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_336
|
Troy Baker is known for his video game role as Booker DeWitt in a first-person shooter video game developed by Irrational Games and published by 3 Games.
|
{
"evidence": [
"Troy Baker",
"BioShock Infinite"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nTroy Baker is known for his video game role as Booker DeWitt in a first-person shooter video game developed by Irrational Games and published by 3 Games.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Troy Baker",
"BioShock Infinite"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3584,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
71712
|
VinylPlus succeeded Vinyl 2010 and it was set up in 2004 around the principles of Responsible Care.
|
{
"evidence": [
"VinylPlus_sentence_0",
"VinylPlus_sentence_8"
],
"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:\nVinylPlus succeeded Vinyl 2010 and it was set up in 2004 around the principles of Responsible Care.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"VinylPlus_sentence_0",
"VinylPlus_sentence_8"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3585,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.751",
"value": "?search="
},
{
"operation": "search",
"time": "15.502",
"value": "VinylPlus"
},
{
"operation": "Now on",
"time": "16.335",
"value": "VinylPlus"
},
{
"operation": "Highlighting",
"time": "30.798",
"value": "VinylPlus_sentence_0"
},
{
"operation": "Highlighting",
"time": "56.936",
"value": "VinylPlus_sentence_8"
},
{
"operation": "finish",
"time": "96.952",
"value": "finish"
}
]
}
|
exfever_3701
|
Perth is the capital and largest city of the Australian state between the Pacific Ocean and a low escarpment that runs from north-south to the east of the Swan Coastal Plain and Perth, Western Australia.
|
{
"evidence": [
"Perth",
"Darling Scarp"
],
"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:\nPerth is the capital and largest city of the Australian state between the Pacific Ocean and a low escarpment that runs from north-south to the east of the Swan Coastal Plain and Perth, Western Australia.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Perth",
"Darling Scarp"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3586,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_345
|
The Lion was directed by an Australian television, advertising, and film director.
|
{
"evidence": [
"Lion (2016 film)",
"Garth Davis"
],
"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 Lion was directed by an Australian television, advertising, and film director.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Lion (2016 film)",
"Garth Davis"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3587,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_8803
|
Daniel Radcliffe also won at Scream Awards.
|
{
"evidence": [
"Daniel Radcliffe",
"The Woman in Black (2012 film)"
],
"label": "NOT ENOUGH INFO"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nDaniel Radcliffe also won at Scream Awards.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Daniel Radcliffe",
"The Woman in Black (2012 film)"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3588,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
75663
|
Dodecaborate ion salts do not react with hot aqueous sodium hydroxide, also known as lye and caustic soda.
|
{
"evidence": [
"Dodecaborate_sentence_8",
"Sodium hydroxide_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:\nDodecaborate ion salts do not react with hot aqueous sodium hydroxide, also known as lye and caustic soda.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Dodecaborate_sentence_8",
"Sodium hydroxide_sentence_0"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3589,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.69",
"value": "?search="
},
{
"operation": "search",
"time": "7.509",
"value": "Dodecaborate"
},
{
"operation": "Now on",
"time": "8.046",
"value": "Dodecaborate"
},
{
"operation": "Highlighting",
"time": "56.791",
"value": "Dodecaborate_sentence_8"
},
{
"operation": "search",
"time": "125.768",
"value": "Hydrochloric acid"
},
{
"operation": "Now on",
"time": "126.857",
"value": "Hydrochloric acid"
},
{
"operation": "search",
"time": "135.247",
"value": "lye"
},
{
"operation": "Now on",
"time": "135.893",
"value": "Lye"
},
{
"operation": "Highlighting",
"time": "139.675",
"value": "Lye_sentence_2"
},
{
"operation": "Highlighting",
"time": "139.676",
"value": "Lye_title"
},
{
"operation": "search",
"time": "147.687",
"value": "Caustic soda"
},
{
"operation": "Now on",
"time": "148.912",
"value": "Sodium hydroxide"
},
{
"operation": "Highlighting",
"time": "152.394",
"value": "Sodium hydroxide_sentence_0"
},
{
"operation": "finish",
"time": "203.177",
"value": "finish"
}
]
}
|
25969
|
Sundarbans East Wildlife Sanctuary altitude varies between 0 and 3 meters above sea level and the soil is much less fertile than that of Sundarbans West Wildlife Sanctuary.
|
{
"evidence": [
"Sundarbans East Wildlife Sanctuary_sentence_10"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nSundarbans East Wildlife Sanctuary altitude varies between 0 and 3 meters above sea level and the soil is much less fertile than that of Sundarbans West Wildlife Sanctuary.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Sundarbans East Wildlife Sanctuary_sentence_10"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3590,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.725",
"value": "?search="
},
{
"operation": "search",
"time": "10.898",
"value": "Sundarbans East Wildlife Sanctuary"
},
{
"operation": "Now on",
"time": "11.812",
"value": "Sundarbans East Wildlife Sanctuary"
},
{
"operation": "Highlighting",
"time": "21.796",
"value": "Sundarbans East Wildlife Sanctuary_sentence_9"
},
{
"operation": "Highlighting",
"time": "35.722",
"value": "Sundarbans East Wildlife Sanctuary_sentence_10"
},
{
"operation": "Highlighting deleted",
"time": "36.93",
"value": "Sundarbans East Wildlife Sanctuary_sentence_9"
},
{
"operation": "finish",
"time": "45.402",
"value": "finish"
}
]
}
|
exfever_2215
|
Hapalochlaena is the taxonomic rank used in the biological classification of living and fossil organisms for blue-ringed octopuses.
|
{
"evidence": [
"Blue-ringed octopus",
"Genus"
],
"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:\nHapalochlaena is the taxonomic rank used in the biological classification of living and fossil organisms for blue-ringed octopuses.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Blue-ringed octopus",
"Genus"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3591,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
72955
|
Miah Dennehy was transferred to Walsall F.C. in July 1975 and in three years with Walsall he made a further 128 league appearances and scored 22 goals.
|
{
"evidence": [
"Miah Dennehy_sentence_14",
"Miah Dennehy_sentence_15"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nMiah Dennehy was transferred to Walsall F.C. in July 1975 and in three years with Walsall he made a further 128 league appearances and scored 22 goals.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Miah Dennehy_sentence_14",
"Miah Dennehy_sentence_15"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3592,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.29",
"value": "?search="
},
{
"operation": "search",
"time": "10.536",
"value": "Miah Dennehy"
},
{
"operation": "Now on",
"time": "12.721",
"value": "Miah Dennehy"
},
{
"operation": "Highlighting",
"time": "97.089",
"value": "Miah Dennehy_sentence_14"
},
{
"operation": "Highlighting",
"time": "101.649",
"value": "Miah Dennehy_sentence_15"
},
{
"operation": "finish",
"time": "112.129",
"value": "finish"
}
]
}
|
91363
|
Karl Toft(30 June 1936-28 April 2018) was a Canadian convicted sex predator and pederast regarded as one of Canada's worst sex offenders, who was fired in 1968 and imprisoned the sexual and physical abuse of his foster child.
|
{
"evidence": [
"Karl Toft_sentence_0",
"Karl Toft_sentence_1",
"Karl Toft_sentence_6",
"Karl Toft_sentence_8",
"Karl Toft_sentence_9"
],
"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:\nKarl Toft(30 June 1936-28 April 2018) was a Canadian convicted sex predator and pederast regarded as one of Canada's worst sex offenders, who was fired in 1968 and imprisoned the sexual and physical abuse of his foster child.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Karl Toft_sentence_0",
"Karl Toft_sentence_1",
"Karl Toft_sentence_6",
"Karl Toft_sentence_8",
"Karl Toft_sentence_9"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3593,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.959",
"value": "?search="
},
{
"operation": "search",
"time": "13.578",
"value": "Karl Toft"
},
{
"operation": "Now on",
"time": "14.305",
"value": "Karl Toft"
},
{
"operation": "Highlighting",
"time": "32.132",
"value": "Karl Toft_sentence_0"
},
{
"operation": "Highlighting",
"time": "34.125",
"value": "Karl Toft_sentence_1"
},
{
"operation": "Highlighting",
"time": "48.015",
"value": "Karl Toft_sentence_6"
},
{
"operation": "Highlighting",
"time": "155.942",
"value": "Karl Toft_sentence_8"
},
{
"operation": "Highlighting",
"time": "170.731",
"value": "Karl Toft_sentence_9"
},
{
"operation": "finish",
"time": "273.756",
"value": "finish"
}
]
}
|
40480
|
Loubet, born on 30 December 1838, admitted to the Parisian bar in 1862; took his doctorate in law the next year; then settled down to exercise his profession in Montélimar.
|
{
"evidence": [
"Émile Loubet_sentence_0",
"Émile Loubet_sentence_7",
"Émile Loubet_sentence_9"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nLoubet, born on 30 December 1838, admitted to the Parisian bar in 1862; took his doctorate in law the next year; then settled down to exercise his profession in Montélimar.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"Émile Loubet_sentence_0",
"Émile Loubet_sentence_7",
"Émile Loubet_sentence_9"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3594,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.838",
"value": "?search="
},
{
"operation": "search",
"time": "14.916",
"value": "Loubet, Emile Francois"
},
{
"operation": "Now on",
"time": "16.42",
"value": "Émile Loubet"
},
{
"operation": "Highlighting",
"time": "39.298",
"value": "Émile Loubet_sentence_0"
},
{
"operation": "Highlighting",
"time": "52.115",
"value": "Émile Loubet_sentence_7"
},
{
"operation": "Highlighting",
"time": "58.585",
"value": "Émile Loubet_sentence_9"
},
{
"operation": "finish",
"time": "105.676",
"value": "finish"
}
]
}
|
31723
|
Manuel Goded Llopis had a part in the 1919 Rif War, an armed conflict fought between Spain and Berber tribes in Morocco.
|
{
"evidence": [
"Rif War_sentence_2"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nManuel Goded Llopis had a part in the 1919 Rif War, an armed conflict fought between Spain and Berber tribes in Morocco.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Rif War_sentence_2"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3595,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.991",
"value": "?search="
},
{
"operation": "Now on",
"time": "11.51",
"value": "?search="
},
{
"operation": "search",
"time": "39.069",
"value": "Manuel Goded Llopis"
},
{
"operation": "Now on",
"time": "39.813",
"value": "Manuel Goded Llopis"
},
{
"operation": "Highlighting",
"time": "66.661",
"value": "Manuel Goded Llopis_sentence_3"
},
{
"operation": "Highlighting",
"time": "66.662",
"value": "Manuel Goded Llopis_title"
},
{
"operation": "hyperlink",
"time": "71.699",
"value": "Rif War"
},
{
"operation": "Now on",
"time": "72.763",
"value": "Rif War"
},
{
"operation": "Highlighting",
"time": "85.116",
"value": "Rif War_sentence_2"
},
{
"operation": "back-button-clicked",
"time": "112.292",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "112.381",
"value": "Manuel Goded Llopis"
},
{
"operation": "Highlighting deleted",
"time": "117.956",
"value": "Manuel Goded Llopis_sentence_3"
},
{
"operation": "Highlighting deleted",
"time": "117.957",
"value": "Manuel Goded Llopis_title"
},
{
"operation": "finish",
"time": "131.076",
"value": "finish"
}
]
}
|
exfever_6119
|
Harley Quinn is a comic relief henchwoman for Joker that is a variant of the virtuous stock character that is the antithesis of a superhero.
|
{
"evidence": [
"Harley Quinn",
"Supervillain"
],
"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:\nHarley Quinn is a comic relief henchwoman for Joker that is a variant of the virtuous stock character that is the antithesis of a superhero.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Harley Quinn",
"Supervillain"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3596,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
48637
|
Khalilzad was appointed to serve as US ambassador to Iraq in November 2003; and during that time, he oversaw the drafting of the constitution of Afghanistan was involved with the country's first elections.
