Pruning Pretrained Encoders with a Multitask Objective
This addresses the problem of computational expense in multitask settings for NLP practitioners, offering an incremental improvement over existing pruning methods.
The paper tackles the challenge of using large pretrained language models for multiple downstream tasks by pruning a single encoder with a multitask objective, finding that it outperforms single-task ensembles on average and is competitive individually, with effectiveness for low-resource tasks.
The sizes of pretrained language models make them challenging and expensive to use when there are multiple desired downstream tasks. In this work, we adopt recent strategies for model pruning during finetuning to explore the question of whether it is possible to prune a single encoder so that it can be used for multiple tasks. We allocate a fixed parameter budget and compare pruning a single model with a multitask objective against the best ensemble of single-task models. We find that under two pruning strategies (element-wise and rank pruning), the approach with the multitask objective outperforms training models separately when averaged across all tasks, and it is competitive on each individual one. Additional analysis finds that using a multitask objective during pruning can also be an effective method for reducing model sizes for low-resource tasks.