id
stringlengths 1
12
| question
stringlengths 22
566
| golden_answers
dict | data_source
stringclasses 2
values | prompt
listlengths 1
1
| ability
stringclasses 6
values | reward_model
dict | extra_info
dict | metadata
dict |
|---|---|---|---|---|---|---|---|---|
exfever_8600
|
One World Trade Center is the main building of the rebuilt World Trade Center complex in a state located in the most populous city in the United Kingdom.
|
{
"evidence": [
"One World Trade Center",
"New York City"
],
"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 World Trade Center is the main building of the rebuilt World Trade Center complex in a state located in the most populous city in the United Kingdom.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"One World Trade Center",
"New York City"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2884,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_1992
|
The Hundred-Foot Journey is a 2014 French comedy-drama film directed by Lasse Hallström and starring an English actress who received an Academy Award and a British Academy Film Award for her portrayal of Queen Elizabeth II in The Queen.
|
{
"evidence": [
"The Hundred-Foot Journey (film)",
"Helen Mirren"
],
"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 Hundred-Foot Journey is a 2014 French comedy-drama film directed by Lasse Hallström and starring an English actress who received an Academy Award and a British Academy Film Award for her portrayal of Queen Elizabeth II in The Queen.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Hundred-Foot Journey (film)",
"Helen Mirren"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2885,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
54151
|
Íte, an early Irish nun grew up as an only child leading her to live in the Killeedy community where she was offered a big grant of land.
|
{
"evidence": [
"Íte of Killeedy_sentence_16"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nÍte, an early Irish nun grew up as an only child leading her to live in the Killeedy community where she was offered a big grant of land.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Íte of Killeedy_sentence_16"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2886,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.35",
"value": "?search="
},
{
"operation": "search",
"time": "9.579",
"value": "ite nun"
},
{
"operation": "Now on",
"time": "11.066",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "15.114",
"value": "Eadgyth of Aylesbury"
},
{
"operation": "Now on",
"time": "16.389",
"value": "Eadgyth of Aylesbury"
},
{
"operation": "back-button-clicked",
"time": "48.078",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "49.02",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "52.026",
"value": "Íte of Killeedy"
},
{
"operation": "Now on",
"time": "53.261",
"value": "Íte of Killeedy"
},
{
"operation": "Highlighting",
"time": "55.176",
"value": "Íte of Killeedy_sentence_0"
},
{
"operation": "Highlighting",
"time": "138.438",
"value": "Íte of Killeedy_sentence_16"
},
{
"operation": "Highlighting deleted",
"time": "145.12",
"value": "Íte of Killeedy_sentence_0"
},
{
"operation": "finish",
"time": "154.862",
"value": "finish"
}
]
}
|
93635
|
The earthquake of April 15, 1185 at the sixth hour of the day in the region of East Midlands, estimated at around five Mw magnitude with a depth of ten km, was one of the largest and most damaging earthquakes the British Isles had ever encountered.
|
{
"evidence": [
"List of earthquakes in the British Isles_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:\nThe earthquake of April 15, 1185 at the sixth hour of the day in the region of East Midlands, estimated at around five Mw magnitude with a depth of ten km, was one of the largest and most damaging earthquakes the British Isles had ever encountered.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"List of earthquakes in the British Isles_sentence_0"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2887,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.994",
"value": "?search="
},
{
"operation": "search",
"time": "254.803",
"value": "list of earthquakes in england"
},
{
"operation": "Now on",
"time": "258.307",
"value": "undefined"
},
{
"operation": "search",
"time": "272.073",
"value": "List of earthquakes in the British Isles"
},
{
"operation": "Now on",
"time": "273.048",
"value": "List of earthquakes in the British Isles"
},
{
"operation": "Highlighting",
"time": "281.792",
"value": "List of earthquakes in the British Isles_sentence_0"
},
{
"operation": "finish",
"time": "297.12",
"value": "finish"
}
]
}
|
exfever_8011
|
The Aegean Islands can be divided into several island groups.
|
{
"evidence": [
"Aegean Sea",
"Balkans"
],
"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 Aegean Islands can be divided into several island groups.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Aegean Sea",
"Balkans"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2888,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
24771
|
Jaimie Isaac, a Winnipeg-based Anishinaabe artist and curator, and a founding member of The Ephemerals Collective, also belongs to the boards of numerous Canadian art organizations.
|
{
"evidence": [
"Jaimie Isaac_sentence_0",
"Jaimie Isaac_sentence_4",
"Jaimie Isaac_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:\nJaimie Isaac, a Winnipeg-based Anishinaabe artist and curator, and a founding member of The Ephemerals Collective, also belongs to the boards of numerous Canadian art organizations.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Jaimie Isaac_sentence_0",
"Jaimie Isaac_sentence_4",
"Jaimie Isaac_sentence_5"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2889,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.853",
"value": "?search="
},
{
"operation": "search",
"time": "21.087",
"value": "Jamie Isaac"
},
{
"operation": "Now on",
"time": "21.748",
"value": "Jamie Isaac"
},
{
"operation": "Page search",
"time": "59.998",
"value": "ephemerals collective"
},
{
"operation": "search",
"time": "73.246",
"value": "The Ephemerals Collective"
},
{
"operation": "Now on",
"time": "74.592",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "79.078",
"value": "Jaimie Isaac"
},
{
"operation": "Now on",
"time": "79.508",
"value": "Jaimie Isaac"
},
{
"operation": "Highlighting",
"time": "81.031",
"value": "Jaimie Isaac_sentence_0"
},
{
"operation": "Highlighting",
"time": "91.901",
"value": "Jaimie Isaac_sentence_4"
},
{
"operation": "Highlighting",
"time": "98.421",
"value": "Jaimie Isaac_sentence_5"
},
{
"operation": "finish",
"time": "135.444",
"value": "finish"
}
]
}
|
exfever_7401
|
The Rolling Stones are an English rock band who pioneered the gritty, heavier-driven sound that came to define a loosely defined sub-genre of rock music typified by a heavy use of aggressive instruments such as the lowest-pitched member of the guitar family.
|
{
"evidence": [
"The Rolling Stones",
"Hard rock",
"Bass guitar"
],
"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 Rolling Stones are an English rock band who pioneered the gritty, heavier-driven sound that came to define a loosely defined sub-genre of rock music typified by a heavy use of aggressive instruments such as the lowest-pitched member of the guitar family.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Rolling Stones",
"Hard rock",
"Bass guitar"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2890,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
86377
|
Major active free-thinking parties in Nepal include the Nepal Communist party and Rastriya Janamorcha, both founded in 2018; Nepal Majdoor Kisan Party founded in 1975 and Communist Part of Nepal (Marxist-Leninist) founded in 2006.
|
{
"evidence": [
"Communist Party of Nepal (2013)_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:\nMajor active free-thinking parties in Nepal include the Nepal Communist party and Rastriya Janamorcha, both founded in 2018; Nepal Majdoor Kisan Party founded in 1975 and Communist Part of Nepal (Marxist-Leninist) founded in 2006.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Communist Party of Nepal (2013)_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2891,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.235",
"value": "?search="
},
{
"operation": "search",
"time": "19.688",
"value": "free-thinking parties in Nepal"
},
{
"operation": "Now on",
"time": "22.501",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "27.424",
"value": "Nepal Freedom of speech by country_section_24"
},
{
"operation": "Now on",
"time": "29.148",
"value": "Freedom of speech by country#Nepal Freedom of speech by country section 24"
},
{
"operation": "hyperlink",
"time": "31.857",
"value": "Censorship in Nepal"
},
{
"operation": "Now on",
"time": "32.455",
"value": "Censorship in Nepal"
},
{
"operation": "hyperlink",
"time": "40.607",
"value": "2 Political Censorship Censorship in Nepal_section_4"
},
{
"operation": "search",
"time": "98.053",
"value": "Nepal Communist party"
},
{
"operation": "Now on",
"time": "99.857",
"value": "Nepal Communist Party"
},
{
"operation": "Highlighting",
"time": "136.791",
"value": "Nepal Communist Party_sentence_3"
},
{
"operation": "Highlighting",
"time": "136.793",
"value": "Nepal Communist Party_title"
},
{
"operation": "hyperlink",
"time": "146.864",
"value": "Communist Party of Nepal (2013)"
},
{
"operation": "Now on",
"time": "147.726",
"value": "Communist Party of Nepal (2013)"
},
{
"operation": "Highlighting",
"time": "152.289",
"value": "Communist Party of Nepal (2013)_sentence_0"
},
{
"operation": "finish",
"time": "199.843",
"value": "finish"
}
]
}
|
exfever_6229
|
Wentworth is an Australian television drama series set in the modern-day and initially focuses on the early days in prison of Bea Smith, a character portrayed by a New Zealand stage and screen actress known for her role as Amazon Ephiny.
|
{
"evidence": [
"Wentworth (TV series)",
"Danielle Cormack"
],
"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:\nWentworth is an Australian television drama series set in the modern-day and initially focuses on the early days in prison of Bea Smith, a character portrayed by a New Zealand stage and screen actress known for her role as Amazon Ephiny.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Wentworth (TV series)",
"Danielle Cormack"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2892,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
46449
|
Dirba is a city in Malaysia with a recorded population in 2015 of 40,310.
|
{
"evidence": [
"Dirba_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:\nDirba is a city in Malaysia with a recorded population in 2015 of 40,310.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Dirba_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2893,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.59",
"value": "?search="
},
{
"operation": "search",
"time": "52.341",
"value": "Dirba"
},
{
"operation": "Now on",
"time": "53.261",
"value": "Dirba"
},
{
"operation": "Highlighting",
"time": "56.131",
"value": "Dirba_sentence_0"
},
{
"operation": "search",
"time": "77.387",
"value": "Dirba is a city in Malaysia"
},
{
"operation": "Now on",
"time": "78.634",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "86.183",
"value": "dibra is a city in malaysia"
},
{
"operation": "Now on",
"time": "87.07",
"value": "undefined"
},
{
"operation": "search",
"time": "101.48",
"value": "dibra"
},
{
"operation": "Now on",
"time": "102.696",
"value": "Dibra"
},
{
"operation": "search",
"time": "122.222",
"value": "dibra"
},
{
"operation": "Now on",
"time": "122.982",
"value": "Dibra"
},
{
"operation": "finish",
"time": "167.157",
"value": "finish"
}
]
}
|
exfever_4634
|
The Emmy Awards are the second-oldest of the four major annual American entertainment awards.
|
{
"evidence": [
"84th Academy Awards",
"Academy Awards",
"Douglas Fairbanks"
],
"label": "NOT ENOUGH INFO"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Emmy Awards are the second-oldest of the four major annual American entertainment awards.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"84th Academy Awards",
"Academy Awards",
"Douglas Fairbanks"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2894,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
19399
|
MacGregor, also called "the Wayne Gretzky of hockey writing", has a longstanding interest in the life of Tom Thomson (a Canadian artist active in the early 20th century), and has written both a novel and a biography exploring the artist's life and mysterious death.
|
{
"evidence": [
"Roy MacGregor_sentence_4",
"Roy MacGregor_sentence_3",
"Tom Thomson_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:\nMacGregor, also called \"the Wayne Gretzky of hockey writing\", has a longstanding interest in the life of Tom Thomson (a Canadian artist active in the early 20th century), and has written both a novel and a biography exploring the artist's life and mysterious death.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"Roy MacGregor_sentence_4",
"Roy MacGregor_sentence_3",
"Tom Thomson_sentence_1"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2895,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.733",
"value": "?search="
},
{
"operation": "search",
"time": "9.569",
"value": "\"the Wayne Gretzky of hockey writing\""
},
{
"operation": "Now on",
"time": "10.45",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "12.011",
"value": "Roy MacGregor"
},
{
"operation": "Now on",
"time": "12.772",
"value": "Roy MacGregor"
},
{
"operation": "Page search",
"time": "18.459",
"value": "the Wayne Gretzky of hockey writing"
},
{
"operation": "Highlighting",
"time": "19.878",
"value": "Roy MacGregor_sentence_4"
},
{
"operation": "Highlighting",
"time": "26.576",
"value": "Roy MacGregor_sentence_3"
},
{
"operation": "hyperlink",
"time": "33.73",
"value": "Tom Thomson"
},
{
"operation": "Now on",
"time": "36.543",
"value": "Tom Thomson"
},
{
"operation": "Highlighting",
"time": "37.806",
"value": "Tom Thomson_sentence_1"
},
{
"operation": "finish",
"time": "70.062",
"value": "finish"
}
]
}
|
exfever_3833
|
The song We Found Love also appears on DJ Calvin Harris' third studio album, which focuses more on the genre of electronic dance music.
