Dolly 2.0 is an open source, instruction-following large language model (LLM) fine-tuned on a high-quality human-generated data set licensed for research and commercial use. Developed by Databricks, this 12-billion parameter language model aims to empower organizations to create, own, and customize powerful LLMs that can interact with people without relying on third-party APIs. The dataset, databricks-dolly-15k, consists of 15,000 human-generated prompt/response pairs designed for instruction tuning LLMs, and is available under a Creative Commons Attribution-ShareAlike 3.0 Unported License. By crowdsourcing among Databricks employees, the dataset represents a wide range of behaviors and provides a commercially viable model for organizations looking to harness the power of LLMs.

Read more on the Databricks blog or Ars Technica.