model parameters
The internal variables of a model that are learned from the training data, influencing the model's predictions. Parameters are adjusted during training to minimize the prediction error and improve accuracy.
- ALINE: Joint Amortization for Bayesian Inference and Active Data Acquisition
- Channel Matters: Estimating Channel Influence for Multivariate Time Series
- Continual Knowledge Adaptation for Reinforcement Learning
- DIsoN: Decentralized Isolation Networks for Out-of-Distribution Detection in Medical Imaging
- Deep Learning with Plausible Deniability
- Dynamical Properties of Tokens in Self-Attention and Effects of Positional Encoding
- Exploiting Vocabulary Frequency Imbalance in Language Model Pre-training
- Generalized Top-k Mallows Model for Ranked Choices
- Gradient Descent as Loss Landscape Navigation: a Normative Framework for Deriving Learning Rules
- MARS: A Malignity-Aware Backdoor Defense in Federated Learning
- Machine Unlearning Doesn't Do What You Think: Lessons for Generative AI Policy and Research
- Mamba Only Glances Once (MOGO): A Lightweight Framework for Efficient Video Action Detection
- On the Effect of Negative Gradient in Group Relative Deep Reinforcement Optimization
- Pay Attention to Small Weights
- Polar Sparsity: High Throughput Batched LLM Inferencing with Scalable Contextual Sparsity
- Prior Forgetting and In-Context Overfitting
- Projection-based Lyapunov method for fully heterogeneous weakly-coupled MDPs
- Statistical Inference under Performativity
- Transformer Key-Value Memories Are Nearly as Interpretable as Sparse Autoencoders