id
stringlengths 1
12
| question
stringlengths 22
566
| golden_answers
dict | data_source
stringclasses 2
values | prompt
listlengths 1
1
| ability
stringclasses 6
values | reward_model
dict | extra_info
dict | metadata
dict |
|---|---|---|---|---|---|---|---|---|
exfever_7237
|
The L.A. Guns lineup currently consists of Tracii Guns, Phil Lewis, Ace Von Johnson, Johnny Martin, Scot Coogan, and Adam Hamilton, is an American band using a subgenre of heavy metal that features pop-influenced hooks and guitar riffs that evolved directly from the glam rock movement of the 1980s.
|
{
"evidence": [
"L.A. Guns",
"Glam metal"
],
"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 L.A. Guns lineup currently consists of Tracii Guns, Phil Lewis, Ace Von Johnson, Johnny Martin, Scot Coogan, and Adam Hamilton, is an American band using a subgenre of heavy metal that features pop-influenced hooks and guitar riffs that evolved directly from the glam rock movement of the 1980s.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"L.A. Guns",
"Glam metal"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7184,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_2214
|
One More Light is the highest-selling studio album by Linkin Park.
|
{
"evidence": [
"One More Light",
"Warner Bros. Records"
],
"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:\nOne More Light is the highest-selling studio album by Linkin Park.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"One More Light",
"Warner Bros. Records"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7185,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
38528
|
The Grogol River, which is located beside the 150,000 seat Gelora Bung Karno Stadium, flows in the northwest area of Chernobyl and the warmest month is March.
|
{
"evidence": [
"Grogol River_sentence_12"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Grogol River, which is located beside the 150,000 seat Gelora Bung Karno Stadium, flows in the northwest area of Chernobyl and the warmest month is March.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Grogol River_sentence_12"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7186,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.985",
"value": "?search="
},
{
"operation": "search",
"time": "8.824",
"value": "Grogol River"
},
{
"operation": "Now on",
"time": "9.331",
"value": "Grogol River"
},
{
"operation": "Page search",
"time": "16.344",
"value": "Stadium"
},
{
"operation": "Highlighting",
"time": "25.587",
"value": "Grogol River_sentence_7"
},
{
"operation": "Highlighting",
"time": "25.588",
"value": "Grogol River_section_0"
},
{
"operation": "Page search",
"time": "41.967",
"value": "northeast"
},
{
"operation": "Page search",
"time": "47.416",
"value": "northwest"
},
{
"operation": "Page search",
"time": "49.327",
"value": "northwest"
},
{
"operation": "Highlighting",
"time": "51.639",
"value": "Grogol River_sentence_12"
},
{
"operation": "Highlighting",
"time": "60.614",
"value": "Grogol River_sentence_14"
},
{
"operation": "Page search",
"time": "65.743",
"value": "chernobyl"
},
{
"operation": "search",
"time": "78.176",
"value": "Java"
},
{
"operation": "Now on",
"time": "79.244",
"value": "Java"
},
{
"operation": "search",
"time": "97.6",
"value": "Chernobyl"
},
{
"operation": "Now on",
"time": "98.208",
"value": "Chernobyl"
},
{
"operation": "back-button-clicked",
"time": "103.758",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "103.954",
"value": "Java"
},
{
"operation": "back-button-clicked",
"time": "107.436",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "107.443",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "107.578",
"value": "Grogol River"
},
{
"operation": "Highlighting",
"time": "112.107",
"value": "Grogol River_sentence_12"
},
{
"operation": "finish",
"time": "125.563",
"value": "finish"
}
]
}
|
exfever_812
|
Henry Cavill is known for his portrayal in Showtime's The Tudors as the 1st Duke of Suffolk, an American military leader and courtier.
|
{
"evidence": [
"Henry Cavill",
"Charles Brandon, 1st Duke of Suffolk"
],
"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:\nHenry Cavill is known for his portrayal in Showtime's The Tudors as the 1st Duke of Suffolk, an American military leader and courtier.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Henry Cavill",
"Charles Brandon, 1st Duke of Suffolk"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7187,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_5976
|
Bosnia and Herzegovina's largest city is surrounded by the Dinaric Alps and situated along the Miljacka River in the heart of the Balkans, a region of Southern Europe.
|
{
"evidence": [
"Bosnia and Herzegovina",
"Sarajevo"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nBosnia and Herzegovina's largest city is surrounded by the Dinaric Alps and situated along the Miljacka River in the heart of the Balkans, a region of Southern Europe.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Bosnia and Herzegovina",
"Sarajevo"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7188,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
50354
|
Gertrude Parsons converted to Catholicism in 1888 and married her husband, who had converted in 1843.
|
{
"evidence": [
"Gertrude Parsons_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:\nGertrude Parsons converted to Catholicism in 1888 and married her husband, who had converted in 1843.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Gertrude Parsons_sentence_1"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7189,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "2.213",
"value": "?search="
},
{
"operation": "search",
"time": "121.444",
"value": "Gertrude Parsons"
},
{
"operation": "Now on",
"time": "123.713",
"value": "Gertrude Parsons"
},
{
"operation": "Highlighting",
"time": "148.643",
"value": "Gertrude Parsons_sentence_1"
},
{
"operation": "Highlighting",
"time": "192.697",
"value": "Gertrude Parsons_sentence_7"
},
{
"operation": "Highlighting",
"time": "192.699",
"value": "Gertrude Parsons_section_0"
},
{
"operation": "Highlighting deleted",
"time": "243.146",
"value": "Gertrude Parsons_sentence_7"
},
{
"operation": "Highlighting deleted",
"time": "243.148",
"value": "Gertrude Parsons_section_0"
},
{
"operation": "finish",
"time": "254.193",
"value": "finish"
}
]
}
|
exfever_7422
|
Thomas Franco is the youngest founder of an art collective.
|
{
"evidence": [
"Tom Franco",
"Berkeley, California"
],
"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:\nThomas Franco is the youngest founder of an art collective.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Tom Franco",
"Berkeley, California"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7190,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_1421
|
The Book of Mormon is a musical comedy by Robert Lopez, Matt Stone, and an American actor, animator, writer, producer, director, and composer.
|
{
"evidence": [
"The Book of Mormon (musical)",
"Trey Parker"
],
"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 Book of Mormon is a musical comedy by Robert Lopez, Matt Stone, and an American actor, animator, writer, producer, director, and composer.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Book of Mormon (musical)",
"Trey Parker"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7191,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
63470
|
Brian Finch longest relationship was with the ITV1 soap opera, Coronation Street, for which he wrote 120 scripts between 1980 and 1999 and he also helped the development of All Creatures Great and Small, The Tomorrow People, and Heartbeat (a British police drama series).
|
{
"evidence": [
"Brian Finch_sentence_1",
"Brian Finch_sentence_12"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nBrian Finch longest relationship was with the ITV1 soap opera, Coronation Street, for which he wrote 120 scripts between 1980 and 1999 and he also helped the development of All Creatures Great and Small, The Tomorrow People, and Heartbeat (a British police drama series).\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Brian Finch_sentence_1",
"Brian Finch_sentence_12"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7192,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "2.214",
"value": "?search="
},
{
"operation": "search",
"time": "6.873",
"value": "Brian Finch"
},
{
"operation": "Now on",
"time": "7.8",
"value": "Brian Finch"
},
{
"operation": "Highlighting",
"time": "18.771",
"value": "Brian Finch_sentence_1"
},
{
"operation": "Highlighting",
"time": "35.046",
"value": "Brian Finch_sentence_12"
},
{
"operation": "finish",
"time": "37.495",
"value": "finish"
}
]
}
|
41270
|
Robert Ilyasov(born 8 October 1973) is a Russian rugby league footballer who began his career for Strela Kazan and played for Strela and Lokomotiv rugby league teams from 1995 to 2001 and also served as the captain of the Russian national rugby league team.
|
{
"evidence": [
"Robert Ilyasov_sentence_0",
"Robert Ilyasov_sentence_4",
"Robert Ilyasov_sentence_5",
"Robert Ilyasov_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:\nRobert Ilyasov(born 8 October 1973) is a Russian rugby league footballer who began his career for Strela Kazan and played for Strela and Lokomotiv rugby league teams from 1995 to 2001 and also served as the captain of the Russian national rugby league team.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Robert Ilyasov_sentence_0",
"Robert Ilyasov_sentence_4",
"Robert Ilyasov_sentence_5",
"Robert Ilyasov_sentence_2"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7193,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.316",
"value": "?search="
},
{
"operation": "search",
"time": "9.141",
"value": "Robert Ilyasov"
},
{
"operation": "Now on",
"time": "11.6",
"value": "Robert Ilyasov"
},
{
"operation": "Highlighting",
"time": "13.625",
"value": "Robert Ilyasov_sentence_0"
},
{
"operation": "Highlighting",
"time": "25.32",
"value": "Robert Ilyasov_sentence_4"
},
{
"operation": "Highlighting",
"time": "28.398",
"value": "Robert Ilyasov_sentence_5"
},
{
"operation": "Highlighting",
"time": "48.368",
"value": "Robert Ilyasov_sentence_2"
},
{
"operation": "finish",
"time": "75.156",
"value": "finish"
}
]
}
|
exfever_6926
|
Daniel Edward Aykroyd is a Canadian actor who performed in a musical sketch on SNL with his American actor friend, best known for being one of the seven original cast members of the CNN sketch comedy show Saturday Night Live.
|
{
"evidence": [
"Dan Aykroyd",
"John Belushi"
],
"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:\nDaniel Edward Aykroyd is a Canadian actor who performed in a musical sketch on SNL with his American actor friend, best known for being one of the seven original cast members of the CNN sketch comedy show Saturday Night Live.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Dan Aykroyd",
"John Belushi"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7194,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_5994
|
Bring It On is a 2000 American teen cheerleading comedy film directed by an American television and film director with a 2008 romantic comedy film credited to him.
|
{
"evidence": [
"Bring It On (film)",
"Peyton Reed",
"Yes Man (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:\nBring It On is a 2000 American teen cheerleading comedy film directed by an American television and film director with a 2008 romantic comedy film credited to him.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Bring It On (film)",
"Peyton Reed",
"Yes Man (film)"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7195,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_740
|
Stefani Joanne Angelina Germanotta was one of the highest-paid artists in 2019.
|
{
"evidence": [
"Grace Jones",
"Lady Gaga",
"The Fame"
],
"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:\nStefani Joanne Angelina Germanotta was one of the highest-paid artists in 2019.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Grace Jones",
"Lady Gaga",
"The Fame"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7196,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
16654
|
Interplanetary Transport Network (a collection of gravitationally determined pathways through the Solar System) a trip from Earth to Venus would only take 3 months and the transfers are low-energy.
|
{
"evidence": [
"Interplanetary Transport Network_sentence_1",
"Interplanetary Transport Network_sentence_37"
],
"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:\nInterplanetary Transport Network (a collection of gravitationally determined pathways through the Solar System) a trip from Earth to Venus would only take 3 months and the transfers are low-energy.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Interplanetary Transport Network_sentence_1",
"Interplanetary Transport Network_sentence_37"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7197,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.515",
"value": "?search="
},
{
"operation": "search",
"time": "13.536",
"value": "Interplanetary Transport Network"
},
{
"operation": "Now on",
"time": "15.258",
"value": "Interplanetary Transport Network"
},
{
"operation": "Highlighting",
"time": "30.091",
"value": "Interplanetary Transport Network_sentence_1"
},
{
"operation": "Page search",
"time": "49.662",
"value": "venus"
},
{
"operation": "Highlighting",
"time": "78.493",
"value": "Interplanetary Transport Network_sentence_2"
},
{
"operation": "Highlighting deleted",
"time": "93.655",
"value": "Interplanetary Transport Network_sentence_2"
},
{
"operation": "Page search",
"time": "117.573",
"value": "3 monthd"
},
{
"operation": "Page search",
"time": "119.187",
"value": "3 months"
},
{
"operation": "Page search",
"time": "126.755",
"value": "months"
},
{
"operation": "Highlighting",
"time": "154.191",
"value": "Interplanetary Transport Network_sentence_37"
},
{
"operation": "finish",
"time": "184.3",
"value": "finish"
}
]
}
|
exfever_277
|
L.A. Reid previously served as the chairman and CEO of an British record label group.
