AI Infra2025
Extends RAG beyond text: parses PDFs and Office files containing images, tables, equations, and charts, then queries them through one multimodal knowledge graph. Built on LightRAG, it replaces separate parsing and retrieval tools.
Embodied AI2025
Provides a unified Python interface to collect data, train visual/dynamics world models, and evaluate them with model-predictive control across many standardized environments. Includes reference baselines, planning solvers, dataset converters, and LanceDB-backed formats for reproducible experiments. Best suited for researchers benchmarking world-model algorithms.
AI Image2025
Detects, segments, and tracks every instance of an open-vocabulary concept in images and video from a text phrase or visual exemplar, not just one object per prompt. An 848M-param model reaching ~75-80% of human accuracy across 270K concepts.
AI Dataset2025
A collection of ready-to-run Hugging Face Jobs OCR scripts that add a markdown column (or structured JSON) to image datasets, with model switching, layout detection, server-mode serving, and per-model options for table/form extraction.
AI Image2025
Self-supervised vision foundation model producing dense, patch-level features that transfer to classification, segmentation, depth, and detection with a frozen backbone. Spans ViT-S (21M) to ViT-7B (6.7B params), plus ConvNeXt and satellite variants.
AI Agent2025
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.
AI Video2025
Generates explorable, 3D-consistent virtual worlds from a single image or short video. Includes official implementations of Lyra‑1 (feed‑forward 3D/4D scene generation via video-diffusion self-distillation) and Lyra‑2 (long-horizon, explorable generative 3D worlds). Best for research and creative prototyping; requires substantial GPU compute.
AI Video Papers2025
Argues a single web-scale generative video model handles vision tasks zero-shot the way LLMs handle language. Probes Veo 3 on segmentation, edge detection, image editing, physical and affordance reasoning, and puzzles like maze solving and symmetry.
AI Agent2025
Automates multi-step web tasks by perceiving webpages as pixels and issuing low-level mouse, keyboard and scroll actions. A 7B-parameter multimodal agent trained on 145K synthetic trajectories (FaraGen), designed for on-device deployment and efficient task completion (~16 steps/task).
AI Model2025
Performs open-vocabulary promptable segmentation and tracking on images and videos, returning instance masks and unique IDs from short noun-phrase or exemplar prompts. Trained on a 4M-concept, 52M-mask dataset; uses a presence head and a decoupled detector–tracker design, with near real-time inference for multi-object video.
AI Agent2026
Grows a personal skill tree by crystallizing each solved task into reusable skills; a ~3K-line autonomous agent framework that gives an LLM system-level control of browser, terminal, filesystem, input/vision and mobile (ADB) via nine atomic tools, optimized for low token cost.
Reduces object-driven shortcut learning in zero-shot compositional action recognition by enforcing temporal verb cues and regularizing against frequent object-verb co-occurrence priors. Proposes RCORE with Co-occurrence Prior Regularization (treats frequent co-occurrences as hard negatives) and Temporal Order Regularization. Evaluated on Sth-com and EK100-com with improved compositional generalization.