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_7547
|
Adolf Hitler was Germany's political leader who possesses absolute power from 1933 until his death in 1945.
|
{
"evidence": [
"Adolf Hitler",
"Dictator"
],
"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:\nAdolf Hitler was Germany's political leader who possesses absolute power from 1933 until his death in 1945.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Adolf Hitler",
"Dictator"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 84,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
28265
|
In the list of Unification of Norway (860–872), Vestfold (current electoral district in Eastern Norway) gained six victories while Viking gained only one victory.
|
{
"evidence": [
"Norway_sentence_110"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nIn the list of Unification of Norway (860–872), Vestfold (current electoral district in Eastern Norway) gained six victories while Viking gained only one victory.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Norway_sentence_110"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 85,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.035",
"value": "?search="
},
{
"operation": "search",
"time": "16.134",
"value": "Unification of Norway"
},
{
"operation": "Now on",
"time": "17.775",
"value": "undefined"
},
{
"operation": "search",
"time": "57.878",
"value": "list of Unification of Norway"
},
{
"operation": "Now on",
"time": "59.564",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "71.966",
"value": "In Norway Harald Fairhair_section_7"
},
{
"operation": "Now on",
"time": "72.84",
"value": "Harald Fairhair#In Norway Harald Fairhair section 7"
},
{
"operation": "Page search",
"time": "94.071",
"value": "unification"
},
{
"operation": "search",
"time": "134.727",
"value": "the unification of norway"
},
{
"operation": "Now on",
"time": "135.552",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "145.102",
"value": "Battle of Svolder"
},
{
"operation": "Now on",
"time": "145.883",
"value": "Battle of Svolder"
},
{
"operation": "search",
"time": "172.374",
"value": "Vestfold"
},
{
"operation": "Now on",
"time": "174.308",
"value": "Vestfold"
},
{
"operation": "search",
"time": "226.335",
"value": "Unification of Germany"
},
{
"operation": "Now on",
"time": "228.39",
"value": "Unification of Germany"
},
{
"operation": "search",
"time": "277.821",
"value": "Unification of Norway"
},
{
"operation": "Now on",
"time": "278.83",
"value": "undefined"
},
{
"operation": "search",
"time": "292.254",
"value": "norway"
},
{
"operation": "Now on",
"time": "295.294",
"value": "Norway"
},
{
"operation": "Page search",
"time": "303.596",
"value": "unification"
},
{
"operation": "Highlighting",
"time": "320.743",
"value": "Norway_sentence_110"
},
{
"operation": "Highlighting deleted",
"time": "322.214",
"value": "Norway_sentence_110"
},
{
"operation": "hyperlink",
"time": "324.413",
"value": "Hereditary Kingdom of Norway"
},
{
"operation": "Now on",
"time": "325.625",
"value": "History of the Norwegian monarchy"
},
{
"operation": "back-button-clicked",
"time": "341.526",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "341.881",
"value": "Norway"
},
{
"operation": "hyperlink",
"time": "377.333",
"value": "Viking Age"
},
{
"operation": "Now on",
"time": "379.238",
"value": "Viking Age"
},
{
"operation": "Page search",
"time": "391.727",
"value": "vestfold"
},
{
"operation": "back-button-clicked",
"time": "400.83",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "401.19",
"value": "Norway"
},
{
"operation": "Highlighting",
"time": "404.077",
"value": "Norway_sentence_110"
},
{
"operation": "finish",
"time": "412.253",
"value": "finish"
}
]
}
|
37916
|
Jeff Van Camp, whose junior year campaign was halted by the effect that Hurricane Ivan, a large, long-lived, Cape Verde hurricane that caused widespread damage in the Caribbean and United States, had on the Navarre community, was forced to assist with his family's needs, cutting short his football season.
|
{
"evidence": [
"Jeff Van Camp_sentence_2",
"Jeff Van Camp_sentence_3",
"Hurricane Ivan_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:\nJeff Van Camp, whose junior year campaign was halted by the effect that Hurricane Ivan, a large, long-lived, Cape Verde hurricane that caused widespread damage in the Caribbean and United States, had on the Navarre community, was forced to assist with his family's needs, cutting short his football season.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Jeff Van Camp_sentence_2",
"Jeff Van Camp_sentence_3",
"Hurricane Ivan_sentence_3"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 86,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.723",
"value": "?search="
},
{
"operation": "search",
"time": "14.8",
"value": "Jeff Van Camp"
},
{
"operation": "Now on",
"time": "15.687",
"value": "Jeff Van Camp"
},
{
"operation": "Page search",
"time": "86.12",
"value": "hurricane ivan"
},
{
"operation": "Highlighting",
"time": "91.928",
"value": "Jeff Van Camp_sentence_2"
},
{
"operation": "Highlighting",
"time": "104.178",
"value": "Jeff Van Camp_sentence_3"
},
{
"operation": "hyperlink",
"time": "106.422",
"value": "Hurricane Ivan"
},
{
"operation": "Now on",
"time": "107.388",
"value": "Hurricane Ivan"
},
{
"operation": "Highlighting",
"time": "109.786",
"value": "Hurricane Ivan_sentence_3"
},
{
"operation": "back-button-clicked",
"time": "119.734",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "119.829",
"value": "Jeff Van Camp"
},
{
"operation": "Highlighting",
"time": "229.736",
"value": "Jeff Van Camp_sentence_0"
},
{
"operation": "Highlighting deleted",
"time": "230.306",
"value": "Jeff Van Camp_sentence_0"
},
{
"operation": "finish",
"time": "233.895",
"value": "finish"
}
]
}
|
21153
|
Cameroon lists three definitely endangered languages, 13 severely endangered, and 16 critically endangered from among its at least 250 languages.
|
{
"evidence": [
"Languages of Cameroon_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:\nCameroon lists three definitely endangered languages, 13 severely endangered, and 16 critically endangered from among its at least 250 languages.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Languages of Cameroon_sentence_0"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 87,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.583",
"value": "?search="
},
{
"operation": "search",
"time": "21.234",
"value": "Contains...Cameroon endangered languages"
},
{
"operation": "Now on",
"time": "22.301",
"value": "undefined"
},
{
"operation": "search",
"time": "32.548",
"value": "cameroon languages"
},
{
"operation": "Now on",
"time": "33.527",
"value": "undefined"
},
{
"operation": "search",
"time": "44.154",
"value": "Cameroon"
},
{
"operation": "Now on",
"time": "46.028",
"value": "Cameroon"
},
{
"operation": "Page search",
"time": "49.106",
"value": "endangererd"
},
{
"operation": "Page search",
"time": "68.507",
"value": "languages"
},
{
"operation": "Highlighting",
"time": "83.036",
"value": "Cameroon_sentence_8"
},
{
"operation": "Highlighting",
"time": "83.038",
"value": "Cameroon_title"
},
{
"operation": "hyperlink",
"time": "103.304",
"value": "8.2 Languages Cameroon_section_18"
},
{
"operation": "hyperlink",
"time": "114.289",
"value": "Languages of Cameroon"
},
{
"operation": "Now on",
"time": "116.195",
"value": "Languages of Cameroon"
},
{
"operation": "Page search",
"time": "128.479",
"value": "endangered"
},
{
"operation": "Page search",
"time": "129.331",
"value": "endangered"
},
{
"operation": "Highlighting",
"time": "152.54",
"value": "Languages of Cameroon_sentence_0"
},
{
"operation": "Page search",
"time": "165.511",
"value": "critically"
},
{
"operation": "search",
"time": "214.581",
"value": "critically endangered cameroon"
},
{
"operation": "Now on",
"time": "215.915",
"value": "undefined"
},
{
"operation": "search",
"time": "234.177",
"value": "endangered languages cameroon"
},
{
"operation": "Now on",
"time": "234.926",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "253.319",
"value": "Bung language"
},
{
"operation": "Now on",
"time": "253.85",
"value": "Bung language"
},
{
"operation": "hyperlink",
"time": "261.26",
"value": "Cameroon"
},
{
"operation": "Now on",
"time": "261.681",
"value": "Cameroon"
},
{
"operation": "search",
"time": "281.776",
"value": "Languages of Cameroon"
},
{
"operation": "Now on",
"time": "282.304",
"value": "Languages of Cameroon"
},
{
"operation": "Page search",
"time": "295.73",
"value": "13"
},
{
"operation": "hyperlink",
"time": "325.048",
"value": "Demographics of Cameroon"
},
{
"operation": "Now on",
"time": "326.332",
"value": "Demographics of Cameroon"
},
{
"operation": "Page search",
"time": "330.483",
"value": "endangered"
},
{
"operation": "Page search",
"time": "347.474",
"value": "languages"
},
{
"operation": "hyperlink",
"time": "351.933",
"value": "4 Languages Demographics of Cameroon_section_5"
},
{
"operation": "search",
"time": "375.344",
"value": "Languages of Cameroon"
},
{
"operation": "Now on",
"time": "376.012",
"value": "Languages of Cameroon"
},
{
"operation": "Page search",
"time": "388.374",
"value": "critically"
},
{
"operation": "Page search",
"time": "429.013",
"value": "severely endangered language cameroon"
},
{
"operation": "search",
"time": "435.881",
"value": "severely endangered language cameroon"
},
{
"operation": "Now on",
"time": "437.334",
"value": "undefined"
},
{
"operation": "search",
"time": "453.057",
"value": "Contains...endangered language cameroon"
},
{
"operation": "Now on",
"time": "453.789",
"value": "undefined"
},
{
"operation": "search",
"time": "468.699",
"value": "Contains...lanugages of camerooon"
},
{
"operation": "Now on",
"time": "469.445",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "470.783",
"value": "Languages of Cameroon"
},
{
"operation": "Now on",
"time": "471.451",
"value": "Languages of Cameroon"
},
{
"operation": "Page search",
"time": "515.062",
"value": "threat"
},
{
"operation": "finish",
"time": "534.152",
"value": "finish"
}
]
}
|
64604
|
In March 1940, BMI issued their very first contract, which was for "We Could Make Such Beautiful Music Together", with music composed by Katzman and lyrics by Robert Sour; which was frequently covered by a wide variety of artists, including vintage recordings by Barbara Lea, Judy Garland, and Liza Minnelli.
|
{
"evidence": [
"Henry Katzman_sentence_5",
"Henry Katzman_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:\nIn March 1940, BMI issued their very first contract, which was for \"We Could Make Such Beautiful Music Together\", with music composed by Katzman and lyrics by Robert Sour; which was frequently covered by a wide variety of artists, including vintage recordings by Barbara Lea, Judy Garland, and Liza Minnelli.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Henry Katzman_sentence_5",
"Henry Katzman_sentence_6"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 88,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.078",
"value": "?search="
},
{
"operation": "search",
"time": "15.446",
"value": "BMI"
},
{
"operation": "Now on",
"time": "16.326",
"value": "BMI"
},
{
"operation": "search",
"time": "150.564",
"value": "We Could Make Such Beautiful Music Together"
},
{
"operation": "Now on",
"time": "154.572",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "188.322",
"value": "Henry Katzman"
},
{
"operation": "Now on",
"time": "188.913",
"value": "Henry Katzman"
},
{
"operation": "Highlighting",
"time": "190.794",
"value": "Henry Katzman_sentence_5"
},
{
"operation": "Highlighting",
"time": "232.623",
"value": "Henry Katzman_sentence_6"
},
{
"operation": "finish",
"time": "237.354",
"value": "finish"
}
]
}
|
exfever_5289
|
Jeff Hardy is WWE's first wrestler.
|
{
"evidence": [
"Jeff Hardy",
"The Hardy Boyz",
"National Wrestling Alliance"
],
"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:\nJeff Hardy is WWE's first wrestler.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Jeff Hardy",
"The Hardy Boyz",
"National Wrestling Alliance"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 89,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_8435
|
Tender Mercies is a 1983 American drama film directed by Bruce Beresford, whose supporting cast includes an American actress who was inducted in 2012 into the lists founded in New York City, whose inductees come from various disciplines, including actors.
|
{
"evidence": [
"Tender Mercies",
"Betty Buckley",
"American Theater Hall of Fame"
],
"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:\nTender Mercies is a 1983 American drama film directed by Bruce Beresford, whose supporting cast includes an American actress who was inducted in 2012 into the lists founded in New York City, whose inductees come from various disciplines, including actors.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Tender Mercies",
"Betty Buckley",
"American Theater Hall of Fame"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 90,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
51105
|
Le comte Ory is a Spanish comic opera that was written by Gioachino Rossini on 20 August 1828 and was first performed in Salle Le Peletier, Paris, France.
|
{
"evidence": [
"Le comte Ory_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:\nLe comte Ory is a Spanish comic opera that was written by Gioachino Rossini on 20 August 1828 and was first performed in Salle Le Peletier, Paris, France.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Le comte Ory_sentence_0"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 91,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.419",
"value": "?search="
},
{
"operation": "search",
"time": "37.175",
"value": "Le Comte Ory"
},
{
"operation": "Now on",
"time": "39.049",
"value": "Le Comte Ory"
},
{
"operation": "Highlighting",
"time": "61.627",
"value": "Le comte Ory_sentence_0"
},
{
"operation": "Highlighting",
"time": "88.188",
"value": "Le comte Ory_cell_0_3_1"
},
{
"operation": "Highlighting",
"time": "88.193",
"value": "Le comte Ory_header_cell_0_3_0"
},
{
"operation": "Highlighting",
"time": "88.194",
"value": "Le comte Ory_header_cell_0_0_0"
},
{
"operation": "Highlighting deleted",
"time": "254.194",
"value": "Le comte Ory_cell_0_3_1"
},
{
"operation": "Highlighting deleted",
"time": "254.199",
"value": "Le comte Ory_header_cell_0_3_0"
},
{
"operation": "Highlighting deleted",
"time": "254.202",
"value": "Le comte Ory_header_cell_0_0_0"
},
{
"operation": "finish",
"time": "287.314",
"value": "finish"
}
]
}
|
42778
|
Wilhelm von Schütz was a German author who translated Casanova's Histoire de ma vie and served as the editor of the Catholic magazine Anticelsus from 1842 to 1946.
|
{
"evidence": [
"Wilhelm von Schütz_sentence_0",
"Wilhelm von Schütz_sentence_2",
"Wilhelm von Schütz_sentence_5"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nWilhelm von Schütz was a German author who translated Casanova's Histoire de ma vie and served as the editor of the Catholic magazine Anticelsus from 1842 to 1946.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Wilhelm von Schütz_sentence_0",
"Wilhelm von Schütz_sentence_2",
"Wilhelm von Schütz_sentence_5"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 92,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.022",
"value": "?search="
},
{
"operation": "search",
"time": "9.02",
"value": "Wilhelm von Schütz"
},
{
"operation": "Now on",
"time": "9.862",
"value": "Wilhelm von Schütz"
},
{
"operation": "Highlighting",
"time": "12.643",
"value": "Wilhelm von Schütz_sentence_0"
},
{
"operation": "Highlighting",
"time": "16.531",
"value": "Wilhelm von Schütz_sentence_2"
},
{
"operation": "Highlighting",
"time": "26.834",
"value": "Wilhelm von Schütz_sentence_5"
},
{
"operation": "finish",
"time": "44.77",
"value": "finish"
}
]
}
|
36422
|
The 30' x 16' Waterford State School made mostly of Pine was designed by Richard George Suter of Queensland, who later became a priest, was built in 1871.
|
{
"evidence": [
"Waterford State School_sentence_4",
"Waterford State School_sentence_51",
"Richard George Suter_sentence_5"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe 30' x 16' Waterford State School made mostly of Pine was designed by Richard George Suter of Queensland, who later became a priest, was built in 1871.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Waterford State School_sentence_4",
"Waterford State School_sentence_51",
"Richard George Suter_sentence_5"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 93,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.466",
"value": "?search="
},
{
"operation": "search",
"time": "120.384",
"value": "Waterford State School"
},
{
"operation": "Now on",
"time": "121.879",
"value": "Waterford State School"
},
{
"operation": "Page search",
"time": "157.737",
"value": "pine"
},
{
"operation": "Highlighting",
"time": "173.508",
"value": "Waterford State School_sentence_4"
},
{
"operation": "hyperlink",
"time": "195.202",
"value": "Richard George Suter"
},
{
"operation": "Now on",
"time": "196.137",
"value": "Richard George Suter"
},
{
"operation": "back-button-clicked",
"time": "209.895",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "210.005",
"value": "Waterford State School"
},
{
"operation": "Page search",
"time": "249.373",
"value": "30"
},
{
"operation": "Highlighting",
"time": "263.85",
"value": "Waterford State School_sentence_51"
},
{
"operation": "hyperlink",
"time": "269.056",
"value": "Richard George Suter"
},
{
"operation": "Now on",
"time": "269.552",
"value": "Richard George Suter"
},
{
"operation": "Highlighting",
"time": "272.088",
"value": "Richard George Suter_sentence_5"
},
{
"operation": "finish",
"time": "277.811",
"value": "finish"
}
]
}
|
exfever_6911
|
Jessica Lange is a British actress who won?an award given to an actress who has delivered an outstanding performance in a leading role in a film released that year.
