Dataset Viewer
Auto-converted to Parquet Duplicate
prompt
listlengths
2
2
completion
listlengths
1
1
label
bool
2 classes
reasoning
stringlengths
213
11.3k
answer
stringclasses
4 values
gt_answer
stringclasses
3 values
paper_id
stringclasses
300 values
anchor_id
stringlengths
13
13
bin
int64
0
9
input_gt
float64
1
8.67
anchor_gt
float64
0.5
8
similarity
float64
0.68
0.92
run_idx
int64
0
2
[ { "content": "You are given two reviews of two different papers. Your job is to determine which paper is better, judging from its review — evaluate the papers themselves, not the quality of the reviews. The first review is AI generated and the second one is human written; the AI reviewer can sometimes be too ni...
[ { "content": "<think>\nWe are given two reviews. Review 1 is AI-generated (likely from a template) and says the paper is incomplete, cuts off mid-sentence, missing sections, etc. It gives a score of -100 and decision Error. But we are not asked to score, just to compare which paper is better based on the review...
false
We are given two reviews. Review 1 is AI-generated (likely from a template) and says the paper is incomplete, cuts off mid-sentence, missing sections, etc. It gives a score of -100 and decision Error. But we are not asked to score, just to compare which paper is better based on the reviews. Review 2 is human-written an...
1
0
DFSb67ksVr
dHdXvu5ehy.md
5
6.67
4.75
0.760485
2
[ { "content": "You are given two reviews of two different papers. Your job is to determine which paper is better, judging from its review — evaluate the papers themselves, not the quality of the reviews. The first review is AI generated and the second one is human written; the AI reviewer can sometimes be too ni...
[ { "content": "<think>\nWe need to decide which paper is better based on the reviews. The first review is AI-generated and says the paper is incomplete (truncated), so it cannot be evaluated. The second paper (Review 2) has multiple human reviews, indicating it is a complete paper with some weaknesses but also s...
false
We need to decide which paper is better based on the reviews. The first review is AI-generated and says the paper is incomplete (truncated), so it cannot be evaluated. The second paper (Review 2) has multiple human reviews, indicating it is a complete paper with some weaknesses but also strengths. The first paper's rev...
1
0
DFSb67ksVr
GFzmAKw3RW.md
4
6.67
3.75
0.759844
2
[ { "content": "You are given two reviews of two different papers. Your job is to determine which paper is better, judging from its review — evaluate the papers themselves, not the quality of the reviews. The first review is AI generated and the second one is human written; the AI reviewer can sometimes be too ni...
[ { "content": "<think>\nWe are given two reviews. The first review is for a paper titled \"MXNET\" about estimating clique number. The review says the paper is severely truncated, missing entire sections including experiments. It gives a score of -100 and decision Error, meaning the paper is incomplete and canno...
false
We are given two reviews. The first review is for a paper titled "MXNET" about estimating clique number. The review says the paper is severely truncated, missing entire sections including experiments. It gives a score of -100 and decision Error, meaning the paper is incomplete and cannot be evaluated. The second review...
1
0
DFSb67ksVr
XWfjugkXzN.md
1
6.67
1.666667
0.743143
2
[ { "content": "You are given two reviews of two different papers. Your job is to determine which paper is better, judging from its review — evaluate the papers themselves, not the quality of the reviews. The first review is AI generated and the second one is human written; the AI reviewer can sometimes be too ni...
[ { "content": "<think>\nWe are given two reviews. Review 1 is about a paper called MXNET. The review states that the paper is severely truncated; the experiments and most method sections are missing. The reviewer says the paper cannot be evaluated and returns an error. Essentially, the paper as provided is incom...
false
We are given two reviews. Review 1 is about a paper called MXNET. The review states that the paper is severely truncated; the experiments and most method sections are missing. The reviewer says the paper cannot be evaluated and returns an error. Essentially, the paper as provided is incomplete, so it's not a valid pape...
1
0
DFSb67ksVr
F8l0llkMk0.md
3
6.67
3.333333
0.770122
1
[ { "content": "You are given two reviews of two different papers. Your job is to determine which paper is better, judging from its review — evaluate the papers themselves, not the quality of the reviews. The first review is AI generated and the second one is human written; the AI reviewer can sometimes be too ni...
[ { "content": "<think>\nWe are given two reviews. The first review is for a paper that is severely truncated, missing major sections. The review states it's incomplete and returns Error. The second review is for a paper that also seems incomplete, with human reviewers noting it's only an abstract and partial int...
false
We are given two reviews. The first review is for a paper that is severely truncated, missing major sections. The review states it's incomplete and returns Error. The second review is for a paper that also seems incomplete, with human reviewers noting it's only an abstract and partial introduction, lacking methods and ...
tie
0
DFSb67ksVr
5lUdTogEL3.md
0
6.67
1
0.773686
0
[ { "content": "You are given two reviews of two different papers. Your job is to determine which paper is better, judging from its review — evaluate the papers themselves, not the quality of the reviews. The first review is AI generated and the second one is human written; the AI reviewer can sometimes be too ni...
