Skip to content

Instantly share code, notes, and snippets.

@elumixor

elumixor/TRPO_max_step.py

Last active May 24, 2020 23:30
Show Gist options
  • Save elumixor/bf29c25a20972b45ffc3c73a1f96235b to your computer and use it in GitHub Desktop.
Save elumixor/bf29c25a20972b45ffc3c73a1f96235b to your computer and use it in GitHub Desktop.
Download ZIP
Raw
TRPO_max_step.py
delta = 0.01. # Should be low (approximately betwween 0.01 and 0.05
max_length = torch.sqrt(2 * delta / (search_dir @ HVP(search_dir)))
max_step = max_length * search_dir
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment