• Towards Fully Automated and Efficient Model Compression
• Requirements
• Running

Towards Fully Automated and Efficient Model Compression

Requirements

• DALI
• Apex
• torchprofile
• other requirements, running requirements.txt

pip install -r requirements.txt

Running

feature extract

# [optional]cache imagenet dataset in RAM for accelerting I/O
code_path='/ITPruner/Imagenet/'
chmod +x ${code_path}/prep_imagenet.sh
cd ${code_path}
echo "preparing data"
bash ${code_path}/prep_imagenet.sh >> /dev/null
echo "preparing data finished"

python3 -m torch.distributed.launch --nproc_per_node=1 feature_extract.py \
 --model "mobilenet" \
 --path "Exp_base/mobilenet_base" \
 --dataset "imagenet" \
 --save_path 'your_data_path' \ 
 --target_flops 150000000 \
 --beta 243

or

bash ./run_scripts/feature_extract/extract_mobilenet_150m.sh

Train

Because of random seed, cfg obtained through feature extraction may have a little difference from ours. Our cfg are given in .sh files.

# [optional]cache imagenet dataset in RAM for accelerting I/O
code_path='/ITPruner/Imagenet/'
chmod +x ${code_path}/prep_imagenet.sh
cd ${code_path}
echo "preparing data"
bash ${code_path}/prep_imagenet.sh >> /dev/null
echo "preparing data finished"

python3 -m torch.distributed.launch --nproc_per_node=4 train.py --train \
 --model "mobilenet" \
 --cfg "[23, 42, 63, 67, 132, 134, 275, 194, 202, 202, 194, 277, 522, 687]" \
 --path "Exp_train/train_mobilenet_150m_${RANDOM}" \
 --dataset "imagenet" \
 --save_path 'your_data_path' \
 --base_path "Exp_base/mobilenet_base" \
 --warm_epoch 1 \
 --sync_bn \
 --n_epochs 250 \
 --label_smoothing 0.1

or

bash ./run_scripts/train/train_mobilenet_150m.sh

Evaluate

We provide some pretrain_models which we list in the paper.

python3 -m torch.distributed.launch --nproc_per_node=1 evaluate.py \
 --model "mobilenet" \
 --path "pretrain_models/train_mobilenet_150m_31752" \
 --dataset "imagenet" \
 --save_path 'your_data_path' \
 --cfg "[23, 42, 63, 67, 132, 134, 275, 194, 202, 202, 194, 277, 522, 687]"

or

bash ./run_scripts/evaluate/evaluate_mobilenet_150m.sh