Inference of [Facebook's LLaMA](https://github.com/facebookresearch/llama) model in pure C/C++
**!!! IMPORTANT !!!**
**Hot topics**
Commit [007a8f6f459c6eb56678fdee4c09219ddb85b640](https://github.com/ggerganov/llama.cpp/commit/007a8f6f459c6eb56678fdee4c09219ddb85b640) added support for all LLaMA models, but introduced breaking changes. If you generated any models before that commit, you must regenerate them after updating to latest master.
- Running on Windows: https://github.com/ggerganov/llama.cpp/issues/22