AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

Robust Optimization for Multilingual Translation with Imbalanced Data

Xian Li, Hongyu Gong

NeurIPS 2021 · Architectures and efficiency › RNNs and normalization · approach: Neural scaling laws · 6 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
Use curvature-aware learned task weights instead of T=1 samplingMultilingual translation · TED relatedaverage BLEU (All): 20.0 -> 20.7Improved
Use curvature-aware learned task weights instead of T=1 samplingMultilingual translation · TED diverseaverage BLEU (Lo): 15.7 -> 16.1Improved
Use curvature-aware learned task weights instead of T=5 samplingMultilingual translation · WMTaverage BLEU (Lo): +2.2 BLEUImproved
Use curvature-aware learned task weights instead of T=5 samplingMultilingual translation · WMTaverage BLEU (All): 20.9 -> 23.5Improved
Use curvature-aware learned task weights instead of T=5 samplingMultilingual translation · OPUS-100average BLEU (Lo): +1.3 BLEUImproved
Remove LayerNorm parameters (ablation)TED diverse translation with T=1 baseline · TED diverseaverage BLEU (Lo): 22.8 -> 23.1Improved
Add CATS while removing LayerNorm parameters (ablation)TED diverse translation · TED diverseaverage BLEU (Lo): 23.1 -> 23.6Improved
Upsample low-resource data with T=5 rather than proportional samplingMultilingual translation · TED relatedaverage BLEU (All): 20.0 -> 19.9No clear effect
Upsample low-resource data with T=5 rather than proportional samplingMultilingual translation · TED diverseaverage BLEU (All): 23.5 -> 22.5Hurt

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