Batch upload part 3
Browse files- nl_tasks/exps/run_ex01/ft/adapter_config.json +18 -0
- nl_tasks/exps/run_ex01/ft/added_tokens.json +3 -0
- nl_tasks/exps/run_ex01/ft/special_tokens_map.json +30 -0
- nl_tasks/exps/run_ex01/ft/tokenizer.json +0 -0
- nl_tasks/exps/run_ex01/ft/tokenizer.model +3 -0
- nl_tasks/exps/run_ex01/ft/tokenizer_config.json +51 -0
- nl_tasks/exps/run_ex01/ft/training_args.bin +3 -0
- nl_tasks/exps/run_ex01/ft2/adapter_config.json +18 -0
- nl_tasks/exps/run_ex01/ft2/adapter_model.bin +3 -0
- nl_tasks/exps/run_ex03/ft/adapter_config.json +18 -0
- nl_tasks/exps/run_ex03/ft/added_tokens.json +3 -0
- nl_tasks/exps/run_ex03/ft/special_tokens_map.json +30 -0
- nl_tasks/exps/run_ex03/ft/tokenizer.json +0 -0
- nl_tasks/exps/run_ex03/ft/tokenizer.model +3 -0
- nl_tasks/exps/run_ex03/ft/tokenizer_config.json +51 -0
- nl_tasks/exps/run_ex03/ft2/adapter_config.json +18 -0
- nl_tasks/exps/run_ex03/ft2/adapter_model.bin +3 -0
- nl_tasks/exps/run_ex03/trainer_state.json +914 -0
- nl_tasks/exps/run_ex04/ft/adapter_config.json +18 -0
- nl_tasks/exps/run_ex04/ft/added_tokens.json +3 -0
- nl_tasks/exps/run_ex04/ft/special_tokens_map.json +30 -0
- nl_tasks/exps/run_ex04/ft/tokenizer.json +0 -0
- nl_tasks/exps/run_ex04/ft/tokenizer.model +3 -0
- nl_tasks/exps/run_ex04/ft/tokenizer_config.json +51 -0
- nl_tasks/exps/run_ex04/ft2/adapter_config.json +18 -0
- nl_tasks/exps/run_ex04/ft2/adapter_model.bin +3 -0
- nl_tasks/exps/run_ex04/trainer_state.json +914 -0
- nl_tasks/exps/run_ex16/ft/adapter_config.json +18 -0
- nl_tasks/exps/run_ex16/ft/special_tokens_map.json +24 -0
- nl_tasks/exps/run_ex16/ft/tokenizer.json +0 -0
- nl_tasks/exps/run_ex16/ft/tokenizer.model +3 -0
- nl_tasks/exps/run_ex16/ft/tokenizer_config.json +43 -0
- nl_tasks/exps/run_ex16/ft2/adapter_config.json +18 -0
- nl_tasks/exps/run_ex16/ft2/adapter_model.bin +3 -0
- nl_tasks/exps/run_ex16/trainer_state.json +783 -0
- nl_tasks/exps/run_ex17/ft/adapter_config.json +18 -0
- nl_tasks/exps/run_ex17/ft/special_tokens_map.json +24 -0
- nl_tasks/exps/run_ex17/ft/tokenizer.json +0 -0
- nl_tasks/exps/run_ex17/ft/tokenizer.model +3 -0
- nl_tasks/exps/run_ex17/ft/tokenizer_config.json +43 -0
- nl_tasks/exps/run_ex17/ft2/adapter_config.json +18 -0
- nl_tasks/exps/run_ex17/ft2/adapter_model.bin +3 -0
- nl_tasks/exps/run_ex17/trainer_state.json +783 -0
- nl_tasks/exps/run_ex18/ft/adapter_config.json +18 -0
- nl_tasks/exps/run_ex18/ft/special_tokens_map.json +24 -0
- nl_tasks/exps/run_ex18/ft/tokenizer.json +0 -0
- nl_tasks/exps/run_ex18/ft/tokenizer.model +3 -0
- nl_tasks/exps/run_ex18/ft/tokenizer_config.json +43 -0
- nl_tasks/exps/run_ex18/ft2/adapter_model.bin +3 -0
- nl_tasks/exps/run_ex18/trainer_state.json +1149 -0
nl_tasks/exps/run_ex01/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": 2,
|
| 9 |
+
"peft_type": "ROTATION",
|
| 10 |
+
"r": 8,
|
| 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_ex01/ft/added_tokens.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"[PAD]": 32000
|
| 3 |
+
}
|
nl_tasks/exps/run_ex01/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_ex01/ft/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
nl_tasks/exps/run_ex01/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_ex01/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_ex01/ft/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:438a99af8739bac7142583f612088b1caf8956d01054899d18b7f598f7f40a1e
|
| 3 |
+
size 6545
|
nl_tasks/exps/run_ex01/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": 2,
|
| 9 |
+
"peft_type": "ROTATION",
|
| 10 |
+
"r": 8,
|
| 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_ex01/ft2/adapter_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ace64702b654bbcccd8d66456649ec0a62321f993e1bf61471451756185b8f39
|
| 3 |
+
size 33602659
|
nl_tasks/exps/run_ex03/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": 2,
|
| 9 |
+
"peft_type": "ROTATION",
|
| 10 |
+
"r": 8,
|
| 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_ex03/ft/added_tokens.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"[PAD]": 32000
|
| 3 |
+
}
|
nl_tasks/exps/run_ex03/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_ex03/ft/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
nl_tasks/exps/run_ex03/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_ex03/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_ex03/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": 2,
|
| 9 |
+
"peft_type": "ROTATION",
|
| 10 |
+
"r": 8,
|
| 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_ex03/ft2/adapter_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e1040c5e94f4d67faa5e11f1c00b64969064451d38f6e1a836fe3b4154048ed6
|
| 3 |
+
size 33602659
|
nl_tasks/exps/run_ex03/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": 0.2224287986755371,
|
| 15 |
+
"learning_rate": 9.836065573770491e-05,
|
| 16 |
+
"loss": 0.68,
|
| 17 |
+
"step": 25
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.04101722723543889,
|
| 21 |
+
"grad_norm": 0.20727577805519104,
|
| 22 |
+
"learning_rate": 0.00020081967213114754,
|
| 23 |
+
"loss": 0.4173,
|
| 24 |
+
"step": 50
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.06152584085315833,
|
| 28 |
+
"grad_norm": 0.14471279084682465,
|
| 29 |
+
"learning_rate": 0.0003032786885245902,
|
| 30 |
+
"loss": 0.3742,
|
| 31 |
+
"step": 75
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.08203445447087777,
|
| 35 |
+
"grad_norm": 0.1944092959165573,
|
| 36 |
+
"learning_rate": 0.0004057377049180328,
|
| 37 |
+
"loss": 0.337,
|
| 38 |
+
"step": 100
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.08203445447087777,
|
| 42 |
+
"eval_loss": 0.32692742347717285,
|
| 43 |
+
"eval_runtime": 19.6173,
|
| 44 |
+
"eval_samples_per_second": 50.975,
|
| 45 |
+
"eval_steps_per_second": 0.816,
|
| 46 |
+
"step": 100
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"epoch": 0.10254306808859721,
|
| 50 |
+
"grad_norm": 0.16892839968204498,
|
| 51 |
+
"learning_rate": 0.0005081967213114754,
|
| 52 |
+
"loss": 0.3414,
|
| 53 |
+
"step": 125
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 0.12305168170631665,
|
| 57 |
+
"grad_norm": 0.17221364378929138,
|
| 58 |
+
"learning_rate": 0.000610655737704918,
|
| 59 |
+
"loss": 0.3233,
|
| 60 |
+
"step": 150
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 0.1435602953240361,
|
| 64 |
+
"grad_norm": 0.1988595426082611,
|
| 65 |
+
"learning_rate": 0.0007131147540983607,
|
| 66 |
+
"loss": 0.3142,
|
| 67 |
+
"step": 175
|
| 68 |
+
},
|
| 69 |
+
{
|
| 70 |
+
"epoch": 0.16406890894175555,
|
| 71 |
+
"grad_norm": 0.22134506702423096,
|
| 72 |
+
"learning_rate": 0.0008155737704918033,
|
| 73 |
+
"loss": 0.3261,
|
| 74 |
+
"step": 200
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.16406890894175555,
|
| 78 |
+
"eval_loss": 0.31184008717536926,
|
| 79 |
+
"eval_runtime": 19.48,
|
| 80 |
+
"eval_samples_per_second": 51.335,
|
| 81 |
+
"eval_steps_per_second": 0.821,
|
| 82 |
+
"step": 200
|
| 83 |
+
},
|
| 84 |
+
{
|
| 85 |
+
"epoch": 0.184577522559475,
|
| 86 |
+
"grad_norm": 0.21120023727416992,
|
| 87 |
+
"learning_rate": 0.0009180327868852459,
|
| 88 |
+
"loss": 0.3234,
|
| 89 |
+
"step": 225
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 0.20508613617719443,
|
| 93 |
+
"grad_norm": 0.29418930411338806,
|
| 94 |
+
"learning_rate": 0.0009999871854116063,
|
| 95 |
+
"loss": 0.3055,
|
| 96 |
+
"step": 250
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"epoch": 0.22559474979491387,
|
| 100 |
+
"grad_norm": 0.27492791414260864,
|
| 101 |
+
"learning_rate": 0.0009995387437838027,
|
| 102 |
+
"loss": 0.3066,
|
| 103 |
+
"step": 275
|
| 104 |
+
},
|
| 105 |
+
{
|
| 106 |
+
"epoch": 0.2461033634126333,
|
| 107 |
+
"grad_norm": 1.5484436750411987,
|
| 108 |
+
"learning_rate": 0.000998450229439693,
|
| 109 |
+
"loss": 0.3111,
|
| 110 |
+
"step": 300
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.2461033634126333,
|
| 114 |
+
"eval_loss": 0.3055231273174286,
|
| 115 |
+
"eval_runtime": 19.4782,
|
| 116 |
+
"eval_samples_per_second": 51.339,
|
| 117 |
+
"eval_steps_per_second": 0.821,
|
| 118 |
+
"step": 300
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 0.2666119770303528,
|
| 122 |
+
"grad_norm": 0.37643224000930786,
|
| 123 |
+
"learning_rate": 0.000996723037122612,
|
| 124 |
+
"loss": 0.3486,
|
| 125 |
+
"step": 325
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 0.2871205906480722,
|
| 129 |
+
"grad_norm": 0.29540306329727173,
|
| 130 |
+
"learning_rate": 0.0009943593799315263,
|
| 131 |
+
"loss": 0.3161,
|
| 132 |
+
"step": 350
|
| 133 |
+
},
|
| 134 |
+
{
|
| 135 |
+
"epoch": 0.30762920426579166,
|
| 136 |
+
"grad_norm": 0.2830330729484558,
|
| 137 |
+
"learning_rate": 0.0009913622864853324,
|
| 138 |
+
"loss": 0.3189,
|
| 139 |
+
"step": 375
|
| 140 |
+
},
|
| 141 |
+
{
|
| 142 |
+
"epoch": 0.3281378178835111,
|
| 143 |
+
"grad_norm": 0.28148511052131653,
|
| 144 |
+
"learning_rate": 0.0009877355970422024,
|
| 145 |
+
"loss": 0.2974,
|
| 146 |
+
"step": 400
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.3281378178835111,
|
| 150 |
+
"eval_loss": 0.2851388454437256,
|
| 151 |
+
"eval_runtime": 19.4841,
|
| 152 |
+
"eval_samples_per_second": 51.324,
|
| 153 |
+
"eval_steps_per_second": 0.821,
|
| 154 |
+
"step": 400
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 0.34864643150123054,
|
| 158 |
+
"grad_norm": 0.2346981018781662,
|
| 159 |
+
"learning_rate": 0.0009834839585789559,
|
| 160 |
+
"loss": 0.3053,
|
| 161 |
+
"step": 425
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 0.36915504511895,
|
| 165 |
+
"grad_norm": 0.26579293608665466,
|
| 166 |
+
"learning_rate": 0.000978612818836762,
|
| 167 |
+
"loss": 0.3007,
|
| 168 |
+
"step": 450
|
| 169 |
+
},
|
| 170 |
+
{
|
| 171 |
+
"epoch": 0.3896636587366694,
|
| 172 |
+
"grad_norm": 0.24696120619773865,
|
| 173 |
+
"learning_rate": 0.0009731284193407981,
|
| 174 |
+
"loss": 0.2895,
|
| 175 |
+
"step": 475
|
| 176 |
+
},
|
| 177 |
+
{
|
| 178 |
+
"epoch": 0.41017227235438886,
|
| 179 |
+
"grad_norm": 0.20220448076725006,
|
| 180 |
+
"learning_rate": 0.0009670377874028117,
|
| 181 |
+
"loss": 0.2733,
|
| 182 |
+
"step": 500
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.41017227235438886,
|
| 186 |
+
"eval_loss": 0.275390625,
|
| 187 |
+
"eval_runtime": 19.4742,
|
| 188 |
+
"eval_samples_per_second": 51.35,
|
| 189 |
+
"eval_steps_per_second": 0.822,
|
| 190 |
+
"step": 500
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 0.4306808859721083,
|
| 194 |
+
"grad_norm": 0.2370428740978241,
|
| 195 |
+
"learning_rate": 0.0009603487271168336,
|
| 196 |
+
"loss": 0.2739,
|
| 197 |
+
"step": 525
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 0.45118949958982774,
|
| 201 |
+
"grad_norm": 0.2254423201084137,
|
| 202 |
+
"learning_rate": 0.0009530698093595781,
|
| 203 |
+
"loss": 0.283,
|
| 204 |
+
"step": 550
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 0.4716981132075472,
|
| 208 |
+
"grad_norm": 0.2541421055793762,
|
| 209 |
+
"learning_rate": 0.0009452103608083418,
|
| 210 |
+
"loss": 0.2731,
|
| 211 |
+
"step": 575
|
| 212 |
+
},
|
| 213 |
+
{
|
| 214 |
+
"epoch": 0.4922067268252666,
|
| 215 |
+
"grad_norm": 0.22776468098163605,
|
| 216 |
+
"learning_rate": 0.0009367804519904775,
|
| 217 |
+
"loss": 0.2911,
|
| 218 |
+
"step": 600
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.4922067268252666,
|
| 222 |
+
"eval_loss": 0.2668289244174957,
|
| 223 |
+
"eval_runtime": 19.4831,
|
| 224 |
+
"eval_samples_per_second": 51.326,
|
| 225 |
+
"eval_steps_per_second": 0.821,
|
| 226 |
+
"step": 600
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.5127153404429861,
|
| 230 |
+
"grad_norm": 0.2263737916946411,
|
| 231 |
+
"learning_rate": 0.0009277908843797492,
|
| 232 |
+
"loss": 0.2704,
|
| 233 |
+
"step": 625
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.5332239540607056,
|
| 237 |
+
"grad_norm": 0.2699456214904785,
|
| 238 |
+
"learning_rate": 0.0009182531765561084,
|
| 239 |
+
"loss": 0.2622,
|
| 240 |
+
"step": 650
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.5537325676784249,
|
| 244 |
+
"grad_norm": 0.22075457870960236,
|
| 245 |
+
"learning_rate": 0.0009081795494466201,
|
| 246 |
+
"loss": 0.2736,
|
| 247 |
+
"step": 675
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.5742411812961444,
|
| 251 |
+
"grad_norm": 0.19850771129131317,
|
| 252 |
+
"learning_rate": 0.0008975829106664539,
|
| 253 |
+
"loss": 0.2711,
|
| 254 |
+
"step": 700
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.5742411812961444,
|
| 258 |
+
"eval_loss": 0.2581631541252136,
|
| 259 |
+
"eval_runtime": 19.4903,
|
| 260 |
+
"eval_samples_per_second": 51.307,
|
| 261 |
+
"eval_steps_per_second": 0.821,
|
| 262 |
+
"step": 700
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.5947497949138638,
|
| 266 |
+
"grad_norm": 0.19109679758548737,
|
| 267 |
+
"learning_rate": 0.0008864768379800017,
|
| 268 |
+
"loss": 0.2682,
|
| 269 |
+
"step": 725
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.6152584085315833,
|
| 273 |
+
"grad_norm": 0.20022593438625336,
|
| 274 |
+
"learning_rate": 0.0008748755619033153,
|
| 275 |
+
"loss": 0.2649,
|
| 276 |
+
"step": 750
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.6357670221493027,
|
| 280 |
+
"grad_norm": 0.2298661172389984,
|
| 281 |
+
"learning_rate": 0.000862793947470155,
|
| 282 |
+
"loss": 0.2629,
|
| 283 |
+
"step": 775
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.6562756357670222,
|
| 287 |
+
"grad_norm": 0.2014313042163849,
|
| 288 |
+
"learning_rate": 0.0008502474751850142,
|
| 289 |
+
"loss": 0.2648,
|
| 290 |
+
"step": 800
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.6562756357670222,
|
| 294 |
+
"eval_loss": 0.25314998626708984,
|
| 295 |
+
"eval_runtime": 19.4877,
|
| 296 |
+
"eval_samples_per_second": 51.315,
|
| 297 |
+
"eval_steps_per_second": 0.821,
|
| 298 |
+
"step": 800
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 0.6767842493847416,
|
| 302 |
+
"grad_norm": 0.17902116477489471,
|
| 303 |
+
"learning_rate": 0.0008372522211875224,
|
| 304 |
+
"loss": 0.2572,
|
| 305 |
+
"step": 825
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 0.6972928630024611,
|
| 309 |
+
"grad_norm": 0.19330015778541565,
|
| 310 |
+
"learning_rate": 0.0008238248366536473,
|
| 311 |
+
"loss": 0.2606,
|
| 312 |
+
"step": 850
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 0.7178014766201805,
|
| 316 |
+
"grad_norm": 0.15420861542224884,
|
| 317 |
+
"learning_rate": 0.0008099825264600842,
|
| 318 |
+
"loss": 0.2495,
|
| 319 |
+
"step": 875
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 0.7383100902379,
|
| 323 |
+
"grad_norm": 0.20329487323760986,
|
| 324 |
+
"learning_rate": 0.0007957430271391761,
|
| 325 |
+
"loss": 0.2534,
|
| 326 |
+
"step": 900
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 0.7383100902379,
|
| 330 |
+
"eval_loss": 0.2430313378572464,
|
| 331 |
+
"eval_runtime": 19.4636,
|
| 332 |
+
"eval_samples_per_second": 51.378,
|
| 333 |
+
"eval_steps_per_second": 0.822,
|
| 334 |
+
"step": 900
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"epoch": 0.7588187038556193,
|
| 338 |
+
"grad_norm": 0.22887705266475677,
|
| 339 |
+
"learning_rate": 0.0007811245841526062,
|
| 340 |
+
"loss": 0.2542,
|
| 341 |
+
"step": 925
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 0.7793273174733388,
|
| 345 |
+
"grad_norm": 0.19081521034240723,
|
| 346 |
+
"learning_rate": 0.0007661459285129879,
|
| 347 |
+
"loss": 0.2437,
|
| 348 |
+
"step": 950
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"epoch": 0.7998359310910582,
|
| 352 |
+
"grad_norm": 0.1877547800540924,
|
| 353 |
+
"learning_rate": 0.0007508262527833029,
|
| 354 |
+
"loss": 0.2499,
|
| 355 |
+
"step": 975
|
| 356 |
+
},
|
| 357 |
+
{
|
| 358 |
+
"epoch": 0.8203445447087777,
|
| 359 |
+
"grad_norm": 0.18896251916885376,
|
| 360 |
+
"learning_rate": 0.000735185186484943,
|
| 361 |
+
"loss": 0.2422,
|
| 362 |
+
"step": 1000
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.8203445447087777,
|
| 366 |
+
"eval_loss": 0.23874206840991974,
|
| 367 |
+
"eval_runtime": 19.462,
|
| 368 |
+
"eval_samples_per_second": 51.382,
|
| 369 |
+
"eval_steps_per_second": 0.822,
|
| 370 |
+
"step": 1000
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 0.8408531583264971,
|
| 374 |
+
"grad_norm": 0.1943541318178177,
|
| 375 |
+
"learning_rate": 0.0007192427709458656,
|
| 376 |
+
"loss": 0.2422,
|
| 377 |
+
"step": 1025
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 0.8613617719442166,
|
| 381 |
+
"grad_norm": 0.1920219510793686,
|
| 382 |
+
"learning_rate": 0.0007030194336210887,
|
| 383 |
+
"loss": 0.2395,
|
| 384 |
+
"step": 1050
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.881870385561936,
|
| 388 |
+
"grad_norm": 0.18021953105926514,
|
| 389 |
+
"learning_rate": 0.0006865359619184331,
|
| 390 |
+
"loss": 0.245,
|
| 391 |
+
"step": 1075
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 0.9023789991796555,
|
| 395 |
+
"grad_norm": 0.1822519749403,
|
| 396 |
+
"learning_rate": 0.0006698134765630434,
|
| 397 |
+
"loss": 0.2433,
|
| 398 |
+
"step": 1100
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 0.9023789991796555,
|
| 402 |
+
"eval_loss": 0.2355821281671524,
|
| 403 |
+
"eval_runtime": 19.4606,
|
| 404 |
+
"eval_samples_per_second": 51.386,
|
| 405 |
+
"eval_steps_per_second": 0.822,
|
| 406 |
+
"step": 1100
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 0.9228876127973749,
|
| 410 |
+
"grad_norm": 0.17558462917804718,
|
| 411 |
+
"learning_rate": 0.0006528734045348248,
|
| 412 |
+
"loss": 0.239,
|
| 413 |
+
"step": 1125
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 0.9433962264150944,
|
| 417 |
+
"grad_norm": 0.1492016613483429,
|
| 418 |
+
"learning_rate": 0.0006357374516134643,
|
| 419 |
+
"loss": 0.2382,
|
| 420 |
+
"step": 1150
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 0.9639048400328137,
|
| 424 |
+
"grad_norm": 0.17199264466762543,
|
| 425 |
+
"learning_rate": 0.0006184275745662179,
|
| 426 |
+
"loss": 0.2441,
|
| 427 |
+
"step": 1175
|
| 428 |
+
},
|
| 429 |
+
{
|
| 430 |
+
"epoch": 0.9844134536505332,
|
| 431 |
+
"grad_norm": 0.17206618189811707,
|
| 432 |
+
"learning_rate": 0.0006009659530141031,
|
| 433 |
+
"loss": 0.2491,
|
| 434 |
+
"step": 1200
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 0.9844134536505332,
|
| 438 |
+
"eval_loss": 0.2298414260149002,
|
| 439 |
+
"eval_runtime": 19.4633,
|
| 440 |
+
"eval_samples_per_second": 51.379,
|
| 441 |
+
"eval_steps_per_second": 0.822,
