mirror
/
learn-python
kopia lustrzana https://github.com/animator/learn-python
1
0
Forkuj 0
Kod Zgłoszenia Packages Projekty Wydania Wiki Aktywność

Update PyTorch_Fundamentals.md

pull/1157/head
Krishna Kaushik 2024-06-13 09:24:26 +05:30 zatwierdzone przez GitHub
rodzic 03e9338a91
commit 71bf180dd5
Nie znaleziono w bazie danych klucza dla tego podpisu
ID klucza GPG: B5690EEEBB952194
1 zmienionych plików z 64 dodań i 107 usunięć
Pokaż statystyki Ściągnij plik aktualizacji Ściągnij plik porównania Expand all files Collapse all files

171
contrib/machine-learning/PyTorch_Fundamentals.md
Wyświetl plik

@ -5,9 +5,9 @@
# Import pytorch in our codespace # Import pytorch in our codespace
import torch import torch
print(torch.__version__) print(torch.__version__)
```
2.3.0+cu121 output -> 2.3.0+cu121
```
2.3.0 is the pytorch version and 121 is the cuda version 2.3.0 is the pytorch version and 121 is the cuda version
@ -21,27 +21,26 @@ Now you have already seen how to create a tensor in pytorch. In this notebook i
# Scalar tensor ( a zero dimension tensor) # Scalar tensor ( a zero dimension tensor)
scalar = torch.tensor(7) scalar = torch.tensor(7)
print(scalar) print(scalar)
```
tensor(7) output -> tensor(7)
```
```python ```python
# Check the dimension of the above tensor # Check the dimension of the above tensor
print(scalar.ndim) print(scalar.ndim)
output -> 0
``` ```
0
```python ```python
# To retrieve the number from the tensor we use `item()` # To retrieve the number from the tensor we use `item()`
print(scalar.item()) print(scalar.item())
```
7 output -> 7
```
@ -49,27 +48,25 @@ print(scalar.item())
# Vector (It is a single dimension tensor but contain many numbers) # Vector (It is a single dimension tensor but contain many numbers)
vector = torch.tensor([1,2]) vector = torch.tensor([1,2])
print(vector) print(vector)
output -> tensor([1, 2])
``` ```
tensor([1, 2])
```python ```python
# Check the dimensions # Check the dimensions
print(vector.ndim) print(vector.ndim)
output -> 1
``` ```
1
```python ```python
# Check the shape of the vector # Check the shape of the vector
print(vector.shape) print(vector.shape)
```
torch.Size([2]) output -> torch.Size([2])
```
The above returns torch.Size([2]) which means our vector has a shape of [2]. This is because of the two elements we placed inside the square brackets ([1,2]) The above returns torch.Size([2]) which means our vector has a shape of [2]. This is because of the two elements we placed inside the square brackets ([1,2])
@ -85,31 +82,28 @@ You can tell the number of dimensions a tensor in PyTorch has by the number of s
MATRIX = torch.tensor([[1,2], MATRIX = torch.tensor([[1,2],
[4,5]]) [4,5]])
print(MATRIX) print(MATRIX)
```
tensor([[1, 2], output -> tensor([[1, 2],
[4, 5]]) [4, 5]])
```
There are two brackets so it must be 2 dimensions , lets check There are two brackets so it must be 2 dimensions , lets check
```python ```python
print(MATRIX.ndim) print(MATRIX.ndim)
output -> 2
``` ```
2
```python ```python
# Shape # Shape
print(MATRIX.shape) print(MATRIX.shape)
output -> torch.Size([2, 2])
``` ```
torch.Size([2, 2])
It means MATRIX has 2 rows and 2 columns. It means MATRIX has 2 rows and 2 columns.
