datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
huatuo_knowledge_graph_qa
Dataset Card for Huatuo_knowledge_graph_qa
Dataset Summary
We built this QA dataset based on the medical knowledge map, with a total of 798,444 pieces of data, in which the questions are constructed by means of templates, and the answers are the contents of the entries in the knowledge map.
Dataset Creation
Source Data… See the full description on the dataset page: https://huggingface.co/datasets/FreedomIntelligence/huatuo_knowledge_graph_qa.huatuo_knowledge_graph_qa
Dataset Card for Huatuo_knowledge_graph_qa
Dataset Summary
We built this QA dataset based on the medical knowledge map, with a total of 798,444 pieces of data, in which the questions are constructed by means of templates, and the answers are the contents of the entries in the knowledge map.
Dataset Creation
Source Data… See the full description on the dataset page: https://huggingface.co/datasets/wuwu616/huatuo_knowledge_graph_qa.GraphInstruct-Testomnimcp_graphrag_knowledge_teaser
🔬 INSPECT THE DEEPSEEK-R1 REASONING CHAIN LIVE:
Zero hallucinations. Null syntax errors. 100% AST compiler validated.🌐 Live Interactive Reasoning & Code Inspector: https://emgena.com/trainingslager🎁 Claim your Free Starter Kit (Code: STARTER100): https://emgena.com/trainingslager🏷️ Launch Discount: Get 20 € OFF any 500-incident production suite with code LAUNCH20!
📜 Enterprise Compliance: EU AI Act Articles 50 & 53 certified • 100% DSGVO / GDPR clean • Commercial EULA… See the full description on the dataset page: https://huggingface.co/datasets/emgena/omnimcp_graphrag_knowledge_teaser.qualora-workforce-skills-graph
Qualora Workforce Skills Graph (Representative Sample)
Rights-clean, provenance-tracked vocational learning data, rebuilt from roughly $2B of U.S. Department of Labor funded open courseware into a labeled skills graph: cleaned courses and lessons, Bloom-tagged assessment items with answer rationales and learning objectives, and a content-grounded course to skill to career graph with salary context. Built for post-training and evaluation, not pretraining bulk.
This repository is… See the full description on the dataset page: https://huggingface.co/datasets/qualora-data-labs/qualora-workforce-skills-graph.fin-cfa-graphgen
Fin-CFA-GraphGen: 785K Knowledge-Guided Financial QA Examples
Fin-CFA-GraphGen is a large-scale English dataset for financial instruction tuning, financial question answering, and domain-specific language-model post-training. It contains 785,149 synthetic question–answer examples generated from CFA curriculum and exam-preparation books with the GraphGen knowledge-driven data-generation method.
The dataset and its role in the post-training pipeline are described in Data-Centric… See the full description on the dataset page: https://huggingface.co/datasets/whoisjiji/fin-cfa-graphgen.islamic-corpus-graph
QuranLab — Qur'an & Hadith Structured Corpus and Knowledge Graph
A unified, verse- and ḥadīth-aligned structured corpus for the Qur'an and the canonical Sunnah,
assembled by volunteers under the QuranLab effort. It links Qur'anic verses, multilingual
translations, classical tafsīr, word-level morphology, and ḥadīth text with normalized authenticity
grades into one consistent graph, alongside retrieval passages, grounded question–answer pairs and a
held-out evaluation set. Every… See the full description on the dataset page: https://huggingface.co/datasets/quranlab/islamic-corpus-graph.omnimcp_graphrag_neo4j_cypher_teaser
🔬 INSPECT THE DEEPSEEK-R1 REASONING CHAIN LIVE:
Zero hallucinations. Null syntax errors. 100% AST compiler validated.🌐 Live Interactive Reasoning & Code Inspector: https://emgena.com/trainingslager🎁 Claim your Free Starter Kit (Code: STARTER100): https://emgena.com/trainingslager🏷️ Launch Discount: Get 20 € OFF any 500-incident production suite with code LAUNCH20!
