gpt-imdb-sigmoid-beta_0.1
This model is a fine-tuned version of lvwerra/gpt2-imdb on an unknown dataset. It achieves the following results on the evaluation set:
- Step: 7000
- Loss: 0.1445
- Rewards/chosen: -5.6156
- Rewards/rejected: -11.9139
- Rewards/accuracies: 0.9354
- Rewards/margins: 6.2982
- Logps/rejected: -382.8238
- Logps/chosen: -291.4216
- Logits/rejected: -44.3728
- Logits/chosen: -46.3321
Model description
More information needed
Intended uses & limitations
More information needed
Training and evaluation data
More information needed
Training procedure
Training hyperparameters
The following hyperparameters were used during training:
- learning_rate: 1e-05
- train_batch_size: 24
- eval_batch_size: 24
- seed: 42
- optimizer: Adam with betas=(0.9,0.99) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_steps: 150
- num_epochs: 3
Training results
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|---|---|---|---|---|---|---|---|---|---|---|---|
| 0.2741 | 0.21 | 500 | 0.3546 | -0.7644 | -2.6310 | 0.8604 | 1.8666 | -289.9951 | -242.9089 | -34.2705 | -35.4568 |
| 0.3403 | 0.42 | 1000 | 0.2963 | -1.6755 | -4.3008 | 0.8687 | 2.6253 | -306.6930 | -252.0203 | -40.9205 | -42.3105 |
| 0.1939 | 0.63 | 1500 | 0.2596 | -3.1297 | -6.7295 | 0.8771 | 3.5998 | -330.9802 | -266.5624 | -37.6829 | -39.1821 |
| 0.2094 | 0.83 | 2000 | 0.1941 | -2.9414 | -6.9143 | 0.9292 | 3.9728 | -332.8280 | -264.6796 | -38.0792 | -39.7464 |
| 0.1481 | 1.04 | 2500 | 0.1744 | -3.7473 | -8.3469 | 0.9333 | 4.5996 | -347.1542 | -272.7383 | -40.9252 | -42.5164 |
| 0.2862 | 1.25 | 3000 | 0.1750 | -4.5825 | -9.7147 | 0.9292 | 5.1322 | -360.8324 | -281.0905 | -41.9790 | -44.0717 |
| 0.304 | 1.46 | 3500 | 0.1652 | -4.3291 | -9.8200 | 0.9333 | 5.4909 | -361.8853 | -278.5559 | -44.1786 | -46.1418 |
| 0.2167 | 1.67 | 4000 | 0.1580 | -4.6175 | -10.0305 | 0.9354 | 5.4130 | -363.9903 | -281.4398 | -43.6324 | -45.4854 |
| 0.1396 | 1.88 | 4500 | 0.1518 | -4.5940 | -10.1635 | 0.9396 | 5.5696 | -365.3205 | -281.2049 | -41.9461 | -43.8060 |
| 0.1575 | 2.08 | 5000 | 0.1525 | -5.3119 | -11.3685 | 0.9292 | 6.0566 | -377.3703 | -288.3840 | -43.4045 | -45.2127 |
| 0.0338 | 2.29 | 5500 | 0.1472 | -5.2545 | -11.3863 | 0.9333 | 6.1319 | -377.5485 | -287.8099 | -43.2283 | -45.1626 |
| 0.1631 | 2.5 | 6000 | 0.1496 | -5.6862 | -11.9852 | 0.9333 | 6.2991 | -383.5375 | -292.1269 | -43.6007 | -45.5693 |
| 0.1177 | 2.71 | 6500 | 0.1473 | -5.6329 | -11.9588 | 0.9417 | 6.3259 | -383.2729 | -291.5939 | -44.3503 | -46.3168 |
| 0.2342 | 2.92 | 7000 | 0.1445 | -5.6156 | -11.9139 | 0.9354 | 6.2982 | -382.8238 | -291.4216 | -44.3728 | -46.3321 |
Framework versions
- Transformers 4.35.2
- Pytorch 2.1.1
- Datasets 2.15.0
- Tokenizers 0.15.0
- Downloads last month
- 12
Inference Providers
NEW
This model is not currently available via any of the supported third-party Inference Providers, and
the model is not deployed on the HF Inference API.
Model tree for Myashka/gpt-imdb-sigmoid-beta_0.1
Base model
lvwerra/gpt2-imdb