Qwen-Image-2.1-LoRA-experiments / experiments /assistant-regularisation.md
bghira's picture
Squash history while preserving current repository contents
7aa9202
|
Raw History Blame Contribute Delete
22 kB
# Combining an assistant with regularisation
[Contents](https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/blob/main/README.md) 路 [Previous: Synthetic regularisation](https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/blob/main/experiments/synthetic-regularisation.md) 路 [Next: Multi-scale training and REPA](https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/blob/main/experiments/multiscale-repa.md)
Combining assistant v2 with synthetic regularisation produces coherent scenes, portraits, and fox illustrations in the inspected final samples. At both output sizes, the 512px-trained adapter expresses the character in the beach, scarf, and bicycle scenes. Body proportions and bicycle interactions remain imperfect.
| Setting | Value |
| --- | --- |
| Training resolution | Separate 512px and 1024px runs |
| Assistant | Frozen v2, strength 1; disabled for validation |
| Regularisation target | **Base + frozen assistant v2** |
| Student | Base + frozen assistant v2 + trainable concept |
| Data | 27 Domokun images + 834 synthetic images per resolution |
| Learning rate | `1e-4` |
| Updates at midpoint | 439 Domokun + 561 regularisation |
| Updates at final | 904 Domokun + 1,096 regularisation |
The runs share initial concept weights and the backend-choice sequence. GPU audits verified the assisted target and restoration of the frozen-assistant training stack.
## Final checkpoints
The 1024px-trained bicycle sample at 1024px output resembles a human in a creature costume. Its scarf colour also changes with output resolution.
CPU-generator seed 42 路 40 inference steps 路 CFG 1 路 Details-fixed VAE (madebyollin) 路 full-frame decoding. Concept adapter only. Columns show prompts; rows show checkpoints or methods. Click a sample to view it at full resolution.
### 512px output
<div style="overflow-x: auto;">
<table>
<thead><tr><th scope="col">Checkpoint / method</th><th scope="col">Beach</th><th scope="col">Red scarf</th><th scope="col">Bicycle</th><th scope="col">Fox</th><th scope="col">Portrait</th></tr></thead>
<tbody>
<tr><th scope="row">Base</th><td><a href="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/737f04c4481c7dd531247755.png"><img alt="Base, 512px" loading="lazy" src="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/737f04c4481c7dd531247755.png" style="min-width: 160px; max-width: none;" width="160"/></a></td><td><a href="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/7b669b4d0905fc96612c178d.png"><img alt="Base, 512px" loading="lazy" src="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/7b669b4d0905fc96612c178d.png" style="min-width: 160px; max-width: none;" width="160"/></a></td><td><a href="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/9c762f2126b8f1092be142bd.png"><img alt="Base, 512px" loading="lazy" src="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/9c762f2126b8f1092be142bd.png" style="min-width: 160px; max-width: none;" width="160"/></a></td><td><a href="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/c150cc084b9dae047545cc64.png"><img alt="Base, 512px" loading="lazy" src="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/c150cc084b9dae047545cc64.png" style="min-width: 160px; max-width: none;" width="160"/></a></td><td><a href="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/998d84865ba20d52b1f69b71.png"><img alt="Base, 512px" loading="lazy" src="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/998d84865ba20d52b1f69b71.png" style="min-width: 160px; max-width: none;" width="160"/></a></td></tr>
<tr><th scope="row">512px training</th><td><a href="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/cbd0c4446629a26558ae81c6.png"><img alt="512px training, 512px" loading="lazy" src="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/cbd0c4446629a26558ae81c6.png" style="min-width: 160px; max-width: none;" width="160"/></a></td><td><a href="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/8f8979f11ff3d2ee4675afae.png"><img alt="512px training, 512px" loading="lazy" src="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/8f8979f11ff3d2ee4675afae.png" style="min-width: 160px; max-width: none;" width="160"/></a></td><td><a href="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/c7a5966da09ff73f6ee5360b.png"><img alt="512px training, 512px" loading="lazy" src="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/c7a5966da09ff73f6ee5360b.png" style="min-width: 160px; max-width: none;" width="160"/></a></td><td><a href="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/1d6c37b4cbe544c2051431ee.png"><img alt="512px training, 512px" loading="lazy" src="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/1d6c37b4cbe544c2051431ee.png" style="min-width: 160px; max-width: none;" width="160"/></a></td><td><a href="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/4d7f9ea46921f601fdbc25ac.png"><img alt="512px training, 512px" loading="lazy" src="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/4d7f9ea46921f601fdbc25ac.png" style="min-width: 160px; max-width: none;" width="160"/></a></td></tr>
