Part of Advances in Neural Information Processing Systems 32 (NeurIPS 2019)
Ravi Kumar, Manish Purohit, Zoya Svitkina, Erik Vee, Joshua Wang
When training complex neural networks, memory usage can be an important bottleneck. The question of when to rematerialize, i.e., to recompute intermediate values rather than retaining them in memory, becomes critical to achieving the best time and space efficiency. In this work we consider the rematerialization problem and devise efficient algorithms that use structural characterizations of computation graphs---treewidth and pathwidth---to obtain provably efficient rematerialization schedules. Our experiments demonstrate the performance of these algorithms on many common deep learning models.