Batch upload part 4
Browse files- nl_tasks/exps/run_ex08/ft/adapter_config.json +18 -0
- nl_tasks/exps/run_ex08/ft/added_tokens.json +3 -0
- nl_tasks/exps/run_ex08/ft/special_tokens_map.json +30 -0
- nl_tasks/exps/run_ex08/ft/tokenizer.json +0 -0
- nl_tasks/exps/run_ex08/ft/tokenizer.model +3 -0
- nl_tasks/exps/run_ex08/ft/tokenizer_config.json +51 -0
- nl_tasks/exps/run_ex08/ft2/adapter_config.json +18 -0
- nl_tasks/exps/run_ex08/ft2/adapter_model.bin +3 -0
- nl_tasks/exps/run_ex08/trainer_state.json +914 -0
- nl_tasks/exps/run_ex09/ft/adapter_config.json +18 -0
- nl_tasks/exps/run_ex09/ft/added_tokens.json +3 -0
- nl_tasks/exps/run_ex09/ft/special_tokens_map.json +30 -0
- nl_tasks/exps/run_ex09/ft/tokenizer.json +0 -0
- nl_tasks/exps/run_ex09/ft/tokenizer.model +3 -0
- nl_tasks/exps/run_ex09/ft/tokenizer_config.json +51 -0
- nl_tasks/exps/run_ex09/ft2/adapter_config.json +18 -0
- nl_tasks/exps/run_ex09/ft2/adapter_model.bin +3 -0
- nl_tasks/exps/run_ex09/trainer_state.json +914 -0
- nl_tasks/exps/run_ex10/ft/adapter_config.json +18 -0
- nl_tasks/exps/run_ex10/ft/added_tokens.json +3 -0
- nl_tasks/exps/run_ex10/ft/special_tokens_map.json +30 -0
- nl_tasks/exps/run_ex10/ft/tokenizer.json +0 -0
- nl_tasks/exps/run_ex10/ft/tokenizer.model +3 -0
- nl_tasks/exps/run_ex10/ft/tokenizer_config.json +51 -0
- nl_tasks/exps/run_ex10/ft2/adapter_config.json +18 -0
- nl_tasks/exps/run_ex10/ft2/adapter_model.bin +3 -0
- nl_tasks/exps/run_ex10/trainer_state.json +914 -0
- nl_tasks/exps/run_ex11/ft/adapter_config.json +18 -0
- nl_tasks/exps/run_ex11/ft/added_tokens.json +3 -0
- nl_tasks/exps/run_ex11/ft/special_tokens_map.json +30 -0
- nl_tasks/exps/run_ex11/ft/tokenizer.json +0 -0
- nl_tasks/exps/run_ex11/ft/tokenizer.model +3 -0
- nl_tasks/exps/run_ex11/ft/tokenizer_config.json +51 -0
- nl_tasks/exps/run_ex11/ft2/adapter_config.json +18 -0
- nl_tasks/exps/run_ex11/ft2/adapter_model.bin +3 -0
- nl_tasks/exps/run_ex11/trainer_state.json +914 -0
- nl_tasks/exps/run_ex12/ft/added_tokens.json +3 -0
- nl_tasks/exps/run_ex12/ft/special_tokens_map.json +30 -0
- nl_tasks/exps/run_ex12/ft/tokenizer_config.json +51 -0
- nl_tasks/exps/run_ex12/trainer_state.json +914 -0
- nl_tasks/exps/run_ex18/ft2/adapter_config.json +18 -0
- nl_tasks/exps/run_exps7/ft/adapter_config.json +18 -0
- nl_tasks/exps/run_exps7/ft/added_tokens.json +3 -0
- nl_tasks/exps/run_exps7/ft/special_tokens_map.json +30 -0
- nl_tasks/exps/run_exps7/ft/tokenizer.json +0 -0
- nl_tasks/exps/run_exps7/ft/tokenizer.model +3 -0
- nl_tasks/exps/run_exps7/ft/tokenizer_config.json +51 -0
- nl_tasks/exps/run_exps7/ft2/adapter_config.json +18 -0
- nl_tasks/exps/run_exps7/ft2/adapter_model.bin +3 -0
- nl_tasks/exps/run_exps7/trainer_state.json +914 -0
nl_tasks/exps/run_ex08/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_ex08/ft/added_tokens.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"[PAD]": 32000
|
| 3 |
+
}
|
nl_tasks/exps/run_ex08/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_ex08/ft/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
nl_tasks/exps/run_ex08/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_ex08/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_ex08/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_ex08/ft2/adapter_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:32b44881f924f323595ebd29d2c231d9f1a8aff2440a568ca8d1d09a5a9daa50
|
| 3 |
+
size 33602915
|
nl_tasks/exps/run_ex08/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": 0.1558552384376526,
|
| 15 |
+
"learning_rate": 0.00019672131147540983,
|
| 16 |
+
"loss": 0.6261,
|
| 17 |
+
"step": 25
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.04101722723543889,
|
| 21 |
+
"grad_norm": 0.22824394702911377,
|
| 22 |
+
"learning_rate": 0.00040163934426229507,
|
| 23 |
+
"loss": 0.3851,
|
| 24 |
+
"step": 50
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.06152584085315833,
|
| 28 |
+
"grad_norm": 0.15604457259178162,
|
| 29 |
+
"learning_rate": 0.0006065573770491804,
|
| 30 |
+
"loss": 0.3599,
|
| 31 |
+
"step": 75
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.08203445447087777,
|
| 35 |
+
"grad_norm": 0.17529062926769257,
|
| 36 |
+
"learning_rate": 0.0008114754098360656,
|
| 37 |
+
"loss": 0.3281,
|
| 38 |
+
"step": 100
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.08203445447087777,
|
| 42 |
+
"eval_loss": 0.3204536736011505,
|
| 43 |
+
"eval_runtime": 19.5187,
|
| 44 |
+
"eval_samples_per_second": 51.233,
|
| 45 |
+
"eval_steps_per_second": 0.82,
|
| 46 |
+
"step": 100
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"epoch": 0.10254306808859721,
|
| 50 |
+
"grad_norm": 0.2388022094964981,
|
| 51 |
+
"learning_rate": 0.0010163934426229509,
|
| 52 |
+
"loss": 0.3386,
|
| 53 |
+
"step": 125
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 0.12305168170631665,
|
| 57 |
+
"grad_norm": 0.3735959529876709,
|
| 58 |
+
"learning_rate": 0.001221311475409836,
|
| 59 |
+
"loss": 0.3296,
|
| 60 |
+
"step": 150
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 0.1435602953240361,
|
| 64 |
+
"grad_norm": 0.2982408106327057,
|
| 65 |
+
"learning_rate": 0.0014262295081967215,
|
| 66 |
+
"loss": 0.3274,
|
| 67 |
+
"step": 175
|
| 68 |
+
},
|
| 69 |
+
{
|
| 70 |
+
"epoch": 0.16406890894175555,
|
| 71 |
+
"grad_norm": 0.46184009313583374,
|
| 72 |
+
"learning_rate": 0.0016311475409836065,
|
| 73 |
+
"loss": 0.3522,
|
| 74 |
+
"step": 200
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.16406890894175555,
|
| 78 |
+
"eval_loss": 0.33685189485549927,
|
| 79 |
+
"eval_runtime": 19.3745,
|
| 80 |
+
"eval_samples_per_second": 51.614,
|
| 81 |
+
"eval_steps_per_second": 0.826,
|
| 82 |
+
"step": 200
|
| 83 |
+
},
|
| 84 |
+
{
|
| 85 |
+
"epoch": 0.184577522559475,
|
| 86 |
+
"grad_norm": 0.3153756558895111,
|
| 87 |
+
"learning_rate": 0.0018360655737704918,
|
| 88 |
+
"loss": 0.3548,
|
| 89 |
+
"step": 225
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 0.20508613617719443,
|
| 93 |
+
"grad_norm": 0.3366014063358307,
|
| 94 |
+
"learning_rate": 0.0019999743708232127,
|
| 95 |
+
"loss": 0.3323,
|
| 96 |
+
"step": 250
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"epoch": 0.22559474979491387,
|
| 100 |
+
"grad_norm": 0.27126145362854004,
|
| 101 |
+
"learning_rate": 0.0019990774875676054,
|
| 102 |
+
"loss": 0.3353,
|
| 103 |
+
"step": 275
|
| 104 |
+
},
|
| 105 |
+
{
|
| 106 |
+
"epoch": 0.2461033634126333,
|
| 107 |
+
"grad_norm": 0.302554190158844,
|
| 108 |
+
"learning_rate": 0.001996900458879386,
|
| 109 |
+
"loss": 0.3289,
|
| 110 |
+
"step": 300
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.2461033634126333,
|
| 114 |
+
"eval_loss": 0.3138977587223053,
|
| 115 |
+
"eval_runtime": 19.3554,
|
| 116 |
+
"eval_samples_per_second": 51.665,
|
| 117 |
+
"eval_steps_per_second": 0.827,
|
| 118 |
+
"step": 300
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 0.2666119770303528,
|
| 122 |
+
"grad_norm": 0.5873791575431824,
|
| 123 |
+
"learning_rate": 0.001993446074245224,
|
| 124 |
+
"loss": 0.3284,
|
| 125 |
+
"step": 325
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 0.2871205906480722,
|
| 129 |
+
"grad_norm": 0.2076379358768463,
|
| 130 |
+
"learning_rate": 0.0019887187598630527,
|
| 131 |
+
"loss": 0.3125,
|
| 132 |
+
"step": 350
|
| 133 |
+
},
|
| 134 |
+
{
|
| 135 |
+
"epoch": 0.30762920426579166,
|
| 136 |
+
"grad_norm": 0.21529337763786316,
|
| 137 |
+
"learning_rate": 0.0019827245729706648,
|
| 138 |
+
"loss": 0.3206,
|
| 139 |
+
"step": 375
|
| 140 |
+
},
|
| 141 |
+
{
|
| 142 |
+
"epoch": 0.3281378178835111,
|
| 143 |
+
"grad_norm": 0.24433602392673492,
|
| 144 |
+
"learning_rate": 0.0019754711940844047,
|
| 145 |
+
"loss": 0.2986,
|
| 146 |
+
"step": 400
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.3281378178835111,
|
| 150 |
+
"eval_loss": 0.28526049852371216,
|
| 151 |
+
"eval_runtime": 19.3517,
|
| 152 |
+
"eval_samples_per_second": 51.675,
|
| 153 |
+
"eval_steps_per_second": 0.827,
|
| 154 |
+
"step": 400
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 0.34864643150123054,
|
| 158 |
+
"grad_norm": 0.18738172948360443,
|
| 159 |
+
"learning_rate": 0.0019669679171579117,
|
| 160 |
+
"loss": 0.3028,
|
| 161 |
+
"step": 425
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 0.36915504511895,
|
| 165 |
+
"grad_norm": 0.16099169850349426,
|
| 166 |
+
"learning_rate": 0.001957225637673524,
|
| 167 |
+
"loss": 0.2977,
|
| 168 |
+
"step": 450
|
| 169 |
+
},
|
| 170 |
+
{
|
| 171 |
+
"epoch": 0.3896636587366694,
|
| 172 |
+
"grad_norm": 0.26189756393432617,
|
| 173 |
+
"learning_rate": 0.0019462568386815961,
|
| 174 |
+
"loss": 0.2846,
|
| 175 |
+
"step": 475
|
| 176 |
+
},
|
| 177 |
+
{
|
| 178 |
+
"epoch": 0.41017227235438886,
|
| 179 |
+
"grad_norm": 0.13307002186775208,
|
| 180 |
+
"learning_rate": 0.0019340755748056234,
|
| 181 |
+
"loss": 0.2676,
|
| 182 |
+
"step": 500
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.41017227235438886,
|
| 186 |
+
"eval_loss": 0.2698473036289215,
|
| 187 |
+
"eval_runtime": 19.3722,
|
| 188 |
+
"eval_samples_per_second": 51.62,
|
| 189 |
+
"eval_steps_per_second": 0.826,
|
| 190 |
+
"step": 500
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 0.4306808859721083,
|
| 194 |
+
"grad_norm": 0.16488012671470642,
|
| 195 |
+
"learning_rate": 0.0019206974542336672,
|
| 196 |
+
"loss": 0.2663,
|
| 197 |
+
"step": 525
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 0.45118949958982774,
|
| 201 |
+
"grad_norm": 0.16170910000801086,
|
| 202 |
+
"learning_rate": 0.0019061396187191563,
|
| 203 |
+
"loss": 0.2789,
|
| 204 |
+
"step": 550
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 0.4716981132075472,
|
| 208 |
+
"grad_norm": 0.1601390689611435,
|
| 209 |
+
"learning_rate": 0.0018904207216166836,
|
| 210 |
+
"loss": 0.2676,
|
| 211 |
+
"step": 575
|
| 212 |
+
},
|
| 213 |
+
{
|
| 214 |
+
"epoch": 0.4922067268252666,
|
| 215 |
+
"grad_norm": 0.1533547341823578,
|
| 216 |
+
"learning_rate": 0.001873560903980955,
|
| 217 |
+
"loss": 0.2843,
|
| 218 |
+
"step": 600
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.4922067268252666,
|
| 222 |
+
"eval_loss": 0.25965753197669983,
|
| 223 |
+
"eval_runtime": 19.3478,
|
| 224 |
+
"eval_samples_per_second": 51.685,
|
| 225 |
+
"eval_steps_per_second": 0.827,
|
| 226 |
+
"step": 600
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.5127153404429861,
|
| 230 |
+
"grad_norm": 0.14545953273773193,
|
| 231 |
+
"learning_rate": 0.0018555817687594984,
|
| 232 |
+
"loss": 0.2636,
|
| 233 |
+
"step": 625
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.5332239540607056,
|
| 237 |
+
"grad_norm": 0.1371300369501114,
|
| 238 |
+
"learning_rate": 0.0018365063531122169,
|
| 239 |
+
"loss": 0.2545,
|
| 240 |
+
"step": 650
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.5537325676784249,
|
| 244 |
+
"grad_norm": 0.1269655078649521,
|
| 245 |
+
"learning_rate": 0.0018163590988932402,
|
| 246 |
+
"loss": 0.2664,
|
| 247 |
+
"step": 675
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.5742411812961444,
|
| 251 |
+
"grad_norm": 0.13067300617694855,
|
| 252 |
+
"learning_rate": 0.0017951658213329078,
|
| 253 |
+
"loss": 0.2641,
|
| 254 |
+
"step": 700
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.5742411812961444,
|
| 258 |
+
"eval_loss": 0.2512499690055847,
|
| 259 |
+
"eval_runtime": 19.3628,
|
| 260 |
+
"eval_samples_per_second": 51.645,
|
| 261 |
+
"eval_steps_per_second": 0.826,
|
| 262 |
+
"step": 700
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.5947497949138638,
|
| 266 |
+
"grad_norm": 0.13466237485408783,
|
| 267 |
+
"learning_rate": 0.0017729536759600033,
|
| 268 |
+
"loss": 0.2618,
|
| 269 |
+
"step": 725
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.6152584085315833,
|
| 273 |
+
"grad_norm": 0.13332244753837585,
|
| 274 |
+
"learning_rate": 0.0017497511238066307,
|
| 275 |
+
"loss": 0.2581,
|
| 276 |
+
"step": 750
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.6357670221493027,
|
| 280 |
+
"grad_norm": 0.14165574312210083,
|
| 281 |
+
"learning_rate": 0.00172558789494031,
|
| 282 |
+
"loss": 0.2552,
|
| 283 |
+
"step": 775
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.6562756357670222,
|
| 287 |
+
"grad_norm": 0.13256995379924774,
|
| 288 |
+
"learning_rate": 0.0017004949503700284,
|
| 289 |
+
"loss": 0.2595,
|
| 290 |
+
"step": 800
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.6562756357670222,
|
| 294 |
+
"eval_loss": 0.2461422085762024,
|
| 295 |
+
"eval_runtime": 19.3355,
|
| 296 |
+
"eval_samples_per_second": 51.718,
|
| 297 |
+
"eval_steps_per_second": 0.827,
|
| 298 |
+
"step": 800
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 0.6767842493847416,
|
| 302 |
+
"grad_norm": 0.12130759656429291,
|
| 303 |
+
"learning_rate": 0.0016745044423750449,
|
| 304 |
+
"loss": 0.2517,
|
| 305 |
+
"step": 825
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 0.6972928630024611,
|
| 309 |
+
"grad_norm": 0.12769611179828644,
|
| 310 |
+
"learning_rate": 0.0016476496733072946,
|
| 311 |
+
"loss": 0.2554,
|
| 312 |
+
"step": 850
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 0.7178014766201805,
|
| 316 |
+
"grad_norm": 0.12426191568374634,
|
| 317 |
+
"learning_rate": 0.0016199650529201684,
|
| 318 |
+
"loss": 0.2426,
|
| 319 |
+
"step": 875
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 0.7383100902379,
|
| 323 |
+
"grad_norm": 0.1303705871105194,
|
| 324 |
+
"learning_rate": 0.0015914860542783522,
|
| 325 |
+
"loss": 0.2484,
|
| 326 |
+
"step": 900
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 0.7383100902379,
|
| 330 |
+
"eval_loss": 0.23876096308231354,
|
| 331 |
+
"eval_runtime": 19.3557,
|
| 332 |
+
"eval_samples_per_second": 51.664,
|
| 333 |
+
"eval_steps_per_second": 0.827,
|
| 334 |
+
"step": 900
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"epoch": 0.7588187038556193,
|
| 338 |
+
"grad_norm": 0.1256195455789566,
|
| 339 |
+
"learning_rate": 0.0015622491683052124,
|
| 340 |
+
"loss": 0.2514,
|
| 341 |
+
"step": 925
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 0.7793273174733388,
|
| 345 |
+
"grad_norm": 0.13517719507217407,
|
| 346 |
+
"learning_rate": 0.0015322918570259759,
|
| 347 |
+
"loss": 0.2388,
|
| 348 |
+
"step": 950
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"epoch": 0.7998359310910582,
|
| 352 |
+
"grad_norm": 0.12305447459220886,
|
| 353 |
+
"learning_rate": 0.0015016525055666057,
|
| 354 |
+
"loss": 0.2457,
|
| 355 |
+
"step": 975
|
| 356 |
+
},
|
| 357 |
+
{
|
| 358 |
+
"epoch": 0.8203445447087777,
|
| 359 |
+
"grad_norm": 0.13518399000167847,
|
| 360 |
+
"learning_rate": 0.001470370372969886,
|
| 361 |
+
"loss": 0.238,
|
| 362 |
+
"step": 1000
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.8203445447087777,
|
| 366 |
+
"eval_loss": 0.23509961366653442,
|
| 367 |
+
"eval_runtime": 19.3367,
|
| 368 |
+
"eval_samples_per_second": 51.715,
|
| 369 |
+
"eval_steps_per_second": 0.827,
|
| 370 |
+
"step": 1000
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 0.8408531583264971,
|
| 374 |
+
"grad_norm": 0.1178346797823906,
|
| 375 |
+
"learning_rate": 0.0014384855418917311,
|
| 376 |
+
"loss": 0.2383,
|
| 377 |
+
"step": 1025
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 0.8613617719442166,
|
| 381 |
+
"grad_norm": 0.10460968315601349,
|
| 382 |
+
"learning_rate": 0.0014060388672421775,
|
| 383 |
+
"loss": 0.2359,
|
| 384 |
+
"step": 1050
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.881870385561936,
|
| 388 |
+
"grad_norm": 0.11884552240371704,
|
| 389 |
+
"learning_rate": 0.0013730719238368662,
|
| 390 |
+
"loss": 0.2411,
|
| 391 |
+
"step": 1075
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 0.9023789991796555,
|
| 395 |
+
"grad_norm": 0.1170521080493927,
|
| 396 |
+
"learning_rate": 0.0013396269531260867,
|
| 397 |
+
"loss": 0.2401,
|
| 398 |
+
"step": 1100
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 0.9023789991796555,
|
| 402 |
+
"eval_loss": 0.22944915294647217,
|
| 403 |
+
"eval_runtime": 19.337,
|
| 404 |
+
"eval_samples_per_second": 51.714,
|
| 405 |
+
"eval_steps_per_second": 0.827,
|
| 406 |
+
"step": 1100
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 0.9228876127973749,
|
| 410 |
+
"grad_norm": 0.10574015229940414,
|
| 411 |
+
"learning_rate": 0.0013057468090696496,
|
| 412 |
+
"loss": 0.2363,
|
| 413 |
+
"step": 1125
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 0.9433962264150944,
|
| 417 |
+
"grad_norm": 0.10140049457550049,
|
| 418 |
+
"learning_rate": 0.0012714749032269287,
|
| 419 |
+
"loss": 0.2353,
|
| 420 |
+
"step": 1150
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 0.9639048400328137,
|
| 424 |
+
"grad_norm": 0.09746623784303665,
|
| 425 |
+
"learning_rate": 0.0012368551491324358,
|
| 426 |
+
"loss": 0.2405,
|
| 427 |
+
"step": 1175
|
| 428 |
+
},
|
| 429 |
+
{
|
| 430 |
+
"epoch": 0.9844134536505332,
|
| 431 |
+
"grad_norm": 0.11055821180343628,
|
| 432 |
+
"learning_rate": 0.0012019319060282063,
|
| 433 |
+
"loss": 0.2457,
|
| 434 |
+
"step": 1200
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 0.9844134536505332,
|
| 438 |
+
"eval_loss": 0.2251633256673813,
|
| 439 |
+
"eval_runtime": 19.3599,
|
| 440 |
+
"eval_samples_per_second": 51.653,
|
| 441 |
+
"eval_steps_per_second": 0.826,
|
| 442 |
+
"step": 1200
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 1.0049220672682526,
|
| 446 |
+
"grad_norm": 0.10832362622022629,
|
| 447 |
+
"learning_rate": 0.0011667499220250803,
|
| 448 |
+
"loss": 0.2256,
|
| 449 |
+
"step": 1225
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 1.0254306808859721,
|
| 453 |
+
"grad_norm": 0.11733749508857727,
|
| 454 |
+
"learning_rate": 0.0011313542767657204,
|
| 455 |
+
"loss": 0.2042,
|
| 456 |
+
"step": 1250
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 1.0459392945036916,
|
| 460 |
+
"grad_norm": 0.11244157701730728,
|
| 461 |
+
"learning_rate": 0.0010957903236628267,
|
| 462 |
+
"loss": 0.2048,
|
| 463 |
+
"step": 1275
|
| 464 |
+
},
|
| 465 |
+
{
|
| 466 |
+
"epoch": 1.066447908121411,
|
| 467 |
+
"grad_norm": 0.09949418157339096,
|
| 468 |
+
"learning_rate": 0.001060103631786563,
|
| 469 |
+
"loss": 0.2091,
|
| 470 |
+
"step": 1300
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 1.066447908121411,
|
| 474 |
+
"eval_loss": 0.22334164381027222,
|
| 475 |
+
"eval_runtime": 19.3527,
|
| 476 |
+
"eval_samples_per_second": 51.672,
|
| 477 |
+
"eval_steps_per_second": 0.827,
|
| 478 |
+
"step": 1300
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 1.0869565217391304,
|
| 482 |
+
"grad_norm": 0.11762440949678421,
|
| 483 |
+
"learning_rate": 0.0010243399274756564,
|
| 484 |
+
"loss": 0.2061,
|
| 485 |
+
"step": 1325
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 1.1074651353568499,
|
| 489 |
+
"grad_norm": 0.09624772518873215,
|
| 490 |
+
"learning_rate": 0.0009885450357469806,
|
| 491 |
+
"loss": 0.1978,
|
| 492 |
+
"step": 1350
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 1.1279737489745694,
|
| 496 |
+
"grad_norm": 0.11852416396141052,
|
| 497 |
+
"learning_rate": 0.0009527648215787065,
|
| 498 |
+
"loss": 0.1996,
|
| 499 |
+
"step": 1375
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 1.1484823625922886,
|
| 503 |
+
"grad_norm": 0.12652769684791565,
|
| 504 |
+
"learning_rate": 0.000917045131142242,
|
| 505 |
+
"loss": 0.1944,
|
| 506 |
+
"step": 1400
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 1.1484823625922886,
|
| 510 |
+
"eval_loss": 0.22206895053386688,
|
| 511 |
+
"eval_runtime": 19.3869,
|
| 512 |
+
"eval_samples_per_second": 51.581,
|
| 513 |
+
"eval_steps_per_second": 0.825,
|
| 514 |
+
"step": 1400
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 1.1689909762100081,
|
| 518 |
+
"grad_norm": 0.12416432797908783,
|
| 519 |
+
"learning_rate": 0.0008814317330582753,
|
| 520 |
+
"loss": 0.2033,
|
| 521 |
+
"step": 1425
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 1.1894995898277276,
|
| 525 |
+
"grad_norm": 0.11753161251544952,
|
| 526 |
+
"learning_rate": 0.000845970259752183,
|
| 527 |
+
"loss": 0.2043,
|
| 528 |
+
"step": 1450
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 1.2100082034454471,
|
| 532 |
+
"grad_norm": 0.14326679706573486,
|
| 533 |
+
"learning_rate": 0.0008107061489839498,
|
| 534 |
+
"loss": 0.2004,
|
| 535 |
+
"step": 1475
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 1.2305168170631666,
|
| 539 |
+
"grad_norm": 0.11070103943347931,
|
| 540 |
+
"learning_rate": 0.0007756845856275194,
|
| 541 |
+
"loss": 0.2095,
|
| 542 |
+
"step": 1500
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 1.2305168170631666,
|
| 546 |
+
"eval_loss": 0.21846410632133484,
|
| 547 |
+
"eval_runtime": 19.3592,
|
| 548 |
+
"eval_samples_per_second": 51.655,
|
| 549 |
+
"eval_steps_per_second": 0.826,
|
| 550 |
+
"step": 1500
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 1.251025430680886,
|
| 554 |
+
"grad_norm": 0.12698207795619965,
|
| 555 |
+
"learning_rate": 0.0007409504437741722,
|
| 556 |
+
"loss": 0.2036,
|
| 557 |
+
"step": 1525
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 1.2715340442986054,
|
| 561 |
+
"grad_norm": 0.0978737324476242,
|
| 562 |
+
"learning_rate": 0.0007065482292341205,
|
| 563 |
+
"loss": 0.1978,
|
| 564 |
+
"step": 1550
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 1.2920426579163249,
|
| 568 |
+
"grad_norm": 0.09450080245733261,
|
| 569 |
+
"learning_rate": 0.0006725220225099911,
|
| 570 |
+
"loss": 0.1997,
|
| 571 |
+
"step": 1575
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 1.3125512715340442,
|
| 575 |
+
"grad_norm": 0.10328544676303864,
|
| 576 |
+
"learning_rate": 0.0006389154223152666,
|
| 577 |
+
"loss": 0.1923,
|
| 578 |
+
"step": 1600
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 1.3125512715340442,
|
| 582 |
+
"eval_loss": 0.21600544452667236,
|
| 583 |
+
"eval_runtime": 19.3364,
|
| 584 |
+
"eval_samples_per_second": 51.716,
|
| 585 |
+
"eval_steps_per_second": 0.827,
|
| 586 |
+
"step": 1600
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 1.3330598851517639,
|
| 590 |
+
"grad_norm": 0.09850370138883591,
|
| 591 |
+
"learning_rate": 0.0006057714897100551,
|
| 592 |
