Batch upload part 5
Browse files- nl_tasks/exps/run_ex12/ft/adapter_config.json +18 -0
- nl_tasks/exps/run_ex12/ft/tokenizer.json +0 -0
- nl_tasks/exps/run_ex12/ft/tokenizer.model +3 -0
- nl_tasks/exps/run_ex12/ft2/adapter_config.json +18 -0
- nl_tasks/exps/run_ex12/ft2/adapter_model.bin +3 -0
- nl_tasks/exps/run_ex13/ft/adapter_config.json +18 -0
- nl_tasks/exps/run_ex13/ft/added_tokens.json +3 -0
- nl_tasks/exps/run_ex13/ft/special_tokens_map.json +30 -0
- nl_tasks/exps/run_ex13/ft/tokenizer.json +0 -0
- nl_tasks/exps/run_ex13/ft/tokenizer.model +3 -0
- nl_tasks/exps/run_ex13/ft/tokenizer_config.json +51 -0
- nl_tasks/exps/run_ex13/ft2/adapter_config.json +18 -0
- nl_tasks/exps/run_ex13/ft2/adapter_model.bin +3 -0
- nl_tasks/exps/run_ex13/trainer_state.json +914 -0
- nl_tasks/exps/run_ex13_3ep/ft/adapter_config.json +18 -0
- nl_tasks/exps/run_ex13_3ep/ft/added_tokens.json +3 -0
- nl_tasks/exps/run_ex13_3ep/ft/special_tokens_map.json +30 -0
- nl_tasks/exps/run_ex13_3ep/ft/tokenizer.json +0 -0
- nl_tasks/exps/run_ex13_3ep/ft/tokenizer.model +3 -0
- nl_tasks/exps/run_ex13_3ep/ft/tokenizer_config.json +51 -0
- nl_tasks/exps/run_ex13_3ep/ft2/adapter_config.json +18 -0
- nl_tasks/exps/run_ex13_3ep/ft2/adapter_model.bin +3 -0
- nl_tasks/exps/run_ex13_3ep/trainer_state.json +1209 -0
- nl_tasks/exps/run_ex14_3ep/ft/adapter_config.json +18 -0
- nl_tasks/exps/run_ex14_3ep/ft/added_tokens.json +3 -0
- nl_tasks/exps/run_ex14_3ep/ft/special_tokens_map.json +30 -0
- nl_tasks/exps/run_ex14_3ep/ft/tokenizer.json +0 -0
- nl_tasks/exps/run_ex14_3ep/ft/tokenizer.model +3 -0
- nl_tasks/exps/run_ex14_3ep/ft/tokenizer_config.json +51 -0
- nl_tasks/exps/run_ex14_3ep/ft2/adapter_config.json +18 -0
- nl_tasks/exps/run_ex14_3ep/ft2/adapter_model.bin +3 -0
- nl_tasks/exps/run_ex14_3ep/trainer_state.json +1209 -0
- nl_tasks/exps/run_ex15_3ep/ft/adapter_config.json +18 -0
- nl_tasks/exps/run_ex15_3ep/ft/added_tokens.json +3 -0
- nl_tasks/exps/run_ex15_3ep/ft/special_tokens_map.json +30 -0
- nl_tasks/exps/run_ex15_3ep/ft/tokenizer.json +0 -0
- nl_tasks/exps/run_ex15_3ep/ft/tokenizer.model +3 -0
- nl_tasks/exps/run_ex15_3ep/ft/tokenizer_config.json +51 -0
- nl_tasks/exps/run_ex15_3ep/ft2/adapter_config.json +18 -0
- nl_tasks/exps/run_ex15_3ep/ft2/adapter_model.bin +3 -0
- nl_tasks/exps/run_ex15_3ep/trainer_state.json +1209 -0
- nl_tasks/exps/run_ex16_3ep/ft/adapter_config.json +18 -0
- nl_tasks/exps/run_ex16_3ep/ft/added_tokens.json +3 -0
- nl_tasks/exps/run_ex16_3ep/ft/special_tokens_map.json +30 -0
- nl_tasks/exps/run_ex16_3ep/ft/tokenizer.json +0 -0
- nl_tasks/exps/run_ex16_3ep/ft/tokenizer.model +3 -0
- nl_tasks/exps/run_ex16_3ep/ft/tokenizer_config.json +51 -0
- nl_tasks/exps/run_ex16_3ep/ft2/adapter_config.json +18 -0
- nl_tasks/exps/run_ex16_3ep/ft2/adapter_model.bin +3 -0
- nl_tasks/exps/run_ex16_3ep/trainer_state.json +1209 -0
nl_tasks/exps/run_ex12/ft/adapter_config.json
ADDED
|
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"T": 1.0,
|
| 3 |
+
"base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
|
| 4 |
+
"bias": "none",
|
| 5 |
+
"inference_mode": false,
|
| 6 |
+
"layers_to_transform": null,
|
| 7 |
+
"modules_to_save": null,
|
| 8 |
+
"num_rotations": 1,
|
| 9 |
+
"peft_type": "ROTATION",
|
| 10 |
+
"r": 16,
|
| 11 |
+
"revision": null,
|
| 12 |
+
"target_modules": [
|
| 13 |
+
"v_proj",
|
| 14 |
+
"q_proj"
|
| 15 |
+
],
|
| 16 |
+
"target_modules_to_skip": null,
|
| 17 |
+
"task_type": "CAUSAL_LM"
|
| 18 |
+
}
|
nl_tasks/exps/run_ex12/ft/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
nl_tasks/exps/run_ex12/ft/tokenizer.model
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
|
| 3 |
+
size 499723
|
nl_tasks/exps/run_ex12/ft2/adapter_config.json
ADDED
|
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"T": 1.0,
|
| 3 |
+
"base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
|
| 4 |
+
"bias": "none",
|
| 5 |
+
"inference_mode": true,
|
| 6 |
+
"layers_to_transform": null,
|
| 7 |
+
"modules_to_save": null,
|
| 8 |
+
"num_rotations": 1,
|
| 9 |
+
"peft_type": "ROTATION",
|
| 10 |
+
"r": 16,
|
| 11 |
+
"revision": null,
|
| 12 |
+
"target_modules": [
|
| 13 |
+
"v_proj",
|
| 14 |
+
"q_proj"
|
| 15 |
+
],
|
| 16 |
+
"target_modules_to_skip": null,
|
| 17 |
+
"task_type": "CAUSAL_LM"
|
| 18 |
+
}
|
nl_tasks/exps/run_ex12/ft2/adapter_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:f15a7067acc59209fae49f5b4409bb98642b3c387e6eb4164485aa5cd5a95beb
|
| 3 |
+
size 33602915
|
nl_tasks/exps/run_ex13/ft/adapter_config.json
ADDED
|
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"T": 1.0,
|
| 3 |
+
"base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
|
| 4 |
+
"bias": "none",
|
| 5 |
+
"inference_mode": false,
|
| 6 |
+
"layers_to_transform": null,
|
| 7 |
+
"modules_to_save": null,
|
| 8 |
+
"num_rotations": 1,
|
| 9 |
+
"peft_type": "ROTATION",
|
| 10 |
+
"r": 16,
|
| 11 |
+
"revision": null,
|
| 12 |
+
"target_modules": [
|
| 13 |
+
"q_proj",
|
| 14 |
+
"v_proj"
|
| 15 |
+
],
|
| 16 |
+
"target_modules_to_skip": null,
|
| 17 |
+
"task_type": "CAUSAL_LM"
|
| 18 |
+
}
|
nl_tasks/exps/run_ex13/ft/added_tokens.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"[PAD]": 32000
|
| 3 |
+
}
|
nl_tasks/exps/run_ex13/ft/special_tokens_map.json
ADDED
|
@@ -0,0 +1,30 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": {
|
| 3 |
+
"content": "</s>",
|
| 4 |
+
"lstrip": false,
|
| 5 |
+
"normalized": false,
|
| 6 |
+
"rstrip": false,
|
| 7 |
+
"single_word": false
|
| 8 |
+
},
|
| 9 |
+
"eos_token": {
|
| 10 |
+
"content": "</s>",
|
| 11 |
+
"lstrip": false,
|
| 12 |
+
"normalized": false,
|
| 13 |
+
"rstrip": false,
|
| 14 |
+
"single_word": false
|
| 15 |
+
},
|
| 16 |
+
"pad_token": {
|
| 17 |
+
"content": "[PAD]",
|
| 18 |
+
"lstrip": false,
|
| 19 |
+
"normalized": false,
|
| 20 |
+
"rstrip": false,
|
| 21 |
+
"single_word": false
|
| 22 |
+
},
|
| 23 |
+
"unk_token": {
|
| 24 |
+
"content": "</s>",
|
| 25 |
+
"lstrip": false,
|
| 26 |
+
"normalized": false,
|
| 27 |
+
"rstrip": false,
|
| 28 |
+
"single_word": false
|
| 29 |
+
}
|
| 30 |
+
}
|
nl_tasks/exps/run_ex13/ft/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
nl_tasks/exps/run_ex13/ft/tokenizer.model
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
|
| 3 |
+
size 499723
|
nl_tasks/exps/run_ex13/ft/tokenizer_config.json
ADDED
|
@@ -0,0 +1,51 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_bos_token": true,
|
| 3 |
+
"add_eos_token": false,
|
| 4 |
+
"add_prefix_space": null,
|
| 5 |
+
"added_tokens_decoder": {
|
| 6 |
+
"0": {
|
| 7 |
+
"content": "<unk>",
|
| 8 |
+
"lstrip": false,
|
| 9 |
+
"normalized": false,
|
| 10 |
+
"rstrip": false,
|
| 11 |
+
"single_word": false,
|
| 12 |
+
"special": true
|
| 13 |
+
},
|
| 14 |
+
"1": {
|
| 15 |
+
"content": "<s>",
|
| 16 |
+
"lstrip": false,
|
| 17 |
+
"normalized": false,
|
| 18 |
+
"rstrip": false,
|
| 19 |
+
"single_word": false,
|
| 20 |
+
"special": true
|
| 21 |
+
},
|
| 22 |
+
"2": {
|
| 23 |
+
"content": "</s>",
|
| 24 |
+
"lstrip": false,
|
| 25 |
+
"normalized": false,
|
| 26 |
+
"rstrip": false,
|
| 27 |
+
"single_word": false,
|
| 28 |
+
"special": true
|
| 29 |
+
},
|
| 30 |
+
"32000": {
|
| 31 |
+
"content": "[PAD]",
|
| 32 |
+
"lstrip": false,
|
| 33 |
+
"normalized": false,
|
| 34 |
+
"rstrip": false,
|
| 35 |
+
"single_word": false,
|
| 36 |
+
"special": true
|
| 37 |
+
}
|
| 38 |
+
},
|
| 39 |
+
"bos_token": "</s>",
|
| 40 |
+
"clean_up_tokenization_spaces": false,
|
| 41 |
+
"eos_token": "</s>",
|
| 42 |
+
"extra_special_tokens": {},
|
| 43 |
+
"legacy": false,
|
| 44 |
+
"model_max_length": 512,
|
| 45 |
+
"pad_token": "[PAD]",
|
| 46 |
+
"padding_side": "right",
|
| 47 |
+
"sp_model_kwargs": {},
|
| 48 |
+
"tokenizer_class": "LlamaTokenizer",
|
| 49 |
+
"unk_token": "</s>",
|
| 50 |
+
"use_default_system_prompt": false
|
| 51 |
+
}
|
nl_tasks/exps/run_ex13/ft2/adapter_config.json
ADDED
|
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"T": 1.0,
|
| 3 |
+
"base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
|
| 4 |
+
"bias": "none",
|
| 5 |
+
"inference_mode": true,
|
| 6 |
+
"layers_to_transform": null,
|
| 7 |
+
"modules_to_save": null,
|
| 8 |
+
"num_rotations": 1,
|
| 9 |
+
"peft_type": "ROTATION",
|
| 10 |
+
"r": 16,
|
| 11 |
+
"revision": null,
|
| 12 |
+
"target_modules": [
|
| 13 |
+
"q_proj",
|
| 14 |
+
"v_proj"
|
| 15 |
+
],
|
| 16 |
+
"target_modules_to_skip": null,
|
| 17 |
+
"task_type": "CAUSAL_LM"
|
| 18 |
+
}
|
nl_tasks/exps/run_ex13/ft2/adapter_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:631c0b21e554b83e3404c4f3d4dfaa0e49fb9017ac23829bc45e4b81e3e709ae
|
| 3 |
+
size 33602915
|
nl_tasks/exps/run_ex13/trainer_state.json
ADDED
|
@@ -0,0 +1,914 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 2.0,
|
| 6 |
+
"eval_steps": 100,
|
| 7 |
+
"global_step": 2438,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.020508613617719443,
|
| 14 |
+
"grad_norm": 0.23063355684280396,
|
| 15 |
+
"learning_rate": 9.836065573770491e-05,
|
| 16 |
+
"loss": 0.6812,
|
| 17 |
+
"step": 25
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.04101722723543889,
|
| 21 |
+
"grad_norm": 0.18014760315418243,
|
| 22 |
+
"learning_rate": 0.00020081967213114754,
|
| 23 |
+
"loss": 0.4176,
|
| 24 |
+
"step": 50
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.06152584085315833,
|
| 28 |
+
"grad_norm": 0.1530769020318985,
|
| 29 |
+
"learning_rate": 0.0003032786885245902,
|
| 30 |
+
"loss": 0.3726,
|
| 31 |
+
"step": 75
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.08203445447087777,
|
| 35 |
+
"grad_norm": 0.18688490986824036,
|
| 36 |
+
"learning_rate": 0.0004057377049180328,
|
| 37 |
+
"loss": 0.3352,
|
| 38 |
+
"step": 100
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.08203445447087777,
|
| 42 |
+
"eval_loss": 0.325385183095932,
|
| 43 |
+
"eval_runtime": 19.475,
|
| 44 |
+
"eval_samples_per_second": 51.348,
|
| 45 |
+
"eval_steps_per_second": 0.822,
|
| 46 |
+
"step": 100
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"epoch": 0.10254306808859721,
|
| 50 |
+
"grad_norm": 0.1708248257637024,
|
| 51 |
+
"learning_rate": 0.0005081967213114754,
|
| 52 |
+
"loss": 0.3388,
|
| 53 |
+
"step": 125
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 0.12305168170631665,
|
| 57 |
+
"grad_norm": 0.1983977109193802,
|
| 58 |
+
"learning_rate": 0.000610655737704918,
|
| 59 |
+
"loss": 0.3211,
|
| 60 |
+
"step": 150
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 0.1435602953240361,
|
| 64 |
+
"grad_norm": 0.1805795133113861,
|
| 65 |
+
"learning_rate": 0.0007131147540983607,
|
| 66 |
+
"loss": 0.312,
|
| 67 |
+
"step": 175
|
| 68 |
+
},
|
| 69 |
+
{
|
| 70 |
+
"epoch": 0.16406890894175555,
|
| 71 |
+
"grad_norm": 0.23340001702308655,
|
| 72 |
+
"learning_rate": 0.0008155737704918033,
|
| 73 |
+
"loss": 0.3232,
|
| 74 |
+
"step": 200
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.16406890894175555,
|
| 78 |
+
"eval_loss": 0.3051459491252899,
|
| 79 |
+
"eval_runtime": 19.3022,
|
| 80 |
+
"eval_samples_per_second": 51.808,
|
| 81 |
+
"eval_steps_per_second": 0.829,
|
| 82 |
+
"step": 200
|
| 83 |
+
},
|
| 84 |
+
{
|
| 85 |
+
"epoch": 0.184577522559475,
|
| 86 |
+
"grad_norm": 0.2863401770591736,
|
| 87 |
+
"learning_rate": 0.0009180327868852459,
|
| 88 |
+
"loss": 0.3226,
|
| 89 |
+
"step": 225
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 0.20508613617719443,
|
| 93 |
+
"grad_norm": 0.30703943967819214,
|
| 94 |
+
"learning_rate": 0.0009999871854116063,
|
| 95 |
+
"loss": 0.3001,
|
| 96 |
+
"step": 250
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"epoch": 0.22559474979491387,
|
| 100 |
+
"grad_norm": 0.21251602470874786,
|
| 101 |
+
"learning_rate": 0.0009995387437838027,
|
| 102 |
+
"loss": 0.3027,
|
| 103 |
+
"step": 275
|
| 104 |
+
},
|
| 105 |
+
{
|
| 106 |
+
"epoch": 0.2461033634126333,
|
| 107 |
+
"grad_norm": 0.25236549973487854,
|
| 108 |
+
"learning_rate": 0.000998450229439693,
|
| 109 |
+
"loss": 0.3044,
|
| 110 |
+
"step": 300
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.2461033634126333,
|
| 114 |
+
"eval_loss": 0.29398414492607117,
|
| 115 |
+
"eval_runtime": 19.3425,
|
| 116 |
+
"eval_samples_per_second": 51.7,
|
| 117 |
+
"eval_steps_per_second": 0.827,
|
| 118 |
+
"step": 300
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 0.2666119770303528,
|
| 122 |
+
"grad_norm": 0.2558315098285675,
|
| 123 |
+
"learning_rate": 0.000996723037122612,
|
| 124 |
+
"loss": 0.3077,
|
| 125 |
+
"step": 325
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 0.2871205906480722,
|
| 129 |
+
"grad_norm": 0.384848415851593,
|
| 130 |
+
"learning_rate": 0.0009943593799315263,
|
| 131 |
+
"loss": 0.2991,
|
| 132 |
+
"step": 350
|
| 133 |
+
},
|
| 134 |
+
{
|
| 135 |
+
"epoch": 0.30762920426579166,
|
| 136 |
+
"grad_norm": 0.2492862343788147,
|
| 137 |
+
"learning_rate": 0.0009913622864853324,
|
| 138 |
+
"loss": 0.3093,
|
| 139 |
+
"step": 375
|
| 140 |
+
},
|
| 141 |
+
{
|
| 142 |
+
"epoch": 0.3281378178835111,
|
| 143 |
+
"grad_norm": 0.34253203868865967,
|
| 144 |
+
"learning_rate": 0.0009877355970422024,
|
| 145 |
+
"loss": 0.301,
|
| 146 |
+
"step": 400
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.3281378178835111,
|
| 150 |
+
"eval_loss": 0.28543996810913086,
|
| 151 |
+
"eval_runtime": 19.2681,
|
| 152 |
+
"eval_samples_per_second": 51.899,
|
| 153 |
+
"eval_steps_per_second": 0.83,
|
| 154 |
+
"step": 400
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 0.34864643150123054,
|
| 158 |
+
"grad_norm": 0.23710455000400543,
|
| 159 |
+
"learning_rate": 0.0009834839585789559,
|
| 160 |
+
"loss": 0.304,
|
| 161 |
+
"step": 425
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 0.36915504511895,
|
| 165 |
+
"grad_norm": 0.22950544953346252,
|
| 166 |
+
"learning_rate": 0.000978612818836762,
|
| 167 |
+
"loss": 0.2995,
|
| 168 |
+
"step": 450
|
| 169 |
+
},
|
| 170 |
+
{
|
| 171 |
+
"epoch": 0.3896636587366694,
|
| 172 |
+
"grad_norm": 0.23045484721660614,
|
| 173 |
+
"learning_rate": 0.0009731284193407981,
|
| 174 |
+
"loss": 0.2856,
|
| 175 |
+
"step": 475
|
| 176 |
+
},
|
| 177 |
+
{
|
| 178 |
+
"epoch": 0.41017227235438886,
|
| 179 |
+
"grad_norm": 0.19551247358322144,
|
| 180 |
+
"learning_rate": 0.0009670377874028117,
|
| 181 |
+
"loss": 0.2725,
|
| 182 |
+
"step": 500
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.41017227235438886,
|
| 186 |
+
"eval_loss": 0.2717110514640808,
|
| 187 |
+
"eval_runtime": 19.3136,
|
| 188 |
+
"eval_samples_per_second": 51.777,
|
| 189 |
+
"eval_steps_per_second": 0.828,
|
| 190 |
+
"step": 500
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 0.4306808859721083,
|
| 194 |
+
"grad_norm": 0.22144560515880585,
|
| 195 |
+
"learning_rate": 0.0009603487271168336,
|
| 196 |
+
"loss": 0.2704,
|
| 197 |
+
"step": 525
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 0.45118949958982774,
|
| 201 |
+
"grad_norm": 0.26910459995269775,
|
| 202 |
+
"learning_rate": 0.0009530698093595781,
|
| 203 |
+
"loss": 0.28,
|
| 204 |
+
"step": 550
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 0.4716981132075472,
|
| 208 |
+
"grad_norm": 0.23758633434772491,
|
| 209 |
+
"learning_rate": 0.0009452103608083418,
|
| 210 |
+
"loss": 0.2711,
|
| 211 |
+
"step": 575
|
| 212 |
+
},
|
| 213 |
+
{
|
| 214 |
+
"epoch": 0.4922067268252666,
|
| 215 |
+
"grad_norm": 0.2066866010427475,
|
| 216 |
+
"learning_rate": 0.0009367804519904775,
|
| 217 |
+
"loss": 0.2876,
|
| 218 |
+
"step": 600
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.4922067268252666,
|
| 222 |
+
"eval_loss": 0.2623426020145416,
|
| 223 |
+
"eval_runtime": 19.2854,
|
| 224 |
+
"eval_samples_per_second": 51.853,
|
| 225 |
+
"eval_steps_per_second": 0.83,
|
| 226 |
+
"step": 600
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.5127153404429861,
|
| 230 |
+
"grad_norm": 0.21168355643749237,
|
| 231 |
+
"learning_rate": 0.0009277908843797492,
|
| 232 |
+
"loss": 0.2663,
|
| 233 |
+
"step": 625
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.5332239540607056,
|
| 237 |
+
"grad_norm": 0.2064104527235031,
|
| 238 |
+
"learning_rate": 0.0009182531765561084,
|
| 239 |
+
"loss": 0.2594,
|
| 240 |
+
"step": 650
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.5537325676784249,
|
| 244 |
+
"grad_norm": 0.21850290894508362,
|
| 245 |
+
"learning_rate": 0.0009081795494466201,
|
| 246 |
+
"loss": 0.2706,
|
| 247 |
+
"step": 675
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.5742411812961444,
|
| 251 |
+
"grad_norm": 0.21628743410110474,
|
| 252 |
+
"learning_rate": 0.0008975829106664539,
|
| 253 |
+
"loss": 0.2696,
|
| 254 |
+
"step": 700
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.5742411812961444,
|
| 258 |
+
"eval_loss": 0.2571398913860321,
|
| 259 |
+
"eval_runtime": 19.259,
|
| 260 |
+
"eval_samples_per_second": 51.924,
|
| 261 |
+
"eval_steps_per_second": 0.831,
|
| 262 |
+
"step": 700
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.5947497949138638,
|
| 266 |
+
"grad_norm": 0.19227014482021332,
|
| 267 |
+
"learning_rate": 0.0008864768379800017,
|
| 268 |
+
"loss": 0.2669,
|
| 269 |
+
"step": 725
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.6152584085315833,
|
| 273 |
+
"grad_norm": 0.19298115372657776,
|
| 274 |
+
"learning_rate": 0.0008748755619033153,
|
| 275 |
+
"loss": 0.2628,
|
| 276 |
+
"step": 750
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.6357670221493027,
|
| 280 |
+
"grad_norm": 0.19282013177871704,
|
| 281 |
+
"learning_rate": 0.000862793947470155,
|
| 282 |
+
"loss": 0.2605,
|
| 283 |
+
"step": 775
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.6562756357670222,
|
| 287 |
+
"grad_norm": 0.2024482786655426,
|
| 288 |
+
"learning_rate": 0.0008502474751850142,
|
| 289 |
+
"loss": 0.2632,
|
| 290 |
+
"step": 800
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.6562756357670222,
|
| 294 |
+
"eval_loss": 0.24924395978450775,
|
| 295 |
+
"eval_runtime": 19.2743,
|
| 296 |
+
"eval_samples_per_second": 51.883,
|
| 297 |
+
"eval_steps_per_second": 0.83,
|
| 298 |
+
"step": 800
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 0.6767842493847416,
|
| 302 |
+
"grad_norm": 0.1860484927892685,
|
| 303 |
+
"learning_rate": 0.0008372522211875224,
|
| 304 |
+
"loss": 0.2554,
|
| 305 |
+
"step": 825
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 0.6972928630024611,
|
| 309 |
+
"grad_norm": 0.19293159246444702,
|
| 310 |
+
"learning_rate": 0.0008238248366536473,
|
| 311 |
+
"loss": 0.259,
|
| 312 |
+
"step": 850
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 0.7178014766201805,
|
| 316 |
+
"grad_norm": 0.16820554435253143,
|
| 317 |
+
"learning_rate": 0.0008099825264600842,
|
| 318 |
+
"loss": 0.2458,
|
| 319 |
+
"step": 875
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 0.7383100902379,
|
| 323 |
+
"grad_norm": 0.20481383800506592,
|
| 324 |
+
"learning_rate": 0.0007957430271391761,
|
| 325 |
+
"loss": 0.2525,
|
| 326 |
+
"step": 900
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 0.7383100902379,
|
| 330 |
+
"eval_loss": 0.24079462885856628,
|
| 331 |
+
"eval_runtime": 19.3447,
|
| 332 |
+
"eval_samples_per_second": 51.694,
|
| 333 |
+
"eval_steps_per_second": 0.827,
|
| 334 |
+
"step": 900
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"epoch": 0.7588187038556193,
|
| 338 |
+
"grad_norm": 0.18768063187599182,
|
| 339 |
+
"learning_rate": 0.0007811245841526062,
|
| 340 |
+
"loss": 0.2535,
|
| 341 |
+
"step": 925
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 0.7793273174733388,
|
| 345 |
+
"grad_norm": 0.17618238925933838,
|
| 346 |
+
"learning_rate": 0.0007661459285129879,
|
| 347 |
+
"loss": 0.241,
|
| 348 |
+
"step": 950
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"epoch": 0.7998359310910582,
|
| 352 |
+
"grad_norm": 0.20532366633415222,
|
| 353 |
+
"learning_rate": 0.0007508262527833029,
|
| 354 |
+
"loss": 0.2475,
|
| 355 |
+
"step": 975
|
| 356 |
+
},
|
| 357 |
+
{
|
| 358 |
+
"epoch": 0.8203445447087777,
|
| 359 |
+
"grad_norm": 0.18020130693912506,
|
| 360 |
+
"learning_rate": 0.000735185186484943,
|
| 361 |
+
"loss": 0.2401,
|
| 362 |
+
"step": 1000
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.8203445447087777,
|
| 366 |
+
"eval_loss": 0.23719018697738647,
|
| 367 |
+
"eval_runtime": 19.3631,
|
| 368 |
+
"eval_samples_per_second": 51.645,
|
| 369 |
+
"eval_steps_per_second": 0.826,
|
| 370 |
+
"step": 1000
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 0.8408531583264971,
|
| 374 |
+
"grad_norm": 0.1882888227701187,
|
| 375 |
+
"learning_rate": 0.0007192427709458656,
|
| 376 |
+
"loss": 0.2409,
|
| 377 |
+
"step": 1025
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 0.8613617719442166,
|
| 381 |
+
"grad_norm": 0.17976818978786469,
|
| 382 |
+
"learning_rate": 0.0007030194336210887,
|
| 383 |
+
"loss": 0.2381,
|
| 384 |
+
"step": 1050
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.881870385561936,
|
| 388 |
+
"grad_norm": 0.1889801174402237,
|
| 389 |
+
"learning_rate": 0.0006865359619184331,
|
| 390 |
+
"loss": 0.2432,
|
| 391 |
+
"step": 1075
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 0.9023789991796555,
|
| 395 |
+
"grad_norm": 0.178939089179039,
|
| 396 |
+
"learning_rate": 0.0006698134765630434,
|
| 397 |
+
"loss": 0.244,
|
| 398 |
+
"step": 1100
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 0.9023789991796555,
|
| 402 |
+
"eval_loss": 0.23325777053833008,
|
| 403 |
+
"eval_runtime": 19.3754,
|
| 404 |
+
"eval_samples_per_second": 51.612,
|
| 405 |
+
"eval_steps_per_second": 0.826,
|
| 406 |
+
"step": 1100
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 0.9228876127973749,
|
| 410 |
+
"grad_norm": 0.17410777509212494,
|
| 411 |
+
"learning_rate": 0.0006528734045348248,
|
| 412 |
+
"loss": 0.2375,
|
| 413 |
+
"step": 1125
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 0.9433962264150944,
|
| 417 |
+
"grad_norm": 0.15016716718673706,
|
| 418 |
+
"learning_rate": 0.0006357374516134643,
|
| 419 |
+
"loss": 0.2371,
|
| 420 |
+
"step": 1150
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 0.9639048400328137,
|
| 424 |
+
"grad_norm": 0.1614176481962204,
|
| 425 |
+
"learning_rate": 0.0006184275745662179,
|
| 426 |
+
"loss": 0.2429,
|
| 427 |
+
"step": 1175
|
| 428 |
+
},
|
| 429 |
+
{
|
| 430 |
+
"epoch": 0.9844134536505332,
|
| 431 |
+
"grad_norm": 0.1717890501022339,
|
| 432 |
+
"learning_rate": 0.0006009659530141031,
|
| 433 |
+
"loss": 0.2478,
|
| 434 |
+
"step": 1200
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 0.9844134536505332,
|
| 438 |
+
"eval_loss": 0.22848205268383026,
|
| 439 |
+
"eval_runtime": 19.3521,
|
| 440 |
+
"eval_samples_per_second": 51.674,
|
| 441 |
+
"eval_steps_per_second": 0.827,
|
| 442 |
+
"step": 1200
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 1.0049220672682526,
|
| 446 |
+
"grad_norm": 0.153545543551445,
|
| 447 |
+
"learning_rate": 0.0005833749610125402,
|
| 448 |
+
"loss": 0.2277,
|
| 449 |
+
"step": 1225
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 1.0254306808859721,
|
| 453 |
+
"grad_norm": 0.16761425137519836,
|
| 454 |
+
"learning_rate": 0.0005656771383828602,
|
| 455 |
+
"loss": 0.2067,
|
| 456 |
+
"step": 1250
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 1.0459392945036916,
|
| 460 |
+
"grad_norm": 0.19482481479644775,
|
| 461 |
+
"learning_rate": 0.0005478951618314134,
|
| 462 |
+
"loss": 0.2074,
|
| 463 |
+
"step": 1275
|
| 464 |
+
},
|
| 465 |
+
{
|
| 466 |
+
"epoch": 1.066447908121411,
|
| 467 |
+
"grad_norm": 0.15821483731269836,
|
| 468 |
+
"learning_rate": 0.0005300518158932815,
|
| 469 |
+
"loss": 0.2104,
|
| 470 |
+
"step": 1300
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 1.066447908121411,
|
| 474 |
+
"eval_loss": 0.2253958135843277,
|
| 475 |
+
"eval_runtime": 19.2878,
|
| 476 |
+
"eval_samples_per_second": 51.846,
|
| 477 |
+
"eval_steps_per_second": 0.83,
|
| 478 |
+
"step": 1300
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 1.0869565217391304,
|
| 482 |
+
"grad_norm": 0.18940797448158264,
|
| 483 |
+
"learning_rate": 0.0005121699637378282,
|
| 484 |
+
"loss": 0.2073,
|
| 485 |
+
"step": 1325
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 1.1074651353568499,
|
| 489 |
+
"grad_norm": 0.1431218981742859,
|
| 490 |
+
"learning_rate": 0.0004942725178734903,
|
| 491 |
+
"loss": 0.2004,
|
| 492 |
+
"step": 1350
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 1.1279737489745694,
|
| 496 |
+
"grad_norm": 0.18451210856437683,
|
| 497 |
+
"learning_rate": 0.00047638241078935324,
|
| 498 |
+
"loss": 0.2017,
|
| 499 |
+
"step": 1375
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 1.1484823625922886,
|
| 503 |
+
"grad_norm": 0.16710275411605835,
|
| 504 |
+
"learning_rate": 0.000458522565571121,
|
| 505 |
+
"loss": 0.1947,
|
| 506 |
+
"step": 1400
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 1.1484823625922886,
|
| 510 |
+
"eval_loss": 0.2227575033903122,
|
| 511 |
+
"eval_runtime": 19.3385,
|
| 512 |
+
"eval_samples_per_second": 51.71,
|
| 513 |
+
"eval_steps_per_second": 0.827,
|
| 514 |
+
"step": 1400
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 1.1689909762100081,
|
| 518 |
+
"grad_norm": 0.17341965436935425,
|
| 519 |
+
"learning_rate": 0.00044071586652913767,
|
| 520 |
+
"loss": 0.2063,
|
| 521 |
+
"step": 1425
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 1.1894995898277276,
|
| 525 |
+
"grad_norm": 0.18759435415267944,
|
| 526 |
+
"learning_rate": 0.0004229851298760915,
|
| 527 |
+
"loss": 0.2057,
|
| 528 |
+
"step": 1450
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 1.2100082034454471,
|
| 532 |
+
"grad_norm": 0.18075019121170044,
|
| 533 |
+
"learning_rate": 0.0004053530744919749,
|
| 534 |
+
"loss": 0.2018,
|
| 535 |
+
"step": 1475
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 1.2305168170631666,
|
| 539 |
+
"grad_norm": 0.1759309619665146,
|
| 540 |
+
"learning_rate": 0.0003878422928137597,
|
| 541 |
+
"loss": 0.211,
|
| 542 |
+
"step": 1500
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 1.2305168170631666,
|
| 546 |
+
"eval_loss": 0.2198861539363861,
|
| 547 |
+
"eval_runtime": 19.3184,
|
| 548 |
+
"eval_samples_per_second": 51.764,
|
| 549 |
+
"eval_steps_per_second": 0.828,
|
