Modified AlexNet-based PyTorch model for CIFAR-10 classification with modern training techniques (label smoothing, dropout, LR scheduling). Achieves ~88.63% test accuracy and includes a Gradio-based Hugging Face deployment.
learning-rate-scheduling modified-alexnet-cifar10 small-image-cnn-classification cnn-regularization-techniques label-smoothing-training dropout-generalization gradient-clipping-stability data-augmentation-cnn image-resize-crop-preprocessing adaptive-average-pooling batch-normalization-effect early-stopping-training sgd-optimizer-analysis feature-extraction-cnns confusion-matrix-evaluation overfitting-control-methods pytorch-training-pipeline cifar10-benchmark-model model-deployment-gradio real-time-image-inference
-
Updated
May 1, 2026 - Jupyter Notebook