Share your thoughts, 1 month free Claude Pro on us
See more
Home
/
Benchmarks
Question Generation on Human Evaluation 2-hop questions (test)
Loading...
74
Well-formed Rate (Yes)
DCQG framework
26.16
38.58
51
63.42
May 25, 2021
Well-formed Rate (Yes)
Well-formed Rate (Acceptable)
Well-formed Rate (No)
Concise Rate (Yes)
Concise Rate (Acceptable)
Concise Rate (No)
Answerable Rate (Yes)
Answerable Rate (No)
Answer Matching Rate (Yes)
Answer Matching Rate (No)
Updated 1mo ago
Evaluation Results
Method
Method
Links
Well-formed Rate (Yes)
Well-formed Rate (Acceptable)
Well-formed Rate (No)
Concise Rate (Yes)
Concise Rate (Acceptable)
Concise Rate (No)
Answerable Rate (Yes)
Answerable Rate (No)
Answer Matching Rate (Yes)
Answer Matching Rate (No)
DCQG framework
difficulty=2-hop
2021.05
74
19
7
67
30
3
78
22
69
31
Gold
difficulty=2-hop
2021.05
72
22
6
56
40
4
92
8
87
13
GPT2
difficulty=2-hop
2021.05
57
34
9
47
50
3
69
31
66
34
DP-Graph
difficulty=2-hop
2021.05
28
41
31
41
53
6
49
51
39
61
Feedback
Search any
task
Search any
task