|
{
"evidence": [
"L.A. Reid",
"The Island Def Jam Music Group"
],
"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:\nL.A. Reid previously served as the chairman and CEO of an British record label group.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"L.A. Reid",
"The Island Def Jam Music Group"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7198,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_8287
|
Men in Black II is a 2002 American action comedy film that serves as the second film of American action comedy film whose genre uses speculative, fictional science-based depictions of phenomena that are not fully accepted by mainstream science.
|
{
"evidence": [
"Men in Black II",
"Men in Black (film series)",
"Science fiction 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:\nMen in Black II is a 2002 American action comedy film that serves as the second film of American action comedy film whose genre uses speculative, fictional science-based depictions of phenomena that are not fully accepted by mainstream science.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Men in Black II",
"Men in Black (film series)",
"Science fiction film"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7199,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
35337
|
The 1994 three 64-gun third-rate ships were cut down to 44-gun fifth-rate frigates and razees was in response to rumours then circulating of very large American frigates supposed to be under construction and by Admiralty Order of 11 September 1994, two 64-gun ships of the Intrepid class - Anson and Magnanime - were to be cut down by one deck level.
|
{
"evidence": [
"1794 razees_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:\nThe 1994 three 64-gun third-rate ships were cut down to 44-gun fifth-rate frigates and razees was in response to rumours then circulating of very large American frigates supposed to be under construction and by Admiralty Order of 11 September 1994, two 64-gun ships of the Intrepid class - Anson and Magnanime - were to be cut down by one deck level.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"1794 razees_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7200,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.12",
"value": "?search="
},
{
"operation": "search",
"time": "152.976",
"value": "1794 razees"
},
{
"operation": "Now on",
"time": "153.884",
"value": "1794 razees"
},
{
"operation": "Highlighting",
"time": "159.964",
"value": "1794 razees_sentence_0"
},
{
"operation": "Highlighting",
"time": "163.844",
"value": "1794 razees_sentence_1"
},
{
"operation": "Highlighting deleted",
"time": "165.815",
"value": "1794 razees_sentence_1"
},
{
"operation": "finish",
"time": "177.896",
"value": "finish"
}
]
}
|
76417
|
Adobe After Effects has extensive plug-in support and a variety of plug-in styles exist.
|
{
"evidence": [
"Adobe After Effects_sentence_8"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nAdobe After Effects has extensive plug-in support and a variety of plug-in styles exist.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Adobe After Effects_sentence_8"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7201,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.707",
"value": "?search="
},
{
"operation": "search",
"time": "4.799",
"value": "Adobe After Effects"
},
{
"operation": "Now on",
"time": "5.943",
"value": "Adobe After Effects"
},
{
"operation": "Highlighting",
"time": "206.88",
"value": "Adobe After Effects_sentence_8"
},
{
"operation": "finish",
"time": "213.618",
"value": "finish"
}
]
}
|
66605
|
Niono is a town and commune located in Timbuktu where at least two major humanitarian projects have attempted to ameliorate the lives of the people.
|
{
"evidence": [
"Niono_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:\nNiono is a town and commune located in Timbuktu where at least two major humanitarian projects have attempted to ameliorate the lives of the people.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Niono_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7202,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.287",
"value": "?search="
},
{
"operation": "search",
"time": "23.31",
"value": "Niono"
},
{
"operation": "Now on",
"time": "24.428",
"value": "Niono"
},
{
"operation": "Highlighting",
"time": "27.113",
"value": "Niono_sentence_0"
},
{
"operation": "hyperlink",
"time": "79.705",
"value": "Ségou Region"
},
{
"operation": "Now on",
"time": "83.792",
"value": "Ségou Region"
},
{
"operation": "back-button-clicked",
"time": "87.728",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "87.791",
"value": "Niono"
},
{
"operation": "hyperlink",
"time": "92.886",
"value": "Timbuktu"
},
{
"operation": "Now on",
"time": "94.062",
"value": "Timbuktu"
},
{
"operation": "back-button-clicked",
"time": "95.947",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "96.019",
"value": "Niono"
},
{
"operation": "hyperlink",
"time": "146.085",
"value": "Timbuktu"
},
{
"operation": "Now on",
"time": "146.43",
"value": "Timbuktu"
},
{
"operation": "back-button-clicked",
"time": "154.632",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "154.702",
"value": "Niono"
},
{
"operation": "finish",
"time": "169.936",
"value": "finish"
}
]
}
|
50987
|
Mithraculus sculptus is wrongly thought to be in the Majidae Family in the Animalia Kingdom.
|
{
"evidence": [
"Mithraculus sculptus_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:\nMithraculus sculptus is wrongly thought to be in the Majidae Family in the Animalia Kingdom.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Mithraculus sculptus_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7203,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.237",
"value": "?search="
},
{
"operation": "search",
"time": "10.042",
"value": "Mithraculus sculptus"
},
{
"operation": "Now on",
"time": "10.987",
"value": "Mithraculus sculptus"
},
{
"operation": "Highlighting",
"time": "17.273",
"value": "Mithraculus sculptus_sentence_0"
},
{
"operation": "finish",
"time": "33.158",
"value": "finish"
}
]
}
|
exfever_817
|
The Rear Window teleplay was made by two writers, including an American producer who had made uncredited contributions to a film by an American animator.
|
{
"evidence": [
"Rear Window (1998 film)",
"Larry Gross",
"Ralph Bakshi"
],
"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 Rear Window teleplay was made by two writers, including an American producer who had made uncredited contributions to a film by an American animator.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Rear Window (1998 film)",
"Larry Gross",
"Ralph Bakshi"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7204,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
26935
|
Bernard Moss coined the term "virokine" but is best known for research on the poxvirus family.
|
{
"evidence": [
"Bernard Moss_sentence_12",
"Bernard Moss_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:\nBernard Moss coined the term \"virokine\" but is best known for research on the poxvirus family.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Bernard Moss_sentence_12",
"Bernard Moss_sentence_9"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7205,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.88",
"value": "?search="
},
{
"operation": "search",
"time": "20.943",
"value": "Bernard Moss"
},
{
"operation": "Now on",
"time": "21.8",
"value": "Bernard Moss"
},
{
"operation": "Page search",
"time": "58.6",
"value": "virokine"
},
{
"operation": "Highlighting",
"time": "62.716",
"value": "Bernard Moss_sentence_12"
},
{
"operation": "Highlighting",
"time": "69.145",
"value": "Bernard Moss_sentence_9"
},
{
"operation": "finish",
"time": "80.972",
"value": "finish"
}
]
}
|
exfever_3160
|
Robert Oppenheimer was Project Y's first director, succeeded by an American physicist in charge of the final assembly of "The Gadget," detonated in July 1946 for the test of the first detonation of a nuclear weapon.
|
{
"evidence": [
"Project Y",
"Norris Bradbury",
"Trinity (nuclear test)"
],
"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:\nRobert Oppenheimer was Project Y's first director, succeeded by an American physicist in charge of the final assembly of \"The Gadget,\" detonated in July 1946 for the test of the first detonation of a nuclear weapon.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Project Y",
"Norris Bradbury",
"Trinity (nuclear test)"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7206,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
36229
|
Virginia State Route 5 has many major intersections, including Osborne Turnpike, S. Boundary Street, and W. Francis Street.
|
{
"evidence": [
"Virginia State Route 5_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:\nVirginia State Route 5 has many major intersections, including Osborne Turnpike, S. Boundary Street, and W. Francis Street.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Virginia State Route 5_sentence_0"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7207,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.036",
"value": "?search="
},
{
"operation": "Now on",
"time": "6.85",
"value": "?search="
},
{
"operation": "search",
"time": "16.219",
"value": "Virginia State Route 5"
},
{
"operation": "Now on",
"time": "17.098",
"value": "Virginia State Route 5"
},
{
"operation": "Page search",
"time": "89.721",
"value": "osborne"
},
{
"operation": "Page search",
"time": "157.945",
"value": "intersection"
},
{
"operation": "hyperlink",
"time": "183.586",
"value": "4 Major intersections Virginia State Route 5_section_3"
},
{
"operation": "Highlighting",
"time": "214.181",
"value": "Virginia State Route 5_sentence_0"
},
{
"operation": "finish",
"time": "219.483",
"value": "finish"
}
]
}
|
5162
|
Dennis Epple's research focuses on political economics, the economics of education, and public finance; more specifically, the Dennis Epple's research focuses on political economics, the economics of education, and public finance; more specifically, the Budget-maximizing model, public decision-making, financing and provision of private and public goods and services in local settings, public decision-making, financing and provision of private and public goods and services in local settings
|
{
"evidence": [
"Dennis Epple_sentence_15",
"Dennis Epple_sentence_16"
],
"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:\nDennis Epple's research focuses on political economics, the economics of education, and public finance; more specifically, the Dennis Epple's research focuses on political economics, the economics of education, and public finance; more specifically, the Budget-maximizing model, public decision-making, financing and provision of private and public goods and services in local settings, public decision-making, financing and provision of private and public goods and services in local settings\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"Dennis Epple_sentence_15",
"Dennis Epple_sentence_16"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7208,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "search",
"time": "44.338",
"value": "Dennis Epple"
},
{
"operation": "hyperlink",
"time": "51.844",
"value": "2 Research Dennis Epple_section_1"
},
{
"operation": "Highlighting",
"time": "57.011",
"value": "Dennis Epple_sentence_15"
},
{
"operation": "Page search",
"time": "85.996",
"value": "budget"
},
{
"operation": "Page search",
"time": "103.883",
"value": "budget-maximizing"
},
{
"operation": "Page search",
"time": "114.498",
"value": "public decisio"
},
{
"operation": "Highlighting",
"time": "122.985",
"value": "Dennis Epple_sentence_16"
},
{
"operation": "Page search",
"time": "142.48",
"value": "maximizing"
},
{
"operation": "finish",
"time": "164.799",
"value": "finish"
}
]
}
|
exfever_5687
|
The Avenger is a 2012 American superhero film distributed by the American film distribution studio in the Walt Disney Company's Disney Media and Entertainment Distribution division.
|
{
"evidence": [
"The Avengers (2012 film)",
"Walt Disney Studios Motion Pictures"
],
"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 Avenger is a 2012 American superhero film distributed by the American film distribution studio in the Walt Disney Company's Disney Media and Entertainment Distribution division.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Avengers (2012 film)",
"Walt Disney Studios Motion Pictures"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7209,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_6255
|
Genghis Khan was the most powerful emperor of the Mongol Empire.
|
{
"evidence": [
"Genghis Khan",
"Mongol invasions and conquests",
"Mongol Empire"
],
"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:\nGenghis Khan was the most powerful emperor of the Mongol Empire.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Genghis Khan",
"Mongol invasions and conquests",
"Mongol Empire"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7210,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
65598
|
Marcus Paschal was signed by the Philadelphia Eagles as an undrafted free agent in May 2007 and was released by the Eagles during final cuts on August 30, 2008.
