Databricks

lastmod 2024-06-04created 2023-04-182 edits

Released Dolly 2.0, similar to ChatGPT, as open source

Features

• The world's first truly open Instruction-Tuned LLM
• Releases all of the training code, dataset, and model weights. That is, any individual or company can create and own their own powerful LLM
• Fine-tuned on databricks-dolly-15k, a human-generated instruction dataset
  • 15,000 prompt/response pairs. Anyone can modify/extend it, and it can also be used for commercial purposes
    • (Alpaca, Koala, GPT4All, Vicuna, etc. all cannot be used commercially)
  • This data was written by 5,000 databricks employees themselves
• Based on the EleutherAI pythia 12B parameter language model

See also

• ChatGPT
• Dolly 2.0
• Machine Learning Platforms
• Dataiku
• Alteryx
• Knime
• Datarobot
• Databricks
• Sagemaker

Favorite site

• Free Dolly: Introducing the World's First Open and Commercially Viable Instruction-Tuned LLM - The Databricks Blog