bzha5848

Results 6 comments of bzha5848

Error on Colab: ModuleNotFoundError: No module named 'airdet'

I managed to solved the last error with !python -m tool.demo ....., but still get errors below. Just wondering how to solve this sort of path issues on colab? ![image](https://user-images.githubusercontent.com/49177682/181813387-6996b364-a5e3-4eff-b732-c8b6d64c170a.png)

config for resnet and mobilenet

Yeah, but I wanna keep the backbone of resnet and efficient and fine-tuning them with the head of yolov5, will that be possible?

RuntimeError: adaptive_max_pool2d_backward_cuda does not have a deterministic implementation, but you set 'torch.use_deterministic_algorithms(True)'

Can I ask why this error happens? Is there any method that I can avoid that issue?

Follow up Question RuntimeError: adaptive_max_pool2d_backward_cuda does not have a deterministic implementation, but you set 'torch.use_deterministic_algorithms(True)

Yeah I did, but can I ask why it happens? And is there any method to avoid this issue while modify the config file for model?

Follow up Question RuntimeError: adaptive_max_pool2d_backward_cuda does not have a deterministic implementation, but you set 'torch.use_deterministic_algorithms(True)

For this issue, is that related to the number of classes I used to train? Coz when I use the dataset with few classes on the same modified model config...

RuntimeError: Failed to run command: <built-in function compile>, errors: Command 'cd /home/tianyu/.cache/paddle_extensions && /home/tianyu/miniconda3/envs/paddle3/bin/python3 ms_deform_attn_setup.py build' returned non-zero exit status 1. LAUNCH INFO 2023-08-24 11:00:59,752 Exit code 1

When I check the log it gives me the info: 2023-08-24 13:13:42,775 - WARNING - No custom op ms_deform_attn found, try JIT build 2023-08-24 13:13:42,775 - WARNING - No custom...