search

ggerganov / llama.cpp

LLM inference in C/C++
MIT License
67.22k stars 9.65k forks source link

binaries don't work? #9794

Open 0wwafa opened 3 weeks ago

0wwafa commented 3 weeks ago

on google colab (even after apt update && apt upgrade

./build/bin/llama-quantize: /lib/x86_64-linux-gnu/libstdc++.so.6: version `GLIBCXX_3.4.32' not found (required by ./build/bin/llama-quantize)
./build/bin/llama-quantize: /lib/x86_64-linux-gnu/libc.so.6: version `GLIBC_2.38' not found (required by ./build/bin/llama-quantize)
rumourscape commented 2 weeks ago

Same issue here

ericdaltman commented 2 weeks ago

GLIBC requirement is beyond the capabilities of the still supported previous Ubuntu LTS. A code downgrade would be required or at the least, if we could know which version of the software we'd have to regress to in order for this to remain working on Ubuntu 22.

0wwafa commented 2 weeks ago

GLIBC requirement is beyond the capabilities of the still supported previous Ubuntu LTS. A code downgrade would be required or at the least, if we could know which version of the software we'd have to regress to in order for this to remain working on Ubuntu 22.

just compile it static for ubuntu 20. And it will work almost everywhere.

0wwafa commented 2 weeks ago

GLIBC requirement is beyond the capabilities of the still supported previous Ubuntu LTS. A code downgrade would be required or at the least, if we could know which version of the software we'd have to regress to in order for this to remain working on Ubuntu 22.

just compile it static for ubuntu 20. And it will work almost everywhere.

Anything that works out of the box on google colab is fine for me :D