Meta The grand finale of this year's AI models is here.Meta was released yesterday (December 6) Llama 3.3.There are 70 billion parameters, but performance is comparable to Llama 3.1 with 405 billion parameters.
Meta emphasizes that Llama 3.3 models are more efficient, less expensive, and can run on standard workstations, reducing operational costs while delivering high-quality text AI solutions.
The Llama 3.3 model focuses on optimizing multi-language support, with support for 8 languages: English, German, French, Italian, Portuguese, Hindi, Spanish and Thai.
In terms of architecture, Llama 3.3 is an autoregressive (autoregressive) language model that uses an optimized transformer architecture, and its fine-tuned version uses Supervised Fine-Tuning (SFT) and Reinforcement Learning Based on Human Feedback (RLHF) to align it with human preferences for usefulness and safety.
Llama 3.3 contexts are 128K in length, support multiple tool usage formats, and can be integrated with external tools and services to extend the functionality of the model.
In terms of security, Meta employs data filtering, model fine-tuning, and system-level security to reduce the risk of model misuse; in addition, Meta encourages developers to deploy Llama 3.3 with the necessary security measures, such as Llama Guard 3, Prompt Guard, and Code Shield, to ensure responsible use of the model.
1AI Attach reference address