ComfyUI Node: Realtime LoRA Trainer (FLUX Klein - Musubi Tuner)

Authored by shootthesound

Created

Updated

535 stars

Run ComfyUI workflows without the setup

No installs, no CUDA version roulette, no GPU sitting idle on your bill. Bring a workflow and run it in the browser.

Category

loaders

Inputs

model_variant
  • Klein Base 4B
  • Klein Base 9B
inputcount INT
images_path STRING
musubi_path STRING
dit_model
    vae_model
      text_encoder
      • (no text encoders found)
      caption STRING
      training_steps INT
      learning_rate FLOAT
      lora_rank INT
      blocks_to_swap INT
      vram_mode
      • Max (1256px)
      • Max (1256px) fp8
      • Max (1256px) fp8 offload
      • Medium (1024px)
      • Medium (1024px) fp8
      • Medium (1024px) fp8 offload
      • Low (768px)
      • Min (512px)
      keep_lora BOOLEAN
      output_name STRING
      custom_python_exe STRING
      image_1 IMAGE
      caption_1 STRING
      image_2 IMAGE
      caption_2 STRING
      image_3 IMAGE
      caption_3 STRING
      image_4 IMAGE
      caption_4 STRING

      Outputs

      STRING

      Extension: Realtime LoRA Trainer

      Train, analyze, selectively load, and extract/merge LoRAs inside ComfyUI. Supports Z-Image, Qwen Image, Qwen Image Edit, SDXL, FLUX, FLUX Klein, Wan 2.2, and SD 1.5. Includes LoRA layer analyzers, selective block loaders, model layer editors, and text-encoder debiasing/inspection tools.

      Authored by shootthesound

      View Nodes

      Looking for a different node?

      More nodes in Realtime LoRA Trainer

      Other loaders nodes

      Run ComfyUI workflows without the setup

      No installs, no CUDA version roulette, no GPU sitting idle on your bill. Bring a workflow and run it in the browser.

      Learn more