Hugging Face's logo

psp-dada
/
LLaVA-v1.6-Vicuna-7B-SENTINEL

Image-Text-to-Text
Transformers
Safetensors
English
lora
llava
conversational
Model card Files
xet
Community
1
psp_dada commited on
Commit
181bcd7
·
1 Parent(s): 06d2098

add file

Browse files
Files changed (15) hide show
  1. adapter_config.json +26 -0
  2. adapter_model.safetensors +3 -0
  3. added_tokens.json +3 -0
  4. all_results.json +8 -0
  5. chat_template.json +3 -0
  6. preprocessor_config.json +52 -0
  7. processor_config.json +6 -0
  8. special_tokens_map.json +30 -0
  9. tokenizer.json +0 -0
  10. tokenizer.model +3 -0
  11. tokenizer_config.json +58 -0
  12. train_results.json +8 -0
  13. trainer_log.jsonl +110 -0
  14. trainer_state.json +1677 -0
  15. training_args.bin +3 -0
adapter_config.json ADDED
@@ -0,0 +1,26 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alpha_pattern": {},
3
+ "auto_mapping": null,
4
+ "base_model_name_or_path": "/psp/llm/utils/models/repo/llava-v1.6-vicuna-7b-hf",
5
+ "bias": "none",
6
+ "fan_in_fan_out": false,
7
+ "inference_mode": true,
8
+ "init_lora_weights": true,
9
+ "layer_replication": null,
10
+ "layers_pattern": null,
11
+ "layers_to_transform": null,
12
+ "loftq_config": {},
13
+ "lora_alpha": 256,
14
+ "lora_dropout": 0.0,
15
+ "megatron_config": null,
16
+ "megatron_core": "megatron.core",
17
+ "modules_to_save": null,
18
+ "peft_type": "LORA",
19
+ "r": 128,
20
+ "rank_pattern": {},
21
+ "revision": null,
22
+ "target_modules": "^(?!.*vision_tower).*(?:v_proj|gate_proj|q_proj|o_proj|k_proj|down_proj|up_proj).*",
23
+ "task_type": "CAUSAL_LM",
24
+ "use_dora": false,
25
+ "use_rslora": false
26
+ }
adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fa280d6fe41589e12eacc4894606895f965366336678018e855fa10f45cba40a
3
+ size 639699488
added_tokens.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ {
2
+ "<image>": 32000
3
+ }
all_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 0.9965714285714286,
3
+ "total_flos": 1.5002634544501228e+18,
4
+ "train_loss": 0.6781991911590646,
5
+ "train_runtime": 6621.3264,
6
+ "train_samples_per_second": 1.057,
7
+ "train_steps_per_second": 0.016
8
+ }
chat_template.json ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ {
2
+ "chat_template": "{% for message in messages %}{% if message['role'] != 'system' %}{{ message['role'].upper() + ': '}}{% endif %}{# Render all images first #}{% for content in message['content'] | selectattr('type', 'equalto', 'image') %}{{ '<image>\n' }}{% endfor %}{# Render all text next #}{% if message['role'] != 'assistant' %}{% for content in message['content'] | selectattr('type', 'equalto', 'text') %}{{ content['text'] + ' '}}{% endfor %}{% else %}{% for content in message['content'] | selectattr('type', 'equalto', 'text') %}{% generation %}{{ content['text'] + ' '}}{% endgeneration %}{% endfor %}{% endif %}{% endfor %}{% if add_generation_prompt %}{{ 'ASSISTANT:' }}{% endif %}"
3
+ }
preprocessor_config.json ADDED
@@ -0,0 +1,52 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "aspect_ratio_setting": "anyres",
3
+ "crop_size": {
4
+ "height": 336,
5
+ "width": 336
6
+ },
7
+ "do_center_crop": true,
8
+ "do_convert_rgb": true,
9
+ "do_normalize": true,
10
+ "do_pad": true,
11
+ "do_rescale": true,
12
+ "do_resize": true,
13
+ "image_grid_pinpoints": [
14
+ [
15
+ 336,
16
+ 672
17
+ ],
18
+ [
19
+ 672,
20
+ 336
21
+ ],
22
+ [
23
+ 672,
24
+ 672
25
+ ],
26
+ [
27
+ 1008,
28
+ 336
29
+ ],
30
+ [
31
+ 336,
32
+ 1008
33
+ ]
34
+ ],
35
+ "image_mean": [
36
+ 0.48145466,
37
+ 0.4578275,
38
+ 0.40821073
39
+ ],
40
+ "image_processor_type": "LlavaNextImageProcessor",
41
+ "image_std": [
42
+ 0.26862954,
43
+ 0.26130258,
44
+ 0.27577711
45
+ ],
46
+ "processor_class": "LlavaNextProcessor",
47
+ "resample": 3,
48
+ "rescale_factor": 0.00392156862745098,
49
+ "size": {
50
+ "shortest_edge": 336
51
+ }
52
+ }
processor_config.json ADDED
@@ -0,0 +1,6 @@
 
 
 
 
 
 
 
1
+ {
2
+ "image_token": "<image>",
3
+ "patch_size": 14,
4
+ "processor_class": "LlavaNextProcessor",
5
+ "vision_feature_select_strategy": "default"
6
+ }
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": "<unk>",
18
+ "lstrip": false,
19
+ "normalized": false,
20
+ "rstrip": false,
21
+ "single_word": false
22
+ },
23
+ "unk_token": {
24
+ "content": "<unk>",
25
+ "lstrip": false,
26
+ "normalized": false,
27
+ "rstrip": false,
28
+ "single_word": false
29
+ }
30
+ }
tokenizer.json ADDED
The diff for this file is too large to render. See raw diff
 
tokenizer.model ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:9e556afd44213b6bd1be2b850ebbbd98f5481437a8021afaf58ee7fb1818d347
3
+ size 499723
tokenizer_config.json ADDED
@@ -0,0 +1,58 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "add_prefix_space": true,
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": "<image>",
32
+ "lstrip": false,
33
+ "normalized": false,
34
+ "rstrip": false,
35
+ "single_word": false,
36
+ "special": true
37
+ }
38
+ },
39
+ "bos_token": "<s>",
40
+ "chat_template": "{% set system_message = 'A chat between a curious user and an artificial intelligence assistant. The assistant gives helpful, detailed, and polite answers to the user\\'s questions.' %}{% if messages[0]['role'] == 'system' %}{% set loop_messages = messages[1:] %}{% set system_message = messages[0]['content'] %}{% else %}{% set loop_messages = messages %}{% endif %}{% if system_message is defined %}{{ system_message }}{% endif %}{% for message in loop_messages %}{% set content = message['content'] %}{% if message['role'] == 'user' %}{{ 'USER: ' + content + ' ASSISTANT:' }}{% elif message['role'] == 'assistant' %}{{ content + '</s>' }}{% endif %}{% endfor %}",
41
+ "clean_up_tokenization_spaces": false,
42
+ "eos_token": "</s>",
43
+ "extra_special_tokens": {
44
+ "image_token": "<image>"
45
+ },
46
+ "image_token": "<image>",
47
+ "legacy": false,
48
+ "model_max_length": 4096,
49
+ "pad_token": "<unk>",
50
+ "padding_side": "right",
51
+ "processor_class": "LlavaNextProcessor",
52
+ "sp_model_kwargs": {},
53
+ "spaces_between_special_tokens": false,
54
+ "split_special_tokens": false,
55
+ "tokenizer_class": "LlamaTokenizer",
56
+ "unk_token": "<unk>",
57
+ "use_default_system_prompt": false
58
+ }
train_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 0.9965714285714286,
3
+ "total_flos": 1.5002634544501228e+18,
4
+ "train_loss": 0.6781991911590646,
5
+ "train_runtime": 6621.3264,
6
+ "train_samples_per_second": 1.057,
7
+ "train_steps_per_second": 0.016
8
+ }
trainer_log.jsonl ADDED
@@ -0,0 +1,110 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {"current_steps": 1, "total_steps": 109, "loss": 0.6931, "accuracy": 0.0, "lr": 1.999584676323851e-06, "epoch": 0.009142857142857144, "percentage": 0.92, "elapsed_time": "0:00:59", "remaining_time": "1:46:19"}
2
+ {"current_steps": 2, "total_steps": 109, "loss": 0.695, "accuracy": 0.453125, "lr": 1.9983390502829166e-06, "epoch": 0.018285714285714287, "percentage": 1.83, "elapsed_time": "0:01:58", "remaining_time": "1:46:02"}
3
+ {"current_steps": 3, "total_steps": 109, "loss": 0.6918, "accuracy": 0.5625, "lr": 1.996264156553169e-06, "epoch": 0.027428571428571427, "percentage": 2.75, "elapsed_time": "0:03:00", "remaining_time": "1:46:06"}
4
+ {"current_steps": 4, "total_steps": 109, "loss": 0.6928, "accuracy": 0.515625, "lr": 1.9933617186395913e-06, "epoch": 0.036571428571428574, "percentage": 3.67, "elapsed_time": "0:04:02", "remaining_time": "1:46:03"}
5
+ {"current_steps": 5, "total_steps": 109, "loss": 0.6929, "accuracy": 0.484375, "lr": 1.9896341474445524e-06, "epoch": 0.045714285714285714, "percentage": 4.59, "elapsed_time": "0:05:03", "remaining_time": "1:45:05"}
6
+ {"current_steps": 6, "total_steps": 109, "loss": 0.6927, "accuracy": 0.484375, "lr": 1.9850845392651948e-06, "epoch": 0.054857142857142854, "percentage": 5.5, "elapsed_time": "0:06:01", "remaining_time": "1:43:29"}
7
+ {"current_steps": 7, "total_steps": 109, "loss": 0.6928, "accuracy": 0.609375, "lr": 1.9797166732215073e-06, "epoch": 0.064, "percentage": 6.42, "elapsed_time": "0:07:04", "remaining_time": "1:43:02"}
8
+ {"current_steps": 8, "total_steps": 109, "loss": 0.6932, "accuracy": 0.5, "lr": 1.9735350081172067e-06, "epoch": 0.07314285714285715, "percentage": 7.34, "elapsed_time": "0:08:04", "remaining_time": "1:41:50"}
9
+ {"current_steps": 9, "total_steps": 109, "loss": 0.6911, "accuracy": 0.5625, "lr": 1.966544678736044e-06, "epoch": 0.08228571428571428, "percentage": 8.26, "elapsed_time": "0:09:03", "remaining_time": "1:40:43"}
10
+ {"current_steps": 10, "total_steps": 109, "loss": 0.6896, "accuracy": 0.65625, "lr": 1.9587514915766123e-06, "epoch": 0.09142857142857143, "percentage": 9.17, "elapsed_time": "0:10:05", "remaining_time": "1:39:57"}
11
+ {"current_steps": 11, "total_steps": 109, "loss": 0.6908, "accuracy": 0.546875, "lr": 1.9501619200291904e-06, "epoch": 0.10057142857142858, "percentage": 10.09, "elapsed_time": "0:11:05", "remaining_time": "1:38:48"}
12
+ {"current_steps": 12, "total_steps": 109, "loss": 0.69, "accuracy": 0.5625, "lr": 1.9407830989986425e-06, "epoch": 0.10971428571428571, "percentage": 11.01, "elapsed_time": "0:12:07", "remaining_time": "1:37:57"}
13
+ {"current_steps": 13, "total_steps": 109, "loss": 0.6915, "accuracy": 0.546875, "lr": 1.930622818977825e-06, "epoch": 0.11885714285714286, "percentage": 11.93, "elapsed_time": "0:13:08", "remaining_time": "1:37:00"}
14
+ {"current_steps": 14, "total_steps": 109, "loss": 0.6901, "accuracy": 0.59375, "lr": 1.9196895195764358e-06, "epoch": 0.128, "percentage": 12.84, "elapsed_time": "0:14:11", "remaining_time": "1:36:16"}
15
+ {"current_steps": 15, "total_steps": 109, "loss": 0.692, "accuracy": 0.5, "lr": 1.907992282510675e-06, "epoch": 0.13714285714285715, "percentage": 13.76, "elapsed_time": "0:15:10", "remaining_time": "1:35:07"}
16
+ {"current_steps": 16, "total_steps": 109, "loss": 0.6891, "accuracy": 0.640625, "lr": 1.8955408240595392e-06, "epoch": 0.1462857142857143, "percentage": 14.68, "elapsed_time": "0:16:11", "remaining_time": "1:34:06"}
17
+ {"current_steps": 17, "total_steps": 109, "loss": 0.6897, "accuracy": 0.625, "lr": 1.882345486994024e-06, "epoch": 0.15542857142857142, "percentage": 15.6, "elapsed_time": "0:17:12", "remaining_time": "1:33:05"}
18
+ {"current_steps": 18, "total_steps": 109, "loss": 0.6908, "accuracy": 0.640625, "lr": 1.8684172319859257e-06, "epoch": 0.16457142857142856, "percentage": 16.51, "elapsed_time": "0:18:11", "remaining_time": "1:31:56"}
19
+ {"current_steps": 19, "total_steps": 109, "loss": 0.6905, "accuracy": 0.53125, "lr": 1.8537676285033885e-06, "epoch": 0.1737142857142857, "percentage": 17.43, "elapsed_time": "0:19:11", "remaining_time": "1:30:52"}
20
+ {"current_steps": 20, "total_steps": 109, "loss": 0.6892, "accuracy": 0.65625, "lr": 1.8384088452007577e-06, "epoch": 0.18285714285714286, "percentage": 18.35, "elapsed_time": "0:20:13", "remaining_time": "1:29:59"}
21
+ {"current_steps": 21, "total_steps": 109, "loss": 0.6841, "accuracy": 0.84375, "lr": 1.8223536398107172e-06, "epoch": 0.192, "percentage": 19.27, "elapsed_time": "0:21:12", "remaining_time": "1:28:52"}
22
+ {"current_steps": 22, "total_steps": 109, "loss": 0.689, "accuracy": 0.578125, "lr": 1.8056153485471165e-06, "epoch": 0.20114285714285715, "percentage": 20.18, "elapsed_time": "0:22:13", "remaining_time": "1:27:54"}
