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_5160
|
How to Train Your Dragon is a 2011 American film loosely based on the 2003 book of the same name by Cressida Cowell. Cressida Cowell FRSL is a British children's author.
|
{
"evidence": [
"How to Train Your Dragon (film)",
"Cressida Cowell"
],
"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:\nHow to Train Your Dragon is a 2011 American film loosely based on the 2003 book of the same name by Cressida Cowell. Cressida Cowell FRSL is a British children's author.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"How to Train Your Dragon (film)",
"Cressida Cowell"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7284,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
53775
|
In 2010 Biotie Therapies attempted but ultimately failed to acquire Elbion GmbH, a German pharmaceutical company in Radebeul, a suburb of Dresden.
|
{
"evidence": [
"Biotie Therapies_sentence_4",
"Radebeul_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:\nIn 2010 Biotie Therapies attempted but ultimately failed to acquire Elbion GmbH, a German pharmaceutical company in Radebeul, a suburb of Dresden.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Biotie Therapies_sentence_4",
"Radebeul_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7285,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.759",
"value": "?search="
},
{
"operation": "search",
"time": "15.319",
"value": "Biotie Therapies"
},
{
"operation": "Now on",
"time": "17.002",
"value": "Biotie Therapies"
},
{
"operation": "Highlighting",
"time": "31.628",
"value": "Biotie Therapies_sentence_4"
},
{
"operation": "hyperlink",
"time": "69.996",
"value": "Radebeul"
},
{
"operation": "Now on",
"time": "71.736",
"value": "Radebeul"
},
{
"operation": "Highlighting",
"time": "73.524",
"value": "Radebeul_sentence_0"
},
{
"operation": "finish",
"time": "77.562",
"value": "finish"
}
]
}
|
exfever_4622
|
7 Things is a song by American singer Miley Cyrus and produced by John Fields, which was released on June 17, 2008, by an American record label of the Disney Music Group.
|
{
"evidence": [
"7 Things",
"Hollywood Records"
],
"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:\n7 Things is a song by American singer Miley Cyrus and produced by John Fields, which was released on June 17, 2008, by an American record label of the Disney Music Group.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"7 Things",
"Hollywood Records"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7286,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
85793
|
The Contribution of the Aviation Industry to the UK Economy was published in 1999, The Economic Contribution of the Aviation Industry in the UK was published in October 2006, and in December 2006 the government published the Air Transport White Paper Progress Report 2006 to report on progress made in "… delivering a sustainable future for aviation."
|
{
"evidence": [
"Air transport in the United Kingdom_sentence_57"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Contribution of the Aviation Industry to the UK Economy was published in 1999, The Economic Contribution of the Aviation Industry in the UK was published in October 2006, and in December 2006 the government published the Air Transport White Paper Progress Report 2006 to report on progress made in \"… delivering a sustainable future for aviation.\"\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Air transport in the United Kingdom_sentence_57"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7287,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.323",
"value": "?search="
},
{
"operation": "search",
"time": "421.337",
"value": "Contribution of the Aviation Industry to the UK Economy"
},
{
"operation": "Now on",
"time": "423.933",
"value": "undefined"
},
{
"operation": "search",
"time": "489.513",
"value": "The Economic Contribution of the Aviation Industry in the UK"
},
{
"operation": "Now on",
"time": "491.003",
"value": "undefined"
},
{
"operation": "search",
"time": "546.572",
"value": "Air Transport White Paper Progress Report 2006"
},
{
"operation": "Now on",
"time": "548.59",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "563.333",
"value": "Air transport in the United Kingdom"
},
{
"operation": "Now on",
"time": "564.175",
"value": "Air transport in the United Kingdom"
},
{
"operation": "Page search",
"time": "586.908",
"value": "2006"
},
{
"operation": "Highlighting",
"time": "637.184",
"value": "Air transport in the United Kingdom_sentence_57"
},
{
"operation": "Page search",
"time": "739.213",
"value": "economic"
},
{
"operation": "search",
"time": "818.438",
"value": "The Economic Contribution of the Aviation Industry in the UK"
},
{
"operation": "Now on",
"time": "819.686",
"value": "undefined"
},
{
"operation": "finish",
"time": "1012.295",
"value": "finish"
}
]
}
|
exfever_8737
|
Reg Watson was an American television producer and screenwriter who created an Australian television soap opera produced by the Grundy Organisation, which was conceived by an Australian television producer and screenwriter.
|
{
"evidence": [
"Reg Watson",
"Prisoner (TV series)",
"Reg Watson"
],
"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:\nReg Watson was an American television producer and screenwriter who created an Australian television soap opera produced by the Grundy Organisation, which was conceived by an Australian television producer and screenwriter.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Reg Watson",
"Prisoner (TV series)",
"Reg Watson"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7288,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
42228
|
After overcoming his alcoholic addiction, Eric Clapton (born 30 March 1945) realized that he had wasted five years of his life, and later on he liked the recordings.
|
{
"evidence": [
"Eric Clapton_sentence_0",
"Eric Clapton_sentence_18",
"Eric Clapton_sentence_170",
"Eric Clapton_sentence_177",
"Eric Clapton_sentence_183",
"Eric Clapton_sentence_198",
"Eric Clapton_sentence_199",
"Eric Clapton_sentence_202",
"Eric Clapton_sentence_203",
"Eric Clapton_sentence_204",
"Eric Clapton_sentence_211",
"Eric Clapton_sentence_422",
"Eric Clapton_sentence_456",
"Eric Clapton_sentence_499",
"Eric Clapton_sentence_502"
],
"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:\nAfter overcoming his alcoholic addiction, Eric Clapton (born 30 March 1945) realized that he had wasted five years of his life, and later on he liked the recordings.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Eric Clapton_sentence_0",
"Eric Clapton_sentence_18",
"Eric Clapton_sentence_170",
"Eric Clapton_sentence_177",
"Eric Clapton_sentence_183",
"Eric Clapton_sentence_198",
"Eric Clapton_sentence_199",
"Eric Clapton_sentence_202",
"Eric Clapton_sentence_203",
"Eric Clapton_sentence_204",
"Eric Clapton_sentence_211",
"Eric Clapton_sentence_422",
"Eric Clapton_sentence_456",
"Eric Clapton_sentence_499",
"Eric Clapton_sentence_502"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7289,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "5.569",
"value": "?search="
},
{
"operation": "search",
"time": "14.836",
"value": "Eric Clapton and the Powerhouse"
},
{
"operation": "Now on",
"time": "19.278",
"value": "Eric Clapton and the Powerhouse"
},
{
"operation": "search",
"time": "25.079",
"value": "Eric Clapton"
},
{
"operation": "Now on",
"time": "34.675",
"value": "Eric Clapton"
},
{
"operation": "Highlighting",
"time": "38.021",
"value": "Eric Clapton_sentence_0"
},
{
"operation": "Page search",
"time": "63.376",
"value": "addiction"
},
{
"operation": "Page search",
"time": "259.988",
"value": "waste"
},
{
"operation": "Highlighting",
"time": "307.545",
"value": "Eric Clapton_sentence_457"
},
{
"operation": "Highlighting deleted",
"time": "326.336",
"value": "Eric Clapton_sentence_457"
},
{
"operation": "Page search",
"time": "366.398",
"value": "alcohol"
},
{
"operation": "Highlighting",
"time": "373.46",
"value": "Eric Clapton_sentence_18"
},
{
"operation": "Highlighting",
"time": "404.451",
"value": "Eric Clapton_sentence_170"
},
{
"operation": "Highlighting",
"time": "418.97",
"value": "Eric Clapton_sentence_177"
},
{
"operation": "Highlighting",
"time": "433.019",
"value": "Eric Clapton_sentence_183"
},
{
"operation": "Highlighting",
"time": "450.618",
"value": "Eric Clapton_sentence_198"
},
{
"operation": "Highlighting",
"time": "453.426",
"value": "Eric Clapton_sentence_199"
},
{
"operation": "Highlighting",
"time": "460.714",
"value": "Eric Clapton_sentence_202"
},
{
"operation": "Highlighting",
"time": "464.636",
"value": "Eric Clapton_sentence_203"
},
{
"operation": "Highlighting",
"time": "467.496",
"value": "Eric Clapton_sentence_204"
},
{
"operation": "Highlighting",
"time": "486.832",
"value": "Eric Clapton_sentence_211"
},
{
"operation": "Highlighting",
"time": "628.163",
"value": "Eric Clapton_sentence_422"
},
{
"operation": "Highlighting",
"time": "645.77",
"value": "Eric Clapton_sentence_456"
},
{
"operation": "Highlighting",
"time": "660.106",
"value": "Eric Clapton_sentence_499"
},
{
"operation": "Highlighting",
"time": "662.848",
"value": "Eric Clapton_sentence_502"
},
{
"operation": "finish",
"time": "677.735",
"value": "finish"
}
]
}
|
93542
|
Testeagles were formed as a techno rock trio in Adelaide (the capital city of the state of South Australia) in 1994, and were signed by Stuart MacQueen of Adelaide-based independent label, Krell Records.
|
{
"evidence": [
"Testeagles_sentence_9",
"Testeagles_sentence_12",
"Adelaide_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:\nTesteagles were formed as a techno rock trio in Adelaide (the capital city of the state of South Australia) in 1994, and were signed by Stuart MacQueen of Adelaide-based independent label, Krell Records.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Testeagles_sentence_9",
"Testeagles_sentence_12",
"Adelaide_sentence_3"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7290,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.91",
"value": "?search="
},
{
"operation": "search",
"time": "15.218",
"value": "Testeagles"
},
{
"operation": "Now on",
"time": "15.916",
"value": "Testeagles"
},
{
"operation": "Highlighting",
"time": "33.53",
"value": "Testeagles_sentence_9"
},
{
"operation": "Highlighting",
"time": "51.849",
"value": "Testeagles_sentence_12"
},
{
"operation": "hyperlink",
"time": "55.959",
"value": "Adelaide"
},
{
"operation": "Now on",
"time": "57.888",
"value": "Adelaide"
},
{
"operation": "Highlighting",
"time": "65.172",
"value": "Adelaide_sentence_3"
},
{
"operation": "finish",
"time": "75.745",
"value": "finish"
}
]
}
|
68760
|
Mason Pearson, who was called by a Yorkshireman in the mid 1860s to come to work at the British Steam Brush Works, in the East End of London, invented the "pneumatic" rubber-cushion hairbrush.
|
{
"evidence": [
"Mason Pearson Brushes_sentence_1",
"Mason Pearson Brushes_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:\nMason Pearson, who was called by a Yorkshireman in the mid 1860s to come to work at the British Steam Brush Works, in the East End of London, invented the \"pneumatic\" rubber-cushion hairbrush.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Mason Pearson Brushes_sentence_1",
"Mason Pearson Brushes_sentence_2"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7291,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.555",
"value": "?search="
},
{
"operation": "search",
"time": "12.69",
"value": "Contains...Mason Pearson"
},
{
"operation": "Now on",
"time": "14.04",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "31.436",
"value": "Mason Pearson Brushes"
},
{
"operation": "Now on",
"time": "31.98",
"value": "Mason Pearson Brushes"
},
{
"operation": "Highlighting",
"time": "51.651",
"value": "Mason Pearson Brushes_sentence_1"
},
{
"operation": "Highlighting",
"time": "59.683",
"value": "Mason Pearson Brushes_sentence_2"
},
{
"operation": "finish",
"time": "74.907",
"value": "finish"
}
]
}
|
58763
|
Balach Marri, born on January 17, 1966, was the leader of the Baloch Liberation Army (listed as a terrorist organization by Pakistan, United Kingdom and the United States) who fought in the Insurgency in Balochistan.
|
{
"evidence": [
"Balach Marri_sentence_11",
"Balach Marri_sentence_0",
"Separatist movements of Pakistan_sentence_24"
],
"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:\nBalach Marri, born on January 17, 1966, was the leader of the Baloch Liberation Army (listed as a terrorist organization by Pakistan, United Kingdom and the United States) who fought in the Insurgency in Balochistan.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Balach Marri_sentence_11",
"Balach Marri_sentence_0",
"Separatist movements of Pakistan_sentence_24"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7292,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0",
"value": "?search="
},
{
"operation": "search",
"time": "0.16",
"value": "Balach Marri"
},
{
"operation": "Now on",
"time": "0.161",
"value": "Balach Marri"
},
{
"operation": "Highlighting",
"time": "0.168",
"value": "Balach Marri_sentence_0"
},
{
"operation": "Highlighting deleted",
"time": "0.17",
"value": "Balach Marri_sentence_0"
},
{
"operation": "Highlighting",
"time": "0.172",
"value": "Balach Marri_header_cell_0_3_0"
},
{
"operation": "Highlighting",
"time": "0.173",
"value": "Balach Marri_cell_0_3_1"
},
{
"operation": "Highlighting",
"time": "0.173",
"value": "Balach Marri_header_cell_0_3_0"
},
{
"operation": "Highlighting",
"time": "0.173",
"value": "Balach Marri_header_cell_0_0_0"
},
{
"operation": "Highlighting deleted",
"time": "0.19",
"value": "Balach Marri_cell_0_3_1"
},
{
"operation": "Highlighting deleted",
"time": "0.19",
"value": "Balach Marri_header_cell_0_3_0"
},
{
"operation": "Highlighting deleted",
"time": "0.19",
"value": "Balach Marri_header_cell_0_0_0"
},
{
"operation": "Highlighting",
"time": "0.194",
"value": "Balach Marri_sentence_11"
},
{
"operation": "Page search",
"time": "0.2",
"value": "unsurg"
},
{
"operation": "Highlighting",
"time": "0.216",
"value": "Balach Marri_sentence_0"
},
{
"operation": "search",
"time": "0.231",
"value": "Contains...Balochistan Liberation Army"
},
{
"operation": "Now on",
"time": "0.233",
"value": "undefined"
},
{
"operation": "search",
"time": "0.247",
"value": "Baloch Liberation Army"
},
{
"operation": "Now on",
"time": "0.247",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "0.263",
"value": "Separatist movements of Pakistan"
},
{
"operation": "Now on",
"time": "0.265",
"value": "Separatist movements of Pakistan"
},
{
"operation": "Page search",
"time": "0.272",
"value": "terrorist"
},
{
"operation": "Highlighting",
"time": "0.279",
"value": "Separatist movements of Pakistan_sentence_24"
},
{
"operation": "Page search",
"time": "0.294",
"value": "united"
},
{
"operation": "Page search",
"time": "0.314",
"value": "us"
},
{
"operation": "Page search",
"time": "0.324",
"value": "terrorist"
},
{
"operation": "Page search",
"time": "0.324",
"value": "terrorist"
},
{
"operation": "finish",
"time": "0.34",
"value": "finish"
},
{
"operation": "Now on",
"time": "1621444561.45",
"value": "Balach Marri"
},
{
"operation": "finish",
"time": "1621444563.662",
"value": "finish"
}
]
}
|
70980
|
George Akerlof's idea in the late 1990s has attracted the attention of some on both sides of the debate over legal abortion, which he has described as a phenomenon labeled "reproductive technology shock."
