Reconchess-CS748

Chess Game Replay and Analysis

This repository includes scripts for replaying chess games, evaluating move scores, and analyzing performance against different bots. Additionally, Jupyter Notebooks are provided for calculating win percentages and training the weights of the lc0 network.

Usage

1. Activate Virtual Environment:

   conda activate <path-to-your-virtualenv>

2. Replace Files:

   Copy and replace the following files in your project with their buffer counterparts:

   • fianchetto_bot.py
   • strangefish_bot.py
   • multiprocessing_strategies.py

3. Run Scripts:

   • To generate a replay buffer:

     python scripts/buffer.py <Path to the game you want to replay>

     For example:

     python scripts/buffer.py /home/puranjay/Fianchetto_IIT_Bombay/Fianchetto_fast/655504.json

   • To replay a game and display move scores and top board probabilities:

     python scripts/buffer_replay.py <Path to game you want to replay>

     For example:

     python scripts/buffer_replay.py /home/puranjay/Fianchetto_IIT_Bombay/Fianchetto_fast/655504.json

Jupyter Notebooks

• Winrate_Fianchetto.ipynb: Use this notebook to find the win percentage against various bots on the RBC server.

• NewOpponentModel.ipynb: This notebook was used to train the weights of the lc0 network.