Are you sure you want to delete this task? Once this task is deleted, it cannot be recovered.
jasondhuang 7c4513de0f | 1 year ago | |
---|---|---|
.. | ||
README.md | 1 year ago | |
README_CN.md | 1 year ago | |
repmlp.png | 1 year ago | |
repmlp_T224.yaml | 1 year ago | |
repmlpblock.png | 1 year ago |
RepMLPNet: Hierarchical Vision MLP with Re-parameterized Locality
Accepted to CVPR-2022!
The latest version: https://openaccess.thecvf.com/content/CVPR2022/papers/Ding_RepMLPNet_Hierarchical_Vision_MLP_With_Re-Parameterized_Locality_CVPR_2022_paper.pdf
Compared to the old version, we no longer use RepMLP Block as a plug-in component in traditional ConvNets. Instead, we build an MLP architecture with RepMLP Block with a hierarchical design. RepMLPNet shows favorable performance, compared to the other vision MLP models including MLP-Mixer, ResMLP, gMLP, S2-MLP, etc.
Of course, you may also use it in your model as a building block.
The overlap between the two versions is the Structural Re-parameterization method (Localtiy Injection) that equivalently merges conv into FC. The architectural designs presented in the latest version significantly differ from the old version (ResNet-50 + RepMLP).
Model | Context | Top-1 (%) | Top-5 (%) | Params (M) | Train T. | Infer T. | Download | Config | Log |
---|---|---|---|---|---|---|---|---|---|
repmlp_t224 | D910x8 | 76.649 | 38.3 | 1011s/epoch | 15.8ms/step | model | cfg | log |
Please refer to the installation instruction in MindCV.
Please download the ImageNet-1K dataset for model training and validation.
Hyper-parameters. The hyper-parameter configurations for producing the reported results are stored in the yaml files in mindcv/configs/repmlp
folder. For example, to train with one of these configurations, you can run:
# train repmlp_t224 on 8 Ascends
bash ./scripts/run_distribution_ascend.sh ./scripts/rank_table_8pcs.json [DATASET_PATH] ./config/repmlp/repmlp_T224.yaml
Note that the number of GPUs/Ascends and batch size will influence the training results. To reproduce the training result at most, it is recommended to use the same number of GPUs/Ascneds with the same batch size.
Detailed adjustable parameters and their default value can be seen in config.py.
To validate the trained model, you can use validate.py
. Here is an example for densenet121 to verify the accuracy of
pretrained weights.
python validate.py --model=RepMLPNet_T224 --data_dir=imagenet_dir --val_split=val --ckpt_path
Please refer to the deployment tutorial in MindCV.
No Description
Jupyter Notebook Python Markdown Text
Dear OpenI User
Thank you for your continuous support to the Openl Qizhi Community AI Collaboration Platform. In order to protect your usage rights and ensure network security, we updated the Openl Qizhi Community AI Collaboration Platform Usage Agreement in January 2024. The updated agreement specifies that users are prohibited from using intranet penetration tools. After you click "Agree and continue", you can continue to use our services. Thank you for your cooperation and understanding.
For more agreement content, please refer to the《Openl Qizhi Community AI Collaboration Platform Usage Agreement》