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 |
|---|---|---|---|---|---|---|---|---|
25393
|
After leaving PS Kemi, Pasher trained with Lansing United of the National Premier Soccer League; then was home for the summer and decided to join the club, of which a good friend was assistant coach.
|
{
"evidence": [
"Tyler Pasher_sentence_19",
"Tyler Pasher_sentence_20"
],
"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:\nAfter leaving PS Kemi, Pasher trained with Lansing United of the National Premier Soccer League; then was home for the summer and decided to join the club, of which a good friend was assistant coach.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Tyler Pasher_sentence_19",
"Tyler Pasher_sentence_20"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6284,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.073",
"value": "?search="
},
{
"operation": "search",
"time": "7.003",
"value": "PS Kemi"
},
{
"operation": "Now on",
"time": "7.929",
"value": "Kemi City F.C."
},
{
"operation": "search",
"time": "27.287",
"value": "Pasher soccer"
},
{
"operation": "Now on",
"time": "28.365",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "31.227",
"value": "Tyler Pasher"
},
{
"operation": "Now on",
"time": "32.236",
"value": "Tyler Pasher"
},
{
"operation": "Highlighting",
"time": "58.621",
"value": "Tyler Pasher_sentence_19"
},
{
"operation": "Highlighting",
"time": "63.638",
"value": "Tyler Pasher_sentence_20"
},
{
"operation": "finish",
"time": "92.247",
"value": "finish"
}
]
}
|
exfever_4740
|
Prison Break revolves around two brothers, including a principal character who plays a prominent role in the series.
|
{
"evidence": [
"Prison Break",
"Lincoln Burrows"
],
"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:\nPrison Break revolves around two brothers, including a principal character who plays a prominent role in the series.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Prison Break",
"Lincoln Burrows"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6285,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_4961
|
The Legend of Tarzan is a 2018 adventure film starring an Australian actress and producer best known for the black comedy film The Wolf of Wall Street.
|
{
"evidence": [
"The Legend of Tarzan (film)",
"Margot Robbie"
],
"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 Legend of Tarzan is a 2018 adventure film starring an Australian actress and producer best known for the black comedy film The Wolf of Wall Street.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"The Legend of Tarzan (film)",
"Margot Robbie"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 6286,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_1262
|
The National Library of India is the second-largest library in Asia.
|
{
"evidence": [
"National Library of India",
"Belvedere Estate"
],
"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 National Library of India is the second-largest library in Asia.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"National Library of India",
"Belvedere Estate"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 6287,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_8801
|
Matthew William Sorum is an American drummer who was also a member of a British-American rock musical group whose members were successful as solo artists or as members of other successful groups.
|
{
"evidence": [
"Matt Sorum",
"Neurotic Outsiders",
"Supergroup (music)"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nMatthew William Sorum is an American drummer who was also a member of a British-American rock musical group whose members were successful as solo artists or as members of other successful groups.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Matt Sorum",
"Neurotic Outsiders",
"Supergroup (music)"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6288,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_976
|
Leonardo da Vinci epitomized the Renaissance humanist ideal through his exceptional intellectual ability.
|
{
"evidence": [
"Leonardo da Vinci",
"Genius"
],
"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:\nLeonardo da Vinci epitomized the Renaissance humanist ideal through his exceptional intellectual ability.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Leonardo da Vinci",
"Genius"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6289,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_1235
|
American Sniper: The Autobiography of the Most Lethal Sniper in U.S. Military History is a work of autobiographical creative nonfiction by United States Navy SEAL Chris Kyle that inspired the 2014 American biographical war drama film directed by an American actor, film director rose to international fame with his role as the "Man with No Name" in Sergio Leone's "Dollars Trilogy" of Spaghetti Westerns during the mid-1960s.
|
{
"evidence": [
"American Sniper (book)",
"American Sniper",
"Clint Eastwood"
],
"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:\nAmerican Sniper: The Autobiography of the Most Lethal Sniper in U.S. Military History is a work of autobiographical creative nonfiction by United States Navy SEAL Chris Kyle that inspired the 2014 American biographical war drama film directed by an American actor, film director rose to international fame with his role as the \"Man with No Name\" in Sergio Leone's \"Dollars Trilogy\" of Spaghetti Westerns during the mid-1960s.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"American Sniper (book)",
"American Sniper",
"Clint Eastwood"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6290,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_3170
|
Saif Ali Khan is an Indian actor who starred in the 1994 Indian action-comedy film directed by Sameer Malkan.
|
{
"evidence": [
"Saif Ali Khan",
"Main Khiladi Tu Anari"
],
"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:\nSaif Ali Khan is an Indian actor who starred in the 1994 Indian action-comedy film directed by Sameer Malkan.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Saif Ali Khan",
"Main Khiladi Tu Anari"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6291,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
48799
|
Tridrepana marginata is a moth in the family Drepanidae and its wingspan is 27-32.2 mm for males and 30.2–39 mm for females.
|
{
"evidence": [
"Tridrepana marginata_sentence_0",
"Tridrepana marginata_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:\nTridrepana marginata is a moth in the family Drepanidae and its wingspan is 27-32.2 mm for males and 30.2–39 mm for females.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"Tridrepana marginata_sentence_0",
"Tridrepana marginata_sentence_3"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6292,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.927",
"value": "?search="
},
{
"operation": "search",
"time": "22.154",
"value": "Tridrepana marginata"
},
{
"operation": "Now on",
"time": "22.999",
"value": "Tridrepana marginata"
},
{
"operation": "Highlighting",
"time": "25.236",
"value": "Tridrepana marginata_sentence_0"
},
{
"operation": "Highlighting",
"time": "40.887",
"value": "Tridrepana marginata_sentence_3"
},
{
"operation": "Highlighting",
"time": "56.837",
"value": "Tridrepana marginata_cell_0_6_0"
},
{
"operation": "Highlighting",
"time": "56.839",
"value": "Tridrepana marginata_header_cell_0_1_0"
},
{
"operation": "Highlighting",
"time": "56.84",
"value": "Tridrepana marginata_header_cell_0_0_0"
},
{
"operation": "Highlighting",
"time": "57.873",
"value": "Tridrepana marginata_cell_0_6_1"
},
{
"operation": "Highlighting",
"time": "63.316",
"value": "Tridrepana marginata_sentence_3"
},
{
"operation": "Highlighting",
"time": "73.968",
"value": "Tridrepana marginata_sentence_0"
},
{
"operation": "Highlighting deleted",
"time": "92.973",
"value": "Tridrepana marginata_cell_0_6_0"
},
{
"operation": "Highlighting deleted",
"time": "100.041",
"value": "Tridrepana marginata_cell_0_6_1"
},
{
"operation": "Highlighting deleted",
"time": "100.044",
"value": "Tridrepana marginata_header_cell_0_1_0"
},
{
"operation": "Highlighting deleted",
"time": "100.045",
"value": "Tridrepana marginata_header_cell_0_0_0"
},
{
"operation": "finish",
"time": "141.628",
"value": "finish"
}
]
}
|
exfever_6287
|
Bradley Cooper is the youngest American actor and filmmaker.
|
{
"evidence": [
"Bradley Cooper",
"Master of Fine Arts"
],
"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:\nBradley Cooper is the youngest American actor and filmmaker.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Bradley Cooper",
"Master of Fine Arts"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 6293,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
94420
|
Judoka must undergo a year long qualifying process, resulting in two judoka per division.
|
{
"evidence": [
"Judo at the 2020 Summer Olympics – Qualification_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:\nJudoka must undergo a year long qualifying process, resulting in two judoka per division.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Judo at the 2020 Summer Olympics – Qualification_sentence_4"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 6294,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.923",
"value": "?search="
},
{
"operation": "search",
"time": "8.258",
"value": "Judoka"
},
{
"operation": "Now on",
"time": "9.586",
"value": "Judo"
},
{
"operation": "Page search",
"time": "18.382",
"value": "qualifying"
},
{
"operation": "Page search",
"time": "36.111",
"value": "year"
},
{
"operation": "Page search",
"time": "121.388",
"value": "division"
},
{
"operation": "hyperlink",
"time": "147.602",
"value": "List of judoka"
},
{
"operation": "Now on",
"time": "148.453",
"value": "List of judoka"
},
{
"operation": "Page search",
"time": "155.951",
"value": "year"
},
{
"operation": "back-button-clicked",
"time": "171.648",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "172.751",
"value": "Judo"
},
{
"operation": "search",
"time": "183.648",
"value": "Judoka"
},
{
"operation": "Now on",
"time": "184.337",
"value": "Judoka"
},
{
"operation": "Page search",
"time": "192.194",
"value": "Judoka"
},
{
"operation": "search",
"time": "272.22",
"value": "Judoka must undergo a year long qualifying process"
},
{
"operation": "Now on",
"time": "275.17",
"value": "undefined"
},
{
"operation": "search",
"time": "318.858",
"value": "two judoka per division"
},
{
"operation": "Now on",
"time": "320.726",
"value": "undefined"
},
{
"operation": "search",
"time": "334.596",
"value": "Contains...Judoka"
},
{
"operation": "Now on",
"time": "335.272",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "351.394",
"value": "Judoka"
},
{
"operation": "Now on",
"time": "351.864",
"value": "Judoka"
},
{
"operation": "back-button-clicked",
"time": "366.49",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "366.986",
"value": "undefined"
},
{
"operation": "search",
"time": "397.634",
"value": "become Judoka"
},
{
"operation": "Now on",
"time": "398.911",
"value": "undefined"
},
{
"operation": "search",
"time": "452.068",
"value": "qualifying process for the judoka"
},
{
"operation": "Now on",
"time": "453.299",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "472.258",
"value": "Judo at the 2020 Summer Olympics – Qualification"
},
{
"operation": "Now on",
"time": "475.718",
"value": "Judo at the 2020 Summer Olympics – Qualification"
},
{
"operation": "Page search",
"time": "480.84",
"value": "judoka"
},
{
"operation": "Page search",
"time": "490.545",
"value": "qualifying"
},
{
"operation": "Highlighting",
"time": "525.333",
"value": "Judo at the 2020 Summer Olympics – Qualification_sentence_4"
},
{
"operation": "finish",
"time": "546.849",
"value": "finish"
}
]
}
|
44217
|
The 2012 Ghanaian parliamentary election in Buem resulted in Henry Ford Kofi Kamel achieving 73.8% of the votes for the National Democratic Congress.
|
{
"evidence": [
"Henry Ford Kamel_sentence_10",
"Henry Ford Kamel_sentence_11",
"Henry Ford Kamel_sentence_7"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe 2012 Ghanaian parliamentary election in Buem resulted in Henry Ford Kofi Kamel achieving 73.8% of the votes for the National Democratic Congress.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Henry Ford Kamel_sentence_10",
"Henry Ford Kamel_sentence_11",
"Henry Ford Kamel_sentence_7"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 6295,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.968",
"value": "?search="
},
{
"operation": "search",
"time": "7.067",
"value": "2012 Ghanaian parliamentary election"
},
{
"operation": "Now on",
"time": "9.237",
"value": "2012 Ghanaian general election"
},
{
"operation": "Page search",
"time": "98.672",
"value": "buem"
},
{
"operation": "hyperlink",
"time": "122.779",
"value": "2012 Ghanaian parliamentary election"
},
{
"operation": "Now on",
"time": "124.254",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "125.486",
"value": "2012 Ghanaian general election"
},
{
"operation": "Now on",
"time": "127.032",
"value": "2012 Ghanaian general election"
},
{
"operation": "search",
"time": "132.258",
"value": "2012 Ghanaian parliamentary election"
},
{
"operation": "Now on",
"time": "133.27",
"value": "2012 Ghanaian parliamentary election"
},
{
"operation": "Page search",
"time": "172.862",
"value": "Henry Ford Kofi Kamel"
},
{
"operation": "Page search",
"time": "177.076",
"value": "Kofi Kamel"
},
{
"operation": "hyperlink",
"time": "181.276",
"value": "List of MPs elected in the 2012 Ghanaian parliamentary election"
},
{
"operation": "Now on",
"time": "182.312",
"value": "List of MPs elected in the 2012 Ghanaian parliamentary election"
},
{
"operation": "Page search",
"time": "190.453",
"value": "Kofi Kamel"
},
{
"operation": "search",
"time": "212.38",
"value": "Henry Ford Kofi Kamel"
},
{
"operation": "Now on",
"time": "213.529",
"value": "Henry Ford Kamel"
},
{
"operation": "Highlighting",
"time": "236.262",
"value": "Henry Ford Kamel_sentence_10"
},
{
"operation": "Highlighting",
"time": "239.254",
"value": "Henry Ford Kamel_sentence_11"
},
{
"operation": "Highlighting",
"time": "241.706",
"value": "Henry Ford Kamel_sentence_7"
},
{
"operation": "hyperlink",
"time": "244.599",
"value": "2008 Ghanaian general election"
},
{
"operation": "Now on",
"time": "245.699",
"value": "2008 Ghanaian general election"
},
{
"operation": "back-button-clicked",
"time": "254.274",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "255.026",
"value": "Henry Ford Kamel"
},
{
"operation": "finish",
"time": "258.037",
"value": "finish"
}
]
}
|
exfever_3704
|
La La Anthony was the youngest host of the VH1 reality television show, Flavor of Love.
