AmberYifan commited on
Commit
57dbea5
1 Parent(s): bcb7f04

Model save

Browse files
README.md ADDED
@@ -0,0 +1,85 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: llama2
3
+ base_model: meta-llama/Llama-2-7b-hf
4
+ tags:
5
+ - generated_from_trainer
6
+ model-index:
7
+ - name: Llama-2-7b-spin-rephrased-10k
8
+ results: []
9
+ ---
10
+
11
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
12
+ should probably proofread and complete it, then remove this comment. -->
13
+
14
+ # Llama-2-7b-spin-rephrased-10k
15
+
16
+ This model is a fine-tuned version of [meta-llama/Llama-2-7b-hf](https://huggingface.co/meta-llama/Llama-2-7b-hf) on an unknown dataset.
17
+ It achieves the following results on the evaluation set:
18
+ - Loss: 0.1071
19
+ - Rewards/real: 10.2171
20
+ - Rewards/generated: -7.6243
21
+ - Rewards/accuracies: 1.0
22
+ - Rewards/margins: 17.8413
23
+ - Logps/generated: -358.9117
24
+ - Logps/real: -104.6875
25
+ - Logits/generated: -0.8781
26
+ - Logits/real: -1.4494
27
+
28
+ ## Model description
29
+
30
+ More information needed
31
+
32
+ ## Intended uses & limitations
33
+
34
+ More information needed
35
+
36
+ ## Training and evaluation data
37
+
38
+ More information needed
39
+
40
+ ## Training procedure
41
+
42
+ ### Training hyperparameters
43
+
44
+ The following hyperparameters were used during training:
45
+ - learning_rate: 5e-07
46
+ - train_batch_size: 4
47
+ - eval_batch_size: 4
48
+ - seed: 42
49
+ - distributed_type: multi-GPU
50
+ - num_devices: 4
51
+ - gradient_accumulation_steps: 2
52
+ - total_train_batch_size: 32
53
+ - total_eval_batch_size: 16
54
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
55
+ - lr_scheduler_type: linear
56
+ - lr_scheduler_warmup_ratio: 0.1
57
+ - num_epochs: 3
58
+
59
+ ### Training results
60
+
61
+ | Training Loss | Epoch | Step | Validation Loss | Rewards/real | Rewards/generated | Rewards/accuracies | Rewards/margins | Logps/generated | Logps/real | Logits/generated | Logits/real |
62
+ |:-------------:|:------:|:----:|:---------------:|:------------:|:-----------------:|:------------------:|:---------------:|:---------------:|:----------:|:----------------:|:-----------:|
63
+ | 0.1687 | 0.1984 | 62 | 0.1554 | 5.2053 | -5.2548 | 1.0 | 10.4601 | -335.2168 | -154.8048 | -0.7218 | -0.4019 |
64
+ | 0.1204 | 0.3968 | 124 | 0.1153 | 9.3697 | -4.5235 | 1.0 | 13.8932 | -327.9041 | -113.1613 | -0.8262 | -1.1627 |
65
+ | 0.1114 | 0.5952 | 186 | 0.1125 | 9.6740 | -5.3166 | 1.0 | 14.9906 | -335.8354 | -110.1185 | -0.8446 | -1.2393 |
66
+ | 0.1094 | 0.7936 | 248 | 0.1110 | 9.8335 | -5.4853 | 1.0 | 15.3188 | -337.5219 | -108.5231 | -0.8538 | -1.2560 |
67
+ | 0.1115 | 0.992 | 310 | 0.1100 | 9.9127 | -6.4827 | 1.0 | 16.3954 | -347.4966 | -107.7317 | -0.8658 | -1.3304 |
68
+ | 0.1046 | 1.1904 | 372 | 0.1093 | 9.9819 | -6.6707 | 1.0 | 16.6526 | -349.3765 | -107.0395 | -0.8656 | -1.3633 |
69
+ | 0.1067 | 1.3888 | 434 | 0.1089 | 10.0127 | -7.5740 | 1.0 | 17.5868 | -358.4094 | -106.7308 | -0.8814 | -1.3898 |
70
+ | 0.1038 | 1.5872 | 496 | 0.1083 | 10.0730 | -7.0038 | 1.0 | 17.0768 | -352.7069 | -106.1281 | -0.8755 | -1.3615 |
71
+ | 0.0996 | 1.7856 | 558 | 0.1079 | 10.1219 | -7.0176 | 1.0 | 17.1396 | -352.8456 | -105.6391 | -0.8467 | -1.3431 |
72
+ | 0.1058 | 1.984 | 620 | 0.1077 | 10.1479 | -7.4808 | 1.0 | 17.6287 | -357.4770 | -105.3797 | -0.8821 | -1.4055 |
73
+ | 0.0995 | 2.1824 | 682 | 0.1074 | 10.1669 | -7.1947 | 1.0 | 17.3617 | -354.6166 | -105.1890 | -0.8781 | -1.4102 |
74
+ | 0.1017 | 2.3808 | 744 | 0.1073 | 10.1849 | -7.6243 | 1.0 | 17.8092 | -358.9117 | -105.0093 | -0.8806 | -1.4228 |
75
+ | 0.1031 | 2.5792 | 806 | 0.1072 | 10.2106 | -7.6581 | 1.0 | 17.8687 | -359.2500 | -104.7519 | -0.8787 | -1.4391 |
76
+ | 0.1025 | 2.7776 | 868 | 0.1071 | 10.2105 | -7.6804 | 1.0 | 17.8909 | -359.4730 | -104.7534 | -0.8824 | -1.4506 |
77
+ | 0.1067 | 2.976 | 930 | 0.1071 | 10.2171 | -7.6243 | 1.0 | 17.8413 | -358.9117 | -104.6875 | -0.8781 | -1.4494 |
78
+
79
+
80
+ ### Framework versions
81
+
82
+ - Transformers 4.43.3
83
+ - Pytorch 2.2.2+cu121
84
+ - Datasets 2.20.0
85
+ - Tokenizers 0.19.1
all_results.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 2.9952,
3
+ "total_flos": 0.0,
4
+ "train_loss": 0.1350623204285263,
5
+ "train_runtime": 11872.5465,
6
+ "train_samples": 9988,
7
+ "train_samples_per_second": 2.524,
8
+ "train_steps_per_second": 0.079
9
+ }
generation_config.json ADDED
@@ -0,0 +1,10 @@
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token_id": 1,
3
+ "do_sample": true,
4
+ "eos_token_id": 2,
5
+ "max_length": 4096,
6
+ "pad_token_id": 0,
7
+ "temperature": 0.6,
8
+ "top_p": 0.9,
9
+ "transformers_version": "4.43.3"
10
+ }
model-00001-of-00003.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:01517963eec4749074362db8de3e1e9f748fdd69613dd143d7c3315a87e1bf56
3
  size 4938985352
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:fd3a5ffd2e394a1c4a127aa1119ed1128e55b230e20363bccd8337a8cbc696bf
3
  size 4938985352
model-00002-of-00003.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:6b57301ca1713ff5ce2fb0b800d8b6285544cf47ea74744238e28c33346878c1
3
  size 4947390880
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:e1c06afe5bb0298a340b5b9c47a17cc9b15f2325ae647528b658fdaa59f4bdec
3
  size 4947390880
model-00003-of-00003.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:2af8a63985358103b12ea30d4fd55b6bc89662e8e2ca52c7b8212e4a87bbccaf
3
  size 3590488816
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:1e4a819d37345ad2d4030c990780abae1d8af68a6537ca196126865a64ada364
3
  size 3590488816
train_results.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 2.9952,
3
+ "total_flos": 0.0,
4
+ "train_loss": 0.1350623204285263,
5
+ "train_runtime": 11872.5465,
6
+ "train_samples": 9988,
7
+ "train_samples_per_second": 2.524,
8
+ "train_steps_per_second": 0.079
9
+ }
trainer_state.json ADDED
@@ -0,0 +1,1692 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": null,
3
+ "best_model_checkpoint": null,
4
+ "epoch": 2.9952,
5
+ "eval_steps": 62,
6
+ "global_step": 936,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "epoch": 0.0032,
13
+ "grad_norm": 76.02843905366466,
14
+ "learning_rate": 5.3191489361702125e-09,
15
+ "logits/generated": -1.098149061203003,
16
+ "logits/real": -0.27942562103271484,
17
+ "logps/generated": -261.58599853515625,
18
+ "logps/real": -202.4606475830078,
19
+ "loss": 0.9167,
20
+ "rewards/accuracies": 0.0,
21
+ "rewards/generated": 0.0,
22
+ "rewards/margins": 0.0,
23
+ "rewards/real": 0.0,
24
+ "step": 1
25
+ },
26
+ {
27
+ "epoch": 0.032,
28
+ "grad_norm": 83.5080388615708,
29
+ "learning_rate": 5.3191489361702123e-08,
30
+ "logits/generated": -0.964782178401947,
31
+ "logits/real": -0.43690696358680725,
32
+ "logps/generated": -277.5035400390625,
33
+ "logps/real": -213.53489685058594,
34
+ "loss": 0.8941,
35
+ "rewards/accuracies": 0.3888888955116272,
36
+ "rewards/generated": 0.007520654238760471,
37
+ "rewards/margins": -0.009737681597471237,
38
+ "rewards/real": -0.002217027824372053,
39
+ "step": 10
40
+ },
41
+ {
42
+ "epoch": 0.064,
43
+ "grad_norm": 66.11667224751555,
44
+ "learning_rate": 1.0638297872340425e-07,
45
+ "logits/generated": -1.0118372440338135,
46
+ "logits/real": -0.4319210946559906,
47
+ "logps/generated": -281.8486633300781,
48
+ "logps/real": -229.50259399414062,
49
+ "loss": 0.8482,
50
+ "rewards/accuracies": 0.887499988079071,
51
+ "rewards/generated": -0.08688100427389145,
52
+ "rewards/margins": 0.1324445903301239,
53
+ "rewards/real": 0.04556358605623245,
54
+ "step": 20
55
+ },
56
+ {
57
+ "epoch": 0.096,
58
+ "grad_norm": 83.1043523213915,
59
+ "learning_rate": 1.5957446808510638e-07,
60
+ "logits/generated": -0.911736011505127,
61
+ "logits/real": -0.6389068365097046,
62
+ "logps/generated": -287.8795471191406,
63
+ "logps/real": -189.71298217773438,
64
+ "loss": 0.6691,
65
+ "rewards/accuracies": 1.0,
66
+ "rewards/generated": -0.3395097553730011,
67
+ "rewards/margins": 0.5189359188079834,