📜 Enterprise Compliance: EU AI Act Articles 50 & 53 certified • 100% DSGVO / GDPR clean • Commercial EULA… See the full description on the dataset page: https://huggingface.co/datasets/emgena/omnimcp_graphrag_neo4j_cypher_teaser.Graph-R1-RFT-COT-30K
Dataset Card: Graph-CoT-30k
Dataset Details
Dataset Name: Graph-CoT-30k
Dataset Creator: HKUST-DSAIL
Dataset Version: 1.0
Release Date: August 2025
Description
Graph-CoT-30k is a large-scale, high-quality instruction tuning dataset designed to enhance the reasoning capabilities of large language models (LLMs) on complex graph-theoretic problems. It contains 30,000 question-answer (QA) pairs, each featuring ultra-long chain-of-thought (CoT) reasoning… See the full description on the dataset page: https://huggingface.co/datasets/HKUST-DSAIL/Graph-R1-RFT-COT-30K.omnimcp_graphrag_grounded_answer_teaser
🔬 INSPECT THE DEEPSEEK-R1 REASONING CHAIN LIVE:
Zero hallucinations. Null syntax errors. 100% AST compiler validated.🌐 Live Interactive Reasoning & Code Inspector: https://emgena.com/trainingslager🎁 Claim your Free Starter Kit (Code: STARTER100): https://emgena.com/trainingslager🏷️ Launch Discount: Get 20 € OFF any 500-incident production suite with code LAUNCH20!
📜 Enterprise Compliance: EU AI Act Articles 50 & 53 certified • 100% DSGVO / GDPR clean • Commercial EULA… See the full description on the dataset page: https://huggingface.co/datasets/emgena/omnimcp_graphrag_grounded_answer_teaser.CTNSG-Graph-Curriculum
CTNSG Graph Curriculum Dataset
This dataset contains preprocessed graphs from WebNLG (v3.0), ATOMIC, and Spider.
It is explicitly designed for the Canonical Tractable Neuro-Symbolic Generation (CTNSG) framework.
Preprocessing
All raw data has been parsed into continuous node and edge embeddings using sentence-transformers/all-MiniLM-L6-v2.
Crucially, the graphs have been mathematically canonicalized using the Reverse Cuthill-McKee (RCM) algorithm.
This minimizes… See the full description on the dataset page: https://huggingface.co/datasets/Borisz42/CTNSG-Graph-Curriculum.qualora-workforce-skills-graph
Qualora Workforce Skills Graph (Representative Sample)
Rights-clean, provenance-tracked vocational learning data, rebuilt from roughly $2B of U.S. Department of Labor funded open courseware into a labeled skills graph: cleaned courses and lessons, Bloom-tagged assessment items with answer rationales and learning objectives, and a content-grounded course to skill to career graph with salary context. Built for post-training and evaluation, not pretraining bulk.
This repository is… See the full description on the dataset page: https://huggingface.co/datasets/VegasIGN/qualora-workforce-skills-graph.Graph-R1-dataset-complete
Graph-R1 Complete Dataset
This dataset contains the complete Graph-R1 graph reasoning dataset with all difficulty levels (1-5).
Files
train_graph_all_levels.parquet: Combined training data from all levels (with level column)
test_graph_and_math_all_levels.parquet: Combined test data from all levels (with level column)
test_graph_mixedsize_cleaned.parquet: Mixed size test data (with level='mixed')
Usage
import pandas as pd
# Load combined data… See the full description on the dataset page: https://huggingface.co/datasets/HKUST-DSAIL/Graph-R1-dataset-complete.omnimcp_graphrag_hybrid_rrf_rerank_teaser
🔬 INSPECT THE DEEPSEEK-R1 REASONING CHAIN LIVE:
Zero hallucinations. Null syntax errors. 100% AST compiler validated.🌐 Live Interactive Reasoning & Code Inspector: https://emgena.com/trainingslager🎁 Claim your Free Starter Kit (Code: STARTER100): https://emgena.com/trainingslager🏷️ Launch Discount: Get 20 € OFF any 500-incident production suite with code LAUNCH20!
📜 Enterprise Compliance: EU AI Act Articles 50 & 53 certified • 100% DSGVO / GDPR clean • Commercial EULA… See the full description on the dataset page: https://huggingface.co/datasets/emgena/omnimcp_graphrag_hybrid_rrf_rerank_teaser.omnimcp_graphrag_triplet_extractor_teaser
🔬 INSPECT THE DEEPSEEK-R1 REASONING CHAIN LIVE:
Zero hallucinations. Null syntax errors. 100% AST compiler validated.🌐 Live Interactive Reasoning & Code Inspector: https://emgena.com/trainingslager🎁 Claim your Free Starter Kit (Code: STARTER100): https://emgena.com/trainingslager🏷️ Launch Discount: Get 20 € OFF any 500-incident production suite with code LAUNCH20!
