Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Preprint |
| Published: |
2026
|
| Subjects: | |
| Online Access: | https://arxiv.org/abs/2605.09463 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| _version_ | 1866913108133412864 |
|---|---|
| author | Tang, Jiwei Huang, Zhijing Zhang, Xinyu Zhang, Chen Jason Yu, Jianxing Zheng, Libin Meng, Rui Yin, Jian |
| author_facet | Tang, Jiwei Huang, Zhijing Zhang, Xinyu Zhang, Chen Jason Yu, Jianxing Zheng, Libin Meng, Rui Yin, Jian |
| contents | Large Language Models (LLMs) have demonstrated exceptional performance across diverse tasks. However, their deployment in long-context scenarios faces high computational overhead and information redundancy. While soft prompt compression has emerged as a promising way to mitigate these costs by compressing sequences into compact embeddings, existing paradigms remain fundamentally constrained by position bias: they primarily rely on learnable tokens insertion at fixed positions or group tokens according to their physical token layout, thereby inducing performance instability and semantic fragmentation. To overcome this bottleneck, we propose Semantic Consistency Context Compression (SeCo), a method that shifts context compression from position-driven to semantic-driven. Rather than constraint by physical token layout, SeCo dynamically anchors compression directly in the semantic space by selecting query-relevant tokens as semantic centers and aggregating remaining tokens via consistency-weighted merging. This design inherently preserves semantic consistency while eliminating position bias. Extensive experiments on 14 benchmarks across two backbone models demonstrate that SeCo consistently shows superiority in downstream tasks, inference latency, and out-of-domain robustness. The code is available at https://anonymous.4open.science/r/seco-EE5E. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2605_09463 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Beyond Position Bias: Shifting Context Compression from Position-Driven to Semantic-Driven Tang, Jiwei Huang, Zhijing Zhang, Xinyu Zhang, Chen Jason Yu, Jianxing Zheng, Libin Meng, Rui Yin, Jian Computation and Language Large Language Models (LLMs) have demonstrated exceptional performance across diverse tasks. However, their deployment in long-context scenarios faces high computational overhead and information redundancy. While soft prompt compression has emerged as a promising way to mitigate these costs by compressing sequences into compact embeddings, existing paradigms remain fundamentally constrained by position bias: they primarily rely on learnable tokens insertion at fixed positions or group tokens according to their physical token layout, thereby inducing performance instability and semantic fragmentation. To overcome this bottleneck, we propose Semantic Consistency Context Compression (SeCo), a method that shifts context compression from position-driven to semantic-driven. Rather than constraint by physical token layout, SeCo dynamically anchors compression directly in the semantic space by selecting query-relevant tokens as semantic centers and aggregating remaining tokens via consistency-weighted merging. This design inherently preserves semantic consistency while eliminating position bias. Extensive experiments on 14 benchmarks across two backbone models demonstrate that SeCo consistently shows superiority in downstream tasks, inference latency, and out-of-domain robustness. The code is available at https://anonymous.4open.science/r/seco-EE5E. |
| title | Beyond Position Bias: Shifting Context Compression from Position-Driven to Semantic-Driven |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2605.09463 |