Realtime LoRA Trainer (Qwen Image - Musubi Tuner)

MusubiQwenImageLoraTrainer

Trains a Qwen Image LoRA from images using Musubi Tuner. For style/subject LoRAs without control images.

Pack: Realtime LoRA Trainer

custom_nodes.comfyUI-Realtime-Lora

Inputs (24)

Name	Type	Required
inputcount	INT	required
images_path	STRING	required
musubi_path	STRING	required
model_mode	COMBO	required
dit_model	COMBO	required
vae_model	COMBO	required
text_encoder	COMBO	required
caption	STRING	required
training_steps	INT	required
learning_rate	FLOAT	required
lora_rank	INT	required
vram_mode	COMBO	required
blocks_to_swap	COMBO	required
keep_lora	BOOLEAN	required
output_name	STRING	required
custom_python_exe	STRING	required
image_1	IMAGE	optional
caption_1	STRING	optional
image_2	IMAGE	optional
caption_2	STRING	optional
image_3	IMAGE	optional
caption_3	STRING	optional
image_4	IMAGE	optional
caption_4	STRING	optional

Outputs (1)

Name	Type
lora_path	STRING