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 |
|---|---|---|---|---|---|---|---|---|
21607
|
Odisha FC Reserves and Academy was first headed by Francisco Perez Lazaro as coach and the 2017–18 I-League 2nd Division, which commenced on March 14, 2018, was their first and last season in the league.
|
{
"evidence": [
"Odisha FC Reserves and Academy_sentence_4",
"Odisha FC Reserves and Academy_sentence_5",
"2017–18 I-League 2nd Division_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:\nOdisha FC Reserves and Academy was first headed by Francisco Perez Lazaro as coach and the 2017–18 I-League 2nd Division, which commenced on March 14, 2018, was their first and last season in the league.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Odisha FC Reserves and Academy_sentence_4",
"Odisha FC Reserves and Academy_sentence_5",
"2017–18 I-League 2nd Division_sentence_1"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5084,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.642",
"value": "?search="
},
{
"operation": "search",
"time": "10.181",
"value": "Odisha FC Reserves and Academy"
},
{
"operation": "Now on",
"time": "11.09",
"value": "Odisha FC Reserves and Academy"
},
{
"operation": "Highlighting",
"time": "33.225",
"value": "Odisha FC Reserves and Academy_sentence_4"
},
{
"operation": "Highlighting",
"time": "50.744",
"value": "Odisha FC Reserves and Academy_sentence_5"
},
{
"operation": "hyperlink",
"time": "53.473",
"value": "2017–18 I-League 2nd Division"
},
{
"operation": "Now on",
"time": "54.812",
"value": "2017–18 I-League 2nd Division"
},
{
"operation": "Highlighting",
"time": "60.335",
"value": "2017–18 I-League 2nd Division_sentence_1"
},
{
"operation": "finish",
"time": "84.291",
"value": "finish"
}
]
}
|
68932
|
William Polson initiated the Meat Export Control Act, he had married Florence Ada and clashed with the reform government which was established in 1912.
|
{
"evidence": [
"William Polson_sentence_8",
"William Polson_sentence_9",
"William Polson_sentence_7",
"Reform Government of New Zealand_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:\nWilliam Polson initiated the Meat Export Control Act, he had married Florence Ada and clashed with the reform government which was established in 1912.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"William Polson_sentence_8",
"William Polson_sentence_9",
"William Polson_sentence_7",
"Reform Government of New Zealand_sentence_0"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5085,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.955",
"value": "undefined"
},
{
"operation": "search",
"time": "19.587",
"value": "William Polson"
},
{
"operation": "Now on",
"time": "21.279",
"value": "William Polson"
},
{
"operation": "Highlighting",
"time": "33.912",
"value": "William Polson_sentence_8"
},
{
"operation": "Highlighting",
"time": "47.546",
"value": "William Polson_sentence_9"
},
{
"operation": "Highlighting",
"time": "80.916",
"value": "William Polson_sentence_7"
},
{
"operation": "hyperlink",
"time": "83.008",
"value": "Reform government"
},
{
"operation": "Now on",
"time": "98.721",
"value": "William Polson"
},
{
"operation": "hyperlink",
"time": "105.433",
"value": "Reform government"
},
{
"operation": "Now on",
"time": "116.968",
"value": "William Polson"
},
{
"operation": "hyperlink",
"time": "120.185",
"value": "Reform government"
},
{
"operation": "Now on",
"time": "122.857",
"value": "Reform Government of New Zealand"
},
{
"operation": "Highlighting",
"time": "125.731",
"value": "Reform Government of New Zealand_sentence_0"
},
{
"operation": "finish",
"time": "144.821",
"value": "finish"
}
]
}
|
exfever_2154
|
21 Jump Street is an adaptation of the 1987–1991 television series by an American television producer whose credits include an American comedy-drama superhero television series that aired on ABC.
|
{
"evidence": [
"21 Jump Street (film)",
"Stephen J. Cannell",
"The Greatest American Hero"
],
"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:\n21 Jump Street is an adaptation of the 1987–1991 television series by an American television producer whose credits include an American comedy-drama superhero television series that aired on ABC.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"21 Jump Street (film)",
"Stephen J. Cannell",
"The Greatest American Hero"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5086,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_5109
|
Middle-earth is the fictional setting of much of J. R. R. Tolkien's fantasy, which term is equivalent to the old Filipino Middangeard, inhabited by and known to humans in early Germanic cosmology.
|
{
"evidence": [
"Middle-earth",
"Midgard"
],
"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:\nMiddle-earth is the fictional setting of much of J. R. R. Tolkien's fantasy, which term is equivalent to the old Filipino Middangeard, inhabited by and known to humans in early Germanic cosmology.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Middle-earth",
"Midgard"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5087,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7671
|
A Perfect Circle is an American rock band formed in a city extending through the coastal mountain range next to the Atlantic Ocean.
|
{
"evidence": [
"A Perfect Circle",
"Los Angeles",
"Santa Monica Mountains"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nA Perfect Circle is an American rock band formed in a city extending through the coastal mountain range next to the Atlantic Ocean.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"A Perfect Circle",
"Los Angeles",
"Santa Monica Mountains"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5088,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
52373
|
Art Griffiths was a World War II pilot and a professional rower.
|
{
"evidence": [
"Art Griffiths_sentence_3",
"Art Griffiths_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:\nArt Griffiths was a World War II pilot and a professional rower.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Art Griffiths_sentence_3",
"Art Griffiths_sentence_0"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5089,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.885",
"value": "?search="
},
{
"operation": "search",
"time": "12.351",
"value": "Art Griffiths"
},
{
"operation": "Now on",
"time": "12.989",
"value": "Art Griffiths"
},
{
"operation": "Highlighting",
"time": "23.082",
"value": "Art Griffiths_sentence_3"
},
{
"operation": "Highlighting",
"time": "27.212",
"value": "Art Griffiths_sentence_0"
},
{
"operation": "finish",
"time": "47.099",
"value": "finish"
}
]
}
|
48223
|
The Quadrangular tower or Keep of the Fortress of Baku, is a 16 m historic tower located on Kichik Gala street, in the city of Baku, in Katex, Azerbaijan.
|
{
"evidence": [
"Katex_sentence_1",
"Baku_sentence_1"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Quadrangular tower or Keep of the Fortress of Baku, is a 16 m historic tower located on Kichik Gala street, in the city of Baku, in Katex, Azerbaijan.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Katex_sentence_1",
"Baku_sentence_1"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5090,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.775",
"value": "?search="
},
{
"operation": "search",
"time": "19.638",
"value": "fortress of baku"
},
{
"operation": "Now on",
"time": "21.484",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "25.571",
"value": "Maiden Tower (Baku)"
},
{
"operation": "Now on",
"time": "26.644",
"value": "Maiden Tower (Baku)"
},
{
"operation": "hyperlink",
"time": "43.04",
"value": "Baku"
},
{
"operation": "Now on",
"time": "45.604",
"value": "Baku"
},
{
"operation": "back-button-clicked",
"time": "48.456",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "48.575",
"value": "Maiden Tower (Baku)"
},
{
"operation": "search",
"time": "73.697",
"value": "Quadrangular Tower"
},
{
"operation": "Now on",
"time": "74.798",
"value": "Quadrangular Tower"
},
{
"operation": "Highlighting",
"time": "77.241",
"value": "Quadrangular Tower_sentence_0"
},
{
"operation": "Highlighting",
"time": "77.243",
"value": "Quadrangular Tower_title"
},
{
"operation": "Highlighting deleted",
"time": "79.495",
"value": "Quadrangular Tower_sentence_0"
},
{
"operation": "Highlighting deleted",
"time": "79.497",
"value": "Quadrangular Tower_title"
},
{
"operation": "Highlighting",
"time": "80.514",
"value": "Quadrangular Tower_sentence_0"
},
{
"operation": "Highlighting",
"time": "80.515",
"value": "Quadrangular Tower_title"
},
{
"operation": "Highlighting deleted",
"time": "81.115",
"value": "Quadrangular Tower_sentence_0"
},
{
"operation": "Highlighting deleted",
"time": "81.117",
"value": "Quadrangular Tower_title"
},
{
"operation": "Highlighting",
"time": "81.74",
"value": "Quadrangular Tower_sentence_0"
},
{
"operation": "Highlighting",
"time": "81.741",
"value": "Quadrangular Tower_title"
},
{
"operation": "Highlighting",
"time": "89.553",
"value": "Quadrangular Tower_sentence_2"
},
{
"operation": "Highlighting",
"time": "106.471",
"value": "Quadrangular Tower_header_cell_0_12_0"
},
{
"operation": "Highlighting",
"time": "107.158",
"value": "Quadrangular Tower_cell_0_12_1"
},
{
"operation": "Highlighting",
"time": "107.171",
"value": "Quadrangular Tower_header_cell_0_12_0"
},
{
"operation": "Highlighting",
"time": "107.171",
"value": "Quadrangular Tower_header_cell_0_2_0"
},
{
"operation": "search",
"time": "113.023",
"value": "katex"
},
{
"operation": "Now on",
"time": "113.92",
"value": "Katex"
},
{
"operation": "Highlighting",
"time": "118.576",
"value": "Katex_sentence_1"
},
{
"operation": "back-button-clicked",
"time": "132.125",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "132.199",
"value": "Quadrangular Tower"
},
{
"operation": "Highlighting deleted",
"time": "142.286",
"value": "Quadrangular Tower_sentence_0"
},
{
"operation": "hyperlink",
"time": "144.769",
"value": "Baku"
},
{
"operation": "Now on",
"time": "145.365",
"value": "Baku"
},
{
"operation": "Highlighting",
"time": "148.862",
"value": "Baku_sentence_1"
},
{
"operation": "back-button-clicked",
"time": "151.404",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "151.49",
"value": "Quadrangular Tower"
},
{
"operation": "Highlighting deleted",
"time": "153.041",
"value": "Quadrangular Tower_sentence_2"
},
{
"operation": "Highlighting deleted",
"time": "155.27",
"value": "Quadrangular Tower_cell_0_12_1"
},
{
"operation": "Highlighting deleted",
"time": "155.272",
"value": "Quadrangular Tower_title"
},
{
"operation": "Highlighting deleted",
"time": "155.281",
"value": "Quadrangular Tower_header_cell_0_12_0"
},
{
"operation": "Highlighting deleted",
"time": "155.283",
"value": "Quadrangular Tower_header_cell_0_2_0"
},
{
"operation": "finish",
"time": "173.023",
"value": "finish"
}
]
}
|
exfever_955
|
Alexander Rae Baldwin III is an American actor, film producer, comedian, and political activist who appeared in a 1989 American fantasy comedy film directed by an American film director known for his gothic fantasy and horror films.
|
{
"evidence": [
"Alec Baldwin",
"Beetlejuice",
"Tim Burton"
],
"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:\nAlexander Rae Baldwin III is an American actor, film producer, comedian, and political activist who appeared in a 1989 American fantasy comedy film directed by an American film director known for his gothic fantasy and horror films.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Alec Baldwin",
"Beetlejuice",
"Tim Burton"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5091,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
79143
|
LT-α plays a role to activate signaling pathway.
|
{
"evidence": [
"Lymphotoxin alpha_sentence_24"
],
"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:\nLT-α plays a role to activate signaling pathway.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Lymphotoxin alpha_sentence_24"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5092,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.179",
"value": "?search="
},
{
"operation": "search",
"time": "12.986",
"value": "Contains...LT-α"
},
{
"operation": "Now on",
"time": "14.767",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "17.739",
"value": "LT-α mediated signaling pathway Lymphotoxin alpha_section_3"
},
{
"operation": "Now on",
"time": "19.399",
"value": "Lymphotoxin alpha#LT-.CE.B1 mediated signaling pathway Lymphotoxin alpha section 3"
},
{
"operation": "Page search",
"time": "31.483",
"value": "activate"
},
{
"operation": "Highlighting",
"time": "47.243",
"value": "Lymphotoxin alpha_sentence_24"
},
{
"operation": "finish",
"time": "60.668",
"value": "finish"
}
]
}
|
31037
|
In the 20th century, the "khaki" uniform was introduced; subsequently, all French and Pakistani military regiments have adopted the uniform for active service and summer dress.
|
{
"evidence": [
"Khaki_sentence_12",
"Khaki_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 20th century, the \"khaki\" uniform was introduced; subsequently, all French and Pakistani military regiments have adopted the uniform for active service and summer dress.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"Khaki_sentence_12",
"Khaki_sentence_11"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5093,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.927",
"value": "undefined"
},
{
"operation": "search",
"time": "37.361",
"value": "Khaki"
},
{
"operation": "Now on",
"time": "38.741",
"value": "Khaki"
},
{
"operation": "Highlighting",
"time": "50.529",
"value": "Khaki_sentence_4"
},
{
"operation": "Highlighting deleted",
"time": "51.327",
"value": "Khaki_sentence_4"
},
{
"operation": "Highlighting",
"time": "66.912",
"value": "Khaki_sentence_12"
},
{
"operation": "Highlighting",
"time": "71.065",
"value": "Khaki_sentence_11"
},
{
"operation": "finish",
"time": "91.807",
"value": "finish"
}
]
}
|
exfever_5086
|
Pearl Jam is an American girl band formed in a seaport city on the West Coast of the United States and the seat of King County, Washington.
