Day #223 - Pytorch - Sessions (9-12)

	#Session #1
	import torch
	import numpy as np
	data = np.array([1,2,2])
	#Class constructor
	#Create Copy
	t1 = torch.Tensor(data)
	#Factor function
	#Return object
	#Create Copy
	t2 = torch.tensor(data)
	#Share Memory
	t3 = torch.as_tensor(data)
	t4 = torch.from_numpy(data)
	print(t1)
	print(t1.dtype)
	#type inferences
	print(torch.tensor(data))
	print(torch.tensor(data,dtype=torch.float64))
	#Session #2
	#Tensor Operations
	#Reshape
	# 3 x 4 tensor
	t = torch.tensor([[1,1,1,1],[2,2,2,2],[3,3,3,3]],dtype=torch.float32)
	print(t.size())
	print(t.shape)
	#Rank - Check length of shape
	print(len(t.shape))
	#ElementWise
	print(torch.tensor(t.shape).prod())
	#Number of elements
	print(t.numel())
	#Change tensor by squeezing and unsqueezing them
	#Removes all aexs with length of 1
	#Build a flatten function
	print(t.reshape(1,12).squeeze())
	print(t.reshape(1,12).squeeze().shape)
	#Flatten function to get 1D Array
	def flatten(t):
	#Any shape -1
	#Figure out value based on elements
	#Reshape will find 12
	t = t.reshape(1,-1)
	t = t.squeeze()
	return t
	print(flatten(t))
	print(t.reshape(1,12))
	#Ways to concatenate
	t1 = torch.tensor([[1,1,1,1],[1,1,1,1],[1,1,1,1],[1,1,1,1]])
	t2 = torch.tensor([[2,2,2,2],[2,2,2,2],[2,2,2,2],[2,2,2,2]])
	t3 = torch.tensor([[3,3,3,3],[3,3,3,3],[3,3,3,3],[3,3,3,3]])
	t = torch.stack((t1,t2,t3))
	print(t.shape)
	#Batch 3, 4 x 4 Images
	#torch.Size([3, 4, 4])
	#Add Axes Color Channels
	t= t.reshape(3,1,4,4)
	print(t)
	#First Image
	print(t[0])
	#First Channel
	print(t[0][0])
	#First row of First Color Channel
	print(t[0][0][0])
	#Flatten Images in the batch
	#Flatten Image tensor within batch tensor
	print(t.reshape(1,-1)[0])
	print(t.reshape(-1))
	print(t.view(t.numel()))
	#Built in method
	print(t.flatten())
	#We need individual predictions
	#Flatten with Batch Axes
	print(t.flatten(start_dim=1).shape)
	print(t.flatten(start_dim=1))
	#Elementwise operation
	t1 = torch.tensor([[1,2],[3,4]])
	t2 = torch.tensor([[9,8],[7,6]])
	#First Axes
	print(t1[0])
	#Second Axes
	print(t1[0][0])
	#Perform element-wise operation
	print(t1+t2)
	print(t1*t2)
	print(t1-t2)
	print(t1/t2)
	#Lower Rank Transferred to Higher Rank
	t1 = torch.tensor([[1,1],[1,1]],dtype=torch.float32)
	t2 = torch.tensor([2,4],dtype=torch.float32)
	print(t1+t2)
	#Reduction Ops
	print(t1.sum())
	print(t1.mean())
	print(t1.std())
	print(t1.sum(dim=1))
	print(t1.max(dim=1))

view raw PytorchLessons.py hosted with ❤ by GitHub

Link - 13th Session

Happy Mastering DL!!!