parameter space
The parameter space in machine learning describes the multidimensional space defined by all possible values that the model parameters can take, crucial for understanding model behavior and conducting optimization.
- Amortized Variational Transdimensional Inference
- Efficient Multi-modal Large Language Models via Progressive Consistency Distillation
- HM3: Hierarchical Multi-Objective Model Merging for Pretrained Models
- Learning (Approximately) Equivariant Networks via Constrained Optimization
- Learning (Approximately) Equivariant Networks via Constrained Optimization
- Long-tailed Recognition with Model Rebalancing
- On Linear Mode Connectivity of Mixture-of-Experts Architectures
- Sharper Convergence Rates for Nonconvex Optimisation via Reduction Mappings
- Understanding Adam Requires Better Rotation Dependent Assumptions