|
{
"evidence": [
"Pearl Jam",
"Seattle"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nPearl Jam is an American girl band formed in a seaport city on the West Coast of the United States and the seat of King County, Washington.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Pearl Jam",
"Seattle"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5094,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
58672
|
Dario Anselmo (born on May 2, 1962) is a former member of the Colorado House of Representatives who grew up in Orono, MN and attended Whittier College, Boston University, and the University of Minnesota.
|
{
"evidence": [
"Dario Anselmo_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:\nDario Anselmo (born on May 2, 1962) is a former member of the Colorado House of Representatives who grew up in Orono, MN and attended Whittier College, Boston University, and the University of Minnesota.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Dario Anselmo_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5095,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.699",
"value": "?search="
},
{
"operation": "search",
"time": "8.116",
"value": "Dario Anselmo"
},
{
"operation": "Now on",
"time": "9.485",
"value": "Dario Anselmo"
},
{
"operation": "Highlighting",
"time": "54.212",
"value": "Dario Anselmo_sentence_2"
},
{
"operation": "Highlighting",
"time": "54.212",
"value": "Dario Anselmo_section_0"
},
{
"operation": "Highlighting deleted",
"time": "59.95",
"value": "Dario Anselmo_sentence_2"
},
{
"operation": "Highlighting deleted",
"time": "59.952",
"value": "Dario Anselmo_section_0"
},
{
"operation": "Highlighting",
"time": "61.298",
"value": "Dario Anselmo_sentence_0"
},
{
"operation": "finish",
"time": "262.979",
"value": "finish"
}
]
}
|
exfever_5752
|
21 Jump Street is a 2012 American buddy cop action comedy series written by an American actor, filmmaker, and comedian who is known for his comedic roles in films including Superbad.
|
{
"evidence": [
"21 Jump Street (film)",
"Jonah Hill"
],
"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:\n21 Jump Street is a 2012 American buddy cop action comedy series written by an American actor, filmmaker, and comedian who is known for his comedic roles in films including Superbad.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"21 Jump Street (film)",
"Jonah Hill"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5096,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_934
|
Charles I was England's fourth and longest-serving monarch.
|
{
"evidence": [
"Charles I of England",
"Kingdom of Ireland"
],
"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:\nCharles I was England's fourth and longest-serving monarch.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Charles I of England",
"Kingdom of Ireland"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5097,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_197
|
Year Zero is the third studio album by American industrial rock band Nine Inch Nails, released by an American record label owned by a Dutch–American multinational music corporation.
|
{
"evidence": [
"Year Zero (album)",
"Interscope Records",
"Universal Music Group"
],
"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:\nYear Zero is the third studio album by American industrial rock band Nine Inch Nails, released by an American record label owned by a Dutch–American multinational music corporation.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Year Zero (album)",
"Interscope Records",
"Universal Music Group"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5098,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_8995
|
Albania is a country in Southeastern Europe that shares land borders with a partially recognized state that declared its independence from a landlocked country in Southeast Europe.
|
{
"evidence": [
"Albania",
"Kosovo",
"Serbia"
],
"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:\nAlbania is a country in Southeastern Europe that shares land borders with a partially recognized state that declared its independence from a landlocked country in Southeast Europe.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Albania",
"Kosovo",
"Serbia"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5099,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
48596
|
The championship has a back-door format for the first two rounds before proceeding to a knock-out format, and generally, any team to lose two matches will be knocked out of the championship.
|
{
"evidence": [
"Boxing_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 championship has a back-door format for the first two rounds before proceeding to a knock-out format, and generally, any team to lose two matches will be knocked out of the championship.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Boxing_sentence_2"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5100,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.127",
"value": "?search="
},
{
"operation": "Now on",
"time": "73.461",
"value": "?search="
},
{
"operation": "search",
"time": "124.304",
"value": "boxing"
},
{
"operation": "Now on",
"time": "125.797",
"value": "Boxing"
},
{
"operation": "Page search",
"time": "205.353",
"value": "championship"
},
{
"operation": "Highlighting",
"time": "310.778",
"value": "Boxing_sentence_2"
},
{
"operation": "finish",
"time": "338.676",
"value": "finish"
}
]
}
|
exfever_5061
|
The Last of Us Part II is a 2020 action-adventure game published by Sony Interactive Entertainment for the home video game console that is announced as the successor to the PlayStation 2.
|
{
"evidence": [
"The Last of Us Part II",
"PlayStation 4"
],
"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 Last of Us Part II is a 2020 action-adventure game published by Sony Interactive Entertainment for the home video game console that is announced as the successor to the PlayStation 2.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Last of Us Part II",
"PlayStation 4"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5101,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
62969
|
In the 1987 season, the Cleveland Indians were being predicted by several publications such as Sports Illustrated (an American sports magazine owned by Authentic Brands Group), as the best team in baseball on the back of their two 30+ home run hitters, but fall well short of SI's bold prediction as they were not above .500 even once all season, and an 8-20 May ended any realistic hope of contention.
|
{
"evidence": [
"1987 Cleveland Indians season_sentence_1",
"1987 Cleveland Indians season_sentence_8",
"1987 Cleveland Indians season_sentence_10",
"1987 Cleveland Indians season_sentence_11",
"Sports Illustrated_sentence_0"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nIn the 1987 season, the Cleveland Indians were being predicted by several publications such as Sports Illustrated (an American sports magazine owned by Authentic Brands Group), as the best team in baseball on the back of their two 30+ home run hitters, but fall well short of SI's bold prediction as they were not above .500 even once all season, and an 8-20 May ended any realistic hope of contention.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"1987 Cleveland Indians season_sentence_1",
"1987 Cleveland Indians season_sentence_8",
"1987 Cleveland Indians season_sentence_10",
"1987 Cleveland Indians season_sentence_11",
"Sports Illustrated_sentence_0"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5102,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.408",
"value": "undefined"
},
{
"operation": "search",
"time": "36.874",
"value": "Cleveland Indians 1987"
},
{
"operation": "Now on",
"time": "39.161",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "49.563",
"value": "Cleveland Indians"
},
{
"operation": "Now on",
"time": "53.703",
"value": "Cleveland Indians"
},
{
"operation": "hyperlink",
"time": "218.772",
"value": "List of Cleveland Indians seasons"
},
{
"operation": "Now on",
"time": "222.211",
"value": "List of Cleveland Indians seasons"
},
{
"operation": "hyperlink",
"time": "237.568",
"value": "1987"
},
{
"operation": "Now on",
"time": "239.996",
"value": "1987 Cleveland Indians season"
},
{
"operation": "Highlighting",
"time": "246.129",
"value": "1987 Cleveland Indians season_sentence_1"
},
{
"operation": "Highlighting",
"time": "265.874",
"value": "1987 Cleveland Indians season_sentence_8"
},
{
"operation": "Highlighting",
"time": "279.543",
"value": "1987 Cleveland Indians season_sentence_10"
},
{
"operation": "Highlighting",
"time": "281.766",
"value": "1987 Cleveland Indians season_sentence_11"
},
{
"operation": "hyperlink",
"time": "288.32",
"value": "Sports Illustrated"
},
{
"operation": "Now on",
"time": "289.967",
"value": "Sports Illustrated"
},
{
"operation": "Highlighting",
"time": "292.231",
"value": "Sports Illustrated_sentence_0"
},
{
"operation": "finish",
"time": "317.785",
"value": "finish"
}
]
}
|
exfever_4636
|
Nina Jacobson was the eighth woman to head a Hollywood film studio.
|
{
"evidence": [
"Nina Jacobson",
"The Walt Disney Company"
],
"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:\nNina Jacobson was the eighth woman to head a Hollywood film studio.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Nina Jacobson",
"The Walt Disney Company"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5103,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_8066
|
Andy Kaufman is the most well-known American professional wrestler.
|
{
"evidence": [
"Andy Kaufman",
"Anti-humor"
],
"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:\nAndy Kaufman is the most well-known American professional wrestler.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Andy Kaufman",
"Anti-humor"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5104,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_6210
|
Alexander Hamilton was an American revolutionary who was the first chief financial officer of the federal government of the United States.
|
{
"evidence": [
"Alexander Hamilton",
"United States Secretary of the Treasury"
],
"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:\nAlexander Hamilton was an American revolutionary who was the first chief financial officer of the federal government of the United States.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Alexander Hamilton",
"United States Secretary of the Treasury"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5105,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_3783
|
Death in Paradise is a British–French crime drama television series, starring an English actor who played Colin Frissell in a 2003 Christmas-themed romantic comedy-drama film, written and directed by Richard Curtis.
|
{
"evidence": [
"Death in Paradise (TV series)",
"Kris Marshall",
"Love Actually"
],
"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:\nDeath in Paradise is a British–French crime drama television series, starring an English actor who played Colin Frissell in a 2003 Christmas-themed romantic comedy-drama film, written and directed by Richard Curtis.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Death in Paradise (TV series)",
"Kris Marshall",
"Love Actually"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5106,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_8301
|
Sheryl Lee was the first actress to venture into filmmaking.
|
{
"evidence": [
"Sheryl Lee",
"David Lynch",
"Venice Film Festival"
],
"label": "NOT ENOUGH INFO"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nSheryl Lee was the first actress to venture into filmmaking.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Sheryl Lee",
"David Lynch",
"Venice Film Festival"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5107,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_4034
|
Daniel Day-Lewis is a retired English actor who has won an Oscar, for which the winner was awarded a copy of a gold statuette that depicts a knight rendered in the style of visual art that first appeared in Italy.
|
{
"evidence": [
"Daniel Day-Lewis",
"Academy Awards",
"Art Deco"
],
"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:\nDaniel Day-Lewis is a retired English actor who has won an Oscar, for which the winner was awarded a copy of a gold statuette that depicts a knight rendered in the style of visual art that first appeared in Italy.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Daniel Day-Lewis",
"Academy Awards",
"Art Deco"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5108,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_5309
|
Berlin is Germany's municipality holding secondary status in a country, state, province, department, or other sub-national entity, usually as its seat of the government.
|
{
"evidence": [
"Berlin",
"Capital city"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nBerlin is Germany's municipality holding secondary status in a country, state, province, department, or other sub-national entity, usually as its seat of the government.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Berlin",
"Capital city"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5109,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_1586
|
Erich von Manstein was a German Field Marshal of the unified armed forces of Nazi Germany from 1935 to 1945.
|
{
"evidence": [
"Erich von Manstein",
"Wehrmacht"
],
"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:\nErich von Manstein was a German Field Marshal of the unified armed forces of Nazi Germany from 1935 to 1945.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Erich von Manstein",
"Wehrmacht"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5110,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_923
|
On Stranger Tides is a 1979 historical fantasy novel that features an English sea captain known as the captain of the vessel that rescued marooned Alexander Selkirk, whose plight is generally believed to have inspired a novel first published on 25 April 1719.
|
{
"evidence": [
"On Stranger Tides",
"Woodes Rogers",
"Robinson Crusoe"
],
"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:\nOn Stranger Tides is a 1979 historical fantasy novel that features an English sea captain known as the captain of the vessel that rescued marooned Alexander Selkirk, whose plight is generally believed to have inspired a novel first published on 25 April 1719.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"On Stranger Tides",
"Woodes Rogers",
"Robinson Crusoe"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5111,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
36357
|
The Karadjordje Cup, named after a Serbian revolutionary, has been played semiannually since 1987 and includes football clubs from around Australia.
|
{
"evidence": [
"Karadjordje Cup_sentence_0"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Karadjordje Cup, named after a Serbian revolutionary, has been played semiannually since 1987 and includes football clubs from around Australia.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Karadjordje Cup_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5112,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.656",
"value": "?search="
},
{
"operation": "search",
"time": "30.567",
"value": "Karadjordje Cup"
},
{
"operation": "Now on",
"time": "31.525",
"value": "Karadjordje Cup"
},
{
"operation": "Page search",
"time": "49.144",
"value": "revolution"
},
{
"operation": "Page search",
"time": "90.798",
"value": "named"
},
{
"operation": "hyperlink",
"time": "121.648",
"value": "Karadjordje"
},
{
"operation": "Now on",
"time": "122.797",
"value": "Karađorđe"
},
{
"operation": "Highlighting",
"time": "132.258",
"value": "Karađorđe_sentence_2"
},
{
"operation": "Highlighting",
"time": "132.26",
"value": "Karađorđe_title"
},
{
"operation": "back-button-clicked",
"time": "136.602",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "136.816",
"value": "Karadjordje Cup"
},
{
"operation": "Page search",
"time": "146.474",
"value": "semi"
},
{
"operation": "Highlighting",
"time": "161.771",
"value": "Karadjordje Cup_sentence_0"
},
{
"operation": "finish",
"time": "318.867",
"value": "finish"
}
]
}
|
exfever_2512
|
The Singing Detective is a radio drama in the national broadcaster of the United Kingdom.