| 550 |
+
"step": 1500
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 1.251025430680886,
|
| 554 |
+
"grad_norm": 0.20878763496875763,
|
| 555 |
+
"learning_rate": 0.0003704752218870861,
|
| 556 |
+
"loss": 0.2064,
|
| 557 |
+
"step": 1525
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 1.2715340442986054,
|
| 561 |
+
"grad_norm": 0.14704816043376923,
|
| 562 |
+
"learning_rate": 0.00035327411461706025,
|
| 563 |
+
"loss": 0.2,
|
| 564 |
+
"step": 1550
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 1.2920426579163249,
|
| 568 |
+
"grad_norm": 0.14631816744804382,
|
| 569 |
+
"learning_rate": 0.00033626101125499555,
|
| 570 |
+
"loss": 0.201,
|
| 571 |
+
"step": 1575
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 1.3125512715340442,
|
| 575 |
+
"grad_norm": 0.16362226009368896,
|
| 576 |
+
"learning_rate": 0.0003194577111576333,
|
| 577 |
+
"loss": 0.1944,
|
| 578 |
+
"step": 1600
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 1.3125512715340442,
|
| 582 |
+
"eval_loss": 0.21637172996997833,
|
| 583 |
+
"eval_runtime": 19.3279,
|
| 584 |
+
"eval_samples_per_second": 51.739,
|
| 585 |
+
"eval_steps_per_second": 0.828,
|
| 586 |
+
"step": 1600
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 1.3330598851517639,
|
| 590 |
+
"grad_norm": 0.1545446664094925,
|
| 591 |
+
"learning_rate": 0.00030288574485502756,
|
| 592 |
+
"loss": 0.1967,
|
| 593 |
+
"step": 1625
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 1.3535684987694832,
|
| 597 |
+
"grad_norm": 0.1899978667497635,
|
| 598 |
+
"learning_rate": 0.00028656634646288565,
|
| 599 |
+
"loss": 0.1975,
|
| 600 |
+
"step": 1650
|
| 601 |
+
},
|
| 602 |
+
{
|
| 603 |
+
"epoch": 1.3740771123872026,
|
| 604 |
+
"grad_norm": 0.18251560628414154,
|
| 605 |
+
"learning_rate": 0.00027052042647471254,
|
| 606 |
+
"loss": 0.1947,
|
| 607 |
+
"step": 1675
|
| 608 |
+
},
|
| 609 |
+
{
|
| 610 |
+
"epoch": 1.3945857260049221,
|
| 611 |
+
"grad_norm": 0.15624067187309265,
|
| 612 |
+
"learning_rate": 0.0002547685449686206,
|
| 613 |
+
"loss": 0.1955,
|
| 614 |
+
"step": 1700
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 1.3945857260049221,
|
| 618 |
+
"eval_loss": 0.21375645697116852,
|
| 619 |
+
"eval_runtime": 19.3278,
|
| 620 |
+
"eval_samples_per_second": 51.739,
|
| 621 |
+
"eval_steps_per_second": 0.828,
|
| 622 |
+
"step": 1700
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"epoch": 1.4150943396226414,
|
| 626 |
+
"grad_norm": 0.16290026903152466,
|
| 627 |
+
"learning_rate": 0.0002393308852631373,
|
| 628 |
+
"loss": 0.1929,
|
| 629 |
+
"step": 1725
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 1.435602953240361,
|
| 633 |
+
"grad_norm": 0.15477831661701202,
|
| 634 |
+
"learning_rate": 0.0002242272280557645,
|
| 635 |
+
"loss": 0.1921,
|
| 636 |
+
"step": 1750
|
| 637 |
+
},
|
| 638 |
+
{
|
| 639 |
+
"epoch": 1.4561115668580804,
|
| 640 |
+
"grad_norm": 0.1626901626586914,
|
| 641 |
+
"learning_rate": 0.0002094769260774262,
|
| 642 |
+
"loss": 0.1982,
|
| 643 |
+
"step": 1775
|
| 644 |
+
},
|
| 645 |
+
{
|
| 646 |
+
"epoch": 1.4766201804758,
|
| 647 |
+
"grad_norm": 0.1413515955209732,
|
| 648 |
+
"learning_rate": 0.00019509887929528458,
|
| 649 |
+
"loss": 0.1878,
|
| 650 |
+
"step": 1800
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 1.4766201804758,
|
| 654 |
+
"eval_loss": 0.21210877597332,
|
| 655 |
+
"eval_runtime": 19.2688,
|
| 656 |
+
"eval_samples_per_second": 51.897,
|
| 657 |
+
"eval_steps_per_second": 0.83,
|
| 658 |
+
"step": 1800
|
| 659 |
+
},
|
| 660 |
+
{
|
| 661 |
+
"epoch": 1.4971287940935194,
|
| 662 |
+
"grad_norm": 0.2152208685874939,
|
| 663 |
+
"learning_rate": 0.0001811115106956918,
|
| 664 |
+
"loss": 0.1879,
|
| 665 |
+
"step": 1825
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 1.5176374077112387,
|
| 669 |
+
"grad_norm": 0.14923618733882904,
|
| 670 |
+
"learning_rate": 0.00016753274267831115,
|
| 671 |
+
"loss": 0.1891,
|
| 672 |
+
"step": 1850
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 1.5381460213289582,
|
| 676 |
+
"grad_norm": 0.1509970724582672,
|
| 677 |
+
"learning_rate": 0.00015437997409165478,
|
| 678 |
+
"loss": 0.1904,
|
| 679 |
+
"step": 1875
|
| 680 |
+
},
|
| 681 |
+
{
|
| 682 |
+
"epoch": 1.5586546349466777,
|
| 683 |
+
"grad_norm": 0.16167961061000824,
|
| 684 |
+
"learning_rate": 0.00014167005793946035,
|
| 685 |
+
"loss": 0.1933,
|
| 686 |
+
"step": 1900
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 1.5586546349466777,
|
| 690 |
+
"eval_loss": 0.209316223859787,
|
| 691 |
+
"eval_runtime": 19.2819,
|
| 692 |
+
"eval_samples_per_second": 51.862,
|
| 693 |
+
"eval_steps_per_second": 0.83,
|
| 694 |
+
"step": 1900
|
| 695 |
+
},
|
| 696 |
+
{
|
| 697 |
+
"epoch": 1.579163248564397,
|
| 698 |
+
"grad_norm": 0.154046893119812,
|
| 699 |
+
"learning_rate": 0.00012941927978647527,
|
| 700 |
+
"loss": 0.1942,
|
| 701 |
+
"step": 1925
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"epoch": 1.5996718621821167,
|
| 705 |
+
"grad_norm": 0.1619158685207367,
|
| 706 |
+
"learning_rate": 0.00011764333689131385,
|
| 707 |
+
"loss": 0.1928,
|
| 708 |
+
"step": 1950
|
| 709 |
+
},
|
| 710 |
+
{
|
| 711 |
+
"epoch": 1.620180475799836,
|
| 712 |
+
"grad_norm": 0.15459883213043213,
|
| 713 |
+
"learning_rate": 0.00010635731809312993,
|
| 714 |
+
"loss": 0.1877,
|
| 715 |
+
"step": 1975
|
| 716 |
+
},
|
| 717 |
+
{
|
| 718 |
+
"epoch": 1.6406890894175554,
|
| 719 |
+
"grad_norm": 0.16406521201133728,
|
| 720 |
+
"learning_rate": 9.557568447787201e-05,
|
| 721 |
+
"loss": 0.195,
|
| 722 |
+
"step": 2000
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 1.6406890894175554,
|
| 726 |
+
"eval_loss": 0.20751364529132843,
|
| 727 |
+
"eval_runtime": 19.2815,
|
| 728 |
+
"eval_samples_per_second": 51.863,
|
| 729 |
+
"eval_steps_per_second": 0.83,
|
| 730 |
+
"step": 2000
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 1.661197703035275,
|
| 734 |
+
"grad_norm": 0.1447853147983551,
|
| 735 |
+
"learning_rate": 8.531225084889654e-05,
|
| 736 |
+
"loss": 0.1848,
|
| 737 |
+
"step": 2025
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 1.6817063166529942,
|
| 741 |
+
"grad_norm": 0.17156699299812317,
|
| 742 |
+
"learning_rate": 7.558016802568091e-05,
|
| 743 |
+
"loss": 0.1859,
|
| 744 |
+
"step": 2050
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 1.7022149302707137,
|
| 748 |
+
"grad_norm": 0.20112192630767822,
|
| 749 |
+
"learning_rate": 6.639190599331746e-05,
|
| 750 |
+
"loss": 0.1919,
|
| 751 |
+
"step": 2075
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 1.7227235438884332,
|
| 755 |
+
"grad_norm": 0.15536586940288544,
|
| 756 |
+
"learning_rate": 5.775923792437865e-05,
|
| 757 |
+
"loss": 0.1822,
|
| 758 |
+
"step": 2100
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 1.7227235438884332,
|
| 762 |
+
"eval_loss": 0.20659959316253662,
|
| 763 |
+
"eval_runtime": 19.2769,
|
| 764 |
+
"eval_samples_per_second": 51.876,
|
| 765 |
+
"eval_steps_per_second": 0.83,
|
| 766 |
+
"step": 2100
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 1.7432321575061525,
|
| 770 |
+
"grad_norm": 0.14990898966789246,
|
| 771 |
+
"learning_rate": 4.9693225093627616e-05,
|
| 772 |
+
"loss": 0.1861,
|
| 773 |
+
"step": 2125
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 1.7637407711238722,
|
| 777 |
+
"grad_norm": 0.13435500860214233,
|
| 778 |
+
"learning_rate": 4.220420270490294e-05,
|
| 779 |
+
"loss": 0.1839,
|
| 780 |
+
"step": 2150
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 1.7842493847415914,
|
| 784 |
+
"grad_norm": 0.15417076647281647,
|
| 785 |
+
"learning_rate": 3.530176664833834e-05,
|
| 786 |
+
"loss": 0.1928,
|
| 787 |
+
"step": 2175
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 1.804757998359311,
|
| 791 |
+
"grad_norm": 0.15742765367031097,
|
| 792 |
+
"learning_rate": 2.8994761204884756e-05,
|
| 793 |
+
"loss": 0.1868,
|
| 794 |
+
"step": 2200
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 1.804757998359311,
|
| 798 |
+
"eval_loss": 0.20532982051372528,
|
| 799 |
+
"eval_runtime": 19.3321,
|
| 800 |
+
"eval_samples_per_second": 51.728,
|
| 801 |
+
"eval_steps_per_second": 0.828,
|
| 802 |
+
"step": 2200
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 1.8252666119770304,
|
| 806 |
+
"grad_norm": 0.1742992252111435,
|
| 807 |
+
"learning_rate": 2.329126771388995e-05,
|
| 808 |
+
"loss": 0.1868,
|
| 809 |
+
"step": 2225
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 1.8457752255947497,
|
| 813 |
+
"grad_norm": 0.16154895722866058,
|
| 814 |
+
"learning_rate": 1.8198594218256815e-05,
|
| 815 |
+
"loss": 0.1891,
|
| 816 |
+
"step": 2250
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"epoch": 1.8662838392124692,
|
| 820 |
+
"grad_norm": 0.14913508296012878,
|
| 821 |
+
"learning_rate": 1.3723266100447052e-05,
|
| 822 |
+
"loss": 0.1932,
|
| 823 |
+
"step": 2275
|
| 824 |
+
},
|
| 825 |
+
{
|
| 826 |
+
"epoch": 1.8867924528301887,
|
| 827 |
+
"grad_norm": 0.15223710238933563,
|
| 828 |
+
"learning_rate": 9.871017721329201e-06,
|
| 829 |
+
"loss": 0.1857,
|
| 830 |
+
"step": 2300
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 1.8867924528301887,
|
| 834 |
+
"eval_loss": 0.20510254800319672,
|
| 835 |
+
"eval_runtime": 19.3501,
|
| 836 |
+
"eval_samples_per_second": 51.679,
|
| 837 |
+
"eval_steps_per_second": 0.827,
|
| 838 |
+
"step": 2300
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 1.907301066447908,
|
| 842 |
+
"grad_norm": 0.17794349789619446,
|
| 843 |
+
"learning_rate": 6.646785072584871e-06,
|
| 844 |
+
"loss": 0.194,
|
| 845 |
+
"step": 2325
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 1.9278096800656277,
|
| 849 |
+
"grad_norm": 0.18157444894313812,
|
| 850 |
+
"learning_rate": 4.054699452086641e-06,
|
| 851 |
+
"loss": 0.1792,
|
| 852 |
+
"step": 2350
|
| 853 |
+
},
|
| 854 |
+
{
|
| 855 |
+
"epoch": 1.948318293683347,
|
| 856 |
+
"grad_norm": 0.1601715087890625,
|
| 857 |
+
"learning_rate": 2.0980821703527886e-06,
|
| 858 |
+
"loss": 0.1848,
|
| 859 |
+
"step": 2375
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 1.9688269073010665,
|
| 863 |
+
"grad_norm": 0.14718155562877655,
|
| 864 |
+
"learning_rate": 7.794402948607671e-07,
|
| 865 |
+
"loss": 0.1862,
|
| 866 |
+
"step": 2400
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 1.9688269073010665,
|
| 870 |
+
"eval_loss": 0.20495718717575073,
|
| 871 |
+
"eval_runtime": 19.2663,
|
| 872 |
+
"eval_samples_per_second": 51.904,
|
| 873 |
+
"eval_steps_per_second": 0.83,
|
| 874 |
+
"step": 2400
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"epoch": 1.989335520918786,
|
| 878 |
+
"grad_norm": 0.15204986929893494,
|
| 879 |
+
"learning_rate": 1.0046343767294853e-07,
|
| 880 |
+
"loss": 0.1818,
|
| 881 |
+
"step": 2425
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 2.0,
|
| 885 |
+
"step": 2438,
|
| 886 |
+
"total_flos": 1.58523627405312e+18,
|
| 887 |
+
"train_loss": 0.24099373250246245,
|
| 888 |
+
"train_runtime": 2702.6542,
|
| 889 |
+
"train_samples_per_second": 28.861,
|
| 890 |
+
"train_steps_per_second": 0.902
|
| 891 |
+
}
|
| 892 |
+
],
|
| 893 |
+
"logging_steps": 25,
|
| 894 |
+
"max_steps": 2438,
|
| 895 |
+
"num_input_tokens_seen": 0,
|
| 896 |
+
"num_train_epochs": 2,
|
| 897 |
+
"save_steps": 0,
|
| 898 |
+
"stateful_callbacks": {
|
| 899 |
+
"TrainerControl": {
|
| 900 |
+
"args": {
|
| 901 |
+
"should_epoch_stop": false,
|
| 902 |
+
"should_evaluate": false,
|
| 903 |
+
"should_log": false,
|
| 904 |
+
"should_save": true,
|
| 905 |
+
"should_training_stop": true
|
| 906 |
+
},
|
| 907 |
+
"attributes": {}
|
| 908 |
+
}
|
| 909 |
+
},
|
| 910 |
+
"total_flos": 1.58523627405312e+18,
|
| 911 |
+
"train_batch_size": 32,
|
| 912 |
+
"trial_name": null,
|
| 913 |
+
"trial_params": null
|
| 914 |
+
}
|
nl_tasks/exps/run_ex13_3ep/ft/adapter_config.json
ADDED
|
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"T": 1.0,
|
| 3 |
+
"base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
|
| 4 |
+
"bias": "none",
|
| 5 |
+
"inference_mode": false,
|
| 6 |
+
"layers_to_transform": null,
|
| 7 |
+
"modules_to_save": null,
|
| 8 |
+
"num_rotations": 1,
|
| 9 |
+
"peft_type": "ROTATION",
|
| 10 |
+
"r": 16,
|
| 11 |
+
"revision": null,
|
| 12 |
+
"target_modules": [
|
| 13 |
+
"q_proj",
|
| 14 |
+
"v_proj"
|
| 15 |
+
],
|
| 16 |
+
"target_modules_to_skip": null,
|
| 17 |
+
"task_type": "CAUSAL_LM"
|
| 18 |
+
}
|
nl_tasks/exps/run_ex13_3ep/ft/added_tokens.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"[PAD]": 32000
|
| 3 |
+
}
|
nl_tasks/exps/run_ex13_3ep/ft/special_tokens_map.json
ADDED
|
@@ -0,0 +1,30 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": {
|
| 3 |
+
"content": "</s>",
|
| 4 |
+
"lstrip": false,
|
| 5 |
+
"normalized": false,
|
| 6 |
+
"rstrip": false,
|
| 7 |
+
"single_word": false
|
| 8 |
+
},
|
| 9 |
+
"eos_token": {
|
| 10 |
+
"content": "</s>",
|
| 11 |
+
"lstrip": false,
|
| 12 |
+
"normalized": false,
|
| 13 |
+
"rstrip": false,
|
| 14 |
+
"single_word": false
|
| 15 |
+
},
|
| 16 |
+
"pad_token": {
|
| 17 |
+
"content": "[PAD]",
|
| 18 |
+
"lstrip": false,
|
| 19 |
+
"normalized": false,
|
| 20 |
+
"rstrip": false,
|
| 21 |
+
"single_word": false
|
| 22 |
+
},
|
| 23 |
+
"unk_token": {
|
| 24 |
+
"content": "</s>",
|
| 25 |
+
"lstrip": false,
|
| 26 |
+
"normalized": false,
|
| 27 |
+
"rstrip": false,
|
| 28 |
+
"single_word": false
|
| 29 |
+
}
|
| 30 |
+
}
|
nl_tasks/exps/run_ex13_3ep/ft/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
nl_tasks/exps/run_ex13_3ep/ft/tokenizer.model
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
|
| 3 |
+
size 499723
|
nl_tasks/exps/run_ex13_3ep/ft/tokenizer_config.json
ADDED
|
@@ -0,0 +1,51 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_bos_token": true,
|
| 3 |
+
"add_eos_token": false,
|
| 4 |
+
"add_prefix_space": null,
|
| 5 |
+
"added_tokens_decoder": {
|
| 6 |
+
"0": {
|
| 7 |
+
"content": "<unk>",
|
| 8 |
+
"lstrip": false,
|
| 9 |
+
"normalized": false,
|
| 10 |
+
"rstrip": false,
|
| 11 |
+
"single_word": false,
|
| 12 |
+
"special": true
|
| 13 |
+
},
|
| 14 |
+
"1": {
|
| 15 |
+
"content": "<s>",
|
| 16 |
+
"lstrip": false,
|
| 17 |
+
"normalized": false,
|
| 18 |
+
"rstrip": false,
|
| 19 |
+
"single_word": false,
|
| 20 |
+
"special": true
|
| 21 |
+
},
|
| 22 |
+
"2": {
|
| 23 |
+
"content": "</s>",
|
| 24 |
+
"lstrip": false,
|
| 25 |
+
"normalized": false,
|
| 26 |
+
"rstrip": false,
|
| 27 |
+
"single_word": false,
|
| 28 |
+
"special": true
|
| 29 |
+
},
|
| 30 |
+
"32000": {
|
| 31 |
+
"content": "[PAD]",
|
| 32 |
+
"lstrip": false,
|
| 33 |
+
"normalized": false,
|
| 34 |
+
"rstrip": false,
|
| 35 |
+
"single_word": false,
|
| 36 |
+
"special": true
|
| 37 |
+
}
|
| 38 |
+
},
|
| 39 |
+
"bos_token": "</s>",
|
| 40 |
+
"clean_up_tokenization_spaces": false,
|
| 41 |
+
"eos_token": "</s>",
|
| 42 |
+
"extra_special_tokens": {},
|
| 43 |
+
"legacy": false,
|
| 44 |
+
"model_max_length": 512,
|
| 45 |
+
"pad_token": "[PAD]",
|
| 46 |
+
"padding_side": "right",
|
| 47 |
+
"sp_model_kwargs": {},
|
| 48 |
+
"tokenizer_class": "LlamaTokenizer",
|
| 49 |
+
"unk_token": "</s>",
|
| 50 |
+
"use_default_system_prompt": false
|
| 51 |
+
}
|
nl_tasks/exps/run_ex13_3ep/ft2/adapter_config.json
ADDED
|
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"T": 1.0,
|
| 3 |
+
"base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
|
| 4 |
+
"bias": "none",
|
| 5 |
+
"inference_mode": true,
|
| 6 |
+
"layers_to_transform": null,
|
| 7 |
+
"modules_to_save": null,
|
| 8 |
+
"num_rotations": 1,
|
| 9 |
+
"peft_type": "ROTATION",
|
| 10 |
+
"r": 16,
|
| 11 |
+
"revision": null,
|
| 12 |
+
"target_modules": [
|
| 13 |
+
"q_proj",
|
| 14 |
+
"v_proj"
|
| 15 |
+
],
|
| 16 |
+
"target_modules_to_skip": null,
|
| 17 |
+
"task_type": "CAUSAL_LM"
|
| 18 |
+
}
|
nl_tasks/exps/run_ex13_3ep/ft2/adapter_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:658dc7a6f10546312e15dafec6425b0d7e38ba5698daa76abe2135fcbe7df5d1
|
| 3 |
+
size 33602915
|
nl_tasks/exps/run_ex13_3ep/trainer_state.json
ADDED
|
@@ -0,0 +1,1209 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 3.0,
|
| 6 |
+
"eval_steps": 200,
|
| 7 |
+
"global_step": 3657,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.020508613617719443,
|
| 14 |
+
"grad_norm": 0.26317867636680603,
|
| 15 |
+
"learning_rate": 6.557377049180328e-05,
|
| 16 |
+
"loss": 0.709,
|
| 17 |
+
"step": 25
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.04101722723543889,
|
| 21 |
+
"grad_norm": 0.2067350596189499,
|
| 22 |
+
"learning_rate": 0.00013387978142076503,
|
| 23 |
+
"loss": 0.4475,
|
| 24 |
+
"step": 50
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.06152584085315833,
|
| 28 |
+
"grad_norm": 0.15732155740261078,
|
| 29 |
+
"learning_rate": 0.00020218579234972678,
|
| 30 |
+
"loss": 0.3862,
|
| 31 |
+
"step": 75
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.08203445447087777,
|
| 35 |
+
"grad_norm": 0.22207772731781006,
|
| 36 |
+
"learning_rate": 0.00027049180327868856,
|
| 37 |
+
"loss": 0.3441,
|
| 38 |
+
"step": 100
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.10254306808859721,
|
| 42 |
+
"grad_norm": 0.17426303029060364,
|
| 43 |
+
"learning_rate": 0.00033879781420765025,
|
| 44 |
+
"loss": 0.3462,
|
| 45 |
+
"step": 125
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.12305168170631665,
|
| 49 |
+
"grad_norm": 0.17706584930419922,
|
| 50 |
+
"learning_rate": 0.00040710382513661205,
|
| 51 |
+
"loss": 0.3274,
|
| 52 |
+
"step": 150
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.1435602953240361,
|
| 56 |
+
"grad_norm": 0.15795481204986572,
|
| 57 |
+
"learning_rate": 0.00047540983606557375,
|
| 58 |
+
"loss": 0.3171,
|
| 59 |
+
"step": 175
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.16406890894175555,
|
| 63 |
+
"grad_norm": 0.2113538384437561,
|
| 64 |
+
"learning_rate": 0.0005437158469945356,
|
| 65 |
+
"loss": 0.3253,
|
| 66 |
+
"step": 200
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.16406890894175555,
|
| 70 |
+
"eval_loss": 0.3043476641178131,
|
| 71 |
+
"eval_runtime": 19.5568,
|
| 72 |
+
"eval_samples_per_second": 51.133,
|
| 73 |
+
"eval_steps_per_second": 0.818,
|
| 74 |
+
"step": 200
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.184577522559475,
|
| 78 |
+
"grad_norm": 0.1723247766494751,
|
| 79 |
+
"learning_rate": 0.0006120218579234972,
|
| 80 |
+
"loss": 0.3192,
|
| 81 |
+
"step": 225
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.20508613617719443,
|
| 85 |
+
"grad_norm": 0.19019053876399994,
|
| 86 |
+
"learning_rate": 0.000680327868852459,
|
| 87 |
+
"loss": 0.2976,
|
| 88 |
+
"step": 250
|
| 89 |
+
},
|
| 90 |
+
{
|
| 91 |
+
"epoch": 0.22559474979491387,
|
| 92 |
+
"grad_norm": 0.21279513835906982,
|
| 93 |
+
"learning_rate": 0.0007486338797814209,
|
| 94 |
+
"loss": 0.3061,
|
| 95 |
+
"step": 275
|
| 96 |
+
},
|
| 97 |
+
{
|
| 98 |
+
"epoch": 0.2461033634126333,
|
| 99 |
+
"grad_norm": 0.23976510763168335,
|
| 100 |
+
"learning_rate": 0.0008169398907103825,
|
| 101 |
+
"loss": 0.3044,
|
| 102 |
+
"step": 300
|
| 103 |
+
},
|
| 104 |
+
{
|
| 105 |
+
"epoch": 0.2666119770303528,
|
| 106 |
+
"grad_norm": 0.22460055351257324,
|
| 107 |
+
"learning_rate": 0.0008852459016393442,
|
| 108 |
+
"loss": 0.3072,
|
| 109 |
+
"step": 325
|
| 110 |
+
},
|
| 111 |
+
{
|
| 112 |
+
"epoch": 0.2871205906480722,
|
| 113 |
+
"grad_norm": 0.23715358972549438,
|
| 114 |
+
"learning_rate": 0.0009535519125683059,
|
| 115 |
+
"loss": 0.3017,
|
| 116 |
+
"step": 350
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.30762920426579166,
|
| 120 |
+
"grad_norm": 0.23479431867599487,
|
| 121 |
+
"learning_rate": 0.0009999854198546752,
|
| 122 |
+
"loss": 0.3159,
|
| 123 |
+
"step": 375
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 0.3281378178835111,
|
| 127 |
+
"grad_norm": 0.3158360421657562,
|
| 128 |
+
"learning_rate": 0.0009997519290247507,
|
| 129 |
+
"loss": 0.3025,
|
| 130 |
+
"step": 400
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 0.3281378178835111,
|
| 134 |
+
"eval_loss": 0.2903132438659668,
|
| 135 |
+
"eval_runtime": 19.3059,
|
| 136 |
+
"eval_samples_per_second": 51.798,
|
| 137 |
+
"eval_steps_per_second": 0.829,
|
| 138 |
+
"step": 400
|
| 139 |
+
},
|
| 140 |
+
{
|
| 141 |
+
"epoch": 0.34864643150123054,
|
| 142 |
+
"grad_norm": 0.22819840908050537,
|
| 143 |
+
"learning_rate": 0.0009992338231424006,
|
| 144 |
+
"loss": 0.3139,
|
| 145 |
+
"step": 425
|
| 146 |
+
},
|
| 147 |
+
{
|
| 148 |
+
"epoch": 0.36915504511895,
|
| 149 |
+
"grad_norm": 0.23836681246757507,
|
| 150 |
+
"learning_rate": 0.000998431397275486,
|
| 151 |
+
"loss": 0.3098,
|
| 152 |
+
"step": 450
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.3896636587366694,
|
| 156 |
+
"grad_norm": 0.22939693927764893,
|
| 157 |
+
"learning_rate": 0.0009973451084157006,
|
| 158 |
+
"loss": 0.2955,
|
| 159 |
+
"step": 475
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.41017227235438886,
|
| 163 |
+
"grad_norm": 0.21114112436771393,
|
| 164 |
+
"learning_rate": 0.0009959755752183062,
|
| 165 |
+
"loss": 0.2807,
|
| 166 |
+
"step": 500
|
| 167 |
+
},
|
| 168 |
+
{
|
| 169 |
+
"epoch": 0.4306808859721083,
|
| 170 |
+
"grad_norm": 0.23353807628154755,
|
| 171 |
+
"learning_rate": 0.0009943235776498026,
|
| 172 |
+
"loss": 0.278,
|
| 173 |
+
"step": 525
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 0.45118949958982774,
|
| 177 |
+
"grad_norm": 0.21267421543598175,
|
| 178 |
+
"learning_rate": 0.0009923900565437262,
|
| 179 |
+
"loss": 0.2863,
|
| 180 |
+
"step": 550
|
| 181 |
+
},
|
| 182 |
+
{
|
| 183 |
+
"epoch": 0.4716981132075472,
|
| 184 |
+
"grad_norm": 0.25392886996269226,
|
| 185 |
+
"learning_rate": 0.0009901761130648326,
|
| 186 |
+
"loss": 0.2795,
|
| 187 |
+
"step": 575
|
| 188 |
+
},
|
| 189 |
+
{
|
| 190 |
+
"epoch": 0.4922067268252666,
|
| 191 |
+
"grad_norm": 0.22405794262886047,
|
| 192 |
+
"learning_rate": 0.0009876830080819703,
|
| 193 |
+
"loss": 0.2965,
|
| 194 |
+
"step": 600
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 0.4922067268252666,
|
| 198 |
+
"eval_loss": 0.27075350284576416,
|
| 199 |
+
"eval_runtime": 19.3269,
|
| 200 |
+
"eval_samples_per_second": 51.741,
|
| 201 |
+
"eval_steps_per_second": 0.828,
|
| 202 |
+
"step": 600
|
| 203 |
+
},
|
| 204 |
+
{
|
| 205 |
+
"epoch": 0.5127153404429861,
|
| 206 |
+
"grad_norm": 0.22697776556015015,
|
| 207 |
+
"learning_rate": 0.00098491216145,
|
| 208 |
+
"loss": 0.2764,
|
| 209 |
+
"step": 625
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 0.5332239540607056,
|
| 213 |
+
"grad_norm": 0.2539340853691101,
|
| 214 |
+
"learning_rate": 0.0009818651512011685,
|
| 215 |
+
"loss": 0.2671,
|
| 216 |
+
"step": 650
|
| 217 |
+
},
|
| 218 |
+
{
|
| 219 |
+
"epoch": 0.5537325676784249,
|
| 220 |
+
"grad_norm": 0.2137545496225357,
|
| 221 |
+
"learning_rate": 0.0009785437126463995,
|
| 222 |
+
"loss": 0.2798,
|
| 223 |
+
"step": 675
|
| 224 |
+
},
|
| 225 |
+
{
|
| 226 |
+
"epoch": 0.5742411812961444,
|
| 227 |
+
"grad_norm": 0.22169587016105652,
|
| 228 |
+
"learning_rate": 0.0009749497373870129,
|
| 229 |
+
"loss": 0.2765,
|
| 230 |
+
"step": 700
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 0.5947497949138638,
|
| 234 |
+
"grad_norm": 0.21494679152965546,
|
| 235 |
+
"learning_rate": 0.0009710852722374326,
|
| 236 |
+
"loss": 0.2765,
|
| 237 |
+
"step": 725
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 0.6152584085315833,
|
| 241 |
+
"grad_norm": 0.2177286595106125,
|
| 242 |
+
"learning_rate": 0.0009669525180595009,
|
| 243 |
+
"loss": 0.2725,
|
| 244 |
+
"step": 750
|
| 245 |
+
},
|
| 246 |
+
{
|
| 247 |
+
"epoch": 0.6357670221493027,
|
| 248 |
+
"grad_norm": 0.22016122937202454,
|
| 249 |
+
"learning_rate": 0.0009625538285090593,
|
| 250 |
+
"loss": 0.2682,
|
| 251 |
+
"step": 775
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 0.6562756357670222,
|
| 255 |
+
"grad_norm": 0.20299573242664337,
|
| 256 |
+
"learning_rate": 0.0009578917086955117,
|
| 257 |
+
"loss": 0.2709,
|
| 258 |
+
"step": 800
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 0.6562756357670222,
|
| 262 |
+
"eval_loss": 0.2572968006134033,
|
| 263 |
+
"eval_runtime": 19.3237,
|
| 264 |
+
"eval_samples_per_second": 51.75,
|
| 265 |
+
"eval_steps_per_second": 0.828,
|
| 266 |
+
"step": 800
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 0.6767842493847416,
|
| 270 |
+
"grad_norm": 0.19535581767559052,
|
| 271 |
+
"learning_rate": 0.0009529688137551328,
|
| 272 |
+
"loss": 0.264,
|
| 273 |
+
"step": 825
|
| 274 |
+
},
|
| 275 |
+
{
|
| 276 |
+
"epoch": 0.6972928630024611,
|
| 277 |
+
"grad_norm": 0.21757008135318756,
|
| 278 |
+
"learning_rate": 0.0009477879473389345,
|
| 279 |
+
"loss": 0.267,
|
| 280 |
+
"step": 850
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 0.7178014766201805,
|
| 284 |
+
"grad_norm": 0.16861338913440704,
|
| 285 |
+
"learning_rate": 0.0009423520600159513,
|
| 286 |
+
"loss": 0.2537,
|
| 287 |
+
"step": 875
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 0.7383100902379,
|
| 291 |
+
"grad_norm": 0.2269093096256256,
|
| 292 |
+
"learning_rate": 0.0009366642475928535,
|
| 293 |
+
"loss": 0.2606,
|
| 294 |
+
"step": 900
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 0.7588187038556193,
|
| 298 |
+
"grad_norm": 0.1875704526901245,
|
| 299 |
+
"learning_rate": 0.0009307277493508466,
|
| 300 |
+
"loss": 0.2609,
|
| 301 |
+
"step": 925
|
| 302 |
+
},
|
| 303 |
+
{
|
| 304 |
+
"epoch": 0.7793273174733388,
|
| 305 |
+
"grad_norm": 0.17958270013332367,
|
| 306 |
+
"learning_rate": 0.0009245459462008587,
|
| 307 |
+
"loss": 0.2486,
|
| 308 |
+
"step": 950
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 0.7998359310910582,
|
| 312 |
+
"grad_norm": 0.2056453377008438,
|
| 313 |
+
"learning_rate": 0.0009181223587580702,
|
| 314 |
+
"loss": 0.2554,
|
| 315 |
+
"step": 975
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 0.8203445447087777,
|
| 319 |
+
"grad_norm": 0.19606703519821167,
|
| 320 |
+
"learning_rate": 0.0009114606453368778,
|
| 321 |
+
"loss": 0.2493,
|
| 322 |
+
"step": 1000
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 0.8203445447087777,
|
| 326 |
+
"eval_loss": 0.24378660321235657,