68
+ "rewards/real": 0.1794261634349823,
69
+ "step": 30
70
+ },
71
+ {
72
+ "epoch": 0.128,
73
+ "grad_norm": 17.588838075583457,
74
+ "learning_rate": 2.127659574468085e-07,
75
+ "logits/generated": -0.9573461413383484,
76
+ "logits/real": -0.4479493498802185,
77
+ "logps/generated": -294.65081787109375,
78
+ "logps/real": -203.56137084960938,
79
+ "loss": 0.4483,
80
+ "rewards/accuracies": 1.0,
81
+ "rewards/generated": -0.8644129633903503,
82
+ "rewards/margins": 1.436708688735962,
83
+ "rewards/real": 0.5722957849502563,
84
+ "step": 40
85
+ },
86
+ {
87
+ "epoch": 0.16,
88
+ "grad_norm": 3.9686616342963887,
89
+ "learning_rate": 2.659574468085106e-07,
90
+ "logits/generated": -0.8112080693244934,
91
+ "logits/real": -0.5195661187171936,
92
+ "logps/generated": -303.88958740234375,
93
+ "logps/real": -186.9166259765625,
94
+ "loss": 0.2406,
95
+ "rewards/accuracies": 1.0,
96
+ "rewards/generated": -1.8978115320205688,
97
+ "rewards/margins": 3.3566880226135254,
98
+ "rewards/real": 1.458876371383667,
99
+ "step": 50
100
+ },
101
+ {
102
+ "epoch": 0.192,
103
+ "grad_norm": 0.873608373543979,
104
+ "learning_rate": 3.1914893617021275e-07,
105
+ "logits/generated": -0.7203534841537476,
106
+ "logits/real": -0.339523047208786,
107
+ "logps/generated": -325.28619384765625,
108
+ "logps/real": -161.00442504882812,
109
+ "loss": 0.1687,
110
+ "rewards/accuracies": 1.0,
111
+ "rewards/generated": -4.045280933380127,
112
+ "rewards/margins": 7.439988613128662,
113
+ "rewards/real": 3.3947081565856934,
114
+ "step": 60
115
+ },
116
+ {
117
+ "epoch": 0.1984,
118
+ "eval_logits/generated": -0.7218369841575623,
119
+ "eval_logits/real": -0.40187305212020874,
120
+ "eval_logps/generated": -335.216796875,
121
+ "eval_logps/real": -154.80484008789062,
122
+ "eval_loss": 0.155403733253479,
123
+ "eval_rewards/accuracies": 1.0,
124
+ "eval_rewards/generated": -5.2547712326049805,
125
+ "eval_rewards/margins": 10.460118293762207,
126
+ "eval_rewards/real": 5.205347537994385,
127
+ "eval_runtime": 27.3659,
128
+ "eval_samples_per_second": 7.308,
129
+ "eval_steps_per_second": 0.475,
130
+ "step": 62
131
+ },
132
+ {
133
+ "epoch": 0.224,
134
+ "grad_norm": 0.9090466290588308,
135
+ "learning_rate": 3.7234042553191484e-07,
136
+ "logits/generated": -0.7592553496360779,
137
+ "logits/real": -0.5271034240722656,
138
+ "logps/generated": -341.49017333984375,
139
+ "logps/real": -148.29876708984375,
140
+ "loss": 0.1549,
141
+ "rewards/accuracies": 1.0,
142
+ "rewards/generated": -4.969511985778809,
143
+ "rewards/margins": 10.437345504760742,
144
+ "rewards/real": 5.467833042144775,
145
+ "step": 70
146
+ },
147
+ {
148
+ "epoch": 0.256,
149
+ "grad_norm": 0.5548494311175409,
150
+ "learning_rate": 4.25531914893617e-07,
151
+ "logits/generated": -0.7383196949958801,
152
+ "logits/real": -0.8009077906608582,
153
+ "logps/generated": -321.6851806640625,
154
+ "logps/real": -123.8510971069336,
155
+ "loss": 0.1358,
156
+ "rewards/accuracies": 1.0,
157
+ "rewards/generated": -3.824869155883789,
158
+ "rewards/margins": 10.858232498168945,
159
+ "rewards/real": 7.033363342285156,
160
+ "step": 80
161
+ },
162
+ {
163
+ "epoch": 0.288,
164
+ "grad_norm": 0.42292249756861683,
165
+ "learning_rate": 4.787234042553192e-07,
166
+ "logits/generated": -0.8187915682792664,
167
+ "logits/real": -1.0330232381820679,
168
+ "logps/generated": -318.3337707519531,
169
+ "logps/real": -120.26998138427734,
170
+ "loss": 0.1213,
171
+ "rewards/accuracies": 1.0,
172
+ "rewards/generated": -3.1653218269348145,
173
+ "rewards/margins": 11.506210327148438,
174
+ "rewards/real": 8.340888977050781,
175
+ "step": 90
176
+ },
177
+ {
178
+ "epoch": 0.32,
179
+ "grad_norm": 0.41767196123848027,
180
+ "learning_rate": 4.96437054631829e-07,
181
+ "logits/generated": -0.8708025217056274,
182
+ "logits/real": -1.1798157691955566,
183
+ "logps/generated": -309.2374572753906,
184
+ "logps/real": -115.8353042602539,
185
+ "loss": 0.1134,
186
+ "rewards/accuracies": 1.0,
187
+ "rewards/generated": -3.7763671875,
188
+ "rewards/margins": 12.563192367553711,
189
+ "rewards/real": 8.786824226379395,
190
+ "step": 100
191
+ },
192
+ {
193
+ "epoch": 0.352,
194
+ "grad_norm": 0.4453179950421757,
195
+ "learning_rate": 4.904988123515439e-07,
196
+ "logits/generated": -0.8516510725021362,
197
+ "logits/real": -1.1489499807357788,
198
+ "logps/generated": -326.2418518066406,
199
+ "logps/real": -125.87767028808594,
200
+ "loss": 0.1137,
201
+ "rewards/accuracies": 1.0,
202
+ "rewards/generated": -4.001976013183594,
203
+ "rewards/margins": 13.618423461914062,
204
+ "rewards/real": 9.616447448730469,
205
+ "step": 110
206
+ },
207
+ {
208
+ "epoch": 0.384,
209
+ "grad_norm": 0.39875861828610537,
210
+ "learning_rate": 4.845605700712589e-07,
211
+ "logits/generated": -0.7948419451713562,
212
+ "logits/real": -1.1261526346206665,
213
+ "logps/generated": -331.29888916015625,
214
+ "logps/real": -118.65771484375,
215
+ "loss": 0.1204,
216
+ "rewards/accuracies": 1.0,
217
+ "rewards/generated": -4.266651153564453,
218
+ "rewards/margins": 13.141134262084961,
219
+ "rewards/real": 8.874483108520508,
220
+ "step": 120
221
+ },
222
+ {
223
+ "epoch": 0.3968,
224
+ "eval_logits/generated": -0.8262335658073425,
225
+ "eval_logits/real": -1.1627280712127686,
226
+ "eval_logps/generated": -327.9040832519531,
227
+ "eval_logps/real": -113.16130065917969,
228
+ "eval_loss": 0.11531589180231094,
229
+ "eval_rewards/accuracies": 1.0,
230
+ "eval_rewards/generated": -4.523497581481934,
231
+ "eval_rewards/margins": 13.89319896697998,
232
+ "eval_rewards/real": 9.369701385498047,
233
+ "eval_runtime": 31.0275,
234
+ "eval_samples_per_second": 6.446,
235
+ "eval_steps_per_second": 0.419,
236
+ "step": 124
237
+ },
238
+ {
239
+ "epoch": 0.416,
240
+ "grad_norm": 0.35711714781872617,
241
+ "learning_rate": 4.786223277909738e-07,
242
+ "logits/generated": -0.8075119853019714,
243
+ "logits/real": -1.1974109411239624,
244
+ "logps/generated": -326.8999938964844,
245
+ "logps/real": -121.57466125488281,
246
+ "loss": 0.1108,
247
+ "rewards/accuracies": 1.0,
248
+ "rewards/generated": -4.337772369384766,
249
+ "rewards/margins": 13.636154174804688,
250
+ "rewards/real": 9.298380851745605,
251
+ "step": 130
252
+ },
253
+ {
254
+ "epoch": 0.448,
255
+ "grad_norm": 0.3901220032340114,
256
+ "learning_rate": 4.7268408551068883e-07,
257
+ "logits/generated": -0.8377895355224609,
258
+ "logits/real": -1.1887013912200928,
259
+ "logps/generated": -325.7310791015625,
260
+ "logps/real": -123.4073715209961,
261
+ "loss": 0.1129,
262
+ "rewards/accuracies": 1.0,
263
+ "rewards/generated": -4.29754114151001,
264
+ "rewards/margins": 13.601455688476562,
265
+ "rewards/real": 9.303915023803711,
266
+ "step": 140
267
+ },
268
+ {
269
+ "epoch": 0.48,
270
+ "grad_norm": 0.3874502118142099,
271
+ "learning_rate": 4.667458432304038e-07,
272
+ "logits/generated": -0.8494235277175903,
273
+ "logits/real": -1.280548334121704,
274
+ "logps/generated": -323.9864807128906,
275
+ "logps/real": -115.0036849975586,
276
+ "loss": 0.1119,
277
+ "rewards/accuracies": 1.0,
278
+ "rewards/generated": -4.178435325622559,
279
+ "rewards/margins": 13.569964408874512,
280
+ "rewards/real": 9.391529083251953,
281
+ "step": 150
282
+ },
283
+ {
284
+ "epoch": 0.512,
285
+ "grad_norm": 0.408082769634744,
286
+ "learning_rate": 4.6080760095011875e-07,
287
+ "logits/generated": -0.8601406812667847,
288
+ "logits/real": -1.2258448600769043,
289
+ "logps/generated": -328.79156494140625,
290
+ "logps/real": -124.53865814208984,
291
+ "loss": 0.117,
292
+ "rewards/accuracies": 1.0,
293
+ "rewards/generated": -4.461178779602051,
294
+ "rewards/margins": 14.086248397827148,
295
+ "rewards/real": 9.625069618225098,
296
+ "step": 160
297
+ },
298
+ {
299
+ "epoch": 0.544,
300
+ "grad_norm": 0.36331526860516933,
301
+ "learning_rate": 4.548693586698337e-07,
302
+ "logits/generated": -0.8272736668586731,
303
+ "logits/real": -1.278106451034546,
304