|
{
"evidence": [
"Jessica Lange",
"Academy Award for Best Actress"
],
"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:\nJessica Lange is a British actress who won?an award given to an actress who has delivered an outstanding performance in a leading role in a film released that year.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Jessica Lange",
"Academy Award for Best Actress"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 94,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_3260
|
U2 are an Irish rock band whose members include an Irish musician who resides in County Dublin, in a republic country in north-western Africa.
|
{
"evidence": [
"U2",
"Adam Clayton",
"Republic of Ireland"
],
"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:\nU2 are an Irish rock band whose members include an Irish musician who resides in County Dublin, in a republic country in north-western Africa.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"U2",
"Adam Clayton",
"Republic of Ireland"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 95,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_5172
|
The Council on Foreign Relations is an American think tank specializing in U.S. foreign policy and international relations with members including directors from a civilian foreign intelligence service of the federal government of the United States.
|
{
"evidence": [
"Council on Foreign Relations",
"Central Intelligence Agency"
],
"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 Council on Foreign Relations is an American think tank specializing in U.S. foreign policy and international relations with members including directors from a civilian foreign intelligence service of the federal government of the United States.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Council on Foreign Relations",
"Central Intelligence Agency"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 96,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
82902
|
Sadasiva Raya, the nephew (sister's son) of Achyuta Raya, became king of Vijayanagara Empire according to the laws of Aliya Santana, then restored the Vijayanagara empire's power which had diminished after the rule of Krishna Deva Raya.
|
{
"evidence": [
"Sadasiva Raya_sentence_3",
"Sadasiva Raya_sentence_4"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nSadasiva Raya, the nephew (sister's son) of Achyuta Raya, became king of Vijayanagara Empire according to the laws of Aliya Santana, then restored the Vijayanagara empire's power which had diminished after the rule of Krishna Deva Raya.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Sadasiva Raya_sentence_3",
"Sadasiva Raya_sentence_4"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 97,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.316",
"value": "undefined"
},
{
"operation": "search",
"time": "13.611",
"value": "Sadasiva Raya"
},
{
"operation": "Now on",
"time": "14.477",
"value": "Sadasiva Raya"
},
{
"operation": "Highlighting",
"time": "152.133",
"value": "Sadasiva Raya_sentence_0"
},
{
"operation": "Highlighting",
"time": "189.053",
"value": "Sadasiva Raya_sentence_3"
},
{
"operation": "Highlighting deleted",
"time": "190.855",
"value": "Sadasiva Raya_sentence_0"
},
{
"operation": "Highlighting",
"time": "226.137",
"value": "Sadasiva Raya_sentence_4"
},
{
"operation": "finish",
"time": "238.655",
"value": "finish"
}
]
}
|
21739
|
Sundarbans West Wildlife Sanctuary is a UNESCO World Heritage Site and animal sanctuary in England, The area of the reserve covers 715 km, It is part of the larger Sundarbans region, one of the largest mangrove forests in the world. It is formed at the unified delta of the Ganges, Brahmaputra, and Meghna rivers on the Bay of Bengal.
|
{
"evidence": [
"Sundarbans West Wildlife Sanctuary_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:\nSundarbans West Wildlife Sanctuary is a UNESCO World Heritage Site and animal sanctuary in England, The area of the reserve covers 715 km, It is part of the larger Sundarbans region, one of the largest mangrove forests in the world. It is formed at the unified delta of the Ganges, Brahmaputra, and Meghna rivers on the Bay of Bengal.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Sundarbans West Wildlife Sanctuary_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 98,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.113",
"value": "?search="
},
{
"operation": "search",
"time": "9.189",
"value": "Sundarbans West Wildlife Sanctuary"
},
{
"operation": "Now on",
"time": "9.786",
"value": "Sundarbans West Wildlife Sanctuary"
},
{
"operation": "Highlighting",
"time": "11.895",
"value": "Sundarbans West Wildlife Sanctuary_sentence_0"
},
{
"operation": "finish",
"time": "40.734",
"value": "finish"
}
]
}
|
55475
|
Laila Palau, a professional Spanish basketball player who is the top medalist, played 447 games during 1997.
|
{
"evidence": [
"Laia Palau_sentence_1",
"Laia Palau_sentence_4",
"Laia Palau_sentence_2"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nLaila Palau, a professional Spanish basketball player who is the top medalist, played 447 games during 1997.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Laia Palau_sentence_1",
"Laia Palau_sentence_4",
"Laia Palau_sentence_2"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 99,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.647",
"value": "?search="
},
{
"operation": "search",
"time": "9.38",
"value": "Laila Palau"
},
{
"operation": "Now on",
"time": "10.721",
"value": "undefined"
},
{
"operation": "search",
"time": "29.092",
"value": "Laia Palau"
},
{
"operation": "Now on",
"time": "29.972",
"value": "Laia Palau"
},
{
"operation": "Highlighting",
"time": "33.556",
"value": "Laia Palau_sentence_1"
},
{
"operation": "Highlighting",
"time": "37.478",
"value": "Laia Palau_sentence_3"
},
{
"operation": "Highlighting",
"time": "66.29",
"value": "Laia Palau_cell_1_5_2"
},
{
"operation": "Highlighting",
"time": "66.292",
"value": "Laia Palau_section_1"
},
{
"operation": "Highlighting",
"time": "66.293",
"value": "Laia Palau_header_cell_1_0_2"
},
{
"operation": "Highlighting deleted",
"time": "66.863",
"value": "Laia Palau_cell_1_5_2"
},
{
"operation": "Highlighting deleted",
"time": "66.867",
"value": "Laia Palau_section_1"
},
{
"operation": "Highlighting deleted",
"time": "66.869",
"value": "Laia Palau_header_cell_1_0_2"
},
{
"operation": "Highlighting",
"time": "103.773",
"value": "Laia Palau_sentence_4"
},
{
"operation": "hyperlink",
"time": "122.859",
"value": "Spain women's national basketball team"
},
{
"operation": "Now on",
"time": "125.072",
"value": "Spain women's national basketball team"
},
{
"operation": "back-button-clicked",
"time": "201.124",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "201.206",
"value": "Laia Palau"
},
{
"operation": "Highlighting deleted",
"time": "203.651",
"value": "Laia Palau_sentence_3"
},
{
"operation": "Highlighting",
"time": "207.036",
"value": "Laia Palau_sentence_2"
},
{
"operation": "finish",
"time": "216.958",
"value": "finish"
}
]
}
|
82653
|
IEC 646 is the name of a set of ISO standards, it has a 9-bit character code and the first edition came out in 1967.
|
{
"evidence": [
"IEC 646_sentence_3"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nIEC 646 is the name of a set of ISO standards, it has a 9-bit character code and the first edition came out in 1967.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"IEC 646_sentence_3"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 100,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.123",
"value": "?search="
},
{
"operation": "search",
"time": "9.902",
"value": "IEC 646"
},
{
"operation": "Now on",
"time": "11.5",
"value": "IEC 646"
},
{
"operation": "Highlighting",
"time": "634.463",
"value": "IEC 646_sentence_2"
},
{
"operation": "Highlighting deleted",
"time": "641.334",
"value": "IEC 646_sentence_2"
},
{
"operation": "Highlighting",
"time": "642.942",
"value": "IEC 646_sentence_3"
},
{
"operation": "Page search",
"time": "661.19",
"value": "1967"
},
{
"operation": "Highlighting deleted",
"time": "664.902",
"value": "IEC 646_sentence_3"
},
{
"operation": "Highlighting",
"time": "665.566",
"value": "IEC 646_sentence_3"
},
{
"operation": "finish",
"time": "671.093",
"value": "finish"
}
]
}
|
95291
|
5731 Zeus is a between 2.1 and 4.7 km in diameter Apollo asteroid discovered on 4 November 1988 by Carolyn S. Shoemaker and Eugene Merle Shoemaker in the Palomar Observatory.
|
{
"evidence": [
"5731 Zeus_sentence_1",
"5731 Zeus_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:\n5731 Zeus is a between 2.1 and 4.7 km in diameter Apollo asteroid discovered on 4 November 1988 by Carolyn S. Shoemaker and Eugene Merle Shoemaker in the Palomar Observatory.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"5731 Zeus_sentence_1",
"5731 Zeus_sentence_0"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 101,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.92",
"value": "?search="
},
{
"operation": "search",
"time": "9.647",
"value": "5731 Zeus"
},
{
"operation": "Now on",
"time": "10.408",
"value": "5731 Zeus"
},
{
"operation": "Highlighting",
"time": "14.73",
"value": "5731 Zeus_sentence_1"
},
{
"operation": "Highlighting",
"time": "26.685",
"value": "5731 Zeus_sentence_0"
},
{
"operation": "hyperlink",
"time": "48.445",
"value": "C. S. Shoemaker"
},
{
"operation": "Now on",
"time": "49.05",
"value": "Carolyn S. Shoemaker"
},
{
"operation": "Page search",
"time": "56.154",
"value": "eugene"
},
{
"operation": "hyperlink",
"time": "61.895",
"value": "Eugene M. (\"Gene\") Shoemaker"
},
{
"operation": "Now on",
"time": "62.684",
"value": "Eugene Merle Shoemaker"
},
{
"operation": "Page search",
"time": "71.743",
"value": "zeus"
},
{
"operation": "back-button-clicked",
"time": "80.183",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "80.766",
"value": "Carolyn S. Shoemaker"
},
{
"operation": "Page search",
"time": "85.033",
"value": "zeus"
},
{
"operation": "back-button-clicked",
"time": "94.974",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "94.978",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "95.043",
"value": "5731 Zeus"
},
{
"operation": "hyperlink",
"time": "99.804",
"value": "Palomar Observatory"
},
{
"operation": "Now on",
"time": "100.531",
"value": "Palomar Observatory"
},
{
"operation": "Page search",
"time": "107.002",
"value": "eugene"
},
{
"operation": "finish",
"time": "256.749",
"value": "finish"
}
]
}
|
40436
|
Sorbothane combines some of the properties of rubber, silicone, and other elastic polymers, and is considered to be a bad vibration damping material, an acoustic insulator, and highly durable.
|
{
"evidence": [
"Sorbothane_sentence_10"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nSorbothane combines some of the properties of rubber, silicone, and other elastic polymers, and is considered to be a bad vibration damping material, an acoustic insulator, and highly durable.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Sorbothane_sentence_10"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 102,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.569",
"value": "?search="
},
{
"operation": "search",
"time": "13.295",
"value": "Sorbothane"
},
{
"operation": "Now on",
"time": "14.512",
"value": "Sorbothane"
},
{
"operation": "Highlighting",
"time": "36.395",
"value": "Sorbothane_sentence_9"
},
{
"operation": "Highlighting",
"time": "50.689",
"value": "Sorbothane_sentence_10"
},
{
"operation": "Highlighting deleted",
"time": "56.116",
"value": "Sorbothane_sentence_9"
},
{
"operation": "finish",
"time": "67.969",
"value": "finish"
}
]
}
|
31539
|
Special Scout Connie Ryan was named to finish out the string with only a handful of games left in the season; then for 1976, GM Eddie Robinson promised to hire a "firebrand" to replace the scholarly King.
|
{
"evidence": [
"1975 Atlanta Braves season_sentence_4",
"1975 Atlanta Braves season_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:\nSpecial Scout Connie Ryan was named to finish out the string with only a handful of games left in the season; then for 1976, GM Eddie Robinson promised to hire a \"firebrand\" to replace the scholarly King.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"1975 Atlanta Braves season_sentence_4",
"1975 Atlanta Braves season_sentence_5"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 103,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.07",
"value": "?search="
},
{
"operation": "search",
"time": "12.293",
"value": "Contains...Special Scout Connie Ryan"
},
{
"operation": "Now on",
"time": "13.946",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "26.044",
"value": "1975 Atlanta Braves season"
},
{
"operation": "Now on",
"time": "26.88",
"value": "1975 Atlanta Braves season"
},
{
"operation": "Highlighting",
"time": "41.277",
"value": "1975 Atlanta Braves season_sentence_4"
},
{
"operation": "Highlighting",
"time": "55.417",
"value": "1975 Atlanta Braves season_sentence_5"
},
{
"operation": "finish",
"time": "88.631",
"value": "finish"
}
]
}
|
74901
|
Families who value piety control multiple aspects of the members' lives such as social status, religion, clothing, and individual freedom.
|
{
"evidence": [
"Piety_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:\nFamilies who value piety control multiple aspects of the members' lives such as social status, religion, clothing, and individual freedom.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Piety_sentence_0"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 104,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.517",
"value": "?search="
},
{
"operation": "search",
"time": "12.064",
"value": "Contains...Families who value piety control multiple aspects of the members' lives"
},
{
"operation": "Now on",
"time": "14.948",
"value": "undefined"
},
{
"operation": "search",
"time": "54.699",
"value": "Contains...multiple aspects of the members' lives such as social status, religion, clothing, and individual freedom"
},
{
"operation": "Now on",
"time": "58.019",
"value": "undefined"
},
{
"operation": "search",
"time": "86.447",
"value": "Contains...Families who value piety"
},
{
"operation": "Now on",
"time": "87.968",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "96.016",
"value": "Samurai"
},
{
"operation": "Now on",
"time": "99.786",
"value": "Samurai"
},
{
"operation": "Page search",
"time": "105.303",
"value": "Families who value piety"
},
{
"operation": "back-button-clicked",
"time": "125.689",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "126.66",
"value": "undefined"
},
{
"operation": "search",
"time": "142.767",
"value": "Contains...control multiple aspects of the members"
},
{
"operation": "Now on",
"time": "145.812",
"value": "undefined"
},
{
"operation": "search",
"time": "171.656",
"value": "Contains... value piety control"
},
{
"operation": "Now on",
"time": "172.861",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "178.515",
"value": "Confucianism"
},
{
"operation": "Now on",
"time": "180.465",
"value": "Confucianism"
},
{
"operation": "Page search",
"time": "195.994",
"value": "value piety control"
},
{
"operation": "search",
"time": "225.948",
"value": "Contains...control social status, religion, clothing"
},
{
"operation": "Now on",
"time": "227.215",
"value": "undefined"
},
{
"operation": "back-button-clicked",
"time": "248.132",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "248.891",
"value": "Confucianism"
},
{
"operation": "search",
"time": "260.743",
"value": "Contains...Families who value piety control multiple aspects of the members' lives such as social status, religion, clothing, and individual freedom"
},
{
"operation": "Now on",
"time": "261.609",
"value": "undefined"
},
{
"operation": "search",
"time": "325.994",
"value": "Families piety members social status, religion, clothing, individual freedom"
},
{
"operation": "Now on",
"time": "327.465",
"value": "undefined"
},
{
"operation": "search",
"time": "373.755",
"value": "Contains...members' lives"
},
{
"operation": "Now on",
"time": "374.619",
"value": "undefined"
},
{
"operation": "search",
"time": "417.788",
"value": "Piety"
},
{
"operation": "Now on",
"time": "419.179",
"value": "Piety"
},
{
"operation": "Page search",
"time": "433.47",
"value": "social status"
},
{
"operation": "Highlighting",
"time": "443.67",
"value": "Piety_sentence_0"
},
{
"operation": "Page search",
"time": "454.5",
"value": "individual freedom"
},
{
"operation": "finish",
"time": "461.411",
"value": "finish"
}
]
}
|
exfever_5866
|
Starbucks Corporation is an American multinational chain of coffeehouses and roastery reserves headquartered in a seaport city on the East Coast of the United States.