|
{
"evidence": [
"We Found Love",
"18 Months",
"Electro house"
],
"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 song We Found Love also appears on DJ Calvin Harris' third studio album, which focuses more on the genre of electronic dance music.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"We Found Love",
"18 Months",
"Electro house"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2896,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_1671
|
Franklin Delano Roosevelt was the most renowned president of America.
|
{
"evidence": [
"Franklin D. Roosevelt",
"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:\nFranklin Delano Roosevelt was the most renowned president of America.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Franklin D. Roosevelt",
"President of the United States"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2897,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
66000
|
Kenneth Maillard is the co-founder and managing director of Belvedere Management Limited, a financial services group based in Mauritius that claims to have $16 billion of assets under administration, and also a director of Fulhold Pharma plc subsidiary Ful Hold Ltd.
|
{
"evidence": [
"Kenneth Maillard_sentence_2",
"Belvedere Management_sentence_0",
"Kenneth Maillard_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:\nKenneth Maillard is the co-founder and managing director of Belvedere Management Limited, a financial services group based in Mauritius that claims to have $16 billion of assets under administration, and also a director of Fulhold Pharma plc subsidiary Ful Hold Ltd.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Kenneth Maillard_sentence_2",
"Belvedere Management_sentence_0",
"Kenneth Maillard_sentence_4"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2898,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.958",
"value": "?search="
},
{
"operation": "search",
"time": "34.042",
"value": "Kenneth Maillard"
},
{
"operation": "Now on",
"time": "34.823",
"value": "Kenneth Maillard"
},
{
"operation": "Page search",
"time": "69.76",
"value": "co-fo"
},
{
"operation": "Highlighting",
"time": "71.739",
"value": "Kenneth Maillard_sentence_2"
},
{
"operation": "Page search",
"time": "142.51",
"value": "financial"
},
{
"operation": "hyperlink",
"time": "148.947",
"value": "Belvedere Management group"
},
{
"operation": "Now on",
"time": "150.66",
"value": "Belvedere Management"
},
{
"operation": "Highlighting",
"time": "151.724",
"value": "Belvedere Management_sentence_0"
},
{
"operation": "back-button-clicked",
"time": "161.193",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "161.263",
"value": "Kenneth Maillard"
},
{
"operation": "Highlighting",
"time": "163.716",
"value": "Kenneth Maillard_sentence_4"
},
{
"operation": "finish",
"time": "200.77",
"value": "finish"
}
]
}
|
exfever_5589
|
Christopher Nolan's 2015 film Batman Begins focuses on the actions of individuals who usually possess extraordinary abilities.
|
{
"evidence": [
"Batman Begins",
"Superhero 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:\nChristopher Nolan's 2015 film Batman Begins focuses on the actions of individuals who usually possess extraordinary abilities.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Batman Begins",
"Superhero film"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2899,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_5403
|
The X-Files has the highest revenue of any Television show.
|
{
"evidence": [
"The X-Files (franchise)",
"Media franchise",
"Media (communication)"
],
"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 X-Files has the highest revenue of any Television show.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The X-Files (franchise)",
"Media franchise",
"Media (communication)"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2900,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
78122
|
There are 19 Silver Diner restaurants in the mid-Atlantic region of the United States, including an airport location in the AirMall at Baltimore/Washington International Thurgood Marshall Airport.
|
{
"evidence": [
"Silver Diner_sentence_5",
"Silver Diner_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:\nThere are 19 Silver Diner restaurants in the mid-Atlantic region of the United States, including an airport location in the AirMall at Baltimore/Washington International Thurgood Marshall Airport.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Silver Diner_sentence_5",
"Silver Diner_sentence_6"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2901,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.368",
"value": "?search="
},
{
"operation": "search",
"time": "25.035",
"value": "Silver Diner"
},
{
"operation": "Now on",
"time": "25.804",
"value": "Silver Diner"
},
{
"operation": "Highlighting",
"time": "28.913",
"value": "Silver Diner_cell_0_4_1"
},
{
"operation": "Highlighting",
"time": "28.916",
"value": "Silver Diner_header_cell_0_4_0"
},
{
"operation": "Page search",
"time": "49.41",
"value": "mid"
},
{
"operation": "Highlighting",
"time": "50.984",
"value": "Silver Diner_sentence_5"
},
{
"operation": "Highlighting deleted",
"time": "53.881",
"value": "Silver Diner_cell_0_4_1"
},
{
"operation": "Highlighting deleted",
"time": "53.884",
"value": "Silver Diner_header_cell_0_4_0"
},
{
"operation": "Highlighting",
"time": "87.113",
"value": "Silver Diner_sentence_6"
},
{
"operation": "finish",
"time": "98.558",
"value": "finish"
}
]
}
|
exfever_1820
|
Hungarians are a nation and an ethnic group from a landlocked Central American country.
|
{
"evidence": [
"Hungarians",
"Hungary"
],
"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:\nHungarians are a nation and an ethnic group from a landlocked Central American country.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Hungarians",
"Hungary"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2902,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_2355
|
The Walt Disney Company, commonly known as Disney, is an American multinational form of activity that holds the attention and interest of an audience or gives pleasure and delight.
|
{
"evidence": [
"The Walt Disney Company",
"Entertainment"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Walt Disney Company, commonly known as Disney, is an American multinational form of activity that holds the attention and interest of an audience or gives pleasure and delight.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Walt Disney Company",
"Entertainment"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2903,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_2816
|
Boston is the most populous city in a region comprising six states in the Northeastern United States.
|
{
"evidence": [
"Boston",
"New England"
],
"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:\nBoston is the most populous city in a region comprising six states in the Northeastern United States.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Boston",
"New England"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2904,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_6787
|
The Who are an English rock band that consists of a bass guitarist and a singer nicknamed The Ox.
|
{
"evidence": [
"The Who",
"John Entwistle"
],
"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 Who are an English rock band that consists of a bass guitarist and a singer nicknamed The Ox.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Who",
"John Entwistle"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2905,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_293
|
Iron Man 3 is a 2013 American superhero film based on Iron Man by a publisher of American comic books and related media which lately acquired by The Walt Disney Company.
|
{
"evidence": [
"Iron Man 3",
"Marvel Comics"
],
"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:\nIron Man 3 is a 2013 American superhero film based on Iron Man by a publisher of American comic books and related media which lately acquired by The Walt Disney Company.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Iron Man 3",
"Marvel Comics"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2906,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_6581
|
Dangal is a 2016 Indian Hindi-language biographical sports drama film produced by an Indian actor, film director, and producer who works in the Indian Hindi-language film industry based in Delhi popularly known as Bollywood.
|
{
"evidence": [
"Dangal (film)",
"Aamir Khan",
"Bollywood"
],
"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:\nDangal is a 2016 Indian Hindi-language biographical sports drama film produced by an Indian actor, film director, and producer who works in the Indian Hindi-language film industry based in Delhi popularly known as Bollywood.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Dangal (film)",
"Aamir Khan",
"Bollywood"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2907,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
92847
|
200,000 people invaded the Hyde Park but the soldiers held back and the meeting proceeded as planned
|
{
"evidence": [
"Reform League_sentence_46",
"Reform League_sentence_48",
"Reform League_sentence_49"
],
"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:\n200,000 people invaded the Hyde Park but the soldiers held back and the meeting proceeded as planned\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"Reform League_sentence_46",
"Reform League_sentence_48",
"Reform League_sentence_49"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2908,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.634",
"value": "undefined"
},
{
"operation": "search",
"time": "25.11",
"value": "Hyde Park and Regent's Park bombings"
},
{
"operation": "Now on",
"time": "26.185",
"value": "Hyde Park and Regent's Park bombings"
},
{
"operation": "search",
"time": "52.897",
"value": "Hyde Park"
},
{
"operation": "Now on",
"time": "53.746",
"value": "Hyde Park"
},
{
"operation": "search",
"time": "79.906",
"value": "Contains...200,000 people invaded the Hyde Park"
},
{
"operation": "Now on",
"time": "82.361",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "88.236",
"value": "Hyde Park demonstration of 1866 Reform League_section_3"
},
{
"operation": "Now on",
"time": "89.129",
"value": "Reform League#Hyde Park demonstration of 1866 Reform League section 3"
},
{
"operation": "Highlighting",
"time": "94.483",
"value": "Reform League_sentence_46"
},
{
"operation": "Highlighting",
"time": "103.996",
"value": "Reform League_sentence_48"
},
{
"operation": "Highlighting",
"time": "110.036",
"value": "Reform League_sentence_49"
},
{
"operation": "finish",
"time": "127.064",
"value": "finish"
}
]
}
|
40116
|
Helga Moddansdóttir's sibilings were Angus the Generous and sisters Frakkök and Þorleif, all four children having both Norse and Gaelic names as their father was Celtic and their mother was Norse.
|
{
"evidence": [
"Helga Moddansdóttir_sentence_11",
"Helga Moddansdóttir_sentence_12",
"Helga Moddansdóttir_sentence_13"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nHelga Moddansdóttir's sibilings were Angus the Generous and sisters Frakkök and Þorleif, all four children having both Norse and Gaelic names as their father was Celtic and their mother was Norse.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Helga Moddansdóttir_sentence_11",
"Helga Moddansdóttir_sentence_12",
"Helga Moddansdóttir_sentence_13"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2909,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.165",
"value": "?search="
},
{
"operation": "search",
"time": "56.895",
"value": "Helga Moddansdóttir"
},
{
"operation": "Now on",
"time": "57.966",
"value": "Helga Moddansdóttir"
},
{
"operation": "Highlighting",
"time": "75.646",
"value": "Helga Moddansdóttir_sentence_11"
},
{
"operation": "Highlighting",
"time": "116.359",
"value": "Helga Moddansdóttir_sentence_12"
},
{
"operation": "Highlighting",
"time": "118.879",
"value": "Helga Moddansdóttir_sentence_13"
},
{
"operation": "finish",
"time": "136.111",
"value": "finish"
}
]
}
|
69834
|
Costa Serena had its inaugural event on May 19, 2007 in Marseille, France.
|
{
"evidence": [
"Costa Serena_sentence_7",
"Costa Serena_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:\nCosta Serena had its inaugural event on May 19, 2007 in Marseille, France.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Costa Serena_sentence_7",
"Costa Serena_sentence_8"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2910,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.454",
"value": "undefined"
},
{
"operation": "search",
"time": "10.883",
"value": "Costa Serena"
},
{
"operation": "Now on",
"time": "12.065",
"value": "Costa Serena"
},
{
"operation": "Highlighting",
"time": "44.443",
"value": "Costa Serena_sentence_7"
},
{
"operation": "Highlighting",
"time": "66.009",
"value": "Costa Serena_sentence_8"
},
{
"operation": "finish",
"time": "106.63",
"value": "finish"
}
]
}
|
52988
|
Nezamysl was one of seven mythical princes born to Premysl the Plloughman, who was a free peasant who married Libuse and becamse prince of the Bohemian Czechs.