+ "logps/generated": -331.06683349609375,
305
+ "logps/real": -119.9251480102539,
306
+ "loss": 0.1135,
307
+ "rewards/accuracies": 1.0,
308
+ "rewards/generated": -4.8740129470825195,
309
+ "rewards/margins": 14.372535705566406,
310
+ "rewards/real": 9.498522758483887,
311
+ "step": 170
312
+ },
313
+ {
314
+ "epoch": 0.576,
315
+ "grad_norm": 0.445865795650114,
316
+ "learning_rate": 4.4893111638954866e-07,
317
+ "logits/generated": -0.8073589205741882,
318
+ "logits/real": -1.2638658285140991,
319
+ "logps/generated": -342.365966796875,
320
+ "logps/real": -106.32196044921875,
321
+ "loss": 0.1114,
322
+ "rewards/accuracies": 1.0,
323
+ "rewards/generated": -5.440592288970947,
324
+ "rewards/margins": 14.317230224609375,
325
+ "rewards/real": 8.876636505126953,
326
+ "step": 180
327
+ },
328
+ {
329
+ "epoch": 0.5952,
330
+ "eval_logits/generated": -0.8445993065834045,
331
+ "eval_logits/real": -1.2393124103546143,
332
+ "eval_logps/generated": -335.83538818359375,
333
+ "eval_logps/real": -110.11853790283203,
334
+ "eval_loss": 0.11254960298538208,
335
+ "eval_rewards/accuracies": 1.0,
336
+ "eval_rewards/generated": -5.31662654876709,
337
+ "eval_rewards/margins": 14.990604400634766,
338
+ "eval_rewards/real": 9.673977851867676,
339
+ "eval_runtime": 25.7798,
340
+ "eval_samples_per_second": 7.758,
341
+ "eval_steps_per_second": 0.504,
342
+ "step": 186
343
+ },
344
+ {
345
+ "epoch": 0.608,
346
+ "grad_norm": 0.39320789077558344,
347
+ "learning_rate": 4.429928741092636e-07,
348
+ "logits/generated": -0.8406023979187012,
349
+ "logits/real": -1.2843071222305298,
350
+ "logps/generated": -316.73291015625,
351
+ "logps/real": -109.3148193359375,
352
+ "loss": 0.1103,
353
+ "rewards/accuracies": 1.0,
354
+ "rewards/generated": -4.513542175292969,
355
+ "rewards/margins": 14.111068725585938,
356
+ "rewards/real": 9.597526550292969,
357
+ "step": 190
358
+ },
359
+ {
360
+ "epoch": 0.64,
361
+ "grad_norm": 0.34126559985788285,
362
+ "learning_rate": 4.3705463182897863e-07,
363
+ "logits/generated": -0.8079656362533569,
364
+ "logits/real": -1.2299845218658447,
365
+ "logps/generated": -333.931396484375,
366
+ "logps/real": -113.5406723022461,
367
+ "loss": 0.1079,
368
+ "rewards/accuracies": 1.0,
369
+ "rewards/generated": -5.271535873413086,
370
+ "rewards/margins": 14.859156608581543,
371
+ "rewards/real": 9.587618827819824,
372
+ "step": 200
373
+ },
374
+ {
375
+ "epoch": 0.672,
376
+ "grad_norm": 0.37211551392957165,
377
+ "learning_rate": 4.311163895486936e-07,
378
+ "logits/generated": -0.8258484601974487,
379
+ "logits/real": -1.2792431116104126,
380
+ "logps/generated": -326.34539794921875,
381
+ "logps/real": -110.4327621459961,
382
+ "loss": 0.1062,
383
+ "rewards/accuracies": 1.0,
384
+ "rewards/generated": -4.544249534606934,
385
+ "rewards/margins": 13.822877883911133,
386
+ "rewards/real": 9.278627395629883,
387
+ "step": 210
388
+ },
389
+ {
390
+ "epoch": 0.704,
391
+ "grad_norm": 0.40528624225351695,
392
+ "learning_rate": 4.251781472684085e-07,
393
+ "logits/generated": -0.9005807638168335,
394
+ "logits/real": -1.367760419845581,
395
+ "logps/generated": -334.90667724609375,
396
+ "logps/real": -108.89170837402344,
397
+ "loss": 0.1082,
398
+ "rewards/accuracies": 1.0,
399
+ "rewards/generated": -5.304292678833008,
400
+ "rewards/margins": 14.677830696105957,
401
+ "rewards/real": 9.37353801727295,
402
+ "step": 220
403
+ },
404
+ {
405
+ "epoch": 0.736,
406
+ "grad_norm": 0.4299824458037775,
407
+ "learning_rate": 4.192399049881235e-07,
408
+ "logits/generated": -0.871414065361023,
409
+ "logits/real": -1.2680821418762207,
410
+ "logps/generated": -327.0166015625,
411
+ "logps/real": -125.11517333984375,
412
+ "loss": 0.1201,
413
+ "rewards/accuracies": 1.0,
414
+ "rewards/generated": -5.102071285247803,
415
+ "rewards/margins": 15.333366394042969,
416
+ "rewards/real": 10.231294631958008,
417
+ "step": 230
418
+ },
419
+ {
420
+ "epoch": 0.768,
421
+ "grad_norm": 0.3200345894830024,
422
+ "learning_rate": 4.1330166270783846e-07,
423
+ "logits/generated": -0.8396976590156555,
424
+ "logits/real": -1.2651922702789307,
425
+ "logps/generated": -349.18804931640625,
426
+ "logps/real": -117.8864517211914,
427
+ "loss": 0.1094,
428
+ "rewards/accuracies": 1.0,
429
+ "rewards/generated": -5.743192672729492,
430
+ "rewards/margins": 15.722867965698242,
431
+ "rewards/real": 9.979674339294434,
432
+ "step": 240
433
+ },
434
+ {
435
+ "epoch": 0.7936,
436
+ "eval_logits/generated": -0.8538265228271484,
437
+ "eval_logits/real": -1.2560151815414429,
438
+ "eval_logps/generated": -337.5219421386719,
439
+ "eval_logps/real": -108.52310180664062,
440
+ "eval_loss": 0.11097312718629837,
441
+ "eval_rewards/accuracies": 1.0,
442
+ "eval_rewards/generated": -5.485285758972168,
443
+ "eval_rewards/margins": 15.318806648254395,
444
+ "eval_rewards/real": 9.83352279663086,
445
+ "eval_runtime": 32.0233,
446
+ "eval_samples_per_second": 6.245,
447
+ "eval_steps_per_second": 0.406,
448
+ "step": 248
449
+ },
450
+ {
451
+ "epoch": 0.8,
452
+ "grad_norm": 0.37657892572843127,
453
+ "learning_rate": 4.0736342042755347e-07,
454
+ "logits/generated": -0.8381448984146118,
455
+ "logits/real": -1.293889045715332,
456
+ "logps/generated": -344.9915466308594,
457
+ "logps/real": -111.15998840332031,
458
+ "loss": 0.1113,
459
+ "rewards/accuracies": 1.0,
460
+ "rewards/generated": -5.526784896850586,
461
+ "rewards/margins": 15.003395080566406,
462
+ "rewards/real": 9.476609230041504,
463
+ "step": 250
464
+ },
465
+ {
466
+ "epoch": 0.832,
467
+ "grad_norm": 0.3794924953311087,
468
+ "learning_rate": 4.0142517814726837e-07,
469
+ "logits/generated": -0.8616489171981812,
470
+ "logits/real": -1.342829704284668,
471
+ "logps/generated": -333.8689270019531,
472
+ "logps/real": -110.18806457519531,
473
+ "loss": 0.1068,
474
+ "rewards/accuracies": 1.0,
475
+ "rewards/generated": -5.346303939819336,
476
+ "rewards/margins": 15.174580574035645,
477
+ "rewards/real": 9.828276634216309,
478
+ "step": 260
479
+ },
480
+ {
481
+ "epoch": 0.864,
482
+ "grad_norm": 0.38432288454636715,
483
+ "learning_rate": 3.9548693586698333e-07,
484
+ "logits/generated": -0.8341599702835083,
485
+ "logits/real": -1.2853684425354004,
486
+ "logps/generated": -340.5439147949219,
487
+ "logps/real": -111.25767517089844,
488
+ "loss": 0.1119,
489
+ "rewards/accuracies": 1.0,
490
+ "rewards/generated": -5.1574320793151855,
491
+ "rewards/margins": 14.693034172058105,
492
+ "rewards/real": 9.535603523254395,
493
+ "step": 270
494
+ },
495
+ {
496
+ "epoch": 0.896,
497
+ "grad_norm": 0.4988860977100256,
498
+ "learning_rate": 3.8954869358669834e-07,
499
+ "logits/generated": -0.842201828956604,
500
+ "logits/real": -1.3558833599090576,
501
+ "logps/generated": -338.8299560546875,
502
+ "logps/real": -110.4022216796875,
503
+ "loss": 0.1111,
504
+ "rewards/accuracies": 1.0,
505
+ "rewards/generated": -5.072219371795654,
506
+ "rewards/margins": 14.635177612304688,
507
+ "rewards/real": 9.562958717346191,
508
+ "step": 280
509
+ },
510
+ {
511
+ "epoch": 0.928,
512
+ "grad_norm": 0.38841887432580663,
513
+ "learning_rate": 3.836104513064133e-07,
514
+ "logits/generated": -0.8865805864334106,
515
+ "logits/real": -1.2532027959823608,
516
+ "logps/generated": -338.3769226074219,
517
+ "logps/real": -120.43833923339844,
518
+ "loss": 0.1084,
519
+ "rewards/accuracies": 1.0,
520
+ "rewards/generated": -5.086993217468262,
521
+ "rewards/margins": 15.214749336242676,
522
+ "rewards/real": 10.127756118774414,
523
+ "step": 290
524
+ },
525
+ {
526
+ "epoch": 0.96,
527
+ "grad_norm": 0.3616732188669415,
528
+ "learning_rate": 3.7767220902612825e-07,
529
+ "logits/generated": -0.8655015826225281,
530
+ "logits/real": -1.3518953323364258,
531
+ "logps/generated": -325.9730529785156,
532
+ "logps/real": -95.09385681152344,
533
+ "loss": 0.1028,
534
+ "rewards/accuracies": 1.0,
535
+ "rewards/generated": -5.991705417633057,
536
+ "rewards/margins": 14.74760627746582,
537
