Search

Generative AI - Courses - IBM Developer

Generative AI involves deep-learning models that produce high-quality content like text and images, based on training data; the page refers to IBM Developer courses on this topic.

[2605.18736] Spectral Progressive Diffusion for Efficient Image and Video Generation

Spectral Progressive Diffusion enhances diffusion-based image and video generation by progressively increasing resolution during denoising, guided by frequency-domain insights and the model's power spectrum. It introduces a spectral noise expansion, an optimal resolution schedule, a training-free acceleration path, and a fine-tuning recipe that boost speed while maintaining quality.

Generative AI Is Becoming Something Much Bigger Than a Chatbot

Generative AI is moving from creating outputs to executing tasks and interacting with the real world, driven by larger data sets, better models, more compute, and memory/tool-use capabilities. This shift emphasizes goal-directed intelligence, with transformers and attention enabling complex reasoning, world models, and robotics, beyond mere next-token generation.

PedroMarinhoDev/Hy-Nomas-ComfyUI · Hugging Face

Hy-Nomas for ComfyUI provides a first-release set of nodes to drive Tencent's Hy-Nomas world model, enabling image- or prompt-driven, segment-by-segment video generation with camera control and drive-mode playback. It includes INT8/FP8 components, setup instructions, and a roadmap for enhancements and additional checkpoints.

SocialAI Generator - Create Viral Content in Seconds

SocialAI Generator makes posts, captions, and hashtags for several social platforms. It is free to use with an OpenRouter key saved in the browser, and donations are optional.

monsterovich/yue2-keygen-music-lora · Hugging Face

Two-stage LoRA setup enhances YuE2-3B for keygen-style chiptune music, using AR for planning and NAR for timbre. The adapters are trained on instrumental keygen data and used together via a LoRA pipeline or CLI, with memory considerations and specific stylized constraints.

Googleが動画生成AI「Gemini Omni 1.2 Flash」をリリース、最大15秒の動画を出力可能 - GIGAZINE

Google announces Gemini Omni 1.2 Flash, a video generation model powering Google Vids, offering up to three 15-second videos per month at no cost for users with a Google account or Google Workspace subscription.

Google DeepMind introduced Genie 4: Real-time simulation

Translated title. Original: Google DeepMind Genie 4’ü tanıttı: Gerçek zamanlı simülasyon

Google DeepMind introduced Genie 4, a model for real-time interactive simulations that blend dynamic worlds with immediate audio. Details on capabilities, access, and use cases are not disclosed, but the system focuses on responsive environments rather than static content.

Google introduced Genie 4, a real-time voice and dialog-based world model

Translated title. Original: Google Genie 4'ü tanıttı, gerçek zamanlı ses ve diyaloglu dünya modeli

Genie 4 is a real-time world model that supports voice- and dialog-based interaction, showing dynamic environments that respond to sounds and human actions in real time.

Google presents Genie 4, the world model with native audio and real-time dialogues

Translated title. Original: Google presenta Genie 4, el world model con audio nativo y diálogos en tiempo real

Google unveils Genie 4, a world model that enables real-time, interactive exploration of dynamic environments with realistic audio, motion, and dialogue of people.