id
stringlengths 1
12
| question
stringlengths 22
566
| golden_answers
dict | data_source
stringclasses 2
values | prompt
listlengths 1
1
| ability
stringclasses 6
values | reward_model
dict | extra_info
dict | metadata
dict |
|---|---|---|---|---|---|---|---|---|
exfever_5913
|
To Die For is a 1997 film directed by Gus Van Sant that stars a British actress who is one of the world's highest-paid actresses.
|
{
"evidence": [
"To Die For",
"Nicole Kidman"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nTo Die For is a 1997 film directed by Gus Van Sant that stars a British actress who is one of the world's highest-paid actresses.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"To Die For",
"Nicole Kidman"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 1584,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
76599
|
The winners of Nallur Polling Division, one of Jaffna Electoral District's eleven polling divisions, has matched its results 5/7 times but has failed to match the national result everytime from 1989-2015.
|
{
"evidence": [
"Nallur Polling Division_sentence_1"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe winners of Nallur Polling Division, one of Jaffna Electoral District's eleven polling divisions, has matched its results 5/7 times but has failed to match the national result everytime from 1989-2015.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Nallur Polling Division_sentence_1"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 1585,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.823",
"value": "?search="
},
{
"operation": "search",
"time": "12.795",
"value": "Nallur Polling Division"
},
{
"operation": "Now on",
"time": "14.242",
"value": "Nallur Polling Division"
},
{
"operation": "Highlighting",
"time": "18.496",
"value": "Nallur Polling Division_sentence_0"
},
{
"operation": "Highlighting",
"time": "33.014",
"value": "Nallur Polling Division_sentence_1"
},
{
"operation": "Highlighting",
"time": "47.126",
"value": "Nallur Polling Division_sentence_2"
},
{
"operation": "Highlighting",
"time": "47.129",
"value": "Nallur Polling Division_section_10"
},
{
"operation": "Highlighting",
"time": "47.13",
"value": "Nallur Polling Division_section_11"
},
{
"operation": "Highlighting deleted",
"time": "58.091",
"value": "Nallur Polling Division_sentence_0"
},
{
"operation": "Highlighting deleted",
"time": "96.329",
"value": "Nallur Polling Division_sentence_2"
},
{
"operation": "Highlighting deleted",
"time": "96.336",
"value": "Nallur Polling Division_section_10"
},
{
"operation": "Highlighting deleted",
"time": "96.338",
"value": "Nallur Polling Division_section_11"
},
{
"operation": "finish",
"time": "113.511",
"value": "finish"
}
]
}
|
exfever_2428
|
Paris, a mythological nobleman, is the son of Queen Hecuba and the legendary and first king of Troy during the Trojan War.
|
{
"evidence": [
"Paris (mythology)",
"Priam"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nParis, a mythological nobleman, is the son of Queen Hecuba and the legendary and first king of Troy during the Trojan War.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Paris (mythology)",
"Priam"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 1586,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
80401
|
Hackpfüffe is a village in Germany that was first mentioned as Bablide, it was documented as Pefelde in 1131 and in 1231 it was known as imperial estate Peflede.
|
{
"evidence": [
"Hackpfüffel_sentence_0",
"Hackpfüffel_sentence_4",
"Hackpfüffel_sentence_5",
"Hackpfüffel_sentence_7"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nHackpfüffe is a village in Germany that was first mentioned as Bablide, it was documented as Pefelde in 1131 and in 1231 it was known as imperial estate Peflede.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Hackpfüffel_sentence_0",
"Hackpfüffel_sentence_4",
"Hackpfüffel_sentence_5",
"Hackpfüffel_sentence_7"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 1587,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.776",
"value": "?search="
},
{
"operation": "search",
"time": "311.619",
"value": "Hackpfüffe"
},
{
"operation": "Now on",
"time": "312.497",
"value": "undefined"
},
{
"operation": "search",
"time": "327.083",
"value": "Bablide"
},
{
"operation": "Now on",
"time": "327.723",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "335.321",
"value": "Hackpfüffel"
},
{
"operation": "Now on",
"time": "335.908",
"value": "Hackpfüffel"
},
{
"operation": "Highlighting",
"time": "350.402",
"value": "Hackpfüffel_sentence_0"
},
{
"operation": "Page search",
"time": "355.22",
"value": "Bablide"
},
{
"operation": "Highlighting",
"time": "357.47",
"value": "Hackpfüffel_sentence_4"
},
{
"operation": "Page search",
"time": "371.812",
"value": "Pefelde"
},
{
"operation": "Highlighting",
"time": "374.459",
"value": "Hackpfüffel_sentence_5"
},
{
"operation": "Highlighting",
"time": "417.417",
"value": "Hackpfüffel_sentence_7"
},
{
"operation": "finish",
"time": "451.772",
"value": "finish"
}
]
}
|
exfever_7908
|
Coronary artery disease is the most common cardiovascular disease, which includes what is commonly known as a heart attack, which occurs when blood flow decreases or stops to the coronary artery of the heart, causing damage to the stomach muscle.
|
{
"evidence": [
"Coronary artery disease",
"Myocardial infarction"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nCoronary artery disease is the most common cardiovascular disease, which includes what is commonly known as a heart attack, which occurs when blood flow decreases or stops to the coronary artery of the heart, causing damage to the stomach muscle.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Coronary artery disease",
"Myocardial infarction"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 1588,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_6238
|
Popeye the Movie is a 1981 American musical comedy film directed by an American film director, screenwriter, and producer known as a five-time nominee of the Academy Award for Best Director.
|
{
"evidence": [
"Popeye (film)",
"Robert Altman"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nPopeye the Movie is a 1981 American musical comedy film directed by an American film director, screenwriter, and producer known as a five-time nominee of the Academy Award for Best Director.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Popeye (film)",
"Robert Altman"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 1589,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
21022
|
Matsudaira Yasuhide was born to the high-ranking hatamoto Matsudaira Yasuzumi, a samurai in the direct service of the Tokugawa shogunate, and later took the name of Daimyō.
|
{
"evidence": [
"Matsudaira Yasuhide_sentence_3",
"Hatamoto_sentence_0"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nMatsudaira Yasuhide was born to the high-ranking hatamoto Matsudaira Yasuzumi, a samurai in the direct service of the Tokugawa shogunate, and later took the name of Daimyō.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Matsudaira Yasuhide_sentence_3",
"Hatamoto_sentence_0"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 1590,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.597",
"value": "?search="
},
{
"operation": "search",
"time": "6.051",
"value": "Matsudaira Yasuhide"
},
{
"operation": "Now on",
"time": "6.722",
"value": "Matsudaira Yasuhide"
},
{
"operation": "Page search",
"time": "15.258",
"value": "Matsudaira Yasuzumi"
},
{
"operation": "Highlighting",
"time": "19.446",
"value": "Matsudaira Yasuhide_sentence_3"
},
{
"operation": "Highlighting deleted",
"time": "36.88",
"value": "Matsudaira Yasuhide_sentence_3"
},
{
"operation": "Highlighting",
"time": "37.348",
"value": "Matsudaira Yasuhide_sentence_3"
},
{
"operation": "Page search",
"time": "41.177",
"value": "samurai"
},
{
"operation": "Highlighting deleted",
"time": "43.275",
"value": "Matsudaira Yasuhide_sentence_3"
},
{
"operation": "Highlighting",
"time": "44.028",
"value": "Matsudaira Yasuhide_sentence_3"
},
{
"operation": "hyperlink",
"time": "55.354",
"value": "hatamoto"
},
{
"operation": "Now on",
"time": "56.132",
"value": "Hatamoto"
},
{
"operation": "Page search",
"time": "59.597",
"value": "Matsudaira Yasuzumi"
},
{
"operation": "Page search",
"time": "64.876",
"value": "samurai"
},
{
"operation": "Highlighting",
"time": "67.783",
"value": "Hatamoto_sentence_0"
},
{
"operation": "back-button-clicked",
"time": "73.285",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "73.373",
"value": "Matsudaira Yasuhide"
},
{
"operation": "Page search",
"time": "92.11",
"value": "Daimyō"
},
{
"operation": "hyperlink",
"time": "114.036",
"value": "hatamoto"
},
{
"operation": "Now on",
"time": "114.446",
"value": "Hatamoto"
},
{
"operation": "Page search",
"time": "118.214",
"value": "Daimyō"
},
{
"operation": "back-button-clicked",
"time": "130.679",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "130.759",
"value": "Matsudaira Yasuhide"
},
{
"operation": "Page search",
"time": "135.886",
"value": "Daimyō"
},
{
"operation": "Page search",
"time": "144.337",
"value": "name"
},
{
"operation": "finish",
"time": "191",
"value": "finish"
}
]
}
|
81678
|
Pancuran Tujuh is in Kemutung Lor VIllage, 1.6 miles west of the Baturraden Tourist Resort.
|
{
"evidence": [
"Pancuran Tujuh_sentence_2",
"Pancuran Tujuh_sentence_3"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nPancuran Tujuh is in Kemutung Lor VIllage, 1.6 miles west of the Baturraden Tourist Resort.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Pancuran Tujuh_sentence_2",
"Pancuran Tujuh_sentence_3"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 1591,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.878",
"value": "?search="
},
{
"operation": "search",
"time": "10.595",
"value": "Pancuran Tujuh"
},
{
"operation": "Now on",
"time": "11.688",
"value": "Pancuran Tujuh"
},
{
"operation": "Highlighting",
"time": "29.696",
"value": "Pancuran Tujuh_sentence_2"
},
{
"operation": "Highlighting",
"time": "41.714",
"value": "Pancuran Tujuh_sentence_3"
},
{
"operation": "finish",
"time": "47.248",
"value": "finish"
}
]
}
|
exfever_5879
|
Mediterranean cuisine is the food and methods of preparation used by the people of the Mediterranean Basin; writers define the three core elements of the cuisine as the olive, wheat, and the grape, yielding olive oil, bread and a type of food typically made from an unleavened dough of wheat flour mixed with water or eggs, formed into sheets or other shapes, then cooked by boiling or baking.
|
{
"evidence": [
"Mediterranean cuisine",
"Pasta"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nMediterranean cuisine is the food and methods of preparation used by the people of the Mediterranean Basin; writers define the three core elements of the cuisine as the olive, wheat, and the grape, yielding olive oil, bread and a type of food typically made from an unleavened dough of wheat flour mixed with water or eggs, formed into sheets or other shapes, then cooked by boiling or baking.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Mediterranean cuisine",
"Pasta"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 1592,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_4989
|
Broadcast News is a romantic comedy-drama film that stars an American actor who became a well-known supporting actor in the 1980s.
|
{
"evidence": [
"Broadcast News (film)",
"Robert Prosky"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nBroadcast News is a romantic comedy-drama film that stars an American actor who became a well-known supporting actor in the 1980s.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Broadcast News (film)",
"Robert Prosky"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 1593,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
37235
|
In the cult of Amalberga of Temse, her vita has a connection to Charlemagne, the monarch who united western and central Europe in the middle ages.
|
{
"evidence": [
"Amalberga of Temse_sentence_6",
"Charlemagne_sentence_2"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nIn the cult of Amalberga of Temse, her vita has a connection to Charlemagne, the monarch who united western and central Europe in the middle ages.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Amalberga of Temse_sentence_6",
"Charlemagne_sentence_2"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 1594,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.691",
"value": "?search="
},
{
"operation": "search",
"time": "17.083",
"value": "Contains...the cult of Amalberga of Temse"
},
{
"operation": "Now on",
"time": "17.783",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "33.097",
"value": "Amalberga of Temse"
},
{
"operation": "Now on",
"time": "33.629",
"value": "Amalberga of Temse"
},
{
"operation": "Highlighting",
"time": "47.662",
"value": "Amalberga of Temse_sentence_6"
},
{
"operation": "hyperlink",
"time": "79.661",
"value": "Charlemagne"
},
{
"operation": "Now on",
"time": "81.909",
"value": "Charlemagne"
},
{
"operation": "Highlighting",
"time": "116.637",
"value": "Charlemagne_sentence_2"
},
{
"operation": "finish",
"time": "143.303",
"value": "finish"
}
]
}
|
exfever_73
|
The Royal Prussian Academy of Sciences was an academy established in Berlin, Germany on 11 July 1900, which was a institution of a Romance language of the Indo-European family.
|
{
"evidence": [
"Prussian Academy of Sciences",
"French language"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Royal Prussian Academy of Sciences was an academy established in Berlin, Germany on 11 July 1900, which was a institution of a Romance language of the Indo-European family.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Prussian Academy of Sciences",
"French language"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 1595,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_1179
|
The Starrcade was a repeated wrestling event held by an Italian professional wrestling promotion founded by Ted Turner.
|
{
"evidence": [
"Starrcade",
"World Championship Wrestling"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Starrcade was a repeated wrestling event held by an Italian professional wrestling promotion founded by Ted Turner.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Starrcade",
"World Championship Wrestling"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 1596,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_3348
|
Corey Todd Taylor is an American singer known as the lead vocalist and lyricist of the American heavy metal band formed in Des Moines, Iowa, in 1994.
|
{
"evidence": [
"Corey Taylor",
"Slipknot (band)"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nCorey Todd Taylor is an American singer known as the lead vocalist and lyricist of the American heavy metal band formed in Des Moines, Iowa, in 1994.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Corey Taylor",
"Slipknot (band)"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 1597,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7210
|
Venice is the most visited city in the Padua-Treviso-Venice Metropolitan Area.
