<h2 id="lora-based-on-agm-:-https:x.comagm86997980">LoRA based on AGM : <a target="_blank" rel="ugc" href="https://x.com/hachifukuDZ">https://x.com/AGM86997980</a></h2><h2 id="trained-using-wai-il-v15">Trained using WAI IL V15</h2><h2 id="trigger-words:-agm-stylerx"><span style="color:rgb(250, 82, 82)">Trigger words:</span> agm stylerx</h2><h2 id="recommended-weight:-0.7-0.9"><strong>Recommended weight: 0.7-0.9</strong></h2><p>Note:</p><p>This is trained with mixed sfw and nsfw dataset, excels in nsfw AGM style</p><p>This style work best with plump body(Style bias)</p><p>Lower weight tend to have better background, you might lose the style the lower you go.</p>

[Style | illustrious-XL] AGM

<p>Trial run, double the steps, half as light weight as fukuro RXS1</p><p>Style and autonomy are great for v1, still need to fix background performance with high weights</p><p>Generation in 2:3 may have out of frame issues, should be addressed in RXS2</p>

Files with SHA256: 3fe2ec3c5c7a7738ec938398e1b54c0016c870db9eb626a3448945753bdc6dc5