Curricular Transfer Learning for Sentence Encoded Tasks
This work addresses a specific issue in NLP for conversational AI, offering an incremental improvement over existing methods.
The paper tackles the problem of performance degradation when fine-tuning language models on target tasks with distributional drift from the source task, such as in conversational environments, by proposing a curriculum of pre-training steps guided by data hacking and grammar analysis, achieving considerable improvement on the MultiWoZ task compared to other pre-training approaches.
Fine-tuning language models in a downstream task is the standard approach for many state-of-the-art methodologies in the field of NLP. However, when the distribution between the source task and target task drifts, \textit{e.g.}, conversational environments, these gains tend to be diminished. This article proposes a sequence of pre-training steps (a curriculum) guided by "data hacking" and grammar analysis that allows further gradual adaptation between pre-training distributions. In our experiments, we acquire a considerable improvement from our method compared to other known pre-training approaches for the MultiWoZ task.