We've talked a lot about it in the group chat so I figured I'd add this issue so that we don't forget about it. Obviously it's a feature we would add after we're done with our thesis.
As in the title: allow for different epsilon_decay values for different tasks. This could especially useful when used in tandem with exploration_reset_value.
We can discuss the exact implementation details later.
We've talked a lot about it in the group chat so I figured I'd add this issue so that we don't forget about it. Obviously it's a feature we would add after we're done with our thesis.
As in the title: allow for different
epsilon_decay
values for different tasks. This could especially useful when used in tandem withexploration_reset_value
.We can discuss the exact implementation details later.