New Delhi: After Microsoft ChatGPT and Google’s Bard, Meta AI joins the race for chatbots with its state-of-the-art foundational big language model designed to help researchers advance their work in the field of artificial intelligence . However, Meta’s Large Language Model Meta AI (LLaMA) isn’t quite like ChatGPT-powered Bing at the moment because it can’t talk to humans yet but will help researchers.
“Smaller, higher-performing models like LLAMA enable others in the research community, who may not have access to the large amount of infrastructure required to study these models, to perform this important, rapid research,” Meta said in a statement. further democratizes access to an ever-changing field.” ,Also read: “Dear SBI User…:” Are you also getting this SMS? Check The Truth About SBI Fake Message Scam,
Meta is making the LLaMA available in several sizes (7 billion, 13 billion, 33 billion and 65 billion parameters). Large language models – natural language processing (NLP) systems with billions of parameters – have opened up new capabilities for generating creative text, solving mathematical theorems, predicting protein structures, answering reading comprehension questions, and much more. are shown. ,Also read: From SBI to BOB: Here are 5 public sector bank FDs compared – Check latest fixed deposit rates for senior citizens,
“They are one of the clearest cases of the substantial potential benefits that large-scale AI could offer to billions of people,” Meta said. Smaller models trained on more tokens—pieces of words—are easier to retrain and fine-tune for specific potential product use cases.
The meta has trained LLaMA 65 billion and LLaMA 33 billion on 1.4 trillion tokens. “Our smallest model, LLaMA 7B, is trained on one trillion tokens,” the company said. Like other large language models, LLaMA works by taking a sequence of words as input and predicting the next word to generate text recursively.
“To train our model, we chose text from one of the 20 languages with the highest number of speakers, focusing on those with the Latin and Cyrillic alphabets,” explained Meta. To maintain integrity and prevent misuse, the company said it is releasing the model under a non-commercial license focused on research use cases at this time.