23
+ {"current_steps": 23, "total_steps": 109, "loss": 0.6881, "accuracy": 0.609375, "lr": 1.788207875027274e-06, "epoch": 0.2102857142857143, "percentage": 21.1, "elapsed_time": "0:23:11", "remaining_time": "1:26:43"}
24
+ {"current_steps": 24, "total_steps": 109, "loss": 0.6867, "accuracy": 0.65625, "lr": 1.7701456787229802e-06, "epoch": 0.21942857142857142, "percentage": 22.02, "elapsed_time": "0:24:11", "remaining_time": "1:25:42"}
25
+ {"current_steps": 25, "total_steps": 109, "loss": 0.6872, "accuracy": 0.65625, "lr": 1.751443762949772e-06, "epoch": 0.22857142857142856, "percentage": 22.94, "elapsed_time": "0:25:12", "remaining_time": "1:24:42"}
26
+ {"current_steps": 26, "total_steps": 109, "loss": 0.6869, "accuracy": 0.640625, "lr": 1.7321176624044686e-06, "epoch": 0.2377142857142857, "percentage": 23.85, "elapsed_time": "0:26:13", "remaining_time": "1:23:41"}
27
+ {"current_steps": 27, "total_steps": 109, "loss": 0.687, "accuracy": 0.703125, "lr": 1.7121834302613185e-06, "epoch": 0.24685714285714286, "percentage": 24.77, "elapsed_time": "0:27:13", "remaining_time": "1:22:40"}
28
+ {"current_steps": 28, "total_steps": 109, "loss": 0.6894, "accuracy": 0.609375, "lr": 1.6916576248374717e-06, "epoch": 0.256, "percentage": 25.69, "elapsed_time": "0:28:16", "remaining_time": "1:21:46"}
29
+ {"current_steps": 29, "total_steps": 109, "loss": 0.6849, "accuracy": 0.65625, "lr": 1.6705572958388573e-06, "epoch": 0.2651428571428571, "percentage": 26.61, "elapsed_time": "0:29:14", "remaining_time": "1:20:39"}
30
+ {"current_steps": 30, "total_steps": 109, "loss": 0.6845, "accuracy": 0.71875, "lr": 1.6488999701978903e-06, "epoch": 0.2742857142857143, "percentage": 27.52, "elapsed_time": "0:30:15", "remaining_time": "1:19:40"}
31
+ {"current_steps": 31, "total_steps": 109, "loss": 0.6849, "accuracy": 0.671875, "lr": 1.6267036375147724e-06, "epoch": 0.2834285714285714, "percentage": 28.44, "elapsed_time": "0:31:11", "remaining_time": "1:18:30"}
32
+ {"current_steps": 32, "total_steps": 109, "loss": 0.6837, "accuracy": 0.75, "lr": 1.6039867351144776e-06, "epoch": 0.2925714285714286, "percentage": 29.36, "elapsed_time": "0:32:10", "remaining_time": "1:17:26"}
33
+ {"current_steps": 33, "total_steps": 109, "loss": 0.6887, "accuracy": 0.609375, "lr": 1.5807681327318369e-06, "epoch": 0.3017142857142857, "percentage": 30.28, "elapsed_time": "0:33:11", "remaining_time": "1:16:26"}
34
+ {"current_steps": 34, "total_steps": 109, "loss": 0.6863, "accuracy": 0.6875, "lr": 1.5570671168374437e-06, "epoch": 0.31085714285714283, "percentage": 31.19, "elapsed_time": "0:34:10", "remaining_time": "1:15:22"}
35
+ {"current_steps": 35, "total_steps": 109, "loss": 0.6875, "accuracy": 0.59375, "lr": 1.5329033746173974e-06, "epoch": 0.32, "percentage": 32.11, "elapsed_time": "0:35:10", "remaining_time": "1:14:21"}
36
+ {"current_steps": 36, "total_steps": 109, "loss": 0.684, "accuracy": 0.734375, "lr": 1.5082969776201945e-06, "epoch": 0.3291428571428571, "percentage": 33.03, "elapsed_time": "0:36:10", "remaining_time": "1:13:20"}
37
+ {"current_steps": 37, "total_steps": 109, "loss": 0.6835, "accuracy": 0.703125, "lr": 1.4832683650843507e-06, "epoch": 0.3382857142857143, "percentage": 33.94, "elapsed_time": "0:37:13", "remaining_time": "1:12:25"}
38
+ {"current_steps": 38, "total_steps": 109, "loss": 0.6852, "accuracy": 0.640625, "lr": 1.4578383269606003e-06, "epoch": 0.3474285714285714, "percentage": 34.86, "elapsed_time": "0:38:12", "remaining_time": "1:11:24"}
39
+ {"current_steps": 39, "total_steps": 109, "loss": 0.6862, "accuracy": 0.609375, "lr": 1.4320279866427795e-06, "epoch": 0.3565714285714286, "percentage": 35.78, "elapsed_time": "0:39:11", "remaining_time": "1:10:21"}
40
+ {"current_steps": 40, "total_steps": 109, "loss": 0.6832, "accuracy": 0.734375, "lr": 1.4058587834217354e-06, "epoch": 0.3657142857142857, "percentage": 36.7, "elapsed_time": "0:40:13", "remaining_time": "1:09:23"}
41
+ {"current_steps": 41, "total_steps": 109, "loss": 0.6789, "accuracy": 0.734375, "lr": 1.3793524546768356e-06, "epoch": 0.37485714285714283, "percentage": 37.61, "elapsed_time": "0:41:08", "remaining_time": "1:08:14"}
42
+ {"current_steps": 42, "total_steps": 109, "loss": 0.6792, "accuracy": 0.75, "lr": 1.3525310178198707e-06, "epoch": 0.384, "percentage": 38.53, "elapsed_time": "0:42:09", "remaining_time": "1:07:15"}
43
+ {"current_steps": 43, "total_steps": 109, "loss": 0.6789, "accuracy": 0.734375, "lr": 1.3254167520063509e-06, "epoch": 0.3931428571428571, "percentage": 39.45, "elapsed_time": "0:43:12", "remaining_time": "1:06:18"}
44
+ {"current_steps": 44, "total_steps": 109, "loss": 0.6815, "accuracy": 0.71875, "lr": 1.2980321796293836e-06, "epoch": 0.4022857142857143, "percentage": 40.37, "elapsed_time": "0:44:11", "remaining_time": "1:05:16"}
45
+ {"current_steps": 45, "total_steps": 109, "loss": 0.6834, "accuracy": 0.75, "lr": 1.2704000476115078e-06, "epoch": 0.4114285714285714, "percentage": 41.28, "elapsed_time": "0:45:12", "remaining_time": "1:04:17"}
46
+ {"current_steps": 46, "total_steps": 109, "loss": 0.6813, "accuracy": 0.6875, "lr": 1.2425433085100224e-06, "epoch": 0.4205714285714286, "percentage": 42.2, "elapsed_time": "0:46:12", "remaining_time": "1:03:17"}
47
+ {"current_steps": 47, "total_steps": 109, "loss": 0.6801, "accuracy": 0.8125, "lr": 1.2144851014515054e-06, "epoch": 0.4297142857142857, "percentage": 43.12, "elapsed_time": "0:47:12", "remaining_time": "1:02:16"}
48
+ {"current_steps": 48, "total_steps": 109, "loss": 0.6821, "accuracy": 0.640625, "lr": 1.1862487329113604e-06, "epoch": 0.43885714285714283, "percentage": 44.04, "elapsed_time": "0:48:14", "remaining_time": "1:01:17"}
49
+ {"current_steps": 49, "total_steps": 109, "loss": 0.6817, "accuracy": 0.6875, "lr": 1.157857657354354e-06, "epoch": 0.448, "percentage": 44.95, "elapsed_time": "0:49:13", "remaining_time": "1:00:16"}
50
+ {"current_steps": 50, "total_steps": 109, "loss": 0.6771, "accuracy": 0.71875, "lr": 1.1293354577522264e-06, "epoch": 0.45714285714285713, "percentage": 45.87, "elapsed_time": "0:50:13", "remaining_time": "0:59:16"}
51
+ {"current_steps": 51, "total_steps": 109, "loss": 0.6767, "accuracy": 0.765625, "lr": 1.1007058259945583e-06, "epoch": 0.4662857142857143, "percentage": 46.79, "elapsed_time": "0:51:15", "remaining_time": "0:58:17"}
52
+ {"current_steps": 52, "total_steps": 109, "loss": 0.6762, "accuracy": 0.828125, "lr": 1.071992543209167e-06, "epoch": 0.4754285714285714, "percentage": 47.71, "elapsed_time": "0:52:17", "remaining_time": "0:57:18"}
53
+ {"current_steps": 53, "total_steps": 109, "loss": 0.6807, "accuracy": 0.71875, "lr": 1.0432194600083739e-06, "epoch": 0.4845714285714286, "percentage": 48.62, "elapsed_time": "0:53:17", "remaining_time": "0:56:18"}
54
+ {"current_steps": 54, "total_steps": 109, "loss": 0.6774, "accuracy": 0.6875, "lr": 1.0144104766775572e-06, "epoch": 0.4937142857142857, "percentage": 49.54, "elapsed_time": "0:54:16", "remaining_time": "0:55:16"}
55
+ {"current_steps": 55, "total_steps": 109, "loss": 0.6781, "accuracy": 0.828125, "lr": 9.85589523322443e-07, "epoch": 0.5028571428571429, "percentage": 50.46, "elapsed_time": "0:55:18", "remaining_time": "0:54:18"}
56
+ {"current_steps": 56, "total_steps": 109, "loss": 0.6736, "accuracy": 0.828125, "lr": 9.567805399916258e-07, "epoch": 0.512, "percentage": 51.38, "elapsed_time": "0:56:19", "remaining_time": "0:53:18"}
57
+ {"current_steps": 57, "total_steps": 109, "loss": 0.6735, "accuracy": 0.75, "lr": 9.280074567908329e-07, "epoch": 0.5211428571428571, "percentage": 52.29, "elapsed_time": "0:57:18", "remaining_time": "0:52:16"}
58
+ {"current_steps": 58, "total_steps": 109, "loss": 0.6759, "accuracy": 0.703125, "lr": 8.992941740054417e-07, "epoch": 0.5302857142857142, "percentage": 53.21, "elapsed_time": "0:58:21", "remaining_time": "0:51:18"}
59
+ {"current_steps": 59, "total_steps": 109, "loss": 0.6751, "accuracy": 0.65625, "lr": 8.706645422477738e-07, "epoch": 0.5394285714285715, "percentage": 54.13, "elapsed_time": "0:59:22", "remaining_time": "0:50:18"}
60
+ {"current_steps": 60, "total_steps": 109, "loss": 0.6724, "accuracy": 0.796875, "lr": 8.42142342645646e-07, "epoch": 0.5485714285714286, "percentage": 55.05, "elapsed_time": "1:00:25", "remaining_time": "0:49:20"}
61
+ {"current_steps": 61, "total_steps": 109, "loss": 0.6766, "accuracy": 0.75, "lr": 8.137512670886396e-07, "epoch": 0.5577142857142857, "percentage": 55.96, "elapsed_time": "1:01:25", "remaining_time": "0:48:20"}
62
+ {"current_steps": 62, "total_steps": 109, "loss": 0.6766, "accuracy": 0.734375, "lr": 7.855148985484945e-07, "epoch": 0.5668571428571428, "percentage": 56.88, "elapsed_time": "1:02:26", "remaining_time": "0:47:19"}
63
+ {"current_steps": 63, "total_steps": 109, "loss": 0.6769, "accuracy": 0.703125, "lr": 7.574566914899778e-07, "epoch": 0.576, "percentage": 57.8, "elapsed_time": "1:03:24", "remaining_time": "0:46:17"}
64
+ {"current_steps": 64, "total_steps": 109, "loss": 0.6712, "accuracy": 0.78125, "lr": 7.29599952388492e-07, "epoch": 0.5851428571428572, "percentage": 58.72, "elapsed_time": "1:04:24", "remaining_time": "0:45:17"}
65
+ {"current_steps": 65, "total_steps": 109, "loss": 0.6749, "accuracy": 0.671875, "lr": 7.019678203706163e-07, "epoch": 0.5942857142857143, "percentage": 59.63, "elapsed_time": "1:05:25", "remaining_time": "0:44:17"}
66
+ {"current_steps": 66, "total_steps": 109, "loss": 0.6739, "accuracy": 0.78125, "lr": 6.74583247993649e-07, "epoch": 0.6034285714285714, "percentage": 60.55, "elapsed_time": "1:06:25", "remaining_time": "0:43:16"}
67
+ {"current_steps": 67, "total_steps": 109, "loss": 0.6759, "accuracy": 0.734375, "lr": 6.474689821801295e-07, "epoch": 0.6125714285714285, "percentage": 61.47, "elapsed_time": "1:07:26", "remaining_time": "0:42:16"}
68
+ {"current_steps": 68, "total_steps": 109, "loss": 0.6736, "accuracy": 0.6875, "lr": 6.206475453231643e-07, "epoch": 0.6217142857142857, "percentage": 62.39, "elapsed_time": "1:08:26", "remaining_time": "0:41:15"}
69
+ {"current_steps": 69, "total_steps": 109, "loss": 0.6666, "accuracy": 0.796875, "lr": 5.941412165782645e-07, "epoch": 0.6308571428571429, "percentage": 63.3, "elapsed_time": "1:09:26", "remaining_time": "0:40:15"}
70
+ {"current_steps": 70, "total_steps": 109, "loss": 0.6701, "accuracy": 0.796875, "lr": 5.679720133572206e-07, "epoch": 0.64, "percentage": 64.22, "elapsed_time": "1:10:28", "remaining_time": "0:39:15"}
71
+ {"current_steps": 71, "total_steps": 109, "loss": 0.6698, "accuracy": 0.765625, "lr": 5.421616730394e-07, "epoch": 0.6491428571428571, "percentage": 65.14, "elapsed_time": "1:11:31", "remaining_time": "0:38:16"}
72
+ {"current_steps": 72, "total_steps": 109, "loss": 0.6795, "accuracy": 0.625, "lr": 5.167316349156494e-07, "epoch": 0.6582857142857143, "percentage": 66.06, "elapsed_time": "1:12:30", "remaining_time": "0:37:15"}
73
+ {"current_steps": 73, "total_steps": 109, "loss": 0.6642, "accuracy": 0.828125, "lr": 4.917030223798057e-07, "epoch": 0.6674285714285715, "percentage": 66.97, "elapsed_time": "1:13:33", "remaining_time": "0:36:16"}
74
