10:24
09:52
11:01
10:52
01:59
01:33
10:24
09:52
11:01
10:52
01:59
01:33
10:24
09:52
11:01
10:52
01:59
01:33
10:24
09:52
11:01
10:52
01:59
01:33
Meta is hoping that others will utilize this research tool to tackle some of the issues that afflict AI language models, rather than being a conversational system.
Recently, the attention in the tech industry has been centered on the language models developed by big companies such as Google, Microsoft, and OpenAI. However, Meta, the parent company of Facebook, is also actively engaged in this field and has just unveiled their new AI language generator, LLaMA.
Unlike ChatGPT or Bing, LLaMA is not a chatbot that people can talk to. It is a research tool that Meta intends to share with the aim of making this important, rapidly evolving field more accessible. The company hopes that LLaMA will help researchers tackle the issues associated with AI language models, including bias, toxicity, and the generation of false information.
Meta has released LLaMA, which is composed of four different-sized models, under a noncommercial license focused on research use cases. Groups such as universities, NGOs, and industry labs will be granted access to this tool. Meta believes that the AI community, including academic researchers, civil society, policymakers, and industry, must collaborate to develop clear guidelines for responsible AI, specifically responsible large language models.
Meta claims that the second-smallest version of the LLaMA model, LLaMA-13B, performs better than OpenAI's GPT-3 model on most benchmarks, while the largest, LLaMA-65B, is competitive with the best models like DeepMind's Chinchilla70B and Google's PaLM 540B. LLaMA-13B can even be run on a single data center-grade Nvidia Tesla V100 GPU, making it more accessible for smaller institutions.
Meta's release of LLaMA is noteworthy, as it differs from the company's previous forays into the AI chatbot space, which received criticism for their performance. With LLaMA, Meta hopes to receive a more positive reception from the research community.
In a Facebook post, CEO Mark Zuckerberg stated, "Meta is committed to this open model of research, and we'll make our new model available to the AI research community."