hZzy commited on
Commit
f04c3bc
1 Parent(s): f1df128

Model save

Browse files
README.md ADDED
@@ -0,0 +1,93 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
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-DPO-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/tkmcvc2n)
17
+ # qwen2.5-0.5b-expo-DPO-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: 225.2907
22
+ - Logps: -80.3480
23
+ - Logits: -0.5542
24
+ - Objective: 216.5803
25
+ - Dpo Loss: 216.5803
26
+ - Regularize: 216.5803
27
+ - Ranking Simple: 0.5419
28
+ - Ranking Idealized: 0.5212
29
+ - Ranking Idealized Expo: 0.5212
30
+ - Wo Beta: 6.8429
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 | Validation Loss | Logps | Logits | Objective | Dpo Loss | Regularize | Ranking Simple | Ranking Idealized | Ranking Idealized Expo | Wo Beta |
66
+ |:-------------:|:------:|:----:|:---------------:|:--------:|:-------:|:---------:|:--------:|:----------:|:--------------:|:-----------------:|:----------------------:|:-------:|
67
+ | 17.3723 | 0.1417 | 50 | 32.1125 | -90.9520 | -1.4391 | 31.4854 | 31.4854 | 31.4854 | 0.5264 | 0.5212 | 0.5212 | 7.6851 |
68
+ | 60.4454 | 0.2834 | 100 | 70.9968 | -86.6000 | -1.4386 | 70.7719 | 70.7719 | 70.7719 | 0.5305 | 0.5212 | 0.5212 | 7.5289 |
69
+ | 100.2237 | 0.4251 | 150 | 129.8928 | -85.7303 | -1.2892 | 126.8845 | 126.8845 | 126.8845 | 0.5321 | 0.5212 | 0.5212 | 7.4641 |
70
+ | 120.8284 | 0.5668 | 200 | 164.0152 | -75.5542 | -1.3195 | 159.5013 | 159.5013 | 159.5013 | 0.5357 | 0.5212 | 0.5212 | 7.1836 |
71
+ | 134.8217 | 0.7085 | 250 | 195.7212 | -79.3891 | -1.2058 | 190.8510 | 190.8510 | 190.8510 | 0.5285 | 0.5212 | 0.5212 | 7.2711 |
72
+ | 119.0273 | 0.8503 | 300 | 192.5231 | -84.2971 | -0.9945 | 188.0580 | 188.0580 | 188.0580 | 0.5357 | 0.5212 | 0.5212 | 6.9382 |
73
+ | 114.0792 | 0.9920 | 350 | 205.7797 | -82.1125 | -1.0045 | 192.3920 | 192.3920 | 192.3920 | 0.5409 | 0.5212 | 0.5212 | 6.9235 |
74
+ | 72.4145 | 1.1337 | 400 | 212.6613 | -82.8156 | -0.7120 | 204.8122 | 204.8122 | 204.8122 | 0.5409 | 0.5212 | 0.5212 | 7.0485 |
75
+ | 76.9668 | 1.2754 | 450 | 210.2291 | -82.4190 | -0.7807 | 203.0261 | 203.0261 | 203.0261 | 0.5383 | 0.5212 | 0.5212 | 6.9244 |
76
+ | 77.9261 | 1.4171 | 500 | 211.3156 | -81.3728 | -0.7438 | 202.1569 | 202.1569 | 202.1569 | 0.5362 | 0.5212 | 0.5212 | 6.8863 |
77
+ | 70.5755 | 1.5588 | 550 | 212.6468 | -82.3296 | -0.6838 | 200.1410 | 200.1410 | 200.1410 | 0.5430 | 0.5212 | 0.5212 | 6.7241 |
78
+ | 69.6026 | 1.7005 | 600 | 212.0254 | -80.7129 | -0.5569 | 196.9669 | 196.9669 | 196.9669 | 0.5419 | 0.5212 | 0.5212 | 6.6975 |
79
+ | 69.7829 | 1.8422 | 650 | 222.2766 | -79.4968 | -0.7062 | 209.6782 | 209.6782 | 209.6782 | 0.5404 | 0.5212 | 0.5212 | 6.6541 |
80
+ | 62.7864 | 1.9839 | 700 | 226.3468 | -80.2667 | -0.6269 | 213.3031 | 213.3031 | 213.3031 | 0.5399 | 0.5212 | 0.5212 | 6.6215 |
81
+ | 37.3326 | 2.1256 | 750 | 219.7785 | -80.5665 | -0.7007 | 208.8723 | 208.8723 | 208.8723 | 0.5440 | 0.5212 | 0.5212 | 6.7265 |
82
+ | 33.2099 | 2.2674 | 800 | 221.8786 | -81.8901 | -0.5673 | 207.6881 | 207.6881 | 207.6881 | 0.5450 | 0.5212 | 0.5212 | 6.6717 |
83
+ | 33.915 | 2.4091 | 850 | 217.6955 | -81.9134 | -0.5178 | 205.0515 | 205.0515 | 205.0515 | 0.5424 | 0.5212 | 0.5212 | 6.7249 |
84
+ | 35.3572 | 2.5508 | 900 | 224.5402 | -81.5880 | -0.4729 | 214.5052 | 214.5052 | 214.5052 | 0.5435 | 0.5212 | 0.5212 | 6.8278 |
85
+ | 31.032 | 2.6925 | 950 | 225.2907 | -80.3480 | -0.5542 | 216.5803 | 216.5803 | 216.5803 | 0.5419 | 0.5212 | 0.5212 | 6.8429 |
86
+
87
+
88
+ ### Framework versions
