search

RobertTLange / gymnax

RL Environments in JAX 🌍
Apache License 2.0
585 stars 54 forks source link

Fix inconsistent dtype for reward in MemoryChain environment #24

Closed smonsays closed 2 years ago

smonsays commented 2 years ago

Hey Robert, I tried to jax.lax.scan over the bsuite-MemoryChain environment and noticed that the reward returned by this env is an int32 instead of a float32 as for the other environments. Was this on purpose? If not here would be the minimal fix.

Best, Simon

RobertTLange commented 2 years ago

Thank you Simon!