Unable to run new gemma3 models `unknown model architecture: 'gemma3'`

[hanthor]

 ramalama run gemma3:12b
 
:: initializing oneAPI environment ...
   entrypoint.sh: BASH_VERSION = 5.2.32(1)-release
   args: Using "$@" for setvars.sh arguments: llama-run -c 2048 --temp 0.8 --ngl 999 /mnt/models/model.file
:: compiler -- latest
:: mkl -- latest
:: tbb -- latest
:: umf -- latest
:: oneAPI environment initialized ::
 
Loading modelget_memory_info: [warning] ext_intel_free_memory is not supported (export/set ZES_ENABLE_SYSMAN=1 to support), use total memory as free memory
llama_model_load: error loading model: error loading model architecture: unknown model architecture: 'gemma3'
llama_model_load_from_file_impl: failed to load model
initialize_model: error: unable to load model from file: /mnt/models/model.file

[hanthor]

ramalama list --json
[{"name": "ollama://gemma3:12b", "modified": "2025-03-14T17:30:05.122487+00:00", "size": 8149180864}, {"name": "ollama://deepseek-r1:7b", "modified": "2025-03-07T06:59:28.841018+00:00", "size": 4683073184}, {"name": "ollama://llama3.2:3b", "modified": "2025-03-05T10:31:40.511003+00:00", "size": 2019377376}, {"name": "huggingface://bartowski/DeepSeek-R1-Distill-Llama-8B-GGUF/DeepSeek-R1-Distill-Llama-8B-Q6_K.gguf", "modified": "2025-02-18T05:55:48.179422+00:00", "size": 6596008800}]

[hanthor]

just got added upstream 4 days ago ggml-org/llama.cpp#12343

[hanthor]

I probably need to try the gguf from huggingface

[ericcurtin]

It will be in the next release of RamaLama #973

[rhatdan]

That is merged so closing.