The latest generation of Meta’s open-source large language model, Llama 3, has been introduced to the public. Two text-based models trained with 8B and 70B parameters have been released, providing state-of-the-art performance and improvements in reasoning, code generation, and instruction capabilities compared to their predecessors.
Initially planned for May release, Llama 3 was delayed due to unforeseen circumstances. However, the company has now made these models available to the public. Llama 3 has been trained on a data set seven times larger than its predecessor, using over 15T tokens collected from publicly available sources. The models have undergone rigorous filtering processes to ensure data quality and reliability. Additionally, more than 5% of the pre-training data set includes information in languages other than English, preparing for upcoming multilingual use cases.
To promote safe interactions with Llama 3, Meta has introduced resources like Llama Guard 2, Code Shield, and CyberSecEval 2. These resources are designed to help users navigate the complexities of working with large language models responsibly. The latest Llama 3 models have been integrated into the Meta AI assistant across platforms like Facebook, Instagram, WhatsApp, and Messenger. Users can download Llama 3 models and access them on various platforms including Amazon Web Services (AWS), Google Cloud Platform (GCP), IBM WatsonX, Microsoft Azure Machine Learning (Azure ML), among others.
Meta plans to introduce new capabilities for Llama 3 in the coming months such as longer context windows and additional model sizes. Research work on the model will also be shared by the company in order to encourage collaboration within the open source community.
Overall, Meta’s latest release of Llama 3 represents a significant leap forward in AI technology and its potential applications across various industries and tools.
Meta is committed to driving innovation in AI across various applications such as natural language processing (NLP), computer vision (CV), robotics (ROBOTICS) etc., while ensuring responsible use through resources like Llama Guard