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_8140
|
Sylvester Stallone's commercial success started as an actor with his role as a fictional character and the protagonist of the 1976 Rocky film series.
|
{
"evidence": [
"Sylvester Stallone",
"Rocky Balboa"
],
"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:\nSylvester Stallone's commercial success started as an actor with his role as a fictional character and the protagonist of the 1976 Rocky film series.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Sylvester Stallone",
"Rocky Balboa"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8084,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
36091
|
Each episode on That Sunday Night Show was 30 minutes long, Chiles was joined by three guests and had high ratings.
|
{
"evidence": [
"That Sunday Night Show_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:\nEach episode on That Sunday Night Show was 30 minutes long, Chiles was joined by three guests and had high ratings.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"That Sunday Night Show_sentence_4"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 8085,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.645",
"value": "undefined"
},
{
"operation": "search",
"time": "19.5",
"value": "That Sunday Night Show"
},
{
"operation": "Now on",
"time": "20.412",
"value": "That Sunday Night Show"
},
{
"operation": "Highlighting",
"time": "42.384",
"value": "That Sunday Night Show_sentence_1"
},
{
"operation": "Page search",
"time": "46.198",
"value": "Chiles"
},
{
"operation": "Page search",
"time": "46.369",
"value": "Chiles"
},
{
"operation": "Page search",
"time": "46.54",
"value": "Chiles"
},
{
"operation": "Page search",
"time": "46.808",
"value": "Chiles"
},
{
"operation": "Page search",
"time": "47.239",
"value": "Chiles"
},
{
"operation": "Page search",
"time": "47.418",
"value": "Chile"
},
{
"operation": "Page search",
"time": "47.598",
"value": "Chil"
},
{
"operation": "Page search",
"time": "47.82",
"value": "Chil"
},
{
"operation": "Highlighting",
"time": "51.071",
"value": "That Sunday Night Show_sentence_2"
},
{
"operation": "Page search",
"time": "57.556",
"value": "high ratings"
},
{
"operation": "Page search",
"time": "57.744",
"value": "high ratings"
},
{
"operation": "Page search",
"time": "57.921",
"value": "high ratings"
},
{
"operation": "Page search",
"time": "58.153",
"value": "high rating"
},
{
"operation": "Page search",
"time": "58.348",
"value": "high ratin"
},
{
"operation": "Page search",
"time": "58.552",
"value": "high rati"
},
{
"operation": "Page search",
"time": "58.749",
"value": "high rat"
},
{
"operation": "Page search",
"time": "58.943",
"value": "high ra"
},
{
"operation": "Page search",
"time": "59.137",
"value": "high r"
},
{
"operation": "Page search",
"time": "59.342",
"value": "high"
},
{
"operation": "Page search",
"time": "59.541",
"value": "high"
},
{
"operation": "Page search",
"time": "59.743",
"value": "hig"
},
{
"operation": "Page search",
"time": "59.948",
"value": "hi"
},
{
"operation": "Page search",
"time": "60.176",
"value": "h"
},
{
"operation": "page-search-reset",
"time": "60.351",
"value": "page-search-reset"
},
{
"operation": "page-search-reset",
"time": "60.533",
"value": "page-search-reset"
},
{
"operation": "Page search",
"time": "69.77",
"value": "rat"
},
{
"operation": "Highlighting",
"time": "73.197",
"value": "That Sunday Night Show_sentence_4"
},
{
"operation": "Highlighting deleted",
"time": "74.098",
"value": "That Sunday Night Show_sentence_2"
},
{
"operation": "Highlighting deleted",
"time": "74.402",
"value": "That Sunday Night Show_sentence_1"
},
{
"operation": "finish",
"time": "81.83",
"value": "finish"
}
]
}
|
86230
|
Werder Bremen made 10 appearances in the 2006 season with Palmeiras, a Brazilian professional football club based in the city of São Paulo, in the district of Perdizes.
|
{
"evidence": [
"SV Werder Bremen_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:\nWerder Bremen made 10 appearances in the 2006 season with Palmeiras, a Brazilian professional football club based in the city of São Paulo, in the district of Perdizes.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"SV Werder Bremen_sentence_1"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 8086,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.1",
"value": "?search="
},
{
"operation": "search",
"time": "19.363",
"value": "Werder Bremen"
},
{
"operation": "Now on",
"time": "22.309",
"value": "SV Werder Bremen"
},
{
"operation": "search",
"time": "44.932",
"value": "Werder Bremen II"
},
{
"operation": "Now on",
"time": "46.364",
"value": "SV Werder Bremen II"
},
{
"operation": "search",
"time": "55.3",
"value": "Werder Bremen"
},
{
"operation": "Now on",
"time": "56.357",
"value": "Werder Bremen"
},
{
"operation": "Highlighting",
"time": "65.024",
"value": "SV Werder Bremen_sentence_1"
},
{
"operation": "search",
"time": "87.729",
"value": "FC palmeiras"
},
{
"operation": "Now on",
"time": "89.622",
"value": "undefined"
},
{
"operation": "search",
"time": "100.355",
"value": "palmeiras FC"
},
{
"operation": "Now on",
"time": "101.3",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "106.926",
"value": "C.R.D. Libolo"
},
{
"operation": "Now on",
"time": "108.444",
"value": "C.R.D. Libolo"
},
{
"operation": "search",
"time": "119.135",
"value": "bremen"
},
{
"operation": "Now on",
"time": "123.369",
"value": "Bremen"
},
{
"operation": "search",
"time": "140.859",
"value": "palmeiras"
},
{
"operation": "Now on",
"time": "146.012",
"value": "Sociedade Esportiva Palmeiras"
},
{
"operation": "Highlighting",
"time": "152.834",
"value": "Sociedade Esportiva Palmeiras_sentence_2"
},
{
"operation": "Highlighting",
"time": "152.836",
"value": "Sociedade Esportiva Palmeiras_title"
},
{
"operation": "hyperlink",
"time": "181.516",
"value": "5 Players Sociedade Esportiva Palmeiras_section_35"
},
{
"operation": "Page search",
"time": "197.362",
"value": "2006"
},
{
"operation": "Highlighting",
"time": "206.046",
"value": "Sociedade Esportiva Palmeiras_sentence_54"
},
{
"operation": "Highlighting",
"time": "206.053",
"value": "Sociedade Esportiva Palmeiras_section_0"
},
{
"operation": "Highlighting",
"time": "206.055",
"value": "Sociedade Esportiva Palmeiras_section_4"
},
{
"operation": "Highlighting",
"time": "222.189",
"value": "Sociedade Esportiva Palmeiras_sentence_135"
},
{
"operation": "Highlighting",
"time": "222.198",
"value": "Sociedade Esportiva Palmeiras_section_10"
},
{
"operation": "search",
"time": "238.833",
"value": "werder"
},
{
"operation": "Now on",
"time": "241.26",
"value": "Werder"
},
{
"operation": "hyperlink",
"time": "243.659",
"value": "1 People Werder_section_0"
},
{
"operation": "Highlighting",
"time": "248.707",
"value": "Werder_item_0_0"
},
{
"operation": "Highlighting",
"time": "248.709",
"value": "Werder_title"
},
{
"operation": "Highlighting",
"time": "248.71",
"value": "Werder_section_0"
},
{
"operation": "Highlighting",
"time": "249.586",
"value": "Werder_item_0_1"
},
{
"operation": "Highlighting",
"time": "250.479",
"value": "Werder_item_0_2"
},
{
"operation": "Highlighting",
"time": "251.445",
"value": "Werder_item_0_3"
},
{
"operation": "Highlighting deleted",
"time": "259.842",
"value": "Werder_item_0_0"
},
{
"operation": "Highlighting deleted",
"time": "260.464",
"value": "Werder_item_0_1"
},
{
"operation": "Highlighting deleted",
"time": "261.065",
"value": "Werder_item_0_2"
},
{
"operation": "Highlighting deleted",
"time": "261.672",
"value": "Werder_item_0_3"
},
{
"operation": "Highlighting deleted",
"time": "261.676",
"value": "Werder_title"
},
{
"operation": "Highlighting deleted",
"time": "261.68",
"value": "Werder_section_0"
},
{
"operation": "search",
"time": "265.754",
"value": "bremen"
},
{
"operation": "Now on",
"time": "268.348",
"value": "Bremen"
},
{
"operation": "search",
"time": "309.554",
"value": "palmeiras FC"
},
{
"operation": "Now on",
"time": "311.207",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "315.34",
"value": "C.R.D. Libolo"
},
{
"operation": "Now on",
"time": "316.328",
"value": "C.R.D. Libolo"
},
{
"operation": "search",
"time": "325.426",
"value": "FC palmeiras"
},
{
"operation": "Now on",
"time": "326.374",
"value": "undefined"
},
{
"operation": "finish",
"time": "341.075",
"value": "finish"
}
]
}
|
exfever_1054
|
Stephen Rider is an American actor known for playing Admiral Stephen W. Rochon in the 2013 historical drama film directed and co-produced by Lee Daniels.
|
{
"evidence": [
"Stephen Rider",
"The Butler"
],
"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:\nStephen Rider is an American actor known for playing Admiral Stephen W. Rochon in the 2013 historical drama film directed and co-produced by Lee Daniels.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Stephen Rider",
"The Butler"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8087,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
25390
|
A fire plough are three wires rubbed together that eventually makes ember due to the friction and heat.
|
{
"evidence": [
"Fire plough_sentence_1",
"Fire plough_sentence_3"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nA fire plough are three wires rubbed together that eventually makes ember due to the friction and heat.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Fire plough_sentence_1",
"Fire plough_sentence_3"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 8088,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "6.325",
"value": "?search="
},
{
"operation": "search",
"time": "16.389",
"value": "Fire plough"
},
{
"operation": "Now on",
"time": "17.956",
"value": "Fire plough"
},
{
"operation": "Highlighting",
"time": "45.114",
"value": "Fire plough_sentence_1"
},
{
"operation": "Highlighting",
"time": "46.877",
"value": "Fire plough_sentence_2"
},
{
"operation": "Highlighting deleted",
"time": "52.132",
"value": "Fire plough_sentence_2"
},
{
"operation": "Highlighting",
"time": "54.4",
"value": "Fire plough_sentence_3"
},
{
"operation": "finish",
"time": "85.789",
"value": "finish"
}
]
}
|
exfever_3831
|
The Super Nintendo Entertainment System is a 16-bit game console that is designed to be connected to a display device, developed by Nintendo that was released in 1990 in Japan, which have been numerous game consoles since the first game console that was designed by a small team led by Ralph H. Baer, in 1972.
|
{
"evidence": [
"Super Nintendo Entertainment System",
"Home video game console",
"Magnavox Odyssey"
],
"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 Super Nintendo Entertainment System is a 16-bit game console that is designed to be connected to a display device, developed by Nintendo that was released in 1990 in Japan, which have been numerous game consoles since the first game console that was designed by a small team led by Ralph H. Baer, in 1972.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Super Nintendo Entertainment System",
"Home video game console",
"Magnavox Odyssey"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8089,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7343
|
Just Dance" is the debut single by an American singer, songwriter, and actress who released the debut studio album on August 19, 2008.
|
{
"evidence": [
"Just Dance (song)",
"Lady Gaga",
"The Fame"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nJust Dance\" is the debut single by an American singer, songwriter, and actress who released the debut studio album on August 19, 2008.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Just Dance (song)",
"Lady Gaga",
"The Fame"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8090,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_4390
|
Led Zeppelin II's production was credited to the band's lead guitarist, an English musician with a varied style that involves various alternative melodies on guitar, which are the assignments of pitches.
|
{
"evidence": [
"Led Zeppelin II",
"Jimmy Page",
"Guitar tunings"
],
"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:\nLed Zeppelin II's production was credited to the band's lead guitarist, an English musician with a varied style that involves various alternative melodies on guitar, which are the assignments of pitches.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Led Zeppelin II",
"Jimmy Page",
"Guitar tunings"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8091,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_6893
|
Bryan Cranston is a Canadian actor known for portraying the main character of the American crime drama television series Breaking Bad.
|
{
"evidence": [
"Bryan Cranston",
"Walter White (Breaking Bad)"
],
"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:\nBryan Cranston is a Canadian actor known for portraying the main character of the American crime drama television series Breaking Bad.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Bryan Cranston",
"Walter White (Breaking Bad)"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 8092,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
68697
|
Bell 707 role is Multipurpose utility helicopter and the manufacturer is Bell Helicopter which cost US$6.1M million.
|
{
"evidence": [
"Bell Textron_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:\nBell 707 role is Multipurpose utility helicopter and the manufacturer is Bell Helicopter which cost US$6.1M million.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Bell Textron_sentence_0"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 8093,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.886",
"value": "?search="
},
{
"operation": "search",
"time": "15.859",
"value": "bell 707"
},
{
"operation": "Now on",
"time": "17.193",
"value": "undefined"
},
{
"operation": "search",
"time": "45.003",
"value": "Bell Helicopters"
},
{
"operation": "Now on",
"time": "46.209",
"value": "Bell Textron"
},
{
"operation": "search",
"time": "149.729",
"value": "manufacturer is Bell Helicopter which cost US$6.1M million"
},
{
"operation": "Now on",
"time": "152.46",
"value": "undefined"
},
{
"operation": "search",
"time": "188.702",
"value": "bell 707 utility helicopter"
},
{
"operation": "Now on",
"time": "190.954",
"value": "undefined"
},
{
"operation": "search",
"time": "267.097",
"value": "Bell Helicopters"
},
{
"operation": "Now on",
"time": "268.171",
"value": "Bell Helicopters"
},
{
"operation": "Highlighting",
"time": "307.138",
"value": "Bell Textron_sentence_0"
},
{
"operation": "finish",
"time": "312.492",
"value": "finish"
}
]
}
|
50093
|
Gerd Janne Kristoffersen was elected to the Norwegian Parliament and she had previously served as a deputy representative.
|
{
"evidence": [
"Gerd Janne Kristoffersen_sentence_1",
"Gerd Janne Kristoffersen_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:\nGerd Janne Kristoffersen was elected to the Norwegian Parliament and she had previously served as a deputy representative.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Gerd Janne Kristoffersen_sentence_1",
"Gerd Janne Kristoffersen_sentence_2"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8094,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.001",
"value": "?search="
},
{
"operation": "search",
"time": "77.721",
"value": "Gerd Janne Kristoffersen"
},
{
"operation": "Now on",
"time": "78.482",
"value": "Gerd Janne Kristoffersen"
},
{
"operation": "Highlighting",
"time": "100.863",
"value": "Gerd Janne Kristoffersen_sentence_1"
},
{
"operation": "Highlighting",
"time": "103.559",
"value": "Gerd Janne Kristoffersen_sentence_2"
},
{
"operation": "finish",
"time": "109.222",
"value": "finish"
}
]
}
|
exfever_314
|
The Taj Mahal was commissioned by the Mughal emperor Shah Jahan to house the tomb of his favourite wife, which is the centrepiece of the complex that includes a place of worship for Christians.
|
{
"evidence": [
"Taj Mahal",
"Mosque"
],
"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 Taj Mahal was commissioned by the Mughal emperor Shah Jahan to house the tomb of his favourite wife, which is the centrepiece of the complex that includes a place of worship for Christians.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Taj Mahal",
"Mosque"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 8095,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_634
|
Spider-Man is a 2002 American superhero film based on the character of the same name by a publisher of American comic books and related media.
|
{
"evidence": [
"Spider-Man (2002 film)",
"Marvel Comics"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nSpider-Man is a 2002 American superhero film based on the character of the same name by a publisher of American comic books and related media.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Spider-Man (2002 film)",
"Marvel Comics"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8096,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
64137
|
In the 2001 Wimbledon Championships, Men's Singles won a total of £455,000 and the Women's Singles won a total of £409,500.
|
{
"evidence": [
"2001 Wimbledon Championships_sentence_11"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nIn the 2001 Wimbledon Championships, Men's Singles won a total of £455,000 and the Women's Singles won a total of £409,500.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"2001 Wimbledon Championships_sentence_11"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 8097,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "2.451",
"value": "?search="
},
{
"operation": "search",
"time": "10.732",
"value": "2001 Wimbledon Championships"
},
{
"operation": "Now on",
"time": "13.235",
"value": "2001 Wimbledon Championships"
},
{
"operation": "Highlighting",
"time": "99.267",
"value": "2001 Wimbledon Championships_cell_0_1_0"
},
{
"operation": "Highlighting",
"time": "99.862",
"value": "2001 Wimbledon Championships_cell_0_1_1"
},
{
"operation": "Highlighting deleted",
"time": "101.549",
"value": "2001 Wimbledon Championships_cell_0_1_0"
},
{
"operation": "Highlighting",
"time": "102.048",
"value": "2001 Wimbledon Championships_cell_0_0_0"
},
{
"operation": "Highlighting",
"time": "103.311",
"value": "2001 Wimbledon Championships_cell_0_0_1"
},
{
"operation": "Highlighting deleted",
"time": "104.431",
"value": "2001 Wimbledon Championships_cell_0_1_1"
},
{
"operation": "Highlighting",
"time": "105.44",
"value": "2001 Wimbledon Championships_cell_0_1_0"
},
{
"operation": "Highlighting",
"time": "105.932",
"value": "2001 Wimbledon Championships_cell_0_1_1"
},
{
"operation": "Highlighting",
"time": "114.095",
"value": "2001 Wimbledon Championships_cell_0_2_0"
},
{
"operation": "Highlighting",
"time": "114.57",
"value": "2001 Wimbledon Championships_cell_0_2_1"
},
{
"operation": "Highlighting deleted",
"time": "121.678",
"value": "2001 Wimbledon Championships_cell_0_2_1"
},
{
"operation": "Highlighting deleted",
"time": "122.235",
"value": "2001 Wimbledon Championships_cell_0_2_0"
},
{
"operation": "Highlighting deleted",
"time": "123.111",
"value": "2001 Wimbledon Championships_cell_0_1_0"
},
{
"operation": "Highlighting deleted",
"time": "123.64",
"value": "2001 Wimbledon Championships_cell_0_1_1"
},
{
"operation": "Highlighting deleted",
"time": "123.992",
"value": "2001 Wimbledon Championships_cell_0_0_1"
},
{
"operation": "Highlighting deleted",
"time": "124.42",
"value": "2001 Wimbledon Championships_cell_0_0_0"
},
{
"operation": "Highlighting",
"time": "126.17",
"value": "2001 Wimbledon Championships_sentence_11"
},
{
"operation": "finish",
"time": "138.87",
"value": "finish"
}
]
}
|
exfever_450
|
Match Point is a psychological thriller film starring an English actress known for playing Homily in a BBC TV adaptation first broadcast on BBC2.
|
{
"evidence": [
"Match Point",
"Penelope Wilton",
"The Borrowers (miniseries)"
],
"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:\nMatch Point is a psychological thriller film starring an English actress known for playing Homily in a BBC TV adaptation first broadcast on BBC2.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Match Point",
"Penelope Wilton",
"The Borrowers (miniseries)"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8098,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
57927
|
Garrison, a Canadian politician born on August 27, 1951, advocated for stricter targets of greenhouse gas emissions reduction, and lobbied to get the municipality to adopt a living wage policy.
|
{
"evidence": [
"Randall Garrison_sentence_0",
"Randall Garrison_sentence_34",
"Randall Garrison_sentence_35"
],
"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:\nGarrison, a Canadian politician born on August 27, 1951, advocated for stricter targets of greenhouse gas emissions reduction, and lobbied to get the municipality to adopt a living wage policy.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"Randall Garrison_sentence_0",
"Randall Garrison_sentence_34",
"Randall Garrison_sentence_35"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8099,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.916",
"value": "?search="
},
{
"operation": "search",
"time": "18.262",
"value": "Contains...Garrison, a Canadian politician"
},
{
"operation": "Now on",
"time": "20.287",
"value": "undefined"
},
{
"operation": "search",
"time": "50.406",
"value": "\"Garrison, a Canadian politician\""
},
{
"operation": "Now on",
"time": "51.23",
"value": "undefined"
},
{
"operation": "search",
"time": "61.636",
"value": "Garrison canadian"
},
{
"operation": "Now on",
"time": "62.684",
"value": "undefined"
},
{
"operation": "search",
"time": "93.029",
"value": "Garrison born August 27, 1951"
},
{
"operation": "Now on",
"time": "94.997",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "105.013",
"value": "Randall Garrison"
},
{
"operation": "Now on",
"time": "106.208",
"value": "Randall Garrison"
},
{
"operation": "Highlighting",
"time": "116.096",
"value": "Randall Garrison_sentence_0"
},
{
"operation": "Highlighting",
"time": "143.811",
"value": "Randall Garrison_sentence_34"
},
{
"operation": "Highlighting",
"time": "157.807",
"value": "Randall Garrison_sentence_35"
},
{
"operation": "finish",
"time": "194.769",
"value": "finish"
}
]
}
|
exfever_899
|
Roman Reigns was the second-highest-paid American wrestler in 2014.
|
{
"evidence": [
"Roman Reigns",
"WWE"
],
"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:\nRoman Reigns was the second-highest-paid American wrestler in 2014.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Roman Reigns",
"WWE"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 8100,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
37346
|
Mack Furlong has worked as a host of CBC Radio program, a co-creator of The Great Eastern radio show, which aired from 1994 to 1999,, and a voice actor for minor characters in The Great Eastern radio show.
|
{
"evidence": [
"Mack Furlong_sentence_2",
"Mack Furlong_sentence_3",
"Mack Furlong_sentence_4"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nMack Furlong has worked as a host of CBC Radio program, a co-creator of The Great Eastern radio show, which aired from 1994 to 1999,, and a voice actor for minor characters in The Great Eastern radio show.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Mack Furlong_sentence_2",
"Mack Furlong_sentence_3",
"Mack Furlong_sentence_4"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8101,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.198",
"value": "?search="
},
{
"operation": "search",
"time": "14.752",
"value": "Mack Furlong"
},
{
"operation": "Now on",
"time": "16.056",
"value": "Mack Furlong"
},
{
"operation": "Highlighting",
"time": "23.444",
"value": "Mack Furlong_sentence_2"
},
{
"operation": "Highlighting",
"time": "39.015",
"value": "Mack Furlong_sentence_3"
},
{
"operation": "Highlighting",
"time": "79.013",
"value": "Mack Furlong_sentence_4"
},
{
"operation": "finish",
"time": "112.198",
"value": "finish"
}
]
}
|
exfever_3199
|
Alia Bhatt is the daughter of an Indian film director, producer, and screenwriter known for his works in Hindi cinema.
|
{
"evidence": [
"Alia Bhatt",
"Mahesh Bhatt"
],
"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:\nAlia Bhatt is the daughter of an Indian film director, producer, and screenwriter known for his works in Hindi cinema.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Alia Bhatt",
"Mahesh Bhatt"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8102,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
82613
|
Geghuni Chitchian is an Armenian composer and singer who became an Honoured Representative of the Arts in 1980.
|
{
"evidence": [
"Geghuni Chitchian_sentence_0",
"Geghuni Chitchian_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:\nGeghuni Chitchian is an Armenian composer and singer who became an Honoured Representative of the Arts in 1980.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Geghuni Chitchian_sentence_0",
"Geghuni Chitchian_sentence_5"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 8103,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.652",
"value": "?search="
},
{
"operation": "search",
"time": "13.226",
"value": "Geghuni Chitchian"
},
{
"operation": "Now on",
"time": "13.835",
"value": "Geghuni Chitchian"
},
{
"operation": "Highlighting",
"time": "18.876",
"value": "Geghuni Chitchian_sentence_0"
},
{
"operation": "Page search",
"time": "30.184",
"value": "sing"
},
{
"operation": "Highlighting",
"time": "70.281",
"value": "Geghuni Chitchian_sentence_5"
},
{
"operation": "finish",
"time": "234.129",
"value": "finish"
}
]
}
|
exfever_5948
|
The Angkorian Empire is the term that historians use to refer to Cambodia from the 9th to the 15th century uses an Austroasiatic language the official and national language of Cambodia.
|
{
"evidence": [
"Khmer Empire",
"Khmer language"
],
"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 Angkorian Empire is the term that historians use to refer to Cambodia from the 9th to the 15th century uses an Austroasiatic language the official and national language of Cambodia.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Khmer Empire",
"Khmer language"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8104,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
90292
|
Hopeless Fountain Kingdom is the album by American singer-songwriter Hasley that was released by June 2, 2017.
|
{
"evidence": [
"Hopeless Fountain Kingdom_sentence_0",
"Hopeless Fountain Kingdom_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:\nHopeless Fountain Kingdom is the album by American singer-songwriter Hasley that was released by June 2, 2017.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Hopeless Fountain Kingdom_sentence_0",
"Hopeless Fountain Kingdom_sentence_1"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8105,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.568",
"value": "?search="
},
{
"operation": "search",
"time": "15.048",
"value": "Hopeless Fountain Kingdom"
},
{
"operation": "Now on",
"time": "16.016",
"value": "Hopeless Fountain Kingdom"
},
{
"operation": "Highlighting",
"time": "19.708",
"value": "Hopeless Fountain Kingdom_sentence_0"
},
{
"operation": "Highlighting",
"time": "23.4",
"value": "Hopeless Fountain Kingdom_sentence_1"
},
{
"operation": "finish",
"time": "30.649",
"value": "finish"
}
]
}
|
86369
|
John Papaloizou not only discovered the Papaloizou-Pringle instability in 1984, but also made major contributions in various areas; such as the radial-orbit instability, toroidal modes in stars and different instabilities in accretion disks.
|
{
"evidence": [
"John Papaloizou_sentence_6",
"John Papaloizou_sentence_7"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nJohn Papaloizou not only discovered the Papaloizou-Pringle instability in 1984, but also made major contributions in various areas; such as the radial-orbit instability, toroidal modes in stars and different instabilities in accretion disks.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"John Papaloizou_sentence_6",
"John Papaloizou_sentence_7"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8106,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.862",
"value": "?search="
},
{
"operation": "search",
"time": "6.785",
"value": "John Papaloizou"
},
{
"operation": "Now on",
"time": "7.347",
"value": "John Papaloizou"
},
{
"operation": "Highlighting",
"time": "13.432",
"value": "John Papaloizou_sentence_6"
},
{
"operation": "Highlighting",
"time": "29.915",
"value": "John Papaloizou_sentence_7"
},
{
"operation": "finish",
"time": "51.985",
"value": "finish"
}
]
}
|
exfever_4654
|
Arrow is a television series and a costumed crime-fighter created by a European comics artist who also co-created the Green Arrow character with a comic book editor best known for editing DC Comics' Superman during the mid-1950s to 1960s.
|
{
"evidence": [
"Arrow (TV series)",
"George Papp",
"Mort Weisinger"
],
"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:\nArrow is a television series and a costumed crime-fighter created by a European comics artist who also co-created the Green Arrow character with a comic book editor best known for editing DC Comics' Superman during the mid-1950s to 1960s.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Arrow (TV series)",
"George Papp",
"Mort Weisinger"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 8107,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_6690
|
John Lennon's songwriting partnership with an Asian singer known for his melodic approach to bass-playing remains the most successful in history.
|
{
"evidence": [
"John Lennon",
"Paul McCartney"
],
"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:\nJohn Lennon's songwriting partnership with an Asian singer known for his melodic approach to bass-playing remains the most successful in history.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"John Lennon",
"Paul McCartney"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 8108,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
47957
|
In the Ninth Dynasty of Egypt according to Haynes, Mreyibre Khety I was of no importance in the dynasty.
|
{
"evidence": [
"Meryibre Khety_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:\nIn the Ninth Dynasty of Egypt according to Haynes, Mreyibre Khety I was of no importance in the dynasty.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Meryibre Khety_sentence_4"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 8109,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.917",
"value": "?search="
},
{
"operation": "search",
"time": "18.521",
"value": "Ninth Dynasty of Egypt"
},
{
"operation": "Now on",
"time": "19.211",
"value": "Ninth Dynasty of Egypt"
},
{
"operation": "Page search",
"time": "33.608",
"value": "haynes"
},
{
"operation": "Page search",
"time": "43.471",
"value": "haynes"
},
{
"operation": "search",
"time": "60.056",
"value": "Mreyibre Kety"
},
{
"operation": "Now on",
"time": "60.782",
"value": "undefined"
},
{
"operation": "search",
"time": "81.919",
"value": "Khety I"
},
{
"operation": "Now on",
"time": "83.018",
"value": "Khety I"
},
{
"operation": "Page search",
"time": "96.359",
"value": "haynes"
},
{
"operation": "search",
"time": "151.79",
"value": "Haynes Ninth Dynasty"
},
{
"operation": "Now on",
"time": "152.686",
"value": "undefined"
},
{
"operation": "search",
"time": "161.997",
"value": "Haynes Khety"
},
{
"operation": "Now on",
"time": "162.309",
"value": "undefined"
},
{
"operation": "search",
"time": "173.262",
"value": "Khety I (nomarch)"
},
{
"operation": "Now on",
"time": "173.902",
"value": "Khety I (nomarch)"
},
{
"operation": "Page search",
"time": "189.013",
"value": "Haynes"
},
{
"operation": "search",
"time": "225.926",
"value": "Khety I"
},
{
"operation": "Now on",
"time": "226.345",
"value": "Khety I"
},
{
"operation": "search",
"time": "244.213",
"value": "Meryibtawy"
},
{
"operation": "Now on",
"time": "244.687",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "251.862",
"value": "List of ancient Egyptians"
},
{
"operation": "Now on",
"time": "254.22",
"value": "List of ancient Egyptians"
},
{
"operation": "Page search",
"time": "259.809",
"value": "Khety"
},
{
"operation": "hyperlink",
"time": "288.027",
"value": "Khety I"
},
{
"operation": "Now on",
"time": "288.399",
"value": "Khety I (nomarch)"
},
{
"operation": "search",
"time": "338.899",
"value": "The Ninth Dynasty of Egypt"
},
{
"operation": "Now on",
"time": "340.565",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "354.268",
"value": "Ninth Dynasty of Egypt"
},
{
"operation": "Now on",
"time": "354.803",
"value": "Ninth Dynasty of Egypt"
},
{
"operation": "hyperlink",
"time": "466.689",
"value": "Meryibre Khety"
},
{
"operation": "Now on",
"time": "467.269",
"value": "Meryibre Khety"
},
{
"operation": "Highlighting",
"time": "493.185",
"value": "Meryibre Khety_sentence_4"
},
{
"operation": "finish",
"time": "533.975",
"value": "finish"
}
]
}
|
exfever_7410
|
Meek Mill is an American rapper born in the largest city in the Commonwealth of Pennsylvania in a country primarily located in North America.
|
{
"evidence": [
"Meek Mill",
"Philadelphia",
"United States"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nMeek Mill is an American rapper born in the largest city in the Commonwealth of Pennsylvania in a country primarily located in North America.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Meek Mill",
"Philadelphia",
"United States"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8110,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_6131
|
Django Unchained is a 2017 American revisionist Western film written and directed by Quentin Tarantino and starring an American actor who is one of the most widely recognized actors of his generation.
|
{
"evidence": [
"Django Unchained",
"Samuel L. Jackson"
],
"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:\nDjango Unchained is a 2017 American revisionist Western film written and directed by Quentin Tarantino and starring an American actor who is one of the most widely recognized actors of his generation.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Django Unchained",
"Samuel L. Jackson"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 8111,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_1352
|
The Boys from Brazil is a 1978 British-American science fiction thriller film starring an American actor and one of the most popular film stars from the 1940s to the 1960s.
|
{
"evidence": [
"The Boys from Brazil (film)",
"Gregory Peck"
],
"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 Boys from Brazil is a 1978 British-American science fiction thriller film starring an American actor and one of the most popular film stars from the 1940s to the 1960s.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Boys from Brazil (film)",
"Gregory Peck"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8112,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
41430
|
There were no significant lunar events in 1942.
|
{
"evidence": [
"Lunar_sentence_0"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThere were no significant lunar events in 1942.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Lunar_sentence_0"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 8113,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.796",
"value": "?search="
},
{
"operation": "search",
"time": "14.883",
"value": "lunar events"
},
{
"operation": "Now on",
"time": "16.399",
"value": "undefined"
},
{
"operation": "search",
"time": "23.697",
"value": "lunar events 1942"
},
{
"operation": "Now on",
"time": "24.957",
"value": "undefined"
},
{
"operation": "search",
"time": "60.344",
"value": "1942 lunar events"
},
{
"operation": "Now on",
"time": "60.929",
"value": "undefined"
},
{
"operation": "search",
"time": "74.86",
"value": "Lunar"
},
{
"operation": "Now on",
"time": "75.661",
"value": "Lunar"
},
{
"operation": "Highlighting",
"time": "116.383",
"value": "Lunar_sentence_0"
},
{
"operation": "finish",
"time": "120.734",
"value": "finish"
}
]
}
|
27932
|
Opponents of same sex marriage says it undermines traditional family and may lead to legalization of polygamy but the LGBT movements are supported unanimously.
|
{
"evidence": [
"LGBT social movements_sentence_208"
],
"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:\nOpponents of same sex marriage says it undermines traditional family and may lead to legalization of polygamy but the LGBT movements are supported unanimously.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"LGBT social movements_sentence_208"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 8114,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.152",
"value": "?search="
},
{
"operation": "search",
"time": "239.174",
"value": "Contains...undermines traditional family and may lead to legalization of polygamy"
},
{
"operation": "Now on",
"time": "241.383",
"value": "undefined"
},
{
"operation": "search",
"time": "332.638",
"value": "lgbt colorado"
},
{
"operation": "Now on",
"time": "334.976",
"value": "undefined"
},
{
"operation": "search",
"time": "343.537",
"value": "LGBT"
},
{
"operation": "Now on",
"time": "345.21",
"value": "LGBT"
},
{
"operation": "Page search",
"time": "351.391",
"value": "polygamy"
},
{
"operation": "search",
"time": "365.427",
"value": "LGBT social movements"
},
{
"operation": "Now on",
"time": "367.585",
"value": "LGBT social movements"
},
{
"operation": "Page search",
"time": "371.942",
"value": "polygamy"
},
{
"operation": "Highlighting",
"time": "386.323",
"value": "LGBT social movements_sentence_210"
},
{
"operation": "Highlighting",
"time": "400.443",
"value": "LGBT social movements_sentence_208"
},
{
"operation": "Highlighting deleted",
"time": "401.965",
"value": "LGBT social movements_sentence_210"
},
{
"operation": "finish",
"time": "410.055",
"value": "finish"
}
]
}
|
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": 8115,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
77405
|
In 2011, with an estimated cost of less than US$100 million, the Guggenheim in Abu Dhabi had begun its construction with a Roman architectural style but was later suspended.
|
{
"evidence": [
"Guggenheim Abu Dhabi_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:\nIn 2011, with an estimated cost of less than US$100 million, the Guggenheim in Abu Dhabi had begun its construction with a Roman architectural style but was later suspended.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Guggenheim Abu Dhabi_sentence_3"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 8116,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.815",
"value": "?search="
},
{
"operation": "search",
"time": "13.988",
"value": "Contains...Guggenheim in Abu Dhabi"
},
{
"operation": "Now on",
"time": "15.157",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "30.359",
"value": "Early history Guggenheim Abu Dhabi_section_0"
},
{
"operation": "Now on",
"time": "31.223",
"value": "Guggenheim Abu Dhabi#Early history Guggenheim Abu Dhabi section 0"
},
{
"operation": "Page search",
"time": "56.01",
"value": "roman"
},
{
"operation": "Page search",
"time": "107.294",
"value": "100"
},
{
"operation": "search",
"time": "120.99",
"value": "with an estimated cost of less than US$100 million, the Guggenheim in Abu Dhabi had begun its construction with a Roman architectural style but was later suspended"
},
{
"operation": "Now on",
"time": "124.373",
"value": "undefined"
},
{
"operation": "search",
"time": "134.597",
"value": "begun its construction with a Roman architectural style but was later suspended"
},
{
"operation": "Now on",
"time": "137.361",
"value": "undefined"
},
{
"operation": "search",
"time": "152.919",
"value": "Contains...Guggenheim in Abu Dhabi"
},
{
"operation": "Now on",
"time": "153.686",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "163.66",
"value": "Gulf Labor"
},
{
"operation": "Now on",
"time": "164.493",
"value": "Gulf Labor"
},
{
"operation": "back-button-clicked",
"time": "170.98",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "171.874",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "174.739",
"value": "Guggenheim Abu Dhabi"
},
{
"operation": "Now on",
"time": "175.546",
"value": "Guggenheim Abu Dhabi"
},
{
"operation": "Highlighting",
"time": "179.519",
"value": "Guggenheim Abu Dhabi_sentence_3"
},
{
"operation": "finish",
"time": "319.582",
"value": "finish"
}
]
}
|
exfever_3630
|
Your Highness is a 2011 American stoner comic fantasy film starring an American actor, comedian, producer, and writer who appeared in a 2008 American stoner comedy film directed by David Gordon Green.
|
{
"evidence": [
"Your Highness",
"Danny McBride",
"Pineapple Express (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:\nYour Highness is a 2011 American stoner comic fantasy film starring an American actor, comedian, producer, and writer who appeared in a 2008 American stoner comedy film directed by David Gordon Green.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Your Highness",
"Danny McBride",
"Pineapple Express (film)"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8117,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
50269
|
Polyamides, a polymer with repeating units linked by amide bonds, has 13–16% DecaBDE and brominated polystyrene; while Polystyrene foam has 0.8–4% HBCD.
|
{
"evidence": [
"Polyamide_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:\nPolyamides, a polymer with repeating units linked by amide bonds, has 13–16% DecaBDE and brominated polystyrene; while Polystyrene foam has 0.8–4% HBCD.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Polyamide_sentence_0"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 8118,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.4",
"value": "?search="
},
{
"operation": "search",
"time": "9.245",
"value": "Polyamides"
},
{
"operation": "Now on",
"time": "10.138",
"value": "Polyamide"
},
{
"operation": "Highlighting",
"time": "15.163",
"value": "Polyamide_sentence_0"
},
{
"operation": "hyperlink",
"time": "64.239",
"value": "amide"
},
{
"operation": "Now on",
"time": "65.082",
"value": "Amide"
},
{
"operation": "back-button-clicked",
"time": "75.655",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "76.176",
"value": "Polyamide"
},
{
"operation": "search",
"time": "102.331",
"value": "Contains...13–16% DecaBDE and brominated polystyrene"
},
{
"operation": "Now on",
"time": "103.589",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "111.708",
"value": "Decabromodiphenyl ether"
},
{
"operation": "Now on",
"time": "112.553",
"value": "Decabromodiphenyl ether"
},
{
"operation": "search",
"time": "144.962",
"value": "Contains...DecaBDE and brominated polystyrene"
},
{
"operation": "Now on",
"time": "145.453",
"value": "undefined"
},
{
"operation": "search",
"time": "169.603",
"value": "Contains...Polystyrene foam has 0.8–4% HBCD"
},
{
"operation": "Now on",
"time": "170.698",
"value": "undefined"
},
{
"operation": "finish",
"time": "186.015",
"value": "finish"
}
]
}
|
10463
|
Jill Haworth made her film debut in a feature film by Preminger, an Austro-Hungarian-born theatre and film director whose full name is Otto Ludwig Preminger.
|
{
"evidence": [
"Jill Haworth_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:\nJill Haworth made her film debut in a feature film by Preminger, an Austro-Hungarian-born theatre and film director whose full name is Otto Ludwig Preminger.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Jill Haworth_sentence_8"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 8119,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.993",
"value": "?search="
},
{
"operation": "search",
"time": "11.826",
"value": "Jill Haworth"
},
{
"operation": "Now on",
"time": "12.863",
"value": "Jill Haworth"
},
{
"operation": "Page search",
"time": "50.546",
"value": "debut"
},
{
"operation": "hyperlink",
"time": "79.593",
"value": "Exodus"
},
{
"operation": "Now on",
"time": "80.332",
"value": "Exodus (1960 film)"
},
{
"operation": "back-button-clicked",
"time": "90.857",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "91.029",
"value": "Jill Haworth"
},
{
"operation": "Highlighting",
"time": "107.331",
"value": "Jill Haworth_sentence_8"
},
{
"operation": "finish",
"time": "122.049",
"value": "finish"
}
]
}
|
71336
|
The 2018 Arkansas elections was held in the US state of Arkansas on November 6, 2018 and all of Arkansas' executive officers were up for election.