|
| 442 |
+
"step": 1200
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 1.0049220672682526,
|
| 446 |
+
"grad_norm": 0.17285768687725067,
|
| 447 |
+
"learning_rate": 0.0005833749610125402,
|
| 448 |
+
"loss": 0.2294,
|
| 449 |
+
"step": 1225
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 1.0254306808859721,
|
| 453 |
+
"grad_norm": 0.1635085940361023,
|
| 454 |
+
"learning_rate": 0.0005656771383828602,
|
| 455 |
+
"loss": 0.2082,
|
| 456 |
+
"step": 1250
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 1.0459392945036916,
|
| 460 |
+
"grad_norm": 0.17534485459327698,
|
| 461 |
+
"learning_rate": 0.0005478951618314134,
|
| 462 |
+
"loss": 0.2099,
|
| 463 |
+
"step": 1275
|
| 464 |
+
},
|
| 465 |
+
{
|
| 466 |
+
"epoch": 1.066447908121411,
|
| 467 |
+
"grad_norm": 0.1634574681520462,
|
| 468 |
+
"learning_rate": 0.0005300518158932815,
|
| 469 |
+
"loss": 0.2124,
|
| 470 |
+
"step": 1300
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 1.066447908121411,
|
| 474 |
+
"eval_loss": 0.2263764590024948,
|
| 475 |
+
"eval_runtime": 19.4607,
|
| 476 |
+
"eval_samples_per_second": 51.386,
|
| 477 |
+
"eval_steps_per_second": 0.822,
|
| 478 |
+
"step": 1300
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 1.0869565217391304,
|
| 482 |
+
"grad_norm": 0.20147192478179932,
|
| 483 |
+
"learning_rate": 0.0005121699637378282,
|
| 484 |
+
"loss": 0.2079,
|
| 485 |
+
"step": 1325
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 1.1074651353568499,
|
| 489 |
+
"grad_norm": 0.1501941829919815,
|
| 490 |
+
"learning_rate": 0.0004942725178734903,
|
| 491 |
+
"loss": 0.2023,
|
| 492 |
+
"step": 1350
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 1.1279737489745694,
|
| 496 |
+
"grad_norm": 0.1833459585905075,
|
| 497 |
+
"learning_rate": 0.00047638241078935324,
|
| 498 |
+
"loss": 0.2037,
|
| 499 |
+
"step": 1375
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 1.1484823625922886,
|
| 503 |
+
"grad_norm": 0.17751918733119965,
|
| 504 |
+
"learning_rate": 0.000458522565571121,
|
| 505 |
+
"loss": 0.1977,
|
| 506 |
+
"step": 1400
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 1.1484823625922886,
|
| 510 |
+
"eval_loss": 0.22482682764530182,
|
| 511 |
+
"eval_runtime": 19.468,
|
| 512 |
+
"eval_samples_per_second": 51.366,
|
| 513 |
+
"eval_steps_per_second": 0.822,
|
| 514 |
+
"step": 1400
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 1.1689909762100081,
|
| 518 |
+
"grad_norm": 0.1845618039369583,
|
| 519 |
+
"learning_rate": 0.00044071586652913767,
|
| 520 |
+
"loss": 0.2075,
|
| 521 |
+
"step": 1425
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 1.1894995898277276,
|
| 525 |
+
"grad_norm": 0.19091343879699707,
|
| 526 |
+
"learning_rate": 0.0004229851298760915,
|
| 527 |
+
"loss": 0.2077,
|
| 528 |
+
"step": 1450
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 1.2100082034454471,
|
| 532 |
+
"grad_norm": 0.17038100957870483,
|
| 533 |
+
"learning_rate": 0.0004053530744919749,
|
| 534 |
+
"loss": 0.2051,
|
| 535 |
+
"step": 1475
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 1.2305168170631666,
|
| 539 |
+
"grad_norm": 0.18054074048995972,
|
| 540 |
+
"learning_rate": 0.0003878422928137597,
|
| 541 |
+
"loss": 0.213,
|
| 542 |
+
"step": 1500
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 1.2305168170631666,
|
| 546 |
+
"eval_loss": 0.22022739052772522,
|
| 547 |
+
"eval_runtime": 19.5059,
|
| 548 |
+
"eval_samples_per_second": 51.267,
|
| 549 |
+
"eval_steps_per_second": 0.82,
|
| 550 |
+
"step": 1500
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 1.251025430680886,
|
| 554 |
+
"grad_norm": 0.19870440661907196,
|
| 555 |
+
"learning_rate": 0.0003704752218870861,
|
| 556 |
+
"loss": 0.2084,
|
| 557 |
+
"step": 1525
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 1.2715340442986054,
|
| 561 |
+
"grad_norm": 0.15106436610221863,
|
| 562 |
+
"learning_rate": 0.00035327411461706025,
|
| 563 |
+
"loss": 0.2018,
|
| 564 |
+
"step": 1550
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 1.2920426579163249,
|
| 568 |
+
"grad_norm": 0.14452490210533142,
|
| 569 |
+
"learning_rate": 0.00033626101125499555,
|
| 570 |
+
"loss": 0.2028,
|
| 571 |
+
"step": 1575
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 1.3125512715340442,
|
| 575 |
+
"grad_norm": 0.15930652618408203,
|
| 576 |
+
"learning_rate": 0.0003194577111576333,
|
| 577 |
+
"loss": 0.1959,
|
| 578 |
+
"step": 1600
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 1.3125512715340442,
|
| 582 |
+
"eval_loss": 0.21710796654224396,
|
| 583 |
+
"eval_runtime": 19.4771,
|
| 584 |
+
"eval_samples_per_second": 51.342,
|
| 585 |
+
"eval_steps_per_second": 0.821,
|
| 586 |
+
"step": 1600
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 1.3330598851517639,
|
| 590 |
+
"grad_norm": 0.16485872864723206,
|
| 591 |
+
"learning_rate": 0.00030288574485502756,
|
| 592 |
+
"loss": 0.1997,
|
| 593 |
+
"step": 1625
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 1.3535684987694832,
|
| 597 |
+
"grad_norm": 0.17522995173931122,
|
| 598 |
+
"learning_rate": 0.00028656634646288565,
|
| 599 |
+
"loss": 0.1996,
|
| 600 |
+
"step": 1650
|
| 601 |
+
},
|
| 602 |
+
{
|
| 603 |
+
"epoch": 1.3740771123872026,
|
| 604 |
+
"grad_norm": 0.1736440658569336,
|
| 605 |
+
"learning_rate": 0.00027052042647471254,
|
| 606 |
+
"loss": 0.1963,
|
| 607 |
+
"step": 1675
|
| 608 |
+
},
|
| 609 |
+
{
|
| 610 |
+
"epoch": 1.3945857260049221,
|
| 611 |
+
"grad_norm": 0.21140924096107483,
|
| 612 |
+
"learning_rate": 0.0002547685449686206,
|
| 613 |
+
"loss": 0.1975,
|
| 614 |
+
"step": 1700
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 1.3945857260049221,
|
| 618 |
+
"eval_loss": 0.2152220606803894,
|
| 619 |
+
"eval_runtime": 19.4995,
|
| 620 |
+
"eval_samples_per_second": 51.283,
|
| 621 |
+
"eval_steps_per_second": 0.821,
|
| 622 |
+
"step": 1700
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"epoch": 1.4150943396226414,
|
| 626 |
+
"grad_norm": 0.1721457540988922,
|
| 627 |
+
"learning_rate": 0.0002393308852631373,
|
| 628 |
+
"loss": 0.1946,
|
| 629 |
+
"step": 1725
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 1.435602953240361,
|
| 633 |
+
"grad_norm": 0.169590562582016,
|
| 634 |
+
"learning_rate": 0.0002242272280557645,
|
| 635 |
+
"loss": 0.193,
|
| 636 |
+
"step": 1750
|
| 637 |
+
},
|
| 638 |
+
{
|
| 639 |
+
"epoch": 1.4561115668580804,
|
| 640 |
+
"grad_norm": 0.16912959516048431,
|
| 641 |
+
"learning_rate": 0.0002094769260774262,
|
| 642 |
+
"loss": 0.2004,
|
| 643 |
+
"step": 1775
|
| 644 |
+
},
|
| 645 |
+
{
|
| 646 |
+
"epoch": 1.4766201804758,
|
| 647 |
+
"grad_norm": 0.14390693604946136,
|
| 648 |
+
"learning_rate": 0.00019509887929528458,
|
| 649 |
+
"loss": 0.1893,
|
| 650 |
+
"step": 1800
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 1.4766201804758,
|
| 654 |
+
"eval_loss": 0.21334853768348694,
|
| 655 |
+
"eval_runtime": 19.483,
|
| 656 |
+
"eval_samples_per_second": 51.327,
|
| 657 |
+
"eval_steps_per_second": 0.821,
|
| 658 |
+
"step": 1800
|
| 659 |
+
},
|
| 660 |
+
{
|
| 661 |
+
"epoch": 1.4971287940935194,
|
| 662 |
+
"grad_norm": 0.15655267238616943,
|
| 663 |
+
"learning_rate": 0.0001811115106956918,
|
| 664 |
+
"loss": 0.1897,
|
| 665 |
+
"step": 1825
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 1.5176374077112387,
|
| 669 |
+
"grad_norm": 0.14342047274112701,
|
| 670 |
+
"learning_rate": 0.00016753274267831115,
|
| 671 |
+
"loss": 0.192,
|
| 672 |
+
"step": 1850
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 1.5381460213289582,
|
| 676 |
+
"grad_norm": 0.1619173139333725,
|
| 677 |
+
"learning_rate": 0.00015437997409165478,
|
| 678 |
+
"loss": 0.1929,
|
| 679 |
+
"step": 1875
|
| 680 |
+
},
|
| 681 |
+
{
|
| 682 |
+
"epoch": 1.5586546349466777,
|
| 683 |
+
"grad_norm": 0.17437753081321716,
|
| 684 |
+
"learning_rate": 0.00014167005793946035,
|
| 685 |
+
"loss": 0.1954,
|
| 686 |
+
"step": 1900
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 1.5586546349466777,
|
| 690 |
+
"eval_loss": 0.21089071035385132,
|
| 691 |
+
"eval_runtime": 19.4624,
|
| 692 |
+
"eval_samples_per_second": 51.381,
|
| 693 |
+
"eval_steps_per_second": 0.822,
|
| 694 |
+
"step": 1900
|
| 695 |
+
},
|
| 696 |
+
{
|
| 697 |
+
"epoch": 1.579163248564397,
|
| 698 |
+
"grad_norm": 0.16736829280853271,
|
| 699 |
+
"learning_rate": 0.00012941927978647527,
|
| 700 |
+
"loss": 0.1955,
|
| 701 |
+
"step": 1925
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"epoch": 1.5996718621821167,
|
| 705 |
+
"grad_norm": 0.16830800473690033,
|
| 706 |
+
"learning_rate": 0.00011764333689131385,
|
| 707 |
+
"loss": 0.1941,
|
| 708 |
+
"step": 1950
|
| 709 |
+
},
|
| 710 |
+
{
|
| 711 |
+
"epoch": 1.620180475799836,
|
| 712 |
+
"grad_norm": 0.16350291669368744,
|
| 713 |
+
"learning_rate": 0.00010635731809312993,
|
| 714 |
+
"loss": 0.1905,
|
| 715 |
+
"step": 1975
|
| 716 |
+
},
|
| 717 |
+
{
|
| 718 |
+
"epoch": 1.6406890894175554,
|
| 719 |
+
"grad_norm": 0.16963137686252594,
|
| 720 |
+
"learning_rate": 9.557568447787201e-05,
|
| 721 |
+
"loss": 0.1962,
|
| 722 |
+
"step": 2000
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 1.6406890894175554,
|
| 726 |
+
"eval_loss": 0.20912133157253265,
|
| 727 |
+
"eval_runtime": 19.4674,
|
| 728 |
+
"eval_samples_per_second": 51.368,
|
| 729 |
+
"eval_steps_per_second": 0.822,
|
| 730 |
+
"step": 2000
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 1.661197703035275,
|
| 734 |
+
"grad_norm": 0.1589893102645874,
|
| 735 |
+
"learning_rate": 8.531225084889654e-05,
|
| 736 |
+
"loss": 0.1866,
|
| 737 |
+
"step": 2025
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 1.6817063166529942,
|
| 741 |
+
"grad_norm": 0.18232178688049316,
|
| 742 |
+
"learning_rate": 7.558016802568091e-05,
|
| 743 |
+
"loss": 0.1877,
|
| 744 |
+
"step": 2050
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 1.7022149302707137,
|
| 748 |
+
"grad_norm": 0.20000113546848297,
|
| 749 |
+
"learning_rate": 6.639190599331746e-05,
|
| 750 |
+
"loss": 0.1937,
|
| 751 |
+
"step": 2075
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 1.7227235438884332,
|
| 755 |
+
"grad_norm": 0.16770529747009277,
|
| 756 |
+
"learning_rate": 5.775923792437865e-05,
|
| 757 |
+
"loss": 0.1844,
|
| 758 |
+
"step": 2100
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 1.7227235438884332,
|
| 762 |
+
"eval_loss": 0.20784646272659302,
|
| 763 |
+
"eval_runtime": 19.4616,
|
| 764 |
+
"eval_samples_per_second": 51.383,
|
| 765 |
+
"eval_steps_per_second": 0.822,
|
| 766 |
+
"step": 2100
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 1.7432321575061525,
|
| 770 |
+
"grad_norm": 0.13503488898277283,
|
| 771 |
+
"learning_rate": 4.9693225093627616e-05,
|
| 772 |
+
"loss": 0.1879,
|
| 773 |
+
"step": 2125
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 1.7637407711238722,
|
| 777 |
+
"grad_norm": 0.1371070146560669,
|
| 778 |
+
"learning_rate": 4.220420270490294e-05,
|
| 779 |
+
"loss": 0.1865,
|
| 780 |
+
"step": 2150
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 1.7842493847415914,
|
| 784 |
+
"grad_norm": 0.14497581124305725,
|
| 785 |
+
"learning_rate": 3.530176664833834e-05,
|
| 786 |
+
"loss": 0.1955,
|
| 787 |
+
"step": 2175
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 1.804757998359311,
|
| 791 |
+
"grad_norm": 0.15688098967075348,
|
| 792 |
+
"learning_rate": 2.8994761204884756e-05,
|
| 793 |
+
"loss": 0.1884,
|
| 794 |
+
"step": 2200
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 1.804757998359311,
|
| 798 |
+
"eval_loss": 0.2068057507276535,
|
| 799 |
+
"eval_runtime": 19.4687,
|
| 800 |
+
"eval_samples_per_second": 51.365,
|
| 801 |
+
"eval_steps_per_second": 0.822,
|
| 802 |
+
"step": 2200
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 1.8252666119770304,
|
| 806 |
+
"grad_norm": 0.16965006291866302,
|
| 807 |
+
"learning_rate": 2.329126771388995e-05,
|
| 808 |
+
"loss": 0.1888,
|
| 809 |
+
"step": 2225
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 1.8457752255947497,
|
| 813 |
+
"grad_norm": 0.15735957026481628,
|
| 814 |
+
"learning_rate": 1.8198594218256815e-05,
|
| 815 |
+
"loss": 0.1901,
|
| 816 |
+
"step": 2250
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"epoch": 1.8662838392124692,
|
| 820 |
+
"grad_norm": 0.1545264571905136,
|
| 821 |
+
"learning_rate": 1.3723266100447052e-05,
|
| 822 |
+
"loss": 0.1955,
|
| 823 |
+
"step": 2275
|
| 824 |
+
},
|
| 825 |
+
{
|
| 826 |
+
"epoch": 1.8867924528301887,
|
| 827 |
+
"grad_norm": 0.14708541333675385,
|
| 828 |
+
"learning_rate": 9.871017721329201e-06,
|
| 829 |
+
"loss": 0.188,
|
| 830 |
+
"step": 2300
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 1.8867924528301887,
|
| 834 |
+
"eval_loss": 0.20641565322875977,
|
| 835 |
+
"eval_runtime": 19.4658,
|
| 836 |
+
"eval_samples_per_second": 51.372,
|
| 837 |
+
"eval_steps_per_second": 0.822,
|
| 838 |
+
"step": 2300
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 1.907301066447908,
|
| 842 |
+
"grad_norm": 0.17863567173480988,
|
| 843 |
+
"learning_rate": 6.646785072584871e-06,
|
| 844 |
+
"loss": 0.1957,
|
| 845 |
+
"step": 2325
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 1.9278096800656277,
|
| 849 |
+
"grad_norm": 0.16463416814804077,
|
| 850 |
+
"learning_rate": 4.054699452086641e-06,
|
| 851 |
+
"loss": 0.1818,
|
| 852 |
+
"step": 2350
|
| 853 |
+
},
|
| 854 |
+
{
|
| 855 |
+
"epoch": 1.948318293683347,
|
| 856 |
+
"grad_norm": 0.16183963418006897,
|
| 857 |
+
"learning_rate": 2.0980821703527886e-06,
|
| 858 |
+
"loss": 0.1861,
|
| 859 |
+
"step": 2375
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 1.9688269073010665,
|
| 863 |
+
"grad_norm": 0.15463247895240784,
|
| 864 |
+
"learning_rate": 7.794402948607671e-07,
|
| 865 |
+
"loss": 0.188,
|
| 866 |
+
"step": 2400
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 1.9688269073010665,
|
| 870 |
+
"eval_loss": 0.20634081959724426,
|
| 871 |
+
"eval_runtime": 19.4722,
|
| 872 |
+
"eval_samples_per_second": 51.355,
|
| 873 |
+
"eval_steps_per_second": 0.822,
|
| 874 |
+
"step": 2400
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"epoch": 1.989335520918786,
|
| 878 |
+
"grad_norm": 0.1495952308177948,
|
| 879 |
+
"learning_rate": 1.0046343767294853e-07,
|
| 880 |
+
"loss": 0.1839,
|
| 881 |
+
"step": 2425
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 2.0,
|
| 885 |
+
"step": 2438,
|
| 886 |
+
"total_flos": 1.58523627405312e+18,
|
| 887 |
+
"train_loss": 0.2435657412816111,
|
| 888 |
+
"train_runtime": 2897.9813,
|
| 889 |
+
"train_samples_per_second": 26.915,
|
| 890 |
+
"train_steps_per_second": 0.841
|
| 891 |
+
}
|
| 892 |
+
],
|
| 893 |
+
"logging_steps": 25,
|
| 894 |
+
"max_steps": 2438,
|
| 895 |
+
"num_input_tokens_seen": 0,
|
| 896 |
+
"num_train_epochs": 2,
|
| 897 |
+
"save_steps": 500,
|
| 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_ex04/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": 4,
|
| 9 |
+
"peft_type": "ROTATION",
|
| 10 |
+
"r": 4,
|
| 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_ex04/ft/added_tokens.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"[PAD]": 32000
|
| 3 |
+
}
|
nl_tasks/exps/run_ex04/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_ex04/ft/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
nl_tasks/exps/run_ex04/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_ex04/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_ex04/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": 4,
|
| 9 |
+
"peft_type": "ROTATION",
|
| 10 |
+
"r": 4,
|
| 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_ex04/ft2/adapter_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e97045977537f28a9fced78ef82f5f89d893e93c955eba64f7e7d14a9ce14548
|
| 3 |
+
size 33602659
|
nl_tasks/exps/run_ex04/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": 0.15271559357643127,
|
| 15 |
+
"learning_rate": 0.00019672131147540983,
|
| 16 |
+
"loss": 0.6265,
|
| 17 |
+
"step": 25
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.04101722723543889,
|
| 21 |
+
"grad_norm": 0.21552442014217377,
|
| 22 |
+
"learning_rate": 0.00040163934426229507,
|
| 23 |
+
"loss": 0.3855,
|
| 24 |
+
"step": 50
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.06152584085315833,
|
| 28 |
+
"grad_norm": 0.1532142609357834,
|
| 29 |
+
"learning_rate": 0.0006065573770491804,
|
| 30 |
+
"loss": 0.3602,
|
| 31 |
+
"step": 75
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.08203445447087777,
|
| 35 |
+
"grad_norm": 0.18029707670211792,
|
| 36 |
+
"learning_rate": 0.0008114754098360656,
|
| 37 |
+
"loss": 0.3288,
|
| 38 |
+
"step": 100
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.08203445447087777,
|
| 42 |
+
"eval_loss": 0.32045745849609375,
|
| 43 |
+
"eval_runtime": 19.8422,
|
| 44 |
+
"eval_samples_per_second": 50.398,
|
| 45 |
+
"eval_steps_per_second": 0.806,
|
| 46 |
+
"step": 100
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"epoch": 0.10254306808859721,
|
| 50 |
+
"grad_norm": 0.22050030529499054,
|
| 51 |
+
"learning_rate": 0.0010163934426229509,
|
| 52 |
+
"loss": 0.4764,
|
| 53 |
+
"step": 125
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 0.12305168170631665,
|
| 57 |
+
"grad_norm": 0.5393442511558533,
|
| 58 |
+
"learning_rate": 0.001221311475409836,
|
| 59 |
+
"loss": 0.3511,
|
| 60 |
+
"step": 150
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 0.1435602953240361,
|
| 64 |
+
"grad_norm": 10.050250053405762,
|
| 65 |
+
"learning_rate": 0.0014262295081967215,
|
| 66 |
+
"loss": 0.6435,
|
| 67 |
+
"step": 175
|
| 68 |
+
},
|
| 69 |
+
{
|
| 70 |
+
"epoch": 0.16406890894175555,
|
| 71 |
+
"grad_norm": 0.4373627007007599,
|
| 72 |
+
"learning_rate": 0.0016311475409836065,
|
| 73 |
+
"loss": 0.3664,
|
| 74 |
+
"step": 200
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.16406890894175555,
|
| 78 |
+
"eval_loss": 0.3410474956035614,
|
| 79 |
+
"eval_runtime": 19.7036,
|
| 80 |
+
"eval_samples_per_second": 50.752,
|
| 81 |
+
"eval_steps_per_second": 0.812,
|
| 82 |
+
"step": 200
|
| 83 |
+
},
|
| 84 |
+
{
|
| 85 |
+
"epoch": 0.184577522559475,
|
| 86 |
+
"grad_norm": 0.5684108734130859,
|
| 87 |
+
"learning_rate": 0.0018360655737704918,
|
| 88 |
+
"loss": 0.3623,
|
| 89 |
+
"step": 225
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 0.20508613617719443,
|
| 93 |
+
"grad_norm": 0.48954978585243225,
|
| 94 |
+
"learning_rate": 0.0019999743708232127,
|
| 95 |
+
"loss": 0.3526,
|
| 96 |
+
"step": 250
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"epoch": 0.22559474979491387,
