Hugging Face's logo

EllieS
/
zephyr-7b-dpo-lora-timedial

PEFT
TensorBoard
Safetensors
mistral
alignment-handbook
Generated from Trainer
trl
dpo
Model card Files
xet
 Metrics Community
EllieS commited on
Commit
183c12e
·
verified ·
1 Parent(s): e8a09bb

Model save

Browse files
Files changed (13) hide show
  1. README.md +70 -0
  2. adapter_config.json +31 -0
  3. adapter_model.safetensors +3 -0
  4. all_results.json +21 -0
  5. eval_results.json +16 -0
  6. runs/Mar29_03-45-16_586cb8b6da8c/events.out.tfevents.1711684032.586cb8b6da8c.32938.0 +3 -0
  7. runs/Mar29_03-45-16_586cb8b6da8c/events.out.tfevents.1711684553.586cb8b6da8c.32938.1 +3 -0
  8. special_tokens_map.json +30 -0
  9. tokenizer.json +0 -0
  10. tokenizer_config.json +47 -0
  11. train_results.json +8 -0
  12. trainer_state.json +842 -0
  13. training_args.bin +3 -0
README.md ADDED
@@ -0,0 +1,70 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: apache-2.0
3
+ library_name: peft
4
+ tags:
5
+ - trl
6
+ - dpo
7
+ - generated_from_trainer
8
+ base_model: alignment-handbook/zephyr-7b-sft-full
9
+ model-index:
10
+ - name: zephyr-7b-dpo-lora-timedial
11
+ results: []
12
+ ---
13
+
14
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
15
+ should probably proofread and complete it, then remove this comment. -->
16
+
17
+ # zephyr-7b-dpo-lora-timedial
18
+
19
+ This model is a fine-tuned version of [alignment-handbook/zephyr-7b-sft-full](https://huggingface.co/alignment-handbook/zephyr-7b-sft-full) on the None dataset.
20
+ It achieves the following results on the evaluation set:
21
+ - Loss: 0.0056
22
+ - Rewards/chosen: -0.5375
23
+ - Rewards/rejected: -5.7320
24
+ - Rewards/accuracies: 1.0
25
+ - Rewards/margins: 5.1945
26
+ - Logps/rejected: -611.0701
27
+ - Logps/chosen: -80.2486
28
+ - Logits/rejected: -2.2508
29
+ - Logits/chosen: -2.4113
30
+
31
+ ## Model description
32
+
33
+ More information needed
34
+
35
+ ## Intended uses & limitations
36
+
37
+ More information needed
38
+
39
+ ## Training and evaluation data
40
+
41
+ More information needed
42
+
43
+ ## Training procedure
44
+
45
+ ### Training hyperparameters
46
+
47
+ The following hyperparameters were used during training:
48
+ - learning_rate: 5e-06
49
+ - train_batch_size: 1
50
+ - eval_batch_size: 1
51
+ - seed: 42
52
+ - distributed_type: multi-GPU
53
+ - gradient_accumulation_steps: 2
54
+ - total_train_batch_size: 2
55
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
56
+ - lr_scheduler_type: cosine
57
+ - lr_scheduler_warmup_ratio: 0.1
58
+ - num_epochs: 1
59
+
60
+ ### Training results
61
+
62
+
63
+
64
+ ### Framework versions
65
+
66
+ - PEFT 0.7.1
67
+ - Transformers 4.36.2
68
+ - Pytorch 2.1.2+cu121
69
+ - Datasets 2.14.6
70
+ - Tokenizers 0.15.2
adapter_config.json ADDED
@@ -0,0 +1,31 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "alpha_pattern": {},
3
+ "auto_mapping": null,
4
+ "base_model_name_or_path": "alignment-handbook/zephyr-7b-sft-full",
5
+ "bias": "none",
6
+ "fan_in_fan_out": false,
7
+ "inference_mode": true,
8
+ "init_lora_weights": true,
9
+ "layers_pattern": null,
10
+ "layers_to_transform": null,
11
+ "loftq_config": {},
12
+ "lora_alpha": 16,
13
+ "lora_dropout": 0.05,
14
+ "megatron_config": null,
15
+ "megatron_core": "megatron.core",
16
+ "modules_to_save": null,
17
+ "peft_type": "LORA",
18
+ "r": 16,
19
+ "rank_pattern": {},
20
+ "revision": null,
21
+ "target_modules": [
22
+ "q_proj",
23
+ "o_proj",
24
+ "gate_proj",
25
+ "up_proj",
26
+ "down_proj",
27
+ "k_proj",
28
+ "v_proj"
29
+ ],
30
+ "task_type": "CAUSAL_LM"
31
+ }
adapter_model.safetensors ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:5cf5a8a88f369fef8232d894d2256a2d28c8157b19cd745b3f8b5b3db7251524
3
+ size 83946192
all_results.json ADDED
@@ -0,0 +1,21 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 1.0,
3
+ "eval_logits/chosen": -2.411261796951294,
4
+ "eval_logits/rejected": -2.2507548332214355,
5
+ "eval_logps/chosen": -80.2485580444336,
6
+ "eval_logps/rejected": -611.070068359375,
7
+ "eval_loss": 0.005633037071675062,
8
+ "eval_rewards/accuracies": 1.0,
9
+ "eval_rewards/chosen": -0.5375131964683533,
10
+ "eval_rewards/margins": 5.194511413574219,
11
+ "eval_rewards/rejected": -5.732025146484375,
12
+ "eval_runtime": 53.953,
13
+ "eval_samples": 289,
14
+ "eval_samples_per_second": 5.357,
15
+ "eval_steps_per_second": 5.357,
16
+ "train_loss": 0.14528509169481824,
17
+ "train_runtime": 467.3095,
18
