Upload folder using huggingface_hub
Browse files- chat_template.jinja +2 -0
- config.json +164 -0
- generation_config.json +10 -0
- model.safetensors +3 -0
- optimizer.pt +3 -0
- processor_config.json +79 -0
- rng_state.pth +3 -0
- scheduler.pt +3 -0
- tokenizer.json +0 -0
- tokenizer_config.json +31 -0
- trainer_state.json +984 -0
- training_args.bin +3 -0
chat_template.jinja
ADDED
|
@@ -0,0 +1,2 @@
|
|
|
|
|
|
|
|
|
|
| 1 |
+
<|im_start|>{% for message in messages %}{{message['role'] | capitalize}}{% if message['content'][0]['type'] == 'image' %}{{':'}}{% else %}{{': '}}{% endif %}{% for line in message['content'] %}{% if line['type'] == 'text' %}{{line['text']}}{% elif line['type'] == 'image' %}{{ '<image>' }}{% endif %}{% endfor %}<end_of_utterance>
|
| 2 |
+
{% endfor %}{% if add_generation_prompt %}{{ 'Assistant:' }}{% endif %}
|
config.json
ADDED
|
@@ -0,0 +1,164 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"architectures": [
|
| 3 |
+
"SmolVLMForConditionalGeneration"
|
| 4 |
+
],
|
| 5 |
+
"bos_token_id": 1,
|
| 6 |
+
"dtype": "bfloat16",
|
| 7 |
+
"eos_token_id": 49279,
|
| 8 |
+
"image_token_id": 49190,
|
| 9 |
+
"model_type": "smolvlm",
|
| 10 |
+
"pad_token_id": 2,
|
| 11 |
+
"scale_factor": 4,
|
| 12 |
+
"text_config": {
|
| 13 |
+
"_flash_attn_2_enabled": true,
|
| 14 |
+
"_name_or_path": "None",
|
| 15 |
+
"architectures": [
|
| 16 |
+
"VLlama3ForCausalLM"
|
| 17 |
+
],
|
| 18 |
+
"attention_bias": false,
|
| 19 |
+
"attention_dropout": 0.0,
|
| 20 |
+
"bos_token_id": 1,
|
| 21 |
+
"dtype": "bfloat16",
|
| 22 |
+
"eos_token_id": 2,
|
| 23 |
+
"head_dim": 64,
|
| 24 |
+
"hidden_act": "silu",
|
| 25 |
+
"hidden_size": 960,
|
| 26 |
+
"initializer_range": 0.02,
|
| 27 |
+
"intermediate_size": 2560,
|
| 28 |
+
"is_llama_config": true,
|
| 29 |
+
"max_position_embeddings": 8192,
|
| 30 |
+
"mlp_bias": false,
|
| 31 |
+
"model_type": "llama",
|
| 32 |
+
"neftune_noise_alpha": 0.0,
|
| 33 |
+
"num_attention_heads": 15,
|
| 34 |
+
"num_hidden_layers": 32,
|
| 35 |
+
"num_key_value_heads": 5,
|
| 36 |
+
"pad_token_id": 2,
|
| 37 |
+
"perceiver_config": {
|
| 38 |
+
"_attn_implementation_autoset": false,
|
| 39 |
+
"_name_or_path": "",
|
| 40 |
+
"add_cross_attention": false,
|
| 41 |
+
"architectures": null,
|
| 42 |
+
"attention_dropout": 0.0,
|
| 43 |
+
"bad_words_ids": null,
|
| 44 |
+
"begin_suppress_tokens": null,
|
| 45 |
+
"bos_token_id": null,
|
| 46 |
+
"chunk_size_feed_forward": 0,
|
| 47 |
+
"cross_attention_hidden_size": null,
|
| 48 |
+
"decoder_start_token_id": null,
|
| 49 |
+
"diversity_penalty": 0.0,
|
| 50 |
+
"do_sample": false,
|
| 51 |
+
"early_stopping": false,
|
| 52 |
+
"encoder_no_repeat_ngram_size": 0,
|
| 53 |
+
"eos_token_id": null,
|
| 54 |
+
"exponential_decay_length_penalty": null,
|
| 55 |
+
"finetuning_task": null,
|
| 56 |
+
"forced_bos_token_id": null,
|
| 57 |
+
"forced_eos_token_id": null,
|
| 58 |
+
"hidden_act": "silu",
|
| 59 |
+
"id2label": {
|
| 60 |
+
"0": "LABEL_0",
|
| 61 |
+
"1": "LABEL_1"
|
| 62 |
+
},
|
| 63 |
+
"is_decoder": false,
|
| 64 |
+
"is_encoder_decoder": false,
|
| 65 |
+
"label2id": {
|
| 66 |
+
"LABEL_0": 0,
|
| 67 |
+
"LABEL_1": 1
|
| 68 |
+
},
|
| 69 |
+
"length_penalty": 1.0,
|
| 70 |
+
"max_length": 20,
|
| 71 |
+
"min_length": 0,
|
| 72 |
+
"model_type": "vllama3",
|
| 73 |
+
"no_repeat_ngram_size": 0,
|
| 74 |
+
"num_beam_groups": 1,
|
| 75 |
+
"num_beams": 1,
|
| 76 |
+
"num_key_value_heads": 1,
|
| 77 |
+
"num_return_sequences": 1,
|
| 78 |
+
"output_attentions": false,
|
| 79 |
+
"output_hidden_states": false,
|
| 80 |
+
"output_scores": false,
|
| 81 |
+
"pad_token_id": null,
|
| 82 |
+
"prefix": null,
|
| 83 |
+
"problem_type": null,
|
| 84 |
+
"pruned_heads": {},
|
| 85 |
+
"qk_layer_norms_perceiver": false,
|
| 86 |
+
"remove_invalid_values": false,
|
| 87 |
+
"repetition_penalty": 1.0,
|
| 88 |
+
"resampler_depth": 6,
|
| 89 |
+
"resampler_head_dim": 96,
|
| 90 |
+
"resampler_n_heads": 16,
|
| 91 |
+
"resampler_n_latents": 64,
|
| 92 |
+
"return_dict": true,
|
| 93 |
+
"return_dict_in_generate": false,
|
| 94 |
+
"sep_token_id": null,
|
| 95 |
+
"suppress_tokens": null,
|
| 96 |
+
"task_specific_params": null,
|
| 97 |
+
"temperature": 1.0,
|
| 98 |
+
"tf_legacy_loss": false,
|
| 99 |
+
"tie_encoder_decoder": false,
|
| 100 |
+
"tie_word_embeddings": true,
|
| 101 |
+
"tokenizer_class": null,
|
| 102 |
+
"top_k": 50,
|
| 103 |
+
"top_p": 1.0,
|
| 104 |
+
"torch_dtype": null,
|
| 105 |
+
"torchscript": false,
|
| 106 |
+
"transformers_version": "4.46.0",
|
| 107 |
+
"typical_p": 1.0,
|
| 108 |
+
"use_bfloat16": false
|
| 109 |
+
},
|
| 110 |
+
"pixel_shuffle_factor": 4,
|
| 111 |
+
"pretraining_tp": 1,
|
| 112 |
+
"qk_layer_norms": false,
|
| 113 |
+
"rms_norm_eps": 1e-05,
|
| 114 |
+
"rope_interleaved": false,
|
| 115 |
+
"rope_parameters": {
|
| 116 |
+
"rope_theta": 100000,
|
| 117 |
+
"rope_type": "default"
|
| 118 |
+
},
|
| 119 |
+
"tie_word_embeddings": false,
|
| 120 |
+
"transformers.js_config": {
|
| 121 |
+
"kv_cache_dtype": {
|
| 122 |
+
"fp16": "float16",
|
| 123 |
+
"q4f16": "float16"
|
| 124 |
+
}
|
| 125 |
+
},
|
| 126 |
+
"use_cache": true,
|
| 127 |
+
"use_resampler": false,
|
| 128 |
+
"vocab_size": 49280
|
| 129 |
+
},
|
| 130 |
+
"tie_word_embeddings": false,
|
| 131 |
+
"transformers.js_config": {
|
| 132 |
+
"kv_cache_dtype": {
|
| 133 |
+
"fp16": "float16",
|
| 134 |
+
"q4f16": "float16"
|
| 135 |
+
}
|
| 136 |
+
},
|
| 137 |
+
"transformers_version": "5.0.0",
|
| 138 |
+
"use_cache": false,
|
| 139 |
+
"use_reentrant_checkpointing": false,
|
| 140 |
+
"vision_config": {
