`server`: add `--reasoning-budget 0` to disable thinking (incl. qwen3 w/ enable_thinking:false)#13771

Merged

ochafik merged 13 commits intoggml-org:masterfrom

ochafik:enable-thinking

May 25, 2025

Commits on May 25, 2025

server: fix/test add_generation_prompt
ochafik
committed
tools: enable hermes2/qwen chat logic even w/o tools
ochafik
committed
server: add --reasoning-format=disabled to disable thinking (incl. qwen3 w/ enable_thinking:false)
ochafik
committed
Update README.md
ochafik
committed
Add models/templates/Qwen-Qwen3-0.6B.jinja
ochafik
committed
update --reasoning-format={disabled -> nothink} as suggested
ochafik
committed
fix command r7b's nothink w/ official template
ochafik
committed
rewrite docs as list as suggested
ochafik
committed
Update common/chat.cpp

ochafik
and
ngxson
authored
Merge branch 'enable-thinking' of github.com:ochafik/llama.cpp into enable-thinking
ochafik
committed
const char* return for chat enum name helpers
ochafik
committed
switch to --reasoning-budget flag
ochafik
committed
Merge branch 'fix-gen-prompt' into enable-thinking
ochafik
committed