A hybrid model combining <a target="_blank" rel="ugc" href="https://github.com/FlexGen/FlexGen">FLEX.1 alpha </a>and <a target="_blank" rel="ugc" href="https://github.com/Kaoru8/T5XXL-Unchained">T5XXL Unchained</a>, trained on ~1 million curated anime-style images from Danbooru (1 epoch). Prompt Format Example:Artwork by [artist], 1girl, rating: {general|sensitive|questionable}, [character details]... (Danbooru tag) [Natural language description]... Known IssuesAnatomy inconsistencies in complex posesRating tags (general/sensitive/questionable) show limited effectiveness Requirements <ul><li>T5XXL tokenizer patch MUST be applied (<a target="_blank" rel="ugc" href="https://github.com/Kaoru8/T5XXL-Unchained">https://github.com/Kaoru8/T5XXL-Unchained</a>)</li><li>CFG is needed for this model, the distilled module was broken during training. Recommended CFG Scale is around 4. </li></ul>Updated versions may be released pending continued development.

ICE_FLEXanime

Experimental checkpoint trained based on modified models.

Files with SHA256: dda969b6c7d538ce202463ee07b7064ac8eb82421c6a4675e32f383977aa98be