|
| 141 |
+
"attention_dropout": 0.0,
|
| 142 |
+
"dtype": "bfloat16",
|
| 143 |
+
"hidden_act": "gelu_pytorch_tanh",
|
| 144 |
+
"hidden_size": 768,
|
| 145 |
+
"image_size": 512,
|
| 146 |
+
"initializer_range": 0.02,
|
| 147 |
+
"intermediate_size": 3072,
|
| 148 |
+
"layer_norm_eps": 1e-06,
|
| 149 |
+
"max_image_size": {
|
| 150 |
+
"longest_edge": 512
|
| 151 |
+
},
|
| 152 |
+
"model_type": "smolvlm_vision",
|
| 153 |
+
"num_attention_heads": 12,
|
| 154 |
+
"num_channels": 3,
|
| 155 |
+
"num_hidden_layers": 12,
|
| 156 |
+
"patch_size": 16,
|
| 157 |
+
"size": {
|
| 158 |
+
"longest_edge": 2048
|
| 159 |
+
},
|
| 160 |
+
"tie_word_embeddings": false,
|
| 161 |
+
"use_base_siglip": false
|
| 162 |
+
},
|
| 163 |
+
"vocab_size": 49280
|
| 164 |
+
}
|
generation_config.json
ADDED
|
@@ -0,0 +1,10 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"_from_model_config": true,
|
| 3 |
+
"bos_token_id": 1,
|
| 4 |
+
"eos_token_id": [
|
| 5 |
+
49279,
|
| 6 |
+
49279
|
| 7 |
+
],
|
| 8 |
+
"pad_token_id": 2,
|
| 9 |
+
"transformers_version": "5.0.0"
|
| 10 |
+
}
|
model.safetensors
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:12c8c5e8d4cf1b4a0a834bcb539e3d8389e63a6306a0ce97d3d67be2ec4c4d48
|
| 3 |
+
size 1015025832
|
optimizer.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a34a45c16f8d91835b8f6085165a440fe23896373ed2c70df14ddd4964d761dd
|
| 3 |
+
size 1637261579
|
processor_config.json
ADDED
|
@@ -0,0 +1,79 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"image_processor": {
|
| 3 |
+
"do_convert_rgb": true,
|
| 4 |
+
"do_image_splitting": true,
|
| 5 |
+
"do_normalize": true,
|
| 6 |
+
"do_pad": true,
|
| 7 |
+
"do_rescale": true,
|
| 8 |
+
"do_resize": true,
|
| 9 |
+
"image_mean": [
|
| 10 |
+
0.5,
|
| 11 |
+
0.5,
|
| 12 |
+
0.5
|
| 13 |
+
],
|
| 14 |
+
"image_processor_type": "SmolVLMImageProcessor",
|
| 15 |
+
"image_std": [
|
| 16 |
+
0.5,
|
| 17 |
+
0.5,
|
| 18 |
+
0.5
|
| 19 |
+
],
|
| 20 |
+
"max_image_size": {
|
| 21 |
+
"longest_edge": 512
|
| 22 |
+
},
|
| 23 |
+
"resample": 1,
|
| 24 |
+
"rescale_factor": 0.00392156862745098,
|
| 25 |
+
"size": {
|
| 26 |
+
"longest_edge": 2048
|
| 27 |
+
},
|
| 28 |
+
"video_sampling": {
|
| 29 |
+
"fps": 1,
|
| 30 |
+
"max_frames": 64,
|
| 31 |
+
"video_size": {
|
| 32 |
+
"longest_edge": 512
|
| 33 |
+
}
|
| 34 |
+
}
|
| 35 |
+
},
|
| 36 |
+
"image_seq_len": 64,
|
| 37 |
+
"processor_class": "SmolVLMProcessor",
|
| 38 |
+
"video_processor": {
|
| 39 |
+
"data_format": "channels_first",
|
| 40 |
+
"default_to_square": true,
|
| 41 |
+
"do_convert_rgb": true,
|
| 42 |
+
"do_image_splitting": true,
|
| 43 |
+
"do_normalize": true,
|
| 44 |
+
"do_pad": true,
|
| 45 |
+
"do_rescale": true,
|
| 46 |
+
"do_resize": true,
|
| 47 |
+
"do_sample_frames": false,
|
| 48 |
+
"fps": 1,
|
| 49 |
+
"image_mean": [
|
| 50 |
+
0.5,
|
| 51 |
+
0.5,
|
| 52 |
+
0.5
|
| 53 |
+
],
|
| 54 |
+
"image_processor_type": "SmolVLMImageProcessor",
|
| 55 |
+
"image_std": [
|
| 56 |
+
0.5,
|
| 57 |
+
0.5,
|
| 58 |
+
0.5
|
| 59 |
+
],
|
| 60 |
+
"max_image_size": {
|
| 61 |
+
"longest_edge": 512
|
| 62 |
+
},
|
| 63 |
+
"num_frames": 64,
|
| 64 |
+
"resample": 1,
|
| 65 |
+
"rescale_factor": 0.00392156862745098,
|
| 66 |
+
"return_metadata": false,
|
| 67 |
+
"size": {
|
| 68 |
+
"longest_edge": 2048
|
| 69 |
+
},
|
| 70 |
+
"video_processor_type": "SmolVLMVideoProcessor",
|
| 71 |
+
"video_sampling": {
|
| 72 |
+
"fps": 1,
|
| 73 |
+
"max_frames": 64,
|
| 74 |
+
"video_size": {
|
| 75 |
+
"longest_edge": 2048
|
| 76 |
+
}
|
| 77 |
+
}
|
| 78 |
+
}
|
| 79 |
+
}
|
rng_state.pth
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:098b29492211804ab324a36f37466821d948280bb74fce4ba895c03f13ecd878
|
| 3 |
+
size 14645
|
scheduler.pt
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f3d2378bc5b746c762327d4822ca1447f1f50eb97a95a925d360512fec0cf2bc
|
| 3 |
+
size 1465
|
tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
tokenizer_config.json
ADDED
|
@@ -0,0 +1,31 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_prefix_space": false,
|
| 3 |
+
"backend": "tokenizers",
|
| 4 |
+
"bos_token": "<|im_start|>",
|
| 5 |
+
"clean_up_tokenization_spaces": false,
|
| 6 |
+
"end_of_utterance_token": "<end_of_utterance>",
|
| 7 |
+
"eos_token": "<end_of_utterance>",
|
| 8 |
+
"extra_special_tokens": [
|
| 9 |
+
"<fake_token_around_image>",
|
| 10 |
+
"<image>",
|
| 11 |
+
"<end_of_utterance>"
|
| 12 |
+
],
|
| 13 |
+
"fake_image_token": "<fake_token_around_image>",
|
| 14 |
+
"global_image_token": "<global-img>",
|
| 15 |
+
"image_token": "<image>",
|
| 16 |
+
"is_local": false,
|
| 17 |
+
"legacy": false,
|
| 18 |
+
"model_max_length": 8192,
|
| 19 |
+
"model_specific_special_tokens": {
|
| 20 |
+
"end_of_utterance_token": "<end_of_utterance>",
|
| 21 |
+
"fake_image_token": "<fake_token_around_image>",
|
| 22 |
+
"global_image_token": "<global-img>",
|
| 23 |
+
"image_token": "<image>"
|
| 24 |
+
},
|
| 25 |
+
"pad_token": "<|im_end|>",
|
| 26 |
+
"processor_class": "SmolVLMProcessor",
|
| 27 |
+
"tokenizer_class": "TokenizersBackend",
|
| 28 |
+
"truncation_side": "left",
|
| 29 |
+
"unk_token": "<|endoftext|>",
|
| 30 |
+
"vocab_size": 49152
|
| 31 |
+
}
|
trainer_state.json
ADDED
|
@@ -0,0 +1,984 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 2.0,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 950,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"entropy": 1.79375,
|
| 14 |
+
"epoch": 0.021052631578947368,
|
| 15 |
+
"grad_norm": 6.71875,
|
| 16 |
+
"learning_rate": 4.9526315789473685e-05,
|
| 17 |
+
