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/

Zendy Plus

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

Zendy Tools

Zendy Open

The Transformer-Kernel (TK) model has demonstrated strong rerankingperformance on the TREC Deep Learning benchmark -- and can be considered to bean efficient (but slightly less effective) alternative to otherTransformer-based architectures that employ (i) large-scale pretraining (hightraining cost), (ii) joint encoding of query and document (high inferencecost), and (iii) larger number of Transformer layers (both high training andhigh inference costs). Since, a variant of the TK model -- called TKL -- hasbeen developed that incorporates local self-attention to efficiently processlonger input sequences in the context of document ranking. In this work, wepropose a novel Conformer layer as an alternative approach to scale TK tolonger input sequences. Furthermore, we incorporate query term independence andexplicit term matching to extend the model to the full retrieval setting. Webenchmark our models under the strictly blind evaluation setting of the TREC2020 Deep Learning track and find that our proposed architecture changes leadto improved retrieval quality over TKL. Our best model also outperforms allnon-neural runs ("trad") and two-thirds of the pretrained Transformer-basedruns ("nnlm") on NDCG@10.

Improving Transformer-Kernel Ranking Model Using Conformer and Query Term Independence