datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
3DCode
Project page
Paper
Code
3dcodebench.com
arXiv:2606.01057
gaoypeng/3dcodebench
News
[06/01/2026] Paper released on arXiv: 3DCodeBench: Benchmarking Agentic Procedural 3D Modeling Via Code.
Note. This is an open-source reproduction of 3DCodeBench.
⚠️ Under final check. The 3DCodeData/ code is still undergoing final
quality review and may contain occasional issues (non-executable scripts, mismatched
captions/renders, or imperfect geometry). If you run… See the full description on the dataset page: https://huggingface.co/datasets/YipengGao/3DCode.LIBERO-datasets
LIBERO Datasets
This is a repo that stores the LIBERO datasets. The structure of the dataset can be found below:
libero_object/
libero_spatial/
libero_goal/
libero_90/
libero_10/
Demonstrations of each task is stored in a hdf5 file. Please refer to download script from the official LIBERO repo for more details.
TexVerse
TexVerse: A Universe of 3D Objects with High-Resolution Textures
Yibo Zhang1,2, Li Zhang1,3, Rui Ma2 *, Nan Cao1,4
1Shanghai Innovation Institute
2Jilin University
3Fudan University
4Tongji University
* Corresponding Author
TexVerse is a large-scale 3D dataset featuring high-resolution textures. Its key characteristics include:
Scale & Source: TexVerse dataset has 858,669 unique 3D models curated from… See the full description on the dataset page: https://huggingface.co/datasets/YiboZhang2001/TexVerse.TexVerse-1Kcode-world-model-project-page-videos
Code World Model Project Page Videos
Public research-demo video assets used by the Code World Model project page.
The gallery/ directory contains aligned RGB and proxy videos for interactive comparison.
MASIV
MASIV Multi-Sequence Dataset
Toward Material-Agnostic System Identification from Videos
ICCV 2025
Yizhou Zhao1, Haoyu Chen1, Chunjiang Liu1, Zhenyang Li2, Charles Herrmann3, Junhwa Hur3, Yinxiao Li3, Ming‑Hsuan Yang4, Bhiksha Raj1, Min Xu1*
1Carnegie Mellon University 2University of Alabama at Birmingham 3Google 4UC Merced
Introduction
The MASIV Multi-Sequence Dataset is a synthetic dataset generated by Genesis to evaluate the generalization of data-driven… See the full description on the dataset page: https://huggingface.co/datasets/yizhouz/MASIV.PanoCity
PanoCity Dataset
A Large-Scale Aerial Panoramic Dataset for 3D Scene Understanding
📊 Dataset Statistics
Attribute
Value
Total Size
1.4 TB
Cities
Beijing (20 blocks), Jinan (76 blocks), Ningbo (41 blocks)
Panoramic RGB Images
119,537 (2048×4096)
Panoramic Depth Maps
119,537 (2048×4096)
Total Images
239,074
Image Format
PNG
📂 Data Structure
PanoCity/
├── splits_config.json # Official train/test splits
├──… See the full description on the dataset page: https://huggingface.co/datasets/YijingGuo/PanoCity.testing-imagesVOST-TAS
[NeurIPS 2025] Tracking and Understanding Object Transformations
If you like our project, please give us a star ⭐ on GitHub for the latest update.
💡 Description
Dataset Visualizations: GitHub
Paper: arXiv:2511.04678
Project Page: tubelet-graph.github.io
Project Repository: GitHub
Point of Contact: Yihong Sun
📊 Dataset Overview
VOST-TAS (TrackAnyState) is an extended version of the VOST validation set with explicit transformation annotations for tracking and… See the full description on the dataset page: https://huggingface.co/datasets/yihongs/VOST-TAS.cspred-dataEgoSAT
EgoSAT
Dataset Description
EgoSAT is an ECCV 2026 benchmark for egocentric streaming interaction understanding:
EgoSAT: A Comprehensive Benchmark of Egocentric Streaming Interaction Understanding
The benchmark evaluates vision-language models under streaming constraints, including retrospective reasoning, present-time interaction understanding, and prospective reasoning over future actions or state transitions.
