Typeerror: argument of type "non type" is not iterable

[Hrusikesh-AIML]

Hi Abraham,

@AbrahamSanders
I added some data in csv_data and ran the model using command (python train.py --datasetdir=datasets\csv) got below error.

" typeerror: argument of type "non type" is not iterable "

below is the data which i have kept in csv_data.csv and trained

question,answer
your question,your answer
what is your name,bob
shall we start the class,"yes,we can"
can you help me with the notes,check the below description
I love pizza,Pizza might be good to eat .lets get back into the topic
I want to play football,will play after the lession

Please help me to resolve this issue.

Thank you :)

Regards,
Hrusikesh
+91-9886036753

[sudheer007]

Hi @AbrahamSanders :
am also getting the same error..kindly help with this..!

Thanks,
[email protected]

[AbrahamSanders]

Hi @pandahrusikesh88 and @sudheer007

It is interesting that in your screenshot I can see it printed "Training Complete!" which is the last line in the train.py script. This indicates that all of the epochs have finished and the model checkpoint has been successfully updated, meaning that the training script worked correctly.

Perhaps the error that you are receiving is peripheral, relating to clearing some kind of internal cache (the function in the traceback is named "TensorCacheDeleter"). I have never encountered this issue before, but then again I typically terminate training long before it completes the full number of epochs. What version of TF do you have installed?

Check to see if you can chat with your trained model without an error. It should work.

Also, a side note - if what you pasted is the entire dataset (6 lines), your model results will most likely be extreme overfitting. I hope you have a much larger dataset to train with.