dut3062796s / lit-llama

Implementation of the LLaMA language model based on nanoGPT. Supports flash attention, quantization, LoRA fine-tuning, pre-training. Apache 2.0-licensed.

Geek Repo

Github PK Tool