(Change Default Outputs) feature: change default learning rate to 3e-6 * sqrt(batch_size)

DeNA / HandyRL

HandyRL is a handy and simple framework based on Python and PyTorch for distributed reinforcement learning that is applicable to your own environments.

MIT License

282 stars 39 forks source link

(Change Default Outputs) feature: change default learning rate to 3e-6 * sqrt(batch_size) #321

Open YuriCat opened 2 years ago

YuriCat commented 2 years ago

The learning rate proportional to batch size looks strange.