研究人员推出了 FastKernels，这是一个新的基准测试，旨在更好地评估生产 LLM 推理中使用的 GPU 核生成代理。现有的基准测试与实际系统不匹配，导致代理生成的核在测试环境之外表现不佳。FastKernels 旨在通过作为一个生产级推理框架来弥合这一差距，该框架反映了实际部署需求，并涵盖了绝大多数 HuggingFace Transformers 架构。 AI
影响 通过改进 GPU 核生成基准测试与生产系统的对齐，解决了 LLM 推理中的关键瓶颈。
排序理由 该集群包含一篇学术论文，介绍了一个用于评估与 AI 相关的基础设施的新基准测试和框架。
AI 生成摘要 · Google Gemini · 来自 3 个来源。 我们如何撰写摘要 →
arXiv:2605.23215v1 Announce Type: cross Abstract: LLM-based agents for GPU kernel generation are advancing rapidly, yet their progress is fundamentally constrained by the benchmarks they optimize against. Existing benchmarks are poorly aligned with production inference frameworks…
LLM-based agents for GPU kernel generation are advancing rapidly, yet their progress is fundamentally constrained by the benchmarks they optimize against. Existing benchmarks are poorly aligned with production inference frameworks: they evaluate kernels on a single GPU with synth…
LLM-based agents for GPU kernel generation are advancing rapidly, yet their progress is fundamentally constrained by the benchmarks they optimize against. Existing benchmarks are poorly aligned with production inference frameworks: they evaluate kernels on a single GPU with synth…