mlx-examples
mlx-examples copied to clipboard
What's a good enough loss value, in LoRA fine-tuning?
How small should loss reach, 0.1 or 1.0?