89
+
90
+ - Transformers 4.42.0
91
+ - Pytorch 2.3.0+cu121
92
+ - Datasets 2.19.1
93
+ - Tokenizers 0.19.1
all_results.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 2.69248937175248,
3
+ "total_flos": 0.0,
4
+ "train_loss": 70.38767440143384,
5
+ "train_runtime": 26138.5983,
6
+ "train_samples": 50802,
7
+ "train_samples_per_second": 9.718,
8
+ "train_steps_per_second": 0.067
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 sha256:46bfc417acdfaf7e2369ea2197e82165d91d59245316c53a58ca4e463153fd9a
3
  size 1975192208
 
1
  version https://git-lfs.github.com/spec/v1
2
+ oid sha256:40c2173f568ec0d993facc821ef5eba342349bfccf94f837a567a43dfbbfa0ab
3
  size 1975192208
train_results.json ADDED
@@ -0,0 +1,9 @@
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "epoch": 2.69248937175248,
3
+ "total_flos": 0.0,
4
+ "train_loss": 70.38767440143384,
5
+ "train_runtime": 26138.5983,
6
+ "train_samples": 50802,
7
+ "train_samples_per_second": 9.718,
8
+ "train_steps_per_second": 0.067
9
+ }
trainer_state.json ADDED
@@ -0,0 +1,694 @@
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ {
2
+ "best_metric": 6.621464252471924,
3
+ "best_model_checkpoint": "./qwen2.5-0.5b/qwen2.5-0.5b-expo-DPO-ES-100/checkpoint-700",
4
+ "epoch": 2.69248937175248,
5
+ "eval_steps": 50,
6
+ "global_step": 950,
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": 18442.54709688657,
15
+ "learning_rate": 2.840909090909091e-08,
16
+ "logits": -1.359458565711975,
17
+ "logps": -84.69721221923828,
18
+ "loss": 0.6931,
19
+ "objective": 0.6931471824645996,
20
+ "ranking_idealized": 0.5833333134651184,
21
+ "ranking_idealized_expo": 0.5833333134651184,
22
+ "ranking_simple": 0.5833333134651184,
23
+ "regularize": 0.6931471824645996,
24
+ "step": 1,
25
+ "wo_beta": 5.271125316619873
26
+ },
27
+ {
28
+ "dpo_loss": 16.993518829345703,
29
+ "epoch": 0.14170996693434104,
30
+ "grad_norm": 20039.682390740076,
31
+ "learning_rate": 1.4204545454545458e-06,
32
+ "logits": -1.4496889114379883,
33
+ "logps": -84.4629898071289,
34
+ "loss": 17.3723,
35
+ "objective": 16.993518829345703,
36
+ "ranking_idealized": 0.5221088528633118,
37
+ "ranking_idealized_expo": 0.5216836929321289,
38
+ "ranking_simple": 0.5242347121238708,
39
+ "regularize": 16.993518829345703,
40
+ "step": 50,
41
+ "wo_beta": 7.0965657234191895
42
+ },
43
+ {
44
+ "epoch": 0.14170996693434104,
45
+ "eval_dpo_loss": 31.485410690307617,
46
+ "eval_logits": -1.4391247034072876,
47
+ "eval_logps": -90.95195770263672,
48
+ "eval_loss": 32.11253356933594,
49
+ "eval_objective": 31.485410690307617,
50
+ "eval_ranking_idealized": 0.5212215185165405,
51
+ "eval_ranking_idealized_expo": 0.5212215185165405,
52
+ "eval_ranking_simple": 0.5263975262641907,
53
+ "eval_regularize": 31.485410690307617,
54
+ "eval_runtime": 309.3922,
55
+ "eval_samples_per_second": 18.714,
56
+ "eval_steps_per_second": 1.561,
57
+ "eval_wo_beta": 7.685111999511719,
58
+ "step": 50
59
+ },
60
+ {
61
+ "dpo_loss": 59.36994934082031,
62
+ "epoch": 0.2834199338686821,
63
+ "grad_norm": 16985.157671530273,
64
+ "learning_rate": 2.8409090909090916e-06,
65
+ "logits": -1.4624311923980713,
66
+ "logps": -81.97602844238281,
67
+ "loss": 60.4454,
68
+ "objective": 59.36994934082031,
69
+ "ranking_idealized": 0.5137500166893005,
70
+ "ranking_idealized_expo": 0.5137500166893005,
71
+ "ranking_simple": 0.5433333516120911,
72
+ "regularize": 59.36994934082031,
73
+ "step": 100,
74
+ "wo_beta": 6.49465799331665
75
+ },
76
+ {
77
+ "epoch": 0.2834199338686821,
78
+ "eval_dpo_loss": 70.77193450927734,
79
