]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
chat : nemotron thinking & toolcalling support (#15676)
authorPiotr Wilkin (ilintar) <redacted>
Thu, 4 Sep 2025 23:22:22 +0000 (01:22 +0200)
committerGitHub <redacted>
Thu, 4 Sep 2025 23:22:22 +0000 (01:22 +0200)
commitb2426e469e2fdb6c44216d56baa4cfff4f39ae00
treeb8c6cfc808cf06602833cbefa88edf0c3cb16307
parent9e2b1e83c68a38ea0c64f726dd979439bd02189b
chat : nemotron thinking & toolcalling support (#15676)

* feat: nemotron thinking & toolcalling support

* Trailing whitespaces

* Corrected template for Nemotron

* Template and parser fixes

* Final template and grammar changes

* Whitespace

* Always do lazy grammar processing since </think> tag will always be there.

* Allow extra content after toolcall

* Whitespace

* New tests: thinking + tools, tools + content, thinking + tools + content (new!)

* Whitespace

* Remove cURL test script
common/chat.cpp
common/chat.h
models/templates/NVIDIA-Nemotron-Nano-v2.jinja [new file with mode: 0644]
tests/test-chat.cpp