|
{
"evidence": [
"Venice",
"Venetian Lagoon"
],
"label": "NOT ENOUGH INFO"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nVenice is the most visited city in the Padua-Treviso-Venice Metropolitan Area.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Venice",
"Venetian Lagoon"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 1598,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_1334
|
Anne Bancroft was an actress who received an award that was first presented in 1929 at a private dinner hosted by Douglas Fairbanks in a historic hotel located at 7000 Hollywood Boulevard.
|
{
"evidence": [
"Anne Bancroft",
"Academy Awards",
"The Hollywood Roosevelt Hotel"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nAnne Bancroft was an actress who received an award that was first presented in 1929 at a private dinner hosted by Douglas Fairbanks in a historic hotel located at 7000 Hollywood Boulevard.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Anne Bancroft",
"Academy Awards",
"The Hollywood Roosevelt Hotel"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 1599,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
54003
|
The Cuba–Haiti Maritime Boundary Agreement is a 20th Century treaty between Cuba and Haiti which delimits the maritime boundary between the two countries.
|
{
"evidence": [
"Cuba–Haiti Maritime Boundary Agreement_sentence_0"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Cuba–Haiti Maritime Boundary Agreement is a 20th Century treaty between Cuba and Haiti which delimits the maritime boundary between the two countries.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Cuba–Haiti Maritime Boundary Agreement_sentence_0"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 1600,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.448",
"value": "?search="
},
{
"operation": "search",
"time": "11.914",
"value": "Cuba-Haiti Maritime Boundary Agreement"
},
{
"operation": "Now on",
"time": "13.46",
"value": "Cuba–Haiti Maritime Boundary Agreement"
},
{
"operation": "Highlighting",
"time": "18.544",
"value": "Cuba–Haiti Maritime Boundary Agreement_sentence_0"
},
{
"operation": "finish",
"time": "57.152",
"value": "finish"
}
]
}
|
exfever_532
|
Thomas Jacob Black is an American actor who voices the giant panda named Po from Kung Fu Panda films by an American animation studio that produces animated films and television programs and is a subsidiary of Universal Pictures, a division of NBCUniversal.
|
{
"evidence": [
"Jack Black",
"DreamWorks Animation"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThomas Jacob Black is an American actor who voices the giant panda named Po from Kung Fu Panda films by an American animation studio that produces animated films and television programs and is a subsidiary of Universal Pictures, a division of NBCUniversal.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Jack Black",
"DreamWorks Animation"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 1601,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
47221
|
Geneviève de Brunelle dismantles the royalist Chouannerie in Normandy, and in 1808, in return for her loyalty to the monarchy, she was given the title "Lady".
|
{
"evidence": [
"Geneviève de Brunelle_sentence_2"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nGeneviève de Brunelle dismantles the royalist Chouannerie in Normandy, and in 1808, in return for her loyalty to the monarchy, she was given the title \"Lady\".\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Geneviève de Brunelle_sentence_2"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 1602,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.321",
"value": "?search="
},
{
"operation": "search",
"time": "15.895",
"value": "Contains...Geneviève de Brunelle dismantles the royalist Chouannerie in Normandy"
},
{
"operation": "Now on",
"time": "17.695",
"value": "undefined"
},
{
"operation": "search",
"time": "37.474",
"value": "Geneviève de Brunelle"
},
{
"operation": "Now on",
"time": "38.589",
"value": "Geneviève de Brunelle"
},
{
"operation": "Highlighting",
"time": "57.952",
"value": "Geneviève de Brunelle_sentence_1"
},
{
"operation": "hyperlink",
"time": "82.798",
"value": "Chouannerie"
},
{
"operation": "Now on",
"time": "83.745",
"value": "Chouannerie"
},
{
"operation": "Page search",
"time": "97.204",
"value": "Lady"
},
{
"operation": "back-button-clicked",
"time": "121.927",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "122.241",
"value": "Geneviève de Brunelle"
},
{
"operation": "Highlighting",
"time": "129.717",
"value": "Geneviève de Brunelle_sentence_2"
},
{
"operation": "Highlighting deleted",
"time": "169.321",
"value": "Geneviève de Brunelle_sentence_1"
},
{
"operation": "finish",
"time": "173.791",
"value": "finish"
}
]
}
|
exfever_2707
|
Andrea Joy Cook is a Canadian actress who appeared in a 1999 American psychological drama film written and directed by Mel Gibson.
|
{
"evidence": [
"A. J. Cook",
"The Virgin Suicides (film)"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nAndrea Joy Cook is a Canadian actress who appeared in a 1999 American psychological drama film written and directed by Mel Gibson.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"A. J. Cook",
"The Virgin Suicides (film)"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 1603,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
75181
|
Karin Aijmer is a member of Cambridge Grammar preference panel and on the committee for ICAME which was founded in 1977 as an international group working in corpus linguistics to digitize English text.
|
{
"evidence": [
"Karin Aijmer_sentence_5",
"Karin Aijmer_sentence_4",
"International Computer Archive of Modern and Medieval English_sentence_1"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nKarin Aijmer is a member of Cambridge Grammar preference panel and on the committee for ICAME which was founded in 1977 as an international group working in corpus linguistics to digitize English text.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Karin Aijmer_sentence_5",
"Karin Aijmer_sentence_4",
"International Computer Archive of Modern and Medieval English_sentence_1"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 1604,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.886",
"value": "?search="
},
{
"operation": "search",
"time": "54.954",
"value": "Karin Aijmer"
},
{
"operation": "Now on",
"time": "55.818",
"value": "Karin Aijmer"
},
{
"operation": "Highlighting",
"time": "78.13",
"value": "Karin Aijmer_sentence_5"
},
{
"operation": "Highlighting",
"time": "113.219",
"value": "Karin Aijmer_sentence_4"
},
{
"operation": "hyperlink",
"time": "129.843",
"value": "ICAME"
},
{
"operation": "Now on",
"time": "130.348",
"value": "International Computer Archive of Modern and Medieval English"
},
{
"operation": "Highlighting",
"time": "133.011",
"value": "International Computer Archive of Modern and Medieval English_sentence_1"
},
{
"operation": "finish",
"time": "146.529",
"value": "finish"
}
]
}
|
57953
|
Lenore Glen Offord (October 24, 1905 in Washington – April 24, 1991 in Oregon) was an American author and reviewer of detective fiction who authored twelve books and received the Edgar Award for Outstanding Criticism in 1952.
|
{
"evidence": [
"Lenore Glen Offord_sentence_0",
"Lenore Glen Offord_sentence_1",
"Lenore Glen Offord_sentence_4",
"Lenore Glen Offord_sentence_6"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nLenore Glen Offord (October 24, 1905 in Washington – April 24, 1991 in Oregon) was an American author and reviewer of detective fiction who authored twelve books and received the Edgar Award for Outstanding Criticism in 1952.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Lenore Glen Offord_sentence_0",
"Lenore Glen Offord_sentence_1",
"Lenore Glen Offord_sentence_4",
"Lenore Glen Offord_sentence_6"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 1605,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.734",
"value": "?search="
},
{
"operation": "search",
"time": "8.383",
"value": "Lenore Glen Offord"
},
{
"operation": "Now on",
"time": "9.494",
"value": "Lenore Glen Offord"
},
{
"operation": "Highlighting",
"time": "13.407",
"value": "Lenore Glen Offord_sentence_0"
},
{
"operation": "Highlighting",
"time": "18.869",
"value": "Lenore Glen Offord_sentence_1"
},
{
"operation": "Highlighting",
"time": "28.629",
"value": "Lenore Glen Offord_sentence_4"
},
{
"operation": "Highlighting",
"time": "37.157",
"value": "Lenore Glen Offord_sentence_6"
},
{
"operation": "Page search",
"time": "52.788",
"value": "Oregon"
},
{
"operation": "Page search",
"time": "53.248",
"value": "Oregon"
},
{
"operation": "Page search",
"time": "55.466",
"value": "Oregon"
},
{
"operation": "Highlighting deleted",
"time": "73.071",
"value": "Lenore Glen Offord_sentence_0"
},
{
"operation": "Highlighting deleted",
"time": "75.814",
"value": "Lenore Glen Offord_sentence_6"
},
{
"operation": "Highlighting deleted",
"time": "76.391",
"value": "Lenore Glen Offord_sentence_4"
},
{
"operation": "Highlighting deleted",
"time": "77.989",
"value": "Lenore Glen Offord_sentence_1"
},
{
"operation": "Highlighting",
"time": "78.982",
"value": "Lenore Glen Offord_sentence_0"
},
{
"operation": "Highlighting",
"time": "81.022",
"value": "Lenore Glen Offord_sentence_1"
},
{
"operation": "Highlighting",
"time": "90.534",
"value": "Lenore Glen Offord_sentence_4"
},
{
"operation": "Highlighting",
"time": "94.598",
"value": "Lenore Glen Offord_sentence_6"
},
{
"operation": "Now on",
"time": "111.354",
"value": "Lenore Glen Offord"
},
{
"operation": "finish",
"time": "126.84",
"value": "finish"
}
]
}
|
2163
|
Rozlyne Clarke released her second single in 1991 from an album produced by Eddy Beatboxing.
|
{
"evidence": [
"Rozlyne Clarke_sentence_18",
"Rozlyne Clarke_sentence_19"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nRozlyne Clarke released her second single in 1991 from an album produced by Eddy Beatboxing.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Rozlyne Clarke_sentence_18",
"Rozlyne Clarke_sentence_19"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 1606,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "search",
"time": "17.911",
"value": "Rozlyne Clarke"
},
{
"operation": "Highlighting",
"time": "39.951",
"value": "Rozlyne Clarke_sentence_18"
},
{
"operation": "Highlighting",
"time": "53.599",
"value": "Rozlyne Clarke_sentence_19"
},
{
"operation": "finish",
"time": "70.648",
"value": "finish"
}
]
}
|
33732
|
Richard Vissing was co-owner of the Vissings Dodge, established in 1968, and ran a wrecker service where he pulled in many cars during his days, with the most famous being that of John Dillinger.
|
{
"evidence": [
"Richard Vissing_sentence_5",
"Richard Vissing_sentence_8",
"Richard Vissing_sentence_9"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nRichard Vissing was co-owner of the Vissings Dodge, established in 1968, and ran a wrecker service where he pulled in many cars during his days, with the most famous being that of John Dillinger.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Richard Vissing_sentence_5",
"Richard Vissing_sentence_8",
"Richard Vissing_sentence_9"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 1607,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.755",
"value": "?search="
},
{
"operation": "search",
"time": "9.218",
"value": "Richard Vissing"
},
{
"operation": "Now on",
"time": "9.967",
"value": "Richard Vissing"
},
{
"operation": "Highlighting",
"time": "24.973",
"value": "Richard Vissing_sentence_5"
},
{
"operation": "Highlighting",
"time": "27.95",
"value": "Richard Vissing_sentence_8"
},
{
"operation": "Highlighting",
"time": "44.445",
"value": "Richard Vissing_sentence_9"
},
{
"operation": "finish",
"time": "56.341",
"value": "finish"
}
]
}
|
exfever_3382
|
Arkham Asylum is a fictional psychiatric hospital/prison that later appeared in thick books that saw a gradual decline in the wake of the telecommunication medium used for transmitting moving images.
|
{
"evidence": [
"Arkham Asylum",
"American comic book",
"Television"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nArkham Asylum is a fictional psychiatric hospital/prison that later appeared in thick books that saw a gradual decline in the wake of the telecommunication medium used for transmitting moving images.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Arkham Asylum",
"American comic book",
"Television"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 1608,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_8334
|
Jefferson Finis Davis was an American politician who represents Mississippi in the United States Senate and the House of Representatives before the civil war in the United States between the Union and the Confederacy.
|
{
"evidence": [
"Jefferson Davis",
"American Civil War"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nJefferson Finis Davis was an American politician who represents Mississippi in the United States Senate and the House of Representatives before the civil war in the United States between the Union and the Confederacy.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Jefferson Davis",
"American Civil War"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 1609,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
33291
|
Wafi City Mall is the premier life of glass specially selected from mosques and temples across the Islamic world, Opened in 1991, the mall includes over 200 stores.
|
{
"evidence": [
"Wafi City_sentence_7",
"Wafi City_sentence_6"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nWafi City Mall is the premier life of glass specially selected from mosques and temples across the Islamic world, Opened in 1991, the mall includes over 200 stores.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Wafi City_sentence_7",
"Wafi City_sentence_6"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 1610,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.134",
"value": "?search="
},
{
"operation": "search",
"time": "13.212",
"value": "Wafi City Mall"
},
{
"operation": "Now on",
"time": "14.539",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "16.908",
"value": "Wafi City"
},
{
"operation": "Now on",
"time": "17.348",
"value": "Wafi City"
},
{
"operation": "Highlighting",
"time": "26.361",
"value": "Wafi City_sentence_7"
},
{
"operation": "Highlighting",
"time": "37.364",
"value": "Wafi City_sentence_6"
},
{
"operation": "finish",
"time": "43.423",
"value": "finish"
}
]
}
|
exfever_545
|
The Ottoman Empire came before the Mughal Empire.
|
{
"evidence": [
"Mughal Empire",
"Kashmir"
],
"label": "NOT ENOUGH INFO"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Ottoman Empire came before the Mughal Empire.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Mughal Empire",
"Kashmir"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 1611,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
27653
|
The Valvata piscinalis is a European valve snail that was introduced to Lake Tahoe in 1897, and in 40 years, dispersed to Lake Erie (the fourth-largest lake of the five Great Lakes in North America).
