ACL'2023: DiffusionBERT: Improving Generative Masked Language Models with Diffusion Models

High-performance Image Tokenizers for VAR and AR

Update-to-data resources for conditional content generation, including human motion generation, image or video generation and editing.

This is the official implementation of our neural-network-based fast diffuse room impulse response generator (FAST-RIR) for generating room impulse responses (RIRs) for a given acoustic environment.

This is the official implementation for ControlVAR.

Few-Shot Diffusion Models

Text-driven human motion generation surveys, datasets and models.

Repository for the paper: 'Diffusion-based Conditional ECG Generation with Structured State Space Models'

Code for "Optimal Transport-Guided Conditional Score-Based Diffusion Model (NeurIPS, 8,7,7,6)"

CVPR 2025 Workshop on CVEU.

[NeurIPS 2023] VPP: Efficient Conditional 3D Generation via Voxel-Point Progressive Representation

The code for the NeurIPS 2021 paper "A Unified View of cGANs with and without Classifiers".

Controllable Face Generation via pretrained Conditional Adversarial Latent Autoencoder (ALAE)

Official PyTorch implementation of "Stochastic Conditional Diffusion Models for Robust Semantic Image Synthesis" (ICML 2024).

A PyTorch implementation of various deep generative models, including Diffusion (DDPM), GAN, cGAN, and VAE.

[IGARSS 2026] Generates synthetic image–mask pairs by denoising joint latents conditioned on a real input image (VAE) and a class prompt (CLIP), enabling targeted oversampling of minority classes (e.g., barren) for LoveDA semantic segmentation.

[ICLR 2022] Denoising Likelihood Score Matching for Conditional Score-based Data Generation

Forward-backward conditional sampling

Controllable Sequence Editing for Counterfactual Generation

Quantitative framework for measuring how conditioning effectiveness varies with noise level in diffusion model inference (SD 1.5 & SDXL)