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 |
|---|---|---|---|---|---|---|---|---|
6210
|
Carlo Beenakker was born in 1960 and worked in theoretical physics, winning an award in 1999 for his work.
|
{
"evidence": [
"Carlo Beenakker_sentence_0",
"Carlo Beenakker_sentence_3",
"Carlo Beenakker_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:\nCarlo Beenakker was born in 1960 and worked in theoretical physics, winning an award in 1999 for his work.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Carlo Beenakker_sentence_0",
"Carlo Beenakker_sentence_3",
"Carlo Beenakker_sentence_7"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 384,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.576",
"value": "?search="
},
{
"operation": "search",
"time": "5.924",
"value": "Carlo Beenakker"
},
{
"operation": "Now on",
"time": "7.645",
"value": "Carlo Beenakker"
},
{
"operation": "Highlighting",
"time": "8.936",
"value": "Carlo Beenakker_sentence_0"
},
{
"operation": "Page search",
"time": "18.851",
"value": "theoretical"
},
{
"operation": "Highlighting",
"time": "20.887",
"value": "Carlo Beenakker_sentence_3"
},
{
"operation": "Highlighting",
"time": "26.511",
"value": "Carlo Beenakker_sentence_7"
},
{
"operation": "finish",
"time": "369.992",
"value": "finish"
}
]
}
|
exfever_2544
|
Lamniformes belong to a family of mackerel sharks named after a Greek mythical creature, a fictional entity.
|
{
"evidence": [
"Lamniformes",
"Lamnidae",
"Legendary creature"
],
"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:\nLamniformes belong to a family of mackerel sharks named after a Greek mythical creature, a fictional entity.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Lamniformes",
"Lamnidae",
"Legendary creature"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 385,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
84907
|
She was born on August 26, 1991 in Devonport, New Zealand studied in Queen's High School and became a member of the Queen's madrigal choir.
|
{
"evidence": [
"Nadia Reid_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:\nShe was born on August 26, 1991 in Devonport, New Zealand studied in Queen's High School and became a member of the Queen's madrigal choir.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Nadia Reid_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 386,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.717",
"value": "?search="
},
{
"operation": "search",
"time": "19.398",
"value": "Contains...She was born on August 26, 1991 in Devonport, New Zealand studied in Queen's High School and became a member of the Queen's madrigal choir."
},
{
"operation": "Now on",
"time": "21.576",
"value": "undefined"
},
{
"operation": "search",
"time": "43.843",
"value": "Contains...she became a member of the Queen's madrigal choir"
},
{
"operation": "Now on",
"time": "45.329",
"value": "undefined"
},
{
"operation": "search",
"time": "111.131",
"value": "Contains...Queen's High School and became a member of the Queen's madrigal choir"
},
{
"operation": "Now on",
"time": "111.892",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "122.882",
"value": "Gustav Holst"
},
{
"operation": "Now on",
"time": "123.853",
"value": "Gustav Holst"
},
{
"operation": "back-button-clicked",
"time": "128.048",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "128.692",
"value": "undefined"
},
{
"operation": "search",
"time": "178.942",
"value": "Nadia Reid"
},
{
"operation": "Now on",
"time": "179.807",
"value": "Nadia Reid"
},
{
"operation": "Highlighting",
"time": "183.538",
"value": "Nadia Reid_sentence_0"
},
{
"operation": "Page search",
"time": "200.495",
"value": "madrigal"
},
{
"operation": "finish",
"time": "301.01",
"value": "finish"
}
]
}
|
exfever_6853
|
Amy Winehouse was an English singer and songwriter known for her deep, expressive contralto vocals and her eclectic mix of musical genres including a popular music genre that originated in the African American community throughout the United Kingdom in the 1950s and early 1960s.
|
{
"evidence": [
"Amy Winehouse",
"Soul music"
],
"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:\nAmy Winehouse was an English singer and songwriter known for her deep, expressive contralto vocals and her eclectic mix of musical genres including a popular music genre that originated in the African American community throughout the United Kingdom in the 1950s and early 1960s.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Amy Winehouse",
"Soul music"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 387,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
93850
|
On 29 November 2011, Syamsudin Noor Airport's radar detected Dassault Falcon 900 EX, but Kohanudnas let the aircraft enter since it has granted clearance by the minister of foreign affairs of Indonesia.
|
{
"evidence": [
"Indonesian National Air Defense Forces Command_sentence_22",
"Indonesian National Air Defense Forces Command_sentence_23",
"Indonesian National Air Defense Forces Command_sentence_24"
],
"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:\nOn 29 November 2011, Syamsudin Noor Airport's radar detected Dassault Falcon 900 EX, but Kohanudnas let the aircraft enter since it has granted clearance by the minister of foreign affairs of Indonesia.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Indonesian National Air Defense Forces Command_sentence_22",
"Indonesian National Air Defense Forces Command_sentence_23",
"Indonesian National Air Defense Forces Command_sentence_24"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 388,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.402",
"value": "?search="
},
{
"operation": "search",
"time": "14.36",
"value": "\"Kohanudnas let the aircraft enter\""
},
{
"operation": "Now on",
"time": "19.175",
"value": "undefined"
},
{
"operation": "search",
"time": "37.594",
"value": "\"radar detected Dassault Falcon 900 EX\""
},
{
"operation": "Now on",
"time": "39.079",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "41.331",
"value": "Indonesian National Air Defense Forces Command"
},
{
"operation": "Now on",
"time": "42.888",
"value": "Indonesian National Air Defense Forces Command"
},
{
"operation": "Page search",
"time": "45.789",
"value": "radar detected Dassault Falcon 900 EX"
},
{
"operation": "Highlighting",
"time": "52.574",
"value": "Indonesian National Air Defense Forces Command_sentence_20"
},
{
"operation": "Highlighting",
"time": "73.076",
"value": "Indonesian National Air Defense Forces Command_sentence_22"
},
{
"operation": "Highlighting",
"time": "81.66",
"value": "Indonesian National Air Defense Forces Command_sentence_23"
},
{
"operation": "Highlighting deleted",
"time": "104.456",
"value": "Indonesian National Air Defense Forces Command_sentence_20"
},
{
"operation": "Highlighting",
"time": "112.807",
"value": "Indonesian National Air Defense Forces Command_sentence_24"
},
{
"operation": "finish",
"time": "128.093",
"value": "finish"
}
]
}
|
14481
|
Sid Lowe is an English journalist who covers Spanish football.
|
{
"evidence": [
"Sid Lowe_sentence_0",
"Sid Lowe_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:\nSid Lowe is an English journalist who covers Spanish football.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Sid Lowe_sentence_0",
"Sid Lowe_sentence_1"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 389,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.16",
"value": "?search="
},
{
"operation": "search",
"time": "11.817",
"value": "Sid Lowe"
},
{
"operation": "Now on",
"time": "12.573",
"value": "Sid Lowe"
},
{
"operation": "Highlighting",
"time": "14.744",
"value": "Sid Lowe_sentence_0"
},
{
"operation": "Highlighting",
"time": "17.912",
"value": "Sid Lowe_sentence_1"
},
{
"operation": "finish",
"time": "23.627",
"value": "finish"
}
]
}
|
exfever_5954
|
Boyd Holbrook is an American model who appeared in an American film genre combining the thriller and psychological fiction genres.
|
{
"evidence": [
"Boyd Holbrook",
"Gone Girl (film)",
"Psychological thriller"
],
"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:\nBoyd Holbrook is an American model who appeared in an American film genre combining the thriller and psychological fiction genres.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Boyd Holbrook",
"Gone Girl (film)",
"Psychological thriller"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 390,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7034
|
Helen Catherine Hardwicke is an German film director, production designer, and screenwriter best known for co-writing the 2003 film Thirteen with an American actress best known for her role as vampire Rosalie Hale in The Twilight Saga.
|
{
"evidence": [
"Catherine Hardwicke",
"Nikki Reed"
],
"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:\nHelen Catherine Hardwicke is an German film director, production designer, and screenwriter best known for co-writing the 2003 film Thirteen with an American actress best known for her role as vampire Rosalie Hale in The Twilight Saga.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Catherine Hardwicke",
"Nikki Reed"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 391,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
28482
|
The 3823 Yorii, a carbonaceous asteroid was discovered in central Japen solely by H. Mori during 1988.
|
{
"evidence": [
"3823 Yorii_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 3823 Yorii, a carbonaceous asteroid was discovered in central Japen solely by H. Mori during 1988.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"3823 Yorii_sentence_1"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 392,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.85",
"value": "?search="
},
{
"operation": "search",
"time": "14.509",
"value": "3823 Yorii"
},
{
"operation": "Now on",
"time": "15.215",
"value": "3823 Yorii"
},
{
"operation": "Highlighting",
"time": "61.317",
"value": "3823 Yorii_sentence_0"
},
{
"operation": "Highlighting",
"time": "97.981",
"value": "3823 Yorii_sentence_1"
},
{
"operation": "Highlighting deleted",
"time": "104.145",
"value": "3823 Yorii_sentence_0"
},
{
"operation": "finish",
"time": "110.385",
"value": "finish"
}
]
}
|
57563
|
During the 2017-2018 season, the Liberty Lady Flames basketball team won the Big South Women's Basketball Tournament for Liberty University and proceeded to the NCAA Women's Tournament.
|
{
"evidence": [
"2017–18 Liberty Lady Flames basketball team_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:\nDuring the 2017-2018 season, the Liberty Lady Flames basketball team won the Big South Women's Basketball Tournament for Liberty University and proceeded to the NCAA Women's Tournament.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"2017–18 Liberty Lady Flames basketball team_sentence_3"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 393,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.886",
"value": "?search="
},
{
"operation": "search",
"time": "15.129",
"value": "Contains...Liberty Lady Flames basketball team"
},
{
"operation": "Now on",
"time": "17.067",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "18.876",
"value": "2018–19 Liberty Lady Flames basketball team"
},
{
"operation": "Now on",
"time": "20.03",
"value": "2018–19 Liberty Lady Flames basketball team"
},
{
"operation": "search",
"time": "120.699",
"value": "Contains...2017-2018 Liberty Lady Flames basketball team"
},
{
"operation": "Now on",
"time": "122.421",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "126.582",
"value": "2017–18 Liberty Lady Flames basketball team"
},
{
"operation": "Now on",
"time": "127.316",
"value": "2017–18 Liberty Lady Flames basketball team"
},
{
"operation": "Highlighting",
"time": "178.343",
"value": "2017–18 Liberty Lady Flames basketball team_sentence_3"
},
{
"operation": "finish",
"time": "197.252",
"value": "finish"
}
]
}
|
exfever_2179
|
Andrew Stanton has also produced short stories preserved by the Academy Film Archive.
|
{
"evidence": [
"Andrew Stanton",
"Finding Dory"
],
"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:\nAndrew Stanton has also produced short stories preserved by the Academy Film Archive.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Andrew Stanton",
"Finding Dory"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 394,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_6040
|
NL is the largest province in Canada.
|
{
"evidence": [
"Newfoundland and Labrador",
"Newfoundland French"
],
"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:\nNL is the largest province in Canada.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Newfoundland and Labrador",
"Newfoundland French"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 395,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
68732
|
The 2015 Ironman World Championship was organized by the World Triathlon Corporation (WTC), a sports event promotion company owned by conglomerate Advance Publications, that produces the Ironman Triathlon, Ironman 70.3, the 5150 series of triathlon races, and other sports events.
|
{
"evidence": [
"2015 Ironman World Championship_sentence_2",
"World Triathlon Corporation_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 2015 Ironman World Championship was organized by the World Triathlon Corporation (WTC), a sports event promotion company owned by conglomerate Advance Publications, that produces the Ironman Triathlon, Ironman 70.3, the 5150 series of triathlon races, and other sports events.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"2015 Ironman World Championship_sentence_2",
"World Triathlon Corporation_sentence_0"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 396,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.778",
"value": "?search="
},
{
"operation": "search",
"time": "14.762",
"value": "2015 Ironman World Championship"
},
{
"operation": "Now on",
"time": "16.117",
"value": "2015 Ironman World Championship"
},
{
"operation": "Highlighting",
"time": "27.049",
"value": "2015 Ironman World Championship_sentence_2"
},
{
"operation": "hyperlink",
"time": "36.245",
"value": "World Triathlon Corporation"
},
{
"operation": "Now on",
"time": "37.148",
"value": "World Triathlon Corporation"
},
{
"operation": "Highlighting",
"time": "39.799",
"value": "World Triathlon Corporation_sentence_0"
},
{
"operation": "finish",
"time": "59.662",
"value": "finish"
}
]
}
|
61725
|
Kollam Canal is a part of Kollam Canal development project that is expected to meet the water transportation and irrigation needs of the people of Kollam Urban Agglomeration in future.
