This is a full restoration of <a target="_blank" rel="ugc" href="https://huggingface.co/zer0int">Zero0Int </a>and <a target="_blank" rel="ugc" href="https://civarchive.com/user/AbstractPhila">AbstractPila</a> work on the CLIP-L model.<ul><li>In 1000's of test images the FP32 model corrected bad anatomy in cases where the FP16 model failed (Seed to Seed)</li><li>In most cases the images have little difference but when differences exist the FP32 model was reliably more accurate (As it should be mathematically)</li><li>Note the vision models use the VIT-L vision blocks Zer0Int has a new finetune of the <a rel="ugc" href="https://huggingface.co/zer0int/CLIP-Registers-Gated_MLP-ViT-L-14">VIT/Vision model here</a></li></ul>MIT LicenseCopyright (c) 2021 OpenAIPermission is hereby granted, free of charge, to any person obtaining a copyof this software and associated documentation files (the "Software"), to dealin the Software without restriction, including without limitation the rightsto use, copy, modify, merge, publish, distribute, sublicense, and/or sellcopies of the Software, and to permit persons to whom the Software isfurnished to do so, subject to the following conditions:The above copyright notice and this permission notice shall be included in allcopies or substantial portions of the Software.THE SOFTWARE IS PROVIDED "AS IS", WITHOUT WARRANTY OF ANY KIND, EXPRESS ORIMPLIED, INCLUDING BUT NOT LIMITED TO THE WARRANTIES OF MERCHANTABILITY,FITNESS FOR A PARTICULAR PURPOSE AND NONINFRINGEMENT. IN NO EVENT SHALL THEAUTHORS OR COPYRIGHT HOLDERS BE LIABLE FOR ANY CLAIM, DAMAGES OR OTHERLIABILITY, WHETHER IN AN ACTION OF CONTRACT, TORT OR OTHERWISE, ARISING FROM,OUT OF OR IN CONNECTION WITH THE SOFTWARE OR THE USE OR OTHER DEALINGS IN THESOFTWARE.

CLIP-L & CLIP-G Full FP32 (Zer0Int & Simulacrum)

clipLCLIPGFullFP32_simulacrumCLIPGFP32.safetensors

Simulacrum-CLIP-G-FP32

Files with SHA256: eb0693caa06216a6daf9512c000633f8d98a53137ee118c82f5e404d94d446ff