File size: 4,442 Bytes
2c5674f 557861d 2c5674f 557861d 2c5674f |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 |
---
base_model: lvwerra/gpt2-imdb
tags:
- generated_from_trainer
model-index:
- name: gpt-imdb-kto-beta_0.1
results: []
---
<!-- This model card has been generated automatically according to the information the Trainer had access to. You
should probably proofread and complete it, then remove this comment. -->
# gpt-imdb-kto-beta_0.1
This model is a fine-tuned version of [lvwerra/gpt2-imdb](https://huggingface.co/lvwerra/gpt2-imdb) on an unknown dataset.
It achieves the following results on the evaluation set:
- Loss: 0.2062
- Rewards/chosen: 2.5179
- Rewards/rejected: -0.0433
- Rewards/accuracies: 0.8250
- Rewards/margins: 2.5611
- Logps/rejected: -264.1180
- Logps/chosen: -210.0866
- Logits/rejected: -30.4371
- Logits/chosen: -31.3849
## 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
- training_steps: 7197
### Training results
| Training Loss | Epoch | Step | Validation Loss | Rewards/chosen | Rewards/rejected | Rewards/accuracies | Rewards/margins | Logps/rejected | Logps/chosen | Logits/rejected | Logits/chosen |
|:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:|
| 0.2522 | 0.21 | 500 | 0.2884 | 1.2801 | -0.2634 | 0.7875 | 1.5434 | -266.3188 | -222.4644 | -37.5496 | -38.4713 |
| 0.3335 | 0.42 | 1000 | 0.2696 | 1.5869 | -0.1616 | 0.7917 | 1.7485 | -265.3008 | -219.3961 | -37.8624 | -38.7817 |
| 0.2435 | 0.63 | 1500 | 0.2472 | 1.8228 | -0.2033 | 0.7896 | 2.0260 | -265.7180 | -217.0376 | -33.3680 | -34.1467 |
| 0.3162 | 0.83 | 2000 | 0.2497 | 2.2013 | 0.3606 | 0.7729 | 1.8407 | -260.0789 | -213.2520 | -33.0705 | -33.7146 |
| 0.1409 | 1.04 | 2500 | 0.2301 | 2.0789 | -0.0950 | 0.8042 | 2.1738 | -264.6351 | -214.4766 | -34.2110 | -35.1256 |
| 0.2415 | 1.25 | 3000 | 0.2221 | 2.1406 | -0.2423 | 0.8042 | 2.3829 | -266.1087 | -213.8594 | -35.0880 | -35.8295 |
| 0.1549 | 1.46 | 3500 | 0.2173 | 2.2945 | -0.0445 | 0.7979 | 2.3390 | -264.1307 | -212.3203 | -31.2025 | -32.0702 |
| 0.1764 | 1.67 | 4000 | 0.2117 | 2.3347 | -0.2551 | 0.8250 | 2.5898 | -266.2365 | -211.9187 | -31.0530 | -31.9754 |
| 0.131 | 1.88 | 4500 | 0.2101 | 2.3080 | -0.3171 | 0.8062 | 2.6251 | -266.8560 | -212.1852 | -30.9535 | -31.9058 |
| 0.2463 | 2.08 | 5000 | 0.2131 | 2.5808 | 0.2215 | 0.8167 | 2.3593 | -261.4699 | -209.4572 | -31.7099 | -32.5262 |
| 0.1536 | 2.29 | 5500 | 0.2084 | 2.5201 | -0.0034 | 0.8125 | 2.5236 | -263.7196 | -210.0640 | -30.3275 | -31.2806 |
| 0.2473 | 2.5 | 6000 | 0.2057 | 2.4813 | -0.1087 | 0.8188 | 2.5899 | -264.7721 | -210.4527 | -30.2259 | -31.1935 |
| 0.2168 | 2.71 | 6500 | 0.2060 | 2.5255 | -0.0304 | 0.8146 | 2.5559 | -263.9893 | -210.0102 | -30.4678 | -31.4146 |
| 0.1669 | 2.92 | 7000 | 0.2062 | 2.5179 | -0.0433 | 0.8250 | 2.5611 | -264.1180 | -210.0866 | -30.4371 | -31.3849 |
### Framework versions
- Transformers 4.35.2
- Pytorch 2.1.1
- Datasets 2.15.0
- Tokenizers 0.15.0
|