Databricks tool helps businesses create their own AI model

Data lakehouse and AI company Databricks has released a new tool for allowing businesses to develop large language models with capabilities evocative of ChatGPT at a low cost.

The AI model has the same instruction following abilities of ChatGPT, but uses the open source EleutherAI model, which has only six billion parameters compared to the 175 billion in GPT-3.

Dolly modifies the model slightly to elicit instruction-following capabilities such as brainstorming and text generation not present in the original model, using data from Alpaca, according to Databricks co-founder and CEO Ali Ghodsi.

“We’re calling the model Dolly — after Dolly the sheep, the first cloned mammal — because it’s an open source clone of an Alpaca, inspired by a LLaMA (Large Language Model Meta AI),” he said.

“We’re in the earliest days of the democratisation of AI for the enterprise, and much work remains to be done, but we believe the technology underlying Dolly represents an exciting new opportunity for companies that want to cheaply build their own instruction-following models.”

He said many companies will prefer to build their own model rather than sending data to a centralised large language model provider such as GPT-3. These include protecting their sensitive and proprietary data and intellectual property rather than handing it over to a third party.

Companies are also likely to have different trade-offs between model quality, cost and desired behaviour, Ghodsi added.

Image credit: iStock.com/1715d1db_3