|
{
"evidence": [
"Marcus Paschal_sentence_4",
"Marcus Paschal_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:\nMarcus Paschal was signed by the Philadelphia Eagles as an undrafted free agent in May 2007 and was released by the Eagles during final cuts on August 30, 2008.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Marcus Paschal_sentence_4",
"Marcus Paschal_sentence_7"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7211,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "7.98",
"value": "?search="
},
{
"operation": "search",
"time": "18.669",
"value": "Marcus Paschal"
},
{
"operation": "Now on",
"time": "19.762",
"value": "Marcus Paschal"
},
{
"operation": "Highlighting",
"time": "33.506",
"value": "Marcus Paschal_sentence_4"
},
{
"operation": "Highlighting",
"time": "47.402",
"value": "Marcus Paschal_sentence_7"
},
{
"operation": "finish",
"time": "55.247",
"value": "finish"
}
]
}
|
exfever_5517
|
UFC 216 was the last MMA event that was held in T-Mobile Arena.
|
{
"evidence": [
"UFC 216",
"Ultimate Fighting Championship"
],
"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:\nUFC 216 was the last MMA event that was held in T-Mobile Arena.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"UFC 216",
"Ultimate Fighting Championship"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7212,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
84904
|
The Pegram Plantation House is a 7.8 acres (3.2 ha) one-story Greek Revival-style galleried house built in 1850 in New Orleans, Louisiana that was added in the National Register of Historic Places on October 24, 2003
|
{
"evidence": [
"Pegram Plantation House_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:\nThe Pegram Plantation House is a 7.8 acres (3.2 ha) one-story Greek Revival-style galleried house built in 1850 in New Orleans, Louisiana that was added in the National Register of Historic Places on October 24, 2003\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Pegram Plantation House_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7213,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "4.501",
"value": "?search="
},
{
"operation": "search",
"time": "12.75",
"value": "Pegram Plantation House"
},
{
"operation": "Now on",
"time": "13.537",
"value": "Pegram Plantation House"
},
{
"operation": "Highlighting",
"time": "18.342",
"value": "Pegram Plantation House_cell_0_2_1"
},
{
"operation": "Highlighting",
"time": "18.348",
"value": "Pegram Plantation House_header_cell_0_2_0"
},
{
"operation": "Highlighting",
"time": "23.333",
"value": "Pegram Plantation House_sentence_2"
},
{
"operation": "Highlighting",
"time": "27.527",
"value": "Pegram Plantation House_sentence_0"
},
{
"operation": "hyperlink",
"time": "32.916",
"value": "Lecompte, Louisiana"
},
{
"operation": "Now on",
"time": "35.109",
"value": "Lecompte, Louisiana"
},
{
"operation": "back-button-clicked",
"time": "40.903",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "40.988",
"value": "Pegram Plantation House"
},
{
"operation": "Highlighting deleted",
"time": "44.076",
"value": "Pegram Plantation House_sentence_2"
},
{
"operation": "Highlighting deleted",
"time": "45.524",
"value": "Pegram Plantation House_cell_0_2_1"
},
{
"operation": "Highlighting deleted",
"time": "45.532",
"value": "Pegram Plantation House_header_cell_0_2_0"
},
{
"operation": "hyperlink",
"time": "49.98",
"value": "Lecompte, Louisiana"
},
{
"operation": "Now on",
"time": "50.462",
"value": "Lecompte, Louisiana"
},
{
"operation": "back-button-clicked",
"time": "59.758",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "59.844",
"value": "Pegram Plantation House"
},
{
"operation": "finish",
"time": "72.059",
"value": "finish"
}
]
}
|
exfever_1585
|
The Flash is an American tv series that aired in an American commercial broadcast network that distributes over the air program content.
|
{
"evidence": [
"The Flash (1990 TV series)",
"CBS",
"Television network"
],
"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 Flash is an American tv series that aired in an American commercial broadcast network that distributes over the air program content.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Flash (1990 TV series)",
"CBS",
"Television network"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7214,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
91603
|
Wakaba Girl tells about Wakaba Kohashi who is a "a slightly out-of-tune super" daughter of a well-to-do family and an admirer of the trendy "gyaru" fashion subculture.
|
{
"evidence": [
"Wakaba Girl_sentence_2",
"Wakaba Girl_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:\nWakaba Girl tells about Wakaba Kohashi who is a \"a slightly out-of-tune super\" daughter of a well-to-do family and an admirer of the trendy \"gyaru\" fashion subculture.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Wakaba Girl_sentence_2",
"Wakaba Girl_sentence_3"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7215,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.027",
"value": "?search="
},
{
"operation": "search",
"time": "18.907",
"value": "Wakaba Girl"
},
{
"operation": "Now on",
"time": "19.59",
"value": "Wakaba Girl"
},
{
"operation": "Highlighting",
"time": "27.346",
"value": "Wakaba Girl_sentence_2"
},
{
"operation": "Highlighting",
"time": "36.899",
"value": "Wakaba Girl_sentence_3"
},
{
"operation": "finish",
"time": "48.968",
"value": "finish"
}
]
}
|
45203
|
Safety sign started as text only signs, and the European Council (a collegiate body that defines the overall political directions and priorities) Directive replaced the old one, which had information on how to utilize safety signs.
|
{
"evidence": [
"Safety sign_sentence_16",
"Safety sign_sentence_17",
"Safety sign_sentence_18",
"European Council_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:\nSafety sign started as text only signs, and the European Council (a collegiate body that defines the overall political directions and priorities) Directive replaced the old one, which had information on how to utilize safety signs.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Safety sign_sentence_16",
"Safety sign_sentence_17",
"Safety sign_sentence_18",
"European Council_sentence_1"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7216,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.406",
"value": "?search="
},
{
"operation": "search",
"time": "18.33",
"value": "Safety sign"
},
{
"operation": "Now on",
"time": "19.364",
"value": "Safety sign"
},
{
"operation": "Page search",
"time": "441.363",
"value": "text only"
},
{
"operation": "Page search",
"time": "447.665",
"value": "text only"
},
{
"operation": "Highlighting",
"time": "461.495",
"value": "Safety sign_sentence_16"
},
{
"operation": "Highlighting",
"time": "481.175",
"value": "Safety sign_sentence_17"
},
{
"operation": "Highlighting",
"time": "482.904",
"value": "Safety sign_sentence_18"
},
{
"operation": "hyperlink",
"time": "485.416",
"value": "European Council"
},
{
"operation": "Now on",
"time": "486.398",
"value": "European Council"
},
{
"operation": "Highlighting",
"time": "490.395",
"value": "European Council_sentence_1"
},
{
"operation": "finish",
"time": "498.578",
"value": "finish"
}
]
}
|
exfever_1945
|
Hamlet was the highest-grossing film of 1990.
|
{
"evidence": [
"Hamlet (1990 film)",
"William 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:\nHamlet was the highest-grossing film of 1990.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Hamlet (1990 film)",
"William Shakespeare"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7217,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7123
|
Get Out is a 2017 American horror film starring an American actor who starred as Jimmy James on the television sitcom NewsRadio.
|
{
"evidence": [
"Get Out (film)",
"Stephen Root"
],
"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:\nGet Out is a 2017 American horror film starring an American actor who starred as Jimmy James on the television sitcom NewsRadio.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Get Out (film)",
"Stephen Root"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7218,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
14067
|
Myles Kennedy's full discography shows that he was the frontman of Queen's touring group, and is currently a member of the rock band Alter Bridge.
|
{
"evidence": [
"Myles Kennedy discography_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:\nMyles Kennedy's full discography shows that he was the frontman of Queen's touring group, and is currently a member of the rock band Alter Bridge.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Myles Kennedy discography_sentence_2"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7219,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.12",
"value": "?search="
},
{
"operation": "search",
"time": "18.483",
"value": "Contains...Myles Kennedy's full discography"
},
{
"operation": "Now on",
"time": "20.576",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "23.637",
"value": "Discography Mark Tremonti_section_11"
},
{
"operation": "Now on",
"time": "24.696",
"value": "Mark Tremonti#Discography Mark Tremonti section 11"
},
{
"operation": "back-button-clicked",
"time": "28.131",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "28.985",
"value": "undefined"
},
{
"operation": "search",
"time": "44.79",
"value": "Myles Kennedy discography"
},
{
"operation": "Now on",
"time": "46.462",
"value": "Myles Kennedy discography"
},
{
"operation": "Page search",
"time": "57.909",
"value": "frontman"
},
{
"operation": "Highlighting",
"time": "63.203",
"value": "Myles Kennedy discography_sentence_2"
},
{
"operation": "hyperlink",
"time": "89.629",
"value": "Slash"
},
{
"operation": "Now on",
"time": "91.055",
"value": "Slash (musician)"
},
{
"operation": "finish",
"time": "114.393",
"value": "finish"
}
]
}
|
exfever_6005
|
Wish Upon is a 2017 American supernatural horror film, which stars an American actor who also played Billy Douglas on the soap opera One Life to Live.
|
{
"evidence": [
"Wish Upon",
"Ryan Phillippe"
],
"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:\nWish Upon is a 2017 American supernatural horror film, which stars an American actor who also played Billy Douglas on the soap opera One Life to Live.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Wish Upon",
"Ryan Phillippe"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7220,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7599
|
Shawn Carlson is the most influential American physicist.
|
{
"evidence": [
"Shawn Carlson",
"Physicist"
],
"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:\nShawn Carlson is the most influential American physicist.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Shawn Carlson",
"Physicist"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7221,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
47152
|
Unlike other Mughal tombs, which are built in gardens, the Tomb of Nadira Begum, a Mughal era tomb in the city of Lahore, Pakistan, does not have a dome and was built in a sunken water tank that is "200 by 200 Mughal gaz in size"
|
{
"evidence": [
"Tomb of Nadira Begum_sentence_7",
"Tomb of Nadira Begum_sentence_0",
"Tomb of Nadira Begum_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:\nUnlike other Mughal tombs, which are built in gardens, the Tomb of Nadira Begum, a Mughal era tomb in the city of Lahore, Pakistan, does not have a dome and was built in a sunken water tank that is \"200 by 200 Mughal gaz in size\"\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Tomb of Nadira Begum_sentence_7",
"Tomb of Nadira Begum_sentence_0",
"Tomb of Nadira Begum_sentence_9"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7222,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.186",
"value": "?search="
},
{
"operation": "search",
"time": "13.317",
"value": "Tomb of Nadira Begum"
},
{
"operation": "Now on",
"time": "14.402",
"value": "Tomb of Nadira Begum"
},
{
"operation": "Highlighting",
"time": "26.74",
"value": "Tomb of Nadira Begum_sentence_7"
},
{
"operation": "Highlighting",
"time": "40.354",
"value": "Tomb of Nadira Begum_sentence_0"
},
{
"operation": "hyperlink",
"time": "48.961",
"value": "gaz"
},
{
"operation": "Now on",
"time": "49.549",
"value": "Guz"
},
{
"operation": "back-button-clicked",
"time": "50.241",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "50.3",
"value": "Tomb of Nadira Begum"
},
{
"operation": "Highlighting",
"time": "51.538",
"value": "Tomb of Nadira Begum_sentence_9"
},
{
"operation": "finish",
"time": "58.704",
"value": "finish"
}
]
}
|
exfever_1732
|
Game of Thrones is the most popular television series that features a large ensemble cast.