|
| 100 |
+
"grad_norm": 0.3746850788593292,
|
| 101 |
+
"learning_rate": 0.0019990774875676054,
|
| 102 |
+
"loss": 0.3427,
|
| 103 |
+
"step": 275
|
| 104 |
+
},
|
| 105 |
+
{
|
| 106 |
+
"epoch": 0.2461033634126333,
|
| 107 |
+
"grad_norm": 0.2990626394748688,
|
| 108 |
+
"learning_rate": 0.001996900458879386,
|
| 109 |
+
"loss": 0.3356,
|
| 110 |
+
"step": 300
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.2461033634126333,
|
| 114 |
+
"eval_loss": 0.3184976875782013,
|
| 115 |
+
"eval_runtime": 19.6936,
|
| 116 |
+
"eval_samples_per_second": 50.778,
|
| 117 |
+
"eval_steps_per_second": 0.812,
|
| 118 |
+
"step": 300
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 0.2666119770303528,
|
| 122 |
+
"grad_norm": 0.25793319940567017,
|
| 123 |
+
"learning_rate": 0.001993446074245224,
|
| 124 |
+
"loss": 0.3332,
|
| 125 |
+
"step": 325
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 0.2871205906480722,
|
| 129 |
+
"grad_norm": 0.23329094052314758,
|
| 130 |
+
"learning_rate": 0.0019887187598630527,
|
| 131 |
+
"loss": 0.3153,
|
| 132 |
+
"step": 350
|
| 133 |
+
},
|
| 134 |
+
{
|
| 135 |
+
"epoch": 0.30762920426579166,
|
| 136 |
+
"grad_norm": 0.20248772203922272,
|
| 137 |
+
"learning_rate": 0.0019827245729706648,
|
| 138 |
+
"loss": 0.3214,
|
| 139 |
+
"step": 375
|
| 140 |
+
},
|
| 141 |
+
{
|
| 142 |
+
"epoch": 0.3281378178835111,
|
| 143 |
+
"grad_norm": 0.19776009023189545,
|
| 144 |
+
"learning_rate": 0.0019754711940844047,
|
| 145 |
+
"loss": 0.2986,
|
| 146 |
+
"step": 400
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.3281378178835111,
|
| 150 |
+
"eval_loss": 0.2885306477546692,
|
| 151 |
+
"eval_runtime": 19.7082,
|
| 152 |
+
"eval_samples_per_second": 50.74,
|
| 153 |
+
"eval_steps_per_second": 0.812,
|
| 154 |
+
"step": 400
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 0.34864643150123054,
|
| 158 |
+
"grad_norm": 0.1707829236984253,
|
| 159 |
+
"learning_rate": 0.0019669679171579117,
|
| 160 |
+
"loss": 0.3049,
|
| 161 |
+
"step": 425
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 0.36915504511895,
|
| 165 |
+
"grad_norm": 0.18409931659698486,
|
| 166 |
+
"learning_rate": 0.001957225637673524,
|
| 167 |
+
"loss": 0.299,
|
| 168 |
+
"step": 450
|
| 169 |
+
},
|
| 170 |
+
{
|
| 171 |
+
"epoch": 0.3896636587366694,
|
| 172 |
+
"grad_norm": 0.18975894153118134,
|
| 173 |
+
"learning_rate": 0.0019462568386815961,
|
| 174 |
+
"loss": 0.2864,
|
| 175 |
+
"step": 475
|
| 176 |
+
},
|
| 177 |
+
{
|
| 178 |
+
"epoch": 0.41017227235438886,
|
| 179 |
+
"grad_norm": 0.15703104436397552,
|
| 180 |
+
"learning_rate": 0.0019340755748056234,
|
| 181 |
+
"loss": 0.2706,
|
| 182 |
+
"step": 500
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.41017227235438886,
|
| 186 |
+
"eval_loss": 0.27103933691978455,
|
| 187 |
+
"eval_runtime": 19.6945,
|
| 188 |
+
"eval_samples_per_second": 50.776,
|
| 189 |
+
"eval_steps_per_second": 0.812,
|
| 190 |
+
"step": 500
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 0.4306808859721083,
|
| 194 |
+
"grad_norm": 0.15644876658916473,
|
| 195 |
+
"learning_rate": 0.0019206974542336672,
|
| 196 |
+
"loss": 0.2697,
|
| 197 |
+
"step": 525
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 0.45118949958982774,
|
| 201 |
+
"grad_norm": 0.16264480352401733,
|
| 202 |
+
"learning_rate": 0.0019061396187191563,
|
| 203 |
+
"loss": 0.2796,
|
| 204 |
+
"step": 550
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 0.4716981132075472,
|
| 208 |
+
"grad_norm": 0.17740413546562195,
|
| 209 |
+
"learning_rate": 0.0018904207216166836,
|
| 210 |
+
"loss": 0.2694,
|
| 211 |
+
"step": 575
|
| 212 |
+
},
|
| 213 |
+
{
|
| 214 |
+
"epoch": 0.4922067268252666,
|
| 215 |
+
"grad_norm": 0.1589346081018448,
|
| 216 |
+
"learning_rate": 0.001873560903980955,
|
| 217 |
+
"loss": 0.2856,
|
| 218 |
+
"step": 600
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.4922067268252666,
|
| 222 |
+
"eval_loss": 0.2602124512195587,
|
| 223 |
+
"eval_runtime": 19.6965,
|
| 224 |
+
"eval_samples_per_second": 50.77,
|
| 225 |
+
"eval_steps_per_second": 0.812,
|
| 226 |
+
"step": 600
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.5127153404429861,
|
| 230 |
+
"grad_norm": 0.14808161556720734,
|
| 231 |
+
"learning_rate": 0.0018555817687594984,
|
| 232 |
+
"loss": 0.2657,
|
| 233 |
+
"step": 625
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.5332239540607056,
|
| 237 |
+
"grad_norm": 0.16308413445949554,
|
| 238 |
+
"learning_rate": 0.0018365063531122169,
|
| 239 |
+
"loss": 0.2576,
|
| 240 |
+
"step": 650
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.5537325676784249,
|
| 244 |
+
"grad_norm": 0.14011067152023315,
|
| 245 |
+
"learning_rate": 0.0018163590988932402,
|
| 246 |
+
"loss": 0.2697,
|
| 247 |
+
"step": 675
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.5742411812961444,
|
| 251 |
+
"grad_norm": 0.1317146271467209,
|
| 252 |
+
"learning_rate": 0.0017951658213329078,
|
| 253 |
+
"loss": 0.2667,
|
| 254 |
+
"step": 700
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.5742411812961444,
|
| 258 |
+
"eval_loss": 0.25465214252471924,
|
| 259 |
+
"eval_runtime": 19.6966,
|
| 260 |
+
"eval_samples_per_second": 50.77,
|
| 261 |
+
"eval_steps_per_second": 0.812,
|
| 262 |
+
"step": 700
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.5947497949138638,
|
| 266 |
+
"grad_norm": 0.13956981897354126,
|
| 267 |
+
"learning_rate": 0.0017729536759600033,
|
| 268 |
+
"loss": 0.2646,
|
| 269 |
+
"step": 725
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.6152584085315833,
|
| 273 |
+
"grad_norm": 0.13741987943649292,
|
| 274 |
+
"learning_rate": 0.0017497511238066307,
|
| 275 |
+
"loss": 0.2598,
|
| 276 |
+
"step": 750
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.6357670221493027,
|
| 280 |
+
"grad_norm": 0.12589550018310547,
|
| 281 |
+
"learning_rate": 0.00172558789494031,
|
| 282 |
+
"loss": 0.258,
|
| 283 |
+
"step": 775
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.6562756357670222,
|
| 287 |
+
"grad_norm": 0.13543251156806946,
|
| 288 |
+
"learning_rate": 0.0017004949503700284,
|
| 289 |
+
"loss": 0.2607,
|
| 290 |
+
"step": 800
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.6562756357670222,
|
| 294 |
+
"eval_loss": 0.24764010310173035,
|
| 295 |
+
"eval_runtime": 19.6874,
|
| 296 |
+
"eval_samples_per_second": 50.794,
|
| 297 |
+
"eval_steps_per_second": 0.813,
|
| 298 |
+
"step": 800
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 0.6767842493847416,
|
| 302 |
+
"grad_norm": 0.12597475945949554,
|
| 303 |
+
"learning_rate": 0.0016745044423750449,
|
| 304 |
+
"loss": 0.2549,
|
| 305 |
+
"step": 825
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 0.6972928630024611,
|
| 309 |
+
"grad_norm": 0.11402735114097595,
|
| 310 |
+
"learning_rate": 0.0016476496733072946,
|
| 311 |
+
"loss": 0.2569,
|
| 312 |
+
"step": 850
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 0.7178014766201805,
|
| 316 |
+
"grad_norm": 0.1090826764702797,
|
| 317 |
+
"learning_rate": 0.0016199650529201684,
|
| 318 |
+
"loss": 0.2452,
|
| 319 |
+
"step": 875
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 0.7383100902379,
|
| 323 |
+
"grad_norm": 0.129683256149292,
|
| 324 |
+
"learning_rate": 0.0015914860542783522,
|
| 325 |
+
"loss": 0.2513,
|
| 326 |
+
"step": 900
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 0.7383100902379,
|
| 330 |
+
"eval_loss": 0.24070946872234344,
|
| 331 |
+
"eval_runtime": 19.7027,
|
| 332 |
+
"eval_samples_per_second": 50.754,
|
| 333 |
+
"eval_steps_per_second": 0.812,
|
| 334 |
+
"step": 900
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"epoch": 0.7588187038556193,
|
| 338 |
+
"grad_norm": 0.12267950922250748,
|
| 339 |
+
"learning_rate": 0.0015622491683052124,
|
| 340 |
+
"loss": 0.2529,
|
| 341 |
+
"step": 925
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 0.7793273174733388,
|
| 345 |
+
"grad_norm": 0.1230987012386322,
|
| 346 |
+
"learning_rate": 0.0015322918570259759,
|
| 347 |
+
"loss": 0.2406,
|
| 348 |
+
"step": 950
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"epoch": 0.7998359310910582,
|
| 352 |
+
"grad_norm": 0.1198684573173523,
|
| 353 |
+
"learning_rate": 0.0015016525055666057,
|
| 354 |
+
"loss": 0.2484,
|
| 355 |
+
"step": 975
|
| 356 |
+
},
|
| 357 |
+
{
|
| 358 |
+
"epoch": 0.8203445447087777,
|
| 359 |
+
"grad_norm": 0.1296621710062027,
|
| 360 |
+
"learning_rate": 0.001470370372969886,
|
| 361 |
+
"loss": 0.2414,
|
| 362 |
+
"step": 1000
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.8203445447087777,
|
| 366 |
+
"eval_loss": 0.23713567852973938,
|
| 367 |
+
"eval_runtime": 19.6967,
|
| 368 |
+
"eval_samples_per_second": 50.77,
|
| 369 |
+
"eval_steps_per_second": 0.812,
|
| 370 |
+
"step": 1000
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 0.8408531583264971,
|
| 374 |
+
"grad_norm": 0.12584251165390015,
|
| 375 |
+
"learning_rate": 0.0014384855418917311,
|
| 376 |
+
"loss": 0.2406,
|
| 377 |
+
"step": 1025
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 0.8613617719442166,
|
| 381 |
+
"grad_norm": 0.10856054723262787,
|
| 382 |
+
"learning_rate": 0.0014060388672421775,
|
| 383 |
+
"loss": 0.2392,
|
| 384 |
+
"step": 1050
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.881870385561936,
|
| 388 |
+
"grad_norm": 0.13140545785427094,
|
| 389 |
+
"learning_rate": 0.0013730719238368662,
|
| 390 |
+
"loss": 0.2433,
|
| 391 |
+
"step": 1075
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 0.9023789991796555,
|
| 395 |
+
"grad_norm": 0.11010289937257767,
|
| 396 |
+
"learning_rate": 0.0013396269531260867,
|
| 397 |
+
"loss": 0.2427,
|
| 398 |
+
"step": 1100
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 0.9023789991796555,
|
| 402 |
+
"eval_loss": 0.23183801770210266,
|
| 403 |
+
"eval_runtime": 19.6983,
|
| 404 |
+
"eval_samples_per_second": 50.766,
|
| 405 |
+
"eval_steps_per_second": 0.812,
|
| 406 |
+
"step": 1100
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 0.9228876127973749,
|
| 410 |
+
"grad_norm": 0.10757762938737869,
|
| 411 |
+
"learning_rate": 0.0013057468090696496,
|
| 412 |
+
"loss": 0.2378,
|
| 413 |
+
"step": 1125
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 0.9433962264150944,
|
| 417 |
+
"grad_norm": 0.09145833551883698,
|
| 418 |
+
"learning_rate": 0.0012714749032269287,
|
| 419 |
+
"loss": 0.2376,
|
| 420 |
+
"step": 1150
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 0.9639048400328137,
|
| 424 |
+
"grad_norm": 0.10331886261701584,
|
| 425 |
+
"learning_rate": 0.0012368551491324358,
|
| 426 |
+
"loss": 0.2436,
|
| 427 |
+
"step": 1175
|
| 428 |
+
},
|
| 429 |
+
{
|
| 430 |
+
"epoch": 0.9844134536505332,
|
| 431 |
+
"grad_norm": 0.10736266523599625,
|
| 432 |
+
"learning_rate": 0.0012019319060282063,
|
| 433 |
+
"loss": 0.2484,
|
| 434 |
+
"step": 1200
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 0.9844134536505332,
|
| 438 |
+
"eval_loss": 0.22849774360656738,
|
| 439 |
+
"eval_runtime": 19.6888,
|
| 440 |
+
"eval_samples_per_second": 50.79,
|
| 441 |
+
"eval_steps_per_second": 0.813,
|
| 442 |
+
"step": 1200
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 1.0049220672682526,
|
| 446 |
+
"grad_norm": 0.1206483393907547,
|
| 447 |
+
"learning_rate": 0.0011667499220250803,
|
| 448 |
+
"loss": 0.2279,
|
| 449 |
+
"step": 1225
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 1.0254306808859721,
|
| 453 |
+
"grad_norm": 0.127241313457489,
|
| 454 |
+
"learning_rate": 0.0011313542767657204,
|
| 455 |
+
"loss": 0.2071,
|
| 456 |
+
"step": 1250
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 1.0459392945036916,
|
| 460 |
+
"grad_norm": 0.13410116732120514,
|
| 461 |
+
"learning_rate": 0.0010957903236628267,
|
| 462 |
+
"loss": 0.2081,
|
| 463 |
+
"step": 1275
|
| 464 |
+
},
|
| 465 |
+
{
|
| 466 |
+
"epoch": 1.066447908121411,
|
| 467 |
+
"grad_norm": 0.10205130279064178,
|
| 468 |
+
"learning_rate": 0.001060103631786563,
|
| 469 |
+
"loss": 0.2126,
|
| 470 |
+
"step": 1300
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 1.066447908121411,
|
| 474 |
+
"eval_loss": 0.2257159948348999,
|
| 475 |
+
"eval_runtime": 19.6933,
|
| 476 |
+
"eval_samples_per_second": 50.779,
|
| 477 |
+
"eval_steps_per_second": 0.812,
|
| 478 |
+
"step": 1300
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 1.0869565217391304,
|
| 482 |
+
"grad_norm": 0.11324121803045273,
|
| 483 |
+
"learning_rate": 0.0010243399274756564,
|
| 484 |
+
"loss": 0.2084,
|
| 485 |
+
"step": 1325
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 1.1074651353568499,
|
| 489 |
+
"grad_norm": 0.09809240698814392,
|
| 490 |
+
"learning_rate": 0.0009885450357469806,
|
| 491 |
+
"loss": 0.2012,
|
| 492 |
+
"step": 1350
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 1.1279737489745694,
|
| 496 |
+
"grad_norm": 0.12032611668109894,
|
| 497 |
+
"learning_rate": 0.0009527648215787065,
|
| 498 |
+
"loss": 0.2023,
|
| 499 |
+
"step": 1375
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 1.1484823625922886,
|
| 503 |
+
"grad_norm": 0.11899235099554062,
|
| 504 |
+
"learning_rate": 0.000917045131142242,
|
| 505 |
+
"loss": 0.1966,
|
| 506 |
+
"step": 1400
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 1.1484823625922886,
|
| 510 |
+
"eval_loss": 0.22418975830078125,
|
| 511 |
+
"eval_runtime": 19.6934,
|
| 512 |
+
"eval_samples_per_second": 50.779,
|
| 513 |
+
"eval_steps_per_second": 0.812,
|
| 514 |
+
"step": 1400
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 1.1689909762100081,
|
| 518 |
+
"grad_norm": 0.1138472855091095,
|
| 519 |
+
"learning_rate": 0.0008814317330582753,
|
| 520 |
+
"loss": 0.2075,
|
| 521 |
+
"step": 1425
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 1.1894995898277276,
|
| 525 |
+
"grad_norm": 0.12254813313484192,
|
| 526 |
+
"learning_rate": 0.000845970259752183,
|
| 527 |
+
"loss": 0.2063,
|
| 528 |
+
"step": 1450
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 1.2100082034454471,
|
| 532 |
+
"grad_norm": 0.1570509970188141,
|
| 533 |
+
"learning_rate": 0.0008107061489839498,
|
| 534 |
+
"loss": 0.2048,
|
| 535 |
+
"step": 1475
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 1.2305168170631666,
|
| 539 |
+
"grad_norm": 0.12186683714389801,
|
| 540 |
+
"learning_rate": 0.0007756845856275194,
|
| 541 |
+
"loss": 0.2134,
|
| 542 |
+
"step": 1500
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 1.2305168170631666,
|
| 546 |
+
"eval_loss": 0.22148898243904114,
|
| 547 |
+
"eval_runtime": 19.702,
|
| 548 |
+
"eval_samples_per_second": 50.756,
|
| 549 |
+
"eval_steps_per_second": 0.812,
|
| 550 |
+
"step": 1500
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 1.251025430680886,
|
| 554 |
+
"grad_norm": 0.13762012124061584,
|
| 555 |
+
"learning_rate": 0.0007409504437741722,
|
| 556 |
+
"loss": 0.2084,
|
| 557 |
+
"step": 1525
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 1.2715340442986054,
|
| 561 |
+
"grad_norm": 0.09893760830163956,
|
| 562 |
+
"learning_rate": 0.0007065482292341205,
|
| 563 |
+
"loss": 0.2017,
|
| 564 |
+
"step": 1550
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 1.2920426579163249,
|
| 568 |
+
"grad_norm": 0.093617744743824,
|
| 569 |
+
"learning_rate": 0.0006725220225099911,
|
| 570 |
+
"loss": 0.2029,
|
| 571 |
+
"step": 1575
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 1.3125512715340442,
|
| 575 |
+
"grad_norm": 0.10118701308965683,
|
| 576 |
+
"learning_rate": 0.0006389154223152666,
|
| 577 |
+
"loss": 0.1957,
|
| 578 |
+
"step": 1600
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 1.3125512715340442,
|
| 582 |
+
"eval_loss": 0.21814565360546112,
|
| 583 |
+
"eval_runtime": 19.6919,
|
| 584 |
+
"eval_samples_per_second": 50.782,
|
| 585 |
+
"eval_steps_per_second": 0.813,
|
| 586 |
+
"step": 1600
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 1.3330598851517639,
|
| 590 |
+
"grad_norm": 0.09370667487382889,
|
| 591 |
+
"learning_rate": 0.0006057714897100551,
|
| 592 |
+
"loss": 0.1987,
|
| 593 |
+
"step": 1625
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 1.3535684987694832,
|
| 597 |
+
"grad_norm": 0.1048697754740715,
|
| 598 |
+
"learning_rate": 0.0005731326929257713,
|
| 599 |
+
"loss": 0.2003,
|
| 600 |
+
"step": 1650
|
| 601 |
+
},
|
| 602 |
+
{
|
| 603 |
+
"epoch": 1.3740771123872026,
|
| 604 |
+
"grad_norm": 0.12721873819828033,
|
| 605 |
+
"learning_rate": 0.0005410408529494251,
|
| 606 |
+
"loss": 0.1966,
|
| 607 |
+
"step": 1675
|
| 608 |
+
},
|
| 609 |
+
{
|
| 610 |
+
"epoch": 1.3945857260049221,
|
| 611 |
+
"grad_norm": 0.10494060069322586,
|
| 612 |
+
"learning_rate": 0.0005095370899372412,
|
| 613 |
+
"loss": 0.1973,
|
| 614 |
+
"step": 1700
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 1.3945857260049221,
|
| 618 |
+
"eval_loss": 0.21557489037513733,
|
| 619 |
+
"eval_runtime": 19.7084,
|
| 620 |
+
"eval_samples_per_second": 50.74,
|
| 621 |
+
"eval_steps_per_second": 0.812,
|
| 622 |
+
"step": 1700
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"epoch": 1.4150943396226414,
|
| 626 |
+
"grad_norm": 0.12225538492202759,
|
| 627 |
+
"learning_rate": 0.0004786617705262746,
|
| 628 |
+
"loss": 0.1945,
|
| 629 |
+
"step": 1725
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 1.435602953240361,
|
| 633 |
+
"grad_norm": 0.10446431487798691,
|
| 634 |
+
"learning_rate": 0.000448454456111529,
|
| 635 |
+
"loss": 0.1935,
|
| 636 |
+
"step": 1750
|
| 637 |
+
},
|
| 638 |
+
{
|
| 639 |
+
"epoch": 1.4561115668580804,
|
| 640 |
+
"grad_norm": 0.10042782127857208,
|
| 641 |
+
"learning_rate": 0.0004189538521548524,
|
| 642 |
+
"loss": 0.201,
|
| 643 |
+
"step": 1775
|
| 644 |
+
},
|
| 645 |
+
{
|
| 646 |
+
"epoch": 1.4766201804758,
|
| 647 |
+
"grad_norm": 0.08880774676799774,
|
| 648 |
+
"learning_rate": 0.00039019775859056916,
|
| 649 |
+
"loss": 0.1892,
|
| 650 |
+
"step": 1800
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 1.4766201804758,
|
| 654 |
+
"eval_loss": 0.2143358439207077,
|
| 655 |
+
"eval_runtime": 19.6963,
|
| 656 |
+
"eval_samples_per_second": 50.771,
|
| 657 |
+
"eval_steps_per_second": 0.812,
|
| 658 |
+
"step": 1800
|
| 659 |
+
},
|
| 660 |
+
{
|
| 661 |
+
"epoch": 1.4971287940935194,
|
| 662 |
+
"grad_norm": 0.1210920587182045,
|
| 663 |
+
"learning_rate": 0.0003622230213913836,
|
| 664 |
+
"loss": 0.1893,
|
| 665 |
+
"step": 1825
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 1.5176374077112387,
|
| 669 |
+
"grad_norm": 0.09757224470376968,
|
| 670 |
+
"learning_rate": 0.0003350654853566223,
|
| 671 |
+
"loss": 0.1919,
|
| 672 |
+
