bupticybee / alphanlholdem Goto Github PK

Hi! icybee,
The file agi.mis in train_league.py might be missing.

Thanks for the great work!

Would you mind sharing more details about the issues in the RLCard environment? We want to fix them.

Thanks for your wonderful project. But I wonder why Trinal-Clip PPO in Alphaholdem is not used.

ERROR: Could not find a version that satisfies the requirement tensorflow==1.15.2 (from versions: 2.8.0rc0, 2.8.0rc1, 2.8.0, 2.8.1, 2.8.2, 2.8.3, 2.8.4, 2.9.0rc0, 2.9.0rc1, 2.9.0rc2, 2.9.0, 2.9.1, 2.9.2, 2.9.3, 2.10.0rc0, 2.10.0rc1, 2.10.0rc2, 2.10.0rc3, 2.10.0, 2.10.1, 2.11.0rc0, 2.11.0rc1, 2.11.0rc2, 2.11.0, 2.11.1, 2.12.0rc0, 2.12.0rc1, 2.12.0, 2.12.1, 2.13.0rc0, 2.13.0rc1, 2.13.0rc2, 2.13.0, 2.13.1, 2.14.0rc0, 2.14.0rc1, 2.14.0)
ERROR: No matching distribution found for tensorflow==1.15.2

This was apparently written in 2023 but is using a version of ray released in early 2020. Not sure why that is but it makes it impossible to use because the ray 0.8.3 branch does not contain tune. The only version of ray[tune] that is available is for the current releases and not the legacy releases.

Is there any way to make this work on any of the Ray releases that are available for install using pip? The furthest version that I can go back to is 0.8.6, but there seems to be a lot of unsupported code when I tried that version.

Is 0.8.3 a must? If so, are there any sources for it still available that I can compile for my machine?

Thanks in advance!

Even after ~ 1 billion self-play, over 1000 checkpoints, the model seems still not converge

Thanks for your wonderful project. May I ask how do you judge the convergency of a self-play game model? Are there any evaluation metrics recommended?

I don't think evaluating with rewards (or utilities) is a good choice, because it is continuously self-improving on both sides at the training stage.