+ "rewards/real": 8.755899429321289,
538
+ "step": 300
539
+ },
540
+ {
541
+ "epoch": 0.992,
542
+ "grad_norm": 0.39359159204366234,
543
+ "learning_rate": 3.717339667458432e-07,
544
+ "logits/generated": -0.8543545603752136,
545
+ "logits/real": -1.3398587703704834,
546
+ "logps/generated": -338.96466064453125,
547
+ "logps/real": -116.5450668334961,
548
+ "loss": 0.1115,
549
+ "rewards/accuracies": 1.0,
550
+ "rewards/generated": -5.719244480133057,
551
+ "rewards/margins": 15.74243450164795,
552
+ "rewards/real": 10.02319049835205,
553
+ "step": 310
554
+ },
555
+ {
556
+ "epoch": 0.992,
557
+ "eval_logits/generated": -0.865822434425354,
558
+ "eval_logits/real": -1.33043372631073,
559
+ "eval_logps/generated": -347.4965515136719,
560
+ "eval_logps/real": -107.73165130615234,
561
+ "eval_loss": 0.11000669747591019,
562
+ "eval_rewards/accuracies": 1.0,
563
+ "eval_rewards/generated": -6.482741832733154,
564
+ "eval_rewards/margins": 16.39541244506836,
565
+ "eval_rewards/real": 9.912667274475098,
566
+ "eval_runtime": 24.7895,
567
+ "eval_samples_per_second": 8.068,
568
+ "eval_steps_per_second": 0.524,
569
+ "step": 310
570
+ },
571
+ {
572
+ "epoch": 1.024,
573
+ "grad_norm": 0.39684443120667556,
574
+ "learning_rate": 3.6579572446555817e-07,
575
+ "logits/generated": -0.9057809710502625,
576
+ "logits/real": -1.3521413803100586,
577
+ "logps/generated": -349.14471435546875,
578
+ "logps/real": -111.87225341796875,
579
+ "loss": 0.1073,
580
+ "rewards/accuracies": 1.0,
581
+ "rewards/generated": -5.6079206466674805,
582
+ "rewards/margins": 15.645317077636719,
583
+ "rewards/real": 10.037394523620605,
584
+ "step": 320
585
+ },
586
+ {
587
+ "epoch": 1.056,
588
+ "grad_norm": 0.39872204323404287,
589
+ "learning_rate": 3.598574821852731e-07,
590
+ "logits/generated": -0.8963452577590942,
591
+ "logits/real": -1.4075753688812256,
592
+ "logps/generated": -350.3981018066406,
593
+ "logps/real": -111.39154052734375,
594
+ "loss": 0.1099,
595
+ "rewards/accuracies": 1.0,
596
+ "rewards/generated": -6.171029090881348,
597
+ "rewards/margins": 16.38304328918457,
598
+ "rewards/real": 10.212014198303223,
599
+ "step": 330
600
+ },
601
+ {
602
+ "epoch": 1.088,
603
+ "grad_norm": 0.3296521457207489,
604
+ "learning_rate": 3.5391923990498813e-07,
605
+ "logits/generated": -0.882519543170929,
606
+ "logits/real": -1.3683688640594482,
607
+ "logps/generated": -344.4193115234375,
608
+ "logps/real": -114.99576568603516,
609
+ "loss": 0.1121,
610
+ "rewards/accuracies": 1.0,
611
+ "rewards/generated": -6.464304447174072,
612
+ "rewards/margins": 16.844728469848633,
613
+ "rewards/real": 10.380424499511719,
614
+ "step": 340
615
+ },
616
+ {
617
+ "epoch": 1.12,
618
+ "grad_norm": 0.40794606558455315,
619
+ "learning_rate": 3.479809976247031e-07,
620
+ "logits/generated": -0.8548433184623718,
621
+ "logits/real": -1.33034086227417,
622
+ "logps/generated": -345.41497802734375,
623
+ "logps/real": -108.29139709472656,
624
+ "loss": 0.1084,
625
+ "rewards/accuracies": 1.0,
626
+ "rewards/generated": -5.70036506652832,
627
+ "rewards/margins": 15.80578899383545,
628
+ "rewards/real": 10.105424880981445,
629
+ "step": 350
630
+ },
631
+ {
632
+ "epoch": 1.152,
633
+ "grad_norm": 0.34179390042713315,
634
+ "learning_rate": 3.42042755344418e-07,
635
+ "logits/generated": -0.8854677081108093,
636
+ "logits/real": -1.3697372674942017,
637
+ "logps/generated": -349.7736511230469,
638
+ "logps/real": -117.81876373291016,
639
+ "loss": 0.1098,
640
+ "rewards/accuracies": 1.0,
641
+ "rewards/generated": -6.514307975769043,
642
+ "rewards/margins": 16.870037078857422,
643
+ "rewards/real": 10.355731010437012,
644
+ "step": 360
645
+ },
646
+ {
647
+ "epoch": 1.184,
648
+ "grad_norm": 0.38985325345599603,
649
+ "learning_rate": 3.36104513064133e-07,
650
+ "logits/generated": -0.8000022172927856,
651
+ "logits/real": -1.3874753713607788,
652
+ "logps/generated": -347.55010986328125,
653
+ "logps/real": -113.49405670166016,
654
+ "loss": 0.1046,
655
+ "rewards/accuracies": 1.0,
656
+ "rewards/generated": -6.464725494384766,
657
+ "rewards/margins": 16.653573989868164,
658
+ "rewards/real": 10.188849449157715,
659
+ "step": 370
660
+ },
661
+ {
662
+ "epoch": 1.1904,
663
+ "eval_logits/generated": -0.8655997514724731,
664
+ "eval_logits/real": -1.3632643222808838,
665
+ "eval_logps/generated": -349.3764953613281,
666
+ "eval_logps/real": -107.03945922851562,
667
+ "eval_loss": 0.10925532132387161,
668
+ "eval_rewards/accuracies": 1.0,
669
+ "eval_rewards/generated": -6.670737266540527,
670
+ "eval_rewards/margins": 16.652626037597656,
671
+ "eval_rewards/real": 9.981886863708496,
672
+ "eval_runtime": 35.169,
673
+ "eval_samples_per_second": 5.687,
674
+ "eval_steps_per_second": 0.37,
675
+ "step": 372
676
+ },
677
+ {
678
+ "epoch": 1.216,
679
+ "grad_norm": 0.3722532991523644,
680
+ "learning_rate": 3.3016627078384796e-07,
681
+ "logits/generated": -0.864538311958313,
682
+ "logits/real": -1.3105075359344482,
683
+ "logps/generated": -335.27362060546875,
684
+ "logps/real": -106.48799133300781,
685
+ "loss": 0.1035,
686
+ "rewards/accuracies": 1.0,
687
+ "rewards/generated": -6.397665977478027,
688
+ "rewards/margins": 16.154638290405273,
689
+ "rewards/real": 9.756973266601562,
690
+ "step": 380
691
+ },
692
+ {
693
+ "epoch": 1.248,
694
+ "grad_norm": 0.4038074067883792,
695
+ "learning_rate": 3.2422802850356297e-07,
696
+ "logits/generated": -0.8852449655532837,
697
+ "logits/real": -1.4460315704345703,
698
+ "logps/generated": -344.63311767578125,
699
+ "logps/real": -106.79902648925781,
700
+ "loss": 0.1027,
701
+ "rewards/accuracies": 1.0,
702
+ "rewards/generated": -5.748571395874023,
703
+ "rewards/margins": 15.521743774414062,
704
+ "rewards/real": 9.773170471191406,
705
+ "step": 390
706
+ },
707
+ {
708
+ "epoch": 1.28,
709
+ "grad_norm": 0.28676972353775737,
710
+ "learning_rate": 3.182897862232779e-07,
711
+ "logits/generated": -0.8121256828308105,
712
+ "logits/real": -1.354479432106018,
713
+ "logps/generated": -345.53326416015625,
714
+ "logps/real": -107.84037017822266,
715
+ "loss": 0.1003,
716
+ "rewards/accuracies": 1.0,
717
+ "rewards/generated": -6.011273384094238,
718
+ "rewards/margins": 15.705144882202148,
719
+ "rewards/real": 9.693869590759277,
720
+ "step": 400
721
+ },
722
+ {
723
+ "epoch": 1.312,
724
+ "grad_norm": 0.3326501083643602,
725
+ "learning_rate": 3.1235154394299283e-07,
726
+ "logits/generated": -0.8699803352355957,
727
+ "logits/real": -1.3926770687103271,
728
+ "logps/generated": -339.2923583984375,
729
+ "logps/real": -113.30909729003906,
730
+ "loss": 0.1069,
731
+ "rewards/accuracies": 1.0,
732
+ "rewards/generated": -5.364248275756836,
733
+ "rewards/margins": 15.17718505859375,
734
+ "rewards/real": 9.81293773651123,
735
+ "step": 410
736
+ },
737
+ {
738
+ "epoch": 1.3439999999999999,
739
+ "grad_norm": 0.41187476133988626,
740
+ "learning_rate": 3.0641330166270784e-07,
741
+ "logits/generated": -0.8451802134513855,
742
+ "logits/real": -1.3541207313537598,
743
+ "logps/generated": -344.1119384765625,
744
+ "logps/real": -99.13960266113281,
745
+ "loss": 0.103,
746
+ "rewards/accuracies": 1.0,
747
+ "rewards/generated": -6.0690155029296875,
748
+ "rewards/margins": 15.549135208129883,
749
+ "rewards/real": 9.480117797851562,
750
+ "step": 420
751
+ },
752
+ {
753
+ "epoch": 1.376,
754
+ "grad_norm": 0.37308896269833713,
755
+ "learning_rate": 3.004750593824228e-07,
756
+ "logits/generated": -0.8354321718215942,
757
+ "logits/real": -1.397528052330017,
758
+ "logps/generated": -354.12841796875,
759
+ "logps/real": -108.86185455322266,
760
+ "loss": 0.1067,
761
+ "rewards/accuracies": 1.0,
762
+ "rewards/generated": -6.957143306732178,
763
+ "rewards/margins": 16.80294418334961,
764
+ "rewards/real": 9.845802307128906,
765
+ "step": 430
766
+ },
767
+ {
768
+ "epoch": 1.3888,
769
+ "eval_logits/generated": -0.881392776966095,
770
+ "eval_logits/real": -1.3897513151168823,
771
+ "eval_logps/generated": -358.4093933105469,
