← Optimization & Theory
Deep Learning › Optimization & Theory ›

Neural Network Optimization

902 directly classified papers

Papers per year

Papers

Spectral Scaling Laws in Language Models: emphHow Effectively Do Feed-Forward Networks Use Their Latent Space? EMNLP 2025
Stepwise Reasoning Checkpoint Analysis: A Test Time Scaling Method to Enhance LLMs’ Reasoning EMNLP 2025
A Proactive Reliability Metric for Detecting Failures in Language Model Training EMNLP 2025
Exploring smaller batch sizes for a high-performing BabyLM model architecture EMNLP 2025
Not All Parameters Matter: Masking Diffusion Models for Enhancing Generation Ability CVPR 2025
Lookahead Q-Cache: Achieving More Consistent KV Cache Eviction via Pseudo Query EMNLP 2025
Language Models Grow Less Humanlike beyond Phase Transition ACL 2025
Low-Rank Interconnected Adaptation across Layers ACL 2025
Segment-Based Attention Masking for GPTs ACL 2025
Value Residual Learning ACL 2025
Training Dynamics Underlying Language Model Scaling Laws: Loss Deceleration and Zero-Sum Learning ACL 2025
Efficient Many-Shot In-Context Learning with Dynamic Block-Sparse Attention ACL 2025
Neural Parameter Search for Slimmer Fine-Tuned Models and Better Transfer ACL 2025
A Training-Free Length Extrapolation Approach for LLMs: Greedy Attention Logit Interpolation EMNLP 2025
FREE: Fast and Robust Vision Language Models with Early Exits ACL 2025
IG-Pruning: Input-Guided Block Pruning for Large Language Models EMNLP 2025
Sketch-of-Thought: Efficient LLM Reasoning with Adaptive Cognitive-Inspired Sketching EMNLP 2025
Variance Sensitivity Induces Attention Entropy Collapse and Instability in Transformers EMNLP 2025
Where to show Demos in Your Prompt: A Positional Bias of In-Context Learning EMNLP 2025
GRASP: Replace Redundant Layers with Adaptive Singular Parameters for Efficient Model Compression EMNLP 2025
The Strawberry Problem: Emergence of Character-level Understanding in Tokenized Language Models EMNLP 2025
LightThinker: Thinking Step-by-Step Compression EMNLP 2025
Revealing the Deceptiveness of Knowledge Editing: A Mechanistic Analysis of Superficial Editing ACL 2025
NeuroAda: Activating Each Neuron’s Potential for Parameter-Efficient Fine-Tuning EMNLP 2025
EcoTune: Token-Efficient Multi-Fidelity Hyperparameter Optimization for Large Language Model Inference EMNLP 2025