On June 27, Hunyuan-a13B, the first hybrid inference MoE model open source by Tencent, debuted in the Magic Match community. The model has a total parameter of 80B and an activation parameter of only 13B. It is the industry's first 13B-level MoE open source hybrid inference model, and its effectiveness is comparable to leading open source models with similar architectures. Developers can obtain leading-edge model capabilities at a lower threshold.

Zhitongcaijing · 06/27 09:57
On June 27, Hunyuan-a13B, the first hybrid inference MoE model open source by Tencent, debuted in the Magic Match community. The model has a total parameter of 80B and an activation parameter of only 13B. It is the industry's first 13B-level MoE open source hybrid inference model, and its effectiveness is comparable to leading open source models with similar architectures. Developers can obtain leading-edge model capabilities at a lower threshold.