|
{
"evidence": [
"La La Anthony",
"For the Love of Ray J",
"Flavor of Love"
],
"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:\nLa La Anthony was the youngest host of the VH1 reality television show, Flavor of Love.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"La La Anthony",
"For the Love of Ray J",
"Flavor of Love"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 6296,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
41873
|
The Lumbini Zone, one of Nepal's fourteen provinces, includes districts such as Kapilvatsu and Rupandehi.
|
{
"evidence": [
"Lumbini Zone_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 Lumbini Zone, one of Nepal's fourteen provinces, includes districts such as Kapilvatsu and Rupandehi.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Lumbini Zone_sentence_0"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 6297,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.696",
"value": "?search="
},
{
"operation": "search",
"time": "13.468",
"value": "Lumbini Zone"
},
{
"operation": "Now on",
"time": "14.344",
"value": "Lumbini Zone"
},
{
"operation": "Highlighting",
"time": "29.327",
"value": "Lumbini Zone_sentence_0"
},
{
"operation": "finish",
"time": "64.254",
"value": "finish"
}
]
}
|
94427
|
Petra Vlhová was the champion for the alpine skiing Girls' Slalom event in the Norway 2012 Innsbruck.
|
{
"evidence": [
"Petra Vlhová_sentence_0",
"Petra Vlhová_sentence_2",
"Petra Vlhová_sentence_4"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nPetra Vlhová was the champion for the alpine skiing Girls' Slalom event in the Norway 2012 Innsbruck.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Petra Vlhová_sentence_0",
"Petra Vlhová_sentence_2",
"Petra Vlhová_sentence_4"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 6298,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.987",
"value": "?search="
},
{
"operation": "search",
"time": "9.995",
"value": "Petra Vlhová"
},
{
"operation": "Now on",
"time": "11.18",
"value": "Petra Vlhová"
},
{
"operation": "Highlighting",
"time": "14.312",
"value": "Petra Vlhová_sentence_0"
},
{
"operation": "Highlighting",
"time": "69.018",
"value": "Petra Vlhová_sentence_2"
},
{
"operation": "Highlighting",
"time": "73.981",
"value": "Petra Vlhová_sentence_4"
},
{
"operation": "finish",
"time": "81.608",
"value": "finish"
}
]
}
|
exfever_330
|
Role Models is an American film that follows two salesmen who sell a type of drink containing stimulant compounds.
|
{
"evidence": [
"Role Models",
"Energy drink"
],
"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:\nRole Models is an American film that follows two salesmen who sell a type of drink containing stimulant compounds.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Role Models",
"Energy drink"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6299,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
2173
|
The Pittsburgh Steelers won eight of their sixteen games in their 1987 season.
|
{
"evidence": [
"1987 Pittsburgh Steelers season_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 Pittsburgh Steelers won eight of their sixteen games in their 1987 season.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"1987 Pittsburgh Steelers season_sentence_1"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 6300,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "search",
"time": "12.76",
"value": "Pittsburgh Steelers"
},
{
"operation": "Page search",
"time": "25.482",
"value": "1987"
},
{
"operation": "search",
"time": "83.341",
"value": "Pittsburgh Steelers 1987 season"
},
{
"operation": "search",
"time": "111.844",
"value": "1987 Pittsburgh Steelers season"
},
{
"operation": "Highlighting",
"time": "208.864",
"value": "1987 Pittsburgh Steelers season_sentence_1"
},
{
"operation": "finish",
"time": "235.73",
"value": "finish"
}
]
}
|
exfever_3896
|
Gia was the first American biographical drama television film directed by Michael Cristofer.
|
{
"evidence": [
"Gia",
"Supermodel"
],
"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:\nGia was the first American biographical drama television film directed by Michael Cristofer.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Gia",
"Supermodel"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 6301,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
91359
|
I Knew It Was You is an American documentary film released on 16 January 2009,directed by Richard Shepard and produced by Brett Ratner, Stacey Reiss and Richard Shepard and distributed by HBO and Oscilloscope.
|
{
"evidence": [
"I Knew It Was You_sentence_4",
"2009 Sundance Film Festival_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:\nI Knew It Was You is an American documentary film released on 16 January 2009,directed by Richard Shepard and produced by Brett Ratner, Stacey Reiss and Richard Shepard and distributed by HBO and Oscilloscope.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"I Knew It Was You_sentence_4",
"2009 Sundance Film Festival_sentence_0"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 6302,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.913",
"value": "?search="
},
{
"operation": "search",
"time": "23.015",
"value": "I Knew It Was You"
},
{
"operation": "Now on",
"time": "23.825",
"value": "I Knew It Was You"
},
{
"operation": "Highlighting",
"time": "45.292",
"value": "I Knew It Was You_sentence_0"
},
{
"operation": "Page search",
"time": "89.209",
"value": "january"
},
{
"operation": "Highlighting",
"time": "174.419",
"value": "I Knew It Was You_sentence_4"
},
{
"operation": "Highlighting deleted",
"time": "177.337",
"value": "I Knew It Was You_sentence_0"
},
{
"operation": "hyperlink",
"time": "202.984",
"value": "2009 Sundance Film Festival"
},
{
"operation": "Now on",
"time": "203.856",
"value": "2009 Sundance Film Festival"
},
{
"operation": "Highlighting",
"time": "251.826",
"value": "2009 Sundance Film Festival_sentence_0"
},
{
"operation": "finish",
"time": "275.719",
"value": "finish"
}
]
}
|
84556
|
Hillmorton is a suburb in the Borough of Paddox , Warwickshire, England.
|
{
"evidence": [
"Hillmorton_sentence_1",
"Hillmorton_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:\nHillmorton is a suburb in the Borough of Paddox , Warwickshire, England.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"Hillmorton_sentence_1",
"Hillmorton_sentence_2"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 6303,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.925",
"value": "?search="
},
{
"operation": "search",
"time": "15.111",
"value": "Hillmorton"
},
{
"operation": "Now on",
"time": "16.352",
"value": "Hillmorton"
},
{
"operation": "Highlighting",
"time": "46.639",
"value": "Hillmorton_sentence_1"
},
{
"operation": "Highlighting",
"time": "55.496",
"value": "Hillmorton_sentence_2"
},
{
"operation": "finish",
"time": "65.349",
"value": "finish"
}
]
}
|
exfever_1092
|
Jesus died by crucifixion as a sacrifice to achieve atonement for sin.
|
{
"evidence": [
"Jesus",
"God the Son"
],
"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:\nJesus died by crucifixion as a sacrifice to achieve atonement for sin.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Jesus",
"God the Son"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 6304,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_167
|
Daniel Carey is the drummer for an American rock band from Los Angeles.
|
{
"evidence": [
"Danny Carey",
"Tool (band)"
],
"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:\nDaniel Carey is the drummer for an American rock band from Los Angeles.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Danny Carey",
"Tool (band)"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6305,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_6748
|
Jamaicans for Justice was founded in 1999 in the capital and largest city of an island country situated in the Caribbean Sea.
|
{
"evidence": [
"Jamaicans for Justice",
"Kingston, Jamaica",
"Jamaica"
],
"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:\nJamaicans for Justice was founded in 1999 in the capital and largest city of an island country situated in the Caribbean Sea.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Jamaicans for Justice",
"Kingston, Jamaica",
"Jamaica"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6306,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
51611
|
Frenchy Uhalt played for the Oakland Oaks, he spent a season and a half with the American Association Milwaukee Brewers but never served as player-manager.
|
{
"evidence": [
"Frenchy Uhalt_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:\nFrenchy Uhalt played for the Oakland Oaks, he spent a season and a half with the American Association Milwaukee Brewers but never served as player-manager.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Frenchy Uhalt_sentence_5"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 6307,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.429",
"value": "?search="
},
{
"operation": "search",
"time": "19.353",
"value": "Frenchy Uhalt"
},
{
"operation": "Now on",
"time": "20.688",
"value": "Frenchy Uhalt"
},
{
"operation": "Highlighting",
"time": "213.428",
"value": "Frenchy Uhalt_sentence_3"
},
{
"operation": "Highlighting",
"time": "216.83",
"value": "Frenchy Uhalt_sentence_4"
},
{
"operation": "Highlighting",
"time": "222.221",
"value": "Frenchy Uhalt_sentence_5"
},
{
"operation": "Highlighting deleted",
"time": "223.7",
"value": "Frenchy Uhalt_sentence_4"
},
{
"operation": "Highlighting deleted",
"time": "224.382",
"value": "Frenchy Uhalt_sentence_3"
},
{
"operation": "finish",
"time": "237.768",
"value": "finish"
}
]
}
|
43409
|
Mark Henig was a British politician and businessman who served as the chairman of the Association of Municipal Corporations from 1966 to 1967 and Lord Mayor of Leicester from 1967 to 1968 .
|
{
"evidence": [
"Mark Henig_sentence_0",
"Mark Henig_sentence_8",
"Mark Henig_sentence_9"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nMark Henig was a British politician and businessman who served as the chairman of the Association of Municipal Corporations from 1966 to 1967 and Lord Mayor of Leicester from 1967 to 1968 .\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Mark Henig_sentence_0",
"Mark Henig_sentence_8",
"Mark Henig_sentence_9"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6308,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.976",
"value": "?search="
},
{
"operation": "search",
"time": "10.46",
"value": "Mark Henig"
},
{
"operation": "Now on",
"time": "11.273",
"value": "Mark Henig"
},
{
"operation": "Highlighting",
"time": "12.699",
"value": "Mark Henig_sentence_0"
},
{
"operation": "Highlighting",
"time": "23.15",
"value": "Mark Henig_sentence_8"
},
{
"operation": "Highlighting",
"time": "28.99",
"value": "Mark Henig_sentence_9"
},
{
"operation": "finish",
"time": "57.954",
"value": "finish"
}
]
}
|
exfever_5903
|
Life Is Peachy is the second studio album by American nu metal band Korn and was produced by an American record producer who discovered acts such as Korn, Glassjaw, The Blood Brothers, Slipknot, and Limp Bizkit.
|
{
"evidence": [
"Life Is Peachy",
"Ross Robinson"
],
"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:\nLife Is Peachy is the second studio album by American nu metal band Korn and was produced by an American record producer who discovered acts such as Korn, Glassjaw, The Blood Brothers, Slipknot, and Limp Bizkit.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Life Is Peachy",
"Ross Robinson"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6309,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7568
|
Little Mix is a music act featuring several male singers who generally harmonize together.
|
{
"evidence": [
"Little Mix",
"Girl 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:\nLittle Mix is a music act featuring several male singers who generally harmonize together.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Little Mix",
"Girl group"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 6310,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
44847
|
Open Government Partnership consecutively held global summits from 2012 to 2019 wherein in 2015 the summit was held in Mexico City, Mexico while the incoming summit for 2021 will be held in Seoul, South Korea.
|
{
"evidence": [
"Open Government Partnership_sentence_72"
],
"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:\nOpen Government Partnership consecutively held global summits from 2012 to 2019 wherein in 2015 the summit was held in Mexico City, Mexico while the incoming summit for 2021 will be held in Seoul, South Korea.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Open Government Partnership_sentence_72"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 6311,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.701",
"value": "?search="
},
{
"operation": "search",
"time": "9.261",
"value": "Open Government Partnership"
},
{
"operation": "Now on",
"time": "10.329",
"value": "Open Government Partnership"
},
{
"operation": "Highlighting",
"time": "27.52",
"value": "Open Government Partnership_sentence_13"
},
{
"operation": "Highlighting",
"time": "27.521",
"value": "Open Government Partnership_section_0"
},
{
"operation": "Highlighting",
"time": "52.102",
"value": "Open Government Partnership_cell_2_1_0"
},
{
"operation": "Highlighting",
"time": "52.104",
"value": "Open Government Partnership_header_cell_2_0_0"
},
{
"operation": "Highlighting",
"time": "55.803",
"value": "Open Government Partnership_cell_2_1_1"
},
{
"operation": "Highlighting",
"time": "55.805",
"value": "Open Government Partnership_header_cell_2_0_1"
},
{
"operation": "Highlighting",
"time": "62.519",
"value": "Open Government Partnership_cell_2_2_0"
},
{
"operation": "Highlighting",
"time": "63.802",
"value": "Open Government Partnership_cell_2_2_1"
},
{
"operation": "Highlighting",
"time": "65.028",
"value": "Open Government Partnership_cell_2_3_0"
},
{
"operation": "Highlighting",
"time": "66.513",
"value": "Open Government Partnership_cell_2_3_1"
},
{
"operation": "Highlighting",
"time": "67.897",
"value": "Open Government Partnership_cell_2_3_2"
},
{
"operation": "Highlighting",
"time": "67.899",
"value": "Open Government Partnership_header_cell_2_0_2"
},
{
"operation": "Highlighting",
"time": "69.798",
"value": "Open Government Partnership_cell_2_4_0"
},
{
"operation": "Highlighting",
"time": "72.397",
"value": "Open Government Partnership_cell_2_4_1"
},
{
"operation": "Highlighting",
"time": "75.457",
"value": "Open Government Partnership_cell_2_5_0"
},
{
"operation": "Highlighting deleted",
"time": "80.891",
"value": "Open Government Partnership_cell_2_5_0"
},
{
"operation": "Highlighting",
"time": "81.543",
"value": "Open Government Partnership_cell_2_5_0"
},
{
"operation": "Highlighting deleted",
"time": "82.488",
"value": "Open Government Partnership_cell_2_4_0"
},
{
"operation": "Highlighting",
"time": "83.625",
"value": "Open Government Partnership_cell_2_4_0"
},
{
"operation": "Highlighting deleted",
"time": "84.727",
"value": "Open Government Partnership_cell_2_1_0"
},
{
"operation": "Highlighting deleted",
"time": "85.436",
"value": "Open Government Partnership_cell_2_1_1"
},
{
"operation": "Highlighting deleted",
"time": "86.257",
"value": "Open Government Partnership_cell_2_2_1"
},
{
"operation": "Highlighting deleted",
"time": "86.92",
"value": "Open Government Partnership_cell_2_3_1"
},
{
"operation": "Highlighting",
"time": "98.416",
"value": "Open Government Partnership_sentence_72"
},
{
"operation": "finish",
"time": "111.993",
"value": "finish"
}
]
}
|
85487
|
Although the Spark Arena hosted many entertainment events in 2014, none of these acts were female musical artists.
