SelfCite: Self-Supervised Alignment for Context Attribution in Large Language Models
We introduce SelfCite, a novel self-supervised approach that aligns LLMs to generate high-quality, fine-grained, sentence-level citations for the statements in their generated responses. Instead of only relying on costly and labor-intensive annotations, SelfCite leverages a reward signal provided by the LLM itself through context ablation: If a citation is necessary, removing the cited text from the context should prevent the same response; if sufficient, retaining the cited text alone should preserve the same response. This reward can guide the inference-time best-of-N sampling strategy to improve citation quality significantly, as well as be used in preference optimization to directly fine-tune the models for generating better citations. The effectiveness of SelfCite is demonstrated by increasing citation F1 up to 5.3 points on the LongBench-Cite benchmark across five long-form question answering tasks.
Code (1)
Tasks
Long Form Question AnsweringQuestion AnsweringSentenceSimilar Papers 제목 키워드 기반
Unsupervised Synthetic Image Attribution: Alignment and Disentanglement
As the quality of synthetic images improves, identifying the underlying concepts of model-generated images is becoming increasingly crucial for copyright protection and ensuring model transparency. Existing methods achie…
Self-Supervised LearningImage AttributionA Sober Look at Agentic Misalignment in Automated Workflows
We study a class of emergent misalignment in multi-agent systems (MAS), with a focus on automated workflows, which we refer to agentic misalignment. Although these systems can solve complex tasks, they often fail because…
A Self-supervised Representation Learning of Sentence Structure for Authorship Attribution
Syntactic structure of sentences in a document substantially informs about its authorial writing style. Sentence representation learning has been widely explored in recent years and it has been shown that it improves the…
Authorship AttributionRepresentation LearningSentenceWord EmbeddingsLeveraging Ensembles and Self-Supervised Learning for Fully-Unsupervised Person Re-Identification and Text Authorship Attribution
Learning from fully-unlabeled data is challenging in Multimedia Forensics problems, such as Person Re-Identification and Text Authorship Attribution. Recent self-supervised learning methods have shown to be effective whe…
Authorship AttributionAuthorship VerificationPerson Re-IdentificationSelf-Supervised Learning+1TAME: Token Attribution and Masking for Emergent misalignment
Fine-tuning an aligned language model on narrow, flawed data can induce harmful behavior far outside the training domain, known as emergent misalignment (EM). Prior work has localized EM in model weights, activations, an…