|
{
"evidence": [
"The Singing Detective",
"BBC"
],
"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 Singing Detective is a radio drama in the national broadcaster of the United Kingdom.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Singing Detective",
"BBC"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5113,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
17449
|
Mithi Chhapri is the only Indian municipality with a hundred percent literacy rate.
|
{
"evidence": [
"Mithi Chhapri_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:\nMithi Chhapri is the only Indian municipality with a hundred percent literacy rate.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Mithi Chhapri_sentence_3"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5114,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.032",
"value": "?search="
},
{
"operation": "search",
"time": "12.375",
"value": "Mithi Chhapri"
},
{
"operation": "Now on",
"time": "13.294",
"value": "Mithi Chhapri"
},
{
"operation": "Highlighting",
"time": "28.406",
"value": "Mithi Chhapri_sentence_3"
},
{
"operation": "finish",
"time": "44.669",
"value": "finish"
}
]
}
|
17351
|
Democratic Party candidate John E. Miles withdrew from the 1940 New Mexico gubernatorial election.
|
{
"evidence": [
"1940 New Mexico gubernatorial election_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:\nDemocratic Party candidate John E. Miles withdrew from the 1940 New Mexico gubernatorial election.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"1940 New Mexico gubernatorial election_sentence_1"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5115,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "2.605",
"value": "?search="
},
{
"operation": "search",
"time": "22.817",
"value": "1940 New Mexico gubernatorial election"
},
{
"operation": "Now on",
"time": "24.537",
"value": "1940 New Mexico gubernatorial election"
},
{
"operation": "Page search",
"time": "30.252",
"value": "miles"
},
{
"operation": "Highlighting",
"time": "37.619",
"value": "1940 New Mexico gubernatorial election_sentence_1"
},
{
"operation": "finish",
"time": "67.886",
"value": "finish"
}
]
}
|
24255
|
USP responds apart from other stress responses and the induction of UPS proteins are used in changes in the entire phenotype but not used in metabolic or growth processes.
|
{
"evidence": [
"Universal stress protein_sentence_45"
],
"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:\nUSP responds apart from other stress responses and the induction of UPS proteins are used in changes in the entire phenotype but not used in metabolic or growth processes.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Universal stress protein_sentence_45"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5116,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.809",
"value": "?search="
},
{
"operation": "search",
"time": "16.19",
"value": "UPS"
},
{
"operation": "Now on",
"time": "16.607",
"value": "UPS"
},
{
"operation": "Page search",
"time": "38.06",
"value": "induction"
},
{
"operation": "search",
"time": "57.852",
"value": "UPS protiens"
},
{
"operation": "Now on",
"time": "58.511",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "62.364",
"value": "ups proteins"
},
{
"operation": "Now on",
"time": "63.438",
"value": "undefined"
},
{
"operation": "search",
"time": "89.236",
"value": "Ubiquitin proteasome system"
},
{
"operation": "Now on",
"time": "90.115",
"value": "Proteasome"
},
{
"operation": "Page search",
"time": "117.147",
"value": "UPS"
},
{
"operation": "search",
"time": "208.121",
"value": "UPS proteins are used in changes in the entire phenotype"
},
{
"operation": "Now on",
"time": "209.208",
"value": "undefined"
},
{
"operation": "search",
"time": "217.488",
"value": "UPS"
},
{
"operation": "Now on",
"time": "218.009",
"value": "UPS"
},
{
"operation": "search",
"time": "248.119",
"value": "USP"
},
{
"operation": "Now on",
"time": "248.848",
"value": "USP"
},
{
"operation": "hyperlink",
"time": "266.062",
"value": "Ultraspiracle protein"
},
{
"operation": "Now on",
"time": "266.646",
"value": "Ecdysone receptor"
},
{
"operation": "Page search",
"time": "294.341",
"value": "phenotype"
},
{
"operation": "Page search",
"time": "307.508",
"value": "stress responses"
},
{
"operation": "back-button-clicked",
"time": "318.637",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "318.691",
"value": "USP"
},
{
"operation": "hyperlink",
"time": "321.263",
"value": "Universal stress protein"
},
{
"operation": "Now on",
"time": "321.801",
"value": "Universal stress protein"
},
{
"operation": "Highlighting",
"time": "361.1",
"value": "Universal stress protein_sentence_1"
},
{
"operation": "Highlighting",
"time": "414.594",
"value": "Universal stress protein_sentence_7"
},
{
"operation": "Highlighting",
"time": "414.595",
"value": "Universal stress protein_section_0"
},
{
"operation": "Highlighting",
"time": "437.138",
"value": "Universal stress protein_sentence_16"
},
{
"operation": "Highlighting deleted",
"time": "438.521",
"value": "Universal stress protein_sentence_7"
},
{
"operation": "Highlighting deleted",
"time": "439.977",
"value": "Universal stress protein_sentence_1"
},
{
"operation": "Page search",
"time": "467.2",
"value": "phenotype"
},
{
"operation": "Highlighting",
"time": "487.136",
"value": "Universal stress protein_sentence_45"
},
{
"operation": "Page search",
"time": "509.527",
"value": "metabolic"
},
{
"operation": "finish",
"time": "543.054",
"value": "finish"
}
]
}
|
16744
|
MAP2K6 is an enzyme and a member of the dual specificity protein kinase family that is encoded by the MAP2K6 gene on chromosome 17, act as an integration point for multiple biochemical signals and mediates apoptotic cell death in thymocytes.
|
{
"evidence": [
"MAP2K6_sentence_1",
"MAP2K6_sentence_2",
"MAP2K6_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:\nMAP2K6 is an enzyme and a member of the dual specificity protein kinase family that is encoded by the MAP2K6 gene on chromosome 17, act as an integration point for multiple biochemical signals and mediates apoptotic cell death in thymocytes.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"MAP2K6_sentence_1",
"MAP2K6_sentence_2",
"MAP2K6_sentence_3"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5117,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.753",
"value": "?search="
},
{
"operation": "search",
"time": "14.018",
"value": "MAP2K6"
},
{
"operation": "Now on",
"time": "14.641",
"value": "MAP2K6"
},
{
"operation": "Highlighting",
"time": "90.293",
"value": "MAP2K6_sentence_1"
},
{
"operation": "Highlighting",
"time": "94.773",
"value": "MAP2K6_sentence_2"
},
{
"operation": "Highlighting",
"time": "116.181",
"value": "MAP2K6_sentence_3"
},
{
"operation": "Page search",
"time": "158.323",
"value": "thymocyte"
},
{
"operation": "hyperlink",
"time": "175.26",
"value": "chromosome 17"
},
{
"operation": "Now on",
"time": "176.595",
"value": "Chromosome 17"
},
{
"operation": "Page search",
"time": "181.074",
"value": "thymocyte"
},
{
"operation": "back-button-clicked",
"time": "187.727",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "187.922",
"value": "MAP2K6"
},
{
"operation": "hyperlink",
"time": "196.093",
"value": "gene"
},
{
"operation": "Now on",
"time": "197.426",
"value": "Gene"
},
{
"operation": "back-button-clicked",
"time": "200.999",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "201.215",
"value": "MAP2K6"
},
{
"operation": "hyperlink",
"time": "303.236",
"value": "mitogen-activated protein"
},
{
"operation": "Now on",
"time": "304.757",
"value": "Mitogen-activated protein kinase"
},
{
"operation": "Page search",
"time": "314.691",
"value": "thymocyte"
},
{
"operation": "finish",
"time": "325.833",
"value": "finish"
}
]
}
|
exfever_8571
|
Greenwich became the first Royal Naval College in 1873.
|
{
"evidence": [
"Greenwich",
"Ceremonial counties of England"
],
"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:\nGreenwich became the first Royal Naval College in 1873.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Greenwich",
"Ceremonial counties of England"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5118,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
25889
|
The Alaska Kid, a Ukranian-Russian-Polish television series consisted of 13 episodes aired in 1993.
|
{
"evidence": [
"The Alaska Kid_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 Alaska Kid, a Ukranian-Russian-Polish television series consisted of 13 episodes aired in 1993.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"The Alaska Kid_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5119,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.139",
"value": "?search="
},
{
"operation": "search",
"time": "10.841",
"value": "The Alaska Kid"
},
{
"operation": "Now on",
"time": "12.272",
"value": "The Alaska Kid"
},
{
"operation": "Highlighting",
"time": "14.76",
"value": "The Alaska Kid_sentence_0"
},
{
"operation": "finish",
"time": "38.601",
"value": "finish"
}
]
}
|
40697
|
In 1834, George Alexander Anstey went back to England with one of his sisters; on his return to Tasmania, he was shipwrecked in the D'Entrecasteaux Channel and in 1837 he took sheep to Port Phillip, sold them to John and Somerville Learmonth, and returned to Oatlands, his father's estate.
|
{
"evidence": [
"George Alexander Anstey_sentence_4",
"George Alexander Anstey_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:\nIn 1834, George Alexander Anstey went back to England with one of his sisters; on his return to Tasmania, he was shipwrecked in the D'Entrecasteaux Channel and in 1837 he took sheep to Port Phillip, sold them to John and Somerville Learmonth, and returned to Oatlands, his father's estate.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"George Alexander Anstey_sentence_4",
"George Alexander Anstey_sentence_5"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5120,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.447",
"value": "?search="
},
{
"operation": "search",
"time": "292.283",
"value": "George Alexander Anstey"
},
{
"operation": "Now on",
"time": "293.436",
"value": "George Alexander Anstey"
},
{
"operation": "Highlighting",
"time": "320.232",
"value": "George Alexander Anstey_sentence_4"
},
{
"operation": "Highlighting",
"time": "335.054",
"value": "George Alexander Anstey_sentence_5"
},
{
"operation": "finish",
"time": "349.248",
"value": "finish"
}
]
}
|
83884
|
Friedrich Heinrich Ranke, the brother of Leopold von Ranke, studied theology and philology at the University of Jena and philosophy at the University of Halle and eventually became a private schoolteacher in Frankfurt an der Oder.
|
{
"evidence": [
"Friedrich Heinrich Ranke_sentence_1",
"Friedrich Heinrich Ranke_sentence_2",
"Friedrich Heinrich Ranke_sentence_3"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nFriedrich Heinrich Ranke, the brother of Leopold von Ranke, studied theology and philology at the University of Jena and philosophy at the University of Halle and eventually became a private schoolteacher in Frankfurt an der Oder.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Friedrich Heinrich Ranke_sentence_1",
"Friedrich Heinrich Ranke_sentence_2",
"Friedrich Heinrich Ranke_sentence_3"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5121,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.114",
"value": "?search="
},
{
"operation": "search",
"time": "13.506",
"value": "Friedrich Heinrich Ranke"
},
{
"operation": "Now on",
"time": "14.214",
"value": "Friedrich Heinrich Ranke"
},
{
"operation": "Highlighting",
"time": "19.009",
"value": "Friedrich Heinrich Ranke_sentence_1"
},
{
"operation": "Highlighting",
"time": "31.378",
"value": "Friedrich Heinrich Ranke_sentence_2"
},
{
"operation": "Highlighting",
"time": "34.984",
"value": "Friedrich Heinrich Ranke_sentence_3"
},
{
"operation": "finish",
"time": "60.925",
"value": "finish"
}
]
}
|
76030
|
Anzor Sitchinava was born in 1995 who plays wing for Academy Tbilisi on the Georgia national team and was penalized in his first match but still selected for the 2018 team.
|
{
"evidence": [
"Anzor Sitchinava_sentence_0",
"Anzor Sitchinava_sentence_1",
"Anzor Sitchinava_sentence_2",
"Anzor Sitchinava_sentence_3"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nAnzor Sitchinava was born in 1995 who plays wing for Academy Tbilisi on the Georgia national team and was penalized in his first match but still selected for the 2018 team.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Anzor Sitchinava_sentence_0",
"Anzor Sitchinava_sentence_1",
"Anzor Sitchinava_sentence_2",
"Anzor Sitchinava_sentence_3"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5122,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.782",
"value": "?search="
},
{
"operation": "search",
"time": "11.832",
"value": "Anzor Sitchinava"
},
{
"operation": "Now on",
"time": "12.423",
"value": "Anzor Sitchinava"
},
{
"operation": "Highlighting",
"time": "16.453",
"value": "Anzor Sitchinava_sentence_0"
},
{
"operation": "Highlighting",
"time": "22.953",
"value": "Anzor Sitchinava_sentence_1"
},
{
"operation": "Highlighting",
"time": "35.086",
"value": "Anzor Sitchinava_sentence_2"
},
{
"operation": "Highlighting",
"time": "65.432",
"value": "Anzor Sitchinava_sentence_3"
},
{
"operation": "finish",
"time": "93.168",
"value": "finish"
}
]
}
|
37940
|
Delmer Berg was an American soldier who volunteered to serve with the XV International Brigade (nicknamed the Abraham Lincoln Brigade) during the Spanish Civil War that has four and who began serving as a union organizer in the 1950s, then became an official of the National Association for the Advancement of Colored People.
