EffB4.cmd

  
REM #!/usr/bin/env bash

REM # Train model ligt aug, First try,  size = 456, batch = 8


REM # Train 4 folds on this data

python train.py --seed 322 --use-current -l tempared_log_loss -t1 0.2 -t2 4.0 -optim adam -v -m efficientnet-b4 -b 8 -e 300 -s 380 -f 0 -a kaggle_light -lr 3e-4 -accum 4 -metric cosine_loss