Top-P Masking for Cross Language Information Retrieval

Fuente: arXiv
Gespeichert in:
Bibliographische Detailangaben
Hauptverfasser: Casale, Joseph, Silverschotz, Andrew, DeSimone, Joseph
Format: Preprint
Veröffentlicht: 2025
Schlagworte:
Online-Zugang:
Tags: Tag hinzufügen
Keine Tags, Fügen Sie den ersten Tag hinzu!
_version_ 1866914108230598656
author Casale, Joseph
Silverschotz, Andrew
DeSimone, Joseph
author_facet Casale, Joseph
Silverschotz, Andrew
DeSimone, Joseph
contents Top-K masking schemes have been proposed as a method to promote sparse representations in Information Retrieval (IR) tasks, as a simple alternative to Floating Point Operations per Second (FLOPS) regularization. Algorithms such as Bilingual Lexical and Document Expansion Model (BLADE), adopt this approach as a post-processing stage. We propose using Top-P Dynamic Masking similar to Nucleus Sampling in Large Language Models, and demonstrate better performance than Top-K masking. Specifically, we evaluate our methods in the domain of Cross Language Information Retrieval (CLIR)
format Preprint
id arxiv_https___arxiv_org_abs_2510_19758
institution arXiv
publishDate 2025
record_format arxiv
spellingShingle Top-P Masking for Cross Language Information Retrieval
Casale, Joseph
Silverschotz, Andrew
DeSimone, Joseph
Information Retrieval
Top-K masking schemes have been proposed as a method to promote sparse representations in Information Retrieval (IR) tasks, as a simple alternative to Floating Point Operations per Second (FLOPS) regularization. Algorithms such as Bilingual Lexical and Document Expansion Model (BLADE), adopt this approach as a post-processing stage. We propose using Top-P Dynamic Masking similar to Nucleus Sampling in Large Language Models, and demonstrate better performance than Top-K masking. Specifically, we evaluate our methods in the domain of Cross Language Information Retrieval (CLIR)
title Top-P Masking for Cross Language Information Retrieval
topic Information Retrieval
url https://arxiv.org/abs/2510.19758