|
{
"evidence": [
"Valvata piscinalis_sentence_0",
"Valvata piscinalis_sentence_22",
"Valvata piscinalis_sentence_23",
"Lake Erie_sentence_1"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Valvata piscinalis is a European valve snail that was introduced to Lake Tahoe in 1897, and in 40 years, dispersed to Lake Erie (the fourth-largest lake of the five Great Lakes in North America).\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Valvata piscinalis_sentence_0",
"Valvata piscinalis_sentence_22",
"Valvata piscinalis_sentence_23",
"Lake Erie_sentence_1"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 1612,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.675",
"value": "?search="
},
{
"operation": "search",
"time": "6.679",
"value": "Valvata piscinalis"
},
{
"operation": "Now on",
"time": "7.397",
"value": "Valvata piscinalis"
},
{
"operation": "Highlighting",
"time": "12.256",
"value": "Valvata piscinalis_sentence_0"
},
{
"operation": "Highlighting",
"time": "36.91",
"value": "Valvata piscinalis_sentence_22"
},
{
"operation": "Highlighting",
"time": "44.678",
"value": "Valvata piscinalis_sentence_23"
},
{
"operation": "hyperlink",
"time": "50.392",
"value": "Lake Erie"
},
{
"operation": "Now on",
"time": "51.333",
"value": "Lake Erie"
},
{
"operation": "Highlighting",
"time": "59.086",
"value": "Lake Erie_sentence_1"
},
{
"operation": "finish",
"time": "85.631",
"value": "finish"
}
]
}
|
exfever_7688
|
The Philadelphia Museum of Art is an art museum originally chartered in 1876 for the first official World's Fair to be held in the United States to celebrate the 105th anniversary of the signing of the Declaration of Independence in Philadelphia.
|
{
"evidence": [
"Philadelphia Museum of Art",
"Centennial Exposition"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Philadelphia Museum of Art is an art museum originally chartered in 1876 for the first official World's Fair to be held in the United States to celebrate the 105th anniversary of the signing of the Declaration of Independence in Philadelphia.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Philadelphia Museum of Art",
"Centennial Exposition"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 1613,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_8522
|
The Greenland shark is the most harmless extant species of shark.
|
{
"evidence": [
"Greenland shark",
"Species distribution"
],
"label": "NOT ENOUGH INFO"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Greenland shark is the most harmless extant species of shark.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Greenland shark",
"Species distribution"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 1614,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_6325
|
Superman is a superhero whose ship landed in the American countryside, near the fictional town in American comic books published by DC Comics.
|
{
"evidence": [
"Superman",
"Smallville (comics)"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nSuperman is a superhero whose ship landed in the American countryside, near the fictional town in American comic books published by DC Comics.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Superman",
"Smallville (comics)"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 1615,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_3236
|
Leati Joseph "Joe" Anoaʻi is a Japanese professional wrestler who is a part of the Anoaʻi family, which also includes his father, an American Samoan retired professional wrestler, best known as one-half of the tag team The Wild Samoans with his brother Afa.
|
{
"evidence": [
"Roman Reigns",
"Sika Anoaʻi"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nLeati Joseph \"Joe\" Anoaʻi is a Japanese professional wrestler who is a part of the Anoaʻi family, which also includes his father, an American Samoan retired professional wrestler, best known as one-half of the tag team The Wild Samoans with his brother Afa.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Roman Reigns",
"Sika Anoaʻi"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 1616,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
66696
|
The SWAC commissioner announced that the game had been officially sanctioned, however the bowl was not revived and although Grambling was selected as the consensus black college football national champion.
|
{
"evidence": [
"Pelican Bowl_sentence_13",
"Pelican Bowl_sentence_14",
"Pelican Bowl_sentence_19"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe SWAC commissioner announced that the game had been officially sanctioned, however the bowl was not revived and although Grambling was selected as the consensus black college football national champion.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Pelican Bowl_sentence_13",
"Pelican Bowl_sentence_14",
"Pelican Bowl_sentence_19"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 1617,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.498",
"value": "?search="
},
{
"operation": "search",
"time": "18.705",
"value": "\"SWAC commissioner\""
},
{
"operation": "Now on",
"time": "20.071",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "23.311",
"value": "Pelican Bowl"
},
{
"operation": "Now on",
"time": "24.383",
"value": "Pelican Bowl"
},
{
"operation": "Page search",
"time": "27.447",
"value": "SWAC commissioner"
},
{
"operation": "Highlighting",
"time": "31.457",
"value": "Pelican Bowl_sentence_13"
},
{
"operation": "Highlighting",
"time": "38.986",
"value": "Pelican Bowl_sentence_14"
},
{
"operation": "Highlighting",
"time": "42.518",
"value": "Pelican Bowl_sentence_16"
},
{
"operation": "Highlighting",
"time": "57.64",
"value": "Pelican Bowl_sentence_19"
},
{
"operation": "Highlighting deleted",
"time": "58.528",
"value": "Pelican Bowl_sentence_16"
},
{
"operation": "finish",
"time": "99.062",
"value": "finish"
}
]
}
|
exfever_875
|
Mark Burg is the co-founder of Evolution Entertainment and produced the Saw film series as well as an American television sitcom that aired on CBS for twelve seasons from September 22, 2003, to February 19, 2006.
|
{
"evidence": [
"Mark Burg",
"Two and a Half Men"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nMark Burg is the co-founder of Evolution Entertainment and produced the Saw film series as well as an American television sitcom that aired on CBS for twelve seasons from September 22, 2003, to February 19, 2006.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Mark Burg",
"Two and a Half Men"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 1618,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
7515
|
The plot of La Suegra tells the story of Victoria Maldonado from Bogota, Colombia, who decides to go to Miami to pursue her modeling career leaving her boyfriend Fernando behind.
|
{
"evidence": [
"La suegra_sentence_1",
"La suegra_sentence_3",
"La suegra_sentence_4"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe plot of La Suegra tells the story of Victoria Maldonado from Bogota, Colombia, who decides to go to Miami to pursue her modeling career leaving her boyfriend Fernando behind.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"La suegra_sentence_1",
"La suegra_sentence_3",
"La suegra_sentence_4"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 1619,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.774",
"value": "?search="
},
{
"operation": "search",
"time": "24.867",
"value": "La suegra"
},
{
"operation": "Now on",
"time": "25.916",
"value": "La suegra"
},
{
"operation": "Highlighting",
"time": "38.162",
"value": "La suegra_sentence_1"
},
{
"operation": "Highlighting",
"time": "47.483",
"value": "La suegra_sentence_3"
},
{
"operation": "Page search",
"time": "79.508",
"value": "fernando"
},
{
"operation": "Highlighting",
"time": "108.075",
"value": "La suegra_sentence_4"
},
{
"operation": "finish",
"time": "195.545",
"value": "finish"
}
]
}
|
exfever_7235
|
Sarah Palin served as the 9th governor of Alaska from 2006 until being replaced by Lieutenant Sean Parnell as the governor of a state located in the Western United States.
|
{
"evidence": [
"Sarah Palin",
"Resignation of Sarah Palin",
"Alaska"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nSarah Palin served as the 9th governor of Alaska from 2006 until being replaced by Lieutenant Sean Parnell as the governor of a state located in the Western United States.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Sarah Palin",
"Resignation of Sarah Palin",
"Alaska"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 1620,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
8790
|
Panshi is a city China.
|
{
"evidence": [
"Panshi_sentence_1"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nPanshi is a city China.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"Panshi_sentence_1"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 1621,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.842",
"value": "undefined"
},
{
"operation": "search",
"time": "5.306",
"value": "panshi"
},
{
"operation": "Now on",
"time": "6.745",
"value": "Panshi"
},
{
"operation": "Highlighting",
"time": "9.678",
"value": "Panshi_sentence_1"
},
{
"operation": "finish",
"time": "18.895",
"value": "finish"
}
]
}
|
46107
|
Shoal Bay Coastal Reserve (a protected area in the Territory of Africa) is bordered by the Atlantic Ocean on the west and the R27 coastal road, and surrounding it are numerous lagoons.
|
{
"evidence": [
"Shoal Bay Coastal Reserve_sentence_0",
"Shoal Bay Coastal Reserve_sentence_2"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nShoal Bay Coastal Reserve (a protected area in the Territory of Africa) is bordered by the Atlantic Ocean on the west and the R27 coastal road, and surrounding it are numerous lagoons.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Shoal Bay Coastal Reserve_sentence_0",
"Shoal Bay Coastal Reserve_sentence_2"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 1622,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.891",
"value": "?search="
},
{
"operation": "search",
"time": "19.504",
"value": "Shoal Bay coastal reserve"
},
{
"operation": "Now on",
"time": "20.148",
"value": "Shoal Bay Coastal Reserve"
},
{
"operation": "Highlighting",
"time": "29.191",
"value": "Shoal Bay Coastal Reserve_sentence_0"
},
{
"operation": "Highlighting",
"time": "48.27",
"value": "Shoal Bay Coastal Reserve_sentence_2"
},
{
"operation": "finish",
"time": "58.749",
"value": "finish"
}
]
}
|
27953
|
In the drama Romantic Repertoire, Sing-hong uses all kinds of tactics to try to make Kan Kit move out of his flat but eventually becomes distant to Kit when he found out about her interests.
|
{
"evidence": [
"Romantic Repertoire_sentence_9"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nIn the drama Romantic Repertoire, Sing-hong uses all kinds of tactics to try to make Kan Kit move out of his flat but eventually becomes distant to Kit when he found out about her interests.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Romantic Repertoire_sentence_9"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 1623,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.69",
"value": "?search="
},
{
"operation": "search",
"time": "9.721",
"value": "Romantic Repertoire"
},
{
"operation": "Now on",
"time": "10.624",
"value": "Romantic Repertoire"
},
{
"operation": "Page search",
"time": "19.543",
"value": "sing-hong"
},
{
"operation": "Highlighting",
"time": "77.192",
"value": "Romantic Repertoire_sentence_7"
},
{
"operation": "Highlighting",
"time": "167.098",
"value": "Romantic Repertoire_sentence_9"
},
{
"operation": "Highlighting deleted",
"time": "243.538",
"value": "Romantic Repertoire_sentence_7"
},
{
"operation": "finish",
"time": "278.764",
"value": "finish"
}
]
}
|
exfever_7891
|
Willie Hugh Nelson is an American musician, actor, and activist who has made several albums that were released in April 1975, including ten songs that consist entirely of pop standards that Nelson picked from among his favorites.
|
{
"evidence": [
"Willie Nelson",
"Stardust (Willie Nelson album)"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nWillie Hugh Nelson is an American musician, actor, and activist who has made several albums that were released in April 1975, including ten songs that consist entirely of pop standards that Nelson picked from among his favorites.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Willie Nelson",
"Stardust (Willie Nelson album)"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 1624,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
48673
|
KQAV station became KPXD and adopted a jazz format.
|
{
"evidence": [
"KQAV_sentence_0",
"KQAV_sentence_2"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nKQAV station became KPXD and adopted a jazz format.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"KQAV_sentence_0",
"KQAV_sentence_2"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 1625,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.549",
"value": "?search="
},
{
"operation": "search",
"time": "29.397",
"value": "Contains...KQAV station"
},
{
"operation": "Now on",
"time": "30.343",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "32.585",
"value": "KQAV"
},
{
"operation": "Now on",
"time": "33.622",
"value": "KQAV"
},
{
"operation": "Highlighting",
"time": "39.298",
"value": "KQAV_sentence_0"
},
{
"operation": "Page search",
"time": "54.88",
"value": "KPXD"
},
{
"operation": "Highlighting",
"time": "85.353",
"value": "KQAV_sentence_2"
},
{
"operation": "Page search",
"time": "106.808",
"value": "jaz"
},
{
"operation": "Highlighting deleted",
"time": "132.071",
"value": "KQAV_sentence_2"
},
{
"operation": "Highlighting",
"time": "133.019",
"value": "KQAV_sentence_2"
},
{
"operation": "search",
"time": "177.799",
"value": "\"KQAV\" \"jazz\""
},
{
"operation": "Now on",
"time": "178.886",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "184.442",
"value": "KMVE"
},
{
"operation": "Now on",
"time": "185.425",
"value": "KMVE"
},
{
"operation": "Page search",
"time": "190.654",
"value": "KQAV"
},
{
"operation": "Page search",
"time": "205.168",
"value": "jaz"
},
{
"operation": "finish",
"time": "236.538",
"value": "finish"
}
]
}
|
exfever_1222
|
Steven Gerrard is an English professional football manager and former player who manages the English professional football club that competes at the top level of the American football league system.
|
{
"evidence": [
"Steven Gerrard",
"Aston Villa F.C.",
"Premier League"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nSteven Gerrard is an English professional football manager and former player who manages the English professional football club that competes at the top level of the American football league system.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Steven Gerrard",
"Aston Villa F.C.",
"Premier League"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 1626,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
91357
|
Colin Bond from the Holden Dealer Team, driving a Holden LH Torana SL/R 5000 L34, has a qualification time of 2:27.4, followed by Allan Moffat of Allan Moffat Racing, driving a Ford XB Falcon GT Hardtop (produced by Ford Germany from 1973 to 1976), with a qualification time of 2:27.5.