|
{
"evidence": [
"Zalmay Khalilzad_sentence_21"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nKhalilzad was appointed to serve as US ambassador to Iraq in November 2003; and during that time, he oversaw the drafting of the constitution of Afghanistan was involved with the country's first elections.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Zalmay Khalilzad_sentence_21"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3597,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.328",
"value": "?search="
},
{
"operation": "search",
"time": "14.886",
"value": "US Ambassador to Iraq"
},
{
"operation": "Now on",
"time": "16.735",
"value": "US Ambassador to Iraq"
},
{
"operation": "Page search",
"time": "27.955",
"value": "Khalilzad"
},
{
"operation": "hyperlink",
"time": "37.251",
"value": "Zalmay Khalilzad"
},
{
"operation": "Now on",
"time": "38.356",
"value": "Zalmay Khalilzad"
},
{
"operation": "Page search",
"time": "56.501",
"value": "ambassador to Iraq"
},
{
"operation": "Highlighting",
"time": "58.771",
"value": "Zalmay Khalilzad_sentence_3"
},
{
"operation": "Page search",
"time": "83.663",
"value": "appointed"
},
{
"operation": "Highlighting",
"time": "85.834",
"value": "Zalmay Khalilzad_sentence_5"
},
{
"operation": "hyperlink",
"time": "118.267",
"value": "United States Ambassador to Iraq"
},
{
"operation": "Now on",
"time": "119.009",
"value": "United States Ambassador to Iraq"
},
{
"operation": "search",
"time": "137.27",
"value": "Contains...appointed to serve as US ambassador to Iraq in November 2003"
},
{
"operation": "Now on",
"time": "143.181",
"value": "undefined"
},
{
"operation": "search",
"time": "166.234",
"value": "Contains...constitution of Afghanistan"
},
{
"operation": "Now on",
"time": "168.031",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "173.934",
"value": "Bonn Agreement (Afghanistan)"
},
{
"operation": "Now on",
"time": "174.64",
"value": "Bonn Agreement (Afghanistan)"
},
{
"operation": "back-button-clicked",
"time": "203.879",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "204.643",
"value": "undefined"
},
{
"operation": "back-button-clicked",
"time": "205.382",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "205.387",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "206.204",
"value": "undefined"
},
{
"operation": "back-button-clicked",
"time": "206.95",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "206.955",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "207.517",
"value": "List of ambassadors of the United States to Iraq"
},
{
"operation": "back-button-clicked",
"time": "208.547",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "208.562",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "208.882",
"value": "Zalmay Khalilzad"
},
{
"operation": "Page search",
"time": "226.634",
"value": "November"
},
{
"operation": "Highlighting",
"time": "310.779",
"value": "Zalmay Khalilzad_sentence_21"
},
{
"operation": "finish",
"time": "358.338",
"value": "finish"
}
]
}
|
exfever_5414
|
The Menace II Society is a 1993 American film directed by the Hughes Brothers, whose genre originated in the United States and features aspects of urban African American culture.
|
{
"evidence": [
"Menace II Society",
"Hood 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:\nThe Menace II Society is a 1993 American film directed by the Hughes Brothers, whose genre originated in the United States and features aspects of urban African American culture.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Menace II Society",
"Hood film"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3598,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
65092
|
Ada Hayden (who had died of cancer at age 65) had taught Botany at Iowa State in 1911 as an instructor, then became an assistant professor of botany in 1920.
|
{
"evidence": [
"Ada Hayden_sentence_19",
"Ada Hayden_sentence_10",
"Ada Hayden_sentence_11"
],
"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:\nAda Hayden (who had died of cancer at age 65) had taught Botany at Iowa State in 1911 as an instructor, then became an assistant professor of botany in 1920.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Ada Hayden_sentence_19",
"Ada Hayden_sentence_10",
"Ada Hayden_sentence_11"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3599,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.07",
"value": "?search="
},
{
"operation": "search",
"time": "11.898",
"value": "Ada Hayden"
},
{
"operation": "Now on",
"time": "12.729",
"value": "Ada Hayden"
},
{
"operation": "Page search",
"time": "30.624",
"value": "cancer"
},
{
"operation": "Highlighting",
"time": "33.385",
"value": "Ada Hayden_sentence_19"
},
{
"operation": "Highlighting",
"time": "42.664",
"value": "Ada Hayden_sentence_10"
},
{
"operation": "Highlighting",
"time": "48.824",
"value": "Ada Hayden_sentence_11"
},
{
"operation": "finish",
"time": "81.737",
"value": "finish"
}
]
}
|
exfever_1622
|
Jeff Goldblum has also released an album, I Shouldn't Be Telling You This.
|
{
"evidence": [
"Jeff Goldblum",
"David Cronenberg",
"Scanners"
],
"label": "NOT ENOUGH INFO"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nJeff Goldblum has also released an album, I Shouldn't Be Telling You This.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Jeff Goldblum",
"David Cronenberg",
"Scanners"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3600,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
63409
|
Personal Ordinariate of Our Lady of Walsingham was considered by the Catholic Bishops' Conference of England and Wales after the resignation of Bishops Burnham and Newton who, together with Broadhurst and their wives, were received into the Roman Catholic Church.
|
{
"evidence": [
"Personal Ordinariate of Our Lady of Walsingham_sentence_24",
"Personal Ordinariate of Our Lady of Walsingham_sentence_23",
"Personal Ordinariate of Our Lady of Walsingham_sentence_25"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nPersonal Ordinariate of Our Lady of Walsingham was considered by the Catholic Bishops' Conference of England and Wales after the resignation of Bishops Burnham and Newton who, together with Broadhurst and their wives, were received into the Roman Catholic Church.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Personal Ordinariate of Our Lady of Walsingham_sentence_24",
"Personal Ordinariate of Our Lady of Walsingham_sentence_23",
"Personal Ordinariate of Our Lady of Walsingham_sentence_25"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3601,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.178",
"value": "?search="
},
{
"operation": "search",
"time": "280.787",
"value": "Personal Ordinariate of Our Lady of Walsingham"
},
{
"operation": "Now on",
"time": "281.99",
"value": "Personal Ordinariate of Our Lady of Walsingham"
},
{
"operation": "Page search",
"time": "890.624",
"value": "considered"
},
{
"operation": "Highlighting",
"time": "912.688",
"value": "Personal Ordinariate of Our Lady of Walsingham_sentence_24"
},
{
"operation": "Highlighting",
"time": "949.878",
"value": "Personal Ordinariate of Our Lady of Walsingham_sentence_23"
},
{
"operation": "Highlighting",
"time": "956.603",
"value": "Personal Ordinariate of Our Lady of Walsingham_sentence_25"
},
{
"operation": "finish",
"time": "968.225",
"value": "finish"
}
]
}
|
4833
|
The 1951 Swiss Grand Prix was a car racing event.
|
{
"evidence": [
"1951 Swiss Grand Prix_sentence_0",
"Formula One_sentence_2"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe 1951 Swiss Grand Prix was a car racing event.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"1951 Swiss Grand Prix_sentence_0",
"Formula One_sentence_2"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3602,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "search",
"time": "11.69",
"value": "1951 Swiss Grand Prix"
},
{
"operation": "Highlighting",
"time": "20.701",
"value": "1951 Swiss Grand Prix_sentence_0"
},
{
"operation": "hyperlink",
"time": "26.958",
"value": "Formula One"
},
{
"operation": "Highlighting",
"time": "36.859",
"value": "Formula One_sentence_2"
},
{
"operation": "finish",
"time": "49.356",
"value": "finish"
}
]
}
|
exfever_8999
|
Cary Elwes is an English actor who performs in a film released by Paramount Pictures.
|
{
"evidence": [
"Cary Elwes",
"Days of Thunder"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nCary Elwes is an English actor who performs in a film released by Paramount Pictures.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Cary Elwes",
"Days of Thunder"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3603,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
36638
|
Samuel Bryan, a resident of Pennsylvania and Anti-Federalist author, is generally ascribed by historians to the Letters of Centinel wherein he commended and endorsed the proposed Constitution of the United States.
|
{
"evidence": [
"Samuel Bryan_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:\nSamuel Bryan, a resident of Pennsylvania and Anti-Federalist author, is generally ascribed by historians to the Letters of Centinel wherein he commended and endorsed the proposed Constitution of the United States.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Samuel Bryan_sentence_2"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3604,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.488",
"value": "?search="
},
{
"operation": "search",
"time": "13.202",
"value": "Samuel Bryan"
},
{
"operation": "Now on",
"time": "14.205",
"value": "Samuel Bryan"
},
{
"operation": "Highlighting",
"time": "34.65",
"value": "Samuel Bryan_sentence_0"
},
{
"operation": "Highlighting",
"time": "49.882",
"value": "Samuel Bryan_sentence_1"
},
{
"operation": "Highlighting",
"time": "55.426",
"value": "Samuel Bryan_sentence_2"
},
{
"operation": "Highlighting deleted",
"time": "57.418",
"value": "Samuel Bryan_sentence_1"
},
{
"operation": "Highlighting deleted",
"time": "58.042",
"value": "Samuel Bryan_sentence_0"
},
{
"operation": "Now on",
"time": "98.863",
"value": "Samuel Bryan"
},
{
"operation": "finish",
"time": "113.883",
"value": "finish"
}
]
}
|
exfever_8869
|
Black Panther is a 2019 American superhero film that stars an American actor and playwright who began his career in theatre, winning a Drama League Directing Fellowship and an acting AUDELCO.
|
{
"evidence": [
"Black Panther (film)",
"Chadwick Boseman"
],
"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:\nBlack Panther is a 2019 American superhero film that stars an American actor and playwright who began his career in theatre, winning a Drama League Directing Fellowship and an acting AUDELCO.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Black Panther (film)",
"Chadwick Boseman"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3605,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_4279
|
Black Sabbath were an American rock band formed by a British guitarist and founder of Heaven & Hell, which disbanded after the death of an American heavy metal singer.
|
{
"evidence": [
"Black Sabbath",
"Tony Iommi",
"Ronnie James Dio"
],
"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:\nBlack Sabbath were an American rock band formed by a British guitarist and founder of Heaven & Hell, which disbanded after the death of an American heavy metal singer.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Black Sabbath",
"Tony Iommi",
"Ronnie James Dio"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3606,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_962
|
Trolls is a computer-animated musical comedy film produced by an animation studio that is a subsidiary of Universal Pictures, a division of NBCUniversal, which is itself a division of a multinational telecommunications conglomerate.
|
{
"evidence": [
"Trolls (film)",
"DreamWorks Animation",
"Comcast"
],
"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:\nTrolls is a computer-animated musical comedy film produced by an animation studio that is a subsidiary of Universal Pictures, a division of NBCUniversal, which is itself a division of a multinational telecommunications conglomerate.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Trolls (film)",
"DreamWorks Animation",
"Comcast"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3607,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7006
|
The Wolfman in 2010 is a remake of a horror film written by a Filipino-American novelist and screenwriter.
|
{
"evidence": [
"The Wolfman (2010 film)",
"The Wolf Man (1941 film)",
"Curt Siodmak"
],
"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 Wolfman in 2010 is a remake of a horror film written by a Filipino-American novelist and screenwriter.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Wolfman (2010 film)",
"The Wolf Man (1941 film)",
"Curt Siodmak"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3608,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
54413
|
Sakiusa Masinivanua (Masi) Matadigo, born on May 8, 1982 in Suva, Fiji, is a rugby union footballer who plays back row for Montpellier Hérault RC, and has been capped for the full Fiji national side.
|
{
"evidence": [
"Sakiusa Matadigo_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:\nSakiusa Masinivanua (Masi) Matadigo, born on May 8, 1982 in Suva, Fiji, is a rugby union footballer who plays back row for Montpellier Hérault RC, and has been capped for the full Fiji national side.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Sakiusa Matadigo_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3609,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.151",
"value": "?search="
},
{
"operation": "search",
"time": "15.382",
"value": "Sakiusa Masinivanua (Masi) Matadigo"
},
{
"operation": "Now on",
"time": "16.125",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "20.446",
"value": "Sakiusa Matadigo"
},
{
"operation": "Now on",
"time": "20.973",
"value": "Sakiusa Matadigo"
},
{
"operation": "Highlighting",
"time": "56.247",
"value": "Sakiusa Matadigo_sentence_0"
},
{
"operation": "finish",
"time": "66.814",
"value": "finish"
}
]
}
|
2558
|
KQRR is an FM radio station broadcasting a Russian Christian format, their first two call letters were KGON, which they used when they first went on the air on July 4, 1947, and then changed to KYMN in February 1, 1965.
|
{
"evidence": [
"KQRR_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:\nKQRR is an FM radio station broadcasting a Russian Christian format, their first two call letters were KGON, which they used when they first went on the air on July 4, 1947, and then changed to KYMN in February 1, 1965.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"KQRR_sentence_1"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3610,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "search",
"time": "8.362",
"value": "KQRR"
},
{
"operation": "Highlighting",
"time": "20.864",
"value": "KQRR_sentence_1"
},
{
"operation": "finish",
"time": "57.088",
"value": "finish"
}
]
}
|
exfever_1359
|
George Edward Foreman is an American former professional boxer who retired after a loss to an American heavyweight boxer who fought twice and outpointed a boxer known for his punching power and crowd-pleasing fighting style.
|
{
"evidence": [
"George Foreman",
"Jimmy Young (boxer)",
"Ron Lyle"
],
"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:\nGeorge Edward Foreman is an American former professional boxer who retired after a loss to an American heavyweight boxer who fought twice and outpointed a boxer known for his punching power and crowd-pleasing fighting style.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"George Foreman",
"Jimmy Young (boxer)",
"Ron Lyle"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3611,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
66734
|
Yakoruda lies in the Rhodope Mountains and was proclaimed a city on 9 September 1964.
