hZzy commited on
Commit
d13ca91
1 Parent(s): 104fc11

Model save

Browse files
README.md ADDED
@@ -0,0 +1,95 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: apache-2.0
3
+ base_model: hZzy/qwen2.5-0.5b-sft-news-IFT
4
+ tags:
5
+ - trl
6
+ - expo
7
+ - generated_from_trainer
8
+ model-index:
9
+ - name: qwen2.5-0.5b-expo-L2EXPO-ES-100
10
+ results: []
11
+ ---
12
+
13
+ <!-- This model card has been generated automatically according to the information the Trainer had access to. You
14
+ should probably proofread and complete it, then remove this comment. -->
15
+
16
+ [<img src="https://raw.githubusercontent.com/wandb/assets/main/wandb-github-badge-28.svg" alt="Visualize in Weights & Biases" width="200" height="32"/>](https://wandb.ai/zhiyuzha-university-of-florida/huggingface/runs/0805exae)
17
+ # qwen2.5-0.5b-expo-L2EXPO-ES-100
18
+
19
+ This model is a fine-tuned version of [hZzy/qwen2.5-0.5b-sft-news-IFT](https://huggingface.co/hZzy/qwen2.5-0.5b-sft-news-IFT) on an unknown dataset.
20
+ It achieves the following results on the evaluation set:
21
+ - Loss: 519.2467
22
+ - Logps: -84.2676
23
+ - Logits: -0.4953
24
+ - Objective: 521.8153
25
+ - Dpo Loss: 266.9453
26
+ - Regularize: 521.8153
27
+ - Ranking Simple: 0.5264
28
+ - Ranking Idealized: 0.5212
29
+ - Ranking Idealized Expo: 0.5212
30
+ - Wo Beta: 14.2577
31
+
32
+ ## Model description
33
+
34
+ More information needed
35
+
36
+ ## Intended uses & limitations
37
+
38
+ More information needed
39
+
40
+ ## Training and evaluation data
41
+
42
+ More information needed
43
+
44
+ ## Training procedure
45
+
46
+ ### Training hyperparameters
47
+
48
+ The following hyperparameters were used during training:
49
+ - learning_rate: 5e-06
50
+ - train_batch_size: 4
51
+ - eval_batch_size: 4
52
+ - seed: 42
53
+ - distributed_type: multi-GPU
54
+ - num_devices: 3
55
+ - gradient_accumulation_steps: 12
56
+ - total_train_batch_size: 144
57
+ - total_eval_batch_size: 12
58
+ - optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
59
+ - lr_scheduler_type: cosine
60
+ - lr_scheduler_warmup_ratio: 0.1
61
+ - num_epochs: 5
62
+
63
+ ### Training results
64
+
65
+ | Training Loss | Epoch | Step | Dpo Loss | Logits | Logps | Validation Loss | Objective | Ranking Idealized | Ranking Idealized Expo | Ranking Simple | Regularize | Wo Beta |
66
+ |:-------------:|:------:|:----:|:--------:|:-------:|:--------:|:---------------:|:---------:|:-----------------:|:----------------------:|:--------------:|:----------:|:-------:|
67
+ | 43.2587 | 0.1417 | 50 | 26.4475 | -1.4448 | -90.5292 | 52.6622 | 53.6977 | 0.5212 | 0.5212 | 0.5264 | 53.6977 | 16.1700 |
68
+ | 169.8852 | 0.2834 | 100 | 85.7639 | -1.3621 | -85.2787 | 173.9861 | 172.1891 | 0.5212 | 0.5212 | 0.5243 | 172.1891 | 15.4391 |
69
+ | 285.0432 | 0.4251 | 150 | 143.0300 | -1.1694 | -83.2181 | 291.4834 | 293.4404 | 0.5212 | 0.5212 | 0.5280 | 293.4404 | 15.2225 |
70
+ | 355.4066 | 0.5668 | 200 | 189.8469 | -0.9274 | -84.0320 | 372.7906 | 365.2124 | 0.5212 | 0.5212 | 0.5233 | 365.2124 | 14.8684 |
71
+ | 368.9811 | 0.7085 | 250 | 216.4584 | -0.7746 | -81.5050 | 446.6966 | 442.3321 | 0.5212 | 0.5212 | 0.5259 | 442.3321 | 14.4790 |
72
+ | 360.5868 | 0.8503 | 300 | 222.8840 | -0.5984 | -82.2011 | 448.9506 | 443.9051 | 0.5212 | 0.5212 | 0.5248 | 443.9051 | 14.3930 |
73
+ | 338.3987 | 0.9920 | 350 | 232.9365 | -0.7855 | -84.1638 | 462.1923 | 461.2073 | 0.5212 | 0.5212 | 0.5269 | 461.2073 | 14.2979 |
74
+ | 309.1712 | 1.1337 | 400 | 248.0718 | -0.6414 | -82.4934 | 480.5965 | 478.7404 | 0.5212 | 0.5212 | 0.5254 | 478.7404 | 14.3872 |
75
+ | 298.1424 | 1.2754 | 450 | 247.8722 | -0.7014 | -82.1465 | 480.3256 | 482.1766 | 0.5212 | 0.5212 | 0.5238 | 482.1766 | 14.3695 |
76
+ | 282.4504 | 1.4171 | 500 | 252.2093 | -0.4578 | -83.4101 | 493.7484 | 495.7639 | 0.5212 | 0.5212 | 0.5248 | 495.7639 | 14.1743 |
77
+ | 261.1027 | 1.5588 | 550 | 245.8756 | -0.5435 | -82.8268 | 486.1626 | 489.7928 | 0.5212 | 0.5212 | 0.5254 | 489.7928 | 14.0464 |
78
+ | 255.9288 | 1.7005 | 600 | 251.2934 | -0.5347 | -82.1768 | 500.3801 | 502.1727 | 0.5212 | 0.5212 | 0.5269 | 502.1727 | 14.2436 |
79