|
{
"evidence": [
"Game of Thrones (season 1)",
"HBO"
],
"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:\nGame of Thrones is the most popular television series that features a large ensemble cast.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Game of Thrones (season 1)",
"HBO"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7223,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
48503
|
Born on January 16, 1866, John Robichaux played jazz music, and was a bandleader, drummer, and violinist.
|
{
"evidence": [
"John Robichaux_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:\nBorn on January 16, 1866, John Robichaux played jazz music, and was a bandleader, drummer, and violinist.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"John Robichaux_sentence_0"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7224,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.92",
"value": "?search="
},
{
"operation": "search",
"time": "9.18",
"value": "John Robichaux"
},
{
"operation": "Now on",
"time": "10.035",
"value": "John Robichaux"
},
{
"operation": "Highlighting",
"time": "14.065",
"value": "John Robichaux_sentence_0"
},
{
"operation": "Highlighting",
"time": "25.168",
"value": "John Robichaux_header_cell_0_4_0"
},
{
"operation": "Highlighting",
"time": "26.673",
"value": "John Robichaux_cell_0_4_1"
},
{
"operation": "Highlighting",
"time": "26.684",
"value": "John Robichaux_header_cell_0_4_0"
},
{
"operation": "Highlighting",
"time": "26.685",
"value": "John Robichaux_header_cell_0_1_0"
},
{
"operation": "Highlighting",
"time": "26.686",
"value": "John Robichaux_header_cell_0_0_0"
},
{
"operation": "Highlighting",
"time": "29.177",
"value": "John Robichaux_cell_0_5_1"
},
{
"operation": "Highlighting",
"time": "29.183",
"value": "John Robichaux_header_cell_0_5_0"
},
{
"operation": "Highlighting deleted",
"time": "50.763",
"value": "John Robichaux_cell_0_4_1"
},
{
"operation": "Highlighting deleted",
"time": "50.771",
"value": "John Robichaux_header_cell_0_4_0"
},
{
"operation": "Highlighting deleted",
"time": "51.353",
"value": "John Robichaux_cell_0_5_1"
},
{
"operation": "Highlighting deleted",
"time": "51.36",
"value": "John Robichaux_header_cell_0_5_0"
},
{
"operation": "Highlighting deleted",
"time": "51.362",
"value": "John Robichaux_header_cell_0_1_0"
},
{
"operation": "Highlighting deleted",
"time": "51.364",
"value": "John Robichaux_header_cell_0_0_0"
},
{
"operation": "finish",
"time": "64.64",
"value": "finish"
}
]
}
|
63378
|
Martin was baptized in Olney, Buckinghamshire, England on May 30, 1621; then on August 11, 1646 at Salisbury, she married a blacksmith named George Martin.
|
{
"evidence": [
"Susannah Martin_sentence_4"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nMartin was baptized in Olney, Buckinghamshire, England on May 30, 1621; then on August 11, 1646 at Salisbury, she married a blacksmith named George Martin.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Susannah Martin_sentence_4"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7225,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.077",
"value": "?search="
},
{
"operation": "search",
"time": "45.871",
"value": "Susannah Martin"
},
{
"operation": "Now on",
"time": "47.138",
"value": "Susannah Martin"
},
{
"operation": "Highlighting",
"time": "61.679",
"value": "Susannah Martin_sentence_5"
},
{
"operation": "Highlighting",
"time": "67.055",
"value": "Susannah Martin_sentence_4"
},
{
"operation": "Highlighting",
"time": "71.385",
"value": "Susannah Martin_sentence_4"
},
{
"operation": "finish",
"time": "81.524",
"value": "finish"
}
]
}
|
exfever_6956
|
Suicide Kings is directed by Johnny Galecki that stars an American actor who has appeared in more than 100 films and television programs, including Annie Hall.
|
{
"evidence": [
"Suicide Kings",
"Christopher Walken"
],
"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:\nSuicide Kings is directed by Johnny Galecki that stars an American actor who has appeared in more than 100 films and television programs, including Annie Hall.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Suicide Kings",
"Christopher Walken"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7226,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
83955
|
Vem cards are the smart card system for bus, train, and metro use in the Recife metropolitan area, the eighth largest metropolitan area in Brazil with a population of about four million people with an area of about 1,069 square miles.
|
{
"evidence": [
"Recife_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:\nVem cards are the smart card system for bus, train, and metro use in the Recife metropolitan area, the eighth largest metropolitan area in Brazil with a population of about four million people with an area of about 1,069 square miles.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Recife_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7227,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.418",
"value": "?search="
},
{
"operation": "search",
"time": "23.323",
"value": "Contains...Vem cards"
},
{
"operation": "Now on",
"time": "24.199",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "35.554",
"value": "Vem"
},
{
"operation": "Now on",
"time": "36.252",
"value": "Vem"
},
{
"operation": "Highlighting",
"time": "41.166",
"value": "Vem_sentence_1"
},
{
"operation": "Highlighting",
"time": "41.167",
"value": "Vem_title"
},
{
"operation": "hyperlink",
"time": "71.385",
"value": "Recife"
},
{
"operation": "Now on",
"time": "72.887",
"value": "Recife"
},
{
"operation": "Highlighting",
"time": "75.617",
"value": "Recife_sentence_0"
},
{
"operation": "back-button-clicked",
"time": "96.997",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "97.171",
"value": "Vem"
},
{
"operation": "Highlighting deleted",
"time": "99.152",
"value": "Vem_sentence_1"
},
{
"operation": "Highlighting deleted",
"time": "99.155",
"value": "Vem_title"
},
{
"operation": "finish",
"time": "119.604",
"value": "finish"
}
]
}
|
exfever_6807
|
The Fame is Lady Gaga's debut studio album, which was produced by different producers, including Cherry Cherry Boom Boom, a British music personality.
|
{
"evidence": [
"The Fame",
"Martin Kierszenbaum"
],
"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 Fame is Lady Gaga's debut studio album, which was produced by different producers, including Cherry Cherry Boom Boom, a British music personality.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Fame",
"Martin Kierszenbaum"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7228,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
66698
|
The Casino Theatre presented an English fairy play, Little Red Riding Hood, in January 1900, wherein Jackson was 'Riding Hood', with Madge Lessing as 'Little Boy Blue', and Kitty Mitchell as 'Little Jack Horner'.
|
{
"evidence": [
"Ethel Jackson_sentence_26",
"Ethel Jackson_sentence_27"
],
"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 Casino Theatre presented an English fairy play, Little Red Riding Hood, in January 1900, wherein Jackson was 'Riding Hood', with Madge Lessing as 'Little Boy Blue', and Kitty Mitchell as 'Little Jack Horner'.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Ethel Jackson_sentence_26",
"Ethel Jackson_sentence_27"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7229,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.023",
"value": "?search="
},
{
"operation": "search",
"time": "29.413",
"value": "Little Red Riding Hood (disambiguation)"
},
{
"operation": "Now on",
"time": "30.749",
"value": "Little Red Riding Hood (disambiguation)"
},
{
"operation": "search",
"time": "60.68",
"value": "\"Casino Theatre presented an English fairy play\""
},
{
"operation": "Now on",
"time": "61.779",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "63.633",
"value": "Ethel Jackson"
},
{
"operation": "Now on",
"time": "64.486",
"value": "Ethel Jackson"
},
{
"operation": "Page search",
"time": "66.944",
"value": "Casino Theatre presented an English fairy play"
},
{
"operation": "Highlighting",
"time": "71.122",
"value": "Ethel Jackson_sentence_26"
},
{
"operation": "Highlighting",
"time": "74.701",
"value": "Ethel Jackson_sentence_27"
},
{
"operation": "finish",
"time": "92.099",
"value": "finish"
}
]
}
|
58653
|
Horticultural societies typically engage in small-scale gardening with simple tools, these methods allow for higher population densities and the common type of horticulture is slash-and-burn cultivation.
|
{
"evidence": [
"Horticulture_sentence_16",
"Horticulture_sentence_22",
"Horticulture_sentence_19",
"Horticulture_sentence_23",
"Royal Horticultural Society_sentence_0",
"Horticulture_sentence_3",
"Horticulture_sentence_8",
"Horticulture_sentence_14"
],
"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:\nHorticultural societies typically engage in small-scale gardening with simple tools, these methods allow for higher population densities and the common type of horticulture is slash-and-burn cultivation.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"Horticulture_sentence_16",
"Horticulture_sentence_22",
"Horticulture_sentence_19",
"Horticulture_sentence_23",
"Royal Horticultural Society_sentence_0",
"Horticulture_sentence_3",
"Horticulture_sentence_8",
"Horticulture_sentence_14"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7230,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "2.169",
"value": "?search="
},
{
"operation": "search",
"time": "18.693",
"value": "horticultural societies"
},
{
"operation": "Now on",
"time": "20.176",
"value": "undefined"
},
{
"operation": "search",
"time": "30.572",
"value": "Horticultural Society of London"
},
{
"operation": "Now on",
"time": "32.756",
"value": "Royal Horticultural Society"
},
{
"operation": "search",
"time": "89.502",
"value": "Horticulture"
},
{
"operation": "Now on",
"time": "91.7",
"value": "Horticulture"
},
{
"operation": "hyperlink",
"time": "100.325",
"value": "4 Organizations Horticulture_section_4"
},
{
"operation": "Page search",
"time": "119.744",
"value": "small"
},
{
"operation": "Highlighting",
"time": "139.081",
"value": "Horticulture_sentence_16"
},
{
"operation": "Highlighting",
"time": "156.562",
"value": "Horticulture_sentence_20"
},
{
"operation": "Highlighting deleted",
"time": "161.138",
"value": "Horticulture_sentence_20"
},
{
"operation": "Highlighting",
"time": "162.365",
"value": "Horticulture_sentence_21"
},
{
"operation": "Highlighting",
"time": "163.314",
"value": "Horticulture_sentence_22"
},
{
"operation": "Highlighting",
"time": "180.767",
"value": "Horticulture_sentence_19"
},
{
"operation": "Highlighting",
"time": "192.615",
"value": "Horticulture_sentence_23"
},
{
"operation": "Page search",
"time": "205.569",
"value": "slash"
},
{
"operation": "Page search",
"time": "227.889",
"value": "tool"
},
{
"operation": "Page search",
"time": "249.949",
"value": "densi*"
},
{
"operation": "Page search",
"time": "273.825",
"value": "densit"
},
{
"operation": "Page search",
"time": "276.7",
"value": "dens"
},
{
"operation": "search",
"time": "290.509",
"value": "slash-and-burn"
},
{
"operation": "Now on",
"time": "292.923",
"value": "undefined"
},
{
"operation": "search",
"time": "314.19",
"value": "horticultural society"
},
{
"operation": "Now on",
"time": "316.761",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "323.202",
"value": "Singapore Botanic Gardens"
},
{
"operation": "Now on",
"time": "328.765",
"value": "Singapore Botanic Gardens"
},
{
"operation": "back-button-clicked",
"time": "333.445",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "334.578",
"value": "undefined"
},
{
"operation": "search",
"time": "340.536",
"value": "horticultural societ"
},
{
"operation": "Now on",
"time": "341.289",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "344.761",
"value": "horticultural society"
},
{
"operation": "Now on",
"time": "346.758",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "349.608",
"value": "Francis Walker (entomologist)"
},
{
"operation": "Now on",
"time": "351.055",
"value": "Francis Walker (entomologist)"
},
{
"operation": "Page search",
"time": "362.131",
"value": "horticul"
},
{
"operation": "hyperlink",
"time": "368.6",
"value": "Royal Horticultural Society"
},
{
"operation": "Now on",
"time": "370.129",
"value": "Royal Horticultural Society"
},
{
"operation": "Highlighting",
"time": "372.002",
"value": "Royal Horticultural Society_sentence_0"
},
{
"operation": "search",
"time": "379.181",
"value": "horticulture"
},
{
"operation": "Now on",
"time": "380.828",
"value": "Horticulture"
},
{
"operation": "Highlighting",
"time": "389.835",
"value": "Horticulture_sentence_3"
},
{
"operation": "Highlighting",
"time": "401.363",
"value": "Horticulture_sentence_8"
},
{
"operation": "Highlighting",
"time": "418.38",
"value": "Horticulture_sentence_14"
},
{
"operation": "Highlighting deleted",
"time": "432.569",
"value": "Horticulture_sentence_21"
},
{
"operation": "finish",
"time": "460.781",
"value": "finish"
}
]
}
|
6537
|
The 1887 Kentucky gubernatorial election held on August 1 resulted to a win for Democratic candidate Simon Bolivar Buckner who won against closest competitor William O'Connell Bradley.
