Project Palbot

This is a Github repository to build a RAG based agent for the game called Palbot. The goal of building the chatbot was to have a single chatbot to answer all queries regarding Palbot game, which included pals, items and detail combinations.

Data Collection

For getting relevant data related to Palworld game, following were my data sources.
Pals: https://palworld.fandom.com/wiki/Paldeck
Items: https://palworld.gg/items
Pal Breed: https://www.rockpapershotgun.com/palworld-breeding-combos

Embeddings

For the embeddings, I used the all-MiniLM-L6-v2 model for creating embeddings. For storing the embeddings, I used chromadb and stored the embeddings locally on my drive.

Query Response

I have used the openAI 3.5 turbo instruct model to generate answer from the most similar embeddings.

UI

I've used streamlit for a very simple UI.

How to run

Follow the steps to run the application

1. Create a new python env and install the requirements.txt file
2. run the command 'streamlit run app.py'
3. Now click scrape data to save data locally
4. Click generate embeddings to save embeddings locally
5. Ask queries regarding pals eg: "How to get anubis"