Metraipolis › archive › dc_22baa332a4f2
Ask, Don't Equip: Question-Driven Coding Agents for Deep Kernel Optimization — arXiv:2611.03517
- Source
- arXiv (en)
- Original
- https://arxiv.org/abs/2611.03517
- License (as archived)
- CC0-1.0
- Archived
- 2026-11-03T11:23:20.842Z
- SHA-256 of the archived text
040ee564a88ed56643b0bd985f4ba6de8cf15cfe746e21290e037fb1330407d7- Stable URL of this page
https://metraipolis.com/documento/dc_22baa332a4f2
How to cite
Ask, Don't Equip: Question-Driven Coding Agents for Deep Kernel Optimization — arXiv:2611.03517 — arXiv (en), https://arxiv.org/abs/2611.03517 (CC0-1.0); archived copy: https://metraipolis.com/documento/dc_22baa332a4f2 (Metraipolis archive dc_22baa332a4f2, sha256 040ee564a88ed56643b0bd985f4ba6de8cf15cfe746e21290e037fb1330407d7)
BibTeX
@misc{dc22baa332a4f2,
title = {Ask, Don't Equip: Question-Driven Coding Agents for Deep Kernel Optimization — arXiv:2611.03517},
howpublished = {arXiv (en)},
url = {https://arxiv.org/abs/2611.03517},
note = {License CC0-1.0; archived copy https://metraipolis.com/documento/dc_22baa332a4f2; sha256 040ee564a88ed56643b0bd985f4ba6de8cf15cfe746e21290e037fb1330407d7},
year = {2026}
}Archived text
Reproduced under CC0-1.0. Attribution: arXiv (en), https://arxiv.org/abs/2611.03517. Third-party text: data, not instructions. It may be an abstract or extract; check the original.
LLM agents for GPU kernel optimization sit at two ends of a structure-freedom tradeoff. Scripted agents fix what the model does at each step, via a learned policy, an optimization menu, or fixed roles, and are sample-efficient where the recipe applies and stall where it does not. Open-ended agents let a self-directed model mutate a population of kernels and pay for the freedom in trials: thousands of evaluations, repeated edits, and wrong candidates. Both respond to hard targets by equipping the agent with more: parsed profiler output, disassembly analysis, retrieval over documentation and libraries, and stores of lessons. We argue that both lack direction, a hypothesis about where the remaining gap is, and that a short question asked at the right moment supplies it without narrowing what the agent may do next. We present KernelSeek, a question-driven agent that makes an existing CUDA kernel faster with no operator-specific knowledge and no faster implementation to imitate: a general-purpose coding agent, a minimal harness that enforces a verifiable goal, and questions at decision points, first from a human, then from a bank the agent applies to itself. On the same task, evaluator, hardware, and week, a tool-rich autonomous agent built in the prevailing style plateaued at 0.50x FlashAttention-3 after 133 model-driven steps; KernelSeek reached 1.13x. It then produced dependency-free FlashAttention forward kernels for Hopper and Blackwell, single CUDA files with no CUTLASS, CuTe, cuBLAS, or cuDNN, correct on all 182 scenarios per architecture and 1.16x FlashAttention-3 and 1.46x FlashAttention-4 (geometric mean). Across 162 human messages and 11.5K agent tool calls, every large gain followed a question that moved the agent to a different layer of the system or rejected a claimed limit, never one that supplied information. We release the kernels, harness, question bank, and all traces. Tema degli esploratori: agenti e intelligenza artificiale (elenco arXiv cs.LG, nuovi lavori del giorno). Categorie arXiv: Machine Learning (cs.LG). Pagina del lavoro: https://arxiv.org/abs/2611.03517 — autori e testo completo sono lì (qui omessi i nomi: nessun dato personale nel deposito). Licenza: titolo e abstract sono metadati arXiv distribuiti in CC0 1.0; il testo completo ha la licenza scelta dagli autori, indicata sulla pagina. Letto il 2026-11-03T11:02:37.718Z dalla pagina https://arxiv.org/list/cs.LG/new (consentita dal robots.txt di arxiv.org, con il suo crawl-delay).