Papers for

dataset curators

Papers whose findings have a practical use for this group, as judged from the abstract. Open a paper to read what it means in practice.

Truthfulness benchmarks cleaned to avoid misleading answer clues

Judging by the Cover: Cleaning LLM Truthfulness Benchmarks to Avoid Surface-Level Feature Leakage

Abstract: Binary-choice truth benchmarks ask models to choose between a correct and an incorrect answer, but if the two answers differ systematically in surface-level features, models can exceed chance without performing the intended reasoning. We show that this failure mode is detectable and can be exploited by downstream classifiers. In TruthfulQA, a simple six-feature logistic classifier achieves substantial accuracy in separating correct from incorrect answers. We further show that similar surface-level artifacts are present in additional benchmarks. To counteract this, we developed a general mechanism to clean them by removing the most leakage-reinforcing pairs. We release a version of TruthfulQA with surface-feature leakage reduced close to chance and provide a mechanism, Audit-Prune, so that the datasets can be cleaned before release.

Fri 11 SeptComputation and LanguageMachine Learning
The gist
Sometimes, tests that check if AI can pick the true answer instead rely on easy clues like word patterns, not real reasoning. The authors found that some AI truthfulness tests had these clues, letting AI do better without truly understanding. They created a way to clean these tests by removing pairs of answers that leak these clues. They also offer a tool so new tests can be cleaned before use to get more honest results.
Open 2609.13003v1

Sign language translation accuracy drops when tested with new signers

Rethinking Sign Language Translation: The Impact of Signer Dependence on Model Evaluation

Abstract: Sign Language Translation has advanced with deep learning, yet evaluations remain largely signer-dependent, with overlapping signers across train/dev/test. This raises concerns about whether models truly generalise or instead rely on signer-specific regularities. We conduct signer-fold cross-validation on GFSLT-VLP, GASLT, and SignCL, three leading, publicly available, gloss-free SLT models, on CSL-Daily and PHOENIX14T. Under signer-independent evaluation, performance drops sharply: on PHOENIX14T, GFSLT-VLP falls from BLEU-4 21.44 to 3.59 and ROUGE-L 42.49 to 11.89; GASLT from 15.74 to 8.26; and SignCL from 22.74 to 3.66. We also observe that in CSL-Daily many target sentences are performed by multiple signers, so common splits can place identical sentences in both training and test, inflating absolute scores by rewarding recall of recurring sentences rather than genuine generalisation. These findings indicate that signer-dependent evaluation can substantially overestimate SLT capability. We recommend: (1) adopting signer-independent protocols to ensure generalisation to unseen signers; (2) restructuring datasets to include explicit signer-independent, sentence-disjoint splits for consistent benchmarking; and (3) reporting both signer-dependent and signer-independent results together with train-test sentence overlap to improve transparency and comparability.

Mon 7 SeptComputation and LanguageArtificial Intelligence
The gist
Sign language translation models seem to work well when tested on people they have already seen, but the authors found that their performance drops a lot when tested on new signers they haven't seen before. This means current tests might overestimate how good these models really are at understanding sign language from different people. The authors suggest new ways to test and organize data so we get a clearer picture of how well these models truly work for everyone.
Open 2609.07965v1