AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

Investigating Data Pruning for Pretraining Biological Foundation Models at Scale

Yifan Wu, Jiyue Jiang, Xichen Ye, Yiqi Wang, Chang Zhou, Yitao Xu, Jiayang Chen, He Hu, Weizhong Zhang, Cheng Jin, Jiao Yuan, Yu Li

AAAI 2026 · Architectures and efficiency › RNNs and normalization · approach: Neural scaling laws · 2 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 CCI coreset (0.2M) instead of random 0.2MRNA-FM pretraining · CRISPR On-TargetSpearman corr (%): 26.67 -> 32.90Improved
use CCI coreset instead of random 0.2MRNA-FM pretraining · RNA Type Classificationaccuracy (%): 82.15 -> 82.88Improved
use Top I coreset instead of random 0.2MRNA-FM pretraining · RNA distance mapR2 (%): 76.71 -> 79.25Improved
use Top I coreset instead of random 0.2MRNA-FM pretraining · RNA contact map (L:1.0L)top precision (%): 94.18 -> 94.36Improved
use influence coresets of 0.2M instead of random 2M (10x more data)RNA-FM pretraining · RNA Type Classificationaccuracy (%): 82.21 -> 82.88Improved
add fine-tuning on random subset before influence estimation (CCI) (ablation)influence-guided selection · CRISPR On-TargetSpearman corr (%): 26.96 -> 32.90Improved
add fine-tuning before influence estimation (Top I) (ablation)influence-guided selection · RNA Type Classificationaccuracy (%): 81.07 -> 82.51Improved
use CCI coreset instead of random 0.2MRNA-FM pretraining · RBP-RNA interactionAUC (%): 75.97 -> 76.04No clear effect
use Top I vs CCI coresetRNA-FM pretraining · bpRNA secondary structureF1 (%): 57.05 vs 56.36No clear effect
use CCI coreset instead of random 0.2MESM-C pretraining · Binary localizationaccuracy (%): 73.64 -> 79.25Improved
use CCI coreset instead of random 0.2MESM-C pretraining · Secondary structure (protein)accuracy (%): 66.18 -> 71.48Improved
use influence coreset 0.2M vs full pretraining dataESM-C pretraining · Binary localizationaccuracy (%): 91.63 -> 79.25Hurt

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