+
"loss": 0.1961,
|
| 593 |
+
"step": 1625
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 1.3535684987694832,
|
| 597 |
+
"grad_norm": 0.10001111030578613,
|
| 598 |
+
"learning_rate": 0.0005731326929257713,
|
| 599 |
+
"loss": 0.1966,
|
| 600 |
+
"step": 1650
|
| 601 |
+
},
|
| 602 |
+
{
|
| 603 |
+
"epoch": 1.3740771123872026,
|
| 604 |
+
"grad_norm": 0.1204497367143631,
|
| 605 |
+
"learning_rate": 0.0005410408529494251,
|
| 606 |
+
"loss": 0.1936,
|
| 607 |
+
"step": 1675
|
| 608 |
+
},
|
| 609 |
+
{
|
| 610 |
+
"epoch": 1.3945857260049221,
|
| 611 |
+
"grad_norm": 0.1069173812866211,
|
| 612 |
+
"learning_rate": 0.0005095370899372412,
|
| 613 |
+
"loss": 0.194,
|
| 614 |
+
"step": 1700
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 1.3945857260049221,
|
| 618 |
+
"eval_loss": 0.21333414316177368,
|
| 619 |
+
"eval_runtime": 19.3528,
|
| 620 |
+
"eval_samples_per_second": 51.672,
|
| 621 |
+
"eval_steps_per_second": 0.827,
|
| 622 |
+
"step": 1700
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"epoch": 1.4150943396226414,
|
| 626 |
+
"grad_norm": 0.11550671607255936,
|
| 627 |
+
"learning_rate": 0.0004786617705262746,
|
| 628 |
+
"loss": 0.1913,
|
| 629 |
+
"step": 1725
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 1.435602953240361,
|
| 633 |
+
"grad_norm": 0.10375545173883438,
|
| 634 |
+
"learning_rate": 0.000448454456111529,
|
| 635 |
+
"loss": 0.1907,
|
| 636 |
+
"step": 1750
|
| 637 |
+
},
|
| 638 |
+
{
|
| 639 |
+
"epoch": 1.4561115668580804,
|
| 640 |
+
"grad_norm": 0.1066800057888031,
|
| 641 |
+
"learning_rate": 0.0004189538521548524,
|
| 642 |
+
"loss": 0.197,
|
| 643 |
+
"step": 1775
|
| 644 |
+
},
|
| 645 |
+
{
|
| 646 |
+
"epoch": 1.4766201804758,
|
| 647 |
+
"grad_norm": 0.10582635551691055,
|
| 648 |
+
"learning_rate": 0.00039019775859056916,
|
| 649 |
+
"loss": 0.1862,
|
| 650 |
+
"step": 1800
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 1.4766201804758,
|
| 654 |
+
"eval_loss": 0.2119527906179428,
|
| 655 |
+
"eval_runtime": 19.3667,
|
| 656 |
+
"eval_samples_per_second": 51.635,
|
| 657 |
+
"eval_steps_per_second": 0.826,
|
| 658 |
+
"step": 1800
|
| 659 |
+
},
|
| 660 |
+
{
|
| 661 |
+
"epoch": 1.4971287940935194,
|
| 662 |
+
"grad_norm": 0.2869551479816437,
|
| 663 |
+
"learning_rate": 0.0003622230213913836,
|
| 664 |
+
"loss": 0.1856,
|
| 665 |
+
"step": 1825
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 1.5176374077112387,
|
| 669 |
+
"grad_norm": 0.09285558015108109,
|
| 670 |
+
"learning_rate": 0.0003350654853566223,
|
| 671 |
+
"loss": 0.1894,
|
| 672 |
+
"step": 1850
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 1.5381460213289582,
|
| 676 |
+
"grad_norm": 0.09621681272983551,
|
| 677 |
+
"learning_rate": 0.00030875994818330957,
|
| 678 |
+
"loss": 0.1898,
|
| 679 |
+
"step": 1875
|
| 680 |
+
},
|
| 681 |
+
{
|
| 682 |
+
"epoch": 1.5586546349466777,
|
| 683 |
+
"grad_norm": 0.11009588092565536,
|
| 684 |
+
"learning_rate": 0.0002833401158789207,
|
| 685 |
+
"loss": 0.192,
|
| 686 |
+
"step": 1900
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 1.5586546349466777,
|
| 690 |
+
"eval_loss": 0.2094431221485138,
|
| 691 |
+
"eval_runtime": 19.3715,
|
| 692 |
+
"eval_samples_per_second": 51.622,
|
| 693 |
+
"eval_steps_per_second": 0.826,
|
| 694 |
+
"step": 1900
|
| 695 |
+
},
|
| 696 |
+
{
|
| 697 |
+
"epoch": 1.579163248564397,
|
| 698 |
+
"grad_norm": 0.10520603507757187,
|
| 699 |
+
"learning_rate": 0.00025883855957295053,
|
| 700 |
+
"loss": 0.1927,
|
| 701 |
+
"step": 1925
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"epoch": 1.5996718621821167,
|
| 705 |
+
"grad_norm": 0.10646632313728333,
|
| 706 |
+
"learning_rate": 0.0002352866737826277,
|
| 707 |
+
"loss": 0.1912,
|
| 708 |
+
"step": 1950
|
| 709 |
+
},
|
| 710 |
+
{
|
| 711 |
+
"epoch": 1.620180475799836,
|
| 712 |
+
"grad_norm": 0.09942781180143356,
|
| 713 |
+
"learning_rate": 0.00021271463618625986,
|
| 714 |
+
"loss": 0.1869,
|
| 715 |
+
"step": 1975
|
| 716 |
+
},
|
| 717 |
+
{
|
| 718 |
+
"epoch": 1.6406890894175554,
|
| 719 |
+
"grad_norm": 0.10553489625453949,
|
| 720 |
+
"learning_rate": 0.00019115136895574402,
|
| 721 |
+
"loss": 0.1932,
|
| 722 |
+
"step": 2000
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 1.6406890894175554,
|
| 726 |
+
"eval_loss": 0.20747876167297363,
|
| 727 |
+
"eval_runtime": 19.3506,
|
| 728 |
+
"eval_samples_per_second": 51.678,
|
| 729 |
+
"eval_steps_per_second": 0.827,
|
| 730 |
+
"step": 2000
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 1.661197703035275,
|
| 734 |
+
"grad_norm": 0.0937742292881012,
|
| 735 |
+
"learning_rate": 0.0001706245016977931,
|
| 736 |
+
"loss": 0.1837,
|
| 737 |
+
"step": 2025
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 1.6817063166529942,
|
| 741 |
+
"grad_norm": 0.10875856131315231,
|
| 742 |
+
"learning_rate": 0.00015116033605136182,
|
| 743 |
+
"loss": 0.1851,
|
| 744 |
+
"step": 2050
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 1.7022149302707137,
|
| 748 |
+
"grad_norm": 0.12695828080177307,
|
| 749 |
+
"learning_rate": 0.00013278381198663492,
|
| 750 |
+
"loss": 0.1911,
|
| 751 |
+
"step": 2075
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 1.7227235438884332,
|
| 755 |
+
"grad_norm": 0.10288714617490768,
|
| 756 |
+
"learning_rate": 0.0001155184758487573,
|
| 757 |
+
"loss": 0.1811,
|
| 758 |
+
"step": 2100
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 1.7227235438884332,
|
| 762 |
+
"eval_loss": 0.20649734139442444,
|
| 763 |
+
"eval_runtime": 19.344,
|
| 764 |
+
"eval_samples_per_second": 51.696,
|
| 765 |
+
"eval_steps_per_second": 0.827,
|
| 766 |
+
"step": 2100
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 1.7432321575061525,
|
| 770 |
+
"grad_norm": 0.09159814566373825,
|
| 771 |
+
"learning_rate": 9.938645018725523e-05,
|
| 772 |
+
"loss": 0.1848,
|
| 773 |
+
"step": 2125
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 1.7637407711238722,
|
| 777 |
+
"grad_norm": 0.08860918134450912,
|
| 778 |
+
"learning_rate": 8.440840540980587e-05,
|
| 779 |
+
"loss": 0.1832,
|
| 780 |
+
"step": 2150
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 1.7842493847415914,
|
| 784 |
+
"grad_norm": 0.09137848764657974,
|
| 785 |
+
"learning_rate": 7.060353329667668e-05,
|
| 786 |
+
"loss": 0.1918,
|
| 787 |
+
"step": 2175
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 1.804757998359311,
|
| 791 |
+
"grad_norm": 0.10140997171401978,
|
| 792 |
+
"learning_rate": 5.798952240976951e-05,
|
| 793 |
+
"loss": 0.1851,
|
| 794 |
+
"step": 2200
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 1.804757998359311,
|
| 798 |
+
"eval_loss": 0.20535807311534882,
|
| 799 |
+
"eval_runtime": 19.3355,
|
| 800 |
+
"eval_samples_per_second": 51.718,
|
| 801 |
+
"eval_steps_per_second": 0.827,
|
| 802 |
+
"step": 2200
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 1.8252666119770304,
|
| 806 |
+
"grad_norm": 0.10309699922800064,
|
| 807 |
+
"learning_rate": 4.65825354277799e-05,
|
| 808 |
+
"loss": 0.1856,
|
| 809 |
+
"step": 2225
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 1.8457752255947497,
|
| 813 |
+
"grad_norm": 0.10617036372423172,
|
| 814 |
+
"learning_rate": 3.639718843651363e-05,
|
| 815 |
+
"loss": 0.1873,
|
| 816 |
+
"step": 2250
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"epoch": 1.8662838392124692,
|
| 820 |
+
"grad_norm": 0.09746799618005753,
|
| 821 |
+
"learning_rate": 2.7446532200894104e-05,
|
| 822 |
+
"loss": 0.1917,
|
| 823 |
+
"step": 2275
|
| 824 |
+
},
|
| 825 |
+
{
|
| 826 |
+
"epoch": 1.8867924528301887,
|
| 827 |
+
"grad_norm": 0.08962379395961761,
|
| 828 |
+
"learning_rate": 1.9742035442658403e-05,
|
| 829 |
+
"loss": 0.1846,
|
| 830 |
+
"step": 2300
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 1.8867924528301887,
|
| 834 |
+
"eval_loss": 0.2049131840467453,
|
| 835 |
+
"eval_runtime": 19.3544,
|
| 836 |
+
"eval_samples_per_second": 51.668,
|
| 837 |
+
"eval_steps_per_second": 0.827,
|
| 838 |
+
"step": 2300
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 1.907301066447908,
|
| 842 |
+
"grad_norm": 0.12254218012094498,
|
| 843 |
+
"learning_rate": 1.3293570145169742e-05,
|
| 844 |
+
"loss": 0.1933,
|
| 845 |
+
"step": 2325
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 1.9278096800656277,
|
| 849 |
+
"grad_norm": 0.09826746582984924,
|
| 850 |
+
"learning_rate": 8.109398904173282e-06,
|
| 851 |
+
"loss": 0.1773,
|
| 852 |
+
"step": 2350
|
| 853 |
+
},
|
| 854 |
+
{
|
| 855 |
+
"epoch": 1.948318293683347,
|
| 856 |
+
"grad_norm": 0.08864423632621765,
|
| 857 |
+
"learning_rate": 4.196164340705577e-06,
|
| 858 |
+
"loss": 0.183,
|
| 859 |
+
"step": 2375
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 1.9688269073010665,
|
| 863 |
+
"grad_norm": 0.09590858966112137,
|
| 864 |
+
"learning_rate": 1.5588805897215342e-06,
|
| 865 |
+
"loss": 0.184,
|
| 866 |
+
"step": 2400
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 1.9688269073010665,
|
| 870 |
+
"eval_loss": 0.2048163115978241,
|
| 871 |
+
"eval_runtime": 19.3386,
|
| 872 |
+
"eval_samples_per_second": 51.71,
|
| 873 |
+
"eval_steps_per_second": 0.827,
|
| 874 |
+
"step": 2400
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"epoch": 1.989335520918786,
|
| 878 |
+
"grad_norm": 0.09918979555368423,
|
| 879 |
+
"learning_rate": 2.0092687534589705e-07,
|
| 880 |
+
"loss": 0.1813,
|
| 881 |
+
"step": 2425
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 2.0,
|
| 885 |
+
"step": 2438,
|
| 886 |
+
"total_flos": 1.58523627405312e+18,
|
| 887 |
+
"train_loss": 0.24037269298124744,
|
| 888 |
+
"train_runtime": 2884.3835,
|
| 889 |
+
"train_samples_per_second": 27.042,
|
| 890 |
+
"train_steps_per_second": 0.845
|
| 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_ex09/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": 16,
|
| 9 |
+
"peft_type": "ROTATION",
|
| 10 |
+
"r": 1,
|
| 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_ex09/ft/added_tokens.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"[PAD]": 32000
|
| 3 |
+
}
|
nl_tasks/exps/run_ex09/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_ex09/ft/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
nl_tasks/exps/run_ex09/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_ex09/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_ex09/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": 16,
|
| 9 |
+
"peft_type": "ROTATION",
|
| 10 |
+
"r": 1,
|
| 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_ex09/ft2/adapter_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:8cf11ff27c491ce13212598d50e355fbf732f6a88d35a2d4435cd4e45ef10aef
|
| 3 |
+
size 33602659
|
nl_tasks/exps/run_ex09/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": 0.1543785035610199,
|
| 15 |
+
"learning_rate": 0.00019672131147540983,
|
| 16 |
+
"loss": 0.6266,
|
| 17 |
+
"step": 25
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.04101722723543889,
|
| 21 |
+
"grad_norm": 0.2001742422580719,
|
| 22 |
+
"learning_rate": 0.00040163934426229507,
|
| 23 |
+
"loss": 0.3858,
|
| 24 |
+
"step": 50
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.06152584085315833,
|
| 28 |
+
"grad_norm": 0.14139799773693085,
|
| 29 |
+
"learning_rate": 0.0006065573770491804,
|
| 30 |
+
"loss": 0.3604,
|
| 31 |
+
"step": 75
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.08203445447087777,
|
| 35 |
+
"grad_norm": 0.20014838874340057,
|
| 36 |
+
"learning_rate": 0.0008114754098360656,
|
| 37 |
+
"loss": 0.3285,
|
| 38 |
+
"step": 100
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.08203445447087777,
|
| 42 |
+
"eval_loss": 0.3202356696128845,
|
| 43 |
+
"eval_runtime": 22.7105,
|
| 44 |
+
"eval_samples_per_second": 44.033,
|
| 45 |
+
"eval_steps_per_second": 0.705,
|
| 46 |
+
"step": 100
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"epoch": 0.10254306808859721,
|
| 50 |
+
"grad_norm": 0.2310003638267517,
|
| 51 |
+
"learning_rate": 0.0010163934426229509,
|
| 52 |
+
"loss": 0.341,
|
| 53 |
+
"step": 125
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 0.12305168170631665,
|
| 57 |
+
"grad_norm": 0.3138357102870941,
|
| 58 |
+
"learning_rate": 0.001221311475409836,
|
| 59 |
+
"loss": 0.3325,
|
| 60 |
+
"step": 150
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 0.1435602953240361,
|
| 64 |
+
"grad_norm": 1.425217628479004,
|
| 65 |
+
"learning_rate": 0.0014262295081967215,
|
| 66 |
+
"loss": 1.0626,
|
| 67 |
+
"step": 175
|
| 68 |
+
},
|
| 69 |
+
{
|
| 70 |
+
"epoch": 0.16406890894175555,
|
| 71 |
+
"grad_norm": 0.6279019713401794,
|
| 72 |
+
"learning_rate": 0.0016311475409836065,
|
| 73 |
+
"loss": 0.4087,
|
| 74 |
+
"step": 200
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.16406890894175555,
|
| 78 |
+
"eval_loss": 0.38264134526252747,
|
| 79 |
+
"eval_runtime": 22.5381,
|
| 80 |
+
"eval_samples_per_second": 44.369,
|
| 81 |
+
"eval_steps_per_second": 0.71,
|
| 82 |
+
"step": 200
|
| 83 |
+
},
|
| 84 |
+
{
|
| 85 |
+
"epoch": 0.184577522559475,
|
| 86 |
+
"grad_norm": 0.38462212681770325,
|
| 87 |
+
"learning_rate": 0.0018360655737704918,
|
| 88 |
+
"loss": 0.4034,
|
| 89 |
+
"step": 225
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 0.20508613617719443,
|
| 93 |
+
"grad_norm": 0.3672288954257965,
|
| 94 |
+
"learning_rate": 0.0019999743708232127,
|
| 95 |
+
"loss": 0.3633,
|
| 96 |
+
"step": 250
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"epoch": 0.22559474979491387,
|
| 100 |
+
"grad_norm": 0.3607560694217682,
|
| 101 |
+
"learning_rate": 0.0019990774875676054,
|
| 102 |
+
"loss": 0.3487,
|
| 103 |
+
"step": 275
|
| 104 |
+
},
|
| 105 |
+
{
|
| 106 |
+
"epoch": 0.2461033634126333,
|
| 107 |
+
"grad_norm": 0.3346173167228699,
|
| 108 |
+
"learning_rate": 0.001996900458879386,
|
| 109 |
+
"loss": 0.3371,
|
| 110 |
+
"step": 300
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.2461033634126333,
|
| 114 |
+
"eval_loss": 0.3209039270877838,
|
| 115 |
+
"eval_runtime": 22.5783,
|
| 116 |
+
"eval_samples_per_second": 44.29,
|
| 117 |
+
"eval_steps_per_second": 0.709,
|
| 118 |
+
"step": 300
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 0.2666119770303528,
|
| 122 |
+
"grad_norm": 0.22805160284042358,
|
| 123 |
+
"learning_rate": 0.001993446074245224,
|
| 124 |
+
"loss": 0.3296,
|
| 125 |
+
"step": 325
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 0.2871205906480722,
|
| 129 |
+
"grad_norm": 0.18590985238552094,
|
| 130 |
+
"learning_rate": 0.0019887187598630527,
|
| 131 |
+
"loss": 0.3134,
|
| 132 |
+
"step": 350
|
| 133 |
+
},
|
| 134 |
+
{
|
| 135 |
+
"epoch": 0.30762920426579166,
|
| 136 |
+
"grad_norm": 0.21088963747024536,
|
| 137 |
+
"learning_rate": 0.0019827245729706648,
|
| 138 |
+
"loss": 0.3199,
|
| 139 |
+
"step": 375
|
| 140 |
+
},
|
| 141 |
+
{
|
| 142 |
+
"epoch": 0.3281378178835111,
|
| 143 |
+
"grad_norm": 0.18156805634498596,
|
| 144 |
+
"learning_rate": 0.0019754711940844047,
|
| 145 |
+
"loss": 0.2996,
|
| 146 |
+
"step": 400
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.3281378178835111,
|
| 150 |
+
"eval_loss": 0.28689703345298767,
|
| 151 |
+
"eval_runtime": 22.5553,
|
| 152 |
+
"eval_samples_per_second": 44.335,
|
| 153 |
+
"eval_steps_per_second": 0.709,
|
| 154 |
+
"step": 400
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 0.34864643150123054,
|
| 158 |
+
"grad_norm": 0.15950609743595123,
|
| 159 |
+
"learning_rate": 0.0019669679171579117,
|
| 160 |
+
"loss": 0.3044,
|
| 161 |
+
"step": 425
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 0.36915504511895,
|
| 165 |
+
"grad_norm": 0.16445936262607574,
|
| 166 |
+
"learning_rate": 0.001957225637673524,
|
| 167 |
+
"loss": 0.3019,
|
| 168 |
+
"step": 450
|
| 169 |
+
},
|
| 170 |
+
{
|
| 171 |
+
"epoch": 0.3896636587366694,
|
| 172 |
+
"grad_norm": 0.16957086324691772,
|
| 173 |
+
"learning_rate": 0.0019462568386815961,
|
| 174 |
+
"loss": 0.2863,
|
| 175 |
+
"step": 475
|
| 176 |
+
},
|
| 177 |
+
{
|
| 178 |
+
"epoch": 0.41017227235438886,
|
| 179 |
+
"grad_norm": 0.12954926490783691,
|
| 180 |
+
"learning_rate": 0.0019340755748056234,
|
| 181 |
+
"loss": 0.2701,
|
| 182 |
+
"step": 500
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.41017227235438886,
|
| 186 |
+
"eval_loss": 0.2733325660228729,
|
| 187 |
+
"eval_runtime": 22.5518,
|
| 188 |
+
"eval_samples_per_second": 44.342,
|
| 189 |
+
"eval_steps_per_second": 0.709,
|
| 190 |
+
"step": 500
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 0.4306808859721083,
|
| 194 |
+
"grad_norm": 0.1369732916355133,
|
| 195 |
+
"learning_rate": 0.0019206974542336672,
|
| 196 |
+
"loss": 0.271,
|
| 197 |
+
"step": 525
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 0.45118949958982774,
|
| 201 |
+
"grad_norm": 0.15917326509952545,
|
| 202 |
+
"learning_rate": 0.0019061396187191563,
|
| 203 |
+
"loss": 0.2802,
|
| 204 |
+
"step": 550
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 0.4716981132075472,
|
| 208 |
+
"grad_norm": 0.16746191680431366,
|
| 209 |
+
"learning_rate": 0.0018904207216166836,
|
| 210 |
+
"loss": 0.2691,
|
| 211 |
+
"step": 575
|
| 212 |
+
},
|
| 213 |
+
{
|
| 214 |
+
"epoch": 0.4922067268252666,
|
| 215 |
+
"grad_norm": 0.1554066687822342,
|
| 216 |
+
"learning_rate": 0.001873560903980955,
|
| 217 |
+
"loss": 0.286,
|
| 218 |
+
"step": 600
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.4922067268252666,
|
| 222 |
+
"eval_loss": 0.26212847232818604,
|
| 223 |
+
"eval_runtime": 22.5566,
|
| 224 |
+
"eval_samples_per_second": 44.333,
|
| 225 |
+
"eval_steps_per_second": 0.709,
|
| 226 |
+
"step": 600
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.5127153404429861,
|
| 230 |
+
"grad_norm": 0.13422970473766327,
|
| 231 |
+
"learning_rate": 0.0018555817687594984,
|
| 232 |
+
"loss": 0.2655,
|
| 233 |
+
"step": 625
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.5332239540607056,
|
| 237 |
+
"grad_norm": 0.14770525693893433,
|
| 238 |
+
"learning_rate": 0.0018365063531122169,
|
| 239 |
+
"loss": 0.26,
|
| 240 |
+
"step": 650
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.5537325676784249,
|
| 244 |
+
"grad_norm": 0.12729965150356293,
|
| 245 |
+
"learning_rate": 0.0018163590988932402,
|
| 246 |
+
"loss": 0.2694,
|
| 247 |
+
"step": 675
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.5742411812961444,
|
| 251 |
+
"grad_norm": 0.1334213763475418,
|
| 252 |
+
"learning_rate": 0.0017951658213329078,
|
| 253 |
+
"loss": 0.268,
|
| 254 |
+
"step": 700
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.5742411812961444,
|
| 258 |
+
"eval_loss": 0.25402218103408813,
|
| 259 |
+
"eval_runtime": 22.5274,
|
| 260 |
+
"eval_samples_per_second": 44.39,
|
| 261 |
+
"eval_steps_per_second": 0.71,
|
| 262 |
+
"step": 700
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.5947497949138638,
|
| 266 |
+
"grad_norm": 0.12552621960639954,
|
| 267 |
+
"learning_rate": 0.0017729536759600033,
|
| 268 |
+
"loss": 0.266,
|
| 269 |
+
"step": 725
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.6152584085315833,
|
| 273 |
+
"grad_norm": 0.13384173810482025,
|
| 274 |
+
"learning_rate": 0.0017497511238066307,
|
| 275 |
+
"loss": 0.2631,
|
| 276 |
+
"step": 750
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.6357670221493027,
|
| 280 |
+
"grad_norm": 0.12846872210502625,
|
| 281 |
+
"learning_rate": 0.00172558789494031,
|
| 282 |
+
"loss": 0.2588,
|
| 283 |
+
"step": 775
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.6562756357670222,
|
| 287 |
+
"grad_norm": 0.16066329181194305,
|
| 288 |
+
"learning_rate": 0.0017004949503700284,
|
| 289 |
+
"loss": 0.2636,
|
| 290 |
+
"step": 800
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.6562756357670222,
|
| 294 |
+
"eval_loss": 0.24892009794712067,
|
| 295 |
+
"eval_runtime": 22.5749,
|
| 296 |
+
"eval_samples_per_second": 44.297,
|
| 297 |
+
"eval_steps_per_second": 0.709,
|
| 298 |
+
"step": 800
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 0.6767842493847416,
|
| 302 |
+
"grad_norm": 0.11756884306669235,
|
| 303 |
+
"learning_rate": 0.0016745044423750449,
|
| 304 |
+
"loss": 0.2563,
|
| 305 |
+
"step": 825
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 0.6972928630024611,
|
| 309 |
+
"grad_norm": 0.1094069853425026,
|
| 310 |
+
"learning_rate": 0.0016476496733072946,
|
| 311 |
+
"loss": 0.2581,
|
| 312 |
+
"step": 850
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 0.7178014766201805,
|
| 316 |
+
"grad_norm": 0.10113517194986343,
|
| 317 |
+
"learning_rate": 0.0016199650529201684,
|
| 318 |
+
"loss": 0.2466,
|
| 319 |
+
"step": 875
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 0.7383100902379,
|
| 323 |
+
"grad_norm": 0.12762148678302765,
|
| 324 |
+
"learning_rate": 0.0015914860542783522,
|
| 325 |
+
"loss": 0.2511,
|
| 326 |
+
"step": 900
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 0.7383100902379,
|
| 330 |
+
"eval_loss": 0.24198263883590698,
|
| 331 |
+
"eval_runtime": 22.5893,
|
| 332 |
+
"eval_samples_per_second": 44.269,
|
| 333 |
+
"eval_steps_per_second": 0.708,
|
| 334 |
+
"step": 900
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"epoch": 0.7588187038556193,
|
| 338 |
+
"grad_norm": 0.1412491798400879,
|
| 339 |
+
"learning_rate": 0.0015622491683052124,
|
| 340 |
+
"loss": 0.2538,
|
| 341 |
+
"step": 925
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 0.7793273174733388,
|
| 345 |
+
"grad_norm": 0.1309656947851181,
|
| 346 |
+
"learning_rate": 0.0015322918570259759,
|
| 347 |
+
"loss": 0.2417,
|
| 348 |
+
"step": 950
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"epoch": 0.7998359310910582,
|
| 352 |
+
"grad_norm": 0.12559030950069427,
|
| 353 |
+
"learning_rate": 0.0015016525055666057,
|
| 354 |
+
"loss": 0.2498,
|
| 355 |
+
"step": 975
|
| 356 |
+
},
|
| 357 |
+
{
|
| 358 |
+
"epoch": 0.8203445447087777,
|
| 359 |
+
"grad_norm": 0.12614794075489044,
|
| 360 |
+
"learning_rate": 0.001470370372969886,
|
| 361 |
+
"loss": 0.2417,
|
| 362 |
+
"step": 1000
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.8203445447087777,
|
| 366 |
+
"eval_loss": 0.2378261834383011,
|
| 367 |
+
"eval_runtime": 22.5548,
|
| 368 |
+
"eval_samples_per_second": 44.336,
|
| 369 |
+