|
{
"evidence": [
"Kollam Canal_sentence_8",
"Kollam Canal_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:\nKollam Canal is a part of Kollam Canal development project that is expected to meet the water transportation and irrigation needs of the people of Kollam Urban Agglomeration in future.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Kollam Canal_sentence_8",
"Kollam Canal_sentence_11"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 397,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.725",
"value": "?search="
},
{
"operation": "search",
"time": "13.435",
"value": "Kollam Canal"
},
{
"operation": "Now on",
"time": "14.319",
"value": "Kollam Canal"
},
{
"operation": "Highlighting",
"time": "54.08",
"value": "Kollam Canal_sentence_8"
},
{
"operation": "Highlighting",
"time": "60.75",
"value": "Kollam Canal_sentence_11"
},
{
"operation": "finish",
"time": "97.187",
"value": "finish"
}
]
}
|
exfever_5674
|
Shifty is an American urban crime thriller filmed in Borehamwood, starring a veteran actress who is a six-time BAFTA TV Award nominee and won the 1979 BAFTA TV Award for Best Actress for the ITV series Lillie.
|
{
"evidence": [
"Shifty (film)",
"Francesca Annis"
],
"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:\nShifty is an American urban crime thriller filmed in Borehamwood, starring a veteran actress who is a six-time BAFTA TV Award nominee and won the 1979 BAFTA TV Award for Best Actress for the ITV series Lillie.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Shifty (film)",
"Francesca Annis"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 398,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_4444
|
Steven Zaillian is the oldest American screenwriter to win an Academy Award.
|
{
"evidence": [
"Steven Zaillian",
"Awakenings"
],
"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:\nSteven Zaillian is the oldest American screenwriter to win an Academy Award.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Steven Zaillian",
"Awakenings"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 399,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_2277
|
In 1616, Thomas Quiney married Judith Shakespeare.
|
{
"evidence": [
"Judith Quiney",
"Hamnet Shakespeare"
],
"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:\nIn 1616, Thomas Quiney married Judith Shakespeare.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Judith Quiney",
"Hamnet Shakespeare"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 400,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
35021
|
Fernando Morena was part of the team that won the Copa América in 1983, the oldest international football competition.
|
{
"evidence": [
"Fernando Morena_sentence_14",
"Copa América_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:\nFernando Morena was part of the team that won the Copa América in 1983, the oldest international football competition.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Fernando Morena_sentence_14",
"Copa América_sentence_3"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 401,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.671",
"value": "?search="
},
{
"operation": "search",
"time": "44.769",
"value": "Fernando Morena"
},
{
"operation": "Now on",
"time": "45.363",
"value": "Fernando Morena"
},
{
"operation": "search",
"time": "82.037",
"value": "Copa América in 1983"
},
{
"operation": "Now on",
"time": "83.823",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "88.137",
"value": "1983 Copa América"
},
{
"operation": "Now on",
"time": "88.746",
"value": "1983 Copa América"
},
{
"operation": "hyperlink",
"time": "377.236",
"value": "Uruguay"
},
{
"operation": "Now on",
"time": "379.447",
"value": "Uruguay national football team"
},
{
"operation": "back-button-clicked",
"time": "383.963",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "384.049",
"value": "1983 Copa América"
},
{
"operation": "Highlighting",
"time": "388.099",
"value": "1983 Copa América_cell_0_4_1"
},
{
"operation": "Highlighting",
"time": "388.1",
"value": "1983 Copa América_title"
},
{
"operation": "Highlighting",
"time": "388.104",
"value": "1983 Copa América_header_cell_0_4_0"
},
{
"operation": "Highlighting",
"time": "388.104",
"value": "1983 Copa América_header_cell_0_3_0"
},
{
"operation": "hyperlink",
"time": "410.777",
"value": "Uruguay"
},
{
"operation": "Now on",
"time": "411.587",
"value": "Uruguay national football team"
},
{
"operation": "search",
"time": "456.964",
"value": "Fernando Morena"
},
{
"operation": "Now on",
"time": "457.523",
"value": "Fernando Morena"
},
{
"operation": "Highlighting",
"time": "489.421",
"value": "Fernando Morena_item_2_4"
},
{
"operation": "Highlighting",
"time": "489.422",
"value": "Fernando Morena_section_3"
},
{
"operation": "Highlighting",
"time": "489.422",
"value": "Fernando Morena_section_5"
},
{
"operation": "hyperlink",
"time": "500.656",
"value": "Uruguayan Primera División"
},
{
"operation": "Now on",
"time": "501.509",
"value": "Uruguayan Primera División"
},
{
"operation": "back-button-clicked",
"time": "504.821",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "504.883",
"value": "Fernando Morena"
},
{
"operation": "hyperlink",
"time": "511.711",
"value": "1982"
},
{
"operation": "Now on",
"time": "512.574",
"value": "1982 Uruguayan Primera División"
},
{
"operation": "search",
"time": "525.291",
"value": "1983 Uruguayan Primera División"
},
{
"operation": "Now on",
"time": "525.902",
"value": "1983 Uruguayan Primera División"
},
{
"operation": "back-button-clicked",
"time": "533.363",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "533.414",
"value": "1982 Uruguayan Primera División"
},
{
"operation": "back-button-clicked",
"time": "533.956",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "533.959",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "534.006",
"value": "Fernando Morena"
},
{
"operation": "Highlighting",
"time": "547.593",
"value": "Fernando Morena_sentence_14"
},
{
"operation": "hyperlink",
"time": "559.335",
"value": "Copa América"
},
{
"operation": "Now on",
"time": "560.563",
"value": "Copa América"
},
{
"operation": "Highlighting",
"time": "566.085",
"value": "Copa América_sentence_3"
},
{
"operation": "finish",
"time": "572.484",
"value": "finish"
}
]
}
|
47880
|
Jerry knows about the US$750,000 in his bank account and the illicit items in his apartment that he manages to evade the FBI arrest.
|
{
"evidence": [
"Eagle Eye_sentence_10"
],
"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:\nJerry knows about the US$750,000 in his bank account and the illicit items in his apartment that he manages to evade the FBI arrest.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Eagle Eye_sentence_10"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 402,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.961",
"value": "?search="
},
{
"operation": "search",
"time": "30.493",
"value": "jerry knows about the US$750,000"
},
{
"operation": "Now on",
"time": "32.621",
"value": "undefined"
},
{
"operation": "search",
"time": "44.121",
"value": "Contains...jerry knows about the US$750,000"
},
{
"operation": "Now on",
"time": "44.992",
"value": "undefined"
},
{
"operation": "search",
"time": "68.805",
"value": "Contains...Jerry knows about the US$750,000 in his bank account and the illicit items in his apartment that he manages to evade the FBI arrest."
},
{
"operation": "Now on",
"time": "70.339",
"value": "undefined"
},
{
"operation": "back-button-clicked",
"time": "74.796",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "75.675",
"value": "undefined"
},
{
"operation": "search",
"time": "186.173",
"value": "Jerry knows about the $750,000 in his bank account"
},
{
"operation": "Now on",
"time": "187.774",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "199.541",
"value": "Elvis Presley"
},
{
"operation": "Now on",
"time": "201.878",
"value": "Elvis Presley"
},
{
"operation": "Page search",
"time": "206.558",
"value": "bank"
},
{
"operation": "back-button-clicked",
"time": "225.738",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "226.884",
"value": "undefined"
},
{
"operation": "search",
"time": "303.21",
"value": "fargo"
},
{
"operation": "Now on",
"time": "304.227",
"value": "Fargo"
},
{
"operation": "hyperlink",
"time": "312.351",
"value": "Fargo (film)"
},
{
"operation": "Now on",
"time": "313.257",
"value": "Fargo (film)"
},
{
"operation": "Page search",
"time": "317.951",
"value": "jerry"
},
{
"operation": "Page search",
"time": "339.586",
"value": "bank"
},
{
"operation": "back-button-clicked",
"time": "350.95",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "351.124",
"value": "Fargo"
},
{
"operation": "Highlighting",
"time": "353.389",
"value": "Fargo_item_0_2"
},
{
"operation": "Highlighting",
"time": "353.39",
"value": "Fargo_title"
},
{
"operation": "Highlighting deleted",
"time": "355.051",
"value": "Fargo_item_0_2"
},
{
"operation": "Highlighting deleted",
"time": "355.052",
"value": "Fargo_title"
},
{
"operation": "hyperlink",
"time": "356.371",
"value": "Fargo (TV series)"
},
{
"operation": "Now on",
"time": "357.422",
"value": "Fargo (TV series)"
},
{
"operation": "Page search",
"time": "360.181",
"value": "bank"
},
{
"operation": "search",
"time": "393.484",
"value": "Eagle eye"
},
{
"operation": "Now on",
"time": "394.318",
"value": "Eagle eye"
},
{
"operation": "hyperlink",
"time": "398.216",
"value": "Eagle Eye (disambiguation)"
},
{
"operation": "Now on",
"time": "398.572",
"value": "Eagle Eye (disambiguation)"
},
{
"operation": "Highlighting",
"time": "401.007",
"value": "Eagle Eye (disambiguation)_sentence_0"
},
{
"operation": "Highlighting",
"time": "401.008",
"value": "Eagle Eye (disambiguation)_title"
},
{
"operation": "Highlighting deleted",
"time": "401.694",
"value": "Eagle Eye (disambiguation)_sentence_0"
},
{
"operation": "Highlighting deleted",
"time": "401.695",
"value": "Eagle Eye (disambiguation)_title"
},
{
"operation": "hyperlink",
"time": "403.258",
"value": "Eagle Eye"
},
{
"operation": "Now on",
"time": "404.028",
"value": "Eagle Eye"
},
{
"operation": "Page search",
"time": "410.412",
"value": "bank"
},
{
"operation": "Highlighting",
"time": "435.577",
"value": "Eagle Eye_sentence_8"
},
{
"operation": "Page search",
"time": "442.169",
"value": "fbi"
},
{
"operation": "Highlighting",
"time": "448.156",
"value": "Eagle Eye_sentence_10"
},
{
"operation": "Highlighting deleted",
"time": "450.683",
"value": "Eagle Eye_sentence_8"
},
{
"operation": "finish",
"time": "462.135",
"value": "finish"
}
]
}
|
exfever_6418
|
Konidela Production Company is an Indian film production company established by one of the highest-paid actors in India.
|
{
"evidence": [
"Konidela Production Company",
"Ram Charan"
],
"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:\nKonidela Production Company is an Indian film production company established by one of the highest-paid actors in India.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Konidela Production Company",
"Ram Charan"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 403,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
2309
|
Democratic Alliance, the official opposition to the ruling party African National Congress (ANC), got the most vote by party-list meanwhile Freedom Front Plus got the least number of votes by party-list.
|
{
"evidence": [
"Democratic Alliance (South Africa)_sentence_0",
"Democratic Alliance (South Africa)_sentence_45"
],
"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:\nDemocratic Alliance, the official opposition to the ruling party African National Congress (ANC), got the most vote by party-list meanwhile Freedom Front Plus got the least number of votes by party-list.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Democratic Alliance (South Africa)_sentence_0",
"Democratic Alliance (South Africa)_sentence_45"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 404,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "search",
"time": "21.082",
"value": "Democratic Alliance (South Africa)"
},
{
"operation": "Highlighting",
"time": "27.103",
"value": "Democratic Alliance (South Africa)_sentence_0"
},
{
"operation": "Highlighting",
"time": "68.602",
"value": "Democratic Alliance (South Africa)_sentence_45"
},
{
"operation": "search",
"time": "112.897",
"value": "Freedom Front Plus"
},
{
"operation": "finish",
"time": "154.354",
"value": "finish"
}
]
}
|
exfever_5765
|
Patton Oswalt is an American stand-up comedian who narrated the sitcom The Goldbergs as an adult Canadian television and film producer, and writer best known as the creator and showrunner of The Goldbergs.
|
{
"evidence": [
"Patton Oswalt",
"Adam F. Goldberg"
],
"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:\nPatton Oswalt is an American stand-up comedian who narrated the sitcom The Goldbergs as an adult Canadian television and film producer, and writer best known as the creator and showrunner of The Goldbergs.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Patton Oswalt",
"Adam F. Goldberg"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 405,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_947
|
Asceticism is the most popular lifestyle to pursue spiritual goals.?
|
{
"evidence": [
"Asceticism",
"Buddhism"
],
"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:\nAsceticism is the most popular lifestyle to pursue spiritual goals.?\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Asceticism",
"Buddhism"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 406,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
72661
|
Naichi, a term to distinguish the area of Japan from its outlying territories during the prewar period, has an estimated population of 72,000,000.
