Taskone Raw Groot 2nd

KHandsome/Taskone-raw-groot-2nd is an open-source foundation model designed for robotics applications, with a particular focus on generalized humanoid robot reasoning and skill execution. The model, referred to as GR00T N1.7, is described as a cross-embodiment foundation model originating from NVIDIA, intended to enable a range of reasoning and skill capabilities in humanoid robots. It is built on a Cosmos-Reason2/Qwen3-VL backbone and incorporates a flow-matching action transformer, which allows it to predict actions based on combined inputs from vision, language, and proprioception. The model's policy can be run on physical robots, and it also supports training custom policies. The evidence notes that the model has been trained and published using the LeRobot framework, and users are directed to resources such as Google Colab and Kaggle notebooks for practical usage. Instructions are available for integrating and operating the model with LeRobot, suggesting that the intended audience includes robotics practitioners and researchers who wish to deploy or further train the model for their own robots. KHandsome/Taskone-raw-groot-2nd is distributed under the Apache 2.0 license, which allows for open use and modification. The model is made accessible via the Hugging Face platform, where users can find additional documentation, guides, and model files. The tool is positioned within the class of open foundation models for robotics and is designed to facilitate generalized reasoning and skill learning in humanoid robots by leveraging multi-modal sensory inputs. No specific details are provided in the evidence about supported robot hardware, performance benchmarks, or detailed application scenarios beyond the general focus on humanoid robotics and generalized reasoning. The model is available for download and integration through the Hugging Face Hub, and users are encouraged to consult the LeRobot groot guide for further instructions on training and deployment.