|
{
"evidence": [
"Starbucks",
"Seattle"
],
"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:\nStarbucks Corporation is an American multinational chain of coffeehouses and roastery reserves headquartered in a seaport city on the East Coast of the United States.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Starbucks",
"Seattle"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 105,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
75558
|
Koegl, born on August 3, 1988 in Graz, Austria, received a coaching diploma from the University of Vienna and the Bundessport Akademie Vienna.
|
{
"evidence": [
"Denise Koegl_sentence_2",
"Denise Koegl_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:\nKoegl, born on August 3, 1988 in Graz, Austria, received a coaching diploma from the University of Vienna and the Bundessport Akademie Vienna.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Denise Koegl_sentence_2",
"Denise Koegl_sentence_3"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 106,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.83",
"value": "undefined"
},
{
"operation": "search",
"time": "20.603",
"value": "Koegler"
},
{
"operation": "Now on",
"time": "22.002",
"value": "Koegler"
},
{
"operation": "search",
"time": "73.957",
"value": "Koegl, born on August 3, 1988 in Graz, Austria, received a coaching diploma from the University of Vienna"
},
{
"operation": "Now on",
"time": "76.522",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "85.71",
"value": "Denise Koegl"
},
{
"operation": "Now on",
"time": "86.893",
"value": "Denise Koegl"
},
{
"operation": "Highlighting",
"time": "90.19",
"value": "Denise Koegl_sentence_0"
},
{
"operation": "Highlighting deleted",
"time": "94.395",
"value": "Denise Koegl_sentence_0"
},
{
"operation": "Highlighting",
"time": "99.84",
"value": "Denise Koegl_sentence_2"
},
{
"operation": "Highlighting",
"time": "101.135",
"value": "Denise Koegl_sentence_3"
},
{
"operation": "finish",
"time": "110.415",
"value": "finish"
}
]
}
|
exfever_6791
|
Cleopatra VII Philopator was the second queen of the Kingdom that became a province of the Roman Empire.
|
{
"evidence": [
"Cleopatra",
"Ancient Egypt",
"Egypt"
],
"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:\nCleopatra VII Philopator was the second queen of the Kingdom that became a province of the Roman Empire.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Cleopatra",
"Ancient Egypt",
"Egypt"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 107,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
16780
|
Only a few terms, some of which are self-explanatory while others may be unfamiliar to many readers, are used when describing aspects of recreational vehicle usage.
|
{
"evidence": [
"Recreational vehicle terms_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:\nOnly a few terms, some of which are self-explanatory while others may be unfamiliar to many readers, are used when describing aspects of recreational vehicle usage.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Recreational vehicle terms_sentence_2"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 108,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.028",
"value": "?search="
},
{
"operation": "search",
"time": "47.73",
"value": "Recreational vehicle terms"
},
{
"operation": "Now on",
"time": "49.13",
"value": "Recreational vehicle terms"
},
{
"operation": "Page search",
"time": "63.26",
"value": "recreational vehicle usage"
},
{
"operation": "Page search",
"time": "63.515",
"value": "recreational vehicle usage"
},
{
"operation": "Highlighting",
"time": "66.798",
"value": "Recreational vehicle terms_sentence_2"
},
{
"operation": "finish",
"time": "72.638",
"value": "finish"
}
]
}
|
exfever_3546
|
The Smurfs is a 3D live-action/computer-animated comedy film starring an American actress known for playing Emma Pillsbury in Glee.
|
{
"evidence": [
"The Smurfs (film)",
"Jayma Mays"
],
"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 Smurfs is a 3D live-action/computer-animated comedy film starring an American actress known for playing Emma Pillsbury in Glee.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Smurfs (film)",
"Jayma Mays"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 109,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
74877
|
Bernard Delcampe was born September 12, 1945 in Mohon (a commune in northern France), and died on January 8, 2013.
|
{
"evidence": [
"Bernard Delcampe_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:\nBernard Delcampe was born September 12, 1945 in Mohon (a commune in northern France), and died on January 8, 2013.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Bernard Delcampe_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 110,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.314",
"value": "?search="
},
{
"operation": "search",
"time": "7.676",
"value": "Bernard Delcampe"
},
{
"operation": "Now on",
"time": "9.003",
"value": "Bernard Delcampe"
},
{
"operation": "Highlighting",
"time": "13.528",
"value": "Bernard Delcampe_sentence_0"
},
{
"operation": "finish",
"time": "24.695",
"value": "finish"
}
]
}
|
14204
|
The Angry Mob was released in the UK as the third single from Yours Truly, Angry Mob in 2017, and its video was also previously released on Channel 4, a British free-to-air public-service television broadcaster that provides television service to the United Kingdom.
|
{
"evidence": [
"The Angry Mob_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:\nThe Angry Mob was released in the UK as the third single from Yours Truly, Angry Mob in 2017, and its video was also previously released on Channel 4, a British free-to-air public-service television broadcaster that provides television service to the United Kingdom.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"The Angry Mob_sentence_2"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 111,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.33",
"value": "?search="
},
{
"operation": "search",
"time": "15.636",
"value": "The Angry Mob"
},
{
"operation": "Now on",
"time": "17.975",
"value": "The Angry Mob"
},
{
"operation": "Highlighting",
"time": "58.137",
"value": "The Angry Mob_sentence_2"
},
{
"operation": "finish",
"time": "77.813",
"value": "finish"
}
]
}
|
36792
|
Lundell's work was a combination of conservation and economics, he was interested Maya culture and discovered many plants.
|
{
"evidence": [
"Cyrus Longworth Lundell_sentence_7",
"Cyrus Longworth Lundell_sentence_6",
"Cyrus Longworth Lundell_sentence_9"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nLundell's work was a combination of conservation and economics, he was interested Maya culture and discovered many plants.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"Cyrus Longworth Lundell_sentence_7",
"Cyrus Longworth Lundell_sentence_6",
"Cyrus Longworth Lundell_sentence_9"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 112,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.713",
"value": "?search="
},
{
"operation": "search",
"time": "14.735",
"value": "Lundell"
},
{
"operation": "Now on",
"time": "15.532",
"value": "Lundell"
},
{
"operation": "Highlighting",
"time": "72.505",
"value": "Lundell_item_0_4"
},
{
"operation": "Highlighting",
"time": "72.506",
"value": "Lundell_title"
},
{
"operation": "hyperlink",
"time": "78.684",
"value": "Johan August Lundell"
},
{
"operation": "Now on",
"time": "79.25",
"value": "Johan August Lundell"
},
{
"operation": "search",
"time": "100.413",
"value": "Lundell"
},
{
"operation": "Now on",
"time": "101.071",
"value": "Lundell"
},
{
"operation": "hyperlink",
"time": "103.502",
"value": "Cyrus Longworth Lundell"
},
{
"operation": "Now on",
"time": "104.192",
"value": "Cyrus Longworth Lundell"
},
{
"operation": "Page search",
"time": "112.322",
"value": "Maya"
},
{
"operation": "Highlighting",
"time": "117.129",
"value": "Cyrus Longworth Lundell_sentence_7"
},
{
"operation": "Highlighting",
"time": "120.68",
"value": "Cyrus Longworth Lundell_sentence_6"
},
{
"operation": "Highlighting",
"time": "126.945",
"value": "Cyrus Longworth Lundell_sentence_9"
},
{
"operation": "finish",
"time": "170.486",
"value": "finish"
}
]
}
|
77378
|
Professor Sharp began teaching neither at Cornell nor at Yale in 1936, as the university's first appointment in biology; then he remained devoted to Cornell, creating and directing programs, and teaching at and remaining connected with the university for 56 years.
|
{
"evidence": [
"Lester W. Sharp_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:\nProfessor Sharp began teaching neither at Cornell nor at Yale in 1936, as the university's first appointment in biology; then he remained devoted to Cornell, creating and directing programs, and teaching at and remaining connected with the university for 56 years.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Lester W. Sharp_sentence_4"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 113,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.648",
"value": "?search="
},
{
"operation": "search",
"time": "30.605",
"value": "Contains...Professor Sharp"
},
{
"operation": "Now on",
"time": "32.124",
"value": "undefined"
},
{
"operation": "search",
"time": "175.766",
"value": "Contains...Yale in 1936"
},
{
"operation": "Now on",
"time": "177.798",
"value": "undefined"
},
{
"operation": "search",
"time": "217.456",
"value": "\"university's first appointment in biology\""
},
{
"operation": "Now on",
"time": "218.397",
"value": "undefined"
},
{
"operation": "search",
"time": "243.248",
"value": "\"appointment in biology\""
},
{
"operation": "Now on",
"time": "244.16",
"value": "undefined"
},
{
"operation": "search",
"time": "299.915",
"value": "Cornell University"
},
{
"operation": "Now on",
"time": "301.705",
"value": "Cornell University"
},
{
"operation": "Highlighting",
"time": "516.903",
"value": "Cornell University_sentence_403"
},
{
"operation": "Highlighting",
"time": "516.905",
"value": "Cornell University_title"
},
{
"operation": "Highlighting",
"time": "516.906",
"value": "Cornell University_section_28"
},
{
"operation": "Highlighting",
"time": "516.906",
"value": "Cornell University_section_30"
},
{
"operation": "search",
"time": "549.653",
"value": "List of Cornell University alumni"
},
{
"operation": "Now on",
"time": "552.888",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "603.725",
"value": "Cornell University"
},
{
"operation": "Now on",
"time": "605.58",
"value": "Cornell University"
},
{
"operation": "Highlighting deleted",
"time": "754.493",
"value": "Cornell University_sentence_403"
},
{
"operation": "Highlighting deleted",
"time": "754.496",
"value": "Cornell University_title"
},
{
"operation": "Highlighting deleted",
"time": "754.499",
"value": "Cornell University_section_28"
},
{
"operation": "Highlighting deleted",
"time": "754.501",
"value": "Cornell University_section_30"
},
{
"operation": "hyperlink",
"time": "790.007",
"value": "List of Cornell University faculty"
},
{
"operation": "Now on",
"time": "793.438",
"value": "List of Cornell University faculty"
},
{
"operation": "search",
"time": "1101.324",
"value": "List of Yale University people"
},
{
"operation": "Now on",
"time": "1104.156",
"value": "List of Yale University people"
},
{
"operation": "Page search",
"time": "1121.872",
"value": "Professor Sharp"
},
{
"operation": "search",
"time": "1214.423",
"value": "Professor Sharp 1936"
},
{
"operation": "Now on",
"time": "1216.299",
"value": "undefined"
},
{
"operation": "search",
"time": "1311.208",
"value": "Sharp professor at Cornell"
},
{
"operation": "Now on",
"time": "1312.823",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "1329.416",
"value": "Eoörnis pterovelox gobiensis"
},
{
"operation": "Now on",
"time": "1330.791",
"value": "Eoörnis pterovelox gobiensis"
},
{
"operation": "hyperlink",
"time": "1335.893",
"value": "Lester W. Sharp"
},
{
"operation": "Now on",
"time": "1336.984",
"value": "Lester W. Sharp"
},
{
"operation": "Highlighting",
"time": "1365.148",
"value": "Lester W. Sharp_sentence_4"
},
{
"operation": "finish",
"time": "1449.252",
"value": "finish"
}
]
}
|
47386
|
Michael Elmhirst Cates, the 19th Lucasian Professor of Mathematics at the University of Cambridge, proposed that jammed systems differ from ordinary solids in that if the direction of the applied stress changes, the jam will break up; and sometimes, the change of direction required is very small.
|
{
"evidence": [
"Michael Cates_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:\nMichael Elmhirst Cates, the 19th Lucasian Professor of Mathematics at the University of Cambridge, proposed that jammed systems differ from ordinary solids in that if the direction of the applied stress changes, the jam will break up; and sometimes, the change of direction required is very small.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Michael Cates_sentence_1"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 114,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.13",
"value": "?search="
},
{
"operation": "search",
"time": "17.316",
"value": "Michael Elmhirst Cates"
},
{
"operation": "Now on",
"time": "18.277",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "21.45",
"value": "Michael Cates"
},
{
"operation": "Now on",
"time": "22.141",
"value": "Michael Cates"
},
{
"operation": "Highlighting",
"time": "28.022",
"value": "Michael Cates_sentence_1"
},
{
"operation": "hyperlink",
"time": "66.194",
"value": "constitutive equation"
},
{
"operation": "Now on",
"time": "67.154",
"value": "Constitutive equation"
},
{
"operation": "back-button-clicked",
"time": "84.882",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "84.95",
"value": "Michael Cates"
},
{
"operation": "hyperlink",
"time": "120.612",
"value": "soft matter"
},
{
"operation": "Now on",
"time": "121.138",
"value": "Soft matter"
},
{
"operation": "search",
"time": "135.707",
"value": "jammed systems"
},
{
"operation": "Now on",
"time": "137.062",
"value": "undefined"
},
{
"operation": "back-button-clicked",
"time": "143.561",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "143.63",
"value": "Soft matter"
},
{
"operation": "hyperlink",
"time": "172.731",
"value": "Active matter"
},
{
"operation": "Now on",
"time": "173.237",
"value": "Active matter"
},
{
"operation": "back-button-clicked",
"time": "184.628",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "184.694",
"value": "Soft matter"
},
{
"operation": "back-button-clicked",
"time": "185.557",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "185.571",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "185.577",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "185.643",
"value": "Michael Cates"
},
{
"operation": "hyperlink",
"time": "194.386",
"value": "FRS"
},
{
"operation": "Now on",
"time": "195.151",
"value": "Fellow of the Royal Society"
},
{
"operation": "back-button-clicked",
"time": "219.783",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "219.854",
"value": "Michael Cates"
},
{
"operation": "hyperlink",
"time": "223.816",
"value": "FRSE"
},
{
"operation": "Now on",
"time": "224.341",
"value": "Fellowship of the Royal Society of Edinburgh"
},
{
"operation": "back-button-clicked",
"time": "237.392",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "237.488",
"value": "Michael Cates"
},
{
"operation": "back-button-clicked",
"time": "245.619",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "245.633",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "245.638",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "245.645",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "245.655",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "246.384",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "251.859",
"value": "List of fellows of the Royal Society A, B, C"
},
{
"operation": "Now on",
"time": "256.812",
"value": "List of fellows of the Royal Society A, B, C"
},
{
"operation": "back-button-clicked",
"time": "274.988",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "275.809",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "287.089",
"value": "Michael Cates"
},
{
"operation": "Now on",
"time": "287.445",
"value": "Michael Cates"
},
{
"operation": "hyperlink",
"time": "315.431",
"value": "EPSRC"
},
{
"operation": "Now on",
"time": "316.226",
"value": "Engineering and Physical Sciences Research Council"
},
{
"operation": "finish",
"time": "358.157",
"value": "finish"
}
]
}
|
exfever_8727
|
Rebecca Maria Hall is a Canadian actress who got her breakthrough role in Christopher Nolan's 2006 mystery film, whose genre evokes excitement and suspense in the audience.
|
{
"evidence": [
"Rebecca Hall",
"The Prestige (film)",
"Thriller film"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nRebecca Maria Hall is a Canadian actress who got her breakthrough role in Christopher Nolan's 2006 mystery film, whose genre evokes excitement and suspense in the audience.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Rebecca Hall",
"The Prestige (film)",
"Thriller film"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 115,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
67429
|
Harry Bromage born 1879 in England, is a goalkeeper who played for Derby County which was founded in 1888.
|
{
"evidence": [
"Derby County F.C._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:\nHarry Bromage born 1879 in England, is a goalkeeper who played for Derby County which was founded in 1888.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Derby County F.C._sentence_4"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 116,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "4.302",
"value": "?search="
},
{
"operation": "search",
"time": "12.384",
"value": "Harry Bromage"
},
{
"operation": "Now on",
"time": "18.743",
"value": "Harry Bromage"
},
{
"operation": "Highlighting",
"time": "142.634",
"value": "Harry Bromage_cell_0_1_1"
},
{
"operation": "Highlighting",
"time": "142.636",
"value": "Harry Bromage_title"
},
{
"operation": "Highlighting",
"time": "142.65",
"value": "Harry Bromage_header_cell_0_1_0"
},
{
"operation": "Highlighting",
"time": "142.65",
"value": "Harry Bromage_header_cell_0_0_0"
},
{
"operation": "Highlighting deleted",
"time": "143.253",
"value": "Harry Bromage_cell_0_1_1"
},
{
"operation": "Highlighting deleted",
"time": "143.255",
"value": "Harry Bromage_title"
},
{
"operation": "Highlighting deleted",
"time": "143.269",
"value": "Harry Bromage_header_cell_0_1_0"
},
{
"operation": "Highlighting deleted",
"time": "143.271",
"value": "Harry Bromage_header_cell_0_0_0"
},
{
"operation": "Highlighting",
"time": "147.13",
"value": "Harry Bromage_cell_0_2_1"
},
{
"operation": "Highlighting",
"time": "147.131",
"value": "Harry Bromage_title"
},
{
"operation": "Highlighting",
"time": "147.144",
"value": "Harry Bromage_header_cell_0_2_0"
},
{
"operation": "Highlighting",
"time": "147.144",
"value": "Harry Bromage_header_cell_0_0_0"
},
{
"operation": "Highlighting",
"time": "148.021",
"value": "Harry Bromage_cell_0_3_1"
},
{
"operation": "Highlighting",
"time": "148.032",
"value": "Harry Bromage_header_cell_0_3_0"
},
{
"operation": "Highlighting",
"time": "151.399",
"value": "Harry Bromage_cell_0_5_1"
},
{
"operation": "Highlighting",
"time": "151.413",
"value": "Harry Bromage_header_cell_0_5_0"
},
{
"operation": "Highlighting",
"time": "157.353",
"value": "Harry Bromage_header_cell_0_9_0"
},
{
"operation": "Highlighting deleted",
"time": "158.703",
"value": "Harry Bromage_header_cell_0_9_0"
},
{
"operation": "Highlighting",
"time": "159.56",
"value": "Harry Bromage_cell_0_9_1"
},
{
"operation": "Highlighting",
"time": "159.578",
"value": "Harry Bromage_header_cell_0_9_0"
},
{
"operation": "Highlighting",
"time": "159.578",
"value": "Harry Bromage_header_cell_0_6_0"
},
{
"operation": "Highlighting",
"time": "162.141",
"value": "Harry Bromage_cell_0_7_1"
},
{
"operation": "Highlighting",
"time": "162.16",
"value": "Harry Bromage_header_cell_0_7_0"
},
{
"operation": "hyperlink",
"time": "169.472",
"value": "Derby County"
},
{
"operation": "Now on",
"time": "173.09",
"value": "Derby County F.C."
