Mini-project
Retail, E-commerce & Consumer Goods
Fine-tuned review summariser
Fine-tune a small model to summarise reviews into pros and cons, and compare it with a large model.
The brief
Shoppers do not read 500 reviews; they want the pros and cons. Fine-tune a small open model to summarise a product's reviews into pros and cons, and compare it with a large hosted model on faithfulness, speed and cost. Good looks like: a training set, a faithfulness check against the reviews, and a recommendation.
What you will build
- Build review sets and reference summaries
- Fine-tune a small model with LoRA
- Summarise with both models
- Check faithfulness with an LLM judge and spot checks
- Compare quality, speed and cost
Free, open data to build it with
- Amazon reviews 2023 ↗ Hugging Face (McAuley Lab)
Check each dataset's licence before you publish your work. Your mentor gives your team its own version of the brief (data, scope or region), so every team's project is its own.