+ "eval_logits": -1.4385725259780884,
80
+ "eval_logps": -86.5999984741211,
81
+ "eval_loss": 70.99683380126953,
82
+ "eval_objective": 70.77193450927734,
83
+ "eval_ranking_idealized": 0.5212215185165405,
84
+ "eval_ranking_idealized_expo": 0.5212215185165405,
85
+ "eval_ranking_simple": 0.5305383205413818,
86
+ "eval_regularize": 70.77193450927734,
87
+ "eval_runtime": 308.6565,
88
+ "eval_samples_per_second": 18.759,
89
+ "eval_steps_per_second": 1.565,
90
+ "eval_wo_beta": 7.528871059417725,
91
+ "step": 100
92
+ },
93
+ {
94
+ "dpo_loss": 101.89765930175781,
95
+ "epoch": 0.42512990080302315,
96
+ "grad_norm": 14194.684745919038,
97
+ "learning_rate": 4.2613636363636365e-06,
98
+ "logits": -1.386507511138916,
99
+ "logps": -78.6588134765625,
100
+ "loss": 100.2237,
101
+ "objective": 101.89765930175781,
102
+ "ranking_idealized": 0.527916669845581,
103
+ "ranking_idealized_expo": 0.527916669845581,
104
+ "ranking_simple": 0.5612499713897705,
105
+ "regularize": 101.89765930175781,
106
+ "step": 150,
107
+ "wo_beta": 6.392862319946289
108
+ },
109
+ {
110
+ "epoch": 0.42512990080302315,
111
+ "eval_dpo_loss": 126.88445281982422,
112
+ "eval_logits": -1.2892017364501953,
113
+ "eval_logps": -85.73033905029297,
114
+ "eval_loss": 129.89280700683594,
115
+ "eval_objective": 126.88445281982422,
116
+ "eval_ranking_idealized": 0.5212215185165405,
117
+ "eval_ranking_idealized_expo": 0.5212215185165405,
118
+ "eval_ranking_simple": 0.5320910811424255,
119
+ "eval_regularize": 126.88445281982422,
120
+ "eval_runtime": 308.7582,
121
+ "eval_samples_per_second": 18.753,
122
+ "eval_steps_per_second": 1.564,
123
+ "eval_wo_beta": 7.464066505432129,
124
+ "step": 150
125
+ },
126
+ {
127
+ "dpo_loss": 119.16841125488281,
128
+ "epoch": 0.5668398677373642,
129
+ "grad_norm": 12833.152797223363,
130
+ "learning_rate": 4.997168347957521e-06,
131
+ "logits": -1.3282432556152344,
132
+ "logps": -75.89098358154297,
133
+ "loss": 120.8284,
134
+ "objective": 119.16841125488281,
135
+ "ranking_idealized": 0.51541668176651,
136
+ "ranking_idealized_expo": 0.51541668176651,
137
+ "ranking_simple": 0.5562499761581421,
138
+ "regularize": 119.16841125488281,
139
+ "step": 200,
140
+ "wo_beta": 6.29520320892334
141
+ },
142
+ {
143
+ "epoch": 0.5668398677373642,
144
+ "eval_dpo_loss": 159.50131225585938,
145
+ "eval_logits": -1.3194636106491089,
146
+ "eval_logps": -75.5542221069336,
147
+ "eval_loss": 164.0151824951172,
148
+ "eval_objective": 159.50131225585938,
149
+ "eval_ranking_idealized": 0.5212215185165405,
150
+ "eval_ranking_idealized_expo": 0.5212215185165405,
151
+ "eval_ranking_simple": 0.5357142686843872,
152
+ "eval_regularize": 159.50131225585938,
153
+ "eval_runtime": 308.6776,
154
+ "eval_samples_per_second": 18.757,
155
+ "eval_steps_per_second": 1.565,
156
+ "eval_wo_beta": 7.183592319488525,
157
+ "step": 200
158
+ },
159
+ {
160
+ "dpo_loss": 134.0004119873047,
161
+ "epoch": 0.7085498346717053,
162
+ "grad_norm": 13188.097817084132,
163
+ "learning_rate": 4.973122855144066e-06,
164
+ "logits": -1.138743281364441,
165
+ "logps": -72.49617004394531,
166
+ "loss": 134.8217,
167
+ "objective": 134.0004119873047,
168
+ "ranking_idealized": 0.5162500143051147,
169
+ "ranking_idealized_expo": 0.5162500143051147,
170
+ "ranking_simple": 0.5795833468437195,
171
+ "regularize": 134.0004119873047,
172
+ "step": 250,
173
+ "wo_beta": 6.357597827911377
174
+ },
175
+ {
176
+ "epoch": 0.7085498346717053,
177
+ "eval_dpo_loss": 190.8510284423828,
178
+ "eval_logits": -1.2058484554290771,
179
+ "eval_logps": -79.3890609741211,
180