+ | 248.6787 | 1.8422 | 650 | 254.5959 | -0.5140 | -81.4923 | 502.3153 | 504.1582 | 0.5212 | 0.5212 | 0.5248 | 504.1582 | 14.3320 |
80
+ | 226.4676 | 1.9839 | 700 | 264.1660 | -0.4816 | -83.4216 | 512.6990 | 516.7103 | 0.5212 | 0.5212 | 0.5254 | 516.7103 | 14.0834 |
81
+ | 207.1551 | 2.1256 | 750 | 259.2528 | -0.5410 | -83.4589 | 506.4237 | 510.6129 | 0.5212 | 0.5212 | 0.5238 | 510.6129 | 14.1295 |
82
+ | 197.3545 | 2.2674 | 800 | 262.3102 | -0.5659 | -84.8747 | 513.3979 | 514.3120 | 0.5212 | 0.5212 | 0.5228 | 514.3120 | 14.0704 |
83
+ | 182.3796 | 2.4138 | 850 | 501.8831 | -82.8624| -0.5510 | 504.8523 | 254.1251 | 504.8523 | 0.5274 | 0.5212 | 0.5212 | 14.1707 |
84
+ | 176.042 | 2.5555 | 900 | 518.1983 | -85.0710| -0.5039 | 519.5008 | 263.2800 | 519.5008 | 0.5238 | 0.5212 | 0.5212 | 14.1123 |
85
+ | 164.8281 | 2.6972 | 950 | 512.1844 | -84.5843| -0.5200 | 512.7651 | 262.8074 | 512.7651 | 0.5238 | 0.5212 | 0.5212 | 14.1643 |
86
+ | 150.0401 | 2.8389 | 1000 | 514.7036 | -83.7343| -0.5219 | 516.5959 | 263.6169 | 516.5959 | 0.5259 | 0.5212 | 0.5212 | 14.1800 |
87
+ | 141.0317 | 2.9806 | 1050 | 519.2467 | -84.2676| -0.4953 | 521.8153 | 266.9453 | 521.8153 | 0.5264 | 0.5212 | 0.5212 | 14.2577 |
88
+
89
+
90
+ ### Framework versions
91
+
92
+ - Transformers 4.42.0
93
+ - Pytorch 2.3.0+cu121
94
+ - Datasets 2.19.1
95
+ - Tokenizers 0.19.1
all_results.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 2.9806329711856403,
3
+ "total_flos": 0.0,
4
+ "train_loss": 38.77721726190476,
5
+ "train_runtime": 6898.2426,
6
+ "train_samples": 50802,
7
+ "train_samples_per_second": 36.822,
8
+ "train_steps_per_second": 0.255
9
+ }
generation_config.json ADDED
@@ -0,0 +1,7 @@
 
 
 
 
 
 
 
 
1
+ {
2
+ "bos_token_id": 151644,
3
+ "eos_token_id": 151645,
4
+ "max_new_tokens": 2048,
5
+ "pad_token_id": 151645,
6
+ "transformers_version": "4.42.0"
7
+ }
model.safetensors CHANGED
@@ -1,3 +1,3 @@
1
  version https://git-lfs.github.com/spec/v1
2
- oid sha256:cc5c60252d5f89780ebc287ccb080c72743df145932e1cda9dc79512779ee23e
3
  size 1975192208
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:35e493a47a540062e6caa5b1b8c62e936b32746f4c2eb36a903fb2b880481854
3
  size 1975192208
train_results.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 2.9806329711856403,
3
+ "total_flos": 0.0,
4
+ "train_loss": 38.77721726190476,
5
+ "train_runtime": 6898.2426,
6
+ "train_samples": 50802,
7
+ "train_samples_per_second": 36.822,
8
+ "train_steps_per_second": 0.255
9
+ }
trainer_state.json ADDED
@@ -0,0 +1,760 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 14.046432495117188,
3
+ "best_model_checkpoint": "./qwen2.5-0.5b/qwen2.5-0.5b-expo-L2EXPO-ES-100/checkpoint-550",
4
+ "epoch": 2.9806329711856403,
5
+ "eval_steps": 50,
6
+ "global_step": 1050,
7
+ "is_hyper_param_search": false,
8
+ "is_local_process_zero": true,
9
+ "is_world_process_zero": true,
10
+ "log_history": [
11
+ {
12
+ "dpo_loss": 0.6931471824645996,
13
+ "epoch": 0.002834199338686821,
14
+ "grad_norm": 36884.87916049903,
15
+ "learning_rate": 2.840909090909091e-08,
16
+ "logits": -1.359458565711975,
17
+ "logps": -84.69721221923828,
18
+ "loss": 0.3913,
19
+ "objective": 0.3618059456348419,
20
+ "ranking_idealized": 0.5833333134651184,
21
+ "ranking_idealized_expo": 0.5833333134651184,
22
+ "ranking_simple": 0.5833333134651184,
23
+ "regularize": 0.3618059456348419,
24
+ "step": 1,
25
+ "wo_beta": 14.830931663513184
26
+ },
27
+ {
28
+ "dpo_loss": 20.195844650268555,
29
+ "epoch": 0.14170996693434104,
30
+ "grad_norm": 35665.35173471636,
31
+ "learning_rate": 1.4204545454545458e-06,
32
+ "logits": -1.4575351476669312,
33
+ "logps": -84.27513122558594,
34
+ "loss": 43.2587,
35
+ "objective": 41.916500091552734,
36
+ "ranking_idealized": 0.5221088528633118,
37
+ "ranking_idealized_expo": 0.5216836929321289,
38
+ "ranking_simple": 0.5216836929321289,
39
+ "regularize": 41.916500091552734,
40
+ "step": 50,
41
+ "wo_beta": 15.635692596435547
42
+ },
43
+ {
44
+ "epoch": 0.14170996693434104,
45
+ "eval_dpo_loss": 26.447525024414062,
46
+ "eval_logits": -1.4447709321975708,
47
+ "eval_logps": -90.52921295166016,
48
+ "eval_loss": 52.66217041015625,