+ "train_samples": 1157,
19
+ "train_samples_per_second": 2.476,
20
+ "train_steps_per_second": 1.237
21
+ }
eval_results.json ADDED
@@ -0,0 +1,16 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 1.0,
3
+ "eval_logits/chosen": -2.411261796951294,
4
+ "eval_logits/rejected": -2.2507548332214355,
5
+ "eval_logps/chosen": -80.2485580444336,
6
+ "eval_logps/rejected": -611.070068359375,
7
+ "eval_loss": 0.005633037071675062,
8
+ "eval_rewards/accuracies": 1.0,
9
+ "eval_rewards/chosen": -0.5375131964683533,
10
+ "eval_rewards/margins": 5.194511413574219,
11
+ "eval_rewards/rejected": -5.732025146484375,
12
+ "eval_runtime": 53.953,
13
+ "eval_samples": 289,
14
+ "eval_samples_per_second": 5.357,
15
+ "eval_steps_per_second": 5.357
16
+ }
runs/Mar29_03-45-16_586cb8b6da8c/events.out.tfevents.1711684032.586cb8b6da8c.32938.0 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:8bc403f9c6cb3902407724247dd223e6ea191587556050dd3f96b3baa7783859
3
+ size 41567
runs/Mar29_03-45-16_586cb8b6da8c/events.out.tfevents.1711684553.586cb8b6da8c.32938.1 ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:b061cd613b146e0563bcd883ee6dba7891fa98b34e2c84246aae84b56b05c2c0
3
+ size 828
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": "</s>",
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_config.json ADDED
@@ -0,0 +1,47 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "add_bos_token": true,
3
+ "add_eos_token": false,
4
+ "added_tokens_decoder": {
5
+ "0": {
6
+ "content": "<unk>",
7
+ "lstrip": false,
8
+ "normalized": false,
9
+ "rstrip": false,
10
+ "single_word": false,
11
+ "special": true
12
+ },
13
+ "1": {
14
+ "content": "<s>",
15
+ "lstrip": false,
16
+ "normalized": false,
17
+ "rstrip": false,
18
+ "single_word": false,
19
+ "special": true
20
+ },
21
+ "2": {
22
+ "content": "</s>",
23
+ "lstrip": false,
24
+ "normalized": false,
25
+ "rstrip": false,
26
+ "single_word": false,
27
+ "special": true
28
+ }
29
+ },
30
+ "additional_special_tokens": [],
31
+ "bos_token": "<s>",
32
+ "chat_template": "{% for message in messages %}\n{% if message['role'] == 'user' %}\n{{ '<|user|>\n' + message['content'] + eos_token }}\n{% elif message['role'] == 'system' %}\n{{ '<|system|>\n' + message['content'] + eos_token }}\n{% elif message['role'] == 'assistant' %}\n{{ '<|assistant|>\n' + message['content'] + eos_token }}\n{% endif %}\n{% if loop.last and add_generation_prompt %}\n{{ '<|assistant|>' }}\n{% endif %}\n{% endfor %}",
33
+ "clean_up_tokenization_spaces": false,
34
+ "eos_token": "</s>",
35
+ "legacy": true,
36
+ "max_length": 2048,
37
+ "model_max_length": 2048,
38
+ "pad_token": "</s>",
39
+ "sp_model_kwargs": {},
40
+ "spaces_between_special_tokens": false,
41
+ "stride": 0,
42
+ "tokenizer_class": "LlamaTokenizer",
43
+ "truncation_side": "left",
44
+ "truncation_strategy": "longest_first",
45
+ "unk_token": "<unk>",
46
+ "use_default_system_prompt": false
47
+ }
train_results.json ADDED
@@ -0,0 +1,8 @@
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 1.0,
3
+ "train_loss": 0.14528509169481824,
4
+ "train_runtime": 467.3095,
5
+ "train_samples": 1157,
6
+ "train_samples_per_second": 2.476,
7
+ "train_steps_per_second": 1.237
8
+ }
trainer_state.json ADDED
@@ -0,0 +1,842 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 0.9991356957649092,
5
+ "eval_steps": 3000,
6
+ "global_step": 578,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.0,
13
+ "learning_rate": 8.620689655172414e-08,
14
+ "logits/chosen": -3.0497775077819824,
15
+ "logits/rejected": -3.057526111602783,
16
+ "logps/chosen": -25.830108642578125,
17
+ "logps/rejected": -43.69984436035156,
18
+ "loss": 0.6931,
19
+ "rewards/accuracies": 0.0,
20
+ "rewards/chosen": 0.0,
21
+ "rewards/margins": 0.0,
22
+ "rewards/rejected": 0.0,
23
+ "step": 1
24
+ },
25
+ {
26
+ "epoch": 0.02,
27
+ "learning_rate": 8.620689655172415e-07,
28
+ "logits/chosen": -2.9667391777038574,
29
+ "logits/rejected": -2.9639124870300293,
30
+ "logps/chosen": -26.11766815185547,
31
+ "logps/rejected": -37.49580383300781,
32
+ "loss": 0.6935,
33
+ "rewards/accuracies": 0.3333333432674408,
34
+ "rewards/chosen": -0.00040692646871320903,
35
+ "rewards/margins": -0.0007521946681663394,
36
+ "rewards/rejected": 0.0003452682576607913,
37
+ "step": 10
38
+ },
39
+ {
40
+ "epoch": 0.03,
41
+ "learning_rate": 1.724137931034483e-06,
42
+ "logits/chosen": -2.960737705230713,
43
+ "logits/rejected": -2.9585623741149902,
44
