Parameter-Efficient Tuning Makes a Good Classification Head
This addresses a specific bottleneck in natural language understanding for researchers and practitioners, offering a stable performance gain but is incremental in nature.
The paper tackles the problem of ineffective classification head pretraining due to changes in backbone output during finetuning, finding that parameter-efficient tuning enables a good classification head that improves performance on 9 GLUE and SuperGLUE tasks.
In recent years, pretrained models revolutionized the paradigm of natural language understanding (NLU), where we append a randomly initialized classification head after the pretrained backbone, e.g. BERT, and finetune the whole model. As the pretrained backbone makes a major contribution to the improvement, we naturally expect a good pretrained classification head can also benefit the training. However, the final-layer output of the backbone, i.e. the input of the classification head, will change greatly during finetuning, making the usual head-only pretraining (LP-FT) ineffective. In this paper, we find that parameter-efficient tuning makes a good classification head, with which we can simply replace the randomly initialized heads for a stable performance gain. Our experiments demonstrate that the classification head jointly pretrained with parameter-efficient tuning consistently improves the performance on 9 tasks in GLUE and SuperGLUE.