Revisiting Long-context Modeling from Context Denoising Perspective
Fuente:
arXiv
Enregistré dans:
| Auteurs principaux: | , , , , , |
|---|---|
| Format: | Preprint |
| Publié: |
2025
|
| Sujets: | |
| Accès en ligne: | |
| Tags: |
Ajouter un tag
Pas de tags, Soyez le premier à ajouter un tag!
|
| _version_ | 1866911247823273984 |
|---|---|
| author | Tang, Zecheng Ji, Baibei Li, Juntao Wu, Lijun Gui, Haijia Zhang, Min |
| author_facet | Tang, Zecheng Ji, Baibei Li, Juntao Wu, Lijun Gui, Haijia Zhang, Min |
| contents | Long-context models (LCMs) have demonstrated great potential in processing long sequences, facilitating many real-world applications. The success of LCMs can be attributed to their ability to locate implicit critical information within the context for further prediction. However, recent research reveals that LCMs are often susceptible to contextual noise, i.e., irrelevant tokens, that can mislead model attention. In this paper, we conduct a fine-grained analysis of the context noise and propose an effective metric, the Integrated Gradient (IG) score, to detect and quantify the noise information within the context. Our findings reveal that even simple mitigation of detected context noise can substantially boost the model's attention on critical tokens and benefit subsequent predictions. Building on this insight, we propose Context Denoising Training (CDT), a straightforward yet effective training strategy that improves attention on critical tokens while reinforcing their influence on model predictions. Extensive experiments across four tasks, under both context window scaling and long-context alignment settings, demonstrate the superiority of CDT. Notably, when trained with CDT, an open-source 8B model can achieve performance (50.92) comparable to GPT-4o (51.00). |
| format | Preprint |
| id |
arxiv_https___arxiv_org_abs_2510_05862 |
| institution | arXiv |
| publishDate | 2025 |
| record_format | arxiv |
| spellingShingle | Revisiting Long-context Modeling from Context Denoising Perspective Tang, Zecheng Ji, Baibei Li, Juntao Wu, Lijun Gui, Haijia Zhang, Min Computation and Language Artificial Intelligence Long-context models (LCMs) have demonstrated great potential in processing long sequences, facilitating many real-world applications. The success of LCMs can be attributed to their ability to locate implicit critical information within the context for further prediction. However, recent research reveals that LCMs are often susceptible to contextual noise, i.e., irrelevant tokens, that can mislead model attention. In this paper, we conduct a fine-grained analysis of the context noise and propose an effective metric, the Integrated Gradient (IG) score, to detect and quantify the noise information within the context. Our findings reveal that even simple mitigation of detected context noise can substantially boost the model's attention on critical tokens and benefit subsequent predictions. Building on this insight, we propose Context Denoising Training (CDT), a straightforward yet effective training strategy that improves attention on critical tokens while reinforcing their influence on model predictions. Extensive experiments across four tasks, under both context window scaling and long-context alignment settings, demonstrate the superiority of CDT. Notably, when trained with CDT, an open-source 8B model can achieve performance (50.92) comparable to GPT-4o (51.00). |
| title | Revisiting Long-context Modeling from Context Denoising Perspective |
| topic | Computation and Language Artificial Intelligence |
| url | https://arxiv.org/abs/2510.05862 |