|
{
"evidence": [
"Spark Arena_sentence_12",
"Spark Arena_sentence_40"
],
"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:\nAlthough the Spark Arena hosted many entertainment events in 2014, none of these acts were female musical artists.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Spark Arena_sentence_12",
"Spark Arena_sentence_40"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 6312,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.845",
"value": "?search="
},
{
"operation": "search",
"time": "20.438",
"value": "Spark Arena"
},
{
"operation": "Now on",
"time": "21.435",
"value": "Spark Arena"
},
{
"operation": "hyperlink",
"time": "56.821",
"value": "5 Other events Spark Arena_section_4"
},
{
"operation": "search",
"time": "77.457",
"value": "2014 Spark Arena event"
},
{
"operation": "Now on",
"time": "80.19",
"value": "undefined"
},
{
"operation": "search",
"time": "94.082",
"value": "Spark Arena"
},
{
"operation": "Now on",
"time": "95.174",
"value": "Spark Arena"
},
{
"operation": "Highlighting",
"time": "99.901",
"value": "Spark Arena_sentence_12"
},
{
"operation": "Highlighting",
"time": "108.092",
"value": "Spark Arena_sentence_40"
},
{
"operation": "finish",
"time": "122.146",
"value": "finish"
}
]
}
|
exfever_6577
|
Jiang Wen is well known internationally as an actor, having starred as Baze Malbus in the American epic space opera film produced by an American film and television production company founded by filmmaker George Lucas.
|
{
"evidence": [
"Jiang Wen",
"Rogue One",
"Lucasfilm"
],
"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:\nJiang Wen is well known internationally as an actor, having starred as Baze Malbus in the American epic space opera film produced by an American film and television production company founded by filmmaker George Lucas.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Jiang Wen",
"Rogue One",
"Lucasfilm"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6313,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7937
|
The New York Giants are a professional American football team based in Las Vegas with eight NFL championship titles, who?joined in the sixth regular season of the National Football League.
|
{
"evidence": [
"New York Giants",
"1925 NFL season"
],
"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 New York Giants are a professional American football team based in Las Vegas with eight NFL championship titles, who?joined in the sixth regular season of the National Football League.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"New York Giants",
"1925 NFL season"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 6314,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
63777
|
Hannah Levien, who was headhunted by a manager in 2011 to work in Los Angeles, was promptly cast in the film Children of Sorrow, a 2012 horror film directed by Jourdan McClure and part of the After Dark Originals series, to play Ellen, a young woman who enters a cult to find answers into the mysterious disappearance of her sister.
|
{
"evidence": [
"Hannah Levien_sentence_8",
"Children of Sorrow_sentence_0",
"Hannah Levien_sentence_9"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nHannah Levien, who was headhunted by a manager in 2011 to work in Los Angeles, was promptly cast in the film Children of Sorrow, a 2012 horror film directed by Jourdan McClure and part of the After Dark Originals series, to play Ellen, a young woman who enters a cult to find answers into the mysterious disappearance of her sister.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Hannah Levien_sentence_8",
"Children of Sorrow_sentence_0",
"Hannah Levien_sentence_9"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6315,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.751",
"value": "?search="
},
{
"operation": "search",
"time": "5.209",
"value": "Hannah Levien"
},
{
"operation": "Now on",
"time": "6.214",
"value": "Hannah Levien"
},
{
"operation": "Highlighting",
"time": "57.445",
"value": "Hannah Levien_sentence_8"
},
{
"operation": "hyperlink",
"time": "63.46",
"value": "Children of Sorrow"
},
{
"operation": "Now on",
"time": "64.551",
"value": "Children of Sorrow"
},
{
"operation": "Highlighting",
"time": "65.391",
"value": "Children of Sorrow_sentence_0"
},
{
"operation": "back-button-clicked",
"time": "83.625",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "83.73",
"value": "Hannah Levien"
},
{
"operation": "Highlighting",
"time": "84.906",
"value": "Hannah Levien_sentence_9"
},
{
"operation": "finish",
"time": "97.035",
"value": "finish"
}
]
}
|
exfever_6788
|
K2 lies in the mountain range in Kashmir that extends into a landlocked country.
|
{
"evidence": [
"K2",
"Karakoram",
"Afghanistan"
],
"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:\nK2 lies in the mountain range in Kashmir that extends into a landlocked country.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"K2",
"Karakoram",
"Afghanistan"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6316,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_6171
|
Mel Columcille Gerard Gibson was born in a city that lies on a bay along the east side of the Hudson River, across from a former hamlet located in the town of Stony Point in Rockland County.
|
{
"evidence": [
"Mel Gibson",
"Peekskill, New York",
"Jones Point, New York"
],
"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:\nMel Columcille Gerard Gibson was born in a city that lies on a bay along the east side of the Hudson River, across from a former hamlet located in the town of Stony Point in Rockland County.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Mel Gibson",
"Peekskill, New York",
"Jones Point, New York"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6317,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7670
|
Gone Girl is a 2014 American psychological horror film directed by David Fincher and written by an American writer.
|
{
"evidence": [
"Gone Girl (film)",
"Gillian Flynn"
],
"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:\nGone Girl is a 2014 American psychological horror film directed by David Fincher and written by an American writer.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Gone Girl (film)",
"Gillian Flynn"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 6318,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
6645
|
Sirin Pancaroglu studied harp at the Geneva Conservatory (the oldest music educaton institution in Geneva, Switzerland) and at Indiana University School of Music (a multi-campus system in Indiana) and receieved has Masters in Education in 1992.
|
{
"evidence": [
"Şirin Pancaroğlu_sentence_8"
],
"label": "REFUTE"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nSirin Pancaroglu studied harp at the Geneva Conservatory (the oldest music educaton institution in Geneva, Switzerland) and at Indiana University School of Music (a multi-campus system in Indiana) and receieved has Masters in Education in 1992.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Şirin Pancaroğlu_sentence_8"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 6319,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.706",
"value": "?search="
},
{
"operation": "search",
"time": "23.84",
"value": "Sirin Pancaroglu"
},
{
"operation": "Now on",
"time": "24.958",
"value": "Şirin Pancaroğlu"
},
{
"operation": "Highlighting",
"time": "52.212",
"value": "Şirin Pancaroğlu_sentence_7"
},
{
"operation": "hyperlink",
"time": "54.938",
"value": "Geneva Conservatory"
},
{
"operation": "Now on",
"time": "55.601",
"value": "Geneva Conservatory"
},
{
"operation": "Highlighting",
"time": "60.18",
"value": "Conservatoire de Musique de Genève_sentence_4"
},
{
"operation": "Highlighting",
"time": "60.189",
"value": "Conservatoire de Musique de Genève_title"
},
{
"operation": "Highlighting",
"time": "60.196",
"value": "Conservatoire de Musique de Genève_section_0"
},
{
"operation": "back-button-clicked",
"time": "61.735",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "62.385",
"value": "Şirin Pancaroğlu"
},
{
"operation": "Highlighting",
"time": "68.156",
"value": "Şirin Pancaroğlu_sentence_8"
},
{
"operation": "hyperlink",
"time": "70.933",
"value": "Indiana University"
},
{
"operation": "Now on",
"time": "71.944",
"value": "Indiana University"
},
{
"operation": "Highlighting",
"time": "86.54",
"value": "Indiana University_sentence_3"
},
{
"operation": "Highlighting",
"time": "86.617",
"value": "Indiana University_title"
},
{
"operation": "back-button-clicked",
"time": "89.305",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "89.561",
"value": "Şirin Pancaroğlu"
},
{
"operation": "Highlighting deleted",
"time": "102.753",
"value": "Şirin Pancaroğlu_sentence_7"
},
{
"operation": "finish",
"time": "130.741",
"value": "finish"
}
]
}
|
exfever_1555
|
The Carolina Panthers are a professional American football team that competes in a professional American football league that consists of 32 teams split between the American Football Conference and the National Football Conference.
|
{
"evidence": [
"Carolina Panthers",
"National Football League"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Carolina Panthers are a professional American football team that competes in a professional American football league that consists of 32 teams split between the American Football Conference and the National Football Conference.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Carolina Panthers",
"National Football League"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6320,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_5220
|
James Blake is an English singer, and record producer best known for his third album which was released on 6 May 2017, by Polydor Records.
|
{
"evidence": [
"James Blake (musician)",
"The Colour in Anything"
],
"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:\nJames Blake is an English singer, and record producer best known for his third album which was released on 6 May 2017, by Polydor Records.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"James Blake (musician)",
"The Colour in Anything"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 6321,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
88337
|
Terri Priest was surrounded with multiple controversies about corruption and infidelity but was a supporter of the Worcester Art Museum (houses over 38,000 works of art dating from antiquity to the present day).
|
{
"evidence": [
"Terri Priest_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:\nTerri Priest was surrounded with multiple controversies about corruption and infidelity but was a supporter of the Worcester Art Museum (houses over 38,000 works of art dating from antiquity to the present day).\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Terri Priest_sentence_10"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 6322,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.85",
"value": "?search="
},
{
"operation": "search",
"time": "10.683",
"value": "Terri Priest"
},
{
"operation": "Now on",
"time": "11.299",
"value": "Terri Priest"
},
{
"operation": "Page search",
"time": "68.392",
"value": "corruption"
},
{
"operation": "Page search",
"time": "111.352",
"value": "Worcester Art Museum"
},
{
"operation": "Highlighting",
"time": "120.183",
"value": "Terri Priest_sentence_10"
},
{
"operation": "finish",
"time": "184.464",
"value": "finish"
}
]
}
|
27520
|
Alfonso beat Newfel Ouatah but lost to Russian Evgeny Romanov at the World Junior Championships 2004; then was also defeated nationally by compatriot superstar Odlanier Solis twice in 2004.
|
{
"evidence": [
"Robert Alfonso_sentence_3",
"Robert Alfonso_sentence_4"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nAlfonso beat Newfel Ouatah but lost to Russian Evgeny Romanov at the World Junior Championships 2004; then was also defeated nationally by compatriot superstar Odlanier Solis twice in 2004.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"Robert Alfonso_sentence_3",
"Robert Alfonso_sentence_4"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6323,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.457",
"value": "?search="
},
{
"operation": "search",
"time": "16.509",
"value": "World Junior Championships 2004"
},
{
"operation": "Now on",
"time": "17.7",
"value": "2004 World Junior Championships"
},
{
"operation": "search",
"time": "85.481",
"value": "Newfel Ouatah"
},
{
"operation": "Now on",
"time": "86.319",
"value": "Newfel Ouatah"
},
{
"operation": "back-button-clicked",
"time": "104.195",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "104.988",
"value": "2004 World Junior Championships"
},
{
"operation": "search",
"time": "113.252",
"value": "Newfel Ouatah"
},
{
"operation": "Now on",
"time": "114.21",
"value": "Newfel Ouatah"
},
{
"operation": "search",
"time": "148.629",
"value": "robert alfonso"
},
{
"operation": "Now on",
"time": "149.875",
"value": "Robert Alfonso"
},
{
"operation": "Highlighting",
"time": "156.446",
"value": "Robert Alfonso_sentence_3"
},
{
"operation": "Highlighting",
"time": "162.321",
"value": "Robert Alfonso_sentence_4"
},
{
"operation": "finish",
"time": "168.941",
"value": "finish"
}
]
}
|
exfever_2085
|
Cosmetics have various purposes such as personal hygiene, personal grooming, and beautification.
|
{
"evidence": [
"Cosmetics",
"Personal care"
],
"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:\nCosmetics have various purposes such as personal hygiene, personal grooming, and beautification.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Cosmetics",
"Personal care"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6324,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_2808
|
Abby Sciuto was introduced in two episodes of the television show created by Donald P. Bellisario.
|
{
"evidence": [
"Abby Sciuto",
"JAG (TV series)"
],
"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:\nAbby Sciuto was introduced in two episodes of the television show created by Donald P. Bellisario.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Abby Sciuto",
"JAG (TV series)"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6325,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
20558
|
Myriam de la Riva's work has participated in over 500 collective exhibitions in both Mexico and abroad in venues such as the Palacio de Bellas Artes, a prominent cultural center in Mexico City.
