extensive experiments
In AI research, extensive experiments refer to rigorous testing and validation of models or algorithms across diverse datasets, conditions, and configurations to ensure robustness, generalizability, and reliability of the claims made regarding their performance.
- Compiler-R1: Towards Agentic Compiler Auto-tuning with Reinforcement Learning
- DataRater: Meta-Learned Dataset Curation
- Deno-IF: Unsupervised Noisy Visible and Infrared Image Fusion Method
- FedGPS: Statistical Rectification Against Data Heterogeneity in Federated Learning
- Investigating Hallucinations of Time Series Foundation Models through Signal Subspace Analysis
- LLM at Network Edge: A Layer-wise Efficient Federated Fine-tuning Approach
- Latent Space Factorization in LoRA
- Learning from Videos for 3D World: Enhancing MLLMs with 3D Vision Geometry Priors
- Long-Tailed Recognition via Information-Preservable Two-Stage Learning
- Mesh Interpolation Graph Network for Dynamic and Spatially Irregular Global Weather Forecasting
- Mitigating Forgetting in LLM Fine-Tuning via Low-Perplexity Token Learning
- ObCLIP: Oblivious CLoud-Device Hybrid Image Generation with Privacy Preservation
- On the Stability of Graph Convolutional Neural Networks: A Probabilistic Perspective
- Permissioned LLMs: Enforcing Access Control in Large Language Models
- Personalized Federated Conformal Prediction with Localization
- Prompt-Guided Alignment with Information Bottleneck Makes Image Compression Also a Restorer
- Provable Scaling Laws for the Test-Time Compute of Large Language Models
- Retrv-R1: A Reasoning-Driven MLLM Framework for Universal and Efficient Multimodal Retrieval
- RidgeLoRA: Matrix Ridge Enhanced Low-Rank Adaptation of Large Language Models
- SkyLadder: Better and Faster Pretraining via Context Window Scheduling
- TANDEM: Bi-Level Data Mixture Optimization with Twin Networks
- Token-Level Self-Play with Importance-Aware Guidance for Large Language Models
- UniZyme: A Unified Protein Cleavage Site Predictor Enhanced with Enzyme Active-Site Knowledge
- Unveiling Extraneous Sampling Bias with Data Missing-Not-At-Random
- WorldMem: Long-term Consistent World Simulation with Memory