OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Developer Tools & SDKs

Deformable Attention

Deformable Attention is a Python library implementing PyTorch deformable-attention modules for one-, two- and three-dimensional tensor inputs. It includes configurable attention and offset parameters and describes a continuous positional embedding modification.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Deformable Attention — OpenViST project profile in AI Developer Tools & SDKs.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use it when implementing a PyTorch model and you need a reusable deformable-attention component for 1D, 2D or 3D tensor inputs.

Who it is for

  • PyTorch developers building computer-vision or transformer models.
  • Researchers experimenting with deformable attention on image, video or sequence inputs.

Documented uses

  • Add a deformable-attention module to a PyTorch architecture.
  • Process feature maps, video-shaped tensors or sequences using the corresponding dimensional module.
  • Adjust documented attention, downsampling and offset parameters.
Before you choose

Requirements & limits

The public GitHub repository lists an MIT license. The README documents installation with pip.

  • Select the 1D, 2D or 3D class according to the structure of the input tensor.
  • The 2D and 3D examples expose settings for downsampling and offset behavior; the 2D example also shows head configuration.
  • This is a model component, not a packaged training workflow or ready-to-use application.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Install the library and prepare an example input

Install the package and create a PyTorch attention module for a documented input shape.

Before you start

  • Python environment with PyTorch available, as used in the README examples

Steps

  1. Install the package with the command below.

    Documented command
    pip install deformable-attention

  2. Import torch and the appropriate class: DeformableAttention, DeformableAttention1D, or DeformableAttention3D.

  3. Initialize the module with settings suited to the input dimensions; the README provides example settings for image, video-shaped and sequence tensors.

  4. Prepare a tensor with the corresponding documented shape and pass it to the module.

Expected outcome

A configured module and an example input/output shape for adding deformable attention to a PyTorch project.

Guide limitations

  • The examples demonstrate module use, not model training or a complete application.
  • The README does not prescribe settings for a particular downstream model.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
14 October 2026
Original website title
lucidrains/deformable-attention · Official homepage title · retained in its source language
How we review sourcesSuggest a correction