|
{
"evidence": [
"Nezamysl_sentence_0",
"Přemysl the Ploughman_sentence_3",
"Přemysl the Ploughman_sentence_7"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nNezamysl was one of seven mythical princes born to Premysl the Plloughman, who was a free peasant who married Libuse and becamse prince of the Bohemian Czechs.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Nezamysl_sentence_0",
"Přemysl the Ploughman_sentence_3",
"Přemysl the Ploughman_sentence_7"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2911,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.772",
"value": "?search="
},
{
"operation": "search",
"time": "14.32",
"value": "Nezamysl"
},
{
"operation": "Now on",
"time": "14.973",
"value": "Nezamysl"
},
{
"operation": "Highlighting",
"time": "53.371",
"value": "Nezamysl_sentence_0"
},
{
"operation": "hyperlink",
"time": "112.884",
"value": "Přemysl the Ploughman"
},
{
"operation": "Now on",
"time": "113.55",
"value": "Přemysl the Ploughman"
},
{
"operation": "hyperlink",
"time": "165.842",
"value": "Přemyslid dynasty"
},
{
"operation": "Now on",
"time": "169.541",
"value": "Přemyslid dynasty"
},
{
"operation": "Highlighting",
"time": "183.235",
"value": "Přemyslid dynasty_sentence_5"
},
{
"operation": "Highlighting",
"time": "183.236",
"value": "Přemyslid dynasty_title"
},
{
"operation": "Highlighting",
"time": "183.237",
"value": "Přemyslid dynasty_section_0"
},
{
"operation": "Highlighting deleted",
"time": "186.303",
"value": "Přemyslid dynasty_sentence_5"
},
{
"operation": "Highlighting deleted",
"time": "186.305",
"value": "Přemyslid dynasty_title"
},
{
"operation": "Highlighting deleted",
"time": "186.311",
"value": "Přemyslid dynasty_section_0"
},
{
"operation": "back-button-clicked",
"time": "189.473",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "190.086",
"value": "Přemysl the Ploughman"
},
{
"operation": "back-button-clicked",
"time": "232.169",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "232.176",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "232.237",
"value": "Nezamysl"
},
{
"operation": "hyperlink",
"time": "274.053",
"value": "Přemysl the Ploughman"
},
{
"operation": "Now on",
"time": "275.055",
"value": "Přemysl, the Ploughman"
},
{
"operation": "Highlighting",
"time": "278.461",
"value": "Přemysl the Ploughman_sentence_3"
},
{
"operation": "Highlighting",
"time": "289.733",
"value": "Přemysl the Ploughman_sentence_7"
},
{
"operation": "finish",
"time": "298.958",
"value": "finish"
}
]
}
|
exfever_7866
|
Michael Edward Love co-founded the Beach Boys with his cousins Brian, Dennis, and an American musician, singer, and songwriter.
|
{
"evidence": [
"Mike Love",
"Carl Wilson"
],
"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:\nMichael Edward Love co-founded the Beach Boys with his cousins Brian, Dennis, and an American musician, singer, and songwriter.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Mike Love",
"Carl Wilson"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2912,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
58564
|
Italian monitor Alfredo Cappellini (wrecked off Ancona on 16 November 1917) displaced 2000 long tons and was powered by one 265-indicated-horsepower engine it also reached maximum speed of 3.76 knots
|
{
"evidence": [
"Italian monitor Alfredo Cappellini_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:\nItalian monitor Alfredo Cappellini (wrecked off Ancona on 16 November 1917) displaced 2000 long tons and was powered by one 265-indicated-horsepower engine it also reached maximum speed of 3.76 knots\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"Italian monitor Alfredo Cappellini_sentence_5"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2913,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.19",
"value": "?search="
},
{
"operation": "search",
"time": "15.139",
"value": "Alfredo Cappellini (S507)"
},
{
"operation": "Now on",
"time": "16.816",
"value": "USS Capitaine (SS-336)"
},
{
"operation": "hyperlink",
"time": "36.243",
"value": "1.3 Alfredo Cappellini (S 507) USS Capitaine (SS-336)_section_3"
},
{
"operation": "Highlighting",
"time": "49.585",
"value": "USS Capitaine (SS-336)_sentence_18"
},
{
"operation": "Highlighting",
"time": "49.586",
"value": "USS Capitaine (SS-336)_title"
},
{
"operation": "Highlighting",
"time": "49.587",
"value": "USS Capitaine (SS-336)_section_0"
},
{
"operation": "Highlighting",
"time": "49.588",
"value": "USS Capitaine (SS-336)_section_3"
},
{
"operation": "search",
"time": "109.203",
"value": "Contains...Alfredo Cappellini"
},
{
"operation": "Now on",
"time": "110.384",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "119.112",
"value": "Service history Italian submarine Comandante Cappellini (1939)_section_0"
},
{
"operation": "Now on",
"time": "119.928",
"value": "Italian submarine Comandante Cappellini (1939)#Service history Italian submarine Comandante Cappellini .281939.29 section 0"
},
{
"operation": "search",
"time": "155.178",
"value": "Contains...wrecked off Ancona on 16 November 1917"
},
{
"operation": "Now on",
"time": "157.091",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "161.074",
"value": "Italian monitor Alfredo Cappellini"
},
{
"operation": "Now on",
"time": "161.592",
"value": "Italian monitor Alfredo Cappellini"
},
{
"operation": "Highlighting",
"time": "176.339",
"value": "Italian monitor Alfredo Cappellini_sentence_14"
},
{
"operation": "Highlighting",
"time": "176.342",
"value": "Italian monitor Alfredo Cappellini_section_1"
},
{
"operation": "Highlighting",
"time": "191.456",
"value": "Italian monitor Alfredo Cappellini_cell_0_14_1"
},
{
"operation": "Highlighting",
"time": "191.464",
"value": "Italian monitor Alfredo Cappellini_header_cell_0_13_0"
},
{
"operation": "Highlighting",
"time": "204.633",
"value": "Italian monitor Alfredo Cappellini_sentence_5"
},
{
"operation": "Highlighting deleted",
"time": "206.656",
"value": "Italian monitor Alfredo Cappellini_cell_0_14_1"
},
{
"operation": "Highlighting deleted",
"time": "206.661",
"value": "Italian monitor Alfredo Cappellini_header_cell_0_13_0"
},
{
"operation": "Highlighting deleted",
"time": "207.931",
"value": "Italian monitor Alfredo Cappellini_sentence_14"
},
{
"operation": "Highlighting deleted",
"time": "207.933",
"value": "Italian monitor Alfredo Cappellini_section_1"
},
{
"operation": "finish",
"time": "219.926",
"value": "finish"
}
]
}
|
exfever_476
|
Jeffrey Nero Hardy is an American professional wrestler widely known for his time with World Wrestling Entertainment, Inc., an American professional boxing promotion.
|
{
"evidence": [
"Jeff Hardy",
"WWE"
],
"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:\nJeffrey Nero Hardy is an American professional wrestler widely known for his time with World Wrestling Entertainment, Inc., an American professional boxing promotion.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Jeff Hardy",
"WWE"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2914,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
4238
|
Italian monitor Alfredo Cappellini displaced 2000 long tons and was powered by one 265-indicated-horsepower engine it also reached maximum speed of 3.76 knots
|
{
"evidence": [
"Italian monitor Alfredo Cappellini_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:\nItalian monitor Alfredo Cappellini displaced 2000 long tons and was powered by one 265-indicated-horsepower engine it also reached maximum speed of 3.76 knots\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Italian monitor Alfredo Cappellini_sentence_5"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2915,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "search",
"time": "22.915",
"value": "Alfredo Cappellini (S507)"
},
{
"operation": "Page search",
"time": "57.245",
"value": "monitor"
},
{
"operation": "Page search",
"time": "111.077",
"value": "2000"
},
{
"operation": "hyperlink",
"time": "128.77",
"value": "Italian ship Alfredo Cappellini"
},
{
"operation": "Highlighting",
"time": "136.002",
"value": "Italian ship Alfredo Cappellini_item_0_0"
},
{
"operation": "Highlighting",
"time": "136.008",
"value": "Italian ship Alfredo Cappellini_title"
},
{
"operation": "hyperlink",
"time": "141.05",
"value": "Italian monitor Alfredo Cappellini"
},
{
"operation": "Highlighting",
"time": "160.902",
"value": "Italian monitor Alfredo Cappellini_sentence_5"
},
{
"operation": "finish",
"time": "182.921",
"value": "finish"
}
]
}
|
exfever_6944
|
Conan is the first comic book by Louise Simonson.
|
{
"evidence": [
"Louise Simonson",
"Cable (comics)"
],
"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:\nConan is the first comic book by Louise Simonson.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Louise Simonson",
"Cable (comics)"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2916,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_2058
|
The House of the Spirits is a 1993 period drama film starring a Spanish actor born on August 10, 1960.
|
{
"evidence": [
"The House of the Spirits (film)",
"Antonio Banderas"
],
"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 House of the Spirits is a 1993 period drama film starring a Spanish actor born on August 10, 1960.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The House of the Spirits (film)",
"Antonio Banderas"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2917,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_1053
|
Christopher Michael Benoit was an Australian professional wrestler who worked for an American professional wrestling promotion founded by Ted Turner.
|
{
"evidence": [
"Chris Benoit",
"World Championship Wrestling"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nChristopher Michael Benoit was an Australian professional wrestler who worked for an American professional wrestling promotion founded by Ted Turner.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Chris Benoit",
"World Championship Wrestling"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2918,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
75257
|
Hicks, an American politician and attorney born on May 29, 1915, enrolled in the University of Washington School of Law, where he earned a Juris Doctor in 1948; then he established a small private practice in Pierce County, Washington.
|
{
"evidence": [
"Floyd Hicks_sentence_0",
"Floyd Hicks_sentence_7",
"Floyd Hicks_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:\nHicks, an American politician and attorney born on May 29, 1915, enrolled in the University of Washington School of Law, where he earned a Juris Doctor in 1948; then he established a small private practice in Pierce County, Washington.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Floyd Hicks_sentence_0",
"Floyd Hicks_sentence_7",
"Floyd Hicks_sentence_8"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2919,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.342",
"value": "?search="
},
{
"operation": "search",
"time": "43.971",
"value": "enrolled in the University of Washington School of Law"
},
{
"operation": "Now on",
"time": "48.631",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "51.456",
"value": "Floyd Hicks"
},
{
"operation": "Now on",
"time": "52.14",
"value": "Floyd Hicks"
},
{
"operation": "Highlighting",
"time": "62.925",
"value": "Floyd Hicks_sentence_0"
},
{
"operation": "Highlighting",
"time": "95.812",
"value": "Floyd Hicks_sentence_7"
},
{
"operation": "Highlighting",
"time": "107.988",
"value": "Floyd Hicks_sentence_8"
},
{
"operation": "finish",
"time": "119.878",
"value": "finish"
}
]
}
|
exfever_4615
|
The Punisher, who appeared in American comic books published by Marvel Comics, does perform actions that are morally correct, but it is not always for the right reasons and he is known as a protagonist that makes key decisions that affect the plot, primarily influencing the story and propelling it forward.
|
{
"evidence": [
"Punisher",
"Antihero",
"Protagonist"
],
"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 Punisher, who appeared in American comic books published by Marvel Comics, does perform actions that are morally correct, but it is not always for the right reasons and he is known as a protagonist that makes key decisions that affect the plot, primarily influencing the story and propelling it forward.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Punisher",
"Antihero",
"Protagonist"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2920,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_214
|
Underworld is a series of dark fantasy action horror films starring an Asian actress and model who appeared in a 1994 drama adventure film co-written and directed by Gabriel Axel.
|
{
"evidence": [
"Underworld (film series)",
"Kate Beckinsale",
"Prince of Jutland"
],
"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:\nUnderworld is a series of dark fantasy action horror films starring an Asian actress and model who appeared in a 1994 drama adventure film co-written and directed by Gabriel Axel.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Underworld (film series)",
"Kate Beckinsale",
"Prince of Jutland"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2921,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_4139
|
Geraldine Leigh Chaplin is the daughter of an English comic actor, filmmaker, and composer who rose to fame in the era of silent film.
|
{
"evidence": [
"Geraldine Chaplin",
"Charlie Chaplin"
],
"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:\nGeraldine Leigh Chaplin is the daughter of an English comic actor, filmmaker, and composer who rose to fame in the era of silent film.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Geraldine Chaplin",
"Charlie Chaplin"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2922,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_349
|
The Golden Globe Awards were earned most by Barbra Streisand.
|
{
"evidence": [
"Geraldine Chaplin",
"Golden Globe Award",
"Academy Awards"
],
"label": "NOT ENOUGH INFO"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Golden Globe Awards were earned most by Barbra Streisand.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Geraldine Chaplin",
"Golden Globe Award",
"Academy Awards"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2923,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
45513
|
Wrangel Mansion, built in 1885 by architect Nikolay Doroshenko, was nationalized and used to house a kindergarten after the end of the Civil War but was closed in the 1990s.
