A good instruction dataset in Italian, generated with LLM-aided translation.
The starting point has been mlabonne/FineTome-100k: a subset of arcee-ai/The-Tome, re-filtered using HuggingFaceFW/fineweb-edu-classifier to identify examples with good educational value. FineTome is a high quality dataset containing diverse elements such as conversations, reasoning problems, and more.
The translation process includes these… See the full description on the dataset page:
https://huggingface.co/datasets/anakin87/fine-instructions-ita-70k.
