블로그 글 추가: 2026-06-18-from-minutes-to-seconds-llm-guided-autotuning-for-helion-kernels, 분 단위에서 초 단위로: Helion 커널을 위한 LLM 기반 자동 튜닝#95
Open
9bow wants to merge 2 commits into
Open
Conversation
PR Preview빌드가 완료되었습니다! 아래 링크에서 변경사항을 확인할 수 있습니다. 미리보기: https://pytorchkr-pr-preview-95.surge.sh
|
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
번역 글 소개
분 단위에서 초 단위로: Helion 커널을 위한 LLM 기반 자동 튜닝 번역 글을 추가합니다.
PyTorch의 커널 DSL인 Helion의 자동 튜닝을 LLM으로 가속하는 방법을 다룹니다. 기존 기본값인 LFBO(가능도 없는 베이지안 최적화) 탐색이 커널마다 수백 번의 컴파일·벤치마크를 거치는 데 비해, LLM 기반 자동 튜너는 LFBO 수준의 성능(기하평균 1.009배)을 유지하면서 약 10배 적은 구성을 벤치마크하고 약 6.7배 짧은 실제 소요 시간에 수렴합니다.
LLM이 뒤처지는 소수 커널은 LLM 초기값 생성 후 LFBO로 정교화하는 하이브리드 전략으로 격차를 메우며, Opus-4.8·gpt-5.5·Sonnet-4.6 간 결과 차이가 거의 없어 모델에 독립적인 실용적 접근임을 B200 33개 커널 벤치마크로 보여줍니다.