+ "logps/chosen": -26.109699249267578,
45
+ "logps/rejected": -38.54250717163086,
46
+ "loss": 0.6928,
47
+ "rewards/accuracies": 0.5,
48
+ "rewards/chosen": 0.0008904800051823258,
49
+ "rewards/margins": 0.0006467189523391426,
50
+ "rewards/rejected": 0.0002437610673950985,
51
+ "step": 20
52
+ },
53
+ {
54
+ "epoch": 0.05,
55
+ "learning_rate": 2.5862068965517246e-06,
56
+ "logits/chosen": -3.008089542388916,
57
+ "logits/rejected": -3.0093681812286377,
58
+ "logps/chosen": -25.44864273071289,
59
+ "logps/rejected": -37.36754608154297,
60
+ "loss": 0.6914,
61
+ "rewards/accuracies": 0.8999999761581421,
62
+ "rewards/chosen": 0.0050292485393583775,
63
+ "rewards/margins": 0.003431679680943489,
64
+ "rewards/rejected": 0.001597568392753601,
65
+ "step": 30
66
+ },
67
+ {
68
+ "epoch": 0.07,
69
+ "learning_rate": 3.448275862068966e-06,
70
+ "logits/chosen": -3.019864797592163,
71
+ "logits/rejected": -3.0186901092529297,
72
+ "logps/chosen": -24.686126708984375,
73
+ "logps/rejected": -36.383323669433594,
74
+ "loss": 0.6878,
75
+ "rewards/accuracies": 1.0,
76
+ "rewards/chosen": 0.016769181936979294,
77
+ "rewards/margins": 0.010735146701335907,
78
+ "rewards/rejected": 0.006034034304320812,
79
+ "step": 40
80
+ },
81
+ {
82
+ "epoch": 0.09,
83
+ "learning_rate": 4.310344827586207e-06,
84
+ "logits/chosen": -2.9620559215545654,
85
+ "logits/rejected": -2.960641622543335,
86
+ "logps/chosen": -22.043102264404297,
87
+ "logps/rejected": -37.594871520996094,
88
+ "loss": 0.6765,
89
+ "rewards/accuracies": 1.0,
90
+ "rewards/chosen": 0.04144766926765442,
91
+ "rewards/margins": 0.03353805094957352,
92
+ "rewards/rejected": 0.0079096220433712,
93
+ "step": 50
94
+ },
95
+ {
96
+ "epoch": 0.1,
97
+ "learning_rate": 4.999817502139027e-06,
98
+ "logits/chosen": -2.9981789588928223,
99
+ "logits/rejected": -2.9974167346954346,
100
+ "logps/chosen": -17.9765625,
101
+ "logps/rejected": -38.53766632080078,
102
+ "loss": 0.659,
103
+ "rewards/accuracies": 1.0,
104
+ "rewards/chosen": 0.08436249941587448,
105
+ "rewards/margins": 0.06955190747976303,
106
+ "rewards/rejected": 0.014810582622885704,
107
+ "step": 60
108
+ },
109
+ {
110
+ "epoch": 0.12,
111
+ "learning_rate": 4.9934328742287285e-06,
112
+ "logits/chosen": -2.984670400619507,
113
+ "logits/rejected": -2.9880006313323975,
114
+ "logps/chosen": -13.898828506469727,
115
+ "logps/rejected": -39.7334098815918,
116
+ "loss": 0.6268,
117
+ "rewards/accuracies": 1.0,
118
+ "rewards/chosen": 0.12379348278045654,
119
+ "rewards/margins": 0.13776807487010956,
120
+ "rewards/rejected": -0.01397457905113697,
121
+ "step": 70
122
+ },
123
+ {
124
+ "epoch": 0.14,
125
+ "learning_rate": 4.977949980164773e-06,
126
+ "logits/chosen": -2.9665162563323975,
127
+ "logits/rejected": -2.97159743309021,
128
+ "logps/chosen": -11.102731704711914,
129
+ "logps/rejected": -47.608909606933594,
130
+ "loss": 0.5781,
131
+ "rewards/accuracies": 1.0,
132
+ "rewards/chosen": 0.15097340941429138,
133
+ "rewards/margins": 0.2455867975950241,
134
+ "rewards/rejected": -0.09461338818073273,
135
+ "step": 80
136
+ },
137
+ {
138
+ "epoch": 0.16,
139
+ "learning_rate": 4.953425315348534e-06,
140
+ "logits/chosen": -2.940497875213623,
141
+ "logits/rejected": -2.944275379180908,
142
+ "logps/chosen": -5.016680717468262,
143
+ "logps/rejected": -57.90388107299805,
144
+ "loss": 0.5062,
145
+ "rewards/accuracies": 1.0,
146
+ "rewards/chosen": 0.2140476256608963,
147
+ "rewards/margins": 0.41967692971229553,
148
+ "rewards/rejected": -0.20562927424907684,
149
+ "step": 90
150
+ },
151
+ {
152
+ "epoch": 0.17,
153
+ "learning_rate": 4.919948367622307e-06,
154
+ "logits/chosen": -2.9766781330108643,
155
+ "logits/rejected": -2.9802069664001465,
156
+ "logps/chosen": -2.6169793605804443,
157
+ "logps/rejected": -70.75330352783203,
158
+ "loss": 0.452,
159
+ "rewards/accuracies": 1.0,
160
+ "rewards/chosen": 0.23872241377830505,
161
+ "rewards/margins": 0.5613620281219482,
162
+ "rewards/rejected": -0.32263970375061035,
163
+ "step": 100
164
+ },
165
+ {
166
+ "epoch": 0.19,
167
+ "learning_rate": 4.8776412907378845e-06,
168
+ "logits/chosen": -2.9479832649230957,
169
+ "logits/rejected": -2.949766159057617,
170
+ "logps/chosen": -1.910538911819458,
171
+ "logps/rejected": -89.78424835205078,
172
+ "loss": 0.3907,
173
+ "rewards/accuracies": 1.0,
174
+ "rewards/chosen": 0.24221813678741455,
175
+ "rewards/margins": 0.7395305037498474,
176
+ "rewards/rejected": -0.4973124563694,