772
+ "eval_logps/real": -106.73082733154297,
773
+ "eval_loss": 0.10892020910978317,
774
+ "eval_rewards/accuracies": 1.0,
775
+ "eval_rewards/generated": -7.574033260345459,
776
+ "eval_rewards/margins": 17.586782455444336,
777
+ "eval_rewards/real": 10.012747764587402,
778
+ "eval_runtime": 25.6933,
779
+ "eval_samples_per_second": 7.784,
780
+ "eval_steps_per_second": 0.506,
781
+ "step": 434
782
+ },
783
+ {
784
+ "epoch": 1.408,
785
+ "grad_norm": 0.35037689567096764,
786
+ "learning_rate": 2.9453681710213776e-07,
787
+ "logits/generated": -0.8212960362434387,
788
+ "logits/real": -1.3137534856796265,
789
+ "logps/generated": -355.6631774902344,
790
+ "logps/real": -114.98287200927734,
791
+ "loss": 0.1095,
792
+ "rewards/accuracies": 1.0,
793
+ "rewards/generated": -6.6356201171875,
794
+ "rewards/margins": 16.64114761352539,
795
+ "rewards/real": 10.00552749633789,
796
+ "step": 440
797
+ },
798
+ {
799
+ "epoch": 1.44,
800
+ "grad_norm": 0.36349550750442733,
801
+ "learning_rate": 2.885985748218527e-07,
802
+ "logits/generated": -0.8298400044441223,
803
+ "logits/real": -1.2980700731277466,
804
+ "logps/generated": -342.68341064453125,
805
+ "logps/real": -89.83566284179688,
806
+ "loss": 0.1046,
807
+ "rewards/accuracies": 1.0,
808
+ "rewards/generated": -6.842597961425781,
809
+ "rewards/margins": 15.478503227233887,
810
+ "rewards/real": 8.635905265808105,
811
+ "step": 450
812
+ },
813
+ {
814
+ "epoch": 1.472,
815
+ "grad_norm": 0.37277388095763225,
816
+ "learning_rate": 2.8266033254156767e-07,
817
+ "logits/generated": -0.891290009021759,
818
+ "logits/real": -1.394723653793335,
819
+ "logps/generated": -341.93218994140625,
820
+ "logps/real": -103.62208557128906,
821
+ "loss": 0.1079,
822
+ "rewards/accuracies": 1.0,
823
+ "rewards/generated": -6.340664863586426,
824
+ "rewards/margins": 15.834001541137695,
825
+ "rewards/real": 9.493334770202637,
826
+ "step": 460
827
+ },
828
+ {
829
+ "epoch": 1.504,
830
+ "grad_norm": 0.3303199018618516,
831
+ "learning_rate": 2.7672209026128263e-07,
832
+ "logits/generated": -0.8582679629325867,
833
+ "logits/real": -1.4615408182144165,
834
+ "logps/generated": -349.00909423828125,
835
+ "logps/real": -104.46484375,
836
+ "loss": 0.0966,
837
+ "rewards/accuracies": 1.0,
838
+ "rewards/generated": -6.171267986297607,
839
+ "rewards/margins": 16.01020622253418,
840
+ "rewards/real": 9.838939666748047,
841
+ "step": 470
842
+ },
843
+ {
844
+ "epoch": 1.536,
845
+ "grad_norm": 0.3813272818464241,
846
+ "learning_rate": 2.7078384798099764e-07,
847
+ "logits/generated": -0.8334212303161621,
848
+ "logits/real": -1.4117028713226318,
849
+ "logps/generated": -349.4712829589844,
850
+ "logps/real": -107.6310806274414,
851
+ "loss": 0.1069,
852
+ "rewards/accuracies": 1.0,
853
+ "rewards/generated": -6.388515472412109,
854
+ "rewards/margins": 16.422515869140625,
855
+ "rewards/real": 10.034002304077148,
856
+ "step": 480
857
+ },
858
+ {
859
+ "epoch": 1.568,
860
+ "grad_norm": 0.3427218965024406,
861
+ "learning_rate": 2.648456057007126e-07,
862
+ "logits/generated": -0.8402020335197449,
863
+ "logits/real": -1.3591063022613525,
864
+ "logps/generated": -349.86529541015625,
865
+ "logps/real": -102.61138916015625,
866
+ "loss": 0.1038,
867
+ "rewards/accuracies": 1.0,
868
+ "rewards/generated": -6.4038848876953125,
869
+ "rewards/margins": 16.043880462646484,
870
+ "rewards/real": 9.639993667602539,
871
+ "step": 490
872
+ },
873
+ {
874
+ "epoch": 1.5872000000000002,
875
+ "eval_logits/generated": -0.8754745125770569,
876
+ "eval_logits/real": -1.3614826202392578,
877
+ "eval_logps/generated": -352.7068786621094,
878
+ "eval_logps/real": -106.12812042236328,
879
+ "eval_loss": 0.10832550376653671,
880
+ "eval_rewards/accuracies": 1.0,
881
+ "eval_rewards/generated": -7.003779888153076,
882
+ "eval_rewards/margins": 17.076799392700195,
883
+ "eval_rewards/real": 10.07302188873291,
884
+ "eval_runtime": 33.3124,
885
+ "eval_samples_per_second": 6.004,
886
+ "eval_steps_per_second": 0.39,
887
+ "step": 496
888
+ },
889
+ {
890
+ "epoch": 1.6,
891
+ "grad_norm": 0.38914741214890197,
892
+ "learning_rate": 2.589073634204275e-07,
893
+ "logits/generated": -0.8448807597160339,
894
+ "logits/real": -1.3226372003555298,
895
+ "logps/generated": -351.1488037109375,
896
+ "logps/real": -107.77713775634766,
897
+ "loss": 0.1085,
898
+ "rewards/accuracies": 1.0,
899
+ "rewards/generated": -6.437747955322266,
900
+ "rewards/margins": 16.31161117553711,
901
+ "rewards/real": 9.87386417388916,
902
+ "step": 500
903
+ },
904
+ {
905
+ "epoch": 1.6320000000000001,
906
+ "grad_norm": 0.3937890384115947,
907
+ "learning_rate": 2.529691211401425e-07,
908
+ "logits/generated": -0.8879863619804382,
909
+ "logits/real": -1.3836486339569092,
910
+ "logps/generated": -340.003662109375,
911
+ "logps/real": -106.28749084472656,
912
+ "loss": 0.1051,
913
+ "rewards/accuracies": 1.0,
914
+ "rewards/generated": -6.393700122833252,
915
+ "rewards/margins": 16.35400390625,
916
+ "rewards/real": 9.960305213928223,
917
+ "step": 510
918
+ },
919
+ {
920
+ "epoch": 1.6640000000000001,
921
+ "grad_norm": 0.38541180062987296,
922
+ "learning_rate": 2.4703087885985747e-07,
923
+ "logits/generated": -0.8842099905014038,
924
+ "logits/real": -1.3454530239105225,
925
+ "logps/generated": -357.91693115234375,
926
+ "logps/real": -115.9624252319336,
927
+ "loss": 0.1108,
928
+ "rewards/accuracies": 1.0,
929
+ "rewards/generated": -7.042295932769775,
930
+ "rewards/margins": 17.503582000732422,
931
+ "rewards/real": 10.461287498474121,
932
+ "step": 520
933
+ },
934
+ {
935
+ "epoch": 1.696,
936
+ "grad_norm": 0.3409231189845364,
937
+ "learning_rate": 2.410926365795724e-07,
938
+ "logits/generated": -0.8585386276245117,
939
+ "logits/real": -1.3780685663223267,
940
+ "logps/generated": -352.55780029296875,
941
+ "logps/real": -104.75419616699219,
942
+ "loss": 0.1091,
943
+ "rewards/accuracies": 1.0,
944
+ "rewards/generated": -6.910445213317871,
945
+ "rewards/margins": 16.95829963684082,
946
+ "rewards/real": 10.047855377197266,
947
+ "step": 530
948
+ },
949
+ {
950
+ "epoch": 1.728,
951
+ "grad_norm": 0.36170229656289943,
952
+ "learning_rate": 2.351543942992874e-07,
953
+ "logits/generated": -0.8263759613037109,
954
+ "logits/real": -1.477992057800293,
955
+ "logps/generated": -342.4795227050781,
956
+ "logps/real": -107.40743255615234,
957
+ "loss": 0.1066,
958
+ "rewards/accuracies": 1.0,
959
+ "rewards/generated": -6.372612953186035,
960
+ "rewards/margins": 16.573341369628906,
961
+ "rewards/real": 10.200728416442871,
962
+ "step": 540
963
+ },
964
+ {
965
+ "epoch": 1.76,
966
+ "grad_norm": 0.3413340610553401,
967
+ "learning_rate": 2.2921615201900234e-07,
968
+ "logits/generated": -0.8356834650039673,
969
+ "logits/real": -1.391550898551941,
970
+ "logps/generated": -350.9759521484375,
971
+ "logps/real": -100.52735137939453,
972
+ "loss": 0.0996,
973
+ "rewards/accuracies": 1.0,
974
+ "rewards/generated": -6.971991539001465,
975
+ "rewards/margins": 16.87733268737793,
976
+ "rewards/real": 9.905343055725098,
977
+ "step": 550
978
+ },
979
+ {
980
+ "epoch": 1.7856,
981
+ "eval_logits/generated": -0.8467180728912354,
982
+ "eval_logits/real": -1.3430825471878052,
983
+ "eval_logps/generated": -352.8455810546875,
984
+ "eval_logps/real": -105.63909149169922,
985
+ "eval_loss": 0.10793981701135635,
986
+ "eval_rewards/accuracies": 1.0,
987
+ "eval_rewards/generated": -7.0176496505737305,
988
+ "eval_rewards/margins": 17.139570236206055,
989
+ "eval_rewards/real": 10.121922492980957,
990
+ "eval_runtime": 26.3578,
991
+ "eval_samples_per_second": 7.588,
992
+ "eval_steps_per_second": 0.493,
993
+ "step": 558
994
+ },
995
+ {
996
+ "epoch": 1.792,
997
+ "grad_norm": 0.3508645377160027,
998
+ "learning_rate": 2.2327790973871732e-07,
999
+ "logits/generated": -0.8774440884590149,
1000
+ "logits/real": -1.3573105335235596,
1001
+ "logps/generated": -351.45697021484375,
1002
+ "logps/real": -119.85599517822266,
1003