"eval_steps_per_second": 0.709,
|
| 370 |
+
"step": 1000
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 0.8408531583264971,
|
| 374 |
+
"grad_norm": 0.134114608168602,
|
| 375 |
+
"learning_rate": 0.0014384855418917311,
|
| 376 |
+
"loss": 0.2452,
|
| 377 |
+
"step": 1025
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 0.8613617719442166,
|
| 381 |
+
"grad_norm": 0.11434811353683472,
|
| 382 |
+
"learning_rate": 0.0014060388672421775,
|
| 383 |
+
"loss": 0.2412,
|
| 384 |
+
"step": 1050
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.881870385561936,
|
| 388 |
+
"grad_norm": 0.11284555494785309,
|
| 389 |
+
"learning_rate": 0.0013730719238368662,
|
| 390 |
+
"loss": 0.245,
|
| 391 |
+
"step": 1075
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 0.9023789991796555,
|
| 395 |
+
"grad_norm": 0.13569487631320953,
|
| 396 |
+
"learning_rate": 0.0013396269531260867,
|
| 397 |
+
"loss": 0.246,
|
| 398 |
+
"step": 1100
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 0.9023789991796555,
|
| 402 |
+
"eval_loss": 0.2345089465379715,
|
| 403 |
+
"eval_runtime": 22.5349,
|
| 404 |
+
"eval_samples_per_second": 44.376,
|
| 405 |
+
"eval_steps_per_second": 0.71,
|
| 406 |
+
"step": 1100
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 0.9228876127973749,
|
| 410 |
+
"grad_norm": 0.09805800765752792,
|
| 411 |
+
"learning_rate": 0.0013057468090696496,
|
| 412 |
+
"loss": 0.2414,
|
| 413 |
+
"step": 1125
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 0.9433962264150944,
|
| 417 |
+
"grad_norm": 0.0931050032377243,
|
| 418 |
+
"learning_rate": 0.0012714749032269287,
|
| 419 |
+
"loss": 0.2404,
|
| 420 |
+
"step": 1150
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 0.9639048400328137,
|
| 424 |
+
"grad_norm": 0.10308840870857239,
|
| 425 |
+
"learning_rate": 0.0012368551491324358,
|
| 426 |
+
"loss": 0.245,
|
| 427 |
+
"step": 1175
|
| 428 |
+
},
|
| 429 |
+
{
|
| 430 |
+
"epoch": 0.9844134536505332,
|
| 431 |
+
"grad_norm": 0.10258302837610245,
|
| 432 |
+
"learning_rate": 0.0012019319060282063,
|
| 433 |
+
"loss": 0.2509,
|
| 434 |
+
"step": 1200
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 0.9844134536505332,
|
| 438 |
+
"eval_loss": 0.22989174723625183,
|
| 439 |
+
"eval_runtime": 22.5404,
|
| 440 |
+
"eval_samples_per_second": 44.365,
|
| 441 |
+
"eval_steps_per_second": 0.71,
|
| 442 |
+
"step": 1200
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 1.0049220672682526,
|
| 446 |
+
"grad_norm": 0.11619790643453598,
|
| 447 |
+
"learning_rate": 0.0011667499220250803,
|
| 448 |
+
"loss": 0.2302,
|
| 449 |
+
"step": 1225
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 1.0254306808859721,
|
| 453 |
+
"grad_norm": 0.11248350143432617,
|
| 454 |
+
"learning_rate": 0.0011313542767657204,
|
| 455 |
+
"loss": 0.2105,
|
| 456 |
+
"step": 1250
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 1.0459392945036916,
|
| 460 |
+
"grad_norm": 0.12309166043996811,
|
| 461 |
+
"learning_rate": 0.0010957903236628267,
|
| 462 |
+
"loss": 0.2114,
|
| 463 |
+
"step": 1275
|
| 464 |
+
},
|
| 465 |
+
{
|
| 466 |
+
"epoch": 1.066447908121411,
|
| 467 |
+
"grad_norm": 0.10280752182006836,
|
| 468 |
+
"learning_rate": 0.001060103631786563,
|
| 469 |
+
"loss": 0.2138,
|
| 470 |
+
"step": 1300
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 1.066447908121411,
|
| 474 |
+
"eval_loss": 0.2270548790693283,
|
| 475 |
+
"eval_runtime": 22.5165,
|
| 476 |
+
"eval_samples_per_second": 44.412,
|
| 477 |
+
"eval_steps_per_second": 0.711,
|
| 478 |
+
"step": 1300
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 1.0869565217391304,
|
| 482 |
+
"grad_norm": 0.12517733871936798,
|
| 483 |
+
"learning_rate": 0.0010243399274756564,
|
| 484 |
+
"loss": 0.2111,
|
| 485 |
+
"step": 1325
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 1.1074651353568499,
|
| 489 |
+
"grad_norm": 0.09303736686706543,
|
| 490 |
+
"learning_rate": 0.0009885450357469806,
|
| 491 |
+
"loss": 0.2043,
|
| 492 |
+
"step": 1350
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 1.1279737489745694,
|
| 496 |
+
"grad_norm": 0.11305887997150421,
|
| 497 |
+
"learning_rate": 0.0009527648215787065,
|
| 498 |
+
"loss": 0.2057,
|
| 499 |
+
"step": 1375
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 1.1484823625922886,
|
| 503 |
+
"grad_norm": 0.1038450226187706,
|
| 504 |
+
"learning_rate": 0.000917045131142242,
|
| 505 |
+
"loss": 0.1984,
|
| 506 |
+
"step": 1400
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 1.1484823625922886,
|
| 510 |
+
"eval_loss": 0.22505907714366913,
|
| 511 |
+
"eval_runtime": 22.5688,
|
| 512 |
+
"eval_samples_per_second": 44.309,
|
| 513 |
+
"eval_steps_per_second": 0.709,
|
| 514 |
+
"step": 1400
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 1.1689909762100081,
|
| 518 |
+
"grad_norm": 0.10388393700122833,
|
| 519 |
+
"learning_rate": 0.0008814317330582753,
|
| 520 |
+
"loss": 0.2092,
|
| 521 |
+
"step": 1425
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 1.1894995898277276,
|
| 525 |
+
"grad_norm": 0.1294604390859604,
|
| 526 |
+
"learning_rate": 0.000845970259752183,
|
| 527 |
+
"loss": 0.2107,
|
| 528 |
+
"step": 1450
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 1.2100082034454471,
|
| 532 |
+
"grad_norm": 0.1135207936167717,
|
| 533 |
+
"learning_rate": 0.0008107061489839498,
|
| 534 |
+
"loss": 0.2069,
|
| 535 |
+
"step": 1475
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 1.2305168170631666,
|
| 539 |
+
"grad_norm": 0.11284071952104568,
|
| 540 |
+
"learning_rate": 0.0007756845856275194,
|
| 541 |
+
"loss": 0.2169,
|
| 542 |
+
"step": 1500
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 1.2305168170631666,
|
| 546 |
+
"eval_loss": 0.2215997278690338,
|
| 547 |
+
"eval_runtime": 22.56,
|
| 548 |
+
"eval_samples_per_second": 44.326,
|
| 549 |
+
"eval_steps_per_second": 0.709,
|
| 550 |
+
"step": 1500
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 1.251025430680886,
|
| 554 |
+
"grad_norm": 0.1196022480726242,
|
| 555 |
+
"learning_rate": 0.0007409504437741722,
|
| 556 |
+
"loss": 0.21,
|
| 557 |
+
"step": 1525
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 1.2715340442986054,
|
| 561 |
+
"grad_norm": 0.10071329027414322,
|
| 562 |
+
"learning_rate": 0.0007065482292341205,
|
| 563 |
+
"loss": 0.2027,
|
| 564 |
+
"step": 1550
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 1.2920426579163249,
|
| 568 |
+
"grad_norm": 0.09619199484586716,
|
| 569 |
+
"learning_rate": 0.0006725220225099911,
|
| 570 |
+
"loss": 0.2052,
|
| 571 |
+
"step": 1575
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 1.3125512715340442,
|
| 575 |
+
"grad_norm": 0.09156788140535355,
|
| 576 |
+
"learning_rate": 0.0006389154223152666,
|
| 577 |
+
"loss": 0.1987,
|
| 578 |
+
"step": 1600
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 1.3125512715340442,
|
| 582 |
+
"eval_loss": 0.21827217936515808,
|
| 583 |
+
"eval_runtime": 22.5451,
|
| 584 |
+
"eval_samples_per_second": 44.356,
|
| 585 |
+
"eval_steps_per_second": 0.71,
|
| 586 |
+
"step": 1600
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 1.3330598851517639,
|
| 590 |
+
"grad_norm": 0.09059920907020569,
|
| 591 |
+
"learning_rate": 0.0006057714897100551,
|
| 592 |
+
"loss": 0.201,
|
| 593 |
+
"step": 1625
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 1.3535684987694832,
|
| 597 |
+
"grad_norm": 0.10633113235235214,
|
| 598 |
+
"learning_rate": 0.0005731326929257713,
|
| 599 |
+
"loss": 0.2022,
|
| 600 |
+
"step": 1650
|
| 601 |
+
},
|
| 602 |
+
{
|
| 603 |
+
"epoch": 1.3740771123872026,
|
| 604 |
+
"grad_norm": 0.10689054429531097,
|
| 605 |
+
"learning_rate": 0.0005410408529494251,
|
| 606 |
+
"loss": 0.2001,
|
| 607 |
+
"step": 1675
|
| 608 |
+
},
|
| 609 |
+
{
|
| 610 |
+
"epoch": 1.3945857260049221,
|
| 611 |
+
"grad_norm": 0.10712600499391556,
|
| 612 |
+
"learning_rate": 0.0005095370899372412,
|
| 613 |
+
"loss": 0.2002,
|
| 614 |
+
"step": 1700
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 1.3945857260049221,
|
| 618 |
+
"eval_loss": 0.2159736305475235,
|
| 619 |
+
"eval_runtime": 22.541,
|
| 620 |
+
"eval_samples_per_second": 44.364,
|
| 621 |
+
"eval_steps_per_second": 0.71,
|
| 622 |
+
"step": 1700
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"epoch": 1.4150943396226414,
|
| 626 |
+
"grad_norm": 0.0958191528916359,
|
| 627 |
+
"learning_rate": 0.0004786617705262746,
|
| 628 |
+
"loss": 0.1979,
|
| 629 |
+
"step": 1725
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 1.435602953240361,
|
| 633 |
+
"grad_norm": 0.097678542137146,
|
| 634 |
+
"learning_rate": 0.000448454456111529,
|
| 635 |
+
"loss": 0.1957,
|
| 636 |
+
"step": 1750
|
| 637 |
+
},
|
| 638 |
+
{
|
| 639 |
+
"epoch": 1.4561115668580804,
|
| 640 |
+
"grad_norm": 0.09663794934749603,
|
| 641 |
+
"learning_rate": 0.0004189538521548524,
|
| 642 |
+
"loss": 0.2034,
|
| 643 |
+
"step": 1775
|
| 644 |
+
},
|
| 645 |
+
{
|
| 646 |
+
"epoch": 1.4766201804758,
|
| 647 |
+
"grad_norm": 0.09802096337080002,
|
| 648 |
+
"learning_rate": 0.00039019775859056916,
|
| 649 |
+
"loss": 0.1927,
|
| 650 |
+
"step": 1800
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 1.4766201804758,
|
| 654 |
+
"eval_loss": 0.2144840955734253,
|
| 655 |
+
"eval_runtime": 22.5527,
|
| 656 |
+
"eval_samples_per_second": 44.341,
|
| 657 |
+
"eval_steps_per_second": 0.709,
|
| 658 |
+
"step": 1800
|
| 659 |
+
},
|
| 660 |
+
{
|
| 661 |
+
"epoch": 1.4971287940935194,
|
| 662 |
+
"grad_norm": 0.09784867614507675,
|
| 663 |
+
"learning_rate": 0.0003622230213913836,
|
| 664 |
+
"loss": 0.1917,
|
| 665 |
+
"step": 1825
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 1.5176374077112387,
|
| 669 |
+
"grad_norm": 0.09651490300893784,
|
| 670 |
+
"learning_rate": 0.0003350654853566223,
|
| 671 |
+
"loss": 0.1944,
|
| 672 |
+
"step": 1850
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 1.5381460213289582,
|
| 676 |
+
"grad_norm": 0.10981660336256027,
|
| 677 |
+
"learning_rate": 0.00030875994818330957,
|
| 678 |
+
"loss": 0.1958,
|
| 679 |
+
"step": 1875
|
| 680 |
+
},
|
| 681 |
+
{
|
| 682 |
+
"epoch": 1.5586546349466777,
|
| 683 |
+
"grad_norm": 0.11364647001028061,
|
| 684 |
+
"learning_rate": 0.0002833401158789207,
|
| 685 |
+
"loss": 0.1985,
|
| 686 |
+
"step": 1900
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 1.5586546349466777,
|
| 690 |
+
"eval_loss": 0.21162408590316772,
|
| 691 |
+
"eval_runtime": 22.5558,
|
| 692 |
+
"eval_samples_per_second": 44.334,
|
| 693 |
+
"eval_steps_per_second": 0.709,
|
| 694 |
+
"step": 1900
|
| 695 |
+
},
|
| 696 |
+
{
|
| 697 |
+
"epoch": 1.579163248564397,
|
| 698 |
+
"grad_norm": 0.10284125059843063,
|
| 699 |
+
"learning_rate": 0.00025883855957295053,
|
| 700 |
+
"loss": 0.1977,
|
| 701 |
+
"step": 1925
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"epoch": 1.5996718621821167,
|
| 705 |
+
"grad_norm": 0.1027180403470993,
|
| 706 |
+
"learning_rate": 0.0002352866737826277,
|
| 707 |
+
"loss": 0.1977,
|
| 708 |
+
"step": 1950
|
| 709 |
+
},
|
| 710 |
+
{
|
| 711 |
+
"epoch": 1.620180475799836,
|
| 712 |
+
"grad_norm": 0.0985800251364708,
|
| 713 |
+
"learning_rate": 0.00021271463618625986,
|
| 714 |
+
"loss": 0.1926,
|
| 715 |
+
"step": 1975
|
| 716 |
+
},
|
| 717 |
+
{
|
| 718 |
+
"epoch": 1.6406890894175554,
|
| 719 |
+
"grad_norm": 0.10515035688877106,
|
| 720 |
+
"learning_rate": 0.00019115136895574402,
|
| 721 |
+
"loss": 0.1991,
|
| 722 |
+
"step": 2000
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 1.6406890894175554,
|
| 726 |
+
"eval_loss": 0.20991793274879456,
|
| 727 |
+
"eval_runtime": 22.5479,
|
| 728 |
+
"eval_samples_per_second": 44.35,
|
| 729 |
+
"eval_steps_per_second": 0.71,
|
| 730 |
+
"step": 2000
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 1.661197703035275,
|
| 734 |
+
"grad_norm": 0.10674113035202026,
|
| 735 |
+
"learning_rate": 0.0001706245016977931,
|
| 736 |
+
"loss": 0.1886,
|
| 737 |
+
"step": 2025
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 1.6817063166529942,
|
| 741 |
+
"grad_norm": 0.11014382541179657,
|
| 742 |
+
"learning_rate": 0.00015116033605136182,
|
| 743 |
+
"loss": 0.191,
|
| 744 |
+
"step": 2050
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 1.7022149302707137,
|
| 748 |
+
"grad_norm": 0.11587057262659073,
|
| 749 |
+
"learning_rate": 0.00013278381198663492,
|
| 750 |
+
"loss": 0.1971,
|
| 751 |
+
"step": 2075
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 1.7227235438884332,
|
| 755 |
+
"grad_norm": 0.11261642724275589,
|
| 756 |
+
"learning_rate": 0.0001155184758487573,
|
| 757 |
+
"loss": 0.1868,
|
| 758 |
+
"step": 2100
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 1.7227235438884332,
|
| 762 |
+
"eval_loss": 0.20869949460029602,
|
| 763 |
+
"eval_runtime": 22.5509,
|
| 764 |
+
"eval_samples_per_second": 44.344,
|
| 765 |
+
"eval_steps_per_second": 0.71,
|
| 766 |
+
"step": 2100
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 1.7432321575061525,
|
| 770 |
+
"grad_norm": 0.0879945233464241,
|
| 771 |
+
"learning_rate": 9.938645018725523e-05,
|
| 772 |
+
"loss": 0.1903,
|
| 773 |
+
"step": 2125
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 1.7637407711238722,
|
| 777 |
+
"grad_norm": 0.08777210116386414,
|
| 778 |
+
"learning_rate": 8.440840540980587e-05,
|
| 779 |
+
"loss": 0.1882,
|
| 780 |
+
"step": 2150
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 1.7842493847415914,
|
| 784 |
+
"grad_norm": 0.09200013428926468,
|
| 785 |
+
"learning_rate": 7.060353329667668e-05,
|
| 786 |
+
"loss": 0.197,
|
| 787 |
+
"step": 2175
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 1.804757998359311,
|
| 791 |
+
"grad_norm": 0.09770681709051132,
|
| 792 |
+
"learning_rate": 5.798952240976951e-05,
|
| 793 |
+
"loss": 0.1905,
|
| 794 |
+
"step": 2200
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 1.804757998359311,
|
| 798 |
+
"eval_loss": 0.2074345052242279,
|
| 799 |
+
"eval_runtime": 22.5502,
|
| 800 |
+
"eval_samples_per_second": 44.345,
|
| 801 |
+
"eval_steps_per_second": 0.71,
|
| 802 |
+
"step": 2200
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 1.8252666119770304,
|
| 806 |
+
"grad_norm": 0.10603570193052292,
|
| 807 |
+
"learning_rate": 4.65825354277799e-05,
|
| 808 |
+
"loss": 0.191,
|
| 809 |
+
"step": 2225
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 1.8457752255947497,
|
| 813 |
+
"grad_norm": 0.09987975656986237,
|
| 814 |
+
"learning_rate": 3.639718843651363e-05,
|
| 815 |
+
"loss": 0.1925,
|
| 816 |
+
"step": 2250
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"epoch": 1.8662838392124692,
|
| 820 |
+
"grad_norm": 0.09468022733926773,
|
| 821 |
+
"learning_rate": 2.7446532200894104e-05,
|
| 822 |
+
"loss": 0.1975,
|
| 823 |
+
"step": 2275
|
| 824 |
+
},
|
| 825 |
+
{
|
| 826 |
+
"epoch": 1.8867924528301887,
|
| 827 |
+
"grad_norm": 0.09474539756774902,
|
| 828 |
+
"learning_rate": 1.9742035442658403e-05,
|
| 829 |
+
"loss": 0.1902,
|
| 830 |
+
"step": 2300
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 1.8867924528301887,
|
| 834 |
+
"eval_loss": 0.20703136920928955,
|
| 835 |
+
"eval_runtime": 22.5531,
|
| 836 |
+
"eval_samples_per_second": 44.34,
|
| 837 |
+
"eval_steps_per_second": 0.709,
|
| 838 |
+
"step": 2300
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 1.907301066447908,
|
| 842 |
+
"grad_norm": 0.11737816035747528,
|
| 843 |
+
"learning_rate": 1.3293570145169742e-05,
|
| 844 |
+
"loss": 0.1983,
|
| 845 |
+
"step": 2325
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 1.9278096800656277,
|
| 849 |
+
"grad_norm": 0.09696778655052185,
|
| 850 |
+
"learning_rate": 8.109398904173282e-06,
|
| 851 |
+
"loss": 0.1836,
|
| 852 |
+
"step": 2350
|
| 853 |
+
},
|
| 854 |
+
{
|
| 855 |
+
"epoch": 1.948318293683347,
|
| 856 |
+
"grad_norm": 0.09188945591449738,
|
| 857 |
+
"learning_rate": 4.196164340705577e-06,
|
| 858 |
+
"loss": 0.1888,
|
| 859 |
+
"step": 2375
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 1.9688269073010665,
|
| 863 |
+
"grad_norm": 0.0918751060962677,
|
| 864 |
+
"learning_rate": 1.5588805897215342e-06,
|
| 865 |
+
"loss": 0.19,
|
| 866 |
+
"step": 2400
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 1.9688269073010665,
|
| 870 |
+
"eval_loss": 0.20687735080718994,
|
| 871 |
+
"eval_runtime": 22.5488,
|
| 872 |
+
"eval_samples_per_second": 44.348,
|
| 873 |
+
"eval_steps_per_second": 0.71,
|
| 874 |
+
"step": 2400
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"epoch": 1.989335520918786,
|
| 878 |
+
"grad_norm": 0.09456487745046616,
|
| 879 |
+
"learning_rate": 2.0092687534589705e-07,
|
| 880 |
+
"loss": 0.1866,
|
| 881 |
+
"step": 2425
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 2.0,
|
| 885 |
+
"step": 2438,
|
| 886 |
+
"total_flos": 1.58523627405312e+18,
|
| 887 |
+
"train_loss": 0.25376511950097774,
|
| 888 |
+
"train_runtime": 3753.3089,
|
| 889 |
+
"train_samples_per_second": 20.782,
|
| 890 |
+
"train_steps_per_second": 0.65
|
| 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_ex10/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": 8,
|
| 9 |
+
"peft_type": "ROTATION",
|
| 10 |
+
"r": 2,
|
| 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_ex10/ft/added_tokens.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"[PAD]": 32000
|
| 3 |
+
}
|
nl_tasks/exps/run_ex10/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_ex10/ft/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
nl_tasks/exps/run_ex10/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_ex10/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_ex10/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": 8,
|
| 9 |
+
"peft_type": "ROTATION",
|
| 10 |
+
"r": 2,
|
| 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_ex10/ft2/adapter_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:027d0ed23f8f468830610ee1ca0346c6399a5df096cde6bfb4c2c5db4cf32523
|
| 3 |
+
size 33602659
|
nl_tasks/exps/run_ex10/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": 0.15523949265480042,
|
| 15 |
+
"learning_rate": 0.00019672131147540983,
|
| 16 |
+
"loss": 0.6265,
|
| 17 |
+
"step": 25
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.04101722723543889,
|
| 21 |
+
"grad_norm": 0.21520094573497772,
|
| 22 |
+
"learning_rate": 0.00040163934426229507,
|
| 23 |
+
"loss": 0.3854,
|
| 24 |
+
"step": 50
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.06152584085315833,
|
| 28 |
+
"grad_norm": 0.151475727558136,
|
| 29 |
+
"learning_rate": 0.0006065573770491804,
|
| 30 |
+
"loss": 0.3599,
|
| 31 |
+
"step": 75
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.08203445447087777,
|
| 35 |
+
"grad_norm": 0.17690438032150269,
|
| 36 |
+
"learning_rate": 0.0008114754098360656,
|
| 37 |
+
"loss": 0.329,
|
| 38 |
+
"step": 100
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.08203445447087777,
|
| 42 |
+
"eval_loss": 0.3202120065689087,
|
| 43 |
+
"eval_runtime": 20.9606,
|
| 44 |
+
"eval_samples_per_second": 47.709,
|
| 45 |
+
"eval_steps_per_second": 0.763,
|
| 46 |
+
"step": 100
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"epoch": 0.10254306808859721,
|
| 50 |
+
"grad_norm": 0.2374378889799118,
|
| 51 |
+
"learning_rate": 0.0010163934426229509,
|
| 52 |
+
"loss": 0.3405,
|
| 53 |
+
"step": 125
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 0.12305168170631665,
|
| 57 |
+
"grad_norm": 0.5405275225639343,
|
| 58 |
+
"learning_rate": 0.001221311475409836,
|
| 59 |
+
"loss": 0.3313,
|
| 60 |
+
"step": 150
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 0.1435602953240361,
|
| 64 |
+
"grad_norm": 0.4715682566165924,
|
| 65 |
+
"learning_rate": 0.0014262295081967215,
|
| 66 |
+
"loss": 0.35,
|
| 67 |
+
"step": 175
|
| 68 |
+
},
|
| 69 |
+
{
|
| 70 |
+
"epoch": 0.16406890894175555,
|
| 71 |
+
"grad_norm": 0.40690892934799194,
|
| 72 |
+
"learning_rate": 0.0016311475409836065,
|
| 73 |
+
"loss": 0.4078,
|
| 74 |
+
"step": 200
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.16406890894175555,
|
| 78 |
+
"eval_loss": 0.34672215580940247,
|
| 79 |
+
"eval_runtime": 20.7767,
|
| 80 |
+
"eval_samples_per_second": 48.131,
|
| 81 |
+
"eval_steps_per_second": 0.77,
|
| 82 |
+
"step": 200
|
| 83 |
+
},
|
| 84 |
+
{
|
| 85 |
+
"epoch": 0.184577522559475,
|
| 86 |
+
"grad_norm": 0.33339014649391174,
|
| 87 |
+
"learning_rate": 0.0018360655737704918,
|
| 88 |
+
"loss": 0.3636,
|
| 89 |
+
"step": 225
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 0.20508613617719443,
|
| 93 |
+
"grad_norm": 0.38985753059387207,
|
| 94 |
+
"learning_rate": 0.0019999743708232127,
|
| 95 |
+
"loss": 0.3419,
|
| 96 |
+
"step": 250
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"epoch": 0.22559474979491387,
|
| 100 |
+
"grad_norm": 0.31976866722106934,
|
| 101 |
+
"learning_rate": 0.0019990774875676054,
|
| 102 |
+
"loss": 0.3374,
|
| 103 |
+
"step": 275
|
| 104 |
+
},
|
| 105 |
+
{
|
| 106 |
+
"epoch": 0.2461033634126333,
|
| 107 |
+
"grad_norm": 0.29505324363708496,
|
| 108 |
+
"learning_rate": 0.001996900458879386,
|
| 109 |
+
"loss": 0.3297,
|
| 110 |
+
"step": 300
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.2461033634126333,
|
| 114 |
+
"eval_loss": 0.3135361075401306,
|
| 115 |
+
"eval_runtime": 20.8074,
|
| 116 |
+
"eval_samples_per_second": 48.06,
|
| 117 |
+
"eval_steps_per_second": 0.769,
|
| 118 |
+
"step": 300
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 0.2666119770303528,
|
| 122 |
+
"grad_norm": 0.2578265964984894,
|
| 123 |
+
"learning_rate": 0.001993446074245224,
|
| 124 |
+
"loss": 0.3261,
|
| 125 |
+
"step": 325
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 0.2871205906480722,
|
| 129 |
+
"grad_norm": 0.21807414293289185,
|
| 130 |
+
"learning_rate": 0.0019887187598630527,
|
| 131 |
+
"loss": 0.3108,
|
| 132 |
+
"step": 350
|
| 133 |
+
},
|
| 134 |
+
{
|
| 135 |
+
"epoch": 0.30762920426579166,
|
| 136 |
+
"grad_norm": 0.21876771748065948,
|
| 137 |
+
"learning_rate": 0.0019827245729706648,
|
| 138 |
+
"loss": 0.3187,
|
| 139 |
+
"step": 375
|
| 140 |
+
},
|
| 141 |
+
{
|
| 142 |
+
"epoch": 0.3281378178835111,
|
| 143 |
+
"grad_norm": 0.24306173622608185,
|
| 144 |
+
"learning_rate": 0.0019754711940844047,
|
| 145 |
+