This HuggingFace dataset contains the main EgoSAT… See the full description on the dataset page: https://huggingface.co/datasets/YijiaLeithu/EgoSAT.MMPreferenceVre10k_pixelsplatActivityNet
Description
Dataset V1-2
v1-2_train.tar.gz and v1-2_val.tar.gz
Data (train and val set) associated with ActivityNet release 1.2
v1-2_test.tar.gz
Data (test set only) associated with ActivityNet release 1.2
Dataset V1-3
v1-3_train_val.tar.gz
Additional videos (train val set) collected for ActivityNet release 1.3
v1-3 is an extension of v1-2, so you also need to download v1-2 data and merge to v1.3
v1-3_test.tar.gz
Additional videos (test set only)… See the full description on the dataset page: https://huggingface.co/datasets/YimuWang/ActivityNet.financial-analyst-data-full
financial-analyst-data-full
A-share historical price + valuation data packaged for financial-analyst —
the 14-agent single-stock deep-dive research workstation.
Published: 2026-05-24
Preset: full — 全 A 股完整包 (含历史退市股). 量化研究员 / 重度用户. lite 全 + TDX 历年财报原始 zip (用户跑 import_tdx_financial.py 解) + F10 原始文本 (公司大事/龙虎榜/主力追踪/最新提示 .txt).
Size: ~14.1 GB
What's included
5450 stocks daily OHLCV + 7 valuation fields (PE/PB/PS/DV/MV/CIRC_MV/turnover_rate)
Date range (daily): 1990-12-19… See the full description on the dataset page: https://huggingface.co/datasets/yifishbossman/financial-analyst-data-full.20000-yingshi-pdfMultiHopRAG
Dataset Card for Dataset Name
A Dataset for Evaluating Retrieval-Augmented Generation Across Documents
Dataset Description
MultiHop-RAG: a QA dataset to evaluate retrieval and reasoning across documents with metadata in the RAG pipelines. It contains 2556 queries, with evidence for each query distributed across 2 to 4 documents. The queries also involve document metadata, reflecting complex scenarios commonly found in real-world RAG applications.
Dataset Sources… See the full description on the dataset page: https://huggingface.co/datasets/yixuantt/MultiHopRAG.fineweb-edu-score-2-minhashPLMCaliperinteractive-world-sim-datacode-world-model-inference-examples-40
Inference examples
This directory contains 40 numbered, independent inference examples.
Every example uses only its public number; source case names and internal paths
are intentionally omitted.
Each numbered directory contains:
first_frame.png: exact 1536x864 generated RGB first frame used by inference.
prompts/*.txt: the exact rolling long-inference prompts used for the result.
condition/*.npz: ordered lossless condition chunks.
metadata.json: frame count, FPS, prompt windows… See the full description on the dataset page: https://huggingface.co/datasets/NTU-yiwen/code-world-model-inference-examples-40.ps4mas-final-test-rollouts-0813
PS4MAS Final Test Rollouts (0813)
Source split: ps4mas-0521-splits final_test_scenarios.jsonl
Each traces/<model>/<model>.jsonl contains the agent-tool-loop output for 200 final_test scenarios × 4 topologies. Most baseline/oracle files are raw traces. GiGPO 0805-r2 step20/40/60/80 evals include OSS-120B scores and summary.json.
Files
Model
Rows
Path
best_rl_gigpo_debate_step40
800
traces/best_rl_gigpo_debate_step40/best_rl_gigpo_debate_step40.jsonl… See the full description on the dataset page: https://huggingface.co/datasets/yinita/ps4mas-final-test-rollouts-0813.LLaSO-Align
LLaSO: A Foundational Framework for Reproducible Research in Large Language and Speech Model
This repository contains the datasets (LLaSO-Align, LLaSO-Instruct, LLaSO-Eval) from the paper LLaSO: A Foundational Framework for Reproducible Research in Large Language and Speech Model.
Fully open corpus + benchmark + reference model for compositional speech-language understanding.
