 357.79 ms (step) remaining time: 0:06:39
|
| 271 |
+
Epoch 1 | iter 5280 step 165 | loss train: 1.421, val: 1.490 | iter time: 360.59 ms (step) remaining time: 0:06:28
|
| 272 |
+
Epoch 1 | iter 5312 step 166 | loss train: 1.395, val: 1.490 | iter time: 357.97 ms (step) remaining time: 0:06:17
|
| 273 |
+
Epoch 1 | iter 5344 step 167 | loss train: 1.363, val: 1.490 | iter time: 359.06 ms (step) remaining time: 0:06:06
|
| 274 |
+
Epoch 1 | iter 5376 step 168 | loss train: 1.416, val: 1.490 | iter time: 360.67 ms (step) remaining time: 0:05:55
|
| 275 |
+
Epoch 1 | iter 5408 step 169 | loss train: 1.430, val: 1.490 | iter time: 360.25 ms (step) remaining time: 0:05:44
|
| 276 |
+
Epoch 1 | iter 5440 step 170 | loss train: 1.386, val: 1.490 | iter time: 360.98 ms (step) remaining time: 0:05:32
|
| 277 |
+
Epoch 1 | iter 5472 step 171 | loss train: 1.478, val: 1.490 | iter time: 360.58 ms (step) remaining time: 0:05:21
|
| 278 |
+
Epoch 1 | iter 5504 step 172 | loss train: 1.396, val: 1.490 | iter time: 360.11 ms (step) remaining time: 0:05:10
|
| 279 |
+
Epoch 1 | iter 5536 step 173 | loss train: 1.390, val: 1.490 | iter time: 359.66 ms (step) remaining time: 0:04:59
|
| 280 |
+
Epoch 1 | iter 5568 step 174 | loss train: 1.377, val: 1.490 | iter time: 359.39 ms (step) remaining time: 0:04:48
|
| 281 |
+
Epoch 1 | iter 5600 step 175 | loss train: 1.370, val: 1.490 | iter time: 360.91 ms (step) remaining time: 0:04:37
|
| 282 |
+
Epoch 1 | iter 5632 step 176 | loss train: 1.407, val: 1.490 | iter time: 360.76 ms (step) remaining time: 0:04:26
|
| 283 |
+
Epoch 1 | iter 5664 step 177 | loss train: 1.474, val: 1.490 | iter time: 358.30 ms (step) remaining time: 0:04:15
|
| 284 |
+
Epoch 1 | iter 5696 step 178 | loss train: 1.480, val: 1.490 | iter time: 359.12 ms (step) remaining time: 0:04:03
|
| 285 |
+
Epoch 1 | iter 5728 step 179 | loss train: 1.418, val: 1.490 | iter time: 361.16 ms (step) remaining time: 0:03:52
|
| 286 |
+
Epoch 1 | iter 5760 step 180 | loss train: 1.447, val: 1.490 | iter time: 359.35 ms (step) remaining time: 0:03:41
|
| 287 |
+
Epoch 1 | iter 5792 step 181 | loss train: 1.380, val: 1.490 | iter time: 359.73 ms (step) remaining time: 0:03:30
|
| 288 |
+
Epoch 1 | iter 5824 step 182 | loss train: 1.433, val: 1.490 | iter time: 359.42 ms (step) remaining time: 0:03:19
|
| 289 |
+
Epoch 1 | iter 5856 step 183 | loss train: 1.377, val: 1.490 | iter time: 359.69 ms (step) remaining time: 0:03:08
|
| 290 |
+
Epoch 1 | iter 5888 step 184 | loss train: 1.391, val: 1.490 | iter time: 358.68 ms (step) remaining time: 0:02:57
|
| 291 |
+
Epoch 1 | iter 5920 step 185 | loss train: 1.303, val: 1.490 | iter time: 360.25 ms (step) remaining time: 0:02:46
|
| 292 |
+
Epoch 1 | iter 5952 step 186 | loss train: 1.414, val: 1.490 | iter time: 357.99 ms (step) remaining time: 0:02:35
|
| 293 |
+
Epoch 1 | iter 5984 step 187 | loss train: 1.396, val: 1.490 | iter time: 360.09 ms (step) remaining time: 0:02:23
|
| 294 |
+
Epoch 1 | iter 6016 step 188 | loss train: 1.407, val: 1.490 | iter time: 360.49 ms (step) remaining time: 0:02:12
|
| 295 |
+
Epoch 1 | iter 6048 step 189 | loss train: 1.336, val: 1.490 | iter time: 359.10 ms (step) remaining time: 0:02:01
|
| 296 |
+
Epoch 1 | iter 6080 step 190 | loss train: 1.436, val: 1.490 | iter time: 359.23 ms (step) remaining time: 0:01:50
|
| 297 |
+
Epoch 1 | iter 6112 step 191 | loss train: 1.408, val: 1.490 | iter time: 359.31 ms (step) remaining time: 0:01:39
|
| 298 |
+
Epoch 1 | iter 6144 step 192 | loss train: 1.466, val: 1.490 | iter time: 359.06 ms (step) remaining time: 0:01:28
|
| 299 |
+
Epoch 1 | iter 6176 step 193 | loss train: 1.413, val: 1.490 | iter time: 359.53 ms (step) remaining time: 0:01:17
|
| 300 |
+
Epoch 1 | iter 6208 step 194 | loss train: 1.356, val: 1.490 | iter time: 361.61 ms (step) remaining time: 0:01:06
|
| 301 |
+
Epoch 1 | iter 6240 step 195 | loss train: 1.480, val: 1.490 | iter time: 361.66 ms (step) remaining time: 0:00:55
|
| 302 |
+
Epoch 1 | iter 6272 step 196 | loss train: 1.454, val: 1.490 | iter time: 361.34 ms (step) remaining time: 0:00:44
|
| 303 |
+
Epoch 1 | iter 6304 step 197 | loss train: 1.361, val: 1.490 | iter time: 358.70 ms (step) remaining time: 0:00:33
|
| 304 |
+
Epoch 1 | iter 6336 step 198 | loss train: 1.430, val: 1.490 | iter time: 360.10 ms (step) remaining time: 0:00:22
|
| 305 |
+
Epoch 1 | iter 6368 step 199 | loss train: 1.464, val: 1.490 | iter time: 359.90 ms (step) remaining time: 0:00:11
|
| 306 |
+
Epoch 1 | iter 6400 step 200 | loss train: 1.453, val: 1.490 | iter time: 359.55 ms (step) remaining time: 0:00:00
|
| 307 |
+
Validating ...
|
| 308 |
+
iter 6400: val loss 1.4696, val time: 6635.54 ms
|
| 309 |
+
Saving checkpoint to 'out/pretrain/tiny-llama-cl-2407/step-00000200/lit_model.pth'
|
| 310 |
+
Validating ...
|
| 311 |
+
Final evaluation | val loss: 1.470 | val ppl: 4.348
|
| 312 |
+
Saving checkpoint to 'out/pretrain/tiny-llama-cl-2407/final/lit_model.pth'
|
| 313 |
+
----------------------------------------
|
| 314 |
+
| Performance
|
| 315 |
+
| - Total tokens : 209,715,200
|
| 316 |
+
| - Training Time : 2269.73 s
|
| 317 |
+
| - Tok/sec : 214.47 tok/s
|
| 318 |
+
| ----------------------------------------
|
| 319 |
+
| Memory Usage
|
| 320 |
+
| - Memory Used : 26.32 GB
|
| 321 |
+
----------------------------------------
|
out/pretrain/tinyllama/teelogs/2407_lr4e-5.txt
ADDED
|
@@ -0,0 +1,334 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
Initializing distributed: GLOBAL_RANK: 0, MEMBER: 1/4
|
| 2 |
+
Initializing distributed: GLOBAL_RANK: 2, MEMBER: 3/4
|
| 3 |
+
Initializing distributed: GLOBAL_RANK: 1, MEMBER: 2/4
|
| 4 |
+
[rank: 2] Seed set to 42
|
| 5 |
+
[rank: 1] Seed set to 42
|
| 6 |
+
Initializing distributed: GLOBAL_RANK: 3, MEMBER: 4/4
|
| 7 |
+
[rank: 3] Seed set to 42
|
| 8 |
+
----------------------------------------------------------------------------------------------------
|
| 9 |
+
distributed_backend=nccl
|
| 10 |
+
All distributed processes registered. Starting with 4 processes
|
| 11 |
+
----------------------------------------------------------------------------------------------------
|
| 12 |
+
|
| 13 |
+
[rank: 0] Seed set to 42
|
| 14 |
+
/mnt/data/litgpt/litgpt/pretrain.py:515: UserWarning: A newer version of litdata is available (0.2.52). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 15 |
+
train_dataloader = data.train_dataloader()
|
| 16 |
+
/mnt/data/litgpt/litgpt/pretrain.py:515: UserWarning: A newer version of litdata is available (0.2.52). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 17 |
+
train_dataloader = data.train_dataloader()
|
| 18 |
+
/mnt/data/litgpt/litgpt/pretrain.py:515: UserWarning: A newer version of litdata is available (0.2.52). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 19 |
+
train_dataloader = data.train_dataloader()
|
| 20 |
+
/mnt/data/litgpt/litgpt/pretrain.py:515: UserWarning: A newer version of litdata is available (0.2.52). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 21 |
+
train_dataloader = data.train_dataloader()
|
| 22 |
+
All GPUs are fully connected via NVLink.
|
| 23 |
+
{'data': {'batch_size': 1,
|
| 24 |
+
'data_path': PosixPath('litgpt/data/arxiv'),
|
| 25 |
+
'num_workers': 0,
|
| 26 |
+
'seed': 42,
|
| 27 |
+
'seq_length': 2048,
|
| 28 |
+
'use_starcoder': True},
|
| 29 |
+
'data_dir': PosixPath('litgpt/data/arxiv/2407'),
|
| 30 |
+
'devices': 'auto',
|
| 31 |
+
'eval': {'evaluate_example': 'first',
|
| 32 |
+
'final_validation': True,
|
| 33 |
+
'initial_validation': True,
|
| 34 |
+
'interval': 50,
|
| 35 |
+
'max_iters': 200,
|
| 36 |
+
'max_new_tokens': None},
|
| 37 |
+
'initial_checkpoint_dir': PosixPath('checkpoints/TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T'),
|
| 38 |
+
'log': {'group': None, 'project': None, 'run': None},
|
| 39 |
+
'logger_name': 'tensorboard',
|
| 40 |
+
'model_config': {'attention_logit_softcapping': None,
|
| 41 |
+
'attention_scores_scalar': None,
|
| 42 |
+
'attn_bias': False,
|
| 43 |
+
'bias': False,
|
| 44 |
+
'block_size': 2048,
|
| 45 |
+
'final_logit_softcapping': None,
|
| 46 |
+
'gelu_approximate': 'none',
|
| 47 |
+
'head_size': 64,
|
| 48 |
+
'hf_config': {'name': 'TinyLlama-1.1B-intermediate-step-1431k-3T',
|
| 49 |
+
'org': 'TinyLlama'},
|
| 50 |
+
'intermediate_size': 5632,
|
| 51 |
+
'lm_head_bias': False,
|
| 52 |
+
'mlp_class_name': 'LLaMAMLP',
|
| 53 |
+
'moe_intermediate_size': None,
|
| 54 |
+
'n_embd': 2048,
|
| 55 |
+
'n_expert': 0,
|
| 56 |
+
'n_expert_per_token': 0,
|
| 57 |
+
'n_head': 32,
|
| 58 |
+
'n_layer': 22,
|
| 59 |
+
'n_query_groups': 4,
|
| 60 |
+
'name': 'tiny-llama-1.1b',
|
| 61 |
+
'norm_1': True,
|
| 62 |
+
'norm_2': True,
|
| 63 |
+
'norm_class_name': 'RMSNorm',
|
| 64 |
+
'norm_eps': 1e-05,
|
| 65 |
+
'norm_qk': False,
|
| 66 |
+
'norm_qk_type': 'default',
|
| 67 |
+
'padded_vocab_size': 32000,
|
| 68 |
+
'padding_multiple': 64,
|
| 69 |
+
'parallel_residual': False,
|
| 70 |
+
'post_attention_norm': False,
|
| 71 |
+
'post_mlp_norm': False,
|
| 72 |
+
'rope_adjustments': None,
|
| 73 |
+
'rope_base': 10000,
|
| 74 |
+
'rope_condense_ratio': 1,
|
| 75 |
+
'rope_indices': None,
|
| 76 |
+
'rope_local_base_freq': None,
|
| 77 |
+
'rotary_percentage': 1.0,
|
| 78 |
+
'scale_embeddings': False,
|
| 79 |
+
'shared_attention_norm': False,
|
| 80 |
+
'sliding_window_indices': None,
|
| 81 |
+
'sliding_window_size': None,
|
| 82 |
+
'vocab_size': 32000},
|
| 83 |
+
'model_name': 'tiny-llama-1.1b',
|
| 84 |
+
'num_nodes': 1,
|
| 85 |
+
'optimizer': "{'class_path': 'torch.optim.AdamW', 'init_args': {'lr': 4e-05, "
|
| 86 |
+
"'weight_decay': 0.1, 'betas': [0.9, 0.95]}}",
|
| 87 |
+
'out_dir': PosixPath('out/pretrain/tinyllama/2407_lr4e-5'),
|
| 88 |
+
'ppl': False,
|
| 89 |
+
'precision': 'bf16-mixed',
|
| 90 |
+
'resume': False,
|
| 91 |
+
'seed': 42,
|
| 92 |
+
'tokenizer_dir': PosixPath('checkpoints/TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T'),
|
| 93 |
+
'train': {'epochs': None,
|
| 94 |
+
'global_batch_size': 512,
|
| 95 |
+
'log_interval': 1,
|
| 96 |
+
'lr_warmup_fraction': None,
|
| 97 |
+
'lr_warmup_steps': 20,
|
| 98 |
+
'max_norm': 1.0,
|
| 99 |
+
'max_seq_length': 2048,
|
| 100 |
+
'max_steps': None,
|
| 101 |
+
'max_tokens': 211812352,
|
| 102 |
+
'micro_batch_size': 4,
|
| 103 |
+
'min_lr': 4e-05,
|
| 104 |
+
'save_interval': 100,
|
| 105 |
+
'tie_embeddings': None}}
|
| 106 |
+
Time to instantiate model: 0.04 seconds.
|
| 107 |
+
Total parameters: 1,100,048,384
|
| 108 |
+
[ok] checkpoints/TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T/lit_model.pth 已是纯权重
|
| 109 |
+
Validating ...
|
| 110 |
+
Measured TFLOPs: 239.66
|
| 111 |
+
Epoch 1 | iter 32 step 1 | loss train: 1.462, val: 1.508 | iter time: 537.76 ms (step) remaining time: 0:39:00
|
| 112 |
+
Epoch 1 | iter 64 step 2 | loss train: 1.481, val: 1.508 | iter time: 356.70 ms (step) remaining time: 0:37:19
|
| 113 |
+
Epoch 1 | iter 96 step 3 | loss train: 1.474, val: 1.508 | iter time: 358.11 ms (step) remaining time: 0:36:46
|
| 114 |
+
Epoch 1 | iter 128 step 4 | loss train: 1.453, val: 1.508 | iter time: 355.68 ms (step) remaining time: 0:36:20
|
| 115 |
+
Epoch 1 | iter 160 step 5 | loss train: 1.454, val: 1.508 | iter time: 358.16 ms (step) remaining time: 0:36:04
|
| 116 |
+
Epoch 1 | iter 192 step 6 | loss train: 1.477, val: 1.508 | iter time: 355.87 ms (step) remaining time: 0:35:48
|
| 117 |
+
Epoch 1 | iter 224 step 7 | loss train: 1.460, val: 1.508 | iter time: 357.58 ms (step) remaining time: 0:35:33
|
| 118 |
+
Epoch 1 | iter 256 step 8 | loss train: 1.446, val: 1.508 | iter time: 359.29 ms (step) remaining time: 0:35:19
|
| 119 |
+
Epoch 1 | iter 288 step 9 | loss train: 1.419, val: 1.508 | iter time: 357.58 ms (step) remaining time: 0:35:06
|
| 120 |
+
Epoch 1 | iter 320 step 10 | loss train: 1.465, val: 1.508 | iter time: 359.52 ms (step) remaining time: 0:34:54
|
| 121 |
+
Epoch 1 | iter 352 step 11 | loss train: 1.426, val: 1.508 | iter time: 359.00 ms (step) remaining time: 0:34:41
|
| 122 |
+
Epoch 1 | iter 384 step 12 | loss train: 1.438, val: 1.508 | iter time: 357.75 ms (step) remaining time: 0:34:29
|
| 123 |
+
Epoch 1 | iter 416 step 13 | loss train: 1.439, val: 1.508 | iter time: 358.76 ms (step) remaining time: 0:34:18
|
| 124 |
+
Epoch 1 | iter 448 step 14 | loss train: 1.486, val: 1.508 | iter time: 358.92 ms (step) remaining time: 0:34:06
|
| 125 |
+
Epoch 1 | iter 480 step 15 | loss train: 1.448, val: 1.508 | iter time: 359.39 ms (step) remaining time: 0:33:55
|
| 126 |
+
Epoch 1 | iter 512 step 16 | loss train: 1.503, val: 1.508 | iter time: 357.43 ms (step) remaining time: 0:33:44
|
| 127 |
+
Epoch 1 | iter 544 step 17 | loss train: 1.446, val: 1.508 | iter time: 358.08 ms (step) remaining time: 0:33:32
|
| 128 |
+
Epoch 1 | iter 576 step 18 | loss train: 1.427, val: 1.508 | iter time: 359.87 ms (step) remaining time: 0:33:21
|
| 129 |
+
Epoch 1 | iter 608 step 19 | loss train: 1.503, val: 1.508 | iter time: 360.56 ms (step) remaining time: 0:33:10
|
| 130 |
+
Epoch 1 | iter 640 step 20 | loss train: 1.482, val: 1.508 | iter time: 358.93 ms (step) remaining time: 0:32:59
|
| 131 |
+
Epoch 1 | iter 672 step 21 | loss train: 1.503, val: 1.508 | iter time: 357.52 ms (step) remaining time: 0:32:47
|
| 132 |
+
Epoch 1 | iter 704 step 22 | loss train: 1.463, val: 1.508 | iter time: 358.45 ms (step) remaining time: 0:32:36
|
| 133 |
+
Epoch 1 | iter 736 step 23 | loss train: 1.421, val: 1.508 | iter time: 360.40 ms (step) remaining time: 0:32:25
|
| 134 |
+
Epoch 1 | iter 768 step 24 | loss train: 1.413, val: 1.508 | iter time: 361.00 ms (step) remaining time: 0:32:14
|
| 135 |
+
Epoch 1 | iter 800 step 25 | loss train: 1.403, val: 1.508 | iter time: 357.88 ms (step) remaining time: 0:32:03
|
| 136 |
+
Epoch 1 | iter 832 step 26 | loss train: 1.445, val: 1.508 | iter time: 358.75 ms (step) remaining time: 0:31:52
|
| 137 |
+
Epoch 1 | iter 864 step 27 | loss train: 1.435, val: 1.508 | iter time: 358.12 ms (step) remaining time: 0:31:41
|
| 138 |
+
Epoch 1 | iter 896 step 28 | loss train: 1.460, val: 1.508 | iter time: 358.23 ms (step) remaining time: 0:31:30
|
| 139 |
+
Epoch 1 | iter 928 step 29 | loss train: 1.451, val: 1.508 | iter time: 361.24 ms (step) remaining time: 0:31:19
|
| 140 |
+
Epoch 1 | iter 960 step 30 | loss train: 1.380, val: 1.508 | iter time: 359.07 ms (step) remaining time: 0:31:08
|
| 141 |
+
Epoch 1 | iter 992 step 31 | loss train: 1.504, val: 1.508 | iter time: 358.70 ms (step) remaining time: 0:30:57
|
| 142 |
+
Epoch 1 | iter 1024 step 32 | loss train: 1.461, val: 1.508 | iter time: 359.33 ms (step) remaining time: 0:30:46
|
| 143 |
+
Epoch 1 | iter 1056 step 33 | loss train: 1.500, val: 1.508 | iter time: 357.92 ms (step) remaining time: 0:30:35
|
| 144 |
+
Epoch 1 | iter 1088 step 34 | loss train: 1.432, val: 1.508 | iter time: 358.42 ms (step) remaining time: 0:30:24
|
| 145 |
+
Epoch 1 | iter 1120 step 35 | loss train: 1.414, val: 1.508 | iter time: 360.10 ms (step) remaining time: 0:30:14
|
| 146 |
+
Epoch 1 | iter 1152 step 36 | loss train: 1.477, val: 1.508 | iter time: 358.07 ms (step) remaining time: 0:30:03
|
| 147 |
+
Epoch 1 | iter 1184 step 37 | loss train: 1.475, val: 1.508 | iter time: 360.08 ms (step) remaining time: 0:29:52
|
| 148 |
+
Epoch 1 | iter 1216 step 38 | loss train: 1.434, val: 1.508 | iter time: 359.17 ms (step) remaining time: 0:29:41
|
| 149 |
+
Epoch 1 | iter 1248 step 39 | loss train: 1.452, val: 1.508 | iter time: 358.45 ms (step) remaining time: 0:29:30
|
| 150 |
+
Epoch 1 | iter 1280 step 40 | loss train: 1.398, val: 1.508 | iter time: 358.07 ms (step) remaining time: 0:29:19
|
| 151 |
+
Epoch 1 | iter 1312 step 41 | loss train: 1.371, val: 1.508 | iter time: 358.66 ms (step) remaining time: 0:29:08
|
| 152 |
+
Epoch 1 | iter 1344 step 42 | loss train: 1.421, val: 1.508 | iter time: 359.26 ms (step) remaining time: 0:28:57
|
| 153 |
+
Epoch 1 | iter 1376 step 43 | loss train: 1.410, val: 1.508 | iter time: 358.69 ms (step) remaining time: 0:28:46
|
| 154 |
+
Epoch 1 | iter 1408 step 44 | loss train: 1.445, val: 1.508 | iter time: 358.95 ms (step) remaining time: 0:28:35
|
| 155 |
+
Epoch 1 | iter 1440 step 45 | loss train: 1.412, val: 1.508 | iter time: 357.91 ms (step) remaining time: 0:28:24
|
| 156 |
+
Epoch 1 | iter 1472 step 46 | loss train: 1.496, val: 1.508 | iter time: 357.62 ms (step) remaining time: 0:28:14
|
| 157 |
+
Epoch 1 | iter 1504 step 47 | loss train: 1.396, val: 1.508 | iter time: 360.43 ms (step) remaining time: 0:28:03
|
| 158 |
+
Epoch 1 | iter 1536 step 48 | loss train: 1.528, val: 1.508 | iter time: 357.49 ms (step) remaining time: 0:27:52
|
| 159 |
+
Epoch 1 | iter 1568 step 49 | loss train: 1.403, val: 1.508 | iter time: 360.38 ms (step) remaining time: 0:27:42
|
| 160 |
+
Epoch 1 | iter 1600 step 50 | loss train: 1.436, val: 1.508 | iter time: 359.54 ms (step) remaining time: 0:27:31
|
| 161 |
+
Validating ...
|
| 162 |
+
iter 1600: val loss 1.4361, val time: 21889.40 ms
|
| 163 |
+
Epoch 1 | iter 1632 step 51 | loss train: 1.399, val: 1.436 | iter time: 359.17 ms (step) remaining time: 0:28:25
|
| 164 |
+
Epoch 1 | iter 1664 step 52 | loss train: 1.365, val: 1.436 | iter time: 357.95 ms (step) remaining time: 0:28:12
|
| 165 |
+
Epoch 1 | iter 1696 step 53 | loss train: 1.386, val: 1.436 | iter time: 358.82 ms (step) remaining time: 0:27:59
|
| 166 |
+
Epoch 1 | iter 1728 step 54 | loss train: 1.465, val: 1.436 | iter time: 358.77 ms (step) remaining time: 0:27:47
|
| 167 |
+
Epoch 1 | iter 1760 step 55 | loss train: 1.426, val: 1.436 | iter time: 743.32 ms (step) remaining time: 0:27:36
|
| 168 |
+
Epoch 1 | iter 1792 step 56 | loss train: 1.446, val: 1.436 | iter time: 359.90 ms (step) remaining time: 0:27:23
|
| 169 |
+
Epoch 1 | iter 1824 step 57 | loss train: 1.410, val: 1.436 | iter time: 358.13 ms (step) remaining time: 0:27:11
|
| 170 |
+
Epoch 1 | iter 1856 step 58 | loss train: 1.328, val: 1.436 | iter time: 358.15 ms (step) remaining time: 0:26:59
|
| 171 |
+
Epoch 1 | iter 1888 step 59 | loss train: 1.427, val: 1.436 | iter time: 361.03 ms (step) remaining time: 0:26:47
|
| 172 |
+
Epoch 1 | iter 1920 step 60 | loss train: 1.401, val: 1.436 | iter time: 358.23 ms (step) remaining time: 0:26:34
|
| 173 |
+
Epoch 1 | iter 1952 step 61 | loss train: 1.462, val: 1.436 | iter time: 360.43 ms (step) remaining time: 0:26:22
|
| 174 |
+
Epoch 1 | iter 1984 step 62 | loss train: 1.468, val: 1.436 | iter time: 359.00 ms (step) remaining time: 0:26:10
|
| 175 |
+
Epoch 1 | iter 2016 step 63 | loss train: 1.465, val: 1.436 | iter time: 359.68 ms (step) remaining time: 0:25:58
|
| 176 |
+
Epoch 1 | iter 2048 step 64 | loss train: 1.435, val: 1.436 | iter time: 358.28 ms (step) remaining time: 0:25:46
|
| 177 |
+
Epoch 1 | iter 2080 step 65 | loss train: 1.422, val: 1.436 | iter time: 358.90 ms (step) remaining time: 0:25:34
|
| 178 |
+
Epoch 1 | iter 2112 step 66 | loss train: 1.490, val: 1.436 | iter time: 360.03 ms (step) remaining time: 0:25:22
|
| 179 |
+
Epoch 1 | iter 2144 step 67 | loss train: 1.377, val: 1.436 | iter time: 358.40 ms (step) remaining time: 0:25:10
|
| 180 |
+
Epoch 1 | iter 2176 step 68 | loss train: 1.425, val: 1.436 | iter time: 359.52 ms (step) remaining time: 0:24:59
|
| 181 |
+
Epoch 1 | iter 2208 step 69 | loss train: 1.444, val: 1.436 | iter time: 359.57 ms (step) remaining time: 0:24:47
|
| 182 |
+
Epoch 1 | iter 2240 step 70 | loss train: 1.378, val: 1.436 | iter time: 360.12 ms (step) remaining time: 0:24:35
|
| 183 |
+
Epoch 1 | iter 2272 step 71 | loss train: 1.439, val: 1.436 | iter time: 358.92 ms (step) remaining time: 0:24:23
|
| 184 |
+
Epoch 1 | iter 2304 step 72 | loss train: 1.468, val: 1.436 | iter time: 359.64 ms (step) remaining time: 0:24:11
|
| 185 |
+
Epoch 1 | iter 2336 step 73 | loss train: 1.421, val: 1.436 | iter time: 358.99 ms (step) remaining time: 0:24:00
|
| 186 |
+
Epoch 1 | iter 2368 step 74 | loss train: 1.446, val: 1.436 | iter time: 359.12 ms (step) remaining time: 0:23:48
|
| 187 |
+
Epoch 1 | iter 2400 step 75 | loss train: 1.320, val: 1.436 | iter time: 359.87 ms (step) remaining time: 0:23:36
|
| 188 |
+
Epoch 1 | iter 2432 step 76 | loss train: 1.437, val: 1.436 | iter time: 358.82 ms (step) remaining time: 0:23:25
|
| 189 |
+
Epoch 1 | iter 2464 step 77 | loss train: 1.389, val: 1.436 | iter time: 360.37 ms (step) remaining time: 0:23:13
|
| 190 |
+
Epoch 1 | iter 2496 step 78 | loss train: 1.423, val: 1.436 | iter time: 357.35 ms (step) remaining time: 0:23:01
|
| 191 |
+
Epoch 1 | iter 2528 step 79 | loss train: 1.434, val: 1.436 | iter time: 358.72 ms (step) remaining time: 0:22:50
|
| 192 |
+
Epoch 1 | iter 2560 step 80 | loss train: 1.422, val: 1.436 | iter time: 358.92 ms (step) remaining time: 0:22:38
|
| 193 |
+
Epoch 1 | iter 2592 step 81 | loss train: 1.413, val: 1.436 | iter time: 359.58 ms (step) remaining time: 0:22:26
|
| 194 |
+
Epoch 1 | iter 2624 step 82 | loss train: 1.347, val: 1.436 | iter time: 360.71 ms (step) remaining time: 0:22:15
|
| 195 |
+
Epoch 1 | iter 2656 step 83 | loss train: 1.413, val: 1.436 | iter time: 360.00 ms (step) remaining time: 0:22:03
|
| 196 |
+
Epoch 1 | iter 2688 step 84 | loss train: 1.431, val: 1.436 | iter time: 360.67 ms (step) remaining time: 0:21:52
|
| 197 |
+
Epoch 1 | iter 2720 step 85 | loss train: 1.463, val: 1.436 | iter time: 359.26 ms (step) remaining time: 0:21:41
|
| 198 |
+
Epoch 1 | iter 2752 step 86 | loss train: 1.453, val: 1.436 | iter time: 358.85 ms (step) remaining time: 0:21:29
|
| 199 |
+
Epoch 1 | iter 2784 step 87 | loss train: 1.369, val: 1.436 | iter time: 358.83 ms (step) remaining time: 0:21:18
|
| 200 |
+
Epoch 1 | iter 2816 step 88 | loss train: 1.450, val: 1.436 | iter time: 360.72 ms (step) remaining time: 0:21:06
|
| 201 |
+
Epoch 1 | iter 2848 step 89 | loss train: 1.467, val: 1.436 | iter time: 360.84 ms (step) remaining time: 0:20:55
|
| 202 |
+
Epoch 1 | iter 2880 step 90 | loss train: 1.429, val: 1.436 | iter time: 360.97 ms (step) remaining time: 0:20:43
|
| 203 |
+
Epoch 1 | iter 2912 step 91 | loss train: 1.427, val: 1.436 | iter time: 360.29 ms (step) remaining time: 0:20:32
|
| 204 |
+
Epoch 1 | iter 2944 step 92 | loss train: 1.341, val: 1.436 | iter time: 359.94 ms (step) remaining time: 0:20:21
|
| 205 |
+
Epoch 1 | iter 2976 step 93 | loss train: 1.379, val: 1.436 | iter time: 358.34 ms (step) remaining time: 0:20:09
|
| 206 |
+
Epoch 1 | iter 3008 step 94 | loss train: 1.442, val: 1.436 | iter time: 359.57 ms (step) remaining time: 0:19:58
|
| 207 |
+
Epoch 1 | iter 3040 step 95 | loss train: 1.353, val: 1.436 | iter time: 359.47 ms (step) remaining time: 0:19:47
|
| 208 |
+
Epoch 1 | iter 3072 step 96 | loss train: 1.358, val: 1.436 | iter time: 359.14 ms (step) remaining time: 0:19:35
|
| 209 |
+
Epoch 1 | iter 3104 step 97 | loss train: 1.383, val: 1.436 | iter time: 359.20 ms (step) remaining time: 0:19:24
|
| 210 |
+
Epoch 1 | iter 3136 step 98 | loss train: 1.347, val: 1.436 | iter time: 358.01 ms (step) remaining time: 0:19:13
|
| 211 |
+
Epoch 1 | iter 3168 step 99 | loss train: 1.429, val: 1.436 | iter time: 355.51 ms (step) remaining time: 0:19:01
|
| 212 |
+
Epoch 1 | iter 3200 step 100 | loss train: 1.425, val: 1.436 | iter time: 358.42 ms (step) remaining time: 0:18:50
|
| 213 |
+
Validating ...
|
| 214 |
+
iter 3200: val loss 1.3808, val time: 21890.43 ms
|
| 215 |
+
Saving checkpoint to 'out/pretrain/tinyllama/2407_lr4e-5/step-00000100/lit_model.pth'
|
| 216 |
+
Epoch 1 | iter 3232 step 101 | loss train: 1.519, val: 1.381 | iter time: 356.30 ms (step) remaining time: 0:19:17
|
| 217 |
+
Epoch 1 | iter 3264 step 102 | loss train: 1.437, val: 1.381 | iter time: 354.98 ms (step) remaining time: 0:19:05
|
| 218 |
+
Epoch 1 | iter 3296 step 103 | loss train: 1.498, val: 1.381 | iter time: 356.48 ms (step) remaining time: 0:18:53
|
| 219 |
+
Epoch 1 | iter 3328 step 104 | loss train: 1.403, val: 1.381 | iter time: 358.68 ms (step) remaining time: 0:18:41
|
| 220 |
+
Epoch 1 | iter 3360 step 105 | loss train: 1.367, val: 1.381 | iter time: 357.85 ms (step) remaining time: 0:18:29
|
| 221 |
+
Epoch 1 | iter 3392 step 106 | loss train: 1.405, val: 1.381 | iter time: 359.50 ms (step) remaining time: 0:18:17
|
| 222 |
+
Epoch 1 | iter 3424 step 107 | loss train: 1.477, val: 1.381 | iter time: 358.88 ms (step) remaining time: 0:18:05
|
| 223 |
+
Epoch 1 | iter 3456 step 108 | loss train: 1.476, val: 1.381 | iter time: 359.27 ms (step) remaining time: 0:17:53
|
| 224 |
+
Epoch 1 | iter 3488 step 109 | loss train: 1.497, val: 1.381 | iter time: 357.47 ms (step) remaining time: 0:17:41
|
| 225 |
+
Epoch 1 | iter 3520 step 110 | loss train: 1.401, val: 1.381 | iter time: 359.16 ms (step) remaining time: 0:17:29
|
| 226 |
+
Epoch 1 | iter 3552 step 111 | loss train: 1.444, val: 1.381 | iter time: 358.70 ms (step) remaining time: 0:17:17
|
| 227 |
+
Epoch 1 | iter 3584 step 112 | loss train: 1.447, val: 1.381 | iter time: 358.16 ms (step) remaining time: 0:17:05
|
| 228 |
+
Epoch 1 | iter 3616 step 113 | loss train: 1.399, val: 1.381 | iter time: 358.70 ms (step) remaining time: 0:16:54
|
| 229 |
+
Epoch 1 | iter 3648 step 114 | loss train: 1.456, val: 1.381 | iter time: 359.15 ms (step) remaining time: 0:16:42
|
| 230 |
+
Epoch 1 | iter 3680 step 115 | loss train: 1.437, val: 1.381 | iter time: 358.86 ms (step) remaining time: 0:16:30
|
| 231 |
+
Epoch 1 | iter 3712 step 116 | loss train: 1.467, val: 1.381 | iter time: 358.82 ms (step) remaining time: 0:16:18
|
| 232 |
+
Epoch 1 | iter 3744 step 117 | loss train: 1.431, val: 1.381 | iter time: 358.59 ms (step) remaining time: 0:16:06
|
| 233 |
+
Epoch 1 | iter 3776 step 118 | loss train: 1.429, val: 1.381 | iter time: 360.06 ms (step) remaining time: 0:15:55
|
| 234 |
+
Epoch 1 | iter 3808 step 119 | loss train: 1.382, val: 1.381 | iter time: 359.83 ms (step) remaining time: 0:15:43
|
| 235 |
+
Epoch 1 | iter 3840 step 120 | loss train: 1.425, val: 1.381 | iter time: 356.77 ms (step) remaining time: 0:15:31
|
| 236 |
+
Epoch 1 | iter 3872 step 121 | loss train: 1.457, val: 1.381 | iter time: 359.41 ms (step) remaining time: 0:15:20
|
| 237 |
+
Epoch 1 | iter 3904 step 122 | loss train: 1.426, val: 1.381 | iter time: 358.54 ms (step) remaining time: 0:15:08
|
| 238 |
+
Epoch 1 | iter 3936 step 123 | loss train: 1.414, val: 1.381 | iter time: 359.50 ms (step) remaining time: 0:14:56
|
| 239 |
+
Epoch 1 | iter 3968 step 124 | loss train: 1.405, val: 1.381 | iter time: 361.57 ms (step) remaining time: 0:14:45
|
| 240 |
+
Epoch 1 | iter 4000 step 125 | loss train: 1.412, val: 1.381 | iter time: 361.58 ms (step) remaining time: 0:14:33
|
| 241 |
+
Epoch 1 | iter 4032 step 126 | loss train: 1.390, val: 1.381 | iter time: 360.64 ms (step) remaining time: 0:14:22
|
| 242 |
+
Epoch 1 | iter 4064 step 127 | loss train: 1.404, val: 1.381 | iter time: 360.52 ms (step) remaining time: 0:14:10
|
| 243 |
+
Epoch 1 | iter 4096 step 128 | loss train: 1.421, val: 1.381 | iter time: 362.00 ms (step) remaining time: 0:13:58
|
| 244 |
+
Epoch 1 | iter 4128 step 129 | loss train: 1.398, val: 1.381 | iter time: 359.93 ms (step) remaining time: 0:13:47
|
| 245 |
+
Epoch 1 | iter 4160 step 130 | loss train: 1.489, val: 1.381 | iter time: 360.07 ms (step) remaining time: 0:13:35
|
| 246 |
+
Epoch 1 | iter 4192 step 131 | loss train: 1.397, val: 1.381 | iter time: 359.08 ms (step) remaining time: 0:13:24
|
| 247 |
+
Epoch 1 | iter 4224 step 132 | loss train: 1.450, val: 1.381 | iter time: 361.62 ms (step) remaining time: 0:13:12
|
| 248 |
+
Epoch 1 | iter 4256 step 133 | loss train: 1.457, val: 1.381 | iter time: 359.58 ms (step) remaining time: 0:13:01
|
| 249 |
+
Epoch 1 | iter 4288 step 134 | loss train: 1.382, val: 1.381 | iter time: 359.90 ms (step) remaining time: 0:12:49
|
| 250 |
+
Epoch 1 | iter 4320 step 135 | loss train: 1.433, val: 1.381 | iter time: 360.97 ms (step) remaining time: 0:12:38
|
| 251 |
+
Epoch 1 | iter 4352 step 136 | loss train: 1.434, val: 1.381 | iter time: 359.51 ms (step) remaining time: 0:12:26
|
| 252 |
+
Epoch 1 | iter 4384 step 137 | loss train: 1.520, val: 1.381 | iter time: 360.19 ms (step) remaining time: 0:12:15
|
| 253 |
+
Epoch 1 | iter 4416 step 138 | loss train: 1.460, val: 1.381 | iter time: 358.76 ms (step) remaining time: 0:12:03
|
| 254 |
+
Epoch 1 | iter 4448 step 139 | loss train: 1.441, val: 1.381 | iter time: 359.84 ms (step) remaining time: 0:11:52
|
| 255 |
+
Epoch 1 | iter 4480 step 140 | loss train: 1.412, val: 1.381 | iter time: 359.28 ms (step) remaining time: 0:11:40
|
| 256 |
+
Epoch 1 | iter 4512 step 141 | loss train: 1.409, val: 1.381 | iter time: 359.51 ms (step) remaining time: 0:11:29
|
| 257 |
+
Epoch 1 | iter 4544 step 142 | loss train: 1.470, val: 1.381 | iter time: 360.79 ms (step) remaining time: 0:11:17
|
| 258 |
+
Epoch 1 | iter 4576 step 143 | loss train: 1.502, val: 1.381 | iter time: 358.40 ms (step) remaining time: 0:11:06
|
| 259 |
+
Epoch 1 | iter 4608 step 144 | loss train: 1.418, val: 1.381 | iter time: 359.06 ms (step) remaining time: 0:10:54
|
| 260 |
+
Epoch 1 | iter 4640 step 145 | loss train: 1.382, val: 1.381 | iter time: 358.36 ms (step) remaining time: 0:10:43
|
| 261 |
+
Epoch 1 | iter 4672 step 146 | loss train: 1.505, val: 1.381 | iter time: 358.19 ms (step) remaining time: 0:10:31
|
| 262 |
+
Epoch 1 | iter 4704 step 147 | loss train: 1.473, val: 1.381 | iter time: 359.93 ms (step) remaining time: 0:10:20
|
| 263 |
+
Epoch 1 | iter 4736 step 148 | loss train: 1.456, val: 1.381 | iter time: 358.15 ms (step) remaining time: 0:10:08
|
| 264 |
+
Epoch 1 | iter 4768 step 149 | loss train: 1.362, val: 1.381 | iter time: 357.96 ms (step) remaining time: 0:09:57
|
| 265 |
+
Epoch 1 | iter 4800 step 150 | loss train: 1.491, val: 1.381 | iter time: 357.98 ms (step) remaining time: 0:09:45
|
| 266 |
+
Validating ...
|
| 267 |
+
iter 4800: val loss 1.3862, val time: 21884.89 ms
|
| 268 |
+
Epoch 1 | iter 4832 step 151 | loss train: 1.477, val: 1.386 | iter time: 361.41 ms (step) remaining time: 0:09:42
|
| 269 |
+
Epoch 1 | iter 4864 step 152 | loss train: 1.443, val: 1.386 | iter time: 359.29 ms (step) remaining time: 0:09:30
|
| 270 |
+
Epoch 1 | iter 4896 step 153 | loss train: 1.419, val: 1.386 | iter time: 360.99 ms (step) remaining time: 0:09:18
|
| 271 |
+
Epoch 1 | iter 4928 step 154 | loss train: 1.400, val: 1.386 | iter time: 359.78 ms (step) remaining time: 0:09:07
|
| 272 |
+
Epoch 1 | iter 4960 step 155 | loss train: 1.409, val: 1.386 | iter time: 358.89 ms (step) remaining time: 0:08:55
|
| 273 |
+
Epoch 1 | iter 4992 step 156 | loss train: 1.495, val: 1.386 | iter time: 359.49 ms (step) remaining time: 0:08:44
|
| 274 |
+
Epoch 1 | iter 5024 step 157 | loss train: 1.392, val: 1.386 | iter time: 360.13 ms (step) remaining time: 0:08:32
|
| 275 |
+
Epoch 1 | iter 5056 step 158 | loss train: 1.416, val: 1.386 | iter time: 360.78 ms (step) remaining time: 0:08:21
|
| 276 |
+
Epoch 1 | iter 5088 step 159 | loss train: 1.372, val: 1.386 | iter time: 359.35 ms (step) remaining time: 0:08:09
|
| 277 |
+
Epoch 1 | iter 5120 step 160 | loss train: 1.405, val: 1.386 | iter time: 360.12 ms (step) remaining time: 0:07:58
|
| 278 |
+
Epoch 1 | iter 5152 step 161 | loss train: 1.435, val: 1.386 | iter time: 357.74 ms (step) remaining time: 0:07:46
|
| 279 |
+
Epoch 1 | iter 5184 step 162 | loss train: 1.424, val: 1.386 | iter time: 360.51 ms (step) remaining time: 0:07:34
|
| 280 |
+
Epoch 1 | iter 5216 step 163 | loss train: 1.463, val: 1.386 | iter time: 357.19 ms (step) remaining time: 0:07:23
|
| 281 |
+
Epoch 1 | iter 5248 step 164 | loss train: 1.399, val: 1.386 | iter time: 359.01 ms (step) remaining time: 0:07:12
|
| 282 |
+
Epoch 1 | iter 5280 step 165 | loss train: 1.414, val: 1.386 | iter time: 359.56 ms (step) remaining time: 0:07:00
|
| 283 |
+
Epoch 1 | iter 5312 step 166 | loss train: 1.443, val: 1.386 | iter time: 358.55 ms (step) remaining time: 0:06:49
|
| 284 |
+
Epoch 1 | iter 5344 step 167 | loss train: 1.411, val: 1.386 | iter time: 359.22 ms (step) remaining time: 0:06:37
|
| 285 |
+
Epoch 1 | iter 5376 step 168 | loss train: 1.386, val: 1.386 | iter time: 359.22 ms (step) remaining time: 0:06:26
|
| 286 |
+
Epoch 1 | iter 5408 step 169 | loss train: 1.408, val: 1.386 | iter time: 359.34 ms (step) remaining time: 0:06:14
|
| 287 |
+
Epoch 1 | iter 5440 step 170 | loss train: 1.394, val: 1.386 | iter time: 359.45 ms (step) remaining time: 0:06:03
|
| 288 |
+
Epoch 1 | iter 5472 step 171 | loss train: 1.445, val: 1.386 | iter time: 360.94 ms (step) remaining time: 0:05:51
|
| 289 |
+
Epoch 1 | iter 5504 step 172 | loss train: 1.413, val: 1.386 | iter time: 364.07 ms (step) remaining time: 0:05:40
|
| 290 |
+
Epoch 1 | iter 5536 step 173 | loss train: 1.430, val: 1.386 | iter time: 360.39 ms (step) remaining time: 0:05:28
|
| 291 |
+
Epoch 1 | iter 5568 step 174 | loss train: 1.449, val: 1.386 | iter time: 358.68 ms (step) remaining time: 0:05:17
|
| 292 |
+
Epoch 1 | iter 5600 step 175 | loss train: 1.434, val: 1.386 | iter time: 358.88 ms (step) remaining time: 0:05:06
|
| 293 |
+
Epoch 1 | iter 5632 step 176 | loss train: 1.372, val: 1.386 | iter time: 358.92 ms (step) remaining time: 0:04:54
|
| 294 |
+
Epoch 1 | iter 5664 step 177 | loss train: 1.382, val: 1.386 | iter time: 358.70 ms (step) remaining time: 0:04:43
|
| 295 |
+
Epoch 1 | iter 5696 step 178 | loss train: 1.395, val: 1.386 | iter time: 359.08 ms (step) remaining time: 0:04:31
|
| 296 |
+
Epoch 1 | iter 5728 step 179 | loss train: 1.432, val: 1.386 | iter time: 360.66 ms (step) remaining time: 0:04:20
|
| 297 |
+
Epoch 1 | iter 5760 step 180 | loss train: 1.405, val: 1.386 | iter time: 357.47 ms (step) remaining time: 0:04:09
|
| 298 |
+
Epoch 1 | iter 5792 step 181 | loss train: 1.406, val: 1.386 | iter time: 359.14 ms (step) remaining time: 0:03:57
|
| 299 |
+
Epoch 1 | iter 5824 step 182 | loss train: 1.435, val: 1.386 | iter time: 360.04 ms (step) remaining time: 0:03:46
|
| 300 |
+
Epoch 1 | iter 5856 step 183 | loss train: 1.459, val: 1.386 | iter time: 359.13 ms (step) remaining time: 0:03:35
|
| 301 |
+
Epoch 1 | iter 5888 step 184 | loss train: 1.389, val: 1.386 | iter time: 358.51 ms (step) remaining time: 0:03:23
|
| 302 |
+
Epoch 1 | iter 5920 step 185 | loss train: 1.427, val: 1.386 | iter time: 359.62 ms (step) remaining time: 0:03:12
|
| 303 |
+
Epoch 1 | iter 5952 step 186 | loss train: 1.407, val: 1.386 | iter time: 358.62 ms (step) remaining time: 0:03:00
|
| 304 |
+
Epoch 1 | iter 5984 step 187 | loss train: 1.366, val: 1.386 | iter time: 359.50 ms (step) remaining time: 0:02:49
|
| 305 |
+
Epoch 1 | iter 6016 step 188 | loss train: 1.361, val: 1.386 | iter time: 358.08 ms (step) remaining time: 0:02:38
|
| 306 |
+
Epoch 1 | iter 6048 step 189 | loss train: 1.419, val: 1.386 | iter time: 359.68 ms (step) remaining time: 0:02:26
|
| 307 |
+
Epoch 1 | iter 6080 step 190 | loss train: 1.437, val: 1.386 | iter time: 359.31 ms (step) remaining time: 0:02:15
|
| 308 |
+
Epoch 1 | iter 6112 step 191 | loss train: 1.411, val: 1.386 | iter time: 359.74 ms (step) remaining time: 0:02:04
|
| 309 |
+
Epoch 1 | iter 6144 step 192 | loss train: 1.377, val: 1.386 | iter time: 358.37 ms (step) remaining time: 0:01:52
|
| 310 |
+
Epoch 1 | iter 6176 step 193 | loss train: 1.469, val: 1.386 | iter time: 359.70 ms (step) remaining time: 0:01:41
|
| 311 |
+
Epoch 1 | iter 6208 step 194 | loss train: 1.386, val: 1.386 | iter time: 588.60 ms (step) remaining time: 0:01:30
|
| 312 |
+
Epoch 1 | iter 6240 step 195 | loss train: 1.365, val: 1.386 | iter time: 358.18 ms (step) remaining time: 0:01:19
|
| 313 |
+
Epoch 1 | iter 6272 step 196 | loss train: 1.379, val: 1.386 | iter time: 358.97 ms (step) remaining time: 0:01:07
|
| 314 |
+
Epoch 1 | iter 6304 step 197 | loss train: 1.389, val: 1.386 | iter time: 358.96 ms (step) remaining time: 0:00:56
|
| 315 |
+
Epoch 1 | iter 6336 step 198 | loss train: 1.476, val: 1.386 | iter time: 359.36 ms (step) remaining time: 0:00:45
|
| 316 |
+
Epoch 1 | iter 6368 step 199 | loss train: 1.538, val: 1.386 | iter time: 359.78 ms (step) remaining time: 0:00:33
|
| 317 |
+
Epoch 1 | iter 6400 step 200 | loss train: 1.453, val: 1.386 | iter time: 356.84 ms (step) remaining time: 0:00:22
|
| 318 |
+
Validating ...
|
| 319 |
+
iter 6400: val loss 1.3940, val time: 21890.79 ms
|
| 320 |
+
Saving checkpoint to 'out/pretrain/tinyllama/2407_lr4e-5/step-00000200/lit_model.pth'
|
| 321 |
+
Epoch 1 | iter 6432 step 201 | loss train: 1.508, val: 1.394 | iter time: 356.12 ms (step) remaining time: 0:00:11
|
| 322 |
+
Epoch 2 | iter 6464 step 202 | loss train: 1.352, val: 1.394 | iter time: 357.03 ms (step) remaining time: 0:00:00
|
| 323 |
+
Validating ...
|
| 324 |
+
Final evaluation | val loss: 1.393 | val ppl: 4.026
|
| 325 |
+
Saving checkpoint to 'out/pretrain/tinyllama/2407_lr4e-5/final/lit_model.pth'
|
| 326 |
+
----------------------------------------
|
| 327 |
+
| Performance
|
| 328 |
+
| - Total tokens : 211,812,352
|
| 329 |
+
| - Training Time : 2377.70 s
|
| 330 |
+
| - Tok/sec : 109.67 tok/s
|
| 331 |
+
| ----------------------------------------
|
| 332 |
+
| Memory Usage
|
| 333 |
+
| - Memory Used : 26.32 GB
|
| 334 |
+
----------------------------------------
|
out/pretrain/tinyllama/teelogs/2408.txt
ADDED
|
@@ -0,0 +1,288 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
Initializing distributed: GLOBAL_RANK: 0, MEMBER: 1/4
|
| 2 |
+
Initializing distributed: GLOBAL_RANK: 3, MEMBER: 4/4
|
| 3 |
+
[rank: 3] Seed set to 42
|
| 4 |
+
Initializing distributed: GLOBAL_RANK: 1, MEMBER: 2/4
|
| 5 |
+
[rank: 1] Seed set to 42
|
| 6 |
+
Initializing distributed: GLOBAL_RANK: 2, MEMBER: 3/4
|
| 7 |
+
[rank: 2] Seed set to 42
|
| 8 |
+
----------------------------------------------------------------------------------------------------
|
| 9 |
+
distributed_backend=nccl
|
| 10 |
+
All distributed processes registered. Starting with 4 processes
|
| 11 |
+
----------------------------------------------------------------------------------------------------
|
| 12 |
+
|
| 13 |
+
[rank: 0] Seed set to 42
|
| 14 |
+
All GPUs are fully connected via NVLink.
|
| 15 |
+
{'data': {'batch_size': 1,
|
| 16 |
+
'data_path': PosixPath('litgpt/data/arxiv'),
|
| 17 |
+
'num_workers': 8,
|
| 18 |
+
'seed': 42,
|
| 19 |
+
'seq_length': 2048,
|
| 20 |
+
'use_starcoder': True},
|
| 21 |
+
'data_dir': PosixPath('litgpt/data/arxiv/2408'),
|
| 22 |
+
'devices': 'auto',
|
| 23 |
+
'eval': {'evaluate_example': 'first',
|
| 24 |
+
'final_validation': True,
|
| 25 |
+
'initial_validation': True,
|
| 26 |
+
'interval': 50,
|
| 27 |
+
'max_iters': 100,
|
| 28 |
+
'max_new_tokens': None},
|
| 29 |
+
'initial_checkpoint_dir': PosixPath('out/pretrain/2407/final'),
|
| 30 |
+
'log': {'group': None, 'project': None, 'run': None},
|
| 31 |
+
'logger_name': 'tensorboard',
|
| 32 |
+
'model_config': {'attention_logit_softcapping': None,
|
| 33 |
+
'attention_scores_scalar': None,
|
| 34 |
+
'attn_bias': False,
|
| 35 |
+
'bias': False,
|
| 36 |
+
'block_size': 2048,
|
| 37 |
+
'final_logit_softcapping': None,
|
| 38 |
+
'gelu_approximate': 'none',
|
| 39 |
+
'head_size': 64,
|
| 40 |
+
'hf_config': {'name': 'TinyLlama-1.1B-intermediate-step-1431k-3T',
|
| 41 |
+
'org': 'TinyLlama'},
|
| 42 |
+
'intermediate_size': 5632,
|
| 43 |
+
'lm_head_bias': False,
|
| 44 |
+
'mlp_class_name': 'LLaMAMLP',
|
| 45 |
+
'moe_intermediate_size': None,
|
| 46 |
+
'n_embd': 2048,
|
| 47 |
+
'n_expert': 0,
|
| 48 |
+
'n_expert_per_token': 0,
|
| 49 |
+
'n_head': 32,
|
| 50 |
+
'n_layer': 22,
|
| 51 |
+
'n_query_groups': 4,
|
| 52 |
+
'name': 'tiny-llama-1.1b',
|
| 53 |
+
'norm_1': True,
|
| 54 |
+
'norm_2': True,
|
| 55 |
+
'norm_class_name': 'RMSNorm',
|
| 56 |
+
'norm_eps': 1e-05,
|
| 57 |
+
'norm_qk': False,
|
| 58 |
+
'norm_qk_type': 'default',
|
| 59 |
+
'padded_vocab_size': 32000,
|
| 60 |
+
'padding_multiple': 64,
|
| 61 |
+
'parallel_residual': False,
|
| 62 |
+
'post_attention_norm': False,
|
| 63 |
+
'post_mlp_norm': False,
|
| 64 |
+
'rope_adjustments': None,
|
| 65 |
+
'rope_base': 10000,
|
| 66 |
+
'rope_condense_ratio': 1,
|
| 67 |
+
'rope_indices': None,
|
| 68 |
+
'rope_local_base_freq': None,
|
| 69 |
+
'rotary_percentage': 1.0,
|
| 70 |
+
'scale_embeddings': False,
|
| 71 |
+
'shared_attention_norm': False,
|
| 72 |
+
'sliding_window_indices': None,
|
| 73 |
+
'sliding_window_size': None,
|
| 74 |
+
'vocab_size': 32000},
|
| 75 |
+
'model_name': 'tiny-llama-1.1b',
|
| 76 |
+
'num_nodes': 1,
|
| 77 |
+
'optimizer': "{'class_path': 'torch.optim.AdamW', 'init_args': {'lr': 0.0004, "
|
| 78 |
+
"'weight_decay': 0.1, 'betas': [0.9, 0.95]}}",
|
| 79 |
+
'out_dir': PosixPath('out/pretrain/2408'),
|
| 80 |
+
'precision': 'bf16-mixed',
|
| 81 |
+
'resume': False,
|
| 82 |
+
'seed': 42,
|
| 83 |
+
'tokenizer_dir': PosixPath('checkpoints/TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T'),
|
| 84 |
+
'train': {'epochs': None,
|
| 85 |
+
'global_batch_size': 512,
|
| 86 |
+
'log_interval': 1,
|
| 87 |
+
'lr_warmup_fraction': None,
|
| 88 |
+
'lr_warmup_steps': 20,
|
| 89 |
+
'max_norm': 1.0,
|
| 90 |
+
'max_seq_length': 2048,
|
| 91 |
+
'max_steps': None,
|
| 92 |
+
'max_tokens': 176160768,
|
| 93 |
+
'micro_batch_size': 4,
|
| 94 |
+
'min_lr': 4e-05,
|
| 95 |
+
'save_interval': 100,
|
| 96 |
+
'tie_embeddings': None}}
|
| 97 |
+
Time to instantiate model: 0.02 seconds.
|
| 98 |
+
Total parameters: 1,100,048,384
|
| 99 |
+
[ok] out/pretrain/2407/final/lit_model.pth 已是纯权重
|
| 100 |
+
Validating ...
|
| 101 |
+
Measured TFLOPs: 239.66
|
| 102 |
+
Epoch 1 | iter 32 step 1 | loss train: 1.454, val: 1.313 | iter time: 545.49 ms (step) remaining time: 0:33:04
|
| 103 |
+
Epoch 1 | iter 64 step 2 | loss train: 1.463, val: 1.313 | iter time: 355.62 ms (step) remaining time: 0:31:19
|
| 104 |
+
Epoch 1 | iter 96 step 3 | loss train: 1.483, val: 1.313 | iter time: 355.27 ms (step) remaining time: 0:30:37
|
| 105 |
+
Epoch 1 | iter 128 step 4 | loss train: 1.514, val: 1.313 | iter time: 359.19 ms (step) remaining time: 0:30:11
|
| 106 |
+
Epoch 1 | iter 160 step 5 | loss train: 1.452, val: 1.313 | iter time: 357.38 ms (step) remaining time: 0:29:52
|
| 107 |
+
Epoch 1 | iter 192 step 6 | loss train: 1.396, val: 1.313 | iter time: 358.99 ms (step) remaining time: 0:29:36
|
| 108 |
+
Epoch 1 | iter 224 step 7 | loss train: 1.378, val: 1.313 | iter time: 357.63 ms (step) remaining time: 0:29:21
|
| 109 |
+
Epoch 1 | iter 256 step 8 | loss train: 1.391, val: 1.313 | iter time: 360.11 ms (step) remaining time: 0:29:08
|
| 110 |
+
Epoch 1 | iter 288 step 9 | loss train: 1.457, val: 1.313 | iter time: 358.34 ms (step) remaining time: 0:28:55
|
| 111 |
+
Epoch 1 | iter 320 step 10 | loss train: 1.460, val: 1.313 | iter time: 358.68 ms (step) remaining time: 0:28:43
|
| 112 |
+
Epoch 1 | iter 352 step 11 | loss train: 1.493, val: 1.313 | iter time: 358.48 ms (step) remaining time: 0:28:32
|
| 113 |
+
Epoch 1 | iter 384 step 12 | loss train: 1.494, val: 1.313 | iter time: 358.84 ms (step) remaining time: 0:28:20
|
| 114 |
+
Epoch 1 | iter 416 step 13 | loss train: 1.476, val: 1.313 | iter time: 360.24 ms (step) remaining time: 0:28:09
|
| 115 |
+
Epoch 1 | iter 448 step 14 | loss train: 1.484, val: 1.313 | iter time: 359.47 ms (step) remaining time: 0:27:57
|
| 116 |
+
Epoch 1 | iter 480 step 15 | loss train: 1.495, val: 1.313 | iter time: 357.80 ms (step) remaining time: 0:27:46
|
| 117 |
+
Epoch 1 | iter 512 step 16 | loss train: 1.462, val: 1.313 | iter time: 358.26 ms (step) remaining time: 0:27:35
|
| 118 |
+
Epoch 1 | iter 544 step 17 | loss train: 1.520, val: 1.313 | iter time: 358.70 ms (step) remaining time: 0:27:24
|
| 119 |
+
Epoch 1 | iter 576 step 18 | loss train: 1.462, val: 1.313 | iter time: 359.21 ms (step) remaining time: 0:27:13
|
| 120 |
+
Epoch 1 | iter 608 step 19 | loss train: 1.503, val: 1.313 | iter time: 360.84 ms (step) remaining time: 0:27:02
|
| 121 |
+
Epoch 1 | iter 640 step 20 | loss train: 1.516, val: 1.313 | iter time: 358.59 ms (step) remaining time: 0:26:51
|
| 122 |
+
Epoch 1 | iter 672 step 21 | loss train: 1.550, val: 1.313 | iter time: 358.53 ms (step) remaining time: 0:26:40
|
| 123 |
+
Epoch 1 | iter 704 step 22 | loss train: 1.521, val: 1.313 | iter time: 358.48 ms (step) remaining time: 0:26:29
|
| 124 |
+
Epoch 1 | iter 736 step 23 | loss train: 1.507, val: 1.313 | iter time: 361.00 ms (step) remaining time: 0:26:18
|
| 125 |
+
Epoch 1 | iter 768 step 24 | loss train: 1.496, val: 1.313 | iter time: 359.17 ms (step) remaining time: 0:26:07
|
| 126 |
+
Epoch 1 | iter 800 step 25 | loss train: 1.516, val: 1.313 | iter time: 358.59 ms (step) remaining time: 0:25:56
|
| 127 |
+
Epoch 1 | iter 832 step 26 | loss train: 1.479, val: 1.313 | iter time: 358.94 ms (step) remaining time: 0:25:45
|
| 128 |
+
Epoch 1 | iter 864 step 27 | loss train: 1.520, val: 1.313 | iter time: 359.95 ms (step) remaining time: 0:25:34
|
| 129 |
+
Epoch 1 | iter 896 step 28 | loss train: 1.584, val: 1.313 | iter time: 358.97 ms (step) remaining time: 0:25:23
|
| 130 |
+
Epoch 1 | iter 928 step 29 | loss train: 1.453, val: 1.313 | iter time: 358.05 ms (step) remaining time: 0:25:12
|
| 131 |
+
Epoch 1 | iter 960 step 30 | loss train: 1.601, val: 1.313 | iter time: 360.72 ms (step) remaining time: 0:25:01
|
| 132 |
+
Epoch 1 | iter 992 step 31 | loss train: 1.561, val: 1.313 | iter time: 361.10 ms (step) remaining time: 0:24:50
|
| 133 |
+
Epoch 1 | iter 1024 step 32 | loss train: 1.432, val: 1.313 | iter time: 358.49 ms (step) remaining time: 0:24:39
|
| 134 |
+
Epoch 1 | iter 1056 step 33 | loss train: 1.495, val: 1.313 | iter time: 360.98 ms (step) remaining time: 0:24:28
|
| 135 |
+
Epoch 1 | iter 1088 step 34 | loss train: 1.560, val: 1.313 | iter time: 358.22 ms (step) remaining time: 0:24:18
|
| 136 |
+
Epoch 1 | iter 1120 step 35 | loss train: 1.490, val: 1.313 | iter time: 358.52 ms (step) remaining time: 0:24:07
|
| 137 |
+
Epoch 1 | iter 1152 step 36 | loss train: 1.493, val: 1.313 | iter time: 360.04 ms (step) remaining time: 0:23:56
|
| 138 |
+
Epoch 1 | iter 1184 step 37 | loss train: 1.412, val: 1.313 | iter time: 357.77 ms (step) remaining time: 0:23:45
|
| 139 |
+
Epoch 1 | iter 1216 step 38 | loss train: 1.451, val: 1.313 | iter time: 360.18 ms (step) remaining time: 0:23:34
|
| 140 |
+
Epoch 1 | iter 1248 step 39 | loss train: 1.396, val: 1.313 | iter time: 361.18 ms (step) remaining time: 0:23:24
|
| 141 |
+
Epoch 1 | iter 1280 step 40 | loss train: 1.498, val: 1.313 | iter time: 360.85 ms (step) remaining time: 0:23:13
|
| 142 |
+
Epoch 1 | iter 1312 step 41 | loss train: 1.472, val: 1.313 | iter time: 360.48 ms (step) remaining time: 0:23:02
|
| 143 |
+
Epoch 1 | iter 1344 step 42 | loss train: 1.481, val: 1.313 | iter time: 359.84 ms (step) remaining time: 0:22:52
|
| 144 |
+
Epoch 1 | iter 1376 step 43 | loss train: 1.498, val: 1.313 | iter time: 360.22 ms (step) remaining time: 0:22:41
|
| 145 |
+
Epoch 1 | iter 1408 step 44 | loss train: 1.439, val: 1.313 | iter time: 359.80 ms (step) remaining time: 0:22:30
|
| 146 |
+
Epoch 1 | iter 1440 step 45 | loss train: 1.463, val: 1.313 | iter time: 359.85 ms (step) remaining time: 0:22:19
|
| 147 |
+
Epoch 1 | iter 1472 step 46 | loss train: 1.489, val: 1.313 | iter time: 360.05 ms (step) remaining time: 0:22:08
|
| 148 |
+
Epoch 1 | iter 1504 step 47 | loss train: 1.545, val: 1.313 | iter time: 359.81 ms (step) remaining time: 0:21:57
|
| 149 |
+
Epoch 1 | iter 1536 step 48 | loss train: 1.525, val: 1.313 | iter time: 359.03 ms (step) remaining time: 0:21:46
|
| 150 |
+
Epoch 1 | iter 1568 step 49 | loss train: 1.371, val: 1.313 | iter time: 358.91 ms (step) remaining time: 0:21:35
|
| 151 |
+
Epoch 1 | iter 1600 step 50 | loss train: 1.438, val: 1.313 | iter time: 360.97 ms (step) remaining time: 0:21:24
|
| 152 |
+
Validating ...
|
| 153 |
+
iter 1600: val loss 1.3687, val time: 5910.61 ms
|
| 154 |
+
Epoch 1 | iter 1632 step 51 | loss train: 1.563, val: 1.369 | iter time: 358.07 ms (step) remaining time: 0:21:27
|
| 155 |
+
Epoch 1 | iter 1664 step 52 | loss train: 1.511, val: 1.369 | iter time: 359.70 ms (step) remaining time: 0:21:15
|
| 156 |
+
Epoch 1 | iter 1696 step 53 | loss train: 1.454, val: 1.369 | iter time: 359.04 ms (step) remaining time: 0:21:04
|
| 157 |
+
Epoch 1 | iter 1728 step 54 | loss train: 1.441, val: 1.369 | iter time: 361.51 ms (step) remaining time: 0:20:53
|
| 158 |
+
Epoch 1 | iter 1760 step 55 | loss train: 1.495, val: 1.369 | iter time: 358.30 ms (step) remaining time: 0:20:42
|
| 159 |
+
Epoch 1 | iter 1792 step 56 | loss train: 1.546, val: 1.369 | iter time: 360.41 ms (step) remaining time: 0:20:30
|
| 160 |
+
Epoch 1 | iter 1824 step 57 | loss train: 1.446, val: 1.369 | iter time: 360.86 ms (step) remaining time: 0:20:19
|
| 161 |
+
Epoch 1 | iter 1856 step 58 | loss train: 1.440, val: 1.369 | iter time: 361.60 ms (step) remaining time: 0:20:08
|
| 162 |
+
Epoch 1 | iter 1888 step 59 | loss train: 1.455, val: 1.369 | iter time: 360.70 ms (step) remaining time: 0:19:57
|
| 163 |
+
Epoch 1 | iter 1920 step 60 | loss train: 1.493, val: 1.369 | iter time: 360.65 ms (step) remaining time: 0:19:45
|
| 164 |
+
Epoch 1 | iter 1952 step 61 | loss train: 1.385, val: 1.369 | iter time: 360.22 ms (step) remaining time: 0:19:34
|
| 165 |
+
Epoch 1 | iter 1984 step 62 | loss train: 1.443, val: 1.369 | iter time: 360.32 ms (step) remaining time: 0:19:23
|
| 166 |
+
Epoch 1 | iter 2016 step 63 | loss train: 1.404, val: 1.369 | iter time: 359.78 ms (step) remaining time: 0:19:12
|
| 167 |
+
Epoch 1 | iter 2048 step 64 | loss train: 1.460, val: 1.369 | iter time: 360.64 ms (step) remaining time: 0:19:01
|
| 168 |
+
Epoch 1 | iter 2080 step 65 | loss train: 1.474, val: 1.369 | iter time: 359.02 ms (step) remaining time: 0:18:50
|
| 169 |
+
Epoch 1 | iter 2112 step 66 | loss train: 1.486, val: 1.369 | iter time: 360.39 ms (step) remaining time: 0:18:39
|
| 170 |
+
Epoch 1 | iter 2144 step 67 | loss train: 1.451, val: 1.369 | iter time: 358.92 ms (step) remaining time: 0:18:27
|
| 171 |
+
Epoch 1 | iter 2176 step 68 | loss train: 1.423, val: 1.369 | iter time: 358.20 ms (step) remaining time: 0:18:16
|
| 172 |
+
Epoch 1 | iter 2208 step 69 | loss train: 1.385, val: 1.369 | iter time: 359.70 ms (step) remaining time: 0:18:05
|
| 173 |
+
Epoch 1 | iter 2240 step 70 | loss train: 1.494, val: 1.369 | iter time: 361.18 ms (step) remaining time: 0:17:54
|
| 174 |
+
Epoch 1 | iter 2272 step 71 | loss train: 1.427, val: 1.369 | iter time: 358.37 ms (step) remaining time: 0:17:43
|
| 175 |
+
Epoch 1 | iter 2304 step 72 | loss train: 1.461, val: 1.369 | iter time: 360.16 ms (step) remaining time: 0:17:32
|
| 176 |
+
Epoch 1 | iter 2336 step 73 | loss train: 1.462, val: 1.369 | iter time: 358.16 ms (step) remaining time: 0:17:21
|
| 177 |
+
Epoch 1 | iter 2368 step 74 | loss train: 1.428, val: 1.369 | iter time: 361.77 ms (step) remaining time: 0:17:10
|
| 178 |
+
Epoch 1 | iter 2400 step 75 | loss train: 1.473, val: 1.369 | iter time: 359.25 ms (step) remaining time: 0:16:59
|
| 179 |
+
Epoch 1 | iter 2432 step 76 | loss train: 1.452, val: 1.369 | iter time: 358.95 ms (step) remaining time: 0:16:48
|
| 180 |
+
Epoch 1 | iter 2464 step 77 | loss train: 1.437, val: 1.369 | iter time: 359.74 ms (step) remaining time: 0:16:36
|
| 181 |
+
Epoch 1 | iter 2496 step 78 | loss train: 1.412, val: 1.369 | iter time: 357.93 ms (step) remaining time: 0:16:25
|
| 182 |
+
Epoch 1 | iter 2528 step 79 | loss train: 1.410, val: 1.369 | iter time: 359.52 ms (step) remaining time: 0:16:14
|
| 183 |
+
Epoch 1 | iter 2560 step 80 | loss train: 1.495, val: 1.369 | iter time: 360.06 ms (step) remaining time: 0:16:03
|
| 184 |
+
Epoch 1 | iter 2592 step 81 | loss train: 1.445, val: 1.369 | iter time: 359.96 ms (step) remaining time: 0:15:52
|
| 185 |
+
Epoch 1 | iter 2624 step 82 | loss train: 1.395, val: 1.369 | iter time: 361.05 ms (step) remaining time: 0:15:41
|
| 186 |
+
Epoch 1 | iter 2656 step 83 | loss train: 1.394, val: 1.369 | iter time: 359.52 ms (step) remaining time: 0:15:30
|
| 187 |
+
Epoch 1 | iter 2688 step 84 | loss train: 1.393, val: 1.369 | iter time: 359.32 ms (step) remaining time: 0:15:19
|
| 188 |
+
Epoch 1 | iter 2720 step 85 | loss train: 1.453, val: 1.369 | iter time: 358.57 ms (step) remaining time: 0:15:08
|
| 189 |
+
Epoch 1 | iter 2752 step 86 | loss train: 1.357, val: 1.369 | iter time: 359.51 ms (step) remaining time: 0:14:57
|
| 190 |
+
Epoch 1 | iter 2784 step 87 | loss train: 1.437, val: 1.369 | iter time: 360.38 ms (step) remaining time: 0:14:46
|
| 191 |
+
Epoch 1 | iter 2816 step 88 | loss train: 1.412, val: 1.369 | iter time: 360.65 ms (step) remaining time: 0:14:35
|
| 192 |
+
Epoch 1 | iter 2848 step 89 | loss train: 1.428, val: 1.369 | iter time: 360.50 ms (step) remaining time: 0:14:24
|
| 193 |
+
Epoch 1 | iter 2880 step 90 | loss train: 1.461, val: 1.369 | iter time: 359.97 ms (step) remaining time: 0:14:13
|
| 194 |
+
Epoch 1 | iter 2912 step 91 | loss train: 1.411, val: 1.369 | iter time: 359.61 ms (step) remaining time: 0:14:02
|
| 195 |
+
Epoch 1 | iter 2944 step 92 | loss train: 1.393, val: 1.369 | iter time: 361.39 ms (step) remaining time: 0:13:51
|
| 196 |
+
Epoch 1 | iter 2976 step 93 | loss train: 1.489, val: 1.369 | iter time: 359.82 ms (step) remaining time: 0:13:40
|
| 197 |
+
Epoch 1 | iter 3008 step 94 | loss train: 1.439, val: 1.369 | iter time: 357.29 ms (step) remaining time: 0:13:29
|
| 198 |
+
Epoch 1 | iter 3040 step 95 | loss train: 1.472, val: 1.369 | iter time: 359.77 ms (step) remaining time: 0:13:18
|
| 199 |
+
Epoch 1 | iter 3072 step 96 | loss train: 1.467, val: 1.369 | iter time: 359.36 ms (step) remaining time: 0:13:07
|
| 200 |
+
Epoch 1 | iter 3104 step 97 | loss train: 1.438, val: 1.369 | iter time: 360.19 ms (step) remaining time: 0:12:56
|
| 201 |
+
Epoch 1 | iter 3136 step 98 | loss train: 1.414, val: 1.369 | iter time: 360.76 ms (step) remaining time: 0:12:45
|
| 202 |
+
Epoch 1 | iter 3168 step 99 | loss train: 1.436, val: 1.369 | iter time: 359.83 ms (step) remaining time: 0:12:34
|
| 203 |
+
Epoch 1 | iter 3200 step 100 | loss train: 1.443, val: 1.369 | iter time: 358.64 ms (step) remaining time: 0:12:23
|
| 204 |
+
Validating ...
|
| 205 |
+
iter 3200: val loss 1.3403, val time: 5880.97 ms
|
| 206 |
+
Saving checkpoint to 'out/pretrain/2408/step-00000100/lit_model.pth'
|
| 207 |
+
Epoch 1 | iter 3232 step 101 | loss train: 1.447, val: 1.340 | iter time: 354.20 ms (step) remaining time: 0:13:14
|
| 208 |
+
Epoch 1 | iter 3264 step 102 | loss train: 1.383, val: 1.340 | iter time: 356.96 ms (step) remaining time: 0:13:01
|
| 209 |
+
Epoch 1 | iter 3296 step 103 | loss train: 1.363, val: 1.340 | iter time: 355.70 ms (step) remaining time: 0:12:49
|
| 210 |
+
Epoch 1 | iter 3328 step 104 | loss train: 1.410, val: 1.340 | iter time: 357.99 ms (step) remaining time: 0:12:36
|
| 211 |
+
Epoch 1 | iter 3360 step 105 | loss train: 1.436, val: 1.340 | iter time: 359.20 ms (step) remaining time: 0:12:24
|
| 212 |
+
Epoch 1 | iter 3392 step 106 | loss train: 1.379, val: 1.340 | iter time: 360.20 ms (step) remaining time: 0:12:12
|
| 213 |
+
Epoch 1 | iter 3424 step 107 | loss train: 1.364, val: 1.340 | iter time: 357.29 ms (step) remaining time: 0:11:59
|
| 214 |
+
Epoch 1 | iter 3456 step 108 | loss train: 1.400, val: 1.340 | iter time: 358.10 ms (step) remaining time: 0:11:47
|
| 215 |
+
Epoch 1 | iter 3488 step 109 | loss train: 1.430, val: 1.340 | iter time: 358.16 ms (step) remaining time: 0:11:35
|
| 216 |
+
Epoch 1 | iter 3520 step 110 | loss train: 1.424, val: 1.340 | iter time: 358.76 ms (step) remaining time: 0:11:22
|
| 217 |
+
Epoch 1 | iter 3552 step 111 | loss train: 1.359, val: 1.340 | iter time: 361.14 ms (step) remaining time: 0:11:10
|
| 218 |
+
Epoch 1 | iter 3584 step 112 | loss train: 1.386, val: 1.340 | iter time: 360.25 ms (step) remaining time: 0:10:58
|
| 219 |
+
Epoch 1 | iter 3616 step 113 | loss train: 1.394, val: 1.340 | iter time: 359.48 ms (step) remaining time: 0:10:46
|
| 220 |
+
Epoch 1 | iter 3648 step 114 | loss train: 1.402, val: 1.340 | iter time: 361.39 ms (step) remaining time: 0:10:33
|
| 221 |
+
Epoch 1 | iter 3680 step 115 | loss train: 1.436, val: 1.340 | iter time: 362.00 ms (step) remaining time: 0:10:21
|
| 222 |
+
Epoch 1 | iter 3712 step 116 | loss train: 1.336, val: 1.340 | iter time: 359.77 ms (step) remaining time: 0:10:09
|
| 223 |
+
Epoch 1 | iter 3744 step 117 | loss train: 1.396, val: 1.340 | iter time: 362.14 ms (step) remaining time: 0:09:57
|
| 224 |
+
Epoch 1 | iter 3776 step 118 | loss train: 1.416, val: 1.340 | iter time: 359.51 ms (step) remaining time: 0:09:45
|
| 225 |
+
Epoch 1 | iter 3808 step 119 | loss train: 1.445, val: 1.340 | iter time: 359.81 ms (step) remaining time: 0:09:33
|
| 226 |
+
Epoch 1 | iter 3840 step 120 | loss train: 1.436, val: 1.340 | iter time: 360.43 ms (step) remaining time: 0:09:21
|
| 227 |
+
Epoch 1 | iter 3872 step 121 | loss train: 1.445, val: 1.340 | iter time: 359.62 ms (step) remaining time: 0:09:09
|
| 228 |
+
Epoch 1 | iter 3904 step 122 | loss train: 1.398, val: 1.340 | iter time: 360.20 ms (step) remaining time: 0:08:57
|
| 229 |
+
Epoch 1 | iter 3936 step 123 | loss train: 1.355, val: 1.340 | iter time: 360.86 ms (step) remaining time: 0:08:45
|
| 230 |
+
Epoch 1 | iter 3968 step 124 | loss train: 1.390, val: 1.340 | iter time: 358.78 ms (step) remaining time: 0:08:33
|
| 231 |
+
Epoch 1 | iter 4000 step 125 | loss train: 1.371, val: 1.340 | iter time: 362.27 ms (step) remaining time: 0:08:21
|
| 232 |
+
Epoch 1 | iter 4032 step 126 | loss train: 1.358, val: 1.340 | iter time: 357.21 ms (step) remaining time: 0:08:09
|
| 233 |
+
Epoch 1 | iter 4064 step 127 | loss train: 1.437, val: 1.340 | iter time: 359.92 ms (step) remaining time: 0:07:57
|
| 234 |
+
Epoch 1 | iter 4096 step 128 | loss train: 1.389, val: 1.340 | iter time: 360.57 ms (step) remaining time: 0:07:45
|
| 235 |
+
Epoch 1 | iter 4128 step 129 | loss train: 1.334, val: 1.340 | iter time: 360.53 ms (step) remaining time: 0:07:33
|
| 236 |
+
Epoch 1 | iter 4160 step 130 | loss train: 1.384, val: 1.340 | iter time: 358.85 ms (step) remaining time: 0:07:21
|
| 237 |
+
Epoch 1 | iter 4192 step 131 | loss train: 1.393, val: 1.340 | iter time: 360.78 ms (step) remaining time: 0:07:10
|
| 238 |
+
Epoch 1 | iter 4224 step 132 | loss train: 1.331, val: 1.340 | iter time: 360.15 ms (step) remaining time: 0:06:58
|
| 239 |
+
Epoch 1 | iter 4256 step 133 | loss train: 1.416, val: 1.340 | iter time: 361.20 ms (step) remaining time: 0:06:46
|
| 240 |
+
Epoch 1 | iter 4288 step 134 | loss train: 1.366, val: 1.340 | iter time: 360.69 ms (step) remaining time: 0:06:34
|
| 241 |
+
Epoch 1 | iter 4320 step 135 | loss train: 1.388, val: 1.340 | iter time: 360.74 ms (step) remaining time: 0:06:22
|
| 242 |
+
Epoch 1 | iter 4352 step 136 | loss train: 1.414, val: 1.340 | iter time: 360.34 ms (step) remaining time: 0:06:11
|
| 243 |
+
Epoch 1 | iter 4384 step 137 | loss train: 1.400, val: 1.340 | iter time: 360.01 ms (step) remaining time: 0:05:59
|
| 244 |
+
Epoch 1 | iter 4416 step 138 | loss train: 1.349, val: 1.340 | iter time: 359.45 ms (step) remaining time: 0:05:47
|
| 245 |
+
Epoch 1 | iter 4448 step 139 | loss train: 1.368, val: 1.340 | iter time: 361.23 ms (step) remaining time: 0:05:35
|
| 246 |
+
Epoch 1 | iter 4480 step 140 | loss train: 1.416, val: 1.340 | iter time: 359.25 ms (step) remaining time: 0:05:24
|
| 247 |
+
Epoch 1 | iter 4512 step 141 | loss train: 1.440, val: 1.340 | iter time: 360.27 ms (step) remaining time: 0:05:12
|
| 248 |
+
Epoch 1 | iter 4544 step 142 | loss train: 1.355, val: 1.340 | iter time: 360.02 ms (step) remaining time: 0:05:00
|
| 249 |
+
Epoch 1 | iter 4576 step 143 | loss train: 1.430, val: 1.340 | iter time: 360.39 ms (step) remaining time: 0:04:49
|
| 250 |
+
Epoch 1 | iter 4608 step 144 | loss train: 1.388, val: 1.340 | iter time: 357.89 ms (step) remaining time: 0:04:37
|
| 251 |
+
Epoch 1 | iter 4640 step 145 | loss train: 1.415, val: 1.340 | iter time: 361.73 ms (step) remaining time: 0:04:25
|
| 252 |
+
Epoch 1 | iter 4672 step 146 | loss train: 1.361, val: 1.340 | iter time: 361.89 ms (step) remaining time: 0:04:14
|
| 253 |
+
Epoch 1 | iter 4704 step 147 | loss train: 1.462, val: 1.340 | iter time: 360.22 ms (step) remaining time: 0:04:02
|
| 254 |
+
Epoch 1 | iter 4736 step 148 | loss train: 1.332, val: 1.340 | iter time: 358.17 ms (step) remaining time: 0:03:50
|
| 255 |
+
Epoch 1 | iter 4768 step 149 | loss train: 1.426, val: 1.340 | iter time: 359.78 ms (step) remaining time: 0:03:39
|
| 256 |
+
Epoch 1 | iter 4800 step 150 | loss train: 1.396, val: 1.340 | iter time: 358.87 ms (step) remaining time: 0:03:27
|
| 257 |
+
Validating ...
|
| 258 |
+
iter 4800: val loss 1.3124, val time: 5894.65 ms
|
| 259 |
+
Epoch 1 | iter 4832 step 151 | loss train: 1.331, val: 1.312 | iter time: 359.88 ms (step) remaining time: 0:03:16
|
| 260 |
+
Epoch 1 | iter 4864 step 152 | loss train: 1.366, val: 1.312 | iter time: 357.72 ms (step) remaining time: 0:03:05
|
| 261 |
+
Epoch 1 | iter 4896 step 153 | loss train: 1.350, val: 1.312 | iter time: 359.10 ms (step) remaining time: 0:02:53
|
| 262 |
+
Epoch 1 | iter 4928 step 154 | loss train: 1.319, val: 1.312 | iter time: 360.79 ms (step) remaining time: 0:02:41
|
| 263 |
+
Epoch 1 | iter 4960 step 155 | loss train: 1.384, val: 1.312 | iter time: 361.43 ms (step) remaining time: 0:02:30
|
| 264 |
+
Epoch 1 | iter 4992 step 156 | loss train: 1.324, val: 1.312 | iter time: 360.17 ms (step) remaining time: 0:02:18
|
| 265 |
+
Epoch 1 | iter 5024 step 157 | loss train: 1.419, val: 1.312 | iter time: 360.86 ms (step) remaining time: 0:02:06
|
| 266 |
+
Epoch 1 | iter 5056 step 158 | loss train: 1.291, val: 1.312 | iter time: 360.09 ms (step) remaining time: 0:01:55
|
| 267 |
+
Epoch 1 | iter 5088 step 159 | loss train: 1.428, val: 1.312 | iter time: 359.59 ms (step) remaining time: 0:01:43
|
| 268 |
+
Epoch 1 | iter 5120 step 160 | loss train: 1.356, val: 1.312 | iter time: 360.29 ms (step) remaining time: 0:01:32
|
| 269 |
+
Epoch 1 | iter 5152 step 161 | loss train: 1.402, val: 1.312 | iter time: 360.16 ms (step) remaining time: 0:01:20
|
| 270 |
+
Epoch 1 | iter 5184 step 162 | loss train: 1.347, val: 1.312 | iter time: 361.50 ms (step) remaining time: 0:01:09
|
| 271 |
+
Epoch 1 | iter 5216 step 163 | loss train: 1.360, val: 1.312 | iter time: 359.61 ms (step) remaining time: 0:00:57
|
| 272 |
+
Epoch 1 | iter 5248 step 164 | loss train: 1.400, val: 1.312 | iter time: 359.27 ms (step) remaining time: 0:00:46
|
| 273 |
+
Epoch 1 | iter 5280 step 165 | loss train: 1.421, val: 1.312 | iter time: 358.93 ms (step) remaining time: 0:00:34
|
| 274 |
+
Epoch 1 | iter 5312 step 166 | loss train: 1.320, val: 1.312 | iter time: 358.83 ms (step) remaining time: 0:00:23
|
| 275 |
+
Epoch 1 | iter 5344 step 167 | loss train: 1.381, val: 1.312 | iter time: 361.81 ms (step) remaining time: 0:00:11
|
| 276 |
+
Epoch 1 | iter 5376 step 168 | loss train: 1.324, val: 1.312 | iter time: 360.58 ms (step) remaining time: 0:00:00
|
| 277 |
+
Validating ...
|
| 278 |
+
Final evaluation | val loss: 1.308 | val ppl: 3.698
|
| 279 |
+
Saving checkpoint to 'out/pretrain/2408/final/lit_model.pth'
|
| 280 |
+
----------------------------------------
|
| 281 |
+
| Performance
|
| 282 |
+
| - Total tokens : 176,160,768
|
| 283 |
+
| - Training Time : 2032.04 s
|
| 284 |
+
| - Tok/sec : 164.71 tok/s
|
| 285 |
+
| ----------------------------------------
|
| 286 |
+
| Memory Usage
|
| 287 |
+
| - Memory Used : 26.32 GB
|
| 288 |
+
----------------------------------------
|
out/pretrain/tinyllama/teelogs/2408_full.txt
ADDED
|
@@ -0,0 +1,291 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
Initializing distributed: GLOBAL_RANK: 0, MEMBER: 1/4
|
| 2 |
+
Initializing distributed: GLOBAL_RANK: 3, MEMBER: 4/4
|
| 3 |
+
[rank: 3] Seed set to 42
|
| 4 |
+
Initializing distributed: GLOBAL_RANK: 2, MEMBER: 3/4
|
| 5 |
+
[rank: 2] Seed set to 42
|
| 6 |
+
Initializing distributed: GLOBAL_RANK: 1, MEMBER: 2/4
|
| 7 |
+
[rank: 1] Seed set to 42
|
| 8 |
+
----------------------------------------------------------------------------------------------------
|
| 9 |
+
distributed_backend=nccl
|
| 10 |
+
All distributed processes registered. Starting with 4 processes
|
| 11 |
+
----------------------------------------------------------------------------------------------------
|
| 12 |
+
|
| 13 |
+
[rank: 0] Seed set to 42
|
| 14 |
+
All GPUs are fully connected via NVLink.
|
| 15 |
+
{'data': {'batch_size': 1,
|
| 16 |
+
'data_path': PosixPath('litgpt/data/arxiv'),
|
| 17 |
+
'num_workers': 8,
|
| 18 |
+
'seed': 42,
|
| 19 |
+
'seq_length': 2048,
|
| 20 |
+
'use_starcoder': True},
|
| 21 |
+
'data_dir': PosixPath('litgpt/data/arxiv/2408'),
|
| 22 |
+
'devices': 'auto',
|
| 23 |
+
'eval': {'evaluate_example': 'first',
|
| 24 |
+
'final_validation': True,
|
| 25 |
+
'initial_validation': True,
|
| 26 |
+
'interval': 50,
|
| 27 |
+
'max_iters': 200,
|
| 28 |
+
'max_new_tokens': None},
|
| 29 |
+
'initial_checkpoint_dir': PosixPath('out/pretrain/2407_full/final'),
|
| 30 |
+
'log': {'group': None, 'project': None, 'run': None},
|
| 31 |
+
'logger_name': 'tensorboard',
|
| 32 |
+
'model_config': {'attention_logit_softcapping': None,
|
| 33 |
+
'attention_scores_scalar': None,
|
| 34 |
+
'attn_bias': False,
|
| 35 |
+
'bias': False,
|
| 36 |
+
'block_size': 2048,
|
| 37 |
+
'final_logit_softcapping': None,
|
| 38 |
+
'gelu_approximate': 'none',
|
| 39 |
+
'head_size': 64,
|
| 40 |
+
'hf_config': {'name': 'TinyLlama-1.1B-intermediate-step-1431k-3T',
|
| 41 |
+
'org': 'TinyLlama'},
|
| 42 |
+
'intermediate_size': 5632,
|
| 43 |
+
'lm_head_bias': False,
|
| 44 |
+
'mlp_class_name': 'LLaMAMLP',
|
| 45 |
+
'moe_intermediate_size': None,
|
| 46 |
+
'n_embd': 2048,
|
| 47 |
+
'n_expert': 0,
|
| 48 |
+
'n_expert_per_token': 0,
|
| 49 |
+
'n_head': 32,
|
| 50 |
+
'n_layer': 22,
|
| 51 |
+
'n_query_groups': 4,
|
| 52 |
+
'name': 'tiny-llama-1.1b',
|
| 53 |
+
'norm_1': True,
|
| 54 |
+
'norm_2': True,
|
| 55 |
+
'norm_class_name': 'RMSNorm',
|
| 56 |
+
'norm_eps': 1e-05,
|
| 57 |
+
'norm_qk': False,
|
| 58 |
+
'norm_qk_type': 'default',
|
| 59 |
+
'padded_vocab_size': 32000,
|
| 60 |
+
'padding_multiple': 64,
|
| 61 |
+
'parallel_residual': False,
|
| 62 |
+
'post_attention_norm': False,
|
| 63 |
+
'post_mlp_norm': False,
|
| 64 |
+
'rope_adjustments': None,
|
| 65 |
+
'rope_base': 10000,
|
| 66 |
+
'rope_condense_ratio': 1,
|
| 67 |
+
'rope_indices': None,
|
| 68 |
+
'rope_local_base_freq': None,
|
| 69 |
+
'rotary_percentage': 1.0,
|
| 70 |
+
'scale_embeddings': False,
|
| 71 |
+
'shared_attention_norm': False,
|
| 72 |
+
'sliding_window_indices': None,
|
| 73 |
+
'sliding_window_size': None,
|
| 74 |
+
'vocab_size': 32000},
|
| 75 |
+
'model_name': 'tiny-llama-1.1b',
|
| 76 |
+
'num_nodes': 1,
|
| 77 |
+
'optimizer': "{'class_path': 'torch.optim.AdamW', 'init_args': {'lr': 0.0004, "
|
| 78 |
+
"'weight_decay': 0.1, 'betas': [0.9, 0.95]}}",
|
| 79 |
+
'out_dir': PosixPath('out/pretrain/2408_full'),
|
| 80 |
+
'ppl': False,
|
| 81 |
+
'precision': 'bf16-mixed',
|
| 82 |
+
'resume': False,
|
| 83 |
+
'seed': 42,
|
| 84 |
+
'tokenizer_dir': PosixPath('checkpoints/TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T'),
|
| 85 |
+
'train': {'epochs': None,
|
| 86 |
+
'global_batch_size': 512,
|
| 87 |
+
'log_interval': 1,
|
| 88 |
+
'lr_warmup_fraction': None,
|
| 89 |
+
'lr_warmup_steps': 20,
|
| 90 |
+
'max_norm': 1.0,
|
| 91 |
+
'max_seq_length': 2048,
|
| 92 |
+
'max_steps': None,
|
| 93 |
+
'max_tokens': 177209344,
|
| 94 |
+
'micro_batch_size': 4,
|
| 95 |
+
'min_lr': 4e-05,
|
| 96 |
+
'save_interval': 100,
|
| 97 |
+
'tie_embeddings': None}}
|
| 98 |
+
Time to instantiate model: 0.02 seconds.
|
| 99 |
+
Total parameters: 1,100,048,384
|
| 100 |
+
[fix] out/pretrain/2407_full/final/lit_model.pth 是整包 state，提取 model 权重并原子覆盖...
|
| 101 |
+
[fix] 已覆盖为纯权重: out/pretrain/2407_full/final/lit_model.pth
|
| 102 |
+
Validating ...
|
| 103 |
+
Measured TFLOPs: 239.66
|
| 104 |
+
Epoch 1 | iter 32 step 1 | loss train: 1.458, val: 1.344 | iter time: 556.79 ms (step) remaining time: 0:34:12
|
| 105 |
+
Epoch 1 | iter 64 step 2 | loss train: 1.465, val: 1.344 | iter time: 357.59 ms (step) remaining time: 0:32:01
|
| 106 |
+
Epoch 1 | iter 96 step 3 | loss train: 1.486, val: 1.344 | iter time: 358.10 ms (step) remaining time: 0:31:10
|
| 107 |
+
Epoch 1 | iter 128 step 4 | loss train: 1.515, val: 1.344 | iter time: 356.50 ms (step) remaining time: 0:30:39
|
| 108 |
+
Epoch 1 | iter 160 step 5 | loss train: 1.452, val: 1.344 | iter time: 358.44 ms (step) remaining time: 0:30:17
|
| 109 |
+
Epoch 1 | iter 192 step 6 | loss train: 1.397, val: 1.344 | iter time: 360.78 ms (step) remaining time: 0:30:00
|
| 110 |
+
Epoch 1 | iter 224 step 7 | loss train: 1.378, val: 1.344 | iter time: 359.16 ms (step) remaining time: 0:29:44
|
| 111 |
+
Epoch 1 | iter 256 step 8 | loss train: 1.393, val: 1.344 | iter time: 359.36 ms (step) remaining time: 0:29:30
|
| 112 |
+
Epoch 1 | iter 288 step 9 | loss train: 1.460, val: 1.344 | iter time: 361.40 ms (step) remaining time: 0:29:16
|
| 113 |
+
Epoch 1 | iter 320 step 10 | loss train: 1.464, val: 1.344 | iter time: 359.52 ms (step) remaining time: 0:29:03
|
| 114 |
+
Epoch 1 | iter 352 step 11 | loss train: 1.491, val: 1.344 | iter time: 359.94 ms (step) remaining time: 0:28:50
|
| 115 |
+
Epoch 1 | iter 384 step 12 | loss train: 1.493, val: 1.344 | iter time: 358.72 ms (step) remaining time: 0:28:38
|
| 116 |
+
Epoch 1 | iter 416 step 13 | loss train: 1.471, val: 1.344 | iter time: 359.25 ms (step) remaining time: 0:28:26
|
| 117 |
+
Epoch 1 | iter 448 step 14 | loss train: 1.476, val: 1.344 | iter time: 360.71 ms (step) remaining time: 0:28:14
|
| 118 |
+
Epoch 1 | iter 480 step 15 | loss train: 1.483, val: 1.344 | iter time: 358.84 ms (step) remaining time: 0:28:03
|
| 119 |
+
Epoch 1 | iter 512 step 16 | loss train: 1.444, val: 1.344 | iter time: 360.49 ms (step) remaining time: 0:27:51
|
| 120 |
+
Epoch 1 | iter 544 step 17 | loss train: 1.509, val: 1.344 | iter time: 359.45 ms (step) remaining time: 0:27:40
|
| 121 |
+
Epoch 1 | iter 576 step 18 | loss train: 1.444, val: 1.344 | iter time: 360.95 ms (step) remaining time: 0:27:28
|
| 122 |
+
Epoch 1 | iter 608 step 19 | loss train: 1.502, val: 1.344 | iter time: 361.80 ms (step) remaining time: 0:27:17
|
| 123 |
+
Epoch 1 | iter 640 step 20 | loss train: 1.585, val: 1.344 | iter time: 360.43 ms (step) remaining time: 0:27:05
|
| 124 |
+
Epoch 1 | iter 672 step 21 | loss train: 1.587, val: 1.344 | iter time: 358.02 ms (step) remaining time: 0:26:54
|
| 125 |
+
Epoch 1 | iter 704 step 22 | loss train: 1.556, val: 1.344 | iter time: 361.13 ms (step) remaining time: 0:26:43
|
| 126 |
+
Epoch 1 | iter 736 step 23 | loss train: 1.535, val: 1.344 | iter time: 362.11 ms (step) remaining time: 0:26:32
|
| 127 |
+
Epoch 1 | iter 768 step 24 | loss train: 1.511, val: 1.344 | iter time: 360.88 ms (step) remaining time: 0:26:21
|
| 128 |
+
Epoch 1 | iter 800 step 25 | loss train: 1.524, val: 1.344 | iter time: 359.76 ms (step) remaining time: 0:26:10
|
| 129 |
+
Epoch 1 | iter 832 step 26 | loss train: 1.486, val: 1.344 | iter time: 361.83 ms (step) remaining time: 0:25:58
|
| 130 |
+
Epoch 1 | iter 864 step 27 | loss train: 1.507, val: 1.344 | iter time: 360.51 ms (step) remaining time: 0:25:47
|
| 131 |
+
Epoch 1 | iter 896 step 28 | loss train: 1.537, val: 1.344 | iter time: 360.39 ms (step) remaining time: 0:25:36
|
| 132 |
+
Epoch 1 | iter 928 step 29 | loss train: 1.457, val: 1.344 | iter time: 360.06 ms (step) remaining time: 0:25:25
|
| 133 |
+
Epoch 1 | iter 960 step 30 | loss train: 1.603, val: 1.344 | iter time: 359.78 ms (step) remaining time: 0:25:14
|
| 134 |
+
Epoch 1 | iter 992 step 31 | loss train: 1.554, val: 1.344 | iter time: 360.81 ms (step) remaining time: 0:25:03
|
| 135 |
+
Epoch 1 | iter 1024 step 32 | loss train: 1.450, val: 1.344 | iter time: 359.00 ms (step) remaining time: 0:24:52
|
| 136 |
+
Epoch 1 | iter 1056 step 33 | loss train: 1.489, val: 1.344 | iter time: 360.18 ms (step) remaining time: 0:24:41
|
| 137 |
+
Epoch 1 | iter 1088 step 34 | loss train: 1.556, val: 1.344 | iter time: 360.34 ms (step) remaining time: 0:24:30
|
| 138 |
+
Epoch 1 | iter 1120 step 35 | loss train: 1.487, val: 1.344 | iter time: 361.29 ms (step) remaining time: 0:24:19
|
| 139 |
+
Epoch 1 | iter 1152 step 36 | loss train: 1.492, val: 1.344 | iter time: 359.15 ms (step) remaining time: 0:24:08
|
| 140 |
+
Epoch 1 | iter 1184 step 37 | loss train: 1.408, val: 1.344 | iter time: 360.39 ms (step) remaining time: 0:23:57
|
| 141 |
+
Epoch 1 | iter 1216 step 38 | loss train: 1.448, val: 1.344 | iter time: 360.25 ms (step) remaining time: 0:23:46
|
| 142 |
+
Epoch 1 | iter 1248 step 39 | loss train: 1.391, val: 1.344 | iter time: 360.56 ms (step) remaining time: 0:23:35
|
| 143 |
+
Epoch 1 | iter 1280 step 40 | loss train: 1.492, val: 1.344 | iter time: 360.26 ms (step) remaining time: 0:23:24
|
| 144 |
+
Epoch 1 | iter 1312 step 41 | loss train: 1.465, val: 1.344 | iter time: 359.92 ms (step) remaining time: 0:23:13
|
| 145 |
+
Epoch 1 | iter 1344 step 42 | loss train: 1.478, val: 1.344 | iter time: 361.10 ms (step) remaining time: 0:23:03
|
| 146 |
+
Epoch 1 | iter 1376 step 43 | loss train: 1.494, val: 1.344 | iter time: 360.51 ms (step) remaining time: 0:22:52
|
| 147 |
+
Epoch 1 | iter 1408 step 44 | loss train: 1.434, val: 1.344 | iter time: 359.39 ms (step) remaining time: 0:22:41
|
| 148 |
+
Epoch 1 | iter 1440 step 45 | loss train: 1.460, val: 1.344 | iter time: 360.42 ms (step) remaining time: 0:22:30
|
| 149 |
+
Epoch 1 | iter 1472 step 46 | loss train: 1.486, val: 1.344 | iter time: 361.43 ms (step) remaining time: 0:22:19
|
| 150 |
+
Epoch 1 | iter 1504 step 47 | loss train: 1.542, val: 1.344 | iter time: 360.10 ms (step) remaining time: 0:22:08
|
| 151 |
+
Epoch 1 | iter 1536 step 48 | loss train: 1.522, val: 1.344 | iter time: 360.44 ms (step) remaining time: 0:21:58
|
| 152 |
+
Epoch 1 | iter 1568 step 49 | loss train: 1.424, val: 1.344 | iter time: 360.86 ms (step) remaining time: 0:21:47
|
| 153 |
+
Epoch 1 | iter 1600 step 50 | loss train: 1.459, val: 1.344 | iter time: 360.89 ms (step) remaining time: 0:21:36
|
| 154 |
+
Validating ...
|
| 155 |
+
iter 1600: val loss 1.3552, val time: 22362.39 ms
|
| 156 |
+
Epoch 1 | iter 1632 step 51 | loss train: 1.595, val: 1.355 | iter time: 359.85 ms (step) remaining time: 0:22:18
|
| 157 |
+
Epoch 1 | iter 1664 step 52 | loss train: 1.544, val: 1.355 | iter time: 361.40 ms (step) remaining time: 0:22:05
|
| 158 |
+
Epoch 1 | iter 1696 step 53 | loss train: 1.481, val: 1.355 | iter time: 358.16 ms (step) remaining time: 0:21:53
|
| 159 |
+
Epoch 1 | iter 1728 step 54 | loss train: 1.469, val: 1.355 | iter time: 361.34 ms (step) remaining time: 0:21:41
|
| 160 |
+
Epoch 1 | iter 1760 step 55 | loss train: 1.547, val: 1.355 | iter time: 358.79 ms (step) remaining time: 0:21:28
|
| 161 |
+
Epoch 1 | iter 1792 step 56 | loss train: 1.819, val: 1.355 | iter time: 360.56 ms (step) remaining time: 0:21:16
|
| 162 |
+
Epoch 1 | iter 1824 step 57 | loss train: 1.541, val: 1.355 | iter time: 358.84 ms (step) remaining time: 0:21:04
|
| 163 |
+
Epoch 1 | iter 1856 step 58 | loss train: 1.501, val: 1.355 | iter time: 359.05 ms (step) remaining time: 0:20:52
|
| 164 |
+
Epoch 1 | iter 1888 step 59 | loss train: 1.502, val: 1.355 | iter time: 359.98 ms (step) remaining time: 0:20:40
|
| 165 |
+
Epoch 1 | iter 1920 step 60 | loss train: 1.523, val: 1.355 | iter time: 360.47 ms (step) remaining time: 0:20:28
|
| 166 |
+
Epoch 1 | iter 1952 step 61 | loss train: 1.408, val: 1.355 | iter time: 360.91 ms (step) remaining time: 0:20:16
|
| 167 |
+
Epoch 1 | iter 1984 step 62 | loss train: 3.327, val: 1.355 | iter time: 360.42 ms (step) remaining time: 0:20:04
|
| 168 |
+
Epoch 1 | iter 2016 step 63 | loss train: 1.724, val: 1.355 | iter time: 358.85 ms (step) remaining time: 0:19:52
|
| 169 |
+
Epoch 1 | iter 2048 step 64 | loss train: 1.597, val: 1.355 | iter time: 357.56 ms (step) remaining time: 0:19:40
|
| 170 |
+
Epoch 1 | iter 2080 step 65 | loss train: 1.699, val: 1.355 | iter time: 360.26 ms (step) remaining time: 0:19:28
|
| 171 |
+
Epoch 1 | iter 2112 step 66 | loss train: 1.603, val: 1.355 | iter time: 362.22 ms (step) remaining time: 0:19:16
|
| 172 |
+
Epoch 1 | iter 2144 step 67 | loss train: 1.530, val: 1.355 | iter time: 359.96 ms (step) remaining time: 0:19:05
|
| 173 |
+
Epoch 1 | iter 2176 step 68 | loss train: 1.480, val: 1.355 | iter time: 360.02 ms (step) remaining time: 0:18:53
|
| 174 |
+
Epoch 1 | iter 2208 step 69 | loss train: 1.426, val: 1.355 | iter time: 358.89 ms (step) remaining time: 0:18:41
|
| 175 |
+
Epoch 1 | iter 2240 step 70 | loss train: 1.540, val: 1.355 | iter time: 360.23 ms (step) remaining time: 0:18:29
|
| 176 |
+
Epoch 1 | iter 2272 step 71 | loss train: 1.459, val: 1.355 | iter time: 359.89 ms (step) remaining time: 0:18:18
|
| 177 |
+
Epoch 1 | iter 2304 step 72 | loss train: 1.489, val: 1.355 | iter time: 360.13 ms (step) remaining time: 0:18:06
|
| 178 |
+
Epoch 1 | iter 2336 step 73 | loss train: 1.490, val: 1.355 | iter time: 359.68 ms (step) remaining time: 0:17:54
|
| 179 |
+
Epoch 1 | iter 2368 step 74 | loss train: 1.455, val: 1.355 | iter time: 359.32 ms (step) remaining time: 0:17:43
|
| 180 |
+
Epoch 1 | iter 2400 step 75 | loss train: 1.498, val: 1.355 | iter time: 360.18 ms (step) remaining time: 0:17:31
|
| 181 |
+
Epoch 1 | iter 2432 step 76 | loss train: 1.472, val: 1.355 | iter time: 359.40 ms (step) remaining time: 0:17:20
|
| 182 |
+
Epoch 1 | iter 2464 step 77 | loss train: 1.456, val: 1.355 | iter time: 360.49 ms (step) remaining time: 0:17:08
|
| 183 |
+
Epoch 1 | iter 2496 step 78 | loss train: 1.461, val: 1.355 | iter time: 359.38 ms (step) remaining time: 0:16:56
|
| 184 |
+
Epoch 1 | iter 2528 step 79 | loss train: 1.450, val: 1.355 | iter time: 359.69 ms (step) remaining time: 0:16:45
|
| 185 |
+
Epoch 1 | iter 2560 step 80 | loss train: 1.558, val: 1.355 | iter time: 359.64 ms (step) remaining time: 0:16:33
|
| 186 |
+
Epoch 1 | iter 2592 step 81 | loss train: 1.482, val: 1.355 | iter time: 360.25 ms (step) remaining time: 0:16:22
|
| 187 |
+
Epoch 1 | iter 2624 step 82 | loss train: 1.444, val: 1.355 | iter time: 358.39 ms (step) remaining time: 0:16:10
|
| 188 |
+
Epoch 1 | iter 2656 step 83 | loss train: 1.424, val: 1.355 | iter time: 358.90 ms (step) remaining time: 0:15:59
|
| 189 |
+
Epoch 1 | iter 2688 step 84 | loss train: 1.422, val: 1.355 | iter time: 361.70 ms (step) remaining time: 0:15:47
|
| 190 |
+
Epoch 1 | iter 2720 step 85 | loss train: 1.477, val: 1.355 | iter time: 360.55 ms (step) remaining time: 0:15:36
|
| 191 |
+
Epoch 1 | iter 2752 step 86 | loss train: 1.381, val: 1.355 | iter time: 358.76 ms (step) remaining time: 0:15:25
|
| 192 |
+
Epoch 1 | iter 2784 step 87 | loss train: 1.458, val: 1.355 | iter time: 358.94 ms (step) remaining time: 0:15:13
|
| 193 |
+
Epoch 1 | iter 2816 step 88 | loss train: 1.430, val: 1.355 | iter time: 361.58 ms (step) remaining time: 0:15:02
|
| 194 |
+
Epoch 1 | iter 2848 step 89 | loss train: 1.447, val: 1.355 | iter time: 359.51 ms (step) remaining time: 0:14:50
|
| 195 |
+
Epoch 1 | iter 2880 step 90 | loss train: 1.477, val: 1.355 | iter time: 359.26 ms (step) remaining time: 0:14:39
|
| 196 |
+
Epoch 1 | iter 2912 step 91 | loss train: 1.429, val: 1.355 | iter time: 359.35 ms (step) remaining time: 0:14:28
|
| 197 |
+
Epoch 1 | iter 2944 step 92 | loss train: 1.407, val: 1.355 | iter time: 359.75 ms (step) remaining time: 0:14:16
|
| 198 |
+
Epoch 1 | iter 2976 step 93 | loss train: 1.504, val: 1.355 | iter time: 359.08 ms (step) remaining time: 0:14:05
|
| 199 |
+
Epoch 1 | iter 3008 step 94 | loss train: 1.450, val: 1.355 | iter time: 360.41 ms (step) remaining time: 0:13:54
|
| 200 |
+
Epoch 1 | iter 3040 step 95 | loss train: 1.485, val: 1.355 | iter time: 360.40 ms (step) remaining time: 0:13:42
|
| 201 |
+
Epoch 1 | iter 3072 step 96 | loss train: 1.481, val: 1.355 | iter time: 360.54 ms (step) remaining time: 0:13:31
|
| 202 |
+
Epoch 1 | iter 3104 step 97 | loss train: 1.449, val: 1.355 | iter time: 360.36 ms (step) remaining time: 0:13:20
|
| 203 |
+
Epoch 1 | iter 3136 step 98 | loss train: 1.428, val: 1.355 | iter time: 359.49 ms (step) remaining time: 0:13:08
|
| 204 |
+
Epoch 1 | iter 3168 step 99 | loss train: 1.448, val: 1.355 | iter time: 359.20 ms (step) remaining time: 0:12:57
|
| 205 |
+
Epoch 1 | iter 3200 step 100 | loss train: 1.454, val: 1.355 | iter time: 360.19 ms (step) remaining time: 0:12:46
|
| 206 |
+
Validating ...