+ "loss": 0.1099,
1004
+ "rewards/accuracies": 1.0,
1005
+ "rewards/generated": -6.625948905944824,
1006
+ "rewards/margins": 17.333721160888672,
1007
+ "rewards/real": 10.707772254943848,
1008
+ "step": 560
1009
+ },
1010
+ {
1011
+ "epoch": 1.8239999999999998,
1012
+ "grad_norm": 0.3654629141180253,
1013
+ "learning_rate": 2.173396674584323e-07,
1014
+ "logits/generated": -0.8835099339485168,
1015
+ "logits/real": -1.4046037197113037,
1016
+ "logps/generated": -346.0320739746094,
1017
+ "logps/real": -121.26094818115234,
1018
+ "loss": 0.1059,
1019
+ "rewards/accuracies": 1.0,
1020
+ "rewards/generated": -6.1836042404174805,
1021
+ "rewards/margins": 16.947559356689453,
1022
+ "rewards/real": 10.763954162597656,
1023
+ "step": 570
1024
+ },
1025
+ {
1026
+ "epoch": 1.8559999999999999,
1027
+ "grad_norm": 0.3990780890154353,
1028
+ "learning_rate": 2.1140142517814726e-07,
1029
+ "logits/generated": -0.8519056439399719,
1030
+ "logits/real": -1.4273611307144165,
1031
+ "logps/generated": -358.67572021484375,
1032
+ "logps/real": -100.94701385498047,
1033
+ "loss": 0.102,
1034
+ "rewards/accuracies": 1.0,
1035
+ "rewards/generated": -6.960443019866943,
1036
+ "rewards/margins": 16.782718658447266,
1037
+ "rewards/real": 9.822273254394531,
1038
+ "step": 580
1039
+ },
1040
+ {
1041
+ "epoch": 1.888,
1042
+ "grad_norm": 0.38514946196243705,
1043
+ "learning_rate": 2.0546318289786222e-07,
1044
+ "logits/generated": -0.8117479085922241,
1045
+ "logits/real": -1.3695944547653198,
1046
+ "logps/generated": -354.63836669921875,
1047
+ "logps/real": -92.74488830566406,
1048
+ "loss": 0.0995,
1049
+ "rewards/accuracies": 1.0,
1050
+ "rewards/generated": -7.354495048522949,
1051
+ "rewards/margins": 16.78274154663086,
1052
+ "rewards/real": 9.428245544433594,
1053
+ "step": 590
1054
+ },
1055
+ {
1056
+ "epoch": 1.92,
1057
+ "grad_norm": 0.3924023599162714,
1058
+ "learning_rate": 1.9952494061757718e-07,
1059
+ "logits/generated": -0.8672367930412292,
1060
+ "logits/real": -1.42955482006073,
1061
+ "logps/generated": -358.61041259765625,
1062
+ "logps/real": -108.92237854003906,
1063
+ "loss": 0.1051,
1064
+ "rewards/accuracies": 1.0,
1065
+ "rewards/generated": -7.023502349853516,
1066
+ "rewards/margins": 17.467809677124023,
1067
+ "rewards/real": 10.444307327270508,
1068
+ "step": 600
1069
+ },
1070
+ {
1071
+ "epoch": 1.952,
1072
+ "grad_norm": 0.4306414835630299,
1073
+ "learning_rate": 1.9358669833729216e-07,
1074
+ "logits/generated": -0.8639044761657715,
1075
+ "logits/real": -1.4375250339508057,
1076
+ "logps/generated": -356.2347412109375,
1077
+ "logps/real": -110.20645904541016,
1078
+ "loss": 0.1031,
1079
+ "rewards/accuracies": 1.0,
1080
+ "rewards/generated": -6.728466987609863,
1081
+ "rewards/margins": 17.04363441467285,
1082
+ "rewards/real": 10.315167427062988,
1083
+ "step": 610
1084
+ },
1085
+ {
1086
+ "epoch": 1.984,
1087
+ "grad_norm": 0.3444950108876968,
1088
+ "learning_rate": 1.876484560570071e-07,
1089
+ "logits/generated": -0.9002432823181152,
1090
+ "logits/real": -1.4994289875030518,
1091
+ "logps/generated": -355.31854248046875,
1092
+ "logps/real": -111.2657699584961,
1093
+ "loss": 0.1058,
1094
+ "rewards/accuracies": 1.0,
1095
+ "rewards/generated": -6.687992095947266,
1096
+ "rewards/margins": 16.953479766845703,
1097
+ "rewards/real": 10.265488624572754,
1098
+ "step": 620
1099
+ },
1100
+ {
1101
+ "epoch": 1.984,
1102
+ "eval_logits/generated": -0.8820578455924988,
1103
+ "eval_logits/real": -1.4055497646331787,
1104
+ "eval_logps/generated": -357.4770202636719,
1105
+ "eval_logps/real": -105.37969207763672,
1106
+ "eval_loss": 0.10766133666038513,
1107
+ "eval_rewards/accuracies": 1.0,
1108
+ "eval_rewards/generated": -7.480791091918945,
1109
+ "eval_rewards/margins": 17.628650665283203,
1110
+ "eval_rewards/real": 10.14786148071289,
1111
+ "eval_runtime": 37.0785,
1112
+ "eval_samples_per_second": 5.394,
1113
+ "eval_steps_per_second": 0.351,
1114
+ "step": 620
1115
+ },
1116
+ {
1117
+ "epoch": 2.016,
1118
+ "grad_norm": 0.3780935929863943,
1119
+ "learning_rate": 1.8171021377672207e-07,
1120
+ "logits/generated": -0.8811729550361633,
1121
+ "logits/real": -1.3763405084609985,
1122
+ "logps/generated": -336.79595947265625,
1123
+ "logps/real": -113.61727142333984,
1124
+ "loss": 0.1078,
1125
+ "rewards/accuracies": 1.0,
1126
+ "rewards/generated": -6.133264064788818,
1127
+ "rewards/margins": 16.935718536376953,
1128
+ "rewards/real": 10.802453994750977,
1129
+ "step": 630
1130
+ },
1131
+ {
1132
+ "epoch": 2.048,
1133
+ "grad_norm": 0.3519668839255736,
1134
+ "learning_rate": 1.7577197149643706e-07,
1135
+ "logits/generated": -0.843653678894043,
1136
+ "logits/real": -1.4190059900283813,
1137
+ "logps/generated": -348.73944091796875,
1138
+ "logps/real": -98.12186431884766,
1139
+ "loss": 0.1021,
1140
+ "rewards/accuracies": 1.0,
1141
+ "rewards/generated": -7.104917049407959,
1142
+ "rewards/margins": 16.62804412841797,
1143
+ "rewards/real": 9.523126602172852,
1144
+ "step": 640
1145
+ },
1146
+ {
1147
+ "epoch": 2.08,
1148
+ "grad_norm": 0.3518523871601281,
1149
+ "learning_rate": 1.6983372921615202e-07,
1150
+ "logits/generated": -0.8700889348983765,
1151
+ "logits/real": -1.4218956232070923,
1152
+ "logps/generated": -359.59912109375,
1153
+ "logps/real": -104.86639404296875,
1154
+ "loss": 0.0964,
1155
+ "rewards/accuracies": 1.0,
1156
+ "rewards/generated": -6.900285243988037,
1157
+ "rewards/margins": 17.489273071289062,
1158
+ "rewards/real": 10.588987350463867,
1159
+ "step": 650
1160
+ },
1161
+ {
1162
+ "epoch": 2.112,
1163
+ "grad_norm": 0.33640275650323853,
1164
+ "learning_rate": 1.6389548693586697e-07,
1165
+ "logits/generated": -0.9481474161148071,
1166
+ "logits/real": -1.4541842937469482,
1167
+ "logps/generated": -341.5894470214844,
1168
+ "logps/real": -103.59893798828125,
1169
+ "loss": 0.1018,
1170
+ "rewards/accuracies": 1.0,
1171
+ "rewards/generated": -6.416829586029053,
1172
+ "rewards/margins": 16.72159194946289,
1173
+ "rewards/real": 10.30476188659668,
1174
+ "step": 660
1175
+ },
1176
+ {
1177
+ "epoch": 2.144,
1178
+ "grad_norm": 0.3422223879953069,
1179
+ "learning_rate": 1.5795724465558193e-07,
1180
+ "logits/generated": -0.9395056962966919,
1181
+ "logits/real": -1.5020724534988403,
1182
+ "logps/generated": -336.66485595703125,
1183
+ "logps/real": -115.76351165771484,
1184
+ "loss": 0.0999,
1185
+ "rewards/accuracies": 1.0,
1186
+ "rewards/generated": -6.646719455718994,
1187
+ "rewards/margins": 17.59234619140625,
1188
+ "rewards/real": 10.945624351501465,
1189
+ "step": 670
1190
+ },
1191
+ {
1192
+ "epoch": 2.176,
1193
+ "grad_norm": 0.4031943667703452,
1194
+ "learning_rate": 1.520190023752969e-07,
1195
+ "logits/generated": -0.8811619877815247,
1196
+ "logits/real": -1.4061198234558105,
1197
+ "logps/generated": -358.12457275390625,
1198
+ "logps/real": -102.46736907958984,
1199
+ "loss": 0.0995,
1200
+ "rewards/accuracies": 1.0,
1201
+ "rewards/generated": -7.244171142578125,
1202
+ "rewards/margins": 17.505325317382812,
1203
+ "rewards/real": 10.261154174804688,
1204
+ "step": 680
1205
+ },
1206
+ {
1207
+ "epoch": 2.1824,
1208
+ "eval_logits/generated": -0.8780685067176819,
1209
+ "eval_logits/real": -1.4101550579071045,
1210
+ "eval_logps/generated": -354.6165771484375,
1211
+ "eval_logps/real": -105.18895721435547,
1212
+ "eval_loss": 0.10742899030447006,
1213
+ "eval_rewards/accuracies": 1.0,
1214
+ "eval_rewards/generated": -7.194748878479004,
1215
+ "eval_rewards/margins": 17.36168670654297,
1216
+ "eval_rewards/real": 10.166935920715332,
1217
+ "eval_runtime": 25.4941,
1218
+ "eval_samples_per_second": 7.845,
1219
+ "eval_steps_per_second": 0.51,
1220
+ "step": 682
1221
+ },
1222
+ {
1223
+ "epoch": 2.208,
1224
+ "grad_norm": 0.3658391972106121,
1225
+ "learning_rate": 1.4608076009501184e-07,
1226
+ "logits/generated": -0.7777714133262634,
1227
+ "logits/real": -1.4327569007873535,
1228
+ "logps/generated": -349.45123291015625,
1229
