DataLoader

Batched iteration over datasets with optional shuffling.

DataLoader(x, y, batch_size=32, shuffle=True, drop_last=False)

from minitorch import DataLoader
import numpy as np

X = np.random.randn(1000, 784).astype(np.float32)
Y = np.random.randn(1000, 10).astype(np.float32)

loader = DataLoader(X, Y, batch_size=64, shuffle=True)

for x_batch, y_batch in loader:
    # x_batch: (64, 784), y_batch: (64, 10)
    pass

print(len(loader))  # number of batches

Parameters

• x - numpy array of inputs
• y - numpy array of targets
• batch_size - samples per batch (default 32)
• shuffle - shuffle indices each epoch (default True)
• drop_last - if True, drop the last batch when it is smaller than batch_size (default False)