|
{
"evidence": [
"1887 Kentucky gubernatorial election_sentence_0",
"1887 Kentucky gubernatorial election_sentence_1"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe 1887 Kentucky gubernatorial election held on August 1 resulted to a win for Democratic candidate Simon Bolivar Buckner who won against closest competitor William O'Connell Bradley.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"1887 Kentucky gubernatorial election_sentence_0",
"1887 Kentucky gubernatorial election_sentence_1"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7231,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.726",
"value": "?search="
},
{
"operation": "search",
"time": "16.544",
"value": "1887 Kentucky gubernatorial election"
},
{
"operation": "Now on",
"time": "17.337",
"value": "1887 Kentucky gubernatorial election"
},
{
"operation": "Highlighting",
"time": "23.533",
"value": "1887 Kentucky gubernatorial election_sentence_0"
},
{
"operation": "Highlighting",
"time": "28.096",
"value": "1887 Kentucky gubernatorial election_sentence_1"
},
{
"operation": "finish",
"time": "49.194",
"value": "finish"
}
]
}
|
exfever_4078
|
NBA is the biggest basketball association in the world.
|
{
"evidence": [
"Kevin Durant",
"Golden State Warriors"
],
"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:\nNBA is the biggest basketball association in the world.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Kevin Durant",
"Golden State Warriors"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7232,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_1533
|
Artemis is the goddess of the hunt and the natural environments on Earth that are nonurbanized land not under extensive agricultural cultivation.
|
{
"evidence": [
"Artemis",
"Wilderness"
],
"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:\nArtemis is the goddess of the hunt and the natural environments on Earth that are nonurbanized land not under extensive agricultural cultivation.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Artemis",
"Wilderness"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7233,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_5047
|
Sue Sylvester is a fictional character that was developed by Glee's creator, an American television dancer best known for creating and producing a number of television series, including Popular.
|
{
"evidence": [
"Sue Sylvester",
"Ryan Murphy (writer)"
],
"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:\nSue Sylvester is a fictional character that was developed by Glee's creator, an American television dancer best known for creating and producing a number of television series, including Popular.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Sue Sylvester",
"Ryan Murphy (writer)"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7234,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
53736
|
An alternative approach is exemplified by the Scheffler Reflector, named after Wolfgang Scheffler (Special Recognition Awardee at the 2006 Nuclear-Free Future Award) a paraboloidal mirror which is rotated about axes that pass through its centre of mass.
|
{
"evidence": [
"Parabolic reflector_sentence_15",
"Parabolic reflector_sentence_16",
"Wolfgang Scheffler (inventor)_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:\nAn alternative approach is exemplified by the Scheffler Reflector, named after Wolfgang Scheffler (Special Recognition Awardee at the 2006 Nuclear-Free Future Award) a paraboloidal mirror which is rotated about axes that pass through its centre of mass.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Parabolic reflector_sentence_15",
"Parabolic reflector_sentence_16",
"Wolfgang Scheffler (inventor)_sentence_2"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7235,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.674",
"value": "?search="
},
{
"operation": "search",
"time": "18.134",
"value": "Contains...Scheffler Reflector"
},
{
"operation": "Now on",
"time": "18.831",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "22.885",
"value": "Scheffler reflector Parabolic reflector_section_3"
},
{
"operation": "Now on",
"time": "24.195",
"value": "Parabolic reflector#Scheffler reflector Parabolic reflector section 3"
},
{
"operation": "Highlighting",
"time": "31.262",
"value": "Parabolic reflector_sentence_15"
},
{
"operation": "Highlighting",
"time": "42.957",
"value": "Parabolic reflector_sentence_16"
},
{
"operation": "hyperlink",
"time": "55.725",
"value": "Wolfgang Scheffler"
},
{
"operation": "Now on",
"time": "57.092",
"value": "Wolfgang Scheffler (inventor)"
},
{
"operation": "Highlighting",
"time": "62.032",
"value": "Wolfgang Scheffler (inventor)_sentence_2"
},
{
"operation": "finish",
"time": "75.372",
"value": "finish"
}
]
}
|
exfever_2145
|
Dracula is a novel with no single antagonist, but opens with solicitor Jonathan Harker taking a business trip to stay at the castle of a historical region in central Romania.
|
{
"evidence": [
"Dracula",
"Transylvania"
],
"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:\nDracula is a novel with no single antagonist, but opens with solicitor Jonathan Harker taking a business trip to stay at the castle of a historical region in central Romania.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Dracula",
"Transylvania"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7236,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
49881
|
Oberrohrdorf is in the Baden district of Sweden, a small district with less than five municipalities.
|
{
"evidence": [
"Oberrohrdorf_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:\nOberrohrdorf is in the Baden district of Sweden, a small district with less than five municipalities.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Oberrohrdorf_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7237,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.988",
"value": "?search="
},
{
"operation": "search",
"time": "18.254",
"value": "Oberrohrdorf"
},
{
"operation": "Now on",
"time": "19.187",
"value": "Oberrohrdorf"
},
{
"operation": "Highlighting",
"time": "23.128",
"value": "Oberrohrdorf_sentence_0"
},
{
"operation": "finish",
"time": "35.025",
"value": "finish"
}
]
}
|
exfever_5940
|
The Clocks Higher Education World University Rankings is a yearly publication by Quacquarelli Symonds (QS).
|
{
"evidence": [
"Times Higher Education World University Rankings",
"QS World University Rankings"
],
"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 Clocks Higher Education World University Rankings is a yearly publication by Quacquarelli Symonds (QS).\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Times Higher Education World University Rankings",
"QS World University Rankings"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7238,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_1364
|
Harry Edward Styles was the youngest singer to appear solo on the cover of Vogue magazine.
|
{
"evidence": [
"Harry Styles",
"One Direction"
],
"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:\nHarry Edward Styles was the youngest singer to appear solo on the cover of Vogue magazine.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Harry Styles",
"One Direction"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7239,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_18
|
David Scott Jones' best-selling video game is Grand Theft Auto.
|
{
"evidence": [
"David Jones (video game developer)",
"Video game developer"
],
"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:\nDavid Scott Jones' best-selling video game is Grand Theft Auto.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"David Jones (video game developer)",
"Video game developer"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7240,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
45583
|
BraunAbility, founded by an owner of Braun Racing, established two new divisions in 1977 before a fire swept through their headquarters in 1979.
|
{
"evidence": [
"BraunAbility_sentence_21",
"BraunAbility_sentence_22"
],
"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:\nBraunAbility, founded by an owner of Braun Racing, established two new divisions in 1977 before a fire swept through their headquarters in 1979.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"BraunAbility_sentence_21",
"BraunAbility_sentence_22"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7241,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.001",
"value": "?search="
},
{
"operation": "search",
"time": "0.007",
"value": "BraunAbility"
},
{
"operation": "Now on",
"time": "0.008",
"value": "BraunAbility"
},
{
"operation": "Highlighting",
"time": "0.015",
"value": "BraunAbility_sentence_2"
},
{
"operation": "hyperlink",
"time": "0.02",
"value": "Ralph Braun"
},
{
"operation": "Now on",
"time": "0.02",
"value": "Ralph Braun"
},
{
"operation": "back-button-clicked",
"time": "0.031",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "0.032",
"value": "BraunAbility"
},
{
"operation": "search",
"time": "0.037",
"value": "braun racing"
},
{
"operation": "Now on",
"time": "0.041",
"value": "Braun Motorsports"
},
{
"operation": "Highlighting",
"time": "0.068",
"value": "Braun Motorsports_sentence_5"
},
{
"operation": "Highlighting",
"time": "0.068",
"value": "Braun Motorsports_title"
},
{
"operation": "back-button-clicked",
"time": "0.074",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "0.074",
"value": "BraunAbility"
},
{
"operation": "Highlighting",
"time": "0.089",
"value": "BraunAbility_sentence_21"
},
{
"operation": "Highlighting",
"time": "0.093",
"value": "BraunAbility_sentence_22"
},
{
"operation": "Highlighting",
"time": "0.102",
"value": "BraunAbility_sentence_21"
},
{
"operation": "Highlighting",
"time": "0.103",
"value": "BraunAbility_sentence_22"
},
{
"operation": "finish",
"time": "0.109",
"value": "finish"
},
{
"operation": "Now on",
"time": "1620846487.709",
"value": "BraunAbility"
},
{
"operation": "finish",
"time": "1620846491.945",
"value": "finish"
}
]
}
|
32459
|
The comic strip, the Brilliant Mind of Edison Lee, is about a ten-year old boy genius who discusses politics and creates inventions; the strip is drawn imprecisely.
|
{
"evidence": [
"The Brilliant Mind of Edison Lee_sentence_21"
],
"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 comic strip, the Brilliant Mind of Edison Lee, is about a ten-year old boy genius who discusses politics and creates inventions; the strip is drawn imprecisely.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"The Brilliant Mind of Edison Lee_sentence_21"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7242,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.182",
"value": "?search="
},
{
"operation": "search",
"time": "9.098",
"value": "Brilliant Mind of Edison Lee"
},
{
"operation": "Now on",
"time": "10.885",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "12.386",
"value": "The Brilliant Mind of Edison Lee"
},
{
"operation": "Now on",
"time": "12.954",
"value": "The Brilliant Mind of Edison Lee"
},
{
"operation": "Highlighting",
"time": "22.688",
"value": "The Brilliant Mind of Edison Lee_sentence_8"
},
{
"operation": "Highlighting",
"time": "22.69",
"value": "The Brilliant Mind of Edison Lee_section_0"
},
{
"operation": "Highlighting",
"time": "22.691",
"value": "The Brilliant Mind of Edison Lee_section_1"
},
{
"operation": "Highlighting",
"time": "27.932",
"value": "The Brilliant Mind of Edison Lee_sentence_9"
},
{
"operation": "Page search",
"time": "45.424",
"value": "drawn"
},
{
"operation": "Highlighting",
"time": "49.87",
"value": "The Brilliant Mind of Edison Lee_sentence_21"
},
{
"operation": "finish",
"time": "60.554",
"value": "finish"
}
]
}
|
exfever_5113
|
After Winona Ryder's film debut in an American romanctic film written and directed by Leo McCarey, she gained attention with her performance in Tim Burton's Beetlejuice 1988.
|
{
"evidence": [
"Winona Ryder",
"Lucas (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:\nAfter Winona Ryder's film debut in an American romanctic film written and directed by Leo McCarey, she gained attention with her performance in Tim Burton's Beetlejuice 1988.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Winona Ryder",
"Lucas (film)"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7243,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
34538
|
Kenneth Riegel engaged at the New York City Opera from 1969 to 1974 before making his first appearance at the Salzburg Festival in 1975.
