|
--- |
|
base_model: lvwerra/gpt2-imdb |
|
tags: |
|
- generated_from_trainer |
|
model-index: |
|
- name: gpt-imdb-hinge-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-hinge-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: 0.1683 |
|
- Rewards/chosen: -2.7548 |
|
- Rewards/rejected: -6.4029 |
|
- Rewards/accuracies: 0.9271 |
|
- Rewards/margins: 3.6481 |
|
- Logps/rejected: -327.7139 |
|
- Logps/chosen: -262.8134 |
|
- Logits/rejected: -45.7026 |
|
- Logits/chosen: -46.9404 |
|
|
|
## 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 | |
|
|:-------------:|:-----:|:----:|:---------------:|:--------------:|:----------------:|:------------------:|:---------------:|:--------------:|:------------:|:---------------:|:-------------:| |
|
| 0.3746 | 0.21 | 500 | 0.3940 | -0.4768 | -1.9553 | 0.8562 | 1.4785 | -283.2387 | -240.0334 | -33.1236 | -34.2065 | |
|
| 0.3627 | 0.42 | 1000 | 0.3395 | -1.0759 | -2.9896 | 0.8646 | 1.9137 | -293.5812 | -246.0238 | -41.8545 | -42.9940 | |
|
| 0.2687 | 0.63 | 1500 | 0.3229 | -1.7235 | -4.1025 | 0.8729 | 2.3790 | -304.7103 | -252.5004 | -39.8423 | -41.2043 | |
|
| 0.1878 | 0.83 | 2000 | 0.2360 | -1.6708 | -4.3940 | 0.9104 | 2.7231 | -307.6249 | -251.9736 | -41.4970 | -42.6933 | |
|
| 0.1936 | 1.04 | 2500 | 0.2124 | -1.9623 | -4.8688 | 0.9250 | 2.9066 | -312.3736 | -254.8880 | -42.8807 | -43.9675 | |
|
| 0.2302 | 1.25 | 3000 | 0.2062 | -2.1959 | -5.2559 | 0.9021 | 3.0600 | -316.2442 | -257.2241 | -45.2090 | -46.3997 | |
|
| 0.2137 | 1.46 | 3500 | 0.2235 | -2.1054 | -5.4204 | 0.9208 | 3.3150 | -317.8889 | -256.3190 | -46.5366 | -47.7024 | |
|
| 0.2231 | 1.67 | 4000 | 0.1884 | -2.3281 | -5.6096 | 0.9208 | 3.2815 | -319.7815 | -258.5467 | -45.7720 | -46.8600 | |
|
| 0.2269 | 1.88 | 4500 | 0.1785 | -2.5145 | -6.0015 | 0.9292 | 3.4871 | -323.7006 | -260.4101 | -45.7220 | -46.8746 | |
|
| 0.1831 | 2.08 | 5000 | 0.1727 | -2.6850 | -6.2801 | 0.9312 | 3.5951 | -326.4862 | -262.1152 | -45.0514 | -46.1610 | |
|
| 0.0112 | 2.29 | 5500 | 0.1682 | -2.5613 | -6.0913 | 0.9312 | 3.5300 | -324.5987 | -260.8782 | -45.3410 | -46.5522 | |
|
| 0.1894 | 2.5 | 6000 | 0.1706 | -2.7334 | -6.3632 | 0.9271 | 3.6298 | -327.3174 | -262.5995 | -45.2020 | -46.4449 | |
|
| 0.13 | 2.71 | 6500 | 0.1685 | -2.7681 | -6.4203 | 0.9250 | 3.6522 | -327.8886 | -262.9462 | -45.5580 | -46.8017 | |
|
| 0.2717 | 2.92 | 7000 | 0.1683 | -2.7548 | -6.4029 | 0.9271 | 3.6481 | -327.7139 | -262.8134 | -45.7026 | -46.9404 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.35.2 |
|
- Pytorch 2.1.1 |
|
- Datasets 2.15.0 |
|
- Tokenizers 0.15.0 |
|
|
