Arch-Net: Model Distillation for Architecture Agnostic Model Deployment

The official implementation of Arch-Net: Model Distillation for Architecture Agnostic Model Deployment


TL;DR Arch-Net is a family of neural networks made up of simple and efficient operators. When a Arch-Net is produced, less common network constructs, like Layer Normalization and Embedding Layers, are eliminated in a progressive manner through label-free Blockwise Model Distillation, while performing sub-eight bit quantization at the same time to maximize performance. For the classification task, only 30k unlabeled images randomly sampled from ImageNet dataset is needed.

Main Results

ImageNet Classification

Model Bit Width Top1 Top5
Arch-Net_Resnet18 32w32a 69.76 89.08
Arch-Net_Resnet18 2w4a 68.77 88.66
Arch-Net_Resnet34 32w32a 73.30 91.42
Arch-Net_Resnet34 2w4a 72.40 91.01
Arch-Net_Resnet50 32w32a 76.13 92.86
Arch-Net_Resnet50 2w4a 74.56 92.39
Arch-Net_MobilenetV1 32w32a 68.79 88.68
Arch-Net_MobilenetV1 2w4a 67.29 88.07
Arch-Net_MobilenetV2 32w32a 71.88 90.29
Arch-Net_MobilenetV2 2w4a 69.09 89.13

Multi30k Machine Translation

Model translation direction Bit Width BLEU
Transformer English to Gemany 32w32a 32.44
Transformer English to Gemany 2w4a 33.75
Transformer English to Gemany 4w4a 34.35
Transformer English to Gemany 8w8a 36.44
Transformer Gemany to English 32w32a 30.32
Transformer Gemany to English 2w4a 32.50
Transformer Gemany to English 4w4a 34.34
Transformer Gemany to English 8w8a 34.05


python == 3.6

refer to requirements.txt for more details

Data Preparation

Download ImageNet and multi30k data(google drive or BaiduYun, code: 8brd) and put them in ./arch-net/data/ as follow:

├── imagenet
│   ├── train
│   ├── val
├── multi30k

Download teacher models at google drive or BaiduYun(code: 57ew) and put them in ./arch-net/models/teacher/pretrained_models/

Get Started

ImageNet Classification (take archnet_resnet18 as an example)

train and evaluate

cd ./train_imagenet

python3 -m torch.distributed.launch --nproc_per_node=8  -j 8 --weight-bit 2 --feature-bit 4 --lr 0.001 --num_gpus 8 --sync-bn

evaluate if you already have the trained models

python3 -m torch.distributed.launch --nproc_per_node=8  -j 8 --weight-bit 2 --feature-bit 4 --lr 0.001 --num_gpus 8 --sync-bn --evaluate

Machine Translation

train a arch-net_transformer of 2w4a

cd ./train_transformer

python3 --translate_direction en2de --teacher_model_path ../models/teacher/pretrained_models/transformer_en_de.chkpt --data_pkl ../data/multi30k/m30k_ende_shr.pkl --batch_size 48 --final_epochs 50 --weight_bit 2 --feature_bit 4 --lr 1e-3 --weight_decay 1e-6 --label_smoothing
  • for arch-net_transformer of 8w8a, use the lr of 1e-3 and the weight decay of 1e-4


cd ./evaluate

python3 --data_pkl ./data/multi30k/m30k_ende_shr.pkl --model path_to_the_outptu_directory/model_max_acc.chkpt
  • to get the BLEU of the evaluated results, go to this website, and then upload ‘predictions.txt’ in the output directory and the ‘gt_en.txt’ or ‘gt_de.txt’ in ./arch-net/data_gt/multi30k/


If you find this project useful for your research, please consider citing the paper.

      title={Arch-Net: Model Distillation for Architecture Agnostic Model Deployment}, 
      author={Weixin Xu and Zipeng Feng and Shuangkang Fang and Song Yuan and Yi Yang and Shuchang Zhou},






If you have any questions, feel free to open an issue or contact us at [email protected].


View Github