|
{
"evidence": [
"Delmer Berg_sentence_0",
"Delmer Berg_sentence_9",
"Delmer Berg_sentence_10"
],
"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:\nDelmer Berg was an American soldier who volunteered to serve with the XV International Brigade (nicknamed the Abraham Lincoln Brigade) during the Spanish Civil War that has four and who began serving as a union organizer in the 1950s, then became an official of the National Association for the Advancement of Colored People.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Delmer Berg_sentence_0",
"Delmer Berg_sentence_9",
"Delmer Berg_sentence_10"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5123,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.638",
"value": "?search="
},
{
"operation": "search",
"time": "15.402",
"value": "Delmer Berg"
},
{
"operation": "Now on",
"time": "16.29",
"value": "Delmer Berg"
},
{
"operation": "Highlighting",
"time": "26.14",
"value": "Delmer Berg_sentence_0"
},
{
"operation": "Highlighting",
"time": "51.653",
"value": "Delmer Berg_sentence_9"
},
{
"operation": "Highlighting",
"time": "62.929",
"value": "Delmer Berg_sentence_10"
},
{
"operation": "finish",
"time": "89.269",
"value": "finish"
}
]
}
|
exfever_4757
|
Rowan Sebastian Atkinson is an Irish actor, comedian and writer who played the title role in a series of four period British sitcoms, plus several one-off installments, which originally aired on BBC One from 1983 to 1989.
|
{
"evidence": [
"Rowan Atkinson",
"Blackadder"
],
"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:\nRowan Sebastian Atkinson is an Irish actor, comedian and writer who played the title role in a series of four period British sitcoms, plus several one-off installments, which originally aired on BBC One from 1983 to 1989.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Rowan Atkinson",
"Blackadder"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5124,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
13763
|
Neos Finance started as Finemiro Banca from a predecessor company in 1997; it became part Cardine Banca group in 2000 and part of Sanpaolo IMI in 2002; and by the end of 2004, Sanpaolo held a minority share.
|
{
"evidence": [
"Neos Finance_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:\nNeos Finance started as Finemiro Banca from a predecessor company in 1997; it became part Cardine Banca group in 2000 and part of Sanpaolo IMI in 2002; and by the end of 2004, Sanpaolo held a minority share.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Neos Finance_sentence_5"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5125,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.025",
"value": "?search="
},
{
"operation": "search",
"time": "19.21",
"value": "Neos Finance"
},
{
"operation": "Now on",
"time": "20.215",
"value": "Neos Finance"
},
{
"operation": "Highlighting",
"time": "155.611",
"value": "Neos Finance_sentence_2"
},
{
"operation": "Highlighting",
"time": "171.324",
"value": "Neos Finance_sentence_3"
},
{
"operation": "Highlighting",
"time": "177.552",
"value": "Neos Finance_sentence_4"
},
{
"operation": "Highlighting",
"time": "196.497",
"value": "Neos Finance_sentence_5"
},
{
"operation": "Highlighting deleted",
"time": "204.909",
"value": "Neos Finance_sentence_4"
},
{
"operation": "Highlighting deleted",
"time": "210.215",
"value": "Neos Finance_sentence_3"
},
{
"operation": "Highlighting",
"time": "212.844",
"value": "Neos Finance_sentence_3"
},
{
"operation": "Highlighting deleted",
"time": "220.962",
"value": "Neos Finance_sentence_3"
},
{
"operation": "Highlighting",
"time": "221.528",
"value": "Neos Finance_sentence_3"
},
{
"operation": "Highlighting deleted",
"time": "224.836",
"value": "Neos Finance_sentence_2"
},
{
"operation": "hyperlink",
"time": "225.702",
"value": "Casse Emiliano Romagnole"
},
{
"operation": "Now on",
"time": "226.831",
"value": "Casse Emiliano Romagnole"
},
{
"operation": "back-button-clicked",
"time": "226.966",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "227.174",
"value": "Neos Finance"
},
{
"operation": "Highlighting deleted",
"time": "227.73",
"value": "Neos Finance_sentence_3"
},
{
"operation": "finish",
"time": "232.924",
"value": "finish"
}
]
}
|
exfever_7713
|
Tina Fey is an American comedian who is best known for creating the comedy series that?aired from October 11, 2006, to January 31, 2013.?
|
{
"evidence": [
"Tina Fey",
"30 Rock"
],
"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:\nTina Fey is an American comedian who is best known for creating the comedy series that?aired from October 11, 2006, to January 31, 2013.?\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Tina Fey",
"30 Rock"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5126,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
49299
|
Hermann Kuhnt was a German ophthalmologist, he studied medicine nowhere near the Bonn, Berlin and Würzburg.
|
{
"evidence": [
"Hermann Kuhnt_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:\nHermann Kuhnt was a German ophthalmologist, he studied medicine nowhere near the Bonn, Berlin and Würzburg.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Hermann Kuhnt_sentence_1"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5127,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.184",
"value": "?search="
},
{
"operation": "search",
"time": "77.668",
"value": "Hermann Kuhnt"
},
{
"operation": "Now on",
"time": "78.539",
"value": "Hermann Kuhnt"
},
{
"operation": "Highlighting",
"time": "84.838",
"value": "Hermann Kuhnt_sentence_0"
},
{
"operation": "Highlighting deleted",
"time": "88.792",
"value": "Hermann Kuhnt_sentence_0"
},
{
"operation": "Highlighting",
"time": "91.812",
"value": "Hermann Kuhnt_sentence_1"
},
{
"operation": "finish",
"time": "118.932",
"value": "finish"
}
]
}
|
exfever_4323
|
Marvel's Inhumans is an American television series produced by an American production company responsible for live-action through an Indian?animation production company.
|
{
"evidence": [
"Inhumans (TV series)",
"Marvel Television",
"Marvel Animation"
],
"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:\nMarvel's Inhumans is an American television series produced by an American production company responsible for live-action through an Indian?animation production company.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Inhumans (TV series)",
"Marvel Television",
"Marvel Animation"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5128,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
78433
|
Moe Brooker is a recipient of the James Van Der Zee Lifetime Achievement Award that celebrates the achievements of African Americans in the visual arts.
|
{
"evidence": [
"Moe Brooker_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:\nMoe Brooker is a recipient of the James Van Der Zee Lifetime Achievement Award that celebrates the achievements of African Americans in the visual arts.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Moe Brooker_sentence_2"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5129,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.751",
"value": "?search="
},
{
"operation": "search",
"time": "7.32",
"value": "Moe Brooker"
},
{
"operation": "Now on",
"time": "8.239",
"value": "Moe Brooker"
},
{
"operation": "Page search",
"time": "51.611",
"value": "James Van Der Zee"
},
{
"operation": "Highlighting",
"time": "64.471",
"value": "Moe Brooker_sentence_2"
},
{
"operation": "hyperlink",
"time": "69.929",
"value": "James Van Der Zee"
},
{
"operation": "Now on",
"time": "70.348",
"value": "James Van Der Zee"
},
{
"operation": "back-button-clicked",
"time": "89.744",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "89.797",
"value": "Moe Brooker"
},
{
"operation": "Highlighting",
"time": "102.447",
"value": "Moe Brooker_cell_0_5_1"
},
{
"operation": "Highlighting",
"time": "102.452",
"value": "Moe Brooker_header_cell_0_5_0"
},
{
"operation": "Highlighting",
"time": "102.452",
"value": "Moe Brooker_header_cell_0_0_0"
},
{
"operation": "Highlighting deleted",
"time": "104.202",
"value": "Moe Brooker_cell_0_5_1"
},
{
"operation": "Highlighting deleted",
"time": "104.206",
"value": "Moe Brooker_header_cell_0_5_0"
},
{
"operation": "Highlighting deleted",
"time": "104.207",
"value": "Moe Brooker_header_cell_0_0_0"
},
{
"operation": "Highlighting",
"time": "106.046",
"value": "Moe Brooker_cell_0_5_1"
},
{
"operation": "Highlighting",
"time": "106.049",
"value": "Moe Brooker_header_cell_0_5_0"
},
{
"operation": "Highlighting",
"time": "106.05",
"value": "Moe Brooker_header_cell_0_0_0"
},
{
"operation": "Highlighting deleted",
"time": "109.185",
"value": "Moe Brooker_cell_0_5_1"
},
{
"operation": "Highlighting deleted",
"time": "109.189",
"value": "Moe Brooker_header_cell_0_5_0"
},
{
"operation": "Highlighting deleted",
"time": "109.19",
"value": "Moe Brooker_header_cell_0_0_0"
},
{
"operation": "search",
"time": "145.559",
"value": "James Van Der Zee lifetime achievement award"
},
{
"operation": "Now on",
"time": "147.225",
"value": "undefined"
},
{
"operation": "back-button-clicked",
"time": "155.109",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "155.162",
"value": "Moe Brooker"
},
{
"operation": "hyperlink",
"time": "157.562",
"value": "James Van Der Zee"
},
{
"operation": "Now on",
"time": "157.928",
"value": "James Van Der Zee"
},
{
"operation": "Page search",
"time": "218.836",
"value": "achievements"
},
{
"operation": "Page search",
"time": "219.198",
"value": "achievements"
},
{
"operation": "back-button-clicked",
"time": "221.007",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "221.061",
"value": "Moe Brooker"
},
{
"operation": "Page search",
"time": "223.299",
"value": "achievements"
},
{
"operation": "search",
"time": "275.563",
"value": "celebrates the achievements of African Americans in the visual arts"
},
{
"operation": "Now on",
"time": "278.125",
"value": "undefined"
},
{
"operation": "finish",
"time": "290.5",
"value": "finish"
}
]
}
|
exfever_235
|
Rebecca Hall directed the psychological thriller The Gift in 2015.
|
{
"evidence": [
"Rebecca Hall",
"Christopher Nolan"
],
"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:\nRebecca Hall directed the psychological thriller The Gift in 2015.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Rebecca Hall",
"Christopher Nolan"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5130,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7856
|
Mauritius is an island nation in the Indian Ocean that includes the main island, as well as an Indian Ocean archipelago.
|
{
"evidence": [
"Mauritius",
"St. Brandon"
],
"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:\nMauritius is an island nation in the Indian Ocean that includes the main island, as well as an Indian Ocean archipelago.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Mauritius",
"St. Brandon"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5131,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_4826
|
David Yates directed a 2016 adventure film about The Legend of a fictional character raised in the African jungle.
|
{
"evidence": [
"The Legend of Tarzan (film)",
"Tarzan"
],
"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:\nDavid Yates directed a 2016 adventure film about The Legend of a fictional character raised in the African jungle.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Legend of Tarzan (film)",
"Tarzan"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5132,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_258
|
Patrick Ness's first film is A Monster Calls.
|
{
"evidence": [
"A Monster Calls (film)",
"A Monster Calls"
],
"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:\nPatrick Ness's first film is A Monster Calls.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"A Monster Calls (film)",
"A Monster Calls"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5133,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
66532
|
After the Treaty of Versaille, the border between Romania and Serbia was so distinct that no Romanian was left within the borders of the Kingdom of Serbs, Croats, and Slovenes.
|
{
"evidence": [
"Romanians of Serbia_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:\nAfter the Treaty of Versaille, the border between Romania and Serbia was so distinct that no Romanian was left within the borders of the Kingdom of Serbs, Croats, and Slovenes.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Romanians of Serbia_sentence_6"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5134,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.583",
"value": "?search="
},
{
"operation": "search",
"time": "19.173",
"value": "the Treaty of Versaille"
},
{
"operation": "Now on",
"time": "24.652",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "28.124",
"value": "Treaty of Versailles"
},
{
"operation": "Now on",
"time": "34.402",
"value": "Treaty of Versailles"
},
{
"operation": "search",
"time": "65.515",
"value": "Treaty of Trianon"
},
{
"operation": "Now on",
"time": "102.027",
"value": "Treaty of Versailles"
},
{
"operation": "search",
"time": "105.591",
"value": "Treaty of Trianon"
},
{
"operation": "Now on",
"time": "159.063",
"value": "Treaty of Trianon"
},
{
"operation": "search",
"time": "262.665",
"value": "Romanians of Serbia"
},
{
"operation": "Now on",
"time": "264.687",
"value": "Romanians of Serbia"
},
{
"operation": "Highlighting",
"time": "288.345",
"value": "Romanians of Serbia_sentence_6"
},
{
"operation": "finish",
"time": "315.141",
"value": "finish"
}
]
}
|
71583
|
Vittus Qujaukitsoq was not popular among the citizens.