|
{
"evidence": [
"Mainland Japan_sentence_0",
"Mainland Japan_sentence_1",
"Mainland Japan_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:\nNaichi, a term to distinguish the area of Japan from its outlying territories during the prewar period, has an estimated population of 72,000,000.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Mainland Japan_sentence_0",
"Mainland Japan_sentence_1",
"Mainland Japan_sentence_2"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 407,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.065",
"value": "?search="
},
{
"operation": "Now on",
"time": "8.761",
"value": "?search="
},
{
"operation": "search",
"time": "105.043",
"value": "Naichi"
},
{
"operation": "Now on",
"time": "105.888",
"value": "Naichi"
},
{
"operation": "Highlighting",
"time": "119.583",
"value": "Mainland Japan_sentence_1"
},
{
"operation": "Highlighting",
"time": "133.949",
"value": "Mainland Japan_sentence_2"
},
{
"operation": "search",
"time": "496.215",
"value": "Naichi"
},
{
"operation": "Now on",
"time": "497.288",
"value": "Naichi"
},
{
"operation": "Highlighting deleted",
"time": "526.237",
"value": "Mainland Japan_sentence_2"
},
{
"operation": "Highlighting deleted",
"time": "526.724",
"value": "Mainland Japan_sentence_1"
},
{
"operation": "Highlighting",
"time": "527.772",
"value": "Mainland Japan_sentence_0"
},
{
"operation": "Highlighting",
"time": "528.781",
"value": "Mainland Japan_sentence_1"
},
{
"operation": "Highlighting",
"time": "529.34",
"value": "Mainland Japan_sentence_2"
},
{
"operation": "finish",
"time": "1487.151",
"value": "finish"
}
]
}
|
27534
|
Rapid economic development in Hong Kong during the 1970s is attributed to a surge in the labor force.
|
{
"evidence": [
"1970s in Hong Kong_sentence_0",
"1970s in Hong Kong_sentence_1",
"1970s in Hong Kong_sentence_94"
],
"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:\nRapid economic development in Hong Kong during the 1970s is attributed to a surge in the labor force.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"1970s in Hong Kong_sentence_0",
"1970s in Hong Kong_sentence_1",
"1970s in Hong Kong_sentence_94"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 408,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.067",
"value": "?search="
},
{
"operation": "search",
"time": "60.984",
"value": "Economy of Hong Kong"
},
{
"operation": "Now on",
"time": "62.71",
"value": "Economy of Hong Kong"
},
{
"operation": "Page search",
"time": "118.551",
"value": "Rapid economic development"
},
{
"operation": "search",
"time": "141.74",
"value": "Rapid economic development in Hong Kong during the 1970s is attributed to a surge in the labor force"
},
{
"operation": "Now on",
"time": "145.019",
"value": "undefined"
},
{
"operation": "back-button-clicked",
"time": "223.249",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "223.817",
"value": "Economy of Hong Kong"
},
{
"operation": "Page search",
"time": "230.858",
"value": "1970s"
},
{
"operation": "search",
"time": "259.317",
"value": "1970s in Hong Kong"
},
{
"operation": "Now on",
"time": "260.729",
"value": "1970s in Hong Kong"
},
{
"operation": "Highlighting",
"time": "264.728",
"value": "1970s in Hong Kong_sentence_0"
},
{
"operation": "Highlighting",
"time": "266.217",
"value": "1970s in Hong Kong_sentence_1"
},
{
"operation": "Page search",
"time": "283.065",
"value": "economic development"
},
{
"operation": "Page search",
"time": "301.051",
"value": "labor"
},
{
"operation": "hyperlink",
"time": "356.672",
"value": "5 Economy 1970s in Hong Kong_section_13"
},
{
"operation": "Highlighting",
"time": "390.386",
"value": "1970s in Hong Kong_sentence_94"
},
{
"operation": "finish",
"time": "409.964",
"value": "finish"
}
]
}
|
6022
|
Jillian Armsbury, a singer, performer, songwriter and activist who was born and raised in Washington and Oregon, started participating in community theatre at the age of nine and studied dance for several years.
|
{
"evidence": [
"Jillian Armsbury_sentence_0",
"Jillian Armsbury_sentence_6",
"Jillian Armsbury_sentence_8",
"Jillian Armsbury_sentence_9"
],
"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:\nJillian Armsbury, a singer, performer, songwriter and activist who was born and raised in Washington and Oregon, started participating in community theatre at the age of nine and studied dance for several years.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Jillian Armsbury_sentence_0",
"Jillian Armsbury_sentence_6",
"Jillian Armsbury_sentence_8",
"Jillian Armsbury_sentence_9"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 409,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.531",
"value": "?search="
},
{
"operation": "search",
"time": "16.044",
"value": "Contains...Jillian Armsbury"
},
{
"operation": "Now on",
"time": "17.053",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "30.858",
"value": "Jillian Armsbury"
},
{
"operation": "Now on",
"time": "31.628",
"value": "Jillian Armsbury"
},
{
"operation": "Highlighting",
"time": "43.05",
"value": "Jillian Armsbury_sentence_0"
},
{
"operation": "Highlighting",
"time": "62.361",
"value": "Jillian Armsbury_sentence_6"
},
{
"operation": "Highlighting",
"time": "68.312",
"value": "Jillian Armsbury_sentence_8"
},
{
"operation": "Highlighting",
"time": "69.425",
"value": "Jillian Armsbury_sentence_9"
},
{
"operation": "finish",
"time": "83.071",
"value": "finish"
}
]
}
|
exfever_2662
|
The Belko Experiment is an Australian horror-thriller film starring an American actor and musician known for originating the role of Moritz Stiefel in Duncan Sheik and Steven Sater's rock musical Spring Awakening.
|
{
"evidence": [
"The Belko Experiment",
"John Gallagher Jr."
],
"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 Belko Experiment is an Australian horror-thriller film starring an American actor and musician known for originating the role of Moritz Stiefel in Duncan Sheik and Steven Sater's rock musical Spring Awakening.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Belko Experiment",
"John Gallagher Jr."
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 410,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7146
|
Stardust is a 2007 romantic fantasy adventure film based on a 1999 novel by an English author of short fiction and novels who was born on November 10, 1960.
|
{
"evidence": [
"Stardust (2007 film)",
"Neil Gaiman"
],
"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:\nStardust is a 2007 romantic fantasy adventure film based on a 1999 novel by an English author of short fiction and novels who was born on November 10, 1960.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Stardust (2007 film)",
"Neil Gaiman"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 411,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
2606
|
Despite offers from around Europe, James stayed in Apulia for 2014–15, and would later make his European debut, along with the club, in the Europe-wide third tier EuroChallenge.
|
{
"evidence": [
"Delroy James_sentence_38",
"Delroy James_sentence_39"
],
"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:\nDespite offers from around Europe, James stayed in Apulia for 2014–15, and would later make his European debut, along with the club, in the Europe-wide third tier EuroChallenge.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Delroy James_sentence_38",
"Delroy James_sentence_39"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 412,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "search",
"time": "189.294",
"value": "delroy james"
},
{
"operation": "Page search",
"time": "196.285",
"value": "offer"
},
{
"operation": "Highlighting",
"time": "204.231",
"value": "Delroy James_sentence_8"
},
{
"operation": "Highlighting",
"time": "204.247",
"value": "Delroy James_section_1"
},
{
"operation": "Highlighting",
"time": "208.614",
"value": "Delroy James_sentence_38"
},
{
"operation": "Highlighting deleted",
"time": "211.26",
"value": "Delroy James_sentence_8"
},
{
"operation": "Highlighting deleted",
"time": "211.287",
"value": "Delroy James_section_1"
},
{
"operation": "Highlighting",
"time": "216.164",
"value": "Delroy James_sentence_39"
},
{
"operation": "finish",
"time": "227.123",
"value": "finish"
}
]
}
|
exfever_7794
|
Boyd Holbrook is an actor who starred in a television series based on the story of the founder and sole leader of the Medellín Cartel.
|
{
"evidence": [
"Boyd Holbrook",
"Narcos",
"Pablo Escobar"
],
"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:\nBoyd Holbrook is an actor who starred in a television series based on the story of the founder and sole leader of the Medellín Cartel.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Boyd Holbrook",
"Narcos",
"Pablo Escobar"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 413,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_3876
|
Aishwarya Rai is the fourth-highest-paid actress in Asia.
|
{
"evidence": [
"Aishwarya Rai",
"Filmfare Awards",
"India"
],
"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:\nAishwarya Rai is the fourth-highest-paid actress in Asia.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Aishwarya Rai",
"Filmfare Awards",
"India"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 414,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_8050
|
Canada is the wealthiest country in the world.
|
{
"evidence": [
"Canada",
"Ottawa"
],
"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:\nCanada is the wealthiest country in the world.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Canada",
"Ottawa"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 415,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_1070
|
Poland is a country in a continent located entirely in the Northern Hemisphere and mostly in the Northern Hemisphere.
|
{
"evidence": [
"Poland",
"Europe"
],
"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:\nPoland is a country in a continent located entirely in the Northern Hemisphere and mostly in the Northern Hemisphere.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Poland",
"Europe"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 416,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
68783
|
Afradapis is a genus of adapiform primate and is one of four typically European caenopithecines (a family within the extinct primate family Adapidae) it also lost its anterior premolar
|
{
"evidence": [
"Afradapis_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:\nAfradapis is a genus of adapiform primate and is one of four typically European caenopithecines (a family within the extinct primate family Adapidae) it also lost its anterior premolar\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Afradapis_sentence_1"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 417,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.755",
"value": "?search="
},
{
"operation": "search",
"time": "10.417",
"value": "Afradapis"
},
{
"operation": "Now on",
"time": "11.378",
"value": "Afradapis"
},
{
"operation": "Highlighting",
"time": "16.796",
"value": "Afradapis_sentence_0"
},
{
"operation": "Highlighting",
"time": "21.395",
"value": "Afradapis_sentence_1"
},
{
"operation": "Highlighting deleted",
"time": "23.883",
"value": "Afradapis_sentence_0"
},
{
"operation": "finish",
"time": "30.329",
"value": "finish"
}
]
}
|
4908
|
Physiologic phimosis, common in males five years of age and younger, is normal, and does require intervention and non-retractile foreskin usually becomes retractable during the course of puberty.
|
{
"evidence": [
"Phimosis_sentence_35"
],
"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:\nPhysiologic phimosis, common in males five years of age and younger, is normal, and does require intervention and non-retractile foreskin usually becomes retractable during the course of puberty.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Phimosis_sentence_35"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 418,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "search",
"time": "48.901",
"value": "physiologic phimosis"
},
{
"operation": "hyperlink",
"time": "69.164",
"value": "Paraphimosis"
},
{
"operation": "Page search",
"time": "83.911",
"value": "physiologic phimosis"
},
{
"operation": "Page search",
"time": "1161.155",
"value": "phimosis"
},
{
"operation": "Page search",
"time": "1174.5",
"value": "physiologic phimosis"
},
{
"operation": "hyperlink",
"time": "1243.472",
"value": "Phimosis"
},
{
"operation": "Page search",
"time": "1256.067",
"value": "physiologic"
},
{
"operation": "Highlighting",
"time": "1285.206",
"value": "Phimosis_sentence_35"
},
{
"operation": "finish",
"time": "1291.501",
"value": "finish"
}
]
}
|
32019
|
Guye District in China does not include the city of Tangshan.
|
{
"evidence": [
"Guye District_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:\nGuye District in China does not include the city of Tangshan.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"Guye District_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 419,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.36",
"value": "?search="
},
{
"operation": "search",
"time": "15.741",
"value": "Guye District"
},
{
"operation": "Now on",
"time": "17.036",
"value": "Guye District"
},
{
"operation": "Highlighting",
"time": "21.536",
"value": "Guye District_sentence_0"
},
{
"operation": "finish",
"time": "34.83",
"value": "finish"
}
]
}
|
40428
|
Igor Kenk, born on April 7, 1959 in Maribor, PR Slovenia, FPR Yugoslavia, has 16 theft charges and a penalty of 30 months in jail.
