Dolly logo

Dolly

Paid

Automate data migration, transfer securely & quickly, between databases, cloud storage, various formats & types.

4.2
Inputs: textOutputs: text
Type
Saas
Company
Databricks

About Dolly

Dolly is an open source project from Databricks Labs that makes the process of moving data between different data sources faster, simpler, and more secure. With Dolly, users can easily and quickly move data between databases, cloud storage, and data lakes, with the assurance of strong authentication and encryption. Dolly supports a wide range of data formats and types, so users can conveniently move data between any two compatible sources. What’s more, Dolly is incredibly easy to use; users can set up and configure their data transfers in just a few clicks. With Dolly, users can be assured that their data transfers will be secure, efficient, and reliable.

Key Features

Automate data migration between databases and cloud storage.
Move data between various data formats and types.
Transfer data securely and quickly with simple setup.

Pros & Cons

Pros
  • Surprisingly high-quality instruction following for a 12B parameter model
  • Open-source and freely available with commercial license
  • Fine-tuned on diverse, human-generated instruction data covering multiple domains
  • Easy to access via Hugging Face and GitHub
  • Trained on a reproducible platform (Databricks)
Cons
  • Not state-of-the-art; lags behind larger or more modern models
  • Struggles with complex prompts, programming, mathematics, factual accuracy, and open-ended QA
  • Prone to hallucination and biases inherited from training data (The Pile)
  • Lacks some capabilities present in the base model (e.g., well-formatted letter writing)
  • Limited dataset size (~15k records) compared to many contemporary LLMs

Best For

Automate data migration between databases and cloud storage.Move data between various data formats and types.Transfer data securely and quickly with simple setup.

Alternatives to Dolly

FAQ

What is Dolly?
Dolly is an open-source, instruction-following large language model created by Databricks, based on Pythia-12b and fine-tuned on ~15k human-generated instruction/response pairs.
What can Dolly be used for?
Dolly is designed for instruction-following tasks including brainstorming, classification, closed and open QA, generation, information extraction, and summarization.
Is Dolly free to use?
Yes, Dolly is open-source and licensed for commercial use. The model is available on Hugging Face and the training code and data are on GitHub.
What are the limitations of Dolly?
Dolly is not state-of-the-art and struggles with complex syntax, programming, math, factual accuracy, date/time handling, open-ended QA, hallucination, list enumeration, stylistic mimicry, and humor. It also reflects biases from its training corpus.