Batch upload part 9
Browse files- nl_tasks/exps/run_ex34/ft/adapter_config.json +18 -0
- nl_tasks/exps/run_ex34/ft/special_tokens_map.json +24 -0
- nl_tasks/exps/run_ex34/ft/tokenizer.json +0 -0
- nl_tasks/exps/run_ex34/ft/tokenizer.model +3 -0
- nl_tasks/exps/run_ex34/ft/tokenizer_config.json +43 -0
- nl_tasks/exps/run_ex34/ft2/adapter_config.json +18 -0
- nl_tasks/exps/run_ex34/ft2/adapter_model.bin +3 -0
- nl_tasks/run_all/exnr14/ft/adapter_config.json +18 -0
- nl_tasks/run_all/exnr14/ft/added_tokens.json +3 -0
- nl_tasks/run_all/exnr14/ft/special_tokens_map.json +30 -0
- nl_tasks/run_all/exnr14/ft/tokenizer.json +0 -0
- nl_tasks/run_all/exnr14/ft/tokenizer.model +3 -0
- nl_tasks/run_all/exnr14/ft/tokenizer_config.json +51 -0
- nl_tasks/run_all/exnr14/ft/training_args.bin +3 -0
- nl_tasks/run_all/exnr14/ft2/adapter_config.json +18 -0
- nl_tasks/run_all/exnr14/ft2/adapter_model.bin +3 -0
- nl_tasks/run_all/exnr14/trainer_state.json +1106 -0
- nl_tasks/run_all/exnr15/ft/adapter_config.json +18 -0
- nl_tasks/run_all/exnr15/ft/added_tokens.json +3 -0
- nl_tasks/run_all/exnr15/ft/special_tokens_map.json +30 -0
- nl_tasks/run_all/exnr15/ft/tokenizer.json +0 -0
- nl_tasks/run_all/exnr15/ft/tokenizer.model +3 -0
- nl_tasks/run_all/exnr15/ft/tokenizer_config.json +51 -0
- nl_tasks/run_all/exnr15/ft/training_args.bin +3 -0
- nl_tasks/run_all/exnr15/ft2/adapter_config.json +18 -0
- nl_tasks/run_all/exnr15/ft2/adapter_model.bin +3 -0
- nl_tasks/run_all/exnr15/trainer_state.json +135 -0
- nl_tasks/run_all/run_exnr10/ft/adapter_config.json +18 -0
- nl_tasks/run_all/run_exnr10/ft/added_tokens.json +3 -0
- nl_tasks/run_all/run_exnr10/ft/special_tokens_map.json +30 -0
- nl_tasks/run_all/run_exnr10/ft/tokenizer.json +0 -0
- nl_tasks/run_all/run_exnr10/ft/tokenizer.model +3 -0
- nl_tasks/run_all/run_exnr10/ft/tokenizer_config.json +51 -0
- nl_tasks/run_all/run_exnr10/ft/training_args.bin +3 -0
- nl_tasks/run_all/run_exnr10/ft2/adapter_config.json +18 -0
- nl_tasks/run_all/run_exnr10/ft2/adapter_model.bin +3 -0
- nl_tasks/run_all/run_exnr10/trainer_state.json +1106 -0
- nl_tasks/run_all/run_exnr11/ft/adapter_config.json +18 -0
- nl_tasks/run_all/run_exnr11/ft/added_tokens.json +3 -0
- nl_tasks/run_all/run_exnr11/ft/special_tokens_map.json +30 -0
- nl_tasks/run_all/run_exnr11/ft/tokenizer.json +0 -0
- nl_tasks/run_all/run_exnr11/ft/tokenizer.model +3 -0
- nl_tasks/run_all/run_exnr11/ft/tokenizer_config.json +51 -0
- nl_tasks/run_all/run_exnr11/ft/training_args.bin +3 -0
- nl_tasks/run_all/run_exnr11/ft2/adapter_config.json +18 -0
- nl_tasks/run_all/run_exnr11/ft2/adapter_model.bin +3 -0
- nl_tasks/run_all/run_exnr11/trainer_state.json +1106 -0
- nl_tasks/run_all/run_exnr12/ft/special_tokens_map.json +30 -0
- nl_tasks/run_all/run_exnr12/ft/tokenizer_config.json +51 -0
- nl_tasks/run_all/run_exnr12/trainer_state.json +1106 -0
nl_tasks/exps/run_ex34/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_ex34/ft/special_tokens_map.json
ADDED
|
@@ -0,0 +1,24 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"bos_token": {
|
| 3 |
+
"content": "<s>",
|
| 4 |
+
"lstrip": false,
|
| 5 |
+
"normalized": false,
|
| 6 |
+
"rstrip": false,
|
| 7 |
+
"single_word": false
|
| 8 |
+
},
|
| 9 |
+
"eos_token": {
|
| 10 |
+
"content": "</s>",
|
| 11 |
+
"lstrip": false,
|
| 12 |
+
"normalized": false,
|
| 13 |
+
"rstrip": false,
|
| 14 |
+
"single_word": false
|
| 15 |
+
},
|
| 16 |
+
"pad_token": "<unk>",
|
| 17 |
+
"unk_token": {
|
| 18 |
+
"content": "<unk>",
|
| 19 |
+
"lstrip": false,
|
| 20 |
+
"normalized": false,
|
| 21 |
+
"rstrip": false,
|
| 22 |
+
"single_word": false
|
| 23 |
+
}
|
| 24 |
+
}
|
nl_tasks/exps/run_ex34/ft/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
nl_tasks/exps/run_ex34/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_ex34/ft/tokenizer_config.json
ADDED
|
@@ -0,0 +1,43 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"add_bos_token": true,
|
| 3 |
+
"add_eos_token": false,
|
| 4 |
+
"add_prefix_space": null,
|
| 5 |
+
"added_tokens_decoder": {
|
| 6 |
+
"0": {
|
| 7 |
+
"content": "<unk>",
|
| 8 |
+
"lstrip": false,
|
| 9 |
+
"normalized": false,
|
| 10 |
+
"rstrip": false,
|
| 11 |
+
"single_word": false,
|
| 12 |
+
"special": true
|
| 13 |
+
},
|
| 14 |
+
"1": {
|
| 15 |
+
"content": "<s>",
|
| 16 |
+
"lstrip": false,
|
| 17 |
+
"normalized": false,
|
| 18 |
+
"rstrip": false,
|
| 19 |
+
"single_word": false,
|
| 20 |
+
"special": true
|
| 21 |
+
},
|
| 22 |
+
"2": {
|
| 23 |
+
"content": "</s>",
|
| 24 |
+
"lstrip": false,
|
| 25 |
+
"normalized": false,
|
| 26 |
+
"rstrip": false,
|
| 27 |
+
"single_word": false,
|
| 28 |
+
"special": true
|
| 29 |
+
}
|
| 30 |
+
},
|
| 31 |
+
"bos_token": "<s>",
|
| 32 |
+
"clean_up_tokenization_spaces": false,
|
| 33 |
+
"eos_token": "</s>",
|
| 34 |
+
"extra_special_tokens": {},
|
| 35 |
+
"legacy": false,
|
| 36 |
+
"model_max_length": 512,
|
| 37 |
+
"pad_token": "<unk>",
|
| 38 |
+
"padding_side": "right",
|
| 39 |
+
"sp_model_kwargs": {},
|
| 40 |
+
"tokenizer_class": "LlamaTokenizer",
|
| 41 |
+
"unk_token": "<unk>",
|
| 42 |
+
"use_default_system_prompt": false
|
| 43 |
+
}
|
nl_tasks/exps/run_ex34/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_ex34/ft2/adapter_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:0d263a69315b36f2b766405381d1a2c89d10e2377e9d5f50e9ff5f74bfa0c189
|
| 3 |
+
size 33602915
|
nl_tasks/run_all/exnr14/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/run_all/exnr14/ft/added_tokens.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"[PAD]": 32000
|
| 3 |
+
}
|
nl_tasks/run_all/exnr14/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/run_all/exnr14/ft/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
nl_tasks/run_all/exnr14/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/run_all/exnr14/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/run_all/exnr14/ft/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:ba427f2d53ed75e43f98a703940afe9017468efadf9a60725d88bde4c1c2c303
|
| 3 |
+
size 6545
|
nl_tasks/run_all/exnr14/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/run_all/exnr14/ft2/adapter_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:29b2d1c56e46e1472d61486ef0f7c4570c1678d2f3ea6b0c9652a5e9a979117e
|
| 3 |
+
size 33602915
|
nl_tasks/run_all/exnr14/trainer_state.json
ADDED
|
@@ -0,0 +1,1106 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 2.0,
|
| 6 |
+
"eval_steps": 50,
|
| 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": 0.16105684638023376,
|
| 15 |
+
"learning_rate": 0.00019672131147540983,
|
| 16 |
+
"loss": 0.6249,
|
| 17 |
+
"step": 25
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.04101722723543889,
|
| 21 |
+
"grad_norm": 0.19086705148220062,
|
| 22 |
+
"learning_rate": 0.00040163934426229507,
|
| 23 |
+
"loss": 0.3842,
|
| 24 |
+
"step": 50
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.04101722723543889,
|
| 28 |
+
"eval_loss": 0.3523162603378296,
|
| 29 |
+
"eval_runtime": 19.5722,
|
| 30 |
+
"eval_samples_per_second": 51.093,
|
| 31 |
+
"eval_steps_per_second": 0.817,
|
| 32 |
+
"step": 50
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.06152584085315833,
|
| 36 |
+
"grad_norm": 0.40827786922454834,
|
| 37 |
+
"learning_rate": 0.0006065573770491804,
|
| 38 |
+
"loss": 0.3562,
|
| 39 |
+
"step": 75
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.08203445447087777,
|
| 43 |
+
"grad_norm": 0.2028273344039917,
|
| 44 |
+
"learning_rate": 0.0008114754098360656,
|
| 45 |
+
"loss": 0.3428,
|
| 46 |
+
"step": 100
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"epoch": 0.08203445447087777,
|
| 50 |
+
"eval_loss": 0.32449883222579956,
|
| 51 |
+
"eval_runtime": 19.5267,
|
| 52 |
+
"eval_samples_per_second": 51.212,
|
| 53 |
+
"eval_steps_per_second": 0.819,
|
| 54 |
+
"step": 100
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.10254306808859721,
|
| 58 |
+
"grad_norm": 0.2950042188167572,
|
| 59 |
+
"learning_rate": 0.0010163934426229509,
|
| 60 |
+
"loss": 0.3332,
|
| 61 |
+
"step": 125
|
| 62 |
+
},
|
| 63 |
+
{
|
| 64 |
+
"epoch": 0.12305168170631665,
|
| 65 |
+
"grad_norm": 2.916337013244629,
|
| 66 |
+
"learning_rate": 0.001221311475409836,
|
| 67 |
+
"loss": 0.5147,
|
| 68 |
+
"step": 150
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.12305168170631665,
|
| 72 |
+
"eval_loss": 0.410254567861557,
|
| 73 |
+
"eval_runtime": 19.5194,
|
| 74 |
+
"eval_samples_per_second": 51.231,
|
| 75 |
+
"eval_steps_per_second": 0.82,
|
| 76 |
+
"step": 150
|
| 77 |
+
},
|
| 78 |
+
{
|
| 79 |
+
"epoch": 0.1435602953240361,
|
| 80 |
+
"grad_norm": 0.40234318375587463,
|
| 81 |
+
"learning_rate": 0.0014262295081967215,
|
| 82 |
+
"loss": 0.3616,
|
| 83 |
+
"step": 175
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"epoch": 0.16406890894175555,
|
| 87 |
+
"grad_norm": 0.5344768762588501,
|
| 88 |
+
"learning_rate": 0.0016311475409836065,
|
| 89 |
+
"loss": 0.3605,
|
| 90 |
+
"step": 200
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 0.16406890894175555,
|
| 94 |
+
"eval_loss": 0.3446679413318634,
|
| 95 |
+
"eval_runtime": 19.5174,
|
| 96 |
+
"eval_samples_per_second": 51.236,
|
| 97 |
+
"eval_steps_per_second": 0.82,
|
| 98 |
+
"step": 200
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.184577522559475,
|
| 102 |
+
"grad_norm": 392.2628173828125,
|
| 103 |
+
"learning_rate": 0.0018360655737704918,
|
| 104 |
+
"loss": 0.4885,
|
| 105 |
+
"step": 225
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 0.20508613617719443,
|
| 109 |
+
"grad_norm": 0.4538051187992096,
|
| 110 |
+
"learning_rate": 0.0019999743708232127,
|
| 111 |
+
"loss": 0.4975,
|
| 112 |
+
"step": 250
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 0.20508613617719443,
|
| 116 |
+
"eval_loss": 0.3520921468734741,
|
| 117 |
+
"eval_runtime": 19.5139,
|
| 118 |
+
"eval_samples_per_second": 51.245,
|
| 119 |
+
"eval_steps_per_second": 0.82,
|
| 120 |
+
"step": 250
|
| 121 |
+
},
|
| 122 |
+
{
|
| 123 |
+
"epoch": 0.22559474979491387,
|
| 124 |
+
"grad_norm": 0.6222809553146362,
|
| 125 |
+
"learning_rate": 0.0019990774875676054,
|
| 126 |
+
"loss": 0.3541,
|
| 127 |
+
"step": 275
|
| 128 |
+
},
|
| 129 |
+
{
|
| 130 |
+
"epoch": 0.2461033634126333,
|
| 131 |
+
"grad_norm": 0.44931286573410034,
|
| 132 |
+
"learning_rate": 0.001996900458879386,
|
| 133 |
+
"loss": 0.3521,
|
| 134 |
+
"step": 300
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.2461033634126333,
|
| 138 |
+
"eval_loss": 0.3311863839626312,
|
| 139 |
+
"eval_runtime": 19.5345,
|
| 140 |
+
"eval_samples_per_second": 51.192,
|
| 141 |
+
"eval_steps_per_second": 0.819,
|
| 142 |
+
"step": 300
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.2666119770303528,
|
| 146 |
+
"grad_norm": 0.23350577056407928,
|
| 147 |
+
"learning_rate": 0.001993446074245224,
|
| 148 |
+
"loss": 0.3325,
|
| 149 |
+
"step": 325
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.2871205906480722,
|
| 153 |
+
"grad_norm": 0.21402505040168762,
|
| 154 |
+
"learning_rate": 0.0019887187598630527,
|
| 155 |
+
"loss": 0.3221,
|
| 156 |
+
"step": 350
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.2871205906480722,
|
| 160 |
+
"eval_loss": 0.2994612455368042,
|
| 161 |
+
"eval_runtime": 19.5154,
|
| 162 |
+
"eval_samples_per_second": 51.242,
|
| 163 |
+
"eval_steps_per_second": 0.82,
|
| 164 |
+
"step": 350
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.30762920426579166,
|
| 168 |
+
"grad_norm": 0.17019234597682953,
|
| 169 |
+
"learning_rate": 0.0019827245729706648,
|
| 170 |
+
"loss": 0.3031,
|
| 171 |
+
"step": 375
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.3281378178835111,
|
| 175 |
+
"grad_norm": 0.17441657185554504,
|
| 176 |
+
"learning_rate": 0.0019754711940844047,
|
| 177 |
+
"loss": 0.3047,
|
| 178 |
+
"step": 400
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.3281378178835111,
|
| 182 |
+
"eval_loss": 0.28233733773231506,
|
| 183 |
+
"eval_runtime": 19.5238,
|
| 184 |
+
"eval_samples_per_second": 51.22,
|
| 185 |
+
"eval_steps_per_second": 0.82,
|
| 186 |
+
"step": 400
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 0.34864643150123054,
|
| 190 |
+
"grad_norm": 0.12981465458869934,
|
| 191 |
+
"learning_rate": 0.0019669679171579117,
|
| 192 |
+
"loss": 0.2964,
|
| 193 |
+
"step": 425
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 0.36915504511895,
|
| 197 |
+
"grad_norm": 0.14443928003311157,
|
| 198 |
+
"learning_rate": 0.001957225637673524,
|
| 199 |
+
"loss": 0.2879,
|
| 200 |
+
"step": 450
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 0.36915504511895,
|
| 204 |
+
"eval_loss": 0.27039000391960144,
|
| 205 |
+
"eval_runtime": 19.544,
|
| 206 |
+
"eval_samples_per_second": 51.167,
|
| 207 |
+
"eval_steps_per_second": 0.819,
|
| 208 |
+
"step": 450
|
| 209 |
+
},
|
| 210 |
+
{
|
| 211 |
+
"epoch": 0.3896636587366694,
|
| 212 |
+
"grad_norm": 0.1316242665052414,
|
| 213 |
+
"learning_rate": 0.0019462568386815961,
|
| 214 |
+
"loss": 0.2839,
|
| 215 |
+
"step": 475
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 0.41017227235438886,
|
| 219 |
+
"grad_norm": 0.12701553106307983,
|
| 220 |
+
"learning_rate": 0.0019340755748056234,
|
| 221 |
+
"loss": 0.2759,
|
| 222 |
+
"step": 500
|
| 223 |
+
},
|
| 224 |
+
{
|
| 225 |
+
"epoch": 0.41017227235438886,
|
| 226 |
+
"eval_loss": 0.2636236250400543,
|
| 227 |
+
"eval_runtime": 19.5187,
|
| 228 |
+
"eval_samples_per_second": 51.233,
|
| 229 |
+
"eval_steps_per_second": 0.82,
|
| 230 |
+
"step": 500
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 0.4306808859721083,
|
| 234 |
+
"grad_norm": 0.1544770896434784,
|
| 235 |
+
"learning_rate": 0.0019206974542336672,
|
| 236 |
+
"loss": 0.2795,
|
| 237 |
+
"step": 525
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 0.45118949958982774,
|
| 241 |
+
"grad_norm": 0.11463268101215363,
|
| 242 |
+
"learning_rate": 0.0019061396187191563,
|
| 243 |
+
"loss": 0.2735,
|
| 244 |
+
"step": 550
|
| 245 |
+
},
|
| 246 |
+
{
|
| 247 |
+
"epoch": 0.45118949958982774,
|
| 248 |
+
"eval_loss": 0.2576003074645996,
|
| 249 |
+
"eval_runtime": 19.5029,
|
| 250 |
+
"eval_samples_per_second": 51.274,
|
| 251 |
+
"eval_steps_per_second": 0.82,
|
| 252 |
+
"step": 550
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 0.4716981132075472,
|
| 256 |
+
"grad_norm": 0.12292572110891342,
|
| 257 |
+
"learning_rate": 0.0018904207216166836,
|
| 258 |
+
"loss": 0.2611,
|
| 259 |
+
"step": 575
|
| 260 |
+
},
|
| 261 |
+
{
|
| 262 |
+
"epoch": 0.4922067268252666,
|
| 263 |
+
"grad_norm": 0.11286451667547226,
|
| 264 |
+
"learning_rate": 0.001873560903980955,
|
| 265 |
+
"loss": 0.2699,
|
| 266 |
+
"step": 600
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 0.4922067268252666,
|
| 270 |
+
"eval_loss": 0.2554730474948883,
|
| 271 |
+
"eval_runtime": 19.529,
|
| 272 |
+
"eval_samples_per_second": 51.206,
|
| 273 |
+
"eval_steps_per_second": 0.819,
|
| 274 |
+
"step": 600
|
| 275 |
+
},
|
| 276 |
+
{
|
| 277 |
+
"epoch": 0.5127153404429861,
|
| 278 |
+
"grad_norm": 0.09815208613872528,
|
| 279 |
+
"learning_rate": 0.0018555817687594984,
|
| 280 |
+
"loss": 0.2573,
|
| 281 |
+
"step": 625
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 0.5332239540607056,
|
| 285 |
+
"grad_norm": 0.10865656286478043,
|
| 286 |
+
"learning_rate": 0.0018365063531122169,
|
| 287 |
+
"loss": 0.2564,
|
| 288 |
+
"step": 650
|
| 289 |
+
},
|
| 290 |
+
{
|
| 291 |
+
"epoch": 0.5332239540607056,
|
| 292 |
+
"eval_loss": 0.25100600719451904,
|
| 293 |
+
"eval_runtime": 19.5294,
|
| 294 |
+
"eval_samples_per_second": 51.205,
|
| 295 |
+
"eval_steps_per_second": 0.819,
|
| 296 |
+
"step": 650
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.5537325676784249,
|
| 300 |
+
"grad_norm": 0.11892469227313995,
|
| 301 |
+
"learning_rate": 0.0018163590988932402,
|
| 302 |
+
"loss": 0.2556,
|
| 303 |
+
"step": 675
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.5742411812961444,
|
| 307 |
+
"grad_norm": 0.08931335061788559,
|
| 308 |
+
"learning_rate": 0.0017951658213329078,
|
| 309 |
+
"loss": 0.2507,
|
| 310 |
+
"step": 700
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.5742411812961444,
|
| 314 |
+
"eval_loss": 0.2487325370311737,
|
| 315 |
+
"eval_runtime": 19.5503,
|
| 316 |
+
"eval_samples_per_second": 51.15,
|
| 317 |
+
"eval_steps_per_second": 0.818,
|
| 318 |
+
"step": 700
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.5947497949138638,
|
| 322 |
+
"grad_norm": 0.09335417300462723,
|
| 323 |
+
"learning_rate": 0.0017729536759600033,
|
| 324 |
+
"loss": 0.2479,
|
| 325 |
+
"step": 725
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.6152584085315833,
|
| 329 |
+
"grad_norm": 0.08527883887290955,
|
| 330 |
+
"learning_rate": 0.0017497511238066307,
|
| 331 |
+
"loss": 0.2571,
|
| 332 |
+
"step": 750
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.6152584085315833,
|
| 336 |
+
"eval_loss": 0.24877095222473145,
|
| 337 |
+
"eval_runtime": 19.5122,
|
| 338 |
+
"eval_samples_per_second": 51.25,
|
| 339 |
+
"eval_steps_per_second": 0.82,
|
| 340 |
+
"step": 750
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 0.6357670221493027,
|
| 344 |
+
"grad_norm": 0.10340578854084015,
|
| 345 |
+
"learning_rate": 0.00172558789494031,
|
| 346 |
+
"loss": 0.2621,
|
| 347 |
+
"step": 775
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 0.6562756357670222,
|
| 351 |
+
"grad_norm": 0.10272800922393799,
|
| 352 |
+
"learning_rate": 0.0017004949503700284,
|
| 353 |
+
"loss": 0.2519,
|
| 354 |
+
"step": 800
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.6562756357670222,
|
| 358 |
+
"eval_loss": 0.24430014193058014,
|
| 359 |
+
"eval_runtime": 19.5231,
|
| 360 |
+
"eval_samples_per_second": 51.221,
|
| 361 |
+
"eval_steps_per_second": 0.82,
|
| 362 |
+
"step": 800
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.6767842493847416,
|
| 366 |
+
"grad_norm": 0.11489123106002808,
|
| 367 |
+
"learning_rate": 0.0016745044423750449,
|
| 368 |
+
"loss": 0.2472,
|
| 369 |
+
"step": 825
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.6972928630024611,
|
| 373 |
+
"grad_norm": 0.07966622710227966,
|
| 374 |
+
"learning_rate": 0.0016476496733072946,
|
| 375 |
+
"loss": 0.261,
|
| 376 |
+
"step": 850
|
| 377 |
+
},
|
| 378 |
+
{
|
| 379 |
+
"epoch": 0.6972928630024611,
|
| 380 |
+
"eval_loss": 0.24359455704689026,
|
| 381 |
+
"eval_runtime": 19.5061,
|
| 382 |
+
"eval_samples_per_second": 51.266,
|
| 383 |
+
"eval_steps_per_second": 0.82,
|
| 384 |
+
"step": 850
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.7178014766201805,
|
| 388 |
+
"grad_norm": 0.07211313396692276,
|
| 389 |
+
"learning_rate": 0.0016199650529201684,
|
| 390 |
+
"loss": 0.2548,
|
| 391 |
+
"step": 875
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 0.7383100902379,
|
| 395 |
+
"grad_norm": 0.08206778764724731,
|
| 396 |
+
"learning_rate": 0.0015914860542783522,
|
| 397 |
+
"loss": 0.2479,
|
| 398 |
+
"step": 900
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 0.7383100902379,
|
| 402 |
+
"eval_loss": 0.24049904942512512,
|
| 403 |
+
"eval_runtime": 19.5237,
|
| 404 |
+
"eval_samples_per_second": 51.22,
|
| 405 |
+
"eval_steps_per_second": 0.82,
|
| 406 |
+
"step": 900
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 0.7588187038556193,
|
| 410 |
+
"grad_norm": 0.08992116898298264,
|
| 411 |
+
"learning_rate": 0.0015622491683052124,
|
| 412 |
+
"loss": 0.2502,
|
| 413 |
+
"step": 925
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 0.7793273174733388,
|
| 417 |
+
"grad_norm": 0.08781281113624573,
|
| 418 |
+
"learning_rate": 0.0015322918570259759,
|
| 419 |
+
"loss": 0.2341,
|
| 420 |
+
"step": 950
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 0.7793273174733388,
|
| 424 |
+
"eval_loss": 0.23792409896850586,
|
| 425 |
+
"eval_runtime": 19.5338,
|
| 426 |
+
"eval_samples_per_second": 51.193,
|
| 427 |
+
"eval_steps_per_second": 0.819,
|
| 428 |
+
"step": 950
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 0.7998359310910582,
|
| 432 |
+
"grad_norm": 0.08975056558847427,
|
| 433 |
+
"learning_rate": 0.0015016525055666057,
|
| 434 |
+
"loss": 0.2471,
|
| 435 |
+
"step": 975
|
| 436 |
+
},
|
| 437 |
+
{
|
| 438 |
+
"epoch": 0.8203445447087777,
|
| 439 |
+
"grad_norm": 0.08536435663700104,
|
| 440 |
+
"learning_rate": 0.001470370372969886,
|
| 441 |
+
"loss": 0.2501,
|
| 442 |
+
"step": 1000
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 0.8203445447087777,
|
| 446 |
+
"eval_loss": 0.23709805309772491,
|
| 447 |
+
"eval_runtime": 19.5182,
|
| 448 |
+
"eval_samples_per_second": 51.234,
|
| 449 |
+
"eval_steps_per_second": 0.82,
|
| 450 |
+
"step": 1000
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.8408531583264971,
|
| 454 |
+
"grad_norm": 0.07575884461402893,
|
| 455 |
+
"learning_rate": 0.0014384855418917311,
|
| 456 |
+
"loss": 0.2366,
|
| 457 |
+
"step": 1025
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 0.8613617719442166,
|
| 461 |
+
"grad_norm": 0.08482314646244049,
|
| 462 |
+
"learning_rate": 0.0014060388672421775,
|
| 463 |
+
"loss": 0.2432,
|
| 464 |
+
"step": 1050
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.8613617719442166,
|
| 468 |
+
"eval_loss": 0.23468729853630066,
|
| 469 |
+
"eval_runtime": 19.4897,
|
| 470 |
+
"eval_samples_per_second": 51.309,
|
| 471 |
+
"eval_steps_per_second": 0.821,
|
| 472 |
+
"step": 1050
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 0.881870385561936,
|
| 476 |
+
"grad_norm": 0.08808961510658264,
|
| 477 |
+
"learning_rate": 0.0013730719238368662,
|
| 478 |
+
"loss": 0.248,
|
| 479 |
+
"step": 1075
