Positional Encoding

Positional Encoding is a deterministic or learned vector representation added to token embeddings in sequence models to provide information about the relative or absolute positions of tokens within a sequence, distinguishing it from other embedding components by explicitly encoding sequential order in contexts where such information is otherwise absent.

317

Publications

35.6K

Citations

1.1K

Authors

389

Institutions

Publications per year

2017–2026

206

Authors

1.1K

  1. LW
    Lewis Wolpert

    University College London, United Kingdom

    7 publications · 3.7K citations

    7

    h-index

  2. SJ
    Sen Jia

    University of Bristol, United Kingdom

    4 publications · 191 citations

    4

    h-index

  3. ND
    Neil D. B. Bruce

    University of Manitoba, Canada

    4 publications · 191 citations

    4

    h-index

  4. DC
    Dongdong Chen

    University of Science and Technology of China, China

    4 publications · 1.7K citations

    4

    h-index

  5. PW
    Peter W. Reddien

    Howard Hughes Medical Institute, United States

    3 publications · 193 citations

    3

    h-index

1–5 of 1.1K

Venues

  1. arXiv (Cornell University)

    Repository · 49 publications · 6.9K citations

    28

    h-index

  2. Developmental Biology

    10 publications · 483 citations

    10

    h-index

  3. Proceedings of the AAAI Conference on Artificial Intelligence

    Conference · 9 publications · 3.2K citations

    9

    h-index

  4. 9

    h-index

  5. Development

    6 publications · 208 citations

    6

    h-index

1–5