📜 Enterprise Compliance: EU AI Act Articles 50 & 53 certified • 100% DSGVO / GDPR clean • Commercial EULA… See the full description on the dataset page: https://huggingface.co/datasets/emgena/omnimcp_graphrag_triplet_extractor_teaser.DesignBench
Dataset Card for DesignBench
Dataset Summary
DesignBench is a multi-framework, multi-task benchmark for evaluating MLLM-based front-end engineering. The paper targets limitations of prior UI code generation benchmarks by covering React, Vue, Angular, and vanilla HTML/CSS, and by evaluating generation, edit, and repair workflows. The full benchmark contains 900 webpage samples spanning multiple topics, edit types, and issue categories.
Supported Tasks… See the full description on the dataset page: https://huggingface.co/datasets/creative-graphic-design/DesignBench.praxis-chess-evidence-graphs
Praxis Chess Evidence Graphs
Built for, and used to
train and test,
praxis-chess-reasoner-qwen3.5-2b-lora
and its 4B sibling.
Chess mistakes from public Lichess data, each with an evidence graph of
engine-computed facts about the move, and a verified explanation: a typed
reasoning chain that a program can check claim by claim, the labels (what
happened, and why) and two or three sentences of prose.
The dataset exists to test one question: does a small model explain mistakes… See the full description on the dataset page: https://huggingface.co/datasets/praxis-chess/praxis-chess-evidence-graphs.GraphSilo-TestGraphOmni
📊 GraphOmni Dataset
Project Page | Paper | GitHub
GraphOmni is a comprehensive benchmark designed to evaluate the reasoning capabilities of Large Language Models (LLMs) on graph-theoretic tasks articulated in natural language. It encompasses diverse graph types, serialization formats, and prompting schemes, providing a robust foundation for advancing research in LLM-based graph reasoning.
🧠 Key Features
Supports six graph algorithm tasks:Connectivity, Bfsorder… See the full description on the dataset page: https://huggingface.co/datasets/G-A-I/GraphOmni.GraphInstruct-RFT-72KZ3-Verified-Reasoning-Graphs
Z3-Verified Constraint Reasoning Dataset
5k Baseline · Production-Ready · Zero Label Noise
The Problem This Solves
Most synthetic reasoning datasets only show the "happy path". Real reasoning requires knowing when to backtrack.
Open-source LLMs hallucinate on constraint satisfaction problems because they are trained on fluent-sounding but logically inconsistent traces. This dataset is different:
❌ No LLM-generated reasoning — zero hallucinations, zero label noise
✅… See the full description on the dataset page: https://huggingface.co/datasets/nagygabor/Z3-Verified-Reasoning-Graphs.kg-triplet-graphrag
kg-triplet-graphrag
Open-text passages annotated with {{entities, relationships}} in the
Microsoft GraphRAG knowledge-model format.
Part of kg-triplet-sft: https://github.com/Alex-tangt/kg-triplet-sft
Size: 3,349 labeled passages — train 2,575 / validation 75 / test 699.
(The split file lists 700 test ids; one, wikipedia-01183, had no teacher labels and is omitted.)
Domain / language: English; Wikipedia 60% + arXiv 40%; sentence-boundary chunks (100–300 words).
Labels: produced… See the full description on the dataset page: https://huggingface.co/datasets/Alextgt/kg-triplet-graphrag.graphinfer
GraphInfer
A benchmark for evaluating an LLM's ability to infer over a graph — to produce an answer that
jointly leverages a node's attributes, its neighbours' attributes, and the edges connecting them.
GraphInfer probes this capability along two axes — Description (what is a region of the graph?)
and Comparison (how do regions of the graph differ?) — over five tasks and six structurally
distinct real-world graphs.
Dataset Summary
GraphInfer contains 42,000… See the full description on the dataset page: https://huggingface.co/datasets/graphinfer/graphinfer.Graph-R1-dataset-level-2
Graph-R1 Dataset Level 2
This dataset contains graph reasoning problems at difficulty level 2.
