Schema2QA 2.0

A large-scale question answering dataset on real-world Schema.org data

View the Project on GitHub stanford-oval/schema2qa

Accuracy on dev set

Schema2QA

  restaurants people movies books music hotels average
BART 76.5% 79.2% 83.4% 84.7% 77.9% 72.5% 79.0%
BERT-LSTM 64.6% 69.7% 74.0% 64.4% 65.7% 58.7% 66.2%

AutoQA

  restaurants people movies books music hotels average
BART 78.8% 73.2% 81.6% 71.7% 71.5% 72.0% 74.8%
BERT-LSTM 65.3% 53.5% 64.2% 61.9% 59.0% 61.4% 60.9%