<tr><th scope="row">1024px training</th><td><a href="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/dd47a5f1e88d6f7539f090ed.png"><img alt="1024px training, 512px" loading="lazy" src="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/dd47a5f1e88d6f7539f090ed.png" style="min-width: 160px; max-width: none;" width="160"/></a></td><td><a href="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/574e8ec7b6e6cfcf199af2ce.png"><img alt="1024px training, 512px" loading="lazy" src="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/574e8ec7b6e6cfcf199af2ce.png" style="min-width: 160px; max-width: none;" width="160"/></a></td><td><a href="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/e8a4ab20f69d38dea33cb12a.png"><img alt="1024px training, 512px" loading="lazy" src="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/e8a4ab20f69d38dea33cb12a.png" style="min-width: 160px; max-width: none;" width="160"/></a></td><td><a href="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/7c7a8320d1cf4d1584e842a8.png"><img alt="1024px training, 512px" loading="lazy" src="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/7c7a8320d1cf4d1584e842a8.png" style="min-width: 160px; max-width: none;" width="160"/></a></td><td><a href="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/1462580e1b35f6c75ee16a73.png"><img alt="1024px training, 512px" loading="lazy" src="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/1462580e1b35f6c75ee16a73.png" style="min-width: 160px; max-width: none;" width="160"/></a></td></tr>
</tbody>
</table>
</div>
### 1024px output
<div style="overflow-x: auto;">
<table>
<thead><tr><th scope="col">Checkpoint / method</th><th scope="col">Beach</th><th scope="col">Red scarf</th><th scope="col">Bicycle</th><th scope="col">Fox</th><th scope="col">Portrait</th></tr></thead>
<tbody>
<tr><th scope="row">Base</th><td><a href="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/b1741b857e5db2d2283e4e21.png"><img alt="Base, 1024px" loading="lazy" src="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/b1741b857e5db2d2283e4e21.png" style="min-width: 160px; max-width: none;" width="160"/></a></td><td><a href="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/bd898237725b4fc0e64346b0.png"><img alt="Base, 1024px" loading="lazy" src="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/bd898237725b4fc0e64346b0.png" style="min-width: 160px; max-width: none;" width="160"/></a></td><td><a href="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/372c1645fc744d64c12c4deb.png"><img alt="Base, 1024px" loading="lazy" src="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/372c1645fc744d64c12c4deb.png" style="min-width: 160px; max-width: none;" width="160"/></a></td><td><a href="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/0b328d8b9790d272b229f31a.png"><img alt="Base, 1024px" loading="lazy" src="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/0b328d8b9790d272b229f31a.png" style="min-width: 160px; max-width: none;" width="160"/></a></td><td><a href="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/41c1294975514ec5bc6fc61f.png"><img alt="Base, 1024px" loading="lazy" src="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/41c1294975514ec5bc6fc61f.png" style="min-width: 160px; max-width: none;" width="160"/></a></td></tr>