"step": 1850
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 1.5381460213289582,
|
| 676 |
+
"grad_norm": 0.0975116491317749,
|
| 677 |
+
"learning_rate": 0.00030875994818330957,
|
| 678 |
+
"loss": 0.1921,
|
| 679 |
+
"step": 1875
|
| 680 |
+
},
|
| 681 |
+
{
|
| 682 |
+
"epoch": 1.5586546349466777,
|
| 683 |
+
"grad_norm": 0.12393087893724442,
|
| 684 |
+
"learning_rate": 0.0002833401158789207,
|
| 685 |
+
"loss": 0.1953,
|
| 686 |
+
"step": 1900
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 1.5586546349466777,
|
| 690 |
+
"eval_loss": 0.21136102080345154,
|
| 691 |
+
"eval_runtime": 19.7233,
|
| 692 |
+
"eval_samples_per_second": 50.702,
|
| 693 |
+
"eval_steps_per_second": 0.811,
|
| 694 |
+
"step": 1900
|
| 695 |
+
},
|
| 696 |
+
{
|
| 697 |
+
"epoch": 1.579163248564397,
|
| 698 |
+
"grad_norm": 0.10960584133863449,
|
| 699 |
+
"learning_rate": 0.00025883855957295053,
|
| 700 |
+
"loss": 0.1963,
|
| 701 |
+
"step": 1925
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"epoch": 1.5996718621821167,
|
| 705 |
+
"grad_norm": 0.1282205730676651,
|
| 706 |
+
"learning_rate": 0.0002352866737826277,
|
| 707 |
+
"loss": 0.1951,
|
| 708 |
+
"step": 1950
|
| 709 |
+
},
|
| 710 |
+
{
|
| 711 |
+
"epoch": 1.620180475799836,
|
| 712 |
+
"grad_norm": 0.09847164899110794,
|
| 713 |
+
"learning_rate": 0.00021271463618625986,
|
| 714 |
+
"loss": 0.1906,
|
| 715 |
+
"step": 1975
|
| 716 |
+
},
|
| 717 |
+
{
|
| 718 |
+
"epoch": 1.6406890894175554,
|
| 719 |
+
"grad_norm": 0.100802943110466,
|
| 720 |
+
"learning_rate": 0.00019115136895574402,
|
| 721 |
+
"loss": 0.1962,
|
| 722 |
+
"step": 2000
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 1.6406890894175554,
|
| 726 |
+
"eval_loss": 0.20942208170890808,
|
| 727 |
+
"eval_runtime": 19.7225,
|
| 728 |
+
"eval_samples_per_second": 50.703,
|
| 729 |
+
"eval_steps_per_second": 0.811,
|
| 730 |
+
"step": 2000
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 1.661197703035275,
|
| 734 |
+
"grad_norm": 0.09639886766672134,
|
| 735 |
+
"learning_rate": 0.0001706245016977931,
|
| 736 |
+
"loss": 0.1865,
|
| 737 |
+
"step": 2025
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 1.6817063166529942,
|
| 741 |
+
"grad_norm": 0.12261033058166504,
|
| 742 |
+
"learning_rate": 0.00015116033605136182,
|
| 743 |
+
"loss": 0.1881,
|
| 744 |
+
"step": 2050
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 1.7022149302707137,
|
| 748 |
+
"grad_norm": 0.1281300187110901,
|
| 749 |
+
"learning_rate": 0.00013278381198663492,
|
| 750 |
+
"loss": 0.1929,
|
| 751 |
+
"step": 2075
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 1.7227235438884332,
|
| 755 |
+
"grad_norm": 0.10178139060735703,
|
| 756 |
+
"learning_rate": 0.0001155184758487573,
|
| 757 |
+
"loss": 0.185,
|
| 758 |
+
"step": 2100
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 1.7227235438884332,
|
| 762 |
+
"eval_loss": 0.20836122334003448,
|
| 763 |
+
"eval_runtime": 19.6891,
|
| 764 |
+
"eval_samples_per_second": 50.789,
|
| 765 |
+
"eval_steps_per_second": 0.813,
|
| 766 |
+
"step": 2100
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 1.7432321575061525,
|
| 770 |
+
"grad_norm": 0.08918900787830353,
|
| 771 |
+
"learning_rate": 9.938645018725523e-05,
|
| 772 |
+
"loss": 0.1875,
|
| 773 |
+
"step": 2125
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 1.7637407711238722,
|
| 777 |
+
"grad_norm": 0.08251411467790604,
|
| 778 |
+
"learning_rate": 8.440840540980587e-05,
|
| 779 |
+
"loss": 0.1863,
|
| 780 |
+
"step": 2150
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 1.7842493847415914,
|
| 784 |
+
"grad_norm": 0.09932556003332138,
|
| 785 |
+
"learning_rate": 7.060353329667668e-05,
|
| 786 |
+
"loss": 0.1946,
|
| 787 |
+
"step": 2175
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 1.804757998359311,
|
| 791 |
+
"grad_norm": 0.1116296797990799,
|
| 792 |
+
"learning_rate": 5.798952240976951e-05,
|
| 793 |
+
"loss": 0.1882,
|
| 794 |
+
"step": 2200
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 1.804757998359311,
|
| 798 |
+
"eval_loss": 0.20705059170722961,
|
| 799 |
+
"eval_runtime": 19.7069,
|
| 800 |
+
"eval_samples_per_second": 50.744,
|
| 801 |
+
"eval_steps_per_second": 0.812,
|
| 802 |
+
"step": 2200
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 1.8252666119770304,
|
| 806 |
+
"grad_norm": 0.10513120889663696,
|
| 807 |
+
"learning_rate": 4.65825354277799e-05,
|
| 808 |
+
"loss": 0.1889,
|
| 809 |
+
"step": 2225
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 1.8457752255947497,
|
| 813 |
+
"grad_norm": 0.11519550532102585,
|
| 814 |
+
"learning_rate": 3.639718843651363e-05,
|
| 815 |
+
"loss": 0.1898,
|
| 816 |
+
"step": 2250
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"epoch": 1.8662838392124692,
|
| 820 |
+
"grad_norm": 0.08914636820554733,
|
| 821 |
+
"learning_rate": 2.7446532200894104e-05,
|
| 822 |
+
"loss": 0.1955,
|
| 823 |
+
"step": 2275
|
| 824 |
+
},
|
| 825 |
+
{
|
| 826 |
+
"epoch": 1.8867924528301887,
|
| 827 |
+
"grad_norm": 0.09381356835365295,
|
| 828 |
+
"learning_rate": 1.9742035442658403e-05,
|
| 829 |
+
"loss": 0.188,
|
| 830 |
+
"step": 2300
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 1.8867924528301887,
|
| 834 |
+
"eval_loss": 0.20672424137592316,
|
| 835 |
+
"eval_runtime": 19.6977,
|
| 836 |
+
"eval_samples_per_second": 50.767,
|
| 837 |
+
"eval_steps_per_second": 0.812,
|
| 838 |
+
"step": 2300
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 1.907301066447908,
|
| 842 |
+
"grad_norm": 0.11715273559093475,
|
| 843 |
+
"learning_rate": 1.3293570145169742e-05,
|
| 844 |
+
"loss": 0.1967,
|
| 845 |
+
"step": 2325
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 1.9278096800656277,
|
| 849 |
+
"grad_norm": 0.10186558961868286,
|
| 850 |
+
"learning_rate": 8.109398904173282e-06,
|
| 851 |
+
"loss": 0.1809,
|
| 852 |
+
"step": 2350
|
| 853 |
+
},
|
| 854 |
+
{
|
| 855 |
+
"epoch": 1.948318293683347,
|
| 856 |
+
"grad_norm": 0.08567159622907639,
|
| 857 |
+
"learning_rate": 4.196164340705577e-06,
|
| 858 |
+
"loss": 0.1851,
|
| 859 |
+
"step": 2375
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 1.9688269073010665,
|
| 863 |
+
"grad_norm": 0.10170499980449677,
|
| 864 |
+
"learning_rate": 1.5588805897215342e-06,
|
| 865 |
+
"loss": 0.1878,
|
| 866 |
+
"step": 2400
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 1.9688269073010665,
|
| 870 |
+
"eval_loss": 0.20661690831184387,
|
| 871 |
+
"eval_runtime": 19.7014,
|
| 872 |
+
"eval_samples_per_second": 50.758,
|
| 873 |
+
"eval_steps_per_second": 0.812,
|
| 874 |
+
"step": 2400
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"epoch": 1.989335520918786,
|
| 878 |
+
"grad_norm": 0.10212118923664093,
|
| 879 |
+
"learning_rate": 2.0092687534589705e-07,
|
| 880 |
+
"loss": 0.1837,
|
| 881 |
+
"step": 2425
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 2.0,
|
| 885 |
+
"step": 2438,
|
| 886 |
+
"total_flos": 1.58523627405312e+18,
|
| 887 |
+
"train_loss": 0.24830095028662114,
|
| 888 |
+
"train_runtime": 3038.3364,
|
| 889 |
+
"train_samples_per_second": 25.672,
|
| 890 |
+
"train_steps_per_second": 0.802
|
| 891 |
+
}
|
| 892 |
+
],
|
| 893 |
+
"logging_steps": 25,
|
| 894 |
+
"max_steps": 2438,
|
| 895 |
+
"num_input_tokens_seen": 0,
|
| 896 |
+
"num_train_epochs": 2,
|
| 897 |
+
"save_steps": 500,
|
| 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_ex16/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/ft/special_tokens_map.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 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": "<unk>",
|
| 17 |
+
"unk_token": {
|
| 18 |
+
"content": "<unk>",
|
| 19 |
+
"lstrip": false,
|
| 20 |
+
"normalized": false,
|
| 21 |
+
"rstrip": false,
|
| 22 |
+
"single_word": false
|
| 23 |
+
}
|
| 24 |
+
}
|
nl_tasks/exps/run_ex16/ft/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
nl_tasks/exps/run_ex16/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/ft/tokenizer_config.json
ADDED
|
@@ -0,0 +1,43 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 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 |
+
},
|
| 31 |
+
"bos_token": "<s>",
|
| 32 |
+
"clean_up_tokenization_spaces": false,
|
| 33 |
+
"eos_token": "</s>",
|
| 34 |
+
"extra_special_tokens": {},
|
| 35 |
+
"legacy": false,
|
| 36 |
+
"model_max_length": 512,
|
| 37 |
+
"pad_token": "<unk>",
|
| 38 |
+
"padding_side": "right",
|
| 39 |
+
"sp_model_kwargs": {},
|
| 40 |
+
"tokenizer_class": "LlamaTokenizer",
|
| 41 |
+
"unk_token": "<unk>",
|
| 42 |
+
"use_default_system_prompt": false
|
| 43 |
+
}
|
nl_tasks/exps/run_ex16/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/ft2/adapter_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a0d8826d6f9277a1337cdad07896fa808601a05a4d275f9c1aeadb7b56c8f2fe
|
| 3 |
+
size 33602915
|
nl_tasks/exps/run_ex16/trainer_state.json
ADDED
|
@@ -0,0 +1,783 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 2.0,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 2500,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.02,
|
| 14 |
+
"grad_norm": 0.2387196570634842,
|
| 15 |
+
"learning_rate": 9.6e-05,
|
| 16 |
+
"loss": 0.6774,
|
| 17 |
+
"step": 25
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.04,
|
| 21 |
+
"grad_norm": 0.2025206983089447,
|
| 22 |
+
"learning_rate": 0.00019600000000000002,
|
| 23 |
+
"loss": 0.4087,
|
| 24 |
+
"step": 50
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.06,
|
| 28 |
+
"grad_norm": 0.18883879482746124,
|
| 29 |
+
"learning_rate": 0.000296,
|
| 30 |
+
"loss": 0.372,
|
| 31 |
+
"step": 75
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.08,
|
| 35 |
+
"grad_norm": 0.17579536139965057,
|
| 36 |
+
"learning_rate": 0.00039600000000000003,
|
| 37 |
+
"loss": 0.3376,
|
| 38 |
+
"step": 100
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.1,
|
| 42 |
+
"grad_norm": 0.1530855894088745,
|
| 43 |
+
"learning_rate": 0.000496,
|
| 44 |
+
"loss": 0.3353,
|
| 45 |
+
"step": 125
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.12,
|
| 49 |
+
"grad_norm": 0.3101711571216583,
|
| 50 |
+
"learning_rate": 0.000596,
|
| 51 |
+
"loss": 0.3248,
|
| 52 |
+
"step": 150
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.14,
|
| 56 |
+
"grad_norm": 0.17804986238479614,
|
| 57 |
+
"learning_rate": 0.000696,
|
| 58 |
+
"loss": 0.3234,
|
| 59 |
+
"step": 175
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.16,
|
| 63 |
+
"grad_norm": 0.17129909992218018,
|
| 64 |
+
"learning_rate": 0.000796,
|
| 65 |
+
"loss": 0.3115,
|
| 66 |
+
"step": 200
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.18,
|
| 70 |
+
"grad_norm": 0.21792300045490265,
|
| 71 |
+
"learning_rate": 0.000896,
|
| 72 |
+
"loss": 0.3106,
|
| 73 |
+
"step": 225
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.2,
|
| 77 |
+
"grad_norm": 0.4547678232192993,
|
| 78 |
+
"learning_rate": 0.000996,
|
| 79 |
+
"loss": 0.3779,
|
| 80 |
+
"step": 250
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.22,
|
| 84 |
+
"grad_norm": 0.3176048696041107,
|
| 85 |
+
"learning_rate": 0.0009997192908557321,
|
| 86 |
+
"loss": 0.3116,
|
| 87 |
+
"step": 275
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.24,
|
| 91 |
+
"grad_norm": 0.22115208208560944,
|
| 92 |
+
"learning_rate": 0.000998830238119205,
|
| 93 |
+
"loss": 0.3065,
|
| 94 |
+
"step": 300
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.26,
|
| 98 |
+
"grad_norm": 0.2293553352355957,
|
| 99 |
+
"learning_rate": 0.000997333437576437,
|
| 100 |
+
"loss": 0.3044,
|
| 101 |
+
"step": 325
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.28,
|
| 105 |
+
"grad_norm": 0.23258264362812042,
|
| 106 |
+
"learning_rate": 0.0009952307128483257,
|
| 107 |
+
"loss": 0.2985,
|
| 108 |
+
"step": 350
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.3,
|
| 112 |
+
"grad_norm": 0.21534672379493713,
|
| 113 |
+
"learning_rate": 0.0009925246257810518,
|
| 114 |
+
"loss": 0.2921,
|
| 115 |
+
"step": 375
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.32,
|
| 119 |
+
"grad_norm": 0.26899391412734985,
|
| 120 |
+
"learning_rate": 0.0009892184733248665,
|
| 121 |
+
"loss": 0.3071,
|
| 122 |
+
"step": 400
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.34,
|
| 126 |
+
"grad_norm": 0.24814312160015106,
|
| 127 |
+
"learning_rate": 0.0009853162835172637,
|
| 128 |
+
"loss": 0.2839,
|
| 129 |
+
"step": 425
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.36,
|
| 133 |
+
"grad_norm": 0.21491532027721405,
|
| 134 |
+
"learning_rate": 0.0009808228105754376,
|
| 135 |
+
"loss": 0.2982,
|
| 136 |
+
"step": 450
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.38,
|
| 140 |
+
"grad_norm": 0.24979755282402039,
|
| 141 |
+
"learning_rate": 0.0009757435291040016,
|
| 142 |
+
"loss": 0.2947,
|
| 143 |
+
"step": 475
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.4,
|
| 147 |
+
"grad_norm": 0.23734457790851593,
|
| 148 |
+
"learning_rate": 0.0009700846274250251,
|
| 149 |
+
"loss": 0.2813,
|
| 150 |
+
"step": 500
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.4,
|
| 154 |
+
"eval_loss": 0.28625112771987915,
|
| 155 |
+
"eval_runtime": 18.9412,
|
| 156 |
+
"eval_samples_per_second": 54.009,
|
| 157 |
+
"eval_steps_per_second": 0.845,
|
| 158 |
+
"step": 500
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.42,
|
| 162 |
+
"grad_norm": 0.2224954217672348,
|
| 163 |
+
"learning_rate": 0.000963853000038517,
|
| 164 |
+
"loss": 0.2829,
|
| 165 |
+
"step": 525
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 0.44,
|
| 169 |
+
"grad_norm": 0.2074827402830124,
|
| 170 |
+
"learning_rate": 0.0009570562392225395,
|
| 171 |
+
"loss": 0.2656,
|
| 172 |
+
"step": 550
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"epoch": 0.46,
|
| 176 |
+
"grad_norm": 0.2435082346200943,
|
| 177 |
+
"learning_rate": 0.0009497026257831855,
|
| 178 |
+
"loss": 0.2741,
|
| 179 |
+
"step": 575
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 0.48,
|
| 183 |
+
"grad_norm": 0.2333240807056427,
|
| 184 |
+
"learning_rate": 0.0009418011189656941,
|
| 185 |
+
"loss": 0.2726,
|
| 186 |
+
"step": 600
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 0.5,
|
| 190 |
+
"grad_norm": 0.18849073350429535,
|
| 191 |
+
"learning_rate": 0.0009333613455389882,
|
| 192 |
+
"loss": 0.2695,
|
| 193 |
+
"step": 625
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 0.52,
|
| 197 |
+
"grad_norm": 0.2095824033021927,
|
| 198 |
+
"learning_rate": 0.000924393588066941,
|
| 199 |
+
"loss": 0.2734,
|
| 200 |
+
"step": 650
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 0.54,
|
| 204 |
+
"grad_norm": 0.2257271260023117,
|
| 205 |
+
"learning_rate": 0.0009149087723806549,
|
| 206 |
+
"loss": 0.281,
|
| 207 |
+
"step": 675
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.56,
|
| 211 |
+
"grad_norm": 0.2157047986984253,
|
| 212 |
+
"learning_rate": 0.0009049184542670199,
|
| 213 |
+
"loss": 0.2702,
|
| 214 |
+
"step": 700
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"epoch": 0.58,
|
| 218 |
+
"grad_norm": 0.22659732401371002,
|
| 219 |
+
"learning_rate": 0.0008944348053897671,
|
| 220 |
+
"loss": 0.2793,
|
| 221 |
+
"step": 725
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 0.6,
|
| 225 |
+
"grad_norm": 0.20790499448776245,
|
| 226 |
+
"learning_rate": 0.0008834705984601709,
|
| 227 |
+
"loss": 0.273,
|
| 228 |
+
"step": 750
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 0.62,
|
| 232 |
+
"grad_norm": 0.1830059289932251,
|
| 233 |
+
"learning_rate": 0.0008720391916754683,
|
| 234 |
+
"loss": 0.2574,
|
| 235 |
+
"step": 775
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 0.64,
|
| 239 |
+
"grad_norm": 0.20472463965415955,
|
| 240 |
+
"learning_rate": 0.0008601545124439535,
|
| 241 |
+
"loss": 0.2668,
|
| 242 |
+
"step": 800
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 0.66,
|
| 246 |
+
"grad_norm": 0.20751415193080902,
|
| 247 |
+
"learning_rate": 0.0008478310404165754,
|
| 248 |
+
"loss": 0.2531,
|
| 249 |
+
"step": 825
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.68,
|
| 253 |
+
"grad_norm": 0.18556305766105652,
|
| 254 |
+
"learning_rate": 0.0008350837898457143,
|
| 255 |
+
"loss": 0.2603,
|
| 256 |
+
"step": 850
|
| 257 |
+
},
|
| 258 |
+
{
|
| 259 |
+
"epoch": 0.7,
|
| 260 |
+
"grad_norm": 0.2075965851545334,
|
| 261 |
+
"learning_rate": 0.0008219282912926269,
|
| 262 |
+
"loss": 0.2604,
|
| 263 |
+
"step": 875
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 0.72,
|
| 267 |
+
"grad_norm": 0.18890772759914398,
|
| 268 |
+
"learning_rate": 0.0008083805727058513,
|
| 269 |
+
"loss": 0.2685,
|
| 270 |
+
"step": 900
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 0.74,
|
| 274 |
+
"grad_norm": 0.16792570054531097,
|
| 275 |
+
"learning_rate": 0.0007944571398936193,
|
| 276 |
+
"loss": 0.2476,
|
| 277 |
+
"step": 925
|
| 278 |
+
},
|
| 279 |
+
{
|
| 280 |
+
"epoch": 0.76,
|
| 281 |
+
"grad_norm": 0.18068134784698486,
|
| 282 |
+
"learning_rate": 0.0007801749564140723,
|
| 283 |
+
"loss": 0.2584,
|
| 284 |
+
"step": 950
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 0.78,
|
| 288 |
+
"grad_norm": 0.1824864149093628,
|
| 289 |
+
"learning_rate": 0.0007655514229077783,
|
| 290 |
+
"loss": 0.2562,
|
| 291 |
+
"step": 975
|
| 292 |
+
},
|
| 293 |
+
{
|
| 294 |
+
"epoch": 0.8,
|
| 295 |
+
"grad_norm": 0.18031607568264008,
|
| 296 |
+
"learning_rate": 0.0007506043558977322,
|
| 297 |
+
"loss": 0.2553,
|
| 298 |
+
"step": 1000
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 0.8,
|
| 302 |
+
"eval_loss": 0.24725794792175293,
|
| 303 |
+
"eval_runtime": 18.714,
|
| 304 |
+
"eval_samples_per_second": 54.665,
|
| 305 |
+
"eval_steps_per_second": 0.855,
|
| 306 |
+
"step": 1000
|
| 307 |
+
},
|
| 308 |
+
{
|
| 309 |
+
"epoch": 0.82,
|
| 310 |
+
"grad_norm": 0.17037659883499146,
|
| 311 |
+
"learning_rate": 0.0007353519660826664,
|
| 312 |
+
"loss": 0.2401,
|
| 313 |
+
"step": 1025
|
| 314 |
+
},
|
| 315 |
+
{
|
| 316 |
+
"epoch": 0.84,
|
| 317 |
+
"grad_norm": 0.24567973613739014,
|
| 318 |
+
"learning_rate": 0.00071981283615012,
|
| 319 |
+
"loss": 0.2495,
|
| 320 |
+
"step": 1050
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 0.86,
|
| 324 |
+
"grad_norm": 0.17724287509918213,
|
| 325 |
+
"learning_rate": 0.0007040058981362964,
|
| 326 |
+
"loss": 0.2422,
|
| 327 |
+
"step": 1075
|
| 328 |
+
},
|
| 329 |
+
{
|
| 330 |
+
"epoch": 0.88,
|
| 331 |
+
"grad_norm": 0.194717139005661,
|
| 332 |
+
"learning_rate": 0.0006879504103602934,
|
| 333 |
+
"loss": 0.2433,
|
| 334 |
+
"step": 1100
