Skip to content

Yuan2.0-M32

A mixture of experts model with attention routing

Visit official site
Category AI models

What yuan2-0-m32 is

The model is presented as a mixture of experts language model that uses a new routing network for more efficient expert selection.

What you can do with it

  • Use it as an aid for domain knowledge and reasoning
  • Run data preprocessing with the supplied scripts

Pros & cons

✓ What we like

  • Reported scores include MATH 55.9 percent and ARC-Challenge 95.8 percent
  • Coding, math and specialized domains are listed as competitive

! What to watch out for

  • A recent docker image is recommended for the environment
  • Inference service deployment follows the vllm plan

FAQ

How many experts are active?

Two out of 32 experts are active.

How large is the training set?

The card states 2000B tokens from scratch.

Last reviewed: 2026-11-15

More LLM API platform tools

View all →

How we review