Model not learning

[SS8188]

Timestamp 7:50:00
PyTorch workflow section exercise
Issue:My model is not updating its parameters(weight & bias) to minimize the Loss when I run the training loop.(Note that I have created training & testing split using shuffling of indices instead of straight away slicing the dataset into 2 parts)
Notebook:https://colab.research.google.com/drive/19LTNh8uftVZ5V71I2HEpIZz3t9HtISND#scrollTo=NhScv2GWbUxL

I tried setting the random_seed value to the same number(42) everywhere,The learning rate is set to 0.01(mentioned in the exercise),I tried changing it but I am not able to fix this(I did the saving & loading part of the exercise mindlessly even when the model isn't working).
I believe all the data is of the same d-type with requires_grad=True and on the same device as the model...
I am facing the same issue in the next few topics

[LuluW8071]

Looks like you gave wrong input data weight and bias. I ran your code by changing the input parameters it ran fine.

weight = 0.3
bias = 0.7

start = 0    
end = 1
step = 0.02

# or you can do 
# start, end, step = 1, 0, -0.02

X=torch.arange(start,end,step).unsqueeze(dim=1)
y=(weight*X)+bias

[SS8188]

I am facing the same error for multiclass classification
Time Stamp:13:30:00
As Instructed by Daniel,I tried coding it myself & fixed the errors my way.I used float32 dtype for the labels & predictions Tensors.But my loss is not going down
Note how each of the parameters in the model are None.
Changing the LR or epochs is not helping.
I didn't understand what changes you made to the linear regression notebook.I don't think weight & bias were the culprits for loss not decreasing.
Notebook link:https://colab.research.google.com/drive/19LTNh8uftVZ5V71I2HEpIZz3t9HtISND?usp=sharing