相关论文: Mind the Gap: How Elicitation Protocols Shape the …

Code LLMs often default to particular programming languages and libraries under neutral prompts. We investigate whether these preferences are encoded as approximately linear directions in activation space that can be manipulated at…

Large Language Models (LLMs), though shown to be effective in many applications, can vary significantly in their response quality. In this paper, we investigate this problem of prompt fairness: specifically, the phrasing of a prompt by…

Self-play preference optimization has emerged as a prominent paradigm for aligning large language models (LLMs). It typically involves a language model to generate on-policy responses for prompts and a reward model (RM) to guide the…

A key requirement in developing Generative Language Models (GLMs) is to have their values aligned with human values. Preference-based alignment is a widely used paradigm for this purpose, in which preferences over generation pairs are first…

Fine-tuning a pretrained language model on a curated dataset can produce spurious correlations between the fine-tuning task and unintended latent factors -- such as misaligned personas or political slant -- that the curation procedure has…

LLM-as-a-judge has become the de facto approach for evaluating LLM outputs. However, judges are known to exhibit self-preference bias (SPB): they tend to favor outputs produced by themselves or by models from their own family. This skews…

Reinforcement Learning from Human Feedback (RLHF) has achieved considerable success in aligning large language models (LLMs) by modeling human preferences with a learnable reward model and employing a reinforcement learning algorithm to…

Recent state-of-the-art recommender systems predominantly rely on either implicit or explicit feedback from users to suggest new items. While effective in recommending novel options, many recommender systems often use uninterpretable…

Aligning large language models (LLMs) with human preferences has been recognized as the key to improving LLMs' interaction quality. However, in this pluralistic world, human preferences can be diversified due to annotators' different…

Exposure bias describes the phenomenon that a language model trained under the teacher forcing schema may perform poorly at the inference stage when its predictions are conditioned on its previous predictions unseen from the training…

Preference alignment is a critical step in making Large Language Models (LLMs) useful and aligned with (human) preferences. Existing approaches such as Reinforcement Learning from Human Feedback or Direct Preference Optimization typically…

Conversational Recommendation Systems (CRSs) have recently started to leverage pretrained language models (LM) such as BERT for their ability to semantically interpret a wide range of preference statement variations. However, pretrained LMs…

Large Language Models (LLMs) have been shown to exhibit various biases and stereotypes in their generated content. While extensive research has investigated biases in LLMs, prior work has predominantly focused on explicit bias, with minimal…

In this work we show how large language models (LLMs) can learn statistical dependencies between otherwise unconditionally independent variables due to dataset selection bias. To demonstrate the effect, we developed a masked gender task…

While various approaches have recently been studied for bias identification, little is known about how implicit language that does not explicitly convey a viewpoint affects bias amplification in large language models. To examine the…

LLM confidence calibration is often evaluated by comparing two signals: token-probability scores and verbalized confidence. These signals are sometimes treated as direct readouts of model uncertainty, but their comparison depends on…

Recent advances in reinforcement learning from human feedback (RLHF) and preference optimization have substantially improved the usability, coherence, and safety of large language models. However, recurring behaviors such as performative…

Ensuring that large language models (LLMs) are both helpful and harmless is a critical challenge, as overly strict constraints can lead to excessive refusals, while permissive models risk generating harmful content. Existing approaches,…

As AI systems approach superhuman capabilities, scalable oversight increasingly relies on LLM-as-a-judge frameworks where models evaluate and guide each other's training. A core assumption is that binary preference labels provide only…

Large language models (LLMs) have shown promising abilities as cost-effective and reference-free evaluators for assessing language generation quality. In particular, pairwise LLM evaluators, which compare two generated texts and determine…