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 |
|---|---|---|---|---|---|---|---|---|
14989
|
Cullingworth is a village and civil parish 7 miles west of Bradford and 3 miles south of Keighley, 2 miles to the east of Haworth and is situated near the head of Harden valley in West Yorkshire, England.
|
{
"evidence": [
"Cullingworth_sentence_0",
"Cullingworth_sentence_1",
"Haworth_sentence_1",
"Cullingworth_sentence_4",
"Keighley_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:\nCullingworth is a village and civil parish 7 miles west of Bradford and 3 miles south of Keighley, 2 miles to the east of Haworth and is situated near the head of Harden valley in West Yorkshire, England.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Cullingworth_sentence_0",
"Cullingworth_sentence_1",
"Haworth_sentence_1",
"Cullingworth_sentence_4",
"Keighley_sentence_1"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5884,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.724",
"value": "?search="
},
{
"operation": "search",
"time": "14.692",
"value": "Cullingworth"
},
{
"operation": "Now on",
"time": "16.208",
"value": "Cullingworth"
},
{
"operation": "Highlighting",
"time": "24.966",
"value": "Cullingworth_sentence_0"
},
{
"operation": "Highlighting",
"time": "31.872",
"value": "Cullingworth_sentence_1"
},
{
"operation": "Page search",
"time": "86.315",
"value": "haworth"
},
{
"operation": "search",
"time": "108.693",
"value": "Haworth"
},
{
"operation": "Now on",
"time": "110.702",
"value": "Haworth"
},
{
"operation": "Highlighting",
"time": "129.747",
"value": "Haworth_sentence_1"
},
{
"operation": "search",
"time": "166.204",
"value": "west yorkshire"
},
{
"operation": "Now on",
"time": "168.66",
"value": "West Yorkshire"
},
{
"operation": "search",
"time": "191.749",
"value": "harden valley"
},
{
"operation": "Now on",
"time": "193.25",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "212.174",
"value": "hardin valley"
},
{
"operation": "Now on",
"time": "213.843",
"value": "undefined"
},
{
"operation": "search",
"time": "222.04",
"value": "Cullingworth"
},
{
"operation": "Now on",
"time": "223.241",
"value": "Cullingworth"
},
{
"operation": "Highlighting",
"time": "239.754",
"value": "Cullingworth_sentence_4"
},
{
"operation": "hyperlink",
"time": "253.286",
"value": "Keighley"
},
{
"operation": "Now on",
"time": "254.31",
"value": "Keighley railway station"
},
{
"operation": "back-button-clicked",
"time": "271.648",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "271.793",
"value": "Cullingworth"
},
{
"operation": "hyperlink",
"time": "276.783",
"value": "Keighley"
},
{
"operation": "Now on",
"time": "279.946",
"value": "Keighley"
},
{
"operation": "Highlighting",
"time": "285.963",
"value": "Keighley_sentence_1"
},
{
"operation": "finish",
"time": "306.265",
"value": "finish"
}
]
}
|
exfever_3290
|
Hungarians are also incorporating local European elements and minor Central Asian-derived traditions.
|
{
"evidence": [
"Hungarians",
"Ethnic group",
"Homeland"
],
"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:\nHungarians are also incorporating local European elements and minor Central Asian-derived traditions.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Hungarians",
"Ethnic group",
"Homeland"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5885,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_6443
|
Crystal Palace F.C. is a club of a team sport that is played in accordance with a set of codified rules that have been maintained by the IFAB.
|
{
"evidence": [
"Crystal Palace F.C.",
"Association football",
"Laws of the Game (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:\nCrystal Palace F.C. is a club of a team sport that is played in accordance with a set of codified rules that have been maintained by the IFAB.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Crystal Palace F.C.",
"Association football",
"Laws of the Game (association football)"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5886,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
23300
|
The 2018 U Sports University Cup Queen's Cup championship game was held in Ontario, but Alberta was penalized for an ineligible player.
|
{
"evidence": [
"2018 U Sports University Cup_sentence_17"
],
"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 2018 U Sports University Cup Queen's Cup championship game was held in Ontario, but Alberta was penalized for an ineligible player.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"2018 U Sports University Cup_sentence_17"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5887,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.023",
"value": "?search="
},
{
"operation": "search",
"time": "16.808",
"value": "2018 U Sports University Cup Queen's Cup"
},
{
"operation": "Now on",
"time": "19.106",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "39.82",
"value": "David Johnston University Cup"
},
{
"operation": "Now on",
"time": "40.794",
"value": "David Johnston University Cup"
},
{
"operation": "search",
"time": "123.099",
"value": "2018 Queen's Cup"
},
{
"operation": "Now on",
"time": "124.857",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "130.129",
"value": "Miguel Tabuena"
},
{
"operation": "Now on",
"time": "130.681",
"value": "Miguel Tabuena"
},
{
"operation": "search",
"time": "155.697",
"value": "2018 U Sports University Cup"
},
{
"operation": "Now on",
"time": "156.365",
"value": "2018 U Sports University Cup"
},
{
"operation": "Highlighting",
"time": "213.745",
"value": "2018 U Sports University Cup_sentence_17"
},
{
"operation": "hyperlink",
"time": "218.704",
"value": "Queen's Cup"
},
{
"operation": "Now on",
"time": "219.466",
"value": "Queen's Cup (ice hockey)"
},
{
"operation": "back-button-clicked",
"time": "240.339",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "240.451",
"value": "2018 U Sports University Cup"
},
{
"operation": "finish",
"time": "267.72",
"value": "finish"
}
]
}
|
exfever_2526
|
Get Out is an American horror film written and directed by a director that was hired as a cast member on the Fox sketch comedy series originally inspired by Mad magazine.
|
{
"evidence": [
"Get Out (film)",
"Jordan Peele",
"Mad TV"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nGet Out is an American horror film written and directed by a director that was hired as a cast member on the Fox sketch comedy series originally inspired by Mad magazine.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Get Out (film)",
"Jordan Peele",
"Mad TV"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5888,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
55143
|
Pascal Johansen made his first professional appearance against FC Metz and the next year competed in, and lost, the Coupe de France.
|
{
"evidence": [
"Pascal Johansen_sentence_4",
"Pascal Johansen_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:\nPascal Johansen made his first professional appearance against FC Metz and the next year competed in, and lost, the Coupe de France.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Pascal Johansen_sentence_4",
"Pascal Johansen_sentence_5"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5889,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.694",
"value": "?search="
},
{
"operation": "search",
"time": "16.562",
"value": "Contains...pascal jphansen"
},
{
"operation": "Now on",
"time": "17.38",
"value": "undefined"
},
{
"operation": "search",
"time": "25.333",
"value": "Pascal Johansen"
},
{
"operation": "Now on",
"time": "26.714",
"value": "Pascal Johansen"
},
{
"operation": "Highlighting",
"time": "39.136",
"value": "Pascal Johansen_sentence_4"
},
{
"operation": "Highlighting",
"time": "64.317",
"value": "Pascal Johansen_sentence_6"
},
{
"operation": "Highlighting deleted",
"time": "67.233",
"value": "Pascal Johansen_sentence_6"
},
{
"operation": "Highlighting",
"time": "68.092",
"value": "Pascal Johansen_sentence_5"
},
{
"operation": "Highlighting deleted",
"time": "71.191",
"value": "Pascal Johansen_sentence_5"
},
{
"operation": "Highlighting",
"time": "71.979",
"value": "Pascal Johansen_sentence_5"
},
{
"operation": "Highlighting",
"time": "74.976",
"value": "Pascal Johansen_sentence_6"
},
{
"operation": "Highlighting deleted",
"time": "82.268",
"value": "Pascal Johansen_sentence_6"
},
{
"operation": "finish",
"time": "93.366",
"value": "finish"
}
]
}
|
exfever_1489
|
Helios is the God and personification of the moon that is form of the Greek language which was dialect of Ancient Greek consisting mainly of Ionic and Aeolics.
|
{
"evidence": [
"Helios",
"Homeric Greek"
],
"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:\nHelios is the God and personification of the moon that is form of the Greek language which was dialect of Ancient Greek consisting mainly of Ionic and Aeolics.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Helios",
"Homeric Greek"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5890,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_1485
|
William Jefferson Clinton is the husband of an African politician who served to represent New York from 2001 to 2009, in the upper chamber of the United States Congress, with the House of Representatives being the lower chamber.
|
{
"evidence": [
"Bill Clinton",
"Hillary Clinton",
"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:\nWilliam Jefferson Clinton is the husband of an African politician who served to represent New York from 2001 to 2009, in the upper chamber of the United States Congress, with the House of Representatives being the lower chamber.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Bill Clinton",
"Hillary Clinton",
"United States Senate"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5891,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_3774
|
"What a Girl Wants" is a song recorded by Christina Aguilera for her eponymous debut studio album that was released in 1999.
|
{
"evidence": [
"What a Girl Wants (Christina Aguilera song)",
"Christina Aguilera (album)"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\n\"What a Girl Wants\" is a song recorded by Christina Aguilera for her eponymous debut studio album that was released in 1999.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"What a Girl Wants (Christina Aguilera song)",
"Christina Aguilera (album)"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5892,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_6404
|
Cary Grant is an actor who appeared in a 1933 crime/comedy film whose other genre is a brief era in the American film industry.
|
{
"evidence": [
"Cary Grant",
"She Done Him Wrong",
"Pre-Code Hollywood"
],
"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:\nCary Grant is an actor who appeared in a 1933 crime/comedy film whose other genre is a brief era in the American film industry.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Cary Grant",
"She Done Him Wrong",
"Pre-Code Hollywood"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5893,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_3684
|
Jason Bateman was also inducted as Hasty Pudding Man of the Year.
|
{
"evidence": [
"Jason Bateman",
"NBC",
"30 Rockefeller Plaza"
],
"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:\nJason Bateman was also inducted as Hasty Pudding Man of the Year.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Jason Bateman",
"NBC",
"30 Rockefeller Plaza"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5894,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
77046
|
OnePlus 2, a phone revealed on July 28, 2015, via virtual reality, features a revamped design meant to put it at the top with the rest of its flagship competitors while the handset incorporates an aluminum-magnesium alloy chassis.
|
{
"evidence": [
"OnePlus 2_sentence_2",
"OnePlus 2_sentence_22",
"OnePlus 2_sentence_23"
],
"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:\nOnePlus 2, a phone revealed on July 28, 2015, via virtual reality, features a revamped design meant to put it at the top with the rest of its flagship competitors while the handset incorporates an aluminum-magnesium alloy chassis.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"OnePlus 2_sentence_2",
"OnePlus 2_sentence_22",
"OnePlus 2_sentence_23"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5895,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.159",
"value": "?search="
},
{
"operation": "search",
"time": "19.997",
"value": "OnePlus 2"
},
{
"operation": "Now on",
"time": "20.829",
"value": "OnePlus 2"
},
{
"operation": "Highlighting",
"time": "27.786",
"value": "OnePlus 2_sentence_2"
},
{
"operation": "Highlighting",
"time": "61.022",
"value": "OnePlus 2_sentence_22"
},
{
"operation": "Highlighting",
"time": "68.942",
"value": "OnePlus 2_sentence_23"
},
{
"operation": "finish",
"time": "81.903",
"value": "finish"
}
]
}
|
exfever_3391
|
Toy Story is a 1996 American computer-animated comedy film co-written by an American animator, film director, screenwriter, producer, and voice actor who directed a 2016 American computer-animated adventure film produced by Pixar Animation Studios.
|
{
"evidence": [
"Toy Story",
"Andrew Stanton",
"Finding Dory"
],
"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:\nToy Story is a 1996 American computer-animated comedy film co-written by an American animator, film director, screenwriter, producer, and voice actor who directed a 2016 American computer-animated adventure film produced by Pixar Animation Studios.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Toy Story",
"Andrew Stanton",
"Finding Dory"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5896,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_919
|
Warcraft is a 2016 American action fantasy film written by an American scenarist who practices the craft of screenwriting, writing screenplays on which mass media, such as films, television programs, and video games, are based.
|
{
"evidence": [
"Warcraft (film)",
"Charles Leavitt",
"Screenwriter"
],
"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:\nWarcraft is a 2016 American action fantasy film written by an American scenarist who practices the craft of screenwriting, writing screenplays on which mass media, such as films, television programs, and video games, are based.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Warcraft (film)",
"Charles Leavitt",
"Screenwriter"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5897,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7919
|
A pet is a companion multicellular organism in the biological kingdom Plantae that is kept primarily for a person's company.
|
{
"evidence": [
"Pet",
"Animal"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nA pet is a companion multicellular organism in the biological kingdom Plantae that is kept primarily for a person's company.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Pet",
"Animal"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5898,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
2433
|
In 2007 the Wax Simulacra was released for download on the iTunes Store and the Zune Marketplace and in 2009, it was performed on MTV Live.
|
{
"evidence": [
"Wax Simulacra_sentence_6",
"Wax Simulacra_sentence_8"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nIn 2007 the Wax Simulacra was released for download on the iTunes Store and the Zune Marketplace and in 2009, it was performed on MTV Live.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Wax Simulacra_sentence_6",
"Wax Simulacra_sentence_8"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5899,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "search",
"time": "9.498",
"value": "Wax Simulacra"
},
{
"operation": "Highlighting",
"time": "31.935",
"value": "Wax Simulacra_sentence_6"
},
{
"operation": "Highlighting",
"time": "35.062",
"value": "Wax Simulacra_sentence_8"
},
{
"operation": "finish",
"time": "63.346",
"value": "finish"
}
]
}
|
exfever_7524
|
Paris fatally wounds the son of Nereus and Doris' daughter.
|
{
"evidence": [
"Paris (mythology)",
"Achilles",
"Thetis"
],
"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:\nParis fatally wounds the son of Nereus and Doris' daughter.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Paris (mythology)",
"Achilles",
"Thetis"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5900,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
33744
|
Ron Dart is one of the primary experts on the life and thought of both Stephen Leacock and George Grant and is a board member of the Thomas Merton Society of Canada.
|
{
"evidence": [
"Ron Dart_sentence_4",
"Ron Dart_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:\nRon Dart is one of the primary experts on the life and thought of both Stephen Leacock and George Grant and is a board member of the Thomas Merton Society of Canada.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Ron Dart_sentence_4",
"Ron Dart_sentence_5"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5901,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.681",
"value": "?search="
},
{
"operation": "search",
"time": "8.797",
"value": "Ron Dart"
},
{
"operation": "Now on",
"time": "9.439",
"value": "Ron Dart"
},
{
"operation": "Page search",
"time": "18.916",
"value": "Stephen Leacock"
},
{
"operation": "Page search",
"time": "19.133",
"value": "Stephen Leacock"
},
{
"operation": "Highlighting",
"time": "21.174",
"value": "Ron Dart_sentence_4"
},
{
"operation": "Highlighting",
"time": "30.531",
"value": "Ron Dart_sentence_5"
},
{
"operation": "finish",
"time": "42.815",
"value": "finish"
}
]
}
|
exfever_5820
|
Titanic is a 1997 Italian style of filmmaking with a large-scale, sweeping scope, and spectacle.?
|
{
"evidence": [
"Titanic (1997 film)",
"Epic 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:\nTitanic is a 1997 Italian style of filmmaking with a large-scale, sweeping scope, and spectacle.?\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Titanic (1997 film)",
"Epic film"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5902,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
95466
|
Both Fly-SAX and Jambojet has flights to Nairobi–Jomo Kenyatta, along with RwandAir and Kenya Airways.
|
{
"evidence": [
"Jomo Kenyatta International Airport_sentence_0",
"Jomo Kenyatta International Airport_sentence_1",
"Jomo Kenyatta International Airport_sentence_2",
"Jomo Kenyatta International Airport_sentence_3",
"Jomo Kenyatta International Airport_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:\nBoth Fly-SAX and Jambojet has flights to Nairobi–Jomo Kenyatta, along with RwandAir and Kenya Airways.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Jomo Kenyatta International Airport_sentence_0",
"Jomo Kenyatta International Airport_sentence_1",
"Jomo Kenyatta International Airport_sentence_2",
"Jomo Kenyatta International Airport_sentence_3",
"Jomo Kenyatta International Airport_sentence_4"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5903,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.687",
"value": "?search="
},
{
"operation": "Now on",
"time": "10.709",
"value": "?search="
},
{
"operation": "search",
"time": "29.045",
"value": "Contains...flights to Nairobi–Jomo Kenyatta"
},
{
"operation": "Now on",
"time": "31.579",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "55.773",
"value": "Jomo Kenyatta International Airport"
},
{
"operation": "Now on",
"time": "57.215",
"value": "Jomo Kenyatta International Airport"
},
{
"operation": "Highlighting",
"time": "75.6",
"value": "Jomo Kenyatta International Airport_sentence_0"
},
{
"operation": "Highlighting",
"time": "76.515",
"value": "Jomo Kenyatta International Airport_sentence_1"
},
{
"operation": "Highlighting",
"time": "76.938",
"value": "Jomo Kenyatta International Airport_sentence_2"
},
{
"operation": "Highlighting",
"time": "78.356",
"value": "Jomo Kenyatta International Airport_sentence_3"
},
{
"operation": "Highlighting",
"time": "78.806",
"value": "Jomo Kenyatta International Airport_sentence_4"
},
{
"operation": "finish",
"time": "85.271",
"value": "finish"
}
]
}
|
79740
|
State Rep. Lopez Romero, a member of the Democratic party (one of the two major contemporary United States political parties), has sponsored 72 bills and currently serves on the State House Education, Retirement, and State Planning and Community Affairs committees.
|
{
"evidence": [
"Brenda Lopez Romero_sentence_2",
"Brenda Lopez Romero_sentence_3",
"Brenda Lopez Romero_sentence_4",
"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:\nState Rep. Lopez Romero, a member of the Democratic party (one of the two major contemporary United States political parties), has sponsored 72 bills and currently serves on the State House Education, Retirement, and State Planning and Community Affairs committees.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Brenda Lopez Romero_sentence_2",
"Brenda Lopez Romero_sentence_3",
"Brenda Lopez Romero_sentence_4",
"Democratic Party (United States)_sentence_1"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5904,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.693",
"value": "?search="
},
{
"operation": "search",
"time": "12.178",
"value": "Contains...Lopez Romero"
},
{
"operation": "Now on",
"time": "13.308",
"value": "undefined"
},
{
"operation": "search",
"time": "103.381",
"value": "Brenda Lopez Romero"
},
{
"operation": "Now on",
"time": "104.379",
"value": "Brenda Lopez Romero"
},
{
"operation": "Highlighting",
"time": "114.207",
"value": "Brenda Lopez Romero_sentence_2"
},
{
"operation": "Highlighting",
"time": "118.434",
"value": "Brenda Lopez Romero_sentence_3"
},
{
"operation": "Highlighting",
"time": "119.277",
"value": "Brenda Lopez Romero_sentence_4"
},
{
"operation": "hyperlink",
"time": "133.449",
"value": "Democratic"
},
{
"operation": "Now on",
"time": "135.47",
"value": "Democratic Party (United States)"
},
{
"operation": "Highlighting",
"time": "136.846",
"value": "Democratic Party (United States)_sentence_1"
},
{
"operation": "finish",
"time": "143.975",
"value": "finish"
}
]
}
|
71925
|
Yaniv Erlich published two studies on genetic privacy and reported the possibility of finding out a males surname by looking at their Y chromosome.
|
{
"evidence": [
"Yaniv Erlich_sentence_12",
"Yaniv Erlich_sentence_13"
],
"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:\nYaniv Erlich published two studies on genetic privacy and reported the possibility of finding out a males surname by looking at their Y chromosome.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Yaniv Erlich_sentence_12",
"Yaniv Erlich_sentence_13"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5905,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.172",
"value": "?search="
},
{
"operation": "search",
"time": "10.715",
"value": "Yaniv Erlich"
},
{
"operation": "Now on",
"time": "11.426",
"value": "Yaniv Erlich"
},
{
"operation": "Page search",
"time": "53.374",
"value": "privacy"
},
{
"operation": "Highlighting",
"time": "67.107",
"value": "Yaniv Erlich_sentence_12"
},
{
"operation": "Highlighting",
"time": "72.778",
"value": "Yaniv Erlich_sentence_13"
},
{
"operation": "finish",
"time": "88.146",
"value": "finish"
}
]
}
|
79899
|
The Methodist Diaconal Order is both a religious order and an order of ministry Methodist Diaconal Order where currently, the order reports to have about 280 deacons, who follow a common Rule of Life.
|
{
"evidence": [
"Methodist Diaconal Order_sentence_1",
"Methodist Diaconal Order_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:\nThe Methodist Diaconal Order is both a religious order and an order of ministry Methodist Diaconal Order where currently, the order reports to have about 280 deacons, who follow a common Rule of Life.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Methodist Diaconal Order_sentence_1",
"Methodist Diaconal Order_sentence_4"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5906,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.117",
"value": "?search="
},
{
"operation": "search",
"time": "9.683",
"value": "Methodist Diaconal Order"
},
{
"operation": "Now on",
"time": "10.62",
"value": "Methodist Diaconal Order"
},
{
"operation": "Highlighting",
"time": "133.439",
"value": "Methodist Diaconal Order_sentence_1"
},
{
"operation": "Page search",
"time": "193.592",
"value": "280"
},
{
"operation": "Highlighting",
"time": "196.476",
"value": "Methodist Diaconal Order_sentence_4"
},
{
"operation": "hyperlink",
"time": "203.143",
"value": "Rule of Life"
},
{
"operation": "Now on",
"time": "203.942",
"value": "Monasticism"
},
{
"operation": "finish",
"time": "212.833",
"value": "finish"
}
]
}
|
13242
|
The AFIB Technology has been tested by clinical studies and the Microlife blood pressure monitor which uses AFIB Technology had a sensitivity of 100%.
|
{
"evidence": [
"AFIB Technology_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 AFIB Technology has been tested by clinical studies and the Microlife blood pressure monitor which uses AFIB Technology had a sensitivity of 100%.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"AFIB Technology_sentence_10"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5907,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "2.072",
"value": "?search="
},
{
"operation": "search",
"time": "11.974",
"value": "AFIB Technology"
},
{
"operation": "Now on",
"time": "13.387",
"value": "AFIB Technology"
},
{
"operation": "Highlighting",
"time": "23.868",
"value": "AFIB Technology_sentence_2"
},
{
"operation": "Highlighting",
"time": "50.388",
"value": "AFIB Technology_sentence_8"
},
{
"operation": "Highlighting",
"time": "55.212",
"value": "AFIB Technology_sentence_9"
},
{
"operation": "Highlighting deleted",
"time": "66.523",
"value": "AFIB Technology_sentence_2"
},
{
"operation": "Highlighting",
"time": "80.741",
"value": "AFIB Technology_sentence_7"
},
{
"operation": "Highlighting deleted",
"time": "82.087",
"value": "AFIB Technology_sentence_7"
},
{
"operation": "Highlighting deleted",
"time": "87.767",
"value": "AFIB Technology_sentence_8"
},
{
"operation": "Highlighting",
"time": "95.577",
"value": "AFIB Technology_sentence_8"
},
{
"operation": "Highlighting",
"time": "107.253",
"value": "AFIB Technology_sentence_10"
},
{
"operation": "Highlighting deleted",
"time": "130.121",
"value": "AFIB Technology_sentence_9"
},
{
"operation": "Highlighting deleted",
"time": "131.886",
"value": "AFIB Technology_sentence_8"
},
{
"operation": "finish",
"time": "166.899",
"value": "finish"
}
]
}
|
87960
|
On the 6th, reinforced by the rest of the Norfolk Regiment, the 7th Rajputs and 120th Infantry and some mountain guns, they tried again and on the 8th the 104th and 110th Infantry were sent up the Tigris to find a place to cross.
|
{
"evidence": [
"Battle of Qurna_sentence_13",
"Battle of Qurna_sentence_16"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nOn the 6th, reinforced by the rest of the Norfolk Regiment, the 7th Rajputs and 120th Infantry and some mountain guns, they tried again and on the 8th the 104th and 110th Infantry were sent up the Tigris to find a place to cross.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Battle of Qurna_sentence_13",
"Battle of Qurna_sentence_16"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5908,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.922",
"value": "?search="
},
{
"operation": "Now on",
"time": "2.222",
"value": "?search="
},
{
"operation": "search",
"time": "24.54",
"value": "Infantry were sent up the Tigris to find a place to cross"
},
{
"operation": "Now on",
"time": "27.58",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "31.817",
"value": "Battle of Qurna"
},
{
"operation": "Now on",
"time": "32.726",
"value": "Battle of Qurna"
},
{
"operation": "Highlighting",
"time": "39.839",
"value": "Battle of Qurna_sentence_13"
},
{
"operation": "Highlighting",
"time": "50.912",
"value": "Battle of Qurna_sentence_16"
},
{
"operation": "finish",
"time": "67.842",
"value": "finish"
}
]
}
|
exfever_658
|
Alessia Cara is a Canadian singer-songwriter who signed with an American multinational record label owned by the Universal Music Group.
|
{
"evidence": [
"Alessia Cara",
"Def Jam Recordings"
],
"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:\nAlessia Cara is a Canadian singer-songwriter who signed with an American multinational record label owned by the Universal Music Group.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Alessia Cara",
"Def Jam Recordings"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5909,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
49269
|
Sori Choi (최소리) is a Korean traditional percussionist, she was born in South Korea and began touring internationally with the Korean Music Project.
|
{
"evidence": [
"Sori Choi_sentence_1",
"Sori Choi_sentence_2",
"Sori Choi_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:\nSori Choi (최소리) is a Korean traditional percussionist, she was born in South Korea and began touring internationally with the Korean Music Project.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"Sori Choi_sentence_1",
"Sori Choi_sentence_2",
"Sori Choi_sentence_6"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5910,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.902",
"value": "?search="
},
{
"operation": "search",
"time": "16.5",
"value": "Sori Choi"
},
{
"operation": "Now on",
"time": "17.401",
"value": "Sori Choi"
},
{
"operation": "Highlighting",
"time": "24.881",
"value": "Sori Choi_sentence_1"
},
{
"operation": "Highlighting",
"time": "30.647",
"value": "Sori Choi_sentence_2"
},
{
"operation": "Highlighting",
"time": "43.436",
"value": "Sori Choi_sentence_6"
},
{
"operation": "finish",
"time": "50.708",
"value": "finish"
}
]
}
|
41430
|
There were no significant lunar events in 1942.
|
{
"evidence": [
"Lunar_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 were no significant lunar events in 1942.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Lunar_sentence_0"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5911,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.796",
"value": "?search="
},
{
"operation": "search",
"time": "14.883",
"value": "lunar events"
},
{
"operation": "Now on",
"time": "16.399",
"value": "undefined"
},
{
"operation": "search",
"time": "23.697",
"value": "lunar events 1942"
},
{
"operation": "Now on",
"time": "24.957",
"value": "undefined"
},
{
"operation": "search",
"time": "60.344",
"value": "1942 lunar events"
},
{
"operation": "Now on",
"time": "60.929",
"value": "undefined"
},
{
"operation": "search",
"time": "74.86",
"value": "Lunar"
},
{
"operation": "Now on",
"time": "75.661",
"value": "Lunar"
},
{
"operation": "Highlighting",
"time": "116.383",
"value": "Lunar_sentence_0"
},
{
"operation": "finish",
"time": "120.734",
"value": "finish"
}
]
}
|
exfever_3117
|
Betty Buckley won the 1983 Oscar Award for Best Featured Actress in a Musical for her role as Grizabella in the original Broadway production of a sung-through musical composed by Andrew Lloyd Webber.
