Skip to content

Misc. bug: WebUI: Reasoning think tags not parsed correctly for GLM4.5 - response in 'Reasoning' section #20356

@MarkErik

Description

@MarkErik

Name and Version

llama-server --version
version: 8250 (0beb8db)
built with AppleClang 17.0.0.17000604 for Darwin arm64

Operating systems

Mac

Which llama.cpp modules do you know to be affected?

llama-server

Command line

llama-server -m /Volumes/LLM/GLM-4.5-UD-Q8_K_XL-unsloth-GGUF/GLM-4.5-UD-Q8_K_XL-00001-of-00008.gguf --ctx-size 0 --temp 0.6 --top-p 0.95 --min-p 0 --host 0.0.0.0 --port 8080 -b 2048 -ub 2048

Problem description & steps to reproduce

  1. Started llama-server (with the GLM4.5 model, tried both 8bit and 4bit quant, same issue)
  2. Navigate to WebUI
  3. Ask it anything e.g. 'why is the sky blue'
  4. Open the 'Reasoning' section and see that tags are visible. And then after the the response is printed in the 'Reasoning' section.

Thought this might be useful:
srv params_from_: Chat format: peg-native

Image Image

First Bad Commit

No response

Relevant log output

Metadata

Metadata

Assignees

Labels

bugSomething isn't workingchat parserIssues related to the chat parser and chat templatesregressionA regression introduced in a new build (something that was previously working correctly)

Type

No type

Projects

No projects

Milestone

No milestone

Relationships

None yet

Development

No branches or pull requests

Issue actions