Finetune Llama 3.1, Mistral, Phi & Gemma LLMs 2-5x faster with 80% less memory - View it on GitHub