|
{
"evidence": [
"Betty Buckley",
"Cats (musical)"
],
"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:\nBetty Buckley won the 1983 Oscar Award for Best Featured Actress in a Musical for her role as Grizabella in the original Broadway production of a sung-through musical composed by Andrew Lloyd Webber.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Betty Buckley",
"Cats (musical)"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5912,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
11967
|
Rania Antonopoulou lost the chance to be appointed by the Greek Government as the Permanent Representative of Greece and a member of the Hellenic Parliament.
|
{
"evidence": [
"Rania Antonopoulou_sentence_3",
"Rania Antonopoulou_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:\nRania Antonopoulou lost the chance to be appointed by the Greek Government as the Permanent Representative of Greece and a member of the Hellenic Parliament.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Rania Antonopoulou_sentence_3",
"Rania Antonopoulou_sentence_4"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5913,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.134",
"value": "?search="
},
{
"operation": "search",
"time": "18.139",
"value": "Rania Antonopoulou"
},
{
"operation": "Now on",
"time": "19.455",
"value": "Rania Antonopoulou"
},
{
"operation": "Highlighting",
"time": "43.244",
"value": "Rania Antonopoulou_sentence_3"
},
{
"operation": "Page search",
"time": "85.292",
"value": "Hellenic"
},
{
"operation": "Highlighting",
"time": "93.266",
"value": "Rania Antonopoulou_sentence_4"
},
{
"operation": "Page search",
"time": "109.209",
"value": "appoin"
},
{
"operation": "finish",
"time": "186.631",
"value": "finish"
}
]
}
|
exfever_7490
|
The Ten Commandments is a 1956 epic religious drama film shot in a higher resolution, widescreen variant of the 35 mm motion picture film format, which was created in 1955.
|
{
"evidence": [
"The Ten Commandments (1956 film)",
"VistaVision"
],
"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 Ten Commandments is a 1956 epic religious drama film shot in a higher resolution, widescreen variant of the 35 mm motion picture film format, which was created in 1955.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Ten Commandments (1956 film)",
"VistaVision"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5914,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_8354
|
Alan Walbridge Ladd was often paired with Veronica Lake in film noir, such as in a film directed by Frank Tuttle, starring a Korean stage and film actor best known for his villainous performances in films such as I Wake Up Screaming.
|
{
"evidence": [
"Alan Ladd",
"This Gun for Hire",
"Laird Cregar"
],
"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:\nAlan Walbridge Ladd was often paired with Veronica Lake in film noir, such as in a film directed by Frank Tuttle, starring a Korean stage and film actor best known for his villainous performances in films such as I Wake Up Screaming.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Alan Ladd",
"This Gun for Hire",
"Laird Cregar"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5915,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_4586
|
7 Things is a song that was released by a Canadian record label with a current roster of artists, including an American pop band.
|
{
"evidence": [
"7 Things",
"Hollywood Records",
"Ocean Park Standoff"
],
"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:\n7 Things is a song that was released by a Canadian record label with a current roster of artists, including an American pop band.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"7 Things",
"Hollywood Records",
"Ocean Park Standoff"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5916,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_8265
|
Futurama was premiered on an American commercial broadcast television network with Big Three television networks such as the radio network which is owned by Comcast through ABSUniversal.
|
{
"evidence": [
"Futurama",
"Fox Broadcasting Company",
"NBC"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nFuturama was premiered on an American commercial broadcast television network with Big Three television networks such as the radio network which is owned by Comcast through ABSUniversal.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Futurama",
"Fox Broadcasting Company",
"NBC"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5917,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
86472
|
Florentine (a variety of Tuscan language, a Romance language) can be distinguished from Standard Italian by differences in phonology, morphology, syntax and lexicon; the most commonly noticed is the gorgia toscana, a consonant-weakening rule widespread in Tuscany in which the voiceless plosive phonemes /k/, /t/, /p/ are pronounced between vowels as fricatives [h], [θ], [ɸ] respectively.
|
{
"evidence": [
"Florentine dialect_sentence_5",
"Florentine dialect_sentence_1",
"Florentine dialect_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:\nFlorentine (a variety of Tuscan language, a Romance language) can be distinguished from Standard Italian by differences in phonology, morphology, syntax and lexicon; the most commonly noticed is the gorgia toscana, a consonant-weakening rule widespread in Tuscany in which the voiceless plosive phonemes /k/, /t/, /p/ are pronounced between vowels as fricatives [h], [θ], [ɸ] respectively.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Florentine dialect_sentence_5",
"Florentine dialect_sentence_1",
"Florentine dialect_sentence_6"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5918,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.842",
"value": "?search="
},
{
"operation": "search",
"time": "23.674",
"value": "Contains...distinguished from Standard Italian by differences"
},
{
"operation": "Now on",
"time": "26.202",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "33.818",
"value": "Florentine dialect"
},
{
"operation": "Now on",
"time": "34.72",
"value": "Florentine dialect"
},
{
"operation": "Highlighting",
"time": "44.74",
"value": "Florentine dialect_sentence_5"
},
{
"operation": "Highlighting",
"time": "52.763",
"value": "Florentine dialect_sentence_1"
},
{
"operation": "Highlighting",
"time": "68.013",
"value": "Florentine dialect_sentence_6"
},
{
"operation": "finish",
"time": "91.045",
"value": "finish"
}
]
}
|
45283
|
MXGP The Official Motocross Videogame (MXGP Pro) was a planned released in November 2011 is available for Microsoft Windows.
|
{
"evidence": [
"MXGP The Official Motocross Videogame_sentence_1"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nMXGP The Official Motocross Videogame (MXGP Pro) was a planned released in November 2011 is available for Microsoft Windows.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"MXGP The Official Motocross Videogame_sentence_1"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5919,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.737",
"value": "?search="
},
{
"operation": "search",
"time": "77.919",
"value": "MXGP The Official Motocross Videogame (MXGP Pro)"
},
{
"operation": "Now on",
"time": "79.245",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "80.784",
"value": "MXGP The Official Motocross Videogame"
},
{
"operation": "Now on",
"time": "81.393",
"value": "MXGP The Official Motocross Videogame"
},
{
"operation": "Page search",
"time": "90.662",
"value": "planned"
},
{
"operation": "Highlighting",
"time": "274.862",
"value": "MXGP The Official Motocross Videogame_sentence_1"
},
{
"operation": "finish",
"time": "281.56",
"value": "finish"
}
]
}
|
36769
|
The word "henchman" has appeared in the name of Hengest, the Saxon chieftain, which was often rendered as Henxman in medieval English, then the word became obsolete for grooms in English from the middle of the 18th century, but was retained in Scots as "personal attendant of a Highland chief".
|
{
"evidence": [
"Henchman_sentence_9"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe word \"henchman\" has appeared in the name of Hengest, the Saxon chieftain, which was often rendered as Henxman in medieval English, then the word became obsolete for grooms in English from the middle of the 18th century, but was retained in Scots as \"personal attendant of a Highland chief\".\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Henchman_sentence_9"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5920,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.726",
"value": "?search="
},
{
"operation": "search",
"time": "24.5",
"value": "Henchman"
},
{
"operation": "Now on",
"time": "25.241",
"value": "Henchman"
},
{
"operation": "Highlighting",
"time": "59.843",
"value": "Henchman_sentence_6"
},
{
"operation": "Highlighting",
"time": "64.852",
"value": "Henchman_sentence_7"
},
{
"operation": "Highlighting",
"time": "78.354",
"value": "Henchman_sentence_9"
},
{
"operation": "Highlighting deleted",
"time": "82.199",
"value": "Henchman_sentence_7"
},
{
"operation": "Highlighting deleted",
"time": "83.066",
"value": "Henchman_sentence_6"
},
{
"operation": "finish",
"time": "91.738",
"value": "finish"
}
]
}
|
exfever_8656
|
Margot Betti Frank's deportation order from the official secret police of Nazi Germany and in German-occupied Europe hastened the Frank family into hiding.
|
{
"evidence": [
"Margot Frank",
"Gestapo"
],
"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:\nMargot Betti Frank's deportation order from the official secret police of Nazi Germany and in German-occupied Europe hastened the Frank family into hiding.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Margot Frank",
"Gestapo"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5921,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
95466
|
Both Fly-SAX and Jambojet has flights to Nairobi–Jomo Kenyatta, along with RwandAir and Kenya Airways.
|
{
"evidence": [
"Jomo Kenyatta International Airport_sentence_0",
"Jomo Kenyatta International Airport_sentence_1",
"Jomo Kenyatta International Airport_sentence_2",
"Jomo Kenyatta International Airport_sentence_3",
"Jomo Kenyatta International Airport_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:\nBoth Fly-SAX and Jambojet has flights to Nairobi–Jomo Kenyatta, along with RwandAir and Kenya Airways.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Jomo Kenyatta International Airport_sentence_0",
"Jomo Kenyatta International Airport_sentence_1",
"Jomo Kenyatta International Airport_sentence_2",
"Jomo Kenyatta International Airport_sentence_3",
"Jomo Kenyatta International Airport_sentence_4"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5922,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.687",
"value": "?search="
},
{
"operation": "Now on",
"time": "10.709",
"value": "?search="
},
{
"operation": "search",
"time": "29.045",
"value": "Contains...flights to Nairobi–Jomo Kenyatta"
},
{
"operation": "Now on",
"time": "31.579",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "55.773",
"value": "Jomo Kenyatta International Airport"
},
{
"operation": "Now on",
"time": "57.215",
"value": "Jomo Kenyatta International Airport"
},
{
"operation": "Highlighting",
"time": "75.6",
"value": "Jomo Kenyatta International Airport_sentence_0"
},
{
"operation": "Highlighting",
"time": "76.515",
"value": "Jomo Kenyatta International Airport_sentence_1"
},
{
"operation": "Highlighting",
"time": "76.938",
"value": "Jomo Kenyatta International Airport_sentence_2"
},
{
"operation": "Highlighting",
"time": "78.356",
"value": "Jomo Kenyatta International Airport_sentence_3"
},
{
"operation": "Highlighting",
"time": "78.806",
"value": "Jomo Kenyatta International Airport_sentence_4"
},
{
"operation": "finish",
"time": "85.271",
"value": "finish"
}
]
}
|
4967
|
Yala National Park is a park situated in the dry semi-arid climatic region and rain is received mainly during the northeast monsoon in Vietnam having six national parks and three wildlife sanctuaries in the vicinity of Yala, where the largest is Lunugamvehera National Park (declared in 1995, with the intention of protecting the catchment area of the Lunugamvehera reservoir and wildlife of the area).
|
{
"evidence": [
"Yala National Park_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:\nYala National Park is a park situated in the dry semi-arid climatic region and rain is received mainly during the northeast monsoon in Vietnam having six national parks and three wildlife sanctuaries in the vicinity of Yala, where the largest is Lunugamvehera National Park (declared in 1995, with the intention of protecting the catchment area of the Lunugamvehera reservoir and wildlife of the area).\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Yala National Park_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5923,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "search",
"time": "24.477",
"value": "Yala National Park"
},
{
"operation": "Page search",
"time": "76.544",
"value": "dry"
},
{
"operation": "Highlighting",
"time": "78.957",
"value": "Yala National Park_sentence_11"
},
{
"operation": "Highlighting",
"time": "79.478",
"value": "Yala National Park_sentence_10"
},
{
"operation": "Highlighting deleted",
"time": "79.939",
"value": "Yala National Park_sentence_11"
},
{
"operation": "Page search",
"time": "88.943",
"value": "rain"
},
{
"operation": "Highlighting",
"time": "103.749",
"value": "Yala National Park_sentence_0"
},
{
"operation": "Highlighting deleted",
"time": "115.376",
"value": "Yala National Park_sentence_10"
},
{
"operation": "Page search",
"time": "142.824",
"value": "lunugamvehera"
},
{
"operation": "hyperlink",
"time": "160.659",
"value": "Lunugamvehera National Park"
},
{
"operation": "finish",
"time": "167.877",
"value": "finish"
}
]
}
|
exfever_257
|
Pearl Jam is an American rock band formed in Seattle had a member who later joined in 1998 and used to be the lead guitarist for Seattle-based rock band Soundgarden.
|
{
"evidence": [
"Pearl Jam",
"Matt Cameron"
],
"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:\nPearl Jam is an American rock band formed in Seattle had a member who later joined in 1998 and used to be the lead guitarist for Seattle-based rock band Soundgarden.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Pearl Jam",
"Matt Cameron"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5924,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
65049
|
During the 1968 Delaware gubernatorial election, Republican candidate Russell W. Peterson (who would later be president of the National Audubon Society) defeated incumbent Governor Charles L. Terry Jr.
|
{
"evidence": [
"1968 Delaware gubernatorial election_sentence_1",
"Russell W. Peterson_sentence_52"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nDuring the 1968 Delaware gubernatorial election, Republican candidate Russell W. Peterson (who would later be president of the National Audubon Society) defeated incumbent Governor Charles L. Terry Jr.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"1968 Delaware gubernatorial election_sentence_1",
"Russell W. Peterson_sentence_52"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5925,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.561",
"value": "?search="
},
{
"operation": "search",
"time": "16.02",
"value": "1968 Delaware gubernatorial election"
},
{
"operation": "Now on",
"time": "17.203",
"value": "1968 Delaware gubernatorial election"
},
{
"operation": "Highlighting",
"time": "44.519",
"value": "1968 Delaware gubernatorial election_sentence_1"
},
{
"operation": "hyperlink",
"time": "60.617",
"value": "Russell W. Peterson"
},
{
"operation": "Now on",
"time": "61.712",
"value": "Russell W. Peterson"
},
{
"operation": "Highlighting",
"time": "75.296",
"value": "Russell W. Peterson_sentence_5"
},
{
"operation": "Highlighting",
"time": "148.374",
"value": "Russell W. Peterson_sentence_52"
},
{
"operation": "Highlighting deleted",
"time": "191.895",
"value": "Russell W. Peterson_sentence_5"
},
{
"operation": "back-button-clicked",
"time": "196.162",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "196.348",
"value": "1968 Delaware gubernatorial election"
},
{
"operation": "finish",
"time": "267.42",
"value": "finish"
}
]
}
|
exfever_7775
|
Joshua Michael Homme is an American musician who was formerly the co-founder and guitarist of an American rock band, formed in Palm Desert, California, in 1987.
