Dependencies
- numpy
- opencv-python
- tensorflow=1.13.1
Environments
A training process can be started as follows with appropriate argument(s):
python main.py --experiment-setup <experiment setup>
The list of all hyperparameters can be found in main.py.
Format of experiment-setup:
This an integer argument with three digits abc (except for no advising which takes 0) which defines the setup of the experiment in terms of action advising method, replay memory regulation method, and action advising budget to be used.
- a: Action advising method
- 1: Early advising
- 2: Uniformly random advising
- b: Replay memory regulation method
- 0: None
- 1: Counter
- 2: RND
- c: Budget (these are defined in
executor.py)- 0: 500
- 1: 1000
- 2: 2500
- 3: 5000
- 4: 10000
- 5: 25000
- 6: 50000
- 7: 100000
Experiment setups used in the study:
-
LavaWorld
- No Advising: 0
- Early Advising: 101, 104, 105, 111, 114, 115, 121, 124, 125
- Uniformly Random: 201, 204, 205, 211, 214, 215, 221, 224, 225
-
MinAtar (for each game)
- No Advising: 0
- Early Advising: 104, 105, 107, 124, 125, 127
- Uniformly Random: 204, 205, 207, 224, 225, 227