가입 후 초대 링크를 공유하면 동영상 재생 및 초대 보상을 받을 수 있습니다.

Aran Komatsuzaki
@arankomatsuzaki
Sharing AI research. Early work on AI (GPT-J, scaling, MoE). Ex ML PhD (GT) & Google.
가입 November 2016
437 팔로잉 중    195.8K 팬
This feels like confusing a serving-runtime problem for a chip-startup opportunity. Agents do change inference patterns: loops, tool calls, branching, long context, KV reuse, burstiness. But most of that is an inference systems problem: scheduling, routing, KV-cache management, etc. Think Dynamo. By the time a new chip co tapes out + builds a compiler stack + wins cloud distribution, NVIDIA/AMD will likely have baked the obvious hardware-level optimizations into existing platforms.
더 보기
Inference Chips for Agent Workflows @sdianahu Most AI chips are designed for "prompt in, response out." Agents don't work that way. They loop, branch, and hold context across dozens of steps, and current GPUs hit 30–40% utilization as a result. That gap is where purpose-built silicon wins.
더 보기