|
{
"evidence": [
"Josh Homme",
"Kyuss"
],
"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:\nJoshua Michael Homme is an American musician who was formerly the co-founder and guitarist of an American rock band, formed in Palm Desert, California, in 1987.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Josh Homme",
"Kyuss"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5926,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
20425
|
Oreta miltodes is a moth in the Drepanidae family (which has about 660 species worldwide, many of which have a hook-shaped apex to the forewing), which have substantially carmine pink forewings and hindwings, and grey whitish costa.
|
{
"evidence": [
"Oreta miltodes_sentence_5",
"Oreta miltodes_sentence_7",
"Oreta miltodes_sentence_8",
"Oreta miltodes_sentence_0",
"Drepanidae_sentence_3",
"Drepanidae_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:\nOreta miltodes is a moth in the Drepanidae family (which has about 660 species worldwide, many of which have a hook-shaped apex to the forewing), which have substantially carmine pink forewings and hindwings, and grey whitish costa.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Oreta miltodes_sentence_5",
"Oreta miltodes_sentence_7",
"Oreta miltodes_sentence_8",
"Oreta miltodes_sentence_0",
"Drepanidae_sentence_3",
"Drepanidae_sentence_6"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5927,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.942",
"value": "?search="
},
{
"operation": "search",
"time": "10.395",
"value": "Oreta miltodes"
},
{
"operation": "Now on",
"time": "11.119",
"value": "Oreta miltodes"
},
{
"operation": "Highlighting",
"time": "64.417",
"value": "Oreta miltodes_sentence_5"
},
{
"operation": "Highlighting",
"time": "88.909",
"value": "Oreta miltodes_sentence_7"
},
{
"operation": "Highlighting",
"time": "106.903",
"value": "Oreta miltodes_sentence_8"
},
{
"operation": "Highlighting",
"time": "112.827",
"value": "Oreta miltodes_sentence_0"
},
{
"operation": "hyperlink",
"time": "116.679",
"value": "Drepanidae"
},
{
"operation": "Now on",
"time": "117.21",
"value": "Drepanidae"
},
{
"operation": "Highlighting",
"time": "121.317",
"value": "Drepanidae_sentence_3"
},
{
"operation": "Highlighting",
"time": "126.78",
"value": "Drepanidae_sentence_6"
},
{
"operation": "finish",
"time": "140.557",
"value": "finish"
}
]
}
|
exfever_336
|
Troy Baker is known for his video game role as Booker DeWitt in a first-person shooter video game developed by Irrational Games and published by 3 Games.
|
{
"evidence": [
"Troy Baker",
"BioShock Infinite"
],
"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:\nTroy Baker is known for his video game role as Booker DeWitt in a first-person shooter video game developed by Irrational Games and published by 3 Games.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Troy Baker",
"BioShock Infinite"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5928,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_3631
|
Pawn Sacrifice was directed by an American filmmaker who has worked primarily in comedy-drama films, such as About Last Night.
|
{
"evidence": [
"Pawn Sacrifice",
"Edward Zwick"
],
"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:\nPawn Sacrifice was directed by an American filmmaker who has worked primarily in comedy-drama films, such as About Last Night.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Pawn Sacrifice",
"Edward Zwick"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5929,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
21173
|
Nguyễn Khuyến's first teacher in his quest to be a mandarin was Pham Van Nghi, he got a Masters but failed his doctorate so he went to the Imperial Academy in France which was built in 1070 and Vietnam's first national university.
|
{
"evidence": [
"Nguyễn Khuyến_sentence_9"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nNguyễn Khuyến's first teacher in his quest to be a mandarin was Pham Van Nghi, he got a Masters but failed his doctorate so he went to the Imperial Academy in France which was built in 1070 and Vietnam's first national university.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Nguyễn Khuyến_sentence_9"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5930,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.863",
"value": "?search="
},
{
"operation": "search",
"time": "11.042",
"value": "Nguyễn Khuyến"
},
{
"operation": "Now on",
"time": "11.954",
"value": "Nguyễn Khuyến"
},
{
"operation": "Page search",
"time": "15.113",
"value": "teacher"
},
{
"operation": "Highlighting",
"time": "22.678",
"value": "Nguyễn Khuyến_sentence_5"
},
{
"operation": "Highlighting",
"time": "23.544",
"value": "Nguyễn Khuyến_sentence_6"
},
{
"operation": "Highlighting",
"time": "36.672",
"value": "Nguyễn Khuyến_sentence_8"
},
{
"operation": "Highlighting",
"time": "38.2",
"value": "Nguyễn Khuyến_sentence_9"
},
{
"operation": "Highlighting deleted",
"time": "86.199",
"value": "Nguyễn Khuyến_sentence_5"
},
{
"operation": "Highlighting deleted",
"time": "86.948",
"value": "Nguyễn Khuyến_sentence_6"
},
{
"operation": "Highlighting deleted",
"time": "90.284",
"value": "Nguyễn Khuyến_sentence_8"
},
{
"operation": "finish",
"time": "99.221",
"value": "finish"
}
]
}
|
44649
|
Peter Noble discovered five years later that he had suffered a broken clavicle, he was the club's top scorer in 1969/1970 and scored in the Anglo-Italian League Cup Final.
|
{
"evidence": [
"Peter Noble_sentence_8",
"Scapula_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:\nPeter Noble discovered five years later that he had suffered a broken clavicle, he was the club's top scorer in 1969/1970 and scored in the Anglo-Italian League Cup Final.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Peter Noble_sentence_8",
"Scapula_sentence_5"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5931,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.205",
"value": "?search="
},
{
"operation": "search",
"time": "14.553",
"value": "Peter Noble"
},
{
"operation": "Now on",
"time": "15.483",
"value": "Peter Noble"
},
{
"operation": "Page search",
"time": "23.105",
"value": "clavicle"
},
{
"operation": "Highlighting",
"time": "31.385",
"value": "Peter Noble_sentence_8"
},
{
"operation": "search",
"time": "47.241",
"value": "clavicle"
},
{
"operation": "Now on",
"time": "48.063",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "64.137",
"value": "Shoulder girdle"
},
{
"operation": "Now on",
"time": "64.884",
"value": "Shoulder girdle"
},
{
"operation": "search",
"time": "89.968",
"value": "shoulder blade"
},
{
"operation": "Now on",
"time": "90.882",
"value": "Scapula"
},
{
"operation": "Highlighting",
"time": "100.049",
"value": "Scapula_sentence_5"
},
{
"operation": "finish",
"time": "120.576",
"value": "finish"
}
]
}
|
exfever_2386
|
Saddam Hussein was an Iraqi politician who espoused Ba'athism, a mix of Arab nationalism and a political ideology that was coined by Michel Aflaq, the principal founder of the political party founded in Syria.
|
{
"evidence": [
"Saddam Hussein",
"Arab socialism",
"Ba'ath Party"
],
"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:\nSaddam Hussein was an Iraqi politician who espoused Ba'athism, a mix of Arab nationalism and a political ideology that was coined by Michel Aflaq, the principal founder of the political party founded in Syria.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Saddam Hussein",
"Arab socialism",
"Ba'ath Party"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5932,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
21782
|
Joe Lebowski had admitted he'd given a letter to James Francis Jewell Archibald, a man who would later be arrested after WWI.
|
{
"evidence": [
"James Francis Jewell Archibald_sentence_6"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nJoe Lebowski had admitted he'd given a letter to James Francis Jewell Archibald, a man who would later be arrested after WWI.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"James Francis Jewell Archibald_sentence_6"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5933,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.597",
"value": "?search="
},
{
"operation": "search",
"time": "5.643",
"value": "Contains...Joe Lebowski"
},
{
"operation": "Now on",
"time": "6.608",
"value": "undefined"
},
{
"operation": "search",
"time": "22.597",
"value": "James Francis Jewell Archibald"
},
{
"operation": "Now on",
"time": "23.43",
"value": "James Francis Jewell Archibald"
},
{
"operation": "Page search",
"time": "27.142",
"value": "Joe Lebowski"
},
{
"operation": "Page search",
"time": "33.151",
"value": "letter"
},
{
"operation": "Highlighting",
"time": "50.4",
"value": "James Francis Jewell Archibald_sentence_6"
},
{
"operation": "finish",
"time": "68.803",
"value": "finish"
}
]
}
|
exfever_1860
|
Aristotle says Hebe is an ancient Greek religion.
|
{
"evidence": [
"Hebe (mythology)",
"Mount Olympus"
],
"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:\nAristotle says Hebe is an ancient Greek religion.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Hebe (mythology)",
"Mount Olympus"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5934,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
13316
|
The 1949 Florida State Seminoles football team won nine out of its ten games that sesaon, and went to the SuperBowl, which was played at Phillips Field and sponsored as a fundraiser by Egypt Temple Shrine until 1955.
|
{
"evidence": [
"1949 Florida State Seminoles football team_sentence_1"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe 1949 Florida State Seminoles football team won nine out of its ten games that sesaon, and went to the SuperBowl, which was played at Phillips Field and sponsored as a fundraiser by Egypt Temple Shrine until 1955.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"1949 Florida State Seminoles football team_sentence_1"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5935,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.786",
"value": "?search="
},
{
"operation": "search",
"time": "10.474",
"value": "1949 Florida State Seminoles football team"
},
{
"operation": "Now on",
"time": "12.2",
"value": "1949 Florida State Seminoles football team"
},
{
"operation": "Highlighting",
"time": "26.646",
"value": "1949 Florida State Seminoles football team_header_cell_0_4_0"
},
{
"operation": "Highlighting",
"time": "27.838",
"value": "1949 Florida State Seminoles football team_cell_0_4_1"
},
{
"operation": "Highlighting",
"time": "27.849",
"value": "1949 Florida State Seminoles football team_header_cell_0_4_0"
},
{
"operation": "Highlighting",
"time": "27.85",
"value": "1949 Florida State Seminoles football team_header_cell_0_2_0"
},
{
"operation": "Highlighting",
"time": "27.851",
"value": "1949 Florida State Seminoles football team_header_cell_0_1_0"
},
{
"operation": "Highlighting",
"time": "27.852",
"value": "1949 Florida State Seminoles football team_header_cell_0_0_0"
},
{
"operation": "Highlighting",
"time": "34.664",
"value": "1949 Florida State Seminoles football team_header_cell_1_0_3"
},
{
"operation": "Highlighting",
"time": "34.666",
"value": "1949 Florida State Seminoles football team_section_0"
},
{
"operation": "Highlighting",
"time": "36.233",
"value": "1949 Florida State Seminoles football team_cell_1_1_3"
},
{
"operation": "Highlighting",
"time": "36.244",
"value": "1949 Florida State Seminoles football team_header_cell_1_0_3"
},
{
"operation": "Highlighting",
"time": "36.904",
"value": "1949 Florida State Seminoles football team_cell_1_2_3"
},
{
"operation": "Highlighting",
"time": "37.537",
"value": "1949 Florida State Seminoles football team_cell_1_3_3"
},
{
"operation": "Highlighting",
"time": "38.17",
"value": "1949 Florida State Seminoles football team_cell_1_4_3"
},
{
"operation": "Highlighting",
"time": "38.777",
"value": "1949 Florida State Seminoles football team_cell_1_5_3"
},
{
"operation": "Highlighting",
"time": "39.431",
"value": "1949 Florida State Seminoles football team_cell_1_6_3"
},
{
"operation": "Highlighting",
"time": "40.128",
"value": "1949 Florida State Seminoles football team_cell_1_7_3"
},
{
"operation": "Highlighting",
"time": "40.762",
"value": "1949 Florida State Seminoles football team_cell_1_8_3"
},
{
"operation": "Highlighting",
"time": "41.275",
"value": "1949 Florida State Seminoles football team_cell_1_9_3"
},
{
"operation": "Highlighting",
"time": "41.909",
"value": "1949 Florida State Seminoles football team_cell_1_10_3"
},
{
"operation": "Highlighting deleted",
"time": "52.813",
"value": "1949 Florida State Seminoles football team_cell_1_1_3"
},
{
"operation": "Highlighting deleted",
"time": "53.336",
"value": "1949 Florida State Seminoles football team_cell_1_2_3"
},
{
"operation": "Highlighting deleted",
"time": "53.855",
"value": "1949 Florida State Seminoles football team_cell_1_3_3"
},
{
"operation": "Highlighting deleted",
"time": "54.418",
"value": "1949 Florida State Seminoles football team_cell_1_4_3"
},
{
"operation": "Highlighting deleted",
"time": "55.536",
"value": "1949 Florida State Seminoles football team_cell_1_5_3"
},
{
"operation": "Highlighting deleted",
"time": "56.034",
"value": "1949 Florida State Seminoles football team_cell_1_6_3"
},
{
"operation": "Highlighting deleted",
"time": "57.145",
"value": "1949 Florida State Seminoles football team_cell_1_7_3"
},
{
"operation": "Highlighting deleted",
"time": "57.672",
"value": "1949 Florida State Seminoles football team_cell_1_8_3"
},
{
"operation": "Highlighting deleted",
"time": "58.193",
"value": "1949 Florida State Seminoles football team_cell_1_9_3"
},
{
"operation": "Highlighting deleted",
"time": "58.712",
"value": "1949 Florida State Seminoles football team_cell_1_10_3"
},
{
"operation": "Highlighting deleted",
"time": "58.717",
"value": "1949 Florida State Seminoles football team_section_0"
},
{
"operation": "Highlighting deleted",
"time": "58.723",
"value": "1949 Florida State Seminoles football team_header_cell_1_0_3"
},
{
"operation": "Highlighting",
"time": "61.617",
"value": "1949 Florida State Seminoles football team_sentence_1"
},
{
"operation": "hyperlink",
"time": "68.637",
"value": "Cigar Bowl"
},
{
"operation": "Now on",
"time": "69.528",
"value": "Cigar Bowl"
},
{
"operation": "Highlighting",
"time": "77.13",
"value": "Cigar Bowl_header_cell_0_0_0"
},
{
"operation": "Highlighting",
"time": "77.132",
"value": "Cigar Bowl_title"
},
{
"operation": "Highlighting",
"time": "77.985",
"value": "Cigar Bowl_cell_0_1_1"
},
{
"operation": "Highlighting",
"time": "77.992",
"value": "Cigar Bowl_header_cell_0_1_0"
},
{
"operation": "Highlighting",
"time": "77.997",
"value": "Cigar Bowl_header_cell_0_0_0"
},
{
"operation": "Highlighting deleted",
"time": "81.026",
"value": "Cigar Bowl_cell_0_1_1"
},
{
"operation": "Highlighting deleted",
"time": "81.029",
"value": "Cigar Bowl_title"
},
{
"operation": "Highlighting deleted",
"time": "81.034",
"value": "Cigar Bowl_header_cell_0_1_0"
},
{
"operation": "Highlighting deleted",
"time": "81.037",
"value": "Cigar Bowl_header_cell_0_0_0"
},
{
"operation": "back-button-clicked",
"time": "100.916",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "101.049",
"value": "1949 Florida State Seminoles football team"
},
{
"operation": "Highlighting",
"time": "102.803",
"value": "1949 Florida State Seminoles football team_sentence_1"
},
{
"operation": "finish",
"time": "110.658",
"value": "finish"
}
]
}
|
75587
|
In 1696, Joseph Christophe painted as a "mai" for Notre-Dame the Miracle of the Loaves and Fishes and was received into the Academy in 1702, and in 1724 became painter to the Archduke Leopold, for whom he executed many portraits.
