TabFact is a new 118K-statement benchmark for verifying text against Wikipedia tables, where baseline models score around 65 percent versus 92 percent for humans.
all of the trains are departing in the morning
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CL 1years
2019 1verdicts
ACCEPT 1representative citing papers
citing papers explorer
-
TabFact: A Large-scale Dataset for Table-based Fact Verification
TabFact is a new 118K-statement benchmark for verifying text against Wikipedia tables, where baseline models score around 65 percent versus 92 percent for humans.