"loss": 1.7312183380126953,
|
| 18 |
+
"mean_token_accuracy": 0.6867853045463562,
|
| 19 |
+
"num_tokens": 840.0,
|
| 20 |
+
"step": 10
|
| 21 |
+
},
|
| 22 |
+
{
|
| 23 |
+
"entropy": 1.52421875,
|
| 24 |
+
"epoch": 0.042105263157894736,
|
| 25 |
+
"grad_norm": 3.90625,
|
| 26 |
+
"learning_rate": 4.9e-05,
|
| 27 |
+
"loss": 1.3260906219482422,
|
| 28 |
+
"mean_token_accuracy": 0.7451180815696716,
|
| 29 |
+
"num_tokens": 1573.0,
|
| 30 |
+
"step": 20
|
| 31 |
+
},
|
| 32 |
+
{
|
| 33 |
+
"entropy": 1.69921875,
|
| 34 |
+
"epoch": 0.06315789473684211,
|
| 35 |
+
"grad_norm": 6.5625,
|
| 36 |
+
"learning_rate": 4.847368421052632e-05,
|
| 37 |
+
"loss": 1.6661336898803711,
|
| 38 |
+
"mean_token_accuracy": 0.7098163902759552,
|
| 39 |
+
"num_tokens": 2338.0,
|
| 40 |
+
"step": 30
|
| 41 |
+
},
|
| 42 |
+
{
|
| 43 |
+
"entropy": 1.5484375,
|
| 44 |
+
"epoch": 0.08421052631578947,
|
| 45 |
+
"grad_norm": 7.4375,
|
| 46 |
+
"learning_rate": 4.794736842105264e-05,
|
| 47 |
+
"loss": 1.5569435119628907,
|
| 48 |
+
"mean_token_accuracy": 0.7198988318443298,
|
| 49 |
+
"num_tokens": 3038.0,
|
| 50 |
+
"step": 40
|
| 51 |
+
},
|
| 52 |
+
{
|
| 53 |
+
"entropy": 1.59296875,
|
| 54 |
+
"epoch": 0.10526315789473684,
|
| 55 |
+
"grad_norm": 8.1875,
|
| 56 |
+
"learning_rate": 4.742105263157895e-05,
|
| 57 |
+
"loss": 1.5420040130615233,
|
| 58 |
+
"mean_token_accuracy": 0.7149245262145996,
|
| 59 |
+
"num_tokens": 3682.0,
|
| 60 |
+
"step": 50
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"entropy": 1.80390625,
|
| 64 |
+
"epoch": 0.12631578947368421,
|
| 65 |
+
"grad_norm": 9.9375,
|
| 66 |
+
"learning_rate": 4.6894736842105264e-05,
|
| 67 |
+
"loss": 1.6832901000976563,
|
| 68 |
+
"mean_token_accuracy": 0.7065998375415802,
|
| 69 |
+
"num_tokens": 4617.0,
|
| 70 |
+
"step": 60
|
| 71 |
+
},
|
| 72 |
+
{
|
| 73 |
+
"entropy": 1.66796875,
|
| 74 |
+
"epoch": 0.14736842105263157,
|
| 75 |
+
"grad_norm": 8.375,
|
| 76 |
+
"learning_rate": 4.6368421052631584e-05,
|
| 77 |
+
"loss": 1.6378042221069335,
|
| 78 |
+
"mean_token_accuracy": 0.7060423612594604,
|
| 79 |
+
"num_tokens": 5349.0,
|
| 80 |
+
"step": 70
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"entropy": 1.5109375,
|
| 84 |
+
"epoch": 0.16842105263157894,
|
| 85 |
+
"grad_norm": 6.03125,
|
| 86 |
+
"learning_rate": 4.58421052631579e-05,
|
| 87 |
+
"loss": 1.5179196357727052,
|
| 88 |
+
"mean_token_accuracy": 0.7291716754436492,
|
| 89 |
+
"num_tokens": 6122.0,
|
| 90 |
+
"step": 80
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"entropy": 1.721875,
|
| 94 |
+
"epoch": 0.18947368421052632,
|
| 95 |
+
"grad_norm": 4.0625,
|
| 96 |
+
"learning_rate": 4.531578947368421e-05,
|
| 97 |
+
"loss": 1.8176845550537108,
|
| 98 |
+
"mean_token_accuracy": 0.6842875778675079,
|
| 99 |
+
"num_tokens": 7197.0,
|
| 100 |
+
"step": 90
|
| 101 |
+
},
|
| 102 |
+
{
|
| 103 |
+
"entropy": 1.52265625,
|
| 104 |
+
"epoch": 0.21052631578947367,
|
| 105 |
+
"grad_norm": 6.46875,
|
| 106 |
+
"learning_rate": 4.478947368421053e-05,
|
| 107 |
+
"loss": 1.5252375602722168,
|
| 108 |
+
"mean_token_accuracy": 0.711300152540207,
|
| 109 |
+
"num_tokens": 8046.0,
|
| 110 |
+
"step": 100
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"entropy": 1.6515625,
|
| 114 |
+
"epoch": 0.23157894736842105,
|
| 115 |
+
"grad_norm": 7.0,
|
| 116 |
+
"learning_rate": 4.426315789473684e-05,
|
| 117 |
+
"loss": 1.611851119995117,
|
| 118 |
+
"mean_token_accuracy": 0.7179319798946381,
|
| 119 |
+
"num_tokens": 8698.0,
|
| 120 |
+
"step": 110
|
| 121 |
+
},
|
| 122 |
+
{
|
| 123 |
+
"entropy": 1.54140625,
|
| 124 |
+
"epoch": 0.25263157894736843,
|
| 125 |
+
"grad_norm": 6.90625,
|
| 126 |
+
"learning_rate": 4.373684210526316e-05,
|
| 127 |
+
"loss": 1.5053813934326172,
|
| 128 |
+
"mean_token_accuracy": 0.7282716870307923,
|
| 129 |
+
"num_tokens": 9455.0,
|
| 130 |
+
"step": 120
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"entropy": 1.5375,
|
| 134 |
+
"epoch": 0.2736842105263158,
|
| 135 |
+
"grad_norm": 7.25,
|
| 136 |
+
"learning_rate": 4.3210526315789475e-05,
|
| 137 |
+
"loss": 1.5904606819152831,
|
| 138 |
+
"mean_token_accuracy": 0.7234093546867371,
|
| 139 |
+
"num_tokens": 10191.0,
|
| 140 |
+
"step": 130
|
| 141 |
+
},
|
| 142 |
+
{
|
| 143 |
+
"entropy": 1.58984375,
|
| 144 |
+
"epoch": 0.29473684210526313,
|
| 145 |
+
"grad_norm": 9.375,
|
| 146 |
+
"learning_rate": 4.2684210526315795e-05,
|
| 147 |
+
"loss": 1.6070585250854492,
|
| 148 |
+
"mean_token_accuracy": 0.7228378415107727,
|
| 149 |
+
"num_tokens": 10859.0,
|
| 150 |
+
"step": 140
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"entropy": 1.80234375,
|
| 154 |
+
"epoch": 0.3157894736842105,
|
| 155 |
+
"grad_norm": 8.3125,
|
| 156 |
+
"learning_rate": 4.215789473684211e-05,
|
| 157 |
+
"loss": 1.7423076629638672,
|
| 158 |
+
"mean_token_accuracy": 0.6761326909065246,
|
| 159 |
+
"num_tokens": 11846.0,
|
| 160 |
+
"step": 150
|
| 161 |
+
},
|
| 162 |
+
{
|
| 163 |
+
"entropy": 1.592578125,
|
| 164 |
+
"epoch": 0.3368421052631579,
|
| 165 |
+
"grad_norm": 5.1875,
|
| 166 |
+
"learning_rate": 4.163157894736842e-05,
|
| 167 |
+
"loss": 1.6300687789916992,
|
| 168 |
+
"mean_token_accuracy": 0.7145272672176362,
|
| 169 |
+
"num_tokens": 12722.0,
|
| 170 |
+
"step": 160
|
| 171 |
+
},
|
| 172 |
+
{
|
| 173 |
+
"entropy": 1.5953125,
|
| 174 |
+
"epoch": 0.35789473684210527,
|
| 175 |
+
"grad_norm": 6.125,
|
| 176 |
+
"learning_rate": 4.110526315789474e-05,
|
| 177 |
+
"loss": 1.44827880859375,
|
| 178 |
+
"mean_token_accuracy": 0.725699108839035,
|
| 179 |
+
"num_tokens": 13458.0,
|
| 180 |