|
{
"evidence": [
"Myriam de la Riva_sentence_10",
"Myriam de la Riva_sentence_11",
"Palacio de Bellas Artes_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:\nMyriam de la Riva's work has participated in over 500 collective exhibitions in both Mexico and abroad in venues such as the Palacio de Bellas Artes, a prominent cultural center in Mexico City.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Myriam de la Riva_sentence_10",
"Myriam de la Riva_sentence_11",
"Palacio de Bellas Artes_sentence_0"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6326,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.63",
"value": "?search="
},
{
"operation": "search",
"time": "15.646",
"value": "Myriam de la Riva"
},
{
"operation": "Now on",
"time": "16.758",
"value": "Myriam de la Riva"
},
{
"operation": "Highlighting",
"time": "30.026",
"value": "Myriam de la Riva_sentence_3"
},
{
"operation": "Highlighting",
"time": "72.299",
"value": "Myriam de la Riva_sentence_10"
},
{
"operation": "Highlighting",
"time": "81.499",
"value": "Myriam de la Riva_sentence_11"
},
{
"operation": "hyperlink",
"time": "87.307",
"value": "Palacio de Bellas Artes"
},
{
"operation": "Now on",
"time": "89.673",
"value": "Palacio de Bellas Artes"
},
{
"operation": "Highlighting",
"time": "91.443",
"value": "Palacio de Bellas Artes_sentence_0"
},
{
"operation": "back-button-clicked",
"time": "106.988",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "107.08",
"value": "Myriam de la Riva"
},
{
"operation": "Now on",
"time": "112.791",
"value": "Myriam de la Riva"
},
{
"operation": "finish",
"time": "135.499",
"value": "finish"
}
]
}
|
exfever_129
|
Sarah Silverman is an American comedian who has won two awards bestowed by the Academy of Television Arts & Sciences to recognize excellence in American primetime television programming.
|
{
"evidence": [
"Sarah Silverman",
"Primetime Emmy Award"
],
"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:\nSarah Silverman is an American comedian who has won two awards bestowed by the Academy of Television Arts & Sciences to recognize excellence in American primetime television programming.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Sarah Silverman",
"Primetime Emmy Award"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6327,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_1922
|
"Hot Right Now" was the most successful single by DJ Fresh.
|
{
"evidence": [
"Hot Right Now",
"Nextlevelism"
],
"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:\n\"Hot Right Now\" was the most successful single by DJ Fresh.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Hot Right Now",
"Nextlevelism"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 6328,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_3420
|
Howard Allan Stern is an American radio and television personality known for his American radio show, whose other prominent staff members include an executive producer born on March 14, 1961.
|
{
"evidence": [
"Howard Stern",
"The Howard Stern Show",
"Gary Dell'Abate"
],
"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:\nHoward Allan Stern is an American radio and television personality known for his American radio show, whose other prominent staff members include an executive producer born on March 14, 1961.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Howard Stern",
"The Howard Stern Show",
"Gary Dell'Abate"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6329,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
70373
|
Mesabi Trail, a 132-mile paved bicycle trail, goes through many small towns like Marble and Keewatin.
|
{
"evidence": [
"Mesabi Trail_sentence_0",
"Mesabi Trail_sentence_2"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nMesabi Trail, a 132-mile paved bicycle trail, goes through many small towns like Marble and Keewatin.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Mesabi Trail_sentence_0",
"Mesabi Trail_sentence_2"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6330,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.875",
"value": "?search="
},
{
"operation": "search",
"time": "8.283",
"value": "Mesabi Trail"
},
{
"operation": "Now on",
"time": "9.289",
"value": "Mesabi Trail"
},
{
"operation": "Highlighting",
"time": "14.003",
"value": "Mesabi Trail_sentence_0"
},
{
"operation": "Highlighting",
"time": "17.564",
"value": "Mesabi Trail_sentence_2"
},
{
"operation": "finish",
"time": "23.075",
"value": "finish"
}
]
}
|
exfever_8138
|
War of the Worlds is a 2015 film directed by Steven Spielberg and written by Josh Friedman and ninth most successful screenwriter of all time in terms of U.S. box office receipts with a total gross of over $2.3 billion.
|
{
"evidence": [
"War of the Worlds (2005 film)",
"David Koepp"
],
"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:\nWar of the Worlds is a 2015 film directed by Steven Spielberg and written by Josh Friedman and ninth most successful screenwriter of all time in terms of U.S. box office receipts with a total gross of over $2.3 billion.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"War of the Worlds (2005 film)",
"David Koepp"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 6331,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
78264
|
Diuris curta is a species of edible orchid that flowers from August to early October and is endemic to eastern Australia, its labellum at 10–12 mm (0.39–0.47 in) long has three lobes, with the centre lobe being egg-shaped to wedge-shaped, and both side lobes being oblong to egg-shaped.
|
{
"evidence": [
"Diuris curta_sentence_0",
"Diuris curta_sentence_11",
"Diuris curta_sentence_8",
"Diuris curta_sentence_9"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nDiuris curta is a species of edible orchid that flowers from August to early October and is endemic to eastern Australia, its labellum at 10–12 mm (0.39–0.47 in) long has three lobes, with the centre lobe being egg-shaped to wedge-shaped, and both side lobes being oblong to egg-shaped.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Diuris curta_sentence_0",
"Diuris curta_sentence_11",
"Diuris curta_sentence_8",
"Diuris curta_sentence_9"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 6332,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.819",
"value": "?search="
},
{
"operation": "search",
"time": "18.104",
"value": "Diuris curta"
},
{
"operation": "Now on",
"time": "18.801",
"value": "Diuris curta"
},
{
"operation": "Highlighting",
"time": "65.418",
"value": "Diuris curta_sentence_0"
},
{
"operation": "Highlighting",
"time": "76.345",
"value": "Diuris curta_sentence_11"
},
{
"operation": "Highlighting",
"time": "89.791",
"value": "Diuris curta_sentence_8"
},
{
"operation": "Highlighting",
"time": "100.381",
"value": "Diuris curta_sentence_9"
},
{
"operation": "finish",
"time": "150.056",
"value": "finish"
}
]
}
|
94124
|
Turners Falls Canal was a canal which is a waterway cannel or artificial waterway for water conveyance or to service water transport vehicles.
|
{
"evidence": [
"Turners Falls Canal_sentence_0",
"Canal_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:\nTurners Falls Canal was a canal which is a waterway cannel or artificial waterway for water conveyance or to service water transport vehicles.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Turners Falls Canal_sentence_0",
"Canal_sentence_1"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6333,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.124",
"value": "?search="
},
{
"operation": "search",
"time": "10.428",
"value": "Turners Falls Canal"
},
{
"operation": "Now on",
"time": "11.133",
"value": "Turners Falls Canal"
},
{
"operation": "Highlighting",
"time": "72.98",
"value": "Turners Falls Canal_sentence_0"
},
{
"operation": "hyperlink",
"time": "74.895",
"value": "canal"
},
{
"operation": "Now on",
"time": "76.211",
"value": "Canal"
},
{
"operation": "Highlighting",
"time": "83.753",
"value": "Canal_sentence_1"
},
{
"operation": "finish",
"time": "89.233",
"value": "finish"
}
]
}
|
83751
|
Histone acetylation/deacetylation alters chromosome structure and affects transcription factor access to RNA, while the protein encoded by this gene belongs to class II of the histone deacetylase/acuc/apha family.
|
{
"evidence": [
"HDAC8_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:\nHistone acetylation/deacetylation alters chromosome structure and affects transcription factor access to RNA, while the protein encoded by this gene belongs to class II of the histone deacetylase/acuc/apha family.\n",
"role": "user"
}
] |
Search terms not in claim
|
{
"ground_truth": {
"evidence": [
"HDAC8_sentence_2"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 6334,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.508",
"value": "?search="
},
{
"operation": "search",
"time": "14.518",
"value": "Histone acetylation"
},
{
"operation": "Now on",
"time": "17.151",
"value": "Histone acetyltransferase"
},
{
"operation": "Page search",
"time": "31.343",
"value": "deacetylation"
},
{
"operation": "Highlighting",
"time": "107.411",
"value": "Histone acetyltransferase_sentence_1"
},
{
"operation": "Highlighting",
"time": "107.412",
"value": "Histone acetyltransferase_title"
},
{
"operation": "search",
"time": "127.076",
"value": "Contains...Histone acetylation/deacetylation"
},
{
"operation": "Now on",
"time": "128.479",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "135.376",
"value": "HDAC8"
},
{
"operation": "Now on",
"time": "135.901",
"value": "HDAC8"
},
{
"operation": "Highlighting",
"time": "140.001",
"value": "HDAC8_sentence_2"
},
{
"operation": "finish",
"time": "151.98",
"value": "finish"
}
]
}
|
exfever_3809
|
Colin James Farrell is an Irish actor who starred in a 2002 American psychological thriller film directed by Joel Schumacher.
|
{
"evidence": [
"Colin Farrell",
"Phone Booth (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:\nColin James Farrell is an Irish actor who starred in a 2002 American psychological thriller film directed by Joel Schumacher.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Colin Farrell",
"Phone Booth (film)"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 6335,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
53322
|
Brahmi numerals are the ancestors of the Hindu-Arabic glyphs, and although mathematicians did not agree on the form of the numerals, most agreed to train themselves with the forms now know as Eaatern Arabic Numerals.
|
{
"evidence": [
"Hindu–Arabic numeral system_sentence_60"
],
"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:\nBrahmi numerals are the ancestors of the Hindu-Arabic glyphs, and although mathematicians did not agree on the form of the numerals, most agreed to train themselves with the forms now know as Eaatern Arabic Numerals.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Hindu–Arabic numeral system_sentence_60"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 6336,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.621",
"value": "?search="
},
{
"operation": "search",
"time": "44.196",
"value": "Brahmi numerals"
},
{
"operation": "Now on",
"time": "45.97",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "62.57",
"value": "Bhāskara I"
},
{
"operation": "Now on",
"time": "63.283",
"value": "Bhāskara I"
},
{
"operation": "Page search",
"time": "77.105",
"value": "brahmi"
},
{
"operation": "back-button-clicked",
"time": "101.921",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "102.552",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "117.68",
"value": "Brahmi script"
},
{
"operation": "Now on",
"time": "134.394",
"value": "Brahmi script"
},
{
"operation": "Highlighting",
"time": "137.985",
"value": "Brahmi script_sentence_1"
},
{
"operation": "Highlighting",
"time": "137.986",
"value": "Brahmi script_title"
},
{
"operation": "Highlighting deleted",
"time": "139.928",
"value": "Brahmi script_sentence_1"
},
{
"operation": "Highlighting deleted",
"time": "139.928",
"value": "Brahmi script_title"
},
{
"operation": "Page search",
"time": "143.611",
"value": "numerals"
},
{
"operation": "hyperlink",
"time": "170.079",
"value": "Hindu-Arabic numeral system"
},
{
"operation": "Now on",
"time": "172.435",
"value": "Hindu–Arabic numeral system"
},
{
"operation": "Page search",
"time": "178.175",
"value": "brahmi"
},
{
"operation": "Highlighting",
"time": "182.47",
"value": "Hindu–Arabic numeral system_sentence_7"
},
{
"operation": "Highlighting",
"time": "225.445",
"value": "Hindu–Arabic numeral system_sentence_19"
},
{
"operation": "Highlighting",
"time": "225.446",
"value": "Hindu–Arabic numeral system_section_2"
},
{
"operation": "Highlighting deleted",
"time": "231.964",
"value": "Hindu–Arabic numeral system_sentence_7"
},
{
"operation": "Highlighting",
"time": "236.62",
"value": "Hindu–Arabic numeral system_sentence_7"
},
{
"operation": "Highlighting",
"time": "277.795",
"value": "Hindu–Arabic numeral system_sentence_27"
},
{
"operation": "Highlighting",
"time": "277.796",
"value": "Hindu–Arabic numeral system_section_5"
},
{
"operation": "Highlighting",
"time": "285.035",
"value": "Hindu–Arabic numeral system_sentence_27"
},
{
"operation": "Highlighting",
"time": "285.035",
"value": "Hindu–Arabic numeral system_section_5"
},
{
"operation": "Page search",
"time": "319.562",
"value": "form of the numerals"
},
{
"operation": "Highlighting",
"time": "384.343",
"value": "Hindu–Arabic numeral system_sentence_60"
},
{
"operation": "Highlighting deleted",
"time": "385.295",
"value": "Hindu–Arabic numeral system_sentence_60"
},
{
"operation": "Highlighting",
"time": "401.478",
"value": "Hindu–Arabic numeral system_sentence_60"
},
{
"operation": "hyperlink",
"time": "424.461",
"value": "Eastern Arabic numerals"
},
{
"operation": "Now on",
"time": "425.105",
"value": "Eastern Arabic numerals"
},
{
"operation": "search",
"time": "513.268",
"value": "most agreed to train themselves with the forms now known as"
},
{
"operation": "Now on",
"time": "520.056",
"value": "undefined"
},
{
"operation": "finish",
"time": "542.314",
"value": "finish"
}
]
}
|
3875
|
Los Gaiteros de San Jacinto members were born in San Jacinto, Colombia which included Miguel Antonio Hernández Vásquez, Juan and José Lara, Pedro Nolasco Mejía and Manuel de Jesus "Mañe" Serpa and was managed by writer and researcher Manuel Zapata Olivella in the 1950s.