|
{
"evidence": [
"Wrangel Mansion_sentence_0",
"Wrangel Mansion_sentence_9",
"Wrangel Mansion_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:\nWrangel Mansion, built in 1885 by architect Nikolay Doroshenko, was nationalized and used to house a kindergarten after the end of the Civil War but was closed in the 1990s.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Wrangel Mansion_sentence_0",
"Wrangel Mansion_sentence_9",
"Wrangel Mansion_sentence_10"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2924,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.287",
"value": "?search="
},
{
"operation": "search",
"time": "11.951",
"value": "Wrangel Mansion"
},
{
"operation": "Now on",
"time": "12.66",
"value": "Wrangel Mansion"
},
{
"operation": "Highlighting",
"time": "19.439",
"value": "Wrangel Mansion_sentence_0"
},
{
"operation": "Highlighting",
"time": "54.207",
"value": "Wrangel Mansion_sentence_9"
},
{
"operation": "Highlighting",
"time": "61.119",
"value": "Wrangel Mansion_sentence_10"
},
{
"operation": "finish",
"time": "84.15",
"value": "finish"
}
]
}
|
13464
|
Telephone numbers in Bahrain, which are regularly updated by the Bahrain Telecommunications Regulatory Authority, are generally eight digits long and majority of the fixed land line number start with 2.
|
{
"evidence": [
"Telephone numbers in Bahrain_sentence_3"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nTelephone numbers in Bahrain, which are regularly updated by the Bahrain Telecommunications Regulatory Authority, are generally eight digits long and majority of the fixed land line number start with 2.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Telephone numbers in Bahrain_sentence_3"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2925,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.594",
"value": "?search="
},
{
"operation": "search",
"time": "21.974",
"value": "Telephone numbers in Bahrain"
},
{
"operation": "Now on",
"time": "22.852",
"value": "Telephone numbers in Bahrain"
},
{
"operation": "Highlighting",
"time": "30.248",
"value": "Telephone numbers in Bahrain_sentence_2"
},
{
"operation": "Highlighting deleted",
"time": "31.312",
"value": "Telephone numbers in Bahrain_sentence_2"
},
{
"operation": "Highlighting",
"time": "34.974",
"value": "Telephone numbers in Bahrain_sentence_0"
},
{
"operation": "Highlighting",
"time": "58.029",
"value": "Telephone numbers in Bahrain_sentence_2"
},
{
"operation": "Highlighting",
"time": "70.739",
"value": "Telephone numbers in Bahrain_sentence_3"
},
{
"operation": "Highlighting deleted",
"time": "73.229",
"value": "Telephone numbers in Bahrain_sentence_2"
},
{
"operation": "Highlighting deleted",
"time": "74.022",
"value": "Telephone numbers in Bahrain_sentence_0"
},
{
"operation": "finish",
"time": "88.237",
"value": "finish"
}
]
}
|
exfever_3399
|
Alexander Hamilton was the youngest influential interpreter and promoter of the U.S. Constitution.
|
{
"evidence": [
"Alexander Hamilton",
"American Revolutionary War"
],
"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:\nAlexander Hamilton was the youngest influential interpreter and promoter of the U.S. Constitution.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Alexander Hamilton",
"American Revolutionary War"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2926,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
54805
|
The Shaw Library, or the Founder's Room, is a general-purpose library and a common room at the London School of Economics and Political Science located on the sixth floor of the Old Building.
|
{
"evidence": [
"LSE Shaw Library_sentence_0",
"LSE Shaw Library_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 Shaw Library, or the Founder's Room, is a general-purpose library and a common room at the London School of Economics and Political Science located on the sixth floor of the Old Building.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"LSE Shaw Library_sentence_0",
"LSE Shaw Library_sentence_1"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2927,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.062",
"value": "?search="
},
{
"operation": "search",
"time": "9.275",
"value": "Shaw Library"
},
{
"operation": "Now on",
"time": "13.273",
"value": "Shaw Library"
},
{
"operation": "search",
"time": "102.561",
"value": "The Founder's Room"
},
{
"operation": "Now on",
"time": "104.965",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "107.039",
"value": "LSE Shaw Library"
},
{
"operation": "Now on",
"time": "107.732",
"value": "LSE Shaw Library"
},
{
"operation": "Highlighting",
"time": "108.928",
"value": "LSE Shaw Library_sentence_0"
},
{
"operation": "Highlighting",
"time": "116.103",
"value": "LSE Shaw Library_sentence_1"
},
{
"operation": "finish",
"time": "127.055",
"value": "finish"
}
]
}
|
75594
|
Cyrus of Panopolis was an author of eulogies, epigrams and epic poetry who never had a chance to serve bureaucratic positions in the palace.
|
{
"evidence": [
"Cyrus of Panopolis_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:\nCyrus of Panopolis was an author of eulogies, epigrams and epic poetry who never had a chance to serve bureaucratic positions in the palace.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Cyrus of Panopolis_sentence_4"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2928,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.979",
"value": "?search="
},
{
"operation": "search",
"time": "7.404",
"value": "Cyrus of Panopolis"
},
{
"operation": "Now on",
"time": "7.925",
"value": "Cyrus of Panopolis"
},
{
"operation": "Highlighting",
"time": "39.011",
"value": "Cyrus of Panopolis_sentence_3"
},
{
"operation": "Highlighting",
"time": "47.62",
"value": "Cyrus of Panopolis_sentence_4"
},
{
"operation": "Highlighting deleted",
"time": "49.77",
"value": "Cyrus of Panopolis_sentence_3"
},
{
"operation": "finish",
"time": "129.818",
"value": "finish"
}
]
}
|
exfever_2456
|
Akira Toriyama is a character designer for a 2007 role-playing video game developed by Mistwalker and Artoon.
|
{
"evidence": [
"Akira Toriyama",
"Blue Dragon (video game)"
],
"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:\nAkira Toriyama is a character designer for a 2007 role-playing video game developed by Mistwalker and Artoon.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Akira Toriyama",
"Blue Dragon (video game)"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2929,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
14801
|
Evi Nemeth joined the University of Colorado Boulder in 1980 before serving a computing facility manager for five years.
|
{
"evidence": [
"Evi Nemeth_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:\nEvi Nemeth joined the University of Colorado Boulder in 1980 before serving a computing facility manager for five years.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Evi Nemeth_sentence_6"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2930,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.281",
"value": "?search="
},
{
"operation": "search",
"time": "24.66",
"value": "Evi Nemeth"
},
{
"operation": "Now on",
"time": "25.356",
"value": "Evi Nemeth"
},
{
"operation": "Highlighting",
"time": "55.755",
"value": "Evi Nemeth_sentence_5"
},
{
"operation": "Highlighting",
"time": "59.21",
"value": "Evi Nemeth_sentence_6"
},
{
"operation": "Highlighting deleted",
"time": "70.208",
"value": "Evi Nemeth_sentence_5"
},
{
"operation": "finish",
"time": "75.352",
"value": "finish"
}
]
}
|
exfever_1205
|
John Clayton Mayer is an American singer, songwriter, and guitarist who was born and raised in a county that contains four of the state's largest cities.
|
{
"evidence": [
"John Mayer",
"Fairfield County, Connecticut"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nJohn Clayton Mayer is an American singer, songwriter, and guitarist who was born and raised in a county that contains four of the state's largest cities.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"John Mayer",
"Fairfield County, Connecticut"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2931,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_896
|
Audi is a German automotive manufacturer headquartered in a place where they founded a name given to several groups, both real and fictitious.
|
{
"evidence": [
"Audi",
"Ingolstadt",
"Illuminati"
],
"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:\nAudi is a German automotive manufacturer headquartered in a place where they founded a name given to several groups, both real and fictitious.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Audi",
"Ingolstadt",
"Illuminati"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2932,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_8036
|
The Man in the High Castle was the second alternative history novel by Philip K. Dick.
|
{
"evidence": [
"The Man in the High Castle",
"Ward Moore",
"The Encyclopedia of Science Fiction"
],
"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 Man in the High Castle was the second alternative history novel by Philip K. Dick.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Man in the High Castle",
"Ward Moore",
"The Encyclopedia of Science Fiction"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2933,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7920
|
"Diamonds" is a song recorded by Rihanna for her seventh studio album that was recorded between June and November 2012, during the promotion of her sixth studio album, Talk That Talk.
|
{
"evidence": [
"Diamonds (Rihanna song)",
"Unapologetic"
],
"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\"Diamonds\" is a song recorded by Rihanna for her seventh studio album that was recorded between June and November 2012, during the promotion of her sixth studio album, Talk That Talk.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Diamonds (Rihanna song)",
"Unapologetic"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2934,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
87513
|
Amazing Adventures of a Nobody is a reality-travel entertainment TV series that follows Leon, who was presented a new five pound note for each new day, and the public could buy him things, but he was never allowed to be given money directly.
|
{
"evidence": [
"Amazing Adventures of a Nobody_sentence_0",
"Amazing Adventures of a Nobody_sentence_5",
"Amazing Adventures of a Nobody_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:\nAmazing Adventures of a Nobody is a reality-travel entertainment TV series that follows Leon, who was presented a new five pound note for each new day, and the public could buy him things, but he was never allowed to be given money directly.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Amazing Adventures of a Nobody_sentence_0",
"Amazing Adventures of a Nobody_sentence_5",
"Amazing Adventures of a Nobody_sentence_6"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2935,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.146",
"value": "?search="
},
{
"operation": "search",
"time": "13.077",
"value": "Amazing Adventures of a Nobody"
},
{
"operation": "Now on",
"time": "13.986",
"value": "Amazing Adventures of a Nobody"
},
{
"operation": "Highlighting",
"time": "32.637",
"value": "Amazing Adventures of a Nobody_sentence_0"
},
{
"operation": "Highlighting",
"time": "41.062",
"value": "Amazing Adventures of a Nobody_sentence_1"
},
{
"operation": "Highlighting deleted",
"time": "65.605",
"value": "Amazing Adventures of a Nobody_sentence_1"
},
{
"operation": "Highlighting",
"time": "68.53",
"value": "Amazing Adventures of a Nobody_sentence_5"
},
{
"operation": "Highlighting",
"time": "72.47",
"value": "Amazing Adventures of a Nobody_sentence_6"
},
{
"operation": "finish",
"time": "84.564",
"value": "finish"
}
]
}
|
41135
|
Edward Charles Stirling had four daughters, he introduced a formal bill for women's suffrage and he was chairman of the New Zealand Museum committee.
|
{
"evidence": [
"Edward Charles Stirling_sentence_26",
"Edward Charles Stirling_sentence_22"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nEdward Charles Stirling had four daughters, he introduced a formal bill for women's suffrage and he was chairman of the New Zealand Museum committee.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Edward Charles Stirling_sentence_26",
"Edward Charles Stirling_sentence_22"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2936,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.672",
"value": "?search="
},
{
"operation": "search",
"time": "8.673",
"value": "Edward Charles Stirling"
},
{
"operation": "Now on",
"time": "10.703",
"value": "Edward Charles Stirling"
},
{
"operation": "hyperlink",
"time": "14.22",
"value": "4 Family Edward Charles Stirling_section_3"
},
{
"operation": "Highlighting",
"time": "35.702",
"value": "Edward Charles Stirling_sentence_16"
},
{
"operation": "Highlighting",
"time": "39.055",
"value": "Edward Charles Stirling_sentence_15"
},
{
"operation": "Highlighting",
"time": "43.443",
"value": "Edward Charles Stirling_sentence_17"
},
{
"operation": "Highlighting",
"time": "46.683",
"value": "Edward Charles Stirling_sentence_18"
},
{
"operation": "Highlighting deleted",
"time": "47.345",
"value": "Edward Charles Stirling_sentence_16"
},
{
"operation": "Highlighting deleted",
"time": "47.747",
"value": "Edward Charles Stirling_sentence_15"
},
{
"operation": "Highlighting",
"time": "83.349",
"value": "Edward Charles Stirling_sentence_22"
},
{
"operation": "Highlighting deleted",
"time": "92.964",
"value": "Edward Charles Stirling_sentence_22"
},
{
"operation": "Highlighting",
"time": "95.84",
"value": "Edward Charles Stirling_sentence_26"
},
{
"operation": "Highlighting deleted",
"time": "101.678",
"value": "Edward Charles Stirling_sentence_18"
},
{
"operation": "Highlighting deleted",
"time": "102.996",
"value": "Edward Charles Stirling_sentence_17"
},
{
"operation": "Highlighting",
"time": "103.99",
"value": "Edward Charles Stirling_sentence_22"
},
{
"operation": "finish",
"time": "117.223",
"value": "finish"
}
]
}
|
exfever_331
|
iO Theater was an improv theater and training center in Rockford that had notable alumni, including an American comedian who is also the co-founder of the improvisational-comedy troupe Upright Citizen Brigade.