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"epoch": 0.9,
|
| 338 |
+
"grad_norm": 0.1920037716627121,
|
| 339 |
+
"learning_rate": 0.0006716659339608077,
|
| 340 |
+
"loss": 0.2569,
|
| 341 |
+
"step": 1125
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 0.92,
|
| 345 |
+
"grad_norm": 0.17713096737861633,
|
| 346 |
+
"learning_rate": 0.0006551723090639006,
|
| 347 |
+
"loss": 0.2356,
|
| 348 |
+
"step": 1150
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"epoch": 0.94,
|
| 352 |
+
"grad_norm": 0.15435126423835754,
|
| 353 |
+
"learning_rate": 0.0006384896306108612,
|
| 354 |
+
"loss": 0.24,
|
| 355 |
+
"step": 1175
|
| 356 |
+
},
|
| 357 |
+
{
|
| 358 |
+
"epoch": 0.96,
|
| 359 |
+
"grad_norm": 0.15589569509029388,
|
| 360 |
+
"learning_rate": 0.0006216382238756146,
|
| 361 |
+
"loss": 0.2288,
|
| 362 |
+
"step": 1200
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.98,
|
| 366 |
+
"grad_norm": 0.15030650794506073,
|
| 367 |
+
"learning_rate": 0.0006046386197015076,
|
| 368 |
+
"loss": 0.2378,
|
| 369 |
+
"step": 1225
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 1.0,
|
| 373 |
+
"grad_norm": 0.18009062111377716,
|
| 374 |
+
"learning_rate": 0.0005875115294876381,
|
| 375 |
+
"loss": 0.2308,
|
| 376 |
+
"step": 1250
|
| 377 |
+
},
|
| 378 |
+
{
|
| 379 |
+
"epoch": 1.02,
|
| 380 |
+
"grad_norm": 0.1822308897972107,
|
| 381 |
+
"learning_rate": 0.0005702778199552054,
|
| 382 |
+
"loss": 0.2019,
|
| 383 |
+
"step": 1275
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 1.04,
|
| 387 |
+
"grad_norm": 0.16461312770843506,
|
| 388 |
+
"learning_rate": 0.000552958487724626,
|
| 389 |
+
"loss": 0.1967,
|
| 390 |
+
"step": 1300
|
| 391 |
+
},
|
| 392 |
+
{
|
| 393 |
+
"epoch": 1.06,
|
| 394 |
+
"grad_norm": 0.16238459944725037,
|
| 395 |
+
"learning_rate": 0.0005355746337343836,
|
| 396 |
+
"loss": 0.219,
|
| 397 |
+
"step": 1325
|
| 398 |
+
},
|
| 399 |
+
{
|
| 400 |
+
"epoch": 1.08,
|
| 401 |
+
"grad_norm": 0.15343919396400452,
|
| 402 |
+
"learning_rate": 0.0005181474375327879,
|
| 403 |
+
"loss": 0.1934,
|
| 404 |
+
"step": 1350
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 1.1,
|
| 408 |
+
"grad_norm": 0.16025611758232117,
|
| 409 |
+
"learning_rate": 0.0005006981314739573,
|
| 410 |
+
"loss": 0.2117,
|
| 411 |
+
"step": 1375
|
| 412 |
+
},
|
| 413 |
+
{
|
| 414 |
+
"epoch": 1.12,
|
| 415 |
+
"grad_norm": 0.16724270582199097,
|
| 416 |
+
"learning_rate": 0.00048324797484946424,
|
| 417 |
+
"loss": 0.2075,
|
| 418 |
+
"step": 1400
|
| 419 |
+
},
|
| 420 |
+
{
|
| 421 |
+
"epoch": 1.1400000000000001,
|
| 422 |
+
"grad_norm": 0.15130473673343658,
|
| 423 |
+
"learning_rate": 0.0004658182279871657,
|
| 424 |
+
"loss": 0.1992,
|
| 425 |
+
"step": 1425
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 1.16,
|
| 429 |
+
"grad_norm": 0.15824459493160248,
|
| 430 |
+
"learning_rate": 0.00044843012634876645,
|
| 431 |
+
"loss": 0.2112,
|
| 432 |
+
"step": 1450
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"epoch": 1.18,
|
| 436 |
+
"grad_norm": 0.17905935645103455,
|
| 437 |
+
"learning_rate": 0.000431104854657681,
|
| 438 |
+
"loss": 0.2015,
|
| 439 |
+
"step": 1475
|
| 440 |
+
},
|
| 441 |
+
{
|
| 442 |
+
"epoch": 1.2,
|
| 443 |
+
"grad_norm": 0.17049947381019592,
|
| 444 |
+
"learning_rate": 0.0004138635210887117,
|
| 445 |
+
"loss": 0.2047,
|
| 446 |
+
"step": 1500
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 1.2,
|
| 450 |
+
"eval_loss": 0.23122769594192505,
|
| 451 |
+
"eval_runtime": 18.7123,
|
| 452 |
+
"eval_samples_per_second": 54.67,
|
| 453 |
+
"eval_steps_per_second": 0.855,
|
| 454 |
+
"step": 1500
|
| 455 |
+
},
|
| 456 |
+
{
|
| 457 |
+
"epoch": 1.22,
|
| 458 |
+
"grad_norm": 0.13479383289813995,
|
| 459 |
+
"learning_rate": 0.0003967271315509884,
|
| 460 |
+
"loss": 0.2009,
|
| 461 |
+
"step": 1525
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 1.24,
|
| 465 |
+
"grad_norm": 0.14500467479228973,
|
| 466 |
+
"learning_rate": 0.0003797165640955041,
|
| 467 |
+
"loss": 0.1987,
|
| 468 |
+
"step": 1550
|
| 469 |
+
},
|
| 470 |
+
{
|
| 471 |
+
"epoch": 1.26,
|
| 472 |
+
"grad_norm": 0.1770029217004776,
|
| 473 |
+
"learning_rate": 0.0003628525434784268,
|
| 474 |
+
"loss": 0.2028,
|
| 475 |
+
"step": 1575
|
| 476 |
+
},
|
| 477 |
+
{
|
| 478 |
+
"epoch": 1.28,
|
| 479 |
+
"grad_norm": 0.16876231133937836,
|
| 480 |
+
"learning_rate": 0.0003461556159111748,
|
| 481 |
+
"loss": 0.2058,
|
| 482 |
+
"step": 1600
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 1.3,
|
| 486 |
+
"grad_norm": 0.16517798602581024,
|
| 487 |
+
"learning_rate": 0.0003296461240280242,
|
| 488 |
+
"loss": 0.1898,
|
| 489 |
+
"step": 1625
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 1.32,
|
| 493 |
+
"grad_norm": 0.16309712827205658,
|
| 494 |
+
"learning_rate": 0.00031334418210174266,
|
| 495 |
+
"loss": 0.2098,
|
| 496 |
+
"step": 1650
|
| 497 |
+
},
|
| 498 |
+
{
|
| 499 |
+
"epoch": 1.34,
|
| 500 |
+
"grad_norm": 0.1465151309967041,
|
| 501 |
+
"learning_rate": 0.0002972696515374455,
|
| 502 |
+
"loss": 0.1953,
|
| 503 |
+
"step": 1675
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 1.3599999999999999,
|
| 507 |
+
"grad_norm": 0.14359064400196075,
|
| 508 |
+
"learning_rate": 0.00028144211667453366,
|
| 509 |
+
"loss": 0.199,
|
| 510 |
+
"step": 1700
|
| 511 |
+
},
|
| 512 |
+
{
|
| 513 |
+
"epoch": 1.38,
|
| 514 |
+
"grad_norm": 0.15689799189567566,
|
| 515 |
+
"learning_rate": 0.00026588086092619277,
|
| 516 |
+
"loss": 0.2043,
|
| 517 |
+
"step": 1725
|
| 518 |
+
},
|
| 519 |
+
{
|
| 520 |
+
"epoch": 1.4,
|
| 521 |
+
"grad_norm": 0.17195850610733032,
|
| 522 |
+
"learning_rate": 0.00025060484328552466,
|
| 523 |
+
"loss": 0.2001,
|
| 524 |
+
"step": 1750
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 1.42,
|
| 528 |
+
"grad_norm": 0.16419224441051483,
|
| 529 |
+
"learning_rate": 0.00023563267522693415,
|
| 530 |
+
"loss": 0.1934,
|
| 531 |
+
"step": 1775
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 1.44,
|
| 535 |
+
"grad_norm": 0.3002654016017914,
|
| 536 |
+
"learning_rate": 0.0002209825980309151,
|
| 537 |
+
"loss": 0.2008,
|
| 538 |
+
"step": 1800
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"epoch": 1.46,
|
| 542 |
+
"grad_norm": 0.16668134927749634,
|
| 543 |
+
"learning_rate": 0.00020667246055985938,
|
| 544 |
+
"loss": 0.2023,
|
| 545 |
+
"step": 1825
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 1.48,
|
| 549 |
+
"grad_norm": 0.16942422091960907,
|
| 550 |
+
"learning_rate": 0.00019271969751196778,
|
| 551 |
+
"loss": 0.1865,
|
| 552 |
+
"step": 1850
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 1.5,
|
| 556 |
+
"grad_norm": 0.16257637739181519,
|
| 557 |
+
"learning_rate": 0.00017914130817975592,
|
| 558 |
+
"loss": 0.1996,
|
| 559 |
+
"step": 1875
|
| 560 |
+
},
|
| 561 |
+
{
|
| 562 |
+
"epoch": 1.52,
|
| 563 |
+
"grad_norm": 0.15791665017604828,
|
| 564 |
+
"learning_rate": 0.00016595383573903412,
|
| 565 |
+
"loss": 0.1926,
|
| 566 |
+
"step": 1900
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 1.54,
|
| 570 |
+
"grad_norm": 0.16652727127075195,
|
| 571 |
+
"learning_rate": 0.0001531733470935976,
|
| 572 |
+
"loss": 0.1941,
|
| 573 |
+
"step": 1925
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 1.56,
|
| 577 |
+
"grad_norm": 0.14352388679981232,
|
| 578 |
+
"learning_rate": 0.00014081541330017704,
|
| 579 |
+
"loss": 0.1854,
|
| 580 |
+
"step": 1950
|
| 581 |
+
},
|
| 582 |
+
{
|
| 583 |
+
"epoch": 1.58,
|
| 584 |
+
"grad_norm": 0.16893626749515533,
|
| 585 |
+
"learning_rate": 0.00012889509059750602,
|
| 586 |
+
"loss": 0.1887,
|
| 587 |
+
"step": 1975
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 1.6,
|
| 591 |
+
"grad_norm": 0.17489072680473328,
|
| 592 |
+
"learning_rate": 0.00011742690206261292,
|
| 593 |
+
"loss": 0.1931,
|
| 594 |
+
"step": 2000
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 1.6,
|
| 598 |
+
"eval_loss": 0.21856220066547394,
|
| 599 |
+
"eval_runtime": 18.7163,
|
| 600 |
+
"eval_samples_per_second": 54.658,
|
| 601 |
+
"eval_steps_per_second": 0.855,
|
| 602 |
+
"step": 2000
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 1.62,
|
| 606 |
+
"grad_norm": 0.16639254987239838,
|
| 607 |
+
"learning_rate": 0.0001064248199166884,
|
| 608 |
+
"loss": 0.1827,
|
| 609 |
+
"step": 2025
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 1.6400000000000001,
|
| 613 |
+
"grad_norm": 0.16985327005386353,
|
| 614 |
+
"learning_rate": 9.590224850208645e-05,
|
| 615 |
+
"loss": 0.1895,
|
| 616 |
+
"step": 2050
|
| 617 |
+
},
|
| 618 |
+
{
|
| 619 |
+
"epoch": 1.6600000000000001,
|
| 620 |
+
"grad_norm": 0.15598806738853455,
|
| 621 |
+
"learning_rate": 8.587200795119792e-05,
|
| 622 |
+
"loss": 0.1949,
|
| 623 |
+
"step": 2075
|
| 624 |
+
},
|
| 625 |
+
{
|
| 626 |
+
"epoch": 1.6800000000000002,
|
| 627 |
+
"grad_norm": 0.19577306509017944,
|
| 628 |
+
"learning_rate": 7.634631856709389e-05,
|
| 629 |
+
"loss": 0.196,
|
| 630 |
+
"step": 2100
|
| 631 |
+
},
|
| 632 |
+
{
|
| 633 |
+
"epoch": 1.7,
|
| 634 |
+
"grad_norm": 0.17040973901748657,
|
| 635 |
+
"learning_rate": 6.733678593496901e-05,
|
| 636 |
+
"loss": 0.1957,
|
| 637 |
+
"step": 2125
|
| 638 |
+
},
|
| 639 |
+
{
|
| 640 |
+
"epoch": 1.72,
|
| 641 |
+
"grad_norm": 0.1549983024597168,
|
| 642 |
+
"learning_rate": 5.885438678252342e-05,
|
| 643 |
+
"loss": 0.1904,
|
| 644 |
+
"step": 2150
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 1.74,
|
| 648 |
+
"grad_norm": 0.14541535079479218,
|
| 649 |
+
"learning_rate": 5.0909455606510726e-05,
|
| 650 |
+
"loss": 0.1857,
|
| 651 |
+
"step": 2175
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 1.76,
|
| 655 |
+
"grad_norm": 0.16388991475105286,
|
| 656 |
+
"learning_rate": 4.3511672081746386e-05,
|
| 657 |
+
"loss": 0.1891,
|
| 658 |
+
"step": 2200
|
| 659 |
+
},
|
| 660 |
+
{
|
| 661 |
+
"epoch": 1.78,
|
| 662 |
+
"grad_norm": 0.1539498269557953,
|
| 663 |
+
"learning_rate": 3.667004926791395e-05,
|
| 664 |
+
"loss": 0.1922,
|
| 665 |
+
"step": 2225
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 1.8,
|
| 669 |
+
"grad_norm": 0.1366104632616043,
|
| 670 |
+
"learning_rate": 3.0392922628540875e-05,
|
| 671 |
+
"loss": 0.1868,
|
| 672 |
+
"step": 2250
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 1.8199999999999998,
|
| 676 |
+
"grad_norm": 0.14811132848262787,
|
| 677 |
+
"learning_rate": 2.468793987551998e-05,
|
| 678 |
+
"loss": 0.1848,
|
| 679 |
+
"step": 2275
|
| 680 |
+
},
|
| 681 |
+
{
|
| 682 |
+
"epoch": 1.8399999999999999,
|
| 683 |
+
"grad_norm": 0.16613063216209412,
|
| 684 |
+
"learning_rate": 1.9562051651550784e-05,
|
| 685 |
+
"loss": 0.1808,
|
| 686 |
+
"step": 2300
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 1.8599999999999999,
|
| 690 |
+
"grad_norm": 0.15500903129577637,
|
| 691 |
+
"learning_rate": 1.5021503061851349e-05,
|
| 692 |
+
"loss": 0.1902,
|
| 693 |
+
"step": 2325
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 1.88,
|
| 697 |
+
"grad_norm": 0.15229476988315582,
|
| 698 |
+
"learning_rate": 1.1071826065460589e-05,
|
| 699 |
+
"loss": 0.1783,
|
| 700 |
+
"step": 2350
|
| 701 |
+
},
|
| 702 |
+
{
|
| 703 |
+
"epoch": 1.9,
|
| 704 |
+
"grad_norm": 0.18280497193336487,
|
| 705 |
+
"learning_rate": 7.717832735397334e-06,
|
| 706 |
+
"loss": 0.1851,
|
| 707 |
+
"step": 2375
|
| 708 |
+
},
|
| 709 |
+
{
|
| 710 |
+
"epoch": 1.92,
|
| 711 |
+
"grad_norm": 0.15305499732494354,
|
| 712 |
+
"learning_rate": 4.963609395891299e-06,
|
| 713 |
+
"loss": 0.1835,
|
| 714 |
+
"step": 2400
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 1.94,
|
| 718 |
+
"grad_norm": 0.15484832227230072,
|
| 719 |
+
"learning_rate": 2.81251164382601e-06,
|
| 720 |
+
"loss": 0.1853,
|
| 721 |
+
"step": 2425
|
| 722 |
+
},
|
| 723 |
+
{
|
| 724 |
+
"epoch": 1.96,
|
| 725 |
+
"grad_norm": 0.15939371287822723,
|
| 726 |
+
"learning_rate": 1.267160260461253e-06,
|
| 727 |
+
"loss": 0.1758,
|
| 728 |
+
"step": 2450
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 1.98,
|
| 732 |
+
"grad_norm": 0.15895092487335205,
|
| 733 |
+
"learning_rate": 3.2943801841439634e-07,
|
| 734 |
+
"loss": 0.1833,
|
| 735 |
+
"step": 2475
|
| 736 |
+
},
|
| 737 |
+
{
|
| 738 |
+
"epoch": 2.0,
|
| 739 |
+
"grad_norm": 0.13426445424556732,
|
| 740 |
+
"learning_rate": 4.873877924582715e-10,
|
| 741 |
+
"loss": 0.1847,
|
| 742 |
+
"step": 2500
|
| 743 |
+
},
|
| 744 |
+
{
|
| 745 |
+
"epoch": 2.0,
|
| 746 |
+
"eval_loss": 0.21446388959884644,
|
| 747 |
+
"eval_runtime": 18.7159,
|
| 748 |
+
"eval_samples_per_second": 54.659,
|
| 749 |
+
"eval_steps_per_second": 0.855,
|
| 750 |
+
"step": 2500
|
| 751 |
+
},
|
| 752 |
+
{
|
| 753 |
+
"epoch": 2.0,
|
| 754 |
+
"step": 2500,
|
| 755 |
+
"total_flos": 1.62588235137024e+18,
|
| 756 |
+
"train_loss": 0.242121671295166,
|
| 757 |
+
"train_runtime": 2369.7199,
|
| 758 |
+
"train_samples_per_second": 33.759,
|
| 759 |
+
"train_steps_per_second": 1.055
|
| 760 |
+
}
|
| 761 |
+
],
|
| 762 |
+
"logging_steps": 25,
|
| 763 |
+
"max_steps": 2500,
|
| 764 |
+
"num_input_tokens_seen": 0,
|
| 765 |
+
"num_train_epochs": 2,
|
| 766 |
+
"save_steps": 0,
|
| 767 |
+
"stateful_callbacks": {
|
| 768 |
+
"TrainerControl": {
|
| 769 |
+
"args": {
|
| 770 |
+
"should_epoch_stop": false,
|
| 771 |
+
"should_evaluate": false,
|
| 772 |
+
"should_log": false,
|
| 773 |
+
"should_save": true,
|
| 774 |
+
"should_training_stop": true
|
| 775 |
+
},
|
| 776 |
+
"attributes": {}
|
| 777 |
+
}
|
| 778 |
+
},
|
| 779 |
+
"total_flos": 1.62588235137024e+18,
|
| 780 |
+
"train_batch_size": 32,
|
| 781 |
+
"trial_name": null,
|
| 782 |
+
"trial_params": null
|
| 783 |
+
}
|
nl_tasks/exps/run_ex17/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_ex17/ft/special_tokens_map.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 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": "<unk>",
|
| 17 |
+
"unk_token": {
|
| 18 |
+
"content": "<unk>",
|
| 19 |
+
"lstrip": false,
|
| 20 |
+
"normalized": false,
|
| 21 |
+
"rstrip": false,
|
| 22 |
+
"single_word": false
|
| 23 |
+
}
|
| 24 |
+
}
|
nl_tasks/exps/run_ex17/ft/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
nl_tasks/exps/run_ex17/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_ex17/ft/tokenizer_config.json
ADDED
|
@@ -0,0 +1,43 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 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 |
+
},
|
| 31 |
+
"bos_token": "<s>",
|
| 32 |
+
"clean_up_tokenization_spaces": false,
|
| 33 |
+
"eos_token": "</s>",
|
| 34 |
+
"extra_special_tokens": {},
|
| 35 |
+
"legacy": false,
|
| 36 |
+
"model_max_length": 512,
|
| 37 |
+
"pad_token": "<unk>",
|
| 38 |
+
"padding_side": "right",
|
| 39 |
+
"sp_model_kwargs": {},
|
| 40 |
+
"tokenizer_class": "LlamaTokenizer",
|
| 41 |
+
"unk_token": "<unk>",
|
| 42 |
+
"use_default_system_prompt": false
|
| 43 |
+
}
|
nl_tasks/exps/run_ex17/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_ex17/ft2/adapter_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:369c2266de02f26ca225488bf2ceedbafac7904204543319e17a3a6726f49a0b
|
| 3 |
+
size 33602915
|
nl_tasks/exps/run_ex17/trainer_state.json
ADDED
|
@@ -0,0 +1,783 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 2.0,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 2500,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.02,
|
| 14 |
+
"grad_norm": 0.6064024567604065,
|
| 15 |
+
"learning_rate": 9.6e-05,
|
| 16 |
+
"loss": 0.7526,
|
| 17 |
+
"step": 25
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.04,
|
| 21 |
+
"grad_norm": 0.1889362633228302,
|
| 22 |
+
"learning_rate": 0.00019600000000000002,
|
| 23 |
+
"loss": 0.431,
|
| 24 |
+
"step": 50
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.06,
|
| 28 |
+
"grad_norm": 0.18094255030155182,
|
| 29 |
+
"learning_rate": 0.000296,
|
| 30 |
+
"loss": 0.3866,
|
| 31 |
+
"step": 75
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.08,
|
| 35 |
+
"grad_norm": 0.17763806879520416,
|
| 36 |
+
"learning_rate": 0.00039600000000000003,
|
| 37 |
+
"loss": 0.3514,
|
| 38 |
+
"step": 100
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.1,
|
| 42 |
+
"grad_norm": 0.22015658020973206,
|
| 43 |
+
"learning_rate": 0.000496,
|
| 44 |
+
"loss": 0.348,
|
| 45 |
+
"step": 125
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.12,
|
| 49 |
+
"grad_norm": 0.1794177144765854,
|
| 50 |
+
"learning_rate": 0.000596,
|
| 51 |
+
"loss": 0.3366,
|
| 52 |
+
"step": 150
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.14,
|
| 56 |
+
"grad_norm": 0.18052445352077484,
|
| 57 |
+
"learning_rate": 0.000696,
|
| 58 |
+
"loss": 0.3388,
|
| 59 |
+
"step": 175
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.16,
|
| 63 |
+
"grad_norm": 0.18147766590118408,
|
| 64 |
+
"learning_rate": 0.000796,
|
| 65 |
+
"loss": 0.3276,
|
| 66 |
+
"step": 200
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.18,
|
| 70 |
+
"grad_norm": 0.17726336419582367,
|
| 71 |
+
"learning_rate": 0.000896,
|
| 72 |
+
"loss": 0.3222,
|
| 73 |
+
"step": 225
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.2,
|
| 77 |
+
"grad_norm": 0.2130950391292572,
|
| 78 |
+
"learning_rate": 0.000996,
|
| 79 |
+
"loss": 0.327,
|
| 80 |
+
"step": 250
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.22,
|
| 84 |
+
"grad_norm": 0.21618539094924927,
|
| 85 |
+
"learning_rate": 0.0009997192908557321,
|
| 86 |
+
"loss": 0.3165,
|
| 87 |
+
"step": 275
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.24,
|
| 91 |
+
"grad_norm": 0.2549876868724823,
|
| 92 |
+
"learning_rate": 0.000998830238119205,
|
| 93 |
+
"loss": 0.3124,
|
| 94 |
+
"step": 300
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.26,
|
| 98 |
+
"grad_norm": 0.2416405975818634,
|
| 99 |
+
"learning_rate": 0.000997333437576437,
|
| 100 |
+
"loss": 0.3106,
|
| 101 |
+
"step": 325
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.28,
|
| 105 |
+
"grad_norm": 0.20741847157478333,
|
| 106 |
+
"learning_rate": 0.0009952307128483257,
|
| 107 |
+
"loss": 0.3045,
|
| 108 |
+
"step": 350
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.3,
|
| 112 |
+
"grad_norm": 0.2067340612411499,
|
| 113 |
+
"learning_rate": 0.0009925246257810518,
|
| 114 |
+
"loss": 0.2954,
|
| 115 |
+
"step": 375
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.32,
|
| 119 |
+
"grad_norm": 0.24208734929561615,
|
| 120 |
+
"learning_rate": 0.0009892184733248665,
|
| 121 |
+
"loss": 0.3109,
|
| 122 |
+
"step": 400
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.34,
|
| 126 |
+
"grad_norm": 0.2122250646352768,
|
| 127 |
+