|
{
"evidence": [
"Yakoruda_sentence_2",
"Yakoruda_sentence_3"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nYakoruda lies in the Rhodope Mountains and was proclaimed a city on 9 September 1964.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Yakoruda_sentence_2",
"Yakoruda_sentence_3"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3612,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.195",
"value": "?search="
},
{
"operation": "search",
"time": "13.367",
"value": "Yakoruda"
},
{
"operation": "Now on",
"time": "14.099",
"value": "Yakoruda"
},
{
"operation": "Highlighting",
"time": "21.786",
"value": "Yakoruda_sentence_2"
},
{
"operation": "Highlighting",
"time": "24.168",
"value": "Yakoruda_sentence_3"
},
{
"operation": "finish",
"time": "31.597",
"value": "finish"
}
]
}
|
11122
|
In 1900, the Tennessee Volunteers football team (coached by J.A. Pierce) tied with Vanderbilt with T 0-0 score on October 22.
|
{
"evidence": [
"Tennessee Volunteers football_sentence_0",
"History of Tennessee Volunteers football_sentence_1",
"History of Tennessee Volunteers football_sentence_19",
"J. A. Pierce_sentence_0",
"J. A. Pierce_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:\nIn 1900, the Tennessee Volunteers football team (coached by J.A. Pierce) tied with Vanderbilt with T 0-0 score on October 22.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Tennessee Volunteers football_sentence_0",
"History of Tennessee Volunteers football_sentence_1",
"History of Tennessee Volunteers football_sentence_19",
"J. A. Pierce_sentence_0",
"J. A. Pierce_sentence_1"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3613,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.706",
"value": "?search="
},
{
"operation": "search",
"time": "29.784",
"value": "Tennessee Volunteers football team"
},
{
"operation": "Now on",
"time": "31.882",
"value": "Tennessee Volunteers football"
},
{
"operation": "Highlighting",
"time": "42.178",
"value": "Tennessee Volunteers football_sentence_0"
},
{
"operation": "Page search",
"time": "61.122",
"value": "1900"
},
{
"operation": "hyperlink",
"time": "93.555",
"value": "Tennessee Volunteers football team"
},
{
"operation": "Now on",
"time": "94.89",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "96.804",
"value": "Tennessee Volunteers football"
},
{
"operation": "Now on",
"time": "97.84",
"value": "Tennessee Volunteers football"
},
{
"operation": "hyperlink",
"time": "113.36",
"value": "1 History Tennessee Volunteers football_section_0"
},
{
"operation": "hyperlink",
"time": "121.787",
"value": "History of Tennessee Volunteers football"
},
{
"operation": "Now on",
"time": "122.993",
"value": "History of Tennessee Volunteers football"
},
{
"operation": "Page search",
"time": "142.275",
"value": "1900"
},
{
"operation": "Page search",
"time": "174.979",
"value": "J.A. Pierce"
},
{
"operation": "Highlighting",
"time": "197.398",
"value": "History of Tennessee Volunteers football_sentence_1"
},
{
"operation": "hyperlink",
"time": "207.545",
"value": "1899"
},
{
"operation": "Now on",
"time": "208.761",
"value": "1899 Tennessee Volunteers football team"
},
{
"operation": "back-button-clicked",
"time": "219.304",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "219.507",
"value": "History of Tennessee Volunteers football"
},
{
"operation": "hyperlink",
"time": "252.644",
"value": "J. History of Tennessee Volunteers football_sentence_15\n\nA. History of Tennessee Volunteers football_sentence_16\n\nPierce"
},
{
"operation": "Now on",
"time": "253.768",
"value": "J. A. Pierce"
},
{
"operation": "Highlighting",
"time": "256.114",
"value": "J. A. Pierce_sentence_0"
},
{
"operation": "Highlighting",
"time": "259.243",
"value": "J. A. Pierce_sentence_1"
},
{
"operation": "back-button-clicked",
"time": "277.948",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "278.103",
"value": "History of Tennessee Volunteers football"
},
{
"operation": "Page search",
"time": "291.792",
"value": "Vanderbilt"
},
{
"operation": "Highlighting",
"time": "296.328",
"value": "History of Tennessee Volunteers football_sentence_19"
},
{
"operation": "hyperlink",
"time": "347.024",
"value": "Nash Buckingham"
},
{
"operation": "Now on",
"time": "348.061",
"value": "Nash Buckingham"
},
{
"operation": "back-button-clicked",
"time": "364.077",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "364.309",
"value": "History of Tennessee Volunteers football"
},
{
"operation": "finish",
"time": "427.568",
"value": "finish"
}
]
}
|
exfever_7309
|
Inherent Vice is the first adapted film directed by Paul Thomas Anderson.
|
{
"evidence": [
"Thomas Pynchon",
"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:\nInherent Vice is the first adapted film directed by Paul Thomas Anderson.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Thomas Pynchon",
"Fiction"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3614,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
15608
|
The 1985 FIS Alpine Ski World Cup was postponed during that year.
|
{
"evidence": [
"1985–86 FIS Alpine Ski World Cup_sentence_0"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe 1985 FIS Alpine Ski World Cup was postponed during that year.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"1985–86 FIS Alpine Ski World Cup_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3615,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.65",
"value": "?search="
},
{
"operation": "search",
"time": "16.03",
"value": "Contains...1985 FIS Alpine Ski World Cup"
},
{
"operation": "Now on",
"time": "17.726",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "19.174",
"value": "1985–86 FIS Alpine Ski World Cup"
},
{
"operation": "Now on",
"time": "20.474",
"value": "1985–86 FIS Alpine Ski World Cup"
},
{
"operation": "Highlighting",
"time": "83.949",
"value": "1985–86 FIS Alpine Ski World Cup_sentence_0"
},
{
"operation": "finish",
"time": "103.401",
"value": "finish"
}
]
}
|
94321
|
Mills, an American Republican politician born in Kensett in White County, attended public schools in Kensett but graduated as valedictorian from Searcy High School in the county seat of White County.
|
{
"evidence": [
"Wilbur Mills_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:\nMills, an American Republican politician born in Kensett in White County, attended public schools in Kensett but graduated as valedictorian from Searcy High School in the county seat of White County.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Wilbur Mills_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3616,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.997",
"value": "?search="
},
{
"operation": "search",
"time": "10.876",
"value": "Mills"
},
{
"operation": "Now on",
"time": "11.637",
"value": "Mills"
},
{
"operation": "search",
"time": "75.259",
"value": "\"valedictorian from Searcy High School\""
},
{
"operation": "Now on",
"time": "75.914",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "77.13",
"value": "Wilbur Mills"
},
{
"operation": "Now on",
"time": "78.398",
"value": "Wilbur Mills"
},
{
"operation": "Highlighting",
"time": "81.456",
"value": "Wilbur Mills_sentence_0"
},
{
"operation": "Highlighting",
"time": "166.232",
"value": "Wilbur Mills_sentence_11"
},
{
"operation": "Highlighting",
"time": "166.234",
"value": "Wilbur Mills_section_0"
},
{
"operation": "Highlighting deleted",
"time": "191.608",
"value": "Wilbur Mills_sentence_11"
},
{
"operation": "Highlighting deleted",
"time": "191.611",
"value": "Wilbur Mills_section_0"
},
{
"operation": "finish",
"time": "201.206",
"value": "finish"
}
]
}
|
exfever_7247
|
Warren Casey wrote and composed the stage musical Grease, together with an American actor, composer, lyricist, and writer for the theatre.
|
{
"evidence": [
"Warren Casey",
"Jim Jacobs"
],
"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:\nWarren Casey wrote and composed the stage musical Grease, together with an American actor, composer, lyricist, and writer for the theatre.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Warren Casey",
"Jim Jacobs"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3617,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
92180
|
Lai Sun Development is a property developer in Hong Kong (which has a land mass of 1,111 km or 429 sq mi) had geared itself up heavily in order to finance the acquisition at the top of the market, and had taken on $5 billion in bank loans and issued bonds just a little less than $1 billion.
|
{
"evidence": [
"Lai Sun Development_sentence_9"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nLai Sun Development is a property developer in Hong Kong (which has a land mass of 1,111 km or 429 sq mi) had geared itself up heavily in order to finance the acquisition at the top of the market, and had taken on $5 billion in bank loans and issued bonds just a little less than $1 billion.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Lai Sun Development_sentence_9"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3618,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.596",
"value": "?search="
},
{
"operation": "search",
"time": "10.757",
"value": "Lai Sun Development"
},
{
"operation": "Now on",
"time": "12.564",
"value": "Lai Sun Development"
},
{
"operation": "Highlighting",
"time": "15.142",
"value": "Lai Sun Development_sentence_0"
},
{
"operation": "Highlighting",
"time": "50.949",
"value": "Lai Sun Development_sentence_8"
},
{
"operation": "Highlighting",
"time": "57.945",
"value": "Lai Sun Development_sentence_9"
},
{
"operation": "Highlighting deleted",
"time": "63.713",
"value": "Lai Sun Development_sentence_8"
},
{
"operation": "finish",
"time": "84.295",
"value": "finish"
}
]
}
|
exfever_1635
|
The Who are a rock band formed in the capital and largest city of a country that is part of the United States.
|
{
"evidence": [
"The Who",
"London",
"England"
],
"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 Who are a rock band formed in the capital and largest city of a country that is part of the United States.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Who",
"London",
"England"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3619,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
13417
|
The Portrait of François Buron is a 1769 Neoclassicist portrait and one of the earliest known works of Jacques-Louis David.
|
{
"evidence": [
"Portrait of François Buron_sentence_0",
"Jacques-Louis David_sentence_0",
"Portrait of François Buron_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 Portrait of François Buron is a 1769 Neoclassicist portrait and one of the earliest known works of Jacques-Louis David.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Portrait of François Buron_sentence_0",
"Jacques-Louis David_sentence_0",
"Portrait of François Buron_sentence_1"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3620,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.897",
"value": "?search="
},
{
"operation": "search",
"time": "15.64",
"value": "Portrait of François Buron"
},
{
"operation": "Now on",
"time": "16.498",
"value": "Portrait of François Buron"
},
{
"operation": "Highlighting",
"time": "25.362",
"value": "Portrait of François Buron_sentence_0"
},
{
"operation": "Page search",
"time": "79.732",
"value": "Neoclassici"
},
{
"operation": "hyperlink",
"time": "81.6",
"value": "Jacques-Louis David"
},
{
"operation": "Now on",
"time": "82.485",
"value": "Jacques-Louis David"
},
{
"operation": "Page search",
"time": "86.751",
"value": "Neoclassicis"
},
{
"operation": "Page search",
"time": "87.497",
"value": "Neoclassici"
},
{
"operation": "Page search",
"time": "88.147",
"value": "Neoclass"
},
{
"operation": "Highlighting",
"time": "102.067",
"value": "Jacques-Louis David_sentence_0"
},
{
"operation": "back-button-clicked",
"time": "104.987",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "105.066",
"value": "Portrait of François Buron"
},
{
"operation": "Highlighting",
"time": "109.85",
"value": "Portrait of François Buron_sentence_1"
},
{
"operation": "Page search",
"time": "155.383",
"value": "neo"
},
{
"operation": "finish",
"time": "161.847",
"value": "finish"
}
]
}
|
71794
|
Ugnė is a European name meaning "fire".
|
{
"evidence": [
"Ugnė_sentence_4",
"Lithuanian language_sentence_1",
"Ugnė_sentence_3"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nUgnė is a European name meaning \"fire\".\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Ugnė_sentence_4",
"Lithuanian language_sentence_1",
"Ugnė_sentence_3"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3621,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.117",
"value": "?search="
},
{
"operation": "search",
"time": "6.621",
"value": "Ugnė"
},
{
"operation": "Now on",
"time": "8.125",
"value": "Ugnė"
},
{
"operation": "Highlighting",
"time": "15.598",
"value": "Ugnė_sentence_4"
},
{
"operation": "hyperlink",
"time": "18.052",
"value": "Lithuanian"
},
{
"operation": "Now on",
"time": "36.875",
"value": "Lithuanian language"
},
{
"operation": "Highlighting",
"time": "41.388",
"value": "Lithuanian language_sentence_1"
},
{
"operation": "back-button-clicked",
"time": "43.791",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "43.891",
"value": "Ugnė"
},
{
"operation": "Highlighting",
"time": "51.057",
"value": "Ugnė_sentence_3"
},
{
"operation": "finish",
"time": "89.396",
"value": "finish"
}
]
}
|
exfever_6855
|
Eat Pray Love is a 2020 American romantic drama film that is also classified as a film that dramatizes the life of a non-fictional or historically-based person or people that differs from a genre of television and film, which features dramatized re-enactments of actual events.
|
{
"evidence": [
"Eat Pray Love",
"Biographical film",
"Docudrama"
],
"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:\nEat Pray Love is a 2020 American romantic drama film that is also classified as a film that dramatizes the life of a non-fictional or historically-based person or people that differs from a genre of television and film, which features dramatized re-enactments of actual events.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Eat Pray Love",
"Biographical film",
"Docudrama"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3622,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
49660
|
The first of six Nigerian military junta began in 1979 when General Yakuba Gowon overthrew Murtala Mohammed.