|
{
"evidence": [
"Igor Kenk_sentence_0",
"Igor Kenk_sentence_14",
"Igor Kenk_sentence_16"
],
"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:\nIgor Kenk, born on April 7, 1959 in Maribor, PR Slovenia, FPR Yugoslavia, has 16 theft charges and a penalty of 30 months in jail.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"Igor Kenk_sentence_0",
"Igor Kenk_sentence_14",
"Igor Kenk_sentence_16"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 420,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.374",
"value": "?search="
},
{
"operation": "search",
"time": "32.767",
"value": "Igor Kenk"
},
{
"operation": "Now on",
"time": "34.049",
"value": "Igor Kenk"
},
{
"operation": "Highlighting",
"time": "48.535",
"value": "Igor Kenk_sentence_0"
},
{
"operation": "Highlighting",
"time": "63.469",
"value": "Igor Kenk_sentence_4"
},
{
"operation": "Highlighting",
"time": "71.548",
"value": "Igor Kenk_cell_0_1_1"
},
{
"operation": "Highlighting",
"time": "71.551",
"value": "Igor Kenk_header_cell_0_1_0"
},
{
"operation": "Highlighting",
"time": "71.551",
"value": "Igor Kenk_header_cell_0_0_0"
},
{
"operation": "Highlighting deleted",
"time": "74.638",
"value": "Igor Kenk_cell_0_1_1"
},
{
"operation": "Highlighting deleted",
"time": "74.64",
"value": "Igor Kenk_header_cell_0_1_0"
},
{
"operation": "Highlighting deleted",
"time": "74.641",
"value": "Igor Kenk_header_cell_0_0_0"
},
{
"operation": "Highlighting deleted",
"time": "76.333",
"value": "Igor Kenk_sentence_4"
},
{
"operation": "Highlighting",
"time": "101.734",
"value": "Igor Kenk_sentence_14"
},
{
"operation": "Highlighting",
"time": "106.845",
"value": "Igor Kenk_sentence_16"
},
{
"operation": "finish",
"time": "221.721",
"value": "finish"
}
]
}
|
exfever_3841
|
Jacob Benjamin Gyllenhaal was a successful American actor for Spider-Man: Far From Home (2019).
|
{
"evidence": [
"Jake Gyllenhaal",
"Homer Hickam"
],
"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:\nJacob Benjamin Gyllenhaal was a successful American actor for Spider-Man: Far From Home (2019).\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Jake Gyllenhaal",
"Homer Hickam"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 421,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_8372
|
Ella Purnell is a French actress best known for her performance as Gwyn in the animated series Star Trek: Prodigy on an American pay television channel that launched on April 1, 1979, as the first cable channel for children.
|
{
"evidence": [
"Ella Purnell",
"Nickelodeon"
],
"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:\nElla Purnell is a French actress best known for her performance as Gwyn in the animated series Star Trek: Prodigy on an American pay television channel that launched on April 1, 1979, as the first cable channel for children.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Ella Purnell",
"Nickelodeon"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 422,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
66013
|
Émile Marie Laurent served in the military from October 25, 1875 to October 5, 1877, but failed to be appointed as the Prefect of Police of Paris.
|
{
"evidence": [
"Émile Marie Laurent_sentence_0",
"Émile Marie Laurent_sentence_3"
],
"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:\nÉmile Marie Laurent served in the military from October 25, 1875 to October 5, 1877, but failed to be appointed as the Prefect of Police of Paris.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Émile Marie Laurent_sentence_0",
"Émile Marie Laurent_sentence_3"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 423,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.218",
"value": "?search="
},
{
"operation": "search",
"time": "14.889",
"value": "Emile Marie Laurent"
},
{
"operation": "Now on",
"time": "15.649",
"value": "Émile Marie Laurent"
},
{
"operation": "Highlighting",
"time": "17.817",
"value": "Émile Marie Laurent_sentence_0"
},
{
"operation": "Highlighting",
"time": "26.161",
"value": "Émile Marie Laurent_sentence_3"
},
{
"operation": "finish",
"time": "36.984",
"value": "finish"
}
]
}
|
45275
|
O'Connor and Krans began writing and performing together in 1998, friends began referring to them as "Jeneric" but they never moved.
|
{
"evidence": [
"The Parlor_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:\nO'Connor and Krans began writing and performing together in 1998, friends began referring to them as \"Jeneric\" but they never moved.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"The Parlor_sentence_8"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 424,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.36",
"value": "?search="
},
{
"operation": "search",
"time": "13.268",
"value": "Jeneric"
},
{
"operation": "Now on",
"time": "14.008",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "17.927",
"value": "The Parlor"
},
{
"operation": "Now on",
"time": "19.157",
"value": "The Parlor"
},
{
"operation": "Highlighting",
"time": "21.087",
"value": "The Parlor_sentence_0"
},
{
"operation": "Highlighting deleted",
"time": "24.406",
"value": "The Parlor_sentence_0"
},
{
"operation": "Highlighting",
"time": "36.032",
"value": "The Parlor_sentence_5"
},
{
"operation": "Highlighting",
"time": "39.999",
"value": "The Parlor_sentence_7"
},
{
"operation": "Highlighting",
"time": "48.61",
"value": "The Parlor_sentence_8"
},
{
"operation": "Highlighting deleted",
"time": "49.621",
"value": "The Parlor_sentence_7"
},
{
"operation": "Highlighting deleted",
"time": "50.502",
"value": "The Parlor_sentence_5"
},
{
"operation": "finish",
"time": "73.88",
"value": "finish"
}
]
}
|
53055
|
Bartłomiej Nowodworski was born 1552 in Krakow, in recognition of his contribution the school became known as Collegium Nowodworskiego and in 1898 the school relocated from the university area.
|
{
"evidence": [
"Bartłomiej Nowodworski High School_sentence_6",
"Bartłomiej Nowodworski High School_sentence_7",
"Bartłomiej Nowodworski High School_sentence_9"
],
"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:\nBartłomiej Nowodworski was born 1552 in Krakow, in recognition of his contribution the school became known as Collegium Nowodworskiego and in 1898 the school relocated from the university area.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Bartłomiej Nowodworski High School_sentence_6",
"Bartłomiej Nowodworski High School_sentence_7",
"Bartłomiej Nowodworski High School_sentence_9"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 425,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.729",
"value": "?search="
},
{
"operation": "search",
"time": "4.938",
"value": "Bartłomiej Nowodworski"
},
{
"operation": "Now on",
"time": "6.919",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "9.722",
"value": "Agata Kornhauser-Duda"
},
{
"operation": "Now on",
"time": "14.03",
"value": "Bartłomiej Nowodworski High School"
},
{
"operation": "Highlighting",
"time": "27.878",
"value": "Bartłomiej Nowodworski High School_sentence_6"
},
{
"operation": "Highlighting",
"time": "56.705",
"value": "Bartłomiej Nowodworski High School_sentence_7"
},
{
"operation": "Highlighting",
"time": "78.245",
"value": "Bartłomiej Nowodworski High School_sentence_9"
},
{
"operation": "finish",
"time": "130.17",
"value": "finish"
}
]
}
|
exfever_8970
|
At the 2008 Brit Awards, Kate Marie Nash won the award that was first presented in 1977, given by the British recorded music industry's Trade Association.
|
{
"evidence": [
"Kate Nash",
"Brit Award for British Female Solo Artist",
"British Phonographic Industry"
],
"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:\nAt the 2008 Brit Awards, Kate Marie Nash won the award that was first presented in 1977, given by the British recorded music industry's Trade Association.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Kate Nash",
"Brit Award for British Female Solo Artist",
"British Phonographic Industry"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 426,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
88013
|
Sillamäe was developed into a resort village.
|
{
"evidence": [
"Sillamäe_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:\nSillamäe was developed into a resort village.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Sillamäe_sentence_5"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 427,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.306",
"value": "undefined"
},
{
"operation": "search",
"time": "10.771",
"value": "Sillamäe"
},
{
"operation": "Now on",
"time": "12.215",
"value": "Sillamäe"
},
{
"operation": "Highlighting",
"time": "24.19",
"value": "Sillamäe_sentence_5"
},
{
"operation": "finish",
"time": "38.27",
"value": "finish"
}
]
}
|
exfever_2577
|
Rob Cavallo is primarily known for his production work with Green Day and has also worked with?a Canadian rock band formed in Charlottesville, Virginia..
|
{
"evidence": [
"Rob Cavallo",
"Dave Matthews 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:\nRob Cavallo is primarily known for his production work with Green Day and has also worked with?a Canadian rock band formed in Charlottesville, Virginia..\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Rob Cavallo",
"Dave Matthews Band"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 428,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
80307
|
The Convincer was released in United States by Yep Roc Records.
|
{
"evidence": [
"The Convincer_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:\nThe Convincer was released in United States by Yep Roc Records.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"The Convincer_sentence_2"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 429,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.739",
"value": "?search="
},
{
"operation": "search",
"time": "11.278",
"value": "The Convincer"
},
{
"operation": "Now on",
"time": "11.893",
"value": "The Convincer"
},
{
"operation": "Highlighting",
"time": "24.242",
"value": "The Convincer_sentence_2"
},
{
"operation": "finish",
"time": "276.137",
"value": "finish"
}
]
}
|
exfever_8486
|
Reba Nell McEntire is an American country music singer who was interested during her childhood in a rodeo event in which a horse and rider attempted to run a cloverleaf pattern around preset barrels in the fastest time.
|
{
"evidence": [
"Reba McEntire",
"Barrel racing"
],
"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:\nReba Nell McEntire is an American country music singer who was interested during her childhood in a rodeo event in which a horse and rider attempted to run a cloverleaf pattern around preset barrels in the fastest time.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Reba McEntire",
"Barrel racing"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 430,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
46689
|
Shortly after parting ways with Lava/Republic in 2013, Tiz partnered with songwriting/production duo Rock Mafia; then released a cover of Rock Mafia's song "The Big Bang", which debuted on the Billboard Hot 100 on May 17, 2014.
|
{
"evidence": [
"Katy Tiz_sentence_5",
"Katy Tiz_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:\nShortly after parting ways with Lava/Republic in 2013, Tiz partnered with songwriting/production duo Rock Mafia; then released a cover of Rock Mafia's song \"The Big Bang\", which debuted on the Billboard Hot 100 on May 17, 2014.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"Katy Tiz_sentence_5",
"Katy Tiz_sentence_6"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 431,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.167",
"value": "?search="
},
{
"operation": "search",
"time": "53.717",
"value": "The Big Bang (song)"
},
{
"operation": "Now on",
"time": "54.731",
"value": "The Big Bang (song)"
},
{
"operation": "hyperlink",
"time": "61.073",
"value": "Katy Tiz"
},
{
"operation": "Now on",
"time": "61.821",
"value": "Katy Tiz"
},
{
"operation": "Highlighting",
"time": "66.06",
"value": "Katy Tiz_sentence_5"
},
{
"operation": "Highlighting",
"time": "72.778",
"value": "Katy Tiz_sentence_6"
},
{
"operation": "finish",
"time": "87.022",
"value": "finish"
}
]
}
|
exfever_5741
|
Katherine Marie Heigl is an American actress, producer and former fashion model who appeared in a 1990 French-American comedy-drama directed by Steve Miner.
|
{
"evidence": [
"Katherine Heigl",
"My Father the Hero (1994 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:\nKatherine Marie Heigl is an American actress, producer and former fashion model who appeared in a 1990 French-American comedy-drama directed by Steve Miner.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Katherine Heigl",
"My Father the Hero (1994 film)"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 432,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_8168
|
Neighbors is a 2014 American comedy film directed by a British-born American director, producer, and screenwriter.
|
{
"evidence": [
"Neighbors (2014 film)",
"Nicholas Stoller"
],
"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:\nNeighbors is a 2014 American comedy film directed by a British-born American director, producer, and screenwriter.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Neighbors (2014 film)",
"Nicholas Stoller"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 433,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_1550
|
Anna Gunn starred in the film Bachelor Party.
|
{
"evidence": [
"Sully (film)",
"Tom Hanks"
],
"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:\nAnna Gunn starred in the film Bachelor Party.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Sully (film)",
"Tom Hanks"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 434,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
40262
|
Faridnagar is a town in the district of Ghaziabad in the Indian state of Uttar Pradesh, the almost deserted state in India.
|
{
"evidence": [
"Uttar Pradesh_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:\nFaridnagar is a town in the district of Ghaziabad in the Indian state of Uttar Pradesh, the almost deserted state in India.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Uttar Pradesh_sentence_1"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 435,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.859",
"value": "?search="
},
{
"operation": "search",
"time": "17.346",
"value": "Faridnagar"
},
{
"operation": "Now on",
"time": "18.539",
"value": "Faridnagar"
},
{
"operation": "Highlighting",
"time": "21.081",
"value": "Faridnagar_sentence_0"
},
{
"operation": "Highlighting",
"time": "21.082",
"value": "Faridnagar_title"
},
{
"operation": "hyperlink",
"time": "39.962",
"value": "Uttar Pradesh"
},
{
"operation": "Now on",
"time": "42.307",
"value": "Uttar Pradesh"
},
{
"operation": "Highlighting",
"time": "47.561",
"value": "Uttar Pradesh_sentence_1"
},
{
"operation": "back-button-clicked",
"time": "49.113",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "49.213",
"value": "Faridnagar"
},
{
"operation": "Highlighting deleted",
"time": "50.216",
"value": "Faridnagar_sentence_0"
},
{
"operation": "Highlighting deleted",
"time": "50.218",
"value": "Faridnagar_title"
},
{
"operation": "finish",
"time": "75.214",
"value": "finish"
}
]
}
|
exfever_1368
|
The Grey Album gained notoriety when EMI attempted to halt its distribution despite approval of the project from Jay-Z and one of the members of the Beatles, an English singer who is one of the most successful composers and performers of all time.
