Li, Li, Prakash, Wang, 2020. Toward Interpretability of Dual-Encoder Models for Dialogue Response Suggestions.