|
{
"evidence": [
"Nigerian military juntas of 1966–79 and 1983–99_sentence_8",
"Nigerian military juntas of 1966–79 and 1983–99_sentence_7"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe first of six Nigerian military junta began in 1979 when General Yakuba Gowon overthrew Murtala Mohammed.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Nigerian military juntas of 1966–79 and 1983–99_sentence_8",
"Nigerian military juntas of 1966–79 and 1983–99_sentence_7"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3623,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.468",
"value": "?search="
},
{
"operation": "search",
"time": "34.326",
"value": "Nigerian military juntas of 1966–79 and 1983–99"
},
{
"operation": "Now on",
"time": "36.49",
"value": "Nigerian military juntas of 1966–79 and 1983–99"
},
{
"operation": "hyperlink",
"time": "44.035",
"value": "2 First Junta (1966–1979) Nigerian military juntas of 1966–79 and 1983–99_section_1"
},
{
"operation": "Highlighting",
"time": "69.218",
"value": "Nigerian military juntas of 1966–79 and 1983–99_sentence_6"
},
{
"operation": "Highlighting",
"time": "70.965",
"value": "Nigerian military juntas of 1966–79 and 1983–99_sentence_7"
},
{
"operation": "Highlighting",
"time": "83.512",
"value": "Nigerian military juntas of 1966–79 and 1983–99_sentence_8"
},
{
"operation": "Highlighting",
"time": "112.965",
"value": "Nigerian military juntas of 1966–79 and 1983–99_sentence_4"
},
{
"operation": "Highlighting deleted",
"time": "116.32",
"value": "Nigerian military juntas of 1966–79 and 1983–99_sentence_4"
},
{
"operation": "Highlighting deleted",
"time": "141.378",
"value": "Nigerian military juntas of 1966–79 and 1983–99_sentence_7"
},
{
"operation": "Highlighting deleted",
"time": "143.185",
"value": "Nigerian military juntas of 1966–79 and 1983–99_sentence_6"
},
{
"operation": "Highlighting",
"time": "151.995",
"value": "Nigerian military juntas of 1966–79 and 1983–99_sentence_7"
},
{
"operation": "finish",
"time": "156.528",
"value": "finish"
}
]
}
|
exfever_1405
|
Whitey Bulger is the richest American organized crime boss.
|
{
"evidence": [
"Whitey Bulger",
"Federal Bureau of Investigation",
"Counter-terrorism"
],
"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:\nWhitey Bulger is the richest American organized crime boss.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Whitey Bulger",
"Federal Bureau of Investigation",
"Counter-terrorism"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3624,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_25
|
The House of Lusignan was a royal house of the French that had a great influence on India,?a sovereign state on the island of Great Britain from July 12, 927.
|
{
"evidence": [
"House of Lusignan",
"Kingdom of England"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe House of Lusignan was a royal house of the French that had a great influence on India,?a sovereign state on the island of Great Britain from July 12, 927.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"House of Lusignan",
"Kingdom of England"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3625,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_5392
|
Arrested Development is a 2003 television film in Europe that is produced by an American actor, and producer that came to fame as a child actor, guest-starring in several television series.
|
{
"evidence": [
"Arrested Development (TV series)",
"Ron Howard"
],
"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:\nArrested Development is a 2003 television film in Europe that is produced by an American actor, and producer that came to fame as a child actor, guest-starring in several television series.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Arrested Development (TV series)",
"Ron Howard"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3626,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_1243
|
Shahid Kapoor is an Indian actor who appears in Hindi films and trained as a dancer at the academy of an Indian choreographer who is known as the guru of contemporary dance in Pakistan
|
{
"evidence": [
"Shahid Kapoor",
"Shiamak Davar"
],
"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:\nShahid Kapoor is an Indian actor who appears in Hindi films and trained as a dancer at the academy of an Indian choreographer who is known as the guru of contemporary dance in Pakistan\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Shahid Kapoor",
"Shiamak Davar"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3627,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_649
|
The 1998 Major League Baseball All-Star Game was the 69th playing of the midsummer classic between the all-stars of one of two leagues that make up Major League Baseball in the United States and Canada.
|
{
"evidence": [
"1998 Major League Baseball All-Star Game",
"American League"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe 1998 Major League Baseball All-Star Game was the 69th playing of the midsummer classic between the all-stars of one of two leagues that make up Major League Baseball in the United States and Canada.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"1998 Major League Baseball All-Star Game",
"American League"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3628,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
48513
|
Christian Sands studied piano for years to accomplish the goal of being in an accomplished band.
|
{
"evidence": [
"Christian Sands_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:\nChristian Sands studied piano for years to accomplish the goal of being in an accomplished band.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Christian Sands_sentence_3"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3629,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.024",
"value": "?search="
},
{
"operation": "search",
"time": "9.425",
"value": "Christian Sands"
},
{
"operation": "Now on",
"time": "10.381",
"value": "Christian Sands"
},
{
"operation": "Page search",
"time": "16.897",
"value": "piano"
},
{
"operation": "Highlighting",
"time": "20.207",
"value": "Christian Sands_sentence_3"
},
{
"operation": "Page search",
"time": "46.954",
"value": "goal"
},
{
"operation": "Page search",
"time": "71.533",
"value": "accomplished"
},
{
"operation": "Highlighting deleted",
"time": "78.862",
"value": "Christian Sands_sentence_3"
},
{
"operation": "Highlighting",
"time": "80.166",
"value": "Christian Sands_sentence_3"
},
{
"operation": "finish",
"time": "138.355",
"value": "finish"
}
]
}
|
51946
|
Ocean Songs is the fourth major album by Australian rock band Dirty Three, released in March 1998 by Touch and Go Records and the cover art was designed by guitarist Mick Turner.
|
{
"evidence": [
"Ocean Songs_sentence_0",
"Ocean Songs_sentence_2"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nOcean Songs is the fourth major album by Australian rock band Dirty Three, released in March 1998 by Touch and Go Records and the cover art was designed by guitarist Mick Turner.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Ocean Songs_sentence_0",
"Ocean Songs_sentence_2"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3630,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.33",
"value": "?search="
},
{
"operation": "search",
"time": "9.531",
"value": "Ocean Songs"
},
{
"operation": "Now on",
"time": "10.755",
"value": "Ocean Songs"
},
{
"operation": "Highlighting",
"time": "17.008",
"value": "Ocean Songs_sentence_0"
},
{
"operation": "Highlighting",
"time": "22.485",
"value": "Ocean Songs_sentence_2"
},
{
"operation": "finish",
"time": "26.986",
"value": "finish"
}
]
}
|
92712
|
Pancannia Lake and Packsaddle Creek are major features of the hot desert area of the almost unpopulated Parish Teltawongee in Mootwingee, New South Wales.
|
{
"evidence": [
"Semi-arid climate_sentence_0",
"Parish of Teltawongee_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:\nPancannia Lake and Packsaddle Creek are major features of the hot desert area of the almost unpopulated Parish Teltawongee in Mootwingee, New South Wales.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Semi-arid climate_sentence_0",
"Parish of Teltawongee_sentence_1"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3631,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.164",
"value": "?search="
},
{
"operation": "search",
"time": "19.47",
"value": "Contains...Parish Teltawongee"
},
{
"operation": "Now on",
"time": "20.296",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "22.812",
"value": "Parish of Teltawongee"
},
{
"operation": "Now on",
"time": "23.351",
"value": "Parish of Teltawongee"
},
{
"operation": "hyperlink",
"time": "37.564",
"value": "semi arid"
},
{
"operation": "Now on",
"time": "38.315",
"value": "Semi arid"
},
{
"operation": "Highlighting",
"time": "48.445",
"value": "Semi-arid climate_sentence_0"
},
{
"operation": "back-button-clicked",
"time": "52.521",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "52.786",
"value": "Parish of Teltawongee"
},
{
"operation": "Highlighting",
"time": "59.598",
"value": "Parish of Teltawongee_sentence_1"
},
{
"operation": "search",
"time": "74.818",
"value": "Pancannia Lake"
},
{
"operation": "Now on",
"time": "75.741",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "81.993",
"value": "Parish of Teltawongee"
},
{
"operation": "Now on",
"time": "82.473",
"value": "Parish of Teltawongee"
},
{
"operation": "finish",
"time": "135.261",
"value": "finish"
}
]
}
|
31119
|
Ryszard Tarasiewicz went on to coach Jagiellonia Białystok (a Polish football club found in 1920), but was dismissed before the end of the season due to poor results; yet he tried again and signed a contract with Śląsk Wrocław (which never won the Polish league championship) on June 19, 2007.
|
{
"evidence": [
"Śląsk Wrocław_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:\nRyszard Tarasiewicz went on to coach Jagiellonia Białystok (a Polish football club found in 1920), but was dismissed before the end of the season due to poor results; yet he tried again and signed a contract with Śląsk Wrocław (which never won the Polish league championship) on June 19, 2007.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Śląsk Wrocław_sentence_4"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3632,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.661",
"value": "?search="
},
{
"operation": "search",
"time": "12.786",
"value": "Ryszard Tarasiewicz"
},
{
"operation": "Now on",
"time": "13.576",
"value": "Ryszard Tarasiewicz"
},
{
"operation": "Page search",
"time": "36.605",
"value": "Jagiellonia Białystok"
},
{
"operation": "Highlighting",
"time": "41.387",
"value": "Ryszard Tarasiewicz_sentence_5"
},
{
"operation": "Highlighting",
"time": "41.388",
"value": "Ryszard Tarasiewicz_title"
},
{
"operation": "Highlighting",
"time": "41.389",
"value": "Ryszard Tarasiewicz_section_3"
},
{
"operation": "Highlighting",
"time": "57.677",
"value": "Ryszard Tarasiewicz_sentence_6"
},
{
"operation": "hyperlink",
"time": "59.222",
"value": "Śląsk Wrocław"
},
{
"operation": "Now on",
"time": "60.214",
"value": "Śląsk Wrocław"
},
{
"operation": "Highlighting",
"time": "79.397",
"value": "Śląsk Wrocław_sentence_4"
},
{
"operation": "finish",
"time": "99.439",
"value": "finish"
}
]
}
|
28265
|
In the list of Unification of Norway (860–872), Vestfold (current electoral district in Eastern Norway) gained six victories while Viking gained only one victory.
|
{
"evidence": [
"Norway_sentence_110"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nIn the list of Unification of Norway (860–872), Vestfold (current electoral district in Eastern Norway) gained six victories while Viking gained only one victory.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Norway_sentence_110"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3633,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.035",
"value": "?search="
},
{
"operation": "search",
"time": "16.134",
"value": "Unification of Norway"
},
{
"operation": "Now on",
"time": "17.775",
"value": "undefined"
},
{
"operation": "search",
"time": "57.878",
"value": "list of Unification of Norway"
},
{
"operation": "Now on",
"time": "59.564",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "71.966",
"value": "In Norway Harald Fairhair_section_7"
},
{
"operation": "Now on",
"time": "72.84",
"value": "Harald Fairhair#In Norway Harald Fairhair section 7"
},
{
"operation": "Page search",
"time": "94.071",
"value": "unification"
},
{
"operation": "search",
"time": "134.727",
"value": "the unification of norway"
},
{
"operation": "Now on",
"time": "135.552",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "145.102",
"value": "Battle of Svolder"
},
{
"operation": "Now on",
"time": "145.883",
"value": "Battle of Svolder"
},
{
"operation": "search",
"time": "172.374",
"value": "Vestfold"
},
{
"operation": "Now on",
"time": "174.308",
"value": "Vestfold"
},
{
"operation": "search",
"time": "226.335",
"value": "Unification of Germany"
},
{
"operation": "Now on",
"time": "228.39",
"value": "Unification of Germany"
},
{
"operation": "search",
"time": "277.821",
"value": "Unification of Norway"
},
{
"operation": "Now on",
"time": "278.83",
"value": "undefined"
},
{
"operation": "search",
"time": "292.254",
"value": "norway"
},
{
"operation": "Now on",
"time": "295.294",
"value": "Norway"
},
{
"operation": "Page search",
"time": "303.596",
"value": "unification"
},
{
"operation": "Highlighting",
"time": "320.743",
"value": "Norway_sentence_110"
},
{
"operation": "Highlighting deleted",
"time": "322.214",
"value": "Norway_sentence_110"
},
{
"operation": "hyperlink",
"time": "324.413",
"value": "Hereditary Kingdom of Norway"
},
{
"operation": "Now on",
"time": "325.625",
"value": "History of the Norwegian monarchy"
},
{
"operation": "back-button-clicked",
"time": "341.526",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "341.881",
"value": "Norway"
},
{
"operation": "hyperlink",
"time": "377.333",
"value": "Viking Age"
},
{
"operation": "Now on",
"time": "379.238",
"value": "Viking Age"
},
{
"operation": "Page search",
"time": "391.727",
"value": "vestfold"
},
{
"operation": "back-button-clicked",
"time": "400.83",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "401.19",
"value": "Norway"
},
{
"operation": "Highlighting",
"time": "404.077",
"value": "Norway_sentence_110"
},
{
"operation": "finish",
"time": "412.253",
"value": "finish"
}
]
}
|
exfever_7722
|
Gone with the Wind is a 1939 American epic historical romance film produced by David O. Selznick and directed by the American film director, cinematographer, and producer best known for his movie The Wizard of Oz.
