<h2 id="alita:-battle-angel-dollbody-z7cj4xbjb"><strong><span style="color:rgb(130, 201, 30)">Alita: Battle Angel DOLLBODY</span></strong></h2><p>Trained Locally on Kohya-Trainer: RTX 3090</p><p>4K HDR Screenshots downscaled to: 1024x900</p><p>Number of Trained Images : 468</p><p>Captioned Locally: Joy Caption</p><p></p><h2 id="training-parameter-7z31y9l2u"><span style="color:rgb(250, 82, 82)">Training Parameter</span></h2><p><span style="color:rgb(229, 231, 235)">Repeats: 2 (since we have a plenty of images, no need to go higher)</span></p><p><span style="color:rgb(229, 231, 235)">Epochs: 10 ( I found my sweet spot here)</span></p><p><span style="color:rgb(229, 231, 235)">Train batch size: 2 (I don't recommend going higher than 2 or 3, better to keep it at 1)</span></p><p><span style="color:rgb(229, 231, 235)">Rank: 32 ( Rank is all about details/Traits, main feature of our subject (Dimensionality of our subject to embed for the Latent space), For Alita's body a Rank of 32 is enough for me as of now. Lack of resources for training :( I would go 128k</span></p><p><span style="color:rgb(229, 231, 235)">steps: 4680 (Took about 7+Hrs to train)</span></p><p></p><h1 id="prompting-rruzbw6sh"><span style="color:rgb(121, 80, 242)">Prompting</span></h1><p><span style="color:rgb(130, 201, 30)">Trigger word</span> : AL1IT0A</p><p><span style="color:rgb(130, 201, 30)">CFG:</span> 1 -1.3 could work fine.</p><p></p><p><span style="color:#82c91e">Dollbody:</span> "The image is a high-resolution photograph depicting a woman with fair skin and straight, shoulder-length dark brown hair. She has a medium build and is dressed in a form-fitting, white bodysuit with intricate, metallic, and organic-looking patterns. The bodysuit is made of a shiny, futuristic material with a slight sheen, suggesting advanced technology. Her expression is contemplative."</p><p></p><p>[Below this prompt you can add your own]</p><p></p><p>or it would be much better if you use the <span style="color:rgb(250, 176, 5)">joy caption texts</span> in the dataset I provided.</p><h2 id="implementations-ju8e3opy6"><span style="color:rgb(34, 139, 230)">Implementations</span></h2><ul><li><p>Not only you can create Alita with this lora, but you can also merge or mix her Cyborg Berserker body , in this example with Dua Lipa by <a target="_blank" rel="ugc" href="https://civarchive.com/user/steffangund">@steffangund</a></p></li></ul><p><a target="_blank" rel="ugc" href="https://civarchive.com/user/steffangund">https://civarchive.com/user/steffangund</a></p><p></p><ul><li><p>I also provided the training data so you can do your own experiment &amp; share your results as always.</p></li></ul><p></p><p>DuaLipa swan song</p><div data-youtube-video><iframe width="640" height="480" allowfullscreen="true" autoplay="false" disablekbcontrols="false" enableiframeapi="false" endtime="0" ivloadpolicy="0" loop="false" modestbranding="false" origin playlist src="https://www.youtube.com/embed/kO8fTk6oKQg" start="0"></iframe></div><p></p><p>Trailer</p><div data-youtube-video><iframe width="640" height="480" allowfullscreen="true" autoplay="false" disablekbcontrols="false" enableiframeapi="false" endtime="0" ivloadpolicy="0" loop="false" modestbranding="false" origin playlist src="https://www.youtube.com/embed/5Ww0AUWdXp4" start="0"></iframe></div><p></p><h2 id="i'm-just-an-insignificant-lora-creator-but-a-buzz-would-help)-65qjli7qq"><span style="color:rgb(130, 201, 30)">I'm just an insignificant Lora Creator, but a Buzz would help;)</span></h2>

Files with SHA256: f1544c43d7e5b4b954377c8083197435710f03bf6d3483c7a8c61d88859b4f57