Interpretable - Githubissues

riiswa / kanrl

Kolmogorov-Arnold Network for Reinforcement Leaning, initial experiments

268 stars 33 forks source link

Closed corentinlger closed 6 months ago

corentinlger commented 6 months ago

Preliminary file to generate a dataset and record videos from an expert policy (loaded from hugging face)

TODO :

Add num_episodes instead of num_steps to collect dataset
Potentially change the system to map algo_name to sb3 algo (and add more to the current dict)
Update requirements.txt
Add streamlit app