Fine-tuning Llama-3 to get 90% of GPT-4's performance at a fraction of the costtogether.ai3 points·mnoorfawi··0 commentsOpen articleSaveView on HN