Pytorch Torch Cumsum
# PyTorch torch.cumsum Function
* * PyTorch torch Reference Manual](#)
`torch.cumsum` is a PyTorch function for calculating cumulative sums. It returns the cumulative sum along a specified dimension, representing the sum of all elements from the start to the current position.
### Function Definition
torch.cumsum(input, dim, dtype=None)
* * *
## Example Usage
## Example
import torch
# Calculate cumulative sum
x = torch.tensor([1,2,3,4,5])
result = torch.cumsum(x, dim=0)
print("Input:", x)
print("Cumulative sum:", result)
# Output: tensor([1, 3, 6, 10, 15])
# Explanation: 1, 1+2=3, 1+2+3=6, 1+2+3+4=10, 1+2+3+4+5=15
# 2D tensor
x = torch.tensor([[1,2,3],[4,5,6]], dtype=torch.float32)
# Column-wise cumulative sum
result_col = torch.cumsum(x, dim=0)
print("nColumn-wise cumulative sum:")
print(result_col)
# tensor([[ 1, 2, 3],
# [ 5, 7, 9]])
# Row-wise cumulative sum
result_row = torch.cumsum(x, dim=1)
print("nRow-wise cumulative sum:")
print(result_row)
# tensor([[ 1, 3, 6],
# [ 4, 9, 15]])
* * PyTorch torch Reference Manual](#)
YouTip