triple1 listlengths 3 3 | triple2 listlengths 3 3 | triple3 listlengths 3 3 | triple1_labels listlengths 3 3 | triple2_labels listlengths 3 3 | triple3_labels listlengths 3 3 | description stringclasses 10
values | keep_triple bool 1
class | query_prompt stringclasses 10
values | query stringclasses 10
values | entity_a stringclasses 10
values | entity_b stringclasses 10
values | rel_b stringclasses 9
values | prompt stringclasses 10
values | raw_responses stringclasses 10
values |
|---|---|---|---|---|---|---|---|---|---|---|---|---|---|---|
[
"Q96581197",
"P57",
"Q100719246"
] | [
"Q96581188",
"P57",
"Q100719246"
] | [
"Q100719246",
"P1327",
"Q96581197"
] | [
"Jerémy Trouilh",
"director",
"Gagarine"
] | [
"Fanny Liatard",
"director",
"Gagarine"
] | [
"Gagarine",
"partner in business or sport",
"Jerémy Trouilh"
] | Q100719246__P57 | false | Task: Generate a Natural Language Query.
You will be given input in the following format:
Concept A: entity1
Concept B: entity2
Concept C: (concept B, rel, entity3)
Concept A and Concept C are connected via an independent concept B. Concept C is a property of Concept B, which the person is supposed to guess.
You... | What are different ways of connecting Jerémy Trouilh, a notable figure, and someone who has been a partner in business or sport with him? | Jerémy Trouilh | Jerémy Trouilh | partner in business or sport | Query: What are different ways of connecting Jerémy Trouilh, a notable figure, and someone who has been a partner in business or sport with him?
Task: Identify how two real-world entities are connected by producing MANY connection paths. A connection path is a sequence of factual triples (head, relationship, tail) for... | {"final_response": "<think>\n\n</think>\n\n<answer>\n{\n \"1\": [\n [\n \"Jer\u00e9my Trouilh\",\n \"collaborated with\",\n \"Fanny Liatard\"\n ],\n [\n \"Fanny Liatard\",\n \"partner in business or sport\",\n \"Jer\u00e9my Trouilh\"\n ]\n ],\n \"2\": [\n [\n \"Jer\u... |
[
"Q18048619",
"P2293",
"Q131755"
] | [
"Q18045544",
"P2293",
"Q131755"
] | [
"Q18045544",
"P5572",
"Q107244"
] | [
"TMEM132C",
"genetic association",
"bipolar disorder"
] | [
"CDH23",
"genetic association",
"bipolar disorder"
] | [
"CDH23",
"expressed in",
"monocyte"
] | Q131755__P2293 | false | Task: Generate a Natural Language Query.
You will be given input in the following format:
Concept A: entity1
Concept B: entity2
Concept C: (concept B, rel, entity3)
Concept A and Concept C are connected via an independent concept B. Concept C is a property of Concept B, which the person is supposed to guess.
You... | What are different ways of connecting TMEM132C, a gene associated with certain biological functions, and a substance that is expressed in monocytes? | TMEM132C | monocyte | expressed in | Query: What are different ways of connecting TMEM132C, a gene associated with certain biological functions, and a substance that is expressed in monocytes?
Task: Identify how two real-world entities are connected by producing MANY connection paths. A connection path is a sequence of factual triples (head, relationship... | {"final_response": "<think>\n\n</think>\n\n<answer>\n{\n \"1\": [\n [\n \"TMEM132C\",\n \"associated with\",\n \"hearing loss\"\n ],\n [\n \"hearing loss\",\n \"associated with\",\n \"GJB2\"\n ],\n [\n \"GJB2\",\n \"expressed in\",\n \"monocyte\"\n ]\n ],... |
[
"Q422606",
"P2175",
"Q12199"
] | [
"Q304330",
"P2175",
"Q12199"
] | [
"Q304330",
"P527",
"Q623"
] | [
"didanosine",
"medical condition treated",
"HIV/AIDS"
] | [
"abacavir",
"medical condition treated",
"HIV/AIDS"
] | [
"abacavir",
"has part(s)",
"carbon"
] | Q12199__P2175 | false | Task: Generate a Natural Language Query.
You will be given input in the following format:
Concept A: entity1
Concept B: entity2
Concept C: (concept B, rel, entity3)
Concept A and Concept C are connected via an independent concept B. Concept C is a property of Concept B, which the person is supposed to guess.
