The ant releases the first 100B diffusion language model LLADA2.0

On December 12th, Ant Technology Research Institute officially launched the LLaDA2.0 series of discrete diffusion large language models, and concurrently released the technical report. The previously open-sourced LLaDA2.0 includes two versions with MoE architecture, 16B and 100B. Ant has expanded the parameter scale of the Diffusion model to the 100B level for the first time.