v0.0.0 (Experimental / Test Release) This is a test LoRA created using Civitai's online training platform with just a single training image. The image caption consists entirely of natural language descriptions.According to my test using Civitai's default settings:<ul><li>At around 600 steps, the model began to memorize the training image.</li><li>At around 1000 steps, the model fully converged.</li></ul>

The Cat That Lived a Million Times

Initial. (Experimental).A LoRA weight of 1.0 or higher is recommended.LORA, MOSS Image Base, dim32/alpha32, AdamW8bit, LR 1e-4, cosine 1700steps, 3images, 1024² res.Captioning: <a target="_blank" rel="ugc" href="https://github.com/OpenMOSS/MOSS-Image/blob/main/USERGUIDE.md#caption-prompt">Caption Prompt of USERGUIDE.md</a>(using Qwen3.5-4B-Uncensored-HauhauCS-Aggressive-Q8_0).

Files with SHA256: db40543cf405349487f89a1b9893f76a5a996b3b96e5e53179f1803c655c48c8