rlworkgroup/garage

Split GaussianMLPModel into several variants

Ouverte

#587 ouverte le 19 mars 2019

 (0 commentaire) (0 réaction) (0 personne assignée)Python (297 forks)batch import
good first issuetf

Métriques du dépôt

Stars
 (1 728 étoiles)
Métriques de merge PR
 (Aucune PR mergée en 30 j)

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

Guide contributeur