|
{
"evidence": [
"Vittus Qujaukitsoq_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:\nVittus Qujaukitsoq was not popular among the citizens.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Vittus Qujaukitsoq_sentence_0"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5135,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.774",
"value": "?search="
},
{
"operation": "search",
"time": "12.333",
"value": "Vittus Qujaukitsoq"
},
{
"operation": "Now on",
"time": "12.948",
"value": "Vittus Qujaukitsoq"
},
{
"operation": "Page search",
"time": "34.418",
"value": "popular"
},
{
"operation": "Page search",
"time": "47.626",
"value": "citizens"
},
{
"operation": "Highlighting",
"time": "134.518",
"value": "Vittus Qujaukitsoq_sentence_0"
},
{
"operation": "finish",
"time": "142.298",
"value": "finish"
}
]
}
|
exfever_691
|
Kate Walsh is an American actress best known for her role as Dr. Addison Montgomery in the American medical drama television series that premiered on the American Broadcasting Company.
|
{
"evidence": [
"Kate Walsh (actress)",
"Grey's Anatomy"
],
"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:\nKate Walsh is an American actress best known for her role as Dr. Addison Montgomery in the American medical drama television series that premiered on the American Broadcasting Company.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Kate Walsh (actress)",
"Grey's Anatomy"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5136,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_3660
|
Ultimate Spider-Man was the most famous American superhero animated television series.
|
{
"evidence": [
"Ultimate Spider-Man (TV series)",
"Ultimate Spider-Man"
],
"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:\nUltimate Spider-Man was the most famous American superhero animated television series.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Ultimate Spider-Man (TV series)",
"Ultimate Spider-Man"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5137,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
68873
|
The bombings in Biloxi, Mississippi, which were referred to as Bombingham, and which targeted African-Americans, were never linked to any particular group of perpetrators.
|
{
"evidence": [
"Bombingham_sentence_2",
"Bombingham_sentence_7"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe bombings in Biloxi, Mississippi, which were referred to as Bombingham, and which targeted African-Americans, were never linked to any particular group of perpetrators.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Bombingham_sentence_2",
"Bombingham_sentence_7"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5138,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.252",
"value": "?search="
},
{
"operation": "search",
"time": "16.285",
"value": "Bombingham"
},
{
"operation": "Now on",
"time": "17.401",
"value": "Bombingham"
},
{
"operation": "Highlighting",
"time": "55.554",
"value": "Bombingham_sentence_2"
},
{
"operation": "Highlighting",
"time": "78.482",
"value": "Bombingham_sentence_7"
},
{
"operation": "finish",
"time": "92.851",
"value": "finish"
}
]
}
|
exfever_139
|
24 was the fourth film directed by Vikram Kumar.
|
{
"evidence": [
"24 (2016 film)",
"Suriya"
],
"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:\n24 was the fourth film directed by Vikram Kumar.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"24 (2016 film)",
"Suriya"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5139,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
49518
|
Six teams participated in the 1992–93 Central Hockey League, hailing Tulsa Oilers (under Head coach Rob Murray) as champion and Oklahoma City Blazers as the team with the top records.
|
{
"evidence": [
"Rob Murray_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:\nSix teams participated in the 1992–93 Central Hockey League, hailing Tulsa Oilers (under Head coach Rob Murray) as champion and Oklahoma City Blazers as the team with the top records.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"Rob Murray_sentence_19"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5140,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.706",
"value": "?search="
},
{
"operation": "search",
"time": "14.033",
"value": "1992–93 Central Hockey League season"
},
{
"operation": "Now on",
"time": "14.867",
"value": "1992–93 CHL season"
},
{
"operation": "Highlighting",
"time": "23.468",
"value": "1992–93 CHL season_sentence_0"
},
{
"operation": "Highlighting",
"time": "23.469",
"value": "1992–93 CHL season_title"
},
{
"operation": "Highlighting",
"time": "36.318",
"value": "1992–93 CHL season_cell_1_1_0"
},
{
"operation": "Highlighting",
"time": "36.32",
"value": "1992–93 CHL season_section_0"
},
{
"operation": "Highlighting",
"time": "36.321",
"value": "1992–93 CHL season_section_1"
},
{
"operation": "Highlighting",
"time": "36.322",
"value": "1992–93 CHL season_header_cell_1_0_0"
},
{
"operation": "Highlighting",
"time": "37.003",
"value": "1992–93 CHL season_cell_1_2_0"
},
{
"operation": "Highlighting",
"time": "37.702",
"value": "1992–93 CHL season_cell_1_3_0"
},
{
"operation": "Highlighting",
"time": "38.507",
"value": "1992–93 CHL season_cell_1_4_0"
},
{
"operation": "Highlighting",
"time": "39.34",
"value": "1992–93 CHL season_cell_1_6_0"
},
{
"operation": "Highlighting",
"time": "40.095",
"value": "1992–93 CHL season_cell_1_7_0"
},
{
"operation": "Highlighting",
"time": "59.076",
"value": "1992–93 CHL season_cell_0_8_1"
},
{
"operation": "Highlighting",
"time": "59.08",
"value": "1992–93 CHL season_header_cell_0_8_0"
},
{
"operation": "Highlighting",
"time": "59.08",
"value": "1992–93 CHL season_header_cell_0_7_0"
},
{
"operation": "hyperlink",
"time": "109.314",
"value": "Tulsa Oilers"
},
{
"operation": "Now on",
"time": "110.089",
"value": "Tulsa Oilers"
},
{
"operation": "Page search",
"time": "127.183",
"value": "rob murray"
},
{
"operation": "search",
"time": "150.047",
"value": "Rob Murray (ice hockey)"
},
{
"operation": "Now on",
"time": "151.018",
"value": "Robert Murray"
},
{
"operation": "hyperlink",
"time": "192.189",
"value": "Bob Murray (ice hockey, born 1954)"
},
{
"operation": "Now on",
"time": "192.988",
"value": "Bob Murray (ice hockey, born 1954)"
},
{
"operation": "back-button-clicked",
"time": "196.66",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "197.025",
"value": "Robert Murray"
},
{
"operation": "hyperlink",
"time": "203.911",
"value": "Robert Murray (ice hockey, born 1951)"
},
{
"operation": "Now on",
"time": "204.67",
"value": "Robert Murray (ice hockey, born 1951)"
},
{
"operation": "back-button-clicked",
"time": "210.94",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "211.054",
"value": "Robert Murray"
},
{
"operation": "search",
"time": "229.028",
"value": "head coach tulsa oilers"
},
{
"operation": "Now on",
"time": "231.01",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "247.457",
"value": "History Tulsa Oilers_section_0"
},
{
"operation": "Now on",
"time": "247.957",
"value": "Tulsa Oilers#History Tulsa Oilers section 0"
},
{
"operation": "Highlighting",
"time": "283.841",
"value": "Tulsa Oilers_cell_0_10_1"
},
{
"operation": "Highlighting",
"time": "283.844",
"value": "Tulsa Oilers_title"
},
{
"operation": "Highlighting",
"time": "283.852",
"value": "Tulsa Oilers_header_cell_0_10_0"
},
{
"operation": "Highlighting",
"time": "283.852",
"value": "Tulsa Oilers_header_cell_0_0_0"
},
{
"operation": "hyperlink",
"time": "286.479",
"value": "Rob Murray"
},
{
"operation": "Now on",
"time": "287.29",
"value": "Rob Murray"
},
{
"operation": "Highlighting",
"time": "320.47",
"value": "Rob Murray_sentence_19"
},
{
"operation": "finish",
"time": "371.577",
"value": "finish"
}
]
}
|
exfever_4388
|
Saratoga is a 1937 American romantic drama film starring an American actress known for her portrayal of "bad girl" characters.
|
{
"evidence": [
"Saratoga (film)",
"Jean Harlow"
],
"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:\nSaratoga is a 1937 American romantic drama film starring an American actress known for her portrayal of \"bad girl\" characters.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Saratoga (film)",
"Jean Harlow"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5141,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_6718
|
James Marshall "Jimi" Hendrix was a record producer.
|
{
"evidence": [
"Jimi Hendrix",
"Clarksville, Tennessee"
],
"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:\nJames Marshall \"Jimi\" Hendrix was a record producer.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Jimi Hendrix",
"Clarksville, Tennessee"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5142,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_2449
|
Neil Patrick Harris has received multiple setbacks throughout his career including The Antoinette Perry Award for Excellence in Broadway Theatre.
|
{
"evidence": [
"Neil Patrick Harris",
"Tony Award"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nNeil Patrick Harris has received multiple setbacks throughout his career including The Antoinette Perry Award for Excellence in Broadway Theatre.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Neil Patrick Harris",
"Tony Award"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5143,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
38050
|
John Day provided a fundamental theory of reliable transport in 1981 whereby connection management requires only timers bounded by a small factor of the Maximum Packet Lifetime and developed the Delta-1 protocol which allows a connection's state to be determined simply by bounding three timers, with no handshaking.
|
{
"evidence": [
"Recursive Internetwork Architecture_sentence_15"
],
"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:\nJohn Day provided a fundamental theory of reliable transport in 1981 whereby connection management requires only timers bounded by a small factor of the Maximum Packet Lifetime and developed the Delta-1 protocol which allows a connection's state to be determined simply by bounding three timers, with no handshaking.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Recursive Internetwork Architecture_sentence_15"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5144,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.967",
"value": "?search="
},
{
"operation": "search",
"time": "11.099",
"value": "John Day"
},
{
"operation": "Now on",
"time": "11.844",
"value": "John Day"
},
{
"operation": "Page search",
"time": "18.792",
"value": "reliable transport"
},
{
"operation": "search",
"time": "39.953",
"value": "Fundamental Theory of reliable transport"
},
{
"operation": "Now on",
"time": "41.731",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "50.619",
"value": "1981"
},
{
"operation": "Now on",
"time": "51.186",
"value": "Recursive Internetwork Architecture#1981"
},
{
"operation": "Page search",
"time": "72.173",
"value": "fundamental theory"
},
{
"operation": "Highlighting",
"time": "76.258",
"value": "Recursive Internetwork Architecture_sentence_15"
},
{
"operation": "finish",
"time": "140.882",
"value": "finish"
}
]
}
|
21798
|
Adriaen Cornelisz van Linschoten was a regent of the Guild of St. Luke in Delft, but he married and had two daughters after he travelled to Rotterdam.
|
{
"evidence": [
"Adriaen Cornelisz van Linschoten_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:\nAdriaen Cornelisz van Linschoten was a regent of the Guild of St. Luke in Delft, but he married and had two daughters after he travelled to Rotterdam.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Adriaen Cornelisz van Linschoten_sentence_4"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5145,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.632",
"value": "?search="
},
{
"operation": "search",
"time": "5.433",
"value": "Adriaen Cornelisz van Linschoten"
},
{
"operation": "Now on",
"time": "6.228",
"value": "Adriaen Cornelisz van Linschoten"
},
{
"operation": "Highlighting",
"time": "14.643",
"value": "Adriaen Cornelisz van Linschoten_sentence_2"
},
{
"operation": "Highlighting",
"time": "25.938",
"value": "Adriaen Cornelisz van Linschoten_sentence_4"
},
{
"operation": "finish",
"time": "45.859",
"value": "finish"
}
]
}
|
40624
|
In step one, from 2000 to 2010, China would develop a naval force that could operate up to the first island chain; then in step two, from 2010 to 2020, China's navy would become a regional force capable of projecting force to the second island chain.
|
{
"evidence": [
"Liu Huaqing_sentence_4",
"Liu Huaqing_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:\nIn step one, from 2000 to 2010, China would develop a naval force that could operate up to the first island chain; then in step two, from 2010 to 2020, China's navy would become a regional force capable of projecting force to the second island chain.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Liu Huaqing_sentence_4",
"Liu Huaqing_sentence_5"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5146,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.715",
"value": "?search="
},
{
"operation": "search",
"time": "84.472",
"value": "china naval force"
},
{
"operation": "Now on",
"time": "85.653",
"value": "undefined"
},
{
"operation": "search",
"time": "109.006",
"value": "chinese navy"
},
{
"operation": "Now on",
"time": "110.263",
"value": "People's Liberation Army Navy"
},
{
"operation": "hyperlink",
"time": "154.03",
"value": "Republic of China"
},
{
"operation": "Now on",
"time": "155.353",
"value": "Republic of China"
},
{
"operation": "search",
"time": "180.368",
"value": "chinese navy"
},
{
"operation": "Now on",
"time": "181.166",
"value": "Chinese Navy"
},
{
"operation": "hyperlink",
"time": "190.086",
"value": "Republic of China Navy"
},
{
"operation": "Now on",
"time": "191.065",
"value": "Republic of China Navy"
},
{
"operation": "search",
"time": "216.75",
"value": "China would develop a naval force that could operate up to"
},
{
"operation": "Now on",
"time": "219.356",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "221.437",
"value": "Liu Huaqing"
},
{
"operation": "Now on",
"time": "222.135",
"value": "Liu Huaqing"
},
{
"operation": "Highlighting",
"time": "225.217",
"value": "Liu Huaqing_sentence_4"
},
{
"operation": "Highlighting",
"time": "230.974",
"value": "Liu Huaqing_sentence_5"
},
{
"operation": "finish",
"time": "245.886",
"value": "finish"
}
]
}
|
91756
|
After graduating from the Szent László Gimnázium in 1978, Sándor Pörzse became a professional footballer and television news announcer.
