datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
Wikidata_Vectors_0.2
Wikidata Entity Embeddings 0.2
Dataset Summary
Wikidata Entity Embeddings is a dataset of embedding vectors for Wikidata entities. Each vector represents a Wikidata item (Q...) or property (P...) based on textual information extracted from Wikidata.
The dataset is part of the Wikidata Embedding Project, an initiative led by Wikimedia Deutschland in collaboration with Jina AI and IBM DataStax. The project provides a publicly accessible Wikidata Vector Database to… See the full description on the dataset page: https://huggingface.co/datasets/philippesaade/Wikidata_Vectors_0.2.v32-vectorsopeniti-vectors
OpenITI Vector Database — maktabati.ai
🇬🇧 English
This dataset contains the fully vectorized OpenITI RELEASE 2025-1-9 collection of classical Islamic texts, prepared for semantic search (RAG).
Each entry represents a text chunk from one of 8,943 works in the OpenITI corpus, together with its embedding vector and complete metadata.
Statistics:
4,696,703 chunks
8,943 works (primary editions only, status=pri from OpenITI TSV)
approx. 470 Parquet files (approx.… See the full description on the dataset page: https://huggingface.co/datasets/Maktabati/openiti-vectors.vecforge-paper-corpus
Note (rebuild in progress): figures are being re-extracted with a fixed extractor (cleaner crops). The image-preview config (Parquet with an inline column + difficulty/type/score labels) returns after re-classification. The config (paper metadata + links) is live now.
VecForge Paper Corpus
A pristine, deduplicated collection of 59,732 top-venue AI/ML/CV/NLP/Robotics papers (2020-2024) with
every captioned figure and full paper text, for research on figure understanding… See the full description on the dataset page: https://huggingface.co/datasets/debajyotidasgupta/vecforge-paper-corpus.shamela-vectors
🇬🇧 English
The largest open-source vector database of the complete al-Maktaba al-Shamela (المكتبة الشاملة) Islamic text corpus, prepared for semantic search and Retrieval-Augmented Generation (RAG). Includes the full Quran text (6,236 verses, Hafs 'an 'Asim).
Statistics:
11,482,164 chunks from 8,589 classical Islamic books
6,236 Quran verses (one verse = one chunk, included in total)
40 categories covering the full breadth of Islamic scholarship
Period: 1st century AH to 15th… See the full description on the dataset page: https://huggingface.co/datasets/Maktabati/shamela-vectors.soda-vec-data-full_pmc_title_abstract
SODA-VEC Clean Dataset
This is a cleaned and filtered version of the SODA-VEC dataset, containing high-quality biomedical title-abstract pairs from PubMed Central (PMC) articles.
Dataset Overview
Total examples: 26,573,900
Training set: 26,473,900 examples (99.6%)
Validation set: 50,000 examples (0.2%)
Test set: 50,000 examples (0.2%)
Quality Filtering Applied
This dataset has been processed with the following quality filters:
Abstract Length… See the full description on the dataset page: https://huggingface.co/datasets/EMBO/soda-vec-data-full_pmc_title_abstract.hackernews-vector-search-datasetThe Hacker News dataset contains 28.74 million postings and their vector embeddings. The embeddings were generated using SentenceTransformers model all-MiniLM-L6-v2. The dimension of each embedding vector is 384.
Created by clickhouse more info: https://clickhouse.com/docs/getting-started/example-datasets/hackernews-vector-search-dataset
route_red_yellow_vector_subtasks_pi05
Route Red-Yellow Vector Subtasks for pi0.5
This is a LeRobot v2.1 transformation of DistantSky/route at commit aced1e96f6b8bf98ffaa0754407636e82442b084.
The dataset contains 212 real-robot episodes, 135699 frames, three cameras, and 14-dimensional actions at 100 Hz.
Conditioning
Only observation.images.video_overhead is modified. The left and right videos are byte-identical to the source dataset.
The overhead image receives one fixed selected-connector pose glyph… See the full description on the dataset page: https://huggingface.co/datasets/ajaysri/route_red_yellow_vector_subtasks_pi05.rss_vectorsamazon-reviewsshamela-vectors
🇬🇧 English
The largest open-source vector database of the complete al-Maktaba al-Shamela (المكتبة الشاملة) Islamic text corpus, prepared for semantic search and Retrieval-Augmented Generation (RAG). Includes the full Quran text (6,236 verses, Hafs 'an 'Asim).
Statistics:
11,482,164 chunks from 8,589 classical Islamic books
6,236 Quran verses (one verse = one chunk, included in total)
40 categories covering the full breadth of Islamic scholarship
Period: 1st century AH to 15th… See the full description on the dataset page: https://huggingface.co/datasets/Kandil7/shamela-vectors.pi07_cable_three_vector_v1
Three-holder cable routing with vector goals
Real-robot demonstrations for a vector-conditioned low-level policy: place three holders and route a cable through each holder. This is the validated LeRobot v3 dataset prepared for the first Pi0.7 four-camera-goal cable policy.
Property
Value
Source recordings
140
Subtask episodes
840
Frames
335,897
Sampling rate
100 Hz
Robot
ARX bimanual
Recorded state / action
14 / 14 dimensions
Video resolution
448 × 448… See the full description on the dataset page: https://huggingface.co/datasets/ajaysri/pi07_cable_three_vector_v1.details_Delta-Vector__Odin-9B
Dataset Card for Evaluation run of Delta-Vector/Odin-9B
Dataset automatically created during the evaluation run of model Delta-Vector/Odin-9B.
The dataset is composed of 136 configuration, each one coresponding to one of the evaluated task.
The dataset has been created from 1 run(s). Each run can be found as a specific split in each configuration, the split being named using the timestamp of the run.The "train" split is always pointing to the latest results.