"loss": 0.3001,
|
| 146 |
+
"step": 400
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.3281378178835111,
|
| 150 |
+
"eval_loss": 0.2858148515224457,
|
| 151 |
+
"eval_runtime": 20.8154,
|
| 152 |
+
"eval_samples_per_second": 48.041,
|
| 153 |
+
"eval_steps_per_second": 0.769,
|
| 154 |
+
"step": 400
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 0.34864643150123054,
|
| 158 |
+
"grad_norm": 0.18788783252239227,
|
| 159 |
+
"learning_rate": 0.0019669679171579117,
|
| 160 |
+
"loss": 0.3047,
|
| 161 |
+
"step": 425
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 0.36915504511895,
|
| 165 |
+
"grad_norm": 0.15331535041332245,
|
| 166 |
+
"learning_rate": 0.001957225637673524,
|
| 167 |
+
"loss": 0.2987,
|
| 168 |
+
"step": 450
|
| 169 |
+
},
|
| 170 |
+
{
|
| 171 |
+
"epoch": 0.3896636587366694,
|
| 172 |
+
"grad_norm": 0.16196906566619873,
|
| 173 |
+
"learning_rate": 0.0019462568386815961,
|
| 174 |
+
"loss": 0.2855,
|
| 175 |
+
"step": 475
|
| 176 |
+
},
|
| 177 |
+
{
|
| 178 |
+
"epoch": 0.41017227235438886,
|
| 179 |
+
"grad_norm": 0.16386869549751282,
|
| 180 |
+
"learning_rate": 0.0019340755748056234,
|
| 181 |
+
"loss": 0.2694,
|
| 182 |
+
"step": 500
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.41017227235438886,
|
| 186 |
+
"eval_loss": 0.271730899810791,
|
| 187 |
+
"eval_runtime": 20.8073,
|
| 188 |
+
"eval_samples_per_second": 48.06,
|
| 189 |
+
"eval_steps_per_second": 0.769,
|
| 190 |
+
"step": 500
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 0.4306808859721083,
|
| 194 |
+
"grad_norm": 0.17943786084651947,
|
| 195 |
+
"learning_rate": 0.0019206974542336672,
|
| 196 |
+
"loss": 0.2685,
|
| 197 |
+
"step": 525
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 0.45118949958982774,
|
| 201 |
+
"grad_norm": 0.1618247926235199,
|
| 202 |
+
"learning_rate": 0.0019061396187191563,
|
| 203 |
+
"loss": 0.278,
|
| 204 |
+
"step": 550
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 0.4716981132075472,
|
| 208 |
+
"grad_norm": 0.15906882286071777,
|
| 209 |
+
"learning_rate": 0.0018904207216166836,
|
| 210 |
+
"loss": 0.2689,
|
| 211 |
+
"step": 575
|
| 212 |
+
},
|
| 213 |
+
{
|
| 214 |
+
"epoch": 0.4922067268252666,
|
| 215 |
+
"grad_norm": 0.1632314771413803,
|
| 216 |
+
"learning_rate": 0.001873560903980955,
|
| 217 |
+
"loss": 0.2863,
|
| 218 |
+
"step": 600
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.4922067268252666,
|
| 222 |
+
"eval_loss": 0.2613782286643982,
|
| 223 |
+
"eval_runtime": 20.8203,
|
| 224 |
+
"eval_samples_per_second": 48.03,
|
| 225 |
+
"eval_steps_per_second": 0.768,
|
| 226 |
+
"step": 600
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.5127153404429861,
|
| 230 |
+
"grad_norm": 0.13561119139194489,
|
| 231 |
+
"learning_rate": 0.0018555817687594984,
|
| 232 |
+
"loss": 0.2661,
|
| 233 |
+
"step": 625
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.5332239540607056,
|
| 237 |
+
"grad_norm": 0.12830308079719543,
|
| 238 |
+
"learning_rate": 0.0018365063531122169,
|
| 239 |
+
"loss": 0.2565,
|
| 240 |
+
"step": 650
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.5537325676784249,
|
| 244 |
+
"grad_norm": 0.13453687727451324,
|
| 245 |
+
"learning_rate": 0.0018163590988932402,
|
| 246 |
+
"loss": 0.2695,
|
| 247 |
+
"step": 675
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.5742411812961444,
|
| 251 |
+
"grad_norm": 0.11978261917829514,
|
| 252 |
+
"learning_rate": 0.0017951658213329078,
|
| 253 |
+
"loss": 0.2661,
|
| 254 |
+
"step": 700
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.5742411812961444,
|
| 258 |
+
"eval_loss": 0.2536289393901825,
|
| 259 |
+
"eval_runtime": 20.8091,
|
| 260 |
+
"eval_samples_per_second": 48.056,
|
| 261 |
+
"eval_steps_per_second": 0.769,
|
| 262 |
+
"step": 700
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.5947497949138638,
|
| 266 |
+
"grad_norm": 0.12471190840005875,
|
| 267 |
+
"learning_rate": 0.0017729536759600033,
|
| 268 |
+
"loss": 0.2633,
|
| 269 |
+
"step": 725
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.6152584085315833,
|
| 273 |
+
"grad_norm": 0.13270163536071777,
|
| 274 |
+
"learning_rate": 0.0017497511238066307,
|
| 275 |
+
"loss": 0.2606,
|
| 276 |
+
"step": 750
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.6357670221493027,
|
| 280 |
+
"grad_norm": 0.12368787080049515,
|
| 281 |
+
"learning_rate": 0.00172558789494031,
|
| 282 |
+
"loss": 0.257,
|
| 283 |
+
"step": 775
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.6562756357670222,
|
| 287 |
+
"grad_norm": 0.13092003762722015,
|
| 288 |
+
"learning_rate": 0.0017004949503700284,
|
| 289 |
+
"loss": 0.2608,
|
| 290 |
+
"step": 800
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.6562756357670222,
|
| 294 |
+
"eval_loss": 0.24879904091358185,
|
| 295 |
+
"eval_runtime": 20.8195,
|
| 296 |
+
"eval_samples_per_second": 48.032,
|
| 297 |
+
"eval_steps_per_second": 0.769,
|
| 298 |
+
"step": 800
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 0.6767842493847416,
|
| 302 |
+
"grad_norm": 0.12192385643720627,
|
| 303 |
+
"learning_rate": 0.0016745044423750449,
|
| 304 |
+
"loss": 0.2549,
|
| 305 |
+
"step": 825
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 0.6972928630024611,
|
| 309 |
+
"grad_norm": 0.10963451862335205,
|
| 310 |
+
"learning_rate": 0.0016476496733072946,
|
| 311 |
+
"loss": 0.2582,
|
| 312 |
+
"step": 850
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 0.7178014766201805,
|
| 316 |
+
"grad_norm": 0.10306356102228165,
|
| 317 |
+
"learning_rate": 0.0016199650529201684,
|
| 318 |
+
"loss": 0.2449,
|
| 319 |
+
"step": 875
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 0.7383100902379,
|
| 323 |
+
"grad_norm": 0.12656815350055695,
|
| 324 |
+
"learning_rate": 0.0015914860542783522,
|
| 325 |
+
"loss": 0.2498,
|
| 326 |
+
"step": 900
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 0.7383100902379,
|
| 330 |
+
"eval_loss": 0.24096110463142395,
|
| 331 |
+
"eval_runtime": 20.8014,
|
| 332 |
+
"eval_samples_per_second": 48.074,
|
| 333 |
+
"eval_steps_per_second": 0.769,
|
| 334 |
+
"step": 900
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"epoch": 0.7588187038556193,
|
| 338 |
+
"grad_norm": 0.1108182966709137,
|
| 339 |
+
"learning_rate": 0.0015622491683052124,
|
| 340 |
+
"loss": 0.2524,
|
| 341 |
+
"step": 925
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 0.7793273174733388,
|
| 345 |
+
"grad_norm": 0.11574766784906387,
|
| 346 |
+
"learning_rate": 0.0015322918570259759,
|
| 347 |
+
"loss": 0.2411,
|
| 348 |
+
"step": 950
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"epoch": 0.7998359310910582,
|
| 352 |
+
"grad_norm": 0.11123888194561005,
|
| 353 |
+
"learning_rate": 0.0015016525055666057,
|
| 354 |
+
"loss": 0.2467,
|
| 355 |
+
"step": 975
|
| 356 |
+
},
|
| 357 |
+
{
|
| 358 |
+
"epoch": 0.8203445447087777,
|
| 359 |
+
"grad_norm": 0.12683457136154175,
|
| 360 |
+
"learning_rate": 0.001470370372969886,
|
| 361 |
+
"loss": 0.2399,
|
| 362 |
+
"step": 1000
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.8203445447087777,
|
| 366 |
+
"eval_loss": 0.23732976615428925,
|
| 367 |
+
"eval_runtime": 20.8008,
|
| 368 |
+
"eval_samples_per_second": 48.075,
|
| 369 |
+
"eval_steps_per_second": 0.769,
|
| 370 |
+
"step": 1000
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 0.8408531583264971,
|
| 374 |
+
"grad_norm": 0.11887123435735703,
|
| 375 |
+
"learning_rate": 0.0014384855418917311,
|
| 376 |
+
"loss": 0.2402,
|
| 377 |
+
"step": 1025
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 0.8613617719442166,
|
| 381 |
+
"grad_norm": 0.1123674139380455,
|
| 382 |
+
"learning_rate": 0.0014060388672421775,
|
| 383 |
+
"loss": 0.2376,
|
| 384 |
+
"step": 1050
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.881870385561936,
|
| 388 |
+
"grad_norm": 0.10185591876506805,
|
| 389 |
+
"learning_rate": 0.0013730719238368662,
|
| 390 |
+
"loss": 0.243,
|
| 391 |
+
"step": 1075
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 0.9023789991796555,
|
| 395 |
+
"grad_norm": 0.10748770833015442,
|
| 396 |
+
"learning_rate": 0.0013396269531260867,
|
| 397 |
+
"loss": 0.2432,
|
| 398 |
+
"step": 1100
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 0.9023789991796555,
|
| 402 |
+
"eval_loss": 0.23266206681728363,
|
| 403 |
+
"eval_runtime": 20.7925,
|
| 404 |
+
"eval_samples_per_second": 48.094,
|
| 405 |
+
"eval_steps_per_second": 0.77,
|
| 406 |
+
"step": 1100
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 0.9228876127973749,
|
| 410 |
+
"grad_norm": 0.10115822404623032,
|
| 411 |
+
"learning_rate": 0.0013057468090696496,
|
| 412 |
+
"loss": 0.2373,
|
| 413 |
+
"step": 1125
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 0.9433962264150944,
|
| 417 |
+
"grad_norm": 0.09614510089159012,
|
| 418 |
+
"learning_rate": 0.0012714749032269287,
|
| 419 |
+
"loss": 0.2378,
|
| 420 |
+
"step": 1150
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 0.9639048400328137,
|
| 424 |
+
"grad_norm": 0.09826623648405075,
|
| 425 |
+
"learning_rate": 0.0012368551491324358,
|
| 426 |
+
"loss": 0.2437,
|
| 427 |
+
"step": 1175
|
| 428 |
+
},
|
| 429 |
+
{
|
| 430 |
+
"epoch": 0.9844134536505332,
|
| 431 |
+
"grad_norm": 0.10577023774385452,
|
| 432 |
+
"learning_rate": 0.0012019319060282063,
|
| 433 |
+
"loss": 0.2481,
|
| 434 |
+
"step": 1200
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 0.9844134536505332,
|
| 438 |
+
"eval_loss": 0.22884155809879303,
|
| 439 |
+
"eval_runtime": 20.8114,
|
| 440 |
+
"eval_samples_per_second": 48.051,
|
| 441 |
+
"eval_steps_per_second": 0.769,
|
| 442 |
+
"step": 1200
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 1.0049220672682526,
|
| 446 |
+
"grad_norm": 0.10433684289455414,
|
| 447 |
+
"learning_rate": 0.0011667499220250803,
|
| 448 |
+
"loss": 0.228,
|
| 449 |
+
"step": 1225
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 1.0254306808859721,
|
| 453 |
+
"grad_norm": 0.1063525378704071,
|
| 454 |
+
"learning_rate": 0.0011313542767657204,
|
| 455 |
+
"loss": 0.2073,
|
| 456 |
+
"step": 1250
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 1.0459392945036916,
|
| 460 |
+
"grad_norm": 0.10519979894161224,
|
| 461 |
+
"learning_rate": 0.0010957903236628267,
|
| 462 |
+
"loss": 0.2087,
|
| 463 |
+
"step": 1275
|
| 464 |
+
},
|
| 465 |
+
{
|
| 466 |
+
"epoch": 1.066447908121411,
|
| 467 |
+
"grad_norm": 0.09157584607601166,
|
| 468 |
+
"learning_rate": 0.001060103631786563,
|
| 469 |
+
"loss": 0.2124,
|
| 470 |
+
"step": 1300
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 1.066447908121411,
|
| 474 |
+
"eval_loss": 0.22583141922950745,
|
| 475 |
+
"eval_runtime": 20.7975,
|
| 476 |
+
"eval_samples_per_second": 48.083,
|
| 477 |
+
"eval_steps_per_second": 0.769,
|
| 478 |
+
"step": 1300
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 1.0869565217391304,
|
| 482 |
+
"grad_norm": 0.10928776115179062,
|
| 483 |
+
"learning_rate": 0.0010243399274756564,
|
| 484 |
+
"loss": 0.2075,
|
| 485 |
+
"step": 1325
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 1.1074651353568499,
|
| 489 |
+
"grad_norm": 0.08832115679979324,
|
| 490 |
+
"learning_rate": 0.0009885450357469806,
|
| 491 |
+
"loss": 0.202,
|
| 492 |
+
"step": 1350
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 1.1279737489745694,
|
| 496 |
+
"grad_norm": 0.11413701623678207,
|
| 497 |
+
"learning_rate": 0.0009527648215787065,
|
| 498 |
+
"loss": 0.2035,
|
| 499 |
+
"step": 1375
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 1.1484823625922886,
|
| 503 |
+
"grad_norm": 0.10007026791572571,
|
| 504 |
+
"learning_rate": 0.000917045131142242,
|
| 505 |
+
"loss": 0.1968,
|
| 506 |
+
"step": 1400
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 1.1484823625922886,
|
| 510 |
+
"eval_loss": 0.22363296151161194,
|
| 511 |
+
"eval_runtime": 20.8077,
|
| 512 |
+
"eval_samples_per_second": 48.059,
|
| 513 |
+
"eval_steps_per_second": 0.769,
|
| 514 |
+
"step": 1400
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 1.1689909762100081,
|
| 518 |
+
"grad_norm": 0.10129231214523315,
|
| 519 |
+
"learning_rate": 0.0008814317330582753,
|
| 520 |
+
"loss": 0.2074,
|
| 521 |
+
"step": 1425
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 1.1894995898277276,
|
| 525 |
+
"grad_norm": 0.11438316851854324,
|
| 526 |
+
"learning_rate": 0.000845970259752183,
|
| 527 |
+
"loss": 0.2074,
|
| 528 |
+
"step": 1450
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 1.2100082034454471,
|
| 532 |
+
"grad_norm": 0.11733181774616241,
|
| 533 |
+
"learning_rate": 0.0008107061489839498,
|
| 534 |
+
"loss": 0.2034,
|
| 535 |
+
"step": 1475
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 1.2305168170631666,
|
| 539 |
+
"grad_norm": 0.1269940584897995,
|
| 540 |
+
"learning_rate": 0.0007756845856275194,
|
| 541 |
+
"loss": 0.2137,
|
| 542 |
+
"step": 1500
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 1.2305168170631666,
|
| 546 |
+
"eval_loss": 0.2212778776884079,
|
| 547 |
+
"eval_runtime": 20.7941,
|
| 548 |
+
"eval_samples_per_second": 48.09,
|
| 549 |
+
"eval_steps_per_second": 0.769,
|
| 550 |
+
"step": 1500
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 1.251025430680886,
|
| 554 |
+
"grad_norm": 0.11956606060266495,
|
| 555 |
+
"learning_rate": 0.0007409504437741722,
|
| 556 |
+
"loss": 0.208,
|
| 557 |
+
"step": 1525
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 1.2715340442986054,
|
| 561 |
+
"grad_norm": 0.1041678637266159,
|
| 562 |
+
"learning_rate": 0.0007065482292341205,
|
| 563 |
+
"loss": 0.2022,
|
| 564 |
+
"step": 1550
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 1.2920426579163249,
|
| 568 |
+
"grad_norm": 0.09209053963422775,
|
| 569 |
+
"learning_rate": 0.0006725220225099911,
|
| 570 |
+
"loss": 0.2035,
|
| 571 |
+
"step": 1575
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 1.3125512715340442,
|
| 575 |
+
"grad_norm": 0.09528414160013199,
|
| 576 |
+
"learning_rate": 0.0006389154223152666,
|
| 577 |
+
"loss": 0.1963,
|
| 578 |
+
"step": 1600
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 1.3125512715340442,
|
| 582 |
+
"eval_loss": 0.21767942607402802,
|
| 583 |
+
"eval_runtime": 20.7917,
|
| 584 |
+
"eval_samples_per_second": 48.096,
|
| 585 |
+
"eval_steps_per_second": 0.77,
|
| 586 |
+
"step": 1600
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 1.3330598851517639,
|
| 590 |
+
"grad_norm": 0.09382138401269913,
|
| 591 |
+
"learning_rate": 0.0006057714897100551,
|
| 592 |
+
"loss": 0.1988,
|
| 593 |
+
"step": 1625
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 1.3535684987694832,
|
| 597 |
+
"grad_norm": 0.10053402185440063,
|
| 598 |
+
"learning_rate": 0.0005731326929257713,
|
| 599 |
+
"loss": 0.1998,
|
| 600 |
+
"step": 1650
|
| 601 |
+
},
|
| 602 |
+
{
|
| 603 |
+
"epoch": 1.3740771123872026,
|
| 604 |
+
"grad_norm": 0.11902140825986862,
|
| 605 |
+
"learning_rate": 0.0005410408529494251,
|
| 606 |
+
"loss": 0.1967,
|
| 607 |
+
"step": 1675
|
| 608 |
+
},
|
| 609 |
+
{
|
| 610 |
+
"epoch": 1.3945857260049221,
|
| 611 |
+
"grad_norm": 0.0958513393998146,
|
| 612 |
+
"learning_rate": 0.0005095370899372412,
|
| 613 |
+
"loss": 0.1977,
|
| 614 |
+
"step": 1700
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 1.3945857260049221,
|
| 618 |
+
"eval_loss": 0.21560746431350708,
|
| 619 |
+
"eval_runtime": 20.8207,
|
| 620 |
+
"eval_samples_per_second": 48.029,
|
| 621 |
+
"eval_steps_per_second": 0.768,
|
| 622 |
+
"step": 1700
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"epoch": 1.4150943396226414,
|
| 626 |
+
"grad_norm": 0.09698429703712463,
|
| 627 |
+
"learning_rate": 0.0004786617705262746,
|
| 628 |
+
"loss": 0.1948,
|
| 629 |
+
"step": 1725
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 1.435602953240361,
|
| 633 |
+
"grad_norm": 0.09873553365468979,
|
| 634 |
+
"learning_rate": 0.000448454456111529,
|
| 635 |
+
"loss": 0.1937,
|
| 636 |
+
"step": 1750
|
| 637 |
+
},
|
| 638 |
+
{
|
| 639 |
+
"epoch": 1.4561115668580804,
|
| 640 |
+
"grad_norm": 0.10226655006408691,
|
| 641 |
+
"learning_rate": 0.0004189538521548524,
|
| 642 |
+
"loss": 0.201,
|
| 643 |
+
"step": 1775
|
| 644 |
+
},
|
| 645 |
+
{
|
| 646 |
+
"epoch": 1.4766201804758,
|
| 647 |
+
"grad_norm": 0.08719471842050552,
|
| 648 |
+
"learning_rate": 0.00039019775859056916,
|
| 649 |
+
"loss": 0.1898,
|
| 650 |
+
"step": 1800
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 1.4766201804758,
|
| 654 |
+
"eval_loss": 0.21353399753570557,
|
| 655 |
+
"eval_runtime": 20.799,
|
| 656 |
+
"eval_samples_per_second": 48.079,
|
| 657 |
+
"eval_steps_per_second": 0.769,
|
| 658 |
+
"step": 1800
|
| 659 |
+
},
|
| 660 |
+
{
|
| 661 |
+
"epoch": 1.4971287940935194,
|
| 662 |
+
"grad_norm": 0.10064075142145157,
|
| 663 |
+
"learning_rate": 0.0003622230213913836,
|
| 664 |
+
"loss": 0.1885,
|
| 665 |
+
"step": 1825
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 1.5176374077112387,
|
| 669 |
+
"grad_norm": 0.08833151310682297,
|
| 670 |
+
"learning_rate": 0.0003350654853566223,
|
| 671 |
+
"loss": 0.1923,
|
| 672 |
+
"step": 1850
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 1.5381460213289582,
|
| 676 |
+
"grad_norm": 0.09507884830236435,
|
| 677 |
+
"learning_rate": 0.00030875994818330957,
|
| 678 |
+
"loss": 0.1927,
|
| 679 |
+
"step": 1875
|
| 680 |
+
},
|
| 681 |
+
{
|
| 682 |
+
"epoch": 1.5586546349466777,
|
| 683 |
+
"grad_norm": 0.11718809604644775,
|
| 684 |
+
"learning_rate": 0.0002833401158789207,
|
| 685 |
+
"loss": 0.1953,
|
| 686 |
+
"step": 1900
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 1.5586546349466777,
|
| 690 |
+
"eval_loss": 0.2108285129070282,
|
| 691 |
+
"eval_runtime": 20.7933,
|
| 692 |
+
"eval_samples_per_second": 48.092,
|
| 693 |
+
"eval_steps_per_second": 0.769,
|
| 694 |
+
"step": 1900
|
| 695 |
+
},
|
| 696 |
+
{
|
| 697 |
+
"epoch": 1.579163248564397,
|
| 698 |
+
"grad_norm": 0.09886983036994934,
|
| 699 |
+
"learning_rate": 0.00025883855957295053,
|
| 700 |
+
"loss": 0.1955,
|
| 701 |
+
"step": 1925
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"epoch": 1.5996718621821167,
|
| 705 |
+
"grad_norm": 0.10471342504024506,
|
| 706 |
+
"learning_rate": 0.0002352866737826277,
|
| 707 |
+
"loss": 0.1944,
|
| 708 |
+
"step": 1950
|
| 709 |
+
},
|
| 710 |
+
{
|
| 711 |
+
"epoch": 1.620180475799836,
|
| 712 |
+
"grad_norm": 0.09086298942565918,
|
| 713 |
+
"learning_rate": 0.00021271463618625986,
|
| 714 |
+
"loss": 0.1905,
|
| 715 |
+
"step": 1975
|
| 716 |
+
},
|
| 717 |
+
{
|
| 718 |
+
"epoch": 1.6406890894175554,
|
| 719 |
+
"grad_norm": 0.10715344548225403,
|
| 720 |
+
"learning_rate": 0.00019115136895574402,
|
| 721 |
+
"loss": 0.1958,
|
| 722 |
+
"step": 2000
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 1.6406890894175554,
|
| 726 |
+
"eval_loss": 0.20930151641368866,
|
| 727 |
+
"eval_runtime": 20.8145,
|
| 728 |
+
"eval_samples_per_second": 48.043,
|
| 729 |
+
"eval_steps_per_second": 0.769,
|
| 730 |
+
"step": 2000
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 1.661197703035275,
|
| 734 |
+
"grad_norm": 0.09305039048194885,
|
| 735 |
+
"learning_rate": 0.0001706245016977931,
|
| 736 |
+
"loss": 0.1872,
|
| 737 |
+
"step": 2025
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 1.6817063166529942,
|
| 741 |
+
"grad_norm": 0.11277252435684204,
|
| 742 |
+
"learning_rate": 0.00015116033605136182,
|
| 743 |
+
"loss": 0.1877,
|
| 744 |
+
"step": 2050
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 1.7022149302707137,
|
| 748 |
+
"grad_norm": 0.12639889121055603,
|
| 749 |
+
"learning_rate": 0.00013278381198663492,
|
| 750 |
+
"loss": 0.1938,
|
| 751 |
+
"step": 2075
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 1.7227235438884332,
|
| 755 |
+
"grad_norm": 0.09616611152887344,
|
| 756 |
+
"learning_rate": 0.0001155184758487573,
|
| 757 |
+
"loss": 0.1847,
|
| 758 |
+
"step": 2100
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 1.7227235438884332,
|
| 762 |
+
"eval_loss": 0.2082420438528061,
|
| 763 |
+
"eval_runtime": 20.7986,
|
| 764 |
+
"eval_samples_per_second": 48.08,
|
| 765 |
+
"eval_steps_per_second": 0.769,
|
| 766 |
+
"step": 2100
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 1.7432321575061525,
|
| 770 |
+
"grad_norm": 0.08913201838731766,
|
| 771 |
+
"learning_rate": 9.938645018725523e-05,
|
| 772 |
+
"loss": 0.1884,
|
| 773 |
+
"step": 2125
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 1.7637407711238722,
|
| 777 |
+
"grad_norm": 0.08310745656490326,
|
| 778 |
+
"learning_rate": 8.440840540980587e-05,
|
| 779 |
+
"loss": 0.1869,
|
| 780 |
+
"step": 2150
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 1.7842493847415914,
|
| 784 |
+
"grad_norm": 0.09385063499212265,
|
| 785 |
+
"learning_rate": 7.060353329667668e-05,
|
| 786 |
+
"loss": 0.1947,
|
| 787 |
+
"step": 2175
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 1.804757998359311,
|
| 791 |
+
"grad_norm": 0.10753221064805984,
|
| 792 |
+
"learning_rate": 5.798952240976951e-05,
|
| 793 |
+
"loss": 0.1888,
|
| 794 |
+
"step": 2200
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 1.804757998359311,
|
| 798 |
+
"eval_loss": 0.20682819187641144,
|
| 799 |
+
"eval_runtime": 20.8031,
|
| 800 |
+
"eval_samples_per_second": 48.07,
|
| 801 |
+
"eval_steps_per_second": 0.769,
|
| 802 |
+
"step": 2200
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 1.8252666119770304,
|
| 806 |
+
"grad_norm": 0.10841704905033112,
|
| 807 |
+
"learning_rate": 4.65825354277799e-05,
|
| 808 |
+
"loss": 0.1889,
|
| 809 |
+
"step": 2225
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 1.8457752255947497,
|
| 813 |
+
"grad_norm": 0.10467419773340225,
|
| 814 |
+
"learning_rate": 3.639718843651363e-05,
|
| 815 |
+
"loss": 0.1895,
|
| 816 |
+
"step": 2250
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"epoch": 1.8662838392124692,
|
| 820 |
+
"grad_norm": 0.09112375229597092,
|
| 821 |
+
"learning_rate": 2.7446532200894104e-05,
|
| 822 |
+
"loss": 0.1955,
|
| 823 |
+
"step": 2275
|
| 824 |
+
},
|
| 825 |
+
{
|
| 826 |
+
"epoch": 1.8867924528301887,
|
| 827 |
+
"grad_norm": 0.10222024470567703,
|
| 828 |
+
"learning_rate": 1.9742035442658403e-05,
|
| 829 |