|
{
"evidence": [
"2018 Arkansas elections_sentence_2"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe 2018 Arkansas elections was held in the US state of Arkansas on November 6, 2018 and all of Arkansas' executive officers were up for election.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"2018 Arkansas elections_sentence_2"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 8120,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.672",
"value": "?search="
},
{
"operation": "search",
"time": "9.358",
"value": "2018 Arkansas elections"
},
{
"operation": "Now on",
"time": "11.105",
"value": "2018 Arkansas elections"
},
{
"operation": "Highlighting",
"time": "33.241",
"value": "2018 Arkansas elections_sentence_2"
},
{
"operation": "Highlighting deleted",
"time": "52.506",
"value": "2018 Arkansas elections_sentence_2"
},
{
"operation": "Highlighting",
"time": "53.243",
"value": "2018 Arkansas elections_sentence_02018 Arkansas elections_sentence_1"
},
{
"operation": "Highlighting",
"time": "55.737",
"value": "2018 Arkansas elections_sentence_2"
},
{
"operation": "Highlighting deleted",
"time": "56.546",
"value": "2018 Arkansas elections_sentence_02018 Arkansas elections_sentence_1"
},
{
"operation": "finish",
"time": "62.784",
"value": "finish"
}
]
}
|
exfever_1170
|
"Born This Way" is a lead song by African singer Lady Gaga, who produced it along with an American music producer based in Chicago, Illinois.
|
{
"evidence": [
"Born This Way (song)",
"DJ White Shadow"
],
"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:\n\"Born This Way\" is a lead song by African singer Lady Gaga, who produced it along with an American music producer based in Chicago, Illinois.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Born This Way (song)",
"DJ White Shadow"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 8121,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_3886
|
Trolls is a 2016 American computer-animated musical comedy series directed by an American film director who directed Shrek Forever After.
|
{
"evidence": [
"Trolls (film)",
"Mike Mitchell (director)"
],
"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:\nTrolls is a 2016 American computer-animated musical comedy series directed by an American film director who directed Shrek Forever After.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Trolls (film)",
"Mike Mitchell (director)"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 8122,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7702
|
Stephanie Daley is a 2006 film starring the youngest recipient of the Academy Award for Best Supporting Actor.
|
{
"evidence": [
"Stephanie Daley",
"Timothy Hutton"
],
"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:\nStephanie Daley is a 2006 film starring the youngest recipient of the Academy Award for Best Supporting Actor.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Stephanie Daley",
"Timothy Hutton"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8123,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
40747
|
Krastyo Lazarov was recruited by the pro-Bulgarian Internal Macedonian Revolutionary Organization, and was initially part of the ceta of general Ivan Tsonchev.
|
{
"evidence": [
"Krsta Kovačević_sentence_5",
"Krsta Kovačević_sentence_6"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nKrastyo Lazarov was recruited by the pro-Bulgarian Internal Macedonian Revolutionary Organization, and was initially part of the ceta of general Ivan Tsonchev.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Krsta Kovačević_sentence_5",
"Krsta Kovačević_sentence_6"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 8124,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.722",
"value": "?search="
},
{
"operation": "search",
"time": "19.136",
"value": "Krastyo Lazarov"
},
{
"operation": "Now on",
"time": "19.878",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "24.715",
"value": "Krsta Kovačević"
},
{
"operation": "Now on",
"time": "25.423",
"value": "Krsta Kovačević"
},
{
"operation": "back-button-clicked",
"time": "32.518",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "33.113",
"value": "undefined"
},
{
"operation": "search",
"time": "46.546",
"value": "Krastyo Lazarov"
},
{
"operation": "Now on",
"time": "47.222",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "61.663",
"value": "Krsta Kovačević"
},
{
"operation": "Now on",
"time": "62.085",
"value": "Krsta Kovačević"
},
{
"operation": "Page search",
"time": "65.22",
"value": "Krastyo Lazarov"
},
{
"operation": "Page search",
"time": "65.787",
"value": "Krastyo Lazarov"
},
{
"operation": "Page search",
"time": "66.255",
"value": "Krastyo Lazarov"
},
{
"operation": "Page search",
"time": "76.113",
"value": "Krastyo Lazaro"
},
{
"operation": "search",
"time": "114.082",
"value": "pro-Bulgarian Internal Macedonian Revolutionary Organization"
},
{
"operation": "Now on",
"time": "115.932",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "139.283",
"value": "Krsta Kovačević"
},
{
"operation": "Now on",
"time": "139.682",
"value": "Krsta Kovačević"
},
{
"operation": "Page search",
"time": "143.815",
"value": "pro-Bulgarian Internal Macedonian Revolutionary Organization"
},
{
"operation": "Page search",
"time": "162.912",
"value": "recruit"
},
{
"operation": "Highlighting",
"time": "182.255",
"value": "Krsta Kovačević_sentence_5"
},
{
"operation": "Highlighting",
"time": "189.872",
"value": "Krsta Kovačević_sentence_6"
},
{
"operation": "finish",
"time": "224.122",
"value": "finish"
}
]
}
|
exfever_2806
|
Walter Pater was the greatest writer in 1880s.
|
{
"evidence": [
"Aestheticism",
"Oscar Wilde",
"Dublin"
],
"label": "NOT ENOUGH INFO"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nWalter Pater was the greatest writer in 1880s.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Aestheticism",
"Oscar Wilde",
"Dublin"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 8125,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
35012
|
Lundtoftbjerg is an area 1 mile south of the small village of Lundtoft which was a municipality in South Jutland County on the east coast of the Jutland peninsula in south Denmark.
|
{
"evidence": [
"Lundtoftbjerg_sentence_0",
"Lundtoft Municipality_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:\nLundtoftbjerg is an area 1 mile south of the small village of Lundtoft which was a municipality in South Jutland County on the east coast of the Jutland peninsula in south Denmark.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Lundtoftbjerg_sentence_0",
"Lundtoft Municipality_sentence_0"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8126,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.931",
"value": "?search="
},
{
"operation": "search",
"time": "13.225",
"value": "Contains...undtoftbjerg"
},
{
"operation": "Now on",
"time": "13.579",
"value": "undefined"
},
{
"operation": "search",
"time": "21.523",
"value": "Lundtoftbjerg"
},
{
"operation": "Now on",
"time": "22.155",
"value": "Lundtoftbjerg"
},
{
"operation": "Highlighting",
"time": "23.648",
"value": "Lundtoftbjerg_sentence_0"
},
{
"operation": "hyperlink",
"time": "31.364",
"value": "Lundtoft"
},
{
"operation": "Now on",
"time": "31.876",
"value": "Lundtoft"
},
{
"operation": "Highlighting",
"time": "39.144",
"value": "Lundtoft Municipality_sentence_0"
},
{
"operation": "finish",
"time": "73.799",
"value": "finish"
}
]
}
|
exfever_1821
|
Chris Benoit has worked for numerous pro-wrestling promotions, including most notably, with a shortened pseudonym, WWE.
|
{
"evidence": [
"Chris Benoit",
"WWE",
"Trade name"
],
"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:\nChris Benoit has worked for numerous pro-wrestling promotions, including most notably, with a shortened pseudonym, WWE.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Chris Benoit",
"WWE",
"Trade name"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 8127,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_8864
|
Leopold II was the third King of the Belgians who claimed the Congo with the help of a Welsh-American explorer who is mainly known for his search for the sources of the major north-flowing river in northeastern Africa.
|
{
"evidence": [
"Leopold II of Belgium",
"Henry Morton Stanley",
"Nile"
],
"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:\nLeopold II was the third King of the Belgians who claimed the Congo with the help of a Welsh-American explorer who is mainly known for his search for the sources of the major north-flowing river in northeastern Africa.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Leopold II of Belgium",
"Henry Morton Stanley",
"Nile"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 8128,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
94103
|
The summer transfer window, which is the period during the year in which a football club can transfer players from other playing staff into their playing staff. that opened on 1 July 2015 and ended on 1 September 2015.
|
{
"evidence": [
"Transfer window_sentence_0",
"Transfer window_sentence_16"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe summer transfer window, which is the period during the year in which a football club can transfer players from other playing staff into their playing staff. that opened on 1 July 2015 and ended on 1 September 2015.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Transfer window_sentence_0",
"Transfer window_sentence_16"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 8129,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.358",
"value": "?search="
},
{
"operation": "search",
"time": "11.454",
"value": "Contains...summer transfer window"
},
{
"operation": "Now on",
"time": "14.709",
"value": "undefined"
},
{
"operation": "search",
"time": "25.606",
"value": "\"summer transfer window\""
},
{
"operation": "Now on",
"time": "27.141",
"value": "undefined"
},
{
"operation": "search",
"time": "48.56",
"value": "\"summer transfer window 2015\""
},
{
"operation": "Now on",
"time": "49.885",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "62.456",
"value": "List of Norwegian football transfers summer 2015"
},
{
"operation": "Now on",
"time": "63.919",
"value": "List of Norwegian football transfers summer 2015"
},
{
"operation": "search",
"time": "90.065",
"value": "\"period during the year in which a football club can transfer players from other playing staff\""
},
{
"operation": "Now on",
"time": "91.945",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "93.571",
"value": "Transfer window"
},
{
"operation": "Now on",
"time": "94.575",
"value": "Transfer window"
},
{
"operation": "Highlighting",
"time": "97.162",
"value": "Transfer window_sentence_0"
},
{
"operation": "Page search",
"time": "121.836",
"value": "summer"
},
{
"operation": "Highlighting",
"time": "199.106",
"value": "Transfer window_cell_0_8_0"
},
{
"operation": "Highlighting",
"time": "199.111",
"value": "Transfer window_header_cell_0_0_0"
},
{
"operation": "Highlighting deleted",
"time": "205.267",
"value": "Transfer window_cell_0_8_0"
},
{
"operation": "Highlighting deleted",
"time": "205.273",
"value": "Transfer window_header_cell_0_0_0"
},
{
"operation": "Highlighting",
"time": "215.447",
"value": "Transfer window_sentence_16"
},
{
"operation": "finish",
"time": "224.775",
"value": "finish"
}
]
}
|
exfever_5915
|
Marilyn Monroe starred in three movies that were released in 1953.
|
{
"evidence": [
"Marilyn Monroe",
"AFI's 100 Years...100 Stars",
"Katharine Hepburn"
],
"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:\nMarilyn Monroe starred in three movies that were released in 1953.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Marilyn Monroe",
"AFI's 100 Years...100 Stars",
"Katharine Hepburn"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 8130,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_6960
|
Colin Edward Quinn's notable film work includes his role as Dooey in a 1998 American comedy film.
|
{
"evidence": [
"Colin Quinn",
"A Night at the Roxbury"
],
"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:\nColin Edward Quinn's notable film work includes his role as Dooey in a 1998 American comedy film.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Colin Quinn",
"A Night at the Roxbury"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8131,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_3455
|
Judd Apatow is a director who developed an Asian sitcom that aired on Fox during the 2001–02 season.
|
{
"evidence": [
"Judd Apatow",
"Undeclared"
],
"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:\nJudd Apatow is a director who developed an Asian sitcom that aired on Fox during the 2001–02 season.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Judd Apatow",
"Undeclared"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 8132,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_6651
|
All My Children is an American television soap opera, created by Will Smith, set in Pine Valley, Pennsylvania, a fictional suburb of the largest city in the Commonwealth of Pennsylvania in the United States.
|
{
"evidence": [
"All My Children",
"Philadelphia"
],
"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:\nAll My Children is an American television soap opera, created by Will Smith, set in Pine Valley, Pennsylvania, a fictional suburb of the largest city in the Commonwealth of Pennsylvania in the United States.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"All My Children",
"Philadelphia"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 8133,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
67244
|
Hans Memling was a painter in Flanders who painted works such as Scenes of the Passion of Christ and The Last Judgement.
|
{
"evidence": [
"Hans Memling_sentence_2",
"Hans Memling_sentence_16",
"Hans Memling_sentence_17"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nHans Memling was a painter in Flanders who painted works such as Scenes of the Passion of Christ and The Last Judgement.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Hans Memling_sentence_2",
"Hans Memling_sentence_16",
"Hans Memling_sentence_17"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8134,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.769",
"value": "?search="
},
{
"operation": "search",
"time": "44.082",
"value": "Hans Memling"
},
{
"operation": "Now on",
"time": "144.149",
"value": "Hans Memling"
},
{
"operation": "Highlighting",
"time": "150.32",
"value": "Hans Memling_sentence_2"
},
{
"operation": "Highlighting",
"time": "164.395",
"value": "Hans Memling_sentence_16"
},
{
"operation": "Highlighting",
"time": "165.65",
"value": "Hans Memling_sentence_17"
},
{
"operation": "Now on",
"time": "286.508",
"value": "Hans Memling"
},
{
"operation": "finish",
"time": "304.082",
"value": "finish"
}
]
}
|
31855
|
Maharaja Theme Park, an amusement park in Tamil Nadu, India (the southernmost part of the Indian subcontinent), was opened to the public in September 2010.
|
{
"evidence": [
"Maharaja Theme Park_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:\nMaharaja Theme Park, an amusement park in Tamil Nadu, India (the southernmost part of the Indian subcontinent), was opened to the public in September 2010.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Maharaja Theme Park_sentence_2"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 8135,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.738",
"value": "?search="
},
{
"operation": "search",
"time": "9.235",
"value": "Maharaja Theme Park"
},
{
"operation": "Now on",
"time": "10.034",
"value": "Maharaja Theme Park"
},
{
"operation": "Highlighting",
"time": "21.694",
"value": "Maharaja Theme Park_sentence_2"
},
{
"operation": "hyperlink",
"time": "43.496",
"value": "India"
},
{
"operation": "Now on",
"time": "45.276",
"value": "India"
},
{
"operation": "search",
"time": "65.598",
"value": "Maharaja Theme Park"
},
{
"operation": "Now on",
"time": "66.102",
"value": "Maharaja Theme Park"
},
{
"operation": "finish",
"time": "75.262",
"value": "finish"
}
]
}
|
40313
|
Javier Patiño scored his first international goal against Bahrain on June 11, 2015, during the 2018 FIFA World Cup qualification held at the Philippine Sports Stadium, Bulacan, Philippines.
