bemorept
(Wonwoo Park)
1
Hi, All.
I have met this error when run pytorch inference in Jetson nano.
"AttributeError: ‘ReLU’ object has no attribute ‘threshold’ ".
What wrong to me?
Jetson nano, Ubuntu 18.04LTS aarch64, Pytorch 1.1.0, Torchvision 0.3.0
Thanks.
@bemoregt.
Could you post the code snippet which raises this error?
Have you solved it?I also ran into it when I called the model
I ran into the issue when I am running the following code using the export.pkl model I had saved in my directory:
import fastai
from fastai.vision import *
def main():
classes = ['Potato___Late_blight',
'Cherry_(including_sour)___healthy',
'Apple___Black_rot',
'Apple___Apple_scab',
'Cherry_(including_sour)___Powdery_mildew',
'Potato___healthy',
'Apple___Cedar_apple_rust',
'Potato___Early_blight',
'Apple___healthy']
#load predictor
predictor = load_learner('.')
img = open_image('apple_black_rot1.jpg')
pred,pred_idx,probs = predictor.predict(img)
print(classes[pred_idx])
if __name__=="__main__":
main()
The same error pops up along with various warnings. Is there any fix to this yet?