AI Paper Atlas

Research map / Large language models / Language, safety and interpretability

Learning to Compress: Unlocking the Potential of Large Language Models for Text Representation

Yeqin Zhang, Yizheng Zhao, Chen Hu, Binxing Jiao, Daxin Jiang, Ruihang Miao, Cam-Tu Nguyen

AAAI 2026 · Large language models › Language, safety and interpretability · approach: LLM unlearning and benchmarks · 1 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 knowledge-distillation (KL) loss in continuation pretext instead of NLL (ablation)LLM2Comp pretraining · MTEB 14-task avgaverage score: 46.22 -> 52.49Improved
use continuation+KD instead of reconstruction pretext (ablation)LLM2Comp pretraining · MTEB 14-task avgaverage score: 35.79 -> 52.49Improved
replace MNTP pretext with CTKD compressionLLM2Vec · MTEB 14-task avg (pretraining only)average score: 46.87 -> 52.49Improved
KD loss instead of NLL: training stability (ablation)LLM2Comp pretraining · MTEB 5-run stdstd: 5.32 -> 1.37Improved
increase memory tokens from 8 to 16 (ablation)LLM2CompKL · MTEB retrievalretrieval scoreHurt
vary memory tokens within 1 to 8 (ablation)LLM2CompKL · MTEB non-retrieval tasksscoreNo clear effect
increase pretraining samples from 32k to 128k (ablation)LLM2CompNLL · MTEB 14-task avgaverage score: 42.95 -> 48.38Improved
pool only de-duplicated effective tokens (ablation)LLM2CompNLL (128k) · MTEB 14-task avgaverage score: 48.38 -> 51.09Improved
add unsupervised contrastive learning after CTKD pretraining (ablation)LLM2CompKL · MTEB 14-task avgaverage score: 52.49 -> 58.51Improved
add supervised contrastive learning after UCL (ablation)LLM2CompKL · MTEB 14-task avgaverage score: 58.51 -> 66.78Improved
use CTKD pretext instead of MNTP before UCLLLM2Comp vs LLM2Vec · MTEB 14-task avgaverage score: 57.82 -> 58.51Improved
use CTKD pretext instead of MNTP before SCL, with 0.36M vs 1.16M samplesLLM2Comp vs LLM2Vec · MTEB 14-task avgaverage score: 66.11 -> 66.78Improved

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