|
| 327 |
+
"eval_runtime": 19.3021,
|
| 328 |
+
"eval_samples_per_second": 51.808,
|
| 329 |
+
"eval_steps_per_second": 0.829,
|
| 330 |
+
"step": 1000
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 0.8408531583264971,
|
| 334 |
+
"grad_norm": 0.192683607339859,
|
| 335 |
+
"learning_rate": 0.0009045645998674388,
|
| 336 |
+
"loss": 0.2492,
|
| 337 |
+
"step": 1025
|
| 338 |
+
},
|
| 339 |
+
{
|
| 340 |
+
"epoch": 0.8613617719442166,
|
| 341 |
+
"grad_norm": 0.19006161391735077,
|
| 342 |
+
"learning_rate": 0.000897438149734979,
|
| 343 |
+
"loss": 0.2452,
|
| 344 |
+
"step": 1050
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 0.881870385561936,
|
| 348 |
+
"grad_norm": 0.1865209937095642,
|
| 349 |
+
"learning_rate": 0.0008900853535430986,
|
| 350 |
+
"loss": 0.2501,
|
| 351 |
+
"step": 1075
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 0.9023789991796555,
|
| 355 |
+
"grad_norm": 0.20048469305038452,
|
| 356 |
+
"learning_rate": 0.0008825103988023442,
|
| 357 |
+
"loss": 0.2504,
|
| 358 |
+
"step": 1100
|
| 359 |
+
},
|
| 360 |
+
{
|
| 361 |
+
"epoch": 0.9228876127973749,
|
| 362 |
+
"grad_norm": 0.18759459257125854,
|
| 363 |
+
"learning_rate": 0.0008747175995453695,
|
| 364 |
+
"loss": 0.2444,
|
| 365 |
+
"step": 1125
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 0.9433962264150944,
|
| 369 |
+
"grad_norm": 0.1509057879447937,
|
| 370 |
+
"learning_rate": 0.0008667113938700395,
|
| 371 |
+
"loss": 0.2444,
|
| 372 |
+
"step": 1150
|
| 373 |
+
},
|
| 374 |
+
{
|
| 375 |
+
"epoch": 0.9639048400328137,
|
| 376 |
+
"grad_norm": 0.17233215272426605,
|
| 377 |
+
"learning_rate": 0.0008584963414118765,
|
| 378 |
+
"loss": 0.2508,
|
| 379 |
+
"step": 1175
|
| 380 |
+
},
|
| 381 |
+
{
|
| 382 |
+
"epoch": 0.9844134536505332,
|
| 383 |
+
"grad_norm": 0.16307514905929565,
|
| 384 |
+
"learning_rate": 0.0008500771207472907,
|
| 385 |
+
"loss": 0.2558,
|
| 386 |
+
"step": 1200
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 0.9844134536505332,
|
| 390 |
+
"eval_loss": 0.23552247881889343,
|
| 391 |
+
"eval_runtime": 19.3122,
|
| 392 |
+
"eval_samples_per_second": 51.781,
|
| 393 |
+
"eval_steps_per_second": 0.828,
|
| 394 |
+
"step": 1200
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 1.0049220672682526,
|
| 398 |
+
"grad_norm": 0.1652350276708603,
|
| 399 |
+
"learning_rate": 0.0008414585267290715,
|
| 400 |
+
"loss": 0.2353,
|
| 401 |
+
"step": 1225
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 1.0254306808859721,
|
| 405 |
+
"grad_norm": 0.16925707459449768,
|
| 406 |
+
"learning_rate": 0.0008326454677556577,
|
| 407 |
+
"loss": 0.2165,
|
| 408 |
+
"step": 1250
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 1.0459392945036916,
|
| 412 |
+
"grad_norm": 0.193923220038414,
|
| 413 |
+
"learning_rate": 0.0008236429629757432,
|
| 414 |
+
"loss": 0.2191,
|
| 415 |
+
"step": 1275
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 1.066447908121411,
|
| 419 |
+
"grad_norm": 0.14884328842163086,
|
| 420 |
+
"learning_rate": 0.0008144561394298075,
|
| 421 |
+
"loss": 0.2216,
|
| 422 |
+
"step": 1300
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 1.0869565217391304,
|
| 426 |
+
"grad_norm": 0.18515324592590332,
|
| 427 |
+
"learning_rate": 0.0008050902291302019,
|
| 428 |
+
"loss": 0.2174,
|
| 429 |
+
"step": 1325
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 1.1074651353568499,
|
| 433 |
+
"grad_norm": 0.14039944112300873,
|
| 434 |
+
"learning_rate": 0.0007955505660814515,
|
| 435 |
+
"loss": 0.2122,
|
| 436 |
+
"step": 1350
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 1.1279737489745694,
|
| 440 |
+
"grad_norm": 0.17923711240291595,
|
| 441 |
+
"learning_rate": 0.0007858425832424727,
|
| 442 |
+
"loss": 0.2143,
|
| 443 |
+
"step": 1375
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 1.1484823625922886,
|
| 447 |
+
"grad_norm": 0.17652627825737,
|
| 448 |
+
"learning_rate": 0.000775971809432434,
|
| 449 |
+
"loss": 0.2065,
|
| 450 |
+
"step": 1400
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 1.1484823625922886,
|
| 454 |
+
"eval_loss": 0.23243418335914612,
|
| 455 |
+
"eval_runtime": 19.325,
|
| 456 |
+
"eval_samples_per_second": 51.746,
|
| 457 |
+
"eval_steps_per_second": 0.828,
|
| 458 |
+
"step": 1400
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 1.1689909762100081,
|
| 462 |
+
"grad_norm": 0.17934605479240417,
|
| 463 |
+
"learning_rate": 0.0007659438661820235,
|
| 464 |
+
"loss": 0.2174,
|
| 465 |
+
"step": 1425
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 1.1894995898277276,
|
| 469 |
+
"grad_norm": 0.1746005117893219,
|
| 470 |
+
"learning_rate": 0.0007557644645319157,
|
| 471 |
+
"loss": 0.2185,
|
| 472 |
+
"step": 1450
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 1.2100082034454471,
|
| 476 |
+
"grad_norm": 0.17862309515476227,
|
| 477 |
+
"learning_rate": 0.0007454394017802622,
|
| 478 |
+
"loss": 0.2143,
|
| 479 |
+
"step": 1475
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 1.2305168170631666,
|
| 483 |
+
"grad_norm": 0.17890584468841553,
|
| 484 |
+
"learning_rate": 0.0007349745581810565,
|
| 485 |
+
"loss": 0.2239,
|
| 486 |
+
"step": 1500
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 1.251025430680886,
|
| 490 |
+
"grad_norm": 0.19279733300209045,
|
| 491 |
+
"learning_rate": 0.0007243758935952547,
|
| 492 |
+
"loss": 0.2194,
|
| 493 |
+
"step": 1525
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 1.2715340442986054,
|
| 497 |
+
"grad_norm": 0.14889924228191376,
|
| 498 |
+
"learning_rate": 0.0007136494440965596,
|
| 499 |
+
"loss": 0.2138,
|
| 500 |
+
"step": 1550
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 1.2920426579163249,
|
| 504 |
+
"grad_norm": 0.13340285420417786,
|
| 505 |
+
"learning_rate": 0.0007028013185338001,
|
| 506 |
+
"loss": 0.2139,
|
| 507 |
+
"step": 1575
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 1.3125512715340442,
|
| 511 |
+
"grad_norm": 0.1602306365966797,
|
| 512 |
+
"learning_rate": 0.000691837695051865,
|
| 513 |
+
"loss": 0.2074,
|
| 514 |
+
"step": 1600
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 1.3125512715340442,
|
| 518 |
+
"eval_loss": 0.22446581721305847,
|
| 519 |
+
"eval_runtime": 19.3197,
|
| 520 |
+
"eval_samples_per_second": 51.761,
|
| 521 |
+
"eval_steps_per_second": 0.828,
|
| 522 |
+
"step": 1600
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 1.3330598851517639,
|
| 526 |
+
"grad_norm": 0.1418837457895279,
|
| 527 |
+
"learning_rate": 0.00068076481757317,
|
| 528 |
+
"loss": 0.2104,
|
| 529 |
+
"step": 1625
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 1.3535684987694832,
|
| 533 |
+
"grad_norm": 0.16610907018184662,
|
| 534 |
+
"learning_rate": 0.0006695889922416659,
|
| 535 |
+
"loss": 0.2116,
|
| 536 |
+
"step": 1650
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 1.3740771123872026,
|
| 540 |
+
"grad_norm": 0.16588063538074493,
|
| 541 |
+
"learning_rate": 0.0006583165838314095,
|
| 542 |
+
"loss": 0.2089,
|
| 543 |
+
"step": 1675
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 1.3945857260049221,
|
| 547 |
+
"grad_norm": 0.14670899510383606,
|
| 548 |
+
"learning_rate": 0.0006469540121217438,
|
| 549 |
+
"loss": 0.2097,
|
| 550 |
+
"step": 1700
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 1.4150943396226414,
|
| 554 |
+
"grad_norm": 0.1579999029636383,
|
| 555 |
+
"learning_rate": 0.0006355077482411526,
|
| 556 |
+
"loss": 0.2064,
|
| 557 |
+
"step": 1725
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 1.435602953240361,
|
| 561 |
+
"grad_norm": 0.14437806606292725,
|
| 562 |
+
"learning_rate": 0.0006239843109818716,
|
| 563 |
+
"loss": 0.2061,
|
| 564 |
+
"step": 1750
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 1.4561115668580804,
|
| 568 |
+
"grad_norm": 0.16887524724006653,
|
| 569 |
+
"learning_rate": 0.0006123902630873521,
|
| 570 |
+
"loss": 0.2123,
|
| 571 |
+
"step": 1775
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 1.4766201804758,
|
| 575 |
+
"grad_norm": 0.14332818984985352,
|
| 576 |
+
"learning_rate": 0.0006007322075146968,
|
| 577 |
+
"loss": 0.2023,
|
| 578 |
+
"step": 1800
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 1.4766201804758,
|
| 582 |
+
"eval_loss": 0.2198713719844818,
|
| 583 |
+
"eval_runtime": 19.2983,
|
| 584 |
+
"eval_samples_per_second": 51.818,
|
| 585 |
+
"eval_steps_per_second": 0.829,
|
| 586 |
+
"step": 1800
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 1.4971287940935194,
|
| 590 |
+
"grad_norm": 0.14545492827892303,
|
| 591 |
+
"learning_rate": 0.0005890167836741919,
|
| 592 |
+
"loss": 0.2023,
|
| 593 |
+
"step": 1825
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 1.5176374077112387,
|
| 597 |
+
"grad_norm": 0.13525480031967163,
|
| 598 |
+
"learning_rate": 0.0005772506636480781,
|
| 599 |
+
"loss": 0.2055,
|
| 600 |
+
"step": 1850
|
| 601 |
+
},
|
| 602 |
+
{
|
| 603 |
+
"epoch": 1.5381460213289582,
|
| 604 |
+
"grad_norm": 0.13384069502353668,
|
| 605 |
+
"learning_rate": 0.0005654405483907163,
|
| 606 |
+
"loss": 0.2049,
|
| 607 |
+
"step": 1875
|
| 608 |
+
},
|
| 609 |
+
{
|
| 610 |
+
"epoch": 1.5586546349466777,
|
| 611 |
+
"grad_norm": 0.16189663112163544,
|
| 612 |
+
"learning_rate": 0.0005535931639123083,
|
| 613 |
+
"loss": 0.2076,
|
| 614 |
+
"step": 1900
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 1.579163248564397,
|
| 618 |
+
"grad_norm": 0.15479648113250732,
|
| 619 |
+
"learning_rate": 0.00054171525744835,
|
| 620 |
+
"loss": 0.2094,
|
| 621 |
+
"step": 1925
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 1.5996718621821167,
|
| 625 |
+
"grad_norm": 0.15555700659751892,
|
| 626 |
+
"learning_rate": 0.0005298135936169937,
|
| 627 |
+
"loss": 0.2069,
|
| 628 |
+
"step": 1950
|
| 629 |
+
},
|
| 630 |
+
{
|
| 631 |
+
"epoch": 1.620180475799836,
|
| 632 |
+
"grad_norm": 0.1431935578584671,
|
| 633 |
+
"learning_rate": 0.000517894950566514,
|
| 634 |
+
"loss": 0.2023,
|
| 635 |
+
"step": 1975
|
| 636 |
+
},
|
| 637 |
+
{
|
| 638 |
+
"epoch": 1.6406890894175554,
|
| 639 |
+
"grad_norm": 0.15172646939754486,
|
| 640 |
+
"learning_rate": 0.0005059661161150655,
|
| 641 |
+
"loss": 0.2089,
|
| 642 |
+
"step": 2000
|
| 643 |
+
},
|
| 644 |
+
{
|
| 645 |
+
"epoch": 1.6406890894175554,
|
| 646 |
+
"eval_loss": 0.21559131145477295,
|
| 647 |
+
"eval_runtime": 19.3061,
|
| 648 |
+
"eval_samples_per_second": 51.797,
|
| 649 |
+
"eval_steps_per_second": 0.829,
|
| 650 |
+
"step": 2000
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 1.661197703035275,
|
| 654 |
+
"grad_norm": 0.13010752201080322,
|
| 655 |
+
"learning_rate": 0.0004940338838849347,
|
| 656 |
+
"loss": 0.1987,
|
| 657 |
+
"step": 2025
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 1.6817063166529942,
|
| 661 |
+
"grad_norm": 0.1643010377883911,
|
| 662 |
+
"learning_rate": 0.0004821050494334861,
|
| 663 |
+
"loss": 0.2005,
|
| 664 |
+
"step": 2050
|
| 665 |
+
},
|
| 666 |
+
{
|
| 667 |
+
"epoch": 1.7022149302707137,
|
| 668 |
+
"grad_norm": 0.16517673432826996,
|
| 669 |
+
"learning_rate": 0.0004701864063830064,
|
| 670 |
+
"loss": 0.2056,
|
| 671 |
+
"step": 2075
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 1.7227235438884332,
|
| 675 |
+
"grad_norm": 0.14333464205265045,
|
| 676 |
+
"learning_rate": 0.0004582847425516501,
|
| 677 |
+
"loss": 0.1961,
|
| 678 |
+
"step": 2100
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 1.7432321575061525,
|
| 682 |
+
"grad_norm": 0.12679831683635712,
|
| 683 |
+
"learning_rate": 0.00044640683608769186,
|
| 684 |
+
"loss": 0.2002,
|
| 685 |
+
"step": 2125
|
| 686 |
+
},
|
| 687 |
+
{
|
| 688 |
+
"epoch": 1.7637407711238722,
|
| 689 |
+
"grad_norm": 0.11452236771583557,
|
| 690 |
+
"learning_rate": 0.0004345594516092838,
|
| 691 |
+
"loss": 0.1982,
|
| 692 |
+
"step": 2150
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 1.7842493847415914,
|
| 696 |
+
"grad_norm": 0.1682896614074707,
|
| 697 |
+
"learning_rate": 0.0004227493363519221,
|
| 698 |
+
"loss": 0.2069,
|
| 699 |
+
"step": 2175
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 1.804757998359311,
|
| 703 |
+
"grad_norm": 0.1387612521648407,
|
| 704 |
+
"learning_rate": 0.0004109832163258082,
|
| 705 |
+
"loss": 0.1986,
|
| 706 |
+
"step": 2200
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 1.804757998359311,
|
| 710 |
+
"eval_loss": 0.21137093007564545,
|
| 711 |
+
"eval_runtime": 19.3527,
|
| 712 |
+
"eval_samples_per_second": 51.672,
|
| 713 |
+
"eval_steps_per_second": 0.827,
|
| 714 |
+
"step": 2200
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 1.8252666119770304,
|
| 718 |
+
"grad_norm": 0.16713084280490875,
|
| 719 |
+
"learning_rate": 0.0003992677924853032,
|
| 720 |
+
"loss": 0.2008,
|
| 721 |
+
"step": 2225
|
| 722 |
+
},
|
| 723 |
+
{
|
| 724 |
+
"epoch": 1.8457752255947497,
|
| 725 |
+
"grad_norm": 0.1538151055574417,
|
| 726 |
+
"learning_rate": 0.0003876097369126479,
|
| 727 |
+
"loss": 0.2017,
|
| 728 |
+
"step": 2250
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 1.8662838392124692,
|
| 732 |
+
"grad_norm": 0.12900249660015106,
|
| 733 |
+
"learning_rate": 0.0003760156890181283,
|
| 734 |
+
"loss": 0.2061,
|
| 735 |
+
"step": 2275
|
| 736 |
+
},
|
| 737 |
+
{
|
| 738 |
+
"epoch": 1.8867924528301887,
|
| 739 |
+
"grad_norm": 0.1338413655757904,
|
| 740 |
+
"learning_rate": 0.0003644922517588474,
|
| 741 |
+
"loss": 0.1975,
|
| 742 |
+
"step": 2300
|
| 743 |
+
},
|
| 744 |
+
{
|
| 745 |
+
"epoch": 1.907301066447908,
|
| 746 |
+
"grad_norm": 0.15099306404590607,
|
| 747 |
+
"learning_rate": 0.0003530459878782563,
|
| 748 |
+
"loss": 0.2064,
|
| 749 |
+
"step": 2325
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 1.9278096800656277,
|
| 753 |
+
"grad_norm": 0.14049404859542847,
|
| 754 |
+
"learning_rate": 0.00034168341616859065,
|
| 755 |
+
"loss": 0.1899,
|
| 756 |
+
"step": 2350
|
| 757 |
+
},
|
| 758 |
+
{
|
| 759 |
+
"epoch": 1.948318293683347,
|
| 760 |
+
"grad_norm": 0.12242951989173889,
|
| 761 |
+
"learning_rate": 0.0003304110077583341,
|
| 762 |
+
"loss": 0.1949,
|
| 763 |
+
"step": 2375
|
| 764 |
+
},
|
| 765 |
+
{
|
| 766 |
+
"epoch": 1.9688269073010665,
|
| 767 |
+
"grad_norm": 0.14392268657684326,
|
| 768 |
+
"learning_rate": 0.00031923518242683,
|
| 769 |
+
"loss": 0.1959,
|
| 770 |
+
"step": 2400
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 1.9688269073010665,
|
| 774 |
+
"eval_loss": 0.2068263292312622,
|
| 775 |
+
"eval_runtime": 19.3392,
|
| 776 |
+
"eval_samples_per_second": 51.708,
|
| 777 |
+
"eval_steps_per_second": 0.827,
|
| 778 |
+
"step": 2400
|
| 779 |
+
},
|
| 780 |
+
{
|
| 781 |
+
"epoch": 1.989335520918786,
|
| 782 |
+
"grad_norm": 0.13493984937667847,
|
| 783 |
+
"learning_rate": 0.000308162304948135,
|
| 784 |
+
"loss": 0.1924,
|
| 785 |
+
"step": 2425
|
| 786 |
+
},
|
| 787 |
+
{
|
| 788 |
+
"epoch": 2.0098441345365052,
|
| 789 |
+
"grad_norm": 0.13458842039108276,
|
| 790 |
+
"learning_rate": 0.00029719868146619997,
|
| 791 |
+
"loss": 0.1804,
|
| 792 |
+
"step": 2450
|
| 793 |
+
},
|
| 794 |
+
{
|
| 795 |
+
"epoch": 2.030352748154225,
|
| 796 |
+
"grad_norm": 0.14248383045196533,
|
| 797 |
+
"learning_rate": 0.0002863505559034406,
|
| 798 |
+
"loss": 0.1643,
|
| 799 |
+
"step": 2475
|
| 800 |
+
},
|
| 801 |
+
{
|
| 802 |
+
"epoch": 2.0508613617719442,
|
| 803 |
+
"grad_norm": 0.13979841768741608,
|
| 804 |
+
"learning_rate": 0.0002756241064047456,
|
| 805 |
+
"loss": 0.164,
|
| 806 |
+
"step": 2500
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 2.0713699753896635,
|
| 810 |
+
"grad_norm": 0.14228808879852295,
|
| 811 |
+
"learning_rate": 0.00026502544181894374,
|
| 812 |
+
"loss": 0.1602,
|
| 813 |
+
"step": 2525
|
| 814 |
+
},
|
| 815 |
+
{
|
| 816 |
+
"epoch": 2.091878589007383,
|
| 817 |
+
"grad_norm": 0.141134575009346,
|
| 818 |
+
"learning_rate": 0.00025456059821973783,
|
| 819 |
+
"loss": 0.169,
|
| 820 |
+
"step": 2550
|
| 821 |
+
},
|
| 822 |
+
{
|
| 823 |
+
"epoch": 2.1123872026251025,
|
| 824 |
+
"grad_norm": 0.13459524512290955,
|
| 825 |
+
"learning_rate": 0.00024423553546808424,
|
| 826 |
+
"loss": 0.161,
|
| 827 |
+
"step": 2575
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"epoch": 2.132895816242822,
|
| 831 |
+
"grad_norm": 0.1448785364627838,
|
| 832 |
+
"learning_rate": 0.00023405613381797665,
|
| 833 |
+
"loss": 0.1543,
|
| 834 |
+
"step": 2600
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 2.132895816242822,
|
| 838 |
+
"eval_loss": 0.2100238800048828,
|
| 839 |
+
"eval_runtime": 19.3232,
|
| 840 |
+
"eval_samples_per_second": 51.751,
|
| 841 |
+
"eval_steps_per_second": 0.828,
|
| 842 |
+
"step": 2600
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 2.1534044298605415,
|
| 846 |
+
"grad_norm": 0.15394213795661926,
|
| 847 |
+
"learning_rate": 0.00022402819056756602,
|
| 848 |
+
"loss": 0.158,
|
| 849 |
+
"step": 2625
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 2.1739130434782608,
|
| 853 |
+
"grad_norm": 0.1316463053226471,
|
| 854 |
+
"learning_rate": 0.00021415741675752742,
|
| 855 |
+
"loss": 0.1572,
|
| 856 |
+
"step": 2650
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 2.1944216570959805,
|
| 860 |
+
"grad_norm": 0.14899882674217224,
|
| 861 |
+
"learning_rate": 0.00020444943391854866,
|
| 862 |
+
"loss": 0.168,
|
| 863 |
+
"step": 2675
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 2.2149302707136997,
|
| 867 |
+
"grad_norm": 0.12964214384555817,
|
| 868 |
+
"learning_rate": 0.00019490977086979822,
|
| 869 |
+
"loss": 0.1639,
|
| 870 |
+
"step": 2700
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 2.235438884331419,
|
| 874 |
+
"grad_norm": 0.15520571172237396,
|
| 875 |
+
"learning_rate": 0.0001855438605701925,
|
| 876 |
+
"loss": 0.1526,
|
| 877 |
+
"step": 2725
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 2.2559474979491387,
|
| 881 |
+
"grad_norm": 0.1541876345872879,
|
| 882 |
+
"learning_rate": 0.00017635703702425682,
|
| 883 |
+
"loss": 0.1662,
|
| 884 |
+
"step": 2750
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 2.276456111566858,
|
| 888 |
+
"grad_norm": 0.16255255043506622,
|
| 889 |
+
"learning_rate": 0.0001673545322443424,
|
| 890 |
+
"loss": 0.1633,
|
| 891 |
+
"step": 2775
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 2.2969647251845773,
|
| 895 |
+
"grad_norm": 0.15506723523139954,
|
| 896 |
+
"learning_rate": 0.00015854147327092855,
|
| 897 |
+
"loss": 0.1583,
|
| 898 |
+
"step": 2800
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 2.2969647251845773,
|
| 902 |
+
"eval_loss": 0.20912893116474152,
|
| 903 |
+
"eval_runtime": 19.2905,
|
| 904 |
+
"eval_samples_per_second": 51.839,
|
| 905 |
+
"eval_steps_per_second": 0.829,
|
| 906 |
+
"step": 2800
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 2.317473338802297,
|
| 910 |
+
"grad_norm": 0.14822597801685333,
|
| 911 |
+
"learning_rate": 0.00014992287925270936,
|
| 912 |
+
"loss": 0.1547,
|
| 913 |
+
"step": 2825
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 2.3379819524200163,
|
| 917 |
+
"grad_norm": 0.1518256664276123,
|
| 918 |
+
"learning_rate": 0.00014150365858812353,
|
| 919 |
+
"loss": 0.157,
|
| 920 |
+
"step": 2850
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 2.358490566037736,
|
| 924 |
+
"grad_norm": 0.1334943324327469,
|
| 925 |
+
"learning_rate": 0.00013328860612996053,
|
| 926 |
+
"loss": 0.1594,
|
| 927 |
+
"step": 2875
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 2.3789991796554553,
|
| 931 |
+
"grad_norm": 0.13774850964546204,
|
| 932 |
+
"learning_rate": 0.00012528240045463047,
|
| 933 |
+
"loss": 0.1616,
|
| 934 |
+
"step": 2900
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 2.399507793273175,
|
| 938 |
+
"grad_norm": 0.17961899936199188,
|
| 939 |
+
"learning_rate": 0.00011748960119765595,
|
| 940 |
+
"loss": 0.1597,
|
| 941 |
+
"step": 2925
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 2.4200164068908943,
|
| 945 |
+
"grad_norm": 0.1587635725736618,
|
| 946 |
+
"learning_rate": 0.00010991464645690142,
|
| 947 |
+
"loss": 0.1628,
|
| 948 |
+
"step": 2950
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 2.4405250205086135,
|
| 952 |
+
"grad_norm": 0.14210981130599976,
|
| 953 |
+
"learning_rate": 0.00010256185026502096,
|
| 954 |
+
"loss": 0.1599,
|
| 955 |
+
"step": 2975
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 2.4610336341263332,
|
| 959 |
+
"grad_norm": 0.1479632705450058,
|
| 960 |
+
"learning_rate": 9.543540013256136e-05,
|
| 961 |
+
"loss": 0.1554,
|
| 962 |
+
"step": 3000
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 2.4610336341263332,
|
| 966 |
+
"eval_loss": 0.20673906803131104,
|
| 967 |
+
"eval_runtime": 19.2847,
|
| 968 |
+
"eval_samples_per_second": 51.855,
|
| 969 |
+
"eval_steps_per_second": 0.83,
|
| 970 |
+
"step": 3000
|
| 971 |
+
},
|
| 972 |
+
{
|
| 973 |
+
"epoch": 2.4815422477440525,
|
| 974 |
+
"grad_norm": 0.13706982135772705,
|
| 975 |
+
"learning_rate": 8.853935466312224e-05,
|
| 976 |
+
"loss": 0.1579,
|
| 977 |
+
"step": 3025
|
| 978 |
+
},
|
| 979 |
+
{
|
| 980 |
+
"epoch": 2.502050861361772,
|
| 981 |
+
"grad_norm": 0.1355781853199005,
|
| 982 |
+
"learning_rate": 8.187764124192981e-05,
|
| 983 |
+
"loss": 0.1543,
|
| 984 |
+
"step": 3050
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 2.5225594749794915,
|
| 988 |
+
"grad_norm": 0.15336012840270996,
|
| 989 |
+
"learning_rate": 7.545405379914138e-05,
|
| 990 |
+
"loss": 0.15,
|
| 991 |
+
"step": 3075
|
| 992 |
+
},
|
| 993 |
+
{
|
| 994 |
+
"epoch": 2.543068088597211,
|
| 995 |
+
"grad_norm": 0.14177563786506653,
|
| 996 |
+
"learning_rate": 6.927225064915349e-05,
|
| 997 |
+
"loss": 0.1617,
|
| 998 |
+
"step": 3100
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 2.56357670221493,
|
| 1002 |
+
"grad_norm": 0.1608775109052658,
|
| 1003 |
+
"learning_rate": 6.333575240714656e-05,
|
| 1004 |
+
"loss": 0.1506,
|
| 1005 |
+
"step": 3125
|
| 1006 |
+
},
|
| 1007 |
+
{
|
| 1008 |
+
"epoch": 2.5840853158326498,
|
| 1009 |
+
"grad_norm": 0.1622104048728943,
|
| 1010 |
+
"learning_rate": 5.764793998404877e-05,
|
| 1011 |
+
"loss": 0.1539,
|
| 1012 |
+
"step": 3150
|
| 1013 |
+
},
|
| 1014 |
+
{
|
| 1015 |
+
"epoch": 2.604593929450369,
|
| 1016 |
+
"grad_norm": 0.14182351529598236,
|
| 1017 |
+
"learning_rate": 5.2212052661065654e-05,
|
| 1018 |
+
"loss": 0.1563,
|
| 1019 |
+
"step": 3175
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 2.6251025430680883,
|
| 1023 |
+
"grad_norm": 0.13961577415466309,
|
| 1024 |
+
"learning_rate": 4.703118624486735e-05,
|
| 1025 |
+
"loss": 0.1601,
|
| 1026 |
+
"step": 3200
|
| 1027 |
+
},
|
| 1028 |
+
{
|
| 1029 |
+
"epoch": 2.6251025430680883,
|
| 1030 |
+
"eval_loss": 0.2056739181280136,
|
| 1031 |
+
"eval_runtime": 19.3188,
|
| 1032 |
+
"eval_samples_per_second": 51.763,
|
| 1033 |
+
"eval_steps_per_second": 0.828,
|
| 1034 |
+
"step": 3200
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 2.645611156685808,
|
| 1038 |
+
"grad_norm": 0.1635577529668808,
|
| 1039 |
+
"learning_rate": 4.210829130448834e-05,
|
| 1040 |
+
"loss": 0.1673,
|
| 1041 |
+
"step": 3225
|
| 1042 |
+
},
|
| 1043 |
+
{
|
| 1044 |
+
"epoch": 2.6661197703035278,
|
| 1045 |
+
"grad_norm": 0.15194955468177795,
|
| 1046 |
+
"learning_rate": 3.7446171490940704e-05,
|
| 1047 |
+
"loss": 0.1541,
|
| 1048 |
+
"step": 3250
|
| 1049 |
+
},
|
| 1050 |
+
{
|
| 1051 |
+
"epoch": 2.686628383921247,
|
| 1052 |
+
"grad_norm": 0.15756186842918396,
|
| 1053 |
+
"learning_rate": 3.3047481940499055e-05,
|
| 1054 |
+
"loss": 0.159,
|
| 1055 |
+
"step": 3275
|
| 1056 |
+
},
|
| 1057 |
+
{
|
| 1058 |
+
"epoch": 2.7071369975389663,
|
| 1059 |
+
"grad_norm": 0.13421984016895294,
|
| 1060 |
+
"learning_rate": 2.8914727762567482e-05,
|
| 1061 |
+
"loss": 0.1527,
|
| 1062 |
+
"step": 3300
|
| 1063 |
+
},
|
| 1064 |
+
{
|
| 1065 |
+
"epoch": 2.727645611156686,
|
| 1066 |
+
"grad_norm": 0.13674722611904144,
|
| 1067 |
+
"learning_rate": 2.5050262612987206e-05,
|
| 1068 |
+
"loss": 0.1512,
|
| 1069 |
+
"step": 3325
|
| 1070 |
+
},
|
| 1071 |
+
{
|
| 1072 |
+
"epoch": 2.7481542247744053,
|
| 1073 |
+
"grad_norm": 0.13958343863487244,
|
| 1074 |
+
"learning_rate": 2.14562873536005e-05,
|
| 1075 |
+
"loss": 0.1481,
|
| 1076 |
+
"step": 3350
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 2.7686628383921246,
|
| 1080 |
+
"grad_norm": 0.17911262810230255,
|
| 1081 |
+
"learning_rate": 1.8134848798831537e-05,
|
| 1082 |
+
"loss": 0.1639,
|
| 1083 |
+
"step": 3375
|
| 1084 |
+
},
|
| 1085 |
+
{
|
| 1086 |
+
"epoch": 2.7891714520098443,
|
| 1087 |
+
"grad_norm": 0.14383310079574585,
|
| 1088 |
+
"learning_rate": 1.5087838549999955e-05,
|
| 1089 |
+
"loss": 0.1528,
|
| 1090 |
+
"step": 3400
|
| 1091 |
+
},
|
| 1092 |
+
{
|
| 1093 |
+
"epoch": 2.7891714520098443,
|
| 1094 |
+
"eval_loss": 0.20520254969596863,
|
| 1095 |
+
"eval_runtime": 19.3265,
|
| 1096 |
+
"eval_samples_per_second": 51.742,
|
| 1097 |
+
"eval_steps_per_second": 0.828,
|
| 1098 |
+
"step": 3400
|
| 1099 |
+
},
|
| 1100 |
+
{
|
| 1101 |
+
"epoch": 2.8096800656275636,
|
| 1102 |
+
"grad_norm": 0.13070909678936005,
|
| 1103 |
+
"learning_rate": 1.2316991918029653e-05,
|
| 1104 |
+
"loss": 0.1609,
|
| 1105 |
+
"step": 3425
|
| 1106 |
+
},
|
| 1107 |
+
{
|
| 1108 |
+
"epoch": 2.830188679245283,
|
| 1109 |
+
"grad_norm": 0.14770615100860596,
|
| 1110 |
+
"learning_rate": 9.823886935167502e-06,
|
| 1111 |
+
"loss": 0.1519,
|
| 1112 |
+
"step": 3450
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 2.8506972928630026,
|
| 1116 |
+
"grad_norm": 0.1517811119556427,
|
| 1117 |
+
"learning_rate": 7.60994345627386e-06,
|
| 1118 |
+
"loss": 0.1582,
|
| 1119 |
+
"step": 3475
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 2.871205906480722,
|
| 1123 |
+
"grad_norm": 0.12982696294784546,
|
| 1124 |
+
"learning_rate": 5.67642235019733e-06,
|
| 1125 |
+
"loss": 0.1504,
|
| 1126 |
+
"step": 3500
|
| 1127 |
+
},
|
| 1128 |
+
{
|
| 1129 |
+
"epoch": 2.891714520098441,
|
| 1130 |
+
"grad_norm": 0.14978957176208496,
|
| 1131 |
+
"learning_rate": 4.024424781693792e-06,
|
| 1132 |
+
"loss": 0.1537,
|
| 1133 |
+
"step": 3525
|
| 1134 |
+
},
|
| 1135 |
+
{
|
| 1136 |
+
"epoch": 2.912223133716161,
|
| 1137 |
+
"grad_norm": 0.14696450531482697,