|
{
"evidence": [
"Joseph Christophe_sentence_1",
"Joseph Christophe_sentence_2"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nIn 1696, Joseph Christophe painted as a \"mai\" for Notre-Dame the Miracle of the Loaves and Fishes and was received into the Academy in 1702, and in 1724 became painter to the Archduke Leopold, for whom he executed many portraits.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Joseph Christophe_sentence_1",
"Joseph Christophe_sentence_2"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5936,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.032",
"value": "?search="
},
{
"operation": "search",
"time": "14.356",
"value": "Joseph Christophe"
},
{
"operation": "Now on",
"time": "15.694",
"value": "Joseph Christophe"
},
{
"operation": "Highlighting",
"time": "19.074",
"value": "Joseph Christophe_sentence_1"
},
{
"operation": "Highlighting",
"time": "25.235",
"value": "Joseph Christophe_sentence_2"
},
{
"operation": "finish",
"time": "56.195",
"value": "finish"
}
]
}
|
exfever_7769
|
Barack Hussein Obama II was named the 2009 Nobel Peace Prize laureate, a decision that drew a mixture of praise and criticism.
|
{
"evidence": [
"Barack Obama",
"President of the United States"
],
"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:\nBarack Hussein Obama II was named the 2009 Nobel Peace Prize laureate, a decision that drew a mixture of praise and criticism.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Barack Obama",
"President of the United States"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5937,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
84071
|
The 2000 Munster Senior Hurling Championship Final (sponsored by Budweiser) was a hurling match between Cork and Tipperary held on July 1, 2000 at the Semple Stadium, Thurles, County Tipperary.
|
{
"evidence": [
"2000 Munster Senior Hurling Championship Final_sentence_0"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe 2000 Munster Senior Hurling Championship Final (sponsored by Budweiser) was a hurling match between Cork and Tipperary held on July 1, 2000 at the Semple Stadium, Thurles, County Tipperary.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"2000 Munster Senior Hurling Championship Final_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5938,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.784",
"value": "?search="
},
{
"operation": "search",
"time": "50.9",
"value": "2000 Munster Senior Hurling Championship Final"
},
{
"operation": "Now on",
"time": "51.667",
"value": "2000 Munster Senior Hurling Championship Final"
},
{
"operation": "Highlighting",
"time": "59.367",
"value": "2000 Munster Senior Hurling Championship Final_sentence_0"
},
{
"operation": "finish",
"time": "67.008",
"value": "finish"
}
]
}
|
exfever_2142
|
The International Space Station is a space station in low Mars orbit that has a design principle that subdivides a system into smaller parts called modules.
|
{
"evidence": [
"International Space Station",
"Modular design"
],
"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 International Space Station is a space station in low Mars orbit that has a design principle that subdivides a system into smaller parts called modules.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"International Space Station",
"Modular design"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5939,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
80583
|
Camassia howellii can be found everywhere in the USA except in Oregon, making around three billion plants worldwide.
|
{
"evidence": [
"Camassia howellii_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:\nCamassia howellii can be found everywhere in the USA except in Oregon, making around three billion plants worldwide.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Camassia howellii_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5940,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.188",
"value": "?search="
},
{
"operation": "search",
"time": "15.591",
"value": "Camassia howellii"
},
{
"operation": "Now on",
"time": "16.588",
"value": "Camassia howellii"
},
{
"operation": "Highlighting",
"time": "21.483",
"value": "Camassia howellii_sentence_0"
},
{
"operation": "finish",
"time": "26.74",
"value": "finish"
}
]
}
|
15709
|
Alice Morgan Wright exhibited domestically at the Art Institute of Chicago, she was also a member of the Society of Independent Artists.
|
{
"evidence": [
"Alice Morgan Wright_sentence_9",
"Alice Morgan Wright_sentence_10"
],
"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:\nAlice Morgan Wright exhibited domestically at the Art Institute of Chicago, she was also a member of the Society of Independent Artists.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Alice Morgan Wright_sentence_9",
"Alice Morgan Wright_sentence_10"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5941,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.364",
"value": "?search="
},
{
"operation": "search",
"time": "10.796",
"value": "Alice Morgan Wright"
},
{
"operation": "Now on",
"time": "12.423",
"value": "Alice Morgan Wright"
},
{
"operation": "Page search",
"time": "25.454",
"value": "chicago"
},
{
"operation": "Highlighting",
"time": "32.255",
"value": "Alice Morgan Wright_sentence_9"
},
{
"operation": "Highlighting",
"time": "40.031",
"value": "Alice Morgan Wright_sentence_10"
},
{
"operation": "finish",
"time": "54.082",
"value": "finish"
}
]
}
|
exfever_5144
|
One Direction are an English-Irish pop boy band formed in London, England in 2011, which signed with the record label Syco Records by an English television personality who also judged The X Factor UK.
|
{
"evidence": [
"One Direction",
"Simon Cowell"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nOne Direction are an English-Irish pop boy band formed in London, England in 2011, which signed with the record label Syco Records by an English television personality who also judged The X Factor UK.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"One Direction",
"Simon Cowell"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5942,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
79145
|
Town Pump is a Montana-based chain of truck stops in services since 1953 and expanded into car washes in the 1980s.
|
{
"evidence": [
"Town Pump_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:\nTown Pump is a Montana-based chain of truck stops in services since 1953 and expanded into car washes in the 1980s.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Town Pump_sentence_2"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5943,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.755",
"value": "?search="
},
{
"operation": "search",
"time": "7.067",
"value": "Town Pump"
},
{
"operation": "Now on",
"time": "8.323",
"value": "Town Pump"
},
{
"operation": "Highlighting",
"time": "17.459",
"value": "Town Pump_sentence_0"
},
{
"operation": "Highlighting",
"time": "21.34",
"value": "Town Pump_sentence_1"
},
{
"operation": "Highlighting",
"time": "28.713",
"value": "Town Pump_sentence_2"
},
{
"operation": "Highlighting deleted",
"time": "61.614",
"value": "Town Pump_sentence_1"
},
{
"operation": "Highlighting deleted",
"time": "62.383",
"value": "Town Pump_sentence_0"
},
{
"operation": "finish",
"time": "79.818",
"value": "finish"
}
]
}
|
exfever_8325
|
The Patriot Act was enacted following the September 11 attacks and is related to the use of violence and fear to achieve an ideological aim.
|
{
"evidence": [
"Patriot Act",
"Terrorism"
],
"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 Patriot Act was enacted following the September 11 attacks and is related to the use of violence and fear to achieve an ideological aim.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Patriot Act",
"Terrorism"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5944,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_1045
|
Divergent is a 2014 film directed by Neil Burger, based on the debut novel of Italian novelist Veronica Roth, published by Harper Collins Children's Books in 2011.
|
{
"evidence": [
"Divergent (film)",
"Divergent (novel)"
],
"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:\nDivergent is a 2014 film directed by Neil Burger, based on the debut novel of Italian novelist Veronica Roth, published by Harper Collins Children's Books in 2011.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Divergent (film)",
"Divergent (novel)"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5945,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_3947
|
The Wars of the Roses were fought between supporters of two rival male-line descendants in which the family's major assets, including a territory over sovereign rules, had been passed from a father to his firstborn son.
|
{
"evidence": [
"Wars of the Roses",
"Cadet branch",
"Realm"
],
"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 Wars of the Roses were fought between supporters of two rival male-line descendants in which the family's major assets, including a territory over sovereign rules, had been passed from a father to his firstborn son.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Wars of the Roses",
"Cadet branch",
"Realm"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5946,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
30449
|
In Europe, the company manufactures in Imola, Italy since 2001 through local business TeaPak and has a branch in Hamburg, Germany, where the company name is YOGI TEA GmbH which means "company with limited liability".
|
{
"evidence": [
"Yogi Tea_sentence_6",
"Yogi Tea_sentence_7",
"Gesellschaft mit beschränkter Haftung_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:\nIn Europe, the company manufactures in Imola, Italy since 2001 through local business TeaPak and has a branch in Hamburg, Germany, where the company name is YOGI TEA GmbH which means \"company with limited liability\".\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Yogi Tea_sentence_6",
"Yogi Tea_sentence_7",
"Gesellschaft mit beschränkter Haftung_sentence_0"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5947,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.763",
"value": "?search="
},
{
"operation": "search",
"time": "14.811",
"value": "Yogi Tea"
},
{
"operation": "Now on",
"time": "16.05",
"value": "Yogi Tea"
},
{
"operation": "Highlighting",
"time": "28.285",
"value": "Yogi Tea_sentence_6"
},
{
"operation": "Highlighting",
"time": "37.015",
"value": "Yogi Tea_sentence_7"
},
{
"operation": "hyperlink",
"time": "45.98",
"value": "GmbH"
},
{
"operation": "Now on",
"time": "47.268",
"value": "Gesellschaft mit beschränkter Haftung"
},
{
"operation": "Highlighting",
"time": "51.913",
"value": "Gesellschaft mit beschränkter Haftung_sentence_0"
},
{
"operation": "back-button-clicked",
"time": "60.952",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "61.092",
"value": "Yogi Tea"
},
{
"operation": "finish",
"time": "102.832",
"value": "finish"
}
]
}
|
exfever_2839
|
A medical school is a tertiary educational institution whose professional degrees for physicians include a BMBS, which is the primary medical degree awarded by medical schools.
|
{
"evidence": [
"Medical school",
"Bachelor of Medicine, Bachelor of Surgery"
],
"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:\nA medical school is a tertiary educational institution whose professional degrees for physicians include a BMBS, which is the primary medical degree awarded by medical schools.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Medical school",
"Bachelor of Medicine, Bachelor of Surgery"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5948,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_4470
|
Wikipedia is a multilingual open online encyclopedia hosted by a 501 nonprofit organization headquartered in San Francisco, California, registered as a foundation under US law.
|
{
"evidence": [
"Wikipedia",
"Wikimedia Foundation"
],
"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:\nWikipedia is a multilingual open online encyclopedia hosted by a 501 nonprofit organization headquartered in San Francisco, California, registered as a foundation under US law.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Wikipedia",
"Wikimedia Foundation"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5949,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_5680
|
The Connaught Tunnel is in a southeastern Canadian province that borders one of the three prairie provinces.
|
{
"evidence": [
"Connaught Tunnel",
"British Columbia",
"Alberta"
],
"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 Connaught Tunnel is in a southeastern Canadian province that borders one of the three prairie provinces.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Connaught Tunnel",
"British Columbia",
"Alberta"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5950,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
76831
|
Zoran Zekić is a former striker who currently works for Maksimir.
|
{
"evidence": [
"Zoran Zekić_sentence_1"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nZoran Zekić is a former striker who currently works for Maksimir.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Zoran Zekić_sentence_1"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5951,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.043",
"value": "?search="
},
{
"operation": "search",
"time": "8.815",
"value": "Zoran Zekić"
},
{
"operation": "Now on",
"time": "9.568",
"value": "Zoran Zekić"
},
{
"operation": "Highlighting",
"time": "13.9",
"value": "Zoran Zekić_cell_0_5_1"
},
{
"operation": "Highlighting",
"time": "13.934",
"value": "Zoran Zekić_header_cell_0_5_0"
},
{
"operation": "Highlighting",
"time": "13.934",
"value": "Zoran Zekić_header_cell_0_0_0"
},
{
"operation": "Highlighting",
"time": "25.823",
"value": "Zoran Zekić_sentence_1"
},
{
"operation": "finish",
"time": "33.287",
"value": "finish"
}
]
}
|
78176
|
During the 37th International Film Festival of India, special presentations on film restoration and digital animation were made by the technicians from Italy, while it also paid homage to 11 Indian film personalities.