"learning_rate": 0.0009853162835172637,
|
| 128 |
+
"loss": 0.29,
|
| 129 |
+
"step": 425
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.36,
|
| 133 |
+
"grad_norm": 0.27720949053764343,
|
| 134 |
+
"learning_rate": 0.0009808228105754376,
|
| 135 |
+
"loss": 0.3032,
|
| 136 |
+
"step": 450
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.38,
|
| 140 |
+
"grad_norm": 0.2765713334083557,
|
| 141 |
+
"learning_rate": 0.0009757435291040016,
|
| 142 |
+
"loss": 0.3023,
|
| 143 |
+
"step": 475
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.4,
|
| 147 |
+
"grad_norm": 0.24158088862895966,
|
| 148 |
+
"learning_rate": 0.0009700846274250251,
|
| 149 |
+
"loss": 0.2895,
|
| 150 |
+
"step": 500
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.4,
|
| 154 |
+
"eval_loss": 0.2855643928050995,
|
| 155 |
+
"eval_runtime": 18.9222,
|
| 156 |
+
"eval_samples_per_second": 54.064,
|
| 157 |
+
"eval_steps_per_second": 0.846,
|
| 158 |
+
"step": 500
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.42,
|
| 162 |
+
"grad_norm": 0.23705452680587769,
|
| 163 |
+
"learning_rate": 0.000963853000038517,
|
| 164 |
+
"loss": 0.2898,
|
| 165 |
+
"step": 525
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 0.44,
|
| 169 |
+
"grad_norm": 0.21587742865085602,
|
| 170 |
+
"learning_rate": 0.0009570562392225395,
|
| 171 |
+
"loss": 0.2737,
|
| 172 |
+
"step": 550
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"epoch": 0.46,
|
| 176 |
+
"grad_norm": 0.23051080107688904,
|
| 177 |
+
"learning_rate": 0.0009497026257831855,
|
| 178 |
+
"loss": 0.2823,
|
| 179 |
+
"step": 575
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 0.48,
|
| 183 |
+
"grad_norm": 0.21226011216640472,
|
| 184 |
+
"learning_rate": 0.0009418011189656941,
|
| 185 |
+
"loss": 0.2784,
|
| 186 |
+
"step": 600
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 0.5,
|
| 190 |
+
"grad_norm": 0.2159241884946823,
|
| 191 |
+
"learning_rate": 0.0009333613455389882,
|
| 192 |
+
"loss": 0.2779,
|
| 193 |
+
"step": 625
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 0.52,
|
| 197 |
+
"grad_norm": 0.23518994450569153,
|
| 198 |
+
"learning_rate": 0.000924393588066941,
|
| 199 |
+
"loss": 0.282,
|
| 200 |
+
"step": 650
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 0.54,
|
| 204 |
+
"grad_norm": 0.24779067933559418,
|
| 205 |
+
"learning_rate": 0.0009149087723806549,
|
| 206 |
+
"loss": 0.29,
|
| 207 |
+
"step": 675
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.56,
|
| 211 |
+
"grad_norm": 0.2609941065311432,
|
| 212 |
+
"learning_rate": 0.0009049184542670199,
|
| 213 |
+
"loss": 0.2769,
|
| 214 |
+
"step": 700
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"epoch": 0.58,
|
| 218 |
+
"grad_norm": 0.23452705144882202,
|
| 219 |
+
"learning_rate": 0.0008944348053897671,
|
| 220 |
+
"loss": 0.2848,
|
| 221 |
+
"step": 725
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 0.6,
|
| 225 |
+
"grad_norm": 0.20832745730876923,
|
| 226 |
+
"learning_rate": 0.0008834705984601709,
|
| 227 |
+
"loss": 0.28,
|
| 228 |
+
"step": 750
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 0.62,
|
| 232 |
+
"grad_norm": 0.20192205905914307,
|
| 233 |
+
"learning_rate": 0.0008720391916754683,
|
| 234 |
+
"loss": 0.265,
|
| 235 |
+
"step": 775
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 0.64,
|
| 239 |
+
"grad_norm": 0.22747072577476501,
|
| 240 |
+
"learning_rate": 0.0008601545124439535,
|
| 241 |
+
"loss": 0.2754,
|
| 242 |
+
"step": 800
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 0.66,
|
| 246 |
+
"grad_norm": 0.22271917760372162,
|
| 247 |
+
"learning_rate": 0.0008478310404165754,
|
| 248 |
+
"loss": 0.2608,
|
| 249 |
+
"step": 825
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.68,
|
| 253 |
+
"grad_norm": 0.1978376805782318,
|
| 254 |
+
"learning_rate": 0.0008350837898457143,
|
| 255 |
+
"loss": 0.269,
|
| 256 |
+
"step": 850
|
| 257 |
+
},
|
| 258 |
+
{
|
| 259 |
+
"epoch": 0.7,
|
| 260 |
+
"grad_norm": 0.21664343774318695,
|
| 261 |
+
"learning_rate": 0.0008219282912926269,
|
| 262 |
+
"loss": 0.2685,
|
| 263 |
+
"step": 875
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 0.72,
|
| 267 |
+
"grad_norm": 0.23120196163654327,
|
| 268 |
+
"learning_rate": 0.0008083805727058513,
|
| 269 |
+
"loss": 0.2754,
|
| 270 |
+
"step": 900
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 0.74,
|
| 274 |
+
"grad_norm": 0.17918695509433746,
|
| 275 |
+
"learning_rate": 0.0007944571398936193,
|
| 276 |
+
"loss": 0.2559,
|
| 277 |
+
"step": 925
|
| 278 |
+
},
|
| 279 |
+
{
|
| 280 |
+
"epoch": 0.76,
|
| 281 |
+
"grad_norm": 0.19703985750675201,
|
| 282 |
+
"learning_rate": 0.0007801749564140723,
|
| 283 |
+
"loss": 0.2667,
|
| 284 |
+
"step": 950
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 0.78,
|
| 288 |
+
"grad_norm": 0.306613951921463,
|
| 289 |
+
"learning_rate": 0.0007655514229077783,
|
| 290 |
+
"loss": 0.3008,
|
| 291 |
+
"step": 975
|
| 292 |
+
},
|
| 293 |
+
{
|
| 294 |
+
"epoch": 0.8,
|
| 295 |
+
"grad_norm": 0.16847214102745056,
|
| 296 |
+
"learning_rate": 0.0007506043558977322,
|
| 297 |
+
"loss": 0.267,
|
| 298 |
+
"step": 1000
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 0.8,
|
| 302 |
+
"eval_loss": 0.25149136781692505,
|
| 303 |
+
"eval_runtime": 18.7169,
|
| 304 |
+
"eval_samples_per_second": 54.657,
|
| 305 |
+
"eval_steps_per_second": 0.855,
|
| 306 |
+
"step": 1000
|
| 307 |
+
},
|
| 308 |
+
{
|
| 309 |
+
"epoch": 0.82,
|
| 310 |
+
"grad_norm": 0.2017536610364914,
|
| 311 |
+
"learning_rate": 0.0007353519660826664,
|
| 312 |
+
"loss": 0.2488,
|
| 313 |
+
"step": 1025
|
| 314 |
+
},
|
| 315 |
+
{
|
| 316 |
+
"epoch": 0.84,
|
| 317 |
+
"grad_norm": 0.26620712876319885,
|
| 318 |
+
"learning_rate": 0.00071981283615012,
|
| 319 |
+
"loss": 0.2578,
|
| 320 |
+
"step": 1050
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 0.86,
|
| 324 |
+
"grad_norm": 0.18335208296775818,
|
| 325 |
+
"learning_rate": 0.0007040058981362964,
|
| 326 |
+
"loss": 0.2515,
|
| 327 |
+
"step": 1075
|
| 328 |
+
},
|
| 329 |
+
{
|
| 330 |
+
"epoch": 0.88,
|
| 331 |
+
"grad_norm": 0.19252872467041016,
|
| 332 |
+
"learning_rate": 0.0006879504103602934,
|
| 333 |
+
"loss": 0.2517,
|
| 334 |
+
"step": 1100
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"epoch": 0.9,
|
| 338 |
+
"grad_norm": 0.2431110143661499,
|
| 339 |
+
"learning_rate": 0.0006716659339608077,
|
| 340 |
+
"loss": 0.2654,
|
| 341 |
+
"step": 1125
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 0.92,
|
| 345 |
+
"grad_norm": 0.22655946016311646,
|
| 346 |
+
"learning_rate": 0.0006551723090639006,
|
| 347 |
+
"loss": 0.2429,
|
| 348 |
+
"step": 1150
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"epoch": 0.94,
|
| 352 |
+
"grad_norm": 0.1737266182899475,
|
| 353 |
+
"learning_rate": 0.0006384896306108612,
|
| 354 |
+
"loss": 0.2482,
|
| 355 |
+
"step": 1175
|
| 356 |
+
},
|
| 357 |
+
{
|
| 358 |
+
"epoch": 0.96,
|
| 359 |
+
"grad_norm": 0.1587584763765335,
|
| 360 |
+
"learning_rate": 0.0006216382238756146,
|
| 361 |
+
"loss": 0.236,
|
| 362 |
+
"step": 1200
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.98,
|
| 366 |
+
"grad_norm": 0.18093238770961761,
|
| 367 |
+
"learning_rate": 0.0006046386197015076,
|
| 368 |
+
"loss": 0.2447,
|
| 369 |
+
"step": 1225
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 1.0,
|
| 373 |
+
"grad_norm": 0.16756398975849152,
|
| 374 |
+
"learning_rate": 0.0005875115294876381,
|
| 375 |
+
"loss": 0.2374,
|
| 376 |
+
"step": 1250
|
| 377 |
+
},
|
| 378 |
+
{
|
| 379 |
+
"epoch": 1.02,
|
| 380 |
+
"grad_norm": 0.17455460131168365,
|
| 381 |
+
"learning_rate": 0.0005702778199552054,
|
| 382 |
+
"loss": 0.2133,
|
| 383 |
+
"step": 1275
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 1.04,
|
| 387 |
+
"grad_norm": 0.1666153520345688,
|
| 388 |
+
"learning_rate": 0.000552958487724626,
|
| 389 |
+
"loss": 0.2073,
|
| 390 |
+
"step": 1300
|
| 391 |
+
},
|
| 392 |
+
{
|
| 393 |
+
"epoch": 1.06,
|
| 394 |
+
"grad_norm": 0.1747789978981018,
|
| 395 |
+
"learning_rate": 0.0005355746337343836,
|
| 396 |
+
"loss": 0.2312,
|
| 397 |
+
"step": 1325
|
| 398 |
+
},
|
| 399 |
+
{
|
| 400 |
+
"epoch": 1.08,
|
| 401 |
+
"grad_norm": 0.16713109612464905,
|
| 402 |
+
"learning_rate": 0.0005181474375327879,
|
| 403 |
+
"loss": 0.2048,
|
| 404 |
+
"step": 1350
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 1.1,
|
| 408 |
+
"grad_norm": 0.1740274429321289,
|
| 409 |
+
"learning_rate": 0.0005006981314739573,
|
| 410 |
+
"loss": 0.2226,
|
| 411 |
+
"step": 1375
|
| 412 |
+
},
|
| 413 |
+
{
|
| 414 |
+
"epoch": 1.12,
|
| 415 |
+
"grad_norm": 0.16742606461048126,
|
| 416 |
+
"learning_rate": 0.00048324797484946424,
|
| 417 |
+
"loss": 0.2187,
|
| 418 |
+
"step": 1400
|
| 419 |
+
},
|
| 420 |
+
{
|
| 421 |
+
"epoch": 1.1400000000000001,
|
| 422 |
+
"grad_norm": 0.16527721285820007,
|
| 423 |
+
"learning_rate": 0.0004658182279871657,
|
| 424 |
+
"loss": 0.2108,
|
| 425 |
+
"step": 1425
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 1.16,
|
| 429 |
+
"grad_norm": 0.16400551795959473,
|
| 430 |
+
"learning_rate": 0.00044843012634876645,
|
| 431 |
+
"loss": 0.2219,
|
| 432 |
+
"step": 1450
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"epoch": 1.18,
|
| 436 |
+
"grad_norm": 0.2440529763698578,
|
| 437 |
+
"learning_rate": 0.000431104854657681,
|
| 438 |
+
"loss": 0.2135,
|
| 439 |
+
"step": 1475
|
| 440 |
+
},
|
| 441 |
+
{
|
| 442 |
+
"epoch": 1.2,
|
| 443 |
+
"grad_norm": 0.17534707486629486,
|
| 444 |
+
"learning_rate": 0.0004138635210887117,
|
| 445 |
+
"loss": 0.2177,
|
| 446 |
+
"step": 1500
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 1.2,
|
| 450 |
+
"eval_loss": 0.2326875478029251,
|
| 451 |
+
"eval_runtime": 18.7207,
|
| 452 |
+
"eval_samples_per_second": 54.645,
|
| 453 |
+
"eval_steps_per_second": 0.855,
|
| 454 |
+
"step": 1500
|
| 455 |
+
},
|
| 456 |
+
{
|
| 457 |
+
"epoch": 1.22,
|
| 458 |
+
"grad_norm": 0.15731997787952423,
|
| 459 |
+
"learning_rate": 0.0003967271315509884,
|
| 460 |
+
"loss": 0.2116,
|
| 461 |
+
"step": 1525
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 1.24,
|
| 465 |
+
"grad_norm": 0.16014693677425385,
|
| 466 |
+
"learning_rate": 0.0003797165640955041,
|
| 467 |
+
"loss": 0.209,
|
| 468 |
+
"step": 1550
|
| 469 |
+
},
|
| 470 |
+
{
|
| 471 |
+
"epoch": 1.26,
|
| 472 |
+
"grad_norm": 0.18441081047058105,
|
| 473 |
+
"learning_rate": 0.0003628525434784268,
|
| 474 |
+
"loss": 0.2148,
|
| 475 |
+
"step": 1575
|
| 476 |
+
},
|
| 477 |
+
{
|
| 478 |
+
"epoch": 1.28,
|
| 479 |
+
"grad_norm": 0.16412386298179626,
|
| 480 |
+
"learning_rate": 0.0003461556159111748,
|
| 481 |
+
"loss": 0.2174,
|
| 482 |
+
"step": 1600
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 1.3,
|
| 486 |
+
"grad_norm": 0.1740948110818863,
|
| 487 |
+
"learning_rate": 0.0003296461240280242,
|
| 488 |
+
"loss": 0.2012,
|
| 489 |
+
"step": 1625
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 1.32,
|
| 493 |
+
"grad_norm": 0.16411401331424713,
|
| 494 |
+
"learning_rate": 0.00031334418210174266,
|
| 495 |
+
"loss": 0.2229,
|
| 496 |
+
"step": 1650
|
| 497 |
+
},
|
| 498 |
+
{
|
| 499 |
+
"epoch": 1.34,
|
| 500 |
+
"grad_norm": 0.15569698810577393,
|
| 501 |
+
"learning_rate": 0.0002972696515374455,
|
| 502 |
+
"loss": 0.2052,
|
| 503 |
+
"step": 1675
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 1.3599999999999999,
|
| 507 |
+
"grad_norm": 0.1418256163597107,
|
| 508 |
+
"learning_rate": 0.00028144211667453366,
|
| 509 |
+
"loss": 0.2094,
|
| 510 |
+
"step": 1700
|
| 511 |
+
},
|
| 512 |
+
{
|
| 513 |
+
"epoch": 1.38,
|
| 514 |
+
"grad_norm": 0.16271263360977173,
|
| 515 |
+
"learning_rate": 0.00026588086092619277,
|
| 516 |
+
"loss": 0.2157,
|
| 517 |
+
"step": 1725
|
| 518 |
+
},
|
| 519 |
+
{
|
| 520 |
+
"epoch": 1.4,
|
| 521 |
+
"grad_norm": 0.18504226207733154,
|
| 522 |
+
"learning_rate": 0.00025060484328552466,
|
| 523 |
+
"loss": 0.2117,
|
| 524 |
+
"step": 1750
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 1.42,
|
| 528 |
+
"grad_norm": 0.16541805863380432,
|
| 529 |
+
"learning_rate": 0.00023563267522693415,
|
| 530 |
+
"loss": 0.2038,
|
| 531 |
+
"step": 1775
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 1.44,
|
| 535 |
+
"grad_norm": 0.1936596930027008,
|
| 536 |
+
"learning_rate": 0.0002209825980309151,
|
| 537 |
+
"loss": 0.213,
|
| 538 |
+
"step": 1800
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"epoch": 1.46,
|
| 542 |
+
"grad_norm": 0.16703884303569794,
|
| 543 |
+
"learning_rate": 0.00020667246055985938,
|
| 544 |
+
"loss": 0.2127,
|
| 545 |
+
"step": 1825
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 1.48,
|
| 549 |
+
"grad_norm": 0.16527888178825378,
|
| 550 |
+
"learning_rate": 0.00019271969751196778,
|
| 551 |
+
"loss": 0.1969,
|
| 552 |
+
"step": 1850
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 1.5,
|
| 556 |
+
"grad_norm": 0.18421709537506104,
|
| 557 |
+
"learning_rate": 0.00017914130817975592,
|
| 558 |
+
"loss": 0.2095,
|
| 559 |
+
"step": 1875
|
| 560 |
+
},
|
| 561 |
+
{
|
| 562 |
+
"epoch": 1.52,
|
| 563 |
+
"grad_norm": 0.13981053233146667,
|
| 564 |
+
"learning_rate": 0.00016595383573903412,
|
| 565 |
+
"loss": 0.2037,
|
| 566 |
+
"step": 1900
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 1.54,
|
| 570 |
+
"grad_norm": 0.18436624109745026,
|
| 571 |
+
"learning_rate": 0.0001531733470935976,
|
| 572 |
+
"loss": 0.2062,
|
| 573 |
+
"step": 1925
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 1.56,
|
| 577 |
+
"grad_norm": 0.13563580811023712,
|
| 578 |
+
"learning_rate": 0.00014081541330017704,
|
| 579 |
+
"loss": 0.1951,
|
| 580 |
+
"step": 1950
|
| 581 |
+
},
|
| 582 |
+
{
|
| 583 |
+
"epoch": 1.58,
|
| 584 |
+
"grad_norm": 0.16344304382801056,
|
| 585 |
+
"learning_rate": 0.00012889509059750602,
|
| 586 |
+
"loss": 0.1994,
|
| 587 |
+
"step": 1975
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 1.6,
|
| 591 |
+
"grad_norm": 0.16763299703598022,
|
| 592 |
+
"learning_rate": 0.00011742690206261292,
|
| 593 |
+
"loss": 0.204,
|
| 594 |
+
"step": 2000
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 1.6,
|
| 598 |
+
"eval_loss": 0.21985949575901031,
|
| 599 |
+
"eval_runtime": 18.7145,
|
| 600 |
+
"eval_samples_per_second": 54.663,
|
| 601 |
+
"eval_steps_per_second": 0.855,
|
| 602 |
+
"step": 2000
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 1.62,
|
| 606 |
+
"grad_norm": 0.16287456452846527,
|
| 607 |
+
"learning_rate": 0.0001064248199166884,
|
| 608 |
+
"loss": 0.1933,
|
| 609 |
+
"step": 2025
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 1.6400000000000001,
|
| 613 |
+
"grad_norm": 0.1603902280330658,
|
| 614 |
+
"learning_rate": 9.590224850208645e-05,
|
| 615 |
+
"loss": 0.2027,
|
| 616 |
+
"step": 2050
|
| 617 |
+
},
|
| 618 |
+
{
|
| 619 |
+
"epoch": 1.6600000000000001,
|
| 620 |
+
"grad_norm": 0.15367600321769714,
|
| 621 |
+
"learning_rate": 8.587200795119792e-05,
|
| 622 |
+
"loss": 0.2057,
|
| 623 |
+
"step": 2075
|
| 624 |
+
},
|
| 625 |
+
{
|
| 626 |
+
"epoch": 1.6800000000000002,
|
| 627 |
+
"grad_norm": 0.16175620257854462,
|
| 628 |
+
"learning_rate": 7.634631856709389e-05,
|
| 629 |
+
"loss": 0.2073,
|
| 630 |
+
"step": 2100
|
| 631 |
+
},
|
| 632 |
+
{
|
| 633 |
+
"epoch": 1.7,
|
| 634 |
+
"grad_norm": 0.17625470459461212,
|
| 635 |
+
"learning_rate": 6.733678593496901e-05,
|
| 636 |
+
"loss": 0.2066,
|
| 637 |
+
"step": 2125
|
| 638 |
+
},
|
| 639 |
+
{
|
| 640 |
+
"epoch": 1.72,
|
| 641 |
+
"grad_norm": 0.15195347368717194,
|
| 642 |
+
"learning_rate": 5.885438678252342e-05,
|
| 643 |
+
"loss": 0.2008,
|
| 644 |
+
"step": 2150
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 1.74,
|
| 648 |
+
"grad_norm": 0.14848165214061737,
|
| 649 |
+
"learning_rate": 5.0909455606510726e-05,
|
| 650 |
+
"loss": 0.1973,
|
| 651 |
+
"step": 2175
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 1.76,
|
| 655 |
+
"grad_norm": 0.16471602022647858,
|
| 656 |
+
"learning_rate": 4.3511672081746386e-05,
|
| 657 |
+
"loss": 0.2003,
|
| 658 |
+
"step": 2200
|
| 659 |
+
},
|
| 660 |
+
{
|
| 661 |
+
"epoch": 1.78,
|
| 662 |
+
"grad_norm": 0.1609528362751007,
|
| 663 |
+
"learning_rate": 3.667004926791395e-05,
|
| 664 |
+
"loss": 0.204,
|
| 665 |
+
"step": 2225
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 1.8,
|
| 669 |
+
"grad_norm": 0.14461441338062286,
|
| 670 |
+
"learning_rate": 3.0392922628540875e-05,
|
| 671 |
+
"loss": 0.1982,
|
| 672 |
+
"step": 2250
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 1.8199999999999998,
|
| 676 |
+
"grad_norm": 0.14768414199352264,
|
| 677 |
+
"learning_rate": 2.468793987551998e-05,
|
| 678 |
+
"loss": 0.1964,
|
| 679 |
+
"step": 2275
|
| 680 |
+
},
|
| 681 |
+
{
|
| 682 |
+
"epoch": 1.8399999999999999,
|
| 683 |
+
"grad_norm": 0.16912218928337097,
|
| 684 |
+
"learning_rate": 1.9562051651550784e-05,
|
| 685 |
+
"loss": 0.1917,
|
| 686 |
+
"step": 2300
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 1.8599999999999999,
|
| 690 |
+
"grad_norm": 0.1520395576953888,
|
| 691 |
+
"learning_rate": 1.5021503061851349e-05,
|
| 692 |
+
"loss": 0.2021,
|
| 693 |
+
"step": 2325
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 1.88,
|
| 697 |
+
"grad_norm": 0.15839679539203644,
|
| 698 |
+
"learning_rate": 1.1071826065460589e-05,
|
| 699 |
+
"loss": 0.1894,
|
| 700 |
+
"step": 2350
|
| 701 |
+
},
|
| 702 |
+
{
|
| 703 |
+
"epoch": 1.9,
|
| 704 |
+
"grad_norm": 0.16376681625843048,
|
| 705 |
+
"learning_rate": 7.717832735397334e-06,
|
| 706 |
+
"loss": 0.1948,
|
| 707 |
+
"step": 2375
|
| 708 |
+
},
|
| 709 |
+
{
|
| 710 |
+
"epoch": 1.92,
|
| 711 |
+
"grad_norm": 0.1577637493610382,
|
| 712 |
+
"learning_rate": 4.963609395891299e-06,
|
| 713 |
+
"loss": 0.1954,
|
| 714 |
+
"step": 2400
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 1.94,
|
| 718 |
+
"grad_norm": 0.158824160695076,
|
| 719 |
+
"learning_rate": 2.81251164382601e-06,
|
| 720 |
+
"loss": 0.1978,
|
| 721 |
+
"step": 2425
|
| 722 |
+
},
|
| 723 |
+
{
|
| 724 |
+
"epoch": 1.96,
|
| 725 |
+
"grad_norm": 0.16813521087169647,
|
| 726 |
+
"learning_rate": 1.267160260461253e-06,
|
| 727 |
+
"loss": 0.1857,
|
| 728 |
+
"step": 2450
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 1.98,
|
| 732 |
+
"grad_norm": 0.15693625807762146,
|
| 733 |
+
"learning_rate": 3.2943801841439634e-07,
|
| 734 |
+
"loss": 0.1944,
|
| 735 |
+
"step": 2475
|
| 736 |
+
},
|
| 737 |
+
{
|
| 738 |
+
"epoch": 2.0,
|
| 739 |
+
"grad_norm": 0.14145386219024658,
|
| 740 |
+
"learning_rate": 4.873877924582715e-10,
|
