Pytorch Golden Template

Class Diagram

Features

Setup

1. Use the template by clicking the green button or directly clone this repo

2. Install miniconda/anaconda, a package for package/environment management

wget repo.continuum.io/miniconda/Miniconda3-latest-Linux-x86_64.sh
bash Miniconda3-latest-Linux-x86_64.sh

3. Build conda environment from file

cd pytorch-golden-template
conda create -n PytorchGoldenTemplate python">=3.6" pytorch">=1.0" torchvision tensorboard pillow">=6.1" pandas coloredlogs imageio scipy -c pytorch
source activate PytorchGoldenTemplate
pip install attrdict tensorboardX
# Or
conda env create -f environment.yaml

4. Update submodules

git submodule update --init --recursive

5. Activate the environment

source activate <environemnt name>

Usage

usage: main.py [-h] [-tc TEMPLATE_CONFIG]
               [-sc SPECIFIED_CONFIGS [SPECIFIED_CONFIGS ...]] [-r RESUME]
               [-p PRETRAINED] [-d DEVICE] [--mode {train,test,eval}]
               [--saved_keys SAVED_KEYS [SAVED_KEYS ...]]
               [--ckpts_subdir CKPTS_SUBDIR] [--outputs_subdir OUTPUTS_SUBDIR]

PyTorch Template

optional arguments:
  -h, --help            show this help message and exit
  -tc TEMPLATE_CONFIG, --template_config TEMPLATE_CONFIG
                        Template configuraion file. It should contain all
                        default configuration and will be overwritten by
                        specified config.
  -sc SPECIFIED_CONFIGS [SPECIFIED_CONFIGS ...], --specified_configs SPECIFIED_CONFIGS [SPECIFIED_CONFIGS ...]
                        Specified configuraion files. They serve as experiemnt
                        controls and will overwrite template configs.
  -r RESUME, --resume RESUME
                        path to latest checkpoint (default: None)
  -p PRETRAINED, --pretrained PRETRAINED
                        path to pretrained checkpoint (default: None)
  -d DEVICE, --device DEVICE
                        indices of GPUs to enable (default: all)
  --mode {train,test,eval}
  --saved_keys SAVED_KEYS [SAVED_KEYS ...]
                        Specify the keys to save at testing mode.
  --ckpts_subdir CKPTS_SUBDIR
                        Subdir name for ckpts saving.
  --outputs_subdir OUTPUTS_SUBDIR
                        Subdir name for outputs saving.

Example: train a new model

python main.py

Example: train a new GAN model

python main.py -tc configs/template_GAN_train_config.json

Example: resume from a checkpoint, inference and save outputs

python main.py --resume ./saved/ckpts/template_config+CrossEntropy/0723_180600/ckpt-ep1-valid_mnist_avg_loss0.2885-best.pth --mode test

Example: generate testing results

# For submission to the leaderboard, etc.
python main.py --mode test -p <pretrained_weight>

Example: evaluate results

# You could evaluate results by other models (corresponding data loader needs to be defined)
# See the result_data_loaders in configs/template_eval_config.json for details
python main.py --mode eval

Tensorboard Integration

By default, Tensorboard logs are recoded under saved/runs.

Please run tensorboard --logidr saved/runs and go to localhost:6006 to see the training progress on tensorboard.

Folder Structure

Authors

• Ya-Liang Chang (Allen) amjltc295
• Zhe Yu Liu Nash2325138