00. PyTorch Fundamentals - Exercises

[hashraf96]

Hi, I could only verify answers to the questions you (mrdbourke) have posted the answers to.

While going through the exercise for "00. PyTorch Fundamentals - Exercises", I had some troubles:

import torch
seed = 0
torch.manual_seed(seed)
tensor_random = torch.rand(7,7)

torch.manual_seed(seed)
tensor_random_second = torch.rand(1,7)
multiplied = torch.mm(tensor_random, tensor_random_second.T)
multiplied,multiplied.shape
print(f"Tensor A: ",tensor_random )
print(f"\nTensor B: ",tensor_random_second )
print(f"\nTensor B.T: ",tensor_random_second.T )
print(f"\nTensor A * Tensor B.T: ",multiplied )

Gives the output:

        [0.8964, 0.4556, 0.6323, 0.3489, 0.4017, 0.0223, 0.1689],
        [0.2939, 0.5185, 0.6977, 0.8000, 0.1610, 0.2823, 0.6816],
        [0.9152, 0.3971, 0.8742, 0.4194, 0.5529, 0.9527, 0.0362],
        [0.1852, 0.3734, 0.3051, 0.9320, 0.1759, 0.2698, 0.1507],
        [0.0317, 0.2081, 0.9298, 0.7231, 0.7423, 0.5263, 0.2437],
        [0.5846, 0.0332, 0.1387, 0.2422, 0.8155, 0.7932, 0.2783]])

Tensor B:  tensor([[0.4963, 0.7682, 0.0885, 0.1320, 0.3074, 0.6341, 0.4901]])

Tensor B.T:  tensor([[0.4963],
        [0.7682],
        [0.0885],
        [0.1320],
        [0.3074],
        [0.6341],
        [0.4901]])

Tensor A * Tensor B.T:  tensor([[1.5985],
        [1.1173],
        [1.2741],
        [1.6838],
        [0.8279],
        [1.0347],
        [1.2498]])

While the author to the question says the answer should be:

(tensor([[1.8542],
         [1.9611],
         [2.2884],
         [3.0481],
         [1.7067],
         [2.5290],
         [1.7989]]), torch.Size([7, 1]))

I don't know what I am doing wrong, can somebody help me with it?

[aronvandepol]

From what I can tell, only the values are different, which is likely because of a different manual seed passed onto torch.rand(). I believe @mrdbourke always uses 42, the answer to the universe.

[mrdbourke]

You're right I generally always use 42 as the random seed but the exercises say to try out various seed values: https://www.learnpytorch.io/00_pytorch_fundamentals/#exercises

[mrdbourke]

Hi @hashraf96,

Your code is so close!

The error is only minor.

The different results come from calling torch.manual_seed() more than once in a single code cell.

You generally only need to call it once per code cell.

See the changes I've made to your code:

import torch
seed = 0

# Only set seed once
torch.manual_seed(seed)

# Create random tensors
tensor_random = torch.rand(7, 7)
tensor_random_second = torch.rand(1, 7)

# Perform matrix multiplication
multiplied = torch.mm(tensor_random, tensor_random_second.T)
multiplied, multiplied.shape

# Print outputs
print(f"Tensor A: ",tensor_random )
print(f"\nTensor B: ",tensor_random_second )
print(f"\nTensor B.T: ",tensor_random_second.T )
print(f"\nTensor A * Tensor B.T: ",multiplied )