Phi-3-mini: a small language model for edge devices

A compact model from Microsoft for settings where memory, compute or latency are tight. It reasons and does maths well for its size.

• Trained on filtered web and code data mixed with synthetic, textbook-style material
• 4K-token context; a separate 128K-context variant is available
• Works best in English

Summary written by SolxLab. Weights and code belong to their authors and are distributed under the license shown; always check the official page before use.