LLMTune: 4-Bit finetuning of 65B LLAMA models on a single consumer GPUgithub.com/kuleshov-group3 pointsvolodia3 years ago