</> Code with Papers by Syntology

Methods › General › Attention Mechanisms › Global-Local Attention

Global-Local Attention

56 papers tagged archive 2025-07-28

archive 2025-07-28 Description, source and code snippet are the archive's method entry.

Global-Local Attention is a type of attention mechanism used in the ETC architecture. ETC receives two separate input sequences: the global input xᵍ = (xᵍ₁, …, xᵍ_(n_g)) and the long input xˡ = (xˡ₁, …xˡ_(nₗ)). Typically, the long input contains the input a standard Transformer would receive, while the global input contains a much smaller number of auxiliary tokens (n_g ≪nₗ). Attention is then split into four separate pieces: global-to-global (g2g), global-tolong (g2l), long-to-global (l2g), and long-to-long (l2l). Attention in the l2l piece (the most computationally expensive piece) is restricted to a fixed radius r ≪nₗ. To compensate for this limited attention span, the tokens in the global input have unrestricted attention, and thus long input tokens can transfer information to each other through global input tokens. Accordingly, g2g, g2l, and l2g pieces of attention are unrestricted.

Source: ETC: Encoding Long and Structured Inputs in Transformers

Papers archive 2025-07-28

30 shown of 56, newest first. Repository counts are the archive's code-links table. A Syntology line states what Syntology ran from that paper's harvested code; it is per sample and not a correctness claim.

Tasks archive 2025-07-28

20 shown of 92 tasks the archive attaches to papers tagged with this method, by distinct papers. A task without a page in the catalog is plain text.

TaskPapers
Classification3
Language Modelling3
Object Detection3
Scheduling3
object-detection3
Computational Efficiency2
Emotion Recognition2
GPU2
Mamba2
Management2
Multi-Armed Bandits2
Position2
Question Answering2
Reinforcement Learning (RL)2
Representation Learning2
Retrieval2
Scene Segmentation2
Segmentation2
Semantic Segmentation2
Thompson Sampling2

Usage over time archive 2025-07-28

Papers per year tagged with Global-Local Attention: 2020 to 2025, peak 13 13 0 2020: 8 papers 2020 2021: 13 papers 2021 2022: 9 papers 2022 2023: 8 papers 2023 2024: 13 papers 2024 2025: 5 papers 2025
Papers per year the archive tags with this method, by the paper's archive date (56 dated). Bars are counts, not a trend claim.

Components: the archive holds no method-to-method composition, so PwC's Components table cannot be rebuilt; the Papers list carries no Results column for the same reason (the archive does not join its leaderboard rows to method tags).

Categories archive 2025-07-28

Attention Mechanisms

Report a problem or propose a change · a person checks every report against the paper or source before anything changes; decisions are listed on /corrections