Share your thoughts, 1 month free Claude Pro on usSee more
WorkDL logo mark

SciTLDR

Benchmarks

Task NameDataset NameSOTA ResultTrend
Language Model CalibrationSCITLDR (dev)
smECE0.072
21
Confidence EstimationSCITLDR (dev)
smECE0.08
21
Extreme SummarizationSciTLDR 1.0 (test)
ROUGE-152.4
20
Scientific Extreme SummarizationSciTLDR (test)
ROUGE-147.7
14
Topic-Controlled SummarisationSciTLDR (test)
BERTScore0.8799
11
SummarizationSciTLDR
BERTScore0.783
10
SummarizationSciTLDR (test)
RL Score35.99
10
Scientific document summarizationSCITLDR (dev)
smECE0.08
9
Abstractive SummarizationSciTLDR AIC
ROUGE-144.9
8
Abstractive SummarizationSciTLDR Abstracts
ROUGE-143.8
8
Abstractive SummarizationSciTLDR Full Text
ROUGE-136.92
6
Text CompressionSciTLDR (test)
ROUGE-1 F144.9
4
Text SummarizationSciTLDR 76 gold papers
MRR0.6
4
Abstractive SummarizationSciTLDR AIC (test)
Mean ROUGE-132.1
4
Showing 14 of 14 rows