Zobrazeno 1 - 10
of 2 655
pro vyhledávání: '"Document modeling"'
Akademický článek
Tento výsledek nelze pro nepřihlášené uživatele zobrazit.
K zobrazení výsledku je třeba se přihlásit.
K zobrazení výsledku je třeba se přihlásit.
The integration of multi-document pre-training objectives into language models has resulted in remarkable improvements in multi-document downstream tasks. In this work, we propose extending this idea by pre-training a generic multi-document model fro
Externí odkaz:
http://arxiv.org/abs/2305.15387
Utilizing pre-trained language models has achieved great success for neural document ranking. Limited by the computational and memory requirements, long document modeling becomes a critical issue. Recent works propose to modify the full attention mat
Externí odkaz:
http://arxiv.org/abs/2202.10870
Transformer is important for text modeling. However, it has difficulty in handling long documents due to the quadratic complexity with input text length. In order to handle this problem, we propose a hierarchical interactive Transformer (Hi-Transform
Externí odkaz:
http://arxiv.org/abs/2106.01040
In this paper, we introduce a two-level attention schema, Poolingformer, for long document modeling. Its first level uses a smaller sliding window pattern to aggregate information from neighbors. Its second level employs a larger window to increase r
Externí odkaz:
http://arxiv.org/abs/2105.04371
Publikováno v:
In Information Sciences April 2023 623:40-55
Akademický článek
Tento výsledek nelze pro nepřihlášené uživatele zobrazit.
K zobrazení výsledku je třeba se přihlásit.
K zobrazení výsledku je třeba se přihlásit.
Transformers are not suited for processing long documents, due to their quadratically increasing memory and time consumption. Simply truncating a long document or applying the sparse attention mechanism will incur the context fragmentation problem or
Externí odkaz:
http://arxiv.org/abs/2012.15688
Autor:
Zheng, Bo, Wen, Haoyang, Liang, Yaobo, Duan, Nan, Che, Wanxiang, Jiang, Daxin, Zhou, Ming, Liu, Ting
Natural Questions is a new challenging machine reading comprehension benchmark with two-grained answers, which are a long answer (typically a paragraph) and a short answer (one or more entities inside the long answer). Despite the effectiveness of ex
Externí odkaz:
http://arxiv.org/abs/2005.05806
Publikováno v:
Известия Иркутского государственного университета: Серия "Математика", Vol 32, Iss 1, Pp 79-93 (2020)
The paper considers the methodological and applied aspects of document modeling, a dialect of semantic modeling based on the metaphor of a document as the basic information structure. Since document modeling is designed for a wide range of model deve
Externí odkaz:
https://doaj.org/article/8673ad12a82f48c3890535acd3210131