error loading model architecture: unknown model architecture: 'gemma4'
While trying to test this model I'm getting the error message shown in the title, I'm running 'version: 8610 (2b86e5cae)', probably some update will need be done by llama.cpp
Same error here.
llama_model_load: error loading model: error loading model architecture: unknown model architecture: 'gemma4
solved in the latest release b8642
is docker image update for llama cpp server ?
The paths are still messed up, I spent a good amount gaslighting myself thinking I was doing something wrong (new to local llms), but..
If you take the unsloth gguf and recreate the model file, problem is fixed:
cat > Modelfile << 'EOF'
FROM /home/j/.ollama/models/blobs/sha256-5dd716a53e80db081ace2d4683887f76ad806e04bb6cd4551cb7699da3cf6bcc
EOF
ollama create unsloth-gemma4-xyz -f Modelfile
ollama run unsloth-gemma4-xyz
The paths are still messed up, I spent a good amount gaslighting myself thinking I was doing something wrong (new to local llms), but..
If you take the unsloth gguf and recreate the model file, problem is fixed:
Thank you for sharing the solution.