+ {"current_steps": 74, "total_steps": 109, "loss": 0.6709, "accuracy": 0.796875, "lr": 4.6709662538260264e-07, "epoch": 0.6765714285714286, "percentage": 67.89, "elapsed_time": "1:14:33", "remaining_time": "0:35:15"}
75
+ {"current_steps": 75, "total_steps": 109, "loss": 0.6715, "accuracy": 0.765625, "lr": 4.429328831625565e-07, "epoch": 0.6857142857142857, "percentage": 68.81, "elapsed_time": "1:15:39", "remaining_time": "0:34:18"}
76
+ {"current_steps": 76, "total_steps": 109, "loss": 0.6779, "accuracy": 0.703125, "lr": 4.192318672681631e-07, "epoch": 0.6948571428571428, "percentage": 69.72, "elapsed_time": "1:16:38", "remaining_time": "0:33:16"}
77
+ {"current_steps": 77, "total_steps": 109, "loss": 0.6658, "accuracy": 0.890625, "lr": 3.9601326488552254e-07, "epoch": 0.704, "percentage": 70.64, "elapsed_time": "1:17:39", "remaining_time": "0:32:16"}
78
+ {"current_steps": 78, "total_steps": 109, "loss": 0.6672, "accuracy": 0.8125, "lr": 3.7329636248522746e-07, "epoch": 0.7131428571428572, "percentage": 71.56, "elapsed_time": "1:18:38", "remaining_time": "0:31:15"}
79
+ {"current_steps": 79, "total_steps": 109, "loss": 0.6703, "accuracy": 0.796875, "lr": 3.511000298021097e-07, "epoch": 0.7222857142857143, "percentage": 72.48, "elapsed_time": "1:19:42", "remaining_time": "0:30:16"}
80
+ {"current_steps": 80, "total_steps": 109, "loss": 0.664, "accuracy": 0.828125, "lr": 3.294427041611425e-07, "epoch": 0.7314285714285714, "percentage": 73.39, "elapsed_time": "1:20:44", "remaining_time": "0:29:16"}
81
+ {"current_steps": 81, "total_steps": 109, "loss": 0.664, "accuracy": 0.84375, "lr": 3.0834237516252816e-07, "epoch": 0.7405714285714285, "percentage": 74.31, "elapsed_time": "1:21:43", "remaining_time": "0:28:14"}
82
+ {"current_steps": 82, "total_steps": 109, "loss": 0.6701, "accuracy": 0.734375, "lr": 2.8781656973868117e-07, "epoch": 0.7497142857142857, "percentage": 75.23, "elapsed_time": "1:22:44", "remaining_time": "0:27:14"}
83
+ {"current_steps": 83, "total_steps": 109, "loss": 0.6708, "accuracy": 0.828125, "lr": 2.6788233759553134e-07, "epoch": 0.7588571428571429, "percentage": 76.15, "elapsed_time": "1:23:44", "remaining_time": "0:26:13"}
84
+ {"current_steps": 84, "total_steps": 109, "loss": 0.6753, "accuracy": 0.703125, "lr": 2.485562370502279e-07, "epoch": 0.768, "percentage": 77.06, "elapsed_time": "1:24:47", "remaining_time": "0:25:14"}
85
+ {"current_steps": 85, "total_steps": 109, "loss": 0.6708, "accuracy": 0.71875, "lr": 2.2985432127701941e-07, "epoch": 0.7771428571428571, "percentage": 77.98, "elapsed_time": "1:25:47", "remaining_time": "0:24:13"}
86
+ {"current_steps": 86, "total_steps": 109, "loss": 0.6646, "accuracy": 0.828125, "lr": 2.117921249727258e-07, "epoch": 0.7862857142857143, "percentage": 78.9, "elapsed_time": "1:26:47", "remaining_time": "0:23:12"}
87
+ {"current_steps": 87, "total_steps": 109, "loss": 0.6644, "accuracy": 0.765625, "lr": 1.9438465145288375e-07, "epoch": 0.7954285714285714, "percentage": 79.82, "elapsed_time": "1:27:46", "remaining_time": "0:22:11"}
88
+ {"current_steps": 88, "total_steps": 109, "loss": 0.6605, "accuracy": 0.8125, "lr": 1.7764636018928247e-07, "epoch": 0.8045714285714286, "percentage": 80.73, "elapsed_time": "1:28:49", "remaining_time": "0:21:11"}
89
+ {"current_steps": 89, "total_steps": 109, "loss": 0.6703, "accuracy": 0.78125, "lr": 1.6159115479924256e-07, "epoch": 0.8137142857142857, "percentage": 81.65, "elapsed_time": "1:29:51", "remaining_time": "0:20:11"}
90
+ {"current_steps": 90, "total_steps": 109, "loss": 0.6722, "accuracy": 0.765625, "lr": 1.4623237149661137e-07, "epoch": 0.8228571428571428, "percentage": 82.57, "elapsed_time": "1:30:51", "remaining_time": "0:19:10"}
91
+ {"current_steps": 91, "total_steps": 109, "loss": 0.6724, "accuracy": 0.71875, "lr": 1.315827680140743e-07, "epoch": 0.832, "percentage": 83.49, "elapsed_time": "1:31:49", "remaining_time": "0:18:09"}
92
+ {"current_steps": 92, "total_steps": 109, "loss": 0.6568, "accuracy": 0.875, "lr": 1.1765451300597574e-07, "epoch": 0.8411428571428572, "percentage": 84.4, "elapsed_time": "1:32:51", "remaining_time": "0:17:09"}
93
+ {"current_steps": 93, "total_steps": 109, "loss": 0.6663, "accuracy": 0.78125, "lr": 1.0445917594046072e-07, "epoch": 0.8502857142857143, "percentage": 85.32, "elapsed_time": "1:33:52", "remaining_time": "0:16:09"}
94
+ {"current_steps": 94, "total_steps": 109, "loss": 0.6718, "accuracy": 0.8125, "lr": 9.200771748932512e-08, "epoch": 0.8594285714285714, "percentage": 86.24, "elapsed_time": "1:34:53", "remaining_time": "0:15:08"}
95
+ {"current_steps": 95, "total_steps": 109, "loss": 0.6697, "accuracy": 0.8125, "lr": 8.031048042356392e-08, "epoch": 0.8685714285714285, "percentage": 87.16, "elapsed_time": "1:35:56", "remaining_time": "0:14:08"}
96
+ {"current_steps": 96, "total_steps": 109, "loss": 0.6672, "accuracy": 0.84375, "lr": 6.93771810221746e-08, "epoch": 0.8777142857142857, "percentage": 88.07, "elapsed_time": "1:36:57", "remaining_time": "0:13:07"}
97
+ {"current_steps": 97, "total_steps": 109, "loss": 0.6699, "accuracy": 0.765625, "lr": 5.9216901001357124e-08, "epoch": 0.8868571428571429, "percentage": 88.99, "elapsed_time": "1:37:58", "remaining_time": "0:12:07"}
98
+ {"current_steps": 98, "total_steps": 109, "loss": 0.6775, "accuracy": 0.671875, "lr": 4.9838079970809245e-08, "epoch": 0.896, "percentage": 89.91, "elapsed_time": "1:39:02", "remaining_time": "0:11:06"}
99
+ {"current_steps": 99, "total_steps": 109, "loss": 0.6716, "accuracy": 0.6875, "lr": 4.124850842338779e-08, "epoch": 0.9051428571428571, "percentage": 90.83, "elapsed_time": "1:40:02", "remaining_time": "0:10:06"}
100
+ {"current_steps": 100, "total_steps": 109, "loss": 0.6686, "accuracy": 0.765625, "lr": 3.345532126395578e-08, "epoch": 0.9142857142857143, "percentage": 91.74, "elapsed_time": "1:41:01", "remaining_time": "0:09:05"}
101
+ {"current_steps": 101, "total_steps": 109, "loss": 0.6659, "accuracy": 0.8125, "lr": 2.6464991882793275e-08, "epoch": 0.9234285714285714, "percentage": 92.66, "elapsed_time": "1:42:01", "remaining_time": "0:08:04"}
102
+ {"current_steps": 102, "total_steps": 109, "loss": 0.6669, "accuracy": 0.765625, "lr": 2.028332677849254e-08, "epoch": 0.9325714285714286, "percentage": 93.58, "elapsed_time": "1:43:03", "remaining_time": "0:07:04"}
103
+ {"current_steps": 103, "total_steps": 109, "loss": 0.6654, "accuracy": 0.828125, "lr": 1.4915460734805098e-08, "epoch": 0.9417142857142857, "percentage": 94.5, "elapsed_time": "1:44:03", "remaining_time": "0:06:03"}
104
+ {"current_steps": 104, "total_steps": 109, "loss": 0.6698, "accuracy": 0.765625, "lr": 1.0365852555447641e-08, "epoch": 0.9508571428571428, "percentage": 95.41, "elapsed_time": "1:45:01", "remaining_time": "0:05:02"}
105
+ {"current_steps": 105, "total_steps": 109, "loss": 0.6712, "accuracy": 0.8125, "lr": 6.6382813604083375e-09, "epoch": 0.96, "percentage": 96.33, "elapsed_time": "1:46:05", "remaining_time": "0:04:02"}
106
+ {"current_steps": 106, "total_steps": 109, "loss": 0.6641, "accuracy": 0.84375, "lr": 3.735843446830866e-09, "epoch": 0.9691428571428572, "percentage": 97.25, "elapsed_time": "1:47:05", "remaining_time": "0:03:01"}
107
+ {"current_steps": 107, "total_steps": 109, "loss": 0.664, "accuracy": 0.75, "lr": 1.6609497170834153e-09, "epoch": 0.9782857142857143, "percentage": 98.17, "elapsed_time": "1:48:09", "remaining_time": "0:02:01"}
108
+ {"current_steps": 108, "total_steps": 109, "loss": 0.6604, "accuracy": 0.796875, "lr": 4.153236761488266e-10, "epoch": 0.9874285714285714, "percentage": 99.08, "elapsed_time": "1:49:08", "remaining_time": "0:01:00"}
109
+ {"current_steps": 109, "total_steps": 109, "loss": 0.6703, "accuracy": 0.765625, "lr": 0.0, "epoch": 0.9965714285714286, "percentage": 100.0, "elapsed_time": "1:50:08", "remaining_time": "0:00:00"}
110
+ {"current_steps": 109, "total_steps": 109, "epoch": 0.9965714285714286, "percentage": 100.0, "elapsed_time": "1:50:21", "remaining_time": "0:00:00"}
trainer_state.json ADDED
@@ -0,0 +1,1677 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 0.9965714285714286,
5
+ "eval_steps": 500,
6
+ "global_step": 109,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.009142857142857144,
13
+ "grad_norm": 0.8219454288482666,
14
+ "learning_rate": 1.999584676323851e-06,
15
+ "logits/chosen": -0.4555240273475647,
16
+ "logits/rejected": -0.4536416828632355,
17
+ "logps/chosen": -20.189544677734375,
18
+ "logps/rejected": -23.271862030029297,
19
+ "loss": 0.6931,
20
+ "rewards/accuracies": 0.0,
21
+ "rewards/chosen": 0.0,
22
+ "rewards/margins": 0.0,
23
+ "rewards/rejected": 0.0,
24
+ "step": 1
25
+ },
26
+ {
27
+ "epoch": 0.018285714285714287,
28
+ "grad_norm": 0.8780603408813477,
29
+ "learning_rate": 1.9983390502829166e-06,
30
+ "logits/chosen": -0.5482959151268005,
31
+ "logits/rejected": -0.5472446084022522,
32
+ "logps/chosen": -20.192975997924805,
33
+ "logps/rejected": -24.76065444946289,
34
+ "loss": 0.695,
35
+ "rewards/accuracies": 0.453125,
36
+ "rewards/chosen": -0.0027601076290011406,
37
+ "rewards/margins": -0.00368416216224432,
38
+ "rewards/rejected": 0.0009240537183359265,
39
+ "step": 2
40
+ },
41
+ {
42
+ "epoch": 0.027428571428571427,
43
+ "grad_norm": 0.8463388085365295,
44
+ "learning_rate": 1.996264156553169e-06,
45
+ "logits/chosen": -0.408035010099411,
46
+ "logits/rejected": -0.40724092721939087,
47
+ "logps/chosen": -18.642845153808594,
48
+ "logps/rejected": -21.026813507080078,
49
+ "loss": 0.6918,
50
+ "rewards/accuracies": 0.5625,
51
+ "rewards/chosen": -0.000891003874130547,
52
+ "rewards/margins": 0.0028043005149811506,
53
+ "rewards/rejected": -0.003695304039865732,
54
+ "step": 3
55
+ },
56
+ {
57
+ "epoch": 0.036571428571428574,
58
+ "grad_norm": 0.8569069504737854,
59
+ "learning_rate": 1.9933617186395913e-06,
60
+ "logits/chosen": -0.4994319677352905,
61
+ "logits/rejected": -0.4975876808166504,
62
+ "logps/chosen": -19.638952255249023,
63
+ "logps/rejected": -24.824777603149414,
64
+ "loss": 0.6928,
65
+ "rewards/accuracies": 0.515625,
66
+ "rewards/chosen": -0.0004311114316806197,
67
+ "rewards/margins": 0.0007929832790978253,
68
+ "rewards/rejected": -0.0012240945361554623,
69
+ "step": 4
70
+ },
71
+ {
72
+ "epoch": 0.045714285714285714,
73
+ "grad_norm": 0.8062121272087097,
74
+ "learning_rate": 1.9896341474445524e-06,
75
+ "logits/chosen": -0.5139641165733337,
76
+ "logits/rejected": -0.5159074068069458,
77
+ "logps/chosen": -20.6983642578125,
78
+ "logps/rejected": -21.59255027770996,
79
+ "loss": 0.6929,
80
+ "rewards/accuracies": 0.484375,
81
+ "rewards/chosen": 2.0562736608553678e-05,
82
+ "rewards/margins": 0.0006297557847574353,
83
+ "rewards/rejected": -0.0006091928808018565,
84
+ "step": 5
85
+ },
86
+ {
87
+ "epoch": 0.054857142857142854,
88
+ "grad_norm": 0.8077049851417542,
89
+ "learning_rate": 1.9850845392651948e-06,
90
+ "logits/chosen": -0.4999333620071411,
91
+ "logits/rejected": -0.4988541305065155,
92
+ "logps/chosen": -20.66352081298828,
93
+ "logps/rejected": -22.273176193237305,
94
+ "loss": 0.6927,
95
+ "rewards/accuracies": 0.484375,
96
+ "rewards/chosen": -0.0015631156275048852,
97
+ "rewards/margins": 0.0009301065583713353,
98
+ "rewards/rejected": -0.0024932222440838814,
99
+ "step": 6
100
+ },
101
+ {
102
+ "epoch": 0.064,
103
