Stable Diffusion Xl
HotArt3 SPO
model_id: hotart3spo
<p>This version adds <a target="_blank" rel="ugc" href="https://civitai.com/models/510261/spo-sdxl4k-p10eplorawebui">Step-aware Preference Optimization (SPO) LORA</a>, [<em>a novel post-training approach that independently evaluates and adjusts the denoising performance at each step, using a step-aware preference model and a step-wise resampler to ensure accurate step-aware supervision</em>... blah, blah, blah = <u>makes the output way artfully beautiful, </u><strong><em><u>a.k.a. - NICER</u></em></strong> !!!]</p><p><em>P.S.: The SPO approach is </em><strong><em><u>better than </u></em></strong><a rel="ugc" href="https://civitai.com/models/242825/dpo-direct-preference-optimization-lora-for-xl-and-15-openrail?modelVersionId=273996"><strong><em><u>Direct Preference Optimization (DPO)</u></em></strong></a><em> that has extended its success from aligning large language models (LLMs) to aligning text-to-image diffusion models with human preferences. Unlike most existing DPO methods, which assume that all diffusion steps share a consistent preference order with the final generated images, this assumption neglects step-specific denoising performance and that preference labels should be tailored to each step's contribution.</em></p>
NSFW samples are gated
This model produces NSFW content. Log in and enable NSFW in your account settings to view samples and run the model. The API and model parameters below are accessible regardless.
Use HotArt3 SPO via the ModelsLab API
Send a POST request to the relevant endpoint with your API key and model_id="hotart3spo". The API does not enforce the same NSFW gating as the web playground — content moderation is handled by your account-level policy. Sign up to get an API key.