Attending to Future Tokens For Bidirectional Sequence Generation
This addresses the problem of inefficient sequence generation for conversational AI by enabling bidirectional attention, though it is incremental as it builds on existing attention mechanisms.
The paper tackles the limitation of left-to-right token generation in neural sequence generation by introducing a bidirectional approach using placeholder tokens, which allows consideration of both past and future tokens, and shows experimental effectiveness on conversational tasks with large performance improvements over baselines.
Neural sequence generation is typically performed token-by-token and left-to-right. Whenever a token is generated only previously produced tokens are taken into consideration. In contrast, for problems such as sequence classification, bidirectional attention, which takes both past and future tokens into consideration, has been shown to perform much better. We propose to make the sequence generation process bidirectional by employing special placeholder tokens. Treated as a node in a fully connected graph, a placeholder token can take past and future tokens into consideration when generating the actual output token. We verify the effectiveness of our approach experimentally on two conversational tasks where the proposed bidirectional model outperforms competitive baselines by a large margin.