|
{
"evidence": [
"George Akerlof_sentence_14",
"George Akerlof_sentence_15"
],
"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:\nGeorge Akerlof's idea in the late 1990s has attracted the attention of some on both sides of the debate over legal abortion, which he has described as a phenomenon labeled \"reproductive technology shock.\"\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"George Akerlof_sentence_14",
"George Akerlof_sentence_15"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7293,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.889",
"value": "?search="
},
{
"operation": "search",
"time": "6.543",
"value": "George Akerlof"
},
{
"operation": "Now on",
"time": "7.559",
"value": "George Akerlof"
},
{
"operation": "Highlighting",
"time": "20.953",
"value": "George Akerlof_sentence_14"
},
{
"operation": "Highlighting",
"time": "28.78",
"value": "George Akerlof_sentence_15"
},
{
"operation": "finish",
"time": "33.265",
"value": "finish"
}
]
}
|
exfever_3635
|
The ketogenic diet is adequate-protein, low-carbohydrate, high in ester of fatty acids based on the scientific study of the properties and behavior of matter.
|
{
"evidence": [
"Ketogenic diet",
"Fat",
"Chemistry"
],
"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 ketogenic diet is adequate-protein, low-carbohydrate, high in ester of fatty acids based on the scientific study of the properties and behavior of matter.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Ketogenic diet",
"Fat",
"Chemistry"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7294,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_5053
|
Lisa Marie Presley is an American singer and songwriter who has been married to an American actor and filmmaker who received various awards including an award bestowed by the 105 members of the Hollywood Foreign Press Association beginning in January 1941.
|
{
"evidence": [
"Lisa Marie Presley",
"Nicolas Cage",
"Golden Globe Award"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nLisa Marie Presley is an American singer and songwriter who has been married to an American actor and filmmaker who received various awards including an award bestowed by the 105 members of the Hollywood Foreign Press Association beginning in January 1941.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Lisa Marie Presley",
"Nicolas Cage",
"Golden Globe Award"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7295,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
85443
|
Spec Keene returned to Oregon State to serve as athletic director, he served in that post for 26 years and was President of the Pacific Coast Conference Athletics Directors Association.
|
{
"evidence": [
"Spec Keene_sentence_13",
"Spec Keene_sentence_14",
"Spec Keene_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:\nSpec Keene returned to Oregon State to serve as athletic director, he served in that post for 26 years and was President of the Pacific Coast Conference Athletics Directors Association.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Spec Keene_sentence_13",
"Spec Keene_sentence_14",
"Spec Keene_sentence_16"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7296,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.27",
"value": "?search="
},
{
"operation": "search",
"time": "13.182",
"value": "Spec Keene"
},
{
"operation": "Now on",
"time": "14.133",
"value": "Spec Keene"
},
{
"operation": "Highlighting",
"time": "26.248",
"value": "Spec Keene_sentence_13"
},
{
"operation": "Highlighting",
"time": "30.8",
"value": "Spec Keene_sentence_14"
},
{
"operation": "Highlighting",
"time": "37.828",
"value": "Spec Keene_sentence_16"
},
{
"operation": "finish",
"time": "49.751",
"value": "finish"
}
]
}
|
exfever_6198
|
Little Dorrit is a novel by Charles Dickens whose genre of the visual, literary, and performing arts in which vices, follies, abuses, and shortcomings of both government and society are held up to ridicule.
|
{
"evidence": [
"Little Dorrit",
"Satire"
],
"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:\nLittle Dorrit is a novel by Charles Dickens whose genre of the visual, literary, and performing arts in which vices, follies, abuses, and shortcomings of both government and society are held up to ridicule.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Little Dorrit",
"Satire"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7297,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
12360
|
Black Cat John Brown is the second album by Alamo Race Track which received positive ratings from De Volkskrant, Het Parool and Allmusic.
|
{
"evidence": [
"Black Cat John Brown_sentence_0",
"Black Cat John Brown_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:\nBlack Cat John Brown is the second album by Alamo Race Track which received positive ratings from De Volkskrant, Het Parool and Allmusic.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Black Cat John Brown_sentence_0",
"Black Cat John Brown_sentence_4"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7298,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.639",
"value": "?search="
},
{
"operation": "search",
"time": "8.098",
"value": "Black Cat John Brown"
},
{
"operation": "Now on",
"time": "8.812",
"value": "Black Cat John Brown"
},
{
"operation": "Highlighting",
"time": "19.605",
"value": "Black Cat John Brown_sentence_0"
},
{
"operation": "Highlighting",
"time": "47.814",
"value": "Black Cat John Brown_sentence_4"
},
{
"operation": "hyperlink",
"time": "58.507",
"value": "De Volkskrant"
},
{
"operation": "Now on",
"time": "59.775",
"value": "De Volkskrant"
},
{
"operation": "back-button-clicked",
"time": "63.29",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "63.398",
"value": "Black Cat John Brown"
},
{
"operation": "hyperlink",
"time": "64.626",
"value": "Het Parool"
},
{
"operation": "Now on",
"time": "65.219",
"value": "Het Parool"
},
{
"operation": "back-button-clicked",
"time": "68.856",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "68.976",
"value": "Black Cat John Brown"
},
{
"operation": "hyperlink",
"time": "70.422",
"value": "Allmusic"
},
{
"operation": "Now on",
"time": "71.672",
"value": "Allmusic"
},
{
"operation": "back-button-clicked",
"time": "74.362",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "74.466",
"value": "Black Cat John Brown"
},
{
"operation": "finish",
"time": "93.47",
"value": "finish"
}
]
}
|
exfever_7995
|
Joshua Michael Homme is an American musician who was formerly the co-founder and guitarist of an American rock band, formed in Palm Desert, California, in 1990.
|
{
"evidence": [
"Josh Homme",
"Kyuss"
],
"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:\nJoshua Michael Homme is an American musician who was formerly the co-founder and guitarist of an American rock band, formed in Palm Desert, California, in 1990.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Josh Homme",
"Kyuss"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7299,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_729
|
Other major Silicon Valley cities include a city located in the Santa Clara Valley in northwest Santa Clara County.
|
{
"evidence": [
"Silicon Valley",
"Sunnyvale, 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:\nOther major Silicon Valley cities include a city located in the Santa Clara Valley in northwest Santa Clara County.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Silicon Valley",
"Sunnyvale, California"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7300,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
47298
|
Gengo Al was launched in March 2018 before being acquired by Lionbridge Technologies in January 2020.
|
{
"evidence": [
"Gengo_sentence_11"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nGengo Al was launched in March 2018 before being acquired by Lionbridge Technologies in January 2020.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Gengo_sentence_11"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7301,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.575",
"value": "?search="
},
{
"operation": "search",
"time": "9.767",
"value": "Contains...Gengo Al"
},
{
"operation": "Now on",
"time": "10.305",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "106.424",
"value": "Lionbridge"
},
{
"operation": "Now on",
"time": "107.146",
"value": "Lionbridge"
},
{
"operation": "Page search",
"time": "111.314",
"value": "gengo"
},
{
"operation": "search",
"time": "125.686",
"value": "Gengo"
},
{
"operation": "Now on",
"time": "126.593",
"value": "Gengo"
},
{
"operation": "Highlighting",
"time": "312.594",
"value": "Gengo_sentence_2"
},
{
"operation": "Highlighting",
"time": "331.289",
"value": "Gengo_sentence_11"
},
{
"operation": "Highlighting deleted",
"time": "340.706",
"value": "Gengo_sentence_2"
},
{
"operation": "Highlighting deleted",
"time": "343.574",
"value": "Gengo_sentence_11"
},
{
"operation": "Highlighting",
"time": "344.286",
"value": "Gengo_sentence_10"
},
{
"operation": "Highlighting",
"time": "345.881",
"value": "Gengo_sentence_11"
},
{
"operation": "Highlighting deleted",
"time": "348.906",
"value": "Gengo_sentence_10"
},
{
"operation": "finish",
"time": "361.449",
"value": "finish"
}
]
}
|
exfever_7899
|
RB Leipzig was founded by Red Bull GmbH, which purchased the playing rights of fifth-tier side SSV Markranst?dt with the intent of advancing to a professional league of a team sport played between two teams of 11 players.
|
{
"evidence": [
"RB Leipzig",
"Bundesliga",
"Association football"
],
"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:\nRB Leipzig was founded by Red Bull GmbH, which purchased the playing rights of fifth-tier side SSV Markranst?dt with the intent of advancing to a professional league of a team sport played between two teams of 11 players.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"RB Leipzig",
"Bundesliga",
"Association football"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7302,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_916
|
Mark Alan Ruffalo is an American actor who appeared in a 2000 American drama film written and directed by an American film director who co-wrote the film Gangs of New York.
|
{
"evidence": [
"Mark Ruffalo",
"You Can Count On Me",
"Kenneth Lonergan"
],
"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:\nMark Alan Ruffalo is an American actor who appeared in a 2000 American drama film written and directed by an American film director who co-wrote the film Gangs of New York.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Mark Ruffalo",
"You Can Count On Me",
"Kenneth Lonergan"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7303,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_1253
|
Kirk Douglas’s first American film was "Champion" in 1949.
|
{
"evidence": [
"Champion (1949 film)",
"Drama (film and television)"
],
"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:\nKirk Douglas’s first American film was \"Champion\" in 1949.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Champion (1949 film)",
"Drama (film and television)"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7304,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7531
|
Red is directed by a German screenwriter and director who controls a film's artistic and dramatic aspects and visualizes the screenplay while guiding the film crew and actors in the fulfillment of that vision.
|
{
"evidence": [
"Red (2010 film)",
"Robert Schwentke",
"Film director"
],
"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:\nRed is directed by a German screenwriter and director who controls a film's artistic and dramatic aspects and visualizes the screenplay while guiding the film crew and actors in the fulfillment of that vision.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Red (2010 film)",
"Robert Schwentke",
"Film director"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7305,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_6253
|
Jessica Michelle Chastain is the most successful film producer in the United States.
|
{
"evidence": [
"Jessica Chastain",
"Juilliard School",
"New York City"
],
"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:\nJessica Michelle Chastain is the most successful film producer in the United States.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Jessica Chastain",
"Juilliard School",
"New York City"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7306,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_1776
|
The House of Lusignan was a royal house of English origin, which at various times ruled several principalities in Europe and the Levant, including the kingdoms of Jerusalem, and a kingdom comprised not only the island of Cyprus but also had a foothold on the Anatolian mainland.
|
{
"evidence": [
"House of Lusignan",
"Kingdom of Cyprus"
],
"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 House of Lusignan was a royal house of English origin, which at various times ruled several principalities in Europe and the Levant, including the kingdoms of Jerusalem, and a kingdom comprised not only the island of Cyprus but also had a foothold on the Anatolian mainland.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"House of Lusignan",
"Kingdom of Cyprus"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7307,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_6533
|
Stephenie Meyer wrote The Host because she wanted to continue the story of Bella Swan and Edward Cullen.
|
{
"evidence": [
"The Host (2013 film)",
"Stephenie Meyer"
],
"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:\nStephenie Meyer wrote The Host because she wanted to continue the story of Bella Swan and Edward Cullen.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Host (2013 film)",
"Stephenie Meyer"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7308,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7070
|
Peyton Williams Manning is an American former football quarterback who played in the professional American football league, which consists of 32 teams, divided equally between the American Football Conference and the National Football Conference.
|
{
"evidence": [
"Peyton Manning",
"National Football League"
],
"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:\nPeyton Williams Manning is an American former football quarterback who played in the professional American football league, which consists of 32 teams, divided equally between the American Football Conference and the National Football Conference.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Peyton Manning",
"National Football League"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7309,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7284
|
Cersei Lannister is a fictional character in the series of epic fantasy novels by the American novelist and screenwriter George R. R. Martin, which has a fourth of seven planned novels in the epic fantasy series, A Song of Ice and Fire.
|
{
"evidence": [
"Cersei Lannister",
"A Song of Ice and Fire",
"A Feast for Crows"
],
"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:\nCersei Lannister is a fictional character in the series of epic fantasy novels by the American novelist and screenwriter George R. R. Martin, which has a fourth of seven planned novels in the epic fantasy series, A Song of Ice and Fire.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Cersei Lannister",
"A Song of Ice and Fire",
"A Feast for Crows"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7310,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
51931
|
The dorsal sepal is egg-shaped, 9–13 mm long, 4–6.5 mm wide and curves upwards; while the sepals are oblong, 9–14 mm long, 2–3 mm wide and turned downwards.
