Meta Platforms Inc, which owns Facebook, has announced the availability of a massive language model for artificial intelligence research. According to Meta, its model was the first 175-billion-parameter language model made available to the larger AI research community.

“Large language models” are natural language processing systems that can answer reading comprehension questions or generate new text after being trained on enormous amounts of material.

Meta stated in a blog post that the introduction of its “Open Pretrained Transformer (OPT-175B)” model will increase academics’ understanding of how huge language models function.

Access restrictions to such models, according to Meta, have “hampered progress on efforts to increase their robustness and alleviate known concerns such as bias and toxicity.”

Read This:-What new feature is WhatsApp working on for its users?

Artificial intelligence technology, which is a major field of study and development for numerous major internet platforms, has the potential to perpetuate humans’ cultural prejudices on problems such as race and gender. Some scholars are concerned about the problems that huge language models can propagate.

Meta stated that it “hoped to broaden the range of voices defining the ethical implications of such technology.”

To prevent misuse and “keep integrity,” the tech giant said it was distributing the model under a nonprofit license to focus on research use cases.

According to Meta, academic researchers and others linked with the government, civic society, and academic groups, as well as corporate research facilities, would have access to the model. The pre-trained models, as well as the code to train and utilize them, will be included in the release.

Take a second to support EPOST DESK on Patreon!


Please enter your comment!
Please enter your name here