Deep Learning with PyTorch
summary
Shows how to build and train models with PyTorch end to end: constructs a toy nine‑pixel classifier (learns W and bias by gradient descent), implements MNIST digit models using linear nets (~80% accuracy) and convolutional nets (approaches 99% accuracy), uses CUDA when available, Adam optimizer with small learning rate and batch size 128, binary/softmax cross‑entropy losses, and saves trained weights to ONNX for inference. Demonstrates loading the ONNX model into a .NET WinForms app to run predictions, proving the pipeline works. tensors track gradients ONNX runtime for .NET
AI-written summary · about
chapters (4)
related talks
59mA Friendly Introduction to Deep Learning for Computer Vision
01h 26mDeep Neural Networks with PyTorch
47mDeep Learning with PyTorch
01h 07mDeep Learning with TensorFlow
41mReally Deep Neural Networks with PyTorch
20mLearn You a PyTorch!
57mDeep Learning with CNTK and F#
01h 06mDeep Learning with TensorFlow