EfficientNet
EfficientNet: Rethinking Model Scaling for Convolutional Neural Networks
Introduction
Figure 1 shows the methods from three dimensions -- width, depth, resolution and compound to expand the model. Increasing the model
size solely would cause the model performance to sub-optimal solution. Howerver, if three methods could be applied together into the model
, it is more likely to achieve optimal solution. By using neural architecture search, the best configurations for width scaling, depth scaling
and resolution scaling could be found. EfficientNet could achieve better model performance on ImageNet-1K dataset compared with previous methods.[1]
Figure 1. Architecture of Efficientent [1]
Results
Our reproduced model performance on ImageNet-1K is reported as follows.
Model |
Context |
Top-1 (%) |
Top-5 (%) |
Params (M) |
Recipe |
Download |
efficientnet_b0 |
D910x64-G |
76.89 |
93.16 |
5.33 |
yaml |
weights |
efficientnet_b1 |
D910x64-G |
78.95 |
94.34 |
7.86 |
yaml |
weights |
Notes
- Context: Training context denoted as {device}x{pieces}-{MS mode}, where mindspore mode can be G - graph mode or F - pynative mode with ms function. For example, D910x8-G is for training on 8 pieces of Ascend 910 NPU using graph mode.
- Top-1 and Top-5: Accuracy reported on the validation set of ImageNet-1K.
Quick Start
Preparation
Installation
Please refer to the installation instruction in MindCV.
Dataset Preparation
Please download the ImageNet-1K dataset for model training and validation.
Training
It is easy to reproduce the reported results with the pre-defined training recipe. For distributed training on multiple Ascend 910 devices, please run
# distributed training on multiple GPU/Ascend devices
mpirun -n 64 python train.py --config configs/efficientnet/efficientnet_b0_ascend.yaml --data_dir /path/to/imagenet
If the script is executed by the root user, the --allow-run-as-root
parameter must be added to mpirun
.
Similarly, you can train the model on multiple GPU devices with the above mpirun
command.
For detailed illustration of all hyper-parameters, please refer to config.py.
Note: As the global batch size (batch_size x num_devices) is an important hyper-parameter, it is recommended to keep the global batch size unchanged for reproduction or adjust the learning rate linearly to a new global batch size.
If you want to train or finetune the model on a smaller dataset without distributed training, please run:
# standalone training on a CPU/GPU/Ascend device
python train.py --config configs/efficientnet/efficientnet_b0_ascend.yaml --data_dir /path/to/dataset --distribute False
Validation
To validate the accuracy of the trained model, you can use validate.py
and parse the checkpoint path with --ckpt_path
.
python validate.py -c configs/efficientnet/efficientnet_b0_ascend.yaml --data_dir /path/to/imagenet --ckpt_path /path/to/ckpt
Deployment
Please refer to the deployment tutorial in MindCV.
References
[1] Tan M, Le Q. Efficientnet: Rethinking model scaling for convolutional neural networks[C]//International conference on machine learning. PMLR, 2019: 6105-6114.