Flux Dev To Schnell 4 Step LoRA - FP32 API Documentation

This version is float32 if your card doesn't support BF16 it's better to have this convert to FP16 than BF16. Less precision loss that way.