|
{
"evidence": [
"The Grey Album",
"Paul McCartney"
],
"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 Grey Album gained notoriety when EMI attempted to halt its distribution despite approval of the project from Jay-Z and one of the members of the Beatles, an English singer who is one of the most successful composers and performers of all time.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Grey Album",
"Paul McCartney"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 436,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_6415
|
Planet of the Apes is a 1968 science fiction film starring an American theater actress known for portraying Stella Kowalski in A Streetcar Named Desire.
|
{
"evidence": [
"Planet of the Apes (1968 film)",
"Kim Hunter"
],
"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:\nPlanet of the Apes is a 1968 science fiction film starring an American theater actress known for portraying Stella Kowalski in A Streetcar Named Desire.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Planet of the Apes (1968 film)",
"Kim Hunter"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 437,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
32314
|
1506 Xosa, discovered by C. Jackson (a South African astronomer who failed to graduate at the University of the Witwatersrand) on May 15, 1939 in Johannesburg Obs., was named after the Xhosa people, a Bantu ethnic group.
|
{
"evidence": [
"Cyril Jackson (astronomer)_sentence_2",
"Cyril Jackson (astronomer)_sentence_3"
],
"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:\n1506 Xosa, discovered by C. Jackson (a South African astronomer who failed to graduate at the University of the Witwatersrand) on May 15, 1939 in Johannesburg Obs., was named after the Xhosa people, a Bantu ethnic group.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Cyril Jackson (astronomer)_sentence_2",
"Cyril Jackson (astronomer)_sentence_3"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 438,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.528",
"value": "?search="
},
{
"operation": "search",
"time": "14.332",
"value": "1506 Xosa"
},
{
"operation": "Now on",
"time": "15.406",
"value": "1506 Xosa"
},
{
"operation": "Highlighting",
"time": "210.191",
"value": "1506 Xosa_cell_0_1_1"
},
{
"operation": "Highlighting",
"time": "210.193",
"value": "1506 Xosa_title"
},
{
"operation": "Highlighting",
"time": "210.205",
"value": "1506 Xosa_header_cell_0_1_0"
},
{
"operation": "Highlighting",
"time": "210.205",
"value": "1506 Xosa_header_cell_0_0_0"
},
{
"operation": "Highlighting",
"time": "215.173",
"value": "1506 Xosa_sentence_3"
},
{
"operation": "hyperlink",
"time": "250.522",
"value": "C. Jackson"
},
{
"operation": "Now on",
"time": "251.343",
"value": "Cyril Jackson (astronomer)"
},
{
"operation": "Highlighting",
"time": "261.951",
"value": "Cyril Jackson (astronomer)_sentence_2"
},
{
"operation": "Highlighting",
"time": "262.759",
"value": "Cyril Jackson (astronomer)_sentence_3"
},
{
"operation": "Highlighting",
"time": "267.908",
"value": "Cyril Jackson (astronomer)_sentence_2"
},
{
"operation": "Highlighting",
"time": "268.627",
"value": "Cyril Jackson (astronomer)_sentence_3"
},
{
"operation": "finish",
"time": "329.505",
"value": "finish"
}
]
}
|
exfever_5202
|
Buzz Aldrin was the youngest American fighter pilot.
|
{
"evidence": [
"Buzz Aldrin",
"Neil Armstrong",
"Wapakoneta, Ohio"
],
"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:\nBuzz Aldrin was the youngest American fighter pilot.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Buzz Aldrin",
"Neil Armstrong",
"Wapakoneta, Ohio"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 439,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
34012
|
Kolntriangle is home to various businesses and the top floor has an apartment.
|
{
"evidence": [
"Kölntriangle_sentence_0",
"Kölntriangle_sentence_3"
],
"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:\nKolntriangle is home to various businesses and the top floor has an apartment.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Kölntriangle_sentence_0",
"Kölntriangle_sentence_3"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 440,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.855",
"value": "undefined"
},
{
"operation": "search",
"time": "8.808",
"value": "Kolntriangle"
},
{
"operation": "Now on",
"time": "10.13",
"value": "Kölntriangle"
},
{
"operation": "Highlighting",
"time": "33.938",
"value": "Kölntriangle_sentence_0"
},
{
"operation": "Highlighting deleted",
"time": "65.657",
"value": "Kölntriangle_sentence_0"
},
{
"operation": "Highlighting",
"time": "70.015",
"value": "Kölntriangle_sentence_0"
},
{
"operation": "search",
"time": "75.79",
"value": "LVR-Turm"
},
{
"operation": "Now on",
"time": "76.748",
"value": "LVR-Turm"
},
{
"operation": "Highlighting",
"time": "90.418",
"value": "Kölntriangle_sentence_3"
},
{
"operation": "search",
"time": "104.964",
"value": "KölnTriangle"
},
{
"operation": "Now on",
"time": "105.944",
"value": "KölnTriangle"
},
{
"operation": "finish",
"time": "148.238",
"value": "finish"
}
]
}
|
exfever_8448
|
Bradley Charles Cooper is the first American actor who receives a British Academy Film Award.
|
{
"evidence": [
"Bradley Cooper",
"Academy Awards",
"Douglas Fairbanks"
],
"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:\nBradley Charles Cooper is the first American actor who receives a British Academy Film Award.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Bradley Cooper",
"Academy Awards",
"Douglas Fairbanks"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 441,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
92465
|
A joint citizen of Canada and Morocco, a country in the Maghreb region of North Africa, who was involved in forging a passport and became a computer technician while facing extradition, was Abdellah Ouzghar.
|
{
"evidence": [
"Abdellah Ouzghar_sentence_0",
"Abdellah Ouzghar_sentence_6",
"Abdellah Ouzghar_sentence_8",
"Morocco_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:\nA joint citizen of Canada and Morocco, a country in the Maghreb region of North Africa, who was involved in forging a passport and became a computer technician while facing extradition, was Abdellah Ouzghar.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Abdellah Ouzghar_sentence_0",
"Abdellah Ouzghar_sentence_6",
"Abdellah Ouzghar_sentence_8",
"Morocco_sentence_4"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 442,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.138",
"value": "?search="
},
{
"operation": "search",
"time": "15.028",
"value": "Abdellah Ouzghar"
},
{
"operation": "Now on",
"time": "15.692",
"value": "Abdellah Ouzghar"
},
{
"operation": "Highlighting",
"time": "17.362",
"value": "Abdellah Ouzghar_sentence_0"
},
{
"operation": "Highlighting",
"time": "23.058",
"value": "Abdellah Ouzghar_sentence_1"
},
{
"operation": "Highlighting deleted",
"time": "44.866",
"value": "Abdellah Ouzghar_sentence_1"
},
{
"operation": "Highlighting",
"time": "46.426",
"value": "Abdellah Ouzghar_sentence_6"
},
{
"operation": "Highlighting",
"time": "54.538",
"value": "Abdellah Ouzghar_sentence_8"
},
{
"operation": "hyperlink",
"time": "62.809",
"value": "Morocco"
},
{
"operation": "Now on",
"time": "65.745",
"value": "Morocco"
},
{
"operation": "Highlighting",
"time": "68.906",
"value": "Morocco_sentence_4"
},
{
"operation": "finish",
"time": "77.857",
"value": "finish"
}
]
}
|
89836
|
The Royal National City Park is located in three municipalities in Sweden, including Stockholm and Solna.
|
{
"evidence": [
"Royal National City Park_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 Royal National City Park is located in three municipalities in Sweden, including Stockholm and Solna.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Royal National City Park_sentence_0"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 443,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.447",
"value": "?search="
},
{
"operation": "search",
"time": "12.765",
"value": "Royal National City Park"
},
{
"operation": "Now on",
"time": "13.562",
"value": "Royal National City Park"
},
{
"operation": "Highlighting",
"time": "71.889",
"value": "Royal National City Park_sentence_0"
},
{
"operation": "finish",
"time": "134.465",
"value": "finish"
}
]
}
|
exfever_6935
|
Black Panther is a film that stars an American actor and playwright who had a breakthrough performance as baseball player Jackie Robinson in a 2013 American biographical sports film.
|
{
"evidence": [
"Black Panther (film)",
"Chadwick Boseman",
"42 (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:\nBlack Panther is a film that stars an American actor and playwright who had a breakthrough performance as baseball player Jackie Robinson in a 2013 American biographical sports film.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Black Panther (film)",
"Chadwick Boseman",
"42 (film)"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 444,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
63317
|
Duit on Mon Dei by Ringo Starr received reviews from three sources, one of which was The Essential Rock Discography which gave it a 4/10.
|
{
"evidence": [
"Duit on Mon Dei_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:\nDuit on Mon Dei by Ringo Starr received reviews from three sources, one of which was The Essential Rock Discography which gave it a 4/10.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Duit on Mon Dei_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 445,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "6.137",
"value": "?search="
},
{
"operation": "search",
"time": "12.129",
"value": "Duit on Mon Dei"
},
{
"operation": "Now on",
"time": "13.136",
"value": "Duit on Mon Dei"
},
{
"operation": "Highlighting",
"time": "21.572",
"value": "Duit on Mon Dei_sentence_0"
},
{
"operation": "search",
"time": "35.367",
"value": "Contains...Duit on Mon Dei by Ringo Starr"
},
{
"operation": "Now on",
"time": "44.936",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "53.577",
"value": "Duit on Mon Dei"
},
{
"operation": "Now on",
"time": "54.203",
"value": "Duit on Mon Dei"
},
{
"operation": "finish",
"time": "61.932",
"value": "finish"
}
]
}
|
80649
|
Weaver Street Market is located in Carrboro, North Carolina.
|
{
"evidence": [
"Weaver Street Market_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:\nWeaver Street Market is located in Carrboro, North Carolina.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Weaver Street Market_sentence_1"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 446,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.87",
"value": "?search="
},
{
"operation": "search",
"time": "17.367",
"value": "Weaver Street Market"
},
{
"operation": "Now on",
"time": "18.088",
"value": "Weaver Street Market"
},
{
"operation": "Highlighting",
"time": "21.802",
"value": "Weaver Street Market_sentence_1"
},
{
"operation": "finish",
"time": "82.9",
"value": "finish"
}
]
}
|
2238
|
According to Eugene Moehring, the Sahara Las Vegas has helped establish LA as a major resort city after the World War II, but as the 1970s went by, the hotel seemed to be getting older.
|
{
"evidence": [
"Sahara Las Vegas_sentence_146",
"Sahara Las Vegas_sentence_151",
"Sahara Las Vegas_sentence_155"
],
"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:\nAccording to Eugene Moehring, the Sahara Las Vegas has helped establish LA as a major resort city after the World War II, but as the 1970s went by, the hotel seemed to be getting older.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Sahara Las Vegas_sentence_146",
"Sahara Las Vegas_sentence_151",
"Sahara Las Vegas_sentence_155"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 447,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "search",
"time": "19.209",
"value": "Sahara Las Vegas"
},
{
"operation": "Page search",
"time": "50.126",
"value": "moehring"
},
{
"operation": "Highlighting",
"time": "61.424",
"value": "Sahara Las Vegas_sentence_146"
},
{
"operation": "Page search",
"time": "84.216",
"value": "older"
},
{
"operation": "Highlighting deleted",
"time": "104.42",
"value": "Sahara Las Vegas_sentence_146"
},
{
"operation": "Highlighting",
"time": "106.601",
"value": "Sahara Las Vegas_sentence_146"
},
{
"operation": "Highlighting",
"time": "116.354",
"value": "Sahara Las Vegas_sentence_148"
},
{
"operation": "Highlighting deleted",
"time": "154.051",
"value": "Sahara Las Vegas_sentence_148"
},
{
"operation": "Highlighting",
"time": "191.21",
"value": "Sahara Las Vegas_sentence_156"
},
{
"operation": "Highlighting",
"time": "237.734",
"value": "Sahara Las Vegas_sentence_151"
},
{
"operation": "Highlighting",
"time": "252.28",
"value": "Sahara Las Vegas_sentence_155"
},
{
"operation": "Highlighting deleted",
"time": "281.325",
"value": "Sahara Las Vegas_sentence_156"
},
{
"operation": "finish",
"time": "303.792",
"value": "finish"
}
]
}
|
94644
|
Sydney Rae White is the frontman of the band called The Wild Things.
