Azerbaijani (instruction, response) pairs for supervised fine-tuning (SFT) of Azerbaijani language
models — part of an open Azerbaijani LLM stack. Instruction data is genuinely scarce for Azerbaijani;
this is both training data for our models and a reusable standalone artifact for anyone building
Azerbaijani instruction-following models.
seeds_az.jsonl — 45 hand-authored, high-quality seed pairs spanning 16 task… See the full description on the dataset page:
https://huggingface.co/datasets/kamaalg/azerbaijani-instructions.