@ -119,31 +113,28 @@ TENSOR = torch.tensor([[[1,2,3],
[4,5,6], [4,5,6],
[7,8,9]]]) [7,8,9]]])
print(TENSOR) print(TENSOR)
```
tensor([[[1, 2, 3], output -> tensor([[[1, 2, 3],
[4, 5, 6], [4, 5, 6],
[7, 8, 9]]]) [7, 8, 9]]])
```
```python ```python
# Let's check the dimensions # Let's check the dimensions
print(TENSOR.ndim) print(TENSOR.ndim)
output -> 3
``` ```
3
```python ```python
# shape? # shape?
print(TENSOR.shape) print(TENSOR.shape)
output -> torch.Size([1, 3, 3])
``` ```
torch.Size([1, 3, 3])
The dimensions go outer to inner. The dimensions go outer to inner.
That means there's 1 dimension of 3 by 3. That means there's 1 dimension of 3 by 3.
@ -164,40 +155,36 @@ We can create them using `torch.rand()` and passing in the `size` parameter.
# creating a random tensor of size (3,4) # creating a random tensor of size (3,4)
rand_tensor = torch.rand(size = (3,4)) rand_tensor = torch.rand(size = (3,4))
print(rand_tensor) print(rand_tensor)
```
tensor([[0.7462, 0.4950, 0.7851, 0.8277], output -> tensor([[0.7462, 0.4950, 0.7851, 0.8277],
[0.6112, 0.5159, 0.1728, 0.6847], [0.6112, 0.5159, 0.1728, 0.6847],
[0.4472, 0.1612, 0.6481, 0.3236]]) [0.4472, 0.1612, 0.6481, 0.3236]])
```
```python ```python
# Check the dimensions # Check the dimensions
print(rand_tensor.ndim) print(rand_tensor.ndim)
output -> 2
``` ```
2
```python ```python
# Shape # Shape
print(rand_tensor.shape) print(rand_tensor.shape)
output -> torch.Size([3, 4])
``` ```
torch.Size([3, 4])
```python ```python
# datatype # datatype
print(rand_tensor.dtype) print(rand_tensor.dtype)
output -> torch.float32
``` ```
torch.float32
### Zeros and ones ### Zeros and ones
Here we will create a tensor of any shape filled with zeros and ones Here we will create a tensor of any shape filled with zeros and ones
@ -207,24 +194,22 @@ Here we will create a tensor of any shape filled with zeros and ones
# Create a tensor of all zeros # Create a tensor of all zeros
zeros = torch.zeros(size = (3,4)) zeros = torch.zeros(size = (3,4))
print(zeros) print(zeros)
```
tensor([[0., 0., 0., 0.], output -> tensor([[0., 0., 0., 0.],
[0., 0., 0., 0.], [0., 0., 0., 0.],
[0., 0., 0., 0.]]) [0., 0., 0., 0.]])
```
```python ```python
# create a tensor of ones # create a tensor of ones
ones = torch.ones(size = (3,4)) ones = torch.ones(size = (3,4))
print(ones) print(ones)
```
tensor([[1., 1., 1., 1.], output -> tensor([[1., 1., 1., 1.],
[1., 1., 1., 1.], [1., 1., 1., 1.],
[1., 1., 1., 1.]]) [1., 1., 1., 1.]])
```
### Create a tensor having range of numbers ### Create a tensor having range of numbers
@ -244,11 +229,10 @@ zero_to_ten = torch.arange(start = 0,
end = 10, end = 10,
step = 1) step = 1)
print(zero_to_ten) print(zero_to_ten)
output -> tensor([0, 1, 2, 3, 4, 5, 6, 7, 8, 9])
``` ```
tensor([0, 1, 2, 3, 4, 5, 6, 7, 8, 9])
# 2. Manipulating tensors (tensor operations) # 2. Manipulating tensors (tensor operations)
The operations are : The operations are :
@ -265,22 +249,20 @@ The operations are :
```python ```python
tensor = torch.tensor([1,2,3]) tensor = torch.tensor([1,2,3])
print(tensor+10) print(tensor+10)
output -> tensor([11, 12, 13])
``` ```
tensor([11, 12, 13])
We have add 10 to each tensor element. We have add 10 to each tensor element.