|
{
"evidence": [
"Sydney Rae White_sentence_3"
],
"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:\nSydney Rae White is the frontman of the band called The Wild Things.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Sydney Rae White_sentence_3"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 448,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.283",
"value": "?search="
},
{
"operation": "search",
"time": "9.483",
"value": "Sydney Rae White"
},
{
"operation": "Now on",
"time": "10.572",
"value": "Sydney Rae White"
},
{
"operation": "Page search",
"time": "21.949",
"value": "Wild"
},
{
"operation": "Highlighting",
"time": "34.419",
"value": "Sydney Rae White_sentence_3"
},
{
"operation": "Highlighting",
"time": "38.948",
"value": "Sydney Rae White_sentence_5"
},
{
"operation": "Highlighting deleted",
"time": "42.828",
"value": "Sydney Rae White_sentence_5"
},
{
"operation": "Page search",
"time": "74.388",
"value": "frontman"
},
{
"operation": "search",
"time": "84.1",
"value": "The Wild Things"
},
{
"operation": "Now on",
"time": "85.536",
"value": "undefined"
},
{
"operation": "finish",
"time": "108.309",
"value": "finish"
}
]
}
|
exfever_3773
|
Cameron Jibril Thomaz is a rapper, singer, and songwriter signed to a record label founded by an American record executive and record producer.
|
{
"evidence": [
"Wiz Khalifa",
"Atlantic Records",
"Herb Abramson"
],
"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:\nCameron Jibril Thomaz is a rapper, singer, and songwriter signed to a record label founded by an American record executive and record producer.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Wiz Khalifa",
"Atlantic Records",
"Herb Abramson"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 449,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
56920
|
Andrea Ferrucci was an Italian sculptor and among his masterworks are the bas-reliefs surrounding the baptismal font at the Duomo di Amalfi, which is the main religious building in Vatican City, and the Basilica of the Holy Cross in Jerusalem cathedral possessed a marble reredos from his hand.
|
{
"evidence": [
"Andrea Ferrucci_sentence_7"
],
"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:\nAndrea Ferrucci was an Italian sculptor and among his masterworks are the bas-reliefs surrounding the baptismal font at the Duomo di Amalfi, which is the main religious building in Vatican City, and the Basilica of the Holy Cross in Jerusalem cathedral possessed a marble reredos from his hand.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Andrea Ferrucci_sentence_7"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 450,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.972",
"value": "?search="
},
{
"operation": "search",
"time": "23.537",
"value": "andrea ferrucci"
},
{
"operation": "Now on",
"time": "24.06",
"value": "Andrea Ferrucci"
},
{
"operation": "Highlighting",
"time": "31.552",
"value": "Andrea Ferrucci_sentence_0"
},
{
"operation": "Highlighting",
"time": "49.839",
"value": "Andrea Ferrucci_sentence_7"
},
{
"operation": "hyperlink",
"time": "63.623",
"value": "Duomo of Pistoia"
},
{
"operation": "Now on",
"time": "64.33",
"value": "Pistoia Cathedral"
},
{
"operation": "back-button-clicked",
"time": "71.047",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "71.094",
"value": "Andrea Ferrucci"
},
{
"operation": "Highlighting deleted",
"time": "91.758",
"value": "Andrea Ferrucci_sentence_0"
},
{
"operation": "finish",
"time": "119.797",
"value": "finish"
}
]
}
|
73452
|
Corrie Sanders notable achievements include holding the South Africa national heavyweight title and he was nicknamed The Sniper and Wladimir Klitschko labled him the softest puncher.
|
{
"evidence": [
"Corrie Sanders_sentence_7"
],
"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:\nCorrie Sanders notable achievements include holding the South Africa national heavyweight title and he was nicknamed The Sniper and Wladimir Klitschko labled him the softest puncher.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Corrie Sanders_sentence_7"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 451,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.763",
"value": "?search="
},
{
"operation": "search",
"time": "7.127",
"value": "Corrie Sanders"
},
{
"operation": "Now on",
"time": "10.35",
"value": "Corrie Sanders"
},
{
"operation": "Highlighting",
"time": "98.333",
"value": "Corrie Sanders_sentence_1"
},
{
"operation": "Highlighting deleted",
"time": "101.832",
"value": "Corrie Sanders_sentence_1"
},
{
"operation": "Highlighting",
"time": "128.775",
"value": "Corrie Sanders_sentence_6"
},
{
"operation": "Highlighting deleted",
"time": "135.818",
"value": "Corrie Sanders_sentence_6"
},
{
"operation": "Highlighting",
"time": "137.55",
"value": "Corrie Sanders_sentence_7"
},
{
"operation": "finish",
"time": "146.426",
"value": "finish"
}
]
}
|
82831
|
Classica Corsica, a legacy event of the 2013 Tour de France, is an elite men's road bicycle racing event held annually on the island of Corsica, a region of France.
|
{
"evidence": [
"Classica Corsica_sentence_2",
"Classica Corsica_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:\nClassica Corsica, a legacy event of the 2013 Tour de France, is an elite men's road bicycle racing event held annually on the island of Corsica, a region of France.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Classica Corsica_sentence_2",
"Classica Corsica_sentence_0"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 452,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.713",
"value": "?search="
},
{
"operation": "search",
"time": "14.536",
"value": "Classica Corsica"
},
{
"operation": "Now on",
"time": "15.457",
"value": "Classica Corsica"
},
{
"operation": "Highlighting",
"time": "26.079",
"value": "Classica Corsica_sentence_2"
},
{
"operation": "hyperlink",
"time": "27.066",
"value": "2013 Tour de France"
},
{
"operation": "Now on",
"time": "28.903",
"value": "2013 Tour de France"
},
{
"operation": "hyperlink",
"time": "32.608",
"value": "Tour de France"
},
{
"operation": "Now on",
"time": "35.917",
"value": "Tour de France"
},
{
"operation": "back-button-clicked",
"time": "54.461",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "54.673",
"value": "2013 Tour de France"
},
{
"operation": "back-button-clicked",
"time": "73.456",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "73.464",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "73.568",
"value": "Classica Corsica"
},
{
"operation": "Highlighting",
"time": "80.581",
"value": "Classica Corsica_sentence_0"
},
{
"operation": "finish",
"time": "89.923",
"value": "finish"
}
]
}
|
exfever_227
|
Oscars are awarded also to some musical artists.
|
{
"evidence": [
"Susan Sarandon",
"Academy Awards"
],
"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:\nOscars are awarded also to some musical artists.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Susan Sarandon",
"Academy Awards"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 453,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7011
|
Larry Buttrose is hired to write literary works that are officially credited to another person as the author.
|
{
"evidence": [
"Larry Buttrose",
"Ghostwriter"
],
"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:\nLarry Buttrose is hired to write literary works that are officially credited to another person as the author.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Larry Buttrose",
"Ghostwriter"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 454,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
23021
|
Narasimha Raya II became the Emperor of the Vijayanagara Empire when he was a teenager following the violent death of his father.
|
{
"evidence": [
"Narasimha Raya II_sentence_7",
"Narasimha Raya II_sentence_8",
"Narasimha Raya II_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:\nNarasimha Raya II became the Emperor of the Vijayanagara Empire when he was a teenager following the violent death of his father.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Narasimha Raya II_sentence_7",
"Narasimha Raya II_sentence_8",
"Narasimha Raya II_sentence_9"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 455,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.084",
"value": "?search="
},
{
"operation": "search",
"time": "18.528",
"value": "Narasimha Raya II"
},
{
"operation": "Now on",
"time": "19.716",
"value": "Narasimha Raya II"
},
{
"operation": "Highlighting",
"time": "139.966",
"value": "Narasimha Raya II_sentence_7"
},
{
"operation": "Highlighting",
"time": "141.446",
"value": "Narasimha Raya II_sentence_8"
},
{
"operation": "Highlighting",
"time": "147.128",
"value": "Narasimha Raya II_sentence_9"
},
{
"operation": "finish",
"time": "157.881",
"value": "finish"
}
]
}
|
exfever_3756
|
The West Wing is an American serial drama that can describe a TV program that has a political component.
|
{
"evidence": [
"The West Wing",
"Political drama"
],
"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 West Wing is an American serial drama that can describe a TV program that has a political component.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The West Wing",
"Political drama"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 456,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_2037
|
Sigmund Freud was an Austrian neurologist and the founder of psychoanalysis, born in Central-Eastern European multinational great power from 1804 to 1867.
|
{
"evidence": [
"Sigmund Freud",
"Austrian Empire"
],
"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:\nSigmund Freud was an Austrian neurologist and the founder of psychoanalysis, born in Central-Eastern European multinational great power from 1804 to 1867.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Sigmund Freud",
"Austrian Empire"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 457,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
11569
|
WCCG's 104.5 frequency was allocated in 1986 just a year ahead of Dr. James Carson's application with the FCC, which led to the awarding of a construction permit two years later.
|
{
"evidence": [
"WCCG_sentence_7"
],
"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:\nWCCG's 104.5 frequency was allocated in 1986 just a year ahead of Dr. James Carson's application with the FCC, which led to the awarding of a construction permit two years later.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"WCCG_sentence_7"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 458,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.671",
"value": "?search="
},
{
"operation": "search",
"time": "13.198",
"value": "WCCG"
},
{
"operation": "Now on",
"time": "13.817",
"value": "WCCG"
},
{
"operation": "Highlighting",
"time": "20.979",
"value": "WCCG_sentence_5"
},
{
"operation": "Highlighting",
"time": "64.931",
"value": "WCCG_sentence_6"
},
{
"operation": "Highlighting deleted",
"time": "67.664",
"value": "WCCG_sentence_6"
},
{
"operation": "Highlighting",
"time": "69.264",
"value": "WCCG_sentence_7"
},
{
"operation": "Highlighting deleted",
"time": "83.858",
"value": "WCCG_sentence_5"
},
{
"operation": "finish",
"time": "114.12",
"value": "finish"
}
]
}
|
18613
|
Bob Caffrey was part of the Olympics and the MLB Draft.
|
{
"evidence": [
"Bob Caffrey_sentence_5",
"Bob Caffrey_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:\nBob Caffrey was part of the Olympics and the MLB Draft.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Bob Caffrey_sentence_5",
"Bob Caffrey_sentence_7"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 459,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.05",
"value": "?search="
},
{
"operation": "search",
"time": "7.324",
"value": "Bob Caffrey"
},
{
"operation": "Now on",
"time": "8.28",
"value": "Bob Caffrey"
},
{
"operation": "Highlighting",
"time": "28.677",
"value": "Bob Caffrey_sentence_7"
},
{
"operation": "Highlighting deleted",
"time": "29.478",
"value": "Bob Caffrey_sentence_7"
},
{
"operation": "Highlighting",
"time": "34.245",
"value": "Bob Caffrey_sentence_5"
},
{
"operation": "Highlighting",
"time": "36.358",
"value": "Bob Caffrey_sentence_7"
},
{
"operation": "finish",
"time": "53.581",
"value": "finish"
}
]
}
|
50559
|
Kéba Mbaye, a Senegalese judge and member of both the International Olympic Committee and the International Court of Justice, was a member of the International Olympic Committee from 1973 to 2002 and also served as president of the Court of Arbitration for Sport, the highest Olympic tribunal.
|
{
"evidence": [
"Kéba Mbaye_sentence_0",
"Kéba Mbaye_sentence_2",
"Kéba Mbaye_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:\nKéba Mbaye, a Senegalese judge and member of both the International Olympic Committee and the International Court of Justice, was a member of the International Olympic Committee from 1973 to 2002 and also served as president of the Court of Arbitration for Sport, the highest Olympic tribunal.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Kéba Mbaye_sentence_0",
"Kéba Mbaye_sentence_2",
"Kéba Mbaye_sentence_4"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 460,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.064",
"value": "undefined"
},
{
"operation": "search",
"time": "10.796",
"value": "Kéba Mbaye"
},
{
"operation": "Now on",
"time": "12.801",
"value": "Kéba Mbaye"
},
{
"operation": "Highlighting",
"time": "20.606",
"value": "Kéba Mbaye_sentence_0"
},
{
"operation": "Highlighting",
"time": "39.922",
"value": "Kéba Mbaye_sentence_2"
},
{
"operation": "Highlighting",
"time": "45.898",
"value": "Kéba Mbaye_sentence_4"
},
{
"operation": "finish",
"time": "72.1",
"value": "finish"
}
]
}
|
exfever_4904
|
The Menace II Society is a 1993 film directed by the Hughes Brothers, whose genre originated in the United Kingdom and features aspects of urban African American culture.