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 0.9023789991796555,
|
| 483 |
+
"grad_norm": 0.0849333256483078,
|
| 484 |
+
"learning_rate": 0.0013396269531260867,
|
| 485 |
+
"loss": 0.24,
|
| 486 |
+
"step": 1100
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 0.9023789991796555,
|
| 490 |
+
"eval_loss": 0.23305058479309082,
|
| 491 |
+
"eval_runtime": 19.5254,
|
| 492 |
+
"eval_samples_per_second": 51.215,
|
| 493 |
+
"eval_steps_per_second": 0.819,
|
| 494 |
+
"step": 1100
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 0.9228876127973749,
|
| 498 |
+
"grad_norm": 0.13481223583221436,
|
| 499 |
+
"learning_rate": 0.0013057468090696496,
|
| 500 |
+
"loss": 0.2376,
|
| 501 |
+
"step": 1125
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 0.9433962264150944,
|
| 505 |
+
"grad_norm": 0.07862788438796997,
|
| 506 |
+
"learning_rate": 0.0012714749032269287,
|
| 507 |
+
"loss": 0.2511,
|
| 508 |
+
"step": 1150
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 0.9433962264150944,
|
| 512 |
+
"eval_loss": 0.23162627220153809,
|
| 513 |
+
"eval_runtime": 19.5241,
|
| 514 |
+
"eval_samples_per_second": 51.219,
|
| 515 |
+
"eval_steps_per_second": 0.819,
|
| 516 |
+
"step": 1150
|
| 517 |
+
},
|
| 518 |
+
{
|
| 519 |
+
"epoch": 0.9639048400328137,
|
| 520 |
+
"grad_norm": 0.08472246676683426,
|
| 521 |
+
"learning_rate": 0.0012368551491324358,
|
| 522 |
+
"loss": 0.2415,
|
| 523 |
+
"step": 1175
|
| 524 |
+
},
|
| 525 |
+
{
|
| 526 |
+
"epoch": 0.9844134536505332,
|
| 527 |
+
"grad_norm": 0.07140998542308807,
|
| 528 |
+
"learning_rate": 0.0012019319060282063,
|
| 529 |
+
"loss": 0.2432,
|
| 530 |
+
"step": 1200
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 0.9844134536505332,
|
| 534 |
+
"eval_loss": 0.22965233027935028,
|
| 535 |
+
"eval_runtime": 19.5217,
|
| 536 |
+
"eval_samples_per_second": 51.225,
|
| 537 |
+
"eval_steps_per_second": 0.82,
|
| 538 |
+
"step": 1200
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"epoch": 1.0049220672682526,
|
| 542 |
+
"grad_norm": 0.09055250138044357,
|
| 543 |
+
"learning_rate": 0.0011667499220250803,
|
| 544 |
+
"loss": 0.2404,
|
| 545 |
+
"step": 1225
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 1.0254306808859721,
|
| 549 |
+
"grad_norm": 0.08059660345315933,
|
| 550 |
+
"learning_rate": 0.0011313542767657204,
|
| 551 |
+
"loss": 0.2058,
|
| 552 |
+
"step": 1250
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 1.0254306808859721,
|
| 556 |
+
"eval_loss": 0.23091378808021545,
|
| 557 |
+
"eval_runtime": 19.5068,
|
| 558 |
+
"eval_samples_per_second": 51.264,
|
| 559 |
+
"eval_steps_per_second": 0.82,
|
| 560 |
+
"step": 1250
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 1.0459392945036916,
|
| 564 |
+
"grad_norm": 0.08628836274147034,
|
| 565 |
+
"learning_rate": 0.0010957903236628267,
|
| 566 |
+
"loss": 0.2037,
|
| 567 |
+
"step": 1275
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 1.066447908121411,
|
| 571 |
+
"grad_norm": 0.08117768913507462,
|
| 572 |
+
"learning_rate": 0.001060103631786563,
|
| 573 |
+
"loss": 0.2115,
|
| 574 |
+
"step": 1300
|
| 575 |
+
},
|
| 576 |
+
{
|
| 577 |
+
"epoch": 1.066447908121411,
|
| 578 |
+
"eval_loss": 0.22918300330638885,
|
| 579 |
+
"eval_runtime": 19.5375,
|
| 580 |
+
"eval_samples_per_second": 51.184,
|
| 581 |
+
"eval_steps_per_second": 0.819,
|
| 582 |
+
"step": 1300
|
| 583 |
+
},
|
| 584 |
+
{
|
| 585 |
+
"epoch": 1.0869565217391304,
|
| 586 |
+
"grad_norm": 0.0947440043091774,
|
| 587 |
+
"learning_rate": 0.0010243399274756564,
|
| 588 |
+
"loss": 0.2192,
|
| 589 |
+
"step": 1325
|
| 590 |
+
},
|
| 591 |
+
{
|
| 592 |
+
"epoch": 1.1074651353568499,
|
| 593 |
+
"grad_norm": 0.09542486071586609,
|
| 594 |
+
"learning_rate": 0.0009885450357469806,
|
| 595 |
+
"loss": 0.2086,
|
| 596 |
+
"step": 1350
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 1.1074651353568499,
|
| 600 |
+
"eval_loss": 0.2292700856924057,
|
| 601 |
+
"eval_runtime": 19.5283,
|
| 602 |
+
"eval_samples_per_second": 51.208,
|
| 603 |
+
"eval_steps_per_second": 0.819,
|
| 604 |
+
"step": 1350
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 1.1279737489745694,
|
| 608 |
+
"grad_norm": 0.0885721817612648,
|
| 609 |
+
"learning_rate": 0.0009527648215787065,
|
| 610 |
+
"loss": 0.2072,
|
| 611 |
+
"step": 1375
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 1.1484823625922886,
|
| 615 |
+
"grad_norm": 0.10220566391944885,
|
| 616 |
+
"learning_rate": 0.000917045131142242,
|
| 617 |
+
"loss": 0.2056,
|
| 618 |
+
"step": 1400
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 1.1484823625922886,
|
| 622 |
+
"eval_loss": 0.2266603708267212,
|
| 623 |
+
"eval_runtime": 19.5243,
|
| 624 |
+
"eval_samples_per_second": 51.218,
|
| 625 |
+
"eval_steps_per_second": 0.819,
|
| 626 |
+
"step": 1400
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 1.1689909762100081,
|
| 630 |
+
"grad_norm": 0.08700072765350342,
|
| 631 |
+
"learning_rate": 0.0008814317330582753,
|
| 632 |
+
"loss": 0.2087,
|
| 633 |
+
"step": 1425
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 1.1894995898277276,
|
| 637 |
+
"grad_norm": 0.09305619448423386,
|
| 638 |
+
"learning_rate": 0.000845970259752183,
|
| 639 |
+
"loss": 0.2154,
|
| 640 |
+
"step": 1450
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 1.1894995898277276,
|
| 644 |
+
"eval_loss": 0.225525364279747,
|
| 645 |
+
"eval_runtime": 19.536,
|
| 646 |
+
"eval_samples_per_second": 51.188,
|
| 647 |
+
"eval_steps_per_second": 0.819,
|
| 648 |
+
"step": 1450
|
| 649 |
+
},
|
| 650 |
+
{
|
| 651 |
+
"epoch": 1.2100082034454471,
|
| 652 |
+
"grad_norm": 0.08418343216180801,
|
| 653 |
+
"learning_rate": 0.0008107061489839498,
|
| 654 |
+
"loss": 0.2074,
|
| 655 |
+
"step": 1475
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 1.2305168170631666,
|
| 659 |
+
"grad_norm": 0.08762308210134506,
|
| 660 |
+
"learning_rate": 0.0007756845856275194,
|
| 661 |
+
"loss": 0.2142,
|
| 662 |
+
"step": 1500
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 1.2305168170631666,
|
| 666 |
+
"eval_loss": 0.2249392718076706,
|
| 667 |
+
"eval_runtime": 19.5325,
|
| 668 |
+
"eval_samples_per_second": 51.197,
|
| 669 |
+
"eval_steps_per_second": 0.819,
|
| 670 |
+
"step": 1500
|
| 671 |
+
},
|
| 672 |
+
{
|
| 673 |
+
"epoch": 1.251025430680886,
|
| 674 |
+
"grad_norm": 0.08624199777841568,
|
| 675 |
+
"learning_rate": 0.0007409504437741722,
|
| 676 |
+
"loss": 0.2036,
|
| 677 |
+
"step": 1525
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 1.2715340442986054,
|
| 681 |
+
"grad_norm": 0.10422079265117645,
|
| 682 |
+
"learning_rate": 0.0007065482292341205,
|
| 683 |
+
"loss": 0.2064,
|
| 684 |
+
"step": 1550
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 1.2715340442986054,
|
| 688 |
+
"eval_loss": 0.22372664511203766,
|
| 689 |
+
"eval_runtime": 19.4993,
|
| 690 |
+
"eval_samples_per_second": 51.284,
|
| 691 |
+
"eval_steps_per_second": 0.821,
|
| 692 |
+
"step": 1550
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 1.2920426579163249,
|
| 696 |
+
"grad_norm": 0.10072669386863708,
|
| 697 |
+
"learning_rate": 0.0006725220225099911,
|
| 698 |
+
"loss": 0.2026,
|
| 699 |
+
"step": 1575
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 1.3125512715340442,
|
| 703 |
+
"grad_norm": 0.08004007488489151,
|
| 704 |
+
"learning_rate": 0.0006389154223152666,
|
| 705 |
+
"loss": 0.1966,
|
| 706 |
+
"step": 1600
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 1.3125512715340442,
|
| 710 |
+
"eval_loss": 0.22237621247768402,
|
| 711 |
+
"eval_runtime": 19.5461,
|
| 712 |
+
"eval_samples_per_second": 51.161,
|
| 713 |
+
"eval_steps_per_second": 0.819,
|
| 714 |
+
"step": 1600
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 1.3330598851517639,
|
| 718 |
+
"grad_norm": 0.08824878185987473,
|
| 719 |
+
"learning_rate": 0.0006057714897100551,
|
| 720 |
+
"loss": 0.2107,
|
| 721 |
+
"step": 1625
|
| 722 |
+
},
|
| 723 |
+
{
|
| 724 |
+
"epoch": 1.3535684987694832,
|
| 725 |
+
"grad_norm": 0.09968467056751251,
|
| 726 |
+
"learning_rate": 0.0005731326929257713,
|
| 727 |
+
"loss": 0.2025,
|
| 728 |
+
"step": 1650
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 1.3535684987694832,
|
| 732 |
+
"eval_loss": 0.22190262377262115,
|
| 733 |
+
"eval_runtime": 19.5253,
|
| 734 |
+
"eval_samples_per_second": 51.215,
|
| 735 |
+
"eval_steps_per_second": 0.819,
|
| 736 |
+
"step": 1650
|
| 737 |
+
},
|
| 738 |
+
{
|
| 739 |
+
"epoch": 1.3740771123872026,
|
| 740 |
+
"grad_norm": 0.08944033086299896,
|
| 741 |
+
"learning_rate": 0.0005410408529494251,
|
| 742 |
+
"loss": 0.2009,
|
| 743 |
+
"step": 1675
|
| 744 |
+
},
|
| 745 |
+
{
|
| 746 |
+
"epoch": 1.3945857260049221,
|
| 747 |
+
"grad_norm": 0.10983427613973618,
|
| 748 |
+
"learning_rate": 0.0005095370899372412,
|
| 749 |
+
"loss": 0.2047,
|
| 750 |
+
"step": 1700
|
| 751 |
+
},
|
| 752 |
+
{
|
| 753 |
+
"epoch": 1.3945857260049221,
|
| 754 |
+
"eval_loss": 0.2195666879415512,
|
| 755 |
+
"eval_runtime": 19.5298,
|
| 756 |
+
"eval_samples_per_second": 51.204,
|
| 757 |
+
"eval_steps_per_second": 0.819,
|
| 758 |
+
"step": 1700
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 1.4150943396226414,
|
| 762 |
+
"grad_norm": 0.0977800264954567,
|
| 763 |
+
"learning_rate": 0.0004786617705262746,
|
| 764 |
+
"loss": 0.1974,
|
| 765 |
+
"step": 1725
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 1.435602953240361,
|
| 769 |
+
"grad_norm": 0.09195175021886826,
|
| 770 |
+
"learning_rate": 0.000448454456111529,
|
| 771 |
+
"loss": 0.2108,
|
| 772 |
+
"step": 1750
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"epoch": 1.435602953240361,
|
| 776 |
+
"eval_loss": 0.21818041801452637,
|
| 777 |
+
"eval_runtime": 19.4982,
|
| 778 |
+
"eval_samples_per_second": 51.287,
|
| 779 |
+
"eval_steps_per_second": 0.821,
|
| 780 |
+
"step": 1750
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 1.4561115668580804,
|
| 784 |
+
"grad_norm": 0.07893190532922745,
|
| 785 |
+
"learning_rate": 0.0004189538521548524,
|
| 786 |
+
"loss": 0.2039,
|
| 787 |
+
"step": 1775
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 1.4766201804758,
|
| 791 |
+
"grad_norm": 0.08887404948472977,
|
| 792 |
+
"learning_rate": 0.00039019775859056916,
|
| 793 |
+
"loss": 0.2013,
|
| 794 |
+
"step": 1800
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 1.4766201804758,
|
| 798 |
+
"eval_loss": 0.21688690781593323,
|
| 799 |
+
"eval_runtime": 19.5295,
|
| 800 |
+
"eval_samples_per_second": 51.205,
|
| 801 |
+
"eval_steps_per_second": 0.819,
|
| 802 |
+
"step": 1800
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 1.4971287940935194,
|
| 806 |
+
"grad_norm": 0.10588109493255615,
|
| 807 |
+
"learning_rate": 0.0003622230213913836,
|
| 808 |
+
"loss": 0.1978,
|
| 809 |
+
"step": 1825
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 1.5176374077112387,
|
| 813 |
+
"grad_norm": 0.09462971240282059,
|
| 814 |
+
"learning_rate": 0.0003350654853566223,
|
| 815 |
+
"loss": 0.2083,
|
| 816 |
+
"step": 1850
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"epoch": 1.5176374077112387,
|
| 820 |
+
"eval_loss": 0.21524257957935333,
|
| 821 |
+
"eval_runtime": 19.5333,
|
| 822 |
+
"eval_samples_per_second": 51.195,
|
| 823 |
+
"eval_steps_per_second": 0.819,
|
| 824 |
+
"step": 1850
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 1.5381460213289582,
|
| 828 |
+
"grad_norm": 0.092497818171978,
|
| 829 |
+
"learning_rate": 0.00030875994818330957,
|
| 830 |
+
"loss": 0.1978,
|
| 831 |
+
"step": 1875
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 1.5586546349466777,
|
| 835 |
+
"grad_norm": 0.0886370837688446,
|
| 836 |
+
"learning_rate": 0.0002833401158789207,
|
| 837 |
+
"loss": 0.2083,
|
| 838 |
+
"step": 1900
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 1.5586546349466777,
|
| 842 |
+
"eval_loss": 0.21452394127845764,
|
| 843 |
+
"eval_runtime": 19.5059,
|
| 844 |
+
"eval_samples_per_second": 51.266,
|
| 845 |
+
"eval_steps_per_second": 0.82,
|
| 846 |
+
"step": 1900
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 1.579163248564397,
|
| 850 |
+
"grad_norm": 0.09655001759529114,
|
| 851 |
+
"learning_rate": 0.00025883855957295053,
|
| 852 |
+
"loss": 0.192,
|
| 853 |
+
"step": 1925
|
| 854 |
+
},
|
| 855 |
+
{
|
| 856 |
+
"epoch": 1.5996718621821167,
|
| 857 |
+
"grad_norm": 0.08523295074701309,
|
| 858 |
+
"learning_rate": 0.0002352866737826277,
|
| 859 |
+
"loss": 0.2012,
|
| 860 |
+
"step": 1950
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 1.5996718621821167,
|
| 864 |
+
"eval_loss": 0.21258948743343353,
|
| 865 |
+
"eval_runtime": 19.4933,
|
| 866 |
+
"eval_samples_per_second": 51.3,
|
| 867 |
+
"eval_steps_per_second": 0.821,
|
| 868 |
+
"step": 1950
|
| 869 |
+
},
|
| 870 |
+
{
|
| 871 |
+
"epoch": 1.620180475799836,
|
| 872 |
+
"grad_norm": 0.09865439683198929,
|
| 873 |
+
"learning_rate": 0.00021271463618625986,
|
| 874 |
+
"loss": 0.1998,
|
| 875 |
+
"step": 1975
|
| 876 |
+
},
|
| 877 |
+
{
|
| 878 |
+
"epoch": 1.6406890894175554,
|
| 879 |
+
"grad_norm": 0.10992613434791565,
|
| 880 |
+
"learning_rate": 0.00019115136895574402,
|
| 881 |
+
"loss": 0.1974,
|
| 882 |
+
"step": 2000
|
| 883 |
+
},
|
| 884 |
+
{
|
| 885 |
+
"epoch": 1.6406890894175554,
|
| 886 |
+
"eval_loss": 0.21244320273399353,
|
| 887 |
+
"eval_runtime": 19.5047,
|
| 888 |
+
"eval_samples_per_second": 51.27,
|
| 889 |
+
"eval_steps_per_second": 0.82,
|
| 890 |
+
"step": 2000
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 1.661197703035275,
|
| 894 |
+
"grad_norm": 0.09869453310966492,
|
| 895 |
+
"learning_rate": 0.0001706245016977931,
|
| 896 |
+
"loss": 0.1984,
|
| 897 |
+
"step": 2025
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 1.6817063166529942,
|
| 901 |
+
"grad_norm": 0.08547994494438171,
|
| 902 |
+
"learning_rate": 0.00015116033605136182,
|
| 903 |
+
"loss": 0.2059,
|
| 904 |
+
"step": 2050
|
| 905 |
+
},
|
| 906 |
+
{
|
| 907 |
+
"epoch": 1.6817063166529942,
|
| 908 |
+
"eval_loss": 0.2115185409784317,
|
| 909 |
+
"eval_runtime": 19.4906,
|
| 910 |
+
"eval_samples_per_second": 51.307,
|
| 911 |
+
"eval_steps_per_second": 0.821,
|
| 912 |
+
"step": 2050
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 1.7022149302707137,
|
| 916 |
+
"grad_norm": 0.09922394901514053,
|
| 917 |
+
"learning_rate": 0.00013278381198663492,
|
| 918 |
+
"loss": 0.194,
|
| 919 |
+
"step": 2075
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 1.7227235438884332,
|
| 923 |
+
"grad_norm": 0.09176763892173767,
|
| 924 |
+
"learning_rate": 0.0001155184758487573,
|
| 925 |
+
"loss": 0.1929,
|
| 926 |
+
"step": 2100
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 1.7227235438884332,
|
| 930 |
+
"eval_loss": 0.21128395199775696,
|
| 931 |
+
"eval_runtime": 19.5212,
|
| 932 |
+
"eval_samples_per_second": 51.226,
|
| 933 |
+
"eval_steps_per_second": 0.82,
|
| 934 |
+
"step": 2100
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 1.7432321575061525,
|
| 938 |
+
"grad_norm": 0.10438892245292664,
|
| 939 |
+
"learning_rate": 9.938645018725523e-05,
|
| 940 |
+
"loss": 0.1987,
|
| 941 |
+
"step": 2125
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 1.7637407711238722,
|
| 945 |
+
"grad_norm": 0.10125499963760376,
|
| 946 |
+
"learning_rate": 8.440840540980587e-05,
|
| 947 |
+
"loss": 0.192,
|
| 948 |
+
"step": 2150
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 1.7637407711238722,
|
| 952 |
+
"eval_loss": 0.21018683910369873,
|
| 953 |
+
"eval_runtime": 19.5296,
|
| 954 |
+
"eval_samples_per_second": 51.204,
|
| 955 |
+
"eval_steps_per_second": 0.819,
|
| 956 |
+
"step": 2150
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 1.7842493847415914,
|
| 960 |
+
"grad_norm": 0.1060590147972107,
|
| 961 |
+
"learning_rate": 7.060353329667668e-05,
|
| 962 |
+
"loss": 0.205,
|
| 963 |
+
"step": 2175
|
| 964 |
+
},
|
| 965 |
+
{
|
| 966 |
+
"epoch": 1.804757998359311,
|
| 967 |
+
"grad_norm": 0.09252982586622238,
|
| 968 |
+
"learning_rate": 5.798952240976951e-05,
|
| 969 |
+
"loss": 0.197,
|
| 970 |
+
"step": 2200
|
| 971 |
+
},
|
| 972 |
+
{
|
| 973 |
+
"epoch": 1.804757998359311,
|
| 974 |
+
"eval_loss": 0.2095346748828888,
|
| 975 |
+
"eval_runtime": 19.5291,
|
| 976 |
+
"eval_samples_per_second": 51.206,
|
| 977 |
+
"eval_steps_per_second": 0.819,
|
| 978 |
+
"step": 2200
|
| 979 |
+
},
|
| 980 |
+
{
|
| 981 |
+
"epoch": 1.8252666119770304,
|
| 982 |
+
"grad_norm": 0.08407966792583466,
|
| 983 |
+
"learning_rate": 4.65825354277799e-05,
|
| 984 |
+
"loss": 0.1919,
|
| 985 |
+
"step": 2225
|
| 986 |
+
},
|
| 987 |
+
{
|
| 988 |
+
"epoch": 1.8457752255947497,
|
| 989 |
+
"grad_norm": 0.10161825269460678,
|
| 990 |
+
"learning_rate": 3.639718843651363e-05,
|
| 991 |
+
"loss": 0.1963,
|
| 992 |
+
"step": 2250
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 1.8457752255947497,
|
| 996 |
+
"eval_loss": 0.20933018624782562,
|
| 997 |
+
"eval_runtime": 19.5415,
|
| 998 |
+
"eval_samples_per_second": 51.173,
|
| 999 |
+
"eval_steps_per_second": 0.819,
|
| 1000 |
+
"step": 2250
|
| 1001 |
+
},
|
| 1002 |
+
{
|
| 1003 |
+
"epoch": 1.8662838392124692,
|
| 1004 |
+
"grad_norm": 0.09280096739530563,
|
| 1005 |
+
"learning_rate": 2.7446532200894104e-05,
|
| 1006 |
+
"loss": 0.1916,
|
| 1007 |
+
"step": 2275
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"epoch": 1.8867924528301887,
|
| 1011 |
+
"grad_norm": 0.11424104869365692,
|
| 1012 |
+
"learning_rate": 1.9742035442658403e-05,
|
| 1013 |
+
"loss": 0.2008,
|
| 1014 |
+
"step": 2300
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"epoch": 1.8867924528301887,
|
| 1018 |
+
"eval_loss": 0.20905601978302002,
|
| 1019 |
+
"eval_runtime": 19.5492,
|
| 1020 |
+
"eval_samples_per_second": 51.153,
|
| 1021 |
+
"eval_steps_per_second": 0.818,
|
| 1022 |
+
"step": 2300
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 1.907301066447908,
|
| 1026 |
+
"grad_norm": 0.08300191909074783,
|
| 1027 |
+
"learning_rate": 1.3293570145169742e-05,
|
| 1028 |
+
"loss": 0.1947,
|
| 1029 |
+
"step": 2325
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 1.9278096800656277,
|
| 1033 |
+
"grad_norm": 0.08883219957351685,
|
| 1034 |
+
"learning_rate": 8.109398904173282e-06,
|
| 1035 |
+
"loss": 0.1932,
|
| 1036 |
+
"step": 2350
|
| 1037 |
+
},
|
| 1038 |
+
{
|
| 1039 |
+
"epoch": 1.9278096800656277,
|
| 1040 |
+
"eval_loss": 0.20888373255729675,
|
| 1041 |
+
"eval_runtime": 19.541,
|
| 1042 |
+
"eval_samples_per_second": 51.174,
|
| 1043 |
+
"eval_steps_per_second": 0.819,
|
| 1044 |
+
"step": 2350
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 1.948318293683347,
|
| 1048 |
+
"grad_norm": 0.08763596415519714,
|
| 1049 |
+
"learning_rate": 4.196164340705577e-06,
|
| 1050 |
+
"loss": 0.1902,
|
| 1051 |
+
"step": 2375
|
| 1052 |
+
},
|
| 1053 |
+
{
|
| 1054 |
+
"epoch": 1.9688269073010665,
|
| 1055 |
+
"grad_norm": 0.07976502925157547,
|
| 1056 |
+
"learning_rate": 1.5588805897215342e-06,
|
| 1057 |
+
"loss": 0.1963,
|
| 1058 |
+
"step": 2400
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 1.9688269073010665,
|
| 1062 |
+
"eval_loss": 0.2088588923215866,
|
| 1063 |
+
"eval_runtime": 19.5331,
|
| 1064 |
+
"eval_samples_per_second": 51.195,
|
| 1065 |
+
"eval_steps_per_second": 0.819,
|
| 1066 |
+
"step": 2400
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 1.989335520918786,
|
| 1070 |
+
"grad_norm": 0.08866149187088013,
|
| 1071 |
+
"learning_rate": 2.0092687534589705e-07,
|
| 1072 |
+
"loss": 0.1948,
|
| 1073 |
+
"step": 2425
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 2.0,
|
| 1077 |
+
"step": 2438,
|
| 1078 |
+
"total_flos": 1.58523627405312e+18,
|
| 1079 |
+
"train_loss": 0.25077993895207595,
|
| 1080 |
+
"train_runtime": 3357.8107,
|
| 1081 |
+
"train_samples_per_second": 23.229,
|
| 1082 |
+
"train_steps_per_second": 0.726
|
| 1083 |
+
}
|
| 1084 |
+
],
|
| 1085 |
+
"logging_steps": 25,
|
| 1086 |
+
"max_steps": 2438,
|
| 1087 |
+
"num_input_tokens_seen": 0,
|
| 1088 |
+
"num_train_epochs": 2,
|
| 1089 |
+
"save_steps": 500,
|
| 1090 |
+
"stateful_callbacks": {
|
| 1091 |
+
"TrainerControl": {
|
| 1092 |
+
"args": {
|
| 1093 |
+
"should_epoch_stop": false,
|
| 1094 |
+
"should_evaluate": false,
|
| 1095 |
+
"should_log": false,
|
| 1096 |
+
"should_save": true,
|
| 1097 |
+
"should_training_stop": true
|
| 1098 |
+
},
|
| 1099 |
+
"attributes": {}
|
| 1100 |
+
}
|
| 1101 |
+
},
|
| 1102 |
+
"total_flos": 1.58523627405312e+18,
|
| 1103 |
+
"train_batch_size": 32,
|
| 1104 |
+
"trial_name": null,
|
| 1105 |
+
"trial_params": null
|
| 1106 |
+
}
|
nl_tasks/run_all/exnr15/ft/adapter_config.json
ADDED
|
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"T": 1.0,
|
| 3 |
+
"base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
|
| 4 |
+
"bias": "none",
|
| 5 |
+
"inference_mode": false,
|
| 6 |
+
"layers_to_transform": null,
|
| 7 |
+
"modules_to_save": null,
|
| 8 |
+
"num_rotations": 4,
|
| 9 |
+
"peft_type": "ROTATION",
|
| 10 |
+
"r": 4,
|
| 11 |
+
"revision": null,
|
| 12 |
+
"target_modules": [
|
| 13 |
+
"q_proj",
|
| 14 |
+
"v_proj"
|
| 15 |
+
],
|
| 16 |
+
"target_modules_to_skip": null,
|
| 17 |
+
"task_type": "CAUSAL_LM"
|
| 18 |
+
}
|
nl_tasks/run_all/exnr15/ft/added_tokens.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"[PAD]": 32000
|
| 3 |
+
}
|
nl_tasks/run_all/exnr15/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/run_all/exnr15/ft/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
nl_tasks/run_all/exnr15/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/run_all/exnr15/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/run_all/exnr15/ft/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:e78d8cf553e1bd2f18138af07f897785457fbcd326b6f0026cb8b64209a2ab44