|
| 207 |
+
iter 3200: val loss 1.2657, val time: 22383.18 ms
|
| 208 |
+
Saving checkpoint to 'out/pretrain/2408_full/step-00000100/lit_model.pth'
|
| 209 |
+
Epoch 1 | iter 3232 step 101 | loss train: 1.461, val: 1.266 | iter time: 356.17 ms (step) remaining time: 0:13:01
|
| 210 |
+
Epoch 1 | iter 3264 step 102 | loss train: 1.392, val: 1.266 | iter time: 357.68 ms (step) remaining time: 0:12:49
|
| 211 |
+
Epoch 1 | iter 3296 step 103 | loss train: 1.374, val: 1.266 | iter time: 360.23 ms (step) remaining time: 0:12:37
|
| 212 |
+
Epoch 1 | iter 3328 step 104 | loss train: 1.422, val: 1.266 | iter time: 359.53 ms (step) remaining time: 0:12:26
|
| 213 |
+
Epoch 1 | iter 3360 step 105 | loss train: 1.445, val: 1.266 | iter time: 357.92 ms (step) remaining time: 0:12:14
|
| 214 |
+
Epoch 1 | iter 3392 step 106 | loss train: 1.389, val: 1.266 | iter time: 360.29 ms (step) remaining time: 0:12:02
|
| 215 |
+
Epoch 1 | iter 3424 step 107 | loss train: 1.376, val: 1.266 | iter time: 359.41 ms (step) remaining time: 0:11:50
|
| 216 |
+
Epoch 1 | iter 3456 step 108 | loss train: 1.411, val: 1.266 | iter time: 360.28 ms (step) remaining time: 0:11:38
|
| 217 |
+
Epoch 1 | iter 3488 step 109 | loss train: 1.439, val: 1.266 | iter time: 359.12 ms (step) remaining time: 0:11:27
|
| 218 |
+
Epoch 1 | iter 3520 step 110 | loss train: 1.434, val: 1.266 | iter time: 360.70 ms (step) remaining time: 0:11:15
|
| 219 |
+
Epoch 1 | iter 3552 step 111 | loss train: 1.368, val: 1.266 | iter time: 358.52 ms (step) remaining time: 0:11:03
|
| 220 |
+
Epoch 1 | iter 3584 step 112 | loss train: 1.394, val: 1.266 | iter time: 360.97 ms (step) remaining time: 0:10:51
|
| 221 |
+
Epoch 1 | iter 3616 step 113 | loss train: 1.403, val: 1.266 | iter time: 359.62 ms (step) remaining time: 0:10:40
|
| 222 |
+
Epoch 1 | iter 3648 step 114 | loss train: 1.411, val: 1.266 | iter time: 358.79 ms (step) remaining time: 0:10:28
|
| 223 |
+
Epoch 1 | iter 3680 step 115 | loss train: 1.444, val: 1.266 | iter time: 360.07 ms (step) remaining time: 0:10:16
|
| 224 |
+
Epoch 1 | iter 3712 step 116 | loss train: 1.344, val: 1.266 | iter time: 359.49 ms (step) remaining time: 0:10:05
|
| 225 |
+
Epoch 1 | iter 3744 step 117 | loss train: 1.419, val: 1.266 | iter time: 359.36 ms (step) remaining time: 0:09:53
|
| 226 |
+
Epoch 1 | iter 3776 step 118 | loss train: 1.439, val: 1.266 | iter time: 362.05 ms (step) remaining time: 0:09:41
|
| 227 |
+
Epoch 1 | iter 3808 step 119 | loss train: 1.456, val: 1.266 | iter time: 360.09 ms (step) remaining time: 0:09:30
|
| 228 |
+
Epoch 1 | iter 3840 step 120 | loss train: 1.444, val: 1.266 | iter time: 360.96 ms (step) remaining time: 0:09:18
|
| 229 |
+
Epoch 1 | iter 3872 step 121 | loss train: 1.454, val: 1.266 | iter time: 359.45 ms (step) remaining time: 0:09:07
|
| 230 |
+
Epoch 1 | iter 3904 step 122 | loss train: 1.404, val: 1.266 | iter time: 359.08 ms (step) remaining time: 0:08:55
|
| 231 |
+
Epoch 1 | iter 3936 step 123 | loss train: 1.365, val: 1.266 | iter time: 358.56 ms (step) remaining time: 0:08:43
|
| 232 |
+
Epoch 1 | iter 3968 step 124 | loss train: 1.397, val: 1.266 | iter time: 358.56 ms (step) remaining time: 0:08:32
|
| 233 |
+
Epoch 1 | iter 4000 step 125 | loss train: 1.384, val: 1.266 | iter time: 360.95 ms (step) remaining time: 0:08:20
|
| 234 |
+
Epoch 1 | iter 4032 step 126 | loss train: 1.367, val: 1.266 | iter time: 359.92 ms (step) remaining time: 0:08:09
|
| 235 |
+
Epoch 1 | iter 4064 step 127 | loss train: 1.447, val: 1.266 | iter time: 359.99 ms (step) remaining time: 0:07:57
|
| 236 |
+
Epoch 1 | iter 4096 step 128 | loss train: 1.398, val: 1.266 | iter time: 358.36 ms (step) remaining time: 0:07:46
|
| 237 |
+
Epoch 1 | iter 4128 step 129 | loss train: 1.346, val: 1.266 | iter time: 361.45 ms (step) remaining time: 0:07:34
|
| 238 |
+
Epoch 1 | iter 4160 step 130 | loss train: 1.394, val: 1.266 | iter time: 357.60 ms (step) remaining time: 0:07:23
|
| 239 |
+
Epoch 1 | iter 4192 step 131 | loss train: 1.401, val: 1.266 | iter time: 360.07 ms (step) remaining time: 0:07:11
|
| 240 |
+
Epoch 1 | iter 4224 step 132 | loss train: 1.339, val: 1.266 | iter time: 361.45 ms (step) remaining time: 0:07:00
|
| 241 |
+
Epoch 1 | iter 4256 step 133 | loss train: 1.424, val: 1.266 | iter time: 361.24 ms (step) remaining time: 0:06:48
|
| 242 |
+
Epoch 1 | iter 4288 step 134 | loss train: 1.374, val: 1.266 | iter time: 361.40 ms (step) remaining time: 0:06:37
|
| 243 |
+
Epoch 1 | iter 4320 step 135 | loss train: 1.398, val: 1.266 | iter time: 358.66 ms (step) remaining time: 0:06:25
|
| 244 |
+
Epoch 1 | iter 4352 step 136 | loss train: 1.423, val: 1.266 | iter time: 359.68 ms (step) remaining time: 0:06:14
|
| 245 |
+
Epoch 1 | iter 4384 step 137 | loss train: 1.409, val: 1.266 | iter time: 358.44 ms (step) remaining time: 0:06:02
|
| 246 |
+
Epoch 1 | iter 4416 step 138 | loss train: 1.358, val: 1.266 | iter time: 360.90 ms (step) remaining time: 0:05:51
|
| 247 |
+
Epoch 1 | iter 4448 step 139 | loss train: 1.375, val: 1.266 | iter time: 361.52 ms (step) remaining time: 0:05:39
|
| 248 |
+
Epoch 1 | iter 4480 step 140 | loss train: 1.423, val: 1.266 | iter time: 360.35 ms (step) remaining time: 0:05:28
|
| 249 |
+
Epoch 1 | iter 4512 step 141 | loss train: 1.447, val: 1.266 | iter time: 359.10 ms (step) remaining time: 0:05:16
|
| 250 |
+
Epoch 1 | iter 4544 step 142 | loss train: 1.362, val: 1.266 | iter time: 360.46 ms (step) remaining time: 0:05:05
|
| 251 |
+
Epoch 1 | iter 4576 step 143 | loss train: 1.441, val: 1.266 | iter time: 360.89 ms (step) remaining time: 0:04:54
|
| 252 |
+
Epoch 1 | iter 4608 step 144 | loss train: 1.394, val: 1.266 | iter time: 360.13 ms (step) remaining time: 0:04:42
|
| 253 |
+
Epoch 1 | iter 4640 step 145 | loss train: 1.422, val: 1.266 | iter time: 359.09 ms (step) remaining time: 0:04:31
|
| 254 |
+
Epoch 1 | iter 4672 step 146 | loss train: 1.371, val: 1.266 | iter time: 358.57 ms (step) remaining time: 0:04:20
|
| 255 |
+
Epoch 1 | iter 4704 step 147 | loss train: 1.471, val: 1.266 | iter time: 361.35 ms (step) remaining time: 0:04:08
|
| 256 |
+
Epoch 1 | iter 4736 step 148 | loss train: 1.338, val: 1.266 | iter time: 360.11 ms (step) remaining time: 0:03:57
|
| 257 |
+
Epoch 1 | iter 4768 step 149 | loss train: 1.434, val: 1.266 | iter time: 360.77 ms (step) remaining time: 0:03:45
|
| 258 |
+
Epoch 1 | iter 4800 step 150 | loss train: 1.403, val: 1.266 | iter time: 360.44 ms (step) remaining time: 0:03:34
|
| 259 |
+
Validating ...
|
| 260 |
+
iter 4800: val loss 1.2096, val time: 22381.61 ms
|
| 261 |
+
Epoch 1 | iter 4832 step 151 | loss train: 1.339, val: 1.210 | iter time: 359.89 ms (step) remaining time: 0:03:25
|
| 262 |
+
Epoch 1 | iter 4864 step 152 | loss train: 1.374, val: 1.210 | iter time: 360.31 ms (step) remaining time: 0:03:14
|
| 263 |
+
Epoch 1 | iter 4896 step 153 | loss train: 1.354, val: 1.210 | iter time: 357.09 ms (step) remaining time: 0:03:02
|
| 264 |
+
Epoch 1 | iter 4928 step 154 | loss train: 1.325, val: 1.210 | iter time: 360.00 ms (step) remaining time: 0:02:51
|
| 265 |
+
Epoch 1 | iter 4960 step 155 | loss train: 1.394, val: 1.210 | iter time: 360.33 ms (step) remaining time: 0:02:39
|
| 266 |
+
Epoch 1 | iter 4992 step 156 | loss train: 1.333, val: 1.210 | iter time: 360.26 ms (step) remaining time: 0:02:28
|
| 267 |
+
Epoch 1 | iter 5024 step 157 | loss train: 1.425, val: 1.210 | iter time: 359.99 ms (step) remaining time: 0:02:17
|
| 268 |
+
Epoch 1 | iter 5056 step 158 | loss train: 1.298, val: 1.210 | iter time: 360.05 ms (step) remaining time: 0:02:05
|
| 269 |
+
Epoch 1 | iter 5088 step 159 | loss train: 1.434, val: 1.210 | iter time: 361.24 ms (step) remaining time: 0:01:54
|
| 270 |
+
Epoch 1 | iter 5120 step 160 | loss train: 1.362, val: 1.210 | iter time: 361.03 ms (step) remaining time: 0:01:42
|
| 271 |
+
Epoch 1 | iter 5152 step 161 | loss train: 1.410, val: 1.210 | iter time: 358.45 ms (step) remaining time: 0:01:31
|
| 272 |
+
Epoch 1 | iter 5184 step 162 | loss train: 1.355, val: 1.210 | iter time: 359.71 ms (step) remaining time: 0:01:19
|
| 273 |
+
Epoch 1 | iter 5216 step 163 | loss train: 1.368, val: 1.210 | iter time: 358.79 ms (step) remaining time: 0:01:08
|
| 274 |
+
Epoch 1 | iter 5248 step 164 | loss train: 1.409, val: 1.210 | iter time: 360.47 ms (step) remaining time: 0:00:56
|
| 275 |
+
Epoch 1 | iter 5280 step 165 | loss train: 1.429, val: 1.210 | iter time: 360.04 ms (step) remaining time: 0:00:45
|
| 276 |
+
Epoch 1 | iter 5312 step 166 | loss train: 1.328, val: 1.210 | iter time: 358.53 ms (step) remaining time: 0:00:34
|
| 277 |
+
Epoch 1 | iter 5344 step 167 | loss train: 1.389, val: 1.210 | iter time: 359.97 ms (step) remaining time: 0:00:22
|
| 278 |
+
Epoch 1 | iter 5376 step 168 | loss train: 1.330, val: 1.210 | iter time: 360.91 ms (step) remaining time: 0:00:11
|
| 279 |
+
Epoch 2 | iter 5408 step 169 | loss train: 1.329, val: 1.210 | iter time: 363.53 ms (step) remaining time: 0:00:00
|
| 280 |
+
Validating ...
|
| 281 |
+
Final evaluation | val loss: 1.188 | val ppl: 3.280
|
| 282 |
+
Saving checkpoint to 'out/pretrain/2408_full/final/lit_model.pth'
|
| 283 |
+
----------------------------------------
|
| 284 |
+
| Performance
|
| 285 |
+
| - Total tokens : 177,209,344
|
| 286 |
+
| - Training Time : 1988.04 s
|
| 287 |
+
| - Tok/sec : 95.92 tok/s
|
| 288 |
+
| ----------------------------------------
|
| 289 |
+
| Memory Usage
|
| 290 |
+
| - Memory Used : 26.32 GB
|
| 291 |
+
----------------------------------------
|
out/pretrain/tinyllama/teelogs/2408_lr4e-5.txt
ADDED
|
@@ -0,0 +1,298 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
Initializing distributed: GLOBAL_RANK: 0, MEMBER: 1/4
|
| 2 |
+
Initializing distributed: GLOBAL_RANK: 3, MEMBER: 4/4
|
| 3 |
+
[rank: 3] Seed set to 42
|
| 4 |
+
Initializing distributed: GLOBAL_RANK: 2, MEMBER: 3/4
|
| 5 |
+
[rank: 2] Seed set to 42
|
| 6 |
+
Initializing distributed: GLOBAL_RANK: 1, MEMBER: 2/4
|
| 7 |
+
----------------------------------------------------------------------------------------------------
|
| 8 |
+
distributed_backend=nccl
|
| 9 |
+
All distributed processes registered. Starting with 4 processes
|
| 10 |
+
----------------------------------------------------------------------------------------------------
|
| 11 |
+
|
| 12 |
+
[rank: 1] Seed set to 42
|
| 13 |
+
[rank: 0] Seed set to 42
|
| 14 |
+
/mnt/data/litgpt/litgpt/pretrain.py:515: UserWarning: A newer version of litdata is available (0.2.52). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 15 |
+
train_dataloader = data.train_dataloader()
|
| 16 |
+
/mnt/data/litgpt/litgpt/pretrain.py:515: UserWarning: A newer version of litdata is available (0.2.52). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 17 |
+
train_dataloader = data.train_dataloader()
|
| 18 |
+
/mnt/data/litgpt/litgpt/pretrain.py:515: UserWarning: A newer version of litdata is available (0.2.52). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 19 |
+
train_dataloader = data.train_dataloader()
|
| 20 |
+
/mnt/data/litgpt/litgpt/pretrain.py:515: UserWarning: A newer version of litdata is available (0.2.52). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 21 |
+
train_dataloader = data.train_dataloader()
|
| 22 |
+
All GPUs are fully connected via NVLink.
|
| 23 |
+
{'data': {'batch_size': 1,
|
| 24 |
+
'data_path': PosixPath('litgpt/data/arxiv'),
|
| 25 |
+
'num_workers': 0,
|
| 26 |
+
'seed': 42,
|
| 27 |
+
'seq_length': 2048,
|
| 28 |
+
'use_starcoder': True},
|
| 29 |
+
'data_dir': PosixPath('litgpt/data/arxiv/2408'),
|
| 30 |
+
'devices': 'auto',
|
| 31 |
+
'eval': {'evaluate_example': 'first',
|
| 32 |
+
'final_validation': True,
|
| 33 |
+
'initial_validation': True,
|
| 34 |
+
'interval': 50,
|
| 35 |
+
'max_iters': 200,
|
| 36 |
+
'max_new_tokens': None},
|
| 37 |
+
'initial_checkpoint_dir': PosixPath('out/pretrain/tinyllama/2407_full/final'),
|
| 38 |
+
'log': {'group': None, 'project': None, 'run': None},
|
| 39 |
+
'logger_name': 'tensorboard',
|
| 40 |
+
'model_config': {'attention_logit_softcapping': None,
|
| 41 |
+
'attention_scores_scalar': None,
|
| 42 |
+
'attn_bias': False,
|
| 43 |
+
'bias': False,
|
| 44 |
+
'block_size': 2048,
|
| 45 |
+
'final_logit_softcapping': None,
|
| 46 |
+
'gelu_approximate': 'none',
|
| 47 |
+
'head_size': 64,
|
| 48 |
+
'hf_config': {'name': 'TinyLlama-1.1B-intermediate-step-1431k-3T',
|
| 49 |
+
'org': 'TinyLlama'},
|
| 50 |
+
'intermediate_size': 5632,
|
| 51 |
+
'lm_head_bias': False,
|
| 52 |
+
'mlp_class_name': 'LLaMAMLP',
|
| 53 |
+
'moe_intermediate_size': None,
|
| 54 |
+
'n_embd': 2048,
|
| 55 |
+
'n_expert': 0,
|
| 56 |
+
'n_expert_per_token': 0,
|
| 57 |
+
'n_head': 32,
|
| 58 |
+
'n_layer': 22,
|
| 59 |
+
'n_query_groups': 4,
|
| 60 |
+
'name': 'tiny-llama-1.1b',
|
| 61 |
+
'norm_1': True,
|
| 62 |
+
'norm_2': True,
|
| 63 |
+
'norm_class_name': 'RMSNorm',
|
| 64 |
+
'norm_eps': 1e-05,
|
| 65 |
+
'norm_qk': False,
|
| 66 |
+
'norm_qk_type': 'default',
|
| 67 |
+
'padded_vocab_size': 32000,
|
| 68 |
+
'padding_multiple': 64,
|
| 69 |
+
'parallel_residual': False,
|
| 70 |
+
'post_attention_norm': False,
|
| 71 |
+
'post_mlp_norm': False,
|
| 72 |
+
'rope_adjustments': None,
|
| 73 |
+
'rope_base': 10000,
|
| 74 |
+
'rope_condense_ratio': 1,
|
| 75 |
+
'rope_indices': None,
|
| 76 |
+
'rope_local_base_freq': None,
|
| 77 |
+
'rotary_percentage': 1.0,
|
| 78 |
+
'scale_embeddings': False,
|
| 79 |
+
'shared_attention_norm': False,
|
| 80 |
+
'sliding_window_indices': None,
|
| 81 |
+
'sliding_window_size': None,
|
| 82 |
+
'vocab_size': 32000},
|
| 83 |
+
'model_name': 'tiny-llama-1.1b',
|
| 84 |
+
'num_nodes': 1,
|
| 85 |
+
'optimizer': "{'class_path': 'torch.optim.AdamW', 'init_args': {'lr': 4e-05, "
|
| 86 |
+
"'weight_decay': 0.1, 'betas': [0.9, 0.95]}}",
|
| 87 |
+
'out_dir': PosixPath('out/pretrain/tinyllama/2408_lr4e-5'),
|
| 88 |
+
'ppl': False,
|
| 89 |
+
'precision': 'bf16-mixed',
|
| 90 |
+
'resume': False,
|
| 91 |
+
'seed': 42,
|
| 92 |
+
'tokenizer_dir': PosixPath('checkpoints/TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T'),
|
| 93 |
+
'train': {'epochs': None,
|
| 94 |
+
'global_batch_size': 512,
|
| 95 |
+
'log_interval': 1,
|
| 96 |
+
'lr_warmup_fraction': None,
|
| 97 |
+
'lr_warmup_steps': 20,
|
| 98 |
+
'max_norm': 1.0,
|
| 99 |
+
'max_seq_length': 2048,
|
| 100 |
+
'max_steps': None,
|
| 101 |
+
'max_tokens': 177209344,
|
| 102 |
+
'micro_batch_size': 4,
|
| 103 |
+
'min_lr': 4e-05,
|
| 104 |
+
'save_interval': 100,
|
| 105 |
+
'tie_embeddings': None}}
|
| 106 |
+
Time to instantiate model: 0.02 seconds.
|
| 107 |
+
Total parameters: 1,100,048,384
|
| 108 |
+
[ok] out/pretrain/tinyllama/2407_full/final/lit_model.pth 已是纯权重
|
| 109 |
+
Validating ...
|
| 110 |
+
Measured TFLOPs: 239.66
|
| 111 |
+
Epoch 1 | iter 32 step 1 | loss train: 1.451, val: 1.357 | iter time: 564.25 ms (step) remaining time: 0:32:45
|
| 112 |
+
Epoch 1 | iter 64 step 2 | loss train: 1.412, val: 1.357 | iter time: 358.32 ms (step) remaining time: 0:31:16
|
| 113 |
+
Epoch 1 | iter 96 step 3 | loss train: 1.501, val: 1.357 | iter time: 358.58 ms (step) remaining time: 0:30:40
|
| 114 |
+
Epoch 1 | iter 128 step 4 | loss train: 1.405, val: 1.357 | iter time: 357.45 ms (step) remaining time: 0:30:17
|
| 115 |
+
Epoch 1 | iter 160 step 5 | loss train: 1.419, val: 1.357 | iter time: 359.22 ms (step) remaining time: 0:30:00
|
| 116 |
+
Epoch 1 | iter 192 step 6 | loss train: 1.398, val: 1.357 | iter time: 359.88 ms (step) remaining time: 0:29:45
|
| 117 |
+
Epoch 1 | iter 224 step 7 | loss train: 1.454, val: 1.357 | iter time: 358.66 ms (step) remaining time: 0:29:32
|
| 118 |
+
Epoch 1 | iter 256 step 8 | loss train: 1.449, val: 1.357 | iter time: 360.74 ms (step) remaining time: 0:29:19
|
| 119 |
+
Epoch 1 | iter 288 step 9 | loss train: 1.409, val: 1.357 | iter time: 359.51 ms (step) remaining time: 0:29:09
|
| 120 |
+
Epoch 1 | iter 320 step 10 | loss train: 1.417, val: 1.357 | iter time: 360.20 ms (step) remaining time: 0:28:58
|
| 121 |
+
Epoch 1 | iter 352 step 11 | loss train: 1.440, val: 1.357 | iter time: 359.61 ms (step) remaining time: 0:28:46
|
| 122 |
+
Epoch 1 | iter 384 step 12 | loss train: 1.415, val: 1.357 | iter time: 359.53 ms (step) remaining time: 0:28:34
|
| 123 |
+
Epoch 1 | iter 416 step 13 | loss train: 1.479, val: 1.357 | iter time: 358.18 ms (step) remaining time: 0:28:23
|
| 124 |
+
Epoch 1 | iter 448 step 14 | loss train: 1.429, val: 1.357 | iter time: 359.84 ms (step) remaining time: 0:28:11
|
| 125 |
+
Epoch 1 | iter 480 step 15 | loss train: 1.366, val: 1.357 | iter time: 359.35 ms (step) remaining time: 0:28:00
|
| 126 |
+
Epoch 1 | iter 512 step 16 | loss train: 1.497, val: 1.357 | iter time: 361.43 ms (step) remaining time: 0:27:48
|
| 127 |
+
Epoch 1 | iter 544 step 17 | loss train: 1.392, val: 1.357 | iter time: 360.36 ms (step) remaining time: 0:27:37
|
| 128 |
+
Epoch 1 | iter 576 step 18 | loss train: 1.447, val: 1.357 | iter time: 359.41 ms (step) remaining time: 0:27:26
|
| 129 |
+
Epoch 1 | iter 608 step 19 | loss train: 1.374, val: 1.357 | iter time: 459.52 ms (step) remaining time: 0:27:15
|
| 130 |
+
Epoch 1 | iter 640 step 20 | loss train: 1.462, val: 1.357 | iter time: 358.72 ms (step) remaining time: 0:27:04
|
| 131 |
+
Epoch 1 | iter 672 step 21 | loss train: 1.410, val: 1.357 | iter time: 359.14 ms (step) remaining time: 0:26:53
|
| 132 |
+
Epoch 1 | iter 704 step 22 | loss train: 1.441, val: 1.357 | iter time: 359.78 ms (step) remaining time: 0:26:42
|
| 133 |
+
Epoch 1 | iter 736 step 23 | loss train: 1.408, val: 1.357 | iter time: 361.34 ms (step) remaining time: 0:26:31
|
| 134 |
+
Epoch 1 | iter 768 step 24 | loss train: 1.472, val: 1.357 | iter time: 358.33 ms (step) remaining time: 0:26:20
|
| 135 |
+
Epoch 1 | iter 800 step 25 | loss train: 1.457, val: 1.357 | iter time: 359.30 ms (step) remaining time: 0:26:09
|
| 136 |
+
Epoch 1 | iter 832 step 26 | loss train: 1.491, val: 1.357 | iter time: 358.45 ms (step) remaining time: 0:25:58
|
| 137 |
+
Epoch 1 | iter 864 step 27 | loss train: 1.431, val: 1.357 | iter time: 361.55 ms (step) remaining time: 0:25:47
|
| 138 |
+
Epoch 1 | iter 896 step 28 | loss train: 1.428, val: 1.357 | iter time: 361.07 ms (step) remaining time: 0:25:36
|
| 139 |
+
Epoch 1 | iter 928 step 29 | loss train: 1.490, val: 1.357 | iter time: 359.53 ms (step) remaining time: 0:25:25
|
| 140 |
+
Epoch 1 | iter 960 step 30 | loss train: 1.387, val: 1.357 | iter time: 360.94 ms (step) remaining time: 0:25:14
|
| 141 |
+
Epoch 1 | iter 992 step 31 | loss train: 1.451, val: 1.357 | iter time: 359.43 ms (step) remaining time: 0:25:03
|
| 142 |
+
Epoch 1 | iter 1024 step 32 | loss train: 1.402, val: 1.357 | iter time: 359.41 ms (step) remaining time: 0:24:52
|
| 143 |
+
Epoch 1 | iter 1056 step 33 | loss train: 1.385, val: 1.357 | iter time: 359.88 ms (step) remaining time: 0:24:41
|
| 144 |
+
Epoch 1 | iter 1088 step 34 | loss train: 1.421, val: 1.357 | iter time: 359.45 ms (step) remaining time: 0:24:30
|
| 145 |
+
Epoch 1 | iter 1120 step 35 | loss train: 1.453, val: 1.357 | iter time: 360.28 ms (step) remaining time: 0:24:19
|
| 146 |
+
Epoch 1 | iter 1152 step 36 | loss train: 1.469, val: 1.357 | iter time: 359.00 ms (step) remaining time: 0:24:08
|
| 147 |
+
Epoch 1 | iter 1184 step 37 | loss train: 1.333, val: 1.357 | iter time: 358.67 ms (step) remaining time: 0:23:57
|
| 148 |
+
Epoch 1 | iter 1216 step 38 | loss train: 1.385, val: 1.357 | iter time: 359.26 ms (step) remaining time: 0:23:46
|
| 149 |
+
Epoch 1 | iter 1248 step 39 | loss train: 1.457, val: 1.357 | iter time: 360.87 ms (step) remaining time: 0:23:35
|
| 150 |
+
Epoch 1 | iter 1280 step 40 | loss train: 1.468, val: 1.357 | iter time: 359.63 ms (step) remaining time: 0:23:24
|
| 151 |
+
Epoch 1 | iter 1312 step 41 | loss train: 1.363, val: 1.357 | iter time: 360.46 ms (step) remaining time: 0:23:13
|
| 152 |
+
Epoch 1 | iter 1344 step 42 | loss train: 1.413, val: 1.357 | iter time: 360.37 ms (step) remaining time: 0:23:02
|
| 153 |
+
Epoch 1 | iter 1376 step 43 | loss train: 1.411, val: 1.357 | iter time: 360.40 ms (step) remaining time: 0:22:51
|
| 154 |
+
Epoch 1 | iter 1408 step 44 | loss train: 1.393, val: 1.357 | iter time: 360.33 ms (step) remaining time: 0:22:40
|
| 155 |
+
Epoch 1 | iter 1440 step 45 | loss train: 1.374, val: 1.357 | iter time: 359.64 ms (step) remaining time: 0:22:29
|
| 156 |
+
Epoch 1 | iter 1472 step 46 | loss train: 1.402, val: 1.357 | iter time: 360.88 ms (step) remaining time: 0:22:19
|
| 157 |
+
Epoch 1 | iter 1504 step 47 | loss train: 1.378, val: 1.357 | iter time: 358.82 ms (step) remaining time: 0:22:08
|
| 158 |
+
Epoch 1 | iter 1536 step 48 | loss train: 1.377, val: 1.357 | iter time: 358.09 ms (step) remaining time: 0:21:57
|
| 159 |
+
Epoch 1 | iter 1568 step 49 | loss train: 1.385, val: 1.357 | iter time: 358.65 ms (step) remaining time: 0:21:46
|
| 160 |
+
Epoch 1 | iter 1600 step 50 | loss train: 1.343, val: 1.357 | iter time: 359.79 ms (step) remaining time: 0:21:35
|
| 161 |
+
Validating ...
|
| 162 |
+
iter 1600: val loss 1.3176, val time: 21942.50 ms
|
| 163 |
+
Epoch 1 | iter 1632 step 51 | loss train: 1.459, val: 1.318 | iter time: 361.78 ms (step) remaining time: 0:22:15
|
| 164 |
+
Epoch 1 | iter 1664 step 52 | loss train: 1.454, val: 1.318 | iter time: 359.96 ms (step) remaining time: 0:22:03
|
| 165 |
+
Epoch 1 | iter 1696 step 53 | loss train: 1.396, val: 1.318 | iter time: 361.47 ms (step) remaining time: 0:21:50
|
| 166 |
+
Epoch 1 | iter 1728 step 54 | loss train: 1.476, val: 1.318 | iter time: 358.80 ms (step) remaining time: 0:21:38
|
| 167 |
+
Epoch 1 | iter 1760 step 55 | loss train: 1.390, val: 1.318 | iter time: 677.12 ms (step) remaining time: 0:21:26
|
| 168 |
+
Epoch 1 | iter 1792 step 56 | loss train: 1.415, val: 1.318 | iter time: 360.52 ms (step) remaining time: 0:21:14
|
| 169 |
+
Epoch 1 | iter 1824 step 57 | loss train: 1.442, val: 1.318 | iter time: 361.78 ms (step) remaining time: 0:21:02
|
| 170 |
+
Epoch 1 | iter 1856 step 58 | loss train: 1.456, val: 1.318 | iter time: 361.70 ms (step) remaining time: 0:20:50
|
| 171 |
+
Epoch 1 | iter 1888 step 59 | loss train: 1.411, val: 1.318 | iter time: 362.05 ms (step) remaining time: 0:20:38
|
| 172 |
+
Epoch 1 | iter 1920 step 60 | loss train: 1.436, val: 1.318 | iter time: 359.75 ms (step) remaining time: 0:20:26
|
| 173 |
+
Epoch 1 | iter 1952 step 61 | loss train: 1.412, val: 1.318 | iter time: 359.40 ms (step) remaining time: 0:20:14
|
| 174 |
+
Epoch 1 | iter 1984 step 62 | loss train: 1.395, val: 1.318 | iter time: 359.34 ms (step) remaining time: 0:20:02
|
| 175 |
+
Epoch 1 | iter 2016 step 63 | loss train: 1.419, val: 1.318 | iter time: 360.03 ms (step) remaining time: 0:19:50
|
| 176 |
+
Epoch 1 | iter 2048 step 64 | loss train: 1.410, val: 1.318 | iter time: 360.68 ms (step) remaining time: 0:19:39
|
| 177 |
+
Epoch 1 | iter 2080 step 65 | loss train: 1.336, val: 1.318 | iter time: 358.81 ms (step) remaining time: 0:19:27
|
| 178 |
+
Epoch 1 | iter 2112 step 66 | loss train: 1.338, val: 1.318 | iter time: 359.95 ms (step) remaining time: 0:19:15
|
| 179 |
+
Epoch 1 | iter 2144 step 67 | loss train: 1.389, val: 1.318 | iter time: 359.72 ms (step) remaining time: 0:19:03
|
| 180 |
+
Epoch 1 | iter 2176 step 68 | loss train: 1.424, val: 1.318 | iter time: 359.37 ms (step) remaining time: 0:18:52
|
| 181 |
+
Epoch 1 | iter 2208 step 69 | loss train: 1.476, val: 1.318 | iter time: 359.64 ms (step) remaining time: 0:18:40
|
| 182 |
+
Epoch 1 | iter 2240 step 70 | loss train: 1.349, val: 1.318 | iter time: 361.64 ms (step) remaining time: 0:18:28
|
| 183 |
+
Epoch 1 | iter 2272 step 71 | loss train: 1.398, val: 1.318 | iter time: 359.20 ms (step) remaining time: 0:18:17
|
| 184 |
+
Epoch 1 | iter 2304 step 72 | loss train: 1.403, val: 1.318 | iter time: 358.98 ms (step) remaining time: 0:18:05
|
| 185 |
+
Epoch 1 | iter 2336 step 73 | loss train: 1.407, val: 1.318 | iter time: 359.70 ms (step) remaining time: 0:17:53
|
| 186 |
+
Epoch 1 | iter 2368 step 74 | loss train: 1.438, val: 1.318 | iter time: 358.49 ms (step) remaining time: 0:17:42
|
| 187 |
+
Epoch 1 | iter 2400 step 75 | loss train: 1.436, val: 1.318 | iter time: 361.74 ms (step) remaining time: 0:17:30
|
| 188 |
+
Epoch 1 | iter 2432 step 76 | loss train: 1.371, val: 1.318 | iter time: 359.09 ms (step) remaining time: 0:17:19
|
| 189 |
+
Epoch 1 | iter 2464 step 77 | loss train: 1.419, val: 1.318 | iter time: 359.18 ms (step) remaining time: 0:17:07
|
| 190 |
+
Epoch 1 | iter 2496 step 78 | loss train: 1.439, val: 1.318 | iter time: 360.09 ms (step) remaining time: 0:16:55
|
| 191 |
+
Epoch 1 | iter 2528 step 79 | loss train: 1.384, val: 1.318 | iter time: 358.98 ms (step) remaining time: 0:16:44
|
| 192 |
+
Epoch 1 | iter 2560 step 80 | loss train: 1.366, val: 1.318 | iter time: 359.61 ms (step) remaining time: 0:16:33
|
| 193 |
+
Epoch 1 | iter 2592 step 81 | loss train: 1.482, val: 1.318 | iter time: 360.42 ms (step) remaining time: 0:16:21
|
| 194 |
+
Epoch 1 | iter 2624 step 82 | loss train: 1.332, val: 1.318 | iter time: 358.74 ms (step) remaining time: 0:16:10
|
| 195 |
+
Epoch 1 | iter 2656 step 83 | loss train: 1.492, val: 1.318 | iter time: 359.88 ms (step) remaining time: 0:15:58
|
| 196 |
+
Epoch 1 | iter 2688 step 84 | loss train: 1.445, val: 1.318 | iter time: 358.31 ms (step) remaining time: 0:15:47
|
| 197 |
+
Epoch 1 | iter 2720 step 85 | loss train: 1.398, val: 1.318 | iter time: 360.69 ms (step) remaining time: 0:15:35
|
| 198 |
+
Epoch 1 | iter 2752 step 86 | loss train: 1.450, val: 1.318 | iter time: 359.74 ms (step) remaining time: 0:15:24
|
| 199 |
+
Epoch 1 | iter 2784 step 87 | loss train: 1.416, val: 1.318 | iter time: 359.66 ms (step) remaining time: 0:15:13
|
| 200 |
+
Epoch 1 | iter 2816 step 88 | loss train: 1.476, val: 1.318 | iter time: 360.79 ms (step) remaining time: 0:15:01
|
| 201 |
+
Epoch 1 | iter 2848 step 89 | loss train: 1.371, val: 1.318 | iter time: 357.84 ms (step) remaining time: 0:14:50
|
| 202 |
+
Epoch 1 | iter 2880 step 90 | loss train: 1.470, val: 1.318 | iter time: 358.94 ms (step) remaining time: 0:14:39
|
| 203 |
+
Epoch 1 | iter 2912 step 91 | loss train: 1.328, val: 1.318 | iter time: 358.27 ms (step) remaining time: 0:14:27
|
| 204 |
+
Epoch 1 | iter 2944 step 92 | loss train: 1.442, val: 1.318 | iter time: 359.77 ms (step) remaining time: 0:14:16
|
| 205 |
+
Epoch 1 | iter 2976 step 93 | loss train: 1.475, val: 1.318 | iter time: 359.28 ms (step) remaining time: 0:14:05
|
| 206 |
+
Epoch 1 | iter 3008 step 94 | loss train: 1.383, val: 1.318 | iter time: 363.83 ms (step) remaining time: 0:13:53
|
| 207 |
+
Epoch 1 | iter 3040 step 95 | loss train: 1.422, val: 1.318 | iter time: 360.13 ms (step) remaining time: 0:13:42
|
| 208 |
+
Epoch 1 | iter 3072 step 96 | loss train: 1.459, val: 1.318 | iter time: 360.31 ms (step) remaining time: 0:13:31
|
| 209 |
+
Epoch 1 | iter 3104 step 97 | loss train: 1.492, val: 1.318 | iter time: 361.09 ms (step) remaining time: 0:13:19
|
| 210 |
+
Epoch 1 | iter 3136 step 98 | loss train: 1.476, val: 1.318 | iter time: 359.04 ms (step) remaining time: 0:13:08
|
| 211 |
+
Epoch 1 | iter 3168 step 99 | loss train: 1.344, val: 1.318 | iter time: 358.83 ms (step) remaining time: 0:12:57
|
| 212 |
+
Epoch 1 | iter 3200 step 100 | loss train: 1.407, val: 1.318 | iter time: 360.71 ms (step) remaining time: 0:12:46
|
| 213 |
+
Validating ...
|
| 214 |
+
iter 3200: val loss 1.2629, val time: 21940.88 ms
|
| 215 |
+
Saving checkpoint to 'out/pretrain/tinyllama/2408_lr4e-5/step-00000100/lit_model.pth'
|
| 216 |
+
Epoch 1 | iter 3232 step 101 | loss train: 1.465, val: 1.263 | iter time: 355.97 ms (step) remaining time: 0:13:00
|
| 217 |
+
Epoch 1 | iter 3264 step 102 | loss train: 1.429, val: 1.263 | iter time: 359.09 ms (step) remaining time: 0:12:48
|
| 218 |
+
Epoch 1 | iter 3296 step 103 | loss train: 1.436, val: 1.263 | iter time: 359.06 ms (step) remaining time: 0:12:36
|
| 219 |
+
Epoch 1 | iter 3328 step 104 | loss train: 1.424, val: 1.263 | iter time: 359.38 ms (step) remaining time: 0:12:24
|
| 220 |
+
Epoch 1 | iter 3360 step 105 | loss train: 1.425, val: 1.263 | iter time: 359.97 ms (step) remaining time: 0:12:13
|
| 221 |
+
Epoch 1 | iter 3392 step 106 | loss train: 1.426, val: 1.263 | iter time: 359.51 ms (step) remaining time: 0:12:01
|
| 222 |
+
Epoch 1 | iter 3424 step 107 | loss train: 1.420, val: 1.263 | iter time: 358.93 ms (step) remaining time: 0:11:49
|
| 223 |
+
Epoch 1 | iter 3456 step 108 | loss train: 1.394, val: 1.263 | iter time: 359.30 ms (step) remaining time: 0:11:37
|
| 224 |
+
Epoch 1 | iter 3488 step 109 | loss train: 1.424, val: 1.263 | iter time: 360.28 ms (step) remaining time: 0:11:25
|
| 225 |
+
Epoch 1 | iter 3520 step 110 | loss train: 1.407, val: 1.263 | iter time: 360.67 ms (step) remaining time: 0:11:14
|
| 226 |
+
Epoch 1 | iter 3552 step 111 | loss train: 1.416, val: 1.263 | iter time: 360.38 ms (step) remaining time: 0:11:02
|
| 227 |
+
Epoch 1 | iter 3584 step 112 | loss train: 1.397, val: 1.263 | iter time: 361.11 ms (step) remaining time: 0:10:50
|
| 228 |
+
Epoch 1 | iter 3616 step 113 | loss train: 1.408, val: 1.263 | iter time: 601.64 ms (step) remaining time: 0:10:39
|
| 229 |
+
Epoch 1 | iter 3648 step 114 | loss train: 1.481, val: 1.263 | iter time: 360.76 ms (step) remaining time: 0:10:27
|
| 230 |
+
Epoch 1 | iter 3680 step 115 | loss train: 1.409, val: 1.263 | iter time: 360.08 ms (step) remaining time: 0:10:15
|
| 231 |
+
Epoch 1 | iter 3712 step 116 | loss train: 1.380, val: 1.263 | iter time: 360.00 ms (step) remaining time: 0:10:04
|
| 232 |
+
Epoch 1 | iter 3744 step 117 | loss train: 1.389, val: 1.263 | iter time: 360.30 ms (step) remaining time: 0:09:52
|
| 233 |
+
Epoch 1 | iter 3776 step 118 | loss train: 1.352, val: 1.263 | iter time: 360.52 ms (step) remaining time: 0:09:40
|
| 234 |
+
Epoch 1 | iter 3808 step 119 | loss train: 1.371, val: 1.263 | iter time: 358.91 ms (step) remaining time: 0:09:29
|
| 235 |
+
Epoch 1 | iter 3840 step 120 | loss train: 1.421, val: 1.263 | iter time: 360.84 ms (step) remaining time: 0:09:17
|
| 236 |
+
Epoch 1 | iter 3872 step 121 | loss train: 1.337, val: 1.263 | iter time: 361.95 ms (step) remaining time: 0:09:06
|
| 237 |
+
Epoch 1 | iter 3904 step 122 | loss train: 1.489, val: 1.263 | iter time: 358.32 ms (step) remaining time: 0:08:54
|
| 238 |
+
Epoch 1 | iter 3936 step 123 | loss train: 1.377, val: 1.263 | iter time: 359.93 ms (step) remaining time: 0:08:42
|
| 239 |
+
Epoch 1 | iter 3968 step 124 | loss train: 1.439, val: 1.263 | iter time: 358.88 ms (step) remaining time: 0:08:31
|
| 240 |
+
Epoch 1 | iter 4000 step 125 | loss train: 1.362, val: 1.263 | iter time: 359.74 ms (step) remaining time: 0:08:20
|
| 241 |
+
Epoch 1 | iter 4032 step 126 | loss train: 1.406, val: 1.263 | iter time: 359.15 ms (step) remaining time: 0:08:08
|
| 242 |
+
Epoch 1 | iter 4064 step 127 | loss train: 1.366, val: 1.263 | iter time: 358.82 ms (step) remaining time: 0:07:57
|
| 243 |
+
Epoch 1 | iter 4096 step 128 | loss train: 1.380, val: 1.263 | iter time: 359.08 ms (step) remaining time: 0:07:45
|
| 244 |
+
Epoch 1 | iter 4128 step 129 | loss train: 1.384, val: 1.263 | iter time: 359.91 ms (step) remaining time: 0:07:34
|
| 245 |
+
Epoch 1 | iter 4160 step 130 | loss train: 1.419, val: 1.263 | iter time: 360.38 ms (step) remaining time: 0:07:22
|
| 246 |
+
Epoch 1 | iter 4192 step 131 | loss train: 1.391, val: 1.263 | iter time: 360.00 ms (step) remaining time: 0:07:11
|
| 247 |
+
Epoch 1 | iter 4224 step 132 | loss train: 1.495, val: 1.263 | iter time: 360.55 ms (step) remaining time: 0:06:59
|
| 248 |
+
Epoch 1 | iter 4256 step 133 | loss train: 1.401, val: 1.263 | iter time: 359.65 ms (step) remaining time: 0:06:48
|
| 249 |
+
Epoch 1 | iter 4288 step 134 | loss train: 1.386, val: 1.263 | iter time: 359.92 ms (step) remaining time: 0:06:36
|
| 250 |
+
Epoch 1 | iter 4320 step 135 | loss train: 1.427, val: 1.263 | iter time: 360.18 ms (step) remaining time: 0:06:25
|
| 251 |
+
Epoch 1 | iter 4352 step 136 | loss train: 1.408, val: 1.263 | iter time: 360.58 ms (step) remaining time: 0:06:13
|
| 252 |
+
Epoch 1 | iter 4384 step 137 | loss train: 1.412, val: 1.263 | iter time: 359.62 ms (step) remaining time: 0:06:02
|
| 253 |
+
Epoch 1 | iter 4416 step 138 | loss train: 1.416, val: 1.263 | iter time: 362.31 ms (step) remaining time: 0:05:50
|
| 254 |
+
Epoch 1 | iter 4448 step 139 | loss train: 1.433, val: 1.263 | iter time: 359.63 ms (step) remaining time: 0:05:39
|
| 255 |
+
Epoch 1 | iter 4480 step 140 | loss train: 1.450, val: 1.263 | iter time: 359.16 ms (step) remaining time: 0:05:28
|
| 256 |
+
Epoch 1 | iter 4512 step 141 | loss train: 1.384, val: 1.263 | iter time: 360.05 ms (step) remaining time: 0:05:16
|
| 257 |
+
Epoch 1 | iter 4544 step 142 | loss train: 1.348, val: 1.263 | iter time: 359.53 ms (step) remaining time: 0:05:05
|
| 258 |
+
Epoch 1 | iter 4576 step 143 | loss train: 1.396, val: 1.263 | iter time: 359.50 ms (step) remaining time: 0:04:53
|
| 259 |
+
Epoch 1 | iter 4608 step 144 | loss train: 1.374, val: 1.263 | iter time: 360.42 ms (step) remaining time: 0:04:42
|
| 260 |
+
Epoch 1 | iter 4640 step 145 | loss train: 1.386, val: 1.263 | iter time: 358.19 ms (step) remaining time: 0:04:31
|
| 261 |
+
Epoch 1 | iter 4672 step 146 | loss train: 1.401, val: 1.263 | iter time: 358.30 ms (step) remaining time: 0:04:19
|
| 262 |
+
Epoch 1 | iter 4704 step 147 | loss train: 1.368, val: 1.263 | iter time: 358.33 ms (step) remaining time: 0:04:08
|
| 263 |
+
Epoch 1 | iter 4736 step 148 | loss train: 1.373, val: 1.263 | iter time: 359.05 ms (step) remaining time: 0:03:57
|
| 264 |
+
Epoch 1 | iter 4768 step 149 | loss train: 1.351, val: 1.263 | iter time: 357.45 ms (step) remaining time: 0:03:45
|
| 265 |
+
Epoch 1 | iter 4800 step 150 | loss train: 1.422, val: 1.263 | iter time: 361.23 ms (step) remaining time: 0:03:34
|
| 266 |
+
Validating ...
|
| 267 |
+
iter 4800: val loss 1.2647, val time: 21943.29 ms
|
| 268 |
+
Epoch 1 | iter 4832 step 151 | loss train: 1.419, val: 1.265 | iter time: 360.61 ms (step) remaining time: 0:03:25
|
| 269 |
+
Epoch 1 | iter 4864 step 152 | loss train: 1.409, val: 1.265 | iter time: 359.59 ms (step) remaining time: 0:03:14
|
| 270 |
+
Epoch 1 | iter 4896 step 153 | loss train: 1.341, val: 1.265 | iter time: 361.08 ms (step) remaining time: 0:03:02
|
| 271 |
+
Epoch 1 | iter 4928 step 154 | loss train: 1.423, val: 1.265 | iter time: 359.58 ms (step) remaining time: 0:02:51
|
| 272 |
+
Epoch 1 | iter 4960 step 155 | loss train: 1.353, val: 1.265 | iter time: 361.92 ms (step) remaining time: 0:02:39
|
| 273 |
+
Epoch 1 | iter 4992 step 156 | loss train: 1.403, val: 1.265 | iter time: 361.02 ms (step) remaining time: 0:02:28
|
| 274 |
+
Epoch 1 | iter 5024 step 157 | loss train: 1.419, val: 1.265 | iter time: 360.62 ms (step) remaining time: 0:02:16
|
| 275 |
+
Epoch 1 | iter 5056 step 158 | loss train: 1.471, val: 1.265 | iter time: 358.83 ms (step) remaining time: 0:02:05
|
| 276 |
+
Epoch 1 | iter 5088 step 159 | loss train: 1.309, val: 1.265 | iter time: 359.32 ms (step) remaining time: 0:01:53
|
| 277 |
+
Epoch 1 | iter 5120 step 160 | loss train: 1.454, val: 1.265 | iter time: 359.49 ms (step) remaining time: 0:01:42
|
| 278 |
+
Epoch 1 | iter 5152 step 161 | loss train: 1.370, val: 1.265 | iter time: 360.16 ms (step) remaining time: 0:01:31
|
| 279 |
+
Epoch 1 | iter 5184 step 162 | loss train: 1.366, val: 1.265 | iter time: 359.12 ms (step) remaining time: 0:01:19
|
| 280 |
+
Epoch 1 | iter 5216 step 163 | loss train: 1.455, val: 1.265 | iter time: 361.51 ms (step) remaining time: 0:01:08
|
| 281 |
+
Epoch 1 | iter 5248 step 164 | loss train: 1.374, val: 1.265 | iter time: 360.18 ms (step) remaining time: 0:00:56
|
| 282 |
+
Epoch 1 | iter 5280 step 165 | loss train: 1.392, val: 1.265 | iter time: 360.68 ms (step) remaining time: 0:00:45
|
| 283 |
+
Epoch 1 | iter 5312 step 166 | loss train: 1.397, val: 1.265 | iter time: 360.54 ms (step) remaining time: 0:00:34
|
| 284 |
+
Epoch 1 | iter 5344 step 167 | loss train: 1.394, val: 1.265 | iter time: 361.64 ms (step) remaining time: 0:00:22
|
| 285 |
+
Epoch 1 | iter 5376 step 168 | loss train: 1.372, val: 1.265 | iter time: 359.24 ms (step) remaining time: 0:00:11
|
| 286 |
+
Epoch 2 | iter 5408 step 169 | loss train: 1.337, val: 1.265 | iter time: 358.30 ms (step) remaining time: 0:00:00
|
| 287 |
+
Validating ...
|
| 288 |
+
Final evaluation | val loss: 1.265 | val ppl: 3.545
|
| 289 |
+
Saving checkpoint to 'out/pretrain/tinyllama/2408_lr4e-5/final/lit_model.pth'
|
| 290 |
+
----------------------------------------
|
| 291 |
+
| Performance
|
| 292 |
+
| - Total tokens : 177,209,344
|
| 293 |
+
| - Training Time : 1983.42 s
|
| 294 |
+
| - Tok/sec : 91.64 tok/s
|
| 295 |
+
| ----------------------------------------
|
| 296 |
+
| Memory Usage
|
| 297 |
+
| - Memory Used : 26.32 GB
|
| 298 |
+
----------------------------------------
|
out/pretrain/tinyllama/teelogs/2409.txt
ADDED
|
File without changes
|
out/pretrain/tinyllama/teelogs/2409_full.txt
ADDED
|
@@ -0,0 +1,354 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
Initializing distributed: GLOBAL_RANK: 0, MEMBER: 1/4
|
| 2 |
+
Initializing distributed: GLOBAL_RANK: 1, MEMBER: 2/4
|
| 3 |
+
[rank: 1] Seed set to 42
|
| 4 |
+
Initializing distributed: GLOBAL_RANK: 3, MEMBER: 4/4
|
| 5 |
+
Initializing distributed: GLOBAL_RANK: 2, MEMBER: 3/4
|
| 6 |
+
[rank: 2] Seed set to 42
|
| 7 |
+
[rank: 3] Seed set to 42
|
| 8 |
+
----------------------------------------------------------------------------------------------------
|
| 9 |
+
distributed_backend=nccl
|
| 10 |
+
All distributed processes registered. Starting with 4 processes
|
| 11 |
+
----------------------------------------------------------------------------------------------------
|
| 12 |
+
|
| 13 |
+
[rank: 0] Seed set to 42
|
| 14 |
+
All GPUs are fully connected via NVLink.
|
| 15 |
+
{'data': {'batch_size': 1,
|
| 16 |
+
'data_path': PosixPath('litgpt/data/arxiv'),
|
| 17 |
+
'num_workers': 8,
|
| 18 |
+
'seed': 42,
|
| 19 |
+
'seq_length': 2048,
|
| 20 |
+
'use_starcoder': True},
|
| 21 |
+
'data_dir': PosixPath('litgpt/data/arxiv/2409'),
|
| 22 |
+
'devices': 'auto',
|
| 23 |
+
'eval': {'evaluate_example': 'first',
|
| 24 |
+
'final_validation': True,
|
| 25 |
+
'initial_validation': True,
|
| 26 |
+
'interval': 50,
|
| 27 |
+
'max_iters': 200,
|
| 28 |
+
'max_new_tokens': None},
|
| 29 |
+
'initial_checkpoint_dir': PosixPath('out/pretrain/2408_full/final'),
|
| 30 |
+
'log': {'group': None, 'project': None, 'run': None},
|
| 31 |
+
'logger_name': 'tensorboard',
|
| 32 |
+
'model_config': {'attention_logit_softcapping': None,
|
| 33 |
+
'attention_scores_scalar': None,
|
| 34 |
+
'attn_bias': False,
|
| 35 |
+
'bias': False,
|
| 36 |
+
'block_size': 2048,
|
| 37 |
+
'final_logit_softcapping': None,
|
| 38 |
+
'gelu_approximate': 'none',
|
| 39 |
+
'head_size': 64,
|
| 40 |
+
'hf_config': {'name': 'TinyLlama-1.1B-intermediate-step-1431k-3T',
|
| 41 |
+
'org': 'TinyLlama'},
|
| 42 |
+
'intermediate_size': 5632,
|
| 43 |
+
'lm_head_bias': False,
|
| 44 |
+
'mlp_class_name': 'LLaMAMLP',
|
| 45 |
+
'moe_intermediate_size': None,
|
| 46 |
+
'n_embd': 2048,
|
| 47 |
+
'n_expert': 0,
|
| 48 |
+
'n_expert_per_token': 0,
|
| 49 |
+
'n_head': 32,
|
| 50 |
+
'n_layer': 22,
|
| 51 |
+
'n_query_groups': 4,
|
| 52 |
+
'name': 'tiny-llama-1.1b',
|
| 53 |
+
'norm_1': True,
|
| 54 |
+
'norm_2': True,
|
| 55 |
+
'norm_class_name': 'RMSNorm',
|
| 56 |
+
'norm_eps': 1e-05,
|
| 57 |
+
'norm_qk': False,
|
| 58 |
+
'norm_qk_type': 'default',
|
| 59 |
+
'padded_vocab_size': 32000,
|
| 60 |
+
'padding_multiple': 64,
|
| 61 |
+
'parallel_residual': False,
|
| 62 |
+
'post_attention_norm': False,
|
| 63 |
+
'post_mlp_norm': False,
|
| 64 |
+
'rope_adjustments': None,
|
| 65 |
+
'rope_base': 10000,
|
| 66 |
+
'rope_condense_ratio': 1,
|
| 67 |
+
'rope_indices': None,
|
| 68 |
+
'rope_local_base_freq': None,
|
| 69 |
+
'rotary_percentage': 1.0,
|
| 70 |
+
'scale_embeddings': False,
|
| 71 |
+
'shared_attention_norm': False,
|
| 72 |
+
'sliding_window_indices': None,
|
| 73 |
+
'sliding_window_size': None,
|
| 74 |
+
'vocab_size': 32000},
|
| 75 |
+
'model_name': 'tiny-llama-1.1b',
|
| 76 |
+
'num_nodes': 1,
|
| 77 |
+
'optimizer': "{'class_path': 'torch.optim.AdamW', 'init_args': {'lr': 0.0004, "
|
| 78 |
+
"'weight_decay': 0.1, 'betas': [0.9, 0.95]}}",
|
| 79 |
+
'out_dir': PosixPath('out/pretrain/2409_full'),
|
| 80 |
+
'ppl': False,
|
| 81 |
+
'precision': 'bf16-mixed',
|
| 82 |
+
'resume': False,
|
| 83 |
+
'seed': 42,
|
| 84 |
+
'tokenizer_dir': PosixPath('checkpoints/TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T'),
|
| 85 |
+
'train': {'epochs': None,
|
| 86 |
+
'global_batch_size': 512,
|
| 87 |
+
'log_interval': 1,
|
| 88 |
+
'lr_warmup_fraction': None,
|
| 89 |
+
'lr_warmup_steps': 20,
|
| 90 |
+
'max_norm': 1.0,
|
| 91 |
+
'max_seq_length': 2048,
|
| 92 |
+
'max_steps': None,
|
| 93 |
+
'max_tokens': 240123904,
|
| 94 |
+
'micro_batch_size': 4,
|
| 95 |
+
'min_lr': 4e-05,
|
| 96 |
+
'save_interval': 100,
|
| 97 |
+
'tie_embeddings': None}}
|
| 98 |
+
Time to instantiate model: 0.04 seconds.
|
| 99 |
+
Total parameters: 1,100,048,384
|
| 100 |
+
[fix] out/pretrain/2408_full/final/lit_model.pth 是整包 state，提取 model 权重并原子覆盖...
|
| 101 |
+
[fix] 已覆盖为纯权重: out/pretrain/2408_full/final/lit_model.pth
|
| 102 |
+
Validating ...
|
| 103 |
+
Measured TFLOPs: 239.66
|
| 104 |
+
Epoch 1 | iter 32 step 1 | loss train: 1.498, val: 1.439 | iter time: 558.21 ms (step) remaining time: 0:45:30
|
| 105 |
+
Epoch 1 | iter 64 step 2 | loss train: 1.432, val: 1.439 | iter time: 355.65 ms (step) remaining time: 0:43:02
|
| 106 |
+
Epoch 1 | iter 96 step 3 | loss train: 1.408, val: 1.439 | iter time: 357.49 ms (step) remaining time: 0:42:06
|
| 107 |
+
Epoch 1 | iter 128 step 4 | loss train: 1.420, val: 1.439 | iter time: 359.73 ms (step) remaining time: 0:41:33
|
| 108 |
+
Epoch 1 | iter 160 step 5 | loss train: 1.421, val: 1.439 | iter time: 359.07 ms (step) remaining time: 0:41:10
|
| 109 |
+
Epoch 1 | iter 192 step 6 | loss train: 1.451, val: 1.439 | iter time: 360.19 ms (step) remaining time: 0:40:52
|
| 110 |
+
Epoch 1 | iter 224 step 7 | loss train: 1.450, val: 1.439 | iter time: 358.86 ms (step) remaining time: 0:40:35
|
| 111 |
+
Epoch 1 | iter 256 step 8 | loss train: 1.471, val: 1.439 | iter time: 358.43 ms (step) remaining time: 0:40:21
|
| 112 |
+
Epoch 1 | iter 288 step 9 | loss train: 1.468, val: 1.439 | iter time: 360.64 ms (step) remaining time: 0:40:07
|
| 113 |
+
Epoch 1 | iter 320 step 10 | loss train: 1.372, val: 1.439 | iter time: 357.91 ms (step) remaining time: 0:39:54
|
| 114 |
+
Epoch 1 | iter 352 step 11 | loss train: 1.486, val: 1.439 | iter time: 360.52 ms (step) remaining time: 0:39:41
|
| 115 |
+
Epoch 1 | iter 384 step 12 | loss train: 1.443, val: 1.439 | iter time: 358.10 ms (step) remaining time: 0:39:29
|
| 116 |
+
Epoch 1 | iter 416 step 13 | loss train: 1.522, val: 1.439 | iter time: 358.28 ms (step) remaining time: 0:39:17
|
| 117 |
+
Epoch 1 | iter 448 step 14 | loss train: 1.455, val: 1.439 | iter time: 360.51 ms (step) remaining time: 0:39:05
|
| 118 |
+
Epoch 1 | iter 480 step 15 | loss train: 1.406, val: 1.439 | iter time: 360.65 ms (step) remaining time: 0:38:53
|
| 119 |
+
Epoch 1 | iter 512 step 16 | loss train: 1.525, val: 1.439 | iter time: 358.56 ms (step) remaining time: 0:38:42
|
| 120 |
+
Epoch 1 | iter 544 step 17 | loss train: 1.481, val: 1.439 | iter time: 360.99 ms (step) remaining time: 0:38:30
|
| 121 |
+
Epoch 1 | iter 576 step 18 | loss train: 1.462, val: 1.439 | iter time: 361.35 ms (step) remaining time: 0:38:19
|
| 122 |
+
Epoch 1 | iter 608 step 19 | loss train: 1.465, val: 1.439 | iter time: 359.94 ms (step) remaining time: 0:38:07
|
| 123 |
+
Epoch 1 | iter 640 step 20 | loss train: 1.490, val: 1.439 | iter time: 359.97 ms (step) remaining time: 0:37:56
|
| 124 |
+
Epoch 1 | iter 672 step 21 | loss train: 1.507, val: 1.439 | iter time: 361.58 ms (step) remaining time: 0:37:45
|
| 125 |
+
Epoch 1 | iter 704 step 22 | loss train: 1.431, val: 1.439 | iter time: 358.75 ms (step) remaining time: 0:37:34
|
| 126 |
+
Epoch 1 | iter 736 step 23 | loss train: 1.510, val: 1.439 | iter time: 359.75 ms (step) remaining time: 0:37:23
|
| 127 |
+
Epoch 1 | iter 768 step 24 | loss train: 1.512, val: 1.439 | iter time: 360.48 ms (step) remaining time: 0:37:11
|
| 128 |
+
Epoch 1 | iter 800 step 25 | loss train: 1.461, val: 1.439 | iter time: 360.71 ms (step) remaining time: 0:37:00
|
| 129 |
+
Epoch 1 | iter 832 step 26 | loss train: 1.365, val: 1.439 | iter time: 361.25 ms (step) remaining time: 0:36:49
|
| 130 |
+
Epoch 1 | iter 864 step 27 | loss train: 1.463, val: 1.439 | iter time: 360.68 ms (step) remaining time: 0:36:38
|
| 131 |
+
Epoch 1 | iter 896 step 28 | loss train: 1.444, val: 1.439 | iter time: 360.16 ms (step) remaining time: 0:36:27
|
| 132 |
+
Epoch 1 | iter 928 step 29 | loss train: 1.409, val: 1.439 | iter time: 359.23 ms (step) remaining time: 0:36:16
|
| 133 |
+
Epoch 1 | iter 960 step 30 | loss train: 1.474, val: 1.439 | iter time: 359.96 ms (step) remaining time: 0:36:05
|
| 134 |
+
Epoch 1 | iter 992 step 31 | loss train: 1.516, val: 1.439 | iter time: 359.90 ms (step) remaining time: 0:35:54
|
| 135 |
+
Epoch 1 | iter 1024 step 32 | loss train: 1.440, val: 1.439 | iter time: 359.33 ms (step) remaining time: 0:35:43
|
| 136 |
+
Epoch 1 | iter 1056 step 33 | loss train: 1.495, val: 1.439 | iter time: 360.16 ms (step) remaining time: 0:35:32
|
| 137 |
+
Epoch 1 | iter 1088 step 34 | loss train: 1.481, val: 1.439 | iter time: 360.33 ms (step) remaining time: 0:35:21
|
| 138 |
+
Epoch 1 | iter 1120 step 35 | loss train: 1.436, val: 1.439 | iter time: 360.18 ms (step) remaining time: 0:35:10
|
| 139 |
+
Epoch 1 | iter 1152 step 36 | loss train: 1.436, val: 1.439 | iter time: 360.03 ms (step) remaining time: 0:34:59
|
| 140 |
+
Epoch 1 | iter 1184 step 37 | loss train: 1.420, val: 1.439 | iter time: 360.55 ms (step) remaining time: 0:34:48
|
| 141 |
+
Epoch 1 | iter 1216 step 38 | loss train: 1.494, val: 1.439 | iter time: 358.50 ms (step) remaining time: 0:34:37
|
| 142 |
+
Epoch 1 | iter 1248 step 39 | loss train: 1.452, val: 1.439 | iter time: 360.03 ms (step) remaining time: 0:34:26
|
| 143 |
+
Epoch 1 | iter 1280 step 40 | loss train: 1.440, val: 1.439 | iter time: 359.54 ms (step) remaining time: 0:34:15
|
| 144 |
+
Epoch 1 | iter 1312 step 41 | loss train: 1.490, val: 1.439 | iter time: 360.37 ms (step) remaining time: 0:34:04
|
| 145 |
+
Epoch 1 | iter 1344 step 42 | loss train: 1.483, val: 1.439 | iter time: 358.71 ms (step) remaining time: 0:33:54
|
| 146 |
+
Epoch 1 | iter 1376 step 43 | loss train: 1.516, val: 1.439 | iter time: 360.13 ms (step) remaining time: 0:33:43
|
| 147 |
+
Epoch 1 | iter 1408 step 44 | loss train: 1.404, val: 1.439 | iter time: 359.22 ms (step) remaining time: 0:33:33
|
| 148 |
+
Epoch 1 | iter 1440 step 45 | loss train: 1.449, val: 1.439 | iter time: 359.40 ms (step) remaining time: 0:33:22
|
| 149 |
+
Epoch 1 | iter 1472 step 46 | loss train: 1.486, val: 1.439 | iter time: 358.81 ms (step) remaining time: 0:33:11
|
| 150 |
+
Epoch 1 | iter 1504 step 47 | loss train: 1.552, val: 1.439 | iter time: 360.34 ms (step) remaining time: 0:33:00
|
| 151 |
+
Epoch 1 | iter 1536 step 48 | loss train: 1.391, val: 1.439 | iter time: 358.45 ms (step) remaining time: 0:32:49
|
| 152 |
+
Epoch 1 | iter 1568 step 49 | loss train: 1.471, val: 1.439 | iter time: 360.69 ms (step) remaining time: 0:32:38
|
| 153 |
+
Epoch 1 | iter 1600 step 50 | loss train: 1.493, val: 1.439 | iter time: 361.05 ms (step) remaining time: 0:32:27
|
| 154 |
+
Validating ...
|
| 155 |
+
iter 1600: val loss 1.4279, val time: 22370.93 ms
|
| 156 |
+
Epoch 1 | iter 1632 step 51 | loss train: 1.459, val: 1.428 | iter time: 359.62 ms (step) remaining time: 0:33:35
|
| 157 |
+
Epoch 1 | iter 1664 step 52 | loss train: 1.479, val: 1.428 | iter time: 357.51 ms (step) remaining time: 0:33:22
|
| 158 |
+
Epoch 1 | iter 1696 step 53 | loss train: 1.395, val: 1.428 | iter time: 358.65 ms (step) remaining time: 0:33:09
|
| 159 |
+
Epoch 1 | iter 1728 step 54 | loss train: 1.450, val: 1.428 | iter time: 360.68 ms (step) remaining time: 0:32:56
|
| 160 |
+
Epoch 1 | iter 1760 step 55 | loss train: 1.430, val: 1.428 | iter time: 360.09 ms (step) remaining time: 0:32:44
|
| 161 |
+
Epoch 1 | iter 1792 step 56 | loss train: 1.498, val: 1.428 | iter time: 360.17 ms (step) remaining time: 0:32:31
|
| 162 |
+
Epoch 1 | iter 1824 step 57 | loss train: 1.504, val: 1.428 | iter time: 360.66 ms (step) remaining time: 0:32:19
|
| 163 |
+
Epoch 1 | iter 1856 step 58 | loss train: 1.457, val: 1.428 | iter time: 360.74 ms (step) remaining time: 0:32:06
|
| 164 |
+
Epoch 1 | iter 1888 step 59 | loss train: 1.462, val: 1.428 | iter time: 361.99 ms (step) remaining time: 0:31:54
|
| 165 |
+
Epoch 1 | iter 1920 step 60 | loss train: 1.434, val: 1.428 | iter time: 360.23 ms (step) remaining time: 0:31:41
|
| 166 |
+
Epoch 1 | iter 1952 step 61 | loss train: 1.400, val: 1.428 | iter time: 360.74 ms (step) remaining time: 0:31:29
|
| 167 |
+
Epoch 1 | iter 1984 step 62 | loss train: 1.460, val: 1.428 | iter time: 360.52 ms (step) remaining time: 0:31:17
|
| 168 |
+
Epoch 1 | iter 2016 step 63 | loss train: 1.443, val: 1.428 | iter time: 360.72 ms (step) remaining time: 0:31:04
|
| 169 |
+
Epoch 1 | iter 2048 step 64 | loss train: 1.438, val: 1.428 | iter time: 359.46 ms (step) remaining time: 0:30:52
|
| 170 |
+
Epoch 1 | iter 2080 step 65 | loss train: 1.438, val: 1.428 | iter time: 359.72 ms (step) remaining time: 0:30:40
|
| 171 |
+
Epoch 1 | iter 2112 step 66 | loss train: 1.476, val: 1.428 | iter time: 361.45 ms (step) remaining time: 0:30:28
|
| 172 |
+
Epoch 1 | iter 2144 step 67 | loss train: 1.395, val: 1.428 | iter time: 360.85 ms (step) remaining time: 0:30:16
|
| 173 |
+
Epoch 1 | iter 2176 step 68 | loss train: 1.451, val: 1.428 | iter time: 359.96 ms (step) remaining time: 0:30:04
|
| 174 |
+
Epoch 1 | iter 2208 step 69 | loss train: 1.467, val: 1.428 | iter time: 358.44 ms (step) remaining time: 0:29:52
|
| 175 |
+
Epoch 1 | iter 2240 step 70 | loss train: 1.470, val: 1.428 | iter time: 360.59 ms (step) remaining time: 0:29:40
|
| 176 |
+
Epoch 1 | iter 2272 step 71 | loss train: 1.403, val: 1.428 | iter time: 358.49 ms (step) remaining time: 0:29:28
|
| 177 |
+
Epoch 1 | iter 2304 step 72 | loss train: 1.482, val: 1.428 | iter time: 359.42 ms (step) remaining time: 0:29:16
|
| 178 |
+
Epoch 1 | iter 2336 step 73 | loss train: 1.478, val: 1.428 | iter time: 360.80 ms (step) remaining time: 0:29:04
|
| 179 |
+
Epoch 1 | iter 2368 step 74 | loss train: 1.512, val: 1.428 | iter time: 359.14 ms (step) remaining time: 0:28:52
|
| 180 |
+
Epoch 1 | iter 2400 step 75 | loss train: 1.437, val: 1.428 | iter time: 360.57 ms (step) remaining time: 0:28:40
|
| 181 |
+
Epoch 1 | iter 2432 step 76 | loss train: 1.382, val: 1.428 | iter time: 362.33 ms (step) remaining time: 0:28:29
|
| 182 |
+
Epoch 1 | iter 2464 step 77 | loss train: 1.438, val: 1.428 | iter time: 359.09 ms (step) remaining time: 0:28:17
|
| 183 |
+
Epoch 1 | iter 2496 step 78 | loss train: 1.485, val: 1.428 | iter time: 361.08 ms (step) remaining time: 0:28:05
|
| 184 |
+
Epoch 1 | iter 2528 step 79 | loss train: 1.483, val: 1.428 | iter time: 358.16 ms (step) remaining time: 0:27:53
|
| 185 |
+
Epoch 1 | iter 2560 step 80 | loss train: 1.489, val: 1.428 | iter time: 360.06 ms (step) remaining time: 0:27:42
|
| 186 |
+
Epoch 1 | iter 2592 step 81 | loss train: 1.389, val: 1.428 | iter time: 359.91 ms (step) remaining time: 0:27:30
|
| 187 |
+
Epoch 1 | iter 2624 step 82 | loss train: 1.435, val: 1.428 | iter time: 359.27 ms (step) remaining time: 0:27:18
|
| 188 |
+
Epoch 1 | iter 2656 step 83 | loss train: 1.439, val: 1.428 | iter time: 360.49 ms (step) remaining time: 0:27:07
|
| 189 |
+
Epoch 1 | iter 2688 step 84 | loss train: 1.498, val: 1.428 | iter time: 361.39 ms (step) remaining time: 0:26:55
|
| 190 |
+
Epoch 1 | iter 2720 step 85 | loss train: 1.384, val: 1.428 | iter time: 358.58 ms (step) remaining time: 0:26:43
|
| 191 |
+
Epoch 1 | iter 2752 step 86 | loss train: 1.432, val: 1.428 | iter time: 358.85 ms (step) remaining time: 0:26:32
|
| 192 |
+
Epoch 1 | iter 2784 step 87 | loss train: 1.430, val: 1.428 | iter time: 360.26 ms (step) remaining time: 0:26:20
|
| 193 |
+
Epoch 1 | iter 2816 step 88 | loss train: 1.488, val: 1.428 | iter time: 359.59 ms (step) remaining time: 0:26:09
|
| 194 |
+
Epoch 1 | iter 2848 step 89 | loss train: 1.408, val: 1.428 | iter time: 359.99 ms (step) remaining time: 0:25:57
|
| 195 |
+
Epoch 1 | iter 2880 step 90 | loss train: 1.478, val: 1.428 | iter time: 360.41 ms (step) remaining time: 0:25:46
|
| 196 |
+
Epoch 1 | iter 2912 step 91 | loss train: 1.427, val: 1.428 | iter time: 359.02 ms (step) remaining time: 0:25:34
|
| 197 |
+
Epoch 1 | iter 2944 step 92 | loss train: 1.398, val: 1.428 | iter time: 359.95 ms (step) remaining time: 0:25:22
|
| 198 |
+
Epoch 1 | iter 2976 step 93 | loss train: 1.528, val: 1.428 | iter time: 359.32 ms (step) remaining time: 0:25:11
|
| 199 |
+
Epoch 1 | iter 3008 step 94 | loss train: 1.456, val: 1.428 | iter time: 358.77 ms (step) remaining time: 0:25:00
|
| 200 |
+
Epoch 1 | iter 3040 step 95 | loss train: 1.488, val: 1.428 | iter time: 361.67 ms (step) remaining time: 0:24:48
|
| 201 |
+
Epoch 1 | iter 3072 step 96 | loss train: 1.486, val: 1.428 | iter time: 358.49 ms (step) remaining time: 0:24:37
|
| 202 |
+
Epoch 1 | iter 3104 step 97 | loss train: 1.414, val: 1.428 | iter time: 362.04 ms (step) remaining time: 0:24:25
|
| 203 |
+
Epoch 1 | iter 3136 step 98 | loss train: 1.391, val: 1.428 | iter time: 358.65 ms (step) remaining time: 0:24:14
|
| 204 |
+
Epoch 1 | iter 3168 step 99 | loss train: 1.472, val: 1.428 | iter time: 359.13 ms (step) remaining time: 0:24:02
|
| 205 |
+
Epoch 1 | iter 3200 step 100 | loss train: 1.409, val: 1.428 | iter time: 360.12 ms (step) remaining time: 0:23:51
|
| 206 |
+
Validating ...
|
| 207 |
+
iter 3200: val loss 1.3742, val time: 22354.24 ms
|
| 208 |
+
Saving checkpoint to 'out/pretrain/2409_full/step-00000100/lit_model.pth'
|
| 209 |
+
Epoch 1 | iter 3232 step 101 | loss train: 1.449, val: 1.374 | iter time: 357.59 ms (step) remaining time: 0:24:29
|
| 210 |
+
Epoch 1 | iter 3264 step 102 | loss train: 1.449, val: 1.374 | iter time: 357.88 ms (step) remaining time: 0:24:17
|
| 211 |
+
Epoch 1 | iter 3296 step 103 | loss train: 1.492, val: 1.374 | iter time: 359.03 ms (step) remaining time: 0:24:05
|
| 212 |
+
Epoch 1 | iter 3328 step 104 | loss train: 1.441, val: 1.374 | iter time: 359.20 ms (step) remaining time: 0:23:53
|
| 213 |
+
Epoch 1 | iter 3360 step 105 | loss train: 1.439, val: 1.374 | iter time: 359.20 ms (step) remaining time: 0:23:40
|
| 214 |
+
Epoch 1 | iter 3392 step 106 | loss train: 1.409, val: 1.374 | iter time: 359.27 ms (step) remaining time: 0:23:28
|
| 215 |
+
Epoch 1 | iter 3424 step 107 | loss train: 1.403, val: 1.374 | iter time: 360.41 ms (step) remaining time: 0:23:16
|
| 216 |
+
Epoch 1 | iter 3456 step 108 | loss train: 1.461, val: 1.374 | iter time: 359.41 ms (step) remaining time: 0:23:04
|
| 217 |
+
Epoch 1 | iter 3488 step 109 | loss train: 1.417, val: 1.374 | iter time: 360.49 ms (step) remaining time: 0:22:52
|
| 218 |
+
Epoch 1 | iter 3520 step 110 | loss train: 1.439, val: 1.374 | iter time: 359.24 ms (step) remaining time: 0:22:40
|
| 219 |
+
Epoch 1 | iter 3552 step 111 | loss train: 1.391, val: 1.374 | iter time: 358.84 ms (step) remaining time: 0:22:28
|
| 220 |
+
Epoch 1 | iter 3584 step 112 | loss train: 1.442, val: 1.374 | iter time: 357.94 ms (step) remaining time: 0:22:16
|
| 221 |
+
Epoch 1 | iter 3616 step 113 | loss train: 1.433, val: 1.374 | iter time: 359.61 ms (step) remaining time: 0:22:04
|
| 222 |
+
Epoch 1 | iter 3648 step 114 | loss train: 1.371, val: 1.374 | iter time: 361.47 ms (step) remaining time: 0:21:52
|
| 223 |
+
Epoch 1 | iter 3680 step 115 | loss train: 1.525, val: 1.374 | iter time: 359.78 ms (step) remaining time: 0:21:40
|
| 224 |
+
Epoch 1 | iter 3712 step 116 | loss train: 1.445, val: 1.374 | iter time: 360.91 ms (step) remaining time: 0:21:28
|
| 225 |
+
Epoch 1 | iter 3744 step 117 | loss train: 1.500, val: 1.374 | iter time: 360.49 ms (step) remaining time: 0:21:17
|
| 226 |
+
Epoch 1 | iter 3776 step 118 | loss train: 1.409, val: 1.374 | iter time: 360.23 ms (step) remaining time: 0:21:05
|
| 227 |
+
Epoch 1 | iter 3808 step 119 | loss train: 1.398, val: 1.374 | iter time: 358.15 ms (step) remaining time: 0:20:53
|
| 228 |
+
Epoch 1 | iter 3840 step 120 | loss train: 1.438, val: 1.374 | iter time: 357.77 ms (step) remaining time: 0:20:41
|
| 229 |
+
Epoch 1 | iter 3872 step 121 | loss train: 1.486, val: 1.374 | iter time: 362.11 ms (step) remaining time: 0:20:29
|
| 230 |
+
Epoch 1 | iter 3904 step 122 | loss train: 1.431, val: 1.374 | iter time: 361.26 ms (step) remaining time: 0:20:17
|
| 231 |
+
Epoch 1 | iter 3936 step 123 | loss train: 1.366, val: 1.374 | iter time: 360.88 ms (step) remaining time: 0:20:05
|
| 232 |
+
Epoch 1 | iter 3968 step 124 | loss train: 1.379, val: 1.374 | iter time: 359.94 ms (step) remaining time: 0:19:53
|
| 233 |
+
Epoch 1 | iter 4000 step 125 | loss train: 1.492, val: 1.374 | iter time: 359.95 ms (step) remaining time: 0:19:42
|
| 234 |
+
Epoch 1 | iter 4032 step 126 | loss train: 1.510, val: 1.374 | iter time: 360.46 ms (step) remaining time: 0:19:30
|
| 235 |
+
Epoch 1 | iter 4064 step 127 | loss train: 1.388, val: 1.374 | iter time: 359.83 ms (step) remaining time: 0:19:18
|
| 236 |
+
Epoch 1 | iter 4096 step 128 | loss train: 1.392, val: 1.374 | iter time: 360.20 ms (step) remaining time: 0:19:06
|
| 237 |
+
Epoch 1 | iter 4128 step 129 | loss train: 1.424, val: 1.374 | iter time: 360.62 ms (step) remaining time: 0:18:55
|
| 238 |
+
Epoch 1 | iter 4160 step 130 | loss train: 1.378, val: 1.374 | iter time: 359.19 ms (step) remaining time: 0:18:43
|
| 239 |
+
Epoch 1 | iter 4192 step 131 | loss train: 1.381, val: 1.374 | iter time: 360.42 ms (step) remaining time: 0:18:31
|
| 240 |
+
Epoch 1 | iter 4224 step 132 | loss train: 1.448, val: 1.374 | iter time: 357.72 ms (step) remaining time: 0:18:20
|
| 241 |
+
Epoch 1 | iter 4256 step 133 | loss train: 1.352, val: 1.374 | iter time: 360.27 ms (step) remaining time: 0:18:08
|
| 242 |
+
Epoch 1 | iter 4288 step 134 | loss train: 1.428, val: 1.374 | iter time: 357.85 ms (step) remaining time: 0:17:56
|
| 243 |
+
Epoch 1 | iter 4320 step 135 | loss train: 1.455, val: 1.374 | iter time: 360.12 ms (step) remaining time: 0:17:44
|
| 244 |
+
Epoch 1 | iter 4352 step 136 | loss train: 1.404, val: 1.374 | iter time: 358.96 ms (step) remaining time: 0:17:33
|
| 245 |
+
Epoch 1 | iter 4384 step 137 | loss train: 1.408, val: 1.374 | iter time: 360.78 ms (step) remaining time: 0:17:21
|
| 246 |
+
Epoch 1 | iter 4416 step 138 | loss train: 1.399, val: 1.374 | iter time: 359.31 ms (step) remaining time: 0:17:10
|
| 247 |
+
Epoch 1 | iter 4448 step 139 | loss train: 1.377, val: 1.374 | iter time: 360.01 ms (step) remaining time: 0:16:58
|
| 248 |
+
Epoch 1 | iter 4480 step 140 | loss train: 1.430, val: 1.374 | iter time: 361.65 ms (step) remaining time: 0:16:46
|
| 249 |
+
Epoch 1 | iter 4512 step 141 | loss train: 1.430, val: 1.374 | iter time: 359.36 ms (step) remaining time: 0:16:35
|
| 250 |
+
Epoch 1 | iter 4544 step 142 | loss train: 1.393, val: 1.374 | iter time: 359.19 ms (step) remaining time: 0:16:23
|
| 251 |
+
Epoch 1 | iter 4576 step 143 | loss train: 1.414, val: 1.374 | iter time: 361.11 ms (step) remaining time: 0:16:12
|
| 252 |
+
Epoch 1 | iter 4608 step 144 | loss train: 1.416, val: 1.374 | iter time: 359.63 ms (step) remaining time: 0:16:00
|
| 253 |
+
Epoch 1 | iter 4640 step 145 | loss train: 1.502, val: 1.374 | iter time: 360.45 ms (step) remaining time: 0:15:48
|
| 254 |
+
Epoch 1 | iter 4672 step 146 | loss train: 1.401, val: 1.374 | iter time: 360.49 ms (step) remaining time: 0:15:37
|
| 255 |
+
Epoch 1 | iter 4704 step 147 | loss train: 1.426, val: 1.374 | iter time: 359.62 ms (step) remaining time: 0:15:25
|
| 256 |
+
Epoch 1 | iter 4736 step 148 | loss train: 1.469, val: 1.374 | iter time: 359.37 ms (step) remaining time: 0:15:14
|
| 257 |
+
Epoch 1 | iter 4768 step 149 | loss train: 1.379, val: 1.374 | iter time: 361.42 ms (step) remaining time: 0:15:02
|
| 258 |
+
Epoch 1 | iter 4800 step 150 | loss train: 1.434, val: 1.374 | iter time: 360.99 ms (step) remaining time: 0:14:51
|
| 259 |
+
Validating ...
|
| 260 |
+
iter 4800: val loss 1.3083, val time: 22361.91 ms
|
| 261 |
+
Epoch 1 | iter 4832 step 151 | loss train: 1.320, val: 1.308 | iter time: 360.14 ms (step) remaining time: 0:14:51
|
| 262 |
+
Epoch 1 | iter 4864 step 152 | loss train: 1.411, val: 1.308 | iter time: 359.02 ms (step) remaining time: 0:14:39
|
| 263 |
+
Epoch 1 | iter 4896 step 153 | loss train: 1.337, val: 1.308 | iter time: 360.17 ms (step) remaining time: 0:14:28
|
| 264 |
+
Epoch 1 | iter 4928 step 154 | loss train: 1.460, val: 1.308 | iter time: 360.78 ms (step) remaining time: 0:14:16
|
| 265 |
+
Epoch 1 | iter 4960 step 155 | loss train: 1.398, val: 1.308 | iter time: 360.32 ms (step) remaining time: 0:14:04
|
| 266 |
+
Epoch 1 | iter 4992 step 156 | loss train: 1.371, val: 1.308 | iter time: 361.74 ms (step) remaining time: 0:13:53
|
| 267 |
+
Epoch 1 | iter 5024 step 157 | loss train: 1.401, val: 1.308 | iter time: 361.55 ms (step) remaining time: 0:13:41
|
| 268 |
+
Epoch 1 | iter 5056 step 158 | loss train: 1.346, val: 1.308 | iter time: 361.36 ms (step) remaining time: 0:13:29
|
| 269 |
+
Epoch 1 | iter 5088 step 159 | loss train: 1.330, val: 1.308 | iter time: 359.34 ms (step) remaining time: 0:13:18
|
| 270 |
+
Epoch 1 | iter 5120 step 160 | loss train: 1.384, val: 1.308 | iter time: 357.55 ms (step) remaining time: 0:13:06
|
| 271 |
+
Epoch 1 | iter 5152 step 161 | loss train: 1.376, val: 1.308 | iter time: 359.88 ms (step) remaining time: 0:12:54
|
| 272 |
+
Epoch 1 | iter 5184 step 162 | loss train: 1.365, val: 1.308 | iter time: 359.41 ms (step) remaining time: 0:12:43
|
| 273 |
+
Epoch 1 | iter 5216 step 163 | loss train: 1.375, val: 1.308 | iter time: 359.51 ms (step) remaining time: 0:12:31
|
| 274 |
+
Epoch 1 | iter 5248 step 164 | loss train: 1.334, val: 1.308 | iter time: 360.30 ms (step) remaining time: 0:12:20
|
| 275 |
+
Epoch 1 | iter 5280 step 165 | loss train: 1.400, val: 1.308 | iter time: 361.56 ms (step) remaining time: 0:12:08
|
| 276 |
+
Epoch 1 | iter 5312 step 166 | loss train: 1.362, val: 1.308 | iter time: 362.95 ms (step) remaining time: 0:11:56
|
| 277 |
+
Epoch 1 | iter 5344 step 167 | loss train: 1.384, val: 1.308 | iter time: 360.51 ms (step) remaining time: 0:11:45
|
| 278 |
+
Epoch 1 | iter 5376 step 168 | loss train: 1.360, val: 1.308 | iter time: 360.44 ms (step) remaining time: 0:11:33
|
| 279 |
+
Epoch 1 | iter 5408 step 169 | loss train: 1.392, val: 1.308 | iter time: 360.19 ms (step) remaining time: 0:11:22
|
| 280 |
+
Epoch 1 | iter 5440 step 170 | loss train: 1.446, val: 1.308 | iter time: 359.28 ms (step) remaining time: 0:11:10
|
| 281 |
+
Epoch 1 | iter 5472 step 171 | loss train: 1.316, val: 1.308 | iter time: 359.45 ms (step) remaining time: 0:10:59
|
| 282 |
+
Epoch 1 | iter 5504 step 172 | loss train: 1.331, val: 1.308 | iter time: 360.01 ms (step) remaining time: 0:10:47
|
| 283 |
+
Epoch 1 | iter 5536 step 173 | loss train: 1.437, val: 1.308 | iter time: 358.74 ms (step) remaining time: 0:10:36
|
| 284 |
+
Epoch 1 | iter 5568 step 174 | loss train: 1.322, val: 1.308 | iter time: 359.81 ms (step) remaining time: 0:10:24
|
| 285 |
+
Epoch 1 | iter 5600 step 175 | loss train: 1.387, val: 1.308 | iter time: 361.16 ms (step) remaining time: 0:10:13
|
| 286 |
+
Epoch 1 | iter 5632 step 176 | loss train: 1.407, val: 1.308 | iter time: 360.28 ms (step) remaining time: 0:10:01
|
| 287 |
+
Epoch 1 | iter 5664 step 177 | loss train: 1.454, val: 1.308 | iter time: 359.64 ms (step) remaining time: 0:09:50
|
| 288 |
+
Epoch 1 | iter 5696 step 178 | loss train: 1.386, val: 1.308 | iter time: 359.70 ms (step) remaining time: 0:09:38
|
| 289 |
+
Epoch 1 | iter 5728 step 179 | loss train: 1.326, val: 1.308 | iter time: 362.08 ms (step) remaining time: 0:09:27
|
| 290 |
+
Epoch 1 | iter 5760 step 180 | loss train: 1.391, val: 1.308 | iter time: 359.35 ms (step) remaining time: 0:09:15
|
| 291 |
+
Epoch 1 | iter 5792 step 181 | loss train: 1.385, val: 1.308 | iter time: 361.94 ms (step) remaining time: 0:09:04
|
| 292 |
+
Epoch 1 | iter 5824 step 182 | loss train: 1.405, val: 1.308 | iter time: 359.97 ms (step) remaining time: 0:08:52
|
| 293 |
+
Epoch 1 | iter 5856 step 183 | loss train: 1.361, val: 1.308 | iter time: 360.45 ms (step) remaining time: 0:08:41
|
| 294 |
+
Epoch 1 | iter 5888 step 184 | loss train: 1.386, val: 1.308 | iter time: 358.21 ms (step) remaining time: 0:08:30
|
| 295 |
+
Epoch 1 | iter 5920 step 185 | loss train: 1.321, val: 1.308 | iter time: 359.45 ms (step) remaining time: 0:08:18
|
| 296 |
+
Epoch 1 | iter 5952 step 186 | loss train: 1.348, val: 1.308 | iter time: 361.57 ms (step) remaining time: 0:08:07
|
| 297 |
+
Epoch 1 | iter 5984 step 187 | loss train: 1.352, val: 1.308 | iter time: 359.83 ms (step) remaining time: 0:07:55
|
| 298 |
+
Epoch 1 | iter 6016 step 188 | loss train: 1.381, val: 1.308 | iter time: 360.83 ms (step) remaining time: 0:07:44
|
| 299 |
+
Epoch 1 | iter 6048 step 189 | loss train: 1.380, val: 1.308 | iter time: 359.87 ms (step) remaining time: 0:07:32
|
| 300 |
+
Epoch 1 | iter 6080 step 190 | loss train: 1.356, val: 1.308 | iter time: 359.57 ms (step) remaining time: 0:07:21
|
| 301 |
+
Epoch 1 | iter 6112 step 191 | loss train: 1.357, val: 1.308 | iter time: 360.99 ms (step) remaining time: 0:07:10
|
| 302 |
+
Epoch 1 | iter 6144 step 192 | loss train: 1.382, val: 1.308 | iter time: 359.41 ms (step) remaining time: 0:06:58
|
| 303 |
+
Epoch 1 | iter 6176 step 193 | loss train: 1.419, val: 1.308 | iter time: 362.76 ms (step) remaining time: 0:06:47
|
| 304 |
+
Epoch 1 | iter 6208 step 194 | loss train: 1.396, val: 1.308 | iter time: 357.99 ms (step) remaining time: 0:06:35
|
| 305 |
+
Epoch 1 | iter 6240 step 195 | loss train: 1.379, val: 1.308 | iter time: 360.59 ms (step) remaining time: 0:06:24
|
| 306 |
+
Epoch 1 | iter 6272 step 196 | loss train: 1.303, val: 1.308 | iter time: 360.35 ms (step) remaining time: 0:06:13
|
| 307 |
+
Epoch 1 | iter 6304 step 197 | loss train: 1.319, val: 1.308 | iter time: 361.73 ms (step) remaining time: 0:06:01
|
| 308 |
+
Epoch 1 | iter 6336 step 198 | loss train: 1.363, val: 1.308 | iter time: 361.68 ms (step) remaining time: 0:05:50
|
| 309 |
+
Epoch 1 | iter 6368 step 199 | loss train: 1.414, val: 1.308 | iter time: 360.85 ms (step) remaining time: 0:05:38
|
| 310 |
+
Epoch 1 | iter 6400 step 200 | loss train: 1.426, val: 1.308 | iter time: 359.01 ms (step) remaining time: 0:05:27
|
| 311 |
+
Validating ...
|
| 312 |
+
iter 6400: val loss 1.2540, val time: 22363.57 ms
|
| 313 |
+
Saving checkpoint to 'out/pretrain/2409_full/step-00000200/lit_model.pth'
|
| 314 |
+
Epoch 1 | iter 6432 step 201 | loss train: 1.349, val: 1.254 | iter time: 357.41 ms (step) remaining time: 0:05:21
|
| 315 |
+
Epoch 1 | iter 6464 step 202 | loss train: 1.325, val: 1.254 | iter time: 356.63 ms (step) remaining time: 0:05:10
|
| 316 |
+
Epoch 1 | iter 6496 step 203 | loss train: 1.364, val: 1.254 | iter time: 360.22 ms (step) remaining time: 0:04:58
|
| 317 |
+
Epoch 1 | iter 6528 step 204 | loss train: 1.339, val: 1.254 | iter time: 360.66 ms (step) remaining time: 0:04:47
|
| 318 |
+
Epoch 1 | iter 6560 step 205 | loss train: 1.444, val: 1.254 | iter time: 359.77 ms (step) remaining time: 0:04:35
|
| 319 |
+
Epoch 1 | iter 6592 step 206 | loss train: 1.366, val: 1.254 | iter time: 358.90 ms (step) remaining time: 0:04:23
|
| 320 |
+
Epoch 1 | iter 6624 step 207 | loss train: 1.327, val: 1.254 | iter time: 360.62 ms (step) remaining time: 0:04:12
|
| 321 |
+
Epoch 1 | iter 6656 step 208 | loss train: 1.382, val: 1.254 | iter time: 359.44 ms (step) remaining time: 0:04:00
|
| 322 |
+
Epoch 1 | iter 6688 step 209 | loss train: 1.393, val: 1.254 | iter time: 360.20 ms (step) remaining time: 0:03:49
|
| 323 |
+
Epoch 1 | iter 6720 step 210 | loss train: 1.318, val: 1.254 | iter time: 360.84 ms (step) remaining time: 0:03:37
|
| 324 |
+
Epoch 1 | iter 6752 step 211 | loss train: 1.447, val: 1.254 | iter time: 358.12 ms (step) remaining time: 0:03:26
|
| 325 |
+
Epoch 1 | iter 6784 step 212 | loss train: 1.398, val: 1.254 | iter time: 359.93 ms (step) remaining time: 0:03:14
|
| 326 |
+
Epoch 1 | iter 6816 step 213 | loss train: 1.424, val: 1.254 | iter time: 361.55 ms (step) remaining time: 0:03:03
|
| 327 |
+
Epoch 1 | iter 6848 step 214 | loss train: 1.363, val: 1.254 | iter time: 358.52 ms (step) remaining time: 0:02:51
|
| 328 |
+
Epoch 1 | iter 6880 step 215 | loss train: 1.384, val: 1.254 | iter time: 359.94 ms (step) remaining time: 0:02:40
|
| 329 |
+
Epoch 1 | iter 6912 step 216 | loss train: 1.361, val: 1.254 | iter time: 360.33 ms (step) remaining time: 0:02:28
|
| 330 |
+
Epoch 1 | iter 6944 step 217 | loss train: 1.370, val: 1.254 | iter time: 359.63 ms (step) remaining time: 0:02:17
|
| 331 |
+
Epoch 1 | iter 6976 step 218 | loss train: 1.361, val: 1.254 | iter time: 359.02 ms (step) remaining time: 0:02:05
|
| 332 |
+
Epoch 1 | iter 7008 step 219 | loss train: 1.324, val: 1.254 | iter time: 359.74 ms (step) remaining time: 0:01:54
|
| 333 |
+
Epoch 1 | iter 7040 step 220 | loss train: 1.320, val: 1.254 | iter time: 359.73 ms (step) remaining time: 0:01:42
|
| 334 |
+
Epoch 1 | iter 7072 step 221 | loss train: 1.405, val: 1.254 | iter time: 359.86 ms (step) remaining time: 0:01:31
|
| 335 |
+
Epoch 1 | iter 7104 step 222 | loss train: 1.370, val: 1.254 | iter time: 360.54 ms (step) remaining time: 0:01:20
|
| 336 |
+
Epoch 1 | iter 7136 step 223 | loss train: 1.352, val: 1.254 | iter time: 360.82 ms (step) remaining time: 0:01:08
|
| 337 |
+
Epoch 1 | iter 7168 step 224 | loss train: 1.415, val: 1.254 | iter time: 360.62 ms (step) remaining time: 0:00:57
|
| 338 |
+
Epoch 1 | iter 7200 step 225 | loss train: 1.401, val: 1.254 | iter time: 358.78 ms (step) remaining time: 0:00:45
|
| 339 |
+
Epoch 1 | iter 7232 step 226 | loss train: 1.302, val: 1.254 | iter time: 360.02 ms (step) remaining time: 0:00:34
|
| 340 |
+
Epoch 1 | iter 7264 step 227 | loss train: 1.352, val: 1.254 | iter time: 369.07 ms (step) remaining time: 0:00:22
|
| 341 |
+
Epoch 1 | iter 7296 step 228 | loss train: 1.362, val: 1.254 | iter time: 360.42 ms (step) remaining time: 0:00:11
|
| 342 |
+
Epoch 2 | iter 7328 step 229 | loss train: 1.286, val: 1.254 | iter time: 361.71 ms (step) remaining time: 0:00:00
|
| 343 |
+
Validating ...
|
| 344 |
+
Final evaluation | val loss: 1.232 | val ppl: 3.430
|
| 345 |
+
Saving checkpoint to 'out/pretrain/2409_full/final/lit_model.pth'
|
| 346 |
+
----------------------------------------
|
| 347 |
+
| Performance
|
| 348 |
+
| - Total tokens : 240,123,904
|
| 349 |
+
| - Training Time : 2679.88 s
|
| 350 |
+
| - Tok/sec : 129.83 tok/s
|
| 351 |
+
| ----------------------------------------
|
| 352 |
+
| Memory Usage
|
| 353 |
+
| - Memory Used : 26.32 GB
|
| 354 |
+
----------------------------------------
|
out/pretrain/tinyllama/teelogs/2409_lr4e-5.txt
ADDED
|
@@ -0,0 +1,361 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
Initializing distributed: GLOBAL_RANK: 0, MEMBER: 1/4
|
| 2 |
+
Initializing distributed: GLOBAL_RANK: 2, MEMBER: 3/4
|
| 3 |
+
[rank: 2] Seed set to 42
|
| 4 |
+
Initializing distributed: GLOBAL_RANK: 3, MEMBER: 4/4
|
| 5 |
+
[rank: 3] Seed set to 42
|
| 6 |
+
Initializing distributed: GLOBAL_RANK: 1, MEMBER: 2/4
|
| 7 |
+
----------------------------------------------------------------------------------------------------
|
| 8 |
+
distributed_backend=nccl
|
| 9 |
+
All distributed processes registered. Starting with 4 processes
|
| 10 |
+
----------------------------------------------------------------------------------------------------
|
| 11 |
+
|
| 12 |
+
[rank: 1] Seed set to 42
|
| 13 |
+
[rank: 0] Seed set to 42
|
| 14 |
+
/mnt/data/litgpt/litgpt/pretrain.py:515: UserWarning: A newer version of litdata is available (0.2.52). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 15 |
+
train_dataloader = data.train_dataloader()
|
| 16 |
+
/mnt/data/litgpt/litgpt/pretrain.py:515: UserWarning: A newer version of litdata is available (0.2.52). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 17 |
+
train_dataloader = data.train_dataloader()
|
| 18 |
+
/mnt/data/litgpt/litgpt/pretrain.py:515: UserWarning: A newer version of litdata is available (0.2.52). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 19 |
+
train_dataloader = data.train_dataloader()
|
| 20 |
+
/mnt/data/litgpt/litgpt/pretrain.py:515: UserWarning: A newer version of litdata is available (0.2.52). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 21 |
+
train_dataloader = data.train_dataloader()
|
| 22 |
+
All GPUs are fully connected via NVLink.
|
| 23 |
+
{'data': {'batch_size': 1,
|
| 24 |
+
'data_path': PosixPath('litgpt/data/arxiv'),
|
| 25 |
+
'num_workers': 0,
|
| 26 |
+
'seed': 42,
|
| 27 |
+
'seq_length': 2048,
|
| 28 |
+
'use_starcoder': True},
|
| 29 |
+
'data_dir': PosixPath('litgpt/data/arxiv/2409'),
|
| 30 |
+
'devices': 'auto',
|
| 31 |
+
'eval': {'evaluate_example': 'first',
|
| 32 |
+
'final_validation': True,
|
| 33 |
+
'initial_validation': True,
|
| 34 |
+
'interval': 50,
|
| 35 |
+
'max_iters': 200,
|
| 36 |
+
'max_new_tokens': None},
|
| 37 |
+
'initial_checkpoint_dir': PosixPath('out/pretrain/tinyllama/2408_full/final'),
|
| 38 |
+
'log': {'group': None, 'project': None, 'run': None},
|
| 39 |
+
'logger_name': 'tensorboard',
|
| 40 |
+
'model_config': {'attention_logit_softcapping': None,
|
| 41 |
+
'attention_scores_scalar': None,
|
| 42 |
+
'attn_bias': False,
|
| 43 |
+
'bias': False,
|
| 44 |
+
'block_size': 2048,
|
| 45 |
+
'final_logit_softcapping': None,
|
| 46 |
+
'gelu_approximate': 'none',
|
| 47 |
+
'head_size': 64,
|
| 48 |
+
'hf_config': {'name': 'TinyLlama-1.1B-intermediate-step-1431k-3T',
|
| 49 |
+
'org': 'TinyLlama'},
|
| 50 |
+
'intermediate_size': 5632,
|
| 51 |
+
'lm_head_bias': False,
|
| 52 |
+
'mlp_class_name': 'LLaMAMLP',
|
| 53 |
+
'moe_intermediate_size': None,
|
| 54 |
+
'n_embd': 2048,
|
| 55 |
+
'n_expert': 0,
|
| 56 |
+
'n_expert_per_token': 0,
|
| 57 |
+
'n_head': 32,
|
| 58 |
+
'n_layer': 22,
|
| 59 |
+
'n_query_groups': 4,
|
| 60 |
+
'name': 'tiny-llama-1.1b',
|
| 61 |
+
'norm_1': True,
|
| 62 |
+
'norm_2': True,
|
| 63 |
+
'norm_class_name': 'RMSNorm',
|
| 64 |
+
'norm_eps': 1e-05,
|
| 65 |
+
'norm_qk': False,
|
| 66 |
+
'norm_qk_type': 'default',
|
| 67 |
+
'padded_vocab_size': 32000,
|
| 68 |
+
'padding_multiple': 64,
|
| 69 |
+
'parallel_residual': False,
|
| 70 |
+
'post_attention_norm': False,
|
| 71 |
+
'post_mlp_norm': False,
|
| 72 |
+
'rope_adjustments': None,
|
| 73 |
+
'rope_base': 10000,
|
| 74 |
+
'rope_condense_ratio': 1,
|
| 75 |
+
'rope_indices': None,
|
| 76 |
+
'rope_local_base_freq': None,
|
| 77 |
+
'rotary_percentage': 1.0,
|
| 78 |
+
'scale_embeddings': False,
|
| 79 |
+
'shared_attention_norm': False,
|
| 80 |
+
'sliding_window_indices': None,
|
| 81 |
+
'sliding_window_size': None,
|
| 82 |
+
'vocab_size': 32000},
|
| 83 |
+
'model_name': 'tiny-llama-1.1b',
|
| 84 |
+
'num_nodes': 1,
|
| 85 |
+
'optimizer': "{'class_path': 'torch.optim.AdamW', 'init_args': {'lr': 4e-05, "
|
| 86 |
+
"'weight_decay': 0.1, 'betas': [0.9, 0.95]}}",
|
| 87 |
+
'out_dir': PosixPath('out/pretrain/tinyllama/2409_lr4e-5'),
|
| 88 |
+
'ppl': False,
|
| 89 |
+
'precision': 'bf16-mixed',
|
| 90 |
+
'resume': False,
|
| 91 |
+
'seed': 42,
|
| 92 |
+
'tokenizer_dir': PosixPath('checkpoints/TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T'),
|
| 93 |
+
'train': {'epochs': None,
|
| 94 |
+
'global_batch_size': 512,
|
| 95 |
+
'log_interval': 1,
|
| 96 |
+
'lr_warmup_fraction': None,
|
| 97 |
+
'lr_warmup_steps': 20,
|
| 98 |
+
'max_norm': 1.0,
|
| 99 |
+
'max_seq_length': 2048,
|
| 100 |
+
'max_steps': None,
|
| 101 |
+
'max_tokens': 240123904,
|
| 102 |
+
'micro_batch_size': 4,
|
| 103 |
+
'min_lr': 4e-05,
|
| 104 |
+
'save_interval': 100,
|
| 105 |
+
'tie_embeddings': None}}
|
| 106 |
+
Time to instantiate model: 0.04 seconds.
|
| 107 |
+
Total parameters: 1,100,048,384
|
| 108 |
+
[ok] out/pretrain/tinyllama/2408_full/final/lit_model.pth 已是纯权重
|
| 109 |
+
Validating ...