Paper • Code
TL;DR. 25.5M training samples, 20 tasks, 3 modality configurations; 15,044-sample stratified… See the full description on the dataset page: https://huggingface.co/datasets/YirongSun/LLaSO-Align.robomme_preprocessed_data
RoboMME Training Data (Pickle Format)
Arxiv Paper | HF Paper | Website | Benchmark Code | Policy Learning Code
This repo contains preprocessed pickle files for RoboMME training data and npy files for cached image tokens. We use this dataset in our MME-VLA experiments.
.
├── data # zipped pickle files
├── features # zipped precompute siglip embeddings
├── meta # statistics for robomme
├── memer # VLM subgoal training data for MemER (only used for symbolic… See the full description on the dataset page: https://huggingface.co/datasets/Yinpei/robomme_preprocessed_data.robomme_data_h5
RoboMME Training Data (H5 format)
Arxiv Paper | HF Paper | Website | Benchmark Code | Policy Learning Code
H5 format of RoboMME training data. There are a total of 16 tasks, each with 100 episodes.
.
├── README.md
├── record_dataset_BinFill.h5.tar.xz
├── record_dataset_ButtonUnmask.h5.tar.xz
├── record_dataset_ButtonUnmaskSwap.h5.tar.xz
├── record_dataset_InsertPeg.h5.tar.xz
├── record_dataset_MoveCube.h5.tar.xz
├── record_dataset_PatternLock.h5.tar.xz
├──… See the full description on the dataset page: https://huggingface.co/datasets/Yinpei/robomme_data_h5.Skydome_HDRIDownload the waymo_skydome folder and put it in data.
semantic_patch_cache
HeatTok Semantic Patch Cache
Precomputed .pt caches for HeatTok. Use with HEATTOK_SEMANTIC_CACHE_DIR=/path/to/cache.
Filename pattern: {image_hash}_g1_s28.pt or {image_hash}_g1_gd1_s28.pt
semantic_patch_cache_vrsbench
Dataset: VRSBench (512×512)
Caches do not store precomputed global tokens or patch orientations.
Gaussian parameters and patch metadata are stored.
No need to regenerate .pt files — HeatTok computes global tokens and orientations online at load… See the full description on the dataset page: https://huggingface.co/datasets/Yingying11/semantic_patch_cache.RefRef_additionalRefRef: A Synthetic Dataset and Benchmark for Reconstructing Refractive and Reflective Objects
Yue Yin ·
Enze Tao ·
Weijian Deng ·
Dylan Campbell
About
This repository provides additional data for the RefRef dataset.
Citation
@misc{yin2025refrefsyntheticdatasetbenchmark,
title={RefRef: A Synthetic Dataset and Benchmark for Reconstructing Refractive and Reflective Objects},
author={Yue Yin and Enze Tao and… See the full description on the dataset page: https://huggingface.co/datasets/yinyue27/RefRef_additional.financial-analyst-data-lite
financial-analyst-data-lite
EN: A-share historical OHLCV + valuation + financials + TDX F10 events, packaged in Qlib binary + Parquet formats. Companion dataset for financial-analyst — a 14-agent single-stock deep-dive research workstation.
中文: A 股历史行情 + 估值 + 财报 + TDX F10 事件数据集, Qlib 二进制 + Parquet 双格式打包. 配套 financial-analyst — 14 Agent 个股深度研究工作站使用.
Published / 发布: 2026-05-24 · Size / 体量: ~2.72 GB · License: Apache 2.0
📊 Three Preset Tiers / 三档预设
Pick the tier that… See the full description on the dataset page: https://huggingface.co/datasets/yifishbossman/financial-analyst-data-lite.internvid-tg
InternVid-TG
Paper | Code
We automate the annotation of InternVid-FLT video data, transitioning from the original video-text alignment data to temporal grounding data, named InternVid-Temporal-Grounding (InternVid-TG). Currently, we release a subset version of InternVid-TG, which includes 89,440 videos and contains 608k event instances. We will gradually release the full version later. For details about the annotation process of InternVid-TG, you can refer to our paper, Distime.… See the full description on the dataset page: https://huggingface.co/datasets/yingsen/internvid-tg.