---
license: gemma
library_name: peft
tags:
- trl
- reward-trainer
- generated_from_trainer
metrics:
- accuracy
base_model: google/gemma-2b
model-index:
- name: RM-HH-Gemma_harmless_gpt3_20000_gemma2b_shuffleTrue_extractchosenTrue
  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. -->

# RM-HH-Gemma_harmless_gpt3_20000_gemma2b_shuffleTrue_extractchosenTrue

This model is a fine-tuned version of [google/gemma-2b](https://huggingface.co/google/gemma-2b) on the None dataset.
It achieves the following results on the evaluation set:
- Loss: 0.3493
- Accuracy: 0.8350

## 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: 1
- eval_batch_size: 8
- seed: 42
- gradient_accumulation_steps: 4
- total_train_batch_size: 4
- 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 |
|:-------------:|:-----:|:----:|:---------------:|:--------:|
| 0.6993        | 0.03  | 250  | 0.6134          | 0.6557   |
| 0.5635        | 0.06  | 500  | 0.4914          | 0.7369   |
| 0.4753        | 0.08  | 750  | 0.4386          | 0.7647   |
| 0.4581        | 0.11  | 1000 | 0.4201          | 0.7794   |
| 0.4055        | 0.14  | 1250 | 0.4168          | 0.7879   |
| 0.4121        | 0.17  | 1500 | 0.4093          | 0.7922   |
| 0.388         | 0.19  | 1750 | 0.4091          | 0.7932   |
| 0.4249        | 0.22  | 2000 | 0.3978          | 0.8015   |
| 0.4087        | 0.25  | 2250 | 0.3929          | 0.8015   |
| 0.4016        | 0.28  | 2500 | 0.3915          | 0.8045   |
| 0.4309        | 0.31  | 2750 | 0.3702          | 0.8105   |
| 0.4258        | 0.33  | 3000 | 0.3625          | 0.8150   |
| 0.427         | 0.36  | 3250 | 0.3671          | 0.8137   |
| 0.3798        | 0.39  | 3500 | 0.3791          | 0.8132   |
| 0.3759        | 0.42  | 3750 | 0.3685          | 0.8152   |
| 0.4008        | 0.44  | 4000 | 0.3601          | 0.8192   |
| 0.3901        | 0.47  | 4250 | 0.3593          | 0.8220   |
| 0.3791        | 0.5   | 4500 | 0.3608          | 0.8235   |
| 0.3801        | 0.53  | 4750 | 0.3620          | 0.8225   |
| 0.3726        | 0.56  | 5000 | 0.3678          | 0.8225   |
| 0.4122        | 0.58  | 5250 | 0.3654          | 0.8220   |
| 0.363         | 0.61  | 5500 | 0.3647          | 0.8245   |
| 0.3808        | 0.64  | 5750 | 0.3569          | 0.8287   |
| 0.3977        | 0.67  | 6000 | 0.3534          | 0.8295   |
| 0.3492        | 0.69  | 6250 | 0.3551          | 0.8307   |
| 0.4155        | 0.72  | 6500 | 0.3462          | 0.8315   |
| 0.3879        | 0.75  | 6750 | 0.3485          | 0.8322   |
| 0.349         | 0.78  | 7000 | 0.3507          | 0.8312   |
| 0.4138        | 0.81  | 7250 | 0.3465          | 0.8352   |
| 0.3483        | 0.83  | 7500 | 0.3471          | 0.8350   |
| 0.3652        | 0.86  | 7750 | 0.3482          | 0.8355   |
| 0.3899        | 0.89  | 8000 | 0.3468          | 0.8345   |
| 0.3793        | 0.92  | 8250 | 0.3466          | 0.8352   |
| 0.3815        | 0.94  | 8500 | 0.3476          | 0.8352   |
| 0.3371        | 0.97  | 8750 | 0.3493          | 0.8350   |


### Framework versions

- PEFT 0.9.0
- Transformers 4.38.2
- Pytorch 2.1.2
- Datasets 2.18.0
- Tokenizers 0.15.2