|
| 110 |
+
Measured TFLOPs: 239.66
|
| 111 |
+
Epoch 1 | iter 32 step 1 | loss train: 1.492, val: 1.476 | iter time: 560.28 ms (step) remaining time: 0:44:23
|
| 112 |
+
Epoch 1 | iter 64 step 2 | loss train: 1.354, val: 1.476 | iter time: 356.78 ms (step) remaining time: 0:42:29
|
| 113 |
+
Epoch 1 | iter 96 step 3 | loss train: 1.430, val: 1.476 | iter time: 358.40 ms (step) remaining time: 0:41:52
|
| 114 |
+
Epoch 1 | iter 128 step 4 | loss train: 1.396, val: 1.476 | iter time: 358.67 ms (step) remaining time: 0:41:24
|
| 115 |
+
Epoch 1 | iter 160 step 5 | loss train: 1.335, val: 1.476 | iter time: 358.23 ms (step) remaining time: 0:41:03
|
| 116 |
+
Epoch 1 | iter 192 step 6 | loss train: 1.487, val: 1.476 | iter time: 358.89 ms (step) remaining time: 0:40:47
|
| 117 |
+
Epoch 1 | iter 224 step 7 | loss train: 1.408, val: 1.476 | iter time: 358.29 ms (step) remaining time: 0:40:32
|
| 118 |
+
Epoch 1 | iter 256 step 8 | loss train: 1.463, val: 1.476 | iter time: 359.94 ms (step) remaining time: 0:40:18
|
| 119 |
+
Epoch 1 | iter 288 step 9 | loss train: 1.382, val: 1.476 | iter time: 358.60 ms (step) remaining time: 0:40:05
|
| 120 |
+
Epoch 1 | iter 320 step 10 | loss train: 1.388, val: 1.476 | iter time: 359.31 ms (step) remaining time: 0:39:52
|
| 121 |
+
Epoch 1 | iter 352 step 11 | loss train: 1.447, val: 1.476 | iter time: 359.66 ms (step) remaining time: 0:39:40
|
| 122 |
+
Epoch 1 | iter 384 step 12 | loss train: 1.524, val: 1.476 | iter time: 360.06 ms (step) remaining time: 0:39:28
|
| 123 |
+
Epoch 1 | iter 416 step 13 | loss train: 1.444, val: 1.476 | iter time: 358.06 ms (step) remaining time: 0:39:16
|
| 124 |
+
Epoch 1 | iter 448 step 14 | loss train: 1.431, val: 1.476 | iter time: 358.18 ms (step) remaining time: 0:39:05
|
| 125 |
+
Epoch 1 | iter 480 step 15 | loss train: 1.472, val: 1.476 | iter time: 359.05 ms (step) remaining time: 0:38:53
|
| 126 |
+
Epoch 1 | iter 512 step 16 | loss train: 1.468, val: 1.476 | iter time: 360.11 ms (step) remaining time: 0:38:42
|
| 127 |
+
Epoch 1 | iter 544 step 17 | loss train: 1.402, val: 1.476 | iter time: 360.00 ms (step) remaining time: 0:38:30
|
| 128 |
+
Epoch 1 | iter 576 step 18 | loss train: 1.418, val: 1.476 | iter time: 357.75 ms (step) remaining time: 0:38:19
|
| 129 |
+
Epoch 1 | iter 608 step 19 | loss train: 1.514, val: 1.476 | iter time: 358.85 ms (step) remaining time: 0:38:08
|
| 130 |
+
Epoch 1 | iter 640 step 20 | loss train: 1.380, val: 1.476 | iter time: 359.64 ms (step) remaining time: 0:37:56
|
| 131 |
+
Epoch 1 | iter 672 step 21 | loss train: 1.510, val: 1.476 | iter time: 359.77 ms (step) remaining time: 0:37:45
|
| 132 |
+
Epoch 1 | iter 704 step 22 | loss train: 1.443, val: 1.476 | iter time: 359.87 ms (step) remaining time: 0:37:34
|
| 133 |
+
Epoch 1 | iter 736 step 23 | loss train: 1.398, val: 1.476 | iter time: 363.44 ms (step) remaining time: 0:37:23
|
| 134 |
+
Epoch 1 | iter 768 step 24 | loss train: 1.455, val: 1.476 | iter time: 358.04 ms (step) remaining time: 0:37:12
|
| 135 |
+
Epoch 1 | iter 800 step 25 | loss train: 1.467, val: 1.476 | iter time: 359.13 ms (step) remaining time: 0:37:01
|
| 136 |
+
Epoch 1 | iter 832 step 26 | loss train: 1.532, val: 1.476 | iter time: 360.82 ms (step) remaining time: 0:36:50
|
| 137 |
+
Epoch 1 | iter 864 step 27 | loss train: 1.453, val: 1.476 | iter time: 360.87 ms (step) remaining time: 0:36:39
|
| 138 |
+
Epoch 1 | iter 896 step 28 | loss train: 1.494, val: 1.476 | iter time: 360.39 ms (step) remaining time: 0:36:28
|
| 139 |
+
Epoch 1 | iter 928 step 29 | loss train: 1.432, val: 1.476 | iter time: 360.38 ms (step) remaining time: 0:36:17
|
| 140 |
+
Epoch 1 | iter 960 step 30 | loss train: 1.460, val: 1.476 | iter time: 358.57 ms (step) remaining time: 0:36:06
|
| 141 |
+
Epoch 1 | iter 992 step 31 | loss train: 1.431, val: 1.476 | iter time: 359.41 ms (step) remaining time: 0:35:55
|
| 142 |
+
Epoch 1 | iter 1024 step 32 | loss train: 1.461, val: 1.476 | iter time: 361.23 ms (step) remaining time: 0:35:44
|
| 143 |
+
Epoch 1 | iter 1056 step 33 | loss train: 1.468, val: 1.476 | iter time: 361.40 ms (step) remaining time: 0:35:33
|
| 144 |
+
Epoch 1 | iter 1088 step 34 | loss train: 1.357, val: 1.476 | iter time: 361.12 ms (step) remaining time: 0:35:22
|
| 145 |
+
Epoch 1 | iter 1120 step 35 | loss train: 1.400, val: 1.476 | iter time: 359.38 ms (step) remaining time: 0:35:11
|
| 146 |
+
Epoch 1 | iter 1152 step 36 | loss train: 1.451, val: 1.476 | iter time: 360.41 ms (step) remaining time: 0:35:00
|
| 147 |
+
Epoch 1 | iter 1184 step 37 | loss train: 1.423, val: 1.476 | iter time: 358.08 ms (step) remaining time: 0:34:49
|
| 148 |
+
Epoch 1 | iter 1216 step 38 | loss train: 1.326, val: 1.476 | iter time: 359.41 ms (step) remaining time: 0:34:38
|
| 149 |
+
Epoch 1 | iter 1248 step 39 | loss train: 1.431, val: 1.476 | iter time: 360.46 ms (step) remaining time: 0:34:27
|
| 150 |
+
Epoch 1 | iter 1280 step 40 | loss train: 1.335, val: 1.476 | iter time: 358.80 ms (step) remaining time: 0:34:16
|
| 151 |
+
Epoch 1 | iter 1312 step 41 | loss train: 1.472, val: 1.476 | iter time: 359.76 ms (step) remaining time: 0:34:05
|
| 152 |
+
Epoch 1 | iter 1344 step 42 | loss train: 1.416, val: 1.476 | iter time: 360.22 ms (step) remaining time: 0:33:54
|
| 153 |
+
Epoch 1 | iter 1376 step 43 | loss train: 1.439, val: 1.476 | iter time: 361.07 ms (step) remaining time: 0:33:43
|
| 154 |
+
Epoch 1 | iter 1408 step 44 | loss train: 1.322, val: 1.476 | iter time: 358.30 ms (step) remaining time: 0:33:32
|
| 155 |
+
Epoch 1 | iter 1440 step 45 | loss train: 1.465, val: 1.476 | iter time: 358.79 ms (step) remaining time: 0:33:21
|
| 156 |
+
Epoch 1 | iter 1472 step 46 | loss train: 1.410, val: 1.476 | iter time: 359.17 ms (step) remaining time: 0:33:11
|
| 157 |
+
Epoch 1 | iter 1504 step 47 | loss train: 1.391, val: 1.476 | iter time: 359.52 ms (step) remaining time: 0:33:00
|
| 158 |
+
Epoch 1 | iter 1536 step 48 | loss train: 1.442, val: 1.476 | iter time: 358.06 ms (step) remaining time: 0:32:49
|
| 159 |
+
Epoch 1 | iter 1568 step 49 | loss train: 1.426, val: 1.476 | iter time: 359.79 ms (step) remaining time: 0:32:38
|
| 160 |
+
Epoch 1 | iter 1600 step 50 | loss train: 1.405, val: 1.476 | iter time: 357.34 ms (step) remaining time: 0:32:27
|
| 161 |
+
Validating ...
|
| 162 |
+
iter 1600: val loss 1.4626, val time: 21934.90 ms
|
| 163 |
+
Epoch 1 | iter 1632 step 51 | loss train: 1.409, val: 1.463 | iter time: 360.03 ms (step) remaining time: 0:33:33
|
| 164 |
+
Epoch 1 | iter 1664 step 52 | loss train: 1.472, val: 1.463 | iter time: 361.39 ms (step) remaining time: 0:33:20
|
| 165 |
+
Epoch 1 | iter 1696 step 53 | loss train: 1.347, val: 1.463 | iter time: 359.22 ms (step) remaining time: 0:33:07
|
| 166 |
+
Epoch 1 | iter 1728 step 54 | loss train: 1.342, val: 1.463 | iter time: 357.80 ms (step) remaining time: 0:32:55
|
| 167 |
+
Epoch 1 | iter 1760 step 55 | loss train: 1.430, val: 1.463 | iter time: 605.16 ms (step) remaining time: 0:32:43
|
| 168 |
+
Epoch 1 | iter 1792 step 56 | loss train: 1.435, val: 1.463 | iter time: 359.22 ms (step) remaining time: 0:32:30
|
| 169 |
+
Epoch 1 | iter 1824 step 57 | loss train: 1.413, val: 1.463 | iter time: 358.84 ms (step) remaining time: 0:32:18
|
| 170 |
+
Epoch 1 | iter 1856 step 58 | loss train: 1.386, val: 1.463 | iter time: 358.99 ms (step) remaining time: 0:32:05
|
| 171 |
+
Epoch 1 | iter 1888 step 59 | loss train: 1.427, val: 1.463 | iter time: 360.41 ms (step) remaining time: 0:31:53
|
| 172 |
+
Epoch 1 | iter 1920 step 60 | loss train: 1.400, val: 1.463 | iter time: 357.77 ms (step) remaining time: 0:31:40
|
| 173 |
+
Epoch 1 | iter 1952 step 61 | loss train: 1.450, val: 1.463 | iter time: 358.48 ms (step) remaining time: 0:31:28
|
| 174 |
+
Epoch 1 | iter 1984 step 62 | loss train: 1.381, val: 1.463 | iter time: 360.31 ms (step) remaining time: 0:31:16
|
| 175 |
+
Epoch 1 | iter 2016 step 63 | loss train: 1.372, val: 1.463 | iter time: 361.03 ms (step) remaining time: 0:31:04
|
| 176 |
+
Epoch 1 | iter 2048 step 64 | loss train: 1.455, val: 1.463 | iter time: 361.10 ms (step) remaining time: 0:30:51
|
| 177 |
+
Epoch 1 | iter 2080 step 65 | loss train: 1.453, val: 1.463 | iter time: 359.55 ms (step) remaining time: 0:30:39
|
| 178 |
+
Epoch 1 | iter 2112 step 66 | loss train: 1.407, val: 1.463 | iter time: 359.98 ms (step) remaining time: 0:30:27
|
| 179 |
+
Epoch 1 | iter 2144 step 67 | loss train: 1.403, val: 1.463 | iter time: 361.10 ms (step) remaining time: 0:30:15
|
| 180 |
+
Epoch 1 | iter 2176 step 68 | loss train: 1.389, val: 1.463 | iter time: 360.72 ms (step) remaining time: 0:30:03
|
| 181 |
+
Epoch 1 | iter 2208 step 69 | loss train: 1.431, val: 1.463 | iter time: 359.26 ms (step) remaining time: 0:29:51
|
| 182 |
+
Epoch 1 | iter 2240 step 70 | loss train: 1.421, val: 1.463 | iter time: 358.02 ms (step) remaining time: 0:29:39
|
| 183 |
+
Epoch 1 | iter 2272 step 71 | loss train: 1.408, val: 1.463 | iter time: 358.99 ms (step) remaining time: 0:29:27
|
| 184 |
+
Epoch 1 | iter 2304 step 72 | loss train: 1.349, val: 1.463 | iter time: 359.51 ms (step) remaining time: 0:29:15
|
| 185 |
+
Epoch 1 | iter 2336 step 73 | loss train: 1.480, val: 1.463 | iter time: 359.38 ms (step) remaining time: 0:29:03
|
| 186 |
+
Epoch 1 | iter 2368 step 74 | loss train: 1.400, val: 1.463 | iter time: 360.56 ms (step) remaining time: 0:28:52
|
| 187 |
+
Epoch 1 | iter 2400 step 75 | loss train: 1.423, val: 1.463 | iter time: 359.39 ms (step) remaining time: 0:28:40
|
| 188 |
+
Epoch 1 | iter 2432 step 76 | loss train: 1.468, val: 1.463 | iter time: 358.38 ms (step) remaining time: 0:28:28
|
| 189 |
+
Epoch 1 | iter 2464 step 77 | loss train: 1.457, val: 1.463 | iter time: 360.23 ms (step) remaining time: 0:28:16
|
| 190 |
+
Epoch 1 | iter 2496 step 78 | loss train: 1.464, val: 1.463 | iter time: 359.34 ms (step) remaining time: 0:28:04
|
| 191 |
+
Epoch 1 | iter 2528 step 79 | loss train: 1.422, val: 1.463 | iter time: 360.36 ms (step) remaining time: 0:27:53
|
| 192 |
+
Epoch 1 | iter 2560 step 80 | loss train: 1.396, val: 1.463 | iter time: 359.33 ms (step) remaining time: 0:27:41
|
| 193 |
+
Epoch 1 | iter 2592 step 81 | loss train: 1.370, val: 1.463 | iter time: 358.77 ms (step) remaining time: 0:27:29
|
| 194 |
+
Epoch 1 | iter 2624 step 82 | loss train: 1.462, val: 1.463 | iter time: 357.94 ms (step) remaining time: 0:27:18
|
| 195 |
+
Epoch 1 | iter 2656 step 83 | loss train: 1.440, val: 1.463 | iter time: 360.16 ms (step) remaining time: 0:27:06
|
| 196 |
+
Epoch 1 | iter 2688 step 84 | loss train: 1.407, val: 1.463 | iter time: 360.14 ms (step) remaining time: 0:26:54
|
| 197 |
+
Epoch 1 | iter 2720 step 85 | loss train: 1.381, val: 1.463 | iter time: 357.96 ms (step) remaining time: 0:26:43
|
| 198 |
+
Epoch 1 | iter 2752 step 86 | loss train: 1.398, val: 1.463 | iter time: 360.35 ms (step) remaining time: 0:26:31
|
| 199 |
+
Epoch 1 | iter 2784 step 87 | loss train: 1.412, val: 1.463 | iter time: 358.09 ms (step) remaining time: 0:26:20
|
| 200 |
+
Epoch 1 | iter 2816 step 88 | loss train: 1.441, val: 1.463 | iter time: 359.66 ms (step) remaining time: 0:26:08
|
| 201 |
+
Epoch 1 | iter 2848 step 89 | loss train: 1.427, val: 1.463 | iter time: 358.37 ms (step) remaining time: 0:25:56
|
| 202 |
+
Epoch 1 | iter 2880 step 90 | loss train: 1.425, val: 1.463 | iter time: 359.89 ms (step) remaining time: 0:25:45
|
| 203 |
+
Epoch 1 | iter 2912 step 91 | loss train: 1.437, val: 1.463 | iter time: 357.72 ms (step) remaining time: 0:25:33
|
| 204 |
+
Epoch 1 | iter 2944 step 92 | loss train: 1.421, val: 1.463 | iter time: 357.77 ms (step) remaining time: 0:25:22
|
| 205 |
+
Epoch 1 | iter 2976 step 93 | loss train: 1.427, val: 1.463 | iter time: 359.69 ms (step) remaining time: 0:25:10
|
| 206 |
+
Epoch 1 | iter 3008 step 94 | loss train: 1.365, val: 1.463 | iter time: 360.43 ms (step) remaining time: 0:24:59
|
| 207 |
+
Epoch 1 | iter 3040 step 95 | loss train: 1.356, val: 1.463 | iter time: 358.34 ms (step) remaining time: 0:24:47
|
| 208 |
+
Epoch 1 | iter 3072 step 96 | loss train: 1.374, val: 1.463 | iter time: 359.81 ms (step) remaining time: 0:24:36
|
| 209 |
+
Epoch 1 | iter 3104 step 97 | loss train: 1.399, val: 1.463 | iter time: 359.25 ms (step) remaining time: 0:24:25
|
| 210 |
+
Epoch 1 | iter 3136 step 98 | loss train: 1.420, val: 1.463 | iter time: 360.10 ms (step) remaining time: 0:24:13
|
| 211 |
+
Epoch 1 | iter 3168 step 99 | loss train: 1.341, val: 1.463 | iter time: 360.92 ms (step) remaining time: 0:24:02
|
| 212 |
+
Epoch 1 | iter 3200 step 100 | loss train: 1.384, val: 1.463 | iter time: 358.23 ms (step) remaining time: 0:23:50
|
| 213 |
+
Validating ...
|
| 214 |
+
iter 3200: val loss 1.4580, val time: 21920.12 ms
|
| 215 |
+
Saving checkpoint to 'out/pretrain/tinyllama/2409_lr4e-5/step-00000100/lit_model.pth'
|
| 216 |
+
Epoch 1 | iter 3232 step 101 | loss train: 1.406, val: 1.458 | iter time: 357.85 ms (step) remaining time: 0:24:27
|
| 217 |
+
Epoch 1 | iter 3264 step 102 | loss train: 1.475, val: 1.458 | iter time: 359.15 ms (step) remaining time: 0:24:15
|
| 218 |
+
Epoch 1 | iter 3296 step 103 | loss train: 1.458, val: 1.458 | iter time: 356.86 ms (step) remaining time: 0:24:03
|
| 219 |
+
Epoch 1 | iter 3328 step 104 | loss train: 1.400, val: 1.458 | iter time: 359.59 ms (step) remaining time: 0:23:51
|
| 220 |
+
Epoch 1 | iter 3360 step 105 | loss train: 1.399, val: 1.458 | iter time: 360.45 ms (step) remaining time: 0:23:39
|
| 221 |
+
Epoch 1 | iter 3392 step 106 | loss train: 1.417, val: 1.458 | iter time: 359.52 ms (step) remaining time: 0:23:26
|
| 222 |
+
Epoch 1 | iter 3424 step 107 | loss train: 1.451, val: 1.458 | iter time: 359.45 ms (step) remaining time: 0:23:15
|
| 223 |
+
Epoch 1 | iter 3456 step 108 | loss train: 1.403, val: 1.458 | iter time: 357.44 ms (step) remaining time: 0:23:02
|
| 224 |
+
Epoch 1 | iter 3488 step 109 | loss train: 1.366, val: 1.458 | iter time: 358.17 ms (step) remaining time: 0:22:50
|
| 225 |
+
Epoch 1 | iter 3520 step 110 | loss train: 1.384, val: 1.458 | iter time: 359.12 ms (step) remaining time: 0:22:38
|
| 226 |
+
Epoch 1 | iter 3552 step 111 | loss train: 1.378, val: 1.458 | iter time: 360.84 ms (step) remaining time: 0:22:26
|
| 227 |
+
Epoch 1 | iter 3584 step 112 | loss train: 1.423, val: 1.458 | iter time: 357.46 ms (step) remaining time: 0:22:14
|
| 228 |
+
Epoch 1 | iter 3616 step 113 | loss train: 1.323, val: 1.458 | iter time: 596.10 ms (step) remaining time: 0:22:03
|
| 229 |
+
Epoch 1 | iter 3648 step 114 | loss train: 1.356, val: 1.458 | iter time: 359.73 ms (step) remaining time: 0:21:51
|
| 230 |
+
Epoch 1 | iter 3680 step 115 | loss train: 1.409, val: 1.458 | iter time: 361.04 ms (step) remaining time: 0:21:39
|
| 231 |
+
Epoch 1 | iter 3712 step 116 | loss train: 1.437, val: 1.458 | iter time: 359.73 ms (step) remaining time: 0:21:27
|
| 232 |
+
Epoch 1 | iter 3744 step 117 | loss train: 1.408, val: 1.458 | iter time: 359.68 ms (step) remaining time: 0:21:15
|
| 233 |
+
Epoch 1 | iter 3776 step 118 | loss train: 1.356, val: 1.458 | iter time: 360.61 ms (step) remaining time: 0:21:03
|
| 234 |
+
Epoch 1 | iter 3808 step 119 | loss train: 1.425, val: 1.458 | iter time: 359.65 ms (step) remaining time: 0:20:51
|
| 235 |
+
Epoch 1 | iter 3840 step 120 | loss train: 1.417, val: 1.458 | iter time: 359.37 ms (step) remaining time: 0:20:39
|
| 236 |
+
Epoch 1 | iter 3872 step 121 | loss train: 1.429, val: 1.458 | iter time: 359.00 ms (step) remaining time: 0:20:28
|
| 237 |
+
Epoch 1 | iter 3904 step 122 | loss train: 1.404, val: 1.458 | iter time: 359.91 ms (step) remaining time: 0:20:16
|
| 238 |
+
Epoch 1 | iter 3936 step 123 | loss train: 1.426, val: 1.458 | iter time: 361.13 ms (step) remaining time: 0:20:04
|
| 239 |
+
Epoch 1 | iter 3968 step 124 | loss train: 1.385, val: 1.458 | iter time: 359.71 ms (step) remaining time: 0:19:52
|
| 240 |
+
Epoch 1 | iter 4000 step 125 | loss train: 1.389, val: 1.458 | iter time: 358.60 ms (step) remaining time: 0:19:41
|
| 241 |
+
Epoch 1 | iter 4032 step 126 | loss train: 1.388, val: 1.458 | iter time: 359.52 ms (step) remaining time: 0:19:29
|
| 242 |
+
Epoch 1 | iter 4064 step 127 | loss train: 1.445, val: 1.458 | iter time: 359.61 ms (step) remaining time: 0:19:17
|
| 243 |
+
Epoch 1 | iter 4096 step 128 | loss train: 1.386, val: 1.458 | iter time: 358.50 ms (step) remaining time: 0:19:06
|
| 244 |
+
Epoch 1 | iter 4128 step 129 | loss train: 1.404, val: 1.458 | iter time: 358.11 ms (step) remaining time: 0:18:54
|
| 245 |
+
Epoch 1 | iter 4160 step 130 | loss train: 1.408, val: 1.458 | iter time: 357.88 ms (step) remaining time: 0:18:42
|
| 246 |
+
Epoch 1 | iter 4192 step 131 | loss train: 1.440, val: 1.458 | iter time: 360.37 ms (step) remaining time: 0:18:30
|
| 247 |
+
Epoch 1 | iter 4224 step 132 | loss train: 1.380, val: 1.458 | iter time: 358.55 ms (step) remaining time: 0:18:19
|
| 248 |
+
Epoch 1 | iter 4256 step 133 | loss train: 1.398, val: 1.458 | iter time: 359.42 ms (step) remaining time: 0:18:07
|
| 249 |
+
Epoch 1 | iter 4288 step 134 | loss train: 1.378, val: 1.458 | iter time: 360.42 ms (step) remaining time: 0:17:55
|
| 250 |
+
Epoch 1 | iter 4320 step 135 | loss train: 1.364, val: 1.458 | iter time: 358.11 ms (step) remaining time: 0:17:44
|
| 251 |
+
Epoch 1 | iter 4352 step 136 | loss train: 1.416, val: 1.458 | iter time: 360.24 ms (step) remaining time: 0:17:32
|
| 252 |
+
Epoch 1 | iter 4384 step 137 | loss train: 1.386, val: 1.458 | iter time: 359.45 ms (step) remaining time: 0:17:21
|
| 253 |
+
Epoch 1 | iter 4416 step 138 | loss train: 1.408, val: 1.458 | iter time: 361.19 ms (step) remaining time: 0:17:09
|
| 254 |
+
Epoch 1 | iter 4448 step 139 | loss train: 1.459, val: 1.458 | iter time: 360.82 ms (step) remaining time: 0:16:57
|
| 255 |
+
Epoch 1 | iter 4480 step 140 | loss train: 1.402, val: 1.458 | iter time: 358.89 ms (step) remaining time: 0:16:46
|
| 256 |
+
Epoch 1 | iter 4512 step 141 | loss train: 1.382, val: 1.458 | iter time: 360.97 ms (step) remaining time: 0:16:34
|
| 257 |
+
Epoch 1 | iter 4544 step 142 | loss train: 1.371, val: 1.458 | iter time: 358.42 ms (step) remaining time: 0:16:23
|
| 258 |
+
Epoch 1 | iter 4576 step 143 | loss train: 1.418, val: 1.458 | iter time: 359.63 ms (step) remaining time: 0:16:11
|
| 259 |
+
Epoch 1 | iter 4608 step 144 | loss train: 1.351, val: 1.458 | iter time: 359.74 ms (step) remaining time: 0:15:59
|
| 260 |
+
Epoch 1 | iter 4640 step 145 | loss train: 1.333, val: 1.458 | iter time: 360.27 ms (step) remaining time: 0:15:48
|
| 261 |
+
Epoch 1 | iter 4672 step 146 | loss train: 1.401, val: 1.458 | iter time: 358.44 ms (step) remaining time: 0:15:36
|
| 262 |
+
Epoch 1 | iter 4704 step 147 | loss train: 1.430, val: 1.458 | iter time: 359.89 ms (step) remaining time: 0:15:25
|
| 263 |
+
Epoch 1 | iter 4736 step 148 | loss train: 1.327, val: 1.458 | iter time: 359.88 ms (step) remaining time: 0:15:13
|
| 264 |
+
Epoch 1 | iter 4768 step 149 | loss train: 1.426, val: 1.458 | iter time: 360.46 ms (step) remaining time: 0:15:02
|
| 265 |
+
Epoch 1 | iter 4800 step 150 | loss train: 1.401, val: 1.458 | iter time: 360.05 ms (step) remaining time: 0:14:50
|
| 266 |
+
Validating ...
|
| 267 |
+
iter 4800: val loss 1.4516, val time: 21924.21 ms
|
| 268 |
+
Epoch 1 | iter 4832 step 151 | loss train: 1.370, val: 1.452 | iter time: 360.84 ms (step) remaining time: 0:14:50
|
| 269 |
+
Epoch 1 | iter 4864 step 152 | loss train: 1.437, val: 1.452 | iter time: 360.27 ms (step) remaining time: 0:14:38
|
| 270 |
+
Epoch 1 | iter 4896 step 153 | loss train: 1.412, val: 1.452 | iter time: 358.42 ms (step) remaining time: 0:14:27
|
| 271 |
+
Epoch 1 | iter 4928 step 154 | loss train: 1.325, val: 1.452 | iter time: 361.22 ms (step) remaining time: 0:14:15
|
| 272 |
+
Epoch 1 | iter 4960 step 155 | loss train: 1.377, val: 1.452 | iter time: 358.15 ms (step) remaining time: 0:14:03
|
| 273 |
+
Epoch 1 | iter 4992 step 156 | loss train: 1.376, val: 1.452 | iter time: 359.50 ms (step) remaining time: 0:13:52
|
| 274 |
+
Epoch 1 | iter 5024 step 157 | loss train: 1.419, val: 1.452 | iter time: 360.80 ms (step) remaining time: 0:13:40
|
| 275 |
+
Epoch 1 | iter 5056 step 158 | loss train: 1.383, val: 1.452 | iter time: 358.62 ms (step) remaining time: 0:13:28
|
| 276 |
+
Epoch 1 | iter 5088 step 159 | loss train: 1.442, val: 1.452 | iter time: 360.67 ms (step) remaining time: 0:13:17
|
| 277 |
+
Epoch 1 | iter 5120 step 160 | loss train: 1.416, val: 1.452 | iter time: 361.06 ms (step) remaining time: 0:13:05
|
| 278 |
+
Epoch 1 | iter 5152 step 161 | loss train: 1.381, val: 1.452 | iter time: 359.42 ms (step) remaining time: 0:12:54
|
| 279 |
+
Epoch 1 | iter 5184 step 162 | loss train: 1.408, val: 1.452 | iter time: 359.96 ms (step) remaining time: 0:12:42
|
| 280 |
+
Epoch 1 | iter 5216 step 163 | loss train: 1.434, val: 1.452 | iter time: 359.23 ms (step) remaining time: 0:12:30
|
| 281 |
+
Epoch 1 | iter 5248 step 164 | loss train: 1.373, val: 1.452 | iter time: 358.71 ms (step) remaining time: 0:12:19
|
| 282 |
+
Epoch 1 | iter 5280 step 165 | loss train: 1.435, val: 1.452 | iter time: 361.07 ms (step) remaining time: 0:12:07
|
| 283 |
+
Epoch 1 | iter 5312 step 166 | loss train: 1.414, val: 1.452 | iter time: 360.17 ms (step) remaining time: 0:11:56
|
| 284 |
+
Epoch 1 | iter 5344 step 167 | loss train: 1.380, val: 1.452 | iter time: 360.52 ms (step) remaining time: 0:11:44
|
| 285 |
+
Epoch 1 | iter 5376 step 168 | loss train: 1.317, val: 1.452 | iter time: 360.58 ms (step) remaining time: 0:11:33
|
| 286 |
+
Epoch 1 | iter 5408 step 169 | loss train: 1.351, val: 1.452 | iter time: 360.37 ms (step) remaining time: 0:11:21
|
| 287 |
+
Epoch 1 | iter 5440 step 170 | loss train: 1.424, val: 1.452 | iter time: 359.67 ms (step) remaining time: 0:11:10
|
| 288 |
+
Epoch 1 | iter 5472 step 171 | loss train: 1.347, val: 1.452 | iter time: 358.93 ms (step) remaining time: 0:10:58
|
| 289 |
+
Epoch 1 | iter 5504 step 172 | loss train: 1.405, val: 1.452 | iter time: 357.71 ms (step) remaining time: 0:10:46
|
| 290 |
+
Epoch 1 | iter 5536 step 173 | loss train: 1.371, val: 1.452 | iter time: 361.79 ms (step) remaining time: 0:10:35
|
| 291 |
+
Epoch 1 | iter 5568 step 174 | loss train: 1.490, val: 1.452 | iter time: 359.08 ms (step) remaining time: 0:10:23
|
| 292 |
+
Epoch 1 | iter 5600 step 175 | loss train: 1.386, val: 1.452 | iter time: 359.42 ms (step) remaining time: 0:10:12
|
| 293 |
+
Epoch 1 | iter 5632 step 176 | loss train: 1.329, val: 1.452 | iter time: 358.66 ms (step) remaining time: 0:10:00
|
| 294 |
+
Epoch 1 | iter 5664 step 177 | loss train: 1.423, val: 1.452 | iter time: 360.97 ms (step) remaining time: 0:09:49
|
| 295 |
+
Epoch 1 | iter 5696 step 178 | loss train: 1.364, val: 1.452 | iter time: 360.18 ms (step) remaining time: 0:09:38
|
| 296 |
+
Epoch 1 | iter 5728 step 179 | loss train: 1.362, val: 1.452 | iter time: 360.96 ms (step) remaining time: 0:09:26
|
| 297 |
+
Epoch 1 | iter 5760 step 180 | loss train: 1.379, val: 1.452 | iter time: 359.09 ms (step) remaining time: 0:09:15
|
| 298 |
+
Epoch 1 | iter 5792 step 181 | loss train: 1.378, val: 1.452 | iter time: 359.78 ms (step) remaining time: 0:09:03
|
| 299 |
+
Epoch 1 | iter 5824 step 182 | loss train: 1.406, val: 1.452 | iter time: 360.20 ms (step) remaining time: 0:08:52
|
| 300 |
+
Epoch 1 | iter 5856 step 183 | loss train: 1.436, val: 1.452 | iter time: 359.22 ms (step) remaining time: 0:08:40
|
| 301 |
+
Epoch 1 | iter 5888 step 184 | loss train: 1.378, val: 1.452 | iter time: 360.23 ms (step) remaining time: 0:08:29
|
| 302 |
+
Epoch 1 | iter 5920 step 185 | loss train: 1.415, val: 1.452 | iter time: 360.64 ms (step) remaining time: 0:08:17
|
| 303 |
+
Epoch 1 | iter 5952 step 186 | loss train: 1.337, val: 1.452 | iter time: 359.25 ms (step) remaining time: 0:08:06
|
| 304 |
+
Epoch 1 | iter 5984 step 187 | loss train: 1.320, val: 1.452 | iter time: 360.09 ms (step) remaining time: 0:07:55
|
| 305 |
+
Epoch 1 | iter 6016 step 188 | loss train: 1.446, val: 1.452 | iter time: 360.62 ms (step) remaining time: 0:07:43
|
| 306 |
+
Epoch 1 | iter 6048 step 189 | loss train: 1.330, val: 1.452 | iter time: 360.41 ms (step) remaining time: 0:07:32
|
| 307 |
+
Epoch 1 | iter 6080 step 190 | loss train: 1.385, val: 1.452 | iter time: 361.33 ms (step) remaining time: 0:07:20
|
| 308 |
+
Epoch 1 | iter 6112 step 191 | loss train: 1.373, val: 1.452 | iter time: 360.04 ms (step) remaining time: 0:07:09
|
| 309 |
+
Epoch 1 | iter 6144 step 192 | loss train: 1.391, val: 1.452 | iter time: 361.01 ms (step) remaining time: 0:06:58
|
| 310 |
+
Epoch 1 | iter 6176 step 193 | loss train: 1.398, val: 1.452 | iter time: 360.77 ms (step) remaining time: 0:06:46
|
| 311 |
+
Epoch 1 | iter 6208 step 194 | loss train: 1.407, val: 1.452 | iter time: 604.28 ms (step) remaining time: 0:06:35
|
| 312 |
+
Epoch 1 | iter 6240 step 195 | loss train: 1.342, val: 1.452 | iter time: 359.34 ms (step) remaining time: 0:06:24
|
| 313 |
+
Epoch 1 | iter 6272 step 196 | loss train: 1.441, val: 1.452 | iter time: 362.32 ms (step) remaining time: 0:06:12
|
| 314 |
+
Epoch 1 | iter 6304 step 197 | loss train: 1.441, val: 1.452 | iter time: 359.83 ms (step) remaining time: 0:06:01
|
| 315 |
+
Epoch 1 | iter 6336 step 198 | loss train: 1.373, val: 1.452 | iter time: 358.74 ms (step) remaining time: 0:05:49
|
| 316 |
+
Epoch 1 | iter 6368 step 199 | loss train: 1.436, val: 1.452 | iter time: 359.92 ms (step) remaining time: 0:05:38
|
| 317 |
+
Epoch 1 | iter 6400 step 200 | loss train: 1.352, val: 1.452 | iter time: 359.42 ms (step) remaining time: 0:05:27
|
| 318 |
+
Validating ...
|
| 319 |
+
iter 6400: val loss 1.4481, val time: 21929.64 ms
|
| 320 |
+
Saving checkpoint to 'out/pretrain/tinyllama/2409_lr4e-5/step-00000200/lit_model.pth'
|
| 321 |
+
Epoch 1 | iter 6432 step 201 | loss train: 1.359, val: 1.448 | iter time: 358.05 ms (step) remaining time: 0:05:21
|
| 322 |
+
Epoch 1 | iter 6464 step 202 | loss train: 1.331, val: 1.448 | iter time: 357.87 ms (step) remaining time: 0:05:09
|
| 323 |
+
Epoch 1 | iter 6496 step 203 | loss train: 1.441, val: 1.448 | iter time: 357.22 ms (step) remaining time: 0:04:58
|
| 324 |
+
Epoch 1 | iter 6528 step 204 | loss train: 1.402, val: 1.448 | iter time: 359.81 ms (step) remaining time: 0:04:46
|
| 325 |
+
Epoch 1 | iter 6560 step 205 | loss train: 1.405, val: 1.448 | iter time: 360.59 ms (step) remaining time: 0:04:35
|
| 326 |
+
Epoch 1 | iter 6592 step 206 | loss train: 1.356, val: 1.448 | iter time: 358.81 ms (step) remaining time: 0:04:23
|
| 327 |
+
Epoch 1 | iter 6624 step 207 | loss train: 1.385, val: 1.448 | iter time: 360.74 ms (step) remaining time: 0:04:11
|
| 328 |
+
Epoch 1 | iter 6656 step 208 | loss train: 1.364, val: 1.448 | iter time: 360.52 ms (step) remaining time: 0:04:00
|
| 329 |
+
Epoch 1 | iter 6688 step 209 | loss train: 1.383, val: 1.448 | iter time: 359.06 ms (step) remaining time: 0:03:48
|
| 330 |
+
Epoch 1 | iter 6720 step 210 | loss train: 1.430, val: 1.448 | iter time: 359.83 ms (step) remaining time: 0:03:37
|
| 331 |
+
Epoch 1 | iter 6752 step 211 | loss train: 1.330, val: 1.448 | iter time: 361.80 ms (step) remaining time: 0:03:25
|
| 332 |
+
Epoch 1 | iter 6784 step 212 | loss train: 1.422, val: 1.448 | iter time: 359.61 ms (step) remaining time: 0:03:14
|
| 333 |
+
Epoch 1 | iter 6816 step 213 | loss train: 1.382, val: 1.448 | iter time: 358.98 ms (step) remaining time: 0:03:02
|
| 334 |
+
Epoch 1 | iter 6848 step 214 | loss train: 1.410, val: 1.448 | iter time: 359.65 ms (step) remaining time: 0:02:51
|
| 335 |
+
Epoch 1 | iter 6880 step 215 | loss train: 1.425, val: 1.448 | iter time: 359.73 ms (step) remaining time: 0:02:40
|
| 336 |
+
Epoch 1 | iter 6912 step 216 | loss train: 1.337, val: 1.448 | iter time: 360.00 ms (step) remaining time: 0:02:28
|
| 337 |
+
Epoch 1 | iter 6944 step 217 | loss train: 1.382, val: 1.448 | iter time: 361.35 ms (step) remaining time: 0:02:17
|
| 338 |
+
Epoch 1 | iter 6976 step 218 | loss train: 1.289, val: 1.448 | iter time: 360.33 ms (step) remaining time: 0:02:05
|
| 339 |
+
Epoch 1 | iter 7008 step 219 | loss train: 1.404, val: 1.448 | iter time: 359.64 ms (step) remaining time: 0:01:54
|
| 340 |
+
Epoch 1 | iter 7040 step 220 | loss train: 1.327, val: 1.448 | iter time: 458.26 ms (step) remaining time: 0:01:42
|
| 341 |
+
Epoch 1 | iter 7072 step 221 | loss train: 1.339, val: 1.448 | iter time: 359.29 ms (step) remaining time: 0:01:31
|
| 342 |
+
Epoch 1 | iter 7104 step 222 | loss train: 1.324, val: 1.448 | iter time: 360.20 ms (step) remaining time: 0:01:19
|
| 343 |
+
Epoch 1 | iter 7136 step 223 | loss train: 1.408, val: 1.448 | iter time: 360.63 ms (step) remaining time: 0:01:08
|
| 344 |
+
Epoch 1 | iter 7168 step 224 | loss train: 1.403, val: 1.448 | iter time: 357.99 ms (step) remaining time: 0:00:57
|
| 345 |
+
Epoch 1 | iter 7200 step 225 | loss train: 1.408, val: 1.448 | iter time: 361.14 ms (step) remaining time: 0:00:45
|
| 346 |
+
Epoch 1 | iter 7232 step 226 | loss train: 1.354, val: 1.448 | iter time: 359.66 ms (step) remaining time: 0:00:34
|
| 347 |
+
Epoch 1 | iter 7264 step 227 | loss train: 1.414, val: 1.448 | iter time: 358.26 ms (step) remaining time: 0:00:22
|
| 348 |
+
Epoch 1 | iter 7296 step 228 | loss train: 1.287, val: 1.448 | iter time: 360.67 ms (step) remaining time: 0:00:11
|
| 349 |
+
Epoch 2 | iter 7328 step 229 | loss train: 1.357, val: 1.448 | iter time: 359.97 ms (step) remaining time: 0:00:00
|
| 350 |
+
Validating ...
|
| 351 |
+
Final evaluation | val loss: 1.388 | val ppl: 4.009
|
| 352 |
+
Saving checkpoint to 'out/pretrain/tinyllama/2409_lr4e-5/final/lit_model.pth'
|
| 353 |
+
----------------------------------------
|
| 354 |
+
| Performance
|
| 355 |
+
| - Total tokens : 240,123,904
|
| 356 |
+
| - Training Time : 2672.28 s
|
| 357 |
+
| - Tok/sec : 124.03 tok/s
|
| 358 |
+
| ----------------------------------------
|
| 359 |
+
| Memory Usage
|
| 360 |
+
| - Memory Used : 26.32 GB
|
| 361 |
+
----------------------------------------
|
out/pretrain/tinyllama/teelogs/2410.txt
ADDED
|
@@ -0,0 +1,368 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
Initializing distributed: GLOBAL_RANK: 0, MEMBER: 1/4
|
| 2 |
+
Initializing distributed: GLOBAL_RANK: 3, MEMBER: 4/4
|
| 3 |
+
[rank: 3] Seed set to 42
|
| 4 |
+
Initializing distributed: GLOBAL_RANK: 2, MEMBER: 3/4
|
| 5 |
+
[rank: 2] Seed set to 42
|
| 6 |
+
Initializing distributed: GLOBAL_RANK: 1, MEMBER: 2/4
|
| 7 |
+
[rank: 1] Seed set to 42
|
| 8 |
+
----------------------------------------------------------------------------------------------------
|
| 9 |
+
distributed_backend=nccl
|
| 10 |
+
All distributed processes registered. Starting with 4 processes
|
| 11 |
+
----------------------------------------------------------------------------------------------------
|
| 12 |
+
|
| 13 |
+
[rank: 0] Seed set to 42
|
| 14 |
+
All GPUs are fully connected via NVLink.
|
| 15 |
+
{'data': {'batch_size': 1,
|
| 16 |
+
'data_path': PosixPath('litgpt/data/arxiv'),
|
| 17 |
+
'num_workers': 8,
|
| 18 |
+
'seed': 42,
|
| 19 |
+
'seq_length': 2048,
|
| 20 |
+
'use_starcoder': True},
|
| 21 |
+
'data_dir': PosixPath('litgpt/data/arxiv/2410'),
|
| 22 |
+
'devices': 'auto',
|
| 23 |
+
'eval': {'evaluate_example': 'first',
|
| 24 |
+
'final_validation': True,
|
| 25 |
+
'initial_validation': True,
|
| 26 |
+
'interval': 50,
|
| 27 |
+
'max_iters': 100,
|
| 28 |
+
'max_new_tokens': None},
|
| 29 |
+
'initial_checkpoint_dir': PosixPath('out/pretrain/2409/final'),
|
| 30 |
+
'log': {'group': None, 'project': None, 'run': None},
|
| 31 |
+
'logger_name': 'tensorboard',
|
| 32 |
+
'model_config': {'attention_logit_softcapping': None,
|
| 33 |
+
'attention_scores_scalar': None,
|
| 34 |
+
'attn_bias': False,
|
| 35 |
+
'bias': False,
|
| 36 |
+
'block_size': 2048,
|
| 37 |
+
'final_logit_softcapping': None,
|
| 38 |
+
'gelu_approximate': 'none',
|
| 39 |
+
'head_size': 64,
|
| 40 |
+
'hf_config': {'name': 'TinyLlama-1.1B-intermediate-step-1431k-3T',
|
| 41 |
+
'org': 'TinyLlama'},
|
| 42 |
+
'intermediate_size': 5632,
|
| 43 |
+
'lm_head_bias': False,
|
| 44 |
+
'mlp_class_name': 'LLaMAMLP',
|
| 45 |
+
'moe_intermediate_size': None,
|
| 46 |
+
'n_embd': 2048,
|
| 47 |
+
'n_expert': 0,
|
| 48 |
+
'n_expert_per_token': 0,
|
| 49 |
+
'n_head': 32,
|
| 50 |
+
'n_layer': 22,
|
| 51 |
+
'n_query_groups': 4,
|
| 52 |
+
'name': 'tiny-llama-1.1b',
|
| 53 |
+
'norm_1': True,
|
| 54 |
+
'norm_2': True,
|
| 55 |
+
'norm_class_name': 'RMSNorm',
|
| 56 |
+
'norm_eps': 1e-05,
|
| 57 |
+
'norm_qk': False,
|
| 58 |
+
'norm_qk_type': 'default',
|
| 59 |
+
'padded_vocab_size': 32000,
|
| 60 |
+
'padding_multiple': 64,
|
| 61 |
+
'parallel_residual': False,
|
| 62 |
+
'post_attention_norm': False,
|
| 63 |
+
'post_mlp_norm': False,
|
| 64 |
+
'rope_adjustments': None,
|
| 65 |
+
'rope_base': 10000,
|
| 66 |
+
'rope_condense_ratio': 1,
|
| 67 |
+
'rope_indices': None,
|
| 68 |
+
'rope_local_base_freq': None,
|
| 69 |
+
'rotary_percentage': 1.0,
|
| 70 |
+
'scale_embeddings': False,
|
| 71 |
+
'shared_attention_norm': False,
|
| 72 |
+
'sliding_window_indices': None,
|
| 73 |
+
'sliding_window_size': None,
|
| 74 |
+
'vocab_size': 32000},
|
| 75 |
+
'model_name': 'tiny-llama-1.1b',
|
| 76 |
+
'num_nodes': 1,
|
| 77 |
+
'optimizer': "{'class_path': 'torch.optim.AdamW', 'init_args': {'lr': 0.0004, "
|
| 78 |
+
"'weight_decay': 0.1, 'betas': [0.9, 0.95]}}",
|
| 79 |
+
'out_dir': PosixPath('out/pretrain/2410'),
|
| 80 |
+
'precision': 'bf16-mixed',
|
| 81 |
+
'resume': False,
|
| 82 |
+
'seed': 42,
|
| 83 |
+
'tokenizer_dir': PosixPath('checkpoints/TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T'),
|
| 84 |
+
'train': {'epochs': None,
|
| 85 |
+
'global_batch_size': 512,
|
| 86 |
+
'log_interval': 1,
|
| 87 |
+
'lr_warmup_fraction': None,
|
| 88 |
+
'lr_warmup_steps': 20,
|
| 89 |
+
'max_norm': 1.0,
|
| 90 |
+
'max_seq_length': 2048,
|
| 91 |
+
'max_steps': None,
|
| 92 |
+
'max_tokens': 255852544,
|
| 93 |
+
'micro_batch_size': 4,
|
| 94 |
+
'min_lr': 4e-05,
|
| 95 |
+
'save_interval': 100,
|
| 96 |
+
'tie_embeddings': None}}
|
| 97 |
+
Time to instantiate model: 0.02 seconds.
|
| 98 |
+
Total parameters: 1,100,048,384
|
| 99 |
+
[fix] out/pretrain/2409/final/lit_model.pth 是整包 state，提取 model 权重并原子覆盖...
|
| 100 |
+
[fix] 已覆盖为纯权重: out/pretrain/2409/final/lit_model.pth
|
| 101 |
+
Validating ...
|
| 102 |
+
Measured TFLOPs: 239.66
|
| 103 |
+
Epoch 1 | iter 32 step 1 | loss train: 1.466, val: 1.433 | iter time: 538.97 ms (step) remaining time: 0:48:06
|
| 104 |
+
Epoch 1 | iter 64 step 2 | loss train: 1.464, val: 1.433 | iter time: 357.93 ms (step) remaining time: 0:45:38
|
| 105 |
+
Epoch 1 | iter 96 step 3 | loss train: 1.443, val: 1.433 | iter time: 356.34 ms (step) remaining time: 0:44:42
|
| 106 |
+
Epoch 1 | iter 128 step 4 | loss train: 1.445, val: 1.433 | iter time: 360.75 ms (step) remaining time: 0:44:13
|
| 107 |
+
Epoch 1 | iter 160 step 5 | loss train: 1.402, val: 1.433 | iter time: 359.07 ms (step) remaining time: 0:43:52
|
| 108 |
+
Epoch 1 | iter 192 step 6 | loss train: 1.479, val: 1.433 | iter time: 359.01 ms (step) remaining time: 0:43:34
|
| 109 |
+
Epoch 1 | iter 224 step 7 | loss train: 1.492, val: 1.433 | iter time: 358.53 ms (step) remaining time: 0:43:18
|
| 110 |
+
Epoch 1 | iter 256 step 8 | loss train: 1.459, val: 1.433 | iter time: 359.98 ms (step) remaining time: 0:43:03
|
| 111 |
+
Epoch 1 | iter 288 step 9 | loss train: 1.464, val: 1.433 | iter time: 358.87 ms (step) remaining time: 0:42:49
|
| 112 |
+
Epoch 1 | iter 320 step 10 | loss train: 1.448, val: 1.433 | iter time: 358.91 ms (step) remaining time: 0:42:36
|
| 113 |
+
Epoch 1 | iter 352 step 11 | loss train: 1.456, val: 1.433 | iter time: 357.55 ms (step) remaining time: 0:42:24
|
| 114 |
+
Epoch 1 | iter 384 step 12 | loss train: 1.470, val: 1.433 | iter time: 359.43 ms (step) remaining time: 0:42:12
|
| 115 |
+
Epoch 1 | iter 416 step 13 | loss train: 1.490, val: 1.433 | iter time: 360.05 ms (step) remaining time: 0:41:59
|
| 116 |
+
Epoch 1 | iter 448 step 14 | loss train: 1.440, val: 1.433 | iter time: 360.50 ms (step) remaining time: 0:41:47
|
| 117 |
+
Epoch 1 | iter 480 step 15 | loss train: 1.453, val: 1.433 | iter time: 360.78 ms (step) remaining time: 0:41:36
|
| 118 |
+
Epoch 1 | iter 512 step 16 | loss train: 1.419, val: 1.433 | iter time: 360.18 ms (step) remaining time: 0:41:24
|
| 119 |
+
Epoch 1 | iter 544 step 17 | loss train: 1.485, val: 1.433 | iter time: 360.97 ms (step) remaining time: 0:41:13
|
| 120 |
+
Epoch 1 | iter 576 step 18 | loss train: 1.457, val: 1.433 | iter time: 359.06 ms (step) remaining time: 0:41:02
|
| 121 |
+
Epoch 1 | iter 608 step 19 | loss train: 1.491, val: 1.433 | iter time: 358.67 ms (step) remaining time: 0:40:50
|
| 122 |
+
Epoch 1 | iter 640 step 20 | loss train: 1.485, val: 1.433 | iter time: 359.53 ms (step) remaining time: 0:40:39
|
| 123 |
+
Epoch 1 | iter 672 step 21 | loss train: 1.460, val: 1.433 | iter time: 360.01 ms (step) remaining time: 0:40:28
|
| 124 |
+
Epoch 1 | iter 704 step 22 | loss train: 1.486, val: 1.433 | iter time: 358.20 ms (step) remaining time: 0:40:17
|
| 125 |
+
Epoch 1 | iter 736 step 23 | loss train: 1.431, val: 1.433 | iter time: 358.85 ms (step) remaining time: 0:40:06
|
| 126 |
+
Epoch 1 | iter 768 step 24 | loss train: 1.432, val: 1.433 | iter time: 358.47 ms (step) remaining time: 0:39:55
|
| 127 |
+
Epoch 1 | iter 800 step 25 | loss train: 1.430, val: 1.433 | iter time: 359.47 ms (step) remaining time: 0:39:44
|
| 128 |
+
Epoch 1 | iter 832 step 26 | loss train: 1.482, val: 1.433 | iter time: 360.26 ms (step) remaining time: 0:39:32
|
| 129 |
+
Epoch 1 | iter 864 step 27 | loss train: 1.422, val: 1.433 | iter time: 358.35 ms (step) remaining time: 0:39:21
|
| 130 |
+
Epoch 1 | iter 896 step 28 | loss train: 1.398, val: 1.433 | iter time: 359.50 ms (step) remaining time: 0:39:10
|
| 131 |
+
Epoch 1 | iter 928 step 29 | loss train: 1.424, val: 1.433 | iter time: 360.78 ms (step) remaining time: 0:38:59
|
| 132 |
+
Epoch 1 | iter 960 step 30 | loss train: 1.510, val: 1.433 | iter time: 360.66 ms (step) remaining time: 0:38:49
|
| 133 |
+
Epoch 1 | iter 992 step 31 | loss train: 1.376, val: 1.433 | iter time: 358.93 ms (step) remaining time: 0:38:38
|
| 134 |
+
Epoch 1 | iter 1024 step 32 | loss train: 1.508, val: 1.433 | iter time: 359.29 ms (step) remaining time: 0:38:28
|
| 135 |
+
Epoch 1 | iter 1056 step 33 | loss train: 1.424, val: 1.433 | iter time: 360.59 ms (step) remaining time: 0:38:17
|
| 136 |
+
Epoch 1 | iter 1088 step 34 | loss train: 1.426, val: 1.433 | iter time: 361.23 ms (step) remaining time: 0:38:06
|
| 137 |
+
Epoch 1 | iter 1120 step 35 | loss train: 1.462, val: 1.433 | iter time: 361.19 ms (step) remaining time: 0:37:55
|
| 138 |
+
Epoch 1 | iter 1152 step 36 | loss train: 1.465, val: 1.433 | iter time: 360.33 ms (step) remaining time: 0:37:44
|
| 139 |
+
Epoch 1 | iter 1184 step 37 | loss train: 1.506, val: 1.433 | iter time: 360.63 ms (step) remaining time: 0:37:33
|
| 140 |
+
Epoch 1 | iter 1216 step 38 | loss train: 1.472, val: 1.433 | iter time: 359.68 ms (step) remaining time: 0:37:22
|
| 141 |
+
Epoch 1 | iter 1248 step 39 | loss train: 1.373, val: 1.433 | iter time: 574.68 ms (step) remaining time: 0:37:12
|
| 142 |
+
Epoch 1 | iter 1280 step 40 | loss train: 1.489, val: 1.433 | iter time: 361.46 ms (step) remaining time: 0:37:01
|
| 143 |
+
Epoch 1 | iter 1312 step 41 | loss train: 1.445, val: 1.433 | iter time: 360.36 ms (step) remaining time: 0:36:50
|
| 144 |
+
Epoch 1 | iter 1344 step 42 | loss train: 1.536, val: 1.433 | iter time: 360.28 ms (step) remaining time: 0:36:39
|
| 145 |
+
Epoch 1 | iter 1376 step 43 | loss train: 1.466, val: 1.433 | iter time: 360.25 ms (step) remaining time: 0:36:28
|
| 146 |
+
Epoch 1 | iter 1408 step 44 | loss train: 1.475, val: 1.433 | iter time: 360.48 ms (step) remaining time: 0:36:17
|
| 147 |
+
Epoch 1 | iter 1440 step 45 | loss train: 1.467, val: 1.433 | iter time: 360.39 ms (step) remaining time: 0:36:06
|
| 148 |
+
Epoch 1 | iter 1472 step 46 | loss train: 1.461, val: 1.433 | iter time: 360.02 ms (step) remaining time: 0:35:55
|
| 149 |
+
Epoch 1 | iter 1504 step 47 | loss train: 1.331, val: 1.433 | iter time: 360.49 ms (step) remaining time: 0:35:44
|
| 150 |
+
Epoch 1 | iter 1536 step 48 | loss train: 1.468, val: 1.433 | iter time: 358.78 ms (step) remaining time: 0:35:33
|
| 151 |
+
Epoch 1 | iter 1568 step 49 | loss train: 1.398, val: 1.433 | iter time: 358.70 ms (step) remaining time: 0:35:22
|
| 152 |
+
Epoch 1 | iter 1600 step 50 | loss train: 1.427, val: 1.433 | iter time: 359.64 ms (step) remaining time: 0:35:11
|
| 153 |
+
Validating ...
|
| 154 |
+
iter 1600: val loss 1.4872, val time: 9390.22 ms
|
| 155 |
+
Epoch 1 | iter 1632 step 51 | loss train: 1.448, val: 1.487 | iter time: 361.16 ms (step) remaining time: 0:35:36
|
| 156 |
+
Epoch 1 | iter 1664 step 52 | loss train: 1.374, val: 1.487 | iter time: 361.60 ms (step) remaining time: 0:35:24
|
| 157 |
+
Epoch 1 | iter 1696 step 53 | loss train: 1.434, val: 1.487 | iter time: 358.98 ms (step) remaining time: 0:35:12
|
| 158 |
+
Epoch 1 | iter 1728 step 54 | loss train: 1.340, val: 1.487 | iter time: 359.62 ms (step) remaining time: 0:35:00
|
| 159 |
+
Epoch 1 | iter 1760 step 55 | loss train: 1.432, val: 1.487 | iter time: 359.79 ms (step) remaining time: 0:34:49
|
| 160 |
+
Epoch 1 | iter 1792 step 56 | loss train: 1.433, val: 1.487 | iter time: 358.30 ms (step) remaining time: 0:34:37
|
| 161 |
+
Epoch 1 | iter 1824 step 57 | loss train: 1.463, val: 1.487 | iter time: 357.91 ms (step) remaining time: 0:34:25
|
| 162 |
+
Epoch 1 | iter 1856 step 58 | loss train: 1.435, val: 1.487 | iter time: 361.36 ms (step) remaining time: 0:34:14
|
| 163 |
+
Epoch 1 | iter 1888 step 59 | loss train: 1.523, val: 1.487 | iter time: 360.72 ms (step) remaining time: 0:34:02
|
| 164 |
+
Epoch 1 | iter 1920 step 60 | loss train: 1.391, val: 1.487 | iter time: 358.96 ms (step) remaining time: 0:33:51
|
| 165 |
+
Epoch 1 | iter 1952 step 61 | loss train: 1.361, val: 1.487 | iter time: 358.85 ms (step) remaining time: 0:33:39
|
| 166 |
+
Epoch 1 | iter 1984 step 62 | loss train: 1.409, val: 1.487 | iter time: 360.52 ms (step) remaining time: 0:33:27
|
| 167 |
+
Epoch 1 | iter 2016 step 63 | loss train: 1.473, val: 1.487 | iter time: 359.89 ms (step) remaining time: 0:33:16
|
| 168 |
+
Epoch 1 | iter 2048 step 64 | loss train: 1.455, val: 1.487 | iter time: 360.09 ms (step) remaining time: 0:33:04
|
| 169 |
+
Epoch 1 | iter 2080 step 65 | loss train: 1.425, val: 1.487 | iter time: 360.64 ms (step) remaining time: 0:32:53
|
| 170 |
+
Epoch 1 | iter 2112 step 66 | loss train: 1.549, val: 1.487 | iter time: 361.60 ms (step) remaining time: 0:32:42
|
| 171 |
+
Epoch 1 | iter 2144 step 67 | loss train: 1.412, val: 1.487 | iter time: 361.50 ms (step) remaining time: 0:32:30
|
| 172 |
+
Epoch 1 | iter 2176 step 68 | loss train: 1.370, val: 1.487 | iter time: 358.55 ms (step) remaining time: 0:32:19
|
| 173 |
+
Epoch 1 | iter 2208 step 69 | loss train: 1.455, val: 1.487 | iter time: 360.57 ms (step) remaining time: 0:32:07
|
| 174 |
+
Epoch 1 | iter 2240 step 70 | loss train: 1.382, val: 1.487 | iter time: 360.11 ms (step) remaining time: 0:31:56
|
| 175 |
+
Epoch 1 | iter 2272 step 71 | loss train: 1.461, val: 1.487 | iter time: 359.48 ms (step) remaining time: 0:31:45
|
| 176 |
+
Epoch 1 | iter 2304 step 72 | loss train: 1.423, val: 1.487 | iter time: 359.50 ms (step) remaining time: 0:31:33
|
| 177 |
+
Epoch 1 | iter 2336 step 73 | loss train: 1.441, val: 1.487 | iter time: 358.78 ms (step) remaining time: 0:31:22
|
| 178 |
+
Epoch 1 | iter 2368 step 74 | loss train: 1.479, val: 1.487 | iter time: 359.35 ms (step) remaining time: 0:31:11
|
| 179 |
+
Epoch 1 | iter 2400 step 75 | loss train: 1.377, val: 1.487 | iter time: 360.39 ms (step) remaining time: 0:30:59
|
| 180 |
+
Epoch 1 | iter 2432 step 76 | loss train: 1.402, val: 1.487 | iter time: 360.92 ms (step) remaining time: 0:30:48
|
| 181 |
+
Epoch 1 | iter 2464 step 77 | loss train: 1.344, val: 1.487 | iter time: 360.01 ms (step) remaining time: 0:30:37
|
| 182 |
+
Epoch 1 | iter 2496 step 78 | loss train: 1.350, val: 1.487 | iter time: 359.29 ms (step) remaining time: 0:30:25
|
| 183 |
+
Epoch 1 | iter 2528 step 79 | loss train: 1.396, val: 1.487 | iter time: 359.02 ms (step) remaining time: 0:30:14
|
| 184 |
+
Epoch 1 | iter 2560 step 80 | loss train: 1.420, val: 1.487 | iter time: 358.99 ms (step) remaining time: 0:30:03
|
| 185 |
+
Epoch 1 | iter 2592 step 81 | loss train: 1.383, val: 1.487 | iter time: 360.35 ms (step) remaining time: 0:29:52
|
| 186 |
+
Epoch 1 | iter 2624 step 82 | loss train: 1.333, val: 1.487 | iter time: 361.65 ms (step) remaining time: 0:29:40
|
| 187 |
+
Epoch 1 | iter 2656 step 83 | loss train: 1.346, val: 1.487 | iter time: 361.18 ms (step) remaining time: 0:29:29
|
| 188 |
+
Epoch 1 | iter 2688 step 84 | loss train: 1.514, val: 1.487 | iter time: 360.03 ms (step) remaining time: 0:29:18
|
| 189 |
+
Epoch 1 | iter 2720 step 85 | loss train: 1.409, val: 1.487 | iter time: 360.69 ms (step) remaining time: 0:29:07
|
| 190 |
+
Epoch 1 | iter 2752 step 86 | loss train: 1.385, val: 1.487 | iter time: 362.28 ms (step) remaining time: 0:28:56
|
| 191 |
+
Epoch 1 | iter 2784 step 87 | loss train: 1.415, val: 1.487 | iter time: 359.15 ms (step) remaining time: 0:28:44
|
| 192 |
+
Epoch 1 | iter 2816 step 88 | loss train: 1.414, val: 1.487 | iter time: 358.59 ms (step) remaining time: 0:28:33
|
| 193 |
+
Epoch 1 | iter 2848 step 89 | loss train: 1.431, val: 1.487 | iter time: 361.58 ms (step) remaining time: 0:28:22
|
| 194 |
+
Epoch 1 | iter 2880 step 90 | loss train: 1.403, val: 1.487 | iter time: 360.82 ms (step) remaining time: 0:28:11
|
| 195 |
+
Epoch 1 | iter 2912 step 91 | loss train: 1.380, val: 1.487 | iter time: 361.98 ms (step) remaining time: 0:28:00
|
| 196 |
+
Epoch 1 | iter 2944 step 92 | loss train: 1.343, val: 1.487 | iter time: 359.92 ms (step) remaining time: 0:27:49
|
| 197 |
+
Epoch 1 | iter 2976 step 93 | loss train: 1.462, val: 1.487 | iter time: 362.02 ms (step) remaining time: 0:27:38
|
| 198 |
+
Epoch 1 | iter 3008 step 94 | loss train: 1.381, val: 1.487 | iter time: 359.42 ms (step) remaining time: 0:27:27
|
| 199 |
+
Epoch 1 | iter 3040 step 95 | loss train: 1.448, val: 1.487 | iter time: 359.42 ms (step) remaining time: 0:27:16
|
| 200 |
+
Epoch 1 | iter 3072 step 96 | loss train: 1.388, val: 1.487 | iter time: 361.60 ms (step) remaining time: 0:27:04
|
| 201 |
+
Epoch 1 | iter 3104 step 97 | loss train: 1.327, val: 1.487 | iter time: 359.88 ms (step) remaining time: 0:26:53
|
| 202 |
+
Epoch 1 | iter 3136 step 98 | loss train: 1.466, val: 1.487 | iter time: 361.19 ms (step) remaining time: 0:26:42
|
| 203 |
+
Epoch 1 | iter 3168 step 99 | loss train: 1.329, val: 1.487 | iter time: 360.18 ms (step) remaining time: 0:26:31
|
| 204 |
+
Epoch 1 | iter 3200 step 100 | loss train: 1.350, val: 1.487 | iter time: 359.05 ms (step) remaining time: 0:26:20
|
| 205 |
+
Validating ...
|
| 206 |
+
iter 3200: val loss 1.4740, val time: 9384.93 ms
|
| 207 |
+
Saving checkpoint to 'out/pretrain/2410/step-00000100/lit_model.pth'
|
| 208 |
+
Epoch 1 | iter 3232 step 101 | loss train: 1.296, val: 1.474 | iter time: 357.38 ms (step) remaining time: 0:26:46
|
| 209 |
+
Epoch 1 | iter 3264 step 102 | loss train: 1.428, val: 1.474 | iter time: 355.69 ms (step) remaining time: 0:26:34
|
| 210 |
+
Epoch 1 | iter 3296 step 103 | loss train: 1.448, val: 1.474 | iter time: 359.48 ms (step) remaining time: 0:26:22
|
| 211 |
+
Epoch 1 | iter 3328 step 104 | loss train: 1.404, val: 1.474 | iter time: 359.12 ms (step) remaining time: 0:26:11
|
| 212 |
+
Epoch 1 | iter 3360 step 105 | loss train: 1.388, val: 1.474 | iter time: 358.96 ms (step) remaining time: 0:25:59
|
| 213 |
+
Epoch 1 | iter 3392 step 106 | loss train: 1.379, val: 1.474 | iter time: 359.76 ms (step) remaining time: 0:25:47
|
| 214 |
+
Epoch 1 | iter 3424 step 107 | loss train: 1.381, val: 1.474 | iter time: 358.67 ms (step) remaining time: 0:25:36
|
| 215 |
+
Epoch 1 | iter 3456 step 108 | loss train: 1.447, val: 1.474 | iter time: 361.40 ms (step) remaining time: 0:25:24
|
| 216 |
+
Epoch 1 | iter 3488 step 109 | loss train: 1.409, val: 1.474 | iter time: 361.35 ms (step) remaining time: 0:25:12
|
| 217 |
+
Epoch 1 | iter 3520 step 110 | loss train: 1.466, val: 1.474 | iter time: 359.23 ms (step) remaining time: 0:25:01
|
| 218 |
+
Epoch 1 | iter 3552 step 111 | loss train: 1.318, val: 1.474 | iter time: 360.34 ms (step) remaining time: 0:24:49
|
| 219 |
+
Epoch 1 | iter 3584 step 112 | loss train: 1.326, val: 1.474 | iter time: 360.75 ms (step) remaining time: 0:24:38
|
| 220 |
+
Epoch 1 | iter 3616 step 113 | loss train: 1.371, val: 1.474 | iter time: 360.46 ms (step) remaining time: 0:24:26
|
| 221 |
+
Epoch 1 | iter 3648 step 114 | loss train: 1.399, val: 1.474 | iter time: 360.60 ms (step) remaining time: 0:24:14
|
| 222 |
+
Epoch 1 | iter 3680 step 115 | loss train: 1.436, val: 1.474 | iter time: 359.09 ms (step) remaining time: 0:24:03
|
| 223 |
+
Epoch 1 | iter 3712 step 116 | loss train: 1.362, val: 1.474 | iter time: 360.72 ms (step) remaining time: 0:23:51
|
| 224 |
+
Epoch 1 | iter 3744 step 117 | loss train: 1.312, val: 1.474 | iter time: 362.13 ms (step) remaining time: 0:23:40
|
| 225 |
+
Epoch 1 | iter 3776 step 118 | loss train: 1.358, val: 1.474 | iter time: 360.14 ms (step) remaining time: 0:23:28
|
| 226 |
+
Epoch 1 | iter 3808 step 119 | loss train: 1.335, val: 1.474 | iter time: 359.47 ms (step) remaining time: 0:23:17
|
| 227 |
+
Epoch 1 | iter 3840 step 120 | loss train: 1.353, val: 1.474 | iter time: 361.59 ms (step) remaining time: 0:23:05
|
| 228 |
+
Epoch 1 | iter 3872 step 121 | loss train: 1.338, val: 1.474 | iter time: 358.56 ms (step) remaining time: 0:22:54
|
| 229 |
+
Epoch 1 | iter 3904 step 122 | loss train: 1.269, val: 1.474 | iter time: 357.74 ms (step) remaining time: 0:22:42
|
| 230 |
+
Epoch 1 | iter 3936 step 123 | loss train: 1.350, val: 1.474 | iter time: 359.74 ms (step) remaining time: 0:22:31
|
| 231 |
+
Epoch 1 | iter 3968 step 124 | loss train: 1.408, val: 1.474 | iter time: 363.03 ms (step) remaining time: 0:22:19
|
| 232 |
+
Epoch 1 | iter 4000 step 125 | loss train: 1.413, val: 1.474 | iter time: 361.10 ms (step) remaining time: 0:22:08
|
| 233 |
+
Epoch 1 | iter 4032 step 126 | loss train: 1.361, val: 1.474 | iter time: 361.09 ms (step) remaining time: 0:21:56
|
| 234 |
+
Epoch 1 | iter 4064 step 127 | loss train: 1.339, val: 1.474 | iter time: 359.38 ms (step) remaining time: 0:21:45
|
| 235 |
+
Epoch 1 | iter 4096 step 128 | loss train: 1.353, val: 1.474 | iter time: 359.48 ms (step) remaining time: 0:21:34
|
| 236 |
+
Epoch 1 | iter 4128 step 129 | loss train: 1.406, val: 1.474 | iter time: 360.43 ms (step) remaining time: 0:21:22
|
| 237 |
+
Epoch 1 | iter 4160 step 130 | loss train: 1.392, val: 1.474 | iter time: 361.54 ms (step) remaining time: 0:21:11
|
| 238 |
+
Epoch 1 | iter 4192 step 131 | loss train: 1.394, val: 1.474 | iter time: 361.00 ms (step) remaining time: 0:20:59
|
| 239 |
+
Epoch 1 | iter 4224 step 132 | loss train: 1.390, val: 1.474 | iter time: 360.47 ms (step) remaining time: 0:20:48
|
| 240 |
+
Epoch 1 | iter 4256 step 133 | loss train: 1.364, val: 1.474 | iter time: 361.32 ms (step) remaining time: 0:20:37
|
| 241 |
+
Epoch 1 | iter 4288 step 134 | loss train: 1.315, val: 1.474 | iter time: 360.98 ms (step) remaining time: 0:20:25
|
| 242 |
+
Epoch 1 | iter 4320 step 135 | loss train: 1.338, val: 1.474 | iter time: 360.69 ms (step) remaining time: 0:20:14
|
| 243 |
+
Epoch 1 | iter 4352 step 136 | loss train: 1.309, val: 1.474 | iter time: 361.63 ms (step) remaining time: 0:20:03
|
| 244 |
+
Epoch 1 | iter 4384 step 137 | loss train: 1.281, val: 1.474 | iter time: 360.86 ms (step) remaining time: 0:19:51
|
| 245 |
+
Epoch 1 | iter 4416 step 138 | loss train: 1.396, val: 1.474 | iter time: 359.86 ms (step) remaining time: 0:19:40
|
| 246 |
+
Epoch 1 | iter 4448 step 139 | loss train: 1.356, val: 1.474 | iter time: 359.44 ms (step) remaining time: 0:19:29
|
| 247 |
+
Epoch 1 | iter 4480 step 140 | loss train: 1.416, val: 1.474 | iter time: 360.20 ms (step) remaining time: 0:19:17
|
| 248 |
+
Epoch 1 | iter 4512 step 141 | loss train: 1.449, val: 1.474 | iter time: 362.18 ms (step) remaining time: 0:19:06
|
| 249 |
+
Epoch 1 | iter 4544 step 142 | loss train: 1.456, val: 1.474 | iter time: 358.76 ms (step) remaining time: 0:18:55
|
| 250 |
+
Epoch 1 | iter 4576 step 143 | loss train: 1.363, val: 1.474 | iter time: 360.93 ms (step) remaining time: 0:18:44
|
| 251 |
+
Epoch 1 | iter 4608 step 144 | loss train: 1.339, val: 1.474 | iter time: 359.04 ms (step) remaining time: 0:18:32
|
| 252 |
+
Epoch 1 | iter 4640 step 145 | loss train: 1.378, val: 1.474 | iter time: 361.47 ms (step) remaining time: 0:18:21
|
| 253 |
+
Epoch 1 | iter 4672 step 146 | loss train: 1.417, val: 1.474 | iter time: 361.31 ms (step) remaining time: 0:18:10
|
| 254 |
+
Epoch 1 | iter 4704 step 147 | loss train: 1.392, val: 1.474 | iter time: 359.46 ms (step) remaining time: 0:17:58
|
| 255 |
+
Epoch 1 | iter 4736 step 148 | loss train: 1.326, val: 1.474 | iter time: 359.44 ms (step) remaining time: 0:17:47
|
| 256 |
+
Epoch 1 | iter 4768 step 149 | loss train: 1.336, val: 1.474 | iter time: 360.85 ms (step) remaining time: 0:17:36
|
| 257 |
+
Epoch 1 | iter 4800 step 150 | loss train: 1.320, val: 1.474 | iter time: 361.59 ms (step) remaining time: 0:17:25
|
| 258 |
+
Validating ...
|
| 259 |
+
iter 4800: val loss 1.4520, val time: 9382.39 ms
|
| 260 |
+
Epoch 1 | iter 4832 step 151 | loss train: 1.357, val: 1.452 | iter time: 361.72 ms (step) remaining time: 0:17:19
|
| 261 |
+
Epoch 1 | iter 4864 step 152 | loss train: 1.390, val: 1.452 | iter time: 361.82 ms (step) remaining time: 0:17:08
|
| 262 |
+
Epoch 1 | iter 4896 step 153 | loss train: 1.325, val: 1.452 | iter time: 359.57 ms (step) remaining time: 0:16:56
|
| 263 |
+
Epoch 1 | iter 4928 step 154 | loss train: 1.330, val: 1.452 | iter time: 358.68 ms (step) remaining time: 0:16:45
|
| 264 |
+
Epoch 1 | iter 4960 step 155 | loss train: 1.378, val: 1.452 | iter time: 359.74 ms (step) remaining time: 0:16:34
|
| 265 |
+
Epoch 1 | iter 4992 step 156 | loss train: 1.380, val: 1.452 | iter time: 359.44 ms (step) remaining time: 0:16:22
|
| 266 |
+
Epoch 1 | iter 5024 step 157 | loss train: 1.326, val: 1.452 | iter time: 360.72 ms (step) remaining time: 0:16:11
|
| 267 |
+
Epoch 1 | iter 5056 step 158 | loss train: 1.331, val: 1.452 | iter time: 359.16 ms (step) remaining time: 0:16:00
|
| 268 |
+
Epoch 1 | iter 5088 step 159 | loss train: 1.312, val: 1.452 | iter time: 360.73 ms (step) remaining time: 0:15:48
|
| 269 |
+
Epoch 1 | iter 5120 step 160 | loss train: 1.320, val: 1.452 | iter time: 358.48 ms (step) remaining time: 0:15:37
|
| 270 |
+
Epoch 1 | iter 5152 step 161 | loss train: 1.425, val: 1.452 | iter time: 361.04 ms (step) remaining time: 0:15:26
|
| 271 |
+
Epoch 1 | iter 5184 step 162 | loss train: 1.359, val: 1.452 | iter time: 359.68 ms (step) remaining time: 0:15:15
|
| 272 |
+
Epoch 1 | iter 5216 step 163 | loss train: 1.322, val: 1.452 | iter time: 359.98 ms (step) remaining time: 0:15:03
|
| 273 |
+
Epoch 1 | iter 5248 step 164 | loss train: 1.362, val: 1.452 | iter time: 360.02 ms (step) remaining time: 0:14:52
|
| 274 |
+
Epoch 1 | iter 5280 step 165 | loss train: 1.416, val: 1.452 | iter time: 360.96 ms (step) remaining time: 0:14:41
|
| 275 |
+
Epoch 1 | iter 5312 step 166 | loss train: 1.361, val: 1.452 | iter time: 360.85 ms (step) remaining time: 0:14:29
|
| 276 |
+
Epoch 1 | iter 5344 step 167 | loss train: 1.284, val: 1.452 | iter time: 361.93 ms (step) remaining time: 0:14:18
|
| 277 |
+
Epoch 1 | iter 5376 step 168 | loss train: 1.341, val: 1.452 | iter time: 361.32 ms (step) remaining time: 0:14:07
|
| 278 |
+
Epoch 1 | iter 5408 step 169 | loss train: 1.354, val: 1.452 | iter time: 360.35 ms (step) remaining time: 0:13:56
|
| 279 |
+
Epoch 1 | iter 5440 step 170 | loss train: 1.324, val: 1.452 | iter time: 360.10 ms (step) remaining time: 0:13:44
|
| 280 |
+
Epoch 1 | iter 5472 step 171 | loss train: 1.276, val: 1.452 | iter time: 360.18 ms (step) remaining time: 0:13:33
|
| 281 |
+
Epoch 1 | iter 5504 step 172 | loss train: 1.378, val: 1.452 | iter time: 358.51 ms (step) remaining time: 0:13:22
|
| 282 |
+
Epoch 1 | iter 5536 step 173 | loss train: 1.440, val: 1.452 | iter time: 360.06 ms (step) remaining time: 0:13:11
|
| 283 |
+
Epoch 1 | iter 5568 step 174 | loss train: 1.353, val: 1.452 | iter time: 361.86 ms (step) remaining time: 0:12:59
|
| 284 |
+
Epoch 1 | iter 5600 step 175 | loss train: 1.355, val: 1.452 | iter time: 359.46 ms (step) remaining time: 0:12:48
|
| 285 |
+
Epoch 1 | iter 5632 step 176 | loss train: 1.359, val: 1.452 | iter time: 360.40 ms (step) remaining time: 0:12:37
|
| 286 |
+
Epoch 1 | iter 5664 step 177 | loss train: 1.381, val: 1.452 | iter time: 362.49 ms (step) remaining time: 0:12:26
|
| 287 |
+
Epoch 1 | iter 5696 step 178 | loss train: 1.301, val: 1.452 | iter time: 360.00 ms (step) remaining time: 0:12:14
|
| 288 |
+
Epoch 1 | iter 5728 step 179 | loss train: 1.365, val: 1.452 | iter time: 361.17 ms (step) remaining time: 0:12:03
|
| 289 |
+
Epoch 1 | iter 5760 step 180 | loss train: 1.239, val: 1.452 | iter time: 361.51 ms (step) remaining time: 0:11:52
|
| 290 |
+
Epoch 1 | iter 5792 step 181 | loss train: 1.363, val: 1.452 | iter time: 358.99 ms (step) remaining time: 0:11:41
|
| 291 |
+
Epoch 1 | iter 5824 step 182 | loss train: 1.392, val: 1.452 | iter time: 361.74 ms (step) remaining time: 0:11:29
|
| 292 |
+
Epoch 1 | iter 5856 step 183 | loss train: 1.356, val: 1.452 | iter time: 359.16 ms (step) remaining time: 0:11:18
|
| 293 |
+
Epoch 1 | iter 5888 step 184 | loss train: 1.277, val: 1.452 | iter time: 360.23 ms (step) remaining time: 0:11:07
|
| 294 |
+
Epoch 1 | iter 5920 step 185 | loss train: 1.300, val: 1.452 | iter time: 360.14 ms (step) remaining time: 0:10:56
|
| 295 |
+
Epoch 1 | iter 5952 step 186 | loss train: 1.326, val: 1.452 | iter time: 361.37 ms (step) remaining time: 0:10:45
|
| 296 |
+
Epoch 1 | iter 5984 step 187 | loss train: 1.311, val: 1.452 | iter time: 360.23 ms (step) remaining time: 0:10:33
|
| 297 |
+
Epoch 1 | iter 6016 step 188 | loss train: 1.275, val: 1.452 | iter time: 359.46 ms (step) remaining time: 0:10:22
|
| 298 |
+
Epoch 1 | iter 6048 step 189 | loss train: 1.363, val: 1.452 | iter time: 360.62 ms (step) remaining time: 0:10:11
|
| 299 |
+
Epoch 1 | iter 6080 step 190 | loss train: 1.369, val: 1.452 | iter time: 360.38 ms (step) remaining time: 0:10:00
|
| 300 |
+
Epoch 1 | iter 6112 step 191 | loss train: 1.333, val: 1.452 | iter time: 360.70 ms (step) remaining time: 0:09:49
|
| 301 |
+
Epoch 1 | iter 6144 step 192 | loss train: 1.371, val: 1.452 | iter time: 360.33 ms (step) remaining time: 0:09:38
|
| 302 |
+
Epoch 1 | iter 6176 step 193 | loss train: 1.425, val: 1.452 | iter time: 360.49 ms (step) remaining time: 0:09:26
|
| 303 |
+
Epoch 1 | iter 6208 step 194 | loss train: 1.347, val: 1.452 | iter time: 360.85 ms (step) remaining time: 0:09:15
|
| 304 |
+
Epoch 1 | iter 6240 step 195 | loss train: 1.347, val: 1.452 | iter time: 359.56 ms (step) remaining time: 0:09:04
|
| 305 |
+
Epoch 1 | iter 6272 step 196 | loss train: 1.288, val: 1.452 | iter time: 360.48 ms (step) remaining time: 0:08:53
|
| 306 |
+
Epoch 1 | iter 6304 step 197 | loss train: 1.360, val: 1.452 | iter time: 360.64 ms (step) remaining time: 0:08:42
|
| 307 |
+
Epoch 1 | iter 6336 step 198 | loss train: 1.344, val: 1.452 | iter time: 361.55 ms (step) remaining time: 0:08:31
|
| 308 |
+
Epoch 1 | iter 6368 step 199 | loss train: 1.407, val: 1.452 | iter time: 360.21 ms (step) remaining time: 0:08:19
|
| 309 |
+
Epoch 1 | iter 6400 step 200 | loss train: 1.399, val: 1.452 | iter time: 362.19 ms (step) remaining time: 0:08:08
|
| 310 |
+
Validating ...
|
| 311 |
+
iter 6400: val loss 1.4320, val time: 9372.73 ms
|
| 312 |
+
Saving checkpoint to 'out/pretrain/2410/step-00000200/lit_model.pth'
|
| 313 |
+
Epoch 1 | iter 6432 step 201 | loss train: 1.332, val: 1.432 | iter time: 357.97 ms (step) remaining time: 0:08:03
|
| 314 |
+
Epoch 1 | iter 6464 step 202 | loss train: 1.368, val: 1.432 | iter time: 358.87 ms (step) remaining time: 0:07:51
|
| 315 |
+
Epoch 1 | iter 6496 step 203 | loss train: 1.373, val: 1.432 | iter time: 358.71 ms (step) remaining time: 0:07:40
|
| 316 |
+
Epoch 1 | iter 6528 step 204 | loss train: 1.342, val: 1.432 | iter time: 361.87 ms (step) remaining time: 0:07:29
|
| 317 |
+
Epoch 1 | iter 6560 step 205 | loss train: 1.275, val: 1.432 | iter time: 360.69 ms (step) remaining time: 0:07:17
|
| 318 |
+
Epoch 1 | iter 6592 step 206 | loss train: 1.338, val: 1.432 | iter time: 363.39 ms (step) remaining time: 0:07:06
|
| 319 |
+
Epoch 1 | iter 6624 step 207 | loss train: 1.325, val: 1.432 | iter time: 361.00 ms (step) remaining time: 0:06:55
|
| 320 |
+
Epoch 1 | iter 6656 step 208 | loss train: 1.316, val: 1.432 | iter time: 361.32 ms (step) remaining time: 0:06:44
|
| 321 |
+
Epoch 1 | iter 6688 step 209 | loss train: 1.314, val: 1.432 | iter time: 358.59 ms (step) remaining time: 0:06:32
|
| 322 |
+
Epoch 1 | iter 6720 step 210 | loss train: 1.293, val: 1.432 | iter time: 359.86 ms (step) remaining time: 0:06:21
|
| 323 |
+
Epoch 1 | iter 6752 step 211 | loss train: 1.294, val: 1.432 | iter time: 359.83 ms (step) remaining time: 0:06:10
|
| 324 |
+
Epoch 1 | iter 6784 step 212 | loss train: 1.266, val: 1.432 | iter time: 359.81 ms (step) remaining time: 0:05:58
|
| 325 |
+
Epoch 1 | iter 6816 step 213 | loss train: 1.242, val: 1.432 | iter time: 359.95 ms (step) remaining time: 0:05:47
|
| 326 |
+
Epoch 1 | iter 6848 step 214 | loss train: 1.295, val: 1.432 | iter time: 360.34 ms (step) remaining time: 0:05:36
|
| 327 |
+
Epoch 1 | iter 6880 step 215 | loss train: 1.375, val: 1.432 | iter time: 358.12 ms (step) remaining time: 0:05:25
|
| 328 |
+
Epoch 1 | iter 6912 step 216 | loss train: 1.295, val: 1.432 | iter time: 359.80 ms (step) remaining time: 0:05:13
|
| 329 |
+
Epoch 1 | iter 6944 step 217 | loss train: 1.303, val: 1.432 | iter time: 358.80 ms (step) remaining time: 0:05:02
|
| 330 |
+
Epoch 1 | iter 6976 step 218 | loss train: 1.301, val: 1.432 | iter time: 358.60 ms (step) remaining time: 0:04:51
|
| 331 |
+
Epoch 1 | iter 7008 step 219 | loss train: 1.315, val: 1.432 | iter time: 359.93 ms (step) remaining time: 0:04:40
|
| 332 |
+
Epoch 1 | iter 7040 step 220 | loss train: 1.302, val: 1.432 | iter time: 359.42 ms (step) remaining time: 0:04:28
|
| 333 |
+
Epoch 1 | iter 7072 step 221 | loss train: 1.305, val: 1.432 | iter time: 358.54 ms (step) remaining time: 0:04:17
|
| 334 |
+
Epoch 1 | iter 7104 step 222 | loss train: 1.300, val: 1.432 | iter time: 360.42 ms (step) remaining time: 0:04:06
|
| 335 |
+
Epoch 1 | iter 7136 step 223 | loss train: 1.326, val: 1.432 | iter time: 360.95 ms (step) remaining time: 0:03:55
|
| 336 |
+
Epoch 1 | iter 7168 step 224 | loss train: 1.359, val: 1.432 | iter time: 359.79 ms (step) remaining time: 0:03:43
|
| 337 |
+
Epoch 1 | iter 7200 step 225 | loss train: 1.401, val: 1.432 | iter time: 360.84 ms (step) remaining time: 0:03:32
|
| 338 |
+
Epoch 1 | iter 7232 step 226 | loss train: 1.359, val: 1.432 | iter time: 361.45 ms (step) remaining time: 0:03:21
|
| 339 |
+
Epoch 1 | iter 7264 step 227 | loss train: 1.390, val: 1.432 | iter time: 360.51 ms (step) remaining time: 0:03:10
|
| 340 |
+
Epoch 1 | iter 7296 step 228 | loss train: 1.358, val: 1.432 | iter time: 360.34 ms (step) remaining time: 0:02:59
|
| 341 |
+
Epoch 1 | iter 7328 step 229 | loss train: 1.322, val: 1.432 | iter time: 359.47 ms (step) remaining time: 0:02:47
|
| 342 |
+
Epoch 1 | iter 7360 step 230 | loss train: 1.335, val: 1.432 | iter time: 359.69 ms (step) remaining time: 0:02:36
|
| 343 |
+
Epoch 1 | iter 7392 step 231 | loss train: 1.316, val: 1.432 | iter time: 361.00 ms (step) remaining time: 0:02:25
|
| 344 |
+
Epoch 1 | iter 7424 step 232 | loss train: 1.289, val: 1.432 | iter time: 359.70 ms (step) remaining time: 0:02:14
|
| 345 |
+
Epoch 1 | iter 7456 step 233 | loss train: 1.184, val: 1.432 | iter time: 358.77 ms (step) remaining time: 0:02:03
|
| 346 |
+
Epoch 1 | iter 7488 step 234 | loss train: 1.281, val: 1.432 | iter time: 361.00 ms (step) remaining time: 0:01:51
|
| 347 |
+
Epoch 1 | iter 7520 step 235 | loss train: 1.274, val: 1.432 | iter time: 360.44 ms (step) remaining time: 0:01:40
|
| 348 |
+
Epoch 1 | iter 7552 step 236 | loss train: 1.313, val: 1.432 | iter time: 359.87 ms (step) remaining time: 0:01:29
|
| 349 |
+
Epoch 1 | iter 7584 step 237 | loss train: 1.313, val: 1.432 | iter time: 360.86 ms (step) remaining time: 0:01:18
|
| 350 |
+
Epoch 1 | iter 7616 step 238 | loss train: 1.296, val: 1.432 | iter time: 361.32 ms (step) remaining time: 0:01:07
|
| 351 |
+
Epoch 1 | iter 7648 step 239 | loss train: 1.318, val: 1.432 | iter time: 359.93 ms (step) remaining time: 0:00:55
|
| 352 |
+
Epoch 1 | iter 7680 step 240 | loss train: 1.270, val: 1.432 | iter time: 358.38 ms (step) remaining time: 0:00:44
|
| 353 |
+
Epoch 1 | iter 7712 step 241 | loss train: 1.349, val: 1.432 | iter time: 361.13 ms (step) remaining time: 0:00:33
|
| 354 |
+
Epoch 1 | iter 7744 step 242 | loss train: 1.299, val: 1.432 | iter time: 359.81 ms (step) remaining time: 0:00:22
|
| 355 |
+
Epoch 1 | iter 7776 step 243 | loss train: 1.362, val: 1.432 | iter time: 360.26 ms (step) remaining time: 0:00:11
|
| 356 |
+
Epoch 1 | iter 7808 step 244 | loss train: 1.311, val: 1.432 | iter time: 360.30 ms (step) remaining time: 0:00:00
|
| 357 |
+
Validating ...
|
| 358 |
+
Final evaluation | val loss: 1.422 | val ppl: 4.145
|
| 359 |
+
Saving checkpoint to 'out/pretrain/2410/final/lit_model.pth'
|
| 360 |
+
----------------------------------------
|
| 361 |
+
| Performance
|
| 362 |
+
| - Total tokens : 255,852,544
|
| 363 |
+
| - Training Time : 2764.77 s
|
| 364 |
+
| - Tok/sec : 229.35 tok/s
|
| 365 |
+
| ----------------------------------------
|
| 366 |
+
| Memory Usage
|
| 367 |
+
| - Memory Used : 26.32 GB
|
| 368 |
+
----------------------------------------
|
out/pretrain/tinyllama/teelogs/2410_full.txt
ADDED
|
@@ -0,0 +1,380 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
Initializing distributed: GLOBAL_RANK: 0, MEMBER: 1/4
|
| 2 |
+
Initializing distributed: GLOBAL_RANK: 3, MEMBER: 4/4
|
| 3 |
+
[rank: 3] Seed set to 42
|
| 4 |
+
Initializing distributed: GLOBAL_RANK: 2, MEMBER: 3/4
|
| 5 |
+
Initializing distributed: GLOBAL_RANK: 1, MEMBER: 2/4
|
| 6 |
+
[rank: 2] Seed set to 42
|
| 7 |
+
----------------------------------------------------------------------------------------------------
|
| 8 |
+
distributed_backend=nccl
|
| 9 |
+
All distributed processes registered. Starting with 4 processes
|
| 10 |
+
----------------------------------------------------------------------------------------------------
|
| 11 |
+
|
| 12 |
+
[rank: 1] Seed set to 42
|
| 13 |
+
[rank: 0] Seed set to 42
|
| 14 |
+
/mnt/data/litgpt/litgpt/pretrain.py:515: UserWarning: A newer version of litdata is available (0.2.52). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 15 |
+
train_dataloader = data.train_dataloader()
|
| 16 |
+
/mnt/data/litgpt/litgpt/pretrain.py:515: UserWarning: A newer version of litdata is available (0.2.52). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 17 |
+
train_dataloader = data.train_dataloader()
|
| 18 |
+
/mnt/data/litgpt/litgpt/pretrain.py:515: UserWarning: A newer version of litdata is available (0.2.52). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 19 |
+
train_dataloader = data.train_dataloader()
|
| 20 |
+
/mnt/data/litgpt/litgpt/pretrain.py:515: UserWarning: A newer version of litdata is available (0.2.52). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 21 |
+
train_dataloader = data.train_dataloader()
|
| 22 |
+
All GPUs are fully connected via NVLink.
|
| 23 |
+
{'data': {'batch_size': 1,
|
| 24 |
+
'data_path': PosixPath('litgpt/data/arxiv'),
|
| 25 |
+
'num_workers': 8,
|
| 26 |
+
'seed': 42,
|
| 27 |
+
'seq_length': 2048,
|
| 28 |
+
'use_starcoder': True},
|
| 29 |
+
'data_dir': PosixPath('litgpt/data/arxiv/2410'),
|
| 30 |
+
'devices': 'auto',
|
| 31 |
+
'eval': {'evaluate_example': 'first',
|
| 32 |
+
'final_validation': True,
|
| 33 |
+
'initial_validation': True,
|
| 34 |
+
'interval': 50,
|
| 35 |
+
'max_iters': 200,
|
| 36 |
+
'max_new_tokens': None},
|
| 37 |
+
'initial_checkpoint_dir': PosixPath('out/pretrain/2409_full/final'),
|
| 38 |
+
'log': {'group': None, 'project': None, 'run': None},
|
| 39 |
+
'logger_name': 'tensorboard',
|
| 40 |
+
'model_config': {'attention_logit_softcapping': None,
|
| 41 |
+
'attention_scores_scalar': None,
|
| 42 |
+
'attn_bias': False,
|
| 43 |
+
'bias': False,
|
| 44 |
+
'block_size': 2048,
|
| 45 |
+
'final_logit_softcapping': None,
|
| 46 |
+
'gelu_approximate': 'none',
|
| 47 |
+
'head_size': 64,
|
| 48 |
+
'hf_config': {'name': 'TinyLlama-1.1B-intermediate-step-1431k-3T',
|
| 49 |
+
'org': 'TinyLlama'},
|
| 50 |
+
'intermediate_size': 5632,
|
| 51 |
+
'lm_head_bias': False,
|
| 52 |
+
'mlp_class_name': 'LLaMAMLP',
|
| 53 |
+
'moe_intermediate_size': None,
|
| 54 |
+
'n_embd': 2048,
|
| 55 |
+
'n_expert': 0,
|
| 56 |
+
'n_expert_per_token': 0,
|
| 57 |
+
'n_head': 32,
|
| 58 |
+
'n_layer': 22,
|
| 59 |
+
'n_query_groups': 4,
|
| 60 |
+
'name': 'tiny-llama-1.1b',
|
| 61 |
+
'norm_1': True,
|
| 62 |
+
'norm_2': True,
|
| 63 |
+
'norm_class_name': 'RMSNorm',
|
| 64 |
+
'norm_eps': 1e-05,
|
| 65 |
+
'norm_qk': False,
|
| 66 |
+
'norm_qk_type': 'default',
|
| 67 |
+
'padded_vocab_size': 32000,
|
| 68 |
+
'padding_multiple': 64,
|
| 69 |
+
'parallel_residual': False,
|
| 70 |
+
'post_attention_norm': False,
|
| 71 |
+
'post_mlp_norm': False,
|
| 72 |
+
'rope_adjustments': None,
|
| 73 |
+
'rope_base': 10000,
|
| 74 |
+
'rope_condense_ratio': 1,
|
| 75 |
+
'rope_indices': None,
|
| 76 |
+
'rope_local_base_freq': None,
|
| 77 |
+
'rotary_percentage': 1.0,
|
| 78 |
+
'scale_embeddings': False,
|
| 79 |
+
'shared_attention_norm': False,
|
| 80 |
+
'sliding_window_indices': None,
|
| 81 |
+
'sliding_window_size': None,
|
| 82 |
+
'vocab_size': 32000},
|
| 83 |
+
'model_name': 'tiny-llama-1.1b',
|
| 84 |
+
'num_nodes': 1,
|
| 85 |
+
'optimizer': "{'class_path': 'torch.optim.AdamW', 'init_args': {'lr': 0.0004, "
|
| 86 |
+
"'weight_decay': 0.1, 'betas': [0.9, 0.95]}}",
|
| 87 |
+
'out_dir': PosixPath('out/pretrain/2410_full'),
|
| 88 |
+
'ppl': False,
|
| 89 |
+
'precision': 'bf16-mixed',
|
| 90 |
+
'resume': False,
|
| 91 |
+
'seed': 42,
|
| 92 |
+
'tokenizer_dir': PosixPath('checkpoints/TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T'),
|
| 93 |
+
'train': {'epochs': None,
|
| 94 |
+
'global_batch_size': 512,
|
| 95 |
+
'log_interval': 1,
|
| 96 |
+
'lr_warmup_fraction': None,
|
| 97 |
+
'lr_warmup_steps': 20,
|
| 98 |
+
'max_norm': 1.0,
|
| 99 |
+
'max_seq_length': 2048,
|
| 100 |
+
'max_steps': None,
|
| 101 |
+
'max_tokens': 258998272,
|
| 102 |
+
'micro_batch_size': 4,
|
| 103 |
+
'min_lr': 4e-05,
|
| 104 |
+
'save_interval': 100,
|
| 105 |
+
'tie_embeddings': None}}
|
| 106 |
+
Time to instantiate model: 0.03 seconds.
|
| 107 |
+
Total parameters: 1,100,048,384
|
| 108 |
+
[fix] out/pretrain/2409_full/final/lit_model.pth 是整包 state，提取 model 权重并原子覆盖...
|
| 109 |
+
[fix] 已覆盖为纯权重: out/pretrain/2409_full/final/lit_model.pth
|
| 110 |
+
Validating ...
|
| 111 |
+
Measured TFLOPs: 239.66
|
| 112 |
+
Epoch 1 | iter 32 step 1 | loss train: 1.468, val: 1.434 | iter time: 549.16 ms (step) remaining time: 0:49:22
|
| 113 |
+
Epoch 1 | iter 64 step 2 | loss train: 1.464, val: 1.434 | iter time: 357.02 ms (step) remaining time: 0:46:36
|
| 114 |
+
Epoch 1 | iter 96 step 3 | loss train: 1.444, val: 1.434 | iter time: 358.26 ms (step) remaining time: 0:45:35
|
| 115 |
+
Epoch 1 | iter 128 step 4 | loss train: 1.446, val: 1.434 | iter time: 359.15 ms (step) remaining time: 0:45:01
|
| 116 |
+
Epoch 1 | iter 160 step 5 | loss train: 1.403, val: 1.434 | iter time: 358.13 ms (step) remaining time: 0:44:36
|
| 117 |
+
Epoch 1 | iter 192 step 6 | loss train: 1.478, val: 1.434 | iter time: 358.23 ms (step) remaining time: 0:44:16
|
| 118 |
+
Epoch 1 | iter 224 step 7 | loss train: 1.494, val: 1.434 | iter time: 358.74 ms (step) remaining time: 0:43:59
|
| 119 |
+
Epoch 1 | iter 256 step 8 | loss train: 1.459, val: 1.434 | iter time: 358.28 ms (step) remaining time: 0:43:43
|
| 120 |
+
Epoch 1 | iter 288 step 9 | loss train: 1.464, val: 1.434 | iter time: 359.45 ms (step) remaining time: 0:43:29
|
| 121 |
+
Epoch 1 | iter 320 step 10 | loss train: 1.448, val: 1.434 | iter time: 358.44 ms (step) remaining time: 0:43:15
|
| 122 |
+
Epoch 1 | iter 352 step 11 | loss train: 1.456, val: 1.434 | iter time: 358.89 ms (step) remaining time: 0:43:02
|
| 123 |
+
Epoch 1 | iter 384 step 12 | loss train: 1.468, val: 1.434 | iter time: 359.20 ms (step) remaining time: 0:42:50
|
| 124 |
+
Epoch 1 | iter 416 step 13 | loss train: 1.488, val: 1.434 | iter time: 358.92 ms (step) remaining time: 0:42:37
|
| 125 |
+
Epoch 1 | iter 448 step 14 | loss train: 1.440, val: 1.434 | iter time: 359.96 ms (step) remaining time: 0:42:25
|
| 126 |
+
Epoch 1 | iter 480 step 15 | loss train: 1.453, val: 1.434 | iter time: 359.32 ms (step) remaining time: 0:42:13
|
| 127 |
+
Epoch 1 | iter 512 step 16 | loss train: 1.418, val: 1.434 | iter time: 359.56 ms (step) remaining time: 0:42:01
|
| 128 |
+
Epoch 1 | iter 544 step 17 | loss train: 1.485, val: 1.434 | iter time: 360.81 ms (step) remaining time: 0:41:50
|
| 129 |
+
Epoch 1 | iter 576 step 18 | loss train: 1.456, val: 1.434 | iter time: 361.53 ms (step) remaining time: 0:41:38
|
| 130 |
+
Epoch 1 | iter 608 step 19 | loss train: 1.488, val: 1.434 | iter time: 358.12 ms (step) remaining time: 0:41:27
|
| 131 |
+
Epoch 1 | iter 640 step 20 | loss train: 1.483, val: 1.434 | iter time: 358.79 ms (step) remaining time: 0:41:15
|
| 132 |
+
Epoch 1 | iter 672 step 21 | loss train: 1.458, val: 1.434 | iter time: 360.80 ms (step) remaining time: 0:41:04
|
| 133 |
+
Epoch 1 | iter 704 step 22 | loss train: 1.487, val: 1.434 | iter time: 360.74 ms (step) remaining time: 0:40:53
|
| 134 |
+
Epoch 1 | iter 736 step 23 | loss train: 1.429, val: 1.434 | iter time: 360.30 ms (step) remaining time: 0:40:41
|
| 135 |
+
Epoch 1 | iter 768 step 24 | loss train: 1.426, val: 1.434 | iter time: 359.60 ms (step) remaining time: 0:40:30
|
| 136 |
+
Epoch 1 | iter 800 step 25 | loss train: 1.426, val: 1.434 | iter time: 360.50 ms (step) remaining time: 0:40:19
|
| 137 |
+
Epoch 1 | iter 832 step 26 | loss train: 1.480, val: 1.434 | iter time: 360.38 ms (step) remaining time: 0:40:08
|
| 138 |
+
Epoch 1 | iter 864 step 27 | loss train: 1.421, val: 1.434 | iter time: 359.78 ms (step) remaining time: 0:39:57
|
| 139 |
+
Epoch 1 | iter 896 step 28 | loss train: 1.395, val: 1.434 | iter time: 361.38 ms (step) remaining time: 0:39:45
|
| 140 |
+
Epoch 1 | iter 928 step 29 | loss train: 1.423, val: 1.434 | iter time: 359.38 ms (step) remaining time: 0:39:34
|
| 141 |
+
Epoch 1 | iter 960 step 30 | loss train: 1.510, val: 1.434 | iter time: 360.48 ms (step) remaining time: 0:39:23
|
| 142 |
+
Epoch 1 | iter 992 step 31 | loss train: 1.375, val: 1.434 | iter time: 357.96 ms (step) remaining time: 0:39:12
|
| 143 |
+
Epoch 1 | iter 1024 step 32 | loss train: 1.507, val: 1.434 | iter time: 362.19 ms (step) remaining time: 0:39:01
|
| 144 |
+
Epoch 1 | iter 1056 step 33 | loss train: 1.426, val: 1.434 | iter time: 360.28 ms (step) remaining time: 0:38:50
|
| 145 |
+
Epoch 1 | iter 1088 step 34 | loss train: 1.427, val: 1.434 | iter time: 360.18 ms (step) remaining time: 0:38:39
|
| 146 |
+
Epoch 1 | iter 1120 step 35 | loss train: 1.463, val: 1.434 | iter time: 361.08 ms (step) remaining time: 0:38:28
|
| 147 |
+
Epoch 1 | iter 1152 step 36 | loss train: 1.465, val: 1.434 | iter time: 361.35 ms (step) remaining time: 0:38:17
|
| 148 |
+
Epoch 1 | iter 1184 step 37 | loss train: 1.501, val: 1.434 | iter time: 361.16 ms (step) remaining time: 0:38:06
|
| 149 |
+
Epoch 1 | iter 1216 step 38 | loss train: 1.468, val: 1.434 | iter time: 361.31 ms (step) remaining time: 0:37:55
|
| 150 |
+
Epoch 1 | iter 1248 step 39 | loss train: 1.369, val: 1.434 | iter time: 361.01 ms (step) remaining time: 0:37:44
|
| 151 |
+
Epoch 1 | iter 1280 step 40 | loss train: 1.486, val: 1.434 | iter time: 359.81 ms (step) remaining time: 0:37:33
|
| 152 |
+
Epoch 1 | iter 1312 step 41 | loss train: 1.444, val: 1.434 | iter time: 360.68 ms (step) remaining time: 0:37:22
|
| 153 |
+
Epoch 1 | iter 1344 step 42 | loss train: 1.534, val: 1.434 | iter time: 358.77 ms (step) remaining time: 0:37:13
|
| 154 |
+
Epoch 1 | iter 1376 step 43 | loss train: 1.465, val: 1.434 | iter time: 358.65 ms (step) remaining time: 0:37:02
|
| 155 |
+
Epoch 1 | iter 1408 step 44 | loss train: 1.473, val: 1.434 | iter time: 359.95 ms (step) remaining time: 0:36:51
|
| 156 |
+
Epoch 1 | iter 1440 step 45 | loss train: 1.463, val: 1.434 | iter time: 359.10 ms (step) remaining time: 0:36:40
|
| 157 |
+
Epoch 1 | iter 1472 step 46 | loss train: 1.460, val: 1.434 | iter time: 358.77 ms (step) remaining time: 0:36:29
|
| 158 |
+
Epoch 1 | iter 1504 step 47 | loss train: 1.331, val: 1.434 | iter time: 360.40 ms (step) remaining time: 0:36:18
|
| 159 |
+
Epoch 1 | iter 1536 step 48 | loss train: 1.467, val: 1.434 | iter time: 363.89 ms (step) remaining time: 0:36:07
|
| 160 |
+
Epoch 1 | iter 1568 step 49 | loss train: 1.397, val: 1.434 | iter time: 360.78 ms (step) remaining time: 0:35:56
|
| 161 |
+
Epoch 1 | iter 1600 step 50 | loss train: 1.426, val: 1.434 | iter time: 359.65 ms (step) remaining time: 0:35:45
|
| 162 |
+
Validating ...
|
| 163 |
+
iter 1600: val loss 1.4374, val time: 22390.38 ms
|
| 164 |
+
Epoch 1 | iter 1632 step 51 | loss train: 1.448, val: 1.437 | iter time: 361.75 ms (step) remaining time: 0:37:03
|
| 165 |
+
Epoch 1 | iter 1664 step 52 | loss train: 1.374, val: 1.437 | iter time: 360.65 ms (step) remaining time: 0:36:49
|
| 166 |
+
Epoch 1 | iter 1696 step 53 | loss train: 1.432, val: 1.437 | iter time: 358.93 ms (step) remaining time: 0:36:36
|
| 167 |
+
Epoch 1 | iter 1728 step 54 | loss train: 1.338, val: 1.437 | iter time: 360.48 ms (step) remaining time: 0:36:23
|
| 168 |
+
Epoch 1 | iter 1760 step 55 | loss train: 1.431, val: 1.437 | iter time: 359.98 ms (step) remaining time: 0:36:11
|
| 169 |
+
Epoch 1 | iter 1792 step 56 | loss train: 1.432, val: 1.437 | iter time: 361.41 ms (step) remaining time: 0:35:58
|
| 170 |
+
Epoch 1 | iter 1824 step 57 | loss train: 1.460, val: 1.437 | iter time: 359.26 ms (step) remaining time: 0:35:45
|
| 171 |
+
Epoch 1 | iter 1856 step 58 | loss train: 1.434, val: 1.437 | iter time: 361.14 ms (step) remaining time: 0:35:32
|
| 172 |
+
Epoch 1 | iter 1888 step 59 | loss train: 1.520, val: 1.437 | iter time: 358.19 ms (step) remaining time: 0:35:20
|
| 173 |
+
Epoch 1 | iter 1920 step 60 | loss train: 1.389, val: 1.437 | iter time: 359.54 ms (step) remaining time: 0:35:07
|
| 174 |
+
Epoch 1 | iter 1952 step 61 | loss train: 1.361, val: 1.437 | iter time: 358.54 ms (step) remaining time: 0:34:55
|
| 175 |
+
Epoch 1 | iter 1984 step 62 | loss train: 1.410, val: 1.437 | iter time: 359.31 ms (step) remaining time: 0:34:42
|
| 176 |
+
Epoch 1 | iter 2016 step 63 | loss train: 1.472, val: 1.437 | iter time: 359.21 ms (step) remaining time: 0:34:30
|
| 177 |
+
Epoch 1 | iter 2048 step 64 | loss train: 1.452, val: 1.437 | iter time: 359.36 ms (step) remaining time: 0:34:17
|
| 178 |
+
Epoch 1 | iter 2080 step 65 | loss train: 1.424, val: 1.437 | iter time: 360.00 ms (step) remaining time: 0:34:05
|
| 179 |
+
Epoch 1 | iter 2112 step 66 | loss train: 1.547, val: 1.437 | iter time: 358.38 ms (step) remaining time: 0:33:53
|
| 180 |
+
Epoch 1 | iter 2144 step 67 | loss train: 1.412, val: 1.437 | iter time: 361.05 ms (step) remaining time: 0:33:41
|
| 181 |
+
Epoch 1 | iter 2176 step 68 | loss train: 1.370, val: 1.437 | iter time: 360.14 ms (step) remaining time: 0:33:28
|
| 182 |
+
Epoch 1 | iter 2208 step 69 | loss train: 1.456, val: 1.437 | iter time: 358.71 ms (step) remaining time: 0:33:16
|
| 183 |
+
Epoch 1 | iter 2240 step 70 | loss train: 1.383, val: 1.437 | iter time: 357.78 ms (step) remaining time: 0:33:04
|
| 184 |
+
Epoch 1 | iter 2272 step 71 | loss train: 1.462, val: 1.437 | iter time: 360.52 ms (step) remaining time: 0:32:52
|
| 185 |
+
Epoch 1 | iter 2304 step 72 | loss train: 1.421, val: 1.437 | iter time: 359.15 ms (step) remaining time: 0:32:40
|
| 186 |
+
Epoch 1 | iter 2336 step 73 | loss train: 1.441, val: 1.437 | iter time: 359.55 ms (step) remaining time: 0:32:28
|
| 187 |
+
Epoch 1 | iter 2368 step 74 | loss train: 1.481, val: 1.437 | iter time: 358.50 ms (step) remaining time: 0:32:16
|
| 188 |
+
Epoch 1 | iter 2400 step 75 | loss train: 1.378, val: 1.437 | iter time: 359.39 ms (step) remaining time: 0:32:04
|
| 189 |
+
Epoch 1 | iter 2432 step 76 | loss train: 1.404, val: 1.437 | iter time: 358.19 ms (step) remaining time: 0:31:52
|
| 190 |
+
Epoch 1 | iter 2464 step 77 | loss train: 1.344, val: 1.437 | iter time: 360.52 ms (step) remaining time: 0:31:40
|
| 191 |
+
Epoch 1 | iter 2496 step 78 | loss train: 1.351, val: 1.437 | iter time: 360.53 ms (step) remaining time: 0:31:28
|
| 192 |
+
Epoch 1 | iter 2528 step 79 | loss train: 1.395, val: 1.437 | iter time: 359.22 ms (step) remaining time: 0:31:17
|
| 193 |
+
Epoch 1 | iter 2560 step 80 | loss train: 1.420, val: 1.437 | iter time: 360.32 ms (step) remaining time: 0:31:05
|
| 194 |
+
Epoch 1 | iter 2592 step 81 | loss train: 1.384, val: 1.437 | iter time: 358.82 ms (step) remaining time: 0:30:53
|
| 195 |
+
Epoch 1 | iter 2624 step 82 | loss train: 1.332, val: 1.437 | iter time: 360.52 ms (step) remaining time: 0:30:41
|
| 196 |
+
Epoch 1 | iter 2656 step 83 | loss train: 1.345, val: 1.437 | iter time: 359.73 ms (step) remaining time: 0:30:29
|
| 197 |
+
Epoch 1 | iter 2688 step 84 | loss train: 1.511, val: 1.437 | iter time: 359.19 ms (step) remaining time: 0:30:18
|
| 198 |
+
Epoch 1 | iter 2720 step 85 | loss train: 1.409, val: 1.437 | iter time: 360.14 ms (step) remaining time: 0:30:06
|
| 199 |
+
Epoch 1 | iter 2752 step 86 | loss train: 1.385, val: 1.437 | iter time: 359.63 ms (step) remaining time: 0:29:54
|
| 200 |
+
Epoch 1 | iter 2784 step 87 | loss train: 1.413, val: 1.437 | iter time: 359.79 ms (step) remaining time: 0:29:43
|
| 201 |
+
Epoch 1 | iter 2816 step 88 | loss train: 1.415, val: 1.437 | iter time: 358.56 ms (step) remaining time: 0:29:31
|
| 202 |
+
Epoch 1 | iter 2848 step 89 | loss train: 1.431, val: 1.437 | iter time: 360.75 ms (step) remaining time: 0:29:19
|
| 203 |
+
Epoch 1 | iter 2880 step 90 | loss train: 1.402, val: 1.437 | iter time: 362.31 ms (step) remaining time: 0:29:08
|
| 204 |
+
Epoch 1 | iter 2912 step 91 | loss train: 1.379, val: 1.437 | iter time: 360.99 ms (step) remaining time: 0:28:56
|
| 205 |
+
Epoch 1 | iter 2944 step 92 | loss train: 1.340, val: 1.437 | iter time: 360.45 ms (step) remaining time: 0:28:45
|
| 206 |
+
Epoch 1 | iter 2976 step 93 | loss train: 1.460, val: 1.437 | iter time: 358.90 ms (step) remaining time: 0:28:33
|
| 207 |
+
Epoch 1 | iter 3008 step 94 | loss train: 1.381, val: 1.437 | iter time: 360.97 ms (step) remaining time: 0:28:21
|
| 208 |
+
Epoch 1 | iter 3040 step 95 | loss train: 1.449, val: 1.437 | iter time: 361.44 ms (step) remaining time: 0:28:10
|
| 209 |
+
Epoch 1 | iter 3072 step 96 | loss train: 1.392, val: 1.437 | iter time: 360.08 ms (step) remaining time: 0:27:58
|
| 210 |
+
Epoch 1 | iter 3104 step 97 | loss train: 1.328, val: 1.437 | iter time: 358.03 ms (step) remaining time: 0:27:47
|
| 211 |
+
Epoch 1 | iter 3136 step 98 | loss train: 1.466, val: 1.437 | iter time: 360.39 ms (step) remaining time: 0:27:35
|
| 212 |
+
Epoch 1 | iter 3168 step 99 | loss train: 1.329, val: 1.437 | iter time: 359.32 ms (step) remaining time: 0:27:24
|
| 213 |
+
Epoch 1 | iter 3200 step 100 | loss train: 1.350, val: 1.437 | iter time: 360.09 ms (step) remaining time: 0:27:12
|
| 214 |
+
Validating ...
