<h1 id="sdxl-4-step-(fp32-with-improved-unet)-cbquyc3ie">SDXL 4 Step (FP32 with Improved UNET)</h1><ul><li>Note I used the <a target="_blank" rel="ugc" href="https://huggingface.co/stabilityai/stable-diffusion-xl-refiner-1.0/resolve/main/text_encoder_2/model.safetensors?download=true">32bit CLIP-G</a></li><li><a target="_blank" rel="ugc" href="https://civarchive.com/models/929400/fp32-sdxlflux-refiner-2gb-gguf">Refiner</a> used in workflow also 32bit GGUF</li><li>Updated <a target="_blank" rel="ugc" href="https://huggingface.co/zer0int/CLIP-GmP-ViT-L-14/blob/main/ViT-L-14-TEXT-detail-improved-hiT-GmP-TE-only-HF.safetensors">CLIP-L</a></li><li>For 4 Step use at CFG 1.0 - Load a image for workflow</li><li>For NSFW images the refiner should not be used</li><li><a rel="ugc" href="https://drive.google.com/drive/folders/13kfr3qny7S2xwG9h7v95F5mkWs0OmU0D">BRSGAN 2x</a> can be found on Google Drive</li></ul><h2 id="2gb-gguf-from-fp32-t1u3nh4zh">2GB GGUF from FP32</h2>Note 2GB requires separate clip and GGUF support, 4GB FP8 is ready to use in any SD GUI<ul><li>Refined with baked FP32 Lora's</li><li>Quantized from the FP32 SDXL model for less loss</li><li>For GGUF Download separate SDXL <a target="_blank" rel="ugc" href="https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0/blob/main/text_encoder_2/model.fp16.safetensors">CLIP-G</a> and <a target="_blank" rel="ugc" href="https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0/blob/main/text_encoder/model.fp16.safetensors">CLIP-L</a> and <a target="_blank" rel="ugc" href="https://huggingface.co/stabilityai/stable-diffusion-xl-base-1.0/blob/main/vae/diffusion_pytorch_model.fp16.safetensors">VAE</a></li></ul><h2 id="4gb-sdxl-(full-checkpoint)-9kert3qca">4GB SDXL (Full Checkpoint)</h2><ul><li>Custom CLIP is not Quantized</li><li>Custom UNET quantized to FP8 allowing for a balance of size and quality</li><li>Works in FORGE, Comfy-UI and Auto-1111</li><li>Works with LORA's</li><li>Beta/Deis is a good choice for img 2 img up-scaling</li></ul>Both models have improved anatomy (Uncensored) for females however GGUF version does not do well with males.

SDXL 4GB/2GB (Improved FP8 & GGUF)

Files with SHA256: a4c62699fd5dae8b72e350c3f0dd4bcf624582a2d964fba556267fc51652689a