Meta to launch AI language model LLaMA to help researchers and take on ChatGPT
Meta announced on Friday it is releasing a new language model based on artificial intelligence as the generative AI race heats up, CEO Mark Zuckerberg said on Friday. The new language model is intended for use in research and will be available for academic researchers, government, civil society and academic organizations, and industry research.
The language model, called LLaMA, is designed to generate text and conversations, summarize written material, and perform complicated tasks like solving math theorems or predicting protein structures. In a Facebook post, Meta said:
“We introduce LLaMA, a collection of foundation language models ranging from 7B to 65B parameters. We train our models on trillions of tokens, and show that it is possible to train state-of-the-art models using publicly available datasets exclusively, without resorting to proprietary and inaccessible datasets. In particular, LLaMA-13B outperforms GPT-3 (175B) on most benchmarks, and LLaMA-65B is competitive with the best models, Chinchilla70B and PaLM-540B. We release all our models to the research community.”
Also in a Twitter post, Research Scientist at Facebook AI Research Guillaume Lample said:
“Today we release LLaMA, 4 foundation models ranging from 7B to 65B parameters. LLaMA-13B outperforms OPT and GPT-3 175B on most benchmarks. LLaMA-65B is competitive with Chinchilla 70B and PaLM 540B.”
Today we release LLaMA, 4 foundation models ranging from 7B to 65B parameters.
LLaMA-13B outperforms OPT and GPT-3 175B on most benchmarks. LLaMA-65B is competitive with Chinchilla 70B and PaLM 540B.
The weights for all models are open and available at https://t.co/q51f2oPZlE
1/n pic.twitter.com/DPyJFBfWEq— Guillaume Lample @ ICLR 2024 (@GuillaumeLample) February 24, 2023
The ChatGPT soared in popularity after its release on November 30, 2022. The OpenAI’s AI-powered chatbot could do virtually everything from writing poetry, and correcting coding mistakes with detailed examples, to generating AI art prompts. It can even explain the worst-case time complexity of the bubble sort algorithm.
In just two months, ChatGPT reached 100 million monthly active users in January, making it the fastest-growing consumer application in history. Last week, the company announced it was launching ChatGPT Plus, a $20 per month pilot subscription plan for its popular AI-powered chatbot. The sudden success of ChatGPT has put more pressure on other tech companies including Google, Baidu, and Alibaba.
ChatGPT is a dialogue-based AI chat interface for its GPT-3 family of large language models. The mind-blowing ChatGPT is a successor to the Generative Pre-trained Transformer 3 (GPT-3), which is an autoregressive language model that uses deep learning to produce human-like text.
ChatGPT was adapted from the GPT-3.5 model but trained to provide more conversational answers. However, unlike the GPT-3, the new ChatGPT demonstrates an array of new capabilities including the ability to engage in conversations with users, write software codes, tell stories, and provide solutions to complex mathematical equations. The beauty of it all is that you can test out the new chatbot for yourself right here.
You can download a PDF copy of the LLaMA Whitepaper below.
[pdf-embedder url=”https://techstartups.com/wp-content/uploads/2023/02/333078981_693988129081760_4712707815225756708_n.pdf”]