|
| 215 |
+
iter 3200: val loss 1.3736, val time: 22397.14 ms
|
| 216 |
+
Saving checkpoint to 'out/pretrain/2410_full/step-00000100/lit_model.pth'
|
| 217 |
+
Epoch 1 | iter 3232 step 101 | loss train: 1.296, val: 1.374 | iter time: 356.74 ms (step) remaining time: 0:27:58
|
| 218 |
+
Epoch 1 | iter 3264 step 102 | loss train: 1.429, val: 1.374 | iter time: 358.51 ms (step) remaining time: 0:27:46
|
| 219 |
+
Epoch 1 | iter 3296 step 103 | loss train: 1.447, val: 1.374 | iter time: 360.03 ms (step) remaining time: 0:27:33
|
| 220 |
+
Epoch 1 | iter 3328 step 104 | loss train: 1.404, val: 1.374 | iter time: 361.34 ms (step) remaining time: 0:27:21
|
| 221 |
+
Epoch 1 | iter 3360 step 105 | loss train: 1.390, val: 1.374 | iter time: 358.50 ms (step) remaining time: 0:27:09
|
| 222 |
+
Epoch 1 | iter 3392 step 106 | loss train: 1.379, val: 1.374 | iter time: 360.28 ms (step) remaining time: 0:26:57
|
| 223 |
+
Epoch 1 | iter 3424 step 107 | loss train: 1.382, val: 1.374 | iter time: 361.12 ms (step) remaining time: 0:26:45
|
| 224 |
+
Epoch 1 | iter 3456 step 108 | loss train: 1.447, val: 1.374 | iter time: 361.09 ms (step) remaining time: 0:26:32
|
| 225 |
+
Epoch 1 | iter 3488 step 109 | loss train: 1.408, val: 1.374 | iter time: 360.69 ms (step) remaining time: 0:26:20
|
| 226 |
+
Epoch 1 | iter 3520 step 110 | loss train: 1.467, val: 1.374 | iter time: 362.05 ms (step) remaining time: 0:26:08
|
| 227 |
+
Epoch 1 | iter 3552 step 111 | loss train: 1.317, val: 1.374 | iter time: 358.93 ms (step) remaining time: 0:25:56
|
| 228 |
+
Epoch 1 | iter 3584 step 112 | loss train: 1.327, val: 1.374 | iter time: 358.60 ms (step) remaining time: 0:25:44
|
| 229 |
+
Epoch 1 | iter 3616 step 113 | loss train: 1.371, val: 1.374 | iter time: 358.74 ms (step) remaining time: 0:25:31
|
| 230 |
+
Epoch 1 | iter 3648 step 114 | loss train: 1.401, val: 1.374 | iter time: 357.60 ms (step) remaining time: 0:25:19
|
| 231 |
+
Epoch 1 | iter 3680 step 115 | loss train: 1.435, val: 1.374 | iter time: 360.41 ms (step) remaining time: 0:25:08
|
| 232 |
+
Epoch 1 | iter 3712 step 116 | loss train: 1.363, val: 1.374 | iter time: 360.91 ms (step) remaining time: 0:24:56
|
| 233 |
+
Epoch 1 | iter 3744 step 117 | loss train: 1.313, val: 1.374 | iter time: 360.18 ms (step) remaining time: 0:24:44
|
| 234 |
+
Epoch 1 | iter 3776 step 118 | loss train: 1.358, val: 1.374 | iter time: 358.93 ms (step) remaining time: 0:24:32
|
| 235 |
+
Epoch 1 | iter 3808 step 119 | loss train: 1.335, val: 1.374 | iter time: 358.73 ms (step) remaining time: 0:24:20
|
| 236 |
+
Epoch 1 | iter 3840 step 120 | loss train: 1.353, val: 1.374 | iter time: 358.42 ms (step) remaining time: 0:24:08
|
| 237 |
+
Epoch 1 | iter 3872 step 121 | loss train: 1.339, val: 1.374 | iter time: 358.72 ms (step) remaining time: 0:23:56
|
| 238 |
+
Epoch 1 | iter 3904 step 122 | loss train: 1.269, val: 1.374 | iter time: 358.77 ms (step) remaining time: 0:23:44
|
| 239 |
+
Epoch 1 | iter 3936 step 123 | loss train: 1.350, val: 1.374 | iter time: 360.59 ms (step) remaining time: 0:23:32
|
| 240 |
+
Epoch 1 | iter 3968 step 124 | loss train: 1.408, val: 1.374 | iter time: 357.90 ms (step) remaining time: 0:23:20
|
| 241 |
+
Epoch 1 | iter 4000 step 125 | loss train: 1.412, val: 1.374 | iter time: 360.35 ms (step) remaining time: 0:23:08
|
| 242 |
+
Epoch 1 | iter 4032 step 126 | loss train: 1.362, val: 1.374 | iter time: 360.72 ms (step) remaining time: 0:22:56
|
| 243 |
+
Epoch 1 | iter 4064 step 127 | loss train: 1.340, val: 1.374 | iter time: 359.27 ms (step) remaining time: 0:22:44
|
| 244 |
+
Epoch 1 | iter 4096 step 128 | loss train: 1.354, val: 1.374 | iter time: 360.72 ms (step) remaining time: 0:22:32
|
| 245 |
+
Epoch 1 | iter 4128 step 129 | loss train: 1.406, val: 1.374 | iter time: 359.88 ms (step) remaining time: 0:22:21
|
| 246 |
+
Epoch 1 | iter 4160 step 130 | loss train: 1.392, val: 1.374 | iter time: 360.16 ms (step) remaining time: 0:22:09
|
| 247 |
+
Epoch 1 | iter 4192 step 131 | loss train: 1.393, val: 1.374 | iter time: 359.85 ms (step) remaining time: 0:21:57
|
| 248 |
+
Epoch 1 | iter 4224 step 132 | loss train: 1.390, val: 1.374 | iter time: 360.97 ms (step) remaining time: 0:21:45
|
| 249 |
+
Epoch 1 | iter 4256 step 133 | loss train: 1.364, val: 1.374 | iter time: 359.47 ms (step) remaining time: 0:21:33
|
| 250 |
+
Epoch 1 | iter 4288 step 134 | loss train: 1.314, val: 1.374 | iter time: 360.33 ms (step) remaining time: 0:21:22
|
| 251 |
+
Epoch 1 | iter 4320 step 135 | loss train: 1.338, val: 1.374 | iter time: 359.13 ms (step) remaining time: 0:21:10
|
| 252 |
+
Epoch 1 | iter 4352 step 136 | loss train: 1.309, val: 1.374 | iter time: 360.45 ms (step) remaining time: 0:20:58
|
| 253 |
+
Epoch 1 | iter 4384 step 137 | loss train: 1.281, val: 1.374 | iter time: 361.21 ms (step) remaining time: 0:20:46
|
| 254 |
+
Epoch 1 | iter 4416 step 138 | loss train: 1.397, val: 1.374 | iter time: 360.56 ms (step) remaining time: 0:20:35
|
| 255 |
+
Epoch 1 | iter 4448 step 139 | loss train: 1.357, val: 1.374 | iter time: 360.16 ms (step) remaining time: 0:20:23
|
| 256 |
+
Epoch 1 | iter 4480 step 140 | loss train: 1.417, val: 1.374 | iter time: 359.42 ms (step) remaining time: 0:20:11
|
| 257 |
+
Epoch 1 | iter 4512 step 141 | loss train: 1.448, val: 1.374 | iter time: 361.35 ms (step) remaining time: 0:20:00
|
| 258 |
+
Epoch 1 | iter 4544 step 142 | loss train: 1.455, val: 1.374 | iter time: 359.39 ms (step) remaining time: 0:19:48
|
| 259 |
+
Epoch 1 | iter 4576 step 143 | loss train: 1.363, val: 1.374 | iter time: 360.30 ms (step) remaining time: 0:19:36
|
| 260 |
+
Epoch 1 | iter 4608 step 144 | loss train: 1.340, val: 1.374 | iter time: 361.42 ms (step) remaining time: 0:19:25
|
| 261 |
+
Epoch 1 | iter 4640 step 145 | loss train: 1.379, val: 1.374 | iter time: 360.63 ms (step) remaining time: 0:19:13
|
| 262 |
+
Epoch 1 | iter 4672 step 146 | loss train: 1.417, val: 1.374 | iter time: 358.98 ms (step) remaining time: 0:19:02
|
| 263 |
+
Epoch 1 | iter 4704 step 147 | loss train: 1.391, val: 1.374 | iter time: 360.11 ms (step) remaining time: 0:18:50
|
| 264 |
+
Epoch 1 | iter 4736 step 148 | loss train: 1.326, val: 1.374 | iter time: 359.22 ms (step) remaining time: 0:18:38
|
| 265 |
+
Epoch 1 | iter 4768 step 149 | loss train: 1.335, val: 1.374 | iter time: 360.32 ms (step) remaining time: 0:18:27
|
| 266 |
+
Epoch 1 | iter 4800 step 150 | loss train: 1.320, val: 1.374 | iter time: 359.80 ms (step) remaining time: 0:18:15
|
| 267 |
+
Validating ...
|
| 268 |
+
iter 4800: val loss 1.3219, val time: 22430.10 ms
|
| 269 |
+
Epoch 1 | iter 4832 step 151 | loss train: 1.357, val: 1.322 | iter time: 361.07 ms (step) remaining time: 0:18:18
|
| 270 |
+
Epoch 1 | iter 4864 step 152 | loss train: 1.390, val: 1.322 | iter time: 359.62 ms (step) remaining time: 0:18:06
|
| 271 |
+
Epoch 1 | iter 4896 step 153 | loss train: 1.325, val: 1.322 | iter time: 358.47 ms (step) remaining time: 0:17:54
|
| 272 |
+
Epoch 1 | iter 4928 step 154 | loss train: 1.332, val: 1.322 | iter time: 359.33 ms (step) remaining time: 0:17:42
|
| 273 |
+
Epoch 1 | iter 4960 step 155 | loss train: 1.378, val: 1.322 | iter time: 360.33 ms (step) remaining time: 0:17:31
|
| 274 |
+
Epoch 1 | iter 4992 step 156 | loss train: 1.380, val: 1.322 | iter time: 361.17 ms (step) remaining time: 0:17:19
|
| 275 |
+
Epoch 1 | iter 5024 step 157 | loss train: 1.327, val: 1.322 | iter time: 359.73 ms (step) remaining time: 0:17:07
|
| 276 |
+
Epoch 1 | iter 5056 step 158 | loss train: 1.332, val: 1.322 | iter time: 360.24 ms (step) remaining time: 0:16:55
|
| 277 |
+
Epoch 1 | iter 5088 step 159 | loss train: 1.312, val: 1.322 | iter time: 358.35 ms (step) remaining time: 0:16:44
|
| 278 |
+
Epoch 1 | iter 5120 step 160 | loss train: 1.322, val: 1.322 | iter time: 359.24 ms (step) remaining time: 0:16:32
|
| 279 |
+
Epoch 1 | iter 5152 step 161 | loss train: 1.427, val: 1.322 | iter time: 359.44 ms (step) remaining time: 0:16:20
|
| 280 |
+
Epoch 1 | iter 5184 step 162 | loss train: 1.359, val: 1.322 | iter time: 360.82 ms (step) remaining time: 0:16:09
|
| 281 |
+
Epoch 1 | iter 5216 step 163 | loss train: 1.322, val: 1.322 | iter time: 360.22 ms (step) remaining time: 0:15:57
|
| 282 |
+
Epoch 1 | iter 5248 step 164 | loss train: 1.361, val: 1.322 | iter time: 360.12 ms (step) remaining time: 0:15:45
|
| 283 |
+
Epoch 1 | iter 5280 step 165 | loss train: 1.417, val: 1.322 | iter time: 360.10 ms (step) remaining time: 0:15:34
|
| 284 |
+
Epoch 1 | iter 5312 step 166 | loss train: 1.361, val: 1.322 | iter time: 360.82 ms (step) remaining time: 0:15:22
|
| 285 |
+
Epoch 1 | iter 5344 step 167 | loss train: 1.283, val: 1.322 | iter time: 361.00 ms (step) remaining time: 0:15:10
|
| 286 |
+
Epoch 1 | iter 5376 step 168 | loss train: 1.343, val: 1.322 | iter time: 360.32 ms (step) remaining time: 0:14:59
|
| 287 |
+
Epoch 1 | iter 5408 step 169 | loss train: 1.354, val: 1.322 | iter time: 358.65 ms (step) remaining time: 0:14:47
|
| 288 |
+
Epoch 1 | iter 5440 step 170 | loss train: 1.324, val: 1.322 | iter time: 359.05 ms (step) remaining time: 0:14:36
|
| 289 |
+
Epoch 1 | iter 5472 step 171 | loss train: 1.276, val: 1.322 | iter time: 359.95 ms (step) remaining time: 0:14:24
|
| 290 |
+
Epoch 1 | iter 5504 step 172 | loss train: 1.377, val: 1.322 | iter time: 360.81 ms (step) remaining time: 0:14:12
|
| 291 |
+
Epoch 1 | iter 5536 step 173 | loss train: 1.440, val: 1.322 | iter time: 360.85 ms (step) remaining time: 0:14:01
|
| 292 |
+
Epoch 1 | iter 5568 step 174 | loss train: 1.354, val: 1.322 | iter time: 360.22 ms (step) remaining time: 0:13:49
|
| 293 |
+
Epoch 1 | iter 5600 step 175 | loss train: 1.356, val: 1.322 | iter time: 361.09 ms (step) remaining time: 0:13:38
|
| 294 |
+
Epoch 1 | iter 5632 step 176 | loss train: 1.360, val: 1.322 | iter time: 359.31 ms (step) remaining time: 0:13:26
|
| 295 |
+
Epoch 1 | iter 5664 step 177 | loss train: 1.380, val: 1.322 | iter time: 359.61 ms (step) remaining time: 0:13:15
|
| 296 |
+
Epoch 1 | iter 5696 step 178 | loss train: 1.301, val: 1.322 | iter time: 359.06 ms (step) remaining time: 0:13:03
|
| 297 |
+
Epoch 1 | iter 5728 step 179 | loss train: 1.366, val: 1.322 | iter time: 589.34 ms (step) remaining time: 0:12:52
|
| 298 |
+
Epoch 1 | iter 5760 step 180 | loss train: 1.240, val: 1.322 | iter time: 358.76 ms (step) remaining time: 0:12:40
|
| 299 |
+
Epoch 1 | iter 5792 step 181 | loss train: 1.363, val: 1.322 | iter time: 360.55 ms (step) remaining time: 0:12:29
|
| 300 |
+
Epoch 1 | iter 5824 step 182 | loss train: 1.392, val: 1.322 | iter time: 361.28 ms (step) remaining time: 0:12:17
|
| 301 |
+
Epoch 1 | iter 5856 step 183 | loss train: 1.356, val: 1.322 | iter time: 358.77 ms (step) remaining time: 0:12:06
|
| 302 |
+
Epoch 1 | iter 5888 step 184 | loss train: 1.278, val: 1.322 | iter time: 359.74 ms (step) remaining time: 0:11:54
|
| 303 |
+
Epoch 1 | iter 5920 step 185 | loss train: 1.300, val: 1.322 | iter time: 360.26 ms (step) remaining time: 0:11:43
|
| 304 |
+
Epoch 1 | iter 5952 step 186 | loss train: 1.326, val: 1.322 | iter time: 358.85 ms (step) remaining time: 0:11:31
|
| 305 |
+
Epoch 1 | iter 5984 step 187 | loss train: 1.312, val: 1.322 | iter time: 360.39 ms (step) remaining time: 0:11:20
|
| 306 |
+
Epoch 1 | iter 6016 step 188 | loss train: 1.275, val: 1.322 | iter time: 360.87 ms (step) remaining time: 0:11:08
|
| 307 |
+
Epoch 1 | iter 6048 step 189 | loss train: 1.363, val: 1.322 | iter time: 358.40 ms (step) remaining time: 0:10:57
|
| 308 |
+
Epoch 1 | iter 6080 step 190 | loss train: 1.369, val: 1.322 | iter time: 360.01 ms (step) remaining time: 0:10:45
|
| 309 |
+
Epoch 1 | iter 6112 step 191 | loss train: 1.333, val: 1.322 | iter time: 358.96 ms (step) remaining time: 0:10:34
|
| 310 |
+
Epoch 1 | iter 6144 step 192 | loss train: 1.372, val: 1.322 | iter time: 360.33 ms (step) remaining time: 0:10:22
|
| 311 |
+
Epoch 1 | iter 6176 step 193 | loss train: 1.426, val: 1.322 | iter time: 361.20 ms (step) remaining time: 0:10:11
|
| 312 |
+
Epoch 1 | iter 6208 step 194 | loss train: 1.349, val: 1.322 | iter time: 359.21 ms (step) remaining time: 0:09:59
|
| 313 |
+
Epoch 1 | iter 6240 step 195 | loss train: 1.348, val: 1.322 | iter time: 361.67 ms (step) remaining time: 0:09:48
|
| 314 |
+
Epoch 1 | iter 6272 step 196 | loss train: 1.288, val: 1.322 | iter time: 360.96 ms (step) remaining time: 0:09:36
|
| 315 |
+
Epoch 1 | iter 6304 step 197 | loss train: 1.360, val: 1.322 | iter time: 360.50 ms (step) remaining time: 0:09:25
|
| 316 |
+
Epoch 1 | iter 6336 step 198 | loss train: 1.343, val: 1.322 | iter time: 359.55 ms (step) remaining time: 0:09:14
|
| 317 |
+
Epoch 1 | iter 6368 step 199 | loss train: 1.407, val: 1.322 | iter time: 359.32 ms (step) remaining time: 0:09:02
|
| 318 |
+
Epoch 1 | iter 6400 step 200 | loss train: 1.399, val: 1.322 | iter time: 358.83 ms (step) remaining time: 0:08:51
|
| 319 |
+
Validating ...
|
| 320 |
+
iter 6400: val loss 1.2789, val time: 22400.75 ms
|
| 321 |
+
Saving checkpoint to 'out/pretrain/2410_full/step-00000200/lit_model.pth'
|
| 322 |
+
Epoch 1 | iter 6432 step 201 | loss train: 1.332, val: 1.279 | iter time: 355.09 ms (step) remaining time: 0:08:48
|
| 323 |
+
Epoch 1 | iter 6464 step 202 | loss train: 1.369, val: 1.279 | iter time: 356.26 ms (step) remaining time: 0:08:37
|
| 324 |
+
Epoch 1 | iter 6496 step 203 | loss train: 1.373, val: 1.279 | iter time: 358.11 ms (step) remaining time: 0:08:25
|
| 325 |
+
Epoch 1 | iter 6528 step 204 | loss train: 1.341, val: 1.279 | iter time: 358.52 ms (step) remaining time: 0:08:13
|
| 326 |
+
Epoch 1 | iter 6560 step 205 | loss train: 1.276, val: 1.279 | iter time: 358.79 ms (step) remaining time: 0:08:02
|
| 327 |
+
Epoch 1 | iter 6592 step 206 | loss train: 1.338, val: 1.279 | iter time: 359.64 ms (step) remaining time: 0:07:50
|
| 328 |
+
Epoch 1 | iter 6624 step 207 | loss train: 1.326, val: 1.279 | iter time: 360.01 ms (step) remaining time: 0:07:39
|
| 329 |
+
Epoch 1 | iter 6656 step 208 | loss train: 1.316, val: 1.279 | iter time: 360.84 ms (step) remaining time: 0:07:27
|
| 330 |
+
Epoch 1 | iter 6688 step 209 | loss train: 1.314, val: 1.279 | iter time: 360.15 ms (step) remaining time: 0:07:15
|
| 331 |
+
Epoch 1 | iter 6720 step 210 | loss train: 1.293, val: 1.279 | iter time: 359.60 ms (step) remaining time: 0:07:04
|
| 332 |
+
Epoch 1 | iter 6752 step 211 | loss train: 1.293, val: 1.279 | iter time: 359.32 ms (step) remaining time: 0:06:52
|
| 333 |
+
Epoch 1 | iter 6784 step 212 | loss train: 1.265, val: 1.279 | iter time: 360.92 ms (step) remaining time: 0:06:41
|
| 334 |
+
Epoch 1 | iter 6816 step 213 | loss train: 1.241, val: 1.279 | iter time: 358.27 ms (step) remaining time: 0:06:29
|
| 335 |
+
Epoch 1 | iter 6848 step 214 | loss train: 1.295, val: 1.279 | iter time: 359.99 ms (step) remaining time: 0:06:18
|
| 336 |
+
Epoch 1 | iter 6880 step 215 | loss train: 1.375, val: 1.279 | iter time: 361.18 ms (step) remaining time: 0:06:06
|
| 337 |
+
Epoch 1 | iter 6912 step 216 | loss train: 1.296, val: 1.279 | iter time: 360.69 ms (step) remaining time: 0:05:55
|
| 338 |
+
Epoch 1 | iter 6944 step 217 | loss train: 1.302, val: 1.279 | iter time: 359.04 ms (step) remaining time: 0:05:43
|
| 339 |
+
Epoch 1 | iter 6976 step 218 | loss train: 1.301, val: 1.279 | iter time: 361.16 ms (step) remaining time: 0:05:31
|
| 340 |
+
Epoch 1 | iter 7008 step 219 | loss train: 1.315, val: 1.279 | iter time: 359.14 ms (step) remaining time: 0:05:20
|
| 341 |
+
Epoch 1 | iter 7040 step 220 | loss train: 1.302, val: 1.279 | iter time: 359.59 ms (step) remaining time: 0:05:08
|
| 342 |
+
Epoch 1 | iter 7072 step 221 | loss train: 1.304, val: 1.279 | iter time: 360.41 ms (step) remaining time: 0:04:57
|
| 343 |
+
Epoch 1 | iter 7104 step 222 | loss train: 1.300, val: 1.279 | iter time: 359.67 ms (step) remaining time: 0:04:45
|
| 344 |
+
Epoch 1 | iter 7136 step 223 | loss train: 1.325, val: 1.279 | iter time: 358.14 ms (step) remaining time: 0:04:34
|
| 345 |
+
Epoch 1 | iter 7168 step 224 | loss train: 1.360, val: 1.279 | iter time: 358.44 ms (step) remaining time: 0:04:22
|
| 346 |
+
Epoch 1 | iter 7200 step 225 | loss train: 1.401, val: 1.279 | iter time: 360.85 ms (step) remaining time: 0:04:11
|
| 347 |
+
Epoch 1 | iter 7232 step 226 | loss train: 1.359, val: 1.279 | iter time: 360.32 ms (step) remaining time: 0:03:59
|
| 348 |
+
Epoch 1 | iter 7264 step 227 | loss train: 1.389, val: 1.279 | iter time: 360.97 ms (step) remaining time: 0:03:48
|
| 349 |
+
Epoch 1 | iter 7296 step 228 | loss train: 1.358, val: 1.279 | iter time: 359.79 ms (step) remaining time: 0:03:36
|
| 350 |
+
Epoch 1 | iter 7328 step 229 | loss train: 1.323, val: 1.279 | iter time: 358.85 ms (step) remaining time: 0:03:25
|
| 351 |
+
Epoch 1 | iter 7360 step 230 | loss train: 1.334, val: 1.279 | iter time: 359.17 ms (step) remaining time: 0:03:14
|
| 352 |
+
Epoch 1 | iter 7392 step 231 | loss train: 1.316, val: 1.279 | iter time: 360.60 ms (step) remaining time: 0:03:02
|
| 353 |
+
Epoch 1 | iter 7424 step 232 | loss train: 1.289, val: 1.279 | iter time: 359.59 ms (step) remaining time: 0:02:51
|
| 354 |
+
Epoch 1 | iter 7456 step 233 | loss train: 1.183, val: 1.279 | iter time: 360.47 ms (step) remaining time: 0:02:39
|
| 355 |
+
Epoch 1 | iter 7488 step 234 | loss train: 1.281, val: 1.279 | iter time: 360.38 ms (step) remaining time: 0:02:28
|
| 356 |
+
Epoch 1 | iter 7520 step 235 | loss train: 1.275, val: 1.279 | iter time: 358.83 ms (step) remaining time: 0:02:16
|
| 357 |
+
Epoch 1 | iter 7552 step 236 | loss train: 1.313, val: 1.279 | iter time: 359.05 ms (step) remaining time: 0:02:05
|
| 358 |
+
Epoch 1 | iter 7584 step 237 | loss train: 1.312, val: 1.279 | iter time: 359.89 ms (step) remaining time: 0:01:54
|
| 359 |
+
Epoch 1 | iter 7616 step 238 | loss train: 1.295, val: 1.279 | iter time: 359.48 ms (step) remaining time: 0:01:42
|
| 360 |
+
Epoch 1 | iter 7648 step 239 | loss train: 1.317, val: 1.279 | iter time: 361.31 ms (step) remaining time: 0:01:31
|
| 361 |
+
Epoch 1 | iter 7680 step 240 | loss train: 1.271, val: 1.279 | iter time: 359.98 ms (step) remaining time: 0:01:19
|
| 362 |
+
Epoch 1 | iter 7712 step 241 | loss train: 1.348, val: 1.279 | iter time: 360.80 ms (step) remaining time: 0:01:08
|
| 363 |
+
Epoch 1 | iter 7744 step 242 | loss train: 1.297, val: 1.279 | iter time: 360.05 ms (step) remaining time: 0:00:56
|
| 364 |
+
Epoch 1 | iter 7776 step 243 | loss train: 1.360, val: 1.279 | iter time: 360.56 ms (step) remaining time: 0:00:45
|
| 365 |
+
Epoch 1 | iter 7808 step 244 | loss train: 1.309, val: 1.279 | iter time: 358.93 ms (step) remaining time: 0:00:34
|
| 366 |
+
Epoch 1 | iter 7840 step 245 | loss train: 1.294, val: 1.279 | iter time: 360.23 ms (step) remaining time: 0:00:22
|
| 367 |
+
Epoch 1 | iter 7872 step 246 | loss train: 1.249, val: 1.279 | iter time: 360.23 ms (step) remaining time: 0:00:11
|
| 368 |
+
Epoch 2 | iter 7904 step 247 | loss train: 1.242, val: 1.279 | iter time: 359.87 ms (step) remaining time: 0:00:00
|
| 369 |
+
Validating ...
|
| 370 |
+
Final evaluation | val loss: 1.247 | val ppl: 3.480
|
| 371 |
+
Saving checkpoint to 'out/pretrain/2410_full/final/lit_model.pth'
|
| 372 |
+
----------------------------------------
|
| 373 |
+
| Performance
|
| 374 |
+
| - Total tokens : 258,998,272
|
| 375 |
+
| - Training Time : 2876.07 s
|
| 376 |
+
| - Tok/sec : 139.82 tok/s
|
| 377 |
+
| ----------------------------------------
|
| 378 |
+
| Memory Usage
|
| 379 |
+
| - Memory Used : 26.32 GB
|
| 380 |
+
----------------------------------------
|
out/pretrain/tinyllama/teelogs/2410_lr4e-5.txt
ADDED
|
@@ -0,0 +1,379 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
Initializing distributed: GLOBAL_RANK: 0, MEMBER: 1/4
|
| 2 |
+
Initializing distributed: GLOBAL_RANK: 1, MEMBER: 2/4
|
| 3 |
+
[rank: 1] Seed set to 42
|
| 4 |
+
Initializing distributed: GLOBAL_RANK: 2, MEMBER: 3/4
|
| 5 |
+
[rank: 2] Seed set to 42
|
| 6 |
+
Initializing distributed: GLOBAL_RANK: 3, MEMBER: 4/4
|
| 7 |
+
[rank: 3] Seed set to 42
|
| 8 |
+
----------------------------------------------------------------------------------------------------
|
| 9 |
+
distributed_backend=nccl
|
| 10 |
+
All distributed processes registered. Starting with 4 processes
|
| 11 |
+
----------------------------------------------------------------------------------------------------
|
| 12 |
+
|
| 13 |
+
[rank: 0] Seed set to 42
|
| 14 |
+
/mnt/data/litgpt/litgpt/pretrain.py:515: UserWarning: A newer version of litdata is available (0.2.52). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 15 |
+
train_dataloader = data.train_dataloader()
|
| 16 |
+
/mnt/data/litgpt/litgpt/pretrain.py:515: UserWarning: A newer version of litdata is available (0.2.52). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 17 |
+
train_dataloader = data.train_dataloader()
|
| 18 |
+
/mnt/data/litgpt/litgpt/pretrain.py:515: UserWarning: A newer version of litdata is available (0.2.52). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 19 |
+
train_dataloader = data.train_dataloader()
|
| 20 |
+
/mnt/data/litgpt/litgpt/pretrain.py:515: UserWarning: A newer version of litdata is available (0.2.52). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 21 |
+
train_dataloader = data.train_dataloader()
|
| 22 |
+
All GPUs are fully connected via NVLink.
|
| 23 |
+
{'data': {'batch_size': 1,
|
| 24 |
+
'data_path': PosixPath('litgpt/data/arxiv'),
|
| 25 |
+
'num_workers': 0,
|
| 26 |
+
'seed': 42,
|
| 27 |
+
'seq_length': 2048,
|
| 28 |
+
'use_starcoder': True},
|
| 29 |
+
'data_dir': PosixPath('litgpt/data/arxiv/2410'),
|
| 30 |
+
'devices': 'auto',
|
| 31 |
+
'eval': {'evaluate_example': 'first',
|
| 32 |
+
'final_validation': True,
|
| 33 |
+
'initial_validation': True,
|
| 34 |
+
'interval': 50,
|
| 35 |
+
'max_iters': 200,
|
| 36 |
+
'max_new_tokens': None},
|
| 37 |
+
'initial_checkpoint_dir': PosixPath('out/pretrain/tinyllama/2409_full/final'),
|
| 38 |
+
'log': {'group': None, 'project': None, 'run': None},
|
| 39 |
+
'logger_name': 'tensorboard',
|
| 40 |
+
'model_config': {'attention_logit_softcapping': None,
|
| 41 |
+
'attention_scores_scalar': None,
|
| 42 |
+
'attn_bias': False,
|
| 43 |
+
'bias': False,
|
| 44 |
+
'block_size': 2048,
|
| 45 |
+
'final_logit_softcapping': None,
|
| 46 |
+
'gelu_approximate': 'none',
|
| 47 |
+
'head_size': 64,
|
| 48 |
+
'hf_config': {'name': 'TinyLlama-1.1B-intermediate-step-1431k-3T',
|
| 49 |
+
'org': 'TinyLlama'},
|
| 50 |
+
'intermediate_size': 5632,
|
| 51 |
+
'lm_head_bias': False,
|
| 52 |
+
'mlp_class_name': 'LLaMAMLP',
|
| 53 |
+
'moe_intermediate_size': None,
|
| 54 |
+
'n_embd': 2048,
|
| 55 |
+
'n_expert': 0,
|
| 56 |
+
'n_expert_per_token': 0,
|
| 57 |
+
'n_head': 32,
|
| 58 |
+
'n_layer': 22,
|
| 59 |
+
'n_query_groups': 4,
|
| 60 |
+
'name': 'tiny-llama-1.1b',
|
| 61 |
+
'norm_1': True,
|
| 62 |
+
'norm_2': True,
|
| 63 |
+
'norm_class_name': 'RMSNorm',
|
| 64 |
+
'norm_eps': 1e-05,
|
| 65 |
+
'norm_qk': False,
|
| 66 |
+
'norm_qk_type': 'default',
|
| 67 |
+
'padded_vocab_size': 32000,
|
| 68 |
+
'padding_multiple': 64,
|
| 69 |
+
'parallel_residual': False,
|
| 70 |
+
'post_attention_norm': False,
|
| 71 |
+
'post_mlp_norm': False,
|
| 72 |
+
'rope_adjustments': None,
|
| 73 |
+
'rope_base': 10000,
|
| 74 |
+
'rope_condense_ratio': 1,
|
| 75 |
+
'rope_indices': None,
|
| 76 |
+
'rope_local_base_freq': None,
|
| 77 |
+
'rotary_percentage': 1.0,
|
| 78 |
+
'scale_embeddings': False,
|
| 79 |
+
'shared_attention_norm': False,
|
| 80 |
+
'sliding_window_indices': None,
|
| 81 |
+
'sliding_window_size': None,
|
| 82 |
+
'vocab_size': 32000},
|
| 83 |
+
'model_name': 'tiny-llama-1.1b',
|
| 84 |
+
'num_nodes': 1,
|
| 85 |
+
'optimizer': "{'class_path': 'torch.optim.AdamW', 'init_args': {'lr': 4e-05, "
|
| 86 |
+
"'weight_decay': 0.1, 'betas': [0.9, 0.95]}}",
|
| 87 |
+
'out_dir': PosixPath('out/pretrain/tinyllama/2410_lr4e-5'),
|
| 88 |
+
'ppl': False,
|
| 89 |
+
'precision': 'bf16-mixed',
|
| 90 |
+
'resume': False,
|
| 91 |
+
'seed': 42,
|
| 92 |
+
'tokenizer_dir': PosixPath('checkpoints/TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T'),
|
| 93 |
+
'train': {'epochs': None,
|
| 94 |
+
'global_batch_size': 512,
|
| 95 |
+
'log_interval': 1,
|
| 96 |
+
'lr_warmup_fraction': None,
|
| 97 |
+
'lr_warmup_steps': 20,
|
| 98 |
+
'max_norm': 1.0,
|
| 99 |
+
'max_seq_length': 2048,
|
| 100 |
+
'max_steps': None,
|
| 101 |
+
'max_tokens': 258998272,
|
| 102 |
+
'micro_batch_size': 4,
|
| 103 |
+
'min_lr': 4e-05,
|
| 104 |
+
'save_interval': 100,
|
| 105 |
+
'tie_embeddings': None}}
|
| 106 |
+
Time to instantiate model: 0.02 seconds.
|
| 107 |
+
Total parameters: 1,100,048,384
|
| 108 |
+
[ok] out/pretrain/tinyllama/2409_full/final/lit_model.pth 已是纯权重
|
| 109 |
+
Validating ...
|
| 110 |
+
Measured TFLOPs: 239.66
|
| 111 |
+
Epoch 1 | iter 32 step 1 | loss train: 1.579, val: 1.379 | iter time: 536.97 ms (step) remaining time: 0:47:57
|
| 112 |
+
Epoch 1 | iter 64 step 2 | loss train: 1.535, val: 1.379 | iter time: 357.22 ms (step) remaining time: 0:45:54
|
| 113 |
+
Epoch 1 | iter 96 step 3 | loss train: 1.516, val: 1.379 | iter time: 359.31 ms (step) remaining time: 0:45:07
|
| 114 |
+
Epoch 1 | iter 128 step 4 | loss train: 1.584, val: 1.379 | iter time: 358.93 ms (step) remaining time: 0:44:39
|
| 115 |
+
Epoch 1 | iter 160 step 5 | loss train: 1.584, val: 1.379 | iter time: 358.89 ms (step) remaining time: 0:44:19
|
| 116 |
+
Epoch 1 | iter 192 step 6 | loss train: 1.559, val: 1.379 | iter time: 358.77 ms (step) remaining time: 0:44:02
|
| 117 |
+
Epoch 1 | iter 224 step 7 | loss train: 1.581, val: 1.379 | iter time: 359.92 ms (step) remaining time: 0:43:48
|
| 118 |
+
Epoch 1 | iter 256 step 8 | loss train: 1.523, val: 1.379 | iter time: 358.95 ms (step) remaining time: 0:43:40
|
| 119 |
+
Epoch 1 | iter 288 step 9 | loss train: 1.454, val: 1.379 | iter time: 359.82 ms (step) remaining time: 0:43:26
|
| 120 |
+
Epoch 1 | iter 320 step 10 | loss train: 1.428, val: 1.379 | iter time: 359.87 ms (step) remaining time: 0:43:13
|
| 121 |
+
Epoch 1 | iter 352 step 11 | loss train: 1.415, val: 1.379 | iter time: 359.11 ms (step) remaining time: 0:43:00
|
| 122 |
+
Epoch 1 | iter 384 step 12 | loss train: 1.433, val: 1.379 | iter time: 358.79 ms (step) remaining time: 0:42:48
|
| 123 |
+
Epoch 1 | iter 416 step 13 | loss train: 1.412, val: 1.379 | iter time: 359.60 ms (step) remaining time: 0:42:36
|
| 124 |
+
Epoch 1 | iter 448 step 14 | loss train: 1.447, val: 1.379 | iter time: 359.45 ms (step) remaining time: 0:42:24
|
| 125 |
+
Epoch 1 | iter 480 step 15 | loss train: 1.393, val: 1.379 | iter time: 360.32 ms (step) remaining time: 0:42:12
|
| 126 |
+
Epoch 1 | iter 512 step 16 | loss train: 1.469, val: 1.379 | iter time: 359.34 ms (step) remaining time: 0:42:01
|
| 127 |
+
Epoch 1 | iter 544 step 17 | loss train: 1.330, val: 1.379 | iter time: 359.33 ms (step) remaining time: 0:41:49
|
| 128 |
+
Epoch 1 | iter 576 step 18 | loss train: 1.370, val: 1.379 | iter time: 359.64 ms (step) remaining time: 0:41:38
|
| 129 |
+
Epoch 1 | iter 608 step 19 | loss train: 1.357, val: 1.379 | iter time: 360.17 ms (step) remaining time: 0:41:26
|
| 130 |
+
Epoch 1 | iter 640 step 20 | loss train: 1.517, val: 1.379 | iter time: 360.47 ms (step) remaining time: 0:41:15
|
| 131 |
+
Epoch 1 | iter 672 step 21 | loss train: 1.409, val: 1.379 | iter time: 361.47 ms (step) remaining time: 0:41:03
|
| 132 |
+
Epoch 1 | iter 704 step 22 | loss train: 1.443, val: 1.379 | iter time: 359.09 ms (step) remaining time: 0:40:52
|
| 133 |
+
Epoch 1 | iter 736 step 23 | loss train: 1.436, val: 1.379 | iter time: 360.50 ms (step) remaining time: 0:40:41
|
| 134 |
+
Epoch 1 | iter 768 step 24 | loss train: 1.375, val: 1.379 | iter time: 359.23 ms (step) remaining time: 0:40:30
|
| 135 |
+
Epoch 1 | iter 800 step 25 | loss train: 1.423, val: 1.379 | iter time: 358.73 ms (step) remaining time: 0:40:19
|
| 136 |
+
Epoch 1 | iter 832 step 26 | loss train: 1.417, val: 1.379 | iter time: 360.01 ms (step) remaining time: 0:40:08
|
| 137 |
+
Epoch 1 | iter 864 step 27 | loss train: 1.422, val: 1.379 | iter time: 360.65 ms (step) remaining time: 0:39:57
|
| 138 |
+
Epoch 1 | iter 896 step 28 | loss train: 1.372, val: 1.379 | iter time: 359.86 ms (step) remaining time: 0:39:46
|
| 139 |
+
Epoch 1 | iter 928 step 29 | loss train: 1.409, val: 1.379 | iter time: 360.36 ms (step) remaining time: 0:39:34
|
| 140 |
+
Epoch 1 | iter 960 step 30 | loss train: 1.403, val: 1.379 | iter time: 357.53 ms (step) remaining time: 0:39:23
|
| 141 |
+
Epoch 1 | iter 992 step 31 | loss train: 1.405, val: 1.379 | iter time: 359.55 ms (step) remaining time: 0:39:12
|
| 142 |
+
Epoch 1 | iter 1024 step 32 | loss train: 1.332, val: 1.379 | iter time: 360.30 ms (step) remaining time: 0:39:01
|
| 143 |
+
Epoch 1 | iter 1056 step 33 | loss train: 1.379, val: 1.379 | iter time: 360.59 ms (step) remaining time: 0:38:50
|
| 144 |
+
Epoch 1 | iter 1088 step 34 | loss train: 1.369, val: 1.379 | iter time: 360.79 ms (step) remaining time: 0:38:39
|
| 145 |
+
Epoch 1 | iter 1120 step 35 | loss train: 1.325, val: 1.379 | iter time: 359.95 ms (step) remaining time: 0:38:28
|
| 146 |
+
Epoch 1 | iter 1152 step 36 | loss train: 1.408, val: 1.379 | iter time: 359.67 ms (step) remaining time: 0:38:17
|
| 147 |
+
Epoch 1 | iter 1184 step 37 | loss train: 1.338, val: 1.379 | iter time: 359.97 ms (step) remaining time: 0:38:06
|
| 148 |
+
Epoch 1 | iter 1216 step 38 | loss train: 1.405, val: 1.379 | iter time: 358.23 ms (step) remaining time: 0:37:55
|
| 149 |
+
Epoch 1 | iter 1248 step 39 | loss train: 1.398, val: 1.379 | iter time: 360.06 ms (step) remaining time: 0:37:44
|
| 150 |
+
Epoch 1 | iter 1280 step 40 | loss train: 1.390, val: 1.379 | iter time: 360.98 ms (step) remaining time: 0:37:33
|
| 151 |
+
Epoch 1 | iter 1312 step 41 | loss train: 1.357, val: 1.379 | iter time: 358.70 ms (step) remaining time: 0:37:22
|
| 152 |
+
Epoch 1 | iter 1344 step 42 | loss train: 1.413, val: 1.379 | iter time: 359.47 ms (step) remaining time: 0:37:11
|
| 153 |
+
Epoch 1 | iter 1376 step 43 | loss train: 1.390, val: 1.379 | iter time: 359.25 ms (step) remaining time: 0:37:00
|
| 154 |
+
Epoch 1 | iter 1408 step 44 | loss train: 1.381, val: 1.379 | iter time: 358.82 ms (step) remaining time: 0:36:49
|
| 155 |
+
Epoch 1 | iter 1440 step 45 | loss train: 1.391, val: 1.379 | iter time: 359.25 ms (step) remaining time: 0:36:38
|
| 156 |
+
Epoch 1 | iter 1472 step 46 | loss train: 1.452, val: 1.379 | iter time: 359.81 ms (step) remaining time: 0:36:29
|
| 157 |
+
Epoch 1 | iter 1504 step 47 | loss train: 1.389, val: 1.379 | iter time: 360.08 ms (step) remaining time: 0:36:18
|
| 158 |
+
Epoch 1 | iter 1536 step 48 | loss train: 1.364, val: 1.379 | iter time: 359.99 ms (step) remaining time: 0:36:07
|
| 159 |
+
Epoch 1 | iter 1568 step 49 | loss train: 1.368, val: 1.379 | iter time: 364.93 ms (step) remaining time: 0:35:56
|
| 160 |
+
Epoch 1 | iter 1600 step 50 | loss train: 1.431, val: 1.379 | iter time: 359.03 ms (step) remaining time: 0:35:45
|
| 161 |
+
Validating ...
|
| 162 |
+
iter 1600: val loss 1.3427, val time: 21931.36 ms
|
| 163 |
+
Epoch 1 | iter 1632 step 51 | loss train: 1.318, val: 1.343 | iter time: 359.29 ms (step) remaining time: 0:36:58
|
| 164 |
+
Epoch 1 | iter 1664 step 52 | loss train: 1.498, val: 1.343 | iter time: 360.88 ms (step) remaining time: 0:36:45
|
| 165 |
+
Epoch 1 | iter 1696 step 53 | loss train: 1.443, val: 1.343 | iter time: 360.17 ms (step) remaining time: 0:36:32
|
| 166 |
+
Epoch 1 | iter 1728 step 54 | loss train: 1.368, val: 1.343 | iter time: 358.24 ms (step) remaining time: 0:36:19
|
| 167 |
+
Epoch 1 | iter 1760 step 55 | loss train: 1.428, val: 1.343 | iter time: 720.31 ms (step) remaining time: 0:36:08
|
| 168 |
+
Epoch 1 | iter 1792 step 56 | loss train: 1.382, val: 1.343 | iter time: 358.18 ms (step) remaining time: 0:35:55
|
| 169 |
+
Epoch 1 | iter 1824 step 57 | loss train: 1.369, val: 1.343 | iter time: 357.93 ms (step) remaining time: 0:35:42
|
| 170 |
+
Epoch 1 | iter 1856 step 58 | loss train: 1.337, val: 1.343 | iter time: 360.48 ms (step) remaining time: 0:35:30
|
| 171 |
+
Epoch 1 | iter 1888 step 59 | loss train: 1.411, val: 1.343 | iter time: 358.99 ms (step) remaining time: 0:35:17
|
| 172 |
+
Epoch 1 | iter 1920 step 60 | loss train: 1.419, val: 1.343 | iter time: 361.85 ms (step) remaining time: 0:35:05
|
| 173 |
+
Epoch 1 | iter 1952 step 61 | loss train: 1.348, val: 1.343 | iter time: 359.02 ms (step) remaining time: 0:34:52
|
| 174 |
+
Epoch 1 | iter 1984 step 62 | loss train: 1.420, val: 1.343 | iter time: 359.25 ms (step) remaining time: 0:34:40
|
| 175 |
+
Epoch 1 | iter 2016 step 63 | loss train: 1.403, val: 1.343 | iter time: 359.86 ms (step) remaining time: 0:34:27
|
| 176 |
+
Epoch 1 | iter 2048 step 64 | loss train: 1.392, val: 1.343 | iter time: 360.11 ms (step) remaining time: 0:34:15
|
| 177 |
+
Epoch 1 | iter 2080 step 65 | loss train: 1.390, val: 1.343 | iter time: 360.07 ms (step) remaining time: 0:34:03
|
| 178 |
+
Epoch 1 | iter 2112 step 66 | loss train: 1.367, val: 1.343 | iter time: 360.96 ms (step) remaining time: 0:33:51
|
| 179 |
+
Epoch 1 | iter 2144 step 67 | loss train: 1.403, val: 1.343 | iter time: 360.24 ms (step) remaining time: 0:33:38
|
| 180 |
+
Epoch 1 | iter 2176 step 68 | loss train: 1.337, val: 1.343 | iter time: 358.68 ms (step) remaining time: 0:33:26
|
| 181 |
+
Epoch 1 | iter 2208 step 69 | loss train: 1.410, val: 1.343 | iter time: 361.21 ms (step) remaining time: 0:33:14
|
| 182 |
+
Epoch 1 | iter 2240 step 70 | loss train: 1.352, val: 1.343 | iter time: 359.72 ms (step) remaining time: 0:33:02
|
| 183 |
+
Epoch 1 | iter 2272 step 71 | loss train: 1.358, val: 1.343 | iter time: 359.27 ms (step) remaining time: 0:32:50
|
| 184 |
+
Epoch 1 | iter 2304 step 72 | loss train: 1.328, val: 1.343 | iter time: 360.73 ms (step) remaining time: 0:32:38
|
| 185 |
+
Epoch 1 | iter 2336 step 73 | loss train: 1.409, val: 1.343 | iter time: 358.84 ms (step) remaining time: 0:32:26
|
| 186 |
+
Epoch 1 | iter 2368 step 74 | loss train: 1.349, val: 1.343 | iter time: 360.37 ms (step) remaining time: 0:32:14
|
| 187 |
+
Epoch 1 | iter 2400 step 75 | loss train: 1.409, val: 1.343 | iter time: 360.47 ms (step) remaining time: 0:32:02
|
| 188 |
+
Epoch 1 | iter 2432 step 76 | loss train: 1.433, val: 1.343 | iter time: 359.26 ms (step) remaining time: 0:31:50
|
| 189 |
+
Epoch 1 | iter 2464 step 77 | loss train: 1.365, val: 1.343 | iter time: 360.33 ms (step) remaining time: 0:31:39
|
| 190 |
+
Epoch 1 | iter 2496 step 78 | loss train: 1.414, val: 1.343 | iter time: 358.72 ms (step) remaining time: 0:31:27
|
| 191 |
+
Epoch 1 | iter 2528 step 79 | loss train: 1.334, val: 1.343 | iter time: 360.70 ms (step) remaining time: 0:31:15
|
| 192 |
+
Epoch 1 | iter 2560 step 80 | loss train: 1.348, val: 1.343 | iter time: 360.65 ms (step) remaining time: 0:31:03
|
| 193 |
+
Epoch 1 | iter 2592 step 81 | loss train: 1.435, val: 1.343 | iter time: 360.49 ms (step) remaining time: 0:30:51
|
| 194 |
+
Epoch 1 | iter 2624 step 82 | loss train: 1.382, val: 1.343 | iter time: 360.39 ms (step) remaining time: 0:30:40
|
| 195 |
+
Epoch 1 | iter 2656 step 83 | loss train: 1.479, val: 1.343 | iter time: 358.98 ms (step) remaining time: 0:30:28
|
| 196 |
+
Epoch 1 | iter 2688 step 84 | loss train: 1.392, val: 1.343 | iter time: 361.67 ms (step) remaining time: 0:30:16
|
| 197 |
+
Epoch 1 | iter 2720 step 85 | loss train: 1.437, val: 1.343 | iter time: 359.09 ms (step) remaining time: 0:30:05
|
| 198 |
+
Epoch 1 | iter 2752 step 86 | loss train: 1.411, val: 1.343 | iter time: 359.38 ms (step) remaining time: 0:29:53
|
| 199 |
+
Epoch 1 | iter 2784 step 87 | loss train: 1.325, val: 1.343 | iter time: 360.46 ms (step) remaining time: 0:29:41
|
| 200 |
+
Epoch 1 | iter 2816 step 88 | loss train: 1.329, val: 1.343 | iter time: 359.10 ms (step) remaining time: 0:29:30
|
| 201 |
+
Epoch 1 | iter 2848 step 89 | loss train: 1.355, val: 1.343 | iter time: 358.21 ms (step) remaining time: 0:29:18
|
| 202 |
+
Epoch 1 | iter 2880 step 90 | loss train: 1.348, val: 1.343 | iter time: 360.35 ms (step) remaining time: 0:29:07
|
| 203 |
+
Epoch 1 | iter 2912 step 91 | loss train: 1.291, val: 1.343 | iter time: 359.44 ms (step) remaining time: 0:28:55
|
| 204 |
+
Epoch 1 | iter 2944 step 92 | loss train: 1.345, val: 1.343 | iter time: 359.07 ms (step) remaining time: 0:28:43
|
| 205 |
+
Epoch 1 | iter 2976 step 93 | loss train: 1.396, val: 1.343 | iter time: 359.81 ms (step) remaining time: 0:28:32
|
| 206 |
+
Epoch 1 | iter 3008 step 94 | loss train: 1.364, val: 1.343 | iter time: 361.59 ms (step) remaining time: 0:28:20
|
| 207 |
+
Epoch 1 | iter 3040 step 95 | loss train: 1.356, val: 1.343 | iter time: 360.84 ms (step) remaining time: 0:28:09
|
| 208 |
+
Epoch 1 | iter 3072 step 96 | loss train: 1.395, val: 1.343 | iter time: 360.73 ms (step) remaining time: 0:27:57
|
| 209 |
+
Epoch 1 | iter 3104 step 97 | loss train: 1.398, val: 1.343 | iter time: 359.35 ms (step) remaining time: 0:27:46
|
| 210 |
+
Epoch 1 | iter 3136 step 98 | loss train: 1.362, val: 1.343 | iter time: 361.78 ms (step) remaining time: 0:27:34
|
| 211 |
+
Epoch 1 | iter 3168 step 99 | loss train: 1.406, val: 1.343 | iter time: 359.70 ms (step) remaining time: 0:27:23
|
| 212 |
+
Epoch 1 | iter 3200 step 100 | loss train: 1.339, val: 1.343 | iter time: 360.47 ms (step) remaining time: 0:27:11
|
| 213 |
+
Validating ...
|
| 214 |
+
iter 3200: val loss 1.3158, val time: 21942.52 ms
|
| 215 |
+
Saving checkpoint to 'out/pretrain/tinyllama/2410_lr4e-5/step-00000100/lit_model.pth'
|
| 216 |
+
Epoch 1 | iter 3232 step 101 | loss train: 1.381, val: 1.316 | iter time: 358.25 ms (step) remaining time: 0:27:55
|
| 217 |
+
Epoch 1 | iter 3264 step 102 | loss train: 1.368, val: 1.316 | iter time: 359.11 ms (step) remaining time: 0:27:43
|
| 218 |
+
Epoch 1 | iter 3296 step 103 | loss train: 1.344, val: 1.316 | iter time: 357.26 ms (step) remaining time: 0:27:30
|
| 219 |
+
Epoch 1 | iter 3328 step 104 | loss train: 1.366, val: 1.316 | iter time: 360.14 ms (step) remaining time: 0:27:18
|
| 220 |
+
Epoch 1 | iter 3360 step 105 | loss train: 1.404, val: 1.316 | iter time: 358.91 ms (step) remaining time: 0:27:06
|
| 221 |
+
Epoch 1 | iter 3392 step 106 | loss train: 1.349, val: 1.316 | iter time: 358.59 ms (step) remaining time: 0:26:53
|
| 222 |
+
Epoch 1 | iter 3424 step 107 | loss train: 1.283, val: 1.316 | iter time: 360.36 ms (step) remaining time: 0:26:41
|
| 223 |
+
Epoch 1 | iter 3456 step 108 | loss train: 1.374, val: 1.316 | iter time: 358.78 ms (step) remaining time: 0:26:29
|
| 224 |
+
Epoch 1 | iter 3488 step 109 | loss train: 1.427, val: 1.316 | iter time: 359.47 ms (step) remaining time: 0:26:17
|
| 225 |
+
Epoch 1 | iter 3520 step 110 | loss train: 1.350, val: 1.316 | iter time: 360.06 ms (step) remaining time: 0:26:05
|
| 226 |
+
Epoch 1 | iter 3552 step 111 | loss train: 1.363, val: 1.316 | iter time: 359.22 ms (step) remaining time: 0:25:53
|
| 227 |
+
Epoch 1 | iter 3584 step 112 | loss train: 1.376, val: 1.316 | iter time: 359.21 ms (step) remaining time: 0:25:40
|
| 228 |
+
Epoch 1 | iter 3616 step 113 | loss train: 1.390, val: 1.316 | iter time: 840.22 ms (step) remaining time: 0:25:29
|
| 229 |
+
Epoch 1 | iter 3648 step 114 | loss train: 1.346, val: 1.316 | iter time: 360.40 ms (step) remaining time: 0:25:17
|
| 230 |
+
Epoch 1 | iter 3680 step 115 | loss train: 1.420, val: 1.316 | iter time: 359.43 ms (step) remaining time: 0:25:05
|
| 231 |
+
Epoch 1 | iter 3712 step 116 | loss train: 1.386, val: 1.316 | iter time: 359.26 ms (step) remaining time: 0:24:53
|
| 232 |
+
Epoch 1 | iter 3744 step 117 | loss train: 1.410, val: 1.316 | iter time: 359.65 ms (step) remaining time: 0:24:41
|
| 233 |
+
Epoch 1 | iter 3776 step 118 | loss train: 1.330, val: 1.316 | iter time: 360.43 ms (step) remaining time: 0:24:29
|
| 234 |
+
Epoch 1 | iter 3808 step 119 | loss train: 1.406, val: 1.316 | iter time: 357.91 ms (step) remaining time: 0:24:17
|
| 235 |
+
Epoch 1 | iter 3840 step 120 | loss train: 1.424, val: 1.316 | iter time: 358.77 ms (step) remaining time: 0:24:05
|
| 236 |
+
Epoch 1 | iter 3872 step 121 | loss train: 1.317, val: 1.316 | iter time: 359.10 ms (step) remaining time: 0:23:53
|
| 237 |
+
Epoch 1 | iter 3904 step 122 | loss train: 1.387, val: 1.316 | iter time: 359.86 ms (step) remaining time: 0:23:41
|
| 238 |
+
Epoch 1 | iter 3936 step 123 | loss train: 1.346, val: 1.316 | iter time: 360.78 ms (step) remaining time: 0:23:29
|
| 239 |
+
Epoch 1 | iter 3968 step 124 | loss train: 1.341, val: 1.316 | iter time: 358.04 ms (step) remaining time: 0:23:17
|
| 240 |
+
Epoch 1 | iter 4000 step 125 | loss train: 1.385, val: 1.316 | iter time: 358.34 ms (step) remaining time: 0:23:06
|
| 241 |
+
Epoch 1 | iter 4032 step 126 | loss train: 1.359, val: 1.316 | iter time: 358.82 ms (step) remaining time: 0:22:54
|
| 242 |
+
Epoch 1 | iter 4064 step 127 | loss train: 1.450, val: 1.316 | iter time: 358.75 ms (step) remaining time: 0:22:43
|
| 243 |
+
Epoch 1 | iter 4096 step 128 | loss train: 1.368, val: 1.316 | iter time: 359.37 ms (step) remaining time: 0:22:31
|
| 244 |
+
Epoch 1 | iter 4128 step 129 | loss train: 1.357, val: 1.316 | iter time: 359.95 ms (step) remaining time: 0:22:19
|
| 245 |
+
Epoch 1 | iter 4160 step 130 | loss train: 1.385, val: 1.316 | iter time: 357.61 ms (step) remaining time: 0:22:07
|
| 246 |
+
Epoch 1 | iter 4192 step 131 | loss train: 1.397, val: 1.316 | iter time: 359.87 ms (step) remaining time: 0:21:56
|
| 247 |
+
Epoch 1 | iter 4224 step 132 | loss train: 1.335, val: 1.316 | iter time: 357.53 ms (step) remaining time: 0:21:44
|
| 248 |
+
Epoch 1 | iter 4256 step 133 | loss train: 1.376, val: 1.316 | iter time: 358.33 ms (step) remaining time: 0:21:32
|
| 249 |
+
Epoch 1 | iter 4288 step 134 | loss train: 1.363, val: 1.316 | iter time: 359.74 ms (step) remaining time: 0:21:20
|
| 250 |
+
Epoch 1 | iter 4320 step 135 | loss train: 1.286, val: 1.316 | iter time: 358.13 ms (step) remaining time: 0:21:09
|
| 251 |
+
Epoch 1 | iter 4352 step 136 | loss train: 1.322, val: 1.316 | iter time: 360.76 ms (step) remaining time: 0:20:57
|
| 252 |
+
Epoch 1 | iter 4384 step 137 | loss train: 1.327, val: 1.316 | iter time: 359.59 ms (step) remaining time: 0:20:45
|
| 253 |
+
Epoch 1 | iter 4416 step 138 | loss train: 1.345, val: 1.316 | iter time: 359.11 ms (step) remaining time: 0:20:34
|
| 254 |
+
Epoch 1 | iter 4448 step 139 | loss train: 1.369, val: 1.316 | iter time: 358.76 ms (step) remaining time: 0:20:22
|
| 255 |
+
Epoch 1 | iter 4480 step 140 | loss train: 1.340, val: 1.316 | iter time: 360.79 ms (step) remaining time: 0:20:10
|
| 256 |
+
Epoch 1 | iter 4512 step 141 | loss train: 1.381, val: 1.316 | iter time: 359.95 ms (step) remaining time: 0:19:59
|
| 257 |
+
Epoch 1 | iter 4544 step 142 | loss train: 1.411, val: 1.316 | iter time: 359.07 ms (step) remaining time: 0:19:47
|
| 258 |
+
Epoch 1 | iter 4576 step 143 | loss train: 1.394, val: 1.316 | iter time: 359.39 ms (step) remaining time: 0:19:35
|
| 259 |
+
Epoch 1 | iter 4608 step 144 | loss train: 1.304, val: 1.316 | iter time: 359.09 ms (step) remaining time: 0:19:24
|
| 260 |
+
Epoch 1 | iter 4640 step 145 | loss train: 1.331, val: 1.316 | iter time: 358.95 ms (step) remaining time: 0:19:12
|
| 261 |
+
Epoch 1 | iter 4672 step 146 | loss train: 1.285, val: 1.316 | iter time: 359.53 ms (step) remaining time: 0:19:00
|
| 262 |
+
Epoch 1 | iter 4704 step 147 | loss train: 1.324, val: 1.316 | iter time: 358.91 ms (step) remaining time: 0:18:49
|
| 263 |
+
Epoch 1 | iter 4736 step 148 | loss train: 1.313, val: 1.316 | iter time: 359.92 ms (step) remaining time: 0:18:37
|
| 264 |
+
Epoch 1 | iter 4768 step 149 | loss train: 1.374, val: 1.316 | iter time: 358.69 ms (step) remaining time: 0:18:26
|
| 265 |
+
Epoch 1 | iter 4800 step 150 | loss train: 1.386, val: 1.316 | iter time: 358.81 ms (step) remaining time: 0:18:14
|
| 266 |
+
Validating ...
|
| 267 |
+
iter 4800: val loss 1.3066, val time: 21939.04 ms
|
| 268 |
+
Epoch 1 | iter 4832 step 151 | loss train: 1.398, val: 1.307 | iter time: 357.83 ms (step) remaining time: 0:18:16
|
| 269 |
+
Epoch 1 | iter 4864 step 152 | loss train: 1.343, val: 1.307 | iter time: 358.51 ms (step) remaining time: 0:18:05
|
| 270 |
+
Epoch 1 | iter 4896 step 153 | loss train: 1.341, val: 1.307 | iter time: 359.38 ms (step) remaining time: 0:17:53
|
| 271 |
+
Epoch 1 | iter 4928 step 154 | loss train: 1.384, val: 1.307 | iter time: 361.23 ms (step) remaining time: 0:17:41
|
| 272 |
+
Epoch 1 | iter 4960 step 155 | loss train: 1.322, val: 1.307 | iter time: 360.36 ms (step) remaining time: 0:17:29
|
| 273 |
+
Epoch 1 | iter 4992 step 156 | loss train: 1.345, val: 1.307 | iter time: 359.18 ms (step) remaining time: 0:17:18
|
| 274 |
+
Epoch 1 | iter 5024 step 157 | loss train: 1.326, val: 1.307 | iter time: 358.44 ms (step) remaining time: 0:17:06
|
| 275 |
+
Epoch 1 | iter 5056 step 158 | loss train: 1.440, val: 1.307 | iter time: 360.00 ms (step) remaining time: 0:16:54
|
| 276 |
+
Epoch 1 | iter 5088 step 159 | loss train: 1.359, val: 1.307 | iter time: 358.61 ms (step) remaining time: 0:16:43
|
| 277 |
+
Epoch 1 | iter 5120 step 160 | loss train: 1.389, val: 1.307 | iter time: 358.95 ms (step) remaining time: 0:16:31
|
| 278 |
+
Epoch 1 | iter 5152 step 161 | loss train: 1.365, val: 1.307 | iter time: 361.49 ms (step) remaining time: 0:16:19
|
| 279 |
+
Epoch 1 | iter 5184 step 162 | loss train: 1.256, val: 1.307 | iter time: 358.73 ms (step) remaining time: 0:16:08
|
| 280 |
+
Epoch 1 | iter 5216 step 163 | loss train: 1.318, val: 1.307 | iter time: 359.95 ms (step) remaining time: 0:15:56
|
| 281 |
+
Epoch 1 | iter 5248 step 164 | loss train: 1.389, val: 1.307 | iter time: 359.48 ms (step) remaining time: 0:15:44
|
| 282 |
+
Epoch 1 | iter 5280 step 165 | loss train: 1.329, val: 1.307 | iter time: 359.49 ms (step) remaining time: 0:15:33
|
| 283 |
+
Epoch 1 | iter 5312 step 166 | loss train: 1.317, val: 1.307 | iter time: 359.08 ms (step) remaining time: 0:15:21
|
| 284 |
+
Epoch 1 | iter 5344 step 167 | loss train: 1.367, val: 1.307 | iter time: 359.26 ms (step) remaining time: 0:15:09
|
| 285 |
+
Epoch 1 | iter 5376 step 168 | loss train: 1.252, val: 1.307 | iter time: 359.39 ms (step) remaining time: 0:14:58
|
| 286 |
+
Epoch 1 | iter 5408 step 169 | loss train: 1.337, val: 1.307 | iter time: 359.35 ms (step) remaining time: 0:14:46
|
| 287 |
+
Epoch 1 | iter 5440 step 170 | loss train: 1.365, val: 1.307 | iter time: 360.12 ms (step) remaining time: 0:14:35
|
| 288 |
+
Epoch 1 | iter 5472 step 171 | loss train: 1.270, val: 1.307 | iter time: 360.55 ms (step) remaining time: 0:14:23
|
| 289 |
+
Epoch 1 | iter 5504 step 172 | loss train: 1.398, val: 1.307 | iter time: 360.00 ms (step) remaining time: 0:14:11
|
| 290 |
+
Epoch 1 | iter 5536 step 173 | loss train: 1.412, val: 1.307 | iter time: 359.76 ms (step) remaining time: 0:14:00
|
| 291 |
+
Epoch 1 | iter 5568 step 174 | loss train: 1.338, val: 1.307 | iter time: 360.71 ms (step) remaining time: 0:13:48
|
| 292 |
+
Epoch 1 | iter 5600 step 175 | loss train: 1.366, val: 1.307 | iter time: 360.22 ms (step) remaining time: 0:13:37
|
| 293 |
+
Epoch 1 | iter 5632 step 176 | loss train: 1.326, val: 1.307 | iter time: 359.66 ms (step) remaining time: 0:13:25
|
| 294 |
+
Epoch 1 | iter 5664 step 177 | loss train: 1.432, val: 1.307 | iter time: 359.91 ms (step) remaining time: 0:13:14
|
| 295 |
+
Epoch 1 | iter 5696 step 178 | loss train: 1.405, val: 1.307 | iter time: 360.93 ms (step) remaining time: 0:13:02
|
| 296 |
+
Epoch 1 | iter 5728 step 179 | loss train: 1.292, val: 1.307 | iter time: 358.00 ms (step) remaining time: 0:12:51
|
| 297 |
+
Epoch 1 | iter 5760 step 180 | loss train: 1.379, val: 1.307 | iter time: 359.26 ms (step) remaining time: 0:12:39
|
| 298 |
+
Epoch 1 | iter 5792 step 181 | loss train: 1.291, val: 1.307 | iter time: 359.61 ms (step) remaining time: 0:12:27
|
| 299 |
+
Epoch 1 | iter 5824 step 182 | loss train: 1.354, val: 1.307 | iter time: 358.58 ms (step) remaining time: 0:12:16
|
| 300 |
+
Epoch 1 | iter 5856 step 183 | loss train: 1.341, val: 1.307 | iter time: 358.16 ms (step) remaining time: 0:12:05
|
| 301 |
+
Epoch 1 | iter 5888 step 184 | loss train: 1.337, val: 1.307 | iter time: 361.95 ms (step) remaining time: 0:11:53
|
| 302 |
+
Epoch 1 | iter 5920 step 185 | loss train: 1.403, val: 1.307 | iter time: 359.13 ms (step) remaining time: 0:11:42
|
| 303 |
+
Epoch 1 | iter 5952 step 186 | loss train: 1.327, val: 1.307 | iter time: 360.21 ms (step) remaining time: 0:11:30
|
| 304 |
+
Epoch 1 | iter 5984 step 187 | loss train: 1.338, val: 1.307 | iter time: 357.91 ms (step) remaining time: 0:11:19
|
| 305 |
+
Epoch 1 | iter 6016 step 188 | loss train: 1.345, val: 1.307 | iter time: 359.86 ms (step) remaining time: 0:11:07
|
| 306 |
+
Epoch 1 | iter 6048 step 189 | loss train: 1.340, val: 1.307 | iter time: 358.22 ms (step) remaining time: 0:10:56
|
| 307 |
+
Epoch 1 | iter 6080 step 190 | loss train: 1.316, val: 1.307 | iter time: 358.41 ms (step) remaining time: 0:10:44
|
| 308 |
+
Epoch 1 | iter 6112 step 191 | loss train: 1.280, val: 1.307 | iter time: 359.95 ms (step) remaining time: 0:10:33
|
| 309 |
+
Epoch 1 | iter 6144 step 192 | loss train: 1.297, val: 1.307 | iter time: 361.44 ms (step) remaining time: 0:10:21
|
| 310 |
+
Epoch 1 | iter 6176 step 193 | loss train: 1.335, val: 1.307 | iter time: 359.54 ms (step) remaining time: 0:10:10
|
| 311 |
+
Epoch 1 | iter 6208 step 194 | loss train: 1.325, val: 1.307 | iter time: 601.91 ms (step) remaining time: 0:09:59
|
| 312 |
+
Epoch 1 | iter 6240 step 195 | loss train: 1.319, val: 1.307 | iter time: 360.10 ms (step) remaining time: 0:09:47
|
| 313 |
+
Epoch 1 | iter 6272 step 196 | loss train: 1.436, val: 1.307 | iter time: 361.06 ms (step) remaining time: 0:09:36
|
| 314 |
+
Epoch 1 | iter 6304 step 197 | loss train: 1.370, val: 1.307 | iter time: 359.46 ms (step) remaining time: 0:09:24
|
| 315 |
+
Epoch 1 | iter 6336 step 198 | loss train: 1.334, val: 1.307 | iter time: 359.54 ms (step) remaining time: 0:09:13
|
| 316 |
+
Epoch 1 | iter 6368 step 199 | loss train: 1.338, val: 1.307 | iter time: 360.19 ms (step) remaining time: 0:09:02
|
| 317 |
+
Epoch 1 | iter 6400 step 200 | loss train: 1.291, val: 1.307 | iter time: 358.99 ms (step) remaining time: 0:08:50
|
| 318 |
+
Validating ...
|
| 319 |
+
iter 6400: val loss 1.3077, val time: 21938.48 ms
|
| 320 |
+
Saving checkpoint to 'out/pretrain/tinyllama/2410_lr4e-5/step-00000200/lit_model.pth'
|
| 321 |
+
Epoch 1 | iter 6432 step 201 | loss train: 1.266, val: 1.308 | iter time: 358.15 ms (step) remaining time: 0:08:48
|
| 322 |
+
Epoch 1 | iter 6464 step 202 | loss train: 1.420, val: 1.308 | iter time: 358.33 ms (step) remaining time: 0:08:36
|
| 323 |
+
Epoch 1 | iter 6496 step 203 | loss train: 1.386, val: 1.308 | iter time: 357.83 ms (step) remaining time: 0:08:24
|
| 324 |
+
Epoch 1 | iter 6528 step 204 | loss train: 1.381, val: 1.308 | iter time: 358.65 ms (step) remaining time: 0:08:13
|
| 325 |
+
Epoch 1 | iter 6560 step 205 | loss train: 1.332, val: 1.308 | iter time: 358.33 ms (step) remaining time: 0:08:01
|
| 326 |
+
Epoch 1 | iter 6592 step 206 | loss train: 1.342, val: 1.308 | iter time: 357.48 ms (step) remaining time: 0:07:49
|
| 327 |
+
Epoch 1 | iter 6624 step 207 | loss train: 1.273, val: 1.308 | iter time: 359.42 ms (step) remaining time: 0:07:38
|
| 328 |
+
Epoch 1 | iter 6656 step 208 | loss train: 1.347, val: 1.308 | iter time: 359.79 ms (step) remaining time: 0:07:26
|
| 329 |
+
Epoch 1 | iter 6688 step 209 | loss train: 1.321, val: 1.308 | iter time: 359.60 ms (step) remaining time: 0:07:15
|
| 330 |
+
Epoch 1 | iter 6720 step 210 | loss train: 1.356, val: 1.308 | iter time: 357.11 ms (step) remaining time: 0:07:03
|
| 331 |
+
Epoch 1 | iter 6752 step 211 | loss train: 1.302, val: 1.308 | iter time: 357.99 ms (step) remaining time: 0:06:52
|
| 332 |
+
Epoch 1 | iter 6784 step 212 | loss train: 1.326, val: 1.308 | iter time: 358.70 ms (step) remaining time: 0:06:40
|
| 333 |
+
Epoch 1 | iter 6816 step 213 | loss train: 1.310, val: 1.308 | iter time: 358.43 ms (step) remaining time: 0:06:29
|
| 334 |
+
Epoch 1 | iter 6848 step 214 | loss train: 1.315, val: 1.308 | iter time: 359.56 ms (step) remaining time: 0:06:17
|
| 335 |
+
Epoch 1 | iter 6880 step 215 | loss train: 1.358, val: 1.308 | iter time: 359.18 ms (step) remaining time: 0:06:06
|
| 336 |
+
Epoch 1 | iter 6912 step 216 | loss train: 1.358, val: 1.308 | iter time: 360.38 ms (step) remaining time: 0:05:54
|
| 337 |
+
Epoch 1 | iter 6944 step 217 | loss train: 1.273, val: 1.308 | iter time: 360.52 ms (step) remaining time: 0:05:42
|
| 338 |
+
Epoch 1 | iter 6976 step 218 | loss train: 1.302, val: 1.308 | iter time: 359.96 ms (step) remaining time: 0:05:31
|
| 339 |
+
Epoch 1 | iter 7008 step 219 | loss train: 1.278, val: 1.308 | iter time: 360.58 ms (step) remaining time: 0:05:19
|
| 340 |
+
Epoch 1 | iter 7040 step 220 | loss train: 1.260, val: 1.308 | iter time: 359.10 ms (step) remaining time: 0:05:08
|
| 341 |
+
Epoch 1 | iter 7072 step 221 | loss train: 1.382, val: 1.308 | iter time: 359.99 ms (step) remaining time: 0:04:56
|
| 342 |
+
Epoch 1 | iter 7104 step 222 | loss train: 1.306, val: 1.308 | iter time: 359.81 ms (step) remaining time: 0:04:45
|
| 343 |
+
Epoch 1 | iter 7136 step 223 | loss train: 1.324, val: 1.308 | iter time: 359.68 ms (step) remaining time: 0:04:34
|
| 344 |
+
Epoch 1 | iter 7168 step 224 | loss train: 1.346, val: 1.308 | iter time: 359.18 ms (step) remaining time: 0:04:22
|
| 345 |
+
Epoch 1 | iter 7200 step 225 | loss train: 1.341, val: 1.308 | iter time: 358.65 ms (step) remaining time: 0:04:11
|
| 346 |
+
Epoch 1 | iter 7232 step 226 | loss train: 1.374, val: 1.308 | iter time: 359.97 ms (step) remaining time: 0:03:59
|
| 347 |
+
Epoch 1 | iter 7264 step 227 | loss train: 1.348, val: 1.308 | iter time: 359.89 ms (step) remaining time: 0:03:48
|
| 348 |
+
Epoch 1 | iter 7296 step 228 | loss train: 1.372, val: 1.308 | iter time: 358.29 ms (step) remaining time: 0:03:36
|
| 349 |
+
Epoch 1 | iter 7328 step 229 | loss train: 1.357, val: 1.308 | iter time: 358.90 ms (step) remaining time: 0:03:25
|
| 350 |
+
Epoch 1 | iter 7360 step 230 | loss train: 1.391, val: 1.308 | iter time: 358.95 ms (step) remaining time: 0:03:13
|
| 351 |
+
Epoch 1 | iter 7392 step 231 | loss train: 1.294, val: 1.308 | iter time: 358.25 ms (step) remaining time: 0:03:02
|
| 352 |
+
Epoch 1 | iter 7424 step 232 | loss train: 1.358, val: 1.308 | iter time: 360.24 ms (step) remaining time: 0:02:50
|
| 353 |
+
Epoch 1 | iter 7456 step 233 | loss train: 1.320, val: 1.308 | iter time: 358.27 ms (step) remaining time: 0:02:39
|
| 354 |
+
Epoch 1 | iter 7488 step 234 | loss train: 1.354, val: 1.308 | iter time: 360.41 ms (step) remaining time: 0:02:28
|
| 355 |
+
Epoch 1 | iter 7520 step 235 | loss train: 1.355, val: 1.308 | iter time: 361.19 ms (step) remaining time: 0:02:16
|
| 356 |
+
Epoch 1 | iter 7552 step 236 | loss train: 1.352, val: 1.308 | iter time: 359.60 ms (step) remaining time: 0:02:05
|
| 357 |
+
Epoch 1 | iter 7584 step 237 | loss train: 1.353, val: 1.308 | iter time: 359.13 ms (step) remaining time: 0:01:53
|
| 358 |
+
Epoch 1 | iter 7616 step 238 | loss train: 1.296, val: 1.308 | iter time: 361.19 ms (step) remaining time: 0:01:42
|
| 359 |
+
Epoch 1 | iter 7648 step 239 | loss train: 1.425, val: 1.308 | iter time: 359.71 ms (step) remaining time: 0:01:31
|
| 360 |
+
Epoch 1 | iter 7680 step 240 | loss train: 1.341, val: 1.308 | iter time: 360.70 ms (step) remaining time: 0:01:19
|
| 361 |
+
Epoch 1 | iter 7712 step 241 | loss train: 1.389, val: 1.308 | iter time: 358.63 ms (step) remaining time: 0:01:08
|
| 362 |
+
Epoch 1 | iter 7744 step 242 | loss train: 1.308, val: 1.308 | iter time: 359.20 ms (step) remaining time: 0:00:56
|
| 363 |
+
Epoch 1 | iter 7776 step 243 | loss train: 1.372, val: 1.308 | iter time: 360.04 ms (step) remaining time: 0:00:45
|
| 364 |
+
Epoch 1 | iter 7808 step 244 | loss train: 1.375, val: 1.308 | iter time: 360.05 ms (step) remaining time: 0:00:34
|
| 365 |
+
Epoch 1 | iter 7840 step 245 | loss train: 1.313, val: 1.308 | iter time: 360.37 ms (step) remaining time: 0:00:22
|
| 366 |
+
Epoch 1 | iter 7872 step 246 | loss train: 1.275, val: 1.308 | iter time: 363.17 ms (step) remaining time: 0:00:11
|
| 367 |
+
Epoch 2 | iter 7904 step 247 | loss train: 1.325, val: 1.308 | iter time: 358.22 ms (step) remaining time: 0:00:00
|
| 368 |
+
Validating ...
|
| 369 |
+
Final evaluation | val loss: 1.303 | val ppl: 3.680
|
| 370 |
+
Saving checkpoint to 'out/pretrain/tinyllama/2410_lr4e-5/final/lit_model.pth'
|
| 371 |
+
----------------------------------------
|
| 372 |
+
| Performance
|
| 373 |
+
| - Total tokens : 258,998,272
|
| 374 |
+
| - Training Time : 2869.92 s
|
| 375 |
+
| - Tok/sec : 133.59 tok/s
|
| 376 |
+
| ----------------------------------------
|
| 377 |
+
| Memory Usage
|
| 378 |
+
| - Memory Used : 26.32 GB
|
| 379 |
+
----------------------------------------
|
out/pretrain/tinyllama/teelogs/2411.txt
ADDED
|
@@ -0,0 +1,360 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
Initializing distributed: GLOBAL_RANK: 0, MEMBER: 1/4
|
| 2 |
+
Initializing distributed: GLOBAL_RANK: 3, MEMBER: 4/4
|
| 3 |
+
[rank: 3] Seed set to 42
|
| 4 |
+
Initializing distributed: GLOBAL_RANK: 2, MEMBER: 3/4
|
| 5 |
+
[rank: 2] Seed set to 42
|
| 6 |
+
Initializing distributed: GLOBAL_RANK: 1, MEMBER: 2/4
|
| 7 |
+
[rank: 1] Seed set to 42
|
| 8 |
+
----------------------------------------------------------------------------------------------------
|
| 9 |
+
distributed_backend=nccl
|
| 10 |
+
All distributed processes registered. Starting with 4 processes
|
| 11 |
+
----------------------------------------------------------------------------------------------------
|
| 12 |
+
|
| 13 |
+
[rank: 0] Seed set to 42
|
| 14 |
+
All GPUs are fully connected via NVLink.
|
| 15 |
+
{'data': {'batch_size': 1,
|
| 16 |
+
'data_path': PosixPath('litgpt/data/arxiv'),
|
| 17 |
+
'num_workers': 8,
|
| 18 |
+
'seed': 42,
|
| 19 |
+
'seq_length': 2048,
|
| 20 |
+
'use_starcoder': True},
|
| 21 |
+
'data_dir': PosixPath('litgpt/data/arxiv/2411'),
|
| 22 |
+
'devices': 'auto',
|
| 23 |
+
'eval': {'evaluate_example': 'first',
|
| 24 |
+
'final_validation': True,
|
| 25 |
+
'initial_validation': True,
|
| 26 |
+
'interval': 50,
|
| 27 |
+
'max_iters': 100,
|
| 28 |
+
'max_new_tokens': None},
|
| 29 |
+
'initial_checkpoint_dir': PosixPath('out/pretrain/2410/final'),
|
| 30 |
+
'log': {'group': None, 'project': None, 'run': None},
|
| 31 |
+
'logger_name': 'tensorboard',
|
| 32 |
+
'model_config': {'attention_logit_softcapping': None,
|
| 33 |
+
'attention_scores_scalar': None,
|
| 34 |
+
'attn_bias': False,
|
| 35 |
+
'bias': False,
|
| 36 |
+
'block_size': 2048,
|
| 37 |
+
'final_logit_softcapping': None,
|
| 38 |
+
'gelu_approximate': 'none',
|
| 39 |
+
'head_size': 64,
|
| 40 |
+
'hf_config': {'name': 'TinyLlama-1.1B-intermediate-step-1431k-3T',
|
| 41 |
+
'org': 'TinyLlama'},
|
| 42 |
+
'intermediate_size': 5632,
|
| 43 |
+
'lm_head_bias': False,
|
| 44 |
+
'mlp_class_name': 'LLaMAMLP',
|
| 45 |
+
'moe_intermediate_size': None,
|
| 46 |
+
'n_embd': 2048,
|
| 47 |
+
'n_expert': 0,
|
| 48 |
+
'n_expert_per_token': 0,
|
| 49 |
+
'n_head': 32,
|
| 50 |
+
'n_layer': 22,
|
| 51 |
+
'n_query_groups': 4,
|
| 52 |
+
'name': 'tiny-llama-1.1b',
|
| 53 |
+
'norm_1': True,
|
| 54 |
+
'norm_2': True,
|
| 55 |
+
'norm_class_name': 'RMSNorm',
|
| 56 |
+
'norm_eps': 1e-05,
|
| 57 |
+
'norm_qk': False,
|
| 58 |
+
'norm_qk_type': 'default',
|
| 59 |
+
'padded_vocab_size': 32000,
|
| 60 |
+
'padding_multiple': 64,
|
| 61 |
+
'parallel_residual': False,
|
| 62 |
+
'post_attention_norm': False,
|
| 63 |
+
'post_mlp_norm': False,
|
| 64 |
+
'rope_adjustments': None,
|
| 65 |
+
'rope_base': 10000,
|
| 66 |
+
'rope_condense_ratio': 1,
|
| 67 |
+
'rope_indices': None,
|
| 68 |
+
'rope_local_base_freq': None,
|
| 69 |
+
'rotary_percentage': 1.0,
|
| 70 |
+
'scale_embeddings': False,
|
| 71 |
+
'shared_attention_norm': False,
|
| 72 |
+
'sliding_window_indices': None,
|
| 73 |
+
'sliding_window_size': None,
|
| 74 |
+
'vocab_size': 32000},
|
| 75 |
+
'model_name': 'tiny-llama-1.1b',
|
| 76 |
+
'num_nodes': 1,
|
| 77 |
+
'optimizer': "{'class_path': 'torch.optim.AdamW', 'init_args': {'lr': 0.0004, "
|
| 78 |
+
"'weight_decay': 0.1, 'betas': [0.9, 0.95]}}",
|
| 79 |
+
'out_dir': PosixPath('out/pretrain/2411'),
|
| 80 |
+
'precision': 'bf16-mixed',
|
| 81 |
+
'resume': False,
|
| 82 |
+
'seed': 42,
|
| 83 |
+
'tokenizer_dir': PosixPath('checkpoints/TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T'),
|
| 84 |
+
'train': {'epochs': None,
|
| 85 |
+
'global_batch_size': 512,
|
| 86 |
+
'log_interval': 1,
|
| 87 |
+
'lr_warmup_fraction': None,
|
| 88 |
+
'lr_warmup_steps': 20,
|
| 89 |
+
'max_norm': 1.0,
|
| 90 |
+
'max_seq_length': 2048,
|
| 91 |
+
'max_steps': None,
|
| 92 |
+
'max_tokens': 247463936,
|
| 93 |
+
'micro_batch_size': 4,
|
| 94 |
+
'min_lr': 4e-05,
|
| 95 |
+
'save_interval': 100,
|
| 96 |
+
'tie_embeddings': None}}
|
| 97 |
+
Time to instantiate model: 0.02 seconds.
|
| 98 |
+
Total parameters: 1,100,048,384
|
| 99 |
+
[fix] out/pretrain/2410/final/lit_model.pth 是整包 state，提取 model 权重并原子覆盖...
|
| 100 |
+
[fix] 已覆盖为纯权重: out/pretrain/2410/final/lit_model.pth
|
| 101 |
+
Validating ...
|
| 102 |
+
Measured TFLOPs: 239.66
|
| 103 |
+
Epoch 1 | iter 32 step 1 | loss train: 1.384, val: 1.368 | iter time: 533.08 ms (step) remaining time: 0:46:31
|
| 104 |
+
Epoch 1 | iter 64 step 2 | loss train: 1.336, val: 1.368 | iter time: 358.01 ms (step) remaining time: 0:44:05
|
| 105 |
+
Epoch 1 | iter 96 step 3 | loss train: 1.355, val: 1.368 | iter time: 357.77 ms (step) remaining time: 0:43:12
|
| 106 |
+
Epoch 1 | iter 128 step 4 | loss train: 1.369, val: 1.368 | iter time: 354.34 ms (step) remaining time: 0:42:41
|
| 107 |
+
Epoch 1 | iter 160 step 5 | loss train: 1.488, val: 1.368 | iter time: 357.29 ms (step) remaining time: 0:42:19
|
| 108 |
+
Epoch 1 | iter 192 step 6 | loss train: 1.450, val: 1.368 | iter time: 359.57 ms (step) remaining time: 0:42:01
|
| 109 |
+
Epoch 1 | iter 224 step 7 | loss train: 1.324, val: 1.368 | iter time: 358.56 ms (step) remaining time: 0:41:46
|
| 110 |
+
Epoch 1 | iter 256 step 8 | loss train: 1.463, val: 1.368 | iter time: 358.41 ms (step) remaining time: 0:41:32
|
| 111 |
+
Epoch 1 | iter 288 step 9 | loss train: 1.444, val: 1.368 | iter time: 359.05 ms (step) remaining time: 0:41:18
|
| 112 |
+
Epoch 1 | iter 320 step 10 | loss train: 1.408, val: 1.368 | iter time: 361.15 ms (step) remaining time: 0:41:06
|
| 113 |
+
Epoch 1 | iter 352 step 11 | loss train: 1.416, val: 1.368 | iter time: 359.05 ms (step) remaining time: 0:40:53
|
| 114 |
+
Epoch 1 | iter 384 step 12 | loss train: 1.422, val: 1.368 | iter time: 358.24 ms (step) remaining time: 0:40:41
|
| 115 |
+
Epoch 1 | iter 416 step 13 | loss train: 1.382, val: 1.368 | iter time: 360.11 ms (step) remaining time: 0:40:29
|
| 116 |
+
Epoch 1 | iter 448 step 14 | loss train: 1.470, val: 1.368 | iter time: 358.91 ms (step) remaining time: 0:40:18
|
| 117 |
+
Epoch 1 | iter 480 step 15 | loss train: 1.357, val: 1.368 | iter time: 356.48 ms (step) remaining time: 0:40:06
|
| 118 |
+
Epoch 1 | iter 512 step 16 | loss train: 1.457, val: 1.368 | iter time: 360.12 ms (step) remaining time: 0:39:55
|
| 119 |
+
Epoch 1 | iter 544 step 17 | loss train: 1.420, val: 1.368 | iter time: 360.19 ms (step) remaining time: 0:39:43
|
| 120 |
+
Epoch 1 | iter 576 step 18 | loss train: 1.354, val: 1.368 | iter time: 358.56 ms (step) remaining time: 0:39:32
|
| 121 |
+
Epoch 1 | iter 608 step 19 | loss train: 1.424, val: 1.368 | iter time: 361.34 ms (step) remaining time: 0:39:21
|
| 122 |
+
Epoch 1 | iter 640 step 20 | loss train: 1.486, val: 1.368 | iter time: 359.60 ms (step) remaining time: 0:39:10
|
| 123 |
+
Epoch 1 | iter 672 step 21 | loss train: 1.439, val: 1.368 | iter time: 359.33 ms (step) remaining time: 0:38:59
|
| 124 |
+
Epoch 1 | iter 704 step 22 | loss train: 1.404, val: 1.368 | iter time: 359.51 ms (step) remaining time: 0:38:48
|
| 125 |
+
Epoch 1 | iter 736 step 23 | loss train: 1.415, val: 1.368 | iter time: 360.30 ms (step) remaining time: 0:38:37
|
| 126 |
+
Epoch 1 | iter 768 step 24 | loss train: 1.389, val: 1.368 | iter time: 359.13 ms (step) remaining time: 0:38:26
|
| 127 |
+
Epoch 1 | iter 800 step 25 | loss train: 1.428, val: 1.368 | iter time: 360.96 ms (step) remaining time: 0:38:15
|
| 128 |
+
Epoch 1 | iter 832 step 26 | loss train: 1.476, val: 1.368 | iter time: 359.89 ms (step) remaining time: 0:38:04
|
| 129 |
+
Epoch 1 | iter 864 step 27 | loss train: 1.432, val: 1.368 | iter time: 361.16 ms (step) remaining time: 0:37:53
|
| 130 |
+
Epoch 1 | iter 896 step 28 | loss train: 1.451, val: 1.368 | iter time: 359.82 ms (step) remaining time: 0:37:42
|
| 131 |
+
Epoch 1 | iter 928 step 29 | loss train: 1.438, val: 1.368 | iter time: 358.61 ms (step) remaining time: 0:37:31
|
| 132 |
+
Epoch 1 | iter 960 step 30 | loss train: 1.334, val: 1.368 | iter time: 358.93 ms (step) remaining time: 0:37:20
|
| 133 |
+
Epoch 1 | iter 992 step 31 | loss train: 1.447, val: 1.368 | iter time: 359.07 ms (step) remaining time: 0:37:09
|
| 134 |
+
Epoch 1 | iter 1024 step 32 | loss train: 1.439, val: 1.368 | iter time: 358.27 ms (step) remaining time: 0:36:59
|
| 135 |
+
Epoch 1 | iter 1056 step 33 | loss train: 1.355, val: 1.368 | iter time: 358.88 ms (step) remaining time: 0:36:48
|
| 136 |
+
Epoch 1 | iter 1088 step 34 | loss train: 1.349, val: 1.368 | iter time: 359.38 ms (step) remaining time: 0:36:37
|
| 137 |
+
Epoch 1 | iter 1120 step 35 | loss train: 1.491, val: 1.368 | iter time: 361.12 ms (step) remaining time: 0:36:26
|
| 138 |
+
Epoch 1 | iter 1152 step 36 | loss train: 1.405, val: 1.368 | iter time: 360.52 ms (step) remaining time: 0:36:15
|
| 139 |
+
Epoch 1 | iter 1184 step 37 | loss train: 1.444, val: 1.368 | iter time: 359.66 ms (step) remaining time: 0:36:04
|
| 140 |
+
Epoch 1 | iter 1216 step 38 | loss train: 1.419, val: 1.368 | iter time: 361.19 ms (step) remaining time: 0:35:53
|
| 141 |
+
Epoch 1 | iter 1248 step 39 | loss train: 1.442, val: 1.368 | iter time: 740.53 ms (step) remaining time: 0:35:44
|
| 142 |
+
Epoch 1 | iter 1280 step 40 | loss train: 1.410, val: 1.368 | iter time: 359.23 ms (step) remaining time: 0:35:33
|
| 143 |
+
Epoch 1 | iter 1312 step 41 | loss train: 1.447, val: 1.368 | iter time: 359.81 ms (step) remaining time: 0:35:22
|
| 144 |
+
Epoch 1 | iter 1344 step 42 | loss train: 1.471, val: 1.368 | iter time: 360.20 ms (step) remaining time: 0:35:11
|
| 145 |
+
Epoch 1 | iter 1376 step 43 | loss train: 1.330, val: 1.368 | iter time: 360.02 ms (step) remaining time: 0:35:00
|
| 146 |
+
Epoch 1 | iter 1408 step 44 | loss train: 1.475, val: 1.368 | iter time: 360.05 ms (step) remaining time: 0:34:50
|
| 147 |
+
Epoch 1 | iter 1440 step 45 | loss train: 1.403, val: 1.368 | iter time: 358.60 ms (step) remaining time: 0:34:39
|
| 148 |
+
Epoch 1 | iter 1472 step 46 | loss train: 1.459, val: 1.368 | iter time: 359.23 ms (step) remaining time: 0:34:28
|
| 149 |
+
Epoch 1 | iter 1504 step 47 | loss train: 1.436, val: 1.368 | iter time: 360.80 ms (step) remaining time: 0:34:17
|
| 150 |
+
Epoch 1 | iter 1536 step 48 | loss train: 1.388, val: 1.368 | iter time: 359.82 ms (step) remaining time: 0:34:06
|
| 151 |
+
Epoch 1 | iter 1568 step 49 | loss train: 1.371, val: 1.368 | iter time: 361.93 ms (step) remaining time: 0:33:55
|
| 152 |
+
Epoch 1 | iter 1600 step 50 | loss train: 1.383, val: 1.368 | iter time: 358.59 ms (step) remaining time: 0:33:44
|
| 153 |
+
Validating ...
|
| 154 |
+
iter 1600: val loss 1.4162, val time: 9248.40 ms
|
| 155 |
+
Epoch 1 | iter 1632 step 51 | loss train: 1.403, val: 1.416 | iter time: 359.61 ms (step) remaining time: 0:34:06
|
| 156 |
+
Epoch 1 | iter 1664 step 52 | loss train: 1.407, val: 1.416 | iter time: 358.95 ms (step) remaining time: 0:33:55
|
| 157 |
+
Epoch 1 | iter 1696 step 53 | loss train: 1.438, val: 1.416 | iter time: 360.01 ms (step) remaining time: 0:33:43
|
| 158 |
+
Epoch 1 | iter 1728 step 54 | loss train: 1.404, val: 1.416 | iter time: 360.49 ms (step) remaining time: 0:33:31
|
| 159 |
+
Epoch 1 | iter 1760 step 55 | loss train: 1.431, val: 1.416 | iter time: 359.98 ms (step) remaining time: 0:33:20
|
| 160 |
+
Epoch 1 | iter 1792 step 56 | loss train: 1.512, val: 1.416 | iter time: 361.11 ms (step) remaining time: 0:33:08
|
| 161 |
+
Epoch 1 | iter 1824 step 57 | loss train: 1.416, val: 1.416 | iter time: 360.76 ms (step) remaining time: 0:32:56
|
| 162 |
+
Epoch 1 | iter 1856 step 58 | loss train: 1.393, val: 1.416 | iter time: 360.52 ms (step) remaining time: 0:32:45
|
| 163 |
+
Epoch 1 | iter 1888 step 59 | loss train: 1.444, val: 1.416 | iter time: 360.60 ms (step) remaining time: 0:32:33
|
| 164 |
+
Epoch 1 | iter 1920 step 60 | loss train: 1.461, val: 1.416 | iter time: 360.63 ms (step) remaining time: 0:32:22
|
| 165 |
+
Epoch 1 | iter 1952 step 61 | loss train: 1.432, val: 1.416 | iter time: 360.83 ms (step) remaining time: 0:32:10
|
| 166 |
+
Epoch 1 | iter 1984 step 62 | loss train: 1.344, val: 1.416 | iter time: 359.97 ms (step) remaining time: 0:31:59
|
| 167 |
+
Epoch 1 | iter 2016 step 63 | loss train: 1.396, val: 1.416 | iter time: 359.48 ms (step) remaining time: 0:31:47
|
| 168 |
+
Epoch 1 | iter 2048 step 64 | loss train: 1.507, val: 1.416 | iter time: 358.87 ms (step) remaining time: 0:31:36
|
| 169 |
+
Epoch 1 | iter 2080 step 65 | loss train: 1.382, val: 1.416 | iter time: 361.07 ms (step) remaining time: 0:31:24
|
| 170 |
+
Epoch 1 | iter 2112 step 66 | loss train: 1.400, val: 1.416 | iter time: 359.70 ms (step) remaining time: 0:31:13
|
| 171 |
+
Epoch 1 | iter 2144 step 67 | loss train: 1.398, val: 1.416 | iter time: 360.04 ms (step) remaining time: 0:31:01
|
| 172 |
+
Epoch 1 | iter 2176 step 68 | loss train: 1.426, val: 1.416 | iter time: 361.01 ms (step) remaining time: 0:30:50
|
| 173 |
+
Epoch 1 | iter 2208 step 69 | loss train: 1.490, val: 1.416 | iter time: 359.98 ms (step) remaining time: 0:30:39
|
| 174 |
+
Epoch 1 | iter 2240 step 70 | loss train: 1.478, val: 1.416 | iter time: 360.47 ms (step) remaining time: 0:30:27
|
| 175 |
+
Epoch 1 | iter 2272 step 71 | loss train: 1.422, val: 1.416 | iter time: 359.49 ms (step) remaining time: 0:30:16
|
| 176 |
+
Epoch 1 | iter 2304 step 72 | loss train: 1.412, val: 1.416 | iter time: 360.69 ms (step) remaining time: 0:30:05
|
| 177 |
+
Epoch 1 | iter 2336 step 73 | loss train: 1.459, val: 1.416 | iter time: 359.38 ms (step) remaining time: 0:29:53
|
| 178 |
+
Epoch 1 | iter 2368 step 74 | loss train: 1.406, val: 1.416 | iter time: 359.04 ms (step) remaining time: 0:29:42
|
| 179 |
+
Epoch 1 | iter 2400 step 75 | loss train: 1.379, val: 1.416 | iter time: 359.77 ms (step) remaining time: 0:29:31
|
| 180 |
+
Epoch 1 | iter 2432 step 76 | loss train: 1.362, val: 1.416 | iter time: 360.03 ms (step) remaining time: 0:29:19
|
| 181 |
+
Epoch 1 | iter 2464 step 77 | loss train: 1.351, val: 1.416 | iter time: 358.80 ms (step) remaining time: 0:29:08
|
| 182 |
+
Epoch 1 | iter 2496 step 78 | loss train: 1.492, val: 1.416 | iter time: 359.87 ms (step) remaining time: 0:28:57
|
| 183 |
+
Epoch 1 | iter 2528 step 79 | loss train: 1.343, val: 1.416 | iter time: 359.52 ms (step) remaining time: 0:28:45
|
| 184 |
+
Epoch 1 | iter 2560 step 80 | loss train: 1.408, val: 1.416 | iter time: 361.03 ms (step) remaining time: 0:28:34
|
| 185 |
+
Epoch 1 | iter 2592 step 81 | loss train: 1.454, val: 1.416 | iter time: 358.81 ms (step) remaining time: 0:28:23
|
| 186 |
+
Epoch 1 | iter 2624 step 82 | loss train: 1.472, val: 1.416 | iter time: 359.54 ms (step) remaining time: 0:28:12
|
| 187 |
+
Epoch 1 | iter 2656 step 83 | loss train: 1.431, val: 1.416 | iter time: 361.34 ms (step) remaining time: 0:28:01
|
| 188 |
+
Epoch 1 | iter 2688 step 84 | loss train: 1.431, val: 1.416 | iter time: 360.81 ms (step) remaining time: 0:27:50
|
| 189 |
+
Epoch 1 | iter 2720 step 85 | loss train: 1.389, val: 1.416 | iter time: 361.35 ms (step) remaining time: 0:27:38
|
| 190 |
+
Epoch 1 | iter 2752 step 86 | loss train: 1.345, val: 1.416 | iter time: 361.94 ms (step) remaining time: 0:27:27
|
| 191 |
+
Epoch 1 | iter 2784 step 87 | loss train: 1.425, val: 1.416 | iter time: 359.44 ms (step) remaining time: 0:27:16
|
| 192 |
+
Epoch 1 | iter 2816 step 88 | loss train: 1.421, val: 1.416 | iter time: 362.02 ms (step) remaining time: 0:27:05
|
| 193 |
+
Epoch 1 | iter 2848 step 89 | loss train: 1.388, val: 1.416 | iter time: 360.05 ms (step) remaining time: 0:26:54
|
| 194 |
+
Epoch 1 | iter 2880 step 90 | loss train: 1.451, val: 1.416 | iter time: 361.13 ms (step) remaining time: 0:26:42
|
| 195 |
+
Epoch 1 | iter 2912 step 91 | loss train: 1.421, val: 1.416 | iter time: 358.93 ms (step) remaining time: 0:26:32
|
| 196 |
+
Epoch 1 | iter 2944 step 92 | loss train: 1.320, val: 1.416 | iter time: 360.94 ms (step) remaining time: 0:26:21
|
| 197 |
+
Epoch 1 | iter 2976 step 93 | loss train: 1.451, val: 1.416 | iter time: 358.59 ms (step) remaining time: 0:26:09
|
| 198 |
+
Epoch 1 | iter 3008 step 94 | loss train: 1.450, val: 1.416 | iter time: 359.83 ms (step) remaining time: 0:25:58
|
| 199 |
+
Epoch 1 | iter 3040 step 95 | loss train: 1.451, val: 1.416 | iter time: 360.40 ms (step) remaining time: 0:25:47
|
| 200 |
+
Epoch 1 | iter 3072 step 96 | loss train: 1.413, val: 1.416 | iter time: 360.42 ms (step) remaining time: 0:25:36
|
| 201 |
+
Epoch 1 | iter 3104 step 97 | loss train: 1.410, val: 1.416 | iter time: 360.21 ms (step) remaining time: 0:25:25
|
| 202 |
+
Epoch 1 | iter 3136 step 98 | loss train: 1.428, val: 1.416 | iter time: 360.18 ms (step) remaining time: 0:25:14
|
| 203 |
+
Epoch 1 | iter 3168 step 99 | loss train: 1.398, val: 1.416 | iter time: 358.78 ms (step) remaining time: 0:25:03
|
| 204 |
+
Epoch 1 | iter 3200 step 100 | loss train: 1.362, val: 1.416 | iter time: 359.73 ms (step) remaining time: 0:24:51
|
| 205 |
+
Validating ...