177
+ "step": 110
178
+ },
179
+ {
180
+ "epoch": 0.21,
181
+ "learning_rate": 4.8266584586307555e-06,
182
+ "logits/chosen": -2.9272332191467285,
183
+ "logits/rejected": -2.9193713665008545,
184
+ "logps/chosen": -1.9234750270843506,
185
+ "logps/rejected": -99.62001037597656,
186
+ "loss": 0.3562,
187
+ "rewards/accuracies": 1.0,
188
+ "rewards/chosen": 0.2426760196685791,
189
+ "rewards/margins": 0.8515622019767761,
190
+ "rewards/rejected": -0.6088861227035522,
191
+ "step": 120
192
+ },
193
+ {
194
+ "epoch": 0.22,
195
+ "learning_rate": 4.7671859021263635e-06,
196
+ "logits/chosen": -2.900512456893921,
197
+ "logits/rejected": -2.8844547271728516,
198
+ "logps/chosen": -6.084620952606201,
199
+ "logps/rejected": -132.8142852783203,
200
+ "loss": 0.2791,
201
+ "rewards/accuracies": 1.0,
202
+ "rewards/chosen": 0.20018012821674347,
203
+ "rewards/margins": 1.139133095741272,
204
+ "rewards/rejected": -0.9389530420303345,
205
+ "step": 130
206
+ },
207
+ {
208
+ "epoch": 0.24,
209
+ "learning_rate": 4.699440630133794e-06,
210
+ "logits/chosen": -2.902156114578247,
211
+ "logits/rejected": -2.8723764419555664,
212
+ "logps/chosen": -4.3537917137146,
213
+ "logps/rejected": -166.32823181152344,
214
+ "loss": 0.2031,
215
+ "rewards/accuracies": 1.0,
216
+ "rewards/chosen": 0.2199852466583252,
217
+ "rewards/margins": 1.4954392910003662,
218
+ "rewards/rejected": -1.275454044342041,
219
+ "step": 140
220
+ },
221
+ {
222
+ "epoch": 0.26,
223
+ "learning_rate": 4.623669837803803e-06,
224
+ "logits/chosen": -2.8253445625305176,
225
+ "logits/rejected": -2.785407304763794,
226
+ "logps/chosen": -7.810912132263184,
227
+ "logps/rejected": -187.6137237548828,
228
+ "loss": 0.1766,
229
+ "rewards/accuracies": 1.0,
230
+ "rewards/chosen": 0.1820736676454544,
231
+ "rewards/margins": 1.6826276779174805,
232
+ "rewards/rejected": -1.500554084777832,
233
+ "step": 150
234
+ },
235
+ {
236
+ "epoch": 0.28,
237
+ "learning_rate": 4.5401500045405126e-06,
238
+ "logits/chosen": -2.769723653793335,
239
+ "logits/rejected": -2.7112717628479004,
240
+ "logps/chosen": -12.122523307800293,
241
+ "logps/rejected": -232.715576171875,
242
+ "loss": 0.119,
243
+ "rewards/accuracies": 1.0,
244
+ "rewards/chosen": 0.13884484767913818,
245
+ "rewards/margins": 2.0785183906555176,
246
+ "rewards/rejected": -1.9396737813949585,
247
+ "step": 160
248
+ },
249
+ {
250
+ "epoch": 0.29,
251
+ "learning_rate": 4.449185885158056e-06,
252
+ "logits/chosen": -2.666938304901123,
253
+ "logits/rejected": -2.58597993850708,
254
+ "logps/chosen": -15.82372760772705,
255
+ "logps/rejected": -269.86346435546875,
256
+ "loss": 0.0872,
257
+ "rewards/accuracies": 1.0,
258
+ "rewards/chosen": 0.10262912511825562,
259
+ "rewards/margins": 2.4057211875915527,
260
+ "rewards/rejected": -2.3030917644500732,
261
+ "step": 170
262
+ },
263
+ {
264
+ "epoch": 0.31,
265
+ "learning_rate": 4.351109397863309e-06,
266
+ "logits/chosen": -2.642421007156372,
267
+ "logits/rejected": -2.571937084197998,
268
+ "logps/chosen": -18.459171295166016,
269
+ "logps/rejected": -301.59368896484375,
270
+ "loss": 0.0649,
271
+ "rewards/accuracies": 1.0,
272
+ "rewards/chosen": 0.08005906641483307,
273
+ "rewards/margins": 2.7106738090515137,
274
+ "rewards/rejected": -2.630614757537842,
275
+ "step": 180
276
+ },
277
+ {
278
+ "epoch": 0.33,
279
+ "learning_rate": 4.246278413122344e-06,
280
+ "logits/chosen": -2.5845489501953125,
281
+ "logits/rejected": -2.4841175079345703,
282
+ "logps/chosen": -21.19426727294922,
283
+ "logps/rejected": -328.6267395019531,
284
+ "loss": 0.0505,
285
+ "rewards/accuracies": 1.0,
286
+ "rewards/chosen": 0.05268295854330063,
287
+ "rewards/margins": 2.9683151245117188,
288
+ "rewards/rejected": -2.9156322479248047,
289
+ "step": 190
290
+ },
291
+ {
292
+ "epoch": 0.35,
293
+ "learning_rate": 4.135075447829912e-06,
294
+ "logits/chosen": -2.586606979370117,
295
+ "logits/rejected": -2.4812800884246826,
296
+ "logps/chosen": -24.817462921142578,
297
+ "logps/rejected": -357.98626708984375,
298
+ "loss": 0.0398,
299
+ "rewards/accuracies": 1.0,
300
+ "rewards/chosen": 0.016275839880108833,
301
+ "rewards/margins": 3.2163288593292236,
302
+ "rewards/rejected": -3.2000529766082764,
303
+ "step": 200
304
+ },
305
+ {
306
+ "epoch": 0.36,
307
+ "learning_rate": 4.017906269546778e-06,
308
+ "logits/chosen": -2.5376362800598145,
309
+ "logits/rejected": -2.4408886432647705,
310