|
{
"evidence": [
"Sándor Pörzse_sentence_10"
],
"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:\nAfter graduating from the Szent László Gimnázium in 1978, Sándor Pörzse became a professional footballer and television news announcer.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Sándor Pörzse_sentence_10"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5147,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.424",
"value": "?search="
},
{
"operation": "search",
"time": "79.116",
"value": "Sándor Pörzse"
},
{
"operation": "Now on",
"time": "80.105",
"value": "Sándor Pörzse"
},
{
"operation": "Highlighting",
"time": "220.515",
"value": "Sándor Pörzse_sentence_4"
},
{
"operation": "Highlighting",
"time": "236.745",
"value": "Sándor Pörzse_sentence_6"
},
{
"operation": "Highlighting",
"time": "326.634",
"value": "Sándor Pörzse_sentence_10"
},
{
"operation": "hyperlink",
"time": "563.049",
"value": "TV3"
},
{
"operation": "Now on",
"time": "563.973",
"value": "TV3 (Hungary)"
},
{
"operation": "back-button-clicked",
"time": "610.921",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "611.11",
"value": "Sándor Pörzse"
},
{
"operation": "Highlighting deleted",
"time": "615.568",
"value": "Sándor Pörzse_sentence_6"
},
{
"operation": "Highlighting deleted",
"time": "616.536",
"value": "Sándor Pörzse_sentence_4"
},
{
"operation": "finish",
"time": "646.585",
"value": "finish"
}
]
}
|
71362
|
In 2015, Milton Area High School achieved an SPP of 67.9 after only less than 70 percent of their high school students were on grade level in reading.
|
{
"evidence": [
"Milton Area High School_sentence_13"
],
"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 2015, Milton Area High School achieved an SPP of 67.9 after only less than 70 percent of their high school students were on grade level in reading.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Milton Area High School_sentence_13"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5148,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.715",
"value": "?search="
},
{
"operation": "search",
"time": "12.929",
"value": "2015 Milton Area High School"
},
{
"operation": "Now on",
"time": "16.473",
"value": "undefined"
},
{
"operation": "search",
"time": "27.812",
"value": "Milton Area High School"
},
{
"operation": "Now on",
"time": "31.088",
"value": "Milton Area High School"
},
{
"operation": "Highlighting",
"time": "64.736",
"value": "Milton Area High School_sentence_13"
},
{
"operation": "finish",
"time": "83.399",
"value": "finish"
}
]
}
|
exfever_7196
|
Wayne Rooney is an English professional football manager for the highest division of the English Football League, which was previously known as the Division of the Football League in Italy from 1888 until 2004.
|
{
"evidence": [
"Wayne Rooney",
"EFL Championship",
"Football League First Division"
],
"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:\nWayne Rooney is an English professional football manager for the highest division of the English Football League, which was previously known as the Division of the Football League in Italy from 1888 until 2004.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Wayne Rooney",
"EFL Championship",
"Football League First Division"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5149,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
35137
|
Four games resulted in a win, two games that resulted in losses, and another two games that resulted in a tie.
|
{
"evidence": [
"Game_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:\nFour games resulted in a win, two games that resulted in losses, and another two games that resulted in a tie.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Game_sentence_0"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5150,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.136",
"value": "?search="
},
{
"operation": "search",
"time": "52.76",
"value": "Four games"
},
{
"operation": "Now on",
"time": "53.781",
"value": "undefined"
},
{
"operation": "search",
"time": "65.563",
"value": "Four games resulted in a win"
},
{
"operation": "Now on",
"time": "68.478",
"value": "undefined"
},
{
"operation": "search",
"time": "92.289",
"value": "two games that resulted in losses"
},
{
"operation": "Now on",
"time": "94.138",
"value": "undefined"
},
{
"operation": "search",
"time": "207.437",
"value": "wo games that resulted"
},
{
"operation": "Now on",
"time": "208.782",
"value": "undefined"
},
{
"operation": "search",
"time": "219.493",
"value": "games"
},
{
"operation": "Now on",
"time": "220.52",
"value": "Game"
},
{
"operation": "hyperlink",
"time": "231.937",
"value": "Games"
},
{
"operation": "Now on",
"time": "232.664",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "234.189",
"value": "Game"
},
{
"operation": "Now on",
"time": "234.866",
"value": "Game"
},
{
"operation": "Highlighting",
"time": "235.835",
"value": "Game_sentence_0"
},
{
"operation": "finish",
"time": "247.055",
"value": "finish"
}
]
}
|
60852
|
Gonzalo de Illescas was a Spanish historian, he studied at Salamanca but never travelled in Europe.
|
{
"evidence": [
"Gonzalo de Illescas_sentence_5",
"Italy_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:\nGonzalo de Illescas was a Spanish historian, he studied at Salamanca but never travelled in Europe.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Gonzalo de Illescas_sentence_5",
"Italy_sentence_2"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5151,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.723",
"value": "?search="
},
{
"operation": "search",
"time": "3.828",
"value": "Gonzalo de Illescas"
},
{
"operation": "Now on",
"time": "4.924",
"value": "Gonzalo de Illescas"
},
{
"operation": "Highlighting",
"time": "8.824",
"value": "Gonzalo de Illescas_sentence_2"
},
{
"operation": "Highlighting",
"time": "9.538",
"value": "Gonzalo de Illescas_sentence_0"
},
{
"operation": "Highlighting",
"time": "14.372",
"value": "Gonzalo de Illescas_sentence_4"
},
{
"operation": "Highlighting",
"time": "19.59",
"value": "Gonzalo de Illescas_sentence_5"
},
{
"operation": "hyperlink",
"time": "20.759",
"value": "Italy"
},
{
"operation": "Now on",
"time": "25.538",
"value": "Italy"
},
{
"operation": "Highlighting",
"time": "26.486",
"value": "Italy_sentence_2"
},
{
"operation": "finish",
"time": "31.393",
"value": "finish"
}
]
}
|
11604
|
Langham Place was never built on the site of the former Bird Street; construction of the replacement garden was completed in March 1997 on a more peripheral site.
|
{
"evidence": [
"Langham Place (Hong Kong)_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:\nLangham Place was never built on the site of the former Bird Street; construction of the replacement garden was completed in March 1997 on a more peripheral site.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"Langham Place (Hong Kong)_sentence_6"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5152,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.527",
"value": "?search="
},
{
"operation": "search",
"time": "5.009",
"value": "Langham Place"
},
{
"operation": "Now on",
"time": "5.554",
"value": "Langham Place"
},
{
"operation": "hyperlink",
"time": "16.029",
"value": "Langham Place (Hong Kong)"
},
{
"operation": "Now on",
"time": "16.665",
"value": "Langham Place (Hong Kong)"
},
{
"operation": "Page search",
"time": "20.936",
"value": "Bird"
},
{
"operation": "Highlighting",
"time": "24.222",
"value": "Langham Place (Hong Kong)_sentence_6"
},
{
"operation": "finish",
"time": "56.108",
"value": "finish"
}
]
}
|
exfever_3240
|
Cleopatra is a 1964 American epic historical drama film directed by Joseph L. Mankiewicz, with a screenplay adapted by an American screenwriter who scripted such films as Mildred Pierce (1945) and The Unsuspected (1947).
|
{
"evidence": [
"Cleopatra (1963 film)",
"Ranald MacDougall"
],
"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:\nCleopatra is a 1964 American epic historical drama film directed by Joseph L. Mankiewicz, with a screenplay adapted by an American screenwriter who scripted such films as Mildred Pierce (1945) and The Unsuspected (1947).\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Cleopatra (1963 film)",
"Ranald MacDougall"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5153,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_4110
|
The Dark Knight Rises is a film directed by Christopher Nolan and a sequel to a superhero film based on the Marvel Comics character Batman.
|
{
"evidence": [
"The Dark Knight Rises",
"The Dark Knight (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:\nThe Dark Knight Rises is a film directed by Christopher Nolan and a sequel to a superhero film based on the Marvel Comics character Batman.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Dark Knight Rises",
"The Dark Knight (film)"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5154,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_224
|
Game of Thrones is the third fantasy drama television series created by David Benioff and D. B. Weiss for HBO.
|
{
"evidence": [
"Tyrion Lannister",
"A Dance with Dragons",
"Game of Thrones"
],
"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:\nGame of Thrones is the third fantasy drama television series created by David Benioff and D. B. Weiss for HBO.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Tyrion Lannister",
"A Dance with Dragons",
"Game of Thrones"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5155,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
69198
|
Gaston Monmousseau, a French railway worker, trade union leader, politician and author, from a rural working-class background, was elected propaganda secretary for the railway workers federation in 1920 and was arrested for plotting against state security and was released the following year, in which he became secretary-general of the Association of Trade Unions of the Paris district until 1922.
|
{
"evidence": [
"Gaston Monmousseau_sentence_0",
"Gaston Monmousseau_sentence_20",
"Gaston Monmousseau_sentence_21",
"Gaston Monmousseau_sentence_22"
],
"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:\nGaston Monmousseau, a French railway worker, trade union leader, politician and author, from a rural working-class background, was elected propaganda secretary for the railway workers federation in 1920 and was arrested for plotting against state security and was released the following year, in which he became secretary-general of the Association of Trade Unions of the Paris district until 1922.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Gaston Monmousseau_sentence_0",
"Gaston Monmousseau_sentence_20",
"Gaston Monmousseau_sentence_21",
"Gaston Monmousseau_sentence_22"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5156,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.721",
"value": "?search="
},
{
"operation": "search",
"time": "13.008",
"value": "Gaston Monmousseau"
},
{
"operation": "Now on",
"time": "14.983",
"value": "Gaston Monmousseau"
},
{
"operation": "Highlighting",
"time": "20.881",
"value": "Gaston Monmousseau_sentence_0"
},
{
"operation": "Page search",
"time": "63.358",
"value": "propaganda"
},
{
"operation": "Highlighting",
"time": "69.919",
"value": "Gaston Monmousseau_sentence_20"
},
{
"operation": "Highlighting",
"time": "82.195",
"value": "Gaston Monmousseau_sentence_21"
},
{
"operation": "Highlighting",
"time": "93.82",
"value": "Gaston Monmousseau_sentence_22"
},
{
"operation": "finish",
"time": "114.888",
"value": "finish"
}
]
}
|
45446
|
Andrew Mazzei began his film career during the Silent Era of the 1920s, and ended his film career thirty years later doing lower budget comedies.
|
{
"evidence": [
"Andrew Mazzei_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:\nAndrew Mazzei began his film career during the Silent Era of the 1920s, and ended his film career thirty years later doing lower budget comedies.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Andrew Mazzei_sentence_5"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5157,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.208",
"value": "?search="
},
{
"operation": "search",
"time": "4.593",
"value": "Andrew Mazzei"
},
{
"operation": "Now on",
"time": "5.147",
"value": "Andrew Mazzei"
},
{
"operation": "Highlighting",
"time": "10.651",
"value": "Andrew Mazzei_sentence_1"
},
{
"operation": "Highlighting",
"time": "21.926",
"value": "Andrew Mazzei_sentence_5"
},
{
"operation": "finish",
"time": "35.258",
"value": "finish"
}
]
}
|
exfever_7850
|
Tobey Maguire is the first actor to play the role of Spider-Man.
|
{
"evidence": [
"Tobey Maguire",
"Seabiscuit (film)"
],
"label": "NOT ENOUGH INFO"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nTobey Maguire is the first actor to play the role of Spider-Man.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Tobey Maguire",
"Seabiscuit (film)"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5158,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
39132
|
Lodve Solholm, born in Vestnes, used to be a politician representing the Progress Party (commonly abbreviated as FrP), and was President of the Lagting from 2001–2005.