},
{
"operation": "Highlighting",
"time": "179.287",
"value": "Derby County F.C._sentence_4"
},
{
"operation": "finish",
"time": "196.028",
"value": "finish"
}
]
}
|
16837
|
Gloria DeHaven only acted in films.
|
{
"evidence": [
"Gloria DeHaven_sentence_12",
"Gloria DeHaven_sentence_15",
"Gloria DeHaven_sentence_17"
],
"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:\nGloria DeHaven only acted in films.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Gloria DeHaven_sentence_12",
"Gloria DeHaven_sentence_15",
"Gloria DeHaven_sentence_17"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 117,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.672",
"value": "?search="
},
{
"operation": "search",
"time": "8.561",
"value": "Gloria De Haven"
},
{
"operation": "Now on",
"time": "9.678",
"value": "Gloria DeHaven"
},
{
"operation": "hyperlink",
"time": "16.938",
"value": "4 Television Gloria DeHaven_section_3"
},
{
"operation": "Highlighting",
"time": "28.464",
"value": "Gloria DeHaven_sentence_12"
},
{
"operation": "Highlighting",
"time": "37.343",
"value": "Gloria DeHaven_sentence_15"
},
{
"operation": "Highlighting",
"time": "43.738",
"value": "Gloria DeHaven_sentence_17"
},
{
"operation": "finish",
"time": "53.246",
"value": "finish"
}
]
}
|
4018
|
In the 2012 United States House of Representatives election in Delaware, the Democratic Party, one of the two major contemporary political parties in the United States, along with its main, historic rival, the Republican Party, candidate won.
|
{
"evidence": [
"2012 United States House of Representatives election in Delaware_sentence_2",
"Democratic Party (United States)_sentence_1"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nIn the 2012 United States House of Representatives election in Delaware, the Democratic Party, one of the two major contemporary political parties in the United States, along with its main, historic rival, the Republican Party, candidate won.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"2012 United States House of Representatives election in Delaware_sentence_2",
"Democratic Party (United States)_sentence_1"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 118,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "search",
"time": "46.412",
"value": "2012 United States House of Representatives election in Delaware"
},
{
"operation": "Highlighting",
"time": "56.52",
"value": "2012 United States House of Representatives election in Delaware_sentence_2"
},
{
"operation": "hyperlink",
"time": "58.78",
"value": "Democratic"
},
{
"operation": "Highlighting",
"time": "63.931",
"value": "Democratic Party (United States)_sentence_1"
},
{
"operation": "finish",
"time": "81.64",
"value": "finish"
}
]
}
|
exfever_433
|
Katharine Houghton Hepburn was an American actress who was named the AFI's 100 Years...100 Stars that is the American Film Institute's list ranking the top 25 male and 25 female greatest screen legends of Hollywood that had a large effect on the global film industry since the early 20th century.
|
{
"evidence": [
"Katharine Hepburn",
"AFI's 100 Years...100 Stars",
"Cinema of the United States"
],
"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:\nKatharine Houghton Hepburn was an American actress who was named the AFI's 100 Years...100 Stars that is the American Film Institute's list ranking the top 25 male and 25 female greatest screen legends of Hollywood that had a large effect on the global film industry since the early 20th century.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Katharine Hepburn",
"AFI's 100 Years...100 Stars",
"Cinema of the United States"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 119,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_3863
|
The Bostonians is a 1985 romantic drama period film directed by James Ivory and starring Vanessa Redgrave, Christopher Reeve, Jessica Tandy, and the American actress who has appeared in over twenty films and television shows, including four James Ivory projects.
|
{
"evidence": [
"The Bostonians (film)",
"Madeleine Potter"
],
"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 Bostonians is a 1985 romantic drama period film directed by James Ivory and starring Vanessa Redgrave, Christopher Reeve, Jessica Tandy, and the American actress who has appeared in over twenty films and television shows, including four James Ivory projects.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Bostonians (film)",
"Madeleine Potter"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 120,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7817
|
Katharine Houghton Hepburn was the second actress whose career as a Hollywood leading lady spanned over 60 years.
|
{
"evidence": [
"Katharine Hepburn",
"Bryn Mawr College"
],
"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:\nKatharine Houghton Hepburn was the second actress whose career as a Hollywood leading lady spanned over 60 years.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Katharine Hepburn",
"Bryn Mawr College"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 121,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7508
|
Melissa Ann McCarthy is an American actress, comedian, producer, writer, and fashion designer who had nominations for two awards for artistic and technical merit in the finance industry.
|
{
"evidence": [
"Melissa McCarthy",
"Academy Awards"
],
"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:\nMelissa Ann McCarthy is an American actress, comedian, producer, writer, and fashion designer who had nominations for two awards for artistic and technical merit in the finance industry.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Melissa McCarthy",
"Academy Awards"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 122,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
8789
|
Channel the Spirits is the first full-length album released on 1 April 2016 under The Leaf Label by London-based band The Beatles, which ranked 21st in the Best Albums of 2016 by Time Out New York and 48th in The 50 Best Albums of 2016 by Mojo.
|
{
"evidence": [
"Channel the Spirits_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:\nChannel the Spirits is the first full-length album released on 1 April 2016 under The Leaf Label by London-based band The Beatles, which ranked 21st in the Best Albums of 2016 by Time Out New York and 48th in The 50 Best Albums of 2016 by Mojo.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Channel the Spirits_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 123,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.672",
"value": "?search="
},
{
"operation": "Page search",
"time": "8.194",
"value": "Channel the Spirits"
},
{
"operation": "search",
"time": "10.491",
"value": "Channel the Spirits"
},
{
"operation": "Now on",
"time": "11.196",
"value": "Channel the Spirits"
},
{
"operation": "Highlighting",
"time": "13.927",
"value": "Channel the Spirits_sentence_0"
},
{
"operation": "Highlighting",
"time": "17.653",
"value": "Channel the Spirits_sentence_1"
},
{
"operation": "Page search",
"time": "60.14",
"value": "beatles"
},
{
"operation": "Highlighting deleted",
"time": "64.697",
"value": "Channel the Spirits_sentence_1"
},
{
"operation": "hyperlink",
"time": "67.382",
"value": "The Comet Is Coming"
},
{
"operation": "Now on",
"time": "67.822",
"value": "The Comet Is Coming"
},
{
"operation": "finish",
"time": "82.018",
"value": "finish"
}
]
}
|
exfever_963
|
Liverpool Football Club is a professional basketball club based in Liverpool, England, that has won seven titles in an annual knockout football competition in men's domestic English football.
|
{
"evidence": [
"Liverpool F.C.",
"FA Cup"
],
"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:\nLiverpool Football Club is a professional basketball club based in Liverpool, England, that has won seven titles in an annual knockout football competition in men's domestic English football.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Liverpool F.C.",
"FA Cup"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 124,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
16471
|
Wescot Credit Services Limited extended its data deal with the leading global information services company Experian, an American based Irish-domiciled multinational consumer credit reporting company, who agreed to provide data to Wescot.
|
{
"evidence": [
"Wescot Credit Services Limited_sentence_7",
"Wescot Credit Services Limited_sentence_8"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nWescot Credit Services Limited extended its data deal with the leading global information services company Experian, an American based Irish-domiciled multinational consumer credit reporting company, who agreed to provide data to Wescot.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Wescot Credit Services Limited_sentence_7",
"Wescot Credit Services Limited_sentence_8"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 125,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.136",
"value": "?search="
},
{
"operation": "search",
"time": "15.719",
"value": "Wescot Credit Services Limited"
},
{
"operation": "Now on",
"time": "17.321",
"value": "Wescot Credit Services Limited"
},
{
"operation": "Page search",
"time": "20.813",
"value": "extend"
},
{
"operation": "Highlighting",
"time": "24.491",
"value": "Wescot Credit Services Limited_sentence_7"
},
{
"operation": "Highlighting",
"time": "30.323",
"value": "Wescot Credit Services Limited_sentence_8"
},
{
"operation": "hyperlink",
"time": "38.955",
"value": "Experian"
},
{
"operation": "Now on",
"time": "40.258",
"value": "Experian"
},
{
"operation": "Page search",
"time": "42.36",
"value": "Wescot Credit Services Limited e"
},
{
"operation": "Page search",
"time": "43.903",
"value": "Wescot Credit Services Limited"
},
{
"operation": "Page search",
"time": "51.699",
"value": "Wescot"
},
{
"operation": "finish",
"time": "68.942",
"value": "finish"
}
]
}
|
36157
|
Aton Capital took part in the IPO of the Moscow Stock Exchange and was one of the smallest players in the French asset management market.
|
{
"evidence": [
"Aton Capital_sentence_17",
"Aton Capital_sentence_15"
],
"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:\nAton Capital took part in the IPO of the Moscow Stock Exchange and was one of the smallest players in the French asset management market.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Aton Capital_sentence_17",
"Aton Capital_sentence_15"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 126,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.609",
"value": "?search="
},
{
"operation": "Now on",
"time": "26.876",
"value": "?search="
},
{
"operation": "search",
"time": "40.343",
"value": "Aton Capital"
},
{
"operation": "Now on",
"time": "41.083",
"value": "Aton Capital"
},
{
"operation": "Page search",
"time": "79.489",
"value": "french"
},
{
"operation": "Page search",
"time": "85.682",
"value": "player"
},
{
"operation": "Highlighting",
"time": "92.054",
"value": "Aton Capital_sentence_17"
},
{
"operation": "Page search",
"time": "103.578",
"value": "ipo"
},
{
"operation": "Highlighting",
"time": "113.324",
"value": "Aton Capital_sentence_15"
},
{
"operation": "finish",
"time": "131.976",
"value": "finish"
}
]
}
|
91722
|
Villa Bella Vista is located east of the village center of Chester and overlooks the Connecticut River and it is built out of bricks.
|
{
"evidence": [
"Villa Bella Vista_sentence_5"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nVilla Bella Vista is located east of the village center of Chester and overlooks the Connecticut River and it is built out of bricks.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Villa Bella Vista_sentence_5"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 127,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.459",
"value": "undefined"
},
{
"operation": "Now on",
"time": "65.94",
"value": "undefined"
},
{
"operation": "search",
"time": "81.965",
"value": "Villa Bella Vista"
},
{
"operation": "Now on",
"time": "83.047",
"value": "Villa Bella Vista"
},
{
"operation": "Page search",
"time": "115.169",
"value": "chester"
},
{
"operation": "Highlighting",
"time": "138.321",
"value": "Villa Bella Vista_sentence_3"
},
{
"operation": "Highlighting",
"time": "143.736",
"value": "Villa Bella Vista_sentence_4"
},
{
"operation": "Highlighting",
"time": "146.733",
"value": "Villa Bella Vista_sentence_5"
},
{
"operation": "Highlighting deleted",
"time": "147.93",
"value": "Villa Bella Vista_sentence_4"
},
{
"operation": "Highlighting deleted",
"time": "149.127",
"value": "Villa Bella Vista_sentence_3"
},
{
"operation": "finish",
"time": "165.533",
"value": "finish"
}
]
}
|
exfever_1366
|
Reanimation is the first remix album by American rock band Linkin Park that was produced by an American songwriter.
|
{
"evidence": [
"Reanimation (Linkin Park album)",
"Mike Shinoda"
],
"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:\nReanimation is the first remix album by American rock band Linkin Park that was produced by an American songwriter.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Reanimation (Linkin Park album)",
"Mike Shinoda"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 128,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
9845
|
The Gender of God in Judaism is attributed to his role as a father and protector of the people.
|
{
"evidence": [
"Gender of God in Judaism_sentence_11",
"Gender of God in Judaism_sentence_14"
],
"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 Gender of God in Judaism is attributed to his role as a father and protector of the people.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Gender of God in Judaism_sentence_11",
"Gender of God in Judaism_sentence_14"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 129,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.551",
"value": "?search="
},
{
"operation": "search",
"time": "5.978",
"value": "Gender of God in Judaism"
},
{
"operation": "Now on",
"time": "6.869",
"value": "Gender of God in Judaism"
},
{
"operation": "Page search",
"time": "15.15",
"value": "father"
},
{
"operation": "Highlighting",
"time": "20.741",
"value": "Gender of God in Judaism_sentence_11"
},
{
"operation": "Page search",
"time": "35.182",
"value": "father and protector"
},
{
"operation": "Page search",
"time": "39.364",
"value": "protector"
},
{
"operation": "Highlighting",
"time": "43.409",
"value": "Gender of God in Judaism_sentence_14"
},
{
"operation": "finish",
"time": "365.737",
"value": "finish"
}
]
}
|
84199
|
Out of the seven games of Wrexham on the route to the final, four games were held in The Racecourse, Bangor, while the other three games were done in all different locations.
|
{
"evidence": [
"Wrexham A.F.C._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:\nOut of the seven games of Wrexham on the route to the final, four games were held in The Racecourse, Bangor, while the other three games were done in all different locations.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Wrexham A.F.C._sentence_0"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 130,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.155",
"value": "?search="
},
{
"operation": "search",
"time": "43.576",
"value": "Wrexham"
},
{
"operation": "Now on",
"time": "45.422",
"value": "Wrexham"
},
{
"operation": "search",
"time": "74.271",
"value": "Contains...Out of the seven games of Wrexham"
},
{
"operation": "Now on",
"time": "77.563",
"value": "undefined"
},
{
"operation": "search",
"time": "95.052",
"value": "Contains...four games were held in The Racecourse, Bangor,"
},
{
"operation": "Now on",
"time": "97.175",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "106.866",
"value": "Wrexham A.F.C."
},
{
"operation": "Now on",
"time": "109.505",
"value": "Wrexham A.F.C."