|
{
"evidence": [
"Javier Patiño_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:\nJavier Patiño scored his first international goal against Bahrain on June 11, 2015, during the 2018 FIFA World Cup qualification held at the Philippine Sports Stadium, Bulacan, Philippines.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Javier Patiño_sentence_19"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 8136,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.002",
"value": "?search="
},
{
"operation": "search",
"time": "0.013",
"value": "Javier Patiño"
},
{
"operation": "Now on",
"time": "0.015",
"value": "Javier Patiño"
},
{
"operation": "hyperlink",
"time": "0.029",
"value": "3 International career Javier Patiño_section_6"
},
{
"operation": "Highlighting",
"time": "0.055",
"value": "Javier Patiño_sentence_19"
},
{
"operation": "Highlighting",
"time": "0.09",
"value": "Javier Patiño_item_1_1"
},
{
"operation": "Highlighting",
"time": "0.09",
"value": "Javier Patiño_section_7"
},
{
"operation": "Highlighting",
"time": "0.09",
"value": "Javier Patiño_section_9"
},
{
"operation": "Highlighting deleted",
"time": "0.091",
"value": "Javier Patiño_item_1_1"
},
{
"operation": "Highlighting deleted",
"time": "0.091",
"value": "Javier Patiño_section_7"
},
{
"operation": "Highlighting deleted",
"time": "0.091",
"value": "Javier Patiño_section_9"
},
{
"operation": "finish",
"time": "0.111",
"value": "finish"
},
{
"operation": "Now on",
"time": "1620676055.736",
"value": "Javier Patiño"
},
{
"operation": "finish",
"time": "1620676058.255",
"value": "finish"
}
]
}
|
34469
|
Two of the longest commissioners for central vigilance, an Italian governmental body created in 1964 to address governmental corruption, are T. U. Vijayasekharan (April 04, 1990 - April 03, 1995) and B. K. Acharya (September 09, 1972 - September 08, 1977).
|
{
"evidence": [
"Central Vigilance Commission_sentence_0"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nTwo of the longest commissioners for central vigilance, an Italian governmental body created in 1964 to address governmental corruption, are T. U. Vijayasekharan (April 04, 1990 - April 03, 1995) and B. K. Acharya (September 09, 1972 - September 08, 1977).\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"Central Vigilance Commission_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 8137,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.923",
"value": "?search="
},
{
"operation": "search",
"time": "20.171",
"value": "Contains...commissioners for central vigilance"
},
{
"operation": "Now on",
"time": "21.697",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "23.771",
"value": "Central Vigilance Commissioners Central Vigilance Commission_section_9"
},
{
"operation": "Now on",
"time": "25.024",
"value": "Central Vigilance Commission#Central Vigilance Commissioners Central Vigilance Commission section 9"
},
{
"operation": "Highlighting",
"time": "35.192",
"value": "Central Vigilance Commission_sentence_0"
},
{
"operation": "finish",
"time": "44.974",
"value": "finish"
}
]
}
|
11543
|
Before being a member of the Knesset for the Progressive Party and Liberal Party between 1949 and 1963, Idov Cohen (born on 1909 in Botoșani County, Romania) was a Romanian-Israeli politician who worked as a bookkeeper and as a journalist.
|
{
"evidence": [
"Idov Cohen_sentence_2",
"Idov Cohen_sentence_1",
"Idov Cohen_sentence_3",
"Idov Cohen_sentence_4",
"Idov Cohen_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:\nBefore being a member of the Knesset for the Progressive Party and Liberal Party between 1949 and 1963, Idov Cohen (born on 1909 in Botoșani County, Romania) was a Romanian-Israeli politician who worked as a bookkeeper and as a journalist.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Idov Cohen_sentence_2",
"Idov Cohen_sentence_1",
"Idov Cohen_sentence_3",
"Idov Cohen_sentence_4",
"Idov Cohen_sentence_0"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8138,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.867",
"value": "?search="
},
{
"operation": "search",
"time": "10.312",
"value": "Idov Cohen"
},
{
"operation": "Now on",
"time": "11.474",
"value": "Idov Cohen"
},
{
"operation": "Highlighting",
"time": "15.357",
"value": "Idov Cohen_cell_0_1_1"
},
{
"operation": "Highlighting",
"time": "15.367",
"value": "Idov Cohen_header_cell_0_1_0"
},
{
"operation": "Highlighting",
"time": "15.367",
"value": "Idov Cohen_header_cell_0_0_0"
},
{
"operation": "Highlighting",
"time": "23.336",
"value": "Idov Cohen_sentence_2"
},
{
"operation": "Highlighting deleted",
"time": "26.627",
"value": "Idov Cohen_cell_0_1_1"
},
{
"operation": "Highlighting deleted",
"time": "26.634",
"value": "Idov Cohen_header_cell_0_1_0"
},
{
"operation": "Highlighting deleted",
"time": "26.636",
"value": "Idov Cohen_header_cell_0_0_0"
},
{
"operation": "Page search",
"time": "38.897",
"value": "knesset"
},
{
"operation": "Highlighting",
"time": "43.937",
"value": "Idov Cohen_sentence_1"
},
{
"operation": "Page search",
"time": "54.966",
"value": "politican"
},
{
"operation": "Highlighting",
"time": "60.43",
"value": "Idov Cohen_sentence_3"
},
{
"operation": "Highlighting",
"time": "61.355",
"value": "Idov Cohen_sentence_4"
},
{
"operation": "Highlighting",
"time": "75.455",
"value": "Idov Cohen_sentence_0"
},
{
"operation": "finish",
"time": "123.536",
"value": "finish"
}
]
}
|
70283
|
Narasimha Raya II became the Emperor of the Vijayanagara Empire when he was a teenager following the violent death of his elder brother.
|
{
"evidence": [
"Narasimha Raya II_sentence_9",
"Narasimha Raya II_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:\nNarasimha Raya II became the Emperor of the Vijayanagara Empire when he was a teenager following the violent death of his elder brother.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Narasimha Raya II_sentence_9",
"Narasimha Raya II_sentence_10"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8139,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.367",
"value": "?search="
},
{
"operation": "search",
"time": "11.193",
"value": "Narasimha Raya II"
},
{
"operation": "Now on",
"time": "11.956",
"value": "Narasimha Raya II"
},
{
"operation": "Page search",
"time": "26.585",
"value": "emperor"
},
{
"operation": "Highlighting",
"time": "29.378",
"value": "Narasimha Raya II_sentence_4"
},
{
"operation": "Highlighting",
"time": "29.379",
"value": "Narasimha Raya II_section_0"
},
{
"operation": "Highlighting",
"time": "46.853",
"value": "Narasimha Raya II_sentence_9"
},
{
"operation": "Highlighting deleted",
"time": "48.295",
"value": "Narasimha Raya II_sentence_4"
},
{
"operation": "Highlighting deleted",
"time": "48.297",
"value": "Narasimha Raya II_section_0"
},
{
"operation": "Highlighting",
"time": "53.489",
"value": "Narasimha Raya II_sentence_10"
},
{
"operation": "finish",
"time": "73.842",
"value": "finish"
}
]
}
|
exfever_69
|
The Football League Cup Final was played between Leicester City and an amateur association football club based in Birkenhead, Merseyside, England.
|
{
"evidence": [
"2000 Football League Cup Final",
"Tranmere Rovers F.C."
],
"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 Football League Cup Final was played between Leicester City and an amateur association football club based in Birkenhead, Merseyside, England.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"2000 Football League Cup Final",
"Tranmere Rovers F.C."
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 8140,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
81514
|
Mayan graffiti, scratched into surfaces using pointed tools, portray a variety of themes and range of topics, and preliminary studies suggest that the authors of graffiti were not trained artists.
|
{
"evidence": [
"Ancient Maya graffiti_sentence_19",
"Ancient Maya graffiti_sentence_37",
"Ancient Maya graffiti_sentence_43"
],
"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:\nMayan graffiti, scratched into surfaces using pointed tools, portray a variety of themes and range of topics, and preliminary studies suggest that the authors of graffiti were not trained artists.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Ancient Maya graffiti_sentence_19",
"Ancient Maya graffiti_sentence_37",
"Ancient Maya graffiti_sentence_43"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8141,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.521",
"value": "undefined"
},
{
"operation": "search",
"time": "13.257",
"value": "Mayan grafitti"
},
{
"operation": "Now on",
"time": "13.956",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "16.865",
"value": "mayan graffiti"
},
{
"operation": "Now on",
"time": "17.759",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "28.046",
"value": "Maya civilization"
},
{
"operation": "Now on",
"time": "29.259",
"value": "Maya civilization"
},
{
"operation": "Page search",
"time": "34.11",
"value": "graffitti"
},
{
"operation": "Page search",
"time": "48.315",
"value": "graffiti"
},
{
"operation": "Highlighting",
"time": "617.803",
"value": "Maya civilization_sentence_404"
},
{
"operation": "Highlighting",
"time": "617.806",
"value": "Maya civilization_title"
},
{
"operation": "Highlighting",
"time": "617.808",
"value": "Maya civilization_section_20"
},
{
"operation": "Highlighting deleted",
"time": "619.029",
"value": "Maya civilization_sentence_404"
},
{
"operation": "Highlighting deleted",
"time": "619.032",
"value": "Maya civilization_title"
},
{
"operation": "Highlighting deleted",
"time": "619.036",
"value": "Maya civilization_section_20"
},
{
"operation": "Highlighting",
"time": "658.883",
"value": "Maya civilization_sentence_406"
},
{
"operation": "Highlighting",
"time": "658.885",
"value": "Maya civilization_title"
},
{
"operation": "Highlighting",
"time": "658.886",
"value": "Maya civilization_section_20"
},
{
"operation": "Highlighting",
"time": "667.684",
"value": "Maya civilization_sentence_405"
},
{
"operation": "search",
"time": "695.764",
"value": "Mayan art"
},
{
"operation": "Now on",
"time": "697.036",
"value": "Ancient Maya art"
},
{
"operation": "Page search",
"time": "709.489",
"value": "graffiti"
},
{
"operation": "search",
"time": "908.314",
"value": "graffiti"
},
{
"operation": "Now on",
"time": "909.135",
"value": "undefined"
},
{
"operation": "search",
"time": "914.09",
"value": "graffiti"
},
{
"operation": "Now on",
"time": "914.619",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "928.185",
"value": "Listings of graffiti Graffito (archaeology)_section_0"
},
{
"operation": "Now on",
"time": "928.748",
"value": "Graffito (archaeology)#Listings of graffiti Graffito .28archaeology.29 section 0"
},
{
"operation": "hyperlink",
"time": "949.825",
"value": "Ancient Maya graffiti"
},
{
"operation": "Now on",
"time": "950.38",
"value": "Ancient Maya graffiti"
},
{
"operation": "Highlighting",
"time": "955.648",
"value": "Ancient Maya graffiti_sentence_1"
},
{
"operation": "Highlighting deleted",
"time": "973.626",
"value": "Ancient Maya graffiti_sentence_1"
},
{
"operation": "Highlighting",
"time": "1003.339",
"value": "Ancient Maya graffiti_sentence_19"
},
{
"operation": "Highlighting",
"time": "1023.412",
"value": "Ancient Maya graffiti_sentence_37"
},
{
"operation": "Highlighting",
"time": "1029.71",
"value": "Ancient Maya graffiti_sentence_43"
},
{
"operation": "finish",
"time": "1040.584",
"value": "finish"
}
]
}
|
83540
|
Art Hirahara is a pianist and composer who studied under a famous pianist at Oberlin.
|
{
"evidence": [
"Art Hirahara_sentence_0",
"Art Hirahara_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:\nArt Hirahara is a pianist and composer who studied under a famous pianist at Oberlin.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Art Hirahara_sentence_0",
"Art Hirahara_sentence_3"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 8142,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.262",
"value": "?search="
},
{
"operation": "search",
"time": "77.687",
"value": "Art Hirahara"
},
{
"operation": "Now on",
"time": "78.269",
"value": "Art Hirahara"
},
{
"operation": "Highlighting",
"time": "85.081",
"value": "Art Hirahara_sentence_0"
},
{
"operation": "Highlighting",
"time": "100.067",
"value": "Art Hirahara_sentence_3"
},
{
"operation": "Highlighting deleted",
"time": "123.44",
"value": "Art Hirahara_sentence_3"
},
{
"operation": "Highlighting",
"time": "125.58",
"value": "Art Hirahara_sentence_3"
},
{
"operation": "search",
"time": "131.309",
"value": "Neal Creque"
},
{
"operation": "Now on",
"time": "132.246",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "138.038",
"value": "2000s in jazz"
},
{
"operation": "Now on",
"time": "138.898",
"value": "2000s in jazz"
},
{
"operation": "Page search",
"time": "141.701",
"value": "Neal Creque"
},
{
"operation": "Page search",
"time": "142.599",
"value": "Neal Creque"
},
{
"operation": "back-button-clicked",
"time": "150.979",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "151.488",
"value": "undefined"
},
{
"operation": "back-button-clicked",
"time": "153.57",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "153.574",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "153.632",
"value": "Art Hirahara"
},
{
"operation": "Page search",
"time": "161.123",
"value": "famous"
},
{
"operation": "finish",
"time": "196.491",
"value": "finish"
}
]
}
|
exfever_4285
|
The Boston Celtics are an American professional basketball team based in Boston that competes in the National Basketball Association as a member of the league's Eastern Conference division in the Eastern Conference of the National Basketball Association consisting of five teams including an American professional basketball team based in the New York City borough of Brooklyn.
