Encoder-Agnostic Adaptation for Conditional Language Generation
This work addresses a key challenge in making pretrained models effective for conditional generation tasks, offering a practical solution for researchers and practitioners in natural language processing.
The paper tackled the problem of adapting pretrained language models for conditional text generation without relying on task-specific encoders, proposing pseudo self attention to inject conditioning into self-attention. The method outperformed strong baselines across four diverse tasks, demonstrating data efficiency and coherent generation.
Large pretrained language models have changed the way researchers approach discriminative natural language understanding tasks, leading to the dominance of approaches that adapt a pretrained model for arbitrary downstream tasks. However it is an open-question how to use similar techniques for language generation. Early results in the encoder-agnostic setting have been mostly negative. In this work we explore methods for adapting a pretrained language model to arbitrary conditional input. We observe that pretrained transformer models are sensitive to large parameter changes during tuning. We therefore propose an adaptation that directly injects arbitrary conditioning into self attention, an approach we call pseudo self attention. Through experiments on four diverse conditional text generation tasks we show that this encoder-agnostic technique outperforms strong baselines, produces coherent generations, and is data efficient.