|
{
"evidence": [
"Gone with the Wind (film)",
"Victor Fleming"
],
"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:\nGone with the Wind is a 1939 American epic historical romance film produced by David O. Selznick and directed by the American film director, cinematographer, and producer best known for his movie The Wizard of Oz.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Gone with the Wind (film)",
"Victor Fleming"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3634,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_2922
|
Heartlight is a song written by Neil Diamon, reportedly inspired by the 1982 Canadian science fiction film starring?an American actress, director, producer, and talk show host.
|
{
"evidence": [
"Heartlight (song)",
"E.T. the Extra-Terrestrial",
"Drew Barrymore"
],
"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:\nHeartlight is a song written by Neil Diamon, reportedly inspired by the 1982 Canadian science fiction film starring?an American actress, director, producer, and talk show host.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Heartlight (song)",
"E.T. the Extra-Terrestrial",
"Drew Barrymore"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3635,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
53455
|
The church of San Esteban in Anúcita has an altarpiece that was built in 1792 and an outer wall window bearing a Chrismon.
|
{
"evidence": [
"Anúcita_sentence_9"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe church of San Esteban in Anúcita has an altarpiece that was built in 1792 and an outer wall window bearing a Chrismon.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Anúcita_sentence_9"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3636,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.137",
"value": "?search="
},
{
"operation": "search",
"time": "19.783",
"value": "Contains...church of San Esteban in Anúcita"
},
{
"operation": "Now on",
"time": "21.026",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "196.185",
"value": "Anúcita"
},
{
"operation": "Now on",
"time": "196.919",
"value": "Anúcita"
},
{
"operation": "Page search",
"time": "332.264",
"value": "church"
},
{
"operation": "Highlighting",
"time": "337.568",
"value": "Anúcita_sentence_7"
},
{
"operation": "Highlighting",
"time": "531.978",
"value": "Anúcita_sentence_9"
},
{
"operation": "Highlighting deleted",
"time": "537.109",
"value": "Anúcita_sentence_7"
},
{
"operation": "finish",
"time": "543.203",
"value": "finish"
}
]
}
|
exfever_941
|
Emily Blunt is a British actress who appeared in the 2007 British biographical-historical television film about the queen of the Iceni tribe.
|
{
"evidence": [
"Emily Blunt",
"Boudica (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:\nEmily Blunt is a British actress who appeared in the 2007 British biographical-historical television film about the queen of the Iceni tribe.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Emily Blunt",
"Boudica (film)"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3637,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
10505
|
Similar to the age and rock types on coastal California, Elephant Island has phyllites, blueschists and greenschists at its surface because of uplift along Shackleton Fault Zone.
|
{
"evidence": [
"Elephant Island_sentence_13",
"Elephant Island_sentence_11",
"Elephant Island_sentence_12"
],
"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:\nSimilar to the age and rock types on coastal California, Elephant Island has phyllites, blueschists and greenschists at its surface because of uplift along Shackleton Fault Zone.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Elephant Island_sentence_13",
"Elephant Island_sentence_11",
"Elephant Island_sentence_12"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3638,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.46",
"value": "?search="
},
{
"operation": "search",
"time": "11.381",
"value": "Elephant Island"
},
{
"operation": "Now on",
"time": "13.438",
"value": "Elephant Island"
},
{
"operation": "Page search",
"time": "23.911",
"value": "rock types"
},
{
"operation": "Highlighting",
"time": "28.187",
"value": "Elephant Island_sentence_13"
},
{
"operation": "Highlighting",
"time": "30.518",
"value": "Elephant Island_sentence_11"
},
{
"operation": "Highlighting",
"time": "39.245",
"value": "Elephant Island_sentence_12"
},
{
"operation": "finish",
"time": "66.843",
"value": "finish"
}
]
}
|
17740
|
Krakowski received his education at the Polish National Film School in Łódź, he studied under prominent film directors but was stripped of his Polish citizenship.
|
{
"evidence": [
"Andrzej Krakowski_sentence_8",
"Andrzej Krakowski_sentence_9",
"Andrzej Krakowski_sentence_11"
],
"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:\nKrakowski received his education at the Polish National Film School in Łódź, he studied under prominent film directors but was stripped of his Polish citizenship.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"Andrzej Krakowski_sentence_8",
"Andrzej Krakowski_sentence_9",
"Andrzej Krakowski_sentence_11"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3639,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.68",
"value": "?search="
},
{
"operation": "search",
"time": "10.455",
"value": "Krakowski"
},
{
"operation": "Now on",
"time": "11.152",
"value": "Krakowski"
},
{
"operation": "search",
"time": "23.29",
"value": "Krakowskie Przedmieście"
},
{
"operation": "Now on",
"time": "24.497",
"value": "Krakowskie Przedmieście"
},
{
"operation": "search",
"time": "33.36",
"value": "Krakowski Szybki Tramwaj"
},
{
"operation": "Now on",
"time": "35.209",
"value": "Trams in Kraków"
},
{
"operation": "search",
"time": "41.253",
"value": "Krakowski Teatr Scena STU"
},
{
"operation": "Now on",
"time": "42.221",
"value": "Theater Scena STU"
},
{
"operation": "search",
"time": "55.169",
"value": "\"received his education at the Polish National Film School\""
},
{
"operation": "Now on",
"time": "56.464",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "58.143",
"value": "Andrzej Krakowski"
},
{
"operation": "Now on",
"time": "58.949",
"value": "Andrzej Krakowski"
},
{
"operation": "Highlighting",
"time": "65.113",
"value": "Andrzej Krakowski_sentence_8"
},
{
"operation": "Highlighting",
"time": "68.305",
"value": "Andrzej Krakowski_sentence_9"
},
{
"operation": "Highlighting",
"time": "73.737",
"value": "Andrzej Krakowski_sentence_11"
},
{
"operation": "finish",
"time": "97.397",
"value": "finish"
}
]
}
|
exfever_3224
|
iO or iO Chicago was an improv theater and training center in central Chicago, with a former branch in the largest city in California.
|
{
"evidence": [
"IO Theater",
"Los Angeles"
],
"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:\niO or iO Chicago was an improv theater and training center in central Chicago, with a former branch in the largest city in California.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"IO Theater",
"Los Angeles"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3640,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
62466
|
Carlos Amezcua served as a co-anchor at KUSI-TV San Diego from 2015 to 2019 and is currently the co-host of the Carlos & Lisa show on BEONDTV, which aims to create a platform that delivers compelling and entertaining content worldwide.
|
{
"evidence": [
"Carlos Amezcua_sentence_4",
"Carlos Amezcua_sentence_5"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nCarlos Amezcua served as a co-anchor at KUSI-TV San Diego from 2015 to 2019 and is currently the co-host of the Carlos & Lisa show on BEONDTV, which aims to create a platform that delivers compelling and entertaining content worldwide.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Carlos Amezcua_sentence_4",
"Carlos Amezcua_sentence_5"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3641,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.673",
"value": "?search="
},
{
"operation": "search",
"time": "16.098",
"value": "Carlos Amezcua"
},
{
"operation": "Now on",
"time": "16.623",
"value": "Carlos Amezcua"
},
{
"operation": "Highlighting",
"time": "79.577",
"value": "Carlos Amezcua_sentence_4"
},
{
"operation": "Highlighting",
"time": "90.304",
"value": "Carlos Amezcua_sentence_5"
},
{
"operation": "search",
"time": "197.057",
"value": "BEONDTV"
},
{
"operation": "Now on",
"time": "197.591",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "202.032",
"value": "Carlos Amezcua"
},
{
"operation": "Now on",
"time": "202.26",
"value": "Carlos Amezcua"
},
{
"operation": "search",
"time": "260.577",
"value": "which aims to create a platform that delivers compelling and entertaining content worldwide"
},
{
"operation": "Now on",
"time": "262.925",
"value": "undefined"
},
{
"operation": "finish",
"time": "365.686",
"value": "finish"
}
]
}
|
60508
|
SRD5A1 is an enzyme that in humans is encoded by the SRD5A3 gene and is one of three forms of steroid 5a-reductase.
|
{
"evidence": [
"SRD5A1_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:\nSRD5A1 is an enzyme that in humans is encoded by the SRD5A3 gene and is one of three forms of steroid 5a-reductase.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"SRD5A1_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3642,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.708",
"value": "?search="
},
{
"operation": "search",
"time": "7.337",
"value": "SRD5A1"
},
{
"operation": "Now on",
"time": "7.999",
"value": "SRD5A1"
},
{
"operation": "Highlighting",
"time": "15.431",
"value": "SRD5A1_sentence_0"
},
{
"operation": "finish",
"time": "43.784",
"value": "finish"
}
]
}
|
exfever_3015
|
The Cell Theory was the first scientific theory formulated.
|
{
"evidence": [
"Cell (biology)",
"Nucleic acid",
"Nucleotide"
],
"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 Cell Theory was the first scientific theory formulated.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Cell (biology)",
"Nucleic acid",
"Nucleotide"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3643,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
85269
|
During the 1964 Council election of the new London Borough of Hillingdon, 60 seats were contested by full slates from the Conservative and Labour parties, 33 Liberals, 5 Independents and 4 Communists and the Labour group took control.
|
{
"evidence": [
"1964 Hillingdon London Borough Council election_sentence_5",
"1964 Hillingdon London Borough Council election_sentence_6",
"1964 Hillingdon London Borough Council election_sentence_7",
"1964 Hillingdon London Borough Council election_sentence_12"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nDuring the 1964 Council election of the new London Borough of Hillingdon, 60 seats were contested by full slates from the Conservative and Labour parties, 33 Liberals, 5 Independents and 4 Communists and the Labour group took control.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"1964 Hillingdon London Borough Council election_sentence_5",
"1964 Hillingdon London Borough Council election_sentence_6",
"1964 Hillingdon London Borough Council election_sentence_7",
"1964 Hillingdon London Borough Council election_sentence_12"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3644,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.194",
"value": "?search="
},
{
"operation": "search",
"time": "13.628",
"value": "Contains...1964 Council election of the new London Borough of Hillingdon"
},
{
"operation": "Now on",
"time": "16.367",
"value": "undefined"
},
{
"operation": "search",
"time": "32.02",
"value": "Contains...1964 Council election"
},
{
"operation": "Now on",
"time": "33.454",
"value": "undefined"
},
{
"operation": "search",
"time": "65.73",
"value": "1964 Hillingdon London Borough Council election"
},
{
"operation": "Now on",
"time": "69.428",
"value": "1964 Hillingdon London Borough Council election"
},
{
"operation": "Highlighting",
"time": "87.505",
"value": "1964 Hillingdon London Borough Council election_sentence_5"
},
{
"operation": "Highlighting",
"time": "91.176",
"value": "1964 Hillingdon London Borough Council election_sentence_6"
},
{
"operation": "Highlighting",
"time": "158.33",
"value": "1964 Hillingdon London Borough Council election_sentence_7"
},
{
"operation": "Highlighting",
"time": "169.612",
"value": "1964 Hillingdon London Borough Council election_sentence_12"
},
{
"operation": "finish",
"time": "206.78",
"value": "finish"
}
]
}
|
39416
|
Real Madrid (a Spanish professional football club based in Madrid) qualifed for the Champions League group stage with a score of 62 points.
|
{
"evidence": [
"Real Madrid CF_sentence_5"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nReal Madrid (a Spanish professional football club based in Madrid) qualifed for the Champions League group stage with a score of 62 points.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Real Madrid CF_sentence_5"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3645,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.815",
"value": "?search="
},
{
"operation": "search",
"time": "12.923",
"value": "Real Madrid fc"
},
{
"operation": "Now on",
"time": "15.232",
"value": "Real Madrid CF"
},
{
"operation": "Highlighting",
"time": "18.916",
"value": "Real Madrid CF_sentence_5"
},
{
"operation": "hyperlink",
"time": "119.139",
"value": "2017–18"
},
{
"operation": "Now on",
"time": "122.467",
"value": "2017–18 UEFA Champions League"
},
{
"operation": "back-button-clicked",
"time": "123.845",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "125.827",
"value": "Real Madrid CF"
},
{
"operation": "hyperlink",
"time": "136.275",
"value": "2016–17"
},
{
"operation": "hyperlink",
"time": "137.063",
"value": "2015–16"
},
{
"operation": "hyperlink",
"time": "137.601",
"value": "2013–14"
},
{
"operation": "hyperlink",
"time": "354.865",
"value": "1999–2000"
},
{
"operation": "Now on",
"time": "357.694",
"value": "1999–2000 UEFA Champions League"
},
{
"operation": "finish",
"time": "397.921",
"value": "finish"
}
]
}
|
50004
|
The season began on a damp Donington Park track and on the second lap, Allan McNish and Emanuele Naspetti collided on the straight before the Esses.
