Radar: Fast Long-Context Decoding for Any Transformer
Fuente:
arXiv
Gespeichert in:
| Hauptverfasser: | , , , , |
|---|---|
| Format: | Preprint |
| Veröffentlicht: |
2025
|
| Schlagworte: | |
| Online-Zugang: | |
| Tags: |
Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
|
| _version_ | 1866929758546165760 |
|---|---|
| author | Hao, Yongchang Zhai, Mengyao Hajimirsadeghi, Hossein Hosseini, Sepidehsadat Tung, Frederick |
| author_facet | Hao, Yongchang Zhai, Mengyao Hajimirsadeghi, Hossein Hosseini, Sepidehsadat Tung, Frederick |
| contents | Transformer models have demonstrated exceptional performance across a wide range of applications. Though forming the foundation of Transformer models, the dot-product attention does not scale well to long-context data since its time requirement grows quadratically with context length. In this work, we propose Radar, a training-free approach that accelerates inference by dynamically searching for the most important context tokens. For any pre-trained Transformer, Radar can reduce the decoding time complexity without training or heuristically evicting tokens. Moreover, we provide theoretical justification for our approach, demonstrating that Radar can reliably identify the most important tokens with high probability. We conduct extensive comparisons with the previous methods on a wide range of tasks. The results demonstrate that Radar achieves the state-of-the-art performance across different architectures with reduced time complexity, offering a practical solution for efficient long-context processing of Transformers. |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2503_10571 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Radar: Fast Long-Context Decoding for Any Transformer Hao, Yongchang Zhai, Mengyao Hajimirsadeghi, Hossein Hosseini, Sepidehsadat Tung, Frederick Machine Learning Transformer models have demonstrated exceptional performance across a wide range of applications. Though forming the foundation of Transformer models, the dot-product attention does not scale well to long-context data since its time requirement grows quadratically with context length. In this work, we propose Radar, a training-free approach that accelerates inference by dynamically searching for the most important context tokens. For any pre-trained Transformer, Radar can reduce the decoding time complexity without training or heuristically evicting tokens. Moreover, we provide theoretical justification for our approach, demonstrating that Radar can reliably identify the most important tokens with high probability. We conduct extensive comparisons with the previous methods on a wide range of tasks. The results demonstrate that Radar achieves the state-of-the-art performance across different architectures with reduced time complexity, offering a practical solution for efficient long-context processing of Transformers. |
| title | Radar: Fast Long-Context Decoding for Any Transformer |
| topic | Machine Learning |
| url | https://arxiv.org/abs/2503.10571 |