Question Answering on KILT: Natural Questions

Metric: EM (higher is better)

LeaderboardDataset

Loading chart...

Results

Sort:

#	Model↕	EM▼	Extra Data	Paper	Date↕	Code
1	intersect	53.74	No	-	-	-
2	Re2G	51.73	No	Re2G: Retrieve, Rerank, Generate	2022-07-13	Code
3	Wikipedia	51.59	No	-	-	-
4	Sphere	46.05	No	-	-	-
5	KGI_0	45.22	No	-	-	-
6	RAG	44.39	No	-	-	-
7	BART + DPR	41.27	No	-	-	-
8	Multitask DPR + BART	39.75	No	-	-	-
9	BERT + DPR	38.64	No	-	-	-
10	BART	21.75	No	-	-	-
11	T5-base	19.6	No	KILT: a Benchmark for Knowledge Intensive Langua...	2020-09-04	Code
12	multi-task small	0.35	No	-	-	-
13	TABi	0	No	-	-	-
14	chriskuei	0	No	-	-	-
15	GENRE	0	No	-	-	-
16	Multi-task DPR	0	No	-	-	-

#1intersect
53.74
EM
No paper
#2Re2GSOTA
51.73
EM· 2022-07-13
Re2G: Retrieve, Rerank, Generate Code
#3Wikipedia
51.59
EM
No paper
#4Sphere
46.05
EM
No paper
#5KGI_0
45.22
EM
No paper
#6RAG
44.39
EM
No paper
#7BART + DPR
41.27
EM
No paper
#8Multitask DPR + BART
39.75
EM
No paper
#9BERT + DPR
38.64
EM
No paper
#10BART
21.75
EM
No paper
#11T5-baseSOTA
19.6
EM· 2020-09-04
KILT: a Benchmark for Knowledge Intensive Language Tasks Code
#12multi-task small
0.35
EM
No paper
#13TABi
0
EM
No paper
#14chriskuei
0
EM
No paper
#15GENRE
0
EM
No paper
#16Multi-task DPR
0
EM
No paper