How justinnitoi/digit_neural_net works
This small Python program loads and preprocesses MNIST, trains a 784–20–10 sigmoid network with per-example backpropagation, and prints epoch accuracy. The excerpts show the loader, dataset access, and training/reporting logic; the user-triggered entry point is not documented, so that relationship is marked unverified.
Main entry points: nn.py Train Network · data.py Load MNIST · mnist.npz MNIST Archive
Components 5
MNIST Data
- Load MNIST
data.py - MNIST Archive
data/mnist.npz
Network Runtime
- Train Network
nn.py - Epoch Accuracy
nn.py
Other components
- User
Connections 4
- User → Train Network: runs
- Train Network → Load MNIST: loads data
nn.py - Load MNIST → MNIST Archive: reads
data.py - Train Network → Epoch Accuracy: prints
nn.py