|
| 3 |
+
size 6545
|
nl_tasks/run_all/exnr15/ft2/adapter_config.json
ADDED
|
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"T": 1.0,
|
| 3 |
+
"base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
|
| 4 |
+
"bias": "none",
|
| 5 |
+
"inference_mode": true,
|
| 6 |
+
"layers_to_transform": null,
|
| 7 |
+
"modules_to_save": null,
|
| 8 |
+
"num_rotations": 4,
|
| 9 |
+
"peft_type": "ROTATION",
|
| 10 |
+
"r": 4,
|
| 11 |
+
"revision": null,
|
| 12 |
+
"target_modules": [
|
| 13 |
+
"q_proj",
|
| 14 |
+
"v_proj"
|
| 15 |
+
],
|
| 16 |
+
"target_modules_to_skip": null,
|
| 17 |
+
"task_type": "CAUSAL_LM"
|
| 18 |
+
}
|
nl_tasks/run_all/exnr15/ft2/adapter_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:80a7fb292572659a561e65a657b939feb3d3203477c08f5e6ace0852fb4af627
|
| 3 |
+
size 33602659
|
nl_tasks/run_all/exnr15/trainer_state.json
ADDED
|
@@ -0,0 +1,135 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 0.16406890894175555,
|
| 6 |
+
"eval_steps": 50,
|
| 7 |
+
"global_step": 200,
|
| 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": 0.0006118718883953989,
|
| 15 |
+
"learning_rate": 0.0009987820251299122,
|
| 16 |
+
"loss": 0.7714,
|
| 17 |
+
"step": 25
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.04101722723543889,
|
| 21 |
+
"grad_norm": 0.0006668232963420451,
|
| 22 |
+
"learning_rate": 0.0009373098535696979,
|
| 23 |
+
"loss": 0.7596,
|
| 24 |
+
"step": 50
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.04101722723543889,
|
| 28 |
+
"eval_loss": 0.7446768283843994,
|
| 29 |
+
"eval_model_preparation_time": 0.009,
|
| 30 |
+
"eval_runtime": 53.4492,
|
| 31 |
+
"eval_samples_per_second": 18.709,
|
| 32 |
+
"eval_steps_per_second": 0.299,
|
| 33 |
+
"step": 50
|
| 34 |
+
},
|
| 35 |
+
{
|
| 36 |
+
"epoch": 0.06152584085315833,
|
| 37 |
+
"grad_norm": 0.0006718848599120975,
|
| 38 |
+
"learning_rate": 0.0007938926261462366,
|
| 39 |
+
"loss": 0.7554,
|
| 40 |
+
"step": 75
|
| 41 |
+
},
|
| 42 |
+
{
|
| 43 |
+
"epoch": 0.08203445447087777,
|
| 44 |
+
"grad_norm": 0.0006972206756472588,
|
| 45 |
+
"learning_rate": 0.0005954044976882724,
|
| 46 |
+
"loss": 0.7346,
|
| 47 |
+
"step": 100
|
| 48 |
+
},
|
| 49 |
+
{
|
| 50 |
+
"epoch": 0.08203445447087777,
|
| 51 |
+
"eval_loss": 0.7248194813728333,
|
| 52 |
+
"eval_model_preparation_time": 0.009,
|
| 53 |
+
"eval_runtime": 19.7467,
|
| 54 |
+
"eval_samples_per_second": 50.641,
|
| 55 |
+
"eval_steps_per_second": 0.81,
|
| 56 |
+
"step": 100
|
| 57 |
+
},
|
| 58 |
+
{
|
| 59 |
+
"epoch": 0.10254306808859721,
|
| 60 |
+
"grad_norm": 0.0006274359184317291,
|
| 61 |
+
"learning_rate": 0.0003790390522001662,
|
| 62 |
+
"loss": 0.7366,
|
| 63 |
+
"step": 125
|
| 64 |
+
},
|
| 65 |
+
{
|
| 66 |
+
"epoch": 0.12305168170631665,
|
| 67 |
+
"grad_norm": 0.0005907765007577837,
|
| 68 |
+
"learning_rate": 0.00018533980447508135,
|
| 69 |
+
"loss": 0.7368,
|
| 70 |
+
"step": 150
|
| 71 |
+
},
|
| 72 |
+
{
|
| 73 |
+
"epoch": 0.12305168170631665,
|
| 74 |
+
"eval_loss": 0.7148993015289307,
|
| 75 |
+
"eval_model_preparation_time": 0.009,
|
| 76 |
+
"eval_runtime": 19.6656,
|
| 77 |
+
"eval_samples_per_second": 50.85,
|
| 78 |
+
"eval_steps_per_second": 0.814,
|
| 79 |
+
"step": 150
|
| 80 |
+
},
|
| 81 |
+
{
|
| 82 |
+
"epoch": 0.1435602953240361,
|
| 83 |
+
"grad_norm": 0.0007314748945645988,
|
| 84 |
+
"learning_rate": 5.060297685041659e-05,
|
| 85 |
+
"loss": 0.7228,
|
| 86 |
+
"step": 175
|
| 87 |
+
},
|
| 88 |
+
{
|
| 89 |
+
"epoch": 0.16406890894175555,
|
| 90 |
+
"grad_norm": 0.0006935550482012331,
|
| 91 |
+
"learning_rate": 7.615242180436521e-08,
|
| 92 |
+
"loss": 0.7275,
|
| 93 |
+
"step": 200
|
| 94 |
+
},
|
| 95 |
+
{
|
| 96 |
+
"epoch": 0.16406890894175555,
|
| 97 |
+
"eval_loss": 0.7132272720336914,
|
| 98 |
+
"eval_model_preparation_time": 0.009,
|
| 99 |
+
"eval_runtime": 19.665,
|
| 100 |
+
"eval_samples_per_second": 50.852,
|
| 101 |
+
"eval_steps_per_second": 0.814,
|
| 102 |
+
"step": 200
|
| 103 |
+
},
|
| 104 |
+
{
|
| 105 |
+
"epoch": 0.16406890894175555,
|
| 106 |
+
"step": 200,
|
| 107 |
+
"total_flos": 1.30070668640256e+17,
|
| 108 |
+
"train_loss": 0.7430903720855713,
|
| 109 |
+
"train_runtime": 454.6042,
|
| 110 |
+
"train_samples_per_second": 14.078,
|
| 111 |
+
"train_steps_per_second": 0.44
|
| 112 |
+
}
|
| 113 |
+
],
|
| 114 |
+
"logging_steps": 25,
|
| 115 |
+
"max_steps": 200,
|
| 116 |
+
"num_input_tokens_seen": 0,
|
| 117 |
+
"num_train_epochs": 1,
|
| 118 |
+
"save_steps": 500,
|
| 119 |
+
"stateful_callbacks": {
|
| 120 |
+
"TrainerControl": {
|
| 121 |
+
"args": {
|
| 122 |
+
"should_epoch_stop": false,
|
| 123 |
+
"should_evaluate": false,
|
| 124 |
+
"should_log": false,
|
| 125 |
+
"should_save": true,
|
| 126 |
+
"should_training_stop": true
|
| 127 |
+
},
|
| 128 |
+
"attributes": {}
|
| 129 |
+
}
|
| 130 |
+
},
|
| 131 |
+
"total_flos": 1.30070668640256e+17,
|
| 132 |
+
"train_batch_size": 32,
|
| 133 |
+
"trial_name": null,
|
| 134 |
+
"trial_params": null
|
| 135 |
+
}
|
nl_tasks/run_all/run_exnr10/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 |
+
"v_proj",
|
| 14 |
+
"q_proj"
|
| 15 |
+
],
|
| 16 |
+
"target_modules_to_skip": null,
|
| 17 |
+
"task_type": "CAUSAL_LM"
|
| 18 |
+
}
|
nl_tasks/run_all/run_exnr10/ft/added_tokens.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"[PAD]": 32000
|
| 3 |
+
}
|
nl_tasks/run_all/run_exnr10/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/run_all/run_exnr10/ft/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
nl_tasks/run_all/run_exnr10/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/run_all/run_exnr10/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/run_all/run_exnr10/ft/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:80200b92f1eba2aa222a8b5e89fb435709191eb6ae0eb5fad588e902f3ef01b1
|
| 3 |
+
size 6481
|
nl_tasks/run_all/run_exnr10/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 |
+
"v_proj",
|
| 14 |
+
"q_proj"
|
| 15 |
+
],
|
| 16 |
+
"target_modules_to_skip": null,
|
| 17 |
+
"task_type": "CAUSAL_LM"
|
| 18 |
+
}
|
nl_tasks/run_all/run_exnr10/ft2/adapter_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:373029eeee0417e9247131c310db05bcef947e17ea80622d6498c15ab27e173e
|
| 3 |
+
size 33602659
|
nl_tasks/run_all/run_exnr10/trainer_state.json
ADDED
|
@@ -0,0 +1,1106 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 2.0,
|
| 6 |
+
"eval_steps": 50,
|
| 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": 0.23067046701908112,
|
| 15 |
+
"learning_rate": 9.836065573770491e-05,
|
| 16 |
+
"loss": 0.6777,
|
| 17 |
+
"step": 25
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.04101722723543889,
|
| 21 |
+
"grad_norm": 0.18888919055461884,
|
| 22 |
+
"learning_rate": 0.00020081967213114754,
|
| 23 |
+
"loss": 0.4155,
|
| 24 |
+
"step": 50
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.04101722723543889,
|
| 28 |
+
"eval_loss": 0.37254372239112854,
|
| 29 |
+
"eval_runtime": 21.1288,
|
| 30 |
+
"eval_samples_per_second": 47.329,
|
| 31 |
+
"eval_steps_per_second": 0.757,
|
| 32 |
+
"step": 50
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.06152584085315833,
|
| 36 |
+
"grad_norm": 0.18046127259731293,
|
| 37 |
+
"learning_rate": 0.0003032786885245902,
|
| 38 |
+
"loss": 0.3704,
|
| 39 |
+
"step": 75
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.08203445447087777,
|
| 43 |
+
"grad_norm": 0.19047358632087708,
|
| 44 |
+
"learning_rate": 0.0004057377049180328,
|
| 45 |
+
"loss": 0.3503,
|
| 46 |
+
"step": 100
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"epoch": 0.08203445447087777,
|
| 50 |
+
"eval_loss": 0.32894426584243774,
|
| 51 |
+
"eval_runtime": 20.8119,
|
| 52 |
+
"eval_samples_per_second": 48.049,
|
| 53 |
+
"eval_steps_per_second": 0.769,
|
| 54 |
+
"step": 100
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.10254306808859721,
|
| 58 |
+
"grad_norm": 0.1786128282546997,
|
| 59 |
+
"learning_rate": 0.0005081967213114754,
|
| 60 |
+
"loss": 0.3301,
|
| 61 |
+
"step": 125
|
| 62 |
+
},
|
| 63 |
+
{
|
| 64 |
+
"epoch": 0.12305168170631665,
|
| 65 |
+
"grad_norm": 0.17241524159908295,
|
| 66 |
+
"learning_rate": 0.000610655737704918,
|
| 67 |
+
"loss": 0.3158,
|
| 68 |
+
"step": 150
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.12305168170631665,
|
| 72 |
+
"eval_loss": 0.31440791487693787,
|
| 73 |
+
"eval_runtime": 20.778,
|
| 74 |
+
"eval_samples_per_second": 48.128,
|
| 75 |
+
"eval_steps_per_second": 0.77,
|
| 76 |
+
"step": 150
|
| 77 |
+
},
|
| 78 |
+
{
|
| 79 |
+
"epoch": 0.1435602953240361,
|
| 80 |
+
"grad_norm": 0.18751497566699982,
|
| 81 |
+
"learning_rate": 0.0007131147540983607,
|
| 82 |
+
"loss": 0.313,
|
| 83 |
+
"step": 175
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"epoch": 0.16406890894175555,
|
| 87 |
+
"grad_norm": 0.24119554460048676,
|
| 88 |
+
"learning_rate": 0.0008155737704918033,
|
| 89 |
+
"loss": 0.3209,
|
| 90 |
+
"step": 200
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 0.16406890894175555,
|
| 94 |
+
"eval_loss": 0.30967944860458374,
|
| 95 |
+
"eval_runtime": 20.8466,
|
| 96 |
+
"eval_samples_per_second": 47.969,
|
| 97 |
+
"eval_steps_per_second": 0.768,
|
| 98 |
+
"step": 200
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.184577522559475,
|
| 102 |
+
"grad_norm": 0.28959155082702637,
|
| 103 |
+
"learning_rate": 0.0009180327868852459,
|
| 104 |
+
"loss": 0.3282,
|
| 105 |
+
"step": 225
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 0.20508613617719443,
|
| 109 |
+
"grad_norm": 0.2528744637966156,
|
| 110 |
+
"learning_rate": 0.0009999871854116063,
|
| 111 |
+
"loss": 0.3229,
|
| 112 |
+
"step": 250
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 0.20508613617719443,
|
| 116 |
+
"eval_loss": 0.30937159061431885,
|
| 117 |
+
"eval_runtime": 20.8274,
|
| 118 |
+
"eval_samples_per_second": 48.014,
|
| 119 |
+
"eval_steps_per_second": 0.768,
|
| 120 |
+
"step": 250
|
| 121 |
+
},
|
| 122 |
+
{
|
| 123 |
+
"epoch": 0.22559474979491387,
|
| 124 |
+
"grad_norm": 0.5322397351264954,
|
| 125 |
+
"learning_rate": 0.0009995387437838027,
|
| 126 |
+
"loss": 0.3225,
|
| 127 |
+
"step": 275
|
| 128 |
+
},
|
| 129 |
+
{
|
| 130 |
+
"epoch": 0.2461033634126333,
|
| 131 |
+
"grad_norm": 13.223990440368652,
|
| 132 |
+
"learning_rate": 0.000998450229439693,
|
| 133 |
+
"loss": 0.354,
|
| 134 |
+
"step": 300
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.2461033634126333,
|
| 138 |
+
"eval_loss": 0.3210923969745636,
|
| 139 |
+
"eval_runtime": 20.8168,
|
| 140 |
+
"eval_samples_per_second": 48.038,
|
| 141 |
+
"eval_steps_per_second": 0.769,
|
| 142 |
+
"step": 300
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.2666119770303528,
|
| 146 |
+
"grad_norm": 0.3021000325679779,
|
| 147 |
+
"learning_rate": 0.000996723037122612,
|
| 148 |
+
"loss": 0.3151,
|
| 149 |
+
"step": 325
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.2871205906480722,
|
| 153 |
+
"grad_norm": 0.21538248658180237,
|
| 154 |
+
"learning_rate": 0.0009943593799315263,
|
| 155 |
+
"loss": 0.3059,
|
| 156 |
+
"step": 350
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.2871205906480722,
|
| 160 |
+
"eval_loss": 0.29110386967658997,
|
| 161 |
+
"eval_runtime": 20.8154,
|
| 162 |
+
"eval_samples_per_second": 48.041,
|
| 163 |
+
"eval_steps_per_second": 0.769,
|
| 164 |
+
"step": 350
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.30762920426579166,
|
| 168 |
+
"grad_norm": 0.2078874111175537,
|
| 169 |
+
"learning_rate": 0.0009913622864853324,
|
| 170 |
+
"loss": 0.2942,
|
| 171 |
+
"step": 375
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.3281378178835111,
|
| 175 |
+
"grad_norm": 0.2866668105125427,
|
| 176 |
+
"learning_rate": 0.0009877355970422024,
|
| 177 |
+
"loss": 0.3038,
|
| 178 |
+
"step": 400
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.3281378178835111,
|
| 182 |
+
"eval_loss": 0.28084269165992737,
|
| 183 |
+
"eval_runtime": 20.8237,
|
| 184 |
+
"eval_samples_per_second": 48.022,
|
| 185 |
+
"eval_steps_per_second": 0.768,
|
| 186 |
+
"step": 400
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 0.34864643150123054,
|
| 190 |
+
"grad_norm": 0.16703106462955475,
|
| 191 |
+
"learning_rate": 0.0009834839585789559,
|
| 192 |
+
"loss": 0.2962,
|
| 193 |
+
"step": 425
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 0.36915504511895,
|
| 197 |
+
"grad_norm": 0.18550001084804535,
|
| 198 |
+
"learning_rate": 0.000978612818836762,
|
| 199 |
+
"loss": 0.292,
|
| 200 |
+
"step": 450
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 0.36915504511895,
|
| 204 |
+
"eval_loss": 0.2726050913333893,
|
| 205 |
+
"eval_runtime": 20.8143,
|
| 206 |
+
"eval_samples_per_second": 48.044,
|
| 207 |
+
"eval_steps_per_second": 0.769,
|
| 208 |
+
"step": 450
|
| 209 |
+
},
|
| 210 |
+
{
|
| 211 |
+
"epoch": 0.3896636587366694,
|
| 212 |
+
"grad_norm": 0.18142051994800568,
|
| 213 |
+
"learning_rate": 0.0009731284193407981,
|
| 214 |
+
"loss": 0.287,
|
| 215 |
+
"step": 475
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 0.41017227235438886,
|
| 219 |
+
"grad_norm": 0.15791696310043335,
|
| 220 |
+
"learning_rate": 0.0009670377874028117,
|
| 221 |
+
"loss": 0.2788,
|
| 222 |
+
"step": 500
|
| 223 |
+
},
|
| 224 |
+
{
|
| 225 |
+
"epoch": 0.41017227235438886,
|
| 226 |
+
"eval_loss": 0.2660056948661804,
|
| 227 |
+
"eval_runtime": 20.7894,
|
| 228 |
+
"eval_samples_per_second": 48.101,
|
| 229 |
+
"eval_steps_per_second": 0.77,
|
| 230 |
+
"step": 500
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 0.4306808859721083,
|
| 234 |
+
"grad_norm": 0.17865672707557678,
|
| 235 |
+
"learning_rate": 0.0009603487271168336,
|
| 236 |
+
"loss": 0.2818,
|
| 237 |
+
"step": 525
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 0.45118949958982774,
|
| 241 |
+
"grad_norm": 0.1716611534357071,
|
| 242 |
+
"learning_rate": 0.0009530698093595781,
|
| 243 |
+
"loss": 0.2754,
|
| 244 |
+
"step": 550
|
| 245 |
+
},
|
| 246 |
+
{
|
| 247 |
+
"epoch": 0.45118949958982774,
|
| 248 |
+
"eval_loss": 0.25951725244522095,
|
| 249 |
+
"eval_runtime": 20.6867,
|
| 250 |
+
"eval_samples_per_second": 48.34,
|
| 251 |
+
"eval_steps_per_second": 0.773,
|
| 252 |
+
"step": 550
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 0.4716981132075472,
|
| 256 |
+
"grad_norm": 0.15952439606189728,
|
| 257 |
+
"learning_rate": 0.0009452103608083418,
|
| 258 |
+
"loss": 0.2624,
|
| 259 |
+
"step": 575
|
| 260 |
+
},
|
| 261 |
+
{
|
| 262 |
+
"epoch": 0.4922067268252666,
|
| 263 |
+
"grad_norm": 0.15508660674095154,
|
| 264 |
+
"learning_rate": 0.0009367804519904775,
|
| 265 |
+
"loss": 0.2707,
|
| 266 |
+
"step": 600
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 0.4922067268252666,
|
| 270 |
+
"eval_loss": 0.2577713131904602,
|
| 271 |
+
"eval_runtime": 20.8355,
|
| 272 |
+
"eval_samples_per_second": 47.995,
|
| 273 |
+
"eval_steps_per_second": 0.768,
|
| 274 |
+
"step": 600
|
| 275 |
+
},
|
| 276 |
+
{
|
| 277 |
+
"epoch": 0.5127153404429861,
|
| 278 |
+
"grad_norm": 0.14390869438648224,
|
| 279 |
+
"learning_rate": 0.0009277908843797492,
|
| 280 |
+
"loss": 0.258,
|
| 281 |
+
"step": 625
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 0.5332239540607056,
|
| 285 |
+
"grad_norm": 0.15374627709388733,
|
| 286 |
+
"learning_rate": 0.0009182531765561084,
|
| 287 |
+
"loss": 0.2575,
|
| 288 |
+
"step": 650
|
| 289 |
+
},
|
| 290 |
+
{
|
| 291 |
+
"epoch": 0.5332239540607056,
|
| 292 |
+
"eval_loss": 0.25287148356437683,
|
| 293 |
+
"eval_runtime": 20.8679,
|
| 294 |
+
"eval_samples_per_second": 47.921,
|
| 295 |
+
"eval_steps_per_second": 0.767,
|
| 296 |
+
"step": 650
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.5537325676784249,
|
| 300 |
+
"grad_norm": 0.13985492289066315,
|
| 301 |
+
"learning_rate": 0.0009081795494466201,
|
| 302 |
+
"loss": 0.2589,
|
| 303 |
+
"step": 675
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.5742411812961444,
|
| 307 |
+
"grad_norm": 0.11946714669466019,
|
| 308 |
+
"learning_rate": 0.0008975829106664539,
|
| 309 |
+
"loss": 0.2502,
|
| 310 |
+
"step": 700
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.5742411812961444,
|
| 314 |
+
"eval_loss": 0.24979238212108612,
|
| 315 |
+
"eval_runtime": 20.8494,
|
| 316 |
+
"eval_samples_per_second": 47.963,
|
| 317 |
+
"eval_steps_per_second": 0.767,
|
| 318 |
+
"step": 700
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.5947497949138638,
|
| 322 |
+
"grad_norm": 0.11909514665603638,
|
| 323 |
+
"learning_rate": 0.0008864768379800017,
|
| 324 |
+
"loss": 0.2475,
|
| 325 |
+
"step": 725
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.6152584085315833,
|
| 329 |
+
"grad_norm": 0.12616503238677979,
|
| 330 |
+
"learning_rate": 0.0008748755619033153,
|
| 331 |
+
"loss": 0.257,
|
| 332 |
+
"step": 750
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.6152584085315833,
|
| 336 |
+
"eval_loss": 0.24706576764583588,
|
| 337 |
+
"eval_runtime": 20.8598,
|
| 338 |
+
"eval_samples_per_second": 47.939,
|
| 339 |
+
"eval_steps_per_second": 0.767,
|
| 340 |
+
"step": 750
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 0.6357670221493027,
|
| 344 |
+
"grad_norm": 0.1338769644498825,
|
| 345 |
+
"learning_rate": 0.000862793947470155,
|
| 346 |
+
"loss": 0.262,
|
| 347 |
+
"step": 775
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 0.6562756357670222,
|
| 351 |
+
"grad_norm": 0.1327950358390808,
|
| 352 |
+
"learning_rate": 0.0008502474751850142,
|
| 353 |
+
"loss": 0.2512,
|
| 354 |
+
"step": 800
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.6562756357670222,
|
| 358 |
+
"eval_loss": 0.24406881630420685,
|
| 359 |
+
"eval_runtime": 20.8661,
|
| 360 |
+
"eval_samples_per_second": 47.925,
|
| 361 |
+
"eval_steps_per_second": 0.767,
|
| 362 |
+
"step": 800
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.6767842493847416,
|
| 366 |
+
"grad_norm": 0.11789478361606598,
|
| 367 |
+
"learning_rate": 0.0008372522211875224,
|
| 368 |
+
"loss": 0.2468,
|
| 369 |
+
"step": 825
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.6972928630024611,
|
| 373 |
+
"grad_norm": 0.114653080701828,
|
| 374 |
+
"learning_rate": 0.0008238248366536473,
|
| 375 |
+
"loss": 0.2593,
|
| 376 |
+
"step": 850
|
| 377 |
+
},
|
| 378 |
+
{
|
| 379 |
+
"epoch": 0.6972928630024611,
|
| 380 |
+
"eval_loss": 0.24199624359607697,
|
| 381 |
+
"eval_runtime": 20.8571,
|
| 382 |
+
"eval_samples_per_second": 47.945,
|
| 383 |
+
"eval_steps_per_second": 0.767,
|
| 384 |
+
"step": 850
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.7178014766201805,
|
| 388 |
+
"grad_norm": 0.11074954271316528,
|
| 389 |
+
"learning_rate": 0.0008099825264600842,
|
| 390 |
+
"loss": 0.2541,
|
| 391 |
+
"step": 875
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 0.7383100902379,
|
| 395 |
+
"grad_norm": 0.16259630024433136,
|
| 396 |
+
"learning_rate": 0.0007957430271391761,
|
| 397 |
+
"loss": 0.2466,
|
| 398 |
+
"step": 900
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 0.7383100902379,
|
| 402 |
+
"eval_loss": 0.23889920115470886,
|
| 403 |
+
"eval_runtime": 20.8704,
|
| 404 |
+
"eval_samples_per_second": 47.915,
|
| 405 |
+
"eval_steps_per_second": 0.767,
|
| 406 |
+
"step": 900
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 0.7588187038556193,
|
| 410 |
+
"grad_norm": 0.11665979772806168,
|
| 411 |
+
"learning_rate": 0.0007811245841526062,
|
| 412 |
+
"loss": 0.2499,
|
| 413 |
+
"step": 925
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 0.7793273174733388,
|
| 417 |
+
"grad_norm": 0.12356989085674286,
|
| 418 |
+
"learning_rate": 0.0007661459285129879,
|
| 419 |
+
"loss": 0.233,
|
| 420 |
+
"step": 950
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 0.7793273174733388,
|
| 424 |
+
"eval_loss": 0.23713098466396332,
|
| 425 |
+
"eval_runtime": 20.8545,
|
| 426 |
+
"eval_samples_per_second": 47.951,
|
| 427 |
+
"eval_steps_per_second": 0.767,
|
| 428 |
+
"step": 950
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 0.7998359310910582,
|
| 432 |
+
"grad_norm": 0.10465991497039795,
|
| 433 |
+
"learning_rate": 0.0007508262527833029,
|
| 434 |
+
"loss": 0.2465,
|
| 435 |
+
"step": 975
|
| 436 |
+
},
|
| 437 |
+
{
|
| 438 |
+
"epoch": 0.8203445447087777,
|
| 439 |
+
"grad_norm": 0.10666616261005402,
|
| 440 |
+
"learning_rate": 0.000735185186484943,
|
| 441 |
+
"loss": 0.2486,
|
| 442 |
+
"step": 1000
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 0.8203445447087777,
|
| 446 |
+
"eval_loss": 0.23608271777629852,
|
| 447 |
+
"eval_runtime": 20.8714,
|
| 448 |
+
"eval_samples_per_second": 47.912,
|
| 449 |
+
"eval_steps_per_second": 0.767,
|
| 450 |
+
"step": 1000
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.8408531583264971,
|
| 454 |
+
"grad_norm": 0.10759830474853516,
|
| 455 |
+
"learning_rate": 0.0007192427709458656,
|
| 456 |
+
"loss": 0.2363,
|
| 457 |
+
"step": 1025
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 0.8613617719442166,
|
| 461 |
+
"grad_norm": 0.11578875035047531,
|
| 462 |
+
"learning_rate": 0.0007030194336210887,
|
| 463 |
+
"loss": 0.2407,
|
| 464 |
+
"step": 1050
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.8613617719442166,
|
| 468 |
+
"eval_loss": 0.23380084335803986,
|
| 469 |
+
"eval_runtime": 20.8115,
|
| 470 |
+
"eval_samples_per_second": 48.05,
|
| 471 |
+
"eval_steps_per_second": 0.769,
|
| 472 |
+
"step": 1050
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 0.881870385561936,
|
| 476 |
+
"grad_norm": 0.11421164870262146,
|
| 477 |
