Swift-edit

ML / Computer Vision · Python, diffusion models, image editing

Overview

A reimplementation of the SwiftEdit paper with additional performance optimizations and architectural improvements for fast text-guided image editing.

GitHub Writeup

Details

SwiftEdit accelerates diffusion-based editing by reducing redundant computation during inference. This project reproduces the core method from the paper and experiments with changes to the pipeline and model usage to improve runtime and memory efficiency while preserving edit quality.