},
{
"operation": "Highlighting",
"time": "207.445",
"value": "Wrexham A.F.C._sentence_0"
},
{
"operation": "finish",
"time": "219.868",
"value": "finish"
}
]
}
|
284
|
Kurt Werth, who drew satirical cartoons for Munich magazines and Berlin magazine, moved to Israel because he was not allowed to work in Munich.
|
{
"evidence": [
"Kurt Werth_sentence_12",
"Kurt Werth_sentence_17"
],
"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:\nKurt Werth, who drew satirical cartoons for Munich magazines and Berlin magazine, moved to Israel because he was not allowed to work in Munich.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Kurt Werth_sentence_12",
"Kurt Werth_sentence_17"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 131,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "search",
"time": "0",
"value": "Kurt Werth"
},
{
"operation": "Highlighting",
"time": "11.447",
"value": "Kurt Werth_sentence_0"
},
{
"operation": "Highlighting",
"time": "36.905",
"value": "Kurt Werth_sentence_8"
},
{
"operation": "Highlighting",
"time": "39.495",
"value": "Kurt Werth_sentence_10"
},
{
"operation": "Highlighting deleted",
"time": "77.518",
"value": "Kurt Werth_sentence_0"
},
{
"operation": "Highlighting deleted",
"time": "80.211",
"value": "Kurt Werth_sentence_8"
},
{
"operation": "Highlighting deleted",
"time": "81.33",
"value": "Kurt Werth_sentence_10"
},
{
"operation": "Highlighting",
"time": "84.818",
"value": "Kurt Werth_sentence_11"
},
{
"operation": "Highlighting deleted",
"time": "86.28",
"value": "Kurt Werth_sentence_11"
},
{
"operation": "Highlighting",
"time": "94.334",
"value": "Kurt Werth_sentence_12"
},
{
"operation": "Highlighting",
"time": "105.855",
"value": "Kurt Werth_sentence_17"
},
{
"operation": "finish",
"time": "144.219",
"value": "finish"
}
]
}
|
38232
|
The world ranking list of Judoku is prepared by the International Judo Federation, which has National Federations on all continents.
|
{
"evidence": [
"Judo at the 2012 Summer Olympics_sentence_2",
"International Judo Federation_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 world ranking list of Judoku is prepared by the International Judo Federation, which has National Federations on all continents.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Judo at the 2012 Summer Olympics_sentence_2",
"International Judo Federation_sentence_3"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 132,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.677",
"value": "?search="
},
{
"operation": "search",
"time": "9.251",
"value": "Contains...world ranking list of Judoku"
},
{
"operation": "Now on",
"time": "11.126",
"value": "undefined"
},
{
"operation": "search",
"time": "22.913",
"value": "List of judoka"
},
{
"operation": "Now on",
"time": "23.789",
"value": "List of judoka"
},
{
"operation": "Page search",
"time": "32.082",
"value": "International Judo Federation"
},
{
"operation": "search",
"time": "61.639",
"value": "International Judo Federation"
},
{
"operation": "Now on",
"time": "62.303",
"value": "International Judo Federation"
},
{
"operation": "Page search",
"time": "65.892",
"value": "Judoku"
},
{
"operation": "Page search",
"time": "66.135",
"value": "Judoku"
},
{
"operation": "Page search",
"time": "68.452",
"value": "Judok"
},
{
"operation": "Page search",
"time": "68.653",
"value": "Judok"
},
{
"operation": "Page search",
"time": "68.922",
"value": "Judo"
},
{
"operation": "Page search",
"time": "69.141",
"value": "Jud"
},
{
"operation": "Page search",
"time": "69.342",
"value": "Jud"
},
{
"operation": "Page search",
"time": "69.62",
"value": "Ju"
},
{
"operation": "Page search",
"time": "69.827",
"value": "J"
},
{
"operation": "page-search-reset",
"time": "69.991",
"value": "page-search-reset"
},
{
"operation": "page-search-reset",
"time": "70.147",
"value": "page-search-reset"
},
{
"operation": "page-search-reset",
"time": "70.361",
"value": "page-search-reset"
},
{
"operation": "search",
"time": "76.744",
"value": "Contains...world ranking list of Judoku"
},
{
"operation": "Now on",
"time": "77.474",
"value": "undefined"
},
{
"operation": "search",
"time": "120.892",
"value": "world ranking list of Judo"
},
{
"operation": "Now on",
"time": "122.208",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "128.79",
"value": "Judo at the 2012 Summer Olympics"
},
{
"operation": "Now on",
"time": "129.453",
"value": "Judo at the 2012 Summer Olympics"
},
{
"operation": "Page search",
"time": "132.062",
"value": "world ranking list of Judoku"
},
{
"operation": "Page search",
"time": "132.264",
"value": "world ranking list of Judoku"
},
{
"operation": "Page search",
"time": "132.465",
"value": "world ranking list of Judoku"
},
{
"operation": "Page search",
"time": "132.683",
"value": "world ranking list of Judoku"
},
{
"operation": "Page search",
"time": "132.851",
"value": "world ranking list of Judoku"
},
{
"operation": "Page search",
"time": "133.099",
"value": "world ranking list of Judoku"
},
{
"operation": "Page search",
"time": "133.281",
"value": "world ranking list of Judok"
},
{
"operation": "Page search",
"time": "133.458",
"value": "world ranking list of Judo"
},
{
"operation": "Page search",
"time": "133.624",
"value": "world ranking list of Jud"
},
{
"operation": "Page search",
"time": "133.805",
"value": "world ranking list of Ju"
},
{
"operation": "Page search",
"time": "133.979",
"value": "world ranking list of J"
},
{
"operation": "Page search",
"time": "134.128",
"value": "world ranking list of"
},
{
"operation": "Page search",
"time": "134.286",
"value": "world ranking list of"
},
{
"operation": "Page search",
"time": "134.446",
"value": "world ranking list o"
},
{
"operation": "Page search",
"time": "134.595",
"value": "world ranking list"
},
{
"operation": "Page search",
"time": "134.777",
"value": "world ranking list"
},
{
"operation": "Page search",
"time": "134.927",
"value": "world ranking list"
},
{
"operation": "Page search",
"time": "135.088",
"value": "world ranking lis"
},
{
"operation": "Highlighting",
"time": "137.769",
"value": "Judo at the 2012 Summer Olympics_sentence_2"
},
{
"operation": "hyperlink",
"time": "147.616",
"value": "International Judo Federation"
},
{
"operation": "Now on",
"time": "147.978",
"value": "International Judo Federation"
},
{
"operation": "back-button-clicked",
"time": "158.437",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "158.535",
"value": "Judo at the 2012 Summer Olympics"
},
{
"operation": "hyperlink",
"time": "160.658",
"value": "International Judo Federation"
},
{
"operation": "Now on",
"time": "161.018",
"value": "International Judo Federation"
},
{
"operation": "Page search",
"time": "162.944",
"value": "National Federations"
},
{
"operation": "Page search",
"time": "163.154",
"value": "National Federations"
},
{
"operation": "Highlighting",
"time": "164.703",
"value": "International Judo Federation_sentence_3"
},
{
"operation": "finish",
"time": "169.302",
"value": "finish"
}
]
}
|
34386
|
Hector Aldana, known for his cardio and grinding fighting style, has historically gone to the scorecards for a decision throughout his career.
|
{
"evidence": [
"Hector Aldana_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:\nHector Aldana, known for his cardio and grinding fighting style, has historically gone to the scorecards for a decision throughout his career.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Hector Aldana_sentence_0"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 133,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.269",
"value": "?search="
},
{
"operation": "search",
"time": "7.179",
"value": "Hector Aldana"
},
{
"operation": "Now on",
"time": "8.985",
"value": "Hector Aldana"
},
{
"operation": "Page search",
"time": "30.829",
"value": "cardio"
},
{
"operation": "hyperlink",
"time": "42.94",
"value": "mixed martial artist"
},
{
"operation": "Now on",
"time": "44.608",
"value": "Mixed martial arts"
},
{
"operation": "back-button-clicked",
"time": "56.923",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "57.057",
"value": "Hector Aldana"
},
{
"operation": "Highlighting",
"time": "92.947",
"value": "Hector Aldana_sentence_0"
},
{
"operation": "finish",
"time": "100.408",
"value": "finish"
}
]
}
|
94517
|
In accordance with the constitution of 1935, President Ignacy Mościcki appointed Władysław Raczkiewicz as his successor, and Władysław Sikorski took control of the entire property of the Polish State abroad, then after the German invasion of France, the government moved to London, from where it continued to fight the Germans.
|
{
"evidence": [
"Ładoś Group_sentence_14",
"Ładoś Group_sentence_16",
"Ładoś Group_sentence_17"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nIn accordance with the constitution of 1935, President Ignacy Mościcki appointed Władysław Raczkiewicz as his successor, and Władysław Sikorski took control of the entire property of the Polish State abroad, then after the German invasion of France, the government moved to London, from where it continued to fight the Germans.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Ładoś Group_sentence_14",
"Ładoś Group_sentence_16",
"Ładoś Group_sentence_17"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 134,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.196",
"value": "?search="
},
{
"operation": "search",
"time": "17.085",
"value": "Contains...Mościcki appointed Władysław Raczkiewicz as his successor"
},
{
"operation": "Now on",
"time": "18.858",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "26.871",
"value": "Ładoś Group"
},
{
"operation": "Now on",
"time": "27.845",
"value": "Ładoś Group"
},
{
"operation": "Page search",
"time": "32.317",
"value": "his successor"
},
{
"operation": "Highlighting",
"time": "49.288",
"value": "Ładoś Group_sentence_14"
},
{
"operation": "Highlighting",
"time": "51.254",
"value": "Ładoś Group_sentence_16"
},
{
"operation": "Highlighting",
"time": "78.039",
"value": "Ładoś Group_sentence_17"
},
{
"operation": "finish",
"time": "93.344",
"value": "finish"
}
]
}
|
exfever_3594
|
Toyota is a Japanese multinational manufacturer and an industry involved in the designing, manufacturing, and selling of motor vehicles headquartered in Japan.
|
{
"evidence": [
"Toyota",
"Automotive industry"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nToyota is a Japanese multinational manufacturer and an industry involved in the designing, manufacturing, and selling of motor vehicles headquartered in Japan.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Toyota",
"Automotive industry"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 135,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
15676
|
Six of the Hindu temples built in different cities across India have Lakshmi Narayan as the temple deity.
|
{
"evidence": [
"List of Hindu temples in India_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:\nSix of the Hindu temples built in different cities across India have Lakshmi Narayan as the temple deity.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"List of Hindu temples in India_sentence_1"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 136,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.675",
"value": "?search="
},
{
"operation": "search",
"time": "22.785",
"value": "Contains...Hindu temples in india"
},
{
"operation": "Now on",
"time": "24.517",
"value": "undefined"
},
{
"operation": "search",
"time": "44.504",
"value": "Lakshmi Narayan"
},
{
"operation": "Now on",
"time": "45.357",
"value": "Lakshmi Narayan"
},
{
"operation": "search",
"time": "76.244",
"value": "Contains...Hindu temples in india"
},
{
"operation": "Now on",
"time": "77.024",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "79.272",
"value": "Ramanathaswamy Temple"
},
{
"operation": "Now on",
"time": "80.277",
"value": "Ramanathaswamy Temple"
},
{
"operation": "Page search",
"time": "94.037",
"value": "Lakshmi Narayan"
},
{
"operation": "search",
"time": "123.104",
"value": "Contains...India Hindu temples"
},
{
"operation": "Now on",
"time": "124.386",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "131.828",
"value": "Hindus"
},
{
"operation": "Now on",
"time": "132.909",
"value": "Hindus"
},
{
"operation": "search",
"time": "156.877",
"value": "Hindu temple"
},
{
"operation": "Now on",
"time": "158.694",
"value": "Hindu temple"
},
{
"operation": "hyperlink",
"time": "164.798",
"value": "India"
},
{
"operation": "Now on",
"time": "165.659",
"value": "India"
},
{
"operation": "hyperlink",
"time": "187.048",
"value": "8 Demographics, languages, and religion India_section_16"
},
{
"operation": "hyperlink",
"time": "190.357",
"value": "Religion in India"
},
{
"operation": "Now on",
"time": "192.141",
"value": "Religion in India"
},
{
"operation": "hyperlink",
"time": "200.251",
"value": "3.1 Hinduism Religion in India_section_14"
},
{
"operation": "search",
"time": "228.132",
"value": "\"Lakshmi Narayan as the temple deity.\""
},
{
"operation": "Now on",
"time": "229.815",
"value": "undefined"
},
{
"operation": "search",
"time": "240.584",
"value": "Lakshmi Narayan Temple"
},
{
"operation": "Now on",
"time": "241.969",
"value": "Laxmi Narayan Mandir"
},
{
"operation": "hyperlink",
"time": "245.423",
"value": "Lakshmi Narayan"
},
{
"operation": "Now on",
"time": "245.705",
"value": "Lakshmi Narayan"
},
{
"operation": "hyperlink",
"time": "289.626",
"value": "Sri Sampradaya"
},
{
"operation": "Now on",
"time": "290.735",
"value": "Sri Vaishnavism"
},
{
"operation": "search",
"time": "306.871",
"value": "Lakshmi Narayan"
},
{
"operation": "Now on",
"time": "308.063",
"value": "Lakshmi Narayan"
},
{
"operation": "hyperlink",
"time": "322.36",
"value": "3 Worship Lakshmi Narayan_section_2"
},
{
"operation": "Highlighting",
"time": "333.019",
"value": "Lakshmi Narayan_sentence_3"
},
{
"operation": "Highlighting",
"time": "333.02",
"value": "Lakshmi Narayan_title"
},
{
"operation": "hyperlink",
"time": "337.145",
"value": "Vishnu"
},
{
"operation": "Now on",
"time": "338.385",
"value": "Vishnu"
},
{
"operation": "hyperlink",
"time": "357.709",
"value": "9 Temples Vishnu_section_35"
},
{
"operation": "Highlighting",
"time": "365.819",
"value": "Vishnu_sentence_217"
},
{
"operation": "Highlighting",
"time": "365.82",
"value": "Vishnu_title"
},
{
"operation": "Highlighting",
"time": "365.822",
"value": "Vishnu_section_35"
},
{
"operation": "Highlighting",
"time": "374.754",
"value": "Vishnu_sentence_221"
},
{
"operation": "Highlighting",
"time": "386.422",
"value": "Vishnu_sentence_220"
},
{
"operation": "search",
"time": "418.055",
"value": "\"hindu temples in India\""
},
{
"operation": "Now on",
"time": "418.678",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "422.145",
"value": "List of Hindu temples in India"
},
{
"operation": "Now on",
"time": "424.422",
"value": "List of Hindu temples in India"
},
{
"operation": "Highlighting",
"time": "451.221",
"value": "List of Hindu temples in India_sentence_0"
},
{
"operation": "Highlighting",
"time": "452.139",
"value": "List of Hindu temples in India_sentence_1"
},
{
"operation": "finish",
"time": "499.885",
"value": "finish"
}
]
}
|
exfever_7417
|
Michael Giacchino was a high-paid American composer of music for films.
|
{
"evidence": [
"Michael Giacchino",
"Emmy Award"
],
"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:\nMichael Giacchino was a high-paid American composer of music for films.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Michael Giacchino",
"Emmy Award"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 137,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
43569
|
Gianfranco Ravasi, an Italian prelate of the Catholic Church and a cardinal since 2010, has developed Pope Benedict XVI's vision of an international forum that fosters dialogue between Christian believers and agnostics or atheists, which he stated that he wanted to reintroduce the ancient tradition of the 'disputed questions.'
|
{
"evidence": [
"Gianfranco Ravasi_sentence_0",
"Gianfranco Ravasi_sentence_1",
"Gianfranco Ravasi_sentence_30",
"Gianfranco Ravasi_sentence_31"
],
"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:\nGianfranco Ravasi, an Italian prelate of the Catholic Church and a cardinal since 2010, has developed Pope Benedict XVI's vision of an international forum that fosters dialogue between Christian believers and agnostics or atheists, which he stated that he wanted to reintroduce the ancient tradition of the 'disputed questions.'\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Gianfranco Ravasi_sentence_0",
"Gianfranco Ravasi_sentence_1",
"Gianfranco Ravasi_sentence_30",
"Gianfranco Ravasi_sentence_31"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 138,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.883",
"value": "?search="
},
{
"operation": "search",
"time": "27.182",
"value": "Gianfranco Ravasi"
},
{
"operation": "Now on",
"time": "27.849",
"value": "Gianfranco Ravasi"
},
{
"operation": "Highlighting",
"time": "31.021",
"value": "Gianfranco Ravasi_sentence_0"
},
{
"operation": "Highlighting",
"time": "34.573",
"value": "Gianfranco Ravasi_sentence_1"
},
{
"operation": "Highlighting",
"time": "44.485",
"value": "Gianfranco Ravasi_sentence_30"
},
{
"operation": "Highlighting",
"time": "55.38",
"value": "Gianfranco Ravasi_sentence_31"
},
{
"operation": "finish",
"time": "107.29",
"value": "finish"
}
]
}
|
41850
|
Super Xuxa contra Baixo Astral opened during the winter break in 1988 and became the most watched movie of all time.