You... | What are different ways of connecting didanosine, a medication used to treat HIV, and a substance that has carbon as one of its components? | didanosine | carbon | has part(s) | Query: What are different ways of connecting didanosine, a medication used to treat HIV, and a substance that has carbon as one of its components?
Task: Identify how two real-world entities are connected by producing MANY connection paths. A connection path is a sequence of factual triples (head, relationship, tail) f... | {"final_response": "<think>\n\n</think>\n\n<answer>\n{\n \"1\": [\n [\n \"didanosine\",\n \"chemical formula\",\n \"C10H12N4O3\"\n ],\n [\n \"C10H12N4O3\",\n \"has part(s)\",\n \"carbon\"\n ]\n ],\n \"2\": [\n [\n \"didanosine\",\n \"is a\",\n \"nucleoside ... |
[
"Q1702841",
"P54",
"Q219714"
] | [
"Q19865964",
"P54",
"Q219714"
] | [
"Q19865964",
"P413",
"Q528145"
] | [
"Jon Runyan",
"member of sports team",
"Philadelphia Eagles"
] | [
"Ephesians Bartley",
"member of sports team",
"Philadelphia Eagles"
] | [
"Ephesians Bartley",
"position played on team / speciality",
"linebacker"
] | Q219714__P54 | false | Task: Generate a Natural Language Query.
You will be given input in the following format:
Concept A: entity1
Concept B: entity2
Concept C: (concept B, rel, entity3)
Concept A and Concept C are connected via an independent concept B. Concept C is a property of Concept B, which the person is supposed to guess.
You... | What are different ways of connecting Jon Runyan, the former American football player, and someone who played as a linebacker on a football team? | Jon Runyan | linebacker | position played on team / speciality | Query: What are different ways of connecting Jon Runyan, the former American football player, and someone who played as a linebacker on a football team?
Task: Identify how two real-world entities are connected by producing MANY connection paths. A connection path is a sequence of factual triples (head, relationship, t... | {"final_response": "<think>\n\n</think>\n\n<answer>\n{\n \"1\": [\n [\n \"Jon Runyan\",\n \"child\",\n \"Jon Runyan Jr.\"\n ],\n [\n \"Jon Runyan Jr.\",\n \"teammate\",\n \"Rashan Gary\"\n ],\n [\n \"Rashan Gary\",\n \"position played on team / speciality\",\n ... |
[
"Q313875",
"P463",
"Q254138"
] | [
"Q18218142",
"P463",
"Q254138"
] | [
"Q18218142",
"P1303",
"Q6607"
] | [
"Shawn Crahan",
"member of",
"Slipknot"
] | [
"Alessandro Venturella",
"member of",
"Slipknot"
] | [
"Alessandro Venturella",
"instrument",
"guitar"
] | Q254138__P463 | false | Task: Generate a Natural Language Query.
You will be given input in the following format:
Concept A: entity1
Concept B: entity2
Concept C: (concept B, rel, entity3)
Concept A and Concept C are connected via an independent concept B. Concept C is a property of Concept B, which the person is supposed to guess.
You... | What are different ways of connecting Shawn Crahan, the musician known for his work with Slipknot, and someone who plays the guitar? | Shawn Crahan | guitar | instrument | Query: What are different ways of connecting Shawn Crahan, the musician known for his work with Slipknot, and someone who plays the guitar?
Task: Identify how two real-world entities are connected by producing MANY connection paths. A connection path is a sequence of factual triples (head, relationship, tail) forming ... | {"final_response": "<think>\n\n</think>\n\n<answer>\n{\n \"1\": [\n [\n \"Shawn Crahan\",\n \"member of\",\n \"Slipknot\"\n ],\n [\n \"Slipknot\",\n \"has member\",\n \"Mick Thomson\"\n ],\n [\n \"Mick Thomson\",\n \"instrument\",\n \"guitar\"\n ]\n ],\n ... |
[
"Q10328595",
"P58",
"Q2848040"
] | [
"Q10328595",
"P58",
"Q1704125"
] | [
"Q1704125",
"P106",
"Q36180"
] | [
"Memórias Póstumas",
"screenwriter",
"André Klotzel"
] | [
"Memórias Póstumas",
"screenwriter",
"José Roberto Torero"
] | [
"José Roberto Torero",
"occupation",
"writer"
] | Q10328595__P58 | false | Task: Generate a Natural Language Query.