+
"learning_rate": 0.0006865359619184331,
|
| 478 |
+
"loss": 0.2464,
|
| 479 |
+
"step": 1075
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 0.9023789991796555,
|
| 483 |
+
"grad_norm": 0.1076551228761673,
|
| 484 |
+
"learning_rate": 0.0006698134765630434,
|
| 485 |
+
"loss": 0.2387,
|
| 486 |
+
"step": 1100
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 0.9023789991796555,
|
| 490 |
+
"eval_loss": 0.23312732577323914,
|
| 491 |
+
"eval_runtime": 20.8233,
|
| 492 |
+
"eval_samples_per_second": 48.023,
|
| 493 |
+
"eval_steps_per_second": 0.768,
|
| 494 |
+
"step": 1100
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 0.9228876127973749,
|
| 498 |
+
"grad_norm": 0.10048159211874008,
|
| 499 |
+
"learning_rate": 0.0006528734045348248,
|
| 500 |
+
"loss": 0.2361,
|
| 501 |
+
"step": 1125
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 0.9433962264150944,
|
| 505 |
+
"grad_norm": 0.1083035096526146,
|
| 506 |
+
"learning_rate": 0.0006357374516134643,
|
| 507 |
+
"loss": 0.2506,
|
| 508 |
+
"step": 1150
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 0.9433962264150944,
|
| 512 |
+
"eval_loss": 0.2312103807926178,
|
| 513 |
+
"eval_runtime": 20.7981,
|
| 514 |
+
"eval_samples_per_second": 48.081,
|
| 515 |
+
"eval_steps_per_second": 0.769,
|
| 516 |
+
"step": 1150
|
| 517 |
+
},
|
| 518 |
+
{
|
| 519 |
+
"epoch": 0.9639048400328137,
|
| 520 |
+
"grad_norm": 0.11083986610174179,
|
| 521 |
+
"learning_rate": 0.0006184275745662179,
|
| 522 |
+
"loss": 0.2401,
|
| 523 |
+
"step": 1175
|
| 524 |
+
},
|
| 525 |
+
{
|
| 526 |
+
"epoch": 0.9844134536505332,
|
| 527 |
+
"grad_norm": 0.10034681111574173,
|
| 528 |
+
"learning_rate": 0.0006009659530141031,
|
| 529 |
+
"loss": 0.2428,
|
| 530 |
+
"step": 1200
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 0.9844134536505332,
|
| 534 |
+
"eval_loss": 0.22971327602863312,
|
| 535 |
+
"eval_runtime": 20.7973,
|
| 536 |
+
"eval_samples_per_second": 48.083,
|
| 537 |
+
"eval_steps_per_second": 0.769,
|
| 538 |
+
"step": 1200
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"epoch": 1.0049220672682526,
|
| 542 |
+
"grad_norm": 0.10005165636539459,
|
| 543 |
+
"learning_rate": 0.0005833749610125402,
|
| 544 |
+
"loss": 0.2395,
|
| 545 |
+
"step": 1225
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 1.0254306808859721,
|
| 549 |
+
"grad_norm": 0.10959237813949585,
|
| 550 |
+
"learning_rate": 0.0005656771383828602,
|
| 551 |
+
"loss": 0.2135,
|
| 552 |
+
"step": 1250
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 1.0254306808859721,
|
| 556 |
+
"eval_loss": 0.23102633655071259,
|
| 557 |
+
"eval_runtime": 20.7627,
|
| 558 |
+
"eval_samples_per_second": 48.163,
|
| 559 |
+
"eval_steps_per_second": 0.771,
|
| 560 |
+
"step": 1250
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 1.0459392945036916,
|
| 564 |
+
"grad_norm": 0.103731170296669,
|
| 565 |
+
"learning_rate": 0.0005478951618314134,
|
| 566 |
+
"loss": 0.2094,
|
| 567 |
+
"step": 1275
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 1.066447908121411,
|
| 571 |
+
"grad_norm": 0.11381426453590393,
|
| 572 |
+
"learning_rate": 0.0005300518158932815,
|
| 573 |
+
"loss": 0.2108,
|
| 574 |
+
"step": 1300
|
| 575 |
+
},
|
| 576 |
+
{
|
| 577 |
+
"epoch": 1.066447908121411,
|
| 578 |
+
"eval_loss": 0.2284233421087265,
|
| 579 |
+
"eval_runtime": 20.7865,
|
| 580 |
+
"eval_samples_per_second": 48.108,
|
| 581 |
+
"eval_steps_per_second": 0.77,
|
| 582 |
+
"step": 1300
|
| 583 |
+
},
|
| 584 |
+
{
|
| 585 |
+
"epoch": 1.0869565217391304,
|
| 586 |
+
"grad_norm": 0.10289537161588669,
|
| 587 |
+
"learning_rate": 0.0005121699637378282,
|
| 588 |
+
"loss": 0.2098,
|
| 589 |
+
"step": 1325
|
| 590 |
+
},
|
| 591 |
+
{
|
| 592 |
+
"epoch": 1.1074651353568499,
|
| 593 |
+
"grad_norm": 0.11907949298620224,
|
| 594 |
+
"learning_rate": 0.0004942725178734903,
|
| 595 |
+
"loss": 0.2152,
|
| 596 |
+
"step": 1350
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 1.1074651353568499,
|
| 600 |
+
"eval_loss": 0.22748351097106934,
|
| 601 |
+
"eval_runtime": 20.7803,
|
| 602 |
+
"eval_samples_per_second": 48.123,
|
| 603 |
+
"eval_steps_per_second": 0.77,
|
| 604 |
+
"step": 1350
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 1.1279737489745694,
|
| 608 |
+
"grad_norm": 0.11696625500917435,
|
| 609 |
+
"learning_rate": 0.00047638241078935324,
|
| 610 |
+
"loss": 0.2121,
|
| 611 |
+
"step": 1375
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 1.1484823625922886,
|
| 615 |
+
"grad_norm": 0.10638347268104553,
|
| 616 |
+
"learning_rate": 0.000458522565571121,
|
| 617 |
+
"loss": 0.2201,
|
| 618 |
+
"step": 1400
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 1.1484823625922886,
|
| 622 |
+
"eval_loss": 0.22692248225212097,
|
| 623 |
+
"eval_runtime": 20.7646,
|
| 624 |
+
"eval_samples_per_second": 48.159,
|
| 625 |
+
"eval_steps_per_second": 0.771,
|
| 626 |
+
"step": 1400
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 1.1689909762100081,
|
| 630 |
+
"grad_norm": 0.10754355788230896,
|
| 631 |
+
"learning_rate": 0.00044071586652913767,
|
| 632 |
+
"loss": 0.2035,
|
| 633 |
+
"step": 1425
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 1.1894995898277276,
|
| 637 |
+
"grad_norm": 0.11690429598093033,
|
| 638 |
+
"learning_rate": 0.0004229851298760915,
|
| 639 |
+
"loss": 0.2135,
|
| 640 |
+
"step": 1450
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 1.1894995898277276,
|
| 644 |
+
"eval_loss": 0.2264111191034317,
|
| 645 |
+
"eval_runtime": 20.7449,
|
| 646 |
+
"eval_samples_per_second": 48.205,
|
| 647 |
+
"eval_steps_per_second": 0.771,
|
| 648 |
+
"step": 1450
|
| 649 |
+
},
|
| 650 |
+
{
|
| 651 |
+
"epoch": 1.2100082034454471,
|
| 652 |
+
"grad_norm": 0.105661541223526,
|
| 653 |
+
"learning_rate": 0.0004053530744919749,
|
| 654 |
+
"loss": 0.2125,
|
| 655 |
+
"step": 1475
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 1.2305168170631666,
|
| 659 |
+
"grad_norm": 0.11686920374631882,
|
| 660 |
+
"learning_rate": 0.0003878422928137597,
|
| 661 |
+
"loss": 0.2158,
|
| 662 |
+
"step": 1500
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 1.2305168170631666,
|
| 666 |
+
"eval_loss": 0.22509609162807465,
|
| 667 |
+
"eval_runtime": 20.7795,
|
| 668 |
+
"eval_samples_per_second": 48.124,
|
| 669 |
+
"eval_steps_per_second": 0.77,
|
| 670 |
+
"step": 1500
|
| 671 |
+
},
|
| 672 |
+
{
|
| 673 |
+
"epoch": 1.251025430680886,
|
| 674 |
+
"grad_norm": 0.11210618168115616,
|
| 675 |
+
"learning_rate": 0.0003704752218870861,
|
| 676 |
+
"loss": 0.2205,
|
| 677 |
+
"step": 1525
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 1.2715340442986054,
|
| 681 |
+
"grad_norm": 0.10161525756120682,
|
| 682 |
+
"learning_rate": 0.00035327411461706025,
|
| 683 |
+
"loss": 0.203,
|
| 684 |
+
"step": 1550
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 1.2715340442986054,
|
| 688 |
+
"eval_loss": 0.22456735372543335,
|
| 689 |
+
"eval_runtime": 20.7538,
|
| 690 |
+
"eval_samples_per_second": 48.184,
|
| 691 |
+
"eval_steps_per_second": 0.771,
|
| 692 |
+
"step": 1550
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 1.2920426579163249,
|
| 696 |
+
"grad_norm": 0.11276591569185257,
|
| 697 |
+
"learning_rate": 0.00033626101125499555,
|
| 698 |
+
"loss": 0.2126,
|
| 699 |
+
"step": 1575
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 1.3125512715340442,
|
| 703 |
+
"grad_norm": 0.10445020347833633,
|
| 704 |
+
"learning_rate": 0.0003194577111576333,
|
| 705 |
+
"loss": 0.2081,
|
| 706 |
+
"step": 1600
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 1.3125512715340442,
|
| 710 |
+
"eval_loss": 0.22300216555595398,
|
| 711 |
+
"eval_runtime": 20.7911,
|
| 712 |
+
"eval_samples_per_second": 48.098,
|
| 713 |
+
"eval_steps_per_second": 0.77,
|
| 714 |
+
"step": 1600
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 1.3330598851517639,
|
| 718 |
+
"grad_norm": 0.09758764505386353,
|
| 719 |
+
"learning_rate": 0.00030288574485502756,
|
| 720 |
+
"loss": 0.2076,
|
| 721 |
+
"step": 1625
|
| 722 |
+
},
|
| 723 |
+
{
|
| 724 |
+
"epoch": 1.3535684987694832,
|
| 725 |
+
"grad_norm": 0.10864421725273132,
|
| 726 |
+
"learning_rate": 0.00028656634646288565,
|
| 727 |
+
"loss": 0.2085,
|
| 728 |
+
"step": 1650
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 1.3535684987694832,
|
| 732 |
+
"eval_loss": 0.2210860252380371,
|
| 733 |
+
"eval_runtime": 20.7702,
|
| 734 |
+
"eval_samples_per_second": 48.146,
|
| 735 |
+
"eval_steps_per_second": 0.77,
|
| 736 |
+
"step": 1650
|
| 737 |
+
},
|
| 738 |
+
{
|
| 739 |
+
"epoch": 1.3740771123872026,
|
| 740 |
+
"grad_norm": 0.12502917647361755,
|
| 741 |
+
"learning_rate": 0.00027052042647471254,
|
| 742 |
+
"loss": 0.1977,
|
| 743 |
+
"step": 1675
|
| 744 |
+
},
|
| 745 |
+
{
|
| 746 |
+
"epoch": 1.3945857260049221,
|
| 747 |
+
"grad_norm": 0.10719209909439087,
|
| 748 |
+
"learning_rate": 0.0002547685449686206,
|
| 749 |
+
"loss": 0.206,
|
| 750 |
+
"step": 1700
|
| 751 |
+
},
|
| 752 |
+
{
|
| 753 |
+
"epoch": 1.3945857260049221,
|
| 754 |
+
"eval_loss": 0.21972452104091644,
|
| 755 |
+
"eval_runtime": 20.7618,
|
| 756 |
+
"eval_samples_per_second": 48.165,
|
| 757 |
+
"eval_steps_per_second": 0.771,
|
| 758 |
+
"step": 1700
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 1.4150943396226414,
|
| 762 |
+
"grad_norm": 0.13277575373649597,
|
| 763 |
+
"learning_rate": 0.0002393308852631373,
|
| 764 |
+
"loss": 0.2175,
|
| 765 |
+
"step": 1725
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 1.435602953240361,
|
| 769 |
+
"grad_norm": 0.10963447391986847,
|
| 770 |
+
"learning_rate": 0.0002242272280557645,
|
| 771 |
+
"loss": 0.2119,
|
| 772 |
+
"step": 1750
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"epoch": 1.435602953240361,
|
| 776 |
+
"eval_loss": 0.21942101418972015,
|
| 777 |
+
"eval_runtime": 20.8,
|
| 778 |
+
"eval_samples_per_second": 48.077,
|
| 779 |
+
"eval_steps_per_second": 0.769,
|
| 780 |
+
"step": 1750
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 1.4561115668580804,
|
| 784 |
+
"grad_norm": 0.1101914569735527,
|
| 785 |
+
"learning_rate": 0.0002094769260774262,
|
| 786 |
+
"loss": 0.2077,
|
| 787 |
+
"step": 1775
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 1.4766201804758,
|
| 791 |
+
"grad_norm": 0.1031966507434845,
|
| 792 |
+
"learning_rate": 0.00019509887929528458,
|
| 793 |
+
"loss": 0.2116,
|
| 794 |
+
"step": 1800
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 1.4766201804758,
|
| 798 |
+
"eval_loss": 0.21845056116580963,
|
| 799 |
+
"eval_runtime": 20.795,
|
| 800 |
+
"eval_samples_per_second": 48.089,
|
| 801 |
+
"eval_steps_per_second": 0.769,
|
| 802 |
+
"step": 1800
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 1.4971287940935194,
|
| 806 |
+
"grad_norm": 0.13558058440685272,
|
| 807 |
+
"learning_rate": 0.0001811115106956918,
|
| 808 |
+
"loss": 0.2028,
|
| 809 |
+
"step": 1825
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 1.5176374077112387,
|
| 813 |
+
"grad_norm": 0.11538273841142654,
|
| 814 |
+
"learning_rate": 0.00016753274267831115,
|
| 815 |
+
"loss": 0.2119,
|
| 816 |
+
"step": 1850
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"epoch": 1.5176374077112387,
|
| 820 |
+
"eval_loss": 0.21733731031417847,
|
| 821 |
+
"eval_runtime": 20.7805,
|
| 822 |
+
"eval_samples_per_second": 48.122,
|
| 823 |
+
"eval_steps_per_second": 0.77,
|
| 824 |
+
"step": 1850
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 1.5381460213289582,
|
| 828 |
+
"grad_norm": 0.13104985654354095,
|
| 829 |
+
"learning_rate": 0.00015437997409165478,
|
| 830 |
+
"loss": 0.2022,
|
| 831 |
+
"step": 1875
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 1.5586546349466777,
|
| 835 |
+
"grad_norm": 0.11488507688045502,
|
| 836 |
+
"learning_rate": 0.00014167005793946035,
|
| 837 |
+
"loss": 0.2015,
|
| 838 |
+
"step": 1900
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 1.5586546349466777,
|
| 842 |
+
"eval_loss": 0.21705850958824158,
|
| 843 |
+
"eval_runtime": 20.7621,
|
| 844 |
+
"eval_samples_per_second": 48.165,
|
| 845 |
+
"eval_steps_per_second": 0.771,
|
| 846 |
+
"step": 1900
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 1.579163248564397,
|
| 850 |
+
"grad_norm": 0.11999308317899704,
|
| 851 |
+
"learning_rate": 0.00012941927978647527,
|
| 852 |
+
"loss": 0.2038,
|
| 853 |
+
"step": 1925
|
| 854 |
+
},
|
| 855 |
+
{
|
| 856 |
+
"epoch": 1.5996718621821167,
|
| 857 |
+
"grad_norm": 0.1013152152299881,
|
| 858 |
+
"learning_rate": 0.00011764333689131385,
|
| 859 |
+
"loss": 0.2095,
|
| 860 |
+
"step": 1950
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 1.5996718621821167,
|
| 864 |
+
"eval_loss": 0.2159809172153473,
|
| 865 |
+
"eval_runtime": 20.8631,
|
| 866 |
+
"eval_samples_per_second": 47.932,
|
| 867 |
+
"eval_steps_per_second": 0.767,
|
| 868 |
+
"step": 1950
|
| 869 |
+
},
|
| 870 |
+
{
|
| 871 |
+
"epoch": 1.620180475799836,
|
| 872 |
+
"grad_norm": 0.10026060789823532,
|
| 873 |
+
"learning_rate": 0.00010635731809312993,
|
| 874 |
+
"loss": 0.2058,
|
| 875 |
+
"step": 1975
|
| 876 |
+
},
|
| 877 |
+
{
|
| 878 |
+
"epoch": 1.6406890894175554,
|
| 879 |
+
"grad_norm": 0.11767016351222992,
|
| 880 |
+
"learning_rate": 9.557568447787201e-05,
|
| 881 |
+
"loss": 0.2056,
|
| 882 |
+
"step": 2000
|
| 883 |
+
},
|
| 884 |
+
{
|
| 885 |
+
"epoch": 1.6406890894175554,
|
| 886 |
+
"eval_loss": 0.2152351588010788,
|
| 887 |
+
"eval_runtime": 20.9101,
|
| 888 |
+
"eval_samples_per_second": 47.824,
|
| 889 |
+
"eval_steps_per_second": 0.765,
|
| 890 |
+
"step": 2000
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 1.661197703035275,
|
| 894 |
+
"grad_norm": 0.10698919743299484,
|
| 895 |
+
"learning_rate": 8.531225084889654e-05,
|
| 896 |
+
"loss": 0.1993,
|
| 897 |
+
"step": 2025
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 1.6817063166529942,
|
| 901 |
+
"grad_norm": 0.13556532561779022,
|
| 902 |
+
"learning_rate": 7.558016802568091e-05,
|
| 903 |
+
"loss": 0.2021,
|
| 904 |
+
"step": 2050
|
| 905 |
+
},
|
| 906 |
+
{
|
| 907 |
+
"epoch": 1.6817063166529942,
|
| 908 |
+
"eval_loss": 0.21463391184806824,
|
| 909 |
+
"eval_runtime": 20.821,
|
| 910 |
+
"eval_samples_per_second": 48.029,
|
| 911 |
+
"eval_steps_per_second": 0.768,
|
| 912 |
+
"step": 2050
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 1.7022149302707137,
|
| 916 |
+
"grad_norm": 0.1280520260334015,
|
| 917 |
+
"learning_rate": 6.639190599331746e-05,
|
| 918 |
+
"loss": 0.2029,
|
| 919 |
+
"step": 2075
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 1.7227235438884332,
|
| 923 |
+
"grad_norm": 0.11126288026571274,
|
| 924 |
+
"learning_rate": 5.775923792437865e-05,
|
| 925 |
+
"loss": 0.1991,
|
| 926 |
+
"step": 2100
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 1.7227235438884332,
|
| 930 |
+
"eval_loss": 0.21391206979751587,
|
| 931 |
+
"eval_runtime": 20.7954,
|
| 932 |
+
"eval_samples_per_second": 48.088,
|
| 933 |
+
"eval_steps_per_second": 0.769,
|
| 934 |
+
"step": 2100
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 1.7432321575061525,
|
| 938 |
+
"grad_norm": 0.11311797052621841,
|
| 939 |
+
"learning_rate": 4.9693225093627616e-05,
|
| 940 |
+
"loss": 0.1992,
|
| 941 |
+
"step": 2125
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 1.7637407711238722,
|
| 945 |
+
"grad_norm": 0.11362408846616745,
|
| 946 |
+
"learning_rate": 4.220420270490294e-05,
|
| 947 |
+
"loss": 0.1932,
|
| 948 |
+
"step": 2150
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 1.7637407711238722,
|
| 952 |
+
"eval_loss": 0.21360714733600616,
|
| 953 |
+
"eval_runtime": 20.7865,
|
| 954 |
+
"eval_samples_per_second": 48.108,
|
| 955 |
+
"eval_steps_per_second": 0.77,
|
| 956 |
+
"step": 2150
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 1.7842493847415914,
|
| 960 |
+
"grad_norm": 0.1285392940044403,
|
| 961 |
+
"learning_rate": 3.530176664833834e-05,
|
| 962 |
+
"loss": 0.196,
|
| 963 |
+
"step": 2175
|
| 964 |
+
},
|
| 965 |
+
{
|
| 966 |
+
"epoch": 1.804757998359311,
|
| 967 |
+
"grad_norm": 0.12596482038497925,
|
| 968 |
+
"learning_rate": 2.8994761204884756e-05,
|
| 969 |
+
"loss": 0.2048,
|
| 970 |
+
"step": 2200
|
| 971 |
+
},
|
| 972 |
+
{
|
| 973 |
+
"epoch": 1.804757998359311,
|
| 974 |
+
"eval_loss": 0.21343722939491272,
|
| 975 |
+
"eval_runtime": 20.8106,
|
| 976 |
+
"eval_samples_per_second": 48.052,
|
| 977 |
+
"eval_steps_per_second": 0.769,
|
| 978 |
+
"step": 2200
|
| 979 |
+
},
|
| 980 |
+
{
|
| 981 |
+
"epoch": 1.8252666119770304,
|
| 982 |
+
"grad_norm": 0.12076670676469803,
|
| 983 |
+
"learning_rate": 2.329126771388995e-05,
|
| 984 |
+
"loss": 0.1957,
|
| 985 |
+
"step": 2225
|
| 986 |
+
},
|
| 987 |
+
{
|
| 988 |
+
"epoch": 1.8457752255947497,
|
| 989 |
+
"grad_norm": 0.10855985432863235,
|
| 990 |
+
"learning_rate": 1.8198594218256815e-05,
|
| 991 |
+
"loss": 0.1971,
|
| 992 |
+
"step": 2250
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 1.8457752255947497,
|
| 996 |
+
"eval_loss": 0.21339941024780273,
|
| 997 |
+
"eval_runtime": 20.7802,
|
| 998 |
+
"eval_samples_per_second": 48.123,
|
| 999 |
+
"eval_steps_per_second": 0.77,
|
| 1000 |
+
"step": 2250
|
| 1001 |
+
},
|
| 1002 |
+
{
|
| 1003 |
+
"epoch": 1.8662838392124692,
|
| 1004 |
+
"grad_norm": 0.12243843078613281,
|
| 1005 |
+
"learning_rate": 1.3723266100447052e-05,
|
| 1006 |
+
"loss": 0.2006,
|
| 1007 |
+
"step": 2275
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"epoch": 1.8867924528301887,
|
| 1011 |
+
"grad_norm": 0.14360643923282623,
|
| 1012 |
+
"learning_rate": 9.871017721329201e-06,
|
| 1013 |
+
"loss": 0.2083,
|
| 1014 |
+
"step": 2300
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"epoch": 1.8867924528301887,
|
| 1018 |
+
"eval_loss": 0.2131664752960205,
|
| 1019 |
+
"eval_runtime": 20.7557,
|
| 1020 |
+
"eval_samples_per_second": 48.18,
|
| 1021 |
+
"eval_steps_per_second": 0.771,
|
| 1022 |
+
"step": 2300
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 1.907301066447908,
|
| 1026 |
+
"grad_norm": 0.11693233996629715,
|
| 1027 |
+
"learning_rate": 6.646785072584871e-06,
|
| 1028 |
+
"loss": 0.2031,
|
| 1029 |
+
"step": 2325
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 1.9278096800656277,
|
| 1033 |
+
"grad_norm": 0.10460177809000015,
|
| 1034 |
+
"learning_rate": 4.054699452086641e-06,
|
| 1035 |
+
"loss": 0.1966,
|
| 1036 |
+
"step": 2350
|
| 1037 |
+
},
|
| 1038 |
+
{
|
| 1039 |
+
"epoch": 1.9278096800656277,
|
| 1040 |
+
"eval_loss": 0.21305271983146667,
|
| 1041 |
+
"eval_runtime": 20.8057,
|
| 1042 |
+
"eval_samples_per_second": 48.064,
|
| 1043 |
+
"eval_steps_per_second": 0.769,
|
| 1044 |
+
"step": 2350
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 1.948318293683347,
|
| 1048 |
+
"grad_norm": 0.1053781509399414,
|
| 1049 |
+
"learning_rate": 2.0980821703527886e-06,
|
| 1050 |
+
"loss": 0.211,
|
| 1051 |
+
"step": 2375
|
| 1052 |
+
},
|
| 1053 |
+
{
|
| 1054 |
+
"epoch": 1.9688269073010665,
|
| 1055 |
+
"grad_norm": 0.1103985607624054,
|
| 1056 |
+
"learning_rate": 7.794402948607671e-07,
|
| 1057 |
+
"loss": 0.2071,
|
| 1058 |
+
"step": 2400
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 1.9688269073010665,
|
| 1062 |
+
"eval_loss": 0.21304111182689667,
|
| 1063 |
+
"eval_runtime": 20.8063,
|
| 1064 |
+
"eval_samples_per_second": 48.062,
|
| 1065 |
+
"eval_steps_per_second": 0.769,
|
| 1066 |
+
"step": 2400
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 1.989335520918786,
|
| 1070 |
+
"grad_norm": 0.13697576522827148,
|
| 1071 |
+
"learning_rate": 1.0046343767294853e-07,
|
| 1072 |
+
"loss": 0.1965,
|
| 1073 |
+
"step": 2425
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 2.0,
|
| 1077 |
+
"step": 2438,
|
| 1078 |
+
"total_flos": 1.58523627405312e+18,
|
| 1079 |
+
"train_loss": 0.24706831303065288,
|
| 1080 |
+
"train_runtime": 3792.5555,
|
| 1081 |
+
"train_samples_per_second": 20.567,
|
| 1082 |
+
"train_steps_per_second": 0.643
|
| 1083 |
+
}
|
| 1084 |
+
],
|
| 1085 |
+
"logging_steps": 25,
|
| 1086 |
+
"max_steps": 2438,
|
| 1087 |
+
"num_input_tokens_seen": 0,
|
| 1088 |
+
"num_train_epochs": 2,
|
| 1089 |
+
"save_steps": 500,
|
| 1090 |
+
"stateful_callbacks": {
|
| 1091 |
+
"TrainerControl": {
|
| 1092 |
+
"args": {
|
| 1093 |
+
"should_epoch_stop": false,
|
| 1094 |
+
"should_evaluate": false,
|
| 1095 |
+
"should_log": false,
|
| 1096 |
+
"should_save": true,
|
| 1097 |
+
"should_training_stop": true
|
| 1098 |
+
},
|
| 1099 |
+
"attributes": {}
|
| 1100 |
+
}
|
| 1101 |
+
},
|
| 1102 |
+
"total_flos": 1.58523627405312e+18,
|
| 1103 |
+
"train_batch_size": 32,
|
| 1104 |
+
"trial_name": null,
|
| 1105 |
+
"trial_params": null
|
| 1106 |
+
}
|
nl_tasks/run_all/run_exnr11/ft/adapter_config.json
ADDED
|
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"T": 1.0,
|
| 3 |
+
"base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
|
| 4 |
+
"bias": "none",
|
| 5 |
+
"inference_mode": false,
|
| 6 |
+
"layers_to_transform": null,
|
| 7 |
+
"modules_to_save": null,
|
| 8 |
+
"num_rotations": 2,
|
| 9 |
+
"peft_type": "ROTATION",
|
| 10 |
+
"r": 8,
|
| 11 |
+
"revision": null,
|
| 12 |
+
"target_modules": [
|
| 13 |
+
"v_proj",
|
| 14 |
+
"q_proj"
|
| 15 |
+
],
|
| 16 |
+
"target_modules_to_skip": null,
|
| 17 |
+
"task_type": "CAUSAL_LM"
|
| 18 |
+
}
|
nl_tasks/run_all/run_exnr11/ft/added_tokens.json
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"[PAD]": 32000
|
| 3 |
+
}
|
nl_tasks/run_all/run_exnr11/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/run_all/run_exnr11/ft/tokenizer.json
ADDED
|
The diff for this file is too large to render.
