datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
AfriGuard-inst
AfriGuard-inst
Alpaca-style instruction-tuning data for safety-aligned fine-tuning, built from
the train/validation splits of
israel/AfriGuard and
israel/AfriGuard-XL.
Test splits are excluded and reserved for evaluation.
Construction
AfriGuard (10 languages): each row yields TWO items — one English
(prompt/response) and one native-language
(prompt_translated/response_translated).
AfriGuard-XL (34 split configs, English-only): each row yields one item.
Rows with… See the full description on the dataset page: https://huggingface.co/datasets/israel/AfriGuard-inst.SNETSNET_Archive