|
{
"evidence": [
"Colin Bond_sentence_1",
"Colin Bond_sentence_19"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nColin Bond from the Holden Dealer Team, driving a Holden LH Torana SL/R 5000 L34, has a qualification time of 2:27.4, followed by Allan Moffat of Allan Moffat Racing, driving a Ford XB Falcon GT Hardtop (produced by Ford Germany from 1973 to 1976), with a qualification time of 2:27.5.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Colin Bond_sentence_1",
"Colin Bond_sentence_19"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 1627,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.167",
"value": "?search="
},
{
"operation": "search",
"time": "20.005",
"value": "Colin Bond"
},
{
"operation": "Now on",
"time": "21.932",
"value": "Colin Bond"
},
{
"operation": "Highlighting",
"time": "25.482",
"value": "Colin Bond_sentence_1"
},
{
"operation": "Highlighting",
"time": "91.02",
"value": "Colin Bond_sentence_19"
},
{
"operation": "hyperlink",
"time": "92.963",
"value": "Australian Manufacturers' Championship"
},
{
"operation": "Now on",
"time": "94.324",
"value": "1974 Australian Manufacturers' Championship"
},
{
"operation": "finish",
"time": "145.201",
"value": "finish"
}
]
}
|
56422
|
James Alexander Jermyn, a Methodist priest in the first half of the 20th century, was educated at the University of New Zealand and ordained deacon in 1915.
|
{
"evidence": [
"James Jermyn_sentence_0"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nJames Alexander Jermyn, a Methodist priest in the first half of the 20th century, was educated at the University of New Zealand and ordained deacon in 1915.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"James Jermyn_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 1628,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.706",
"value": "?search="
},
{
"operation": "search",
"time": "17.423",
"value": "James Alexander Jermyn"
},
{
"operation": "Now on",
"time": "17.966",
"value": "James Jermyn"
},
{
"operation": "Highlighting",
"time": "20.109",
"value": "James Jermyn_sentence_0"
},
{
"operation": "hyperlink",
"time": "25.133",
"value": "Anglican"
},
{
"operation": "Now on",
"time": "26.621",
"value": "Anglicanism"
},
{
"operation": "hyperlink",
"time": "68.17",
"value": "Anglican Church (disambiguation)"
},
{
"operation": "Now on",
"time": "68.588",
"value": "Anglican Church (disambiguation)"
},
{
"operation": "search",
"time": "78.318",
"value": "James Alexander Jermyn"
},
{
"operation": "Now on",
"time": "78.758",
"value": "James Alexander Jermyn"
},
{
"operation": "hyperlink",
"time": "95.63",
"value": "Anglican"
},
{
"operation": "Now on",
"time": "96.085",
"value": "Anglican"
},
{
"operation": "hyperlink",
"time": "108.172",
"value": "Methodism"
},
{
"operation": "Now on",
"time": "109.558",
"value": "Methodism"
},
{
"operation": "finish",
"time": "127.124",
"value": "finish"
}
]
}
|
exfever_4624
|
Bungie was established in May 1992 by Alex Seropian who later brought in programmer who co-founded the video game studio
|
{
"evidence": [
"Bungie",
"Jason Jones (programmer)"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nBungie was established in May 1992 by Alex Seropian who later brought in programmer who co-founded the video game studio\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Bungie",
"Jason Jones (programmer)"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 1629,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
51887
|
The first game of the 1998 Philadelphia Phillies season was a loss against the New York Mets.
|
{
"evidence": [
"1998 Philadelphia Phillies season_sentence_1",
"1998 Philadelphia Phillies season_sentence_3"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe first game of the 1998 Philadelphia Phillies season was a loss against the New York Mets.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"1998 Philadelphia Phillies season_sentence_1",
"1998 Philadelphia Phillies season_sentence_3"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 1630,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.774",
"value": "?search="
},
{
"operation": "search",
"time": "12.475",
"value": "1998 Philadelphia Phillies season"
},
{
"operation": "Now on",
"time": "14.514",
"value": "1998 Philadelphia Phillies season"
},
{
"operation": "Highlighting",
"time": "21.593",
"value": "1998 Philadelphia Phillies season_sentence_1"
},
{
"operation": "Highlighting",
"time": "36.248",
"value": "1998 Philadelphia Phillies season_sentence_3"
},
{
"operation": "finish",
"time": "51.169",
"value": "finish"
}
]
}
|
31693
|
During the Second World War, SS British Premier was used on convoy duties to bring oil from the Persian Gulf (a mediterranean sea in Western Asia) around the Cape of Good Hope and to the United Kingdom; then her final convoy was to have taken her from Abadan to Swansea.
|
{
"evidence": [
"SS British Premier_sentence_1",
"SS British Premier_sentence_2",
"Persian Gulf_sentence_5"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nDuring the Second World War, SS British Premier was used on convoy duties to bring oil from the Persian Gulf (a mediterranean sea in Western Asia) around the Cape of Good Hope and to the United Kingdom; then her final convoy was to have taken her from Abadan to Swansea.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"SS British Premier_sentence_1",
"SS British Premier_sentence_2",
"Persian Gulf_sentence_5"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 1631,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.663",
"value": "?search="
},
{
"operation": "search",
"time": "113.716",
"value": "SS British Premier"
},
{
"operation": "Now on",
"time": "114.673",
"value": "SS British Premier"
},
{
"operation": "Highlighting",
"time": "122.434",
"value": "SS British Premier_sentence_1"
},
{
"operation": "Highlighting",
"time": "134.204",
"value": "SS British Premier_sentence_2"
},
{
"operation": "hyperlink",
"time": "135.946",
"value": "Persian Gulf"
},
{
"operation": "Now on",
"time": "136.731",
"value": "Persian Gulf"
},
{
"operation": "Highlighting",
"time": "149.707",
"value": "Persian Gulf_sentence_5"
},
{
"operation": "finish",
"time": "157.925",
"value": "finish"
}
]
}
|
exfever_154
|
The AK-47 is a gas-operated assault rifle chambered for the 7.62×39mm cartridge, developed by Vladimir Lenin in the socialist and communist state that spanned much of Eurasia from 1922 to 1991.
|
{
"evidence": [
"AK-47",
"Soviet Union"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe AK-47 is a gas-operated assault rifle chambered for the 7.62×39mm cartridge, developed by Vladimir Lenin in the socialist and communist state that spanned much of Eurasia from 1922 to 1991.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"AK-47",
"Soviet Union"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 1632,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
61639
|
Gheorghe Bibescu (born 1804-1 June 1848) was a hospodar of Wallachia between 1 January 1843 – 25 June 1848.
|
{
"evidence": [
"Gheorghe Bibescu_sentence_0"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nGheorghe Bibescu (born 1804-1 June 1848) was a hospodar of Wallachia between 1 January 1843 – 25 June 1848.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Gheorghe Bibescu_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 1633,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.998",
"value": "?search="
},
{
"operation": "search",
"time": "8.856",
"value": "Gheorghe Bibescu"
},
{
"operation": "Now on",
"time": "9.958",
"value": "Gheorghe Bibescu"
},
{
"operation": "Highlighting",
"time": "16.822",
"value": "Gheorghe Bibescu_sentence_0"
},
{
"operation": "finish",
"time": "43.198",
"value": "finish"
}
]
}
|
exfever_6748
|
Jamaicans for Justice was founded in 1999 in the capital and largest city of an island country situated in the Caribbean Sea.
|
{
"evidence": [
"Jamaicans for Justice",
"Kingston, Jamaica",
"Jamaica"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nJamaicans for Justice was founded in 1999 in the capital and largest city of an island country situated in the Caribbean Sea.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Jamaicans for Justice",
"Kingston, Jamaica",
"Jamaica"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 1634,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_3426
|
Danny Elfman is an American screenwriter who has frequently worked with directors, including an American film producer.
|
{
"evidence": [
"Danny Elfman",
"Tim Burton"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nDanny Elfman is an American screenwriter who has frequently worked with directors, including an American film producer.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Danny Elfman",
"Tim Burton"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 1635,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
57858
|
Operation Choke Point was the focus of the hearing held by the House Financial Services Committee on April 2014 wherein four months after, a bill that limits the means of law enforcement to restrict access to banking system was introduced as an opposing response.
|
{
"evidence": [
"Operation Choke Point_sentence_22",
"Operation Choke Point_sentence_21"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nOperation Choke Point was the focus of the hearing held by the House Financial Services Committee on April 2014 wherein four months after, a bill that limits the means of law enforcement to restrict access to banking system was introduced as an opposing response.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Operation Choke Point_sentence_22",
"Operation Choke Point_sentence_21"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 1636,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.135",
"value": "?search="
},
{
"operation": "search",
"time": "8.874",
"value": "Operation Choke Point"
},
{
"operation": "Now on",
"time": "9.554",
"value": "Operation Choke Point"
},
{
"operation": "Highlighting",
"time": "35.535",
"value": "Operation Choke Point_sentence_22"
},
{
"operation": "Highlighting",
"time": "43.927",
"value": "Operation Choke Point_sentence_21"
},
{
"operation": "finish",
"time": "57.684",
"value": "finish"
}
]
}
|
exfever_238
|
James Garner played the leadimg role in Great Escape with an American actor.
|
{
"evidence": [
"James Garner",
"Steve McQueen"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nJames Garner played the leadimg role in Great Escape with an American actor.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"James Garner",
"Steve McQueen"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 1637,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
49681
|
Wilderich von Walderdorff, a member of the Walderdorff family, first studied in Würzburg, Speyer and the Collegium Germanicum in Rome and was Vicar-General of Mainz from 1647 to 1669.
|
{
"evidence": [
"Wilderich von Walderdorff_sentence_1",
"Wilderich von Walderdorff_sentence_5",
"Wilderich von Walderdorff_sentence_6"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nWilderich von Walderdorff, a member of the Walderdorff family, first studied in Würzburg, Speyer and the Collegium Germanicum in Rome and was Vicar-General of Mainz from 1647 to 1669.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Wilderich von Walderdorff_sentence_1",
"Wilderich von Walderdorff_sentence_5",
"Wilderich von Walderdorff_sentence_6"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 1638,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.378",
"value": "?search="
},
{
"operation": "Now on",
"time": "12.423",
"value": "?search="
},
{
"operation": "search",
"time": "20.044",
"value": "Wilderich von Walderdorff"
},
{
"operation": "Now on",
"time": "21.069",
"value": "Wilderich von Walderdorff"
},
{
"operation": "Highlighting",
"time": "28.624",
"value": "Wilderich von Walderdorff_sentence_1"
},
{
"operation": "Highlighting",
"time": "52.508",
"value": "Wilderich von Walderdorff_sentence_5"
},
{
"operation": "Highlighting",
"time": "61.36",
"value": "Wilderich von Walderdorff_sentence_6"
},
{
"operation": "finish",
"time": "94.443",
"value": "finish"
}
]
}
|
exfever_7265
|
Friends with Benefits is a 2011 film that features an award-winning American actress.
|
{
"evidence": [
"Friends with Benefits (film)",
"Patricia Clarkson"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nFriends with Benefits is a 2011 film that features an award-winning American actress.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Friends with Benefits (film)",
"Patricia Clarkson"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 1639,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
47071
|
Kristjánsson took charge of Icelandic premier league team Fram on June 30, 2004, replacing Ion Geolgău (a former Romanian football midfielder and manager); then he was in charge again for the 2005 season, but left the club after relegation.
|
{
"evidence": [
"Ólafur Kristjánsson_sentence_6",
"Ion Geolgău_sentence_0",
"Ólafur Kristjánsson_sentence_8"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nKristjánsson took charge of Icelandic premier league team Fram on June 30, 2004, replacing Ion Geolgău (a former Romanian football midfielder and manager); then he was in charge again for the 2005 season, but left the club after relegation.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"Ólafur Kristjánsson_sentence_6",
"Ion Geolgău_sentence_0",
"Ólafur Kristjánsson_sentence_8"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 1640,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.209",
"value": "?search="
},
{
"operation": "search",
"time": "16.847",
"value": "Kristjánsson"
},
{
"operation": "Now on",
"time": "17.735",
"value": "Kristjánsson"
},
{
"operation": "search",
"time": "41.777",
"value": "Fram"
},
{
"operation": "Now on",
"time": "42.666",
"value": "Fram"
},
{
"operation": "hyperlink",
"time": "47.702",
"value": "Fram (disambiguation)"
},
{
"operation": "Now on",
"time": "48.208",
"value": "Fram (disambiguation)"
},
{
"operation": "hyperlink",
"time": "66.974",
"value": "IF Fram Tórshavn"
},
{
"operation": "Now on",
"time": "67.482",
"value": "FC Hoyvík"
},
{
"operation": "back-button-clicked",
"time": "69.534",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "69.599",
"value": "Fram (disambiguation)"
},
{
"operation": "search",
"time": "89.637",
"value": "Ólafur Kristjánsson"
},
{
"operation": "Now on",
"time": "90.35",
"value": "Ólafur Kristjánsson"
},
{
"operation": "Highlighting",
"time": "96.215",
"value": "Ólafur Kristjánsson_sentence_6"
},
{
"operation": "hyperlink",
"time": "170.946",
"value": "Ion Geolgău"
},
{
"operation": "Now on",
"time": "171.497",
"value": "Ion Geolgău"
},
{
"operation": "Highlighting",
"time": "172.679",
"value": "Ion Geolgău_sentence_0"
},
{
"operation": "back-button-clicked",
"time": "174.442",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "174.503",
"value": "Ólafur Kristjánsson"
},
{
"operation": "Highlighting",
"time": "179.689",
"value": "Ólafur Kristjánsson_sentence_8"
},
{
"operation": "search",
"time": "192.882",
"value": "Kristjánsson"
},
{
"operation": "Now on",
"time": "193.557",
"value": "Kristjánsson"
},
{
"operation": "finish",
"time": "210.613",
"value": "finish"
}
]
}
|
37629
|
The village has a total number of 382 houses and the population of 2233 of which 1154 are males while 1179 are females according to the report published by Census India in 2011 and the literacy rate of the village is 77.53%.
