AI Paper Atlas

Research map / Architectures and efficiency / RNNs and normalization

Bring Future Vision: Dynamic Computation Allocation Guided by Lightweight Feature Forecaster

Chao Han, Yijuan Liang, Zihao Xuan, Daokuan Wu, Wei Zhang, Xiaoyu Shen

ICML 2026 · Architectures and efficiency › RNNs and normalization · approach: Neural scaling laws

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 Lightweight Feature Forecaster (informed routing) instead of skipping (ablation)SkipGPT router, Llama3.1-8B · WikiText-2, 25% sparsityperplexity: 20.76 -> 9.43Improved
add Lightweight Feature Forecaster instead of skipping (ablation)SkipGPT router, Llama3.1-8B · WikiText-2, 40% sparsityperplexity: 53.94 -> 13.87Improved
add LFF (ablation)SkipGPT router, Llama3.1-8B · Reasoning average, 25% sparsityaverage accuracy (%): 51.23 -> 64.62Improved
add LFF (ablation)SkipGPT router, Llama3.1-8B · Reasoning average, 40% sparsityaverage accuracy (%): 46.75 -> 56.02Improved
add LFF, after LoRA fine-tuning (ablation)SkipGPT-LoRA · Reasoning average, 25% sparsityaverage accuracy (%): 60.43 -> 66.16Improved
add LFF, after LoRA fine-tuning (ablation)SkipGPT-LoRA · Reasoning average, 40% sparsityaverage accuracy (%): 58.85 -> 60.24Improved
add LFF with balanced attention/FFN sparsity (ablation)SkipGPT-balance · Balanced reasoningaverage accuracy (%): 58.42 -> 63.49Improved
add LFF with balanced attention/FFN sparsity (ablation)SkipGPT-balance · Balanced WikiText-2perplexity: 11.40 -> 9.49Improved
add LFF under KV cache reduction (ablation)SkipGPT · WikiText-2 PPL, 25% sparsity, KV reductionperplexity: 11.68 -> 10.21Improved
enforce KV cache reduction (mask skipped tokens from attention)LFF · WikiText-2 PPL, 25% sparsityperplexity: 8.93 -> 10.21Hurt
add a linear LFF to FFN units (ablation)FFN units · WikiText-2 PPL without fine-tuningperplexity: 36.50 -> 12.19Improved
add ReLU non-linearity in the LFF middle layer (ablation)linear FFN LFF · WikiText-2 PPL without fine-tuningperplexity: 12.19 -> 11.99No 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