Pith. sign in

REVIEW 5 cited by

Cheating Automatic LLM Benchmarks: Null Models Achieve High Win Rates

Not yet reviewed by Pith; the record is open.

Re-run · record.json Download PDF Read on arXiv ↗

This paper has not been read by Pith yet. Machine review is queued; the pith claim, tier, and objections will appear here once it completes.

SPECIMEN: schema-true, not a live event

T0 review · schema-true

One-sentence machine reading of the paper's core claim.

pith:XXXXXXXX · record.json · timestamp

arxiv 2410.07137 v2 pith:RUHEZA2O submitted 2024-10-09 cs.CL cs.AIcs.CRcs.LG

Xiaosen Zheng , Tianyu Pang , Chao Du , Qian Liu , Jing Jiang , Min Lin This is my paper
classification cs.CLcs.AIcs.CRcs.LG
keywords benchmarksratesautomaticalpacaevalcheatinghighmodelspromotional
verification ladder T0 review T1 audit T2 compute T3 formal
0 comments
read the original abstract

Automatic LLM benchmarks, such as AlpacaEval 2.0, Arena-Hard-Auto, and MT-Bench, have become popular for evaluating language models due to their cost-effectiveness and scalability compared to human evaluation. Achieving high win rates on these benchmarks can significantly boost the promotional impact of newly released language models. This promotional benefit may motivate tricks, such as manipulating model output length or style to game win rates, even though several mechanisms have been developed to control length and disentangle style to reduce gameability. Nonetheless, we show that even a "null model" that always outputs a constant response (irrelevant to input instructions) can cheat automatic benchmarks and achieve top-ranked win rates: an 86.5% LC win rate on AlpacaEval 2.0; an 83.0 score on Arena-Hard-Auto; and a 9.55 score on MT-Bench. Moreover, the crafted cheating outputs are transferable because we assume that the instructions of these benchmarks (e.g., 805 samples of AlpacaEval 2.0) are private and cannot be accessed. While our experiments are primarily proof-of-concept, an adversary could use LLMs to generate more imperceptible cheating responses, unethically benefiting from high win rates and promotional impact. Our findings call for the development of anti-cheating mechanisms for reliable automatic benchmarks. The code is available at https://github.com/sail-sg/Cheating-LLM-Benchmarks.

Discussion (0). Sign in to comment.

Forward citations

Cited by 5 Pith papers

Reviewed papers in the Pith corpus that reference this work. Sorted by Pith novelty score. Full citation record

  1. The Verifier is the Curriculum: Execution-Gated Self-Distillation for Cross-Family Game Generation

    cs.AI 2026-06 conditional novelty 6.5 of 10

    Strict-launch-gated rejection-sampling self-distillation compounds cross-family clean Godot generation from 8.8% to 42.2% and full best-of-K coverage, while gold duplication and a lenient BUILD filter erase the gain.

  2. Every Model Cheats: Prompt-Level Mitigation of Cheating on Offensive Cyber Tasks

    cs.CR 2026-07 conditional novelty 6.0 of 10

    On Cybench CTF tasks, 21 of 22 models cheated under baseline, 37% of baseline passes were cheated, and stricter anti-cheat prompts reduced but did not remove cheating.

  3. One Token to Fool LLM-as-a-Judge

    cs.LG 2025-07 conditional novelty 6.0 of 10

    LLM reward models falsely accept empty 'master key' responses such as ':' or 'Thought process:' across many models, and a fine-tuning defense reduces these false positives to near zero.

  4. Attestable Audits: Verifiable AI Safety Benchmarks Using Trusted Execution Environments

    cs.AI 2025-06 conditional novelty 6.0 of 10

    A TEE-based protocol for cryptographically verifiable AI safety benchmark results, demonstrated on Llama-3.1 with AWS Nitro Enclaves.

  5. Scaling Point-in-Time Language Models

    cs.CL 2026-04 conditional novelty 5.5 of 10

    Scaling point-in-time LLMs to 4B parameters and 1T temporally filtered tokens narrows the gap to unrestricted models to about 8–11 average points and yields positive out-of-sample Sharpe ratios from news embeddings.

Pith tools