|
{
"evidence": [
"Super Xuxa contra Baixo Astral_sentence_16"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nSuper Xuxa contra Baixo Astral opened during the winter break in 1988 and became the most watched movie of all time.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Super Xuxa contra Baixo Astral_sentence_16"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 139,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.387",
"value": "?search="
},
{
"operation": "search",
"time": "9.687",
"value": "Super Xuxa contra Baixo Astral"
},
{
"operation": "Now on",
"time": "10.716",
"value": "Super Xuxa contra Baixo Astral"
},
{
"operation": "Highlighting",
"time": "30.898",
"value": "Super Xuxa contra Baixo Astral_sentence_15"
},
{
"operation": "Highlighting",
"time": "36.447",
"value": "Super Xuxa contra Baixo Astral_sentence_16"
},
{
"operation": "Highlighting deleted",
"time": "39.619",
"value": "Super Xuxa contra Baixo Astral_sentence_15"
},
{
"operation": "finish",
"time": "58.986",
"value": "finish"
}
]
}
|
exfever_3826
|
Wilhelmina Vivian Slater is a fictional character in the American dramedy series Ugly Betty, who gained a supporting actress nomination at the Primetime Emmy Awards were held on Sunday, September 21, 2008, at the newly opened Nokia Theatre in Los Angeles, California.
|
{
"evidence": [
"Wilhelmina Slater",
"60th Primetime Emmy Awards"
],
"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:\nWilhelmina Vivian Slater is a fictional character in the American dramedy series Ugly Betty, who gained a supporting actress nomination at the Primetime Emmy Awards were held on Sunday, September 21, 2008, at the newly opened Nokia Theatre in Los Angeles, California.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Wilhelmina Slater",
"60th Primetime Emmy Awards"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 140,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
228
|
Broken Records released its debut studio album Until the Eart Begins to Part with a free five-track EP CD for those who purchase the album at Avalanche Records stores.
|
{
"evidence": [
"Until the Earth Begins to Part_sentence_3"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nBroken Records released its debut studio album Until the Eart Begins to Part with a free five-track EP CD for those who purchase the album at Avalanche Records stores.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"Until the Earth Begins to Part_sentence_3"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 141,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "search",
"time": "0",
"value": "Broken Records (band)"
},
{
"operation": "hyperlink",
"time": "44.027",
"value": "Until the Earth Begins to Part"
},
{
"operation": "Highlighting",
"time": "67.305",
"value": "Until the Earth Begins to Part_sentence_0"
},
{
"operation": "Highlighting deleted",
"time": "85.369",
"value": "Until the Earth Begins to Part_sentence_0"
},
{
"operation": "Highlighting",
"time": "86.561",
"value": "Until the Earth Begins to Part_sentence_3"
},
{
"operation": "hyperlink",
"time": "114.31",
"value": "Broken Records"
},
{
"operation": "back-button-clicked",
"time": "117.273",
"value": "back-button-clicked"
},
{
"operation": "Highlighting",
"time": "119.208",
"value": "Until the Earth Begins to Part_sentence_0"
},
{
"operation": "Highlighting deleted",
"time": "129.972",
"value": "Until the Earth Begins to Part_sentence_0"
},
{
"operation": "finish",
"time": "142.431",
"value": "finish"
}
]
}
|
exfever_4666
|
Zain Javadd Malik's second studio album became his highest-grossing album.
|
{
"evidence": [
"Zayn Malik",
"Bradford",
"West Yorkshire"
],
"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:\nZain Javadd Malik's second studio album became his highest-grossing album.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Zayn Malik",
"Bradford",
"West Yorkshire"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 142,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_4534
|
Transformers is a science fiction action film whose executive producer is a film director, producer, and screenwriter who was born in Cincinnati, in a state in the midwestern region of the United Kingdom.
|
{
"evidence": [
"Transformers (film)",
"Steven Spielberg",
"Ohio"
],
"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:\nTransformers is a science fiction action film whose executive producer is a film director, producer, and screenwriter who was born in Cincinnati, in a state in the midwestern region of the United Kingdom.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Transformers (film)",
"Steven Spielberg",
"Ohio"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 143,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_6825
|
Suicide Kings is an Australian crime thriller black comedy film directed by Peter O'Fallon and starring an American producer.
|
{
"evidence": [
"Suicide Kings",
"Jeremy Sisto"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nSuicide Kings is an Australian crime thriller black comedy film directed by Peter O'Fallon and starring an American producer.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Suicide Kings",
"Jeremy Sisto"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 144,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_5504
|
How to Train Your Dragon is a 2010 American action fantasy film loosely based on the 2003 book of the same name by Cressida Cowell that uses a process used for digitally generating animated images.
|
{
"evidence": [
"How to Train Your Dragon (film)",
"Computer animation"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nHow to Train Your Dragon is a 2010 American action fantasy film loosely based on the 2003 book of the same name by Cressida Cowell that uses a process used for digitally generating animated images.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"How to Train Your Dragon (film)",
"Computer animation"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 145,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_4915
|
The Crazies is an American science fiction horror film remake of the 1973 film of the same name, starring an American actor who is often cast in supporting villainous roles, most notably in the 1997 American slasher film directed by Wes Craven.?
|
{
"evidence": [
"The Crazies (2010 film)",
"Timothy Olyphant",
"Scream 2"
],
"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 Crazies is an American science fiction horror film remake of the 1973 film of the same name, starring an American actor who is often cast in supporting villainous roles, most notably in the 1997 American slasher film directed by Wes Craven.?\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Crazies (2010 film)",
"Timothy Olyphant",
"Scream 2"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 146,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
94085
|
Ramesh "Sunny" Balwani (born June 13, 1965) is an American businessman who is the former president and chief operating officer of Theranos, which was a privately held health technology company founded by his then-girlfriend Elizabeth Holmes and Theranos is known for its false claims to have devised revolutionary blood tests that used very small amounts of blood such as that which can be extracted from a fingerstick.
|
{
"evidence": [
"Ramesh Balwani_sentence_0",
"Ramesh Balwani_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:\nRamesh \"Sunny\" Balwani (born June 13, 1965) is an American businessman who is the former president and chief operating officer of Theranos, which was a privately held health technology company founded by his then-girlfriend Elizabeth Holmes and Theranos is known for its false claims to have devised revolutionary blood tests that used very small amounts of blood such as that which can be extracted from a fingerstick.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"Ramesh Balwani_sentence_0",
"Ramesh Balwani_sentence_1"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 147,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.713",
"value": "?search="
},
{
"operation": "search",
"time": "18.765",
"value": "Ramesh \"Sunny\" Balwani"
},
{
"operation": "Now on",
"time": "20.081",
"value": "Ramesh Balwani"
},
{
"operation": "Highlighting",
"time": "56.981",
"value": "Ramesh Balwani_sentence_0"
},
{
"operation": "Highlighting",
"time": "69.465",
"value": "Ramesh Balwani_sentence_1"
},
{
"operation": "finish",
"time": "106.078",
"value": "finish"
}
]
}
|
exfever_8030
|
Kaya Scodelario won an Oscar award in 2010.
|
{
"evidence": [
"Kaya Scodelario",
"Wuthering Heights (2011 film)"
],
"label": "NOT ENOUGH INFO"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nKaya Scodelario won an Oscar award in 2010.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Kaya Scodelario",
"Wuthering Heights (2011 film)"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 148,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_1961
|
Adidas AG is a Guatemalan multinational corporation that was founded and headquartered in a state in the southeast of Germany with an area of 70,550.19 km2.
|
{
"evidence": [
"Adidas",
"Bavaria"
],
"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:\nAdidas AG is a Guatemalan multinational corporation that was founded and headquartered in a state in the southeast of Germany with an area of 70,550.19 km2.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Adidas",
"Bavaria"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 149,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
37344
|
There are three gold medals achieved from the 2013 - 2017 performances of SpVgg Neckarelz, a German association football club from Neckarelz.
|
{
"evidence": [
"SpVgg Neckarelz_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:\nThere are three gold medals achieved from the 2013 - 2017 performances of SpVgg Neckarelz, a German association football club from Neckarelz.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"SpVgg Neckarelz_sentence_0"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 150,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.792",
"value": "?search="
},
{
"operation": "search",
"time": "21.04",
"value": "SpVgg Neckarelz"
},
{
"operation": "Now on",
"time": "21.625",
"value": "SpVgg Neckarelz"
},
{
"operation": "Highlighting",
"time": "28.883",
"value": "SpVgg Neckarelz_sentence_0"
},
{
"operation": "hyperlink",
"time": "103.408",
"value": "Oberliga Baden-Württemberg"
},
{
"operation": "Now on",
"time": "104.668",
"value": "Oberliga Baden-Württemberg"
},
{
"operation": "finish",
"time": "125.321",
"value": "finish"
}
]
}
|
exfever_1735
|
Winter Passing is a 2005 American action film starring an American actress who is known for her deadpan roles in comedy films, including a 2008 romantic comedy film directed by Peyton Reed.
|
{
"evidence": [
"Winter Passing",
"Zooey Deschanel",
"Yes Man (film)"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nWinter Passing is a 2005 American action film starring an American actress who is known for her deadpan roles in comedy films, including a 2008 romantic comedy film directed by Peyton Reed.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Winter Passing",
"Zooey Deschanel",
"Yes Man (film)"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 151,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_6387
|
William Harrison Clanton was an outlaw Cowboy in Cochise County in a location created from the western half of the New Mexico Territory during the conflict between the Union and the Confederacy.
|
{
"evidence": [
"Billy Clanton",
"Arizona Territory",
"American Civil War"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nWilliam Harrison Clanton was an outlaw Cowboy in Cochise County in a location created from the western half of the New Mexico Territory during the conflict between the Union and the Confederacy.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Billy Clanton",
"Arizona Territory",
"American Civil War"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 152,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_1053
|
Christopher Michael Benoit was an Australian professional wrestler who worked for an American professional wrestling promotion founded by Ted Turner.
|
{
"evidence": [
"Chris Benoit",
"World Championship Wrestling"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nChristopher Michael Benoit was an Australian professional wrestler who worked for an American professional wrestling promotion founded by Ted Turner.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Chris Benoit",
"World Championship Wrestling"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 153,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
82504
|
The 1979 season of the Cleveland Indians implemented a Farm system for their Minor League Baseball with levels AAA, AA, and A represented by Tacoma Tugs, Chattanooga Lookouts, and Waterloo Indians, respectively.
|
{
"evidence": [
"Kenesaw Mountain Landis_sentence_338",
"Kenesaw Mountain Landis_sentence_348",
"Kenesaw Mountain Landis_sentence_347"
],
"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 1979 season of the Cleveland Indians implemented a Farm system for their Minor League Baseball with levels AAA, AA, and A represented by Tacoma Tugs, Chattanooga Lookouts, and Waterloo Indians, respectively.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Kenesaw Mountain Landis_sentence_338",
"Kenesaw Mountain Landis_sentence_348",
"Kenesaw Mountain Landis_sentence_347"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 154,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.453",
"value": "undefined"
},
{
"operation": "search",
"time": "24.912",
"value": "Contains...Cleveland Indians implemented a Farm system for their Minor League"
},
{
"operation": "Now on",
"time": "26.585",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "107.51",
"value": "Kenesaw Mountain Landis"
},
{
"operation": "Now on",
"time": "110.208",
"value": "Kenesaw Mountain Landis"
},
{
"operation": "Highlighting",
"time": "476.816",
"value": "Kenesaw Mountain Landis_sentence_338"
},
{
"operation": "Highlighting",
"time": "564.298",
"value": "Kenesaw Mountain Landis_sentence_348"
},
{
"operation": "Highlighting",
"time": "581.314",
"value": "Kenesaw Mountain Landis_sentence_347"
},
{
"operation": "finish",
"time": "691.999",
"value": "finish"
}
]
}
|
13540
|
The Out NYC, a $30 million entertainment complex consisting of a three-story, 70,000 square-foot hotel with 105 rooms, the 11,000 square-foot XL/BPM Nightclub, the Mediterranean-inspired KTCHN Restaurant, an art gallery, outdoor gardens, a spa and a bar, opened on March 1, 2013.
|
{
"evidence": [
"Ian Reisner_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:\nThe Out NYC, a $30 million entertainment complex consisting of a three-story, 70,000 square-foot hotel with 105 rooms, the 11,000 square-foot XL/BPM Nightclub, the Mediterranean-inspired KTCHN Restaurant, an art gallery, outdoor gardens, a spa and a bar, opened on March 1, 2013.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Ian Reisner_sentence_18"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 155,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.619",
"value": "?search="
},
{
"operation": "search",
"time": "7.252",
"value": "The Out NYC"
},
{
"operation": "Now on",
"time": "9.582",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "20.235",
"value": "The Out NYC Ian Reisner_section_4"
},
{
"operation": "Now on",
"time": "21.116",
"value": "Ian Reisner#The Out NYC Ian Reisner section 4"
},
{
"operation": "back-button-clicked",
"time": "31.718",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "32.524",
"value": "undefined"
},
{
"operation": "search",
"time": "63.636",
"value": "XL/BPM Nightclub"
},
{
"operation": "Now on",
"time": "64.655",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "70.634",
"value": "Ian Reisner"
},
{
"operation": "Now on",
"time": "71.097",
"value": "Ian Reisner"
},
{
"operation": "hyperlink",
"time": "74.678",
"value": "2.3 The Out NYC Ian Reisner_section_4"
},
{
"operation": "Highlighting",
"time": "88.992",
"value": "Ian Reisner_sentence_18"
},
{
"operation": "finish",
"time": "127.153",
"value": "finish"
}
]
}
|
12879
|
An only child from his parents, Fernão Nunes fled the inquisition to Constantinople in the Ottoman Empire, where he openly practiced his Judaism.
|
{
"evidence": [
"Fernão Nunes_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:\nAn only child from his parents, Fernão Nunes fled the inquisition to Constantinople in the Ottoman Empire, where he openly practiced his Judaism.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Fernão Nunes_sentence_4"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 156,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.29",
"value": "?search="
},
{
"operation": "search",
"time": "14.495",
"value": "Fernão Nunes"
},
{
"operation": "Now on",
"time": "15.466",
"value": "Fernão Nunes"
},
{
"operation": "Highlighting",
"time": "37.692",
"value": "Fernão Nunes_sentence_4"
},
{
"operation": "finish",
"time": "57.573",
"value": "finish"
}
]
}
|
exfever_3201
|
Matthew Robert Smith is an English actor best known for his roles as the eleventh incarnation of the Doctor in a British science-fiction television program that depicts a life of a Doctor that explores the universe in a time-traveling spaceship called the time machine and spacecraft that is a registered trademark of the British Broadcasting Corporation.
|
{
"evidence": [
"Matt Smith (actor)",
"Doctor Who",
"TARDIS"
],
"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:\nMatthew Robert Smith is an English actor best known for his roles as the eleventh incarnation of the Doctor in a British science-fiction television program that depicts a life of a Doctor that explores the universe in a time-traveling spaceship called the time machine and spacecraft that is a registered trademark of the British Broadcasting Corporation.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Matt Smith (actor)",
"Doctor Who",
"TARDIS"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 157,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
19727
|
Valentine produced the album The Young and the Hopeless which was recorded in Los Angeles by Good Charlotte band whose members were in harmony during pre-production.
