performance enhancement
Improvements made to an AI system that increase its effectiveness, efficiency, or accuracy in performing tasks.
- Accelerated Sampling from Masked Diffusion Models via Entropy Bounded Unmasking
- AceReason-Nemotron: Advancing Math and Code Reasoning through Reinforcement Learning
- Activation-Informed Merging of Large Language Models
- AdaDetectGPT: Adaptive Detection of LLM-Generated Text with Statistical Guarantees
- Advancing Expert Specialization for Better MoE
- An Effective Levelling Paradigm for Unlabeled Scenarios
- Broken Tokens? Your Language Model can Secretly Handle Non-Canonical Tokenizations
- Continual Optimization with Symmetry Teleportation for Multi-Task Learning
- DeltaPhi: Physical States Residual Learning for Neural Operators in Data-Limited PDE Solving
- EAGLE-3: Scaling up Inference Acceleration of Large Language Models via Training-Time Test
- Epistemic Uncertainty Estimation in Regression Ensemble Models with Pairwise Epistemic Estimators
- Exploiting the Asymmetric Uncertainty Structure of Pre-trained VLMs on the Unit Hypersphere
- FairNet: Dynamic Fairness Correction without Performance Loss via Contrastive Conditional LoRA
- From Self-Check to Consensus: Bayesian Strategic Decoding in Large Language Models
- Geometry-Aware Collaborative Multi-Solutions Optimizer for Model Fine-Tuning with Parameter Efficiency
- Hawaii: Hierarchical Visual Knowledge Transfer for Efficient Vision-Language Models
- High-order Interactions Modeling for Interpretable Multi-Agent Q-Learning
- Highlighting What Matters: Promptable Embeddings for Attribute-Focused Image Retrieval
- INST-IT: Boosting Instance Understanding via Explicit Visual Prompt Instruction Tuning
- Improving Deep Learning for Accelerated MRI With Data Filtering
- Improving Diffusion-based Inverse Algorithms under Few-Step Constraint via Linear Extrapolation
- Jury-and-Judge Chain-of-Thought for Uncovering Toxic Data in 3D Visual Grounding
- Learning to Better Search with Language Models via Guided Reinforced Self-Training
- Low-Rank Head Avatar Personalization with Registers
- Mitigating Reward Over-optimization in Direct Alignment Algorithms with Importance Sampling
- Modality-Aware SAM: Sharpness-Aware-Minimization Driven Gradient Modulation for Harmonized Multimodal Learning
- OmniBench: Towards The Future of Universal Omni-Language Models
- Order-Level Attention Similarity Across Language Models: A Latent Commonality
- PolyJuice Makes It Real: Black-Box, Universal Red Teaming for Synthetic Image Detectors
- Puzzles: Unbounded Video-Depth Augmentation for Scalable End-to-End 3D Reconstruction
- QiMeng-MuPa: Mutual-Supervised Learning for Sequential-to-Parallel Code Translation
- Reward Reasoning Models
- SAVVY: Spatial Awareness via Audio-Visual LLMs through Seeing and Hearing
- SAVVY: Spatial Awareness via Audio-Visual LLMs through Seeing and Hearing
- SGCD: Stain-Guided CycleDiffusion for Unsupervised Domain Adaptation of Histopathology Image Classification
- SPACE: SPike-Aware Consistency Enhancement for Test-Time Adaptation in Spiking Neural Networks
- SiriuS: Self-improving Multi-agent Systems via Bootstrapped Reasoning
- Spectral Conditioning of Attention Improves Transformer Performance
- Spik-NeRF: Spiking Neural Networks for Neural Radiance Fields
- Tackling Feature-Classifier Mismatch in Federated Learning via Prompt-Driven Feature Transformation
- The Curse of Multi-Modalities: Evaluating Hallucinations of Large Multimodal Models across Language, Visual, and Audio
- Thinking in Character: Advancing Role-Playing Agents with Role-Aware Reasoning
- TreeSynth: Synthesizing Diverse Data from Scratch via Tree-Guided Subspace Partitioning
- Universal Cross-Tokenizer Distillation via Approximate Likelihood Matching
- Vicinity-Guided Discriminative Latent Diffusion for Privacy-Preserving Domain Adaptation
- Vision Transformers Don't Need Trained Registers
- metaTextGrad: Automatically optimizing language model optimizers