{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/sparsifying-transformer-models-with","title":"Sparsifying Transformer Models with Trainable Representation Pooling","arxiv_id":"2009.05169","date":"2020-09-10","proceeding":"ACL 2022 5","authors":["Michał Pietruszka","Łukasz Borchmann","Łukasz Garncarek"],"abstract":"We propose a novel method to sparsify attention in the Transformer model by learning to select the most-informative token representations during the training process, thus focusing on the task-specific parts of an input. A reduction of quadratic time and memory complexity to sublinear was achieved due to a robust trainable top-$k$ operator. Our experiments on a challenging long document summarization task show that even our simple baseline performs comparably to the current SOTA, and with trainable pooling, we can retain its top quality, while being $1.8\\times$ faster during training, $4.5\\times$ faster during inference, and up to $13\\times$ more computationally efficient in the decoder.","url_abs":"https://arxiv.org/abs/2009.05169v4","url_pdf":"https://arxiv.org/pdf/2009.05169v4.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"sparsifying-transformer-models-with","repo_url":"https://github.com/applicaai/pyramidions","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":1,"framework":"pytorch","reach":{"status":"ok"}}],"tasks":[{"task_slug":"decoder","task_name":"Decoder"},{"task_slug":"document-summarization","task_name":"Document Summarization"},{"task_slug":"summarization","task_name":"Summarization"},{"task_slug":"text-summarization","task_name":"Text Summarization"}],"methods":[{"method_slug":"absolute-position-encodings","method_name":"Absolute Position Encodings"},{"method_slug":"adam","method_name":"Adam"},{"method_slug":"attention","method_name":"Attention"},{"method_slug":"attention-dropout","method_name":"Attention Dropout"},{"method_slug":"bpe","method_name":"BPE"},{"method_slug":"cosine-annealing","method_name":"Cosine Annealing"},{"method_slug":"dense-connections","method_name":"Dense Connections"},{"method_slug":"dropout","method_name":"Dropout"},{"method_slug":"label-smoothing","method_name":"Label Smoothing"},{"method_slug":"layer-normalization","method_name":"Layer Normalization"},{"method_slug":"linear-layer","method_name":"Linear Layer"},{"method_slug":"linear-warmup-with-cosine-annealing","method_name":"Linear Warmup With Cosine Annealing"},{"method_slug":"multi-head-attention","method_name":"Multi-Head Attention"},{"method_slug":"position-wise-feed-forward-layer","method_name":"Position-Wise Feed-Forward Layer"},{"method_slug":"residual-connection","method_name":"Residual Connection"},{"method_slug":"softmax","method_name":"Softmax"},{"method_slug":"sparse-transformer","method_name":"Sparse Transformer"},{"method_slug":"transformer","method_name":"Transformer"},{"method_slug":"weight-decay","method_name":"Weight Decay"}],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/text-summarization-on-pubmed-1","task":"Text Summarization","dataset":"Pubmed","model":"DeepPyramidion","rank_in_archive_order":11,"of":29,"metrics":{"ROUGE-1":"47.81","ROUGE-2":"21.14"},"uses_additional_data":false},{"leaderboard":"/sota/text-summarization-on-arxiv-summarization","task":"Text Summarization","dataset":"arXiv Summarization Dataset","model":"DeepPyramidion","rank_in_archive_order":2,"of":4,"metrics":{"ROUGE-1":"47.15","ROUGE-2":"19.99"},"uses_additional_data":false},{"leaderboard":"/sota/text-summarization-on-arxiv-summarization","task":"Text Summarization","dataset":"arXiv Summarization Dataset","model":"Blockwise (baseline)","rank_in_archive_order":3,"of":4,"metrics":{"ROUGE-1":"46.85","ROUGE-2":"19.39"},"uses_additional_data":false}],"syntology":{"atlas_url":"https://app.syntology.ai/?focus=2009.05169","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}