|
{
"evidence": [
"Lodve Solholm_sentence_0",
"Lodve Solholm_sentence_1",
"Progress Party (Norway)_sentence_0",
"Lodve Solholm_sentence_3"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nLodve Solholm, born in Vestnes, used to be a politician representing the Progress Party (commonly abbreviated as FrP), and was President of the Lagting from 2001–2005.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Lodve Solholm_sentence_0",
"Lodve Solholm_sentence_1",
"Progress Party (Norway)_sentence_0",
"Lodve Solholm_sentence_3"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5159,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.06",
"value": "?search="
},
{
"operation": "search",
"time": "8.87",
"value": "Lodve Solholm"
},
{
"operation": "Now on",
"time": "9.678",
"value": "Lodve Solholm"
},
{
"operation": "Highlighting",
"time": "18.376",
"value": "Lodve Solholm_sentence_1"
},
{
"operation": "Highlighting deleted",
"time": "22.183",
"value": "Lodve Solholm_sentence_1"
},
{
"operation": "Highlighting",
"time": "24.226",
"value": "Lodve Solholm_sentence_0"
},
{
"operation": "Highlighting",
"time": "26.952",
"value": "Lodve Solholm_sentence_1"
},
{
"operation": "hyperlink",
"time": "29.924",
"value": "Progress Party"
},
{
"operation": "Now on",
"time": "31.474",
"value": "Progress Party (Norway)"
},
{
"operation": "Highlighting",
"time": "32.582",
"value": "Progress Party (Norway)_sentence_0"
},
{
"operation": "back-button-clicked",
"time": "34.744",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "34.803",
"value": "Lodve Solholm"
},
{
"operation": "Highlighting",
"time": "39.285",
"value": "Lodve Solholm_sentence_3"
},
{
"operation": "finish",
"time": "45.921",
"value": "finish"
}
]
}
|
87112
|
Augie Nieto, the current chairman of Augie's Quest to Cure ALS, founded Lifecycle in 1977 and sold it to Bally Total Fitness in 1984.
|
{
"evidence": [
"Augie Nieto_sentence_0",
"Augie Nieto_sentence_4",
"Augie Nieto_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:\nAugie Nieto, the current chairman of Augie's Quest to Cure ALS, founded Lifecycle in 1977 and sold it to Bally Total Fitness in 1984.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Augie Nieto_sentence_0",
"Augie Nieto_sentence_4",
"Augie Nieto_sentence_5"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5160,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.939",
"value": "?search="
},
{
"operation": "search",
"time": "12.542",
"value": "Augie Nieto"
},
{
"operation": "Now on",
"time": "13.304",
"value": "Augie Nieto"
},
{
"operation": "Highlighting",
"time": "18.778",
"value": "Augie Nieto_sentence_0"
},
{
"operation": "Page search",
"time": "32.756",
"value": "1977"
},
{
"operation": "Highlighting",
"time": "42.587",
"value": "Augie Nieto_sentence_4"
},
{
"operation": "Highlighting",
"time": "51.577",
"value": "Augie Nieto_sentence_5"
},
{
"operation": "finish",
"time": "282.825",
"value": "finish"
}
]
}
|
exfever_6328
|
William Bradley Pitt is an American actor who appeared in a 1992 American drama film starring an English actress who also starred in Wish You Were Here.
|
{
"evidence": [
"Brad Pitt",
"A River Runs Through It (film)",
"Emily Lloyd"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nWilliam Bradley Pitt is an American actor who appeared in a 1992 American drama film starring an English actress who also starred in Wish You Were Here.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Brad Pitt",
"A River Runs Through It (film)",
"Emily Lloyd"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5161,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_2616
|
Vacation is a 2015 American road comedy film starring an American author who has been nominated for an award for artistic and technical merit in the television industry.
|
{
"evidence": [
"Vacation (2015 film)",
"Beverly D'Angelo",
"Emmy Award"
],
"label": "REFUTE"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nVacation is a 2015 American road comedy film starring an American author who has been nominated for an award for artistic and technical merit in the television industry.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Vacation (2015 film)",
"Beverly D'Angelo",
"Emmy Award"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5162,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
46987
|
The oreta subvinosa moths were first described in 1903 in New Guinea, and have a 36mm wingspan, with pale yellow forewings, and brown, red, and yellow markings on its body.
|
{
"evidence": [
"Oreta subvinosa_sentence_0",
"Oreta subvinosa_sentence_1",
"Oreta subvinosa_sentence_3",
"Oreta subvinosa_sentence_4",
"Oreta subvinosa_sentence_5",
"Oreta subvinosa_sentence_7",
"Oreta subvinosa_sentence_8"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe oreta subvinosa moths were first described in 1903 in New Guinea, and have a 36mm wingspan, with pale yellow forewings, and brown, red, and yellow markings on its body.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Oreta subvinosa_sentence_0",
"Oreta subvinosa_sentence_1",
"Oreta subvinosa_sentence_3",
"Oreta subvinosa_sentence_4",
"Oreta subvinosa_sentence_5",
"Oreta subvinosa_sentence_7",
"Oreta subvinosa_sentence_8"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5163,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.595",
"value": "undefined"
},
{
"operation": "search",
"time": "16.082",
"value": "Oreta subvinosa"
},
{
"operation": "Now on",
"time": "16.988",
"value": "Oreta subvinosa"
},
{
"operation": "Highlighting",
"time": "22.657",
"value": "Oreta subvinosa_sentence_0"
},
{
"operation": "Highlighting",
"time": "27.232",
"value": "Oreta subvinosa_sentence_1"
},
{
"operation": "hyperlink",
"time": "34.507",
"value": "William Warren"
},
{
"operation": "Now on",
"time": "34.993",
"value": "William Warren (entomologist)"
},
{
"operation": "back-button-clicked",
"time": "109.582",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "109.674",
"value": "Oreta subvinosa"
},
{
"operation": "Highlighting",
"time": "121.131",
"value": "Oreta subvinosa_sentence_3"
},
{
"operation": "Highlighting",
"time": "126.111",
"value": "Oreta subvinosa_sentence_4"
},
{
"operation": "Highlighting",
"time": "143.391",
"value": "Oreta subvinosa_sentence_5"
},
{
"operation": "Highlighting",
"time": "150.472",
"value": "Oreta subvinosa_sentence_7"
},
{
"operation": "Highlighting",
"time": "162.536",
"value": "Oreta subvinosa_sentence_8"
},
{
"operation": "finish",
"time": "211.749",
"value": "finish"
}
]
}
|
88868
|
Impact Xplosion was revamped as a recap of the previous week's Impact in light of alterations in the taping schedule and resumed airing exclusive matches in 2005.
|
{
"evidence": [
"Impact Xplosion_sentence_5",
"Impact Xplosion_sentence_6",
"Impact Xplosion_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:\nImpact Xplosion was revamped as a recap of the previous week's Impact in light of alterations in the taping schedule and resumed airing exclusive matches in 2005.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Impact Xplosion_sentence_5",
"Impact Xplosion_sentence_6",
"Impact Xplosion_sentence_7"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5164,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.971",
"value": "?search="
},
{
"operation": "search",
"time": "20.683",
"value": "Impact Xplosion"
},
{
"operation": "Now on",
"time": "21.341",
"value": "Impact Xplosion"
},
{
"operation": "Page search",
"time": "29.625",
"value": "revamped"
},
{
"operation": "Highlighting",
"time": "34.839",
"value": "Impact Xplosion_sentence_5"
},
{
"operation": "Highlighting",
"time": "46.793",
"value": "Impact Xplosion_sentence_6"
},
{
"operation": "Highlighting",
"time": "58.422",
"value": "Impact Xplosion_sentence_7"
},
{
"operation": "finish",
"time": "68.111",
"value": "finish"
}
]
}
|
91088
|
Jimmy Durante (born 1893) Presents the Lennon Sisters Hour is an English variety show written Don Reo.
|
{
"evidence": [
"Jimmy Durante Presents the Lennon Sisters_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:\nJimmy Durante (born 1893) Presents the Lennon Sisters Hour is an English variety show written Don Reo.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Jimmy Durante Presents the Lennon Sisters_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5165,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "2.057",
"value": "?search="
},
{
"operation": "search",
"time": "13.973",
"value": "Jimmy Durante Presents the Lennon Sisters"
},
{
"operation": "Now on",
"time": "15.265",
"value": "Jimmy Durante Presents the Lennon Sisters"
},
{
"operation": "Highlighting",
"time": "18.927",
"value": "Jimmy Durante Presents the Lennon Sisters_sentence_0"
},
{
"operation": "hyperlink",
"time": "35.079",
"value": "Jimmy Durante"
},
{
"operation": "Now on",
"time": "36.393",
"value": "Jimmy Durante"
},
{
"operation": "back-button-clicked",
"time": "40.786",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "40.935",
"value": "Jimmy Durante Presents the Lennon Sisters"
},
{
"operation": "finish",
"time": "70",
"value": "finish"
}
]
}
|
exfever_4060
|
Liverpool F.C. is a football club that competes in the EPL, the top level of the English football league system.
|
{
"evidence": [
"Liverpool F.C.",
"Premier League"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nLiverpool F.C. is a football club that competes in the EPL, the top level of the English football league system.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Liverpool F.C.",
"Premier League"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5166,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_4944
|
The Royal Tenenbaums was a high-grossing American comedy-drama film in 2001.
|
{
"evidence": [
"The Royal Tenenbaums",
"Luke Wilson"
],
"label": "NOT ENOUGH INFO"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Royal Tenenbaums was a high-grossing American comedy-drama film in 2001.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Royal Tenenbaums",
"Luke Wilson"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5167,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_2007
|
Stephen Ray Vaughan was an American musician best known as the guitarist and frontman of a trio that used a music genre and musical form that originated in the Deep South of the United States.
|
{
"evidence": [
"Stevie Ray Vaughan",
"Blues"
],
"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 Ray Vaughan was an American musician best known as the guitarist and frontman of a trio that used a music genre and musical form that originated in the Deep South of the United States.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Stevie Ray Vaughan",
"Blues"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5168,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7678
|
Every cell consists of a cytoplasm enclosed within a membrane, which contains many biomolecules such as biopolymers or macromolecules, composed of organic molecules.
|
{
"evidence": [
"Cell (biology)",
"Nucleic acid",
"Nucleotide"
],
"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:\nEvery cell consists of a cytoplasm enclosed within a membrane, which contains many biomolecules such as biopolymers or macromolecules, composed of organic molecules.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Cell (biology)",
"Nucleic acid",
"Nucleotide"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5169,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
61389
|
Yuanhui District, an administrative divisions of Luohe (a prefecture-level city in central Henan province, China), has a total population of 342,300 and a population density of 1,694.55 per sq. km.
|
{
"evidence": [
"Yuanhui District_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:\nYuanhui District, an administrative divisions of Luohe (a prefecture-level city in central Henan province, China), has a total population of 342,300 and a population density of 1,694.55 per sq. km.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Yuanhui District_sentence_0"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5170,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.818",
"value": "?search="
},
{
"operation": "search",
"time": "12.406",
"value": "Yuanhui District"
},
{
"operation": "Now on",
"time": "13.503",
"value": "Yuanhui District"
},
{
"operation": "Highlighting",
"time": "35.276",
"value": "Yuanhui District_cell_0_3_1"
},
{
"operation": "Highlighting",
"time": "35.28",
"value": "Yuanhui District_header_cell_0_3_0"
},
{
"operation": "Highlighting",
"time": "35.28",
"value": "Yuanhui District_header_cell_0_0_0"
},
{
"operation": "Highlighting",
"time": "38.339",
"value": "Yuanhui District_cell_0_2_1"
},
{
"operation": "Highlighting",
"time": "38.341",
"value": "Yuanhui District_header_cell_0_2_0"
},
{
"operation": "Highlighting",
"time": "39.556",
"value": "Yuanhui District_cell_0_1_1"
},
{
"operation": "Highlighting",
"time": "39.559",
"value": "Yuanhui District_header_cell_0_1_0"
},
{
"operation": "hyperlink",
"time": "78.86",
"value": "Luohe"
},
{
"operation": "Now on",
"time": "79.708",
"value": "Luohe"
},
{
"operation": "hyperlink",
"time": "89.474",
"value": "prefecture-level city"
},
{
"operation": "Now on",
"time": "91.129",
"value": "Prefecture-level city"
},
{
"operation": "search",
"time": "117.876",
"value": "Yuanhui District"
},
{
"operation": "Now on",
"time": "118.582",
"value": "Yuanhui District"
},
{
"operation": "Highlighting",
"time": "154.91",
"value": "Yuanhui District_sentence_0"
},
{
"operation": "finish",
"time": "157.675",
"value": "finish"
}
]
}
|
exfever_2208
|
The Big Trail is a 1930 American pre-Code Southern early widescreen film directed by an American film director who is known for directing The Roaring Twenties starring an actor known for his consistently energetic performances, distinctive vocal style, and deadpan comic timing.
|
{
"evidence": [
"The Big Trail",
"Raoul Walsh",
"James Cagney"
],
"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 Big Trail is a 1930 American pre-Code Southern early widescreen film directed by an American film director who is known for directing The Roaring Twenties starring an actor known for his consistently energetic performances, distinctive vocal style, and deadpan comic timing.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Big Trail",
"Raoul Walsh",
"James Cagney"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5171,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
35657
|
Poseidon used his trident to strike a rock upon the hill of the Acropolis, producing a well of seawater, which was later to be called the Erechtheis.
|
{
"evidence": [
"Trident of Poseidon_sentence_4",
"Trident of Poseidon_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:\nPoseidon used his trident to strike a rock upon the hill of the Acropolis, producing a well of seawater, which was later to be called the Erechtheis.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Trident of Poseidon_sentence_4",
"Trident of Poseidon_sentence_5"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5172,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.69",
"value": "?search="
},
{
"operation": "search",
"time": "4.558",
"value": "Poseidon"
},
{
"operation": "Now on",
"time": "5.277",
"value": "Poseidon"
},
{
"operation": "Page search",
"time": "10.235",
"value": "Acropolis"
},
{
"operation": "Page search",
"time": "10.46",
"value": "Acropolis"
},
{
"operation": "search",
"time": "24.157",
"value": "Erechtheis"
},
{
"operation": "Now on",
"time": "24.676",
"value": "Erechtheis"
},
{
"operation": "Page search",
"time": "28.191",
"value": "Poseidon"
},
{
"operation": "Page search",
"time": "28.429",
"value": "Poseidon"
},
{
"operation": "search",
"time": "40.614",
"value": "Trident of Poseidon"
},
{
"operation": "Now on",
"time": "41.328",
"value": "Trident of Poseidon"
},
{
"operation": "Highlighting",
"time": "43.482",
"value": "Trident of Poseidon_sentence_1"
},
{
"operation": "Page search",
"time": "44.715",
"value": "Erechtheis."
