At an event in London on Tuesday, Meta confirmed that it plans an initial release of Llama 3, the next generation of its large language model used to power generative ai assistants, within the next month.
This confirms a report published on Monday by The Information that Meta was approaching the launch..
“Within the next month, actually less, hopefully in a very short period of time, we hope to begin rolling out our new set of next-generation base models, Llama 3,” said Nick Clegg, president of global affairs at Meta. He described what appears to be the release of several different iterations or versions of the product. “There will be several different models with different capabilities and different versatility (launched) over the course of this year, starting very soon.”
The plan, added Chris Cox, Meta's chief product officer, will be to push multiple products on Meta with Llama 3.
Meta has been struggling to catch up with OpenAI, taking it and other big tech companies like Google by surprise when it launched ChatGPT more than a year ago and the app went viral, turning ai generative questions and answers into everyday experiences. and conventional.
Meta has largely taken a very cautious approach to ai, but that hasn't gone down well with audiences, and previous versions of Llama were criticized for being too limited. (Llama 2 was released publicly in July 2023. The first version of Llama was not released to the public, but was still leaked online.)
Llama 3, which has a larger scope than its predecessors, is expected to address this, with capabilities not only to answer questions more accurately but also to answer a broader range of questions that could include more controversial topics. He hopes this will make the product successful among users.
“Our long-term goal is to make a Llama-powered Meta ai the world's most useful assistant,” said Joelle Pineau, vice president of ai research. “There's quite a bit of work ahead to get there.” The company did not talk about the size of the parameters it is using in Llama 3, nor did it offer demonstrations of how it would work. It is expected to have around 140 billion parameters, compared to 70 billion for the larger Llama 2 model.
In particular, the Llama de Meta families, created as open source products, represent a different philosophical approach to how ai should be developed as a broader technology. By doing so, Meta hopes to gain favor with developers over more proprietary models.
But it seems that Meta is also acting more cautiously, especially when it comes to other generative ai beyond text generation. The company has not yet released Emu, its imaging tool, Pineau said.
“Latency is very important, along with security and ease of use, to generate images that you are proud of and that represent whatever your creative context is,” Cox said.
Ironically, or perhaps unsurprisingly (heh), even as Meta works to release Llama 3, it has some major generative ai skeptics at home.
Yann LeCun, the celebrated ai academic who is also Meta's chief ai scientist, criticized the limitations of generative ai in general and said his bet is on what comes next. He predicts there will be a Joint Integrated Prediction Architecture (JEPA), a different approach to both training models and producing results, which Meta has been using to build more accurate predictive ai in the area of ai.meta.com/blog/v-jepa-yann-lecun-ai-model-video-joint-embedding-predictive-architecture/” target=”_blank” rel=”noopener”>image generation.
“The future of ai is JEPA. It’s not generative ai,” he said. “We'll have to rename Chris's product division.”