See raw diff
|
|
|
nl_tasks/run_all/run_exnr11/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/run_all/run_exnr11/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/run_all/run_exnr11/ft/training_args.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:3c3569eea58fa21482a4bc85e514a16f90ed59354c7c0b80262bb053b7c12c9c
|
| 3 |
+
size 6481
|
nl_tasks/run_all/run_exnr11/ft2/adapter_config.json
ADDED
|
@@ -0,0 +1,18 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"T": 1.0,
|
| 3 |
+
"base_model_name_or_path": "meta-llama/Llama-2-7b-hf",
|
| 4 |
+
"bias": "none",
|
| 5 |
+
"inference_mode": true,
|
| 6 |
+
"layers_to_transform": null,
|
| 7 |
+
"modules_to_save": null,
|
| 8 |
+
"num_rotations": 2,
|
| 9 |
+
"peft_type": "ROTATION",
|
| 10 |
+
"r": 8,
|
| 11 |
+
"revision": null,
|
| 12 |
+
"target_modules": [
|
| 13 |
+
"v_proj",
|
| 14 |
+
"q_proj"
|
| 15 |
+
],
|
| 16 |
+
"target_modules_to_skip": null,
|
| 17 |
+
"task_type": "CAUSAL_LM"
|
| 18 |
+
}
|
nl_tasks/run_all/run_exnr11/ft2/adapter_model.bin
ADDED
|
@@ -0,0 +1,3 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
version https://git-lfs.github.com/spec/v1
|
| 2 |
+
oid sha256:a354ea299786492a94c791601240dfb633666078b15a8ebdf0181221f18adb01
|
| 3 |
+
size 33602659
|
nl_tasks/run_all/run_exnr11/trainer_state.json
ADDED
|
@@ -0,0 +1,1106 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 2.0,
|
| 6 |
+
"eval_steps": 50,
|
| 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": 0.22536601126194,
|
| 15 |
+
"learning_rate": 9.836065573770491e-05,
|
| 16 |
+
"loss": 0.6776,
|
| 17 |
+
"step": 25
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.04101722723543889,
|
| 21 |
+
"grad_norm": 0.18631383776664734,
|
| 22 |
+
"learning_rate": 0.00020081967213114754,
|
| 23 |
+
"loss": 0.4154,
|
| 24 |
+
"step": 50
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.04101722723543889,
|
| 28 |
+
"eval_loss": 0.3722397983074188,
|
| 29 |
+
"eval_runtime": 19.8213,
|
| 30 |
+
"eval_samples_per_second": 50.451,
|
| 31 |
+
"eval_steps_per_second": 0.807,
|
| 32 |
+
"step": 50
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.06152584085315833,
|
| 36 |
+
"grad_norm": 0.17987015843391418,
|
| 37 |
+
"learning_rate": 0.0003032786885245902,
|
| 38 |
+
"loss": 0.3704,
|
| 39 |
+
"step": 75
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.08203445447087777,
|
| 43 |
+
"grad_norm": 0.19490021467208862,
|
| 44 |
+
"learning_rate": 0.0004057377049180328,
|
| 45 |
+
"loss": 0.3506,
|
| 46 |
+
"step": 100
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"epoch": 0.08203445447087777,
|
| 50 |
+
"eval_loss": 0.3291980028152466,
|
| 51 |
+
"eval_runtime": 19.4889,
|
| 52 |
+
"eval_samples_per_second": 51.311,
|
| 53 |
+
"eval_steps_per_second": 0.821,
|
| 54 |
+
"step": 100
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.10254306808859721,
|
| 58 |
+
"grad_norm": 1.9897618293762207,
|
| 59 |
+
"learning_rate": 0.0005081967213114754,
|
| 60 |
+
"loss": 0.3309,
|
| 61 |
+
"step": 125
|
| 62 |
+
},
|
| 63 |
+
{
|
| 64 |
+
"epoch": 0.12305168170631665,
|
| 65 |
+
"grad_norm": 0.17177057266235352,
|
| 66 |
+
"learning_rate": 0.000610655737704918,
|
| 67 |
+
"loss": 0.3153,
|
| 68 |
+
"step": 150
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.12305168170631665,
|
| 72 |
+
"eval_loss": 0.31367334723472595,
|
| 73 |
+
"eval_runtime": 19.4853,
|
| 74 |
+
"eval_samples_per_second": 51.321,
|
| 75 |
+
"eval_steps_per_second": 0.821,
|
| 76 |
+
"step": 150
|
| 77 |
+
},
|
| 78 |
+
{
|
| 79 |
+
"epoch": 0.1435602953240361,
|
| 80 |
+
"grad_norm": 0.22781579196453094,
|
| 81 |
+
"learning_rate": 0.0007131147540983607,
|
| 82 |
+
"loss": 0.314,
|
| 83 |
+
"step": 175
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"epoch": 0.16406890894175555,
|
| 87 |
+
"grad_norm": 0.3163987994194031,
|
| 88 |
+
"learning_rate": 0.0008155737704918033,
|
| 89 |
+
"loss": 0.3208,
|
| 90 |
+
"step": 200
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 0.16406890894175555,
|
| 94 |
+
"eval_loss": 0.31177034974098206,
|
| 95 |
+
"eval_runtime": 19.4994,
|
| 96 |
+
"eval_samples_per_second": 51.284,
|
| 97 |
+
"eval_steps_per_second": 0.821,
|
| 98 |
+
"step": 200
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.184577522559475,
|
| 102 |
+
"grad_norm": 0.32001861929893494,
|
| 103 |
+
"learning_rate": 0.0009180327868852459,
|
| 104 |
+
"loss": 0.3285,
|
| 105 |
+
"step": 225
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 0.20508613617719443,
|
| 109 |
+
"grad_norm": 0.2669612169265747,
|
| 110 |
+
"learning_rate": 0.0009999871854116063,
|
| 111 |
+
"loss": 0.3195,
|
| 112 |
+
"step": 250
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 0.20508613617719443,
|
| 116 |
+
"eval_loss": 0.30614417791366577,
|
| 117 |
+
"eval_runtime": 19.4864,
|
| 118 |
+
"eval_samples_per_second": 51.318,
|
| 119 |
+
"eval_steps_per_second": 0.821,
|
| 120 |
+
"step": 250
|
| 121 |
+
},
|
| 122 |
+
{
|
| 123 |
+
"epoch": 0.22559474979491387,
|
| 124 |
+
"grad_norm": 0.35766294598579407,
|
| 125 |
+
"learning_rate": 0.0009995387437838027,
|
| 126 |
+
"loss": 0.3164,
|
| 127 |
+
"step": 275
|
| 128 |
+
},
|
| 129 |
+
{
|
| 130 |
+
"epoch": 0.2461033634126333,
|
| 131 |
+
"grad_norm": 0.2564176619052887,
|
| 132 |
+
"learning_rate": 0.000998450229439693,
|
| 133 |
+
"loss": 0.3137,
|
| 134 |
+
"step": 300
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.2461033634126333,
|
| 138 |
+
"eval_loss": 0.3005053699016571,
|
| 139 |
+
"eval_runtime": 19.4856,
|
| 140 |
+
"eval_samples_per_second": 51.32,
|
| 141 |
+
"eval_steps_per_second": 0.821,
|
| 142 |
+
"step": 300
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.2666119770303528,
|
| 146 |
+
"grad_norm": 0.21307945251464844,
|
| 147 |
+
"learning_rate": 0.000996723037122612,
|
| 148 |
+
"loss": 0.2997,
|
| 149 |
+
"step": 325
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.2871205906480722,
|
| 153 |
+
"grad_norm": 0.24557501077651978,
|
| 154 |
+
"learning_rate": 0.0009943593799315263,
|
| 155 |
+
"loss": 0.3067,
|
| 156 |
+
"step": 350
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.2871205906480722,
|
| 160 |
+
"eval_loss": 0.29183846712112427,
|
| 161 |
+
"eval_runtime": 19.4754,
|
| 162 |
+
"eval_samples_per_second": 51.347,
|
| 163 |
+
"eval_steps_per_second": 0.822,
|
| 164 |
+
"step": 350
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.30762920426579166,
|
| 168 |
+
"grad_norm": 0.22399385273456573,
|
| 169 |
+
"learning_rate": 0.0009913622864853324,
|
| 170 |
+
"loss": 0.2946,
|
| 171 |
+
"step": 375
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.3281378178835111,
|
| 175 |
+
"grad_norm": 0.28306031227111816,
|
| 176 |
+
"learning_rate": 0.0009877355970422024,
|
| 177 |
+
"loss": 0.3047,
|
| 178 |
+
"step": 400
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.3281378178835111,
|
| 182 |
+
"eval_loss": 0.2847941815853119,
|
| 183 |
+
"eval_runtime": 19.4709,
|
| 184 |
+
"eval_samples_per_second": 51.359,
|
| 185 |
+
"eval_steps_per_second": 0.822,
|
| 186 |
+
"step": 400
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 0.34864643150123054,
|
| 190 |
+
"grad_norm": 0.18770238757133484,
|
| 191 |
+
"learning_rate": 0.0009834839585789559,
|
| 192 |
+
"loss": 0.2983,
|
| 193 |
+
"step": 425
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 0.36915504511895,
|
| 197 |
+
"grad_norm": 0.2074132114648819,
|
| 198 |
+
"learning_rate": 0.000978612818836762,
|
| 199 |
+
"loss": 0.2957,
|
| 200 |
+
"step": 450
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 0.36915504511895,
|
| 204 |
+
"eval_loss": 0.2784390449523926,
|
| 205 |
+
"eval_runtime": 19.4671,
|
| 206 |
+
"eval_samples_per_second": 51.369,
|
| 207 |
+
"eval_steps_per_second": 0.822,
|
| 208 |
+
"step": 450
|
| 209 |
+
},
|
| 210 |
+
{
|
| 211 |
+
"epoch": 0.3896636587366694,
|
| 212 |
+
"grad_norm": 0.1998339742422104,
|
| 213 |
+
"learning_rate": 0.0009731284193407981,
|
| 214 |
+
"loss": 0.2922,
|
| 215 |
+
"step": 475
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 0.41017227235438886,
|
| 219 |
+
"grad_norm": 0.19609296321868896,
|
| 220 |
+
"learning_rate": 0.0009670377874028117,
|
| 221 |
+
"loss": 0.2838,
|
| 222 |
+
"step": 500
|
| 223 |
+
},
|
| 224 |
+
{
|
| 225 |
+
"epoch": 0.41017227235438886,
|
| 226 |
+
"eval_loss": 0.2710179090499878,
|
| 227 |
+
"eval_runtime": 19.4596,
|
| 228 |
+
"eval_samples_per_second": 51.389,
|
| 229 |
+
"eval_steps_per_second": 0.822,
|
| 230 |
+
"step": 500
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 0.4306808859721083,
|
| 234 |
+
"grad_norm": 0.22822904586791992,
|
| 235 |
+
"learning_rate": 0.0009603487271168336,
|
| 236 |
+
"loss": 0.2859,
|
| 237 |
+
"step": 525
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 0.45118949958982774,
|
| 241 |
+
"grad_norm": 0.21479056775569916,
|
| 242 |
+
"learning_rate": 0.0009530698093595781,
|
| 243 |
+
"loss": 0.2805,
|
| 244 |
+
"step": 550
|
| 245 |
+
},
|
| 246 |
+
{
|
| 247 |
+
"epoch": 0.45118949958982774,
|
| 248 |
+
"eval_loss": 0.2640228867530823,
|
| 249 |
+
"eval_runtime": 19.4506,
|
| 250 |
+
"eval_samples_per_second": 51.412,
|
| 251 |
+
"eval_steps_per_second": 0.823,
|
| 252 |
+
"step": 550
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 0.4716981132075472,
|
| 256 |
+
"grad_norm": 0.21766699850559235,
|
| 257 |
+
"learning_rate": 0.0009452103608083418,
|
| 258 |
+
"loss": 0.2659,
|
| 259 |
+
"step": 575
|
| 260 |
+
},
|
| 261 |
+
{
|
| 262 |
+
"epoch": 0.4922067268252666,
|
| 263 |
+
"grad_norm": 0.21340589225292206,
|
| 264 |
+
"learning_rate": 0.0009367804519904775,
|
| 265 |
+
"loss": 0.2761,
|
| 266 |
+
"step": 600
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 0.4922067268252666,
|
| 270 |
+
"eval_loss": 0.26190468668937683,
|
| 271 |
+
"eval_runtime": 19.4654,
|
| 272 |
+
"eval_samples_per_second": 51.373,
|
| 273 |
+
"eval_steps_per_second": 0.822,
|
| 274 |
+
"step": 600
|
| 275 |
+
},
|
| 276 |
+
{
|
| 277 |
+
"epoch": 0.5127153404429861,
|
| 278 |
+
"grad_norm": 0.1762474924325943,
|
| 279 |
+
"learning_rate": 0.0009277908843797492,
|
| 280 |
+
"loss": 0.2622,
|
| 281 |
+
"step": 625
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 0.5332239540607056,
|
| 285 |
+
"grad_norm": 0.19957980513572693,
|
| 286 |
+
"learning_rate": 0.0009182531765561084,
|
| 287 |
+
"loss": 0.2618,
|
| 288 |
+
"step": 650
|
| 289 |
+
},
|
| 290 |
+
{
|
| 291 |
+
"epoch": 0.5332239540607056,
|
| 292 |
+
"eval_loss": 0.2553676664829254,
|
| 293 |
+
"eval_runtime": 19.4847,
|
| 294 |
+
"eval_samples_per_second": 51.322,
|
| 295 |
+
"eval_steps_per_second": 0.821,
|
| 296 |
+
"step": 650
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.5537325676784249,
|
| 300 |
+
"grad_norm": 0.1787998378276825,
|
| 301 |
+
"learning_rate": 0.0009081795494466201,
|
| 302 |
+
"loss": 0.261,
|
| 303 |
+
"step": 675
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.5742411812961444,
|
| 307 |
+
"grad_norm": 0.1660076081752777,
|
| 308 |
+
"learning_rate": 0.0008975829106664539,
|
| 309 |
+
"loss": 0.2534,
|
| 310 |
+
"step": 700
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.5742411812961444,
|
| 314 |
+
"eval_loss": 0.2539336383342743,
|
| 315 |
+
"eval_runtime": 19.4661,
|
| 316 |
+
"eval_samples_per_second": 51.371,
|
| 317 |
+
"eval_steps_per_second": 0.822,
|
| 318 |
+
"step": 700
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.5947497949138638,
|
| 322 |
+
"grad_norm": 0.1471703201532364,
|
| 323 |
+
"learning_rate": 0.0008864768379800017,
|
| 324 |
+
"loss": 0.2508,
|
| 325 |
+
"step": 725
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.6152584085315833,
|
| 329 |
+
"grad_norm": 0.16628079116344452,
|
| 330 |
+
"learning_rate": 0.0008748755619033153,
|
| 331 |
+
"loss": 0.2608,
|
| 332 |
+
"step": 750
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.6152584085315833,
|
| 336 |
+
"eval_loss": 0.25101426243782043,
|
| 337 |
+
"eval_runtime": 19.4998,
|
| 338 |
+
"eval_samples_per_second": 51.283,
|
| 339 |
+
"eval_steps_per_second": 0.821,
|
| 340 |
+
"step": 750
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 0.6357670221493027,
|
| 344 |
+
"grad_norm": 0.180254727602005,
|
| 345 |
+
"learning_rate": 0.000862793947470155,
|
| 346 |
+
"loss": 0.2647,
|
| 347 |
+
"step": 775
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 0.6562756357670222,
|
| 351 |
+
"grad_norm": 0.16186201572418213,
|
| 352 |
+
"learning_rate": 0.0008502474751850142,
|
| 353 |
+
"loss": 0.2538,
|
| 354 |
+
"step": 800
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.6562756357670222,
|
| 358 |
+
"eval_loss": 0.24444915354251862,
|
| 359 |
+
"eval_runtime": 19.4781,
|
| 360 |
+
"eval_samples_per_second": 51.34,
|
| 361 |
+
"eval_steps_per_second": 0.821,
|
| 362 |
+
"step": 800
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.6767842493847416,
|
| 366 |
+
"grad_norm": 0.1586069017648697,
|
| 367 |
+
"learning_rate": 0.0008372522211875224,
|
| 368 |
+
"loss": 0.248,
|
| 369 |
+
"step": 825
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.6972928630024611,
|
| 373 |
+
"grad_norm": 0.1476527601480484,
|
| 374 |
+
"learning_rate": 0.0008238248366536473,
|
| 375 |
+
"loss": 0.2611,
|
| 376 |
+
"step": 850
|
| 377 |
+
},
|
| 378 |
+
{
|
| 379 |
+
"epoch": 0.6972928630024611,
|
| 380 |
+
"eval_loss": 0.24326753616333008,
|
| 381 |
+
"eval_runtime": 19.4366,
|
| 382 |
+
"eval_samples_per_second": 51.449,
|
| 383 |
+
"eval_steps_per_second": 0.823,
|
| 384 |
+
"step": 850
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.7178014766201805,
|
| 388 |
+
"grad_norm": 0.11883819103240967,
|
| 389 |
+
"learning_rate": 0.0008099825264600842,
|
| 390 |
+
"loss": 0.255,
|
| 391 |
+
"step": 875
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 0.7383100902379,
|
| 395 |
+
"grad_norm": 0.1317686289548874,
|
| 396 |
+
"learning_rate": 0.0007957430271391761,
|
| 397 |
+
"loss": 0.2468,
|
| 398 |
+
"step": 900
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 0.7383100902379,
|
| 402 |
+
"eval_loss": 0.2392330914735794,
|
| 403 |
+
"eval_runtime": 19.493,
|
| 404 |
+
"eval_samples_per_second": 51.301,
|
| 405 |
+
"eval_steps_per_second": 0.821,
|
| 406 |
+
"step": 900
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 0.7588187038556193,
|
| 410 |
+
"grad_norm": 0.1467055082321167,
|
| 411 |
+
"learning_rate": 0.0007811245841526062,
|
| 412 |
+
"loss": 0.25,
|
| 413 |
+
"step": 925
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 0.7793273174733388,
|
| 417 |
+
"grad_norm": 0.15291206538677216,
|
| 418 |
+
"learning_rate": 0.0007661459285129879,
|
| 419 |
+
"loss": 0.2334,
|
| 420 |
+
"step": 950
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 0.7793273174733388,
|
| 424 |
+
"eval_loss": 0.2371567040681839,
|
| 425 |
+
"eval_runtime": 19.4844,
|
| 426 |
+
"eval_samples_per_second": 51.323,
|
| 427 |
+
"eval_steps_per_second": 0.821,
|
| 428 |
+
"step": 950
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 0.7998359310910582,
|
| 432 |
+
"grad_norm": 0.1438579112291336,
|
| 433 |
+
"learning_rate": 0.0007508262527833029,
|
| 434 |
+
"loss": 0.2458,
|
| 435 |
+
"step": 975
|
| 436 |
+
},
|
| 437 |
+
{
|
| 438 |
+
"epoch": 0.8203445447087777,
|
| 439 |
+
"grad_norm": 0.1307683140039444,
|
| 440 |
+
"learning_rate": 0.000735185186484943,
|
| 441 |
+
"loss": 0.2478,
|
| 442 |
+
"step": 1000
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 0.8203445447087777,
|
| 446 |
+
"eval_loss": 0.23642279207706451,
|
| 447 |
+
"eval_runtime": 19.474,
|
| 448 |
+
"eval_samples_per_second": 51.35,
|
| 449 |
+
"eval_steps_per_second": 0.822,
|
| 450 |
+
"step": 1000
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.8408531583264971,
|
| 454 |
+
"grad_norm": 0.12361195683479309,
|
| 455 |
+
"learning_rate": 0.0007192427709458656,
|
| 456 |
+
"loss": 0.2358,
|
| 457 |
+
"step": 1025
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 0.8613617719442166,
|
| 461 |
+
"grad_norm": 0.12630796432495117,
|
| 462 |
+
"learning_rate": 0.0007030194336210887,
|
| 463 |
+
"loss": 0.2409,
|
| 464 |
+
"step": 1050
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.8613617719442166,
|
| 468 |
+
"eval_loss": 0.23247265815734863,
|
| 469 |
+
"eval_runtime": 19.4414,
|
| 470 |
+
"eval_samples_per_second": 51.437,
|
| 471 |
+
"eval_steps_per_second": 0.823,
|
| 472 |
+
"step": 1050
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 0.881870385561936,
|
| 476 |
+
"grad_norm": 0.13411730527877808,
|
| 477 |
+
"learning_rate": 0.0006865359619184331,
|
| 478 |
+
"loss": 0.2447,
|
| 479 |
+
"step": 1075
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 0.9023789991796555,
|
| 483 |
+
"grad_norm": 0.12585557997226715,
|
| 484 |
+
"learning_rate": 0.0006698134765630434,
|
| 485 |
+
"loss": 0.2379,
|
| 486 |
+
"step": 1100
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 0.9023789991796555,
|
| 490 |
+
"eval_loss": 0.2312391996383667,
|
| 491 |
+
"eval_runtime": 19.4906,
|
| 492 |
+
"eval_samples_per_second": 51.307,
|
| 493 |
+
"eval_steps_per_second": 0.821,
|
| 494 |
+
"step": 1100
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 0.9228876127973749,
|
| 498 |
+
"grad_norm": 0.10996991395950317,
|
| 499 |
+
"learning_rate": 0.0006528734045348248,
|
| 500 |
+
"loss": 0.2347,
|
| 501 |
+
"step": 1125
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 0.9433962264150944,
|
| 505 |
+
"grad_norm": 0.1385001838207245,
|
| 506 |
+
"learning_rate": 0.0006357374516134643,
|
| 507 |
+
"loss": 0.2489,
|
| 508 |
+
"step": 1150
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 0.9433962264150944,
|
| 512 |
+
"eval_loss": 0.2296425998210907,
|
| 513 |
+
"eval_runtime": 19.4676,
|
| 514 |
+
"eval_samples_per_second": 51.367,
|
| 515 |
+
"eval_steps_per_second": 0.822,
|
| 516 |
+
"step": 1150
|
| 517 |
+
},
|
| 518 |
+
{
|
| 519 |
+
"epoch": 0.9639048400328137,
|
| 520 |
+
"grad_norm": 0.1425400674343109,
|
| 521 |
+
"learning_rate": 0.0006184275745662179,
|
| 522 |
+
"loss": 0.2376,
|
| 523 |
+
"step": 1175
|
| 524 |
+
},
|
| 525 |
+
{
|
| 526 |
+
"epoch": 0.9844134536505332,
|
| 527 |
+
"grad_norm": 0.12151045352220535,
|
| 528 |
+
"learning_rate": 0.0006009659530141031,
|
| 529 |
+
"loss": 0.2408,
|
| 530 |
