Examples

You can find examples here.

Basic examples

• Quick start.

• MNIST example.

• MNIST VAE example.

• CIFAR example.

• Model loading.

• Sequential LR scheduler.

Advanced examples

• CIFAR with DDP, mixed precision and gradient accumulation.

  Single GPU training:

  python cifar_advanced.py --batch_size 256 --lr 0.001
  

  Single machine 2 GPUs distributed data parallel training:

  ./cifar_advanced.sh 2 --batch_size 128 --lr 0.0005
  

  DDP training with mixed precision and gradient accumulation:

  ./cifar_advanced.sh 2 --batch_size 128 --lr 0.0005 --amp --iter_size 2
  

• Custom callback events.

• Custom build methods for creation of model parts.

Solutions of competitions

• 1st place solution for Sensorium Competition at NeurIPS 2023

• 1st place solution for SoccerNet Ball Action Spotting Challenge at CVPR 2023

• 1st place solution for SoccerNet Camera Calibration Challenge at CVPR 2023

• 1st place solution for Freesound Audio Tagging 2019 at Kaggle

• 14th place solution for TGS Salt Identification Challenge at Kaggle

• 22nd place solution for RANZCR CLiP - Catheter and Line Position Challenge at Kaggle

• 45th place solution for RANZCR CLiP - Catheter and Line Position Challenge at Kaggle

• 50th place solution for Quick, Draw! Doodle Recognition Challenge at Kaggle

• 66th place solution for Airbus Ship Detection Challenge at Kaggle

• Community Prize solution for Seismic Facies Identification Challenge at AIcrowd

• Solution for Deep Chimpact: Depth Estimation for Wildlife Conservation at DrivenData

• Solution for Humpback Whale Identification at Kaggle

• Solution for VSB Power Line Fault Detection at Kaggle

• Solution for Bengali.AI Handwritten Grapheme Classification at Kaggle

• Solution for ALASKA2 Image Steganalysis competition at Kaggle