|
{
"evidence": [
"IO Theater",
"Amy Poehler"
],
"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:\niO Theater was an improv theater and training center in Rockford that had notable alumni, including an American comedian who is also the co-founder of the improvisational-comedy troupe Upright Citizen Brigade.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"IO Theater",
"Amy Poehler"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2937,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_6796
|
Annie Parisse is a British actress who starred in the soap opera As the World Turns, for which she was nominated for an American accolade bestowed by the New York-based National Academy of Television Arts and Sciences in recognition of excellence in American daytime television programming.
|
{
"evidence": [
"Annie Parisse",
"Daytime Emmy Award"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nAnnie Parisse is a British actress who starred in the soap opera As the World Turns, for which she was nominated for an American accolade bestowed by the New York-based National Academy of Television Arts and Sciences in recognition of excellence in American daytime television programming.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Annie Parisse",
"Daytime Emmy Award"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2938,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_1038
|
Temple of the Dog was an American rock supergroup consisting of members such as an American singer best known as the lead vocalist and rhythm guitarist for an American rock band formed in Seattle, Washington, who helped to popularize grunge music, a style of alternative rock that developed in the American Pacific Northwest.
|
{
"evidence": [
"Temple of the Dog",
"Chris Cornell",
"Soundgarden"
],
"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:\nTemple of the Dog was an American rock supergroup consisting of members such as an American singer best known as the lead vocalist and rhythm guitarist for an American rock band formed in Seattle, Washington, who helped to popularize grunge music, a style of alternative rock that developed in the American Pacific Northwest.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Temple of the Dog",
"Chris Cornell",
"Soundgarden"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2939,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
59076
|
Metallica Stadium Tour was a co-headlining concert tour by American rock bands Guns N'Roses
|
{
"evidence": [
"Metallica Stadium Tour_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:\nMetallica Stadium Tour was a co-headlining concert tour by American rock bands Guns N'Roses\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Metallica Stadium Tour_sentence_0"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2940,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.357",
"value": "?search="
},
{
"operation": "search",
"time": "11.028",
"value": "Metallica Stadium Tour"
},
{
"operation": "Now on",
"time": "12.767",
"value": "Metallica Stadium Tour"
},
{
"operation": "Highlighting",
"time": "26.771",
"value": "Metallica Stadium Tour_sentence_0"
},
{
"operation": "finish",
"time": "47.595",
"value": "finish"
}
]
}
|
exfever_6667
|
L.A. Law is an American legal drama television series that ran for eight seasons on an American English-language commercial broadcast television and radio network headquartered at 30 Rockefeller Plaza in New York City.
|
{
"evidence": [
"L.A. Law",
"NBC"
],
"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:\nL.A. Law is an American legal drama television series that ran for eight seasons on an American English-language commercial broadcast television and radio network headquartered at 30 Rockefeller Plaza in New York City.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"L.A. Law",
"NBC"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2941,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_8480
|
A hate group is a dance group that advocates and practices angry emotional responses towards certain people or ideas, usually related to disliking something.
|
{
"evidence": [
"Hate group",
"Hatred"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nA hate group is a dance group that advocates and practices angry emotional responses towards certain people or ideas, usually related to disliking something.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Hate group",
"Hatred"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2942,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
38051
|
The Kocatepe Mosque, the largest mosque in Ankara, Turkey, was built with the help of Ahmet Hamdi Akseki and the founding members of the "Society to Build a Mosque in Yenişehir, Ankara.”
|
{
"evidence": [
"Kocatepe Mosque_sentence_0",
"Kocatepe Mosque_sentence_1",
"Kocatepe Mosque_sentence_3",
"Kocatepe Mosque_sentence_4",
"Kocatepe Mosque_sentence_5",
"Kocatepe Mosque_sentence_6"
],
"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 Kocatepe Mosque, the largest mosque in Ankara, Turkey, was built with the help of Ahmet Hamdi Akseki and the founding members of the \"Society to Build a Mosque in Yenişehir, Ankara.”\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Kocatepe Mosque_sentence_0",
"Kocatepe Mosque_sentence_1",
"Kocatepe Mosque_sentence_3",
"Kocatepe Mosque_sentence_4",
"Kocatepe Mosque_sentence_5",
"Kocatepe Mosque_sentence_6"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2943,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.166",
"value": "undefined"
},
{
"operation": "search",
"time": "18.288",
"value": "Kocatepe Mosque"
},
{
"operation": "Now on",
"time": "19.833",
"value": "Kocatepe Mosque"
},
{
"operation": "Highlighting",
"time": "25.648",
"value": "Kocatepe Mosque_sentence_0"
},
{
"operation": "Highlighting",
"time": "36.858",
"value": "Kocatepe Mosque_sentence_1"
},
{
"operation": "Highlighting deleted",
"time": "41.423",
"value": "Kocatepe Mosque_sentence_1"
},
{
"operation": "Page search",
"time": "109.105",
"value": "ahmet"
},
{
"operation": "Page search",
"time": "109.263",
"value": "ahmet"
},
{
"operation": "Page search",
"time": "109.432",
"value": "ahmet"
},
{
"operation": "Page search",
"time": "109.594",
"value": "ahmet"
},
{
"operation": "Page search",
"time": "109.783",
"value": "ahmet"
},
{
"operation": "Page search",
"time": "109.952",
"value": "ahmet"
},
{
"operation": "Highlighting",
"time": "123.721",
"value": "Kocatepe Mosque_sentence_3"
},
{
"operation": "Highlighting",
"time": "209.211",
"value": "Kocatepe Mosque_sentence_4"
},
{
"operation": "Highlighting deleted",
"time": "294.585",
"value": "Kocatepe Mosque_sentence_0"
},
{
"operation": "Highlighting deleted",
"time": "326.619",
"value": "Kocatepe Mosque_sentence_4"
},
{
"operation": "Highlighting",
"time": "328.048",
"value": "Kocatepe Mosque_sentence_0"
},
{
"operation": "Highlighting",
"time": "328.54",
"value": "Kocatepe Mosque_sentence_1"
},
{
"operation": "Highlighting deleted",
"time": "330.584",
"value": "Kocatepe Mosque_sentence_3"
},
{
"operation": "Highlighting",
"time": "335.349",
"value": "Kocatepe Mosque_sentence_3"
},
{
"operation": "Highlighting",
"time": "341.282",
"value": "Kocatepe Mosque_sentence_4"
},
{
"operation": "Highlighting",
"time": "363.429",
"value": "Kocatepe Mosque_sentence_5"
},
{
"operation": "Highlighting",
"time": "363.902",
"value": "Kocatepe Mosque_sentence_6"
},
{
"operation": "finish",
"time": "374.575",
"value": "finish"
}
]
}
|
40034
|
In April 2014, Joel Nouble (who plays for the Aldershot Town Football Club, a professional association football club based in Aldershot, Hampshire, England) was offered his first professional contract with Dagenham, and was later voted the Dagenham & Redbridge Academy Player of the Year award for the 2012–13 season.
|
{
"evidence": [
"Joel Nouble_sentence_19"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nIn April 2014, Joel Nouble (who plays for the Aldershot Town Football Club, a professional association football club based in Aldershot, Hampshire, England) was offered his first professional contract with Dagenham, and was later voted the Dagenham & Redbridge Academy Player of the Year award for the 2012–13 season.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Joel Nouble_sentence_19"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2944,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.695",
"value": "?search="
},
{
"operation": "search",
"time": "81.442",
"value": "Joel Nouble"
},
{
"operation": "Now on",
"time": "82.456",
"value": "Joel Nouble"
},
{
"operation": "Highlighting",
"time": "86.733",
"value": "Joel Nouble_sentence_0"
},
{
"operation": "hyperlink",
"time": "90.548",
"value": "Aldershot Town"
},
{
"operation": "Now on",
"time": "91.542",
"value": "Aldershot Town F.C."
},
{
"operation": "Highlighting",
"time": "95.617",
"value": "Aldershot Town F.C._sentence_0"
},
{
"operation": "Highlighting",
"time": "95.617",
"value": "Aldershot Town F.C._title"
},
{
"operation": "back-button-clicked",
"time": "100.279",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "100.415",
"value": "Joel Nouble"
},
{
"operation": "Highlighting",
"time": "115.684",
"value": "Joel Nouble_sentence_17"
},
{
"operation": "Highlighting deleted",
"time": "131.939",
"value": "Joel Nouble_sentence_17"
},
{
"operation": "Highlighting",
"time": "132.795",
"value": "Joel Nouble_sentence_19"
},
{
"operation": "finish",
"time": "151.128",
"value": "finish"
}
]
}
|
88359
|
Floyd Clymer became the youngest Ford dealer in the USA at age 19, and began racing motorcycles in the 1910s.
|
{
"evidence": [
"Floyd Clymer_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:\nFloyd Clymer became the youngest Ford dealer in the USA at age 19, and began racing motorcycles in the 1910s.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Floyd Clymer_sentence_2"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2945,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.655",
"value": "?search="
},
{
"operation": "search",
"time": "6.642",
"value": "Floyd Clymer"
},
{
"operation": "Now on",
"time": "7.326",
"value": "Floyd Clymer"
},
{
"operation": "Highlighting",
"time": "15.323",
"value": "Floyd Clymer_sentence_2"
},
{
"operation": "Now on",
"time": "31.136",
"value": "Floyd Clymer"
},
{
"operation": "finish",
"time": "46.812",
"value": "finish"
}
]
}
|
27156
|
During the campaign against Highgate Rabbit Farm by the Animal liberation movement, the activists' efforts never led to any facility closures.
|
{
"evidence": [
"Campaign against Highgate Rabbit Farm_sentence_0",
"Campaign against Highgate Rabbit Farm_sentence_2"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nDuring the campaign against Highgate Rabbit Farm by the Animal liberation movement, the activists' efforts never led to any facility closures.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Campaign against Highgate Rabbit Farm_sentence_0",
"Campaign against Highgate Rabbit Farm_sentence_2"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2946,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.209",
"value": "?search="
},
{
"operation": "search",
"time": "159.09",
"value": "Highgate Rabbit Farm"
},
{
"operation": "Now on",
"time": "159.845",
"value": "Campaign against Highgate Rabbit Farm"
},
{
"operation": "Highlighting",
"time": "167.035",
"value": "Campaign against Highgate Rabbit Farm_sentence_0"
},
{
"operation": "Highlighting",
"time": "172.017",
"value": "Campaign against Highgate Rabbit Farm_sentence_2"
},
{
"operation": "Page search",
"time": "183.854",
"value": "facility close"
},
{
"operation": "Page search",
"time": "239.689",
"value": "close"
},
{
"operation": "finish",
"time": "269.498",
"value": "finish"
}
]
}
|
exfever_4186
|
Lauren Graham born on March 16, 1967 is an American dancer best known for her role as Lorelai Gilmore on an American comedy-drama television series created by Amy Sherman-Palladino.
|
{
"evidence": [
"Lauren Graham",
"Gilmore Girls"
],
"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:\nLauren Graham born on March 16, 1967 is an American dancer best known for her role as Lorelai Gilmore on an American comedy-drama television series created by Amy Sherman-Palladino.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Lauren Graham",
"Gilmore Girls"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2947,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
20246
|
CDC6 homolog is a protein, it localizes in the cell nucleus during cell cycle but plays no role in DNA replication.