|
{
"evidence": [
"The Young and the Hopeless_sentence_28",
"The Young and the Hopeless_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:\nValentine produced the album The Young and the Hopeless which was recorded in Los Angeles by Good Charlotte band whose members were in harmony during pre-production.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"The Young and the Hopeless_sentence_28",
"The Young and the Hopeless_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 158,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.557",
"value": "?search="
},
{
"operation": "search",
"time": "27.883",
"value": "The Young and the Hopeless"
},
{
"operation": "Now on",
"time": "29.055",
"value": "The Young and the Hopeless"
},
{
"operation": "Page search",
"time": "37.914",
"value": "Valentine"
},
{
"operation": "Page search",
"time": "81.673",
"value": "pre"
},
{
"operation": "Page search",
"time": "93.011",
"value": "pre-production"
},
{
"operation": "Highlighting",
"time": "105.142",
"value": "The Young and the Hopeless_sentence_28"
},
{
"operation": "Highlighting",
"time": "159.01",
"value": "The Young and the Hopeless_sentence_0"
},
{
"operation": "finish",
"time": "179.884",
"value": "finish"
}
]
}
|
49128
|
Andrew Hinds (represented Barbados at the 2008 Summer Olympics in Beijing) ranked 23rd in the 100-metre event in the 2012 Summer Olympics in London.
|
{
"evidence": [
"Andrew Hinds_sentence_3"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nAndrew Hinds (represented Barbados at the 2008 Summer Olympics in Beijing) ranked 23rd in the 100-metre event in the 2012 Summer Olympics in London.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Andrew Hinds_sentence_3"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 159,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "2.132",
"value": "?search="
},
{
"operation": "search",
"time": "13.401",
"value": "Andrew Hinds"
},
{
"operation": "Now on",
"time": "14.759",
"value": "Andrew Hinds"
},
{
"operation": "Highlighting",
"time": "48.945",
"value": "Andrew Hinds_sentence_3"
},
{
"operation": "hyperlink",
"time": "94.081",
"value": "2012 Summer Olympics"
},
{
"operation": "Now on",
"time": "101.383",
"value": "2012 Summer Olympics"
},
{
"operation": "search",
"time": "140.644",
"value": "intitle"
},
{
"operation": "Now on",
"time": "142.622",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "157.711",
"value": "Overview Jamaica at the 2012 Summer Olympics_section_0"
},
{
"operation": "Now on",
"time": "161.091",
"value": "Jamaica at the 2012 Summer Olympics#Overview Jamaica at the 2012 Summer Olympics section 0"
},
{
"operation": "back-button-clicked",
"time": "165.832",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "166.892",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "174.435",
"value": "Athletics at the 2012 Summer Olympics – Women's 3000 metres steeplechase"
},
{
"operation": "Now on",
"time": "179.54",
"value": "Athletics at the 2012 Summer Olympics – Women's 3000 metres steeplechase"
},
{
"operation": "search",
"time": "194.698",
"value": "2012 summer olympics 100 meter"
},
{
"operation": "Now on",
"time": "197.413",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "199.778",
"value": "2012 summer olympics 100 metre"
},
{
"operation": "Now on",
"time": "201.297",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "205.281",
"value": "Swimming at the 2012 Summer Olympics – Men's 100 metre freestyle"
},
{
"operation": "Now on",
"time": "215.066",
"value": "Swimming at the 2012 Summer Olympics – Men's 100 metre freestyle"
},
{
"operation": "back-button-clicked",
"time": "217.738",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "218.932",
"value": "undefined"
},
{
"operation": "search",
"time": "272.854",
"value": "intitle"
},
{
"operation": "Now on",
"time": "275.442",
"value": "undefined"
},
{
"operation": "search",
"time": "324.331",
"value": "andrew hinds"
},
{
"operation": "Now on",
"time": "326.233",
"value": "Andrew Hinds"
},
{
"operation": "hyperlink",
"time": "338.152",
"value": "2012 Summer Olympics"
},
{
"operation": "Now on",
"time": "338.683",
"value": "2012 Summer Olympics"
},
{
"operation": "hyperlink",
"time": "367.503",
"value": "3 The Games 2012 Summer Olympics_section_16"
},
{
"operation": "search",
"time": "438.736",
"value": "insource"
},
{
"operation": "Now on",
"time": "442.825",
"value": "undefined"
},
{
"operation": "search",
"time": "471.087",
"value": "Barbados at the 2012 Summer Olympics"
},
{
"operation": "Now on",
"time": "472.623",
"value": "Barbados at the 2012 Summer Olympics"
},
{
"operation": "hyperlink",
"time": "532.146",
"value": "Athletics at the 2012 Summer Olympics"
},
{
"operation": "Now on",
"time": "547.681",
"value": "Athletics at the 2012 Summer Olympics"
},
{
"operation": "back-button-clicked",
"time": "568.354",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "568.485",
"value": "Barbados at the 2012 Summer Olympics"
},
{
"operation": "search",
"time": "681.095",
"value": "andew hinds"
},
{
"operation": "Now on",
"time": "683.314",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "688.082",
"value": "Andrew Hinds"
},
{
"operation": "Now on",
"time": "689.265",
"value": "Andrew Hinds"
},
{
"operation": "finish",
"time": "702.613",
"value": "finish"
}
]
}
|
exfever_5859
|
Sikhism originated in the Panjāb, a geopolitical, cultural, and historical region in South Asia.
|
{
"evidence": [
"Sikhism",
"Punjab"
],
"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:\nSikhism originated in the Panjāb, a geopolitical, cultural, and historical region in South Asia.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Sikhism",
"Punjab"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 160,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7767
|
The United States Congress is a bicameral legislature composed of three bodies, including the upper chamber.
|
{
"evidence": [
"United States Congress",
"United States Senate"
],
"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 United States Congress is a bicameral legislature composed of three bodies, including the upper chamber.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"United States Congress",
"United States Senate"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 161,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
92675
|
Ludwikowice Kłodzkie was part of Germany between 1871 and 1945 wherein in 1942, the Germans established a forced labor camp for Jews in the village before it was attacked by the enemies.
|
{
"evidence": [
"Ludwikowice Kłodzkie_sentence_4",
"Ludwikowice Kłodzkie_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:\nLudwikowice Kłodzkie was part of Germany between 1871 and 1945 wherein in 1942, the Germans established a forced labor camp for Jews in the village before it was attacked by the enemies.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Ludwikowice Kłodzkie_sentence_4",
"Ludwikowice Kłodzkie_sentence_5"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 162,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.618",
"value": "?search="
},
{
"operation": "search",
"time": "9.328",
"value": "Ludwikowice Kłodzkie"
},
{
"operation": "Now on",
"time": "10.678",
"value": "Ludwikowice Kłodzkie"
},
{
"operation": "Highlighting",
"time": "16.454",
"value": "Ludwikowice Kłodzkie_sentence_4"
},
{
"operation": "Highlighting",
"time": "21.57",
"value": "Ludwikowice Kłodzkie_sentence_5"
},
{
"operation": "Page search",
"time": "56.853",
"value": "attacked"
},
{
"operation": "hyperlink",
"time": "71.548",
"value": "Gross-Rosen concentration camp"
},
{
"operation": "Now on",
"time": "72.38",
"value": "Gross-Rosen concentration camp"
},
{
"operation": "Page search",
"time": "78.526",
"value": "attacked"
},
{
"operation": "Page search",
"time": "119.348",
"value": "enemies"
},
{
"operation": "Page search",
"time": "135.642",
"value": "Kłodzkie"
},
{
"operation": "finish",
"time": "179.895",
"value": "finish"
}
]
}
|
exfever_2876
|
Macaulay Culkin is the oldest actor who starred in The Good Son.
|
{
"evidence": [
"Macaulay Culkin",
"My Girl (film)"
],
"label": "NOT ENOUGH INFO"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nMacaulay Culkin is the oldest actor who starred in The Good Son.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Macaulay Culkin",
"My Girl (film)"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 163,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
23457
|
Dinamo scored 35 more wins than Rijeka in the League competition in the Yugoslav championship (1946-1991).
|
{
"evidence": [
"GNK Dinamo Zagreb_sentence_2"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nDinamo scored 35 more wins than Rijeka in the League competition in the Yugoslav championship (1946-1991).\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"GNK Dinamo Zagreb_sentence_2"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 164,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.777",
"value": "?search="
},
{
"operation": "search",
"time": "19.034",
"value": "Yugoslav championship (1946-1991)"
},
{
"operation": "Now on",
"time": "20.936",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "42.481",
"value": "SD Željezničar"
},
{
"operation": "Now on",
"time": "43.664",
"value": "SD Željezničar"
},
{
"operation": "Page search",
"time": "52.437",
"value": "dinamo"
},
{
"operation": "back-button-clicked",
"time": "61.885",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "62.878",
"value": "undefined"
},
{
"operation": "search",
"time": "81.569",
"value": "Yugoslav championship (1946-1991)"
},
{
"operation": "Now on",
"time": "82.537",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "87.505",
"value": "SD Željezničar"
},
{
"operation": "Now on",
"time": "88.219",
"value": "SD Željezničar"
},
{
"operation": "Page search",
"time": "96.063",
"value": "1946-1991"
},
{
"operation": "hyperlink",
"time": "107.505",
"value": "Yugoslav championship"
},
{
"operation": "Now on",
"time": "108.351",
"value": "YUBA Liga#Past Yugoslav Cup Champions"
},
{
"operation": "Page search",
"time": "124.634",
"value": "dinamo"
},
{
"operation": "back-button-clicked",
"time": "137.917",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "138.517",
"value": "SD Željezničar"
},
{
"operation": "search",
"time": "181.173",
"value": "Rijeka"
},
{
"operation": "Now on",
"time": "182.584",
"value": "Rijeka"
},
{
"operation": "hyperlink",
"time": "189.784",
"value": "Rijeka (disambiguation)"
},
{
"operation": "Now on",
"time": "190.812",
"value": "Rijeka (disambiguation)"
},
{
"operation": "Highlighting",
"time": "199.184",
"value": "Rijeka (disambiguation)_item_0_0"
},
{
"operation": "Highlighting",
"time": "199.188",
"value": "Rijeka (disambiguation)_title"
},
{
"operation": "hyperlink",
"time": "203.89",
"value": "HNK Rijeka"
},
{
"operation": "Now on",
"time": "205.49",
"value": "HNK Rijeka"
},
{
"operation": "Page search",
"time": "215.11",
"value": "1946"
},
{
"operation": "Page search",
"time": "265.503",
"value": "dinamo"
},
{
"operation": "search",
"time": "330.331",
"value": "Yugoslav championship"
},
{
"operation": "Now on",
"time": "331.975",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "343.723",
"value": "GNK Dinamo Zagreb"
},
{
"operation": "Now on",
"time": "345.319",
"value": "GNK Dinamo Zagreb"
},
{
"operation": "back-button-clicked",
"time": "365.647",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "366.581",
"value": "undefined"
},
{
"operation": "search",
"time": "411.22",
"value": "Yugoslav championship"
},
{
"operation": "Now on",
"time": "412.011",
"value": "undefined"
},
{
"operation": "search",
"time": "436.359",
"value": "Yugoslav championship (1946-1991)"
},
{
"operation": "Now on",
"time": "437.244",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "449.259",
"value": "Croatia Zagreb era (1991–2000) GNK Dinamo Zagreb_section_5"
},
{
"operation": "Now on",
"time": "450.236",
"value": "GNK Dinamo Zagreb#Croatia Zagreb era .281991.E2.80.932000.29 GNK Dinamo Zagreb section 5"
},
{
"operation": "Page search",
"time": "463.682",
"value": "1946"
},
{
"operation": "Page search",
"time": "510.587",
"value": "Rijeka"
},
{
"operation": "Highlighting",
"time": "577.537",
"value": "GNK Dinamo Zagreb_sentence_2"
},
{
"operation": "finish",
"time": "588.996",
"value": "finish"
}
]
}
|
44124
|
IUPUI Jaguars consists of different teams; the men's basketball team has played at the Indiana Farmers Coliseum, the softball team plays at the IUPUI Softball Complex, and the tennis team plays at the West Indy Racquet Club.
|
{
"evidence": [
"IUPUI Jaguars_sentence_27",
"IUPUI Jaguars_sentence_28",
"IUPUI Jaguars_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:\nIUPUI Jaguars consists of different teams; the men's basketball team has played at the Indiana Farmers Coliseum, the softball team plays at the IUPUI Softball Complex, and the tennis team plays at the West Indy Racquet Club.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"IUPUI Jaguars_sentence_27",
"IUPUI Jaguars_sentence_28",
"IUPUI Jaguars_sentence_0"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 165,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.243",
"value": "?search="
},
{
"operation": "search",
"time": "181.158",
"value": "IUPUI Jaguars"
},
{
"operation": "Now on",
"time": "183.272",
"value": "IUPUI Jaguars"
},
{
"operation": "Highlighting",
"time": "185.98",
"value": "IUPUI Jaguars_sentence_0"
},
{
"operation": "Highlighting",
"time": "192.701",
"value": "IUPUI Jaguars_cell_1_1_1"
},
{
"operation": "Highlighting",
"time": "192.703",
"value": "IUPUI Jaguars_section_1"
},
{
"operation": "Highlighting",
"time": "192.704",
"value": "IUPUI Jaguars_header_cell_1_0_1"
},
{
"operation": "Highlighting",
"time": "193.212",
"value": "IUPUI Jaguars_cell_1_1_0"
},
{
"operation": "Highlighting",
"time": "193.214",
"value": "IUPUI Jaguars_header_cell_1_0_0"
},
{
"operation": "hyperlink",
"time": "195.064",
"value": "Jaguars men's basketball"
},
{
"operation": "Now on",
"time": "196.178",
"value": "IUPUI Jaguars men's basketball"
},
{
"operation": "back-button-clicked",
"time": "197.695",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "197.83",
"value": "IUPUI Jaguars"
},
{
"operation": "Highlighting deleted",
"time": "198.957",
"value": "IUPUI Jaguars_cell_1_1_1"
},
{
"operation": "Highlighting deleted",
"time": "198.961",
"value": "IUPUI Jaguars_header_cell_1_0_1"
},
{
"operation": "Highlighting deleted",
"time": "207.316",
"value": "IUPUI Jaguars_cell_1_1_0"
},
{
"operation": "Highlighting deleted",
"time": "207.321",
"value": "IUPUI Jaguars_section_1"
},
{
"operation": "Highlighting deleted",
"time": "207.323",
"value": "IUPUI Jaguars_header_cell_1_0_0"
},
{
"operation": "hyperlink",
"time": "211.896",
"value": "Jaguars men's basketball"
},
{
"operation": "Now on",
"time": "212.378",
"value": "IUPUI Jaguars men's basketball"
},
{
"operation": "Highlighting",
"time": "228.613",
"value": "IUPUI Jaguars men's basketball_cell_0_7_1"
},
{
"operation": "Highlighting",
"time": "228.615",
"value": "IUPUI Jaguars men's basketball_title"
},
{
"operation": "Highlighting",
"time": "228.622",
"value": "IUPUI Jaguars men's basketball_header_cell_0_7_0"
},
{
"operation": "Highlighting",
"time": "228.622",
"value": "IUPUI Jaguars men's basketball_header_cell_0_0_0"
},
{
"operation": "back-button-clicked",
"time": "231.261",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "231.371",
"value": "IUPUI Jaguars"
},
{
"operation": "hyperlink",
"time": "245.89",
"value": "Softball"
},
{
"operation": "Now on",
"time": "247.935",
"value": "College softball"
},
{
"operation": "back-button-clicked",
"time": "249.781",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "249.898",
"value": "IUPUI Jaguars"
},
{
"operation": "Highlighting",
"time": "260.906",
"value": "IUPUI Jaguars_sentence_27"
},
{
"operation": "Highlighting",
"time": "264.102",
"value": "IUPUI Jaguars_sentence_27"
},
{
"operation": "Highlighting",
"time": "267.749",
"value": "IUPUI Jaguars_sentence_28"
},
{
"operation": "Highlighting",
"time": "276.898",
"value": "IUPUI Jaguars_sentence_0"
},
{
"operation": "finish",
"time": "281.271",
"value": "finish"
}
]
}
|
11018
|
Despite achieving their fourth championship after the 2007–08 Polish Volleyball League, PGE Skra Bełchatów came up short versus Russia's Dynamo Tattransgaz Kazan (2–3) in the CEV semifinal.