|
| 1138 |
+
"learning_rate": 2.6548915842993793e-06,
|
| 1139 |
+
"loss": 0.1535,
|
| 1140 |
+
"step": 3550
|
| 1141 |
+
},
|
| 1142 |
+
{
|
| 1143 |
+
"epoch": 2.93273174733388,
|
| 1144 |
+
"grad_norm": 0.15015673637390137,
|
| 1145 |
+
"learning_rate": 1.5686027245138123e-06,
|
| 1146 |
+
"loss": 0.1505,
|
| 1147 |
+
"step": 3575
|
| 1148 |
+
},
|
| 1149 |
+
{
|
| 1150 |
+
"epoch": 2.9532403609516,
|
| 1151 |
+
"grad_norm": 0.13611365854740143,
|
| 1152 |
+
"learning_rate": 7.661768575994432e-07,
|
| 1153 |
+
"loss": 0.1576,
|
| 1154 |
+
"step": 3600
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 2.9532403609516,
|
| 1158 |
+
"eval_loss": 0.2049596607685089,
|
| 1159 |
+
"eval_runtime": 19.3142,
|
| 1160 |
+
"eval_samples_per_second": 51.775,
|
| 1161 |
+
"eval_steps_per_second": 0.828,
|
| 1162 |
+
"step": 3600
|
| 1163 |
+
},
|
| 1164 |
+
{
|
| 1165 |
+
"epoch": 2.973748974569319,
|
| 1166 |
+
"grad_norm": 0.13153861463069916,
|
| 1167 |
+
"learning_rate": 2.480709752493704e-07,
|
| 1168 |
+
"loss": 0.1532,
|
| 1169 |
+
"step": 3625
|
| 1170 |
+
},
|
| 1171 |
+
{
|
| 1172 |
+
"epoch": 2.994257588187039,
|
| 1173 |
+
"grad_norm": 0.14799685776233673,
|
| 1174 |
+
"learning_rate": 1.4580145324849525e-08,
|
| 1175 |
+
"loss": 0.1517,
|
| 1176 |
+
"step": 3650
|
| 1177 |
+
},
|
| 1178 |
+
{
|
| 1179 |
+
"epoch": 3.0,
|
| 1180 |
+
"step": 3657,
|
| 1181 |
+
"total_flos": 2.37785441107968e+18,
|
| 1182 |
+
"train_loss": 0.2201500027956192,
|
| 1183 |
+
"train_runtime": 3679.5729,
|
| 1184 |
+
"train_samples_per_second": 31.797,
|
| 1185 |
+
"train_steps_per_second": 0.994
|
| 1186 |
+
}
|
| 1187 |
+
],
|
| 1188 |
+
"logging_steps": 25,
|
| 1189 |
+
"max_steps": 3657,
|
| 1190 |
+
"num_input_tokens_seen": 0,
|
| 1191 |
+
"num_train_epochs": 3,
|
| 1192 |
+
"save_steps": 0,
|
| 1193 |
+
"stateful_callbacks": {
|
| 1194 |
+
"TrainerControl": {
|
| 1195 |
+
"args": {
|
| 1196 |
+
"should_epoch_stop": false,
|
| 1197 |
+
"should_evaluate": false,
|
| 1198 |
+
"should_log": false,
|
| 1199 |
+
"should_save": true,
|
| 1200 |
+
"should_training_stop": true
|
| 1201 |
+
},
|
| 1202 |
+
"attributes": {}
|
| 1203 |
+
}
|
| 1204 |
+
},
|
| 1205 |
+
"total_flos": 2.37785441107968e+18,
|
| 1206 |
+
"train_batch_size": 32,
|
| 1207 |
+
"trial_name": null,
|
| 1208 |
+
"trial_params": null
|
| 1209 |
+
}
|
nl_tasks/exps/run_ex14_3ep/ft/adapter_config.json
ADDED
|
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"T": 1.0,
|
| 3 |
+
"base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
|
| 4 |
+
"bias": "none",
|
| 5 |
+
"inference_mode": false,
|
| 6 |
+
"layers_to_transform": null,
|
| 7 |
+
"modules_to_save": null,
|
| 8 |
+
"num_rotations": 1,
|
| 9 |
+
"peft_type": "ROTATION",
|
| 10 |
+
"r": 16,
|
| 11 |
+
"revision": null,
|
| 12 |
+
"target_modules": [
|
| 13 |
+
"v_proj",
|
| 14 |
+
"q_proj"
|
| 15 |
+
],
|
| 16 |
+
"target_modules_to_skip": null,
|
| 17 |
+
"task_type": "CAUSAL_LM"
|
| 18 |
+
}
|
nl_tasks/exps/run_ex14_3ep/ft/added_tokens.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"[PAD]": 32000
|
| 3 |
+
}
|
nl_tasks/exps/run_ex14_3ep/ft/special_tokens_map.json
ADDED
|
@@ -0,0 +1,30 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": {
|
| 3 |
+
"content": "</s>",
|
| 4 |
+
"lstrip": false,
|
| 5 |
+
"normalized": false,
|
| 6 |
+
"rstrip": false,
|
| 7 |
+
"single_word": false
|
| 8 |
+
},
|
| 9 |
+
"eos_token": {
|
| 10 |
+
"content": "</s>",
|
| 11 |
+
"lstrip": false,
|
| 12 |
+
"normalized": false,
|
| 13 |
+
"rstrip": false,
|
| 14 |
+
"single_word": false
|
| 15 |
+
},
|
| 16 |
+
"pad_token": {
|
| 17 |
+
"content": "[PAD]",
|
| 18 |
+
"lstrip": false,
|
| 19 |
+
"normalized": false,
|
| 20 |
+
"rstrip": false,
|
| 21 |
+
"single_word": false
|
| 22 |
+
},
|
| 23 |
+
"unk_token": {
|
| 24 |
+
"content": "</s>",
|
| 25 |
+
"lstrip": false,
|
| 26 |
+
"normalized": false,
|
| 27 |
+
"rstrip": false,
|
| 28 |
+
"single_word": false
|
| 29 |
+
}
|
| 30 |
+
}
|
nl_tasks/exps/run_ex14_3ep/ft/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
nl_tasks/exps/run_ex14_3ep/ft/tokenizer.model
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
|
| 3 |
+
size 499723
|
nl_tasks/exps/run_ex14_3ep/ft/tokenizer_config.json
ADDED
|
@@ -0,0 +1,51 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_bos_token": true,
|
| 3 |
+
"add_eos_token": false,
|
| 4 |
+
"add_prefix_space": null,
|
| 5 |
+
"added_tokens_decoder": {
|
| 6 |
+
"0": {
|
| 7 |
+
"content": "<unk>",
|
| 8 |
+
"lstrip": false,
|
| 9 |
+
"normalized": false,
|
| 10 |
+
"rstrip": false,
|
| 11 |
+
"single_word": false,
|
| 12 |
+
"special": true
|
| 13 |
+
},
|
| 14 |
+
"1": {
|
| 15 |
+
"content": "<s>",
|
| 16 |
+
"lstrip": false,
|
| 17 |
+
"normalized": false,
|
| 18 |
+
"rstrip": false,
|
| 19 |
+
"single_word": false,
|
| 20 |
+
"special": true
|
| 21 |
+
},
|
| 22 |
+
"2": {
|
| 23 |
+
"content": "</s>",
|
| 24 |
+
"lstrip": false,
|
| 25 |
+
"normalized": false,
|
| 26 |
+
"rstrip": false,
|
| 27 |
+
"single_word": false,
|
| 28 |
+
"special": true
|
| 29 |
+
},
|
| 30 |
+
"32000": {
|
| 31 |
+
"content": "[PAD]",
|
| 32 |
+
"lstrip": false,
|
| 33 |
+
"normalized": false,
|
| 34 |
+
"rstrip": false,
|
| 35 |
+
"single_word": false,
|
| 36 |
+
"special": true
|
| 37 |
+
}
|
| 38 |
+
},
|
| 39 |
+
"bos_token": "</s>",
|
| 40 |
+
"clean_up_tokenization_spaces": false,
|
| 41 |
+
"eos_token": "</s>",
|
| 42 |
+
"extra_special_tokens": {},
|
| 43 |
+
"legacy": false,
|
| 44 |
+
"model_max_length": 512,
|
| 45 |
+
"pad_token": "[PAD]",
|
| 46 |
+
"padding_side": "right",
|
| 47 |
+
"sp_model_kwargs": {},
|
| 48 |
+
"tokenizer_class": "LlamaTokenizer",
|
| 49 |
+
"unk_token": "</s>",
|
| 50 |
+
"use_default_system_prompt": false
|
| 51 |
+
}
|
nl_tasks/exps/run_ex14_3ep/ft2/adapter_config.json
ADDED
|
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"T": 1.0,
|
| 3 |
+
"base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
|
| 4 |
+
"bias": "none",
|
| 5 |
+
"inference_mode": true,
|
| 6 |
+
"layers_to_transform": null,
|
| 7 |
+
"modules_to_save": null,
|
| 8 |
+
"num_rotations": 1,
|
| 9 |
+
"peft_type": "ROTATION",
|
| 10 |
+
"r": 16,
|
| 11 |
+
"revision": null,
|
| 12 |
+
"target_modules": [
|
| 13 |
+
"v_proj",
|
| 14 |
+
"q_proj"
|
| 15 |
+
],
|
| 16 |
+
"target_modules_to_skip": null,
|
| 17 |
+
"task_type": "CAUSAL_LM"
|
| 18 |
+
}
|
nl_tasks/exps/run_ex14_3ep/ft2/adapter_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0a45591967cec76d0355af4114fe571940349ed79455a6edefbafc7a00143d30
|
| 3 |
+
size 33602915
|
nl_tasks/exps/run_ex14_3ep/trainer_state.json
ADDED
|
@@ -0,0 +1,1209 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 3.0,
|
| 6 |
+
"eval_steps": 200,
|
| 7 |
+
"global_step": 3657,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.020508613617719443,
|
| 14 |
+
"grad_norm": 0.12512381374835968,
|
| 15 |
+
"learning_rate": 1.3114754098360657e-05,
|
| 16 |
+
"loss": 0.7486,
|
| 17 |
+
"step": 25
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.04101722723543889,
|
| 21 |
+
"grad_norm": 0.25120121240615845,
|
| 22 |
+
"learning_rate": 2.677595628415301e-05,
|
| 23 |
+
"loss": 0.6624,
|
| 24 |
+
"step": 50
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.06152584085315833,
|
| 28 |
+
"grad_norm": 0.14835110306739807,
|
| 29 |
+
"learning_rate": 4.0437158469945356e-05,
|
| 30 |
+
"loss": 0.4793,
|
| 31 |
+
"step": 75
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.08203445447087777,
|
| 35 |
+
"grad_norm": 0.17858591675758362,
|
| 36 |
+
"learning_rate": 5.409836065573771e-05,
|
| 37 |
+
"loss": 0.4029,
|
| 38 |
+
"step": 100
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.10254306808859721,
|
| 42 |
+
"grad_norm": 0.18509596586227417,
|
| 43 |
+
"learning_rate": 6.775956284153006e-05,
|
| 44 |
+
"loss": 0.386,
|
| 45 |
+
"step": 125
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.12305168170631665,
|
| 49 |
+
"grad_norm": 0.20203717052936554,
|
| 50 |
+
"learning_rate": 8.14207650273224e-05,
|
| 51 |
+
"loss": 0.3596,
|
| 52 |
+
"step": 150
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.1435602953240361,
|
| 56 |
+
"grad_norm": 0.21976588666439056,
|
| 57 |
+
"learning_rate": 9.508196721311476e-05,
|
| 58 |
+
"loss": 0.3431,
|
| 59 |
+
"step": 175
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.16406890894175555,
|
| 63 |
+
"grad_norm": 0.2748827636241913,
|
| 64 |
+
"learning_rate": 0.00010874316939890712,
|
| 65 |
+
"loss": 0.3476,
|
| 66 |
+
"step": 200
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.16406890894175555,
|
| 70 |
+
"eval_loss": 0.32849055528640747,
|
| 71 |
+
"eval_runtime": 19.5315,
|
| 72 |
+
"eval_samples_per_second": 51.199,
|
| 73 |
+
"eval_steps_per_second": 0.819,
|
| 74 |
+
"step": 200
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.184577522559475,
|
| 78 |
+
"grad_norm": 0.2196296602487564,
|
| 79 |
+
"learning_rate": 0.00012240437158469945,
|
| 80 |
+
"loss": 0.3411,
|
| 81 |
+
"step": 225
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.20508613617719443,
|
| 85 |
+
"grad_norm": 0.21974113583564758,
|
| 86 |
+
"learning_rate": 0.0001360655737704918,
|
| 87 |
+
"loss": 0.3171,
|
| 88 |
+
"step": 250
|
| 89 |
+
},
|
| 90 |
+
{
|
| 91 |
+
"epoch": 0.22559474979491387,
|
| 92 |
+
"grad_norm": 0.19429394602775574,
|
| 93 |
+
"learning_rate": 0.00014972677595628418,
|
| 94 |
+
"loss": 0.3136,
|
| 95 |
+
"step": 275
|
| 96 |
+
},
|
| 97 |
+
{
|
| 98 |
+
"epoch": 0.2461033634126333,
|
| 99 |
+
"grad_norm": 0.21085946261882782,
|
| 100 |
+
"learning_rate": 0.0001633879781420765,
|
| 101 |
+
"loss": 0.3184,
|
| 102 |
+
"step": 300
|
| 103 |
+
},
|
| 104 |
+
{
|
| 105 |
+
"epoch": 0.2666119770303528,
|
| 106 |
+
"grad_norm": 0.22968199849128723,
|
| 107 |
+
"learning_rate": 0.00017704918032786885,
|
| 108 |
+
"loss": 0.3208,
|
| 109 |
+
"step": 325
|
| 110 |
+
},
|
| 111 |
+
{
|
| 112 |
+
"epoch": 0.2871205906480722,
|
| 113 |
+
"grad_norm": 0.1966412514448166,
|
| 114 |
+
"learning_rate": 0.0001907103825136612,
|
| 115 |
+
"loss": 0.3119,
|
| 116 |
+
"step": 350
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.30762920426579166,
|
| 120 |
+
"grad_norm": 0.21624255180358887,
|
| 121 |
+
"learning_rate": 0.00019999708397093504,
|
| 122 |
+
"loss": 0.3226,
|
| 123 |
+
"step": 375
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 0.3281378178835111,
|
| 127 |
+
"grad_norm": 0.20411108434200287,
|
| 128 |
+
"learning_rate": 0.00019995038580495014,
|
| 129 |
+
"loss": 0.3055,
|
| 130 |
+
"step": 400
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 0.3281378178835111,
|
| 134 |
+
"eval_loss": 0.295260488986969,
|
| 135 |
+
"eval_runtime": 19.3475,
|
| 136 |
+
"eval_samples_per_second": 51.686,
|
| 137 |
+
"eval_steps_per_second": 0.827,
|
| 138 |
+
"step": 400
|
| 139 |
+
},
|
| 140 |
+
{
|
| 141 |
+
"epoch": 0.34864643150123054,
|
| 142 |
+
"grad_norm": 0.1912434846162796,
|
| 143 |
+
"learning_rate": 0.0001998467646284801,
|
| 144 |
+
"loss": 0.3129,
|
| 145 |
+
"step": 425
|
| 146 |
+
},
|
| 147 |
+
{
|
| 148 |
+
"epoch": 0.36915504511895,
|
| 149 |
+
"grad_norm": 0.18060387670993805,
|
| 150 |
+
"learning_rate": 0.00019968627945509725,
|
| 151 |
+
"loss": 0.309,
|
| 152 |
+
"step": 450
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.3896636587366694,
|
| 156 |
+
"grad_norm": 0.18696913123130798,
|
| 157 |
+
"learning_rate": 0.00019946902168314012,
|
| 158 |
+
"loss": 0.2988,
|
| 159 |
+
"step": 475
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.41017227235438886,
|
| 163 |
+
"grad_norm": 0.1760314702987671,
|
| 164 |
+
"learning_rate": 0.00019919511504366125,
|
| 165 |
+
"loss": 0.2817,
|
| 166 |
+
"step": 500
|
| 167 |
+
},
|
| 168 |
+
{
|
| 169 |
+
"epoch": 0.4306808859721083,
|
| 170 |
+
"grad_norm": 0.1863497942686081,
|
| 171 |
+
"learning_rate": 0.00019886471552996054,
|
| 172 |
+
"loss": 0.2823,
|
| 173 |
+
"step": 525
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 0.45118949958982774,
|
| 177 |
+
"grad_norm": 0.17864450812339783,
|
| 178 |
+
"learning_rate": 0.00019847801130874523,
|
| 179 |
+
"loss": 0.2905,
|
| 180 |
+
"step": 550
|
| 181 |
+
},
|
| 182 |
+
{
|
| 183 |
+
"epoch": 0.4716981132075472,
|
| 184 |
+
"grad_norm": 0.19246388971805573,
|
| 185 |
+
"learning_rate": 0.00019803522261296652,
|
| 186 |
+
"loss": 0.2842,
|
| 187 |
+
"step": 575
|
| 188 |
+
},
|
| 189 |
+
{
|
| 190 |
+
"epoch": 0.4922067268252666,
|
| 191 |
+
"grad_norm": 0.1875869631767273,
|
| 192 |
+
"learning_rate": 0.00019753660161639407,
|
| 193 |
+
"loss": 0.3009,
|
| 194 |
+
"step": 600
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 0.4922067268252666,
|
| 198 |
+
"eval_loss": 0.27711474895477295,
|
| 199 |
+
"eval_runtime": 19.3054,
|
| 200 |
+
"eval_samples_per_second": 51.799,
|
| 201 |
+
"eval_steps_per_second": 0.829,
|
| 202 |
+
"step": 600
|
| 203 |
+
},
|
| 204 |
+
{
|
| 205 |
+
"epoch": 0.5127153404429861,
|
| 206 |
+
"grad_norm": 0.17517873644828796,
|
| 207 |
+
"learning_rate": 0.00019698243229000003,
|
| 208 |
+
"loss": 0.279,
|
| 209 |
+
"step": 625
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 0.5332239540607056,
|
| 213 |
+
"grad_norm": 0.19444359838962555,
|
| 214 |
+
"learning_rate": 0.0001963730302402337,
|
| 215 |
+
"loss": 0.2718,
|
| 216 |
+
"step": 650
|
| 217 |
+
},
|
| 218 |
+
{
|
| 219 |
+
"epoch": 0.5537325676784249,
|
| 220 |
+
"grad_norm": 0.17459408938884735,
|
| 221 |
+
"learning_rate": 0.0001957087425292799,
|
| 222 |
+
"loss": 0.2846,
|
| 223 |
+
"step": 675
|
| 224 |
+
},
|
| 225 |
+
{
|
| 226 |
+
"epoch": 0.5742411812961444,
|
| 227 |
+
"grad_norm": 0.19231507182121277,
|
| 228 |
+
"learning_rate": 0.0001949899474774026,
|
| 229 |
+
"loss": 0.2822,
|
| 230 |
+
"step": 700
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 0.5947497949138638,
|
| 234 |
+
"grad_norm": 0.17913177609443665,
|
| 235 |
+
"learning_rate": 0.00019421705444748653,
|
| 236 |
+
"loss": 0.2803,
|
| 237 |
+
"step": 725
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 0.6152584085315833,
|
| 241 |
+
"grad_norm": 0.1698766052722931,
|
| 242 |
+
"learning_rate": 0.0001933905036119002,
|
| 243 |
+
"loss": 0.2789,
|
| 244 |
+
"step": 750
|
| 245 |
+
},
|
| 246 |
+
{
|
| 247 |
+
"epoch": 0.6357670221493027,
|
| 248 |
+
"grad_norm": 0.1915999799966812,
|
| 249 |
+
"learning_rate": 0.0001925107657018119,
|
| 250 |
+
"loss": 0.2746,
|
| 251 |
+
"step": 775
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 0.6562756357670222,
|
| 255 |
+
"grad_norm": 0.18614713847637177,
|
| 256 |
+
"learning_rate": 0.00019157834173910235,
|
| 257 |
+
"loss": 0.2797,
|
| 258 |
+
"step": 800
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 0.6562756357670222,
|
| 262 |
+
"eval_loss": 0.2656422257423401,
|
| 263 |
+
"eval_runtime": 19.3095,
|
| 264 |
+
"eval_samples_per_second": 51.788,
|
| 265 |
+
"eval_steps_per_second": 0.829,
|
| 266 |
+
"step": 800
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 0.6767842493847416,
|
| 270 |
+
"grad_norm": 0.1689230501651764,
|
| 271 |
+
"learning_rate": 0.00019059376275102656,
|
| 272 |
+
"loss": 0.2718,
|
| 273 |
+
"step": 825
|
| 274 |
+
},
|
| 275 |
+
{
|
| 276 |
+
"epoch": 0.6972928630024611,
|
| 277 |
+
"grad_norm": 0.18635261058807373,
|
| 278 |
+
"learning_rate": 0.0001895575894677869,
|
| 279 |
+
"loss": 0.2744,
|
| 280 |
+
"step": 850
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 0.7178014766201805,
|
| 284 |
+
"grad_norm": 0.17009536921977997,
|
| 285 |
+
"learning_rate": 0.00018847041200319026,
|
| 286 |
+
"loss": 0.2644,
|
| 287 |
+
"step": 875
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 0.7383100902379,
|
| 291 |
+
"grad_norm": 0.20701751112937927,
|
| 292 |
+
"learning_rate": 0.0001873328495185707,
|
| 293 |
+
"loss": 0.271,
|
| 294 |
+
"step": 900
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 0.7588187038556193,
|
| 298 |
+
"grad_norm": 0.1744987815618515,
|
| 299 |
+
"learning_rate": 0.0001861455498701693,
|
| 300 |
+
"loss": 0.2701,
|
| 301 |
+
"step": 925
|
| 302 |
+
},
|
| 303 |
+
{
|
| 304 |
+
"epoch": 0.7793273174733388,
|
| 305 |
+
"grad_norm": 0.18133293092250824,
|
| 306 |
+
"learning_rate": 0.00018490918924017177,
|
| 307 |
+
"loss": 0.2595,
|
| 308 |
+
"step": 950
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 0.7998359310910582,
|
| 312 |
+
"grad_norm": 0.20107491314411163,
|
| 313 |
+
"learning_rate": 0.00018362447175161403,
|
| 314 |
+
"loss": 0.2648,
|
| 315 |
+
"step": 975
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 0.8203445447087777,
|
| 319 |
+
"grad_norm": 0.18567617237567902,
|
| 320 |
+
"learning_rate": 0.00018229212906737557,
|
| 321 |
+
"loss": 0.2575,
|
| 322 |
+
"step": 1000
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 0.8203445447087777,
|
| 326 |
+
"eval_loss": 0.25553619861602783,
|
| 327 |
+
"eval_runtime": 19.3337,
|
| 328 |
+
"eval_samples_per_second": 51.723,
|
| 329 |
+
"eval_steps_per_second": 0.828,
|
| 330 |
+
"step": 1000
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 0.8408531583264971,
|
| 334 |
+
"grad_norm": 0.20225924253463745,
|
| 335 |
+
"learning_rate": 0.00018091291997348775,
|
| 336 |
+
"loss": 0.2595,
|
| 337 |
+
"step": 1025
|
| 338 |
+
},
|
| 339 |
+
{
|
| 340 |
+
"epoch": 0.8613617719442166,
|
| 341 |
+
"grad_norm": 0.20732153952121735,
|
| 342 |
+
"learning_rate": 0.0001794876299469958,
|
| 343 |
+
"loss": 0.2584,
|
| 344 |
+
"step": 1050
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 0.881870385561936,
|
| 348 |
+
"grad_norm": 0.20521634817123413,
|
| 349 |
+
"learning_rate": 0.00017801707070861972,
|
| 350 |
+
"loss": 0.2631,
|
| 351 |
+
"step": 1075
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 0.9023789991796555,
|
| 355 |
+
"grad_norm": 0.2013619989156723,
|
| 356 |
+
"learning_rate": 0.00017650207976046883,
|
| 357 |
+
"loss": 0.2628,
|
| 358 |
+
"step": 1100
|
| 359 |
+
},
|
| 360 |
+
{
|
| 361 |
+
"epoch": 0.9228876127973749,
|
| 362 |
+
"grad_norm": 0.18652208149433136,
|
| 363 |
+
"learning_rate": 0.0001749435199090739,
|
| 364 |
+
"loss": 0.2579,
|
| 365 |
+
"step": 1125
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 0.9433962264150944,
|
| 369 |
+
"grad_norm": 0.1730799525976181,
|
| 370 |
+
"learning_rate": 0.00017334227877400791,
|
| 371 |
+
"loss": 0.257,
|
| 372 |
+
"step": 1150
|
| 373 |
+
},
|
| 374 |
+
{
|
| 375 |
+
"epoch": 0.9639048400328137,
|
| 376 |
+
"grad_norm": 0.17982584238052368,
|
| 377 |
+
"learning_rate": 0.0001716992682823753,
|
| 378 |
+
"loss": 0.2632,
|
| 379 |
+
"step": 1175
|
| 380 |
+
},
|
| 381 |
+
{
|
| 382 |
+
"epoch": 0.9844134536505332,
|
| 383 |
+
"grad_norm": 0.18876557052135468,
|
| 384 |
+
"learning_rate": 0.00017001542414945815,
|
| 385 |
+
"loss": 0.2692,
|
| 386 |
+
"step": 1200
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 0.9844134536505332,
|
| 390 |
+
"eval_loss": 0.24926717579364777,
|
| 391 |
+
"eval_runtime": 19.3053,
|
| 392 |
+
"eval_samples_per_second": 51.799,
|
| 393 |
+
"eval_steps_per_second": 0.829,
|
| 394 |
+
"step": 1200
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 1.0049220672682526,
|
| 398 |
+
"grad_norm": 0.18879544734954834,
|
| 399 |
+
"learning_rate": 0.0001682917053458143,
|
| 400 |
+
"loss": 0.2531,
|
| 401 |
+
"step": 1225
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 1.0254306808859721,
|
| 405 |
+
"grad_norm": 0.18688759207725525,
|
| 406 |
+
"learning_rate": 0.00016652909355113153,
|
| 407 |
+
"loss": 0.2454,
|
| 408 |
+
"step": 1250
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 1.0459392945036916,
|
| 412 |
+
"grad_norm": 0.20011909306049347,
|
| 413 |
+
"learning_rate": 0.00016472859259514862,
|
| 414 |
+
"loss": 0.2472,
|
| 415 |
+
"step": 1275
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 1.066447908121411,
|
| 419 |
+
"grad_norm": 0.17656545341014862,
|
| 420 |
+
"learning_rate": 0.0001628912278859615,
|
| 421 |
+
"loss": 0.2501,
|
| 422 |
+
"step": 1300
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 1.0869565217391304,
|
| 426 |
+
"grad_norm": 0.19816625118255615,
|
| 427 |
+
"learning_rate": 0.00016101804582604036,
|
| 428 |
+
"loss": 0.2466,
|
| 429 |
+
"step": 1325
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 1.1074651353568499,
|
| 433 |
+
"grad_norm": 0.18136921525001526,
|
| 434 |
+
"learning_rate": 0.0001591101132162903,
|
| 435 |
+
"loss": 0.2372,
|
| 436 |
+
"step": 1350
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 1.1279737489745694,
|
| 440 |
+
"grad_norm": 0.22778378427028656,
|
| 441 |
+
"learning_rate": 0.00015716851664849456,
|
| 442 |
+
"loss": 0.2429,
|
| 443 |
+
"step": 1375
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 1.1484823625922886,
|
| 447 |
+
"grad_norm": 0.18908445537090302,
|
| 448 |
+
"learning_rate": 0.0001551943618864868,
|
| 449 |
+
"loss": 0.2323,
|
| 450 |
+
"step": 1400
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 1.1484823625922886,
|
| 454 |
+
"eval_loss": 0.24481597542762756,
|
| 455 |
+
"eval_runtime": 19.3242,
|
| 456 |
+
"eval_samples_per_second": 51.749,
|
| 457 |
+
"eval_steps_per_second": 0.828,
|
| 458 |
+
"step": 1400
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 1.1689909762100081,
|
| 462 |
+
"grad_norm": 0.19832967221736908,
|
| 463 |
+
"learning_rate": 0.0001531887732364047,
|
| 464 |
+
"loss": 0.2438,
|
| 465 |
+
"step": 1425
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 1.1894995898277276,
|
| 469 |
+
"grad_norm": 0.2384573370218277,
|
| 470 |
+
"learning_rate": 0.00015115289290638316,
|
| 471 |
+
"loss": 0.2472,
|
| 472 |
+
"step": 1450
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 1.2100082034454471,
|
| 476 |
+
"grad_norm": 0.2025284469127655,
|
| 477 |
+
"learning_rate": 0.00014908788035605247,
|
| 478 |
+
"loss": 0.2422,
|
| 479 |
+
"step": 1475
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 1.2305168170631666,
|
| 483 |
+
"grad_norm": 0.2065984159708023,
|
| 484 |
+
"learning_rate": 0.0001469949116362113,
|
| 485 |
+
"loss": 0.2513,
|
| 486 |
+
"step": 1500
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 1.251025430680886,
|
| 490 |
+
"grad_norm": 0.24289533495903015,
|
| 491 |
+
"learning_rate": 0.00014487517871905094,
|
| 492 |
+
"loss": 0.2464,
|
| 493 |
+
"step": 1525
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 1.2715340442986054,
|
| 497 |
+
"grad_norm": 0.1909610629081726,
|
| 498 |
+
"learning_rate": 0.0001427298888193119,
|
| 499 |
+
"loss": 0.2393,
|
| 500 |
+
"step": 1550
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 1.2920426579163249,
|
| 504 |
+
"grad_norm": 0.18158946931362152,
|
| 505 |
+
"learning_rate": 0.00014056026370676004,
|
| 506 |
+
"loss": 0.2414,
|
| 507 |
+
"step": 1575
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 1.3125512715340442,
|
| 511 |
+
"grad_norm": 0.19775278866291046,
|
| 512 |
+
"learning_rate": 0.000138367539010373,
|
| 513 |
+
"loss": 0.2331,
|
| 514 |
+
"step": 1600
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 1.3125512715340442,
|
| 518 |
+
"eval_loss": 0.23973043262958527,
|
| 519 |
+
"eval_runtime": 19.3129,
|
| 520 |
+
"eval_samples_per_second": 51.779,
|
| 521 |
+
"eval_steps_per_second": 0.828,
|
| 522 |
+
"step": 1600
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 1.3330598851517639,
|
| 526 |
+
"grad_norm": 0.1940084993839264,
|
| 527 |
+
"learning_rate": 0.000136152963514634,
|
| 528 |
+
"loss": 0.2364,
|
| 529 |
+
"step": 1625
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 1.3535684987694832,
|
| 533 |
+
"grad_norm": 0.21692197024822235,
|
| 534 |
+
"learning_rate": 0.0001339177984483332,
|
| 535 |
+
"loss": 0.2372,
|
| 536 |
+
"step": 1650
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 1.3740771123872026,
|
| 540 |
+
"grad_norm": 0.22663699090480804,
|
| 541 |
+
"learning_rate": 0.0001316633167662819,
|
| 542 |
+
"loss": 0.2347,
|
| 543 |
+
"step": 1675
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 1.3945857260049221,
|
| 547 |
+
"grad_norm": 0.20252764225006104,
|
| 548 |
+
"learning_rate": 0.00012939080242434876,
|
| 549 |
+
"loss": 0.2336,
|
| 550 |
+
"step": 1700
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 1.4150943396226414,
|
| 554 |
+
"grad_norm": 0.2129640132188797,
|
| 555 |
+
"learning_rate": 0.00012710154964823054,
|
| 556 |
+
"loss": 0.2312,
|
| 557 |
+
"step": 1725
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 1.435602953240361,
|
| 561 |
+
"grad_norm": 0.1916554719209671,
|
| 562 |
+
"learning_rate": 0.00012479686219637432,
|
| 563 |
+
"loss": 0.2319,
|
| 564 |
+
"step": 1750
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 1.4561115668580804,
|
| 568 |
+
"grad_norm": 0.2715451419353485,
|
| 569 |
+
"learning_rate": 0.00012247805261747044,
|
| 570 |
+
"loss": 0.2386,
|
| 571 |
+
"step": 1775
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 1.4766201804758,
|
| 575 |
+
"grad_norm": 0.196442648768425,
|
| 576 |
+
"learning_rate": 0.00012014644150293936,
|
| 577 |
+
"loss": 0.2273,
|
| 578 |
+
"step": 1800
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 1.4766201804758,
|
| 582 |
+
"eval_loss": 0.23566412925720215,
|
| 583 |
+
"eval_runtime": 19.3249,
|
| 584 |
+
"eval_samples_per_second": 51.747,
|
| 585 |
+
"eval_steps_per_second": 0.828,
|
| 586 |
+
"step": 1800
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 1.4971287940935194,
|
| 590 |
+
"grad_norm": 0.19579745829105377,
|
| 591 |
+
"learning_rate": 0.00011780335673483838,
|
| 592 |
+
"loss": 0.2282,
|
| 593 |
+
"step": 1825
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 1.5176374077112387,
|
| 597 |
+
"grad_norm": 0.19303074479103088,
|
| 598 |
+
"learning_rate": 0.00011545013272961563,
|
| 599 |
+
"loss": 0.2302,
|
| 600 |
+
"step": 1850
|
| 601 |
+
},
|
| 602 |
+
{
|
| 603 |
+
"epoch": 1.5381460213289582,
|
| 604 |
+
"grad_norm": 0.19902493059635162,
|
| 605 |
+
"learning_rate": 0.00011308810967814325,
|
| 606 |
+
"loss": 0.23,
|
| 607 |
+
"step": 1875
|
| 608 |
+
},
|
| 609 |
+
{
|
| 610 |
+
"epoch": 1.5586546349466777,
|
| 611 |
+
"grad_norm": 0.2139587551355362,
|
| 612 |
+
"learning_rate": 0.00011071863278246166,
|
| 613 |
+
"loss": 0.2341,
|
| 614 |
+
"step": 1900
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 1.579163248564397,
|
| 618 |
+
