🛰️ Daily AI Frontier
‹ back to 2026-08-10

Zero Gap Is Not Restoration: Stratified Per-Question Probability Evaluation and Step-wise Mitigation of Benchmark Contamination

Research LLMs & Foundation Models

Ranking

Overall 66
Content 65
Popularity 68

Observed public metrics from 1 member.

Merged summary

TL;DR - An arXiv paper arguing that the standard G-AP metric for judging benchmark-contamination mitigation is misleading, and proposing both a better metric (SA-PPG) and a decoding-time mitigation method (RailCap). It matters because it suggests prior claims of "restoring" a contaminated model's true capability are substantially overstated.

  • G-AP critique: discrete correct/incorrect readouts hide per-question behavior, averaging before differencing lets over- and under-suppression cancel, and uniform per-question weighting can be gamed by shifting solve probabilities onto the clean model's high-frequency values.
  • SA-PPG: estimates each question's solve probability by sampling, differences it against the clean model per question, then aggregates within strata defined by the clean model's solve probability.
  • RailCap: judges contamination during generation rather than pre-estimating where it lies — when a sample falls back onto the greedy trajectory, the next trajectory token is capped to the runner-up, accumulating suppression until the response distribution is sufficiently dispersed.
  • Findings: across multiple contaminated models and benchmarks, SA-PPG shows prior strategies' restoration is overestimated, and RailCap achieves the lowest SA-PPG (no numeric results given in the provided abstract).

Sources (1)

Zero Gap Is Not Restoration: Stratified Per-Question Probability Evaluation and Step-wise Mitigation of Benchmark Contamination

arXiv cs.CL Ruijie Hou, Yueyang Jiao, Zhao Wang, Yingming Li 2026-08-07 arXiv:2608.07341
Public signals Hugging Face upvotes 4
Providers: Hugging Face · Upvotes 4 OpenAlex · N/A Publisher · N/A Semantic Scholar · N/A X · N/A Fetched 2026-09-09 08:18:11.518199 UTC

TL;DR - An arXiv paper arguing that the standard G-AP metric for judging benchmark-contamination mitigation is misleading, and proposing both a better metric (SA-PPG) and a decoding-time mitigation method (RailCap). It matters because it suggests prior claims of "restoring" a contaminated model's true capability are substantially overstated.

  • G-AP critique: discrete correct/incorrect readouts hide per-question behavior, averaging before differencing lets over- and under-suppression cancel, and uniform per-question weighting can be gamed by shifting solve probabilities onto the clean model's high-frequency values.
  • SA-PPG: estimates each question's solve probability by sampling, differences it against the clean model per question, then aggregates within strata defined by the clean model's solve probability.
  • RailCap: judges contamination during generation rather than pre-estimating where it lies — when a sample falls back onto the greedy trajectory, the next trajectory token is capped to the runner-up, accumulating suppression until the response distribution is sufficiently dispersed.
  • Findings: across multiple contaminated models and benchmarks, SA-PPG shows prior strategies' restoration is overestimated, and RailCap achieves the lowest SA-PPG (no numeric results given in the provided abstract).
item →