Hello im a fast.ai student and im currently doing a kaggle competition UBC Ovarian Cancer Subtype Classification and Outlier Detection (UBC-OCEAN)

fastai
1

The metrics used for the competition is balanced accuracy and im getting an error when im using fastai balanced accuracy
TypeError: Exception occured in Recorder when calling event after_batch:
unsupported operand type(s) for *: ‘AccumMetric’ and ‘int’

2

can someon please help

3

Please show your code

4

learn = vision_learner(dls,‘resnet26d’,metrics=BalancedAccuracy).to_fp16()
learn.fine_tune(5,0.01)

5

Not sure but shouldn’t this be BalancedAccuracy() with the brackets instead ?

6

i tried but it still not working when i use other metrics like accuracy or error_rate it works

7

I would like to team up with you @Michael_bb for this competition. Feel free to email me at Kamalmohapy@gmail.com

8

Put brackets for it to work. Doing it this way BalancedAccuracy() works for me.

If the issue persists, paste the error you’re getting on here.

9

alright cool

10

Thanks it works

11

Thank you it works

12

That’s awesome @Michael_bb . Would you like us to team up for this competition and hopefully kill it

13

that will be great