YOLOX_pruning

YOLOX model prune

Train origin yolox-l model

python3 tools/train.py -f ../sample/my_hand_voc_l.py -b 32 --fp16

Train sparsity yolox-l model

add depend to tools/train.py

import sys
from pathlib import Path 
sys.path.append(str(Path(__file__).resolve().parents[2]/'prun_train'))
from my_train import MY_Trainer_Loose

and then change trainer = Trainer(exp, args) to trainer = MY_Trainer_Loose(exp, args) and then train the model

python3 tools/train.py -f ../sample/my_hand_voc_l.py -b 32 --fp16

pruning and retrain

python3 ../prun_train/prune.py -c ~/Downloads/latest_ckpt.pth -f ../sample/my_hand_voc_l.py -expn pruned_model