|
|
--- |
|
|
library_name: transformers |
|
|
tags: |
|
|
- text2text-generation |
|
|
- summarization |
|
|
- product-review |
|
|
- sentiment-analysis |
|
|
- t5-small |
|
|
- huggingface |
|
|
--- |
|
|
|
|
|
# 📝 Review Summariser GPT - Config1 |
|
|
|
|
|
A fine-tuned `t5-small` model that generates concise summaries from product reviews. This model is part of the ReviewSummariserGPT project and also pairs well with a Hugging Face sentiment analysis pipeline for classifying tone. |
|
|
|
|
|
--- |
|
|
|
|
|
## Model Details |
|
|
|
|
|
### Model Description |
|
|
|
|
|
This is a sequence-to-sequence Transformer model based on `t5-small`, fine-tuned on a dataset of 1,000 product review → summary pairs. It is designed to take in a review (e.g., from Amazon or Yelp) and output a short, helpful summary. |
|
|
|
|
|
- **Developed by:** Manish Kumar Kondoju |
|
|
- **Finetuned from model:** [`t5-small`](https://huggingface.co/t5-small) |
|
|
- **Language(s) (NLP):** English |
|
|
- **License:** Apache 2.0 |
|
|
- **Model type:** Seq2Seq Transformer (Text-to-Text Generation) |
|
|
- **Shared by:** Manish014 |
|
|
|
|
|
### Model Sources |
|
|
|
|
|
- **Repository:** [https://huggingface.co/Manish014/review-summariser-gpt-config1](https://huggingface.co/Manish014/review-summariser-gpt-config1) |
|
|
- **Demo:** Coming soon via Gradio Hugging Face Space |
|
|
|
|
|
--- |
|
|
|
|
|
## Uses |
|
|
|
|
|
### Direct Use |
|
|
|
|
|
- Generate summaries for user reviews to improve content digestibility. |
|
|
- Enhance e-commerce UX by auto-summarizing reviews. |
|
|
|
|
|
```python |
|
|
from transformers import AutoTokenizer, AutoModelForSeq2SeqLM |
|
|
|
|
|
tokenizer = AutoTokenizer.from_pretrained("Manish014/review-summariser-gpt-config1") |
|
|
model = AutoModelForSeq2SeqLM.from_pretrained("Manish014/review-summariser-gpt-config1") |
|
|
|
|
|
input_text = "summarize: The build quality is terrible and the support team was unhelpful." |
|
|
inputs = tokenizer(input_text, return_tensors="pt", truncation=True) |
|
|
outputs = model.generate(inputs["input_ids"], max_length=60) |
|
|
summary = tokenizer.decode(outputs[0], skip_special_tokens=True) |
|
|
|