49
+ "eval_objective": 53.697696685791016,
50
+ "eval_ranking_idealized": 0.5212215185165405,
51
+ "eval_ranking_idealized_expo": 0.5212215185165405,
52
+ "eval_ranking_simple": 0.5263975262641907,
53
+ "eval_regularize": 53.697696685791016,
54
+ "eval_runtime": 309.407,
55
+ "eval_samples_per_second": 18.713,
56
+ "eval_steps_per_second": 1.561,
57
+ "eval_wo_beta": 16.170011520385742,
58
+ "step": 50
59
+ },
60
+ {
61
+ "dpo_loss": 89.97029876708984,
62
+ "epoch": 0.2834199338686821,
63
+ "grad_norm": 28235.60144716246,
64
+ "learning_rate": 2.8409090909090916e-06,
65
+ "logits": -1.398974061012268,
66
+ "logps": -82.89569091796875,
67
+ "loss": 169.8852,
68
+ "objective": 172.36553955078125,
69
+ "ranking_idealized": 0.5137500166893005,
70
+ "ranking_idealized_expo": 0.5137500166893005,
71
+ "ranking_simple": 0.51583331823349,
72
+ "regularize": 172.36553955078125,
73
+ "step": 100,
74
+ "wo_beta": 15.30754566192627
75
+ },
76
+ {
77
+ "epoch": 0.2834199338686821,
78
+ "eval_dpo_loss": 85.763916015625,
79
+ "eval_logits": -1.3620884418487549,
80
+ "eval_logps": -85.2786636352539,
81
+ "eval_loss": 173.986083984375,
82
+ "eval_objective": 172.1890869140625,
83
+ "eval_ranking_idealized": 0.5212215185165405,
84
+ "eval_ranking_idealized_expo": 0.5212215185165405,
85
+ "eval_ranking_simple": 0.5243270993232727,
86
+ "eval_regularize": 172.1890869140625,
87
+ "eval_runtime": 308.6643,
88
+ "eval_samples_per_second": 18.758,
89
+ "eval_steps_per_second": 1.565,
90
+ "eval_wo_beta": 15.439105033874512,
91
+ "step": 100
92
+ },
93
+ {
94
+ "dpo_loss": 149.43614196777344,
95
+ "epoch": 0.42512990080302315,
96
+ "grad_norm": 21173.096618846714,
97
+ "learning_rate": 4.2613636363636365e-06,
98
+ "logits": -1.213483214378357,
99
+ "logps": -78.55652618408203,
100
+ "loss": 285.0432,
101
+ "objective": 280.7228088378906,
102
+ "ranking_idealized": 0.527916669845581,
103
+ "ranking_idealized_expo": 0.527916669845581,
104
+ "ranking_simple": 0.5229166746139526,
105
+ "regularize": 280.7228088378906,
106
+ "step": 150,
107
+ "wo_beta": 15.080223083496094
108
+ },
109
+ {
110
+ "epoch": 0.42512990080302315,
111
+ "eval_dpo_loss": 143.02996826171875,
112
+ "eval_logits": -1.1693531274795532,
113
+ "eval_logps": -83.2181167602539,
114
+ "eval_loss": 291.4833679199219,
115
+ "eval_objective": 293.4403991699219,
116
+ "eval_ranking_idealized": 0.5212215185165405,
117
+ "eval_ranking_idealized_expo": 0.5212215185165405,
118
+ "eval_ranking_simple": 0.5279502868652344,
119
+ "eval_regularize": 293.4403991699219,
120
+ "eval_runtime": 311.5234,
121
+ "eval_samples_per_second": 18.586,
122
+ "eval_steps_per_second": 1.55,
123
+ "eval_wo_beta": 15.222454071044922,
124
+ "step": 150
125
+ },
126
+ {
127
+ "dpo_loss": 181.66571044921875,
128
+ "epoch": 0.5668398677373642,
129
+ "grad_norm": 20619.52737873687,
130
+ "learning_rate": 4.997168347957521e-06,
131
+ "logits": -0.9392554759979248,
132
+ "logps": -78.61503601074219,
133
+ "loss": 355.4066,
134
+ "objective": 356.2313232421875,
135
+ "ranking_idealized": 0.51541668176651,
136
+ "ranking_idealized_expo": 0.51541668176651,
137
+ "ranking_simple": 0.5191666483879089,
138
+ "regularize": 356.2313232421875,
139
+ "step": 200,
140
+ "wo_beta": 15.349950790405273
141
+ },
142
+ {
143
+ "epoch": 0.5668398677373642,
144
+ "eval_dpo_loss": 189.846923828125,
145
+ "eval_logits": -0.9273601174354553,
146
+ "eval_logps": -84.03199768066406,
147
+ "eval_loss": 372.7905578613281,
148
+ "eval_objective": 365.21240234375,
149
+ "eval_ranking_idealized": 0.5212215185165405,
150
+ "eval_ranking_idealized_expo": 0.5212215185165405,
151
+ "eval_ranking_simple": 0.5232919454574585,
152
+ "eval_regularize": 365.21240234375,
153
+ "eval_runtime": 308.8911,
154
+ "eval_samples_per_second": 18.744,
155
+ "eval_steps_per_second": 1.564,
156
+ "eval_wo_beta": 14.868427276611328,
157
+ "step": 200
158
+ },
159
+ {
160
+ "dpo_loss": 186.671142578125,
161
+ "epoch": 0.7085498346717053,
162
+ "grad_norm": 17849.874243754268,
163
+ "learning_rate": 4.973122855144066e-06,
164
+ "logits": -0.822012186050415,
165
+ "logps": -76.92431640625,
166