+ "eval_loss": 195.72117614746094,
181
+ "eval_objective": 190.8510284423828,
182
+ "eval_ranking_idealized": 0.5212215185165405,
183
+ "eval_ranking_idealized_expo": 0.5212215185165405,
184
+ "eval_ranking_simple": 0.5284678936004639,
185
+ "eval_regularize": 190.8510284423828,
186
+ "eval_runtime": 308.7239,
187
+ "eval_samples_per_second": 18.755,
188
+ "eval_steps_per_second": 1.565,
189
+ "eval_wo_beta": 7.27105188369751,
190
+ "step": 250
191
+ },
192
+ {
193
+ "dpo_loss": 114.66310119628906,
194
+ "epoch": 0.8502598016060463,
195
+ "grad_norm": 8558.083204689276,
196
+ "learning_rate": 4.924776641419513e-06,
197
+ "logits": -1.090308666229248,
198
+ "logps": -76.0093765258789,
199
+ "loss": 119.0273,
200
+ "objective": 114.66310119628906,
201
+ "ranking_idealized": 0.4950000047683716,
202
+ "ranking_idealized_expo": 0.4950000047683716,
203
+ "ranking_simple": 0.5687500238418579,
204
+ "regularize": 114.66310119628906,
205
+ "step": 300,
206
+ "wo_beta": 6.086948871612549
207
+ },
208
+ {
209
+ "epoch": 0.8502598016060463,
210
+ "eval_dpo_loss": 188.05804443359375,
211
+ "eval_logits": -0.994479775428772,
212
+ "eval_logps": -84.29710388183594,
213
+ "eval_loss": 192.52310180664062,
214
+ "eval_objective": 188.05804443359375,
215
+ "eval_ranking_idealized": 0.5212215185165405,
216
+ "eval_ranking_idealized_expo": 0.5212215185165405,
217
+ "eval_ranking_simple": 0.5357142686843872,
218
+ "eval_regularize": 188.05804443359375,
219
+ "eval_runtime": 308.8243,
220
+ "eval_samples_per_second": 18.749,
221
+ "eval_steps_per_second": 1.564,
222
+ "eval_wo_beta": 6.938241958618164,
223
+ "step": 300
224
+ },
225
+ {
226
+ "dpo_loss": 116.39104461669922,
227
+ "epoch": 0.9919697685403873,
228
+ "grad_norm": 14112.952284083942,
229
+ "learning_rate": 4.8526047530778175e-06,
230
+ "logits": -0.9387365579605103,
231
+ "logps": -79.04769897460938,
232
+ "loss": 114.0792,
233
+ "objective": 116.39104461669922,
234
+ "ranking_idealized": 0.5254166722297668,
235
+ "ranking_idealized_expo": 0.5254166722297668,
236
+ "ranking_simple": 0.5975000262260437,
237
+ "regularize": 116.39104461669922,
238
+ "step": 350,
239
+ "wo_beta": 5.450421333312988
240
+ },
241
+ {
242
+ "epoch": 0.9919697685403873,
243
+ "eval_dpo_loss": 192.3919677734375,
244
+ "eval_logits": -1.0045075416564941,
245
+ "eval_logps": -82.11246490478516,
246
+ "eval_loss": 205.77967834472656,
247
+ "eval_objective": 192.3919677734375,
248
+ "eval_ranking_idealized": 0.5212215185165405,
249
+ "eval_ranking_idealized_expo": 0.5212215185165405,
250
+ "eval_ranking_simple": 0.5408902764320374,
251
+ "eval_regularize": 192.3919677734375,
252
+ "eval_runtime": 308.4486,
253
+ "eval_samples_per_second": 18.771,
254
+ "eval_steps_per_second": 1.566,
255
+ "eval_wo_beta": 6.923487663269043,
256
+ "step": 350
257
+ },
258
+ {
259
+ "dpo_loss": 73.98444366455078,
260
+ "epoch": 1.1336797354747283,
261
+ "grad_norm": 7172.9062533852075,
262
+ "learning_rate": 4.757316345716554e-06,
263
+ "logits": -0.8197985887527466,
264
+ "logps": -76.91300201416016,
265
+ "loss": 72.4145,
266
+ "objective": 73.98444366455078,
267
+ "ranking_idealized": 0.5320833325386047,
268
+ "ranking_idealized_expo": 0.5320833325386047,
269
+ "ranking_simple": 0.621666669845581,
270
+ "regularize": 73.98444366455078,
271
+ "step": 400,
272
+ "wo_beta": 4.771157264709473
273
+ },
274
+ {
275
+ "epoch": 1.1336797354747283,
276
+ "eval_dpo_loss": 204.8121795654297,
277
+ "eval_logits": -0.7119522094726562,
278
+ "eval_logps": -82.81564331054688,
279
+ "eval_loss": 212.66128540039062,
280