|
{
"evidence": [
"37th International Film Festival of India_sentence_2",
"37th International Film Festival of India_sentence_3"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nDuring the 37th International Film Festival of India, special presentations on film restoration and digital animation were made by the technicians from Italy, while it also paid homage to 11 Indian film personalities.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"37th International Film Festival of India_sentence_2",
"37th International Film Festival of India_sentence_3"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5952,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.299",
"value": "?search="
},
{
"operation": "search",
"time": "15.146",
"value": "37th International Film Festival of India"
},
{
"operation": "Now on",
"time": "15.882",
"value": "37th International Film Festival of India"
},
{
"operation": "Highlighting",
"time": "74.466",
"value": "37th International Film Festival of India_sentence_2"
},
{
"operation": "Highlighting",
"time": "76.477",
"value": "37th International Film Festival of India_sentence_3"
},
{
"operation": "finish",
"time": "86.716",
"value": "finish"
}
]
}
|
1675
|
Joe Kowalczyk is the Deputy Department Head of the Fleet Imagery Support Team (FIST) division of Nimitz Operational Intelligence Center, which serves as the "eyes" of the US Navy.
|
{
"evidence": [
"Nimitz Operational Intelligence Center_sentence_20",
"Nimitz Operational Intelligence Center_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:\nJoe Kowalczyk is the Deputy Department Head of the Fleet Imagery Support Team (FIST) division of Nimitz Operational Intelligence Center, which serves as the \"eyes\" of the US Navy.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Nimitz Operational Intelligence Center_sentence_20",
"Nimitz Operational Intelligence Center_sentence_17"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5953,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "search",
"time": "10.194",
"value": "Joe Kowalczyk"
},
{
"operation": "search",
"time": "19.538",
"value": "Fleet Imagery Support Team"
},
{
"operation": "hyperlink",
"time": "26.375",
"value": "Fleet Imagery Support Team (FIST) Nimitz Operational Intelligence Center_section_5"
},
{
"operation": "Highlighting",
"time": "42.063",
"value": "Nimitz Operational Intelligence Center_sentence_20"
},
{
"operation": "Highlighting deleted",
"time": "79.956",
"value": "Nimitz Operational Intelligence Center_sentence_20"
},
{
"operation": "hyperlink",
"time": "112.212",
"value": "2.1.2 Transnational Threat Department (TNT) Nimitz Operational Intelligence Center_section_4"
},
{
"operation": "Highlighting",
"time": "139.214",
"value": "Nimitz Operational Intelligence Center_sentence_20"
},
{
"operation": "search",
"time": "160.851",
"value": "Nimitz Operational Intelligence Center"
},
{
"operation": "Page search",
"time": "167.554",
"value": "eyes"
},
{
"operation": "Highlighting",
"time": "175.718",
"value": "Nimitz Operational Intelligence Center_sentence_17"
},
{
"operation": "finish",
"time": "189.415",
"value": "finish"
}
]
}
|
exfever_7714
|
The Statue of Liberty is a colossal neoclassical sculpture on a federally owned island in Upper New York Bay in the United States.
|
{
"evidence": [
"Statue of Liberty",
"Liberty Island"
],
"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 Statue of Liberty is a colossal neoclassical sculpture on a federally owned island in Upper New York Bay in the United States.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Statue of Liberty",
"Liberty Island"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5954,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
7108
|
Alexandra Zabelina, who first aspired in the sport gymnastics, missed the 1964 Summer Olympics but won individual Silver Prize at 1961 and 1966 world championships.
|
{
"evidence": [
"Alexandra Zabelina_sentence_5",
"Alexandra Zabelina_sentence_4",
"Alexandra Zabelina_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:\nAlexandra Zabelina, who first aspired in the sport gymnastics, missed the 1964 Summer Olympics but won individual Silver Prize at 1961 and 1966 world championships.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Alexandra Zabelina_sentence_5",
"Alexandra Zabelina_sentence_4",
"Alexandra Zabelina_sentence_3"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5955,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.862",
"value": "?search="
},
{
"operation": "search",
"time": "15.087",
"value": "Alexandra Zabelina"
},
{
"operation": "Now on",
"time": "15.932",
"value": "Alexandra Zabelina"
},
{
"operation": "Highlighting",
"time": "28.281",
"value": "Alexandra Zabelina_sentence_5"
},
{
"operation": "Highlighting",
"time": "34.671",
"value": "Alexandra Zabelina_sentence_4"
},
{
"operation": "Highlighting",
"time": "46.829",
"value": "Alexandra Zabelina_sentence_3"
},
{
"operation": "finish",
"time": "65.547",
"value": "finish"
}
]
}
|
exfever_4211
|
The West Wing is an American series of political dramas set in the White House during the administration of President Josiah Bartlet, which supported a political philosophy and a reform movement in the early 19th century.
|
{
"evidence": [
"The West Wing",
"Democratic Party (United States)",
"Progressivism in the United States"
],
"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 West Wing is an American series of political dramas set in the White House during the administration of President Josiah Bartlet, which supported a political philosophy and a reform movement in the early 19th century.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The West Wing",
"Democratic Party (United States)",
"Progressivism in the United States"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5956,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7948
|
The Last of Us Part II is a 2020 action-adventure game that focuses on two playable characters, Abby and the character portrayed by Ashley Johnson, that was escorted by a fictional character and main protagonist in the first game.
|
{
"evidence": [
"The Last of Us Part II",
"Ellie (The Last of Us)"
],
"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 Last of Us Part II is a 2020 action-adventure game that focuses on two playable characters, Abby and the character portrayed by Ashley Johnson, that was escorted by a fictional character and main protagonist in the first game.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Last of Us Part II",
"Ellie (The Last of Us)"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5957,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
89515
|
The text and musical notation of the Sharrer Parchment are complete, and the parchment is written with Latin letters.
|
{
"evidence": [
"Pergaminho Sharrer_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:\nThe text and musical notation of the Sharrer Parchment are complete, and the parchment is written with Latin letters.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Pergaminho Sharrer_sentence_5"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5958,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.518",
"value": "?search="
},
{
"operation": "search",
"time": "23.373",
"value": "Contains...Sharrer Parchment"
},
{
"operation": "Now on",
"time": "24.501",
"value": "undefined"
},
{
"operation": "search",
"time": "63.506",
"value": "Sharer Parchment"
},
{
"operation": "Now on",
"time": "64.465",
"value": "undefined"
},
{
"operation": "search",
"time": "105.58",
"value": "Contains...Sharrer Parchment"
},
{
"operation": "Now on",
"time": "106.345",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "108.49",
"value": "Pergaminho Sharrer"
},
{
"operation": "Now on",
"time": "109.421",
"value": "Pergaminho Sharrer"
},
{
"operation": "Highlighting",
"time": "129.039",
"value": "Pergaminho Sharrer_sentence_5"
},
{
"operation": "finish",
"time": "142.425",
"value": "finish"
}
]
}
|
exfever_8669
|
Ava Marie DuVernay is one of the world's highest-paid filmmakers.
|
{
"evidence": [
"Ava DuVernay",
"Middle of Nowhere (2012 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:\nAva Marie DuVernay is one of the world's highest-paid filmmakers.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Ava DuVernay",
"Middle of Nowhere (2012 film)"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5959,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_5889
|
Kitti's hog-nosed bat is the smallest species of bat and arguably the world's smallest mammal that occupies caves of carbonate sedimentary rocks that are mostly composed of the minerals calcite and aragonite.
|
{
"evidence": [
"Kitti's hog-nosed bat",
"Limestone"
],
"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:\nKitti's hog-nosed bat is the smallest species of bat and arguably the world's smallest mammal that occupies caves of carbonate sedimentary rocks that are mostly composed of the minerals calcite and aragonite.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Kitti's hog-nosed bat",
"Limestone"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5960,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
17411
|
The New Frontier was used in presidential candidate John F. Kennedy's acceptance speech in the 1960 United States presidential election for the 44th presidential election held Tuesday, November 8, 1960.
|
{
"evidence": [
"New Frontier_sentence_1",
"1960 United States presidential election_sentence_1",
"1960 United States presidential election_sentence_2"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe New Frontier was used in presidential candidate John F. Kennedy's acceptance speech in the 1960 United States presidential election for the 44th presidential election held Tuesday, November 8, 1960.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"New Frontier_sentence_1",
"1960 United States presidential election_sentence_1",
"1960 United States presidential election_sentence_2"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5961,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.577",
"value": "?search="
},
{
"operation": "search",
"time": "5.826",
"value": "New Frontier"
},
{
"operation": "Now on",
"time": "6.801",
"value": "New Frontier"
},
{
"operation": "Page search",
"time": "11.23",
"value": "John"
},
{
"operation": "Highlighting",
"time": "23.56",
"value": "New Frontier_sentence_1"
},
{
"operation": "Page search",
"time": "72.223",
"value": "presidential"
},
{
"operation": "hyperlink",
"time": "90.326",
"value": "1960 United States presidential election"
},
{
"operation": "Now on",
"time": "91.66",
"value": "1960 United States presidential election"
},
{
"operation": "Highlighting",
"time": "94.928",
"value": "1960 United States presidential election_sentence_1"
},
{
"operation": "Highlighting",
"time": "106.164",
"value": "1960 United States presidential election_sentence_2"
},
{
"operation": "finish",
"time": "111.032",
"value": "finish"
}
]
}
|
78195
|
Annie Dillard not only received bachelor's and master's degrees at Hollin's College but also met her husband, creative writing professor Richard Henry Wilde Dillard, who she claimed taught her everything she knew about writing.
|
{
"evidence": [
"Annie Dillard_sentence_18",
"Annie Dillard_sentence_24",
"Annie Dillard_sentence_92"
],
"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:\nAnnie Dillard not only received bachelor's and master's degrees at Hollin's College but also met her husband, creative writing professor Richard Henry Wilde Dillard, who she claimed taught her everything she knew about writing.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Annie Dillard_sentence_18",
"Annie Dillard_sentence_24",
"Annie Dillard_sentence_92"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5962,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.823",
"value": "?search="
},
{
"operation": "search",
"time": "19.172",
"value": "Annie Dillard"
},
{
"operation": "Now on",
"time": "20.025",
"value": "Annie Dillard"
},
{
"operation": "Highlighting",
"time": "69.742",
"value": "Annie Dillard_sentence_18"
},
{
"operation": "Highlighting",
"time": "75.104",
"value": "Annie Dillard_sentence_24"
},
{
"operation": "Page search",
"time": "171.942",
"value": "everything she knew"
},
{
"operation": "Highlighting",
"time": "209.446",
"value": "Annie Dillard_sentence_92"
},
{
"operation": "Page search",
"time": "230.185",
"value": "everything"
},
{
"operation": "Highlighting",
"time": "295.055",
"value": "Annie Dillard_sentence_21"
},
{
"operation": "Highlighting deleted",
"time": "305.168",
"value": "Annie Dillard_sentence_21"
},
{
"operation": "Highlighting",
"time": "341.695",
"value": "Annie Dillard_sentence_22"
},
{
"operation": "Highlighting deleted",
"time": "364.274",
"value": "Annie Dillard_sentence_22"
},
{
"operation": "finish",
"time": "385.614",
"value": "finish"
}
]
}
|
exfever_3155
|
The Conquest of the Planet of the Apes explores how the apes rebelled from humanity's ill treatment following a 1971 American science fiction film that starred several actors, including an American actor.
|
{
"evidence": [
"Conquest of the Planet of the Apes",
"Escape from the Planet of the Apes",
"Bradford Dillman"
],
"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 Conquest of the Planet of the Apes explores how the apes rebelled from humanity's ill treatment following a 1971 American science fiction film that starred several actors, including an American actor.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Conquest of the Planet of the Apes",
"Escape from the Planet of the Apes",
"Bradford Dillman"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5963,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
12075
|
In 1995 Tony Archer, a wildlife consultant working in Zanzibar, acquired a dried skin and skull in the village of Kitogani, in south-central Unguja, which was subsequently described as a new subspecies of servaline genet (G. s. archeri), a recently discovered subspecies of servaline genet endemic to Unguja Island in the Tanzanian archipelago of Zanzibar, named in honour of Archer.
|
{
"evidence": [
"Zanzibar servaline genet_sentence_6",
"Zanzibar servaline genet_sentence_7",
"Zanzibar servaline genet_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:\nIn 1995 Tony Archer, a wildlife consultant working in Zanzibar, acquired a dried skin and skull in the village of Kitogani, in south-central Unguja, which was subsequently described as a new subspecies of servaline genet (G. s. archeri), a recently discovered subspecies of servaline genet endemic to Unguja Island in the Tanzanian archipelago of Zanzibar, named in honour of Archer.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"Zanzibar servaline genet_sentence_6",
"Zanzibar servaline genet_sentence_7",
"Zanzibar servaline genet_sentence_0"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5964,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "5.132",
"value": "undefined"
},
{
"operation": "search",
"time": "26.644",
"value": "Tony Archer"
},
{
"operation": "Now on",
"time": "28.859",
"value": "Tony Archer"
},
{
"operation": "search",
"time": "54.246",
"value": "Servaline genet"
},
{
"operation": "Now on",
"time": "55.78",
"value": "Servaline genet"
},
{
"operation": "Highlighting",
"time": "144.432",
"value": "Servaline genet_item_0_4"
},
{
"operation": "Highlighting",
"time": "144.435",
"value": "Servaline genet_title"
},
{
"operation": "Highlighting",
"time": "144.437",
"value": "Servaline genet_section_2"
},
{
"operation": "search",
"time": "176.24",
"value": "Zanzibar servaline genet"
},
{
"operation": "Now on",
"time": "177.402",
"value": "Zanzibar servaline genet"
},
{
"operation": "Highlighting",
"time": "196.133",
"value": "Zanzibar servaline genet_sentence_6"
},
{
"operation": "Highlighting",
"time": "200.94",
"value": "Zanzibar servaline genet_sentence_7"
},
{
"operation": "Highlighting",
"time": "235.936",
"value": "Zanzibar servaline genet_sentence_0"
},
{
"operation": "finish",
"time": "301.686",
"value": "finish"
}
]
}
|
15506
|
Ethanolamine oxidaseis is a hormone that belongs to the family of oxidoreductases and has three subtrates.