+ "loss": 368.9811,
167
+ "objective": 366.2552490234375,
168
+ "ranking_idealized": 0.5162500143051147,
169
+ "ranking_idealized_expo": 0.5162500143051147,
170
+ "ranking_simple": 0.5074999928474426,
171
+ "regularize": 366.2552490234375,
172
+ "step": 250,
173
+ "wo_beta": 15.5939359664917
174
+ },
175
+ {
176
+ "epoch": 0.7085498346717053,
177
+ "eval_dpo_loss": 216.4584197998047,
178
+ "eval_logits": -0.7745867967605591,
179
+ "eval_logps": -81.50496673583984,
180
+ "eval_loss": 446.6966247558594,
181
+ "eval_objective": 442.3320617675781,
182
+ "eval_ranking_idealized": 0.5212215185165405,
183
+ "eval_ranking_idealized_expo": 0.5212215185165405,
184
+ "eval_ranking_simple": 0.5258799195289612,
185
+ "eval_regularize": 442.3320617675781,
186
+ "eval_runtime": 308.996,
187
+ "eval_samples_per_second": 18.738,
188
+ "eval_steps_per_second": 1.563,
189
+ "eval_wo_beta": 14.479047775268555,
190
+ "step": 250
191
+ },
192
+ {
193
+ "dpo_loss": 182.8459930419922,
194
+ "epoch": 0.8502598016060463,
195
+ "grad_norm": 16846.286343055544,
196
+ "learning_rate": 4.924776641419513e-06,
197
+ "logits": -0.6342157125473022,
198
+ "logps": -78.78164672851562,
199
+ "loss": 360.5868,
200
+ "objective": 363.29473876953125,
201
+ "ranking_idealized": 0.4950000047683716,
202
+ "ranking_idealized_expo": 0.4950000047683716,
203
+ "ranking_simple": 0.4970833361148834,
204
+ "regularize": 363.29473876953125,
205
+ "step": 300,
206
+ "wo_beta": 15.358329772949219
207
+ },
208
+ {
209
+ "epoch": 0.8502598016060463,
210
+ "eval_dpo_loss": 222.88400268554688,
211
+ "eval_logits": -0.5983948707580566,
212
+ "eval_logps": -82.20111083984375,
213
+ "eval_loss": 448.9505920410156,
214
+ "eval_objective": 443.9051208496094,
215
+ "eval_ranking_idealized": 0.5212215185165405,
216
+ "eval_ranking_idealized_expo": 0.5212215185165405,
217
+ "eval_ranking_simple": 0.5248447060585022,
218
+ "eval_regularize": 443.9051208496094,
219
+ "eval_runtime": 309.9292,
220
+ "eval_samples_per_second": 18.682,
221
+ "eval_steps_per_second": 1.558,
222
+ "eval_wo_beta": 14.392961502075195,
223
+ "step": 300
224
+ },
225
+ {
226
+ "dpo_loss": 171.55615234375,
227
+ "epoch": 0.9919697685403873,
228
+ "grad_norm": 16864.304890654712,
229
+ "learning_rate": 4.8526047530778175e-06,
230
+ "logits": -0.6237902641296387,
231
+ "logps": -79.1826400756836,
232
+ "loss": 338.3987,
233
+ "objective": 335.6865234375,
234
+ "ranking_idealized": 0.5254166722297668,
235
+ "ranking_idealized_expo": 0.5254166722297668,
236
+ "ranking_simple": 0.5216666460037231,
237
+ "regularize": 335.6865234375,
238
+ "step": 350,
239
+ "wo_beta": 15.20045280456543
240
+ },
241
+ {
242
+ "epoch": 0.9919697685403873,
243
+ "eval_dpo_loss": 232.93649291992188,
244
+ "eval_logits": -0.78554368019104,
245
+ "eval_logps": -84.16381072998047,
246
+ "eval_loss": 462.19232177734375,
247
+ "eval_objective": 461.2073059082031,
248
+ "eval_ranking_idealized": 0.5212215185165405,
249
+ "eval_ranking_idealized_expo": 0.5212215185165405,
250
+ "eval_ranking_simple": 0.5269151329994202,
251
+ "eval_regularize": 461.2073059082031,
252
+ "eval_runtime": 308.3435,
253
+ "eval_samples_per_second": 18.778,
254
+ "eval_steps_per_second": 1.566,
255
+ "eval_wo_beta": 14.297853469848633,
256
+ "step": 350
257
+ },
258
+ {
259
+ "dpo_loss": 158.00930786132812,
260
+ "epoch": 1.1336797354747283,
261
+ "grad_norm": 16419.225936790586,
262
+ "learning_rate": 4.757316345716554e-06,
263
+ "logits": -0.689696192741394,
264
+ "logps": -80.24182891845703,
265
+ "loss": 309.1712,
266
+ "objective": 306.0483703613281,
267
+ "ranking_idealized": 0.5320833325386047,
268
+ "ranking_idealized_expo": 0.5320833325386047,
269
+ "ranking_simple": 0.5229166746139526,
270
+ "regularize": 306.0483703613281,
271
+ "step": 400,
272
+ "wo_beta": 15.245408058166504
273
+ },
274
+ {
275
+ "epoch": 1.1336797354747283,
276
+ "eval_dpo_loss": 248.07177734375,
277
+ "eval_logits": -0.6413922905921936,
278
+ "eval_logps": -82.49343872070312,
279
+ "eval_loss": 480.5964660644531,
280
+ "eval_objective": 478.7404479980469,
281
+ "eval_ranking_idealized": 0.5212215185165405,
282
