Dataset · Hugging Face

XDUImageLab/SandThink

Name: SandThink
Creator: XDUImageLab
License: mit
Keywords: task_categories:robotics, task_categories:visual-question-answering, task_categories:text-generation, language:en, language:zh, license:mit, size_categories:n<1K, format:imagefolder, modality:image, library:datasets, library:mlcroissant, region:us, embodied-ai, vla, chain-of-thought

SandThink Dataset (v1.0) SandThink 是一个专为具身智能 (Embodied AI) 任务设计的大规模指令微调与偏好对齐数据集。该数据集通过结构化的 Chain-of-Thought (CoT) 推理过程，显著提升了 Vision-Language-Action (VLA) 模型在复杂环境下的任务拆解、路径规划和动作执行能力。 📊 数据集概览 (Dataset Summary) 本数据集包含三个核心组件，总计约 37,000 条高质量数据：文件名类型规模说明 Mars_CoT.jsonl SFT / CoT ~17,000 条针对特定复杂环境（如类地/火星沙盘）的高级推理数据。 iter_DPO_data.jsonl DPO / RLHF ~16,000 条包含 Chosen 和 Rejected 的偏好对，用于模型对齐与优化。 train_CoT.jsonl SFT / CoT ~4,000 条基础任务的思维链训练数据，涵盖避障、导航等指令。… See the full description on the dataset page: https://huggingface.co/datasets/XDUImageLab/SandThink.

View on Hugging FaceReport an issue

Author / organization	XDUImageLab
License (as declared)	mit
Downloads	338 (as reported by Hugging Face)
Likes	1
Created	4 March 2026
Last updated at source	5 March 2026

embodied-aivlachain-of-thoughtdpo

Embed this entry

<iframe src="https://silytics.si/embed/item/0190708f-37d6-4f1f-86aa-45d224a7b313" width="420" height="190" style="border:0" loading="lazy" title="Silytics entry"></iframe>

Free, no tracking. More about feeds and embeds.

More from XDUImageLab

XDUImageLab/SandGODataset

Silytics shows public metadata from Hugging Face and links to the original. Information may be incomplete or out of date. Last refreshed 16 October 2026. Licenses and terms of the original apply to the work itself. Not financial, legal or professional advice.