Revision History
KernelBench · 4 revisions
Sizes are character counts of the article source. The signed number is the change from the previous revision.
Recent edit summaries
Detailed summaries recorded by editors. Generic maintenance summaries are omitted here; every recorded revision remains below. Dates describe the edit, not necessarily the event it covers.
- Minor (verifier): level ordering is by primitive-operation count, not measured difficulty (Level 2 was easiest for reasoning models in the paper's one-shot results); NVIDIA 10-minute budget scoped to Level 1; baseline-timing guidance cited to the GitHub README (new ref 6); gpu_kernel red link removed
Version 4 · Sep 16, 2026, 03:04 PM
- Correction: KernelBench was first presented in a Dec 3, 2024 Stanford blog (not introduced by the Feb 2025 paper); NVIDIA's loop budget was more than 10 minutes, not 15; one-shot correctness restated from Table 1; Ré spelling. Also add Dream-RSI use of four KernelBench tasks
Version 3 · Sep 16, 2026, 02:52 PM
- Added 6 contextual internal links
Version 2 · Jul 23, 2026, 04:03 PM