Skip to content

Latest commit

 

History

History
9 lines (6 loc) · 464 Bytes

README.md

File metadata and controls

9 lines (6 loc) · 464 Bytes

LA-LLAMA3

LA-LLAMA3 is a minimalist implementation of the LLAMA3 algorithm, primarily using linear algebra operations, and its main program is confined to just 100 lines of code. This project serves as a tool to provide a clear understanding of the fundamental workings of the LLAMA3(and generally, Transformer) architecture.

And it actually works.

For a version with KV Cache, check out the kv_cache branch.