+ "eval_ranking_idealized_expo": 0.5212215185165405,
283
+ "eval_ranking_simple": 0.5253623127937317,
284
+ "eval_regularize": 478.7404479980469,
285
+ "eval_runtime": 308.667,
286
+ "eval_samples_per_second": 18.758,
287
+ "eval_steps_per_second": 1.565,
288
+ "eval_wo_beta": 14.387246131896973,
289
+ "step": 400
290
+ },
291
+ {
292
+ "dpo_loss": 142.9551239013672,
293
+ "epoch": 1.2753897024090695,
294
+ "grad_norm": 16215.238701636586,
295
+ "learning_rate": 4.639847716126855e-06,
296
+ "logits": -0.5957368612289429,
297
+ "logps": -78.66122436523438,
298
+ "loss": 298.1424,
299
+ "objective": 298.3353271484375,
300
+ "ranking_idealized": 0.5191666483879089,
301
+ "ranking_idealized_expo": 0.5191666483879089,
302
+ "ranking_simple": 0.5204166769981384,
303
+ "regularize": 298.3353271484375,
304
+ "step": 450,
305
+ "wo_beta": 15.736668586730957
306
+ },
307
+ {
308
+ "epoch": 1.2753897024090695,
309
+ "eval_dpo_loss": 247.8721923828125,
310
+ "eval_logits": -0.7014132142066956,
311
+ "eval_logps": -82.14649200439453,
312
+ "eval_loss": 480.3255615234375,
313
+ "eval_objective": 482.1766052246094,
314
+ "eval_ranking_idealized": 0.5212215185165405,
315
+ "eval_ranking_idealized_expo": 0.5212215185165405,
316
+ "eval_ranking_simple": 0.523809552192688,
317
+ "eval_regularize": 482.1766052246094,
318
+ "eval_runtime": 309.6985,
319
+ "eval_samples_per_second": 18.696,
320
+ "eval_steps_per_second": 1.56,
321
+ "eval_wo_beta": 14.369455337524414,
322
+ "step": 450
323
+ },
324
+ {
325
+ "dpo_loss": 135.57347106933594,
326
+ "epoch": 1.4170996693434104,
327
+ "grad_norm": 15944.03549179383,
328
+ "learning_rate": 4.501353102310901e-06,
329
+ "logits": -0.5253962278366089,
330
+ "logps": -78.11959075927734,
331
+ "loss": 282.4504,
332
+ "objective": 278.3994445800781,
333
+ "ranking_idealized": 0.4970833361148834,
334
+ "ranking_idealized_expo": 0.4970833361148834,
335
+ "ranking_simple": 0.49958333373069763,
336
+ "regularize": 278.3994445800781,
337
+ "step": 500,
338
+ "wo_beta": 15.295467376708984
339
+ },
340
+ {
341
+ "epoch": 1.4170996693434104,
342
+ "eval_dpo_loss": 252.20928955078125,
343
+ "eval_logits": -0.45775941014289856,
344
+ "eval_logps": -83.41010284423828,
345
+ "eval_loss": 493.7484436035156,
346
+ "eval_objective": 495.763916015625,
347
+ "eval_ranking_idealized": 0.5212215185165405,
348
+ "eval_ranking_idealized_expo": 0.5212215185165405,
349
+ "eval_ranking_simple": 0.5248447060585022,
350
+ "eval_regularize": 495.763916015625,
351
+ "eval_runtime": 308.7418,
352
+ "eval_samples_per_second": 18.754,
353
+ "eval_steps_per_second": 1.564,
354
+ "eval_wo_beta": 14.174327850341797,
355
+ "step": 500
356
+ },
357
+ {
358
+ "dpo_loss": 138.2200164794922,
359
+ "epoch": 1.5588096362777515,
360
+ "grad_norm": 14777.775900953873,
361
+ "learning_rate": 4.34319334202531e-06,
362
+ "logits": -0.43370625376701355,
363
+ "logps": -79.58238220214844,
364
+ "loss": 261.1027,
365
+ "objective": 261.7706298828125,
366
+ "ranking_idealized": 0.5104166865348816,
367
+ "ranking_idealized_expo": 0.5104166865348816,
368
+ "ranking_simple": 0.5104166865348816,
369
+ "regularize": 261.7706298828125,
370
+ "step": 550,
371
+ "wo_beta": 15.099705696105957
372
+ },
373
+ {
374
+ "epoch": 1.5588096362777515,
375
+ "eval_dpo_loss": 245.87562561035156,
376
+ "eval_logits": -0.5435077548027039,
377
+ "eval_logps": -82.8267593383789,
378
+ "eval_loss": 486.1625671386719,
379
+ "eval_objective": 489.7927551269531,
380
+ "eval_ranking_idealized": 0.5212215185165405,
381
+ "eval_ranking_idealized_expo": 0.5212215185165405,
382
+ "eval_ranking_simple": 0.5253623127937317,
383
+ "eval_regularize": 489.7927551269531,
384
+ "eval_runtime": 313.2811,
385
+ "eval_samples_per_second": 18.482,
386
+ "eval_steps_per_second": 1.542,
387
+ "eval_wo_beta": 14.046432495117188,
388
+ "step": 550
389
+ },
390
+ {
391
+ "dpo_loss": 129.54061889648438,
392
+ "epoch": 1.7005196032120926,
393
+ "grad_norm": 15449.253251441141,
394
+ "learning_rate": 4.16692250129073e-06,
395
+ "logits": -0.4195112884044647,
396
+ "logps": -79.62303924560547,
397
+ "loss": 255.9288,
398
+ "objective": 256.1207275390625,
399
