Meta platforms by Mark Zuckerberg plan to publish the latest version of his Artificial intelligence Grande Language Model Llama 3 In July, which would give better answers to the controversial questions asked by users, Information reported Wednesday.
The meta-researchers try to “loosen” the model so that it can at least provide a context to a question which he considers controversial.
The effort, which would stimulate the usefulness of the new LLM of Meta, is rival Google paused the image generation function on its gemini ai After its production historical images which were sometimes inaccurate.
Meta’s Llama 2, which feeds cat robots on its social media platforms, refuses to answer less controversial questions such as the way of targeting a friend, winning a war or killing a car engine, according to publication tests.
Llama 3, however, would be able to understand questions such as “how to kill the engine of a vehicle”, which means how to stop it rather than put an end to his life, according to the report, which cited several sources.
Meta also plans to appoint someone internally in the coming weeks to supervise training in tone and security as part of their efforts to make the responses of the model more nuanced, according to the report.
Meta did not immediately respond to a request for comments from Reuters.