+
"step": 1200
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 0.9844134536505332,
|
| 534 |
+
"eval_loss": 0.22871758043766022,
|
| 535 |
+
"eval_runtime": 19.4794,
|
| 536 |
+
"eval_samples_per_second": 51.336,
|
| 537 |
+
"eval_steps_per_second": 0.821,
|
| 538 |
+
"step": 1200
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"epoch": 1.0049220672682526,
|
| 542 |
+
"grad_norm": 0.1182444617152214,
|
| 543 |
+
"learning_rate": 0.0005833749610125402,
|
| 544 |
+
"loss": 0.237,
|
| 545 |
+
"step": 1225
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 1.0254306808859721,
|
| 549 |
+
"grad_norm": 0.11846429109573364,
|
| 550 |
+
"learning_rate": 0.0005656771383828602,
|
| 551 |
+
"loss": 0.2092,
|
| 552 |
+
"step": 1250
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 1.0254306808859721,
|
| 556 |
+
"eval_loss": 0.22994151711463928,
|
| 557 |
+
"eval_runtime": 19.4266,
|
| 558 |
+
"eval_samples_per_second": 51.476,
|
| 559 |
+
"eval_steps_per_second": 0.824,
|
| 560 |
+
"step": 1250
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 1.0459392945036916,
|
| 564 |
+
"grad_norm": 0.12247344851493835,
|
| 565 |
+
"learning_rate": 0.0005478951618314134,
|
| 566 |
+
"loss": 0.2043,
|
| 567 |
+
"step": 1275
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 1.066447908121411,
|
| 571 |
+
"grad_norm": 0.11501295119524002,
|
| 572 |
+
"learning_rate": 0.0005300518158932815,
|
| 573 |
+
"loss": 0.2051,
|
| 574 |
+
"step": 1300
|
| 575 |
+
},
|
| 576 |
+
{
|
| 577 |
+
"epoch": 1.066447908121411,
|
| 578 |
+
"eval_loss": 0.22652946412563324,
|
| 579 |
+
"eval_runtime": 19.4689,
|
| 580 |
+
"eval_samples_per_second": 51.364,
|
| 581 |
+
"eval_steps_per_second": 0.822,
|
| 582 |
+
"step": 1300
|
| 583 |
+
},
|
| 584 |
+
{
|
| 585 |
+
"epoch": 1.0869565217391304,
|
| 586 |
+
"grad_norm": 0.12041299045085907,
|
| 587 |
+
"learning_rate": 0.0005121699637378282,
|
| 588 |
+
"loss": 0.2046,
|
| 589 |
+
"step": 1325
|
| 590 |
+
},
|
| 591 |
+
{
|
| 592 |
+
"epoch": 1.1074651353568499,
|
| 593 |
+
"grad_norm": 0.13483497500419617,
|
| 594 |
+
"learning_rate": 0.0004942725178734903,
|
| 595 |
+
"loss": 0.2088,
|
| 596 |
+
"step": 1350
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 1.1074651353568499,
|
| 600 |
+
"eval_loss": 0.22511404752731323,
|
| 601 |
+
"eval_runtime": 19.4668,
|
| 602 |
+
"eval_samples_per_second": 51.37,
|
| 603 |
+
"eval_steps_per_second": 0.822,
|
| 604 |
+
"step": 1350
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 1.1279737489745694,
|
| 608 |
+
"grad_norm": 0.12721161544322968,
|
| 609 |
+
"learning_rate": 0.00047638241078935324,
|
| 610 |
+
"loss": 0.2069,
|
| 611 |
+
"step": 1375
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 1.1484823625922886,
|
| 615 |
+
"grad_norm": 0.1224995031952858,
|
| 616 |
+
"learning_rate": 0.000458522565571121,
|
| 617 |
+
"loss": 0.2145,
|
| 618 |
+
"step": 1400
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 1.1484823625922886,
|
| 622 |
+
"eval_loss": 0.2244964987039566,
|
| 623 |
+
"eval_runtime": 19.4669,
|
| 624 |
+
"eval_samples_per_second": 51.369,
|
| 625 |
+
"eval_steps_per_second": 0.822,
|
| 626 |
+
"step": 1400
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 1.1689909762100081,
|
| 630 |
+
"grad_norm": 0.11426686495542526,
|
| 631 |
+
"learning_rate": 0.00044071586652913767,
|
| 632 |
+
"loss": 0.1983,
|
| 633 |
+
"step": 1425
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 1.1894995898277276,
|
| 637 |
+
"grad_norm": 0.13497664034366608,
|
| 638 |
+
"learning_rate": 0.0004229851298760915,
|
| 639 |
+
"loss": 0.2079,
|
| 640 |
+
"step": 1450
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 1.1894995898277276,
|
| 644 |
+
"eval_loss": 0.22429363429546356,
|
| 645 |
+
"eval_runtime": 19.4906,
|
| 646 |
+
"eval_samples_per_second": 51.307,
|
| 647 |
+
"eval_steps_per_second": 0.821,
|
| 648 |
+
"step": 1450
|
| 649 |
+
},
|
| 650 |
+
{
|
| 651 |
+
"epoch": 1.2100082034454471,
|
| 652 |
+
"grad_norm": 0.11498471349477768,
|
| 653 |
+
"learning_rate": 0.0004053530744919749,
|
| 654 |
+
"loss": 0.2069,
|
| 655 |
+
"step": 1475
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 1.2305168170631666,
|
| 659 |
+
"grad_norm": 0.13178111612796783,
|
| 660 |
+
"learning_rate": 0.0003878422928137597,
|
| 661 |
+
"loss": 0.2103,
|
| 662 |
+
"step": 1500
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 1.2305168170631666,
|
| 666 |
+
"eval_loss": 0.22244225442409515,
|
| 667 |
+
"eval_runtime": 19.4709,
|
| 668 |
+
"eval_samples_per_second": 51.359,
|
| 669 |
+
"eval_steps_per_second": 0.822,
|
| 670 |
+
"step": 1500
|
| 671 |
+
},
|
| 672 |
+
{
|
| 673 |
+
"epoch": 1.251025430680886,
|
| 674 |
+
"grad_norm": 0.1308068484067917,
|
| 675 |
+
"learning_rate": 0.0003704752218870861,
|
| 676 |
+
"loss": 0.2151,
|
| 677 |
+
"step": 1525
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 1.2715340442986054,
|
| 681 |
+
"grad_norm": 0.11254964768886566,
|
| 682 |
+
"learning_rate": 0.00035327411461706025,
|
| 683 |
+
"loss": 0.1971,
|
| 684 |
+
"step": 1550
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 1.2715340442986054,
|
| 688 |
+
"eval_loss": 0.22179578244686127,
|
| 689 |
+
"eval_runtime": 19.4274,
|
| 690 |
+
"eval_samples_per_second": 51.474,
|
| 691 |
+
"eval_steps_per_second": 0.824,
|
| 692 |
+
"step": 1550
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 1.2920426579163249,
|
| 696 |
+
"grad_norm": 0.13904231786727905,
|
| 697 |
+
"learning_rate": 0.00033626101125499555,
|
| 698 |
+
"loss": 0.2065,
|
| 699 |
+
"step": 1575
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 1.3125512715340442,
|
| 703 |
+
"grad_norm": 0.11037751287221909,
|
| 704 |
+
"learning_rate": 0.0003194577111576333,
|
| 705 |
+
"loss": 0.2013,
|
| 706 |
+
"step": 1600
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 1.3125512715340442,
|
| 710 |
+
"eval_loss": 0.21995632350444794,
|
| 711 |
+
"eval_runtime": 19.4997,
|
| 712 |
+
"eval_samples_per_second": 51.283,
|
| 713 |
+
"eval_steps_per_second": 0.821,
|
| 714 |
+
"step": 1600
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 1.3330598851517639,
|
| 718 |
+
"grad_norm": 0.10675220936536789,
|
| 719 |
+
"learning_rate": 0.00030288574485502756,
|
| 720 |
+
"loss": 0.2017,
|
| 721 |
+
"step": 1625
|
| 722 |
+
},
|
| 723 |
+
{
|
| 724 |
+
"epoch": 1.3535684987694832,
|
| 725 |
+
"grad_norm": 0.11563979089260101,
|
| 726 |
+
"learning_rate": 0.00028656634646288565,
|
| 727 |
+
"loss": 0.2033,
|
| 728 |
+
"step": 1650
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 1.3535684987694832,
|
| 732 |
+
"eval_loss": 0.21779730916023254,
|
| 733 |
+
"eval_runtime": 19.4741,
|
| 734 |
+
"eval_samples_per_second": 51.35,
|
| 735 |
+
"eval_steps_per_second": 0.822,
|
| 736 |
+
"step": 1650
|
| 737 |
+
},
|
| 738 |
+
{
|
| 739 |
+
"epoch": 1.3740771123872026,
|
| 740 |
+
"grad_norm": 0.13573266565799713,
|
| 741 |
+
"learning_rate": 0.00027052042647471254,
|
| 742 |
+
"loss": 0.1916,
|
| 743 |
+
"step": 1675
|
| 744 |
+
},
|
| 745 |
+
{
|
| 746 |
+
"epoch": 1.3945857260049221,
|
| 747 |
+
"grad_norm": 0.12692494690418243,
|
| 748 |
+
"learning_rate": 0.0002547685449686206,
|
| 749 |
+
"loss": 0.1999,
|
| 750 |
+
"step": 1700
|
| 751 |
+
},
|
| 752 |
+
{
|
| 753 |
+
"epoch": 1.3945857260049221,
|
| 754 |
+
"eval_loss": 0.21664589643478394,
|
| 755 |
+
"eval_runtime": 19.4774,
|
| 756 |
+
"eval_samples_per_second": 51.341,
|
| 757 |
+
"eval_steps_per_second": 0.821,
|
| 758 |
+
"step": 1700
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 1.4150943396226414,
|
| 762 |
+
"grad_norm": 0.14164209365844727,
|
| 763 |
+
"learning_rate": 0.0002393308852631373,
|
| 764 |
+
"loss": 0.2108,
|
| 765 |
+
"step": 1725
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 1.435602953240361,
|
| 769 |
+
"grad_norm": 0.12433931976556778,
|
| 770 |
+
"learning_rate": 0.0002242272280557645,
|
| 771 |
+
"loss": 0.2056,
|
| 772 |
+
"step": 1750
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"epoch": 1.435602953240361,
|
| 776 |
+
"eval_loss": 0.2161126285791397,
|
| 777 |
+
"eval_runtime": 19.4746,
|
| 778 |
+
"eval_samples_per_second": 51.349,
|
| 779 |
+
"eval_steps_per_second": 0.822,
|
| 780 |
+
"step": 1750
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 1.4561115668580804,
|
| 784 |
+
"grad_norm": 0.1255362182855606,
|
| 785 |
+
"learning_rate": 0.0002094769260774262,
|
| 786 |
+
"loss": 0.2021,
|
| 787 |
+
"step": 1775
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 1.4766201804758,
|
| 791 |
+
"grad_norm": 0.12211991846561432,
|
| 792 |
+
"learning_rate": 0.00019509887929528458,
|
| 793 |
+
"loss": 0.2056,
|
| 794 |
+
"step": 1800
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 1.4766201804758,
|
| 798 |
+
"eval_loss": 0.21495358645915985,
|
| 799 |
+
"eval_runtime": 19.4956,
|
| 800 |
+
"eval_samples_per_second": 51.294,
|
| 801 |
+
"eval_steps_per_second": 0.821,
|
| 802 |
+
"step": 1800
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 1.4971287940935194,
|
| 806 |
+
"grad_norm": 0.13951410353183746,
|
| 807 |
+
"learning_rate": 0.0001811115106956918,
|
| 808 |
+
"loss": 0.1964,
|
| 809 |
+
"step": 1825
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 1.5176374077112387,
|
| 813 |
+
"grad_norm": 0.13265341520309448,
|
| 814 |
+
"learning_rate": 0.00016753274267831115,
|
| 815 |
+
"loss": 0.206,
|
| 816 |
+
"step": 1850
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"epoch": 1.5176374077112387,
|
| 820 |
+
"eval_loss": 0.21398000419139862,
|
| 821 |
+
"eval_runtime": 19.4829,
|
| 822 |
+
"eval_samples_per_second": 51.327,
|
| 823 |
+
"eval_steps_per_second": 0.821,
|
| 824 |
+
"step": 1850
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 1.5381460213289582,
|
| 828 |
+
"grad_norm": 0.13726522028446198,
|
| 829 |
+
"learning_rate": 0.00015437997409165478,
|
| 830 |
+
"loss": 0.1964,
|
| 831 |
+
"step": 1875
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 1.5586546349466777,
|
| 835 |
+
"grad_norm": 0.11551380902528763,
|
| 836 |
+
"learning_rate": 0.00014167005793946035,
|
| 837 |
+
"loss": 0.1952,
|
| 838 |
+
"step": 1900
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 1.5586546349466777,
|
| 842 |
+
"eval_loss": 0.2141638696193695,
|
| 843 |
+
"eval_runtime": 19.4999,
|
| 844 |
+
"eval_samples_per_second": 51.282,
|
| 845 |
+
"eval_steps_per_second": 0.821,
|
| 846 |
+
"step": 1900
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 1.579163248564397,
|
| 850 |
+
"grad_norm": 0.13091330230236053,
|
| 851 |
+
"learning_rate": 0.00012941927978647527,
|
| 852 |
+
"loss": 0.1976,
|
| 853 |
+
"step": 1925
|
| 854 |
+
},
|
| 855 |
+
{
|
| 856 |
+
"epoch": 1.5996718621821167,
|
| 857 |
+
"grad_norm": 0.11574764549732208,
|
| 858 |
+
"learning_rate": 0.00011764333689131385,
|
| 859 |
+
"loss": 0.2036,
|
| 860 |
+
"step": 1950
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 1.5996718621821167,
|
| 864 |
+
"eval_loss": 0.2131749391555786,
|
| 865 |
+
"eval_runtime": 19.4949,
|
| 866 |
+
"eval_samples_per_second": 51.295,
|
| 867 |
+
"eval_steps_per_second": 0.821,
|
| 868 |
+
"step": 1950
|
| 869 |
+
},
|
| 870 |
+
{
|
| 871 |
+
"epoch": 1.620180475799836,
|
| 872 |
+
"grad_norm": 0.11186928302049637,
|
| 873 |
+
"learning_rate": 0.00010635731809312993,
|
| 874 |
+
"loss": 0.1993,
|
| 875 |
+
"step": 1975
|
| 876 |
+
},
|
| 877 |
+
{
|
| 878 |
+
"epoch": 1.6406890894175554,
|
| 879 |
+
"grad_norm": 0.13052970170974731,
|
| 880 |
+
"learning_rate": 9.557568447787201e-05,
|
| 881 |
+
"loss": 0.1989,
|
| 882 |
+
"step": 2000
|
| 883 |
+
},
|
| 884 |
+
{
|
| 885 |
+
"epoch": 1.6406890894175554,
|
| 886 |
+
"eval_loss": 0.21262580156326294,
|
| 887 |
+
"eval_runtime": 19.4819,
|
| 888 |
+
"eval_samples_per_second": 51.33,
|
| 889 |
+
"eval_steps_per_second": 0.821,
|
| 890 |
+
"step": 2000
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 1.661197703035275,
|
| 894 |
+
"grad_norm": 0.11890964955091476,
|
| 895 |
+
"learning_rate": 8.531225084889654e-05,
|
| 896 |
+
"loss": 0.1925,
|
| 897 |
+
"step": 2025
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 1.6817063166529942,
|
| 901 |
+
"grad_norm": 0.13581904768943787,
|
| 902 |
+
"learning_rate": 7.558016802568091e-05,
|
| 903 |
+
"loss": 0.1966,
|
| 904 |
+
"step": 2050
|
| 905 |
+
},
|
| 906 |
+
{
|
| 907 |
+
"epoch": 1.6817063166529942,
|
| 908 |
+
"eval_loss": 0.21162261068820953,
|
| 909 |
+
"eval_runtime": 19.4663,
|
| 910 |
+
"eval_samples_per_second": 51.371,
|
| 911 |
+
"eval_steps_per_second": 0.822,
|
| 912 |
+
"step": 2050
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 1.7022149302707137,
|
| 916 |
+
"grad_norm": 0.12677335739135742,
|
| 917 |
+
"learning_rate": 6.639190599331746e-05,
|
| 918 |
+
"loss": 0.1964,
|
| 919 |
+
"step": 2075
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 1.7227235438884332,
|
| 923 |
+
"grad_norm": 0.13097846508026123,
|
| 924 |
+
"learning_rate": 5.775923792437865e-05,
|
| 925 |
+
"loss": 0.1918,
|
| 926 |
+
"step": 2100
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 1.7227235438884332,
|
| 930 |
+
"eval_loss": 0.2108549326658249,
|
| 931 |
+
"eval_runtime": 19.5146,
|
| 932 |
+
"eval_samples_per_second": 51.244,
|
| 933 |
+
"eval_steps_per_second": 0.82,
|
| 934 |
+
"step": 2100
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 1.7432321575061525,
|
| 938 |
+
"grad_norm": 0.12936541438102722,
|
| 939 |
+
"learning_rate": 4.9693225093627616e-05,
|
| 940 |
+
"loss": 0.1931,
|
| 941 |
+
"step": 2125
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 1.7637407711238722,
|
| 945 |
+
"grad_norm": 0.12237502634525299,
|
| 946 |
+
"learning_rate": 4.220420270490294e-05,
|
| 947 |
+
"loss": 0.1874,
|
| 948 |
+
"step": 2150
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 1.7637407711238722,
|
| 952 |
+
"eval_loss": 0.2103663980960846,
|
| 953 |
+
"eval_runtime": 19.4878,
|
| 954 |
+
"eval_samples_per_second": 51.314,
|
| 955 |
+
"eval_steps_per_second": 0.821,
|
| 956 |
+
"step": 2150
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 1.7842493847415914,
|
| 960 |
+
"grad_norm": 0.13166476786136627,
|
| 961 |
+
"learning_rate": 3.530176664833834e-05,
|
| 962 |
+
"loss": 0.1901,
|
| 963 |
+
"step": 2175
|
| 964 |
+
},
|
| 965 |
+
{
|
| 966 |
+
"epoch": 1.804757998359311,
|
| 967 |
+
"grad_norm": 0.14463044703006744,
|
| 968 |
+
"learning_rate": 2.8994761204884756e-05,
|
| 969 |
+
"loss": 0.1986,
|
| 970 |
+
"step": 2200
|
| 971 |
+
},
|
| 972 |
+
{
|
| 973 |
+
"epoch": 1.804757998359311,
|
| 974 |
+
"eval_loss": 0.2102488875389099,
|
| 975 |
+
"eval_runtime": 19.4852,
|
| 976 |
+
"eval_samples_per_second": 51.321,
|
| 977 |
+
"eval_steps_per_second": 0.821,
|
| 978 |
+
"step": 2200
|
| 979 |
+
},
|
| 980 |
+
{
|
| 981 |
+
"epoch": 1.8252666119770304,
|
| 982 |
+
"grad_norm": 0.1415223479270935,
|
| 983 |
+
"learning_rate": 2.329126771388995e-05,
|
| 984 |
+
"loss": 0.1903,
|
| 985 |
+
"step": 2225
|
| 986 |
+
},
|
| 987 |
+
{
|
| 988 |
+
"epoch": 1.8457752255947497,
|
| 989 |
+
"grad_norm": 0.11420800536870956,
|
| 990 |
+
"learning_rate": 1.8198594218256815e-05,
|
| 991 |
+
"loss": 0.1901,
|
| 992 |
+
"step": 2250
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 1.8457752255947497,
|
| 996 |
+
"eval_loss": 0.2102055698633194,
|
| 997 |
+
"eval_runtime": 19.4587,
|
| 998 |
+
"eval_samples_per_second": 51.391,
|
| 999 |
+
"eval_steps_per_second": 0.822,
|
| 1000 |
+
"step": 2250
|
| 1001 |
+
},
|
| 1002 |
+
{
|
| 1003 |
+
"epoch": 1.8662838392124692,
|
| 1004 |
+
"grad_norm": 0.1356535106897354,
|
| 1005 |
+
"learning_rate": 1.3723266100447052e-05,
|
| 1006 |
+
"loss": 0.1943,
|
| 1007 |
+
"step": 2275
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"epoch": 1.8867924528301887,
|
| 1011 |
+
"grad_norm": 0.15679548680782318,
|
| 1012 |
+
"learning_rate": 9.871017721329201e-06,
|
| 1013 |
+
"loss": 0.2016,
|
| 1014 |
+
"step": 2300
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"epoch": 1.8867924528301887,
|
| 1018 |
+
"eval_loss": 0.2100086808204651,
|
| 1019 |
+
"eval_runtime": 19.461,
|
| 1020 |
+
"eval_samples_per_second": 51.385,
|
| 1021 |
+
"eval_steps_per_second": 0.822,
|
| 1022 |
+
"step": 2300
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 1.907301066447908,
|
| 1026 |
+
"grad_norm": 0.13155241310596466,
|
| 1027 |
+
"learning_rate": 6.646785072584871e-06,
|
| 1028 |
+
"loss": 0.1959,
|
| 1029 |
+
"step": 2325
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 1.9278096800656277,
|
| 1033 |
+
"grad_norm": 0.11297722160816193,
|
| 1034 |
+
"learning_rate": 4.054699452086641e-06,
|
| 1035 |
+
"loss": 0.1904,
|
| 1036 |
+
"step": 2350
|
| 1037 |
+
},
|
| 1038 |
+
{
|
| 1039 |
+
"epoch": 1.9278096800656277,
|
| 1040 |
+
"eval_loss": 0.2098853588104248,
|
| 1041 |
+
"eval_runtime": 19.4561,
|
| 1042 |
+
"eval_samples_per_second": 51.398,
|
| 1043 |
+
"eval_steps_per_second": 0.822,
|
| 1044 |
+
"step": 2350
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 1.948318293683347,
|
| 1048 |
+
"grad_norm": 0.13836362957954407,
|
| 1049 |
+
"learning_rate": 2.0980821703527886e-06,
|
| 1050 |
+
"loss": 0.2049,
|
| 1051 |
+
"step": 2375
|
| 1052 |
+
},
|
| 1053 |
+
{
|
| 1054 |
+
"epoch": 1.9688269073010665,
|
| 1055 |
+
"grad_norm": 0.12997499108314514,
|
| 1056 |
+
"learning_rate": 7.794402948607671e-07,
|
| 1057 |
+
"loss": 0.1999,
|
| 1058 |
+
"step": 2400
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 1.9688269073010665,
|
| 1062 |
+
"eval_loss": 0.2098313570022583,
|
| 1063 |
+
"eval_runtime": 19.5038,
|
| 1064 |
+
"eval_samples_per_second": 51.272,
|
| 1065 |
+
"eval_steps_per_second": 0.82,
|
| 1066 |
+
"step": 2400
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 1.989335520918786,
|
| 1070 |
+
"grad_norm": 0.14497987926006317,
|
| 1071 |
+
"learning_rate": 1.0046343767294853e-07,
|
| 1072 |
+
"loss": 0.1901,
|
| 1073 |
+
"step": 2425
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 2.0,
|
| 1077 |
+
"step": 2438,
|
| 1078 |
+
"total_flos": 1.58523627405312e+18,
|
| 1079 |
+
"train_loss": 0.24394705248622253,
|
| 1080 |
+
"train_runtime": 3518.0931,
|
| 1081 |
+
"train_samples_per_second": 22.171,
|
| 1082 |
+
"train_steps_per_second": 0.693
|
| 1083 |
+
}
