-
Notifications
You must be signed in to change notification settings - Fork 13.3k
Description
Name and Version
version: 0 (unknown)
built with gcc-12 (Debian 12.4.0-5) 12.4.0 for x86_64-linux-gnu
the build is latest version b6529
Operating systems
Linux
Which llama.cpp modules do you know to be affected?
llama-server
Command line
llama-server -m GLM-4.5-UD-Q4_K_XL-00001-of-00005.gguf --threads,temp,penalty,ngl,tensor-split,etc etc etc --flash-attn on --swa-full --jinja
Problem description & steps to reproduce
Hello, the thought process is not displayed in the new Web UI. Neither enabling nor disabling "Show thought in progress" works, the chat screen just shows "Processing..." and ticking context tokens until the model starts outputting the result. I have checked with 2 different browsers: Firefox which I usually use for llama.cpp web UI (so all settings are saved in the LocalStorage/IndexedDB) and a clean fresh Chromium profile without any cookies or saved settings.
I have also tried to remove --jinja
from the command line but it did not help.
First Bad Commit
I don't know the first commit but in b6517 this function is also broken.