Context Dependence and Reliability in Autoregressive Language Models.
Poushali Sengupta, Shashi Raj Pandey, Sabita Maharjan, Frank Eliassen · 2026
Read Context Dependence and Reliability in Autoregressive Language Models. by Poushali Sengupta, Shashi Raj Pandey, Sabita Maharjan, Frank Eliassen (2026) in the Tower of Babel, a free online library of public-domain works preserved in clean, searchable text. No sign-up. The full text is served as a plain-text file at a predictable URL, where text is present: https://antarctic-labs.com/tower-of-babel/arxiv-2602-01378.txt.
Description
Poushali Sengupta and co-authors build RISE, a redundancy-proof score for explaining language-model outputs. Big models draw on rich context full of repeats from prompts, retrieved passages, and chat history, yet standard explanation scores wobble when context parts overlap. Tiny input tweaks can swing credit wildly. RISE gauges each input's unique pull relative to the rest, damping repeat effects for clearer, steadier credit. Tests show RISE beating classic explanation plans on robustness, proving conditional info matters for trustworthy explanation and watch.
Catalog record
| CREATOR | Poushali Sengupta, Shashi Raj Pandey, Sabita Maharjan, Frank Eliassen |
|---|---|
| YEAR | 2026 |
| COLLECTION | PAPERS |
| FORMAT | TXT |
| FILE SIZE | 123687 |
| LICENSE | http://creativecommons.org/licenses/by/4.0/ |
| RIGHTS | CC BY |
| SOURCE | arXiv |
Source: https://arxiv.org/abs/2602.01378