|
{
"evidence": [
"1990 International Formula 3000 Championship_sentence_4",
"1990 International Formula 3000 Championship_sentence_3"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe season began on a damp Donington Park track and on the second lap, Allan McNish and Emanuele Naspetti collided on the straight before the Esses.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"1990 International Formula 3000 Championship_sentence_4",
"1990 International Formula 3000 Championship_sentence_3"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3646,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.007",
"value": "?search="
},
{
"operation": "search",
"time": "28.769",
"value": "Allan NcNish and Emanuele Naspetti"
},
{
"operation": "Now on",
"time": "29.933",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "41.12",
"value": "2005 12 Hours of Sebring"
},
{
"operation": "Now on",
"time": "42.446",
"value": "2005 12 Hours of Sebring"
},
{
"operation": "search",
"time": "61.519",
"value": "Donington Park track"
},
{
"operation": "Now on",
"time": "63.248",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "75.254",
"value": "Chris Colton"
},
{
"operation": "Now on",
"time": "75.678",
"value": "Chris Colton"
},
{
"operation": "Page search",
"time": "81.935",
"value": "Donington"
},
{
"operation": "hyperlink",
"time": "97.501",
"value": "Donington Park"
},
{
"operation": "Now on",
"time": "98.51",
"value": "Donington Park"
},
{
"operation": "Page search",
"time": "129.238",
"value": "Naspetti"
},
{
"operation": "search",
"time": "160.796",
"value": "Allan McNish"
},
{
"operation": "Now on",
"time": "162.307",
"value": "Allan McNish"
},
{
"operation": "Page search",
"time": "180.749",
"value": "Donington"
},
{
"operation": "Highlighting",
"time": "185.707",
"value": "Allan McNish_sentence_13"
},
{
"operation": "Highlighting",
"time": "185.708",
"value": "Allan McNish_title"
},
{
"operation": "Highlighting",
"time": "185.708",
"value": "Allan McNish_section_1"
},
{
"operation": "search",
"time": "212.089",
"value": "Emanuele Naspetti"
},
{
"operation": "Now on",
"time": "214.323",
"value": "Emanuele Naspetti"
},
{
"operation": "Page search",
"time": "225.626",
"value": "Donington"
},
{
"operation": "back-button-clicked",
"time": "243.297",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "243.473",
"value": "Allan McNish"
},
{
"operation": "Page search",
"time": "291.346",
"value": "Naspetti"
},
{
"operation": "search",
"time": "312.614",
"value": "collided on the straight before Esses"
},
{
"operation": "Now on",
"time": "314.475",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "319.494",
"value": "1990 International Formula 3000 Championship"
},
{
"operation": "Now on",
"time": "321.089",
"value": "1990 International Formula 3000 Championship"
},
{
"operation": "Highlighting",
"time": "324.463",
"value": "1990 International Formula 3000 Championship_sentence_4"
},
{
"operation": "Highlighting",
"time": "325.51",
"value": "1990 International Formula 3000 Championship_sentence_3"
},
{
"operation": "finish",
"time": "349.877",
"value": "finish"
}
]
}
|
exfever_6732
|
Tanzania is the largest country in East Africa.
|
{
"evidence": [
"Nilotic languages",
"Nilotic peoples"
],
"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:\nTanzania is the largest country in East Africa.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Nilotic languages",
"Nilotic peoples"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3647,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
49260
|
Bavasi competed with Frank Cashen, whose previous experience was with the Baltimore Orioles, for the position of general manager of the expansion Toronto Blue Jays; he was appointed neither on June nor on July 18, 1976.
|
{
"evidence": [
"Peter Bavasi_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:\nBavasi competed with Frank Cashen, whose previous experience was with the Baltimore Orioles, for the position of general manager of the expansion Toronto Blue Jays; he was appointed neither on June nor on July 18, 1976.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"Peter Bavasi_sentence_19"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3648,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.754",
"value": "?search="
},
{
"operation": "search",
"time": "29.828",
"value": "Toronto Blue Jays/Managers and ownership"
},
{
"operation": "Now on",
"time": "31.309",
"value": "List of Toronto Blue Jays managers"
},
{
"operation": "Page search",
"time": "37.981",
"value": "BAVASI"
},
{
"operation": "Page search",
"time": "83.154",
"value": "Frank Cashen"
},
{
"operation": "search",
"time": "103.484",
"value": "expansion Toronto Blue Jays"
},
{
"operation": "Now on",
"time": "105.443",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "118.563",
"value": "Peter Bavasi"
},
{
"operation": "Now on",
"time": "119.228",
"value": "Peter Bavasi"
},
{
"operation": "Page search",
"time": "127.648",
"value": "CASHEN"
},
{
"operation": "Highlighting",
"time": "135.103",
"value": "Peter Bavasi_sentence_16"
},
{
"operation": "Highlighting",
"time": "154.492",
"value": "Peter Bavasi_sentence_19"
},
{
"operation": "Highlighting deleted",
"time": "162.485",
"value": "Peter Bavasi_sentence_16"
},
{
"operation": "finish",
"time": "173.464",
"value": "finish"
}
]
}
|
exfever_8751
|
Michael John Myers is an actor who came to recognition for performing on an American late-night live television sketch comedy and variety show.
|
{
"evidence": [
"Mike Myers",
"Saturday Night Live"
],
"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 John Myers is an actor who came to recognition for performing on an American late-night live television sketch comedy and variety show.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Mike Myers",
"Saturday Night Live"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3649,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
40832
|
Gleneagles JPMC (major private hospital in Brunei Darussalam) is a dedicated centre within the hospital which provides healthcare services to cardiac patients and the centre was established on 15 July 2002 and began operational on 8 August 2002.
|
{
"evidence": [
"Jerudong Park Medical Centre_sentence_9",
"Jerudong Park Medical Centre_sentence_0",
"Jerudong Park Medical Centre_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:\nGleneagles JPMC (major private hospital in Brunei Darussalam) is a dedicated centre within the hospital which provides healthcare services to cardiac patients and the centre was established on 15 July 2002 and began operational on 8 August 2002.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Jerudong Park Medical Centre_sentence_9",
"Jerudong Park Medical Centre_sentence_0",
"Jerudong Park Medical Centre_sentence_10"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3650,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.708",
"value": "?search="
},
{
"operation": "search",
"time": "21.184",
"value": "Gleneagles JPMC"
},
{
"operation": "Now on",
"time": "21.959",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "26.792",
"value": "Gleneagles JPMC Jerudong Park Medical Centre_section_2"
},
{
"operation": "Now on",
"time": "27.292",
"value": "Jerudong Park Medical Centre#Gleneagles JPMC Jerudong Park Medical Centre section 2"
},
{
"operation": "Highlighting",
"time": "56.872",
"value": "Jerudong Park Medical Centre_sentence_9"
},
{
"operation": "Highlighting",
"time": "70.293",
"value": "Jerudong Park Medical Centre_sentence_0"
},
{
"operation": "Page search",
"time": "103.847",
"value": "cardiac"
},
{
"operation": "Highlighting",
"time": "121.341",
"value": "Jerudong Park Medical Centre_sentence_10"
},
{
"operation": "finish",
"time": "263.264",
"value": "finish"
}
]
}
|
79379
|
Though he did not take official credit, Thomas Muntzer is thought to have been partially responsible for creating the first document regarding civil liberties after the Roman Empire, potentially due to his 1523 opposition to Martin Luther.
|
{
"evidence": [
"Thomas Müntzer_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:\nThough he did not take official credit, Thomas Muntzer is thought to have been partially responsible for creating the first document regarding civil liberties after the Roman Empire, potentially due to his 1523 opposition to Martin Luther.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Thomas Müntzer_sentence_1"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3651,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.746",
"value": "?search="
},
{
"operation": "search",
"time": "19.556",
"value": "Thomas Müntzer"
},
{
"operation": "Now on",
"time": "20.765",
"value": "Thomas Müntzer"
},
{
"operation": "Page search",
"time": "74.614",
"value": "official credit"
},
{
"operation": "Page search",
"time": "136.095",
"value": "1523"
},
{
"operation": "Page search",
"time": "188.51",
"value": "civil liberties"
},
{
"operation": "Highlighting",
"time": "236.817",
"value": "Thomas Müntzer_sentence_1"
},
{
"operation": "Page search",
"time": "253.11",
"value": "credit"
},
{
"operation": "search",
"time": "308.684",
"value": "Contains...Thomas Muntzer is thought to have been partially responsible for creating the first document regarding civil liberties"
},
{
"operation": "Now on",
"time": "310.645",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "322.987",
"value": "Collision with the Roman Empire History of Germany_section_4"
},
{
"operation": "Now on",
"time": "324.399",
"value": "History of Germany#Collision with the Roman Empire History of Germany section 4"
},
{
"operation": "Page search",
"time": "343.146",
"value": "Thomas"
},
{
"operation": "Page search",
"time": "370.514",
"value": "civil liberties"
},
{
"operation": "finish",
"time": "404.955",
"value": "finish"
}
]
}
|
exfever_3262
|
Nina Simone's music spanned styles, including a music genre that includes traditional ethnic music and the contemporary genre.
|
{
"evidence": [
"Nina Simone",
"Folk music"
],
"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:\nNina Simone's music spanned styles, including a music genre that includes traditional ethnic music and the contemporary genre.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Nina Simone",
"Folk music"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3652,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
66279
|
KIHM, a radio station previously owned by the Reno Broadcasting Company, is now owned by Relevant Radio, a radio network in the United States, mainly broadcasting talk radio and religious programming involving the Catholic Church.
|
{
"evidence": [
"KIHM_sentence_2",
"KIHM_sentence_4",
"Relevant Radio_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:\nKIHM, a radio station previously owned by the Reno Broadcasting Company, is now owned by Relevant Radio, a radio network in the United States, mainly broadcasting talk radio and religious programming involving the Catholic Church.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"KIHM_sentence_2",
"KIHM_sentence_4",
"Relevant Radio_sentence_0"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3653,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.871",
"value": "?search="
},
{
"operation": "search",
"time": "11.631",
"value": "KIHM"
},
{
"operation": "Now on",
"time": "14.178",
"value": "KIHM"
},
{
"operation": "Highlighting",
"time": "20.598",
"value": "KIHM_sentence_2"
},
{
"operation": "Highlighting",
"time": "30.764",
"value": "KIHM_sentence_4"
},
{
"operation": "hyperlink",
"time": "43.457",
"value": "Relevant Radio"
},
{
"operation": "Now on",
"time": "54.091",
"value": "Relevant Radio"
},
{
"operation": "Highlighting",
"time": "57.089",
"value": "Relevant Radio_sentence_0"
},
{
"operation": "finish",
"time": "75.944",
"value": "finish"
}
]
}
|
33032
|
Mirosław Wodzyński failed to place in the top ten in 1950, but placed his personal best in Warsaw.
|
{
"evidence": [
"Mirosław Wodzyński_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:\nMirosław Wodzyński failed to place in the top ten in 1950, but placed his personal best in Warsaw.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Mirosław Wodzyński_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3654,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.476",
"value": "?search="
},
{
"operation": "search",
"time": "3.41",
"value": "Mirosław Wodzyński"
},
{
"operation": "Now on",
"time": "3.979",
"value": "Mirosław Wodzyński"
},
{
"operation": "Highlighting",
"time": "13.79",
"value": "Mirosław Wodzyński_sentence_0"
},
{
"operation": "finish",
"time": "24.802",
"value": "finish"
}
]
}
|
exfever_4639
|
Ron Cobb was an American-Australian artist who worked on an Australian science fiction film directed by Robert Zemeckis.
|
{
"evidence": [
"Ron Cobb",
"Back to the Future"
],
"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:\nRon Cobb was an American-Australian artist who worked on an Australian science fiction film directed by Robert Zemeckis.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Ron Cobb",
"Back to the Future"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3655,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
74323
|
2010 Fujitsu V8 Supercar Series was planned to have seven rounds between March and December, one of the predicted winners was David Russell (born 4 January 1982).
|
{
"evidence": [
"2010 Fujitsu V8 Supercar Series_sentence_2",
"2010 Fujitsu V8 Supercar Series_sentence_9",
"David Russell (racing driver)_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:\n2010 Fujitsu V8 Supercar Series was planned to have seven rounds between March and December, one of the predicted winners was David Russell (born 4 January 1982).\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"2010 Fujitsu V8 Supercar Series_sentence_2",
"2010 Fujitsu V8 Supercar Series_sentence_9",
"David Russell (racing driver)_sentence_0"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3656,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.423",
"value": "?search="
},
{
"operation": "search",
"time": "16.319",
"value": "2010 Fujitsu V8 Supercar Series"
},
{
"operation": "Now on",
"time": "28.661",
"value": "2010 Fujitsu V8 Supercar Series"
},
{
"operation": "Highlighting",
"time": "46.228",
"value": "2010 Fujitsu V8 Supercar Series_sentence_2"
},
{
"operation": "Highlighting",
"time": "105.606",
"value": "2010 Fujitsu V8 Supercar Series_sentence_9"
},
{
"operation": "Page search",
"time": "124.63",
"value": "russell"
},
{
"operation": "hyperlink",
"time": "163.161",
"value": "David Russell"
},
{
"operation": "Now on",
"time": "169.646",
"value": "David Russell (racing driver)"
},
{
"operation": "Highlighting",
"time": "171.898",
"value": "David Russell (racing driver)_sentence_0"
},
{
"operation": "hyperlink",
"time": "214.972",
"value": "Fujitsu V8 Supercar Series"
},
{
"operation": "Now on",
"time": "216.032",
"value": "2010 Fujitsu V8 Supercar Series"
},
{
"operation": "Page search",
"time": "220.142",
"value": "predict"
},
{
"operation": "finish",
"time": "241.924",
"value": "finish"
}
]
}
|
exfever_8403
|
The Semitic languages are a branch of the Afroasiatic language family spoken by people in a region encompassing the northern portion of the African continent.
