|
--- |
|
license: other |
|
base_model: facebook/opt-350m |
|
tags: |
|
- trl |
|
- reward-trainer |
|
- generated_from_trainer |
|
metrics: |
|
- accuracy |
|
model-index: |
|
- name: reward_modeling_anthropic_hh |
|
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. --> |
|
|
|
# reward_modeling_anthropic_hh |
|
|
|
This model is a fine-tuned version of [facebook/opt-350m](https://huggingface.co/facebook/opt-350m) on an unknown dataset. |
|
It achieves the following results on the evaluation set: |
|
- Loss: 0.5931 |
|
- Accuracy: 0.6723 |
|
|
|
## 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: 1.41e-05 |
|
- train_batch_size: 16 |
|
- eval_batch_size: 8 |
|
- seed: 42 |
|
- gradient_accumulation_steps: 2 |
|
- total_train_batch_size: 32 |
|
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08 |
|
- lr_scheduler_type: linear |
|
- num_epochs: 1.0 |
|
|
|
### Training results |
|
|
|
| Training Loss | Epoch | Step | Validation Loss | Accuracy | |
|
|:-------------:|:------:|:----:|:---------------:|:--------:| |
|
| 0.6502 | 0.1087 | 500 | 0.6495 | 0.6120 | |
|
| 0.6042 | 0.2174 | 1000 | 0.6465 | 0.6373 | |
|
| 0.6325 | 0.3262 | 1500 | 0.6252 | 0.6438 | |
|
| 0.6129 | 0.4349 | 2000 | 0.6130 | 0.6546 | |
|
| 0.6244 | 0.5436 | 2500 | 0.6117 | 0.6607 | |
|
| 0.6359 | 0.6523 | 3000 | 0.6028 | 0.6686 | |
|
| 0.629 | 0.7610 | 3500 | 0.6015 | 0.6645 | |
|
| 0.6253 | 0.8698 | 4000 | 0.5956 | 0.6712 | |
|
| 0.5928 | 0.9785 | 4500 | 0.5931 | 0.6723 | |
|
|
|
|
|
### Framework versions |
|
|
|
- Transformers 4.40.2 |
|
- Pytorch 2.4.0+cu121 |
|
- Datasets 2.19.1 |
|
- Tokenizers 0.19.1 |
|
|