|
{
"evidence": [
"CDC6_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:\nCDC6 homolog is a protein, it localizes in the cell nucleus during cell cycle but plays no role in DNA replication.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"CDC6_sentence_2"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2948,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.666",
"value": "?search="
},
{
"operation": "search",
"time": "14.251",
"value": "CDC6 homolog"
},
{
"operation": "Now on",
"time": "15.154",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "29.844",
"value": "Cdc6"
},
{
"operation": "Now on",
"time": "30.677",
"value": "Cdc6"
},
{
"operation": "Page search",
"time": "54.067",
"value": "homolog"
},
{
"operation": "hyperlink",
"time": "65.427",
"value": "Cdc6 homolog in humans"
},
{
"operation": "Now on",
"time": "66.223",
"value": "CDC6"
},
{
"operation": "Highlighting",
"time": "73.372",
"value": "CDC6_sentence_0"
},
{
"operation": "Highlighting",
"time": "91.371",
"value": "CDC6_sentence_2"
},
{
"operation": "Highlighting deleted",
"time": "95.764",
"value": "CDC6_sentence_0"
},
{
"operation": "finish",
"time": "107.819",
"value": "finish"
}
]
}
|
exfever_3389
|
The Boot Monument is a Japanese Revolutionary War memorial located in Saratoga National Historical Park in the state in the Northeastern United States bordered by few states including the southernmost state in the New England region of the United States.
|
{
"evidence": [
"Boot Monument",
"New York (state)",
"Connecticut"
],
"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 Boot Monument is a Japanese Revolutionary War memorial located in Saratoga National Historical Park in the state in the Northeastern United States bordered by few states including the southernmost state in the New England region of the United States.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Boot Monument",
"New York (state)",
"Connecticut"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2949,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_218
|
As the World Turns is a soap opera on NBC that can be watched using a telecommunication medium used for transmitting moving images in black-and-white.
|
{
"evidence": [
"As the World Turns",
"Television"
],
"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:\nAs the World Turns is a soap opera on NBC that can be watched using a telecommunication medium used for transmitting moving images in black-and-white.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"As the World Turns",
"Television"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2950,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_6879
|
The law of value is a central concept in Karl Marx's critique of political economy first expounded in his polemic, a book by Karl Marx published in Paris and a region of Belgium in 1847, comprising 19 municipalities, located in the central portion of the country and is a part of both the French Community of Belgium and the Flemish Community.
|
{
"evidence": [
"Law of value",
"The Poverty of Philosophy",
"Brussels"
],
"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 law of value is a central concept in Karl Marx's critique of political economy first expounded in his polemic, a book by Karl Marx published in Paris and a region of Belgium in 1847, comprising 19 municipalities, located in the central portion of the country and is a part of both the French Community of Belgium and the Flemish Community.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Law of value",
"The Poverty of Philosophy",
"Brussels"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2951,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_5934
|
Selena Marie Gomez appeared in a 2010 American made-for-television comedy-drama fantasy film.
|
{
"evidence": [
"Selena Gomez",
"Wizards of Waverly Place: The Movie"
],
"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:\nSelena Marie Gomez appeared in a 2010 American made-for-television comedy-drama fantasy film.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Selena Gomez",
"Wizards of Waverly Place: The Movie"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2952,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_6422
|
Houston is the most populated city in a state in the South Central region of the United States.
|
{
"evidence": [
"Houston",
"Texas"
],
"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:\nHouston is the most populated city in a state in the South Central region of the United States.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Houston",
"Texas"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2953,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_135
|
Joni Mitchell's first original song is Urge for Going.
|
{
"evidence": [
"Joni Mitchell",
"Nightclub"
],
"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:\nJoni Mitchell's first original song is Urge for Going.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Joni Mitchell",
"Nightclub"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2954,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
58778
|
Erna Baumbauer (8 January 1919 – 30 January 2010) was a casting agent and honorary awardee of the German Film Award from Germany who was called the "grand dame of German casting agents" by The Hollywood Reporter".
|
{
"evidence": [
"Erna Baumbauer_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:\nErna Baumbauer (8 January 1919 – 30 January 2010) was a casting agent and honorary awardee of the German Film Award from Germany who was called the \"grand dame of German casting agents\" by The Hollywood Reporter\".\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Erna Baumbauer_sentence_0"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2955,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.928",
"value": "?search="
},
{
"operation": "search",
"time": "11.996",
"value": "Erna Baumbauer"
},
{
"operation": "Now on",
"time": "13.021",
"value": "Erna Baumbauer"
},
{
"operation": "hyperlink",
"time": "45.977",
"value": "German"
},
{
"operation": "Now on",
"time": "47.906",
"value": "Erna Baumbauer"
},
{
"operation": "Highlighting",
"time": "49.051",
"value": "Erna Baumbauer_sentence_0"
},
{
"operation": "Page search",
"time": "93.765",
"value": "honorary"
},
{
"operation": "search",
"time": "135.071",
"value": "German Film Award"
},
{
"operation": "Now on",
"time": "139.416",
"value": "Deutscher Filmpreis"
},
{
"operation": "Page search",
"time": "181.617",
"value": "baum"
},
{
"operation": "back-button-clicked",
"time": "209.441",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "209.507",
"value": "Erna Baumbauer"
},
{
"operation": "Highlighting",
"time": "211.889",
"value": "Erna Baumbauer_sentence_1"
},
{
"operation": "Highlighting",
"time": "214.495",
"value": "Erna Baumbauer_sentence_2"
},
{
"operation": "Highlighting deleted",
"time": "215.001",
"value": "Erna Baumbauer_sentence_2"
},
{
"operation": "Highlighting deleted",
"time": "215.507",
"value": "Erna Baumbauer_sentence_1"
},
{
"operation": "search",
"time": "234.325",
"value": "honorary awardee of the German Film Award"
},
{
"operation": "Now on",
"time": "235.517",
"value": "undefined"
},
{
"operation": "back-button-clicked",
"time": "249.159",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "249.208",
"value": "Erna Baumbauer"
},
{
"operation": "hyperlink",
"time": "253.472",
"value": "The Hollywood Reporter"
},
{
"operation": "Now on",
"time": "254.467",
"value": "The Hollywood Reporter"
},
{
"operation": "back-button-clicked",
"time": "256.003",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "256.051",
"value": "Erna Baumbauer"
},
{
"operation": "Page search",
"time": "263.534",
"value": "German Film Award"
},
{
"operation": "hyperlink",
"time": "274.197",
"value": "lifetime achievement award"
},
{
"operation": "Now on",
"time": "274.618",
"value": "Lifetime achievement award"
},
{
"operation": "back-button-clicked",
"time": "284.365",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "284.418",
"value": "Erna Baumbauer"
},
{
"operation": "finish",
"time": "346.385",
"value": "finish"
}
]
}
|
3511
|
Dandenong Rangers, which is a professional Australian rules football club based in Melbourne, Victoria, was a team where Alison Downie was part of from 2017 to 2018 where she only appeared in one game.
|
{
"evidence": [
"Dandenong Rangers_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:\nDandenong Rangers, which is a professional Australian rules football club based in Melbourne, Victoria, was a team where Alison Downie was part of from 2017 to 2018 where she only appeared in one game.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Dandenong Rangers_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2956,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "search",
"time": "10.092",
"value": "Dandenong Ranges"
},
{
"operation": "search",
"time": "22.842",
"value": "dandenong rangers"
},
{
"operation": "Highlighting",
"time": "27.931",
"value": "Dandenong Rangers_sentence_0"
},
{
"operation": "Page search",
"time": "46.09",
"value": "alison"
},
{
"operation": "hyperlink",
"time": "58.17",
"value": "Alison Downie"
},
{
"operation": "finish",
"time": "77.008",
"value": "finish"
}
]
}
|
81974
|
Gali Muddu Krishnama Naidu was an Indian politician from Uttar Pradesh who joined politics after a successful career in teaching and was elected five times to the Andhra Pradesh Legislative Assembly.
|
{
"evidence": [
"Gali Muddu Krishnama Naidu_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:\nGali Muddu Krishnama Naidu was an Indian politician from Uttar Pradesh who joined politics after a successful career in teaching and was elected five times to the Andhra Pradesh Legislative Assembly.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Gali Muddu Krishnama Naidu_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2957,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.835",
"value": "?search="
},
{
"operation": "search",
"time": "13.501",
"value": "Gali Muddu Krishnama Naidu"
},
{
"operation": "Now on",
"time": "14.227",
"value": "Gali Muddu Krishnama Naidu"
},
{
"operation": "Highlighting",
"time": "18.801",
"value": "Gali Muddu Krishnama Naidu_sentence_0"
},
{
"operation": "finish",
"time": "27.678",
"value": "finish"
}
]
}
|
39604
|
During the 2015 Sam Houston State Bearkats football team match against the Stephen F. Austin Lumberjacks (the 90th time the teams faced each other), SFA scored only 4 points before half time.
|
{
"evidence": [
"2015 Sam Houston State Bearkats football team_sentence_18"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nDuring the 2015 Sam Houston State Bearkats football team match against the Stephen F. Austin Lumberjacks (the 90th time the teams faced each other), SFA scored only 4 points before half time.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"2015 Sam Houston State Bearkats football team_sentence_18"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2958,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.249",
"value": "?search="
},
{
"operation": "search",
"time": "9.68",
"value": "2015 Sam Houston State Bearkats football team"
},
{
"operation": "Now on",
"time": "12.164",
"value": "2015 Sam Houston State Bearkats football team"
},
{
"operation": "hyperlink",
"time": "40.782",
"value": "2.4 Stephen F. Austin 2015 Sam Houston State Bearkats football team_section_5"
},
{
"operation": "Highlighting",
"time": "44.728",
"value": "2015 Sam Houston State Bearkats football team_sentence_12"
},
{
"operation": "Highlighting",
"time": "92.544",
"value": "2015 Sam Houston State Bearkats football team_sentence_18"
},
{
"operation": "Highlighting",
"time": "106.483",
"value": "2015 Sam Houston State Bearkats football team_sentence_17"
},
{
"operation": "Highlighting deleted",
"time": "114.135",
"value": "2015 Sam Houston State Bearkats football team_sentence_17"
},
{
"operation": "Highlighting deleted",
"time": "123.174",
"value": "2015 Sam Houston State Bearkats football team_sentence_12"
},
{
"operation": "Highlighting",
"time": "147.173",
"value": "2015 Sam Houston State Bearkats football team_sentence_17"
},
{
"operation": "Highlighting deleted",
"time": "163.335",
"value": "2015 Sam Houston State Bearkats football team_sentence_17"
},
{
"operation": "finish",
"time": "171.974",
"value": "finish"
}
]
}
|
66873
|
On Avenue 43, which is one of nine avenues in the Los Angeles Avenues, one can find the Lummis House, a Los Angeles Historic-Cultural Monument.
|
{
"evidence": [
"Los Angeles Avenues_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:\nOn Avenue 43, which is one of nine avenues in the Los Angeles Avenues, one can find the Lummis House, a Los Angeles Historic-Cultural Monument.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Los Angeles Avenues_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2959,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.814",
"value": "?search="
},
{
"operation": "search",
"time": "40.822",
"value": "Avenue 43"
},
{
"operation": "Now on",
"time": "42.991",
"value": "undefined"
},
{
"operation": "search",
"time": "67.158",
"value": "Lummis House"
},
{
"operation": "Now on",
"time": "68.114",
"value": "Lummis House"
},
{
"operation": "Highlighting",
"time": "82.014",
"value": "Lummis House_sentence_2"
},
{
"operation": "Highlighting",
"time": "82.015",
"value": "Lummis House_title"
},
{
"operation": "Highlighting",
"time": "92.518",
"value": "Lummis House_cell_0_0_1"
},
{
"operation": "Highlighting",
"time": "92.522",
"value": "Lummis House_header_cell_0_0_0"
},
{
"operation": "Page search",
"time": "137.638",
"value": "avenue"
},
{
"operation": "search",
"time": "165.23",
"value": "Los Angeles Avenues"
},
{
"operation": "Now on",
"time": "166.146",
"value": "Los Angeles Avenues"
},
{
"operation": "Highlighting",
"time": "168.949",
"value": "Los Angeles Avenues_sentence_0"
},
{
"operation": "finish",
"time": "189.205",
"value": "finish"
}
]
}
|
exfever_6384
|
Cape Fear is a 1991 American psychological thriller film as a remake of the 1962 American noir psychological thriller film starring Gregory Peck, Robert Mitchum, and Polly Bergen.
