Meta is making waves in the AI community by announcing the latest version of its open source Llama artificial intelligence (AI) model.
As the AI arms race continues to heat up, industry experts are taking a closer look at the capabilities of Meta’s Llama 3 model and comparing it to competing products from tech giants like Google, OpenAI, and Anthropic. With the promise of improved performance and accessibility, the Llama 3 model has the potential to reshape the competitive landscape and democratize access to AI.
“Llama 3 released by Meta is poised to set a new benchmark in the open source generative AI market with advanced features.” Moses GutmanCEO of an AI company clear MLhe told PYMNTS. “By pushing the boundaries of what open source models can achieve, Llama 3 challenges other contributors to improve their own services and accelerate innovation across the board. We reaffirm our belief that this technology has the potential to compete with closed-source alternatives.”
Chatbot upgrade?
Meta CEO Mark Zuckerberg Published in thread Announcing the new Llama 3 AI model. This powers Meta AI and is now integrated into various Meta products such as Messenger and Instagram as well as dedicated websites.
“We are upgrading MetaAI with the new cutting-edge Llama 3 AI model that we are open sourcing,” Zuckerberg wrote. “We believe that this new model makes Meta AI the most intelligent AI assistant at our disposal.”
Zuckerberg said two smaller versions of the Llama 3 model, with 8 billion and 70 billion parameters, respectively, are better than other free models based on standard performance benchmarks used to assess model quality. It was announced that it showed good performance in comparison. He also noted that the largest version, Llama 3, with 400 billion parameters, is still in training.
Llama 3 is trained on a dataset that is 7x larger than its predecessor, Llama 2. In fact, companies prioritize both dataset size as well as dataset quality and diversity in order to obtain satisfactory output. Nevertheless, the larger the set of parameters on which the model is trained, the better it will be able to understand the context between words.
Cloris ChenCEO of cogito financediscussed with PYMNTS the benchmarks that Meta highlighted in its announcement, noting that they include standard industry metrics such as MMLU and Meta’s proprietary assessment tests. In these tests, the Llama 3 70B Instruct model reportedly outperformed competing products such as GPT-3.5 and Claude Sonnet. Chen noted that while Meta has already addressed these concerns, this has sparked a debate about the objectivity of these tests.
One of the main advantages of Llama 3 is that it comes in two sizes: a small model and a large model. Austin VanceCEO of a digital transformation company intensive labhe told PYMTS. He said this smaller model is better than the accessible version of his GPT 3.5 Turbo in OpenAI, offering faster inference times and the benefits of being open source.
“This allows businesses to deploy the model into their infrastructure at a relatively affordable price.” This gives you control over your security environment, data, and development workflows without the significant costs associated with OpenAI. ” he added.
Vance said the larger Llama 3 model is also impressive, surpassing all previous open source models in terms of information recall, mathematical ability, and conversational style. He said the recall ability was particularly noteworthy, and the natural flow of conversation was evidence of the model’s sophistication.
“With Llama3, Meta positions itself as a challenger to Google and Microsoft in the open source AI space, fulfilling the role that Elon Musk envisioned for OpenAI,” he added.
mike finleyco-founder of answer rockettold PYMNTS that the new llama model has about 8 billion neurons, comparable to the brain of the largest parrot. He said the model appears to have been trained for a much longer period of time than other open source models.
“In fact, this means we are getting closer to a model that can live on. edge device (Phones and tablets first, then POS [point of sale] devices and cars, then chip cards and pacemakers),” he added.
Growing open source AI movement
Gutmann emphasized that Meta’s new AI model is compelling evidence that open source software can rival proprietary or “closed” models in performance and innovation. He highlighted how Meta’s approach to making his company’s AI technology more accessible could challenge the dominance of closed systems.
“Furthermore, as the capabilities of these models improve, they will foster more collaborative projects and partnerships within the open source community, fostering a more vibrant ecosystem,” he added.
Meta is one of several companies driving open source AI software efforts. Similarly, Elon Musk’s xAI is making waves by open sourcing its AI chatbot Grok, a step that could democratize AI technology and foster innovation in commercial use. Many people think so.
Grok models are available to X Premium+ subscribers via platforms such as GitHub and BitTorrent. The release coincides with Musk’s broader criticism of OpenAI’s lack of transparency, highlighted by his legal standoff with OpenAI over its AI models.
“Llama 3 is a very important development in the open source community, and one of the main reasons for that is the unique size of the dataset (as was the case with the 70B version),” said Chen. “Interestingly, when analyzing individual results across standard benchmarks, the 70B Instruct model is nearly on par with Google’s Gemini.”