|
{
"evidence": [
"2011 Census of India_sentence_8"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe village has a total number of 382 houses and the population of 2233 of which 1154 are males while 1179 are females according to the report published by Census India in 2011 and the literacy rate of the village is 77.53%.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"2011 Census of India_sentence_8"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 1641,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.684",
"value": "?search="
},
{
"operation": "search",
"time": "235.479",
"value": "Census India in 2011"
},
{
"operation": "Now on",
"time": "237.531",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "245.66",
"value": "2011 Census of India"
},
{
"operation": "Now on",
"time": "246.561",
"value": "2011 Census of India"
},
{
"operation": "Page search",
"time": "251.983",
"value": "village"
},
{
"operation": "Page search",
"time": "267.103",
"value": "village"
},
{
"operation": "Highlighting",
"time": "273.78",
"value": "2011 Census of India_sentence_8"
},
{
"operation": "finish",
"time": "276.838",
"value": "finish"
}
]
}
|
11355
|
Cristina Luca Boico moved to Bucharest to study medicine after completing her secondary education, though she joined the Student’s Democratic Front, an organization aligned with the Communist Party, she didn’t have a problem graduating.
|
{
"evidence": [
"Cristina Luca Boico_sentence_11"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nCristina Luca Boico moved to Bucharest to study medicine after completing her secondary education, though she joined the Student’s Democratic Front, an organization aligned with the Communist Party, she didn’t have a problem graduating.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Cristina Luca Boico_sentence_11"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 1642,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.527",
"value": "?search="
},
{
"operation": "search",
"time": "8.509",
"value": "Cristina Luca Boico"
},
{
"operation": "Now on",
"time": "9.352",
"value": "Cristina Luca Boico"
},
{
"operation": "Page search",
"time": "12.742",
"value": "Bucharest"
},
{
"operation": "Highlighting",
"time": "21.83",
"value": "Cristina Luca Boico_sentence_8"
},
{
"operation": "Highlighting",
"time": "41.851",
"value": "Cristina Luca Boico_sentence_9"
},
{
"operation": "Highlighting",
"time": "77.27",
"value": "Cristina Luca Boico_sentence_11"
},
{
"operation": "Highlighting deleted",
"time": "122.395",
"value": "Cristina Luca Boico_sentence_9"
},
{
"operation": "Highlighting deleted",
"time": "123.429",
"value": "Cristina Luca Boico_sentence_8"
},
{
"operation": "finish",
"time": "134.508",
"value": "finish"
}
]
}
|
exfever_3870
|
The Vampire Diaries is an American fantasy teen drama television series that was developed by an American screenwriter, director, and producer best known for developing and writing the slasher film Scream.
|
{
"evidence": [
"The Vampire Diaries",
"Kevin Williamson (screenwriter)"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Vampire Diaries is an American fantasy teen drama television series that was developed by an American screenwriter, director, and producer best known for developing and writing the slasher film Scream.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Vampire Diaries",
"Kevin Williamson (screenwriter)"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 1643,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
18854
|
Amita is a name both used for a brand of fruit juice and for a health system called AMITA Health which is a joint operating company formed between Alexian Brothers Health System ad Adventist Midwest Health..
|
{
"evidence": [
"Amita_sentence_1",
"Amita_sentence_2",
"AMITA Health_sentence_2"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nAmita is a name both used for a brand of fruit juice and for a health system called AMITA Health which is a joint operating company formed between Alexian Brothers Health System ad Adventist Midwest Health..\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Amita_sentence_1",
"Amita_sentence_2",
"AMITA Health_sentence_2"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 1644,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.721",
"value": "?search="
},
{
"operation": "search",
"time": "9.822",
"value": "Amita"
},
{
"operation": "Now on",
"time": "10.577",
"value": "Amita"
},
{
"operation": "Highlighting",
"time": "15.286",
"value": "Amita_sentence_1"
},
{
"operation": "Highlighting",
"time": "17.728",
"value": "Amita_sentence_2"
},
{
"operation": "hyperlink",
"time": "22.679",
"value": "AMITA Health"
},
{
"operation": "Now on",
"time": "23.098",
"value": "AMITA Health"
},
{
"operation": "Highlighting",
"time": "26.026",
"value": "AMITA Health_sentence_2"
},
{
"operation": "finish",
"time": "74.222",
"value": "finish"
}
]
}
|
51328
|
Mika Häkkinen was the champion of the 1999 German Grand Prix with a time of 1:31:18.785; followed by Michael Schumacher with a time of 1:31:23.8.
|
{
"evidence": [
"1999 German Grand Prix_sentence_0",
"Mika Häkkinen_sentence_1"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nMika Häkkinen was the champion of the 1999 German Grand Prix with a time of 1:31:18.785; followed by Michael Schumacher with a time of 1:31:23.8.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"1999 German Grand Prix_sentence_0",
"Mika Häkkinen_sentence_1"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 1645,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.475",
"value": "?search="
},
{
"operation": "search",
"time": "38.892",
"value": "Mika Hakkinen"
},
{
"operation": "Now on",
"time": "42.602",
"value": "Mika Häkkinen"
},
{
"operation": "Highlighting",
"time": "218.286",
"value": "Mika Häkkinen_sentence_67"
},
{
"operation": "Highlighting",
"time": "218.294",
"value": "Mika Häkkinen_section_1"
},
{
"operation": "Highlighting",
"time": "218.296",
"value": "Mika Häkkinen_section_2"
},
{
"operation": "Highlighting",
"time": "218.296",
"value": "Mika Häkkinen_section_4"
},
{
"operation": "hyperlink",
"time": "262.541",
"value": "German Grand Prix"
},
{
"operation": "Now on",
"time": "264.353",
"value": "1992 German Grand Prix"
},
{
"operation": "search",
"time": "358.396",
"value": "1999 German Grand Prix"
},
{
"operation": "Now on",
"time": "359.796",
"value": "1999 German Grand Prix"
},
{
"operation": "Highlighting",
"time": "370.888",
"value": "1999 German Grand Prix_sentence_0"
},
{
"operation": "Highlighting",
"time": "399.037",
"value": "1999 German Grand Prix_sentence_4"
},
{
"operation": "Highlighting deleted",
"time": "419.867",
"value": "1999 German Grand Prix_sentence_0"
},
{
"operation": "Highlighting",
"time": "587.615",
"value": "1999 German Grand Prix_sentence_0"
},
{
"operation": "hyperlink",
"time": "600.47",
"value": "Mika Häkkinen"
},
{
"operation": "Now on",
"time": "603.412",
"value": "Mika Häkkinen"
},
{
"operation": "Highlighting",
"time": "610.151",
"value": "Mika Häkkinen_sentence_1"
},
{
"operation": "hyperlink",
"time": "721.843",
"value": "German Grand Prix"
},
{
"operation": "Now on",
"time": "722.793",
"value": "1999 German Grand Prix"
},
{
"operation": "Highlighting",
"time": "806.67",
"value": "1999 German Grand Prix_cell_0_5_5"
},
{
"operation": "Highlighting",
"time": "806.674",
"value": "1999 German Grand Prix_section_0"
},
{
"operation": "Highlighting",
"time": "806.675",
"value": "1999 German Grand Prix_section_2"
},
{
"operation": "Highlighting",
"time": "806.718",
"value": "1999 German Grand Prix_header_cell_0_5_0"
},
{
"operation": "Highlighting",
"time": "806.719",
"value": "1999 German Grand Prix_header_cell_0_0_5"
},
{
"operation": "Highlighting deleted",
"time": "829.007",
"value": "1999 German Grand Prix_cell_0_5_5"
},
{
"operation": "Highlighting deleted",
"time": "829.011",
"value": "1999 German Grand Prix_section_0"
},
{
"operation": "Highlighting deleted",
"time": "829.012",
"value": "1999 German Grand Prix_section_2"
},
{
"operation": "Highlighting deleted",
"time": "829.056",
"value": "1999 German Grand Prix_header_cell_0_5_0"
},
{
"operation": "Highlighting deleted",
"time": "829.058",
"value": "1999 German Grand Prix_header_cell_0_0_5"
},
{
"operation": "finish",
"time": "1112.513",
"value": "finish"
}
]
}
|
exfever_1156
|
Bosnia and Herzegovina, known informally as Bosnia, is a country at the crossroads of the south and southeast in a continent that is also recognized as a part of Eurasia that shares the continental landmass of Afro-Eurasia with both Asia and the world's second-largest and second-most-populous continent, after Asia in both cases.
|
{
"evidence": [
"Bosnia and Herzegovina",
"Europe",
"Africa"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nBosnia and Herzegovina, known informally as Bosnia, is a country at the crossroads of the south and southeast in a continent that is also recognized as a part of Eurasia that shares the continental landmass of Afro-Eurasia with both Asia and the world's second-largest and second-most-populous continent, after Asia in both cases.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Bosnia and Herzegovina",
"Europe",
"Africa"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 1646,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
26396
|
Green consumption (related to notions of sustainable consumer behaviour) is a concept that ascribes to consumers co-responsibility for addressing environmental problems, with American "green" brands beginning to appear in the 1970s and exploding on the American market until the 1990s.
|
{
"evidence": [
"Green consumption_sentence_4"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nGreen consumption (related to notions of sustainable consumer behaviour) is a concept that ascribes to consumers co-responsibility for addressing environmental problems, with American \"green\" brands beginning to appear in the 1970s and exploding on the American market until the 1990s.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Green consumption_sentence_4"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 1647,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.488",
"value": "undefined"
},
{
"operation": "search",
"time": "9.031",
"value": "Green consumption"
},
{
"operation": "Now on",
"time": "9.769",
"value": "Green consumption"
},
{
"operation": "Highlighting",
"time": "42.584",
"value": "Green consumption_sentence_0"
},
{
"operation": "Highlighting deleted",
"time": "42.791",
"value": "Green consumption_sentence_0"
},
{
"operation": "Highlighting",
"time": "43.206",
"value": "Green consumption_sentence_0"
},
{
"operation": "Highlighting",
"time": "51.813",
"value": "Green consumption_sentence_2"
},
{
"operation": "Highlighting",
"time": "68.51",
"value": "Green consumption_sentence_4"
},
{
"operation": "Highlighting deleted",
"time": "71.241",
"value": "Green consumption_sentence_4"
},
{
"operation": "Highlighting",
"time": "74.491",
"value": "Green consumption_sentence_4"
},
{
"operation": "Highlighting deleted",
"time": "128.177",
"value": "Green consumption_sentence_0"
},
{
"operation": "Highlighting deleted",
"time": "129.182",
"value": "Green consumption_sentence_2"
},
{
"operation": "finish",
"time": "144.72",
"value": "finish"
}
]
}
|
38378
|
With the support of José María Farfán and Cristina Farfán, Rita Cetina Gutiérrez, who wrote poetry from about 1860 which was published in many local journals and newspapers, opened La Siempreviva (Everlasting), an institution consisting of Mexico's first secular school for poor girls as well as an art college for young women.
|
{
"evidence": [
"Rita Cetina Gutiérrez_sentence_8"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nWith the support of José María Farfán and Cristina Farfán, Rita Cetina Gutiérrez, who wrote poetry from about 1860 which was published in many local journals and newspapers, opened La Siempreviva (Everlasting), an institution consisting of Mexico's first secular school for poor girls as well as an art college for young women.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Rita Cetina Gutiérrez_sentence_8"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 1648,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.698",
"value": "?search="
},
{
"operation": "search",
"time": "9.132",
"value": "Rita Cetina Gutiérrez"
},
{
"operation": "Now on",
"time": "10.172",
"value": "Rita Cetina Gutiérrez"
},
{
"operation": "Highlighting",
"time": "14.627",
"value": "Rita Cetina Gutiérrez_sentence_6"
},
{
"operation": "Highlighting",
"time": "32.1",
"value": "Rita Cetina Gutiérrez_sentence_8"
},
{
"operation": "finish",
"time": "38.86",
"value": "finish"
}
]
}
|
exfever_5310
|
Italian is a language that evolved from the Vulgar Latin of the post-Republican period of Roman civilization.
|
{
"evidence": [
"Italian language",
"Ancient Rome"
],
"label": "NOT ENOUGH INFO"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nItalian is a language that evolved from the Vulgar Latin of the post-Republican period of Roman civilization.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Italian language",
"Ancient Rome"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 1649,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7208
|
This Is Us is an American family drama television series that was piloted on an American English-language commercial radio network.?
|
{
"evidence": [
"This Is Us (TV series)",
"NBC"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThis Is Us is an American family drama television series that was piloted on an American English-language commercial radio network.?\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"This Is Us (TV series)",
"NBC"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 1650,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_1694
|
The Golden Girls is an American sitcom aired on NBC, starring Bea Arthur, Betty White, Estelle Getty, and an American actress and comedian best known for her role as Vivian Harmon on Maude.
|
{
"evidence": [
"The Golden Girls",
"Rue McClanahan"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Golden Girls is an American sitcom aired on NBC, starring Bea Arthur, Betty White, Estelle Getty, and an American actress and comedian best known for her role as Vivian Harmon on Maude.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Golden Girls",
"Rue McClanahan"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 1651,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
87169
|
In 2001, Gulben Ergen played Melek in Dadi and won a Best Actress award at the Golden Butterflys; the next year, a lead single from her album, Sade v Sadece received an award at the Turkey Music Awards, which used to be one of the most prestigious awards in Turkey until news emerged about winners being changed to favor relatives.