|
| 206 |
+
iter 3200: val loss 1.4050, val time: 9254.27 ms
|
| 207 |
+
Saving checkpoint to 'out/pretrain/2411/step-00000100/lit_model.pth'
|
| 208 |
+
Epoch 1 | iter 3232 step 101 | loss train: 1.439, val: 1.405 | iter time: 357.45 ms (step) remaining time: 0:25:15
|
| 209 |
+
Epoch 1 | iter 3264 step 102 | loss train: 1.414, val: 1.405 | iter time: 356.66 ms (step) remaining time: 0:25:03
|
| 210 |
+
Epoch 1 | iter 3296 step 103 | loss train: 1.393, val: 1.405 | iter time: 359.91 ms (step) remaining time: 0:24:52
|
| 211 |
+
Epoch 1 | iter 3328 step 104 | loss train: 1.375, val: 1.405 | iter time: 357.56 ms (step) remaining time: 0:24:40
|
| 212 |
+
Epoch 1 | iter 3360 step 105 | loss train: 1.360, val: 1.405 | iter time: 360.33 ms (step) remaining time: 0:24:28
|
| 213 |
+
Epoch 1 | iter 3392 step 106 | loss train: 1.416, val: 1.405 | iter time: 359.51 ms (step) remaining time: 0:24:17
|
| 214 |
+
Epoch 1 | iter 3424 step 107 | loss train: 1.431, val: 1.405 | iter time: 359.40 ms (step) remaining time: 0:24:05
|
| 215 |
+
Epoch 1 | iter 3456 step 108 | loss train: 1.393, val: 1.405 | iter time: 360.01 ms (step) remaining time: 0:23:53
|
| 216 |
+
Epoch 1 | iter 3488 step 109 | loss train: 1.456, val: 1.405 | iter time: 358.84 ms (step) remaining time: 0:23:42
|
| 217 |
+
Epoch 1 | iter 3520 step 110 | loss train: 1.387, val: 1.405 | iter time: 360.06 ms (step) remaining time: 0:23:30
|
| 218 |
+
Epoch 1 | iter 3552 step 111 | loss train: 1.333, val: 1.405 | iter time: 361.29 ms (step) remaining time: 0:23:19
|
| 219 |
+
Epoch 1 | iter 3584 step 112 | loss train: 1.362, val: 1.405 | iter time: 359.46 ms (step) remaining time: 0:23:07
|
| 220 |
+
Epoch 1 | iter 3616 step 113 | loss train: 1.385, val: 1.405 | iter time: 360.14 ms (step) remaining time: 0:22:56
|
| 221 |
+
Epoch 1 | iter 3648 step 114 | loss train: 1.411, val: 1.405 | iter time: 359.16 ms (step) remaining time: 0:22:44
|
| 222 |
+
Epoch 1 | iter 3680 step 115 | loss train: 1.415, val: 1.405 | iter time: 358.68 ms (step) remaining time: 0:22:32
|
| 223 |
+
Epoch 1 | iter 3712 step 116 | loss train: 1.399, val: 1.405 | iter time: 358.66 ms (step) remaining time: 0:22:21
|
| 224 |
+
Epoch 1 | iter 3744 step 117 | loss train: 1.375, val: 1.405 | iter time: 359.46 ms (step) remaining time: 0:22:09
|
| 225 |
+
Epoch 1 | iter 3776 step 118 | loss train: 1.354, val: 1.405 | iter time: 360.39 ms (step) remaining time: 0:21:58
|
| 226 |
+
Epoch 1 | iter 3808 step 119 | loss train: 1.386, val: 1.405 | iter time: 360.57 ms (step) remaining time: 0:21:47
|
| 227 |
+
Epoch 1 | iter 3840 step 120 | loss train: 1.388, val: 1.405 | iter time: 359.30 ms (step) remaining time: 0:21:35
|
| 228 |
+
Epoch 1 | iter 3872 step 121 | loss train: 1.384, val: 1.405 | iter time: 359.24 ms (step) remaining time: 0:21:24
|
| 229 |
+
Epoch 1 | iter 3904 step 122 | loss train: 1.404, val: 1.405 | iter time: 361.93 ms (step) remaining time: 0:21:12
|
| 230 |
+
Epoch 1 | iter 3936 step 123 | loss train: 1.447, val: 1.405 | iter time: 359.03 ms (step) remaining time: 0:21:01
|
| 231 |
+
Epoch 1 | iter 3968 step 124 | loss train: 1.338, val: 1.405 | iter time: 361.84 ms (step) remaining time: 0:20:49
|
| 232 |
+
Epoch 1 | iter 4000 step 125 | loss train: 1.384, val: 1.405 | iter time: 359.67 ms (step) remaining time: 0:20:38
|
| 233 |
+
Epoch 1 | iter 4032 step 126 | loss train: 1.360, val: 1.405 | iter time: 361.66 ms (step) remaining time: 0:20:26
|
| 234 |
+
Epoch 1 | iter 4064 step 127 | loss train: 1.383, val: 1.405 | iter time: 359.82 ms (step) remaining time: 0:20:15
|
| 235 |
+
Epoch 1 | iter 4096 step 128 | loss train: 1.347, val: 1.405 | iter time: 361.78 ms (step) remaining time: 0:20:04
|
| 236 |
+
Epoch 1 | iter 4128 step 129 | loss train: 1.381, val: 1.405 | iter time: 360.25 ms (step) remaining time: 0:19:52
|
| 237 |
+
Epoch 1 | iter 4160 step 130 | loss train: 1.412, val: 1.405 | iter time: 360.70 ms (step) remaining time: 0:19:41
|
| 238 |
+
Epoch 1 | iter 4192 step 131 | loss train: 1.339, val: 1.405 | iter time: 360.27 ms (step) remaining time: 0:19:30
|
| 239 |
+
Epoch 1 | iter 4224 step 132 | loss train: 1.359, val: 1.405 | iter time: 359.63 ms (step) remaining time: 0:19:18
|
| 240 |
+
Epoch 1 | iter 4256 step 133 | loss train: 1.337, val: 1.405 | iter time: 360.16 ms (step) remaining time: 0:19:07
|
| 241 |
+
Epoch 1 | iter 4288 step 134 | loss train: 1.426, val: 1.405 | iter time: 359.02 ms (step) remaining time: 0:18:56
|
| 242 |
+
Epoch 1 | iter 4320 step 135 | loss train: 1.286, val: 1.405 | iter time: 360.57 ms (step) remaining time: 0:18:44
|
| 243 |
+
Epoch 1 | iter 4352 step 136 | loss train: 1.323, val: 1.405 | iter time: 358.27 ms (step) remaining time: 0:18:33
|
| 244 |
+
Epoch 1 | iter 4384 step 137 | loss train: 1.329, val: 1.405 | iter time: 359.05 ms (step) remaining time: 0:18:22
|
| 245 |
+
Epoch 1 | iter 4416 step 138 | loss train: 1.333, val: 1.405 | iter time: 359.39 ms (step) remaining time: 0:18:10
|
| 246 |
+
Epoch 1 | iter 4448 step 139 | loss train: 1.379, val: 1.405 | iter time: 358.93 ms (step) remaining time: 0:17:59
|
| 247 |
+
Epoch 1 | iter 4480 step 140 | loss train: 1.415, val: 1.405 | iter time: 359.14 ms (step) remaining time: 0:17:48
|
| 248 |
+
Epoch 1 | iter 4512 step 141 | loss train: 1.342, val: 1.405 | iter time: 360.54 ms (step) remaining time: 0:17:36
|
| 249 |
+
Epoch 1 | iter 4544 step 142 | loss train: 1.388, val: 1.405 | iter time: 361.40 ms (step) remaining time: 0:17:25
|
| 250 |
+
Epoch 1 | iter 4576 step 143 | loss train: 1.349, val: 1.405 | iter time: 359.42 ms (step) remaining time: 0:17:14
|
| 251 |
+
Epoch 1 | iter 4608 step 144 | loss train: 1.347, val: 1.405 | iter time: 361.98 ms (step) remaining time: 0:17:03
|
| 252 |
+
Epoch 1 | iter 4640 step 145 | loss train: 1.391, val: 1.405 | iter time: 359.69 ms (step) remaining time: 0:16:52
|
| 253 |
+
Epoch 1 | iter 4672 step 146 | loss train: 1.371, val: 1.405 | iter time: 361.41 ms (step) remaining time: 0:16:40
|
| 254 |
+
Epoch 1 | iter 4704 step 147 | loss train: 1.404, val: 1.405 | iter time: 360.51 ms (step) remaining time: 0:16:29
|
| 255 |
+
Epoch 1 | iter 4736 step 148 | loss train: 1.366, val: 1.405 | iter time: 361.40 ms (step) remaining time: 0:16:18
|
| 256 |
+
Epoch 1 | iter 4768 step 149 | loss train: 1.365, val: 1.405 | iter time: 360.73 ms (step) remaining time: 0:16:06
|
| 257 |
+
Epoch 1 | iter 4800 step 150 | loss train: 1.466, val: 1.405 | iter time: 359.57 ms (step) remaining time: 0:15:55
|
| 258 |
+
Validating ...
|
| 259 |
+
iter 4800: val loss 1.3828, val time: 9254.19 ms
|
| 260 |
+
Epoch 1 | iter 4832 step 151 | loss train: 1.345, val: 1.383 | iter time: 360.62 ms (step) remaining time: 0:15:49
|
| 261 |
+
Epoch 1 | iter 4864 step 152 | loss train: 1.357, val: 1.383 | iter time: 358.48 ms (step) remaining time: 0:15:38
|
| 262 |
+
Epoch 1 | iter 4896 step 153 | loss train: 1.303, val: 1.383 | iter time: 359.80 ms (step) remaining time: 0:15:26
|
| 263 |
+
Epoch 1 | iter 4928 step 154 | loss train: 1.347, val: 1.383 | iter time: 360.01 ms (step) remaining time: 0:15:15
|
| 264 |
+
Epoch 1 | iter 4960 step 155 | loss train: 1.382, val: 1.383 | iter time: 358.21 ms (step) remaining time: 0:15:04
|
| 265 |
+
Epoch 1 | iter 4992 step 156 | loss train: 1.290, val: 1.383 | iter time: 361.00 ms (step) remaining time: 0:14:53
|
| 266 |
+
Epoch 1 | iter 5024 step 157 | loss train: 1.422, val: 1.383 | iter time: 360.10 ms (step) remaining time: 0:14:41
|
| 267 |
+
Epoch 1 | iter 5056 step 158 | loss train: 1.391, val: 1.383 | iter time: 361.31 ms (step) remaining time: 0:14:30
|
| 268 |
+
Epoch 1 | iter 5088 step 159 | loss train: 1.343, val: 1.383 | iter time: 359.19 ms (step) remaining time: 0:14:19
|
| 269 |
+
Epoch 1 | iter 5120 step 160 | loss train: 1.417, val: 1.383 | iter time: 357.94 ms (step) remaining time: 0:14:07
|
| 270 |
+
Epoch 1 | iter 5152 step 161 | loss train: 1.394, val: 1.383 | iter time: 358.74 ms (step) remaining time: 0:13:56
|
| 271 |
+
Epoch 1 | iter 5184 step 162 | loss train: 1.357, val: 1.383 | iter time: 358.74 ms (step) remaining time: 0:13:45
|
| 272 |
+
Epoch 1 | iter 5216 step 163 | loss train: 1.381, val: 1.383 | iter time: 359.33 ms (step) remaining time: 0:13:33
|
| 273 |
+
Epoch 1 | iter 5248 step 164 | loss train: 1.393, val: 1.383 | iter time: 359.34 ms (step) remaining time: 0:13:22
|
| 274 |
+
Epoch 1 | iter 5280 step 165 | loss train: 1.362, val: 1.383 | iter time: 360.92 ms (step) remaining time: 0:13:11
|
| 275 |
+
Epoch 1 | iter 5312 step 166 | loss train: 1.380, val: 1.383 | iter time: 359.72 ms (step) remaining time: 0:13:00
|
| 276 |
+
Epoch 1 | iter 5344 step 167 | loss train: 1.358, val: 1.383 | iter time: 359.68 ms (step) remaining time: 0:12:48
|
| 277 |
+
Epoch 1 | iter 5376 step 168 | loss train: 1.351, val: 1.383 | iter time: 361.31 ms (step) remaining time: 0:12:37
|
| 278 |
+
Epoch 1 | iter 5408 step 169 | loss train: 1.357, val: 1.383 | iter time: 359.44 ms (step) remaining time: 0:12:26
|
| 279 |
+
Epoch 1 | iter 5440 step 170 | loss train: 1.396, val: 1.383 | iter time: 358.26 ms (step) remaining time: 0:12:15
|
| 280 |
+
Epoch 1 | iter 5472 step 171 | loss train: 1.394, val: 1.383 | iter time: 359.21 ms (step) remaining time: 0:12:03
|
| 281 |
+
Epoch 1 | iter 5504 step 172 | loss train: 1.385, val: 1.383 | iter time: 360.31 ms (step) remaining time: 0:11:52
|
| 282 |
+
Epoch 1 | iter 5536 step 173 | loss train: 1.365, val: 1.383 | iter time: 361.09 ms (step) remaining time: 0:11:41
|
| 283 |
+
Epoch 1 | iter 5568 step 174 | loss train: 1.345, val: 1.383 | iter time: 359.66 ms (step) remaining time: 0:11:30
|
| 284 |
+
Epoch 1 | iter 5600 step 175 | loss train: 1.288, val: 1.383 | iter time: 360.00 ms (step) remaining time: 0:11:18
|
| 285 |
+
Epoch 1 | iter 5632 step 176 | loss train: 1.343, val: 1.383 | iter time: 360.03 ms (step) remaining time: 0:11:07
|
| 286 |
+
Epoch 1 | iter 5664 step 177 | loss train: 1.352, val: 1.383 | iter time: 359.72 ms (step) remaining time: 0:10:56
|
| 287 |
+
Epoch 1 | iter 5696 step 178 | loss train: 1.371, val: 1.383 | iter time: 361.38 ms (step) remaining time: 0:10:45
|
| 288 |
+
Epoch 1 | iter 5728 step 179 | loss train: 1.364, val: 1.383 | iter time: 358.27 ms (step) remaining time: 0:10:34
|
| 289 |
+
Epoch 1 | iter 5760 step 180 | loss train: 1.341, val: 1.383 | iter time: 359.06 ms (step) remaining time: 0:10:22
|
| 290 |
+
Epoch 1 | iter 5792 step 181 | loss train: 1.406, val: 1.383 | iter time: 357.99 ms (step) remaining time: 0:10:11
|
| 291 |
+
Epoch 1 | iter 5824 step 182 | loss train: 1.419, val: 1.383 | iter time: 358.49 ms (step) remaining time: 0:10:00
|
| 292 |
+
Epoch 1 | iter 5856 step 183 | loss train: 1.361, val: 1.383 | iter time: 360.05 ms (step) remaining time: 0:09:49
|
| 293 |
+
Epoch 1 | iter 5888 step 184 | loss train: 1.406, val: 1.383 | iter time: 359.90 ms (step) remaining time: 0:09:38
|
| 294 |
+
Epoch 1 | iter 5920 step 185 | loss train: 1.304, val: 1.383 | iter time: 359.92 ms (step) remaining time: 0:09:26
|
| 295 |
+
Epoch 1 | iter 5952 step 186 | loss train: 1.379, val: 1.383 | iter time: 359.36 ms (step) remaining time: 0:09:15
|
| 296 |
+
Epoch 1 | iter 5984 step 187 | loss train: 1.330, val: 1.383 | iter time: 360.07 ms (step) remaining time: 0:09:04
|
| 297 |
+
Epoch 1 | iter 6016 step 188 | loss train: 1.376, val: 1.383 | iter time: 360.45 ms (step) remaining time: 0:08:53
|
| 298 |
+
Epoch 1 | iter 6048 step 189 | loss train: 1.328, val: 1.383 | iter time: 360.58 ms (step) remaining time: 0:08:42
|
| 299 |
+
Epoch 1 | iter 6080 step 190 | loss train: 1.340, val: 1.383 | iter time: 359.85 ms (step) remaining time: 0:08:31
|
| 300 |
+
Epoch 1 | iter 6112 step 191 | loss train: 1.291, val: 1.383 | iter time: 360.37 ms (step) remaining time: 0:08:19
|
| 301 |
+
Epoch 1 | iter 6144 step 192 | loss train: 1.360, val: 1.383 | iter time: 359.17 ms (step) remaining time: 0:08:08
|
| 302 |
+
Epoch 1 | iter 6176 step 193 | loss train: 1.341, val: 1.383 | iter time: 357.75 ms (step) remaining time: 0:07:57
|
| 303 |
+
Epoch 1 | iter 6208 step 194 | loss train: 1.360, val: 1.383 | iter time: 360.25 ms (step) remaining time: 0:07:46
|
| 304 |
+
Epoch 1 | iter 6240 step 195 | loss train: 1.347, val: 1.383 | iter time: 358.13 ms (step) remaining time: 0:07:35
|
| 305 |
+
Epoch 1 | iter 6272 step 196 | loss train: 1.321, val: 1.383 | iter time: 360.99 ms (step) remaining time: 0:07:24
|
| 306 |
+
Epoch 1 | iter 6304 step 197 | loss train: 1.358, val: 1.383 | iter time: 359.30 ms (step) remaining time: 0:07:13
|
| 307 |
+
Epoch 1 | iter 6336 step 198 | loss train: 1.309, val: 1.383 | iter time: 359.97 ms (step) remaining time: 0:07:01
|
| 308 |
+
Epoch 1 | iter 6368 step 199 | loss train: 1.285, val: 1.383 | iter time: 362.01 ms (step) remaining time: 0:06:50
|
| 309 |
+
Epoch 1 | iter 6400 step 200 | loss train: 1.358, val: 1.383 | iter time: 360.60 ms (step) remaining time: 0:06:39
|
| 310 |
+
Validating ...
|
| 311 |
+
iter 6400: val loss 1.3643, val time: 9276.42 ms
|
| 312 |
+
Saving checkpoint to 'out/pretrain/2411/step-00000200/lit_model.pth'
|
| 313 |
+
Epoch 1 | iter 6432 step 201 | loss train: 1.301, val: 1.364 | iter time: 357.65 ms (step) remaining time: 0:06:33
|
| 314 |
+
Epoch 1 | iter 6464 step 202 | loss train: 1.383, val: 1.364 | iter time: 357.12 ms (step) remaining time: 0:06:21
|
| 315 |
+
Epoch 1 | iter 6496 step 203 | loss train: 1.345, val: 1.364 | iter time: 357.83 ms (step) remaining time: 0:06:10
|
| 316 |
+
Epoch 1 | iter 6528 step 204 | loss train: 1.373, val: 1.364 | iter time: 359.57 ms (step) remaining time: 0:05:59
|
| 317 |
+
Epoch 1 | iter 6560 step 205 | loss train: 1.372, val: 1.364 | iter time: 359.81 ms (step) remaining time: 0:05:47
|
| 318 |
+
Epoch 1 | iter 6592 step 206 | loss train: 1.298, val: 1.364 | iter time: 358.90 ms (step) remaining time: 0:05:36
|
| 319 |
+
Epoch 1 | iter 6624 step 207 | loss train: 1.378, val: 1.364 | iter time: 361.14 ms (step) remaining time: 0:05:25
|
| 320 |
+
Epoch 1 | iter 6656 step 208 | loss train: 1.347, val: 1.364 | iter time: 360.97 ms (step) remaining time: 0:05:14
|
| 321 |
+
Epoch 1 | iter 6688 step 209 | loss train: 1.376, val: 1.364 | iter time: 360.26 ms (step) remaining time: 0:05:02
|
| 322 |
+
Epoch 1 | iter 6720 step 210 | loss train: 1.289, val: 1.364 | iter time: 360.51 ms (step) remaining time: 0:04:51
|
| 323 |
+
Epoch 1 | iter 6752 step 211 | loss train: 1.302, val: 1.364 | iter time: 359.02 ms (step) remaining time: 0:04:40
|
| 324 |
+
Epoch 1 | iter 6784 step 212 | loss train: 1.287, val: 1.364 | iter time: 358.60 ms (step) remaining time: 0:04:29
|
| 325 |
+
Epoch 1 | iter 6816 step 213 | loss train: 1.340, val: 1.364 | iter time: 361.35 ms (step) remaining time: 0:04:17
|
| 326 |
+
Epoch 1 | iter 6848 step 214 | loss train: 1.362, val: 1.364 | iter time: 360.09 ms (step) remaining time: 0:04:06
|
| 327 |
+
Epoch 1 | iter 6880 step 215 | loss train: 1.334, val: 1.364 | iter time: 359.07 ms (step) remaining time: 0:03:55
|
| 328 |
+
Epoch 1 | iter 6912 step 216 | loss train: 1.293, val: 1.364 | iter time: 358.31 ms (step) remaining time: 0:03:44
|
| 329 |
+
Epoch 1 | iter 6944 step 217 | loss train: 1.274, val: 1.364 | iter time: 359.84 ms (step) remaining time: 0:03:32
|
| 330 |
+
Epoch 1 | iter 6976 step 218 | loss train: 1.331, val: 1.364 | iter time: 360.24 ms (step) remaining time: 0:03:21
|
| 331 |
+
Epoch 1 | iter 7008 step 219 | loss train: 1.396, val: 1.364 | iter time: 359.51 ms (step) remaining time: 0:03:10
|
| 332 |
+
Epoch 1 | iter 7040 step 220 | loss train: 1.305, val: 1.364 | iter time: 359.02 ms (step) remaining time: 0:02:59
|
| 333 |
+
Epoch 1 | iter 7072 step 221 | loss train: 1.274, val: 1.364 | iter time: 360.91 ms (step) remaining time: 0:02:47
|
| 334 |
+
Epoch 1 | iter 7104 step 222 | loss train: 1.315, val: 1.364 | iter time: 360.72 ms (step) remaining time: 0:02:36
|
| 335 |
+
Epoch 1 | iter 7136 step 223 | loss train: 1.319, val: 1.364 | iter time: 359.08 ms (step) remaining time: 0:02:25
|
| 336 |
+
Epoch 1 | iter 7168 step 224 | loss train: 1.274, val: 1.364 | iter time: 358.49 ms (step) remaining time: 0:02:14
|
| 337 |
+
Epoch 1 | iter 7200 step 225 | loss train: 1.383, val: 1.364 | iter time: 358.98 ms (step) remaining time: 0:02:03
|
| 338 |
+
Epoch 1 | iter 7232 step 226 | loss train: 1.332, val: 1.364 | iter time: 361.54 ms (step) remaining time: 0:01:51
|
| 339 |
+
Epoch 1 | iter 7264 step 227 | loss train: 1.358, val: 1.364 | iter time: 359.08 ms (step) remaining time: 0:01:40
|
| 340 |
+
Epoch 1 | iter 7296 step 228 | loss train: 1.330, val: 1.364 | iter time: 359.41 ms (step) remaining time: 0:01:29
|
| 341 |
+
Epoch 1 | iter 7328 step 229 | loss train: 1.388, val: 1.364 | iter time: 360.89 ms (step) remaining time: 0:01:18
|
| 342 |
+
Epoch 1 | iter 7360 step 230 | loss train: 1.324, val: 1.364 | iter time: 359.52 ms (step) remaining time: 0:01:07
|
| 343 |
+
Epoch 1 | iter 7392 step 231 | loss train: 1.350, val: 1.364 | iter time: 360.05 ms (step) remaining time: 0:00:55
|
| 344 |
+
Epoch 1 | iter 7424 step 232 | loss train: 1.282, val: 1.364 | iter time: 360.44 ms (step) remaining time: 0:00:44
|
| 345 |
+
Epoch 1 | iter 7456 step 233 | loss train: 1.389, val: 1.364 | iter time: 358.88 ms (step) remaining time: 0:00:33
|
| 346 |
+
Epoch 1 | iter 7488 step 234 | loss train: 1.346, val: 1.364 | iter time: 358.37 ms (step) remaining time: 0:00:22
|
| 347 |
+
Epoch 1 | iter 7520 step 235 | loss train: 1.326, val: 1.364 | iter time: 359.94 ms (step) remaining time: 0:00:11
|
| 348 |
+
Epoch 1 | iter 7552 step 236 | loss train: 1.333, val: 1.364 | iter time: 360.17 ms (step) remaining time: 0:00:00
|
| 349 |
+
Validating ...
|
| 350 |
+
Final evaluation | val loss: 1.357 | val ppl: 3.883
|
| 351 |
+
Saving checkpoint to 'out/pretrain/2411/final/lit_model.pth'
|
| 352 |
+
----------------------------------------
|
| 353 |
+
| Performance
|
| 354 |
+
| - Total tokens : 247,463,936
|
| 355 |
+
| - Training Time : 2675.12 s
|
| 356 |
+
| - Tok/sec : 221.25 tok/s
|
| 357 |
+
| ----------------------------------------
|
| 358 |
+
| Memory Usage
|
| 359 |
+
| - Memory Used : 26.32 GB
|
| 360 |
+
----------------------------------------
|
out/pretrain/tinyllama/teelogs/2411_full.txt
ADDED
|
@@ -0,0 +1,372 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
Initializing distributed: GLOBAL_RANK: 0, MEMBER: 1/4
|
| 2 |
+
Initializing distributed: GLOBAL_RANK: 2, MEMBER: 3/4
|
| 3 |
+
[rank: 2] Seed set to 42
|
| 4 |
+
Initializing distributed: GLOBAL_RANK: 3, MEMBER: 4/4
|
| 5 |
+
[rank: 3] Seed set to 42
|
| 6 |
+
Initializing distributed: GLOBAL_RANK: 1, MEMBER: 2/4
|
| 7 |
+
----------------------------------------------------------------------------------------------------
|
| 8 |
+
distributed_backend=nccl
|
| 9 |
+
All distributed processes registered. Starting with 4 processes
|
| 10 |
+
----------------------------------------------------------------------------------------------------
|
| 11 |
+
|
| 12 |
+
[rank: 1] Seed set to 42
|
| 13 |
+
[rank: 0] Seed set to 42
|
| 14 |
+
/mnt/data/litgpt/litgpt/pretrain.py:515: UserWarning: A newer version of litdata is available (0.2.52). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 15 |
+
train_dataloader = data.train_dataloader()
|
| 16 |
+
/mnt/data/litgpt/litgpt/pretrain.py:515: UserWarning: A newer version of litdata is available (0.2.52). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 17 |
+
train_dataloader = data.train_dataloader()
|
| 18 |
+
/mnt/data/litgpt/litgpt/pretrain.py:515: UserWarning: A newer version of litdata is available (0.2.52). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 19 |
+
train_dataloader = data.train_dataloader()
|
| 20 |
+
/mnt/data/litgpt/litgpt/pretrain.py:515: UserWarning: A newer version of litdata is available (0.2.52). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 21 |
+
train_dataloader = data.train_dataloader()
|
| 22 |
+
All GPUs are fully connected via NVLink.
|
| 23 |
+
{'data': {'batch_size': 1,
|
| 24 |
+
'data_path': PosixPath('litgpt/data/arxiv'),
|
| 25 |
+
'num_workers': 8,
|
| 26 |
+
'seed': 42,
|
| 27 |
+
'seq_length': 2048,
|
| 28 |
+
'use_starcoder': True},
|
| 29 |
+
'data_dir': PosixPath('litgpt/data/arxiv/2411'),
|
| 30 |
+
'devices': 'auto',
|
| 31 |
+
'eval': {'evaluate_example': 'first',
|
| 32 |
+
'final_validation': True,
|
| 33 |
+
'initial_validation': True,
|
| 34 |
+
'interval': 50,
|
| 35 |
+
'max_iters': 200,
|
| 36 |
+
'max_new_tokens': None},
|
| 37 |
+
'initial_checkpoint_dir': PosixPath('out/pretrain/2410_full/final'),
|
| 38 |
+
'log': {'group': None, 'project': None, 'run': None},
|
| 39 |
+
'logger_name': 'tensorboard',
|
| 40 |
+
'model_config': {'attention_logit_softcapping': None,
|
| 41 |
+
'attention_scores_scalar': None,
|
| 42 |
+
'attn_bias': False,
|
| 43 |
+
'bias': False,
|
| 44 |
+
'block_size': 2048,
|
| 45 |
+
'final_logit_softcapping': None,
|
| 46 |
+
'gelu_approximate': 'none',
|
| 47 |
+
'head_size': 64,
|
| 48 |
+
'hf_config': {'name': 'TinyLlama-1.1B-intermediate-step-1431k-3T',
|
| 49 |
+
'org': 'TinyLlama'},
|
| 50 |
+
'intermediate_size': 5632,
|
| 51 |
+
'lm_head_bias': False,
|
| 52 |
+
'mlp_class_name': 'LLaMAMLP',
|
| 53 |
+
'moe_intermediate_size': None,
|
| 54 |
+
'n_embd': 2048,
|
| 55 |
+
'n_expert': 0,
|
| 56 |
+
'n_expert_per_token': 0,
|
| 57 |
+
'n_head': 32,
|
| 58 |
+
'n_layer': 22,
|
| 59 |
+
'n_query_groups': 4,
|
| 60 |
+
'name': 'tiny-llama-1.1b',
|
| 61 |
+
'norm_1': True,
|
| 62 |
+
'norm_2': True,
|
| 63 |
+
'norm_class_name': 'RMSNorm',
|
| 64 |
+
'norm_eps': 1e-05,
|
| 65 |
+
'norm_qk': False,
|
| 66 |
+
'norm_qk_type': 'default',
|
| 67 |
+
'padded_vocab_size': 32000,
|
| 68 |
+
'padding_multiple': 64,
|
| 69 |
+
'parallel_residual': False,
|
| 70 |
+
'post_attention_norm': False,
|
| 71 |
+
'post_mlp_norm': False,
|
| 72 |
+
'rope_adjustments': None,
|
| 73 |
+
'rope_base': 10000,
|
| 74 |
+
'rope_condense_ratio': 1,
|
| 75 |
+
'rope_indices': None,
|
| 76 |
+
'rope_local_base_freq': None,
|
| 77 |
+
'rotary_percentage': 1.0,
|
| 78 |
+
'scale_embeddings': False,
|
| 79 |
+
'shared_attention_norm': False,
|
| 80 |
+
'sliding_window_indices': None,
|
| 81 |
+
'sliding_window_size': None,
|
| 82 |
+
'vocab_size': 32000},
|
| 83 |
+
'model_name': 'tiny-llama-1.1b',
|
| 84 |
+
'num_nodes': 1,
|
| 85 |
+
'optimizer': "{'class_path': 'torch.optim.AdamW', 'init_args': {'lr': 0.0004, "
|
| 86 |
+
"'weight_decay': 0.1, 'betas': [0.9, 0.95]}}",
|
| 87 |
+
'out_dir': PosixPath('out/pretrain/2411_full'),
|
| 88 |
+
'ppl': False,
|
| 89 |
+
'precision': 'bf16-mixed',
|
| 90 |
+
'resume': False,
|
| 91 |
+
'seed': 42,
|
| 92 |
+
'tokenizer_dir': PosixPath('checkpoints/TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T'),
|
| 93 |
+
'train': {'epochs': None,
|
| 94 |
+
'global_batch_size': 512,
|
| 95 |
+
'log_interval': 1,
|
| 96 |
+
'lr_warmup_fraction': None,
|
| 97 |
+
'lr_warmup_steps': 20,
|
| 98 |
+
'max_norm': 1.0,
|
| 99 |
+
'max_seq_length': 2048,
|
| 100 |
+
'max_steps': None,
|
| 101 |
+
'max_tokens': 250609664,
|
| 102 |
+
'micro_batch_size': 4,
|
| 103 |
+
'min_lr': 4e-05,
|
| 104 |
+
'save_interval': 100,
|
| 105 |
+
'tie_embeddings': None}}
|
| 106 |
+
Time to instantiate model: 0.04 seconds.
|
| 107 |
+
Total parameters: 1,100,048,384
|
| 108 |
+
[fix] out/pretrain/2410_full/final/lit_model.pth 是整包 state，提取 model 权重并原子覆盖...
|
| 109 |
+
[fix] 已覆盖为纯权重: out/pretrain/2410_full/final/lit_model.pth
|
| 110 |
+
Validating ...
|
| 111 |
+
Measured TFLOPs: 239.66
|
| 112 |
+
Epoch 1 | iter 32 step 1 | loss train: 1.385, val: 1.312 | iter time: 536.05 ms (step) remaining time: 0:47:19
|
| 113 |
+
Epoch 1 | iter 64 step 2 | loss train: 1.337, val: 1.312 | iter time: 358.46 ms (step) remaining time: 0:44:50
|
| 114 |
+
Epoch 1 | iter 96 step 3 | loss train: 1.356, val: 1.312 | iter time: 358.56 ms (step) remaining time: 0:43:54
|
| 115 |
+
Epoch 1 | iter 128 step 4 | loss train: 1.369, val: 1.312 | iter time: 357.84 ms (step) remaining time: 0:43:21
|
| 116 |
+
Epoch 1 | iter 160 step 5 | loss train: 1.488, val: 1.312 | iter time: 357.15 ms (step) remaining time: 0:42:57
|
| 117 |
+
Epoch 1 | iter 192 step 6 | loss train: 1.451, val: 1.312 | iter time: 357.91 ms (step) remaining time: 0:42:38
|
| 118 |
+
Epoch 1 | iter 224 step 7 | loss train: 1.325, val: 1.312 | iter time: 360.05 ms (step) remaining time: 0:42:23
|
| 119 |
+
Epoch 1 | iter 256 step 8 | loss train: 1.463, val: 1.312 | iter time: 359.24 ms (step) remaining time: 0:42:08
|
| 120 |
+
Epoch 1 | iter 288 step 9 | loss train: 1.442, val: 1.312 | iter time: 360.24 ms (step) remaining time: 0:41:55
|
| 121 |
+
Epoch 1 | iter 320 step 10 | loss train: 1.406, val: 1.312 | iter time: 359.20 ms (step) remaining time: 0:41:42
|
| 122 |
+
Epoch 1 | iter 352 step 11 | loss train: 1.415, val: 1.312 | iter time: 360.21 ms (step) remaining time: 0:41:29
|
| 123 |
+
Epoch 1 | iter 384 step 12 | loss train: 1.419, val: 1.312 | iter time: 359.54 ms (step) remaining time: 0:41:17
|
| 124 |
+
Epoch 1 | iter 416 step 13 | loss train: 1.382, val: 1.312 | iter time: 361.15 ms (step) remaining time: 0:41:05
|
| 125 |
+
Epoch 1 | iter 448 step 14 | loss train: 1.473, val: 1.312 | iter time: 360.67 ms (step) remaining time: 0:40:53
|
| 126 |
+
Epoch 1 | iter 480 step 15 | loss train: 1.360, val: 1.312 | iter time: 359.17 ms (step) remaining time: 0:40:41
|
| 127 |
+
Epoch 1 | iter 512 step 16 | loss train: 1.460, val: 1.312 | iter time: 361.14 ms (step) remaining time: 0:40:30
|
| 128 |
+
Epoch 1 | iter 544 step 17 | loss train: 1.422, val: 1.312 | iter time: 358.70 ms (step) remaining time: 0:40:18
|
| 129 |
+
Epoch 1 | iter 576 step 18 | loss train: 1.353, val: 1.312 | iter time: 359.36 ms (step) remaining time: 0:40:07
|
| 130 |
+
Epoch 1 | iter 608 step 19 | loss train: 1.420, val: 1.312 | iter time: 360.97 ms (step) remaining time: 0:39:56
|
| 131 |
+
Epoch 1 | iter 640 step 20 | loss train: 1.477, val: 1.312 | iter time: 359.06 ms (step) remaining time: 0:39:44
|
| 132 |
+
Epoch 1 | iter 672 step 21 | loss train: 1.434, val: 1.312 | iter time: 359.01 ms (step) remaining time: 0:39:33
|
| 133 |
+
Epoch 1 | iter 704 step 22 | loss train: 1.402, val: 1.312 | iter time: 361.25 ms (step) remaining time: 0:39:22
|
| 134 |
+
Epoch 1 | iter 736 step 23 | loss train: 1.421, val: 1.312 | iter time: 357.72 ms (step) remaining time: 0:39:11
|
| 135 |
+
Epoch 1 | iter 768 step 24 | loss train: 1.394, val: 1.312 | iter time: 359.50 ms (step) remaining time: 0:39:00
|
| 136 |
+
Epoch 1 | iter 800 step 25 | loss train: 1.428, val: 1.312 | iter time: 359.04 ms (step) remaining time: 0:38:49
|
| 137 |
+
Epoch 1 | iter 832 step 26 | loss train: 1.477, val: 1.312 | iter time: 358.50 ms (step) remaining time: 0:38:37
|
| 138 |
+
Epoch 1 | iter 864 step 27 | loss train: 1.431, val: 1.312 | iter time: 360.47 ms (step) remaining time: 0:38:26
|
| 139 |
+
Epoch 1 | iter 896 step 28 | loss train: 1.452, val: 1.312 | iter time: 360.67 ms (step) remaining time: 0:38:15
|
| 140 |
+
Epoch 1 | iter 928 step 29 | loss train: 1.439, val: 1.312 | iter time: 359.16 ms (step) remaining time: 0:38:04
|
| 141 |
+
Epoch 1 | iter 960 step 30 | loss train: 1.332, val: 1.312 | iter time: 359.50 ms (step) remaining time: 0:37:53
|
| 142 |
+
Epoch 1 | iter 992 step 31 | loss train: 1.444, val: 1.312 | iter time: 360.83 ms (step) remaining time: 0:37:42
|
| 143 |
+
Epoch 1 | iter 1024 step 32 | loss train: 1.436, val: 1.312 | iter time: 359.66 ms (step) remaining time: 0:37:31
|
| 144 |
+
Epoch 1 | iter 1056 step 33 | loss train: 1.353, val: 1.312 | iter time: 359.79 ms (step) remaining time: 0:37:20
|
| 145 |
+
Epoch 1 | iter 1088 step 34 | loss train: 1.349, val: 1.312 | iter time: 358.40 ms (step) remaining time: 0:37:09
|
| 146 |
+
Epoch 1 | iter 1120 step 35 | loss train: 1.492, val: 1.312 | iter time: 361.49 ms (step) remaining time: 0:36:58
|
| 147 |
+
Epoch 1 | iter 1152 step 36 | loss train: 1.406, val: 1.312 | iter time: 360.74 ms (step) remaining time: 0:36:48
|
| 148 |
+
Epoch 1 | iter 1184 step 37 | loss train: 1.444, val: 1.312 | iter time: 359.37 ms (step) remaining time: 0:36:37
|
| 149 |
+
Epoch 1 | iter 1216 step 38 | loss train: 1.421, val: 1.312 | iter time: 359.50 ms (step) remaining time: 0:36:26
|
| 150 |
+
Epoch 1 | iter 1248 step 39 | loss train: 1.443, val: 1.312 | iter time: 361.23 ms (step) remaining time: 0:36:15
|
| 151 |
+
Epoch 1 | iter 1280 step 40 | loss train: 1.412, val: 1.312 | iter time: 360.27 ms (step) remaining time: 0:36:04
|
| 152 |
+
Epoch 1 | iter 1312 step 41 | loss train: 1.448, val: 1.312 | iter time: 357.78 ms (step) remaining time: 0:35:53
|
| 153 |
+
Epoch 1 | iter 1344 step 42 | loss train: 1.471, val: 1.312 | iter time: 359.73 ms (step) remaining time: 0:35:43
|
| 154 |
+
Epoch 1 | iter 1376 step 43 | loss train: 1.332, val: 1.312 | iter time: 360.19 ms (step) remaining time: 0:35:32
|
| 155 |
+
Epoch 1 | iter 1408 step 44 | loss train: 1.476, val: 1.312 | iter time: 360.41 ms (step) remaining time: 0:35:21
|
| 156 |
+
Epoch 1 | iter 1440 step 45 | loss train: 1.401, val: 1.312 | iter time: 358.95 ms (step) remaining time: 0:35:10
|
| 157 |
+
Epoch 1 | iter 1472 step 46 | loss train: 1.456, val: 1.312 | iter time: 358.22 ms (step) remaining time: 0:34:59
|
| 158 |
+
Epoch 1 | iter 1504 step 47 | loss train: 1.435, val: 1.312 | iter time: 357.37 ms (step) remaining time: 0:34:48
|
| 159 |
+
Epoch 1 | iter 1536 step 48 | loss train: 1.385, val: 1.312 | iter time: 362.32 ms (step) remaining time: 0:34:37
|
| 160 |
+
Epoch 1 | iter 1568 step 49 | loss train: 1.370, val: 1.312 | iter time: 360.46 ms (step) remaining time: 0:34:26
|
| 161 |
+
Epoch 1 | iter 1600 step 50 | loss train: 1.382, val: 1.312 | iter time: 359.46 ms (step) remaining time: 0:34:15
|
| 162 |
+
Validating ...
|
| 163 |
+
iter 1600: val loss 1.2939, val time: 22370.16 ms
|
| 164 |
+
Epoch 1 | iter 1632 step 51 | loss train: 1.405, val: 1.294 | iter time: 360.53 ms (step) remaining time: 0:35:29
|
| 165 |
+
Epoch 1 | iter 1664 step 52 | loss train: 1.407, val: 1.294 | iter time: 359.34 ms (step) remaining time: 0:35:16
|
| 166 |
+
Epoch 1 | iter 1696 step 53 | loss train: 1.439, val: 1.294 | iter time: 358.07 ms (step) remaining time: 0:35:03
|
| 167 |
+
Epoch 1 | iter 1728 step 54 | loss train: 1.402, val: 1.294 | iter time: 359.39 ms (step) remaining time: 0:34:50
|
| 168 |
+
Epoch 1 | iter 1760 step 55 | loss train: 1.430, val: 1.294 | iter time: 361.28 ms (step) remaining time: 0:34:37
|
| 169 |
+
Epoch 1 | iter 1792 step 56 | loss train: 1.509, val: 1.294 | iter time: 358.62 ms (step) remaining time: 0:34:25
|
| 170 |
+
Epoch 1 | iter 1824 step 57 | loss train: 1.416, val: 1.294 | iter time: 358.34 ms (step) remaining time: 0:34:12
|
| 171 |
+
Epoch 1 | iter 1856 step 58 | loss train: 1.394, val: 1.294 | iter time: 358.73 ms (step) remaining time: 0:33:59
|
| 172 |
+
Epoch 1 | iter 1888 step 59 | loss train: 1.443, val: 1.294 | iter time: 360.27 ms (step) remaining time: 0:33:47
|
| 173 |
+
Epoch 1 | iter 1920 step 60 | loss train: 1.460, val: 1.294 | iter time: 360.60 ms (step) remaining time: 0:33:34
|
| 174 |
+
Epoch 1 | iter 1952 step 61 | loss train: 1.432, val: 1.294 | iter time: 359.26 ms (step) remaining time: 0:33:22
|
| 175 |
+
Epoch 1 | iter 1984 step 62 | loss train: 1.343, val: 1.294 | iter time: 358.09 ms (step) remaining time: 0:33:10
|
| 176 |
+
Epoch 1 | iter 2016 step 63 | loss train: 1.394, val: 1.294 | iter time: 358.02 ms (step) remaining time: 0:32:57
|
| 177 |
+
Epoch 1 | iter 2048 step 64 | loss train: 1.506, val: 1.294 | iter time: 359.22 ms (step) remaining time: 0:32:45
|
| 178 |
+
Epoch 1 | iter 2080 step 65 | loss train: 1.380, val: 1.294 | iter time: 359.89 ms (step) remaining time: 0:32:33
|
| 179 |
+
Epoch 1 | iter 2112 step 66 | loss train: 1.400, val: 1.294 | iter time: 360.79 ms (step) remaining time: 0:32:20
|
| 180 |
+
Epoch 1 | iter 2144 step 67 | loss train: 1.399, val: 1.294 | iter time: 358.23 ms (step) remaining time: 0:32:08
|
| 181 |
+
Epoch 1 | iter 2176 step 68 | loss train: 1.425, val: 1.294 | iter time: 360.82 ms (step) remaining time: 0:31:56
|
| 182 |
+
Epoch 1 | iter 2208 step 69 | loss train: 1.490, val: 1.294 | iter time: 358.42 ms (step) remaining time: 0:31:44
|
| 183 |
+
Epoch 1 | iter 2240 step 70 | loss train: 1.476, val: 1.294 | iter time: 359.49 ms (step) remaining time: 0:31:32
|
| 184 |
+
Epoch 1 | iter 2272 step 71 | loss train: 1.421, val: 1.294 | iter time: 360.85 ms (step) remaining time: 0:31:20
|
| 185 |
+
Epoch 1 | iter 2304 step 72 | loss train: 1.413, val: 1.294 | iter time: 360.18 ms (step) remaining time: 0:31:08
|
| 186 |
+
Epoch 1 | iter 2336 step 73 | loss train: 1.459, val: 1.294 | iter time: 360.26 ms (step) remaining time: 0:30:56
|
| 187 |
+
Epoch 1 | iter 2368 step 74 | loss train: 1.406, val: 1.294 | iter time: 360.15 ms (step) remaining time: 0:30:44
|
| 188 |
+
Epoch 1 | iter 2400 step 75 | loss train: 1.379, val: 1.294 | iter time: 358.32 ms (step) remaining time: 0:30:32
|
| 189 |
+
Epoch 1 | iter 2432 step 76 | loss train: 1.362, val: 1.294 | iter time: 359.50 ms (step) remaining time: 0:30:21
|
| 190 |
+
Epoch 1 | iter 2464 step 77 | loss train: 1.351, val: 1.294 | iter time: 359.54 ms (step) remaining time: 0:30:09
|
| 191 |
+
Epoch 1 | iter 2496 step 78 | loss train: 1.492, val: 1.294 | iter time: 359.40 ms (step) remaining time: 0:29:57
|
| 192 |
+
Epoch 1 | iter 2528 step 79 | loss train: 1.343, val: 1.294 | iter time: 359.06 ms (step) remaining time: 0:29:45
|
| 193 |
+
Epoch 1 | iter 2560 step 80 | loss train: 1.406, val: 1.294 | iter time: 358.65 ms (step) remaining time: 0:29:33
|
| 194 |
+
Epoch 1 | iter 2592 step 81 | loss train: 1.453, val: 1.294 | iter time: 358.60 ms (step) remaining time: 0:29:22
|
| 195 |
+
Epoch 1 | iter 2624 step 82 | loss train: 1.470, val: 1.294 | iter time: 358.12 ms (step) remaining time: 0:29:10
|
| 196 |
+
Epoch 1 | iter 2656 step 83 | loss train: 1.431, val: 1.294 | iter time: 358.82 ms (step) remaining time: 0:28:58
|
| 197 |
+
Epoch 1 | iter 2688 step 84 | loss train: 1.432, val: 1.294 | iter time: 359.61 ms (step) remaining time: 0:28:47
|
| 198 |
+
Epoch 1 | iter 2720 step 85 | loss train: 1.388, val: 1.294 | iter time: 361.41 ms (step) remaining time: 0:28:35
|
| 199 |
+
Epoch 1 | iter 2752 step 86 | loss train: 1.344, val: 1.294 | iter time: 360.90 ms (step) remaining time: 0:28:23
|
| 200 |
+
Epoch 1 | iter 2784 step 87 | loss train: 1.425, val: 1.294 | iter time: 359.16 ms (step) remaining time: 0:28:12
|
| 201 |
+
Epoch 1 | iter 2816 step 88 | loss train: 1.422, val: 1.294 | iter time: 359.69 ms (step) remaining time: 0:28:00
|
| 202 |
+
Epoch 1 | iter 2848 step 89 | loss train: 1.388, val: 1.294 | iter time: 359.55 ms (step) remaining time: 0:27:48
|
| 203 |
+
Epoch 1 | iter 2880 step 90 | loss train: 1.452, val: 1.294 | iter time: 359.84 ms (step) remaining time: 0:27:37
|
| 204 |
+
Epoch 1 | iter 2912 step 91 | loss train: 1.422, val: 1.294 | iter time: 360.12 ms (step) remaining time: 0:27:25
|
| 205 |
+
Epoch 1 | iter 2944 step 92 | loss train: 1.318, val: 1.294 | iter time: 360.34 ms (step) remaining time: 0:27:14
|
| 206 |
+
Epoch 1 | iter 2976 step 93 | loss train: 1.452, val: 1.294 | iter time: 359.09 ms (step) remaining time: 0:27:02
|
| 207 |
+
Epoch 1 | iter 3008 step 94 | loss train: 1.451, val: 1.294 | iter time: 359.24 ms (step) remaining time: 0:26:51
|
| 208 |
+
Epoch 1 | iter 3040 step 95 | loss train: 1.450, val: 1.294 | iter time: 361.68 ms (step) remaining time: 0:26:39
|
| 209 |
+
Epoch 1 | iter 3072 step 96 | loss train: 1.411, val: 1.294 | iter time: 358.21 ms (step) remaining time: 0:26:28
|
| 210 |
+
Epoch 1 | iter 3104 step 97 | loss train: 1.410, val: 1.294 | iter time: 359.25 ms (step) remaining time: 0:26:16
|
| 211 |
+
Epoch 1 | iter 3136 step 98 | loss train: 1.428, val: 1.294 | iter time: 357.92 ms (step) remaining time: 0:26:05
|
| 212 |
+
Epoch 1 | iter 3168 step 99 | loss train: 1.397, val: 1.294 | iter time: 362.02 ms (step) remaining time: 0:25:53
|
| 213 |
+
Epoch 1 | iter 3200 step 100 | loss train: 1.359, val: 1.294 | iter time: 359.59 ms (step) remaining time: 0:25:42
|
| 214 |
+
Validating ...
|
| 215 |
+
iter 3200: val loss 1.2439, val time: 22378.01 ms
|
| 216 |
+
Saving checkpoint to 'out/pretrain/2411_full/step-00000100/lit_model.pth'
|
| 217 |
+
Epoch 1 | iter 3232 step 101 | loss train: 1.438, val: 1.244 | iter time: 358.59 ms (step) remaining time: 0:26:24
|
| 218 |
+
Epoch 1 | iter 3264 step 102 | loss train: 1.414, val: 1.244 | iter time: 359.90 ms (step) remaining time: 0:26:12
|
| 219 |
+
Epoch 1 | iter 3296 step 103 | loss train: 1.393, val: 1.244 | iter time: 358.44 ms (step) remaining time: 0:25:59
|
| 220 |
+
Epoch 1 | iter 3328 step 104 | loss train: 1.375, val: 1.244 | iter time: 360.59 ms (step) remaining time: 0:25:47
|
| 221 |
+
Epoch 1 | iter 3360 step 105 | loss train: 1.360, val: 1.244 | iter time: 359.67 ms (step) remaining time: 0:25:35
|
| 222 |
+
Epoch 1 | iter 3392 step 106 | loss train: 1.415, val: 1.244 | iter time: 358.96 ms (step) remaining time: 0:25:23
|
| 223 |
+
Epoch 1 | iter 3424 step 107 | loss train: 1.429, val: 1.244 | iter time: 359.56 ms (step) remaining time: 0:25:11
|
| 224 |
+
Epoch 1 | iter 3456 step 108 | loss train: 1.394, val: 1.244 | iter time: 358.84 ms (step) remaining time: 0:24:58
|
| 225 |
+
Epoch 1 | iter 3488 step 109 | loss train: 1.455, val: 1.244 | iter time: 359.08 ms (step) remaining time: 0:24:46
|
| 226 |
+
Epoch 1 | iter 3520 step 110 | loss train: 1.388, val: 1.244 | iter time: 359.00 ms (step) remaining time: 0:24:34
|
| 227 |
+
Epoch 1 | iter 3552 step 111 | loss train: 1.333, val: 1.244 | iter time: 358.53 ms (step) remaining time: 0:24:22
|
| 228 |
+
Epoch 1 | iter 3584 step 112 | loss train: 1.359, val: 1.244 | iter time: 360.46 ms (step) remaining time: 0:24:10
|
| 229 |
+
Epoch 1 | iter 3616 step 113 | loss train: 1.384, val: 1.244 | iter time: 359.24 ms (step) remaining time: 0:23:58
|
| 230 |
+
Epoch 1 | iter 3648 step 114 | loss train: 1.412, val: 1.244 | iter time: 360.66 ms (step) remaining time: 0:23:46
|
| 231 |
+
Epoch 1 | iter 3680 step 115 | loss train: 1.415, val: 1.244 | iter time: 361.23 ms (step) remaining time: 0:23:34
|
| 232 |
+
Epoch 1 | iter 3712 step 116 | loss train: 1.406, val: 1.244 | iter time: 359.68 ms (step) remaining time: 0:23:22
|
| 233 |
+
Epoch 1 | iter 3744 step 117 | loss train: 1.377, val: 1.244 | iter time: 361.09 ms (step) remaining time: 0:23:10
|
| 234 |
+
Epoch 1 | iter 3776 step 118 | loss train: 1.355, val: 1.244 | iter time: 359.84 ms (step) remaining time: 0:22:59
|
| 235 |
+
Epoch 1 | iter 3808 step 119 | loss train: 1.388, val: 1.244 | iter time: 358.49 ms (step) remaining time: 0:22:47
|
| 236 |
+
Epoch 1 | iter 3840 step 120 | loss train: 1.387, val: 1.244 | iter time: 359.27 ms (step) remaining time: 0:22:35
|
| 237 |
+
Epoch 1 | iter 3872 step 121 | loss train: 1.385, val: 1.244 | iter time: 359.35 ms (step) remaining time: 0:22:23
|
| 238 |
+
Epoch 1 | iter 3904 step 122 | loss train: 1.405, val: 1.244 | iter time: 360.18 ms (step) remaining time: 0:22:11
|
| 239 |
+
Epoch 1 | iter 3936 step 123 | loss train: 1.448, val: 1.244 | iter time: 358.64 ms (step) remaining time: 0:21:59
|
| 240 |
+
Epoch 1 | iter 3968 step 124 | loss train: 1.338, val: 1.244 | iter time: 360.89 ms (step) remaining time: 0:21:47
|
| 241 |
+
Epoch 1 | iter 4000 step 125 | loss train: 1.384, val: 1.244 | iter time: 358.47 ms (step) remaining time: 0:21:35
|
| 242 |
+
Epoch 1 | iter 4032 step 126 | loss train: 1.360, val: 1.244 | iter time: 358.89 ms (step) remaining time: 0:21:23
|
| 243 |
+
Epoch 1 | iter 4064 step 127 | loss train: 1.383, val: 1.244 | iter time: 361.35 ms (step) remaining time: 0:21:12
|
| 244 |
+
Epoch 1 | iter 4096 step 128 | loss train: 1.347, val: 1.244 | iter time: 361.50 ms (step) remaining time: 0:21:00
|
| 245 |
+
Epoch 1 | iter 4128 step 129 | loss train: 1.382, val: 1.244 | iter time: 358.77 ms (step) remaining time: 0:20:48
|
| 246 |
+
Epoch 1 | iter 4160 step 130 | loss train: 1.412, val: 1.244 | iter time: 358.04 ms (step) remaining time: 0:20:36
|
| 247 |
+
Epoch 1 | iter 4192 step 131 | loss train: 1.340, val: 1.244 | iter time: 359.86 ms (step) remaining time: 0:20:25
|
| 248 |
+
Epoch 1 | iter 4224 step 132 | loss train: 1.360, val: 1.244 | iter time: 358.69 ms (step) remaining time: 0:20:13
|
| 249 |
+
Epoch 1 | iter 4256 step 133 | loss train: 1.337, val: 1.244 | iter time: 359.84 ms (step) remaining time: 0:20:01
|
| 250 |
+
Epoch 1 | iter 4288 step 134 | loss train: 1.427, val: 1.244 | iter time: 357.63 ms (step) remaining time: 0:19:49
|
| 251 |
+
Epoch 1 | iter 4320 step 135 | loss train: 1.287, val: 1.244 | iter time: 358.42 ms (step) remaining time: 0:19:38
|
| 252 |
+
Epoch 1 | iter 4352 step 136 | loss train: 1.322, val: 1.244 | iter time: 360.74 ms (step) remaining time: 0:19:26
|
| 253 |
+
Epoch 1 | iter 4384 step 137 | loss train: 1.330, val: 1.244 | iter time: 360.09 ms (step) remaining time: 0:19:14
|
| 254 |
+
Epoch 1 | iter 4416 step 138 | loss train: 1.333, val: 1.244 | iter time: 359.74 ms (step) remaining time: 0:19:03
|
| 255 |
+
Epoch 1 | iter 4448 step 139 | loss train: 1.379, val: 1.244 | iter time: 358.36 ms (step) remaining time: 0:18:51
|
| 256 |
+
Epoch 1 | iter 4480 step 140 | loss train: 1.417, val: 1.244 | iter time: 360.26 ms (step) remaining time: 0:18:39
|
| 257 |
+
Epoch 1 | iter 4512 step 141 | loss train: 1.342, val: 1.244 | iter time: 359.52 ms (step) remaining time: 0:18:28
|
| 258 |
+
Epoch 1 | iter 4544 step 142 | loss train: 1.389, val: 1.244 | iter time: 360.43 ms (step) remaining time: 0:18:16
|
| 259 |
+
Epoch 1 | iter 4576 step 143 | loss train: 1.347, val: 1.244 | iter time: 360.40 ms (step) remaining time: 0:18:04
|
| 260 |
+
Epoch 1 | iter 4608 step 144 | loss train: 1.347, val: 1.244 | iter time: 360.14 ms (step) remaining time: 0:17:53
|
| 261 |
+
Epoch 1 | iter 4640 step 145 | loss train: 1.392, val: 1.244 | iter time: 358.88 ms (step) remaining time: 0:17:41
|
| 262 |
+
Epoch 1 | iter 4672 step 146 | loss train: 1.371, val: 1.244 | iter time: 361.56 ms (step) remaining time: 0:17:30
|
| 263 |
+
Epoch 1 | iter 4704 step 147 | loss train: 1.404, val: 1.244 | iter time: 360.01 ms (step) remaining time: 0:17:18
|
| 264 |
+
Epoch 1 | iter 4736 step 148 | loss train: 1.366, val: 1.244 | iter time: 358.92 ms (step) remaining time: 0:17:07
|
| 265 |
+
Epoch 1 | iter 4768 step 149 | loss train: 1.365, val: 1.244 | iter time: 359.66 ms (step) remaining time: 0:16:55
|
| 266 |
+
Epoch 1 | iter 4800 step 150 | loss train: 1.466, val: 1.244 | iter time: 358.44 ms (step) remaining time: 0:16:44
|
| 267 |
+
Validating ...
|
| 268 |
+
iter 4800: val loss 1.1958, val time: 22369.24 ms
|
| 269 |
+
Epoch 1 | iter 4832 step 151 | loss train: 1.345, val: 1.196 | iter time: 359.84 ms (step) remaining time: 0:16:45
|
| 270 |
+
Epoch 1 | iter 4864 step 152 | loss train: 1.358, val: 1.196 | iter time: 359.76 ms (step) remaining time: 0:16:33
|
| 271 |
+
Epoch 1 | iter 4896 step 153 | loss train: 1.304, val: 1.196 | iter time: 358.12 ms (step) remaining time: 0:16:22
|
| 272 |
+
Epoch 1 | iter 4928 step 154 | loss train: 1.348, val: 1.196 | iter time: 358.95 ms (step) remaining time: 0:16:10
|
| 273 |
+
Epoch 1 | iter 4960 step 155 | loss train: 1.382, val: 1.196 | iter time: 359.76 ms (step) remaining time: 0:15:58
|
| 274 |
+
Epoch 1 | iter 4992 step 156 | loss train: 1.290, val: 1.196 | iter time: 360.15 ms (step) remaining time: 0:15:46
|
| 275 |
+
Epoch 1 | iter 5024 step 157 | loss train: 1.422, val: 1.196 | iter time: 358.70 ms (step) remaining time: 0:15:35
|
| 276 |
+
Epoch 1 | iter 5056 step 158 | loss train: 1.391, val: 1.196 | iter time: 359.50 ms (step) remaining time: 0:15:23
|
| 277 |
+
Epoch 1 | iter 5088 step 159 | loss train: 1.343, val: 1.196 | iter time: 358.65 ms (step) remaining time: 0:15:11
|
| 278 |
+
Epoch 1 | iter 5120 step 160 | loss train: 1.419, val: 1.196 | iter time: 360.49 ms (step) remaining time: 0:15:00
|
| 279 |
+
Epoch 1 | iter 5152 step 161 | loss train: 1.394, val: 1.196 | iter time: 359.69 ms (step) remaining time: 0:14:48
|
| 280 |
+
Epoch 1 | iter 5184 step 162 | loss train: 1.358, val: 1.196 | iter time: 360.02 ms (step) remaining time: 0:14:36
|
| 281 |
+
Epoch 1 | iter 5216 step 163 | loss train: 1.382, val: 1.196 | iter time: 360.12 ms (step) remaining time: 0:14:25
|
| 282 |
+
Epoch 1 | iter 5248 step 164 | loss train: 1.395, val: 1.196 | iter time: 358.85 ms (step) remaining time: 0:14:13
|
| 283 |
+
Epoch 1 | iter 5280 step 165 | loss train: 1.364, val: 1.196 | iter time: 359.12 ms (step) remaining time: 0:14:01
|
| 284 |
+
Epoch 1 | iter 5312 step 166 | loss train: 1.379, val: 1.196 | iter time: 360.38 ms (step) remaining time: 0:13:50
|
| 285 |
+
Epoch 1 | iter 5344 step 167 | loss train: 1.358, val: 1.196 | iter time: 358.86 ms (step) remaining time: 0:13:38
|
| 286 |
+
Epoch 1 | iter 5376 step 168 | loss train: 1.351, val: 1.196 | iter time: 360.70 ms (step) remaining time: 0:13:27
|
| 287 |
+
Epoch 1 | iter 5408 step 169 | loss train: 1.358, val: 1.196 | iter time: 359.88 ms (step) remaining time: 0:13:15
|
| 288 |
+
Epoch 1 | iter 5440 step 170 | loss train: 1.396, val: 1.196 | iter time: 358.87 ms (step) remaining time: 0:13:04
|
| 289 |
+
Epoch 1 | iter 5472 step 171 | loss train: 1.395, val: 1.196 | iter time: 358.15 ms (step) remaining time: 0:12:52
|
| 290 |
+
Epoch 1 | iter 5504 step 172 | loss train: 1.385, val: 1.196 | iter time: 358.52 ms (step) remaining time: 0:12:41
|
| 291 |
+
Epoch 1 | iter 5536 step 173 | loss train: 1.366, val: 1.196 | iter time: 358.65 ms (step) remaining time: 0:12:29
|
| 292 |
+
Epoch 1 | iter 5568 step 174 | loss train: 1.345, val: 1.196 | iter time: 359.53 ms (step) remaining time: 0:12:17
|
| 293 |
+
Epoch 1 | iter 5600 step 175 | loss train: 1.289, val: 1.196 | iter time: 358.53 ms (step) remaining time: 0:12:06
|
| 294 |
+
Epoch 1 | iter 5632 step 176 | loss train: 1.343, val: 1.196 | iter time: 359.26 ms (step) remaining time: 0:11:54
|
| 295 |
+
Epoch 1 | iter 5664 step 177 | loss train: 1.352, val: 1.196 | iter time: 361.21 ms (step) remaining time: 0:11:43
|
| 296 |
+
Epoch 1 | iter 5696 step 178 | loss train: 1.371, val: 1.196 | iter time: 358.45 ms (step) remaining time: 0:11:31
|
| 297 |
+
Epoch 1 | iter 5728 step 179 | loss train: 1.365, val: 1.196 | iter time: 579.10 ms (step) remaining time: 0:11:20
|
| 298 |
+
Epoch 1 | iter 5760 step 180 | loss train: 1.342, val: 1.196 | iter time: 358.31 ms (step) remaining time: 0:11:08
|
| 299 |
+
Epoch 1 | iter 5792 step 181 | loss train: 1.407, val: 1.196 | iter time: 358.42 ms (step) remaining time: 0:10:57
|
| 300 |
+
Epoch 1 | iter 5824 step 182 | loss train: 1.420, val: 1.196 | iter time: 359.56 ms (step) remaining time: 0:10:45
|
| 301 |
+
Epoch 1 | iter 5856 step 183 | loss train: 1.362, val: 1.196 | iter time: 359.50 ms (step) remaining time: 0:10:34
|
| 302 |
+
Epoch 1 | iter 5888 step 184 | loss train: 1.406, val: 1.196 | iter time: 359.14 ms (step) remaining time: 0:10:23
|
| 303 |
+
Epoch 1 | iter 5920 step 185 | loss train: 1.303, val: 1.196 | iter time: 359.29 ms (step) remaining time: 0:10:11
|
| 304 |
+
Epoch 1 | iter 5952 step 186 | loss train: 1.381, val: 1.196 | iter time: 360.81 ms (step) remaining time: 0:10:00
|
| 305 |
+
Epoch 1 | iter 5984 step 187 | loss train: 1.330, val: 1.196 | iter time: 358.61 ms (step) remaining time: 0:09:48
|
| 306 |
+
Epoch 1 | iter 6016 step 188 | loss train: 1.376, val: 1.196 | iter time: 359.75 ms (step) remaining time: 0:09:37
|
| 307 |
+
Epoch 1 | iter 6048 step 189 | loss train: 1.328, val: 1.196 | iter time: 359.61 ms (step) remaining time: 0:09:25
|
| 308 |
+
Epoch 1 | iter 6080 step 190 | loss train: 1.341, val: 1.196 | iter time: 359.27 ms (step) remaining time: 0:09:14
|
| 309 |
+
Epoch 1 | iter 6112 step 191 | loss train: 1.291, val: 1.196 | iter time: 360.10 ms (step) remaining time: 0:09:02
|
| 310 |
+
Epoch 1 | iter 6144 step 192 | loss train: 1.360, val: 1.196 | iter time: 357.85 ms (step) remaining time: 0:08:51
|
| 311 |
+
Epoch 1 | iter 6176 step 193 | loss train: 1.342, val: 1.196 | iter time: 358.16 ms (step) remaining time: 0:08:40
|
| 312 |
+
Epoch 1 | iter 6208 step 194 | loss train: 1.360, val: 1.196 | iter time: 357.86 ms (step) remaining time: 0:08:28
|
| 313 |
+
Epoch 1 | iter 6240 step 195 | loss train: 1.348, val: 1.196 | iter time: 359.78 ms (step) remaining time: 0:08:17
|
| 314 |
+
Epoch 1 | iter 6272 step 196 | loss train: 1.321, val: 1.196 | iter time: 358.91 ms (step) remaining time: 0:08:05
|
| 315 |
+
Epoch 1 | iter 6304 step 197 | loss train: 1.359, val: 1.196 | iter time: 359.99 ms (step) remaining time: 0:07:54
|
| 316 |
+
Epoch 1 | iter 6336 step 198 | loss train: 1.310, val: 1.196 | iter time: 359.90 ms (step) remaining time: 0:07:43
|
| 317 |
+
Epoch 1 | iter 6368 step 199 | loss train: 1.284, val: 1.196 | iter time: 359.14 ms (step) remaining time: 0:07:31
|
| 318 |
+
Epoch 1 | iter 6400 step 200 | loss train: 1.357, val: 1.196 | iter time: 360.49 ms (step) remaining time: 0:07:20
|
| 319 |
+
Validating ...
|
| 320 |
+
iter 6400: val loss 1.1577, val time: 22364.37 ms
|
| 321 |
+
Saving checkpoint to 'out/pretrain/2411_full/step-00000200/lit_model.pth'
|
| 322 |
+
Epoch 1 | iter 6432 step 201 | loss train: 1.301, val: 1.158 | iter time: 355.83 ms (step) remaining time: 0:07:16
|
| 323 |
+
Epoch 1 | iter 6464 step 202 | loss train: 1.383, val: 1.158 | iter time: 355.74 ms (step) remaining time: 0:07:04
|
| 324 |
+
Epoch 1 | iter 6496 step 203 | loss train: 1.345, val: 1.158 | iter time: 359.40 ms (step) remaining time: 0:06:53
|
| 325 |
+
Epoch 1 | iter 6528 step 204 | loss train: 1.374, val: 1.158 | iter time: 360.00 ms (step) remaining time: 0:06:41
|
| 326 |
+
Epoch 1 | iter 6560 step 205 | loss train: 1.372, val: 1.158 | iter time: 357.77 ms (step) remaining time: 0:06:29
|
| 327 |
+
Epoch 1 | iter 6592 step 206 | loss train: 1.299, val: 1.158 | iter time: 360.10 ms (step) remaining time: 0:06:18
|
| 328 |
+
Epoch 1 | iter 6624 step 207 | loss train: 1.378, val: 1.158 | iter time: 357.40 ms (step) remaining time: 0:06:06
|
| 329 |
+
Epoch 1 | iter 6656 step 208 | loss train: 1.348, val: 1.158 | iter time: 359.02 ms (step) remaining time: 0:05:55
|
| 330 |
+
Epoch 1 | iter 6688 step 209 | loss train: 1.377, val: 1.158 | iter time: 358.85 ms (step) remaining time: 0:05:43
|
| 331 |
+
Epoch 1 | iter 6720 step 210 | loss train: 1.288, val: 1.158 | iter time: 359.82 ms (step) remaining time: 0:05:32
|
| 332 |
+
Epoch 1 | iter 6752 step 211 | loss train: 1.302, val: 1.158 | iter time: 357.84 ms (step) remaining time: 0:05:20
|
| 333 |
+
Epoch 1 | iter 6784 step 212 | loss train: 1.287, val: 1.158 | iter time: 359.81 ms (step) remaining time: 0:05:09
|
| 334 |
+
Epoch 1 | iter 6816 step 213 | loss train: 1.339, val: 1.158 | iter time: 360.20 ms (step) remaining time: 0:04:57
|
| 335 |
+
Epoch 1 | iter 6848 step 214 | loss train: 1.361, val: 1.158 | iter time: 361.63 ms (step) remaining time: 0:04:46
|
| 336 |
+
Epoch 1 | iter 6880 step 215 | loss train: 1.334, val: 1.158 | iter time: 360.10 ms (step) remaining time: 0:04:34
|
| 337 |
+
Epoch 1 | iter 6912 step 216 | loss train: 1.294, val: 1.158 | iter time: 359.26 ms (step) remaining time: 0:04:23
|
| 338 |
+
Epoch 1 | iter 6944 step 217 | loss train: 1.272, val: 1.158 | iter time: 359.47 ms (step) remaining time: 0:04:11
|
| 339 |
+
Epoch 1 | iter 6976 step 218 | loss train: 1.331, val: 1.158 | iter time: 359.73 ms (step) remaining time: 0:04:00
|
| 340 |
+
Epoch 1 | iter 7008 step 219 | loss train: 1.395, val: 1.158 | iter time: 359.78 ms (step) remaining time: 0:03:48
|
| 341 |
+
Epoch 1 | iter 7040 step 220 | loss train: 1.306, val: 1.158 | iter time: 360.37 ms (step) remaining time: 0:03:37
|
| 342 |
+
Epoch 1 | iter 7072 step 221 | loss train: 1.274, val: 1.158 | iter time: 358.43 ms (step) remaining time: 0:03:25
|
| 343 |
+
Epoch 1 | iter 7104 step 222 | loss train: 1.315, val: 1.158 | iter time: 359.68 ms (step) remaining time: 0:03:14
|
| 344 |
+
Epoch 1 | iter 7136 step 223 | loss train: 1.319, val: 1.158 | iter time: 358.74 ms (step) remaining time: 0:03:02
|
| 345 |
+
Epoch 1 | iter 7168 step 224 | loss train: 1.274, val: 1.158 | iter time: 358.32 ms (step) remaining time: 0:02:51
|
| 346 |
+
Epoch 1 | iter 7200 step 225 | loss train: 1.382, val: 1.158 | iter time: 360.61 ms (step) remaining time: 0:02:39
|
| 347 |
+
Epoch 1 | iter 7232 step 226 | loss train: 1.332, val: 1.158 | iter time: 360.36 ms (step) remaining time: 0:02:28
|
| 348 |
+
Epoch 1 | iter 7264 step 227 | loss train: 1.357, val: 1.158 | iter time: 360.40 ms (step) remaining time: 0:02:16
|
| 349 |
+
Epoch 1 | iter 7296 step 228 | loss train: 1.331, val: 1.158 | iter time: 358.67 ms (step) remaining time: 0:02:05
|
| 350 |
+
Epoch 1 | iter 7328 step 229 | loss train: 1.388, val: 1.158 | iter time: 358.12 ms (step) remaining time: 0:01:54
|
| 351 |
+
Epoch 1 | iter 7360 step 230 | loss train: 1.323, val: 1.158 | iter time: 359.22 ms (step) remaining time: 0:01:42
|
| 352 |
+
Epoch 1 | iter 7392 step 231 | loss train: 1.350, val: 1.158 | iter time: 360.58 ms (step) remaining time: 0:01:31
|
| 353 |
+
Epoch 1 | iter 7424 step 232 | loss train: 1.281, val: 1.158 | iter time: 360.66 ms (step) remaining time: 0:01:19
|
| 354 |
+
Epoch 1 | iter 7456 step 233 | loss train: 1.389, val: 1.158 | iter time: 358.50 ms (step) remaining time: 0:01:08
|
| 355 |
+
Epoch 1 | iter 7488 step 234 | loss train: 1.345, val: 1.158 | iter time: 359.13 ms (step) remaining time: 0:00:56
|
| 356 |
+
Epoch 1 | iter 7520 step 235 | loss train: 1.326, val: 1.158 | iter time: 357.69 ms (step) remaining time: 0:00:45
|
| 357 |
+
Epoch 1 | iter 7552 step 236 | loss train: 1.332, val: 1.158 | iter time: 359.30 ms (step) remaining time: 0:00:34
|
| 358 |
+
Epoch 1 | iter 7584 step 237 | loss train: 1.343, val: 1.158 | iter time: 359.36 ms (step) remaining time: 0:00:22
|
| 359 |
+
Epoch 1 | iter 7616 step 238 | loss train: 1.297, val: 1.158 | iter time: 359.37 ms (step) remaining time: 0:00:11
|
| 360 |
+
Epoch 2 | iter 7648 step 239 | loss train: 1.201, val: 1.158 | iter time: 359.80 ms (step) remaining time: 0:00:00
|
| 361 |
+
Validating ...
|
| 362 |
+
Final evaluation | val loss: 1.132 | val ppl: 3.101
|
| 363 |
+
Saving checkpoint to 'out/pretrain/2411_full/final/lit_model.pth'
|
| 364 |
+
----------------------------------------
|
| 365 |
+
| Performance
|
| 366 |
+
| - Total tokens : 250,609,664
|
| 367 |
+
| - Training Time : 2784.71 s
|
| 368 |
+
| - Tok/sec : 135.07 tok/s
|
| 369 |
+
| ----------------------------------------
|
| 370 |
+
| Memory Usage
|
| 371 |
+
| - Memory Used : 26.32 GB
|
| 372 |
+
----------------------------------------
|
out/pretrain/tinyllama/teelogs/2411_lr4e-5.txt
ADDED
|
@@ -0,0 +1,371 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
Initializing distributed: GLOBAL_RANK: 0, MEMBER: 1/4
|
| 2 |
+
Initializing distributed: GLOBAL_RANK: 1, MEMBER: 2/4
|
| 3 |
+
[rank: 1] Seed set to 42
|
| 4 |
+
Initializing distributed: GLOBAL_RANK: 2, MEMBER: 3/4
|
| 5 |
+
[rank: 2] Seed set to 42
|
| 6 |
+
Initializing distributed: GLOBAL_RANK: 3, MEMBER: 4/4
|
| 7 |
+
[rank: 3] Seed set to 42
|
| 8 |
+
----------------------------------------------------------------------------------------------------
|
| 9 |
+
distributed_backend=nccl
|
| 10 |
+
All distributed processes registered. Starting with 4 processes
|
| 11 |
+
----------------------------------------------------------------------------------------------------
|
| 12 |
+
|
| 13 |
+
[rank: 0] Seed set to 42
|
| 14 |
+
/mnt/data/litgpt/litgpt/pretrain.py:515: UserWarning: A newer version of litdata is available (0.2.52). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 15 |
+
train_dataloader = data.train_dataloader()
|
| 16 |
+
/mnt/data/litgpt/litgpt/pretrain.py:515: UserWarning: A newer version of litdata is available (0.2.52). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 17 |
+
train_dataloader = data.train_dataloader()
|
| 18 |
+
/mnt/data/litgpt/litgpt/pretrain.py:515: UserWarning: A newer version of litdata is available (0.2.52). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 19 |
+
train_dataloader = data.train_dataloader()
|
| 20 |
+
/mnt/data/litgpt/litgpt/pretrain.py:515: UserWarning: A newer version of litdata is available (0.2.52). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 21 |
+
train_dataloader = data.train_dataloader()
|
| 22 |
+
All GPUs are fully connected via NVLink.
|
| 23 |
+
{'data': {'batch_size': 1,
|
| 24 |
+
'data_path': PosixPath('litgpt/data/arxiv'),
|
| 25 |
+
'num_workers': 0,
|
| 26 |
+
'seed': 42,
|
| 27 |
+
'seq_length': 2048,
|
| 28 |
+
'use_starcoder': True},
|
| 29 |
+
'data_dir': PosixPath('litgpt/data/arxiv/2411'),
|
| 30 |
+
'devices': 'auto',
|
| 31 |
+
'eval': {'evaluate_example': 'first',
|
| 32 |
+
'final_validation': True,
|
| 33 |
+
'initial_validation': True,
|
| 34 |
+
'interval': 50,
|
| 35 |
+
'max_iters': 200,
|
| 36 |
+
'max_new_tokens': None},
|
| 37 |
+
'initial_checkpoint_dir': PosixPath('out/pretrain/tinyllama/2410_full/final'),
|
| 38 |
+
'log': {'group': None, 'project': None, 'run': None},
|
| 39 |
+
'logger_name': 'tensorboard',
|
| 40 |
+
'model_config': {'attention_logit_softcapping': None,
|
| 41 |
+
'attention_scores_scalar': None,
|
| 42 |
+
'attn_bias': False,
|
| 43 |
+
'bias': False,
|
| 44 |
+
'block_size': 2048,
|
| 45 |
+
'final_logit_softcapping': None,
|
| 46 |
+
'gelu_approximate': 'none',
|
| 47 |
+
'head_size': 64,
|
| 48 |
+
'hf_config': {'name': 'TinyLlama-1.1B-intermediate-step-1431k-3T',
|
| 49 |
+
'org': 'TinyLlama'},
|
| 50 |
+
'intermediate_size': 5632,
|
| 51 |
+
'lm_head_bias': False,
|
| 52 |
+
'mlp_class_name': 'LLaMAMLP',
|
| 53 |
+
'moe_intermediate_size': None,
|
| 54 |
+
'n_embd': 2048,
|
| 55 |
+
'n_expert': 0,
|
| 56 |
+
'n_expert_per_token': 0,
|
| 57 |
+
'n_head': 32,
|
| 58 |
+
'n_layer': 22,
|
| 59 |
+
'n_query_groups': 4,
|
| 60 |
+
'name': 'tiny-llama-1.1b',
|
| 61 |
+
'norm_1': True,
|
| 62 |
+
'norm_2': True,
|
| 63 |
+
'norm_class_name': 'RMSNorm',
|
| 64 |
+
'norm_eps': 1e-05,
|
| 65 |
+
'norm_qk': False,
|
| 66 |
+
'norm_qk_type': 'default',
|
| 67 |
+
'padded_vocab_size': 32000,
|
| 68 |
+
'padding_multiple': 64,
|
| 69 |
+
'parallel_residual': False,
|
| 70 |
+
'post_attention_norm': False,
|
| 71 |
+
'post_mlp_norm': False,
|
| 72 |
+
'rope_adjustments': None,
|
| 73 |
+
'rope_base': 10000,
|
| 74 |
+
'rope_condense_ratio': 1,
|
| 75 |
+
'rope_indices': None,
|
| 76 |
+
'rope_local_base_freq': None,
|
| 77 |
+
'rotary_percentage': 1.0,
|
| 78 |
+
'scale_embeddings': False,
|
| 79 |
+
'shared_attention_norm': False,
|
| 80 |
+
'sliding_window_indices': None,
|
| 81 |
+
'sliding_window_size': None,
|
| 82 |
+
'vocab_size': 32000},
|
| 83 |
+
'model_name': 'tiny-llama-1.1b',
|
| 84 |
+
'num_nodes': 1,
|
| 85 |
+
'optimizer': "{'class_path': 'torch.optim.AdamW', 'init_args': {'lr': 4e-05, "
|
| 86 |
+
"'weight_decay': 0.1, 'betas': [0.9, 0.95]}}",
|
| 87 |
+
'out_dir': PosixPath('out/pretrain/tinyllama/2411_lr4e-5'),
|
| 88 |
+
'ppl': False,
|
| 89 |
+
'precision': 'bf16-mixed',
|
| 90 |
+
'resume': False,
|
| 91 |
+
'seed': 42,
|
| 92 |
+
'tokenizer_dir': PosixPath('checkpoints/TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T'),
|
| 93 |
+
'train': {'epochs': None,
|
| 94 |
+
'global_batch_size': 512,
|
| 95 |
+
'log_interval': 1,
|
| 96 |
+
'lr_warmup_fraction': None,
|
| 97 |
+
'lr_warmup_steps': 20,
|
| 98 |
+
'max_norm': 1.0,
|
| 99 |
+
'max_seq_length': 2048,
|
| 100 |
+
'max_steps': None,
|
| 101 |
+
'max_tokens': 250609664,
|
| 102 |
+
'micro_batch_size': 4,
|
| 103 |
+
'min_lr': 4e-05,
|
| 104 |
+
'save_interval': 100,
|
| 105 |
+
'tie_embeddings': None}}
|
| 106 |
+
Time to instantiate model: 0.04 seconds.
|
| 107 |
+
Total parameters: 1,100,048,384
|
| 108 |
+
[ok] out/pretrain/tinyllama/2410_full/final/lit_model.pth 已是纯权重
|
| 109 |
+
Validating ...
|
| 110 |
+
Measured TFLOPs: 239.66
|
| 111 |
+
Epoch 1 | iter 32 step 1 | loss train: 1.425, val: 1.370 | iter time: 557.26 ms (step) remaining time: 0:46:27
|
| 112 |
+
Epoch 1 | iter 64 step 2 | loss train: 1.357, val: 1.370 | iter time: 358.88 ms (step) remaining time: 0:44:24
|
| 113 |
+
Epoch 1 | iter 96 step 3 | loss train: 1.439, val: 1.370 | iter time: 357.24 ms (step) remaining time: 0:43:38
|
| 114 |
+
Epoch 1 | iter 128 step 4 | loss train: 1.387, val: 1.370 | iter time: 359.44 ms (step) remaining time: 0:43:10
|
| 115 |
+
Epoch 1 | iter 160 step 5 | loss train: 1.403, val: 1.370 | iter time: 358.31 ms (step) remaining time: 0:42:49
|
| 116 |
+
Epoch 1 | iter 192 step 6 | loss train: 1.406, val: 1.370 | iter time: 359.10 ms (step) remaining time: 0:42:32
|
| 117 |
+
Epoch 1 | iter 224 step 7 | loss train: 1.423, val: 1.370 | iter time: 361.44 ms (step) remaining time: 0:42:17
|
| 118 |
+
Epoch 1 | iter 256 step 8 | loss train: 1.407, val: 1.370 | iter time: 358.88 ms (step) remaining time: 0:42:04
|
| 119 |
+
Epoch 1 | iter 288 step 9 | loss train: 1.441, val: 1.370 | iter time: 359.81 ms (step) remaining time: 0:41:50
|
| 120 |
+
Epoch 1 | iter 320 step 10 | loss train: 1.432, val: 1.370 | iter time: 357.69 ms (step) remaining time: 0:41:38
|
| 121 |
+
Epoch 1 | iter 352 step 11 | loss train: 1.422, val: 1.370 | iter time: 359.13 ms (step) remaining time: 0:41:26
|
| 122 |
+
Epoch 1 | iter 384 step 12 | loss train: 1.411, val: 1.370 | iter time: 361.31 ms (step) remaining time: 0:41:14
|
| 123 |
+
Epoch 1 | iter 416 step 13 | loss train: 1.398, val: 1.370 | iter time: 359.58 ms (step) remaining time: 0:41:02
|
| 124 |
+
Epoch 1 | iter 448 step 14 | loss train: 1.334, val: 1.370 | iter time: 358.30 ms (step) remaining time: 0:40:51
|
| 125 |
+
Epoch 1 | iter 480 step 15 | loss train: 1.360, val: 1.370 | iter time: 361.08 ms (step) remaining time: 0:40:40
|
| 126 |
+
Epoch 1 | iter 512 step 16 | loss train: 1.343, val: 1.370 | iter time: 359.32 ms (step) remaining time: 0:40:28
|
| 127 |
+
Epoch 1 | iter 544 step 17 | loss train: 1.461, val: 1.370 | iter time: 359.93 ms (step) remaining time: 0:40:17
|
| 128 |
+
Epoch 1 | iter 576 step 18 | loss train: 1.396, val: 1.370 | iter time: 359.52 ms (step) remaining time: 0:40:06
|
| 129 |
+
Epoch 1 | iter 608 step 19 | loss train: 1.358, val: 1.370 | iter time: 359.94 ms (step) remaining time: 0:39:55
|
| 130 |
+
Epoch 1 | iter 640 step 20 | loss train: 1.376, val: 1.370 | iter time: 360.33 ms (step) remaining time: 0:39:44
|
| 131 |
+
Epoch 1 | iter 672 step 21 | loss train: 1.387, val: 1.370 | iter time: 358.21 ms (step) remaining time: 0:39:32
|
| 132 |
+
Epoch 1 | iter 704 step 22 | loss train: 1.390, val: 1.370 | iter time: 358.39 ms (step) remaining time: 0:39:21
|
| 133 |
+
Epoch 1 | iter 736 step 23 | loss train: 1.424, val: 1.370 | iter time: 359.39 ms (step) remaining time: 0:39:10
|
| 134 |
+
Epoch 1 | iter 768 step 24 | loss train: 1.454, val: 1.370 | iter time: 360.41 ms (step) remaining time: 0:38:59
|
| 135 |
+
Epoch 1 | iter 800 step 25 | loss train: 1.416, val: 1.370 | iter time: 361.36 ms (step) remaining time: 0:38:48
|
| 136 |
+
Epoch 1 | iter 832 step 26 | loss train: 1.419, val: 1.370 | iter time: 359.95 ms (step) remaining time: 0:38:37
|
| 137 |
+
Epoch 1 | iter 864 step 27 | loss train: 1.362, val: 1.370 | iter time: 359.55 ms (step) remaining time: 0:38:26
|
| 138 |
+
Epoch 1 | iter 896 step 28 | loss train: 1.386, val: 1.370 | iter time: 359.90 ms (step) remaining time: 0:38:15
|
| 139 |
+
Epoch 1 | iter 928 step 29 | loss train: 1.414, val: 1.370 | iter time: 360.66 ms (step) remaining time: 0:38:04
|
| 140 |
+
Epoch 1 | iter 960 step 30 | loss train: 1.452, val: 1.370 | iter time: 358.72 ms (step) remaining time: 0:37:53
|
| 141 |
+
Epoch 1 | iter 992 step 31 | loss train: 1.395, val: 1.370 | iter time: 361.33 ms (step) remaining time: 0:37:42
|
| 142 |
+
Epoch 1 | iter 1024 step 32 | loss train: 1.423, val: 1.370 | iter time: 360.06 ms (step) remaining time: 0:37:31
|
| 143 |
+
Epoch 1 | iter 1056 step 33 | loss train: 1.322, val: 1.370 | iter time: 360.32 ms (step) remaining time: 0:37:20
|
| 144 |
+
Epoch 1 | iter 1088 step 34 | loss train: 1.382, val: 1.370 | iter time: 360.10 ms (step) remaining time: 0:37:09
|
| 145 |
+
Epoch 1 | iter 1120 step 35 | loss train: 1.402, val: 1.370 | iter time: 359.58 ms (step) remaining time: 0:36:58
|
| 146 |
+
Epoch 1 | iter 1152 step 36 | loss train: 1.390, val: 1.370 | iter time: 358.14 ms (step) remaining time: 0:36:47
|
| 147 |
+
Epoch 1 | iter 1184 step 37 | loss train: 1.391, val: 1.370 | iter time: 359.54 ms (step) remaining time: 0:36:37
|
| 148 |
+
Epoch 1 | iter 1216 step 38 | loss train: 1.392, val: 1.370 | iter time: 361.43 ms (step) remaining time: 0:36:26
|
| 149 |
+
Epoch 1 | iter 1248 step 39 | loss train: 1.404, val: 1.370 | iter time: 358.98 ms (step) remaining time: 0:36:15
|
| 150 |
+
Epoch 1 | iter 1280 step 40 | loss train: 1.417, val: 1.370 | iter time: 359.41 ms (step) remaining time: 0:36:04
|
| 151 |
+
Epoch 1 | iter 1312 step 41 | loss train: 1.420, val: 1.370 | iter time: 361.02 ms (step) remaining time: 0:35:53
|
| 152 |
+
Epoch 1 | iter 1344 step 42 | loss train: 1.390, val: 1.370 | iter time: 360.54 ms (step) remaining time: 0:35:42
|
| 153 |
+
Epoch 1 | iter 1376 step 43 | loss train: 1.390, val: 1.370 | iter time: 359.14 ms (step) remaining time: 0:35:31
|
| 154 |
+
Epoch 1 | iter 1408 step 44 | loss train: 1.418, val: 1.370 | iter time: 358.42 ms (step) remaining time: 0:35:20
|
| 155 |
+
Epoch 1 | iter 1440 step 45 | loss train: 1.355, val: 1.370 | iter time: 359.41 ms (step) remaining time: 0:35:09
|
| 156 |
+
Epoch 1 | iter 1472 step 46 | loss train: 1.331, val: 1.370 | iter time: 358.29 ms (step) remaining time: 0:34:59
|
| 157 |
+
Epoch 1 | iter 1504 step 47 | loss train: 1.356, val: 1.370 | iter time: 358.49 ms (step) remaining time: 0:34:48
|
| 158 |
+
Epoch 1 | iter 1536 step 48 | loss train: 1.359, val: 1.370 | iter time: 360.59 ms (step) remaining time: 0:34:37
|
| 159 |
+
Epoch 1 | iter 1568 step 49 | loss train: 1.383, val: 1.370 | iter time: 361.03 ms (step) remaining time: 0:34:27
|
| 160 |
+
Epoch 1 | iter 1600 step 50 | loss train: 1.459, val: 1.370 | iter time: 359.73 ms (step) remaining time: 0:34:16
|
| 161 |
+
Validating ...
|
| 162 |
+
iter 1600: val loss 1.3383, val time: 21900.94 ms
|
| 163 |
+
Epoch 1 | iter 1632 step 51 | loss train: 1.385, val: 1.338 | iter time: 358.90 ms (step) remaining time: 0:35:25
|
| 164 |
+
Epoch 1 | iter 1664 step 52 | loss train: 1.389, val: 1.338 | iter time: 359.75 ms (step) remaining time: 0:35:12
|
| 165 |
+
Epoch 1 | iter 1696 step 53 | loss train: 1.427, val: 1.338 | iter time: 358.72 ms (step) remaining time: 0:35:00
|
| 166 |
+
Epoch 1 | iter 1728 step 54 | loss train: 1.336, val: 1.338 | iter time: 359.96 ms (step) remaining time: 0:34:47
|
| 167 |
+
Epoch 1 | iter 1760 step 55 | loss train: 1.348, val: 1.338 | iter time: 635.02 ms (step) remaining time: 0:34:35
|
| 168 |
+
Epoch 1 | iter 1792 step 56 | loss train: 1.421, val: 1.338 | iter time: 358.31 ms (step) remaining time: 0:34:22
|
| 169 |
+
Epoch 1 | iter 1824 step 57 | loss train: 1.421, val: 1.338 | iter time: 358.69 ms (step) remaining time: 0:34:10
|
| 170 |
+
Epoch 1 | iter 1856 step 58 | loss train: 1.427, val: 1.338 | iter time: 361.00 ms (step) remaining time: 0:33:57
|
| 171 |
+
Epoch 1 | iter 1888 step 59 | loss train: 1.435, val: 1.338 | iter time: 358.88 ms (step) remaining time: 0:33:45
|
| 172 |
+
Epoch 1 | iter 1920 step 60 | loss train: 1.418, val: 1.338 | iter time: 360.23 ms (step) remaining time: 0:33:32
|
| 173 |
+
Epoch 1 | iter 1952 step 61 | loss train: 1.373, val: 1.338 | iter time: 359.76 ms (step) remaining time: 0:33:20
|
| 174 |
+
Epoch 1 | iter 1984 step 62 | loss train: 1.432, val: 1.338 | iter time: 358.65 ms (step) remaining time: 0:33:08
|
| 175 |
+
Epoch 1 | iter 2016 step 63 | loss train: 1.288, val: 1.338 | iter time: 358.24 ms (step) remaining time: 0:32:55
|
| 176 |
+
Epoch 1 | iter 2048 step 64 | loss train: 1.354, val: 1.338 | iter time: 359.77 ms (step) remaining time: 0:32:43
|
| 177 |
+
Epoch 1 | iter 2080 step 65 | loss train: 1.350, val: 1.338 | iter time: 359.76 ms (step) remaining time: 0:32:31
|
| 178 |
+
Epoch 1 | iter 2112 step 66 | loss train: 1.341, val: 1.338 | iter time: 358.35 ms (step) remaining time: 0:32:19
|
| 179 |
+
Epoch 1 | iter 2144 step 67 | loss train: 1.413, val: 1.338 | iter time: 361.61 ms (step) remaining time: 0:32:07
|
| 180 |
+
Epoch 1 | iter 2176 step 68 | loss train: 1.380, val: 1.338 | iter time: 358.20 ms (step) remaining time: 0:31:55
|
| 181 |
+
Epoch 1 | iter 2208 step 69 | loss train: 1.374, val: 1.338 | iter time: 358.46 ms (step) remaining time: 0:31:43
|
| 182 |
+
Epoch 1 | iter 2240 step 70 | loss train: 1.466, val: 1.338 | iter time: 360.78 ms (step) remaining time: 0:31:31
|
| 183 |
+
Epoch 1 | iter 2272 step 71 | loss train: 1.385, val: 1.338 | iter time: 358.25 ms (step) remaining time: 0:31:19
|
| 184 |
+
Epoch 1 | iter 2304 step 72 | loss train: 1.379, val: 1.338 | iter time: 359.42 ms (step) remaining time: 0:31:07
|
| 185 |
+
Epoch 1 | iter 2336 step 73 | loss train: 1.410, val: 1.338 | iter time: 358.83 ms (step) remaining time: 0:30:55
|
| 186 |
+
Epoch 1 | iter 2368 step 74 | loss train: 1.296, val: 1.338 | iter time: 359.04 ms (step) remaining time: 0:30:43
|
| 187 |
+
Epoch 1 | iter 2400 step 75 | loss train: 1.381, val: 1.338 | iter time: 359.99 ms (step) remaining time: 0:30:31
|
| 188 |
+
Epoch 1 | iter 2432 step 76 | loss train: 1.411, val: 1.338 | iter time: 358.24 ms (step) remaining time: 0:30:19
|
| 189 |
+
Epoch 1 | iter 2464 step 77 | loss train: 1.360, val: 1.338 | iter time: 360.91 ms (step) remaining time: 0:30:07
|
| 190 |
+
Epoch 1 | iter 2496 step 78 | loss train: 1.420, val: 1.338 | iter time: 358.21 ms (step) remaining time: 0:29:56
|
| 191 |
+
Epoch 1 | iter 2528 step 79 | loss train: 1.315, val: 1.338 | iter time: 361.07 ms (step) remaining time: 0:29:44
|
| 192 |
+
Epoch 1 | iter 2560 step 80 | loss train: 1.351, val: 1.338 | iter time: 358.36 ms (step) remaining time: 0:29:32
|
| 193 |
+
Epoch 1 | iter 2592 step 81 | loss train: 1.430, val: 1.338 | iter time: 357.41 ms (step) remaining time: 0:29:20
|
| 194 |
+
Epoch 1 | iter 2624 step 82 | loss train: 1.388, val: 1.338 | iter time: 360.66 ms (step) remaining time: 0:29:09
|
| 195 |
+
Epoch 1 | iter 2656 step 83 | loss train: 1.356, val: 1.338 | iter time: 361.16 ms (step) remaining time: 0:28:57
|
| 196 |
+
Epoch 1 | iter 2688 step 84 | loss train: 1.398, val: 1.338 | iter time: 358.34 ms (step) remaining time: 0:28:45
|
| 197 |
+
Epoch 1 | iter 2720 step 85 | loss train: 1.425, val: 1.338 | iter time: 360.00 ms (step) remaining time: 0:28:34
|
| 198 |
+
Epoch 1 | iter 2752 step 86 | loss train: 1.379, val: 1.338 | iter time: 361.11 ms (step) remaining time: 0:28:22
|
| 199 |
+
Epoch 1 | iter 2784 step 87 | loss train: 1.336, val: 1.338 | iter time: 360.53 ms (step) remaining time: 0:28:10
|
| 200 |
+
Epoch 1 | iter 2816 step 88 | loss train: 1.273, val: 1.338 | iter time: 360.65 ms (step) remaining time: 0:27:59
|
| 201 |
+
Epoch 1 | iter 2848 step 89 | loss train: 1.362, val: 1.338 | iter time: 359.41 ms (step) remaining time: 0:27:47
|
| 202 |
+
Epoch 1 | iter 2880 step 90 | loss train: 1.361, val: 1.338 | iter time: 360.27 ms (step) remaining time: 0:27:36
|
| 203 |
+
Epoch 1 | iter 2912 step 91 | loss train: 1.392, val: 1.338 | iter time: 359.95 ms (step) remaining time: 0:27:24
|
| 204 |
+
Epoch 1 | iter 2944 step 92 | loss train: 1.367, val: 1.338 | iter time: 358.76 ms (step) remaining time: 0:27:13
|
| 205 |
+
Epoch 1 | iter 2976 step 93 | loss train: 1.412, val: 1.338 | iter time: 359.34 ms (step) remaining time: 0:27:01
|
| 206 |
+
Epoch 1 | iter 3008 step 94 | loss train: 1.381, val: 1.338 | iter time: 359.27 ms (step) remaining time: 0:26:50
|
| 207 |
+
Epoch 1 | iter 3040 step 95 | loss train: 1.361, val: 1.338 | iter time: 359.58 ms (step) remaining time: 0:26:38
|
| 208 |
+
Epoch 1 | iter 3072 step 96 | loss train: 1.366, val: 1.338 | iter time: 357.97 ms (step) remaining time: 0:26:27
|
| 209 |
+
Epoch 1 | iter 3104 step 97 | loss train: 1.349, val: 1.338 | iter time: 358.62 ms (step) remaining time: 0:26:15
|
| 210 |
+
Epoch 1 | iter 3136 step 98 | loss train: 1.381, val: 1.338 | iter time: 360.10 ms (step) remaining time: 0:26:04
|
| 211 |
+
Epoch 1 | iter 3168 step 99 | loss train: 1.358, val: 1.338 | iter time: 358.33 ms (step) remaining time: 0:25:52
|
| 212 |
+
Epoch 1 | iter 3200 step 100 | loss train: 1.398, val: 1.338 | iter time: 359.29 ms (step) remaining time: 0:25:41
|
| 213 |
+
Validating ...
|
| 214 |
+
iter 3200: val loss 1.3139, val time: 21910.85 ms
|
| 215 |
+
Saving checkpoint to 'out/pretrain/tinyllama/2411_lr4e-5/step-00000100/lit_model.pth'
|
| 216 |
+
Epoch 1 | iter 3232 step 101 | loss train: 1.311, val: 1.314 | iter time: 357.56 ms (step) remaining time: 0:26:21
|
| 217 |
+
Epoch 1 | iter 3264 step 102 | loss train: 1.323, val: 1.314 | iter time: 356.72 ms (step) remaining time: 0:26:09
|
| 218 |
+
Epoch 1 | iter 3296 step 103 | loss train: 1.438, val: 1.314 | iter time: 359.89 ms (step) remaining time: 0:25:56
|
| 219 |
+
Epoch 1 | iter 3328 step 104 | loss train: 1.375, val: 1.314 | iter time: 358.43 ms (step) remaining time: 0:25:44
|
| 220 |
+
Epoch 1 | iter 3360 step 105 | loss train: 1.390, val: 1.314 | iter time: 359.49 ms (step) remaining time: 0:25:32
|
| 221 |
+
Epoch 1 | iter 3392 step 106 | loss train: 1.341, val: 1.314 | iter time: 359.77 ms (step) remaining time: 0:25:20
|
| 222 |
+
Epoch 1 | iter 3424 step 107 | loss train: 1.385, val: 1.314 | iter time: 359.95 ms (step) remaining time: 0:25:08
|
| 223 |
+
Epoch 1 | iter 3456 step 108 | loss train: 1.376, val: 1.314 | iter time: 358.45 ms (step) remaining time: 0:24:56
|
| 224 |
+
Epoch 1 | iter 3488 step 109 | loss train: 1.362, val: 1.314 | iter time: 358.93 ms (step) remaining time: 0:24:44
|
| 225 |
+
Epoch 1 | iter 3520 step 110 | loss train: 1.355, val: 1.314 | iter time: 358.55 ms (step) remaining time: 0:24:32
|
| 226 |
+
Epoch 1 | iter 3552 step 111 | loss train: 1.465, val: 1.314 | iter time: 360.07 ms (step) remaining time: 0:24:20
|
| 227 |
+
Epoch 1 | iter 3584 step 112 | loss train: 1.377, val: 1.314 | iter time: 359.74 ms (step) remaining time: 0:24:08
|
| 228 |
+
Epoch 1 | iter 3616 step 113 | loss train: 1.381, val: 1.314 | iter time: 584.82 ms (step) remaining time: 0:23:56
|
| 229 |
+
Epoch 1 | iter 3648 step 114 | loss train: 1.431, val: 1.314 | iter time: 359.99 ms (step) remaining time: 0:23:44
|
| 230 |
+
Epoch 1 | iter 3680 step 115 | loss train: 1.325, val: 1.314 | iter time: 361.19 ms (step) remaining time: 0:23:32
|
| 231 |
+
Epoch 1 | iter 3712 step 116 | loss train: 1.322, val: 1.314 | iter time: 358.81 ms (step) remaining time: 0:23:20
|
| 232 |
+
Epoch 1 | iter 3744 step 117 | loss train: 1.369, val: 1.314 | iter time: 360.20 ms (step) remaining time: 0:23:08
|
| 233 |
+
Epoch 1 | iter 3776 step 118 | loss train: 1.396, val: 1.314 | iter time: 359.36 ms (step) remaining time: 0:22:56
|
| 234 |
+
Epoch 1 | iter 3808 step 119 | loss train: 1.327, val: 1.314 | iter time: 359.73 ms (step) remaining time: 0:22:44
|
| 235 |
+
Epoch 1 | iter 3840 step 120 | loss train: 1.401, val: 1.314 | iter time: 360.06 ms (step) remaining time: 0:22:33
|
| 236 |
+
Epoch 1 | iter 3872 step 121 | loss train: 1.395, val: 1.314 | iter time: 357.15 ms (step) remaining time: 0:22:21
|
| 237 |
+
Epoch 1 | iter 3904 step 122 | loss train: 1.335, val: 1.314 | iter time: 361.27 ms (step) remaining time: 0:22:09
|
| 238 |
+
Epoch 1 | iter 3936 step 123 | loss train: 1.401, val: 1.314 | iter time: 358.41 ms (step) remaining time: 0:21:57
|
| 239 |
+
Epoch 1 | iter 3968 step 124 | loss train: 1.397, val: 1.314 | iter time: 358.62 ms (step) remaining time: 0:21:45
|
| 240 |
+
Epoch 1 | iter 4000 step 125 | loss train: 1.360, val: 1.314 | iter time: 357.98 ms (step) remaining time: 0:21:34
|
| 241 |
+
Epoch 1 | iter 4032 step 126 | loss train: 1.334, val: 1.314 | iter time: 359.19 ms (step) remaining time: 0:21:22
|
| 242 |
+
Epoch 1 | iter 4064 step 127 | loss train: 1.383, val: 1.314 | iter time: 358.96 ms (step) remaining time: 0:21:10
|
| 243 |
+
Epoch 1 | iter 4096 step 128 | loss train: 1.360, val: 1.314 | iter time: 359.49 ms (step) remaining time: 0:20:58
|
| 244 |
+
Epoch 1 | iter 4128 step 129 | loss train: 1.442, val: 1.314 | iter time: 359.41 ms (step) remaining time: 0:20:47
|
| 245 |
+
Epoch 1 | iter 4160 step 130 | loss train: 1.379, val: 1.314 | iter time: 359.43 ms (step) remaining time: 0:20:35
|
| 246 |
+
Epoch 1 | iter 4192 step 131 | loss train: 1.338, val: 1.314 | iter time: 360.82 ms (step) remaining time: 0:20:23
|
| 247 |
+
Epoch 1 | iter 4224 step 132 | loss train: 1.317, val: 1.314 | iter time: 358.28 ms (step) remaining time: 0:20:12
|
| 248 |
+
Epoch 1 | iter 4256 step 133 | loss train: 1.375, val: 1.314 | iter time: 359.99 ms (step) remaining time: 0:20:00
|
| 249 |
+
Epoch 1 | iter 4288 step 134 | loss train: 1.378, val: 1.314 | iter time: 360.52 ms (step) remaining time: 0:19:48
|
| 250 |
+
Epoch 1 | iter 4320 step 135 | loss train: 1.407, val: 1.314 | iter time: 358.86 ms (step) remaining time: 0:19:36
|
| 251 |
+
Epoch 1 | iter 4352 step 136 | loss train: 1.369, val: 1.314 | iter time: 358.37 ms (step) remaining time: 0:19:25
|
| 252 |
+
Epoch 1 | iter 4384 step 137 | loss train: 1.383, val: 1.314 | iter time: 359.87 ms (step) remaining time: 0:19:13
|
| 253 |
+
Epoch 1 | iter 4416 step 138 | loss train: 1.391, val: 1.314 | iter time: 360.27 ms (step) remaining time: 0:19:02
|
| 254 |
+
Epoch 1 | iter 4448 step 139 | loss train: 1.351, val: 1.314 | iter time: 359.80 ms (step) remaining time: 0:18:50
|
| 255 |
+
Epoch 1 | iter 4480 step 140 | loss train: 1.364, val: 1.314 | iter time: 359.44 ms (step) remaining time: 0:18:38
|
| 256 |
+
Epoch 1 | iter 4512 step 141 | loss train: 1.379, val: 1.314 | iter time: 358.38 ms (step) remaining time: 0:18:27
|
| 257 |
+
Epoch 1 | iter 4544 step 142 | loss train: 1.359, val: 1.314 | iter time: 358.63 ms (step) remaining time: 0:18:15
|
| 258 |
+
Epoch 1 | iter 4576 step 143 | loss train: 1.373, val: 1.314 | iter time: 359.62 ms (step) remaining time: 0:18:03
|
| 259 |
+
Epoch 1 | iter 4608 step 144 | loss train: 1.381, val: 1.314 | iter time: 357.44 ms (step) remaining time: 0:17:52
|
| 260 |
+
Epoch 1 | iter 4640 step 145 | loss train: 1.403, val: 1.314 | iter time: 358.92 ms (step) remaining time: 0:17:40
|
| 261 |
+
Epoch 1 | iter 4672 step 146 | loss train: 1.279, val: 1.314 | iter time: 358.93 ms (step) remaining time: 0:17:29
|
| 262 |
+
Epoch 1 | iter 4704 step 147 | loss train: 1.358, val: 1.314 | iter time: 358.71 ms (step) remaining time: 0:17:17
|
| 263 |
+
Epoch 1 | iter 4736 step 148 | loss train: 1.381, val: 1.314 | iter time: 358.61 ms (step) remaining time: 0:17:06
|
| 264 |
+
Epoch 1 | iter 4768 step 149 | loss train: 1.395, val: 1.314 | iter time: 359.43 ms (step) remaining time: 0:16:54
|
| 265 |
+
Epoch 1 | iter 4800 step 150 | loss train: 1.352, val: 1.314 | iter time: 357.49 ms (step) remaining time: 0:16:43
|
| 266 |
+
Validating ...
|
| 267 |
+
iter 4800: val loss 1.3058, val time: 21890.86 ms
|
| 268 |
+
Epoch 1 | iter 4832 step 151 | loss train: 1.349, val: 1.306 | iter time: 357.46 ms (step) remaining time: 0:16:44
|
| 269 |
+
Epoch 1 | iter 4864 step 152 | loss train: 1.342, val: 1.306 | iter time: 359.41 ms (step) remaining time: 0:16:32
|
| 270 |
+
Epoch 1 | iter 4896 step 153 | loss train: 1.351, val: 1.306 | iter time: 359.81 ms (step) remaining time: 0:16:20
|
| 271 |
+
Epoch 1 | iter 4928 step 154 | loss train: 1.299, val: 1.306 | iter time: 358.87 ms (step) remaining time: 0:16:09
|
| 272 |
+
Epoch 1 | iter 4960 step 155 | loss train: 1.376, val: 1.306 | iter time: 358.86 ms (step) remaining time: 0:15:57
|
| 273 |
+
Epoch 1 | iter 4992 step 156 | loss train: 1.328, val: 1.306 | iter time: 360.34 ms (step) remaining time: 0:15:45
|
| 274 |
+
Epoch 1 | iter 5024 step 157 | loss train: 1.400, val: 1.306 | iter time: 358.51 ms (step) remaining time: 0:15:34
|
| 275 |
+
Epoch 1 | iter 5056 step 158 | loss train: 1.365, val: 1.306 | iter time: 358.29 ms (step) remaining time: 0:15:22
|
| 276 |
+
Epoch 1 | iter 5088 step 159 | loss train: 1.407, val: 1.306 | iter time: 360.57 ms (step) remaining time: 0:15:10
|
| 277 |
+
Epoch 1 | iter 5120 step 160 | loss train: 1.391, val: 1.306 | iter time: 358.20 ms (step) remaining time: 0:14:59
|
| 278 |
+
Epoch 1 | iter 5152 step 161 | loss train: 1.380, val: 1.306 | iter time: 357.64 ms (step) remaining time: 0:14:47
|
| 279 |
+
Epoch 1 | iter 5184 step 162 | loss train: 1.345, val: 1.306 | iter time: 358.64 ms (step) remaining time: 0:14:35
|
| 280 |
+
Epoch 1 | iter 5216 step 163 | loss train: 1.341, val: 1.306 | iter time: 359.35 ms (step) remaining time: 0:14:24
|
| 281 |
+
Epoch 1 | iter 5248 step 164 | loss train: 1.271, val: 1.306 | iter time: 359.06 ms (step) remaining time: 0:14:12
|
| 282 |
+
Epoch 1 | iter 5280 step 165 | loss train: 1.330, val: 1.306 | iter time: 358.38 ms (step) remaining time: 0:14:01
|
| 283 |
+
Epoch 1 | iter 5312 step 166 | loss train: 1.346, val: 1.306 | iter time: 360.27 ms (step) remaining time: 0:13:49
|
| 284 |
+
Epoch 1 | iter 5344 step 167 | loss train: 1.325, val: 1.306 | iter time: 358.67 ms (step) remaining time: 0:13:37
|
| 285 |
+
Epoch 1 | iter 5376 step 168 | loss train: 1.397, val: 1.306 | iter time: 360.59 ms (step) remaining time: 0:13:26
|
| 286 |
+
Epoch 1 | iter 5408 step 169 | loss train: 1.329, val: 1.306 | iter time: 359.50 ms (step) remaining time: 0:13:14
|
| 287 |
+
Epoch 1 | iter 5440 step 170 | loss train: 1.366, val: 1.306 | iter time: 360.68 ms (step) remaining time: 0:13:03
|
| 288 |
+
Epoch 1 | iter 5472 step 171 | loss train: 1.365, val: 1.306 | iter time: 358.40 ms (step) remaining time: 0:12:51
|
| 289 |
+
Epoch 1 | iter 5504 step 172 | loss train: 1.384, val: 1.306 | iter time: 358.19 ms (step) remaining time: 0:12:40
|
| 290 |
+
Epoch 1 | iter 5536 step 173 | loss train: 1.356, val: 1.306 | iter time: 358.01 ms (step) remaining time: 0:12:28
|
| 291 |
+
Epoch 1 | iter 5568 step 174 | loss train: 1.381, val: 1.306 | iter time: 358.96 ms (step) remaining time: 0:12:17
|
| 292 |
+
Epoch 1 | iter 5600 step 175 | loss train: 1.273, val: 1.306 | iter time: 358.58 ms (step) remaining time: 0:12:05
|
| 293 |
+
Epoch 1 | iter 5632 step 176 | loss train: 1.348, val: 1.306 | iter time: 361.06 ms (step) remaining time: 0:11:54
|
| 294 |
+
Epoch 1 | iter 5664 step 177 | loss train: 1.291, val: 1.306 | iter time: 357.70 ms (step) remaining time: 0:11:42
|
| 295 |
+
Epoch 1 | iter 5696 step 178 | loss train: 1.348, val: 1.306 | iter time: 359.83 ms (step) remaining time: 0:11:31
|
| 296 |
+
Epoch 1 | iter 5728 step 179 | loss train: 1.367, val: 1.306 | iter time: 358.49 ms (step) remaining time: 0:11:19
|
| 297 |
+
Epoch 1 | iter 5760 step 180 | loss train: 1.295, val: 1.306 | iter time: 357.70 ms (step) remaining time: 0:11:08
|
| 298 |
+
Epoch 1 | iter 5792 step 181 | loss train: 1.335, val: 1.306 | iter time: 359.37 ms (step) remaining time: 0:10:56
|
| 299 |
+
Epoch 1 | iter 5824 step 182 | loss train: 1.340, val: 1.306 | iter time: 358.24 ms (step) remaining time: 0:10:45
|
| 300 |
+
Epoch 1 | iter 5856 step 183 | loss train: 1.318, val: 1.306 | iter time: 355.62 ms (step) remaining time: 0:10:33
|
| 301 |
+
Epoch 1 | iter 5888 step 184 | loss train: 1.372, val: 1.306 | iter time: 358.89 ms (step) remaining time: 0:10:22
|
| 302 |
+
Epoch 1 | iter 5920 step 185 | loss train: 1.386, val: 1.306 | iter time: 357.10 ms (step) remaining time: 0:10:10
|
| 303 |
+
Epoch 1 | iter 5952 step 186 | loss train: 1.386, val: 1.306 | iter time: 360.36 ms (step) remaining time: 0:09:59
|
| 304 |
+
Epoch 1 | iter 5984 step 187 | loss train: 1.323, val: 1.306 | iter time: 360.42 ms (step) remaining time: 0:09:47
|
| 305 |
+
Epoch 1 | iter 6016 step 188 | loss train: 1.316, val: 1.306 | iter time: 359.29 ms (step) remaining time: 0:09:36
|
| 306 |
+
Epoch 1 | iter 6048 step 189 | loss train: 1.380, val: 1.306 | iter time: 357.98 ms (step) remaining time: 0:09:25
|
| 307 |
+
Epoch 1 | iter 6080 step 190 | loss train: 1.337, val: 1.306 | iter time: 360.39 ms (step) remaining time: 0:09:13
|
| 308 |
+
Epoch 1 | iter 6112 step 191 | loss train: 1.375, val: 1.306 | iter time: 359.23 ms (step) remaining time: 0:09:02
|
| 309 |
+
Epoch 1 | iter 6144 step 192 | loss train: 1.444, val: 1.306 | iter time: 358.45 ms (step) remaining time: 0:08:50
|
| 310 |
+
Epoch 1 | iter 6176 step 193 | loss train: 1.324, val: 1.306 | iter time: 359.90 ms (step) remaining time: 0:08:39
|
| 311 |
+
Epoch 1 | iter 6208 step 194 | loss train: 1.323, val: 1.306 | iter time: 612.97 ms (step) remaining time: 0:08:28
|
| 312 |
+
Epoch 1 | iter 6240 step 195 | loss train: 1.316, val: 1.306 | iter time: 360.08 ms (step) remaining time: 0:08:16
|
| 313 |
+
Epoch 1 | iter 6272 step 196 | loss train: 1.370, val: 1.306 | iter time: 356.98 ms (step) remaining time: 0:08:05
|
| 314 |
+
Epoch 1 | iter 6304 step 197 | loss train: 1.388, val: 1.306 | iter time: 361.03 ms (step) remaining time: 0:07:54
|
| 315 |
+
Epoch 1 | iter 6336 step 198 | loss train: 1.365, val: 1.306 | iter time: 358.46 ms (step) remaining time: 0:07:42
|
| 316 |
+
Epoch 1 | iter 6368 step 199 | loss train: 1.332, val: 1.306 | iter time: 360.29 ms (step) remaining time: 0:07:31
|
| 317 |
+
Epoch 1 | iter 6400 step 200 | loss train: 1.451, val: 1.306 | iter time: 359.46 ms (step) remaining time: 0:07:19
|
| 318 |
+
Validating ...
