Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

How to export the SwinT model from BEVFusion? #275

Open
slayerlpj opened this issue Aug 2, 2024 · 2 comments
Open

How to export the SwinT model from BEVFusion? #275

slayerlpj opened this issue Aug 2, 2024 · 2 comments

Comments

@slayerlpj
Copy link

I have trained camera+lidar based on SwinT, I have got a pth. How can I generate PTQ model or FP16 model?

I used the command "python qat/ptq.py --config=configs/nuscenes/det/transfusion/secfpn/camera+lidar/swint_v0p075/configs.yaml --ckpt=model/swint/epoch_30.pth --calibrate_batch 300", the I got a error as follow:

2024-08-02 09:16:14,038 - mmdet - INFO - load checkpoint from http path: https://github.com/SwinTransformer/storage/releases/download/v1.0.0/swin_tiny_patch4_window7_224.pth
I0802 09:16:14.038485 139857851803456 logging.py:101] load checkpoint from http path: https://github.com/SwinTransformer/storage/releases/download/v1.0.0/swin_tiny_patch4_window7_224.pth
load checkpoint from local path: model/swint/epoch_30.pth
Traceback (most recent call last):
File "qat/ptq.py", line 150, in
main()
File "qat/ptq.py", line 130, in main
model = quantize_net(model)
File "qat/ptq.py", line 81, in quantize_net
quantize.quantize_encoders_camera_branch(model.encoders.camera)
File "/home/work/work/lpj/BEVFusion/Lidar_AI_Solution/CUDA-BEVFusion/qat/lean/quantize.py", line 175, in quantize_encoders_camera_branch
major = model_camera_branch.backbone.layer3[0].conv1._input_quantizer
File "/home/work/anaconda3/envs/bevfusion/lib/python3.8/site-packages/torch/nn/modules/module.py", line 1177, in getattr
raise AttributeError("'{}' object has no attribute '{}'".format(
AttributeError: 'SwinTransformer' object has no attribute 'layer3'

How can I fix the problem?

@slayerlpj
Copy link
Author

When I used the command to export fp16 model, I got a another error as follow:

python qat/export-camera.py --ckpt=model/swint/epoch_30.pth --fp16
Traceback (most recent call last):
File "qat/export-camera.py", line 138, in
main()
File "qat/export-camera.py", line 85, in main
model = torch.load(args.ckpt).module
AttributeError: 'dict' object has no attribute 'module'

@evil-master
Copy link

当我使用命令导出 fp16 模型时,我收到了另一个错误,如下所示:

python qat/export-camera.py --ckpt=model/swint/epoch_30.pth --fp16 回溯(最近调用最后一次):文件 “qat/export-camera.py”,第 138 行,在 main() 中 文件“qat/export-camera.py”,第 85 行,在 main model 中 = torch.load(args.ckpt).module AttributeError:“dict”对象没有属性“module”

how do you reslove it?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants