ChatGPT Today we are releasing a new state-of-the-art AI Large Language Model called LLaMA, built to help researchers advance their work,” Zuckerberg wrote on his social media platforms. The LLM has shown great promise in producing language, participating in discussions, summarizing written material, and more challenging tasks, including determining protein structures or solving mathematical theorems. ChatGPT However, the meta exec did not specify which of these tasks, if any, LLaMA can currently perform.
ChatGPT, the only detail Zuckerberg offered about the big language model in Friday’s announcement is that his company is “committed to this open model of research, and we will provide the AI research community with our new model.”
ChatGPT In a company blog post published Friday afternoon that included a link to a full research paper on AI and its Github model map, Meta offered considerably more information. ChatGPT The company stated that “LLaMA works by accepting a series of words as input and predicting the next word to generate the text recursively.” This is how other major language models work. ChatGPT Meta said LLaMA was trained on text from 20 different languages. ChatGPT Its model map notes that the training data consisted of publicly available text from CCNet, C4, Wikipedia, ArXiv, and Stack Exchange.
Compared to other major language models, ChatGPT and its AI “need significantly less computing power and resources,” according to META. It said that the LLAMA would be available in several sizes. ChatGPT The company further emphasized its commitment to transparency and responsible AI development and reiterated that only AI researchers will be granted access.
ChatGPT To maintain the integrity and prevent abuse, we release our model under a non-commercial license focused on research use. Academic researchers will access the model on a case-by-case basis; ChatGPT is affiliated with organizations in government, civil society, and academia; and industrial research laboratories around the world,” the company wrote.
So far, ChatGPT at this stage of research, LLaMA has not been included in any of Meta’s products or platforms (including Instagram and Facebook), a company spokesperson told Gizmodo via email. ChatGPT Den spokesperson further added that the company has nothing to share at this time regarding the public preview or extended public access.
For now, ChatGPT, it’s hard to say how LLAMA will stack up against AI efforts from other tech giants in an increasingly crowded field. However, ChatGPT has come with its snafus in almost every major language model that has been launched recently. Both Google and Microsoft integrated AI text generators into their search platforms, and both advertised unintentional misinformation. ChatGPT Then there are the unresolved questions of copyright, privacy, whether AI can develop “emotions” or not, and how to deal with the many ChatGPT ChatGPT solutions on the Internet looking for ways to circumvent restrictions and generate offensive material or malware.
Meta has acknowledged some ChatGPT of these shortcomings in their blog post. To address the dangers of bias, harmful comments, and ChatGPT hallucinations in large language models, much more research is still needed. Like other models, the LLaMA has these difficulties, the company said. ChatGPT, however, seems like Meta believes its open, research-oriented approach is a step forward toward solving AI’s many, ongoing problems.
Those hoping to make use of ChatGPT from LLaMA themselves can apply via this request form.