CIFAR-10
A widely used dataset for image classification tasks consisting of 60,000 32×32 color images in 10 classes.
Category: Concepts
TinyImageNet
A smaller version of the ImageNet dataset, containing 200 classes with 500 training images per class, used for image classification tasks.
Sparsity-Aware Retraining
A retraining process that takes into account the sparsity of the model, focusing on recovering performance without reactivating pruned connections.
One-Shot Pruning
A pruning strategy that eliminates weights in a single pass rather than through iterative cycles of training and pruning.
Unstructured Pruning
A method of reducing the size of neural networks by removing individual weights based on certain criteria, rather than entire neurons or layers.
Importance Score Estimation
A technique used to evaluate the significance of each parameter in a model to determine which ones can be pruned without significantly affecting performance.
Knowledge Distillation
A process where a smaller model is trained to replicate the behavior of a larger, pre-trained model, effectively transferring knowledge.
Predictive Accuracy
A measure of how accurately a model predicts outcomes compared to actual results.
Nonlinear Recurrent Language Models
A type of language model that utilizes recurrent neural networks with nonlinear activation functions to process sequences of data.
Technical Variables
Quantitative measures derived from historical price and volume data used to forecast future price movements.