instadeepai / Mava

🦁 A research-friendly codebase for fast experimentation of multi-agent reinforcement learning in JAX
Apache License 2.0
719 stars 85 forks source link

[FEATURE] piecewise linear epsilon for exploration #473

Closed nima-siboni closed 2 years ago

nima-siboni commented 2 years ago

Feature

A piecewise linear scheduler for epsilon. With piecewise linear scheduler the user can increase and decrease the epsilon over the desired time intervals.

Proposal

Creating a new class for similar to LinearTimestepScheduler

Testing

The returned value of the epsilon should be tested:

Definition of done

A class is implemented and all the tests are passed.

DriesSmit commented 2 years ago

Closing all TF issues as we are depreciating our TF systems.