+ "grad_norm": 0.792171835899353,
104
+ "learning_rate": 1.9797166732215073e-06,
105
+ "logits/chosen": -0.47023099660873413,
106
+ "logits/rejected": -0.4678763151168823,
107
+ "logps/chosen": -18.65442657470703,
108
+ "logps/rejected": -22.887775421142578,
109
+ "loss": 0.6928,
110
+ "rewards/accuracies": 0.609375,
111
+ "rewards/chosen": 0.0008359567727893591,
112
+ "rewards/margins": 0.0007938339258544147,
113
+ "rewards/rejected": 4.212261410430074e-05,
114
+ "step": 7
115
+ },
116
+ {
117
+ "epoch": 0.07314285714285715,
118
+ "grad_norm": 0.8115561008453369,
119
+ "learning_rate": 1.9735350081172067e-06,
120
+ "logits/chosen": -0.43957412242889404,
121
+ "logits/rejected": -0.4355510473251343,
122
+ "logps/chosen": -21.491243362426758,
123
+ "logps/rejected": -24.725614547729492,
124
+ "loss": 0.6932,
125
+ "rewards/accuracies": 0.5,
126
+ "rewards/chosen": 0.0008824847172945738,
127
+ "rewards/margins": 7.572164759039879e-05,
128
+ "rewards/rejected": 0.0008067631861194968,
129
+ "step": 8
130
+ },
131
+ {
132
+ "epoch": 0.08228571428571428,
133
+ "grad_norm": 0.8728166818618774,
134
+ "learning_rate": 1.966544678736044e-06,
135
+ "logits/chosen": -0.516644299030304,
136
+ "logits/rejected": -0.5143158435821533,
137
+ "logps/chosen": -19.448204040527344,
138
+ "logps/rejected": -23.218055725097656,
139
+ "loss": 0.6911,
140
+ "rewards/accuracies": 0.5625,
141
+ "rewards/chosen": 0.0004068721318617463,
142
+ "rewards/margins": 0.004237363114953041,
143
+ "rewards/rejected": -0.0038304910995066166,
144
+ "step": 9
145
+ },
146
+ {
147
+ "epoch": 0.09142857142857143,
148
+ "grad_norm": 0.8181026577949524,
149
+ "learning_rate": 1.9587514915766123e-06,
150
+ "logits/chosen": -0.5233340263366699,
151
+ "logits/rejected": -0.5205311179161072,
152
+ "logps/chosen": -19.936290740966797,
153
+ "logps/rejected": -24.24879264831543,
154
+ "loss": 0.6896,
155
+ "rewards/accuracies": 0.65625,
156
+ "rewards/chosen": 0.00298746838234365,
157
+ "rewards/margins": 0.0072690267115831375,
158
+ "rewards/rejected": -0.004281558096408844,
159
+ "step": 10
160
+ },
161
+ {
162
+ "epoch": 0.10057142857142858,
163
+ "grad_norm": 0.8092432022094727,
164
+ "learning_rate": 1.9501619200291904e-06,
165
+ "logits/chosen": -0.5212205648422241,
166
+ "logits/rejected": -0.5186525583267212,
167
+ "logps/chosen": -18.66614532470703,
168
+ "logps/rejected": -22.687854766845703,
169
+ "loss": 0.6908,
170
+ "rewards/accuracies": 0.546875,
171
+ "rewards/chosen": 0.0014528490137308836,
172
+ "rewards/margins": 0.0048631783574819565,
173
+ "rewards/rejected": -0.0034103295765817165,
174
+ "step": 11
175
+ },
176
+ {
177
+ "epoch": 0.10971428571428571,
178
+ "grad_norm": 0.8379883170127869,
179
+ "learning_rate": 1.9407830989986425e-06,
180
+ "logits/chosen": -0.4597111940383911,
181
+ "logits/rejected": -0.45854365825653076,
182
+ "logps/chosen": -20.127107620239258,
183
+ "logps/rejected": -23.143390655517578,
184
+ "loss": 0.69,
185
+ "rewards/accuracies": 0.5625,
186
+ "rewards/chosen": 0.0038777701556682587,
187
+ "rewards/margins": 0.006433242931962013,
188
+ "rewards/rejected": -0.002555472543463111,
189
+ "step": 12
190
+ },
191
+ {
192
+ "epoch": 0.11885714285714286,
193
+ "grad_norm": 0.8451613783836365,
194
+ "learning_rate": 1.930622818977825e-06,
195
+ "logits/chosen": -0.4704110026359558,
196
+ "logits/rejected": -0.4691443145275116,
197
+ "logps/chosen": -20.862239837646484,
198
+ "logps/rejected": -22.899272918701172,
199
+ "loss": 0.6915,
200
+ "rewards/accuracies": 0.546875,
201
+ "rewards/chosen": 0.00038755848072469234,
202
+ "rewards/margins": 0.0034983016084879637,
203
+ "rewards/rejected": -0.0031107431277632713,
204
+ "step": 13
205
+ },
206
+ {
207
+ "epoch": 0.128,
208
+ "grad_norm": 0.8429624438285828,
209
+ "learning_rate": 1.9196895195764358e-06,
210
+ "logits/chosen": -0.5267453193664551,
211
+ "logits/rejected": -0.5237237215042114,
212
+ "logps/chosen": -20.44851303100586,
213
+ "logps/rejected": -25.33282470703125,
214
+ "loss": 0.6901,
215
+ "rewards/accuracies": 0.59375,
216
+ "rewards/chosen": 0.0021579659078270197,
217
+ "rewards/margins": 0.0063264258205890656,
218
+ "rewards/rejected": -0.004168459679931402,
219
+ "step": 14
220
+ },
221
+ {
222
+ "epoch": 0.13714285714285715,
223
+ "grad_norm": 0.8316566348075867,
224
+ "learning_rate": 1.907992282510675e-06,
225
+ "logits/chosen": -0.5031190514564514,
226
+ "logits/rejected": -0.5018728971481323,
227
+ "logps/chosen": -21.15545082092285,
228
+ "logps/rejected": -22.290752410888672,
229
+ "loss": 0.692,
230
+ "rewards/accuracies": 0.5,
231
+ "rewards/chosen": -0.0004379315651021898,
232
+ "rewards/margins": 0.002324428642168641,
233
+ "rewards/rejected": -0.0027623602654784918,
234
+ "step": 15
235
+ },
236
+ {
237
+ "epoch": 0.1462857142857143,
238
+ "grad_norm": 0.798302412033081,
239
+ "learning_rate": 1.8955408240595392e-06,
240
+ "logits/chosen": -0.4640355706214905,
241
+ "logits/rejected": -0.46341395378112793,
242
+ "logps/chosen": -17.664894104003906,
243
+ "logps/rejected": -23.190065383911133,
244
+ "loss": 0.6891,
245
+ "rewards/accuracies": 0.640625,
246
+ "rewards/chosen": 0.001957263797521591,
247
+ "rewards/margins": 0.008238941431045532,
248
+ "rewards/rejected": -0.006281676702201366,
249
+ "step": 16
250
+ },
251
+ {
252
+ "epoch": 0.15542857142857142,
253
+ "grad_norm": 0.8422457575798035,
254
+ "learning_rate": 1.882345486994024e-06,
255
+ "logits/chosen": -0.5235487818717957,
256
+ "logits/rejected": -0.5194670557975769,
257
+ "logps/chosen": -19.45635414123535,
258
+ "logps/rejected": -22.29392433166504,
259
+ "loss": 0.6897,
260
+ "rewards/accuracies": 0.625,
261
+ "rewards/chosen": -0.0012477637501433492,
262
+ "rewards/margins": 0.007098089437931776,
263
+ "rewards/rejected": -0.008345852605998516,
264
+ "step": 17
265
+ },
266
+ {
267
+ "epoch": 0.16457142857142856,
268
+ "grad_norm": 0.8501055836677551,
269
+ "learning_rate": 1.8684172319859257e-06,
270
+ "logits/chosen": -0.47542956471443176,
271
+ "logits/rejected": -0.47601693868637085,
272
+ "logps/chosen": -19.992841720581055,
273
+ "logps/rejected": -21.64902687072754,
274
+ "loss": 0.6908,
275
+ "rewards/accuracies": 0.640625,
276
+ "rewards/chosen": 0.0012793272035196424,
277
+ "rewards/margins": 0.004879676271229982,
278
+ "rewards/rejected": -0.003600348485633731,
279
+ "step": 18
280
+ },
281
+ {
282
+ "epoch": 0.1737142857142857,
283
+ "grad_norm": 0.8320617079734802,
284
+ "learning_rate": 1.8537676285033885e-06,
285
+ "logits/chosen": -0.4839419722557068,
286
+ "logits/rejected": -0.48214980959892273,
287
+ "logps/chosen": -20.186752319335938,
288
+ "logps/rejected": -24.680709838867188,
289
+ "loss": 0.6905,
290
+ "rewards/accuracies": 0.53125,
291
+ "rewards/chosen": 0.0007302426965907216,
292
+ "rewards/margins": 0.005332631058990955,
293
+ "rewards/rejected": -0.00460238941013813,
294
+ "step": 19
295
+ },
296
+ {
297
+ "epoch": 0.18285714285714286,
298
+ "grad_norm": 0.8264328837394714,
299
+ "learning_rate": 1.8384088452007577e-06,
300
+ "logits/chosen": -0.4898846447467804,
301
+ "logits/rejected": -0.49135124683380127,
302
+ "logps/chosen": -21.396648406982422,
303
+ "logps/rejected": -21.207351684570312,
304
+ "loss": 0.6892,
305
+ "rewards/accuracies": 0.65625,
306
+ "rewards/chosen": 0.0012042367598041892,
307
+ "rewards/margins": 0.008166770450770855,
308
+ "rewards/rejected": -0.006962533108890057,
309
+ "step": 20
310
+ },
311
+ {
312
+ "epoch": 0.192,
313
+ "grad_norm": 0.8570350408554077,
314
+ "learning_rate": 1.8223536398107172e-06,
315
+ "logits/chosen": -0.45695698261260986,
316
+ "logits/rejected": -0.4537597596645355,
317
+ "logps/chosen": -20.424699783325195,
318
+ "logps/rejected": -24.07827377319336,
319
+ "loss": 0.6841,
320
+ "rewards/accuracies": 0.84375,
321
+ "rewards/chosen": 0.005775772035121918,
322
+ "rewards/margins": 0.01844131201505661,
323
+ "rewards/rejected": -0.012665539048612118,
324
+ "step": 21
325
+ },
326
+ {
327
+ "epoch": 0.20114285714285715,
328
+ "grad_norm": 0.8049766421318054,
329
+ "learning_rate": 1.8056153485471165e-06,
330
+ "logits/chosen": -0.4402849078178406,
331
+ "logits/rejected": -0.4379768371582031,
332
+ "logps/chosen": -19.821779251098633,
333
+ "logps/rejected": -21.597673416137695,
334
+ "loss": 0.689,
335
+ "rewards/accuracies": 0.578125,
336
+ "rewards/chosen": 0.002172305481508374,
337
+ "rewards/margins": 0.00856495276093483,
338
+ "rewards/rejected": -0.006392647512257099,
339
+ "step": 22
340
+ },
341
+ {
342
+ "epoch": 0.2102857142857143,
343
+ "grad_norm": 0.8513593673706055,
344
+ "learning_rate": 1.788207875027274e-06,
345
+ "logits/chosen": -0.5087753534317017,
346
+ "logits/rejected": -0.5060349106788635,
347
+ "logps/chosen": -18.349246978759766,
348
+ "logps/rejected": -24.271512985229492,
349
+ "loss": 0.6881,
350
+ "rewards/accuracies": 0.609375,
351
+ "rewards/chosen": 0.003083178075030446,
352
+ "rewards/margins": 0.010339491069316864,
353
+ "rewards/rejected": -0.007256313227117062,
354
+ "step": 23
355
+ },
356
+ {
357
+ "epoch": 0.21942857142857142,
358
+ "grad_norm": 0.870607852935791,
359
+ "learning_rate": 1.7701456787229802e-06,
360
+ "logits/chosen": -0.4925919771194458,
361
+ "logits/rejected": -0.4890042245388031,
362
+ "logps/chosen": -19.655742645263672,
363
+ "logps/rejected": -24.92131233215332,
364
+ "loss": 0.6867,
365
+ "rewards/accuracies": 0.65625,
366
+ "rewards/chosen": 0.0005849468288943172,
367
+ "rewards/margins": 0.01311715692281723,
368
+ "rewards/rejected": -0.012532209977507591,
369
+ "step": 24
370
+ },
371
+ {
372
+ "epoch": 0.22857142857142856,
373
+ "grad_norm": 0.8988638520240784,
374
+ "learning_rate": 1.751443762949772e-06,
375
+ "logits/chosen": -0.49574416875839233,
376
+ "logits/rejected": -0.49408501386642456,
377
+ "logps/chosen": -21.59381675720215,
378
+ "logps/rejected": -23.790117263793945,
379
+ "loss": 0.6872,
380
+ "rewards/accuracies": 0.65625,
381
+ "rewards/chosen": 0.0035057328641414642,
382
+ "rewards/margins": 0.012268707156181335,
383
+ "rewards/rejected": -0.008762974292039871,
384
+ "step": 25
385
+ },
386
+ {
387
+ "epoch": 0.2377142857142857,
388
+ "grad_norm": 0.8565362095832825,
389
+ "learning_rate": 1.7321176624044686e-06,
390
+ "logits/chosen": -0.4831523597240448,
391
+ "logits/rejected": -0.4798797369003296,
392
+ "logps/chosen": -19.747970581054688,
393
+ "logps/rejected": -25.572662353515625,
394
+ "loss": 0.6869,
395
+ "rewards/accuracies": 0.640625,
396
+ "rewards/chosen": 0.0019180128583684564,
397
+ "rewards/margins": 0.012836613692343235,
398
+ "rewards/rejected": -0.010918600484728813,
399
+ "step": 26
400
+ },
401
+ {
402
+ "epoch": 0.24685714285714286,
403
+ "grad_norm": 0.8045784831047058,
404
+ "learning_rate": 1.7121834302613185e-06,
405
+ "logits/chosen": -0.48217910528182983,
406
+ "logits/rejected": -0.4800165593624115,
407
+ "logps/chosen": -20.369915008544922,
408
+ "logps/rejected": -22.05496597290039,
409
+ "loss": 0.687,
410
+ "rewards/accuracies": 0.703125,
411
+ "rewards/chosen": 0.0015595003496855497,
412
+ "rewards/margins": 0.012484083883464336,
413
+ "rewards/rejected": -0.010924583300948143,
414
+ "step": 27
415
+ },
416
+ {
417
+ "epoch": 0.256,
418
+ "grad_norm": 0.7867112755775452,
419
+ "learning_rate": 1.6916576248374717e-06,
