QREFINE, a BERT- and character-aware Seq2Seq model trained with PPO and answer-aware rewards, generates cleaned questions that improve answer retrieval over previous refinement methods.
Proximal Policy Optimization and its Dynamic Version for Sequence Generation
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
abstract
In sequence generation task, many works use policy gradient for model optimization to tackle the intractable backpropagation issue when maximizing the non-differentiable evaluation metrics or fooling the discriminator in adversarial learning. In this paper, we replace policy gradient with proximal policy optimization (PPO), which is a proved more efficient reinforcement learning algorithm, and propose a dynamic approach for PPO (PPO-dynamic). We demonstrate the efficacy of PPO and PPO-dynamic on conditional sequence generation tasks including synthetic experiment and chit-chat chatbot. The results show that PPO and PPO-dynamic can beat policy gradient by stability and performance.
fields
cs.IR 1years
2019 1verdicts
CONDITIONAL 1representative citing papers
citing papers explorer
-
Generative Question Refinement with Deep Reinforcement Learning in Retrieval-based QA System
QREFINE, a BERT- and character-aware Seq2Seq model trained with PPO and answer-aware rewards, generates cleaned questions that improve answer retrieval over previous refinement methods.