+
"loss": 0.1874,
|
| 830 |
+
"step": 2300
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 1.8867924528301887,
|
| 834 |
+
"eval_loss": 0.20645059645175934,
|
| 835 |
+
"eval_runtime": 20.8102,
|
| 836 |
+
"eval_samples_per_second": 48.053,
|
| 837 |
+
"eval_steps_per_second": 0.769,
|
| 838 |
+
"step": 2300
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 1.907301066447908,
|
| 842 |
+
"grad_norm": 0.11472873389720917,
|
| 843 |
+
"learning_rate": 1.3293570145169742e-05,
|
| 844 |
+
"loss": 0.1953,
|
| 845 |
+
"step": 2325
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 1.9278096800656277,
|
| 849 |
+
"grad_norm": 0.09682969003915787,
|
| 850 |
+
"learning_rate": 8.109398904173282e-06,
|
| 851 |
+
"loss": 0.1815,
|
| 852 |
+
"step": 2350
|
| 853 |
+
},
|
| 854 |
+
{
|
| 855 |
+
"epoch": 1.948318293683347,
|
| 856 |
+
"grad_norm": 0.08157285302877426,
|
| 857 |
+
"learning_rate": 4.196164340705577e-06,
|
| 858 |
+
"loss": 0.1859,
|
| 859 |
+
"step": 2375
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 1.9688269073010665,
|
| 863 |
+
"grad_norm": 0.09527117013931274,
|
| 864 |
+
"learning_rate": 1.5588805897215342e-06,
|
| 865 |
+
"loss": 0.1878,
|
| 866 |
+
"step": 2400
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 1.9688269073010665,
|
| 870 |
+
"eval_loss": 0.20635543763637543,
|
| 871 |
+
"eval_runtime": 20.7984,
|
| 872 |
+
"eval_samples_per_second": 48.081,
|
| 873 |
+
"eval_steps_per_second": 0.769,
|
| 874 |
+
"step": 2400
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"epoch": 1.989335520918786,
|
| 878 |
+
"grad_norm": 0.09093104302883148,
|
| 879 |
+
"learning_rate": 2.0092687534589705e-07,
|
| 880 |
+
"loss": 0.1836,
|
| 881 |
+
"step": 2425
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 2.0,
|
| 885 |
+
"step": 2438,
|
| 886 |
+
"total_flos": 1.58523627405312e+18,
|
| 887 |
+
"train_loss": 0.24367707522801438,
|
| 888 |
+
"train_runtime": 3314.9991,
|
| 889 |
+
"train_samples_per_second": 23.529,
|
| 890 |
+
"train_steps_per_second": 0.735
|
| 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_ex11/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_ex11/ft/added_tokens.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"[PAD]": 32000
|
| 3 |
+
}
|
nl_tasks/exps/run_ex11/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_ex11/ft/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
nl_tasks/exps/run_ex11/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_ex11/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_ex11/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_ex11/ft2/adapter_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:5f8678271e44a27681684478473182e6942ff02f975e79b92df5e6ce87361bc0
|
| 3 |
+
size 33602915
|
nl_tasks/exps/run_ex11/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": 0.18468455970287323,
|
| 15 |
+
"learning_rate": 0.0009836065573770492,
|
| 16 |
+
"loss": 0.5288,
|
| 17 |
+
"step": 25
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.04101722723543889,
|
| 21 |
+
"grad_norm": 0.2326832413673401,
|
| 22 |
+
"learning_rate": 0.0020081967213114752,
|
| 23 |
+
"loss": 0.3549,
|
| 24 |
+
"step": 50
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.06152584085315833,
|
| 28 |
+
"grad_norm": 0.8706040978431702,
|
| 29 |
+
"learning_rate": 0.0030327868852459017,
|
| 30 |
+
"loss": 0.3863,
|
| 31 |
+
"step": 75
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.08203445447087777,
|
| 35 |
+
"grad_norm": 4.948596000671387,
|
| 36 |
+
"learning_rate": 0.004057377049180328,
|
| 37 |
+
"loss": 0.5797,
|
| 38 |
+
"step": 100
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.08203445447087777,
|
| 42 |
+
"eval_loss": 0.6113807559013367,
|
| 43 |
+
"eval_runtime": 19.4524,
|
| 44 |
+
"eval_samples_per_second": 51.407,
|
| 45 |
+
"eval_steps_per_second": 0.823,
|
| 46 |
+
"step": 100
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"epoch": 0.10254306808859721,
|
| 50 |
+
"grad_norm": 0.40604349970817566,
|
| 51 |
+
"learning_rate": 0.005081967213114754,
|
| 52 |
+
"loss": 0.7815,
|
| 53 |
+
"step": 125
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 0.12305168170631665,
|
| 57 |
+
"grad_norm": 0.23787237703800201,
|
| 58 |
+
"learning_rate": 0.00610655737704918,
|
| 59 |
+
"loss": 0.3926,
|
| 60 |
+
"step": 150
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 0.1435602953240361,
|
| 64 |
+
"grad_norm": 0.146455317735672,
|
| 65 |
+
"learning_rate": 0.007131147540983607,
|
| 66 |
+
"loss": 0.3578,
|
| 67 |
+
"step": 175
|
| 68 |
+
},
|
| 69 |
+
{
|
| 70 |
+
"epoch": 0.16406890894175555,
|
| 71 |
+
"grad_norm": 0.14696627855300903,
|
| 72 |
+
"learning_rate": 0.008155737704918033,
|
| 73 |
+
"loss": 0.3532,
|
| 74 |
+
"step": 200
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.16406890894175555,
|
| 78 |
+
"eval_loss": 0.32705461978912354,
|
| 79 |
+
"eval_runtime": 19.2909,
|
| 80 |
+
"eval_samples_per_second": 51.838,
|
| 81 |
+
"eval_steps_per_second": 0.829,
|
| 82 |
+
"step": 200
|
| 83 |
+
},
|
| 84 |
+
{
|
| 85 |
+
"epoch": 0.184577522559475,
|
| 86 |
+
"grad_norm": 0.1279250830411911,
|
| 87 |
+
"learning_rate": 0.009180327868852459,
|
| 88 |
+
"loss": 0.342,
|
| 89 |
+
"step": 225
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 0.20508613617719443,
|
| 93 |
+
"grad_norm": 0.29706206917762756,
|
| 94 |
+
"learning_rate": 0.009999871854116063,
|
| 95 |
+
"loss": 0.3129,
|
| 96 |
+
"step": 250
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"epoch": 0.22559474979491387,
|
| 100 |
+
"grad_norm": 0.09470393508672714,
|
| 101 |
+
"learning_rate": 0.009995387437838026,
|
| 102 |
+
"loss": 0.3085,
|
| 103 |
+
"step": 275
|
| 104 |
+
},
|
| 105 |
+
{
|
| 106 |
+
"epoch": 0.2461033634126333,
|
| 107 |
+
"grad_norm": 0.11185816675424576,
|
| 108 |
+
"learning_rate": 0.009984502294396931,
|
| 109 |
+
"loss": 0.3071,
|
| 110 |
+
"step": 300
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.2461033634126333,
|
| 114 |
+
"eval_loss": 0.2951618731021881,
|
| 115 |
+
"eval_runtime": 19.2841,
|
| 116 |
+
"eval_samples_per_second": 51.856,
|
| 117 |
+
"eval_steps_per_second": 0.83,
|
| 118 |
+
"step": 300
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 0.2666119770303528,
|
| 122 |
+
"grad_norm": 0.10301616042852402,
|
| 123 |
+
"learning_rate": 0.00996723037122612,
|
| 124 |
+
"loss": 0.3052,
|
| 125 |
+
"step": 325
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 0.2871205906480722,
|
| 129 |
+
"grad_norm": 0.08757445216178894,
|
| 130 |
+
"learning_rate": 0.009943593799315264,
|
| 131 |
+
"loss": 0.2966,
|
| 132 |
+
"step": 350
|
| 133 |
+
},
|
| 134 |
+
{
|
| 135 |
+
"epoch": 0.30762920426579166,
|
| 136 |
+
"grad_norm": 0.08772272616624832,
|
| 137 |
+
"learning_rate": 0.009913622864853325,
|
| 138 |
+
"loss": 0.3049,
|
| 139 |
+
"step": 375
|
| 140 |
+
},
|
| 141 |
+
{
|
| 142 |
+
"epoch": 0.3281378178835111,
|
| 143 |
+
"grad_norm": 0.0956786572933197,
|
| 144 |
+
"learning_rate": 0.009877355970422023,
|
| 145 |
+
"loss": 0.2879,
|
| 146 |
+
"step": 400
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.3281378178835111,
|
| 150 |
+
"eval_loss": 0.2759348154067993,
|
| 151 |
+
"eval_runtime": 19.2731,
|
| 152 |
+
"eval_samples_per_second": 51.886,
|
| 153 |
+
"eval_steps_per_second": 0.83,
|
| 154 |
+
"step": 400
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 0.34864643150123054,
|
| 158 |
+
"grad_norm": 0.07560200244188309,
|
| 159 |
+
"learning_rate": 0.009834839585789559,
|
| 160 |
+
"loss": 0.299,
|
| 161 |
+
"step": 425
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 0.36915504511895,
|
| 165 |
+
"grad_norm": 0.07089433819055557,
|
| 166 |
+
"learning_rate": 0.009786128188367622,
|
| 167 |
+
"loss": 0.2938,
|
| 168 |
+
"step": 450
|
| 169 |
+
},
|
| 170 |
+
{
|
| 171 |
+
"epoch": 0.3896636587366694,
|
| 172 |
+
"grad_norm": 0.07281703501939774,
|
| 173 |
+
"learning_rate": 0.00973128419340798,
|
| 174 |
+
"loss": 0.2793,
|
| 175 |
+
"step": 475
|
| 176 |
+
},
|
| 177 |
+
{
|
| 178 |
+
"epoch": 0.41017227235438886,
|
| 179 |
+
"grad_norm": 0.05941835418343544,
|
| 180 |
+
"learning_rate": 0.009670377874028116,
|
| 181 |
+
"loss": 0.2633,
|
| 182 |
+
"step": 500
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.41017227235438886,
|
| 186 |
+
"eval_loss": 0.2671220004558563,
|
| 187 |
+
"eval_runtime": 19.3369,
|
| 188 |
+
"eval_samples_per_second": 51.715,
|
| 189 |
+
"eval_steps_per_second": 0.827,
|
| 190 |
+
"step": 500
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 0.4306808859721083,
|
| 194 |
+
"grad_norm": 0.055430203676223755,
|
| 195 |
+
"learning_rate": 0.009603487271168337,
|
| 196 |
+
"loss": 0.2624,
|
| 197 |
+
"step": 525
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 0.45118949958982774,
|
| 201 |
+
"grad_norm": 0.0656672939658165,
|
| 202 |
+
"learning_rate": 0.009530698093595781,
|
| 203 |
+
"loss": 0.2751,
|
| 204 |
+
"step": 550
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 0.4716981132075472,
|
| 208 |
+
"grad_norm": 0.10994518548250198,
|
| 209 |
+
"learning_rate": 0.009452103608083419,
|
| 210 |
+
"loss": 0.2658,
|
| 211 |
+
"step": 575
|
| 212 |
+
},
|
| 213 |
+
{
|
| 214 |
+
"epoch": 0.4922067268252666,
|
| 215 |
+
"grad_norm": 0.06084736809134483,
|
| 216 |
+
"learning_rate": 0.009367804519904774,
|
| 217 |
+
"loss": 0.2835,
|
| 218 |
+
"step": 600
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.4922067268252666,
|
| 222 |
+
"eval_loss": 0.25752514600753784,
|
| 223 |
+
"eval_runtime": 19.2929,
|
| 224 |
+
"eval_samples_per_second": 51.833,
|
| 225 |
+
"eval_steps_per_second": 0.829,
|
| 226 |
+
"step": 600
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.5127153404429861,
|
| 230 |
+
"grad_norm": 0.05249754339456558,
|
| 231 |
+
"learning_rate": 0.009277908843797492,
|
| 232 |
+
"loss": 0.2609,
|
| 233 |
+
"step": 625
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.5332239540607056,
|
| 237 |
+
"grad_norm": 0.06335076689720154,
|
| 238 |
+
"learning_rate": 0.009182531765561084,
|
| 239 |
+
"loss": 0.2544,
|
| 240 |
+
"step": 650
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.5537325676784249,
|
| 244 |
+
"grad_norm": 0.057296834886074066,
|
| 245 |
+
"learning_rate": 0.0090817954944662,
|
| 246 |
+
"loss": 0.265,
|
| 247 |
+
"step": 675
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.5742411812961444,
|
| 251 |
+
"grad_norm": 0.0538799874484539,
|
| 252 |
+
"learning_rate": 0.00897582910666454,
|
| 253 |
+
"loss": 0.2638,
|
| 254 |
+
"step": 700
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.5742411812961444,
|
| 258 |
+
"eval_loss": 0.24969546496868134,
|
| 259 |
+
"eval_runtime": 19.2581,
|
| 260 |
+
"eval_samples_per_second": 51.926,
|
| 261 |
+
"eval_steps_per_second": 0.831,
|
| 262 |
+
"step": 700
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.5947497949138638,
|
| 266 |
+
"grad_norm": 0.05747291073203087,
|
| 267 |
+
"learning_rate": 0.008864768379800016,
|
| 268 |
+
"loss": 0.2616,
|
| 269 |
+
"step": 725
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.6152584085315833,
|
| 273 |
+
"grad_norm": 0.051392413675785065,
|
| 274 |
+
"learning_rate": 0.008748755619033153,
|
| 275 |
+
"loss": 0.2575,
|
| 276 |
+
"step": 750
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.6357670221493027,
|
| 280 |
+
"grad_norm": 0.04845484718680382,
|
| 281 |
+
"learning_rate": 0.00862793947470155,
|
| 282 |
+
"loss": 0.2541,
|
| 283 |
+
"step": 775
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.6562756357670222,
|
| 287 |
+
"grad_norm": 0.06000267714262009,
|
| 288 |
+
"learning_rate": 0.008502474751850142,
|
| 289 |
+
"loss": 0.2596,
|
| 290 |
+
"step": 800
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.6562756357670222,
|
| 294 |
+
"eval_loss": 0.24642696976661682,
|
| 295 |
+
"eval_runtime": 19.2939,
|
| 296 |
+
"eval_samples_per_second": 51.83,
|
| 297 |
+
"eval_steps_per_second": 0.829,
|
| 298 |
+
"step": 800
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 0.6767842493847416,
|
| 302 |
+
"grad_norm": 0.04552628844976425,
|
| 303 |
+
"learning_rate": 0.008372522211875224,
|
| 304 |
+
"loss": 0.2526,
|
| 305 |
+
"step": 825
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 0.6972928630024611,
|
| 309 |
+
"grad_norm": 0.04838543385267258,
|
| 310 |
+
"learning_rate": 0.008238248366536473,
|
| 311 |
+
"loss": 0.2545,
|
| 312 |
+
"step": 850
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 0.7178014766201805,
|
| 316 |
+
"grad_norm": 0.03977634012699127,
|
| 317 |
+
"learning_rate": 0.008099825264600843,
|
| 318 |
+
"loss": 0.2436,
|
| 319 |
+
"step": 875
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 0.7383100902379,
|
| 323 |
+
"grad_norm": 0.05270398035645485,
|
| 324 |
+
"learning_rate": 0.007957430271391761,
|
| 325 |
+
"loss": 0.2488,
|
| 326 |
+
"step": 900
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 0.7383100902379,
|
| 330 |
+
"eval_loss": 0.23903542757034302,
|
| 331 |
+
"eval_runtime": 19.2555,
|
| 332 |
+
"eval_samples_per_second": 51.933,
|
| 333 |
+
"eval_steps_per_second": 0.831,
|
| 334 |
+
"step": 900
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"epoch": 0.7588187038556193,
|
| 338 |
+
"grad_norm": 0.05134091153740883,
|
| 339 |
+
"learning_rate": 0.007811245841526062,
|
| 340 |
+
"loss": 0.2511,
|
| 341 |
+
"step": 925
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 0.7793273174733388,
|
| 345 |
+
"grad_norm": 0.04741799831390381,
|
| 346 |
+
"learning_rate": 0.00766145928512988,
|
| 347 |
+
"loss": 0.2385,
|
| 348 |
+
"step": 950
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"epoch": 0.7998359310910582,
|
| 352 |
+
"grad_norm": 0.05198933929204941,
|
| 353 |
+
"learning_rate": 0.007508262527833029,
|
| 354 |
+
"loss": 0.246,
|
| 355 |
+
"step": 975
|
| 356 |
+
},
|
| 357 |
+
{
|
| 358 |
+
"epoch": 0.8203445447087777,
|
| 359 |
+
"grad_norm": 0.05085662007331848,
|
| 360 |
+
"learning_rate": 0.00735185186484943,
|
| 361 |
+
"loss": 0.2394,
|
| 362 |
+
"step": 1000
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.8203445447087777,
|
| 366 |
+
"eval_loss": 0.2347901165485382,
|
| 367 |
+
"eval_runtime": 19.3141,
|
| 368 |
+
"eval_samples_per_second": 51.776,
|
| 369 |
+
"eval_steps_per_second": 0.828,
|
| 370 |
+
"step": 1000
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 0.8408531583264971,
|
| 374 |
+
"grad_norm": 0.04405153915286064,
|
| 375 |
+
"learning_rate": 0.007192427709458655,
|
| 376 |
+
"loss": 0.2385,
|
| 377 |
+
"step": 1025
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 0.8613617719442166,
|
| 381 |
+
"grad_norm": 0.04311925917863846,
|
| 382 |
+
"learning_rate": 0.007030194336210888,
|
| 383 |
+
"loss": 0.2373,
|
| 384 |
+
"step": 1050
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.881870385561936,
|
| 388 |
+
"grad_norm": 0.04251931980252266,
|
| 389 |
+
"learning_rate": 0.00686535961918433,
|
| 390 |
+
"loss": 0.2417,
|
| 391 |
+
"step": 1075
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 0.9023789991796555,
|
| 395 |
+
"grad_norm": 0.0427103117108345,
|
| 396 |
+
"learning_rate": 0.006698134765630434,
|
| 397 |
+
"loss": 0.2422,
|
| 398 |
+
"step": 1100
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 0.9023789991796555,
|
| 402 |
+
"eval_loss": 0.23155222833156586,
|
| 403 |
+
"eval_runtime": 19.2861,
|
| 404 |
+
"eval_samples_per_second": 51.851,
|
| 405 |
+
"eval_steps_per_second": 0.83,
|
| 406 |
+
"step": 1100
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 0.9228876127973749,
|
| 410 |
+
"grad_norm": 0.0433138944208622,
|
| 411 |
+
"learning_rate": 0.006528734045348248,
|
| 412 |
+
"loss": 0.2383,
|
| 413 |
+
"step": 1125
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 0.9433962264150944,
|
| 417 |
+
"grad_norm": 0.042825039476156235,
|
| 418 |
+
"learning_rate": 0.006357374516134643,
|
| 419 |
+
"loss": 0.2365,
|
| 420 |
+
"step": 1150
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 0.9639048400328137,
|
| 424 |
+
"grad_norm": 0.0420411117374897,
|
| 425 |
+
"learning_rate": 0.006184275745662179,
|
| 426 |
+
"loss": 0.2416,
|
| 427 |
+
"step": 1175
|
| 428 |
+
},
|
| 429 |
+
{
|
| 430 |
+
"epoch": 0.9844134536505332,
|
| 431 |
+
"grad_norm": 0.04197347164154053,
|
| 432 |
+
"learning_rate": 0.006009659530141031,
|
| 433 |
+
"loss": 0.2474,
|
| 434 |
+
"step": 1200
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 0.9844134536505332,
|
| 438 |
+
"eval_loss": 0.22774319350719452,
|
| 439 |
+
"eval_runtime": 19.2633,
|
| 440 |
+
"eval_samples_per_second": 51.912,
|
| 441 |
+
"eval_steps_per_second": 0.831,
|
| 442 |
+
"step": 1200
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 1.0049220672682526,
|
| 446 |
+
"grad_norm": 0.03893238306045532,
|
| 447 |
+
"learning_rate": 0.005833749610125402,
|
| 448 |
+
"loss": 0.227,
|
| 449 |
+
"step": 1225
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 1.0254306808859721,
|
| 453 |
+
"grad_norm": 0.040400441735982895,
|
| 454 |
+
"learning_rate": 0.005656771383828601,
|
| 455 |
+
"loss": 0.2042,
|
| 456 |
+
"step": 1250
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 1.0459392945036916,
|
| 460 |
+
"grad_norm": 0.04679598659276962,
|
| 461 |
+
"learning_rate": 0.005478951618314133,
|
| 462 |
+
"loss": 0.2071,
|
| 463 |
+
"step": 1275
|
| 464 |
+
},
|
| 465 |
+
{
|
| 466 |
+
"epoch": 1.066447908121411,
|
| 467 |
+
"grad_norm": 0.040728066116571426,
|
| 468 |
+
"learning_rate": 0.0053005181589328145,
|
| 469 |
+
"loss": 0.2088,
|
| 470 |
+
"step": 1300
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 1.066447908121411,
|
| 474 |
+
"eval_loss": 0.22445358335971832,
|
| 475 |
+
"eval_runtime": 19.275,
|
| 476 |
+
"eval_samples_per_second": 51.881,
|
| 477 |
+
"eval_steps_per_second": 0.83,
|
| 478 |
+
"step": 1300
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 1.0869565217391304,
|
| 482 |
+
"grad_norm": 0.04398899897933006,
|
| 483 |
+
"learning_rate": 0.005121699637378282,
|
| 484 |
+
"loss": 0.2062,
|
| 485 |
+
"step": 1325
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 1.1074651353568499,
|
| 489 |
+
"grad_norm": 0.04232119768857956,
|
| 490 |
+
"learning_rate": 0.004942725178734904,
|
| 491 |
+
"loss": 0.1997,
|
| 492 |
+
"step": 1350
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 1.1279737489745694,
|
| 496 |
+
"grad_norm": 0.04680604115128517,
|
| 497 |
+
"learning_rate": 0.0047638241078935325,
|
| 498 |
+
"loss": 0.201,
|
| 499 |
+
"step": 1375
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 1.1484823625922886,
|
| 503 |
+
"grad_norm": 0.04227881878614426,
|
| 504 |
+
"learning_rate": 0.00458522565571121,
|
| 505 |
+
"loss": 0.195,
|
| 506 |
+
"step": 1400
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 1.1484823625922886,
|
| 510 |
+
"eval_loss": 0.22240343689918518,
|
| 511 |
+
"eval_runtime": 19.2743,
|
| 512 |
+
"eval_samples_per_second": 51.883,
|
| 513 |
+
"eval_steps_per_second": 0.83,
|
| 514 |
+
"step": 1400
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 1.1689909762100081,
|
| 518 |
+
"grad_norm": 0.045188840478658676,
|
| 519 |
+
"learning_rate": 0.004407158665291376,
|
| 520 |
+
"loss": 0.2051,
|
| 521 |
+
"step": 1425
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 1.1894995898277276,
|
| 525 |
+
"grad_norm": 0.049927521497011185,
|
| 526 |
+
"learning_rate": 0.004229851298760915,
|
| 527 |
+
"loss": 0.2052,
|
| 528 |
+
"step": 1450
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 1.2100082034454471,
|
| 532 |
+
"grad_norm": 0.04331929609179497,
|
| 533 |
+
"learning_rate": 0.004053530744919749,
|
| 534 |
+
"loss": 0.2015,
|
| 535 |
+
"step": 1475
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 1.2305168170631666,
|
| 539 |
+
"grad_norm": 0.05498148500919342,
|
| 540 |
+
"learning_rate": 0.0038784229281375966,
|
| 541 |
+
"loss": 0.2114,
|
| 542 |
+
"step": 1500
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 1.2305168170631666,
|
| 546 |
+
"eval_loss": 0.2203172892332077,
|
| 547 |
+
"eval_runtime": 19.2909,
|
| 548 |
+
"eval_samples_per_second": 51.838,
|
| 549 |
+
"eval_steps_per_second": 0.829,
|
| 550 |
+
"step": 1500
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 1.251025430680886,
|
| 554 |
+
"grad_norm": 0.0505937822163105,
|
| 555 |
+
"learning_rate": 0.003704752218870861,
|
| 556 |
+
"loss": 0.2054,
|
| 557 |
+
"step": 1525
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 1.2715340442986054,
|
| 561 |
+
"grad_norm": 0.03429531678557396,
|
| 562 |
+
"learning_rate": 0.003532741146170603,
|
| 563 |
+
"loss": 0.1993,
|
| 564 |
+
"step": 1550
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 1.2920426579163249,
|
| 568 |
+
"grad_norm": 0.03555479645729065,
|
| 569 |
+
"learning_rate": 0.0033626101125499553,
|
| 570 |
+
"loss": 0.2017,
|
| 571 |
+
"step": 1575
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 1.3125512715340442,
|
| 575 |
+
"grad_norm": 0.038551222532987595,
|
| 576 |
+
"learning_rate": 0.0031945771115763333,
|
| 577 |
+
"loss": 0.1952,
|
| 578 |
+
"step": 1600
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 1.3125512715340442,
|
| 582 |
+
"eval_loss": 0.21675598621368408,
|
| 583 |
+
"eval_runtime": 19.283,
|
| 584 |
+
"eval_samples_per_second": 51.859,
|
| 585 |
+
"eval_steps_per_second": 0.83,
|
| 586 |
+
"step": 1600
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 1.3330598851517639,
|
| 590 |
+
"grad_norm": 0.03651474416255951,
|
| 591 |
+
"learning_rate": 0.0030288574485502757,
|
| 592 |
+
"loss": 0.1966,
|
| 593 |
+
"step": 1625
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 1.3535684987694832,
|
| 597 |
+
"grad_norm": 0.04078409820795059,
|
| 598 |
+
"learning_rate": 0.0028656634646288567,
|
| 599 |
+
"loss": 0.1979,
|
| 600 |
+
"step": 1650
|
| 601 |
+
},
|
| 602 |
+
{
|
| 603 |
+
"epoch": 1.3740771123872026,
|
| 604 |
+
"grad_norm": 0.040798161178827286,
|
| 605 |
+
"learning_rate": 0.0027052042647471254,
|
| 606 |
+
"loss": 0.1943,
|
| 607 |
+
"step": 1675
|
| 608 |
+
},
|
| 609 |
+
{
|
| 610 |
+
"epoch": 1.3945857260049221,
|
| 611 |
+
"grad_norm": 0.03889540582895279,
|
| 612 |
+
"learning_rate": 0.0025476854496862054,
|
| 613 |
+
"loss": 0.1966,
|
| 614 |
+
"step": 1700