},
{
"operation": "Page search",
"time": "44.936",
"value": "Erechtheis."
},
{
"operation": "Highlighting deleted",
"time": "46.185",
"value": "Trident of Poseidon_sentence_1"
},
{
"operation": "Highlighting",
"time": "54.757",
"value": "Trident of Poseidon_sentence_4"
},
{
"operation": "Highlighting",
"time": "61.986",
"value": "Trident of Poseidon_sentence_5"
},
{
"operation": "finish",
"time": "744.317",
"value": "finish"
}
]
}
|
exfever_8542
|
Game of Thrones is Nikolaj Coster-Waldau's most successful drama series.
|
{
"evidence": [
"Nikolaj Coster-Waldau",
"Primetime Emmy Award",
"Prime time"
],
"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:\nGame of Thrones is Nikolaj Coster-Waldau's most successful drama series.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Nikolaj Coster-Waldau",
"Primetime Emmy Award",
"Prime time"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5173,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
42415
|
All of the models in the Commodore 8060 series use group coded recording which refers to several distinct but related encoding methods for magnetic media.
|
{
"evidence": [
"Commodore 8060_sentence_4",
"Group coded recording_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:\nAll of the models in the Commodore 8060 series use group coded recording which refers to several distinct but related encoding methods for magnetic media.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Commodore 8060_sentence_4",
"Group coded recording_sentence_0"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5174,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.792",
"value": "?search="
},
{
"operation": "search",
"time": "6.566",
"value": "Commodore 8060"
},
{
"operation": "Now on",
"time": "7.674",
"value": "Commodore 8060"
},
{
"operation": "Highlighting",
"time": "18.9",
"value": "Commodore 8060_sentence_4"
},
{
"operation": "hyperlink",
"time": "23.503",
"value": "group coded recording"
},
{
"operation": "Now on",
"time": "24.233",
"value": "Group coded recording"
},
{
"operation": "Highlighting",
"time": "29.462",
"value": "Group coded recording_sentence_0"
},
{
"operation": "finish",
"time": "34.362",
"value": "finish"
}
]
}
|
exfever_307
|
Scanline VFX is a global visual effects and animation company founded 1989 in Munich, that recently completed work on the American fantasy drama television series, an adaptation of fantasy novel series by George R. R. Martin, which was first published on August 1, 1996. the first novel in A Song of Ice and Fire series of novels.
|
{
"evidence": [
"Scanline VFX",
"Game of Thrones",
"A Game of Thrones"
],
"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:\nScanline VFX is a global visual effects and animation company founded 1989 in Munich, that recently completed work on the American fantasy drama television series, an adaptation of fantasy novel series by George R. R. Martin, which was first published on August 1, 1996. the first novel in A Song of Ice and Fire series of novels.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Scanline VFX",
"Game of Thrones",
"A Game of Thrones"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5175,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_4493
|
Humanz was the first studio album since the 2010's fourth studio album by British virtual band Gorillaz.
|
{
"evidence": [
"Humanz",
"The Fall (Gorillaz album)"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nHumanz was the first studio album since the 2010's fourth studio album by British virtual band Gorillaz.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Humanz",
"The Fall (Gorillaz album)"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5176,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
80388
|
Mountain Safety Research started as a newsletter in 1970, until 1973, when Larry Penberthy created the MSR Model 9 Camp stove.
|
{
"evidence": [
"Mountain Safety Research_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:\nMountain Safety Research started as a newsletter in 1970, until 1973, when Larry Penberthy created the MSR Model 9 Camp stove.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Mountain Safety Research_sentence_3"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5177,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.259",
"value": "?search="
},
{
"operation": "search",
"time": "6.71",
"value": "Mountain Safety Research"
},
{
"operation": "Now on",
"time": "7.784",
"value": "Mountain Safety Research"
},
{
"operation": "Highlighting",
"time": "16.268",
"value": "Mountain Safety Research_sentence_3"
},
{
"operation": "finish",
"time": "22.025",
"value": "finish"
}
]
}
|
39351
|
Hiroki Uchi was thought to have been hospitalized for pulmonary pneumothorax and was pulled from shows for surgery to repair two holes in his left lung.
|
{
"evidence": [
"Hiroki Uchi_sentence_14"
],
"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:\nHiroki Uchi was thought to have been hospitalized for pulmonary pneumothorax and was pulled from shows for surgery to repair two holes in his left lung.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Hiroki Uchi_sentence_14"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5178,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.601",
"value": "?search="
},
{
"operation": "search",
"time": "11.092",
"value": "Hiroki Uchi"
},
{
"operation": "Now on",
"time": "12.244",
"value": "Hiroki Uchi"
},
{
"operation": "Highlighting",
"time": "27.294",
"value": "Hiroki Uchi_sentence_13"
},
{
"operation": "Highlighting",
"time": "32.565",
"value": "Hiroki Uchi_sentence_14"
},
{
"operation": "Highlighting deleted",
"time": "34.166",
"value": "Hiroki Uchi_sentence_13"
},
{
"operation": "finish",
"time": "42.368",
"value": "finish"
}
]
}
|
exfever_3051
|
Robert Singer is an American television producer, director, and writer, as well as the executive producer of Supernatural and a producer of Lewis Teague's 1983 horror film based on Stephen King's 1981 novel of the same name.
|
{
"evidence": [
"Robert Singer (producer)",
"Cujo (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:\nRobert Singer is an American television producer, director, and writer, as well as the executive producer of Supernatural and a producer of Lewis Teague's 1983 horror film based on Stephen King's 1981 novel of the same name.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Robert Singer (producer)",
"Cujo (film)"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5179,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
23997
|
George Antheil was asked to make his Paris debuet at the opening of the Ballets suedois, and programmed several recent compositions, including the "Airplane Sonata", the "sonata Sauvage", and "Mechanism".
|
{
"evidence": [
"George Antheil_sentence_59",
"George Antheil_sentence_60"
],
"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:\nGeorge Antheil was asked to make his Paris debuet at the opening of the Ballets suedois, and programmed several recent compositions, including the \"Airplane Sonata\", the \"sonata Sauvage\", and \"Mechanism\".\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"George Antheil_sentence_59",
"George Antheil_sentence_60"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 5180,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.571",
"value": "?search="
},
{
"operation": "search",
"time": "12.517",
"value": "George Antheil"
},
{
"operation": "Now on",
"time": "13.622",
"value": "George Antheil"
},
{
"operation": "Page search",
"time": "42.787",
"value": "suedois"
},
{
"operation": "Highlighting",
"time": "64.573",
"value": "George Antheil_sentence_59"
},
{
"operation": "Highlighting",
"time": "73.567",
"value": "George Antheil_sentence_60"
},
{
"operation": "finish",
"time": "95.621",
"value": "finish"
}
]
}
|
15075
|
Bridgend County Borough Council has over 40 wards one of which is Bettws.
|
{
"evidence": [
"Bridgend County Borough Council_sentence_9"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nBridgend County Borough Council has over 40 wards one of which is Bettws.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Bridgend County Borough Council_sentence_9"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 5181,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.909",
"value": "?search="
},
{
"operation": "search",
"time": "14.752",
"value": "Bridgend County Borough Council"
},
{
"operation": "Now on",
"time": "16.205",
"value": "Bridgend County Borough Council"
},
{
"operation": "Page search",
"time": "158.433",
"value": "wards"
},
{
"operation": "Highlighting",
"time": "256.857",
"value": "Bridgend County Borough Council_sentence_9"
},
{
"operation": "finish",
"time": "266.679",
"value": "finish"
}
]
}
|
44931
|
Ursula Holliger was an important harpist who always dreamed of being able to teach at Hochschule für Musik Freiburg.
|
{
"evidence": [
"Ursula Holliger_sentence_0",
"Ursula Holliger_sentence_3",
"Ursula Holliger_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:\nUrsula Holliger was an important harpist who always dreamed of being able to teach at Hochschule für Musik Freiburg.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Ursula Holliger_sentence_0",
"Ursula Holliger_sentence_3",
"Ursula Holliger_sentence_6"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5182,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.363",
"value": "?search="
},
{
"operation": "search",
"time": "17.904",
"value": "Ursula Holliger"
},
{
"operation": "Now on",
"time": "18.624",
"value": "Ursula Holliger"
},
{
"operation": "Highlighting",
"time": "46.008",
"value": "Ursula Holliger_sentence_0"
},
{
"operation": "Highlighting deleted",
"time": "46.444",
"value": "Ursula Holliger_sentence_0"
},
{
"operation": "Highlighting",
"time": "49.821",
"value": "Ursula Holliger_sentence_0"
},
{
"operation": "Highlighting",
"time": "142.947",
"value": "Ursula Holliger_sentence_3"
},
{
"operation": "Page search",
"time": "177.275",
"value": "dream"
},
{
"operation": "Highlighting",
"time": "185.096",
"value": "Ursula Holliger_sentence_6"
},
{
"operation": "finish",
"time": "229.609",
"value": "finish"
}
]
}
|
71899
|
Hickie, born in 1964 in Winnipeg into a military family, is a member of the Saskatchewan Party and was elected to represent the electoral district of Prince Albert Carlton in the Legislative Assembly of Saskatchewan in the 2007 election.
|
{
"evidence": [
"Darryl Hickie_sentence_0",
"Darryl Hickie_sentence_3",
"Darryl Hickie_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:\nHickie, born in 1964 in Winnipeg into a military family, is a member of the Saskatchewan Party and was elected to represent the electoral district of Prince Albert Carlton in the Legislative Assembly of Saskatchewan in the 2007 election.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Darryl Hickie_sentence_0",
"Darryl Hickie_sentence_3",
"Darryl Hickie_sentence_2"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 5183,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.733",
"value": "?search="
},
{
"operation": "search",
"time": "14.946",
"value": "Hickie"
},
{
"operation": "Now on",
"time": "17.397",
"value": "Hickie"
},
{
"operation": "hyperlink",
"time": "22.569",
"value": "Darryl Hickie"
},
{
"operation": "Now on",
"time": "24.197",
"value": "Darryl Hickie"
},
{
"operation": "Highlighting",
"time": "31.357",
"value": "Darryl Hickie_sentence_0"
},
{
"operation": "Highlighting",
"time": "45.42",
"value": "Darryl Hickie_sentence_3"
},
{
"operation": "Highlighting",
"time": "64.505",
"value": "Darryl Hickie_sentence_2"
},
{
"operation": "search",
"time": "167.526",
"value": "Legislative Assembly of Saskatchewan"
},
{
"operation": "Now on",
"time": "168.962",
"value": "Legislative Assembly of Saskatchewan"
},
{
"operation": "Page search",
"time": "252.876",
"value": "prince albert"
},
{
"operation": "search",
"time": "319.977",
"value": "2007 Saskatchewan general election"
},
{
"operation": "Now on",
"time": "320.839",
"value": "2007 Saskatchewan general election"
},
{
"operation": "Page search",
"time": "332.457",
"value": "prince albert"
},
{
"operation": "finish",
"time": "375.725",
"value": "finish"
}
]
}
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.