apcc-geoslegend / weather4cast2021-swinencoderdecoder

Weather4cast2021-SwinEncoderDecoder (AI4EX Team)

Table of Content

• General Info
• Requirements
• Installation
• Usage
• Inference

General Info

The resipository contains the code and learned model parameters for our submision in Weather4cast2021 stage-1 competition.

Requirements

This resipository depends on the following packages availability

• Pytorch Lightning
• timm
• torch_optimizer
• pytorch_model_summary
• einops

Installation:

unzip folder.zip
cd folder
conda create --name swinencoder_env python=3.6
conda activate swinencoder_env
conda install pytorch=1.9.0 cudatoolkit=10.2 -c pytorch
pip install -r requirements.txt

Usage

• a.1) train from scratch

  python main.py --gpus 0 --use_all_region
  

• a.2) fine tune a model from a checkpoint

  python main.py --gpu_id 1 --use_all_region --mode train --name ALL_real_swinencoder3d_688080 --time-code 20210630T224355 --initial-epoch 58```
  
  

• b.1) evaluate an untrained model (with random weights)

  python main.py --gpus 0 --use_all_region --mode test
  

• b.2) evaluate a trained model from a checkpoint (submitted inference)

  python main.py --gpu_id 1 --use_all_region --mode test --name ALL_real_swinencoder3d_688080 --time-code 20210630T224355 --initial-epoch 58
  

Inference

To generate predictions using our trained model

R=R1
INPUT_PATH=../data
WEIGHTS=logs/ALL_real_swinencoder3d_688080
OUT_PATH=.
python inference.py -d $INPUT_PATH -r $R -w $WEIGHTS -o $OUT_PATH -g 1