Neural machine learning models can successfully model language that is similar to their training distribution, but they are highly susceptible to degradation under distribution shift, which occurs in many practical applications when processing out-of-domain (OOD) text. This has been attributed to "shortcut learning": relying on weak correlations over arbitrary large contexts. We propose a method based on OOD detection with Random Network Distillation to allow an autoregressive language model to automatically disregard OOD context during inference, smoothly transitioning towards a less expressive but more robust model as the data becomes more OOD while retaining its full context capability when operating in-distribution. We apply our method to a GRU architecture, demonstrating improvements on multiple language modeling (LM) datasets.
翻译:神经机学习模型可以成功地模拟与其培训分布相类似的语言,但非常容易在分布转移中退化,这在处理外部域(OOD)文本时会在许多实际应用中发生。这归因于“短期学习 ” : 依赖任意大环境的薄弱关联性。 我们提出了一个基于随机网络蒸馏检测OOOD的方法,允许自动递减语言模型在推断时自动忽略OOD环境,随着数据在分布过程中的运行,平稳地过渡到一个表达性较弱但更健全的模型,同时保留其全部背景能力。我们将我们的方法应用于一个GRU结构,展示多语言模型数据集的改进。