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 Tools annual

Global: Zendy Free Plan

Global: Zendy Tools monthly

Global: Zendy Plus monthly

Model fine-tuning and adaptation have become a common approach for modelspecialization for downstream tasks or domains. Fine-tuning the entire model ora subset of the parameters using light-weight adaptation has shown considerablesuccess across different specialization tasks. Fine-tuning a model for a largenumber of domains typically requires starting a new training job for everydomain posing scaling limitations. Once these models are trained, deployingthem also poses significant scalability challenges for inference for real-timeapplications. In this paper, building upon prior light-weight adaptationtechniques, we propose a modular framework that enables us to substantiallyimprove scalability for model training and inference. We introduce Submodelsthat can be quickly and dynamically loaded for on-the-fly inference. We alsopropose multiple approaches for training those Submodels in parallel using anembedding space in the same training job. We test our framework on an extremeuse-case which is speech model personalization for atypical speech, requiring aSubmodel for each user. We obtain 128x Submodel throughput with a fixedcomputation budget without a loss of accuracy. We also show that learning aspeaker-embedding space can scale further and reduce the amount ofpersonalization training data required per speaker.

A Scalable Model Specialization Framework for Training and Inference using Submodels and its Application to Speech Model Personalization