|
{
"evidence": [
"Boston Celtics",
"Atlantic Division (NBA)",
"Brooklyn Nets"
],
"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 Boston Celtics are an American professional basketball team based in Boston that competes in the National Basketball Association as a member of the league's Eastern Conference division in the Eastern Conference of the National Basketball Association consisting of five teams including an American professional basketball team based in the New York City borough of Brooklyn.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Boston Celtics",
"Atlantic Division (NBA)",
"Brooklyn Nets"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8143,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_6953
|
Gary Oldman is an English actor who gained international fame playing the secret agent James Bond in the film series, beginning with a spy film directed by Martin Campbell, and it was the first Eon-produced Bond film to be co-distributed by Columbia Pictures.
|
{
"evidence": [
"Daniel Craig",
"Casino Royale (2006 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:\nGary Oldman is an English actor who gained international fame playing the secret agent James Bond in the film series, beginning with a spy film directed by Martin Campbell, and it was the first Eon-produced Bond film to be co-distributed by Columbia Pictures.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Daniel Craig",
"Casino Royale (2006 film)"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 8144,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_3678
|
Tom Felton is an English actor best known for playing Draco Malfoy in the film adaptations of the series of seven fantasy novels written by British author J.R.R. Tolkien.
|
{
"evidence": [
"Tom Felton",
"Harry Potter"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nTom Felton is an English actor best known for playing Draco Malfoy in the film adaptations of the series of seven fantasy novels written by British author J.R.R. Tolkien.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Tom Felton",
"Harry Potter"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 8145,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
57600
|
Ashqout is a town in Lebanon that are predominantly Maronite catholic and has three schools, one of which is public and the other two are private.
|
{
"evidence": [
"Ashqout_sentence_0",
"Ashqout_sentence_3",
"Ashqout_sentence_4"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nAshqout is a town in Lebanon that are predominantly Maronite catholic and has three schools, one of which is public and the other two are private.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Ashqout_sentence_0",
"Ashqout_sentence_3",
"Ashqout_sentence_4"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8146,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.723",
"value": "?search="
},
{
"operation": "search",
"time": "324.318",
"value": "Contains...Ashqout"
},
{
"operation": "Now on",
"time": "325.22",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "517.865",
"value": "Ashqout"
},
{
"operation": "Now on",
"time": "518.694",
"value": "Ashqout"
},
{
"operation": "Highlighting",
"time": "531.85",
"value": "Ashqout_sentence_0"
},
{
"operation": "Highlighting",
"time": "546.037",
"value": "Ashqout_sentence_3"
},
{
"operation": "Highlighting",
"time": "553.178",
"value": "Ashqout_sentence_4"
},
{
"operation": "finish",
"time": "563.186",
"value": "finish"
}
]
}
|
5047
|
The Bucharest Metro Line M5 (opened on 24 April 1993) construction was completed on the second section of the M5 and the expected construction cost was $708.6 million.
|
{
"evidence": [
"Bucharest Metro Line M5_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 Bucharest Metro Line M5 (opened on 24 April 1993) construction was completed on the second section of the M5 and the expected construction cost was $708.6 million.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Bucharest Metro Line M5_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 8147,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "search",
"time": "15.145",
"value": "Bucharest Metro Line M5"
},
{
"operation": "Highlighting",
"time": "26.551",
"value": "Bucharest Metro Line M5_sentence_0"
},
{
"operation": "finish",
"time": "57.446",
"value": "finish"
}
]
}
|
exfever_1365
|
Lindsay Lohan's breakthrough came in Walt Disney Pictures' 1998 American romantic comedy film directed and co-written by Nancy Meyers.
|
{
"evidence": [
"Lindsay Lohan",
"The Parent Trap (1998 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:\nLindsay Lohan's breakthrough came in Walt Disney Pictures' 1998 American romantic comedy film directed and co-written by Nancy Meyers.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Lindsay Lohan",
"The Parent Trap (1998 film)"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8148,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_4613
|
Dunkirk is a 2017 crime film directed and produced by Christopher Nolan, whose cast includes an English actor, playwright, and theatre director.
|
{
"evidence": [
"Dunkirk (2017 film)",
"Mark Rylance"
],
"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:\nDunkirk is a 2017 crime film directed and produced by Christopher Nolan, whose cast includes an English actor, playwright, and theatre director.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Dunkirk (2017 film)",
"Mark Rylance"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 8149,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
63523
|
The 1794 razees was in response to rumours then circulating of very large French frigates supposed to be under construction and by Admiralty Order of 11 August 1794, two 64-gun ships of the Intrepid class - Anson and Magnanime - were to be cut down by one deck level.
|
{
"evidence": [
"1794 razees_sentence_1",
"1794 razees_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 1794 razees was in response to rumours then circulating of very large French frigates supposed to be under construction and by Admiralty Order of 11 August 1794, two 64-gun ships of the Intrepid class - Anson and Magnanime - were to be cut down by one deck level.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"1794 razees_sentence_1",
"1794 razees_sentence_2"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8150,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.962",
"value": "?search="
},
{
"operation": "search",
"time": "28.938",
"value": "1794 razees"
},
{
"operation": "Now on",
"time": "29.731",
"value": "1794 razees"
},
{
"operation": "Highlighting",
"time": "36.185",
"value": "1794 razees_sentence_1"
},
{
"operation": "Highlighting",
"time": "43.793",
"value": "1794 razees_sentence_2"
},
{
"operation": "finish",
"time": "61.464",
"value": "finish"
}
]
}
|
exfever_8797
|
Carmelo Anthony, a player for the Boston Celtics, also played for one of the most prestigious intercollegiate men's basketball teams, representing a university in New York.
|
{
"evidence": [
"Carmelo Anthony",
"Syracuse Orange men's basketball"
],
"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:\nCarmelo Anthony, a player for the Boston Celtics, also played for one of the most prestigious intercollegiate men's basketball teams, representing a university in New York.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Carmelo Anthony",
"Syracuse Orange men's basketball"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 8151,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_2510
|
Back to the Future Part II is a 1989 Japanese science fiction film directed by Robert Zemeckis that stars an American actress best known for her role in an American thriller film written by Jonathan Hensleigh.
|
{
"evidence": [
"Back to the Future Part II",
"Elisabeth Shue",
"The Saint (1997 film)"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nBack to the Future Part II is a 1989 Japanese science fiction film directed by Robert Zemeckis that stars an American actress best known for her role in an American thriller film written by Jonathan Hensleigh.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Back to the Future Part II",
"Elisabeth Shue",
"The Saint (1997 film)"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 8152,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
21335
|
When Keegan (born on January 26, 1924) finished at school, instead of a British university, he went to Queen's University at Kingston, Ontario; then in 1944, he travelled to the UK from Canada to join the Fleet Air Arm as a pilot.
|
{
"evidence": [
"Denis Keegan_sentence_0",
"Denis Keegan_sentence_5",
"Denis Keegan_sentence_7"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nWhen Keegan (born on January 26, 1924) finished at school, instead of a British university, he went to Queen's University at Kingston, Ontario; then in 1944, he travelled to the UK from Canada to join the Fleet Air Arm as a pilot.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Denis Keegan_sentence_0",
"Denis Keegan_sentence_5",
"Denis Keegan_sentence_7"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8153,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.884",
"value": "?search="
},
{
"operation": "search",
"time": "25.453",
"value": "ravelled to the UK from Canada to join the Fleet Air Arm as a pilot"
},
{
"operation": "Now on",
"time": "26.64",
"value": "undefined"
},
{
"operation": "search",
"time": "43.669",
"value": "went to Queen's University at Kingston"
},
{
"operation": "Now on",
"time": "46.116",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "47.18",
"value": "Denis Keegan"
},
{
"operation": "Now on",
"time": "47.77",
"value": "Denis Keegan"
},
{
"operation": "Highlighting",
"time": "54.501",
"value": "Denis Keegan_sentence_0"
},
{
"operation": "Highlighting",
"time": "67.331",
"value": "Denis Keegan_sentence_5"
},
{
"operation": "Highlighting",
"time": "73.765",
"value": "Denis Keegan_sentence_7"
},
{
"operation": "finish",
"time": "105.539",
"value": "finish"
}
]
}
|
49066
|
In 1894, the first race was organised between Lévis and the Port of Quebec at the first Quebec Winter Carnival; while the Montreal races were held during the Fête des Neiges de Montréal from 1988 to 1992.
|
{
"evidence": [
"Ice canoeing_sentence_7",
"Ice canoeing_sentence_8"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nIn 1894, the first race was organised between Lévis and the Port of Quebec at the first Quebec Winter Carnival; while the Montreal races were held during the Fête des Neiges de Montréal from 1988 to 1992.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Ice canoeing_sentence_7",
"Ice canoeing_sentence_8"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8154,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.429",
"value": "?search="
},
{
"operation": "search",
"time": "22.696",
"value": "Quebec Winter Carnival"
},
{
"operation": "Now on",
"time": "24.63",
"value": "undefined"
},
{
"operation": "search",
"time": "64.332",
"value": "ice canoeing"
},
{
"operation": "Now on",
"time": "65.188",
"value": "Ice canoeing"
},
{
"operation": "Highlighting",
"time": "70.289",
"value": "Ice canoeing_sentence_7"
},
{
"operation": "Highlighting",
"time": "77.634",
"value": "Ice canoeing_sentence_8"
},
{
"operation": "finish",
"time": "84.951",
"value": "finish"
}
]
}
|
exfever_2357
|
Stardust is a critically acclaimed fantasy movie of 2007.
|
{
"evidence": [
"Stardust (2007 film)",
"Charlie Cox"
],
"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:\nStardust is a critically acclaimed fantasy movie of 2007.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Stardust (2007 film)",
"Charlie Cox"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 8155,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_71
|
Uta Thyra Hagen was a German-American actress and theatre practitioner who was on the Hollywood blacklist because of her association with an American bass-baritone concert artist, stage and film actor, athlete, and activist who became famous both for his cultural accomplishments and for his political stances.
|
{
"evidence": [
"Uta Hagen",
"Paul Robeson"
],
"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:\nUta Thyra Hagen was a German-American actress and theatre practitioner who was on the Hollywood blacklist because of her association with an American bass-baritone concert artist, stage and film actor, athlete, and activist who became famous both for his cultural accomplishments and for his political stances.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Uta Hagen",
"Paul Robeson"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8156,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
35467
|
Born in 1891, Bill Ladbury (an English professional boxer) had his first professional fight in November 1908 and in 1912 he had three fights against Joe Fox, winning two and losing the third.
|
{
"evidence": [
"Bill Ladbury_sentence_0",
"Bill Ladbury_sentence_2",
"Bill Ladbury_sentence_5"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nBorn in 1891, Bill Ladbury (an English professional boxer) had his first professional fight in November 1908 and in 1912 he had three fights against Joe Fox, winning two and losing the third.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Bill Ladbury_sentence_0",
"Bill Ladbury_sentence_2",
"Bill Ladbury_sentence_5"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8157,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.064",
"value": "?search="
},
{
"operation": "search",
"time": "4.89",
"value": "Bill Ladbury"
},
{
"operation": "Now on",
"time": "5.661",
"value": "Bill Ladbury"
},
{
"operation": "Highlighting",
"time": "8.681",
"value": "Bill Ladbury_sentence_0"
},
{
"operation": "Page search",
"time": "17.629",
"value": "1908"
},
{
"operation": "Highlighting",
"time": "23.311",
"value": "Bill Ladbury_sentence_2"
},
{
"operation": "Highlighting",
"time": "33.606",
"value": "Bill Ladbury_sentence_5"
},
{
"operation": "finish",
"time": "95.736",
"value": "finish"
}
]
}
|
exfever_3366
|
The human condition is a very broad topic that has been and continues to be pondered and analyzed from many perspectives, including any collection of written work, but it is also used more narrowly to refer to a diverse range of human activities and resulting products that involve creativity.
|
{
"evidence": [
"Human condition",
"Literature",
"Art"
],
"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 human condition is a very broad topic that has been and continues to be pondered and analyzed from many perspectives, including any collection of written work, but it is also used more narrowly to refer to a diverse range of human activities and resulting products that involve creativity.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Human condition",
"Literature",
"Art"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8158,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_6866
|
Sofia Coppola is an American filmmaker and actress who is the youngest child and only daughter of an Afghan film director, producer, and screenwriter.
|
{
"evidence": [
"Sofia Coppola",
"Francis Ford Coppola"
],
"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:\nSofia Coppola is an American filmmaker and actress who is the youngest child and only daughter of an Afghan film director, producer, and screenwriter.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Sofia Coppola",
"Francis Ford Coppola"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 8159,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_4231
|
Michael Edward Love was an American singer and songwriter who co-founded One Direction with his cousin, best remembered as their drummer.
|
{
"evidence": [
"Mike Love",
"Dennis Wilson"
],
"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:\nMichael Edward Love was an American singer and songwriter who co-founded One Direction with his cousin, best remembered as their drummer.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Mike Love",
"Dennis Wilson"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 8160,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
45684
|
Riblja Čorba guitarist, Bora Đorđević, not only started performing and composing as a teenager; but also sang and played rhythm guitar in his early bands.
|
{
"evidence": [
"Riblja Čorba_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:\nRiblja Čorba guitarist, Bora Đorđević, not only started performing and composing as a teenager; but also sang and played rhythm guitar in his early bands.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Riblja Čorba_sentence_5"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 8161,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.705",
"value": "?search="
},
{
"operation": "search",
"time": "23.887",
"value": "Bora Đorđević"
},
{
"operation": "Now on",
"time": "24.817",
"value": "Bora Đorđević"
},
{
"operation": "Highlighting",
"time": "48.009",
"value": "Bora Đorđević_sentence_2"
},
{
"operation": "Highlighting",
"time": "48.01",
"value": "Bora Đorđević_title"
},
{
"operation": "Page search",
"time": "52.032",
"value": "guitarist"
},
{
"operation": "Page search",
"time": "86.812",
"value": "guitar"
},
{
"operation": "hyperlink",
"time": "131.502",
"value": "Riblja Čorba"
},
{
"operation": "Now on",
"time": "132.777",
"value": "Riblja Čorba"
},
{
"operation": "Highlighting",
"time": "139.536",
"value": "Riblja Čorba_sentence_5"
},
{
"operation": "finish",
"time": "151.172",
"value": "finish"
}
]
}
|
52034
|
The XMPP Core and XMPP IM both are not supported by openfire.