+ "logps/real": -91.0394287109375,
1230
+ "loss": 0.1006,
1231
+ "rewards/accuracies": 1.0,
1232
+ "rewards/generated": -6.516034126281738,
1233
+ "rewards/margins": 16.061874389648438,
1234
+ "rewards/real": 9.5458402633667,
1235
+ "step": 690
1236
+ },
1237
+ {
1238
+ "epoch": 2.24,
1239
+ "grad_norm": 0.37751724197745956,
1240
+ "learning_rate": 1.4014251781472683e-07,
1241
+ "logits/generated": -0.8562400937080383,
1242
+ "logits/real": -1.4605350494384766,
1243
+ "logps/generated": -349.2660827636719,
1244
+ "logps/real": -110.1448745727539,
1245
+ "loss": 0.0985,
1246
+ "rewards/accuracies": 1.0,
1247
+ "rewards/generated": -6.469089508056641,
1248
+ "rewards/margins": 17.073518753051758,
1249
+ "rewards/real": 10.604430198669434,
1250
+ "step": 700
1251
+ },
1252
+ {
1253
+ "epoch": 2.2720000000000002,
1254
+ "grad_norm": 0.3099039440839268,
1255
+ "learning_rate": 1.342042755344418e-07,
1256
+ "logits/generated": -0.9007622003555298,
1257
+ "logits/real": -1.437908411026001,
1258
+ "logps/generated": -345.6163330078125,
1259
+ "logps/real": -93.06938171386719,
1260
+ "loss": 0.0978,
1261
+ "rewards/accuracies": 1.0,
1262
+ "rewards/generated": -6.0835700035095215,
1263
+ "rewards/margins": 15.84107780456543,
1264
+ "rewards/real": 9.757506370544434,
1265
+ "step": 710
1266
+ },
1267
+ {
1268
+ "epoch": 2.304,
1269
+ "grad_norm": 0.3864668793151262,
1270
+ "learning_rate": 1.2826603325415677e-07,
1271
+ "logits/generated": -0.8885849714279175,
1272
+ "logits/real": -1.5355165004730225,
1273
+ "logps/generated": -342.10443115234375,
1274
+ "logps/real": -104.58480072021484,
1275
+ "loss": 0.1019,
1276
+ "rewards/accuracies": 1.0,
1277
+ "rewards/generated": -6.677704811096191,
1278
+ "rewards/margins": 16.409557342529297,
1279
+ "rewards/real": 9.731854438781738,
1280
+ "step": 720
1281
+ },
1282
+ {
1283
+ "epoch": 2.336,
1284
+ "grad_norm": 0.3445053668290169,
1285
+ "learning_rate": 1.2232779097387173e-07,
1286
+ "logits/generated": -0.8385915756225586,
1287
+ "logits/real": -1.397671103477478,
1288
+ "logps/generated": -354.0185241699219,
1289
+ "logps/real": -112.39442443847656,
1290
+ "loss": 0.1017,
1291
+ "rewards/accuracies": 1.0,
1292
+ "rewards/generated": -7.097531795501709,
1293
+ "rewards/margins": 18.07855987548828,
1294
+ "rewards/real": 10.981030464172363,
1295
+ "step": 730
1296
+ },
1297
+ {
1298
+ "epoch": 2.368,
1299
+ "grad_norm": 0.3341439577424557,
1300
+ "learning_rate": 1.163895486935867e-07,
1301
+ "logits/generated": -0.8578903079032898,
1302
+ "logits/real": -1.383490800857544,
1303
+ "logps/generated": -347.2725524902344,
1304
+ "logps/real": -103.35150146484375,
1305
+ "loss": 0.1017,
1306
+ "rewards/accuracies": 1.0,
1307
+ "rewards/generated": -6.981736183166504,
1308
+ "rewards/margins": 17.021286010742188,
1309
+ "rewards/real": 10.039549827575684,
1310
+ "step": 740
1311
+ },
1312
+ {
1313
+ "epoch": 2.3808,
1314
+ "eval_logits/generated": -0.880645751953125,
1315
+ "eval_logits/real": -1.4227505922317505,
1316
+ "eval_logps/generated": -358.9116516113281,
1317
+ "eval_logps/real": -105.00930786132812,
1318
+ "eval_loss": 0.10732075572013855,
1319
+ "eval_rewards/accuracies": 1.0,
1320
+ "eval_rewards/generated": -7.624257564544678,
1321
+ "eval_rewards/margins": 17.80915641784668,
1322
+ "eval_rewards/real": 10.184901237487793,
1323
+ "eval_runtime": 36.3751,
1324
+ "eval_samples_per_second": 5.498,
1325
+ "eval_steps_per_second": 0.357,
1326
+ "step": 744
1327
+ },
1328
+ {
1329
+ "epoch": 2.4,
1330
+ "grad_norm": 0.37872196207505543,
1331
+ "learning_rate": 1.1045130641330165e-07,
1332
+ "logits/generated": -0.9046470522880554,
1333
+ "logits/real": -1.3609508275985718,
1334
+ "logps/generated": -347.01611328125,
1335
+ "logps/real": -113.12504577636719,
1336
+ "loss": 0.1048,
1337
+ "rewards/accuracies": 1.0,
1338
+ "rewards/generated": -6.732972621917725,
1339
+ "rewards/margins": 17.482839584350586,
1340
+ "rewards/real": 10.74986457824707,
1341
+ "step": 750
1342
+ },
1343
+ {
1344
+ "epoch": 2.432,
1345
+ "grad_norm": 0.3456217830065947,
1346
+ "learning_rate": 1.0451306413301662e-07,
1347
+ "logits/generated": -0.9419130086898804,
1348
+ "logits/real": -1.448075294494629,
1349
+ "logps/generated": -350.6387023925781,
1350
+ "logps/real": -104.91294860839844,
1351
+ "loss": 0.1024,
1352
+ "rewards/accuracies": 1.0,
1353
+ "rewards/generated": -6.793516635894775,
1354
+ "rewards/margins": 17.103900909423828,
1355
+ "rewards/real": 10.310384750366211,
1356
+ "step": 760
1357
+ },
1358
+ {
1359
+ "epoch": 2.464,
1360
+ "grad_norm": 0.3510621706063929,
1361
+ "learning_rate": 9.857482185273158e-08,
1362
+ "logits/generated": -0.8172124028205872,
1363
+ "logits/real": -1.428694486618042,
1364
+ "logps/generated": -363.5953674316406,
1365
+ "logps/real": -88.23005676269531,
1366
+ "loss": 0.1012,
1367
+ "rewards/accuracies": 1.0,
1368
+ "rewards/generated": -7.76650857925415,
1369
+ "rewards/margins": 17.137920379638672,
1370
+ "rewards/real": 9.371413230895996,
1371
+ "step": 770
1372
+ },
1373
+ {
1374
+ "epoch": 2.496,
1375
+ "grad_norm": 0.29361560856209884,
1376
+ "learning_rate": 9.263657957244655e-08,
1377
+ "logits/generated": -0.8288412094116211,
1378
+ "logits/real": -1.4823163747787476,
1379
+ "logps/generated": -359.7073974609375,
1380
+ "logps/real": -97.23246002197266,
1381
+ "loss": 0.1021,
1382
+ "rewards/accuracies": 1.0,
1383
+ "rewards/generated": -7.1570844650268555,
1384
+ "rewards/margins": 17.029443740844727,
1385
+ "rewards/real": 9.872357368469238,
1386
+ "step": 780
1387
+ },
1388
+ {
1389
+ "epoch": 2.528,
1390
+ "grad_norm": 0.3698992846301525,
1391
+ "learning_rate": 8.669833729216151e-08,
1392
+ "logits/generated": -0.834811806678772,
1393
+ "logits/real": -1.4748550653457642,
1394
+ "logps/generated": -355.08685302734375,
1395
+ "logps/real": -91.9556884765625,
1396
+ "loss": 0.1018,
1397
+ "rewards/accuracies": 1.0,
1398
+ "rewards/generated": -7.218266487121582,
1399
+ "rewards/margins": 17.2064208984375,
1400
+ "rewards/real": 9.988153457641602,
1401
+ "step": 790
1402
+ },
1403
+ {
1404
+ "epoch": 2.56,
1405
+ "grad_norm": 0.3614102743036081,
1406
+ "learning_rate": 8.076009501187649e-08,
1407
+ "logits/generated": -0.8415037989616394,
1408
+ "logits/real": -1.4521253108978271,
1409
+ "logps/generated": -342.2822265625,
1410
+ "logps/real": -102.14026641845703,
1411
+ "loss": 0.1031,
1412
+ "rewards/accuracies": 1.0,
1413
+ "rewards/generated": -7.008787631988525,
1414
+ "rewards/margins": 16.703975677490234,
1415
+ "rewards/real": 9.695188522338867,
1416
+ "step": 800
1417
+ },
1418
+ {
1419
+ "epoch": 2.5792,
1420
+ "eval_logits/generated": -0.878677248954773,
1421
+ "eval_logits/real": -1.4391472339630127,
1422
+ "eval_logps/generated": -359.2500305175781,
1423
+ "eval_logps/real": -104.75186920166016,
1424
+ "eval_loss": 0.1071779727935791,
1425
+ "eval_rewards/accuracies": 1.0,
1426
+ "eval_rewards/generated": -7.65809440612793,
1427
+ "eval_rewards/margins": 17.868741989135742,
1428
+ "eval_rewards/real": 10.210646629333496,
1429
+ "eval_runtime": 29.5912,
1430
+ "eval_samples_per_second": 6.759,
1431
+ "eval_steps_per_second": 0.439,
1432
+ "step": 806
1433
+ },
1434
+ {
1435
+ "epoch": 2.592,
1436
+ "grad_norm": 0.3510246600808513,
1437
+ "learning_rate": 7.482185273159145e-08,
1438
+ "logits/generated": -0.9100650548934937,
1439
+ "logits/real": -1.4608324766159058,
1440
+ "logps/generated": -355.4903259277344,
1441
+ "logps/real": -110.2839584350586,
1442
+ "loss": 0.1063,
1443
+ "rewards/accuracies": 1.0,
1444
+ "rewards/generated": -7.249466896057129,
1445
+ "rewards/margins": 18.285369873046875,
1446
+ "rewards/real": 11.035902976989746,
1447
+ "step": 810
1448
+ },
1449
+ {
1450
+ "epoch": 2.624,
1451
+ "grad_norm": 0.37436996561336855,
1452
+ "learning_rate": 6.88836104513064e-08,
1453
+ "logits/generated": -0.8607547879219055,
1454
+ "logits/real": -1.4581551551818848,
1455
+ "logps/generated": -353.0670471191406,
1456
+ "logps/real": -93.31352233886719,
