The Smart Atlas jointly with Huawei opensource the first domestic chip training multimodal SOTA model.

Today, Zhilv and Huawei jointly open-sourced the new generation image generation model GLM-Image. The model is based on the Ascend Atlas 800T A2 device and the MindSpore AI framework to complete the entire process from data to training, making it the first SOTA multimodal model to complete training on a domestic chip. According to the introduction, GLM-Image achieves the joint generation of images and language models. In API call mode, generating an image only costs 0.1 yuan, and a speed-optimized version is coming soon.