An additional… See the full description on the dataset page: https://huggingface.co/datasets/OALL/details_Delta-Vector__Odin-9B.MVRBScreenshotRetrieval
MVRB Screenshot Retrieval
MTEB v2 multimodal retrieval dataset layout for the seven Screenshot Retrieval subsets from MVRB.
MobileGym-ConAct-Trajectories
MobileGym-ConAct-Trajectories
Dataset Viewer · MobileGym · MemGUI-Agent · Paper
Abstract
MobileGym-ConAct-Trajectories is a release of successful mobile GUI-agent rollouts collected in the MobileGym simulator. Each trajectory is selected from judge-verified rollouts using a deterministic per-task rule: retain the shortest structurally valid success, then break ties by source run and episode ID. The release preserves screenshots, the rendered prompt supplied to the… See the full description on the dataset page: https://huggingface.co/datasets/Ma-Vector/MobileGym-ConAct-Trajectories.newsdataio_vectorsx86-instruction-test-vectors
x86-64 Instruction Test Vectors
Ground truth behavior of individual x86-64 instructions, captured by executing every encoding on real hardware and recording the resulting register and flag state. This is measured silicon behavior, not a model and not an emulator, so it also reflects implementation specific results such as the values an instruction leaves in flags that the architecture documents as undefined.
How it was generated
Each test case is produced by the… See the full description on the dataset page: https://huggingface.co/datasets/BinPrey/x86-instruction-test-vectors.googlenews_vectorsamazon-reviews-10MHumaniBench
HumaniBench: A Human-Centric Benchmark for Large Multimodal Models Evaluation
**HumaniBench** is a benchmark for evaluating large multimodal models (LMMs) using real-world, human-centric criteria. It consists of 32,000+ image–question pairs across 7 tasks:
✅ Open/closed VQA
🌍 Multilingual QA
📌 Visual grounding
💬 Empathetic captioning
🧠 Robustness, reasoning, and ethics
Each example is annotated with GPT-4o drafts, then verified by experts to ensure quality and… See the full description on the dataset page: https://huggingface.co/datasets/vector-institute/HumaniBench.hotpotqaVectorEdits
VectorEdits: A Dataset and Benchmark for Instruction-Based Editing of Vector Graphics
NOTE: Currently only test set has generated labels, other sets will have them soon
Find the details in our paper: VectorEdits: A Dataset and Benchmark for Instruction-Based Editing of Vector Graphics
Github repository: JosefKuchar/vector-edits
We introduce a large-scale dataset for instruction-guided vector image editing, consisting of over 270,000 pairs of SVG images paired with natural language… See the full description on the dataset page: https://huggingface.co/datasets/mikronai/VectorEdits.LoCoMo
LoCoMo
MTEB v2 text retrieval dataset layout for LoCoMo. The candidates configs store the per-query retrieval pool for each memory retrieval subset.
nvidia-math-vectorizedwiki-image-vectorsVector embeddings of images on Wikipedia using google/siglip2-base-patch16-384
Here's an example of how to use this
# /// script
# requires-python = ">=3.11"
# dependencies = [
# "gradio",
# "torch",
# "transformers",
# "pyarrow",
# "numpy",
# "huggingface_hub",
# ]
# ///
"""
Semantic search over Wikipedia/Commons image embeddings (SigLIP 2).
Run locally:
VECTORS=/path/to/vectors.parquet uv run app.py
Or let it pull from the Hub:
uv run app.py
"""
import… See the full description on the dataset page: https://huggingface.co/datasets/derenrich/wiki-image-vectors.soda-vec-data-full_pmc_title_abstract_paired
SODA-VEC Paired Dataset for Negative Sampling
This is a paired version of the SODA-VEC dataset, specifically formatted for negative sampling training with MultipleNegativesRankingLoss.
Dataset Overview
Total examples: 26,573,900
Format: Paired (anchor-positive) for contrastive learning
Source: EMBO/soda-vec-data-full_pmc_title_abstract
Purpose: Training sentence transformers with negative sampling
Data Format
Each example contains:
anchor (string): The title… See the full description on the dataset page: https://huggingface.co/datasets/EMBO/soda-vec-data-full_pmc_title_abstract_paired.AIME_2024_DeepSeek_R1_0528_Temp_1.0_L_16384Responses of deepseek-ai/DeepSeek-R1-0528 for AIME 2024 (original dataset: Maxwell-Jia/AIME_2024).
Generation temperature is set to 1.0 and maximum token is set to 16384.
scandi-wiki-vector-store
Dataset Card for kardosdrur/scandi-wiki-vector-store
This dataset was created using the vicinity library, a lightweight nearest neighbors library with flexible backends.
It contains a vector space with 3655450 items.
Usage
You can load this dataset using the following code:
from vicinity import Vicinity
vicinity = Vicinity.load_from_hub("kardosdrur/scandi-wiki-vector-store")
After loading the dataset, you can use the vicinity.query method to find the nearest neighbors to… See the full description on the dataset page: https://huggingface.co/datasets/kardosdrur/scandi-wiki-vector-store.vectrix-art-e
Vectrix ART-E: Synthetic Email Agent Benchmark
A fully synthetic email corpus and task dataset for training and evaluating email search agents, built as a drop-in replacement for the Enron corpus used in OpenPipe's ART-E benchmark.
Key Result
A Qwen3.5-35B-A3B fine-tuned via GRPO on this synthetic dataset beats o3 on real Enron emails (86% vs 85%) — despite never seeing a single real email during training.
Dataset Contents
The dataset is available in two… See the full description on the dataset page: https://huggingface.co/datasets/TonicAI/vectrix-art-e.DUDE
DUDE
MTEB v2 multimodal retrieval dataset layout for DUDE.