|
{
"evidence": [
"Gülben Ergen_sentence_3",
"Gülben Ergen_sentence_4",
"Kral Turkey Music Awards_sentence_8"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nIn 2001, Gulben Ergen played Melek in Dadi and won a Best Actress award at the Golden Butterflys; the next year, a lead single from her album, Sade v Sadece received an award at the Turkey Music Awards, which used to be one of the most prestigious awards in Turkey until news emerged about winners being changed to favor relatives.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Gülben Ergen_sentence_3",
"Gülben Ergen_sentence_4",
"Kral Turkey Music Awards_sentence_8"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 1652,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.179",
"value": "?search="
},
{
"operation": "search",
"time": "140.004",
"value": "Gulben Ergen"
},
{
"operation": "Now on",
"time": "141.244",
"value": "Gülben Ergen"
},
{
"operation": "Highlighting",
"time": "150.183",
"value": "Gülben Ergen_cell_1_11_1"
},
{
"operation": "Highlighting",
"time": "150.184",
"value": "Gülben Ergen_section_14"
},
{
"operation": "Highlighting",
"time": "150.186",
"value": "Gülben Ergen_header_cell_1_0_1"
},
{
"operation": "Highlighting",
"time": "150.661",
"value": "Gülben Ergen_cell_1_11_2"
},
{
"operation": "Highlighting",
"time": "150.668",
"value": "Gülben Ergen_header_cell_1_0_2"
},
{
"operation": "Highlighting",
"time": "152.197",
"value": "Gülben Ergen_cell_1_11_0"
},
{
"operation": "Highlighting",
"time": "152.201",
"value": "Gülben Ergen_header_cell_1_0_0"
},
{
"operation": "Highlighting",
"time": "157.927",
"value": "Gülben Ergen_sentence_3"
},
{
"operation": "Highlighting",
"time": "169.185",
"value": "Gülben Ergen_sentence_4"
},
{
"operation": "hyperlink",
"time": "179.222",
"value": "Turkey Music Awards"
},
{
"operation": "Now on",
"time": "180.27",
"value": "Kral Turkey Music Awards"
},
{
"operation": "Highlighting",
"time": "190.976",
"value": "Kral Turkey Music Awards_sentence_8"
},
{
"operation": "finish",
"time": "200.098",
"value": "finish"
}
]
}
|
exfever_1934
|
Nerrukku Ner is a 1997 Indian Tamil-language action rom-com film starring Suriya and an Indian actor, dancer, playback singer, and philanthropist who works predominantly in Tamil cinema.
|
{
"evidence": [
"Nerrukku Ner",
"Vijay (actor)"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nNerrukku Ner is a 1997 Indian Tamil-language action rom-com film starring Suriya and an Indian actor, dancer, playback singer, and philanthropist who works predominantly in Tamil cinema.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Nerrukku Ner",
"Vijay (actor)"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 1653,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
93022
|
Edward English was an English cricketer which is a bat-and-ball game played between two teams of eleven players on a field at the ventre of which is 22-yard pitch a wicket at each end, each comprising two bails balanced on three stumps.
|
{
"evidence": [
"Edward English_sentence_0",
"Cricket_sentence_3"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nEdward English was an English cricketer which is a bat-and-ball game played between two teams of eleven players on a field at the ventre of which is 22-yard pitch a wicket at each end, each comprising two bails balanced on three stumps.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Edward English_sentence_0",
"Cricket_sentence_3"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 1654,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.201",
"value": "?search="
},
{
"operation": "search",
"time": "42.42",
"value": "Edward English"
},
{
"operation": "Now on",
"time": "43.578",
"value": "Edward English"
},
{
"operation": "Highlighting",
"time": "52.184",
"value": "Edward English_sentence_0"
},
{
"operation": "hyperlink",
"time": "104.464",
"value": "cricketer"
},
{
"operation": "Now on",
"time": "106.214",
"value": "Cricket"
},
{
"operation": "Highlighting",
"time": "127.986",
"value": "Cricket_sentence_3"
},
{
"operation": "finish",
"time": "226.654",
"value": "finish"
}
]
}
|
exfever_859
|
Joni Mitchell is a Canadian-American singer-songwriter who has received numerous awards, including an induction into the Rock Hall, a museum and hall of fame located in downtown Columbus, Ohio.
|
{
"evidence": [
"Joni Mitchell",
"Rock and Roll Hall of Fame"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nJoni Mitchell is a Canadian-American singer-songwriter who has received numerous awards, including an induction into the Rock Hall, a museum and hall of fame located in downtown Columbus, Ohio.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Joni Mitchell",
"Rock and Roll Hall of Fame"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 1655,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_6054
|
Mr. Nobody is a 2009 science fiction drama film starring a Welsh actor and musician best known for his role in a 2004 British psychological thriller film.
|
{
"evidence": [
"Mr. Nobody (film)",
"Rhys Ifans",
"Enduring Love (film)"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nMr. Nobody is a 2009 science fiction drama film starring a Welsh actor and musician best known for his role in a 2004 British psychological thriller film.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Mr. Nobody (film)",
"Rhys Ifans",
"Enduring Love (film)"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 1656,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_6331
|
Dennis Quaid is an American actor whose notable credits include a 1983 American epic historical drama film that is based on the 1979 book by Tom Wolfe.
|
{
"evidence": [
"Dennis Quaid",
"The Right Stuff (film)",
"The Right Stuff (book)"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nDennis Quaid is an American actor whose notable credits include a 1983 American epic historical drama film that is based on the 1979 book by Tom Wolfe.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Dennis Quaid",
"The Right Stuff (film)",
"The Right Stuff (book)"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 1657,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_247
|
The Black Panther is the nineteenth fictional character in an American comic book published by Marvel Comics.
|
{
"evidence": [
"Black Panther (comics)",
"Africa"
],
"label": "NOT ENOUGH INFO"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Black Panther is the nineteenth fictional character in an American comic book published by Marvel Comics.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Black Panther (comics)",
"Africa"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 1658,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_6247
|
John Nash was a prominent British architect who designed the London royal residence and administrative headquarters of the United Kingdom's monarch, which is located in the City of Westminster.
|
{
"evidence": [
"John Nash (architect)",
"Buckingham Palace"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nJohn Nash was a prominent British architect who designed the London royal residence and administrative headquarters of the United Kingdom's monarch, which is located in the City of Westminster.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"John Nash (architect)",
"Buckingham Palace"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 1659,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
32976
|
Ingar Solty, a German political writer in neither Das Argument nor Süddeutsche Zeitung and journalist focusing on marxism, socialism, received his PhD from York University in Toronto, Canada.
|
{
"evidence": [
"Ingar Solty_sentence_0",
"Ingar Solty_sentence_4",
"Ingar Solty_sentence_1"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nIngar Solty, a German political writer in neither Das Argument nor Süddeutsche Zeitung and journalist focusing on marxism, socialism, received his PhD from York University in Toronto, Canada.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"Ingar Solty_sentence_0",
"Ingar Solty_sentence_4",
"Ingar Solty_sentence_1"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 1660,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.822",
"value": "?search="
},
{
"operation": "search",
"time": "13.88",
"value": "Ingar Solty"
},
{
"operation": "Now on",
"time": "14.785",
"value": "Ingar Solty"
},
{
"operation": "Highlighting",
"time": "17.312",
"value": "Ingar Solty_sentence_0"
},
{
"operation": "Highlighting",
"time": "49.564",
"value": "Ingar Solty_sentence_4"
},
{
"operation": "Highlighting",
"time": "83.785",
"value": "Ingar Solty_sentence_1"
},
{
"operation": "finish",
"time": "112.779",
"value": "finish"
}
]
}
|
92295
|
30th New York State Legislature ( the Southern District is the stronghold of the Democratic Party for 17 years while the Eastern District was the stronghold of the Republican Party for 25 years.
|
{
"evidence": [
"30th New York State Legislature_sentence_2"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\n30th New York State Legislature ( the Southern District is the stronghold of the Democratic Party for 17 years while the Eastern District was the stronghold of the Republican Party for 25 years.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"30th New York State Legislature_sentence_2"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 1661,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.097",
"value": "?search="
},
{
"operation": "search",
"time": "11.585",
"value": "30th New York State Legislature"
},
{
"operation": "Now on",
"time": "14.058",
"value": "30th New York State Legislature"
},
{
"operation": "Highlighting",
"time": "72.626",
"value": "30th New York State Legislature_sentence_4"
},
{
"operation": "Highlighting deleted",
"time": "74.619",
"value": "30th New York State Legislature_sentence_4"
},
{
"operation": "Page search",
"time": "85.311",
"value": "Southern District"
},
{
"operation": "search",
"time": "196.51",
"value": "southern District is the stronghold of the Democratic Party for 17 years"
},
{
"operation": "Now on",
"time": "202.44",
"value": "undefined"
},
{
"operation": "search",
"time": "245.702",
"value": "Southern District is the stronghold of the Democratic"
},
{
"operation": "Now on",
"time": "247.11",
"value": "undefined"
},
{
"operation": "search",
"time": "477.685",
"value": "30th New York State Legislature"
},
{
"operation": "Now on",
"time": "478.41",
"value": "30th New York State Legislature"
},
{
"operation": "search",
"time": "592.115",
"value": "Southern District of New York"
},
{
"operation": "Now on",
"time": "597.062",
"value": "United States District Court for the Southern District of New York"
},
{
"operation": "hyperlink",
"time": "610.918",
"value": "2 History United States District Court for the Southern District of New York_section_1"
},
{
"operation": "Page search",
"time": "665.586",
"value": "stronghold"
},
{
"operation": "hyperlink",
"time": "685.958",
"value": "Northern"
},
{
"operation": "Now on",
"time": "687.976",
"value": "United States District Court for the Northern District of New York"
},
{
"operation": "search",
"time": "708.978",
"value": "southern district"
},
{
"operation": "Now on",
"time": "709.805",
"value": "Southern District"
},
{
"operation": "search",
"time": "753.914",
"value": "30th New York State Legislature southern district"
},
{
"operation": "Now on",
"time": "757.048",
"value": "undefined"
},
{
"operation": "search",
"time": "767.472",
"value": "30th New York State Legislature"
},
{
"operation": "Now on",
"time": "768.282",
"value": "30th New York State Legislature"
},
{
"operation": "hyperlink",
"time": "798.453",
"value": "4.1 Districts 30th New York State Legislature_section_4"
},
{
"operation": "Highlighting",
"time": "800.613",
"value": "30th New York State Legislature_item_0_0"
},
{
"operation": "Highlighting",
"time": "800.615",
"value": "30th New York State Legislature_section_3"
},
{
"operation": "Highlighting",
"time": "800.615",
"value": "30th New York State Legislature_section_4"
},
{
"operation": "Highlighting",
"time": "860.486",
"value": "30th New York State Legislature_sentence_2"
},
{
"operation": "hyperlink",
"time": "873.037",
"value": "Democratic-Republicans"
},
{
"operation": "Now on",
"time": "874.865",
"value": "Democratic-Republican Party"
},
{
"operation": "Page search",
"time": "931.308",
"value": "district"
},
{
"operation": "finish",
"time": "965.22",
"value": "finish"
}
]
}
|
exfever_4215
|
Oz the Great and Powerful is a 2013 American fantasy film with a form of adventure fiction and a genre of film.
|
{
"evidence": [
"Oz the Great and Powerful",
"Adventure film"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nOz the Great and Powerful is a 2013 American fantasy film with a form of adventure fiction and a genre of film.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Oz the Great and Powerful",
"Adventure film"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 1662,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_1305
|
Lisa Kudrow is an American actress who is popularly known for her role as one of the eight main characters in the American sitcom Friends, created by an American television writer known for her work on the HBO series Dream On.
|
{
"evidence": [
"Lisa Kudrow",
"Phoebe Buffay",
"Marta Kauffman"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nLisa Kudrow is an American actress who is popularly known for her role as one of the eight main characters in the American sitcom Friends, created by an American television writer known for her work on the HBO series Dream On.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Lisa Kudrow",
"Phoebe Buffay",
"Marta Kauffman"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 1663,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
67221
|
The Worldreader app, launched in January 2012 and available on Microsoft Windows Store and Google's Play Store, provides worldwide access to books, educational resources and health information to people with mobile phones.
|
{
"evidence": [
"Worldreader_sentence_12"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Worldreader app, launched in January 2012 and available on Microsoft Windows Store and Google's Play Store, provides worldwide access to books, educational resources and health information to people with mobile phones.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Worldreader_sentence_12"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 1664,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.134",
"value": "undefined"
},
{
"operation": "search",
"time": "6.954",
"value": "Worldreader"
},
{
"operation": "Now on",
"time": "7.769",
"value": "Worldreader"
},
{
"operation": "Highlighting",
"time": "36.357",
"value": "Worldreader_sentence_12"
},
{
"operation": "finish",
"time": "45.9",
"value": "finish"
}
]
}
|
87442
|
In Amanush 2 (music by Arindam Chatterjee who has 17 songs) Raghu's problems begin when his father commits suicide and he seeks revenge, when he is released from juvie he becomes a medical student.
