Implementing DDPM, U-Net, Classifier-Free Guidance, and CLIP-style Cosine Similarity from scratch on FashionMNIST — based on NVIDIA DLI Generative AI with Diffusion Models curriculum. / NVIDIA DLI 'Generative AI with Diffusion Models' 커리큘럼 기반으로 DDPM, U-Net, CFG, CLIP 코사인 유사도를 처음부터 직접 구현한 생성형 AI 포트폴리오.
Related
Saturn-He/DOGAN: Official PyTorch implementation of DOGAN: DINO-based optical-prior-driven GAN for SAR-to-optical image translation (IEEE TGRS 2025).
Official PyTorch implementation of DOGAN: DINO-based optical-prior-driven GAN for SAR-to-optical image translation (IEEE TGRS 2025).
dickoyudha-netizen/adobe-firefly-image-creator: Generate images from text prompts with Adobe Firefly Image Creator.
Generate images from text prompts with Adobe Firefly Image Creator.
alexturkey/VideoTurko: MiniMax H3 on a 4 GB GTX 1050 Ti: stable-diffusion.cpp memory patches, up to 209 frames (8.7 s) in one pass
MiniMax H3 on a 4 GB GTX 1050 Ti: stable-diffusion.cpp memory patches, up to 209 frames (8.7 s) in one pass