User guides

• Installation and support
• Full pipeline
• Phase-staged pipeline
• Modular Diffusers
• Dynamic component loading
• Presets and platforms
• Component and VAE strategies
• Configuration
• Metrics
• Capacity profiles
• Experimental native VMM
• Saved TorchAO FP8 weight-only
• FP8 conversion
• Opt-in dense convolution For model commands and measured comparisons, use the separate examples repository. Selected compatibility is not arbitrary-model or low-RAM acceptance.