TISA

December 1, 2021 ยท View on GitHub

This is the code for the article The Case for Translation-Invariant Self-Attention in Transformer-Based Language Models

Please cite this article as follows:

@inproceedings{wennberg-henter-2021-case,
    title = "The Case for Translation-Invariant Self-Attention in Transformer-Based Language Models",
    author = "Wennberg, Ulme  and
      Henter, Gustav Eje",
    booktitle = "Proceedings of the 59th Annual Meeting of the Association for Computational Linguistics and the 11th International Joint Conference on Natural Language Processing (Volume 2: Short Papers)",
    month = aug,
    year = "2021",
    publisher = "Association for Computational Linguistics",
    url = "https://aclanthology.org/2021.acl-short.18",
    doi = "10.18653/v1/2021.acl-short.18",
    pages = "130--140"
}

Model

The TISA model is located in model/tisa.py

Figures

Bert base uncased

bert-base-uncased

Bert large uncased

bert-large-uncased

Bert base cased

bert-base-cased

Bert large cased

bert-large-cased

Roberta base

roberta-base

Roberta large

roberta-large

Albert base v1

albert-base-v1

Albert base v2

albert-base-v2

Albert large v1

albert-large-v1

Albert large v2

albert-large-v2

Albert xlarge v1

albert-xlarge-v1

Albert xlarge v2

albert-xlarge-v2

Albert xxlarge v1

albert-xxlarge-v1

Albert xxlarge v2

albert-xxlarge-v2

Longformer base

allenai_longformer-base-4096

Longformer base

allenai_longformer-large-4096

KB Bert base swedish cased

KB_bert-base-swedish-cased

KB Albert base swedish cased alpha

KB_albert-base-swedish-cased-alpha