+
"step": 170
|
| 181 |
+
},
|
| 182 |
+
{
|
| 183 |
+
"entropy": 1.8171875,
|
| 184 |
+
"epoch": 0.37894736842105264,
|
| 185 |
+
"grad_norm": 7.125,
|
| 186 |
+
"learning_rate": 4.0578947368421054e-05,
|
| 187 |
+
"loss": 1.7392475128173828,
|
| 188 |
+
"mean_token_accuracy": 0.6949776589870453,
|
| 189 |
+
"num_tokens": 14197.0,
|
| 190 |
+
"step": 180
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"entropy": 1.78671875,
|
| 194 |
+
"epoch": 0.4,
|
| 195 |
+
"grad_norm": 7.8125,
|
| 196 |
+
"learning_rate": 4.0052631578947367e-05,
|
| 197 |
+
"loss": 1.8162834167480468,
|
| 198 |
+
"mean_token_accuracy": 0.6562099277973175,
|
| 199 |
+
"num_tokens": 15056.0,
|
| 200 |
+
"step": 190
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"entropy": 1.77421875,
|
| 204 |
+
"epoch": 0.42105263157894735,
|
| 205 |
+
"grad_norm": 7.53125,
|
| 206 |
+
"learning_rate": 3.9526315789473686e-05,
|
| 207 |
+
"loss": 1.7153417587280273,
|
| 208 |
+
"mean_token_accuracy": 0.6948422849178314,
|
| 209 |
+
"num_tokens": 15807.0,
|
| 210 |
+
"step": 200
|
| 211 |
+
},
|
| 212 |
+
{
|
| 213 |
+
"entropy": 1.6875,
|
| 214 |
+
"epoch": 0.4421052631578947,
|
| 215 |
+
"grad_norm": 9.125,
|
| 216 |
+
"learning_rate": 3.9000000000000006e-05,
|
| 217 |
+
"loss": 1.6782726287841796,
|
| 218 |
+
"mean_token_accuracy": 0.7238920211791993,
|
| 219 |
+
"num_tokens": 16432.0,
|
| 220 |
+
"step": 210
|
| 221 |
+
},
|
| 222 |
+
{
|
| 223 |
+
"entropy": 1.446484375,
|
| 224 |
+
"epoch": 0.4631578947368421,
|
| 225 |
+
"grad_norm": 5.84375,
|
| 226 |
+
"learning_rate": 3.847368421052632e-05,
|
| 227 |
+
"loss": 1.4408910751342774,
|
| 228 |
+
"mean_token_accuracy": 0.7403214454650879,
|
| 229 |
+
"num_tokens": 17103.0,
|
| 230 |
+
"step": 220
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"entropy": 1.76171875,
|
| 234 |
+
"epoch": 0.4842105263157895,
|
| 235 |
+
"grad_norm": 6.375,
|
| 236 |
+
"learning_rate": 3.794736842105263e-05,
|
| 237 |
+
"loss": 1.7578521728515626,
|
| 238 |
+
"mean_token_accuracy": 0.6848492741584777,
|
| 239 |
+
"num_tokens": 18038.0,
|
| 240 |
+
"step": 230
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"entropy": 1.53515625,
|
| 244 |
+
"epoch": 0.5052631578947369,
|
| 245 |
+
"grad_norm": 6.71875,
|
| 246 |
+
"learning_rate": 3.742105263157895e-05,
|
| 247 |
+
"loss": 1.48400821685791,
|
| 248 |
+
"mean_token_accuracy": 0.7131125509738923,
|
| 249 |
+
"num_tokens": 18720.0,
|
| 250 |
+
"step": 240
|
| 251 |
+
},
|
| 252 |
+
{
|
| 253 |
+
"entropy": 1.49140625,
|
| 254 |
+
"epoch": 0.5263157894736842,
|
| 255 |
+
"grad_norm": 6.375,
|
| 256 |
+
"learning_rate": 3.6894736842105265e-05,
|
| 257 |
+
"loss": 1.5496106147766113,
|
| 258 |
+
"mean_token_accuracy": 0.7365909218788147,
|
| 259 |
+
"num_tokens": 19308.0,
|
| 260 |
+
"step": 250
|
| 261 |
+
},
|
| 262 |
+
{
|
| 263 |
+
"entropy": 1.46796875,
|
| 264 |
+
"epoch": 0.5473684210526316,
|
| 265 |
+
"grad_norm": 8.125,
|
| 266 |
+
"learning_rate": 3.636842105263158e-05,
|
| 267 |
+
"loss": 1.5240853309631348,
|
| 268 |
+
"mean_token_accuracy": 0.736497437953949,
|
| 269 |
+
"num_tokens": 19967.0,
|
| 270 |
+
"step": 260
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"entropy": 1.578125,
|
| 274 |
+
"epoch": 0.5684210526315789,
|
| 275 |
+
"grad_norm": 8.3125,
|
| 276 |
+
"learning_rate": 3.58421052631579e-05,
|
| 277 |
+
"loss": 1.5680004119873048,
|
| 278 |
+
"mean_token_accuracy": 0.7332142323255539,
|
| 279 |
+
"num_tokens": 20769.0,
|
| 280 |
+
"step": 270
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"entropy": 1.501953125,
|
| 284 |
+
"epoch": 0.5894736842105263,
|
| 285 |
+
"grad_norm": 8.375,
|
| 286 |
+
"learning_rate": 3.531578947368421e-05,
|
| 287 |
+
"loss": 1.5029385566711426,
|
| 288 |
+
"mean_token_accuracy": 0.7254173457622528,
|
| 289 |
+
"num_tokens": 22503.0,
|
| 290 |
+
"step": 280
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"entropy": 1.5921875,
|
| 294 |
+
"epoch": 0.6105263157894737,
|
| 295 |
+
"grad_norm": 7.0,
|
| 296 |
+
"learning_rate": 3.478947368421053e-05,
|
| 297 |
+
"loss": 1.6090343475341797,
|
| 298 |
+
"mean_token_accuracy": 0.7082146763801574,
|
| 299 |
+
"num_tokens": 23439.0,
|
| 300 |
+
"step": 290
|
| 301 |
+
},
|
| 302 |
+
{
|
| 303 |
+
"entropy": 1.75078125,
|
| 304 |
+
"epoch": 0.631578947368421,
|
| 305 |
+
"grad_norm": 8.25,
|
| 306 |
+
"learning_rate": 3.426315789473684e-05,
|
| 307 |
+
"loss": 1.6250024795532227,
|
| 308 |
+
"mean_token_accuracy": 0.6883616149425507,
|
| 309 |
+
"num_tokens": 24161.0,
|
| 310 |
+
"step": 300
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"entropy": 1.269140625,
|
| 314 |
+
"epoch": 0.6526315789473685,
|
| 315 |
+
"grad_norm": 7.46875,
|
| 316 |
+
"learning_rate": 3.373684210526316e-05,
|
| 317 |
+
"loss": 1.3148769378662108,
|
| 318 |
+
"mean_token_accuracy": 0.7618813216686249,
|
| 319 |
+
"num_tokens": 24716.0,
|
| 320 |
+
"step": 310
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"entropy": 1.809375,
|
| 324 |
+
"epoch": 0.6736842105263158,
|
| 325 |
+
"grad_norm": 7.40625,
|
| 326 |
+
"learning_rate": 3.3210526315789476e-05,
|
| 327 |
+
"loss": 1.9381757736206056,
|
| 328 |
+
"mean_token_accuracy": 0.667089307308197,
|
| 329 |
+
"num_tokens": 25481.0,
|
| 330 |
+
"step": 320
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"entropy": 1.73515625,
|
| 334 |
+
"epoch": 0.6947368421052632,
|
| 335 |
+
"grad_norm": 11.5625,
|
| 336 |
+
"learning_rate": 3.268421052631579e-05,
|
| 337 |
+
"loss": 1.76402587890625,
|
| 338 |
+
"mean_token_accuracy": 0.6992617845535278,
|
| 339 |
+
"num_tokens": 26159.0,