"grad_norm": 0.2020283043384552,
|
| 619 |
+
"learning_rate": 0.00010834305148966999,
|
| 620 |
+
"loss": 0.2367,
|
| 621 |
+
"step": 1925
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 1.5996718621821167,
|
| 625 |
+
"grad_norm": 0.21497027575969696,
|
| 626 |
+
"learning_rate": 0.00010596271872339876,
|
| 627 |
+
"loss": 0.2348,
|
| 628 |
+
"step": 1950
|
| 629 |
+
},
|
| 630 |
+
{
|
| 631 |
+
"epoch": 1.620180475799836,
|
| 632 |
+
"grad_norm": 0.20861054956912994,
|
| 633 |
+
"learning_rate": 0.0001035789901133028,
|
| 634 |
+
"loss": 0.2271,
|
| 635 |
+
"step": 1975
|
| 636 |
+
},
|
| 637 |
+
{
|
| 638 |
+
"epoch": 1.6406890894175554,
|
| 639 |
+
"grad_norm": 0.21709710359573364,
|
| 640 |
+
"learning_rate": 0.0001011932232230131,
|
| 641 |
+
"loss": 0.2348,
|
| 642 |
+
"step": 2000
|
| 643 |
+
},
|
| 644 |
+
{
|
| 645 |
+
"epoch": 1.6406890894175554,
|
| 646 |
+
"eval_loss": 0.2315228432416916,
|
| 647 |
+
"eval_runtime": 19.345,
|
| 648 |
+
"eval_samples_per_second": 51.693,
|
| 649 |
+
"eval_steps_per_second": 0.827,
|
| 650 |
+
"step": 2000
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 1.661197703035275,
|
| 654 |
+
"grad_norm": 0.20213831961154938,
|
| 655 |
+
"learning_rate": 9.880677677698695e-05,
|
| 656 |
+
"loss": 0.2244,
|
| 657 |
+
"step": 2025
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 1.6817063166529942,
|
| 661 |
+
"grad_norm": 0.22401712834835052,
|
| 662 |
+
"learning_rate": 9.642100988669722e-05,
|
| 663 |
+
"loss": 0.2284,
|
| 664 |
+
"step": 2050
|
| 665 |
+
},
|
| 666 |
+
{
|
| 667 |
+
"epoch": 1.7022149302707137,
|
| 668 |
+
"grad_norm": 0.25369125604629517,
|
| 669 |
+
"learning_rate": 9.403728127660129e-05,
|
| 670 |
+
"loss": 0.2337,
|
| 671 |
+
"step": 2075
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 1.7227235438884332,
|
| 675 |
+
"grad_norm": 0.2132670134305954,
|
| 676 |
+
"learning_rate": 9.165694851033002e-05,
|
| 677 |
+
"loss": 0.221,
|
| 678 |
+
"step": 2100
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 1.7432321575061525,
|
| 682 |
+
"grad_norm": 0.18563179671764374,
|
| 683 |
+
"learning_rate": 8.928136721753837e-05,
|
| 684 |
+
"loss": 0.227,
|
| 685 |
+
"step": 2125
|
| 686 |
+
},
|
| 687 |
+
{
|
| 688 |
+
"epoch": 1.7637407711238722,
|
| 689 |
+
"grad_norm": 0.1770583838224411,
|
| 690 |
+
"learning_rate": 8.691189032185678e-05,
|
| 691 |
+
"loss": 0.2248,
|
| 692 |
+
"step": 2150
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 1.7842493847415914,
|
| 696 |
+
"grad_norm": 0.19352003931999207,
|
| 697 |
+
"learning_rate": 8.454986727038442e-05,
|
| 698 |
+
"loss": 0.2325,
|
| 699 |
+
"step": 2175
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 1.804757998359311,
|
| 703 |
+
"grad_norm": 0.1982247531414032,
|
| 704 |
+
"learning_rate": 8.219664326516165e-05,
|
| 705 |
+
"loss": 0.2261,
|
| 706 |
+
"step": 2200
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 1.804757998359311,
|
| 710 |
+
"eval_loss": 0.22743363678455353,
|
| 711 |
+
"eval_runtime": 19.3486,
|
| 712 |
+
"eval_samples_per_second": 51.683,
|
| 713 |
+
"eval_steps_per_second": 0.827,
|
| 714 |
+
"step": 2200
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 1.8252666119770304,
|
| 718 |
+
"grad_norm": 0.21940328180789948,
|
| 719 |
+
"learning_rate": 7.985355849706065e-05,
|
| 720 |
+
"loss": 0.2266,
|
| 721 |
+
"step": 2225
|
| 722 |
+
},
|
| 723 |
+
{
|
| 724 |
+
"epoch": 1.8457752255947497,
|
| 725 |
+
"grad_norm": 0.19998738169670105,
|
| 726 |
+
"learning_rate": 7.752194738252958e-05,
|
| 727 |
+
"loss": 0.2279,
|
| 728 |
+
"step": 2250
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 1.8662838392124692,
|
| 732 |
+
"grad_norm": 0.19756212830543518,
|
| 733 |
+
"learning_rate": 7.520313780362567e-05,
|
| 734 |
+
"loss": 0.2322,
|
| 735 |
+
"step": 2275
|
| 736 |
+
},
|
| 737 |
+
{
|
| 738 |
+
"epoch": 1.8867924528301887,
|
| 739 |
+
"grad_norm": 0.20220190286636353,
|
| 740 |
+
"learning_rate": 7.289845035176947e-05,
|
| 741 |
+
"loss": 0.225,
|
| 742 |
+
"step": 2300
|
| 743 |
+
},
|
| 744 |
+
{
|
| 745 |
+
"epoch": 1.907301066447908,
|
| 746 |
+
"grad_norm": 0.228904590010643,
|
| 747 |
+
"learning_rate": 7.060919757565126e-05,
|
| 748 |
+
"loss": 0.2326,
|
| 749 |
+
"step": 2325
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 1.9278096800656277,
|
| 753 |
+
"grad_norm": 0.20801404118537903,
|
| 754 |
+
"learning_rate": 6.833668323371814e-05,
|
| 755 |
+
"loss": 0.2151,
|
| 756 |
+
"step": 2350
|
| 757 |
+
},
|
| 758 |
+
{
|
| 759 |
+
"epoch": 1.948318293683347,
|
| 760 |
+
"grad_norm": 0.18295502662658691,
|
| 761 |
+
"learning_rate": 6.608220155166682e-05,
|
| 762 |
+
"loss": 0.2217,
|
| 763 |
+
"step": 2375
|
| 764 |
+
},
|
| 765 |
+
{
|
| 766 |
+
"epoch": 1.9688269073010665,
|
| 767 |
+
"grad_norm": 0.20270372927188873,
|
| 768 |
+
"learning_rate": 6.384703648536601e-05,
|
| 769 |
+
"loss": 0.2232,
|
| 770 |
+
"step": 2400
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 1.9688269073010665,
|
| 774 |
+
"eval_loss": 0.2246938943862915,
|
| 775 |
+
"eval_runtime": 19.3031,
|
| 776 |
+
"eval_samples_per_second": 51.805,
|
| 777 |
+
"eval_steps_per_second": 0.829,
|
| 778 |
+
"step": 2400
|
| 779 |
+
},
|
| 780 |
+
{
|
| 781 |
+
"epoch": 1.989335520918786,
|
| 782 |
+
"grad_norm": 0.2036377638578415,
|
| 783 |
+
"learning_rate": 6.1632460989627e-05,
|
| 784 |
+
"loss": 0.2191,
|
| 785 |
+
"step": 2425
|
| 786 |
+
},
|
| 787 |
+
{
|
| 788 |
+
"epoch": 2.0098441345365052,
|
| 789 |
+
"grad_norm": 0.2158953845500946,
|
| 790 |
+
"learning_rate": 5.943973629324e-05,
|
| 791 |
+
"loss": 0.2161,
|
| 792 |
+
"step": 2450
|
| 793 |
+
},
|
| 794 |
+
{
|
| 795 |
+
"epoch": 2.030352748154225,
|
| 796 |
+
"grad_norm": 0.22983403503894806,
|
| 797 |
+
"learning_rate": 5.727011118068812e-05,
|
| 798 |
+
"loss": 0.2091,
|
| 799 |
+
"step": 2475
|
| 800 |
+
},
|
| 801 |
+
{
|
| 802 |
+
"epoch": 2.0508613617719442,
|
| 803 |
+
"grad_norm": 0.24088236689567566,
|
| 804 |
+
"learning_rate": 5.512482128094912e-05,
|
| 805 |
+
"loss": 0.2122,
|
| 806 |
+
"step": 2500
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 2.0713699753896635,
|
| 810 |
+
"grad_norm": 0.21485844254493713,
|
| 811 |
+
"learning_rate": 5.3005088363788744e-05,
|
| 812 |
+
"loss": 0.2046,
|
| 813 |
+
"step": 2525
|
| 814 |
+
},
|
| 815 |
+
{
|
| 816 |
+
"epoch": 2.091878589007383,
|
| 817 |
+
"grad_norm": 0.22207602858543396,
|
| 818 |
+
"learning_rate": 5.0912119643947564e-05,
|
| 819 |
+
"loss": 0.2133,
|
| 820 |
+
"step": 2550
|
| 821 |
+
},
|
| 822 |
+
{
|
| 823 |
+
"epoch": 2.1123872026251025,
|
| 824 |
+
"grad_norm": 0.2140221893787384,
|
| 825 |
+
"learning_rate": 4.8847107093616854e-05,
|
| 826 |
+
"loss": 0.2057,
|
| 827 |
+
"step": 2575
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"epoch": 2.132895816242822,
|
| 831 |
+
"grad_norm": 0.22705940902233124,
|
| 832 |
+
"learning_rate": 4.681122676359533e-05,
|
| 833 |
+
"loss": 0.1985,
|
| 834 |
+
"step": 2600
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 2.132895816242822,
|
| 838 |
+
"eval_loss": 0.22434084117412567,
|
| 839 |
+
"eval_runtime": 19.3818,
|
| 840 |
+
"eval_samples_per_second": 51.595,
|
| 841 |
+
"eval_steps_per_second": 0.826,
|
| 842 |
+
"step": 2600
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 2.1534044298605415,
|
| 846 |
+
"grad_norm": 0.2110925018787384,
|
| 847 |
+
"learning_rate": 4.4805638113513203e-05,
|
| 848 |
+
"loss": 0.2021,
|
| 849 |
+
"step": 2625
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 2.1739130434782608,
|
| 853 |
+
"grad_norm": 0.21700389683246613,
|
| 854 |
+
"learning_rate": 4.2831483351505485e-05,
|
| 855 |
+
"loss": 0.2014,
|
| 856 |
+
"step": 2650
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 2.1944216570959805,
|
| 860 |
+
"grad_norm": 0.2308293730020523,
|
| 861 |
+
"learning_rate": 4.088988678370973e-05,
|
| 862 |
+
"loss": 0.215,
|
| 863 |
+
"step": 2675
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 2.2149302707136997,
|
| 867 |
+
"grad_norm": 0.2091713398694992,
|
| 868 |
+
"learning_rate": 3.8981954173959644e-05,
|
| 869 |
+
"loss": 0.209,
|
| 870 |
+
"step": 2700
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 2.235438884331419,
|
| 874 |
+
"grad_norm": 0.22311793267726898,
|
| 875 |
+
"learning_rate": 3.71087721140385e-05,
|
| 876 |
+
"loss": 0.1963,
|
| 877 |
+
"step": 2725
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 2.2559474979491387,
|
| 881 |
+
"grad_norm": 0.23842178285121918,
|
| 882 |
+
"learning_rate": 3.5271407404851366e-05,
|
| 883 |
+
"loss": 0.2096,
|
| 884 |
+
"step": 2750
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 2.276456111566858,
|
| 888 |
+
"grad_norm": 0.22167986631393433,
|
| 889 |
+
"learning_rate": 3.3470906448868476e-05,
|
| 890 |
+
"loss": 0.2071,
|
| 891 |
+
"step": 2775
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 2.2969647251845773,
|
| 895 |
+
"grad_norm": 0.23508331179618835,
|
| 896 |
+
"learning_rate": 3.170829465418571e-05,
|
| 897 |
+
"loss": 0.2041,
|
| 898 |
+
"step": 2800
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 2.2969647251845773,
|
| 902 |
+
"eval_loss": 0.2228643149137497,
|
| 903 |
+
"eval_runtime": 19.3503,
|
| 904 |
+
"eval_samples_per_second": 51.679,
|
| 905 |
+
"eval_steps_per_second": 0.827,
|
| 906 |
+
"step": 2800
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 2.317473338802297,
|
| 910 |
+
"grad_norm": 0.21861039102077484,
|
| 911 |
+
"learning_rate": 2.9984575850541875e-05,
|
| 912 |
+
"loss": 0.1987,
|
| 913 |
+
"step": 2825
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 2.3379819524200163,
|
| 917 |
+
"grad_norm": 0.23032869398593903,
|
| 918 |
+
"learning_rate": 2.8300731717624707e-05,
|
| 919 |
+
"loss": 0.2025,
|
| 920 |
+
"step": 2850
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 2.358490566037736,
|
| 924 |
+
"grad_norm": 0.20549559593200684,
|
| 925 |
+
"learning_rate": 2.6657721225992105e-05,
|
| 926 |
+
"loss": 0.203,
|
| 927 |
+
"step": 2875
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 2.3789991796554553,
|
| 931 |
+
"grad_norm": 0.2443164736032486,
|
| 932 |
+
"learning_rate": 2.5056480090926092e-05,
|
| 933 |
+
"loss": 0.2074,
|
| 934 |
+
"step": 2900
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 2.399507793273175,
|
| 938 |
+
"grad_norm": 0.2777072787284851,
|
| 939 |
+
"learning_rate": 2.349792023953119e-05,
|
| 940 |
+
"loss": 0.205,
|
| 941 |
+
"step": 2925
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 2.4200164068908943,
|
| 945 |
+
"grad_norm": 0.2372252494096756,
|
| 946 |
+
"learning_rate": 2.1982929291380284e-05,
|
| 947 |
+
"loss": 0.2065,
|
| 948 |
+
"step": 2950
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 2.4405250205086135,
|
| 952 |
+
"grad_norm": 0.22545471787452698,
|
| 953 |
+
"learning_rate": 2.0512370053004193e-05,
|
| 954 |
+
"loss": 0.2063,
|
| 955 |
+
"step": 2975
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 2.4610336341263332,
|
| 959 |
+
"grad_norm": 0.2503770589828491,
|
| 960 |
+
"learning_rate": 1.908708002651227e-05,
|
| 961 |
+
"loss": 0.1983,
|
| 962 |
+
"step": 3000
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 2.4610336341263332,
|
| 966 |
+
"eval_loss": 0.2219480276107788,
|
| 967 |
+
"eval_runtime": 19.2886,
|
| 968 |
+
"eval_samples_per_second": 51.844,
|
| 969 |
+
"eval_steps_per_second": 0.83,
|
| 970 |
+
"step": 3000
|
| 971 |
+
},
|
| 972 |
+
{
|
| 973 |
+
"epoch": 2.4815422477440525,
|
| 974 |
+
"grad_norm": 0.23392586410045624,
|
| 975 |
+
"learning_rate": 1.770787093262445e-05,
|
| 976 |
+
"loss": 0.2023,
|
| 977 |
+
"step": 3025
|
| 978 |
+
},
|
| 979 |
+
{
|
| 980 |
+
"epoch": 2.502050861361772,
|
| 981 |
+
"grad_norm": 0.20792211592197418,
|
| 982 |
+
"learning_rate": 1.6375528248385963e-05,
|
| 983 |
+
"loss": 0.1994,
|
| 984 |
+
"step": 3050
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 2.5225594749794915,
|
| 988 |
+
"grad_norm": 0.24046897888183594,
|
| 989 |
+
"learning_rate": 1.5090810759828278e-05,
|
| 990 |
+
"loss": 0.1952,
|
| 991 |
+
"step": 3075
|
| 992 |
+
},
|
| 993 |
+
{
|
| 994 |
+
"epoch": 2.543068088597211,
|
| 995 |
+
"grad_norm": 0.23085150122642517,
|
| 996 |
+
"learning_rate": 1.3854450129830699e-05,
|
| 997 |
+
"loss": 0.2087,
|
| 998 |
+
"step": 3100
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 2.56357670221493,
|
| 1002 |
+
"grad_norm": 0.253032386302948,
|
| 1003 |
+
"learning_rate": 1.2667150481429313e-05,
|
| 1004 |
+
"loss": 0.1937,
|
| 1005 |
+
"step": 3125
|
| 1006 |
+
},
|
| 1007 |
+
{
|
| 1008 |
+
"epoch": 2.5840853158326498,
|
| 1009 |
+
"grad_norm": 0.24757049977779388,
|
| 1010 |
+
"learning_rate": 1.1529587996809755e-05,
|
| 1011 |
+
"loss": 0.1995,
|
| 1012 |
+
"step": 3150
|
| 1013 |
+
},
|
| 1014 |
+
{
|
| 1015 |
+
"epoch": 2.604593929450369,
|
| 1016 |
+
"grad_norm": 0.21053193509578705,
|
| 1017 |
+
"learning_rate": 1.0442410532213131e-05,
|
| 1018 |
+
"loss": 0.2012,
|
| 1019 |
+
"step": 3175
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 2.6251025430680883,
|
| 1023 |
+
"grad_norm": 0.2175871729850769,
|
| 1024 |
+
"learning_rate": 9.40623724897347e-06,
|
| 1025 |
+
"loss": 0.2068,
|
| 1026 |
+
"step": 3200
|
| 1027 |
+
},
|
| 1028 |
+
{
|
| 1029 |
+
"epoch": 2.6251025430680883,
|
| 1030 |
+
"eval_loss": 0.22113485634326935,
|
| 1031 |
+
"eval_runtime": 19.2803,
|
| 1032 |
+
"eval_samples_per_second": 51.866,
|
| 1033 |
+
"eval_steps_per_second": 0.83,
|
| 1034 |
+
"step": 3200
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 2.645611156685808,
|
| 1038 |
+
"grad_norm": 0.2602500915527344,
|
| 1039 |
+
"learning_rate": 8.421658260897669e-06,
|
| 1040 |
+
"loss": 0.2148,
|
| 1041 |
+
"step": 3225
|
| 1042 |
+
},
|
| 1043 |
+
{
|
| 1044 |
+
"epoch": 2.6661197703035278,
|
| 1045 |
+
"grad_norm": 0.23596608638763428,
|
| 1046 |
+
"learning_rate": 7.489234298188141e-06,
|
| 1047 |
+
"loss": 0.1978,
|
| 1048 |
+
"step": 3250
|
| 1049 |
+
},
|
| 1050 |
+
{
|
| 1051 |
+
"epoch": 2.686628383921247,
|
| 1052 |
+
"grad_norm": 0.3108835220336914,
|
| 1053 |
+
"learning_rate": 6.609496388099812e-06,
|
| 1054 |
+
"loss": 0.2051,
|
| 1055 |
+
"step": 3275
|
| 1056 |
+
},
|
| 1057 |
+
{
|
| 1058 |
+
"epoch": 2.7071369975389663,
|
| 1059 |
+
"grad_norm": 0.22037780284881592,
|
| 1060 |
+
"learning_rate": 5.782945552513497e-06,
|
| 1061 |
+
"loss": 0.1975,
|
| 1062 |
+
"step": 3300
|
| 1063 |
+
},
|
| 1064 |
+
{
|
| 1065 |
+
"epoch": 2.727645611156686,
|
| 1066 |
+
"grad_norm": 0.23090258240699768,
|
| 1067 |
+
"learning_rate": 5.010052522597441e-06,
|
| 1068 |
+
"loss": 0.1965,
|
| 1069 |
+
"step": 3325
|
| 1070 |
+
},
|
| 1071 |
+
{
|
| 1072 |
+
"epoch": 2.7481542247744053,
|
| 1073 |
+
"grad_norm": 0.220258429646492,
|
| 1074 |
+
"learning_rate": 4.2912574707201e-06,
|
| 1075 |
+
"loss": 0.1919,
|
| 1076 |
+
"step": 3350
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 2.7686628383921246,
|
| 1080 |
+
"grad_norm": 0.25809064507484436,
|
| 1081 |
+
"learning_rate": 3.6269697597663076e-06,
|
| 1082 |
+
"loss": 0.2123,
|
| 1083 |
+
"step": 3375
|
| 1084 |
+
},
|
| 1085 |
+
{
|
| 1086 |
+
"epoch": 2.7891714520098443,
|
| 1087 |
+
"grad_norm": 0.21720901131629944,
|
| 1088 |
+
"learning_rate": 3.017567709999991e-06,
|
| 1089 |
+
"loss": 0.1988,
|
| 1090 |
+
"step": 3400
|
| 1091 |
+
},
|
| 1092 |
+
{
|
| 1093 |
+
"epoch": 2.7891714520098443,
|
| 1094 |
+
"eval_loss": 0.22063028812408447,
|
| 1095 |
+
"eval_runtime": 19.273,
|
| 1096 |
+
"eval_samples_per_second": 51.886,
|
| 1097 |
+
"eval_steps_per_second": 0.83,
|
| 1098 |
+
"step": 3400
|
| 1099 |
+
},
|
| 1100 |
+
{
|
| 1101 |
+
"epoch": 2.8096800656275636,
|
| 1102 |
+
"grad_norm": 0.2201199233531952,
|
| 1103 |
+
"learning_rate": 2.4633983836059303e-06,
|
| 1104 |
+
"loss": 0.2075,
|
| 1105 |
+
"step": 3425
|
| 1106 |
+
},
|
| 1107 |
+
{
|
| 1108 |
+
"epoch": 2.830188679245283,
|
| 1109 |
+
"grad_norm": 0.23775076866149902,
|
| 1110 |
+
"learning_rate": 1.9647773870335006e-06,
|
| 1111 |
+
"loss": 0.1961,
|
| 1112 |
+
"step": 3450
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 2.8506972928630026,
|
| 1116 |
+
"grad_norm": 0.2370598465204239,
|
| 1117 |
+
"learning_rate": 1.521988691254772e-06,
|
| 1118 |
+
"loss": 0.2058,
|
| 1119 |
+
"step": 3475
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 2.871205906480722,
|
| 1123 |
+
"grad_norm": 0.2179088294506073,
|
| 1124 |
+
"learning_rate": 1.135284470039466e-06,
|
| 1125 |
+
"loss": 0.1961,
|
| 1126 |
+
"step": 3500
|
| 1127 |
+
},
|
| 1128 |
+
{
|
| 1129 |
+
"epoch": 2.891714520098441,
|
| 1130 |
+
"grad_norm": 0.2624494135379791,
|
| 1131 |
+
"learning_rate": 8.048849563387584e-07,
|
| 1132 |
+
"loss": 0.1981,
|
| 1133 |
+
"step": 3525
|
| 1134 |
+
},
|
| 1135 |
+
{
|
| 1136 |
+
"epoch": 2.912223133716161,
|
| 1137 |
+
"grad_norm": 0.2389053851366043,
|
| 1138 |
+
"learning_rate": 5.309783168598759e-07,
|
| 1139 |
+
"loss": 0.1982,
|
| 1140 |
+
"step": 3550
|
| 1141 |
+
},
|
| 1142 |
+
{
|
| 1143 |
+
"epoch": 2.93273174733388,
|
| 1144 |
+
"grad_norm": 0.24316072463989258,
|
| 1145 |
+
"learning_rate": 3.1372054490276247e-07,
|
| 1146 |
+
"loss": 0.1965,
|
| 1147 |
+
"step": 3575
|
| 1148 |
+
},
|
| 1149 |
+
{
|
| 1150 |
+
"epoch": 2.9532403609516,
|
| 1151 |
+
"grad_norm": 0.2208586186170578,
|
| 1152 |
+
"learning_rate": 1.5323537151988866e-07,
|
| 1153 |
+
"loss": 0.2031,
|
| 1154 |
+
"step": 3600
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 2.9532403609516,
|
| 1158 |
+
"eval_loss": 0.22049948573112488,
|
| 1159 |
+
"eval_runtime": 19.2724,
|
| 1160 |
+
"eval_samples_per_second": 51.888,
|
| 1161 |
+
"eval_steps_per_second": 0.83,
|
| 1162 |
+
"step": 3600
|
| 1163 |
+
},
|
| 1164 |
+
{
|
| 1165 |
+
"epoch": 2.973748974569319,
|
| 1166 |
+
"grad_norm": 0.21298660337924957,
|
| 1167 |
+
"learning_rate": 4.961419504987408e-08,
|
| 1168 |
+
"loss": 0.1983,
|
| 1169 |
+
"step": 3625
|
| 1170 |
+
},
|
| 1171 |
+
{
|
| 1172 |
+
"epoch": 2.994257588187039,
|
| 1173 |
+
"grad_norm": 0.22230713069438934,
|
| 1174 |
+
"learning_rate": 2.9160290649699053e-09,
|
| 1175 |
+
"loss": 0.1976,
|
| 1176 |
+
"step": 3650
|
| 1177 |
+
},
|
| 1178 |
+
{
|
| 1179 |
+
"epoch": 3.0,
|
| 1180 |
+
"step": 3657,
|
| 1181 |
+
"total_flos": 2.37785441107968e+18,
|
| 1182 |
+
"train_loss": 0.2500096781513943,
|
| 1183 |
+
"train_runtime": 3681.5914,
|
| 1184 |
+
"train_samples_per_second": 31.78,
|
| 1185 |
+
"train_steps_per_second": 0.993
|
| 1186 |
+
}
|
| 1187 |
+
],
|
| 1188 |
+
"logging_steps": 25,
|
| 1189 |
+
"max_steps": 3657,
|
| 1190 |
+
"num_input_tokens_seen": 0,
|
| 1191 |
+
"num_train_epochs": 3,
|
| 1192 |
+
"save_steps": 0,
|
| 1193 |
+
"stateful_callbacks": {
|
| 1194 |
+
"TrainerControl": {
|
| 1195 |
+
"args": {
|
| 1196 |
+
"should_epoch_stop": false,
|
| 1197 |
+
"should_evaluate": false,
|
| 1198 |
+
"should_log": false,
|
| 1199 |
+
"should_save": true,
|
| 1200 |
+
"should_training_stop": true
|
| 1201 |
+
},
|
| 1202 |
+
"attributes": {}
|
| 1203 |
+
}
|
| 1204 |
+
},
|
| 1205 |
+
"total_flos": 2.37785441107968e+18,
|
| 1206 |
+
"train_batch_size": 32,
|
| 1207 |
+
"trial_name": null,
|
| 1208 |
+
"trial_params": null
|
| 1209 |
+
}
|
nl_tasks/exps/run_ex15_3ep/ft/adapter_config.json
ADDED
|
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"T": 1.0,
|
| 3 |
+
"base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
|
| 4 |
+
"bias": "none",
|
| 5 |
+
"inference_mode": false,
|
| 6 |
+
"layers_to_transform": null,
|
| 7 |
+
"modules_to_save": null,
|
| 8 |
+
"num_rotations": 1,
|
| 9 |
+
"peft_type": "ROTATION",
|
| 10 |
+
"r": 16,
|
| 11 |
+
"revision": null,
|
| 12 |
+
"target_modules": [
|
| 13 |
+
"v_proj",
|
| 14 |
+
"q_proj"
|
| 15 |
+
],
|
| 16 |
+
"target_modules_to_skip": null,
|
| 17 |
+
"task_type": "CAUSAL_LM"
|
| 18 |
+
}
|
nl_tasks/exps/run_ex15_3ep/ft/added_tokens.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"[PAD]": 32000
|
| 3 |
+
}
|
nl_tasks/exps/run_ex15_3ep/ft/special_tokens_map.json
ADDED
|
@@ -0,0 +1,30 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": {
|
| 3 |
+
"content": "</s>",
|
| 4 |
+
"lstrip": false,
|
| 5 |
+
"normalized": false,
|
| 6 |
+
"rstrip": false,
|
| 7 |
+
"single_word": false
|
| 8 |
+
},
|
| 9 |
+
"eos_token": {
|
| 10 |
+
"content": "</s>",
|
| 11 |
+
"lstrip": false,
|
| 12 |
+
"normalized": false,
|
| 13 |
+
"rstrip": false,
|
| 14 |
+
"single_word": false
|
| 15 |
+
},
|
| 16 |
+
"pad_token": {
|
| 17 |
+
"content": "[PAD]",
|
| 18 |
+
"lstrip": false,
|
| 19 |
+
"normalized": false,
|
| 20 |
+
"rstrip": false,
|
| 21 |
+
"single_word": false
|
| 22 |
+
},
|
| 23 |
+
"unk_token": {
|
| 24 |
+
"content": "</s>",
|
| 25 |
+
"lstrip": false,
|
| 26 |
+
"normalized": false,
|
| 27 |
+
"rstrip": false,
|
| 28 |
+
"single_word": false
|
| 29 |
+
}
|
| 30 |
+
}
|
nl_tasks/exps/run_ex15_3ep/ft/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
nl_tasks/exps/run_ex15_3ep/ft/tokenizer.model
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
|
| 3 |
+
size 499723
|
nl_tasks/exps/run_ex15_3ep/ft/tokenizer_config.json
ADDED
|
@@ -0,0 +1,51 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_bos_token": true,
|
| 3 |
+
"add_eos_token": false,
|
| 4 |
+
"add_prefix_space": null,
|
| 5 |
+
"added_tokens_decoder": {
|
| 6 |
+
"0": {
|
| 7 |
+
"content": "<unk>",
|
| 8 |
+
"lstrip": false,
|
| 9 |
+
"normalized": false,
|
| 10 |
+
"rstrip": false,
|
| 11 |
+
"single_word": false,
|
| 12 |
+
"special": true
|
| 13 |
+
},
|
| 14 |
+
"1": {
|
| 15 |
+
"content": "<s>",
|
| 16 |
+
"lstrip": false,
|
| 17 |
+
"normalized": false,
|
| 18 |
+
"rstrip": false,
|
| 19 |
+
"single_word": false,
|
| 20 |
+
"special": true
|
| 21 |
+
},
|
| 22 |
+
"2": {
|
| 23 |
+
"content": "</s>",
|
| 24 |
+
"lstrip": false,
|
| 25 |
+
"normalized": false,
|
| 26 |
+
"rstrip": false,
|
| 27 |
+
"single_word": false,
|
| 28 |
+
"special": true
|
| 29 |
+
},
|
| 30 |
+
"32000": {
|
| 31 |
+
"content": "[PAD]",
|
| 32 |
+
"lstrip": false,
|
| 33 |
+
"normalized": false,
|
| 34 |
+
"rstrip": false,
|
| 35 |
+
"single_word": false,
|
| 36 |
+
"special": true
|
| 37 |
+
}
|
| 38 |
+
},
|
| 39 |
+
"bos_token": "</s>",
|
| 40 |
+
"clean_up_tokenization_spaces": false,
|
| 41 |
+
"eos_token": "</s>",
|
| 42 |
+
"extra_special_tokens": {},
|
| 43 |
+
"legacy": false,
|
| 44 |
+
"model_max_length": 512,
|
| 45 |
+
"pad_token": "[PAD]",
|
| 46 |
+
"padding_side": "right",
|
| 47 |
+
"sp_model_kwargs": {},
|
| 48 |
+
"tokenizer_class": "LlamaTokenizer",
|
| 49 |
+
"unk_token": "</s>",
|
| 50 |
+
"use_default_system_prompt": false
|
| 51 |
+
}
|
nl_tasks/exps/run_ex15_3ep/ft2/adapter_config.json
ADDED
|
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"T": 1.0,
|
| 3 |
+
"base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
|
| 4 |
+
"bias": "none",
|
| 5 |
+
"inference_mode": true,
|
| 6 |
+
"layers_to_transform": null,
|
| 7 |
+
"modules_to_save": null,
|
| 8 |
+
"num_rotations": 1,
|
| 9 |
+
"peft_type": "ROTATION",
|
| 10 |
+
"r": 16,
|
| 11 |
+
"revision": null,
|
| 12 |
+
"target_modules": [
|
| 13 |
+
"v_proj",
|
| 14 |
+
"q_proj"
|
| 15 |
+
],
|
| 16 |
+
"target_modules_to_skip": null,
|
| 17 |
+
"task_type": "CAUSAL_LM"
|
| 18 |
+
}
|
nl_tasks/exps/run_ex15_3ep/ft2/adapter_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ce697834401f369bf8515155a5849658b34db55939812d8b84d1812bb1bc1d2b
|
| 3 |
+
size 33602915
|
nl_tasks/exps/run_ex15_3ep/trainer_state.json
ADDED
|
@@ -0,0 +1,1209 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 3.0,
|
| 6 |
+
"eval_steps": 200,
|
| 7 |
+
"global_step": 3657,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.020508613617719443,
|
| 14 |
+
"grad_norm": 0.2236936092376709,
|
| 15 |
+
"learning_rate": 3.278688524590164e-05,
|
| 16 |
+
"loss": 0.7378,
|
| 17 |
+
"step": 25
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.04101722723543889,
|
| 21 |
+
"grad_norm": 0.15586046874523163,
|
| 22 |
+
"learning_rate": 6.693989071038252e-05,
|
| 23 |
+
"loss": 0.5263,
|
| 24 |
+
"step": 50
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.06152584085315833,
|
| 28 |
+
"grad_norm": 0.1585606187582016,
|
| 29 |
+
"learning_rate": 0.00010109289617486339,
|
| 30 |
+
"loss": 0.4153,
|
| 31 |
+
"step": 75
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.08203445447087777,
|
| 35 |
+
"grad_norm": 0.21381111443042755,
|
| 36 |
+
"learning_rate": 0.00013524590163934428,
|
| 37 |
+
"loss": 0.3615,
|
| 38 |
+
"step": 100
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.10254306808859721,
|
| 42 |
+
"grad_norm": 0.18566539883613586,
|
| 43 |
+
"learning_rate": 0.00016939890710382513,
|
| 44 |
+
"loss": 0.3582,
|
| 45 |
+
"step": 125
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.12305168170631665,
|
| 49 |
+
"grad_norm": 0.18897324800491333,
|
| 50 |
+
"learning_rate": 0.00020355191256830603,
|
| 51 |
+
"loss": 0.338,
|
| 52 |
+
"step": 150
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.1435602953240361,
|
| 56 |
+
"grad_norm": 0.19681961834430695,
|
| 57 |
+
"learning_rate": 0.00023770491803278687,
|
| 58 |
+
"loss": 0.3266,
|
| 59 |
+
"step": 175
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.16406890894175555,
|
| 63 |
+
"grad_norm": 0.20275168120861053,
|
| 64 |
+
"learning_rate": 0.0002718579234972678,
|
| 65 |
+
"loss": 0.3333,
|
| 66 |
+
"step": 200
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.16406890894175555,
|
| 70 |
+
"eval_loss": 0.3140156865119934,
|
| 71 |
+
"eval_runtime": 19.5725,
|
| 72 |
+
"eval_samples_per_second": 51.092,
|
| 73 |
+
"eval_steps_per_second": 0.817,
|
| 74 |
+
"step": 200
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.184577522559475,
|
| 78 |
+
"grad_norm": 0.18092350661754608,
|
| 79 |
+
"learning_rate": 0.0003060109289617486,
|
| 80 |
+
"loss": 0.3277,
|
| 81 |
+
"step": 225
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.20508613617719443,
|
| 85 |
+
"grad_norm": 0.16045643389225006,
|
| 86 |
+
"learning_rate": 0.0003401639344262295,
|
| 87 |
+
"loss": 0.3029,
|
| 88 |
+
"step": 250
|
| 89 |
+
},
|
| 90 |
+
{
|
| 91 |
+
"epoch": 0.22559474979491387,
|
| 92 |
+
"grad_norm": 0.16111819446086884,
|
| 93 |
+
"learning_rate": 0.0003743169398907104,
|
| 94 |
+
"loss": 0.3003,
|
| 95 |
+
"step": 275
|
| 96 |
+
},
|