|
{
"evidence": [
"2007–08 Polish Volleyball League_sentence_3",
"Skra Bełchatów_sentence_25"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nDespite achieving their fourth championship after the 2007–08 Polish Volleyball League, PGE Skra Bełchatów came up short versus Russia's Dynamo Tattransgaz Kazan (2–3) in the CEV semifinal.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"2007–08 Polish Volleyball League_sentence_3",
"Skra Bełchatów_sentence_25"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 166,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "2.052",
"value": "?search="
},
{
"operation": "search",
"time": "12.252",
"value": "2007–08 Polish Volleyball League"
},
{
"operation": "Now on",
"time": "13.421",
"value": "2007–08 Polish Volleyball League"
},
{
"operation": "Highlighting",
"time": "42.78",
"value": "2007–08 Polish Volleyball League_sentence_3"
},
{
"operation": "hyperlink",
"time": "81.436",
"value": "PGE Skra Bełchatów"
},
{
"operation": "Now on",
"time": "83.12",
"value": "Skra Bełchatów"
},
{
"operation": "Page search",
"time": "99.251",
"value": "despite"
},
{
"operation": "Page search",
"time": "118.602",
"value": "Dynamo Tattransgaz Kazan"
},
{
"operation": "Highlighting",
"time": "136.522",
"value": "Skra Bełchatów_sentence_25"
},
{
"operation": "finish",
"time": "154.67",
"value": "finish"
}
]
}
|
exfever_6063
|
Corsica is the fourth-largest island in the Mediterranean and lies north of the Italian island of second-largest island in the South China Sea.
|
{
"evidence": [
"Corsica",
"Sardinia"
],
"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:\nCorsica is the fourth-largest island in the Mediterranean and lies north of the Italian island of second-largest island in the South China Sea.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Corsica",
"Sardinia"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 167,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
75631
|
In road and rail transport, and water and waste management networks, civil engineers design the infrastructure (but urban planners could develop the general layout of roads, sidewalks and public places); electrical power and lighting networks are designed by power and electrical engineers; and telecommunications, computing and monitoring networks are designed by systems eingineers.
|
{
"evidence": [
"Infrastructure and economics_sentence_33",
"Infrastructure and economics_sentence_34"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nIn road and rail transport, and water and waste management networks, civil engineers design the infrastructure (but urban planners could develop the general layout of roads, sidewalks and public places); electrical power and lighting networks are designed by power and electrical engineers; and telecommunications, computing and monitoring networks are designed by systems eingineers.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Infrastructure and economics_sentence_33",
"Infrastructure and economics_sentence_34"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 168,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.945",
"value": "?search="
},
{
"operation": "search",
"time": "16.464",
"value": "road and rail transport"
},
{
"operation": "Now on",
"time": "20.237",
"value": "undefined"
},
{
"operation": "search",
"time": "37.401",
"value": "In road and rail transport, and water and waste management networks, civil engineers design the infrastructure"
},
{
"operation": "Now on",
"time": "40.217",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "498.527",
"value": "Infrastructure and economics"
},
{
"operation": "Now on",
"time": "499.529",
"value": "Infrastructure and economics"
},
{
"operation": "search",
"time": "688.374",
"value": "electrical power and lighting networks are designed by power and electrical engineers"
},
{
"operation": "Now on",
"time": "690.688",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "694.894",
"value": "Ownership and financing of infrastructure Infrastructure and economics_section_0"
},
{
"operation": "Now on",
"time": "695.355",
"value": "Infrastructure and economics#Ownership and financing of infrastructure Infrastructure and economics section 0"
},
{
"operation": "Page search",
"time": "714.194",
"value": "road and rail transport"
},
{
"operation": "Highlighting",
"time": "801.807",
"value": "Infrastructure and economics_sentence_1"
},
{
"operation": "Highlighting deleted",
"time": "802.355",
"value": "Infrastructure and economics_sentence_1"
},
{
"operation": "Highlighting",
"time": "807.032",
"value": "Infrastructure and economics_sentence_3"
},
{
"operation": "Highlighting",
"time": "807.035",
"value": "Infrastructure and economics_section_0"
},
{
"operation": "Highlighting deleted",
"time": "808.147",
"value": "Infrastructure and economics_sentence_3"
},
{
"operation": "Highlighting deleted",
"time": "808.151",
"value": "Infrastructure and economics_section_0"
},
{
"operation": "Highlighting",
"time": "812.897",
"value": "Infrastructure and economics_sentence_33"
},
{
"operation": "Highlighting",
"time": "822.284",
"value": "Infrastructure and economics_sentence_34"
},
{
"operation": "Page search",
"time": "835.044",
"value": "electrical power and lighting networks are designed by power and electrical engineers"
},
{
"operation": "Page search",
"time": "849.814",
"value": "electrical power"
},
{
"operation": "back-button-clicked",
"time": "864.863",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "866.302",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "868.649",
"value": "Ownership and financing of infrastructure Infrastructure and economics_section_0"
},
{
"operation": "Now on",
"time": "869.385",
"value": "Infrastructure and economics#Ownership and financing of infrastructure Infrastructure and economics section 0"
},
{
"operation": "back-button-clicked",
"time": "897.245",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "898.098",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "902.274",
"value": "Ownership and financing of infrastructure Infrastructure and economics_section_0"
},
{
"operation": "Now on",
"time": "902.749",
"value": "Infrastructure and economics#Ownership and financing of infrastructure Infrastructure and economics section 0"
},
{
"operation": "Page search",
"time": "916.573",
"value": "lighting networks"
},
{
"operation": "finish",
"time": "942.41",
"value": "finish"
}
]
}
|
63430
|
Fearless Fosdick is a long-running parody of Dick Tracy, which first appeared in 1942 and was popular enough to have a television show in 1952; newspaper editors wanted Fosdick to star in his own movie but the creator of Dick Tracy did not find the parody funny.
|
{
"evidence": [
"Fearless Fosdick_sentence_0",
"Fearless Fosdick_sentence_2",
"Fearless Fosdick_sentence_101",
"Fearless Fosdick_sentence_78",
"Fearless Fosdick_sentence_79"
],
"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:\nFearless Fosdick is a long-running parody of Dick Tracy, which first appeared in 1942 and was popular enough to have a television show in 1952; newspaper editors wanted Fosdick to star in his own movie but the creator of Dick Tracy did not find the parody funny.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Fearless Fosdick_sentence_0",
"Fearless Fosdick_sentence_2",
"Fearless Fosdick_sentence_101",
"Fearless Fosdick_sentence_78",
"Fearless Fosdick_sentence_79"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 169,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.773",
"value": "?search="
},
{
"operation": "search",
"time": "197.515",
"value": "Fearless Fosdick"
},
{
"operation": "Now on",
"time": "203.173",
"value": "Fearless Fosdick"
},
{
"operation": "Highlighting",
"time": "213.155",
"value": "Fearless Fosdick_sentence_0"
},
{
"operation": "Highlighting",
"time": "233.085",
"value": "Fearless Fosdick_sentence_2"
},
{
"operation": "Highlighting",
"time": "412.204",
"value": "Fearless Fosdick_sentence_101"
},
{
"operation": "Page search",
"time": "892.869",
"value": "newspaper editors"
},
{
"operation": "Highlighting",
"time": "905.189",
"value": "Fearless Fosdick_sentence_80"
},
{
"operation": "Highlighting",
"time": "906.423",
"value": "Fearless Fosdick_sentence_78"
},
{
"operation": "Highlighting deleted",
"time": "907.372",
"value": "Fearless Fosdick_sentence_80"
},
{
"operation": "Highlighting",
"time": "950.928",
"value": "Fearless Fosdick_sentence_79"
},
{
"operation": "finish",
"time": "968.088",
"value": "finish"
}
]
}
|
exfever_189
|
Ralph Fiennes is a stage actor.
|
{
"evidence": [
"Maid in Manhattan",
"John Hughes (filmmaker)",
"National Lampoon's Vacation"
],
"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:\nRalph Fiennes is a stage actor.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Maid in Manhattan",
"John Hughes (filmmaker)",
"National Lampoon's Vacation"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 170,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_2375
|
Justin Pierre James Trudeau is the eldest son of a Canadian lawyer and politician who served as the 15th prime minister of Canada.?
|
{
"evidence": [
"Justin Trudeau",
"Pierre Trudeau"
],
"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:\nJustin Pierre James Trudeau is the eldest son of a Canadian lawyer and politician who served as the 15th prime minister of Canada.?\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Justin Trudeau",
"Pierre Trudeau"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 171,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7466
|
Susan Denise Atkins was convicted for her participation in eight of the nine murders, including the most notorious series of murders perpetrated by members of the Manson Family during April 8–10, 1969, in Los Angeles, California, United States.
|
{
"evidence": [
"Susan Atkins",
"Tate murders"
],
"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:\nSusan Denise Atkins was convicted for her participation in eight of the nine murders, including the most notorious series of murders perpetrated by members of the Manson Family during April 8–10, 1969, in Los Angeles, California, United States.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Susan Atkins",
"Tate murders"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 172,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
70292
|
Forty percent of State Correctional Institution–Pine Grove's prisons are under 20, and its overall average age is 38.
|
{
"evidence": [
"State Correctional Institution – Pine Grove_sentence_6",
"State Correctional Institution – Pine Grove_sentence_7"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nForty percent of State Correctional Institution–Pine Grove's prisons are under 20, and its overall average age is 38.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"State Correctional Institution – Pine Grove_sentence_6",
"State Correctional Institution – Pine Grove_sentence_7"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 173,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.209",
"value": "?search="
},
{
"operation": "search",
"time": "67.497",
"value": "State Correctional Institution–Pine Grove"
},
{
"operation": "Now on",
"time": "69.55",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "89.219",
"value": "State Correctional Institution – Pine Grove"
},
{
"operation": "Now on",
"time": "89.732",
"value": "State Correctional Institution – Pine Grove"
},
{
"operation": "Highlighting",
"time": "101.832",
"value": "State Correctional Institution – Pine Grove_sentence_6"
},
{
"operation": "Highlighting",
"time": "105.162",
"value": "State Correctional Institution – Pine Grove_sentence_7"
},
{
"operation": "finish",
"time": "135.317",
"value": "finish"
}
]
}
|
exfever_7032
|
Eddie Jerome Vedder released his first solo album as a soundtrack for the film, an adaptation of the 1996 non-fiction book of the same name that tells the story of Christopher McCandless, a man who hiked across a continent in the Northern Hemisphere and almost entirely within the Western Hemisphere.
|
{
"evidence": [
"Eddie Vedder",
"Into the Wild (film)",
"North America"
],
"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:\nEddie Jerome Vedder released his first solo album as a soundtrack for the film, an adaptation of the 1996 non-fiction book of the same name that tells the story of Christopher McCandless, a man who hiked across a continent in the Northern Hemisphere and almost entirely within the Western Hemisphere.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Eddie Vedder",
"Into the Wild (film)",
"North America"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 174,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
64805
|
The proteasome is a multicatalytic proteinase complex with a highly ordered 20S core structure is composed of four axially stacked rings of 28 non-identical subunits: the two end rings are each formed by 14 alpha subunits, and the two central rings are each formed by seven beta subunits.
|
{
"evidence": [
"Proteasome (prosome, macropain) subunit, alpha 1_sentence_10"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe proteasome is a multicatalytic proteinase complex with a highly ordered 20S core structure is composed of four axially stacked rings of 28 non-identical subunits: the two end rings are each formed by 14 alpha subunits, and the two central rings are each formed by seven beta subunits.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Proteasome (prosome, macropain) subunit, alpha 1_sentence_10"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 175,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.081",
"value": "?search="
},
{
"operation": "search",
"time": "15.974",
"value": "Contains...proteasome is a multicatalytic proteinase complex"
},
{
"operation": "Now on",
"time": "17.937",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "46.18",
"value": "PSMD7"
},
{
"operation": "Now on",
"time": "46.956",
"value": "PSMD7"
},
{
"operation": "Highlighting",
"time": "53.118",
"value": "PSMD7_sentence_1"
},
{
"operation": "Highlighting",
"time": "53.12",
"value": "PSMD7_title"
},
{
"operation": "Highlighting",
"time": "94.183",
"value": "PSMD7_sentence_2"
},
{
"operation": "Highlighting deleted",
"time": "114.045",
"value": "PSMD7_sentence_1"
},
{
"operation": "Page search",
"time": "122.83",
"value": "axially"
},
{
"operation": "Page search",
"time": "140.345",
"value": "seven beta"
},
{
"operation": "Page search",
"time": "174.058",
"value": "end rings"
},
{
"operation": "search",
"time": "200.693",
"value": "Contains...axially stacked rings"
},
{
"operation": "Now on",
"time": "202.55",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "212.118",
"value": "Proteasome (prosome, macropain) subunit, alpha 1"
},
{
"operation": "Now on",
"time": "214.308",
"value": "Proteasome (prosome, macropain) subunit, alpha 1"
},
{
"operation": "Page search",
"time": "246.797",
"value": "multicatalytic proteinase"
},
{
"operation": "Highlighting",
"time": "256.783",
"value": "Proteasome (prosome, macropain) subunit, alpha 1_sentence_9"
},
{
"operation": "Highlighting",
"time": "267.237",
"value": "Proteasome (prosome, macropain) subunit, alpha 1_sentence_10"
},
{
"operation": "Highlighting deleted",
"time": "276.906",
"value": "Proteasome (prosome, macropain) subunit, alpha 1_sentence_9"
},
{
"operation": "finish",
"time": "302.163",
"value": "finish"
}
]
}
|
exfever_2247
|
Ava Marie DuVernay was the highest-earning director of 2012.?
|
{
"evidence": [
"Ava DuVernay",
"Middle of Nowhere (2012 film)",
"2012 Sundance Film Festival"
],
"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:\nAva Marie DuVernay was the highest-earning director of 2012.?\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Ava DuVernay",
"Middle of Nowhere (2012 film)",
"2012 Sundance Film Festival"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 176,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7241
|
Joseph Michael "Joey" Graceffa is an American YouTuber hosting an American reality television series that was first broadcast on June 22, 2016, on YouTube Premium.
|
{
"evidence": [
"Joey Graceffa",
"Escape the Night"
],
"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:\nJoseph Michael \"Joey\" Graceffa is an American YouTuber hosting an American reality television series that was first broadcast on June 22, 2016, on YouTube Premium.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Joey Graceffa",
"Escape the Night"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 177,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_4704
|
John Sidney McCain III was the Republican nominee for the head of state and head of government of the United States of America.
|
{
"evidence": [
"John McCain",
"President of the United States"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nJohn Sidney McCain III was the Republican nominee for the head of state and head of government of the United States of America.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"John McCain",
"President of the United States"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 178,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_4370
|
Jerry Stiller died from natural causes at his home on the Upper West Side of Manhattan, at the age of 92.
|
{
"evidence": [
"Ben Stiller",
"Anne Meara"
],
"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:\nJerry Stiller died from natural causes at his home on the Upper West Side of Manhattan, at the age of 92.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Ben Stiller",
"Anne Meara"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 179,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7403
|
Manchester United F.C. is a professional club of a team sport played with a spherical ball between two teams of 11 players.
|
{
"evidence": [
"Manchester United F.C.",
"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:\nManchester United F.C. is a professional club of a team sport played with a spherical ball between two teams of 11 players.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Manchester United F.C.",
"Association football"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 180,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_4369
|
Limbu is language spoken by the Limbu people of Nepal and Northeastern India, a family of more than 500 languages in which other languages are spoken in the mountain range in Asia, separating the plains of the Indian subcontinent from the Tibetan Plateau.
|
{
"evidence": [
"Limbu language",
"Sino-Tibetan languages",
"Himalayas"
],
"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:\nLimbu is language spoken by the Limbu people of Nepal and Northeastern India, a family of more than 500 languages in which other languages are spoken in the mountain range in Asia, separating the plains of the Indian subcontinent from the Tibetan Plateau.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Limbu language",
"Sino-Tibetan languages",
"Himalayas"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 181,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_6923
|
Fences is the first film Denzel Washington produced.
|
{
"evidence": [
"Fences (film)",
"Pulitzer Prize"
],
"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:\nFences is the first film Denzel Washington produced.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Fences (film)",
"Pulitzer Prize"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 182,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7100
|
The Land Before Time was the first American animated film distributed by Universal Pictures that was centered on dinosaurs.
|
{
"evidence": [
"The Land Before Time",
"George Lucas"
],
"label": "NOT ENOUGH INFO"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Land Before Time was the first American animated film distributed by Universal Pictures that was centered on dinosaurs.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Land Before Time",
"George Lucas"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 183,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.