TODO.md

TODO List

• incorporate forgets and split into train data and model names
• improve model statistics with accuracies
• implement negative collisions removal
• when looping, use the training data from the previous loop and do not generate them again
• implement lgb parameters autotune