FewRel is a large-scale few-shot relation classification details. Please refer to our EMNLP18 paper for details.
The evaluation is based on N-way K-shot, where for each episode, N classes, K instances per class and a query example are sampled and given these supporting examples, the models are required to classify the query example into one of the N classes.
Start: Sept. 1, 2018, midnight
You must be logged in to participate in competitions.Sign In