Hyperparameters for expert trajectories

SaminYeasar / Off_Policy_Adversarial_Inverse_Reinforcement_Learning

Implementation of Off Policy Adversarial Inverse Reinforcement Learning

MIT License

20 stars 2 forks source link

Hyperparameters for expert trajectories #3

Closed sff1019 closed 3 years ago

sff1019 commented 3 years ago

The previous issue #2 seems to be closed, hence opening a new one. I saw the link for the expert trajectories, but these does not include the hyperparameters or training script. Will it be possible for you to share these as well? E.g. lr, noise, batch_size, etc.

Thank you,

SaminYeasar commented 3 years ago

Hi, this implementation of soft-actor-critic (https://github.com/denisyarats/pytorch_sac) (with default hyper-parameter) should work fine.

sff1019 commented 3 years ago

Yes, it does seem work! Thank you!