1457
+ "loss": 0.1019,
1458
+ "rewards/accuracies": 1.0,
1459
+ "rewards/generated": -6.994053840637207,
1460
+ "rewards/margins": 16.841081619262695,
1461
+ "rewards/real": 9.847024917602539,
1462
+ "step": 820
1463
+ },
1464
+ {
1465
+ "epoch": 2.656,
1466
+ "grad_norm": 0.33372442568098337,
1467
+ "learning_rate": 6.294536817102138e-08,
1468
+ "logits/generated": -0.932579517364502,
1469
+ "logits/real": -1.4342137575149536,
1470
+ "logps/generated": -356.23455810546875,
1471
+ "logps/real": -117.7547836303711,
1472
+ "loss": 0.1053,
1473
+ "rewards/accuracies": 1.0,
1474
+ "rewards/generated": -7.367926120758057,
1475
+ "rewards/margins": 18.793277740478516,
1476
+ "rewards/real": 11.42535400390625,
1477
+ "step": 830
1478
+ },
1479
+ {
1480
+ "epoch": 2.6879999999999997,
1481
+ "grad_norm": 0.31963742826245944,
1482
+ "learning_rate": 5.700712589073634e-08,
1483
+ "logits/generated": -0.866934597492218,
1484
+ "logits/real": -1.499237060546875,
1485
+ "logps/generated": -363.590576171875,
1486
+ "logps/real": -108.01591491699219,
1487
+ "loss": 0.1013,
1488
+ "rewards/accuracies": 1.0,
1489
+ "rewards/generated": -7.553430080413818,
1490
+ "rewards/margins": 18.140201568603516,
1491
+ "rewards/real": 10.586769104003906,
1492
+ "step": 840
1493
+ },
1494
+ {
1495
+ "epoch": 2.7199999999999998,
1496
+ "grad_norm": 0.34388141431724034,
1497
+ "learning_rate": 5.10688836104513e-08,
1498
+ "logits/generated": -0.8293858766555786,
1499
+ "logits/real": -1.378164291381836,
1500
+ "logps/generated": -356.0320129394531,
1501
+ "logps/real": -100.18760681152344,
1502
+ "loss": 0.0992,
1503
+ "rewards/accuracies": 1.0,
1504
+ "rewards/generated": -6.919466495513916,
1505
+ "rewards/margins": 16.961532592773438,
1506
+ "rewards/real": 10.04206657409668,
1507
+ "step": 850
1508
+ },
1509
+ {
1510
+ "epoch": 2.752,
1511
+ "grad_norm": 0.3665611604871145,
1512
+ "learning_rate": 4.5130641330166267e-08,
1513
+ "logits/generated": -0.9078382253646851,
1514
+ "logits/real": -1.4398049116134644,
1515
+ "logps/generated": -349.5274963378906,
1516
+ "logps/real": -110.4783706665039,
1517
+ "loss": 0.1025,
1518
+ "rewards/accuracies": 1.0,
1519
+ "rewards/generated": -7.484776973724365,
1520
+ "rewards/margins": 18.162715911865234,
1521
+ "rewards/real": 10.677938461303711,
1522
+ "step": 860
1523
+ },
1524
+ {
1525
+ "epoch": 2.7776,
1526
+ "eval_logits/generated": -0.8824076056480408,
1527
+ "eval_logits/real": -1.4506325721740723,
1528
+ "eval_logps/generated": -359.47296142578125,
1529
+ "eval_logps/real": -104.75336456298828,
1530
+ "eval_loss": 0.10711534321308136,
1531
+ "eval_rewards/accuracies": 1.0,
1532
+ "eval_rewards/generated": -7.680384635925293,
1533
+ "eval_rewards/margins": 17.890880584716797,
1534
+ "eval_rewards/real": 10.210494041442871,
1535
+ "eval_runtime": 35.1692,
1536
+ "eval_samples_per_second": 5.687,
1537
+ "eval_steps_per_second": 0.37,
1538
+ "step": 868
1539
+ },
1540
+ {
1541
+ "epoch": 2.784,
1542
+ "grad_norm": 0.3504947718447569,
1543
+ "learning_rate": 3.919239904988123e-08,
1544
+ "logits/generated": -0.8932768702507019,
1545
+ "logits/real": -1.496914267539978,
1546
+ "logps/generated": -356.4810791015625,
1547
+ "logps/real": -118.42008209228516,
1548
+ "loss": 0.1092,
1549
+ "rewards/accuracies": 1.0,
1550
+ "rewards/generated": -7.157601356506348,
1551
+ "rewards/margins": 18.124359130859375,
1552
+ "rewards/real": 10.966758728027344,
1553
+ "step": 870
1554
+ },
1555
+ {
1556
+ "epoch": 2.816,
1557
+ "grad_norm": 0.3398952469610853,
1558
+ "learning_rate": 3.32541567695962e-08,
1559
+ "logits/generated": -0.8756890296936035,
1560
+ "logits/real": -1.4591081142425537,
1561
+ "logps/generated": -350.1907958984375,
1562
+ "logps/real": -116.3914566040039,
1563
+ "loss": 0.1061,
1564
+ "rewards/accuracies": 1.0,
1565
+ "rewards/generated": -6.856751918792725,
1566
+ "rewards/margins": 18.386884689331055,
1567
+ "rewards/real": 11.530131340026855,
1568
+ "step": 880
1569
+ },
1570
+ {
1571
+ "epoch": 2.848,
1572
+ "grad_norm": 0.3405981200610349,
1573
+ "learning_rate": 2.7315914489311164e-08,
1574
+ "logits/generated": -0.884991466999054,
1575
+ "logits/real": -1.523085355758667,
1576
+ "logps/generated": -345.1888732910156,
1577
+ "logps/real": -108.27435302734375,
1578
+ "loss": 0.1036,
1579
+ "rewards/accuracies": 1.0,
1580
+ "rewards/generated": -6.91180419921875,
1581
+ "rewards/margins": 17.881935119628906,
1582
+ "rewards/real": 10.97012996673584,
1583
+ "step": 890
1584
+ },
1585
+ {
1586
+ "epoch": 2.88,
1587
+ "grad_norm": 0.35128059946197576,
1588
+ "learning_rate": 2.1377672209026125e-08,
1589
+ "logits/generated": -0.8905202150344849,
1590
+ "logits/real": -1.5026408433914185,
1591
+ "logps/generated": -359.63897705078125,
1592
+ "logps/real": -102.87252044677734,
1593
+ "loss": 0.0983,
1594
+ "rewards/accuracies": 1.0,
1595
+ "rewards/generated": -6.926169395446777,
1596
+ "rewards/margins": 17.284862518310547,
1597
+ "rewards/real": 10.358692169189453,
1598
+ "step": 900
1599
+ },
1600
+ {
1601
+ "epoch": 2.912,
1602
+ "grad_norm": 0.435980758550885,
1603
+ "learning_rate": 1.5439429928741092e-08,
1604
+ "logits/generated": -0.9093903303146362,
1605
+ "logits/real": -1.4204092025756836,
1606
+ "logps/generated": -344.41156005859375,
1607
+ "logps/real": -117.05790710449219,
1608
+ "loss": 0.1076,
1609
+ "rewards/accuracies": 1.0,
1610
+ "rewards/generated": -6.70886754989624,
1611
+ "rewards/margins": 17.578311920166016,
1612
+ "rewards/real": 10.869443893432617,
1613
+ "step": 910
1614
+ },
1615
+ {
1616
+ "epoch": 2.944,
1617
+ "grad_norm": 0.3248143296769044,
1618
+ "learning_rate": 9.501187648456057e-09,
1619
+ "logits/generated": -0.8297950029373169,
1620
+ "logits/real": -1.4422714710235596,
1621
+ "logps/generated": -344.2285461425781,
1622
+ "logps/real": -97.16529846191406,
1623
+ "loss": 0.0982,
1624
+ "rewards/accuracies": 1.0,
1625
+ "rewards/generated": -6.858038425445557,
1626
+ "rewards/margins": 16.81562614440918,
1627
+ "rewards/real": 9.957587242126465,
1628
+ "step": 920
1629
+ },
1630
+ {
1631
+ "epoch": 2.976,
1632
+ "grad_norm": 0.6321286392468793,
1633
+ "learning_rate": 3.562945368171021e-09,
1634
+ "logits/generated": -0.9097634553909302,
1635
+ "logits/real": -1.5169529914855957,
1636
+ "logps/generated": -343.1878967285156,
1637
+ "logps/real": -112.12968444824219,
1638
+ "loss": 0.1067,
1639
+ "rewards/accuracies": 1.0,
1640
+ "rewards/generated": -6.950539588928223,
1641
+ "rewards/margins": 17.87512969970703,
1642
+ "rewards/real": 10.924592971801758,
1643
+ "step": 930
1644
+ },
1645
+ {
1646
+ "epoch": 2.976,
1647
+ "eval_logits/generated": -0.8780982494354248,
1648
+ "eval_logits/real": -1.4493974447250366,
1649
+ "eval_logps/generated": -358.9117431640625,
1650
+ "eval_logps/real": -104.68750762939453,
1651
+ "eval_loss": 0.10710752755403519,
1652
+ "eval_rewards/accuracies": 1.0,
1653
+ "eval_rewards/generated": -7.624265193939209,
1654
+ "eval_rewards/margins": 17.841346740722656,
1655
+ "eval_rewards/real": 10.217081069946289,
1656
+ "eval_runtime": 25.8818,
1657
+ "eval_samples_per_second": 7.727,
1658
+ "eval_steps_per_second": 0.502,
1659
+ "step": 930
1660
+ },
1661
+ {
1662
+ "epoch": 2.9952,
1663
+ "step": 936,
1664
+ "total_flos": 0.0,
1665
+ "train_loss": 0.1350623204285263,
1666
+ "train_runtime": 11872.5465,
1667
+ "train_samples_per_second": 2.524,
1668
+ "train_steps_per_second": 0.079
1669
+ }
1670
+ ],
1671
+ "logging_steps": 10,
1672
+ "max_steps": 936,
1673
+ "num_input_tokens_seen": 0,
1674
+ "num_train_epochs": 3,
1675
+ "save_steps": 62,
1676
+ "stateful_callbacks": {
1677
+ "TrainerControl": {
1678
+ "args": {
1679
+ "should_epoch_stop": false,
1680
+ "should_evaluate": false,
1681
+ "should_log": false,
1682
+ "should_save": true,
1683
+ "should_training_stop": true
1684
+ },
1685
+ "attributes": {}
1686
+ }
1687
+ },
1688
+ "total_flos": 0.0,
1689
+ "train_batch_size": 4,
1690
+ "trial_name": null,
1691
+ "trial_params": null
1692
+ }