CI for Reinforcement Learning

[nodchip]

Currently the CI for compilation and Supervised Learning were setup. We could also need the CI for Reinforcement Learning. i.e. CI for training data generation with an existing net file, and training with the training from an existing net file.

[vondele]

this should be easy to add in tests/instrumented_learn.sh, we just need to generate a variant of learn01.exp ? Is the only change needed to set 'SkipLoadingEval false' ?

[nodchip]

Is the only change needed to set 'SkipLoadingEval false' ?

In training data generation, we need to load an existing file, and use the raw NNUE eval values. We could need to set SkipLoadingEval to false, set EvalFile to a proper net file path, and set Use NNUE to pure.

In training, we also need to load an existing file, and use the raw NNUE eval values. We could need to set SkipLoadingEval to false, set EvalFile to the one used in the training data generation, and set Use NNUE to pure.

After we setup CI for Reinforcement Learning, CI tests may fail because of https://github.com/nodchip/Stockfish/issues/128. EDIT: I'm investigating https://github.com/nodchip/Stockfish/issues/128.

[nodchip]

128 has been fixed.