AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

AG-REPA: Causal Layer Selection for Representation Alignment in Audio Flow Matching

Pengfei ZHANG, Tianxin Xie, Minghao Yang, Li Liu

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
align FoG-A-selected top-3 layers instead of the best single fixed layer (REPA @ L8)AG-REPA on DiT flow matching · Speech FAD, Config BFAD: 1.58 -> 1.29Improved
add attribution-weighted sparse top-3 selection on top of Shallow REPA (L1-L3) (ablation)AG-REPA · Speech FAD, Config BFAD: 1.36 -> 1.29Improved
align deep representation-rich layers (L20-L22) (ablation)REPA · Speech FAD, Config BFAD: 1.84 -> 1.79No clear effect
align random layers (L5, L14, L19) (ablation)REPA · Speech FAD, Config BFAD: 1.84 -> 1.75No clear effect
select layers by highest LASP score instead of random (ablation)REPA layer selection · Speech FAD, Config BFAD: 1.75 -> 1.68Improved
select layers by FoG-A instead of Gradient Norm (ablation)AG-REPA layer selection · Speech FAD / Audio FADFAD: 1.35/2.71 -> 1.29/2.56Improved
select layers by FoG-A instead of LASP (ablation)AG-REPA layer selection · Steps to reach Speech FAD 1.5convergence steps: 720k -> 220kImproved
apply AG-REPA to VoiceboxVoicebox · VoiceboxFAD: 1.20 -> 0.95Improved
apply AG-REPA to CosyVoiceCosyVoice · CosyVoiceFAD: 0.88 -> 0.72Improved
apply AG-REPA to F5-TTSF5-TTS · F5-TTSFAD: 1.45 -> 1.15Improved
re-run FoG-A per architecture instead of hard-coding shallow layers L1-L3 (ablation)AG-REPA vs Shallow REPA · F5-TTSFAD: 1.34 -> 1.15Improved
continuously refresh layer set S every epoch instead of static selection (ablation)AG-REPA · Speech/Audio FADFAD: 1.29/2.56 -> 1.28/2.55No 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