420
+ "logits/chosen": -0.44315752387046814,
421
+ "logits/rejected": -0.44253185391426086,
422
+ "logps/chosen": -18.988693237304688,
423
+ "logps/rejected": -21.743547439575195,
424
+ "loss": 0.6894,
425
+ "rewards/accuracies": 0.609375,
426
+ "rewards/chosen": 0.0011231115786358714,
427
+ "rewards/margins": 0.00781021686270833,
428
+ "rewards/rejected": -0.00668710470199585,
429
+ "step": 28
430
+ },
431
+ {
432
+ "epoch": 0.2651428571428571,
433
+ "grad_norm": 0.8259056806564331,
434
+ "learning_rate": 1.6705572958388573e-06,
435
+ "logits/chosen": -0.47697725892066956,
436
+ "logits/rejected": -0.47588181495666504,
437
+ "logps/chosen": -20.31436538696289,
438
+ "logps/rejected": -24.063405990600586,
439
+ "loss": 0.6849,
440
+ "rewards/accuracies": 0.65625,
441
+ "rewards/chosen": 0.0021363100968301296,
442
+ "rewards/margins": 0.01682964153587818,
443
+ "rewards/rejected": -0.014693330973386765,
444
+ "step": 29
445
+ },
446
+ {
447
+ "epoch": 0.2742857142857143,
448
+ "grad_norm": 0.8481204509735107,
449
+ "learning_rate": 1.6488999701978903e-06,
450
+ "logits/chosen": -0.5506658554077148,
451
+ "logits/rejected": -0.5491083264350891,
452
+ "logps/chosen": -20.75564956665039,
453
+ "logps/rejected": -22.990718841552734,
454
+ "loss": 0.6845,
455
+ "rewards/accuracies": 0.71875,
456
+ "rewards/chosen": 0.006822012830525637,
457
+ "rewards/margins": 0.01761447824537754,
458
+ "rewards/rejected": -0.010792464017868042,
459
+ "step": 30
460
+ },
461
+ {
462
+ "epoch": 0.2834285714285714,
463
+ "grad_norm": 0.8275832533836365,
464
+ "learning_rate": 1.6267036375147724e-06,
465
+ "logits/chosen": -0.5217517018318176,
466
+ "logits/rejected": -0.5186524987220764,
467
+ "logps/chosen": -19.507129669189453,
468
+ "logps/rejected": -23.376976013183594,
469
+ "loss": 0.6849,
470
+ "rewards/accuracies": 0.671875,
471
+ "rewards/chosen": 0.003803990315645933,
472
+ "rewards/margins": 0.0168306864798069,
473
+ "rewards/rejected": -0.013026697561144829,
474
+ "step": 31
475
+ },
476
+ {
477
+ "epoch": 0.2925714285714286,
478
+ "grad_norm": 0.8895473480224609,
479
+ "learning_rate": 1.6039867351144776e-06,
480
+ "logits/chosen": -0.4692089557647705,
481
+ "logits/rejected": -0.46629345417022705,
482
+ "logps/chosen": -20.55600357055664,
483
+ "logps/rejected": -23.617862701416016,
484
+ "loss": 0.6837,
485
+ "rewards/accuracies": 0.75,
486
+ "rewards/chosen": 0.003948096185922623,
487
+ "rewards/margins": 0.01930196024477482,
488
+ "rewards/rejected": -0.015353864058852196,
489
+ "step": 32
490
+ },
491
+ {
492
+ "epoch": 0.3017142857142857,
493
+ "grad_norm": 0.8436418175697327,
494
+ "learning_rate": 1.5807681327318369e-06,
495
+ "logits/chosen": -0.5051190257072449,
496
+ "logits/rejected": -0.5041450262069702,
497
+ "logps/chosen": -21.08065414428711,
498
+ "logps/rejected": -24.50140380859375,
499
+ "loss": 0.6887,
500
+ "rewards/accuracies": 0.609375,
501
+ "rewards/chosen": 0.0005654746782965958,
502
+ "rewards/margins": 0.009133580140769482,
503
+ "rewards/rejected": -0.008568106219172478,
504
+ "step": 33
505
+ },
506
+ {
507
+ "epoch": 0.31085714285714283,
508
+ "grad_norm": 0.8358535766601562,
509
+ "learning_rate": 1.5570671168374437e-06,
510
+ "logits/chosen": -0.5785030126571655,
511
+ "logits/rejected": -0.5773332715034485,
512
+ "logps/chosen": -22.108564376831055,
513
+ "logps/rejected": -22.211402893066406,
514
+ "loss": 0.6863,
515
+ "rewards/accuracies": 0.6875,
516
+ "rewards/chosen": 0.0015158435562625527,
517
+ "rewards/margins": 0.014024410396814346,
518
+ "rewards/rejected": -0.012508566491305828,
519
+ "step": 34
520
+ },
521
+ {
522
+ "epoch": 0.32,
523
+ "grad_norm": 0.7899988293647766,
524
+ "learning_rate": 1.5329033746173974e-06,
525
+ "logits/chosen": -0.48546528816223145,
526
+ "logits/rejected": -0.48558929562568665,
527
+ "logps/chosen": -19.545730590820312,
528
+ "logps/rejected": -20.367347717285156,
529
+ "loss": 0.6875,
530
+ "rewards/accuracies": 0.59375,
531
+ "rewards/chosen": 0.00038537243381142616,
532
+ "rewards/margins": 0.011459273286163807,
533
+ "rewards/rejected": -0.011073900386691093,
534
+ "step": 35
535
+ },
536
+ {
537
+ "epoch": 0.3291428571428571,
538
+ "grad_norm": 0.848831057548523,
539
+ "learning_rate": 1.5082969776201945e-06,
540
+ "logits/chosen": -0.46931010484695435,
541
+ "logits/rejected": -0.4679415822029114,
542
+ "logps/chosen": -19.450942993164062,
543
+ "logps/rejected": -23.988311767578125,
544
+ "loss": 0.684,
545
+ "rewards/accuracies": 0.734375,
546
+ "rewards/chosen": 0.0018924175528809428,
547
+ "rewards/margins": 0.01862933486700058,
548
+ "rewards/rejected": -0.016736915335059166,
549
+ "step": 36
550
+ },
551
+ {
552
+ "epoch": 0.3382857142857143,
553
+ "grad_norm": 0.8378761410713196,
554
+ "learning_rate": 1.4832683650843507e-06,
555
+ "logits/chosen": -0.5115276575088501,
556
+ "logits/rejected": -0.5111303329467773,
557
+ "logps/chosen": -20.903900146484375,
558
+ "logps/rejected": -23.84684181213379,
559
+ "loss": 0.6835,
560
+ "rewards/accuracies": 0.703125,
561
+ "rewards/chosen": 0.0019410561071708798,
562
+ "rewards/margins": 0.019770702347159386,
563
+ "rewards/rejected": -0.017829647287726402,
564
+ "step": 37
565
+ },
566
+ {
567
+ "epoch": 0.3474285714285714,
568
+ "grad_norm": 0.8993732929229736,
569
+ "learning_rate": 1.4578383269606003e-06,
570
+ "logits/chosen": -0.4959084987640381,
571
+ "logits/rejected": -0.4965103268623352,
572
+ "logps/chosen": -22.240234375,
573
+ "logps/rejected": -24.32564926147461,
574
+ "loss": 0.6852,
575
+ "rewards/accuracies": 0.640625,
576
+ "rewards/chosen": 0.0024603637866675854,
577
+ "rewards/margins": 0.016349542886018753,
578
+ "rewards/rejected": -0.01388918049633503,
579
+ "step": 38
580
+ },
581
+ {
582
+ "epoch": 0.3565714285714286,
583
+ "grad_norm": 0.883084774017334,
584
+ "learning_rate": 1.4320279866427795e-06,
585
+ "logits/chosen": -0.4984406530857086,
586
+ "logits/rejected": -0.4946525990962982,
587
+ "logps/chosen": -21.160072326660156,
588
+ "logps/rejected": -27.482582092285156,
589
+ "loss": 0.6862,
590
+ "rewards/accuracies": 0.609375,
591
+ "rewards/chosen": -0.0018302500247955322,
592
+ "rewards/margins": 0.014431210234761238,
593
+ "rewards/rejected": -0.01626146212220192,
594
+ "step": 39
595
+ },
596
+ {
597
+ "epoch": 0.3657142857142857,
598
+ "grad_norm": 0.8378523588180542,
599
+ "learning_rate": 1.4058587834217354e-06,
600
+ "logits/chosen": -0.45701509714126587,
601
+ "logits/rejected": -0.4572596549987793,
602
+ "logps/chosen": -20.714893341064453,
603
+ "logps/rejected": -21.489299774169922,
604
+ "loss": 0.6832,
605
+ "rewards/accuracies": 0.734375,
606
+ "rewards/chosen": 0.0033057844266295433,
607
+ "rewards/margins": 0.020338119938969612,
608
+ "rewards/rejected": -0.017032336443662643,
609
+ "step": 40
610
+ },
611
+ {
612
+ "epoch": 0.37485714285714283,
613
+ "grad_norm": 0.8495325446128845,
614
+ "learning_rate": 1.3793524546768356e-06,
615
+ "logits/chosen": -0.4954044818878174,
616
+ "logits/rejected": -0.49266135692596436,
617
+ "logps/chosen": -20.177230834960938,
618
+ "logps/rejected": -24.510028839111328,
619
+ "loss": 0.6789,
620
+ "rewards/accuracies": 0.734375,
621
+ "rewards/chosen": 0.004343810956925154,
622
+ "rewards/margins": 0.029092557728290558,
623
+ "rewards/rejected": -0.024748750030994415,
624
+ "step": 41
625
+ },
626
+ {
627
+ "epoch": 0.384,
628
+ "grad_norm": 0.8451035022735596,
629
+ "learning_rate": 1.3525310178198707e-06,
630
+ "logits/chosen": -0.4811052680015564,
631
+ "logits/rejected": -0.4776228070259094,
632
+ "logps/chosen": -19.258869171142578,
633
+ "logps/rejected": -24.437456130981445,
634
+ "loss": 0.6792,
635
+ "rewards/accuracies": 0.75,
636
+ "rewards/chosen": -0.00047845463268458843,
637
+ "rewards/margins": 0.02840927056968212,
638
+ "rewards/rejected": -0.028887726366519928,
639
+ "step": 42
640
+ },
641
+ {
642
+ "epoch": 0.3931428571428571,
643
+ "grad_norm": 0.8372145295143127,
644
+ "learning_rate": 1.3254167520063509e-06,
645
+ "logits/chosen": -0.5291227102279663,
646
+ "logits/rejected": -0.5279653668403625,
647
+ "logps/chosen": -20.010604858398438,
648
+ "logps/rejected": -22.272926330566406,
649
+ "loss": 0.6789,
650
+ "rewards/accuracies": 0.734375,
651
+ "rewards/chosen": 0.006788618862628937,
652
+ "rewards/margins": 0.029079778119921684,
653
+ "rewards/rejected": -0.022291159257292747,
654
+ "step": 43
655
+ },
656
+ {
657
+ "epoch": 0.4022857142857143,
658
+ "grad_norm": 0.8531703352928162,
659
+ "learning_rate": 1.2980321796293836e-06,
660
+ "logits/chosen": -0.5021355152130127,
661
+ "logits/rejected": -0.49966001510620117,
662
+ "logps/chosen": -19.32958984375,
663
+ "logps/rejected": -23.641944885253906,
664
+ "loss": 0.6815,
665
+ "rewards/accuracies": 0.71875,
666
+ "rewards/chosen": -0.0008274047868326306,
667
+ "rewards/margins": 0.023971833288669586,
668
+ "rewards/rejected": -0.02479923702776432,
669
+ "step": 44
670
+ },
671
+ {
672
+ "epoch": 0.4114285714285714,
673
+ "grad_norm": 0.8609844446182251,
674
+ "learning_rate": 1.2704000476115078e-06,
675
+ "logits/chosen": -0.453107088804245,
676
+ "logits/rejected": -0.4526231288909912,
677
+ "logps/chosen": -20.13964080810547,
678
+ "logps/rejected": -21.271961212158203,
679
+ "loss": 0.6834,
680
+ "rewards/accuracies": 0.75,
681
+ "rewards/chosen": -0.001453144708648324,
682
+ "rewards/margins": 0.02003173902630806,
683
+ "rewards/rejected": -0.02148488536477089,
684
+ "step": 45
685
+ },
686
+ {
687
+ "epoch": 0.4205714285714286,
688
+ "grad_norm": 0.8536068201065063,
689
+ "learning_rate": 1.2425433085100224e-06,
690
+ "logits/chosen": -0.5088432431221008,
691
+ "logits/rejected": -0.5059933662414551,
692
+ "logps/chosen": -19.692020416259766,
693
+ "logps/rejected": -25.83331298828125,
694
+ "loss": 0.6813,
695
+ "rewards/accuracies": 0.6875,
696
+ "rewards/chosen": 0.00026753084966912866,
697
+ "rewards/margins": 0.024224810302257538,
698
+ "rewards/rejected": -0.023957280442118645,
699
+ "step": 46
700
+ },
701
+ {
702
+ "epoch": 0.4297142857142857,
703
+ "grad_norm": 0.8311668634414673,
704
+ "learning_rate": 1.2144851014515054e-06,
705
+ "logits/chosen": -0.5062947273254395,
706
+ "logits/rejected": -0.5038907527923584,
707
+ "logps/chosen": -17.751794815063477,
708
+ "logps/rejected": -22.226573944091797,
709
+ "loss": 0.6801,
710
+ "rewards/accuracies": 0.8125,
711
+ "rewards/chosen": 0.0016074711456894875,
712
+ "rewards/margins": 0.026652250438928604,
713
+ "rewards/rejected": -0.025044778361916542,
714
+ "step": 47
715
+ },
716
+ {
717
+ "epoch": 0.43885714285714283,
718
+ "grad_norm": 0.8564451932907104,
719
+ "learning_rate": 1.1862487329113604e-06,
720
+ "logits/chosen": -0.49548226594924927,
721
+ "logits/rejected": -0.495021253824234,
722
+ "logps/chosen": -19.72035026550293,
723
+ "logps/rejected": -22.470474243164062,
724
+ "loss": 0.6821,
725
+ "rewards/accuracies": 0.640625,
726
+ "rewards/chosen": -0.003209448419511318,
727
+ "rewards/margins": 0.02287052944302559,
728
+ "rewards/rejected": -0.026079978793859482,
729
+ "step": 48
730
+ },
731
+ {
732
+ "epoch": 0.448,
733
+ "grad_norm": 0.8162851333618164,
734
+ "learning_rate": 1.157857657354354e-06,
735
+ "logits/chosen": -0.485680490732193,
736
+ "logits/rejected": -0.4866037368774414,