|
{
"evidence": [
"Openfire_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 XMPP Core and XMPP IM both are not supported by openfire.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Openfire_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 8162,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.796",
"value": "?search="
},
{
"operation": "search",
"time": "16.093",
"value": "xMPP Core"
},
{
"operation": "Now on",
"time": "17.212",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "33.836",
"value": "XMPP"
},
{
"operation": "Now on",
"time": "34.645",
"value": "XMPP"
},
{
"operation": "Page search",
"time": "54.166",
"value": "xmpp im"
},
{
"operation": "Page search",
"time": "74.456",
"value": "openfire"
},
{
"operation": "search",
"time": "93.717",
"value": "Openfire"
},
{
"operation": "Now on",
"time": "94.501",
"value": "Openfire"
},
{
"operation": "Page search",
"time": "100.061",
"value": "xmpp"
},
{
"operation": "Page search",
"time": "112.331",
"value": "xmpp core"
},
{
"operation": "Page search",
"time": "130.703",
"value": "xmpp core im"
},
{
"operation": "Page search",
"time": "145.465",
"value": "xmpp im"
},
{
"operation": "Highlighting",
"time": "171.153",
"value": "Openfire_sentence_0"
},
{
"operation": "hyperlink",
"time": "178.357",
"value": "Extensible Messaging and Presence Protocol"
},
{
"operation": "Now on",
"time": "179.174",
"value": "XMPP"
},
{
"operation": "Page search",
"time": "183.114",
"value": "core"
},
{
"operation": "Page search",
"time": "195.176",
"value": "openfire"
},
{
"operation": "Page search",
"time": "208.653",
"value": "core"
},
{
"operation": "Page search",
"time": "234.247",
"value": "openfire"
},
{
"operation": "search",
"time": "239.587",
"value": "openfire"
},
{
"operation": "Now on",
"time": "240.221",
"value": "Openfire"
},
{
"operation": "finish",
"time": "259.012",
"value": "finish"
}
]
}
|
exfever_2170
|
Room is a 2015 film directed by Lenny Abrahamson whose a category of narrative fiction intended to be more serious than humorous in tone.
|
{
"evidence": [
"Room (2015 film)",
"Drama (film and television)"
],
"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:\nRoom is a 2015 film directed by Lenny Abrahamson whose a category of narrative fiction intended to be more serious than humorous in tone.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Room (2015 film)",
"Drama (film and television)"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8163,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_5365
|
Denial is a 2016 biographical film based on the 2005 book History on Trial: My Day in Court with a Holocaust Denier by an American History and Holocaust Studies Professor at Emory University in a state in the Southeastern region of the United States.
|
{
"evidence": [
"Denial (2016 film)",
"Deborah Lipstadt",
"Georgia (U.S. state)"
],
"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:\nDenial is a 2016 biographical film based on the 2005 book History on Trial: My Day in Court with a Holocaust Denier by an American History and Holocaust Studies Professor at Emory University in a state in the Southeastern region of the United States.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Denial (2016 film)",
"Deborah Lipstadt",
"Georgia (U.S. state)"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8164,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
39806
|
Walter Folger Brown was a successful political figure in the 1930s from Toledo, Ohio.
|
{
"evidence": [
"Walter Folger Brown_sentence_6"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nWalter Folger Brown was a successful political figure in the 1930s from Toledo, Ohio.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Walter Folger Brown_sentence_6"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 8165,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.685",
"value": "?search="
},
{
"operation": "search",
"time": "7.853",
"value": "Walter Folger Brown"
},
{
"operation": "Now on",
"time": "8.51",
"value": "Walter Folger Brown"
},
{
"operation": "Highlighting",
"time": "115.454",
"value": "Walter Folger Brown_sentence_3"
},
{
"operation": "Page search",
"time": "140.837",
"value": "succ"
},
{
"operation": "Highlighting",
"time": "151.359",
"value": "Walter Folger Brown_sentence_6"
},
{
"operation": "Highlighting deleted",
"time": "152.944",
"value": "Walter Folger Brown_sentence_3"
},
{
"operation": "finish",
"time": "164.887",
"value": "finish"
}
]
}
|
54498
|
The 1996 National Football League season was the Seattle Seahawks' 15th season to participate; then the team was rented to Paul Walker in 2005.
|
{
"evidence": [
"1996 NFL season_sentence_0"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe 1996 National Football League season was the Seattle Seahawks' 15th season to participate; then the team was rented to Paul Walker in 2005.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"1996 NFL season_sentence_0"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 8166,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.381",
"value": "?search="
},
{
"operation": "search",
"time": "6.979",
"value": "1996 National Football League"
},
{
"operation": "Now on",
"time": "9.615",
"value": "undefined"
},
{
"operation": "search",
"time": "73.832",
"value": "1996 Seattle Seahawks"
},
{
"operation": "Now on",
"time": "76.32",
"value": "1996 Seattle Seahawks season"
},
{
"operation": "search",
"time": "105.991",
"value": "Seattle Seahawks"
},
{
"operation": "Now on",
"time": "114.193",
"value": "1996 Seattle Seahawks season"
},
{
"operation": "search",
"time": "135.566",
"value": "Seattle Seahawks"
},
{
"operation": "Now on",
"time": "137.548",
"value": "Seattle Seahawks"
},
{
"operation": "Page search",
"time": "141.768",
"value": "1996"
},
{
"operation": "Page search",
"time": "190.788",
"value": "paul walker"
},
{
"operation": "Page search",
"time": "207.556",
"value": "rented"
},
{
"operation": "Page search",
"time": "219.721",
"value": "walker"
},
{
"operation": "search",
"time": "230.865",
"value": "team was rented to Paul Walker in 2005"
},
{
"operation": "Now on",
"time": "233.56",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "244.703",
"value": "H. G. Wells"
},
{
"operation": "Now on",
"time": "246.347",
"value": "H. G. Wells"
},
{
"operation": "search",
"time": "295.631",
"value": "Contains... Seattle Seahawks' 15th season to participate"
},
{
"operation": "Now on",
"time": "297.811",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "304.325",
"value": "Week 5"
},
{
"operation": "Now on",
"time": "314.801",
"value": "2011 New York Giants season#Week 5"
},
{
"operation": "search",
"time": "334.573",
"value": "Seattle seahawks seasons"
},
{
"operation": "Now on",
"time": "337.496",
"value": "List of Seattle Seahawks seasons"
},
{
"operation": "Page search",
"time": "343.616",
"value": "15th"
},
{
"operation": "hyperlink",
"time": "350.006",
"value": "History of the Seattle Seahawks"
},
{
"operation": "Now on",
"time": "352.521",
"value": "History of the Seattle Seahawks"
},
{
"operation": "Page search",
"time": "359.538",
"value": "15th season"
},
{
"operation": "Page search",
"time": "375.421",
"value": "walker"
},
{
"operation": "Highlighting",
"time": "386.299",
"value": "History of the Seattle Seahawks_sentence_182"
},
{
"operation": "Highlighting",
"time": "386.301",
"value": "History of the Seattle Seahawks_title"
},
{
"operation": "Highlighting",
"time": "386.304",
"value": "History of the Seattle Seahawks_section_9"
},
{
"operation": "Highlighting",
"time": "386.304",
"value": "History of the Seattle Seahawks_section_14"
},
{
"operation": "Highlighting deleted",
"time": "387.116",
"value": "History of the Seattle Seahawks_sentence_182"
},
{
"operation": "Highlighting deleted",
"time": "387.119",
"value": "History of the Seattle Seahawks_title"
},
{
"operation": "Highlighting deleted",
"time": "387.122",
"value": "History of the Seattle Seahawks_section_9"
},
{
"operation": "Highlighting deleted",
"time": "387.124",
"value": "History of the Seattle Seahawks_section_14"
},
{
"operation": "search",
"time": "396.622",
"value": "1996 National Football League season"
},
{
"operation": "Now on",
"time": "400.328",
"value": "1996 NFL season"
},
{
"operation": "Highlighting",
"time": "425.203",
"value": "1996 NFL season_sentence_0"
},
{
"operation": "finish",
"time": "430.178",
"value": "finish"
}
]
}
|
exfever_1448
|
Loni Love is a former electrical professional who invent, design, analyze, build and test a physical system using power to apply forces and control movement to perform an action.
|
{
"evidence": [
"Loni Love",
"Engineer",
"Machine"
],
"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:\nLoni Love is a former electrical professional who invent, design, analyze, build and test a physical system using power to apply forces and control movement to perform an action.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Loni Love",
"Engineer",
"Machine"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8167,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_8900
|
T-Pain first popularized the creative use of the Auto-Tune pitch correction effect.
|
{
"evidence": [
"T-Pain",
"Travis Scott"
],
"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:\nT-Pain first popularized the creative use of the Auto-Tune pitch correction effect.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"T-Pain",
"Travis Scott"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 8168,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
77143
|
Among the many defunct airlines of Lithuania, Air Lithuania commenced operations in 1996 and ceased them in 2005, while Amber Air commenced operations in 2004 and ended them in 2007.
|
{
"evidence": [
"Air Lithuania_sentence_3",
"Air Lithuania_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:\nAmong the many defunct airlines of Lithuania, Air Lithuania commenced operations in 1996 and ceased them in 2005, while Amber Air commenced operations in 2004 and ended them in 2007.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Air Lithuania_sentence_3",
"Air Lithuania_sentence_4"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 8169,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.736",
"value": "?search="
},
{
"operation": "search",
"time": "20.162",
"value": "Air Lithuania"
},
{
"operation": "Now on",
"time": "21.074",
"value": "Air Lithuania"
},
{
"operation": "Highlighting",
"time": "77.325",
"value": "Air Lithuania_sentence_4"
},
{
"operation": "Highlighting",
"time": "103.689",
"value": "Air Lithuania_sentence_3"
},
{
"operation": "Highlighting deleted",
"time": "105.373",
"value": "Air Lithuania_sentence_4"
},
{
"operation": "Highlighting",
"time": "105.959",
"value": "Air Lithuania_sentence_4"
},
{
"operation": "finish",
"time": "127.975",
"value": "finish"
}
]
}
|
exfever_1316
|
Stephen John Coogan is the co-founder of the production company Baby Cow Productions and co-creator of the comedy character who is a tasteless and inept right-wing broadcaster.
|
{
"evidence": [
"Steve Coogan",
"Alan Partridge"
],
"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:\nStephen John Coogan is the co-founder of the production company Baby Cow Productions and co-creator of the comedy character who is a tasteless and inept right-wing broadcaster.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Steve Coogan",
"Alan Partridge"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8170,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
38726
|
Martin Suckling, a British composer born in November 23, 1981, studied in the Royal Academy of Music and undertook doctoral research at the Yale School of Music.
|
{
"evidence": [
"Martin Suckling_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:\nMartin Suckling, a British composer born in November 23, 1981, studied in the Royal Academy of Music and undertook doctoral research at the Yale School of Music.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Martin Suckling_sentence_5"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 8171,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.341",
"value": "?search="
},
{
"operation": "search",
"time": "116.491",
"value": "Martin Suckling"
},
{
"operation": "Now on",
"time": "117.341",
"value": "Martin Suckling"
},
{
"operation": "Highlighting",
"time": "119.318",
"value": "Martin Suckling_sentence_0"
},
{
"operation": "Highlighting",
"time": "138.734",
"value": "Martin Suckling_sentence_5"
},
{
"operation": "Page search",
"time": "149.309",
"value": "yale"
},
{
"operation": "Highlighting",
"time": "154.666",
"value": "Martin Suckling_sentence_4"
},
{
"operation": "Highlighting deleted",
"time": "163.259",
"value": "Martin Suckling_sentence_4"
},
{
"operation": "Page search",
"time": "173.447",
"value": "yale"
},
{
"operation": "Highlighting deleted",
"time": "186.509",
"value": "Martin Suckling_sentence_0"
},
{
"operation": "finish",
"time": "192.899",
"value": "finish"
}
]
}
|
exfever_4882
|
American Gods is a fantasy novel by an Italian author of comic books who has won a literary award given by the Association for Library Service to Children to the author of "the most distinguished contribution to American literature for children".
|
{
"evidence": [
"American Gods",
"Neil Gaiman",
"Newbery Medal"
],
"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:\nAmerican Gods is a fantasy novel by an Italian author of comic books who has won a literary award given by the Association for Library Service to Children to the author of \"the most distinguished contribution to American literature for children\".\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"American Gods",
"Neil Gaiman",
"Newbery Medal"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 8172,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
48184
|
The Landscape Agency was founded by James in October 1998 after teaming up with Tom Stuart-Smith and Todd Longstaffe-Gowan and James grew the business by seeking additional start-up capital after a year.