|
{
"evidence": [
"Menace II Society",
"Hood 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 Menace II Society is a 1993 film directed by the Hughes Brothers, whose genre originated in the United Kingdom and features aspects of urban African American culture.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Menace II Society",
"Hood film"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 461,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
37068
|
Oswald Houston and his family moved to Tennessee in 1833, then he and his wife raised their family, then he became the first treasurer of the city of Atlanta (who was responsible for liquidity risk management, cash management, issuing and repaying debt, and interest rate risk, and oversight of pension investment management) after its inception in 1847 to the mid-1850s.
|
{
"evidence": [
"Oswald Houston_sentence_6",
"Oswald Houston_sentence_0",
"Oswald Houston_sentence_8",
"Oswald Houston_sentence_11"
],
"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:\nOswald Houston and his family moved to Tennessee in 1833, then he and his wife raised their family, then he became the first treasurer of the city of Atlanta (who was responsible for liquidity risk management, cash management, issuing and repaying debt, and interest rate risk, and oversight of pension investment management) after its inception in 1847 to the mid-1850s.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Oswald Houston_sentence_6",
"Oswald Houston_sentence_0",
"Oswald Houston_sentence_8",
"Oswald Houston_sentence_11"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 462,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.599",
"value": "?search="
},
{
"operation": "search",
"time": "16.735",
"value": "Oswald Houston"
},
{
"operation": "Now on",
"time": "17.323",
"value": "Oswald Houston"
},
{
"operation": "Highlighting",
"time": "30.838",
"value": "Oswald Houston_sentence_6"
},
{
"operation": "Highlighting",
"time": "41.554",
"value": "Oswald Houston_sentence_0"
},
{
"operation": "Highlighting",
"time": "52.246",
"value": "Oswald Houston_sentence_8"
},
{
"operation": "Highlighting",
"time": "77.518",
"value": "Oswald Houston_sentence_11"
},
{
"operation": "search",
"time": "155.997",
"value": "Treasurer City of Atlanta"
},
{
"operation": "Now on",
"time": "157.562",
"value": "undefined"
},
{
"operation": "search",
"time": "171.418",
"value": "Atlanta Treasurer"
},
{
"operation": "Now on",
"time": "171.918",
"value": "undefined"
},
{
"operation": "finish",
"time": "210.873",
"value": "finish"
}
]
}
|
exfever_2474
|
John McCain was a nominee for president of the U.S. in the 2008 election, which he lost to an African-American politician who graduated from a private Ivy League research university in Albany.
|
{
"evidence": [
"John McCain",
"Barack Obama",
"Columbia University"
],
"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:\nJohn McCain was a nominee for president of the U.S. in the 2008 election, which he lost to an African-American politician who graduated from a private Ivy League research university in Albany.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"John McCain",
"Barack Obama",
"Columbia University"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 463,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
65761
|
Patricio Toranzo, a centre midfielder (midfielders are generally positioned on the field between their team's defenders and forwards) for the team Huracán was born on March 13, 1982.
|
{
"evidence": [
"Patricio Toranzo_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:\nPatricio Toranzo, a centre midfielder (midfielders are generally positioned on the field between their team's defenders and forwards) for the team Huracán was born on March 13, 1982.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Patricio Toranzo_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 464,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.972",
"value": "?search="
},
{
"operation": "search",
"time": "12.842",
"value": "Patricio Toranzo"
},
{
"operation": "Now on",
"time": "13.911",
"value": "Patricio Toranzo"
},
{
"operation": "Highlighting",
"time": "36.094",
"value": "Patricio Toranzo_sentence_0"
},
{
"operation": "finish",
"time": "58.083",
"value": "finish"
}
]
}
|
exfever_8925
|
Andrew Stanton is an American animator whose work includes co-directing Pixar's American computer-musical film that stars the voice of an American stand-up comedian and actor.
|
{
"evidence": [
"Andrew Stanton",
"WALL-E",
"Jeff Garlin"
],
"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:\nAndrew Stanton is an American animator whose work includes co-directing Pixar's American computer-musical film that stars the voice of an American stand-up comedian and actor.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Andrew Stanton",
"WALL-E",
"Jeff Garlin"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 465,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_5327
|
Saif Ali Khan, an Indian actor who made his acting debut in Yash Chopra's drama Ek Hasina Thi, is the son of an Indian cricketer and a former captain of the Indian cricket team.
|
{
"evidence": [
"Saif Ali Khan",
"Mansoor Ali Khan Pataudi"
],
"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:\nSaif Ali Khan, an Indian actor who made his acting debut in Yash Chopra's drama Ek Hasina Thi, is the son of an Indian cricketer and a former captain of the Indian cricket team.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Saif Ali Khan",
"Mansoor Ali Khan Pataudi"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 466,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
54732
|
"Boku ga Ichiban Hoshikatta Mono", which debuted at number 9 on the Oricon Singles Chart with 32,000 copies sold, charted for 16 weeks on the Oricon Singles Chart, selling a reported total of 81,000 copies during its run.
|
{
"evidence": [
"Boku ga Ichiban Hoshikatta Mono_sentence_7"
],
"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:\n\"Boku ga Ichiban Hoshikatta Mono\", which debuted at number 9 on the Oricon Singles Chart with 32,000 copies sold, charted for 16 weeks on the Oricon Singles Chart, selling a reported total of 81,000 copies during its run.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Boku ga Ichiban Hoshikatta Mono_sentence_7"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 467,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "2.326",
"value": "?search="
},
{
"operation": "search",
"time": "16.875",
"value": "Boku ga Ichiban Hoshikatta Mono"
},
{
"operation": "Now on",
"time": "19.453",
"value": "Boku ga Ichiban Hoshikatta Mono"
},
{
"operation": "Highlighting",
"time": "24.684",
"value": "Boku ga Ichiban Hoshikatta Mono_sentence_0"
},
{
"operation": "hyperlink",
"time": "40.013",
"value": "1 Chart performance Boku ga Ichiban Hoshikatta Mono_section_0"
},
{
"operation": "Highlighting",
"time": "53.679",
"value": "Boku ga Ichiban Hoshikatta Mono_header_cell_0_1_0"
},
{
"operation": "Highlighting",
"time": "53.681",
"value": "Boku ga Ichiban Hoshikatta Mono_section_2"
},
{
"operation": "Highlighting",
"time": "54.538",
"value": "Boku ga Ichiban Hoshikatta Mono_cell_0_1_1"
},
{
"operation": "Highlighting",
"time": "54.545",
"value": "Boku ga Ichiban Hoshikatta Mono_header_cell_0_1_0"
},
{
"operation": "Highlighting",
"time": "54.545",
"value": "Boku ga Ichiban Hoshikatta Mono_header_cell_0_0_1"
},
{
"operation": "Highlighting",
"time": "60.943",
"value": "Boku ga Ichiban Hoshikatta Mono_header_cell_0_0_0"
},
{
"operation": "Highlighting deleted",
"time": "63.64",
"value": "Boku ga Ichiban Hoshikatta Mono_header_cell_0_0_0"
},
{
"operation": "Highlighting deleted",
"time": "66.642",
"value": "Boku ga Ichiban Hoshikatta Mono_cell_0_1_1"
},
{
"operation": "Highlighting deleted",
"time": "66.645",
"value": "Boku ga Ichiban Hoshikatta Mono_section_2"
},
{
"operation": "Highlighting deleted",
"time": "66.647",
"value": "Boku ga Ichiban Hoshikatta Mono_header_cell_0_1_0"
},
{
"operation": "Highlighting deleted",
"time": "66.648",
"value": "Boku ga Ichiban Hoshikatta Mono_header_cell_0_0_1"
},
{
"operation": "Highlighting",
"time": "67.754",
"value": "Boku ga Ichiban Hoshikatta Mono_header_cell_0_0_0"
},
{
"operation": "Highlighting",
"time": "67.756",
"value": "Boku ga Ichiban Hoshikatta Mono_section_2"
},
{
"operation": "Highlighting",
"time": "68.575",
"value": "Boku ga Ichiban Hoshikatta Mono_header_cell_0_0_1"
},
{
"operation": "Highlighting",
"time": "69.543",
"value": "Boku ga Ichiban Hoshikatta Mono_header_cell_0_1_0"
},
{
"operation": "Highlighting",
"time": "70.285",
"value": "Boku ga Ichiban Hoshikatta Mono_cell_0_1_1"
},
{
"operation": "Highlighting",
"time": "70.289",
"value": "Boku ga Ichiban Hoshikatta Mono_header_cell_0_1_0"
},
{
"operation": "Highlighting",
"time": "70.29",
"value": "Boku ga Ichiban Hoshikatta Mono_header_cell_0_0_1"
},
{
"operation": "hyperlink",
"time": "92.248",
"value": "Oricon"
},
{
"operation": "Now on",
"time": "93.843",
"value": "Oricon"
},
{
"operation": "hyperlink",
"time": "101.8",
"value": "5.2 Past charts Oricon_section_6"
},
{
"operation": "hyperlink",
"time": "119.665",
"value": "Oricon Singles Chart"
},
{
"operation": "Now on",
"time": "120.678",
"value": "Oricon Singles Chart"
},
{
"operation": "back-button-clicked",
"time": "142.472",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "142.55",
"value": "Oricon#Past charts Oricon section 6"
},
{
"operation": "back-button-clicked",
"time": "145.839",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "152.764",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "152.815",
"value": "Boku ga Ichiban Hoshikatta Mono#Chart performance Boku ga Ichiban Hoshikatta Mono section 0"
},
{
"operation": "Highlighting",
"time": "185.386",
"value": "Boku ga Ichiban Hoshikatta Mono_sentence_7"
},
{
"operation": "finish",
"time": "203.206",
"value": "finish"
}
]
}
|
exfever_3354
|
Ivan Lendl, from London, is a former professional player of a racket sport that can be played individually against a single opponent or between two teams of two players each.
|
{
"evidence": [
"Ivan Lendl",
"Tennis"
],
"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:\nIvan Lendl, from London, is a former professional player of a racket sport that can be played individually against a single opponent or between two teams of two players each.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Ivan Lendl",
"Tennis"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 468,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_6160
|
Perry Saturn is an American retired professional wrestler and a former champion in a professional wrestling world tag team championship in Extreme Championship Wrestling established under ECW in 1986.
|
{
"evidence": [
"Perry Saturn",
"ECW World Tag Team Championship"
],
"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:\nPerry Saturn is an American retired professional wrestler and a former champion in a professional wrestling world tag team championship in Extreme Championship Wrestling established under ECW in 1986.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Perry Saturn",
"ECW World Tag Team Championship"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 469,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
90599
|
Benedikt Stilling was a German anatomist who had a private practice in Kassel, received his doctorate from the University of Marburg and made first mention of vasomotor.
|
{
"evidence": [
"Benedict Stilling_sentence_0",
"Benedict Stilling_sentence_2",
"Benedict Stilling_sentence_3",
"Benedict Stilling_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:\nBenedikt Stilling was a German anatomist who had a private practice in Kassel, received his doctorate from the University of Marburg and made first mention of vasomotor.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"Benedict Stilling_sentence_0",
"Benedict Stilling_sentence_2",
"Benedict Stilling_sentence_3",
"Benedict Stilling_sentence_6"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 470,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.174",
"value": "undefined"
},
{
"operation": "search",
"time": "10.38",
"value": "Benedikt Stilling"
},
{
"operation": "Now on",
"time": "12.487",
"value": "Benedict Stilling"
},
{
"operation": "Highlighting",
"time": "17.202",
"value": "Benedict Stilling_sentence_0"
},
{
"operation": "Highlighting",
"time": "38.778",
"value": "Benedict Stilling_sentence_2"
},
{
"operation": "Highlighting",
"time": "39.902",
"value": "Benedict Stilling_sentence_3"
},
{
"operation": "Highlighting",
"time": "47.014",
"value": "Benedict Stilling_sentence_6"
},
{
"operation": "finish",
"time": "63.388",
"value": "finish"
}
]
}
|
66289
|
Plato Tiburtinus was a 12th-century Italian mathematician, astronomer and translator, who lived in Barcelona from 1116 to 1138, best known for translating Hebrew and Arabic documents into Latin, a classical language belonging to the Italic branch of the Indo-European languages.
