Sometimes we just need to go back to basics, this repo tracks my progress while I revisit core PyTorch concepts.

Projects at a glance

01 — LinearRegression.ipynb

• Simple linear regression in PyTorch
• Trained with SGD + MSELoss
• Clean training loop and loss curves

02 — Classification

• Binary (make_circles): MLP trained with BCEWithLogitsLoss (apply sigmoid at inference)
• Multiclass (make_blobs): MLP trained with CrossEntropyLoss (softmax at inference)