|
| 1084 |
+
],
|
| 1085 |
+
"logging_steps": 25,
|
| 1086 |
+
"max_steps": 2438,
|
| 1087 |
+
"num_input_tokens_seen": 0,
|
| 1088 |
+
"num_train_epochs": 2,
|
| 1089 |
+
"save_steps": 500,
|
| 1090 |
+
"stateful_callbacks": {
|
| 1091 |
+
"TrainerControl": {
|
| 1092 |
+
"args": {
|
| 1093 |
+
"should_epoch_stop": false,
|
| 1094 |
+
"should_evaluate": false,
|
| 1095 |
+
"should_log": false,
|
| 1096 |
+
"should_save": true,
|
| 1097 |
+
"should_training_stop": true
|
| 1098 |
+
},
|
| 1099 |
+
"attributes": {}
|
| 1100 |
+
}
|
| 1101 |
+
},
|
| 1102 |
+
"total_flos": 1.58523627405312e+18,
|
| 1103 |
+
"train_batch_size": 32,
|
| 1104 |
+
"trial_name": null,
|
| 1105 |
+
"trial_params": null
|
| 1106 |
+
}
|
nl_tasks/run_all/run_exnr12/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/run_all/run_exnr12/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/run_all/run_exnr12/trainer_state.json
ADDED
|
@@ -0,0 +1,1106 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
{
|
| 2 |
+
"best_global_step": null,
|
| 3 |
+
"best_metric": null,
|
| 4 |
+
"best_model_checkpoint": null,
|
| 5 |
+
"epoch": 2.0,
|
| 6 |
+
"eval_steps": 50,
|
| 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": 0.22555650770664215,
|
| 15 |
+
"learning_rate": 9.836065573770491e-05,
|
| 16 |
+
"loss": 0.6774,
|
| 17 |
+
"step": 25
|
| 18 |
+
},
|
| 19 |
+
{
|
| 20 |
+
"epoch": 0.04101722723543889,
|
| 21 |
+
"grad_norm": 0.18577826023101807,
|
| 22 |
+
"learning_rate": 0.00020081967213114754,
|
| 23 |
+
"loss": 0.4155,
|
| 24 |
+
"step": 50
|
| 25 |
+
},
|
| 26 |
+
{
|
| 27 |
+
"epoch": 0.04101722723543889,
|
| 28 |
+
"eval_loss": 0.37306666374206543,
|
| 29 |
+
"eval_runtime": 22.912,
|
| 30 |
+
"eval_samples_per_second": 43.645,
|
| 31 |
+
"eval_steps_per_second": 0.698,
|
| 32 |
+
"step": 50
|
| 33 |
+
},
|
| 34 |
+
{
|
| 35 |
+
"epoch": 0.06152584085315833,
|
| 36 |
+
"grad_norm": 0.17923958599567413,
|
| 37 |
+
"learning_rate": 0.0003032786885245902,
|
| 38 |
+
"loss": 0.3703,
|
| 39 |
+
"step": 75
|
| 40 |
+
},
|
| 41 |
+
{
|
| 42 |
+
"epoch": 0.08203445447087777,
|
| 43 |
+
"grad_norm": 0.18290868401527405,
|
| 44 |
+
"learning_rate": 0.0004057377049180328,
|
| 45 |
+
"loss": 0.3501,
|
| 46 |
+
"step": 100
|
| 47 |
+
},
|
| 48 |
+
{
|
| 49 |
+
"epoch": 0.08203445447087777,
|
| 50 |
+
"eval_loss": 0.3294435441493988,
|
| 51 |
+
"eval_runtime": 22.6287,
|
| 52 |
+
"eval_samples_per_second": 44.192,
|
| 53 |
+
"eval_steps_per_second": 0.707,
|
| 54 |
+
"step": 100
|
| 55 |
+
},
|
| 56 |
+
{
|
| 57 |
+
"epoch": 0.10254306808859721,
|
| 58 |
+
"grad_norm": 0.1609077900648117,
|
| 59 |
+
"learning_rate": 0.0005081967213114754,
|
| 60 |
+
"loss": 0.3305,
|
| 61 |
+
"step": 125
|
| 62 |
+
},
|
| 63 |
+
{
|
| 64 |
+
"epoch": 0.12305168170631665,
|
| 65 |
+
"grad_norm": 0.1902703195810318,
|
| 66 |
+
"learning_rate": 0.000610655737704918,
|
| 67 |
+
"loss": 0.3159,
|
| 68 |
+
"step": 150
|
| 69 |
+
},
|
| 70 |
+
{
|
| 71 |
+
"epoch": 0.12305168170631665,
|
| 72 |
+
"eval_loss": 0.31451696157455444,
|
| 73 |
+
"eval_runtime": 22.6178,
|
| 74 |
+
"eval_samples_per_second": 44.213,
|
| 75 |
+
"eval_steps_per_second": 0.707,
|
| 76 |
+
"step": 150
|
| 77 |
+
},
|
| 78 |
+
{
|
| 79 |
+
"epoch": 0.1435602953240361,
|
| 80 |
+
"grad_norm": 0.2086164802312851,
|
| 81 |
+
"learning_rate": 0.0007131147540983607,
|
| 82 |
+
"loss": 0.3135,
|
| 83 |
+
"step": 175
|
| 84 |
+
},
|
| 85 |
+
{
|
| 86 |
+
"epoch": 0.16406890894175555,
|
| 87 |
+
"grad_norm": 0.22159546613693237,
|
| 88 |
+
"learning_rate": 0.0008155737704918033,
|
| 89 |
+
"loss": 0.3193,
|
| 90 |
+
"step": 200
|
| 91 |
+
},
|
| 92 |
+
{
|
| 93 |
+
"epoch": 0.16406890894175555,
|
| 94 |
+
"eval_loss": 0.30913859605789185,
|
| 95 |
+
"eval_runtime": 22.6205,
|
| 96 |
+
"eval_samples_per_second": 44.208,
|
| 97 |
+
"eval_steps_per_second": 0.707,
|
| 98 |
+
"step": 200
|
| 99 |
+
},
|
| 100 |
+
{
|
| 101 |
+
"epoch": 0.184577522559475,
|
| 102 |
+
"grad_norm": 2.3699522018432617,
|
| 103 |
+
"learning_rate": 0.0009180327868852459,
|
| 104 |
+
"loss": 0.4187,
|
| 105 |
+
"step": 225
|
| 106 |
+
},
|
| 107 |
+
{
|
| 108 |
+
"epoch": 0.20508613617719443,
|
| 109 |
+
"grad_norm": 0.542046070098877,
|
| 110 |
+
"learning_rate": 0.0009999871854116063,
|
| 111 |
+
"loss": 0.4345,
|
| 112 |
+
"step": 250
|
| 113 |
+
},
|
| 114 |
+
{
|
| 115 |
+
"epoch": 0.20508613617719443,
|
| 116 |
+
"eval_loss": 0.3142073452472687,
|
| 117 |
+
"eval_runtime": 22.564,
|
| 118 |
+
"eval_samples_per_second": 44.318,
|
| 119 |
+
"eval_steps_per_second": 0.709,
|
| 120 |
+
"step": 250
|
| 121 |
+
},
|
| 122 |
+
{
|
| 123 |
+
"epoch": 0.22559474979491387,
|
| 124 |
+
"grad_norm": 0.3608435392379761,
|
| 125 |
+
"learning_rate": 0.0009995387437838027,
|
| 126 |
+
"loss": 0.3233,
|
| 127 |
+
"step": 275
|
| 128 |
+
},
|
| 129 |
+
{
|
| 130 |
+
"epoch": 0.2461033634126333,
|
| 131 |
+
"grad_norm": 0.28521400690078735,
|
| 132 |
+
"learning_rate": 0.000998450229439693,
|
| 133 |
+
"loss": 0.3179,
|
| 134 |
+
"step": 300
|
| 135 |
+
},
|
| 136 |
+
{
|
| 137 |
+
"epoch": 0.2461033634126333,
|
| 138 |
+
"eval_loss": 0.30236080288887024,
|
| 139 |
+
"eval_runtime": 22.6232,
|
| 140 |
+
"eval_samples_per_second": 44.202,
|
| 141 |
+
"eval_steps_per_second": 0.707,
|
| 142 |
+
"step": 300
|
| 143 |
+
},
|
| 144 |
+
{
|
| 145 |
+
"epoch": 0.2666119770303528,
|
| 146 |
+
"grad_norm": 0.22246453166007996,
|
| 147 |
+
"learning_rate": 0.000996723037122612,
|
| 148 |
+
"loss": 0.3011,
|
| 149 |
+
"step": 325
|
| 150 |
+
},
|
| 151 |
+
{
|
| 152 |
+
"epoch": 0.2871205906480722,
|
| 153 |
+
"grad_norm": 0.24288132786750793,
|
| 154 |
+
"learning_rate": 0.0009943593799315263,
|
| 155 |
+
"loss": 0.3078,
|
| 156 |
+
"step": 350
|
| 157 |
+
},
|
| 158 |
+
{
|
| 159 |
+
"epoch": 0.2871205906480722,
|
| 160 |
+
"eval_loss": 0.2915167212486267,
|
| 161 |
+
"eval_runtime": 22.6232,
|
| 162 |
+
"eval_samples_per_second": 44.202,
|
| 163 |
+
"eval_steps_per_second": 0.707,
|
| 164 |
+
"step": 350
|
| 165 |
+
},
|
| 166 |
+
{
|
| 167 |
+
"epoch": 0.30762920426579166,
|
| 168 |
+
"grad_norm": 0.21969842910766602,
|
| 169 |
+
"learning_rate": 0.0009913622864853324,
|
| 170 |
+
"loss": 0.2936,
|
| 171 |
+
"step": 375
|
| 172 |
+
},
|
| 173 |
+
{
|
| 174 |
+
"epoch": 0.3281378178835111,
|
| 175 |
+
"grad_norm": 0.2624405026435852,
|
| 176 |
+
"learning_rate": 0.0009877355970422024,
|
| 177 |
+
"loss": 0.3019,
|
| 178 |
+
"step": 400
|
| 179 |
+
},
|
| 180 |
+
{
|
| 181 |
+
"epoch": 0.3281378178835111,
|
| 182 |
+
"eval_loss": 0.2813864052295685,
|
| 183 |
+
"eval_runtime": 22.6144,
|
| 184 |
+
"eval_samples_per_second": 44.22,
|
| 185 |
+
"eval_steps_per_second": 0.708,
|
| 186 |
+
"step": 400
|
| 187 |
+
},
|
| 188 |
+
{
|
| 189 |
+
"epoch": 0.34864643150123054,
|
| 190 |
+
"grad_norm": 0.18397915363311768,
|
| 191 |
+
"learning_rate": 0.0009834839585789559,
|
| 192 |
+
"loss": 0.2951,
|
| 193 |
+
"step": 425
|
| 194 |
+
},
|
| 195 |
+
{
|
| 196 |
+
"epoch": 0.36915504511895,
|
| 197 |
+
"grad_norm": 0.1632954329252243,
|
| 198 |
+
"learning_rate": 0.000978612818836762,
|
| 199 |
+
"loss": 0.2885,
|
| 200 |
+
"step": 450
|
| 201 |
+
},
|
| 202 |
+
{
|
| 203 |
+
"epoch": 0.36915504511895,
|
| 204 |
+
"eval_loss": 0.26967909932136536,
|
| 205 |
+
"eval_runtime": 22.5519,
|
| 206 |
+
"eval_samples_per_second": 44.342,
|
| 207 |
+
"eval_steps_per_second": 0.709,
|
| 208 |
+
"step": 450
|
| 209 |
+
},
|
| 210 |
+
{
|
| 211 |
+
"epoch": 0.3896636587366694,
|
| 212 |
+
"grad_norm": 0.17423078417778015,
|
| 213 |
+
"learning_rate": 0.0009731284193407981,
|
| 214 |
+
"loss": 0.2864,
|
| 215 |
+
"step": 475
|
| 216 |
+
},
|
| 217 |
+
{
|
| 218 |
+
"epoch": 0.41017227235438886,
|
| 219 |
+
"grad_norm": 0.15992988646030426,
|
| 220 |
+
"learning_rate": 0.0009670377874028117,
|
| 221 |
+
"loss": 0.2763,
|
| 222 |
+
"step": 500
|
| 223 |
+
},
|
| 224 |
+
{
|
| 225 |
+
"epoch": 0.41017227235438886,
|
| 226 |
+
"eval_loss": 0.2657691538333893,
|
| 227 |
+
"eval_runtime": 22.5303,
|
| 228 |
+
"eval_samples_per_second": 44.385,
|
| 229 |
+
"eval_steps_per_second": 0.71,
|
| 230 |
+
"step": 500
|
| 231 |
+
},
|
| 232 |
+
{
|
| 233 |
+
"epoch": 0.4306808859721083,
|
| 234 |
+
"grad_norm": 0.17642049491405487,
|
| 235 |
+
"learning_rate": 0.0009603487271168336,
|
| 236 |
+
"loss": 0.2804,
|
| 237 |
+
"step": 525
|
| 238 |
+
},
|
| 239 |
+
{
|
| 240 |
+
"epoch": 0.45118949958982774,
|
| 241 |
+
"grad_norm": 0.15310442447662354,
|
| 242 |
+
"learning_rate": 0.0009530698093595781,
|
| 243 |
+
"loss": 0.2738,
|
| 244 |
+
"step": 550
|
| 245 |
+
},
|
| 246 |
+
{
|
| 247 |
+
"epoch": 0.45118949958982774,
|
| 248 |
+
"eval_loss": 0.25949400663375854,
|
| 249 |
+
"eval_runtime": 22.5573,
|
| 250 |
+
"eval_samples_per_second": 44.332,
|
| 251 |
+
"eval_steps_per_second": 0.709,
|
| 252 |
+
"step": 550
|
| 253 |
+
},
|
| 254 |
+
{
|
| 255 |
+
"epoch": 0.4716981132075472,
|
| 256 |
+
"grad_norm": 0.16352593898773193,
|
| 257 |
+
"learning_rate": 0.0009452103608083418,
|
| 258 |
+
"loss": 0.2605,
|
| 259 |
+
"step": 575
|
| 260 |
+
},
|
| 261 |
+
{
|
| 262 |
+
"epoch": 0.4922067268252666,
|
| 263 |
+
"grad_norm": 0.16357450187206268,
|
| 264 |
+
"learning_rate": 0.0009367804519904775,
|
| 265 |
+
"loss": 0.2695,
|
| 266 |
+
"step": 600
|
| 267 |
+
},
|
| 268 |
+
{
|
| 269 |
+
"epoch": 0.4922067268252666,
|
| 270 |
+
"eval_loss": 0.2563272714614868,
|
| 271 |
+
"eval_runtime": 22.5256,
|
| 272 |
+
"eval_samples_per_second": 44.394,
|
| 273 |
+
"eval_steps_per_second": 0.71,
|
| 274 |
+
"step": 600
|
| 275 |
+
},
|
| 276 |
+
{
|
| 277 |
+
"epoch": 0.5127153404429861,
|
| 278 |
+
"grad_norm": 0.26224130392074585,
|
| 279 |
+
"learning_rate": 0.0009277908843797492,
|
| 280 |
+
"loss": 0.2572,
|
| 281 |
+
"step": 625
|
| 282 |
+
},
|
| 283 |
+
{
|
| 284 |
+
"epoch": 0.5332239540607056,
|
| 285 |
+
"grad_norm": 0.15522083640098572,
|
| 286 |
+
"learning_rate": 0.0009182531765561084,
|
| 287 |
+
"loss": 0.2557,
|
| 288 |
+
"step": 650
|
| 289 |
+
},
|
| 290 |
+
{
|
| 291 |
+
"epoch": 0.5332239540607056,
|
| 292 |
+
"eval_loss": 0.25189048051834106,
|
| 293 |
+
"eval_runtime": 22.5857,
|
| 294 |
+
"eval_samples_per_second": 44.276,
|
| 295 |
+
"eval_steps_per_second": 0.708,
|
| 296 |
+
"step": 650
|
| 297 |
+
},
|
| 298 |
+
{
|
| 299 |
+
"epoch": 0.5537325676784249,
|
| 300 |
+
"grad_norm": 0.1434183418750763,
|
| 301 |
+
"learning_rate": 0.0009081795494466201,
|
| 302 |
+
"loss": 0.2577,
|
| 303 |
+
"step": 675
|
| 304 |
+
},
|
| 305 |
+
{
|
| 306 |
+
"epoch": 0.5742411812961444,
|
| 307 |
+
"grad_norm": 0.1259659081697464,
|
| 308 |
+
"learning_rate": 0.0008975829106664539,
|
| 309 |
+
"loss": 0.251,
|
| 310 |
+
"step": 700
|
| 311 |
+
},
|
| 312 |
+
{
|
| 313 |
+
"epoch": 0.5742411812961444,
|
| 314 |
+
"eval_loss": 0.25064969062805176,
|
| 315 |
+
"eval_runtime": 22.5416,
|
| 316 |
+
"eval_samples_per_second": 44.362,
|
| 317 |
+
"eval_steps_per_second": 0.71,
|
| 318 |
+
"step": 700
|
| 319 |
+
},
|
| 320 |
+
{
|
| 321 |
+
"epoch": 0.5947497949138638,
|
| 322 |
+
"grad_norm": 0.11997288465499878,
|
| 323 |
+
"learning_rate": 0.0008864768379800017,
|
| 324 |
+
"loss": 0.2478,
|
| 325 |
+
"step": 725
|
| 326 |
+
},
|
| 327 |
+
{
|
| 328 |
+
"epoch": 0.6152584085315833,
|
| 329 |
+
"grad_norm": 0.130789652466774,
|
| 330 |
+
"learning_rate": 0.0008748755619033153,
|
| 331 |
+
"loss": 0.2566,
|
| 332 |
+
"step": 750
|
| 333 |
+
},
|
| 334 |
+
{
|
| 335 |
+
"epoch": 0.6152584085315833,
|
| 336 |
+
"eval_loss": 0.24806569516658783,
|
| 337 |
+
"eval_runtime": 22.5742,
|
| 338 |
+
"eval_samples_per_second": 44.298,
|
| 339 |
+
"eval_steps_per_second": 0.709,
|
| 340 |
+
"step": 750
|
| 341 |
+
},
|
| 342 |
+
{
|
| 343 |
+
"epoch": 0.6357670221493027,
|
| 344 |
+
"grad_norm": 0.14199994504451752,
|
| 345 |
+
"learning_rate": 0.000862793947470155,
|
| 346 |
+
"loss": 0.2606,
|
| 347 |
+
"step": 775
|
| 348 |
+
},
|
| 349 |
+
{
|
| 350 |
+
"epoch": 0.6562756357670222,
|
| 351 |
+
"grad_norm": 0.1386088728904724,
|
| 352 |
+
"learning_rate": 0.0008502474751850142,
|
| 353 |
+
"loss": 0.251,
|
| 354 |
+
"step": 800
|
| 355 |
+
},
|
| 356 |
+
{
|
| 357 |
+
"epoch": 0.6562756357670222,
|
| 358 |
+
"eval_loss": 0.24411910772323608,
|
| 359 |
+
"eval_runtime": 22.5531,
|
| 360 |
+
"eval_samples_per_second": 44.34,
|
| 361 |
+
"eval_steps_per_second": 0.709,
|
| 362 |
+
"step": 800
|
| 363 |
+
},
|
| 364 |
+
{
|
| 365 |
+
"epoch": 0.6767842493847416,
|
| 366 |
+
"grad_norm": 0.12433658540248871,
|
| 367 |
+
"learning_rate": 0.0008372522211875224,
|
| 368 |
+
"loss": 0.2469,
|
| 369 |
+
"step": 825
|
| 370 |
+
},
|
| 371 |
+
{
|
| 372 |
+
"epoch": 0.6972928630024611,
|
| 373 |
+
"grad_norm": 0.13002005219459534,
|
| 374 |
+
"learning_rate": 0.0008238248366536473,
|
| 375 |
+
"loss": 0.2591,
|
| 376 |
+
"step": 850
|
| 377 |
+
},
|
| 378 |
+
{
|
| 379 |
+
"epoch": 0.6972928630024611,
|
| 380 |
+
"eval_loss": 0.24230404198169708,
|
| 381 |
+
"eval_runtime": 22.5724,
|
| 382 |
+
"eval_samples_per_second": 44.302,
|
| 383 |
+
"eval_steps_per_second": 0.709,
|
| 384 |
+
"step": 850
|
| 385 |
+
},
|
| 386 |
+
{
|
| 387 |
+
"epoch": 0.7178014766201805,
|
| 388 |
+
"grad_norm": 0.10630059987306595,
|
| 389 |
+
"learning_rate": 0.0008099825264600842,
|
| 390 |
+
"loss": 0.253,
|
| 391 |
+
"step": 875
|
| 392 |
+
},
|
| 393 |
+
{
|
| 394 |
+
"epoch": 0.7383100902379,
|
| 395 |
+
"grad_norm": 0.11450506001710892,
|
| 396 |
+
"learning_rate": 0.0007957430271391761,
|
| 397 |
+
"loss": 0.2467,
|
| 398 |
+
"step": 900
|
| 399 |
+
},
|
| 400 |
+
{
|
| 401 |
+
"epoch": 0.7383100902379,
|
| 402 |
+
"eval_loss": 0.2398349493741989,
|
| 403 |
+
"eval_runtime": 22.5769,
|
| 404 |
+
"eval_samples_per_second": 44.293,
|
| 405 |
+
"eval_steps_per_second": 0.709,
|
| 406 |
+
"step": 900
|
| 407 |
+
},
|
| 408 |
+
{
|
| 409 |
+
"epoch": 0.7588187038556193,
|
| 410 |
+
"grad_norm": 0.11560004949569702,
|
| 411 |
+
"learning_rate": 0.0007811245841526062,
|
| 412 |
+
"loss": 0.2506,
|
| 413 |
+
"step": 925
|
| 414 |
+
},
|
| 415 |
+
{
|
| 416 |
+
"epoch": 0.7793273174733388,
|
| 417 |
+
"grad_norm": 0.129349485039711,
|
| 418 |
+
"learning_rate": 0.0007661459285129879,
|
| 419 |
+
"loss": 0.234,
|
| 420 |
+
"step": 950
|
| 421 |
+
},
|
| 422 |
+
{
|
| 423 |
+
"epoch": 0.7793273174733388,
|
| 424 |
+
"eval_loss": 0.2383035570383072,
|
| 425 |
+
"eval_runtime": 22.5608,
|
| 426 |
+
"eval_samples_per_second": 44.325,
|
| 427 |
+
"eval_steps_per_second": 0.709,
|
| 428 |
+
"step": 950
|
| 429 |
+
},
|
| 430 |
+
{
|
| 431 |
+
"epoch": 0.7998359310910582,
|
| 432 |
+
"grad_norm": 0.11788657307624817,
|
| 433 |
+
"learning_rate": 0.0007508262527833029,
|
| 434 |
+
"loss": 0.2461,
|
| 435 |
+
"step": 975
|
| 436 |
+
},
|
| 437 |
+
{
|
| 438 |
+
"epoch": 0.8203445447087777,
|
| 439 |
+
"grad_norm": 0.11253403127193451,
|
| 440 |
+
"learning_rate": 0.000735185186484943,
|
| 441 |
+
"loss": 0.2495,
|
| 442 |
+
"step": 1000
|
| 443 |
+
},
|
| 444 |
+
{
|
| 445 |
+
"epoch": 0.8203445447087777,
|
| 446 |
+
"eval_loss": 0.23681500554084778,
|
| 447 |
+
"eval_runtime": 22.5792,
|
| 448 |
+
"eval_samples_per_second": 44.288,
|
| 449 |
+
"eval_steps_per_second": 0.709,
|
| 450 |
+
"step": 1000
|
| 451 |
+
},
|
| 452 |
+
{
|
| 453 |
+
"epoch": 0.8408531583264971,
|
| 454 |
+
"grad_norm": 0.1057242602109909,
|
| 455 |
+
"learning_rate": 0.0007192427709458656,
|
| 456 |
+
"loss": 0.2366,
|
| 457 |
+
"step": 1025
|
| 458 |
+
},
|
| 459 |
+
{
|
| 460 |
+
"epoch": 0.8613617719442166,
|
| 461 |
+
"grad_norm": 0.12001299113035202,
|
| 462 |
+
"learning_rate": 0.0007030194336210887,
|
| 463 |
+
"loss": 0.2426,
|
| 464 |
+
"step": 1050
|
| 465 |
+
},
|
| 466 |
+
{
|
| 467 |
+
"epoch": 0.8613617719442166,
|
| 468 |
+
"eval_loss": 0.23558472096920013,
|
| 469 |
+
"eval_runtime": 22.572,
|
| 470 |
+
"eval_samples_per_second": 44.303,
|
| 471 |
+
"eval_steps_per_second": 0.709,
|
| 472 |
+
"step": 1050
|
| 473 |
+
},
|
| 474 |
+
{
|
| 475 |
+
"epoch": 0.881870385561936,
|
| 476 |
+
"grad_norm": 0.13502635061740875,
|
| 477 |
+
"learning_rate": 0.0006865359619184331,
|
| 478 |
+
"loss": 0.2471,
|
| 479 |
+
"step": 1075
|
| 480 |
+
},
|
| 481 |
+
{
|
| 482 |
+
"epoch": 0.9023789991796555,
|
| 483 |
+
"grad_norm": 0.10965543240308762,
|
| 484 |
+
"learning_rate": 0.0006698134765630434,
|
| 485 |
+
"loss": 0.2397,
|
| 486 |
+
"step": 1100
|
| 487 |
+
},
|
| 488 |
+
{
|
| 489 |
+
"epoch": 0.9023789991796555,
|
| 490 |
+
"eval_loss": 0.2340947687625885,
|
| 491 |
+
"eval_runtime": 22.599,
|
| 492 |
+
"eval_samples_per_second": 44.25,
|
| 493 |
+
"eval_steps_per_second": 0.708,
|
| 494 |
+
"step": 1100
|
| 495 |
+
},
|
| 496 |
+
{
|
| 497 |
+
"epoch": 0.9228876127973749,
|
| 498 |
+
"grad_norm": 0.09839494526386261,
|
| 499 |
+
"learning_rate": 0.0006528734045348248,
|
| 500 |
+
"loss": 0.2376,
|
| 501 |
+
"step": 1125
|
| 502 |
+
},
|
| 503 |
+
{
|
| 504 |
+
"epoch": 0.9433962264150944,
|
| 505 |
+
"grad_norm": 0.11483541131019592,
|
| 506 |
+
"learning_rate": 0.0006357374516134643,
|
| 507 |
+
"loss": 0.2516,
|
| 508 |
+
"step": 1150
|
| 509 |
+
},
|
| 510 |
+
{
|
| 511 |
+
"epoch": 0.9433962264150944,
|
| 512 |
+
"eval_loss": 0.23218606412410736,
|
| 513 |
+
"eval_runtime": 22.5646,
|
| 514 |
+
"eval_samples_per_second": 44.317,
|
| 515 |
+
"eval_steps_per_second": 0.709,
|
| 516 |
+
"step": 1150
|
| 517 |
+
},
|
| 518 |
+
{
|
| 519 |
+
"epoch": 0.9639048400328137,
|
| 520 |
+
"grad_norm": 0.11629696935415268,
|
| 521 |
+
"learning_rate": 0.0006184275745662179,
|
| 522 |
+
"loss": 0.2405,
|
| 523 |
+
"step": 1175
|
| 524 |
+
},
|
| 525 |
+
{
|
| 526 |
+
"epoch": 0.9844134536505332,
|
| 527 |
+
"grad_norm": 0.10627080500125885,
|
| 528 |
+
"learning_rate": 0.0006009659530141031,
|
| 529 |
+
"loss": 0.2434,
|
| 530 |
+
"step": 1200
|
| 531 |
+
},
|
| 532 |
+
{
|
| 533 |
+
"epoch": 0.9844134536505332,
|
| 534 |
+
"eval_loss": 0.2307899445295334,
|
| 535 |
+
"eval_runtime": 22.5891,
|
| 536 |
+
"eval_samples_per_second": 44.269,
|
| 537 |
+
"eval_steps_per_second": 0.708,
|
| 538 |
+
"step": 1200
|
| 539 |
+
},
|
| 540 |
+
{
|
| 541 |
+
"epoch": 1.0049220672682526,
|
| 542 |
+
"grad_norm": 0.10291340202093124,
|
| 543 |
+
"learning_rate": 0.0005833749610125402,
|
| 544 |
+