| 741 |
+
"loss": 0.1952,
|
| 742 |
+
"step": 2500
|
| 743 |
+
},
|
| 744 |
+
{
|
| 745 |
+
"epoch": 2.0,
|
| 746 |
+
"eval_loss": 0.21614596247673035,
|
| 747 |
+
"eval_runtime": 18.7001,
|
| 748 |
+
"eval_samples_per_second": 54.706,
|
| 749 |
+
"eval_steps_per_second": 0.856,
|
| 750 |
+
"step": 2500
|
| 751 |
+
},
|
| 752 |
+
{
|
| 753 |
+
"epoch": 2.0,
|
| 754 |
+
"step": 2500,
|
| 755 |
+
"total_flos": 1.62588235137024e+18,
|
| 756 |
+
"train_loss": 0.25245753231048584,
|
| 757 |
+
"train_runtime": 2494.6259,
|
| 758 |
+
"train_samples_per_second": 32.069,
|
| 759 |
+
"train_steps_per_second": 1.002
|
| 760 |
+
}
|
| 761 |
+
],
|
| 762 |
+
"logging_steps": 25,
|
| 763 |
+
"max_steps": 2500,
|
| 764 |
+
"num_input_tokens_seen": 0,
|
| 765 |
+
"num_train_epochs": 2,
|
| 766 |
+
"save_steps": 0,
|
| 767 |
+
"stateful_callbacks": {
|
| 768 |
+
"TrainerControl": {
|
| 769 |
+
"args": {
|
| 770 |
+
"should_epoch_stop": false,
|
| 771 |
+
"should_evaluate": false,
|
| 772 |
+
"should_log": false,
|
| 773 |
+
"should_save": true,
|
| 774 |
+
"should_training_stop": true
|
| 775 |
+
},
|
| 776 |
+
"attributes": {}
|
| 777 |
+
}
|
| 778 |
+
},
|
| 779 |
+
"total_flos": 1.62588235137024e+18,
|
| 780 |
+
"train_batch_size": 32,
|
| 781 |
+
"trial_name": null,
|
| 782 |
+
"trial_params": null
|
| 783 |
+
}
|
nl_tasks/exps/run_ex18/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_ex18/ft/special_tokens_map.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 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": "<unk>",
|
| 17 |
+
"unk_token": {
|
| 18 |
+
"content": "<unk>",
|
| 19 |
+
"lstrip": false,
|
| 20 |
+
"normalized": false,
|
| 21 |
+
"rstrip": false,
|
| 22 |
+
"single_word": false
|
| 23 |
+
}
|
| 24 |
+
}
|
nl_tasks/exps/run_ex18/ft/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
nl_tasks/exps/run_ex18/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_ex18/ft/tokenizer_config.json
ADDED
|
@@ -0,0 +1,43 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 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 |
+
},
|
| 31 |
+
"bos_token": "<s>",
|
| 32 |
+
"clean_up_tokenization_spaces": false,
|
| 33 |
+
"eos_token": "</s>",
|
| 34 |
+
"extra_special_tokens": {},
|
| 35 |
+
"legacy": false,
|
| 36 |
+
"model_max_length": 512,
|
| 37 |
+
"pad_token": "<unk>",
|
| 38 |
+
"padding_side": "right",
|
| 39 |
+
"sp_model_kwargs": {},
|
| 40 |
+
"tokenizer_class": "LlamaTokenizer",
|
| 41 |
+
"unk_token": "<unk>",
|
| 42 |
+
"use_default_system_prompt": false
|
| 43 |
+
}
|
nl_tasks/exps/run_ex18/ft2/adapter_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:fd6be1170c30d475f79f3891823bb960793d3024ca6e742fff866f704f2d143c
|
| 3 |
+
size 33602915
|
nl_tasks/exps/run_ex18/trainer_state.json
ADDED
|
@@ -0,0 +1,1149 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 3.0,
|
| 6 |
+
"eval_steps": 500,
|
| 7 |
+
"global_step": 3750,
|
| 8 |
+
"is_hyper_param_search": false,
|
| 9 |
+
"is_local_process_zero": true,
|
| 10 |
+
"is_world_process_zero": true,
|
| 11 |
+
"log_history": [
|
| 12 |
+
{
|
| 13 |
+
"epoch": 0.02,
|
| 14 |
+
"grad_norm": 1.1674950122833252,
|
| 15 |
+
"learning_rate": 6.4e-05,
|
| 16 |
+
"loss": 0.8724,
|
| 17 |
+
"step": 25
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.04,
|
| 21 |
+
"grad_norm": 0.2832157015800476,
|
| 22 |
+
"learning_rate": 0.00013066666666666665,
|
| 23 |
+
"loss": 0.4969,
|
| 24 |
+
"step": 50
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.06,
|
| 28 |
+
"grad_norm": 0.20104631781578064,
|
| 29 |
+
"learning_rate": 0.00019733333333333335,
|
| 30 |
+
"loss": 0.4162,
|
| 31 |
+
"step": 75
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.08,
|
| 35 |
+
"grad_norm": 0.2139427810907364,
|
| 36 |
+
"learning_rate": 0.000264,
|
| 37 |
+
"loss": 0.3734,
|
| 38 |
+
"step": 100
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.1,
|
| 42 |
+
"grad_norm": 0.17443759739398956,
|
| 43 |
+
"learning_rate": 0.00033066666666666666,
|
| 44 |
+
"loss": 0.3667,
|
| 45 |
+
"step": 125
|
| 46 |
+
},
|
| 47 |
+
{
|
| 48 |
+
"epoch": 0.12,
|
| 49 |
+
"grad_norm": 0.20704133808612823,
|
| 50 |
+
"learning_rate": 0.0003973333333333333,
|
| 51 |
+
"loss": 0.3543,
|
| 52 |
+
"step": 150
|
| 53 |
+
},
|
| 54 |
+
{
|
| 55 |
+
"epoch": 0.14,
|
| 56 |
+
"grad_norm": 0.1681472808122635,
|
| 57 |
+
"learning_rate": 0.00046400000000000006,
|
| 58 |
+
"loss": 0.3504,
|
| 59 |
+
"step": 175
|
| 60 |
+
},
|
| 61 |
+
{
|
| 62 |
+
"epoch": 0.16,
|
| 63 |
+
"grad_norm": 0.17286404967308044,
|
| 64 |
+
"learning_rate": 0.0005306666666666666,
|
| 65 |
+
"loss": 0.3357,
|
| 66 |
+
"step": 200
|
| 67 |
+
},
|
| 68 |
+
{
|
| 69 |
+
"epoch": 0.18,
|
| 70 |
+
"grad_norm": 0.21512570977210999,
|
| 71 |
+
"learning_rate": 0.0005973333333333334,
|
| 72 |
+
"loss": 0.3304,
|
| 73 |
+
"step": 225
|
| 74 |
+
},
|
| 75 |
+
{
|
| 76 |
+
"epoch": 0.2,
|
| 77 |
+
"grad_norm": 0.17349863052368164,
|
| 78 |
+
"learning_rate": 0.0006640000000000001,
|
| 79 |
+
"loss": 0.3344,
|
| 80 |
+
"step": 250
|
| 81 |
+
},
|
| 82 |
+
{
|
| 83 |
+
"epoch": 0.22,
|
| 84 |
+
"grad_norm": 0.20435509085655212,
|
| 85 |
+
"learning_rate": 0.0007306666666666667,
|
| 86 |
+
"loss": 0.3245,
|
| 87 |
+
"step": 275
|
| 88 |
+
},
|
| 89 |
+
{
|
| 90 |
+
"epoch": 0.24,
|
| 91 |
+
"grad_norm": 0.16571108996868134,
|
| 92 |
+
"learning_rate": 0.0007973333333333334,
|
| 93 |
+
"loss": 0.3215,
|
| 94 |
+
"step": 300
|
| 95 |
+
},
|
| 96 |
+
{
|
| 97 |
+
"epoch": 0.26,
|
| 98 |
+
"grad_norm": 0.22708046436309814,
|
| 99 |
+
"learning_rate": 0.000864,
|
| 100 |
+
"loss": 0.3194,
|
| 101 |
+
"step": 325
|
| 102 |
+
},
|
| 103 |
+
{
|
| 104 |
+
"epoch": 0.28,
|
| 105 |
+
"grad_norm": 0.19495542347431183,
|
| 106 |
+
"learning_rate": 0.0009306666666666667,
|
| 107 |
+
"loss": 0.3147,
|
| 108 |
+
"step": 350
|
| 109 |
+
},
|
| 110 |
+
{
|
| 111 |
+
"epoch": 0.3,
|
| 112 |
+
"grad_norm": 0.19485542178153992,
|
| 113 |
+
"learning_rate": 0.0009973333333333334,
|
| 114 |
+
"loss": 0.308,
|
| 115 |
+
"step": 375
|
| 116 |
+
},
|
| 117 |
+
{
|
| 118 |
+
"epoch": 0.32,
|
| 119 |
+
"grad_norm": 0.24180231988430023,
|
| 120 |
+
"learning_rate": 0.0009998752338940611,
|
| 121 |
+
"loss": 0.3278,
|
| 122 |
+
"step": 400
|
| 123 |
+
},
|
| 124 |
+
{
|
| 125 |
+
"epoch": 0.34,
|
| 126 |
+
"grad_norm": 0.21253246068954468,
|
| 127 |
+
"learning_rate": 0.0009994799931474475,
|
| 128 |
+
"loss": 0.3048,
|
| 129 |
+
"step": 425
|
| 130 |
+
},
|
| 131 |
+
{
|
| 132 |
+
"epoch": 0.36,
|
| 133 |
+
"grad_norm": 0.2390223890542984,
|
| 134 |
+
"learning_rate": 0.0009988142751731796,
|
| 135 |
+
"loss": 0.3199,
|
| 136 |
+
"step": 450
|
| 137 |
+
},
|
| 138 |
+
{
|
| 139 |
+
"epoch": 0.38,
|
| 140 |
+
"grad_norm": 0.24184420704841614,
|
| 141 |
+
"learning_rate": 0.0009978784404692847,
|
| 142 |
+
"loss": 0.3149,
|
| 143 |
+
"step": 475
|
| 144 |
+
},
|
| 145 |
+
{
|
| 146 |
+
"epoch": 0.4,
|
| 147 |
+
"grad_norm": 0.2562459707260132,
|
| 148 |
+
"learning_rate": 0.0009966729958067638,
|
| 149 |
+
"loss": 0.3038,
|
| 150 |
+
"step": 500
|
| 151 |
+
},
|
| 152 |
+
{
|
| 153 |
+
"epoch": 0.4,
|
| 154 |
+
"eval_loss": 0.2947409152984619,
|
| 155 |
+
"eval_runtime": 18.9289,
|
| 156 |
+
"eval_samples_per_second": 54.044,
|
| 157 |
+
"eval_steps_per_second": 0.845,
|
| 158 |
+
"step": 500
|
| 159 |
+
},
|
| 160 |
+
{
|
| 161 |
+
"epoch": 0.42,
|
| 162 |
+
"grad_norm": 0.2443474531173706,
|
| 163 |
+
"learning_rate": 0.0009951985939551673,
|
| 164 |
+
"loss": 0.305,
|
| 165 |
+
"step": 525
|
| 166 |
+
},
|
| 167 |
+
{
|
| 168 |
+
"epoch": 0.44,
|
| 169 |
+
"grad_norm": 0.22000126540660858,
|
| 170 |
+
"learning_rate": 0.0009934560333291077,
|
| 171 |
+
"loss": 0.288,
|
| 172 |
+
"step": 550
|
| 173 |
+
},
|
| 174 |
+
{
|
| 175 |
+
"epoch": 0.46,
|
| 176 |
+
"grad_norm": 0.2495642602443695,
|
| 177 |
+
"learning_rate": 0.0009914462575559045,
|
| 178 |
+
"loss": 0.2983,
|
| 179 |
+
"step": 575
|
| 180 |
+
},
|
| 181 |
+
{
|
| 182 |
+
"epoch": 0.48,
|
| 183 |
+
"grad_norm": 0.6387917995452881,
|
| 184 |
+
"learning_rate": 0.000989170354964594,
|
| 185 |
+
"loss": 0.5488,
|
| 186 |
+
"step": 600
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 0.5,
|
| 190 |
+
"grad_norm": 0.20248781144618988,
|
| 191 |
+
"learning_rate": 0.0009866295579965781,
|
| 192 |
+
"loss": 0.2921,
|
| 193 |
+
"step": 625
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 0.52,
|
| 197 |
+
"grad_norm": 0.3091880679130554,
|
| 198 |
+
"learning_rate": 0.000983825242538238,
|
| 199 |
+
"loss": 0.2965,
|
| 200 |
+
"step": 650
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 0.54,
|
| 204 |
+
"grad_norm": 0.25312355160713196,
|
| 205 |
+
"learning_rate": 0.0009807589271758651,
|
| 206 |
+
"loss": 0.3051,
|
| 207 |
+
"step": 675
|
| 208 |
+
},
|
| 209 |
+
{
|
| 210 |
+
"epoch": 0.56,
|
| 211 |
+
"grad_norm": 0.2332855761051178,
|
| 212 |
+
"learning_rate": 0.0009774322723733215,
|
| 213 |
+
"loss": 0.2895,
|
| 214 |
+
"step": 700
|
| 215 |
+
},
|
| 216 |
+
{
|
| 217 |
+
"epoch": 0.58,
|
| 218 |
+
"grad_norm": 0.23708628118038177,
|
| 219 |
+
"learning_rate": 0.0009738470795728685,
|
| 220 |
+
"loss": 0.2979,
|
| 221 |
+
"step": 725
|
| 222 |
+
},
|
| 223 |
+
{
|
| 224 |
+
"epoch": 0.6,
|
| 225 |
+
"grad_norm": 0.2137805074453354,
|
| 226 |
+
"learning_rate": 0.000970005290219654,
|
| 227 |
+
"loss": 0.2936,
|
| 228 |
+
"step": 750
|
| 229 |
+
},
|
| 230 |
+
{
|
| 231 |
+
"epoch": 0.62,
|
| 232 |
+
"grad_norm": 0.2010948210954666,
|
| 233 |
+
"learning_rate": 0.0009659089847103863,
|
| 234 |
+
"loss": 0.2774,
|
| 235 |
+
"step": 775
|
| 236 |
+
},
|
| 237 |
+
{
|
| 238 |
+
"epoch": 0.64,
|
| 239 |
+
"grad_norm": 0.21842093765735626,
|
| 240 |
+
"learning_rate": 0.0009615603812667617,
|
| 241 |
+
"loss": 0.2887,
|
| 242 |
+
"step": 800
|
| 243 |
+
},
|
| 244 |
+
{
|
| 245 |
+
"epoch": 0.66,
|
| 246 |
+
"grad_norm": 0.24124319851398468,
|
| 247 |
+
"learning_rate": 0.0009569618347342592,
|
| 248 |
+
"loss": 0.2732,
|
| 249 |
+
"step": 825
|
| 250 |
+
},
|
| 251 |
+
{
|
| 252 |
+
"epoch": 0.68,
|
| 253 |
+
"grad_norm": 0.2690409719944,
|
| 254 |
+
"learning_rate": 0.0009521158353069494,
|
| 255 |
+
"loss": 0.2911,
|
| 256 |
+
"step": 850
|
| 257 |
+
},
|
| 258 |
+
{
|
| 259 |
+
"epoch": 0.7,
|
| 260 |
+
"grad_norm": 0.21264797449111938,
|
| 261 |
+
"learning_rate": 0.0009470250071790107,
|
| 262 |
+
"loss": 0.2819,
|
| 263 |
+
"step": 875
|
| 264 |
+
},
|
| 265 |
+
{
|
| 266 |
+
"epoch": 0.72,
|
| 267 |
+
"grad_norm": 0.26778486371040344,
|
| 268 |
+
"learning_rate": 0.0009416921071236822,
|
| 269 |
+
"loss": 0.289,
|
| 270 |
+
"step": 900
|
| 271 |
+
},
|
| 272 |
+
{
|
| 273 |
+
"epoch": 0.74,
|
| 274 |
+
"grad_norm": 0.1802307665348053,
|
| 275 |
+
"learning_rate": 0.0009361200230004219,
|
| 276 |
+
"loss": 0.268,
|
| 277 |
+
"step": 925
|
| 278 |
+
},
|
| 279 |
+
{
|
| 280 |
+
"epoch": 0.76,
|
| 281 |
+
"grad_norm": 0.18479594588279724,
|
| 282 |
+
"learning_rate": 0.0009303117721910801,
|
| 283 |
+
"loss": 0.2807,
|
| 284 |
+
"step": 950
|
| 285 |
+
},
|
| 286 |
+
{
|
| 287 |
+
"epoch": 0.78,
|
| 288 |
+
"grad_norm": 0.19667844474315643,
|
| 289 |
+
"learning_rate": 0.0009242704999659339,
|
| 290 |
+
"loss": 0.2775,
|
| 291 |
+
"step": 975
|
| 292 |
+
},
|
| 293 |
+
{
|
| 294 |
+
"epoch": 0.8,
|
| 295 |
+
"grad_norm": 0.26136162877082825,
|
| 296 |
+
"learning_rate": 0.0009179994777804677,
|
| 297 |
+
"loss": 0.2776,
|
| 298 |
+
"step": 1000
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 0.8,
|
| 302 |
+
"eval_loss": 0.25686538219451904,
|
| 303 |
+
"eval_runtime": 18.7344,
|
| 304 |
+
"eval_samples_per_second": 54.605,
|
| 305 |
+
"eval_steps_per_second": 0.854,
|
| 306 |
+
"step": 1000
|
| 307 |
+
},
|
| 308 |
+
{
|
| 309 |
+
"epoch": 0.82,
|
| 310 |
+
"grad_norm": 0.1717814952135086,
|
| 311 |
+
"learning_rate": 0.0009115021015038226,
|
| 312 |
+
"loss": 0.2585,
|
| 313 |
+
"step": 1025
|
| 314 |
+
},
|
| 315 |
+
{
|
| 316 |
+
"epoch": 0.84,
|
| 317 |
+
"grad_norm": 0.23498405516147614,
|
| 318 |
+
"learning_rate": 0.0009047818895798731,
|
| 319 |
+
"loss": 0.2707,
|
| 320 |
+
"step": 1050
|
| 321 |
+
},
|
| 322 |
+
{
|
| 323 |
+
"epoch": 0.86,
|
| 324 |
+
"grad_norm": 0.18124710023403168,
|
| 325 |
+
"learning_rate": 0.0008978424811219277,
|
| 326 |
+
"loss": 0.2648,
|
| 327 |
+
"step": 1075
|
| 328 |
+
},
|
| 329 |
+
{
|
| 330 |
+
"epoch": 0.88,
|
| 331 |
+
"grad_norm": 0.19796232879161835,
|
| 332 |
+
"learning_rate": 0.0008906876339420851,
|
| 333 |
+
"loss": 0.2642,
|
| 334 |
+
"step": 1100
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"epoch": 0.9,
|
| 338 |
+
"grad_norm": 0.20190514624118805,
|
| 339 |
+
"learning_rate": 0.0008833212225163124,
|
| 340 |
+
"loss": 0.2788,
|
| 341 |
+
"step": 1125
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 0.92,
|
| 345 |
+
"grad_norm": 0.2207169383764267,
|
| 346 |
+
"learning_rate": 0.0008757472358863481,
|
| 347 |
+
"loss": 0.2549,
|
| 348 |
+
"step": 1150
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"epoch": 0.94,
|
| 352 |
+
"grad_norm": 0.15971709787845612,
|
| 353 |
+
"learning_rate": 0.0008679697754995655,
|
| 354 |
+
"loss": 0.262,
|
| 355 |
+
"step": 1175
|
| 356 |
+
},
|
| 357 |
+
{
|
| 358 |
+
"epoch": 0.96,
|
| 359 |
+
"grad_norm": 0.17682687938213348,
|
| 360 |
+
"learning_rate": 0.0008599930529879669,
|
| 361 |
+
"loss": 0.2467,
|
| 362 |
+
"step": 1200
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.98,
|
| 366 |
+
"grad_norm": 0.16468586027622223,
|
| 367 |
+
"learning_rate": 0.0008518213878875102,
|
| 368 |
+
"loss": 0.2576,
|
| 369 |
+
"step": 1225
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 1.0,
|
| 373 |
+
"grad_norm": 0.1838867962360382,
|
| 374 |
+
"learning_rate": 0.0008434592052990044,
|
| 375 |
+
"loss": 0.2514,
|
| 376 |
+
"step": 1250
|
| 377 |
+
},
|
| 378 |
+
{
|
| 379 |
+
"epoch": 1.02,
|
| 380 |
+
"grad_norm": 0.18226249516010284,
|
| 381 |
+
"learning_rate": 0.0008349110334918391,
|
| 382 |
+
"loss": 0.2308,
|
| 383 |
+
"step": 1275
|
| 384 |
+
},
|
| 385 |
+
{
|
| 386 |
+
"epoch": 1.04,
|
| 387 |
+
"grad_norm": 0.164842426776886,
|
| 388 |
+
"learning_rate": 0.0008261815014518466,
|
| 389 |
+
"loss": 0.2249,
|
| 390 |
+
"step": 1300
|
| 391 |
+
},
|
| 392 |
+
{
|
| 393 |
+
"epoch": 1.06,
|
| 394 |
+
"grad_norm": 0.16907496750354767,
|
| 395 |
+
"learning_rate": 0.000817275336374625,
|
| 396 |
+
"loss": 0.2495,
|
| 397 |
+
"step": 1325
|
| 398 |
+
},
|
| 399 |
+
{
|
| 400 |
+
"epoch": 1.08,
|
| 401 |
+
"grad_norm": 0.14416393637657166,
|
| 402 |
+
"learning_rate": 0.0008081973611056783,
|
| 403 |
+
"loss": 0.2217,
|
| 404 |
+
"step": 1350
|
| 405 |
+
},
|
| 406 |
+
{
|
| 407 |
+
"epoch": 1.1,
|
| 408 |
+
"grad_norm": 0.16304358839988708,
|
| 409 |
+
"learning_rate": 0.0007989524915287594,
|
| 410 |
+
"loss": 0.2412,
|
| 411 |
+
"step": 1375
|
| 412 |
+
},
|
| 413 |
+
{
|
| 414 |
+
"epoch": 1.12,
|
| 415 |
+
"grad_norm": 0.17802534997463226,
|
| 416 |
+
"learning_rate": 0.000789545733903834,
|
| 417 |
+
"loss": 0.2371,
|
| 418 |
+
"step": 1400
|
| 419 |
+
},
|
| 420 |
+
{
|
| 421 |
+
"epoch": 1.1400000000000001,
|
| 422 |
+
"grad_norm": 0.1758696585893631,
|
| 423 |
+
"learning_rate": 0.0007799821821561011,
|
| 424 |
+
"loss": 0.2272,
|
| 425 |
+
"step": 1425
|
| 426 |
+
},
|
| 427 |
+
{
|
| 428 |
+
"epoch": 1.16,
|
| 429 |
+
"grad_norm": 0.17655503749847412,
|
| 430 |
+
"learning_rate": 0.0007702670151175434,
|
| 431 |
+
"loss": 0.2412,
|
| 432 |
+
"step": 1450
|
| 433 |
+
},
|
| 434 |
+
{
|
| 435 |
+
"epoch": 1.18,
|
| 436 |
+
"grad_norm": 0.2106119692325592,
|
| 437 |
+
"learning_rate": 0.0007604054937224989,
|
| 438 |
+
"loss": 0.2315,
|
| 439 |
+
"step": 1475
|
| 440 |
+
},
|
| 441 |
+
{
|
| 442 |
+
"epoch": 1.2,
|
| 443 |
+
"grad_norm": 0.17596380412578583,
|
| 444 |
+
"learning_rate": 0.00075040295815877,
|
| 445 |
+
"loss": 0.2372,
|
| 446 |
+
"step": 1500
|
| 447 |
+
},
|
| 448 |
+
{
|
| 449 |
+
"epoch": 1.2,
|
| 450 |
+
"eval_loss": 0.2413705736398697,
|
| 451 |
+
"eval_runtime": 18.735,
|
| 452 |
+
"eval_samples_per_second": 54.604,
|
| 453 |
+
"eval_steps_per_second": 0.854,
|
| 454 |
+
"step": 1500
|
| 455 |
+
},
|
| 456 |
+
{
|
| 457 |
+
"epoch": 1.22,
|
| 458 |
+
"grad_norm": 0.1379537284374237,
|
| 459 |
+
"learning_rate": 0.0007402648249758203,
|
| 460 |
+
"loss": 0.2322,
|
| 461 |
+
"step": 1525
|
| 462 |
+
},
|
| 463 |
+
{
|
| 464 |
+
"epoch": 1.24,
|
| 465 |
+
"grad_norm": 0.15587058663368225,
|
| 466 |
+
"learning_rate": 0.0007299965841516163,
|
| 467 |
+
"loss": 0.2261,
|
| 468 |
+
"step": 1550
|
| 469 |
+
},
|
| 470 |
+
{
|
| 471 |
+
"epoch": 1.26,
|
| 472 |
+
"grad_norm": 0.1809404343366623,
|
| 473 |
+
"learning_rate": 0.0007196037961197066,
|
| 474 |
+
"loss": 0.2333,
|
| 475 |
+
"step": 1575
|
| 476 |
+
},
|
| 477 |
+
{
|
| 478 |
+
"epoch": 1.28,
|
| 479 |
+
"grad_norm": 0.16045919060707092,
|
| 480 |
+
"learning_rate": 0.0007090920887581507,
|
| 481 |
+
"loss": 0.2377,
|
| 482 |
+
"step": 1600
|
| 483 |
+
},
|
| 484 |
+
{
|
| 485 |
+
"epoch": 1.3,
|
| 486 |
+
"grad_norm": 0.16197146475315094,
|
| 487 |
+
"learning_rate": 0.0006984671543419228,
|
| 488 |
+
"loss": 0.2207,
|
| 489 |
+
"step": 1625
|
| 490 |
+
},
|
| 491 |
+
{
|
| 492 |
+
"epoch": 1.32,
|
| 493 |
+
"grad_norm": 0.16172240674495697,
|
| 494 |
+
"learning_rate": 0.0006877347464604445,
|
| 495 |
+
"loss": 0.2433,
|
| 496 |
+
"step": 1650
|
| 497 |
+
},
|
| 498 |
+
{
|
| 499 |
+
"epoch": 1.34,
|
| 500 |
+
"grad_norm": 0.15120625495910645,
|
| 501 |
+
"learning_rate": 0.0006769006769019148,
|
| 502 |
+
"loss": 0.2253,
|
| 503 |
+
"step": 1675
|
| 504 |
+
},
|
| 505 |
+
{
|
| 506 |
+
"epoch": 1.3599999999999999,
|
| 507 |
+
"grad_norm": 0.1332470178604126,
|
| 508 |
+
"learning_rate": 0.0006659708125061241,
|
| 509 |
+
"loss": 0.2293,
|
| 510 |
+
"step": 1700
|
| 511 |
+
},
|
| 512 |
+
{
|
| 513 |
+
"epoch": 1.38,
|
| 514 |
+
"grad_norm": 0.1519145667552948,
|
| 515 |
+
"learning_rate": 0.0006549510719874577,
|
| 516 |
+
"loss": 0.2365,
|
| 517 |
+
"step": 1725
|
| 518 |
+
},
|
| 519 |
+
{
|
| 520 |
+
"epoch": 1.4,
|
| 521 |
+
"grad_norm": 0.16403517127037048,
|
| 522 |
+
"learning_rate": 0.0006438474227298065,
|
| 523 |
+
"loss": 0.2314,
|
| 524 |
+
"step": 1750
|
| 525 |
+
},
|
| 526 |
+
{
|
| 527 |
+
"epoch": 1.42,
|
| 528 |
+
"grad_norm": 0.1657528132200241,
|
| 529 |
+
"learning_rate": 0.0006326658775551235,
|
| 530 |
+
"loss": 0.2258,
|
| 531 |
+
"step": 1775
|
| 532 |
+
},
|
| 533 |
+
{
|
| 534 |
+
"epoch": 1.44,
|
| 535 |
+
"grad_norm": 0.1824166178703308,
|
| 536 |
+
"learning_rate": 0.0006214124914673755,
|
