AI Paper Atlas

Research map / Architectures and efficiency / Attention and transformers

Separating Representation from Reconstruction Enables Scalable Text Encoders

Megi Dervishi, Mathurin VIDEAU, Yann LeCun

ICML 2026 · Architectures and efficiency › Attention and transformers · approach: Language model pretraining and induction · 0 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
Scale BERT model size under frozen evaluationBERT (flat MLM) · frozen GLUE/MTEB(eng, v2)frozen probe score: largest BERTs fall below the smallest model (10x smaller)Hurt
Separate representation from reconstruction with a cross-attention predictor (CrossBERT) (ablation)BERT 239M · GLUE linear probeaverage score: 67.4 -> 73.8Improved
Use CrossBERT architecture (ablation)BERT 239M · MTEB(eng, v1) frozen backbone contrastive finetuningaverage score: 44.3 -> 53.7Improved
Use CrossBERT architecture (ablation)BERT 239M · GLUE full finetuningaverage score: 86.4 -> 86.6No clear effect
Use CrossBERT architecture (ablation)BERT 239M · MTEB(eng, v1) full contrastive finetuningaverage score: 54.1 -> 54.5Improved
Use CrossBERT architecture (ablation)BERT 239M · MTEB(eng, v1) frozen backbone RetrievalRetrieval score: 19.2 -> 40.7Improved
Increase masking ratio from 20% to 50% in CrossBERT (ablation)CrossBERT · frozen GLUE linear probeaverage score: 73.8 -> 73.1No clear effect
Increase masking ratio from 50% to 65% in CrossBERT (ablation)CrossBERT · frozen GLUE linear probeaverage score: 73.1 -> 72.4Hurt
Add Complementary Masking Strategy (CMS) (ablation)CrossBERT · sample efficiency (Appendix B)data needed for same performance: requires half the training dataImproved
Probe earlier layers instead of final layer of BERT (ablation)BERT 239M · frozen GLUE linear probeaverage score: 67.4 -> 70.2Improved
Use CrossBERT's masked-token dropping in the encoderCrossBERT · training throughput on H100tokens per second: 123k -> 207kImproved
Warm-start contrastive adapter with the pretrained predictor (ablation)CrossBERT frozen-backbone finetuning · MTEBMTEB score: approx +2%Improved

Earlier methods this paper compares with

Named baselines from earlier accepted papers in this paper's results tables: better than 0 of 1 on most shared results.

Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.

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