KernelBench-M

The measurement artifact for Measuring the Checker: Mutation Analysis for GPU-Kernel Benchmark Oracles: the mutation operators, the verified CUDA substrates they mutate, the kill witnesses, and the pipeline that produced every number in the paper.

Layout

rules/          124 mutation rules, six families (mutator.py loads all of them)
substrates/     208 gate-verified CUDA implementations, one per KernelBench
                problem: the mutation targets. Never used as oracle.
witnesses/      per-problem kill records: for every mutant, its rule, family,
                mutated site, whether the official protocol kills it, and — when
                it survives — a suite that does
pipeline/       the measurement code (screening, kill matrix, audit, set cover)
summary.json    per-problem mutant / witnessed / official-kill counts
validation/     errata, validation evidence, and quarantined historical records

The substrate count above describes the original release. InstanceNorm's repaired source has passed H100 re-admission; its active witness and summary records now describe the repaired source and expanded regression suites.

Dataset Viewer and library access

The viewer's default configuration contains one row per mutation record in the active witnesses/*.json files, with problem and mutant_id identifying its source. The other columns preserve the original witness fields, including null values. The validation split is a browsing table of all active records; the dev/test assignment for the paper remains in holdout.json.

from datasets import load_dataset

records = load_dataset("Elfsong/KernelBench-M", split="validation")
# Use streaming=True to iterate without downloading the entire table first.

data/witnesses.jsonl is generated by pipeline/export_hf_viewer.py. After updating active witness records, run python pipeline/export_hf_viewer.py from the repository root; --check verifies that the export is up to date. The explicit data_files configuration keeps the separate, heterogeneous reports in validation/ out of the viewer's table.

Supporting records: mechanism_curves.json (Fig. 2b–c), holdout.json (dev/test split), ladder_summary.json (knowledge ladder), gate_report.json (substrate admission), invalid_suites.json (suites rejected for crossing the validity ceiling).

What a substrate is, and is not

KernelBench ships PyTorch references whose GPU execution bottoms out in closed cuDNN/cuBLAS binaries, so there is no source to mutate. A substrate is a correct CUDA implementation of the same computation, used only as a mutation target. The oracle stays the benchmark's own PyTorch reference. Each substrate is admitted by pipeline/gate_candidates.py, which compares it against that reference on every suite before it may enter the pool (gate_report.json records admissions and rejections).

Witness format

"softmax:barrier-drop:3": {
  "rule": "barrier-drop", "family": "sync",
  "site": "__syncthreads();",
  "cls": "high_value",
  "killed_by_official": false,
  "witness": "T2_misaligned_batch"
}

cls is killed_by_T0 (the official inputs detect it), high_value (survives the official inputs, a targeted input detects it), or no_witness (nothing we have detects it — an equivalence or oracle-blindness candidate, excluded from every denominator in the paper).

Reproducing

Requires an NVIDIA GPU with CUDA 12.x and PyTorch. From pipeline/:

python screen_mutants.py     # generate, filter, and screen mutants
python full_matrix.py        # kill matrix over all suites
python audit_matrix.py       # score competing protocols (§5)
python analyze_cover.py      # set-cover suite synthesis (§6)
python analyze_holdout.py    # dev/test holdout (§6)

Each stage writes append-only JSONL journals and resumes from them, so a crashed or preempted run can simply be restarted. SHARD_ID/SHARD_N split problems across GPUs; STEAL=1 lets an idle worker pick up unclaimed problems.

Caveats

Mutation score is adequacy relative to this fault model, not absolute correctness: these operators cannot express faults living in structures the substrates do not contain (tensor-core paths, double-buffered pipelines), nor multi-site interactions. The artifact supports comparative claims between protocols and existence claims about specific faults. It cannot certify that a kernel passing these suites is correct.

Acknowledgments

Thanks to Jayden Vasquez for reporting the InstanceNorm numerical issue and providing a reproducer and initial fix.