rlworkgroup/garage

Split GaussianMLPModel into several variants

Open

#587 opened on Mar 19, 2019

View on GitHub
 (0 comments) (0 reactions) (0 assignees)Python (1,728 stars) (297 forks)batch import
good first issuetf

Description

The current GaussianMLPModel actually represents several architecture variants, which makes it difficult to use and to test. We should split it into several variants based on architecture (and possibly parameterization)

description mean std_param
fixed variance MLP1 parameter (trainable or not)
split network MLP1 MLP2
shared network/multi-headed MLP1/head1 MLP1/head2

Contributor guide

Tech stack
python
Domain
backend
Issue type
refactor
DifficultyEstimated implementation difficulty for a new contributor, from 1 for very small changes to 5 for expert-level work.
3
Estimated timeA rough time range for an experienced contributor to investigate, implement, test, and prepare a pull request.
1-2 days
Activity statusHow available the issue appears right now: fresh, active, stale, blocked, or waiting on maintainer input.
stale
ClarityHow clearly the issue explains the expected change, acceptance criteria, and next step.
mostly clear
Prerequisites
Pythonreinforcement learning fundamentals
Newbie friendlinessA 1-100 score estimating how approachable this issue is for first-time contributors.
40
Research direction
Examine the current GaussianMLPModel class in the garage codebase to understand its architecture and parameterization. Then, propose splitting into three variants as outlined in the table: fixed variance, split network, and shared network/multi headed. Consider creating separate classes or configuration options. Ensure backward compatibility or update tests accordingly.