ing via sentence embeddings, we observe surprisingly good performance with min- imal amounts of supervised training data for a transfer task.
確定！ 回上一頁
