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

The remarkable success of large transformer-based models such as BERT,RoBERTa and XLNet in many NLP tasks comes with a large increase in monetary andenvironmental cost due to their high computational load and energy consumption.In order to reduce this computational load in inference time, we presentTangoBERT, a cascaded model architecture in which instances are first processedby an efficient but less accurate first tier model, and only part of thoseinstances are additionally processed by a less efficient but more accuratesecond tier model. The decision of whether to apply the second tier model isbased on a confidence score produced by the first tier model. Our simple methodhas several appealing practical advantages compared to standard cascadingapproaches based on multi-layered transformer models. First, it enables higherspeedup gains (average lower latency). Second, it takes advantage of batch sizeoptimization for cascading, which increases the relative inference costreductions. We report TangoBERT inference CPU speedup on four textclassification GLUE tasks and on one reading comprehension task. Experimentalresults show that TangoBERT outperforms efficient early exit baseline models;on the the SST-2 task, it achieves an accuracy of 93.9% with a CPU speedup of8.2x.

TangoBERT: Reducing Inference Cost by using Cascaded Architecture