+ "logps/chosen": -29.325183868408203,
311
+ "logps/rejected": -380.6731872558594,
312
+ "loss": 0.0331,
313
+ "rewards/accuracies": 1.0,
314
+ "rewards/chosen": -0.02684500254690647,
315
+ "rewards/margins": 3.4006056785583496,
316
+ "rewards/rejected": -3.427450656890869,
317
+ "step": 210
318
+ },
319
+ {
320
+ "epoch": 0.38,
321
+ "learning_rate": 3.895198415897896e-06,
322
+ "logits/chosen": -2.5046138763427734,
323
+ "logits/rejected": -2.3839638233184814,
324
+ "logps/chosen": -33.11587142944336,
325
+ "logps/rejected": -399.06591796875,
326
+ "loss": 0.0289,
327
+ "rewards/accuracies": 1.0,
328
+ "rewards/chosen": -0.0685315877199173,
329
+ "rewards/margins": 3.547637939453125,
330
+ "rewards/rejected": -3.6161694526672363,
331
+ "step": 220
332
+ },
333
+ {
334
+ "epoch": 0.4,
335
+ "learning_rate": 3.767399634533976e-06,
336
+ "logits/chosen": -2.4707531929016113,
337
+ "logits/rejected": -2.362856388092041,
338
+ "logps/chosen": -38.450321197509766,
339
+ "logps/rejected": -421.177490234375,
340
+ "loss": 0.0243,
341
+ "rewards/accuracies": 1.0,
342
+ "rewards/chosen": -0.12026266753673553,
343
+ "rewards/margins": 3.7206790447235107,
344
+ "rewards/rejected": -3.8409416675567627,
345
+ "step": 230
346
+ },
347
+ {
348
+ "epoch": 0.41,
349
+ "learning_rate": 3.634976249348867e-06,
350
+ "logits/chosen": -2.4691028594970703,
351
+ "logits/rejected": -2.3579373359680176,
352
+ "logps/chosen": -39.478824615478516,
353
+ "logps/rejected": -435.550048828125,
354
+ "loss": 0.0216,
355
+ "rewards/accuracies": 1.0,
356
+ "rewards/chosen": -0.12904781103134155,
357
+ "rewards/margins": 3.84360933303833,
358
+ "rewards/rejected": -3.972656726837158,
359
+ "step": 240
360
+ },
361
+ {
362
+ "epoch": 0.43,
363
+ "learning_rate": 3.4984114589142388e-06,
364
+ "logits/chosen": -2.391221284866333,
365
+ "logits/rejected": -2.26558256149292,
366
+ "logps/chosen": -42.098384857177734,
367
+ "logps/rejected": -447.58306884765625,
368
+ "loss": 0.0194,
369
+ "rewards/accuracies": 1.0,
370
+ "rewards/chosen": -0.1557580828666687,
371
+ "rewards/margins": 3.939157009124756,
372
+ "rewards/rejected": -4.094915390014648,
373
+ "step": 250
374
+ },
375
+ {
376
+ "epoch": 0.45,
377
+ "learning_rate": 3.3582035733403963e-06,
378
+ "logits/chosen": -2.3836541175842285,
379
+ "logits/rejected": -2.2646937370300293,
380
+ "logps/chosen": -45.768280029296875,
381
+ "logps/rejected": -462.4888610839844,
382
+ "loss": 0.0176,
383
+ "rewards/accuracies": 1.0,
384
+ "rewards/chosen": -0.19619223475456238,
385
+ "rewards/margins": 4.036476135253906,
386
+ "rewards/rejected": -4.232667922973633,
387
+ "step": 260
388
+ },
389
+ {
390
+ "epoch": 0.47,
391
+ "learning_rate": 3.214864195996723e-06,
392
+ "logits/chosen": -2.4122414588928223,
393
+ "logits/rejected": -2.3053128719329834,
394
+ "logps/chosen": -49.939971923828125,
395
+ "logps/rejected": -473.1507263183594,
396
+ "loss": 0.0166,
397
+ "rewards/accuracies": 1.0,
398
+ "rewards/chosen": -0.23972025513648987,
399
+ "rewards/margins": 4.099632740020752,
400
+ "rewards/rejected": -4.339352130889893,
401
+ "step": 270
402
+ },
403
+ {
404
+ "epoch": 0.48,
405
+ "learning_rate": 3.068916356726475e-06,
406
+ "logits/chosen": -2.410609483718872,
407
+ "logits/rejected": -2.286745548248291,
408
+ "logps/chosen": -51.68450164794922,
409
+ "logps/rejected": -478.81976318359375,
410
+ "loss": 0.0157,
411
+ "rewards/accuracies": 1.0,
412
+ "rewards/chosen": -0.2583352029323578,
413
+ "rewards/margins": 4.155708312988281,
414
+ "rewards/rejected": -4.414044380187988,
415
+ "step": 280
416
+ },
417
+ {
418
+ "epoch": 0.5,
419
+ "learning_rate": 2.920892603367596e-06,
420
+ "logits/chosen": -2.359619617462158,
421
+ "logits/rejected": -2.2012834548950195,
422
+ "logps/chosen": -57.13511276245117,
423
+ "logps/rejected": -506.31842041015625,
424
+ "loss": 0.0128,
425
+ "rewards/accuracies": 1.0,
426
+ "rewards/chosen": -0.3052617609500885,
427
+ "rewards/margins": 4.382186412811279,
428
+ "rewards/rejected": -4.687448501586914,
429
+ "step": 290
430
+ },
431
+ {
432
+ "epoch": 0.52,
433
+ "learning_rate": 2.771333058543416e-06,
434
+ "logits/chosen": -2.3818418979644775,
435
+ "logits/rejected": -2.2266857624053955,
436
+ "logps/chosen": -60.3787727355957,
437
+ "logps/rejected": -535.8792724609375,
438
+ "loss": 0.0097,
439
+ "rewards/accuracies": 1.0,
440
+ "rewards/chosen": -0.3381674885749817,
441
+ "rewards/margins": 4.6410980224609375,
442
