parameter count
Parameter count indicates the total number of adjustable parameters within a machine learning model. A higher parameter count typically implies a more complex model capable of capturing more intricate patterns but may also increase risk of overfitting and require more computational resources.
- A Implies B: Circuit Analysis in LLMs for Propositional Logical Reasoning
- Analyzing the Power of Chain of Thought through Memorization Capabilities
- CrossSpectra: Exploiting Cross-Layer Smoothness for Parameter-Efficient Fine-Tuning
- Permutation Equivariant Neural Controlled Differential Equations for Dynamic Graph Representation Learning
- Spark Transformer: Reactivating Sparsity in Transformer FFN and Attention
- The $\varphi$ Curve: The Shape of Generalization through the Lens of Norm-based Capacity Control
- The Cost of Robustness: Tighter Bounds on Parameter Complexity for Robust Memorization in ReLU Nets