datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
morpheus-real-world
Morpheus — Real-World Physics Videos
Real-world reference footage for Morpheus, a benchmark that tests whether
video generative models (Wan, CogVideo, LTX-Video, COSMOS-predict1/2,
Pyramid-Flow, Veo3, Kling-Turbo, ...) obey Newtonian mechanics. Object
trajectories are extracted via SAM2 tracking and tested against physical laws
(energy/momentum conservation, equations of motion) rather than pixel-matched
to a single "correct" video.
This repo contains the filmed real-world… See the full description on the dataset page: https://huggingface.co/datasets/physics-from-video/morpheus-real-world.meta13sphere_IRS_DCE_Topological_Dynamics__Boundary_Dissolution_Physics
Resonance Resonance / IRS-DCE
MASTER README (FULL EXTENDED VERSION)
If you need the other data or pdf check on [https://huggingface.co/datasets/meta13sphere/phaseShift_shell_result_pdf]
[2026-09-25 Update]
The Geometry of IRS-DCE Boundary Dissolution v1.0 is now available in Korean and English.
Resolution-Dependent Component Decomposition and Dynamic Rearrangement
This release connects the existing IRS-DCE and BBRCM research to mathematical analysis and recorded… See the full description on the dataset page: https://huggingface.co/datasets/meta13sphere/meta13sphere_IRS_DCE_Topological_Dynamics__Boundary_Dissolution_Physics.vidore_v3_physicsViDoRe V3 : Physics
This dataset, Physics, is a corpus of course slides on bachelor level physics lectures, intended for complex visual understanding tasks. It is one of the 10 corpora comprising the ViDoRe v3 Benchmark.
About ViDoRe v3
ViDoRe V3 is our latest benchmark for RAG evaluation on visually-rich documents from real-world applications. It features 10 datasets with, in total, 26,000 pages and 3099 queries, translated into 6 languages. Each query comes with human-verified… See the full description on the dataset page: https://huggingface.co/datasets/vidore/vidore_v3_physics.video-gen-physics-gallery
Paper material (assets/drive)
Two generated galleries plus a figure asset pack. All reproducible from the repo —
don't hand-edit them, re-run the generator.
gallery
what it shows
generator
real_videos/
the input side of the benchmark: real GT episodes, 5 diverse tasks per (embodiment x view-track x markovian/non-markovian) cell
scripts/make_real_video_gallery.py
method_comparison/
the output side: ONE episode rendered by every acceleration method, so a single dir is… See the full description on the dataset page: https://huggingface.co/datasets/doanh25032004/video-gen-physics-gallery.Physics-Informed-Deformable-Gaussian-Splattingvidore_v3_physics_mteb_format
Vidore3PhysicsRetrieval
An MTEB dataset
Massive Text Embedding Benchmark
Retrieve associated pages according to questions.
Task category
t2i
Domains
Academic
Reference
https://huggingface.co/blog/QuentinJG/introducing-vidore-v3
Source datasets:
vidore/vidore_v3_physics
How to evaluate on this task
You can evaluate an embedding model on this dataset using the following code:
import mteb
task = mteb.get_task("Vidore3PhysicsRetrieval")
evaluator =… See the full description on the dataset page: https://huggingface.co/datasets/vidore/vidore_v3_physics_mteb_format.PhysicsHandwritten-Physics-Notes-Dataset
English Handwritten Physics Notes Dataset
This dataset contains high-resolution images of handwritten physics notes written in English. The collection includes theoretical explanations, formulas, diagrams, derivations, and problem-solving steps. It is designed to support AI research in handwriting recognition, scientific OCR, and document understanding for physics and STEM education.
Contact
For queries or collaborations related to this dataset, contact:… See the full description on the dataset page: https://huggingface.co/datasets/HumynLabs/Handwritten-Physics-Notes-Dataset.WISER[CVPR 2026] WISER: Wider Search, Deeper Thinking, and Adaptive Fusion for Training-Free Zero-Shot Composed Image Retrieval
If you find this project useful, please give us a star 🌟.
🔥 News
[2026/2/23] Our paper has been accepted to CVPR 2026!
[2026/2/27] We release our paper in the arxiv.
[2026/2/27] We release the code.
🔎 Overview
Overview of the proposed WISER framework. (1) Wider Search. We leverage an editor to produce text and image… See the full description on the dataset page: https://huggingface.co/datasets/Physicsmile/WISER.physics-problems
Physics Problems Dataset
Dataset Description:
This dataset contains a collection of physics problems designed for educational and research purposes. Each problem includes a question, relevant equations, and, where applicable, numerical or symbolic solutions. The dataset covers topics such as mechanics, electromagnetism, thermodynamics, optics, and modern physics.
The dataset supports training and evaluation of models in:
Natural Language Processing (NLP) for physics problem-solving… See the full description on the dataset page: https://huggingface.co/datasets/HumynLabs/physics-problems.physics_big
The First Multimodal Chain of Thoughts Dataset for Physics Problems
Dataset Overview
This dataset consists of a diverse collection of physics problems, categorized into different domains:
1000 problems on Kinematics
600 problems on Electricity and Circuits
500 problems on Thermodynamics
Data Sources
All data has been extracted from open sources, ensuring a wide variety of problem types and difficulty levels.
