Details: https://github.com/Nerogar/OneTrainer/blob/master/docs/RamOffloading.md
- Flux LoRA training on 6GB GPUs (at 512px resolution)
- Flux Fine-Tuning on 16GB GPUs (or even less) +64GB of RAM
- SD3.5-M Fine-Tuning on 4GB GPUs (at 1024px resolution)
You must log in or # to comment.