+ "eval_objective": 204.8121795654297,
281
+ "eval_ranking_idealized": 0.5212215185165405,
282
+ "eval_ranking_idealized_expo": 0.5212215185165405,
283
+ "eval_ranking_simple": 0.5408902764320374,
284
+ "eval_regularize": 204.8121795654297,
285
+ "eval_runtime": 308.465,
286
+ "eval_samples_per_second": 18.77,
287
+ "eval_steps_per_second": 1.566,
288
+ "eval_wo_beta": 7.0484819412231445,
289
+ "step": 400
290
+ },
291
+ {
292
+ "dpo_loss": 81.11451721191406,
293
+ "epoch": 1.2753897024090695,
294
+ "grad_norm": 7257.334900105363,
295
+ "learning_rate": 4.639847716126855e-06,
296
+ "logits": -0.6066682934761047,
297
+ "logps": -78.56853485107422,
298
+ "loss": 76.9668,
299
+ "objective": 81.11451721191406,
300
+ "ranking_idealized": 0.5191666483879089,
301
+ "ranking_idealized_expo": 0.5191666483879089,
302
+ "ranking_simple": 0.6295833587646484,
303
+ "regularize": 81.11451721191406,
304
+ "step": 450,
305
+ "wo_beta": 5.272921562194824
306
+ },
307
+ {
308
+ "epoch": 1.2753897024090695,
309
+ "eval_dpo_loss": 203.026123046875,
310
+ "eval_logits": -0.7807133197784424,
311
+ "eval_logps": -82.4189682006836,
312
+ "eval_loss": 210.22909545898438,
313
+ "eval_objective": 203.026123046875,
314
+ "eval_ranking_idealized": 0.5212215185165405,
315
+ "eval_ranking_idealized_expo": 0.5212215185165405,
316
+ "eval_ranking_simple": 0.5383023023605347,
317
+ "eval_regularize": 203.026123046875,
318
+ "eval_runtime": 309.148,
319
+ "eval_samples_per_second": 18.729,
320
+ "eval_steps_per_second": 1.562,
321
+ "eval_wo_beta": 6.924361705780029,
322
+ "step": 450
323
+ },
324
+ {
325
+ "dpo_loss": 76.8044204711914,
326
+ "epoch": 1.4170996693434104,
327
+ "grad_norm": 8969.209638335737,
328
+ "learning_rate": 4.501353102310901e-06,
329
+ "logits": -0.6817887425422668,
330
+ "logps": -77.68151092529297,
331
+ "loss": 77.9261,
332
+ "objective": 76.8044204711914,
333
+ "ranking_idealized": 0.4970833361148834,
334
+ "ranking_idealized_expo": 0.4970833361148834,
335
+ "ranking_simple": 0.6170833110809326,
336
+ "regularize": 76.8044204711914,
337
+ "step": 500,
338
+ "wo_beta": 5.202561855316162
339
+ },
340
+ {
341
+ "epoch": 1.4170996693434104,
342
+ "eval_dpo_loss": 202.15687561035156,
343
+ "eval_logits": -0.7438479661941528,
344
+ "eval_logps": -81.37281036376953,
345
+ "eval_loss": 211.31556701660156,
346
+ "eval_objective": 202.15687561035156,
347
+ "eval_ranking_idealized": 0.5212215185165405,
348
+ "eval_ranking_idealized_expo": 0.5212215185165405,
349
+ "eval_ranking_simple": 0.5362318754196167,
350
+ "eval_regularize": 202.15687561035156,
351
+ "eval_runtime": 308.8567,
352
+ "eval_samples_per_second": 18.747,
353
+ "eval_steps_per_second": 1.564,
354
+ "eval_wo_beta": 6.886315822601318,
355
+ "step": 500
356
+ },
357
+ {
358
+ "dpo_loss": 70.9330062866211,
359
+ "epoch": 1.5588096362777515,
360
+ "grad_norm": 8040.484078292184,
361
+ "learning_rate": 4.34319334202531e-06,
362
+ "logits": -0.636380136013031,
363
+ "logps": -77.03202056884766,
364
+ "loss": 70.5755,
365
+ "objective": 70.9330062866211,
366
+ "ranking_idealized": 0.5104166865348816,
367
+ "ranking_idealized_expo": 0.5104166865348816,
368
+ "ranking_simple": 0.637499988079071,
369
+ "regularize": 70.9330062866211,
370
+ "step": 550,
371
+ "wo_beta": 4.849801063537598
372
+ },
373
+ {
374
+ "epoch": 1.5588096362777515,
375
+ "eval_dpo_loss": 200.1410369873047,
376
+ "eval_logits": -0.6838412880897522,
377
+ "eval_logps": -82.32962799072266,
378
+ "eval_loss": 212.6468048095703,
379
+ "eval_objective": 200.1410369873047,
380
+ "eval_ranking_idealized": 0.5212215185165405,