|
{
"evidence": [
"Crepidium fimbriatum_sentence_7",
"Dendrobium bigibbum_sentence_8",
"Prasophyllum caricetum_sentence_7"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe dorsal sepal is egg-shaped, 9–13 mm long, 4–6.5 mm wide and curves upwards; while the sepals are oblong, 9–14 mm long, 2–3 mm wide and turned downwards.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Crepidium fimbriatum_sentence_7",
"Dendrobium bigibbum_sentence_8",
"Prasophyllum caricetum_sentence_7"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7311,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.715",
"value": "?search="
},
{
"operation": "search",
"time": "20.512",
"value": "dorsal sepal"
},
{
"operation": "Now on",
"time": "24.31",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "27.201",
"value": "dorsal segal"
},
{
"operation": "Now on",
"time": "27.738",
"value": "undefined"
},
{
"operation": "search",
"time": "286.167",
"value": "Crepidium fimbriatum"
},
{
"operation": "Now on",
"time": "287.36",
"value": "Crepidium fimbriatum"
},
{
"operation": "Highlighting",
"time": "595.945",
"value": "Crepidium fimbriatum_sentence_7"
},
{
"operation": "search",
"time": "1007.632",
"value": "sepals oblong"
},
{
"operation": "Now on",
"time": "1010.019",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "1014.572",
"value": "Dendrobium bigibbum"
},
{
"operation": "Now on",
"time": "1015.419",
"value": "Dendrobium bigibbum"
},
{
"operation": "Highlighting",
"time": "1031.016",
"value": "Dendrobium bigibbum_sentence_8"
},
{
"operation": "search",
"time": "1078.541",
"value": "Contains...wide and curves upwards"
},
{
"operation": "Now on",
"time": "1080.838",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "1084.194",
"value": "Taxonomy and naming Prasophyllum caricetum_section_1"
},
{
"operation": "Now on",
"time": "1085.18",
"value": "Prasophyllum caricetum#Taxonomy and naming Prasophyllum caricetum section 1"
},
{
"operation": "Highlighting",
"time": "1095.894",
"value": "Prasophyllum caricetum_sentence_7"
},
{
"operation": "finish",
"time": "1115.649",
"value": "finish"
}
]
}
|
42228
|
After overcoming his alcoholic addiction, Eric Clapton (born 30 March 1945) realized that he had wasted five years of his life, and later on he liked the recordings.
|
{
"evidence": [
"Eric Clapton_sentence_0",
"Eric Clapton_sentence_18",
"Eric Clapton_sentence_170",
"Eric Clapton_sentence_177",
"Eric Clapton_sentence_183",
"Eric Clapton_sentence_198",
"Eric Clapton_sentence_199",
"Eric Clapton_sentence_202",
"Eric Clapton_sentence_203",
"Eric Clapton_sentence_204",
"Eric Clapton_sentence_211",
"Eric Clapton_sentence_422",
"Eric Clapton_sentence_456",
"Eric Clapton_sentence_499",
"Eric Clapton_sentence_502"
],
"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:\nAfter overcoming his alcoholic addiction, Eric Clapton (born 30 March 1945) realized that he had wasted five years of his life, and later on he liked the recordings.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Eric Clapton_sentence_0",
"Eric Clapton_sentence_18",
"Eric Clapton_sentence_170",
"Eric Clapton_sentence_177",
"Eric Clapton_sentence_183",
"Eric Clapton_sentence_198",
"Eric Clapton_sentence_199",
"Eric Clapton_sentence_202",
"Eric Clapton_sentence_203",
"Eric Clapton_sentence_204",
"Eric Clapton_sentence_211",
"Eric Clapton_sentence_422",
"Eric Clapton_sentence_456",
"Eric Clapton_sentence_499",
"Eric Clapton_sentence_502"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7312,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "5.569",
"value": "?search="
},
{
"operation": "search",
"time": "14.836",
"value": "Eric Clapton and the Powerhouse"
},
{
"operation": "Now on",
"time": "19.278",
"value": "Eric Clapton and the Powerhouse"
},
{
"operation": "search",
"time": "25.079",
"value": "Eric Clapton"
},
{
"operation": "Now on",
"time": "34.675",
"value": "Eric Clapton"
},
{
"operation": "Highlighting",
"time": "38.021",
"value": "Eric Clapton_sentence_0"
},
{
"operation": "Page search",
"time": "63.376",
"value": "addiction"
},
{
"operation": "Page search",
"time": "259.988",
"value": "waste"
},
{
"operation": "Highlighting",
"time": "307.545",
"value": "Eric Clapton_sentence_457"
},
{
"operation": "Highlighting deleted",
"time": "326.336",
"value": "Eric Clapton_sentence_457"
},
{
"operation": "Page search",
"time": "366.398",
"value": "alcohol"
},
{
"operation": "Highlighting",
"time": "373.46",
"value": "Eric Clapton_sentence_18"
},
{
"operation": "Highlighting",
"time": "404.451",
"value": "Eric Clapton_sentence_170"
},
{
"operation": "Highlighting",
"time": "418.97",
"value": "Eric Clapton_sentence_177"
},
{
"operation": "Highlighting",
"time": "433.019",
"value": "Eric Clapton_sentence_183"
},
{
"operation": "Highlighting",
"time": "450.618",
"value": "Eric Clapton_sentence_198"
},
{
"operation": "Highlighting",
"time": "453.426",
"value": "Eric Clapton_sentence_199"
},
{
"operation": "Highlighting",
"time": "460.714",
"value": "Eric Clapton_sentence_202"
},
{
"operation": "Highlighting",
"time": "464.636",
"value": "Eric Clapton_sentence_203"
},
{
"operation": "Highlighting",
"time": "467.496",
"value": "Eric Clapton_sentence_204"
},
{
"operation": "Highlighting",
"time": "486.832",
"value": "Eric Clapton_sentence_211"
},
{
"operation": "Highlighting",
"time": "628.163",
"value": "Eric Clapton_sentence_422"
},
{
"operation": "Highlighting",
"time": "645.77",
"value": "Eric Clapton_sentence_456"
},
{
"operation": "Highlighting",
"time": "660.106",
"value": "Eric Clapton_sentence_499"
},
{
"operation": "Highlighting",
"time": "662.848",
"value": "Eric Clapton_sentence_502"
},
{
"operation": "finish",
"time": "677.735",
"value": "finish"
}
]
}
|
exfever_6828
|
The 2002 Royal Rumble was the 14th annual Royal Rumble professional wrestling event produced by the World Wrestling Federation that was held in the capital and most populous city of the U.S. state of Georgia.
|
{
"evidence": [
"Royal Rumble (2002)",
"Atlanta"
],
"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 2002 Royal Rumble was the 14th annual Royal Rumble professional wrestling event produced by the World Wrestling Federation that was held in the capital and most populous city of the U.S. state of Georgia.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Royal Rumble (2002)",
"Atlanta"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7313,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
65430
|
Hamdi Salihi moved for the first time outside joining Panionios (founded in 1890) and played five games before moving back.
|
{
"evidence": [
"Hamdi Salihi_sentence_10",
"Hamdi Salihi_sentence_11",
"Hamdi Salihi_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:\nHamdi Salihi moved for the first time outside joining Panionios (founded in 1890) and played five games before moving back.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Hamdi Salihi_sentence_10",
"Hamdi Salihi_sentence_11",
"Hamdi Salihi_sentence_12"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7314,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.925",
"value": "?search="
},
{
"operation": "search",
"time": "17.395",
"value": "Hamdi Salihi"
},
{
"operation": "Now on",
"time": "19.262",
"value": "Hamdi Salihi"
},
{
"operation": "Highlighting",
"time": "33.926",
"value": "Hamdi Salihi_sentence_10"
},
{
"operation": "Highlighting",
"time": "46.365",
"value": "Hamdi Salihi_sentence_11"
},
{
"operation": "Highlighting",
"time": "47.55",
"value": "Hamdi Salihi_sentence_12"
},
{
"operation": "finish",
"time": "59.372",
"value": "finish"
}
]
}
|
exfever_793
|
Amsterdam is located within the Dutch province of the Netherlands in the northwestern part of the country.
|
{
"evidence": [
"Amsterdam",
"North Holland"
],
"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:\nAmsterdam is located within the Dutch province of the Netherlands in the northwestern part of the country.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Amsterdam",
"North Holland"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7315,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
40469
|
ADAM12 gene encodes a member of the a disintegrin and metalloprotease protein family, are membrane-anchored proteins and have two alternatively spliced transcripts.
|
{
"evidence": [
"ADAM12_sentence_4",
"ADAM12_sentence_5",
"ADAM12_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:\nADAM12 gene encodes a member of the a disintegrin and metalloprotease protein family, are membrane-anchored proteins and have two alternatively spliced transcripts.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"ADAM12_sentence_4",
"ADAM12_sentence_5",
"ADAM12_sentence_3"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7316,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.877",
"value": "?search="
},
{
"operation": "search",
"time": "15.507",
"value": "ADAM12"
},
{
"operation": "Now on",
"time": "16.444",
"value": "ADAM12"
},
{
"operation": "Highlighting",
"time": "41.959",
"value": "ADAM12_sentence_4"
},
{
"operation": "Highlighting",
"time": "42.542",
"value": "ADAM12_sentence_5"
},
{
"operation": "Highlighting",
"time": "48.108",
"value": "ADAM12_sentence_3"
},
{
"operation": "finish",
"time": "89.607",
"value": "finish"
}
]
}
|
61056
|
During WWII, at least 40,000 Chinese laborers were forcibly brought to Japan from China to make up for a nationwide labor shortage, and there were about 9,000 Chinese laborers forced to work in mines in Nagasaki.
|
{
"evidence": [
"Monument to Commemorate Chinese Victims of the Atomic Bombing_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:\nDuring WWII, at least 40,000 Chinese laborers were forcibly brought to Japan from China to make up for a nationwide labor shortage, and there were about 9,000 Chinese laborers forced to work in mines in Nagasaki.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Monument to Commemorate Chinese Victims of the Atomic Bombing_sentence_2"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7317,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.236",
"value": "?search="
},
{
"operation": "search",
"time": "60.794",
"value": "forced labor wwii"
},
{
"operation": "Now on",
"time": "62.871",
"value": "undefined"
},
{
"operation": "search",
"time": "88.654",
"value": "forced labor japan"
},
{
"operation": "Now on",
"time": "90.616",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "96.614",
"value": "Forced labor Japanese war crimes_section_16"
},
{
"operation": "Now on",
"time": "98.393",
"value": "Japanese war crimes#Forced labor Japanese war crimes section 16"
},
{
"operation": "Highlighting",
"time": "112.719",
"value": "Japanese war crimes_sentence_196"
},
{
"operation": "Highlighting",
"time": "112.72",
"value": "Japanese war crimes_title"
},
{
"operation": "Highlighting",
"time": "112.723",
"value": "Japanese war crimes_section_6"
},
{
"operation": "Highlighting",
"time": "112.723",
"value": "Japanese war crimes_section_16"
},
{
"operation": "hyperlink",
"time": "213.611",
"value": "Slavery in Japan"
},
{
"operation": "Now on",
"time": "214.566",
"value": "Slavery in Japan"
},
{
"operation": "search",
"time": "277.729",
"value": "Monument to Commemorate Chinese Victims of the Atomic Bombing"
},
{
"operation": "Now on",
"time": "279.023",
"value": "Monument to Commemorate Chinese Victims of the Atomic Bombing"
},
{
"operation": "Highlighting",
"time": "287.183",
"value": "Monument to Commemorate Chinese Victims of the Atomic Bombing_sentence_1"
},
{
"operation": "Highlighting",
"time": "296.686",
"value": "Monument to Commemorate Chinese Victims of the Atomic Bombing_sentence_2"
},
{
"operation": "Highlighting deleted",
"time": "298.239",
"value": "Monument to Commemorate Chinese Victims of the Atomic Bombing_sentence_1"
},
{
"operation": "finish",
"time": "335.728",
"value": "finish"
}
]
}
|
exfever_4937
|
World War II was the biggest war in the world.
|
{
"evidence": [
"Adolf Hitler",
"Dictator"
],
"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:\nWorld War II was the biggest war in the world.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Adolf Hitler",
"Dictator"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7318,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_3040
|
School 2013 is a 2012 South Korean television series that uses a traditional writing system that consists of Chinese characters incorporated into the native language for about 80 million people, mostly of Korean descent, with Korean pronunciation.
|
{
"evidence": [
"School 2013",
"Hanja",
"Korean language"
],
"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:\nSchool 2013 is a 2012 South Korean television series that uses a traditional writing system that consists of Chinese characters incorporated into the native language for about 80 million people, mostly of Korean descent, with Korean pronunciation.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"School 2013",
"Hanja",
"Korean language"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7319,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
10375
|
The 1996 National Football League season was the Seattle Seahawks' 21st season to participate, and then in 1997, it was sold to Paul Allen, the co-founder of Microsoft.