|
{
"evidence": [
"Los Gaiteros de San Jacinto_sentence_5",
"Los Gaiteros de San Jacinto_sentence_4",
"Los Gaiteros de San Jacinto_sentence_6"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nLos Gaiteros de San Jacinto members were born in San Jacinto, Colombia which included Miguel Antonio Hernández Vásquez, Juan and José Lara, Pedro Nolasco Mejía and Manuel de Jesus \"Mañe\" Serpa and was managed by writer and researcher Manuel Zapata Olivella in the 1950s.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Los Gaiteros de San Jacinto_sentence_5",
"Los Gaiteros de San Jacinto_sentence_4",
"Los Gaiteros de San Jacinto_sentence_6"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6337,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "search",
"time": "0.014",
"value": "Los Gaiteros de San Jacinto"
},
{
"operation": "Highlighting",
"time": "0.023",
"value": "Los Gaiteros de San Jacinto_sentence_4"
},
{
"operation": "Highlighting",
"time": "0.04",
"value": "Los Gaiteros de San Jacinto_header_cell_0_8_0"
},
{
"operation": "Highlighting",
"time": "0.041",
"value": "Los Gaiteros de San Jacinto_cell_0_8_1"
},
{
"operation": "Highlighting",
"time": "0.041",
"value": "Los Gaiteros de San Jacinto_header_cell_0_8_0"
},
{
"operation": "Highlighting",
"time": "0.041",
"value": "Los Gaiteros de San Jacinto_header_cell_0_1_0"
},
{
"operation": "Highlighting",
"time": "0.041",
"value": "Los Gaiteros de San Jacinto_header_cell_0_0_0"
},
{
"operation": "Highlighting deleted",
"time": "0.048",
"value": "Los Gaiteros de San Jacinto_cell_0_8_1"
},
{
"operation": "Highlighting deleted",
"time": "0.048",
"value": "Los Gaiteros de San Jacinto_header_cell_0_8_0"
},
{
"operation": "Highlighting deleted",
"time": "0.048",
"value": "Los Gaiteros de San Jacinto_header_cell_0_1_0"
},
{
"operation": "Highlighting deleted",
"time": "0.048",
"value": "Los Gaiteros de San Jacinto_header_cell_0_0_0"
},
{
"operation": "Highlighting",
"time": "0.06",
"value": "Los Gaiteros de San Jacinto_header_cell_0_8_0"
},
{
"operation": "Highlighting",
"time": "0.062",
"value": "Los Gaiteros de San Jacinto_cell_0_8_1"
},
{
"operation": "Highlighting",
"time": "0.063",
"value": "Los Gaiteros de San Jacinto_header_cell_0_8_0"
},
{
"operation": "Highlighting",
"time": "0.063",
"value": "Los Gaiteros de San Jacinto_header_cell_0_1_0"
},
{
"operation": "Highlighting",
"time": "0.063",
"value": "Los Gaiteros de San Jacinto_header_cell_0_0_0"
},
{
"operation": "Highlighting deleted",
"time": "0.104",
"value": "Los Gaiteros de San Jacinto_sentence_4"
},
{
"operation": "Highlighting",
"time": "0.106",
"value": "Los Gaiteros de San Jacinto_sentence_5"
},
{
"operation": "Highlighting deleted",
"time": "0.112",
"value": "Los Gaiteros de San Jacinto_cell_0_8_1"
},
{
"operation": "Highlighting deleted",
"time": "0.112",
"value": "Los Gaiteros de San Jacinto_header_cell_0_8_0"
},
{
"operation": "Highlighting deleted",
"time": "0.112",
"value": "Los Gaiteros de San Jacinto_header_cell_0_1_0"
},
{
"operation": "Highlighting deleted",
"time": "0.112",
"value": "Los Gaiteros de San Jacinto_header_cell_0_0_0"
},
{
"operation": "Highlighting",
"time": "0.13",
"value": "Los Gaiteros de San Jacinto_sentence_4"
},
{
"operation": "Highlighting",
"time": "0.143",
"value": "Los Gaiteros de San Jacinto_sentence_6"
},
{
"operation": "finish",
"time": "0.166",
"value": "finish"
},
{
"operation": "finish",
"time": "1618779863.591",
"value": "finish"
}
]
}
|
45047
|
Kéba Mbaye was the chairman of the International Olympic Committee from 1973 to 2002 and also served as president of the Court of Arbitration for Sport, the highest Olympic tribunal.
|
{
"evidence": [
"Kéba Mbaye_sentence_2",
"Kéba Mbaye_sentence_4"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nKéba Mbaye was the chairman of the International Olympic Committee from 1973 to 2002 and also served as president of the Court of Arbitration for Sport, the highest Olympic tribunal.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Kéba Mbaye_sentence_2",
"Kéba Mbaye_sentence_4"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 6338,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.129",
"value": "?search="
},
{
"operation": "search",
"time": "29.471",
"value": "Kéba Mbaye"
},
{
"operation": "Now on",
"time": "30.309",
"value": "Kéba Mbaye"
},
{
"operation": "Page search",
"time": "43.558",
"value": "chairman"
},
{
"operation": "Highlighting",
"time": "51.303",
"value": "Kéba Mbaye_sentence_0"
},
{
"operation": "Page search",
"time": "83.593",
"value": "1973"
},
{
"operation": "Highlighting",
"time": "88.257",
"value": "Kéba Mbaye_sentence_2"
},
{
"operation": "Highlighting deleted",
"time": "96.373",
"value": "Kéba Mbaye_sentence_0"
},
{
"operation": "Highlighting",
"time": "111.447",
"value": "Kéba Mbaye_sentence_4"
},
{
"operation": "finish",
"time": "268.075",
"value": "finish"
}
]
}
|
7748
|
Maria Theresa of Naples and Sicily loved entertainment, she enjoyed the Waltz and adapted well to her new home in Vienna the capital of Austria.
|
{
"evidence": [
"Maria Theresa of Naples and Sicily_sentence_13",
"Maria Theresa of Naples and Sicily_sentence_15",
"Maria Theresa of Naples and Sicily_sentence_14",
"Vienna_sentence_4"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nMaria Theresa of Naples and Sicily loved entertainment, she enjoyed the Waltz and adapted well to her new home in Vienna the capital of Austria.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Maria Theresa of Naples and Sicily_sentence_13",
"Maria Theresa of Naples and Sicily_sentence_15",
"Maria Theresa of Naples and Sicily_sentence_14",
"Vienna_sentence_4"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6339,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.923",
"value": "?search="
},
{
"operation": "search",
"time": "15.629",
"value": "Maria Theresa of Naples and Sicily"
},
{
"operation": "Now on",
"time": "16.698",
"value": "Maria Theresa of Naples and Sicily"
},
{
"operation": "Highlighting",
"time": "44.218",
"value": "Maria Theresa of Naples and Sicily_sentence_13"
},
{
"operation": "Highlighting",
"time": "51.25",
"value": "Maria Theresa of Naples and Sicily_sentence_15"
},
{
"operation": "Highlighting",
"time": "54.631",
"value": "Maria Theresa of Naples and Sicily_sentence_14"
},
{
"operation": "search",
"time": "104.294",
"value": "Vienna"
},
{
"operation": "Now on",
"time": "105.632",
"value": "Vienna"
},
{
"operation": "Highlighting",
"time": "110.67",
"value": "Vienna_sentence_4"
},
{
"operation": "finish",
"time": "124.357",
"value": "finish"
}
]
}
|
exfever_5948
|
The Angkorian Empire is the term that historians use to refer to Cambodia from the 9th to the 15th century uses an Austroasiatic language the official and national language of Cambodia.
|
{
"evidence": [
"Khmer Empire",
"Khmer language"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Angkorian Empire is the term that historians use to refer to Cambodia from the 9th to the 15th century uses an Austroasiatic language the official and national language of Cambodia.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Khmer Empire",
"Khmer language"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6340,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
39113
|
The Tabascans took Villahermosa upriver after bombarding the city on their second attempt; the Tabascans formed guerilla groups which, when matched with the inhospitable drug lords, forced U.S. Military to withdraw.
|
{
"evidence": [
"Tabasco_sentence_131",
"Tabasco_sentence_132",
"Tabasco_sentence_133"
],
"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 Tabascans took Villahermosa upriver after bombarding the city on their second attempt; the Tabascans formed guerilla groups which, when matched with the inhospitable drug lords, forced U.S. Military to withdraw.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Tabasco_sentence_131",
"Tabasco_sentence_132",
"Tabasco_sentence_133"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 6341,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.705",
"value": "?search="
},
{
"operation": "search",
"time": "15.578",
"value": "Tabascans"
},
{
"operation": "Now on",
"time": "16.327",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "24.377",
"value": "Tabasco"
},
{
"operation": "Now on",
"time": "25.246",
"value": "Tabasco"
},
{
"operation": "Page search",
"time": "36.683",
"value": "villhermosa"
},
{
"operation": "Page search",
"time": "60.094",
"value": "upriver"
},
{
"operation": "Highlighting",
"time": "94.342",
"value": "Tabasco_sentence_131"
},
{
"operation": "Highlighting",
"time": "102.275",
"value": "Tabasco_sentence_132"
},
{
"operation": "Highlighting",
"time": "113.347",
"value": "Tabasco_sentence_133"
},
{
"operation": "finish",
"time": "153.284",
"value": "finish"
}
]
}
|
exfever_6230
|
Neighbors was the first comedy film directed by Nicholas Stoller.
|
{
"evidence": [
"Neighbors (2014 film)",
"Christopher Mintz-Plasse"
],
"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:\nNeighbors was the first comedy film directed by Nicholas Stoller.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Neighbors (2014 film)",
"Christopher Mintz-Plasse"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 6342,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
83748
|
Druk Air flies to 19 destinations, including Guwahati and Mumbai in India, the seventh-largest country by land area.
|
{
"evidence": [
"Druk Air_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:\nDruk Air flies to 19 destinations, including Guwahati and Mumbai in India, the seventh-largest country by land area.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Druk Air_sentence_3"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 6343,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.074",
"value": "?search="
},
{
"operation": "search",
"time": "15.808",
"value": "Druk Air"
},
{
"operation": "Now on",
"time": "17.177",
"value": "Druk Air"
},
{
"operation": "Page search",
"time": "23.869",
"value": "destin"
},
{
"operation": "Highlighting",
"time": "26.277",
"value": "Druk Air_sentence_3"
},
{
"operation": "finish",
"time": "82.404",
"value": "finish"
}
]
}
|
exfever_1062
|
Miley Cyrus faced backlash for making “slanted eyes” in a photo with friends.
|
{
"evidence": [
"Miley Cyrus",
"Rock music"
],
"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:\nMiley Cyrus faced backlash for making “slanted eyes” in a photo with friends.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Miley Cyrus",
"Rock music"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 6344,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_5521
|
U2 is an Irish rock band from Dublin that consists of Bono, Adam Clayton, Larry Mullen Jr., and an English-born Irish musician, singer, and songwriter, born on 8 August 1961.
|
{
"evidence": [
"U2",
"The Edge"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nU2 is an Irish rock band from Dublin that consists of Bono, Adam Clayton, Larry Mullen Jr., and an English-born Irish musician, singer, and songwriter, born on 8 August 1961.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"U2",
"The Edge"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6345,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7269
|
New Horizons is a space probe engineered by the Johns Hopkins University Applied Physics Laboratory together with an independent and nonprofit organization founded in 1947.
|
{
"evidence": [
"New Horizons",
"Southwest Research Institute"
],
"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:\nNew Horizons is a space probe engineered by the Johns Hopkins University Applied Physics Laboratory together with an independent and nonprofit organization founded in 1947.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"New Horizons",
"Southwest Research Institute"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6346,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_2394
|
Bohemian Rhapsody is a song by the British rock band Queen that has a loosely defined subgenre within the broad genre of popular music that originated as "rock and roll" in the United States.
|
{
"evidence": [
"Bohemian Rhapsody",
"Hard rock",
"Rock music"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nBohemian Rhapsody is a song by the British rock band Queen that has a loosely defined subgenre within the broad genre of popular music that originated as \"rock and roll\" in the United States.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Bohemian Rhapsody",
"Hard rock",
"Rock music"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6347,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_1173
|
Tate Ellington is an American actor who has appeared in television shows, including an American legal drama television series that focuses on the lead character of the CBS television series.
|
{
"evidence": [
"Tate Ellington",
"The Good Wife",
"Alicia Florrick"
],
"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:\nTate Ellington is an American actor who has appeared in television shows, including an American legal drama television series that focuses on the lead character of the CBS television series.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Tate Ellington",
"The Good Wife",
"Alicia Florrick"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6348,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
21632
|
Albert Einstein published theories of general relativity and special relativity, both of which described a unified spacetime in Einstein Field Equations.
