YazidJanati / ding Python
DInG
Decoupled Inpainting Guidance: efficient, training-free image inpainting with pretrained diffusion models.
Research software
Repositories
Research implementations and evaluated learning labs, with code and interactive demonstrations.
ahmedgh970 / ding-editor Python
DInG-Editor
A modular package extending decoupled guidance to image and video editing with pretrained models.
These repositories are maintained by collaborators and accompany the papers listed on the Publications page.
Research and learning labs
Six independently developed projects connecting theory, implementation and measured experiments. Each repository includes reproduction instructions, comparisons and an interactive view of its recorded results.
navidbagheri7 / Diffusion-Models Python
Diffusion Models
Diffusion models, score SDEs, flow matching, conditioning and adversarial objectives, connecting equations with experiments and executable lessons.
navidbagheri7 / Reinforcement-Learning Python
Reinforcement Learning
Planning, online and offline reinforcement learning, control and maze simulations, with evaluated policies and interactive rollouts.
navidbagheri7 / Machine-Learning-and-Deep-Learning Python
Machine Learning and Deep Learning
Classical and modern machine learning, probabilistic inference, image restoration and generative models, with measured comparisons and reproducible evaluations.
navidbagheri7 / MCMC-and-Energy-Based-Models Python
MCMC and Energy-Based Models
Gibbs, Metropolis, HMC, NUTS and Langevin sampling, Bayesian inference and energy-based learning, with distribution views and multichain diagnostics.
navidbagheri7 / Convex-Optimization Python
Convex Optimization
Convex certificates, first-order and primal-dual methods, recommendation and combinatorial optimization, with evaluated solvers and interactive comparisons.
navidbagheri7 / Information-Theory Python
Information Theory
Entropy, compression, coding, capacity, rate-distortion and neural information estimation, with reproducible simulations and interactive comparisons.