+ "rewards/rejected": -4.979265213012695,
443
+ "step": 300
444
+ },
445
+ {
446
+ "epoch": 0.54,
447
+ "learning_rate": 2.620783448813768e-06,
448
+ "logits/chosen": -2.3339812755584717,
449
+ "logits/rejected": -2.1715776920318604,
450
+ "logps/chosen": -61.98762130737305,
451
+ "logps/rejected": -553.9091796875,
452
+ "loss": 0.0085,
453
+ "rewards/accuracies": 1.0,
454
+ "rewards/chosen": -0.35833388566970825,
455
+ "rewards/margins": 4.780001640319824,
456
+ "rewards/rejected": -5.138335704803467,
457
+ "step": 310
458
+ },
459
+ {
460
+ "epoch": 0.55,
461
+ "learning_rate": 2.4697931133779566e-06,
462
+ "logits/chosen": -2.323732852935791,
463
+ "logits/rejected": -2.1703879833221436,
464
+ "logps/chosen": -69.43708801269531,
465
+ "logps/rejected": -569.32080078125,
466
+ "loss": 0.0077,
467
+ "rewards/accuracies": 1.0,
468
+ "rewards/chosen": -0.43295058608055115,
469
+ "rewards/margins": 4.877272129058838,
470
+ "rewards/rejected": -5.3102216720581055,
471
+ "step": 320
472
+ },
473
+ {
474
+ "epoch": 0.57,
475
+ "learning_rate": 2.3189129995955944e-06,
476
+ "logits/chosen": -2.327399730682373,
477
+ "logits/rejected": -2.1635982990264893,
478
+ "logps/chosen": -73.22216796875,
479
+ "logps/rejected": -579.3975830078125,
480
+ "loss": 0.0076,
481
+ "rewards/accuracies": 1.0,
482
+ "rewards/chosen": -0.4656201899051666,
483
+ "rewards/margins": 4.9330034255981445,
484
+ "rewards/rejected": -5.398622989654541,
485
+ "step": 330
486
+ },
487
+ {
488
+ "epoch": 0.59,
489
+ "learning_rate": 2.168693652639432e-06,
490
+ "logits/chosen": -2.413621187210083,
491
+ "logits/rejected": -2.260995388031006,
492
+ "logps/chosen": -70.82634735107422,
493
+ "logps/rejected": -586.3375244140625,
494
+ "loss": 0.0068,
495
+ "rewards/accuracies": 1.0,
496
+ "rewards/chosen": -0.44182008504867554,
497
+ "rewards/margins": 5.035239219665527,
498
+ "rewards/rejected": -5.477059364318848,
499
+ "step": 340
500
+ },
501
+ {
502
+ "epoch": 0.61,
503
+ "learning_rate": 2.019683206615729e-06,
504
+ "logits/chosen": -2.2976067066192627,
505
+ "logits/rejected": -2.1246187686920166,
506
+ "logps/chosen": -72.80682373046875,
507
+ "logps/rejected": -594.421142578125,
508
+ "loss": 0.0062,
509
+ "rewards/accuracies": 1.0,
510
+ "rewards/chosen": -0.46819859743118286,
511
+ "rewards/margins": 5.091480731964111,
512
+ "rewards/rejected": -5.5596795082092285,
513
+ "step": 350
514
+ },
515
+ {
516
+ "epoch": 0.62,
517
+ "learning_rate": 1.872425384482346e-06,
518
+ "logits/chosen": -2.36497163772583,
519
+ "logits/rejected": -2.222611427307129,
520
+ "logps/chosen": -72.7022705078125,
521
+ "logps/rejected": -596.8053588867188,
522
+ "loss": 0.006,
523
+ "rewards/accuracies": 1.0,
524
+ "rewards/chosen": -0.4603179097175598,
525
+ "rewards/margins": 5.118600845336914,
526
+ "rewards/rejected": -5.578917980194092,
527
+ "step": 360
528
+ },
529
+ {
530
+ "epoch": 0.64,
531
+ "learning_rate": 1.7274575140626318e-06,
532
+ "logits/chosen": -2.3522539138793945,
533
+ "logits/rejected": -2.1843767166137695,
534
+ "logps/chosen": -77.02735137939453,
535
+ "logps/rejected": -607.0662841796875,
536
+ "loss": 0.0057,
537
+ "rewards/accuracies": 1.0,
538
+ "rewards/chosen": -0.507721483707428,
539
+ "rewards/margins": 5.175690174102783,
540
+ "rewards/rejected": -5.683411598205566,
541
+ "step": 370
542
+ },
543
+ {
544
+ "epoch": 0.66,
545
+ "learning_rate": 1.5853085673944695e-06,
546
+ "logits/chosen": -2.3788235187530518,
547
+ "logits/rejected": -2.2348849773406982,
548
+ "logps/chosen": -77.29615783691406,
549
+ "logps/rejected": -609.80322265625,
550
+ "loss": 0.0056,
551
+ "rewards/accuracies": 1.0,
552
+ "rewards/chosen": -0.5133799910545349,
553
+ "rewards/margins": 5.1920671463012695,
554
+ "rewards/rejected": -5.705447196960449,
555
+ "step": 380
556
+ },
557
+ {
558
+ "epoch": 0.67,
559
+ "learning_rate": 1.4464972305686778e-06,
560
+ "logits/chosen": -2.3803200721740723,
561
+ "logits/rejected": -2.229212999343872,
562
+ "logps/chosen": -79.22676849365234,
563
+ "logps/rejected": -602.490966796875,
564
+ "loss": 0.0064,
565
+ "rewards/accuracies": 1.0,
566
+ "rewards/chosen": -0.5286643505096436,
567
+ "rewards/margins": 5.111319541931152,
568
+ "rewards/rejected": -5.639983177185059,
569
+ "step": 390
570
+ },
571
+ {
572
+ "epoch": 0.69,
573
+ "learning_rate": 1.3115300110997097e-06,
574
+ "logits/chosen": -2.2747700214385986,
575
+ "logits/rejected": -2.100229024887085,
576
+ "logps/chosen": -77.5864486694336,
