trade-offs
The compromises made when optimizing different objectives or performance metrics in AI systems. For instance, increasing model accuracy may result in longer training times or require more computational resources.
- Balancing Gradient and Hessian Queries in Non-Convex Optimization
- ConStellaration: A dataset of QI-like stellarator plasma boundaries and optimization benchmarks
- Depth-Width Tradeoffs for Transformers on Graph Tasks
- Dynamic Semantic-Aware Correlation Modeling for UAV Tracking
- Embracing Contradiction: Theoretical Inconsistency Will Not Impede the Road of Building Responsible AI Systems
- Emergent Risk Awareness in Rational Agents under Resource Constraints
- Joint‑Embedding vs Reconstruction: Provable Benefits of Latent Space Prediction for Self‑Supervised Learning
- LT-Soups: Bridging Head and Tail Classes via Subsampled Model Soups
- MoE-CAP: Benchmarking Cost, Accuracy and Performance of Sparse Mixture-of-Experts Systems
- Preference Optimization on Pareto Sets: On a Theory of Multi-Objective Optimization
- Principled Fine-tuning of LLMs from User-Edits: A Medley of Preference, Supervision, and Reward
- SAFEPATH: Preventing Harmful Reasoning in Chain-of-Thought via Early Alignment
- Tractable Multinomial Logit Contextual Bandits with Non-Linear Utilities
- Transfer Learning for Benign Overfitting in High-Dimensional Linear Regression