Deep learning + pytorch self study notes
background I am a senior now and have been admitted by Tsinghua University. As the initial direction of graduate students will involve in-depth learning, the tutor said that they will use Python for defect detection, involving precision measurement, advanced manufacturing and other related fielUTF-8...
Posted by perfume on Sun, 16 May 2021 08:50:56 +0930
mmclassification image classification model training and testing
1. Model training 1.1 using the pre training model Because the imagenet is very large, I choose to use cifar for model training first. The cifar model is as follows Links: links: https://github.com/open-mmlab/mmclassification/blob/master/docs/model_zoo.md Download link: link: https://download.UTF-8...
Posted by marconi8 on Thu, 20 May 2021 04:55:58 +0930
PyTorch Foundation: Dataset, Dataloader, torch vision of data loading and preprocessing
PyTorch Foundation: data loading and preprocessing PyTorch passed torch.utils.data encapsulates the common data loading, which can easily realize multi thread data pre reading and batch loading. also torchvision has implemented common image data sets in advance, including CIFAR-10, ImageNet, COUTF-8...
Posted by sri2002 on Mon, 24 May 2021 03:30:24 +0930
PyTorch torch.nn from the beginning
On torch.nn PyTorch provides well-designed modules and classes such as torch.nn, torch.optim, Dataset and DataLoader neural networks. To make the most of their capabilities and customize them for your problems, you need to really understand what they're doing. To establish this understanding, wUTF-8...
Posted by ijmccoy on Mon, 24 May 2021 04:40:24 +0930
Resolve CUDA error: no kernel image is available for execution on the device
1. Computer Configuration GPU 3080 arithmetic 8.6 CUDA 11.1 CUDNN 8.2.0 conda 4.9.2 python 3.8.5 2. Description of the problem First in pytroch website Use pip command according to computer configuration pip3 install torch==1.8.1+cu111 torchvision==0.9.1+cu111 torchaudio==0.8.1 -f https://downlUTF-8...
Posted by nuxy on Tue, 25 May 2021 01:39:52 +0930
Multiplication in pytorch
There are many functions in pytorch for multiplication operation. What are the different functions? torch.mul multiply is an alias for mul, consistent with mul usage torch.mul(input, other, *, out=None) There are two main uses: (1) Multiplication of a number by each element in a tensor a = torcUTF-8...
Posted by Platinumweaver on Sun, 30 May 2021 01:50:25 +0930
Target detection in Python: YOLO-v5
Yolo and ssd are two popular algorithms in target detection. This article records how to use yolo-v5 for target detection. Because of the long space, it is divided into two parts. This is the first part, recording how to run through yolo-v5 code, and training and testing the coco128 data set. TUTF-8...
Posted by vbzoom.com on Sun, 30 May 2021 06:44:09 +0930
AI playing games (DL basis) pytorch framework learning notes
Processing data If we want to process 10000 pictures of data, we can't transfer 10000 pictures at a time, so the video memory will burst, and we can't transfer only one picture at a time, so the efficiency is too low. So we need to send all the pictures in batches. For example, there are 100 bUTF-8...
Posted by BlueMeteor on Tue, 01 Jun 2021 02:39:11 +0930
Artificial intelligence deep learning Python and neural network
1, Basic operation of Python Note: detailed description in the code comments, in order to ensure the integrity of the code, the following code will have repeated parts, the focus will be described outside the title 1.numpy establishes the matrix import numpy as np # Building all 1 matrices of 4UTF-8...
Posted by Dedix on Thu, 03 Jun 2021 04:09:20 +0930
The difference between conventional convolution and causal convolution
0. Preface Today, when learning causal convolution (in WaveNet and TCN), I encountered some problems when looking at the source code, such as why Chomp1d? According to my understanding, if the number of padding is ( k − 1 ) × d / 2 (k-1)\times d/2 (k−1)×d/2( k k k is the convolution kernel sizUTF-8...
Posted by Mr_jmm on Tue, 08 Jun 2021 09:49:09 +0930