|
| 340 |
+
"step": 330
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"entropy": 1.644140625,
|
| 344 |
+
"epoch": 0.7157894736842105,
|
| 345 |
+
"grad_norm": 7.625,
|
| 346 |
+
"learning_rate": 3.215789473684211e-05,
|
| 347 |
+
"loss": 1.6910707473754882,
|
| 348 |
+
"mean_token_accuracy": 0.6974358975887298,
|
| 349 |
+
"num_tokens": 26983.0,
|
| 350 |
+
"step": 340
|
| 351 |
+
},
|
| 352 |
+
{
|
| 353 |
+
"entropy": 1.44140625,
|
| 354 |
+
"epoch": 0.7368421052631579,
|
| 355 |
+
"grad_norm": 7.84375,
|
| 356 |
+
"learning_rate": 3.163157894736842e-05,
|
| 357 |
+
"loss": 1.3823105812072753,
|
| 358 |
+
"mean_token_accuracy": 0.7533604800701141,
|
| 359 |
+
"num_tokens": 27564.0,
|
| 360 |
+
"step": 350
|
| 361 |
+
},
|
| 362 |
+
{
|
| 363 |
+
"entropy": 1.41953125,
|
| 364 |
+
"epoch": 0.7578947368421053,
|
| 365 |
+
"grad_norm": 9.5,
|
| 366 |
+
"learning_rate": 3.1105263157894735e-05,
|
| 367 |
+
"loss": 1.488726806640625,
|
| 368 |
+
"mean_token_accuracy": 0.733670562505722,
|
| 369 |
+
"num_tokens": 28200.0,
|
| 370 |
+
"step": 360
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"entropy": 1.753125,
|
| 374 |
+
"epoch": 0.7789473684210526,
|
| 375 |
+
"grad_norm": 5.9375,
|
| 376 |
+
"learning_rate": 3.0578947368421054e-05,
|
| 377 |
+
"loss": 1.8948373794555664,
|
| 378 |
+
"mean_token_accuracy": 0.675409197807312,
|
| 379 |
+
"num_tokens": 28919.0,
|
| 380 |
+
"step": 370
|
| 381 |
+
},
|
| 382 |
+
{
|
| 383 |
+
"entropy": 1.98828125,
|
| 384 |
+
"epoch": 0.8,
|
| 385 |
+
"grad_norm": 8.5,
|
| 386 |
+
"learning_rate": 3.005263157894737e-05,
|
| 387 |
+
"loss": 2.146737289428711,
|
| 388 |
+
"mean_token_accuracy": 0.6334406793117523,
|
| 389 |
+
"num_tokens": 29894.0,
|
| 390 |
+
"step": 380
|
| 391 |
+
},
|
| 392 |
+
{
|
| 393 |
+
"entropy": 1.85625,
|
| 394 |
+
"epoch": 0.8210526315789474,
|
| 395 |
+
"grad_norm": 6.3125,
|
| 396 |
+
"learning_rate": 2.9526315789473684e-05,
|
| 397 |
+
"loss": 1.858269500732422,
|
| 398 |
+
"mean_token_accuracy": 0.6439402669668197,
|
| 399 |
+
"num_tokens": 30652.0,
|
| 400 |
+
"step": 390
|
| 401 |
+
},
|
| 402 |
+
{
|
| 403 |
+
"entropy": 1.727734375,
|
| 404 |
+
"epoch": 0.8421052631578947,
|
| 405 |
+
"grad_norm": 7.96875,
|
| 406 |
+
"learning_rate": 2.9e-05,
|
| 407 |
+
"loss": 1.7343599319458007,
|
| 408 |
+
"mean_token_accuracy": 0.700012594461441,
|
| 409 |
+
"num_tokens": 31372.0,
|
| 410 |
+
"step": 400
|
| 411 |
+
},
|
| 412 |
+
{
|
| 413 |
+
"entropy": 1.65,
|
| 414 |
+
"epoch": 0.8631578947368421,
|
| 415 |
+
"grad_norm": 8.0625,
|
| 416 |
+
"learning_rate": 2.847368421052632e-05,
|
| 417 |
+
"loss": 1.6553001403808594,
|
| 418 |
+
"mean_token_accuracy": 0.7165175020694733,
|
| 419 |
+
"num_tokens": 32149.0,
|
| 420 |
+
"step": 410
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"entropy": 1.596875,
|
| 424 |
+
"epoch": 0.8842105263157894,
|
| 425 |
+
"grad_norm": 6.59375,
|
| 426 |
+
"learning_rate": 2.7947368421052633e-05,
|
| 427 |
+
"loss": 1.5661128997802733,
|
| 428 |
+
"mean_token_accuracy": 0.7166097521781921,
|
| 429 |
+
"num_tokens": 33203.0,
|
| 430 |
+
"step": 420
|
| 431 |
+
},
|
| 432 |
+
{
|
| 433 |
+
"entropy": 1.6015625,
|
| 434 |
+
"epoch": 0.9052631578947369,
|
| 435 |
+
"grad_norm": 7.1875,
|
| 436 |
+
"learning_rate": 2.7421052631578946e-05,
|
| 437 |
+
"loss": 1.6179162979125976,
|
| 438 |
+
"mean_token_accuracy": 0.7093785464763641,
|
| 439 |
+
"num_tokens": 33889.0,
|
| 440 |
+
"step": 430
|
| 441 |
+
},
|
| 442 |
+
{
|
| 443 |
+
"entropy": 1.75546875,
|
| 444 |
+
"epoch": 0.9263157894736842,
|
| 445 |
+
"grad_norm": 7.21875,
|
| 446 |
+
"learning_rate": 2.6894736842105266e-05,
|
| 447 |
+
"loss": 1.7675729751586915,
|
| 448 |
+
"mean_token_accuracy": 0.6769322335720063,
|
| 449 |
+
"num_tokens": 34862.0,
|
| 450 |
+
"step": 440
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"entropy": 1.92109375,
|
| 454 |
+
"epoch": 0.9473684210526315,
|
| 455 |
+
"grad_norm": 11.0625,
|
| 456 |
+
"learning_rate": 2.6368421052631582e-05,
|
| 457 |
+
"loss": 1.9548963546752929,
|
| 458 |
+
"mean_token_accuracy": 0.648491358757019,
|
| 459 |
+
"num_tokens": 35710.0,
|
| 460 |
+
"step": 450
|
| 461 |
+
},
|
| 462 |
+
{
|
| 463 |
+
"entropy": 1.699609375,
|
| 464 |
+
"epoch": 0.968421052631579,
|
| 465 |
+
"grad_norm": 6.03125,
|
| 466 |
+
"learning_rate": 2.5842105263157895e-05,
|
| 467 |
+
"loss": 1.6432403564453124,
|
| 468 |
+
"mean_token_accuracy": 0.6918732106685639,
|
| 469 |
+
"num_tokens": 36374.0,
|
| 470 |
+
"step": 460
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"entropy": 1.84453125,
|
| 474 |
+
"epoch": 0.9894736842105263,
|
| 475 |
+
"grad_norm": 7.03125,
|
| 476 |
+
"learning_rate": 2.5315789473684208e-05,
|
| 477 |
+
"loss": 1.7591934204101562,
|
| 478 |
+
"mean_token_accuracy": 0.6581568241119384,
|
| 479 |
+
"num_tokens": 37139.0,
|
| 480 |
+
"step": 470
|
| 481 |
+
},
|
| 482 |
+
{
|
| 483 |
+
"entropy": 1.75859375,
|
| 484 |
+
"epoch": 1.0105263157894737,
|
| 485 |
+
"grad_norm": 7.15625,
|
| 486 |
+
"learning_rate": 2.4789473684210528e-05,
|
| 487 |
+
"loss": 1.7067642211914062,
|
| 488 |
+
"mean_token_accuracy": 0.7008034646511078,
|
| 489 |
+
"num_tokens": 38037.0,
|
| 490 |
+
"step": 480
|
| 491 |
+
},
|
| 492 |
+
{
|
| 493 |
+
"entropy": 1.4203125,
|
| 494 |
+
"epoch": 1.0315789473684212,
|
| 495 |
+
"grad_norm": 6.4375,
|
| 496 |
+
"learning_rate": 2.4263157894736844e-05,
|
| 497 |
+
"loss": 1.378176498413086,
|
| 498 |
+
"mean_token_accuracy": 0.7486959993839264,
|
| 499 |
+
"num_tokens": 38668.0,
|
| 500 |
+
"step": 490
