<h2 id="sdxl-1.0-(distilledpredictive)">SDXL 1.0 (Distilled/Predictive)</h2><ul><li>Full FP32 Model</li><li>CLIP-G and CLIP-L are both trained and distilled.</li></ul><hr />Distilled: The clip models have been distilled from larger text/token projections: teacher/student; goal the smaller model learns the latent shape of the larger.Predictive: The clip model had separate training where the padding token was used as a mask. Allowing for the clip model to predict some additional context if the 75 token limit is not used.

SDXL 1.0 (Distilled/Predictive)

SDXL_DP_FP32.safetensors

FP32-SFW

SDXL 1.0 (Distilled/Predictive)<ul><li>Full FP32 Model</li><li>CLIP-G and CLIP-L are both trained and distilled.</li></ul>Distilled: The clip models have been distilled from larger text/token projections to teacher the smaller model the latent shape of the larger.Predictive: The clip model had separate training where the padding token was used as a mask. Allowing for the clip model to predict some additional context if the 75 token limit is not used.

Files with SHA256: 3da4c11dcf1df7c763dd62470be7b348d479d1a1b5cda4485fb3cbb1e46a07b6