Data Distribution Matters: A Data-Centric Perspective on Context Compression for Large Language Model
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , , , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2026
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866915768524865536 |
|---|---|
| author | Lv, Kangtao Tang, Jiwei Liu, Langming Chen, Haibin Zhang, Weidong Liu, Shilei Wang, Yongwei Yuan, Yujin Su, Wenbo Zheng, Bo |
| author_facet | Lv, Kangtao Tang, Jiwei Liu, Langming Chen, Haibin Zhang, Weidong Liu, Shilei Wang, Yongwei Yuan, Yujin Su, Wenbo Zheng, Bo |
| contents | The deployment of Large Language Models (LLMs) in long-context scenarios is hindered by computational inefficiency and significant information redundancy. Although recent advancements have widely adopted context compression to address these challenges, existing research only focus on model-side improvements, the impact of the data distribution itself on context compression remains largely unexplored. To bridge this gap, we are the first to adopt a data-centric perspective to systematically investigate how data distribution impacts compression quality, including two dimensions: input data and intrinsic data (i.e., the model's internal pretrained knowledge). We evaluate the semantic integrity of compressed representations using an autoencoder-based framework to systematically investigate it. Our experimental results reveal that: (1) encoder-measured input entropy negatively correlates with compression quality, while decoder-measured entropy shows no significant relationship under a frozen-decoder setting; and (2) the gap between intrinsic data of the encoder and decoder significantly diminishes compression gains, which is hard to mitigate. Based on these findings, we further present practical guidelines to optimize compression gains. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2602_01778 |
| institution | arXiv |
| publishDate | 2026 |
| record_format | arxiv |
| spellingShingle | Data Distribution Matters: A Data-Centric Perspective on Context Compression for Large Language Model Lv, Kangtao Tang, Jiwei Liu, Langming Chen, Haibin Zhang, Weidong Liu, Shilei Wang, Yongwei Yuan, Yujin Su, Wenbo Zheng, Bo Computation and Language The deployment of Large Language Models (LLMs) in long-context scenarios is hindered by computational inefficiency and significant information redundancy. Although recent advancements have widely adopted context compression to address these challenges, existing research only focus on model-side improvements, the impact of the data distribution itself on context compression remains largely unexplored. To bridge this gap, we are the first to adopt a data-centric perspective to systematically investigate how data distribution impacts compression quality, including two dimensions: input data and intrinsic data (i.e., the model's internal pretrained knowledge). We evaluate the semantic integrity of compressed representations using an autoencoder-based framework to systematically investigate it. Our experimental results reveal that: (1) encoder-measured input entropy negatively correlates with compression quality, while decoder-measured entropy shows no significant relationship under a frozen-decoder setting; and (2) the gap between intrinsic data of the encoder and decoder significantly diminishes compression gains, which is hard to mitigate. Based on these findings, we further present practical guidelines to optimize compression gains. |
| title | Data Distribution Matters: A Data-Centric Perspective on Context Compression for Large Language Model |
| topic | Computation and Language |
| url | https://arxiv.org/abs/2602.01778 |