737
+ "logps/chosen": -20.71297836303711,
738
+ "logps/rejected": -22.378761291503906,
739
+ "loss": 0.6817,
740
+ "rewards/accuracies": 0.6875,
741
+ "rewards/chosen": 0.0007772197714075446,
742
+ "rewards/margins": 0.023563655093312263,
743
+ "rewards/rejected": -0.02278643473982811,
744
+ "step": 49
745
+ },
746
+ {
747
+ "epoch": 0.45714285714285713,
748
+ "grad_norm": 0.830013632774353,
749
+ "learning_rate": 1.1293354577522264e-06,
750
+ "logits/chosen": -0.4831707775592804,
751
+ "logits/rejected": -0.48350754380226135,
752
+ "logps/chosen": -19.254268646240234,
753
+ "logps/rejected": -22.687885284423828,
754
+ "loss": 0.6771,
755
+ "rewards/accuracies": 0.71875,
756
+ "rewards/chosen": 0.0018323032418265939,
757
+ "rewards/margins": 0.03298073261976242,
758
+ "rewards/rejected": -0.031148429960012436,
759
+ "step": 50
760
+ },
761
+ {
762
+ "epoch": 0.4662857142857143,
763
+ "grad_norm": 0.8517441749572754,
764
+ "learning_rate": 1.1007058259945583e-06,
765
+ "logits/chosen": -0.46854254603385925,
766
+ "logits/rejected": -0.468083918094635,
767
+ "logps/chosen": -22.5173397064209,
768
+ "logps/rejected": -21.76658058166504,
769
+ "loss": 0.6767,
770
+ "rewards/accuracies": 0.765625,
771
+ "rewards/chosen": 0.003332484746351838,
772
+ "rewards/margins": 0.03372838720679283,
773
+ "rewards/rejected": -0.030395902693271637,
774
+ "step": 51
775
+ },
776
+ {
777
+ "epoch": 0.4754285714285714,
778
+ "grad_norm": 0.878716230392456,
779
+ "learning_rate": 1.071992543209167e-06,
780
+ "logits/chosen": -0.5000350475311279,
781
+ "logits/rejected": -0.5004590749740601,
782
+ "logps/chosen": -21.257644653320312,
783
+ "logps/rejected": -23.946401596069336,
784
+ "loss": 0.6762,
785
+ "rewards/accuracies": 0.828125,
786
+ "rewards/chosen": 0.008468384854495525,
787
+ "rewards/margins": 0.03463183715939522,
788
+ "rewards/rejected": -0.026163451373577118,
789
+ "step": 52
790
+ },
791
+ {
792
+ "epoch": 0.4845714285714286,
793
+ "grad_norm": 0.9382399320602417,
794
+ "learning_rate": 1.0432194600083739e-06,
795
+ "logits/chosen": -0.42771774530410767,
796
+ "logits/rejected": -0.4264780580997467,
797
+ "logps/chosen": -21.58111572265625,
798
+ "logps/rejected": -22.952890396118164,
799
+ "loss": 0.6807,
800
+ "rewards/accuracies": 0.71875,
801
+ "rewards/chosen": -0.0012989637907594442,
802
+ "rewards/margins": 0.025485645979642868,
803
+ "rewards/rejected": -0.026784608140587807,
804
+ "step": 53
805
+ },
806
+ {
807
+ "epoch": 0.4937142857142857,
808
+ "grad_norm": 0.8357513546943665,
809
+ "learning_rate": 1.0144104766775572e-06,
810
+ "logits/chosen": -0.45827746391296387,
811
+ "logits/rejected": -0.45641711354255676,
812
+ "logps/chosen": -20.121042251586914,
813
+ "logps/rejected": -24.95348358154297,
814
+ "loss": 0.6774,
815
+ "rewards/accuracies": 0.6875,
816
+ "rewards/chosen": 0.0013005671789869666,
817
+ "rewards/margins": 0.03278574347496033,
818
+ "rewards/rejected": -0.031485170125961304,
819
+ "step": 54
820
+ },
821
+ {
822
+ "epoch": 0.5028571428571429,
823
+ "grad_norm": 0.8457587957382202,
824
+ "learning_rate": 9.85589523322443e-07,
825
+ "logits/chosen": -0.4841751456260681,
826
+ "logits/rejected": -0.48546385765075684,
827
+ "logps/chosen": -19.163833618164062,
828
+ "logps/rejected": -21.794513702392578,
829
+ "loss": 0.6781,
830
+ "rewards/accuracies": 0.828125,
831
+ "rewards/chosen": 0.0022508702240884304,
832
+ "rewards/margins": 0.03070840798318386,
833
+ "rewards/rejected": -0.028457539156079292,
834
+ "step": 55
835
+ },
836
+ {
837
+ "epoch": 0.512,
838
+ "grad_norm": 0.8297368288040161,
839
+ "learning_rate": 9.567805399916258e-07,
840
+ "logits/chosen": -0.5258585214614868,
841
+ "logits/rejected": -0.5240969657897949,
842
+ "logps/chosen": -18.361141204833984,
843
+ "logps/rejected": -20.9268741607666,
844
+ "loss": 0.6736,
845
+ "rewards/accuracies": 0.828125,
846
+ "rewards/chosen": 0.004435597453266382,
847
+ "rewards/margins": 0.0400521382689476,
848
+ "rewards/rejected": -0.03561653941869736,
849
+ "step": 56
850
+ },
851
+ {
852
+ "epoch": 0.5211428571428571,
853
+ "grad_norm": 0.8742468953132629,
854
+ "learning_rate": 9.280074567908329e-07,
855
+ "logits/chosen": -0.46377429366111755,
856
+ "logits/rejected": -0.4626997113227844,
857
+ "logps/chosen": -20.75351333618164,
858
+ "logps/rejected": -22.867094039916992,
859
+ "loss": 0.6735,
860
+ "rewards/accuracies": 0.75,
861
+ "rewards/chosen": -0.0022125868126749992,
862
+ "rewards/margins": 0.040575940161943436,
863
+ "rewards/rejected": -0.04278852790594101,
864
+ "step": 57
865
+ },
866
+ {
867
+ "epoch": 0.5302857142857142,
868
+ "grad_norm": 0.8354327082633972,
869
+ "learning_rate": 8.992941740054417e-07,
870
+ "logits/chosen": -0.5173907279968262,
871
+ "logits/rejected": -0.515368640422821,
872
+ "logps/chosen": -18.376188278198242,
873
+ "logps/rejected": -21.78232765197754,
874
+ "loss": 0.6759,
875
+ "rewards/accuracies": 0.703125,
876
+ "rewards/chosen": 0.002152972621843219,
877
+ "rewards/margins": 0.03564020246267319,
878
+ "rewards/rejected": -0.0334872305393219,
879
+ "step": 58
880
+ },
881
+ {
882
+ "epoch": 0.5394285714285715,
883
+ "grad_norm": 0.8686521649360657,
884
+ "learning_rate": 8.706645422477738e-07,
885
+ "logits/chosen": -0.5015859603881836,
886
+ "logits/rejected": -0.500649094581604,
887
+ "logps/chosen": -20.688770294189453,
888
+ "logps/rejected": -25.79022216796875,
889
+ "loss": 0.6751,
890
+ "rewards/accuracies": 0.65625,
891
+ "rewards/chosen": 0.0003949231468141079,
892
+ "rewards/margins": 0.03735088184475899,
893
+ "rewards/rejected": -0.03695595636963844,
894
+ "step": 59
895
+ },
896
+ {
897
+ "epoch": 0.5485714285714286,
898
+ "grad_norm": 0.8363887071609497,
899
+ "learning_rate": 8.42142342645646e-07,
900
+ "logits/chosen": -0.5239295363426208,
901
+ "logits/rejected": -0.5249078869819641,
902
+ "logps/chosen": -20.808338165283203,
903
+ "logps/rejected": -21.720489501953125,
904
+ "loss": 0.6724,
905
+ "rewards/accuracies": 0.796875,
906
+ "rewards/chosen": 0.007244637701660395,
907
+ "rewards/margins": 0.0426597073674202,
908
+ "rewards/rejected": -0.03541507199406624,
909
+ "step": 60
910
+ },
911
+ {
912
+ "epoch": 0.5577142857142857,
913
+ "grad_norm": 0.8207051157951355,
914
+ "learning_rate": 8.137512670886396e-07,
915
+ "logits/chosen": -0.5192184448242188,
916
+ "logits/rejected": -0.5182772874832153,
917
+ "logps/chosen": -20.181354522705078,
918
+ "logps/rejected": -22.01133918762207,
919
+ "loss": 0.6766,
920
+ "rewards/accuracies": 0.75,
921
+ "rewards/chosen": -0.0022195111960172653,
922
+ "rewards/margins": 0.0343991219997406,
923
+ "rewards/rejected": -0.036618635058403015,
924
+ "step": 61
925
+ },
926
+ {
927
+ "epoch": 0.5668571428571428,
928
+ "grad_norm": 0.8342001438140869,
929
+ "learning_rate": 7.855148985484945e-07,
930
+ "logits/chosen": -0.46402209997177124,
931
+ "logits/rejected": -0.46386054158210754,
932
+ "logps/chosen": -19.38882827758789,
933
+ "logps/rejected": -22.53490447998047,
934
+ "loss": 0.6766,
935
+ "rewards/accuracies": 0.734375,
936
+ "rewards/chosen": 0.0009781585540622473,
937
+ "rewards/margins": 0.034154050052165985,
938
+ "rewards/rejected": -0.03317589312791824,
939
+ "step": 62
940
+ },
941
+ {
942
+ "epoch": 0.576,
943
+ "grad_norm": 0.8299075961112976,
944
+ "learning_rate": 7.574566914899778e-07,
945
+ "logits/chosen": -0.46919354796409607,
946
+ "logits/rejected": -0.46564850211143494,
947
+ "logps/chosen": -19.143640518188477,
948
+ "logps/rejected": -23.518924713134766,
949
+ "loss": 0.6769,
950
+ "rewards/accuracies": 0.703125,
951
+ "rewards/chosen": 0.001144832349382341,
952
+ "rewards/margins": 0.03376341611146927,
953
+ "rewards/rejected": -0.03261858597397804,
954
+ "step": 63
955
+ },
956
+ {
957
+ "epoch": 0.5851428571428572,
958
+ "grad_norm": 0.8152530789375305,
959
+ "learning_rate": 7.29599952388492e-07,
960
+ "logits/chosen": -0.4920732378959656,
961
+ "logits/rejected": -0.49123212695121765,
962
+ "logps/chosen": -19.615325927734375,
963
+ "logps/rejected": -22.90446662902832,
964
+ "loss": 0.6712,
965
+ "rewards/accuracies": 0.78125,
966
+ "rewards/chosen": 0.00682541448622942,
967
+ "rewards/margins": 0.04505688697099686,
968
+ "rewards/rejected": -0.03823147341609001,
969
+ "step": 64
970
+ },
971
+ {
972
+ "epoch": 0.5942857142857143,
973
+ "grad_norm": 0.8843557834625244,
974
+ "learning_rate": 7.019678203706163e-07,
975
+ "logits/chosen": -0.5882308483123779,
976
+ "logits/rejected": -0.5857687592506409,
977
+ "logps/chosen": -20.49724006652832,
978
+ "logps/rejected": -24.06455421447754,
979
+ "loss": 0.6749,
980
+ "rewards/accuracies": 0.671875,
981
+ "rewards/chosen": 0.001912413164973259,
982
+ "rewards/margins": 0.03791937604546547,
983
+ "rewards/rejected": -0.03600696474313736,
984
+ "step": 65
985
+ },
986
+ {
987
+ "epoch": 0.6034285714285714,
988
+ "grad_norm": 0.8193322420120239,
989
+ "learning_rate": 6.74583247993649e-07,
990
+ "logits/chosen": -0.4921879172325134,
991
+ "logits/rejected": -0.4918578863143921,
992
+ "logps/chosen": -19.484710693359375,
993
+ "logps/rejected": -21.404190063476562,
994
+ "loss": 0.6739,
995
+ "rewards/accuracies": 0.78125,
996
+ "rewards/chosen": 0.009157883003354073,
997
+ "rewards/margins": 0.03956674411892891,
998
+ "rewards/rejected": -0.030408862978219986,
999
+ "step": 66
1000
+ },
1001
+ {
1002
+ "epoch": 0.6125714285714285,
1003
+ "grad_norm": 0.8606120347976685,
1004
+ "learning_rate": 6.474689821801295e-07,
1005
+ "logits/chosen": -0.49513500928878784,
1006
+ "logits/rejected": -0.49418747425079346,
1007
+ "logps/chosen": -20.472610473632812,
1008
+ "logps/rejected": -23.929628372192383,
1009
+ "loss": 0.6759,
1010
+ "rewards/accuracies": 0.734375,
1011
+ "rewards/chosen": -0.000603815889917314,
1012
+ "rewards/margins": 0.035757146775722504,
1013
+ "rewards/rejected": -0.0363609679043293,
1014
+ "step": 67
1015
+ },
1016
+ {
1017
+ "epoch": 0.6217142857142857,
1018
+ "grad_norm": 0.8370280861854553,
1019
+ "learning_rate": 6.206475453231643e-07,
1020
+ "logits/chosen": -0.4970676600933075,
1021
+ "logits/rejected": -0.4940704107284546,
1022
+ "logps/chosen": -17.735843658447266,
1023
+ "logps/rejected": -22.89034652709961,
1024
+ "loss": 0.6736,
1025
+ "rewards/accuracies": 0.6875,
1026
+ "rewards/chosen": 0.003980270586907864,
1027
+ "rewards/margins": 0.04059094190597534,
1028
+ "rewards/rejected": -0.036610670387744904,
1029
+ "step": 68
1030
+ },
1031
+ {
1032
+ "epoch": 0.6308571428571429,
1033
+ "grad_norm": 0.8599807620048523,
1034
+ "learning_rate": 5.941412165782645e-07,
1035
+ "logits/chosen": -0.4919041097164154,
1036
+ "logits/rejected": -0.49065884947776794,
1037
+ "logps/chosen": -20.158681869506836,
1038
+ "logps/rejected": -22.979007720947266,
1039
+ "loss": 0.6666,
1040
+ "rewards/accuracies": 0.796875,
1041
+ "rewards/chosen": 0.009388677775859833,
1042
+ "rewards/margins": 0.05475625395774841,
1043
+ "rewards/rejected": -0.04536757990717888,
1044
+ "step": 69
1045
+ },
1046
+ {
1047
+ "epoch": 0.64,
1048
+ "grad_norm": 0.8497822880744934,
1049
+ "learning_rate": 5.679720133572206e-07,
1050
+ "logits/chosen": -0.5309776663780212,
1051
+ "logits/rejected": -0.529637336730957,
1052
+ "logps/chosen": -20.97797393798828,
1053
