PiT (Pooling-based Vision Transformer) is an improvement of Vision Transformer (ViT) model proposed by Byeongho Heo in 2021. PiT adds pooling layer on the basis of ViT model, so that the spatial dimension of each layer is reduced like CNN, instead of ViT using the same spatial dimension for all layers. PiT achieves the improved model capability and generalization performance against ViT. [1]
Figure 1. Architecture of PiT [1]
Our reproduced model performance on ImageNet-1K is reported as follows.
| Model | Context | Top-1 (%) | Top-5 (%) | Params (M) | Recipe | Download |
|---|---|---|---|---|---|---|
| pit_ti | D910x8-G | 72.96 | 91.33 | 4.85 | yaml | weights |
| pit_xs | D910x8-G | 78.41 | 94.06 | 10.61 | yaml | weights |
| pit_s | D910x8-G | 80.56 | 94.80 | 23.46 | yaml | weights |
| pit_b | D910x8-G | 81.87 | 95.04 | 73.76 | yaml | weights |
- 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.
Please refer to the installation instruction in MindCV.
Please download the ImageNet-1K dataset for model training and validation.
- Distributed 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 8 python train.py --config configs/pit/pit_xs_ascend.yaml --data_dir /path/to/imagenetIf the script is executed by the root user, the
--allow-run-as-rootparameter must be added tompirun.
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.
- Standalone Training
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/pit/pit_xs_ascend.yaml --data_dir /path/to/dataset --distribute FalseTo 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/pit/pit_xs_ascend.yaml --data_dir /path/to/imagenet --ckpt_path /path/to/ckptPlease refer to the deployment tutorial in MindCV.
[1] Heo B, Yun S, Han D, et al. Rethinking spatial dimensions of vision transformers[C]//Proceedings of the IEEE/CVF International Conference on Computer Vision. 2021: 11936-11945.