<tr><th scope="row">512px training</th><td><a href="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/5c328d498a703f0037535d3d.png"><img alt="512px training, 1024px" loading="lazy" src="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/5c328d498a703f0037535d3d.png" style="min-width: 160px; max-width: none;" width="160"/></a></td><td><a href="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/d1a8c285b643645bca5713ab.png"><img alt="512px training, 1024px" loading="lazy" src="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/d1a8c285b643645bca5713ab.png" style="min-width: 160px; max-width: none;" width="160"/></a></td><td><a href="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/39d06258ffb3356e11fcb90c.png"><img alt="512px training, 1024px" loading="lazy" src="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/39d06258ffb3356e11fcb90c.png" style="min-width: 160px; max-width: none;" width="160"/></a></td><td><a href="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/343ff03e6b62b390bfeda6a0.png"><img alt="512px training, 1024px" loading="lazy" src="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/343ff03e6b62b390bfeda6a0.png" style="min-width: 160px; max-width: none;" width="160"/></a></td><td><a href="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/103973c5e6f763ee2f8f997c.png"><img alt="512px training, 1024px" loading="lazy" src="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/103973c5e6f763ee2f8f997c.png" style="min-width: 160px; max-width: none;" width="160"/></a></td></tr>
<tr><th scope="row">1024px training</th><td><a href="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/5d2fb1a6b1a17bf15b61d818.png"><img alt="1024px training, 1024px" loading="lazy" src="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/5d2fb1a6b1a17bf15b61d818.png" style="min-width: 160px; max-width: none;" width="160"/></a></td><td><a href="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/f32b6761827e14c07549c30f.png"><img alt="1024px training, 1024px" loading="lazy" src="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/f32b6761827e14c07549c30f.png" style="min-width: 160px; max-width: none;" width="160"/></a></td><td><a href="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/bc024332cf2bfa455870399f.png"><img alt="1024px training, 1024px" loading="lazy" src="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/bc024332cf2bfa455870399f.png" style="min-width: 160px; max-width: none;" width="160"/></a></td><td><a href="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/c68d4c182b779f995a5ae4cf.png"><img alt="1024px training, 1024px" loading="lazy" src="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/c68d4c182b779f995a5ae4cf.png" style="min-width: 160px; max-width: none;" width="160"/></a></td><td><a href="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/39ece219417d6acb3eb0151a.png"><img alt="1024px training, 1024px" loading="lazy" src="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/39ece219417d6acb3eb0151a.png" style="min-width: 160px; max-width: none;" width="160"/></a></td></tr>
</tbody>
</table>
</div>
## Training progression
The following columns combine training resolution and prompt; every sample is 1024px output.
### 1024px output
<div style="overflow-x: auto;">
<table>
<thead><tr><th scope="col">Checkpoint / method</th><th scope="col">512px training: beach</th><th scope="col">512px training: bicycle</th><th scope="col">1024px training: beach</th><th scope="col">1024px training: bicycle</th></tr></thead>
<tbody>
<tr><th scope="row">Base</th><td><a href="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/b1741b857e5db2d2283e4e21.png"><img alt="Base, 1024px" loading="lazy" src="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/b1741b857e5db2d2283e4e21.png" style="min-width: 160px; max-width: none;" width="160"/></a></td><td><a href="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/372c1645fc744d64c12c4deb.png"><img alt="Base, 1024px" loading="lazy" src="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/372c1645fc744d64c12c4deb.png" style="min-width: 160px; max-width: none;" width="160"/></a></td><td><a href="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/b1741b857e5db2d2283e4e21.png"><img alt="Base, 1024px" loading="lazy" src="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/b1741b857e5db2d2283e4e21.png" style="min-width: 160px; max-width: none;" width="160"/></a></td><td><a href="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/372c1645fc744d64c12c4deb.png"><img alt="Base, 1024px" loading="lazy" src="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/372c1645fc744d64c12c4deb.png" style="min-width: 160px; max-width: none;" width="160"/></a></td></tr>
