HuggingFaceH4/ultrachat_200k
Viewer • Updated • 515k • 86.2k • 1.18k
Instruction Datasets
Curated SFT datasets for instruction-following and conversational fine-tuning
Note Multi-turn conversations, 515K samples, MIT license - Best for chat
teknium/OpenHermes-2.5
Viewer • Updated • 1M • 13.1k • 928Note GPT-4 generated, 1M samples, Apache 2.0 - High quality general purpose
Open-Orca/OpenOrca
Viewer • Updated • 2.94M • 12.9k • 1.63kNote GPT-3.5/4 augmented FLAN, 2.94M samples, MIT - Large scale training
mlabonne/FineTome-100k
Viewer • Updated • 100k • 20.2k • 290Note Curated subset, 100K samples, Apache 2.0 - Quick experiments
-
bio-nlp-umass/bioinstruct
Viewer • Updated • 25k • 158 • 25 -
CodeDevX/Vibe-Coding-Instruct
Viewer • Updated • 1.1M • 201 • 191 -
ox-ox/mythos-character-distillation
Viewer • Updated • 551 • 257 • 8 -
ansulev/claude-mythos-distilled-25k
Viewer • Updated • 25k • 272 • 6