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 1.3945857260049221,
|
| 618 |
+
"eval_loss": 0.2134680598974228,
|
| 619 |
+
"eval_runtime": 19.2952,
|
| 620 |
+
"eval_samples_per_second": 51.826,
|
| 621 |
+
"eval_steps_per_second": 0.829,
|
| 622 |
+
"step": 1700
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"epoch": 1.4150943396226414,
|
| 626 |
+
"grad_norm": 0.03957779332995415,
|
| 627 |
+
"learning_rate": 0.002393308852631373,
|
| 628 |
+
"loss": 0.1927,
|
| 629 |
+
"step": 1725
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 1.435602953240361,
|
| 633 |
+
"grad_norm": 0.040814198553562164,
|
| 634 |
+
"learning_rate": 0.0022422722805576453,
|
| 635 |
+
"loss": 0.1911,
|
| 636 |
+
"step": 1750
|
| 637 |
+
},
|
| 638 |
+
{
|
| 639 |
+
"epoch": 1.4561115668580804,
|
| 640 |
+
"grad_norm": 0.03768594563007355,
|
| 641 |
+
"learning_rate": 0.002094769260774262,
|
| 642 |
+
"loss": 0.1982,
|
| 643 |
+
"step": 1775
|
| 644 |
+
},
|
| 645 |
+
{
|
| 646 |
+
"epoch": 1.4766201804758,
|
| 647 |
+
"grad_norm": 0.037988074123859406,
|
| 648 |
+
"learning_rate": 0.0019509887929528458,
|
| 649 |
+
"loss": 0.1879,
|
| 650 |
+
"step": 1800
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 1.4766201804758,
|
| 654 |
+
"eval_loss": 0.21205949783325195,
|
| 655 |
+
"eval_runtime": 19.2762,
|
| 656 |
+
"eval_samples_per_second": 51.877,
|
| 657 |
+
"eval_steps_per_second": 0.83,
|
| 658 |
+
"step": 1800
|
| 659 |
+
},
|
| 660 |
+
{
|
| 661 |
+
"epoch": 1.4971287940935194,
|
| 662 |
+
"grad_norm": 0.0394948311150074,
|
| 663 |
+
"learning_rate": 0.0018111151069569182,
|
| 664 |
+
"loss": 0.1868,
|
| 665 |
+
"step": 1825
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 1.5176374077112387,
|
| 669 |
+
"grad_norm": 0.03492829203605652,
|
| 670 |
+
"learning_rate": 0.0016753274267831113,
|
| 671 |
+
"loss": 0.1906,
|
| 672 |
+
"step": 1850
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 1.5381460213289582,
|
| 676 |
+
"grad_norm": 0.036308083683252335,
|
| 677 |
+
"learning_rate": 0.0015437997409165477,
|
| 678 |
+
"loss": 0.191,
|
| 679 |
+
"step": 1875
|
| 680 |
+
},
|
| 681 |
+
{
|
| 682 |
+
"epoch": 1.5586546349466777,
|
| 683 |
+
"grad_norm": 0.04244968667626381,
|
| 684 |
+
"learning_rate": 0.0014167005793946036,
|
| 685 |
+
"loss": 0.1939,
|
| 686 |
+
"step": 1900
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 1.5586546349466777,
|
| 690 |
+
"eval_loss": 0.20898254215717316,
|
| 691 |
+
"eval_runtime": 19.2438,
|
| 692 |
+
"eval_samples_per_second": 51.965,
|
| 693 |
+
"eval_steps_per_second": 0.831,
|
| 694 |
+
"step": 1900
|
| 695 |
+
},
|
| 696 |
+
{
|
| 697 |
+
"epoch": 1.579163248564397,
|
| 698 |
+
"grad_norm": 0.0380813330411911,
|
| 699 |
+
"learning_rate": 0.0012941927978647528,
|
| 700 |
+
"loss": 0.1946,
|
| 701 |
+
"step": 1925
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"epoch": 1.5996718621821167,
|
| 705 |
+
"grad_norm": 0.039436712861061096,
|
| 706 |
+
"learning_rate": 0.0011764333689131385,
|
| 707 |
+
"loss": 0.1921,
|
| 708 |
+
"step": 1950
|
| 709 |
+
},
|
| 710 |
+
{
|
| 711 |
+
"epoch": 1.620180475799836,
|
| 712 |
+
"grad_norm": 0.037560343742370605,
|
| 713 |
+
"learning_rate": 0.0010635731809312992,
|
| 714 |
+
"loss": 0.1878,
|
| 715 |
+
"step": 1975
|
| 716 |
+
},
|
| 717 |
+
{
|
| 718 |
+
"epoch": 1.6406890894175554,
|
| 719 |
+
"grad_norm": 0.043062906712293625,
|
| 720 |
+
"learning_rate": 0.0009557568447787202,
|
| 721 |
+
"loss": 0.1941,
|
| 722 |
+
"step": 2000
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 1.6406890894175554,
|
| 726 |
+
"eval_loss": 0.20627179741859436,
|
| 727 |
+
"eval_runtime": 19.2924,
|
| 728 |
+
"eval_samples_per_second": 51.834,
|
| 729 |
+
"eval_steps_per_second": 0.829,
|
| 730 |
+
"step": 2000
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 1.661197703035275,
|
| 734 |
+
"grad_norm": 0.03676965460181236,
|
| 735 |
+
"learning_rate": 0.0008531225084889654,
|
| 736 |
+
"loss": 0.1833,
|
| 737 |
+
"step": 2025
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 1.6817063166529942,
|
| 741 |
+
"grad_norm": 0.047483399510383606,
|
| 742 |
+
"learning_rate": 0.0007558016802568091,
|
| 743 |
+
"loss": 0.1853,
|
| 744 |
+
"step": 2050
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 1.7022149302707137,
|
| 748 |
+
"grad_norm": 0.05413796380162239,
|
| 749 |
+
"learning_rate": 0.0006639190599331745,
|
| 750 |
+
"loss": 0.1905,
|
| 751 |
+
"step": 2075
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 1.7227235438884332,
|
| 755 |
+
"grad_norm": 0.04098552465438843,
|
| 756 |
+
"learning_rate": 0.0005775923792437865,
|
| 757 |
+
"loss": 0.1818,
|
| 758 |
+
"step": 2100
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 1.7227235438884332,
|
| 762 |
+
"eval_loss": 0.20522451400756836,
|
| 763 |
+
"eval_runtime": 19.2877,
|
| 764 |
+
"eval_samples_per_second": 51.847,
|
| 765 |
+
"eval_steps_per_second": 0.83,
|
| 766 |
+
"step": 2100
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 1.7432321575061525,
|
| 770 |
+
"grad_norm": 0.03664683923125267,
|
| 771 |
+
"learning_rate": 0.0004969322509362762,
|
| 772 |
+
"loss": 0.1849,
|
| 773 |
+
"step": 2125
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 1.7637407711238722,
|
| 777 |
+
"grad_norm": 0.033080730587244034,
|
| 778 |
+
"learning_rate": 0.0004220420270490294,
|
| 779 |
+
"loss": 0.1827,
|
| 780 |
+
"step": 2150
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 1.7842493847415914,
|
| 784 |
+
"grad_norm": 0.03479531407356262,
|
| 785 |
+
"learning_rate": 0.0003530176664833834,
|
| 786 |
+
"loss": 0.1917,
|
| 787 |
+
"step": 2175
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 1.804757998359311,
|
| 791 |
+
"grad_norm": 0.04237658903002739,
|
| 792 |
+
"learning_rate": 0.00028994761204884756,
|
| 793 |
+
"loss": 0.1843,
|
| 794 |
+
"step": 2200
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 1.804757998359311,
|
| 798 |
+
"eval_loss": 0.20377041399478912,
|
| 799 |
+
"eval_runtime": 19.3506,
|
| 800 |
+
"eval_samples_per_second": 51.678,
|
| 801 |
+
"eval_steps_per_second": 0.827,
|
| 802 |
+
"step": 2200
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 1.8252666119770304,
|
| 806 |
+
"grad_norm": 0.03981838747859001,
|
| 807 |
+
"learning_rate": 0.00023291267713889953,
|
| 808 |
+
"loss": 0.1857,
|
| 809 |
+
"step": 2225
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 1.8457752255947497,
|
| 813 |
+
"grad_norm": 0.041474029421806335,
|
| 814 |
+
"learning_rate": 0.00018198594218256814,
|
| 815 |
+
"loss": 0.1873,
|
| 816 |
+
"step": 2250
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"epoch": 1.8662838392124692,
|
| 820 |
+
"grad_norm": 0.03762819990515709,
|
| 821 |
+
"learning_rate": 0.00013723266100447052,
|
| 822 |
+
"loss": 0.192,
|
| 823 |
+
"step": 2275
|
| 824 |
+
},
|
| 825 |
+
{
|
| 826 |
+
"epoch": 1.8867924528301887,
|
| 827 |
+
"grad_norm": 0.03675968199968338,
|
| 828 |
+
"learning_rate": 9.871017721329201e-05,
|
| 829 |
+
"loss": 0.1849,
|
| 830 |
+
"step": 2300
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 1.8867924528301887,
|
| 834 |
+
"eval_loss": 0.20302893221378326,
|
| 835 |
+
"eval_runtime": 19.279,
|
| 836 |
+
"eval_samples_per_second": 51.87,
|
| 837 |
+
"eval_steps_per_second": 0.83,
|
| 838 |
+
"step": 2300
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 1.907301066447908,
|
| 842 |
+
"grad_norm": 0.04405345022678375,
|
| 843 |
+
"learning_rate": 6.64678507258487e-05,
|
| 844 |
+
"loss": 0.193,
|
| 845 |
+
"step": 2325
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 1.9278096800656277,
|
| 849 |
+
"grad_norm": 0.03877412527799606,
|
| 850 |
+
"learning_rate": 4.054699452086641e-05,
|
| 851 |
+
"loss": 0.1773,
|
| 852 |
+
"step": 2350
|
| 853 |
+
},
|
| 854 |
+
{
|
| 855 |
+
"epoch": 1.948318293683347,
|
| 856 |
+
"grad_norm": 0.03657250851392746,
|
| 857 |
+
"learning_rate": 2.0980821703527885e-05,
|
| 858 |
+
"loss": 0.1831,
|
| 859 |
+
"step": 2375
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 1.9688269073010665,
|
| 863 |
+
"grad_norm": 0.03823859989643097,
|
| 864 |
+
"learning_rate": 7.794402948607671e-06,
|
| 865 |
+
"loss": 0.1838,
|
| 866 |
+
"step": 2400
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 1.9688269073010665,
|
| 870 |
+
"eval_loss": 0.20284757018089294,
|
| 871 |
+
"eval_runtime": 19.2943,
|
| 872 |
+
"eval_samples_per_second": 51.829,
|
| 873 |
+
"eval_steps_per_second": 0.829,
|
| 874 |
+
"step": 2400
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"epoch": 1.989335520918786,
|
| 878 |
+
"grad_norm": 0.037098102271556854,
|
| 879 |
+
"learning_rate": 1.0046343767294852e-06,
|
| 880 |
+
"loss": 0.1804,
|
| 881 |
+
"step": 2425
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 2.0,
|
| 885 |
+
"step": 2438,
|
| 886 |
+
"total_flos": 1.58523627405312e+18,
|
| 887 |
+
"train_loss": 0.24610730680507945,
|
| 888 |
+
"train_runtime": 2871.7098,
|
| 889 |
+
"train_samples_per_second": 27.162,
|
| 890 |
+
"train_steps_per_second": 0.849
|
| 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_ex12/ft/added_tokens.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"[PAD]": 32000
|
| 3 |
+
}
|
nl_tasks/exps/run_ex12/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_ex12/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_ex12/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": 0.19958673417568207,
|
| 15 |
+
"learning_rate": 0.0009836065573770492,
|
| 16 |
+
"loss": 0.5277,
|
| 17 |
+
"step": 25
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.04101722723543889,
|
| 21 |
+
"grad_norm": 0.23706433176994324,
|
| 22 |
+
"learning_rate": 0.0020081967213114752,
|
| 23 |
+
"loss": 0.3549,
|
| 24 |
+
"step": 50
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.06152584085315833,
|
| 28 |
+
"grad_norm": 0.4635924994945526,
|
| 29 |
+
"learning_rate": 0.0030327868852459017,
|
| 30 |
+
"loss": 0.3883,
|
| 31 |
+
"step": 75
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.08203445447087777,
|
| 35 |
+
"grad_norm": 2.479024887084961,
|
| 36 |
+
"learning_rate": 0.004057377049180328,
|
| 37 |
+
"loss": 0.4933,
|
| 38 |
+
"step": 100
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.08203445447087777,
|
| 42 |
+
"eval_loss": 0.5168276429176331,
|
| 43 |
+
"eval_runtime": 19.5143,
|
| 44 |
+
"eval_samples_per_second": 51.245,
|
| 45 |
+
"eval_steps_per_second": 0.82,
|
| 46 |
+
"step": 100
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"epoch": 0.10254306808859721,
|
| 50 |
+
"grad_norm": 0.5372624397277832,
|
| 51 |
+
"learning_rate": 0.005081967213114754,
|
| 52 |
+
"loss": 0.5286,
|
| 53 |
+
"step": 125
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 0.12305168170631665,
|
| 57 |
+
"grad_norm": 0.25752052664756775,
|
| 58 |
+
"learning_rate": 0.00610655737704918,
|
| 59 |
+
"loss": 0.3972,
|
| 60 |
+
"step": 150
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 0.1435602953240361,
|
| 64 |
+
"grad_norm": 0.1797451674938202,
|
| 65 |
+
"learning_rate": 0.007131147540983607,
|
| 66 |
+
"loss": 0.3692,
|
| 67 |
+
"step": 175
|
| 68 |
+
},
|
| 69 |
+
{
|
| 70 |
+
"epoch": 0.16406890894175555,
|
| 71 |
+
"grad_norm": 0.18513981997966766,
|
| 72 |
+
"learning_rate": 0.008155737704918033,
|
| 73 |
+
"loss": 0.3546,
|
| 74 |
+
"step": 200
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.16406890894175555,
|
| 78 |
+
"eval_loss": 0.3297361731529236,
|
| 79 |
+
"eval_runtime": 19.3627,
|
| 80 |
+
"eval_samples_per_second": 51.646,
|
| 81 |
+
"eval_steps_per_second": 0.826,
|
| 82 |
+
"step": 200
|
| 83 |
+
},
|
| 84 |
+
{
|
| 85 |
+
"epoch": 0.184577522559475,
|
| 86 |
+
"grad_norm": 0.11278539150953293,
|
| 87 |
+
"learning_rate": 0.009180327868852459,
|
| 88 |
+
"loss": 0.3396,
|
| 89 |
+
"step": 225
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 0.20508613617719443,
|
| 93 |
+
"grad_norm": 0.14479927718639374,
|
| 94 |
+
"learning_rate": 0.009999871854116063,
|
| 95 |
+
"loss": 0.3293,
|
| 96 |
+
"step": 250
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"epoch": 0.22559474979491387,
|
| 100 |
+
"grad_norm": 0.10411059856414795,
|
| 101 |
+
"learning_rate": 0.009995387437838026,
|
| 102 |
+
"loss": 0.3166,
|
| 103 |
+
"step": 275
|
| 104 |
+
},
|
| 105 |
+
{
|
| 106 |
+
"epoch": 0.2461033634126333,
|
| 107 |
+
"grad_norm": 0.12827317416667938,
|
| 108 |
+
"learning_rate": 0.009984502294396931,
|
| 109 |
+
"loss": 0.3086,
|
| 110 |
+
"step": 300
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.2461033634126333,
|
| 114 |
+
"eval_loss": 0.2996523380279541,
|
| 115 |
+
"eval_runtime": 19.2521,
|
| 116 |
+
"eval_samples_per_second": 51.942,
|
| 117 |
+
"eval_steps_per_second": 0.831,
|
| 118 |
+
"step": 300
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 0.2666119770303528,
|
| 122 |
+
"grad_norm": 0.08072741329669952,
|
| 123 |
+
"learning_rate": 0.00996723037122612,
|
| 124 |
+
"loss": 0.3085,
|
| 125 |
+
"step": 325
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 0.2871205906480722,
|
| 129 |
+
"grad_norm": 0.08349168300628662,
|
| 130 |
+
"learning_rate": 0.009943593799315264,
|
| 131 |
+
"loss": 0.2971,
|
| 132 |
+
"step": 350
|
| 133 |
+
},
|
| 134 |
+
{
|
| 135 |
+
"epoch": 0.30762920426579166,
|
| 136 |
+
"grad_norm": 0.08299077302217484,
|
| 137 |
+
"learning_rate": 0.009913622864853325,
|
| 138 |
+
"loss": 0.3057,
|
| 139 |
+
"step": 375
|
| 140 |
+
},
|
| 141 |
+
{
|
| 142 |
+
"epoch": 0.3281378178835111,
|
| 143 |
+
"grad_norm": 0.0772460401058197,
|
| 144 |
+
"learning_rate": 0.009877355970422023,
|
| 145 |
+
"loss": 0.2878,
|
| 146 |
+
"step": 400
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.3281378178835111,
|
| 150 |
+
"eval_loss": 0.27609387040138245,
|
| 151 |
+
"eval_runtime": 19.2868,
|
| 152 |
+
"eval_samples_per_second": 51.849,
|
| 153 |
+
"eval_steps_per_second": 0.83,
|
| 154 |
+
"step": 400
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 0.34864643150123054,
|
| 158 |
+
"grad_norm": 0.07354776561260223,
|
| 159 |
+
"learning_rate": 0.009834839585789559,
|
| 160 |
+
"loss": 0.296,
|
| 161 |
+
"step": 425
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 0.36915504511895,
|
| 165 |
+
"grad_norm": 0.06830737739801407,
|
| 166 |
+
"learning_rate": 0.009786128188367622,
|
| 167 |
+
"loss": 0.2906,
|
| 168 |
+
"step": 450
|
| 169 |
+
},
|
| 170 |
+
{
|
| 171 |
+
"epoch": 0.3896636587366694,
|
| 172 |
+
"grad_norm": 0.05791277438402176,
|
| 173 |
+
"learning_rate": 0.00973128419340798,
|
| 174 |
+
"loss": 0.2783,
|
| 175 |
+
"step": 475
|
| 176 |
+
},
|
| 177 |
+
{
|
| 178 |
+
"epoch": 0.41017227235438886,
|
| 179 |
+
"grad_norm": 0.05781574174761772,
|
| 180 |
+
"learning_rate": 0.009670377874028116,
|
| 181 |
+
"loss": 0.2622,
|
| 182 |
+
"step": 500
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.41017227235438886,
|
| 186 |
+
"eval_loss": 0.26715630292892456,
|
| 187 |
+
"eval_runtime": 19.2839,
|
| 188 |
+
"eval_samples_per_second": 51.857,
|
| 189 |
+
"eval_steps_per_second": 0.83,
|
| 190 |
+
"step": 500
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 0.4306808859721083,
|
| 194 |
+
"grad_norm": 0.057914622128009796,
|
| 195 |
+
"learning_rate": 0.009603487271168337,
|
| 196 |
+
"loss": 0.2623,
|
| 197 |
+
"step": 525
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 0.45118949958982774,
|
| 201 |
+
"grad_norm": 0.06443090736865997,
|
| 202 |
+
"learning_rate": 0.009530698093595781,
|
| 203 |
+
"loss": 0.2734,
|
| 204 |
+
"step": 550
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 0.4716981132075472,
|
| 208 |
+
"grad_norm": 0.09124640375375748,
|
| 209 |
+
"learning_rate": 0.009452103608083419,
|
| 210 |
+
"loss": 0.2633,
|
| 211 |
+
"step": 575
|
| 212 |
+
},
|
| 213 |
+
{
|
| 214 |
+
"epoch": 0.4922067268252666,
|
| 215 |
+
"grad_norm": 0.05553770810365677,
|
| 216 |
+
"learning_rate": 0.009367804519904774,
|
| 217 |
+
"loss": 0.2816,
|
| 218 |
+
"step": 600
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.4922067268252666,
|
| 222 |
+
"eval_loss": 0.25673770904541016,
|
| 223 |
+
"eval_runtime": 19.3172,
|
| 224 |
+
"eval_samples_per_second": 51.767,
|
| 225 |
+
"eval_steps_per_second": 0.828,
|
| 226 |
+
"step": 600
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.5127153404429861,
|
| 230 |
+
"grad_norm": 0.0597306527197361,
|
| 231 |
+
"learning_rate": 0.009277908843797492,
|
| 232 |
+
"loss": 0.2595,
|
| 233 |
+
"step": 625
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.5332239540607056,
|
| 237 |
+
"grad_norm": 0.065106101334095,
|
| 238 |
+
"learning_rate": 0.009182531765561084,
|
| 239 |
+
"loss": 0.2529,
|
| 240 |
+
"step": 650
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.5537325676784249,
|
| 244 |
+
"grad_norm": 0.07067123800516129,
|
| 245 |
+
"learning_rate": 0.0090817954944662,
|
| 246 |
+
"loss": 0.2661,
|
| 247 |
+
"step": 675
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.5742411812961444,
|
| 251 |
+
"grad_norm": 0.05265355110168457,
|
| 252 |
+
"learning_rate": 0.00897582910666454,
|
| 253 |
+
"loss": 0.2626,
|
| 254 |
+
"step": 700
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.5742411812961444,
|
| 258 |
+
"eval_loss": 0.25007763504981995,
|
| 259 |
+
"eval_runtime": 19.346,
|
| 260 |
+
"eval_samples_per_second": 51.69,
|
| 261 |
+
"eval_steps_per_second": 0.827,
|
| 262 |
+
"step": 700
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.5947497949138638,
|
| 266 |
+
"grad_norm": 0.06907053291797638,
|
| 267 |
+
"learning_rate": 0.008864768379800016,
|
| 268 |
+
"loss": 0.2605,
|
| 269 |
+
"step": 725
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.6152584085315833,
|
| 273 |
+
"grad_norm": 0.055843643844127655,
|
| 274 |
+
"learning_rate": 0.008748755619033153,
|
| 275 |
+
"loss": 0.2576,
|
| 276 |
+
"step": 750
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.6357670221493027,
|
| 280 |
+
"grad_norm": 0.05601644888520241,
|
| 281 |
+
"learning_rate": 0.00862793947470155,
|
| 282 |
+
"loss": 0.2541,
|
| 283 |
+
"step": 775
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.6562756357670222,
|
| 287 |
+
"grad_norm": 0.05217461287975311,
|
| 288 |
+
"learning_rate": 0.008502474751850142,
|
| 289 |
+
"loss": 0.2582,
|
| 290 |
+
"step": 800
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.6562756357670222,
|
| 294 |
+
"eval_loss": 0.2471487671136856,
|
| 295 |
+
"eval_runtime": 19.3575,
|
| 296 |
+
"eval_samples_per_second": 51.66,
|
| 297 |
+
"eval_steps_per_second": 0.827,
|
| 298 |
+
"step": 800
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 0.6767842493847416,
|
| 302 |
+
"grad_norm": 0.04733233153820038,
|
| 303 |
+
"learning_rate": 0.008372522211875224,
|
| 304 |
+
"loss": 0.252,
|
| 305 |
+
"step": 825
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 0.6972928630024611,
|
| 309 |
+
"grad_norm": 0.04515957087278366,
|
| 310 |
+
"learning_rate": 0.008238248366536473,
|
| 311 |
+
"loss": 0.2551,
|
| 312 |
+
"step": 850
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 0.7178014766201805,
|
| 316 |
+
"grad_norm": 0.041885215789079666,
|
| 317 |
+
"learning_rate": 0.008099825264600843,
|
| 318 |
+
"loss": 0.2435,
|
| 319 |
+
"step": 875
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 0.7383100902379,
|
| 323 |
+
"grad_norm": 0.052805330604314804,
|
| 324 |
+
"learning_rate": 0.007957430271391761,
|
| 325 |
+
"loss": 0.2491,
|
| 326 |
+
"step": 900
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 0.7383100902379,
|
| 330 |
+
"eval_loss": 0.23912134766578674,
|
| 331 |
+
"eval_runtime": 19.331,
|
| 332 |
+
"eval_samples_per_second": 51.73,
|
| 333 |
+
"eval_steps_per_second": 0.828,
|
| 334 |
+
"step": 900
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"epoch": 0.7588187038556193,
|
| 338 |
+
"grad_norm": 0.04989820346236229,
|
| 339 |
+
"learning_rate": 0.007811245841526062,
|
| 340 |
+
"loss": 0.2507,
|
| 341 |
+
"step": 925
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 0.7793273174733388,
|
| 345 |
+
"grad_norm": 0.05028345808386803,
|
| 346 |
+
"learning_rate": 0.00766145928512988,
|
| 347 |
+
"loss": 0.2383,
|
| 348 |
+
"step": 950
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"epoch": 0.7998359310910582,
|
| 352 |
+
"grad_norm": 0.04753001779317856,
|
| 353 |
+
"learning_rate": 0.007508262527833029,
|
| 354 |
+
"loss": 0.2462,
|
| 355 |
+
"step": 975
|
| 356 |
+
},
|
| 357 |
+
{
|
| 358 |
+
"epoch": 0.8203445447087777,
|
| 359 |
+
"grad_norm": 0.0477447547018528,
|
| 360 |
+
"learning_rate": 0.00735185186484943,
|
| 361 |
+
"loss": 0.2393,
|
| 362 |
+
"step": 1000
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.8203445447087777,
|
| 366 |
+
"eval_loss": 0.23531778156757355,
|
| 367 |
+
"eval_runtime": 19.342,
|
| 368 |
+
"eval_samples_per_second": 51.701,
|
| 369 |
+
"eval_steps_per_second": 0.827,
|
| 370 |
+
"step": 1000
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 0.8408531583264971,
|
| 374 |
+
"grad_norm": 0.04486614465713501,
|
| 375 |
+
"learning_rate": 0.007192427709458655,
|
| 376 |
+
"loss": 0.2382,
|
| 377 |
+
"step": 1025
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 0.8613617719442166,
|
| 381 |
+
"grad_norm": 0.04273708537220955,
|
| 382 |
+
"learning_rate": 0.007030194336210888,
|
| 383 |
+
"loss": 0.2363,
|
| 384 |
+
"step": 1050
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.881870385561936,
|
| 388 |
+
"grad_norm": 0.04644826427102089,
|
| 389 |
+
"learning_rate": 0.00686535961918433,
|
| 390 |
+
"loss": 0.2409,
|
| 391 |
+
"step": 1075
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 0.9023789991796555,
|
| 395 |
+
"grad_norm": 0.04469456896185875,
|
| 396 |
+
"learning_rate": 0.006698134765630434,
|
| 397 |
+
"loss": 0.2419,
|
| 398 |
+
"step": 1100
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 0.9023789991796555,
|
| 402 |
+
"eval_loss": 0.2309047281742096,
|
| 403 |
+
"eval_runtime": 19.2884,
|
| 404 |
+
"eval_samples_per_second": 51.845,
|
| 405 |
+
"eval_steps_per_second": 0.83,
|
| 406 |
+
"step": 1100
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 0.9228876127973749,
|
| 410 |
+
"grad_norm": 0.04691525548696518,
|
| 411 |
+
"learning_rate": 0.006528734045348248,
|
| 412 |
+
"loss": 0.2363,
|
