AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

Uncovering the Latent Potential of Deep Intermediate Representations

Arnesh Batra, Arush Gumber, Aniket Khandelwal, Jashn Khemani, Anubha Gupta

ICML 2026 · Architectures and efficiency › RNNs and normalization · approach: Layer-wise analysis of language models · 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
add GeoReg to LOES during fine-tuning (ablation)LOES (k=3) · Stanford Cars (trainable DINOv2-S/14)top-1 accuracy (%): 81.9 -> 82.7Improved
add GeoReg to LOES when the encoder is frozen (ablation)LOES · frozen encoderaccuracyNo clear effect
add GeoReg to trainable BERT-base fine-tuning (ablation)BERT-base fine-tuning · TweetEval Emojivalidation accuracyImproved
use LOES layer selection instead of last layerCLIP-B/32 frozen · Stanford Carstest accuracy (%): 50.53 -> 59.95Improved
use LOES layer selection instead of last layerDeiT-B/16 frozen · Mini-ImageNettest accuracy (%): 98.63 -> 98.44Hurt
use LOES instead of last layerModernBERT-B · MTEB average (excl. Toxic)accuracy (%): 54.25 -> 64.89Improved
select layers with LOES instead of entropy, curvature or intrinsic-dimension selectionModernBERT-B · MTEB average (excl. Toxic)accuracy (%): 61.15 -> 64.89Improved
LOES selection instead of random 4 layers (ablation)ModernBERT-B · AM-Scenarioaccuracy (%): 77.47 -> 78.51Improved
LOES instead of greedy probe-accuracy selection with 5 epochs (ablation)ModernBERT-B · MTOPaccuracy (%): 94.55 -> 94.48No clear effect
increase k beyond 3-4 selected layers (ablation)LOES · MTOP (ModernBERT-base)accuracy (%): 0.02 point gain from k=3 to k=5No clear effect
increase k from 4 to 10 (ablation)LOES · MTOP (BERT-base)accuracy (%): 0.69 point gain for 2.5x parametersImproved
vary ridge regularizer lambda from 1e-5 to 1e-2 (ablation)LOES (k=4) · MTOP / AM-Scenario (ModernBERT-base)accuracy (%): MTOP 94.27 to 94.80No clear effect

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