ComfyUI Node: Text Encoder Inspector (Qwen3-4B)
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
model_patches/analysis
Inputs
clip CLIP
prompt STRING
analysis_mode
- quick
- ablation_HEAVY
ablation_strength FLOAT
Outputs
STRING
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
Looking for a different node?
More nodes in Realtime LoRA Trainer
- Apply Trained LoRA
- FLUX Analyzer + Selective Loader V2
- FLUX Klein 4B Analyzer + Selective Loader V2
- FLUX Klein 9B Analyzer + Selective Loader V2
- DIT Deep Debiaser (FLUX.2 Klein — Verified)
- FLUX Model Layer Editor
- Selective LoRA Loader (FLUX)
- VAE Deep Debiaser (Flux 2 Klein — 125 Tensors)
- VAE Inspector (Flux 2 Klein — 125 Tensors)
- LoRA Loader + Analyzer
- LoRA Loader + Analyzer V2
- Realtime LoRA Trainer (FLUX Klein - Musubi Tuner)
Other model_patches/analysis 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.