381
+ "eval_ranking_idealized_expo": 0.5212215185165405,
382
+ "eval_ranking_simple": 0.5429606437683105,
383
+ "eval_regularize": 200.1410369873047,
384
+ "eval_runtime": 311.5527,
385
+ "eval_samples_per_second": 18.584,
386
+ "eval_steps_per_second": 1.55,
387
+ "eval_wo_beta": 6.724142551422119,
388
+ "step": 550
389
+ },
390
+ {
391
+ "dpo_loss": 65.89614868164062,
392
+ "epoch": 1.7005196032120926,
393
+ "grad_norm": 7186.548213944491,
394
+ "learning_rate": 4.16692250129073e-06,
395
+ "logits": -0.510294497013092,
396
+ "logps": -78.14505004882812,
397
+ "loss": 69.6026,
398
+ "objective": 65.89614868164062,
399
+ "ranking_idealized": 0.5149999856948853,
400
+ "ranking_idealized_expo": 0.5149999856948853,
401
+ "ranking_simple": 0.6416666507720947,
402
+ "regularize": 65.89614868164062,
403
+ "step": 600,
404
+ "wo_beta": 4.862266540527344
405
+ },
406
+ {
407
+ "epoch": 1.7005196032120926,
408
+ "eval_dpo_loss": 196.96690368652344,
409
+ "eval_logits": -0.5568598508834839,
410
+ "eval_logps": -80.71286010742188,
411
+ "eval_loss": 212.02540588378906,
412
+ "eval_objective": 196.96690368652344,
413
+ "eval_ranking_idealized": 0.5212215185165405,
414
+ "eval_ranking_idealized_expo": 0.5212215185165405,
415
+ "eval_ranking_simple": 0.5419254899024963,
416
+ "eval_regularize": 196.96690368652344,
417
+ "eval_runtime": 308.7949,
418
+ "eval_samples_per_second": 18.75,
419
+ "eval_steps_per_second": 1.564,
420
+ "eval_wo_beta": 6.6975417137146,
421
+ "step": 600
422
+ },
423
+ {
424
+ "dpo_loss": 69.19684600830078,
425
+ "epoch": 1.8422295701464337,
426
+ "grad_norm": 7556.588020373631,
427
+ "learning_rate": 3.974272604254906e-06,
428
+ "logits": -0.5819191336631775,
429
+ "logps": -75.77037048339844,
430
+ "loss": 69.7829,
431
+ "objective": 69.19684600830078,
432
+ "ranking_idealized": 0.527916669845581,
433
+ "ranking_idealized_expo": 0.527916669845581,
434
+ "ranking_simple": 0.6395833492279053,
435
+ "regularize": 69.19684600830078,
436
+ "step": 650,
437
+ "wo_beta": 5.048923969268799
438
+ },
439
+ {
440
+ "epoch": 1.8422295701464337,
441
+ "eval_dpo_loss": 209.67823791503906,
442
+ "eval_logits": -0.7061935663223267,
443
+ "eval_logps": -79.49684143066406,
444
+ "eval_loss": 222.27664184570312,
445
+ "eval_objective": 209.67823791503906,
446
+ "eval_ranking_idealized": 0.5212215185165405,
447
+ "eval_ranking_idealized_expo": 0.5212215185165405,
448
+ "eval_ranking_simple": 0.5403726696968079,
449
+ "eval_regularize": 209.67823791503906,
450
+ "eval_runtime": 308.514,
451
+ "eval_samples_per_second": 18.767,
452
+ "eval_steps_per_second": 1.566,
453
+ "eval_wo_beta": 6.654106140136719,
454
+ "step": 650
455
+ },
456
+ {
457
+ "dpo_loss": 60.93992233276367,
458
+ "epoch": 1.9839395370807746,
459
+ "grad_norm": 7674.886013172465,
460
+ "learning_rate": 3.767136614452458e-06,
461
+ "logits": -0.5295735597610474,
462
+ "logps": -76.11310577392578,
463
+ "loss": 62.7864,
464
+ "objective": 60.93992233276367,
465
+ "ranking_idealized": 0.5112500190734863,
466
+ "ranking_idealized_expo": 0.5108333230018616,
467
+ "ranking_simple": 0.6387500166893005,
468
+ "regularize": 60.93992233276367,
469
+ "step": 700,
470
+ "wo_beta": 4.921832084655762
471
+ },
472
+ {
473
+ "epoch": 1.9839395370807746,
474
+ "eval_dpo_loss": 213.3031463623047,
475
+ "eval_logits": -0.6268911957740784,
476
+ "eval_logps": -80.2667236328125,
477
+ "eval_loss": 226.3468017578125,
478
+ "eval_objective": 213.3031463623047,
479
+ "eval_ranking_idealized": 0.5212215185165405,
480
+ "eval_ranking_idealized_expo": 0.5212215185165405,
