search

carsten-wenderdel / wildbg

Backgammon engine based on neural networks
https://wildbg.shuttleapp.rs/swagger-ui/
Apache License 2.0
18 stars 5 forks source link

Learning rate as parameter #16

Closed carsten-wenderdel closed 8 months ago

carsten-wenderdel commented 11 months ago

The Python script should be called with the learning rate as parameter. This learning rate should also be stored in the file name of the resulting network. By doing this we can train several nets in parallel and then take the best one.

carsten-wenderdel commented 8 months ago

I close this issue as it's not that much needed anymore, since now many onnx files are generated during one learning session. See https://github.com/carsten-wenderdel/wildbg/commit/07bad18b6d43468680b168d2ac9a4b4d164d3d6e and related to that also https://github.com/carsten-wenderdel/wildbg/commit/7781a74d70be8759ec0788daf7cdb182198f0509