<p><strong><span style="color:rgb(250, 82, 82)">I am sharing how I trained this model with full details and even the dataset: please read entire post very carefully.</span></strong></p><p><strong><span style="color:rgb(230, 73, 128)">This model is purely trained for educational and research purposes only for SFW and ethical image generation.</span></strong></p><p><strong><span style="color:rgb(190, 75, 219)">The workflow and the config used in this tutorial can be used to train clothing, items, animals, pets, objects, styles, simply anything.</span></strong></p><p><strong><span style="color:rgb(130, 201, 30)">The uploaded images have SwarmUI metadata and can be re-generated exactly. For generations FP16 model used but FP8 should yield almost same quality. Don't forget to have used yolo face masking model in prompts.</span></strong></p><p></p><h2 id="how-to-use-9u53nkzxb"><strong><span style="color:rgb(250, 82, 82)">How To Use</span></strong></h2><p>Download model into diffusion_models of the SwarmUI. Then you need to use Clip-L and T5-XXL models as well. I recommend T5-XXL FP16 or Scaled FP8 version.</p><p>A newest fully public tutorial here for how to use :</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/-zOKhoO9a5s" start="0"></iframe></div><p></p><p>I have trained both FLUX LoRA and Fine-Tuning / DreamBooth model.</p><p>Activation token / trigger word : <strong>ohwx man</strong></p><p>Each training was up to 200 epochs and once every 10 epoch checkpoints saved and shared on below Hugging Face Repo : <a target="_blank" rel="ugc" href="https://huggingface.co/MonsterMMORPG/Model_Training_Experiments_As_A_Baseline">https://huggingface.co/MonsterMMORPG/Model_Training_Experiments_As_A_Baseline</a></p><p></p><p>This model contains experimental results comparing Fine-Tuning / DreamBooth and LoRA training approaches.</p><h2 id="additional-resources-n45pyl964">Additional Resources</h2><ul><li><p>Installers and Config Files : <a target="_blank" rel="ugc" href="https://www.patreon.com/posts/112099700"><strong><u>https://www.patreon.com/posts/112099700</u></strong></a></p></li><li><p>FLUX Fine-Tuning / DreamBooth Zero-to-Hero Tutorial : <a target="_blank" rel="ugc" href="https://youtu.be/FvpWy1x5etM"><strong><u>https://youtu.be/FvpWy1x5etM</u></strong></a></p></li><li><p></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/FvpWy1x5etM" start="0"></iframe></div></li><li><p>FLUX LoRA Training Zero-to-Hero Tutorial : <a target="_blank" rel="ugc" href="https://youtu.be/nySGu12Y05k"><strong><u>https://youtu.be/nySGu12Y05k</u></strong></a></p></li><li><p></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/nySGu12Y05k" start="0"></iframe></div></li><li><p>Complete Dataset, Training Config Json Files and Testing Prompts : <a target="_blank" rel="ugc" href="https://www.patreon.com/posts/114972274"><strong><u>https://www.patreon.com/posts/114972274</u></strong></a></p></li><li><p>Click below link to download all trained LoRA and Fine-Tuning / DreamBooth checkpoints for free</p></li><li><p><a target="_blank" rel="ugc" href="https://huggingface.co/MonsterMMORPG/Model_Training_Experiments_As_A_Baseline/tree/main"><strong><u>https://huggingface.co/MonsterMMORPG/Model_Training_Experiments_As_A_Baseline/tree/main</u></strong></a></p></li></ul><h2 id="environment-setup-p6altk4bd">Environment Setup</h2><ul><li><p>Kohya GUI Version: <code>021c6f5ae3055320a56967284e759620c349aa56</code></p></li><li><p>Torch: 2.5.1</p></li><li><p>xFormers: 0.0.28.post3</p></li></ul><h2 id="dataset-information-rwzmsasgr">Dataset Information</h2><ul><li><p>Resolution: 1024x1024</p></li><li><p>Dataset Size: 28 images</p></li><li><p>Captions: "ohwx man" (nothing else)</p></li><li><p>Activation Token/Trigger Word: "ohwx man"</p></li></ul><h2 id="fine-tuning-dreambooth-experiment-ht2e2nhat">Fine-Tuning / DreamBooth Experiment</h2><h3 id="configuration-7pxzzo31u">Configuration</h3><ul><li><p>Config File: <code>48GB_GPU_28200MB_6.4_second_it_Tier_1.json</code></p></li><li><p>Training: Up to 200 epochs with consistent config</p></li><li><p>Optimal Result: Epoch 170 (subjective assessment)</p></li></ul><h3 id="results-riqzikzm3">Results</h3><ul><li><p><a target="_blank" rel="ugc" href="https://huggingface.co/MonsterMMORPG/Model_Training_Experiments_As_A_Baseline/blob/main/Dwayne_Fine_Tune_Realism_Test_Part1.jpg"><strong><u>Realism Test Part 1</u></strong></a></p></li><li><p><a target="_blank" rel="ugc" href="https://huggingface.co/MonsterMMORPG/Model_Training_Experiments_As_A_Baseline/blob/main/Dwayne_Fine_Tune_Realism_Test_Part2.jpg"><strong><u>Realism Test Part 2</u></strong></a></p></li></ul><h2 id="lora-experiment-ih7e8pnq8">LoRA Experiment</h2><h3 id="configuration-cva20ctgo">Configuration</h3><ul><li><p>Config File: <code>Rank_1_29500MB_8_85_Second_IT.json</code></p></li><li><p>Training: Up to 200 epochs</p></li><li><p>Optimal Result: Epoch 160 (subjective assessment)</p></li></ul><h3 id="results-o43vfeh04">Results</h3><ul><li><p><a target="_blank" rel="ugc" href="https://huggingface.co/MonsterMMORPG/Model_Training_Experiments_As_A_Baseline/blob/main/Dwayne_LoRA_Realism_Test_Part1.jpg"><strong><u>LoRA Realism Test Part 1</u></strong></a></p></li><li><p><a target="_blank" rel="ugc" href="https://huggingface.co/MonsterMMORPG/Model_Training_Experiments_As_A_Baseline/blob/main/Dwayne_LoRA_Realism_Test_Part2.jpg"><strong><u>LoRA Realism Test Part 2</u></strong></a></p></li></ul><h2 id="comparison-results-o2axklxz3">Comparison Results</h2><ul><li><p><a target="_blank" rel="ugc" href="https://huggingface.co/MonsterMMORPG/Model_Training_Experiments_As_A_Baseline/blob/main/LoRA_90_Epoch_vs_LoRA_160_Epoch_vs_Fine_Tuning_170_Epoch.jpg"><strong><u>LoRA 90 vs 160 vs Fine-Tuning 170 Comparison</u></strong></a></p></li></ul><h3 id="key-observations-jd3j2q451">Key Observations</h3><ul><li><p>LoRA demonstrates excellent realism but shows more obvious overfitting when generating stylized images.</p></li><li><p>Fine-Tuning / DreamBooth is better than LoRA as expected.</p></li></ul><h2 id="model-naming-convention-ftndxqo0e">Model Naming Convention</h2><h3 id="fine-tuning-models-7ajix9qnv">Fine-Tuning Models</h3><ul><li><p><code>Dwayne_Johnson_FLUX_Fine_Tuning-000010.safetensors</code></p><ul><li><p>10 epochs</p></li><li><p>280 steps (28 images × 10 epochs)</p></li><li><p>Batch size: 1</p></li><li><p>Resolution: 1024x1024</p></li></ul></li><li><p><code>Dwayne_Johnson_FLUX_Fine_Tuning-000020.safetensors</code></p><ul><li><p>20 epochs</p></li><li><p>560 steps (28 images × 20 epochs)</p></li><li><p>Batch size: 1</p></li><li><p>Resolution: 1024x1024</p></li></ul></li></ul><h3 id="lora-models-60vq3qq15">LoRA Models</h3><ul><li><p><code>Dwayne_Johnson_FLUX_LoRA-000010.safetensors</code></p><ul><li><p>10 epochs</p></li><li><p>280 steps (28 images × 10 epochs)</p></li><li><p>Batch size: 1</p></li><li><p>Resolution: 1024x1024</p></li></ul></li><li><p><code>Dwayne_Johnson_FLUX_LoRA-000020.safetensors</code></p><ul><li><p>20 epochs</p></li><li><p>560 steps (28 images × 20 epochs)</p></li><li><p>Batch size: 1</p></li><li><p>Resolution: 1024x1024</p></li></ul></li></ul>

Dwayne Johnson aka The Rock FLUX Dev Fine-Tuning / DreamBooth Model for Educational and Research Purposes - Dwayne Johnson aka The Rock FLUX Dev LoRA Model for Educational and Research Purposes - Full Tutorial

<p><strong>For Full Details, Training Dataset, Tutorial, Guide, Configs, Training Json Files, Workflows, Installers, Resources and All Checkpoints &gt; </strong><a target="_blank" rel="ugc" href="https://huggingface.co/MonsterMMORPG/Model_Training_Experiments_As_A_Baseline"><strong>https://huggingface.co/MonsterMMORPG/Model_Training_Experiments_As_A_Baseline</strong></a></p><p><strong>This is FP8 converted version of original FP16 training</strong></p>

Files with SHA256: 7cbfac7158cb1b09650eb0388a9b246ea99e2cecf6411404a95e20d04e15cc5c