This ComfyUI toolkit provides a unified interface for training LoRAs on multiple diffusion model architectures using three different training backends (sd-scripts, Musubi Tuner, AI-Toolkit). It includes trainer nodes, selective block-level LoRA loaders with strength scheduling, model analyzers for comparing trained LoRAs, and model layer editors for fine-tuning. Supports 8 architectures and offers utilities for LoRA extraction, merging, and compression.