← Tuesday, October 13, 2026AgentsPaper

MetaKernelBench measures whether GPU-kernel skills transfer across DSLs

huggingface.co by Xueyi Chen et al. Editor score 8/10

Across six models, transferred skills changed paired performance from -19% to +29%, with regressions on 16% to 45% of attempts. The benchmark tests whether agent memory helps kernel optimization across languages instead of merely scoring isolated attempts.

What

MetaKernelBench evaluates whether GPU-kernel optimization experience transfers between CuTe DSL and TIRx.

Method

Agents distill a source attempt into a natural-language skill, then use it on the same problem in the other DSL; paired lift ranged from -19% to +29%.

Why it matters

It measures whether retained skills help beyond correctness and speed on one attempt, while exposing transfer regressions across models and problems.

Read on huggingface.co → Code · 4 stars

Headline, summary, brief and diagram are written by a model from the source's own text; the link, byline and numbers come from the source. Read the original before relying on a claim.