<tr><th scope="row">Update 1,000</th><td><a href="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/9f7fc1ecf568e4dc15490a8f.png"><img alt="Update 1,000, 1024px" loading="lazy" src="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/9f7fc1ecf568e4dc15490a8f.png" style="min-width: 160px; max-width: none;" width="160"/></a></td><td><a href="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/ea4eab650bfc7dc972e8f16b.png"><img alt="Update 1,000, 1024px" loading="lazy" src="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/ea4eab650bfc7dc972e8f16b.png" style="min-width: 160px; max-width: none;" width="160"/></a></td><td><a href="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/d752265e11f8e0a3fe7a0441.png"><img alt="Update 1,000, 1024px" loading="lazy" src="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/d752265e11f8e0a3fe7a0441.png" style="min-width: 160px; max-width: none;" width="160"/></a></td><td><a href="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/7c03c90e20e923963f0053b9.png"><img alt="Update 1,000, 1024px" loading="lazy" src="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/7c03c90e20e923963f0053b9.png" style="min-width: 160px; max-width: none;" width="160"/></a></td></tr>
<tr><th scope="row">Update 2,000</th><td><a href="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/5c328d498a703f0037535d3d.png"><img alt="Update 2,000, 1024px" loading="lazy" src="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/5c328d498a703f0037535d3d.png" style="min-width: 160px; max-width: none;" width="160"/></a></td><td><a href="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/39d06258ffb3356e11fcb90c.png"><img alt="Update 2,000, 1024px" loading="lazy" src="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/39d06258ffb3356e11fcb90c.png" style="min-width: 160px; max-width: none;" width="160"/></a></td><td><a href="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/5d2fb1a6b1a17bf15b61d818.png"><img alt="Update 2,000, 1024px" loading="lazy" src="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/5d2fb1a6b1a17bf15b61d818.png" style="min-width: 160px; max-width: none;" width="160"/></a></td><td><a href="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/bc024332cf2bfa455870399f.png"><img alt="Update 2,000, 1024px" loading="lazy" src="https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/assets/fixed-vae/bc024332cf2bfa455870399f.png" style="min-width: 160px; max-width: none;" width="160"/></a></td></tr>
</tbody>
</table>
</div>
## Reproduce
512px: [Training configuration](https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/blob/main/recipes/assisted-v2-regularised-512/config.json) 路 [Datasets](https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/blob/main/recipes/assisted-v2-regularised-512/dataloader.json) 路 [Prompts](https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/blob/main/recipes/assisted-v2-regularised-512/prompts.json) 路 [1,000 updates](https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/checkpoints/assisted-v2-regularised-512/step-1000/pytorch_lora_weights.safetensors) 路 [2,000 updates](https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/checkpoints/assisted-v2-regularised-512/step-2000/pytorch_lora_weights.safetensors)
1024px: [Training configuration](https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/blob/main/recipes/assisted-v2-regularised-1024/config.json) 路 [Datasets](https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/blob/main/recipes/assisted-v2-regularised-1024/dataloader.json) 路 [Prompts](https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/blob/main/recipes/assisted-v2-regularised-1024/prompts.json) 路 [1,000 updates](https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/checkpoints/assisted-v2-regularised-1024/step-1000/pytorch_lora_weights.safetensors) 路 [2,000 updates](https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/resolve/main/checkpoints/assisted-v2-regularised-1024/step-2000/pytorch_lora_weights.safetensors)
These recipes require assistant-aware parent capture. Use the recorded [implementation and launch instructions](https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/blob/main/reference/reproduction.md#assistant-aware-regularisation) and [training provenance](https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/blob/main/training_provenance.json).
## Limitations
The synthetic pools differ between training resolutions. [Multi-scale and REPA experiments](https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/blob/main/experiments/multiscale-repa.md) use a bare-base regularisation target, a different training setup.
[Contents](https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/blob/main/README.md) 路 [Previous: Synthetic regularisation](https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/blob/main/experiments/synthetic-regularisation.md) 路 [Next: Multi-scale training and REPA](https://huggingface.co/SimpleTuner/Qwen-Image-2.1-LoRA-experiments/blob/main/experiments/multiscale-repa.md)