481
+ "eval_ranking_simple": 0.5398550629615784,
482
+ "eval_regularize": 213.3031463623047,
483
+ "eval_runtime": 312.7433,
484
+ "eval_samples_per_second": 18.514,
485
+ "eval_steps_per_second": 1.544,
486
+ "eval_wo_beta": 6.621464252471924,
487
+ "step": 700
488
+ },
489
+ {
490
+ "dpo_loss": 37.5257682800293,
491
+ "epoch": 2.1256495040151155,
492
+ "grad_norm": 6013.443827898748,
493
+ "learning_rate": 3.547549834686222e-06,
494
+ "logits": -0.507068932056427,
495
+ "logps": -76.31243133544922,
496
+ "loss": 37.3326,
497
+ "objective": 37.5257682800293,
498
+ "ranking_idealized": 0.5112500190734863,
499
+ "ranking_idealized_expo": 0.5112500190734863,
500
+ "ranking_simple": 0.6583333611488342,
501
+ "regularize": 37.5257682800293,
502
+ "step": 750,
503
+ "wo_beta": 4.600498676300049
504
+ },
505
+ {
506
+ "epoch": 2.1256495040151155,
507
+ "eval_dpo_loss": 208.872314453125,
508
+ "eval_logits": -0.7007076144218445,
509
+ "eval_logps": -80.56648254394531,
510
+ "eval_loss": 219.77853393554688,
511
+ "eval_objective": 208.872314453125,
512
+ "eval_ranking_idealized": 0.5212215185165405,
513
+ "eval_ranking_idealized_expo": 0.5212215185165405,
514
+ "eval_ranking_simple": 0.5439958572387695,
515
+ "eval_regularize": 208.872314453125,
516
+ "eval_runtime": 308.6702,
517
+ "eval_samples_per_second": 18.758,
518
+ "eval_steps_per_second": 1.565,
519
+ "eval_wo_beta": 6.726495265960693,
520
+ "step": 750
521
+ },
522
+ {
523
+ "dpo_loss": 32.680686950683594,
524
+ "epoch": 2.2673594709494567,
525
+ "grad_norm": 5530.199901679942,
526
+ "learning_rate": 3.3176699082935546e-06,
527
+ "logits": -0.5285750031471252,
528
+ "logps": -77.53636932373047,
529
+ "loss": 33.2099,
530
+ "objective": 32.680686950683594,
531
+ "ranking_idealized": 0.512499988079071,
532
+ "ranking_idealized_expo": 0.512499988079071,
533
+ "ranking_simple": 0.6654166579246521,
534
+ "regularize": 32.680686950683594,
535
+ "step": 800,
536
+ "wo_beta": 4.228292465209961
537
+ },
538
+ {
539
+ "epoch": 2.2673594709494567,
540
+ "eval_dpo_loss": 207.68812561035156,
541
+ "eval_logits": -0.5673097968101501,
542
+ "eval_logps": -81.89009857177734,
543
+ "eval_loss": 221.87864685058594,
544
+ "eval_objective": 207.68812561035156,
545
+ "eval_ranking_idealized": 0.5212215185165405,
546
+ "eval_ranking_idealized_expo": 0.5212215185165405,
547
+ "eval_ranking_simple": 0.5450310707092285,
548
+ "eval_regularize": 207.68812561035156,
549
+ "eval_runtime": 308.8922,
550
+ "eval_samples_per_second": 18.744,
551
+ "eval_steps_per_second": 1.564,
552
+ "eval_wo_beta": 6.671654224395752,
553
+ "step": 800
554
+ },
555
+ {
556
+ "dpo_loss": 34.45829772949219,
557
+ "epoch": 2.409069437883798,
558
+ "grad_norm": 4586.247966907105,
559
+ "learning_rate": 3.0797556183036582e-06,
560
+ "logits": -0.49624624848365784,
561
+ "logps": -79.02293395996094,
562
+ "loss": 33.915,
563
+ "objective": 34.45829772949219,
564
+ "ranking_idealized": 0.5141666531562805,
565
+ "ranking_idealized_expo": 0.5133333206176758,
566
+ "ranking_simple": 0.6575000286102295,
567
+ "regularize": 34.45829772949219,
568
+ "step": 850,
569
+ "wo_beta": 4.579156398773193
570
+ },
571
+ {
572
+ "epoch": 2.409069437883798,
573
+ "eval_dpo_loss": 205.05154418945312,
574
+ "eval_logits": -0.5177903771400452,
575
+ "eval_logps": -81.91341400146484,
576
+ "eval_loss": 217.6955108642578,
577
+ "eval_objective": 205.05154418945312,
578
+ "eval_ranking_idealized": 0.5212215185165405,
579
+ "eval_ranking_idealized_expo": 0.5212215185165405,
580
+ "eval_ranking_simple": 0.542443037033081,
