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 Tools annual

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 Tools monthly

Global: Zendy Plus monthly

Presents the access of premium content as premium feature

Premium Content

Global: Zendy Plus annual

Global: Zendy Free Plan

Training large neural networks is known to be time-consuming, with thelearning duration taking days or even weeks. To address this problem,large-batch optimization was introduced. This approach demonstrated thatscaling mini-batch sizes with appropriate learning rate adjustments can speedup the training process by orders of magnitude. While long training time wasnot typically a major issue for model-free deep offline RL algorithms, recentlyintroduced Q-ensemble methods achieving state-of-the-art performance made thisissue more relevant, notably extending the training duration. In this work, wedemonstrate how this class of methods can benefit from large-batchoptimization, which is commonly overlooked by the deep offline RL community. Weshow that scaling the mini-batch size and naively adjusting the learning rateallows for (1) a reduced size of the Q-ensemble, (2) stronger penalization ofout-of-distribution actions, and (3) improved convergence time, effectivelyshortening training duration by 3-4x times on average.

Q-Ensemble for Offline RL: Don't Scale the Ensemble, Scale the Batch Size