<span style="background-color:rgb(26, 27, 30);color:rgb(141, 141, 143);font-family:-apple-system, BlinkMacSystemFont, &quot;Segoe UI&quot;, Roboto, Helvetica, Arial, sans-serif, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;;font-size:16px">Based on Anima (base v1.0).<span style="background-color:rgb(26, 27, 30);color:rgb(141, 141, 143);font-family:-apple-system, BlinkMacSystemFont, &quot;Segoe UI&quot;, Roboto, Helvetica, Arial, sans-serif, &quot;Apple Color Emoji&quot;, &quot;Segoe UI Emoji&quot;;font-size:16px">Trigger word:@PARA7737This is a LoRA for one style of the artist:<a target="_blank" rel="ugc" href="https://www.pixiv.net/users/119152694">PARA7737</a>.I will remove this if requested by the original artist.基于 Anima (base v1.0)。触发词：@PARA7737这是 <a target="_blank" rel="ugc" href="https://www.pixiv.net/users/119152694">PARA7737</a>风格之一的LoRA。如果原作者要求，我会删除此模型To be honest, training this creator's LoRA is quite challenging. I’ve been following them since the release of Chenkin Noob v0.5. After Anima Base came out, I ran numerous training sessions, continuously tweaking parameters and refining the dataset. Yet the outputs never met my expectations. Even the current iteration, which is the best result I’ve achieved so far, still has shortcomings in generalization and fitting performance.说实话，这位作者风格的 LoRA 训练难度很高。我从 Chenkin Noob v0.5 阶段就开始关注他，Anima Base 发布之后，我陆续做了多轮训练，反复调试参数、优化数据集，但成品效果始终达不到预期。即便当前版本已是我所有尝试里效果最优的，在泛化能力和拟合度上依旧存在不足。

Files with SHA256: 7b5ada50cd7d0d1f651e2fb9669e0b56c072a7a5606ffe34a4b55c99dc8c31f7