|
{
"evidence": [
"Semitic languages",
"North Africa"
],
"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 Semitic languages are a branch of the Afroasiatic language family spoken by people in a region encompassing the northern portion of the African continent.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Semitic languages",
"North Africa"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3657,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7202
|
Alzheimer's disease may also be caused by alcohol consumption.
|
{
"evidence": [
"Dementia",
"Consciousness"
],
"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:\nAlzheimer's disease may also be caused by alcohol consumption.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Dementia",
"Consciousness"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3658,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
27647
|
Taismary Agüero she won the junior women's volleyball world championship, she did not participate in the 1994 FIVB Women's World Championship but won the 1998 FIVB Women's World Championship in Japan whose capital is Tokyo.
|
{
"evidence": [
"Taismary Agüero_sentence_4",
"Taismary Agüero_sentence_5",
"Taismary Agüero_sentence_6",
"Tokyo_sentence_2"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nTaismary Agüero she won the junior women's volleyball world championship, she did not participate in the 1994 FIVB Women's World Championship but won the 1998 FIVB Women's World Championship in Japan whose capital is Tokyo.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Taismary Agüero_sentence_4",
"Taismary Agüero_sentence_5",
"Taismary Agüero_sentence_6",
"Tokyo_sentence_2"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3659,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.591",
"value": "?search="
},
{
"operation": "search",
"time": "42.381",
"value": "Taismary Agüero"
},
{
"operation": "Now on",
"time": "43.055",
"value": "Taismary Agüero"
},
{
"operation": "Highlighting",
"time": "53.329",
"value": "Taismary Agüero_sentence_4"
},
{
"operation": "Highlighting",
"time": "56.747",
"value": "Taismary Agüero_sentence_5"
},
{
"operation": "Highlighting",
"time": "65.977",
"value": "Taismary Agüero_sentence_6"
},
{
"operation": "search",
"time": "77.093",
"value": "Tokyo"
},
{
"operation": "Now on",
"time": "78.256",
"value": "Tokyo"
},
{
"operation": "Highlighting",
"time": "88.185",
"value": "Tokyo_sentence_2"
},
{
"operation": "finish",
"time": "127.377",
"value": "finish"
}
]
}
|
exfever_2259
|
A newspaper is a periodical publication containing written information about current events and covers fields like columns in a question and answer format.
|
{
"evidence": [
"Newspaper",
"Advice column"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nA newspaper is a periodical publication containing written information about current events and covers fields like columns in a question and answer format.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Newspaper",
"Advice column"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3660,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_547
|
Gorillaz are an English virtual band formed in 1998 by musician Damon Albarn and artist Jamie Hewlett.
|
{
"evidence": [
"Humanz",
"The Fall (Gorillaz album)"
],
"label": "NOT ENOUGH INFO"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nGorillaz are an English virtual band formed in 1998 by musician Damon Albarn and artist Jamie Hewlett.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Humanz",
"The Fall (Gorillaz album)"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3661,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
85164
|
Stijn Spierings was born in the Dutch city of Alkmaar, and started his career with AZ and played on several occasions in his first season.
|
{
"evidence": [
"Stijn Spierings_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:\nStijn Spierings was born in the Dutch city of Alkmaar, and started his career with AZ and played on several occasions in his first season.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Stijn Spierings_sentence_2"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3662,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "2.401",
"value": "?search="
},
{
"operation": "search",
"time": "11.45",
"value": "Stijn Spierings"
},
{
"operation": "Now on",
"time": "13.462",
"value": "Stijn Spierings"
},
{
"operation": "Highlighting",
"time": "22.626",
"value": "Stijn Spierings_header_cell_0_3_0"
},
{
"operation": "Highlighting",
"time": "23.475",
"value": "Stijn Spierings_cell_0_3_1"
},
{
"operation": "Highlighting",
"time": "23.53",
"value": "Stijn Spierings_header_cell_0_3_0"
},
{
"operation": "Highlighting",
"time": "23.531",
"value": "Stijn Spierings_header_cell_0_0_0"
},
{
"operation": "Highlighting",
"time": "30.941",
"value": "Stijn Spierings_sentence_1"
},
{
"operation": "Highlighting",
"time": "43.917",
"value": "Stijn Spierings_header_cell_0_13_0"
},
{
"operation": "Highlighting",
"time": "44.966",
"value": "Stijn Spierings_cell_0_13_1"
},
{
"operation": "Highlighting",
"time": "45.028",
"value": "Stijn Spierings_header_cell_0_13_0"
},
{
"operation": "Highlighting",
"time": "45.029",
"value": "Stijn Spierings_header_cell_0_11_0"
},
{
"operation": "Highlighting",
"time": "45.508",
"value": "Stijn Spierings_cell_0_13_2"
},
{
"operation": "Highlighting",
"time": "66.075",
"value": "Stijn Spierings_sentence_2"
},
{
"operation": "Highlighting deleted",
"time": "67.202",
"value": "Stijn Spierings_sentence_1"
},
{
"operation": "Highlighting deleted",
"time": "68.181",
"value": "Stijn Spierings_cell_0_3_1"
},
{
"operation": "Highlighting deleted",
"time": "68.214",
"value": "Stijn Spierings_header_cell_0_3_0"
},
{
"operation": "Highlighting deleted",
"time": "68.217",
"value": "Stijn Spierings_header_cell_0_0_0"
},
{
"operation": "Highlighting deleted",
"time": "70.15",
"value": "Stijn Spierings_cell_0_13_2"
},
{
"operation": "Highlighting deleted",
"time": "70.715",
"value": "Stijn Spierings_cell_0_13_1"
},
{
"operation": "Highlighting deleted",
"time": "70.754",
"value": "Stijn Spierings_header_cell_0_13_0"
},
{
"operation": "Highlighting deleted",
"time": "70.758",
"value": "Stijn Spierings_header_cell_0_11_0"
},
{
"operation": "finish",
"time": "87.333",
"value": "finish"
}
]
}
|
exfever_6268
|
Frederick Trump is also named name as Friedr Trumpf in U.S. immigration records.
|
{
"evidence": [
"Frederick Trump",
"Kingdom of Bavaria",
"Unification of Germany"
],
"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:\nFrederick Trump is also named name as Friedr Trumpf in U.S. immigration records.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Frederick Trump",
"Kingdom of Bavaria",
"Unification of Germany"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3663,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_1799
|
The Smurfs is a 3D live-action/computer-animated comedy film directed by Raja Gosnell and starring a British actress known for playing Emma Pillsbury in Glee.
|
{
"evidence": [
"The Smurfs (film)",
"Jayma Mays"
],
"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 Smurfs is a 3D live-action/computer-animated comedy film directed by Raja Gosnell and starring a British actress known for playing Emma Pillsbury in Glee.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Smurfs (film)",
"Jayma Mays"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3664,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_3708
|
Kellogg's is the third food manufacturing company to produce cereal.
|
{
"evidence": [
"Kellogg's",
"Battle Creek, Michigan"
],
"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:\nKellogg's is the third food manufacturing company to produce cereal.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Kellogg's",
"Battle Creek, Michigan"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3665,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
65387
|
John Lindsay set the world record for the Men's 200m TW3 at the 1992 Summer Paralympics (the ninth iteration of the event).
|
{
"evidence": [
"1992 Summer Paralympics_sentence_0",
"John Lindsay (Paralympian)_sentence_4"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nJohn Lindsay set the world record for the Men's 200m TW3 at the 1992 Summer Paralympics (the ninth iteration of the event).\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"1992 Summer Paralympics_sentence_0",
"John Lindsay (Paralympian)_sentence_4"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3666,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.869",
"value": "?search="
},
{
"operation": "search",
"time": "12.499",
"value": "John Lindsay"
},
{
"operation": "Now on",
"time": "22.029",
"value": "John Lindsay"
},
{
"operation": "back-button-clicked",
"time": "42.275",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "42.377",
"value": "?search="
},
{
"operation": "search",
"time": "52.517",
"value": "1992 Summer Paralympics"
},
{
"operation": "Now on",
"time": "53.65",
"value": "1992 Summer Paralympics"
},
{
"operation": "Highlighting",
"time": "87.113",
"value": "1992 Summer Paralympics_sentence_0"
},
{
"operation": "search",
"time": "98.031",
"value": "Contains...Men's 200m TW3 at the 1992 Summer Paralympics"
},
{
"operation": "Now on",
"time": "100.414",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "127.174",
"value": "Canada at the 1992 Summer Paralympics"
},
{
"operation": "Now on",
"time": "128.554",
"value": "Canada at the 1992 Summer Paralympics"
},
{
"operation": "search",
"time": "169.902",
"value": "Contains...Men's 200m TW3"
},
{
"operation": "Now on",
"time": "200.137",
"value": "Canada at the 1992 Summer Paralympics"
},
{
"operation": "search",
"time": "215.105",
"value": "Men's 200m TW3"
},
{
"operation": "Now on",
"time": "216.106",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "220.243",
"value": "John Lindsay (Paralympian)"
},
{
"operation": "Now on",
"time": "220.692",
"value": "John Lindsay (Paralympian)"
},
{
"operation": "Highlighting",
"time": "230.647",
"value": "John Lindsay (Paralympian)_sentence_2"
},
{
"operation": "Highlighting deleted",
"time": "232.298",
"value": "John Lindsay (Paralympian)_sentence_2"
},
{
"operation": "Highlighting",
"time": "250.25",
"value": "John Lindsay (Paralympian)_sentence_4"
},
{
"operation": "hyperlink",
"time": "322.687",
"value": "1996 Summer Paralympics"
},
{
"operation": "Now on",
"time": "323.639",
"value": "1996 Summer Paralympics"
},
{
"operation": "back-button-clicked",
"time": "335.442",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "335.554",
"value": "John Lindsay (Paralympian)"
},
{
"operation": "finish",
"time": "371.332",
"value": "finish"
}
]
}
|
53771
|
Sathorn Unique Tower is at Kowloon Peninsula, Hong Kong that sits roughly beside the Wat Yan Nawa and is near the middle of Sathon Road, where Taksin Bridge and the Saphan Taksin Station of the BTS Skytrain are located, formally known as "the Elevated Train in Commemoration of HM the King's 6th Cycle Birthday."
|
{
"evidence": [
"Sathorn Unique Tower_sentence_14",
"Sathorn Unique Tower_sentence_15",
"Sukhumvit line_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:\nSathorn Unique Tower is at Kowloon Peninsula, Hong Kong that sits roughly beside the Wat Yan Nawa and is near the middle of Sathon Road, where Taksin Bridge and the Saphan Taksin Station of the BTS Skytrain are located, formally known as \"the Elevated Train in Commemoration of HM the King's 6th Cycle Birthday.\"\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Sathorn Unique Tower_sentence_14",
"Sathorn Unique Tower_sentence_15",
"Sukhumvit line_sentence_5"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3667,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.988",
"value": "?search="
},
{
"operation": "search",
"time": "9.094",
"value": "Sathorn Unique Tower"
},
{
"operation": "Now on",
"time": "9.787",
"value": "Sathorn Unique Tower"
},
{
"operation": "Highlighting",
"time": "47.403",
"value": "Sathorn Unique Tower_sentence_14"
},
{
"operation": "Highlighting",
"time": "57.091",
"value": "Sathorn Unique Tower_sentence_15"
},
{
"operation": "search",
"time": "84.541",
"value": "Contains...the Elevated Train"
},
{
"operation": "Now on",
"time": "88.406",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "93.891",
"value": "Sukhumvit line"
},
{
"operation": "Now on",
"time": "96.123",
"value": "Sukhumvit line"
},
{
"operation": "Highlighting",
"time": "101.387",
"value": "Sukhumvit line_sentence_5"
},
{
"operation": "finish",
"time": "130.586",
"value": "finish"
}
]
}
|
exfever_6515
|
Mikhail Dmitrievich Prokhorov, the former chairman of Russia's largest gold producer, was also the former president of Onexim Group, a Russian gold mining company.
|
{
"evidence": [
"Mikhail Prokhorov",
"Polyus Gold"
],
"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:\nMikhail Dmitrievich Prokhorov, the former chairman of Russia's largest gold producer, was also the former president of Onexim Group, a Russian gold mining company.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Mikhail Prokhorov",
"Polyus Gold"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3668,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_4966
|
Pineapple Express is a 2008 African stoner comedy film written by Evan Goldberg and a Canadian actor who was originally a stand-up comedian in Vancouver.