|
{
"evidence": [
"1996 Seattle Seahawks season_sentence_0",
"1996 Seattle Seahawks season_sentence_3",
"Paul Allen_sentence_2"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe 1996 National Football League season was the Seattle Seahawks' 21st season to participate, and then in 1997, it was sold to Paul Allen, the co-founder of Microsoft.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"1996 Seattle Seahawks season_sentence_0",
"1996 Seattle Seahawks season_sentence_3",
"Paul Allen_sentence_2"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7320,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.516",
"value": "?search="
},
{
"operation": "search",
"time": "15.65",
"value": "1996 National Football League season"
},
{
"operation": "Now on",
"time": "16.77",
"value": "1996 NFL season"
},
{
"operation": "search",
"time": "86.448",
"value": "1996 Seattle Seahawks season"
},
{
"operation": "Now on",
"time": "87.345",
"value": "1996 Seattle Seahawks season"
},
{
"operation": "Highlighting",
"time": "89.216",
"value": "1996 Seattle Seahawks season_sentence_0"
},
{
"operation": "Highlighting",
"time": "121.445",
"value": "1996 Seattle Seahawks season_sentence_3"
},
{
"operation": "hyperlink",
"time": "125.193",
"value": "Paul Allen"
},
{
"operation": "Now on",
"time": "126.09",
"value": "Paul Allen"
},
{
"operation": "Highlighting",
"time": "130.9",
"value": "Paul Allen_sentence_2"
},
{
"operation": "finish",
"time": "159.123",
"value": "finish"
}
]
}
|
exfever_4555
|
Legion was a fictional character that was the son of Bruce Lee, who was referred to as the human subspecies, Homo sapiens superior.
|
{
"evidence": [
"Legion (Marvel Comics)",
"Mutant (Marvel Comics)"
],
"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:\nLegion was a fictional character that was the son of Bruce Lee, who was referred to as the human subspecies, Homo sapiens superior.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Legion (Marvel Comics)",
"Mutant (Marvel Comics)"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7321,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
44718
|
Peter Andrew Jones (born 14 December 1951) is a British artist and illustrator of fantasy and science fiction genre who has worked on book jacket covers, film posters, advertising, and games.
|
{
"evidence": [
"Peter Andrew Jones_sentence_0",
"Peter Andrew Jones_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:\nPeter Andrew Jones (born 14 December 1951) is a British artist and illustrator of fantasy and science fiction genre who has worked on book jacket covers, film posters, advertising, and games.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Peter Andrew Jones_sentence_0",
"Peter Andrew Jones_sentence_1"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7322,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.75",
"value": "?search="
},
{
"operation": "search",
"time": "14.88",
"value": "Peter Andrew Jones"
},
{
"operation": "Now on",
"time": "15.533",
"value": "Peter Andrew Jones"
},
{
"operation": "Highlighting",
"time": "19.206",
"value": "Peter Andrew Jones_sentence_0"
},
{
"operation": "Highlighting",
"time": "32.005",
"value": "Peter Andrew Jones_sentence_1"
},
{
"operation": "finish",
"time": "106.455",
"value": "finish"
}
]
}
|
exfever_7460
|
The RMS Titanic was an American passenger liner designed by a shipyard that works primarily in ship repair, located in the capital and largest city of Northern Ireland.
|
{
"evidence": [
"RMS Titanic",
"Harland and Wolff",
"Belfast"
],
"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 RMS Titanic was an American passenger liner designed by a shipyard that works primarily in ship repair, located in the capital and largest city of Northern Ireland.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"RMS Titanic",
"Harland and Wolff",
"Belfast"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7323,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
58180
|
Substitution may be used to integrate functions of several variables and to express the fact that the absolute value, or the non-negative value of x, may be equal to the volume.
|
{
"evidence": [
"Absolute value_sentence_1",
"Integration by substitution_sentence_34",
"Integration by substitution_sentence_37"
],
"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:\nSubstitution may be used to integrate functions of several variables and to express the fact that the absolute value, or the non-negative value of x, may be equal to the volume.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Absolute value_sentence_1",
"Integration by substitution_sentence_34",
"Integration by substitution_sentence_37"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7324,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.841",
"value": "?search="
},
{
"operation": "search",
"time": "17.897",
"value": "Absolute value"
},
{
"operation": "Now on",
"time": "19.19",
"value": "Absolute value"
},
{
"operation": "Page search",
"time": "26.423",
"value": "substitution"
},
{
"operation": "Highlighting",
"time": "49.057",
"value": "Absolute value_sentence_1"
},
{
"operation": "Page search",
"time": "112.874",
"value": "volume"
},
{
"operation": "search",
"time": "133.867",
"value": "substitution"
},
{
"operation": "Now on",
"time": "135.204",
"value": "Substitution"
},
{
"operation": "hyperlink",
"time": "141.34",
"value": "2.2 Mathematics and computing Substitution_section_3"
},
{
"operation": "hyperlink",
"time": "150.599",
"value": "Integration by substitution"
},
{
"operation": "Now on",
"time": "151.327",
"value": "Integration by substitution"
},
{
"operation": "Page search",
"time": "159.527",
"value": "absolute"
},
{
"operation": "Highlighting",
"time": "171.075",
"value": "Integration by substitution_sentence_34"
},
{
"operation": "Highlighting",
"time": "176.816",
"value": "Integration by substitution_sentence_37"
},
{
"operation": "finish",
"time": "227.782",
"value": "finish"
}
]
}
|
70065
|
Catherine Ibarguen scored twelve points for the first time in the Diamond race, which put her in a tie with one other competitor.
|
{
"evidence": [
"Caterine Ibargüen_sentence_10"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nCatherine Ibarguen scored twelve points for the first time in the Diamond race, which put her in a tie with one other competitor.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Caterine Ibargüen_sentence_10"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7325,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "49.714",
"value": "undefined"
},
{
"operation": "search",
"time": "55.003",
"value": "Catherine Ibarguen"
},
{
"operation": "Now on",
"time": "57.372",
"value": "Catherine Ibarguen"
},
{
"operation": "Page search",
"time": "92.687",
"value": "tie"
},
{
"operation": "Page search",
"time": "93.9",
"value": "tie"
},
{
"operation": "Page search",
"time": "120.451",
"value": "points"
},
{
"operation": "hyperlink",
"time": "136.197",
"value": "armed conflict in Colombia"
},
{
"operation": "Now on",
"time": "138.668",
"value": "Colombian conflict (1964–present)"
},
{
"operation": "Page search",
"time": "149.995",
"value": "Catherine"
},
{
"operation": "back-button-clicked",
"time": "162.255",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "163.605",
"value": "Caterine Ibargüen"
},
{
"operation": "Page search",
"time": "185.012",
"value": "Diamond"
},
{
"operation": "Highlighting",
"time": "221.698",
"value": "Caterine Ibargüen_sentence_10"
},
{
"operation": "hyperlink",
"time": "227.215",
"value": "Diamond League"
},
{
"operation": "Now on",
"time": "233.895",
"value": "Caterine Ibargüen"
},
{
"operation": "finish",
"time": "242.945",
"value": "finish"
}
]
}
|
58645
|
The Yoshizawa Randlett system was never mentioned in a book.
|
{
"evidence": [
"Yoshizawa–Randlett system_sentence_13"
],
"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 Yoshizawa Randlett system was never mentioned in a book.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Yoshizawa–Randlett system_sentence_13"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7326,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.851",
"value": "?search="
},
{
"operation": "search",
"time": "13.613",
"value": "Contains... Yoshizawa Randlett system"
},
{
"operation": "Now on",
"time": "14.761",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "17.909",
"value": "Yoshizawa–Randlett system"
},
{
"operation": "Now on",
"time": "18.826",
"value": "Yoshizawa–Randlett system"
},
{
"operation": "Highlighting",
"time": "87.071",
"value": "Yoshizawa–Randlett system_sentence_13"
},
{
"operation": "search",
"time": "109.199",
"value": "samuel randlett"
},
{
"operation": "Now on",
"time": "110.626",
"value": "undefined"
},
{
"operation": "back-button-clicked",
"time": "127.255",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "127.39",
"value": "Yoshizawa–Randlett system"
},
{
"operation": "finish",
"time": "151.008",
"value": "finish"
}
]
}
|
76633
|
Ingelfingen is located in Hohenlohekreis district, Baden-Württemberg state in France with a total area of 66.48 km.
|
{
"evidence": [
"Ingelfingen_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:\nIngelfingen is located in Hohenlohekreis district, Baden-Württemberg state in France with a total area of 66.48 km.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Ingelfingen_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7327,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.957",
"value": "?search="
},
{
"operation": "search",
"time": "10.823",
"value": "Ingelfingen"
},
{
"operation": "Now on",
"time": "14.758",
"value": "Ingelfingen"
},
{
"operation": "Highlighting",
"time": "18.403",
"value": "Ingelfingen_sentence_0"
},
{
"operation": "finish",
"time": "28.862",
"value": "finish"
}
]
}
|
70610
|
China started a bandy development programme by organising educational days in Ürümqi in June 2009, but they did not come as planned to the 2011 Asian Winter Games.
|
{
"evidence": [
"Ürümqi_sentence_154"
],
"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:\nChina started a bandy development programme by organising educational days in Ürümqi in June 2009, but they did not come as planned to the 2011 Asian Winter Games.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Ürümqi_sentence_154"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7328,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.74",
"value": "?search="
},
{
"operation": "search",
"time": "20.126",
"value": "bandy development programme"
},
{
"operation": "Now on",
"time": "22.486",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "26.654",
"value": "Ürümqi"
},
{
"operation": "Now on",
"time": "27.719",
"value": "Ürümqi"
},
{
"operation": "Page search",
"time": "52.026",
"value": "bandy development"
},
{
"operation": "Highlighting",
"time": "63.425",
"value": "Ürümqi_sentence_154"
},
{
"operation": "Page search",
"time": "95.472",
"value": "2011 asian winter"
},
{
"operation": "search",
"time": "117.312",
"value": "2011 Asian Winter Games"
},
{
"operation": "Now on",
"time": "118.422",
"value": "2011 Asian Winter Games"
},
{
"operation": "Page search",
"time": "131",
"value": "china"
},
{
"operation": "Page search",
"time": "157.783",
"value": "did not come as planned to the 2011"
},
{
"operation": "search",
"time": "169.416",
"value": "did not come as planned to the 2011"
},
{
"operation": "Now on",
"time": "173.979",
"value": "undefined"
},
{
"operation": "Now on",
"time": "220.156",
"value": "undefined"
},
{
"operation": "finish",
"time": "229.82",
"value": "finish"
}
]
}
|
exfever_8182
|
Square Enix developed and published Kingdom Hearts III for the eighth-generation console, competing with the PlayStation 4 from a multinational video game and digital entertainment company.
|
{
"evidence": [
"Kingdom Hearts III",
"Xbox One",
"Sony Interactive Entertainment"
],
"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:\nSquare Enix developed and published Kingdom Hearts III for the eighth-generation console, competing with the PlayStation 4 from a multinational video game and digital entertainment company.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Kingdom Hearts III",
"Xbox One",
"Sony Interactive Entertainment"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7329,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
46127
|
Australia's truple world champion Jack Brabham was in the first position at the Classification, finishing 80 laps with a time of 1:49:34.6
|
{
"evidence": [
"Jack Brabham_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:\nAustralia's truple world champion Jack Brabham was in the first position at the Classification, finishing 80 laps with a time of 1:49:34.6\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Jack Brabham_sentence_1"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7330,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.588",
"value": "?search="
},
{
"operation": "search",
"time": "53.659",
"value": "Jack Brabham"
},
{
"operation": "Now on",
"time": "55.529",
"value": "Jack Brabham"
},
{
"operation": "Page search",
"time": "75.527",
"value": "triple world champion"
},
{
"operation": "hyperlink",
"time": "160.856",
"value": "1960"
},
{
"operation": "Now on",
"time": "162.272",
"value": "1960 Formula One season"
},
{
"operation": "back-button-clicked",
"time": "188.023",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "188.284",
"value": "Jack Brabham"
},
{
"operation": "Highlighting",
"time": "207.774",
"value": "Jack Brabham_header_cell_0_7_0"
},
{
"operation": "Highlighting",
"time": "208.517",
"value": "Jack Brabham_cell_0_7_1"
},
{
"operation": "Highlighting",
"time": "208.529",
"value": "Jack Brabham_header_cell_0_7_0"
},
{
"operation": "Highlighting",
"time": "208.53",
"value": "Jack Brabham_header_cell_0_2_0"
},
{
"operation": "Page search",
"time": "225.025",
"value": "Classification"
},
{
"operation": "Highlighting",
"time": "250.253",
"value": "Jack Brabham_sentence_1"
},
{
"operation": "Highlighting deleted",
"time": "253.909",
"value": "Jack Brabham_cell_0_7_1"
},
{
"operation": "Highlighting deleted",
"time": "253.912",
"value": "Jack Brabham_header_cell_0_7_0"
},
{
"operation": "Highlighting deleted",
"time": "253.913",
"value": "Jack Brabham_header_cell_0_2_0"
},
{
"operation": "search",
"time": "274.116",
"value": "1"
},
{
"operation": "Now on",
"time": "276.257",
"value": "undefined"
},
{
"operation": "back-button-clicked",
"time": "305.899",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "306.161",
"value": "Jack Brabham"
},
{
"operation": "finish",
"time": "325.978",
"value": "finish"
}
]
}
|
exfever_7373
|
Liana Liberato is the youngest recipient of the Daytime Emmy Award.
|
{
"evidence": [
"Liana Liberato",
"Hulu"
],
"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:\nLiana Liberato is the youngest recipient of the Daytime Emmy Award.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Liana Liberato",
"Hulu"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7331,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
73057
|
Confederation Beach Park (a full-sized train station until 1978) is a Tourist Area/Park in Hamilton, Ontario, which includes Wild Waterworks, an outdoor waterpark; Adventure Village, an amusement centre with laser tag, bumper cars and an 18-hole mini golf course; and a fast food diner restaurant with a 1950s motif called Hutch's.