|
{
"evidence": [
"Cape Fear (1991 film)",
"Cape Fear (1962 film)"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nCape Fear is a 1991 American psychological thriller film as a remake of the 1962 American noir psychological thriller film starring Gregory Peck, Robert Mitchum, and Polly Bergen.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Cape Fear (1991 film)",
"Cape Fear (1962 film)"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2960,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_8336
|
Charlize Theron is a black American actress who rose to international prominence in the 1970s as the leading lady in Taylor Hackford's 1997 American supernatural horror film.
|
{
"evidence": [
"Charlize Theron",
"The Devil's Advocate (1997 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:\nCharlize Theron is a black American actress who rose to international prominence in the 1970s as the leading lady in Taylor Hackford's 1997 American supernatural horror film.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Charlize Theron",
"The Devil's Advocate (1997 film)"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2961,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_3545
|
Practical Magic is an American fantasy romantic comedy film starring an American actor who starred in an American comedy film written and directed by Barry Levinson.
|
{
"evidence": [
"Practical Magic",
"Aidan Quinn",
"Avalon (1990 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:\nPractical Magic is an American fantasy romantic comedy film starring an American actor who starred in an American comedy film written and directed by Barry Levinson.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Practical Magic",
"Aidan Quinn",
"Avalon (1990 film)"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2962,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
33832
|
Sara Ryan is a Canadian author who won the Leslie Bradshaw Award for Young Adult Literature, which began being awarded in 2002.
|
{
"evidence": [
"Sara Ryan_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:\nSara Ryan is a Canadian author who won the Leslie Bradshaw Award for Young Adult Literature, which began being awarded in 2002.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"Sara Ryan_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2963,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.887",
"value": "?search="
},
{
"operation": "search",
"time": "14.207",
"value": "Sara Ryan"
},
{
"operation": "Now on",
"time": "15.414",
"value": "Sara Ryan"
},
{
"operation": "Highlighting",
"time": "18.03",
"value": "Sara Ryan_sentence_0"
},
{
"operation": "finish",
"time": "31.892",
"value": "finish"
}
]
}
|
exfever_6884
|
Broadcast News is a 1987 American romantic comedy-drama film starring an American actress and former fashion model known for her role as Holly Goodhead in the James Bond film Moonraker.
|
{
"evidence": [
"Broadcast News (film)",
"Lois Chiles"
],
"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:\nBroadcast News is a 1987 American romantic comedy-drama film starring an American actress and former fashion model known for her role as Holly Goodhead in the James Bond film Moonraker.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Broadcast News (film)",
"Lois Chiles"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2964,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
89800
|
CAMK2B has a beta chain product that belongs to the serine/threonine protein kinase family—an esterase enzyme in the family of transferases.
|
{
"evidence": [
"Esterase_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:\nCAMK2B has a beta chain product that belongs to the serine/threonine protein kinase family—an esterase enzyme in the family of transferases.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Esterase_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2965,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.559",
"value": "?search="
},
{
"operation": "search",
"time": "11.32",
"value": "CAMK2B"
},
{
"operation": "Now on",
"time": "12.042",
"value": "CAMK2B"
},
{
"operation": "Highlighting",
"time": "27.951",
"value": "CAMK2B_sentence_4"
},
{
"operation": "Highlighting",
"time": "27.952",
"value": "CAMK2B_title"
},
{
"operation": "Highlighting",
"time": "27.952",
"value": "CAMK2B_section_0"
},
{
"operation": "Highlighting",
"time": "41.454",
"value": "CAMK2B_sentence_1"
},
{
"operation": "search",
"time": "109.165",
"value": "esterase"
},
{
"operation": "Now on",
"time": "110.192",
"value": "Esterase"
},
{
"operation": "hyperlink",
"time": "125.963",
"value": "Transferase"
},
{
"operation": "Now on",
"time": "126.879",
"value": "Transferase"
},
{
"operation": "back-button-clicked",
"time": "137.268",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "137.566",
"value": "Esterase"
},
{
"operation": "Highlighting",
"time": "151.267",
"value": "Esterase_sentence_0"
},
{
"operation": "back-button-clicked",
"time": "166.035",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "166.045",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "166.312",
"value": "CAMK2B"
},
{
"operation": "back-button-clicked",
"time": "186.093",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "186.1",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "186.435",
"value": "?search="
},
{
"operation": "search",
"time": "196.831",
"value": "esterase"
},
{
"operation": "Now on",
"time": "197.866",
"value": "Esterase"
},
{
"operation": "hyperlink",
"time": "215.864",
"value": "Transferase"
},
{
"operation": "Now on",
"time": "216.487",
"value": "Transferase"
},
{
"operation": "finish",
"time": "250.634",
"value": "finish"
}
]
}
|
38635
|
George Seligman (born Abril 30, 1927 in New York)is an American Lie algebras mathematician and Professor Emeritus of Mathematics who served as chairman of the mathematics department of Yale University and a Fulbright Lecturer at the University of Münster.
|
{
"evidence": [
"George Seligman_sentence_0",
"George Seligman_sentence_4",
"George Seligman_sentence_3"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nGeorge Seligman (born Abril 30, 1927 in New York)is an American Lie algebras mathematician and Professor Emeritus of Mathematics who served as chairman of the mathematics department of Yale University and a Fulbright Lecturer at the University of Münster.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"George Seligman_sentence_0",
"George Seligman_sentence_4",
"George Seligman_sentence_3"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2966,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.523",
"value": "?search="
},
{
"operation": "search",
"time": "4.012",
"value": "George Seligman"
},
{
"operation": "Now on",
"time": "4.831",
"value": "George Seligman"
},
{
"operation": "Highlighting",
"time": "68.336",
"value": "George Seligman_sentence_0"
},
{
"operation": "Highlighting",
"time": "98.146",
"value": "George Seligman_sentence_4"
},
{
"operation": "Highlighting",
"time": "98.754",
"value": "George Seligman_sentence_3"
},
{
"operation": "finish",
"time": "384.935",
"value": "finish"
}
]
}
|
exfever_1317
|
Darius Rucker was friends with the father of Taylor Swift when they were in college.
|
{
"evidence": [
"Darius Rucker",
"Capitol Records Nashville"
],
"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:\nDarius Rucker was friends with the father of Taylor Swift when they were in college.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Darius Rucker",
"Capitol Records Nashville"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2967,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_5313
|
U2 is an Irish rock band from the largest city in Ireland.
|
{
"evidence": [
"U2",
"Dublin"
],
"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:\nU2 is an Irish rock band from the largest city in Ireland.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"U2",
"Dublin"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2968,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7559
|
Eric Overmyer,?best known for his work on The Wire, is also known for co-writing The Corner: A Year in the Life of an Inner-City Neighborhood with a novelist who was also a?former Baltimore police detective for the homicide and narcotics divisions.
|
{
"evidence": [
"David Simon",
"Ed Burns"
],
"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:\nEric Overmyer,?best known for his work on The Wire, is also known for co-writing The Corner: A Year in the Life of an Inner-City Neighborhood with a novelist who was also a?former Baltimore police detective for the homicide and narcotics divisions.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"David Simon",
"Ed Burns"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2969,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
2928
|
Ted Cowan (Edward George Cowan), is a French physician and worked as a laboratory assistant prior to World War I.
|
{
"evidence": [
"Ted Cowan_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:\nTed Cowan (Edward George Cowan), is a French physician and worked as a laboratory assistant prior to World War I.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"Ted Cowan_sentence_1"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2970,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "search",
"time": "5.846",
"value": "ted cowan"
},
{
"operation": "Highlighting",
"time": "9.865",
"value": "Ted Cowan_sentence_1"
},
{
"operation": "finish",
"time": "40.069",
"value": "finish"
}
]
}
|
exfever_1808
|
Amy Sherman-Palladino is the creator of an American comedy-drama television series.
|
{
"evidence": [
"Amy Sherman-Palladino",
"Gilmore Girls"
],
"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:\nAmy Sherman-Palladino is the creator of an American comedy-drama television series.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Amy Sherman-Palladino",
"Gilmore Girls"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2971,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
95194
|
The proteasome (protein complexes which degrade unneeded or damaged proteins by proteolysis, a chemical reaction that breaks peptide bonds) is a multicatalytic proteinase complex with a highly ordered 20S core structure is composed of four axially stacked rings of 28 non-identical subunits: the two end rings are each formed by seven alpha subunits, and the two central rings are each formed by seven beta subunits.
|
{
"evidence": [
"Proteasome_sentence_0",
"PSMA2_sentence_7",
"PSMA2_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 proteasome (protein complexes which degrade unneeded or damaged proteins by proteolysis, a chemical reaction that breaks peptide bonds) is a multicatalytic proteinase complex with a highly ordered 20S core structure is composed of four axially stacked rings of 28 non-identical subunits: the two end rings are each formed by seven alpha subunits, and the two central rings are each formed by seven beta subunits.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Proteasome_sentence_0",
"PSMA2_sentence_7",
"PSMA2_sentence_8"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2972,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1",
"value": "?search="
},
{
"operation": "search",
"time": "41.295",
"value": "Proteasome"
},
{
"operation": "Now on",
"time": "44.637",
"value": "Proteasome"
},
{
"operation": "Highlighting",
"time": "46.55",
"value": "Proteasome_sentence_0"
},
{
"operation": "Page search",
"time": "76.109",
"value": "mutlicatalytic"
},
{
"operation": "search",
"time": "159.007",
"value": "PSMA2"
},
{
"operation": "Now on",
"time": "159.693",
"value": "PSMA2"
},
{
"operation": "Highlighting",
"time": "166.55",
"value": "PSMA2_sentence_7"
},
{
"operation": "Highlighting",
"time": "180.197",
"value": "PSMA2_sentence_8"
},
{
"operation": "finish",
"time": "190.653",
"value": "finish"
}
]
}
|
36724
|
The Northwest Film Center offers college credit in association with Portland State University, Marylhurst University, and the Pacific Northwest College of Art, whereas the core program contains more than ten classes covering fundamentals of video, film, and other aspects, while most courses are one day a week and have sections either in the afternoon or at night.
|
{
"evidence": [
"Northwest Film Center_sentence_9",
"Northwest Film Center_sentence_10",
"Northwest Film Center_sentence_13"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Northwest Film Center offers college credit in association with Portland State University, Marylhurst University, and the Pacific Northwest College of Art, whereas the core program contains more than ten classes covering fundamentals of video, film, and other aspects, while most courses are one day a week and have sections either in the afternoon or at night.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Northwest Film Center_sentence_9",
"Northwest Film Center_sentence_10",
"Northwest Film Center_sentence_13"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2973,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.618",
"value": "?search="
},
{
"operation": "search",
"time": "8.21",
"value": "Northwest Film Center"
},
{
"operation": "Now on",
"time": "8.799",
"value": "Northwest Film Center"
},
{
"operation": "Page search",
"time": "17.192",
"value": "Portland State"
},
{
"operation": "Highlighting",
"time": "26.001",
"value": "Northwest Film Center_sentence_9"
},
{
"operation": "Highlighting",
"time": "33.86",
"value": "Northwest Film Center_sentence_10"
},
{
"operation": "Highlighting",
"time": "55.289",
"value": "Northwest Film Center_sentence_13"
},
{
"operation": "Page search",
"time": "114.523",
"value": "course"
},
{
"operation": "finish",
"time": "138.413",
"value": "finish"
}
]
}
|
62160
|
Cercamon composed a Plahn for William X of Aquitaine in 1137 and Lanfanc Cigala composed a Plahn for Raimbaut d'Aurenga in 1173.
|
{
"evidence": [
"Cercamon_sentence_0",
"Cercamon_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:\nCercamon composed a Plahn for William X of Aquitaine in 1137 and Lanfanc Cigala composed a Plahn for Raimbaut d'Aurenga in 1173.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Cercamon_sentence_0",
"Cercamon_sentence_1"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2974,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.052",
"value": "?search="
},
{
"operation": "search",
"time": "32.724",
"value": "Cercamon"
},
{
"operation": "Now on",
"time": "33.403",
"value": "Cercamon"
},
{
"operation": "Page search",
"time": "61.832",
"value": "1137"
},
{
"operation": "Highlighting",
"time": "69.333",
"value": "Cercamon_sentence_0"
},
{
"operation": "Highlighting",
"time": "70.134",
"value": "Cercamon_sentence_1"
},
{
"operation": "finish",
"time": "80.115",
"value": "finish"
}
]
}
|
exfever_2838
|
Dollhouse ends with Echo retaining her selfhood even after everyone's personalities are restored.
