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

We study the evolution of latent space in fine-tuned NLP models. Differentfrom the commonly used probing-framework, we opt for an unsupervised method toanalyze representations. More specifically, we discover latent concepts in therepresentational space using hierarchical clustering. We then use an alignmentfunction to gauge the similarity between the latent space of a pre-trainedmodel and its fine-tuned version. We use traditional linguistic concepts tofacilitate our understanding and also study how the model space transformstowards task-specific information. We perform a thorough analysis, comparingpre-trained and fine-tuned models across three models and three downstreamtasks. The notable findings of our work are: i) the latent space of the higherlayers evolve towards task-specific concepts, ii) whereas the lower layersretain generic concepts acquired in the pre-trained model, iii) we discoveredthat some concepts in the higher layers acquire polarity towards the outputclass, and iv) that these concepts can be used for generating adversarialtriggers.

On the Transformation of Latent Space in Fine-Tuned NLP Models