GitLab

Bump llama.cpp to b2081

runners

disable rocm builds

Accomodate split cuda lib dir

Co-authored-by: Erik Sp <git@aschwa.com>

Harden generate patching model

Only apply patches if we have any, and make sure to cleanup
every file we patched at the end to leave the tree clean

append image tags to user content

structured debug prompt

update slog handler options

- consistent format by using text handler for debug and non-debug
- truncate source file to just the file name

remove unnecessary parse raw

Bump llama.cpp to b1999