"loss": 0.2417,
|
| 545 |
+
"step": 1225
|
| 546 |
+
},
|
| 547 |
+
{
|
| 548 |
+
"epoch": 1.0254306808859721,
|
| 549 |
+
"grad_norm": 0.1100890040397644,
|
| 550 |
+
"learning_rate": 0.0005656771383828602,
|
| 551 |
+
"loss": 0.2157,
|
| 552 |
+
"step": 1250
|
| 553 |
+
},
|
| 554 |
+
{
|
| 555 |
+
"epoch": 1.0254306808859721,
|
| 556 |
+
"eval_loss": 0.23217763006687164,
|
| 557 |
+
"eval_runtime": 22.572,
|
| 558 |
+
"eval_samples_per_second": 44.303,
|
| 559 |
+
"eval_steps_per_second": 0.709,
|
| 560 |
+
"step": 1250
|
| 561 |
+
},
|
| 562 |
+
{
|
| 563 |
+
"epoch": 1.0459392945036916,
|
| 564 |
+
"grad_norm": 0.10660576075315475,
|
| 565 |
+
"learning_rate": 0.0005478951618314134,
|
| 566 |
+
"loss": 0.2115,
|
| 567 |
+
"step": 1275
|
| 568 |
+
},
|
| 569 |
+
{
|
| 570 |
+
"epoch": 1.066447908121411,
|
| 571 |
+
"grad_norm": 0.10155736654996872,
|
| 572 |
+
"learning_rate": 0.0005300518158932815,
|
| 573 |
+
"loss": 0.2131,
|
| 574 |
+
"step": 1300
|
| 575 |
+
},
|
| 576 |
+
{
|
| 577 |
+
"epoch": 1.066447908121411,
|
| 578 |
+
"eval_loss": 0.2298276424407959,
|
| 579 |
+
"eval_runtime": 22.5487,
|
| 580 |
+
"eval_samples_per_second": 44.348,
|
| 581 |
+
"eval_steps_per_second": 0.71,
|
| 582 |
+
"step": 1300
|
| 583 |
+
},
|
| 584 |
+
{
|
| 585 |
+
"epoch": 1.0869565217391304,
|
| 586 |
+
"grad_norm": 0.10922397673130035,
|
| 587 |
+
"learning_rate": 0.0005121699637378282,
|
| 588 |
+
"loss": 0.2115,
|
| 589 |
+
"step": 1325
|
| 590 |
+
},
|
| 591 |
+
{
|
| 592 |
+
"epoch": 1.1074651353568499,
|
| 593 |
+
"grad_norm": 0.11864089965820312,
|
| 594 |
+
"learning_rate": 0.0004942725178734903,
|
| 595 |
+
"loss": 0.2172,
|
| 596 |
+
"step": 1350
|
| 597 |
+
},
|
| 598 |
+
{
|
| 599 |
+
"epoch": 1.1074651353568499,
|
| 600 |
+
"eval_loss": 0.22882254421710968,
|
| 601 |
+
"eval_runtime": 22.5984,
|
| 602 |
+
"eval_samples_per_second": 44.251,
|
| 603 |
+
"eval_steps_per_second": 0.708,
|
| 604 |
+
"step": 1350
|
| 605 |
+
},
|
| 606 |
+
{
|
| 607 |
+
"epoch": 1.1279737489745694,
|
| 608 |
+
"grad_norm": 0.11816546320915222,
|
| 609 |
+
"learning_rate": 0.00047638241078935324,
|
| 610 |
+
"loss": 0.2144,
|
| 611 |
+
"step": 1375
|
| 612 |
+
},
|
| 613 |
+
{
|
| 614 |
+
"epoch": 1.1484823625922886,
|
| 615 |
+
"grad_norm": 0.115484818816185,
|
| 616 |
+
"learning_rate": 0.000458522565571121,
|
| 617 |
+
"loss": 0.2225,
|
| 618 |
+
"step": 1400
|
| 619 |
+
},
|
| 620 |
+
{
|
| 621 |
+
"epoch": 1.1484823625922886,
|
| 622 |
+
"eval_loss": 0.22774316370487213,
|
| 623 |
+
"eval_runtime": 22.573,
|
| 624 |
+
"eval_samples_per_second": 44.301,
|
| 625 |
+
"eval_steps_per_second": 0.709,
|
| 626 |
+
"step": 1400
|
| 627 |
+
},
|
| 628 |
+
{
|
| 629 |
+
"epoch": 1.1689909762100081,
|
| 630 |
+
"grad_norm": 0.10250715166330338,
|
| 631 |
+
"learning_rate": 0.00044071586652913767,
|
| 632 |
+
"loss": 0.206,
|
| 633 |
+
"step": 1425
|
| 634 |
+
},
|
| 635 |
+
{
|
| 636 |
+
"epoch": 1.1894995898277276,
|
| 637 |
+
"grad_norm": 0.11922137439250946,
|
| 638 |
+
"learning_rate": 0.0004229851298760915,
|
| 639 |
+
"loss": 0.2158,
|
| 640 |
+
"step": 1450
|
| 641 |
+
},
|
| 642 |
+
{
|
| 643 |
+
"epoch": 1.1894995898277276,
|
| 644 |
+
"eval_loss": 0.22773854434490204,
|
| 645 |
+
"eval_runtime": 22.5992,
|
| 646 |
+
"eval_samples_per_second": 44.249,
|
| 647 |
+
"eval_steps_per_second": 0.708,
|
| 648 |
+
"step": 1450
|
| 649 |
+
},
|
| 650 |
+
{
|
| 651 |
+
"epoch": 1.2100082034454471,
|
| 652 |
+
"grad_norm": 0.10833777487277985,
|
| 653 |
+
"learning_rate": 0.0004053530744919749,
|
| 654 |
+
"loss": 0.2158,
|
| 655 |
+
"step": 1475
|
| 656 |
+
},
|
| 657 |
+
{
|
| 658 |
+
"epoch": 1.2305168170631666,
|
| 659 |
+
"grad_norm": 0.11317677795886993,
|
| 660 |
+
"learning_rate": 0.0003878422928137597,
|
| 661 |
+
"loss": 0.2187,
|
| 662 |
+
"step": 1500
|
| 663 |
+
},
|
| 664 |
+
{
|
| 665 |
+
"epoch": 1.2305168170631666,
|
| 666 |
+
"eval_loss": 0.2261391282081604,
|
| 667 |
+
"eval_runtime": 22.6015,
|
| 668 |
+
"eval_samples_per_second": 44.245,
|
| 669 |
+
"eval_steps_per_second": 0.708,
|
| 670 |
+
"step": 1500
|
| 671 |
+
},
|
| 672 |
+
{
|
| 673 |
+
"epoch": 1.251025430680886,
|
| 674 |
+
"grad_norm": 0.11307208985090256,
|
| 675 |
+
"learning_rate": 0.0003704752218870861,
|
| 676 |
+
"loss": 0.2236,
|
| 677 |
+
"step": 1525
|
| 678 |
+
},
|
| 679 |
+
{
|
| 680 |
+
"epoch": 1.2715340442986054,
|
| 681 |
+
"grad_norm": 0.10417971014976501,
|
| 682 |
+
"learning_rate": 0.00035327411461706025,
|
| 683 |
+
"loss": 0.2054,
|
| 684 |
+
"step": 1550
|
| 685 |
+
},
|
| 686 |
+
{
|
| 687 |
+
"epoch": 1.2715340442986054,
|
| 688 |
+
"eval_loss": 0.2259172946214676,
|
| 689 |
+
"eval_runtime": 22.5368,
|
| 690 |
+
"eval_samples_per_second": 44.372,
|
| 691 |
+
"eval_steps_per_second": 0.71,
|
| 692 |
+
"step": 1550
|
| 693 |
+
},
|
| 694 |
+
{
|
| 695 |
+
"epoch": 1.2920426579163249,
|
| 696 |
+
"grad_norm": 0.11175528168678284,
|
| 697 |
+
"learning_rate": 0.00033626101125499555,
|
| 698 |
+
"loss": 0.2151,
|
| 699 |
+
"step": 1575
|
| 700 |
+
},
|
| 701 |
+
{
|
| 702 |
+
"epoch": 1.3125512715340442,
|
| 703 |
+
"grad_norm": 0.10208392888307571,
|
| 704 |
+
"learning_rate": 0.0003194577111576333,
|
| 705 |
+
"loss": 0.2113,
|
| 706 |
+
"step": 1600
|
| 707 |
+
},
|
| 708 |
+
{
|
| 709 |
+
"epoch": 1.3125512715340442,
|
| 710 |
+
"eval_loss": 0.2242567092180252,
|
| 711 |
+
"eval_runtime": 22.5352,
|
| 712 |
+
"eval_samples_per_second": 44.375,
|
| 713 |
+
"eval_steps_per_second": 0.71,
|
| 714 |
+
"step": 1600
|
| 715 |
+
},
|
| 716 |
+
{
|
| 717 |
+
"epoch": 1.3330598851517639,
|
| 718 |
+
"grad_norm": 0.10389668494462967,
|
| 719 |
+
"learning_rate": 0.00030288574485502756,
|
| 720 |
+
"loss": 0.2107,
|
| 721 |
+
"step": 1625
|
| 722 |
+
},
|
| 723 |
+
{
|
| 724 |
+
"epoch": 1.3535684987694832,
|
| 725 |
+
"grad_norm": 0.10494068264961243,
|
| 726 |
+
"learning_rate": 0.00028656634646288565,
|
| 727 |
+
"loss": 0.2119,
|
| 728 |
+
"step": 1650
|
| 729 |
+
},
|
| 730 |
+
{
|
| 731 |
+
"epoch": 1.3535684987694832,
|
| 732 |
+
"eval_loss": 0.22249621152877808,
|
| 733 |
+
"eval_runtime": 22.5187,
|
| 734 |
+
"eval_samples_per_second": 44.408,
|
| 735 |
+
"eval_steps_per_second": 0.711,
|
| 736 |
+
"step": 1650
|
| 737 |
+
},
|
| 738 |
+
{
|
| 739 |
+
"epoch": 1.3740771123872026,
|
| 740 |
+
"grad_norm": 0.11896856129169464,
|
| 741 |
+
"learning_rate": 0.00027052042647471254,
|
| 742 |
+
"loss": 0.2009,
|
| 743 |
+
"step": 1675
|
| 744 |
+
},
|
| 745 |
+
{
|
| 746 |
+
"epoch": 1.3945857260049221,
|
| 747 |
+
"grad_norm": 0.10503373295068741,
|
| 748 |
+
"learning_rate": 0.0002547685449686206,
|
| 749 |
+
"loss": 0.2095,
|
| 750 |
+
"step": 1700
|
| 751 |
+
},
|
| 752 |
+
{
|
| 753 |
+
"epoch": 1.3945857260049221,
|
| 754 |
+
"eval_loss": 0.22149282693862915,
|
| 755 |
+
"eval_runtime": 22.5508,
|
| 756 |
+
"eval_samples_per_second": 44.344,
|
| 757 |
+
"eval_steps_per_second": 0.71,
|
| 758 |
+
"step": 1700
|
| 759 |
+
},
|
| 760 |
+
{
|
| 761 |
+
"epoch": 1.4150943396226414,
|
| 762 |
+
"grad_norm": 0.11639218777418137,
|
| 763 |
+
"learning_rate": 0.0002393308852631373,
|
| 764 |
+
"loss": 0.2215,
|
| 765 |
+
"step": 1725
|
| 766 |
+
},
|
| 767 |
+
{
|
| 768 |
+
"epoch": 1.435602953240361,
|
| 769 |
+
"grad_norm": 0.11435659229755402,
|
| 770 |
+
"learning_rate": 0.0002242272280557645,
|
| 771 |
+
"loss": 0.2155,
|
| 772 |
+
"step": 1750
|
| 773 |
+
},
|
| 774 |
+
{
|
| 775 |
+
"epoch": 1.435602953240361,
|
| 776 |
+
"eval_loss": 0.2211078256368637,
|
| 777 |
+
"eval_runtime": 22.5499,
|
| 778 |
+
"eval_samples_per_second": 44.346,
|
| 779 |
+
"eval_steps_per_second": 0.71,
|
| 780 |
+
"step": 1750
|
| 781 |
+
},
|
| 782 |
+
{
|
| 783 |
+
"epoch": 1.4561115668580804,
|
| 784 |
+
"grad_norm": 0.11285687983036041,
|
| 785 |
+
"learning_rate": 0.0002094769260774262,
|
| 786 |
+
"loss": 0.2113,
|
| 787 |
+
"step": 1775
|
| 788 |
+
},
|
| 789 |
+
{
|
| 790 |
+
"epoch": 1.4766201804758,
|
| 791 |
+
"grad_norm": 0.10831128060817719,
|
| 792 |
+
"learning_rate": 0.00019509887929528458,
|
| 793 |
+
"loss": 0.2154,
|
| 794 |
+
"step": 1800
|
| 795 |
+
},
|
| 796 |
+
{
|
| 797 |
+
"epoch": 1.4766201804758,
|
| 798 |
+
"eval_loss": 0.22003982961177826,
|
| 799 |
+
"eval_runtime": 22.5404,
|
| 800 |
+
"eval_samples_per_second": 44.365,
|
| 801 |
+
"eval_steps_per_second": 0.71,
|
| 802 |
+
"step": 1800
|
| 803 |
+
},
|
| 804 |
+
{
|
| 805 |
+
"epoch": 1.4971287940935194,
|
| 806 |
+
"grad_norm": 0.13930928707122803,
|
| 807 |
+
"learning_rate": 0.0001811115106956918,
|
| 808 |
+
"loss": 0.2061,
|
| 809 |
+
"step": 1825
|
| 810 |
+
},
|
| 811 |
+
{
|
| 812 |
+
"epoch": 1.5176374077112387,
|
| 813 |
+
"grad_norm": 0.11574142426252365,
|
| 814 |
+
"learning_rate": 0.00016753274267831115,
|
| 815 |
+
"loss": 0.2158,
|
| 816 |
+
"step": 1850
|
| 817 |
+
},
|
| 818 |
+
{
|
| 819 |
+
"epoch": 1.5176374077112387,
|
| 820 |
+
"eval_loss": 0.21911242604255676,
|
| 821 |
+
"eval_runtime": 22.62,
|
| 822 |
+
"eval_samples_per_second": 44.209,
|
| 823 |
+
"eval_steps_per_second": 0.707,
|
| 824 |
+
"step": 1850
|
| 825 |
+
},
|
| 826 |
+
{
|
| 827 |
+
"epoch": 1.5381460213289582,
|
| 828 |
+
"grad_norm": 0.1241801530122757,
|
| 829 |
+
"learning_rate": 0.00015437997409165478,
|
| 830 |
+
"loss": 0.2061,
|
| 831 |
+
"step": 1875
|
| 832 |
+
},
|
| 833 |
+
{
|
| 834 |
+
"epoch": 1.5586546349466777,
|
| 835 |
+
"grad_norm": 0.10792222619056702,
|
| 836 |
+
"learning_rate": 0.00014167005793946035,
|
| 837 |
+
"loss": 0.2055,
|
| 838 |
+
"step": 1900
|
| 839 |
+
},
|
| 840 |
+
{
|
| 841 |
+
"epoch": 1.5586546349466777,
|
| 842 |
+
"eval_loss": 0.21896851062774658,
|
| 843 |
+
"eval_runtime": 22.5247,
|
| 844 |
+
"eval_samples_per_second": 44.396,
|
| 845 |
+
"eval_steps_per_second": 0.71,
|
| 846 |
+
"step": 1900
|
| 847 |
+
},
|
| 848 |
+
{
|
| 849 |
+
"epoch": 1.579163248564397,
|
| 850 |
+
"grad_norm": 0.12615908682346344,
|
| 851 |
+
"learning_rate": 0.00012941927978647527,
|
| 852 |
+
"loss": 0.2078,
|
| 853 |
+
"step": 1925
|
| 854 |
+
},
|
| 855 |
+
{
|
| 856 |
+
"epoch": 1.5996718621821167,
|
| 857 |
+
"grad_norm": 0.0955195277929306,
|
| 858 |
+
"learning_rate": 0.00011764333689131385,
|
| 859 |
+
"loss": 0.2135,
|
| 860 |
+
"step": 1950
|
| 861 |
+
},
|
| 862 |
+
{
|
| 863 |
+
"epoch": 1.5996718621821167,
|
| 864 |
+
"eval_loss": 0.21795088052749634,
|
| 865 |
+
"eval_runtime": 22.5428,
|
| 866 |
+
"eval_samples_per_second": 44.36,
|
| 867 |
+
"eval_steps_per_second": 0.71,
|
| 868 |
+
"step": 1950
|
| 869 |
+
},
|
| 870 |
+
{
|
| 871 |
+
"epoch": 1.620180475799836,
|
| 872 |
+
"grad_norm": 0.09828197956085205,
|
| 873 |
+
"learning_rate": 0.00010635731809312993,
|
| 874 |
+
"loss": 0.2098,
|
| 875 |
+
"step": 1975
|
| 876 |
+
},
|
| 877 |
+
{
|
| 878 |
+
"epoch": 1.6406890894175554,
|
| 879 |
+
"grad_norm": 0.11805617809295654,
|
| 880 |
+
"learning_rate": 9.557568447787201e-05,
|
| 881 |
+
"loss": 0.2099,
|
| 882 |
+
"step": 2000
|
| 883 |
+
},
|
| 884 |
+
{
|
| 885 |
+
"epoch": 1.6406890894175554,
|
| 886 |
+
"eval_loss": 0.21716704964637756,
|
| 887 |
+
"eval_runtime": 22.5329,
|
| 888 |
+
"eval_samples_per_second": 44.38,
|
| 889 |
+
"eval_steps_per_second": 0.71,
|
| 890 |
+
"step": 2000
|
| 891 |
+
},
|
| 892 |
+
{
|
| 893 |
+
"epoch": 1.661197703035275,
|
| 894 |
+
"grad_norm": 0.11299672722816467,
|
| 895 |
+
"learning_rate": 8.531225084889654e-05,
|
| 896 |
+
"loss": 0.2026,
|
| 897 |
+
"step": 2025
|
| 898 |
+
},
|
| 899 |
+
{
|
| 900 |
+
"epoch": 1.6817063166529942,
|
| 901 |
+
"grad_norm": 0.11950846761465073,
|
| 902 |
+
"learning_rate": 7.558016802568091e-05,
|
| 903 |
+
"loss": 0.2062,
|
| 904 |
+
"step": 2050
|
| 905 |
+
},
|
| 906 |
+
{
|
| 907 |
+
"epoch": 1.6817063166529942,
|
| 908 |
+
"eval_loss": 0.21671602129936218,
|
| 909 |
+
"eval_runtime": 22.5667,
|
| 910 |
+
"eval_samples_per_second": 44.313,
|
| 911 |
+
"eval_steps_per_second": 0.709,
|
| 912 |
+
"step": 2050
|
| 913 |
+
},
|
| 914 |
+
{
|
| 915 |
+
"epoch": 1.7022149302707137,
|
| 916 |
+
"grad_norm": 0.11061406135559082,
|
| 917 |
+
"learning_rate": 6.639190599331746e-05,
|
| 918 |
+
"loss": 0.2073,
|
| 919 |
+
"step": 2075
|
| 920 |
+
},
|
| 921 |
+
{
|
| 922 |
+
"epoch": 1.7227235438884332,
|
| 923 |
+
"grad_norm": 0.10908389836549759,
|
| 924 |
+
"learning_rate": 5.775923792437865e-05,
|
| 925 |
+
"loss": 0.2016,
|
| 926 |
+
"step": 2100
|
| 927 |
+
},
|
| 928 |
+
{
|
| 929 |
+
"epoch": 1.7227235438884332,
|
| 930 |
+
"eval_loss": 0.2159593254327774,
|
| 931 |
+
"eval_runtime": 22.5289,
|
| 932 |
+
"eval_samples_per_second": 44.387,
|
| 933 |
+
"eval_steps_per_second": 0.71,
|
| 934 |
+
"step": 2100
|
| 935 |
+
},
|
| 936 |
+
{
|
| 937 |
+
"epoch": 1.7432321575061525,
|
| 938 |
+
"grad_norm": 0.11665470898151398,
|
| 939 |
+
"learning_rate": 4.9693225093627616e-05,
|
| 940 |
+
"loss": 0.2029,
|
| 941 |
+
"step": 2125
|
| 942 |
+
},
|
| 943 |
+
{
|
| 944 |
+
"epoch": 1.7637407711238722,
|
| 945 |
+
"grad_norm": 0.11755326390266418,
|
| 946 |
+
"learning_rate": 4.220420270490294e-05,
|
| 947 |
+
"loss": 0.1974,
|
| 948 |
+
"step": 2150
|
| 949 |
+
},
|
| 950 |
+
{
|
| 951 |
+
"epoch": 1.7637407711238722,
|
| 952 |
+
"eval_loss": 0.21538600325584412,
|
| 953 |
+
"eval_runtime": 22.6212,
|
| 954 |
+
"eval_samples_per_second": 44.206,
|
| 955 |
+
"eval_steps_per_second": 0.707,
|
| 956 |
+
"step": 2150
|
| 957 |
+
},
|
| 958 |
+
{
|
| 959 |
+
"epoch": 1.7842493847415914,
|
| 960 |
+
"grad_norm": 0.12463412433862686,
|
| 961 |
+
"learning_rate": 3.530176664833834e-05,
|
| 962 |
+
"loss": 0.1999,
|
| 963 |
+
"step": 2175
|
| 964 |
+
},
|
| 965 |
+
{
|
| 966 |
+
"epoch": 1.804757998359311,
|
| 967 |
+
"grad_norm": 0.13732177019119263,
|
| 968 |
+
"learning_rate": 2.8994761204884756e-05,
|
| 969 |
+
"loss": 0.2084,
|
| 970 |
+
"step": 2200
|
| 971 |
+
},
|
| 972 |
+
{
|
| 973 |
+
"epoch": 1.804757998359311,
|
| 974 |
+
"eval_loss": 0.2153223305940628,
|
| 975 |
+
"eval_runtime": 22.5592,
|
| 976 |
+
"eval_samples_per_second": 44.328,
|
| 977 |
+
"eval_steps_per_second": 0.709,
|
| 978 |
+
"step": 2200
|
| 979 |
+
},
|
| 980 |
+
{
|
| 981 |
+
"epoch": 1.8252666119770304,
|
| 982 |
+
"grad_norm": 0.125973179936409,
|
| 983 |
+
"learning_rate": 2.329126771388995e-05,
|
| 984 |
+
"loss": 0.1998,
|
| 985 |
+
"step": 2225
|
| 986 |
+
},
|
| 987 |
+
{
|
| 988 |
+
"epoch": 1.8457752255947497,
|
| 989 |
+
"grad_norm": 0.10488861054182053,
|
| 990 |
+
"learning_rate": 1.8198594218256815e-05,
|
| 991 |
+
"loss": 0.2009,
|
| 992 |
+
"step": 2250
|
| 993 |
+
},
|
| 994 |
+
{
|
| 995 |
+
"epoch": 1.8457752255947497,
|
| 996 |
+
"eval_loss": 0.2151503562927246,
|
| 997 |
+
"eval_runtime": 22.5499,
|
| 998 |
+
"eval_samples_per_second": 44.346,
|
| 999 |
+
"eval_steps_per_second": 0.71,
|
| 1000 |
+
"step": 2250
|
| 1001 |
+
},
|
| 1002 |
+
{
|
| 1003 |
+
"epoch": 1.8662838392124692,
|
| 1004 |
+
"grad_norm": 0.11978505551815033,
|
| 1005 |
+
"learning_rate": 1.3723266100447052e-05,
|
| 1006 |
+
"loss": 0.204,
|
| 1007 |
+
"step": 2275
|
| 1008 |
+
},
|
| 1009 |
+
{
|
| 1010 |
+
"epoch": 1.8867924528301887,
|
| 1011 |
+
"grad_norm": 0.1408199667930603,
|
| 1012 |
+
"learning_rate": 9.871017721329201e-06,
|
| 1013 |
+
"loss": 0.2122,
|
| 1014 |
+
"step": 2300
|
| 1015 |
+
},
|
| 1016 |
+
{
|
| 1017 |
+
"epoch": 1.8867924528301887,
|
| 1018 |
+
"eval_loss": 0.21504977345466614,
|
| 1019 |
+
"eval_runtime": 22.618,
|
| 1020 |
+
"eval_samples_per_second": 44.213,
|
| 1021 |
+
"eval_steps_per_second": 0.707,
|
| 1022 |
+
"step": 2300
|
| 1023 |
+
},
|
| 1024 |
+
{
|
| 1025 |
+
"epoch": 1.907301066447908,
|
| 1026 |
+
"grad_norm": 0.11792837083339691,
|
| 1027 |
+
"learning_rate": 6.646785072584871e-06,
|
| 1028 |
+
"loss": 0.2068,
|
| 1029 |
+
"step": 2325
|
| 1030 |
+
},
|
| 1031 |
+
{
|
| 1032 |
+
"epoch": 1.9278096800656277,
|
| 1033 |
+
"grad_norm": 0.0992947518825531,
|
| 1034 |
+
"learning_rate": 4.054699452086641e-06,
|
| 1035 |
+
"loss": 0.2006,
|
| 1036 |
+
"step": 2350
|
| 1037 |
+
},
|
| 1038 |
+
{
|
| 1039 |
+
"epoch": 1.9278096800656277,
|
| 1040 |
+
"eval_loss": 0.21499131619930267,
|
| 1041 |
+
"eval_runtime": 22.5655,
|
| 1042 |
+
"eval_samples_per_second": 44.315,
|
| 1043 |
+
"eval_steps_per_second": 0.709,
|
| 1044 |
+
"step": 2350
|
| 1045 |
+
},
|
| 1046 |
+
{
|
| 1047 |
+
"epoch": 1.948318293683347,
|
| 1048 |
+
"grad_norm": 0.1237172856926918,
|
| 1049 |
+
"learning_rate": 2.0980821703527886e-06,
|
| 1050 |
+
"loss": 0.2153,
|
| 1051 |
+
"step": 2375
|
| 1052 |
+
},
|
| 1053 |
+
{
|
| 1054 |
+
"epoch": 1.9688269073010665,
|
| 1055 |
+
"grad_norm": 0.11623578518629074,
|
| 1056 |
+
"learning_rate": 7.794402948607671e-07,
|
| 1057 |
+
"loss": 0.2113,
|
| 1058 |
+
"step": 2400
|
| 1059 |
+
},
|
| 1060 |
+
{
|
| 1061 |
+
"epoch": 1.9688269073010665,
|
| 1062 |
+
"eval_loss": 0.21487931907176971,
|
| 1063 |
+
"eval_runtime": 22.5687,
|
| 1064 |
+
"eval_samples_per_second": 44.309,
|
| 1065 |
+
"eval_steps_per_second": 0.709,
|
| 1066 |
+
"step": 2400
|
| 1067 |
+
},
|
| 1068 |
+
{
|
| 1069 |
+
"epoch": 1.989335520918786,
|
| 1070 |
+
"grad_norm": 0.14077042043209076,
|
| 1071 |
+
"learning_rate": 1.0046343767294853e-07,
|
| 1072 |
+
"loss": 0.2009,
|
| 1073 |
+
"step": 2425
|
| 1074 |
+
},
|
| 1075 |
+
{
|
| 1076 |
+
"epoch": 2.0,
|
| 1077 |
+
"step": 2438,
|
| 1078 |
+
"total_flos": 1.58523627405312e+18,
|
| 1079 |
+
"train_loss": 0.2502285066053846,
|
| 1080 |
+
"train_runtime": 4294.1547,
|
| 1081 |
+
"train_samples_per_second": 18.164,
|
| 1082 |
+
"train_steps_per_second": 0.568
|
| 1083 |
+
}
|
| 1084 |
+
],
|
| 1085 |
+
"logging_steps": 25,
|
| 1086 |
+
"max_steps": 2438,
|
| 1087 |
+
"num_input_tokens_seen": 0,
|
| 1088 |
+
"num_train_epochs": 2,
|
| 1089 |
+
"save_steps": 500,
|
| 1090 |
+
"stateful_callbacks": {
|
| 1091 |
+
"TrainerControl": {
|
| 1092 |
+
"args": {
|
| 1093 |
+
"should_epoch_stop": false,
|
| 1094 |
+
"should_evaluate": false,
|
| 1095 |
+
"should_log": false,
|
| 1096 |
+
"should_save": true,
|
| 1097 |
+
"should_training_stop": true
|
| 1098 |
+
},
|
| 1099 |
+
"attributes": {}
|
| 1100 |
+
}
|
| 1101 |
+
},
|
| 1102 |
+
"total_flos": 1.58523627405312e+18,
|
| 1103 |
+
"train_batch_size": 32,
|
| 1104 |
+
"trial_name": null,
|
| 1105 |
+
"trial_params": null
|
| 1106 |
+
}
|