|
{
"evidence": [
"Confederation Beach Park_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:\nConfederation Beach Park (a full-sized train station until 1978) is a Tourist Area/Park in Hamilton, Ontario, which includes Wild Waterworks, an outdoor waterpark; Adventure Village, an amusement centre with laser tag, bumper cars and an 18-hole mini golf course; and a fast food diner restaurant with a 1950s motif called Hutch's.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Confederation Beach Park_sentence_4"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7332,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.001",
"value": "?search="
},
{
"operation": "search",
"time": "0.015",
"value": "Confederation Beach Park"
},
{
"operation": "Now on",
"time": "0.016",
"value": "Confederation Beach Park"
},
{
"operation": "Highlighting",
"time": "0.027",
"value": "Confederation Beach Park_sentence_0"
},
{
"operation": "Highlighting",
"time": "0.029",
"value": "Confederation Beach Park_sentence_1"
},
{
"operation": "finish",
"time": "0.068",
"value": "finish"
},
{
"operation": "Now on",
"time": "1622038911.058",
"value": "Confederation Beach Park"
},
{
"operation": "Highlighting",
"time": "1622038918.463",
"value": "Confederation Beach Park_sentence_4"
},
{
"operation": "Highlighting deleted",
"time": "1622038922.974",
"value": "Confederation Beach Park_sentence_1"
},
{
"operation": "Highlighting deleted",
"time": "1622038923.7",
"value": "Confederation Beach Park_sentence_0"
},
{
"operation": "Highlighting",
"time": "1622038929.757",
"value": "Confederation Beach Park_sentence_5"
},
{
"operation": "Highlighting deleted",
"time": "1622038933.346",
"value": "Confederation Beach Park_sentence_5"
},
{
"operation": "finish",
"time": "1622038938.284",
"value": "finish"
}
]
}
|
72426
|
The Garra nambulica is an extinct shark species that belongs in the Cyprinidae family and Garra Genus.
|
{
"evidence": [
"Garra nambulica_sentence_0",
"Cyprinidae_sentence_1"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Garra nambulica is an extinct shark species that belongs in the Cyprinidae family and Garra Genus.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Garra nambulica_sentence_0",
"Cyprinidae_sentence_1"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7333,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.001",
"value": "?search="
},
{
"operation": "search",
"time": "0.011",
"value": "Garra nambulica"
},
{
"operation": "Now on",
"time": "0.012",
"value": "Garra nambulica"
},
{
"operation": "Highlighting",
"time": "0.019",
"value": "Garra nambulica_sentence_0"
},
{
"operation": "hyperlink",
"time": "0.03",
"value": "Garra"
},
{
"operation": "Now on",
"time": "0.03",
"value": "Garra"
},
{
"operation": "back-button-clicked",
"time": "0.037",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "0.038",
"value": "Garra nambulica"
},
{
"operation": "hyperlink",
"time": "0.039",
"value": "cyprinid fish"
},
{
"operation": "Now on",
"time": "0.04",
"value": "Cyprinidae"
},
{
"operation": "Highlighting",
"time": "0.045",
"value": "Cyprinidae_sentence_1"
},
{
"operation": "back-button-clicked",
"time": "0.048",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "0.048",
"value": "Garra nambulica"
},
{
"operation": "finish",
"time": "0.062",
"value": "finish"
},
{
"operation": "Now on",
"time": "1621993575.242",
"value": "Garra nambulica"
},
{
"operation": "finish",
"time": "1621993576.813",
"value": "finish"
}
]
}
|
73636
|
The Environmentalists for Nuclear Energy Australia was established in 2007, which was based upon supporting economic policies of the Liberal and National parties, greater protection for native foests, reduction of greenhouse gas emissions, and support of the Gunns pulp mill in Tasmania.
|
{
"evidence": [
"Environmentalists for Nuclear Energy Australia_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:\nThe Environmentalists for Nuclear Energy Australia was established in 2007, which was based upon supporting economic policies of the Liberal and National parties, greater protection for native foests, reduction of greenhouse gas emissions, and support of the Gunns pulp mill in Tasmania.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Environmentalists for Nuclear Energy Australia_sentence_7"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7334,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.453",
"value": "?search="
},
{
"operation": "search",
"time": "16.07",
"value": "Environmentalists for Nuclear Energy Australia"
},
{
"operation": "Now on",
"time": "17.254",
"value": "Environmentalists for Nuclear Energy Australia"
},
{
"operation": "Highlighting",
"time": "28.244",
"value": "Environmentalists for Nuclear Energy Australia_cell_0_2_1"
},
{
"operation": "Highlighting",
"time": "28.249",
"value": "Environmentalists for Nuclear Energy Australia_header_cell_0_2_0"
},
{
"operation": "Highlighting",
"time": "28.249",
"value": "Environmentalists for Nuclear Energy Australia_header_cell_0_0_0"
},
{
"operation": "Highlighting",
"time": "40.162",
"value": "Environmentalists for Nuclear Energy Australia_sentence_4"
},
{
"operation": "Highlighting",
"time": "59.073",
"value": "Environmentalists for Nuclear Energy Australia_sentence_7"
},
{
"operation": "Highlighting",
"time": "65.294",
"value": "Environmentalists for Nuclear Energy Australia_sentence_5"
},
{
"operation": "Highlighting deleted",
"time": "85.129",
"value": "Environmentalists for Nuclear Energy Australia_sentence_5"
},
{
"operation": "Highlighting deleted",
"time": "85.632",
"value": "Environmentalists for Nuclear Energy Australia_sentence_4"
},
{
"operation": "Highlighting deleted",
"time": "87.968",
"value": "Environmentalists for Nuclear Energy Australia_cell_0_2_1"
},
{
"operation": "Highlighting deleted",
"time": "87.975",
"value": "Environmentalists for Nuclear Energy Australia_header_cell_0_2_0"
},
{
"operation": "Highlighting deleted",
"time": "87.977",
"value": "Environmentalists for Nuclear Energy Australia_header_cell_0_0_0"
},
{
"operation": "finish",
"time": "107.174",
"value": "finish"
}
]
}
|
74238
|
HD 200964 has a magnitude of 6.64 and is known to have two giant extrasolar planets, but it is too faint to be seen with the naked eye.
|
{
"evidence": [
"HD 200964_sentence_3",
"HD 200964_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:\nHD 200964 has a magnitude of 6.64 and is known to have two giant extrasolar planets, but it is too faint to be seen with the naked eye.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"HD 200964_sentence_3",
"HD 200964_sentence_5"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7335,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.349",
"value": "?search="
},
{
"operation": "search",
"time": "9.057",
"value": "HD 200964"
},
{
"operation": "Now on",
"time": "10.032",
"value": "HD 200964"
},
{
"operation": "Highlighting",
"time": "103.66",
"value": "HD 200964_cell_0_4_1"
},
{
"operation": "Highlighting",
"time": "103.668",
"value": "HD 200964_header_cell_0_4_0"
},
{
"operation": "Highlighting",
"time": "103.669",
"value": "HD 200964_header_cell_0_0_0"
},
{
"operation": "Highlighting deleted",
"time": "125.38",
"value": "HD 200964_cell_0_4_1"
},
{
"operation": "Highlighting deleted",
"time": "125.39",
"value": "HD 200964_header_cell_0_4_0"
},
{
"operation": "Highlighting deleted",
"time": "125.391",
"value": "HD 200964_header_cell_0_0_0"
},
{
"operation": "Highlighting",
"time": "126.929",
"value": "HD 200964_sentence_3"
},
{
"operation": "Highlighting",
"time": "128.11",
"value": "HD 200964_sentence_5"
},
{
"operation": "finish",
"time": "138.152",
"value": "finish"
}
]
}
|
49132
|
Nadia Comăneci is a Romanian gymnast who has won a total of nine Olympic medals, the most of any athlete Romania has ever fielded to the Olympics.
|
{
"evidence": [
"Nadia Comăneci_sentence_5"
],
"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:\nNadia Comăneci is a Romanian gymnast who has won a total of nine Olympic medals, the most of any athlete Romania has ever fielded to the Olympics.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Nadia Comăneci_sentence_5"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7336,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.134",
"value": "?search="
},
{
"operation": "search",
"time": "18.162",
"value": "Nadia Comăneci"
},
{
"operation": "Now on",
"time": "19.071",
"value": "Nadia Comăneci"
},
{
"operation": "Page search",
"time": "35.915",
"value": "Olympic medals"
},
{
"operation": "Highlighting",
"time": "40.082",
"value": "Nadia Comăneci_sentence_5"
},
{
"operation": "Page search",
"time": "75.307",
"value": "most"
},
{
"operation": "Page search",
"time": "118.36",
"value": "Olympic"
},
{
"operation": "Page search",
"time": "133.722",
"value": "Romania"
},
{
"operation": "finish",
"time": "326.552",
"value": "finish"
}
]
}
|
exfever_6649
|
Seattle is a seaport city on the West Coast of the United States, situated on a narrow piece of land between Puget Sound and Lake Washington.
|
{
"evidence": [
"Seattle",
"Isthmus"
],
"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:\nSeattle is a seaport city on the West Coast of the United States, situated on a narrow piece of land between Puget Sound and Lake Washington.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Seattle",
"Isthmus"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7337,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
86308
|
Shadow Forest is a children's novel by Matt Haig who is an English novelist and journalist.
|
{
"evidence": [
"Shadow Forest_sentence_0",
"Matt Haig_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:\nShadow Forest is a children's novel by Matt Haig who is an English novelist and journalist.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Shadow Forest_sentence_0",
"Matt Haig_sentence_0"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7338,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.111",
"value": "?search="
},
{
"operation": "search",
"time": "9.655",
"value": "Shadow Forest"
},
{
"operation": "Now on",
"time": "10.516",
"value": "Shadow Forest"
},
{
"operation": "Highlighting",
"time": "15.382",
"value": "Shadow Forest_sentence_0"
},
{
"operation": "hyperlink",
"time": "17.715",
"value": "Matt Haig"
},
{
"operation": "Now on",
"time": "18.203",
"value": "Matt Haig"
},
{
"operation": "Highlighting",
"time": "19.38",
"value": "Matt Haig_sentence_0"
},
{
"operation": "finish",
"time": "148.104",
"value": "finish"
}
]
}
|
exfever_1495
|
Anna Paquin is an actress who starred in a 1999 drama film starring an American actor and producer.
|
{
"evidence": [
"Anna Paquin",
"A Walk on the Moon",
"Liev Schreiber"
],
"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:\nAnna Paquin is an actress who starred in a 1999 drama film starring an American actor and producer.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Anna Paquin",
"A Walk on the Moon",
"Liev Schreiber"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7339,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_3398
|
Roger Allers is a stage director.
|
{
"evidence": [
"The Lion King",
"Walt Disney Animation Studios"
],
"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:\nRoger Allers is a stage director.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Lion King",
"Walt Disney Animation Studios"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7340,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_4469
|
Christoph Waltz is an Italian and German actor who received two copies of a golden statuette as an award from the Oscars that was originally sculpted by an American art director for the film industry.
|
{
"evidence": [
"Christoph Waltz",
"Academy Awards",
"Cedric Gibbons"
],
"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:\nChristoph Waltz is an Italian and German actor who received two copies of a golden statuette as an award from the Oscars that was originally sculpted by an American art director for the film industry.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Christoph Waltz",
"Academy Awards",
"Cedric Gibbons"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7341,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
47043
|
The Oriental hornet (Vespa orientalis, Linnaeus, 1761) is a social insect of the family Vespidae that is found in Southwest Asia, Northeast Africa, the island of Madagascar, the Middle East, Central Asia, and parts of Southern Europe.
|
{
"evidence": [
"Oriental hornet_sentence_0",
"Oriental hornet_sentence_1",
"Carl Linnaeus_sentence_238",
"Carl Linnaeus_sentence_6",
"Carl Linnaeus_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:\nThe Oriental hornet (Vespa orientalis, Linnaeus, 1761) is a social insect of the family Vespidae that is found in Southwest Asia, Northeast Africa, the island of Madagascar, the Middle East, Central Asia, and parts of Southern Europe.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Oriental hornet_sentence_0",
"Oriental hornet_sentence_1",
"Carl Linnaeus_sentence_238",
"Carl Linnaeus_sentence_6",
"Carl Linnaeus_sentence_4"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7342,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.3",
"value": "?search="
},
{
"operation": "search",
"time": "19.678",
"value": "Contains...The Oriental hornet"
},
{
"operation": "Now on",
"time": "21.325",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "23.794",
"value": "Oriental hornet"
},
{
"operation": "Now on",
"time": "24.417",
"value": "Oriental hornet"
},
{
"operation": "Highlighting",
"time": "28.462",
"value": "Oriental hornet_sentence_0"
},
{
"operation": "Highlighting",
"time": "47.997",
"value": "Oriental hornet_sentence_1"
},
{
"operation": "search",
"time": "75.163",
"value": "Vespa orientalis"
},
{
"operation": "Now on",
"time": "76.473",
"value": "Oriental hornet"
},
{
"operation": "search",
"time": "197.604",
"value": "Linnaeus"
},
{
"operation": "Now on",
"time": "198.823",
"value": "Carl Linnaeus"
},
{
"operation": "hyperlink",
"time": "206.2",
"value": "Linnaeus (disambiguation)"
},
{
"operation": "Now on",
"time": "207.097",
"value": "Linnaeus (disambiguation)"
},
{
"operation": "hyperlink",
"time": "221.278",
"value": "Carl Linnaeus"
},
{
"operation": "Now on",
"time": "222.509",
"value": "Carl Linnaeus"
},
{
"operation": "Highlighting",
"time": "273.189",
"value": "Carl Linnaeus_sentence_238"
},
{
"operation": "Highlighting",
"time": "288.429",
"value": "Carl Linnaeus_sentence_6"
},
{
"operation": "Highlighting",
"time": "300.922",
"value": "Carl Linnaeus_sentence_4"
},
{
"operation": "hyperlink",
"time": "301.715",
"value": "binomial nomenclature"
},
{
"operation": "Now on",
"time": "302.736",
"value": "Binomial nomenclature"
},
{
"operation": "finish",
"time": "317.491",
"value": "finish"
}
]
}
|
33533
|
The Nimbochromis livingstonii dorsal fin may be blue and the males tend to color up and turn a dark blue during breeding.
