search

chainer / chainerrl

ChainerRL is a deep reinforcement learning library built on top of Chainer.
MIT License
1.18k stars 224 forks source link

Bug in Soft Actor Critic's batch_observe_and_train #557

Closed ummavi closed 5 years ago

ummavi commented 5 years ago

env_id is not passed as a parameter when transitions are added to the replay buffer in Soft Actor Critic.

https://github.com/chainer/chainerrl/blob/8bee9da793bffa352e1d2c2fba16b4f148ba0253/chainerrl/agents/soft_actor_critic.py#L377

prabhatnagarajan commented 5 years ago

Look at this PR as a reference: https://github.com/chainer/chainerrl/pull/442/files

ummavi commented 5 years ago

Resolved by #558