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

Hyper-parameter optimization is crucial for pushing the accuracy of a deeplearning model to its limits. A hyper-parameter optimization job, referred toas a study, involves numerous trials of training a model using differenttraining knobs, and therefore is very computation-heavy, typically taking hoursand days to finish. We observe that trials issued from hyper-parameteroptimization algorithms often share common hyper-parameter sequence prefixes.Based on this observation, we propose Hippo, a hyper-parameter optimizationsystem that removes redundancy in the training process to reduce the overallamount of computation significantly. Instead of executing each trialindependently as in existing hyper-parameter optimization systems, Hippo breaksdown the hyper-parameter sequences into stages and merges common stages to forma tree of stages (called a stage-tree), then executes a stage once per tree ona distributed GPU server environment. Hippo is applicable to not only singlestudies, but multi-study scenarios as well, where multiple studies of the samemodel and search space can be formulated as trees of stages. Evaluations showthat Hippo's stage-based execution strategy outperforms trial-based methodssuch as Ray Tune for several models and hyper-parameter optimizationalgorithms, reducing GPU-hours and end-to-end training time significantly.

Hippo: Taming Hyper-parameter Optimization of Deep Learning with Stage Trees