|
{
"evidence": [
"Eliza Dushku",
"Tru Calling"
],
"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:\nDollhouse ends with Echo retaining her selfhood even after everyone's personalities are restored.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Eliza Dushku",
"Tru Calling"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2975,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_4188
|
Fairfield Grammar School was replaced an Academy secondary school in a suburb of the city of Bristol, in souththwest England.
|
{
"evidence": [
"Fairfield Grammar School",
"Fairfield High School, Bristol",
"Horfield"
],
"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:\nFairfield Grammar School was replaced an Academy secondary school in a suburb of the city of Bristol, in souththwest England.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Fairfield Grammar School",
"Fairfield High School, Bristol",
"Horfield"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2976,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
56250
|
Bob Weiskopf was a television writer and producer who was born on March 13, 1914 in Chicago, Illinois, the most populous city in the world.
|
{
"evidence": [
"Chicago_sentence_2",
"Chicago_sentence_3"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nBob Weiskopf was a television writer and producer who was born on March 13, 1914 in Chicago, Illinois, the most populous city in the world.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Chicago_sentence_2",
"Chicago_sentence_3"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2977,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.727",
"value": "?search="
},
{
"operation": "search",
"time": "29.027",
"value": "Bob Weiskopf"
},
{
"operation": "Now on",
"time": "30.414",
"value": "Bob Weiskopf"
},
{
"operation": "Highlighting",
"time": "34.743",
"value": "Bob Weiskopf_sentence_0"
},
{
"operation": "Highlighting",
"time": "34.745",
"value": "Bob Weiskopf_title"
},
{
"operation": "Highlighting",
"time": "44.693",
"value": "Bob Weiskopf_sentence_3"
},
{
"operation": "Highlighting",
"time": "44.695",
"value": "Bob Weiskopf_section_0"
},
{
"operation": "hyperlink",
"time": "49.318",
"value": "Chicago, Illinois"
},
{
"operation": "Now on",
"time": "61.182",
"value": "Chicago"
},
{
"operation": "Highlighting",
"time": "61.58",
"value": "Chicago_sentence_2"
},
{
"operation": "Highlighting",
"time": "66.649",
"value": "Chicago_sentence_3"
},
{
"operation": "finish",
"time": "81.967",
"value": "finish"
}
]
}
|
exfever_8813
|
The domestic dog is a domesticated descendant of the wolf and the final basic unit of classification and a taxonomic rank of an organism. An organism is an organic, living system that functions as an individual entity.
|
{
"evidence": [
"Dog",
"Species",
"Organism"
],
"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 domestic dog is a domesticated descendant of the wolf and the final basic unit of classification and a taxonomic rank of an organism. An organism is an organic, living system that functions as an individual entity.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Dog",
"Species",
"Organism"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2978,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_3499
|
Assassin's Creed is a game mechanic of using a computer-simulated environment.
|
{
"evidence": [
"Open world",
"Virtual world"
],
"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:\nAssassin's Creed is a game mechanic of using a computer-simulated environment.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Open world",
"Virtual world"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2979,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_5408
|
Backstreet Boys are an American vocal group that includes an American singer, songwriter, actor, and model.
|
{
"evidence": [
"Backstreet Boys",
"Kevin Richardson (musician)"
],
"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:\nBackstreet Boys are an American vocal group that includes an American singer, songwriter, actor, and model.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Backstreet Boys",
"Kevin Richardson (musician)"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2980,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_2543
|
Dan Aykroyd was an original member of the "Not Ready for Prime Time Players" on Chappelle's Show, where he performed with his friend, an American comedian, actor, and musician.
|
{
"evidence": [
"Dan Aykroyd",
"John Belushi"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nDan Aykroyd was an original member of the \"Not Ready for Prime Time Players\" on Chappelle's Show, where he performed with his friend, an American comedian, actor, and musician.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Dan Aykroyd",
"John Belushi"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 2981,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
42856
|
Meghan Coffey, born in New Berlin, Wisconsin, U.S., graduated from Marquette University with a Bachelor of Science degree, before joining Miss America 2007.
|
{
"evidence": [
"Meghan Coffey_sentence_10",
"Meghan Coffey_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:\nMeghan Coffey, born in New Berlin, Wisconsin, U.S., graduated from Marquette University with a Bachelor of Science degree, before joining Miss America 2007.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Meghan Coffey_sentence_10",
"Meghan Coffey_sentence_1"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 2982,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.799",
"value": "?search="
},
{
"operation": "search",
"time": "6.127",
"value": "Megan Coffey"
},
{
"operation": "Now on",
"time": "7.565",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "9.603",
"value": "meghan coffey"
},
{
"operation": "Now on",
"time": "10.55",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "13.754",
"value": "Meghan Coffey"
},
{
"operation": "Now on",
"time": "14.571",
"value": "Meghan Coffey"
},
{
"operation": "Highlighting",
"time": "18.84",
"value": "Meghan Coffey_cell_0_1_1"
},
{
"operation": "Highlighting",
"time": "18.842",
"value": "Meghan Coffey_header_cell_0_1_0"
},
{
"operation": "Highlighting",
"time": "18.842",
"value": "Meghan Coffey_header_cell_0_0_0"
},
{
"operation": "Highlighting",
"time": "27.648",
"value": "Meghan Coffey_cell_0_2_1"
},
{
"operation": "Highlighting",
"time": "27.65",
"value": "Meghan Coffey_header_cell_0_2_0"
},
{
"operation": "Highlighting deleted",
"time": "30.056",
"value": "Meghan Coffey_cell_0_2_1"
},
{
"operation": "Highlighting deleted",
"time": "30.059",
"value": "Meghan Coffey_header_cell_0_2_0"
},
{
"operation": "Highlighting",
"time": "32.148",
"value": "Meghan Coffey_sentence_1"
},
{
"operation": "Highlighting",
"time": "45.851",
"value": "Meghan Coffey_sentence_0"
},
{
"operation": "Highlighting deleted",
"time": "53.231",
"value": "Meghan Coffey_cell_0_1_1"
},
{
"operation": "Highlighting deleted",
"time": "53.234",
"value": "Meghan Coffey_header_cell_0_1_0"
},
{
"operation": "Highlighting deleted",
"time": "53.235",
"value": "Meghan Coffey_header_cell_0_0_0"
},
{
"operation": "hyperlink",
"time": "88.669",
"value": "Miss America"
},
{
"operation": "Now on",
"time": "89.622",
"value": "Miss America"
},
{
"operation": "back-button-clicked",
"time": "92.315",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "92.365",
"value": "Meghan Coffey"
},
{
"operation": "hyperlink",
"time": "122.475",
"value": "Miss Wisconsin USA"
},
{
"operation": "Now on",
"time": "123.558",
"value": "Miss Wisconsin USA"
},
{
"operation": "Highlighting",
"time": "182.416",
"value": "Miss Wisconsin USA_cell_1_16_1"
},
{
"operation": "Highlighting",
"time": "182.417",
"value": "Miss Wisconsin USA_title"
},
{
"operation": "Highlighting",
"time": "182.418",
"value": "Miss Wisconsin USA_section_4"
},
{
"operation": "Highlighting",
"time": "182.42",
"value": "Miss Wisconsin USA_header_cell_1_0_1"
},
{
"operation": "Highlighting deleted",
"time": "183.564",
"value": "Miss Wisconsin USA_cell_1_16_1"
},
{
"operation": "Highlighting deleted",
"time": "183.566",
"value": "Miss Wisconsin USA_title"
},
{
"operation": "Highlighting deleted",
"time": "183.567",
"value": "Miss Wisconsin USA_section_4"
},
{
"operation": "Highlighting deleted",
"time": "183.571",
"value": "Miss Wisconsin USA_header_cell_1_0_1"
},
{
"operation": "Highlighting",
"time": "193.193",
"value": "Miss Wisconsin USA_cell_1_16_0"
},
{
"operation": "Highlighting",
"time": "193.195",
"value": "Miss Wisconsin USA_title"
},
{
"operation": "Highlighting",
"time": "193.195",
"value": "Miss Wisconsin USA_section_4"
},
{
"operation": "Highlighting",
"time": "193.197",
"value": "Miss Wisconsin USA_header_cell_1_0_0"
},
{
"operation": "Highlighting",
"time": "199.606",
"value": "Miss Wisconsin USA_cell_1_16_1"
},
{
"operation": "Highlighting",
"time": "199.611",
"value": "Miss Wisconsin USA_header_cell_1_0_1"
},
{
"operation": "Highlighting deleted",
"time": "202.126",
"value": "Miss Wisconsin USA_cell_1_16_1"
},
{
"operation": "Highlighting deleted",
"time": "202.13",
"value": "Miss Wisconsin USA_header_cell_1_0_1"
},
{
"operation": "Highlighting deleted",
"time": "203.352",
"value": "Miss Wisconsin USA_cell_1_16_0"
},
{
"operation": "Highlighting deleted",
"time": "203.353",
"value": "Miss Wisconsin USA_title"
},
{
"operation": "Highlighting deleted",
"time": "203.355",
"value": "Miss Wisconsin USA_section_4"
},
{
"operation": "Highlighting deleted",
"time": "203.357",
"value": "Miss Wisconsin USA_header_cell_1_0_0"
},
{
"operation": "hyperlink",
"time": "205.94",
"value": "Miss America"
},
{
"operation": "Now on",
"time": "206.711",
"value": "Miss America"
},
{
"operation": "hyperlink",
"time": "217.212",
"value": "List of Miss America titleholders"
},
{
"operation": "Now on",
"time": "217.918",
"value": "List of Miss America titleholders"
},
{
"operation": "hyperlink",
"time": "228.012",
"value": "2 Winners List of Miss America titleholders_section_1"
},
{
"operation": "Highlighting",
"time": "245.563",
"value": "List of Miss America titleholders_cell_0_17_0"
},
{
"operation": "Highlighting",
"time": "245.564",
"value": "List of Miss America titleholders_title"
},
{
"operation": "Highlighting",
"time": "245.564",
"value": "List of Miss America titleholders_section_2"
},
{
"operation": "Highlighting",
"time": "245.566",
"value": "List of Miss America titleholders_header_cell_0_0_0"
},
{
"operation": "Highlighting",
"time": "249.804",
"value": "List of Miss America titleholders_cell_0_17_1"
},
{
"operation": "Highlighting",
"time": "249.806",
"value": "List of Miss America titleholders_header_cell_0_0_2"
},
{
"operation": "back-button-clicked",
"time": "258.805",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "258.912",
"value": "Miss America"
},
{
"operation": "back-button-clicked",
"time": "263.102",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "263.106",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "263.203",
"value": "Miss Wisconsin USA"
},
{
"operation": "back-button-clicked",
"time": "265.791",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "265.795",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "265.843",
"value": "Meghan Coffey"
},
{
"operation": "Highlighting",
"time": "285.264",
"value": "Meghan Coffey_sentence_10"
},
{
"operation": "Highlighting deleted",
"time": "303.287",
"value": "Meghan Coffey_sentence_0"
},
{
"operation": "search",
"time": "324.584",
"value": "2007 Miss America"
},
{
"operation": "Now on",
"time": "326.85",
"value": "undefined"
},
{
"operation": "back-button-clicked",
"time": "382.893",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "382.942",
"value": "Meghan Coffey"
},
{
"operation": "Highlighting deleted",
"time": "386.763",
"value": "Meghan Coffey_sentence_1"
},
{
"operation": "Highlighting",
"time": "395.155",
"value": "Meghan Coffey_sentence_1"
},
{
"operation": "finish",
"time": "430.683",
"value": "finish"
}
]
}
|
exfever_3518
|
Elizabeth Barrett was an unbelievable poet.
|
{
"evidence": [
"Robert Browning",
"Vocabulary"
],
"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:\nElizabeth Barrett was an unbelievable poet.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Robert Browning",
"Vocabulary"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 2983,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.