ИИ-агенты
IsaacLab
Isaac Lab — NVIDIA's unified framework for robot learning, with 30+ environments and multi-physics simulation.
★ 8,3 тыс. звёздBSD-3-Clause
Смотреть руководство →Isaac-GR00T — проект с открытым исходным кодом из категории «ИИ-агенты» от NVIDIA (NVIDIA/Isaac-GR00T на GitHub). У него 8 173 звезды на GitHub и 1 486 форков, он распространяется по лицензии Apache-2.0 и последний раз обновлялся на GitHub 7 октября 2026.
NVIDIA Isaac GR00T N1.7 — an open foundation model for generalist robots, ready to fine-tune and deploy.
Check the hardware requirements and your robot/dataset format (LeRobot-style datasets are used in the workflow), and install git-lfs before cloning since demo data uses it.
Clone NVIDIA/Isaac-GR00T from GitHub (or use the install path in the run steps below) and skim its README for the parts relevant to your goal.
Clone with submodules, install with uv (or the pip alternative), and add the system GL libraries the README lists.
Run a documented inference or evaluation example with the demo data, then fine-tune on a small slice of your own dataset.
A pretrained vision-language-action model intended as a base for many robots and tasks, not a single-task policy.
N1.7 replaces the older Eagle backbone with Cosmos-Reason2-2B (Qwen3-VL), encoding images at native aspect ratio without padding.
State and action dimensions expand to 132 with an action horizon of 40, and the flow-matching DiT action head is updated for N1.7.
Fine-tune on multiple dataset paths at once with mixture weighting (ds_weights_alpha) for easier multi-source training.
Full-pipeline ONNX and TensorRT export, with hardened policy serving, rollout recording, evaluation and config validation.
Documented results across RoboCasa (including GR1 tabletop tasks), SimplerEnv and real G1 robot evaluation.
Get the source from the official GitHub repository for Isaac-GR00T.
git clone https://github.com/NVIDIA/Isaac-GR00T.git
cd Isaac-GR00TInstall git-lfs first (required for the demo data), then clone the repository with its submodules as the README directs.
sudo apt install git-lfs && git lfs install
git clone --recurse-submodules https://github.com/NVIDIA/Isaac-GR00TInstall in editable mode with uv (the README's primary path) after cloning; a pip alternative is documented there too.
uv pip install -e . --verbose # pip alternative: pip install -e . — see READMENVIDIA, through the NVIDIA/Isaac-GR00T repository, with the product page at developer.nvidia.com/isaac/gr00t.
Yes — the project is open source under the Apache-2.0 license as recorded for this listing. See the license terms for any conditions on redistribution.
The model and code are free under Apache-2.0 with no API key, but you need capable GPU hardware and your own robot demonstration data for meaningful fine-tuning.
Training from scratch needs far more data and compute for basic competence. GR00T starts from a pretrained generalist model you fine-tune — you still need quality demonstrations and GPUs, but the starting point already understands scenes and language.
Clone the repository (or use the documented installer), follow the install and configure steps on this page, and run the smallest real task first — run a documented inference or evaluation example with the demo data, then fine-tune on a small slice of your own dataset.
Isaac Lab — NVIDIA's unified framework for robot learning, with 30+ environments and multi-physics simulation.
Open-source 7B vision-language-action model and codebase for generalist robot manipulation.
Enterprise LLM app platform: GenAI workflows, RAG, agents, model management, evaluation, fine-tuning.
Open source ETL/ELT you deploy on your own servers or cloud.
Fast, best-practice evaluations for AI model outputs.
An open-source speech language model from Hume.
Другие проекты из категории «ИИ-агенты» с наибольшим числом звёзд в каталоге LibreAI — реалистичный список альтернатив, если Isaac-GR00T вам не подходит.
A personal AI that really does things — on any OS, with your data staying yours.
An agentic skills framework & software development methodology that works.
Agent harness optimization system with skills, memory and security for Claude Code, Codex and more
The agent that grows with you — a Python framework for adaptive AI agents.
DeepSeek's plugin-based agent harness where everything is a plugin.
Соседние проекты из категории «ИИ-агенты» в каталоге, ближайшие по числу звёзд к Isaac-GR00T — продолжайте изучать полку по обе стороны от него.
Isaac Lab — NVIDIA's unified framework for robot learning, with 30+ environments and multi-physics simulation.
Stream's open Vision Agents — build real-time voice and vision agents with any model or video provider.
FIRERPA / LAMDA — full-stack Android device control: remote desktop, automation, MITM, Frida and MCP.
All your agents and workspaces in one place — on every device you own.
Simple workflow patterns for building effective agents on MCP.
Osaurus — own your AI: a native macOS agent harness with any model, memory and fully offline execution.
Бесплатно и открыто — забирайте исходный код, ставьте звезду репозиторию и начинайте создавать.