To-do

Coding

• Check step up
• Check data builder
• Check model builder
• Run a small experiment to know pipeline (data input, data loader, build model, train model, evaluate)
• Check SSFA
• Check VoTr
• Add IA-SSD
• Create Models (mix modules)

Writing

• Introduction
• Related Work

Commands

Training

torchrun --nproc_per_node <number_gpus> train.py <configuration_path> <output_path> --batch_size <batch_size_per_gpu> --num_workers ?

Testing

 torchrun --nproc_per_node <number_gpus> evaluate.py <configuration_path> <checkpoint> <output_path> --batch_size <batch_size_per_gpu> --num_workers ?