AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

MILCO: Learned Sparse Retrieval Across Languages via a Multilingual Connector

Thong Nguyen, Yibin Lei, Jia-Huei Ju, Eugene Yang, Andrew Yates

ICLR 2026 · Label-efficient and robust learning › Recognition and long-tail · approach: Multilingual language representation pretraining · 11 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add LexEcho source view to English-only view (ablation)MILCO · MIRACL devnDCG@10: 69.4 -> 72.3Improved
add LexEcho source view (ablation)MILCO · MLDRnDCG@10: 69.9 -> 74.4Improved
add Sparse Alignment Pretraining before contrastive training (ablation)MILCO · MIRACL devnDCG@10: 59.2 -> 72.3 (with SAP, distill, LexEcho; English-only 59.2 -> 69.4)Improved
add sparse contrastive training after alignment (ablation)MILCO · MIRACL devnDCG@10: 54.5 -> 70.1Improved
add KL distillation from cross-encoder instead of InfoNCE (ablation)MILCO · MIRACL devnDCG@10: 70.1 -> 72.3Improved
add multilingual connector (map to English lexicon) instead of full multilingual MLM head (ablation)MILCO · MIRACL devnDCG@10: 50.7 -> 59.2Improved
use sparse alignment instead of dense alignment (with distill)MILCO · MIRACL devnDCG@10: 67.9 -> 72.3Improved
use MILCO vs Qwen3-Embed-8B on MTEBv2MILCO · Multilingual MTEBv2nDCG@10 avg: 75.59 vs 66.83Hurt
post-hoc mass pruning p=10 (ablation)MILCO · MIRACL English SeismicnDCG@10: 56.4 -> 56.3No clear effect
post-hoc mass pruning p=60 (ablation)MILCO · MIRACL English SeismicnDCG@10: 56.4 -> 50.3Hurt
use MILCO shared English lexical space instead of source-language sparse modelsMILCO · MKQARecall@100: 45.3 -> 76.6Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy