A fine-tuned embedding model for recommending anime from user query.
This project is heavily inspired from manga-recommendation by @jjentaa
Dataset by Md Sazzad Siddique (Version 1)
You can try the web application for the model here
- Using
mistralai/Mistral-7B-Instruct-v0.2for generated query and use it for fine tuning pre-trained model - I use
paraphrase-multilingual-mpnet-base-v2as a base model + LoRA adapter, fine-tuning with anime synopsis data + query
KiruruP/anime-recommendation-multilingual-mpnet-base-v2-peft: Multilingual Fine-tuned model
- Clone this repository
- Install the necessary library follwing the list below
sentence_transformers == 5.1.0
numpy == 2.1.3
torch == 2.7.1
pandas == 2.2.3
streamlit == 1.48.1
peft
or you can install all-in-one in terminal through command
pip install -r requirements.txt
- run this command in terminal, it will open in your browser and enjoy!! :)
streamlit run app.py