Files
train_graph_level_2.parquet: Training data for level 2
test_graph_and_math_level_2.parquet: Test data for level 2
Usage
import pandas as pd
# Load training data
train_df = pd.read_parquet('train_graph_level_2.parquet')
# Load test data
test_df = pd.read_parquet('test_graph_and_math_level_2.parquet')
Citation
If you use this… See the full description on the dataset page: https://huggingface.co/datasets/HKUST-DSAIL/Graph-R1-dataset-level-2.graph-reasoning-messages-11K
Graph Reasoning (Messages)
A collection of chat messages designed for training and evaluating graph-native / structured reasoning behaviors in language models. Each dataset item is a conversation represented as an ordered list of {role, content} messages (OpenAI-style chat format).
What’s inside
Each example is a JSON-like list of messages, e.g.
[
{ "role": "user", "content": "..." },
{ "role": "assistant", "content": "..." }
]
Assistant responses include explicit… See the full description on the dataset page: https://huggingface.co/datasets/lamm-mit/graph-reasoning-messages-11K.Graph-R1-dataset-level-3
Graph-R1 Dataset Level 3
This dataset contains graph reasoning problems at difficulty level 3.
Files
train_graph_level_3.parquet: Training data for level 3
test_graph_and_math_level_3.parquet: Test data for level 3
Usage
import pandas as pd
# Load training data
train_df = pd.read_parquet('train_graph_level_3.parquet')
# Load test data
test_df = pd.read_parquet('test_graph_and_math_level_3.parquet')
Citation
If you use this… See the full description on the dataset page: https://huggingface.co/datasets/HKUST-DSAIL/Graph-R1-dataset-level-3.Cybersecurity_RAG_Knowledge_Graph-25-Topics-75-Articles-200-ChunksCybersecurity RAG Knowledge Graph (25 Topics, 75 Articles, 200 Chunks)
Preview dataset — full commercial package available at:https://automatekc.gumroad.com/l/cybersecurity-rag-graph
Overview
This is a structured, synthetic, commercially‑safe cybersecurity knowledge graph designed for RAG systems, AI copilots, fine‑tuning, and domain‑specific retrieval.
This repo contains a preview only.
The full dataset (25 topics, 75 articles, ~200 chunks, graph metadata, and structured JSON files) is… See the full description on the dataset page: https://huggingface.co/datasets/invinciblejha01/Cybersecurity_RAG_Knowledge_Graph-25-Topics-75-Articles-200-Chunks.Graph-R1-dataset-level-1
Graph-R1 Dataset Level 1
This dataset contains graph reasoning problems at difficulty level 1.
Files
train_graph_level_1.parquet: Training data for level 1
test_graph_and_math_level_1.parquet: Test data for level 1
Usage
import pandas as pd
# Load training data
train_df = pd.read_parquet('train_graph_level_1.parquet')
# Load test data
test_df = pd.read_parquet('test_graph_and_math_level_1.parquet')
Citation
If you use this… See the full description on the dataset page: https://huggingface.co/datasets/HKUST-DSAIL/Graph-R1-dataset-level-1.g4-graphsuite-synth-v1
G4GraphSuite v1
Fully synthetic, seed-reproducible suite for code-graph reasoning research
(paper track: Tasks & Benchmarks / Graph Reasoning). 240 procedural Python
repos (50 of them 150–350-node mega graphs), every bug-localization and
code-generation sample is verified by execution (fail-to-pass guarantee:
the gold repo passes all tests, the buggy repo fails exactly the target test).
Subsets
file
rows
task
verified
graph_qa.jsonl
1138
reachability… See the full description on the dataset page: https://huggingface.co/datasets/Nabidnur/g4-graphsuite-synth-v1.GraphTraceBench
GraphTraceBench
GraphTraceBench is a synthetic benchmark suite for controlled evaluation of long-range graph reasoning. It is designed to test whether graph-native and multimodal models can recover latent computation traces and execute multi-step arithmetic operations over directed graphs.
The benchmark contains graph-execution tasks where the target answer is obtained by following a directed reasoning trace and applying arithmetic operators along the trace.
Tasks… See the full description on the dataset page: https://huggingface.co/datasets/GraphTraceBench/GraphTraceBench.