|
| 319 |
+
iter 6400: val loss 1.3072, val time: 21899.37 ms
|
| 320 |
+
Saving checkpoint to 'out/pretrain/tinyllama/2411_lr4e-5/step-00000200/lit_model.pth'
|
| 321 |
+
Epoch 1 | iter 6432 step 201 | loss train: 1.341, val: 1.307 | iter time: 358.12 ms (step) remaining time: 0:07:15
|
| 322 |
+
Epoch 1 | iter 6464 step 202 | loss train: 1.344, val: 1.307 | iter time: 359.29 ms (step) remaining time: 0:07:04
|
| 323 |
+
Epoch 1 | iter 6496 step 203 | loss train: 1.339, val: 1.307 | iter time: 360.33 ms (step) remaining time: 0:06:52
|
| 324 |
+
Epoch 1 | iter 6528 step 204 | loss train: 1.402, val: 1.307 | iter time: 360.76 ms (step) remaining time: 0:06:41
|
| 325 |
+
Epoch 1 | iter 6560 step 205 | loss train: 1.297, val: 1.307 | iter time: 359.61 ms (step) remaining time: 0:06:29
|
| 326 |
+
Epoch 1 | iter 6592 step 206 | loss train: 1.375, val: 1.307 | iter time: 360.07 ms (step) remaining time: 0:06:17
|
| 327 |
+
Epoch 1 | iter 6624 step 207 | loss train: 1.332, val: 1.307 | iter time: 361.61 ms (step) remaining time: 0:06:06
|
| 328 |
+
Epoch 1 | iter 6656 step 208 | loss train: 1.331, val: 1.307 | iter time: 359.24 ms (step) remaining time: 0:05:54
|
| 329 |
+
Epoch 1 | iter 6688 step 209 | loss train: 1.379, val: 1.307 | iter time: 359.51 ms (step) remaining time: 0:05:43
|
| 330 |
+
Epoch 1 | iter 6720 step 210 | loss train: 1.335, val: 1.307 | iter time: 360.13 ms (step) remaining time: 0:05:31
|
| 331 |
+
Epoch 1 | iter 6752 step 211 | loss train: 1.298, val: 1.307 | iter time: 358.39 ms (step) remaining time: 0:05:20
|
| 332 |
+
Epoch 1 | iter 6784 step 212 | loss train: 1.303, val: 1.307 | iter time: 359.29 ms (step) remaining time: 0:05:08
|
| 333 |
+
Epoch 1 | iter 6816 step 213 | loss train: 1.384, val: 1.307 | iter time: 360.50 ms (step) remaining time: 0:04:57
|
| 334 |
+
Epoch 1 | iter 6848 step 214 | loss train: 1.401, val: 1.307 | iter time: 359.59 ms (step) remaining time: 0:04:45
|
| 335 |
+
Epoch 1 | iter 6880 step 215 | loss train: 1.283, val: 1.307 | iter time: 358.61 ms (step) remaining time: 0:04:34
|
| 336 |
+
Epoch 1 | iter 6912 step 216 | loss train: 1.342, val: 1.307 | iter time: 360.32 ms (step) remaining time: 0:04:22
|
| 337 |
+
Epoch 1 | iter 6944 step 217 | loss train: 1.273, val: 1.307 | iter time: 359.43 ms (step) remaining time: 0:04:11
|
| 338 |
+
Epoch 1 | iter 6976 step 218 | loss train: 1.313, val: 1.307 | iter time: 360.51 ms (step) remaining time: 0:03:59
|
| 339 |
+
Epoch 1 | iter 7008 step 219 | loss train: 1.334, val: 1.307 | iter time: 359.06 ms (step) remaining time: 0:03:48
|
| 340 |
+
Epoch 1 | iter 7040 step 220 | loss train: 1.355, val: 1.307 | iter time: 359.24 ms (step) remaining time: 0:03:36
|
| 341 |
+
Epoch 1 | iter 7072 step 221 | loss train: 1.355, val: 1.307 | iter time: 359.64 ms (step) remaining time: 0:03:25
|
| 342 |
+
Epoch 1 | iter 7104 step 222 | loss train: 1.301, val: 1.307 | iter time: 358.80 ms (step) remaining time: 0:03:13
|
| 343 |
+
Epoch 1 | iter 7136 step 223 | loss train: 1.348, val: 1.307 | iter time: 358.49 ms (step) remaining time: 0:03:02
|
| 344 |
+
Epoch 1 | iter 7168 step 224 | loss train: 1.378, val: 1.307 | iter time: 358.56 ms (step) remaining time: 0:02:51
|
| 345 |
+
Epoch 1 | iter 7200 step 225 | loss train: 1.429, val: 1.307 | iter time: 358.82 ms (step) remaining time: 0:02:39
|
| 346 |
+
Epoch 1 | iter 7232 step 226 | loss train: 1.401, val: 1.307 | iter time: 359.42 ms (step) remaining time: 0:02:28
|
| 347 |
+
Epoch 1 | iter 7264 step 227 | loss train: 1.388, val: 1.307 | iter time: 360.22 ms (step) remaining time: 0:02:16
|
| 348 |
+
Epoch 1 | iter 7296 step 228 | loss train: 1.397, val: 1.307 | iter time: 359.12 ms (step) remaining time: 0:02:05
|
| 349 |
+
Epoch 1 | iter 7328 step 229 | loss train: 1.382, val: 1.307 | iter time: 358.52 ms (step) remaining time: 0:01:53
|
| 350 |
+
Epoch 1 | iter 7360 step 230 | loss train: 1.304, val: 1.307 | iter time: 358.21 ms (step) remaining time: 0:01:42
|
| 351 |
+
Epoch 1 | iter 7392 step 231 | loss train: 1.306, val: 1.307 | iter time: 358.57 ms (step) remaining time: 0:01:31
|
| 352 |
+
Epoch 1 | iter 7424 step 232 | loss train: 1.297, val: 1.307 | iter time: 359.78 ms (step) remaining time: 0:01:19
|
| 353 |
+
Epoch 1 | iter 7456 step 233 | loss train: 1.375, val: 1.307 | iter time: 360.36 ms (step) remaining time: 0:01:08
|
| 354 |
+
Epoch 1 | iter 7488 step 234 | loss train: 1.353, val: 1.307 | iter time: 360.17 ms (step) remaining time: 0:00:56
|
| 355 |
+
Epoch 1 | iter 7520 step 235 | loss train: 1.319, val: 1.307 | iter time: 359.95 ms (step) remaining time: 0:00:45
|
| 356 |
+
Epoch 1 | iter 7552 step 236 | loss train: 1.331, val: 1.307 | iter time: 359.22 ms (step) remaining time: 0:00:34
|
| 357 |
+
Epoch 1 | iter 7584 step 237 | loss train: 1.325, val: 1.307 | iter time: 360.10 ms (step) remaining time: 0:00:22
|
| 358 |
+
Epoch 1 | iter 7616 step 238 | loss train: 1.324, val: 1.307 | iter time: 360.19 ms (step) remaining time: 0:00:11
|
| 359 |
+
Epoch 2 | iter 7648 step 239 | loss train: 1.201, val: 1.307 | iter time: 359.45 ms (step) remaining time: 0:00:00
|
| 360 |
+
Validating ...
|
| 361 |
+
Final evaluation | val loss: 1.301 | val ppl: 3.672
|
| 362 |
+
Saving checkpoint to 'out/pretrain/tinyllama/2411_lr4e-5/final/lit_model.pth'
|
| 363 |
+
----------------------------------------
|
| 364 |
+
| Performance
|
| 365 |
+
| - Total tokens : 250,609,664
|
| 366 |
+
| - Training Time : 2780.44 s
|
| 367 |
+
| - Tok/sec : 129.06 tok/s
|
| 368 |
+
| ----------------------------------------
|
| 369 |
+
| Memory Usage
|
| 370 |
+
| - Memory Used : 26.32 GB
|
| 371 |
+
----------------------------------------
|
out/pretrain/tinyllama/teelogs/2412.txt
ADDED
|
@@ -0,0 +1,414 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
Initializing distributed: GLOBAL_RANK: 0, MEMBER: 1/4
|
| 2 |
+
Initializing distributed: GLOBAL_RANK: 2, MEMBER: 3/4
|
| 3 |
+
[rank: 2] Seed set to 42
|
| 4 |
+
Initializing distributed: GLOBAL_RANK: 3, MEMBER: 4/4
|
| 5 |
+
Initializing distributed: GLOBAL_RANK: 1, MEMBER: 2/4
|
| 6 |
+
----------------------------------------------------------------------------------------------------
|
| 7 |
+
distributed_backend=nccl
|
| 8 |
+
All distributed processes registered. Starting with 4 processes
|
| 9 |
+
----------------------------------------------------------------------------------------------------
|
| 10 |
+
|
| 11 |
+
[rank: 1] Seed set to 42
|
| 12 |
+
[rank: 3] Seed set to 42
|
| 13 |
+
[rank: 0] Seed set to 42
|
| 14 |
+
All GPUs are fully connected via NVLink.
|
| 15 |
+
{'data': {'batch_size': 1,
|
| 16 |
+
'data_path': PosixPath('litgpt/data/arxiv'),
|
| 17 |
+
'num_workers': 8,
|
| 18 |
+
'seed': 42,
|
| 19 |
+
'seq_length': 2048,
|
| 20 |
+
'use_starcoder': True},
|
| 21 |
+
'data_dir': PosixPath('litgpt/data/arxiv/2412'),
|
| 22 |
+
'devices': 'auto',
|
| 23 |
+
'eval': {'evaluate_example': 'first',
|
| 24 |
+
'final_validation': True,
|
| 25 |
+
'initial_validation': True,
|
| 26 |
+
'interval': 50,
|
| 27 |
+
'max_iters': 100,
|
| 28 |
+
'max_new_tokens': None},
|
| 29 |
+
'initial_checkpoint_dir': PosixPath('out/pretrain/2411/final'),
|
| 30 |
+
'log': {'group': None, 'project': None, 'run': None},
|
| 31 |
+
'logger_name': 'tensorboard',
|
| 32 |
+
'model_config': {'attention_logit_softcapping': None,
|
| 33 |
+
'attention_scores_scalar': None,
|
| 34 |
+
'attn_bias': False,
|
| 35 |
+
'bias': False,
|
| 36 |
+
'block_size': 2048,
|
| 37 |
+
'final_logit_softcapping': None,
|
| 38 |
+
'gelu_approximate': 'none',
|
| 39 |
+
'head_size': 64,
|
| 40 |
+
'hf_config': {'name': 'TinyLlama-1.1B-intermediate-step-1431k-3T',
|
| 41 |
+
'org': 'TinyLlama'},
|
| 42 |
+
'intermediate_size': 5632,
|
| 43 |
+
'lm_head_bias': False,
|
| 44 |
+
'mlp_class_name': 'LLaMAMLP',
|
| 45 |
+
'moe_intermediate_size': None,
|
| 46 |
+
'n_embd': 2048,
|
| 47 |
+
'n_expert': 0,
|
| 48 |
+
'n_expert_per_token': 0,
|
| 49 |
+
'n_head': 32,
|
| 50 |
+
'n_layer': 22,
|
| 51 |
+
'n_query_groups': 4,
|
| 52 |
+
'name': 'tiny-llama-1.1b',
|
| 53 |
+
'norm_1': True,
|
| 54 |
+
'norm_2': True,
|
| 55 |
+
'norm_class_name': 'RMSNorm',
|
| 56 |
+
'norm_eps': 1e-05,
|
| 57 |
+
'norm_qk': False,
|
| 58 |
+
'norm_qk_type': 'default',
|
| 59 |
+
'padded_vocab_size': 32000,
|
| 60 |
+
'padding_multiple': 64,
|
| 61 |
+
'parallel_residual': False,
|
| 62 |
+
'post_attention_norm': False,
|
| 63 |
+
'post_mlp_norm': False,
|
| 64 |
+
'rope_adjustments': None,
|
| 65 |
+
'rope_base': 10000,
|
| 66 |
+
'rope_condense_ratio': 1,
|
| 67 |
+
'rope_indices': None,
|
| 68 |
+
'rope_local_base_freq': None,
|
| 69 |
+
'rotary_percentage': 1.0,
|
| 70 |
+
'scale_embeddings': False,
|
| 71 |
+
'shared_attention_norm': False,
|
| 72 |
+
'sliding_window_indices': None,
|
| 73 |
+
'sliding_window_size': None,
|
| 74 |
+
'vocab_size': 32000},
|
| 75 |
+
'model_name': 'tiny-llama-1.1b',
|
| 76 |
+
'num_nodes': 1,
|
| 77 |
+
'optimizer': "{'class_path': 'torch.optim.AdamW', 'init_args': {'lr': 0.0004, "
|
| 78 |
+
"'weight_decay': 0.1, 'betas': [0.9, 0.95]}}",
|
| 79 |
+
'out_dir': PosixPath('out/pretrain/2412'),
|
| 80 |
+
'precision': 'bf16-mixed',
|
| 81 |
+
'resume': False,
|
| 82 |
+
'seed': 42,
|
| 83 |
+
'tokenizer_dir': PosixPath('checkpoints/TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T'),
|
| 84 |
+
'train': {'epochs': None,
|
| 85 |
+
'global_batch_size': 512,
|
| 86 |
+
'log_interval': 1,
|
| 87 |
+
'lr_warmup_fraction': None,
|
| 88 |
+
'lr_warmup_steps': 20,
|
| 89 |
+
'max_norm': 1.0,
|
| 90 |
+
'max_seq_length': 2048,
|
| 91 |
+
'max_steps': None,
|
| 92 |
+
'max_tokens': 301989888,
|
| 93 |
+
'micro_batch_size': 4,
|
| 94 |
+
'min_lr': 4e-05,
|
| 95 |
+
'save_interval': 100,
|
| 96 |
+
'tie_embeddings': None}}
|
| 97 |
+
Time to instantiate model: 0.04 seconds.
|
| 98 |
+
Total parameters: 1,100,048,384
|
| 99 |
+
[fix] out/pretrain/2411/final/lit_model.pth 是整包 state，提取 model 权重并原子覆盖...
|
| 100 |
+
[fix] 已覆盖为纯权重: out/pretrain/2411/final/lit_model.pth
|
| 101 |
+
Validating ...
|
| 102 |
+
Measured TFLOPs: 239.66
|
| 103 |
+
Epoch 1 | iter 32 step 1 | loss train: 1.340, val: 1.515 | iter time: 550.46 ms (step) remaining time: 0:56:58
|
| 104 |
+
Epoch 1 | iter 64 step 2 | loss train: 1.361, val: 1.515 | iter time: 357.72 ms (step) remaining time: 0:54:01
|
| 105 |
+
Epoch 1 | iter 96 step 3 | loss train: 1.446, val: 1.515 | iter time: 357.30 ms (step) remaining time: 0:52:56
|
| 106 |
+
Epoch 1 | iter 128 step 4 | loss train: 1.434, val: 1.515 | iter time: 358.09 ms (step) remaining time: 0:52:19
|
| 107 |
+
Epoch 1 | iter 160 step 5 | loss train: 1.324, val: 1.515 | iter time: 358.59 ms (step) remaining time: 0:51:54
|
| 108 |
+
Epoch 1 | iter 192 step 6 | loss train: 1.345, val: 1.515 | iter time: 357.57 ms (step) remaining time: 0:51:34
|
| 109 |
+
Epoch 1 | iter 224 step 7 | loss train: 1.359, val: 1.515 | iter time: 357.80 ms (step) remaining time: 0:51:17
|
| 110 |
+
Epoch 1 | iter 256 step 8 | loss train: 1.362, val: 1.515 | iter time: 357.59 ms (step) remaining time: 0:51:02
|
| 111 |
+
Epoch 1 | iter 288 step 9 | loss train: 1.357, val: 1.515 | iter time: 360.39 ms (step) remaining time: 0:50:48
|
| 112 |
+
Epoch 1 | iter 320 step 10 | loss train: 1.415, val: 1.515 | iter time: 357.40 ms (step) remaining time: 0:50:34
|
| 113 |
+
Epoch 1 | iter 352 step 11 | loss train: 1.381, val: 1.515 | iter time: 358.29 ms (step) remaining time: 0:50:21
|
| 114 |
+
Epoch 1 | iter 384 step 12 | loss train: 1.444, val: 1.515 | iter time: 360.64 ms (step) remaining time: 0:50:09
|
| 115 |
+
Epoch 1 | iter 416 step 13 | loss train: 1.399, val: 1.515 | iter time: 358.73 ms (step) remaining time: 0:49:57
|
| 116 |
+
Epoch 1 | iter 448 step 14 | loss train: 1.363, val: 1.515 | iter time: 360.75 ms (step) remaining time: 0:49:45
|
| 117 |
+
Epoch 1 | iter 480 step 15 | loss train: 1.498, val: 1.515 | iter time: 359.49 ms (step) remaining time: 0:49:33
|
| 118 |
+
Epoch 1 | iter 512 step 16 | loss train: 1.433, val: 1.515 | iter time: 358.86 ms (step) remaining time: 0:49:21
|
| 119 |
+
Epoch 1 | iter 544 step 17 | loss train: 1.432, val: 1.515 | iter time: 358.03 ms (step) remaining time: 0:49:09
|
| 120 |
+
Epoch 1 | iter 576 step 18 | loss train: 1.403, val: 1.515 | iter time: 360.03 ms (step) remaining time: 0:48:58
|
| 121 |
+
Epoch 1 | iter 608 step 19 | loss train: 1.403, val: 1.515 | iter time: 358.78 ms (step) remaining time: 0:48:46
|
| 122 |
+
Epoch 1 | iter 640 step 20 | loss train: 1.457, val: 1.515 | iter time: 359.36 ms (step) remaining time: 0:48:35
|
| 123 |
+
Epoch 1 | iter 672 step 21 | loss train: 1.389, val: 1.515 | iter time: 358.37 ms (step) remaining time: 0:48:24
|
| 124 |
+
Epoch 1 | iter 704 step 22 | loss train: 1.465, val: 1.515 | iter time: 359.08 ms (step) remaining time: 0:48:13
|
| 125 |
+
Epoch 1 | iter 736 step 23 | loss train: 1.411, val: 1.515 | iter time: 358.30 ms (step) remaining time: 0:48:02
|
| 126 |
+
Epoch 1 | iter 768 step 24 | loss train: 1.469, val: 1.515 | iter time: 359.87 ms (step) remaining time: 0:47:51
|
| 127 |
+
Epoch 1 | iter 800 step 25 | loss train: 1.427, val: 1.515 | iter time: 358.92 ms (step) remaining time: 0:47:40
|
| 128 |
+
Epoch 1 | iter 832 step 26 | loss train: 1.383, val: 1.515 | iter time: 360.24 ms (step) remaining time: 0:47:29
|
| 129 |
+
Epoch 1 | iter 864 step 27 | loss train: 1.419, val: 1.515 | iter time: 360.07 ms (step) remaining time: 0:47:18
|
| 130 |
+
Epoch 1 | iter 896 step 28 | loss train: 1.441, val: 1.515 | iter time: 359.76 ms (step) remaining time: 0:47:07
|
| 131 |
+
Epoch 1 | iter 928 step 29 | loss train: 1.427, val: 1.515 | iter time: 358.88 ms (step) remaining time: 0:46:56
|
| 132 |
+
Epoch 1 | iter 960 step 30 | loss train: 1.467, val: 1.515 | iter time: 359.25 ms (step) remaining time: 0:46:45
|
| 133 |
+
Epoch 1 | iter 992 step 31 | loss train: 1.430, val: 1.515 | iter time: 359.50 ms (step) remaining time: 0:46:34
|
| 134 |
+
Epoch 1 | iter 1024 step 32 | loss train: 1.436, val: 1.515 | iter time: 359.80 ms (step) remaining time: 0:46:24
|
| 135 |
+
Epoch 1 | iter 1056 step 33 | loss train: 1.396, val: 1.515 | iter time: 360.07 ms (step) remaining time: 0:46:13
|
| 136 |
+
Epoch 1 | iter 1088 step 34 | loss train: 1.380, val: 1.515 | iter time: 359.33 ms (step) remaining time: 0:46:02
|
| 137 |
+
Epoch 1 | iter 1120 step 35 | loss train: 1.468, val: 1.515 | iter time: 358.38 ms (step) remaining time: 0:45:51
|
| 138 |
+
Epoch 1 | iter 1152 step 36 | loss train: 1.416, val: 1.515 | iter time: 359.15 ms (step) remaining time: 0:45:40
|
| 139 |
+
Epoch 1 | iter 1184 step 37 | loss train: 1.398, val: 1.515 | iter time: 359.93 ms (step) remaining time: 0:45:29
|
| 140 |
+
Epoch 1 | iter 1216 step 38 | loss train: 1.409, val: 1.515 | iter time: 357.99 ms (step) remaining time: 0:45:18
|
| 141 |
+
Epoch 1 | iter 1248 step 39 | loss train: 1.389, val: 1.515 | iter time: 736.23 ms (step) remaining time: 0:45:10
|
| 142 |
+
Epoch 1 | iter 1280 step 40 | loss train: 1.404, val: 1.515 | iter time: 360.93 ms (step) remaining time: 0:44:59
|
| 143 |
+
Epoch 1 | iter 1312 step 41 | loss train: 1.405, val: 1.515 | iter time: 361.51 ms (step) remaining time: 0:44:48
|
| 144 |
+
Epoch 1 | iter 1344 step 42 | loss train: 1.451, val: 1.515 | iter time: 360.00 ms (step) remaining time: 0:44:37
|
| 145 |
+
Epoch 1 | iter 1376 step 43 | loss train: 1.460, val: 1.515 | iter time: 359.28 ms (step) remaining time: 0:44:26
|
| 146 |
+
Epoch 1 | iter 1408 step 44 | loss train: 1.472, val: 1.515 | iter time: 358.33 ms (step) remaining time: 0:44:14
|
| 147 |
+
Epoch 1 | iter 1440 step 45 | loss train: 1.415, val: 1.515 | iter time: 361.15 ms (step) remaining time: 0:44:03
|
| 148 |
+
Epoch 1 | iter 1472 step 46 | loss train: 1.460, val: 1.515 | iter time: 359.96 ms (step) remaining time: 0:43:52
|
| 149 |
+
Epoch 1 | iter 1504 step 47 | loss train: 1.489, val: 1.515 | iter time: 359.96 ms (step) remaining time: 0:43:41
|
| 150 |
+
Epoch 1 | iter 1536 step 48 | loss train: 1.385, val: 1.515 | iter time: 357.90 ms (step) remaining time: 0:43:30
|
| 151 |
+
Epoch 1 | iter 1568 step 49 | loss train: 1.511, val: 1.515 | iter time: 359.73 ms (step) remaining time: 0:43:19
|
| 152 |
+
Epoch 1 | iter 1600 step 50 | loss train: 1.442, val: 1.515 | iter time: 357.52 ms (step) remaining time: 0:43:08
|
| 153 |
+
Validating ...
|
| 154 |
+
iter 1600: val loss 1.6045, val time: 11279.09 ms
|
| 155 |
+
Epoch 1 | iter 1632 step 51 | loss train: 1.426, val: 1.604 | iter time: 359.26 ms (step) remaining time: 0:43:50
|
| 156 |
+
Epoch 1 | iter 1664 step 52 | loss train: 1.486, val: 1.604 | iter time: 360.42 ms (step) remaining time: 0:43:38
|
| 157 |
+
Epoch 1 | iter 1696 step 53 | loss train: 1.418, val: 1.604 | iter time: 359.57 ms (step) remaining time: 0:43:25
|
| 158 |
+
Epoch 1 | iter 1728 step 54 | loss train: 1.393, val: 1.604 | iter time: 360.09 ms (step) remaining time: 0:43:13
|
| 159 |
+
Epoch 1 | iter 1760 step 55 | loss train: 1.424, val: 1.604 | iter time: 357.67 ms (step) remaining time: 0:43:01
|
| 160 |
+
Epoch 1 | iter 1792 step 56 | loss train: 1.438, val: 1.604 | iter time: 361.34 ms (step) remaining time: 0:42:49
|
| 161 |
+
Epoch 1 | iter 1824 step 57 | loss train: 1.448, val: 1.604 | iter time: 358.53 ms (step) remaining time: 0:42:37
|
| 162 |
+
Epoch 1 | iter 1856 step 58 | loss train: 1.469, val: 1.604 | iter time: 360.29 ms (step) remaining time: 0:42:25
|
| 163 |
+
Epoch 1 | iter 1888 step 59 | loss train: 1.441, val: 1.604 | iter time: 359.49 ms (step) remaining time: 0:42:13
|
| 164 |
+
Epoch 1 | iter 1920 step 60 | loss train: 1.403, val: 1.604 | iter time: 358.95 ms (step) remaining time: 0:42:02
|
| 165 |
+
Epoch 1 | iter 1952 step 61 | loss train: 1.517, val: 1.604 | iter time: 359.77 ms (step) remaining time: 0:41:50
|
| 166 |
+
Epoch 1 | iter 1984 step 62 | loss train: 1.388, val: 1.604 | iter time: 360.10 ms (step) remaining time: 0:41:38
|
| 167 |
+
Epoch 1 | iter 2016 step 63 | loss train: 1.500, val: 1.604 | iter time: 358.88 ms (step) remaining time: 0:41:26
|
| 168 |
+
Epoch 1 | iter 2048 step 64 | loss train: 1.350, val: 1.604 | iter time: 359.68 ms (step) remaining time: 0:41:14
|
| 169 |
+
Epoch 1 | iter 2080 step 65 | loss train: 1.486, val: 1.604 | iter time: 360.35 ms (step) remaining time: 0:41:03
|
| 170 |
+
Epoch 1 | iter 2112 step 66 | loss train: 1.392, val: 1.604 | iter time: 358.01 ms (step) remaining time: 0:40:51
|
| 171 |
+
Epoch 1 | iter 2144 step 67 | loss train: 1.399, val: 1.604 | iter time: 359.99 ms (step) remaining time: 0:40:39
|
| 172 |
+
Epoch 1 | iter 2176 step 68 | loss train: 1.407, val: 1.604 | iter time: 358.84 ms (step) remaining time: 0:40:28
|
| 173 |
+
Epoch 1 | iter 2208 step 69 | loss train: 1.447, val: 1.604 | iter time: 360.85 ms (step) remaining time: 0:40:16
|
| 174 |
+
Epoch 1 | iter 2240 step 70 | loss train: 1.491, val: 1.604 | iter time: 358.11 ms (step) remaining time: 0:40:05
|
| 175 |
+
Epoch 1 | iter 2272 step 71 | loss train: 1.385, val: 1.604 | iter time: 360.35 ms (step) remaining time: 0:39:53
|
| 176 |
+
Epoch 1 | iter 2304 step 72 | loss train: 1.379, val: 1.604 | iter time: 359.89 ms (step) remaining time: 0:39:41
|
| 177 |
+
Epoch 1 | iter 2336 step 73 | loss train: 1.394, val: 1.604 | iter time: 359.85 ms (step) remaining time: 0:39:30
|
| 178 |
+
Epoch 1 | iter 2368 step 74 | loss train: 1.458, val: 1.604 | iter time: 360.69 ms (step) remaining time: 0:39:18
|
| 179 |
+
Epoch 1 | iter 2400 step 75 | loss train: 1.439, val: 1.604 | iter time: 358.82 ms (step) remaining time: 0:39:07
|
| 180 |
+
Epoch 1 | iter 2432 step 76 | loss train: 1.379, val: 1.604 | iter time: 357.71 ms (step) remaining time: 0:38:55
|
| 181 |
+
Epoch 1 | iter 2464 step 77 | loss train: 1.413, val: 1.604 | iter time: 360.82 ms (step) remaining time: 0:38:44
|
| 182 |
+
Epoch 1 | iter 2496 step 78 | loss train: 1.356, val: 1.604 | iter time: 359.71 ms (step) remaining time: 0:38:32
|
| 183 |
+
Epoch 1 | iter 2528 step 79 | loss train: 1.404, val: 1.604 | iter time: 359.23 ms (step) remaining time: 0:38:21
|
| 184 |
+
Epoch 1 | iter 2560 step 80 | loss train: 1.459, val: 1.604 | iter time: 359.05 ms (step) remaining time: 0:38:10
|
| 185 |
+
Epoch 1 | iter 2592 step 81 | loss train: 1.402, val: 1.604 | iter time: 358.61 ms (step) remaining time: 0:37:58
|
| 186 |
+
Epoch 1 | iter 2624 step 82 | loss train: 1.393, val: 1.604 | iter time: 360.20 ms (step) remaining time: 0:37:47
|
| 187 |
+
Epoch 1 | iter 2656 step 83 | loss train: 1.351, val: 1.604 | iter time: 359.56 ms (step) remaining time: 0:37:36
|
| 188 |
+
Epoch 1 | iter 2688 step 84 | loss train: 1.379, val: 1.604 | iter time: 359.61 ms (step) remaining time: 0:37:24
|
| 189 |
+
Epoch 1 | iter 2720 step 85 | loss train: 1.482, val: 1.604 | iter time: 360.17 ms (step) remaining time: 0:37:13
|
| 190 |
+
Epoch 1 | iter 2752 step 86 | loss train: 1.417, val: 1.604 | iter time: 358.16 ms (step) remaining time: 0:37:02
|
| 191 |
+
Epoch 1 | iter 2784 step 87 | loss train: 1.367, val: 1.604 | iter time: 359.33 ms (step) remaining time: 0:36:50
|
| 192 |
+
Epoch 1 | iter 2816 step 88 | loss train: 1.409, val: 1.604 | iter time: 357.66 ms (step) remaining time: 0:36:39
|
| 193 |
+
Epoch 1 | iter 2848 step 89 | loss train: 1.381, val: 1.604 | iter time: 359.77 ms (step) remaining time: 0:36:28
|
| 194 |
+
Epoch 1 | iter 2880 step 90 | loss train: 1.395, val: 1.604 | iter time: 357.83 ms (step) remaining time: 0:36:17
|
| 195 |
+
Epoch 1 | iter 2912 step 91 | loss train: 1.470, val: 1.604 | iter time: 358.83 ms (step) remaining time: 0:36:06
|
| 196 |
+
Epoch 1 | iter 2944 step 92 | loss train: 1.402, val: 1.604 | iter time: 359.46 ms (step) remaining time: 0:35:55
|
| 197 |
+
Epoch 1 | iter 2976 step 93 | loss train: 1.374, val: 1.604 | iter time: 360.89 ms (step) remaining time: 0:35:43
|
| 198 |
+
Epoch 1 | iter 3008 step 94 | loss train: 1.374, val: 1.604 | iter time: 359.28 ms (step) remaining time: 0:35:32
|
| 199 |
+
Epoch 1 | iter 3040 step 95 | loss train: 1.398, val: 1.604 | iter time: 359.69 ms (step) remaining time: 0:35:21
|
| 200 |
+
Epoch 1 | iter 3072 step 96 | loss train: 1.415, val: 1.604 | iter time: 358.53 ms (step) remaining time: 0:35:10
|
| 201 |
+
Epoch 1 | iter 3104 step 97 | loss train: 1.340, val: 1.604 | iter time: 359.64 ms (step) remaining time: 0:34:58
|
| 202 |
+
Epoch 1 | iter 3136 step 98 | loss train: 1.433, val: 1.604 | iter time: 359.08 ms (step) remaining time: 0:34:47
|
| 203 |
+
Epoch 1 | iter 3168 step 99 | loss train: 1.429, val: 1.604 | iter time: 361.53 ms (step) remaining time: 0:34:36
|
| 204 |
+
Epoch 1 | iter 3200 step 100 | loss train: 1.471, val: 1.604 | iter time: 359.28 ms (step) remaining time: 0:34:25
|
| 205 |
+
Validating ...
|
| 206 |
+
iter 3200: val loss 1.5566, val time: 11290.77 ms
|
| 207 |
+
Saving checkpoint to 'out/pretrain/2412/step-00000100/lit_model.pth'
|
| 208 |
+
Epoch 1 | iter 3232 step 101 | loss train: 1.348, val: 1.557 | iter time: 355.70 ms (step) remaining time: 0:35:05
|
| 209 |
+
Epoch 1 | iter 3264 step 102 | loss train: 1.418, val: 1.557 | iter time: 359.22 ms (step) remaining time: 0:34:53
|
| 210 |
+
Epoch 1 | iter 3296 step 103 | loss train: 1.400, val: 1.557 | iter time: 360.08 ms (step) remaining time: 0:34:41
|
| 211 |
+
Epoch 1 | iter 3328 step 104 | loss train: 1.394, val: 1.557 | iter time: 360.18 ms (step) remaining time: 0:34:29
|
| 212 |
+
Epoch 1 | iter 3360 step 105 | loss train: 1.484, val: 1.557 | iter time: 359.43 ms (step) remaining time: 0:34:17
|
| 213 |
+
Epoch 1 | iter 3392 step 106 | loss train: 1.481, val: 1.557 | iter time: 359.06 ms (step) remaining time: 0:34:05
|
| 214 |
+
Epoch 1 | iter 3424 step 107 | loss train: 1.436, val: 1.557 | iter time: 358.11 ms (step) remaining time: 0:33:53
|
| 215 |
+
Epoch 1 | iter 3456 step 108 | loss train: 1.422, val: 1.557 | iter time: 357.37 ms (step) remaining time: 0:33:41
|
| 216 |
+
Epoch 1 | iter 3488 step 109 | loss train: 1.392, val: 1.557 | iter time: 362.25 ms (step) remaining time: 0:33:30
|
| 217 |
+
Epoch 1 | iter 3520 step 110 | loss train: 1.430, val: 1.557 | iter time: 358.26 ms (step) remaining time: 0:33:18
|
| 218 |
+
Epoch 1 | iter 3552 step 111 | loss train: 1.469, val: 1.557 | iter time: 358.26 ms (step) remaining time: 0:33:06
|
| 219 |
+
Epoch 1 | iter 3584 step 112 | loss train: 1.414, val: 1.557 | iter time: 359.00 ms (step) remaining time: 0:32:54
|
| 220 |
+
Epoch 1 | iter 3616 step 113 | loss train: 1.365, val: 1.557 | iter time: 359.17 ms (step) remaining time: 0:32:42
|
| 221 |
+
Epoch 1 | iter 3648 step 114 | loss train: 1.386, val: 1.557 | iter time: 358.07 ms (step) remaining time: 0:32:31
|
| 222 |
+
Epoch 1 | iter 3680 step 115 | loss train: 1.385, val: 1.557 | iter time: 360.50 ms (step) remaining time: 0:32:19
|
| 223 |
+
Epoch 1 | iter 3712 step 116 | loss train: 1.406, val: 1.557 | iter time: 360.24 ms (step) remaining time: 0:32:07
|
| 224 |
+
Epoch 1 | iter 3744 step 117 | loss train: 1.431, val: 1.557 | iter time: 360.89 ms (step) remaining time: 0:31:55
|
| 225 |
+
Epoch 1 | iter 3776 step 118 | loss train: 1.428, val: 1.557 | iter time: 358.67 ms (step) remaining time: 0:31:44
|
| 226 |
+
Epoch 1 | iter 3808 step 119 | loss train: 1.400, val: 1.557 | iter time: 359.95 ms (step) remaining time: 0:31:32
|
| 227 |
+
Epoch 1 | iter 3840 step 120 | loss train: 1.331, val: 1.557 | iter time: 358.79 ms (step) remaining time: 0:31:20
|
| 228 |
+
Epoch 1 | iter 3872 step 121 | loss train: 1.394, val: 1.557 | iter time: 360.93 ms (step) remaining time: 0:31:09
|
| 229 |
+
Epoch 1 | iter 3904 step 122 | loss train: 1.409, val: 1.557 | iter time: 360.29 ms (step) remaining time: 0:30:57
|
| 230 |
+
Epoch 1 | iter 3936 step 123 | loss train: 1.389, val: 1.557 | iter time: 358.96 ms (step) remaining time: 0:30:45
|
| 231 |
+
Epoch 1 | iter 3968 step 124 | loss train: 1.381, val: 1.557 | iter time: 360.04 ms (step) remaining time: 0:30:34
|
| 232 |
+
Epoch 1 | iter 4000 step 125 | loss train: 1.441, val: 1.557 | iter time: 359.37 ms (step) remaining time: 0:30:22
|
| 233 |
+
Epoch 1 | iter 4032 step 126 | loss train: 1.344, val: 1.557 | iter time: 358.96 ms (step) remaining time: 0:30:11
|
| 234 |
+
Epoch 1 | iter 4064 step 127 | loss train: 1.416, val: 1.557 | iter time: 359.99 ms (step) remaining time: 0:29:59
|
| 235 |
+
Epoch 1 | iter 4096 step 128 | loss train: 1.417, val: 1.557 | iter time: 359.98 ms (step) remaining time: 0:29:47
|
| 236 |
+
Epoch 1 | iter 4128 step 129 | loss train: 1.412, val: 1.557 | iter time: 359.21 ms (step) remaining time: 0:29:36
|
| 237 |
+
Epoch 1 | iter 4160 step 130 | loss train: 1.436, val: 1.557 | iter time: 357.40 ms (step) remaining time: 0:29:24
|
| 238 |
+
Epoch 1 | iter 4192 step 131 | loss train: 1.406, val: 1.557 | iter time: 359.45 ms (step) remaining time: 0:29:13
|
| 239 |
+
Epoch 1 | iter 4224 step 132 | loss train: 1.419, val: 1.557 | iter time: 358.23 ms (step) remaining time: 0:29:01
|
| 240 |
+
Epoch 1 | iter 4256 step 133 | loss train: 1.460, val: 1.557 | iter time: 359.71 ms (step) remaining time: 0:28:50
|
| 241 |
+
Epoch 1 | iter 4288 step 134 | loss train: 1.408, val: 1.557 | iter time: 359.28 ms (step) remaining time: 0:28:38
|
| 242 |
+
Epoch 1 | iter 4320 step 135 | loss train: 1.436, val: 1.557 | iter time: 357.52 ms (step) remaining time: 0:28:27
|
| 243 |
+
Epoch 1 | iter 4352 step 136 | loss train: 1.368, val: 1.557 | iter time: 360.51 ms (step) remaining time: 0:28:15
|
| 244 |
+
Epoch 1 | iter 4384 step 137 | loss train: 1.351, val: 1.557 | iter time: 359.69 ms (step) remaining time: 0:28:04
|
| 245 |
+
Epoch 1 | iter 4416 step 138 | loss train: 1.391, val: 1.557 | iter time: 359.32 ms (step) remaining time: 0:27:52
|
| 246 |
+
Epoch 1 | iter 4448 step 139 | loss train: 1.408, val: 1.557 | iter time: 360.94 ms (step) remaining time: 0:27:41
|
| 247 |
+
Epoch 1 | iter 4480 step 140 | loss train: 1.455, val: 1.557 | iter time: 360.43 ms (step) remaining time: 0:27:29
|
| 248 |
+
Epoch 1 | iter 4512 step 141 | loss train: 1.324, val: 1.557 | iter time: 358.18 ms (step) remaining time: 0:27:18
|
| 249 |
+
Epoch 1 | iter 4544 step 142 | loss train: 1.365, val: 1.557 | iter time: 358.46 ms (step) remaining time: 0:27:07
|
| 250 |
+
Epoch 1 | iter 4576 step 143 | loss train: 1.423, val: 1.557 | iter time: 358.26 ms (step) remaining time: 0:26:56
|
| 251 |
+
Epoch 1 | iter 4608 step 144 | loss train: 1.355, val: 1.557 | iter time: 357.67 ms (step) remaining time: 0:26:44
|
| 252 |
+
Epoch 1 | iter 4640 step 145 | loss train: 1.326, val: 1.557 | iter time: 360.79 ms (step) remaining time: 0:26:33
|
| 253 |
+
Epoch 1 | iter 4672 step 146 | loss train: 1.410, val: 1.557 | iter time: 361.28 ms (step) remaining time: 0:26:21
|
| 254 |
+
Epoch 1 | iter 4704 step 147 | loss train: 1.376, val: 1.557 | iter time: 359.61 ms (step) remaining time: 0:26:10
|
| 255 |
+
Epoch 1 | iter 4736 step 148 | loss train: 1.410, val: 1.557 | iter time: 358.55 ms (step) remaining time: 0:25:58
|
| 256 |
+
Epoch 1 | iter 4768 step 149 | loss train: 1.353, val: 1.557 | iter time: 357.80 ms (step) remaining time: 0:25:47
|
| 257 |
+
Epoch 1 | iter 4800 step 150 | loss train: 1.361, val: 1.557 | iter time: 360.19 ms (step) remaining time: 0:25:36
|
| 258 |
+
Validating ...
|
| 259 |
+
iter 4800: val loss 1.5478, val time: 11291.47 ms
|
| 260 |
+
Epoch 1 | iter 4832 step 151 | loss train: 1.345, val: 1.548 | iter time: 359.93 ms (step) remaining time: 0:25:35
|
| 261 |
+
Epoch 1 | iter 4864 step 152 | loss train: 1.329, val: 1.548 | iter time: 361.09 ms (step) remaining time: 0:25:23
|
| 262 |
+
Epoch 1 | iter 4896 step 153 | loss train: 1.380, val: 1.548 | iter time: 359.32 ms (step) remaining time: 0:25:12
|
| 263 |
+
Epoch 1 | iter 4928 step 154 | loss train: 1.323, val: 1.548 | iter time: 360.58 ms (step) remaining time: 0:25:00
|
| 264 |
+
Epoch 1 | iter 4960 step 155 | loss train: 1.432, val: 1.548 | iter time: 360.24 ms (step) remaining time: 0:24:49
|
| 265 |
+
Epoch 1 | iter 4992 step 156 | loss train: 1.333, val: 1.548 | iter time: 358.40 ms (step) remaining time: 0:24:37
|
| 266 |
+
Epoch 1 | iter 5024 step 157 | loss train: 1.420, val: 1.548 | iter time: 359.48 ms (step) remaining time: 0:24:26
|
| 267 |
+
Epoch 1 | iter 5056 step 158 | loss train: 1.411, val: 1.548 | iter time: 359.26 ms (step) remaining time: 0:24:14
|
| 268 |
+
Epoch 1 | iter 5088 step 159 | loss train: 1.357, val: 1.548 | iter time: 356.81 ms (step) remaining time: 0:24:03
|
| 269 |
+
Epoch 1 | iter 5120 step 160 | loss train: 1.417, val: 1.548 | iter time: 360.77 ms (step) remaining time: 0:23:51
|
| 270 |
+
Epoch 1 | iter 5152 step 161 | loss train: 1.360, val: 1.548 | iter time: 358.44 ms (step) remaining time: 0:23:40
|
| 271 |
+
Epoch 1 | iter 5184 step 162 | loss train: 1.403, val: 1.548 | iter time: 359.83 ms (step) remaining time: 0:23:28
|
| 272 |
+
Epoch 1 | iter 5216 step 163 | loss train: 1.412, val: 1.548 | iter time: 359.03 ms (step) remaining time: 0:23:17
|
| 273 |
+
Epoch 1 | iter 5248 step 164 | loss train: 1.378, val: 1.548 | iter time: 361.43 ms (step) remaining time: 0:23:05
|
| 274 |
+
Epoch 1 | iter 5280 step 165 | loss train: 1.328, val: 1.548 | iter time: 358.02 ms (step) remaining time: 0:22:54
|
| 275 |
+
Epoch 1 | iter 5312 step 166 | loss train: 1.311, val: 1.548 | iter time: 359.26 ms (step) remaining time: 0:22:43
|
| 276 |
+
Epoch 1 | iter 5344 step 167 | loss train: 1.428, val: 1.548 | iter time: 359.76 ms (step) remaining time: 0:22:31
|
| 277 |
+
Epoch 1 | iter 5376 step 168 | loss train: 1.457, val: 1.548 | iter time: 361.20 ms (step) remaining time: 0:22:20
|
| 278 |
+
Epoch 1 | iter 5408 step 169 | loss train: 1.413, val: 1.548 | iter time: 358.80 ms (step) remaining time: 0:22:08
|
| 279 |
+
Epoch 1 | iter 5440 step 170 | loss train: 1.322, val: 1.548 | iter time: 359.81 ms (step) remaining time: 0:21:57
|
| 280 |
+
Epoch 1 | iter 5472 step 171 | loss train: 1.370, val: 1.548 | iter time: 358.45 ms (step) remaining time: 0:21:46
|
| 281 |
+
Epoch 1 | iter 5504 step 172 | loss train: 1.416, val: 1.548 | iter time: 360.15 ms (step) remaining time: 0:21:34
|
| 282 |
+
Epoch 1 | iter 5536 step 173 | loss train: 1.394, val: 1.548 | iter time: 360.54 ms (step) remaining time: 0:21:23
|
| 283 |
+
Epoch 1 | iter 5568 step 174 | loss train: 1.362, val: 1.548 | iter time: 358.27 ms (step) remaining time: 0:21:12
|
| 284 |
+
Epoch 1 | iter 5600 step 175 | loss train: 1.391, val: 1.548 | iter time: 360.58 ms (step) remaining time: 0:21:00
|
| 285 |
+
Epoch 1 | iter 5632 step 176 | loss train: 1.385, val: 1.548 | iter time: 358.95 ms (step) remaining time: 0:20:49
|
| 286 |
+
Epoch 1 | iter 5664 step 177 | loss train: 1.388, val: 1.548 | iter time: 359.80 ms (step) remaining time: 0:20:38
|
| 287 |
+
Epoch 1 | iter 5696 step 178 | loss train: 1.350, val: 1.548 | iter time: 359.06 ms (step) remaining time: 0:20:26
|
| 288 |
+
Epoch 1 | iter 5728 step 179 | loss train: 1.408, val: 1.548 | iter time: 357.48 ms (step) remaining time: 0:20:15
|
| 289 |
+
Epoch 1 | iter 5760 step 180 | loss train: 1.395, val: 1.548 | iter time: 359.03 ms (step) remaining time: 0:20:04
|
| 290 |
+
Epoch 1 | iter 5792 step 181 | loss train: 1.325, val: 1.548 | iter time: 359.54 ms (step) remaining time: 0:19:52
|
| 291 |
+
Epoch 1 | iter 5824 step 182 | loss train: 1.400, val: 1.548 | iter time: 359.30 ms (step) remaining time: 0:19:41
|
| 292 |
+
Epoch 1 | iter 5856 step 183 | loss train: 1.367, val: 1.548 | iter time: 360.67 ms (step) remaining time: 0:19:30
|
| 293 |
+
Epoch 1 | iter 5888 step 184 | loss train: 1.324, val: 1.548 | iter time: 360.37 ms (step) remaining time: 0:19:18
|
| 294 |
+
Epoch 1 | iter 5920 step 185 | loss train: 1.377, val: 1.548 | iter time: 361.38 ms (step) remaining time: 0:19:07
|
| 295 |
+
Epoch 1 | iter 5952 step 186 | loss train: 1.345, val: 1.548 | iter time: 360.53 ms (step) remaining time: 0:18:56
|
| 296 |
+
Epoch 1 | iter 5984 step 187 | loss train: 1.424, val: 1.548 | iter time: 357.06 ms (step) remaining time: 0:18:45
|
| 297 |
+
Epoch 1 | iter 6016 step 188 | loss train: 1.313, val: 1.548 | iter time: 361.73 ms (step) remaining time: 0:18:33
|
| 298 |
+
Epoch 1 | iter 6048 step 189 | loss train: 1.412, val: 1.548 | iter time: 359.61 ms (step) remaining time: 0:18:22
|
| 299 |
+
Epoch 1 | iter 6080 step 190 | loss train: 1.355, val: 1.548 | iter time: 360.04 ms (step) remaining time: 0:18:11
|
| 300 |
+
Epoch 1 | iter 6112 step 191 | loss train: 1.322, val: 1.548 | iter time: 359.21 ms (step) remaining time: 0:17:59
|
| 301 |
+
Epoch 1 | iter 6144 step 192 | loss train: 1.365, val: 1.548 | iter time: 359.09 ms (step) remaining time: 0:17:48
|
| 302 |
+
Epoch 1 | iter 6176 step 193 | loss train: 1.329, val: 1.548 | iter time: 357.82 ms (step) remaining time: 0:17:37
|
| 303 |
+
Epoch 1 | iter 6208 step 194 | loss train: 1.402, val: 1.548 | iter time: 360.41 ms (step) remaining time: 0:17:26
|
| 304 |
+
Epoch 1 | iter 6240 step 195 | loss train: 1.307, val: 1.548 | iter time: 358.97 ms (step) remaining time: 0:17:15
|
| 305 |
+
Epoch 1 | iter 6272 step 196 | loss train: 1.265, val: 1.548 | iter time: 358.39 ms (step) remaining time: 0:17:03
|
| 306 |
+
Epoch 1 | iter 6304 step 197 | loss train: 1.385, val: 1.548 | iter time: 358.00 ms (step) remaining time: 0:16:52
|
| 307 |
+
Epoch 1 | iter 6336 step 198 | loss train: 1.375, val: 1.548 | iter time: 358.06 ms (step) remaining time: 0:16:41
|
| 308 |
+
Epoch 1 | iter 6368 step 199 | loss train: 1.381, val: 1.548 | iter time: 360.08 ms (step) remaining time: 0:16:30
|
| 309 |
+
Epoch 1 | iter 6400 step 200 | loss train: 1.368, val: 1.548 | iter time: 361.09 ms (step) remaining time: 0:16:18
|
| 310 |
+
Validating ...
|
| 311 |
+
iter 6400: val loss 1.5167, val time: 11312.35 ms
|
| 312 |
+
Saving checkpoint to 'out/pretrain/2412/step-00000200/lit_model.pth'
|
| 313 |
+
Epoch 1 | iter 6432 step 201 | loss train: 1.359, val: 1.517 | iter time: 357.64 ms (step) remaining time: 0:16:19
|
| 314 |
+
Epoch 1 | iter 6464 step 202 | loss train: 1.352, val: 1.517 | iter time: 357.31 ms (step) remaining time: 0:16:08
|
| 315 |
+
Epoch 1 | iter 6496 step 203 | loss train: 1.301, val: 1.517 | iter time: 357.70 ms (step) remaining time: 0:15:56
|
| 316 |
+
Epoch 1 | iter 6528 step 204 | loss train: 1.306, val: 1.517 | iter time: 357.61 ms (step) remaining time: 0:15:45
|
| 317 |
+
Epoch 1 | iter 6560 step 205 | loss train: 1.393, val: 1.517 | iter time: 367.74 ms (step) remaining time: 0:15:33
|
| 318 |
+
Epoch 1 | iter 6592 step 206 | loss train: 1.352, val: 1.517 | iter time: 358.15 ms (step) remaining time: 0:15:22
|
| 319 |
+
Epoch 1 | iter 6624 step 207 | loss train: 1.411, val: 1.517 | iter time: 358.99 ms (step) remaining time: 0:15:11
|
| 320 |
+
Epoch 1 | iter 6656 step 208 | loss train: 1.366, val: 1.517 | iter time: 358.44 ms (step) remaining time: 0:14:59
|
| 321 |
+
Epoch 1 | iter 6688 step 209 | loss train: 1.313, val: 1.517 | iter time: 358.03 ms (step) remaining time: 0:14:48
|
| 322 |
+
Epoch 1 | iter 6720 step 210 | loss train: 1.350, val: 1.517 | iter time: 359.49 ms (step) remaining time: 0:14:36
|
| 323 |
+
Epoch 1 | iter 6752 step 211 | loss train: 1.347, val: 1.517 | iter time: 359.26 ms (step) remaining time: 0:14:25
|
| 324 |
+
Epoch 1 | iter 6784 step 212 | loss train: 1.426, val: 1.517 | iter time: 359.41 ms (step) remaining time: 0:14:14
|
| 325 |
+
Epoch 1 | iter 6816 step 213 | loss train: 1.344, val: 1.517 | iter time: 361.01 ms (step) remaining time: 0:14:02
|
| 326 |
+
Epoch 1 | iter 6848 step 214 | loss train: 1.337, val: 1.517 | iter time: 357.72 ms (step) remaining time: 0:13:51
|
| 327 |
+
Epoch 1 | iter 6880 step 215 | loss train: 1.410, val: 1.517 | iter time: 358.95 ms (step) remaining time: 0:13:40
|
| 328 |
+
Epoch 1 | iter 6912 step 216 | loss train: 1.297, val: 1.517 | iter time: 359.36 ms (step) remaining time: 0:13:28
|
| 329 |
+
Epoch 1 | iter 6944 step 217 | loss train: 1.370, val: 1.517 | iter time: 359.56 ms (step) remaining time: 0:13:17
|
| 330 |
+
Epoch 1 | iter 6976 step 218 | loss train: 1.322, val: 1.517 | iter time: 361.88 ms (step) remaining time: 0:13:05
|
| 331 |
+
Epoch 1 | iter 7008 step 219 | loss train: 1.346, val: 1.517 | iter time: 359.38 ms (step) remaining time: 0:12:54
|
| 332 |
+
Epoch 1 | iter 7040 step 220 | loss train: 1.328, val: 1.517 | iter time: 361.68 ms (step) remaining time: 0:12:43
|
| 333 |
+
Epoch 1 | iter 7072 step 221 | loss train: 1.312, val: 1.517 | iter time: 359.43 ms (step) remaining time: 0:12:31
|
| 334 |
+
Epoch 1 | iter 7104 step 222 | loss train: 1.336, val: 1.517 | iter time: 358.97 ms (step) remaining time: 0:12:20
|
| 335 |
+
Epoch 1 | iter 7136 step 223 | loss train: 1.351, val: 1.517 | iter time: 358.65 ms (step) remaining time: 0:12:09
|
| 336 |
+
Epoch 1 | iter 7168 step 224 | loss train: 1.311, val: 1.517 | iter time: 359.52 ms (step) remaining time: 0:11:57
|
| 337 |
+
Epoch 1 | iter 7200 step 225 | loss train: 1.326, val: 1.517 | iter time: 359.56 ms (step) remaining time: 0:11:46
|
| 338 |
+
Epoch 1 | iter 7232 step 226 | loss train: 1.391, val: 1.517 | iter time: 357.92 ms (step) remaining time: 0:11:35
|
| 339 |
+
Epoch 1 | iter 7264 step 227 | loss train: 1.327, val: 1.517 | iter time: 361.14 ms (step) remaining time: 0:11:24
|
| 340 |
+
Epoch 1 | iter 7296 step 228 | loss train: 1.391, val: 1.517 | iter time: 360.86 ms (step) remaining time: 0:11:12
|
| 341 |
+
Epoch 1 | iter 7328 step 229 | loss train: 1.355, val: 1.517 | iter time: 360.88 ms (step) remaining time: 0:11:01
|
| 342 |
+
Epoch 1 | iter 7360 step 230 | loss train: 1.387, val: 1.517 | iter time: 360.25 ms (step) remaining time: 0:10:50
|
| 343 |
+
Epoch 1 | iter 7392 step 231 | loss train: 1.347, val: 1.517 | iter time: 360.12 ms (step) remaining time: 0:10:38
|
| 344 |
+
Epoch 1 | iter 7424 step 232 | loss train: 1.324, val: 1.517 | iter time: 359.50 ms (step) remaining time: 0:10:27
|
| 345 |
+
Epoch 1 | iter 7456 step 233 | loss train: 1.345, val: 1.517 | iter time: 361.46 ms (step) remaining time: 0:10:16
|
| 346 |
+
Epoch 1 | iter 7488 step 234 | loss train: 1.301, val: 1.517 | iter time: 360.08 ms (step) remaining time: 0:10:04
|
| 347 |
+
Epoch 1 | iter 7520 step 235 | loss train: 1.383, val: 1.517 | iter time: 360.12 ms (step) remaining time: 0:09:53
|
| 348 |
+
Epoch 1 | iter 7552 step 236 | loss train: 1.362, val: 1.517 | iter time: 359.22 ms (step) remaining time: 0:09:42
|
| 349 |
+
Epoch 1 | iter 7584 step 237 | loss train: 1.446, val: 1.517 | iter time: 358.76 ms (step) remaining time: 0:09:31
|
| 350 |
+
Epoch 1 | iter 7616 step 238 | loss train: 1.408, val: 1.517 | iter time: 358.99 ms (step) remaining time: 0:09:19
|
| 351 |
+
Epoch 1 | iter 7648 step 239 | loss train: 1.319, val: 1.517 | iter time: 357.99 ms (step) remaining time: 0:09:08
|
| 352 |
+
Epoch 1 | iter 7680 step 240 | loss train: 1.305, val: 1.517 | iter time: 359.95 ms (step) remaining time: 0:08:57
|
| 353 |
+
Epoch 1 | iter 7712 step 241 | loss train: 1.373, val: 1.517 | iter time: 360.20 ms (step) remaining time: 0:08:46
|
| 354 |
+
Epoch 1 | iter 7744 step 242 | loss train: 1.302, val: 1.517 | iter time: 358.69 ms (step) remaining time: 0:08:34
|
| 355 |
+
Epoch 1 | iter 7776 step 243 | loss train: 1.306, val: 1.517 | iter time: 360.55 ms (step) remaining time: 0:08:23
|
| 356 |
+
Epoch 1 | iter 7808 step 244 | loss train: 1.351, val: 1.517 | iter time: 361.13 ms (step) remaining time: 0:08:12
|
| 357 |
+
Epoch 1 | iter 7840 step 245 | loss train: 1.318, val: 1.517 | iter time: 358.61 ms (step) remaining time: 0:08:01
|
| 358 |
+
Epoch 1 | iter 7872 step 246 | loss train: 1.363, val: 1.517 | iter time: 359.24 ms (step) remaining time: 0:07:49
|
| 359 |
+
Epoch 1 | iter 7904 step 247 | loss train: 1.371, val: 1.517 | iter time: 361.89 ms (step) remaining time: 0:07:38
|
| 360 |
+
Epoch 1 | iter 7936 step 248 | loss train: 1.304, val: 1.517 | iter time: 361.06 ms (step) remaining time: 0:07:27
|
| 361 |
+
Epoch 1 | iter 7968 step 249 | loss train: 1.301, val: 1.517 | iter time: 359.34 ms (step) remaining time: 0:07:16
|
| 362 |
+
Epoch 1 | iter 8000 step 250 | loss train: 1.362, val: 1.517 | iter time: 360.73 ms (step) remaining time: 0:07:04
|
| 363 |
+
Validating ...
|
| 364 |
+
iter 8000: val loss 1.5129, val time: 11296.47 ms
|
| 365 |
+
Epoch 1 | iter 8032 step 251 | loss train: 1.402, val: 1.513 | iter time: 358.22 ms (step) remaining time: 0:06:55
|
| 366 |
+
Epoch 1 | iter 8064 step 252 | loss train: 1.334, val: 1.513 | iter time: 358.87 ms (step) remaining time: 0:06:44
|
| 367 |
+
Epoch 1 | iter 8096 step 253 | loss train: 1.387, val: 1.513 | iter time: 360.21 ms (step) remaining time: 0:06:32
|
| 368 |
+
Epoch 1 | iter 8128 step 254 | loss train: 1.346, val: 1.513 | iter time: 358.89 ms (step) remaining time: 0:06:21
|
| 369 |
+
Epoch 1 | iter 8160 step 255 | loss train: 1.288, val: 1.513 | iter time: 359.06 ms (step) remaining time: 0:06:10
|
| 370 |
+
Epoch 1 | iter 8192 step 256 | loss train: 1.320, val: 1.513 | iter time: 357.70 ms (step) remaining time: 0:05:59
|
| 371 |
+
Epoch 1 | iter 8224 step 257 | loss train: 1.288, val: 1.513 | iter time: 359.90 ms (step) remaining time: 0:05:47
|
| 372 |
+
Epoch 1 | iter 8256 step 258 | loss train: 1.400, val: 1.513 | iter time: 358.76 ms (step) remaining time: 0:05:36
|
| 373 |
+
Epoch 1 | iter 8288 step 259 | loss train: 1.351, val: 1.513 | iter time: 358.20 ms (step) remaining time: 0:05:25
|
| 374 |
+
Epoch 1 | iter 8320 step 260 | loss train: 1.350, val: 1.513 | iter time: 358.80 ms (step) remaining time: 0:05:14
|
| 375 |
+
Epoch 1 | iter 8352 step 261 | loss train: 1.428, val: 1.513 | iter time: 359.16 ms (step) remaining time: 0:05:02
|
| 376 |
+
Epoch 1 | iter 8384 step 262 | loss train: 1.390, val: 1.513 | iter time: 360.59 ms (step) remaining time: 0:04:51
|
| 377 |
+
Epoch 1 | iter 8416 step 263 | loss train: 1.396, val: 1.513 | iter time: 358.98 ms (step) remaining time: 0:04:40
|
| 378 |
+
Epoch 1 | iter 8448 step 264 | loss train: 1.297, val: 1.513 | iter time: 360.47 ms (step) remaining time: 0:04:29
|
| 379 |
+
Epoch 1 | iter 8480 step 265 | loss train: 1.331, val: 1.513 | iter time: 358.98 ms (step) remaining time: 0:04:17
|
| 380 |
+
Epoch 1 | iter 8512 step 266 | loss train: 1.341, val: 1.513 | iter time: 360.22 ms (step) remaining time: 0:04:06
|
| 381 |
+
Epoch 1 | iter 8544 step 267 | loss train: 1.339, val: 1.513 | iter time: 360.93 ms (step) remaining time: 0:03:55
|
| 382 |
+
Epoch 1 | iter 8576 step 268 | loss train: 1.369, val: 1.513 | iter time: 359.87 ms (step) remaining time: 0:03:44
|
| 383 |
+
Epoch 1 | iter 8608 step 269 | loss train: 1.329, val: 1.513 | iter time: 359.81 ms (step) remaining time: 0:03:32
|
| 384 |
+
Epoch 1 | iter 8640 step 270 | loss train: 1.325, val: 1.513 | iter time: 359.98 ms (step) remaining time: 0:03:21
|
| 385 |
+
Epoch 1 | iter 8672 step 271 | loss train: 1.344, val: 1.513 | iter time: 358.67 ms (step) remaining time: 0:03:10
|
| 386 |
+
Epoch 1 | iter 8704 step 272 | loss train: 1.380, val: 1.513 | iter time: 359.40 ms (step) remaining time: 0:02:59
|
| 387 |
+
Epoch 1 | iter 8736 step 273 | loss train: 1.355, val: 1.513 | iter time: 359.79 ms (step) remaining time: 0:02:47
|
| 388 |
+
Epoch 1 | iter 8768 step 274 | loss train: 1.335, val: 1.513 | iter time: 357.99 ms (step) remaining time: 0:02:36
|
| 389 |
+
Epoch 1 | iter 8800 step 275 | loss train: 1.354, val: 1.513 | iter time: 360.61 ms (step) remaining time: 0:02:25
|
| 390 |
+
Epoch 1 | iter 8832 step 276 | loss train: 1.368, val: 1.513 | iter time: 360.65 ms (step) remaining time: 0:02:14
|
| 391 |
+
Epoch 1 | iter 8864 step 277 | loss train: 1.358, val: 1.513 | iter time: 358.75 ms (step) remaining time: 0:02:03
|
| 392 |
+
Epoch 1 | iter 8896 step 278 | loss train: 1.335, val: 1.513 | iter time: 360.56 ms (step) remaining time: 0:01:51
|
| 393 |
+
Epoch 1 | iter 8928 step 279 | loss train: 1.334, val: 1.513 | iter time: 359.23 ms (step) remaining time: 0:01:40
|
| 394 |
+
Epoch 1 | iter 8960 step 280 | loss train: 1.324, val: 1.513 | iter time: 360.47 ms (step) remaining time: 0:01:29
|
| 395 |
+
Epoch 1 | iter 8992 step 281 | loss train: 1.363, val: 1.513 | iter time: 360.08 ms (step) remaining time: 0:01:18
|
| 396 |
+
Epoch 1 | iter 9024 step 282 | loss train: 1.341, val: 1.513 | iter time: 359.93 ms (step) remaining time: 0:01:07
|
| 397 |
+
Epoch 1 | iter 9056 step 283 | loss train: 1.330, val: 1.513 | iter time: 359.77 ms (step) remaining time: 0:00:55
|
| 398 |
+
Epoch 1 | iter 9088 step 284 | loss train: 1.378, val: 1.513 | iter time: 359.65 ms (step) remaining time: 0:00:44
|
| 399 |
+
Epoch 1 | iter 9120 step 285 | loss train: 1.394, val: 1.513 | iter time: 360.87 ms (step) remaining time: 0:00:33
|
| 400 |
+
Epoch 1 | iter 9152 step 286 | loss train: 1.375, val: 1.513 | iter time: 358.83 ms (step) remaining time: 0:00:22
|
| 401 |
+
Epoch 1 | iter 9184 step 287 | loss train: 1.343, val: 1.513 | iter time: 358.85 ms (step) remaining time: 0:00:11
|
| 402 |
+
Epoch 1 | iter 9216 step 288 | loss train: 1.378, val: 1.513 | iter time: 357.81 ms (step) remaining time: 0:00:00
|
| 403 |
+
Validating ...
|
| 404 |
+
Final evaluation | val loss: 1.499 | val ppl: 4.477
|
| 405 |
+
Saving checkpoint to 'out/pretrain/2412/final/lit_model.pth'
|
| 406 |
+
----------------------------------------
|
| 407 |
+
| Performance
|
| 408 |
+
| - Total tokens : 301,989,888
|
| 409 |
+
| - Training Time : 3261.91 s
|
| 410 |
+
| - Tok/sec : 269.31 tok/s
|
| 411 |
+
| ----------------------------------------
|
| 412 |
+
| Memory Usage
|
| 413 |
+
| - Memory Used : 26.32 GB
|
| 414 |
+
----------------------------------------
|
out/pretrain/tinyllama/teelogs/2412_full.txt
ADDED
|
@@ -0,0 +1,425 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
Initializing distributed: GLOBAL_RANK: 0, MEMBER: 1/4
|
| 2 |
+
Initializing distributed: GLOBAL_RANK: 1, MEMBER: 2/4
|
| 3 |
+
Initializing distributed: GLOBAL_RANK: 2, MEMBER: 3/4
|
| 4 |
+
[rank: 2] Seed set to 42
|
| 5 |
+
[rank: 1] Seed set to 42
|
| 6 |
+
Initializing distributed: GLOBAL_RANK: 3, MEMBER: 4/4
|
| 7 |
+
[rank: 3] Seed set to 42
|
| 8 |
+
----------------------------------------------------------------------------------------------------
|
| 9 |
+
distributed_backend=nccl
|
| 10 |
+
All distributed processes registered. Starting with 4 processes
|
| 11 |
+
----------------------------------------------------------------------------------------------------
|
| 12 |
+
|
| 13 |
+
[rank: 0] Seed set to 42
|
| 14 |
+
/mnt/data/litgpt/litgpt/pretrain.py:515: UserWarning: A newer version of litdata is available (0.2.52). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 15 |
+
train_dataloader = data.train_dataloader()
|
| 16 |
+
/mnt/data/litgpt/litgpt/pretrain.py:515: UserWarning: A newer version of litdata is available (0.2.52). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 17 |
+
train_dataloader = data.train_dataloader()
|
| 18 |
+
/mnt/data/litgpt/litgpt/pretrain.py:515: UserWarning: A newer version of litdata is available (0.2.52). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 19 |
+
train_dataloader = data.train_dataloader()
|
| 20 |
+
/mnt/data/litgpt/litgpt/pretrain.py:515: UserWarning: A newer version of litdata is available (0.2.52). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 21 |
+
train_dataloader = data.train_dataloader()
|
| 22 |
+
All GPUs are fully connected via NVLink.
|
| 23 |
+
{'data': {'batch_size': 1,
|
| 24 |
+
'data_path': PosixPath('litgpt/data/arxiv'),
|
| 25 |
+
'num_workers': 8,
|
| 26 |
+
'seed': 42,
|
| 27 |
+
'seq_length': 2048,
|
| 28 |
+
'use_starcoder': True},
|
| 29 |
+
'data_dir': PosixPath('litgpt/data/arxiv/2412'),
|
| 30 |
+
'devices': 'auto',
|
| 31 |
+
'eval': {'evaluate_example': 'first',
|
| 32 |
+
'final_validation': True,
|
| 33 |
+
'initial_validation': True,
|
| 34 |
+
'interval': 50,
|
| 35 |
+
'max_iters': 200,
|
| 36 |
+
'max_new_tokens': None},
|
| 37 |
+
'initial_checkpoint_dir': PosixPath('out/pretrain/2411_full/final'),
|
| 38 |
+
'log': {'group': None, 'project': None, 'run': None},
|
| 39 |
+
'logger_name': 'tensorboard',
|
| 40 |
+
'model_config': {'attention_logit_softcapping': None,
|
| 41 |
+
'attention_scores_scalar': None,
|
| 42 |
+
'attn_bias': False,
|
| 43 |
+
'bias': False,
|
| 44 |
+
'block_size': 2048,
|
| 45 |
+
'final_logit_softcapping': None,
|
| 46 |
+
'gelu_approximate': 'none',
|
| 47 |
+
'head_size': 64,
|
| 48 |
+
'hf_config': {'name': 'TinyLlama-1.1B-intermediate-step-1431k-3T',
|
| 49 |
+
'org': 'TinyLlama'},
|
| 50 |
+
'intermediate_size': 5632,
|
| 51 |
+
'lm_head_bias': False,
|
| 52 |
+
'mlp_class_name': 'LLaMAMLP',
|
| 53 |
+
'moe_intermediate_size': None,
|
| 54 |
+
'n_embd': 2048,
|
| 55 |
+
'n_expert': 0,
|
| 56 |
+
'n_expert_per_token': 0,
|
| 57 |
+
'n_head': 32,
|
| 58 |
+
'n_layer': 22,
|
| 59 |
+
'n_query_groups': 4,
|
| 60 |
+
'name': 'tiny-llama-1.1b',
|
| 61 |
+
'norm_1': True,
|
| 62 |
+
'norm_2': True,
|
| 63 |
+
'norm_class_name': 'RMSNorm',
|
| 64 |
+
'norm_eps': 1e-05,
|
| 65 |
+
'norm_qk': False,
|
| 66 |
+
'norm_qk_type': 'default',
|
| 67 |
+
'padded_vocab_size': 32000,
|
| 68 |
+
'padding_multiple': 64,
|
| 69 |
+
'parallel_residual': False,
|
| 70 |
+
'post_attention_norm': False,
|
| 71 |
+
'post_mlp_norm': False,
|
| 72 |
+
'rope_adjustments': None,
|
| 73 |
+
'rope_base': 10000,
|
| 74 |
+
'rope_condense_ratio': 1,
|
| 75 |
+
'rope_indices': None,
|
| 76 |
+
'rope_local_base_freq': None,
|
| 77 |
+
'rotary_percentage': 1.0,
|
| 78 |
+
'scale_embeddings': False,
|
| 79 |
+
'shared_attention_norm': False,
|
| 80 |
+
'sliding_window_indices': None,
|
| 81 |
+
'sliding_window_size': None,
|
| 82 |
+
'vocab_size': 32000},
|
| 83 |
+
'model_name': 'tiny-llama-1.1b',
|
| 84 |
+
'num_nodes': 1,
|
| 85 |
+
'optimizer': "{'class_path': 'torch.optim.AdamW', 'init_args': {'lr': 0.0004, "
|
| 86 |
+
"'weight_decay': 0.1, 'betas': [0.9, 0.95]}}",
|
| 87 |
+
'out_dir': PosixPath('out/pretrain/2412_full'),
|
| 88 |
+
'ppl': False,
|
| 89 |
+
'precision': 'bf16-mixed',
|
| 90 |
+
'resume': False,
|
| 91 |
+
'seed': 42,
|
| 92 |
+
'tokenizer_dir': PosixPath('checkpoints/TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T'),
|
| 93 |
+
'train': {'epochs': None,
|
| 94 |
+
'global_batch_size': 512,
|
| 95 |
+
'log_interval': 1,
|
| 96 |
+
'lr_warmup_fraction': None,
|
| 97 |
+
'lr_warmup_steps': 20,
|
| 98 |
+
'max_norm': 1.0,
|
| 99 |
+
'max_seq_length': 2048,
|
| 100 |
+
'max_steps': None,
|
| 101 |
+
'max_tokens': 304087040,
|
| 102 |
+
'micro_batch_size': 4,
|
| 103 |
+
'min_lr': 4e-05,
|
| 104 |
+
'save_interval': 100,
|
| 105 |
+
'tie_embeddings': None}}
|
| 106 |
+
Time to instantiate model: 0.04 seconds.
|
| 107 |
+
Total parameters: 1,100,048,384
|
| 108 |
+
[fix] out/pretrain/2411_full/final/lit_model.pth 是整包 state，提取 model 权重并原子覆盖...
|
| 109 |
+
[fix] 已覆盖为纯权重: out/pretrain/2411_full/final/lit_model.pth
|
| 110 |
+
Validating ...
|
| 111 |
+
Measured TFLOPs: 239.66
|
| 112 |
+
Epoch 1 | iter 32 step 1 | loss train: 1.341, val: 1.354 | iter time: 542.36 ms (step) remaining time: 0:57:31
|
| 113 |
+
Epoch 1 | iter 64 step 2 | loss train: 1.361, val: 1.354 | iter time: 355.55 ms (step) remaining time: 0:54:33
|
| 114 |
+
Epoch 1 | iter 96 step 3 | loss train: 1.447, val: 1.354 | iter time: 356.62 ms (step) remaining time: 0:53:26
|
| 115 |
+
Epoch 1 | iter 128 step 4 | loss train: 1.433, val: 1.354 | iter time: 355.90 ms (step) remaining time: 0:52:49
|
| 116 |
+
Epoch 1 | iter 160 step 5 | loss train: 1.324, val: 1.354 | iter time: 357.80 ms (step) remaining time: 0:52:22
|
| 117 |
+
Epoch 1 | iter 192 step 6 | loss train: 1.344, val: 1.354 | iter time: 356.75 ms (step) remaining time: 0:52:01
|
| 118 |
+
Epoch 1 | iter 224 step 7 | loss train: 1.360, val: 1.354 | iter time: 357.42 ms (step) remaining time: 0:51:44
|
| 119 |
+
Epoch 1 | iter 256 step 8 | loss train: 1.361, val: 1.354 | iter time: 361.01 ms (step) remaining time: 0:51:29
|
| 120 |
+
Epoch 1 | iter 288 step 9 | loss train: 1.357, val: 1.354 | iter time: 358.85 ms (step) remaining time: 0:51:15
|
| 121 |
+
Epoch 1 | iter 320 step 10 | loss train: 1.414, val: 1.354 | iter time: 360.47 ms (step) remaining time: 0:51:01
|
| 122 |
+
Epoch 1 | iter 352 step 11 | loss train: 1.382, val: 1.354 | iter time: 358.64 ms (step) remaining time: 0:50:48
|
| 123 |
+
Epoch 1 | iter 384 step 12 | loss train: 1.444, val: 1.354 | iter time: 359.98 ms (step) remaining time: 0:50:36
|
| 124 |
+
Epoch 1 | iter 416 step 13 | loss train: 1.399, val: 1.354 | iter time: 360.41 ms (step) remaining time: 0:50:23
|
| 125 |
+
Epoch 1 | iter 448 step 14 | loss train: 1.365, val: 1.354 | iter time: 361.38 ms (step) remaining time: 0:50:12
|
| 126 |
+
Epoch 1 | iter 480 step 15 | loss train: 1.500, val: 1.354 | iter time: 359.77 ms (step) remaining time: 0:50:00
|
| 127 |
+
Epoch 1 | iter 512 step 16 | loss train: 1.432, val: 1.354 | iter time: 358.93 ms (step) remaining time: 0:49:48
|
| 128 |
+
Epoch 1 | iter 544 step 17 | loss train: 1.429, val: 1.354 | iter time: 359.86 ms (step) remaining time: 0:49:36
|
| 129 |
+
Epoch 1 | iter 576 step 18 | loss train: 1.400, val: 1.354 | iter time: 358.84 ms (step) remaining time: 0:49:25
|
| 130 |
+
Epoch 1 | iter 608 step 19 | loss train: 1.399, val: 1.354 | iter time: 358.75 ms (step) remaining time: 0:49:13
|
| 131 |
+
Epoch 1 | iter 640 step 20 | loss train: 1.456, val: 1.354 | iter time: 360.98 ms (step) remaining time: 0:49:02
|
| 132 |
+
Epoch 1 | iter 672 step 21 | loss train: 1.386, val: 1.354 | iter time: 358.72 ms (step) remaining time: 0:48:51
|
| 133 |
+
Epoch 1 | iter 704 step 22 | loss train: 1.462, val: 1.354 | iter time: 359.69 ms (step) remaining time: 0:48:39
|
| 134 |
+
Epoch 1 | iter 736 step 23 | loss train: 1.410, val: 1.354 | iter time: 360.88 ms (step) remaining time: 0:48:28
|
| 135 |
+
Epoch 1 | iter 768 step 24 | loss train: 1.469, val: 1.354 | iter time: 359.34 ms (step) remaining time: 0:48:17
|
| 136 |
+
Epoch 1 | iter 800 step 25 | loss train: 1.425, val: 1.354 | iter time: 359.01 ms (step) remaining time: 0:48:06
|
| 137 |
+
Epoch 1 | iter 832 step 26 | loss train: 1.383, val: 1.354 | iter time: 359.88 ms (step) remaining time: 0:47:54
|
| 138 |
+
Epoch 1 | iter 864 step 27 | loss train: 1.419, val: 1.354 | iter time: 360.99 ms (step) remaining time: 0:47:43
|
| 139 |
+
Epoch 1 | iter 896 step 28 | loss train: 1.440, val: 1.354 | iter time: 359.21 ms (step) remaining time: 0:47:32
|
| 140 |
+
Epoch 1 | iter 928 step 29 | loss train: 1.427, val: 1.354 | iter time: 359.62 ms (step) remaining time: 0:47:21
|
| 141 |
+
Epoch 1 | iter 960 step 30 | loss train: 1.465, val: 1.354 | iter time: 360.69 ms (step) remaining time: 0:47:10
|
| 142 |
+
Epoch 1 | iter 992 step 31 | loss train: 1.429, val: 1.354 | iter time: 359.04 ms (step) remaining time: 0:46:59
|
| 143 |
+
Epoch 1 | iter 1024 step 32 | loss train: 1.435, val: 1.354 | iter time: 359.84 ms (step) remaining time: 0:46:48
|
| 144 |
+
Epoch 1 | iter 1056 step 33 | loss train: 1.395, val: 1.354 | iter time: 360.27 ms (step) remaining time: 0:46:37
|
| 145 |
+
Epoch 1 | iter 1088 step 34 | loss train: 1.379, val: 1.354 | iter time: 358.37 ms (step) remaining time: 0:46:26
|
| 146 |
+
Epoch 1 | iter 1120 step 35 | loss train: 1.468, val: 1.354 | iter time: 360.41 ms (step) remaining time: 0:46:15
|
| 147 |
+
Epoch 1 | iter 1152 step 36 | loss train: 1.427, val: 1.354 | iter time: 360.51 ms (step) remaining time: 0:46:04
|
| 148 |
+
Epoch 1 | iter 1184 step 37 | loss train: 1.396, val: 1.354 | iter time: 358.07 ms (step) remaining time: 0:45:53
|
| 149 |
+
Epoch 1 | iter 1216 step 38 | loss train: 1.408, val: 1.354 | iter time: 359.20 ms (step) remaining time: 0:45:42
|
| 150 |
+
Epoch 1 | iter 1248 step 39 | loss train: 1.386, val: 1.354 | iter time: 360.69 ms (step) remaining time: 0:45:31
|
| 151 |
+
Epoch 1 | iter 1280 step 40 | loss train: 1.398, val: 1.354 | iter time: 360.91 ms (step) remaining time: 0:45:20
|
| 152 |
+
Epoch 1 | iter 1312 step 41 | loss train: 1.401, val: 1.354 | iter time: 358.68 ms (step) remaining time: 0:45:09
|
| 153 |
+
Epoch 1 | iter 1344 step 42 | loss train: 1.448, val: 1.354 | iter time: 359.88 ms (step) remaining time: 0:45:00
|
| 154 |
+
Epoch 1 | iter 1376 step 43 | loss train: 1.456, val: 1.354 | iter time: 360.21 ms (step) remaining time: 0:44:49
|
| 155 |
+
Epoch 1 | iter 1408 step 44 | loss train: 1.469, val: 1.354 | iter time: 360.14 ms (step) remaining time: 0:44:38
|
| 156 |
+
Epoch 1 | iter 1440 step 45 | loss train: 1.411, val: 1.354 | iter time: 359.36 ms (step) remaining time: 0:44:26
|
| 157 |
+
Epoch 1 | iter 1472 step 46 | loss train: 1.456, val: 1.354 | iter time: 358.13 ms (step) remaining time: 0:44:15
|
| 158 |
+
Epoch 1 | iter 1504 step 47 | loss train: 1.457, val: 1.354 | iter time: 359.10 ms (step) remaining time: 0:44:04
|
| 159 |
+
Epoch 1 | iter 1536 step 48 | loss train: 1.378, val: 1.354 | iter time: 359.98 ms (step) remaining time: 0:43:53
|
| 160 |
+
Epoch 1 | iter 1568 step 49 | loss train: 1.496, val: 1.354 | iter time: 360.10 ms (step) remaining time: 0:43:42
|
| 161 |
+
Epoch 1 | iter 1600 step 50 | loss train: 1.412, val: 1.354 | iter time: 359.07 ms (step) remaining time: 0:43:31
|
| 162 |
+
Validating ...
|
| 163 |
+
iter 1600: val loss 1.3390, val time: 22296.02 ms
|
| 164 |
+
Epoch 1 | iter 1632 step 51 | loss train: 1.401, val: 1.339 | iter time: 360.83 ms (step) remaining time: 0:45:07
|
| 165 |
+
Epoch 1 | iter 1664 step 52 | loss train: 1.461, val: 1.339 | iter time: 359.43 ms (step) remaining time: 0:44:54
|
| 166 |
+
Epoch 1 | iter 1696 step 53 | loss train: 1.402, val: 1.339 | iter time: 360.41 ms (step) remaining time: 0:44:40
|
| 167 |
+
Epoch 1 | iter 1728 step 54 | loss train: 1.380, val: 1.339 | iter time: 360.50 ms (step) remaining time: 0:44:27
|
| 168 |
+
Epoch 1 | iter 1760 step 55 | loss train: 1.410, val: 1.339 | iter time: 359.11 ms (step) remaining time: 0:44:14
|
| 169 |
+
Epoch 1 | iter 1792 step 56 | loss train: 1.419, val: 1.339 | iter time: 361.42 ms (step) remaining time: 0:44:01
|
| 170 |
+
Epoch 1 | iter 1824 step 57 | loss train: 1.434, val: 1.339 | iter time: 360.24 ms (step) remaining time: 0:43:48
|
| 171 |
+
Epoch 1 | iter 1856 step 58 | loss train: 1.456, val: 1.339 | iter time: 359.49 ms (step) remaining time: 0:43:35
|
| 172 |
+
Epoch 1 | iter 1888 step 59 | loss train: 1.432, val: 1.339 | iter time: 359.70 ms (step) remaining time: 0:43:22
|
| 173 |
+
Epoch 1 | iter 1920 step 60 | loss train: 1.395, val: 1.339 | iter time: 359.47 ms (step) remaining time: 0:43:09
|
| 174 |
+
Epoch 1 | iter 1952 step 61 | loss train: 1.513, val: 1.339 | iter time: 361.67 ms (step) remaining time: 0:42:56
|
| 175 |
+
Epoch 1 | iter 1984 step 62 | loss train: 1.384, val: 1.339 | iter time: 357.72 ms (step) remaining time: 0:42:44
|
| 176 |
+
Epoch 1 | iter 2016 step 63 | loss train: 1.498, val: 1.339 | iter time: 360.05 ms (step) remaining time: 0:42:31
|
| 177 |
+
Epoch 1 | iter 2048 step 64 | loss train: 1.348, val: 1.339 | iter time: 360.51 ms (step) remaining time: 0:42:18
|
| 178 |
+
Epoch 1 | iter 2080 step 65 | loss train: 1.450, val: 1.339 | iter time: 360.83 ms (step) remaining time: 0:42:06
|
| 179 |
+
Epoch 1 | iter 2112 step 66 | loss train: 1.393, val: 1.339 | iter time: 360.45 ms (step) remaining time: 0:41:53
|
| 180 |
+
Epoch 1 | iter 2144 step 67 | loss train: 1.395, val: 1.339 | iter time: 358.96 ms (step) remaining time: 0:41:41
|
| 181 |
+
Epoch 1 | iter 2176 step 68 | loss train: 1.405, val: 1.339 | iter time: 359.69 ms (step) remaining time: 0:41:29
|
| 182 |
+
Epoch 1 | iter 2208 step 69 | loss train: 1.447, val: 1.339 | iter time: 358.44 ms (step) remaining time: 0:41:16
|
| 183 |
+
Epoch 1 | iter 2240 step 70 | loss train: 1.493, val: 1.339 | iter time: 360.57 ms (step) remaining time: 0:41:04
|
| 184 |
+
Epoch 1 | iter 2272 step 71 | loss train: 1.385, val: 1.339 | iter time: 359.78 ms (step) remaining time: 0:40:52
|
| 185 |
+
Epoch 1 | iter 2304 step 72 | loss train: 1.380, val: 1.339 | iter time: 358.99 ms (step) remaining time: 0:40:39
|
| 186 |
+
Epoch 1 | iter 2336 step 73 | loss train: 1.397, val: 1.339 | iter time: 359.30 ms (step) remaining time: 0:40:27
|
| 187 |
+
Epoch 1 | iter 2368 step 74 | loss train: 1.458, val: 1.339 | iter time: 360.61 ms (step) remaining time: 0:40:15
|
| 188 |
+
Epoch 1 | iter 2400 step 75 | loss train: 1.435, val: 1.339 | iter time: 359.19 ms (step) remaining time: 0:40:03
|
| 189 |
+
Epoch 1 | iter 2432 step 76 | loss train: 1.380, val: 1.339 | iter time: 361.20 ms (step) remaining time: 0:39:51
|
| 190 |
+
Epoch 1 | iter 2464 step 77 | loss train: 1.413, val: 1.339 | iter time: 360.82 ms (step) remaining time: 0:39:39
|
| 191 |
+
Epoch 1 | iter 2496 step 78 | loss train: 1.354, val: 1.339 | iter time: 359.73 ms (step) remaining time: 0:39:27
|
| 192 |
+
Epoch 1 | iter 2528 step 79 | loss train: 1.406, val: 1.339 | iter time: 359.10 ms (step) remaining time: 0:39:15
|
| 193 |
+
Epoch 1 | iter 2560 step 80 | loss train: 1.461, val: 1.339 | iter time: 357.56 ms (step) remaining time: 0:39:03
|
| 194 |
+
Epoch 1 | iter 2592 step 81 | loss train: 1.402, val: 1.339 | iter time: 359.97 ms (step) remaining time: 0:38:51
|
| 195 |
+
Epoch 1 | iter 2624 step 82 | loss train: 1.394, val: 1.339 | iter time: 357.17 ms (step) remaining time: 0:38:39
|
| 196 |
+
Epoch 1 | iter 2656 step 83 | loss train: 1.353, val: 1.339 | iter time: 359.08 ms (step) remaining time: 0:38:27
|
| 197 |
+
Epoch 1 | iter 2688 step 84 | loss train: 1.380, val: 1.339 | iter time: 358.63 ms (step) remaining time: 0:38:15
|
| 198 |
+
Epoch 1 | iter 2720 step 85 | loss train: 1.481, val: 1.339 | iter time: 358.72 ms (step) remaining time: 0:38:03
|
| 199 |
+
Epoch 1 | iter 2752 step 86 | loss train: 1.417, val: 1.339 | iter time: 359.02 ms (step) remaining time: 0:37:52
|
| 200 |
+
Epoch 1 | iter 2784 step 87 | loss train: 1.368, val: 1.339 | iter time: 359.64 ms (step) remaining time: 0:37:40
|
| 201 |
+
Epoch 1 | iter 2816 step 88 | loss train: 1.408, val: 1.339 | iter time: 360.00 ms (step) remaining time: 0:37:28
|
| 202 |
+
Epoch 1 | iter 2848 step 89 | loss train: 1.381, val: 1.339 | iter time: 358.61 ms (step) remaining time: 0:37:16
|
| 203 |
+
Epoch 1 | iter 2880 step 90 | loss train: 1.395, val: 1.339 | iter time: 358.45 ms (step) remaining time: 0:37:05
|
| 204 |
+
Epoch 1 | iter 2912 step 91 | loss train: 1.469, val: 1.339 | iter time: 359.47 ms (step) remaining time: 0:36:53
|
| 205 |
+
Epoch 1 | iter 2944 step 92 | loss train: 1.402, val: 1.339 | iter time: 359.16 ms (step) remaining time: 0:36:41
|
| 206 |
+
Epoch 1 | iter 2976 step 93 | loss train: 1.373, val: 1.339 | iter time: 359.54 ms (step) remaining time: 0:36:30
|
| 207 |
+
Epoch 1 | iter 3008 step 94 | loss train: 1.373, val: 1.339 | iter time: 359.46 ms (step) remaining time: 0:36:18
|
| 208 |
+
Epoch 1 | iter 3040 step 95 | loss train: 1.403, val: 1.339 | iter time: 361.33 ms (step) remaining time: 0:36:06
|
| 209 |
+
Epoch 1 | iter 3072 step 96 | loss train: 1.419, val: 1.339 | iter time: 358.25 ms (step) remaining time: 0:35:55
|
| 210 |
+
Epoch 1 | iter 3104 step 97 | loss train: 1.343, val: 1.339 | iter time: 358.54 ms (step) remaining time: 0:35:43
|
| 211 |
+
Epoch 1 | iter 3136 step 98 | loss train: 1.441, val: 1.339 | iter time: 359.17 ms (step) remaining time: 0:35:31
|
| 212 |
+
Epoch 1 | iter 3168 step 99 | loss train: 1.431, val: 1.339 | iter time: 358.26 ms (step) remaining time: 0:35:20
|
| 213 |
+
Epoch 1 | iter 3200 step 100 | loss train: 1.475, val: 1.339 | iter time: 360.35 ms (step) remaining time: 0:35:08
|
| 214 |
+
Validating ...
|
| 215 |
+
iter 3200: val loss 1.3051, val time: 22338.71 ms
|
| 216 |
+
Saving checkpoint to 'out/pretrain/2412_full/step-00000100/lit_model.pth'
|
| 217 |
+
Epoch 1 | iter 3232 step 101 | loss train: 1.356, val: 1.305 | iter time: 357.25 ms (step) remaining time: 0:36:10
|
| 218 |
+
Epoch 1 | iter 3264 step 102 | loss train: 1.421, val: 1.305 | iter time: 355.99 ms (step) remaining time: 0:35:57
|
| 219 |
+
Epoch 1 | iter 3296 step 103 | loss train: 1.403, val: 1.305 | iter time: 357.58 ms (step) remaining time: 0:35:45
|
| 220 |
+
Epoch 1 | iter 3328 step 104 | loss train: 1.402, val: 1.305 | iter time: 360.22 ms (step) remaining time: 0:35:32
|
| 221 |
+
Epoch 1 | iter 3360 step 105 | loss train: 1.499, val: 1.305 | iter time: 360.35 ms (step) remaining time: 0:35:20
|
| 222 |
+
Epoch 1 | iter 3392 step 106 | loss train: 1.496, val: 1.305 | iter time: 358.34 ms (step) remaining time: 0:35:07
|
| 223 |
+
Epoch 1 | iter 3424 step 107 | loss train: 1.446, val: 1.305 | iter time: 360.35 ms (step) remaining time: 0:34:55
|
| 224 |
+
Epoch 1 | iter 3456 step 108 | loss train: 1.433, val: 1.305 | iter time: 359.43 ms (step) remaining time: 0:34:42
|
| 225 |
+
Epoch 1 | iter 3488 step 109 | loss train: 1.400, val: 1.305 | iter time: 359.45 ms (step) remaining time: 0:34:30
|
| 226 |
+
Epoch 1 | iter 3520 step 110 | loss train: 1.435, val: 1.305 | iter time: 359.26 ms (step) remaining time: 0:34:18
|
| 227 |
+
Epoch 1 | iter 3552 step 111 | loss train: 1.474, val: 1.305 | iter time: 358.92 ms (step) remaining time: 0:34:05
|
| 228 |
+
Epoch 1 | iter 3584 step 112 | loss train: 1.418, val: 1.305 | iter time: 359.54 ms (step) remaining time: 0:33:53
|
| 229 |
+
Epoch 1 | iter 3616 step 113 | loss train: 1.369, val: 1.305 | iter time: 358.52 ms (step) remaining time: 0:33:41
|
| 230 |
+
Epoch 1 | iter 3648 step 114 | loss train: 1.389, val: 1.305 | iter time: 359.86 ms (step) remaining time: 0:33:28
|
| 231 |
+
Epoch 1 | iter 3680 step 115 | loss train: 1.388, val: 1.305 | iter time: 358.30 ms (step) remaining time: 0:33:16
|
| 232 |
+
Epoch 1 | iter 3712 step 116 | loss train: 1.421, val: 1.305 | iter time: 358.77 ms (step) remaining time: 0:33:04
|
| 233 |
+
Epoch 1 | iter 3744 step 117 | loss train: 1.442, val: 1.305 | iter time: 358.14 ms (step) remaining time: 0:32:52
|
| 234 |
+
Epoch 1 | iter 3776 step 118 | loss train: 1.436, val: 1.305 | iter time: 360.16 ms (step) remaining time: 0:32:40
|
| 235 |
+
Epoch 1 | iter 3808 step 119 | loss train: 1.407, val: 1.305 | iter time: 361.65 ms (step) remaining time: 0:32:28
|
| 236 |
+
Epoch 1 | iter 3840 step 120 | loss train: 1.337, val: 1.305 | iter time: 360.64 ms (step) remaining time: 0:32:16
|
| 237 |
+
Epoch 1 | iter 3872 step 121 | loss train: 1.400, val: 1.305 | iter time: 357.63 ms (step) remaining time: 0:32:04
|
| 238 |
+
Epoch 1 | iter 3904 step 122 | loss train: 1.412, val: 1.305 | iter time: 359.94 ms (step) remaining time: 0:31:51
|
| 239 |
+
Epoch 1 | iter 3936 step 123 | loss train: 1.392, val: 1.305 | iter time: 359.11 ms (step) remaining time: 0:31:39
|
| 240 |
+
Epoch 1 | iter 3968 step 124 | loss train: 1.383, val: 1.305 | iter time: 361.00 ms (step) remaining time: 0:31:27
|
| 241 |
+
Epoch 1 | iter 4000 step 125 | loss train: 1.445, val: 1.305 | iter time: 357.56 ms (step) remaining time: 0:31:15
|
| 242 |
+
Epoch 1 | iter 4032 step 126 | loss train: 1.344, val: 1.305 | iter time: 359.05 ms (step) remaining time: 0:31:03
|
| 243 |
+
Epoch 1 | iter 4064 step 127 | loss train: 1.416, val: 1.305 | iter time: 358.36 ms (step) remaining time: 0:30:51
|
| 244 |
+
Epoch 1 | iter 4096 step 128 | loss train: 1.418, val: 1.305 | iter time: 357.78 ms (step) remaining time: 0:30:39
|
| 245 |
+
Epoch 1 | iter 4128 step 129 | loss train: 1.414, val: 1.305 | iter time: 360.90 ms (step) remaining time: 0:30:27
|
| 246 |
+
Epoch 1 | iter 4160 step 130 | loss train: 1.438, val: 1.305 | iter time: 358.08 ms (step) remaining time: 0:30:15
|
| 247 |
+
Epoch 1 | iter 4192 step 131 | loss train: 1.408, val: 1.305 | iter time: 360.55 ms (step) remaining time: 0:30:03
|
| 248 |
+
Epoch 1 | iter 4224 step 132 | loss train: 1.420, val: 1.305 | iter time: 360.67 ms (step) remaining time: 0:29:51
|
| 249 |
+
Epoch 1 | iter 4256 step 133 | loss train: 1.462, val: 1.305 | iter time: 358.36 ms (step) remaining time: 0:29:39
|
| 250 |
+
Epoch 1 | iter 4288 step 134 | loss train: 1.410, val: 1.305 | iter time: 360.03 ms (step) remaining time: 0:29:28
|
| 251 |
+
Epoch 1 | iter 4320 step 135 | loss train: 1.436, val: 1.305 | iter time: 359.46 ms (step) remaining time: 0:29:16
|
| 252 |
+
Epoch 1 | iter 4352 step 136 | loss train: 1.366, val: 1.305 | iter time: 360.06 ms (step) remaining time: 0:29:04
|
| 253 |
+
Epoch 1 | iter 4384 step 137 | loss train: 1.352, val: 1.305 | iter time: 360.91 ms (step) remaining time: 0:28:52
|
| 254 |
+
Epoch 1 | iter 4416 step 138 | loss train: 1.390, val: 1.305 | iter time: 359.05 ms (step) remaining time: 0:28:40
|
| 255 |
+
Epoch 1 | iter 4448 step 139 | loss train: 1.408, val: 1.305 | iter time: 360.61 ms (step) remaining time: 0:28:28
|
| 256 |
+
Epoch 1 | iter 4480 step 140 | loss train: 1.454, val: 1.305 | iter time: 358.86 ms (step) remaining time: 0:28:16
|
| 257 |
+
Epoch 1 | iter 4512 step 141 | loss train: 1.325, val: 1.305 | iter time: 361.05 ms (step) remaining time: 0:28:05
|
| 258 |
+
Epoch 1 | iter 4544 step 142 | loss train: 1.364, val: 1.305 | iter time: 358.21 ms (step) remaining time: 0:27:53
|
| 259 |
+
Epoch 1 | iter 4576 step 143 | loss train: 1.424, val: 1.305 | iter time: 359.12 ms (step) remaining time: 0:27:41
|
| 260 |
+
Epoch 1 | iter 4608 step 144 | loss train: 1.354, val: 1.305 | iter time: 360.74 ms (step) remaining time: 0:27:29
|
| 261 |
+
Epoch 1 | iter 4640 step 145 | loss train: 1.325, val: 1.305 | iter time: 358.66 ms (step) remaining time: 0:27:18
|
| 262 |
+
Epoch 1 | iter 4672 step 146 | loss train: 1.403, val: 1.305 | iter time: 359.61 ms (step) remaining time: 0:27:06
|
| 263 |
+
Epoch 1 | iter 4704 step 147 | loss train: 1.376, val: 1.305 | iter time: 358.40 ms (step) remaining time: 0:26:54
|
| 264 |
+
Epoch 1 | iter 4736 step 148 | loss train: 1.410, val: 1.305 | iter time: 357.32 ms (step) remaining time: 0:26:42
|
| 265 |
+
Epoch 1 | iter 4768 step 149 | loss train: 1.352, val: 1.305 | iter time: 358.96 ms (step) remaining time: 0:26:31
|
| 266 |
+
Epoch 1 | iter 4800 step 150 | loss train: 1.362, val: 1.305 | iter time: 358.69 ms (step) remaining time: 0:26:19
|
| 267 |
+
Validating ...
|
| 268 |
+
iter 4800: val loss 1.2641, val time: 22318.71 ms
|
| 269 |
+
Epoch 1 | iter 4832 step 151 | loss train: 1.345, val: 1.264 | iter time: 359.29 ms (step) remaining time: 0:26:28
|
| 270 |
+
Epoch 1 | iter 4864 step 152 | loss train: 1.330, val: 1.264 | iter time: 360.07 ms (step) remaining time: 0:26:16
|
| 271 |
+
Epoch 1 | iter 4896 step 153 | loss train: 1.381, val: 1.264 | iter time: 360.40 ms (step) remaining time: 0:26:04
|
| 272 |
+
Epoch 1 | iter 4928 step 154 | loss train: 1.323, val: 1.264 | iter time: 360.91 ms (step) remaining time: 0:25:52
|
| 273 |
+
Epoch 1 | iter 4960 step 155 | loss train: 1.432, val: 1.264 | iter time: 359.53 ms (step) remaining time: 0:25:40
|
| 274 |
+
Epoch 1 | iter 4992 step 156 | loss train: 1.333, val: 1.264 | iter time: 360.37 ms (step) remaining time: 0:25:28
|
| 275 |
+
Epoch 1 | iter 5024 step 157 | loss train: 1.412, val: 1.264 | iter time: 360.80 ms (step) remaining time: 0:25:17
|
| 276 |
+
Epoch 1 | iter 5056 step 158 | loss train: 1.410, val: 1.264 | iter time: 358.73 ms (step) remaining time: 0:25:05
|
| 277 |
+
Epoch 1 | iter 5088 step 159 | loss train: 1.357, val: 1.264 | iter time: 358.80 ms (step) remaining time: 0:24:53
|
| 278 |
+
Epoch 1 | iter 5120 step 160 | loss train: 1.416, val: 1.264 | iter time: 358.54 ms (step) remaining time: 0:24:41
|
| 279 |
+
Epoch 1 | iter 5152 step 161 | loss train: 1.361, val: 1.264 | iter time: 359.90 ms (step) remaining time: 0:24:29
|
| 280 |
+
Epoch 1 | iter 5184 step 162 | loss train: 1.404, val: 1.264 | iter time: 360.05 ms (step) remaining time: 0:24:17
|
| 281 |
+
Epoch 1 | iter 5216 step 163 | loss train: 1.413, val: 1.264 | iter time: 357.68 ms (step) remaining time: 0:24:06
|
| 282 |
+
Epoch 1 | iter 5248 step 164 | loss train: 1.378, val: 1.264 | iter time: 358.85 ms (step) remaining time: 0:23:54
|
| 283 |
+
Epoch 1 | iter 5280 step 165 | loss train: 1.327, val: 1.264 | iter time: 358.98 ms (step) remaining time: 0:23:42
|
| 284 |
+
Epoch 1 | iter 5312 step 166 | loss train: 1.312, val: 1.264 | iter time: 358.73 ms (step) remaining time: 0:23:30
|
| 285 |
+
Epoch 1 | iter 5344 step 167 | loss train: 1.428, val: 1.264 | iter time: 360.91 ms (step) remaining time: 0:23:18
|
| 286 |
+
Epoch 1 | iter 5376 step 168 | loss train: 1.457, val: 1.264 | iter time: 359.35 ms (step) remaining time: 0:23:07
|
| 287 |
+
Epoch 1 | iter 5408 step 169 | loss train: 1.415, val: 1.264 | iter time: 358.68 ms (step) remaining time: 0:22:55
|
| 288 |
+
Epoch 1 | iter 5440 step 170 | loss train: 1.322, val: 1.264 | iter time: 360.00 ms (step) remaining time: 0:22:43
|
| 289 |
+
Epoch 1 | iter 5472 step 171 | loss train: 1.370, val: 1.264 | iter time: 361.17 ms (step) remaining time: 0:22:32
|
| 290 |
+
Epoch 1 | iter 5504 step 172 | loss train: 1.417, val: 1.264 | iter time: 361.49 ms (step) remaining time: 0:22:20
|
| 291 |
+
Epoch 1 | iter 5536 step 173 | loss train: 1.394, val: 1.264 | iter time: 360.35 ms (step) remaining time: 0:22:08
|
| 292 |
+
Epoch 1 | iter 5568 step 174 | loss train: 1.361, val: 1.264 | iter time: 360.22 ms (step) remaining time: 0:21:57
|
| 293 |
+
Epoch 1 | iter 5600 step 175 | loss train: 1.391, val: 1.264 | iter time: 357.35 ms (step) remaining time: 0:21:45
|
| 294 |
+
Epoch 1 | iter 5632 step 176 | loss train: 1.384, val: 1.264 | iter time: 357.40 ms (step) remaining time: 0:21:33
|
| 295 |
+
Epoch 1 | iter 5664 step 177 | loss train: 1.388, val: 1.264 | iter time: 360.11 ms (step) remaining time: 0:21:22
|
| 296 |
+
Epoch 1 | iter 5696 step 178 | loss train: 1.350, val: 1.264 | iter time: 360.21 ms (step) remaining time: 0:21:10
|
| 297 |
+
Epoch 1 | iter 5728 step 179 | loss train: 1.408, val: 1.264 | iter time: 357.78 ms (step) remaining time: 0:20:58
|
| 298 |
+
Epoch 1 | iter 5760 step 180 | loss train: 1.393, val: 1.264 | iter time: 358.86 ms (step) remaining time: 0:20:47
|
| 299 |
+
Epoch 1 | iter 5792 step 181 | loss train: 1.325, val: 1.264 | iter time: 359.74 ms (step) remaining time: 0:20:35
|
| 300 |
+
Epoch 1 | iter 5824 step 182 | loss train: 1.401, val: 1.264 | iter time: 359.26 ms (step) remaining time: 0:20:24
|
| 301 |
+
Epoch 1 | iter 5856 step 183 | loss train: 1.367, val: 1.264 | iter time: 359.59 ms (step) remaining time: 0:20:12
|
| 302 |
+
Epoch 1 | iter 5888 step 184 | loss train: 1.324, val: 1.264 | iter time: 357.58 ms (step) remaining time: 0:20:00
|
| 303 |
+
Epoch 1 | iter 5920 step 185 | loss train: 1.377, val: 1.264 | iter time: 359.21 ms (step) remaining time: 0:19:49
|
| 304 |
+
Epoch 1 | iter 5952 step 186 | loss train: 1.343, val: 1.264 | iter time: 359.98 ms (step) remaining time: 0:19:37
|
| 305 |
+
Epoch 1 | iter 5984 step 187 | loss train: 1.425, val: 1.264 | iter time: 358.59 ms (step) remaining time: 0:19:26
|
| 306 |
+
Epoch 1 | iter 6016 step 188 | loss train: 1.313, val: 1.264 | iter time: 359.82 ms (step) remaining time: 0:19:14
|
| 307 |
+
Epoch 1 | iter 6048 step 189 | loss train: 1.411, val: 1.264 | iter time: 359.79 ms (step) remaining time: 0:19:03
|
| 308 |
+
Epoch 1 | iter 6080 step 190 | loss train: 1.353, val: 1.264 | iter time: 359.23 ms (step) remaining time: 0:18:51
|
| 309 |
+
Epoch 1 | iter 6112 step 191 | loss train: 1.320, val: 1.264 | iter time: 359.70 ms (step) remaining time: 0:18:39
|
| 310 |
+
Epoch 1 | iter 6144 step 192 | loss train: 1.365, val: 1.264 | iter time: 360.71 ms (step) remaining time: 0:18:28
|
| 311 |
+
Epoch 1 | iter 6176 step 193 | loss train: 1.328, val: 1.264 | iter time: 358.91 ms (step) remaining time: 0:18:16
|
| 312 |
+
Epoch 1 | iter 6208 step 194 | loss train: 1.401, val: 1.264 | iter time: 358.67 ms (step) remaining time: 0:18:05
|
| 313 |
+
Epoch 1 | iter 6240 step 195 | loss train: 1.308, val: 1.264 | iter time: 359.17 ms (step) remaining time: 0:17:53
|
| 314 |
+
Epoch 1 | iter 6272 step 196 | loss train: 1.264, val: 1.264 | iter time: 358.97 ms (step) remaining time: 0:17:42
|
| 315 |
+
Epoch 1 | iter 6304 step 197 | loss train: 1.383, val: 1.264 | iter time: 358.90 ms (step) remaining time: 0:17:30
|
| 316 |
+
Epoch 1 | iter 6336 step 198 | loss train: 1.374, val: 1.264 | iter time: 360.07 ms (step) remaining time: 0:17:19
|
| 317 |
+
Epoch 1 | iter 6368 step 199 | loss train: 1.380, val: 1.264 | iter time: 358.77 ms (step) remaining time: 0:17:07
|
| 318 |
+
Epoch 1 | iter 6400 step 200 | loss train: 1.368, val: 1.264 | iter time: 361.04 ms (step) remaining time: 0:16:56
|
| 319 |
+
Validating ...
