American startup publishes source code for the emergence of new chatbots

launch Data blocks released an open source code this Friday the 24th that will allow companies to build their own chatbots modeled after ChatGPT. The code is a template Intelligence Artificial trained on datasets and can be trained to perform multiple.

With this initiative, the company aims to introduce a viable alternative to AI learning called the big language model, the same one that powers the famous AI robot. Open AI.

Second There godsi, CEO of Databricks, the goal is to enable organizations to train their own chatbots with the startup’s software and without having to share their data. “In the future, everyone will have their own model and will be able to really develop it,” the head said.

However, he also stated that while there are features such as writing blog posts in open source, the company has yet to do benchmark tests to prove it is as effective as ChatGPT.

Cost and size reduction

In the current technology marketplace, with startups raising millions of dollars in venture capital to train AI models, Databricks is presenting an initiative that can capture the attention of entrepreneurs.

At the same time, such gigantic organizations as Google This Target, work on reducing the cost and size of these models to improve their accuracy. “I think eventually you will make fewer and fewer of them, and they will be open source,” Ghodsi said. OpenAI trains its chatbots on supercomputers and with huge amounts of data. The costs of doing so, according to its CEO, are high.

While the Databricks initiative is new and will still need to be used frequently for evaluation, it could provide more options for companies that want to build their own models, following the trend towards more affordable AI technology.

Source: Reuters.

…..