Hi All, I read several post and I already know that there are issues with using jupyter notebook with multi threading. Since those posts were some years ago, I am wondering if this issue was fixed. Below is my machine and a snippet of my code (nested for loop). It runs indefinitely without giving any errors if I put inside Dataloader num_workers > 0 (default is zero) 
Is CUDA supported by this system? True
CUDA version: 12.1
Using device: cuda
ID of current CUDA device: 0
Name of current CUDA device: NVIDIA GeForce RTX 3060
NVIDIA GeForce RTX 3060
It is working on a single thread and is extremely slow and I need to speed the execution up.
Below a snippet of my code: X_PM10_train_pt, Y_PM10_train_pt are cuda tensors, but nothing change if they move on cpu.
train_loader = DataLoader(TensorDataset(X_PM10_train_pt, Y_PM10_train_pt), batch_size=params['batch_size'],shuffle=False)
.....
for epoch in range(num_epochs):
model_mlp2_pm10.train()
for inputs, labels in train_loader:
optimizer.zero_grad() #zero the gradients after updating, before starting a new optimization
outputs = model_mlp2_pm10(inputs) #predict = forward pass with our model
loss = criterion(outputs, labels) #loss
loss.backward() #calculate gradients
optimizer.step()#update weights
# Valutazione del modello sul set di validazione
Please give me any hint how to overcome this issue