|
{
"evidence": [
"Amanush 2_sentence_26",
"Amanush 2_sentence_6",
"Amanush 2_sentence_7",
"Amanush 2_sentence_8",
"Amanush 2_sentence_10"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nIn Amanush 2 (music by Arindam Chatterjee who has 17 songs) Raghu's problems begin when his father commits suicide and he seeks revenge, when he is released from juvie he becomes a medical student.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Amanush 2_sentence_26",
"Amanush 2_sentence_6",
"Amanush 2_sentence_7",
"Amanush 2_sentence_8",
"Amanush 2_sentence_10"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 1665,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.128",
"value": "?search="
},
{
"operation": "search",
"time": "134.482",
"value": "Amanush 2"
},
{
"operation": "Now on",
"time": "135.441",
"value": "Amanush 2"
},
{
"operation": "hyperlink",
"time": "141.37",
"value": "5 Music Amanush 2_section_6"
},
{
"operation": "Highlighting",
"time": "145.993",
"value": "Amanush 2_sentence_26"
},
{
"operation": "search",
"time": "205.129",
"value": "Arindam Chatterjee"
},
{
"operation": "Now on",
"time": "206.131",
"value": "Arindam Chatterjee"
},
{
"operation": "search",
"time": "257.537",
"value": "Contains...Amanush 2"
},
{
"operation": "Now on",
"time": "258.957",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "265.833",
"value": "Soundtrack Amanush_section_2"
},
{
"operation": "Now on",
"time": "266.528",
"value": "Amanush#Soundtrack Amanush section 2"
},
{
"operation": "back-button-clicked",
"time": "274.877",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "275.602",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "327.712",
"value": "List of Bengali films of 2015"
},
{
"operation": "Now on",
"time": "329.57",
"value": "List of Bengali films of 2015"
},
{
"operation": "back-button-clicked",
"time": "343.274",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "344.03",
"value": "undefined"
},
{
"operation": "search",
"time": "372.363",
"value": "AMANUSH 2 SONGS"
},
{
"operation": "Now on",
"time": "373.34",
"value": "undefined"
},
{
"operation": "search",
"time": "431.667",
"value": "Contains...Arindam Chatterjee"
},
{
"operation": "Now on",
"time": "432.485",
"value": "undefined"
},
{
"operation": "search",
"time": "479.44",
"value": "Amanush 2"
},
{
"operation": "Now on",
"time": "480.46",
"value": "Amanush 2"
},
{
"operation": "Highlighting",
"time": "486.358",
"value": "Amanush 2_sentence_6"
},
{
"operation": "Highlighting",
"time": "490.574",
"value": "Amanush 2_sentence_7"
},
{
"operation": "Highlighting",
"time": "504.104",
"value": "Amanush 2_sentence_8"
},
{
"operation": "Highlighting",
"time": "519.1",
"value": "Amanush 2_sentence_10"
},
{
"operation": "finish",
"time": "589.953",
"value": "finish"
}
]
}
|
48424
|
Like previous games, virtual avatars of Taylor Swift, Adam Levine, and the band No Doubt are presented in Band Hero, which received mixed reviews from journalists.
|
{
"evidence": [
"Band Hero_sentence_3",
"Band Hero_sentence_4"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nLike previous games, virtual avatars of Taylor Swift, Adam Levine, and the band No Doubt are presented in Band Hero, which received mixed reviews from journalists.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Band Hero_sentence_3",
"Band Hero_sentence_4"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 1666,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.969",
"value": "?search="
},
{
"operation": "search",
"time": "24.499",
"value": "Band Hero"
},
{
"operation": "Now on",
"time": "26.41",
"value": "Band Hero"
},
{
"operation": "Page search",
"time": "35.013",
"value": "swift"
},
{
"operation": "Highlighting",
"time": "36.822",
"value": "Band Hero_sentence_3"
},
{
"operation": "Highlighting",
"time": "48.474",
"value": "Band Hero_sentence_4"
},
{
"operation": "finish",
"time": "54.272",
"value": "finish"
}
]
}
|
72307
|
Ustyansky District was subordinated to Northern Krai, then the krai was transformed into Northern Oblast.
|
{
"evidence": [
"Ustyansky District_sentence_25",
"Ustyansky District_sentence_26"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nUstyansky District was subordinated to Northern Krai, then the krai was transformed into Northern Oblast.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Ustyansky District_sentence_25",
"Ustyansky District_sentence_26"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 1667,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.841",
"value": "undefined"
},
{
"operation": "search",
"time": "9.172",
"value": "Ustyansky District"
},
{
"operation": "Now on",
"time": "10.562",
"value": "Ustyansky District"
},
{
"operation": "Page search",
"time": "19.066",
"value": "Krai"
},
{
"operation": "Page search",
"time": "19.891",
"value": "Krai"
},
{
"operation": "Highlighting",
"time": "35.743",
"value": "Ustyansky District_sentence_25"
},
{
"operation": "Highlighting",
"time": "46.239",
"value": "Ustyansky District_sentence_26"
},
{
"operation": "finish",
"time": "52.388",
"value": "finish"
}
]
}
|
7918
|
Richard Corish ran as a political member at the 1922 general election and later on served in the Dáil.
|
{
"evidence": [
"Richard Corish_sentence_5",
"Richard Corish_sentence_6"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nRichard Corish ran as a political member at the 1922 general election and later on served in the Dáil.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Richard Corish_sentence_5",
"Richard Corish_sentence_6"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 1668,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.481",
"value": "?search="
},
{
"operation": "search",
"time": "10.132",
"value": "richard corish"
},
{
"operation": "Now on",
"time": "11.454",
"value": "Richard Corish"
},
{
"operation": "Highlighting",
"time": "32.877",
"value": "Richard Corish_sentence_5"
},
{
"operation": "Highlighting",
"time": "45.604",
"value": "Richard Corish_sentence_6"
},
{
"operation": "finish",
"time": "59.173",
"value": "finish"
}
]
}
|
exfever_8574
|
Mexico City is the capital and smallest city of Mexico which is located in the Valley of Mexico, a large valley in the high area of a highland consisting of flat surface that involves the vertical and horizontal dimensions of land surface, that is raised sharply above the surrounding area on at least one side.
|
{
"evidence": [
"Mexico City",
"Plateau",
"Terrain"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nMexico City is the capital and smallest city of Mexico which is located in the Valley of Mexico, a large valley in the high area of a highland consisting of flat surface that involves the vertical and horizontal dimensions of land surface, that is raised sharply above the surrounding area on at least one side.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Mexico City",
"Plateau",
"Terrain"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 1669,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
28962
|
Philip Vassar Hunter born 1883 invented the buoyant cable, he was president of the Institution of Electrical Engineers and the Australian Ice Hockey Association.
|
{
"evidence": [
"Philip Vassar Hunter_sentence_1",
"Philip Vassar Hunter_sentence_3",
"Philip Vassar Hunter_sentence_4",
"Philip Vassar Hunter_sentence_5"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nPhilip Vassar Hunter born 1883 invented the buoyant cable, he was president of the Institution of Electrical Engineers and the Australian Ice Hockey Association.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Philip Vassar Hunter_sentence_1",
"Philip Vassar Hunter_sentence_3",
"Philip Vassar Hunter_sentence_4",
"Philip Vassar Hunter_sentence_5"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 1670,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.139",
"value": "?search="
},
{
"operation": "search",
"time": "13.587",
"value": "Philip Vassar Hunter"
},
{
"operation": "Now on",
"time": "14.232",
"value": "Philip Vassar Hunter"
},
{
"operation": "Highlighting",
"time": "33.13",
"value": "Philip Vassar Hunter_sentence_0"
},
{
"operation": "Highlighting deleted",
"time": "44.794",
"value": "Philip Vassar Hunter_sentence_0"
},
{
"operation": "Highlighting",
"time": "45.586",
"value": "Philip Vassar Hunter_sentence_1"
},
{
"operation": "Highlighting",
"time": "50.794",
"value": "Philip Vassar Hunter_sentence_3"
},
{
"operation": "Highlighting",
"time": "56.106",
"value": "Philip Vassar Hunter_sentence_4"
},
{
"operation": "Highlighting",
"time": "60.378",
"value": "Philip Vassar Hunter_sentence_5"
},
{
"operation": "finish",
"time": "85.33",
"value": "finish"
}
]
}
|
78127
|
In the 2016 Longford Senior Football Championships, there were 14 teams grouped into groups A, B, and C.
|
{
"evidence": [
"2016 Longford Senior Football Championship_sentence_1",
"2016 Longford Senior Football Championship_sentence_9",
"2016 Longford Senior Football Championship_sentence_10",
"2016 Longford Senior Football Championship_sentence_8"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nIn the 2016 Longford Senior Football Championships, there were 14 teams grouped into groups A, B, and C.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"2016 Longford Senior Football Championship_sentence_1",
"2016 Longford Senior Football Championship_sentence_9",
"2016 Longford Senior Football Championship_sentence_10",
"2016 Longford Senior Football Championship_sentence_8"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 1671,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.011",
"value": "?search="
},
{
"operation": "search",
"time": "22.078",
"value": "2016 Longford Senior Football Championship"
},
{
"operation": "Now on",
"time": "22.773",
"value": "2016 Longford Senior Football Championship"
},
{
"operation": "Highlighting",
"time": "101.294",
"value": "2016 Longford Senior Football Championship_sentence_0"
},
{
"operation": "Highlighting deleted",
"time": "102.028",
"value": "2016 Longford Senior Football Championship_sentence_0"
},
{
"operation": "Highlighting",
"time": "104.469",
"value": "2016 Longford Senior Football Championship_sentence_1"
},
{
"operation": "Highlighting",
"time": "121.781",
"value": "2016 Longford Senior Football Championship_sentence_9"
},
{
"operation": "Highlighting",
"time": "122.732",
"value": "2016 Longford Senior Football Championship_sentence_10"
},
{
"operation": "Highlighting",
"time": "143.253",
"value": "2016 Longford Senior Football Championship_sentence_8"
},
{
"operation": "finish",
"time": "196.341",
"value": "finish"
}
]
}
|
16471
|
Wescot Credit Services Limited extended its data deal with the leading global information services company Experian, an American based Irish-domiciled multinational consumer credit reporting company, who agreed to provide data to Wescot.
|
{
"evidence": [
"Wescot Credit Services Limited_sentence_7",
"Wescot Credit Services Limited_sentence_8"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nWescot Credit Services Limited extended its data deal with the leading global information services company Experian, an American based Irish-domiciled multinational consumer credit reporting company, who agreed to provide data to Wescot.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Wescot Credit Services Limited_sentence_7",
"Wescot Credit Services Limited_sentence_8"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 1672,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.136",
"value": "?search="
},
{
"operation": "search",
"time": "15.719",
"value": "Wescot Credit Services Limited"
},
{
"operation": "Now on",
"time": "17.321",
"value": "Wescot Credit Services Limited"
},
{
"operation": "Page search",
"time": "20.813",
"value": "extend"
},
{
"operation": "Highlighting",
"time": "24.491",
"value": "Wescot Credit Services Limited_sentence_7"
},
{
"operation": "Highlighting",
"time": "30.323",
"value": "Wescot Credit Services Limited_sentence_8"
},
{
"operation": "hyperlink",
"time": "38.955",
"value": "Experian"
},
{
"operation": "Now on",
"time": "40.258",
"value": "Experian"
},
{
"operation": "Page search",
"time": "42.36",
"value": "Wescot Credit Services Limited e"
},
{
"operation": "Page search",
"time": "43.903",
"value": "Wescot Credit Services Limited"
},
{
"operation": "Page search",
"time": "51.699",
"value": "Wescot"
},
{
"operation": "finish",
"time": "68.942",
"value": "finish"
}
]
}
|
exfever_7374
|
The Matrix Revolutions is a 2003 American action novel whose genre uses speculative, fictional science-based depictions of phenomena that are not fully accepted by mainstream science.
|
{
"evidence": [
"The Matrix Revolutions",
"Science fiction film"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Matrix Revolutions is a 2003 American action novel whose genre uses speculative, fictional science-based depictions of phenomena that are not fully accepted by mainstream science.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Matrix Revolutions",
"Science fiction film"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 1673,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
11363
|
Joseph Newark founded Congregation B'nai B'rith in Los Angeles, which is now known as the Wilshire Boulevard Temple, he is also known for officiating the first Jewish wedding in California.
