Published:
How to efficiently get one-hot representation with pytorch? Here’s the solution to it.
======
import torch
import torch.nn as nn
import torch.nn.functonal as F
x = F.softmax(torch.rand(10, 5))
pred = torch.argmax(x, dim=1)
one_hot_pred = torch.zeros_like(x).scatter_(1, pred.unsqueeze(1), 1.)
Published:
Learning the basic algorithms is crucial for programmers. Here is a personal summary of details about how to solve algorithm problems in practice with examples from LeetCode. Please check out:
Published:
How to creating a customized colorbar with python? When creating heatmaps, we usually have
Published:
The data visualization projects are done using D3(Data Driven Document). 
Published:
This is the panorama recovered from 25 images taken from an UAV. I used Pixel Congealing to estimate rotation parameters. 
