Leaderboard Logo
Leaderboards
The Ai2 Leaderboard will be archived at the end of April 2025. We plan to show past results, but will not accept new submissions. Currently, new accounts cannot be created.
SciTail: Science Entailment Logo

SciTail

The SciTail dataset is the first entailment dataset derived from an end task (science question answering) with independently authored sentences. To create this dataset, a multiple-choice question and the correct ansmore
Rank
Submission
Created
Accuracy
1
DeBERTa
Microsoft Dynamics 365 AI
10/26/20200.9770
2
CA-MTL
Elhattami, Pilault and Pal
09/27/20200.9675
2
ALUM
Xiaodong Liu
03/19/20200.9675
4
ALICE
Lis Kanashiro Pereira, Xiaodo…
05/15/20200.9624
5
KD-MT-DNN
Microsoft D365 AI & MSR AI
05/19/20190.9610
6
CMTL
Amine Elhattami, Jonathan Pil…
09/23/20200.9595
7
MT-DNN(BigBird)
Microsoft D365 AI & MSR AI
01/29/20190.9407
8
BigBird
Microsoft Dynamics 365 AI Res…
11/29/20180.9384
9
06/12/20180.8830
10
HBMP
University of Helsinki
08/28/20180.8600
11
BiLSTM Max-Out
Allen Institute for Artificia…
09/09/20180.8540
12
ConSeqNet
IBM Research AI
09/27/20180.8520
13
MIMN
Beijing Language and Culture …
08/15/20180.8400
14
CAFE
Nanyang Technological Univers…
09/11/20180.8330
15
DeIsTe
University of Pennsylvania & …
04/25/20180.8210
16
HCRN
Nanyang Technological Univers…
07/14/20180.8000
17
AdvEntuRe
Carnegie Mellon University & …
07/16/20180.7900
18
DGEM
Allen Institute for Artificia…
02/03/20180.7730
19
02/03/20180.7230
20
NGram baseline
Allen Institute for Artificia…
02/03/20180.7060
20
600D ESIM
University of Science and Tec…
02/03/20180.7060
22
Hypothesis-Only
Johns Hopkins University & BI…
05/14/20180.6660
23
02/03/20180.6030

Accuracy Over Time

01/01/201807/01/201801/01/201907/01/201901/01/202007/01/20200.000.200.400.600.801.00
Running BestSubmissionsSubmission DateAccuracy