Adds downfall environment

edbeeching / godot_rl_agents_examples

Example Environments for the Godot RL Agents library

MIT License

37 stars 13 forks source link

Closed edbeeching closed 6 months ago

edbeeching commented 6 months ago

Remaining todos:

[x] add onnx model
[x] add video
[x] add hyperparameters for training (default with 2M steps with 4 parallel envs)
[x] add example to hub and to tests on the godot-rl main repo https://github.com/edbeeching/godot_rl_agents/pull/176

Ivan-267 commented 6 months ago

Nicely made.

I've made a quick test (not full training), it seems to be working OK and is starting to learn.

I'll add some comments just as notes, the env is working fine as is so no alterations are necessary.

Ivan-267 commented 6 months ago

I just retried with the onnx and got an inference error. Solved by renaming the .csproj and .sln files to the name of the project.

It looks good and the agents are able to solve the levels.

edbeeching commented 6 months ago

@Ivan-267 ah thanks. Yes, I renamed these locally. I need to push the new files.