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