|
{
"evidence": [
"Albert Einstein_sentence_18",
"Einstein field equations_sentence_1"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nAlbert Einstein published theories of general relativity and special relativity, both of which described a unified spacetime in Einstein Field Equations.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Albert Einstein_sentence_18",
"Einstein field equations_sentence_1"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 6349,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.486",
"value": "?search="
},
{
"operation": "search",
"time": "11.463",
"value": "Albert Einstein"
},
{
"operation": "Now on",
"time": "13.368",
"value": "Albert Einstein"
},
{
"operation": "Page search",
"time": "27.531",
"value": "general relativity"
},
{
"operation": "Highlighting",
"time": "41.889",
"value": "Albert Einstein_sentence_18"
},
{
"operation": "Page search",
"time": "61.213",
"value": "unified spacetime"
},
{
"operation": "Page search",
"time": "82.606",
"value": "Einstein Field Equations"
},
{
"operation": "hyperlink",
"time": "96.215",
"value": "Einstein field equations"
},
{
"operation": "Now on",
"time": "96.86",
"value": "Einstein field equations"
},
{
"operation": "Page search",
"time": "107.458",
"value": "special"
},
{
"operation": "Page search",
"time": "123.154",
"value": "relativity"
},
{
"operation": "back-button-clicked",
"time": "173.476",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "173.762",
"value": "Albert Einstein"
},
{
"operation": "Page search",
"time": "185.025",
"value": "special relativ"
},
{
"operation": "hyperlink",
"time": "226.679",
"value": "special relativity"
},
{
"operation": "Now on",
"time": "227.94",
"value": "Special relativity"
},
{
"operation": "Page search",
"time": "273.679",
"value": "spacetime"
},
{
"operation": "Page search",
"time": "281.308",
"value": "unified spacetime"
},
{
"operation": "Page search",
"time": "303.139",
"value": "Einstein Field Equations"
},
{
"operation": "Page search",
"time": "305.09",
"value": "Einstein Field Equation"
},
{
"operation": "search",
"time": "410.111",
"value": "Spacetime"
},
{
"operation": "Now on",
"time": "411.611",
"value": "Spacetime"
},
{
"operation": "Page search",
"time": "419.226",
"value": "relativ"
},
{
"operation": "search",
"time": "592.754",
"value": "general relativity and special relativity unified spacetime"
},
{
"operation": "Now on",
"time": "594.618",
"value": "undefined"
},
{
"operation": "search",
"time": "601.729",
"value": "unified spacetime"
},
{
"operation": "Now on",
"time": "602.637",
"value": "undefined"
},
{
"operation": "search",
"time": "633.854",
"value": "Einstein field equations"
},
{
"operation": "Now on",
"time": "634.288",
"value": "Einstein field equations"
},
{
"operation": "Page search",
"time": "639.677",
"value": "unified"
},
{
"operation": "Page search",
"time": "647.67",
"value": "spacetime"
},
{
"operation": "Highlighting",
"time": "677.753",
"value": "Einstein field equations_sentence_1"
},
{
"operation": "Page search",
"time": "681.23",
"value": "relative"
},
{
"operation": "Page search",
"time": "684.55",
"value": "special rela"
},
{
"operation": "Highlighting deleted",
"time": "693.876",
"value": "Einstein field equations_sentence_1"
},
{
"operation": "Highlighting",
"time": "694.358",
"value": "Einstein field equations_sentence_1"
},
{
"operation": "finish",
"time": "702.898",
"value": "finish"
}
]
}
|
27786
|
Hattiesburg School District has a predominantly white student population in the 2006-2007 school year.
|
{
"evidence": [
"Hattiesburg Public School District_sentence_17"
],
"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:\nHattiesburg School District has a predominantly white student population in the 2006-2007 school year.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Hattiesburg Public School District_sentence_17"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 6350,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.559",
"value": "?search="
},
{
"operation": "search",
"time": "7.304",
"value": "Hattiesburg School District"
},
{
"operation": "Now on",
"time": "9.362",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "36.893",
"value": "2006-07 school year Hattiesburg Public School District_section_6"
},
{
"operation": "Now on",
"time": "39.987",
"value": "Hattiesburg Public School District#2006-07 school year Hattiesburg Public School District section 6"
},
{
"operation": "Highlighting",
"time": "47.188",
"value": "Hattiesburg Public School District_sentence_17"
},
{
"operation": "Now on",
"time": "63.245",
"value": "Hattiesburg Public School District#2006-07 school year Hattiesburg Public School District section 6"
},
{
"operation": "finish",
"time": "68.404",
"value": "finish"
}
]
}
|
45157
|
The 1979 Adur District Council election resulted in the conservatives gaining control of the council which is a political party in the United Kingdom and sits on the centre-right of the British political spectrum and have been in government since 2010 and as of 2020 hold an overall majority.
|
{
"evidence": [
"1979 Adur District Council election_sentence_3",
"Conservative Party (UK)_sentence_0",
"Conservative Party (UK)_sentence_1",
"Conservative Party (UK)_sentence_2"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe 1979 Adur District Council election resulted in the conservatives gaining control of the council which is a political party in the United Kingdom and sits on the centre-right of the British political spectrum and have been in government since 2010 and as of 2020 hold an overall majority.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"1979 Adur District Council election_sentence_3",
"Conservative Party (UK)_sentence_0",
"Conservative Party (UK)_sentence_1",
"Conservative Party (UK)_sentence_2"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6351,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "3.304",
"value": "undefined"
},
{
"operation": "search",
"time": "13.51",
"value": "1979 Adur District Council election"
},
{
"operation": "Now on",
"time": "15.071",
"value": "1979 Adur District Council election"
},
{
"operation": "Highlighting",
"time": "21.442",
"value": "1979 Adur District Council election_sentence_3"
},
{
"operation": "hyperlink",
"time": "27.715",
"value": "Conservatives"
},
{
"operation": "Now on",
"time": "29.809",
"value": "Conservative Party (UK)"
},
{
"operation": "Highlighting",
"time": "31.086",
"value": "Conservative Party (UK)_sentence_0"
},
{
"operation": "Highlighting",
"time": "36.536",
"value": "Conservative Party (UK)_sentence_1"
},
{
"operation": "Highlighting",
"time": "42.456",
"value": "Conservative Party (UK)_sentence_2"
},
{
"operation": "finish",
"time": "49.329",
"value": "finish"
}
]
}
|
50159
|
Margie Bowes, who began singing in elementary school and was appearing on television programs at the age of thirteen, was born on March 18, 1941 in Roxboro, North Carolina.
|
{
"evidence": [
"Margie Bowes_sentence_4",
"Margie Bowes_sentence_5",
"Margie Bowes_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:\nMargie Bowes, who began singing in elementary school and was appearing on television programs at the age of thirteen, was born on March 18, 1941 in Roxboro, North Carolina.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Margie Bowes_sentence_4",
"Margie Bowes_sentence_5",
"Margie Bowes_sentence_3"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6352,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.489",
"value": "?search="
},
{
"operation": "search",
"time": "11.748",
"value": "Margie Bowes"
},
{
"operation": "Now on",
"time": "12.5",
"value": "Margie Bowes"
},
{
"operation": "Highlighting",
"time": "17.243",
"value": "Margie Bowes_header_cell_0_2_0"
},
{
"operation": "Highlighting",
"time": "18.044",
"value": "Margie Bowes_cell_0_2_1"
},
{
"operation": "Highlighting",
"time": "18.047",
"value": "Margie Bowes_header_cell_0_2_0"
},
{
"operation": "Highlighting",
"time": "18.047",
"value": "Margie Bowes_header_cell_0_0_0"
},
{
"operation": "Highlighting",
"time": "26.778",
"value": "Margie Bowes_sentence_4"
},
{
"operation": "Highlighting",
"time": "29.362",
"value": "Margie Bowes_sentence_5"
},
{
"operation": "Highlighting deleted",
"time": "35.786",
"value": "Margie Bowes_cell_0_2_1"
},
{
"operation": "Highlighting deleted",
"time": "35.788",
"value": "Margie Bowes_header_cell_0_2_0"
},
{
"operation": "Highlighting deleted",
"time": "35.788",
"value": "Margie Bowes_header_cell_0_0_0"
},
{
"operation": "Highlighting",
"time": "36.914",
"value": "Margie Bowes_sentence_3"
},
{
"operation": "finish",
"time": "41.714",
"value": "finish"
}
]
}
|
40565
|
Diaminopimelate dehydrogenase ( 2007, 4 structures have been solved for this class of enzymes) is an enzyme that catalyzes the chemical reaction and has two substrates.
|
{
"evidence": [
"Diaminopimelate dehydrogenase_sentence_6",
"Diaminopimelate dehydrogenase_sentence_0",
"Diaminopimelate dehydrogenase_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:\nDiaminopimelate dehydrogenase ( 2007, 4 structures have been solved for this class of enzymes) is an enzyme that catalyzes the chemical reaction and has two substrates.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Diaminopimelate dehydrogenase_sentence_6",
"Diaminopimelate dehydrogenase_sentence_0",
"Diaminopimelate dehydrogenase_sentence_1"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 6353,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.669",
"value": "?search="
},
{
"operation": "search",
"time": "11.418",
"value": "Diaminopimelate dehydrogenase"
},
{
"operation": "Now on",
"time": "12.088",
"value": "Diaminopimelate dehydrogenase"
},
{
"operation": "Highlighting",
"time": "17.239",
"value": "Diaminopimelate dehydrogenase_sentence_6"
},
{
"operation": "Highlighting",
"time": "21.415",
"value": "Diaminopimelate dehydrogenase_sentence_0"
},
{
"operation": "Highlighting",
"time": "24.303",
"value": "Diaminopimelate dehydrogenase_sentence_1"
},
{
"operation": "finish",
"time": "32.782",
"value": "finish"
}
]
}
|
exfever_1888
|
Christopher Robert Evans is an American actor who gained attention for his portrayal of a fictional superhero appearing in American comic books published by Marvel Comics.
|
{
"evidence": [
"Chris Evans (actor)",
"Human Torch"
],
"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:\nChristopher Robert Evans is an American actor who gained attention for his portrayal of a fictional superhero appearing in American comic books published by Marvel Comics.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Chris Evans (actor)",
"Human Torch"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6354,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
12599
|
Melissa McMorrow is an American professional boxer which is a prizefighting, regulated, and sanctioned boxing.
|
{
"evidence": [
"Professional boxing_sentence_2",
"Melissa McMorrow_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:\nMelissa McMorrow is an American professional boxer which is a prizefighting, regulated, and sanctioned boxing.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Professional boxing_sentence_2",
"Melissa McMorrow_sentence_0"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6355,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.52",
"value": "?search="
},
{
"operation": "search",
"time": "19.293",
"value": "Melissa McMorrow"
},
{
"operation": "Now on",
"time": "20.207",
"value": "Melissa McMorrow"
},
{
"operation": "hyperlink",
"time": "25.607",
"value": "professional boxer"
},
{
"operation": "Now on",
"time": "26.311",
"value": "Professional boxing"
},
{
"operation": "Highlighting",
"time": "45.377",
"value": "Professional boxing_sentence_2"
},
{
"operation": "back-button-clicked",
"time": "48.887",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "48.932",
"value": "Melissa McMorrow"
},
{
"operation": "Highlighting",
"time": "50.176",
"value": "Melissa McMorrow_sentence_0"
},
{
"operation": "finish",
"time": "68.967",
"value": "finish"
}
]
}
|
55731
|
The Federation Cup was a replacement of the Indian Super Cup, where now they are open to all ten teams in the I-League and Indian Super League.
|
{
"evidence": [
"Super Cup (India)_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:\nThe Federation Cup was a replacement of the Indian Super Cup, where now they are open to all ten teams in the I-League and Indian Super League.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Super Cup (India)_sentence_3"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 6356,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.276",
"value": "?search="
},
{
"operation": "search",
"time": "35.119",
"value": "The Federation Cup was a replacement of the Indian Super Cup"
},
{
"operation": "Now on",
"time": "37.185",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "114.303",
"value": "Super Cup (India)"
},
{
"operation": "Now on",
"time": "115.05",
"value": "Super Cup (India)"
},
{
"operation": "Highlighting",
"time": "215.292",
"value": "Super Cup (India)_sentence_1"
},
{
"operation": "Highlighting deleted",
"time": "218.677",
"value": "Super Cup (India)_sentence_1"
},
{
"operation": "Highlighting",
"time": "229.543",
"value": "Super Cup (India)_sentence_1"
},
{
"operation": "Highlighting",
"time": "230.98",
"value": "Super Cup (India)_sentence_2"
},
{
"operation": "Highlighting",
"time": "238.445",
"value": "Super Cup (India)_sentence_3"
},
{
"operation": "Highlighting deleted",
"time": "241.183",
"value": "Super Cup (India)_sentence_2"
},
{
"operation": "Highlighting deleted",
"time": "241.813",
"value": "Super Cup (India)_sentence_1"
},
{
"operation": "finish",
"time": "251.653",
"value": "finish"
}
]
}
|
70347
|
In 2002, Daniel Lalonde was hired by LVMH (Louis Vuitton Moët Hennessy) as president and CEO of LVMH Watches & Jewelry, and then he recruited celebrities, such as Leonardo DiCaprio and Uma Thurman to take part in watch campaigns.
