SKU SFT-2007 · Sold by External
Databricks Dolly 15k
Product specifications
| SKU | SFT-2007 |
|---|---|
| Data type | SFT dataset |
| Volume | 15011 rows |
| Size on disk | 13.1 MB |
| Format | JSONL |
| Access model | PUBLIC LICENSE |
| Pricing | Free · open-source license |
| Quality score | — |
| License | CC-BY-SA-3.0 |
An open-source, entirely human-generated instruction-following dataset created by more than 5,000 Databricks employees. Records span eight InstructGPT-style categories — brainstorming, classification, closed QA, generation, information extraction, open QA and summarization — with some closed/extraction items citing Wikipedia passages in a context field. Each record has instruction, optional context, response and category fields.