4/ Faster to train! Training with LoRA can easily be 4-8x faster than full 
fine-tuning.

→ View original post on X — @predibase,