This is a fork of LarryJane491's Lora-Training-in-Comfy, which again is based on Akegarasu's lora-scripts project.
- Implemented a bugfix for the Advanced node, correcting typing error on line 299 in the train.py file. (network_moduke changed to network_module.)
- Changed some hidden parameter for the simple node.
- Added Support for regularization dataset.
Make a folder in the data_path directory, and name this new folder reg_data_dir.