SGLang @sgl_project
1 of this account's posts made the board, 0 reached the top 3.
10.7K followersOpen Source 1 Follow on X
Stories from this source
-
SGLang Optimized for Nvidia Rubin
SGLang, with NVIDIA, optimized attention, MoE and speculative verification kernels on early Rubin hardware, speeding Kimi K3 inference up to 20% and end-to-end by 5.9%.