llama.cpp
Which commits of llama.cpp and llama-ccp-python did u use. I don't get it running.
I actually run it by ollama. I tried llama.cpp before, but seems something run about the c compiler.
I get a seg fault using latest llama.cpp, other models work
Yes, when using the latest version of llama.cpp, we encounter the following error: llama_model_load: error loading model: error loading model hyperparameters: key qwen35.rope.dimension_sections has wrong array length; expected 4, got 3.
To resolve this, you need to switch to a specific commit and build an older version - git checkout $(git rev-list -n 1 --before="2026-03-01" master)
To resolve this, you need to switch to a specific commit and build an older version - git checkout $(git rev-list -n 1 --before="2026-03-01" master)
Hmm, I encounter the same error even after building llama.cpp binary from the revision you provided. Do you have a specific commit hash which works?
To resolve this, you need to switch to a specific commit and build an older version - git checkout $(git rev-list -n 1 --before="2026-03-01" master)
Hmm, I encounter the same error even after building llama.cpp binary from the revision you provided. Do you have a specific commit hash which works?
I don't think this is necessary anymore, all the quanta from here already work on the latest version of llama.cpp - https://huggingface.co/HauhauCS/Qwen3.5-9B-Uncensored-HauhauCS-Aggressive