
Chen Liang, Daoping Wu, Donghua Yang, Zhiyu Liang, Ying Zhang, Shuguang Xiong, Hongzhi Wang · arXiv (Cornell University) 2026 · 2026
DOI: 10.48550/arxiv.2610.25877
Counts differ because each database indexes a different set of publications. We treat OpenAlex as the canonical count; Google Scholar is not shown (no API, and crawling it violates its ToS).
We study how words in the context support or oppose a language model's response and present DeltaTrace, a novel attribution framework that explains the effect of input tokens on a fixed model response. DeltaTrace measures how the response's log-probability changes when the context tokens being explained are masked. To distribute this difference among source tokens, it constructs operator-specific local rules for finite changes and composes them backward through the computational graph using effective secant slopes. Each token is assigned a signed attribution as the inner product of its embedding displacement and the corresponding backward-propagated slope. Local conservation makes these contributions sum to the measured score difference and preserves their meaning when aggregated over sentences or passages. DeltaTrace supports both attention and linear-attention models, with auxiliary memory growing linearly with sequence length at fixed model dimensions and chunk size. On the evaluated attention model, DeltaTrace achieves the lowest deletion-faithfulness scores (RISE and MAS) on most retrieval and reasoning tasks and the highest needle-in-a-haystack Recall, exceeding FlashTrace by 5.67-24.33 percentage points. On the evaluated hybrid attention and linear-attention model, DeltaTrace lowers RISE on 9 tasks and MAS on 12 relative to FlashTrace.
No comments yet — start the discussion below.