+ "logps/rejected": -24.416120529174805,
1054
+ "loss": 0.6701,
1055
+ "rewards/accuracies": 0.796875,
1056
+ "rewards/chosen": 0.0016971048898994923,
1057
+ "rewards/margins": 0.047444336116313934,
1058
+ "rewards/rejected": -0.04574722796678543,
1059
+ "step": 70
1060
+ },
1061
+ {
1062
+ "epoch": 0.6491428571428571,
1063
+ "grad_norm": 0.8903601765632629,
1064
+ "learning_rate": 5.421616730394e-07,
1065
+ "logits/chosen": -0.4996778070926666,
1066
+ "logits/rejected": -0.4983339011669159,
1067
+ "logps/chosen": -21.491771697998047,
1068
+ "logps/rejected": -24.228439331054688,
1069
+ "loss": 0.6698,
1070
+ "rewards/accuracies": 0.765625,
1071
+ "rewards/chosen": 0.0040655978955328465,
1072
+ "rewards/margins": 0.048246923834085464,
1073
+ "rewards/rejected": -0.04418133199214935,
1074
+ "step": 71
1075
+ },
1076
+ {
1077
+ "epoch": 0.6582857142857143,
1078
+ "grad_norm": 0.8497235774993896,
1079
+ "learning_rate": 5.167316349156494e-07,
1080
+ "logits/chosen": -0.43921422958374023,
1081
+ "logits/rejected": -0.43771079182624817,
1082
+ "logps/chosen": -21.344398498535156,
1083
+ "logps/rejected": -23.420509338378906,
1084
+ "loss": 0.6795,
1085
+ "rewards/accuracies": 0.625,
1086
+ "rewards/chosen": -0.0017599835991859436,
1087
+ "rewards/margins": 0.028571717441082,
1088
+ "rewards/rejected": -0.030331702902913094,
1089
+ "step": 72
1090
+ },
1091
+ {
1092
+ "epoch": 0.6674285714285715,
1093
+ "grad_norm": 0.8865553736686707,
1094
+ "learning_rate": 4.917030223798057e-07,
1095
+ "logits/chosen": -0.5928893089294434,
1096
+ "logits/rejected": -0.5926108956336975,
1097
+ "logps/chosen": -21.220226287841797,
1098
+ "logps/rejected": -22.727062225341797,
1099
+ "loss": 0.6642,
1100
+ "rewards/accuracies": 0.828125,
1101
+ "rewards/chosen": 0.010776174254715443,
1102
+ "rewards/margins": 0.06004716455936432,
1103
+ "rewards/rejected": -0.04927099123597145,
1104
+ "step": 73
1105
+ },
1106
+ {
1107
+ "epoch": 0.6765714285714286,
1108
+ "grad_norm": 0.8491672873497009,
1109
+ "learning_rate": 4.6709662538260264e-07,
1110
+ "logits/chosen": -0.4956081807613373,
1111
+ "logits/rejected": -0.49190759658813477,
1112
+ "logps/chosen": -20.650142669677734,
1113
+ "logps/rejected": -24.599397659301758,
1114
+ "loss": 0.6709,
1115
+ "rewards/accuracies": 0.796875,
1116
+ "rewards/chosen": -0.0004027389222756028,
1117
+ "rewards/margins": 0.04638548940420151,
1118
+ "rewards/rejected": -0.04678822308778763,
1119
+ "step": 74
1120
+ },
1121
+ {
1122
+ "epoch": 0.6857142857142857,
1123
+ "grad_norm": 0.8251460790634155,
1124
+ "learning_rate": 4.429328831625565e-07,
1125
+ "logits/chosen": -0.5142466425895691,
1126
+ "logits/rejected": -0.5125504732131958,
1127
+ "logps/chosen": -19.210243225097656,
1128
+ "logps/rejected": -22.045814514160156,
1129
+ "loss": 0.6715,
1130
+ "rewards/accuracies": 0.765625,
1131
+ "rewards/chosen": 0.004155635833740234,
1132
+ "rewards/margins": 0.04472273588180542,
1133
+ "rewards/rejected": -0.040567103773355484,
1134
+ "step": 75
1135
+ },
1136
+ {
1137
+ "epoch": 0.6948571428571428,
1138
+ "grad_norm": 0.84358811378479,
1139
+ "learning_rate": 4.192318672681631e-07,
1140
+ "logits/chosen": -0.5178839564323425,
1141
+ "logits/rejected": -0.5178475379943848,
1142
+ "logps/chosen": -21.470251083374023,
1143
+ "logps/rejected": -22.522048950195312,
1144
+ "loss": 0.6779,
1145
+ "rewards/accuracies": 0.703125,
1146
+ "rewards/chosen": 0.0009530810639262199,
1147
+ "rewards/margins": 0.03175386041402817,
1148
+ "rewards/rejected": -0.030800778418779373,
1149
+ "step": 76
1150
+ },
1151
+ {
1152
+ "epoch": 0.704,
1153
+ "grad_norm": 0.8886722326278687,
1154
+ "learning_rate": 3.9601326488552254e-07,
1155
+ "logits/chosen": -0.5036581158638,
1156
+ "logits/rejected": -0.5014926791191101,
1157
+ "logps/chosen": -19.43602752685547,
1158
+ "logps/rejected": -23.601289749145508,
1159
+ "loss": 0.6658,
1160
+ "rewards/accuracies": 0.890625,
1161
+ "rewards/chosen": 0.006434202194213867,
1162
+ "rewards/margins": 0.056305788457393646,
1163
+ "rewards/rejected": -0.04987158998847008,
1164
+ "step": 77
1165
+ },
1166
+ {
1167
+ "epoch": 0.7131428571428572,
1168
+ "grad_norm": 0.8791576027870178,
1169
+ "learning_rate": 3.7329636248522746e-07,
1170
+ "logits/chosen": -0.46405550837516785,
1171
+ "logits/rejected": -0.4603899419307709,
1172
+ "logps/chosen": -18.42851448059082,
1173
+ "logps/rejected": -25.18975830078125,
1174
+ "loss": 0.6672,
1175
+ "rewards/accuracies": 0.8125,
1176
+ "rewards/chosen": -0.0007312432862818241,
1177
+ "rewards/margins": 0.053291577845811844,
1178
+ "rewards/rejected": -0.05402282252907753,
1179
+ "step": 78
1180
+ },
1181
+ {
1182
+ "epoch": 0.7222857142857143,
1183
+ "grad_norm": 0.8433294296264648,
1184
+ "learning_rate": 3.511000298021097e-07,
1185
+ "logits/chosen": -0.49895980954170227,
1186
+ "logits/rejected": -0.49694061279296875,
1187
+ "logps/chosen": -19.6041202545166,
1188
+ "logps/rejected": -25.00043487548828,
1189
+ "loss": 0.6703,
1190
+ "rewards/accuracies": 0.796875,
1191
+ "rewards/chosen": 0.003833733033388853,
1192
+ "rewards/margins": 0.04719376564025879,
1193
+ "rewards/rejected": -0.04336003214120865,
1194
+ "step": 79
1195
+ },
1196
+ {
1197
+ "epoch": 0.7314285714285714,
1198
+ "grad_norm": 0.876516580581665,
1199
+ "learning_rate": 3.294427041611425e-07,
1200
+ "logits/chosen": -0.5160364508628845,
1201
+ "logits/rejected": -0.5145105123519897,
1202
+ "logps/chosen": -19.245559692382812,
1203
+ "logps/rejected": -23.522289276123047,
1204
+ "loss": 0.664,
1205
+ "rewards/accuracies": 0.828125,
1206
+ "rewards/chosen": -0.007931034080684185,
1207
+ "rewards/margins": 0.060558758676052094,
1208
+ "rewards/rejected": -0.06848978996276855,
1209
+ "step": 80
1210
+ },
1211
+ {
1212
+ "epoch": 0.7405714285714285,
1213
+ "grad_norm": 0.8878673315048218,
1214
+ "learning_rate": 3.0834237516252816e-07,
1215
+ "logits/chosen": -0.5207007527351379,
1216
+ "logits/rejected": -0.519152045249939,
1217
+ "logps/chosen": -19.772418975830078,
1218
+ "logps/rejected": -24.96041488647461,
1219
+ "loss": 0.664,
1220
+ "rewards/accuracies": 0.84375,
1221
+ "rewards/chosen": 0.006276377942413092,
1222
+ "rewards/margins": 0.060217950493097305,
1223
+ "rewards/rejected": -0.05394157022237778,
1224
+ "step": 81
1225
+ },
1226
+ {
1227
+ "epoch": 0.7497142857142857,
1228
+ "grad_norm": 0.8819431066513062,
1229
+ "learning_rate": 2.8781656973868117e-07,
1230
+ "logits/chosen": -0.5246998071670532,
1231
+ "logits/rejected": -0.5219220519065857,
1232
+ "logps/chosen": -20.551727294921875,
1233
+ "logps/rejected": -25.322784423828125,
1234
+ "loss": 0.6701,
1235
+ "rewards/accuracies": 0.734375,
1236
+ "rewards/chosen": 0.002442985773086548,
1237
+ "rewards/margins": 0.048708707094192505,
1238
+ "rewards/rejected": -0.04626572132110596,
1239
+ "step": 82
1240
+ },
1241
+ {
1242
+ "epoch": 0.7588571428571429,
1243
+ "grad_norm": 0.885720431804657,
1244
+ "learning_rate": 2.6788233759553134e-07,
1245
+ "logits/chosen": -0.49644091725349426,
1246
+ "logits/rejected": -0.4941262900829315,
1247
+ "logps/chosen": -18.80657958984375,
1248
+ "logps/rejected": -24.829132080078125,
1249
+ "loss": 0.6708,
1250
+ "rewards/accuracies": 0.828125,
1251
+ "rewards/chosen": 0.0027784379199147224,
1252
+ "rewards/margins": 0.04654521495103836,
1253
+ "rewards/rejected": -0.043766774237155914,
1254
+ "step": 83
1255
+ },
1256
+ {
1257
+ "epoch": 0.768,
1258
+ "grad_norm": 0.8168476819992065,
1259
+ "learning_rate": 2.485562370502279e-07,
1260
+ "logits/chosen": -0.4792459011077881,
1261
+ "logits/rejected": -0.4787862300872803,
1262
+ "logps/chosen": -18.839183807373047,
1263
+ "logps/rejected": -23.453889846801758,
1264
+ "loss": 0.6753,
1265
+ "rewards/accuracies": 0.703125,
1266
+ "rewards/chosen": -0.0008009821176528931,
1267
+ "rewards/margins": 0.0374554842710495,
1268
+ "rewards/rejected": -0.03825646638870239,
1269
+ "step": 84
1270
+ },
1271
+ {
1272
+ "epoch": 0.7771428571428571,
1273
+ "grad_norm": 0.8348813056945801,
1274
+ "learning_rate": 2.2985432127701941e-07,
1275
+ "logits/chosen": -0.4961659908294678,
1276
+ "logits/rejected": -0.49521833658218384,
1277
+ "logps/chosen": -19.933313369750977,
1278
+ "logps/rejected": -21.544658660888672,
1279
+ "loss": 0.6708,
1280
+ "rewards/accuracies": 0.71875,
1281
+ "rewards/chosen": 0.0010099338833242655,
1282
+ "rewards/margins": 0.046823300421237946,
1283
+ "rewards/rejected": -0.045813366770744324,
1284
+ "step": 85
1285
+ },
1286
+ {
1287
+ "epoch": 0.7862857142857143,
1288
+ "grad_norm": 0.8756973147392273,
1289
+ "learning_rate": 2.117921249727258e-07,
1290
+ "logits/chosen": -0.4967116415500641,
1291
+ "logits/rejected": -0.49604907631874084,
1292
+ "logps/chosen": -20.46807861328125,
1293
+ "logps/rejected": -22.279346466064453,
1294
+ "loss": 0.6646,
1295
+ "rewards/accuracies": 0.828125,
1296
+ "rewards/chosen": 0.006185826845467091,
1297
+ "rewards/margins": 0.05949004366993904,
1298
+ "rewards/rejected": -0.053304221481084824,
1299
+ "step": 86
1300
+ },
1301
+ {
1302
+ "epoch": 0.7954285714285714,
1303
+ "grad_norm": 0.8629654049873352,
1304
+ "learning_rate": 1.9438465145288375e-07,
1305
+ "logits/chosen": -0.4333621561527252,
1306
+ "logits/rejected": -0.43237873911857605,
1307
+ "logps/chosen": -20.248123168945312,
1308
+ "logps/rejected": -21.335224151611328,
1309
+ "loss": 0.6644,
1310
+ "rewards/accuracies": 0.765625,
1311
+ "rewards/chosen": 0.0038934394251555204,
1312
+ "rewards/margins": 0.059577763080596924,
1313
+ "rewards/rejected": -0.05568432807922363,
1314
+ "step": 87
1315
+ },
1316
+ {
1317
+ "epoch": 0.8045714285714286,
1318
+ "grad_norm": 0.881422221660614,
1319
+ "learning_rate": 1.7764636018928247e-07,
1320
+ "logits/chosen": -0.5081770420074463,
1321
+ "logits/rejected": -0.5073721408843994,
1322
+ "logps/chosen": -20.489046096801758,
1323
+ "logps/rejected": -23.48529815673828,
1324
+ "loss": 0.6605,
1325
+ "rewards/accuracies": 0.8125,
1326
+ "rewards/chosen": 0.016606241464614868,
1327
+ "rewards/margins": 0.06781661510467529,
1328
+ "rewards/rejected": -0.05121037736535072,
1329
+ "step": 88
1330
+ },
1331
+ {
1332
+ "epoch": 0.8137142857142857,
1333
+ "grad_norm": 0.8496101498603821,
1334
+ "learning_rate": 1.6159115479924256e-07,
1335
+ "logits/chosen": -0.5462079048156738,
1336
+ "logits/rejected": -0.5453540086746216,
1337
+ "logps/chosen": -20.12715721130371,
1338
+ "logps/rejected": -24.592514038085938,
1339
+ "loss": 0.6703,
1340
+ "rewards/accuracies": 0.78125,
1341
+ "rewards/chosen": -0.0007760243024677038,
1342
+ "rewards/margins": 0.04710102081298828,
1343
+ "rewards/rejected": -0.04787704348564148,
1344
+ "step": 89
1345
+ },
1346
+ {
1347
+ "epoch": 0.8228571428571428,
1348
+ "grad_norm": 0.8117421269416809,
1349
+ "learning_rate": 1.4623237149661137e-07,
1350
+ "logits/chosen": -0.4740930199623108,
1351
+ "logits/rejected": -0.4749094247817993,
1352
+ "logps/chosen": -20.31298828125,
1353
+ "logps/rejected": -21.797792434692383,
1354
+ "loss": 0.6722,
1355
+ "rewards/accuracies": 0.765625,
1356
+ "rewards/chosen": 0.011264022439718246,
1357
+ "rewards/margins": 0.043472517281770706,
1358
+ "rewards/rejected": -0.03220849484205246,
1359
+ "step": 90
1360
+ },
1361
+ {
1362
+ "epoch": 0.832,
1363
+ "grad_norm": 0.8725610971450806,
