Data

Tools

Indexes

Applications

Experiments

Open Source Science

my-pytorch-template

Code for training and testing artificial intelligence models

https://github.com/kwondoryoung/my-pytorch-template

Science Score: 44.0%

This score indicates how likely this project is to be science-related based on various indicators:

  • ✓
    CITATION.cff file
    Found CITATION.cff file
  • ✓
    codemeta.json file
    Found codemeta.json file
  • ✓
    .zenodo.json file
    Found .zenodo.json file
  • ○
    DOI references
  • ○
    Academic publication links
  • ○
    Academic email domains
  • ○
    Institutional organization owner
  • ○
    JOSS paper metadata
  • ○
    Scientific vocabulary similarity
    Low similarity (9.5%) to scientific vocabulary

Keywords

artificial-neural-networks computer-vision pytorch template-project
Last synced: 6 months ago · JSON representation ·

Repository

Code for training and testing artificial intelligence models

Basic Info
  • Host: GitHub
  • Owner: KwonDoRyoung
  • License: mit
  • Language: Python
  • Default Branch: main
  • Homepage:
  • Size: 146 KB
Statistics
  • Stars: 1
  • Watchers: 1
  • Forks: 0
  • Open Issues: 0
  • Releases: 0
Topics
artificial-neural-networks computer-vision pytorch template-project
Created over 2 years ago · Last pushed almost 2 years ago
Metadata Files
Readme License Citation

README.md

My Pytorch Template

Python(3.11.7) & Package version

  • pytorch==2.1.2 (CUDA==11.8)
  • torchvision==0.16.2
  • timm==0.9.12
  • opencv-python==4.9.0.80
  • pandas==2.2.0
  • matplotlib==3.8.2
  • Django==5.0.1
  • scikit-learn==1.4.0
  • scipy==1.12.0
  • SimpleITK==2.3.1
  • albumentations==1.3.1
  • torchmetrics==1.3.0.post0
  • wandb==0.16.2
  • torchcam==0.4.0
  • openpyxl==3.1.2
  • seaborn==0.13.2
  • tqdm==4.66.2

Dataset

  1. mnist : MNIST
  2. cat_dog : Cat & Dog

Tool

  1. binary Classification
  2. multiclass Classification
  3. binary Segmentation

Model

  1. Classification
    • timm 활용하여 호출함
  2. Segmentation
    • Backbone 필요한 segmentation 모델의 경우, timm 패키지 활용
    • 종류
      • FCN 32, 16, 8
      • DeepLabV3
      • DeepLabV3+
      • U-Net
      • U-Net++

Loss

  1. dice
    • Binary Dice Loss
    • Multiclass Dice Loss
  2. ce+dice
    • Binary Cross Entropy with Dice Loss
    • Multiclass Cross Entropy with Dice Loss

CODE Execution

  1. Use specific GPU bash CUDA_VICUDA_VISIBLE_DEVICES=0 python train_cls.py ... CUDA_VICUDA_VISIBLE_DEVICES=1,2 python train_cls.py ...
  2. Execute the distributed data-parallel in Pytorch
    • Execute the python script in a single server (--standalone) bash torchrun --standalone --nnodes=1 --nproc-per-node=${NUM_TRAINERS} train_cls.py ...

Reference

  1. https://github.com/qubvel/segmentation_models.pytorch/tree/master
  2. https://github.com/facebookresearch/fairseq/tree/main
  3. https://github.com/JunMa11/SegLossOdyssey/tree/master

Citation

If you find this repository useful for your research or if it helps any of your work, please consider citing it. GitHub will automatically generate a citation for you in APA or BibTeX format when you click the 'Cite this repository' button above the file list. @software{Kwon_My_PyTorch_Template, author = {Kwon, Doyoung}, title = {{My PyTorch Template}}, url = {https://github.com/KwonDoRyoung/my-pytorch-template}, version = {0.1} }

Owner

  • Name: KwonDoYoung
  • Login: KwonDoRyoung
  • Kind: user

Citation (CITATION.cff) 

cff-version: 1.2.0
message: "If you use this repository, please cite it as below."
title: "My PyTorch Template"
authors:
  - family-names: "Kwon"
    given-names: "Doyoung"
    affiliation: "Department of Artificial Intelligence, Kyungpook National University, Daegu, South Korea"
    email: "entropy0437@gmail.com"
version: 0.1
year: 2024
url: "https://github.com/KwonDoRyoung/my-pytorch-template"

GitHub Events

Total
Last Year