Model Card for Qomhrá

Qomhrá is a bilingual Irish-English LLM. It is an activation and language aware quantized model. It is based on the training of Qwen3-8B with Irish and English data.

Acknowledgements

I would like to thank my academic supervisor Dr. Barry Devereux of Queen's University Belfast for his sustained guidance for rigorous experimentation. I would also like to thank the Abair (abair.ie) research group for their financial support and Irish language technology expertise. I would like to thank Jerry Sweeney, Managing Director at CloudCIX Limited for the generous donation of computing resources. Finally, I would like to Khan-Tung Tran, PhD Candidate at University College Cork, for the open-sourcing of benchmarks, training data and guidance.

Citation

@misc{mcinerney2025qomhra,
  author       = {Joseph McInerney},
  title        = {Qomhrá: A Bilingual Irish–English Large Language Model},
  year         = {2025}
}
Downloads last month
200
Safetensors
Model size
8.19B params
Tensor type
F16
·
Inference Providers NEW
This model isn't deployed by any Inference Provider. 🙋 Ask for provider support

Model tree for jmcinern/qwen3-8B-cpt-sft-awq

Finetuned
(1)
this model

Space using jmcinern/qwen3-8B-cpt-sft-awq 1