577
+ "logps/rejected": -604.7847900390625,
578
+ "loss": 0.0058,
579
+ "rewards/accuracies": 1.0,
580
+ "rewards/chosen": -0.5120630860328674,
581
+ "rewards/margins": 5.148989200592041,
582
+ "rewards/rejected": -5.6610517501831055,
583
+ "step": 400
584
+ },
585
+ {
586
+ "epoch": 0.71,
587
+ "learning_rate": 1.1808993897346679e-06,
588
+ "logits/chosen": -2.389352321624756,
589
+ "logits/rejected": -2.2429161071777344,
590
+ "logps/chosen": -77.43450927734375,
591
+ "logps/rejected": -608.711181640625,
592
+ "loss": 0.0056,
593
+ "rewards/accuracies": 1.0,
594
+ "rewards/chosen": -0.5119161009788513,
595
+ "rewards/margins": 5.186758995056152,
596
+ "rewards/rejected": -5.69867467880249,
597
+ "step": 410
598
+ },
599
+ {
600
+ "epoch": 0.73,
601
+ "learning_rate": 1.0550820234444627e-06,
602
+ "logits/chosen": -2.3463988304138184,
603
+ "logits/rejected": -2.1632635593414307,
604
+ "logps/chosen": -78.80754852294922,
605
+ "logps/rejected": -609.238037109375,
606
+ "loss": 0.0057,
607
+ "rewards/accuracies": 1.0,
608
+ "rewards/chosen": -0.5245837569236755,
609
+ "rewards/margins": 5.180685043334961,
610
+ "rewards/rejected": -5.705269813537598,
611
+ "step": 420
612
+ },
613
+ {
614
+ "epoch": 0.74,
615
+ "learning_rate": 9.345370061542158e-07,
616
+ "logits/chosen": -2.351940155029297,
617
+ "logits/rejected": -2.187344551086426,
618
+ "logps/chosen": -79.23500061035156,
619
+ "logps/rejected": -604.2271118164062,
620
+ "loss": 0.006,
621
+ "rewards/accuracies": 1.0,
622
+ "rewards/chosen": -0.5345843434333801,
623
+ "rewards/margins": 5.121103286743164,
624
+ "rewards/rejected": -5.65568733215332,
625
+ "step": 430
626
+ },
627
+ {
628
+ "epoch": 0.76,
629
+ "learning_rate": 8.197041935593181e-07,
630
+ "logits/chosen": -2.3927040100097656,
631
+ "logits/rejected": -2.2277729511260986,
632
+ "logps/chosen": -76.28541564941406,
633
+ "logps/rejected": -600.7366943359375,
634
+ "loss": 0.0061,
635
+ "rewards/accuracies": 1.0,
636
+ "rewards/chosen": -0.5005275011062622,
637
+ "rewards/margins": 5.117630958557129,
638
+ "rewards/rejected": -5.618158340454102,
639
+ "step": 440
640
+ },
641
+ {
642
+ "epoch": 0.78,
643
+ "learning_rate": 7.110025981396976e-07,
644
+ "logits/chosen": -2.294013023376465,
645
+ "logits/rejected": -2.102651357650757,
646
+ "logps/chosen": -79.84446716308594,
647
+ "logps/rejected": -609.43310546875,
648
+ "loss": 0.0057,
649
+ "rewards/accuracies": 1.0,
650
+ "rewards/chosen": -0.5404741168022156,
651
+ "rewards/margins": 5.176560401916504,
652
+ "rewards/rejected": -5.717034816741943,
653
+ "step": 450
654
+ },
655
+ {
656
+ "epoch": 0.8,
657
+ "learning_rate": 6.088288602287159e-07,
658
+ "logits/chosen": -2.4395508766174316,
659
+ "logits/rejected": -2.2818832397460938,
660
+ "logps/chosen": -79.54411315917969,
661
+ "logps/rejected": -610.0706176757812,
662
+ "loss": 0.0056,
663
+ "rewards/accuracies": 1.0,
664
+ "rewards/chosen": -0.5314868092536926,
665
+ "rewards/margins": 5.184579849243164,
666
+ "rewards/rejected": -5.716067314147949,
667
+ "step": 460
668
+ },
669
+ {
670
+ "epoch": 0.81,
671
+ "learning_rate": 5.135558007156146e-07,
672
+ "logits/chosen": -2.4146294593811035,
673
+ "logits/rejected": -2.2654051780700684,
674
+ "logps/chosen": -78.92245483398438,
675
+ "logps/rejected": -611.4479370117188,
676
+ "loss": 0.0055,
677
+ "rewards/accuracies": 1.0,
678
+ "rewards/chosen": -0.5293115377426147,
679
+ "rewards/margins": 5.204201698303223,
680
+ "rewards/rejected": -5.733513832092285,
681
+ "step": 470
682
+ },
683
+ {
684
+ "epoch": 0.83,
685
+ "learning_rate": 4.255310606625124e-07,
686
+ "logits/chosen": -2.403562068939209,
687
+ "logits/rejected": -2.2434592247009277,
688
+ "logps/chosen": -78.9774398803711,
689
+ "logps/rejected": -610.0330200195312,
690
+ "loss": 0.0057,
691
+ "rewards/accuracies": 1.0,
692
+ "rewards/chosen": -0.5241142511367798,
693
+ "rewards/margins": 5.185243129730225,
694
+ "rewards/rejected": -5.709357738494873,
695
+ "step": 480
696
+ },
697
+ {
698
+ "epoch": 0.85,
699
+ "learning_rate": 3.450758327998768e-07,
700
+ "logits/chosen": -2.3373961448669434,
701
+ "logits/rejected": -2.1586098670959473,
702
+ "logps/chosen": -80.73445129394531,
703
+ "logps/rejected": -612.7761840820312,
704
+ "loss": 0.0055,
705
+ "rewards/accuracies": 1.0,
706
+ "rewards/chosen": -0.5439419746398926,
707
+ "rewards/margins": 5.210984230041504,
708
+ "rewards/rejected": -5.754925727844238,
