---
license: mit
base_model: cardiffnlp/twitter-roberta-base-sentiment-latest
language:
- en
library_name: transformers
tags:
- Roberta
- Sentiment Analysis
widget:
- text: This product is really great!
- text: This product is really bad!
---

# Fine-tuned RoBERTa for Sentiment Analysis on Reviews

This model is a fine-tuned version of [cardiffnlp/twitter-roberta-base-sentiment-latest](https://huggingface.co/cardiffnlp/twitter-roberta-base-sentiment-latest) on the [Amazon Reviews dataset](https://www.kaggle.com/datasets/bittlingmayer/amazonreviews) for sentiment analysis.

## Model Details

- **Model Name:** `AnkitAI/reviews-roberta-base-sentiment-analysis`
- **Base Model:** `cardiffnlp/twitter-roberta-base-sentiment-latest`
- **Dataset:** [Amazon Reviews](https://www.kaggle.com/datasets/bittlingmayer/amazonreviews)
- **Fine-tuning:** This model was fine-tuned for sentiment analysis with a classification head for binary sentiment classification (positive and negative).

## Training

The model was trained using the following parameters:

- **Learning Rate:** 2e-5
- **Batch Size:** 16
- **Weight Decay:** 0.01
- **Evaluation Strategy:** Epoch

### Training Details

- **Evaluation Loss:** 0.1049
- **Evaluation Runtime:** 3177.538 seconds
- **Evaluation Samples/Second:** 226.591
- **Evaluation Steps/Second:** 7.081
- **Training Runtime:** 110070.6349 seconds
- **Training Samples/Second:** 78.495
- **Training Steps/Second:** 2.453
- **Training Loss:** 0.0858
- **Evaluation Accuracy:** 97.19%
- **Evaluation Precision:** 97.9%
- **Evaluation Recall:** 97.18%
- **Evaluation F1 Score:** 97.19%

## Usage

You can use this model directly with the Hugging Face `transformers` library:

```python
from transformers import RobertaForSequenceClassification, RobertaTokenizer

model_name = "AnkitAI/reviews-roberta-base-sentiment-analysis"
model = RobertaForSequenceClassification.from_pretrained(model_name)
tokenizer = RobertaTokenizer.from_pretrained(model_name)

# Example usage
inputs = tokenizer("This product is great!", return_tensors="pt")
outputs = model(**inputs) # 1 for positive, 0 for negative
```

## License

This model is licensed under the [MIT License](LICENSE).