Sagemaker Notebook instance error AttributeError: 'MaterializedLayer' object has no attribute 'pack_annotations'

Question

I have a dask cluster active

from dask.distributed import Client, progress 

client = Client()
client

When I try to encode my data I get the error:

AttributeError: 'MaterializedLayer' object has no attribute 'pack_annotations'

I encoded the data with the cluster closed and then tried to fit the model and I get the same error.

same problem when using AWS SageMaker. It was working okay till yesterday! — armiro
– armiro, Commented Jun 23, 2022 at 21:54
I am about to use a bigger instance and not go for a distributed approach, I don't see any other options on my end. — Jorge
– Jorge, Commented Jun 26, 2022 at 6:28
In my case, the largest possible instance was not sufficient. I ended up creating a crawler and connecting the data to an Athena database. — armiro
– armiro, Commented Jun 27, 2022 at 17:39
I was even deleting some columns, I managed to train the model, but It was not giving me any predictions, I will just go iddle... Do you mind telling me more about your approach?, I would like to give it a try but I have no idea 🥺 — Jorge
– Jorge, Commented Jun 27, 2022 at 21:06
Actually, my intention was to load a very large dataset and perform some SQL queries to filter some rows and export results, so in this case, I don't think my approach helps you. I suggest you try AWS EMR and create a cluster, then connect it to a SageMaker notebook. Let me know if it helped you. — armiro
– armiro, Commented Jun 27, 2022 at 21:20

Jorge · Accepted Answer · 2022-06-29 16:49:15Z

1

Solution: Sagemaker Kernel most be python V3.8 which is compatible with dask distributed, as soon as I update the kernel to python 3.8. Works just fine.

answered Jun 29, 2022 at 16:49

Jorge

12910 bronze badges

Sign up to request clarification or add additional context in comments.

Collectives™ on Stack Overflow

Sagemaker Notebook instance error AttributeError: 'MaterializedLayer' object has no attribute 'pack_annotations'

1 Answer 1

Comments

Your Answer

Hot Network Questions

Collectives™ on Stack Overflow

1 Answer 1

Comments

Your Answer

Sign up or log in

Post as a guest

Related