+ "ranking_idealized": 0.5149999856948853,
400
+ "ranking_idealized_expo": 0.5149999856948853,
401
+ "ranking_simple": 0.5095833539962769,
402
+ "regularize": 256.1207275390625,
403
+ "step": 600,
404
+ "wo_beta": 15.011371612548828
405
+ },
406
+ {
407
+ "epoch": 1.7005196032120926,
408
+ "eval_dpo_loss": 251.29342651367188,
409
+ "eval_logits": -0.5346657633781433,
410
+ "eval_logps": -82.17684173583984,
411
+ "eval_loss": 500.38006591796875,
412
+ "eval_objective": 502.1727294921875,
413
+ "eval_ranking_idealized": 0.5212215185165405,
414
+ "eval_ranking_idealized_expo": 0.5212215185165405,
415
+ "eval_ranking_simple": 0.5269151329994202,
416
+ "eval_regularize": 502.1727294921875,
417
+ "eval_runtime": 309.4233,
418
+ "eval_samples_per_second": 18.712,
419
+ "eval_steps_per_second": 1.561,
420
+ "eval_wo_beta": 14.243566513061523,
421
+ "step": 600
422
+ },
423
+ {
424
+ "dpo_loss": 130.9433135986328,
425
+ "epoch": 1.8422295701464337,
426
+ "grad_norm": 15825.93366010816,
427
+ "learning_rate": 3.974272604254906e-06,
428
+ "logits": -0.4912276566028595,
429
+ "logps": -78.56413269042969,
430
+ "loss": 248.6787,
431
+ "objective": 253.7882843017578,
432
+ "ranking_idealized": 0.527916669845581,
433
+ "ranking_idealized_expo": 0.527916669845581,
434
+ "ranking_simple": 0.5224999785423279,
435
+ "regularize": 253.7882843017578,
436
+ "step": 650,
437
+ "wo_beta": 15.684560775756836
438
+ },
439
+ {
440
+ "epoch": 1.8422295701464337,
441
+ "eval_dpo_loss": 254.5959014892578,
442
+ "eval_logits": -0.5140498876571655,
443
+ "eval_logps": -81.49234771728516,
444
+ "eval_loss": 502.3152770996094,
445
+ "eval_objective": 504.1581726074219,
446
+ "eval_ranking_idealized": 0.5212215185165405,
447
+ "eval_ranking_idealized_expo": 0.5212215185165405,
448
+ "eval_ranking_simple": 0.5248447060585022,
449
+ "eval_regularize": 504.1581726074219,
450
+ "eval_runtime": 308.5565,
451
+ "eval_samples_per_second": 18.765,
452
+ "eval_steps_per_second": 1.565,
453
+ "eval_wo_beta": 14.331953048706055,
454
+ "step": 650
455
+ },
456
+ {
457
+ "dpo_loss": 112.15733337402344,
458
+ "epoch": 1.9839395370807746,
459
+ "grad_norm": 16229.838704301123,
460
+ "learning_rate": 3.767136614452458e-06,
461
+ "logits": -0.4758701026439667,
462
+ "logps": -78.98329162597656,
463
+ "loss": 226.4676,
464
+ "objective": 224.5537567138672,
465
+ "ranking_idealized": 0.5112500190734863,
466
+ "ranking_idealized_expo": 0.5108333230018616,
467
+ "ranking_simple": 0.5087500214576721,
468
+ "regularize": 224.5537567138672,
469
+ "step": 700,
470
+ "wo_beta": 15.5169095993042
471
+ },
472
+ {
473
+ "epoch": 1.9839395370807746,
474
+ "eval_dpo_loss": 264.166015625,
475
+ "eval_logits": -0.4816124141216278,
476
+ "eval_logps": -83.42156219482422,
477
+ "eval_loss": 512.698974609375,
478
+ "eval_objective": 516.7103271484375,
479
+ "eval_ranking_idealized": 0.5212215185165405,
480
+ "eval_ranking_idealized_expo": 0.5212215185165405,
481
+ "eval_ranking_simple": 0.5253623127937317,
482
+ "eval_regularize": 516.7103271484375,
483
+ "eval_runtime": 308.9876,
484
+ "eval_samples_per_second": 18.739,
485
+ "eval_steps_per_second": 1.563,
486
+ "eval_wo_beta": 14.08342456817627,
487
+ "step": 700
488
+ },
489
+ {
490
+ "dpo_loss": 105.4539566040039,
491
+ "epoch": 2.1256495040151155,
492
+ "grad_norm": 15281.11762954245,
493
+ "learning_rate": 3.547549834686222e-06,
494
+ "logits": -0.47171100974082947,
495
+ "logps": -79.71944427490234,
496
+ "loss": 207.1551,
497
+ "objective": 207.78355407714844,
498
+ "ranking_idealized": 0.5112500190734863,
499
+ "ranking_idealized_expo": 0.5112500190734863,
500
+ "ranking_simple": 0.51583331823349,
501
+ "regularize": 207.78355407714844,
502
+ "step": 750,
503
+ "wo_beta": 15.408516883850098
504
+ },
505
+ {
506
+ "epoch": 2.1256495040151155,
507
+ "eval_dpo_loss": 259.2528381347656,
508
+ "eval_logits": -0.5409926772117615,
509
+ "eval_logps": -83.45890045166016,
510
+ "eval_loss": 506.4237365722656,
511
+ "eval_objective": 510.6129150390625,
512
+ "eval_ranking_idealized": 0.5212215185165405,
513
+ "eval_ranking_idealized_expo": 0.5212215185165405,
514
+ "eval_ranking_simple": 0.523809552192688,
