论文标题
具有自适应组成模块的持续序列产生
Continual Sequence Generation with Adaptive Compositional Modules
论文作者
论文摘要
当需要将模型快速调整到新任务的情况下而又不忘记旧任务的知识时，持续学习对于实际部署至关重要。连续序列生成上的现有工作要么总是重用现有参数学习新任务，这很容易受到灾难性遗忘的遗忘，或者盲目地为每个新任务添加新参数，这可以防止在相似任务之间进行知识共享。为了获得两全其美的最佳，在这项工作中，我们提出了使用自适应组成模块的持续序列生成，以适应性地添加变压器体系结构中的模块，并为新任务组成新旧模块。我们还结合了伪经验重播，以促进这些共享模块中的知识转移。各种生成任务序列的实验结果表明，我们的框架可以根据任务相似性自适应地添加模块或重用模块，从性能和参数效率方面优于最先进的基线。我们在https://github.com/gt-salt/adaptive-compositional-modules上公开代码。
Continual learning is essential for real-world deployment when there is a need to quickly adapt the model to new tasks without forgetting knowledge of old tasks. Existing work on continual sequence generation either always reuses existing parameters to learn new tasks, which is vulnerable to catastrophic forgetting on dissimilar tasks, or blindly adds new parameters for every new task, which could prevent knowledge sharing between similar tasks. To get the best of both worlds, in this work, we propose continual sequence generation with adaptive compositional modules to adaptively add modules in transformer architectures and compose both old and new modules for new tasks. We also incorporate pseudo experience replay to facilitate knowledge transfer in those shared modules. Experiment results on various sequences of generation tasks show that our framework can adaptively add modules or reuse modules based on task similarity, outperforming state-of-the-art baselines in terms of both performance and parameter efficiency. We make our code public at https://github.com/GT-SALT/Adaptive-Compositional-Modules.