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"entropy": 1.56328125,
|
| 504 |
+
"epoch": 1.0526315789473684,
|
| 505 |
+
"grad_norm": 7.15625,
|
| 506 |
+
"learning_rate": 2.373684210526316e-05,
|
| 507 |
+
"loss": 1.4762983322143555,
|
| 508 |
+
"mean_token_accuracy": 0.7175065577030182,
|
| 509 |
+
"num_tokens": 39424.0,
|
| 510 |
+
"step": 500
|
| 511 |
+
},
|
| 512 |
+
{
|
| 513 |
+
"entropy": 1.630078125,
|
| 514 |
+
"epoch": 1.0736842105263158,
|
| 515 |
+
"grad_norm": 4.84375,
|
| 516 |
+
"learning_rate": 2.3210526315789473e-05,
|
| 517 |
+
"loss": 1.5906378746032714,
|
| 518 |
+
"mean_token_accuracy": 0.6876044690608978,
|
| 519 |
+
"num_tokens": 40317.0,
|
| 520 |
+
"step": 510
|
| 521 |
+
},
|
| 522 |
+
{
|
| 523 |
+
"entropy": 1.556640625,
|
| 524 |
+
"epoch": 1.0947368421052632,
|
| 525 |
+
"grad_norm": 6.59375,
|
| 526 |
+
"learning_rate": 2.268421052631579e-05,
|
| 527 |
+
"loss": 1.6246864318847656,
|
| 528 |
+
"mean_token_accuracy": 0.7021546125411987,
|
| 529 |
+
"num_tokens": 41101.0,
|
| 530 |
+
"step": 520
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"entropy": 1.86015625,
|
| 534 |
+
"epoch": 1.1157894736842104,
|
| 535 |
+
"grad_norm": 6.3125,
|
| 536 |
+
"learning_rate": 2.2157894736842106e-05,
|
| 537 |
+
"loss": 1.832110595703125,
|
| 538 |
+
"mean_token_accuracy": 0.6659113824367523,
|
| 539 |
+
"num_tokens": 42080.0,
|
| 540 |
+
"step": 530
|
| 541 |
+
},
|
| 542 |
+
{
|
| 543 |
+
"entropy": 1.50859375,
|
| 544 |
+
"epoch": 1.1368421052631579,
|
| 545 |
+
"grad_norm": 4.75,
|
| 546 |
+
"learning_rate": 2.1631578947368423e-05,
|
| 547 |
+
"loss": 1.495261001586914,
|
| 548 |
+
"mean_token_accuracy": 0.7114485323429107,
|
| 549 |
+
"num_tokens": 43064.0,
|
| 550 |
+
"step": 540
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"entropy": 1.379296875,
|
| 554 |
+
"epoch": 1.1578947368421053,
|
| 555 |
+
"grad_norm": 7.90625,
|
| 556 |
+
"learning_rate": 2.110526315789474e-05,
|
| 557 |
+
"loss": 1.3109845161437987,
|
| 558 |
+
"mean_token_accuracy": 0.7768619418144226,
|
| 559 |
+
"num_tokens": 43638.0,
|
| 560 |
+
"step": 550
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"entropy": 1.671875,
|
| 564 |
+
"epoch": 1.1789473684210527,
|
| 565 |
+
"grad_norm": 8.1875,
|
| 566 |
+
"learning_rate": 2.0578947368421052e-05,
|
| 567 |
+
"loss": 1.7390716552734375,
|
| 568 |
+
"mean_token_accuracy": 0.6966245353221894,
|
| 569 |
+
"num_tokens": 44538.0,
|
| 570 |
+
"step": 560
|
| 571 |
+
},
|
| 572 |
+
{
|
| 573 |
+
"entropy": 1.45390625,
|
| 574 |
+
"epoch": 1.2,
|
| 575 |
+
"grad_norm": 8.5,
|
| 576 |
+
"learning_rate": 2.005263157894737e-05,
|
| 577 |
+
"loss": 1.3690235137939453,
|
| 578 |
+
"mean_token_accuracy": 0.742480456829071,
|
| 579 |
+
"num_tokens": 45159.0,
|
| 580 |
+
"step": 570
|
| 581 |
+
},
|
| 582 |
+
{
|
| 583 |
+
"entropy": 1.8234375,
|
| 584 |
+
"epoch": 1.2210526315789474,
|
| 585 |
+
"grad_norm": 6.65625,
|
| 586 |
+
"learning_rate": 1.9526315789473685e-05,
|
| 587 |
+
"loss": 1.890069580078125,
|
| 588 |
+
"mean_token_accuracy": 0.6900433540344239,
|
| 589 |
+
"num_tokens": 45918.0,
|
| 590 |
+
"step": 580
|
| 591 |
+
},
|
| 592 |
+
{
|
| 593 |
+
"entropy": 1.8640625,
|
| 594 |
+
"epoch": 1.2421052631578948,
|
| 595 |
+
"grad_norm": 8.125,
|
| 596 |
+
"learning_rate": 1.9e-05,
|
| 597 |
+
"loss": 1.7881799697875977,
|
| 598 |
+
"mean_token_accuracy": 0.6537608683109284,
|
| 599 |
+
"num_tokens": 46795.0,
|
| 600 |
+
"step": 590
|
| 601 |
+
},
|
| 602 |
+
{
|
| 603 |
+
"entropy": 1.59453125,
|
| 604 |
+
"epoch": 1.263157894736842,
|
| 605 |
+
"grad_norm": 6.5,
|
| 606 |
+
"learning_rate": 1.8473684210526317e-05,
|
| 607 |
+
"loss": 1.5872214317321778,
|
| 608 |
+
"mean_token_accuracy": 0.722487497329712,
|
| 609 |
+
"num_tokens": 47423.0,
|
| 610 |
+
"step": 600
|
| 611 |
+
},
|
| 612 |
+
{
|
| 613 |
+
"entropy": 1.78828125,
|
| 614 |
+
"epoch": 1.2842105263157895,
|
| 615 |
+
"grad_norm": 7.3125,
|
| 616 |
+
"learning_rate": 1.7947368421052634e-05,
|
| 617 |
+
"loss": 1.6725593566894532,
|
| 618 |
+
"mean_token_accuracy": 0.6980259001255036,
|
| 619 |
+
"num_tokens": 48158.0,
|
| 620 |
+
"step": 610
|
| 621 |
+
},
|
| 622 |
+
{
|
| 623 |
+
"entropy": 1.61015625,
|
| 624 |
+
"epoch": 1.305263157894737,
|
| 625 |
+
"grad_norm": 6.5,
|
| 626 |
+
"learning_rate": 1.742105263157895e-05,
|
| 627 |
+
"loss": 1.5390226364135742,
|
| 628 |
+
"mean_token_accuracy": 0.725999391078949,
|
| 629 |
+
"num_tokens": 48875.0,
|
| 630 |
+
"step": 620
|
| 631 |
+
},
|
| 632 |
+
{
|
| 633 |
+
"entropy": 1.71953125,
|
| 634 |
+
"epoch": 1.3263157894736843,
|
| 635 |
+
"grad_norm": 8.25,
|
| 636 |
+
"learning_rate": 1.6894736842105263e-05,
|
| 637 |
+
"loss": 1.6228282928466797,
|
| 638 |
+
"mean_token_accuracy": 0.7085169553756714,
|
| 639 |
+
"num_tokens": 49697.0,
|
| 640 |
+
"step": 630
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"entropy": 1.437109375,
|
| 644 |
+
"epoch": 1.3473684210526315,
|
| 645 |
+
"grad_norm": 7.5625,
|
| 646 |
+
"learning_rate": 1.636842105263158e-05,
|
| 647 |
+
"loss": 1.4906560897827148,
|
| 648 |
+
"mean_token_accuracy": 0.7471937596797943,
|
| 649 |
+
"num_tokens": 50300.0,
|
| 650 |
+
"step": 640
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"entropy": 1.73359375,
|
| 654 |
+
"epoch": 1.368421052631579,
|
| 655 |
+
"grad_norm": 4.375,
|
| 656 |
+
"learning_rate": 1.5842105263157896e-05,
|
| 657 |
+
"loss": 1.5986823081970214,
|
| 658 |
+
"mean_token_accuracy": 0.699000483751297,
|
| 659 |
+
"num_tokens": 51332.0,
|
| 660 |
+
"step": 650
|
| 661 |
+
},
|
| 662 |
+
{
|
| 663 |
+
"entropy": 1.2671875,
