Pacman with A2C and deep q Learning

My pacman agent will promate high level of skills with the next updates in the future. Right now i am training my pacman agent, trying to teach new skill for it. In the future i will use rl with human fedback to make observations for my agent's learning path for the sake of curiosity

Current: Level 1

level1

for this level, agent just responsible for finding paths without obstacles to eat pellets with small 5*5 grid.

level2

for this level, agent just responsible for finding paths without obstacles to eat pellets with 10*10 grid.

level3

for this level, agent just responsible fpr finding path with small obstacles to eat pellets with 10*10 grid.