```python ```python
tensor1 = torch.tensor([4,5,6]) tensor1 = torch.tensor([4,5,6])
print(tensor+tensor1) print(tensor+tensor1)
output -> tensor([5, 7, 9])
``` ```
tensor([5, 7, 9])
We have added two tensors , remember that addition takes place element wise. We have added two tensors , remember that addition takes place element wise.
### 2. Subtraction ### 2. Subtraction
@ -288,32 +270,29 @@ We have added two tensors , remember that addition takes place element wise.
```python ```python
print(tensor-8) print(tensor-8)
output -> tensor([-7, -6, -5])
``` ```
tensor([-7, -6, -5])
We've subtracted 8 from the above tensor. We've subtracted 8 from the above tensor.
```python ```python
print(tensor-tensor1) print(tensor-tensor1)
output -> tensor([-3, -3, -3])
``` ```
tensor([-3, -3, -3])
### 3. Multiplication ### 3. Multiplication
```python ```python
# Multiply the tensor with 10 (element wise) # Multiply the tensor with 10 (element wise)
print(tensor*10) print(tensor*10)
output -> tensor([10, 20, 30])
``` ```
tensor([10, 20, 30])
Each element of tensor gets multiplied by 10. Each element of tensor gets multiplied by 10.
Note: Note:
@ -324,19 +303,17 @@ PyTorch also has a bunch of built-in functions like `torch.mul()` (short for mul
```python ```python
# let's see them # let's see them
print(torch.add(tensor,10)) print(torch.add(tensor,10))
output -> tensor([11, 12, 13])
``` ```
tensor([11, 12, 13])
```python ```python
print(torch.mul(tensor,10)) print(torch.mul(tensor,10))
output -> tensor([10, 20, 30])
``` ```
tensor([10, 20, 30])
### Matrix multiplication (is all you need) ### Matrix multiplication (is all you need)
One of the most common operations in machine learning and deep learning algorithms (like neural networks) is matrix multiplication. One of the most common operations in machine learning and deep learning algorithms (like neural networks) is matrix multiplication.
@ -366,61 +343,41 @@ tensor2 = torch.tensor([[[1,1,1],
[3,3,3]]]) [3,3,3]]])
print(tensor1) , print(tensor2) print(tensor1) , print(tensor2)
```
tensor([[[1, 2, 3], output1 -> tensor([[[1, 2, 3],
[4, 5, 6], [4, 5, 6],
[7, 8, 9]]]) [7, 8, 9]]])
tensor([[[1, 1, 1], output2 -> tensor([[[1, 1, 1],
[2, 2, 2], [2, 2, 2],
[3, 3, 3]]]) [3, 3, 3]]])
```
(None, None)
```python ```python
# let's check the shape # let's check the shape
print(tensor1.shape) , print(tensor2.shape) print(tensor1.shape) , print(tensor2.shape)
output1 -> torch.Size([1, 3, 3])
output2 ->torch.Size([1, 3, 3])
``` ```
torch.Size([1, 3, 3])
torch.Size([1, 3, 3])
(None, None)
```python ```python
# Matrix multiplication # Matrix multiplication
print(torch.matmul(tensor1, tensor2)) print(torch.matmul(tensor1, tensor2))
```
tensor([[[14, 14, 14], output -> tensor([[[14, 14, 14],
[32, 32, 32], [32, 32, 32],
[50, 50, 50]]]) [50, 50, 50]]])
```
```python ```python
# Can also use the "@" symbol for matrix multiplication, though not recommended # Can also use the "@" symbol for matrix multiplication, though not recommended
print(tensor1 @ tensor2) print(tensor1 @ tensor2)
```
tensor([[[14, 14, 14], output -> tensor([[[14, 14, 14],
[32, 32, 32], [32, 32, 32],
[50, 50, 50]]]) [50, 50, 50]]])
```
Note: Note: