| S.No |
Topic |
Sub-Topics |
| 1 | PyTorch | What is PyTorch, PyTorch vs TensorFlow, Installation, CPU vs GPU, PyTorch Ecosystem |
| 2 | PyTorch Tensors | Tensor creation, Tensor types, Shape & dtype, Indexing & slicing, Tensor operations |
| 3 | Tensor Mathematics | Arithmetic ops, Matrix multiplication, Broadcasting, Reduction ops, In-place ops |
| 4 | Autograd Basics | Computational graph, requires_grad, backward(), grad attribute, detach() |
| 5 | PyTorch CUDA | CUDA tensors, GPU availability, Moving data to GPU, Performance tips, Multi-GPU intro |
| 6 | Neural Network Module | nn.Module, Parameters, Forward method, Model structure, Model summary |
| 7 | Loss Functions | MSELoss, CrossEntropyLoss, NLLLoss, Custom loss, Reduction methods |
| 8 | Optimizers | SGD, Adam, RMSprop, Learning rate, Weight decay |
| 9 | Training Loop | Forward pass, Loss computation, Backpropagation, Optimizer step, Zero gradients |
| 10 | Dataset & DataLoader | Custom Dataset, DataLoader, Batching, Shuffling, num_workers |
| 11 | Data Preprocessing | Normalization, Scaling, Augmentation, Transform pipeline, torchvision.transforms |
| 12 | Linear Regression | Model definition, Loss & optimizer, Training, Evaluation, Visualization |
| 13 | Logistic Regression | Binary classification, Sigmoid, BCE loss, Accuracy metric, Training loop |
| 14 | Feedforward Neural Networks | Dense layers, Activation functions, Hidden layers, Overfitting, Regularization |
| 15 | Activation Functions | ReLU, Sigmoid, Tanh, Softmax, LeakyReLU |
| 16 | Model Evaluation | Train vs eval mode, Validation loop, Metrics, Confusion matrix, Accuracy & loss |
| 17 | Saving & Loading Models | state_dict, torch.save, torch.load, Checkpoints, Best practices |
| 18 | Convolutional Neural Networks | Conv layers, Pooling, Feature maps, CNN architecture, Use cases |
| 19 | Image Classification | torchvision datasets, CNN training, Data augmentation, Evaluation, Inference |
| 20 | Transfer Learning | Pretrained models, Freezing layers, Fine-tuning, ResNet, VGG |
| 21 | Recurrent Neural Networks | RNN basics, Sequence data, Hidden state, Vanishing gradients, Use cases |
| 22 | LSTM & GRU | LSTM architecture, GRU basics, Sequence modeling, Time series, Text data |
| 23 | Text Processing | Tokenization, Embeddings, torchtext, Padding, Sequence batching |
| 24 | Custom Layers & Models | Custom nn.Module, Parameter handling, Reusability, Debugging, Testing |
| 25 | Regularization Techniques | Dropout, BatchNorm, Weight decay, Early stopping, Data augmentation |
| 26 | Learning Rate Scheduling | StepLR, ReduceLROnPlateau, CosineAnnealing, Warm-up, Best practices |
| 27 | Mixed Precision Training | AMP basics, torch.cuda.amp, Performance gain, Memory optimization, Stability |
| 28 | Model Deployment Basics | TorchScript, Tracing, Scripting, Inference optimization, Exporting models |
| 29 | Debugging & Profiling | Gradient issues, NaNs, torch.autograd profiler, Performance tuning, Common errors |
| 30 | PyTorch Best Practices | Code structure, Reproducibility, Experiment tracking, Scaling training, Next steps |
