How to get mini-batches in pytorch in a clean and efficient way?

Question

I was trying to do a simple thing which was train a linear model with Stochastic Gradient Descent (SGD) using torch:

import numpy as np

import torch
from torch.         


        

Answer 1

An alternative could be using pd.DataFrame.sample

train = pd.read_csv(TrainSetPath)
test = pd.read_csv(TestSetPath)

# use df.sample() to shuffle the data frame 
train = train.sample(frac=1)
test = test.sample(frac=1)

for i in range(epochs):
        for j in range(batch_per_epoch):
            train_batch = train.sample(n=BatchSize, axis='index',replace=True)
            y_train = train_batch['Target']
            X_train = train_batch.drop(['Target'], axis=1)
            
            # convert data frames to tensors and send them to GPU (if used)
            X_train = torch.tensor(np.mat(X_train)).float().to(device)
            y_train = torch.tensor(np.mat(y_train)).float().to(device)