- By Vikas Yadav
- Fri, 19 Apr 2024 10:59 AM (IST)
- Source:JND
Meta has introduced Meta Llama 3, its latest open generative AI large language model (LLM). The company shared that it is announcing two models with eight billion and 70 billion parameters that can offer wide use cases. The models get improved reasoning, among other new capabilities. Meta claims these are "the best open source models of their class".
In the future, these text-based models are planned to gain multilingual and multimodal support and offer continued upgrades in coding and reasoning. The company claimed the training dataset for Llama 3 is seven times larger than Llama 2. For high-quality training, the company used a range of data-filtering pipelines.
"These pipelines include using heuristic filters, NSFW filters, semantic deduplication approaches, and text classifiers to predict data quality," Meta said. The new model also has safety features like Llama Guard 2, Code Shield, and CyberSec Eval 2. The social media giant said that Llama 3 will be available soon across major platforms "including cloud providers, model API providers, and much more".
Introducing Meta Llama 3: the most capable openly available LLM to date.
— AI at Meta (@AIatMeta) April 18, 2024
Today we’re releasing 8B & 70B models that deliver on new capabilities such as improved reasoning and set a new state-of-the-art for models of their sizes.
Today's release includes the first two Llama 3… pic.twitter.com/Q80lVTeS7m
The company has integrated the latest models in Meta AI that will be available in Meta apps. Users can use Meta AI on Instagram, Facebook, WhatsApp, Messenger and the web to start a chat. Plus, users will be able to test the multimodal Meta AI on their Ray-Ban Meta smart glasses.
The Llama 3 models will also be available to "AWS, Databricks, Google Cloud, Hugging Face, Kaggle, IBM WatsonX, Microsoft Azure, NVIDIA NIM, and Snowflake, and with support from hardware platforms offered by AMD, AWS, Dell, Intel, NVIDIA, and Qualcomm" soon.
"Our largest models are over 400 billion parameters and, while these models are still training...Over the coming months, we'll release multiple models with new capabilities including multimodality, the ability to converse in multiple languages, a much longer context window, and stronger overall capabilities," Meta noted.