|
| 664 |
+
"epoch": 1.3894736842105262,
|
| 665 |
+
"grad_norm": 11.8125,
|
| 666 |
+
"learning_rate": 1.5315789473684212e-05,
|
| 667 |
+
"loss": 1.234378719329834,
|
| 668 |
+
"mean_token_accuracy": 0.7714344441890717,
|
| 669 |
+
"num_tokens": 51900.0,
|
| 670 |
+
"step": 660
|
| 671 |
+
},
|
| 672 |
+
{
|
| 673 |
+
"entropy": 1.90859375,
|
| 674 |
+
"epoch": 1.4105263157894736,
|
| 675 |
+
"grad_norm": 6.875,
|
| 676 |
+
"learning_rate": 1.4789473684210529e-05,
|
| 677 |
+
"loss": 1.815606689453125,
|
| 678 |
+
"mean_token_accuracy": 0.6656625211238861,
|
| 679 |
+
"num_tokens": 52883.0,
|
| 680 |
+
"step": 670
|
| 681 |
+
},
|
| 682 |
+
{
|
| 683 |
+
"entropy": 1.558203125,
|
| 684 |
+
"epoch": 1.431578947368421,
|
| 685 |
+
"grad_norm": 9.375,
|
| 686 |
+
"learning_rate": 1.4263157894736842e-05,
|
| 687 |
+
"loss": 1.4858011245727538,
|
| 688 |
+
"mean_token_accuracy": 0.7443289816379547,
|
| 689 |
+
"num_tokens": 53576.0,
|
| 690 |
+
"step": 680
|
| 691 |
+
},
|
| 692 |
+
{
|
| 693 |
+
"entropy": 1.64765625,
|
| 694 |
+
"epoch": 1.4526315789473685,
|
| 695 |
+
"grad_norm": 8.9375,
|
| 696 |
+
"learning_rate": 1.373684210526316e-05,
|
| 697 |
+
"loss": 1.534531307220459,
|
| 698 |
+
"mean_token_accuracy": 0.7204049170017243,
|
| 699 |
+
"num_tokens": 54450.0,
|
| 700 |
+
"step": 690
|
| 701 |
+
},
|
| 702 |
+
{
|
| 703 |
+
"entropy": 1.58359375,
|
| 704 |
+
"epoch": 1.4736842105263157,
|
| 705 |
+
"grad_norm": 6.46875,
|
| 706 |
+
"learning_rate": 1.3210526315789473e-05,
|
| 707 |
+
"loss": 1.5365165710449218,
|
| 708 |
+
"mean_token_accuracy": 0.7069519102573395,
|
| 709 |
+
"num_tokens": 55160.0,
|
| 710 |
+
"step": 700
|
| 711 |
+
},
|
| 712 |
+
{
|
| 713 |
+
"entropy": 1.4171875,
|
| 714 |
+
"epoch": 1.4947368421052631,
|
| 715 |
+
"grad_norm": 7.1875,
|
| 716 |
+
"learning_rate": 1.268421052631579e-05,
|
| 717 |
+
"loss": 1.307802391052246,
|
| 718 |
+
"mean_token_accuracy": 0.7627157270908356,
|
| 719 |
+
"num_tokens": 55929.0,
|
| 720 |
+
"step": 710
|
| 721 |
+
},
|
| 722 |
+
{
|
| 723 |
+
"entropy": 1.678125,
|
| 724 |
+
"epoch": 1.5157894736842106,
|
| 725 |
+
"grad_norm": 6.34375,
|
| 726 |
+
"learning_rate": 1.2157894736842105e-05,
|
| 727 |
+
"loss": 1.5420659065246582,
|
| 728 |
+
"mean_token_accuracy": 0.7097006201744079,
|
| 729 |
+
"num_tokens": 56725.0,
|
| 730 |
+
"step": 720
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"entropy": 1.58125,
|
| 734 |
+
"epoch": 1.5368421052631578,
|
| 735 |
+
"grad_norm": 7.0625,
|
| 736 |
+
"learning_rate": 1.1631578947368422e-05,
|
| 737 |
+
"loss": 1.4044910430908204,
|
| 738 |
+
"mean_token_accuracy": 0.7465328335762024,
|
| 739 |
+
"num_tokens": 57398.0,
|
| 740 |
+
"step": 730
|
| 741 |
+
},
|
| 742 |
+
{
|
| 743 |
+
"entropy": 1.576171875,
|
| 744 |
+
"epoch": 1.5578947368421052,
|
| 745 |
+
"grad_norm": 6.03125,
|
| 746 |
+
"learning_rate": 1.1105263157894738e-05,
|
| 747 |
+
"loss": 1.605686569213867,
|
| 748 |
+
"mean_token_accuracy": 0.7333506286144257,
|
| 749 |
+
"num_tokens": 58072.0,
|
| 750 |
+
"step": 740
|
| 751 |
+
},
|
| 752 |
+
{
|
| 753 |
+
"entropy": 1.59140625,
|
| 754 |
+
"epoch": 1.5789473684210527,
|
| 755 |
+
"grad_norm": 6.03125,
|
| 756 |
+
"learning_rate": 1.0578947368421053e-05,
|
| 757 |
+
"loss": 1.4928099632263183,
|
| 758 |
+
"mean_token_accuracy": 0.7072650909423828,
|
| 759 |
+
"num_tokens": 58748.0,
|
| 760 |
+
"step": 750
|
| 761 |
+
},
|
| 762 |
+
{
|
| 763 |
+
"entropy": 1.63046875,
|
| 764 |
+
"epoch": 1.6,
|
| 765 |
+
"grad_norm": 7.1875,
|
| 766 |
+
"learning_rate": 1.005263157894737e-05,
|
| 767 |
+
"loss": 1.624325942993164,
|
| 768 |
+
"mean_token_accuracy": 0.690889635682106,
|
| 769 |
+
"num_tokens": 59658.0,
|
| 770 |
+
"step": 760
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"entropy": 1.803125,
|
| 774 |
+
"epoch": 1.6210526315789475,
|
| 775 |
+
"grad_norm": 6.03125,
|
| 776 |
+
"learning_rate": 9.526315789473684e-06,
|
| 777 |
+
"loss": 1.7750425338745117,
|
| 778 |
+
"mean_token_accuracy": 0.6651369571685791,
|
| 779 |
+
"num_tokens": 60859.0,
|
| 780 |
+
"step": 770
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"entropy": 1.4703125,
|
| 784 |
+
"epoch": 1.6421052631578947,
|
| 785 |
+
"grad_norm": 7.03125,
|
| 786 |
+
"learning_rate": 9e-06,
|
| 787 |
+
"loss": 1.4258437156677246,
|
| 788 |
+
"mean_token_accuracy": 0.7347829401493072,
|
| 789 |
+
"num_tokens": 62439.0,
|
| 790 |
+
"step": 780
|
| 791 |
+
},
|
| 792 |
+
{
|
| 793 |
+
"entropy": 1.671875,
|
| 794 |
+
"epoch": 1.663157894736842,
|
| 795 |
+
"grad_norm": 7.71875,
|
| 796 |
+
"learning_rate": 8.473684210526315e-06,
|
| 797 |
+
"loss": 1.721219825744629,
|
| 798 |
+
"mean_token_accuracy": 0.7035934925079346,
|
| 799 |
+
"num_tokens": 63290.0,
|
| 800 |
+
"step": 790
|
| 801 |
+
},
|
| 802 |
+
{
|
| 803 |
+
"entropy": 1.554296875,
|
| 804 |
+
"epoch": 1.6842105263157894,
|
| 805 |
+
"grad_norm": 9.0625,
|
| 806 |
+
"learning_rate": 7.947368421052633e-06,
|
| 807 |
+
"loss": 1.5237810134887695,
|
| 808 |
+
"mean_token_accuracy": 0.7325670003890992,
|
| 809 |
+
"num_tokens": 64010.0,
|
| 810 |
+
"step": 800
|
| 811 |
+
},
|
| 812 |
+
{
|
| 813 |
+
"entropy": 1.7859375,
|
| 814 |
+
"epoch": 1.7052631578947368,
|
| 815 |
+
"grad_norm": 8.5,
|
| 816 |
+
"learning_rate": 7.421052631578948e-06,
|
| 817 |
+
"loss": 1.7944671630859375,
|
| 818 |
+
"mean_token_accuracy": 0.683906614780426,
|
| 819 |
+
"num_tokens": 64966.0,
|
| 820 |
+
"step": 810
|
| 821 |
+
},
|
| 822 |
+
{
|
| 823 |
+
"entropy": 1.840625,
|
| 824 |
+
"epoch": 1.7263157894736842,
|
| 825 |
+
"grad_norm": 7.53125,
|
| 826 |