[ { "content": "<think>\nWe are asked to compare two papers based on their reviews. The first review is for a paper called \"MXNET\" about estimating clique number. The review states that the paper is severely truncated, missing most of the method details, all experiments, and references. The reviewer gives a dec...
false
We are asked to compare two papers based on their reviews. The first review is for a paper called "MXNET" about estimating clique number. The review states that the paper is severely truncated, missing most of the method details, all experiments, and references. The reviewer gives a decision of "Error" and says the pap...
1
0
DFSb67ksVr
AxYTFpdlvj.md
2
6.67
2
0.761111
1
[ { "content": "You are given two reviews of two different papers. Your job is to determine which paper is better, judging from its review — evaluate the papers themselves, not the quality of the reviews. The first review is AI generated and the second one is human written; the AI reviewer can sometimes be too ni...
[ { "content": "<think>\nWe are asked to compare two papers based on their reviews. The first paper's review is AI-generated and indicates the paper is severely truncated (incomplete). The second paper's review is human-written and gives mixed feedback: some weaknesses but also strengths. The instruction says to ...
false
We are asked to compare two papers based on their reviews. The first paper's review is AI-generated and indicates the paper is severely truncated (incomplete). The second paper's review is human-written and gives mixed feedback: some weaknesses but also strengths. The instruction says to evaluate the papers themselves,...
1
0
DFSb67ksVr
XWfjugkXzN.md
1
6.67
1.666667
0.743143
1
[ { "content": "You are given two reviews of two different papers. Your job is to determine which paper is better, judging from its review — evaluate the papers themselves, not the quality of the reviews. The first review is AI generated and the second one is human written; the AI reviewer can sometimes be too ni...
[ { "content": "<think>\nWe are asked to compare two papers based on their reviews. The first review says the paper is incomplete and gives a decision of \"Error\" with score -100. The second review describes a paper about lifelong person re-identification, but the human reviewers all say the submitted manuscript...
false
We are asked to compare two papers based on their reviews. The first review says the paper is incomplete and gives a decision of "Error" with score -100. The second review describes a paper about lifelong person re-identification, but the human reviewers all say the submitted manuscript is incomplete (only abstract and...
tie
0
DFSb67ksVr
5lUdTogEL3.md
0
6.67
1
0.773686
2
[ { "content": "You are given two reviews of two different papers. Your job is to determine which paper is better, judging from its review — evaluate the papers themselves, not the quality of the reviews. The first review is AI generated and the second one is human written; the AI reviewer can sometimes be too ni...
[ { "content": "<think>\nWe are asked to compare two papers based on their reviews. The first review is AI-generated, the second is human-written. The AI reviewer can be too nice, so we need to account for that.\n\nReview 1 (AI-generated) says the paper is incomplete: it is severely truncated, missing large secti...
false
We are asked to compare two papers based on their reviews. The first review is AI-generated, the second is human-written. The AI reviewer can be too nice, so we need to account for that. Review 1 (AI-generated) says the paper is incomplete: it is severely truncated, missing large sections including experiments. It giv...
tie
0
DFSb67ksVr
5lUdTogEL3.md
0
6.67
1
0.773686
1
[ { "content": "You are given two reviews of two different papers. Your job is to determine which paper is better, judging from its review — evaluate the papers themselves, not the quality of the reviews. The first review is AI generated and the second one is human written; the AI reviewer can sometimes be too ni...
[ { "content": "<think>\nWe are asked to compare two papers based on their reviews. The first review (for paper 1, MXNET) is AI-generated. The second review (for paper 2, \"On the Hölder Stability of Multiset and Graph Neural Networks\") is human-written. We need to determine which paper is better, accounting for...
true
We are asked to compare two papers based on their reviews. The first review (for paper 1, MXNET) is AI-generated. The second review (for paper 2, "On the Hölder Stability of Multiset and Graph Neural Networks") is human-written. We need to determine which paper is better, accounting for the fact that the AI reviewer ca...
1
1
DFSb67ksVr
P7KIGdgW8S.md
9
6.67
8
0.739853
2
[ { "content": "You are given two reviews of two different papers. Your job is to determine which paper is better, judging from its review — evaluate the papers themselves, not the quality of the reviews. The first review is AI generated and the second one is human written; the AI reviewer can sometimes be too ni...
[ { "content": "<think>\nWe need to decide which paper is better based on the reviews. The first review is AI-generated and describes a paper that is incomplete—the file is severely truncated, missing major sections including experiments. The review states the paper cannot be evaluated and gives a decision of \"E...
true
We need to decide which paper is better based on the reviews. The first review is AI-generated and describes a paper that is incomplete—the file is severely truncated, missing major sections including experiments. The review states the paper cannot be evaluated and gives a decision of "Error". This indicates the paper ...
1
1
DFSb67ksVr
P7KIGdgW8S.md
9
6.67
8
0.739853
1
End of preview. Expand in Data Studio
README.md exists but content is empty.
Downloads last month
45