SimpleTuner
An open-source diffusion model training framework built on top of Diffusers, offering dataset caching, multi-resolution bucketing, VRAM optimization, EMA, and unified configuration for LoRA and full fine-tuning, with support for SDXL, FLUX, Kolors, Sana and other architectures.
Timeline (last 90 days)
Training of the SpaceVision LoRA used the SimpleTuner and Diffusers frameworks.
Diffusers is an inference and training library maintained by Hugging Face, providing modular implementations of UNet, Transformer, schedulers, VAE, text encoders, etc.; SimpleTuner is built on top of it and supports architectures such as SDXL, FLUX, Kolors, and Sana.
A reproducible workflow is: collect domain-specific images → generate detailed captions with JoyCaption → train a LoRA with SimpleTuner and Diffusers; this process can be transferred to other vertical domains.
All Facts (3)
Training of the SpaceVision LoRA used the SimpleTuner and Diffusers frameworks.
50%UnverifiedDiffusers is an inference and training library maintained by Hugging Face, providing modular implementations of UNet, Transformer, schedulers, VAE, text encoders, etc.; SimpleTuner is built on top of it and supports architectures such as SDXL, FLUX, Kolors, and Sana.
50%UnverifiedA reproducible workflow is: collect domain-specific images → generate detailed captions with JoyCaption → train a LoRA with SimpleTuner and Diffusers; this process can be transferred to other vertical domains.
50%