|
{
"evidence": [
"Kenneth Riegel_sentence_3",
"Kenneth Riegel_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:\nKenneth Riegel engaged at the New York City Opera from 1969 to 1974 before making his first appearance at the Salzburg Festival in 1975.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Kenneth Riegel_sentence_3",
"Kenneth Riegel_sentence_5"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7244,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "2.209",
"value": "?search="
},
{
"operation": "search",
"time": "14.441",
"value": "Kenneth Riegel"
},
{
"operation": "Now on",
"time": "15.813",
"value": "Kenneth Riegel"
},
{
"operation": "Page search",
"time": "20.413",
"value": "engage"
},
{
"operation": "Highlighting",
"time": "33.432",
"value": "Kenneth Riegel_sentence_3"
},
{
"operation": "Highlighting",
"time": "37.849",
"value": "Kenneth Riegel_sentence_5"
},
{
"operation": "finish",
"time": "45.971",
"value": "finish"
}
]
}
|
19486
|
American character actor Herb Vigran (June 5, 1910 – November 29, 1986) only acted in television shows.
|
{
"evidence": [
"Herb Vigran_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:\nAmerican character actor Herb Vigran (June 5, 1910 – November 29, 1986) only acted in television shows.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Herb Vigran_sentence_1"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7245,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.605",
"value": "?search="
},
{
"operation": "search",
"time": "6.223",
"value": "Herb Vigran"
},
{
"operation": "Now on",
"time": "7.805",
"value": "Herb Vigran"
},
{
"operation": "Highlighting",
"time": "9.03",
"value": "Herb Vigran_sentence_0"
},
{
"operation": "Highlighting",
"time": "21.638",
"value": "Herb Vigran_cell_1_0_1"
},
{
"operation": "Highlighting",
"time": "21.639",
"value": "Herb Vigran_section_7"
},
{
"operation": "Highlighting",
"time": "22.068",
"value": "Herb Vigran_cell_1_1_1"
},
{
"operation": "Highlighting",
"time": "23.357",
"value": "Herb Vigran_cell_2_0_1"
},
{
"operation": "Highlighting",
"time": "23.358",
"value": "Herb Vigran_section_8"
},
{
"operation": "Highlighting",
"time": "23.647",
"value": "Herb Vigran_cell_2_1_1"
},
{
"operation": "Highlighting",
"time": "28.482",
"value": "Herb Vigran_header_cell_4_0_1"
},
{
"operation": "Highlighting",
"time": "28.482",
"value": "Herb Vigran_section_10"
},
{
"operation": "Highlighting",
"time": "28.807",
"value": "Herb Vigran_cell_4_1_1"
},
{
"operation": "Highlighting",
"time": "28.813",
"value": "Herb Vigran_header_cell_4_0_1"
},
{
"operation": "Highlighting",
"time": "47.325",
"value": "Herb Vigran_sentence_1"
},
{
"operation": "Highlighting",
"time": "50.977",
"value": "Herb Vigran_sentence_1"
},
{
"operation": "finish",
"time": "54.324",
"value": "finish"
}
]
}
|
exfever_647
|
Guitar is the mosst played instrument all over the world.
|
{
"evidence": [
"Chet Atkins",
"Les Paul"
],
"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:\nGuitar is the mosst played instrument all over the world.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Chet Atkins",
"Les Paul"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7246,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_6696
|
The Man with the Iron Fists is a 2012 American martial arts film that features a Hong Kong-American person who portrays a character in a performance in film and television.
|
{
"evidence": [
"The Man with the Iron Fists",
"Byron Mann",
"Actor"
],
"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 Man with the Iron Fists is a 2012 American martial arts film that features a Hong Kong-American person who portrays a character in a performance in film and television.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Man with the Iron Fists",
"Byron Mann",
"Actor"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7247,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_2579
|
Sammy Cahn and his collaborators had a series of hit recordings during the singer's tenure at Capitol Records, with an Australian singer and actor, one of the best-selling music artists of all time.
|
{
"evidence": [
"Sammy Cahn",
"Frank Sinatra"
],
"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:\nSammy Cahn and his collaborators had a series of hit recordings during the singer's tenure at Capitol Records, with an Australian singer and actor, one of the best-selling music artists of all time.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Sammy Cahn",
"Frank Sinatra"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7248,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_830
|
Velaiilla Pattadhari 2 is the last season of the same name.
|
{
"evidence": [
"Velaiilla Pattadhari 2",
"Amala Paul"
],
"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:\nVelaiilla Pattadhari 2 is the last season of the same name.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Velaiilla Pattadhari 2",
"Amala Paul"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7249,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_6977
|
Atlantic City is a romantic crime film starring a Canadian actor that was best known for his television roles on a Canadian public broadcaster for both radio and television.
|
{
"evidence": [
"Atlantic City (1980 film)",
"Al Waxman",
"Canadian Broadcasting Corporation"
],
"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:\nAtlantic City is a romantic crime film starring a Canadian actor that was best known for his television roles on a Canadian public broadcaster for both radio and television.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Atlantic City (1980 film)",
"Al Waxman",
"Canadian Broadcasting Corporation"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7250,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
11089
|
Weigang, the first violinist for the Shanghai Quartet since its inception, has been a soloist with the Asian Youth Orchestra, Shanghai Symphony Orchestra, BBC Symphony Orchestra, and BBC Scottish Symphony Orchestra.
|
{
"evidence": [
"Shanghai Quartet_sentence_21",
"Shanghai Quartet_sentence_22"
],
"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:\nWeigang, the first violinist for the Shanghai Quartet since its inception, has been a soloist with the Asian Youth Orchestra, Shanghai Symphony Orchestra, BBC Symphony Orchestra, and BBC Scottish Symphony Orchestra.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"Shanghai Quartet_sentence_21",
"Shanghai Quartet_sentence_22"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7251,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.516",
"value": "?search="
},
{
"operation": "search",
"time": "19.805",
"value": "Contains...Weigang"
},
{
"operation": "Now on",
"time": "20.422",
"value": "undefined"
},
{
"operation": "search",
"time": "35.33",
"value": "Shanghai Quartet"
},
{
"operation": "Now on",
"time": "36.142",
"value": "Shanghai Quartet"
},
{
"operation": "Highlighting",
"time": "45.671",
"value": "Shanghai Quartet_sentence_1"
},
{
"operation": "Highlighting",
"time": "70.803",
"value": "Shanghai Quartet_sentence_21"
},
{
"operation": "Highlighting",
"time": "82.996",
"value": "Shanghai Quartet_sentence_22"
},
{
"operation": "Highlighting deleted",
"time": "98.485",
"value": "Shanghai Quartet_sentence_1"
},
{
"operation": "finish",
"time": "125.69",
"value": "finish"
}
]
}
|
38374
|
1950 Texas gubernatorial election was held on November 7,1950, and the incumbent Governor Allan Shavers win his reelection bid.
|
{
"evidence": [
"1950 Texas gubernatorial election_sentence_0",
"1950 Texas gubernatorial election_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:\n1950 Texas gubernatorial election was held on November 7,1950, and the incumbent Governor Allan Shavers win his reelection bid.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"1950 Texas gubernatorial election_sentence_0",
"1950 Texas gubernatorial election_sentence_1"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7252,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.909",
"value": "?search="
},
{
"operation": "search",
"time": "14.763",
"value": "1950 Texas gubernatorial election"
},
{
"operation": "Now on",
"time": "15.753",
"value": "1950 Texas gubernatorial election"
},
{
"operation": "Highlighting",
"time": "45.137",
"value": "1950 Texas gubernatorial election_sentence_0"
},
{
"operation": "Highlighting",
"time": "49.663",
"value": "1950 Texas gubernatorial election_sentence_1"
},
{
"operation": "finish",
"time": "181.937",
"value": "finish"
}
]
}
|
exfever_8275
|
Brad Wilk is best known as the drummer of an American rock supergroup formed in Glendale, California, in 2000.
|
{
"evidence": [
"Brad Wilk",
"Audioslave"
],
"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:\nBrad Wilk is best known as the drummer of an American rock supergroup formed in Glendale, California, in 2000.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Brad Wilk",
"Audioslave"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7253,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_200
|
Twitter is an American online broadcast medium that exists as a specific form of blogging.
|
{
"evidence": [
"Twitter",
"Microblogging"
],
"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:\nTwitter is an American online broadcast medium that exists as a specific form of blogging.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Twitter",
"Microblogging"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7254,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_2359
|
"Just the Way You Are" is the debut solo single by Bruno Mars and written by American songwriting and record production team established in a state in the Western United States with over 39.5 million residents.
|
{
"evidence": [
"Just the Way You Are (Bruno Mars song)",
"The Smeezingtons",
"California"
],
"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:\n\"Just the Way You Are\" is the debut solo single by Bruno Mars and written by American songwriting and record production team established in a state in the Western United States with over 39.5 million residents.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Just the Way You Are (Bruno Mars song)",
"The Smeezingtons",
"California"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7255,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_5018
|
"Live Your Life" was the most famous song from T.I.'s sixth studio album.
|
{
"evidence": [
"Live Your Life (T.I. song)",
"O-Zone"
],
"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:\n\"Live Your Life\" was the most famous song from T.I.'s sixth studio album.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Live Your Life (T.I. song)",
"O-Zone"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7256,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
79258
|
Hidden Figures was the highest grossing film for the 89th Academy Awards.
|
{
"evidence": [
"Hidden Figures_sentence_140"
],
"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:\nHidden Figures was the highest grossing film for the 89th Academy Awards.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Hidden Figures_sentence_140"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7257,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.153",
"value": "?search="
},
{
"operation": "search",
"time": "43.297",
"value": "Hidden Figures"
},
{
"operation": "Now on",
"time": "44.261",
"value": "Hidden Figures"
},
{
"operation": "Highlighting",
"time": "99.714",
"value": "Hidden Figures_sentence_10"
},
{
"operation": "Highlighting",
"time": "108.226",
"value": "Hidden Figures_sentence_140"
},
{
"operation": "search",
"time": "151.806",
"value": "89th Academy Awards"
},
{
"operation": "Now on",
"time": "152.937",
"value": "89th Academy Awards"
},
{
"operation": "finish",
"time": "191.148",
"value": "finish"
}
]
}
|
35001
|
Caguas (a historic place in central Puerto Rico) had seven listings on the registry of Historic Places, and five of these listings are found in a barrio called Caguas Pueblo.
|
{
"evidence": [
"Caguas, Puerto Rico_sentence_1"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nCaguas (a historic place in central Puerto Rico) had seven listings on the registry of Historic Places, and five of these listings are found in a barrio called Caguas Pueblo.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Caguas, Puerto Rico_sentence_1"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7258,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.308",
"value": "?search="
},
{
"operation": "search",
"time": "9.507",
"value": "Caguas"
},
{
"operation": "Now on",
"time": "11.545",
"value": "Caguas, Puerto Rico"
},
{
"operation": "Highlighting",
"time": "34.419",
"value": "Caguas, Puerto Rico_sentence_1"
},
{
"operation": "search",
"time": "57.563",
"value": "Caguas, Puerto Rico"
},
{
"operation": "Now on",
"time": "58.928",
"value": "Caguas, Puerto Rico"
},
{
"operation": "hyperlink",
"time": "93.124",
"value": "1.1 Hurricane Maria Caguas, Puerto Rico_section_1"
},
{
"operation": "finish",
"time": "117.344",
"value": "finish"
}
]
}
|
exfever_8363
|
Terminator Genisys is a Canadian science fiction action film directed by an American television and film director known for his work on TV shows such as Lost, The West Wing, Six Feet Under, Sex and the City, The Sopranos, Game of Thrones, Boardwalk Empire, Deadwood, and Mad Men.