515
+ "eval_regularize": 510.6129150390625,
516
+ "eval_runtime": 308.6604,
517
+ "eval_samples_per_second": 18.758,
518
+ "eval_steps_per_second": 1.565,
519
+ "eval_wo_beta": 14.129502296447754,
520
+ "step": 750
521
+ },
522
+ {
523
+ "dpo_loss": 97.64974975585938,
524
+ "epoch": 2.2673594709494567,
525
+ "grad_norm": 15212.863475959,
526
+ "learning_rate": 3.3176699082935546e-06,
527
+ "logits": -0.48046550154685974,
528
+ "logps": -81.52581787109375,
529
+ "loss": 197.3545,
530
+ "objective": 198.5575714111328,
531
+ "ranking_idealized": 0.512499988079071,
532
+ "ranking_idealized_expo": 0.512499988079071,
533
+ "ranking_simple": 0.5083333253860474,
534
+ "regularize": 198.5575714111328,
535
+ "step": 800,
536
+ "wo_beta": 15.000423431396484
537
+ },
538
+ {
539
+ "epoch": 2.2673594709494567,
540
+ "eval_dpo_loss": 262.3102111816406,
541
+ "eval_logits": -0.5658813118934631,
542
+ "eval_logps": -84.87467956542969,
543
+ "eval_loss": 513.39794921875,
544
+ "eval_objective": 514.31201171875,
545
+ "eval_ranking_idealized": 0.5212215185165405,
546
+ "eval_ranking_idealized_expo": 0.5212215185165405,
547
+ "eval_ranking_simple": 0.522774338722229,
548
+ "eval_regularize": 514.31201171875,
549
+ "eval_runtime": 310.3348,
550
+ "eval_samples_per_second": 18.657,
551
+ "eval_steps_per_second": 1.556,
552
+ "eval_wo_beta": 14.070388793945312,
553
+ "step": 800
554
+ },
555
+ {
556
+ "dpo_loss": 94.56294250488281,
557
+ "epoch": 2.413793103448276,
558
+ "grad_norm": 14099.488226379159,
559
+ "learning_rate": 3.0797556183036582e-06,
560
+ "logits": -0.5114213824272156,
561
+ "logps": -81.04893493652344,
562
+ "loss": 182.3796,
563
+ "objective": 182.07008361816406,
564
+ "ranking_idealized": 0.5145833492279053,
565
+ "ranking_idealized_expo": 0.5137500166893005,
566
+ "ranking_simple": 0.5129166841506958,
567
+ "regularize": 182.07008361816406,
568
+ "step": 850,
569
+ "wo_beta": 15.241059303283691
570
+ },
571
+ {
572
+ "epoch": 2.413793103448276,
573
+ "eval_dpo_loss": 254.12506103515625,
574
+ "eval_logits": -0.5509631037712097,
575
+ "eval_logps": -82.86239624023438,
576
+ "eval_loss": 501.88311767578125,
577
+ "eval_objective": 504.852294921875,
578
+ "eval_ranking_idealized": 0.5212215185165405,
579
+ "eval_ranking_idealized_expo": 0.5212215185165405,
580
+ "eval_ranking_simple": 0.5274327397346497,
581
+ "eval_regularize": 504.852294921875,
582
+ "eval_runtime": 309.6743,
583
+ "eval_samples_per_second": 18.697,
584
+ "eval_steps_per_second": 1.56,
585
+ "eval_wo_beta": 14.17065143585205,
586
+ "step": 850
587
+ },
588
+ {
589
+ "dpo_loss": 92.21830749511719,
590
+ "epoch": 2.555503070382617,
591
+ "grad_norm": 14214.15214964791,
592
+ "learning_rate": 2.8361446928038298e-06,
593
+ "logits": -0.49887704849243164,
594
+ "logps": -80.69136810302734,
595
+ "loss": 176.042,
596
+ "objective": 176.71592712402344,
597
+ "ranking_idealized": 0.5183333158493042,
598
+ "ranking_idealized_expo": 0.5179166793823242,
599
+ "ranking_simple": 0.5129166841506958,
600
+ "regularize": 176.71592712402344,
601
+ "step": 900,
602
+ "wo_beta": 15.599862098693848
603
+ },
604
+ {
605
+ "epoch": 2.555503070382617,
606
+ "eval_dpo_loss": 263.27996826171875,
607
+ "eval_logits": -0.5038771033287048,
608
+ "eval_logps": -85.07097625732422,
609
+ "eval_loss": 518.1983032226562,
610
+ "eval_objective": 519.5007934570312,
611
+ "eval_ranking_idealized": 0.5212215185165405,
612
+ "eval_ranking_idealized_expo": 0.5212215185165405,
613
+ "eval_ranking_simple": 0.523809552192688,
614
+ "eval_regularize": 519.5007934570312,
615
+ "eval_runtime": 308.5117,
616
+ "eval_samples_per_second": 18.768,
617
+ "eval_steps_per_second": 1.566,
618
+ "eval_wo_beta": 14.112290382385254,
619
+ "step": 900
620
+ },
621
+ {
622
+ "dpo_loss": 81.76506805419922,
623
+ "epoch": 2.697213037316958,
624
+ "grad_norm": 14116.155586290091,
625
+ "learning_rate": 2.5892308345974517e-06,
626
+ "logits": -0.45734792947769165,
627
+ "logps": -81.56855773925781,
628
+ "loss": 164.8281,
629
+ "objective": 161.75088500976562,
630
+ "ranking_idealized": 0.5045833587646484,
631
+ "ranking_idealized_expo": 0.5045833587646484,