|
{
"evidence": [
"Ethanolamine oxidase_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:\nEthanolamine oxidaseis is a hormone that belongs to the family of oxidoreductases and has three subtrates.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Ethanolamine oxidase_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5965,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.61",
"value": "?search="
},
{
"operation": "search",
"time": "38.297",
"value": "Ethanolamine oxidase"
},
{
"operation": "Now on",
"time": "39.058",
"value": "Ethanolamine oxidase"
},
{
"operation": "Highlighting",
"time": "48.677",
"value": "Ethanolamine oxidase_sentence_0"
},
{
"operation": "finish",
"time": "137.711",
"value": "finish"
}
]
}
|
84138
|
Andrew Brice set up a printing business in 1740 after publishing a collection of stories and poems --- and never published another poem again.
|
{
"evidence": [
"Andrew Brice_sentence_11"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nAndrew Brice set up a printing business in 1740 after publishing a collection of stories and poems --- and never published another poem again.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Andrew Brice_sentence_11"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5966,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.776",
"value": "?search="
},
{
"operation": "search",
"time": "10.38",
"value": "Andrew Brice"
},
{
"operation": "Now on",
"time": "11.346",
"value": "Andrew Brice"
},
{
"operation": "Highlighting",
"time": "29.808",
"value": "Andrew Brice_sentence_8"
},
{
"operation": "Highlighting",
"time": "49.042",
"value": "Andrew Brice_sentence_7"
},
{
"operation": "Highlighting deleted",
"time": "50.132",
"value": "Andrew Brice_sentence_7"
},
{
"operation": "Highlighting deleted",
"time": "50.717",
"value": "Andrew Brice_sentence_8"
},
{
"operation": "Highlighting",
"time": "54.653",
"value": "Andrew Brice_sentence_7"
},
{
"operation": "Highlighting",
"time": "56.391",
"value": "Andrew Brice_sentence_8"
},
{
"operation": "Highlighting",
"time": "58.176",
"value": "Andrew Brice_sentence_9"
},
{
"operation": "Highlighting",
"time": "72.463",
"value": "Andrew Brice_sentence_11"
},
{
"operation": "Highlighting deleted",
"time": "79.318",
"value": "Andrew Brice_sentence_7"
},
{
"operation": "Highlighting deleted",
"time": "80.963",
"value": "Andrew Brice_sentence_8"
},
{
"operation": "Highlighting deleted",
"time": "96.316",
"value": "Andrew Brice_sentence_9"
},
{
"operation": "finish",
"time": "105.329",
"value": "finish"
}
]
}
|
exfever_8919
|
Knight Bachelor is the basic rank granted to a man who has been knighted by the monarch.
|
{
"evidence": [
"Wallander (UK TV series)",
"Kenneth Branagh"
],
"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:\nKnight Bachelor is the basic rank granted to a man who has been knighted by the monarch.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Wallander (UK TV series)",
"Kenneth Branagh"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5967,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7392
|
Thirteen is a 2003 American teen drama novel featuring an American actor and filmmaker born on August 17, 1988.
|
{
"evidence": [
"Thirteen (2003 film)",
"Brady Corbet"
],
"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:\nThirteen is a 2003 American teen drama novel featuring an American actor and filmmaker born on August 17, 1988.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Thirteen (2003 film)",
"Brady Corbet"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5968,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
76099
|
Marlin Camp Carbine is a knife produced between 1985-2003 and it weighed around six pounds.
|
{
"evidence": [
"Marlin Camp Carbine_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:\nMarlin Camp Carbine is a knife produced between 1985-2003 and it weighed around six pounds.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Marlin Camp Carbine_sentence_2"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5969,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.855",
"value": "?search="
},
{
"operation": "search",
"time": "11.376",
"value": "Marlin Camp Carbine"
},
{
"operation": "Now on",
"time": "12.086",
"value": "Marlin Camp Carbine"
},
{
"operation": "Highlighting",
"time": "20.227",
"value": "Marlin Camp Carbine_sentence_2"
},
{
"operation": "finish",
"time": "35.727",
"value": "finish"
}
]
}
|
exfever_6663
|
Alter Ego is the second top-selling album by Tyrese Darnell Gibson.
|
{
"evidence": [
"Tyrese Gibson",
"Tyrese (album)"
],
"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:\nAlter Ego is the second top-selling album by Tyrese Darnell Gibson.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Tyrese Gibson",
"Tyrese (album)"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5970,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
67610
|
Marko Aleksejev is a retired Estonian high jumper from Tartu with a personal best jump at 2.28 m recorded in the International High Jump Meet in Bühl, Germany and placed 31st in the qualifying round of the 2004 Summer Olympics in Rio de Janeiro, Brazil..
|
{
"evidence": [
"Marko Aleksejev_sentence_3",
"Marko Aleksejev_sentence_4",
"Marko Aleksejev_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:\nMarko Aleksejev is a retired Estonian high jumper from Tartu with a personal best jump at 2.28 m recorded in the International High Jump Meet in Bühl, Germany and placed 31st in the qualifying round of the 2004 Summer Olympics in Rio de Janeiro, Brazil..\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Marko Aleksejev_sentence_3",
"Marko Aleksejev_sentence_4",
"Marko Aleksejev_sentence_5"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5971,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.38",
"value": "?search="
},
{
"operation": "search",
"time": "9.185",
"value": "Marko Aleksejev"
},
{
"operation": "Now on",
"time": "9.915",
"value": "Marko Aleksejev"
},
{
"operation": "Highlighting",
"time": "11.988",
"value": "Marko Aleksejev_sentence_0"
},
{
"operation": "Highlighting",
"time": "21.092",
"value": "Marko Aleksejev_cell_0_12_1"
},
{
"operation": "Highlighting",
"time": "21.099",
"value": "Marko Aleksejev_header_cell_0_12_0"
},
{
"operation": "Highlighting",
"time": "21.099",
"value": "Marko Aleksejev_header_cell_0_11_0"
},
{
"operation": "Highlighting deleted",
"time": "26.735",
"value": "Marko Aleksejev_cell_0_12_1"
},
{
"operation": "Highlighting deleted",
"time": "26.742",
"value": "Marko Aleksejev_header_cell_0_12_0"
},
{
"operation": "Highlighting deleted",
"time": "26.742",
"value": "Marko Aleksejev_header_cell_0_11_0"
},
{
"operation": "Highlighting",
"time": "29.057",
"value": "Marko Aleksejev_sentence_2"
},
{
"operation": "Highlighting",
"time": "65.103",
"value": "Marko Aleksejev_sentence_5"
},
{
"operation": "Highlighting deleted",
"time": "69.987",
"value": "Marko Aleksejev_sentence_5"
},
{
"operation": "Highlighting",
"time": "73.978",
"value": "Marko Aleksejev_sentence_3"
},
{
"operation": "Highlighting",
"time": "78.265",
"value": "Marko Aleksejev_sentence_4"
},
{
"operation": "Highlighting",
"time": "80.934",
"value": "Marko Aleksejev_sentence_5"
},
{
"operation": "Highlighting deleted",
"time": "91.173",
"value": "Marko Aleksejev_sentence_2"
},
{
"operation": "Highlighting deleted",
"time": "94.28",
"value": "Marko Aleksejev_sentence_0"
},
{
"operation": "finish",
"time": "102.103",
"value": "finish"
}
]
}
|
exfever_1068
|
Vietnam War raged has been resolved and suddenly go to peace afterwhile.
|
{
"evidence": [
"Bed-In",
"Amsterdam"
],
"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:\nVietnam War raged has been resolved and suddenly go to peace afterwhile.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Bed-In",
"Amsterdam"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5972,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
35951
|
Aeronáutica provides consumer services for airlines.
|
{
"evidence": [
"Aeronautica_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:\nAeronáutica provides consumer services for airlines.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Aeronautica_sentence_0"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5973,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.701",
"value": "?search="
},
{
"operation": "search",
"time": "14.263",
"value": "Aeronautica"
},
{
"operation": "Now on",
"time": "15.149",
"value": "Aeronautica"
},
{
"operation": "hyperlink",
"time": "26.137",
"value": "Aeronáutica (Angola)"
},
{
"operation": "Now on",
"time": "26.752",
"value": "Aeronáutica (Angola)"
},
{
"operation": "search",
"time": "45.384",
"value": "Aeronáutica"
},
{
"operation": "Now on",
"time": "46.151",
"value": "Aeronautica"
},
{
"operation": "hyperlink",
"time": "53.55",
"value": "Aeronáutica Industrial S.A."
},
{
"operation": "Now on",
"time": "54.143",
"value": "Aeronáutica Industrial S.A."
},
{
"operation": "search",
"time": "69.942",
"value": "Aeronáutica provides consumer services for airlines"
},
{
"operation": "Now on",
"time": "71.408",
"value": "undefined"
},
{
"operation": "search",
"time": "81.875",
"value": "consumer service for airlines"
},
{
"operation": "Now on",
"time": "83.191",
"value": "undefined"
},
{
"operation": "search",
"time": "101.166",
"value": "Aeronautica"
},
{
"operation": "Now on",
"time": "102.15",
"value": "Aeronautica"
},
{
"operation": "hyperlink",
"time": "110.198",
"value": "Aeronáutica (Seville Metro)"
},
{
"operation": "Now on",
"time": "110.793",
"value": "Aeronáutica (Seville Metro)"
},
{
"operation": "back-button-clicked",
"time": "116.381",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "116.466",
"value": "Aeronautica"
},
{
"operation": "Highlighting",
"time": "118.692",
"value": "Aeronautica_sentence_0"
},
{
"operation": "finish",
"time": "129.173",
"value": "finish"
}
]
}
|
18823
|
Red Spring Run flows south for several tenths of a mile, turns southwest but has no tributaries.
|
{
"evidence": [
"Red Spring Run_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:\nRed Spring Run flows south for several tenths of a mile, turns southwest but has no tributaries.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Red Spring Run_sentence_4"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5974,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.836",
"value": "?search="
},
{
"operation": "search",
"time": "11.169",
"value": "Red Spring Run"
},
{
"operation": "Now on",
"time": "12.217",
"value": "Red Spring Run"
},
{
"operation": "Highlighting",
"time": "121.895",
"value": "Red Spring Run_sentence_8"
},
{
"operation": "Highlighting",
"time": "121.898",
"value": "Red Spring Run_section_0"
},
{
"operation": "Highlighting",
"time": "133.924",
"value": "Red Spring Run_sentence_4"
},
{
"operation": "Highlighting deleted",
"time": "137.409",
"value": "Red Spring Run_sentence_8"
},
{
"operation": "Highlighting deleted",
"time": "137.416",
"value": "Red Spring Run_section_0"
},
{
"operation": "finish",
"time": "149.452",
"value": "finish"
}
]
}
|
exfever_6110
|
Sarah Michelle Prinze is an Italian actress, producer, and entrepreneur who appeared as Kendall Hart in an American television soap opera that aired on ABC from January 5, 1970, to September 23, 2011, and on The Online Network.
|
{
"evidence": [
"Sarah Michelle Gellar",
"All My Children"
],
"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:\nSarah Michelle Prinze is an Italian actress, producer, and entrepreneur who appeared as Kendall Hart in an American television soap opera that aired on ABC from January 5, 1970, to September 23, 2011, and on The Online Network.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Sarah Michelle Gellar",
"All My Children"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5975,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
49747
|
Duncan Lechesa played for different football teams before retiring in 2011 and becoming a coach.
|
{
"evidence": [
"Duncan Lechesa_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:\nDuncan Lechesa played for different football teams before retiring in 2011 and becoming a coach.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Duncan Lechesa_sentence_2"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5976,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.827",
"value": "?search="
},
{
"operation": "search",
"time": "23.65",
"value": "Duncan Lechesa"
},
{
"operation": "Now on",
"time": "24.723",
"value": "Duncan Lechesa"
},
{
"operation": "Highlighting",
"time": "40.673",
"value": "Duncan Lechesa_sentence_5"
},
{
"operation": "Highlighting",
"time": "81.682",
"value": "Duncan Lechesa_header_cell_0_11_0"
},
{
"operation": "Highlighting",
"time": "82.582",
"value": "Duncan Lechesa_cell_0_11_1"
},
{
"operation": "Highlighting",
"time": "82.601",
"value": "Duncan Lechesa_header_cell_0_11_0"
},
{
"operation": "Highlighting",
"time": "82.601",
"value": "Duncan Lechesa_header_cell_0_9_0"
},
{
"operation": "Highlighting deleted",
"time": "87.514",
"value": "Duncan Lechesa_cell_0_11_1"
},
{
"operation": "Highlighting deleted",
"time": "87.533",
"value": "Duncan Lechesa_header_cell_0_11_0"
},
{
"operation": "Highlighting deleted",
"time": "87.535",
"value": "Duncan Lechesa_header_cell_0_9_0"
},
{
"operation": "Highlighting",
"time": "120.568",
"value": "Duncan Lechesa_sentence_2"
},
{
"operation": "Highlighting",
"time": "139.436",
"value": "Duncan Lechesa_sentence_6"
},
{
"operation": "Highlighting",
"time": "140.665",
"value": "Duncan Lechesa_sentence_7"
},
{
"operation": "Highlighting deleted",
"time": "198.895",
"value": "Duncan Lechesa_sentence_5"
},
{
"operation": "Highlighting deleted",
"time": "199.428",
"value": "Duncan Lechesa_sentence_6"
},
{
"operation": "Highlighting deleted",
"time": "200.011",
"value": "Duncan Lechesa_sentence_7"
},
{
"operation": "finish",
"time": "209.16",
"value": "finish"
}
]
}
|
58763
|
Balach Marri, born on January 17, 1966, was the leader of the Baloch Liberation Army (listed as a terrorist organization by Pakistan, United Kingdom and the United States) who fought in the Insurgency in Balochistan.