|
| 320 |
+
iter 6400: val loss 1.2153, val time: 22309.80 ms
|
| 321 |
+
Saving checkpoint to 'out/pretrain/2412_full/step-00000200/lit_model.pth'
|
| 322 |
+
Epoch 1 | iter 6432 step 201 | loss train: 1.359, val: 1.215 | iter time: 355.81 ms (step) remaining time: 0:17:02
|
| 323 |
+
Epoch 1 | iter 6464 step 202 | loss train: 1.352, val: 1.215 | iter time: 359.06 ms (step) remaining time: 0:16:50
|
| 324 |
+
Epoch 1 | iter 6496 step 203 | loss train: 1.301, val: 1.215 | iter time: 359.85 ms (step) remaining time: 0:16:38
|
| 325 |
+
Epoch 1 | iter 6528 step 204 | loss train: 1.308, val: 1.215 | iter time: 358.79 ms (step) remaining time: 0:16:26
|
| 326 |
+
Epoch 1 | iter 6560 step 205 | loss train: 1.392, val: 1.215 | iter time: 359.65 ms (step) remaining time: 0:16:15
|
| 327 |
+
Epoch 1 | iter 6592 step 206 | loss train: 1.352, val: 1.215 | iter time: 359.50 ms (step) remaining time: 0:16:03
|
| 328 |
+
Epoch 1 | iter 6624 step 207 | loss train: 1.410, val: 1.215 | iter time: 357.80 ms (step) remaining time: 0:15:51
|
| 329 |
+
Epoch 1 | iter 6656 step 208 | loss train: 1.366, val: 1.215 | iter time: 358.14 ms (step) remaining time: 0:15:39
|
| 330 |
+
Epoch 1 | iter 6688 step 209 | loss train: 1.313, val: 1.215 | iter time: 358.15 ms (step) remaining time: 0:15:28
|
| 331 |
+
Epoch 1 | iter 6720 step 210 | loss train: 1.350, val: 1.215 | iter time: 359.08 ms (step) remaining time: 0:15:16
|
| 332 |
+
Epoch 1 | iter 6752 step 211 | loss train: 1.346, val: 1.215 | iter time: 360.07 ms (step) remaining time: 0:15:04
|
| 333 |
+
Epoch 1 | iter 6784 step 212 | loss train: 1.426, val: 1.215 | iter time: 360.54 ms (step) remaining time: 0:14:53
|
| 334 |
+
Epoch 1 | iter 6816 step 213 | loss train: 1.358, val: 1.215 | iter time: 360.48 ms (step) remaining time: 0:14:41
|
| 335 |
+
Epoch 1 | iter 6848 step 214 | loss train: 1.337, val: 1.215 | iter time: 360.18 ms (step) remaining time: 0:14:29
|
| 336 |
+
Epoch 1 | iter 6880 step 215 | loss train: 1.411, val: 1.215 | iter time: 359.64 ms (step) remaining time: 0:14:18
|
| 337 |
+
Epoch 1 | iter 6912 step 216 | loss train: 1.298, val: 1.215 | iter time: 358.69 ms (step) remaining time: 0:14:06
|
| 338 |
+
Epoch 1 | iter 6944 step 217 | loss train: 1.372, val: 1.215 | iter time: 360.14 ms (step) remaining time: 0:13:54
|
| 339 |
+
Epoch 1 | iter 6976 step 218 | loss train: 1.322, val: 1.215 | iter time: 361.39 ms (step) remaining time: 0:13:43
|
| 340 |
+
Epoch 1 | iter 7008 step 219 | loss train: 1.347, val: 1.215 | iter time: 359.88 ms (step) remaining time: 0:13:31
|
| 341 |
+
Epoch 1 | iter 7040 step 220 | loss train: 1.327, val: 1.215 | iter time: 359.27 ms (step) remaining time: 0:13:20
|
| 342 |
+
Epoch 1 | iter 7072 step 221 | loss train: 1.312, val: 1.215 | iter time: 359.89 ms (step) remaining time: 0:13:08
|
| 343 |
+
Epoch 1 | iter 7104 step 222 | loss train: 1.337, val: 1.215 | iter time: 359.85 ms (step) remaining time: 0:12:56
|
| 344 |
+
Epoch 1 | iter 7136 step 223 | loss train: 1.350, val: 1.215 | iter time: 358.61 ms (step) remaining time: 0:12:45
|
| 345 |
+
Epoch 1 | iter 7168 step 224 | loss train: 1.312, val: 1.215 | iter time: 358.78 ms (step) remaining time: 0:12:33
|
| 346 |
+
Epoch 1 | iter 7200 step 225 | loss train: 1.328, val: 1.215 | iter time: 360.38 ms (step) remaining time: 0:12:22
|
| 347 |
+
Epoch 1 | iter 7232 step 226 | loss train: 1.392, val: 1.215 | iter time: 359.40 ms (step) remaining time: 0:12:10
|
| 348 |
+
Epoch 1 | iter 7264 step 227 | loss train: 1.327, val: 1.215 | iter time: 359.22 ms (step) remaining time: 0:11:58
|
| 349 |
+
Epoch 1 | iter 7296 step 228 | loss train: 1.390, val: 1.215 | iter time: 360.04 ms (step) remaining time: 0:11:47
|
| 350 |
+
Epoch 1 | iter 7328 step 229 | loss train: 1.355, val: 1.215 | iter time: 359.28 ms (step) remaining time: 0:11:35
|
| 351 |
+
Epoch 1 | iter 7360 step 230 | loss train: 1.386, val: 1.215 | iter time: 360.59 ms (step) remaining time: 0:11:24
|
| 352 |
+
Epoch 1 | iter 7392 step 231 | loss train: 1.347, val: 1.215 | iter time: 358.41 ms (step) remaining time: 0:11:12
|
| 353 |
+
Epoch 1 | iter 7424 step 232 | loss train: 1.324, val: 1.215 | iter time: 358.11 ms (step) remaining time: 0:11:01
|
| 354 |
+
Epoch 1 | iter 7456 step 233 | loss train: 1.345, val: 1.215 | iter time: 359.01 ms (step) remaining time: 0:10:49
|
| 355 |
+
Epoch 1 | iter 7488 step 234 | loss train: 1.301, val: 1.215 | iter time: 357.50 ms (step) remaining time: 0:10:38
|
| 356 |
+
Epoch 1 | iter 7520 step 235 | loss train: 1.382, val: 1.215 | iter time: 360.48 ms (step) remaining time: 0:10:26
|
| 357 |
+
Epoch 1 | iter 7552 step 236 | loss train: 1.362, val: 1.215 | iter time: 359.96 ms (step) remaining time: 0:10:15
|
| 358 |
+
Epoch 1 | iter 7584 step 237 | loss train: 1.447, val: 1.215 | iter time: 359.46 ms (step) remaining time: 0:10:03
|
| 359 |
+
Epoch 1 | iter 7616 step 238 | loss train: 1.407, val: 1.215 | iter time: 359.74 ms (step) remaining time: 0:09:52
|
| 360 |
+
Epoch 1 | iter 7648 step 239 | loss train: 1.319, val: 1.215 | iter time: 360.08 ms (step) remaining time: 0:09:40
|
| 361 |
+
Epoch 1 | iter 7680 step 240 | loss train: 1.305, val: 1.215 | iter time: 358.30 ms (step) remaining time: 0:09:29
|
| 362 |
+
Epoch 1 | iter 7712 step 241 | loss train: 1.373, val: 1.215 | iter time: 358.91 ms (step) remaining time: 0:09:17
|
| 363 |
+
Epoch 1 | iter 7744 step 242 | loss train: 1.302, val: 1.215 | iter time: 359.06 ms (step) remaining time: 0:09:06
|
| 364 |
+
Epoch 1 | iter 7776 step 243 | loss train: 1.306, val: 1.215 | iter time: 359.05 ms (step) remaining time: 0:08:54
|
| 365 |
+
Epoch 1 | iter 7808 step 244 | loss train: 1.350, val: 1.215 | iter time: 360.22 ms (step) remaining time: 0:08:43
|
| 366 |
+
Epoch 1 | iter 7840 step 245 | loss train: 1.405, val: 1.215 | iter time: 360.15 ms (step) remaining time: 0:08:31
|
| 367 |
+
Epoch 1 | iter 7872 step 246 | loss train: 1.363, val: 1.215 | iter time: 359.16 ms (step) remaining time: 0:08:20
|
| 368 |
+
Epoch 1 | iter 7904 step 247 | loss train: 1.371, val: 1.215 | iter time: 360.34 ms (step) remaining time: 0:08:08
|
| 369 |
+
Epoch 1 | iter 7936 step 248 | loss train: 1.305, val: 1.215 | iter time: 360.11 ms (step) remaining time: 0:07:57
|
| 370 |
+
Epoch 1 | iter 7968 step 249 | loss train: 1.302, val: 1.215 | iter time: 358.10 ms (step) remaining time: 0:07:45
|
| 371 |
+
Epoch 1 | iter 8000 step 250 | loss train: 1.362, val: 1.215 | iter time: 361.27 ms (step) remaining time: 0:07:34
|
| 372 |
+
Validating ...
|
| 373 |
+
iter 8000: val loss 1.1898, val time: 22341.58 ms
|
| 374 |
+
Epoch 1 | iter 8032 step 251 | loss train: 1.403, val: 1.190 | iter time: 360.25 ms (step) remaining time: 0:07:26
|
| 375 |
+
Epoch 1 | iter 8064 step 252 | loss train: 1.336, val: 1.190 | iter time: 359.44 ms (step) remaining time: 0:07:15
|
| 376 |
+
Epoch 1 | iter 8096 step 253 | loss train: 1.387, val: 1.190 | iter time: 358.14 ms (step) remaining time: 0:07:03
|
| 377 |
+
Epoch 1 | iter 8128 step 254 | loss train: 1.346, val: 1.190 | iter time: 358.79 ms (step) remaining time: 0:06:51
|
| 378 |
+
Epoch 1 | iter 8160 step 255 | loss train: 1.287, val: 1.190 | iter time: 359.99 ms (step) remaining time: 0:06:40
|
| 379 |
+
Epoch 1 | iter 8192 step 256 | loss train: 1.320, val: 1.190 | iter time: 358.56 ms (step) remaining time: 0:06:28
|
| 380 |
+
Epoch 1 | iter 8224 step 257 | loss train: 1.288, val: 1.190 | iter time: 359.50 ms (step) remaining time: 0:06:17
|
| 381 |
+
Epoch 1 | iter 8256 step 258 | loss train: 1.401, val: 1.190 | iter time: 358.16 ms (step) remaining time: 0:06:05
|
| 382 |
+
Epoch 1 | iter 8288 step 259 | loss train: 1.352, val: 1.190 | iter time: 359.35 ms (step) remaining time: 0:05:54
|
| 383 |
+
Epoch 1 | iter 8320 step 260 | loss train: 1.350, val: 1.190 | iter time: 359.49 ms (step) remaining time: 0:05:42
|
| 384 |
+
Epoch 1 | iter 8352 step 261 | loss train: 1.428, val: 1.190 | iter time: 359.55 ms (step) remaining time: 0:05:31
|
| 385 |
+
Epoch 1 | iter 8384 step 262 | loss train: 1.390, val: 1.190 | iter time: 359.85 ms (step) remaining time: 0:05:19
|
| 386 |
+
Epoch 1 | iter 8416 step 263 | loss train: 1.397, val: 1.190 | iter time: 360.74 ms (step) remaining time: 0:05:08
|
| 387 |
+
Epoch 1 | iter 8448 step 264 | loss train: 1.296, val: 1.190 | iter time: 359.49 ms (step) remaining time: 0:04:56
|
| 388 |
+
Epoch 1 | iter 8480 step 265 | loss train: 1.330, val: 1.190 | iter time: 360.23 ms (step) remaining time: 0:04:45
|
| 389 |
+
Epoch 1 | iter 8512 step 266 | loss train: 1.342, val: 1.190 | iter time: 359.45 ms (step) remaining time: 0:04:34
|
| 390 |
+
Epoch 1 | iter 8544 step 267 | loss train: 1.339, val: 1.190 | iter time: 360.97 ms (step) remaining time: 0:04:22
|
| 391 |
+
Epoch 1 | iter 8576 step 268 | loss train: 1.369, val: 1.190 | iter time: 360.16 ms (step) remaining time: 0:04:11
|
| 392 |
+
Epoch 1 | iter 8608 step 269 | loss train: 1.330, val: 1.190 | iter time: 360.27 ms (step) remaining time: 0:03:59
|
| 393 |
+
Epoch 1 | iter 8640 step 270 | loss train: 1.327, val: 1.190 | iter time: 359.12 ms (step) remaining time: 0:03:48
|
| 394 |
+
Epoch 1 | iter 8672 step 271 | loss train: 1.323, val: 1.190 | iter time: 357.73 ms (step) remaining time: 0:03:36
|
| 395 |
+
Epoch 1 | iter 8704 step 272 | loss train: 1.382, val: 1.190 | iter time: 358.08 ms (step) remaining time: 0:03:25
|
| 396 |
+
Epoch 1 | iter 8736 step 273 | loss train: 1.356, val: 1.190 | iter time: 359.43 ms (step) remaining time: 0:03:13
|
| 397 |
+
Epoch 1 | iter 8768 step 274 | loss train: 1.337, val: 1.190 | iter time: 359.24 ms (step) remaining time: 0:03:02
|
| 398 |
+
Epoch 1 | iter 8800 step 275 | loss train: 1.355, val: 1.190 | iter time: 359.00 ms (step) remaining time: 0:02:50
|
| 399 |
+
Epoch 1 | iter 8832 step 276 | loss train: 1.368, val: 1.190 | iter time: 359.62 ms (step) remaining time: 0:02:39
|
| 400 |
+
Epoch 1 | iter 8864 step 277 | loss train: 1.358, val: 1.190 | iter time: 358.51 ms (step) remaining time: 0:02:28
|
| 401 |
+
Epoch 1 | iter 8896 step 278 | loss train: 1.335, val: 1.190 | iter time: 358.77 ms (step) remaining time: 0:02:16
|
| 402 |
+
Epoch 1 | iter 8928 step 279 | loss train: 1.335, val: 1.190 | iter time: 359.75 ms (step) remaining time: 0:02:05
|
| 403 |
+
Epoch 1 | iter 8960 step 280 | loss train: 1.324, val: 1.190 | iter time: 358.75 ms (step) remaining time: 0:01:53
|
| 404 |
+
Epoch 1 | iter 8992 step 281 | loss train: 1.363, val: 1.190 | iter time: 360.08 ms (step) remaining time: 0:01:42
|
| 405 |
+
Epoch 1 | iter 9024 step 282 | loss train: 1.342, val: 1.190 | iter time: 357.96 ms (step) remaining time: 0:01:31
|
| 406 |
+
Epoch 1 | iter 9056 step 283 | loss train: 1.330, val: 1.190 | iter time: 359.58 ms (step) remaining time: 0:01:19
|
| 407 |
+
Epoch 1 | iter 9088 step 284 | loss train: 1.378, val: 1.190 | iter time: 359.22 ms (step) remaining time: 0:01:08
|
| 408 |
+
Epoch 1 | iter 9120 step 285 | loss train: 1.395, val: 1.190 | iter time: 359.31 ms (step) remaining time: 0:00:56
|
| 409 |
+
Epoch 1 | iter 9152 step 286 | loss train: 1.375, val: 1.190 | iter time: 358.42 ms (step) remaining time: 0:00:45
|
| 410 |
+
Epoch 1 | iter 9184 step 287 | loss train: 1.343, val: 1.190 | iter time: 359.37 ms (step) remaining time: 0:00:34
|
| 411 |
+
Epoch 1 | iter 9216 step 288 | loss train: 1.376, val: 1.190 | iter time: 359.54 ms (step) remaining time: 0:00:22
|
| 412 |
+
Epoch 2 | iter 9248 step 289 | loss train: 1.302, val: 1.190 | iter time: 358.39 ms (step) remaining time: 0:00:11
|
| 413 |
+
Epoch 2 | iter 9280 step 290 | loss train: 1.167, val: 1.190 | iter time: 358.25 ms (step) remaining time: 0:00:00
|
| 414 |
+
Validating ...
|
| 415 |
+
Final evaluation | val loss: 1.173 | val ppl: 3.232
|
| 416 |
+
Saving checkpoint to 'out/pretrain/2412_full/final/lit_model.pth'
|
| 417 |
+
----------------------------------------
|
| 418 |
+
| Performance
|
| 419 |
+
| - Total tokens : 304,087,040
|
| 420 |
+
| - Training Time : 3361.34 s
|
| 421 |
+
| - Tok/sec : 163.63 tok/s
|
| 422 |
+
| ----------------------------------------
|
| 423 |
+
| Memory Usage
|
| 424 |
+
| - Memory Used : 26.32 GB
|
| 425 |
+
----------------------------------------
|
out/pretrain/tinyllama/teelogs/2412_lr4e-5.txt
ADDED
|
@@ -0,0 +1,424 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
Initializing distributed: GLOBAL_RANK: 0, MEMBER: 1/4
|
| 2 |
+
Initializing distributed: GLOBAL_RANK: 1, MEMBER: 2/4
|
| 3 |
+
[rank: 1] Seed set to 42
|
| 4 |
+
Initializing distributed: GLOBAL_RANK: 3, MEMBER: 4/4
|
| 5 |
+
[rank: 3] Seed set to 42
|
| 6 |
+
Initializing distributed: GLOBAL_RANK: 2, MEMBER: 3/4
|
| 7 |
+
[rank: 2] Seed set to 42
|
| 8 |
+
----------------------------------------------------------------------------------------------------
|
| 9 |
+
distributed_backend=nccl
|
| 10 |
+
All distributed processes registered. Starting with 4 processes
|
| 11 |
+
----------------------------------------------------------------------------------------------------
|
| 12 |
+
|
| 13 |
+
[rank: 0] Seed set to 42
|
| 14 |
+
/mnt/data/litgpt/litgpt/pretrain.py:515: UserWarning: A newer version of litdata is available (0.2.52). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 15 |
+
train_dataloader = data.train_dataloader()
|
| 16 |
+
/mnt/data/litgpt/litgpt/pretrain.py:515: UserWarning: A newer version of litdata is available (0.2.52). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 17 |
+
train_dataloader = data.train_dataloader()
|
| 18 |
+
/mnt/data/litgpt/litgpt/pretrain.py:515: UserWarning: A newer version of litdata is available (0.2.52). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 19 |
+
train_dataloader = data.train_dataloader()
|
| 20 |
+
/mnt/data/litgpt/litgpt/pretrain.py:515: UserWarning: A newer version of litdata is available (0.2.52). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 21 |
+
train_dataloader = data.train_dataloader()
|
| 22 |
+
All GPUs are fully connected via NVLink.
|
| 23 |
+
{'data': {'batch_size': 1,
|
| 24 |
+
'data_path': PosixPath('litgpt/data/arxiv'),
|
| 25 |
+
'num_workers': 0,
|
| 26 |
+
'seed': 42,
|
| 27 |
+
'seq_length': 2048,
|
| 28 |
+
'use_starcoder': True},
|
| 29 |
+
'data_dir': PosixPath('litgpt/data/arxiv/2412'),
|
| 30 |
+
'devices': 'auto',
|
| 31 |
+
'eval': {'evaluate_example': 'first',
|
| 32 |
+
'final_validation': True,
|
| 33 |
+
'initial_validation': True,
|
| 34 |
+
'interval': 50,
|
| 35 |
+
'max_iters': 200,
|
| 36 |
+
'max_new_tokens': None},
|
| 37 |
+
'initial_checkpoint_dir': PosixPath('out/pretrain/tinyllama/2411_full/final'),
|
| 38 |
+
'log': {'group': None, 'project': None, 'run': None},
|
| 39 |
+
'logger_name': 'tensorboard',
|
| 40 |
+
'model_config': {'attention_logit_softcapping': None,
|
| 41 |
+
'attention_scores_scalar': None,
|
| 42 |
+
'attn_bias': False,
|
| 43 |
+
'bias': False,
|
| 44 |
+
'block_size': 2048,
|
| 45 |
+
'final_logit_softcapping': None,
|
| 46 |
+
'gelu_approximate': 'none',
|
| 47 |
+
'head_size': 64,
|
| 48 |
+
'hf_config': {'name': 'TinyLlama-1.1B-intermediate-step-1431k-3T',
|
| 49 |
+
'org': 'TinyLlama'},
|
| 50 |
+
'intermediate_size': 5632,
|
| 51 |
+
'lm_head_bias': False,
|
| 52 |
+
'mlp_class_name': 'LLaMAMLP',
|
| 53 |
+
'moe_intermediate_size': None,
|
| 54 |
+
'n_embd': 2048,
|
| 55 |
+
'n_expert': 0,
|
| 56 |
+
'n_expert_per_token': 0,
|
| 57 |
+
'n_head': 32,
|
| 58 |
+
'n_layer': 22,
|
| 59 |
+
'n_query_groups': 4,
|
| 60 |
+
'name': 'tiny-llama-1.1b',
|
| 61 |
+
'norm_1': True,
|
| 62 |
+
'norm_2': True,
|
| 63 |
+
'norm_class_name': 'RMSNorm',
|
| 64 |
+
'norm_eps': 1e-05,
|
| 65 |
+
'norm_qk': False,
|
| 66 |
+
'norm_qk_type': 'default',
|
| 67 |
+
'padded_vocab_size': 32000,
|
| 68 |
+
'padding_multiple': 64,
|
| 69 |
+
'parallel_residual': False,
|
| 70 |
+
'post_attention_norm': False,
|
| 71 |
+
'post_mlp_norm': False,
|
| 72 |
+
'rope_adjustments': None,
|
| 73 |
+
'rope_base': 10000,
|
| 74 |
+
'rope_condense_ratio': 1,
|
| 75 |
+
'rope_indices': None,
|
| 76 |
+
'rope_local_base_freq': None,
|
| 77 |
+
'rotary_percentage': 1.0,
|
| 78 |
+
'scale_embeddings': False,
|
| 79 |
+
'shared_attention_norm': False,
|
| 80 |
+
'sliding_window_indices': None,
|
| 81 |
+
'sliding_window_size': None,
|
| 82 |
+
'vocab_size': 32000},
|
| 83 |
+
'model_name': 'tiny-llama-1.1b',
|
| 84 |
+
'num_nodes': 1,
|
| 85 |
+
'optimizer': "{'class_path': 'torch.optim.AdamW', 'init_args': {'lr': 4e-05, "
|
| 86 |
+
"'weight_decay': 0.1, 'betas': [0.9, 0.95]}}",
|
| 87 |
+
'out_dir': PosixPath('out/pretrain/tinyllama/2412_lr4e-5'),
|
| 88 |
+
'ppl': False,
|
| 89 |
+
'precision': 'bf16-mixed',
|
| 90 |
+
'resume': False,
|
| 91 |
+
'seed': 42,
|
| 92 |
+
'tokenizer_dir': PosixPath('checkpoints/TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T'),
|
| 93 |
+
'train': {'epochs': None,
|
| 94 |
+
'global_batch_size': 512,
|
| 95 |
+
'log_interval': 1,
|
| 96 |
+
'lr_warmup_fraction': None,
|
| 97 |
+
'lr_warmup_steps': 20,
|
| 98 |
+
'max_norm': 1.0,
|
| 99 |
+
'max_seq_length': 2048,
|
| 100 |
+
'max_steps': None,
|
| 101 |
+
'max_tokens': 304087040,
|
| 102 |
+
'micro_batch_size': 4,
|
| 103 |
+
'min_lr': 4e-05,
|
| 104 |
+
'save_interval': 100,
|
| 105 |
+
'tie_embeddings': None}}
|
| 106 |
+
Time to instantiate model: 0.02 seconds.
|
| 107 |
+
Total parameters: 1,100,048,384
|
| 108 |
+
[ok] out/pretrain/tinyllama/2411_full/final/lit_model.pth 已是纯权重
|
| 109 |
+
Validating ...
|
| 110 |
+
Measured TFLOPs: 239.66
|
| 111 |
+
Epoch 1 | iter 32 step 1 | loss train: 1.336, val: 1.313 | iter time: 565.83 ms (step) remaining time: 0:56:26
|
| 112 |
+
Epoch 1 | iter 64 step 2 | loss train: 1.317, val: 1.313 | iter time: 356.53 ms (step) remaining time: 0:54:00
|
| 113 |
+
Epoch 1 | iter 96 step 3 | loss train: 1.342, val: 1.313 | iter time: 359.62 ms (step) remaining time: 0:53:05
|
| 114 |
+
Epoch 1 | iter 128 step 4 | loss train: 1.431, val: 1.313 | iter time: 358.74 ms (step) remaining time: 0:52:41
|
| 115 |
+
Epoch 1 | iter 160 step 5 | loss train: 1.400, val: 1.313 | iter time: 359.17 ms (step) remaining time: 0:52:17
|
| 116 |
+
Epoch 1 | iter 192 step 6 | loss train: 1.420, val: 1.313 | iter time: 362.10 ms (step) remaining time: 0:51:59
|
| 117 |
+
Epoch 1 | iter 224 step 7 | loss train: 1.456, val: 1.313 | iter time: 362.36 ms (step) remaining time: 0:51:42
|
| 118 |
+
Epoch 1 | iter 256 step 8 | loss train: 1.355, val: 1.313 | iter time: 359.39 ms (step) remaining time: 0:51:28
|
| 119 |
+
Epoch 1 | iter 288 step 9 | loss train: 1.346, val: 1.313 | iter time: 359.07 ms (step) remaining time: 0:51:14
|
| 120 |
+
Epoch 1 | iter 320 step 10 | loss train: 1.458, val: 1.313 | iter time: 360.41 ms (step) remaining time: 0:51:01
|
| 121 |
+
Epoch 1 | iter 352 step 11 | loss train: 1.388, val: 1.313 | iter time: 359.18 ms (step) remaining time: 0:50:48
|
| 122 |
+
Epoch 1 | iter 384 step 12 | loss train: 1.411, val: 1.313 | iter time: 358.91 ms (step) remaining time: 0:50:35
|
| 123 |
+
Epoch 1 | iter 416 step 13 | loss train: 1.331, val: 1.313 | iter time: 360.22 ms (step) remaining time: 0:50:23
|
| 124 |
+
Epoch 1 | iter 448 step 14 | loss train: 1.439, val: 1.313 | iter time: 359.52 ms (step) remaining time: 0:50:11
|
| 125 |
+
Epoch 1 | iter 480 step 15 | loss train: 1.357, val: 1.313 | iter time: 360.56 ms (step) remaining time: 0:49:59
|
| 126 |
+
Epoch 1 | iter 512 step 16 | loss train: 1.350, val: 1.313 | iter time: 360.92 ms (step) remaining time: 0:49:48
|
| 127 |
+
Epoch 1 | iter 544 step 17 | loss train: 1.353, val: 1.313 | iter time: 358.80 ms (step) remaining time: 0:49:36
|
| 128 |
+
Epoch 1 | iter 576 step 18 | loss train: 1.354, val: 1.313 | iter time: 360.08 ms (step) remaining time: 0:49:25
|
| 129 |
+
Epoch 1 | iter 608 step 19 | loss train: 1.396, val: 1.313 | iter time: 357.80 ms (step) remaining time: 0:49:14
|
| 130 |
+
Epoch 1 | iter 640 step 20 | loss train: 1.383, val: 1.313 | iter time: 358.83 ms (step) remaining time: 0:49:04
|
| 131 |
+
Epoch 1 | iter 672 step 21 | loss train: 1.401, val: 1.313 | iter time: 360.35 ms (step) remaining time: 0:48:52
|
| 132 |
+
Epoch 1 | iter 704 step 22 | loss train: 1.397, val: 1.313 | iter time: 358.78 ms (step) remaining time: 0:48:41
|
| 133 |
+
Epoch 1 | iter 736 step 23 | loss train: 1.378, val: 1.313 | iter time: 358.63 ms (step) remaining time: 0:48:30
|
| 134 |
+
Epoch 1 | iter 768 step 24 | loss train: 1.405, val: 1.313 | iter time: 360.41 ms (step) remaining time: 0:48:19
|
| 135 |
+
Epoch 1 | iter 800 step 25 | loss train: 1.383, val: 1.313 | iter time: 358.98 ms (step) remaining time: 0:48:07
|
| 136 |
+
Epoch 1 | iter 832 step 26 | loss train: 1.374, val: 1.313 | iter time: 360.33 ms (step) remaining time: 0:47:56
|
| 137 |
+
Epoch 1 | iter 864 step 27 | loss train: 1.372, val: 1.313 | iter time: 361.14 ms (step) remaining time: 0:47:45
|
| 138 |
+
Epoch 1 | iter 896 step 28 | loss train: 1.336, val: 1.313 | iter time: 359.37 ms (step) remaining time: 0:47:34
|
| 139 |
+
Epoch 1 | iter 928 step 29 | loss train: 1.453, val: 1.313 | iter time: 358.61 ms (step) remaining time: 0:47:23
|
| 140 |
+
Epoch 1 | iter 960 step 30 | loss train: 1.349, val: 1.313 | iter time: 360.64 ms (step) remaining time: 0:47:12
|
| 141 |
+
Epoch 1 | iter 992 step 31 | loss train: 1.352, val: 1.313 | iter time: 359.15 ms (step) remaining time: 0:47:01
|
| 142 |
+
Epoch 1 | iter 1024 step 32 | loss train: 1.361, val: 1.313 | iter time: 358.99 ms (step) remaining time: 0:46:50
|
| 143 |
+
Epoch 1 | iter 1056 step 33 | loss train: 1.384, val: 1.313 | iter time: 360.68 ms (step) remaining time: 0:46:38
|
| 144 |
+
Epoch 1 | iter 1088 step 34 | loss train: 1.324, val: 1.313 | iter time: 361.32 ms (step) remaining time: 0:46:27
|
| 145 |
+
Epoch 1 | iter 1120 step 35 | loss train: 1.345, val: 1.313 | iter time: 360.46 ms (step) remaining time: 0:46:16
|
| 146 |
+
Epoch 1 | iter 1152 step 36 | loss train: 1.323, val: 1.313 | iter time: 358.31 ms (step) remaining time: 0:46:05
|
| 147 |
+
Epoch 1 | iter 1184 step 37 | loss train: 1.361, val: 1.313 | iter time: 360.53 ms (step) remaining time: 0:45:54
|
| 148 |
+
Epoch 1 | iter 1216 step 38 | loss train: 1.354, val: 1.313 | iter time: 361.09 ms (step) remaining time: 0:45:43
|
| 149 |
+
Epoch 1 | iter 1248 step 39 | loss train: 1.399, val: 1.313 | iter time: 358.49 ms (step) remaining time: 0:45:32
|
| 150 |
+
Epoch 1 | iter 1280 step 40 | loss train: 1.402, val: 1.313 | iter time: 359.77 ms (step) remaining time: 0:45:21
|
| 151 |
+
Epoch 1 | iter 1312 step 41 | loss train: 1.361, val: 1.313 | iter time: 360.00 ms (step) remaining time: 0:45:11
|
| 152 |
+
Epoch 1 | iter 1344 step 42 | loss train: 1.378, val: 1.313 | iter time: 360.63 ms (step) remaining time: 0:44:59
|
| 153 |
+
Epoch 1 | iter 1376 step 43 | loss train: 1.383, val: 1.313 | iter time: 358.34 ms (step) remaining time: 0:44:48
|
| 154 |
+
Epoch 1 | iter 1408 step 44 | loss train: 1.435, val: 1.313 | iter time: 359.95 ms (step) remaining time: 0:44:37
|
| 155 |
+
Epoch 1 | iter 1440 step 45 | loss train: 1.423, val: 1.313 | iter time: 360.90 ms (step) remaining time: 0:44:27
|
| 156 |
+
Epoch 1 | iter 1472 step 46 | loss train: 1.322, val: 1.313 | iter time: 360.88 ms (step) remaining time: 0:44:17
|
| 157 |
+
Epoch 1 | iter 1504 step 47 | loss train: 1.326, val: 1.313 | iter time: 360.33 ms (step) remaining time: 0:44:06
|
| 158 |
+
Epoch 1 | iter 1536 step 48 | loss train: 1.383, val: 1.313 | iter time: 358.00 ms (step) remaining time: 0:43:55
|
| 159 |
+
Epoch 1 | iter 1568 step 49 | loss train: 1.384, val: 1.313 | iter time: 358.84 ms (step) remaining time: 0:43:44
|
| 160 |
+
Epoch 1 | iter 1600 step 50 | loss train: 1.391, val: 1.313 | iter time: 358.96 ms (step) remaining time: 0:43:33
|
| 161 |
+
Validating ...
|
| 162 |
+
iter 1600: val loss 1.3029, val time: 21936.26 ms
|
| 163 |
+
Epoch 1 | iter 1632 step 51 | loss train: 1.355, val: 1.303 | iter time: 359.70 ms (step) remaining time: 0:45:05
|
| 164 |
+
Epoch 1 | iter 1664 step 52 | loss train: 1.364, val: 1.303 | iter time: 357.90 ms (step) remaining time: 0:44:51
|
| 165 |
+
Epoch 1 | iter 1696 step 53 | loss train: 1.319, val: 1.303 | iter time: 360.23 ms (step) remaining time: 0:44:38
|
| 166 |
+
Epoch 1 | iter 1728 step 54 | loss train: 1.359, val: 1.303 | iter time: 360.56 ms (step) remaining time: 0:44:25
|
| 167 |
+
Epoch 1 | iter 1760 step 55 | loss train: 1.322, val: 1.303 | iter time: 809.47 ms (step) remaining time: 0:44:13
|
| 168 |
+
Epoch 1 | iter 1792 step 56 | loss train: 1.336, val: 1.303 | iter time: 358.64 ms (step) remaining time: 0:44:00
|
| 169 |
+
Epoch 1 | iter 1824 step 57 | loss train: 1.309, val: 1.303 | iter time: 360.66 ms (step) remaining time: 0:43:47
|
| 170 |
+
Epoch 1 | iter 1856 step 58 | loss train: 1.324, val: 1.303 | iter time: 360.01 ms (step) remaining time: 0:43:34
|
| 171 |
+
Epoch 1 | iter 1888 step 59 | loss train: 1.394, val: 1.303 | iter time: 360.62 ms (step) remaining time: 0:43:21
|
| 172 |
+
Epoch 1 | iter 1920 step 60 | loss train: 1.447, val: 1.303 | iter time: 361.11 ms (step) remaining time: 0:43:09
|
| 173 |
+
Epoch 1 | iter 1952 step 61 | loss train: 1.324, val: 1.303 | iter time: 360.09 ms (step) remaining time: 0:42:56
|
| 174 |
+
Epoch 1 | iter 1984 step 62 | loss train: 1.373, val: 1.303 | iter time: 359.53 ms (step) remaining time: 0:42:43
|
| 175 |
+
Epoch 1 | iter 2016 step 63 | loss train: 1.415, val: 1.303 | iter time: 359.52 ms (step) remaining time: 0:42:31
|
| 176 |
+
Epoch 1 | iter 2048 step 64 | loss train: 1.373, val: 1.303 | iter time: 358.80 ms (step) remaining time: 0:42:18
|
| 177 |
+
Epoch 1 | iter 2080 step 65 | loss train: 1.323, val: 1.303 | iter time: 360.27 ms (step) remaining time: 0:42:06
|
| 178 |
+
Epoch 1 | iter 2112 step 66 | loss train: 1.322, val: 1.303 | iter time: 358.27 ms (step) remaining time: 0:41:53
|
| 179 |
+
Epoch 1 | iter 2144 step 67 | loss train: 1.304, val: 1.303 | iter time: 359.32 ms (step) remaining time: 0:41:41
|
| 180 |
+
Epoch 1 | iter 2176 step 68 | loss train: 1.318, val: 1.303 | iter time: 360.43 ms (step) remaining time: 0:41:28
|
| 181 |
+
Epoch 1 | iter 2208 step 69 | loss train: 1.369, val: 1.303 | iter time: 359.58 ms (step) remaining time: 0:41:16
|
| 182 |
+
Epoch 1 | iter 2240 step 70 | loss train: 1.404, val: 1.303 | iter time: 359.33 ms (step) remaining time: 0:41:04
|
| 183 |
+
Epoch 1 | iter 2272 step 71 | loss train: 1.351, val: 1.303 | iter time: 358.02 ms (step) remaining time: 0:40:52
|
| 184 |
+
Epoch 1 | iter 2304 step 72 | loss train: 1.377, val: 1.303 | iter time: 359.25 ms (step) remaining time: 0:40:40
|
| 185 |
+
Epoch 1 | iter 2336 step 73 | loss train: 1.363, val: 1.303 | iter time: 359.37 ms (step) remaining time: 0:40:27
|
| 186 |
+
Epoch 1 | iter 2368 step 74 | loss train: 1.281, val: 1.303 | iter time: 359.60 ms (step) remaining time: 0:40:15
|
| 187 |
+
Epoch 1 | iter 2400 step 75 | loss train: 1.375, val: 1.303 | iter time: 361.50 ms (step) remaining time: 0:40:03
|
| 188 |
+
Epoch 1 | iter 2432 step 76 | loss train: 1.347, val: 1.303 | iter time: 359.78 ms (step) remaining time: 0:39:51
|
| 189 |
+
Epoch 1 | iter 2464 step 77 | loss train: 1.368, val: 1.303 | iter time: 359.86 ms (step) remaining time: 0:39:39
|
| 190 |
+
Epoch 1 | iter 2496 step 78 | loss train: 1.347, val: 1.303 | iter time: 359.10 ms (step) remaining time: 0:39:27
|
| 191 |
+
Epoch 1 | iter 2528 step 79 | loss train: 1.366, val: 1.303 | iter time: 359.23 ms (step) remaining time: 0:39:15
|
| 192 |
+
Epoch 1 | iter 2560 step 80 | loss train: 1.331, val: 1.303 | iter time: 358.89 ms (step) remaining time: 0:39:03
|
| 193 |
+
Epoch 1 | iter 2592 step 81 | loss train: 1.390, val: 1.303 | iter time: 361.51 ms (step) remaining time: 0:38:51
|
| 194 |
+
Epoch 1 | iter 2624 step 82 | loss train: 1.341, val: 1.303 | iter time: 360.83 ms (step) remaining time: 0:38:39
|
| 195 |
+
Epoch 1 | iter 2656 step 83 | loss train: 1.359, val: 1.303 | iter time: 360.05 ms (step) remaining time: 0:38:28
|
| 196 |
+
Epoch 1 | iter 2688 step 84 | loss train: 1.384, val: 1.303 | iter time: 359.42 ms (step) remaining time: 0:38:16
|
| 197 |
+
Epoch 1 | iter 2720 step 85 | loss train: 1.318, val: 1.303 | iter time: 358.95 ms (step) remaining time: 0:38:04
|
| 198 |
+
Epoch 1 | iter 2752 step 86 | loss train: 1.457, val: 1.303 | iter time: 359.61 ms (step) remaining time: 0:37:52
|
| 199 |
+
Epoch 1 | iter 2784 step 87 | loss train: 1.348, val: 1.303 | iter time: 358.33 ms (step) remaining time: 0:37:40
|
| 200 |
+
Epoch 1 | iter 2816 step 88 | loss train: 1.373, val: 1.303 | iter time: 359.67 ms (step) remaining time: 0:37:29
|
| 201 |
+
Epoch 1 | iter 2848 step 89 | loss train: 1.376, val: 1.303 | iter time: 359.15 ms (step) remaining time: 0:37:17
|
| 202 |
+
Epoch 1 | iter 2880 step 90 | loss train: 1.308, val: 1.303 | iter time: 360.99 ms (step) remaining time: 0:37:05
|
| 203 |
+
Epoch 1 | iter 2912 step 91 | loss train: 1.350, val: 1.303 | iter time: 359.87 ms (step) remaining time: 0:36:54
|
| 204 |
+
Epoch 1 | iter 2944 step 92 | loss train: 1.389, val: 1.303 | iter time: 360.11 ms (step) remaining time: 0:36:42
|
| 205 |
+
Epoch 1 | iter 2976 step 93 | loss train: 1.359, val: 1.303 | iter time: 360.67 ms (step) remaining time: 0:36:30
|
| 206 |
+
Epoch 1 | iter 3008 step 94 | loss train: 1.400, val: 1.303 | iter time: 359.70 ms (step) remaining time: 0:36:19
|
| 207 |
+
Epoch 1 | iter 3040 step 95 | loss train: 1.350, val: 1.303 | iter time: 360.29 ms (step) remaining time: 0:36:07
|
| 208 |
+
Epoch 1 | iter 3072 step 96 | loss train: 1.348, val: 1.303 | iter time: 361.10 ms (step) remaining time: 0:35:55
|
| 209 |
+
Epoch 1 | iter 3104 step 97 | loss train: 1.299, val: 1.303 | iter time: 360.33 ms (step) remaining time: 0:35:44
|
| 210 |
+
Epoch 1 | iter 3136 step 98 | loss train: 1.291, val: 1.303 | iter time: 360.33 ms (step) remaining time: 0:35:32
|
| 211 |
+
Epoch 1 | iter 3168 step 99 | loss train: 1.389, val: 1.303 | iter time: 359.23 ms (step) remaining time: 0:35:21
|
| 212 |
+
Epoch 1 | iter 3200 step 100 | loss train: 1.363, val: 1.303 | iter time: 359.37 ms (step) remaining time: 0:35:09
|
| 213 |
+
Validating ...
|
| 214 |
+
iter 3200: val loss 1.2997, val time: 21941.50 ms
|
| 215 |
+
Saving checkpoint to 'out/pretrain/tinyllama/2412_lr4e-5/step-00000100/lit_model.pth'
|
| 216 |
+
Epoch 1 | iter 3232 step 101 | loss train: 1.409, val: 1.300 | iter time: 356.05 ms (step) remaining time: 0:36:09
|
| 217 |
+
Epoch 1 | iter 3264 step 102 | loss train: 1.378, val: 1.300 | iter time: 358.00 ms (step) remaining time: 0:35:56
|
| 218 |
+
Epoch 1 | iter 3296 step 103 | loss train: 1.386, val: 1.300 | iter time: 360.82 ms (step) remaining time: 0:35:43
|
| 219 |
+
Epoch 1 | iter 3328 step 104 | loss train: 1.346, val: 1.300 | iter time: 359.41 ms (step) remaining time: 0:35:31
|
| 220 |
+
Epoch 1 | iter 3360 step 105 | loss train: 1.372, val: 1.300 | iter time: 360.93 ms (step) remaining time: 0:35:18
|
| 221 |
+
Epoch 1 | iter 3392 step 106 | loss train: 1.320, val: 1.300 | iter time: 359.57 ms (step) remaining time: 0:35:06
|
| 222 |
+
Epoch 1 | iter 3424 step 107 | loss train: 1.369, val: 1.300 | iter time: 358.71 ms (step) remaining time: 0:34:53
|
| 223 |
+
Epoch 1 | iter 3456 step 108 | loss train: 1.364, val: 1.300 | iter time: 359.31 ms (step) remaining time: 0:34:41
|
| 224 |
+
Epoch 1 | iter 3488 step 109 | loss train: 1.312, val: 1.300 | iter time: 361.07 ms (step) remaining time: 0:34:28
|
| 225 |
+
Epoch 1 | iter 3520 step 110 | loss train: 1.353, val: 1.300 | iter time: 358.59 ms (step) remaining time: 0:34:16
|
| 226 |
+
Epoch 1 | iter 3552 step 111 | loss train: 1.331, val: 1.300 | iter time: 359.16 ms (step) remaining time: 0:34:04
|
| 227 |
+
Epoch 1 | iter 3584 step 112 | loss train: 1.360, val: 1.300 | iter time: 357.47 ms (step) remaining time: 0:33:51
|
| 228 |
+
Epoch 1 | iter 3616 step 113 | loss train: 1.284, val: 1.300 | iter time: 358.67 ms (step) remaining time: 0:33:40
|
| 229 |
+
Epoch 1 | iter 3648 step 114 | loss train: 1.303, val: 1.300 | iter time: 360.30 ms (step) remaining time: 0:33:27
|
| 230 |
+
Epoch 1 | iter 3680 step 115 | loss train: 1.309, val: 1.300 | iter time: 360.21 ms (step) remaining time: 0:33:15
|
| 231 |
+
Epoch 1 | iter 3712 step 116 | loss train: 1.314, val: 1.300 | iter time: 360.69 ms (step) remaining time: 0:33:03
|
| 232 |
+
Epoch 1 | iter 3744 step 117 | loss train: 1.395, val: 1.300 | iter time: 359.26 ms (step) remaining time: 0:32:51
|
| 233 |
+
Epoch 1 | iter 3776 step 118 | loss train: 1.342, val: 1.300 | iter time: 358.67 ms (step) remaining time: 0:32:38
|
| 234 |
+
Epoch 1 | iter 3808 step 119 | loss train: 1.335, val: 1.300 | iter time: 360.89 ms (step) remaining time: 0:32:26
|
| 235 |
+
Epoch 1 | iter 3840 step 120 | loss train: 1.376, val: 1.300 | iter time: 359.61 ms (step) remaining time: 0:32:14
|
| 236 |
+
Epoch 1 | iter 3872 step 121 | loss train: 1.390, val: 1.300 | iter time: 360.92 ms (step) remaining time: 0:32:02
|
| 237 |
+
Epoch 1 | iter 3904 step 122 | loss train: 1.353, val: 1.300 | iter time: 362.29 ms (step) remaining time: 0:31:50
|
| 238 |
+
Epoch 1 | iter 3936 step 123 | loss train: 1.321, val: 1.300 | iter time: 358.52 ms (step) remaining time: 0:31:38
|
| 239 |
+
Epoch 1 | iter 3968 step 124 | loss train: 1.308, val: 1.300 | iter time: 359.64 ms (step) remaining time: 0:31:26
|
| 240 |
+
Epoch 1 | iter 4000 step 125 | loss train: 1.405, val: 1.300 | iter time: 361.00 ms (step) remaining time: 0:31:15
|
| 241 |
+
Epoch 1 | iter 4032 step 126 | loss train: 1.355, val: 1.300 | iter time: 359.20 ms (step) remaining time: 0:31:03
|
| 242 |
+
Epoch 1 | iter 4064 step 127 | loss train: 1.379, val: 1.300 | iter time: 360.45 ms (step) remaining time: 0:30:51
|
| 243 |
+
Epoch 1 | iter 4096 step 128 | loss train: 1.381, val: 1.300 | iter time: 354.76 ms (step) remaining time: 0:30:39
|
| 244 |
+
Epoch 1 | iter 4128 step 129 | loss train: 1.356, val: 1.300 | iter time: 360.52 ms (step) remaining time: 0:30:27
|
| 245 |
+
Epoch 1 | iter 4160 step 130 | loss train: 1.332, val: 1.300 | iter time: 360.45 ms (step) remaining time: 0:30:15
|
| 246 |
+
Epoch 1 | iter 4192 step 131 | loss train: 1.392, val: 1.300 | iter time: 360.49 ms (step) remaining time: 0:30:03
|
| 247 |
+
Epoch 1 | iter 4224 step 132 | loss train: 1.309, val: 1.300 | iter time: 361.96 ms (step) remaining time: 0:29:51
|
| 248 |
+
Epoch 1 | iter 4256 step 133 | loss train: 1.348, val: 1.300 | iter time: 358.25 ms (step) remaining time: 0:29:39
|
| 249 |
+
Epoch 1 | iter 4288 step 134 | loss train: 1.386, val: 1.300 | iter time: 358.28 ms (step) remaining time: 0:29:27
|
| 250 |
+
Epoch 1 | iter 4320 step 135 | loss train: 1.332, val: 1.300 | iter time: 358.33 ms (step) remaining time: 0:29:15
|
| 251 |
+
Epoch 1 | iter 4352 step 136 | loss train: 1.395, val: 1.300 | iter time: 358.01 ms (step) remaining time: 0:29:04
|
| 252 |
+
Epoch 1 | iter 4384 step 137 | loss train: 1.333, val: 1.300 | iter time: 359.89 ms (step) remaining time: 0:28:52
|
| 253 |
+
Epoch 1 | iter 4416 step 138 | loss train: 1.349, val: 1.300 | iter time: 360.63 ms (step) remaining time: 0:28:40
|
| 254 |
+
Epoch 1 | iter 4448 step 139 | loss train: 1.407, val: 1.300 | iter time: 361.40 ms (step) remaining time: 0:28:28
|
| 255 |
+
Epoch 1 | iter 4480 step 140 | loss train: 1.380, val: 1.300 | iter time: 359.52 ms (step) remaining time: 0:28:16
|
| 256 |
+
Epoch 1 | iter 4512 step 141 | loss train: 1.394, val: 1.300 | iter time: 358.69 ms (step) remaining time: 0:28:05
|
| 257 |
+
Epoch 1 | iter 4544 step 142 | loss train: 1.403, val: 1.300 | iter time: 357.73 ms (step) remaining time: 0:27:53
|
| 258 |
+
Epoch 1 | iter 4576 step 143 | loss train: 1.450, val: 1.300 | iter time: 359.22 ms (step) remaining time: 0:27:41
|
| 259 |
+
Epoch 1 | iter 4608 step 144 | loss train: 1.413, val: 1.300 | iter time: 358.98 ms (step) remaining time: 0:27:29
|
| 260 |
+
Epoch 1 | iter 4640 step 145 | loss train: 1.424, val: 1.300 | iter time: 360.01 ms (step) remaining time: 0:27:18
|
| 261 |
+
Epoch 1 | iter 4672 step 146 | loss train: 1.338, val: 1.300 | iter time: 359.50 ms (step) remaining time: 0:27:06
|
| 262 |
+
Epoch 1 | iter 4704 step 147 | loss train: 1.383, val: 1.300 | iter time: 359.39 ms (step) remaining time: 0:26:54
|
| 263 |
+
Epoch 1 | iter 4736 step 148 | loss train: 1.414, val: 1.300 | iter time: 358.73 ms (step) remaining time: 0:26:43
|
| 264 |
+
Epoch 1 | iter 4768 step 149 | loss train: 1.394, val: 1.300 | iter time: 359.10 ms (step) remaining time: 0:26:31
|
| 265 |
+
Epoch 1 | iter 4800 step 150 | loss train: 1.386, val: 1.300 | iter time: 359.23 ms (step) remaining time: 0:26:19
|
| 266 |
+
Validating ...
|
| 267 |
+
iter 4800: val loss 1.2971, val time: 21947.04 ms
|
| 268 |
+
Epoch 1 | iter 4832 step 151 | loss train: 1.409, val: 1.297 | iter time: 359.90 ms (step) remaining time: 0:26:28
|
| 269 |
+
Epoch 1 | iter 4864 step 152 | loss train: 1.348, val: 1.297 | iter time: 359.18 ms (step) remaining time: 0:26:16
|
| 270 |
+
Epoch 1 | iter 4896 step 153 | loss train: 1.340, val: 1.297 | iter time: 358.93 ms (step) remaining time: 0:26:04
|
| 271 |
+
Epoch 1 | iter 4928 step 154 | loss train: 1.445, val: 1.297 | iter time: 360.32 ms (step) remaining time: 0:25:52
|
| 272 |
+
Epoch 1 | iter 4960 step 155 | loss train: 1.437, val: 1.297 | iter time: 359.25 ms (step) remaining time: 0:25:40
|
| 273 |
+
Epoch 1 | iter 4992 step 156 | loss train: 1.339, val: 1.297 | iter time: 359.60 ms (step) remaining time: 0:25:28
|
| 274 |
+
Epoch 1 | iter 5024 step 157 | loss train: 1.345, val: 1.297 | iter time: 360.62 ms (step) remaining time: 0:25:16
|
| 275 |
+
Epoch 1 | iter 5056 step 158 | loss train: 1.361, val: 1.297 | iter time: 358.48 ms (step) remaining time: 0:25:05
|
| 276 |
+
Epoch 1 | iter 5088 step 159 | loss train: 1.368, val: 1.297 | iter time: 360.55 ms (step) remaining time: 0:24:53
|
| 277 |
+
Epoch 1 | iter 5120 step 160 | loss train: 1.329, val: 1.297 | iter time: 360.70 ms (step) remaining time: 0:24:41
|
| 278 |
+
Epoch 1 | iter 5152 step 161 | loss train: 1.374, val: 1.297 | iter time: 357.78 ms (step) remaining time: 0:24:29
|
| 279 |
+
Epoch 1 | iter 5184 step 162 | loss train: 1.464, val: 1.297 | iter time: 360.20 ms (step) remaining time: 0:24:17
|
| 280 |
+
Epoch 1 | iter 5216 step 163 | loss train: 1.368, val: 1.297 | iter time: 360.98 ms (step) remaining time: 0:24:05
|
| 281 |
+
Epoch 1 | iter 5248 step 164 | loss train: 1.447, val: 1.297 | iter time: 360.17 ms (step) remaining time: 0:23:54
|
| 282 |
+
Epoch 1 | iter 5280 step 165 | loss train: 1.434, val: 1.297 | iter time: 358.76 ms (step) remaining time: 0:23:42
|
| 283 |
+
Epoch 1 | iter 5312 step 166 | loss train: 1.419, val: 1.297 | iter time: 359.84 ms (step) remaining time: 0:23:30
|
| 284 |
+
Epoch 1 | iter 5344 step 167 | loss train: 1.411, val: 1.297 | iter time: 359.60 ms (step) remaining time: 0:23:18
|
| 285 |
+
Epoch 1 | iter 5376 step 168 | loss train: 1.357, val: 1.297 | iter time: 359.87 ms (step) remaining time: 0:23:07
|
| 286 |
+
Epoch 1 | iter 5408 step 169 | loss train: 1.365, val: 1.297 | iter time: 359.00 ms (step) remaining time: 0:22:55
|
| 287 |
+
Epoch 1 | iter 5440 step 170 | loss train: 1.300, val: 1.297 | iter time: 360.21 ms (step) remaining time: 0:22:43
|
| 288 |
+
Epoch 1 | iter 5472 step 171 | loss train: 1.318, val: 1.297 | iter time: 361.20 ms (step) remaining time: 0:22:32
|
| 289 |
+
Epoch 1 | iter 5504 step 172 | loss train: 1.367, val: 1.297 | iter time: 360.54 ms (step) remaining time: 0:22:20
|
| 290 |
+
Epoch 1 | iter 5536 step 173 | loss train: 1.385, val: 1.297 | iter time: 358.10 ms (step) remaining time: 0:22:08
|
| 291 |
+
Epoch 1 | iter 5568 step 174 | loss train: 1.393, val: 1.297 | iter time: 359.94 ms (step) remaining time: 0:21:56
|
| 292 |
+
Epoch 1 | iter 5600 step 175 | loss train: 1.387, val: 1.297 | iter time: 358.91 ms (step) remaining time: 0:21:45
|
| 293 |
+
Epoch 1 | iter 5632 step 176 | loss train: 1.371, val: 1.297 | iter time: 358.64 ms (step) remaining time: 0:21:33
|
| 294 |
+
Epoch 1 | iter 5664 step 177 | loss train: 1.347, val: 1.297 | iter time: 361.96 ms (step) remaining time: 0:21:21
|
| 295 |
+
Epoch 1 | iter 5696 step 178 | loss train: 1.445, val: 1.297 | iter time: 360.14 ms (step) remaining time: 0:21:10
|
| 296 |
+
Epoch 1 | iter 5728 step 179 | loss train: 1.352, val: 1.297 | iter time: 359.23 ms (step) remaining time: 0:20:58
|
| 297 |
+
Epoch 1 | iter 5760 step 180 | loss train: 1.355, val: 1.297 | iter time: 359.65 ms (step) remaining time: 0:20:47
|
| 298 |
+
Epoch 1 | iter 5792 step 181 | loss train: 1.381, val: 1.297 | iter time: 359.39 ms (step) remaining time: 0:20:35
|
| 299 |
+
Epoch 1 | iter 5824 step 182 | loss train: 1.354, val: 1.297 | iter time: 358.25 ms (step) remaining time: 0:20:23
|
| 300 |
+
Epoch 1 | iter 5856 step 183 | loss train: 1.379, val: 1.297 | iter time: 359.01 ms (step) remaining time: 0:20:12
|
| 301 |
+
Epoch 1 | iter 5888 step 184 | loss train: 1.398, val: 1.297 | iter time: 357.69 ms (step) remaining time: 0:20:00
|
| 302 |
+
Epoch 1 | iter 5920 step 185 | loss train: 1.347, val: 1.297 | iter time: 361.26 ms (step) remaining time: 0:19:49
|
| 303 |
+
Epoch 1 | iter 5952 step 186 | loss train: 1.393, val: 1.297 | iter time: 358.92 ms (step) remaining time: 0:19:37
|
| 304 |
+
Epoch 1 | iter 5984 step 187 | loss train: 1.402, val: 1.297 | iter time: 359.91 ms (step) remaining time: 0:19:26
|
| 305 |
+
Epoch 1 | iter 6016 step 188 | loss train: 1.469, val: 1.297 | iter time: 359.22 ms (step) remaining time: 0:19:14
|
| 306 |
+
Epoch 1 | iter 6048 step 189 | loss train: 1.360, val: 1.297 | iter time: 360.93 ms (step) remaining time: 0:19:02
|
| 307 |
+
Epoch 1 | iter 6080 step 190 | loss train: 1.397, val: 1.297 | iter time: 361.50 ms (step) remaining time: 0:18:51
|
| 308 |
+
Epoch 1 | iter 6112 step 191 | loss train: 1.342, val: 1.297 | iter time: 358.85 ms (step) remaining time: 0:18:39
|
| 309 |
+
Epoch 1 | iter 6144 step 192 | loss train: 1.362, val: 1.297 | iter time: 358.55 ms (step) remaining time: 0:18:28
|
| 310 |
+
Epoch 1 | iter 6176 step 193 | loss train: 1.396, val: 1.297 | iter time: 360.94 ms (step) remaining time: 0:18:16
|
| 311 |
+
Epoch 1 | iter 6208 step 194 | loss train: 1.442, val: 1.297 | iter time: 607.16 ms (step) remaining time: 0:18:05
|
| 312 |
+
Epoch 1 | iter 6240 step 195 | loss train: 1.380, val: 1.297 | iter time: 358.51 ms (step) remaining time: 0:17:53
|
| 313 |
+
Epoch 1 | iter 6272 step 196 | loss train: 1.402, val: 1.297 | iter time: 358.22 ms (step) remaining time: 0:17:42
|
| 314 |
+
Epoch 1 | iter 6304 step 197 | loss train: 1.315, val: 1.297 | iter time: 359.23 ms (step) remaining time: 0:17:30
|
| 315 |
+
Epoch 1 | iter 6336 step 198 | loss train: 1.343, val: 1.297 | iter time: 358.04 ms (step) remaining time: 0:17:19
|
| 316 |
+
Epoch 1 | iter 6368 step 199 | loss train: 1.394, val: 1.297 | iter time: 359.15 ms (step) remaining time: 0:17:07
|
| 317 |
+
Epoch 1 | iter 6400 step 200 | loss train: 1.381, val: 1.297 | iter time: 360.85 ms (step) remaining time: 0:16:56
|
| 318 |
+
Validating ...
|
| 319 |
+
iter 6400: val loss 1.2924, val time: 21942.30 ms
|
| 320 |
+
Saving checkpoint to 'out/pretrain/tinyllama/2412_lr4e-5/step-00000200/lit_model.pth'
|
| 321 |
+
Epoch 1 | iter 6432 step 201 | loss train: 1.364, val: 1.292 | iter time: 356.68 ms (step) remaining time: 0:17:01
|
| 322 |
+
Epoch 1 | iter 6464 step 202 | loss train: 1.424, val: 1.292 | iter time: 357.73 ms (step) remaining time: 0:16:49
|
| 323 |
+
Epoch 1 | iter 6496 step 203 | loss train: 1.355, val: 1.292 | iter time: 358.41 ms (step) remaining time: 0:16:38
|
| 324 |
+
Epoch 1 | iter 6528 step 204 | loss train: 1.373, val: 1.292 | iter time: 359.90 ms (step) remaining time: 0:16:26
|
| 325 |
+
Epoch 1 | iter 6560 step 205 | loss train: 1.342, val: 1.292 | iter time: 359.21 ms (step) remaining time: 0:16:14
|
| 326 |
+
Epoch 1 | iter 6592 step 206 | loss train: 1.420, val: 1.292 | iter time: 357.64 ms (step) remaining time: 0:16:02
|
| 327 |
+
Epoch 1 | iter 6624 step 207 | loss train: 1.355, val: 1.292 | iter time: 360.14 ms (step) remaining time: 0:15:51
|
| 328 |
+
Epoch 1 | iter 6656 step 208 | loss train: 1.328, val: 1.292 | iter time: 358.94 ms (step) remaining time: 0:15:39
|
| 329 |
+
Epoch 1 | iter 6688 step 209 | loss train: 1.408, val: 1.292 | iter time: 358.67 ms (step) remaining time: 0:15:27
|
| 330 |
+
Epoch 1 | iter 6720 step 210 | loss train: 1.340, val: 1.292 | iter time: 358.83 ms (step) remaining time: 0:15:16
|
| 331 |
+
Epoch 1 | iter 6752 step 211 | loss train: 1.366, val: 1.292 | iter time: 360.99 ms (step) remaining time: 0:15:04
|
| 332 |
+
Epoch 1 | iter 6784 step 212 | loss train: 1.425, val: 1.292 | iter time: 360.72 ms (step) remaining time: 0:14:52
|
| 333 |
+
Epoch 1 | iter 6816 step 213 | loss train: 1.419, val: 1.292 | iter time: 359.38 ms (step) remaining time: 0:14:41
|
| 334 |
+
Epoch 1 | iter 6848 step 214 | loss train: 1.332, val: 1.292 | iter time: 357.71 ms (step) remaining time: 0:14:29
|
| 335 |
+
Epoch 1 | iter 6880 step 215 | loss train: 1.437, val: 1.292 | iter time: 360.60 ms (step) remaining time: 0:14:17
|
| 336 |
+
Epoch 1 | iter 6912 step 216 | loss train: 1.399, val: 1.292 | iter time: 359.07 ms (step) remaining time: 0:14:06
|
| 337 |
+
Epoch 1 | iter 6944 step 217 | loss train: 1.338, val: 1.292 | iter time: 360.92 ms (step) remaining time: 0:13:54
|
| 338 |
+
Epoch 1 | iter 6976 step 218 | loss train: 1.336, val: 1.292 | iter time: 359.81 ms (step) remaining time: 0:13:43
|
| 339 |
+
Epoch 1 | iter 7008 step 219 | loss train: 1.346, val: 1.292 | iter time: 360.16 ms (step) remaining time: 0:13:31
|
| 340 |
+
Epoch 1 | iter 7040 step 220 | loss train: 1.305, val: 1.292 | iter time: 359.56 ms (step) remaining time: 0:13:19
|
| 341 |
+
Epoch 1 | iter 7072 step 221 | loss train: 1.328, val: 1.292 | iter time: 358.76 ms (step) remaining time: 0:13:08
|
| 342 |
+
Epoch 1 | iter 7104 step 222 | loss train: 1.400, val: 1.292 | iter time: 360.18 ms (step) remaining time: 0:12:56
|
| 343 |
+
Epoch 1 | iter 7136 step 223 | loss train: 1.305, val: 1.292 | iter time: 359.27 ms (step) remaining time: 0:12:45
|
| 344 |
+
Epoch 1 | iter 7168 step 224 | loss train: 1.370, val: 1.292 | iter time: 358.90 ms (step) remaining time: 0:12:33
|
| 345 |
+
Epoch 1 | iter 7200 step 225 | loss train: 1.322, val: 1.292 | iter time: 360.03 ms (step) remaining time: 0:12:21
|
| 346 |
+
Epoch 1 | iter 7232 step 226 | loss train: 1.406, val: 1.292 | iter time: 360.34 ms (step) remaining time: 0:12:10
|
| 347 |
+
Epoch 1 | iter 7264 step 227 | loss train: 1.360, val: 1.292 | iter time: 358.39 ms (step) remaining time: 0:11:58
|
| 348 |
+
Epoch 1 | iter 7296 step 228 | loss train: 1.296, val: 1.292 | iter time: 360.99 ms (step) remaining time: 0:11:47
|
| 349 |
+
Epoch 1 | iter 7328 step 229 | loss train: 1.506, val: 1.292 | iter time: 358.73 ms (step) remaining time: 0:11:35
|
| 350 |
+
Epoch 1 | iter 7360 step 230 | loss train: 1.343, val: 1.292 | iter time: 359.70 ms (step) remaining time: 0:11:24
|
| 351 |
+
Epoch 1 | iter 7392 step 231 | loss train: 1.392, val: 1.292 | iter time: 360.83 ms (step) remaining time: 0:11:12
|
| 352 |
+
Epoch 1 | iter 7424 step 232 | loss train: 1.325, val: 1.292 | iter time: 359.22 ms (step) remaining time: 0:11:01
|
| 353 |
+
Epoch 1 | iter 7456 step 233 | loss train: 1.350, val: 1.292 | iter time: 360.45 ms (step) remaining time: 0:10:49
|
| 354 |
+
Epoch 1 | iter 7488 step 234 | loss train: 1.432, val: 1.292 | iter time: 360.91 ms (step) remaining time: 0:10:38
|
| 355 |
+
Epoch 1 | iter 7520 step 235 | loss train: 1.354, val: 1.292 | iter time: 358.74 ms (step) remaining time: 0:10:26
|
| 356 |
+
Epoch 1 | iter 7552 step 236 | loss train: 1.391, val: 1.292 | iter time: 360.07 ms (step) remaining time: 0:10:14
|
| 357 |
+
Epoch 1 | iter 7584 step 237 | loss train: 1.333, val: 1.292 | iter time: 359.95 ms (step) remaining time: 0:10:03
|
| 358 |
+
Epoch 1 | iter 7616 step 238 | loss train: 1.344, val: 1.292 | iter time: 359.28 ms (step) remaining time: 0:09:51
|
| 359 |
+
Epoch 1 | iter 7648 step 239 | loss train: 1.386, val: 1.292 | iter time: 362.00 ms (step) remaining time: 0:09:40
|
| 360 |
+
Epoch 1 | iter 7680 step 240 | loss train: 1.430, val: 1.292 | iter time: 359.55 ms (step) remaining time: 0:09:29
|
| 361 |
+
Epoch 1 | iter 7712 step 241 | loss train: 1.328, val: 1.292 | iter time: 359.32 ms (step) remaining time: 0:09:17
|
| 362 |
+
Epoch 1 | iter 7744 step 242 | loss train: 1.419, val: 1.292 | iter time: 360.56 ms (step) remaining time: 0:09:06
|
| 363 |
+
Epoch 1 | iter 7776 step 243 | loss train: 1.343, val: 1.292 | iter time: 361.35 ms (step) remaining time: 0:08:54
|
| 364 |
+
Epoch 1 | iter 7808 step 244 | loss train: 1.374, val: 1.292 | iter time: 360.30 ms (step) remaining time: 0:08:43
|
| 365 |
+
Epoch 1 | iter 7840 step 245 | loss train: 1.376, val: 1.292 | iter time: 361.11 ms (step) remaining time: 0:08:31
|
| 366 |
+
Epoch 1 | iter 7872 step 246 | loss train: 1.384, val: 1.292 | iter time: 359.56 ms (step) remaining time: 0:08:20
|
| 367 |
+
Epoch 1 | iter 7904 step 247 | loss train: 1.308, val: 1.292 | iter time: 359.60 ms (step) remaining time: 0:08:08
|
| 368 |
+
Epoch 1 | iter 7936 step 248 | loss train: 1.346, val: 1.292 | iter time: 358.91 ms (step) remaining time: 0:07:57
|
| 369 |
+
Epoch 1 | iter 7968 step 249 | loss train: 1.375, val: 1.292 | iter time: 359.99 ms (step) remaining time: 0:07:45
|
| 370 |
+
Epoch 1 | iter 8000 step 250 | loss train: 1.336, val: 1.292 | iter time: 360.09 ms (step) remaining time: 0:07:34
|
| 371 |
+
Validating ...
|
| 372 |
+
iter 8000: val loss 1.2909, val time: 21936.55 ms
|
| 373 |
+
Epoch 1 | iter 8032 step 251 | loss train: 1.326, val: 1.291 | iter time: 360.83 ms (step) remaining time: 0:07:26
|
| 374 |
+
Epoch 1 | iter 8064 step 252 | loss train: 1.370, val: 1.291 | iter time: 359.27 ms (step) remaining time: 0:07:14
|
| 375 |
+
Epoch 1 | iter 8096 step 253 | loss train: 1.377, val: 1.291 | iter time: 358.56 ms (step) remaining time: 0:07:03
|
| 376 |
+
Epoch 1 | iter 8128 step 254 | loss train: 1.378, val: 1.291 | iter time: 358.74 ms (step) remaining time: 0:06:51
|
| 377 |
+
Epoch 1 | iter 8160 step 255 | loss train: 1.377, val: 1.291 | iter time: 358.21 ms (step) remaining time: 0:06:40
|
| 378 |
+
Epoch 1 | iter 8192 step 256 | loss train: 1.409, val: 1.291 | iter time: 358.67 ms (step) remaining time: 0:06:28
|
| 379 |
+
Epoch 1 | iter 8224 step 257 | loss train: 1.385, val: 1.291 | iter time: 358.52 ms (step) remaining time: 0:06:17
|
| 380 |
+
Epoch 1 | iter 8256 step 258 | loss train: 1.357, val: 1.291 | iter time: 359.40 ms (step) remaining time: 0:06:05
|
| 381 |
+
Epoch 1 | iter 8288 step 259 | loss train: 1.464, val: 1.291 | iter time: 360.44 ms (step) remaining time: 0:05:54
|
| 382 |
+
Epoch 1 | iter 8320 step 260 | loss train: 1.405, val: 1.291 | iter time: 360.00 ms (step) remaining time: 0:05:42
|
| 383 |
+
Epoch 1 | iter 8352 step 261 | loss train: 1.397, val: 1.291 | iter time: 358.48 ms (step) remaining time: 0:05:31
|
| 384 |
+
Epoch 1 | iter 8384 step 262 | loss train: 1.367, val: 1.291 | iter time: 358.87 ms (step) remaining time: 0:05:19
|
| 385 |
+
Epoch 1 | iter 8416 step 263 | loss train: 1.448, val: 1.291 | iter time: 356.68 ms (step) remaining time: 0:05:08
|
| 386 |
+
Epoch 1 | iter 8448 step 264 | loss train: 1.280, val: 1.291 | iter time: 360.27 ms (step) remaining time: 0:04:56
|
| 387 |
+
Epoch 1 | iter 8480 step 265 | loss train: 1.335, val: 1.291 | iter time: 360.53 ms (step) remaining time: 0:04:45
|
| 388 |
+
Epoch 1 | iter 8512 step 266 | loss train: 1.367, val: 1.291 | iter time: 357.93 ms (step) remaining time: 0:04:33
|
| 389 |
+
Epoch 1 | iter 8544 step 267 | loss train: 1.408, val: 1.291 | iter time: 360.38 ms (step) remaining time: 0:04:22
|
| 390 |
+
Epoch 1 | iter 8576 step 268 | loss train: 1.362, val: 1.291 | iter time: 359.66 ms (step) remaining time: 0:04:11
|
| 391 |
+
Epoch 1 | iter 8608 step 269 | loss train: 1.410, val: 1.291 | iter time: 360.35 ms (step) remaining time: 0:03:59
|
| 392 |
+
Epoch 1 | iter 8640 step 270 | loss train: 1.342, val: 1.291 | iter time: 359.21 ms (step) remaining time: 0:03:48
|
| 393 |
+
Epoch 1 | iter 8672 step 271 | loss train: 1.380, val: 1.291 | iter time: 360.04 ms (step) remaining time: 0:03:36
|
| 394 |
+
Epoch 1 | iter 8704 step 272 | loss train: 1.380, val: 1.291 | iter time: 358.57 ms (step) remaining time: 0:03:25
|
| 395 |
+
Epoch 1 | iter 8736 step 273 | loss train: 1.305, val: 1.291 | iter time: 358.47 ms (step) remaining time: 0:03:13
|
| 396 |
+
Epoch 1 | iter 8768 step 274 | loss train: 1.349, val: 1.291 | iter time: 359.82 ms (step) remaining time: 0:03:02
|
| 397 |
+
Epoch 1 | iter 8800 step 275 | loss train: 1.379, val: 1.291 | iter time: 359.03 ms (step) remaining time: 0:02:50
|
| 398 |
+
Epoch 1 | iter 8832 step 276 | loss train: 1.371, val: 1.291 | iter time: 360.44 ms (step) remaining time: 0:02:39
|
| 399 |
+
Epoch 1 | iter 8864 step 277 | loss train: 1.441, val: 1.291 | iter time: 360.93 ms (step) remaining time: 0:02:28
|
| 400 |
+
Epoch 1 | iter 8896 step 278 | loss train: 1.361, val: 1.291 | iter time: 359.28 ms (step) remaining time: 0:02:16
|
| 401 |
+
Epoch 1 | iter 8928 step 279 | loss train: 1.410, val: 1.291 | iter time: 359.84 ms (step) remaining time: 0:02:05
|
| 402 |
+
Epoch 1 | iter 8960 step 280 | loss train: 1.382, val: 1.291 | iter time: 359.06 ms (step) remaining time: 0:01:53
|
| 403 |
+
Epoch 1 | iter 8992 step 281 | loss train: 1.391, val: 1.291 | iter time: 358.93 ms (step) remaining time: 0:01:42
|
| 404 |
+
Epoch 1 | iter 9024 step 282 | loss train: 1.405, val: 1.291 | iter time: 361.64 ms (step) remaining time: 0:01:31
|
| 405 |
+
Epoch 1 | iter 9056 step 283 | loss train: 1.283, val: 1.291 | iter time: 360.10 ms (step) remaining time: 0:01:19
|
| 406 |
+
Epoch 1 | iter 9088 step 284 | loss train: 1.435, val: 1.291 | iter time: 359.89 ms (step) remaining time: 0:01:08
|
| 407 |
+
Epoch 1 | iter 9120 step 285 | loss train: 1.389, val: 1.291 | iter time: 359.18 ms (step) remaining time: 0:00:56
|
| 408 |
+
Epoch 1 | iter 9152 step 286 | loss train: 1.360, val: 1.291 | iter time: 359.35 ms (step) remaining time: 0:00:45
|
| 409 |
+
Epoch 1 | iter 9184 step 287 | loss train: 1.383, val: 1.291 | iter time: 359.96 ms (step) remaining time: 0:00:34
|
| 410 |
+
Epoch 1 | iter 9216 step 288 | loss train: 1.421, val: 1.291 | iter time: 360.37 ms (step) remaining time: 0:00:22
|
| 411 |
+
Epoch 2 | iter 9248 step 289 | loss train: 1.394, val: 1.291 | iter time: 357.32 ms (step) remaining time: 0:00:11
|
| 412 |
+
Epoch 2 | iter 9280 step 290 | loss train: 1.326, val: 1.291 | iter time: 361.11 ms (step) remaining time: 0:00:00
|
| 413 |
+
Validating ...
|
| 414 |
+
Final evaluation | val loss: 1.227 | val ppl: 3.412
|
| 415 |
+
Saving checkpoint to 'out/pretrain/tinyllama/2412_lr4e-5/final/lit_model.pth'
|
| 416 |
+
----------------------------------------
|
| 417 |
+
| Performance
|
| 418 |
+
| - Total tokens : 304,087,040
|
| 419 |
+
| - Training Time : 3359.33 s
|
| 420 |
+
| - Tok/sec : 156.37 tok/s
|
| 421 |
+
| ----------------------------------------
|
| 422 |
+
| Memory Usage
|
| 423 |
+
| - Memory Used : 26.32 GB
|
| 424 |
+
----------------------------------------
|
out/pretrain/tinyllama/teelogs/2501.txt
ADDED
|
@@ -0,0 +1,378 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
Initializing distributed: GLOBAL_RANK: 0, MEMBER: 1/4
|
| 2 |
+
Initializing distributed: GLOBAL_RANK: 1, MEMBER: 2/4
|
| 3 |
+
[rank: 1] Seed set to 42
|
| 4 |
+
Initializing distributed: GLOBAL_RANK: 3, MEMBER: 4/4
|
| 5 |
+
[rank: 3] Seed set to 42
|
| 6 |
+
Initializing distributed: GLOBAL_RANK: 2, MEMBER: 3/4
|
| 7 |
+
[rank: 2] Seed set to 42
|
| 8 |
+
----------------------------------------------------------------------------------------------------
|
| 9 |
+
distributed_backend=nccl
|
| 10 |
+
All distributed processes registered. Starting with 4 processes
|
| 11 |
+
----------------------------------------------------------------------------------------------------
|
| 12 |
+
|
| 13 |
+
[rank: 0] Seed set to 42
|
| 14 |
+
All GPUs are fully connected via NVLink.
|
| 15 |
+
{'data': {'batch_size': 1,
|
| 16 |
+
'data_path': PosixPath('litgpt/data/arxiv'),
|
| 17 |
+
'num_workers': 8,
|
| 18 |
+
'seed': 42,
|
| 19 |
+
'seq_length': 2048,
|
| 20 |
+
'use_starcoder': True},
|
| 21 |
+
'data_dir': PosixPath('litgpt/data/arxiv/2501'),
|
| 22 |
+
'devices': 'auto',
|
| 23 |
+
'eval': {'evaluate_example': 'first',
|
| 24 |
+
'final_validation': True,
|
| 25 |
+
'initial_validation': True,
|
| 26 |
+
'interval': 50,
|
| 27 |
+
'max_iters': 100,
|
| 28 |
+
'max_new_tokens': None},
|
| 29 |
+
'initial_checkpoint_dir': PosixPath('out/pretrain/2412/final'),
|
| 30 |
+
'log': {'group': None, 'project': None, 'run': None},
|
| 31 |
+
'logger_name': 'tensorboard',
|
| 32 |
+
'model_config': {'attention_logit_softcapping': None,
|
| 33 |
+
'attention_scores_scalar': None,
|
| 34 |
+
'attn_bias': False,
|
| 35 |
+
'bias': False,
|
| 36 |
+
'block_size': 2048,
|
| 37 |
+
'final_logit_softcapping': None,
|
| 38 |
+
'gelu_approximate': 'none',
|
| 39 |
+
'head_size': 64,
|
| 40 |
+
'hf_config': {'name': 'TinyLlama-1.1B-intermediate-step-1431k-3T',
|
| 41 |
+
'org': 'TinyLlama'},
|
| 42 |
+
'intermediate_size': 5632,
|
| 43 |
+
'lm_head_bias': False,
|
| 44 |
+
'mlp_class_name': 'LLaMAMLP',
|
| 45 |
+
'moe_intermediate_size': None,
|
| 46 |
+
'n_embd': 2048,
|
| 47 |
+
'n_expert': 0,
|
| 48 |
+
'n_expert_per_token': 0,
|
| 49 |
+
'n_head': 32,
|
| 50 |
+
'n_layer': 22,
|
| 51 |
+
'n_query_groups': 4,
|
| 52 |
+
'name': 'tiny-llama-1.1b',
|
| 53 |
+
'norm_1': True,
|
| 54 |
+
'norm_2': True,
|
| 55 |
+
'norm_class_name': 'RMSNorm',
|
| 56 |
+
'norm_eps': 1e-05,
|
| 57 |
+
'norm_qk': False,
|
| 58 |
+
'norm_qk_type': 'default',
|
| 59 |
+
'padded_vocab_size': 32000,
|
| 60 |
+
'padding_multiple': 64,
|
| 61 |
+
'parallel_residual': False,
|
| 62 |
+
'post_attention_norm': False,
|
| 63 |
+
'post_mlp_norm': False,
|
| 64 |
+
'rope_adjustments': None,
|
| 65 |
+
'rope_base': 10000,
|
| 66 |
+
'rope_condense_ratio': 1,
|
| 67 |
+
'rope_indices': None,
|
| 68 |
+
'rope_local_base_freq': None,
|
| 69 |
+
'rotary_percentage': 1.0,
|
| 70 |
+
'scale_embeddings': False,
|
| 71 |
+
'shared_attention_norm': False,
|
| 72 |
+
'sliding_window_indices': None,
|
| 73 |
+
'sliding_window_size': None,
|
| 74 |
+
'vocab_size': 32000},
|
| 75 |
+
'model_name': 'tiny-llama-1.1b',
|
| 76 |
+
'num_nodes': 1,
|
| 77 |
+
'optimizer': "{'class_path': 'torch.optim.AdamW', 'init_args': {'lr': 0.0004, "
|
| 78 |
+
"'weight_decay': 0.1, 'betas': [0.9, 0.95]}}",
|
| 79 |
+
'out_dir': PosixPath('out/pretrain/2501'),
|
| 80 |
+
'precision': 'bf16-mixed',
|
| 81 |
+
'resume': False,
|
| 82 |
+
'seed': 42,
|
| 83 |
+
'tokenizer_dir': PosixPath('checkpoints/TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T'),
|
| 84 |
+
'train': {'epochs': None,
|
| 85 |
+
'global_batch_size': 512,
|
| 86 |
+
'log_interval': 1,
|
| 87 |
+
'lr_warmup_fraction': None,
|
| 88 |
+
'lr_warmup_steps': 20,
|
| 89 |
+
'max_norm': 1.0,
|
| 90 |
+
'max_seq_length': 2048,
|
| 91 |
+
'max_steps': None,
|
| 92 |
+
'max_tokens': 264241152,
|
| 93 |
+
'micro_batch_size': 4,
|
| 94 |
+
'min_lr': 4e-05,
|
| 95 |
+
'save_interval': 100,
|
| 96 |
+
'tie_embeddings': None}}
|
| 97 |
+
Time to instantiate model: 0.02 seconds.
|
| 98 |
+
Total parameters: 1,100,048,384
|
| 99 |
+
[fix] out/pretrain/2412/final/lit_model.pth 是整包 state，提取 model 权重并原子覆盖...
|
| 100 |
+
[fix] 已覆盖为纯权重: out/pretrain/2412/final/lit_model.pth
|
| 101 |
+
Validating ...
|
| 102 |
+
Measured TFLOPs: 239.66
|
| 103 |
+
Epoch 1 | iter 32 step 1 | loss train: 1.400, val: 1.332 | iter time: 542.67 ms (step) remaining time: 0:49:50
|
| 104 |
+
Epoch 1 | iter 64 step 2 | loss train: 1.346, val: 1.332 | iter time: 354.95 ms (step) remaining time: 0:47:12
|
| 105 |
+
Epoch 1 | iter 96 step 3 | loss train: 1.349, val: 1.332 | iter time: 356.42 ms (step) remaining time: 0:46:14
|
| 106 |
+
Epoch 1 | iter 128 step 4 | loss train: 1.348, val: 1.332 | iter time: 359.29 ms (step) remaining time: 0:45:41
|
| 107 |
+
Epoch 1 | iter 160 step 5 | loss train: 1.309, val: 1.332 | iter time: 357.98 ms (step) remaining time: 0:45:17
|
| 108 |
+
Epoch 1 | iter 192 step 6 | loss train: 1.392, val: 1.332 | iter time: 357.45 ms (step) remaining time: 0:44:59
|
| 109 |
+
Epoch 1 | iter 224 step 7 | loss train: 1.365, val: 1.332 | iter time: 359.96 ms (step) remaining time: 0:44:43
|
| 110 |
+
Epoch 1 | iter 256 step 8 | loss train: 1.357, val: 1.332 | iter time: 360.01 ms (step) remaining time: 0:44:29
|
| 111 |
+
Epoch 1 | iter 288 step 9 | loss train: 1.361, val: 1.332 | iter time: 359.18 ms (step) remaining time: 0:44:15
|
| 112 |
+
Epoch 1 | iter 320 step 10 | loss train: 1.356, val: 1.332 | iter time: 358.22 ms (step) remaining time: 0:44:02
|
| 113 |
+
Epoch 1 | iter 352 step 11 | loss train: 1.460, val: 1.332 | iter time: 357.89 ms (step) remaining time: 0:43:50
|
| 114 |
+
Epoch 1 | iter 384 step 12 | loss train: 1.357, val: 1.332 | iter time: 358.98 ms (step) remaining time: 0:43:38
|
| 115 |
+
Epoch 1 | iter 416 step 13 | loss train: 1.317, val: 1.332 | iter time: 359.74 ms (step) remaining time: 0:43:26
|
| 116 |
+
Epoch 1 | iter 448 step 14 | loss train: 1.365, val: 1.332 | iter time: 359.22 ms (step) remaining time: 0:43:14
|
| 117 |
+
Epoch 1 | iter 480 step 15 | loss train: 1.356, val: 1.332 | iter time: 360.00 ms (step) remaining time: 0:43:02
|
| 118 |
+
Epoch 1 | iter 512 step 16 | loss train: 1.438, val: 1.332 | iter time: 358.80 ms (step) remaining time: 0:42:51
|
| 119 |
+
Epoch 1 | iter 544 step 17 | loss train: 1.357, val: 1.332 | iter time: 360.71 ms (step) remaining time: 0:42:39
|
| 120 |
+
Epoch 1 | iter 576 step 18 | loss train: 1.392, val: 1.332 | iter time: 359.65 ms (step) remaining time: 0:42:28
|
| 121 |
+
Epoch 1 | iter 608 step 19 | loss train: 1.368, val: 1.332 | iter time: 357.81 ms (step) remaining time: 0:42:17
|
| 122 |
+
Epoch 1 | iter 640 step 20 | loss train: 1.386, val: 1.332 | iter time: 358.55 ms (step) remaining time: 0:42:06
|
| 123 |
+
Epoch 1 | iter 672 step 21 | loss train: 1.378, val: 1.332 | iter time: 359.05 ms (step) remaining time: 0:41:54
|
| 124 |
+
Epoch 1 | iter 704 step 22 | loss train: 1.421, val: 1.332 | iter time: 358.99 ms (step) remaining time: 0:41:43
|
| 125 |
+
Epoch 1 | iter 736 step 23 | loss train: 1.430, val: 1.332 | iter time: 358.35 ms (step) remaining time: 0:41:32
|
| 126 |
+
Epoch 1 | iter 768 step 24 | loss train: 1.417, val: 1.332 | iter time: 360.41 ms (step) remaining time: 0:41:21
|
| 127 |
+
Epoch 1 | iter 800 step 25 | loss train: 1.410, val: 1.332 | iter time: 360.08 ms (step) remaining time: 0:41:10
|
| 128 |
+
Epoch 1 | iter 832 step 26 | loss train: 1.370, val: 1.332 | iter time: 359.93 ms (step) remaining time: 0:40:59
|
| 129 |
+
Epoch 1 | iter 864 step 27 | loss train: 1.462, val: 1.332 | iter time: 358.41 ms (step) remaining time: 0:40:48
|
| 130 |
+
Epoch 1 | iter 896 step 28 | loss train: 1.384, val: 1.332 | iter time: 360.39 ms (step) remaining time: 0:40:37
|
| 131 |
+
Epoch 1 | iter 928 step 29 | loss train: 1.419, val: 1.332 | iter time: 359.62 ms (step) remaining time: 0:40:26
|
| 132 |
+
Epoch 1 | iter 960 step 30 | loss train: 1.397, val: 1.332 | iter time: 360.96 ms (step) remaining time: 0:40:15
|
| 133 |
+
Epoch 1 | iter 992 step 31 | loss train: 1.438, val: 1.332 | iter time: 359.77 ms (step) remaining time: 0:40:04
|
| 134 |
+
Epoch 1 | iter 1024 step 32 | loss train: 1.413, val: 1.332 | iter time: 358.74 ms (step) remaining time: 0:39:54
|
| 135 |
+
Epoch 1 | iter 1056 step 33 | loss train: 1.313, val: 1.332 | iter time: 358.83 ms (step) remaining time: 0:39:43
|
| 136 |
+
Epoch 1 | iter 1088 step 34 | loss train: 1.467, val: 1.332 | iter time: 359.99 ms (step) remaining time: 0:39:32
|
| 137 |
+
Epoch 1 | iter 1120 step 35 | loss train: 1.367, val: 1.332 | iter time: 360.26 ms (step) remaining time: 0:39:21
|
| 138 |
+
Epoch 1 | iter 1152 step 36 | loss train: 1.371, val: 1.332 | iter time: 358.26 ms (step) remaining time: 0:39:10
|
| 139 |
+
Epoch 1 | iter 1184 step 37 | loss train: 1.384, val: 1.332 | iter time: 360.89 ms (step) remaining time: 0:38:59
|
| 140 |
+
Epoch 1 | iter 1216 step 38 | loss train: 1.473, val: 1.332 | iter time: 360.28 ms (step) remaining time: 0:38:48
|
| 141 |
+
Epoch 1 | iter 1248 step 39 | loss train: 1.339, val: 1.332 | iter time: 567.37 ms (step) remaining time: 0:38:38
|
| 142 |
+
Epoch 1 | iter 1280 step 40 | loss train: 1.378, val: 1.332 | iter time: 359.83 ms (step) remaining time: 0:38:27
|
| 143 |
+
Epoch 1 | iter 1312 step 41 | loss train: 1.402, val: 1.332 | iter time: 359.55 ms (step) remaining time: 0:38:16
|
| 144 |
+
Epoch 1 | iter 1344 step 42 | loss train: 1.378, val: 1.332 | iter time: 359.78 ms (step) remaining time: 0:38:05
|
| 145 |
+
Epoch 1 | iter 1376 step 43 | loss train: 1.412, val: 1.332 | iter time: 358.24 ms (step) remaining time: 0:37:54
|
| 146 |
+
Epoch 1 | iter 1408 step 44 | loss train: 1.351, val: 1.332 | iter time: 359.95 ms (step) remaining time: 0:37:43
|
| 147 |
+
Epoch 1 | iter 1440 step 45 | loss train: 1.361, val: 1.332 | iter time: 359.66 ms (step) remaining time: 0:37:32
|
| 148 |
+
Epoch 1 | iter 1472 step 46 | loss train: 1.381, val: 1.332 | iter time: 360.26 ms (step) remaining time: 0:37:21
|
| 149 |
+
Epoch 1 | iter 1504 step 47 | loss train: 1.390, val: 1.332 | iter time: 359.66 ms (step) remaining time: 0:37:10
|
| 150 |
+
Epoch 1 | iter 1536 step 48 | loss train: 1.385, val: 1.332 | iter time: 359.21 ms (step) remaining time: 0:36:59
|
| 151 |
+
Epoch 1 | iter 1568 step 49 | loss train: 1.336, val: 1.332 | iter time: 359.84 ms (step) remaining time: 0:36:48
|
| 152 |
+
Epoch 1 | iter 1600 step 50 | loss train: 1.439, val: 1.332 | iter time: 360.70 ms (step) remaining time: 0:36:37
|
| 153 |
+
Validating ...
|
| 154 |
+
iter 1600: val loss 1.3841, val time: 9333.33 ms
|
| 155 |
+
Epoch 1 | iter 1632 step 51 | loss train: 1.399, val: 1.384 | iter time: 359.46 ms (step) remaining time: 0:37:03
|
| 156 |
+
Epoch 1 | iter 1664 step 52 | loss train: 1.382, val: 1.384 | iter time: 358.16 ms (step) remaining time: 0:36:51
|
| 157 |
+
Epoch 1 | iter 1696 step 53 | loss train: 1.403, val: 1.384 | iter time: 358.36 ms (step) remaining time: 0:36:39
|
| 158 |
+
Epoch 1 | iter 1728 step 54 | loss train: 1.356, val: 1.384 | iter time: 361.38 ms (step) remaining time: 0:36:27
|
| 159 |
+
Epoch 1 | iter 1760 step 55 | loss train: 1.366, val: 1.384 | iter time: 359.32 ms (step) remaining time: 0:36:15
|
| 160 |
+
Epoch 1 | iter 1792 step 56 | loss train: 1.372, val: 1.384 | iter time: 359.30 ms (step) remaining time: 0:36:04
|
| 161 |
+
Epoch 1 | iter 1824 step 57 | loss train: 1.346, val: 1.384 | iter time: 359.75 ms (step) remaining time: 0:35:52
|
| 162 |
+
Epoch 1 | iter 1856 step 58 | loss train: 1.389, val: 1.384 | iter time: 358.76 ms (step) remaining time: 0:35:40
|
| 163 |
+
Epoch 1 | iter 1888 step 59 | loss train: 1.383, val: 1.384 | iter time: 358.26 ms (step) remaining time: 0:35:29
|
| 164 |
+
Epoch 1 | iter 1920 step 60 | loss train: 1.376, val: 1.384 | iter time: 359.70 ms (step) remaining time: 0:35:17
|
| 165 |
+
Epoch 1 | iter 1952 step 61 | loss train: 1.416, val: 1.384 | iter time: 360.33 ms (step) remaining time: 0:35:06
|
| 166 |
+
Epoch 1 | iter 1984 step 62 | loss train: 1.413, val: 1.384 | iter time: 359.84 ms (step) remaining time: 0:34:54
|
| 167 |
+
Epoch 1 | iter 2016 step 63 | loss train: 1.403, val: 1.384 | iter time: 359.84 ms (step) remaining time: 0:34:43
|
| 168 |
+
Epoch 1 | iter 2048 step 64 | loss train: 1.422, val: 1.384 | iter time: 359.36 ms (step) remaining time: 0:34:31
|
| 169 |
+
Epoch 1 | iter 2080 step 65 | loss train: 1.403, val: 1.384 | iter time: 359.96 ms (step) remaining time: 0:34:20
|
| 170 |
+
Epoch 1 | iter 2112 step 66 | loss train: 1.423, val: 1.384 | iter time: 359.89 ms (step) remaining time: 0:34:08
|
| 171 |
+
Epoch 1 | iter 2144 step 67 | loss train: 1.388, val: 1.384 | iter time: 357.93 ms (step) remaining time: 0:33:57
|
| 172 |
+
Epoch 1 | iter 2176 step 68 | loss train: 1.387, val: 1.384 | iter time: 360.34 ms (step) remaining time: 0:33:45
|
| 173 |
+
Epoch 1 | iter 2208 step 69 | loss train: 1.378, val: 1.384 | iter time: 360.62 ms (step) remaining time: 0:33:34
|
| 174 |
+
Epoch 1 | iter 2240 step 70 | loss train: 1.456, val: 1.384 | iter time: 360.33 ms (step) remaining time: 0:33:22
|
| 175 |
+
Epoch 1 | iter 2272 step 71 | loss train: 1.331, val: 1.384 | iter time: 358.28 ms (step) remaining time: 0:33:11
|
| 176 |
+
Epoch 1 | iter 2304 step 72 | loss train: 1.450, val: 1.384 | iter time: 361.11 ms (step) remaining time: 0:33:00
|
| 177 |
+
Epoch 1 | iter 2336 step 73 | loss train: 1.370, val: 1.384 | iter time: 358.94 ms (step) remaining time: 0:32:48
|
| 178 |
+
Epoch 1 | iter 2368 step 74 | loss train: 1.393, val: 1.384 | iter time: 360.33 ms (step) remaining time: 0:32:37
|
| 179 |
+
Epoch 1 | iter 2400 step 75 | loss train: 1.346, val: 1.384 | iter time: 360.23 ms (step) remaining time: 0:32:26
|
| 180 |
+
Epoch 1 | iter 2432 step 76 | loss train: 1.390, val: 1.384 | iter time: 359.55 ms (step) remaining time: 0:32:14
|
| 181 |
+
Epoch 1 | iter 2464 step 77 | loss train: 1.430, val: 1.384 | iter time: 360.50 ms (step) remaining time: 0:32:03
|
| 182 |
+
Epoch 1 | iter 2496 step 78 | loss train: 1.400, val: 1.384 | iter time: 359.88 ms (step) remaining time: 0:31:52
|
| 183 |
+
Epoch 1 | iter 2528 step 79 | loss train: 1.414, val: 1.384 | iter time: 358.00 ms (step) remaining time: 0:31:40
|
| 184 |
+
Epoch 1 | iter 2560 step 80 | loss train: 1.360, val: 1.384 | iter time: 359.93 ms (step) remaining time: 0:31:29
|
| 185 |
+
Epoch 1 | iter 2592 step 81 | loss train: 1.370, val: 1.384 | iter time: 360.49 ms (step) remaining time: 0:31:18
|
| 186 |
+
Epoch 1 | iter 2624 step 82 | loss train: 1.357, val: 1.384 | iter time: 359.63 ms (step) remaining time: 0:31:07
|
| 187 |
+
Epoch 1 | iter 2656 step 83 | loss train: 1.475, val: 1.384 | iter time: 358.50 ms (step) remaining time: 0:30:56
|
| 188 |
+
Epoch 1 | iter 2688 step 84 | loss train: 1.418, val: 1.384 | iter time: 359.98 ms (step) remaining time: 0:30:44
|
| 189 |
+
Epoch 1 | iter 2720 step 85 | loss train: 1.399, val: 1.384 | iter time: 358.52 ms (step) remaining time: 0:30:33
|
| 190 |
+
Epoch 1 | iter 2752 step 86 | loss train: 1.314, val: 1.384 | iter time: 359.80 ms (step) remaining time: 0:30:22
|
| 191 |
+
Epoch 1 | iter 2784 step 87 | loss train: 1.379, val: 1.384 | iter time: 359.09 ms (step) remaining time: 0:30:11
|
| 192 |
+
Epoch 1 | iter 2816 step 88 | loss train: 1.389, val: 1.384 | iter time: 360.58 ms (step) remaining time: 0:30:00
|
| 193 |
+
Epoch 1 | iter 2848 step 89 | loss train: 1.462, val: 1.384 | iter time: 359.08 ms (step) remaining time: 0:29:48
|
| 194 |
+
Epoch 1 | iter 2880 step 90 | loss train: 1.416, val: 1.384 | iter time: 358.58 ms (step) remaining time: 0:29:37
|
| 195 |
+
Epoch 1 | iter 2912 step 91 | loss train: 1.358, val: 1.384 | iter time: 360.25 ms (step) remaining time: 0:29:27
|
| 196 |
+
Epoch 1 | iter 2944 step 92 | loss train: 1.353, val: 1.384 | iter time: 359.92 ms (step) remaining time: 0:29:15
|
| 197 |
+
Epoch 1 | iter 2976 step 93 | loss train: 1.371, val: 1.384 | iter time: 358.83 ms (step) remaining time: 0:29:04
|
| 198 |
+
Epoch 1 | iter 3008 step 94 | loss train: 1.398, val: 1.384 | iter time: 358.68 ms (step) remaining time: 0:28:53
|
| 199 |
+
Epoch 1 | iter 3040 step 95 | loss train: 1.385, val: 1.384 | iter time: 358.23 ms (step) remaining time: 0:28:42
|
| 200 |
+
Epoch 1 | iter 3072 step 96 | loss train: 1.368, val: 1.384 | iter time: 359.37 ms (step) remaining time: 0:28:31
|
| 201 |
+
Epoch 1 | iter 3104 step 97 | loss train: 1.405, val: 1.384 | iter time: 362.03 ms (step) remaining time: 0:28:20
|
| 202 |
+
Epoch 1 | iter 3136 step 98 | loss train: 1.418, val: 1.384 | iter time: 358.84 ms (step) remaining time: 0:28:08
|
| 203 |
+
Epoch 1 | iter 3168 step 99 | loss train: 1.383, val: 1.384 | iter time: 361.50 ms (step) remaining time: 0:27:57
|
| 204 |
+
Epoch 1 | iter 3200 step 100 | loss train: 1.381, val: 1.384 | iter time: 357.64 ms (step) remaining time: 0:27:46
|
| 205 |
+
Validating ...
|
| 206 |
+
iter 3200: val loss 1.3758, val time: 9316.28 ms
|
| 207 |
+
Saving checkpoint to 'out/pretrain/2501/step-00000100/lit_model.pth'
|
| 208 |
+
Epoch 1 | iter 3232 step 101 | loss train: 1.369, val: 1.376 | iter time: 356.94 ms (step) remaining time: 0:28:14
|
| 209 |
+
Epoch 1 | iter 3264 step 102 | loss train: 1.419, val: 1.376 | iter time: 359.22 ms (step) remaining time: 0:28:03
|
| 210 |
+
Epoch 1 | iter 3296 step 103 | loss train: 1.440, val: 1.376 | iter time: 360.22 ms (step) remaining time: 0:27:51
|
| 211 |
+
Epoch 1 | iter 3328 step 104 | loss train: 1.407, val: 1.376 | iter time: 358.35 ms (step) remaining time: 0:27:39
|
| 212 |
+
Epoch 1 | iter 3360 step 105 | loss train: 1.383, val: 1.376 | iter time: 359.82 ms (step) remaining time: 0:27:27
|
| 213 |
+
Epoch 1 | iter 3392 step 106 | loss train: 1.396, val: 1.376 | iter time: 359.48 ms (step) remaining time: 0:27:16
|
| 214 |
+
Epoch 1 | iter 3424 step 107 | loss train: 1.320, val: 1.376 | iter time: 358.96 ms (step) remaining time: 0:27:04
|
| 215 |
+
Epoch 1 | iter 3456 step 108 | loss train: 1.330, val: 1.376 | iter time: 360.58 ms (step) remaining time: 0:26:52
|
| 216 |
+
Epoch 1 | iter 3488 step 109 | loss train: 1.373, val: 1.376 | iter time: 357.68 ms (step) remaining time: 0:26:41
|
| 217 |
+
Epoch 1 | iter 3520 step 110 | loss train: 1.408, val: 1.376 | iter time: 360.18 ms (step) remaining time: 0:26:29
|
| 218 |
+
Epoch 1 | iter 3552 step 111 | loss train: 1.320, val: 1.376 | iter time: 360.34 ms (step) remaining time: 0:26:17
|
| 219 |
+
Epoch 1 | iter 3584 step 112 | loss train: 1.342, val: 1.376 | iter time: 360.17 ms (step) remaining time: 0:26:06
|
| 220 |
+
Epoch 1 | iter 3616 step 113 | loss train: 1.356, val: 1.376 | iter time: 360.13 ms (step) remaining time: 0:25:54
|
| 221 |
+
Epoch 1 | iter 3648 step 114 | loss train: 1.406, val: 1.376 | iter time: 358.94 ms (step) remaining time: 0:25:42
|
| 222 |
+
Epoch 1 | iter 3680 step 115 | loss train: 1.394, val: 1.376 | iter time: 356.49 ms (step) remaining time: 0:25:31
|
| 223 |
+
Epoch 1 | iter 3712 step 116 | loss train: 1.352, val: 1.376 | iter time: 357.91 ms (step) remaining time: 0:25:19
|
| 224 |
+
Epoch 1 | iter 3744 step 117 | loss train: 1.402, val: 1.376 | iter time: 358.79 ms (step) remaining time: 0:25:08
|
| 225 |
+
Epoch 1 | iter 3776 step 118 | loss train: 1.402, val: 1.376 | iter time: 359.61 ms (step) remaining time: 0:24:56
|
| 226 |
+
Epoch 1 | iter 3808 step 119 | loss train: 1.425, val: 1.376 | iter time: 357.89 ms (step) remaining time: 0:24:45
|
| 227 |
+
Epoch 1 | iter 3840 step 120 | loss train: 1.401, val: 1.376 | iter time: 359.65 ms (step) remaining time: 0:24:33
|
| 228 |
+
Epoch 1 | iter 3872 step 121 | loss train: 1.384, val: 1.376 | iter time: 358.79 ms (step) remaining time: 0:24:22
|
| 229 |
+
Epoch 1 | iter 3904 step 122 | loss train: 1.360, val: 1.376 | iter time: 360.40 ms (step) remaining time: 0:24:10
|
| 230 |
+
Epoch 1 | iter 3936 step 123 | loss train: 1.456, val: 1.376 | iter time: 360.81 ms (step) remaining time: 0:23:59
|
| 231 |
+
Epoch 1 | iter 3968 step 124 | loss train: 1.363, val: 1.376 | iter time: 357.82 ms (step) remaining time: 0:23:47
|
| 232 |
+
Epoch 1 | iter 4000 step 125 | loss train: 1.366, val: 1.376 | iter time: 359.29 ms (step) remaining time: 0:23:36
|
| 233 |
+
Epoch 1 | iter 4032 step 126 | loss train: 1.341, val: 1.376 | iter time: 359.66 ms (step) remaining time: 0:23:24
|
| 234 |
+
Epoch 1 | iter 4064 step 127 | loss train: 1.351, val: 1.376 | iter time: 357.20 ms (step) remaining time: 0:23:13
|
| 235 |
+
Epoch 1 | iter 4096 step 128 | loss train: 1.309, val: 1.376 | iter time: 359.26 ms (step) remaining time: 0:23:02
|
| 236 |
+
Epoch 1 | iter 4128 step 129 | loss train: 1.369, val: 1.376 | iter time: 359.71 ms (step) remaining time: 0:22:50
|
| 237 |
+
Epoch 1 | iter 4160 step 130 | loss train: 1.372, val: 1.376 | iter time: 359.24 ms (step) remaining time: 0:22:39
|
| 238 |
+
Epoch 1 | iter 4192 step 131 | loss train: 1.379, val: 1.376 | iter time: 361.33 ms (step) remaining time: 0:22:27
|
| 239 |
+
Epoch 1 | iter 4224 step 132 | loss train: 1.371, val: 1.376 | iter time: 358.67 ms (step) remaining time: 0:22:16
|
| 240 |
+
Epoch 1 | iter 4256 step 133 | loss train: 1.293, val: 1.376 | iter time: 358.61 ms (step) remaining time: 0:22:05
|
| 241 |
+
Epoch 1 | iter 4288 step 134 | loss train: 1.356, val: 1.376 | iter time: 359.37 ms (step) remaining time: 0:21:53
|
| 242 |
+
Epoch 1 | iter 4320 step 135 | loss train: 1.374, val: 1.376 | iter time: 360.14 ms (step) remaining time: 0:21:42
|
| 243 |
+
Epoch 1 | iter 4352 step 136 | loss train: 1.379, val: 1.376 | iter time: 358.83 ms (step) remaining time: 0:21:31
|
| 244 |
+
Epoch 1 | iter 4384 step 137 | loss train: 1.325, val: 1.376 | iter time: 359.97 ms (step) remaining time: 0:21:19
|
| 245 |
+
Epoch 1 | iter 4416 step 138 | loss train: 1.310, val: 1.376 | iter time: 360.51 ms (step) remaining time: 0:21:08
|
| 246 |
+
Epoch 1 | iter 4448 step 139 | loss train: 1.344, val: 1.376 | iter time: 357.83 ms (step) remaining time: 0:20:56
|
| 247 |
+
Epoch 1 | iter 4480 step 140 | loss train: 1.404, val: 1.376 | iter time: 358.43 ms (step) remaining time: 0:20:45
|
| 248 |
+
Epoch 1 | iter 4512 step 141 | loss train: 1.398, val: 1.376 | iter time: 359.62 ms (step) remaining time: 0:20:34
|
| 249 |
+
Epoch 1 | iter 4544 step 142 | loss train: 1.305, val: 1.376 | iter time: 360.75 ms (step) remaining time: 0:20:23
|
| 250 |
+
Epoch 1 | iter 4576 step 143 | loss train: 1.334, val: 1.376 | iter time: 358.80 ms (step) remaining time: 0:20:12
|
| 251 |
+
Epoch 1 | iter 4608 step 144 | loss train: 1.267, val: 1.376 | iter time: 359.82 ms (step) remaining time: 0:20:00
|
| 252 |
+
Epoch 1 | iter 4640 step 145 | loss train: 1.310, val: 1.376 | iter time: 359.23 ms (step) remaining time: 0:19:49
|
| 253 |
+
Epoch 1 | iter 4672 step 146 | loss train: 1.347, val: 1.376 | iter time: 360.55 ms (step) remaining time: 0:19:38
|
| 254 |
+
Epoch 1 | iter 4704 step 147 | loss train: 1.339, val: 1.376 | iter time: 360.26 ms (step) remaining time: 0:19:26
|
| 255 |
+
Epoch 1 | iter 4736 step 148 | loss train: 1.320, val: 1.376 | iter time: 359.16 ms (step) remaining time: 0:19:15
|
| 256 |
+
Epoch 1 | iter 4768 step 149 | loss train: 1.307, val: 1.376 | iter time: 358.87 ms (step) remaining time: 0:19:04
|
| 257 |
+
Epoch 1 | iter 4800 step 150 | loss train: 1.353, val: 1.376 | iter time: 359.04 ms (step) remaining time: 0:18:52
|
| 258 |
+
Validating ...
