English (United Kingdom)

https://curated-unify.zendy.io/wp-json/zendy-region/v1/featured_content/oa?rat=en

https://curated-unify.zendy.io/wp-json/zendy-region/v1/highlighted_journal/

Global: Zendy Plus annual

Presents the access of premium content as premium feature

Premium Content

Presents the keyphrase highlighting as premium feature

Keyphrase Highlighting

Presents the summarisation as premium feature

Summarisation

Insights

Presents the pdf analysis as premium feature

PDF Analysis

Presents the zaia usage as premium feature

ZAIA

Global: Zendy Plus monthly

Global: Zendy Tools annual

Global: Zendy Tools monthly

Global: Zendy Free Plan

Many of the most performant deep learning models today in fields likelanguage and image understanding are fine-tuned models that contain billions ofparameters. In anticipation of workloads that involve serving many of suchlarge models to handle different tasks, we develop Computron, a system thatuses memory swapping to serve multiple distributed models on a shared GPUcluster. Computron implements a model parallel swapping design that takesadvantage of the aggregate CPU-GPU link bandwidth of a cluster to speed upmodel parameter transfers. This design makes swapping large models feasible andcan improve resource utilization. We demonstrate that Computron successfullyparallelizes model swapping on multiple GPUs, and we test it on randomizedworkloads to show how it can tolerate real world variability factors likeburstiness and skewed request rates. Computron's source code is available athttps://github.com/dlzou/computron.

Computron: Serving Distributed Deep Learning Models with Model Parallel Swapping