This file will become your README and also the index of your documentation.
pip install -r requirements
pip install -e .
-
Representation function -
$h_{\theta}(o_1, o_2,...o_t)$ : It takes the history of observations and produces a hidden state$s_o$ of the observations -
Prediction function -
$f_{\theta}(s^k)$ : It takes the hidden state of the observation and predicts the policy$\pi^k$ and the value$v^k$ -
Dynamic function -
$g_{\theta}(s^{k-1}, a^k) = r^k, s^k$ : It takes the current state and an action, it predicts the next state and reward
Monte Carlo Tree Search (MCTS: to guide the exploration of the game state space and select the most promising actions
- A replay buffer share between agents
- Monte Carlo Tree Search
Resources that i used to implement MuZero - MuZero - Mastering Atari, Go, Chess and Shogi by Planning with a Learned Model | RL Paper explained
- MCTS
- Recreating DeepMind's AlphaZero - AI Plays Connect 4 - Part 2: Intro to Monte Carlo Tree Search: https://youtu.be/HikhrP5sgQo