Traning data from Cleale.
This LoRA looks a bit smoother in practice.
If you like it, feel free to drop some BUZZ to support future updates.
