Deep Learning with PyTorch
summary
Builds a convolutional neural network in a Jupyter notebook using PyTorch, loads the CIFAR-10 dataset on Kaggle/Colab with GPU, trains for about ten epochs and shows training and validation loss curves, then uses a pre‑trained/quick model as an image classifier. Code-level details include data batching, relu and max‑pool layers, and gradient descent updates; final results top out around ~76% accuracy, with signs of validation plateauing and mild overfitting.
AI-written summary · about
chapters (17)
- 0:00Introduction
- 2:46Tips
- 9:35Why PyTorch
- 16:17Framework Comparison
- 19:00Building Neural Network
- 20:14Google Collab
- 21:34Goggles
- 22:59Notebook
- 24:03Data Preparation
- 29:39Data Splitting
- 30:39Batches
- 35:24Rectifier
- 36:44Spatial pooling
- 38:09Fully connected layer
- 39:24Convolutional network
- 42:14Train model
- 43:09Evaluation
related talks
01h 39mGoogle Colaboratory and Python: Deep Learning basics for CV tasks
01h 05mDeep Learning with PyTorch
01h 55mGoogle Colaboratory and Python: Deep Learning basics for CV tasks
01h 27mHands-On Computer Vision with PyTorch
01h 32mIntroduction to PyTorch
32mYou Can Do Deep Learning!
35mDeconstructing Deep Learning
01h 06mDeep Learning with TensorFlow