|
{
"evidence": [
"Terminator Genisys",
"Alan Taylor (director)"
],
"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:\nTerminator Genisys is a Canadian science fiction action film directed by an American television and film director known for his work on TV shows such as Lost, The West Wing, Six Feet Under, Sex and the City, The Sopranos, Game of Thrones, Boardwalk Empire, Deadwood, and Mad Men.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Terminator Genisys",
"Alan Taylor (director)"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7259,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_5726
|
Sansa Stark is a fictional character in the A Song of Ice and Fire novel who later on appeared in the second novel in A Song of Ice and Fire, an epic fantasy series by Russian author George R. R. Martin expected to consist of seven volumes.
|
{
"evidence": [
"Sansa Stark",
"A Clash of Kings"
],
"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:\nSansa Stark is a fictional character in the A Song of Ice and Fire novel who later on appeared in the second novel in A Song of Ice and Fire, an epic fantasy series by Russian author George R. R. Martin expected to consist of seven volumes.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Sansa Stark",
"A Clash of Kings"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7260,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
64476
|
Pharmacoepigenetics (an old field that studies the underlying epigenetic marking patterns) loss of function can lead to hypomethylation and hypermethylation of DNA promoters.
|
{
"evidence": [
"Pharmacoepigenetics_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:\nPharmacoepigenetics (an old field that studies the underlying epigenetic marking patterns) loss of function can lead to hypomethylation and hypermethylation of DNA promoters.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Pharmacoepigenetics_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7261,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.352",
"value": "?search="
},
{
"operation": "search",
"time": "9.393",
"value": "Pharmacoepigenetics"
},
{
"operation": "Now on",
"time": "10.544",
"value": "Pharmacoepigenetics"
},
{
"operation": "Highlighting",
"time": "13.495",
"value": "Pharmacoepigenetics_sentence_0"
},
{
"operation": "Page search",
"time": "33.884",
"value": "DNA"
},
{
"operation": "finish",
"time": "127.6",
"value": "finish"
}
]
}
|
70903
|
Hazel Wood Waterman life changed when Waldo died in 1901 and she started studying architectural drafting and was later hired by Gills firm.
|
{
"evidence": [
"Hazel Wood Waterman_sentence_15"
],
"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:\nHazel Wood Waterman life changed when Waldo died in 1901 and she started studying architectural drafting and was later hired by Gills firm.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Hazel Wood Waterman_sentence_15"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7262,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.741",
"value": "?search="
},
{
"operation": "search",
"time": "9.621",
"value": "Hazel Wood Waterman"
},
{
"operation": "Now on",
"time": "13.771",
"value": "Hazel Wood Waterman"
},
{
"operation": "Page search",
"time": "21.525",
"value": "waldo"
},
{
"operation": "Highlighting",
"time": "40.52",
"value": "Hazel Wood Waterman_sentence_15"
},
{
"operation": "Highlighting",
"time": "46.661",
"value": "Hazel Wood Waterman_sentence_16"
},
{
"operation": "Highlighting",
"time": "53.543",
"value": "Hazel Wood Waterman_sentence_18"
},
{
"operation": "finish",
"time": "86.773",
"value": "finish"
}
]
}
|
31148
|
In Éloi Laurent Despeaux's class the regular troops consisted of the 1st Battalions of the 17th and 25th Line Infantry Demi-brigades, and all infantry battalions were numbered between 783 and 919 soldiers.
|
{
"evidence": [
"Éloi Laurent Despeaux_sentence_30",
"Éloi Laurent Despeaux_sentence_32"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nIn Éloi Laurent Despeaux's class the regular troops consisted of the 1st Battalions of the 17th and 25th Line Infantry Demi-brigades, and all infantry battalions were numbered between 783 and 919 soldiers.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Éloi Laurent Despeaux_sentence_30",
"Éloi Laurent Despeaux_sentence_32"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7263,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.825",
"value": "?search="
},
{
"operation": "search",
"time": "51.853",
"value": "In Eloi Laurent Despeaux"
},
{
"operation": "Now on",
"time": "52.704",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "55.124",
"value": "Éloi Laurent Despeaux"
},
{
"operation": "Now on",
"time": "55.718",
"value": "Éloi Laurent Despeaux"
},
{
"operation": "Page search",
"time": "71.291",
"value": "class"
},
{
"operation": "Page search",
"time": "89.435",
"value": "regular"
},
{
"operation": "Highlighting",
"time": "96.948",
"value": "Éloi Laurent Despeaux_sentence_30"
},
{
"operation": "Highlighting",
"time": "107.699",
"value": "Éloi Laurent Despeaux_sentence_32"
},
{
"operation": "finish",
"time": "114.555",
"value": "finish"
}
]
}
|
exfever_1588
|
Berlin is the capital and largest city in Germany by both area and population, and the most populous city in a political and economic union of 27 member states that are located primarily in Asia.
|
{
"evidence": [
"Berlin",
"European Union"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nBerlin is the capital and largest city in Germany by both area and population, and the most populous city in a political and economic union of 27 member states that are located primarily in Asia.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Berlin",
"European Union"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7264,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
53803
|
In the field of passive fluency, commonly known as passive bilingualism, François Grosjean argues that people with no equal competence in multiple languages are not "bilinguals."
|
{
"evidence": [
"Passive fluency_sentence_3",
"Multilingualism_sentence_123",
"François Grosjean_sentence_3",
"François Grosjean_sentence_4"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nIn the field of passive fluency, commonly known as passive bilingualism, François Grosjean argues that people with no equal competence in multiple languages are not \"bilinguals.\"\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Passive fluency_sentence_3",
"Multilingualism_sentence_123",
"François Grosjean_sentence_3",
"François Grosjean_sentence_4"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7265,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.3",
"value": "?search="
},
{
"operation": "search",
"time": "11.726",
"value": "Passive fluency"
},
{
"operation": "Now on",
"time": "12.439",
"value": "Passive fluency"
},
{
"operation": "Highlighting",
"time": "20.047",
"value": "Passive fluency_sentence_3"
},
{
"operation": "Page search",
"time": "39.223",
"value": "François Grosjean"
},
{
"operation": "search",
"time": "82.569",
"value": "François Grosjean argues that people with no equal competence in multiple languages are not \"bilinguals"
},
{
"operation": "Now on",
"time": "84.355",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "96.58",
"value": "Multilingualism"
},
{
"operation": "Now on",
"time": "97.905",
"value": "Multilingualism"
},
{
"operation": "Page search",
"time": "104.851",
"value": "François Grosjean"
},
{
"operation": "Page search",
"time": "105.096",
"value": "François Grosjean"
},
{
"operation": "Page search",
"time": "142.116",
"value": "people with no equal competence in multiple languages are not bilinguals"
},
{
"operation": "Page search",
"time": "142.892",
"value": "people with no equal competence in multiple languages are not bilinguals"
},
{
"operation": "Page search",
"time": "170.046",
"value": "François Grosjean"
},
{
"operation": "Highlighting",
"time": "183.216",
"value": "Multilingualism_sentence_123"
},
{
"operation": "search",
"time": "190.506",
"value": "François Grosjean"
},
{
"operation": "Now on",
"time": "191.239",
"value": "François Grosjean"
},
{
"operation": "Page search",
"time": "204.895",
"value": "people with no equal competence"
},
{
"operation": "Highlighting",
"time": "226.032",
"value": "François Grosjean_sentence_3"
},
{
"operation": "Highlighting",
"time": "228.138",
"value": "François Grosjean_sentence_4"
},
{
"operation": "search",
"time": "298.401",
"value": "myths about bilingualism"
},
{
"operation": "Now on",
"time": "299.915",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "309.42",
"value": "Multilingualism"
},
{
"operation": "Now on",
"time": "310.156",
"value": "Multilingualism"
},
{
"operation": "finish",
"time": "341.311",
"value": "finish"
}
]
}
|
27019
|
Kyrian Nwabueze signed with United Soccer League club Tulsa Roughnecks in March 2016 and in August 2016, Nwabueze signed for Armenian side Ararat Yerevan.
|
{
"evidence": [
"Kyrian Nwabueze_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:\nKyrian Nwabueze signed with United Soccer League club Tulsa Roughnecks in March 2016 and in August 2016, Nwabueze signed for Armenian side Ararat Yerevan.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Kyrian Nwabueze_sentence_3"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7266,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.127",
"value": "?search="
},
{
"operation": "search",
"time": "122.56",
"value": "Kyrian Nwabueze"
},
{
"operation": "Now on",
"time": "123.245",
"value": "Kyrian Nwabueze"
},
{
"operation": "Highlighting",
"time": "185.471",
"value": "Kyrian Nwabueze_sentence_3"
},
{
"operation": "Page search",
"time": "201.782",
"value": "united soccer league"
},
{
"operation": "finish",
"time": "219.153",
"value": "finish"
}
]
}
|
21089
|
Licensing was done with broadcasting leading music being later refused for the Only Fools and Horses home video releases.
|
{
"evidence": [
"List of Only Fools and Horses home video releases_sentence_8",
"List of Only Fools and Horses home video releases_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:\nLicensing was done with broadcasting leading music being later refused for the Only Fools and Horses home video releases.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"List of Only Fools and Horses home video releases_sentence_8",
"List of Only Fools and Horses home video releases_sentence_7"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7267,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.594",
"value": "?search="
},
{
"operation": "search",
"time": "15.417",
"value": "Only Fools and Horses"
},
{
"operation": "Now on",
"time": "17.491",
"value": "Only Fools and Horses"
},
{
"operation": "Page search",
"time": "23.995",
"value": "video releases"
},
{
"operation": "hyperlink",
"time": "35.677",
"value": "List of Only Fools and Horses home video releases"
},
{
"operation": "Now on",
"time": "36.54",
"value": "List of Only Fools and Horses home video releases"
},
{
"operation": "Page search",
"time": "43.123",
"value": "Licensing"
},
{
"operation": "Highlighting",
"time": "107.28",
"value": "List of Only Fools and Horses home video releases_sentence_7"
},
{
"operation": "Highlighting",
"time": "161.468",
"value": "List of Only Fools and Horses home video releases_sentence_8"
},
{
"operation": "page-search-reset",
"time": "194.517",
"value": "page-search-reset"
},
{
"operation": "Highlighting deleted",
"time": "199.899",
"value": "List of Only Fools and Horses home video releases_sentence_7"
},
{
"operation": "Highlighting",
"time": "201.344",
"value": "List of Only Fools and Horses home video releases_sentence_7"
},
{
"operation": "finish",
"time": "271.595",
"value": "finish"
}
]
}
|
68856
|
The Natural Four was signed by ABC Records after their first release, "I Thought You Were Mine" rose to number seven on the charts and sold 30,000 copies; their subsequent releases did not repeat the success of their initial recording.
