Infer-forge:围绕 SGLang 的 Harness、Loop 与 Graph 工程
Blog Infer-forge: Harness, Loop, and Graph Engineering Around SGLang Inference optimization may look local in code, but its validity is global. A kernel, communication path, or scheduling change becomes meaningful only at a specific deployment point defined by the mode... Tianyu Zhang, Hanlin Gao, Yusong Gao, Yun Zhang
AI 导读Infer-forge 是一套围绕 SGLang 推理优化的内部工程系统,通过 MonoRepo、Harness、Task Loop 与 Task Graph 四种结构,将部署点约束链(模型、SLO、拓扑、运行时、加速平台)转化为可复现、可审计的工程流程。
来源:LMSYS Blog · lmsys.org