English
Related papers

Related papers: All Leaks Count, Some Count More: Interpretable Te…

200 papers

Post-cutoff performance decay of LLMs has been widely interpreted as a temporal signal for benchmark contamination, where public information released before the training cutoff may have been included into training corpora and inflated model…

Backtesting large language models on historical events requires reasoning exclusively from information available before a specified cutoff date. Yet models routinely leak post-cutoff knowledge from pre-training into their reasoning,…

Machine Learning · Computer Science 2026-05-20 Zeyu Zhang , Bradly C. Stadie

Large language models (LLMs) face significant challenges in ex-ante reasoning, where analysis, inference, or predictions must be made without access to information from future events. Even with explicit prompts enforcing temporal cutoffs,…

Machine Learning · Computer Science 2025-05-27 Yachuan Liu , Xiaochun Wei , Lin Shi , Xinnuo Li , Bohan Zhang , Paramveer Dhillon , Qiaozhu Mei

Concept-based Models aim to improve interpretability by predicting high-level intermediate concepts, representing a promising approach for deployment in high-risk scenarios. However, they are known to suffer from information leakage,…

Machine Learning · Computer Science 2026-03-25 Enrico Parisini , Tapabrata Chakraborti , Chris Harbron , Ben D. MacArthur , Christopher R. S. Banerji

As AI models grow larger, the demand for accountability and interpretability has become increasingly critical for understanding their decision-making processes. Concept Bottleneck Models (CBMs) have gained attention for enhancing…

Machine Learning · Computer Science 2024-10-10 Angelos Ragkousis , Sonali Parbhoo

Adversarial attacks by malicious users that threaten the safety of large language models (LLMs) can be viewed as attempts to infer a target property $T$ that is unknown when an instruction is issued, and becomes knowable only after the…

Cryptography and Security · Computer Science 2025-10-21 Masahiro Kaneko , Timothy Baldwin

Large Language Models (LLMs) are widely used for temporal prediction, but their reliance on pretraining data raises contamination concerns, as accurate predictions on pre-cutoff test data may reflect memorization rather than reasoning,…

Computation and Language · Computer Science 2025-10-16 Xin Gao , Ruiyi Zhang , Daniel Du , Saurabh Mahindre , Sai Ashish Somayajula , Pengtao Xie

Large Language Models (LLMs) demonstrate potential in complex legal tasks like argument generation, yet their reliability remains a concern. Building upon pilot work assessing LLM generation of 3-ply legal arguments using human evaluation,…

Computation and Language · Computer Science 2025-06-04 Li Zhang , Morgan Gray , Jaromir Savelka , Kevin D. Ashley

Released Large Language Models (LLMs) are often paired with a claimed knowledge cutoff date, or the dates at which training data was gathered. Such information is crucial for applications where the LLM must provide up to date information.…

Computation and Language · Computer Science 2024-09-18 Jeffrey Cheng , Marc Marone , Orion Weller , Dawn Lawrie , Daniel Khashabi , Benjamin Van Durme

Large language models (LLMs) often fail to reason under temporal cutoffs: when prompted to answer from the standpoint of an earlier time, they exploit knowledge that became available only later. We study this failure through the lens of…

Artificial Intelligence · Computer Science 2026-05-15 Chenlu Ding , Jiancan Wu , Yanchen Luo , Zheyuan Liu , Yancheng Yuan , Xiang Wang

As reasoning LLMs increasingly trade tokens for accuracy through deliberation, search, and self-correction, a single accuracy score can no longer tell whether those tokens buy useful reasoning, recovery from hard instances, or unnecessary…

Computation and Language · Computer Science 2026-05-19 Daniel Kaiser , Arnoldo Frigessi , Ali Ramezani-Kebrya , Benjamin Ricaud

Large language models are moving from static text generators toward real-world decision-support systems, where forecasting is a composite capability that links information gathering, evidence integration, situational judgment, and…

Artificial Intelligence · Computer Science 2026-05-06 Yiding Ma , Chengyun Ruan , Kaibo Huang , Zhongliang Yang , Linna Zhou

Chain-of-Thought (CoT) prompting improves LLM reasoning but can increase privacy risk by resurfacing personally identifiable information (PII) from the prompt into reasoning traces and outputs, even under policies that instruct the model…

Computation and Language · Computer Science 2026-03-09 Patrick Ahrend , Tobias Eder , Xiyang Yang , Zhiyi Pan , Georg Groh

Deploying LLMs raises two coupled challenges: (1) monitoring--estimating where a model underperforms as traffic and domains drift--and (2) improvement--prioritizing data acquisition to close the largest performance gaps. We test whether an…

Computation and Language · Computer Science 2026-05-27 Pedro Memoli Buffa , Luciano Del Corro

Clinical natural language processing (NLP) models have shown promise for supporting hospital discharge planning by leveraging narrative clinical documentation. However, note-based models are particularly vulnerable to temporal and lexical…

Computation and Language · Computer Science 2026-02-20 Ha Na Cho , Sairam Sutari , Alexander Lopez , Hansen Bow , Kai Zheng

Large language models (LLMs) are increasingly deployed in settings where the available context is incomplete or degraded. We argue that an LLM generating answers under incomplete context can be viewed as an implicit imputer, and evaluated…

Machine Learning · Statistics 2026-05-14 Stef van Buuren

As language models (LMs) deliver increasing performance on a range of NLP tasks, probing classifiers have become an indispensable technique in the effort to better understand their inner workings. A typical setup involves (1) defining an…

Computation and Language · Computer Science 2024-08-01 Charles Jin , Martin Rinard

Concept Bottleneck Models (CBMs) aim to enhance interpretability by structuring predictions around human-understandable concepts. However, unintended information leakage, where predictive signals bypass the concept bottleneck, compromises…

Machine Learning · Computer Science 2025-07-22 Mikael Makonnen , Moritz Vandenhirtz , Sonia Laguna , Julia E Vogt

Large language models (LLMs) are increasingly exposed to data contamination, i.e., performance gains driven by prior exposure of test datasets rather than generalization. However, in the context of tabular data, this problem is largely…

Computation and Language · Computer Science 2026-03-31 Matteo Silvestri , Fabiano Veglianti , Flavio Giorgi , Fabrizio Silvestri , Gabriele Tolomei

While pre-trained language models (LMs) have brought great improvements in many NLP tasks, there is increasing attention to explore capabilities of LMs and interpret their predictions. However, existing works usually focus only on a certain…

Computation and Language · Computer Science 2022-07-29 Yaozong Shen , Lijie Wang , Ying Chen , Xinyan Xiao , Jing Liu , Hua Wu
‹ Prev 1 2 3 10 Next ›