Here's why this is cool: Dolly is a cheap-to-build LLM that exhibits a surprising degree of the instruction following capabilities exhibited by ChatGPT. Whereas the work from the Alpaca team showed that state-of-the-art models could be coaxed into high quality instruction-following behavior, the Databricks team found that even years-old open source models with much earlier architectures exhibit striking behaviors when fine tuned on a small corpus of instruction training data. Dolly works by taking an existing open source 6 billion parameter model from EleutherAI and modifying it ever so slightly to elicit instruction following capabilities such as brainstorming and text generation not present in the original model, using data from Alpaca. This means that these highly functional LLM patterns can be made available on much older, cheaper to run models, and is likely to continue the trend of democratizing access to generative AI.

Very cool. And Dolly is Apache 2.0 licensed, so it can be used for commercial purpose as well! Although it's based on GPT-J 6B, which is Apache 2.0 and Alpaca which is Creative Commons Non commercial.

this is fantatasttic, upvoted - I signed up by email for the weights a while back but still haven't recieved anything though

Congratulations on the launch of Dolly! It's impressive to see such a powerful language model created in such a short amount of time. Let's discuss how it will change the way we interact with AI!

Fantastic work by your launch team! We launched Edworking today as well! Do support us! (https://www.producthunt.com/post...)

Thank you for this, I think you are doing very important thing. It appears demand on such models is pretty high