|
{
"evidence": [
"Nimbochromis livingstonii_sentence_6",
"Nimbochromis livingstonii_sentence_7",
"Nimbochromis livingstonii_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:\nThe Nimbochromis livingstonii dorsal fin may be blue and the males tend to color up and turn a dark blue during breeding.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Nimbochromis livingstonii_sentence_6",
"Nimbochromis livingstonii_sentence_7",
"Nimbochromis livingstonii_sentence_8"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7343,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.937",
"value": "?search="
},
{
"operation": "search",
"time": "697.347",
"value": "Nimbochromis livingstonii dorsal fin"
},
{
"operation": "Now on",
"time": "699.185",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "709.662",
"value": "Nimbochromis livingstonii"
},
{
"operation": "Now on",
"time": "711.062",
"value": "Nimbochromis livingstonii"
},
{
"operation": "Page search",
"time": "756.895",
"value": "fin"
},
{
"operation": "Page search",
"time": "773.67",
"value": "dorsal"
},
{
"operation": "Highlighting",
"time": "778.205",
"value": "Nimbochromis livingstonii_sentence_6"
},
{
"operation": "Highlighting",
"time": "792.173",
"value": "Nimbochromis livingstonii_sentence_7"
},
{
"operation": "Highlighting",
"time": "792.889",
"value": "Nimbochromis livingstonii_sentence_8"
},
{
"operation": "finish",
"time": "800.023",
"value": "finish"
}
]
}
|
92255
|
Hanley, a New Jersey-born coloratura soprano and teenage protégé of Leonard Bernstein, trained as an opera singer and attended the Juilliard School in New York City.
|
{
"evidence": [
"Trés Hanley_sentence_4",
"Trés Hanley_sentence_5",
"Trés Hanley_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:\nHanley, a New Jersey-born coloratura soprano and teenage protégé of Leonard Bernstein, trained as an opera singer and attended the Juilliard School in New York City.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"Trés Hanley_sentence_4",
"Trés Hanley_sentence_5",
"Trés Hanley_sentence_1"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7344,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.262",
"value": "?search="
},
{
"operation": "search",
"time": "15.713",
"value": "Hanley"
},
{
"operation": "Now on",
"time": "16.245",
"value": "Hanley"
},
{
"operation": "hyperlink",
"time": "20.304",
"value": "Hanley (name)"
},
{
"operation": "Now on",
"time": "20.869",
"value": "Hanley (name)"
},
{
"operation": "Page search",
"time": "31.673",
"value": "opera"
},
{
"operation": "Page search",
"time": "39.2",
"value": "singer"
},
{
"operation": "hyperlink",
"time": "53.729",
"value": "Trés Hanley"
},
{
"operation": "Now on",
"time": "54.283",
"value": "Trés Hanley"
},
{
"operation": "Highlighting",
"time": "61.761",
"value": "Trés Hanley_sentence_4"
},
{
"operation": "Highlighting",
"time": "65.161",
"value": "Trés Hanley_sentence_5"
},
{
"operation": "Highlighting",
"time": "78.873",
"value": "Trés Hanley_sentence_1"
},
{
"operation": "finish",
"time": "214.784",
"value": "finish"
}
]
}
|
exfever_5161
|
My Everything is the second studio album by American singer Ariana Grande, released on August 25, 2014, by a New York City-based American record label owned by a Dutch–American multinational music corporation.
|
{
"evidence": [
"My Everything (Ariana Grande album)",
"Republic Records",
"Universal Music Group"
],
"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:\nMy Everything is the second studio album by American singer Ariana Grande, released on August 25, 2014, by a New York City-based American record label owned by a Dutch–American multinational music corporation.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"My Everything (Ariana Grande album)",
"Republic Records",
"Universal Music Group"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7345,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_1667
|
Catching Fire is the last science fiction young adult novel written by Suzanne Collins.
|
{
"evidence": [
"Catching Fire",
"The Hunger Games (novel)"
],
"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:\nCatching Fire is the last science fiction young adult novel written by Suzanne Collins.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Catching Fire",
"The Hunger Games (novel)"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7346,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
42349
|
The church of San Esteban in Anúcita (a village in which Spanish is the local language) has an altarpiece that was built in 1792 and an outer wall window bearing a Chrismon.
|
{
"evidence": [
"Anúcita_sentence_7"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe church of San Esteban in Anúcita (a village in which Spanish is the local language) has an altarpiece that was built in 1792 and an outer wall window bearing a Chrismon.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Anúcita_sentence_7"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7347,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.873",
"value": "undefined"
},
{
"operation": "search",
"time": "14.985",
"value": "San Esteban in Anúcita"
},
{
"operation": "Now on",
"time": "16.583",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "21.692",
"value": "Anúcita"
},
{
"operation": "Now on",
"time": "22.886",
"value": "Anúcita"
},
{
"operation": "Highlighting",
"time": "42.931",
"value": "Anúcita_sentence_7"
},
{
"operation": "Page search",
"time": "69.753",
"value": "window bearing a Chrismon"
},
{
"operation": "search",
"time": "74.09",
"value": "window bearing a Chrismon"
},
{
"operation": "Now on",
"time": "75.828",
"value": "undefined"
},
{
"operation": "search",
"time": "94.381",
"value": "altarpiece that was built in 1792"
},
{
"operation": "Now on",
"time": "95.911",
"value": "undefined"
},
{
"operation": "search",
"time": "120.853",
"value": "The church of San Esteban in Anúcita ("
},
{
"operation": "Now on",
"time": "121.378",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "125.295",
"value": "Education Anúcita_section_0"
},
{
"operation": "Now on",
"time": "125.69",
"value": "Anúcita#Education An.C3.BAcita section 0"
},
{
"operation": "search",
"time": "137.128",
"value": "alterpice"
},
{
"operation": "Now on",
"time": "137.882",
"value": "undefined"
},
{
"operation": "back-button-clicked",
"time": "144.677",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "144.765",
"value": "Anúcita#Education An.C3.BAcita section 0"
},
{
"operation": "finish",
"time": "225.371",
"value": "finish"
}
]
}
|
exfever_3400
|
Tom Brady is a football quarterback who spent his first 22 seasons with a football team that was a charter member of the American Football League before joining the 51st regular season of the National Football League.
|
{
"evidence": [
"Tom Brady",
"New England Patriots",
"1970 NFL season"
],
"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:\nTom Brady is a football quarterback who spent his first 22 seasons with a football team that was a charter member of the American Football League before joining the 51st regular season of the National Football League.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Tom Brady",
"New England Patriots",
"1970 NFL season"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7348,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
56632
|
The Great Fire of 1911 (took place in Bangor, Maine, United States, on April 30 and May 1, 1911) was Maine's last large-scale urban conflagrations, but resulted in the creation of an early 20th-century urban space comprising of 48 buildings (and three parks); mostly constructed between 1911 and 1915 in the burned area.
|
{
"evidence": [
"Great Fire of 1911 Historic District_sentence_2",
"Great Fire of 1911 Historic District_sentence_3"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Great Fire of 1911 (took place in Bangor, Maine, United States, on April 30 and May 1, 1911) was Maine's last large-scale urban conflagrations, but resulted in the creation of an early 20th-century urban space comprising of 48 buildings (and three parks); mostly constructed between 1911 and 1915 in the burned area.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"Great Fire of 1911 Historic District_sentence_2",
"Great Fire of 1911 Historic District_sentence_3"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7349,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.629",
"value": "?search="
},
{
"operation": "search",
"time": "19.997",
"value": "Great Fire of 1911"
},
{
"operation": "Now on",
"time": "21.22",
"value": "Great Fire of 1911"
},
{
"operation": "Highlighting",
"time": "23.735",
"value": "Great Fire of 1911_sentence_0"
},
{
"operation": "Highlighting",
"time": "23.736",
"value": "Great Fire of 1911_title"
},
{
"operation": "search",
"time": "548.459",
"value": "Contains...The Great Fire of 1911"
},
{
"operation": "Now on",
"time": "552.06",
"value": "undefined"
},
{
"operation": "search",
"time": "567.222",
"value": "Great Fire of 1911 Historic District"
},
{
"operation": "Now on",
"time": "568.684",
"value": "Great Fire of 1911 Historic District"
},
{
"operation": "Highlighting",
"time": "585.507",
"value": "Great Fire of 1911 Historic District_sentence_2"
},
{
"operation": "Highlighting",
"time": "603.275",
"value": "Great Fire of 1911 Historic District_sentence_3"
},
{
"operation": "finish",
"time": "622.17",
"value": "finish"
}
]
}
|
58459
|
Although Pete Correale was a stand up comic, he never released a comedy CD.
|
{
"evidence": [
"Pete Correale_sentence_9"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nAlthough Pete Correale was a stand up comic, he never released a comedy CD.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Pete Correale_sentence_9"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7350,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.902",
"value": "?search="
},
{
"operation": "search",
"time": "10.59",
"value": "Pete Correale"
},
{
"operation": "Now on",
"time": "11.675",
"value": "Pete Correale"
},
{
"operation": "Highlighting",
"time": "31.23",
"value": "Pete Correale_sentence_0"
},
{
"operation": "Page search",
"time": "55.452",
"value": "comedy"
},
{
"operation": "Highlighting",
"time": "119.529",
"value": "Pete Correale_sentence_9"
},
{
"operation": "Highlighting",
"time": "122.673",
"value": "Pete Correale_sentence_9"
},
{
"operation": "finish",
"time": "142.532",
"value": "finish"
}
]
}
|
exfever_1012
|
Alec Baldwin played both leading and supporting roles in a variety of films, including a 1988 Australian romantic comedy-drama film directed by Mike Nichols.
|
{
"evidence": [
"Alec Baldwin",
"Working Girl"
],
"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:\nAlec Baldwin played both leading and supporting roles in a variety of films, including a 1988 Australian romantic comedy-drama film directed by Mike Nichols.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Alec Baldwin",
"Working Girl"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7351,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
6256
|
The Deed Is Done (reissued in 2013 under SPV/SteamHammer) is the second studio album by Molly Hatchet and the first album with only two guitarists.
|
{
"evidence": [
"The Deed Is Done_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 Deed Is Done (reissued in 2013 under SPV/SteamHammer) is the second studio album by Molly Hatchet and the first album with only two guitarists.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"The Deed Is Done_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7352,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.563",
"value": "?search="
},
{
"operation": "search",
"time": "14.516",
"value": "The Deed Is Done"
},
{
"operation": "Now on",
"time": "15.4",
"value": "The Deed Is Done"
},
{
"operation": "Highlighting",
"time": "24.729",
"value": "The Deed Is Done_sentence_0"
},
{
"operation": "finish",
"time": "43.76",
"value": "finish"
}
]
}
|
23274
|
Eliezer Gruenbaum joined the Communist Party of Poland at the age of 16, then fought in the Spanish Civil War (a civil war in Spain fought neither in 1936 nor in 1939) on the Republican side.
|
{
"evidence": [
"Spanish Civil War_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:\nEliezer Gruenbaum joined the Communist Party of Poland at the age of 16, then fought in the Spanish Civil War (a civil war in Spain fought neither in 1936 nor in 1939) on the Republican side.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Spanish Civil War_sentence_2"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7353,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.842",
"value": "?search="
},
{
"operation": "search",
"time": "8.211",
"value": "Eliezer Gruenbaum"
},
{
"operation": "Now on",
"time": "8.881",
"value": "Eliezer Gruenbaum"
},
{
"operation": "Page search",
"time": "24.346",
"value": "communist"
},
{
"operation": "Highlighting",
"time": "32.113",
"value": "Eliezer Gruenbaum_sentence_9"
},
{
"operation": "Highlighting",
"time": "32.114",
"value": "Eliezer Gruenbaum_title"
},
{
"operation": "Highlighting",
"time": "32.114",
"value": "Eliezer Gruenbaum_section_1"
},
{
"operation": "Page search",
"time": "45.633",
"value": "spanish civil war"
},
{
"operation": "Highlighting",
"time": "62.459",
"value": "Eliezer Gruenbaum_sentence_11"
},
{
"operation": "hyperlink",
"time": "64.549",
"value": "Spanish Civil War"
},
{
"operation": "Now on",
"time": "66.818",
"value": "Spanish Civil War"
},
{
"operation": "Page search",
"time": "79.059",
"value": "1936"
},
{
"operation": "Highlighting",
"time": "200.861",
"value": "Spanish Civil War_sentence_2"
},
{
"operation": "finish",
"time": "230.005",
"value": "finish"
}
]
}
|
exfever_2892
|
Gabourey Sidibe was a main cast member of the series, The Big C, in the Canadian premium television network owned by Paramount Media Networks.
|
{
"evidence": [
"Gabourey Sidibe",
"Showtime (TV network)"
],
"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:\nGabourey Sidibe was a main cast member of the series, The Big C, in the Canadian premium television network owned by Paramount Media Networks.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Gabourey Sidibe",
"Showtime (TV network)"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7354,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_1482
|
Patrick Bateman is the greatest fictional character created by novelist Bret Easton Ellis.
|
{
"evidence": [
"Patrick Bateman",
"Serial killer"
],
"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:\nPatrick Bateman is the greatest fictional character created by novelist Bret Easton Ellis.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Patrick Bateman",
"Serial killer"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7355,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
63819
|
Simarik is a double platinum 1997 song by Tarkan, a Turkish singer who released his first album in 1992 at the age of 20, and who has not found much success in his career since that time.
