README.md

This is a port of llama2.c to Fusion. Fusion can be automatically translated to pure C, C++, C#, D, Java, JavaScript, Python, Swift and TypeScript with no additional dependencies.

It uses a similar model format to llama2.c, but with bfloat16 weights, so it occupies half the disk space, half the RAM and is twice as fast. Use export_meta_llama_bin.py to convert the original Meta models or float2bf16.c to convert llama2.c models.