issues
search
cogment
/
cogment-verse
Research platform for Human-in-the-loop learning (HILL) & Multi-Agent Reinforcement Learning (MARL)
https://cogment.ai/cogment_verse
Apache License 2.0
76
stars
14
forks
source link
Make behaviour cloning pong work
#148
Open
cloderic
opened
1 year ago
cloderic
commented
1 year ago
Find / Train an AI pong model (e.g.
https://huggingface.co/sb3
)
Maybe the opportunity to have an "load from hugging face" agent (e.g. script to load a model from hugging face hub or agent that is able to run a model from hugging face hub).
Implement online training / data collection runner
Fully headless or using human actor)
"proper" way to save/import dataset
Implement offline training
LailaElMoujtahid
commented
1 year ago
@wduguay-air --> any status