okuvshynov/slowllama

Finetune llama2-70b and codellama on MacBook Air without quantization

156 best points · 54 comments · 1 Hacker News post · 446 GitHub stars · View on GitHub

Hacker News discussions