| 413 |
+
"step": 1125
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 0.9433962264150944,
|
| 417 |
+
"grad_norm": 0.04146049544215202,
|
| 418 |
+
"learning_rate": 0.006357374516134643,
|
| 419 |
+
"loss": 0.2371,
|
| 420 |
+
"step": 1150
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 0.9639048400328137,
|
| 424 |
+
"grad_norm": 0.043408554047346115,
|
| 425 |
+
"learning_rate": 0.006184275745662179,
|
| 426 |
+
"loss": 0.2422,
|
| 427 |
+
"step": 1175
|
| 428 |
+
},
|
| 429 |
+
{
|
| 430 |
+
"epoch": 0.9844134536505332,
|
| 431 |
+
"grad_norm": 0.0436575785279274,
|
| 432 |
+
"learning_rate": 0.006009659530141031,
|
| 433 |
+
"loss": 0.2477,
|
| 434 |
+
"step": 1200
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 0.9844134536505332,
|
| 438 |
+
"eval_loss": 0.22737036645412445,
|
| 439 |
+
"eval_runtime": 19.2923,
|
| 440 |
+
"eval_samples_per_second": 51.834,
|
| 441 |
+
"eval_steps_per_second": 0.829,
|
| 442 |
+
"step": 1200
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 1.0049220672682526,
|
| 446 |
+
"grad_norm": 0.0417296439409256,
|
| 447 |
+
"learning_rate": 0.005833749610125402,
|
| 448 |
+
"loss": 0.2257,
|
| 449 |
+
"step": 1225
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 1.0254306808859721,
|
| 453 |
+
"grad_norm": 0.04858849197626114,
|
| 454 |
+
"learning_rate": 0.005656771383828601,
|
| 455 |
+
"loss": 0.2048,
|
| 456 |
+
"step": 1250
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 1.0459392945036916,
|
| 460 |
+
"grad_norm": 0.04328998550772667,
|
| 461 |
+
"learning_rate": 0.005478951618314133,
|
| 462 |
+
"loss": 0.2053,
|
| 463 |
+
"step": 1275
|
| 464 |
+
},
|
| 465 |
+
{
|
| 466 |
+
"epoch": 1.066447908121411,
|
| 467 |
+
"grad_norm": 0.05659686401486397,
|
| 468 |
+
"learning_rate": 0.0053005181589328145,
|
| 469 |
+
"loss": 0.2081,
|
| 470 |
+
"step": 1300
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 1.066447908121411,
|
| 474 |
+
"eval_loss": 0.22519998252391815,
|
| 475 |
+
"eval_runtime": 19.2941,
|
| 476 |
+
"eval_samples_per_second": 51.829,
|
| 477 |
+
"eval_steps_per_second": 0.829,
|
| 478 |
+
"step": 1300
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 1.0869565217391304,
|
| 482 |
+
"grad_norm": 0.045947764068841934,
|
| 483 |
+
"learning_rate": 0.005121699637378282,
|
| 484 |
+
"loss": 0.207,
|
| 485 |
+
"step": 1325
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 1.1074651353568499,
|
| 489 |
+
"grad_norm": 0.036780331283807755,
|
| 490 |
+
"learning_rate": 0.004942725178734904,
|
| 491 |
+
"loss": 0.1988,
|
| 492 |
+
"step": 1350
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 1.1279737489745694,
|
| 496 |
+
"grad_norm": 0.04897301271557808,
|
| 497 |
+
"learning_rate": 0.0047638241078935325,
|
| 498 |
+
"loss": 0.201,
|
| 499 |
+
"step": 1375
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 1.1484823625922886,
|
| 503 |
+
"grad_norm": 0.04408912733197212,
|
| 504 |
+
"learning_rate": 0.00458522565571121,
|
| 505 |
+
"loss": 0.1952,
|
| 506 |
+
"step": 1400
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 1.1484823625922886,
|
| 510 |
+
"eval_loss": 0.2233022004365921,
|
| 511 |
+
"eval_runtime": 19.2658,
|
| 512 |
+
"eval_samples_per_second": 51.905,
|
| 513 |
+
"eval_steps_per_second": 0.83,
|
| 514 |
+
"step": 1400
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 1.1689909762100081,
|
| 518 |
+
"grad_norm": 0.049094438552856445,
|
| 519 |
+
"learning_rate": 0.004407158665291376,
|
| 520 |
+
"loss": 0.2056,
|
| 521 |
+
"step": 1425
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 1.1894995898277276,
|
| 525 |
+
"grad_norm": 0.06326494365930557,
|
| 526 |
+
"learning_rate": 0.004229851298760915,
|
| 527 |
+
"loss": 0.2056,
|
| 528 |
+
"step": 1450
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 1.2100082034454471,
|
| 532 |
+
"grad_norm": 0.05762980133295059,
|
| 533 |
+
"learning_rate": 0.004053530744919749,
|
| 534 |
+
"loss": 0.202,
|
| 535 |
+
"step": 1475
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 1.2305168170631666,
|
| 539 |
+
"grad_norm": 0.04594657942652702,
|
| 540 |
+
"learning_rate": 0.0038784229281375966,
|
| 541 |
+
"loss": 0.2109,
|
| 542 |
+
"step": 1500
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 1.2305168170631666,
|
| 546 |
+
"eval_loss": 0.22034893929958344,
|
| 547 |
+
"eval_runtime": 19.2907,
|
| 548 |
+
"eval_samples_per_second": 51.839,
|
| 549 |
+
"eval_steps_per_second": 0.829,
|
| 550 |
+
"step": 1500
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 1.251025430680886,
|
| 554 |
+
"grad_norm": 0.050695791840553284,
|
| 555 |
+
"learning_rate": 0.003704752218870861,
|
| 556 |
+
"loss": 0.2054,
|
| 557 |
+
"step": 1525
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 1.2715340442986054,
|
| 561 |
+
"grad_norm": 0.03723922371864319,
|
| 562 |
+
"learning_rate": 0.003532741146170603,
|
| 563 |
+
"loss": 0.1991,
|
| 564 |
+
"step": 1550
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 1.2920426579163249,
|
| 568 |
+
"grad_norm": 0.036494262516498566,
|
| 569 |
+
"learning_rate": 0.0033626101125499553,
|
| 570 |
+
"loss": 0.2013,
|
| 571 |
+
"step": 1575
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 1.3125512715340442,
|
| 575 |
+
"grad_norm": 0.04479338973760605,
|
| 576 |
+
"learning_rate": 0.0031945771115763333,
|
| 577 |
+
"loss": 0.1941,
|
| 578 |
+
"step": 1600
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 1.3125512715340442,
|
| 582 |
+
"eval_loss": 0.21673938632011414,
|
| 583 |
+
"eval_runtime": 19.2817,
|
| 584 |
+
"eval_samples_per_second": 51.863,
|
| 585 |
+
"eval_steps_per_second": 0.83,
|
| 586 |
+
"step": 1600
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 1.3330598851517639,
|
| 590 |
+
"grad_norm": 0.03847524896264076,
|
| 591 |
+
"learning_rate": 0.0030288574485502757,
|
| 592 |
+
"loss": 0.1972,
|
| 593 |
+
"step": 1625
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 1.3535684987694832,
|
| 597 |
+
"grad_norm": 0.04458480328321457,
|
| 598 |
+
"learning_rate": 0.0028656634646288567,
|
| 599 |
+
"loss": 0.1977,
|
| 600 |
+
"step": 1650
|
| 601 |
+
},
|
| 602 |
+
{
|
| 603 |
+
"epoch": 1.3740771123872026,
|
| 604 |
+
"grad_norm": 0.0475829616189003,
|
| 605 |
+
"learning_rate": 0.0027052042647471254,
|
| 606 |
+
"loss": 0.1942,
|
| 607 |
+
"step": 1675
|
| 608 |
+
},
|
| 609 |
+
{
|
| 610 |
+
"epoch": 1.3945857260049221,
|
| 611 |
+
"grad_norm": 0.03854650631546974,
|
| 612 |
+
"learning_rate": 0.0025476854496862054,
|
| 613 |
+
"loss": 0.1962,
|
| 614 |
+
"step": 1700
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 1.3945857260049221,
|
| 618 |
+
"eval_loss": 0.21376186609268188,
|
| 619 |
+
"eval_runtime": 19.3671,
|
| 620 |
+
"eval_samples_per_second": 51.634,
|
| 621 |
+
"eval_steps_per_second": 0.826,
|
| 622 |
+
"step": 1700
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"epoch": 1.4150943396226414,
|
| 626 |
+
"grad_norm": 0.040612999349832535,
|
| 627 |
+
"learning_rate": 0.002393308852631373,
|
| 628 |
+
"loss": 0.1924,
|
| 629 |
+
"step": 1725
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 1.435602953240361,
|
| 633 |
+
"grad_norm": 0.0439542718231678,
|
| 634 |
+
"learning_rate": 0.0022422722805576453,
|
| 635 |
+
"loss": 0.1909,
|
| 636 |
+
"step": 1750
|
| 637 |
+
},
|
| 638 |
+
{
|
| 639 |
+
"epoch": 1.4561115668580804,
|
| 640 |
+
"grad_norm": 0.0390244759619236,
|
| 641 |
+
"learning_rate": 0.002094769260774262,
|
| 642 |
+
"loss": 0.1973,
|
| 643 |
+
"step": 1775
|
| 644 |
+
},
|
| 645 |
+
{
|
| 646 |
+
"epoch": 1.4766201804758,
|
| 647 |
+
"grad_norm": 0.03593549132347107,
|
| 648 |
+
"learning_rate": 0.0019509887929528458,
|
| 649 |
+
"loss": 0.1883,
|
| 650 |
+
"step": 1800
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 1.4766201804758,
|
| 654 |
+
"eval_loss": 0.21154361963272095,
|
| 655 |
+
"eval_runtime": 19.2617,
|
| 656 |
+
"eval_samples_per_second": 51.916,
|
| 657 |
+
"eval_steps_per_second": 0.831,
|
| 658 |
+
"step": 1800
|
| 659 |
+
},
|
| 660 |
+
{
|
| 661 |
+
"epoch": 1.4971287940935194,
|
| 662 |
+
"grad_norm": 0.03686383366584778,
|
| 663 |
+
"learning_rate": 0.0018111151069569182,
|
| 664 |
+
"loss": 0.1874,
|
| 665 |
+
"step": 1825
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 1.5176374077112387,
|
| 669 |
+
"grad_norm": 0.04321776703000069,
|
| 670 |
+
"learning_rate": 0.0016753274267831113,
|
| 671 |
+
"loss": 0.1901,
|
| 672 |
+
"step": 1850
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 1.5381460213289582,
|
| 676 |
+
"grad_norm": 0.03843892365694046,
|
| 677 |
+
"learning_rate": 0.0015437997409165477,
|
| 678 |
+
"loss": 0.19,
|
| 679 |
+
"step": 1875
|
| 680 |
+
},
|
| 681 |
+
{
|
| 682 |
+
"epoch": 1.5586546349466777,
|
| 683 |
+
"grad_norm": 0.04191848263144493,
|
| 684 |
+
"learning_rate": 0.0014167005793946036,
|
| 685 |
+
"loss": 0.1929,
|
| 686 |
+
"step": 1900
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 1.5586546349466777,
|
| 690 |
+
"eval_loss": 0.20885708928108215,
|
| 691 |
+
"eval_runtime": 19.2521,
|
| 692 |
+
"eval_samples_per_second": 51.942,
|
| 693 |
+
"eval_steps_per_second": 0.831,
|
| 694 |
+
"step": 1900
|
| 695 |
+
},
|
| 696 |
+
{
|
| 697 |
+
"epoch": 1.579163248564397,
|
| 698 |
+
"grad_norm": 0.04100554808974266,
|
| 699 |
+
"learning_rate": 0.0012941927978647528,
|
| 700 |
+
"loss": 0.1939,
|
| 701 |
+
"step": 1925
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"epoch": 1.5996718621821167,
|
| 705 |
+
"grad_norm": 0.04225098714232445,
|
| 706 |
+
"learning_rate": 0.0011764333689131385,
|
| 707 |
+
"loss": 0.1924,
|
| 708 |
+
"step": 1950
|
| 709 |
+
},
|
| 710 |
+
{
|
| 711 |
+
"epoch": 1.620180475799836,
|
| 712 |
+
"grad_norm": 0.03872651979327202,
|
| 713 |
+
"learning_rate": 0.0010635731809312992,
|
| 714 |
+
"loss": 0.1872,
|
| 715 |
+
"step": 1975
|
| 716 |
+
},
|
| 717 |
+
{
|
| 718 |
+
"epoch": 1.6406890894175554,
|
| 719 |
+
"grad_norm": 0.04000687971711159,
|
| 720 |
+
"learning_rate": 0.0009557568447787202,
|
| 721 |
+
"loss": 0.1941,
|
| 722 |
+
"step": 2000
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 1.6406890894175554,
|
| 726 |
+
"eval_loss": 0.20660191774368286,
|
| 727 |
+
"eval_runtime": 19.2828,
|
| 728 |
+
"eval_samples_per_second": 51.86,
|
| 729 |
+
"eval_steps_per_second": 0.83,
|
| 730 |
+
"step": 2000
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 1.661197703035275,
|
| 734 |
+
"grad_norm": 0.03831176087260246,
|
| 735 |
+
"learning_rate": 0.0008531225084889654,
|
| 736 |
+
"loss": 0.1829,
|
| 737 |
+
"step": 2025
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 1.6817063166529942,
|
| 741 |
+
"grad_norm": 0.044930826872587204,
|
| 742 |
+
"learning_rate": 0.0007558016802568091,
|
| 743 |
+
"loss": 0.185,
|
| 744 |
+
"step": 2050
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 1.7022149302707137,
|
| 748 |
+
"grad_norm": 0.05037025734782219,
|
| 749 |
+
"learning_rate": 0.0006639190599331745,
|
| 750 |
+
"loss": 0.191,
|
| 751 |
+
"step": 2075
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 1.7227235438884332,
|
| 755 |
+
"grad_norm": 0.042025428265333176,
|
| 756 |
+
"learning_rate": 0.0005775923792437865,
|
| 757 |
+
"loss": 0.182,
|
| 758 |
+
"step": 2100
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 1.7227235438884332,
|
| 762 |
+
"eval_loss": 0.20513421297073364,
|
| 763 |
+
"eval_runtime": 19.6005,
|
| 764 |
+
"eval_samples_per_second": 51.019,
|
| 765 |
+
"eval_steps_per_second": 0.816,
|
| 766 |
+
"step": 2100
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 1.7432321575061525,
|
| 770 |
+
"grad_norm": 0.03695955500006676,
|
| 771 |
+
"learning_rate": 0.0004969322509362762,
|
| 772 |
+
"loss": 0.185,
|
| 773 |
+
"step": 2125
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 1.7637407711238722,
|
| 777 |
+
"grad_norm": 0.03301229700446129,
|
| 778 |
+
"learning_rate": 0.0004220420270490294,
|
| 779 |
+
"loss": 0.1827,
|
| 780 |
+
"step": 2150
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 1.7842493847415914,
|
| 784 |
+
"grad_norm": 0.03758403658866882,
|
| 785 |
+
"learning_rate": 0.0003530176664833834,
|
| 786 |
+
"loss": 0.1916,
|
| 787 |
+
"step": 2175
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 1.804757998359311,
|
| 791 |
+
"grad_norm": 0.03936561569571495,
|
| 792 |
+
"learning_rate": 0.00028994761204884756,
|
| 793 |
+
"loss": 0.1852,
|
| 794 |
+
"step": 2200
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 1.804757998359311,
|
| 798 |
+
"eval_loss": 0.2034943699836731,
|
| 799 |
+
"eval_runtime": 19.2867,
|
| 800 |
+
"eval_samples_per_second": 51.849,
|
| 801 |
+
"eval_steps_per_second": 0.83,
|
| 802 |
+
"step": 2200
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 1.8252666119770304,
|
| 806 |
+
"grad_norm": 0.040019210427999496,
|
| 807 |
+
"learning_rate": 0.00023291267713889953,
|
| 808 |
+
"loss": 0.1852,
|
| 809 |
+
"step": 2225
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 1.8457752255947497,
|
| 813 |
+
"grad_norm": 0.04008966311812401,
|
| 814 |
+
"learning_rate": 0.00018198594218256814,
|
| 815 |
+
"loss": 0.1873,
|
| 816 |
+
"step": 2250
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"epoch": 1.8662838392124692,
|
| 820 |
+
"grad_norm": 0.03722105920314789,
|
| 821 |
+
"learning_rate": 0.00013723266100447052,
|
| 822 |
+
"loss": 0.1924,
|
| 823 |
+
"step": 2275
|
| 824 |
+
},
|
| 825 |
+
{
|
| 826 |
+
"epoch": 1.8867924528301887,
|
| 827 |
+
"grad_norm": 0.03645153343677521,
|
| 828 |
+
"learning_rate": 9.871017721329201e-05,
|
| 829 |
+
"loss": 0.1845,
|
| 830 |
+
"step": 2300
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 1.8867924528301887,
|
| 834 |
+
"eval_loss": 0.20311051607131958,
|
| 835 |
+
"eval_runtime": 19.2814,
|
| 836 |
+
"eval_samples_per_second": 51.864,
|
| 837 |
+
"eval_steps_per_second": 0.83,
|
| 838 |
+
"step": 2300
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 1.907301066447908,
|
| 842 |
+
"grad_norm": 0.04580986499786377,
|
| 843 |
+
"learning_rate": 6.64678507258487e-05,
|
| 844 |
+
"loss": 0.1922,
|
| 845 |
+
"step": 2325
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 1.9278096800656277,
|
| 849 |
+
"grad_norm": 0.04106990993022919,
|
| 850 |
+
"learning_rate": 4.054699452086641e-05,
|
| 851 |
+
"loss": 0.1773,
|
| 852 |
+
"step": 2350
|
| 853 |
+
},
|
| 854 |
+
{
|
| 855 |
+
"epoch": 1.948318293683347,
|
| 856 |
+
"grad_norm": 0.03752952069044113,
|
| 857 |
+
"learning_rate": 2.0980821703527885e-05,
|
| 858 |
+
"loss": 0.1823,
|
| 859 |
+
"step": 2375
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 1.9688269073010665,
|
| 863 |
+
"grad_norm": 0.03623511642217636,
|
| 864 |
+
"learning_rate": 7.794402948607671e-06,
|
| 865 |
+
"loss": 0.1841,
|
| 866 |
+
"step": 2400
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 1.9688269073010665,
|
| 870 |
+
"eval_loss": 0.20291264355182648,
|
| 871 |
+
"eval_runtime": 19.3204,
|
| 872 |
+
"eval_samples_per_second": 51.759,
|
| 873 |
+
"eval_steps_per_second": 0.828,
|
| 874 |
+
"step": 2400
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"epoch": 1.989335520918786,
|
| 878 |
+
"grad_norm": 0.03788290545344353,
|
| 879 |
+
"learning_rate": 1.0046343767294852e-06,
|
| 880 |
+
"loss": 0.1805,
|
| 881 |
+
"step": 2425
|
| 882 |
+
},
|
| 883 |
+
{
|
| 884 |
+
"epoch": 2.0,
|
| 885 |
+
"step": 2438,
|
| 886 |
+
"total_flos": 1.58523627405312e+18,
|
| 887 |
+
"train_loss": 0.2427779671791834,
|
| 888 |
+
"train_runtime": 2867.9661,
|
| 889 |
+
"train_samples_per_second": 27.197,
|
| 890 |
+
"train_steps_per_second": 0.85
|
| 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_ex18/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_exps7/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_exps7/ft/added_tokens.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"[PAD]": 32000
|
| 3 |
+
}
|
nl_tasks/exps/run_exps7/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_exps7/ft/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
nl_tasks/exps/run_exps7/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_exps7/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_exps7/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_exps7/ft2/adapter_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:93f6652c2396681417a15cf0604fbf5b8fbd7e3734ced348eda5f5148bdce429
|
| 3 |
+
size 33602915
|
nl_tasks/exps/run_exps7/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": 0.22137899696826935,
|
| 15 |
+
"learning_rate": 9.836065573770491e-05,
|
| 16 |
+
"loss": 0.6806,
|
| 17 |
+
"step": 25
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.04101722723543889,
|
| 21 |
+
"grad_norm": 0.2084890753030777,
|
| 22 |
+
"learning_rate": 0.00020081967213114754,
|
| 23 |
+
"loss": 0.4176,
|
| 24 |
+
"step": 50
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.06152584085315833,
|
| 28 |
+
"grad_norm": 0.1516050547361374,
|
| 29 |
+
"learning_rate": 0.0003032786885245902,
|
| 30 |
+
"loss": 0.3741,
|
| 31 |
+
"step": 75
|
| 32 |
+
},
|
| 33 |
+
{
|
| 34 |
+
"epoch": 0.08203445447087777,
|
| 35 |
+
"grad_norm": 0.18930844962596893,
|
| 36 |
+
"learning_rate": 0.0004057377049180328,
|
| 37 |
+
"loss": 0.3369,
|
| 38 |
+
"step": 100
|
| 39 |
+
},
|
| 40 |
+
{
|
| 41 |
+
"epoch": 0.08203445447087777,
|
| 42 |
+
"eval_loss": 0.3269580900669098,
|
| 43 |
+
"eval_runtime": 19.5479,
|
| 44 |
+
"eval_samples_per_second": 51.156,
|
| 45 |
+
"eval_steps_per_second": 0.819,
|
| 46 |
+
"step": 100
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"epoch": 0.10254306808859721,
|
| 50 |
+
"grad_norm": 0.16576114296913147,
|
| 51 |
+
"learning_rate": 0.0005081967213114754,
|
| 52 |
+
"loss": 0.3407,
|
| 53 |
+
"step": 125
|
| 54 |
+
},
|
| 55 |
+
{
|
| 56 |
+
"epoch": 0.12305168170631665,
|
| 57 |
+
"grad_norm": 0.18032360076904297,
|
| 58 |
+
"learning_rate": 0.000610655737704918,
|
| 59 |
+
"loss": 0.3225,
|
| 60 |
+
"step": 150
|
| 61 |
+
},
|
| 62 |
+
{
|
| 63 |
+
"epoch": 0.1435602953240361,
|
| 64 |
+
"grad_norm": 0.21672417223453522,
|
| 65 |
+
"learning_rate": 0.0007131147540983607,
|
| 66 |
+
"loss": 0.3131,
|
| 67 |
+
"step": 175
|
| 68 |
+
},
|
| 69 |
+
{
|
| 70 |
+
"epoch": 0.16406890894175555,
|
| 71 |
+
"grad_norm": 0.2109616994857788,
|
| 72 |
+
"learning_rate": 0.0008155737704918033,
|
| 73 |
+
"loss": 0.3242,
|
| 74 |
+
"step": 200
|
| 75 |
+
},
|
| 76 |
+
{
|
| 77 |
+
"epoch": 0.16406890894175555,
|
| 78 |
+
"eval_loss": 0.30799221992492676,
|
| 79 |
+
"eval_runtime": 19.3498,
|
| 80 |
+
"eval_samples_per_second": 51.68,
|
| 81 |
+
"eval_steps_per_second": 0.827,
|
| 82 |
+
"step": 200
|
| 83 |
+
},
|
| 84 |
+
{
|
| 85 |
+
"epoch": 0.184577522559475,
|
| 86 |
+
"grad_norm": 0.20327340066432953,
|
| 87 |
+
"learning_rate": 0.0009180327868852459,
|
| 88 |
+
"loss": 0.3222,
|
| 89 |
+
"step": 225
|
| 90 |
+
},
|
| 91 |
+
{
|
| 92 |
+
"epoch": 0.20508613617719443,
|
| 93 |
+
"grad_norm": 0.2256147265434265,
|
| 94 |
+
"learning_rate": 0.0009999871854116063,
|
| 95 |
+
"loss": 0.3012,
|
| 96 |
+
"step": 250
|
| 97 |
+
},
|
| 98 |
+
{
|
| 99 |
+
"epoch": 0.22559474979491387,
|
| 100 |
+
"grad_norm": 0.20731832087039948,
|
| 101 |
+
"learning_rate": 0.0009995387437838027,
|
| 102 |
+
"loss": 0.3034,
|
| 103 |
+
"step": 275
|
| 104 |
+
},
|
| 105 |
+
{
|
| 106 |
+
"epoch": 0.2461033634126333,
|
| 107 |
+
"grad_norm": 0.255392462015152,
|
| 108 |
+
"learning_rate": 0.000998450229439693,
|
| 109 |
+
"loss": 0.3062,
|
| 110 |
+
"step": 300
|
| 111 |
+
},
|
| 112 |
+
{
|
| 113 |
+
"epoch": 0.2461033634126333,
|
| 114 |
+
"eval_loss": 0.29782918095588684,
|
| 115 |
+
"eval_runtime": 19.3507,
|
| 116 |
+
"eval_samples_per_second": 51.678,
|
| 117 |
+
"eval_steps_per_second": 0.827,
|
| 118 |
+
"step": 300
|
| 119 |
+
},
|
| 120 |
+
{
|
| 121 |
+
"epoch": 0.2666119770303528,
|
| 122 |
+
"grad_norm": 0.2502018213272095,
|
| 123 |
+
"learning_rate": 0.000996723037122612,
|
| 124 |
+
"loss": 0.3102,
|
| 125 |
+
"step": 325
|
| 126 |
+
},
|
| 127 |
+
{
|
| 128 |
+
"epoch": 0.2871205906480722,
|
| 129 |
+
"grad_norm": 0.2249726802110672,
|
| 130 |
+
"learning_rate": 0.0009943593799315263,
|
| 131 |
+
"loss": 0.3007,
|
| 132 |
+
"step": 350
|
| 133 |
+
},
|
| 134 |
+
{
|
| 135 |
+
"epoch": 0.30762920426579166,
|
| 136 |
+
"grad_norm": 0.24895645678043365,
|
| 137 |
+
"learning_rate": 0.0009913622864853324,
|
| 138 |
+
"loss": 0.3104,
|
| 139 |
+
"step": 375
|
| 140 |
+
},
|
| 141 |
+
{
|
| 142 |
+
"epoch": 0.3281378178835111,
|
| 143 |
+
"grad_norm": 0.2924279272556305,
|
| 144 |
+
"learning_rate": 0.0009877355970422024,
|
| 145 |
+
"loss": 0.2939,
|
| 146 |
+
"step": 400
|
| 147 |
+
},
|
| 148 |
+
{
|
| 149 |
+
"epoch": 0.3281378178835111,
|
| 150 |
+
"eval_loss": 0.2853645384311676,
|
| 151 |
+
"eval_runtime": 19.368,
|
| 152 |
+
"eval_samples_per_second": 51.631,
|
| 153 |
+
"eval_steps_per_second": 0.826,
|
| 154 |
+
"step": 400
|
| 155 |
+
},
|
| 156 |
+
{
|
| 157 |
+
"epoch": 0.34864643150123054,
|
| 158 |
+
"grad_norm": 0.22633150219917297,
|
| 159 |
+
"learning_rate": 0.0009834839585789559,
|
| 160 |
+
"loss": 0.3027,
|
| 161 |
+
"step": 425
|
| 162 |
+
},
|
| 163 |
+
{
|
| 164 |
+
"epoch": 0.36915504511895,
|
| 165 |
+
"grad_norm": 0.25082913041114807,
|
| 166 |
+
"learning_rate": 0.000978612818836762,
|
| 167 |
+
"loss": 0.2995,
|
| 168 |
+
"step": 450
|
| 169 |
+
},
|
| 170 |
+
{
|
| 171 |
+
"epoch": 0.3896636587366694,
|
| 172 |
+
"grad_norm": 0.2795739769935608,
|
| 173 |
+
"learning_rate": 0.0009731284193407981,
|
| 174 |
+
"loss": 0.2877,
|
| 175 |
+
"step": 475
|
| 176 |
+
},
|
| 177 |
+
{
|
| 178 |
+
"epoch": 0.41017227235438886,
|
| 179 |
+
"grad_norm": 0.21229897439479828,
|
| 180 |
+
"learning_rate": 0.0009670377874028117,
|
| 181 |
+
"loss": 0.2735,
|
| 182 |
+
"step": 500
|
| 183 |
+
},
|
| 184 |
+
{
|
| 185 |
+
"epoch": 0.41017227235438886,
|
| 186 |
+
"eval_loss": 0.27481311559677124,
|
| 187 |
+
"eval_runtime": 19.3461,
|
| 188 |
+
"eval_samples_per_second": 51.69,
|
| 189 |
+
"eval_steps_per_second": 0.827,
|
| 190 |
+
"step": 500
|
| 191 |
+
},
|
| 192 |
+
{
|
| 193 |
+
"epoch": 0.4306808859721083,
|
| 194 |
+
"grad_norm": 0.27039724588394165,
|
| 195 |
+
"learning_rate": 0.0009603487271168336,
|
| 196 |
+
"loss": 0.273,