+
"learning_rate": 6.894736842105263e-06,
|
| 827 |
+
"loss": 1.787227249145508,
|
| 828 |
+
"mean_token_accuracy": 0.6670031368732452,
|
| 829 |
+
"num_tokens": 65703.0,
|
| 830 |
+
"step": 820
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"entropy": 1.29765625,
|
| 834 |
+
"epoch": 1.7473684210526317,
|
| 835 |
+
"grad_norm": 9.0625,
|
| 836 |
+
"learning_rate": 6.368421052631579e-06,
|
| 837 |
+
"loss": 1.2813177108764648,
|
| 838 |
+
"mean_token_accuracy": 0.7610228896141052,
|
| 839 |
+
"num_tokens": 66421.0,
|
| 840 |
+
"step": 830
|
| 841 |
+
},
|
| 842 |
+
{
|
| 843 |
+
"entropy": 1.504296875,
|
| 844 |
+
"epoch": 1.768421052631579,
|
| 845 |
+
"grad_norm": 6.5,
|
| 846 |
+
"learning_rate": 5.842105263157895e-06,
|
| 847 |
+
"loss": 1.477138137817383,
|
| 848 |
+
"mean_token_accuracy": 0.7455608665943145,
|
| 849 |
+
"num_tokens": 67005.0,
|
| 850 |
+
"step": 840
|
| 851 |
+
},
|
| 852 |
+
{
|
| 853 |
+
"entropy": 1.60859375,
|
| 854 |
+
"epoch": 1.7894736842105263,
|
| 855 |
+
"grad_norm": 5.03125,
|
| 856 |
+
"learning_rate": 5.315789473684211e-06,
|
| 857 |
+
"loss": 1.6519662857055664,
|
| 858 |
+
"mean_token_accuracy": 0.7005816400051117,
|
| 859 |
+
"num_tokens": 67975.0,
|
| 860 |
+
"step": 850
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"entropy": 1.6765625,
|
| 864 |
+
"epoch": 1.8105263157894735,
|
| 865 |
+
"grad_norm": 6.53125,
|
| 866 |
+
"learning_rate": 4.789473684210526e-06,
|
| 867 |
+
"loss": 1.6701608657836915,
|
| 868 |
+
"mean_token_accuracy": 0.7136963486671448,
|
| 869 |
+
"num_tokens": 68650.0,
|
| 870 |
+
"step": 860
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"entropy": 1.44609375,
|
| 874 |
+
"epoch": 1.831578947368421,
|
| 875 |
+
"grad_norm": 7.1875,
|
| 876 |
+
"learning_rate": 4.2631578947368425e-06,
|
| 877 |
+
"loss": 1.3475713729858398,
|
| 878 |
+
"mean_token_accuracy": 0.7566476047039032,
|
| 879 |
+
"num_tokens": 69317.0,
|
| 880 |
+
"step": 870
|
| 881 |
+
},
|
| 882 |
+
{
|
| 883 |
+
"entropy": 1.71328125,
|
| 884 |
+
"epoch": 1.8526315789473684,
|
| 885 |
+
"grad_norm": 6.90625,
|
| 886 |
+
"learning_rate": 3.736842105263158e-06,
|
| 887 |
+
"loss": 1.7077770233154297,
|
| 888 |
+
"mean_token_accuracy": 0.6919207274913788,
|
| 889 |
+
"num_tokens": 70169.0,
|
| 890 |
+
"step": 880
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"entropy": 1.646875,
|
| 894 |
+
"epoch": 1.8736842105263158,
|
| 895 |
+
"grad_norm": 7.8125,
|
| 896 |
+
"learning_rate": 3.2105263157894735e-06,
|
| 897 |
+
"loss": 1.5933343887329101,
|
| 898 |
+
"mean_token_accuracy": 0.708821702003479,
|
| 899 |
+
"num_tokens": 71138.0,
|
| 900 |
+
"step": 890
|
| 901 |
+
},
|
| 902 |
+
{
|
| 903 |
+
"entropy": 1.437109375,
|
| 904 |
+
"epoch": 1.8947368421052633,
|
| 905 |
+
"grad_norm": 8.625,
|
| 906 |
+
"learning_rate": 2.68421052631579e-06,
|
| 907 |
+
"loss": 1.4426955223083495,
|
| 908 |
+
"mean_token_accuracy": 0.7344056785106658,
|
| 909 |
+
"num_tokens": 71776.0,
|
| 910 |
+
"step": 900
|
| 911 |
+
},
|
| 912 |
+
{
|
| 913 |
+
"entropy": 1.591015625,
|
| 914 |
+
"epoch": 1.9157894736842105,
|
| 915 |
+
"grad_norm": 8.125,
|
| 916 |
+
"learning_rate": 2.1578947368421054e-06,
|
| 917 |
+
"loss": 1.4523811340332031,
|
| 918 |
+
"mean_token_accuracy": 0.7294364452362061,
|
| 919 |
+
"num_tokens": 72390.0,
|
| 920 |
+
"step": 910
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"entropy": 1.48984375,
|
| 924 |
+
"epoch": 1.936842105263158,
|
| 925 |
+
"grad_norm": 8.375,
|
| 926 |
+
"learning_rate": 1.6315789473684212e-06,
|
| 927 |
+
"loss": 1.4202921867370606,
|
| 928 |
+
"mean_token_accuracy": 0.7283547043800354,
|
| 929 |
+
"num_tokens": 73109.0,
|
| 930 |
+
"step": 920
|
| 931 |
+
},
|
| 932 |
+
{
|
| 933 |
+
"entropy": 1.37578125,
|
| 934 |
+
"epoch": 1.9578947368421051,
|
| 935 |
+
"grad_norm": 7.34375,
|
| 936 |
+
"learning_rate": 1.1052631578947369e-06,
|
| 937 |
+
"loss": 1.3212156295776367,
|
| 938 |
+
"mean_token_accuracy": 0.765373581647873,
|
| 939 |
+
"num_tokens": 73665.0,
|
| 940 |
+
"step": 930
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"entropy": 1.69140625,
|
| 944 |
+
"epoch": 1.9789473684210526,
|
| 945 |
+
"grad_norm": 7.5625,
|
| 946 |
+
"learning_rate": 5.789473684210527e-07,
|
| 947 |
+
"loss": 1.5747876167297363,
|
| 948 |
+
"mean_token_accuracy": 0.7353746354579925,
|
| 949 |
+
"num_tokens": 74320.0,
|
| 950 |
+
"step": 940
|
| 951 |
+
},
|
| 952 |
+
{
|
| 953 |
+
"entropy": 1.75390625,
|
| 954 |
+
"epoch": 2.0,
|
| 955 |
+
"grad_norm": 6.46875,
|
| 956 |
+
"learning_rate": 5.263157894736842e-08,
|
| 957 |
+
"loss": 1.7055414199829102,
|
| 958 |
+
"mean_token_accuracy": 0.6862038552761078,
|
| 959 |
+
"num_tokens": 75304.0,
|
| 960 |
+
"step": 950
|
| 961 |
+
}
|
| 962 |
+
],
|
| 963 |
+
"logging_steps": 10,
|
| 964 |
+
"max_steps": 950,
|
| 965 |
+
"num_input_tokens_seen": 0,
|
| 966 |
+
"num_train_epochs": 2,
|
| 967 |
+
"save_steps": 100,
|
| 968 |
+
"stateful_callbacks": {
|
| 969 |
+
"TrainerControl": {
|
| 970 |
+
"args": {
|
| 971 |
+
"should_epoch_stop": false,
|
| 972 |
+
"should_evaluate": false,
|
| 973 |
+
"should_log": false,
|
| 974 |
+
"should_save": true,
|
| 975 |
+
"should_training_stop": true
|
| 976 |
+
},
|
| 977 |
+
"attributes": {}
|
| 978 |
+
}
|
| 979 |
+
},
|
| 980 |
+
"total_flos": 207562104419328.0,
|
| 981 |
+
"train_batch_size": 1,
|
| 982 |
+
"trial_name": null,
|
| 983 |
+
"trial_params": null
|
| 984 |
+
}
|
training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:81fe99ed782ca55dead5a35dfca3b7fd4020c233b42b2db4b7d03721505224d1
|
| 3 |
+
size 5585
|