search

srush / llama2.rs

A fast llama2 decoder in pure Rust.
MIT License
1.01k stars 56 forks source link

Max vocab size #28

Closed srush closed 1 year ago

srush commented 1 year ago

Cuts off some of the additional tokens that are added for llama fine tunes (will support in a future PR).

https://github.com/srush/llama2.rs/issues/22