1364
+ "learning_rate": 1.315827680140743e-07,
1365
+ "logits/chosen": -0.44238367676734924,
1366
+ "logits/rejected": -0.4397670030593872,
1367
+ "logps/chosen": -21.677785873413086,
1368
+ "logps/rejected": -26.820186614990234,
1369
+ "loss": 0.6724,
1370
+ "rewards/accuracies": 0.71875,
1371
+ "rewards/chosen": -0.012972681783139706,
1372
+ "rewards/margins": 0.04338699206709862,
1373
+ "rewards/rejected": -0.0563596747815609,
1374
+ "step": 91
1375
+ },
1376
+ {
1377
+ "epoch": 0.8411428571428572,
1378
+ "grad_norm": 0.9056218266487122,
1379
+ "learning_rate": 1.1765451300597574e-07,
1380
+ "logits/chosen": -0.5040266513824463,
1381
+ "logits/rejected": -0.5000580549240112,
1382
+ "logps/chosen": -19.250755310058594,
1383
+ "logps/rejected": -25.95856475830078,
1384
+ "loss": 0.6568,
1385
+ "rewards/accuracies": 0.875,
1386
+ "rewards/chosen": 0.01886388659477234,
1387
+ "rewards/margins": 0.07544377446174622,
1388
+ "rewards/rejected": -0.056579895317554474,
1389
+ "step": 92
1390
+ },
1391
+ {
1392
+ "epoch": 0.8502857142857143,
1393
+ "grad_norm": 0.8533287644386292,
1394
+ "learning_rate": 1.0445917594046072e-07,
1395
+ "logits/chosen": -0.4758893847465515,
1396
+ "logits/rejected": -0.4730798602104187,
1397
+ "logps/chosen": -19.33151626586914,
1398
+ "logps/rejected": -23.52974510192871,
1399
+ "loss": 0.6663,
1400
+ "rewards/accuracies": 0.78125,
1401
+ "rewards/chosen": -0.0019086365355178714,
1402
+ "rewards/margins": 0.05595230683684349,
1403
+ "rewards/rejected": -0.05786094069480896,
1404
+ "step": 93
1405
+ },
1406
+ {
1407
+ "epoch": 0.8594285714285714,
1408
+ "grad_norm": 0.8568288087844849,
1409
+ "learning_rate": 9.200771748932512e-08,
1410
+ "logits/chosen": -0.5422481894493103,
1411
+ "logits/rejected": -0.5400527715682983,
1412
+ "logps/chosen": -18.446407318115234,
1413
+ "logps/rejected": -23.11469841003418,
1414
+ "loss": 0.6718,
1415
+ "rewards/accuracies": 0.8125,
1416
+ "rewards/chosen": -0.0013355785049498081,
1417
+ "rewards/margins": 0.044257037341594696,
1418
+ "rewards/rejected": -0.04559261351823807,
1419
+ "step": 94
1420
+ },
1421
+ {
1422
+ "epoch": 0.8685714285714285,
1423
+ "grad_norm": 0.8776550889015198,
1424
+ "learning_rate": 8.031048042356392e-08,
1425
+ "logits/chosen": -0.5440014600753784,
1426
+ "logits/rejected": -0.5424464344978333,
1427
+ "logps/chosen": -20.24156951904297,
1428
+ "logps/rejected": -23.428844451904297,
1429
+ "loss": 0.6697,
1430
+ "rewards/accuracies": 0.8125,
1431
+ "rewards/chosen": -0.0018163705244660378,
1432
+ "rewards/margins": 0.04859299957752228,
1433
+ "rewards/rejected": -0.05040936544537544,
1434
+ "step": 95
1435
+ },
1436
+ {
1437
+ "epoch": 0.8777142857142857,
1438
+ "grad_norm": 0.8318793177604675,
1439
+ "learning_rate": 6.93771810221746e-08,
1440
+ "logits/chosen": -0.49489662051200867,
1441
+ "logits/rejected": -0.49194085597991943,
1442
+ "logps/chosen": -18.67500114440918,
1443
+ "logps/rejected": -21.768386840820312,
1444
+ "loss": 0.6672,
1445
+ "rewards/accuracies": 0.84375,
1446
+ "rewards/chosen": 0.007825060747563839,
1447
+ "rewards/margins": 0.05333828553557396,
1448
+ "rewards/rejected": -0.04551322013139725,
1449
+ "step": 96
1450
+ },
1451
+ {
1452
+ "epoch": 0.8868571428571429,
1453
+ "grad_norm": 0.8406640887260437,
1454
+ "learning_rate": 5.9216901001357124e-08,
1455
+ "logits/chosen": -0.5510961413383484,
1456
+ "logits/rejected": -0.5509752035140991,
1457
+ "logps/chosen": -21.143768310546875,
1458
+ "logps/rejected": -23.429378509521484,
1459
+ "loss": 0.6699,
1460
+ "rewards/accuracies": 0.765625,
1461
+ "rewards/chosen": 0.004591323435306549,
1462
+ "rewards/margins": 0.047924332320690155,
1463
+ "rewards/rejected": -0.043333012610673904,
1464
+ "step": 97
1465
+ },
1466
+ {
1467
+ "epoch": 0.896,
1468
+ "grad_norm": 0.8756365180015564,
1469
+ "learning_rate": 4.9838079970809245e-08,
1470
+ "logits/chosen": -0.5205091834068298,
1471
+ "logits/rejected": -0.5211708545684814,
1472
+ "logps/chosen": -22.369707107543945,
1473
+ "logps/rejected": -23.702943801879883,
1474
+ "loss": 0.6775,
1475
+ "rewards/accuracies": 0.671875,
1476
+ "rewards/chosen": -0.0037931082770228386,
1477
+ "rewards/margins": 0.03301382064819336,
1478
+ "rewards/rejected": -0.03680692985653877,
1479
+ "step": 98
1480
+ },
1481
+ {
1482
+ "epoch": 0.9051428571428571,
1483
+ "grad_norm": 0.8306006193161011,
1484
+ "learning_rate": 4.124850842338779e-08,
1485
+ "logits/chosen": -0.5665202140808105,
1486
+ "logits/rejected": -0.5653331279754639,
1487
+ "logps/chosen": -18.529712677001953,
1488
+ "logps/rejected": -22.967239379882812,
1489
+ "loss": 0.6716,
1490
+ "rewards/accuracies": 0.6875,
1491
+ "rewards/chosen": -0.0003311295295134187,
1492
+ "rewards/margins": 0.04485708475112915,
1493
+ "rewards/rejected": -0.045188214629888535,
1494
+ "step": 99
1495
+ },
1496
+ {
1497
+ "epoch": 0.9142857142857143,
1498
+ "grad_norm": 0.8332085013389587,
1499
+ "learning_rate": 3.345532126395578e-08,
1500
+ "logits/chosen": -0.45710164308547974,
1501
+ "logits/rejected": -0.45356103777885437,
1502
+ "logps/chosen": -19.887006759643555,
1503
+ "logps/rejected": -24.004615783691406,
1504
+ "loss": 0.6686,
1505
+ "rewards/accuracies": 0.765625,
1506
+ "rewards/chosen": 0.002424683189019561,
1507
+ "rewards/margins": 0.05114402621984482,
1508
+ "rewards/rejected": -0.04871934652328491,
1509
+ "step": 100
1510
+ },
1511
+ {
1512
+ "epoch": 0.9234285714285714,
1513
+ "grad_norm": 0.8623512387275696,
1514
+ "learning_rate": 2.6464991882793275e-08,
1515
+ "logits/chosen": -0.413399338722229,
1516
+ "logits/rejected": -0.41507232189178467,
1517
+ "logps/chosen": -20.780710220336914,
1518
+ "logps/rejected": -21.81368637084961,
1519
+ "loss": 0.6659,
1520
+ "rewards/accuracies": 0.8125,
1521
+ "rewards/chosen": 0.0015395893715322018,
1522
+ "rewards/margins": 0.05646102502942085,
1523
+ "rewards/rejected": -0.054921433329582214,
1524
+ "step": 101
1525
+ },
1526
+ {
1527
+ "epoch": 0.9325714285714286,
1528
+ "grad_norm": 0.86390620470047,
1529
+ "learning_rate": 2.028332677849254e-08,
1530
+ "logits/chosen": -0.530703067779541,
1531
+ "logits/rejected": -0.5275431275367737,
1532
+ "logps/chosen": -20.840030670166016,
1533
+ "logps/rejected": -25.737503051757812,
1534
+ "loss": 0.6669,
1535
+ "rewards/accuracies": 0.765625,
1536
+ "rewards/chosen": -0.002680886071175337,
1537
+ "rewards/margins": 0.05446278676390648,
1538
+ "rewards/rejected": -0.0571436733007431,
1539
+ "step": 102
1540
+ },
1541
+ {
1542
+ "epoch": 0.9417142857142857,
1543
+ "grad_norm": 0.8835663199424744,
1544
+ "learning_rate": 1.4915460734805098e-08,
1545
+ "logits/chosen": -0.508980929851532,
1546
+ "logits/rejected": -0.5062388777732849,
1547
+ "logps/chosen": -21.180665969848633,
1548
+ "logps/rejected": -24.446884155273438,
1549
+ "loss": 0.6654,
1550
+ "rewards/accuracies": 0.828125,
1551
+ "rewards/chosen": 0.0001750650117173791,
1552
+ "rewards/margins": 0.05733681842684746,
1553
+ "rewards/rejected": -0.05716175213456154,
1554
+ "step": 103
1555
+ },
1556
+ {
1557
+ "epoch": 0.9508571428571428,
1558
+ "grad_norm": 0.8763132095336914,
1559
+ "learning_rate": 1.0365852555447641e-08,
1560
+ "logits/chosen": -0.45249295234680176,
1561
+ "logits/rejected": -0.45072636008262634,
1562
+ "logps/chosen": -19.483192443847656,
1563
+ "logps/rejected": -22.927146911621094,
1564
+ "loss": 0.6698,
1565
+ "rewards/accuracies": 0.765625,
1566
+ "rewards/chosen": 0.005263314116746187,
1567
+ "rewards/margins": 0.04841461777687073,
1568
+ "rewards/rejected": -0.04315130412578583,
1569
+ "step": 104
1570
+ },
1571
+ {
1572
+ "epoch": 0.96,
1573
+ "grad_norm": 0.8393348455429077,
1574
+ "learning_rate": 6.6382813604083375e-09,
1575
+ "logits/chosen": -0.49951499700546265,
1576
+ "logits/rejected": -0.49781399965286255,
1577
+ "logps/chosen": -18.490966796875,
1578
+ "logps/rejected": -23.551212310791016,
1579
+ "loss": 0.6712,
1580
+ "rewards/accuracies": 0.8125,
1581
+ "rewards/chosen": 0.0046125901862978935,
1582
+ "rewards/margins": 0.04545419290661812,
1583
+ "rewards/rejected": -0.04084160551428795,
1584
+ "step": 105
1585
+ },
1586
+ {
1587
+ "epoch": 0.9691428571428572,
1588
+ "grad_norm": 0.8526674509048462,
1589
+ "learning_rate": 3.735843446830866e-09,
1590
+ "logits/chosen": -0.5123409032821655,
1591
+ "logits/rejected": -0.5107146501541138,
1592
+ "logps/chosen": -18.7539119720459,
1593
+ "logps/rejected": -23.78987693786621,
1594
+ "loss": 0.6641,
1595
+ "rewards/accuracies": 0.84375,
1596
+ "rewards/chosen": 0.00847574882209301,
1597
+ "rewards/margins": 0.06031036376953125,
1598
+ "rewards/rejected": -0.05183461308479309,
1599
+ "step": 106
1600
+ },
1601
+ {
1602
+ "epoch": 0.9782857142857143,
1603
+ "grad_norm": 0.9083833694458008,
1604
+ "learning_rate": 1.6609497170834153e-09,
1605
+ "logits/chosen": -0.47583484649658203,
1606
+ "logits/rejected": -0.4731263518333435,
1607
+ "logps/chosen": -19.73035430908203,
1608
+ "logps/rejected": -23.654985427856445,
1609
+ "loss": 0.664,
1610
+ "rewards/accuracies": 0.75,
1611
+ "rewards/chosen": -0.0008802660740911961,
1612
+ "rewards/margins": 0.06081285700201988,
1613
+ "rewards/rejected": -0.06169312447309494,
1614
+ "step": 107
1615
+ },
1616
+ {
1617
+ "epoch": 0.9874285714285714,
1618
+ "grad_norm": 0.8618621826171875,
1619
+ "learning_rate": 4.153236761488266e-10,
1620
+ "logits/chosen": -0.4947027266025543,
1621
+ "logits/rejected": -0.4926915168762207,
1622
+ "logps/chosen": -18.10439109802246,
1623
+ "logps/rejected": -22.617685317993164,
1624
+ "loss": 0.6604,
1625
+ "rewards/accuracies": 0.796875,
1626
+ "rewards/chosen": 0.008419442921876907,
1627
+ "rewards/margins": 0.06772737205028534,
1628
+ "rewards/rejected": -0.059307925403118134,
1629
+ "step": 108
1630
+ },
1631
+ {
1632
+ "epoch": 0.9965714285714286,
1633
+ "grad_norm": 0.8564931154251099,
1634
+ "learning_rate": 0.0,
1635
+ "logits/chosen": -0.48442748188972473,
1636
+ "logits/rejected": -0.4826381206512451,
1637
+ "logps/chosen": -20.38802719116211,
1638
+ "logps/rejected": -24.91822052001953,
1639
+ "loss": 0.6703,
1640
+ "rewards/accuracies": 0.765625,
1641
+ "rewards/chosen": -0.0030104685574769974,
1642
+ "rewards/margins": 0.04760502278804779,
1643
+ "rewards/rejected": -0.05061549320816994,
1644
+ "step": 109
1645
+ },
1646
+ {
1647
+ "epoch": 0.9965714285714286,
1648
+ "step": 109,
1649
+ "total_flos": 1.5002634544501228e+18,
1650
+ "train_loss": 0.6781991911590646,
1651
+ "train_runtime": 6621.3264,
1652
+ "train_samples_per_second": 1.057,
1653
+ "train_steps_per_second": 0.016
1654
+ }
1655
+ ],
1656
+ "logging_steps": 1.0,
1657
+ "max_steps": 109,
1658
+ "num_input_tokens_seen": 0,
1659
+ "num_train_epochs": 1,
1660
+ "save_steps": 500,
1661
+ "stateful_callbacks": {
1662
+ "TrainerControl": {
1663
+ "args": {
1664
+ "should_epoch_stop": false,
1665
+ "should_evaluate": false,
1666
+ "should_log": false,
1667
+ "should_save": true,
1668
+ "should_training_stop": true
1669
+ },
1670
+ "attributes": {}
1671
+ }
1672
+ },
1673
+ "total_flos": 1.5002634544501228e+18,
1674
+ "train_batch_size": 2,
1675
+ "trial_name": null,
1676
+ "trial_params": null
1677
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:d6f06b86b7bae51eae676662811e31fae2cef0bb0ae2e6fd39a00f18e49f4e75
3
+ size 7224