Penalty Decoding: Well Suppress the Self-Reinforcement Effect in Open-Ended Text Generation.

Wenhong Zhu, Hongkun Hao, Rui Wang

Browse the full EMNLP paper archive.