This repository only contains the adapter weights from LoRA fine-tuning. To load the model, the base model Llama-2-13b-chat-hf has to be loaded and used as base to load the adapter weights.

The adapter weights can be merged with the base model. Since this takes much more space tough, only adapter folder was uploaded. If merging is required, please refer to the project repository or the llama-recepies repository by meta research labs (https://github.com/facebookresearch/llama-recipes) for examples.