ResearchTrend.AI
  • Papers
  • Communities
  • Events
  • Blog
  • Pricing
Papers
Communities
Social Events
Terms and Conditions
Pricing
Parameter LabParameter LabTwitterGitHubLinkedInBlueskyYoutube

© 2025 ResearchTrend.AI, All rights reserved.

  1. Home
  2. Papers
  3. 2006.09616
11
93

Dynamic Tensor Rematerialization

17 June 2020
Marisa Kirisame
Steven Lyubomirsky
Altan Haan
Jennifer Brennan
Mike He
Jared Roesch
Tianqi Chen
Zachary Tatlock
ArXivPDFHTML
Abstract

Checkpointing enables the training of deep learning models under restricted memory budgets by freeing intermediate activations from memory and recomputing them on demand. Current checkpointing techniques statically plan these recomputations offline and assume static computation graphs. We demonstrate that a simple online algorithm can achieve comparable performance by introducing Dynamic Tensor Rematerialization (DTR), a greedy online algorithm for checkpointing that is extensible and general, is parameterized by eviction policy, and supports dynamic models. We prove that DTR can train an NNN-layer linear feedforward network on an Ω(N)\Omega(\sqrt{N})Ω(N​) memory budget with only O(N)\mathcal{O}(N)O(N) tensor operations. DTR closely matches the performance of optimal static checkpointing in simulated experiments. We incorporate a DTR prototype into PyTorch merely by interposing on tensor allocations and operator calls and collecting lightweight metadata on tensors.

View on arXiv
Comments on this paper