Saved in:
| Main Authors: | , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2605.27294 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866917536689291264 |
|---|---|
| author | Repantis, Vyzantinos Gawde, Ameya Singh, Harshvardhan Alekar, Rohit Zhang, Cien Karslioglu, Svetlana Vishwakarma, Akash |
| author_facet | Repantis, Vyzantinos Gawde, Ameya Singh, Harshvardhan Alekar, Rohit Zhang, Cien Karslioglu, Svetlana Vishwakarma, Akash |
| contents | Retrieval-augmented generation (RAG) systems can respond incorrectly even when the correct passage was retrieved. The model must still read the retrieved passages and identify which one contains the answer among others that look relevant. This passage-reading model is called the reader. Does it fail simply because the context is longer or because the other passages genuinely compete with the correct one? We introduce and demonstrate a matched-control protocol for RAG reading: we keep the number and length of passages fixed, but replace hard competitors with less competitive real passages. We apply this control across two compact open models on SQuAD. This replacement partially restores performance, with the strongest effects on F1 and answer inclusion. For Phi-2, this recovers +6.0 EM points, +7.0 answer-inclusion points, and +0.057 F1. For Qwen2.5-1.5B, it recovers +4.5 EM points, +9.0 answer-inclusion points, and +0.068 F1. To track how performance changes as competitors accumulate, we also report retention curves and summarize them with a right-censored half-life when the curves do not cross half-retention. Together, these results show the protocol isolates a competition effect distinct from context length, though the effect is clearer for F1 and answer inclusion than for exact match, and also varies with snippet length. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_27294 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Separating Semantic Competition from Context Length in RAG Reading Repantis, Vyzantinos Gawde, Ameya Singh, Harshvardhan Alekar, Rohit Zhang, Cien Karslioglu, Svetlana Vishwakarma, Akash Computation and Language Information Retrieval Retrieval-augmented generation (RAG) systems can respond incorrectly even when the correct passage was retrieved. The model must still read the retrieved passages and identify which one contains the answer among others that look relevant. This passage-reading model is called the reader. Does it fail simply because the context is longer or because the other passages genuinely compete with the correct one? We introduce and demonstrate a matched-control protocol for RAG reading: we keep the number and length of passages fixed, but replace hard competitors with less competitive real passages. We apply this control across two compact open models on SQuAD. This replacement partially restores performance, with the strongest effects on F1 and answer inclusion. For Phi-2, this recovers +6.0 EM points, +7.0 answer-inclusion points, and +0.057 F1. For Qwen2.5-1.5B, it recovers +4.5 EM points, +9.0 answer-inclusion points, and +0.068 F1. To track how performance changes as competitors accumulate, we also report retention curves and summarize them with a right-censored half-life when the curves do not cross half-retention. Together, these results show the protocol isolates a competition effect distinct from context length, though the effect is clearer for F1 and answer inclusion than for exact match, and also varies with snippet length. |
| title | Separating Semantic Competition from Context Length in RAG Reading |
| topic | Computation and Language Information Retrieval |
| url | https://arxiv.org/abs/2605.27294 |