709
+ "step": 490
710
+ },
711
+ {
712
+ "epoch": 0.86,
713
+ "learning_rate": 2.7248368952908055e-07,
714
+ "logits/chosen": -2.3969485759735107,
715
+ "logits/rejected": -2.228907585144043,
716
+ "logps/chosen": -79.6150131225586,
717
+ "logps/rejected": -609.5167236328125,
718
+ "loss": 0.0057,
719
+ "rewards/accuracies": 1.0,
720
+ "rewards/chosen": -0.535262942314148,
721
+ "rewards/margins": 5.172930717468262,
722
+ "rewards/rejected": -5.708194255828857,
723
+ "step": 500
724
+ },
725
+ {
726
+ "epoch": 0.88,
727
+ "learning_rate": 2.0801951170854402e-07,
728
+ "logits/chosen": -2.267817497253418,
729
+ "logits/rejected": -2.0743813514709473,
730
+ "logps/chosen": -77.05073547363281,
731
+ "logps/rejected": -607.4371948242188,
732
+ "loss": 0.0057,
733
+ "rewards/accuracies": 1.0,
734
+ "rewards/chosen": -0.5123855471611023,
735
+ "rewards/margins": 5.169266223907471,
736
+ "rewards/rejected": -5.6816511154174805,
737
+ "step": 510
738
+ },
739
+ {
740
+ "epoch": 0.9,
741
+ "learning_rate": 1.5191852213221198e-07,
742
+ "logits/chosen": -2.4104533195495605,
743
+ "logits/rejected": -2.257448196411133,
744
+ "logps/chosen": -79.09856414794922,
745
+ "logps/rejected": -612.449951171875,
746
+ "loss": 0.0054,
747
+ "rewards/accuracies": 1.0,
748
+ "rewards/chosen": -0.5263990163803101,
749
+ "rewards/margins": 5.221766471862793,
750
+ "rewards/rejected": -5.748165130615234,
751
+ "step": 520
752
+ },
753
+ {
754
+ "epoch": 0.92,
755
+ "learning_rate": 1.0438542722708444e-07,
756
+ "logits/chosen": -2.4659290313720703,
757
+ "logits/rejected": -2.336254596710205,
758
+ "logps/chosen": -79.4029541015625,
759
+ "logps/rejected": -611.0933837890625,
760
+ "loss": 0.0056,
761
+ "rewards/accuracies": 1.0,
762
+ "rewards/chosen": -0.5314724445343018,
763
+ "rewards/margins": 5.193521499633789,
764
+ "rewards/rejected": -5.724993705749512,
765
+ "step": 530
766
+ },
767
+ {
768
+ "epoch": 0.93,
769
+ "learning_rate": 6.559367010166629e-08,
770
+ "logits/chosen": -2.4373106956481934,
771
+ "logits/rejected": -2.2991061210632324,
772
+ "logps/chosen": -77.79346466064453,
773
+ "logps/rejected": -610.4267578125,
774
+ "loss": 0.0056,
775
+ "rewards/accuracies": 1.0,
776
+ "rewards/chosen": -0.514025092124939,
777
+ "rewards/margins": 5.191291809082031,
778
+ "rewards/rejected": -5.705316543579102,
779
+ "step": 540
780
+ },
781
+ {
782
+ "epoch": 0.95,
783
+ "learning_rate": 3.568479767087296e-08,
784
+ "logits/chosen": -2.378970146179199,
785
+ "logits/rejected": -2.2227816581726074,
786
+ "logps/chosen": -79.35990905761719,
787
+ "logps/rejected": -606.5776977539062,
788
+ "loss": 0.0057,
789
+ "rewards/accuracies": 1.0,
790
+ "rewards/chosen": -0.5338703393936157,
791
+ "rewards/margins": 5.170326232910156,
792
+ "rewards/rejected": -5.704196929931641,
793
+ "step": 550
794
+ },
795
+ {
796
+ "epoch": 0.97,
797
+ "learning_rate": 1.4767944166687032e-08,
798
+ "logits/chosen": -2.3564634323120117,
799
+ "logits/rejected": -2.1833720207214355,
800
+ "logps/chosen": -80.16700744628906,
801
+ "logps/rejected": -610.6815185546875,
802
+ "loss": 0.0056,
803
+ "rewards/accuracies": 1.0,
804
+ "rewards/chosen": -0.5410750508308411,
805
+ "rewards/margins": 5.180901527404785,
806
+ "rewards/rejected": -5.721976280212402,
807
+ "step": 560
808
+ },
809
+ {
810
+ "epoch": 0.99,
811
+ "learning_rate": 2.919432919183396e-09,
812
+ "logits/chosen": -2.4240996837615967,
813
+ "logits/rejected": -2.268498659133911,
814
+ "logps/chosen": -79.65086364746094,
815
+ "logps/rejected": -616.8253784179688,
816
+ "loss": 0.0052,
817
+ "rewards/accuracies": 1.0,
818
+ "rewards/chosen": -0.5321217179298401,
819
+ "rewards/margins": 5.257809638977051,
820
+ "rewards/rejected": -5.789931297302246,
821
+ "step": 570
822
+ },
823
+ {
824
+ "epoch": 1.0,
825
+ "step": 578,
826
+ "total_flos": 0.0,
827
+ "train_loss": 0.14528509169481824,
828
+ "train_runtime": 467.3095,
829
+ "train_samples_per_second": 2.476,
830
+ "train_steps_per_second": 1.237
831
+ }
832
+ ],
833
+ "logging_steps": 10,
834
+ "max_steps": 578,
835
+ "num_input_tokens_seen": 0,
836
+ "num_train_epochs": 1,
837
+ "save_steps": 1000,
838
+ "total_flos": 0.0,
839
+ "train_batch_size": 1,
840
+ "trial_name": null,
841
+ "trial_params": null
842
+ }
training_args.bin ADDED
@@ -0,0 +1,3 @@
 
 
 
 
1
+ version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e851b1d4339e2249f6dc808a607c94ed9a80f5350c9647a88aebad26ed98d655
3
+ size 4856