| 97 |
+
{
|
| 98 |
+
"epoch": 0.2461033634126333,
|
| 99 |
+
"grad_norm": 0.17536306381225586,
|
| 100 |
+
"learning_rate": 0.00040846994535519127,
|
| 101 |
+
"loss": 0.3049,
|
| 102 |
+
"step": 300
|
| 103 |
+
},
|
| 104 |
+
{
|
| 105 |
+
"epoch": 0.2666119770303528,
|
| 106 |
+
"grad_norm": 0.18236631155014038,
|
| 107 |
+
"learning_rate": 0.0004426229508196721,
|
| 108 |
+
"loss": 0.308,
|
| 109 |
+
"step": 325
|
| 110 |
+
},
|
| 111 |
+
{
|
| 112 |
+
"epoch": 0.2871205906480722,
|
| 113 |
+
"grad_norm": 0.17001983523368835,
|
| 114 |
+
"learning_rate": 0.00047677595628415297,
|
| 115 |
+
"loss": 0.2999,
|
| 116 |
+
"step": 350
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.30762920426579166,
|
| 120 |
+
"grad_norm": 0.18576228618621826,
|
| 121 |
+
"learning_rate": 0.0004999927099273376,
|
| 122 |
+
"loss": 0.3109,
|
| 123 |
+
"step": 375
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 0.3281378178835111,
|
| 127 |
+
"grad_norm": 0.20945490896701813,
|
| 128 |
+
"learning_rate": 0.0004998759645123753,
|
| 129 |
+
"loss": 0.2948,
|
| 130 |
+
"step": 400
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 0.3281378178835111,
|
| 134 |
+
"eval_loss": 0.2856634557247162,
|
| 135 |
+
"eval_runtime": 19.7912,
|
| 136 |
+
"eval_samples_per_second": 50.527,
|
| 137 |
+
"eval_steps_per_second": 0.808,
|
| 138 |
+
"step": 400
|
| 139 |
+
},
|
| 140 |
+
{
|
| 141 |
+
"epoch": 0.34864643150123054,
|
| 142 |
+
"grad_norm": 0.1730087697505951,
|
| 143 |
+
"learning_rate": 0.0004996169115712003,
|
| 144 |
+
"loss": 0.3036,
|
| 145 |
+
"step": 425
|
| 146 |
+
},
|
| 147 |
+
{
|
| 148 |
+
"epoch": 0.36915504511895,
|
| 149 |
+
"grad_norm": 0.17050108313560486,
|
| 150 |
+
"learning_rate": 0.000499215698637743,
|
| 151 |
+
"loss": 0.2994,
|
| 152 |
+
"step": 450
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.3896636587366694,
|
| 156 |
+
"grad_norm": 0.19767747819423676,
|
| 157 |
+
"learning_rate": 0.0004986725542078503,
|
| 158 |
+
"loss": 0.2872,
|
| 159 |
+
"step": 475
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.41017227235438886,
|
| 163 |
+
"grad_norm": 0.17706479132175446,
|
| 164 |
+
"learning_rate": 0.0004979877876091531,
|
| 165 |
+
"loss": 0.2705,
|
| 166 |
+
"step": 500
|
| 167 |
+
},
|
| 168 |
+
{
|
| 169 |
+
"epoch": 0.4306808859721083,
|
| 170 |
+
"grad_norm": 0.18240663409233093,
|
| 171 |
+
"learning_rate": 0.0004971617888249013,
|
| 172 |
+
"loss": 0.2711,
|
| 173 |
+
"step": 525
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 0.45118949958982774,
|
| 177 |
+
"grad_norm": 0.18252277374267578,
|
| 178 |
+
"learning_rate": 0.0004961950282718631,
|
| 179 |
+
"loss": 0.2811,
|
| 180 |
+
"step": 550
|
| 181 |
+
},
|
| 182 |
+
{
|
| 183 |
+
"epoch": 0.4716981132075472,
|
| 184 |
+
"grad_norm": 0.20153027772903442,
|
| 185 |
+
"learning_rate": 0.0004950880565324163,
|
| 186 |
+
"loss": 0.2733,
|
| 187 |
+
"step": 575
|
| 188 |
+
},
|
| 189 |
+
{
|
| 190 |
+
"epoch": 0.4922067268252666,
|
| 191 |
+
"grad_norm": 0.19355134665966034,
|
| 192 |
+
"learning_rate": 0.0004938415040409851,
|
| 193 |
+
"loss": 0.2901,
|
| 194 |
+
"step": 600
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 0.4922067268252666,
|
| 198 |
+
"eval_loss": 0.2669033110141754,
|
| 199 |
+
"eval_runtime": 19.2998,
|
| 200 |
+
"eval_samples_per_second": 51.814,
|
| 201 |
+
"eval_steps_per_second": 0.829,
|
| 202 |
+
"step": 600
|
| 203 |
+
},
|
| 204 |
+
{
|
| 205 |
+
"epoch": 0.5127153404429861,
|
| 206 |
+
"grad_norm": 0.18456509709358215,
|
| 207 |
+
"learning_rate": 0.000492456080725,
|
| 208 |
+
"loss": 0.2703,
|
| 209 |
+
"step": 625
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 0.5332239540607056,
|
| 213 |
+
"grad_norm": 0.2080974578857422,
|
| 214 |
+
"learning_rate": 0.0004909325756005843,
|
| 215 |
+
"loss": 0.2619,
|
| 216 |
+
"step": 650
|
| 217 |
+
},
|
| 218 |
+
{
|
| 219 |
+
"epoch": 0.5537325676784249,
|
| 220 |
+
"grad_norm": 0.19045989215373993,
|
| 221 |
+
"learning_rate": 0.0004892718563231997,
|
| 222 |
+
"loss": 0.275,
|
| 223 |
+
"step": 675
|
| 224 |
+
},
|
| 225 |
+
{
|
| 226 |
+
"epoch": 0.5742411812961444,
|
| 227 |
+
"grad_norm": 0.2038784921169281,
|
| 228 |
+
"learning_rate": 0.00048747486869350645,
|
| 229 |
+
"loss": 0.2712,
|
| 230 |
+
"step": 700
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 0.5947497949138638,
|
| 234 |
+
"grad_norm": 0.19773368537425995,
|
| 235 |
+
"learning_rate": 0.0004855426361187163,
|
| 236 |
+
"loss": 0.2705,
|
| 237 |
+
"step": 725
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 0.6152584085315833,
|
| 241 |
+
"grad_norm": 0.20288875699043274,
|
| 242 |
+
"learning_rate": 0.00048347625902975044,
|
| 243 |
+
"loss": 0.2676,
|
| 244 |
+
"step": 750
|
| 245 |
+
},
|
| 246 |
+
{
|
| 247 |
+
"epoch": 0.6357670221493027,
|
| 248 |
+
"grad_norm": 0.20831525325775146,
|
| 249 |
+
"learning_rate": 0.00048127691425452966,
|
| 250 |
+
"loss": 0.2634,
|
| 251 |
+
"step": 775
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 0.6562756357670222,
|
| 255 |
+
"grad_norm": 0.21394769847393036,
|
| 256 |
+
"learning_rate": 0.00047894585434775583,
|
| 257 |
+
"loss": 0.2683,
|
| 258 |
+
"step": 800
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 0.6562756357670222,
|
| 262 |
+
"eval_loss": 0.2570738196372986,
|
| 263 |
+
"eval_runtime": 19.2751,
|
| 264 |
+
"eval_samples_per_second": 51.88,
|
| 265 |
+
"eval_steps_per_second": 0.83,
|
| 266 |
+
"step": 800
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 0.6767842493847416,
|
| 270 |
+
"grad_norm": 0.18774528801441193,
|
| 271 |
+
"learning_rate": 0.0004764844068775664,
|
| 272 |
+
"loss": 0.2616,
|
| 273 |
+
"step": 825
|
| 274 |
+
},
|
| 275 |
+
{
|
| 276 |
+
"epoch": 0.6972928630024611,
|
| 277 |
+
"grad_norm": 0.20157219469547272,
|
| 278 |
+
"learning_rate": 0.00047389397366946723,
|
| 279 |
+
"loss": 0.2645,
|
| 280 |
+
"step": 850
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 0.7178014766201805,
|
| 284 |
+
"grad_norm": 0.16799937188625336,
|
| 285 |
+
"learning_rate": 0.00047117603000797567,
|
| 286 |
+
"loss": 0.2531,
|
| 287 |
+
"step": 875
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 0.7383100902379,
|
| 291 |
+
"grad_norm": 0.21612580120563507,
|
| 292 |
+
"learning_rate": 0.00046833212379642674,
|
| 293 |
+
"loss": 0.2596,
|
| 294 |
+
"step": 900
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 0.7588187038556193,
|
| 298 |
+
"grad_norm": 0.19849224388599396,
|
| 299 |
+
"learning_rate": 0.0004653638746754233,
|
| 300 |
+
"loss": 0.2601,
|
| 301 |
+
"step": 925
|
| 302 |
+
},
|
| 303 |
+
{
|
| 304 |
+
"epoch": 0.7793273174733388,
|
| 305 |
+
"grad_norm": 0.20190981030464172,
|
| 306 |
+
"learning_rate": 0.00046227297310042935,
|
| 307 |
+
"loss": 0.2484,
|
| 308 |
+
"step": 950
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 0.7998359310910582,
|
| 312 |
+
"grad_norm": 0.19541241228580475,
|
| 313 |
+
"learning_rate": 0.0004590611793790351,
|
| 314 |
+
"loss": 0.2545,
|
| 315 |
+
"step": 975
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 0.8203445447087777,
|
| 319 |
+
"grad_norm": 0.21503788232803345,
|
| 320 |
+
"learning_rate": 0.0004557303226684389,
|
| 321 |
+
"loss": 0.2483,
|
| 322 |
+
"step": 1000
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 0.8203445447087777,
|
| 326 |
+
"eval_loss": 0.24449799954891205,
|
| 327 |
+
"eval_runtime": 19.3269,
|
| 328 |
+
"eval_samples_per_second": 51.741,
|
| 329 |
+
"eval_steps_per_second": 0.828,
|
| 330 |
+
"step": 1000
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 0.8408531583264971,
|
| 334 |
+
"grad_norm": 0.21388469636440277,
|
| 335 |
+
"learning_rate": 0.0004522822999337194,
|
| 336 |
+
"loss": 0.2485,
|
| 337 |
+
"step": 1025
|
| 338 |
+
},
|
| 339 |
+
{
|
| 340 |
+
"epoch": 0.8613617719442166,
|
| 341 |
+
"grad_norm": 0.20644618570804596,
|
| 342 |
+
"learning_rate": 0.0004487190748674895,
|
| 343 |
+
"loss": 0.2458,
|
| 344 |
+
"step": 1050
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 0.881870385561936,
|
| 348 |
+
"grad_norm": 0.20617376267910004,
|
| 349 |
+
"learning_rate": 0.0004450426767715493,
|
| 350 |
+
"loss": 0.2515,
|
| 351 |
+
"step": 1075
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 0.9023789991796555,
|
| 355 |
+
"grad_norm": 0.20171715319156647,
|
| 356 |
+
"learning_rate": 0.0004412551994011721,
|
| 357 |
+
"loss": 0.2521,
|
| 358 |
+
"step": 1100
|
| 359 |
+
},
|
| 360 |
+
{
|
| 361 |
+
"epoch": 0.9228876127973749,
|
| 362 |
+
"grad_norm": 0.20110057294368744,
|
| 363 |
+
"learning_rate": 0.00043735879977268476,
|
| 364 |
+
"loss": 0.2465,
|
| 365 |
+
"step": 1125
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 0.9433962264150944,
|
| 369 |
+
"grad_norm": 0.1831655353307724,
|
| 370 |
+
"learning_rate": 0.00043335569693501975,
|
| 371 |
+
"loss": 0.2461,
|
| 372 |
+
"step": 1150
|
| 373 |
+
},
|
| 374 |
+
{
|
| 375 |
+
"epoch": 0.9639048400328137,
|
| 376 |
+
"grad_norm": 0.1896771490573883,
|
| 377 |
+
"learning_rate": 0.00042924817070593823,
|
| 378 |
+
"loss": 0.2526,
|
| 379 |
+
"step": 1175
|
| 380 |
+
},
|
| 381 |
+
{
|
| 382 |
+
"epoch": 0.9844134536505332,
|
| 383 |
+
"grad_norm": 0.20726299285888672,
|
| 384 |
+
"learning_rate": 0.00042503856037364536,
|
| 385 |
+
"loss": 0.2582,
|
| 386 |
+
"step": 1200
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 0.9844134536505332,
|
| 390 |
+
"eval_loss": 0.2386789172887802,
|
| 391 |
+
"eval_runtime": 19.2877,
|
| 392 |
+
"eval_samples_per_second": 51.846,
|
| 393 |
+
"eval_steps_per_second": 0.83,
|
| 394 |
+
"step": 1200
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 1.0049220672682526,
|
| 398 |
+
"grad_norm": 0.21773400902748108,
|
| 399 |
+
"learning_rate": 0.00042072926336453575,
|
| 400 |
+
"loss": 0.2396,
|
| 401 |
+
"step": 1225
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 1.0254306808859721,
|
| 405 |
+
"grad_norm": 0.19781112670898438,
|
| 406 |
+
"learning_rate": 0.00041632273387782884,
|
| 407 |
+
"loss": 0.2242,
|
| 408 |
+
"step": 1250
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 1.0459392945036916,
|
| 412 |
+
"grad_norm": 0.21470637619495392,
|
| 413 |
+
"learning_rate": 0.0004118214814878716,
|
| 414 |
+
"loss": 0.2271,
|
| 415 |
+
"step": 1275
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 1.066447908121411,
|
| 419 |
+
"grad_norm": 0.1851588487625122,
|
| 420 |
+
"learning_rate": 0.00040722806971490376,
|
| 421 |
+
"loss": 0.2296,
|
| 422 |
+
"step": 1300
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 1.0869565217391304,
|
| 426 |
+
"grad_norm": 0.20691987872123718,
|
| 427 |
+
"learning_rate": 0.0004025451145651009,
|
| 428 |
+
"loss": 0.2262,
|
| 429 |
+
"step": 1325
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 1.1074651353568499,
|
| 433 |
+
"grad_norm": 0.178948312997818,
|
| 434 |
+
"learning_rate": 0.00039777528304072574,
|
| 435 |
+
"loss": 0.2178,
|
| 436 |
+
"step": 1350
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 1.1279737489745694,
|
| 440 |
+
"grad_norm": 0.22265319526195526,
|
| 441 |
+
"learning_rate": 0.00039292129162123634,
|
| 442 |
+
"loss": 0.2216,
|
| 443 |
+
"step": 1375
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 1.1484823625922886,
|
| 447 |
+
"grad_norm": 0.26052021980285645,
|
| 448 |
+
"learning_rate": 0.000387985904716217,
|
| 449 |
+
"loss": 0.2145,
|
| 450 |
+
"step": 1400
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 1.1484823625922886,
|
| 454 |
+
"eval_loss": 0.23337993025779724,
|
| 455 |
+
"eval_runtime": 19.2972,
|
| 456 |
+
"eval_samples_per_second": 51.821,
|
| 457 |
+
"eval_steps_per_second": 0.829,
|
| 458 |
+
"step": 1400
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 1.1689909762100081,
|
| 462 |
+
"grad_norm": 0.2162986695766449,
|
| 463 |
+
"learning_rate": 0.00038297193309101175,
|
| 464 |
+
"loss": 0.2241,
|
| 465 |
+
"step": 1425
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 1.1894995898277276,
|
| 469 |
+
"grad_norm": 0.23128162324428558,
|
| 470 |
+
"learning_rate": 0.00037788223226595783,
|
| 471 |
+
"loss": 0.2264,
|
| 472 |
+
"step": 1450
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 1.2100082034454471,
|
| 476 |
+
"grad_norm": 0.22870606184005737,
|
| 477 |
+
"learning_rate": 0.0003727197008901311,
|
| 478 |
+
"loss": 0.2225,
|
| 479 |
+
"step": 1475
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 1.2305168170631666,
|
| 483 |
+
"grad_norm": 0.21988478302955627,
|
| 484 |
+
"learning_rate": 0.00036748727909052825,
|
| 485 |
+
"loss": 0.2319,
|
| 486 |
+
"step": 1500
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 1.251025430680886,
|
| 490 |
+
"grad_norm": 0.23757073283195496,
|
| 491 |
+
"learning_rate": 0.00036218794679762737,
|
| 492 |
+
"loss": 0.2268,
|
| 493 |
+
"step": 1525
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 1.2715340442986054,
|
| 497 |
+
"grad_norm": 0.19217915832996368,
|
| 498 |
+
"learning_rate": 0.0003568247220482798,
|
| 499 |
+
"loss": 0.2197,
|
| 500 |
+
"step": 1550
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 1.2920426579163249,
|
| 504 |
+
"grad_norm": 0.18676155805587769,
|
| 505 |
+
"learning_rate": 0.00035140065926690005,
|
| 506 |
+
"loss": 0.2219,
|
| 507 |
+
"step": 1575
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 1.3125512715340442,
|
| 511 |
+
"grad_norm": 0.19034576416015625,
|
| 512 |
+
"learning_rate": 0.0003459188475259325,
|
| 513 |
+
"loss": 0.2145,
|
| 514 |
+
"step": 1600
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 1.3125512715340442,
|
| 518 |
+
"eval_loss": 0.22706548869609833,
|
| 519 |
+
"eval_runtime": 19.3066,
|
| 520 |
+
"eval_samples_per_second": 51.796,
|
| 521 |
+
"eval_steps_per_second": 0.829,
|
| 522 |
+
"step": 1600
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 1.3330598851517639,
|
| 526 |
+
"grad_norm": 0.24615806341171265,
|
| 527 |
+
"learning_rate": 0.000340382408786585,
|
| 528 |
+
"loss": 0.2178,
|
| 529 |
+
"step": 1625
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 1.3535684987694832,
|
| 533 |
+
"grad_norm": 0.21526384353637695,
|
| 534 |
+
"learning_rate": 0.00033479449612083297,
|
| 535 |
+
"loss": 0.2185,
|
| 536 |
+
"step": 1650
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 1.3740771123872026,
|
| 540 |
+
"grad_norm": 0.2186814397573471,
|
| 541 |
+
"learning_rate": 0.00032915829191570474,
|
| 542 |
+
"loss": 0.2156,
|
| 543 |
+
"step": 1675
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 1.3945857260049221,
|
| 547 |
+
"grad_norm": 0.20043250918388367,
|
| 548 |
+
"learning_rate": 0.0003234770060608719,
|
| 549 |
+
"loss": 0.2166,
|
| 550 |
+
"step": 1700
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 1.4150943396226414,
|
| 554 |
+
"grad_norm": 0.2249319851398468,
|
| 555 |
+
"learning_rate": 0.0003177538741205763,
|
| 556 |
+
"loss": 0.2129,
|
| 557 |
+
"step": 1725
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 1.435602953240361,
|
| 561 |
+
"grad_norm": 0.20142365992069244,
|
| 562 |
+
"learning_rate": 0.0003119921554909358,
|
| 563 |
+
"loss": 0.2121,
|
| 564 |
+
"step": 1750
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 1.4561115668580804,
|
| 568 |
+
"grad_norm": 0.21530795097351074,
|
| 569 |
+
"learning_rate": 0.00030619513154367606,
|
| 570 |
+
"loss": 0.2198,
|
| 571 |
+
"step": 1775
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 1.4766201804758,
|
| 575 |
+
"grad_norm": 0.1881314367055893,
|
| 576 |
+
"learning_rate": 0.0003003661037573484,
|
| 577 |
+
"loss": 0.2091,
|
| 578 |
+
"step": 1800
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 1.4766201804758,
|
| 582 |
+
"eval_loss": 0.2217085212469101,
|
| 583 |
+
"eval_runtime": 19.27,
|
| 584 |
+
"eval_samples_per_second": 51.894,
|
| 585 |
+
"eval_steps_per_second": 0.83,
|
| 586 |
+
"step": 1800
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 1.4971287940935194,
|
| 590 |
+
"grad_norm": 0.20787756145000458,
|
| 591 |
+
"learning_rate": 0.00029450839183709594,
|
| 592 |
+
"loss": 0.2101,
|
| 593 |
+
"step": 1825
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 1.5176374077112387,
|
| 597 |
+
"grad_norm": 0.1853906214237213,
|
| 598 |
+
"learning_rate": 0.00028862533182403904,
|
| 599 |
+
"loss": 0.2118,
|
| 600 |
+
"step": 1850
|
| 601 |
+
},
|
| 602 |
+
{
|
| 603 |
+
"epoch": 1.5381460213289582,
|
| 604 |
+
"grad_norm": 0.18963763117790222,
|
| 605 |
+
"learning_rate": 0.0002827202741953581,
|
| 606 |
+
"loss": 0.2105,
|
| 607 |
+
"step": 1875
|
| 608 |
+
},
|
| 609 |
+
{
|
| 610 |
+
"epoch": 1.5586546349466777,
|
| 611 |
+
"grad_norm": 0.21780110895633698,
|
| 612 |
+
"learning_rate": 0.00027679658195615416,
|
| 613 |
+
"loss": 0.2145,
|
| 614 |
+
"step": 1900
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 1.579163248564397,
|
| 618 |
+
"grad_norm": 0.2096894383430481,
|
| 619 |
+
"learning_rate": 0.000270857628724175,
|
| 620 |
+
"loss": 0.2156,
|
| 621 |
+
"step": 1925
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 1.5996718621821167,
|
| 625 |
+
"grad_norm": 0.23090897500514984,
|
| 626 |
+
"learning_rate": 0.00026490679680849687,
|
| 627 |
+
"loss": 0.2137,
|
| 628 |
+
"step": 1950
|
| 629 |
+
},
|
| 630 |
+
{
|
| 631 |
+
"epoch": 1.620180475799836,
|
| 632 |
+
"grad_norm": 0.19595655798912048,
|
| 633 |
+
"learning_rate": 0.000258947475283257,
|
| 634 |
+
"loss": 0.2079,
|
| 635 |
+
"step": 1975
|
| 636 |
+
},
|
| 637 |
+
{
|
| 638 |
+
"epoch": 1.6406890894175554,
|
| 639 |
+
"grad_norm": 0.21419216692447662,
|
| 640 |
+
"learning_rate": 0.00025298305805753277,
|
| 641 |
+
"loss": 0.2156,
|
| 642 |
+
"step": 2000
|
| 643 |
+
},
|
| 644 |
+
{
|
| 645 |
+
"epoch": 1.6406890894175554,
|
| 646 |
+
"eval_loss": 0.21743401885032654,
|
| 647 |
+
"eval_runtime": 19.3038,
|
| 648 |
+
"eval_samples_per_second": 51.803,
|
| 649 |
+
"eval_steps_per_second": 0.829,
|
| 650 |
+
"step": 2000
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 1.661197703035275,
|
| 654 |
+
"grad_norm": 0.1843814104795456,
|
| 655 |
+
"learning_rate": 0.00024701694194246735,
|
| 656 |
+
"loss": 0.2058,
|
| 657 |
+
"step": 2025
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 1.6817063166529942,
|
| 661 |
+
"grad_norm": 0.22441938519477844,
|
| 662 |
+
"learning_rate": 0.00024105252471674305,
|
| 663 |
+
"loss": 0.2063,
|
| 664 |
+
"step": 2050
|
| 665 |
+
},
|
| 666 |
+
{
|
| 667 |
+
"epoch": 1.7022149302707137,
|
| 668 |
+
"grad_norm": 0.22590817511081696,
|
| 669 |
+
"learning_rate": 0.0002350932031915032,
|
| 670 |
+
"loss": 0.2131,
|
| 671 |
+
"step": 2075
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 1.7227235438884332,
|
| 675 |
+
"grad_norm": 0.19390460848808289,
|
| 676 |
+
"learning_rate": 0.00022914237127582505,
|
| 677 |
+
"loss": 0.2015,
|
| 678 |
+
"step": 2100
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 1.7432321575061525,
|
| 682 |
+
"grad_norm": 0.18362314999103546,
|
| 683 |
+
"learning_rate": 0.00022320341804384593,
|
| 684 |
+
"loss": 0.2066,
|
| 685 |
+
"step": 2125
|
| 686 |
+
},
|
| 687 |
+
{
|
| 688 |
+
"epoch": 1.7637407711238722,
|
| 689 |
+
"grad_norm": 0.1673150658607483,
|
| 690 |
+
"learning_rate": 0.0002172797258046419,
|
| 691 |
+
"loss": 0.2033,
|
| 692 |
+
"step": 2150
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 1.7842493847415914,
|
| 696 |
+
"grad_norm": 0.18781742453575134,
|
| 697 |
+
"learning_rate": 0.00021137466817596105,
|
| 698 |
+
"loss": 0.2124,
|
| 699 |
+
"step": 2175
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 1.804757998359311,
|
| 703 |
+
"grad_norm": 0.1934407502412796,
|
| 704 |
+
"learning_rate": 0.0002054916081629041,
|
| 705 |
+
"loss": 0.2051,
|
| 706 |
+
"step": 2200
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 1.804757998359311,
|
| 710 |
+
"eval_loss": 0.21281394362449646,
|
| 711 |
+
"eval_runtime": 19.3022,
|
| 712 |
+
"eval_samples_per_second": 51.808,
|
| 713 |
+
"eval_steps_per_second": 0.829,
|
| 714 |
+
"step": 2200
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 1.8252666119770304,
|
| 718 |
+
"grad_norm": 0.22336915135383606,
|
| 719 |
+
"learning_rate": 0.0001996338962426516,
|
| 720 |
+
"loss": 0.2052,
|
| 721 |
+
"step": 2225
|
| 722 |
+
},
|
| 723 |
+
{
|
| 724 |
+
"epoch": 1.8457752255947497,
|
| 725 |
+
"grad_norm": 0.1918330043554306,
|
| 726 |
+
"learning_rate": 0.00019380486845632395,
|
| 727 |
+
"loss": 0.2071,
|
| 728 |
+
"step": 2250
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 1.8662838392124692,
|
| 732 |
+
"grad_norm": 0.18349102139472961,
|
| 733 |
+
"learning_rate": 0.00018800784450906415,
|
| 734 |
+
"loss": 0.2117,
|
| 735 |
+
"step": 2275
|
| 736 |
+
},
|
| 737 |
+
{
|
| 738 |
+
"epoch": 1.8867924528301887,
|
| 739 |
+
"grad_norm": 0.18384303152561188,
|
| 740 |
+
"learning_rate": 0.0001822461258794237,
|
| 741 |
+
"loss": 0.2033,
|
| 742 |
+
"step": 2300
|
| 743 |
+
},
|
| 744 |
+
{
|
| 745 |
+
"epoch": 1.907301066447908,
|
| 746 |
+
"grad_norm": 0.21255937218666077,
|
| 747 |
+
"learning_rate": 0.00017652299393912816,
|
| 748 |
+
"loss": 0.212,
|
| 749 |
+
"step": 2325
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 1.9278096800656277,
|
| 753 |
+
"grad_norm": 0.18994638323783875,
|
| 754 |
+
"learning_rate": 0.00017084170808429533,
|
| 755 |
+
"loss": 0.1952,
|
| 756 |
+
"step": 2350
|
| 757 |
+
},
|
| 758 |
+
{
|
| 759 |
+
"epoch": 1.948318293683347,
|
| 760 |
+
"grad_norm": 0.19273756444454193,
|
| 761 |
+
"learning_rate": 0.00016520550387916704,
|
| 762 |
+
"loss": 0.2011,
|
| 763 |
+
"step": 2375
|
| 764 |
+
},
|
| 765 |
+
{
|
| 766 |
+
"epoch": 1.9688269073010665,
|
| 767 |
+
"grad_norm": 0.18741604685783386,
|
| 768 |
+
"learning_rate": 0.000159617591213415,
|
| 769 |
+
"loss": 0.2021,
|
| 770 |
+
"step": 2400
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 1.9688269073010665,
|
| 774 |
+
"eval_loss": 0.20898540318012238,
|
| 775 |
+
"eval_runtime": 19.2799,
|
| 776 |
+
"eval_samples_per_second": 51.867,
|
| 777 |
+
"eval_steps_per_second": 0.83,
|
| 778 |
+
"step": 2400
|
| 779 |
+
},
|
| 780 |
+
{
|
| 781 |
+
"epoch": 1.989335520918786,
|
| 782 |
+
"grad_norm": 0.18499332666397095,
|
| 783 |
+
"learning_rate": 0.0001540811524740675,
|
| 784 |
+
"loss": 0.1975,
|
| 785 |
+
"step": 2425
|
| 786 |
+
},
|
| 787 |
+
{
|
| 788 |
+
"epoch": 2.0098441345365052,
|
| 789 |
+
"grad_norm": 0.19958826899528503,
|
| 790 |
+
"learning_rate": 0.00014859934073309998,
|
| 791 |
+
"loss": 0.1877,
|
| 792 |
+
"step": 2450
|
| 793 |
+
},
|
| 794 |
+
{
|
| 795 |
+
"epoch": 2.030352748154225,
|
| 796 |
+
"grad_norm": 0.20758692920207977,
|
| 797 |
+
"learning_rate": 0.0001431752779517203,
|
| 798 |
+
"loss": 0.1735,
|
| 799 |
+
"step": 2475
|
| 800 |
+
},
|
| 801 |
+
{
|
| 802 |
+
"epoch": 2.0508613617719442,
|
| 803 |
+
"grad_norm": 0.21208564937114716,
|
| 804 |
+
"learning_rate": 0.0001378120532023728,
|
| 805 |
+
"loss": 0.1739,
|
| 806 |
+
"step": 2500
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 2.0713699753896635,
|
| 810 |
+
"grad_norm": 0.19463765621185303,
|
| 811 |
+
"learning_rate": 0.00013251272090947187,
|
| 812 |
+
"loss": 0.1695,
|
| 813 |
+
"step": 2525
|
| 814 |
+
},
|
| 815 |
+
{
|
| 816 |
+
"epoch": 2.091878589007383,
|
| 817 |
+
"grad_norm": 0.20465432107448578,
|
| 818 |
+
"learning_rate": 0.00012728029910986891,
|
| 819 |
+
"loss": 0.1777,
|
| 820 |
+
"step": 2550
|
| 821 |
+
},
|
| 822 |
+
{
|
| 823 |
+
"epoch": 2.1123872026251025,
|
| 824 |
+
"grad_norm": 0.20672444999217987,
|
| 825 |
+
"learning_rate": 0.00012211776773404212,
|
| 826 |
+
"loss": 0.1711,
|
| 827 |
+
"step": 2575
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"epoch": 2.132895816242822,
|
| 831 |
+
"grad_norm": 0.20150959491729736,
|
| 832 |
+
"learning_rate": 0.00011702806690898832,
|
| 833 |
+
"loss": 0.1634,
|
| 834 |
+
"step": 2600
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 2.132895816242822,
|
| 838 |
+
"eval_loss": 0.21075494587421417,
|
| 839 |
+
"eval_runtime": 19.2618,
|
| 840 |
+
"eval_samples_per_second": 51.916,
|
| 841 |
+
"eval_steps_per_second": 0.831,
|
| 842 |
+
"step": 2600
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 2.1534044298605415,
|
| 846 |
+
"grad_norm": 0.20416037738323212,
|
| 847 |
+
"learning_rate": 0.00011201409528378301,
|
| 848 |
+
"loss": 0.1672,
|
| 849 |
+
"step": 2625
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 2.1739130434782608,
|
| 853 |
+
"grad_norm": 0.2009623795747757,
|
| 854 |
+
"learning_rate": 0.00010707870837876371,
|
| 855 |
+
"loss": 0.1668,
|
| 856 |
+
"step": 2650
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 2.1944216570959805,
|
| 860 |
+
"grad_norm": 0.2076873630285263,
|
| 861 |
+
"learning_rate": 0.00010222471695927433,
|
| 862 |
+
"loss": 0.1774,
|
| 863 |
+
"step": 2675
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 2.2149302707136997,
|
| 867 |
+
"grad_norm": 0.18090513348579407,
|
| 868 |
+
"learning_rate": 9.745488543489911e-05,
|
| 869 |
+
"loss": 0.1729,
|
| 870 |
+
"step": 2700
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 2.235438884331419,
|
| 874 |
+
"grad_norm": 0.2118251621723175,
|
| 875 |
+
"learning_rate": 9.277193028509625e-05,
|
| 876 |
+
"loss": 0.1615,
|
| 877 |
+
"step": 2725
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 2.2559474979491387,
|
| 881 |
+
"grad_norm": 0.22627845406532288,
|
| 882 |
+
"learning_rate": 8.817851851212841e-05,
|
| 883 |
+
"loss": 0.1747,
|
| 884 |
+
"step": 2750
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 2.276456111566858,
|
| 888 |
+
"grad_norm": 0.2138642519712448,
|
| 889 |
+
"learning_rate": 8.36772661221712e-05,
|
| 890 |
+
"loss": 0.1723,
|
| 891 |
+
"step": 2775
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 2.2969647251845773,
|
| 895 |
+
"grad_norm": 0.20521090924739838,
|
| 896 |
+
"learning_rate": 7.927073663546427e-05,
|
| 897 |
+
"loss": 0.168,
|
| 898 |
+
"step": 2800
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 2.2969647251845773,
|
| 902 |
+
"eval_loss": 0.20893201231956482,
|
| 903 |
+
"eval_runtime": 19.33,
|
| 904 |
+
"eval_samples_per_second": 51.733,
|
| 905 |
+
"eval_steps_per_second": 0.828,
|
| 906 |
+
"step": 2800
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 2.317473338802297,
|
| 910 |
+
"grad_norm": 0.2066967785358429,
|
