Skip to content

Fix 3DBall PPO hard regression #4133

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
merged 10 commits into from
Jul 7, 2020
Merged

Fix 3DBall PPO hard regression #4133

Changes from all commits
Commits
Show all changes
10 commits
Select commit Hold shift + click to select a range
a4d81d3
update config 3dball
andrewcoh Jun 17, 2020
e5bba80
zeroing out beta ppo 3dballhard
andrewcoh Jun 17, 2020
07ad85a
reduce batch size
andrewcoh Jun 17, 2020
d180744
Merge branch 'master' into fix-3dball-regr
andrewcoh Jun 17, 2020
2c113f4
3dball hard ppo 1M timesteps
andrewcoh Jun 18, 2020
5eb188b
revert 1M timestep commit
andrewcoh Jun 18, 2020
e13c4da
reduce gamma
andrewcoh Jun 18, 2020
0f38c36
learning rate to constant
andrewcoh Jun 20, 2020
3b41c2e
reduce steps
andrewcoh Jun 20, 2020
fc59204
constant to linear LR
andrewcoh Jun 21, 2020
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
6 changes: 3 additions & 3 deletions config/ppo/3DBallHard.yaml
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -2,7 +2,7 @@ behaviors:
3DBallHard:
trainer_type: ppo
hyperparameters:
batch_size: 1200
batch_size: 120
buffer_size: 12000
learning_rate: 0.0003
beta: 0.001
Expand All @@ -17,10 +17,10 @@ behaviors:
vis_encode_type: simple
reward_signals:
extrinsic:
gamma: 0.995
gamma: 0.99
strength: 1.0
keep_checkpoints: 5
max_steps: 5000000
max_steps: 500000
time_horizon: 1000
summary_freq: 12000
threaded: true