|
{
"evidence": [
"Joseph Newmark_sentence_1",
"Joseph Newmark_sentence_13"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nJoseph Newark founded Congregation B'nai B'rith in Los Angeles, which is now known as the Wilshire Boulevard Temple, he is also known for officiating the first Jewish wedding in California.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Joseph Newmark_sentence_1",
"Joseph Newmark_sentence_13"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 1674,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.072",
"value": "?search="
},
{
"operation": "search",
"time": "19.241",
"value": "Congregation B'nai B'rith"
},
{
"operation": "Now on",
"time": "20.48",
"value": "Wilshire Boulevard Temple"
},
{
"operation": "hyperlink",
"time": "39.848",
"value": "Congregation B'nai B'rith"
},
{
"operation": "Now on",
"time": "40.762",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "48.512",
"value": "Wilshire Boulevard Temple"
},
{
"operation": "Now on",
"time": "49.302",
"value": "Wilshire Boulevard Temple"
},
{
"operation": "Highlighting",
"time": "50.641",
"value": "Wilshire Boulevard Temple_sentence_0"
},
{
"operation": "Highlighting",
"time": "50.642",
"value": "Wilshire Boulevard Temple_title"
},
{
"operation": "Page search",
"time": "55.616",
"value": "Congregation B'nai B'rith"
},
{
"operation": "Page search",
"time": "72.97",
"value": "Congregation B'nai B'rith"
},
{
"operation": "back-button-clicked",
"time": "78.409",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "78.489",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "84.016",
"value": "Wilshire Boulevard Temple"
},
{
"operation": "Now on",
"time": "84.611",
"value": "Wilshire Boulevard Temple"
},
{
"operation": "Highlighting",
"time": "87.186",
"value": "Wilshire Boulevard Temple_header_cell_0_0_0"
},
{
"operation": "Highlighting",
"time": "90.25",
"value": "Wilshire Boulevard Temple_cell_0_0_1"
},
{
"operation": "Highlighting",
"time": "90.26",
"value": "Wilshire Boulevard Temple_header_cell_0_0_0"
},
{
"operation": "Page search",
"time": "101.547",
"value": "joseph newark"
},
{
"operation": "search",
"time": "122.809",
"value": "Contains...joseph newark"
},
{
"operation": "Now on",
"time": "124.528",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "127.894",
"value": "joseph newmark"
},
{
"operation": "Now on",
"time": "128.569",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "133.022",
"value": "Joseph Newmark"
},
{
"operation": "Now on",
"time": "133.678",
"value": "Joseph Newmark"
},
{
"operation": "Highlighting",
"time": "137.164",
"value": "Joseph Newmark_sentence_1"
},
{
"operation": "Page search",
"time": "158.077",
"value": "officiating"
},
{
"operation": "Page search",
"time": "168.471",
"value": "wedding"
},
{
"operation": "Highlighting",
"time": "173.346",
"value": "Joseph Newmark_sentence_13"
},
{
"operation": "back-button-clicked",
"time": "187.058",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "187.766",
"value": "undefined"
},
{
"operation": "back-button-clicked",
"time": "189.471",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "189.48",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "190.493",
"value": "undefined"
},
{
"operation": "back-button-clicked",
"time": "191.279",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "191.284",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "191.397",
"value": "Wilshire Boulevard Temple"
},
{
"operation": "Highlighting deleted",
"time": "195.782",
"value": "Wilshire Boulevard Temple_sentence_0"
},
{
"operation": "Highlighting deleted",
"time": "203.454",
"value": "Wilshire Boulevard Temple_cell_0_0_1"
},
{
"operation": "Highlighting deleted",
"time": "203.455",
"value": "Wilshire Boulevard Temple_title"
},
{
"operation": "Highlighting deleted",
"time": "203.46",
"value": "Wilshire Boulevard Temple_header_cell_0_0_0"
},
{
"operation": "finish",
"time": "232.491",
"value": "finish"
}
]
}
|
exfever_2566
|
Andy Warhol was a leading figure of pop art who explored the relationship between different media, including a form of marketing communication that is communicated through various mass media.
|
{
"evidence": [
"Andy Warhol",
"Advertising"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nAndy Warhol was a leading figure of pop art who explored the relationship between different media, including a form of marketing communication that is communicated through various mass media.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Andy Warhol",
"Advertising"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 1675,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_3962
|
Daddy Yankee was the wealthiest Puerto Rican rapper.
|
{
"evidence": [
"Daddy Yankee",
"DJ Playero"
],
"label": "NOT ENOUGH INFO"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nDaddy Yankee was the wealthiest Puerto Rican rapper.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Daddy Yankee",
"DJ Playero"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 1676,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
35508
|
The 1981 John Player Special Formula 1 World Series was the inaugural season of Formula 1 Powerboat racing, where Renato Molinari (an 18-time World Champion in different categories) was the series' first-ever champion, winning four of the five races he contested.
|
{
"evidence": [
"1981 F1 Powerboat World Championship_sentence_0",
"1981 F1 Powerboat World Championship_sentence_3",
"Renato Molinari_sentence_1"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe 1981 John Player Special Formula 1 World Series was the inaugural season of Formula 1 Powerboat racing, where Renato Molinari (an 18-time World Champion in different categories) was the series' first-ever champion, winning four of the five races he contested.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"1981 F1 Powerboat World Championship_sentence_0",
"1981 F1 Powerboat World Championship_sentence_3",
"Renato Molinari_sentence_1"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 1677,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.067",
"value": "?search="
},
{
"operation": "search",
"time": "61.489",
"value": "1981 F1 Powerboat World Championship"
},
{
"operation": "Now on",
"time": "62.324",
"value": "1981 F1 Powerboat World Championship"
},
{
"operation": "Highlighting",
"time": "66.751",
"value": "1981 F1 Powerboat World Championship_sentence_0"
},
{
"operation": "Highlighting",
"time": "84.901",
"value": "1981 F1 Powerboat World Championship_sentence_3"
},
{
"operation": "hyperlink",
"time": "93.783",
"value": "Renato Molinari"
},
{
"operation": "Now on",
"time": "94.47",
"value": "Renato Molinari"
},
{
"operation": "Highlighting",
"time": "99.011",
"value": "Renato Molinari_sentence_1"
},
{
"operation": "back-button-clicked",
"time": "104.203",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "104.298",
"value": "1981 F1 Powerboat World Championship"
},
{
"operation": "finish",
"time": "109.644",
"value": "finish"
}
]
}
|
32338
|
Most Somali men have found jobs in the transport and production industries, and a rising number of women in the community do clerical work; while communal support networks have also compressed.
|
{
"evidence": [
"Somali Australians_sentence_12"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nMost Somali men have found jobs in the transport and production industries, and a rising number of women in the community do clerical work; while communal support networks have also compressed.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Somali Australians_sentence_12"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 1678,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.563",
"value": "?search="
},
{
"operation": "search",
"time": "16.081",
"value": "somalia"
},
{
"operation": "Now on",
"time": "17.522",
"value": "Somalia"
},
{
"operation": "search",
"time": "24.122",
"value": "Most Somali men"
},
{
"operation": "Now on",
"time": "25.967",
"value": "undefined"
},
{
"operation": "back-button-clicked",
"time": "41.785",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "42.053",
"value": "Somalia"
},
{
"operation": "Page search",
"time": "44.877",
"value": "Most Somali men"
},
{
"operation": "search",
"time": "68.736",
"value": "rising number of women in the community do clerical work"
},
{
"operation": "Now on",
"time": "70.833",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "75.274",
"value": "Somali community in Australia"
},
{
"operation": "Now on",
"time": "75.687",
"value": "Somali Australians"
},
{
"operation": "Highlighting",
"time": "83.635",
"value": "Somali Australians_sentence_11"
},
{
"operation": "Highlighting",
"time": "90.502",
"value": "Somali Australians_sentence_12"
},
{
"operation": "Highlighting deleted",
"time": "104.431",
"value": "Somali Australians_sentence_11"
},
{
"operation": "finish",
"time": "112.872",
"value": "finish"
}
]
}
|
71548
|
Hamgyeong Line was a railway line which transported passengers and goods on wheeled vehicles running on rails that are located on tracks.
|
{
"evidence": [
"Hamgyeong Line_sentence_0",
"Rail transport_sentence_3"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nHamgyeong Line was a railway line which transported passengers and goods on wheeled vehicles running on rails that are located on tracks.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Hamgyeong Line_sentence_0",
"Rail transport_sentence_3"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 1679,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.923",
"value": "?search="
},
{
"operation": "search",
"time": "7.315",
"value": "Hamgyeong Line"
},
{
"operation": "Now on",
"time": "17.096",
"value": "Hamgyeong Line"
},
{
"operation": "Page search",
"time": "441.569",
"value": "wheeled vehicles"
},
{
"operation": "Highlighting",
"time": "514.737",
"value": "Hamgyeong Line_sentence_4"
},
{
"operation": "Highlighting",
"time": "514.738",
"value": "Hamgyeong Line_section_0"
},
{
"operation": "Highlighting deleted",
"time": "515.361",
"value": "Hamgyeong Line_sentence_4"
},
{
"operation": "Highlighting deleted",
"time": "515.364",
"value": "Hamgyeong Line_section_0"
},
{
"operation": "search",
"time": "588.216",
"value": "transported passengers and goods on wheeled vehicles"
},
{
"operation": "Now on",
"time": "591.153",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "600.311",
"value": "Rail transport"
},
{
"operation": "Now on",
"time": "602.026",
"value": "Rail transport"
},
{
"operation": "Highlighting",
"time": "610.351",
"value": "Rail transport_sentence_3"
},
{
"operation": "back-button-clicked",
"time": "612.984",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "614.129",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "614.354",
"value": "Hamgyeong Line"
},
{
"operation": "Highlighting",
"time": "626.225",
"value": "Hamgyeong Line_sentence_0"
},
{
"operation": "hyperlink",
"time": "626.892",
"value": "railway line"
},
{
"operation": "Now on",
"time": "627.302",
"value": "Rail transport"
},
{
"operation": "Highlighting deleted",
"time": "628.61",
"value": "Rail transport_sentence_3"
},
{
"operation": "Highlighting",
"time": "628.923",
"value": "Rail transport_sentence_3"
},
{
"operation": "finish",
"time": "640.917",
"value": "finish"
}
]
}
|
32275
|
Both Suzuki & Nyima (2016) and Tournadre (2014) classifies Basum, a language spoken by 13.5% of the population of Gongbo'gyamda County, to be an unclassified Bodish language and non-Tibetic language.
|
{
"evidence": [
"Basum language_sentence_11",
"Basum language_sentence_10",
"Basum language_sentence_1"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nBoth Suzuki & Nyima (2016) and Tournadre (2014) classifies Basum, a language spoken by 13.5% of the population of Gongbo'gyamda County, to be an unclassified Bodish language and non-Tibetic language.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Basum language_sentence_11",
"Basum language_sentence_10",
"Basum language_sentence_1"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 1680,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.768",
"value": "?search="
},
{
"operation": "search",
"time": "31.17",
"value": "Suzuki & Nyima"
},
{
"operation": "Now on",
"time": "33.567",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "45.931",
"value": "Lamo language"
},
{
"operation": "Now on",
"time": "47.056",
"value": "Lamo language"
},
{
"operation": "Page search",
"time": "54.44",
"value": "Suzuki & Nyima"
},
{
"operation": "Page search",
"time": "71.316",
"value": "basum"
},
{
"operation": "search",
"time": "92.302",
"value": "basum"
},
{
"operation": "Now on",
"time": "93.594",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "96.35",
"value": "Basum"
},
{
"operation": "Now on",
"time": "97.368",
"value": "Basum"
},
{
"operation": "hyperlink",
"time": "101.311",
"value": "Basum language"
},
{
"operation": "Now on",
"time": "102.337",
"value": "Basum language"
},
{
"operation": "Page search",
"time": "125.57",
"value": "Suzuki & Nyima"
},
{
"operation": "Page search",
"time": "152.278",
"value": "tournadre"
},
{
"operation": "Page search",
"time": "152.676",
"value": "tournadre"
},
{
"operation": "Highlighting",
"time": "161.469",
"value": "Basum language_sentence_11"
},
{
"operation": "Page search",
"time": "184.255",
"value": "suzuki"
},
{
"operation": "Highlighting",
"time": "189.887",
"value": "Basum language_sentence_10"
},
{
"operation": "Highlighting",
"time": "334.883",
"value": "Basum language_sentence_1"
},
{
"operation": "finish",
"time": "356.712",
"value": "finish"
}
]
}
|
46869
|
In 1924 Charles Lethbridge Kingsford was elected a fellow of the British Academy, which is the United Kingdom's national academy for the humanities and the social sciences, then, later on, was recognized "as our greatest modern authority on the history of the late 15th century".
|
{
"evidence": [
"Charles Lethbridge Kingsford_sentence_10",
"British Academy_sentence_1",
"Charles Lethbridge Kingsford_sentence_11"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nIn 1924 Charles Lethbridge Kingsford was elected a fellow of the British Academy, which is the United Kingdom's national academy for the humanities and the social sciences, then, later on, was recognized \"as our greatest modern authority on the history of the late 15th century\".\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Charles Lethbridge Kingsford_sentence_10",
"British Academy_sentence_1",
"Charles Lethbridge Kingsford_sentence_11"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 1681,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.228",
"value": "?search="
},
{
"operation": "search",
"time": "25.286",
"value": "Charles Lethbridge Kingsford"
},
{
"operation": "Now on",
"time": "26.214",
"value": "Charles Lethbridge Kingsford"
},
{
"operation": "Highlighting",
"time": "35.849",
"value": "Charles Lethbridge Kingsford_sentence_10"
},
{
"operation": "hyperlink",
"time": "39.118",
"value": "British Academy"
},
{
"operation": "Now on",
"time": "39.693",
"value": "British Academy"
},
{
"operation": "Highlighting",
"time": "44.95",
"value": "British Academy_sentence_1"
},
{
"operation": "back-button-clicked",
"time": "48.014",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "48.121",
"value": "Charles Lethbridge Kingsford"
},
{
"operation": "Highlighting",
"time": "65.001",
"value": "Charles Lethbridge Kingsford_sentence_11"
},
{
"operation": "finish",
"time": "83.011",
"value": "finish"
}
]
}
|
exfever_4006
|
Tatiana Maslany is a Canadian actress who appeared in a Canadian comedy-drama television series aired on CBC and distributed internationally by CBC Worldwide.
|
{
"evidence": [
"Tatiana Maslany",
"Being Erica"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nTatiana Maslany is a Canadian actress who appeared in a Canadian comedy-drama television series aired on CBC and distributed internationally by CBC Worldwide.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Tatiana Maslany",
"Being Erica"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 1682,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_2404
|
Seth Rogen got a part in a sitcom that was listed in the "25 Best Cult TV Shows of the Past 25 Years" by an American monthly entertainment magazine.
|
{
"evidence": [
"Seth Rogen",
"Undeclared",
"Entertainment Weekly"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nSeth Rogen got a part in a sitcom that was listed in the \"25 Best Cult TV Shows of the Past 25 Years\" by an American monthly entertainment magazine.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Seth Rogen",
"Undeclared",
"Entertainment Weekly"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 1683,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.