Skip to content

Latest commit

 

History

History
40 lines (30 loc) · 1.4 KB

File metadata and controls

40 lines (30 loc) · 1.4 KB

Fashion RecSys

Hybrid fashion recommendation system — combines collaborative filtering with LLM-based reranking to deliver personalized outfit suggestions that understand style context.

Results

Model Precision@10 Recall@10 NDCG@10
Baseline CF 0.24 0.31 0.28
CF + LLM Rerank 0.31 0.38 0.35
Improvement +29% +23% +25%

Quick Start

git clone https://github.com/Jkanishkha0305/fashion-recsys.git
cd fashion-recsys
pip install -r requirements.txt

# Train collaborative filtering model
python train.py --model cf --epochs 20

# Launch recommendation UI
streamlit run app.py

Tech Stack

  • Collaborative Filtering — matrix factorization baseline
  • LLM Reranking — GPT-4 reranks CF candidates using style context
  • PyTorch — model training
  • Streamlit — interactive recommendation UI