You will be given input in the following format:
Concept A: entity1
Concept B: entity2
Concept C: (concept B, rel, entity3)
Concept A and Concept C are connected via an independent concept B. Concept C is a property of Concept B, which the person is supposed to guess.
You... | What are different ways of connecting André Klotzel, a Brazilian filmmaker, and someone who is a writer by profession? | André Klotzel | writer | occupation | Query: What are different ways of connecting André Klotzel, a Brazilian filmmaker, and someone who is a writer by profession?
Task: Identify how two real-world entities are connected by producing MANY connection paths. A connection path is a sequence of factual triples (head, relationship, tail) forming a continuous c... | {"final_response": "<think>\n\n</think>\n\n<answer>\n{\n \"1\": [\n [\"Andr\u00e9 Klotzel\", \"directed\", \"A Marvada Carne\"],\n [\"A Marvada Carne\", \"screenwriter\", \"Andr\u00e9 Klotzel\"],\n [\"Andr\u00e9 Klotzel\", \"occupation\", \"writer\"]\n ],\n \"2\": [\n [\"Andr\u00e9 Klotzel\", \"directed\... |
[
"Q1050065",
"P58",
"Q471000"
] | [
"Q1050065",
"P58",
"Q577959"
] | [
"Q577959",
"P101",
"Q115156386"
] | [
"REC 2",
"screenwriter",
"Jaume Balagueró"
] | [
"REC 2",
"screenwriter",
"Paco Plaza"
] | [
"Paco Plaza",
"field of work",
"film screenwriting"
] | Q1050065__P58 | false | Task: Generate a Natural Language Query.
You will be given input in the following format:
Concept A: entity1
Concept B: entity2
Concept C: (concept B, rel, entity3)
Concept A and Concept C are connected via an independent concept B. Concept C is a property of Concept B, which the person is supposed to guess.
You... | What are different ways of connecting Jaume Balagueró, the Spanish filmmaker, and someone who works in the field of film screenwriting? | Jaume Balagueró | film screenwriting | field of work | Query: What are different ways of connecting Jaume Balagueró, the Spanish filmmaker, and someone who works in the field of film screenwriting?
Task: Identify how two real-world entities are connected by producing MANY connection paths. A connection path is a sequence of factual triples (head, relationship, tail) formi... | {"final_response": "<think>\n\n</think>\n\n<answer>\n{\n \"1\": [\n [\n \"Jaume Balaguer\u00f3\",\n \"field of work\",\n \"film screenwriting\"\n ]\n ],\n \"2\": [\n [\n \"Jaume Balaguer\u00f3\",\n \"co-director\",\n \"Paco Plaza\"\n ],\n [\n \"Paco Plaza\",\n \... |
[
"Q105081740",
"P57",
"Q15831456"
] | [
"Q105081740",
"P57",
"Q3189592"
] | [
"Q3189592",
"P69",
"Q3578463"
] | [
"The Deep House",
"director",
"Alexandre Bustillo"
] | [
"The Deep House",
"director",
"Julien Maury"
] | [
"Julien Maury",
"educated at",
"École Supérieure de Réalisation Audiovisuelle"
] | Q105081740__P57 | false | Task: Generate a Natural Language Query.
You will be given input in the following format:
Concept A: entity1
Concept B: entity2
Concept C: (concept B, rel, entity3)
Concept A and Concept C are connected via an independent concept B. Concept C is a property of Concept B, which the person is supposed to guess.
You... | What are different ways of connecting The Deep House, a horror film, and someone who was educated at the École Supérieure de Réalisation Audiovisuelle? | The Deep House | École Supérieure de Réalisation Audiovisuelle | educated at | Query: What are different ways of connecting The Deep House, a horror film, and someone who was educated at the École Supérieure de Réalisation Audiovisuelle?
Task: Identify how two real-world entities are connected by producing MANY connection paths. A connection path is a sequence of factual triples (head, relations... | {"final_response": "<think>\n\n</think>\n\n<answer>\n{\n \"1\": [\n [\n \"The Deep House\",\n \"directed by\",\n \"Julien Maury\"\n ],\n [\n \"Julien Maury\",\n \"educated at\",\n \"\u00c9cole Sup\u00e9rieure de R\u00e9alisation Audiovisuelle\"\n ]\n ],\n \"2\": [\n [\n ... |
[
"Q66671",
"P40",
"Q1339"
] | [
"Q309470",
"P40",
"Q1339"
] | [
"Q309470",
"P106",
"Q115785313"
] | [
"Maria Elisabeth Lämmerhirt",
"child",
"Johann Sebastian Bach"
] | [
"Johann Ambrosius Bach",
"child",
"Johann Sebastian Bach"
] | [
"Johann Ambrosius Bach",
"occupation",
"court musician"
] | Q1339__P40 | false | Task: Generate a Natural Language Query.