|
{
"evidence": [
"Daniel Lalonde_sentence_1",
"LVMH_sentence_0",
"Daniel Lalonde_sentence_17"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nIn 2002, Daniel Lalonde was hired by LVMH (Louis Vuitton Moët Hennessy) as president and CEO of LVMH Watches & Jewelry, and then he recruited celebrities, such as Leonardo DiCaprio and Uma Thurman to take part in watch campaigns.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Daniel Lalonde_sentence_1",
"LVMH_sentence_0",
"Daniel Lalonde_sentence_17"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 6357,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.395",
"value": "?search="
},
{
"operation": "search",
"time": "26.926",
"value": "LVMH Watches & Jewelry"
},
{
"operation": "Now on",
"time": "28.221",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "36.331",
"value": "Daniel Lalonde"
},
{
"operation": "Now on",
"time": "37.598",
"value": "Daniel Lalonde"
},
{
"operation": "Page search",
"time": "47.93",
"value": "lvmh"
},
{
"operation": "Highlighting",
"time": "56.52",
"value": "Daniel Lalonde_sentence_1"
},
{
"operation": "hyperlink",
"time": "58.612",
"value": "LVMH"
},
{
"operation": "Now on",
"time": "59.404",
"value": "LVMH"
},
{
"operation": "Highlighting",
"time": "64.097",
"value": "LVMH_sentence_0"
},
{
"operation": "back-button-clicked",
"time": "66.67",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "66.846",
"value": "Daniel Lalonde"
},
{
"operation": "Page search",
"time": "87.866",
"value": "leonardo"
},
{
"operation": "Highlighting",
"time": "124.907",
"value": "Daniel Lalonde_sentence_17"
},
{
"operation": "search",
"time": "153.151",
"value": "leonardo ddicaprio lvmh"
},
{
"operation": "Now on",
"time": "153.975",
"value": "undefined"
},
{
"operation": "finish",
"time": "332.016",
"value": "finish"
}
]
}
|
exfever_8977
|
House of 1000 Corpses is a 2004 comedy horror film directed by an American singer, filmmaker, and voice actor.
|
{
"evidence": [
"House of 1000 Corpses",
"Rob Zombie"
],
"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:\nHouse of 1000 Corpses is a 2004 comedy horror film directed by an American singer, filmmaker, and voice actor.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"House of 1000 Corpses",
"Rob Zombie"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 6358,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
449
|
Lucas Scott(One Tree Hill) was played by Chad Michael Murry who felt a connection with the character, what was what the producers planned.
|
{
"evidence": [
"Lucas Scott_sentence_7",
"Lucas Scott_sentence_8",
"Lucas Scott_sentence_12"
],
"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:\nLucas Scott(One Tree Hill) was played by Chad Michael Murry who felt a connection with the character, what was what the producers planned.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Lucas Scott_sentence_7",
"Lucas Scott_sentence_8",
"Lucas Scott_sentence_12"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 6359,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "search",
"time": "27.212",
"value": "Lucas Scott"
},
{
"operation": "Highlighting",
"time": "32.325",
"value": "Lucas Scott_sentence_0"
},
{
"operation": "Highlighting",
"time": "87.657",
"value": "Lucas Scott_sentence_7"
},
{
"operation": "Highlighting",
"time": "95.096",
"value": "Lucas Scott_sentence_8"
},
{
"operation": "Highlighting",
"time": "96.898",
"value": "Lucas Scott_sentence_9"
},
{
"operation": "Highlighting",
"time": "109.74",
"value": "Lucas Scott_sentence_12"
},
{
"operation": "Highlighting deleted",
"time": "121.442",
"value": "Lucas Scott_sentence_0"
},
{
"operation": "Highlighting deleted",
"time": "130.543",
"value": "Lucas Scott_sentence_9"
},
{
"operation": "finish",
"time": "148.403",
"value": "finish"
}
]
}
|
exfever_5552
|
European football clubs are among the best football teams in the world.
|
{
"evidence": [
"RB Leipzig",
"Red Bull Arena (Leipzig)",
"FC Sachsen Leipzig"
],
"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:\nEuropean football clubs are among the best football teams in the world.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"RB Leipzig",
"Red Bull Arena (Leipzig)",
"FC Sachsen Leipzig"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 6360,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_2528
|
Philadelphia is the largest city in the Commonwealth of Pennsylvania in a country primarily located in North America.
|
{
"evidence": [
"Philadelphia",
"United States"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nPhiladelphia is the largest city in the Commonwealth of Pennsylvania in a country primarily located in North America.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Philadelphia",
"United States"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6361,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7393
|
Michael Jackson was the oldest member of the Jackson 5.
|
{
"evidence": [
"Jackson family",
"Five Stairsteps"
],
"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:\nMichael Jackson was the oldest member of the Jackson 5.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Jackson family",
"Five Stairsteps"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 6362,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7837
|
Anneliese van der Pol is a Dutch-born American actress whose films include a 2010 American parody vampire film starring an American actor and model who has gained fame by playing Liam Court in The CW's teen drama series 90210.
|
{
"evidence": [
"Anneliese van der Pol",
"Vampires Suck",
"Matt Lanter"
],
"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:\nAnneliese van der Pol is a Dutch-born American actress whose films include a 2010 American parody vampire film starring an American actor and model who has gained fame by playing Liam Court in The CW's teen drama series 90210.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Anneliese van der Pol",
"Vampires Suck",
"Matt Lanter"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6363,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_6174
|
Elsa Lafuente Medianu appeared in several films, including a 2009 American neo-noir action crime thriller film directed by an English film director.?
|
{
"evidence": [
"Elsa Pataky",
"Give 'Em Hell, Malone",
"Russell Mulcahy"
],
"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:\nElsa Lafuente Medianu appeared in several films, including a 2009 American neo-noir action crime thriller film directed by an English film director.?\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Elsa Pataky",
"Give 'Em Hell, Malone",
"Russell Mulcahy"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 6364,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
35561
|
Mother Teresa, honoured in the Catholic Church as Saint Teresa of Calcutta, is the only Indian Nobel laureate in the field of peace.
|
{
"evidence": [
"Mother Teresa_sentence_9"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nMother Teresa, honoured in the Catholic Church as Saint Teresa of Calcutta, is the only Indian Nobel laureate in the field of peace.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"Mother Teresa_sentence_9"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 6365,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.138",
"value": "undefined"
},
{
"operation": "search",
"time": "29.293",
"value": "Mother Teresa"
},
{
"operation": "Now on",
"time": "30.731",
"value": "Mother Teresa"
},
{
"operation": "Page search",
"time": "79.709",
"value": "Nobel"
},
{
"operation": "Page search",
"time": "80.46",
"value": "Nobel"
},
{
"operation": "Page search",
"time": "80.744",
"value": "Nobel"
},
{
"operation": "search",
"time": "107.809",
"value": "nobel Peace laureates"
},
{
"operation": "Now on",
"time": "109.591",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "112.029",
"value": "Nobel Peace laureates"
},
{
"operation": "Now on",
"time": "112.949",
"value": "List of Nobel Peace Prize laureates"
},
{
"operation": "Highlighting",
"time": "122.709",
"value": "List of Nobel Peace Prize laureates_item_1_5"
},
{
"operation": "Highlighting",
"time": "122.713",
"value": "List of Nobel Peace Prize laureates_title"
},
{
"operation": "Highlighting",
"time": "122.716",
"value": "List of Nobel Peace Prize laureates_section_2"
},
{
"operation": "Highlighting deleted",
"time": "124.344",
"value": "List of Nobel Peace Prize laureates_item_1_5"
},
{
"operation": "Highlighting deleted",
"time": "124.349",
"value": "List of Nobel Peace Prize laureates_title"
},
{
"operation": "Highlighting deleted",
"time": "124.356",
"value": "List of Nobel Peace Prize laureates_section_2"
},
{
"operation": "Highlighting",
"time": "143.475",
"value": "List of Nobel Peace Prize laureates_sentence_3"
},
{
"operation": "Highlighting",
"time": "143.478",
"value": "List of Nobel Peace Prize laureates_title"
},
{
"operation": "Highlighting deleted",
"time": "145.08",
"value": "List of Nobel Peace Prize laureates_sentence_3"
},
{
"operation": "Highlighting deleted",
"time": "145.085",
"value": "List of Nobel Peace Prize laureates_title"
},
{
"operation": "hyperlink",
"time": "172.342",
"value": "List of Nobel laureates"
},
{
"operation": "Now on",
"time": "172.993",
"value": "List of Nobel laureates"
},
{
"operation": "back-button-clicked",
"time": "200.688",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "201.151",
"value": "List of Nobel Peace Prize laureates"
},
{
"operation": "back-button-clicked",
"time": "205.186",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "205.218",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "206.039",
"value": "undefined"
},
{
"operation": "back-button-clicked",
"time": "207.942",
"value": "back-button-clicked"
},
{
"operation": "back-button-clicked",
"time": "207.98",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "208.521",
"value": "Mother Teresa"
},
{
"operation": "search",
"time": "214.947",
"value": "nobel"
},
{
"operation": "Now on",
"time": "215.864",
"value": "Nobel"
},
{
"operation": "back-button-clicked",
"time": "216.876",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "217.542",
"value": "Mother Teresa"
},
{
"operation": "Page search",
"time": "220.474",
"value": "nobel"
},
{
"operation": "Highlighting",
"time": "224.198",
"value": "Mother Teresa_sentence_9"
},
{
"operation": "finish",
"time": "227.979",
"value": "finish"
}
]
}
|
exfever_3323
|
Harvey Kurtzman began to work on the New Trend line of comic books at an American publisher of comic books notably the American bi-monthly drama comic anthology series.
|
{
"evidence": [
"Harvey Kurtzman",
"EC Comics",
"Tales from the Crypt (comics)"
],
"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:\nHarvey Kurtzman began to work on the New Trend line of comic books at an American publisher of comic books notably the American bi-monthly drama comic anthology series.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Harvey Kurtzman",
"EC Comics",
"Tales from the Crypt (comics)"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 6366,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
95434
|
Donna Mae Mims is the first woman to win a Sports Car Club of America, a club that runs many programs for both amateur and professional racers.
|
{
"evidence": [
"Donna Mae Mims_sentence_1",
"Sports Car Club of America_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:\nDonna Mae Mims is the first woman to win a Sports Car Club of America, a club that runs many programs for both amateur and professional racers.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"Donna Mae Mims_sentence_1",
"Sports Car Club of America_sentence_1"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6367,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.655",
"value": "?search="
},
{
"operation": "search",
"time": "8.262",
"value": "Donna Mae Mims"
},
{
"operation": "Now on",
"time": "9.022",
"value": "Donna Mae Mims"
},
{
"operation": "Page search",
"time": "27.112",
"value": "sports car"
},
{
"operation": "Highlighting",
"time": "30.882",
"value": "Donna Mae Mims_sentence_1"
},
{
"operation": "hyperlink",
"time": "33.427",
"value": "Sports Car Club of America"
},
{
"operation": "Now on",
"time": "34.386",
"value": "Sports Car Club of America"
},
{
"operation": "Highlighting",
"time": "38.002",
"value": "Sports Car Club of America_sentence_1"
},
{
"operation": "finish",
"time": "46.019",
"value": "finish"
}
]
}
|
34264
|
2009 The Island 300 was the twelfth event of the 2009 V8 Supercar Championship Series which includes thirteen V8 Supercar Championship Series.
|
{
"evidence": [
"2009 V8 Supercar Championship Series_sentence_0",
"2009 The Island 300_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:\n2009 The Island 300 was the twelfth event of the 2009 V8 Supercar Championship Series which includes thirteen V8 Supercar Championship Series.\n",
"role": "user"
}
] |
Multi-hop Reasoning
|
{
"ground_truth": {
"evidence": [
"2009 V8 Supercar Championship Series_sentence_0",
"2009 The Island 300_sentence_0"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6368,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.51",
"value": "undefined"
},
{
"operation": "search",
"time": "17.521",
"value": "2009 The Island 300"
},
{
"operation": "Now on",
"time": "18.329",
"value": "2009 The Island 300"
},
{
"operation": "Page search",
"time": "34.031",
"value": "V8 Supercar Championship Series"
},
{
"operation": "hyperlink",
"time": "49.32",
"value": "2009 V8 Supercar Championship Series"
},
{
"operation": "Now on",
"time": "50.417",
"value": "2009 V8 Supercar Championship Series"
},
{
"operation": "Highlighting",
"time": "56.835",
"value": "2009 V8 Supercar Championship Series_sentence_0"
},
{
"operation": "search",
"time": "143.858",
"value": "V8 Supercar Championship Series"
},
{
"operation": "Now on",
"time": "145.3",
"value": "Supercars Championship"
},
{
"operation": "back-button-clicked",
"time": "189.715",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "190.311",
"value": "2009 V8 Supercar Championship Series"
},
{
"operation": "search",
"time": "219.702",
"value": "2009 The Island 300"
},
{
"operation": "Now on",
"time": "220.246",
"value": "2009 The Island 300"
},
{
"operation": "Highlighting",
"time": "221.625",
"value": "2009 The Island 300_sentence_0"
},
{
"operation": "finish",
"time": "249.965",
"value": "finish"
}
]
}
|
exfever_4971
|
Dangerous Beauty is a 1998 British biographical drama film starring an English actress on stage and screen who appeared in several films, including Braveheart and The Land Girls.
|
{
"evidence": [
"Dangerous Beauty",
"Catherine McCormack"
],
"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:\nDangerous Beauty is a 1998 British biographical drama film starring an English actress on stage and screen who appeared in several films, including Braveheart and The Land Girls.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Dangerous Beauty",
"Catherine McCormack"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 6369,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_7313
|
Edgar Allan Poe is regarded in the US as a central figure in the artistic, literary, musical, and intellectual movements that originated in Europe towards the end of the 17th century.
|
{
"evidence": [
"Edgar Allan Poe",
"Romanticism"
],
"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:\nEdgar Allan Poe is regarded in the US as a central figure in the artistic, literary, musical, and intellectual movements that originated in Europe towards the end of the 17th century.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Edgar Allan Poe",
"Romanticism"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 6370,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
17005
|
Finch Field, located in Thomasville, North Carolina has been renovated four times since 2002.