|
{
"evidence": [
"The Landscape Agency_sentence_6"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Landscape Agency was founded by James in October 1998 after teaming up with Tom Stuart-Smith and Todd Longstaffe-Gowan and James grew the business by seeking additional start-up capital after a year.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"The Landscape Agency_sentence_6"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 8173,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.064",
"value": "?search="
},
{
"operation": "search",
"time": "17.943",
"value": "The Landscape Agency"
},
{
"operation": "Now on",
"time": "18.684",
"value": "The Landscape Agency"
},
{
"operation": "Highlighting",
"time": "28.021",
"value": "The Landscape Agency_table_caption_0"
},
{
"operation": "Highlighting",
"time": "29.491",
"value": "The Landscape Agency_header_cell_0_1_0"
},
{
"operation": "Highlighting",
"time": "30.211",
"value": "The Landscape Agency_cell_0_1_1"
},
{
"operation": "Highlighting",
"time": "30.215",
"value": "The Landscape Agency_header_cell_0_1_0"
},
{
"operation": "Highlighting",
"time": "32.035",
"value": "The Landscape Agency_header_cell_0_2_0"
},
{
"operation": "Highlighting",
"time": "32.963",
"value": "The Landscape Agency_cell_0_2_1"
},
{
"operation": "Highlighting",
"time": "32.967",
"value": "The Landscape Agency_header_cell_0_2_0"
},
{
"operation": "Page search",
"time": "42.895",
"value": "Tom"
},
{
"operation": "Highlighting",
"time": "46.027",
"value": "The Landscape Agency_sentence_5"
},
{
"operation": "Highlighting",
"time": "61.266",
"value": "The Landscape Agency_sentence_5"
},
{
"operation": "Highlighting",
"time": "74.914",
"value": "The Landscape Agency_sentence_6"
},
{
"operation": "Highlighting deleted",
"time": "88.114",
"value": "The Landscape Agency_sentence_5"
},
{
"operation": "finish",
"time": "117.522",
"value": "finish"
}
]
}
|
93122
|
Henry of Badewide (died ca. 1164) was a Saxon Count of Botwide and Count of Ratzeburg who came from a knightly family from Lüneburg and feuded with Count Adolf II of Schauenburg over Holstein and Wagria
|
{
"evidence": [
"Henry of Badewide_sentence_0",
"Henry of Badewide_sentence_1",
"Henry of Badewide_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:\nHenry of Badewide (died ca. 1164) was a Saxon Count of Botwide and Count of Ratzeburg who came from a knightly family from Lüneburg and feuded with Count Adolf II of Schauenburg over Holstein and Wagria\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"Henry of Badewide_sentence_0",
"Henry of Badewide_sentence_1",
"Henry of Badewide_sentence_10"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8174,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.096",
"value": "?search="
},
{
"operation": "search",
"time": "10.673",
"value": "Henry of Badewide"
},
{
"operation": "Now on",
"time": "11.915",
"value": "Henry of Badewide"
},
{
"operation": "Highlighting",
"time": "20.856",
"value": "Henry of Badewide_sentence_0"
},
{
"operation": "Highlighting",
"time": "22.48",
"value": "Henry of Badewide_sentence_1"
},
{
"operation": "Highlighting",
"time": "36.712",
"value": "Henry of Badewide_sentence_10"
},
{
"operation": "finish",
"time": "46.945",
"value": "finish"
}
]
}
|
exfever_5437
|
The 2002 Royal Rumble was the 15th annual Royal Rumble, which is a performance art of scripted wrestling matches, often attached to a wrestler's in-ring persona that is artificially created in order to draw fan interest.?
|
{
"evidence": [
"Royal Rumble (2002)",
"Professional wrestling",
"Gimmick (professional wrestling)"
],
"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 2002 Royal Rumble was the 15th annual Royal Rumble, which is a performance art of scripted wrestling matches, often attached to a wrestler's in-ring persona that is artificially created in order to draw fan interest.?\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Royal Rumble (2002)",
"Professional wrestling",
"Gimmick (professional wrestling)"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8175,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
73222
|
Brewer, born in 1810, in Norwich to a Baptist schoolmaster, was matriculated at Queen's College, Oxford in 1827, before being ordained in the Church of England in 1837.
|
{
"evidence": [
"John Sherren Brewer_sentence_0",
"John Sherren Brewer_sentence_3",
"John Sherren Brewer_sentence_4",
"John Sherren Brewer_sentence_5"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nBrewer, born in 1810, in Norwich to a Baptist schoolmaster, was matriculated at Queen's College, Oxford in 1827, before being ordained in the Church of England in 1837.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"John Sherren Brewer_sentence_0",
"John Sherren Brewer_sentence_3",
"John Sherren Brewer_sentence_4",
"John Sherren Brewer_sentence_5"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8176,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.889",
"value": "?search="
},
{
"operation": "search",
"time": "13.743",
"value": "Brewer"
},
{
"operation": "Now on",
"time": "16.448",
"value": "Brewing"
},
{
"operation": "hyperlink",
"time": "85.467",
"value": "Brewer (disambiguation)"
},
{
"operation": "Now on",
"time": "86.028",
"value": "Brewer (disambiguation)"
},
{
"operation": "hyperlink",
"time": "102.897",
"value": "Brewer (surname)"
},
{
"operation": "Now on",
"time": "103.42",
"value": "Brewer (surname)"
},
{
"operation": "Page search",
"time": "112.826",
"value": "1810"
},
{
"operation": "hyperlink",
"time": "130.81",
"value": "John Sherren Brewer"
},
{
"operation": "Now on",
"time": "131.376",
"value": "John Sherren Brewer"
},
{
"operation": "Highlighting",
"time": "212.248",
"value": "John Sherren Brewer_sentence_0"
},
{
"operation": "Highlighting",
"time": "216.405",
"value": "John Sherren Brewer_sentence_3"
},
{
"operation": "Highlighting",
"time": "222.117",
"value": "John Sherren Brewer_sentence_4"
},
{
"operation": "Highlighting",
"time": "226.373",
"value": "John Sherren Brewer_sentence_5"
},
{
"operation": "finish",
"time": "240.068",
"value": "finish"
}
]
}
|
exfever_4778
|
The Armenian genocide was the systematic destruction of the Armenian people in the empire that controlled much of Southeast Europe.
|
{
"evidence": [
"Armenian Genocide",
"Ottoman Empire"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Armenian genocide was the systematic destruction of the Armenian people in the empire that controlled much of Southeast Europe.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Armenian Genocide",
"Ottoman Empire"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8177,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
16330
|
Major General Jim Muhwezi (born August 23, 1950 in Uganda) who is currently Uganda's Minister of Information & National Guidance, graduated from Makerere University with a degree of Bachelor of Laws, and Tanzania Police Academy where he finished the Police Cadet Course.
|
{
"evidence": [
"Jim Muhwezi_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:\nMajor General Jim Muhwezi (born August 23, 1950 in Uganda) who is currently Uganda's Minister of Information & National Guidance, graduated from Makerere University with a degree of Bachelor of Laws, and Tanzania Police Academy where he finished the Police Cadet Course.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Jim Muhwezi_sentence_1"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 8178,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.639",
"value": "?search="
},
{
"operation": "search",
"time": "20.973",
"value": "Major General Jim Muhwezi"
},
{
"operation": "Now on",
"time": "22.926",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "25.834",
"value": "Jim Muhwezi"
},
{
"operation": "Now on",
"time": "26.506",
"value": "Jim Muhwezi"
},
{
"operation": "Highlighting",
"time": "29.46",
"value": "Jim Muhwezi_sentence_0"
},
{
"operation": "Highlighting deleted",
"time": "40.779",
"value": "Jim Muhwezi_sentence_0"
},
{
"operation": "Highlighting",
"time": "55.023",
"value": "Jim Muhwezi_sentence_1"
},
{
"operation": "finish",
"time": "79.123",
"value": "finish"
}
]
}
|
43995
|
At the age of 18, a year before her dad died, Blaire White was a victim of date rape by a man, which she later revealed in a video collaboration with Shane Dawson.
|
{
"evidence": [
"Blaire White_sentence_5",
"Blaire White_sentence_6"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nAt the age of 18, a year before her dad died, Blaire White was a victim of date rape by a man, which she later revealed in a video collaboration with Shane Dawson.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Blaire White_sentence_5",
"Blaire White_sentence_6"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8179,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.712",
"value": "?search="
},
{
"operation": "search",
"time": "22.805",
"value": "Blaire White"
},
{
"operation": "Now on",
"time": "23.884",
"value": "Blaire White"
},
{
"operation": "Highlighting",
"time": "41.21",
"value": "Blaire White_sentence_5"
},
{
"operation": "Highlighting",
"time": "59.201",
"value": "Blaire White_sentence_6"
},
{
"operation": "finish",
"time": "440.06",
"value": "finish"
}
]
}
|
exfever_5794
|
Wolfgang Amadeus Mozart was the first prolific and influential composer of the Classical period.
|
{
"evidence": [
"Wolfgang Amadeus Mozart",
"Choir"
],
"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:\nWolfgang Amadeus Mozart was the first prolific and influential composer of the Classical period.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Wolfgang Amadeus Mozart",
"Choir"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 8180,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_2043
|
Elizabeth Hughes Meriwether is known for creating an Asian television sitcom for the Fox Broadcasting Company that stars an American actor, comedian, film producer, and screenwriter.
|
{
"evidence": [
"Elizabeth Meriwether",
"New Girl",
"Jake Johnson"
],
"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:\nElizabeth Hughes Meriwether is known for creating an Asian television sitcom for the Fox Broadcasting Company that stars an American actor, comedian, film producer, and screenwriter.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Elizabeth Meriwether",
"New Girl",
"Jake Johnson"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 8181,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
73593
|
During the 1938-39 Lancashire Cup Final, Jim Sullivan of Wigan and Harold Osbaldestin of Salford both played the fullback position for their respective teams.
|
{
"evidence": [
"Jim Sullivan (Welsh rugby league)_sentence_17",
"Harold Osbaldestin_sentence_6"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nDuring the 1938-39 Lancashire Cup Final, Jim Sullivan of Wigan and Harold Osbaldestin of Salford both played the fullback position for their respective teams.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Jim Sullivan (Welsh rugby league)_sentence_17",
"Harold Osbaldestin_sentence_6"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8182,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "78.603",
"value": "undefined"
},
{
"operation": "search",
"time": "126.831",
"value": "1938-39 Lancashire Cup"
},
{
"operation": "Now on",
"time": "132.346",
"value": "1938-39 Lancashire Cup"
},
{
"operation": "Highlighting",
"time": "140.306",
"value": "1938–39 Lancashire Cup_cell_1_3_0"
},
{
"operation": "Highlighting",
"time": "140.308",
"value": "1938–39 Lancashire Cup_title"
},
{
"operation": "Highlighting",
"time": "140.309",
"value": "1938–39 Lancashire Cup_section_1"
},
{
"operation": "Highlighting",
"time": "140.31",
"value": "1938–39 Lancashire Cup_section_5"
},
{
"operation": "Highlighting",
"time": "140.31",
"value": "1938–39 Lancashire Cup_section_6"
},
{
"operation": "Highlighting",
"time": "140.314",
"value": "1938–39 Lancashire Cup_header_cell_1_0_0"
},
{
"operation": "Highlighting",
"time": "147.485",
"value": "1938–39 Lancashire Cup_cell_1_3_2"
},
{
"operation": "Highlighting",
"time": "147.489",
"value": "1938–39 Lancashire Cup_header_cell_1_0_2"
},
{
"operation": "hyperlink",
"time": "164.86",
"value": "Jim Sullivan"
},
{
"operation": "Now on",
"time": "166.849",
"value": "Jim Sullivan (Welsh rugby league)"
},
{
"operation": "Highlighting",
"time": "175.908",
"value": "Jim Sullivan (Welsh rugby league)_sentence_2"
},
{
"operation": "Highlighting deleted",
"time": "179.346",
"value": "Jim Sullivan (Welsh rugby league)_sentence_2"
},
{
"operation": "hyperlink",
"time": "201.526",
"value": "1938–39 Lancashire Cup"
},
{
"operation": "Now on",
"time": "202.462",
"value": "1938–39 Lancashire Cup"
},
{
"operation": "hyperlink",
"time": "210.588",
"value": "1938–39 Northern Rugby Football League season"
},
{
"operation": "Now on",
"time": "211.442",
"value": "1938–39 Northern Rugby Football League season"
},
{
"operation": "back-button-clicked",
"time": "216.943",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "217.062",
"value": "1938–39 Lancashire Cup"
},
{
"operation": "hyperlink",
"time": "228.449",
"value": "Jim Sullivan"
},
{
"operation": "Now on",
"time": "228.87",
"value": "Jim Sullivan (Welsh rugby league)"
},
{
"operation": "Highlighting",
"time": "243.513",
"value": "Jim Sullivan (Welsh rugby league)_sentence_17"
},
{
"operation": "back-button-clicked",
"time": "257.707",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "257.845",
"value": "1938–39 Lancashire Cup"
},
{
"operation": "hyperlink",
"time": "262.697",
"value": "Harold Osbaldestin"
},
{
"operation": "Now on",
"time": "263.297",
"value": "Harold Osbaldestin"
},
{
"operation": "Highlighting",
"time": "278.909",
"value": "Harold Osbaldestin_sentence_6"
},
{
"operation": "finish",
"time": "317.503",
"value": "finish"
}
]
}
|
61200
|
The 1999 Anguillian general election was held on the fourth of March and the opposition Anguilla National Alliance won three seats.
|
{
"evidence": [
"1999 Anguillian general election_sentence_0",
"1999 Anguillian general election_sentence_2"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe 1999 Anguillian general election was held on the fourth of March and the opposition Anguilla National Alliance won three seats.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"1999 Anguillian general election_sentence_0",
"1999 Anguillian general election_sentence_2"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 8183,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.636",
"value": "?search="
},
{
"operation": "search",
"time": "10.159",
"value": "1999 Anguillian general election"
},
{
"operation": "Now on",
"time": "11.325",
"value": "1999 Anguillian general election"
},
{
"operation": "Highlighting",
"time": "41.409",
"value": "1999 Anguillian general election_sentence_0"
},
{
"operation": "Highlighting",
"time": "45.855",
"value": "1999 Anguillian general election_sentence_2"
},
{
"operation": "finish",
"time": "96.499",
"value": "finish"
}
]
}
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.