Qwen 3.5?

[willhsmit]

Any chance of a llama.cpp update to pick up support for Qwen 3.5?

[Stonedge]

just updating the requirements.txt file to use the latest version of the llamacpp (v0.83.0 https://github.com/oobabooga/llama-cpp-binaries/releases) and apply it (pip install -r requirements.txt) make it work flawlessly.

[willhsmit]

Is that windows + cu124 + nvidia?

Both edit to 0.83 and switching to the dev branch are pretty consistently giving me a

Error loading the model with llama.cpp: Server process terminated unexpectedly with exit code:
3221226505

for models that load just fine in main @0.74.

[Stonedge]

I'm on linux.... but on my side I first start with the 0.80 and it was working.... not sure for windows but here the modified line from the wheel section: https://github.com/oobabooga/llama-cpp-binaries/releases/download/v0.80.0/llama_cpp_binaries-0.80.0+cu124-py3-none-win_amd64.whl; platform_system == "Windows"

[willhsmit]

Followup: current main version as of Mar 7 dev->main pull is working great for me on both old models and 3.5, and I'm finding 3.5 to be a solid upgrade over 3 VL for image captioning tasks.