---
base_model: openai-community/gpt2
library_name: peft
license: mit
metrics:
- accuracy
tags:
- generated_from_trainer
model-index:
- name: GPT2-small-QLoRA-finetuned-amazon-reviews-en-classification
  results: []
datasets:
- mteb/amazon_reviews_multi
language:
- en
widget:
- text: It`s an amazing product
- text: I hate this product
- text: It's ok, but a bit expensive
pipeline_tag: text-classification
---

<!-- 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. -->

# GPT2-small-QLoRA-finetuned-amazon-reviews-en-classification

This model is a fine-tuned version of [openai-community/gpt2](https://huggingface.co/openai-community/gpt2) on [mteb/amazon_reviews_multi](https://huggingface.co/datasets/mteb/amazon_reviews_multi) dataset.

It is the result of the post [Fine tunning SML](https://maximofn.com/qlora/)

It achieves the following results on the evaluation set:
- Loss: 0.8883
- Accuracy: 0.615

## Model description

This model provides classification of reviews in english

## 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: 2e-05
- train_batch_size: 224
- eval_batch_size: 224
- seed: 42
- optimizer: Adam with betas=(0.9,0.999) and epsilon=1e-08
- lr_scheduler_type: cosine
- lr_scheduler_warmup_ratio: 0.1
- num_epochs: 3
- mixed_precision_training: Native AMP

### Training results

| Training Loss | Epoch | Step | Validation Loss | Accuracy |
|:-------------:|:-----:|:----:|:---------------:|:--------:|
| 2.088         | 1.0   | 893  | 0.9780          | 0.5848   |
| 0.9588        | 2.0   | 1786 | 0.8940          | 0.6156   |
| 0.9147        | 3.0   | 2679 | 0.8918          | 0.6168   |


### Framework versions

- PEFT 0.12.0
- Transformers 4.43.3
- Pytorch 2.3.1+cu121
- Datasets 2.20.0
- Tokenizers 0.19.1