|
{
"evidence": [
"Finch Field_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:\nFinch Field, located in Thomasville, North Carolina has been renovated four times since 2002.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Finch Field_sentence_2"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 6371,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "0.995",
"value": "?search="
},
{
"operation": "search",
"time": "11.674",
"value": "Finch Field"
},
{
"operation": "Now on",
"time": "12.377",
"value": "Finch Field"
},
{
"operation": "Highlighting",
"time": "19.138",
"value": "Finch Field_sentence_0"
},
{
"operation": "Highlighting",
"time": "26.505",
"value": "Finch Field_sentence_2"
},
{
"operation": "Highlighting deleted",
"time": "33.69",
"value": "Finch Field_sentence_0"
},
{
"operation": "finish",
"time": "41.433",
"value": "finish"
}
]
}
|
exfever_4445
|
Drop Dead Gorgeous, earn $10,571,408 in Domestic Cumulative Box Office Records.
|
{
"evidence": [
"Drop Dead Gorgeous (film)",
"Brittany Murphy"
],
"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:\nDrop Dead Gorgeous, earn $10,571,408 in Domestic Cumulative Box Office Records.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Drop Dead Gorgeous (film)",
"Brittany Murphy"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 6372,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_8651
|
Awakening is Jackie Evancho's best-selling album.
|
{
"evidence": [
"Jackie Evancho",
"Crossover music",
"Musical composition"
],
"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:\nAwakening is Jackie Evancho's best-selling album.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Jackie Evancho",
"Crossover music",
"Musical composition"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 6373,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_4247
|
Stan Lee and Steve Ditko created Spider-Man, a stock character that possesses a superpower.
|
{
"evidence": [
"Spider-Man",
"Superhero"
],
"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:\nStan Lee and Steve Ditko created Spider-Man, a stock character that possesses a superpower.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Spider-Man",
"Superhero"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6374,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
16918
|
The Province of Udine was a province in Italy near Austria and Slovenia.
|
{
"evidence": [
"Province of Udine_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:\nThe Province of Udine was a province in Italy near Austria and Slovenia.\n",
"role": "user"
}
] |
Entity Disambiguation
|
{
"ground_truth": {
"evidence": [
"Province of Udine_sentence_0"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6375,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.168",
"value": "?search="
},
{
"operation": "search",
"time": "14.871",
"value": "Province of Udine"
},
{
"operation": "Now on",
"time": "16.041",
"value": "Province of Udine"
},
{
"operation": "Highlighting",
"time": "18.944",
"value": "Province of Udine_sentence_0"
},
{
"operation": "Highlighting",
"time": "29.492",
"value": "Province of Udine_sentence_2"
},
{
"operation": "Highlighting deleted",
"time": "52.908",
"value": "Province of Udine_sentence_2"
},
{
"operation": "finish",
"time": "63.495",
"value": "finish"
}
]
}
|
exfever_4805
|
Catherine Keener is the youngest actress in the film Get Out.
|
{
"evidence": [
"Get Out (film)",
"Allison Williams (actress)"
],
"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:\nCatherine Keener is the youngest actress in the film Get Out.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Get Out (film)",
"Allison Williams (actress)"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 6376,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
53802
|
The 1931 Colgate football team under head coach Andrew Kerr won eight of nine games that season.
|
{
"evidence": [
"1931 Colgate football team_sentence_1"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe 1931 Colgate football team under head coach Andrew Kerr won eight of nine games that season.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"1931 Colgate football team_sentence_1"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6377,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.965",
"value": "?search="
},
{
"operation": "search",
"time": "12.773",
"value": "Andrew Kerr"
},
{
"operation": "Now on",
"time": "14.135",
"value": "Andrew Kerr"
},
{
"operation": "hyperlink",
"time": "84.364",
"value": "Colgate"
},
{
"operation": "Now on",
"time": "85.737",
"value": "Colgate Raiders football"
},
{
"operation": "back-button-clicked",
"time": "170.154",
"value": "back-button-clicked"
},
{
"operation": "Now on",
"time": "170.25",
"value": "Andrew Kerr"
},
{
"operation": "Highlighting",
"time": "179.01",
"value": "Andrew Kerr_sentence_3"
},
{
"operation": "Highlighting",
"time": "179.011",
"value": "Andrew Kerr_title"
},
{
"operation": "search",
"time": "204.768",
"value": "1931 Colgate football team"
},
{
"operation": "Now on",
"time": "205.421",
"value": "1931 Colgate football team"
},
{
"operation": "Highlighting",
"time": "222.617",
"value": "1931 Colgate football team_sentence_1"
},
{
"operation": "finish",
"time": "257.174",
"value": "finish"
}
]
}
|
47068
|
At the age of 15, Anastasia Michaelsdotter is a former model, journalist, and creative director that got scouted by several modeling agencies and signed with Model International Management and Elite Model Management, then she moved to Stockholm and lived in both Milan & Paris by the age of 18.
|
{
"evidence": [
"Anastasia Michaelsdotter_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:\nAt the age of 15, Anastasia Michaelsdotter is a former model, journalist, and creative director that got scouted by several modeling agencies and signed with Model International Management and Elite Model Management, then she moved to Stockholm and lived in both Milan & Paris by the age of 18.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Anastasia Michaelsdotter_sentence_7"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 6378,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "2.165",
"value": "?search="
},
{
"operation": "search",
"time": "10.533",
"value": "Anastasia Michaelsdotter"
},
{
"operation": "Now on",
"time": "11.499",
"value": "Anastasia Michaelsdotter"
},
{
"operation": "Highlighting",
"time": "826.112",
"value": "Anastasia Michaelsdotter_sentence_0"
},
{
"operation": "Highlighting deleted",
"time": "826.697",
"value": "Anastasia Michaelsdotter_sentence_0"
},
{
"operation": "Highlighting",
"time": "893.578",
"value": "Anastasia Michaelsdotter_sentence_0"
},
{
"operation": "Highlighting",
"time": "905.567",
"value": "Anastasia Michaelsdotter_sentence_6"
},
{
"operation": "Highlighting",
"time": "909.76",
"value": "Anastasia Michaelsdotter_sentence_7"
},
{
"operation": "Highlighting deleted",
"time": "914.565",
"value": "Anastasia Michaelsdotter_sentence_6"
},
{
"operation": "Highlighting deleted",
"time": "916.789",
"value": "Anastasia Michaelsdotter_sentence_0"
},
{
"operation": "Highlighting",
"time": "993.892",
"value": "Anastasia Michaelsdotter_sentence_0"
},
{
"operation": "Highlighting",
"time": "1001.709",
"value": "Anastasia Michaelsdotter_sentence_6"
},
{
"operation": "Highlighting deleted",
"time": "1057.184",
"value": "Anastasia Michaelsdotter_sentence_6"
},
{
"operation": "Highlighting deleted",
"time": "1058.5",
"value": "Anastasia Michaelsdotter_sentence_0"
},
{
"operation": "finish",
"time": "1060.826",
"value": "finish"
}
]
}
|
62588
|
The Saint Louis Blue and White Football team took a break during the month of July in 1904.
|
{
"evidence": [
"1904 Saint Louis Blue and White football team_sentence_0"
],
"label": "NOT ENOUGH INFO"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Saint Louis Blue and White Football team took a break during the month of July in 1904.\n",
"role": "user"
}
] |
Other
|
{
"ground_truth": {
"evidence": [
"1904 Saint Louis Blue and White football team_sentence_0"
],
"label": "NOT ENOUGH INFO"
},
"style": "rule"
}
|
{
"index": 6379,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.432",
"value": "?search="
},
{
"operation": "search",
"time": "14.219",
"value": "Saint Louis Blue and White Football team"
},
{
"operation": "Now on",
"time": "16.935",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "31.108",
"value": "1904 Saint Louis Blue and White football team"
},
{
"operation": "Now on",
"time": "31.816",
"value": "1904 Saint Louis Blue and White football team"
},
{
"operation": "hyperlink",
"time": "58.337",
"value": "1904 college football season"
},
{
"operation": "Now on",
"time": "59.638",
"value": "1904 college football season"
},
{
"operation": "search",
"time": "75.631",
"value": "Saint Louis Blue and White Football team"
},
{
"operation": "Now on",
"time": "76.826",
"value": "undefined"
},
{
"operation": "hyperlink",
"time": "79.919",
"value": "1904 Saint Louis Blue and White football team"
},
{
"operation": "Now on",
"time": "81.014",
"value": "1904 Saint Louis Blue and White football team"
},
{
"operation": "Highlighting",
"time": "85.58",
"value": "1904 Saint Louis Blue and White football team_sentence_0"
},
{
"operation": "finish",
"time": "96.263",
"value": "finish"
}
]
}
|
92622
|
The Paralympic Oath was not used in the 1950 Summer Paralympics.
|
{
"evidence": [
"Paralympic symbols_sentence_58"
],
"label": "SUPPORT"
}
|
feverous
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\nThe Paralympic Oath was not used in the 1950 Summer Paralympics.\n",
"role": "user"
}
] |
Numerical Reasoning
|
{
"ground_truth": {
"evidence": [
"Paralympic symbols_sentence_58"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6380,
"split": "train_online"
}
|
{
"annotator_operations": [
{
"operation": "start",
"time": "0",
"value": "start"
},
{
"operation": "Now on",
"time": "1.381",
"value": "?search="
},
{
"operation": "search",
"time": "14.227",
"value": "Paralympic Oath"
},
{
"operation": "Now on",
"time": "16.445",
"value": "Paralympic Oath"
},
{
"operation": "hyperlink",
"time": "25.566",
"value": "8 Paralympic Oath Paralympic symbols_section_9"
},
{
"operation": "Highlighting",
"time": "49.918",
"value": "Paralympic symbols_sentence_58"
},
{
"operation": "finish",
"time": "64.331",
"value": "finish"
}
]
}
|
exfever_6909
|
Seth MacFarlane is an American actor, screenwriter, producer, director, and singer who wrote, directed, and starred in the 2016 American comedy film starring Mark Wahlberg.
|
{
"evidence": [
"Seth MacFarlane",
"Ted 2"
],
"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:\nSeth MacFarlane is an American actor, screenwriter, producer, director, and singer who wrote, directed, and starred in the 2016 American comedy film starring Mark Wahlberg.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Seth MacFarlane",
"Ted 2"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 6381,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_8807
|
"See You Again" is the debut single by an American singer who emerged as a celebrity with a large teenage fan base while portraying the title character of the Disney Channel television series, Hannah Montana.
|
{
"evidence": [
"See You Again (Miley Cyrus song)",
"Miley Cyrus",
"Teen idol"
],
"label": "SUPPORT"
}
|
exfever
|
[
{
"content": "You are a claim-verification assistant. You MUST follow this protocol exactly:\n\n<plan>…</plan>\n- Once at the start: sketch your high-level strategy, such as claim decomposition, entity recognition, etc.\n\n<search>…</search>\n- When you need a fact: emit exactly this tag with your query.\n- To make the most of your search turns, don’t repeat identical queries.\n- You can search at most three times.\n\n<information>\n[[e_1]]: info1\n[[e_2]]: info2\n…\n</information>\n- You will be given claim related information in the format above.\n\n<think>…</think> \n- Use for every piece of reasoning; do not state your final verdict here.\n- You must conduct reasoning inside <think> and </think> first every time you get new information.\n\n<answer>\nLabel: SUPPORT / REFUTE / NOT ENOUGH INFO\nEvidence: [[e_1]], [[e_3]], ...\n</answer> \n- Emit exactly once at the end, no extra text or tags.\n- Evidence id such as e_1 will be replaced by real ids from the corpus. You must include useful real ids when answering\n- Evidence outputs must strictly enforce the format [[e_i]], [[e_j]]…\n- Answer Labels respectively stand for:\nSUPPORT: The claim is consistent with the cited evidence and the evidence is sufficient to confirm the claim.\nREFUTE: The claim contradicts the cited evidence and the evidence is sufficient to disprove the claim.\nNOT ENOUGH INFO: The available evidence is insufficient to determine whether the claim is true or false.\n\n- Note that the process should be: plan → (search → information → think) repeat until conclusion → answer\n\nVerify the claim:\n\"See You Again\" is the debut single by an American singer who emerged as a celebrity with a large teenage fan base while portraying the title character of the Disney Channel television series, Hannah Montana.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"See You Again (Miley Cyrus song)",
"Miley Cyrus",
"Teen idol"
],
"label": "SUPPORT"
},
"style": "rule"
}
|
{
"index": 6382,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
exfever_6539
|
Margaret of Valois was a French princess of the Valois dynasty who became Queen of Navarre by marriage to the last monarch of France from the House of Bourbon, a cadet branch of the Capetian dynasty.
|
{
"evidence": [
"Margaret of Valois",
"Henry IV of France"
],
"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:\nMargaret of Valois was a French princess of the Valois dynasty who became Queen of Navarre by marriage to the last monarch of France from the House of Bourbon, a cadet branch of the Capetian dynasty.\n",
"role": "user"
}
] |
EX-FEVER
|
{
"ground_truth": {
"evidence": [
"Margaret of Valois",
"Henry IV of France"
],
"label": "REFUTE"
},
"style": "rule"
}
|
{
"index": 6383,
"split": "train_online"
}
|
{
"annotator_operations": []
}
|
Subsets and Splits
No community queries yet
The top public SQL queries from the community will appear here once available.