Absence Bench: Language Models Can’t See What’s Missing

Ari Holtzman (University of Chicago) · Chenhao Tan (University of Chicago) · Harvey Yiyun Fu (University of Chicago) · Aryan Shrivastava (University of Chicago) · Jared Moore (Stanford University) · Peter West (University of British Columbia)

absencebenchcontext lengthdocument gapsf1-score fundamental limitationsgithub pull requestsinformation omissionmissing information detectionmodel performance evaluationneedle in a haystacknumerical sequencespoetry analysissuperhuman capabilitiestask proximity analysistransformer attention mechanisms

Large language models (LLMs) are increasingly capable of processing long inputs and locating specific information within them, as evidenced by their performance on the Needle in a Haystack (NIAH) test. However, while models excel at recalling surprising information, they still struggle to identify clearly omitted information. We introduce AbsenceBench to assesses LLMs' capacity to detect missing information across three domains: numerical sequences, poetry, and GitHub pull requests. AbsenceBench asks models to identify which pieces of a document were deliberately removed, given access to both the original and edited contexts. Despite the apparent straightforwardness of these tasks, our experiments reveal that even state-of-the-art models like Claude-3.7-Sonnet achieve only 69.6% F1-score with a modest average context length of 5K tokens. Our analysis suggests this poor performance stems from a fundamental limitation: Transformer attention mechanisms cannot easily attend to "gaps" in documents since these absences don't correspond to any specific keys that can be attended to. Overall, our results and analysis provide a case study of the close proximity of tasks where models are already superhuman (NIAH) and tasks where models breakdown unexpectedly (AbsenceBench).