Mark Zuckerberg stated that he is more into AI, disbursing billions of amount for its innovative creation and enhancement (Image: Instagram @Mark Zuckerberg)
Since people are more interested in artificial intelligence, tech giants are combating each other to provide the best models for the masses. Many have introduced AI models and are on a mission to consistently unveil AI-infused smartphones. But Facebook owner Mark Zuckerberg crowns the AI revolution bestowing free and highly capable chatbots. On Monday, Meta rolled out its new Llama 3.1 which is considered the largest language model. Llama 3.1 405B is exceptional among the available models having unmatched flexibility, and control, possessing a wide range of capabilities that rival the best-sourced close models.
The new model allows the users to step into exploring more diverse features unlocking new workflows including model distillation and synthetic data generation. The new 3.1 model is also free of cost. The company didn’t reveal the charge of developing Lama 3.1 but Mark Zuckerberg stated that he is more into AI, disbursing billions of amount for its innovative creation and enhancement.
Through this latest release, Meta is showing that the closed approach favoured by most AI companies is not the only way to develop AI. But the company is also putting itself at the centre of the debate over the dangers posed by releasing AI without controls. Meta trains Llama in a way that prevents the model from producing harmful output by default, but the model can be modified to remove such safeguards. The company claims the new model could be considered the smartest AI on Earth to date.
Also, read| Haven’t You Tried Meta AI On WhatsApp Yet? Here Is A Sneak-Peek
On the release of this new Llama 3.1, the company evaluates performance on over 150 benchmark datasets that come with an array of multiple languages. Additionally, it performs extensive human evaluations that compare Llama 3.1 with viewing models in real-world scenarios. This flagship model might be a rival to leading foundations across a range of tasks, including GPT-4, GPT-4o, and Claude 3.5 Sonnet. In addition, the mini models are competitive with open and closed models with almost similar parametric quantities as the company used the 405B parameter model to improve the post-training quality of its mini versions.
Also, read| Meta In Talks With Apple Over AI Partnership: Report
The new 3.1 Llama model has improved the quality of the data for pre and post-training when it is compared to the previous versions. This may include meticulous pre-processing and data curation for pre-training data. The company stated that the notion of collaborating with industry, and start-ups would help them to define the interface of the fragmented components in a better way as this vision needs to be implemented without further atomisation.
