search

lrw04 / llama2.c-to-ncnn

A converter for llama2.c legacy models to ncnn models.
MIT License
82 stars 10 forks source link

PR patch deleted #4

Open magicse opened 2 months ago

magicse commented 2 months ago

Did you delete your PR patch? https://github.com/lrw04/ncnn/tree/patch-linearint8

lrw04 commented 2 months ago

it's left as an exercise for the reader

magicse commented 2 months ago

Thank You for your work. I added compilation custom linearint8 layer locally without changing ncnn sources. https://github.com/magicse/llama2.c-to-ncnn/tree/master May be later i make GPU support for linearint8 also locally.