PyTorch code and checkpoints release for OFA-KD: https://arxiv.org/abs/2310.19444

Adaptive Length Image Tokenization via Recurrent Allocation | How many tokens is an image worth ?

A minimal mode-line for emacs which is not highly configurable and doesn't work in terminal

Code for NeurIPS 2024 paper - The GAN is dead; long live the GAN! A Modern Baseline GAN - by Huang et al.

Symbol overlay extensions for Emacs multiple cursors

easysession.el - Persist and Restore your Emacs Sessions effortlessly

HunyuanVideo: A Systematic Framework For Large Video Generation Model

SEED-Voken: A Series of Powerful Visual Tokenizers

[CVPR2024] Official Code for the "AdaBM: On-the-Fly Adaptive Bit Mapping for Image Super-Resolution"

An Aspiring Drop-In Replacement for NumPy at Scale

The best OSS video generation models

PyTorch codes for "Real-World Blind Super-Resolution via Feature Matching with Implicit High-Resolution Priors", ACM MM2022 (Oral)

Official PyTorch Implementation of "StarSRGAN: Improving Real-World Blind Super-Resolution" (WSCG 2023)

Single-file Emacs starter kit without 3rd-party packages. Almost vanilla Emacs, with just the right amount of sweetness to flatten the learning curve.

Unofficial implementation of Image Super-Resolution via Iterative Refinement by Pytorch

Conditioning and feature fusion methods such as FiLM, Conditional Layer Norm and AdaIN.

Region-Aware Text-to-Image Generation via Hard Binding and Soft Refinement 🔥

A nicer interface for query-replace on Emacs

Official Implementations for Paper - MagicQuill: An Intelligent Interactive Image Editing System

One-step image-to-image with Stable Diffusion turbo: sketch2image, day2night, and more

document image degradation

[NeurlPS2024] One-Step Effective Diffusion Network for Real-World Image Super-Resolution

[CVPR 2024] Low-Res Leads the Way: Improving Generalization for Super-Resolution by Self-Supervised Learning

[CVPR 2021] Unsupervised Degradation Representation Learning for Blind Super-Resolution

This is the official testing code of the baseline method presented at the CVPR 2023 NTIRE Real-Time 4K Super-Resolution Challenge. We provide model and pre-trained checkpoints.