]> git.djapps.eu Git - pkg/ggml/sources/llama.cpp/commit
model: MTP support for Qwen3-Next (#25589)
authorMasashi Yoshimura <redacted>
Mon, 3 Aug 2026 08:15:01 +0000 (17:15 +0900)
committerGitHub <redacted>
Mon, 3 Aug 2026 08:15:01 +0000 (17:15 +0900)
commit4ed2b13f758ea467282cf0b0e1a938fca7f51211
treef93b0e8e2aba9ef537f49e89b3185e8e39e74206
parent2b63e0610bbc2be990ae1360d5256efcdc3f9efb
model: MTP support for Qwen3-Next (#25589)

* mtp for qwen3nex

* fix for python type-check

* Fix to compute num_mtp from directly mtp layer

* define opt_num_mtp_layers in _QwenMtpMixin and fix some comments

* Fix for python type check

* Update gguf-py/gguf/constants.py

Co-authored-by: Sigbjørn Skjæret <redacted>
* rebase and add load_mtp flags

* Update src/models/qwen3next.cpp

Co-authored-by: Sigbjørn Skjæret <redacted>
* Update src/models/qwen3next.cpp

Co-authored-by: Sigbjørn Skjæret <redacted>
---------

Co-authored-by: Sigbjørn Skjæret <redacted>
conversion/qwen.py
gguf-py/gguf/constants.py
src/llama-model.cpp
src/models/models.h
src/models/qwen3next.cpp