|
{
"evidence": [
"Pineapple Express (film)",
"Seth Rogen"
],
"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:\nPineapple Express is a 2008 African stoner comedy film written by Evan Goldberg and a Canadian actor who was originally a stand-up comedian in Vancouver.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Pineapple Express (film)",
"Seth Rogen"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3669,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_3336
|
James Cromwell is an American actor whose best-known films include an American sports drama film produced and released by Walt Disney Pictures.
|
{
"evidence": [
"James Cromwell",
"Secretariat (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:\nJames Cromwell is an American actor whose best-known films include an American sports drama film produced and released by Walt Disney Pictures.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"James Cromwell",
"Secretariat (film)"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3670,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
60398
|
The Kamloops Heritage Railway is a heritage railway in Kamloops, British Columbia that was operated as living history to re-create or preserve railway scenes of the past.
|
{
"evidence": [
"Kamloops Heritage Railway_sentence_0",
"Heritage railway_sentence_2"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Kamloops Heritage Railway is a heritage railway in Kamloops, British Columbia that was operated as living history to re-create or preserve railway scenes of the past.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Kamloops Heritage Railway_sentence_0",
"Heritage railway_sentence_2"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3671,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.824",
"value": "?search="
},
{
"operation": "search",
"time": "16.124",
"value": "Kamloops Heritage Railway"
},
{
"operation": "Now on",
"time": "17.34",
"value": "Kamloops Heritage Railway"
},
{
"operation": "Highlighting",
"time": "23.336",
"value": "Kamloops Heritage Railway_sentence_0"
},
{
"operation": "hyperlink",
"time": "67.683",
"value": "heritage railway"
},
{
"operation": "Now on",
"time": "68.952",
"value": "Heritage railway"
},
{
"operation": "Highlighting",
"time": "72.205",
"value": "Heritage railway_sentence_2"
},
{
"operation": "finish",
"time": "96.292",
"value": "finish"
}
]
}
|
exfever_2823
|
The song "7 Things" was co-written by Miley Cyrus, Tim James, and an American songwriter, record producer, music producer, and businesswoman.
|
{
"evidence": [
"7 Things",
"Antonina Armato"
],
"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 \"7 Things\" was co-written by Miley Cyrus, Tim James, and an American songwriter, record producer, music producer, and businesswoman.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"7 Things",
"Antonina Armato"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3672,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7807
|
Tenitra Michelle Williams rose to fame in the 2000s as a member of the R&B American girl group that included a Canadian singer, performer, songwriter, actress, and television personality.
|
{
"evidence": [
"Michelle Williams (singer)",
"Destiny's Child",
"Kelly Rowland"
],
"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:\nTenitra Michelle Williams rose to fame in the 2000s as a member of the R&B American girl group that included a Canadian singer, performer, songwriter, actress, and television personality.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Michelle Williams (singer)",
"Destiny's Child",
"Kelly Rowland"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3673,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
94092
|
Tournexia is from the Plantae kingdom and a member of the Asteraceae family.
|
{
"evidence": [
"Asteraceae_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:\nTournexia is from the Plantae kingdom and a member of the Asteraceae family.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Asteraceae_sentence_0"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3674,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.015",
"value": "?search="
},
{
"operation": "search",
"time": "17.257",
"value": "Contains...Tournexia"
},
{
"operation": "Now on",
"time": "18.117",
"value": "undefined"
},
{
"operation": "search",
"time": "24.257",
"value": "Contains...Tournexia"
},
{
"operation": "Now on",
"time": "24.726",
"value": "undefined"
},
{
"operation": "search",
"time": "29.68",
"value": "turnexia"
},
{
"operation": "Now on",
"time": "30.81",
"value": "undefined"
},
{
"operation": "search",
"time": "43.553",
"value": "tuneria"
},
{
"operation": "Now on",
"time": "44.203",
"value": "undefined"
},
{
"operation": "back-button-clicked",
"time": "48.527",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "48.643",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "52.078",
"value": "turneri"
},
{
"operation": "Now on",
"time": "52.764",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "55.747",
"value": "Chrysoritis"
},
{
"operation": "Now on",
"time": "56.617",
"value": "Chrysoritis"
},
{
"operation": "Page search",
"time": "68.772",
"value": "to"
},
{
"operation": "Page search",
"time": "78.234",
"value": "tornuri"
},
{
"operation": "Page search",
"time": "80.966",
"value": "uri"
},
{
"operation": "hyperlink",
"time": "102.763",
"value": "Chrysoritis turneri"
},
{
"operation": "Now on",
"time": "103.766",
"value": "Chrysoritis turneri"
},
{
"operation": "search",
"time": "123.55",
"value": "Asteraceae"
},
{
"operation": "Now on",
"time": "124.851",
"value": "Asteraceae"
},
{
"operation": "Page search",
"time": "138.865",
"value": "turn"
},
{
"operation": "Page search",
"time": "148",
"value": "exia"
},
{
"operation": "Highlighting",
"time": "161.997",
"value": "Asteraceae_sentence_0"
},
{
"operation": "finish",
"time": "165.81",
"value": "finish"
}
]
}
|
exfever_7280
|
Lymelife is a 2009 American comedy-drama film produced by an American film director, producer, and screenwriter who is known for directing the film Who's That Knocking at My Door.
|
{
"evidence": [
"Lymelife",
"Martin Scorsese"
],
"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:\nLymelife is a 2009 American comedy-drama film produced by an American film director, producer, and screenwriter who is known for directing the film Who's That Knocking at My Door.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Lymelife",
"Martin Scorsese"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3675,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_4935
|
Joseph Robinette Biden Jr. is a politician who served as the 47th vice president from 2009 to 2017 under Barack Obama and represented a state in the Mid-Atlantic region of the United States in the United States Senate from 1973 to 2009.
|
{
"evidence": [
"Joe Biden",
"Delaware"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nJoseph Robinette Biden Jr. is a politician who served as the 47th vice president from 2009 to 2017 under Barack Obama and represented a state in the Mid-Atlantic region of the United States in the United States Senate from 1973 to 2009.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Joe Biden",
"Delaware"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3676,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_36
|
Sam Claflin won the Man of the Year from Glamour Award.
|
{
"evidence": [
"Sam Claflin",
"London Academy of Music and Dramatic Art"
],
"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:\nSam Claflin won the Man of the Year from Glamour Award.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Sam Claflin",
"London Academy of Music and Dramatic Art"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3677,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_704
|
Margaret Constance "Maisie" Williams is an English actress who made her acting debut as a lead character in Game of Thrones which was aired on an American pay television network, which is the flagship property of namesake parent-subsidiary Home Box Office, Inc.
|
{
"evidence": [
"Maisie Williams",
"HBO"
],
"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:\nMargaret Constance \"Maisie\" Williams is an English actress who made her acting debut as a lead character in Game of Thrones which was aired on an American pay television network, which is the flagship property of namesake parent-subsidiary Home Box Office, Inc.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Maisie Williams",
"HBO"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3678,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
44636
|
James Chilton (a passenger on the Mayflower–a ship that transported passengers to the New World in 1620) had three daughters who lived to maturity: Mary, Ingel, and Isabella (who was baptized in 1586/87).
|
{
"evidence": [
"James Chilton_sentence_24",
"James Chilton_sentence_25",
"James Chilton_sentence_28",
"James Chilton_sentence_11"
],
"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:\nJames Chilton (a passenger on the Mayflower–a ship that transported passengers to the New World in 1620) had three daughters who lived to maturity: Mary, Ingel, and Isabella (who was baptized in 1586/87).\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"James Chilton_sentence_24",
"James Chilton_sentence_25",
"James Chilton_sentence_28",
"James Chilton_sentence_11"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 3679,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.268",
"value": "?search="
},
{
"operation": "search",
"time": "10.022",
"value": "James Chilton"
},
{
"operation": "Now on",
"time": "10.954",
"value": "James Chilton"
},
{
"operation": "Highlighting",
"time": "63.934",
"value": "James Chilton_sentence_0"
},
{
"operation": "hyperlink",
"time": "69.047",
"value": "Mayflower"
},
{
"operation": "Now on",
"time": "69.935",
"value": "Mayflower"
},
{
"operation": "Highlighting",
"time": "72.248",
"value": "Mayflower_sentence_2"
},
{
"operation": "Highlighting",
"time": "72.249",
"value": "Mayflower_title"
},
{
"operation": "back-button-clicked",
"time": "78.412",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "78.479",
"value": "James Chilton"
},
{
"operation": "hyperlink",
"time": "120.265",
"value": "freeman"
},
{
"operation": "Now on",
"time": "120.982",
"value": "Freeman (Colonial)"
},
{
"operation": "back-button-clicked",
"time": "134.446",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "134.514",
"value": "James Chilton"
},
{
"operation": "Highlighting",
"time": "188.946",
"value": "James Chilton_sentence_24"
},
{
"operation": "Highlighting",
"time": "189.79",
"value": "James Chilton_sentence_25"
},
{
"operation": "Highlighting",
"time": "207.757",
"value": "James Chilton_sentence_28"
},
{
"operation": "Highlighting deleted",
"time": "212.571",
"value": "James Chilton_sentence_0"
},
{
"operation": "Highlighting",
"time": "214.135",
"value": "James Chilton_sentence_0"
},
{
"operation": "Highlighting deleted",
"time": "216.085",
"value": "James Chilton_sentence_0"
},
{
"operation": "hyperlink",
"time": "216.733",
"value": "Mayflower"
},
{
"operation": "Now on",
"time": "217.915",
"value": "Mayflower"
},
{
"operation": "Highlighting deleted",
"time": "224.289",
"value": "Mayflower_sentence_2"
},
{
"operation": "Highlighting deleted",
"time": "224.291",
"value": "Mayflower_title"
},
{
"operation": "back-button-clicked",
"time": "226.358",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "226.432",
"value": "James Chilton"
},
{
"operation": "Highlighting",
"time": "272.667",
"value": "James Chilton_sentence_11"
},
{
"operation": "finish",
"time": "284.692",
"value": "finish"
}
]
}
|
exfever_7863
|
Angelina Jolie is an American actress who has received numerous accolades, including an award for artistic and technical merit in the film industry.
|
{
"evidence": [
"Angelina Jolie",
"Academy Awards"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nAngelina Jolie is an American actress who has received numerous accolades, including an award for artistic and technical merit in the film industry.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Angelina Jolie",
"Academy Awards"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 3680,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_3884
|
As You Like It is a 2007 romance film that stars an English actress who appeared in the film A Doll's House, which earned her Drama Desk Award for Outstanding Actress.
|
{
"evidence": [
"As You Like It (2006 film)",
"Janet McTeer"
],
"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 You Like It is a 2007 romance film that stars an English actress who appeared in the film A Doll's House, which earned her Drama Desk Award for Outstanding Actress.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"As You Like It (2006 film)",
"Janet McTeer"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3681,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
8332
|
The Soundscans of Ministry's discography and the latest by NIN, Rammstein and Marilyn Manson were taken this year; in contrast, the Soundscan sale numbers from Broken are over 15 years old.
|
{
"evidence": [
"Industrial rock sales and awards_sentence_34"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Soundscans of Ministry's discography and the latest by NIN, Rammstein and Marilyn Manson were taken this year; in contrast, the Soundscan sale numbers from Broken are over 15 years old.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Industrial rock sales and awards_sentence_34"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3682,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.017",
"value": "?search="
},
{
"operation": "search",
"time": "25.933",
"value": "Contains...Soundscans of Ministry's discography"
},
{
"operation": "Now on",
"time": "28.574",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "38.389",
"value": "Industrial rock sales and awards"
},
{
"operation": "Now on",
"time": "39.782",
"value": "Industrial rock sales and awards"
},
{
"operation": "Page search",
"time": "46.577",
"value": "Soundscans of Ministry's discography"
},
{
"operation": "Highlighting",
"time": "59.441",
"value": "Industrial rock sales and awards_sentence_33"
},
{
"operation": "Highlighting",
"time": "62.805",
"value": "Industrial rock sales and awards_sentence_34"
},
{
"operation": "Highlighting deleted",
"time": "79.249",
"value": "Industrial rock sales and awards_sentence_33"
},
{
"operation": "finish",
"time": "127.83",
"value": "finish"
}
]
}
|
exfever_3321
|
Boston Legal is an American legal action-drama television series created by David E. Kelley and starring a Canadian actor best known for his role as Captain of the USS Enterprise in the Star Trek franchise, who leads his crew as they explore new worlds, new civilizations, and boldly go where no man has gone before.
|
{
"evidence": [
"Boston Legal",
"William Shatner",
"James T. Kirk"
],
"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:\nBoston Legal is an American legal action-drama television series created by David E. Kelley and starring a Canadian actor best known for his role as Captain of the USS Enterprise in the Star Trek franchise, who leads his crew as they explore new worlds, new civilizations, and boldly go where no man has gone before.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Boston Legal",
"William Shatner",
"James T. Kirk"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 3683,
"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.