arXiv Machine Learning
Jun 9
Decoy-Calibrated Failure Audits for Language Models
arXiv:2606. 09046v1 Announce Type: new Abstract: Useful audits reveal not only how often a model fails, but also where its failures concentrate.
Instrument Effects in Language-Model Honesty Evaluation: An Auditable Single-System Demonstration
arXiv:2607. 14399v1 Announce Type: new Abstract: Evaluations of language-model honesty read the model's verdicts as evidence about the model.
Related stories
arXiv AI
Aug 17
Inducing Reward-Free Judging Rubrics that Reduce Over-Crediting in Agent Evaluation
arXiv:2608. 13564v1 Announce Type: new Abstract: Evaluating language-model agents at scale increasingly relies on a second language model as an automatic judge, because the gold signal, an executable environment reward, is expensive, slow, or unavailable at deployment time.
arXiv AI
Sep 2
Commit-first LLM judging inherits the judge's own errors
The paper investigates whether widely used evaluation frameworks for large language models (LLMs) implement a defense called commit‑first judging, which requires a judge to solve a task itself before accepting a candidate answer. Across 24 configurations in eight popular frameworks, none use the full commit‑first method; nine use a weaker variant that is ineffective. In controlled experiments, the weaker variant allowed systems to game the judge, while the full commit‑first approach eliminated this vulnerability but sometimes worsened evaluation when the judge’s own answer was incorrect.
Hugging Face Trending Papers
Aug 31
Commit-first LLM judging inherits the judge's own errors
LLM judges, models that score another system's output, can be gamed by the systems they score. Recent work identifies one defence that works: the judge solves the task itself first and commits to that...
arXiv AI
2d ago
Kepler: Auditable World Models for ARC-AGI-3
Kepler is an open‑source harness that represents hypotheses as executable world models and validates them through retrospective transition checks and conditional prediction checks. In the ARC‑AGI‑3 benchmark, a frozen Claude Opus 5 configuration achieved a perfect 100.00 RHAE on all 25 public games without per‑game model selection or score‑conditioned reruns, and matched or outperformed median‑human action counts on 181 of 183 levels. The study also identified three evaluation failures and highlighted that public‑set score alone has limited discriminative value, advocating for first‑attempt, cost‑conditioned, and verification‑aware reporting. whyItMatters":"The results demonstrate that a purely score‑based evaluation can be misleading, underscoring the need for more rigorous, cost‑aware, and verification‑aware metrics in AI benchmark assessments."
arXiv Machine Learning
Jul 8
More Convincing, Not More Correct: Self-Play Reward Hacking of Reference-Free LLM Judges
arXiv:2607. 05904v1 Announce Type: new Abstract: Training a language model against its own reference-free judgments (the premise of self-rewarding, self-play, and LLM-as-a-judge pipelines) assumes a model's verdict on a shown answer tracks correctness.
arXiv AI
Aug 24
Structure for Reading, Prose for Writing: Asymmetric Structural Conditioning in Multi-Agent Document Authoring
The paper reports on a deployed multi‑agent tender‑response system that uses an open‑weights language model under sovereignty constraints. In a blind comparison, the system’s answers were judged at least as good as human‑written bids in 40 of 55 sections, with only a few gaps attributable to missing knowledge rather than writing quality. The study also demonstrates an asymmetry in conditioning: while structural markup improves reading tasks, converting instruction material from prose to nested XML degrades answer quality, and naming forbidden constructions concentrates defects.
arXiv AI
Aug 24
Calibrating Criterion Revision in LLM Agents: Failure Modes and a Trace-Anchored Protocol
The paper introduces a framework for evaluating how large language model agents revise their success criteria after failures, defining five non‑compensatory conditions that must be met for a criterion revision to be considered valid. Using the CMB‑0.1 protocol, the authors test twelve cross‑domain scenarios across four system configurations, finding that no model trial satisfies all five conditions and highlighting specific failure modes such as zero‑state reconstruction and inadequate intervention sensitivity. They propose a more stringent trace‑anchored CMB‑0.4 protocol to better isolate and measure criterion revision in future studies.
arXiv Machine Learning
Sep 17
A Calibrated Instrument for Measuring How Inference Optimizations Affect Output Quality
The paper introduces a calibrated instrument for rigorously measuring how inference optimizations—such as quantization, early‑exit, and speculative decoding—affect the output quality of large language models. It uses a formally calibrated LLM judge that verifies no systematic bias between statistically equivalent outputs and includes a null condition to ensure measured differences are zero. Applying this method, the authors find that a 4‑bit model is indistinguishable from its 16‑bit counterpart, while 3‑bit quantization and early‑exit techniques incur measurable quality losses that vary by language and task, and that token‑certainty‑based acceptance rules cannot reliably identify impactful errors.
arXiv AI
Sep 7
When Do Internal Probes Beat Reading the Answer? Miscalibrated Readouts and Behavior-Concealed Knowledge in Language Models
A 0.6B language model consistently answers YES to 1,200 logical tests, yet its behavior shows no discrimination. Linear probes reveal the correct verdict with high AUC (0.96) and transfer to unseen structures, but a single scalar readout fails due to a saturated decision threshold offset by +4.6 σ. Adjusting this threshold restores behavior accuracy from 50 % to 81 % and improves higher‑scale models, demonstrating that miscalibrated readouts, not hidden knowledge loss, drive performance gaps.
arXiv AI
Sep 15
Calibrating Interpretability Instruments Before Trusting Their Verdicts
arXiv:2609.14754v1 Announce Type: cross Abstract: Causal claims about large language model (LLM) internals rest on measurements. Those might include a projection, a cosine, an ablation delta, or an i...
arXiv Computation and Language
Sep 21
JudgeSense: A Benchmark for Prompt Sensitivity in LLM-as-a-Judge Systems
JudgeSense is a benchmark comprising 880 items from human‑labelled corpora, each presented under two differently worded instructions that ask the same question. The study evaluates 25 judges from six providers across four tasks, measuring how rewording affects agreement with the judge’s own verdicts. Results show that rewording reduces agreement on all tasks, with significant effects on two, and that stability varies across tasks and is not predicted by parameter count.