27 billion-parameter model
A class of artificial intelligence models characterized by approximately 27 billion trainable parameters. These models balance computational efficiency with high-capacity reasoning, often deployed for specialized tasks requiring structured output rather than open-ended generation.
Key Implementations
Clef 27B
Clef 27B: Multimodal AI Decision Model for Structured Input Analysis
Developed by Cloudflare, Clef 27B is a multimodal decision model designed for rapid, structured decision-making. It diverges from traditional Large Language Models (LLMs) that generate text by instead returning calibrated probabilities for specific queries.
- Architecture: 27 billion parameters.
- Input Modalities: Text, images, video, and JSON data.
- Output Format: Structured probabilities rather than natural language text.
- Primary Use Case: Decision-making tasks requiring high-speed inference and precise confidence calibration.
- Analysis Source: Clef 27B: Multimodal AI Decision Model for Structured Input Analysis by Fahd Mirza.