Портирование ML-моделей в SD.Next

★ 6.5 · ai-tooling

port-model is a Claude Code skill that ports and integrates new machine learning models into SD.Next using a phased flow: identify the model architecture and artifact layout, select the integration path requiring the least new code, then implement the loader and routing updates. The skill supports three paths — reusing an upstream Diffusers pipeline with SD.Next wiring, creating a custom package under `pipelines/<model>/`, or explicitly remapping weights from a raw checkpoint or safetensors file. Routing changes span `modules/sd_models.py`, `modules/sd_detect.py`, and `modules/modeldata.py`, f