Synapse Espresso: Using OpenAI GPT-3 in Synapse!

Welcome to the 22nd video in our Synapse Espresso series!

In this video, Thomas joins Stijn to show how we can make use of Azure OpenAI GPT-3 from within Spark in Synapse Analytics. GPT-3 is a natural language AI model and it’s able to understand text that you feed into it and then generate now text based on your input.

For the integration between Synapse and Azure OpenAI, SynapseML is used. This open-source library simplifies the creation of scalable Machine Learning pipelines. One of the features it provides is the ability to send requests to Azure Cognitive Services using APIs. It’s this feature that allows to easily access Azure OpenAI from Synapse Analytics.

The spark notebook showcases an example of sentiment analysis on restaurant reviews using GPT-3.

The reviews are loaded into a dataframe and the corresponding prompts are written, the SynapseML OpenAICompletion object is defined and finally the prompts are sent to OpenAI for analysis.

Related blog post: https://techcommunity.microsoft.com/t5/azure-synapse-analytics-blog/using-openai-gpt-in-synapse-analytics/ba-p/3751815 (opens in new tab)
SynapseML: https://microsoft.github.io/SynapseML/ (opens in new tab)
What is Azure OpenAI? https://learn.microsoft.com/en-us/azure/cognitive-services/openai/overview (opens in new tab)

Date: