This dataset consists of 50,000+ instruction-style samples derived from real anonymized legal decisions collected from the e-Məhkəmə portal. The data was carefully processed to create instruction-response pairs tailored to Azerbaijani legal scenarios.
It was used to fine-tune the azlegllm-4b model, which provides legal assistance in Azerbaijani.
question: Legal questions phrased as instructions or real… See the full description on the dataset page:
https://huggingface.co/datasets/nijatzeynalov/azlegllm-dataset.