|
{
"evidence": [
"Tarkan (singer)_sentence_18"
],
"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:\nSimarik is a double platinum 1997 song by Tarkan, a Turkish singer who released his first album in 1992 at the age of 20, and who has not found much success in his career since that time.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Tarkan (singer)_sentence_18"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7356,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.735",
"value": "?search="
},
{
"operation": "search",
"time": "3.116",
"value": "Simarik"
},
{
"operation": "Now on",
"time": "8.727",
"value": "Şımarık"
},
{
"operation": "Highlighting",
"time": "10.582",
"value": "Şımarık_sentence_0"
},
{
"operation": "Highlighting",
"time": "10.583",
"value": "Şımarık_title"
},
{
"operation": "hyperlink",
"time": "13.042",
"value": "Tarkan"
},
{
"operation": "Now on",
"time": "22.772",
"value": "Tarkan (singer)"
},
{
"operation": "Highlighting",
"time": "23.652",
"value": "Tarkan (singer)_sentence_0"
},
{
"operation": "Highlighting",
"time": "30.13",
"value": "Tarkan (singer)_sentence_6"
},
{
"operation": "Highlighting",
"time": "46.94",
"value": "Tarkan (singer)_sentence_8"
},
{
"operation": "Highlighting deleted",
"time": "48.184",
"value": "Tarkan (singer)_sentence_8"
},
{
"operation": "Highlighting",
"time": "49.599",
"value": "Tarkan (singer)_sentence_8"
},
{
"operation": "Highlighting",
"time": "59.029",
"value": "Tarkan (singer)_sentence_18"
},
{
"operation": "finish",
"time": "63.561",
"value": "finish"
}
]
}
|
exfever_5696
|
The Silence of the Lambs is a 1993 American film with a genre focus on mental, emotional, and psychological states to frighten, disturb, and/or unsettle its audience.
|
{
"evidence": [
"The Silence of the Lambs (film)",
"Psychological horror"
],
"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 Silence of the Lambs is a 1993 American film with a genre focus on mental, emotional, and psychological states to frighten, disturb, and/or unsettle its audience.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Silence of the Lambs (film)",
"Psychological horror"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7357,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_105
|
Saving Private Ryan is a 1998 American epic war film that is set during Operation Overlord, the Allied operation that launched the successful invasion of German-occupied Western Europe during World War II.
|
{
"evidence": [
"Saving Private Ryan",
"Invasion of Normandy"
],
"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:\nSaving Private Ryan is a 1998 American epic war film that is set during Operation Overlord, the Allied operation that launched the successful invasion of German-occupied Western Europe during World War II.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Saving Private Ryan",
"Invasion of Normandy"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7358,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_1529
|
Machu Picchu is located within Urubamba Province above the Sacred Valley, which is 80 kilometers northwest of the capital of the Empire, which is also known as the Incan Empire or the Inka Empire.
|
{
"evidence": [
"Machu Picchu",
"Cusco",
"Inca Empire"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nMachu Picchu is located within Urubamba Province above the Sacred Valley, which is 80 kilometers northwest of the capital of the Empire, which is also known as the Incan Empire or the Inka Empire.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Machu Picchu",
"Cusco",
"Inca Empire"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7359,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_5343
|
Ingrid Bergman is the first Swedish actress to win a prestigious award for acting.
|
{
"evidence": [
"Enemy (2013 film)",
"Isabella Rossellini"
],
"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:\nIngrid Bergman is the first Swedish actress to win a prestigious award for acting.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Enemy (2013 film)",
"Isabella Rossellini"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7360,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
29161
|
The 1971 European 10 m Events Championships was the 1st edition of the special 10 m events competition, European 10 m Events Championships which are special shooting sport championships of 10 metre events.
|
{
"evidence": [
"1971 European 10 m Events Championships_sentence_0",
"European 10 m Events Championships_sentence_0"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe 1971 European 10 m Events Championships was the 1st edition of the special 10 m events competition, European 10 m Events Championships which are special shooting sport championships of 10 metre events.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"1971 European 10 m Events Championships_sentence_0",
"European 10 m Events Championships_sentence_0"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7361,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.659",
"value": "?search="
},
{
"operation": "search",
"time": "83.439",
"value": "1971 European 10 m Events Championships"
},
{
"operation": "Now on",
"time": "84.522",
"value": "1971 European 10 m Events Championships"
},
{
"operation": "Highlighting",
"time": "91.294",
"value": "1971 European 10 m Events Championships_sentence_0"
},
{
"operation": "hyperlink",
"time": "95.907",
"value": "European 10 m Events Championships"
},
{
"operation": "Now on",
"time": "97.111",
"value": "European 10 m Events Championships"
},
{
"operation": "Highlighting",
"time": "105.51",
"value": "European 10 m Events Championships_sentence_0"
},
{
"operation": "finish",
"time": "226.585",
"value": "finish"
}
]
}
|
83305
|
According to Moroccan nationality law, citizenship can be granted to those with a father born in Morocco, though this must be approved by the Minister of Justice.
|
{
"evidence": [
"Moroccan nationality law_sentence_12",
"Moroccan nationality law_sentence_13"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nAccording to Moroccan nationality law, citizenship can be granted to those with a father born in Morocco, though this must be approved by the Minister of Justice.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Moroccan nationality law_sentence_12",
"Moroccan nationality law_sentence_13"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7362,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.694",
"value": "undefined"
},
{
"operation": "search",
"time": "197.171",
"value": "Morrocan citizenship"
},
{
"operation": "Now on",
"time": "198.213",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "202.112",
"value": "moroccan citizenship"
},
{
"operation": "Now on",
"time": "203.643",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "208.131",
"value": "Human rights in Morocco"
},
{
"operation": "Now on",
"time": "209.422",
"value": "Human rights in Morocco"
},
{
"operation": "hyperlink",
"time": "220.889",
"value": "Moroccan nationality law"
},
{
"operation": "Now on",
"time": "221.463",
"value": "Moroccan nationality law"
},
{
"operation": "Highlighting",
"time": "229.706",
"value": "Moroccan nationality law_sentence_1"
},
{
"operation": "Highlighting",
"time": "252.412",
"value": "Moroccan nationality law_sentence_11"
},
{
"operation": "Highlighting deleted",
"time": "255.303",
"value": "Moroccan nationality law_sentence_11"
},
{
"operation": "Highlighting",
"time": "257.845",
"value": "Moroccan nationality law_sentence_12"
},
{
"operation": "Highlighting",
"time": "259.623",
"value": "Moroccan nationality law_sentence_13"
},
{
"operation": "Highlighting deleted",
"time": "273.956",
"value": "Moroccan nationality law_sentence_1"
},
{
"operation": "finish",
"time": "305.243",
"value": "finish"
}
]
}
|
43427
|
Rahah railway station in Iran includes Eslamshahrtoward Tehran and Shuttle bus.
|
{
"evidence": [
"Eslamshahr railway station_sentence_0",
"Islamic Republic of Iran Railways_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:\nRahah railway station in Iran includes Eslamshahrtoward Tehran and Shuttle bus.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Eslamshahr railway station_sentence_0",
"Islamic Republic of Iran Railways_sentence_0"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7363,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.15",
"value": "?search="
},
{
"operation": "search",
"time": "10.038",
"value": "Contains...Rahah railway station"
},
{
"operation": "Now on",
"time": "10.757",
"value": "undefined"
},
{
"operation": "search",
"time": "59.563",
"value": "Islamic Republic of Iran Railways"
},
{
"operation": "Now on",
"time": "61.363",
"value": "Islamic Republic of Iran Railways"
},
{
"operation": "Page search",
"time": "68.693",
"value": "rahah"
},
{
"operation": "Page search",
"time": "93.646",
"value": "eslamshah"
},
{
"operation": "Page search",
"time": "147.221",
"value": "eslam"
},
{
"operation": "search",
"time": "196.296",
"value": "Transport in Iran"
},
{
"operation": "Now on",
"time": "197.526",
"value": "Transport in Iran"
},
{
"operation": "Page search",
"time": "207.677",
"value": "eslamshah"
},
{
"operation": "search",
"time": "231.014",
"value": "Eslamshahr railway station"
},
{
"operation": "Now on",
"time": "231.921",
"value": "Eslamshahr railway station"
},
{
"operation": "Highlighting",
"time": "238.269",
"value": "Eslamshahr railway station_sentence_0"
},
{
"operation": "hyperlink",
"time": "273.181",
"value": "IRI Railway"
},
{
"operation": "Now on",
"time": "274.622",
"value": "Islamic Republic of Iran Railways"
},
{
"operation": "Highlighting",
"time": "280.9",
"value": "Islamic Republic of Iran Railways_sentence_0"
},
{
"operation": "finish",
"time": "286.255",
"value": "finish"
}
]
}
|
38545
|
Edwin Percy Baker won 11 county singles over a 37-year period and led Dorset in the Middleton Cup from 1927 until 1969 and won four English National Singles titles in 1932, 1946, 1952 & 1955, setting a record at the time.
|
{
"evidence": [
"Edwin Percy Baker_sentence_4",
"Edwin Percy Baker_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:\nEdwin Percy Baker won 11 county singles over a 37-year period and led Dorset in the Middleton Cup from 1927 until 1969 and won four English National Singles titles in 1932, 1946, 1952 & 1955, setting a record at the time.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Edwin Percy Baker_sentence_4",
"Edwin Percy Baker_sentence_5"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7364,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.053",
"value": "?search="
},
{
"operation": "search",
"time": "20.33",
"value": "Edwin Percy Baker"
},
{
"operation": "Now on",
"time": "20.971",
"value": "Edwin Percy Baker"
},
{
"operation": "Highlighting",
"time": "29.462",
"value": "Edwin Percy Baker_sentence_4"
},
{
"operation": "Page search",
"time": "51.227",
"value": "dorset"
},
{
"operation": "Page search",
"time": "82.318",
"value": "english"
},
{
"operation": "Highlighting",
"time": "91.11",
"value": "Edwin Percy Baker_sentence_5"
},
{
"operation": "finish",
"time": "187.888",
"value": "finish"
}
]
}
|
44706
|
Lyte as a Rock was her first album and in an interview for Okayplayer (an online hip-hop and alternative music website) the lyrics made up on the spot.
|
{
"evidence": [
"Lyte as a Rock_sentence_0",
"Lyte as a Rock_sentence_11"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nLyte as a Rock was her first album and in an interview for Okayplayer (an online hip-hop and alternative music website) the lyrics made up on the spot.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"Lyte as a Rock_sentence_0",
"Lyte as a Rock_sentence_11"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7365,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.732",
"value": "?search="
},
{
"operation": "search",
"time": "12.865",
"value": "Lyte as a Rock"
},
{
"operation": "Now on",
"time": "13.766",
"value": "Lyte as a Rock"
},
{
"operation": "Highlighting",
"time": "17.641",
"value": "Lyte as a Rock_sentence_0"
},
{
"operation": "Highlighting",
"time": "177.557",
"value": "Lyte as a Rock_sentence_11"
},
{
"operation": "finish",
"time": "189.753",
"value": "finish"
}
]
}
|
91406
|
Barry Brill was the youngest person to be appointed director of the New Zealand Broadcasting Corporation, and held many other positions such as student liaison in Victoria LAw School, Trustee of the Wellington Trustee Savings Bank,, and served on the Wellington Free Ambulance Board and Welling Hospital Board during his career.
|
{
"evidence": [
"Barry Brill_sentence_10"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nBarry Brill was the youngest person to be appointed director of the New Zealand Broadcasting Corporation, and held many other positions such as student liaison in Victoria LAw School, Trustee of the Wellington Trustee Savings Bank,, and served on the Wellington Free Ambulance Board and Welling Hospital Board during his career.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Barry Brill_sentence_10"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7366,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.296",
"value": "?search="
},
{
"operation": "search",
"time": "68.013",
"value": "Barry Brill"
},
{
"operation": "Now on",
"time": "68.848",
"value": "Barry Brill"
},
{
"operation": "Highlighting",
"time": "157.849",
"value": "Barry Brill_sentence_11"
},
{
"operation": "Highlighting",
"time": "318.471",
"value": "Barry Brill_sentence_10"
},
{
"operation": "Highlighting deleted",
"time": "319.639",
"value": "Barry Brill_sentence_11"
},
{
"operation": "finish",
"time": "399.335",
"value": "finish"
}
]
}
|
51534
|
The Dresden Monarchs was promoted to the German Football League (a Canadian football league in Germany), but had a difficult 2007 season.
|
{
"evidence": [
"German Football League_sentence_1"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Dresden Monarchs was promoted to the German Football League (a Canadian football league in Germany), but had a difficult 2007 season.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"German Football League_sentence_1"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7367,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.187",
"value": "?search="
},
{
"operation": "search",
"time": "12.51",
"value": "Dresden Monarchs"
},
{
"operation": "Now on",
"time": "13.406",
"value": "Dresden Monarchs"
},
{
"operation": "Page search",
"time": "43.139",
"value": "promo"
},
{
"operation": "Highlighting",
"time": "74.443",
"value": "Dresden Monarchs_sentence_7"
},
{
"operation": "Highlighting",
"time": "74.445",
"value": "Dresden Monarchs_title"
},
{
"operation": "Highlighting",
"time": "74.447",
"value": "Dresden Monarchs_section_0"
},
{
"operation": "hyperlink",
"time": "86.959",
"value": "German Football League"
},
{
"operation": "Now on",
"time": "89.424",
"value": "German Football League"
},
{
"operation": "Highlighting",
"time": "101.022",
"value": "German Football League_sentence_1"
},
{
"operation": "Page search",
"time": "114.872",
"value": "canada"
},
{
"operation": "back-button-clicked",
"time": "131.042",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "131.163",
"value": "Dresden Monarchs"
},
{
"operation": "Highlighting deleted",
"time": "134.3",
"value": "Dresden Monarchs_sentence_7"
},
{
"operation": "Highlighting deleted",
"time": "134.302",
"value": "Dresden Monarchs_title"
},
{
"operation": "Highlighting deleted",
"time": "134.305",
"value": "Dresden Monarchs_section_0"
},
{
"operation": "hyperlink",
"time": "135.169",
"value": "German Football League"
},
{
"operation": "Now on",
"time": "135.918",
"value": "German Football League"
},
{
"operation": "finish",
"time": "179.509",
"value": "finish"
}
]
}
|
23431
|
The first Paralympic Oath was taken by Italian athlete Franco Rossi during the 1950 Summer Paralympics, followed by Shigeo Aono in the succeeding event.