|
| 259 |
+
iter 4800: val loss 1.3584, val time: 9331.75 ms
|
| 260 |
+
Epoch 1 | iter 4832 step 151 | loss train: 1.368, val: 1.358 | iter time: 360.64 ms (step) remaining time: 0:18:47
|
| 261 |
+
Epoch 1 | iter 4864 step 152 | loss train: 1.279, val: 1.358 | iter time: 360.04 ms (step) remaining time: 0:18:36
|
| 262 |
+
Epoch 1 | iter 4896 step 153 | loss train: 1.366, val: 1.358 | iter time: 360.63 ms (step) remaining time: 0:18:25
|
| 263 |
+
Epoch 1 | iter 4928 step 154 | loss train: 1.266, val: 1.358 | iter time: 360.45 ms (step) remaining time: 0:18:13
|
| 264 |
+
Epoch 1 | iter 4960 step 155 | loss train: 1.348, val: 1.358 | iter time: 359.06 ms (step) remaining time: 0:18:02
|
| 265 |
+
Epoch 1 | iter 4992 step 156 | loss train: 1.389, val: 1.358 | iter time: 358.50 ms (step) remaining time: 0:17:51
|
| 266 |
+
Epoch 1 | iter 5024 step 157 | loss train: 1.293, val: 1.358 | iter time: 358.17 ms (step) remaining time: 0:17:39
|
| 267 |
+
Epoch 1 | iter 5056 step 158 | loss train: 1.321, val: 1.358 | iter time: 358.70 ms (step) remaining time: 0:17:28
|
| 268 |
+
Epoch 1 | iter 5088 step 159 | loss train: 1.378, val: 1.358 | iter time: 357.95 ms (step) remaining time: 0:17:17
|
| 269 |
+
Epoch 1 | iter 5120 step 160 | loss train: 1.392, val: 1.358 | iter time: 360.34 ms (step) remaining time: 0:17:05
|
| 270 |
+
Epoch 1 | iter 5152 step 161 | loss train: 1.413, val: 1.358 | iter time: 359.66 ms (step) remaining time: 0:16:54
|
| 271 |
+
Epoch 1 | iter 5184 step 162 | loss train: 1.470, val: 1.358 | iter time: 359.69 ms (step) remaining time: 0:16:43
|
| 272 |
+
Epoch 1 | iter 5216 step 163 | loss train: 1.372, val: 1.358 | iter time: 358.44 ms (step) remaining time: 0:16:31
|
| 273 |
+
Epoch 1 | iter 5248 step 164 | loss train: 1.303, val: 1.358 | iter time: 361.69 ms (step) remaining time: 0:16:20
|
| 274 |
+
Epoch 1 | iter 5280 step 165 | loss train: 1.376, val: 1.358 | iter time: 359.07 ms (step) remaining time: 0:16:09
|
| 275 |
+
Epoch 1 | iter 5312 step 166 | loss train: 1.343, val: 1.358 | iter time: 361.26 ms (step) remaining time: 0:15:57
|
| 276 |
+
Epoch 1 | iter 5344 step 167 | loss train: 1.392, val: 1.358 | iter time: 359.30 ms (step) remaining time: 0:15:46
|
| 277 |
+
Epoch 1 | iter 5376 step 168 | loss train: 1.375, val: 1.358 | iter time: 359.14 ms (step) remaining time: 0:15:35
|
| 278 |
+
Epoch 1 | iter 5408 step 169 | loss train: 1.358, val: 1.358 | iter time: 359.95 ms (step) remaining time: 0:15:24
|
| 279 |
+
Epoch 1 | iter 5440 step 170 | loss train: 1.387, val: 1.358 | iter time: 360.69 ms (step) remaining time: 0:15:12
|
| 280 |
+
Epoch 1 | iter 5472 step 171 | loss train: 1.347, val: 1.358 | iter time: 357.40 ms (step) remaining time: 0:15:01
|
| 281 |
+
Epoch 1 | iter 5504 step 172 | loss train: 1.354, val: 1.358 | iter time: 360.32 ms (step) remaining time: 0:14:50
|
| 282 |
+
Epoch 1 | iter 5536 step 173 | loss train: 1.366, val: 1.358 | iter time: 358.60 ms (step) remaining time: 0:14:39
|
| 283 |
+
Epoch 1 | iter 5568 step 174 | loss train: 1.337, val: 1.358 | iter time: 359.23 ms (step) remaining time: 0:14:27
|
| 284 |
+
Epoch 1 | iter 5600 step 175 | loss train: 1.337, val: 1.358 | iter time: 359.73 ms (step) remaining time: 0:14:16
|
| 285 |
+
Epoch 1 | iter 5632 step 176 | loss train: 1.367, val: 1.358 | iter time: 358.54 ms (step) remaining time: 0:14:05
|
| 286 |
+
Epoch 1 | iter 5664 step 177 | loss train: 1.427, val: 1.358 | iter time: 359.61 ms (step) remaining time: 0:13:54
|
| 287 |
+
Epoch 1 | iter 5696 step 178 | loss train: 1.292, val: 1.358 | iter time: 357.83 ms (step) remaining time: 0:13:42
|
| 288 |
+
Epoch 1 | iter 5728 step 179 | loss train: 1.315, val: 1.358 | iter time: 358.14 ms (step) remaining time: 0:13:31
|
| 289 |
+
Epoch 1 | iter 5760 step 180 | loss train: 1.341, val: 1.358 | iter time: 362.36 ms (step) remaining time: 0:13:20
|
| 290 |
+
Epoch 1 | iter 5792 step 181 | loss train: 1.344, val: 1.358 | iter time: 359.84 ms (step) remaining time: 0:13:09
|
| 291 |
+
Epoch 1 | iter 5824 step 182 | loss train: 1.347, val: 1.358 | iter time: 359.39 ms (step) remaining time: 0:12:58
|
| 292 |
+
Epoch 1 | iter 5856 step 183 | loss train: 1.342, val: 1.358 | iter time: 358.73 ms (step) remaining time: 0:12:46
|
| 293 |
+
Epoch 1 | iter 5888 step 184 | loss train: 1.327, val: 1.358 | iter time: 359.10 ms (step) remaining time: 0:12:35
|
| 294 |
+
Epoch 1 | iter 5920 step 185 | loss train: 1.291, val: 1.358 | iter time: 360.39 ms (step) remaining time: 0:12:24
|
| 295 |
+
Epoch 1 | iter 5952 step 186 | loss train: 1.345, val: 1.358 | iter time: 359.84 ms (step) remaining time: 0:12:13
|
| 296 |
+
Epoch 1 | iter 5984 step 187 | loss train: 1.355, val: 1.358 | iter time: 359.66 ms (step) remaining time: 0:12:02
|
| 297 |
+
Epoch 1 | iter 6016 step 188 | loss train: 1.359, val: 1.358 | iter time: 358.92 ms (step) remaining time: 0:11:50
|
| 298 |
+
Epoch 1 | iter 6048 step 189 | loss train: 1.333, val: 1.358 | iter time: 360.08 ms (step) remaining time: 0:11:39
|
| 299 |
+
Epoch 1 | iter 6080 step 190 | loss train: 1.322, val: 1.358 | iter time: 358.55 ms (step) remaining time: 0:11:28
|
| 300 |
+
Epoch 1 | iter 6112 step 191 | loss train: 1.350, val: 1.358 | iter time: 359.75 ms (step) remaining time: 0:11:17
|
| 301 |
+
Epoch 1 | iter 6144 step 192 | loss train: 1.317, val: 1.358 | iter time: 358.34 ms (step) remaining time: 0:11:06
|
| 302 |
+
Epoch 1 | iter 6176 step 193 | loss train: 1.433, val: 1.358 | iter time: 359.86 ms (step) remaining time: 0:10:54
|
| 303 |
+
Epoch 1 | iter 6208 step 194 | loss train: 1.398, val: 1.358 | iter time: 359.30 ms (step) remaining time: 0:10:43
|
| 304 |
+
Epoch 1 | iter 6240 step 195 | loss train: 1.394, val: 1.358 | iter time: 360.39 ms (step) remaining time: 0:10:32
|
| 305 |
+
Epoch 1 | iter 6272 step 196 | loss train: 1.342, val: 1.358 | iter time: 357.85 ms (step) remaining time: 0:10:21
|
| 306 |
+
Epoch 1 | iter 6304 step 197 | loss train: 1.339, val: 1.358 | iter time: 358.97 ms (step) remaining time: 0:10:10
|
| 307 |
+
Epoch 1 | iter 6336 step 198 | loss train: 1.325, val: 1.358 | iter time: 357.72 ms (step) remaining time: 0:09:59
|
| 308 |
+
Epoch 1 | iter 6368 step 199 | loss train: 1.259, val: 1.358 | iter time: 360.42 ms (step) remaining time: 0:09:48
|
| 309 |
+
Epoch 1 | iter 6400 step 200 | loss train: 1.374, val: 1.358 | iter time: 359.52 ms (step) remaining time: 0:09:36
|
| 310 |
+
Validating ...
|
| 311 |
+
iter 6400: val loss 1.3422, val time: 9310.92 ms
|
| 312 |
+
Saving checkpoint to 'out/pretrain/2501/step-00000200/lit_model.pth'
|
| 313 |
+
Epoch 1 | iter 6432 step 201 | loss train: 1.287, val: 1.342 | iter time: 357.16 ms (step) remaining time: 0:09:32
|
| 314 |
+
Epoch 1 | iter 6464 step 202 | loss train: 1.426, val: 1.342 | iter time: 357.81 ms (step) remaining time: 0:09:21
|
| 315 |
+
Epoch 1 | iter 6496 step 203 | loss train: 1.360, val: 1.342 | iter time: 356.41 ms (step) remaining time: 0:09:09
|
| 316 |
+
Epoch 1 | iter 6528 step 204 | loss train: 1.257, val: 1.342 | iter time: 360.00 ms (step) remaining time: 0:08:58
|
| 317 |
+
Epoch 1 | iter 6560 step 205 | loss train: 1.297, val: 1.342 | iter time: 358.62 ms (step) remaining time: 0:08:47
|
| 318 |
+
Epoch 1 | iter 6592 step 206 | loss train: 1.310, val: 1.342 | iter time: 359.17 ms (step) remaining time: 0:08:35
|
| 319 |
+
Epoch 1 | iter 6624 step 207 | loss train: 1.291, val: 1.342 | iter time: 359.83 ms (step) remaining time: 0:08:24
|
| 320 |
+
Epoch 1 | iter 6656 step 208 | loss train: 1.322, val: 1.342 | iter time: 359.76 ms (step) remaining time: 0:08:13
|
| 321 |
+
Epoch 1 | iter 6688 step 209 | loss train: 1.394, val: 1.342 | iter time: 360.63 ms (step) remaining time: 0:08:01
|
| 322 |
+
Epoch 1 | iter 6720 step 210 | loss train: 1.287, val: 1.342 | iter time: 359.49 ms (step) remaining time: 0:07:50
|
| 323 |
+
Epoch 1 | iter 6752 step 211 | loss train: 1.331, val: 1.342 | iter time: 358.42 ms (step) remaining time: 0:07:39
|
| 324 |
+
Epoch 1 | iter 6784 step 212 | loss train: 1.279, val: 1.342 | iter time: 358.61 ms (step) remaining time: 0:07:28
|
| 325 |
+
Epoch 1 | iter 6816 step 213 | loss train: 1.420, val: 1.342 | iter time: 360.61 ms (step) remaining time: 0:07:16
|
| 326 |
+
Epoch 1 | iter 6848 step 214 | loss train: 1.332, val: 1.342 | iter time: 359.73 ms (step) remaining time: 0:07:05
|
| 327 |
+
Epoch 1 | iter 6880 step 215 | loss train: 1.279, val: 1.342 | iter time: 359.59 ms (step) remaining time: 0:06:54
|
| 328 |
+
Epoch 1 | iter 6912 step 216 | loss train: 1.283, val: 1.342 | iter time: 360.42 ms (step) remaining time: 0:06:43
|
| 329 |
+
Epoch 1 | iter 6944 step 217 | loss train: 1.322, val: 1.342 | iter time: 361.37 ms (step) remaining time: 0:06:31
|
| 330 |
+
Epoch 1 | iter 6976 step 218 | loss train: 1.278, val: 1.342 | iter time: 359.24 ms (step) remaining time: 0:06:20
|
| 331 |
+
Epoch 1 | iter 7008 step 219 | loss train: 1.306, val: 1.342 | iter time: 359.08 ms (step) remaining time: 0:06:09
|
| 332 |
+
Epoch 1 | iter 7040 step 220 | loss train: 1.367, val: 1.342 | iter time: 359.51 ms (step) remaining time: 0:05:58
|
| 333 |
+
Epoch 1 | iter 7072 step 221 | loss train: 1.401, val: 1.342 | iter time: 358.77 ms (step) remaining time: 0:05:46
|
| 334 |
+
Epoch 1 | iter 7104 step 222 | loss train: 1.307, val: 1.342 | iter time: 359.80 ms (step) remaining time: 0:05:35
|
| 335 |
+
Epoch 1 | iter 7136 step 223 | loss train: 1.292, val: 1.342 | iter time: 357.55 ms (step) remaining time: 0:05:24
|
| 336 |
+
Epoch 1 | iter 7168 step 224 | loss train: 1.316, val: 1.342 | iter time: 357.36 ms (step) remaining time: 0:05:13
|
| 337 |
+
Epoch 1 | iter 7200 step 225 | loss train: 1.304, val: 1.342 | iter time: 360.23 ms (step) remaining time: 0:05:01
|
| 338 |
+
Epoch 1 | iter 7232 step 226 | loss train: 1.264, val: 1.342 | iter time: 359.23 ms (step) remaining time: 0:04:50
|
| 339 |
+
Epoch 1 | iter 7264 step 227 | loss train: 1.297, val: 1.342 | iter time: 359.82 ms (step) remaining time: 0:04:39
|
| 340 |
+
Epoch 1 | iter 7296 step 228 | loss train: 1.325, val: 1.342 | iter time: 360.34 ms (step) remaining time: 0:04:28
|
| 341 |
+
Epoch 1 | iter 7328 step 229 | loss train: 1.234, val: 1.342 | iter time: 360.32 ms (step) remaining time: 0:04:17
|
| 342 |
+
Epoch 1 | iter 7360 step 230 | loss train: 1.287, val: 1.342 | iter time: 361.14 ms (step) remaining time: 0:04:05
|
| 343 |
+
Epoch 1 | iter 7392 step 231 | loss train: 1.249, val: 1.342 | iter time: 361.26 ms (step) remaining time: 0:03:54
|
| 344 |
+
Epoch 1 | iter 7424 step 232 | loss train: 1.299, val: 1.342 | iter time: 359.71 ms (step) remaining time: 0:03:43
|
| 345 |
+
Epoch 1 | iter 7456 step 233 | loss train: 1.321, val: 1.342 | iter time: 359.22 ms (step) remaining time: 0:03:32
|
| 346 |
+
Epoch 1 | iter 7488 step 234 | loss train: 1.303, val: 1.342 | iter time: 359.32 ms (step) remaining time: 0:03:21
|
| 347 |
+
Epoch 1 | iter 7520 step 235 | loss train: 1.363, val: 1.342 | iter time: 360.85 ms (step) remaining time: 0:03:09
|
| 348 |
+
Epoch 1 | iter 7552 step 236 | loss train: 1.248, val: 1.342 | iter time: 356.77 ms (step) remaining time: 0:02:58
|
| 349 |
+
Epoch 1 | iter 7584 step 237 | loss train: 1.220, val: 1.342 | iter time: 360.02 ms (step) remaining time: 0:02:47
|
| 350 |
+
Epoch 1 | iter 7616 step 238 | loss train: 1.325, val: 1.342 | iter time: 357.48 ms (step) remaining time: 0:02:36
|
| 351 |
+
Epoch 1 | iter 7648 step 239 | loss train: 1.352, val: 1.342 | iter time: 358.89 ms (step) remaining time: 0:02:25
|
| 352 |
+
Epoch 1 | iter 7680 step 240 | loss train: 1.250, val: 1.342 | iter time: 360.98 ms (step) remaining time: 0:02:13
|
| 353 |
+
Epoch 1 | iter 7712 step 241 | loss train: 1.265, val: 1.342 | iter time: 358.69 ms (step) remaining time: 0:02:02
|
| 354 |
+
Epoch 1 | iter 7744 step 242 | loss train: 1.269, val: 1.342 | iter time: 359.83 ms (step) remaining time: 0:01:51
|
| 355 |
+
Epoch 1 | iter 7776 step 243 | loss train: 1.286, val: 1.342 | iter time: 360.23 ms (step) remaining time: 0:01:40
|
| 356 |
+
Epoch 1 | iter 7808 step 244 | loss train: 1.292, val: 1.342 | iter time: 359.48 ms (step) remaining time: 0:01:29
|
| 357 |
+
Epoch 1 | iter 7840 step 245 | loss train: 1.219, val: 1.342 | iter time: 359.16 ms (step) remaining time: 0:01:18
|
| 358 |
+
Epoch 1 | iter 7872 step 246 | loss train: 1.377, val: 1.342 | iter time: 359.59 ms (step) remaining time: 0:01:06
|
| 359 |
+
Epoch 1 | iter 7904 step 247 | loss train: 1.245, val: 1.342 | iter time: 360.63 ms (step) remaining time: 0:00:55
|
| 360 |
+
Epoch 1 | iter 7936 step 248 | loss train: 1.228, val: 1.342 | iter time: 360.64 ms (step) remaining time: 0:00:44
|
| 361 |
+
Epoch 1 | iter 7968 step 249 | loss train: 1.286, val: 1.342 | iter time: 357.79 ms (step) remaining time: 0:00:33
|
| 362 |
+
Epoch 1 | iter 8000 step 250 | loss train: 1.271, val: 1.342 | iter time: 359.67 ms (step) remaining time: 0:00:22
|
| 363 |
+
Validating ...
|
| 364 |
+
iter 8000: val loss 1.3335, val time: 9326.54 ms
|
| 365 |
+
Epoch 1 | iter 8032 step 251 | loss train: 1.335, val: 1.333 | iter time: 361.00 ms (step) remaining time: 0:00:11
|
| 366 |
+
Epoch 1 | iter 8064 step 252 | loss train: 1.351, val: 1.333 | iter time: 358.17 ms (step) remaining time: 0:00:00
|
| 367 |
+
Validating ...
|
| 368 |
+
Final evaluation | val loss: 1.333 | val ppl: 3.792
|
| 369 |
+
Saving checkpoint to 'out/pretrain/2501/final/lit_model.pth'
|
| 370 |
+
----------------------------------------
|
| 371 |
+
| Performance
|
| 372 |
+
| - Total tokens : 264,241,152
|
| 373 |
+
| - Training Time : 2857.04 s
|
| 374 |
+
| - Tok/sec : 234.93 tok/s
|
| 375 |
+
| ----------------------------------------
|
| 376 |
+
| Memory Usage
|
| 377 |
+
| - Memory Used : 26.32 GB
|
| 378 |
+
----------------------------------------
|
out/pretrain/tinyllama/teelogs/2501_full.txt
ADDED
|
@@ -0,0 +1,389 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
Initializing distributed: GLOBAL_RANK: 0, MEMBER: 1/4
|
| 2 |
+
Initializing distributed: GLOBAL_RANK: 1, MEMBER: 2/4
|
| 3 |
+
[rank: 1] Seed set to 42
|
| 4 |
+
Initializing distributed: GLOBAL_RANK: 3, MEMBER: 4/4
|
| 5 |
+
Initializing distributed: GLOBAL_RANK: 2, MEMBER: 3/4
|
| 6 |
+
[rank: 3] Seed set to 42
|
| 7 |
+
[rank: 2] Seed set to 42
|
| 8 |
+
----------------------------------------------------------------------------------------------------
|
| 9 |
+
distributed_backend=nccl
|
| 10 |
+
All distributed processes registered. Starting with 4 processes
|
| 11 |
+
----------------------------------------------------------------------------------------------------
|
| 12 |
+
|
| 13 |
+
[rank: 0] Seed set to 42
|
| 14 |
+
/mnt/data/litgpt/litgpt/pretrain.py:515: UserWarning: A newer version of litdata is available (0.2.52). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 15 |
+
train_dataloader = data.train_dataloader()
|
| 16 |
+
/mnt/data/litgpt/litgpt/pretrain.py:515: UserWarning: A newer version of litdata is available (0.2.52). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 17 |
+
train_dataloader = data.train_dataloader()
|
| 18 |
+
/mnt/data/litgpt/litgpt/pretrain.py:515: UserWarning: A newer version of litdata is available (0.2.52). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 19 |
+
train_dataloader = data.train_dataloader()
|
| 20 |
+
/mnt/data/litgpt/litgpt/pretrain.py:515: UserWarning: A newer version of litdata is available (0.2.52). Please consider upgrading with `pip install -U litdata`. Not all functionalities of the platform can be guaranteed to work with the current version.
|
| 21 |
+
train_dataloader = data.train_dataloader()
|
| 22 |
+
All GPUs are fully connected via NVLink.
|
| 23 |
+
{'data': {'batch_size': 1,
|
| 24 |
+
'data_path': PosixPath('litgpt/data/arxiv'),
|
| 25 |
+
'num_workers': 8,
|
| 26 |
+
'seed': 42,
|
| 27 |
+
'seq_length': 2048,
|
| 28 |
+
'use_starcoder': True},
|
| 29 |
+
'data_dir': PosixPath('litgpt/data/arxiv/2501'),
|
| 30 |
+
'devices': 'auto',
|
| 31 |
+
'eval': {'evaluate_example': 'first',
|
| 32 |
+
'final_validation': True,
|
| 33 |
+
'initial_validation': True,
|
| 34 |
+
'interval': 50,
|
| 35 |
+
'max_iters': 200,
|
| 36 |
+
'max_new_tokens': None},
|
| 37 |
+
'initial_checkpoint_dir': PosixPath('out/pretrain/2412_full/final'),
|
| 38 |
+
'log': {'group': None, 'project': None, 'run': None},
|
| 39 |
+
'logger_name': 'tensorboard',
|
| 40 |
+
'model_config': {'attention_logit_softcapping': None,
|
| 41 |
+
'attention_scores_scalar': None,
|
| 42 |
+
'attn_bias': False,
|
| 43 |
+
'bias': False,
|
| 44 |
+
'block_size': 2048,
|
| 45 |
+
'final_logit_softcapping': None,
|
| 46 |
+
'gelu_approximate': 'none',
|
| 47 |
+
'head_size': 64,
|
| 48 |
+
'hf_config': {'name': 'TinyLlama-1.1B-intermediate-step-1431k-3T',
|
| 49 |
+
'org': 'TinyLlama'},
|
| 50 |
+
'intermediate_size': 5632,
|
| 51 |
+
'lm_head_bias': False,
|
| 52 |
+
'mlp_class_name': 'LLaMAMLP',
|
| 53 |
+
'moe_intermediate_size': None,
|
| 54 |
+
'n_embd': 2048,
|
| 55 |
+
'n_expert': 0,
|
| 56 |
+
'n_expert_per_token': 0,
|
| 57 |
+
'n_head': 32,
|
| 58 |
+
'n_layer': 22,
|
| 59 |
+
'n_query_groups': 4,
|
| 60 |
+
'name': 'tiny-llama-1.1b',
|
| 61 |
+
'norm_1': True,
|
| 62 |
+
'norm_2': True,
|
| 63 |
+
'norm_class_name': 'RMSNorm',
|
| 64 |
+
'norm_eps': 1e-05,
|
| 65 |
+
'norm_qk': False,
|
| 66 |
+
'norm_qk_type': 'default',
|
| 67 |
+
'padded_vocab_size': 32000,
|
| 68 |
+
'padding_multiple': 64,
|
| 69 |
+
'parallel_residual': False,
|
| 70 |
+
'post_attention_norm': False,
|
| 71 |
+
'post_mlp_norm': False,
|
| 72 |
+
'rope_adjustments': None,
|
| 73 |
+
'rope_base': 10000,
|
| 74 |
+
'rope_condense_ratio': 1,
|
| 75 |
+
'rope_indices': None,
|
| 76 |
+
'rope_local_base_freq': None,
|
| 77 |
+
'rotary_percentage': 1.0,
|
| 78 |
+
'scale_embeddings': False,
|
| 79 |
+
'shared_attention_norm': False,
|
| 80 |
+
'sliding_window_indices': None,
|
| 81 |
+
'sliding_window_size': None,
|
| 82 |
+
'vocab_size': 32000},
|
| 83 |
+
'model_name': 'tiny-llama-1.1b',
|
| 84 |
+
'num_nodes': 1,
|
| 85 |
+
'optimizer': "{'class_path': 'torch.optim.AdamW', 'init_args': {'lr': 0.0004, "
|
| 86 |
+
"'weight_decay': 0.1, 'betas': [0.9, 0.95]}}",
|
| 87 |
+
'out_dir': PosixPath('out/pretrain/2501_full'),
|
| 88 |
+
'ppl': False,
|
| 89 |
+
'precision': 'bf16-mixed',
|
| 90 |
+
'resume': False,
|
| 91 |
+
'seed': 42,
|
| 92 |
+
'tokenizer_dir': PosixPath('checkpoints/TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T'),
|
| 93 |
+
'train': {'epochs': None,
|
| 94 |
+
'global_batch_size': 512,
|
| 95 |
+
'log_interval': 1,
|
| 96 |
+
'lr_warmup_fraction': None,
|
| 97 |
+
'lr_warmup_steps': 20,
|
| 98 |
+
'max_norm': 1.0,
|
| 99 |
+
'max_seq_length': 2048,
|
| 100 |
+
'max_steps': None,
|
| 101 |
+
'max_tokens': 266338304,
|
| 102 |
+
'micro_batch_size': 4,
|
| 103 |
+
'min_lr': 4e-05,
|
| 104 |
+
'save_interval': 100,
|
| 105 |
+
'tie_embeddings': None}}
|
| 106 |
+
Time to instantiate model: 0.04 seconds.
|
| 107 |
+
Total parameters: 1,100,048,384
|
| 108 |
+
[fix] out/pretrain/2412_full/final/lit_model.pth 是整包 state，提取 model 权重并原子覆盖...
|
| 109 |
+
[fix] 已覆盖为纯权重: out/pretrain/2412_full/final/lit_model.pth
|
| 110 |
+
Validating ...
|
| 111 |
+
Measured TFLOPs: 239.66
|
| 112 |
+
Epoch 1 | iter 32 step 1 | loss train: 1.403, val: 1.397 | iter time: 538.33 ms (step) remaining time: 0:50:36
|
| 113 |
+
Epoch 1 | iter 64 step 2 | loss train: 1.346, val: 1.397 | iter time: 356.55 ms (step) remaining time: 0:47:48
|
| 114 |
+
Epoch 1 | iter 96 step 3 | loss train: 1.353, val: 1.397 | iter time: 358.24 ms (step) remaining time: 0:46:47
|
| 115 |
+
Epoch 1 | iter 128 step 4 | loss train: 1.348, val: 1.397 | iter time: 357.23 ms (step) remaining time: 0:46:12
|
| 116 |
+
Epoch 1 | iter 160 step 5 | loss train: 1.310, val: 1.397 | iter time: 358.87 ms (step) remaining time: 0:45:47
|
| 117 |
+
Epoch 1 | iter 192 step 6 | loss train: 1.394, val: 1.397 | iter time: 358.63 ms (step) remaining time: 0:45:27
|
| 118 |
+
Epoch 1 | iter 224 step 7 | loss train: 1.366, val: 1.397 | iter time: 359.22 ms (step) remaining time: 0:45:11
|
| 119 |
+
Epoch 1 | iter 256 step 8 | loss train: 1.357, val: 1.397 | iter time: 360.37 ms (step) remaining time: 0:44:55
|
| 120 |
+
Epoch 1 | iter 288 step 9 | loss train: 1.362, val: 1.397 | iter time: 358.10 ms (step) remaining time: 0:44:41
|
| 121 |
+
Epoch 1 | iter 320 step 10 | loss train: 1.356, val: 1.397 | iter time: 360.02 ms (step) remaining time: 0:44:28
|
| 122 |
+
Epoch 1 | iter 352 step 11 | loss train: 1.460, val: 1.397 | iter time: 358.79 ms (step) remaining time: 0:44:15
|
| 123 |
+
Epoch 1 | iter 384 step 12 | loss train: 1.355, val: 1.397 | iter time: 360.48 ms (step) remaining time: 0:44:02
|
| 124 |
+
Epoch 1 | iter 416 step 13 | loss train: 1.316, val: 1.397 | iter time: 358.54 ms (step) remaining time: 0:43:50
|
| 125 |
+
Epoch 1 | iter 448 step 14 | loss train: 1.365, val: 1.397 | iter time: 359.56 ms (step) remaining time: 0:43:38
|
| 126 |
+
Epoch 1 | iter 480 step 15 | loss train: 1.356, val: 1.397 | iter time: 358.74 ms (step) remaining time: 0:43:26
|
| 127 |
+
Epoch 1 | iter 512 step 16 | loss train: 1.437, val: 1.397 | iter time: 358.07 ms (step) remaining time: 0:43:14
|
| 128 |
+
Epoch 1 | iter 544 step 17 | loss train: 1.358, val: 1.397 | iter time: 359.89 ms (step) remaining time: 0:43:02
|
| 129 |
+
Epoch 1 | iter 576 step 18 | loss train: 1.393, val: 1.397 | iter time: 359.03 ms (step) remaining time: 0:42:51
|
| 130 |
+
Epoch 1 | iter 608 step 19 | loss train: 1.371, val: 1.397 | iter time: 360.24 ms (step) remaining time: 0:42:39
|
| 131 |
+
Epoch 1 | iter 640 step 20 | loss train: 1.386, val: 1.397 | iter time: 360.41 ms (step) remaining time: 0:42:28
|
| 132 |
+
Epoch 1 | iter 672 step 21 | loss train: 1.377, val: 1.397 | iter time: 359.95 ms (step) remaining time: 0:42:17
|
| 133 |
+
Epoch 1 | iter 704 step 22 | loss train: 1.421, val: 1.397 | iter time: 359.80 ms (step) remaining time: 0:42:05
|
| 134 |
+
Epoch 1 | iter 736 step 23 | loss train: 1.428, val: 1.397 | iter time: 359.09 ms (step) remaining time: 0:41:54
|
| 135 |
+
Epoch 1 | iter 768 step 24 | loss train: 1.415, val: 1.397 | iter time: 359.53 ms (step) remaining time: 0:41:43
|
| 136 |
+
Epoch 1 | iter 800 step 25 | loss train: 1.410, val: 1.397 | iter time: 358.80 ms (step) remaining time: 0:41:32
|
| 137 |
+
Epoch 1 | iter 832 step 26 | loss train: 1.366, val: 1.397 | iter time: 359.72 ms (step) remaining time: 0:41:21
|
| 138 |
+
Epoch 1 | iter 864 step 27 | loss train: 1.458, val: 1.397 | iter time: 359.36 ms (step) remaining time: 0:41:10
|
| 139 |
+
Epoch 1 | iter 896 step 28 | loss train: 1.384, val: 1.397 | iter time: 357.36 ms (step) remaining time: 0:40:59
|
| 140 |
+
Epoch 1 | iter 928 step 29 | loss train: 1.417, val: 1.397 | iter time: 360.41 ms (step) remaining time: 0:40:47
|
| 141 |
+
Epoch 1 | iter 960 step 30 | loss train: 1.395, val: 1.397 | iter time: 358.60 ms (step) remaining time: 0:40:36
|
| 142 |
+
Epoch 1 | iter 992 step 31 | loss train: 1.435, val: 1.397 | iter time: 358.80 ms (step) remaining time: 0:40:25
|
| 143 |
+
Epoch 1 | iter 1024 step 32 | loss train: 1.411, val: 1.397 | iter time: 359.41 ms (step) remaining time: 0:40:14
|
| 144 |
+
Epoch 1 | iter 1056 step 33 | loss train: 1.312, val: 1.397 | iter time: 361.37 ms (step) remaining time: 0:40:03
|
| 145 |
+
Epoch 1 | iter 1088 step 34 | loss train: 1.466, val: 1.397 | iter time: 358.91 ms (step) remaining time: 0:39:52
|
| 146 |
+
Epoch 1 | iter 1120 step 35 | loss train: 1.367, val: 1.397 | iter time: 358.94 ms (step) remaining time: 0:39:41
|
| 147 |
+
Epoch 1 | iter 1152 step 36 | loss train: 1.369, val: 1.397 | iter time: 359.73 ms (step) remaining time: 0:39:30
|
| 148 |
+
Epoch 1 | iter 1184 step 37 | loss train: 1.385, val: 1.397 | iter time: 359.10 ms (step) remaining time: 0:39:19
|
| 149 |
+
Epoch 1 | iter 1216 step 38 | loss train: 1.473, val: 1.397 | iter time: 359.89 ms (step) remaining time: 0:39:08
|
| 150 |
+
Epoch 1 | iter 1248 step 39 | loss train: 1.340, val: 1.397 | iter time: 358.85 ms (step) remaining time: 0:38:57
|
| 151 |
+
Epoch 1 | iter 1280 step 40 | loss train: 1.378, val: 1.397 | iter time: 358.18 ms (step) remaining time: 0:38:46
|
| 152 |
+
Epoch 1 | iter 1312 step 41 | loss train: 1.403, val: 1.397 | iter time: 359.14 ms (step) remaining time: 0:38:36
|
| 153 |
+
Epoch 1 | iter 1344 step 42 | loss train: 1.380, val: 1.397 | iter time: 358.61 ms (step) remaining time: 0:38:26
|
| 154 |
+
Epoch 1 | iter 1376 step 43 | loss train: 1.413, val: 1.397 | iter time: 360.20 ms (step) remaining time: 0:38:15
|
| 155 |
+
Epoch 1 | iter 1408 step 44 | loss train: 1.352, val: 1.397 | iter time: 359.32 ms (step) remaining time: 0:38:04
|
| 156 |
+
Epoch 1 | iter 1440 step 45 | loss train: 1.362, val: 1.397 | iter time: 358.92 ms (step) remaining time: 0:37:53
|
| 157 |
+
Epoch 1 | iter 1472 step 46 | loss train: 1.383, val: 1.397 | iter time: 359.46 ms (step) remaining time: 0:37:42
|
| 158 |
+
Epoch 1 | iter 1504 step 47 | loss train: 1.391, val: 1.397 | iter time: 359.28 ms (step) remaining time: 0:37:31
|
| 159 |
+
Epoch 1 | iter 1536 step 48 | loss train: 1.394, val: 1.397 | iter time: 360.60 ms (step) remaining time: 0:37:20
|
| 160 |
+
Epoch 1 | iter 1568 step 49 | loss train: 1.350, val: 1.397 | iter time: 359.75 ms (step) remaining time: 0:37:09
|
| 161 |
+
Epoch 1 | iter 1600 step 50 | loss train: 1.448, val: 1.397 | iter time: 358.81 ms (step) remaining time: 0:36:58
|
| 162 |
+
Validating ...
|
| 163 |
+
iter 1600: val loss 1.3935, val time: 22346.96 ms
|
| 164 |
+
Epoch 1 | iter 1632 step 51 | loss train: 1.403, val: 1.393 | iter time: 359.56 ms (step) remaining time: 0:38:18
|
| 165 |
+
Epoch 1 | iter 1664 step 52 | loss train: 1.387, val: 1.393 | iter time: 361.43 ms (step) remaining time: 0:38:05
|
| 166 |
+
Epoch 1 | iter 1696 step 53 | loss train: 1.407, val: 1.393 | iter time: 361.57 ms (step) remaining time: 0:37:52
|
| 167 |
+
Epoch 1 | iter 1728 step 54 | loss train: 1.370, val: 1.393 | iter time: 360.65 ms (step) remaining time: 0:37:39
|
| 168 |
+
Epoch 1 | iter 1760 step 55 | loss train: 1.373, val: 1.393 | iter time: 360.84 ms (step) remaining time: 0:37:27
|
| 169 |
+
Epoch 1 | iter 1792 step 56 | loss train: 1.378, val: 1.393 | iter time: 358.78 ms (step) remaining time: 0:37:14
|
| 170 |
+
Epoch 1 | iter 1824 step 57 | loss train: 1.349, val: 1.393 | iter time: 358.85 ms (step) remaining time: 0:37:01
|
| 171 |
+
Epoch 1 | iter 1856 step 58 | loss train: 1.390, val: 1.393 | iter time: 359.22 ms (step) remaining time: 0:36:48
|
| 172 |
+
Epoch 1 | iter 1888 step 59 | loss train: 1.387, val: 1.393 | iter time: 360.96 ms (step) remaining time: 0:36:36
|
| 173 |
+
Epoch 1 | iter 1920 step 60 | loss train: 1.375, val: 1.393 | iter time: 359.90 ms (step) remaining time: 0:36:23
|
| 174 |
+
Epoch 1 | iter 1952 step 61 | loss train: 1.416, val: 1.393 | iter time: 358.98 ms (step) remaining time: 0:36:11
|
| 175 |
+
Epoch 1 | iter 1984 step 62 | loss train: 1.415, val: 1.393 | iter time: 360.22 ms (step) remaining time: 0:35:58
|
| 176 |
+
Epoch 1 | iter 2016 step 63 | loss train: 1.403, val: 1.393 | iter time: 360.85 ms (step) remaining time: 0:35:46
|
| 177 |
+
Epoch 1 | iter 2048 step 64 | loss train: 1.423, val: 1.393 | iter time: 361.80 ms (step) remaining time: 0:35:34
|
| 178 |
+
Epoch 1 | iter 2080 step 65 | loss train: 1.405, val: 1.393 | iter time: 360.37 ms (step) remaining time: 0:35:21
|
| 179 |
+
Epoch 1 | iter 2112 step 66 | loss train: 1.425, val: 1.393 | iter time: 361.07 ms (step) remaining time: 0:35:09
|
| 180 |
+
Epoch 1 | iter 2144 step 67 | loss train: 1.389, val: 1.393 | iter time: 358.03 ms (step) remaining time: 0:34:57
|
| 181 |
+
Epoch 1 | iter 2176 step 68 | loss train: 1.388, val: 1.393 | iter time: 359.64 ms (step) remaining time: 0:34:45
|
| 182 |
+
Epoch 1 | iter 2208 step 69 | loss train: 1.378, val: 1.393 | iter time: 360.29 ms (step) remaining time: 0:34:33
|
| 183 |
+
Epoch 1 | iter 2240 step 70 | loss train: 1.458, val: 1.393 | iter time: 360.03 ms (step) remaining time: 0:34:21
|
| 184 |
+
Epoch 1 | iter 2272 step 71 | loss train: 1.333, val: 1.393 | iter time: 359.79 ms (step) remaining time: 0:34:09
|
| 185 |
+
Epoch 1 | iter 2304 step 72 | loss train: 1.449, val: 1.393 | iter time: 359.22 ms (step) remaining time: 0:33:57
|
| 186 |
+
Epoch 1 | iter 2336 step 73 | loss train: 1.368, val: 1.393 | iter time: 361.36 ms (step) remaining time: 0:33:45
|
| 187 |
+
Epoch 1 | iter 2368 step 74 | loss train: 1.391, val: 1.393 | iter time: 360.59 ms (step) remaining time: 0:33:33
|
| 188 |
+
Epoch 1 | iter 2400 step 75 | loss train: 1.346, val: 1.393 | iter time: 360.80 ms (step) remaining time: 0:33:21
|
| 189 |
+
Epoch 1 | iter 2432 step 76 | loss train: 1.389, val: 1.393 | iter time: 358.67 ms (step) remaining time: 0:33:09
|
| 190 |
+
Epoch 1 | iter 2464 step 77 | loss train: 1.429, val: 1.393 | iter time: 362.44 ms (step) remaining time: 0:32:57
|
| 191 |
+
Epoch 1 | iter 2496 step 78 | loss train: 1.402, val: 1.393 | iter time: 359.60 ms (step) remaining time: 0:32:45
|
| 192 |
+
Epoch 1 | iter 2528 step 79 | loss train: 1.414, val: 1.393 | iter time: 360.31 ms (step) remaining time: 0:32:33
|
| 193 |
+
Epoch 1 | iter 2560 step 80 | loss train: 1.360, val: 1.393 | iter time: 360.49 ms (step) remaining time: 0:32:21
|
| 194 |
+
Epoch 1 | iter 2592 step 81 | loss train: 1.373, val: 1.393 | iter time: 361.48 ms (step) remaining time: 0:32:10
|
| 195 |
+
Epoch 1 | iter 2624 step 82 | loss train: 1.359, val: 1.393 | iter time: 358.61 ms (step) remaining time: 0:31:58
|
| 196 |
+
Epoch 1 | iter 2656 step 83 | loss train: 1.476, val: 1.393 | iter time: 358.76 ms (step) remaining time: 0:31:46
|
| 197 |
+
Epoch 1 | iter 2688 step 84 | loss train: 1.417, val: 1.393 | iter time: 360.95 ms (step) remaining time: 0:31:34
|
| 198 |
+
Epoch 1 | iter 2720 step 85 | loss train: 1.400, val: 1.393 | iter time: 359.84 ms (step) remaining time: 0:31:23
|
| 199 |
+
Epoch 1 | iter 2752 step 86 | loss train: 1.314, val: 1.393 | iter time: 360.06 ms (step) remaining time: 0:31:11
|
| 200 |
+
Epoch 1 | iter 2784 step 87 | loss train: 1.378, val: 1.393 | iter time: 359.69 ms (step) remaining time: 0:30:59
|
| 201 |
+
Epoch 1 | iter 2816 step 88 | loss train: 1.388, val: 1.393 | iter time: 358.09 ms (step) remaining time: 0:30:48
|
| 202 |
+
Epoch 1 | iter 2848 step 89 | loss train: 1.460, val: 1.393 | iter time: 358.62 ms (step) remaining time: 0:30:36
|
| 203 |
+
Epoch 1 | iter 2880 step 90 | loss train: 1.413, val: 1.393 | iter time: 358.29 ms (step) remaining time: 0:30:25
|
| 204 |
+
Epoch 1 | iter 2912 step 91 | loss train: 1.357, val: 1.393 | iter time: 359.89 ms (step) remaining time: 0:30:13
|
| 205 |
+
Epoch 1 | iter 2944 step 92 | loss train: 1.353, val: 1.393 | iter time: 360.57 ms (step) remaining time: 0:30:01
|
| 206 |
+
Epoch 1 | iter 2976 step 93 | loss train: 1.370, val: 1.393 | iter time: 359.00 ms (step) remaining time: 0:29:50
|
| 207 |
+
Epoch 1 | iter 3008 step 94 | loss train: 1.397, val: 1.393 | iter time: 358.75 ms (step) remaining time: 0:29:38
|
| 208 |
+
Epoch 1 | iter 3040 step 95 | loss train: 1.385, val: 1.393 | iter time: 358.80 ms (step) remaining time: 0:29:27
|
| 209 |
+
Epoch 1 | iter 3072 step 96 | loss train: 1.370, val: 1.393 | iter time: 361.56 ms (step) remaining time: 0:29:15
|
| 210 |
+
Epoch 1 | iter 3104 step 97 | loss train: 1.405, val: 1.393 | iter time: 360.19 ms (step) remaining time: 0:29:04
|
| 211 |
+
Epoch 1 | iter 3136 step 98 | loss train: 1.418, val: 1.393 | iter time: 359.04 ms (step) remaining time: 0:28:52
|
| 212 |
+
Epoch 1 | iter 3168 step 99 | loss train: 1.384, val: 1.393 | iter time: 358.87 ms (step) remaining time: 0:28:41
|
| 213 |
+
Epoch 1 | iter 3200 step 100 | loss train: 1.380, val: 1.393 | iter time: 360.67 ms (step) remaining time: 0:28:29
|
| 214 |
+
Validating ...
|
| 215 |
+
iter 3200: val loss 1.3468, val time: 22312.98 ms
|
| 216 |
+
Saving checkpoint to 'out/pretrain/2501_full/step-00000100/lit_model.pth'
|
| 217 |
+
Epoch 1 | iter 3232 step 101 | loss train: 1.368, val: 1.347 | iter time: 358.34 ms (step) remaining time: 0:29:17
|
| 218 |
+
Epoch 1 | iter 3264 step 102 | loss train: 1.419, val: 1.347 | iter time: 359.27 ms (step) remaining time: 0:29:05
|
| 219 |
+
Epoch 1 | iter 3296 step 103 | loss train: 1.440, val: 1.347 | iter time: 359.18 ms (step) remaining time: 0:28:52
|
| 220 |
+
Epoch 1 | iter 3328 step 104 | loss train: 1.408, val: 1.347 | iter time: 361.67 ms (step) remaining time: 0:28:40
|
| 221 |
+
Epoch 1 | iter 3360 step 105 | loss train: 1.382, val: 1.347 | iter time: 359.36 ms (step) remaining time: 0:28:28
|
| 222 |
+
Epoch 1 | iter 3392 step 106 | loss train: 1.395, val: 1.347 | iter time: 358.25 ms (step) remaining time: 0:28:16
|
| 223 |
+
Epoch 1 | iter 3424 step 107 | loss train: 1.321, val: 1.347 | iter time: 359.61 ms (step) remaining time: 0:28:03
|
| 224 |
+
Epoch 1 | iter 3456 step 108 | loss train: 1.329, val: 1.347 | iter time: 359.73 ms (step) remaining time: 0:27:51
|
| 225 |
+
Epoch 1 | iter 3488 step 109 | loss train: 1.373, val: 1.347 | iter time: 361.43 ms (step) remaining time: 0:27:39
|
| 226 |
+
Epoch 1 | iter 3520 step 110 | loss train: 1.408, val: 1.347 | iter time: 359.68 ms (step) remaining time: 0:27:27
|
| 227 |
+
Epoch 1 | iter 3552 step 111 | loss train: 1.320, val: 1.347 | iter time: 359.40 ms (step) remaining time: 0:27:14
|
| 228 |
+
Epoch 1 | iter 3584 step 112 | loss train: 1.341, val: 1.347 | iter time: 358.33 ms (step) remaining time: 0:27:02
|
| 229 |
+
Epoch 1 | iter 3616 step 113 | loss train: 1.358, val: 1.347 | iter time: 359.00 ms (step) remaining time: 0:26:50
|
| 230 |
+
Epoch 1 | iter 3648 step 114 | loss train: 1.407, val: 1.347 | iter time: 359.78 ms (step) remaining time: 0:26:38
|
| 231 |
+
Epoch 1 | iter 3680 step 115 | loss train: 1.394, val: 1.347 | iter time: 357.77 ms (step) remaining time: 0:26:26
|
| 232 |
+
Epoch 1 | iter 3712 step 116 | loss train: 1.353, val: 1.347 | iter time: 361.80 ms (step) remaining time: 0:26:14
|
| 233 |
+
Epoch 1 | iter 3744 step 117 | loss train: 1.401, val: 1.347 | iter time: 359.22 ms (step) remaining time: 0:26:02
|
| 234 |
+
Epoch 1 | iter 3776 step 118 | loss train: 1.401, val: 1.347 | iter time: 359.99 ms (step) remaining time: 0:25:50
|
| 235 |
+
Epoch 1 | iter 3808 step 119 | loss train: 1.425, val: 1.347 | iter time: 359.97 ms (step) remaining time: 0:25:38
|
| 236 |
+
Epoch 1 | iter 3840 step 120 | loss train: 1.401, val: 1.347 | iter time: 360.56 ms (step) remaining time: 0:25:26
|
| 237 |
+
Epoch 1 | iter 3872 step 121 | loss train: 1.383, val: 1.347 | iter time: 358.77 ms (step) remaining time: 0:25:14
|
| 238 |
+
Epoch 1 | iter 3904 step 122 | loss train: 1.359, val: 1.347 | iter time: 361.14 ms (step) remaining time: 0:25:02
|
| 239 |
+
Epoch 1 | iter 3936 step 123 | loss train: 1.454, val: 1.347 | iter time: 359.28 ms (step) remaining time: 0:24:51
|
| 240 |
+
Epoch 1 | iter 3968 step 124 | loss train: 1.363, val: 1.347 | iter time: 360.13 ms (step) remaining time: 0:24:39
|
| 241 |
+
Epoch 1 | iter 4000 step 125 | loss train: 1.365, val: 1.347 | iter time: 360.35 ms (step) remaining time: 0:24:27
|
| 242 |
+
Epoch 1 | iter 4032 step 126 | loss train: 1.340, val: 1.347 | iter time: 358.83 ms (step) remaining time: 0:24:15
|
| 243 |
+
Epoch 1 | iter 4064 step 127 | loss train: 1.352, val: 1.347 | iter time: 362.50 ms (step) remaining time: 0:24:03
|
| 244 |
+
Epoch 1 | iter 4096 step 128 | loss train: 1.310, val: 1.347 | iter time: 359.94 ms (step) remaining time: 0:23:51
|
| 245 |
+
Epoch 1 | iter 4128 step 129 | loss train: 1.367, val: 1.347 | iter time: 359.28 ms (step) remaining time: 0:23:39
|
| 246 |
+
Epoch 1 | iter 4160 step 130 | loss train: 1.372, val: 1.347 | iter time: 359.84 ms (step) remaining time: 0:23:27
|
| 247 |
+
Epoch 1 | iter 4192 step 131 | loss train: 1.379, val: 1.347 | iter time: 359.89 ms (step) remaining time: 0:23:16
|
| 248 |
+
Epoch 1 | iter 4224 step 132 | loss train: 1.371, val: 1.347 | iter time: 359.55 ms (step) remaining time: 0:23:04
|
| 249 |
+
Epoch 1 | iter 4256 step 133 | loss train: 1.293, val: 1.347 | iter time: 359.16 ms (step) remaining time: 0:22:52
|
| 250 |
+
Epoch 1 | iter 4288 step 134 | loss train: 1.357, val: 1.347 | iter time: 360.69 ms (step) remaining time: 0:22:40
|
| 251 |
+
Epoch 1 | iter 4320 step 135 | loss train: 1.372, val: 1.347 | iter time: 358.83 ms (step) remaining time: 0:22:28
|
| 252 |
+
Epoch 1 | iter 4352 step 136 | loss train: 1.380, val: 1.347 | iter time: 359.60 ms (step) remaining time: 0:22:17
|
| 253 |
+
Epoch 1 | iter 4384 step 137 | loss train: 1.325, val: 1.347 | iter time: 359.80 ms (step) remaining time: 0:22:05
|
| 254 |
+
Epoch 1 | iter 4416 step 138 | loss train: 1.310, val: 1.347 | iter time: 360.75 ms (step) remaining time: 0:21:53
|
| 255 |
+
Epoch 1 | iter 4448 step 139 | loss train: 1.343, val: 1.347 | iter time: 360.34 ms (step) remaining time: 0:21:42
|
| 256 |
+
Epoch 1 | iter 4480 step 140 | loss train: 1.405, val: 1.347 | iter time: 358.82 ms (step) remaining time: 0:21:30
|
| 257 |
+
Epoch 1 | iter 4512 step 141 | loss train: 1.398, val: 1.347 | iter time: 359.01 ms (step) remaining time: 0:21:18
|
| 258 |
+
Epoch 1 | iter 4544 step 142 | loss train: 1.304, val: 1.347 | iter time: 361.06 ms (step) remaining time: 0:21:07
|
| 259 |
+
Epoch 1 | iter 4576 step 143 | loss train: 1.335, val: 1.347 | iter time: 361.51 ms (step) remaining time: 0:20:55
|
| 260 |
+
Epoch 1 | iter 4608 step 144 | loss train: 1.267, val: 1.347 | iter time: 360.34 ms (step) remaining time: 0:20:43
|
| 261 |
+
Epoch 1 | iter 4640 step 145 | loss train: 1.310, val: 1.347 | iter time: 359.48 ms (step) remaining time: 0:20:32
|
| 262 |
+
Epoch 1 | iter 4672 step 146 | loss train: 1.347, val: 1.347 | iter time: 360.42 ms (step) remaining time: 0:20:20
|
| 263 |
+
Epoch 1 | iter 4704 step 147 | loss train: 1.340, val: 1.347 | iter time: 360.08 ms (step) remaining time: 0:20:08
|
| 264 |
+
Epoch 1 | iter 4736 step 148 | loss train: 1.320, val: 1.347 | iter time: 360.98 ms (step) remaining time: 0:19:57
|
| 265 |
+
Epoch 1 | iter 4768 step 149 | loss train: 1.306, val: 1.347 | iter time: 359.88 ms (step) remaining time: 0:19:45
|
| 266 |
+
Epoch 1 | iter 4800 step 150 | loss train: 1.354, val: 1.347 | iter time: 358.99 ms (step) remaining time: 0:19:33
|
| 267 |
+
Validating ...
|
| 268 |
+
iter 4800: val loss 1.2983, val time: 22355.73 ms
|
| 269 |
+
Epoch 1 | iter 4832 step 151 | loss train: 1.368, val: 1.298 | iter time: 359.06 ms (step) remaining time: 0:19:37
|
| 270 |
+
Epoch 1 | iter 4864 step 152 | loss train: 1.281, val: 1.298 | iter time: 358.58 ms (step) remaining time: 0:19:25
|
| 271 |
+
Epoch 1 | iter 4896 step 153 | loss train: 1.368, val: 1.298 | iter time: 360.00 ms (step) remaining time: 0:19:14
|
| 272 |
+
Epoch 1 | iter 4928 step 154 | loss train: 1.267, val: 1.298 | iter time: 358.61 ms (step) remaining time: 0:19:02
|
| 273 |
+
Epoch 1 | iter 4960 step 155 | loss train: 1.349, val: 1.298 | iter time: 359.30 ms (step) remaining time: 0:18:50
|
| 274 |
+
Epoch 1 | iter 4992 step 156 | loss train: 1.388, val: 1.298 | iter time: 359.72 ms (step) remaining time: 0:18:38
|
| 275 |
+
Epoch 1 | iter 5024 step 157 | loss train: 1.293, val: 1.298 | iter time: 360.07 ms (step) remaining time: 0:18:26
|
| 276 |
+
Epoch 1 | iter 5056 step 158 | loss train: 1.322, val: 1.298 | iter time: 360.49 ms (step) remaining time: 0:18:15
|
| 277 |
+
Epoch 1 | iter 5088 step 159 | loss train: 1.377, val: 1.298 | iter time: 359.55 ms (step) remaining time: 0:18:03
|
| 278 |
+
Epoch 1 | iter 5120 step 160 | loss train: 1.392, val: 1.298 | iter time: 360.04 ms (step) remaining time: 0:17:51
|
| 279 |
+
Epoch 1 | iter 5152 step 161 | loss train: 1.413, val: 1.298 | iter time: 360.48 ms (step) remaining time: 0:17:39
|
| 280 |
+
Epoch 1 | iter 5184 step 162 | loss train: 1.471, val: 1.298 | iter time: 358.96 ms (step) remaining time: 0:17:28
|
| 281 |
+
Epoch 1 | iter 5216 step 163 | loss train: 1.373, val: 1.298 | iter time: 359.37 ms (step) remaining time: 0:17:16
|
| 282 |
+
Epoch 1 | iter 5248 step 164 | loss train: 1.304, val: 1.298 | iter time: 358.78 ms (step) remaining time: 0:17:04
|
| 283 |
+
Epoch 1 | iter 5280 step 165 | loss train: 1.378, val: 1.298 | iter time: 360.73 ms (step) remaining time: 0:16:53
|
| 284 |
+
Epoch 1 | iter 5312 step 166 | loss train: 1.344, val: 1.298 | iter time: 358.77 ms (step) remaining time: 0:16:41
|
| 285 |
+
Epoch 1 | iter 5344 step 167 | loss train: 1.392, val: 1.298 | iter time: 361.08 ms (step) remaining time: 0:16:29
|
| 286 |
+
Epoch 1 | iter 5376 step 168 | loss train: 1.375, val: 1.298 | iter time: 359.46 ms (step) remaining time: 0:16:18
|
| 287 |
+
Epoch 1 | iter 5408 step 169 | loss train: 1.359, val: 1.298 | iter time: 359.83 ms (step) remaining time: 0:16:06
|
| 288 |
+
Epoch 1 | iter 5440 step 170 | loss train: 1.386, val: 1.298 | iter time: 358.48 ms (step) remaining time: 0:15:55
|
| 289 |
+
Epoch 1 | iter 5472 step 171 | loss train: 1.346, val: 1.298 | iter time: 359.73 ms (step) remaining time: 0:15:43
|
| 290 |
+
Epoch 1 | iter 5504 step 172 | loss train: 1.356, val: 1.298 | iter time: 359.56 ms (step) remaining time: 0:15:31
|
| 291 |
+
Epoch 1 | iter 5536 step 173 | loss train: 1.367, val: 1.298 | iter time: 359.31 ms (step) remaining time: 0:15:20
|
| 292 |
+
Epoch 1 | iter 5568 step 174 | loss train: 1.336, val: 1.298 | iter time: 357.40 ms (step) remaining time: 0:15:08
|
| 293 |
+
Epoch 1 | iter 5600 step 175 | loss train: 1.339, val: 1.298 | iter time: 358.61 ms (step) remaining time: 0:14:57
|
| 294 |
+
Epoch 1 | iter 5632 step 176 | loss train: 1.367, val: 1.298 | iter time: 358.90 ms (step) remaining time: 0:14:45
|
| 295 |
+
Epoch 1 | iter 5664 step 177 | loss train: 1.427, val: 1.298 | iter time: 358.81 ms (step) remaining time: 0:14:33
|
| 296 |
+
Epoch 1 | iter 5696 step 178 | loss train: 1.292, val: 1.298 | iter time: 358.37 ms (step) remaining time: 0:14:22
|
| 297 |
+
Epoch 1 | iter 5728 step 179 | loss train: 1.315, val: 1.298 | iter time: 360.13 ms (step) remaining time: 0:14:10
|
| 298 |
+
Epoch 1 | iter 5760 step 180 | loss train: 1.341, val: 1.298 | iter time: 360.82 ms (step) remaining time: 0:13:59
|
| 299 |
+
Epoch 1 | iter 5792 step 181 | loss train: 1.345, val: 1.298 | iter time: 359.64 ms (step) remaining time: 0:13:47
|
| 300 |
+
Epoch 1 | iter 5824 step 182 | loss train: 1.349, val: 1.298 | iter time: 359.87 ms (step) remaining time: 0:13:36
|
| 301 |
+
Epoch 1 | iter 5856 step 183 | loss train: 1.343, val: 1.298 | iter time: 360.25 ms (step) remaining time: 0:13:24
|
| 302 |
+
Epoch 1 | iter 5888 step 184 | loss train: 1.327, val: 1.298 | iter time: 359.87 ms (step) remaining time: 0:13:13
|
| 303 |
+
Epoch 1 | iter 5920 step 185 | loss train: 1.292, val: 1.298 | iter time: 360.61 ms (step) remaining time: 0:13:01
|
| 304 |
+
Epoch 1 | iter 5952 step 186 | loss train: 1.345, val: 1.298 | iter time: 358.24 ms (step) remaining time: 0:12:50
|
| 305 |
+
Epoch 1 | iter 5984 step 187 | loss train: 1.355, val: 1.298 | iter time: 360.62 ms (step) remaining time: 0:12:38
|
| 306 |
+
Epoch 1 | iter 6016 step 188 | loss train: 1.360, val: 1.298 | iter time: 360.38 ms (step) remaining time: 0:12:27
|
| 307 |
+
Epoch 1 | iter 6048 step 189 | loss train: 1.332, val: 1.298 | iter time: 360.11 ms (step) remaining time: 0:12:15
|
| 308 |
+
Epoch 1 | iter 6080 step 190 | loss train: 1.324, val: 1.298 | iter time: 360.02 ms (step) remaining time: 0:12:04
|
| 309 |
+
Epoch 1 | iter 6112 step 191 | loss train: 1.351, val: 1.298 | iter time: 360.18 ms (step) remaining time: 0:11:52
|
| 310 |
+
Epoch 1 | iter 6144 step 192 | loss train: 1.316, val: 1.298 | iter time: 358.14 ms (step) remaining time: 0:11:41
|
| 311 |
+
Epoch 1 | iter 6176 step 193 | loss train: 1.433, val: 1.298 | iter time: 359.33 ms (step) remaining time: 0:11:30
|
| 312 |
+
Epoch 1 | iter 6208 step 194 | loss train: 1.400, val: 1.298 | iter time: 361.19 ms (step) remaining time: 0:11:18
|
| 313 |
+
Epoch 1 | iter 6240 step 195 | loss train: 1.394, val: 1.298 | iter time: 359.00 ms (step) remaining time: 0:11:07
|
| 314 |
+
Epoch 1 | iter 6272 step 196 | loss train: 1.341, val: 1.298 | iter time: 358.43 ms (step) remaining time: 0:10:55
|
| 315 |
+
Epoch 1 | iter 6304 step 197 | loss train: 1.340, val: 1.298 | iter time: 358.66 ms (step) remaining time: 0:10:44
|
| 316 |
+
Epoch 1 | iter 6336 step 198 | loss train: 1.325, val: 1.298 | iter time: 360.47 ms (step) remaining time: 0:10:32
|
| 317 |
+
Epoch 1 | iter 6368 step 199 | loss train: 1.261, val: 1.298 | iter time: 360.92 ms (step) remaining time: 0:10:21
|
| 318 |
+
Epoch 1 | iter 6400 step 200 | loss train: 1.374, val: 1.298 | iter time: 361.31 ms (step) remaining time: 0:10:09
|
| 319 |
+
Validating ...
|
| 320 |
+
iter 6400: val loss 1.2520, val time: 22361.19 ms
|
| 321 |
+
Saving checkpoint to 'out/pretrain/2501_full/step-00000200/lit_model.pth'
|
| 322 |
+
Epoch 1 | iter 6432 step 201 | loss train: 1.287, val: 1.252 | iter time: 356.95 ms (step) remaining time: 0:10:08
|
| 323 |
+
Epoch 1 | iter 6464 step 202 | loss train: 1.426, val: 1.252 | iter time: 357.79 ms (step) remaining time: 0:09:57
|
| 324 |
+
Epoch 1 | iter 6496 step 203 | loss train: 1.360, val: 1.252 | iter time: 356.87 ms (step) remaining time: 0:09:45
|
| 325 |
+
Epoch 1 | iter 6528 step 204 | loss train: 1.257, val: 1.252 | iter time: 360.23 ms (step) remaining time: 0:09:33
|
| 326 |
+
Epoch 1 | iter 6560 step 205 | loss train: 1.297, val: 1.252 | iter time: 360.28 ms (step) remaining time: 0:09:22
|
| 327 |
+
Epoch 1 | iter 6592 step 206 | loss train: 1.312, val: 1.252 | iter time: 360.69 ms (step) remaining time: 0:09:10
|
| 328 |
+
Epoch 1 | iter 6624 step 207 | loss train: 1.291, val: 1.252 | iter time: 358.29 ms (step) remaining time: 0:08:59
|
| 329 |
+
Epoch 1 | iter 6656 step 208 | loss train: 1.322, val: 1.252 | iter time: 358.72 ms (step) remaining time: 0:08:47
|
| 330 |
+
Epoch 1 | iter 6688 step 209 | loss train: 1.394, val: 1.252 | iter time: 359.99 ms (step) remaining time: 0:08:35
|
| 331 |
+
Epoch 1 | iter 6720 step 210 | loss train: 1.287, val: 1.252 | iter time: 357.99 ms (step) remaining time: 0:08:24
|
| 332 |
+
Epoch 1 | iter 6752 step 211 | loss train: 1.333, val: 1.252 | iter time: 359.27 ms (step) remaining time: 0:08:12
|
| 333 |
+
Epoch 1 | iter 6784 step 212 | loss train: 1.279, val: 1.252 | iter time: 360.06 ms (step) remaining time: 0:08:01
|
| 334 |
+
Epoch 1 | iter 6816 step 213 | loss train: 1.419, val: 1.252 | iter time: 358.95 ms (step) remaining time: 0:07:49
|
| 335 |
+
Epoch 1 | iter 6848 step 214 | loss train: 1.332, val: 1.252 | iter time: 359.75 ms (step) remaining time: 0:07:37
|
| 336 |
+
Epoch 1 | iter 6880 step 215 | loss train: 1.278, val: 1.252 | iter time: 359.60 ms (step) remaining time: 0:07:26
|
| 337 |
+
Epoch 1 | iter 6912 step 216 | loss train: 1.283, val: 1.252 | iter time: 360.78 ms (step) remaining time: 0:07:14
|
| 338 |
+
Epoch 1 | iter 6944 step 217 | loss train: 1.324, val: 1.252 | iter time: 360.59 ms (step) remaining time: 0:07:03
|
| 339 |
+
Epoch 1 | iter 6976 step 218 | loss train: 1.277, val: 1.252 | iter time: 361.70 ms (step) remaining time: 0:06:51
|
| 340 |
+
Epoch 1 | iter 7008 step 219 | loss train: 1.306, val: 1.252 | iter time: 357.83 ms (step) remaining time: 0:06:40
|
| 341 |
+
Epoch 1 | iter 7040 step 220 | loss train: 1.367, val: 1.252 | iter time: 358.98 ms (step) remaining time: 0:06:28
|
| 342 |
+
Epoch 1 | iter 7072 step 221 | loss train: 1.402, val: 1.252 | iter time: 360.28 ms (step) remaining time: 0:06:17
|
| 343 |
+
Epoch 1 | iter 7104 step 222 | loss train: 1.307, val: 1.252 | iter time: 360.30 ms (step) remaining time: 0:06:05
|
| 344 |
+
Epoch 1 | iter 7136 step 223 | loss train: 1.291, val: 1.252 | iter time: 359.98 ms (step) remaining time: 0:05:54
|
| 345 |
+
Epoch 1 | iter 7168 step 224 | loss train: 1.316, val: 1.252 | iter time: 361.93 ms (step) remaining time: 0:05:42
|
| 346 |
+
Epoch 1 | iter 7200 step 225 | loss train: 1.304, val: 1.252 | iter time: 359.45 ms (step) remaining time: 0:05:31
|
| 347 |
+
Epoch 1 | iter 7232 step 226 | loss train: 1.264, val: 1.252 | iter time: 359.71 ms (step) remaining time: 0:05:19
|
| 348 |
+
Epoch 1 | iter 7264 step 227 | loss train: 1.299, val: 1.252 | iter time: 360.36 ms (step) remaining time: 0:05:08
|
| 349 |
+
Epoch 1 | iter 7296 step 228 | loss train: 1.326, val: 1.252 | iter time: 358.74 ms (step) remaining time: 0:04:56
|
| 350 |
+
Epoch 1 | iter 7328 step 229 | loss train: 1.233, val: 1.252 | iter time: 359.42 ms (step) remaining time: 0:04:45
|
| 351 |
+
Epoch 1 | iter 7360 step 230 | loss train: 1.288, val: 1.252 | iter time: 360.07 ms (step) remaining time: 0:04:33
|
| 352 |
+
Epoch 1 | iter 7392 step 231 | loss train: 1.249, val: 1.252 | iter time: 360.96 ms (step) remaining time: 0:04:22
|
| 353 |
+
Epoch 1 | iter 7424 step 232 | loss train: 1.300, val: 1.252 | iter time: 361.62 ms (step) remaining time: 0:04:10
|
| 354 |
+
Epoch 1 | iter 7456 step 233 | loss train: 1.320, val: 1.252 | iter time: 359.72 ms (step) remaining time: 0:03:59
|
| 355 |
+
Epoch 1 | iter 7488 step 234 | loss train: 1.304, val: 1.252 | iter time: 358.41 ms (step) remaining time: 0:03:48
|
| 356 |
+
Epoch 1 | iter 7520 step 235 | loss train: 1.363, val: 1.252 | iter time: 360.15 ms (step) remaining time: 0:03:36
|
| 357 |
+
Epoch 1 | iter 7552 step 236 | loss train: 1.247, val: 1.252 | iter time: 359.70 ms (step) remaining time: 0:03:25
|
| 358 |
+
Epoch 1 | iter 7584 step 237 | loss train: 1.220, val: 1.252 | iter time: 358.00 ms (step) remaining time: 0:03:13
|
| 359 |
+
Epoch 1 | iter 7616 step 238 | loss train: 1.325, val: 1.252 | iter time: 359.85 ms (step) remaining time: 0:03:02
|
| 360 |
+
Epoch 1 | iter 7648 step 239 | loss train: 1.351, val: 1.252 | iter time: 357.77 ms (step) remaining time: 0:02:50
|
| 361 |
+
Epoch 1 | iter 7680 step 240 | loss train: 1.251, val: 1.252 | iter time: 360.29 ms (step) remaining time: 0:02:39
|
| 362 |
+
Epoch 1 | iter 7712 step 241 | loss train: 1.265, val: 1.252 | iter time: 360.02 ms (step) remaining time: 0:02:27
|
| 363 |
+
Epoch 1 | iter 7744 step 242 | loss train: 1.270, val: 1.252 | iter time: 361.47 ms (step) remaining time: 0:02:16
|
| 364 |
+
Epoch 1 | iter 7776 step 243 | loss train: 1.287, val: 1.252 | iter time: 359.72 ms (step) remaining time: 0:02:05
|
| 365 |
+
Epoch 1 | iter 7808 step 244 | loss train: 1.293, val: 1.252 | iter time: 359.49 ms (step) remaining time: 0:01:53
|
| 366 |
+
Epoch 1 | iter 7840 step 245 | loss train: 1.219, val: 1.252 | iter time: 358.47 ms (step) remaining time: 0:01:42
|
| 367 |
+
Epoch 1 | iter 7872 step 246 | loss train: 1.378, val: 1.252 | iter time: 360.27 ms (step) remaining time: 0:01:31
|
| 368 |
+
Epoch 1 | iter 7904 step 247 | loss train: 1.245, val: 1.252 | iter time: 357.73 ms (step) remaining time: 0:01:19
|
| 369 |
+
Epoch 1 | iter 7936 step 248 | loss train: 1.227, val: 1.252 | iter time: 358.04 ms (step) remaining time: 0:01:08
|
| 370 |
+
Epoch 1 | iter 7968 step 249 | loss train: 1.286, val: 1.252 | iter time: 359.20 ms (step) remaining time: 0:00:56
|
| 371 |
+
Epoch 1 | iter 8000 step 250 | loss train: 1.272, val: 1.252 | iter time: 359.58 ms (step) remaining time: 0:00:45
|
| 372 |
+
Validating ...
|
| 373 |
+
iter 8000: val loss 1.2134, val time: 22383.99 ms
|
| 374 |
+
Epoch 1 | iter 8032 step 251 | loss train: 1.333, val: 1.213 | iter time: 360.62 ms (step) remaining time: 0:00:34
|
| 375 |
+
Epoch 1 | iter 8064 step 252 | loss train: 1.351, val: 1.213 | iter time: 359.46 ms (step) remaining time: 0:00:22
|
| 376 |
+
Epoch 1 | iter 8096 step 253 | loss train: 1.283, val: 1.213 | iter time: 361.84 ms (step) remaining time: 0:00:11
|
| 377 |
+
Epoch 2 | iter 8128 step 254 | loss train: 1.321, val: 1.213 | iter time: 360.06 ms (step) remaining time: 0:00:00
|
| 378 |
+
Validating ...
|
| 379 |
+
Final evaluation | val loss: 1.210 | val ppl: 3.354
|
| 380 |
+
Saving checkpoint to 'out/pretrain/2501_full/final/lit_model.pth'
|
| 381 |
+
----------------------------------------
|
| 382 |
+
| Performance
|
| 383 |
+
| - Total tokens : 266,338,304
|
| 384 |
+
| - Training Time : 2971.64 s
|
| 385 |
+
| - Tok/sec : 143.04 tok/s
|
| 386 |
+
| ----------------------------------------
|
| 387 |
+
| Memory Usage
|
| 388 |
+
| - Memory Used : 26.32 GB
|
| 389 |
+
----------------------------------------
|
out/pretrain/tinyllama_3_epoch/2407/final/config.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"LlamaForCausalLM"
|
| 4 |
+
],
|
| 5 |
+
"bos_token_id": 1,
|
| 6 |
+
"eos_token_id": 2,
|
| 7 |
+
"hidden_act": "silu",
|
| 8 |
+
"hidden_size": 2048,
|
| 9 |
+
"initializer_range": 0.02,
|
| 10 |
+
"intermediate_size": 5632,
|
| 11 |
+
"max_position_embeddings": 2048,
|
| 12 |
+
"model_type": "llama",
|
| 13 |
+
"num_attention_heads": 32,
|
| 14 |
+
"num_hidden_layers": 22,
|
| 15 |
+
"num_key_value_heads": 4,
|
| 16 |
+
"pretraining_tp": 1,
|
| 17 |
+
"rms_norm_eps": 1e-05,
|
| 18 |
+
"rope_scaling": null,
|
| 19 |
+
"tie_word_embeddings": false,
|
| 20 |
+
"torch_dtype": "float32",
|
| 21 |
+
"transformers_version": "4.31.0.dev0",
|
| 22 |
+
"use_cache": true,
|
| 23 |
+
"vocab_size": 32000
|
| 24 |
+
}
|
out/pretrain/tinyllama_3_epoch/2407/final/generation_config.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token_id": 1,
|
| 3 |
+
"eos_token_id": 2,
|
| 4 |
+
"pad_token_id": 0,
|
| 5 |
+
"max_length": 2048,
|
| 6 |
+
"transformers_version": "4.31.0.dev0"
|
| 7 |
+
}
|
out/pretrain/tinyllama_3_epoch/2407/final/hyperparameters.yaml
ADDED
|
@@ -0,0 +1,44 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
model_name: tiny-llama-1.1b
|
| 2 |
+
out_dir: out/pretrain/tinyllama_3_epoch/2407
|
| 3 |
+
precision: bf16-mixed
|
| 4 |
+
initial_checkpoint_dir: checkpoints/TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T
|
| 5 |
+
resume: false
|
| 6 |
+
data:
|
| 7 |
+
class_path: litgpt.data.Arxiv
|
| 8 |
+
init_args:
|
| 9 |
+
data_path: litgpt/data/arxiv
|
| 10 |
+
seed: 42
|
| 11 |
+
num_workers: 0
|
| 12 |
+
use_starcoder: true
|
| 13 |
+
ppl: false
|
| 14 |
+
data_dir: litgpt/data/arxiv_tinyllama_tokenized/2407
|
| 15 |
+
train:
|
| 16 |
+
save_interval: 100
|
| 17 |
+
log_interval: 1
|
| 18 |
+
global_batch_size: 512
|
| 19 |
+
micro_batch_size: 4
|
| 20 |
+
lr_warmup_steps: 20
|
| 21 |
+
max_tokens: 635437056
|
| 22 |
+
max_seq_length: 2048
|
| 23 |
+
max_norm: 1.0
|
| 24 |
+
min_lr: 4.0e-05
|
| 25 |
+
eval:
|
| 26 |
+
interval: 50
|
| 27 |
+
max_iters: 200
|
| 28 |
+
initial_validation: true
|
| 29 |
+
final_validation: true
|
| 30 |
+
evaluate_example: first
|
| 31 |
+
log: {}
|
| 32 |
+
optimizer:
|
| 33 |
+
class_path: torch.optim.AdamW
|
| 34 |
+
init_args:
|
| 35 |
+
lr: 4.0e-05
|
| 36 |
+
weight_decay: 0.1
|
| 37 |
+
betas:
|
| 38 |
+
- 0.9
|
| 39 |
+
- 0.95
|
| 40 |
+
devices: auto
|
| 41 |
+
num_nodes: 1
|
| 42 |
+
tokenizer_dir: checkpoints/TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T
|
| 43 |
+
logger_name: tensorboard
|
| 44 |
+
seed: 42
|
out/pretrain/tinyllama_3_epoch/2407/final/model_config.yaml
ADDED
|
@@ -0,0 +1,44 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
attention_logit_softcapping: null
|
| 2 |
+
attention_scores_scalar: null
|
| 3 |
+
attn_bias: false
|
| 4 |
+
bias: false
|
| 5 |
+
block_size: 2048
|
| 6 |
+
final_logit_softcapping: null
|
| 7 |
+
gelu_approximate: none
|
| 8 |
+
head_size: 64
|
| 9 |
+
hf_config:
|
| 10 |
+
name: TinyLlama-1.1B-intermediate-step-1431k-3T
|
| 11 |
+
org: TinyLlama
|
| 12 |
+
intermediate_size: 5632
|
| 13 |
+
lm_head_bias: false
|
| 14 |
+
mlp_class_name: LLaMAMLP
|
| 15 |
+
moe_intermediate_size: null
|
| 16 |
+
n_embd: 2048
|
| 17 |
+
n_expert: 0
|
| 18 |
+
n_expert_per_token: 0
|
| 19 |
+
n_head: 32
|
| 20 |
+
n_layer: 22
|
| 21 |
+
n_query_groups: 4
|
| 22 |
+
name: tiny-llama-1.1b
|
| 23 |
+
norm_1: true
|
| 24 |
+
norm_2: true
|
| 25 |
+
norm_class_name: RMSNorm
|
| 26 |
+
norm_eps: 1.0e-05
|
| 27 |
+
norm_qk: false
|
| 28 |
+
norm_qk_type: default
|
| 29 |
+
padded_vocab_size: 32000
|
| 30 |
+
padding_multiple: 64
|
| 31 |
+
parallel_residual: false
|
| 32 |
+
post_attention_norm: false
|
| 33 |
+
post_mlp_norm: false
|
| 34 |
+
rope_adjustments: null
|
| 35 |
+
rope_base: 10000
|
| 36 |
+
rope_condense_ratio: 1
|
| 37 |
+
rope_indices: null
|
| 38 |
+
rope_local_base_freq: null
|
| 39 |
+
rotary_percentage: 1.0
|
| 40 |
+
scale_embeddings: false
|
| 41 |
+
shared_attention_norm: false
|
| 42 |
+
sliding_window_indices: null
|
| 43 |
+
sliding_window_size: null
|
| 44 |
+
vocab_size: 32000
|
out/pretrain/tinyllama_3_epoch/2407/final/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
out/pretrain/tinyllama_3_epoch/2407/final/tokenizer_config.json
ADDED
|
@@ -0,0 +1,35 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_bos_token": true,
|
| 3 |
+
"add_eos_token": false,
|
| 4 |
+
"bos_token": {
|
| 5 |
+
"__type": "AddedToken",
|
| 6 |
+
"content": "<s>",
|
| 7 |
+
"lstrip": false,
|
| 8 |
+
"normalized": false,
|
| 9 |
+
"rstrip": false,
|
| 10 |
+
"single_word": false
|
| 11 |
+
},
|
| 12 |
+
"clean_up_tokenization_spaces": false,
|
| 13 |
+
"eos_token": {
|
| 14 |
+
"__type": "AddedToken",
|
| 15 |
+
"content": "</s>",
|
| 16 |
+
"lstrip": false,
|
| 17 |
+
"normalized": false,
|
| 18 |
+
"rstrip": false,
|
| 19 |
+
"single_word": false
|
| 20 |
+
},
|
| 21 |
+
"legacy": false,
|
| 22 |
+
"model_max_length": 1000000000000000019884624838656,
|
| 23 |
+
"pad_token": null,
|
| 24 |
+
"padding_side": "right",
|
| 25 |
+
"sp_model_kwargs": {},
|
| 26 |
+
"tokenizer_class": "LlamaTokenizer",
|
| 27 |
+
"unk_token": {
|
| 28 |
+
"__type": "AddedToken",
|
| 29 |
+
"content": "<unk>",
|
| 30 |
+
"lstrip": false,
|
| 31 |
+
"normalized": false,
|
| 32 |
+
"rstrip": false,
|
| 33 |
+
"single_word": false
|
| 34 |
+
}
|
| 35 |
+
}
|
out/pretrain/tinyllama_3_epoch/2408/final/config.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"LlamaForCausalLM"
|
| 4 |
+
],
|
| 5 |
+
"bos_token_id": 1,
|
| 6 |
+
"eos_token_id": 2,
|
| 7 |
+
"hidden_act": "silu",
|
| 8 |
+
"hidden_size": 2048,
|
| 9 |
+
"initializer_range": 0.02,
|
| 10 |
+
"intermediate_size": 5632,
|
| 11 |
+
"max_position_embeddings": 2048,
|
| 12 |
+
"model_type": "llama",
|
| 13 |
+
"num_attention_heads": 32,
|
| 14 |
+
"num_hidden_layers": 22,
|
| 15 |
+
"num_key_value_heads": 4,
|
| 16 |
+
"pretraining_tp": 1,
|
| 17 |
+
"rms_norm_eps": 1e-05,
|
| 18 |
+
"rope_scaling": null,
|
| 19 |
+
"tie_word_embeddings": false,
|
| 20 |
+
"torch_dtype": "float32",
|
| 21 |
+
"transformers_version": "4.31.0.dev0",
|
| 22 |
+
"use_cache": true,
|
| 23 |
+
"vocab_size": 32000
|
| 24 |
+
}
|
out/pretrain/tinyllama_3_epoch/2408/final/generation_config.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token_id": 1,
|
| 3 |
+
"eos_token_id": 2,
|
| 4 |
+
"pad_token_id": 0,
|
| 5 |
+
"max_length": 2048,
|
| 6 |
+
"transformers_version": "4.31.0.dev0"
|
| 7 |
+
}
|
out/pretrain/tinyllama_3_epoch/2408/final/hyperparameters.yaml
ADDED
|
@@ -0,0 +1,44 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
model_name: tiny-llama-1.1b
|
| 2 |
+
out_dir: out/pretrain/tinyllama_3_epoch/2408
|
| 3 |
+
precision: bf16-mixed
|
| 4 |
+
initial_checkpoint_dir: out/pretrain/tinyllama_3_epoch/2407/final
|
| 5 |
+
resume: false
|
| 6 |
+
data:
|
| 7 |
+
class_path: litgpt.data.Arxiv
|
| 8 |
+
init_args:
|
| 9 |
+
data_path: litgpt/data/arxiv
|
| 10 |
+
seed: 42
|
| 11 |
+
num_workers: 0
|
| 12 |
+
use_starcoder: true
|
| 13 |
+
ppl: false
|
| 14 |
+
data_dir: litgpt/data/arxiv_tinyllama_tokenized/2408
|
| 15 |
+
train:
|
| 16 |
+
save_interval: 100
|
| 17 |
+
log_interval: 1
|
| 18 |
+
global_batch_size: 512
|
| 19 |
+
micro_batch_size: 4
|
| 20 |
+
lr_warmup_steps: 20
|
| 21 |
+
max_tokens: 531628032
|
| 22 |
+
max_seq_length: 2048
|
| 23 |
+
max_norm: 1.0
|
| 24 |
+
min_lr: 4.0e-05
|
| 25 |
+
eval:
|
| 26 |
+
interval: 50
|
| 27 |
+
max_iters: 200
|
| 28 |
+
initial_validation: true
|
| 29 |
+
final_validation: true
|
| 30 |
+
evaluate_example: first
|
| 31 |
+
log: {}
|
| 32 |
+
optimizer:
|
| 33 |
+
class_path: torch.optim.AdamW
|
| 34 |
+
init_args:
|
| 35 |
+
lr: 4.0e-05
|
| 36 |
+
weight_decay: 0.1
|
| 37 |
+
betas:
|
| 38 |
+
- 0.9
|
| 39 |
+
- 0.95
|
| 40 |
+
devices: auto
|
| 41 |
+
num_nodes: 1
|
| 42 |
+
tokenizer_dir: checkpoints/TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T
|
| 43 |
+
logger_name: tensorboard
|
| 44 |
+
seed: 42
|
out/pretrain/tinyllama_3_epoch/2408/final/model_config.yaml
ADDED
|
@@ -0,0 +1,44 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
attention_logit_softcapping: null
|
| 2 |
+
attention_scores_scalar: null
|
| 3 |
+
attn_bias: false
|
| 4 |
+
bias: false
|
| 5 |
+
block_size: 2048
|
| 6 |
+
final_logit_softcapping: null
|
| 7 |
+
gelu_approximate: none
|
| 8 |
+
head_size: 64
|
| 9 |
+
hf_config:
|
| 10 |
+
name: TinyLlama-1.1B-intermediate-step-1431k-3T
|
| 11 |
+
org: TinyLlama
|
| 12 |
+
intermediate_size: 5632
|
| 13 |
+
lm_head_bias: false
|
| 14 |
+
mlp_class_name: LLaMAMLP
|
| 15 |
+
moe_intermediate_size: null
|
| 16 |
+
n_embd: 2048
|
| 17 |
+
n_expert: 0
|
| 18 |
+
n_expert_per_token: 0
|
| 19 |
+
n_head: 32
|
| 20 |
+
n_layer: 22
|
| 21 |
+
n_query_groups: 4
|
| 22 |
+
name: tiny-llama-1.1b
|
| 23 |
+
norm_1: true
|
| 24 |
+
norm_2: true
|
| 25 |
+
norm_class_name: RMSNorm
|
| 26 |
+
norm_eps: 1.0e-05
|
| 27 |
+
norm_qk: false
|
| 28 |
+
norm_qk_type: default
|
| 29 |
+
padded_vocab_size: 32000
|
| 30 |
+
padding_multiple: 64
|
| 31 |
+
parallel_residual: false
|
| 32 |
+
post_attention_norm: false
|
| 33 |
+
post_mlp_norm: false
|
| 34 |
+
rope_adjustments: null
|
| 35 |
+
rope_base: 10000
|
| 36 |
+
rope_condense_ratio: 1
|
| 37 |
+
rope_indices: null
|
| 38 |
+
rope_local_base_freq: null
|
| 39 |
+
rotary_percentage: 1.0
|
| 40 |
+
scale_embeddings: false
|
| 41 |
+
shared_attention_norm: false
|
| 42 |
+
sliding_window_indices: null
|
| 43 |
+
sliding_window_size: null
|
| 44 |
+
vocab_size: 32000
|
out/pretrain/tinyllama_3_epoch/2408/final/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
out/pretrain/tinyllama_3_epoch/2408/final/tokenizer_config.json
ADDED
|
@@ -0,0 +1,35 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_bos_token": true,
|
| 3 |
+
"add_eos_token": false,
|
| 4 |
+
"bos_token": {
|
| 5 |
+
"__type": "AddedToken",
|
| 6 |
+
"content": "<s>",
|
| 7 |
+
"lstrip": false,
|
| 8 |
+
"normalized": false,
|
| 9 |
+
"rstrip": false,
|
| 10 |
+
"single_word": false
|
| 11 |
+
},
|
| 12 |
+
"clean_up_tokenization_spaces": false,
|
| 13 |
+
"eos_token": {
|
| 14 |
+
"__type": "AddedToken",
|
| 15 |
+
"content": "</s>",
|
| 16 |
+
"lstrip": false,
|
| 17 |
+
"normalized": false,
|
| 18 |
+
"rstrip": false,
|
| 19 |
+
"single_word": false
|
| 20 |
+
},
|
| 21 |
+
"legacy": false,
|
| 22 |
+
"model_max_length": 1000000000000000019884624838656,
|
| 23 |
+
"pad_token": null,
|
| 24 |
+
"padding_side": "right",
|
| 25 |
+
"sp_model_kwargs": {},
|
| 26 |
+
"tokenizer_class": "LlamaTokenizer",
|
| 27 |
+
"unk_token": {
|
| 28 |
+
"__type": "AddedToken",
|
| 29 |
+
"content": "<unk>",
|
| 30 |
+
"lstrip": false,
|
| 31 |
+
"normalized": false,
|
| 32 |
+
"rstrip": false,
|
| 33 |
+
"single_word": false
|
| 34 |
+
}
|
| 35 |
+
}
|
out/pretrain/tinyllama_3_epoch/2409/final/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
out/pretrain/tinyllama_3_epoch/2409/final/tokenizer_config.json
ADDED
|
@@ -0,0 +1,35 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_bos_token": true,
|
| 3 |
+
"add_eos_token": false,
|
| 4 |
+
"bos_token": {
|
| 5 |
+
"__type": "AddedToken",
|
| 6 |
+
"content": "<s>",
|
| 7 |
+
"lstrip": false,
|
| 8 |
+
"normalized": false,
|
| 9 |
+
"rstrip": false,
|
| 10 |
+
"single_word": false
|
| 11 |
+
},
|
| 12 |
+
"clean_up_tokenization_spaces": false,
|
| 13 |
+
"eos_token": {
|
| 14 |
+
"__type": "AddedToken",
|
| 15 |
+
"content": "</s>",
|
| 16 |
+
"lstrip": false,
|
| 17 |
+
"normalized": false,
|
| 18 |
+
"rstrip": false,
|
| 19 |
+
"single_word": false
|
| 20 |
+
},
|
| 21 |
+
"legacy": false,
|
| 22 |
+
"model_max_length": 1000000000000000019884624838656,
|
| 23 |
+
"pad_token": null,
|
| 24 |
+
"padding_side": "right",
|
| 25 |
+
"sp_model_kwargs": {},
|
| 26 |
+
"tokenizer_class": "LlamaTokenizer",
|
| 27 |
+
"unk_token": {
|
| 28 |
+
"__type": "AddedToken",
|
| 29 |
+
"content": "<unk>",
|
| 30 |
+
"lstrip": false,
|
| 31 |
+
"normalized": false,
|
| 32 |
+
"rstrip": false,
|
| 33 |
+
"single_word": false
|
| 34 |
+
}
|
| 35 |
+
}
|
out/pretrain/tinyllama_lr_plus/2501/final/config.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"LlamaForCausalLM"
|
| 4 |
+
],
|
| 5 |
+
"bos_token_id": 1,
|
| 6 |
+
"eos_token_id": 2,
|
| 7 |
+
"hidden_act": "silu",
|
| 8 |
+
"hidden_size": 2048,
|
| 9 |
+
"initializer_range": 0.02,
|
| 10 |
+
"intermediate_size": 5632,
|
| 11 |
+
"max_position_embeddings": 2048,
|
| 12 |
+
"model_type": "llama",
|
| 13 |
+
"num_attention_heads": 32,
|
| 14 |
+
"num_hidden_layers": 22,
|
| 15 |
+
"num_key_value_heads": 4,
|
| 16 |
+
"pretraining_tp": 1,
|
| 17 |
+
"rms_norm_eps": 1e-05,
|
| 18 |
+
"rope_scaling": null,
|
| 19 |
+
"tie_word_embeddings": false,
|
| 20 |
+
"torch_dtype": "float32",
|
| 21 |
+
"transformers_version": "4.31.0.dev0",
|
| 22 |
+
"use_cache": true,
|
| 23 |
+
"vocab_size": 32000
|
| 24 |
+
}
|
out/pretrain/tinyllama_lr_plus/2502/final/config.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"LlamaForCausalLM"
|
| 4 |
+
],
|
| 5 |
+
"bos_token_id": 1,
|
| 6 |
+
"eos_token_id": 2,
|
| 7 |
+
"hidden_act": "silu",
|
| 8 |
+
"hidden_size": 2048,
|
| 9 |
+
"initializer_range": 0.02,
|
| 10 |
+
"intermediate_size": 5632,
|
| 11 |
+
"max_position_embeddings": 2048,
|
| 12 |
+
"model_type": "llama",
|
| 13 |
+
"num_attention_heads": 32,
|
| 14 |
+
"num_hidden_layers": 22,
|
| 15 |
+
"num_key_value_heads": 4,
|
| 16 |
+
"pretraining_tp": 1,
|
| 17 |
+
"rms_norm_eps": 1e-05,
|
| 18 |
+
"rope_scaling": null,
|
| 19 |
+
"tie_word_embeddings": false,
|
| 20 |
+
"torch_dtype": "float32",
|
| 21 |
+
"transformers_version": "4.31.0.dev0",
|
| 22 |
+
"use_cache": true,
|
| 23 |
+
"vocab_size": 32000
|
| 24 |
+
}
|
out/pretrain/tinyllama_lr_plus/2502/final/generation_config.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token_id": 1,
|
| 3 |
+
"eos_token_id": 2,
|
| 4 |
+
"pad_token_id": 0,
|
| 5 |
+
"max_length": 2048,
|
| 6 |
+
"transformers_version": "4.31.0.dev0"
|
| 7 |
+
}
|
out/pretrain/tinyllama_lr_plus/2502/final/hyperparameters.yaml
ADDED
|
@@ -0,0 +1,44 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
model_name: tiny-llama-1.1b
|
| 2 |
+
out_dir: out/pretrain/tinyllama_lr_plus/2502
|
| 3 |
+
precision: bf16-mixed
|
| 4 |
+
initial_checkpoint_dir: out/pretrain/tinyllama_lr_plus/2501/final
|
| 5 |
+
resume: false
|
| 6 |
+
data:
|
| 7 |
+
class_path: litgpt.data.Arxiv
|
| 8 |
+
init_args:
|
| 9 |
+
data_path: litgpt/data/arxiv
|
| 10 |
+
seed: 42
|
| 11 |
+
num_workers: 0
|
| 12 |
+
use_starcoder: true
|
| 13 |
+
ppl: false
|
| 14 |
+
data_dir: litgpt/data/arxiv_tinyllama_tokenized/2502
|
| 15 |
+
train:
|
| 16 |
+
save_interval: 9999
|
| 17 |
+
log_interval: 1
|
| 18 |
+
global_batch_size: 512
|
| 19 |
+
micro_batch_size: 4
|
| 20 |
+
lr_warmup_steps: 20
|
| 21 |
+
max_tokens: 330301440
|
| 22 |
+
max_seq_length: 2048
|
| 23 |
+
max_norm: 1.0
|
| 24 |
+
min_lr: 0.0008
|
| 25 |
+
eval:
|
| 26 |
+
interval: 50
|
| 27 |
+
max_iters: 200
|
| 28 |
+
initial_validation: true
|
| 29 |
+
final_validation: true
|
| 30 |
+
evaluate_example: first
|
| 31 |
+
log: {}
|
| 32 |
+
optimizer:
|
| 33 |
+
class_path: torch.optim.AdamW
|
| 34 |
+
init_args:
|
| 35 |
+
lr: 0.0008
|
| 36 |
+
weight_decay: 0.1
|
| 37 |
+
betas:
|
| 38 |
+
- 0.9
|
| 39 |
+
- 0.95
|
| 40 |
+
devices: auto
|
| 41 |
+
num_nodes: 1
|
| 42 |
+
tokenizer_dir: checkpoints/TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T
|
| 43 |
+
logger_name: tensorboard
|
| 44 |
+
seed: 42
|
out/pretrain/tinyllama_lr_plus/2502/final/model_config.yaml
ADDED
|
@@ -0,0 +1,44 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
attention_logit_softcapping: null
|
| 2 |
+
attention_scores_scalar: null
|
| 3 |
+
attn_bias: false
|
| 4 |
+
bias: false
|
| 5 |
+
block_size: 2048
|
| 6 |
+
final_logit_softcapping: null
|
| 7 |
+
gelu_approximate: none
|
| 8 |
+
head_size: 64
|
| 9 |
+
hf_config:
|
| 10 |
+
name: TinyLlama-1.1B-intermediate-step-1431k-3T
|
| 11 |
+
org: TinyLlama
|
| 12 |
+
intermediate_size: 5632
|
| 13 |
+
lm_head_bias: false
|
| 14 |
+
mlp_class_name: LLaMAMLP
|
| 15 |
+
moe_intermediate_size: null
|
| 16 |
+
n_embd: 2048
|
| 17 |
+
n_expert: 0
|
| 18 |
+
n_expert_per_token: 0
|
| 19 |
+
n_head: 32
|
| 20 |
+
n_layer: 22
|
| 21 |
+
n_query_groups: 4
|
| 22 |
+
name: tiny-llama-1.1b
|
| 23 |
+
norm_1: true
|
| 24 |
+
norm_2: true
|
| 25 |
+
norm_class_name: RMSNorm
|
| 26 |
+
norm_eps: 1.0e-05
|
| 27 |
+
norm_qk: false
|
| 28 |
+
norm_qk_type: default
|
| 29 |
+
padded_vocab_size: 32000
|
| 30 |
+
padding_multiple: 64
|
| 31 |
+
parallel_residual: false
|
| 32 |
+
post_attention_norm: false
|
| 33 |
+
post_mlp_norm: false
|
| 34 |
+
rope_adjustments: null
|
| 35 |
+
rope_base: 10000
|
| 36 |
+
rope_condense_ratio: 1
|
| 37 |
+
rope_indices: null
|
| 38 |
+
rope_local_base_freq: null
|
| 39 |
+
rotary_percentage: 1.0
|
| 40 |
+
scale_embeddings: false
|
| 41 |
+
shared_attention_norm: false
|
| 42 |
+
sliding_window_indices: null
|
| 43 |
+
sliding_window_size: null
|
| 44 |
+
vocab_size: 32000
|
out/pretrain/tinyllama_lr_plus/2502/final/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
out/pretrain/tinyllama_lr_plus/2502/final/tokenizer_config.json
ADDED
|
@@ -0,0 +1,35 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_bos_token": true,
|
| 3 |
+
"add_eos_token": false,
|
| 4 |
+
"bos_token": {
|
| 5 |
+
"__type": "AddedToken",
|
| 6 |
+
"content": "<s>",
|
| 7 |
+
"lstrip": false,
|
| 8 |
+
"normalized": false,
|
| 9 |
+
"rstrip": false,
|
| 10 |
+
"single_word": false
|
| 11 |
+
},
|
| 12 |
+
"clean_up_tokenization_spaces": false,
|
| 13 |
+
"eos_token": {
|
| 14 |
+
"__type": "AddedToken",
|
| 15 |
+
"content": "</s>",
|
| 16 |
+
"lstrip": false,
|
| 17 |
+
"normalized": false,
|
| 18 |
+
"rstrip": false,
|
| 19 |
+
"single_word": false
|
| 20 |
+
},
|
| 21 |
+
"legacy": false,
|
| 22 |
+
"model_max_length": 1000000000000000019884624838656,
|
| 23 |
+
"pad_token": null,
|
| 24 |
+
"padding_side": "right",
|
| 25 |
+
"sp_model_kwargs": {},
|
| 26 |
+
"tokenizer_class": "LlamaTokenizer",
|
| 27 |
+
"unk_token": {
|
| 28 |
+
"__type": "AddedToken",
|
| 29 |
+
"content": "<unk>",
|
| 30 |
+
"lstrip": false,
|
| 31 |
+
"normalized": false,
|
| 32 |
+
"rstrip": false,
|
| 33 |
+
"single_word": false
|
| 34 |
+
}
|
| 35 |
+
}
|
out/pretrain/tinyllama_lr_plus/2503/final/config.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"LlamaForCausalLM"
|
| 4 |
+
],
|
| 5 |
+
"bos_token_id": 1,
|
| 6 |
+
"eos_token_id": 2,
|
| 7 |
+
"hidden_act": "silu",
|
| 8 |
+
"hidden_size": 2048,
|
| 9 |
+
"initializer_range": 0.02,
|
| 10 |
+
"intermediate_size": 5632,
|
| 11 |
+
"max_position_embeddings": 2048,
|
| 12 |
+
"model_type": "llama",
|
| 13 |
+
"num_attention_heads": 32,
|
| 14 |
+
"num_hidden_layers": 22,
|
| 15 |
+
"num_key_value_heads": 4,
|
| 16 |
+
"pretraining_tp": 1,
|
| 17 |
+
"rms_norm_eps": 1e-05,
|
| 18 |
+
"rope_scaling": null,
|
| 19 |
+
"tie_word_embeddings": false,
|
| 20 |
+
"torch_dtype": "float32",
|
| 21 |
+
"transformers_version": "4.31.0.dev0",
|
| 22 |
+
"use_cache": true,
|
| 23 |
+
"vocab_size": 32000
|
| 24 |
+
}
|
out/pretrain/tinyllama_lr_plus/2503/final/generation_config.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token_id": 1,
|
| 3 |
+
"eos_token_id": 2,
|
| 4 |
+
"pad_token_id": 0,
|
| 5 |
+
"max_length": 2048,
|
| 6 |
+
"transformers_version": "4.31.0.dev0"
|
| 7 |
+
}
|
out/pretrain/tinyllama_lr_plus/2503/final/hyperparameters.yaml
ADDED
|
@@ -0,0 +1,44 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
model_name: tiny-llama-1.1b
|
| 2 |
+
out_dir: out/pretrain/tinyllama_lr_plus/2503
|
| 3 |
+
precision: bf16-mixed
|
| 4 |
+
initial_checkpoint_dir: out/pretrain/tinyllama_lr_plus/2502/final
|
| 5 |
+
resume: false
|
| 6 |
+
data:
|
| 7 |
+
class_path: litgpt.data.Arxiv
|
| 8 |
+
init_args:
|
| 9 |
+
data_path: litgpt/data/arxiv
|
| 10 |
+
seed: 42
|
| 11 |
+
num_workers: 0
|
| 12 |
+
use_starcoder: true
|
| 13 |
+
ppl: false
|
| 14 |
+
data_dir: litgpt/data/arxiv_tinyllama_tokenized/2503
|
| 15 |
+
train:
|
| 16 |
+
save_interval: 9999
|
| 17 |
+
log_interval: 1
|
| 18 |
+
global_batch_size: 512
|
| 19 |
+
micro_batch_size: 4
|
| 20 |
+
lr_warmup_steps: 20
|
| 21 |
+
max_tokens: 394264576
|
| 22 |
+
max_seq_length: 2048
|
| 23 |
+
max_norm: 1.0
|
| 24 |
+
min_lr: 0.0008
|
| 25 |
+
eval:
|
| 26 |
+
interval: 50
|
| 27 |
+
max_iters: 200
|
| 28 |
+
initial_validation: true
|
| 29 |
+
final_validation: true
|
| 30 |
+
evaluate_example: first
|
| 31 |
+
log: {}
|
| 32 |
+
optimizer:
|
| 33 |
+
class_path: torch.optim.AdamW
|
| 34 |
+
init_args:
|
| 35 |
+
lr: 0.0008
|
| 36 |
+
weight_decay: 0.1
|
| 37 |
+
betas:
|
| 38 |
+
- 0.9
|
| 39 |
+
- 0.95
|
| 40 |
+
devices: auto
|
| 41 |
+
num_nodes: 1
|
| 42 |
+
tokenizer_dir: checkpoints/TinyLlama/TinyLlama-1.1B-intermediate-step-1431k-3T
|
| 43 |
+
logger_name: tensorboard
|
| 44 |
+
seed: 42
|
out/pretrain/tinyllama_lr_plus/2503/final/model_config.yaml
ADDED
|
@@ -0,0 +1,44 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
attention_logit_softcapping: null
|
| 2 |
+
attention_scores_scalar: null
|
| 3 |
+
attn_bias: false
|
| 4 |
+
bias: false
|
| 5 |
+
block_size: 2048
|
| 6 |
+
final_logit_softcapping: null
|
| 7 |
+
gelu_approximate: none
|
| 8 |
+
head_size: 64
|
| 9 |
+
hf_config:
|
| 10 |
+
name: TinyLlama-1.1B-intermediate-step-1431k-3T
|
| 11 |
+
org: TinyLlama
|
| 12 |
+
intermediate_size: 5632
|
| 13 |
+
lm_head_bias: false
|
| 14 |
+
mlp_class_name: LLaMAMLP
|
| 15 |
+
moe_intermediate_size: null
|
| 16 |
+
n_embd: 2048
|
| 17 |
+
n_expert: 0
|
| 18 |
+
n_expert_per_token: 0
|
| 19 |
+
n_head: 32
|
| 20 |
+
n_layer: 22
|
| 21 |
+
n_query_groups: 4
|
| 22 |
+
name: tiny-llama-1.1b
|
| 23 |
+
norm_1: true
|
| 24 |
+
norm_2: true
|
| 25 |
+
norm_class_name: RMSNorm
|
| 26 |
+
norm_eps: 1.0e-05
|
| 27 |
+
norm_qk: false
|
| 28 |
+
norm_qk_type: default
|
| 29 |
+
padded_vocab_size: 32000
|
| 30 |
+
padding_multiple: 64
|
| 31 |
+
parallel_residual: false
|
| 32 |
+
post_attention_norm: false
|
| 33 |
+
post_mlp_norm: false
|
| 34 |
+
rope_adjustments: null
|
| 35 |
+
rope_base: 10000
|
| 36 |
+
rope_condense_ratio: 1
|
| 37 |
+
rope_indices: null
|
| 38 |
+
rope_local_base_freq: null
|
| 39 |
+
rotary_percentage: 1.0
|
| 40 |
+
scale_embeddings: false
|
| 41 |
+
shared_attention_norm: false
|
| 42 |
+
sliding_window_indices: null
|
| 43 |
+
sliding_window_size: null
|
| 44 |
+
vocab_size: 32000
|
out/pretrain/tinyllama_lr_plus/2503/final/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
out/pretrain/tinyllama_lr_plus/2503/final/tokenizer_config.json
ADDED
|
@@ -0,0 +1,35 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_bos_token": true,
|
| 3 |
+
"add_eos_token": false,
|
| 4 |
+
"bos_token": {
|
| 5 |
+
"__type": "AddedToken",
|
| 6 |
+
"content": "<s>",
|
| 7 |
+
"lstrip": false,
|
| 8 |
+
"normalized": false,
|
| 9 |
+
"rstrip": false,
|
| 10 |
+
"single_word": false
|
| 11 |
+
},
|
| 12 |
+
"clean_up_tokenization_spaces": false,
|
| 13 |
+
"eos_token": {
|
| 14 |
+
"__type": "AddedToken",
|
| 15 |
+
"content": "</s>",
|
| 16 |
+
"lstrip": false,
|
| 17 |
+
"normalized": false,
|
| 18 |
+
"rstrip": false,
|
| 19 |
+
"single_word": false
|
| 20 |
+
},
|
| 21 |
+
"legacy": false,
|
| 22 |
+
"model_max_length": 1000000000000000019884624838656,
|
| 23 |
+
"pad_token": null,
|
| 24 |
+
"padding_side": "right",
|
| 25 |
+
"sp_model_kwargs": {},
|
| 26 |
+
"tokenizer_class": "LlamaTokenizer",
|
| 27 |
+
"unk_token": {
|
| 28 |
+
"__type": "AddedToken",
|
| 29 |
+
"content": "<unk>",
|
| 30 |
+
"lstrip": false,
|
| 31 |
+
"normalized": false,
|
| 32 |
+
"rstrip": false,
|
| 33 |
+
"single_word": false
|
| 34 |
+
}
|
| 35 |
+
}
|
out/pretrain/tinyllama_lr_plus/2504/final/config.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"LlamaForCausalLM"
|
| 4 |
+
],
|
| 5 |
+
"bos_token_id": 1,
|
| 6 |
+
"eos_token_id": 2,
|
| 7 |
+
"hidden_act": "silu",
|
| 8 |
+
"hidden_size": 2048,
|
| 9 |
+
"initializer_range": 0.02,
|
| 10 |
+
"intermediate_size": 5632,
|
| 11 |
+
"max_position_embeddings": 2048,
|
| 12 |
+
"model_type": "llama",
|
| 13 |
+
"num_attention_heads": 32,
|
| 14 |
+
"num_hidden_layers": 22,
|
| 15 |
+
"num_key_value_heads": 4,
|
| 16 |
+
"pretraining_tp": 1,
|
| 17 |
+
"rms_norm_eps": 1e-05,
|
| 18 |
+
"rope_scaling": null,
|
| 19 |
+
"tie_word_embeddings": false,
|
| 20 |
+
"torch_dtype": "float32",
|
| 21 |
+
"transformers_version": "4.31.0.dev0",
|
| 22 |
+
"use_cache": true,
|
| 23 |
+
"vocab_size": 32000
|
| 24 |
+
}
|
out/pretrain/tinyllama_lr_plus/2504/final/generation_config.json
ADDED
|
@@ -0,0 +1,7 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token_id": 1,
|
| 3 |
+
"eos_token_id": 2,
|
| 4 |
+
"pad_token_id": 0,
|
| 5 |
+
"max_length": 2048,
|
| 6 |
+
"transformers_version": "4.31.0.dev0"
|
| 7 |
+
}
|