Structure
Each problem is designed to… See the full description on the dataset page: https://huggingface.co/datasets/Vikhrmodels/physics_big.counterfactual-physics2-000-Physics-Chemistry-And-Biology-Questions-With-Image-Explanations-Dataset
2,000 Physics, Chemistry, and Biology Questions with Image Explanations Dataset
This collection contains 2,000 high-quality physics, chemistry, and biology questions, featuring a core modality of original images paired with text explanations. The data covers various formats, including multiple-choice, fill-in-the-blanks, experimental, and calculation questions. Each record provides the original question image, precise OCR-extracted text, and detailed step-by-step textual… See the full description on the dataset page: https://huggingface.co/datasets/Mobiusi/2-000-Physics-Chemistry-And-Biology-Questions-With-Image-Explanations-Dataset.PhysicsMind
PhysicsMind Video Dataset
This is the video generation (VG) component of the PhysicsMind benchmark, a unified physics-grounded benchmark that evaluates both reasoning and prediction under the same physical laws in both simulated and real-world settings.
About PhysicsMind
PhysicsMind is designed to evaluate whether video generation models can produce physically plausible motion trajectories that obey center-of-mass, torque, and inertial constraints. The benchmark focuses… See the full description on the dataset page: https://huggingface.co/datasets/PhysicsMindBench/PhysicsMind.sora-video-generation-physics-likert-scoring
Rapidata Video Generation Physics Dataset
If you get value from this dataset and would like to see more in the future, please consider liking it.
This dataset was collected in ~1 hour using the Rapidata Python API, accessible to anyone and ideal for large scale data annotation.
Overview
In this dataset, ~6000 human evaluators were asked to rate AI-generated videos based on if gravity and colisions make sense, without seeing the prompts used to generate them.… See the full description on the dataset page: https://huggingface.co/datasets/Rapidata/sora-video-generation-physics-likert-scoring.vidore3_physics_neomme_260m_li
vidore3_physics_neomme_260m_li
Multi-vector (late-interaction) embeddings of ViDoRe physics (vidore/physics), encoded with
Hcompany/NeoMME-260M-Retriever-ST-late at revision 023be2a8ab9d797f5aa76f5bf8b5dde78d819659.
Source data: Hugging Face dataset vidore/vidore_v3_physics at revision a0de276f515acc044b72cae8de53a44bb5a8f1f5, configs corpus / queries / qrels, split test, loaded with datasets. Document, query and qrel ids are the source's own ids,
unchanged.
Every document is… See the full description on the dataset page: https://huggingface.co/datasets/robro612/vidore3_physics_neomme_260m_li.physr1corp
PhysR1Corp — Closed-form RL Training Pool (2,268 records)
The closed-form (numeric / MCQ-gradable) RL training pool used by Physics-R1 (NeurIPS 2026 D&B Track submission). Carved out of PhysCorp-A (the audited 6,432-record corpus) by dropping open-ended questions, then decontaminated against MMMU-Pro Physics (−87 records) and against PhyX-mini + PhysUniBench-en at cos ≥ 0.85 (−78 additional records: 69 PhyX-mini near-duplicates and 9 PhysUniBench-en template duplicates). One… See the full description on the dataset page: https://huggingface.co/datasets/physics-r1-anonymous/physr1corp.physics-conflictsketchvlm-physics-ball-drop
SketchVLM: Physics Ball Drop Dataset
This dataset is part of the SketchVLM project, introduced in the paper: SketchVLM: Vision language models can annotate images to explain thoughts and guide users.
Project Page | GitHub | Interactive Demo
Description
SketchVLM is a training-free, model-agnostic framework that enables vision-language models (VLMs) to produce non-destructive, editable SVG overlays on input images to visually explain their reasoning.
The Physics Ball… See the full description on the dataset page: https://huggingface.co/datasets/loganbolton/sketchvlm-physics-ball-drop.testing-physics-I2VphysicsComputational-physicsEnergy-Physics-Pipeline-lane-state
Zer0pa/Energy-Physics-Pipeline-lane-state
Canonical Hugging Face custody surface for the Energy research lane (slot 08, portfolio insilico, backing repo Zer0pa/Energy-Physics-Pipeline).
This dataset is the durable off-laptop disaster-recovery target for every evidence artefact, every corridor-agent receipt, every product-page render, every research source-doc, and every FPO §A truth source produced for the lane. Together with the GitHub repo and the PyPI package, these three remotes… See the full description on the dataset page: https://huggingface.co/datasets/Zer0pa/Energy-Physics-Pipeline-lane-state.Physics-Images-DatasetDataset Description:
This Physics image dataset is part of a large-scale STEM image dataset containing 383,697 images, designed to support the development and training of advanced computer vision, educational AI, and multimodal learning systems.
Additionally, this dataset can be used in pipelines for Supervised Fine-Tuning (SFT) and Reinforcement Learning with Human Feedback (RLHF) workflows, improving model performance in scientific image understanding, visual reasoning, concept recognition… See the full description on the dataset page: https://huggingface.co/datasets/InfoBayAI/Physics-Images-Dataset.physics_geologyPhysics-Chemistry-Mathematics-Pro
Physics-Chemistry-Mathematics-Pro
Copyright © 2026 Vinay Umrethe umrethevinay@gmail.com.
This dataset is licensed under the Creative Commons Attribution 4.0 International License.
vietnamese_physics_highschool_multimodalphysics11_hardmultimodal_mcq_greek_physicsbrazilian-math-physics-qa-vision
Brazilian Math & Physics QA — Image Dependent
English | Português do Brasil
English
Summary
Brazilian Portuguese educational question-answer pairs whose problem statement or solution depends on one or more images.
Examples: 3,808
Referenced image URLs: 5,094 unique
Language: Brazilian Portuguese (pt-BR)
Schema
{"id":"vqa_...","subject":"matematica","category":"geometria","title":"...","messages":[{"role":"user","content":"...… See the full description on the dataset page: https://huggingface.co/datasets/artificialguybr/brazilian-math-physics-qa-vision.