|
{
"evidence": [
"Summer Paralympic Games_sentence_14"
],
"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 first Paralympic Oath was taken by Italian athlete Franco Rossi during the 1950 Summer Paralympics, followed by Shigeo Aono in the succeeding event.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Summer Paralympic Games_sentence_14"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7368,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.029",
"value": "?search="
},
{
"operation": "search",
"time": "10.968",
"value": "Paralympic Oath"
},
{
"operation": "Now on",
"time": "12.345",
"value": "Paralympic symbols"
},
{
"operation": "search",
"time": "27.415",
"value": "Franco Rossi"
},
{
"operation": "Now on",
"time": "28.346",
"value": "Franco Rossi"
},
{
"operation": "hyperlink",
"time": "39.751",
"value": "Franco Rossi (athlete)"
},
{
"operation": "Now on",
"time": "40.283",
"value": "Franco Rossi (parathlete)"
},
{
"operation": "search",
"time": "67.104",
"value": "1950 Summer Paralympics"
},
{
"operation": "Now on",
"time": "68.624",
"value": "undefined"
},
{
"operation": "search",
"time": "86.546",
"value": "Summer Paralympic Games"
},
{
"operation": "Now on",
"time": "87.746",
"value": "Summer Paralympic Games"
},
{
"operation": "Highlighting",
"time": "102.409",
"value": "Summer Paralympic Games_sentence_14"
},
{
"operation": "finish",
"time": "114.543",
"value": "finish"
}
]
}
|
exfever_2568
|
A hate group is a majority that advocates and practices the use of physical force so as to injure, abuse, damage, or destroy.
|
{
"evidence": [
"Hate group",
"Violence"
],
"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:\nA hate group is a majority that advocates and practices the use of physical force so as to injure, abuse, damage, or destroy.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Hate group",
"Violence"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7369,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_1673
|
To Kill a Mockingbird is Gregory Peck's second film.
|
{
"evidence": [
"To Kill a Mockingbird (film)",
"Robert Mulligan"
],
"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:\nTo Kill a Mockingbird is Gregory Peck's second film.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"To Kill a Mockingbird (film)",
"Robert Mulligan"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7370,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
37982
|
Frederick Borden, the Minister of Militia and Defence, announced in May that a 200-strong force would be deployed to Fort Selkirk; and one of the arguments put forward in favour of this option was that it was much cheaper than sending additional police, who enjoyed higher pay than soldiers.
|
{
"evidence": [
"Frederick William Borden_sentence_1",
"Frederick William Borden_sentence_17",
"Frederick William Borden_sentence_18"
],
"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:\nFrederick Borden, the Minister of Militia and Defence, announced in May that a 200-strong force would be deployed to Fort Selkirk; and one of the arguments put forward in favour of this option was that it was much cheaper than sending additional police, who enjoyed higher pay than soldiers.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Frederick William Borden_sentence_1",
"Frederick William Borden_sentence_17",
"Frederick William Borden_sentence_18"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7371,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "2.076",
"value": "?search="
},
{
"operation": "search",
"time": "14.676",
"value": "Frederick Borden"
},
{
"operation": "Now on",
"time": "16.052",
"value": "Frederick William Borden"
},
{
"operation": "Highlighting",
"time": "33.967",
"value": "Frederick William Borden_sentence_1"
},
{
"operation": "search",
"time": "60.663",
"value": "Fort Selkirk"
},
{
"operation": "Now on",
"time": "61.784",
"value": "Fort Selkirk"
},
{
"operation": "back-button-clicked",
"time": "81.151",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "82.402",
"value": "Frederick William Borden"
},
{
"operation": "hyperlink",
"time": "84.951",
"value": "Second Boer War"
},
{
"operation": "Now on",
"time": "88.615",
"value": "Second Boer War"
},
{
"operation": "search",
"time": "146.294",
"value": "Fort Selkirk"
},
{
"operation": "Now on",
"time": "147.555",
"value": "Fort Selkirk"
},
{
"operation": "back-button-clicked",
"time": "172.849",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "173.233",
"value": "Second Boer War"
},
{
"operation": "back-button-clicked",
"time": "173.885",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "173.895",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "174.092",
"value": "Frederick William Borden"
},
{
"operation": "Highlighting",
"time": "233.757",
"value": "Frederick William Borden_sentence_17"
},
{
"operation": "Highlighting",
"time": "234.407",
"value": "Frederick William Borden_sentence_18"
},
{
"operation": "finish",
"time": "254.048",
"value": "finish"
}
]
}
|
11935
|
Kewal Singh (an Indian diplomat, Foreign Secretary and India's ambassador to the USSR, Pakistan and USA) took office on October 21, 1954 then left on November 16, 1956; before M.K. Kripalani took over on November 17, 1956.
|
{
"evidence": [
"Kewal Singh_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:\nKewal Singh (an Indian diplomat, Foreign Secretary and India's ambassador to the USSR, Pakistan and USA) took office on October 21, 1954 then left on November 16, 1956; before M.K. Kripalani took over on November 17, 1956.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Kewal Singh_sentence_0"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7372,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.787",
"value": "undefined"
},
{
"operation": "search",
"time": "8.301",
"value": "kewal singh"
},
{
"operation": "Now on",
"time": "9.465",
"value": "Kewal Singh"
},
{
"operation": "Highlighting",
"time": "11.891",
"value": "Kewal Singh_sentence_0"
},
{
"operation": "finish",
"time": "126.145",
"value": "finish"
}
]
}
|
exfever_5217
|
Jack Dylan Grazer is an American actor who had a breakthrough playing the role of Eddie Kaspbrak in the film adaptation of the Stephen King novel It, directed by Andy Muschietti and written by Chase Palmer.
|
{
"evidence": [
"Jack Dylan Grazer",
"It (2017 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:\nJack Dylan Grazer is an American actor who had a breakthrough playing the role of Eddie Kaspbrak in the film adaptation of the Stephen King novel It, directed by Andy Muschietti and written by Chase Palmer.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Jack Dylan Grazer",
"It (2017 film)"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7373,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
86198
|
Maroš Šefčovič has been a member of the European Commission since 2009 which is the executive branch of the European Union, responsible for proposing legislation, but does not implement decisions, upholding the EU treaties and managing the day-to-day business of the EU.
|
{
"evidence": [
"European Commission_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:\nMaroš Šefčovič has been a member of the European Commission since 2009 which is the executive branch of the European Union, responsible for proposing legislation, but does not implement decisions, upholding the EU treaties and managing the day-to-day business of the EU.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"European Commission_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7374,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.069",
"value": "?search="
},
{
"operation": "search",
"time": "115.073",
"value": "Maroš Šefčovič"
},
{
"operation": "Now on",
"time": "115.974",
"value": "Maroš Šefčovič"
},
{
"operation": "Highlighting",
"time": "124.903",
"value": "Maroš Šefčovič_sentence_1"
},
{
"operation": "Highlighting",
"time": "124.904",
"value": "Maroš Šefčovič_title"
},
{
"operation": "hyperlink",
"time": "128.443",
"value": "European Commission"
},
{
"operation": "Now on",
"time": "129.652",
"value": "European Commission"
},
{
"operation": "Highlighting",
"time": "136.737",
"value": "European Commission_sentence_0"
},
{
"operation": "finish",
"time": "157.407",
"value": "finish"
}
]
}
|
56649
|
The ice shelf of the Sulzberger Bay (a bay between Fisher Island and Vollmer Island, along the coast of King Edward VII Land) released icebergs within a day of the 2011 Tōhoku earthquake and tsunami, thus some scientists have linked the ice calving to the tsunami reaching the ice shelf, some 13,600 kilometers (8,500 mi) away from the earthquake epicenter.
|
{
"evidence": [
"Sulzberger Bay_sentence_0",
"Sulzberger Bay_sentence_5",
"Sulzberger Bay_sentence_6"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe ice shelf of the Sulzberger Bay (a bay between Fisher Island and Vollmer Island, along the coast of King Edward VII Land) released icebergs within a day of the 2011 Tōhoku earthquake and tsunami, thus some scientists have linked the ice calving to the tsunami reaching the ice shelf, some 13,600 kilometers (8,500 mi) away from the earthquake epicenter.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"Sulzberger Bay_sentence_0",
"Sulzberger Bay_sentence_5",
"Sulzberger Bay_sentence_6"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7375,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.527",
"value": "?search="
},
{
"operation": "search",
"time": "17.822",
"value": "Sulzberger Bay"
},
{
"operation": "Now on",
"time": "19.92",
"value": "Sulzberger Bay"
},
{
"operation": "Highlighting",
"time": "31.176",
"value": "Sulzberger Bay_sentence_0"
},
{
"operation": "Highlighting",
"time": "44.074",
"value": "Sulzberger Bay_sentence_5"
},
{
"operation": "Highlighting",
"time": "62.338",
"value": "Sulzberger Bay_sentence_6"
},
{
"operation": "finish",
"time": "105.077",
"value": "finish"
}
]
}
|
exfever_4011
|
The United States Congress was created by the Constitution of the United States to replace its legislative function as the governing body of the United States of America.
|
{
"evidence": [
"United States Congress",
"Congress of the Confederation"
],
"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 United States Congress was created by the Constitution of the United States to replace its legislative function as the governing body of the United States of America.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"United States Congress",
"Congress of the Confederation"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 7376,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_6322
|
David LeRoy Anderson is an English makeup artist who has won an award for a 1997 American science fiction action comedy film loosely based on the series created by an American comic book author.
|
{
"evidence": [
"David LeRoy Anderson",
"Men in Black (1997 film)",
"Lowell Cunningham"
],
"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 LeRoy Anderson is an English makeup artist who has won an award for a 1997 American science fiction action comedy film loosely based on the series created by an American comic book author.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"David LeRoy Anderson",
"Men in Black (1997 film)",
"Lowell Cunningham"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7377,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_1191
|
Hamburg is the capital and largest city of the Spanish autonomous community, situated on the lower reaches of the only major navigable river in Spain.
|
{
"evidence": [
"Seville",
"Guadalquivir"
],
"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:\nHamburg is the capital and largest city of the Spanish autonomous community, situated on the lower reaches of the only major navigable river in Spain.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Seville",
"Guadalquivir"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7378,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7858
|
"What a Girl Wants" is a song recorded by the youngest American singer, Christina Aguilera.
|
{
"evidence": [
"What a Girl Wants (Christina Aguilera song)",
"Genie in a Bottle"
],
"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\"What a Girl Wants\" is a song recorded by the youngest American singer, Christina Aguilera.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"What a Girl Wants (Christina Aguilera song)",
"Genie in a Bottle"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7379,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_2413
|
Walter Matthau was the most famous American actor.
|
{
"evidence": [
"Walter Matthau",
"King Creole",
"Carolyn Jones"
],
"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:\nWalter Matthau was the most famous American actor.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Walter Matthau",
"King Creole",
"Carolyn Jones"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7380,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7938
|
The Grateful Dead was a Canadian rock band that fused elements such as music genre and musical form which originated in the Deep South of the United States around the 1860s by African-Americans.
|
{
"evidence": [
"Grateful Dead",
"Blues"
],
"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 Grateful Dead was a Canadian rock band that fused elements such as music genre and musical form which originated in the Deep South of the United States around the 1860s by African-Americans.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Grateful Dead",
"Blues"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7381,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
5993
|
Most people in Bhamlada are from Schedule Caste, the town has the Schedule Tribe population and as per census 496 males and 46 females were engaged in work activities.
|
{
"evidence": [
"Bhamlada_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:\nMost people in Bhamlada are from Schedule Caste, the town has the Schedule Tribe population and as per census 496 males and 46 females were engaged in work activities.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Bhamlada_sentence_7"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 7382,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.988",
"value": "?search="
},
{
"operation": "search",
"time": "14.783",
"value": "Bhamlada"
},
{
"operation": "Now on",
"time": "15.7",
"value": "Bhamlada"
},
{
"operation": "Highlighting",
"time": "24.797",
"value": "Bhamlada_sentence_6"
},
{
"operation": "Highlighting",
"time": "34.139",
"value": "Bhamlada_sentence_7"
},
{
"operation": "Highlighting deleted",
"time": "35.555",
"value": "Bhamlada_sentence_6"
},
{
"operation": "finish",
"time": "49.29",
"value": "finish"
}
]
}
|
93109
|
Gangavalli is a town in Salem district in the state of Tamil Nadu, India that is flat.
|
{
"evidence": [
"Gangavalli_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:\nGangavalli is a town in Salem district in the state of Tamil Nadu, India that is flat.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Gangavalli_sentence_1"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 7383,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "9.468",
"value": "?search="
},
{
"operation": "search",
"time": "92.815",
"value": "Gangavalli"
},
{
"operation": "Now on",
"time": "95.629",
"value": "Gangavalli"
},
{
"operation": "Highlighting",
"time": "97.025",
"value": "Gangavalli_sentence_1"
},
{
"operation": "Highlighting",
"time": "124.874",
"value": "Gangavalli_sentence_1"
},
{
"operation": "finish",
"time": "211.168",
"value": "finish"
}
]
}
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.