saymrwulf/stable-baselines3
1
0
Fork 0
mirror of https://github.com/saymrwulf/stable-baselines3.git synced 2026-07-24 19:43:50 +00:00
Code Issues Projects Releases Packages Wiki Activity Actions

Fix infs in SAC by bounding the mean

Browse source
This commit is contained in:
Antonin Raffin 2019-12-18 13:45:33 +01:00
parent 57708a628c
commit e49d97bf98
1 changed files with 2 additions and 0 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
torchy_baselines/sac/policies.py
View file

@ -48,6 +48,8 @@ class Actor(BaseNetwork):
self.mu, self.log_std = self.action_dist.proba_distribution_net(latent_dim=net_arch[-1],
latent_sde_dim=latent_sde_dim,
log_std_init=log_std_init)
# Avoid saturation by limiting the mean of the gaussian to be in [-1, 1]
self.mu = nn.Sequential(self.mu, nn.Tanh())
else:
self.action_dist = SquashedDiagGaussianDistribution(action_dim)
self.mu = nn.Linear(net_arch[-1], action_dim)