|
{
"evidence": [
"Balach Marri_sentence_11",
"Balach Marri_sentence_0",
"Separatist movements of Pakistan_sentence_24"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nBalach Marri, born on January 17, 1966, was the leader of the Baloch Liberation Army (listed as a terrorist organization by Pakistan, United Kingdom and the United States) who fought in the Insurgency in Balochistan.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Balach Marri_sentence_11",
"Balach Marri_sentence_0",
"Separatist movements of Pakistan_sentence_24"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5977,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0",
"value": "?search="
},
{
"operation": "search",
"time": "0.16",
"value": "Balach Marri"
},
{
"operation": "Now on",
"time": "0.161",
"value": "Balach Marri"
},
{
"operation": "Highlighting",
"time": "0.168",
"value": "Balach Marri_sentence_0"
},
{
"operation": "Highlighting deleted",
"time": "0.17",
"value": "Balach Marri_sentence_0"
},
{
"operation": "Highlighting",
"time": "0.172",
"value": "Balach Marri_header_cell_0_3_0"
},
{
"operation": "Highlighting",
"time": "0.173",
"value": "Balach Marri_cell_0_3_1"
},
{
"operation": "Highlighting",
"time": "0.173",
"value": "Balach Marri_header_cell_0_3_0"
},
{
"operation": "Highlighting",
"time": "0.173",
"value": "Balach Marri_header_cell_0_0_0"
},
{
"operation": "Highlighting deleted",
"time": "0.19",
"value": "Balach Marri_cell_0_3_1"
},
{
"operation": "Highlighting deleted",
"time": "0.19",
"value": "Balach Marri_header_cell_0_3_0"
},
{
"operation": "Highlighting deleted",
"time": "0.19",
"value": "Balach Marri_header_cell_0_0_0"
},
{
"operation": "Highlighting",
"time": "0.194",
"value": "Balach Marri_sentence_11"
},
{
"operation": "Page search",
"time": "0.2",
"value": "unsurg"
},
{
"operation": "Highlighting",
"time": "0.216",
"value": "Balach Marri_sentence_0"
},
{
"operation": "search",
"time": "0.231",
"value": "Contains...Balochistan Liberation Army"
},
{
"operation": "Now on",
"time": "0.233",
"value": "undefined"
},
{
"operation": "search",
"time": "0.247",
"value": "Baloch Liberation Army"
},
{
"operation": "Now on",
"time": "0.247",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "0.263",
"value": "Separatist movements of Pakistan"
},
{
"operation": "Now on",
"time": "0.265",
"value": "Separatist movements of Pakistan"
},
{
"operation": "Page search",
"time": "0.272",
"value": "terrorist"
},
{
"operation": "Highlighting",
"time": "0.279",
"value": "Separatist movements of Pakistan_sentence_24"
},
{
"operation": "Page search",
"time": "0.294",
"value": "united"
},
{
"operation": "Page search",
"time": "0.314",
"value": "us"
},
{
"operation": "Page search",
"time": "0.324",
"value": "terrorist"
},
{
"operation": "Page search",
"time": "0.324",
"value": "terrorist"
},
{
"operation": "finish",
"time": "0.34",
"value": "finish"
},
{
"operation": "Now on",
"time": "1621444561.45",
"value": "Balach Marri"
},
{
"operation": "finish",
"time": "1621444563.662",
"value": "finish"
}
]
}
|
36500
|
In the Nisha Sharma dowry case, which is referred to as Dahez in Arabic, the groom's mother, Vidya Dalal, was arrested on 15 May and Navneet Rai came forward claiming to be having a relationship with her for longer than 5 years and having married her secretly.
|
{
"evidence": [
"Dowry system in India_sentence_3",
"Nisha Sharma dowry case_sentence_40",
"Nisha Sharma dowry case_sentence_41"
],
"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 Nisha Sharma dowry case, which is referred to as Dahez in Arabic, the groom's mother, Vidya Dalal, was arrested on 15 May and Navneet Rai came forward claiming to be having a relationship with her for longer than 5 years and having married her secretly.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Dowry system in India_sentence_3",
"Nisha Sharma dowry case_sentence_40",
"Nisha Sharma dowry case_sentence_41"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5978,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.603",
"value": "?search="
},
{
"operation": "search",
"time": "18.296",
"value": "Nisha Sharma dowry case"
},
{
"operation": "Now on",
"time": "19.636",
"value": "Nisha Sharma dowry case"
},
{
"operation": "Page search",
"time": "39.44",
"value": "Dahez"
},
{
"operation": "Page search",
"time": "75.086",
"value": "arabic"
},
{
"operation": "search",
"time": "128.064",
"value": "Dahez in Arabic"
},
{
"operation": "Now on",
"time": "128.911",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "136.143",
"value": "Dowry system in India"
},
{
"operation": "Now on",
"time": "137.749",
"value": "Dowry system in India"
},
{
"operation": "Page search",
"time": "151.094",
"value": "arabic"
},
{
"operation": "Highlighting",
"time": "153.735",
"value": "Dowry system in India_sentence_3"
},
{
"operation": "back-button-clicked",
"time": "156.04",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "157.048",
"value": "undefined"
},
{
"operation": "back-button-clicked",
"time": "158.042",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "158.044",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "158.099",
"value": "Nisha Sharma dowry case"
},
{
"operation": "Page search",
"time": "179.178",
"value": "15"
},
{
"operation": "Highlighting",
"time": "218.981",
"value": "Nisha Sharma dowry case_sentence_40"
},
{
"operation": "Highlighting",
"time": "240.885",
"value": "Nisha Sharma dowry case_sentence_41"
},
{
"operation": "finish",
"time": "281.026",
"value": "finish"
}
]
}
|
72508
|
The Ern Westmore Hollywood Glamour Show is hosted by Ern Westmore, Perc Westmore's twin.
|
{
"evidence": [
"The Ern Westmore Hollywood Glamour Show_sentence_1",
"Ern Westmore_sentence_1"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Ern Westmore Hollywood Glamour Show is hosted by Ern Westmore, Perc Westmore's twin.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"The Ern Westmore Hollywood Glamour Show_sentence_1",
"Ern Westmore_sentence_1"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5979,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.461",
"value": "?search="
},
{
"operation": "search",
"time": "16.627",
"value": "Contains...Ern Westmore Hollywood Glamour Show"
},
{
"operation": "Now on",
"time": "18.18",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "21.647",
"value": "The Ern Westmore Hollywood Glamour Show"
},
{
"operation": "Now on",
"time": "22.621",
"value": "The Ern Westmore Hollywood Glamour Show"
},
{
"operation": "Highlighting",
"time": "64.088",
"value": "The Ern Westmore Hollywood Glamour Show_sentence_1"
},
{
"operation": "hyperlink",
"time": "78.67",
"value": "Ern Westmore"
},
{
"operation": "Now on",
"time": "79.317",
"value": "Ern Westmore"
},
{
"operation": "Highlighting",
"time": "100.619",
"value": "Ern Westmore_sentence_1"
},
{
"operation": "finish",
"time": "134.231",
"value": "finish"
}
]
}
|
exfever_8965
|
The Armenian genocide was the systematic destruction of the Armenian people and identity in the Ottoman Empire during World War I which was implemented primarily through the mass murder of around one million Armenians during the forced march of prisoners of war or other captives or deportees in which individuals are left to die along the way.
|
{
"evidence": [
"Armenian Genocide",
"Death march"
],
"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 Armenian genocide was the systematic destruction of the Armenian people and identity in the Ottoman Empire during World War I which was implemented primarily through the mass murder of around one million Armenians during the forced march of prisoners of war or other captives or deportees in which individuals are left to die along the way.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Armenian Genocide",
"Death march"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5980,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
63287
|
The daughter of Audebert de Noves, Laura, was a great influence on humanist poet Francesco Petrarch's works.
|
{
"evidence": [
"Laura de Noves_sentence_2",
"Laura de Noves_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:\nThe daughter of Audebert de Noves, Laura, was a great influence on humanist poet Francesco Petrarch's works.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Laura de Noves_sentence_2",
"Laura de Noves_sentence_5"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5981,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.9",
"value": "?search="
},
{
"operation": "search",
"time": "17.525",
"value": "Contains...Audebert de Noves"
},
{
"operation": "Now on",
"time": "19.364",
"value": "undefined"
},
{
"operation": "search",
"time": "64.259",
"value": "Laura de Noves"
},
{
"operation": "Now on",
"time": "65.44",
"value": "Laura de Noves"
},
{
"operation": "Highlighting",
"time": "82.869",
"value": "Laura de Noves_sentence_2"
},
{
"operation": "Highlighting",
"time": "118.47",
"value": "Laura de Noves_sentence_5"
},
{
"operation": "finish",
"time": "176.997",
"value": "finish"
}
]
}
|
78907
|
In Nigeria it takes five years of higher education to become a pharmacist, a health professional who specializes in medicines.
|
{
"evidence": [
"National Universities Commission_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:\nIn Nigeria it takes five years of higher education to become a pharmacist, a health professional who specializes in medicines.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"National Universities Commission_sentence_0"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5982,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.873",
"value": "?search="
},
{
"operation": "search",
"time": "18",
"value": "Nigeria"
},
{
"operation": "Now on",
"time": "19.603",
"value": "Nigeria"
},
{
"operation": "hyperlink",
"time": "40.385",
"value": "7 Demographics Nigeria_section_29"
},
{
"operation": "hyperlink",
"time": "75.766",
"value": "Education in Nigeria"
},
{
"operation": "Now on",
"time": "77.239",
"value": "Education in Nigeria"
},
{
"operation": "Highlighting",
"time": "79.789",
"value": "Education in Nigeria_sentence_0"
},
{
"operation": "Highlighting",
"time": "79.79",
"value": "Education in Nigeria_title"
},
{
"operation": "hyperlink",
"time": "92.058",
"value": "5 Tertiary education Education in Nigeria_section_4"
},
{
"operation": "search",
"time": "180.786",
"value": "Contains...In Nigeria it takes five years of higher education to become a pharmacist, a health professional who specializes in medicines."
},
{
"operation": "Now on",
"time": "183.153",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "194.631",
"value": "Health care systems by country"
},
{
"operation": "Now on",
"time": "196.313",
"value": "Health care systems by country"
},
{
"operation": "search",
"time": "256.377",
"value": "nigerian health professionals"
},
{
"operation": "Now on",
"time": "258.005",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "276.428",
"value": "Health Nigeria_section_33"
},
{
"operation": "Now on",
"time": "277.399",
"value": "Nigeria#Health Nigeria section 33"
},
{
"operation": "hyperlink",
"time": "330.722",
"value": "Index of Nigeria-related articles"
},
{
"operation": "Now on",
"time": "332.123",
"value": "Index of Nigeria-related articles"
},
{
"operation": "hyperlink",
"time": "386.754",
"value": "Lists of country-related topics"
},
{
"operation": "Now on",
"time": "388.681",
"value": "Lists of country-related topics"
},
{
"operation": "hyperlink",
"time": "403.805",
"value": "Index of Nigeria-related articles"
},
{
"operation": "Now on",
"time": "404.438",
"value": "Index of Nigeria-related articles"
},
{
"operation": "hyperlink",
"time": "478.922",
"value": "List of universities in Nigeria"
},
{
"operation": "Now on",
"time": "480.704",
"value": "List of universities in Nigeria"
},
{
"operation": "hyperlink",
"time": "528.956",
"value": "National Universities Commission"
},
{
"operation": "Now on",
"time": "529.687",
"value": "National Universities Commission"
},
{
"operation": "Highlighting",
"time": "556.542",
"value": "National Universities Commission_sentence_0"
},
{
"operation": "hyperlink",
"time": "557.621",
"value": "higher education"
},
{
"operation": "Now on",
"time": "558.411",
"value": "Higher education"
},
{
"operation": "back-button-clicked",
"time": "577.649",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "577.764",
"value": "National Universities Commission"
},
{
"operation": "finish",
"time": "592.27",
"value": "finish"
}
]
}
|
11359
|
The Connecticut Route 89 continues north as Westford Road, still parallel to the Mount Hope River, as the road leaves the village, it turns west on Turnpike Road, then north on Ferrence Road.
|
{
"evidence": [
"Connecticut Route 89_sentence_7",
"Connecticut Route 89_sentence_8"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Connecticut Route 89 continues north as Westford Road, still parallel to the Mount Hope River, as the road leaves the village, it turns west on Turnpike Road, then north on Ferrence Road.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Connecticut Route 89_sentence_7",
"Connecticut Route 89_sentence_8"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5983,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.064",
"value": "?search="
},
{
"operation": "search",
"time": "10.235",
"value": "Connecticut Route 89"
},
{
"operation": "Now on",
"time": "11.152",
"value": "Connecticut Route 89"
},
{
"operation": "Page search",
"time": "18.206",
"value": "westford"
},
{
"operation": "Highlighting",
"time": "22.262",
"value": "Connecticut Route 89_sentence_7"
},
{
"operation": "Highlighting",
"time": "33.163",
"value": "Connecticut Route 89_sentence_8"
},
{
"operation": "finish",
"time": "66.339",
"value": "finish"
}
]
}
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.