Tencent

WeDLM

WeDLM: The fastest diffusion language model with standard causal attention and native KV cache compatibility, delivering real speedups over vLLM-optimized baselines.

exlawShaoxunZengaakash-chaddha