ChatGPT, the next-generation chatbot from OpenAI, has drawn a lot of attention to conversational AI. Now, Altostra, which provides a no-code infrastructure platform for cloud application development, has built a template intended to quickly launch ChatGPT-like AI applications on Amazon Web Services.
Applications can be built in fewer than five minutes, the company said.
The company has developed an open source GTC-3 app template for deploying applications based on OpenAI’s GPT-3 text models on AWS.
The open source, serverless web application integrates any of OpenAI’s four GPT-3 models with a web UI. Altostra on February 7 posted instructions on its website, with the process requiring a signup for Altostra.
OpenAPI’s GPT-3 (Generative Pre-trained Transformer 3) models are advanced language message models that can generate human-like text, Altostra noted.
They serve as a type of transformer neural network, a deep learning model for natural language processing.
ChatGPT is an OpenAI model designed to respond to text inputs in a human-like fashion. It is based on the GPT-3 language models and utilises training to generate responses in a natural language format.
Altostra’s GPT-3 applications draw on a number of AWS services.
These include Amazon Simple Storage Service (S3) to store the web UI, Amazon CloudFront to host the web UI, Amazon API Gateway, and three AWS Lambda functions to process user requests and generate text using GPT-3. Amazon Simple Systems Manager is used to securely store the GPT-3 API key.