Tag
Explore by tags
Explains how modern LLMs are trained, tokenized, post-trained, and used, from internet-scale pretraining to RLHF and tool use. The value is a coherent mental model, not a quick product tutorial.
Walks through real LLM workflows across chat, search, deep research, file analysis, coding, voice, images, and generated podcasts. It is most useful as a field guide to the messy AI app layer.
Turns a raw idea, novel, or screenplay into a complete multi-shot video through a multi-agent pipeline that scripts, storyboards, and renders shots while a vision model checks character and scene consistency across the whole story.
Provides PyTorch code, pretrained checkpoints, and evaluation tooling for V-JEPA 2 — a Meta FAIR family of self-supervised video encoders and an action-conditioned world model. Includes training recipes, HuggingFace checkpoints, evaluation probes, and robot post‑training artifacts.
Real-time, streaming world model for interactive long-horizon video and scene generation — provides persistent memory across interactions, streaming inference for live scenarios, and example demos for building interactive environments.
Timeline-based video editor for web, desktop and mobile that integrates AI-assisted workflows (MCP server and generative-model integrations) with a Rust core and plugin-first, cross-platform architecture; supports headless batch rendering and no-watermark exports.
An agentic framework that analyzes, plans, and executes multi-step video understanding and editing workflows using multimodal LLM-driven agents—features intent decomposition, graph-based workflow orchestration, and automated shot planning for long-form video tasks.
Framework for building multi-modal AI agents that watch, listen, and reason over live video, pairing vision models (YOLO, Roboflow, Moondream) with LLMs like Gemini and OpenAI. Agents join calls in ~500ms and keep audio/video latency under 30ms.
Provides an NVFP4‑optimized training and inference infrastructure for long-form video diffusion models — supports multi-shot AR training, KV-cache and NVFP4 quantized inference, sequence-parallelism and async decoding for higher FPS and longer outputs.