Meta releases 5 AI models
The models include image-to-text and text-to-music generation models, a multi-token prediction model, and a technique for detecting AI-generated speech.
The models include image-to-text and text-to-music generation models, a multi-token prediction model, and a technique for detecting AI-generated speech.
According to the company, Meta AI is powered by a custom model that leverages technology from Meta's Llama 2 and the latest large language model (LLM) research.