| 537 |
+
"loss": 0.2336,
|
| 538 |
+
"step": 1800
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"epoch": 1.46,
|
| 542 |
+
"grad_norm": 0.1564626395702362,
|
| 543 |
+
"learning_rate": 0.0006100933583736508,
|
| 544 |
+
"loss": 0.2333,
|
| 545 |
+
"step": 1825
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 1.48,
|
| 549 |
+
"grad_norm": 0.1635698676109314,
|
| 550 |
+
"learning_rate": 0.0005987146077842015,
|
| 551 |
+
"loss": 0.217,
|
| 552 |
+
"step": 1850
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 1.5,
|
| 556 |
+
"grad_norm": 0.17116394639015198,
|
| 557 |
+
"learning_rate": 0.000587282401493205,
|
| 558 |
+
"loss": 0.2293,
|
| 559 |
+
"step": 1875
|
| 560 |
+
},
|
| 561 |
+
{
|
| 562 |
+
"epoch": 1.52,
|
| 563 |
+
"grad_norm": 0.14937834441661835,
|
| 564 |
+
"learning_rate": 0.0005758029302420446,
|
| 565 |
+
"loss": 0.2232,
|
| 566 |
+
"step": 1900
|
| 567 |
+
},
|
| 568 |
+
{
|
| 569 |
+
"epoch": 1.54,
|
| 570 |
+
"grad_norm": 0.15280550718307495,
|
| 571 |
+
"learning_rate": 0.0005642824103669125,
|
| 572 |
+
"loss": 0.2267,
|
| 573 |
+
"step": 1925
|
| 574 |
+
},
|
| 575 |
+
{
|
| 576 |
+
"epoch": 1.56,
|
| 577 |
+
"grad_norm": 0.1290639340877533,
|
| 578 |
+
"learning_rate": 0.000552727080432556,
|
| 579 |
+
"loss": 0.2132,
|
| 580 |
+
"step": 1950
|
| 581 |
+
},
|
| 582 |
+
{
|
| 583 |
+
"epoch": 1.58,
|
| 584 |
+
"grad_norm": 0.1540573388338089,
|
| 585 |
+
"learning_rate": 0.0005411431978539828,
|
| 586 |
+
"loss": 0.2217,
|
| 587 |
+
"step": 1975
|
| 588 |
+
},
|
| 589 |
+
{
|
| 590 |
+
"epoch": 1.6,
|
| 591 |
+
"grad_norm": 0.15700837969779968,
|
| 592 |
+
"learning_rate": 0.0005295370355079614,
|
| 593 |
+
"loss": 0.2251,
|
| 594 |
+
"step": 2000
|
| 595 |
+
},
|
| 596 |
+
{
|
| 597 |
+
"epoch": 1.6,
|
| 598 |
+
"eval_loss": 0.228237584233284,
|
| 599 |
+
"eval_runtime": 18.7264,
|
| 600 |
+
"eval_samples_per_second": 54.629,
|
| 601 |
+
"eval_steps_per_second": 0.854,
|
| 602 |
+
"step": 2000
|
| 603 |
+
},
|
| 604 |
+
{
|
| 605 |
+
"epoch": 1.62,
|
| 606 |
+
"grad_norm": 0.15311604738235474,
|
| 607 |
+
"learning_rate": 0.0005179148783361473,
|
| 608 |
+
"loss": 0.2137,
|
| 609 |
+
"step": 2025
|
| 610 |
+
},
|
| 611 |
+
{
|
| 612 |
+
"epoch": 1.6400000000000001,
|
| 613 |
+
"grad_norm": 0.14809995889663696,
|
| 614 |
+
"learning_rate": 0.0005062830199416764,
|
| 615 |
+
"loss": 0.2225,
|
| 616 |
+
"step": 2050
|
| 617 |
+
},
|
| 618 |
+
{
|
| 619 |
+
"epoch": 1.6600000000000001,
|
| 620 |
+
"grad_norm": 0.14053231477737427,
|
| 621 |
+
"learning_rate": 0.0004946477591810677,
|
| 622 |
+
"loss": 0.2259,
|
| 623 |
+
"step": 2075
|
| 624 |
+
},
|
| 625 |
+
{
|
| 626 |
+
"epoch": 1.6800000000000002,
|
| 627 |
+
"grad_norm": 0.15467585623264313,
|
| 628 |
+
"learning_rate": 0.00048301539675328197,
|
| 629 |
+
"loss": 0.2273,
|
| 630 |
+
"step": 2100
|
| 631 |
+
},
|
| 632 |
+
{
|
| 633 |
+
"epoch": 1.7,
|
| 634 |
+
"grad_norm": 0.17832106351852417,
|
| 635 |
+
"learning_rate": 0.000471392231787781,
|
| 636 |
+
"loss": 0.2267,
|
| 637 |
+
"step": 2125
|
| 638 |
+
},
|
| 639 |
+
{
|
| 640 |
+
"epoch": 1.72,
|
| 641 |
+
"grad_norm": 0.14658208191394806,
|
| 642 |
+
"learning_rate": 0.0004597845584334386,
|
| 643 |
+
"loss": 0.2214,
|
| 644 |
+
"step": 2150
|
| 645 |
+
},
|
| 646 |
+
{
|
| 647 |
+
"epoch": 1.74,
|
| 648 |
+
"grad_norm": 0.13484394550323486,
|
| 649 |
+
"learning_rate": 0.00044819866245014836,
|
| 650 |
+
"loss": 0.2161,
|
| 651 |
+
"step": 2175
|
| 652 |
+
},
|
| 653 |
+
{
|
| 654 |
+
"epoch": 1.76,
|
| 655 |
+
"grad_norm": 0.1637551337480545,
|
| 656 |
+
"learning_rate": 0.0004366408178049728,
|
| 657 |
+
"loss": 0.2199,
|
| 658 |
+
"step": 2200
|
| 659 |
+
},
|
| 660 |
+
{
|
| 661 |
+
"epoch": 1.78,
|
| 662 |
+
"grad_norm": 0.13961051404476166,
|
| 663 |
+
"learning_rate": 0.0004251172832746799,
|
| 664 |
+
"loss": 0.2242,
|
| 665 |
+
"step": 2225
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 1.8,
|
| 669 |
+
"grad_norm": 0.13440121710300446,
|
| 670 |
+
"learning_rate": 0.00041363429905650543,
|
| 671 |
+
"loss": 0.2184,
|
| 672 |
+
"step": 2250
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 1.8199999999999998,
|
| 676 |
+
"grad_norm": 0.14450828731060028,
|
| 677 |
+
"learning_rate": 0.00040219808338897594,
|
| 678 |
+
"loss": 0.2141,
|
| 679 |
+
"step": 2275
|
| 680 |
+
},
|
| 681 |
+
{
|
| 682 |
+
"epoch": 1.8399999999999999,
|
| 683 |
+
"grad_norm": 0.14556901156902313,
|
| 684 |
+
"learning_rate": 0.0003908148291846224,
|
| 685 |
+
"loss": 0.2084,
|
| 686 |
+
"step": 2300
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 1.8599999999999999,
|
| 690 |
+
"grad_norm": 0.14280621707439423,
|
| 691 |
+
"learning_rate": 0.00037949070067640927,
|
| 692 |
+
"loss": 0.2207,
|
| 693 |
+
"step": 2325
|
| 694 |
+
},
|
| 695 |
+
{
|
| 696 |
+
"epoch": 1.88,
|
| 697 |
+
"grad_norm": 0.1389743834733963,
|
| 698 |
+
"learning_rate": 0.00036823183007969373,
|
| 699 |
+
"loss": 0.207,
|
| 700 |
+
"step": 2350
|
| 701 |
+
},
|
| 702 |
+
{
|
| 703 |
+
"epoch": 1.9,
|
| 704 |
+
"grad_norm": 0.1533161848783493,
|
| 705 |
+
"learning_rate": 0.00035704431427152237,
|
| 706 |
+
"loss": 0.2134,
|
| 707 |
+
"step": 2375
|
| 708 |
+
},
|
| 709 |
+
{
|
| 710 |
+
"epoch": 1.92,
|
| 711 |
+
"grad_norm": 0.13798296451568604,
|
| 712 |
+
"learning_rate": 0.00034593421148906525,
|
| 713 |
+
"loss": 0.2102,
|
| 714 |
+
"step": 2400
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 1.94,
|
| 718 |
+
"grad_norm": 0.14728130400180817,
|
| 719 |
+
"learning_rate": 0.0003349075380489731,
|
| 720 |
+
"loss": 0.2132,
|
| 721 |
+
"step": 2425
|
| 722 |
+
},
|
| 723 |
+
{
|
| 724 |
+
"epoch": 1.96,
|
| 725 |
+
"grad_norm": 0.1491023153066635,
|
| 726 |
+
"learning_rate": 0.0003239702650894364,
|
| 727 |
+
"loss": 0.2004,
|
| 728 |
+
"step": 2450
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 1.98,
|
| 732 |
+
"grad_norm": 0.1351625770330429,
|
| 733 |
+
"learning_rate": 0.00031312831533670773,
|
| 734 |
+
"loss": 0.2103,
|
| 735 |
+
"step": 2475
|
| 736 |
+
},
|
| 737 |
+
{
|
| 738 |
+
"epoch": 2.0,
|
| 739 |
+
"grad_norm": 0.11963696032762527,
|
| 740 |
+
"learning_rate": 0.0003023875598978419,
|
| 741 |
+
"loss": 0.2109,
|
| 742 |
+
"step": 2500
|
| 743 |
+
},
|
| 744 |
+
{
|
| 745 |
+
"epoch": 2.0,
|
| 746 |
+
"eval_loss": 0.21772204339504242,
|
| 747 |
+
"eval_runtime": 18.7134,
|
| 748 |
+
"eval_samples_per_second": 54.667,
|
| 749 |
+
"eval_steps_per_second": 0.855,
|
| 750 |
+
"step": 2500
|
| 751 |
+
},
|
| 752 |
+
{
|
| 753 |
+
"epoch": 2.02,
|
| 754 |
+
"grad_norm": 0.14547832310199738,
|
| 755 |
+
"learning_rate": 0.0002917538150813876,
|
| 756 |
+
"loss": 0.181,
|
| 757 |
+
"step": 2525
|
| 758 |
+
},
|
| 759 |
+
{
|
| 760 |
+
"epoch": 2.04,
|
| 761 |
+
"grad_norm": 0.1392798125743866,
|
| 762 |
+
"learning_rate": 0.00028123283924775356,
|
| 763 |
+
"loss": 0.1846,
|
| 764 |
+
"step": 2550
|
| 765 |
+
},
|
| 766 |
+
{
|
| 767 |
+
"epoch": 2.06,
|
| 768 |
+
"grad_norm": 0.14942820370197296,
|
| 769 |
+
"learning_rate": 0.00027083032969095505,
|
| 770 |
+
"loss": 0.1882,
|
| 771 |
+
"step": 2575
|
| 772 |
+
},
|
| 773 |
+
{
|
| 774 |
+
"epoch": 2.08,
|
| 775 |
+
"grad_norm": 0.14887529611587524,
|
| 776 |
+
"learning_rate": 0.0002605519195534288,
|
| 777 |
+
"loss": 0.1877,
|
| 778 |
+
"step": 2600
|
| 779 |
+
},
|
| 780 |
+
{
|
| 781 |
+
"epoch": 2.1,
|
| 782 |
+
"grad_norm": 0.148512601852417,
|
| 783 |
+
"learning_rate": 0.00025040317477558616,
|
| 784 |
+
"loss": 0.1909,
|
| 785 |
+
"step": 2625
|
| 786 |
+
},
|
| 787 |
+
{
|
| 788 |
+
"epoch": 2.12,
|
| 789 |
+
"grad_norm": 0.16728661954402924,
|
| 790 |
+
"learning_rate": 0.00024038959108175928,
|
| 791 |
+
"loss": 0.1866,
|
| 792 |
+
"step": 2650
|
| 793 |
+
},
|
| 794 |
+
{
|
| 795 |
+
"epoch": 2.14,
|
| 796 |
+
"grad_norm": 0.1289970874786377,
|
| 797 |
+
"learning_rate": 0.0002305165910041689,
|
| 798 |
+
"loss": 0.1797,
|
| 799 |
+
"step": 2675
|
| 800 |
+
},
|
| 801 |
+
{
|
| 802 |
+
"epoch": 2.16,
|
| 803 |
+
"grad_norm": 0.1591159850358963,
|
| 804 |
+
"learning_rate": 0.00022078952094652704,
|
| 805 |
+
"loss": 0.1877,
|
| 806 |
+
"step": 2700
|
| 807 |
+
},
|
| 808 |
+
{
|
| 809 |
+
"epoch": 2.18,
|
| 810 |
+
"grad_norm": 0.12650765478610992,
|
| 811 |
+
"learning_rate": 0.0002112136482888663,
|
| 812 |
+
"loss": 0.1859,
|
| 813 |
+
"step": 2725
|
| 814 |
+
},
|
| 815 |
+
{
|
| 816 |
+
"epoch": 2.2,
|
| 817 |
+
"grad_norm": 0.14034953713417053,
|
| 818 |
+
"learning_rate": 0.00020179415853515908,
|
| 819 |
+
"loss": 0.18,
|
| 820 |
+
"step": 2750
|
| 821 |
+
},
|
| 822 |
+
{
|
| 823 |
+
"epoch": 2.22,
|
| 824 |
+
"grad_norm": 0.17186295986175537,
|
| 825 |
+
"learning_rate": 0.00019253615250527738,
|
| 826 |
+
"loss": 0.187,
|
| 827 |
+
"step": 2775
|
| 828 |
+
},
|
| 829 |
+
{
|
| 830 |
+
"epoch": 2.24,
|
| 831 |
+
"grad_norm": 0.13548125326633453,
|
| 832 |
+
"learning_rate": 0.00018344464357280722,
|
| 833 |
+
"loss": 0.1874,
|
| 834 |
+
"step": 2800
|
| 835 |
+
},
|
| 836 |
+
{
|
| 837 |
+
"epoch": 2.26,
|
| 838 |
+
"grad_norm": 0.1618693470954895,
|
| 839 |
+
"learning_rate": 0.00017452455495021963,
|
| 840 |
+
"loss": 0.1877,
|
| 841 |
+
"step": 2825
|
| 842 |
+
},
|
| 843 |
+
{
|
| 844 |
+
"epoch": 2.2800000000000002,
|
| 845 |
+
"grad_norm": 0.15758028626441956,
|
| 846 |
+
"learning_rate": 0.00016578071702286396,
|
| 847 |
+
"loss": 0.1822,
|
| 848 |
+
"step": 2850
|
| 849 |
+
},
|
| 850 |
+
{
|
| 851 |
+
"epoch": 2.3,
|
| 852 |
+
"grad_norm": 0.15973731875419617,
|
| 853 |
+
"learning_rate": 0.00015721786473322824,
|
| 854 |
+
"loss": 0.1849,
|
| 855 |
+
"step": 2875
|
| 856 |
+
},
|
| 857 |
+
{
|
| 858 |
+
"epoch": 2.32,
|
| 859 |
+
"grad_norm": 0.14636465907096863,
|
| 860 |
+
"learning_rate": 0.0001488406350168854,
|
| 861 |
+
"loss": 0.1916,
|
| 862 |
+
"step": 2900
|
| 863 |
+
},
|
| 864 |
+
{
|
| 865 |
+
"epoch": 2.34,
|
| 866 |
+
"grad_norm": 0.1348065584897995,
|
| 867 |
+
"learning_rate": 0.00014065356429151033,
|
| 868 |
+
"loss": 0.191,
|
| 869 |
+
"step": 2925
|
| 870 |
+
},
|
| 871 |
+
{
|
| 872 |
+
"epoch": 2.36,
|
| 873 |
+
"grad_norm": 0.1236218586564064,
|
| 874 |
+
"learning_rate": 0.0001326610860003293,
|
| 875 |
+
"loss": 0.1871,
|
| 876 |
+
"step": 2950
|
| 877 |
+
},
|
| 878 |
+
{
|
| 879 |
+
"epoch": 2.38,
|
| 880 |
+
"grad_norm": 0.14592643082141876,
|
| 881 |
+
"learning_rate": 0.00012486752821133312,
|
| 882 |
+
"loss": 0.1815,
|
| 883 |
+
"step": 2975
|
| 884 |
+
},
|
| 885 |
+
{
|
| 886 |
+
"epoch": 2.4,
|
| 887 |
+
"grad_norm": 0.1528843492269516,
|
| 888 |
+
"learning_rate": 0.00011727711127355117,
|
| 889 |
+
"loss": 0.1805,
|
| 890 |
+
"step": 3000
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 2.4,
|
| 894 |
+
"eval_loss": 0.21503186225891113,
|
| 895 |
+
"eval_runtime": 18.7554,
|
| 896 |
+
"eval_samples_per_second": 54.544,
|
| 897 |
+
"eval_steps_per_second": 0.853,
|
| 898 |
+
"step": 3000
|
| 899 |
+
},
|
| 900 |
+
{
|
| 901 |
+
"epoch": 2.42,
|
| 902 |
+
"grad_norm": 0.14162999391555786,
|
| 903 |
+
"learning_rate": 0.00010989394553165833,
|
| 904 |
+
"loss": 0.1846,
|
| 905 |
+
"step": 3025
|
| 906 |
+
},
|
| 907 |
+
{
|
| 908 |
+
"epoch": 2.44,
|
| 909 |
+
"grad_norm": 0.13103316724300385,
|
| 910 |
+
"learning_rate": 0.00010272202910015082,
|
| 911 |
+
"loss": 0.177,
|
| 912 |
+
"step": 3050
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 2.46,
|
| 916 |
+
"grad_norm": 0.143747016787529,
|
| 917 |
+
"learning_rate": 9.576524569829692e-05,
|
| 918 |
+
"loss": 0.1874,
|
| 919 |
+
"step": 3075
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 2.48,
|
| 923 |
+
"grad_norm": 0.12383788824081421,
|
| 924 |
+
"learning_rate": 8.902736254703348e-05,
|
| 925 |
+
"loss": 0.1875,
|
| 926 |
+
"step": 3100
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 2.5,
|
| 930 |
+
"grad_norm": 0.14952044188976288,
|
| 931 |
+
"learning_rate": 8.251202832895066e-05,
|
| 932 |
+
"loss": 0.1787,
|
| 933 |
+
"step": 3125
|
| 934 |
+
},
|
| 935 |
+
{
|
| 936 |
+
"epoch": 2.52,
|
| 937 |
+
"grad_norm": 0.12611518800258636,
|
| 938 |
+
"learning_rate": 7.622277121246512e-05,
|
| 939 |
+
"loss": 0.1828,
|
| 940 |
+
"step": 3150
|
| 941 |
+
},
|
| 942 |
+
{
|
| 943 |
+
"epoch": 2.54,
|
| 944 |
+
"grad_norm": 0.1494569331407547,
|
| 945 |
+
"learning_rate": 7.01629969412545e-05,
|
| 946 |
+
"loss": 0.1895,
|
| 947 |
+
"step": 3175
|
| 948 |
+
},
|
| 949 |
+
{
|
| 950 |
+
"epoch": 2.56,
|
| 951 |
+
"grad_norm": 0.12987783551216125,
|
| 952 |
+
"learning_rate": 6.433598698998766e-05,
|
| 953 |
+
"loss": 0.1814,
|
| 954 |
+
"step": 3200
|
| 955 |
+
},
|
| 956 |
+
{
|
| 957 |
+
"epoch": 2.58,
|
| 958 |
+
"grad_norm": 0.15549467504024506,
|
| 959 |
+
"learning_rate": 5.874489678734812e-05,
|
| 960 |
+
"loss": 0.1775,
|
| 961 |
+
"step": 3225
|
| 962 |
+
},
|
| 963 |
+
{
|
| 964 |
+
"epoch": 2.6,
|
| 965 |
+
"grad_norm": 0.1471382975578308,
|
| 966 |
+
"learning_rate": 5.3392754007313305e-05,
|
| 967 |
+
"loss": 0.1889,
|
| 968 |
+
"step": 3250
|
| 969 |
+
},
|
| 970 |
+
{
|
| 971 |
+
"epoch": 2.62,
|
| 972 |
+
"grad_norm": 0.14564456045627594,
|
| 973 |
+
"learning_rate": 4.828245692961608e-05,
|
| 974 |
+
"loss": 0.1718,
|
| 975 |
+
"step": 3275
|
| 976 |
+
},
|
| 977 |
+
{
|
| 978 |
+
"epoch": 2.64,
|
| 979 |
+
"grad_norm": 0.14467193186283112,
|
| 980 |
+
"learning_rate": 4.341677287027529e-05,
|
| 981 |
+
"loss": 0.1777,
|
| 982 |
+
"step": 3300
|
| 983 |
+
},
|
| 984 |
+
{
|
| 985 |
+
"epoch": 2.66,
|
| 986 |
+
"grad_norm": 0.15905599296092987,
|
| 987 |
+
"learning_rate": 3.8798336683045054e-05,
|
| 988 |
+
"loss": 0.187,
|
| 989 |
+
"step": 3325
|
| 990 |
+
},
|
| 991 |
+
{
|
| 992 |
+
"epoch": 2.68,
|
| 993 |
+
"grad_norm": 0.15432722866535187,
|
| 994 |
+
"learning_rate": 3.442964933259474e-05,
|
| 995 |
+
"loss": 0.1748,
|
| 996 |
+
"step": 3350
|
| 997 |
+
},
|
| 998 |
+
{
|
| 999 |
+
"epoch": 2.7,
|
| 1000 |
+
"grad_norm": 0.162972554564476,
|
| 1001 |
+
"learning_rate": 3.0313076540192996e-05,
|
| 1002 |
+
"loss": 0.1843,
|
| 1003 |
+
"step": 3375
|
| 1004 |
+
},
|
| 1005 |
+
{
|
| 1006 |
+
"epoch": 2.7199999999999998,
|
| 1007 |
+
"grad_norm": 0.14356060326099396,
|
| 1008 |
+
"learning_rate": 2.6450847502627883e-05,
|
| 1009 |
+
"loss": 0.1826,
|
| 1010 |
+
"step": 3400
|
| 1011 |
+
},
|
| 1012 |
+
{
|
| 1013 |
+
"epoch": 2.74,
|
| 1014 |
+
"grad_norm": 0.1465945690870285,
|
| 1015 |
+
"learning_rate": 2.2845053685056904e-05,
|
| 1016 |
+
"loss": 0.1853,
|
| 1017 |
+
"step": 3425
|
| 1018 |
+
},
|
| 1019 |
+
{
|
| 1020 |
+
"epoch": 2.76,
|
| 1021 |
+
"grad_norm": 0.13656504452228546,
|
| 1022 |
+
"learning_rate": 1.9497647688442477e-05,
|
| 1023 |
+
"loss": 0.186,
|
| 1024 |
+
"step": 3450
|
| 1025 |
+
},
|
| 1026 |
+
{
|
| 1027 |
+
"epoch": 2.7800000000000002,
|
| 1028 |
+
"grad_norm": 0.1474902331829071,
|
| 1029 |
+
"learning_rate": 1.6410442192183573e-05,
|
| 1030 |
+
"loss": 0.1848,
|
| 1031 |
+
"step": 3475
|
| 1032 |
+
},
|
| 1033 |
+
{
|
| 1034 |
+
"epoch": 2.8,
|
| 1035 |
+
"grad_norm": 0.13733918964862823,
|
| 1036 |
+
"learning_rate": 1.3585108972518078e-05,
|
| 1037 |
+
"loss": 0.1842,
|
| 1038 |
+
"step": 3500
|
| 1039 |
+
},
|
| 1040 |
+
{
|
| 1041 |
+
"epoch": 2.8,
|
| 1042 |
+
"eval_loss": 0.21268269419670105,
|
| 1043 |
+
"eval_runtime": 18.7242,
|
| 1044 |
+
"eval_samples_per_second": 54.635,
|
| 1045 |
+
"eval_steps_per_second": 0.855,
|
| 1046 |
+
"step": 3500
|
| 1047 |
+
},
|
| 1048 |
+
{
|
| 1049 |
+
"epoch": 2.82,
|
| 1050 |
+
"grad_norm": 0.12263137847185135,
|
| 1051 |
+
"learning_rate": 1.1023177997226297e-05,
|
| 1052 |
+
"loss": 0.1788,
|
| 1053 |
+
"step": 3525
|
| 1054 |
+
},
|
| 1055 |
+
{
|
| 1056 |
+
"epoch": 2.84,
|
| 1057 |
+
"grad_norm": 0.15410158038139343,
|
| 1058 |
+
"learning_rate": 8.72603659712662e-06,
|
| 1059 |
+
"loss": 0.1841,
|
| 1060 |
+
"step": 3550
|
| 1061 |
+
},
|
| 1062 |
+
{
|
| 1063 |
+
"epoch": 2.86,
|
| 1064 |
+
"grad_norm": 0.12600606679916382,
|
| 1065 |
+
"learning_rate": 6.694928714811255e-06,
|
| 1066 |
+
"loss": 0.1814,
|
| 1067 |
+
"step": 3575
|
| 1068 |
+
},
|
| 1069 |
+
{
|
| 1070 |
+
"epoch": 2.88,
|
| 1071 |
+
"grad_norm": 0.152796670794487,
|
| 1072 |
+
"learning_rate": 4.93095423102935e-06,
|
| 1073 |
+
"loss": 0.1837,
|
| 1074 |
+
"step": 3600
|
| 1075 |
+
},
|
| 1076 |
+
{
|
| 1077 |
+
"epoch": 2.9,
|
| 1078 |
+
"grad_norm": 0.1589384227991104,
|
| 1079 |
+
"learning_rate": 3.435068369082306e-06,
|
| 1080 |
+
"loss": 0.1793,
|
| 1081 |
+
"step": 3625
|
| 1082 |
+
},
|
| 1083 |
+
{
|
| 1084 |
+
"epoch": 2.92,
|
| 1085 |
+
"grad_norm": 0.11970392614603043,
|
| 1086 |
+
"learning_rate": 2.2080811775535003e-06,
|
| 1087 |
+
"loss": 0.1903,
|
| 1088 |
+
"step": 3650
|
| 1089 |
+
},
|
| 1090 |
+
{
|
| 1091 |
+
"epoch": 2.94,
|
| 1092 |
+
"grad_norm": 0.14448143541812897,
|
| 1093 |
+
"learning_rate": 1.2506570916523408e-06,
|
| 1094 |
+
"loss": 0.1809,
|
| 1095 |
+
"step": 3675
|
| 1096 |
+
},
|
| 1097 |
+
{
|
| 1098 |
+
"epoch": 2.96,
|
| 1099 |
+
"grad_norm": 0.1562301218509674,
|
| 1100 |
+
"learning_rate": 5.633145734114664e-07,
|
| 1101 |
+
"loss": 0.1858,
|
| 1102 |
+
"step": 3700
|
| 1103 |
+
},
|
| 1104 |
+
{
|
| 1105 |
+
"epoch": 2.98,
|
| 1106 |
+
"grad_norm": 0.141978919506073,
|
| 1107 |
+
"learning_rate": 1.4642583092999705e-07,
|
| 1108 |
+
"loss": 0.1818,
|
| 1109 |
+
"step": 3725
|
| 1110 |
+
},
|
| 1111 |
+
{
|
| 1112 |
+
"epoch": 3.0,
|
| 1113 |
+
"grad_norm": 0.1364634484052658,
|
| 1114 |
+
"learning_rate": 2.166168162065496e-10,
|
| 1115 |
+
"loss": 0.1815,
|
| 1116 |
+
"step": 3750
|
| 1117 |
+
},
|
| 1118 |
+
{
|
| 1119 |
+
"epoch": 3.0,
|
| 1120 |
+
"step": 3750,
|
| 1121 |
+
"total_flos": 2.43882352705536e+18,
|
| 1122 |
+
"train_loss": 0.24297360801696777,
|
| 1123 |
+
"train_runtime": 3674.409,
|
| 1124 |
+
"train_samples_per_second": 32.658,
|
| 1125 |
+
"train_steps_per_second": 1.021
|
| 1126 |
+
}
|
| 1127 |
+
],
|
| 1128 |
+
"logging_steps": 25,
|
| 1129 |
+
"max_steps": 3750,
|
| 1130 |
+
"num_input_tokens_seen": 0,
|
| 1131 |
+
"num_train_epochs": 3,
|
| 1132 |
+
"save_steps": 0,
|
| 1133 |
+
"stateful_callbacks": {
|
| 1134 |
+
"TrainerControl": {
|
| 1135 |
+
"args": {
|
| 1136 |
+
"should_epoch_stop": false,
|
| 1137 |
+
"should_evaluate": false,
|
| 1138 |
+
"should_log": false,
|
| 1139 |
+
"should_save": true,
|
| 1140 |
+
"should_training_stop": true
|
| 1141 |
+
},
|
| 1142 |
+
"attributes": {}
|
| 1143 |
+
}
|
| 1144 |
+
},
|
| 1145 |
+
"total_flos": 2.43882352705536e+18,
|
| 1146 |
+
"train_batch_size": 32,
|
| 1147 |
+
"trial_name": null,
|
| 1148 |
+
"trial_params": null
|
| 1149 |
+
}
|