|
{
"evidence": [
"The Natural Four_sentence_4",
"The Natural Four_sentence_3",
"The Natural Four_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:\nThe Natural Four was signed by ABC Records after their first release, \"I Thought You Were Mine\" rose to number seven on the charts and sold 30,000 copies; their subsequent releases did not repeat the success of their initial recording.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"The Natural Four_sentence_4",
"The Natural Four_sentence_3",
"The Natural Four_sentence_5"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7268,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.241",
"value": "?search="
},
{
"operation": "search",
"time": "15.28",
"value": "The Natural Four"
},
{
"operation": "Now on",
"time": "16.49",
"value": "The Natural Four"
},
{
"operation": "Highlighting",
"time": "20.73",
"value": "The Natural Four_sentence_4"
},
{
"operation": "Highlighting",
"time": "22.679",
"value": "The Natural Four_sentence_3"
},
{
"operation": "Highlighting",
"time": "41.6",
"value": "The Natural Four_sentence_5"
},
{
"operation": "Highlighting",
"time": "45.799",
"value": "The Natural Four_sentence_6"
},
{
"operation": "Highlighting deleted",
"time": "78.659",
"value": "The Natural Four_sentence_6"
},
{
"operation": "finish",
"time": "91.549",
"value": "finish"
}
]
}
|
exfever_5099
|
LeBron James is a basketball player in a professional basketball league in North America.
|
{
"evidence": [
"LeBron James",
"National Basketball Association"
],
"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:\nLeBron James is a basketball player in a professional basketball league in North America.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"LeBron James",
"National Basketball Association"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7269,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
66055
|
Ancestral Puebloans, an ancient Native American culture that spanned the present-day Four Corners region of the United States and never known for their pottery, have homes and villages that were based on the pit-house, a common feature in the Basketmaker periods.
|
{
"evidence": [
"Ancestral Puebloans_sentence_44"
],
"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:\nAncestral Puebloans, an ancient Native American culture that spanned the present-day Four Corners region of the United States and never known for their pottery, have homes and villages that were based on the pit-house, a common feature in the Basketmaker periods.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Ancestral Puebloans_sentence_44"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7270,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "5.432",
"value": "?search="
},
{
"operation": "search",
"time": "7.744",
"value": "Ancestral Puebloans"
},
{
"operation": "Now on",
"time": "15.396",
"value": "Ancestral Puebloans"
},
{
"operation": "hyperlink",
"time": "40.228",
"value": "8 Anasazi as a cultural label Ancestral Puebloans_section_12"
},
{
"operation": "Page search",
"time": "194.744",
"value": "pottery"
},
{
"operation": "Highlighting",
"time": "199.409",
"value": "Ancestral Puebloans_sentence_44"
},
{
"operation": "Page search",
"time": "217.978",
"value": "basketmaker"
},
{
"operation": "hyperlink",
"time": "239.512",
"value": "Basketmaker"
},
{
"operation": "Now on",
"time": "241.16",
"value": "Basketmaker culture"
},
{
"operation": "finish",
"time": "243.413",
"value": "finish"
}
]
}
|
exfever_7940
|
Seth Aaron Rogen moved to Canada for a part in Judd Apatow's series, a television series created by Paul Feig that aired on NBC during the 1999–2000 television season.
|
{
"evidence": [
"Seth Rogen",
"Freaks and Geeks"
],
"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:\nSeth Aaron Rogen moved to Canada for a part in Judd Apatow's series, a television series created by Paul Feig that aired on NBC during the 1999–2000 television season.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Seth Rogen",
"Freaks and Geeks"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7271,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
6672
|
In the 1997 Seattle Seahawks season, which was the team's 22nd season with the National Football League, they signed a rookie quarterback and traded quarterback Rick Mirer.
|
{
"evidence": [
"1997 Seattle Seahawks season_sentence_0",
"1997 Seattle Seahawks season_sentence_3",
"1997 Seattle Seahawks season_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:\nIn the 1997 Seattle Seahawks season, which was the team's 22nd season with the National Football League, they signed a rookie quarterback and traded quarterback Rick Mirer.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"1997 Seattle Seahawks season_sentence_0",
"1997 Seattle Seahawks season_sentence_3",
"1997 Seattle Seahawks season_sentence_1"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7272,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.781",
"value": "?search="
},
{
"operation": "search",
"time": "9.231",
"value": "1997 Seattle Seahawks season"
},
{
"operation": "Now on",
"time": "10.124",
"value": "1997 Seattle Seahawks season"
},
{
"operation": "Highlighting",
"time": "13.026",
"value": "1997 Seattle Seahawks season_sentence_0"
},
{
"operation": "Highlighting",
"time": "41.074",
"value": "1997 Seattle Seahawks season_sentence_3"
},
{
"operation": "Highlighting",
"time": "49.375",
"value": "1997 Seattle Seahawks season_sentence_1"
},
{
"operation": "finish",
"time": "74.475",
"value": "finish"
}
]
}
|
exfever_1109
|
Collin Farrell is an Irish actor who was discovered by Hollywood upon taking the lead role in an American war drama film directed by Joel Schumacher.
|
{
"evidence": [
"Colin Farrell",
"Tigerland"
],
"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:\nCollin Farrell is an Irish actor who was discovered by Hollywood upon taking the lead role in an American war drama film directed by Joel Schumacher.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Colin Farrell",
"Tigerland"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7273,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_5258
|
Mark Bryan is an American singer who gained fame as the lead vocalist of Hootie & the Blowfish, who released a solo album in 2002 on an American independent record label that was formed in 1998 by Steven McKeever.
|
{
"evidence": [
"Darius Rucker",
"Hidden Beach Recordings"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nMark Bryan is an American singer who gained fame as the lead vocalist of Hootie & the Blowfish, who released a solo album in 2002 on an American independent record label that was formed in 1998 by Steven McKeever.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Darius Rucker",
"Hidden Beach Recordings"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7274,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
94518
|
ATB Financial is a manufacturing company based in Edmonton, Alberta.
|
{
"evidence": [
"ATB Financial_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:\nATB Financial is a manufacturing company based in Edmonton, Alberta.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"ATB Financial_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7275,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.029",
"value": "?search="
},
{
"operation": "search",
"time": "41.389",
"value": "ATB Financial"
},
{
"operation": "Now on",
"time": "42.195",
"value": "ATB Financial"
},
{
"operation": "Highlighting",
"time": "103.294",
"value": "ATB Financial_sentence_0"
},
{
"operation": "finish",
"time": "108.862",
"value": "finish"
}
]
}
|
38267
|
Munni Begum grew up in Pakistan (Bangladesh) in her early life, and studied music for six years before launching her career.
|
{
"evidence": [
"Munni Begum_sentence_6",
"Munni Begum_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:\nMunni Begum grew up in Pakistan (Bangladesh) in her early life, and studied music for six years before launching her career.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Munni Begum_sentence_6",
"Munni Begum_sentence_7"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7276,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.876",
"value": "?search="
},
{
"operation": "search",
"time": "6.338",
"value": "Munni Begum"
},
{
"operation": "Now on",
"time": "7.152",
"value": "Munni Begum"
},
{
"operation": "Highlighting",
"time": "85.839",
"value": "Munni Begum_sentence_6"
},
{
"operation": "Highlighting",
"time": "92.281",
"value": "Munni Begum_sentence_7"
},
{
"operation": "finish",
"time": "96.51",
"value": "finish"
}
]
}
|
exfever_8395
|
David Lawrence Schwimmer is an American actor, comedian, director, and producer who starred in the 1998 American action film based on Stephen King's 1982 novella.
|
{
"evidence": [
"David Schwimmer",
"Apt Pupil (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:\nDavid Lawrence Schwimmer is an American actor, comedian, director, and producer who starred in the 1998 American action film based on Stephen King's 1982 novella.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"David Schwimmer",
"Apt Pupil (film)"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7277,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_167
|
Daniel Carey is the drummer for an American rock band from Los Angeles.
|
{
"evidence": [
"Danny Carey",
"Tool (band)"
],
"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:\nDaniel Carey is the drummer for an American rock band from Los Angeles.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Danny Carey",
"Tool (band)"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7278,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_2621
|
Tyler Perry is an American screenwriter who has created multiple scripted series, including an English crime drama that premiered on May 28, 2013.
|
{
"evidence": [
"Tyler Perry",
"The Haves and the Have Nots (TV series)"
],
"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:\nTyler Perry is an American screenwriter who has created multiple scripted series, including an English crime drama that premiered on May 28, 2013.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Tyler Perry",
"The Haves and the Have Nots (TV series)"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7279,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_1446
|
Saswata Chatterjee is an Indian actor whose work has been praised, including his portrayal of an Indian film director, along with Bengali filmmakers, like one of the most prominent Indian film directors of his time.
|
{
"evidence": [
"Saswata Chatterjee",
"Ritwik Ghatak",
"Tapan Sinha"
],
"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:\nSaswata Chatterjee is an Indian actor whose work has been praised, including his portrayal of an Indian film director, along with Bengali filmmakers, like one of the most prominent Indian film directors of his time.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Saswata Chatterjee",
"Ritwik Ghatak",
"Tapan Sinha"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7280,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_4513
|
Jack Roosevelt Robinson was an American professional baseball player who became the first African American to play in the oldest major professional sports league in the world started in the period from around 1900 to the emergence of Babe Ruth as a power hitter in 1919 when he hit a then-league record 29 home runs.
|
{
"evidence": [
"Jackie Robinson",
"Major League Baseball",
"Dead-ball era"
],
"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:\nJack Roosevelt Robinson was an American professional baseball player who became the first African American to play in the oldest major professional sports league in the world started in the period from around 1900 to the emergence of Babe Ruth as a power hitter in 1919 when he hit a then-league record 29 home runs.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Jackie Robinson",
"Major League Baseball",
"Dead-ball era"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7281,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
14946
|
Praveen Kumar Nishad was selected as the Samajwadi candidate in Gorakhpur, where the Nishad community is the second-largest demographic group,In a major upset, Praveen Nishad wrested the seat from BJP, which had not lost the seat since 1989.
|
{
"evidence": [
"Praveen Kumar Nishad_sentence_5",
"Praveen Kumar Nishad_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:\nPraveen Kumar Nishad was selected as the Samajwadi candidate in Gorakhpur, where the Nishad community is the second-largest demographic group,In a major upset, Praveen Nishad wrested the seat from BJP, which had not lost the seat since 1989.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Praveen Kumar Nishad_sentence_5",
"Praveen Kumar Nishad_sentence_4"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7282,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.695",
"value": "?search="
},
{
"operation": "search",
"time": "11.354",
"value": "Praveen Kumar Nishad"
},
{
"operation": "Now on",
"time": "12.213",
"value": "Praveen Kumar Nishad"
},
{
"operation": "Highlighting",
"time": "104.252",
"value": "Praveen Kumar Nishad_sentence_5"
},
{
"operation": "Highlighting",
"time": "104.989",
"value": "Praveen Kumar Nishad_sentence_4"
},
{
"operation": "finish",
"time": "133.199",
"value": "finish"
}
]
}
|
15841
|
The 1892 United States presidential election in Illinois is where 24 electors were chosen at random leading to a Democratic candidate winner for the Electoral College, which forms every four years for each state.
|
{
"evidence": [
"1892 United States presidential election in Illinois_sentence_3",
"1892 United States presidential election in Illinois_sentence_4"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe 1892 United States presidential election in Illinois is where 24 electors were chosen at random leading to a Democratic candidate winner for the Electoral College, which forms every four years for each state.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"1892 United States presidential election in Illinois_sentence_3",
"1892 United States presidential election in Illinois_sentence_4"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7283,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.611",
"value": "?search="
},
{
"operation": "search",
"time": "12.672",
"value": "1892 united states presidential election in illinois"
},
{
"operation": "Now on",
"time": "13.836",
"value": "1892 United States presidential election in Illinois"
},
{
"operation": "Highlighting",
"time": "28.414",
"value": "1892 United States presidential election in Illinois_sentence_3"
},
{
"operation": "Highlighting",
"time": "29.225",
"value": "1892 United States presidential election in Illinois_sentence_4"
},
{
"operation": "finish",
"time": "49.204",
"value": "finish"
}
]
}
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.