|
| 197 |
+
"step": 525
|
| 198 |
+
},
|
| 199 |
+
{
|
| 200 |
+
"epoch": 0.45118949958982774,
|
| 201 |
+
"grad_norm": 0.2912074625492096,
|
| 202 |
+
"learning_rate": 0.0009530698093595781,
|
| 203 |
+
"loss": 0.2825,
|
| 204 |
+
"step": 550
|
| 205 |
+
},
|
| 206 |
+
{
|
| 207 |
+
"epoch": 0.4716981132075472,
|
| 208 |
+
"grad_norm": 0.3081355392932892,
|
| 209 |
+
"learning_rate": 0.0009452103608083418,
|
| 210 |
+
"loss": 0.2757,
|
| 211 |
+
"step": 575
|
| 212 |
+
},
|
| 213 |
+
{
|
| 214 |
+
"epoch": 0.4922067268252666,
|
| 215 |
+
"grad_norm": 0.21602022647857666,
|
| 216 |
+
"learning_rate": 0.0009367804519904775,
|
| 217 |
+
"loss": 0.2916,
|
| 218 |
+
"step": 600
|
| 219 |
+
},
|
| 220 |
+
{
|
| 221 |
+
"epoch": 0.4922067268252666,
|
| 222 |
+
"eval_loss": 0.2674732804298401,
|
| 223 |
+
"eval_runtime": 19.3679,
|
| 224 |
+
"eval_samples_per_second": 51.632,
|
| 225 |
+
"eval_steps_per_second": 0.826,
|
| 226 |
+
"step": 600
|
| 227 |
+
},
|
| 228 |
+
{
|
| 229 |
+
"epoch": 0.5127153404429861,
|
| 230 |
+
"grad_norm": 0.2135145366191864,
|
| 231 |
+
"learning_rate": 0.0009277908843797492,
|
| 232 |
+
"loss": 0.2696,
|
| 233 |
+
"step": 625
|
| 234 |
+
},
|
| 235 |
+
{
|
| 236 |
+
"epoch": 0.5332239540607056,
|
| 237 |
+
"grad_norm": 0.22797523438930511,
|
| 238 |
+
"learning_rate": 0.0009182531765561084,
|
| 239 |
+
"loss": 0.2606,
|
| 240 |
+
"step": 650
|
| 241 |
+
},
|
| 242 |
+
{
|
| 243 |
+
"epoch": 0.5537325676784249,
|
| 244 |
+
"grad_norm": 0.191336989402771,
|
| 245 |
+
"learning_rate": 0.0009081795494466201,
|
| 246 |
+
"loss": 0.2739,
|
| 247 |
+
"step": 675
|
| 248 |
+
},
|
| 249 |
+
{
|
| 250 |
+
"epoch": 0.5742411812961444,
|
| 251 |
+
"grad_norm": 0.2148035168647766,
|
| 252 |
+
"learning_rate": 0.0008975829106664539,
|
| 253 |
+
"loss": 0.2707,
|
| 254 |
+
"step": 700
|
| 255 |
+
},
|
| 256 |
+
{
|
| 257 |
+
"epoch": 0.5742411812961444,
|
| 258 |
+
"eval_loss": 0.2565022110939026,
|
| 259 |
+
"eval_runtime": 19.3435,
|
| 260 |
+
"eval_samples_per_second": 51.697,
|
| 261 |
+
"eval_steps_per_second": 0.827,
|
| 262 |
+
"step": 700
|
| 263 |
+
},
|
| 264 |
+
{
|
| 265 |
+
"epoch": 0.5947497949138638,
|
| 266 |
+
"grad_norm": 0.21229054033756256,
|
| 267 |
+
"learning_rate": 0.0008864768379800017,
|
| 268 |
+
"loss": 0.2683,
|
| 269 |
+
"step": 725
|
| 270 |
+
},
|
| 271 |
+
{
|
| 272 |
+
"epoch": 0.6152584085315833,
|
| 273 |
+
"grad_norm": 0.22327612340450287,
|
| 274 |
+
"learning_rate": 0.0008748755619033153,
|
| 275 |
+
"loss": 0.2652,
|
| 276 |
+
"step": 750
|
| 277 |
+
},
|
| 278 |
+
{
|
| 279 |
+
"epoch": 0.6357670221493027,
|
| 280 |
+
"grad_norm": 0.1960134655237198,
|
| 281 |
+
"learning_rate": 0.000862793947470155,
|
| 282 |
+
"loss": 0.2608,
|
| 283 |
+
"step": 775
|
| 284 |
+
},
|
| 285 |
+
{
|
| 286 |
+
"epoch": 0.6562756357670222,
|
| 287 |
+
"grad_norm": 0.21620357036590576,
|
| 288 |
+
"learning_rate": 0.0008502474751850142,
|
| 289 |
+
"loss": 0.2649,
|
| 290 |
+
"step": 800
|
| 291 |
+
},
|
| 292 |
+
{
|
| 293 |
+
"epoch": 0.6562756357670222,
|
| 294 |
+
"eval_loss": 0.2519683539867401,
|
| 295 |
+
"eval_runtime": 19.3701,
|
| 296 |
+
"eval_samples_per_second": 51.626,
|
| 297 |
+
"eval_steps_per_second": 0.826,
|
| 298 |
+
"step": 800
|
| 299 |
+
},
|
| 300 |
+
{
|
| 301 |
+
"epoch": 0.6767842493847416,
|
| 302 |
+
"grad_norm": 0.18587523698806763,
|
| 303 |
+
"learning_rate": 0.0008372522211875224,
|
| 304 |
+
"loss": 0.2578,
|
| 305 |
+
"step": 825
|
| 306 |
+
},
|
| 307 |
+
{
|
| 308 |
+
"epoch": 0.6972928630024611,
|
| 309 |
+
"grad_norm": 0.19317218661308289,
|
| 310 |
+
"learning_rate": 0.0008238248366536473,
|
| 311 |
+
"loss": 0.2597,
|
| 312 |
+
"step": 850
|
| 313 |
+
},
|
| 314 |
+
{
|
| 315 |
+
"epoch": 0.7178014766201805,
|
| 316 |
+
"grad_norm": 0.1740579754114151,
|
| 317 |
+
"learning_rate": 0.0008099825264600842,
|
| 318 |
+
"loss": 0.2476,
|
| 319 |
+
"step": 875
|
| 320 |
+
},
|
| 321 |
+
{
|
| 322 |
+
"epoch": 0.7383100902379,
|
| 323 |
+
"grad_norm": 0.21372999250888824,
|
| 324 |
+
"learning_rate": 0.0007957430271391761,
|
| 325 |
+
"loss": 0.2538,
|
| 326 |
+
"step": 900
|
| 327 |
+
},
|
| 328 |
+
{
|
| 329 |
+
"epoch": 0.7383100902379,
|
| 330 |
+
"eval_loss": 0.2426031082868576,
|
| 331 |
+
"eval_runtime": 19.3513,
|
| 332 |
+
"eval_samples_per_second": 51.676,
|
| 333 |
+
"eval_steps_per_second": 0.827,
|
| 334 |
+
"step": 900
|
| 335 |
+
},
|
| 336 |
+
{
|
| 337 |
+
"epoch": 0.7588187038556193,
|
| 338 |
+
"grad_norm": 0.19013312458992004,
|
| 339 |
+
"learning_rate": 0.0007811245841526062,
|
| 340 |
+
"loss": 0.254,
|
| 341 |
+
"step": 925
|
| 342 |
+
},
|
| 343 |
+
{
|
| 344 |
+
"epoch": 0.7793273174733388,
|
| 345 |
+
"grad_norm": 0.17855894565582275,
|
| 346 |
+
"learning_rate": 0.0007661459285129879,
|
| 347 |
+
"loss": 0.2419,
|
| 348 |
+
"step": 950
|
| 349 |
+
},
|
| 350 |
+
{
|
| 351 |
+
"epoch": 0.7998359310910582,
|
| 352 |
+
"grad_norm": 0.19467145204544067,
|
| 353 |
+
"learning_rate": 0.0007508262527833029,
|
| 354 |
+
"loss": 0.249,
|
| 355 |
+
"step": 975
|
| 356 |
+
},
|
| 357 |
+
{
|
| 358 |
+
"epoch": 0.8203445447087777,
|
| 359 |
+
"grad_norm": 0.19387587904930115,
|
| 360 |
+
"learning_rate": 0.000735185186484943,
|
| 361 |
+
"loss": 0.2427,
|
| 362 |
+
"step": 1000
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.8203445447087777,
|
| 366 |
+
"eval_loss": 0.23754794895648956,
|
| 367 |
+
"eval_runtime": 19.3524,
|
| 368 |
+
"eval_samples_per_second": 51.673,
|
| 369 |
+
"eval_steps_per_second": 0.827,
|
| 370 |
+
"step": 1000
|
| 371 |
+
},
|
| 372 |
+
{
|
| 373 |
+
"epoch": 0.8408531583264971,
|
| 374 |
+
"grad_norm": 0.20248214900493622,
|
| 375 |
+
"learning_rate": 0.0007192427709458656,
|
| 376 |
+
"loss": 0.2427,
|
| 377 |
+
"step": 1025
|
| 378 |
+
},
|
| 379 |
+
{
|
| 380 |
+
"epoch": 0.8613617719442166,
|
| 381 |
+
"grad_norm": 0.17144259810447693,
|
| 382 |
+
"learning_rate": 0.0007030194336210887,
|
| 383 |
+
"loss": 0.2385,
|
| 384 |
+
"step": 1050
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.881870385561936,
|
| 388 |
+
"grad_norm": 0.17359495162963867,
|
| 389 |
+
"learning_rate": 0.0006865359619184331,
|
| 390 |
+
"loss": 0.2445,
|
| 391 |
+
"step": 1075
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 0.9023789991796555,
|
| 395 |
+
"grad_norm": 0.19860954582691193,
|
| 396 |
+
"learning_rate": 0.0006698134765630434,
|
| 397 |
+
"loss": 0.2446,
|
| 398 |
+
"step": 1100
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 0.9023789991796555,
|
| 402 |
+
"eval_loss": 0.23382826149463654,
|
| 403 |
+
"eval_runtime": 19.3312,
|
| 404 |
+
"eval_samples_per_second": 51.73,
|
| 405 |
+
"eval_steps_per_second": 0.828,
|
| 406 |
+
"step": 1100
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 0.9228876127973749,
|
| 410 |
+
"grad_norm": 0.17317168414592743,
|
| 411 |
+
"learning_rate": 0.0006528734045348248,
|
| 412 |
+
"loss": 0.2394,
|
| 413 |
+
"step": 1125
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 0.9433962264150944,
|
| 417 |
+
"grad_norm": 0.15615008771419525,
|
| 418 |
+
"learning_rate": 0.0006357374516134643,
|
| 419 |
+
"loss": 0.2373,
|
| 420 |
+
"step": 1150
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 0.9639048400328137,
|
| 424 |
+
"grad_norm": 0.15703895688056946,
|
| 425 |
+
"learning_rate": 0.0006184275745662179,
|
| 426 |
+
"loss": 0.2427,
|
| 427 |
+
"step": 1175
|
| 428 |
+
},
|
| 429 |
+
{
|
| 430 |
+
"epoch": 0.9844134536505332,
|
| 431 |
+
"grad_norm": 0.1694665551185608,
|
| 432 |
+
"learning_rate": 0.0006009659530141031,
|
| 433 |
+
"loss": 0.2496,
|
| 434 |
+
"step": 1200
|
| 435 |
+
},
|
| 436 |
+
{
|
| 437 |
+
"epoch": 0.9844134536505332,
|
| 438 |
+
"eval_loss": 0.2290077954530716,
|
| 439 |
+
"eval_runtime": 19.3632,
|
| 440 |
+
"eval_samples_per_second": 51.644,
|
| 441 |
+
"eval_steps_per_second": 0.826,
|
| 442 |
+
"step": 1200
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 1.0049220672682526,
|
| 446 |
+
"grad_norm": 0.1727023869752884,
|
| 447 |
+
"learning_rate": 0.0005833749610125402,
|
| 448 |
+
"loss": 0.2285,
|
| 449 |
+
"step": 1225
|
| 450 |
+
},
|
| 451 |
+
{
|
| 452 |
+
"epoch": 1.0254306808859721,
|
| 453 |
+
"grad_norm": 0.16593652963638306,
|
| 454 |
+
"learning_rate": 0.0005656771383828602,
|
| 455 |
+
"loss": 0.2077,
|
| 456 |
+
"step": 1250
|
| 457 |
+
},
|
| 458 |
+
{
|
| 459 |
+
"epoch": 1.0459392945036916,
|
| 460 |
+
"grad_norm": 0.17790479958057404,
|
| 461 |
+
"learning_rate": 0.0005478951618314134,
|
| 462 |
+
"loss": 0.2087,
|
| 463 |
+
"step": 1275
|
| 464 |
+
},
|
| 465 |
+
{
|
| 466 |
+
"epoch": 1.066447908121411,
|
| 467 |
+
"grad_norm": 0.16105860471725464,
|
| 468 |
+
"learning_rate": 0.0005300518158932815,
|
| 469 |
+
"loss": 0.2112,
|
| 470 |
+
"step": 1300
|
| 471 |
+
},
|
| 472 |
+
{
|
| 473 |
+
"epoch": 1.066447908121411,
|
| 474 |
+
"eval_loss": 0.2252962589263916,
|
| 475 |
+
"eval_runtime": 19.3256,
|
| 476 |
+
"eval_samples_per_second": 51.745,
|
| 477 |
+
"eval_steps_per_second": 0.828,
|
| 478 |
+
"step": 1300
|
| 479 |
+
},
|
| 480 |
+
{
|
| 481 |
+
"epoch": 1.0869565217391304,
|
| 482 |
+
"grad_norm": 0.19076059758663177,
|
| 483 |
+
"learning_rate": 0.0005121699637378282,
|
| 484 |
+
"loss": 0.2083,
|
| 485 |
+
"step": 1325
|
| 486 |
+
},
|
| 487 |
+
{
|
| 488 |
+
"epoch": 1.1074651353568499,
|
| 489 |
+
"grad_norm": 0.14692914485931396,
|
| 490 |
+
"learning_rate": 0.0004942725178734903,
|
| 491 |
+
"loss": 0.2015,
|
| 492 |
+
"step": 1350
|
| 493 |
+
},
|
| 494 |
+
{
|
| 495 |
+
"epoch": 1.1279737489745694,
|
| 496 |
+
"grad_norm": 0.20046532154083252,
|
| 497 |
+
"learning_rate": 0.00047638241078935324,
|
| 498 |
+
"loss": 0.2031,
|
| 499 |
+
"step": 1375
|
| 500 |
+
},
|
| 501 |
+
{
|
| 502 |
+
"epoch": 1.1484823625922886,
|
| 503 |
+
"grad_norm": 0.16865970194339752,
|
| 504 |
+
"learning_rate": 0.000458522565571121,
|
| 505 |
+
"loss": 0.1972,
|
| 506 |
+
"step": 1400
|
| 507 |
+
},
|
| 508 |
+
{
|
| 509 |
+
"epoch": 1.1484823625922886,
|
| 510 |
+
"eval_loss": 0.2231229841709137,
|
| 511 |
+
"eval_runtime": 19.3499,
|
| 512 |
+
"eval_samples_per_second": 51.68,
|
| 513 |
+
"eval_steps_per_second": 0.827,
|
| 514 |
+
"step": 1400
|
| 515 |
+
},
|
| 516 |
+
{
|
| 517 |
+
"epoch": 1.1689909762100081,
|
| 518 |
+
"grad_norm": 0.17782826721668243,
|
| 519 |
+
"learning_rate": 0.00044071586652913767,
|
| 520 |
+
"loss": 0.2072,
|
| 521 |
+
"step": 1425
|
| 522 |
+
},
|
| 523 |
+
{
|
| 524 |
+
"epoch": 1.1894995898277276,
|
| 525 |
+
"grad_norm": 0.1706642061471939,
|
| 526 |
+
"learning_rate": 0.0004229851298760915,
|
| 527 |
+
"loss": 0.2062,
|
| 528 |
+
"step": 1450
|
| 529 |
+
},
|
| 530 |
+
{
|
| 531 |
+
"epoch": 1.2100082034454471,
|
| 532 |
+
"grad_norm": 0.17019104957580566,
|
| 533 |
+
"learning_rate": 0.0004053530744919749,
|
| 534 |
+
"loss": 0.2037,
|
| 535 |
+
"step": 1475
|
| 536 |
+
},
|
| 537 |
+
{
|
| 538 |
+
"epoch": 1.2305168170631666,
|
| 539 |
+
"grad_norm": 0.1775612235069275,
|
| 540 |
+
"learning_rate": 0.0003878422928137597,
|
| 541 |
+
"loss": 0.2124,
|
| 542 |
+
"step": 1500
|
| 543 |
+
},
|
| 544 |
+
{
|
| 545 |
+
"epoch": 1.2305168170631666,
|
| 546 |
+
"eval_loss": 0.22061145305633545,
|
| 547 |
+
"eval_runtime": 19.3712,
|
| 548 |
+
"eval_samples_per_second": 51.623,
|
| 549 |
+
"eval_steps_per_second": 0.826,
|
| 550 |
+
"step": 1500
|
| 551 |
+
},
|
| 552 |
+
{
|
| 553 |
+
"epoch": 1.251025430680886,
|
| 554 |
+
"grad_norm": 0.19087927043437958,
|
| 555 |
+
"learning_rate": 0.0003704752218870861,
|
| 556 |
+
"loss": 0.2065,
|
| 557 |
+
"step": 1525
|
| 558 |
+
},
|
| 559 |
+
{
|
| 560 |
+
"epoch": 1.2715340442986054,
|
| 561 |
+
"grad_norm": 0.15985721349716187,
|
| 562 |
+
"learning_rate": 0.00035327411461706025,
|
| 563 |
+
"loss": 0.2006,
|
| 564 |
+
"step": 1550
|
| 565 |
+
},
|
| 566 |
+
{
|
| 567 |
+
"epoch": 1.2920426579163249,
|
| 568 |
+
"grad_norm": 0.13923943042755127,
|
| 569 |
+
"learning_rate": 0.00033626101125499555,
|
| 570 |
+
"loss": 0.2021,
|
| 571 |
+
"step": 1575
|
| 572 |
+
},
|
| 573 |
+
{
|
| 574 |
+
"epoch": 1.3125512715340442,
|
| 575 |
+
"grad_norm": 0.16866903007030487,
|
| 576 |
+
"learning_rate": 0.0003194577111576333,
|
| 577 |
+
"loss": 0.1946,
|
| 578 |
+
"step": 1600
|
| 579 |
+
},
|
| 580 |
+
{
|
| 581 |
+
"epoch": 1.3125512715340442,
|
| 582 |
+
"eval_loss": 0.21704842150211334,
|
| 583 |
+
"eval_runtime": 19.3264,
|
| 584 |
+
"eval_samples_per_second": 51.743,
|
| 585 |
+
"eval_steps_per_second": 0.828,
|
| 586 |
+
"step": 1600
|
| 587 |
+
},
|
| 588 |
+
{
|
| 589 |
+
"epoch": 1.3330598851517639,
|
| 590 |
+
"grad_norm": 0.1447484791278839,
|
| 591 |
+
"learning_rate": 0.00030288574485502756,
|
| 592 |
+
"loss": 0.198,
|
| 593 |
+
"step": 1625
|
| 594 |
+
},
|
| 595 |
+
{
|
| 596 |
+
"epoch": 1.3535684987694832,
|
| 597 |
+
"grad_norm": 0.16940434277057648,
|
| 598 |
+
"learning_rate": 0.00028656634646288565,
|
| 599 |
+
"loss": 0.1988,
|
| 600 |
+
"step": 1650
|
| 601 |
+
},
|
| 602 |
+
{
|
| 603 |
+
"epoch": 1.3740771123872026,
|
| 604 |
+
"grad_norm": 0.21641549468040466,
|
| 605 |
+
"learning_rate": 0.00027052042647471254,
|
| 606 |
+
"loss": 0.1964,
|
| 607 |
+
"step": 1675
|
| 608 |
+
},
|
| 609 |
+
{
|
| 610 |
+
"epoch": 1.3945857260049221,
|
| 611 |
+
"grad_norm": 0.15306439995765686,
|
| 612 |
+
"learning_rate": 0.0002547685449686206,
|
| 613 |
+
"loss": 0.1969,
|
| 614 |
+
"step": 1700
|
| 615 |
+
},
|
| 616 |
+
{
|
| 617 |
+
"epoch": 1.3945857260049221,
|
| 618 |
+
"eval_loss": 0.21418742835521698,
|
| 619 |
+
"eval_runtime": 19.3438,
|
| 620 |
+
"eval_samples_per_second": 51.696,
|
| 621 |
+
"eval_steps_per_second": 0.827,
|
| 622 |
+
"step": 1700
|
| 623 |
+
},
|
| 624 |
+
{
|
| 625 |
+
"epoch": 1.4150943396226414,
|
| 626 |
+
"grad_norm": 0.1541595160961151,
|
| 627 |
+
"learning_rate": 0.0002393308852631373,
|
| 628 |
+
"loss": 0.1931,
|
| 629 |
+
"step": 1725
|
| 630 |
+
},
|
| 631 |
+
{
|
| 632 |
+
"epoch": 1.435602953240361,
|
| 633 |
+
"grad_norm": 0.16876709461212158,
|
| 634 |
+
"learning_rate": 0.0002242272280557645,
|
| 635 |
+
"loss": 0.1917,
|
| 636 |
+
"step": 1750
|
| 637 |
+
},
|
| 638 |
+
{
|
| 639 |
+
"epoch": 1.4561115668580804,
|
| 640 |
+
"grad_norm": 0.16322433948516846,
|
| 641 |
+
"learning_rate": 0.0002094769260774262,
|
| 642 |
+
"loss": 0.1992,
|
| 643 |
+
"step": 1775
|
| 644 |
+
},
|
| 645 |
+
{
|
| 646 |
+
"epoch": 1.4766201804758,
|
| 647 |
+
"grad_norm": 0.17060863971710205,
|
| 648 |
+
"learning_rate": 0.00019509887929528458,
|
| 649 |
+
"loss": 0.1887,
|
| 650 |
+
"step": 1800
|
| 651 |
+
},
|
| 652 |
+
{
|
| 653 |
+
"epoch": 1.4766201804758,
|
| 654 |
+
"eval_loss": 0.21251504123210907,
|
| 655 |
+
"eval_runtime": 19.3507,
|
| 656 |
+
"eval_samples_per_second": 51.678,
|
| 657 |
+
"eval_steps_per_second": 0.827,
|
| 658 |
+
"step": 1800
|
| 659 |
+
},
|
| 660 |
+
{
|
| 661 |
+
"epoch": 1.4971287940935194,
|
| 662 |
+
"grad_norm": 0.15198908746242523,
|
| 663 |
+
"learning_rate": 0.0001811115106956918,
|
| 664 |
+
"loss": 0.1883,
|
| 665 |
+
"step": 1825
|
| 666 |
+
},
|
| 667 |
+
{
|
| 668 |
+
"epoch": 1.5176374077112387,
|
| 669 |
+
"grad_norm": 0.1435108780860901,
|
| 670 |
+
"learning_rate": 0.00016753274267831115,
|
| 671 |
+
"loss": 0.1905,
|
| 672 |
+
"step": 1850
|
| 673 |
+
},
|
| 674 |
+
{
|
| 675 |
+
"epoch": 1.5381460213289582,
|
| 676 |
+
"grad_norm": 0.171073317527771,
|
| 677 |
+
"learning_rate": 0.00015437997409165478,
|
| 678 |
+
"loss": 0.1913,
|
| 679 |
+
"step": 1875
|
| 680 |
+
},
|
| 681 |
+
{
|
| 682 |
+
"epoch": 1.5586546349466777,
|
| 683 |
+
"grad_norm": 0.17346729338169098,
|
| 684 |
+
"learning_rate": 0.00014167005793946035,
|
| 685 |
+
"loss": 0.1938,
|
| 686 |
+
"step": 1900
|
| 687 |
+
},
|
| 688 |
+
{
|
| 689 |
+
"epoch": 1.5586546349466777,
|
| 690 |
+
"eval_loss": 0.2099105715751648,
|
| 691 |
+
"eval_runtime": 19.3226,
|
| 692 |
+
"eval_samples_per_second": 51.753,
|
| 693 |
+
"eval_steps_per_second": 0.828,
|
| 694 |
+
"step": 1900
|
| 695 |
+
},
|
| 696 |
+
{
|
| 697 |
+
"epoch": 1.579163248564397,
|
| 698 |
+
"grad_norm": 0.1647835671901703,
|
| 699 |
+
"learning_rate": 0.00012941927978647527,
|
| 700 |
+
"loss": 0.195,
|
| 701 |
+
"step": 1925
|
| 702 |
+
},
|
| 703 |
+
{
|
| 704 |
+
"epoch": 1.5996718621821167,
|
| 705 |
+
"grad_norm": 0.1640302836894989,
|
| 706 |
+
"learning_rate": 0.00011764333689131385,
|
| 707 |
+
"loss": 0.1941,
|
| 708 |
+
"step": 1950
|
| 709 |
+
},
|
| 710 |
+
{
|
| 711 |
+
"epoch": 1.620180475799836,
|
| 712 |
+
"grad_norm": 0.16301721334457397,
|
| 713 |
+
"learning_rate": 0.00010635731809312993,
|
| 714 |
+
"loss": 0.1892,
|
| 715 |
+
"step": 1975
|
| 716 |
+
},
|
| 717 |
+
{
|
| 718 |
+
"epoch": 1.6406890894175554,
|
| 719 |
+
"grad_norm": 0.15858031809329987,
|
| 720 |
+
"learning_rate": 9.557568447787201e-05,
|
| 721 |
+
"loss": 0.1948,
|
| 722 |
+
"step": 2000
|
| 723 |
+
},
|
| 724 |
+
{
|
| 725 |
+
"epoch": 1.6406890894175554,
|
| 726 |
+
"eval_loss": 0.2079882174730301,
|
| 727 |
+
"eval_runtime": 19.3579,
|
| 728 |
+
"eval_samples_per_second": 51.658,
|
| 729 |
+
"eval_steps_per_second": 0.827,
|
| 730 |
+
"step": 2000
|
| 731 |
+
},
|
| 732 |
+
{
|
| 733 |
+
"epoch": 1.661197703035275,
|
| 734 |
+
"grad_norm": 0.14629994332790375,
|
| 735 |
+
"learning_rate": 8.531225084889654e-05,
|
| 736 |
+
"loss": 0.1851,
|
| 737 |
+
"step": 2025
|
| 738 |
+
},
|
| 739 |
+
{
|
| 740 |
+
"epoch": 1.6817063166529942,
|
| 741 |
+
"grad_norm": 0.16934742033481598,
|
| 742 |
+
"learning_rate": 7.558016802568091e-05,
|
| 743 |
+
"loss": 0.1878,
|
| 744 |
+
"step": 2050
|
| 745 |
+
},
|
| 746 |
+
{
|
| 747 |
+
"epoch": 1.7022149302707137,
|
| 748 |
+
"grad_norm": 0.18174977600574493,
|
| 749 |
+
"learning_rate": 6.639190599331746e-05,
|
| 750 |
+
"loss": 0.1929,
|
| 751 |
+
"step": 2075
|
| 752 |
+
},
|
| 753 |
+
{
|
| 754 |
+
"epoch": 1.7227235438884332,
|
| 755 |
+
"grad_norm": 0.16311554610729218,
|
| 756 |
+
"learning_rate": 5.775923792437865e-05,
|
| 757 |
+
"loss": 0.1831,
|
| 758 |
+
"step": 2100
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 1.7227235438884332,
|
| 762 |
+
"eval_loss": 0.20694270730018616,
|
| 763 |
+
"eval_runtime": 19.3451,
|
| 764 |
+
"eval_samples_per_second": 51.693,
|
| 765 |
+
"eval_steps_per_second": 0.827,
|
| 766 |
+
"step": 2100
|
| 767 |
+
},
|
| 768 |
+
{
|
| 769 |
+
"epoch": 1.7432321575061525,
|
| 770 |
+
"grad_norm": 0.1429784595966339,
|
| 771 |
+
"learning_rate": 4.9693225093627616e-05,
|
| 772 |
+
"loss": 0.1865,
|
| 773 |
+
"step": 2125
|
| 774 |
+
},
|
| 775 |
+
{
|
| 776 |
+
"epoch": 1.7637407711238722,
|
| 777 |
+
"grad_norm": 0.1375550776720047,
|
| 778 |
+
"learning_rate": 4.220420270490294e-05,
|
| 779 |
+
"loss": 0.1852,
|
| 780 |
+
"step": 2150
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 1.7842493847415914,
|
| 784 |
+
"grad_norm": 0.14471116662025452,
|
| 785 |
+
"learning_rate": 3.530176664833834e-05,
|
| 786 |
+
"loss": 0.1943,
|
| 787 |
+
"step": 2175
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 1.804757998359311,
|
| 791 |
+
"grad_norm": 0.1561882048845291,
|
| 792 |
+
"learning_rate": 2.8994761204884756e-05,
|
| 793 |
+
"loss": 0.1873,
|
| 794 |
+
"step": 2200
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 1.804757998359311,
|
| 798 |
+
"eval_loss": 0.20590035617351532,
|
| 799 |
+
"eval_runtime": 19.3627,
|
| 800 |
+
"eval_samples_per_second": 51.646,
|
| 801 |
+
"eval_steps_per_second": 0.826,
|
| 802 |
+
"step": 2200
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 1.8252666119770304,
|
| 806 |
+
"grad_norm": 0.16283340752124786,
|
| 807 |
+
"learning_rate": 2.329126771388995e-05,
|
| 808 |
+
"loss": 0.1881,
|
| 809 |
+
"step": 2225
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 1.8457752255947497,
|
| 813 |
+
"grad_norm": 0.15389613807201385,
|
| 814 |
+
"learning_rate": 1.8198594218256815e-05,
|
| 815 |
+
"loss": 0.1903,
|
| 816 |
+
"step": 2250
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"epoch": 1.8662838392124692,
|
| 820 |
+
"grad_norm": 0.15144121646881104,
|
| 821 |
+
"learning_rate": 1.3723266100447052e-05,
|
| 822 |
+
"loss": 0.1943,
|
| 823 |
+
"step": 2275
|
| 824 |
+
},
|
| 825 |
+
{
|
| 826 |
+
"epoch": 1.8867924528301887,
|
| 827 |
+
"grad_norm": 0.1426999270915985,
|
| 828 |
+
"learning_rate": 9.871017721329201e-06,
|
| 829 |
+
"loss": 0.1868,
|
| 830 |
+
"step": 2300
|
| 831 |
+
},
|
| 832 |
+
{
|
| 833 |
+
"epoch": 1.8867924528301887,
|
| 834 |
+
"eval_loss": 0.20548181235790253,
|
| 835 |
+
"eval_runtime": 19.3456,
|
| 836 |
+
"eval_samples_per_second": 51.691,
|
| 837 |
+
"eval_steps_per_second": 0.827,
|
| 838 |
+
"step": 2300
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 1.907301066447908,
|
| 842 |
+
"grad_norm": 0.18528231978416443,
|
| 843 |
+
"learning_rate": 6.646785072584871e-06,
|
| 844 |
+
"loss": 0.1949,
|
| 845 |
+
"step": 2325
|
| 846 |
+
},
|
| 847 |
+
{
|
| 848 |
+
"epoch": 1.9278096800656277,
|
| 849 |
+
"grad_norm": 0.14917585253715515,
|
| 850 |
+
"learning_rate": 4.054699452086641e-06,
|
| 851 |
+
"loss": 0.1795,
|
| 852 |
+
"step": 2350
|
| 853 |
+
},
|
| 854 |
+
{
|
| 855 |
+
"epoch": 1.948318293683347,
|
| 856 |
+
"grad_norm": 0.13997340202331543,
|
| 857 |
+
"learning_rate": 2.0980821703527886e-06,
|
| 858 |
+
"loss": 0.1855,
|
| 859 |
+
"step": 2375
|
| 860 |
+
},
|
| 861 |
+
{
|
| 862 |
+
"epoch": 1.9688269073010665,
|
| 863 |
+
"grad_norm": 0.145187109708786,
|
| 864 |
+
"learning_rate": 7.794402948607671e-07,
|
| 865 |
+
"loss": 0.187,
|
| 866 |
+
"step": 2400
|
| 867 |
+
},
|
| 868 |
+
{
|
| 869 |
+
"epoch": 1.9688269073010665,
|
| 870 |
+
"eval_loss": 0.2054169923067093,
|
| 871 |
+
"eval_runtime": 19.3538,
|
| 872 |
+
"eval_samples_per_second": 51.67,
|
| 873 |
+
"eval_steps_per_second": 0.827,
|
| 874 |
+
"step": 2400
|
| 875 |
+
},
|
| 876 |
+
{
|
| 877 |
+
"epoch": 1.989335520918786,
|
| 878 |
+
"grad_norm": 0.1513894647359848,
|
| 879 |
+
"learning_rate": 1.0046343767294853e-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": 0.2420808641122344,
|
| 888 |
+
"train_runtime": 3047.0565,
|
| 889 |
+
"train_samples_per_second": 25.598,
|
| 890 |
+
"train_steps_per_second": 0.8
|
| 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 |
+
}
|