hey yall, I made this as way to understand what's going on in RL algorithms (REINFORCE, A2C, PPO). It's built on TensorflowJS (although I'd love to move it to JS-Pytorch). It uses your browser's GPU to train live.
I found it helpful to have a "studio" experience like this to get realtime feedback as I was doing a hyperparameter search. There's a lot of parameters to tweak in RL.
Web ML is cool and I hope to see the ecosystem grow more
ultrafro2•1h ago
I found it helpful to have a "studio" experience like this to get realtime feedback as I was doing a hyperparameter search. There's a lot of parameters to tweak in RL.
Web ML is cool and I hope to see the ecosystem grow more