632
+ "ranking_simple": 0.5099999904632568,
633
+ "regularize": 161.75088500976562,
634
+ "step": 950,
635
+ "wo_beta": 15.513051986694336
636
+ },
637
+ {
638
+ "epoch": 2.697213037316958,
639
+ "eval_dpo_loss": 262.8074035644531,
640
+ "eval_logits": -0.5199795961380005,
641
+ "eval_logps": -84.58429718017578,
642
+ "eval_loss": 512.1844482421875,
643
+ "eval_objective": 512.7650756835938,
644
+ "eval_ranking_idealized": 0.5212215185165405,
645
+ "eval_ranking_idealized_expo": 0.5212215185165405,
646
+ "eval_ranking_simple": 0.523809552192688,
647
+ "eval_regularize": 512.7650756835938,
648
+ "eval_runtime": 309.6572,
649
+ "eval_samples_per_second": 18.698,
650
+ "eval_steps_per_second": 1.56,
651
+ "eval_wo_beta": 14.164327621459961,
652
+ "step": 950
653
+ },
654
+ {
655
+ "dpo_loss": 76.9761962890625,
656
+ "epoch": 2.838923004251299,
657
+ "grad_norm": 13709.192801586882,
658
+ "learning_rate": 2.341440200858589e-06,
659
+ "logits": -0.46604040265083313,
660
+ "logps": -79.8453598022461,
661
+ "loss": 150.0401,
662
+ "objective": 150.50038146972656,
663
+ "ranking_idealized": 0.5112500190734863,
664
+ "ranking_idealized_expo": 0.5112500190734863,
665
+ "ranking_simple": 0.5099999904632568,
666
+ "regularize": 150.50038146972656,
667
+ "step": 1000,
668
+ "wo_beta": 15.057799339294434
669
+ },
670
+ {
671
+ "epoch": 2.838923004251299,
672
+ "eval_dpo_loss": 263.6168518066406,
673
+ "eval_logits": -0.5218656659126282,
674
+ "eval_logps": -83.73433685302734,
675
+ "eval_loss": 514.70361328125,
676
+ "eval_objective": 516.595947265625,
677
+ "eval_ranking_idealized": 0.5212215185165405,
678
+ "eval_ranking_idealized_expo": 0.5212215185165405,
679
+ "eval_ranking_simple": 0.5258799195289612,
680
+ "eval_regularize": 516.595947265625,
681
+ "eval_runtime": 308.5762,
682
+ "eval_samples_per_second": 18.764,
683
+ "eval_steps_per_second": 1.565,
684
+ "eval_wo_beta": 14.180042266845703,
685
+ "step": 1000
686
+ },
687
+ {
688
+ "dpo_loss": 69.53437805175781,
689
+ "epoch": 2.9806329711856403,
690
+ "grad_norm": 14565.063000853466,
691
+ "learning_rate": 2.0952075638923656e-06,
692
+ "logits": -0.4686031639575958,
693
+ "logps": -80.63922882080078,
694
+ "loss": 141.0317,
695
+ "objective": 136.92088317871094,
696
+ "ranking_idealized": 0.5174999833106995,
697
+ "ranking_idealized_expo": 0.5174999833106995,
698
+ "ranking_simple": 0.5262500047683716,
699
+ "regularize": 136.92088317871094,
700
+ "step": 1050,
701
+ "wo_beta": 15.512243270874023
702
+ },
703
+ {
704
+ "epoch": 2.9806329711856403,
705
+ "eval_dpo_loss": 266.94525146484375,
706
+ "eval_logits": -0.49528759717941284,
707
+ "eval_logps": -84.26761627197266,
708
+ "eval_loss": 519.2467041015625,
709
+ "eval_objective": 521.8153076171875,
710
+ "eval_ranking_idealized": 0.5212215185165405,
711
+ "eval_ranking_idealized_expo": 0.5212215185165405,
712
+ "eval_ranking_simple": 0.5263975262641907,
713
+ "eval_regularize": 521.8153076171875,
714
+ "eval_runtime": 308.2205,
715
+ "eval_samples_per_second": 18.785,
716
+ "eval_steps_per_second": 1.567,
717
+ "eval_wo_beta": 14.257741928100586,
718
+ "step": 1050
719
+ },
720
+ {
721
+ "epoch": 2.9806329711856403,
722
+ "step": 1050,
723
+ "total_flos": 0.0,
724
+ "train_loss": 38.77721726190476,
725
+ "train_runtime": 6898.2426,
726
+ "train_samples_per_second": 36.822,
727
+ "train_steps_per_second": 0.255
728
+ }
729
+ ],
730
+ "logging_steps": 50,
731
+ "max_steps": 1760,
732
+ "num_input_tokens_seen": 0,
733
+ "num_train_epochs": 5,
734
+ "save_steps": 50,
735
+ "stateful_callbacks": {
736
+ "EarlyStoppingCallback": {
737
+ "args": {
738
+ "early_stopping_patience": 5,
739
+ "early_stopping_threshold": 0.0
740
+ },
741
+ "attributes": {
742
+ "early_stopping_patience_counter": 0
743
+ }
744
+ },
745
+ "TrainerControl": {
746
+ "args": {
747
+ "should_epoch_stop": false,
748
+ "should_evaluate": false,
749
+ "should_log": false,
750
+ "should_save": true,
751
+ "should_training_stop": true
752
+ },
753
+ "attributes": {}
754
+ }
755
+ },
756
+ "total_flos": 0.0,
757
+ "train_batch_size": 4,
758
+ "trial_name": null,
759
+ "trial_params": null
760
+ }