| 911 |
+
"learning_rate": 7.496143962635468e-05,
|
| 912 |
+
"loss": 0.1641,
|
| 913 |
+
"step": 2825
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 2.3379819524200163,
|
| 917 |
+
"grad_norm": 0.22058984637260437,
|
| 918 |
+
"learning_rate": 7.075182929406177e-05,
|
| 919 |
+
"loss": 0.1669,
|
| 920 |
+
"step": 2850
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 2.358490566037736,
|
| 924 |
+
"grad_norm": 0.1837540566921234,
|
| 925 |
+
"learning_rate": 6.664430306498026e-05,
|
| 926 |
+
"loss": 0.1678,
|
| 927 |
+
"step": 2875
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 2.3789991796554553,
|
| 931 |
+
"grad_norm": 0.2143106460571289,
|
| 932 |
+
"learning_rate": 6.264120022731524e-05,
|
| 933 |
+
"loss": 0.1718,
|
| 934 |
+
"step": 2900
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 2.399507793273175,
|
| 938 |
+
"grad_norm": 0.23707209527492523,
|
| 939 |
+
"learning_rate": 5.8744800598827974e-05,
|
| 940 |
+
"loss": 0.1684,
|
| 941 |
+
"step": 2925
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 2.4200164068908943,
|
| 945 |
+
"grad_norm": 0.22467195987701416,
|
| 946 |
+
"learning_rate": 5.495732322845071e-05,
|
| 947 |
+
"loss": 0.1715,
|
| 948 |
+
"step": 2950
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 2.4405250205086135,
|
| 952 |
+
"grad_norm": 0.1992505043745041,
|
| 953 |
+
"learning_rate": 5.128092513251048e-05,
|
| 954 |
+
"loss": 0.1688,
|
| 955 |
+
"step": 2975
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 2.4610336341263332,
|
| 959 |
+
"grad_norm": 0.21786148846149445,
|
| 960 |
+
"learning_rate": 4.771770006628068e-05,
|
| 961 |
+
"loss": 0.1644,
|
| 962 |
+
"step": 3000
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 2.4610336341263332,
|
| 966 |
+
"eval_loss": 0.2070690244436264,
|
| 967 |
+
"eval_runtime": 19.325,
|
| 968 |
+
"eval_samples_per_second": 51.746,
|
| 969 |
+
"eval_steps_per_second": 0.828,
|
| 970 |
+
"step": 3000
|
| 971 |
+
},
|
| 972 |
+
{
|
| 973 |
+
"epoch": 2.4815422477440525,
|
| 974 |
+
"grad_norm": 0.2135370373725891,
|
| 975 |
+
"learning_rate": 4.426967733156112e-05,
|
| 976 |
+
"loss": 0.1666,
|
| 977 |
+
"step": 3025
|
| 978 |
+
},
|
| 979 |
+
{
|
| 980 |
+
"epoch": 2.502050861361772,
|
| 981 |
+
"grad_norm": 0.18468019366264343,
|
| 982 |
+
"learning_rate": 4.0938820620964905e-05,
|
| 983 |
+
"loss": 0.1639,
|
| 984 |
+
"step": 3050
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 2.5225594749794915,
|
| 988 |
+
"grad_norm": 0.2244136482477188,
|
| 989 |
+
"learning_rate": 3.772702689957069e-05,
|
| 990 |
+
"loss": 0.1597,
|
| 991 |
+
"step": 3075
|
| 992 |
+
},
|
| 993 |
+
{
|
| 994 |
+
"epoch": 2.543068088597211,
|
| 995 |
+
"grad_norm": 0.2008039504289627,
|
| 996 |
+
"learning_rate": 3.4636125324576744e-05,
|
| 997 |
+
"loss": 0.1711,
|
| 998 |
+
"step": 3100
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 2.56357670221493,
|
| 1002 |
+
"grad_norm": 0.22959856688976288,
|
| 1003 |
+
"learning_rate": 3.166787620357328e-05,
|
| 1004 |
+
"loss": 0.159,
|
| 1005 |
+
"step": 3125
|
| 1006 |
+
},
|
| 1007 |
+
{
|
| 1008 |
+
"epoch": 2.5840853158326498,
|
| 1009 |
+
"grad_norm": 0.2295537292957306,
|
| 1010 |
+
"learning_rate": 2.8823969992024384e-05,
|
| 1011 |
+
"loss": 0.163,
|
| 1012 |
+
"step": 3150
|
| 1013 |
+
},
|
| 1014 |
+
{
|
| 1015 |
+
"epoch": 2.604593929450369,
|
| 1016 |
+
"grad_norm": 0.19450610876083374,
|
| 1017 |
+
"learning_rate": 2.6106026330532827e-05,
|
| 1018 |
+
"loss": 0.1653,
|
| 1019 |
+
"step": 3175
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 2.6251025430680883,
|
| 1023 |
+
"grad_norm": 0.20349068939685822,
|
| 1024 |
+
"learning_rate": 2.3515593122433676e-05,
|
| 1025 |
+
"loss": 0.1697,
|
| 1026 |
+
"step": 3200
|
| 1027 |
+
},
|
| 1028 |
+
{
|
| 1029 |
+
"epoch": 2.6251025430680883,
|
| 1030 |
+
"eval_loss": 0.2064325213432312,
|
| 1031 |
+
"eval_runtime": 19.3326,
|
| 1032 |
+
"eval_samples_per_second": 51.726,
|
| 1033 |
+
"eval_steps_per_second": 0.828,
|
| 1034 |
+
"step": 3200
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 2.645611156685808,
|
| 1038 |
+
"grad_norm": 0.23804035782814026,
|
| 1039 |
+
"learning_rate": 2.105414565224417e-05,
|
| 1040 |
+
"loss": 0.177,
|
| 1041 |
+
"step": 3225
|
| 1042 |
+
},
|
| 1043 |
+
{
|
| 1044 |
+
"epoch": 2.6661197703035278,
|
| 1045 |
+
"grad_norm": 0.21700942516326904,
|
| 1046 |
+
"learning_rate": 1.8723085745470352e-05,
|
| 1047 |
+
"loss": 0.162,
|
| 1048 |
+
"step": 3250
|
| 1049 |
+
},
|
| 1050 |
+
{
|
| 1051 |
+
"epoch": 2.686628383921247,
|
| 1052 |
+
"grad_norm": 0.2227737456560135,
|
| 1053 |
+
"learning_rate": 1.6523740970249527e-05,
|
| 1054 |
+
"loss": 0.1681,
|
| 1055 |
+
"step": 3275
|
| 1056 |
+
},
|
| 1057 |
+
{
|
| 1058 |
+
"epoch": 2.7071369975389663,
|
| 1059 |
+
"grad_norm": 0.18803074955940247,
|
| 1060 |
+
"learning_rate": 1.4457363881283741e-05,
|
| 1061 |
+
"loss": 0.1623,
|
| 1062 |
+
"step": 3300
|
| 1063 |
+
},
|
| 1064 |
+
{
|
| 1065 |
+
"epoch": 2.727645611156686,
|
| 1066 |
+
"grad_norm": 0.21387532353401184,
|
| 1067 |
+
"learning_rate": 1.2525131306493603e-05,
|
| 1068 |
+
"loss": 0.1611,
|
| 1069 |
+
"step": 3325
|
| 1070 |
+
},
|
| 1071 |
+
{
|
| 1072 |
+
"epoch": 2.7481542247744053,
|
| 1073 |
+
"grad_norm": 0.20090188086032867,
|
| 1074 |
+
"learning_rate": 1.072814367680025e-05,
|
| 1075 |
+
"loss": 0.1573,
|
| 1076 |
+
"step": 3350
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 2.7686628383921246,
|
| 1080 |
+
"grad_norm": 0.24346517026424408,
|
| 1081 |
+
"learning_rate": 9.067424399415769e-06,
|
| 1082 |
+
"loss": 0.1745,
|
| 1083 |
+
"step": 3375
|
| 1084 |
+
},
|
| 1085 |
+
{
|
| 1086 |
+
"epoch": 2.7891714520098443,
|
| 1087 |
+
"grad_norm": 0.19440355896949768,
|
| 1088 |
+
"learning_rate": 7.543919274999978e-06,
|
| 1089 |
+
"loss": 0.1617,
|
| 1090 |
+
"step": 3400
|
| 1091 |
+
},
|
| 1092 |
+
{
|
| 1093 |
+
"epoch": 2.7891714520098443,
|
| 1094 |
+
"eval_loss": 0.20586562156677246,
|
| 1095 |
+
"eval_runtime": 19.3052,
|
| 1096 |
+
"eval_samples_per_second": 51.8,
|
| 1097 |
+
"eval_steps_per_second": 0.829,
|
| 1098 |
+
"step": 3400
|
| 1099 |
+
},
|
| 1100 |
+
{
|
| 1101 |
+
"epoch": 2.8096800656275636,
|
| 1102 |
+
"grad_norm": 0.19979579746723175,
|
| 1103 |
+
"learning_rate": 6.158495959014826e-06,
|
| 1104 |
+
"loss": 0.1698,
|
| 1105 |
+
"step": 3425
|
| 1106 |
+
},
|
| 1107 |
+
{
|
| 1108 |
+
"epoch": 2.830188679245283,
|
| 1109 |
+
"grad_norm": 0.21324074268341064,
|
| 1110 |
+
"learning_rate": 4.911943467583751e-06,
|
| 1111 |
+
"loss": 0.1611,
|
| 1112 |
+
"step": 3450
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 2.8506972928630026,
|
| 1116 |
+
"grad_norm": 0.21548697352409363,
|
| 1117 |
+
"learning_rate": 3.80497172813693e-06,
|
| 1118 |
+
"loss": 0.1689,
|
| 1119 |
+
"step": 3475
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 2.871205906480722,
|
| 1123 |
+
"grad_norm": 0.19309598207473755,
|
| 1124 |
+
"learning_rate": 2.838211175098665e-06,
|
| 1125 |
+
"loss": 0.16,
|
| 1126 |
+
"step": 3500
|
| 1127 |
+
},
|
| 1128 |
+
{
|
| 1129 |
+
"epoch": 2.891714520098441,
|
| 1130 |
+
"grad_norm": 0.21341754496097565,
|
| 1131 |
+
"learning_rate": 2.012212390846896e-06,
|
| 1132 |
+
"loss": 0.1621,
|
| 1133 |
+
"step": 3525
|
| 1134 |
+
},
|
| 1135 |
+
{
|
| 1136 |
+
"epoch": 2.912223133716161,
|
| 1137 |
+
"grad_norm": 0.2006499022245407,
|
| 1138 |
+
"learning_rate": 1.3274457921496896e-06,
|
| 1139 |
+
"loss": 0.1619,
|
| 1140 |
+
"step": 3550
|
| 1141 |
+
},
|
| 1142 |
+
{
|
| 1143 |
+
"epoch": 2.93273174733388,
|
| 1144 |
+
"grad_norm": 0.21944327652454376,
|
| 1145 |
+
"learning_rate": 7.843013622569062e-07,
|
| 1146 |
+
"loss": 0.1605,
|
| 1147 |
+
"step": 3575
|
| 1148 |
+
},
|
| 1149 |
+
{
|
| 1150 |
+
"epoch": 2.9532403609516,
|
| 1151 |
+
"grad_norm": 0.188226118683815,
|
| 1152 |
+
"learning_rate": 3.830884287997216e-07,
|
| 1153 |
+
"loss": 0.1668,
|
| 1154 |
+
"step": 3600
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 2.9532403609516,
|
| 1158 |
+
"eval_loss": 0.20568273961544037,
|
| 1159 |
+
"eval_runtime": 19.3042,
|
| 1160 |
+
"eval_samples_per_second": 51.802,
|
| 1161 |
+
"eval_steps_per_second": 0.829,
|
| 1162 |
+
"step": 3600
|
| 1163 |
+
},
|
| 1164 |
+
{
|
| 1165 |
+
"epoch": 2.973748974569319,
|
| 1166 |
+
"grad_norm": 0.18749408423900604,
|
| 1167 |
+
"learning_rate": 1.240354876246852e-07,
|
| 1168 |
+
"loss": 0.1624,
|
| 1169 |
+
"step": 3625
|
| 1170 |
+
},
|
| 1171 |
+
{
|
| 1172 |
+
"epoch": 2.994257588187039,
|
| 1173 |
+
"grad_norm": 0.21548126637935638,
|
| 1174 |
+
"learning_rate": 7.290072662424763e-09,
|
| 1175 |
+
"loss": 0.1601,
|
| 1176 |
+
"step": 3650
|
| 1177 |
+
},
|
| 1178 |
+
{
|
| 1179 |
+
"epoch": 3.0,
|
| 1180 |
+
"step": 3657,
|
| 1181 |
+
"total_flos": 2.37785441107968e+18,
|
| 1182 |
+
"train_loss": 0.22604482963432665,
|
| 1183 |
+
"train_runtime": 3705.7117,
|
| 1184 |
+
"train_samples_per_second": 31.573,
|
| 1185 |
+
"train_steps_per_second": 0.987
|
| 1186 |
+
}
|
| 1187 |
+
],
|
| 1188 |
+
"logging_steps": 25,
|
| 1189 |
+
"max_steps": 3657,
|
| 1190 |
+
"num_input_tokens_seen": 0,
|
| 1191 |
+
"num_train_epochs": 3,
|
| 1192 |
+
"save_steps": 0,
|
| 1193 |
+
"stateful_callbacks": {
|
| 1194 |
+
"TrainerControl": {
|
| 1195 |
+
"args": {
|
| 1196 |
+
"should_epoch_stop": false,
|
| 1197 |
+
"should_evaluate": false,
|
| 1198 |
+
"should_log": false,
|
| 1199 |
+
"should_save": true,
|
| 1200 |
+
"should_training_stop": true
|
| 1201 |
+
},
|
| 1202 |
+
"attributes": {}
|
| 1203 |
+
}
|
| 1204 |
+
},
|
| 1205 |
+
"total_flos": 2.37785441107968e+18,
|
| 1206 |
+
"train_batch_size": 32,
|
| 1207 |
+
"trial_name": null,
|
| 1208 |
+
"trial_params": null
|
| 1209 |
+
}
|
nl_tasks/exps/run_ex16_3ep/ft/adapter_config.json
ADDED
|
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"T": 1.0,
|
| 3 |
+
"base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
|
| 4 |
+
"bias": "none",
|
| 5 |
+
"inference_mode": false,
|
| 6 |
+
"layers_to_transform": null,
|
| 7 |
+
"modules_to_save": null,
|
| 8 |
+
"num_rotations": 1,
|
| 9 |
+
"peft_type": "ROTATION",
|
| 10 |
+
"r": 16,
|
| 11 |
+
"revision": null,
|
| 12 |
+
"target_modules": [
|
| 13 |
+
"v_proj",
|
| 14 |
+
"q_proj"
|
| 15 |
+
],
|
| 16 |
+
"target_modules_to_skip": null,
|
| 17 |
+
"task_type": "CAUSAL_LM"
|
| 18 |
+
}
|
nl_tasks/exps/run_ex16_3ep/ft/added_tokens.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"[PAD]": 32000
|
| 3 |
+
}
|
nl_tasks/exps/run_ex16_3ep/ft/special_tokens_map.json
ADDED
|
@@ -0,0 +1,30 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": {
|
| 3 |
+
"content": "</s>",
|
| 4 |
+
"lstrip": false,
|
| 5 |
+
"normalized": false,
|
| 6 |
+
"rstrip": false,
|
| 7 |
+
"single_word": false
|
| 8 |
+
},
|
| 9 |
+
"eos_token": {
|
| 10 |
+
"content": "</s>",
|
| 11 |
+
"lstrip": false,
|
| 12 |
+
"normalized": false,
|
| 13 |
+
"rstrip": false,
|
| 14 |
+
"single_word": false
|
| 15 |
+
},
|
| 16 |
+
"pad_token": {
|
| 17 |
+
"content": "[PAD]",
|
| 18 |
+
"lstrip": false,
|
| 19 |
+
"normalized": false,
|
| 20 |
+
"rstrip": false,
|
| 21 |
+
"single_word": false
|
| 22 |
+
},
|
| 23 |
+
"unk_token": {
|
| 24 |
+
"content": "</s>",
|
| 25 |
+
"lstrip": false,
|
| 26 |
+
"normalized": false,
|
| 27 |
+
"rstrip": false,
|
| 28 |
+
"single_word": false
|
| 29 |
+
}
|
| 30 |
+
}
|
nl_tasks/exps/run_ex16_3ep/ft/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
nl_tasks/exps/run_ex16_3ep/ft/tokenizer.model
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
|
| 3 |
+
size 499723
|
nl_tasks/exps/run_ex16_3ep/ft/tokenizer_config.json
ADDED
|
@@ -0,0 +1,51 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_bos_token": true,
|
| 3 |
+
"add_eos_token": false,
|
| 4 |
+
"add_prefix_space": null,
|
| 5 |
+
"added_tokens_decoder": {
|
| 6 |
+
"0": {
|
| 7 |
+
"content": "<unk>",
|
| 8 |
+
"lstrip": false,
|
| 9 |
+
"normalized": false,
|
| 10 |
+
"rstrip": false,
|
| 11 |
+
"single_word": false,
|
| 12 |
+
"special": true
|
| 13 |
+
},
|
| 14 |
+
"1": {
|
| 15 |
+
"content": "<s>",
|
| 16 |
+
"lstrip": false,
|
| 17 |
+
"normalized": false,
|
| 18 |
+
"rstrip": false,
|
| 19 |
+
"single_word": false,
|
| 20 |
+
"special": true
|
| 21 |
+
},
|
| 22 |
+
"2": {
|
| 23 |
+
"content": "</s>",
|
| 24 |
+
"lstrip": false,
|
| 25 |
+
"normalized": false,
|
| 26 |
+
"rstrip": false,
|
| 27 |
+
"single_word": false,
|
| 28 |
+
"special": true
|
| 29 |
+
},
|
| 30 |
+
"32000": {
|
| 31 |
+
"content": "[PAD]",
|
| 32 |
+
"lstrip": false,
|
| 33 |
+
"normalized": false,
|
| 34 |
+
"rstrip": false,
|
| 35 |
+
"single_word": false,
|
| 36 |
+
"special": true
|
| 37 |
+
}
|
| 38 |
+
},
|
| 39 |
+
"bos_token": "</s>",
|
| 40 |
+
"clean_up_tokenization_spaces": false,
|
| 41 |
+
"eos_token": "</s>",
|
| 42 |
+
"extra_special_tokens": {},
|
| 43 |
+
"legacy": false,
|
| 44 |
+
"model_max_length": 512,
|
| 45 |
+
"pad_token": "[PAD]",
|
| 46 |
+
"padding_side": "right",
|
| 47 |
+
"sp_model_kwargs": {},
|
| 48 |
+
"tokenizer_class": "LlamaTokenizer",
|
| 49 |
+
"unk_token": "</s>",
|
| 50 |
+
"use_default_system_prompt": false
|
| 51 |
+
}
|
nl_tasks/exps/run_ex16_3ep/ft2/adapter_config.json
ADDED
|
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"T": 1.0,
|
| 3 |
+
"base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
|
| 4 |
+
"bias": "none",
|
| 5 |
+
"inference_mode": true,
|
| 6 |
+
"layers_to_transform": null,
|
| 7 |
+
"modules_to_save": null,
|
| 8 |
+
"num_rotations": 1,
|
| 9 |
+
"peft_type": "ROTATION",
|
| 10 |
+
"r": 16,
|
| 11 |
+
"revision": null,
|
| 12 |
+
"target_modules": [
|
| 13 |
+
"v_proj",
|
| 14 |
+
"q_proj"
|
| 15 |
+
],
|
| 16 |
+
"target_modules_to_skip": null,
|
| 17 |
+
"task_type": "CAUSAL_LM"
|
| 18 |
+
}
|
nl_tasks/exps/run_ex16_3ep/ft2/adapter_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:9586ec31356c8b78b6439055739a8423e161e76302d4f2e82b4847793d501e7d
|
| 3 |
+
size 33602915
|
nl_tasks/exps/run_ex16_3ep/trainer_state.json
ADDED
|
@@ -0,0 +1,1209 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 3.0,
|
| 6 |
+
"eval_steps": 200,
|
| 7 |
+
"global_step": 3657,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.020508613617719443,
|
| 14 |
+
"grad_norm": 0.5933114290237427,
|
| 15 |
+
"learning_rate": 6.557377049180328e-05,
|
| 16 |
+
"loss": 0.7845,
|
| 17 |
+
"step": 25
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.04101722723543889,
|
| 21 |
+
"grad_norm": 0.20545591413974762,
|
| 22 |
+
"learning_rate": 0.00013387978142076503,
|
| 23 |
+
"loss": 0.4732,
|
| 24 |
+
"step": 50
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.06152584085315833,
|
| 28 |
+
"grad_norm": 0.16550013422966003,
|
| 29 |
+
"learning_rate": 0.00020218579234972678,
|
| 30 |
+
"loss": 0.4016,
|
| 31 |
+
"step": 75
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.08203445447087777,
|
| 35 |
+
"grad_norm": 0.20370103418827057,
|
| 36 |
+
"learning_rate": 0.00027049180327868856,
|
| 37 |
+
"loss": 0.3572,
|
| 38 |
+
"step": 100
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.10254306808859721,
|
| 42 |
+
"grad_norm": 0.18663452565670013,
|
| 43 |
+
"learning_rate": 0.00033879781420765025,
|
| 44 |
+
"loss": 0.3588,
|
| 45 |
+
"step": 125
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.12305168170631665,
|
| 49 |
+
"grad_norm": 0.17238572239875793,
|
| 50 |
+
"learning_rate": 0.00040710382513661205,
|
| 51 |
+
"loss": 0.3399,
|
| 52 |
+
"step": 150
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.1435602953240361,
|
| 56 |
+
"grad_norm": 0.1672045886516571,
|
| 57 |
+
"learning_rate": 0.00047540983606557375,
|
| 58 |
+
"loss": 0.3282,
|
| 59 |
+
"step": 175
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.16406890894175555,
|
| 63 |
+
"grad_norm": 0.2247926890850067,
|
| 64 |
+
"learning_rate": 0.0005437158469945356,
|
| 65 |
+
"loss": 0.3367,
|
| 66 |
+
"step": 200
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.16406890894175555,
|
| 70 |
+
"eval_loss": 0.30540144443511963,
|
| 71 |
+
"eval_runtime": 19.5674,
|
| 72 |
+
"eval_samples_per_second": 51.105,
|
| 73 |
+
"eval_steps_per_second": 0.818,
|
| 74 |
+
"step": 200
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.184577522559475,
|
| 78 |
+
"grad_norm": 0.1778956949710846,
|
| 79 |
+
"learning_rate": 0.0006120218579234972,
|
| 80 |
+
"loss": 0.3304,
|
| 81 |
+
"step": 225
|
| 82 |
+
},
|
| 83 |
+
{
|
| 84 |
+
"epoch": 0.20508613617719443,
|
| 85 |
+
"grad_norm": 0.18393848836421967,
|
| 86 |
+
"learning_rate": 0.000680327868852459,
|
| 87 |
+
"loss": 0.3073,
|
| 88 |
+
"step": 250
|
| 89 |
+
},
|
| 90 |
+
{
|
| 91 |
+
"epoch": 0.22559474979491387,
|
| 92 |
+
"grad_norm": 0.17240218818187714,
|
| 93 |
+
"learning_rate": 0.0007486338797814209,
|
| 94 |
+
"loss": 0.307,
|
| 95 |
+
"step": 275
|
| 96 |
+
},
|
| 97 |
+
{
|
| 98 |
+
"epoch": 0.2461033634126333,
|
| 99 |
+
"grad_norm": 0.2195698320865631,
|
| 100 |
+
"learning_rate": 0.0008169398907103825,
|
| 101 |
+
"loss": 0.3127,
|
| 102 |
+
"step": 300
|
| 103 |
+
},
|
| 104 |
+
{
|
| 105 |
+
"epoch": 0.2666119770303528,
|
| 106 |
+
"grad_norm": 0.21643932163715363,
|
| 107 |
+
"learning_rate": 0.0008852459016393442,
|
| 108 |
+
"loss": 0.3163,
|
| 109 |
+
"step": 325
|
| 110 |
+
},
|
| 111 |
+
{
|
| 112 |
+
"epoch": 0.2871205906480722,
|
| 113 |
+
"grad_norm": 0.20525363087654114,
|
| 114 |
+
"learning_rate": 0.0009535519125683059,
|
| 115 |
+
"loss": 0.3099,
|
| 116 |
+
"step": 350
|
| 117 |
+
},
|
| 118 |
+
{
|
| 119 |
+
"epoch": 0.30762920426579166,
|
| 120 |
+
"grad_norm": 0.22471587359905243,
|
| 121 |
+
"learning_rate": 0.0009999854198546752,
|
| 122 |
+
"loss": 0.323,
|
| 123 |
+
"step": 375
|
| 124 |
+
},
|
| 125 |
+
{
|
| 126 |
+
"epoch": 0.3281378178835111,
|
| 127 |
+
"grad_norm": 0.2640959322452545,
|
| 128 |
+
"learning_rate": 0.0009997519290247507,
|
| 129 |
+
"loss": 0.3093,
|
| 130 |
+
"step": 400
|
| 131 |
+
},
|
| 132 |
+
{
|
| 133 |
+
"epoch": 0.3281378178835111,
|
| 134 |
+
"eval_loss": 0.2868829369544983,
|
| 135 |
+
"eval_runtime": 19.292,
|
| 136 |
+
"eval_samples_per_second": 51.835,
|
| 137 |
+
"eval_steps_per_second": 0.829,
|
| 138 |
+
"step": 400
|
| 139 |
+
},
|
| 140 |
+
{
|
| 141 |
+
"epoch": 0.34864643150123054,
|
| 142 |
+
"grad_norm": 0.19907452166080475,
|
| 143 |
+
"learning_rate": 0.0009992338231424006,
|
| 144 |
+
"loss": 0.3161,
|
| 145 |
+
"step": 425
|
| 146 |
+
},
|
| 147 |
+
{
|
| 148 |
+
"epoch": 0.36915504511895,
|
| 149 |
+
"grad_norm": 0.25168290734291077,
|
| 150 |
+
"learning_rate": 0.000998431397275486,
|
| 151 |
+
"loss": 0.3153,
|
| 152 |
+
"step": 450
|
| 153 |
+
},
|
| 154 |
+
{
|
| 155 |
+
"epoch": 0.3896636587366694,
|
| 156 |
+
"grad_norm": 0.29462915658950806,
|
| 157 |
+
"learning_rate": 0.0009973451084157006,
|
| 158 |
+
"loss": 0.3047,
|
| 159 |
+
"step": 475
|
| 160 |
+
},
|
| 161 |
+
{
|
| 162 |
+
"epoch": 0.41017227235438886,
|
| 163 |
+
"grad_norm": 0.2352636158466339,
|
| 164 |
+
"learning_rate": 0.0009959755752183062,
|
| 165 |
+
"loss": 0.2897,
|
| 166 |
+
"step": 500
|
| 167 |
+
},
|
| 168 |
+
{
|
| 169 |
+
"epoch": 0.4306808859721083,
|
| 170 |
+
"grad_norm": 0.22885267436504364,
|
| 171 |
+
"learning_rate": 0.0009943235776498026,
|
| 172 |
+
"loss": 0.2873,
|
| 173 |
+
"step": 525
|
| 174 |
+
},
|
| 175 |
+
{
|
| 176 |
+
"epoch": 0.45118949958982774,
|
| 177 |
+
"grad_norm": 0.22446544468402863,
|
| 178 |
+
"learning_rate": 0.0009923900565437262,
|
| 179 |
+
"loss": 0.2953,
|
| 180 |
+
"step": 550
|
| 181 |
+
},
|
| 182 |
+
{
|
| 183 |
+
"epoch": 0.4716981132075472,
|
| 184 |
+
"grad_norm": 0.2551325559616089,
|
| 185 |
+
"learning_rate": 0.0009901761130648326,
|
| 186 |
+
"loss": 0.2858,
|
| 187 |
+
"step": 575
|
| 188 |
+
},
|
| 189 |
+
{
|
| 190 |
+
"epoch": 0.4922067268252666,
|
| 191 |
+
"grad_norm": 0.2091352343559265,
|
| 192 |
+
"learning_rate": 0.0009876830080819703,
|
| 193 |
+
"loss": 0.3045,
|
| 194 |
+
"step": 600
|
| 195 |
+
},
|
| 196 |
+
{
|
| 197 |
+
"epoch": 0.4922067268252666,
|
| 198 |
+
"eval_loss": 0.2704485356807709,
|
| 199 |
+
"eval_runtime": 19.2756,
|
| 200 |
+
"eval_samples_per_second": 51.879,
|
| 201 |
+
"eval_steps_per_second": 0.83,
|
| 202 |
+
"step": 600
|
| 203 |
+
},
|
| 204 |
+
{
|
| 205 |
+
"epoch": 0.5127153404429861,
|
| 206 |
+
"grad_norm": 0.2221243530511856,
|
| 207 |
+
"learning_rate": 0.00098491216145,
|
| 208 |
+
"loss": 0.283,
|
| 209 |
+
"step": 625
|
| 210 |
+
},
|
| 211 |
+
{
|
| 212 |
+
"epoch": 0.5332239540607056,
|
| 213 |
+
"grad_norm": 0.2306085228919983,
|
| 214 |
+
"learning_rate": 0.0009818651512011685,
|
| 215 |
+
"loss": 0.2732,
|
| 216 |
+
"step": 650
|
| 217 |
+
},
|
| 218 |
+
{
|
| 219 |
+
"epoch": 0.5537325676784249,
|
| 220 |
+
"grad_norm": 0.22338023781776428,
|
| 221 |
+
"learning_rate": 0.0009785437126463995,
|
| 222 |
+
"loss": 0.2865,
|
| 223 |
+
"step": 675
|
| 224 |
+
},
|
| 225 |
+
{
|
| 226 |
+
"epoch": 0.5742411812961444,
|
| 227 |
+
"grad_norm": 0.3610588312149048,
|
| 228 |
+
"learning_rate": 0.0009749497373870129,
|
| 229 |
+
"loss": 0.2859,
|
| 230 |
+
"step": 700
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 0.5947497949138638,
|
| 234 |
+
"grad_norm": 0.2590906322002411,
|
| 235 |
+
"learning_rate": 0.0009710852722374326,
|
| 236 |
+
"loss": 0.2834,
|
| 237 |
+
"step": 725
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 0.6152584085315833,
|
| 241 |
+
"grad_norm": 0.23781649768352509,
|
| 242 |
+
"learning_rate": 0.0009669525180595009,
|
| 243 |
+
"loss": 0.2797,
|
| 244 |
+
"step": 750
|
| 245 |
+
},
|
| 246 |
+
{
|
| 247 |
+
"epoch": 0.6357670221493027,
|
| 248 |
+
"grad_norm": 0.19784238934516907,
|
| 249 |
+
"learning_rate": 0.0009625538285090593,
|
| 250 |
+
"loss": 0.2747,
|
| 251 |
+
"step": 775
|
| 252 |
+
},
|
| 253 |
+
{
|
| 254 |
+
"epoch": 0.6562756357670222,
|
| 255 |
+
"grad_norm": 0.20952032506465912,
|
| 256 |
+
"learning_rate": 0.0009578917086955117,
|
| 257 |
+
"loss": 0.2795,
|
| 258 |
+
"step": 800
|
| 259 |
+
},
|
| 260 |
+
{
|
| 261 |
+
"epoch": 0.6562756357670222,
|
| 262 |
+
"eval_loss": 0.2591809630393982,
|
| 263 |
+
"eval_runtime": 19.2722,
|
| 264 |
+
"eval_samples_per_second": 51.888,
|
| 265 |
+
"eval_steps_per_second": 0.83,
|
| 266 |
+
"step": 800
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 0.6767842493847416,
|
| 270 |
+
"grad_norm": 0.189390167593956,
|
| 271 |
+
"learning_rate": 0.0009529688137551328,
|
| 272 |
+
"loss": 0.2706,
|
| 273 |
+
"step": 825
|
| 274 |
+
},
|
| 275 |
+
{
|
| 276 |
+
"epoch": 0.6972928630024611,
|
| 277 |
+
"grad_norm": 0.224375382065773,
|
| 278 |
+
"learning_rate": 0.0009477879473389345,
|
| 279 |
+
"loss": 0.2754,
|
| 280 |
+
"step": 850
|
| 281 |
+
},
|
| 282 |
+
{
|
| 283 |
+
"epoch": 0.7178014766201805,
|
| 284 |
+
"grad_norm": 0.24536602199077606,
|
| 285 |
+
"learning_rate": 0.0009423520600159513,
|
| 286 |
+
"loss": 0.2658,
|
| 287 |
+
"step": 875
|
| 288 |
+
},
|
| 289 |
+
{
|
| 290 |
+
"epoch": 0.7383100902379,
|
| 291 |
+
"grad_norm": 0.25930875539779663,
|
| 292 |
+
"learning_rate": 0.0009366642475928535,
|
| 293 |
+
"loss": 0.2708,
|
| 294 |
+
"step": 900
|
| 295 |
+
},
|
| 296 |
+
{
|
| 297 |
+
"epoch": 0.7588187038556193,
|
| 298 |
+
"grad_norm": 0.23591011762619019,
|
| 299 |
+
"learning_rate": 0.0009307277493508466,
|
| 300 |
+
"loss": 0.2712,
|
| 301 |
+
"step": 925
|
| 302 |
+
},
|
| 303 |
+
{
|
| 304 |
+
"epoch": 0.7793273174733388,
|
| 305 |
+
"grad_norm": 0.17592450976371765,
|
| 306 |
+
"learning_rate": 0.0009245459462008587,
|
| 307 |
+
"loss": 0.2557,
|
| 308 |
+
"step": 950
|
| 309 |
+
},
|
| 310 |
+
{
|
| 311 |
+
"epoch": 0.7998359310910582,
|
| 312 |
+
"grad_norm": 0.2024001181125641,
|
| 313 |
+
"learning_rate": 0.0009181223587580702,
|
| 314 |
+
"loss": 0.2645,
|
| 315 |
+
"step": 975
|
| 316 |
+
},
|
| 317 |
+
{
|
| 318 |
+
"epoch": 0.8203445447087777,
|
| 319 |
+
"grad_norm": 0.24190227687358856,
|
| 320 |
+
"learning_rate": 0.0009114606453368778,
|
| 321 |
+
"loss": 0.2572,
|
| 322 |
+
"step": 1000
|
| 323 |
+
},
|
| 324 |
+
{
|
| 325 |
+
"epoch": 0.8203445447087777,
|
| 326 |
+
"eval_loss": 0.24514010548591614,
|
| 327 |
+
"eval_runtime": 19.3033,
|
| 328 |
+
"eval_samples_per_second": 51.805,
|
| 329 |
+
"eval_steps_per_second": 0.829,
|
| 330 |
+
"step": 1000
|
| 331 |
+
},
|
| 332 |
+
{
|
| 333 |
+
"epoch": 0.8408531583264971,
|
| 334 |
+
"grad_norm": 0.2223917841911316,
|
| 335 |
+
"learning_rate": 0.0009045645998674388,
|
| 336 |
+
"loss": 0.2555,
|
| 337 |
+
"step": 1025
|
| 338 |
+
},
|
| 339 |
+
{
|
| 340 |
+
"epoch": 0.8613617719442166,
|
| 341 |
+
"grad_norm": 0.20406895875930786,
|
| 342 |
+
"learning_rate": 0.000897438149734979,
|
| 343 |
+
"loss": 0.254,
|
| 344 |
+
"step": 1050
|
| 345 |
+
},
|
| 346 |
+
{
|
| 347 |
+
"epoch": 0.881870385561936,
|
| 348 |
+
"grad_norm": 0.18564128875732422,
|
| 349 |
+
"learning_rate": 0.0008900853535430986,
|
| 350 |
+
"loss": 0.2591,
|
| 351 |
+
"step": 1075
|
| 352 |
+
},
|
| 353 |
+
{
|
| 354 |
+
"epoch": 0.9023789991796555,
|
| 355 |
+
"grad_norm": 0.18205063045024872,
|
| 356 |
+
"learning_rate": 0.0008825103988023442,
|
| 357 |
+
"loss": 0.2588,
|
| 358 |
+
"step": 1100
|
| 359 |
+
},
|
| 360 |
+
{
|
| 361 |
+
"epoch": 0.9228876127973749,
|
| 362 |
+
"grad_norm": 0.16957780718803406,
|
| 363 |
+
"learning_rate": 0.0008747175995453695,
|
| 364 |
+
"loss": 0.2537,
|
| 365 |
+
"step": 1125
|
| 366 |
+
},
|
| 367 |
+
{
|
| 368 |
+
"epoch": 0.9433962264150944,
|
| 369 |
+
"grad_norm": 0.16273853182792664,
|
| 370 |
+
"learning_rate": 0.0008667113938700395,
|
| 371 |
+
"loss": 0.2533,
|
| 372 |
+
"step": 1150
|
| 373 |
+
},
|
| 374 |
+
{
|
| 375 |
+
"epoch": 0.9639048400328137,
|
| 376 |
+
"grad_norm": 0.16668862104415894,
|
| 377 |
+
"learning_rate": 0.0008584963414118765,
|
| 378 |
+
"loss": 0.2571,
|
| 379 |
+
"step": 1175
|
| 380 |
+
},
|
| 381 |
+
{
|
| 382 |
+
"epoch": 0.9844134536505332,
|
| 383 |
+
"grad_norm": 0.17593467235565186,
|
| 384 |
+
"learning_rate": 0.0008500771207472907,
|
| 385 |
+
"loss": 0.2625,
|
| 386 |
+
"step": 1200
|
| 387 |
+
},
|
| 388 |
+
{
|
| 389 |
+
"epoch": 0.9844134536505332,
|
| 390 |
+
"eval_loss": 0.23641632497310638,
|
| 391 |
+
"eval_runtime": 19.2597,
|
| 392 |
+
"eval_samples_per_second": 51.922,
|
| 393 |
+
"eval_steps_per_second": 0.831,
