相关论文: Layer of Truth: Probing Belief Shifts under Contin…

Large language models are pre-trained on uncurated text datasets consisting of trillions of tokens scraped from the Web. Prior work has shown that: (1) web-scraped pre-training datasets can be practically poisoned by malicious actors; and…

Large Language Models (LLMs) have demonstrated impressive capabilities in reasoning, positioning them as promising tools for supporting human problem-solving. However, what happens when their performance is affected by misinformation, i.e.,…

Through systematic empirical investigation, we uncover a fundamental and concerning property of Large Language Models: while they can safely learn facts that don't contradict their knowledge, attempting to update facts with contradictory…

Large language models (LLMs) encapsulate vast amounts of knowledge but still remain vulnerable to external misinformation. Existing research mainly studied this susceptibility behavior in a single-turn setting. However, belief can change…

Large language models have the potential to generate explanations for their own predictions in a variety of styles based on user instructions. Recent research has examined whether these self-explanations faithfully reflect the models'…

Modern language models can imitate complex patterns through few-shot learning, enabling them to complete challenging tasks without fine-tuning. However, imitation can also lead models to reproduce inaccuracies or harmful content if present…

To reliably assist human decision-making, LLMs must maintain factual internal beliefs against misleading injections. While current models resist explicit misinformation, we uncover a fundamental vulnerability to sophisticated,…

Backdoor data poisoning, inserted within instruction examples used to fine-tune a foundation Large Language Model (LLM) for downstream tasks (\textit{e.g.,} sentiment prediction), is a serious security concern due to the evasive nature of…

The opacity in developing large language models (LLMs) is raising growing concerns about the potential contamination of public benchmarks in the pre-training data. Existing contamination detection methods are typically based on the text…

Large Language Models (LLMs) are capable of recalling multilingual factual knowledge present in their pretraining data. However, most studies evaluate only the final model, leaving the development of factual recall and crosslingual…

The pre-training of large language models (LLMs) relies on massive text datasets sourced from diverse and difficult-to-curate origins. Although membership inference attacks and hidden canaries have been explored to trace data usage, such…

Large Language Models (LLMs) have revolutionized numerous applications, making them an integral part of our digital ecosystem. However, their reliability becomes critical, especially when these models are exposed to misinformation. We…

Given the growing influx of misinformation across news and social media, there is a critical need for systems that can provide effective real-time verification of news claims. Large language or multimodal model based verification has been…

Large Language Models' knowledge of how to perform cyber-security attacks, create bioweapons, and manipulate humans poses risks of misuse. Previous work has proposed methods to unlearn this knowledge. Historically, it has been unclear…

Why do language models trained on contradictory data prefer correct answers? In controlled experiments with small transformers (3.5M--86M parameters), we show that this preference tracks the compressibility structure of errors rather than…

Large language models (LLMs), despite their powerful capabilities, suffer from factual hallucinations where they generate verifiable falsehoods. We identify a root of this issue: the imbalanced data distribution in the pretraining corpus,…

Poisoning attacks can compromise the safety of large language models (LLMs) by injecting malicious documents into their training data. Existing work has studied pretraining poisoning assuming adversaries control a percentage of the training…

Current pre-trained language models have enabled remarkable improvements in downstream tasks, but it remains difficult to distinguish effects of statistical correlation from more systematic logical reasoning grounded on the understanding of…

Large language models may encounter factual knowledge during pre-training yet fail to reliably use that knowledge after fine-tuning. Despite growing empirical evidence that MLP layers store factual associations and fine-tuning affects…

Language models are becoming the default interface to factual knowledge, yet they often verify outputs more reliably than they generate them. This generation-verification gap (GV-gap) underlies many recent advances in self-improvement and…