581
+ "eval_regularize": 205.05154418945312,
582
+ "eval_runtime": 308.6493,
583
+ "eval_samples_per_second": 18.759,
584
+ "eval_steps_per_second": 1.565,
585
+ "eval_wo_beta": 6.724928855895996,
586
+ "step": 850
587
+ },
588
+ {
589
+ "dpo_loss": 39.46757888793945,
590
+ "epoch": 2.550779404818139,
591
+ "grad_norm": 5711.491250646742,
592
+ "learning_rate": 2.8361446928038298e-06,
593
+ "logits": -0.44471094012260437,
594
+ "logps": -76.97750091552734,
595
+ "loss": 35.3572,
596
+ "objective": 39.46757888793945,
597
+ "ranking_idealized": 0.5183333158493042,
598
+ "ranking_idealized_expo": 0.5179166793823242,
599
+ "ranking_simple": 0.6600000262260437,
600
+ "regularize": 39.46757888793945,
601
+ "step": 900,
602
+ "wo_beta": 4.633584022521973
603
+ },
604
+ {
605
+ "epoch": 2.550779404818139,
606
+ "eval_dpo_loss": 214.5051727294922,
607
+ "eval_logits": -0.47291529178619385,
608
+ "eval_logps": -81.58795166015625,
609
+ "eval_loss": 224.5401611328125,
610
+ "eval_objective": 214.5051727294922,
611
+ "eval_ranking_idealized": 0.5212215185165405,
612
+ "eval_ranking_idealized_expo": 0.5212215185165405,
613
+ "eval_ranking_simple": 0.54347825050354,
614
+ "eval_regularize": 214.5051727294922,
615
+ "eval_runtime": 308.5271,
616
+ "eval_samples_per_second": 18.767,
617
+ "eval_steps_per_second": 1.566,
618
+ "eval_wo_beta": 6.827796936035156,
619
+ "step": 900
620
+ },
621
+ {
622
+ "dpo_loss": 30.270259857177734,
623
+ "epoch": 2.69248937175248,
624
+ "grad_norm": 4986.6651394423525,
625
+ "learning_rate": 2.5892308345974517e-06,
626
+ "logits": -0.46939340233802795,
627
+ "logps": -76.20037841796875,
628
+ "loss": 31.032,
629
+ "objective": 30.270259857177734,
630
+ "ranking_idealized": 0.5058333277702332,
631
+ "ranking_idealized_expo": 0.5058333277702332,
632
+ "ranking_simple": 0.6645833253860474,
633
+ "regularize": 30.270259857177734,
634
+ "step": 950,
635
+ "wo_beta": 4.50759220123291
636
+ },
637
+ {
638
+ "epoch": 2.69248937175248,
639
+ "eval_dpo_loss": 216.58029174804688,
640
+ "eval_logits": -0.55417799949646,
641
+ "eval_logps": -80.34796905517578,
642
+ "eval_loss": 225.29067993164062,
643
+ "eval_objective": 216.58029174804688,
644
+ "eval_ranking_idealized": 0.5212215185165405,
645
+ "eval_ranking_idealized_expo": 0.5212215185165405,
646
+ "eval_ranking_simple": 0.5419254899024963,
647
+ "eval_regularize": 216.58029174804688,
648
+ "eval_runtime": 309.3192,
649
+ "eval_samples_per_second": 18.719,
650
+ "eval_steps_per_second": 1.561,
651
+ "eval_wo_beta": 6.8429484367370605,
652
+ "step": 950
653
+ },
654
+ {
655
+ "epoch": 2.69248937175248,
656
+ "step": 950,
657
+ "total_flos": 0.0,
658
+ "train_loss": 70.38767440143384,
659
+ "train_runtime": 26138.5983,
660
+ "train_samples_per_second": 9.718,
661
+ "train_steps_per_second": 0.067
662
+ }
663
+ ],
664
+ "logging_steps": 50,
665
+ "max_steps": 1760,
666
+ "num_input_tokens_seen": 0,
667
+ "num_train_epochs": 5,
668
+ "save_steps": 50,
669
+ "stateful_callbacks": {
670
+ "EarlyStoppingCallback": {
671
+ "args": {
672
+ "early_stopping_patience": 5,
673
+ "early_stopping_threshold": 0.0
674
+ },
675
+ "attributes": {
676
+ "early_stopping_patience_counter": 0
677
+ }
678
+ },
679
+ "TrainerControl": {
680
+ "args": {
681
+ "should_epoch_stop": false,
682
+ "should_evaluate": false,
683
+ "should_log": false,
684
+ "should_save": true,
685
+ "should_training_stop": true
686
+ },
687
+ "attributes": {}
688
+ }
689
+ },
690
+ "total_flos": 0.0,
691
+ "train_batch_size": 4,
692
+ "trial_name": null,
693
+ "trial_params": null
694
+ }