Update export.py - Githubissues

srush / llama2.rs

A fast llama2 decoder in pure Rust.

MIT License

995 stars 54 forks source link

Closed huoyushequ closed 10 months ago

huoyushequ commented 10 months ago

to make sure all the tensors in function rearrange have dtype torch.int32