The paper defines and classifies "Polluted Chinese" (PoC) tokens—Chinese tokens in GPT vocabularies that correspond to pornographic or gambling content. It develops a detector by fine‑tuning an LLM to label such tokens using semantics and search‑engine references, then uses token appearance to speculate on training‑data contamination. Experiments across GPT and 23 other LLMs show widespread PoC tokens, with GPT’s vocabulary containing over 40% long Chinese tokens linked to porn or gambling, and the method is validated on datasets like C4 and Pile.
By Qingjie Zhang, Di Wang, Haoting Qian, Liu Yan, Tianwei Zhang, Ke Xu, Qi Li, Minlie Huang, Hewu Li, Han Qiu
arXiv:2604. 14180v2 Announce Type: replace-cross Abstract: We train a 318M-parameter Transformer language model from scratch on a curated corpus of 1.
By Jiuting Chen, Yuan Lian, Hao Wu, Tianqi Huang, Hiroshi Sasaki, Makoto Kouno, Jongil Choi
arXiv:2609.37594v1 Announce Type: new
Abstract: Uncertainty estimates tell us how unsure a model is, but not why. Without knowing which parts of an input influences a model's uncertainty, we cannot t...
By David Achara, Maryam Sultana, Alexander D. Rast, Fabio Cuzzolin
The paper introduces a pairwise difference learning (PDL) approach for classification, where a model predicts the outcome difference between two instances and uses these predictions to classify a query instance by averaging over training examples. It presents a meta‑learning technique tailored to this setting and tackles uncertainty quantification by decomposing total uncertainty into aleatoric and epistemic components, extending these measures to PDL. Extensive empirical studies evaluate both predictive accuracy and the model’s awareness of uncertainty.
By Mohamed Karim Belaid, Maximilian Rabus, Eyke H\"ullermeier
arXiv:2602. 22221v2 Announce Type: replace-cross Abstract: Search engines and AI-powered systems increasingly mediate access to factual information, yet their reliability remains difficult to evaluate in realistic information-seeking settings.
By Geng Liu, Li Feng, Mengxiao Zhu, Francesco Pierri
The paper investigates Mandarin Chinese reduplicative constructions that repeat two-character base words or their constituents, such as expressions meaning ‘in good health’ or ‘discuss a bit’. Using Tencent word embeddings, the study demonstrates that distributional semantics can recover known semantic and grammatical properties of these reduplications, revealing clear semantic and pragmatic differentiation between the two patterns. Procrustes analysis shows that the overall organization of the base-word space is largely preserved in the reduplication space, with local mismatches indicating discourse-pragmatic reorganization.
By Chaoyi Wu, Yu-Hsiang Tseng, R. Harald Baayen
The paper presents new models for inferring gender from Chinese names, addressing the ambiguity inherent in Pinyin-only representations. Two hanzi-based models—a regression model producing a continuous gender score and a classification model yielding male, neutral, or female labels—are trained on Chinese-language physics publications and achieve high predictive accuracy. A Pinyin pipeline aggregates these hanzi predictions with transliteration and frequency weighting, adding an ambiguity score to flag unresolved cases, and outperforms existing tools by assigning gender to all names in an APS journal sample.
By Lisette Esp\'in-Noboa, Mingrong She, Ana Maria Jaramillo, Liuhuaying Yang
The paper introduces Label-Confidence-Aware Uncertainty Quantification (LCA-UQ), a method that uses Pointwise Kullback-Leibler divergence to align global entropy from multiple stochastic samples with the local confidence of a candidate answer. By bridging this gap, LCA-UQ improves the reliability and stability of uncertainty assessments in natural language generation. Experiments on popular LLMs and NLP datasets show that label sources significantly influence classification and that LCA-UQ outperforms existing uncertainty estimation approaches.
By Qinhong Lin, Yinglun Feng, Yuhao Zhang, Zhongliang Yang, Linna Zhou
arXiv:2512.05461v2 Announce Type: replace-cross
Abstract: Large language models (LLMs) are rapidly being integrated into computational social science research, yet their blackboxed training and desig...
By Bolun Zhang, Linzhuo Li, Yunqi Chen, Qinlin Zhao, Zihan Zhu, Xiaoyuan Yi, Xing Xie
arXiv:2610.08879v1 Announce Type: new
Abstract: Pretraining strategies significantly impact the quality of language models, yet existing comparisons of Masked Language Modeling (MLM), Whole Word Mask...
By Yiping Bai
The neutral, or floating, tone of Mandarin Chinese is a tone with an enigmatic set of properties. It has been described as a reduced tone, or as a tone that sometimes is lexically fixed but that can also be toneless.
arXiv:2607. 08377v1 Announce Type: new Abstract: Uncertainty quantification is central to the reliable deployment of large language models (LLMs), and eigenvalues of semantic embeddings have recently emerged as a key tool in state-of-the-art methods.
By Sebastian G. Gruber, Nassim Walha, Francis Bach, Florian Buettner