Saved in:
| Main Authors: | , |
|---|---|
| Format: | Preprint |
| Published: |
2024
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2412.14368 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866908991259410432 |
|---|---|
| author | Jiang, Yuxuan Ferraro, Francis |
| author_facet | Jiang, Yuxuan Ferraro, Francis |
| contents | Recently, Large Language Models (LLMs) have shown impressive performance in character understanding tasks, such as analyzing the roles, personalities, and relationships of fictional characters. However, the extensive pre-training corpora used by LLMs raise concerns that they may rely on memorizing popular fictional works rather than genuinely understanding and reasoning about them. In this work, we argue that 'gist memory'-capturing essential meaning - should be the primary mechanism for character understanding tasks, as opposed to 'verbatim memory' - exact match of a string. We introduce a simple yet effective method to mitigate mechanized memorization in character understanding evaluations while preserving the essential implicit cues needed for comprehension and reasoning. Our approach reduces memorization-driven performance on popular fictional works from 96% accuracy to 72% and results in up to an 18% drop in accuracy across various character understanding tasks. These findings underscore the issue of data contamination in existing benchmarks, which often measure memorization rather than true character understanding. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2412_14368 |
| institution | arXiv |
| publishDate | 2024 |
| record_format | arxiv |
| spellingShingle | Beyond Math: Stories as a Testbed for Memorization-Constrained Reasoning in LLMs Jiang, Yuxuan Ferraro, Francis Computation and Language Recently, Large Language Models (LLMs) have shown impressive performance in character understanding tasks, such as analyzing the roles, personalities, and relationships of fictional characters. However, the extensive pre-training corpora used by LLMs raise concerns that they may rely on memorizing popular fictional works rather than genuinely understanding and reasoning about them. In this work, we argue that 'gist memory'-capturing essential meaning - should be the primary mechanism for character understanding tasks, as opposed to 'verbatim memory' - exact match of a string. We introduce a simple yet effective method to mitigate mechanized memorization in character understanding evaluations while preserving the essential implicit cues needed for comprehension and reasoning. Our approach reduces memorization-driven performance on popular fictional works from 96% accuracy to 72% and results in up to an 18% drop in accuracy across various character understanding tasks. These findings underscore the issue of data contamination in existing benchmarks, which often measure memorization rather than true character understanding. |
| title | Beyond Math: Stories as a Testbed for Memorization-Constrained Reasoning in LLMs |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2412.14368 |