Meta with support from partners like NVIDIA, Google Cloud, Azure, and AWS, has introduced its latest and largest open-source language model - Llama 3.1 AI Model. The new model comes in three sizes, 8B, 70B, and 405B parameters. Among the three, 405B model is one of the most powerful open-source models available. It shines in general knowledge, steerability, math, tool use, and multilingual translation.
Llama 3.1 has scored 73.8 on the MATH benchmark which is close to GPT-4o’s 76.6 and surpassing Claude 3.5 Sonnet’s 71.1. It supports multiple languages like Spanish, Portuguese, Italian, German, Thai, French, and Hindi, making it useful for various users and applications worldwide.
The notable improvement is its enlarged context window, supporting up to 128,000 tokens compared to Llama 3’s 8,192. This growth enables the model to handle longer texts, improving its ability to perform complex reasoning tasks. In addition, Llama 3.1 405B is a powerful tool for synthetic data generation, setting a new standard for generative AI. It can build high-quality task-specific synthetic data for training other models, enhancing accuracy in fields like finance, retail, and telecommunications.
The remarkable feature of Llama 3.1 405B is its potential for model distillation, which means transferring its capabilities to smaller, more efficient models. This method can produce compact models with similar performance at lower costs and reduced latency.
The Llama 3.1 Instruct models are specifically fine-tuned for effective tool usage, enhancing their capability to interact with supportive programs for tasks such as searches, image generation, code execution, and mathematical reasoning.
Meta has also presented robust security measures, including Llama Guard 3 for input and output moderation and Prompt Guard to detect and respond to prompt injection and jailbreak inputs.
See What’s Next in Tech With the Fast Forward Newsletter
Tweets From @varindiamag
Nothing to see here - yet
When they Tweet, their Tweets will show up here.
