1 paper · 1 filter
Jatin Ganhotra, Robert Moore, Sachindra Joshi +1
Existing benchmarks used to evaluate the performance of end-to-end neural dialog systems lack a key component: natural variation present in human conversations. Most datasets are c…