The paper examines how calibration rules for open‑ended question answering can be misled by the choice of correctness labels, noting that exact‑match labels can produce misleading confidence estimates. It introduces FactForm, a fact‑grounded form grader that combines reference matching with evidence retrieval, achieving lower error and higher AUROC than traditional graders on a large benchmark. FactForm’s design allows a single LLM call per answer and demonstrates stable performance across runs.
Reliable evaluation of open-ended question answering remains a bottleneck for measuring answer correctness of modern LLMs. Unlike multiple-choice tasks, free-form answers may be correct in many surfac...
The paper introduces a semantic correctness taxonomy that categorizes open‑ended QA answers into eight ordered classes, distinguishing between correct, verbose, and hallucinated responses. It releases two datasets—CAP‑Correctness and CAP‑Statements—to support benchmark evaluation and NLI‑based training. The authors also propose CAP (Context‑Aware Precision), a reference‑based metric that scores question‑conditioned statements via bidirectional NLI and demonstrates superior performance under a monotonicity protocol.
arXiv:2610.08026v1 Announce Type: new Abstract: In recent years, several methods for detecting when large language models (LLMs) hallucinate have been developed. These methods are often benchmarked w...
The paper demonstrates that open‑ended Theory‑of‑Mind trackers can produce valid beliefs that are absent from finite reference sets, and that treating unmatched outputs as false can reverse model‑selection rankings. By recoding references for 259 beliefs, the authors show a dramatic drop in weighted prevalence and a reversal of strictly proper Brier risk, with similar distortions observed in a 301‑question NQ‑open DPR‑BERT pipeline. The study further reveals that 90‑96% of audited unmatched beliefs are literally true, and introduces a TriSource‑Restore method that anchors reference labels to a probability‑sampled human pilot to restore calibration and ranking integrity.
arXiv:2609.13824v1 Announce Type: new Abstract: Large language models (LLMs) are increasingly used to evaluate the responses of other language models. This approach, known as LLM-as-a-Judge, is faste...
arXiv:2606. 11477v1 Announce Type: cross Abstract: Correcting handwritten exams by hand is time-consuming and error-prone, particularly for large cohorts, while fully digital exams tend to force a didactic narrowing towards closed question formats.
arXiv:2605. 27752v3 Announce Type: replace Abstract: Is verbalized confidence better calibrated than token likelihood?
The paper investigates whether accuracy and order sensitivity in large language model (LLM) multiple‑choice question (MCQ) evaluation move together. Using proprietary and open‑weight models on MMLU and ARC‑Challenge, the authors measure order sensitivity with flip rate and recall standard deviation (RStd) and find that these metrics often diverge from accuracy and from each other. They conclude that accuracy alone is insufficient and recommend reporting both flip rate and RStd alongside accuracy.
arXiv:2609.37493v1 Announce Type: cross Abstract: Serving an answer from a large language model requires deciding when to abstain, yet a verifier's ranking accuracy alone does not determine the error...
arXiv:2609.15561v1 Announce Type: new Abstract: Evaluating the factual correctness of large language models (LLMs) is vital for many applications. But are our evaluation tools themselves trustworthy?...
arXiv:2609.01354v1 Announce Type: cross Abstract: Reinforcement learning with verifiable rewards (RLVR) and standard benchmark evaluation both rely on an automatic verifier that turns a free text ans...
The paper introduces a penalty‑aware evaluation framework for Retrieval‑Augmented Generation (RAG) systems that uses asymmetric scoring, knowledge‑gap canaries, and a failure‑attribution pipeline. Applying this framework to three commercial RAG products and a baseline on SimpleQA‑Verified, the authors find that while overall accuracy is similar across systems, canary violation rates vary dramatically, showing that systems differ more in when they answer than in what they answer. The study demonstrates that penalty‑aware scoring can reorder system rankings and is robust across different penalty settings.