higgsfield-ai / higgsfield

Fault-tolerant, highly scalable GPU orchestration, and a machine learning framework designed for training models with billions to trillions of parameters
Apache License 2.0
3.29k stars 553 forks source link

Update hyperparameters of PPO and GAIL #31

Closed chungshan closed 1 year ago

chungshan commented 3 years ago

Hi, I update hyperparameters of PPO and GAIL to have a more stable results according to PPO implementation #9

PPO: image

GAIL: image