2048 game simulation for reinforcement learning on decision theory course.
2048
-
Moves: slide to left/r/u/d
-
End state: all cells are full and there are no possibles moves
-
OR reaching 2048 in one cell
-
Random Agent: Generates blocks (power of 2)
-
AI agent: do the valid moves
-
Rewards: the created number out after slide
-
Start with 3x3 and target 64
- get_all_possible_actions
- get_all_possible_states
- _calculate_transition_prob