|
| 394 |
+
"step": 1200
|
| 395 |
+
},
|
| 396 |
+
{
|
| 397 |
+
"epoch": 1.0049220672682526,
|
| 398 |
+
"grad_norm": 0.18487612903118134,
|
| 399 |
+
"learning_rate": 0.0008414585267290715,
|
| 400 |
+
"loss": 0.2442,
|
| 401 |
+
"step": 1225
|
| 402 |
+
},
|
| 403 |
+
{
|
| 404 |
+
"epoch": 1.0254306808859721,
|
| 405 |
+
"grad_norm": 0.1899247169494629,
|
| 406 |
+
"learning_rate": 0.0008326454677556577,
|
| 407 |
+
"loss": 0.2271,
|
| 408 |
+
"step": 1250
|
| 409 |
+
},
|
| 410 |
+
{
|
| 411 |
+
"epoch": 1.0459392945036916,
|
| 412 |
+
"grad_norm": 0.18518805503845215,
|
| 413 |
+
"learning_rate": 0.0008236429629757432,
|
| 414 |
+
"loss": 0.2307,
|
| 415 |
+
"step": 1275
|
| 416 |
+
},
|
| 417 |
+
{
|
| 418 |
+
"epoch": 1.066447908121411,
|
| 419 |
+
"grad_norm": 0.1996040940284729,
|
| 420 |
+
"learning_rate": 0.0008144561394298075,
|
| 421 |
+
"loss": 0.2328,
|
| 422 |
+
"step": 1300
|
| 423 |
+
},
|
| 424 |
+
{
|
| 425 |
+
"epoch": 1.0869565217391304,
|
| 426 |
+
"grad_norm": 0.18206605315208435,
|
| 427 |
+
"learning_rate": 0.0008050902291302019,
|
| 428 |
+
"loss": 0.23,
|
| 429 |
+
"step": 1325
|
| 430 |
+
},
|
| 431 |
+
{
|
| 432 |
+
"epoch": 1.1074651353568499,
|
| 433 |
+
"grad_norm": 0.14665578305721283,
|
| 434 |
+
"learning_rate": 0.0007955505660814515,
|
| 435 |
+
"loss": 0.2216,
|
| 436 |
+
"step": 1350
|
| 437 |
+
},
|
| 438 |
+
{
|
| 439 |
+
"epoch": 1.1279737489745694,
|
| 440 |
+
"grad_norm": 0.18593929708003998,
|
| 441 |
+
"learning_rate": 0.0007858425832424727,
|
| 442 |
+
"loss": 0.2253,
|
| 443 |
+
"step": 1375
|
| 444 |
+
},
|
| 445 |
+
{
|
| 446 |
+
"epoch": 1.1484823625922886,
|
| 447 |
+
"grad_norm": 0.16838900744915009,
|
| 448 |
+
"learning_rate": 0.000775971809432434,
|
| 449 |
+
"loss": 0.2163,
|
| 450 |
+
"step": 1400
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 1.1484823625922886,
|
| 454 |
+
"eval_loss": 0.23211660981178284,
|
| 455 |
+
"eval_runtime": 19.2711,
|
| 456 |
+
"eval_samples_per_second": 51.891,
|
| 457 |
+
"eval_steps_per_second": 0.83,
|
| 458 |
+
"step": 1400
|
| 459 |
+
},
|
| 460 |
+
{
|
| 461 |
+
"epoch": 1.1689909762100081,
|
| 462 |
+
"grad_norm": 0.17335142195224762,
|
| 463 |
+
"learning_rate": 0.0007659438661820235,
|
| 464 |
+
"loss": 0.2278,
|
| 465 |
+
"step": 1425
|
| 466 |
+
},
|
| 467 |
+
{
|
| 468 |
+
"epoch": 1.1894995898277276,
|
| 469 |
+
"grad_norm": 0.1807931661605835,
|
| 470 |
+
"learning_rate": 0.0007557644645319157,
|
| 471 |
+
"loss": 0.2301,
|
| 472 |
+
"step": 1450
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 1.2100082034454471,
|
| 476 |
+
"grad_norm": 0.1775919646024704,
|
| 477 |
+
"learning_rate": 0.0007454394017802622,
|
| 478 |
+
"loss": 0.2249,
|
| 479 |
+
"step": 1475
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 1.2305168170631666,
|
| 483 |
+
"grad_norm": 0.16616088151931763,
|
| 484 |
+
"learning_rate": 0.0007349745581810565,
|
| 485 |
+
"loss": 0.2353,
|
| 486 |
+
"step": 1500
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 1.251025430680886,
|
| 490 |
+
"grad_norm": 0.184806689620018,
|
| 491 |
+
"learning_rate": 0.0007243758935952547,
|
| 492 |
+
"loss": 0.2291,
|
| 493 |
+
"step": 1525
|
| 494 |
+
},
|
| 495 |
+
{
|
| 496 |
+
"epoch": 1.2715340442986054,
|
| 497 |
+
"grad_norm": 0.1499050408601761,
|
| 498 |
+
"learning_rate": 0.0007136494440965596,
|
| 499 |
+
"loss": 0.2223,
|
| 500 |
+
"step": 1550
|
| 501 |
+
},
|
| 502 |
+
{
|
| 503 |
+
"epoch": 1.2920426579163249,
|
| 504 |
+
"grad_norm": 0.13914547860622406,
|
| 505 |
+
"learning_rate": 0.0007028013185338001,
|
| 506 |
+
"loss": 0.2252,
|
| 507 |
+
"step": 1575
|
| 508 |
+
},
|
| 509 |
+
{
|
| 510 |
+
"epoch": 1.3125512715340442,
|
| 511 |
+
"grad_norm": 0.15225553512573242,
|
| 512 |
+
"learning_rate": 0.000691837695051865,
|
| 513 |
+
"loss": 0.2177,
|
| 514 |
+
"step": 1600
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 1.3125512715340442,
|
| 518 |
+
"eval_loss": 0.2241576910018921,
|
| 519 |
+
"eval_runtime": 19.3112,
|
| 520 |
+
"eval_samples_per_second": 51.783,
|
| 521 |
+
"eval_steps_per_second": 0.829,
|
| 522 |
+
"step": 1600
|
| 523 |
+
},
|
| 524 |
+
{
|
| 525 |
+
"epoch": 1.3330598851517639,
|
| 526 |
+
"grad_norm": 0.143367737531662,
|
| 527 |
+
"learning_rate": 0.00068076481757317,
|
| 528 |
+
"loss": 0.2206,
|
| 529 |
+
"step": 1625
|
| 530 |
+
},
|
| 531 |
+
{
|
| 532 |
+
"epoch": 1.3535684987694832,
|
| 533 |
+
"grad_norm": 0.1510310173034668,
|
| 534 |
+
"learning_rate": 0.0006695889922416659,
|
| 535 |
+
"loss": 0.2208,
|
| 536 |
+
"step": 1650
|
| 537 |
+
},
|
| 538 |
+
{
|
| 539 |
+
"epoch": 1.3740771123872026,
|
| 540 |
+
"grad_norm": 0.1739281564950943,
|
| 541 |
+
"learning_rate": 0.0006583165838314095,
|
| 542 |
+
"loss": 0.2192,
|
| 543 |
+
"step": 1675
|
| 544 |
+
},
|
| 545 |
+
{
|
| 546 |
+
"epoch": 1.3945857260049221,
|
| 547 |
+
"grad_norm": 0.14383970201015472,
|
| 548 |
+
"learning_rate": 0.0006469540121217438,
|
| 549 |
+
"loss": 0.2187,
|
| 550 |
+
"step": 1700
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 1.4150943396226414,
|
| 554 |
+
"grad_norm": 0.17553195357322693,
|
| 555 |
+
"learning_rate": 0.0006355077482411526,
|
| 556 |
+
"loss": 0.2171,
|
| 557 |
+
"step": 1725
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 1.435602953240361,
|
| 561 |
+
"grad_norm": 0.14919953048229218,
|
| 562 |
+
"learning_rate": 0.0006239843109818716,
|
| 563 |
+
"loss": 0.2167,
|
| 564 |
+
"step": 1750
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 1.4561115668580804,
|
| 568 |
+
"grad_norm": 0.15127669274806976,
|
| 569 |
+
"learning_rate": 0.0006123902630873521,
|
| 570 |
+
"loss": 0.222,
|
| 571 |
+
"step": 1775
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 1.4766201804758,
|
| 575 |
+
"grad_norm": 0.13554859161376953,
|
| 576 |
+
"learning_rate": 0.0006007322075146968,
|
| 577 |
+
"loss": 0.2127,
|
| 578 |
+
"step": 1800
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 1.4766201804758,
|
| 582 |
+
"eval_loss": 0.2200252115726471,
|
| 583 |
+
"eval_runtime": 19.2774,
|
| 584 |
+
"eval_samples_per_second": 51.874,
|
| 585 |
+
"eval_steps_per_second": 0.83,
|
| 586 |
+
"step": 1800
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 1.4971287940935194,
|
| 590 |
+
"grad_norm": 0.14612731337547302,
|
| 591 |
+
"learning_rate": 0.0005890167836741919,
|
| 592 |
+
"loss": 0.2122,
|
| 593 |
+
"step": 1825
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 1.5176374077112387,
|
| 597 |
+
"grad_norm": 0.14290480315685272,
|
| 598 |
+
"learning_rate": 0.0005772506636480781,
|
| 599 |
+
"loss": 0.2147,
|
| 600 |
+
"step": 1850
|
| 601 |
+
},
|
| 602 |
+
{
|
| 603 |
+
"epoch": 1.5381460213289582,
|
| 604 |
+
"grad_norm": 0.14055421948432922,
|
| 605 |
+
"learning_rate": 0.0005654405483907163,
|
| 606 |
+
"loss": 0.2138,
|
| 607 |
+
"step": 1875
|
| 608 |
+
},
|
| 609 |
+
{
|
| 610 |
+
"epoch": 1.5586546349466777,
|
| 611 |
+
"grad_norm": 0.16261056065559387,
|
| 612 |
+
"learning_rate": 0.0005535931639123083,
|
| 613 |
+
"loss": 0.2177,
|
| 614 |
+
"step": 1900
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 1.579163248564397,
|
| 618 |
+
"grad_norm": 0.1579464077949524,
|
| 619 |
+
"learning_rate": 0.00054171525744835,
|
| 620 |
+
"loss": 0.2201,
|
| 621 |
+
"step": 1925
|
| 622 |
+
},
|
| 623 |
+
{
|
| 624 |
+
"epoch": 1.5996718621821167,
|
| 625 |
+
"grad_norm": 0.1471942812204361,
|
| 626 |
+
"learning_rate": 0.0005298135936169937,
|
| 627 |
+
"loss": 0.2181,
|
| 628 |
+
"step": 1950
|
| 629 |
+
},
|
| 630 |
+
{
|
| 631 |
+
"epoch": 1.620180475799836,
|
| 632 |
+
"grad_norm": 0.13963663578033447,
|
| 633 |
+
"learning_rate": 0.000517894950566514,
|
| 634 |
+
"loss": 0.2117,
|
| 635 |
+
"step": 1975
|
| 636 |
+
},
|
| 637 |
+
{
|
| 638 |
+
"epoch": 1.6406890894175554,
|
| 639 |
+
"grad_norm": 0.14293774962425232,
|
| 640 |
+
"learning_rate": 0.0005059661161150655,
|
| 641 |
+
"loss": 0.2182,
|
| 642 |
+
"step": 2000
|
| 643 |
+
},
|
| 644 |
+
{
|
| 645 |
+
"epoch": 1.6406890894175554,
|
| 646 |
+
"eval_loss": 0.21523867547512054,
|
| 647 |
+
"eval_runtime": 19.2818,
|
| 648 |
+
"eval_samples_per_second": 51.862,
|
| 649 |
+
"eval_steps_per_second": 0.83,
|
| 650 |
+
"step": 2000
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 1.661197703035275,
|
| 654 |
+
"grad_norm": 0.13382603228092194,
|
| 655 |
+
"learning_rate": 0.0004940338838849347,
|
| 656 |
+
"loss": 0.2092,
|
| 657 |
+
"step": 2025
|
| 658 |
+
},
|
| 659 |
+
{
|
| 660 |
+
"epoch": 1.6817063166529942,
|
| 661 |
+
"grad_norm": 0.16645269095897675,
|
| 662 |
+
"learning_rate": 0.0004821050494334861,
|
| 663 |
+
"loss": 0.2106,
|
| 664 |
+
"step": 2050
|
| 665 |
+
},
|
| 666 |
+
{
|
| 667 |
+
"epoch": 1.7022149302707137,
|
| 668 |
+
"grad_norm": 0.16617609560489655,
|
| 669 |
+
"learning_rate": 0.0004701864063830064,
|
| 670 |
+
"loss": 0.2149,
|
| 671 |
+
"step": 2075
|
| 672 |
+
},
|
| 673 |
+
{
|
| 674 |
+
"epoch": 1.7227235438884332,
|
| 675 |
+
"grad_norm": 0.134024977684021,
|
| 676 |
+
"learning_rate": 0.0004582847425516501,
|
| 677 |
+
"loss": 0.2065,
|
| 678 |
+
"step": 2100
|
| 679 |
+
},
|
| 680 |
+
{
|
| 681 |
+
"epoch": 1.7432321575061525,
|
| 682 |
+
"grad_norm": 0.12853078544139862,
|
| 683 |
+
"learning_rate": 0.00044640683608769186,
|
| 684 |
+
"loss": 0.2099,
|
| 685 |
+
"step": 2125
|
| 686 |
+
},
|
| 687 |
+
{
|
| 688 |
+
"epoch": 1.7637407711238722,
|
| 689 |
+
"grad_norm": 0.12432336807250977,
|
| 690 |
+
"learning_rate": 0.0004345594516092838,
|
| 691 |
+
"loss": 0.2075,
|
| 692 |
+
"step": 2150
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 1.7842493847415914,
|
| 696 |
+
"grad_norm": 0.13519197702407837,
|
| 697 |
+
"learning_rate": 0.0004227493363519221,
|
| 698 |
+
"loss": 0.2162,
|
| 699 |
+
"step": 2175
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 1.804757998359311,
|
| 703 |
+
"grad_norm": 0.13607081770896912,
|
| 704 |
+
"learning_rate": 0.0004109832163258082,
|
| 705 |
+
"loss": 0.2081,
|
| 706 |
+
"step": 2200
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 1.804757998359311,
|
| 710 |
+
"eval_loss": 0.21113082766532898,
|
| 711 |
+
"eval_runtime": 19.3118,
|
| 712 |
+
"eval_samples_per_second": 51.782,
|
| 713 |
+
"eval_steps_per_second": 0.829,
|
| 714 |
+
"step": 2200
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 1.8252666119770304,
|
| 718 |
+
"grad_norm": 0.1487966775894165,
|
| 719 |
+
"learning_rate": 0.0003992677924853032,
|
| 720 |
+
"loss": 0.21,
|
| 721 |
+
"step": 2225
|
| 722 |
+
},
|
| 723 |
+
{
|
| 724 |
+
"epoch": 1.8457752255947497,
|
| 725 |
+
"grad_norm": 0.1369887739419937,
|
| 726 |
+
"learning_rate": 0.0003876097369126479,
|
| 727 |
+
"loss": 0.2107,
|
| 728 |
+
"step": 2250
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 1.8662838392124692,
|
| 732 |
+
"grad_norm": 0.1338726282119751,
|
| 733 |
+
"learning_rate": 0.0003760156890181283,
|
| 734 |
+
"loss": 0.2146,
|
| 735 |
+
"step": 2275
|
| 736 |
+
},
|
| 737 |
+
{
|
| 738 |
+
"epoch": 1.8867924528301887,
|
| 739 |
+
"grad_norm": 0.14062516391277313,
|
| 740 |
+
"learning_rate": 0.0003644922517588474,
|
| 741 |
+
"loss": 0.2081,
|
| 742 |
+
"step": 2300
|
| 743 |
+
},
|
| 744 |
+
{
|
| 745 |
+
"epoch": 1.907301066447908,
|
| 746 |
+
"grad_norm": 0.1451762616634369,
|
| 747 |
+
"learning_rate": 0.0003530459878782563,
|
| 748 |
+
"loss": 0.2165,
|
| 749 |
+
"step": 2325
|
| 750 |
+
},
|
| 751 |
+
{
|
| 752 |
+
"epoch": 1.9278096800656277,
|
| 753 |
+
"grad_norm": 0.13617286086082458,
|
| 754 |
+
"learning_rate": 0.00034168341616859065,
|
| 755 |
+
"loss": 0.1994,
|
| 756 |
+
"step": 2350
|
| 757 |
+
},
|
| 758 |
+
{
|
| 759 |
+
"epoch": 1.948318293683347,
|
| 760 |
+
"grad_norm": 0.12289093434810638,
|
| 761 |
+
"learning_rate": 0.0003304110077583341,
|
| 762 |
+
"loss": 0.2047,
|
| 763 |
+
"step": 2375
|
| 764 |
+
},
|
| 765 |
+
{
|
| 766 |
+
"epoch": 1.9688269073010665,
|
| 767 |
+
"grad_norm": 0.13330784440040588,
|
| 768 |
+
"learning_rate": 0.00031923518242683,
|
| 769 |
+
"loss": 0.2055,
|
| 770 |
+
"step": 2400
|
| 771 |
+
},
|
| 772 |
+
{
|
| 773 |
+
"epoch": 1.9688269073010665,
|
| 774 |
+
"eval_loss": 0.20672284066677094,
|
| 775 |
+
"eval_runtime": 19.3165,
|
| 776 |
+
"eval_samples_per_second": 51.769,
|
| 777 |
+
"eval_steps_per_second": 0.828,
|
| 778 |
+
"step": 2400
|
| 779 |
+
},
|
| 780 |
+
{
|
| 781 |
+
"epoch": 1.989335520918786,
|
| 782 |
+
"grad_norm": 0.13013282418251038,
|
| 783 |
+
"learning_rate": 0.000308162304948135,
|
| 784 |
+
"loss": 0.2023,
|
| 785 |
+
"step": 2425
|
| 786 |
+
},
|
| 787 |
+
{
|
| 788 |
+
"epoch": 2.0098441345365052,
|
| 789 |
+
"grad_norm": 0.1487245261669159,
|
| 790 |
+
"learning_rate": 0.00029719868146619997,
|
| 791 |
+
"loss": 0.1913,
|
| 792 |
+
"step": 2450
|
| 793 |
+
},
|
| 794 |
+
{
|
| 795 |
+
"epoch": 2.030352748154225,
|
| 796 |
+
"grad_norm": 0.1466096192598343,
|
| 797 |
+
"learning_rate": 0.0002863505559034406,
|
| 798 |
+
"loss": 0.1776,
|
| 799 |
+
"step": 2475
|
| 800 |
+
},
|
| 801 |
+
{
|
| 802 |
+
"epoch": 2.0508613617719442,
|
| 803 |
+
"grad_norm": 0.16183960437774658,
|
| 804 |
+
"learning_rate": 0.0002756241064047456,
|
| 805 |
+
"loss": 0.1794,
|
| 806 |
+
"step": 2500
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 2.0713699753896635,
|
| 810 |
+
"grad_norm": 0.1348283886909485,
|
| 811 |
+
"learning_rate": 0.00026502544181894374,
|
| 812 |
+
"loss": 0.1757,
|
| 813 |
+
"step": 2525
|
| 814 |
+
},
|
| 815 |
+
{
|
| 816 |
+
"epoch": 2.091878589007383,
|
| 817 |
+
"grad_norm": 0.14854110777378082,
|
| 818 |
+
"learning_rate": 0.00025456059821973783,
|
| 819 |
+
"loss": 0.1842,
|
| 820 |
+
"step": 2550
|
| 821 |
+
},
|
| 822 |
+
{
|
| 823 |
+
"epoch": 2.1123872026251025,
|
| 824 |
+
"grad_norm": 0.1386813372373581,
|
| 825 |
+
"learning_rate": 0.00024423553546808424,
|
| 826 |
+
"loss": 0.1768,
|
| 827 |
+
"step": 2575
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"epoch": 2.132895816242822,
|
| 831 |
+
"grad_norm": 0.1473291516304016,
|
| 832 |
+
"learning_rate": 0.00023405613381797665,
|
| 833 |
+
"loss": 0.1696,
|
| 834 |
+
"step": 2600
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 2.132895816242822,
|
| 838 |
+
"eval_loss": 0.20782257616519928,
|
| 839 |
+
"eval_runtime": 19.3116,
|
| 840 |
+
"eval_samples_per_second": 51.782,
|
| 841 |
+
"eval_steps_per_second": 0.829,
|
| 842 |
+
"step": 2600
|
| 843 |
+
},
|
| 844 |
+
{
|
| 845 |
+
"epoch": 2.1534044298605415,
|
| 846 |
+
"grad_norm": 0.1481078863143921,
|
| 847 |
+
"learning_rate": 0.00022402819056756602,
|
| 848 |
+
"loss": 0.1737,
|
| 849 |
+
"step": 2625
|
| 850 |
+
},
|
| 851 |
+
{
|
| 852 |
+
"epoch": 2.1739130434782608,
|
| 853 |
+
"grad_norm": 0.1274261176586151,
|
| 854 |
+
"learning_rate": 0.00021415741675752742,
|
| 855 |
+
"loss": 0.1725,
|
| 856 |
+
"step": 2650
|
| 857 |
+
},
|
| 858 |
+
{
|
| 859 |
+
"epoch": 2.1944216570959805,
|
| 860 |
+
"grad_norm": 0.15026845037937164,
|
| 861 |
+
"learning_rate": 0.00020444943391854866,
|
| 862 |
+
"loss": 0.1841,
|
| 863 |
+
"step": 2675
|
| 864 |
+
},
|
| 865 |
+
{
|
| 866 |
+
"epoch": 2.2149302707136997,
|
| 867 |
+
"grad_norm": 0.1312485784292221,
|
| 868 |
+
"learning_rate": 0.00019490977086979822,
|
| 869 |
+
"loss": 0.18,
|
| 870 |
+
"step": 2700
|
| 871 |
+
},
|
| 872 |
+
{
|
| 873 |
+
"epoch": 2.235438884331419,
|
| 874 |
+
"grad_norm": 0.15328241884708405,
|
| 875 |
+
"learning_rate": 0.0001855438605701925,
|
| 876 |
+
"loss": 0.1688,
|
| 877 |
+
"step": 2725
|
| 878 |
+
},
|
| 879 |
+
{
|
| 880 |
+
"epoch": 2.2559474979491387,
|
| 881 |
+
"grad_norm": 0.15922562777996063,
|
| 882 |
+
"learning_rate": 0.00017635703702425682,
|
| 883 |
+
"loss": 0.1812,
|
| 884 |
+
"step": 2750
|
| 885 |
+
},
|
| 886 |
+
{
|
| 887 |
+
"epoch": 2.276456111566858,
|
| 888 |
+
"grad_norm": 0.1465969830751419,
|
| 889 |
+
"learning_rate": 0.0001673545322443424,
|
| 890 |
+
"loss": 0.1785,
|
| 891 |
+
"step": 2775
|
| 892 |
+
},
|
| 893 |
+
{
|
| 894 |
+
"epoch": 2.2969647251845773,
|
| 895 |
+
"grad_norm": 0.16353848576545715,
|
| 896 |
+
"learning_rate": 0.00015854147327092855,
|
| 897 |
+
"loss": 0.1743,
|
| 898 |
+
"step": 2800
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 2.2969647251845773,
|
| 902 |
+
"eval_loss": 0.2060222029685974,
|
| 903 |
+
"eval_runtime": 19.3095,
|
| 904 |
+
"eval_samples_per_second": 51.788,
|
| 905 |
+
"eval_steps_per_second": 0.829,
|
| 906 |
+
"step": 2800
|
| 907 |
+
},
|
| 908 |
+
{
|
| 909 |
+
"epoch": 2.317473338802297,
|
| 910 |
+
"grad_norm": 0.14032265543937683,
|
| 911 |
+
"learning_rate": 0.00014992287925270936,
|
| 912 |
+
"loss": 0.1698,
|
| 913 |
+
"step": 2825
|
| 914 |
+
},
|
| 915 |
+
{
|
| 916 |
+
"epoch": 2.3379819524200163,
|
| 917 |
+
"grad_norm": 0.14994294941425323,
|
| 918 |
+
"learning_rate": 0.00014150365858812353,
|
| 919 |
+
"loss": 0.1732,
|
| 920 |
+
"step": 2850
|
| 921 |
+
},
|
| 922 |
+
{
|
| 923 |
+
"epoch": 2.358490566037736,
|
| 924 |
+
"grad_norm": 0.12427656352519989,
|
| 925 |
+
"learning_rate": 0.00013328860612996053,
|
| 926 |
+
"loss": 0.1742,
|
| 927 |
+
"step": 2875
|
| 928 |
+
},
|
| 929 |
+
{
|
| 930 |
+
"epoch": 2.3789991796554553,
|
| 931 |
+
"grad_norm": 0.14301162958145142,
|
| 932 |
+
"learning_rate": 0.00012528240045463047,
|
| 933 |
+
"loss": 0.1777,
|
| 934 |
+
"step": 2900
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 2.399507793273175,
|
| 938 |
+
"grad_norm": 0.16245818138122559,
|
| 939 |
+
"learning_rate": 0.00011748960119765595,
|
| 940 |
+
"loss": 0.1749,
|
| 941 |
+
"step": 2925
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 2.4200164068908943,
|
| 945 |
+
"grad_norm": 0.14960838854312897,
|
| 946 |
+
"learning_rate": 0.00010991464645690142,
|
| 947 |
+
"loss": 0.1779,
|
| 948 |
+
"step": 2950
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 2.4405250205086135,
|
| 952 |
+
"grad_norm": 0.13634172081947327,
|
| 953 |
+
"learning_rate": 0.00010256185026502096,
|
| 954 |
+
"loss": 0.1753,
|
| 955 |
+
"step": 2975
|
| 956 |
+
},
|
| 957 |
+
{
|
| 958 |
+
"epoch": 2.4610336341263332,
|
| 959 |
+
"grad_norm": 0.14704327285289764,
|
| 960 |
+
"learning_rate": 9.543540013256136e-05,
|
| 961 |
+
"loss": 0.17,
|
| 962 |
+
"step": 3000
|
| 963 |
+
},
|
| 964 |
+
{
|
| 965 |
+
"epoch": 2.4610336341263332,
|
| 966 |
+
"eval_loss": 0.2040461301803589,
|
| 967 |
+
"eval_runtime": 19.3166,
|
| 968 |
+
"eval_samples_per_second": 51.769,
|
| 969 |
+
"eval_steps_per_second": 0.828,
|
| 970 |
+
"step": 3000
|
| 971 |
+
},
|
| 972 |
+
{
|
| 973 |
+
"epoch": 2.4815422477440525,
|
| 974 |
+
"grad_norm": 0.14649824798107147,
|
| 975 |
+
"learning_rate": 8.853935466312224e-05,
|
| 976 |
+
"loss": 0.1734,
|
| 977 |
+
"step": 3025
|
| 978 |
+
},
|
| 979 |
+
{
|
| 980 |
+
"epoch": 2.502050861361772,
|
| 981 |
+
"grad_norm": 0.13683316111564636,
|
| 982 |
+
"learning_rate": 8.187764124192981e-05,
|
| 983 |
+
"loss": 0.1696,
|
| 984 |
+
"step": 3050
|
| 985 |
+
},
|
| 986 |
+
{
|
| 987 |
+
"epoch": 2.5225594749794915,
|
| 988 |
+
"grad_norm": 0.14455805718898773,
|
| 989 |
+
"learning_rate": 7.545405379914138e-05,
|
| 990 |
+
"loss": 0.166,
|
| 991 |
+
"step": 3075
|
| 992 |
+
},
|
| 993 |
+
{
|
| 994 |
+
"epoch": 2.543068088597211,
|
| 995 |
+
"grad_norm": 0.14557591080665588,
|
| 996 |
+
"learning_rate": 6.927225064915349e-05,
|
| 997 |
+
"loss": 0.1774,
|
| 998 |
+
"step": 3100
|
| 999 |
+
},
|
| 1000 |
+
{
|
| 1001 |
+
"epoch": 2.56357670221493,
|
| 1002 |
+
"grad_norm": 0.15906178951263428,
|
| 1003 |
+
"learning_rate": 6.333575240714656e-05,
|
| 1004 |
+
"loss": 0.1663,
|
| 1005 |
+
"step": 3125
|
| 1006 |
+
},
|
| 1007 |
+
{
|
| 1008 |
+
"epoch": 2.5840853158326498,
|
| 1009 |
+
"grad_norm": 0.15751130878925323,
|
| 1010 |
+
"learning_rate": 5.764793998404877e-05,
|
| 1011 |
+
"loss": 0.1686,
|
| 1012 |
+
"step": 3150
|
| 1013 |
+
},
|
| 1014 |
+
{
|
| 1015 |
+
"epoch": 2.604593929450369,
|
| 1016 |
+
"grad_norm": 0.13006390631198883,
|
| 1017 |
+
"learning_rate": 5.2212052661065654e-05,
|
| 1018 |
+
"loss": 0.1712,
|
| 1019 |
+
"step": 3175
|
| 1020 |
+
},
|
| 1021 |
+
{
|
| 1022 |
+
"epoch": 2.6251025430680883,
|
| 1023 |
+
"grad_norm": 0.13611075282096863,
|
| 1024 |
+
"learning_rate": 4.703118624486735e-05,
|
| 1025 |
+
"loss": 0.1759,
|
| 1026 |
+
"step": 3200
|
| 1027 |
+
},
|
| 1028 |
+
{
|
| 1029 |
+
"epoch": 2.6251025430680883,
|
| 1030 |
+
"eval_loss": 0.20307449996471405,
|
| 1031 |
+
"eval_runtime": 19.2946,
|
| 1032 |
+
"eval_samples_per_second": 51.828,
|
| 1033 |
+
"eval_steps_per_second": 0.829,
|
| 1034 |
+
"step": 3200
|
| 1035 |
+
},
|
| 1036 |
+
{
|
| 1037 |
+
"epoch": 2.645611156685808,
|
| 1038 |
+
"grad_norm": 0.16608655452728271,
|
| 1039 |
+
"learning_rate": 4.210829130448834e-05,
|
| 1040 |
+
"loss": 0.1833,
|
| 1041 |
+
"step": 3225
|
| 1042 |
+
},
|
| 1043 |
+
{
|
| 1044 |
+
"epoch": 2.6661197703035278,
|
| 1045 |
+
"grad_norm": 0.15309980511665344,
|
| 1046 |
+
"learning_rate": 3.7446171490940704e-05,
|
| 1047 |
+
"loss": 0.1683,
|
| 1048 |
+
"step": 3250
|
| 1049 |
+
},
|
| 1050 |
+
{
|
| 1051 |
+
"epoch": 2.686628383921247,
|
| 1052 |
+
"grad_norm": 0.15862932801246643,
|
| 1053 |
+
"learning_rate": 3.3047481940499055e-05,
|
| 1054 |
+
"loss": 0.1747,
|
| 1055 |
+
"step": 3275
|
| 1056 |
+
},
|
| 1057 |
+
{
|
| 1058 |
+
"epoch": 2.7071369975389663,
|
| 1059 |
+
"grad_norm": 0.1526409238576889,
|
| 1060 |
+
"learning_rate": 2.8914727762567482e-05,
|
| 1061 |
+
"loss": 0.1675,
|
| 1062 |
+
"step": 3300
|
| 1063 |
+
},
|
| 1064 |
+
{
|
| 1065 |
+
"epoch": 2.727645611156686,
|
| 1066 |
+
"grad_norm": 0.145959734916687,
|
| 1067 |
+
"learning_rate": 2.5050262612987206e-05,
|
| 1068 |
+
"loss": 0.1674,
|
| 1069 |
+
"step": 3325
|
| 1070 |
+
},
|
| 1071 |
+
{
|
| 1072 |
+
"epoch": 2.7481542247744053,
|
| 1073 |
+
"grad_norm": 0.14206132292747498,
|
| 1074 |
+
"learning_rate": 2.14562873536005e-05,
|
| 1075 |
+
"loss": 0.1637,
|
| 1076 |
+
"step": 3350
|
| 1077 |
+
},
|
| 1078 |
+
{
|
| 1079 |
+
"epoch": 2.7686628383921246,
|
| 1080 |
+
"grad_norm": 0.18054018914699554,
|
| 1081 |
+
"learning_rate": 1.8134848798831537e-05,
|
| 1082 |
+
"loss": 0.1799,
|
| 1083 |
+
"step": 3375
|
| 1084 |
+
},
|
| 1085 |
+
{
|
| 1086 |
+
"epoch": 2.7891714520098443,
|
| 1087 |
+
"grad_norm": 0.13794836401939392,
|
| 1088 |
+
"learning_rate": 1.5087838549999955e-05,
|
| 1089 |
+
"loss": 0.168,
|
| 1090 |
+
"step": 3400
|
| 1091 |
+
},
|
| 1092 |
+
{
|
| 1093 |
+
"epoch": 2.7891714520098443,
|
| 1094 |
+
"eval_loss": 0.2025303989648819,
|
| 1095 |
+
"eval_runtime": 19.3132,
|
| 1096 |
+
"eval_samples_per_second": 51.778,
|
| 1097 |
+
"eval_steps_per_second": 0.828,
|
| 1098 |
+
"step": 3400
|
| 1099 |
+
},
|
| 1100 |
+
{
|
| 1101 |
+
"epoch": 2.8096800656275636,
|
| 1102 |
+
"grad_norm": 0.13373124599456787,
|
| 1103 |
+
"learning_rate": 1.2316991918029653e-05,
|
| 1104 |
+
"loss": 0.1759,
|
| 1105 |
+
"step": 3425
|
| 1106 |
+
},
|
| 1107 |
+
{
|
| 1108 |
+
"epoch": 2.830188679245283,
|
| 1109 |
+
"grad_norm": 0.1434241682291031,
|
| 1110 |
+
"learning_rate": 9.823886935167502e-06,
|
| 1111 |
+
"loss": 0.1679,
|
| 1112 |
+
"step": 3450
|
| 1113 |
+
},
|
| 1114 |
+
{
|
| 1115 |
+
"epoch": 2.8506972928630026,
|
| 1116 |
+
"grad_norm": 0.1458105742931366,
|
| 1117 |
+
"learning_rate": 7.60994345627386e-06,
|
| 1118 |
+
"loss": 0.1742,
|
| 1119 |
+
"step": 3475
|
| 1120 |
+
},
|
| 1121 |
+
{
|
| 1122 |
+
"epoch": 2.871205906480722,
|
| 1123 |
+
"grad_norm": 0.1493147611618042,
|
| 1124 |
+
"learning_rate": 5.67642235019733e-06,
|
| 1125 |
+
"loss": 0.1654,
|
| 1126 |
+
"step": 3500
|
| 1127 |
+
},
|
| 1128 |
+
{
|
| 1129 |
+
"epoch": 2.891714520098441,
|
| 1130 |
+
"grad_norm": 0.15342405438423157,
|
| 1131 |
+
"learning_rate": 4.024424781693792e-06,
|
| 1132 |
+
"loss": 0.1685,
|
| 1133 |
+
"step": 3525
|
| 1134 |
+
},
|
| 1135 |
+
{
|
| 1136 |
+
"epoch": 2.912223133716161,
|
| 1137 |
+
"grad_norm": 0.15680305659770966,
|
| 1138 |
+
"learning_rate": 2.6548915842993793e-06,
|
| 1139 |
+
"loss": 0.1684,
|
| 1140 |
+
"step": 3550
|
| 1141 |
+
},
|
| 1142 |
+
{
|
| 1143 |
+
"epoch": 2.93273174733388,
|
| 1144 |
+
"grad_norm": 0.1487962007522583,
|
| 1145 |
+
"learning_rate": 1.5686027245138123e-06,
|
| 1146 |
+
"loss": 0.167,
|
| 1147 |
+
"step": 3575
|
| 1148 |
+
},
|
| 1149 |
+
{
|
| 1150 |
+
"epoch": 2.9532403609516,
|
| 1151 |
+
"grad_norm": 0.13475292921066284,
|
| 1152 |
+
"learning_rate": 7.661768575994432e-07,
|
| 1153 |
+
"loss": 0.1733,
|
| 1154 |
+
"step": 3600
|
| 1155 |
+
},
|
| 1156 |
+
{
|
| 1157 |
+
"epoch": 2.9532403609516,
|
| 1158 |
+
"eval_loss": 0.2021828442811966,
|
| 1159 |
+
"eval_runtime": 19.3189,
|
| 1160 |
+
"eval_samples_per_second": 51.763,
|
| 1161 |
+
"eval_steps_per_second": 0.828,
|
| 1162 |
+
"step": 3600
|
| 1163 |
+
},
|
| 1164 |
+
{
|
| 1165 |
+
"epoch": 2.973748974569319,
|
| 1166 |
+
"grad_norm": 0.12766960263252258,
|
| 1167 |
+
"learning_rate": 2.480709752493704e-07,
|
| 1168 |
+
"loss": 0.1678,
|
| 1169 |
+
"step": 3625
|
| 1170 |
+
},
|
| 1171 |
+
{
|
| 1172 |
+
"epoch": 2.994257588187039,
|
| 1173 |
+
"grad_norm": 0.14542804658412933,
|
| 1174 |
+
"learning_rate": 1.4580145324849525e-08,
|
| 1175 |
+
"loss": 0.1669,
|
| 1176 |
+
"step": 3650
|
| 1177 |
+
},
|
| 1178 |
+
{
|
| 1179 |
+
"epoch": 3.0,
|
| 1180 |
+
"step": 3657,
|
| 1181 |
+
"total_flos": 2.37785441107968e+18,
|
| 1182 |
+
"train_loss": 0.2320223105941472,
|
| 1183 |
+
"train_runtime": 3891.0373,
|
| 1184 |
+
"train_samples_per_second": 30.069,
|
| 1185 |
+
"train_steps_per_second": 0.94
|
| 1186 |
+
}
|
| 1187 |
+
],
|
| 1188 |
+
"logging_steps": 25,
|
| 1189 |
+
"max_steps": 3657,
|
| 1190 |
+
"num_input_tokens_seen": 0,
|
| 1191 |
+
"num_train_epochs": 3,
|
| 1192 |
+
"save_steps": 0,
|
| 1193 |
+
"stateful_callbacks": {
|
| 1194 |
+
"TrainerControl": {
|
| 1195 |
+
"args": {
|
| 1196 |
+
"should_epoch_stop": false,
|
| 1197 |
+
"should_evaluate": false,
|
| 1198 |
+
"should_log": false,
|
| 1199 |
+
"should_save": true,
|
| 1200 |
+
"should_training_stop": true
|
| 1201 |
+
},
|
| 1202 |
+
"attributes": {}
|
| 1203 |
+
}
|
| 1204 |
+
},
|
| 1205 |
+
"total_flos": 2.37785441107968e+18,
|
| 1206 |
+
"train_batch_size": 32,
|
| 1207 |
+
"trial_name": null,
|
| 1208 |
+
"trial_params": null
|
| 1209 |
+
}
|