12 papers
On the Limits of Steering Vectors for Preference-Aligned Generation
Melanie Subbiah, Zara Hall, Kathleen McKeown
Steering vectors have emerged as a promising approach to controlled text generation, offering interpretable, training-free mechanisms for shaping model outputs. However, their prac…
Detect, Remask, Repair: Diffusion Editing for Faithful Summarization of Evolving Contexts
Hao Zou, Zachary Horvitz, Chandhru Karthick +2
Summaries of real-world events can become outdated as contexts evolve and new information arrives. A common response is to generate a new summary from the updated context, but full…
Whose Story Gets Told? Positionality and Bias in LLM Summaries of Life Narratives
Melanie Subbiah, Haaris Mian, Nicholas Deas +3
Increasingly, studies are exploring using Large Language Models (LLMs) for accelerated or scaled qualitative analysis of text data. While we can compare LLM accuracy against human…
Computational Representations of Character Significance in Novels
Haaris Mian, Melanie Subbiah, Sharon Marcus +2
Characters in novels have typically been modeled based on their presence in scenes in narrative, considering aspects like their actions, named mentions, and dialogue. This concepti…
Counterfactual Simulatability of LLM Explanations for Generation Tasks
Marvin Limpijankit, Yanda Chen, Melanie Subbiah +2
LLMs can be unpredictable, as even slight alterations to the prompt can cause the output to change in unexpected ways. Thus, the ability of models to accurately explain their behav…
Artificial Impressions: Evaluating Large Language Model Behavior Through the Lens of Trait Impressions
Nicholas Deas, Kathleen McKeown
We introduce and study artificial impressions--patterns in LLMs' internal representations of prompts that resemble human impressions and stereotypes based on language. We fit linea…