Cosmos-Reason1：从物理常识到具身推理

Cosmos-Reason1: From Physical Common Sense To Embodied Reasoning

March 18, 2025

作者: NVIDIA, Alisson Azzolini, Hannah Brandon, Prithvijit Chattopadhyay, Huayu Chen, Jinju Chu, Yin Cui, Jenna Diamond, Yifan Ding, Francesco Ferroni, Rama Govindaraju, Jinwei Gu, Siddharth Gururani, Imad El Hanafi, Zekun Hao, Jacob Huffman, Jingyi Jin, Brendan Johnson, Rizwan Khan, George Kurian, Elena Lantz, Nayeon Lee, Zhaoshuo Li, Xuan Li, Tsung-Yi Lin, Yen-Chen Lin, Ming-Yu Liu, Andrew Mathau, Yun Ni, Lindsey Pavao, Wei Ping, David W. Romero, Misha Smelyanskiy, Shuran Song, Lyne Tchapmi, Andrew Z. Wang, Boxin Wang, Haoxiang Wang, Fangyin Wei, Jiashu Xu, Yao Xu, Xiaodong Yang, Zhuolin Yang, Xiaohui Zeng, Zhe Zhang

cs.AI

摘要

物理AI系统需要感知、理解并在物理世界中执行复杂动作。本文中，我们介绍了Cosmos-Reason1模型，该模型能够理解物理世界，并通过长链思维推理过程生成合适的具身决策（如下一步动作）。我们首先定义了物理AI推理的关键能力，重点关注物理常识与具身推理。为表示物理常识，我们采用了一种层次化本体，捕捉关于空间、时间和物理的基本知识。对于具身推理，我们依赖一个二维本体，该本体能够泛化不同的物理具身形式。基于这些能力，我们开发了两个多模态大语言模型：Cosmos-Reason1-8B和Cosmos-Reason1-56B。我们分四个阶段精心准备数据并训练模型：视觉预训练、通用监督微调（SFT）、物理AI SFT以及作为后训练的物理AI强化学习（RL）。为了评估模型，我们根据本体构建了全面的物理常识与具身推理基准。评估结果表明，物理AI SFT和强化学习带来了显著提升。为推动物理AI的发展，我们将在NVIDIA开放模型许可下，于https://github.com/nvidia-cosmos/cosmos-reason1 公开代码与预训练模型。

English

Physical AI systems need to perceive, understand, and perform complex actions in the physical world. In this paper, we present the Cosmos-Reason1 models that can understand the physical world and generate appropriate embodied decisions (e.g., next step action) in natural language through long chain-of-thought reasoning processes. We begin by defining key capabilities for Physical AI reasoning, with a focus on physical common sense and embodied reasoning. To represent physical common sense, we use a hierarchical ontology that captures fundamental knowledge about space, time, and physics. For embodied reasoning, we rely on a two-dimensional ontology that generalizes across different physical embodiments. Building on these capabilities, we develop two multimodal large language models, Cosmos-Reason1-8B and Cosmos-Reason1-56B. We curate data and train our models in four stages: vision pre-training, general supervised fine-tuning (SFT), Physical AI SFT, and Physical AI reinforcement learning (RL) as the post-training. To evaluate our models, we build comprehensive benchmarks for physical common sense and embodied reasoning according to our ontologies. Evaluation results show that Physical AI SFT and reinforcement learning bring significant improvements. To facilitate the development of Physical AI, we will make our code and pre-trained models available under the NVIDIA Open Model License at https://github.com/nvidia-cosmos/cosmos-reason1.