|
{
"evidence": [
"Plato Tiburtinus_sentence_1",
"Plato Tiburtinus_sentence_2",
"Latin_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:\nPlato Tiburtinus was a 12th-century Italian mathematician, astronomer and translator, who lived in Barcelona from 1116 to 1138, best known for translating Hebrew and Arabic documents into Latin, a classical language belonging to the Italic branch of the Indo-European languages.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Plato Tiburtinus_sentence_1",
"Plato Tiburtinus_sentence_2",
"Latin_sentence_1"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 471,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.139",
"value": "?search="
},
{
"operation": "search",
"time": "9.28",
"value": "Plato Tiburtinus"
},
{
"operation": "Now on",
"time": "10.093",
"value": "Plato Tiburtinus"
},
{
"operation": "Highlighting",
"time": "12.132",
"value": "Plato Tiburtinus_sentence_1"
},
{
"operation": "Highlighting",
"time": "17.764",
"value": "Plato Tiburtinus_sentence_2"
},
{
"operation": "hyperlink",
"time": "22.6",
"value": "Latin"
},
{
"operation": "Now on",
"time": "25.228",
"value": "Latin"
},
{
"operation": "Highlighting",
"time": "27.098",
"value": "Latin_sentence_1"
},
{
"operation": "finish",
"time": "31.832",
"value": "finish"
}
]
}
|
exfever_5388
|
Trolls released another animated series titled Trolls: Trollstopia.
|
{
"evidence": [
"Trolls (film)",
"Musical film"
],
"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:\nTrolls released another animated series titled Trolls: Trollstopia.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Trolls (film)",
"Musical film"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 472,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_2174
|
James Tate Ellington appeared in the 2009 Broadway production of The Philanthropist, which starred an American actor and singer whose roles include the Golden Globe-nominated portrayal of the title character in Ferris Bueller's Day Off.
|
{
"evidence": [
"Tate Ellington",
"Matthew Broderick"
],
"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 Tate Ellington appeared in the 2009 Broadway production of The Philanthropist, which starred an American actor and singer whose roles include the Golden Globe-nominated portrayal of the title character in Ferris Bueller's Day Off.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Tate Ellington",
"Matthew Broderick"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 473,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
64844
|
Harold Strauss only wrote short stories and never associated them with cultural exchanges.
|
{
"evidence": [
"Harold Strauss_sentence_6"
],
"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:\nHarold Strauss only wrote short stories and never associated them with cultural exchanges.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Harold Strauss_sentence_6"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 474,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.809",
"value": "?search="
},
{
"operation": "search",
"time": "12.052",
"value": "Harold Strauss"
},
{
"operation": "Now on",
"time": "12.77",
"value": "Harold Strauss"
},
{
"operation": "Highlighting",
"time": "31.533",
"value": "Harold Strauss_sentence_6"
},
{
"operation": "finish",
"time": "47.612",
"value": "finish"
}
]
}
|
4312
|
The U.S. Army Corps of Engineers manages Pine Creek Lake, which was built in 1969 and covers 3,750 acres in McCurtain County (population 33,151), Pushmataha County (population 11, 572), but is not located near any historically Indian territories in the state.
|
{
"evidence": [
"Pine Creek Lake_sentence_0",
"Pushmataha County, Oklahoma_sentence_35"
],
"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 U.S. Army Corps of Engineers manages Pine Creek Lake, which was built in 1969 and covers 3,750 acres in McCurtain County (population 33,151), Pushmataha County (population 11, 572), but is not located near any historically Indian territories in the state.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Pine Creek Lake_sentence_0",
"Pushmataha County, Oklahoma_sentence_35"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 475,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "search",
"time": "7.108",
"value": "The U.S. Army Corps of Engineers"
},
{
"operation": "hyperlink",
"time": "12.874",
"value": "United States Army Corps of Engineers"
},
{
"operation": "Page search",
"time": "22.78",
"value": "Pine Creek Lake"
},
{
"operation": "Page search",
"time": "35.042",
"value": "Pine Creek"
},
{
"operation": "Page search",
"time": "35.597",
"value": "Pine Creek"
},
{
"operation": "Page search",
"time": "62.669",
"value": "manage"
},
{
"operation": "search",
"time": "82.368",
"value": "pine creek lake"
},
{
"operation": "Highlighting",
"time": "86.725",
"value": "Pine Creek Lake_sentence_4Pine Creek Lake_sentence_5"
},
{
"operation": "Highlighting",
"time": "97.942",
"value": "Pine Creek Lake_sentence_3"
},
{
"operation": "Highlighting",
"time": "106.182",
"value": "Pine Creek Lake_sentence_7"
},
{
"operation": "Highlighting",
"time": "106.187",
"value": "Pine Creek Lake_section_0"
},
{
"operation": "Highlighting",
"time": "111.592",
"value": "Pine Creek Lake_sentence_0"
},
{
"operation": "hyperlink",
"time": "120.065",
"value": "McCurtain County"
},
{
"operation": "Highlighting",
"time": "132.56",
"value": "McCurtain County, Oklahoma_sentence_3"
},
{
"operation": "Highlighting",
"time": "132.569",
"value": "McCurtain County, Oklahoma_title"
},
{
"operation": "back-button-clicked",
"time": "140.234",
"value": "back-button-clicked"
},
{
"operation": "hyperlink",
"time": "144.46",
"value": "Pushmataha County"
},
{
"operation": "Highlighting",
"time": "149.87",
"value": "Pushmataha County, Oklahoma_sentence_1"
},
{
"operation": "back-button-clicked",
"time": "152.757",
"value": "back-button-clicked"
},
{
"operation": "Page search",
"time": "217.257",
"value": "indian"
},
{
"operation": "hyperlink",
"time": "246.31",
"value": "Pushmataha County"
},
{
"operation": "Page search",
"time": "260.462",
"value": "indian"
},
{
"operation": "Highlighting",
"time": "283.722",
"value": "Pushmataha County, Oklahoma_sentence_35"
},
{
"operation": "finish",
"time": "357.677",
"value": "finish"
}
]
}
|
exfever_1925
|
The Woman in Red is a 1984 Spanish film that won one of the awards given annually to people working in the motion picture industry by AMPAS.
|
{
"evidence": [
"The Woman in Red (1984 film)",
"Academy Award for Best Original Song"
],
"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 Woman in Red is a 1984 Spanish film that won one of the awards given annually to people working in the motion picture industry by AMPAS.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Woman in Red (1984 film)",
"Academy Award for Best Original Song"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 476,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
91163
|
Nazas River in South Africa contains a desertic habitat and is home of the ocotillo.
|
{
"evidence": [
"Nazas River_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:\nNazas River in South Africa contains a desertic habitat and is home of the ocotillo.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Nazas River_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 477,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.286",
"value": "?search="
},
{
"operation": "search",
"time": "12.197",
"value": "Nazas River"
},
{
"operation": "Now on",
"time": "45.344",
"value": "Nazas River"
},
{
"operation": "Highlighting",
"time": "48.063",
"value": "Nazas River_sentence_0"
},
{
"operation": "finish",
"time": "80.3",
"value": "finish"
}
]
}
|
exfever_8579
|
Bluebird cafe is the largest cafe in the United states that gathered different music artists.
|
{
"evidence": [
"Nashville Songwriters Association International",
"Music industry"
],
"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:\nBluebird cafe is the largest cafe in the United states that gathered different music artists.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Nashville Songwriters Association International",
"Music industry"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 478,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
15148
|
Nightline anchor Ted Koppel went off on longtime Los Angeles Dodgers executive Al Campanis for the racist implications of his remarks resulting in Koppel's dismissal from ABC.
|
{
"evidence": [
"Al Campanis_sentence_24",
"Ted Koppel_sentence_72"
],
"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:\nNightline anchor Ted Koppel went off on longtime Los Angeles Dodgers executive Al Campanis for the racist implications of his remarks resulting in Koppel's dismissal from ABC.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Al Campanis_sentence_24",
"Ted Koppel_sentence_72"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 479,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.096",
"value": "?search="
},
{
"operation": "search",
"time": "5.779",
"value": "ted koppel"
},
{
"operation": "Now on",
"time": "6.562",
"value": "Ted Koppel"
},
{
"operation": "Highlighting",
"time": "14.332",
"value": "Ted Koppel_sentence_0"
},
{
"operation": "search",
"time": "46.309",
"value": "al campanis"
},
{
"operation": "Now on",
"time": "47.294",
"value": "Al Campanis"
},
{
"operation": "Highlighting",
"time": "64.393",
"value": "Al Campanis_sentence_2"
},
{
"operation": "Highlighting",
"time": "101.504",
"value": "Al Campanis_sentence_31"
},
{
"operation": "Highlighting",
"time": "120.997",
"value": "Al Campanis_sentence_32"
},
{
"operation": "back-button-clicked",
"time": "124.557",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "124.698",
"value": "Ted Koppel"
},
{
"operation": "search",
"time": "167.385",
"value": "al campanis"
},
{
"operation": "Now on",
"time": "168.344",
"value": "Al Campanis"
},
{
"operation": "Highlighting deleted",
"time": "178.49",
"value": "Al Campanis_sentence_32"
},
{
"operation": "Highlighting",
"time": "180.166",
"value": "Al Campanis_sentence_24"
},
{
"operation": "back-button-clicked",
"time": "181.431",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "181.566",
"value": "Ted Koppel"
},
{
"operation": "Highlighting",
"time": "188.073",
"value": "Ted Koppel_sentence_72"
},
{
"operation": "finish",
"time": "194.238",
"value": "finish"
}
]
}
|
exfever_6513
|
Jamie Lee Curtis is an American actress who made her feature film debut in the seminal slasher film Halloween by an American filmmaker generally recognized as one of the greatest masters of the horror genre.
|
{
"evidence": [
"Jamie Lee Curtis",
"John Carpenter"
],
"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:\nJamie Lee Curtis is an American actress who made her feature film debut in the seminal slasher film Halloween by an American filmmaker generally recognized as one of the greatest masters of the horror genre.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Jamie Lee Curtis",
"John Carpenter"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 480,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_1900
|
Spider-Man is a superhero that first appeared in an Italian comic book published by Marvel Comics from 1961 through 1962.
|
{
"evidence": [
"Spider-Man",
"Amazing Fantasy"
],
"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:\nSpider-Man is a superhero that first appeared in an Italian comic book published by Marvel Comics from 1961 through 1962.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Spider-Man",
"Amazing Fantasy"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 481,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
37705
|
Women in Indian Armed Forces were assigned to various branches of the army including a specialist corps in the Indian Army which primarily provides medical services to all Army personnel.
|
{
"evidence": [
"Special forces of India_sentence_1",
"Women in Indian Armed Forces_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:\nWomen in Indian Armed Forces were assigned to various branches of the army including a specialist corps in the Indian Army which primarily provides medical services to all Army personnel.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Special forces of India_sentence_1",
"Women in Indian Armed Forces_sentence_0"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 482,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.107",
"value": "?search="
},
{
"operation": "search",
"time": "16.498",
"value": "Women in Indian Armed Forces"
},
{
"operation": "Now on",
"time": "17.387",
"value": "Women in Indian Armed Forces"
},
{
"operation": "Page search",
"time": "34.157",
"value": "army"
},
{
"operation": "Page search",
"time": "73.978",
"value": "specialist"
},
{
"operation": "Page search",
"time": "123.869",
"value": "medical"
},
{
"operation": "Highlighting",
"time": "229.562",
"value": "Women in Indian Armed Forces_sentence_0"
},
{
"operation": "hyperlink",
"time": "255.947",
"value": "Special Forces of India"
},
{
"operation": "Now on",
"time": "256.661",
"value": "Special forces of India"
},
{
"operation": "Page search",
"time": "268.727",
"value": "medical"
},
{
"operation": "Highlighting",
"time": "315.457",
"value": "Special forces of India_sentence_1"
},
{
"operation": "Highlighting",
"time": "323.626",
"value": "Special forces of India_sentence_1"
},
{
"operation": "back-button-clicked",
"time": "339.989",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "340.258",
"value": "Women in Indian Armed Forces"
},
{
"operation": "Highlighting",
"time": "344.347",
"value": "Women in Indian Armed Forces_sentence_0"
},
{
"operation": "finish",
"time": "350.559",
"value": "finish"
}
]
}
|
14717
|
List of Interstate Highways in Texas, there are eight highways in the list.
|
{
"evidence": [
"List of Interstate Highways in Texas_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:\nList of Interstate Highways in Texas, there are eight highways in the list.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"List of Interstate Highways in Texas_sentence_2"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 483,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.592",
"value": "?search="
},
{
"operation": "search",
"time": "12.351",
"value": "List of Interstate Highways in Texas"
},
{
"operation": "Now on",
"time": "13.764",
"value": "List of Interstate Highways in Texas"
},
{
"operation": "Highlighting",
"time": "62.93",
"value": "List of Interstate Highways in Texas_sentence_2"
},
{
"operation": "finish",
"time": "85.012",
"value": "finish"
}
]
}
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.