Exploring the Role of Task Transferability in Large-Scale Multi-Task Learning
This work addresses the efficiency of multi-task learning for practitioners by showing that task selection can reduce computational overhead without sacrificing performance.
The study investigated how the scale and relatedness of tasks affect multi-task representation learning, finding that while larger task sets generally improve performance, using a smaller set of related tasks can achieve similar results at lower computational cost when target tasks are known.
Recent work has found that multi-task training with a large number of diverse tasks can uniformly improve downstream performance on unseen target tasks. In contrast, literature on task transferability has established that the choice of intermediate tasks can heavily affect downstream task performance. In this work, we aim to disentangle the effect of scale and relatedness of tasks in multi-task representation learning. We find that, on average, increasing the scale of multi-task learning, in terms of the number of tasks, indeed results in better learned representations than smaller multi-task setups. However, if the target tasks are known ahead of time, then training on a smaller set of related tasks is competitive to the large-scale multi-task training at a reduced computational cost.