You will be given input in the following format:
Concept A: entity1
Concept B: entity2
Concept C: (concept B, rel, entity3)
Concept A and Concept C are connected via an independent concept B. Concept C is a property of Concept B, which the person is supposed to guess.
You... | What are different ways of connecting Maria Elisabeth Lämmerhirt and someone who held the occupation of a court musician? | Maria Elisabeth Lämmerhirt | court musician | occupation | Query: What are different ways of connecting Maria Elisabeth Lämmerhirt and someone who held the occupation of a court musician?
Task: Identify how two real-world entities are connected by producing MANY connection paths. A connection path is a sequence of factual triples (head, relationship, tail) forming a continuou... | {"final_response": "<think>\n\n</think>\n\n<answer>\n{\n \"1\": [\n [\n \"Maria Elisabeth L\u00e4mmerhirt\",\n \"mother of\",\n \"Johann Sebastian Bach\"\n ],\n [\n \"Johann Sebastian Bach\",\n \"occupation\",\n \"court musician\"\n ]\n ],\n \"2\": [\n [\n \"Maria El... |
[
"Q1087499",
"P2175",
"Q35869"
] | [
"Q417767",
"P2175",
"Q35869"
] | [
"Q417767",
"P3780",
"Q47521846"
] | [
"troleandomycin",
"medical condition treated",
"asthma"
] | [
"montelukast",
"medical condition treated",
"asthma"
] | [
"montelukast",
"active ingredient in",
"Singulair"
] | Q35869__P2175 | false | Task: Generate a Natural Language Query.
You will be given input in the following format:
Concept A: entity1
Concept B: entity2
Concept C: (concept B, rel, entity3)
Concept A and Concept C are connected via an independent concept B. Concept C is a property of Concept B, which the person is supposed to guess.
You... | What are different ways of connecting troleandomycin, a chemical compound, and a substance that is the active ingredient in Singulair? | troleandomycin | Singulair | active ingredient in | Query: What are different ways of connecting troleandomycin, a chemical compound, and a substance that is the active ingredient in Singulair?
Task: Identify how two real-world entities are connected by producing MANY connection paths. A connection path is a sequence of factual triples (head, relationship, tail) formin... | {"final_response": "<think>\n\n</think>\n\n<answer>\n{\n \"1\": [\n [\n \"troleandomycin\",\n \"drug class\",\n \"macrolide\"\n ],\n [\n \"macrolide\",\n \"interacts with\",\n \"Cytochrome P450 3A4\"\n ],\n [\n \"Cytochrome P450 3A4\",\n \"metabolizes\",\n \"... |
Dataset: connections-dev/connection_queries_jan12
This dataset was generated using the inference script with the following configuration:
Inference Parameters
Model Configuration
- Model Name:
gemini-3-pro-preview - Server URL:
Not specified - API Key:
Not provided - Request Timeout:
30 seconds
Query Configuration
- Query Type:
natural - Query Column:
query - Sampling Type:
original
Generation Parameters
- Temperature:
0.7 - Max Tokens:
16384 - Number of Runs:
1 - Chunk Size:
Entire dataset - Effort Level:
low
Feature Flags
- Step by Step Reasoning:
False - Reasoning Enabled:
False - No Curator:
False - Index Filter:
None (all samples) - Start Index:
70 - End Index:
80
Data Information
- Input File:
connections-dev/connection_queries_jan12 - Number of Samples:
10 - Output Filename:
connection_queries_jan12__natural__s70__e80__original__1__low__0.7__16384__gemini-3-pro-preview.jsonl
Dataset Structure
The dataset contains the following key columns:
query: The original queryentity_a: Starting entityentity_b: Target entityrel_b: Target relationshippath_prediction: Generated connection pathsprompt_for_path: Prompts used for path generation
Usage
You can load this dataset using:
from datasets import load_dataset
dataset = load_dataset("s70_e80_original_1_low_0.7_16384_gemini-3-pro-preview__raw")
Generation Details
This dataset was generated on 2026-01-22 14:45:48 using the inference pipeline with the above configuration.
- Downloads last month
- 2