<h2 id="second-training-run-of-the-prepared-dataset-for-qwen-image-2512">Second Training Run of the Prepared Dataset for Qwen-Image 2512</h2><p>I’ve completed a second training run using the prepared dataset for Qwen-Image 2512.</p><p>The dataset included 400 images that I created and enhanced in Illustrious.<br />This time I tried using <code>rank64</code>, which is why the LoRA file is so large. I also used a high number of steps — 20,000 — which I think was justified for such a small dataset, judging by the tests I ran between checkpoints. I’m more than satisfied with the current results.</p><hr /><h3 id="important:-recommended-settings">⚠️ <span style="color:rgb(253, 126, 20)">Important: Recommended Settings</span></h3><p>If you decide to test this version, I used the <strong>Qwen-Image-2512-4steps-V1</strong> base. For the best results, use the following configuration:</p><ul><li><p><strong>KSampler Steps:</strong> 4-6</p></li><li><p><strong>Sampler:</strong> <code>euler_ancestral</code></p></li><li><p><strong>Scheduler:</strong> <code>beta57</code></p></li></ul><p><em>These settings appear to work best for maintaining possible stability and quality with this specific LoRA.</em></p><p></p><h2 id="current-state-of-the-lora">Current State of the LoRA</h2><p>The LoRA is still in an active development phase, so there may be artifacts and failures to understand many concepts.</p><p>Right now it’s a mix of multiple concepts, but roughly 60% of the dataset is either fully nude characters or characters in very revealing outfits.</p><p>There are very few sexual acts and anus-focused images, so the model often makes mistakes in those areas.</p><hr /><h2 id="concept-gaps-and-dataset-quality">Concept Gaps and Dataset Quality</h2><p>There are still significant gaps in concepts such as:</p><ul><li><p>Sex positions</p></li><li><p>Different types/styles of clothing</p></li></ul><p>The current dataset and captions are not perfect, but they already provide decent training results.</p><hr /><h2 id="future-plans">Future Plans</h2><p>There’s still a huge amount of work to be done, especially if we focus on the main concepts. I plan to keep working on this, gradually improving both the dataset and the LoRA.</p><p></p>

Files with SHA256: 6519be40c81541800e9d50f0b3fbd888ade1f136b4fd25f21496b5401f8dbe7f