Are there plans to implement the rsm parameter for GPU training in CatBoost, and has its impact on overfitting mitigation been evaluated? [closed]
Closed 8 hours ago.
CatboostError for easy training without description
not so long ago I taught the model, got the results and everything approx. Now I needed to restart the project and everything stopped working for me, namely the error on catboost.
Catboost feature importance loses dimension
The model was trained on 444 features and gives result on such data, but feature importance contains only 443 features
Catboost feature importance losses dimension
The model was trained on 444 features and gives result on such data, but feature importance contains only 443 features