Course 4 - Wk 2 - Ass 2 - kernel died - out of memory

Course Q&A Deep Learning Specialization Convolutional Neural Networks
1

Course 4 - Wk 2 - Program Assignment 2 - Transfer Learning
Got error message below:
“The kernel appears to have died. It will restart automatically”.

While running the last part of the fine-tuning assignment:
Epoch 5/10
9/9 [==============================] - 36s 4s/step - loss: 0.6796 - accuracy: 0.7672 - val_loss: 0.3451 - val_accuracy: 0.8462
Epoch 6/10
9/9 [==============================] - 56s 6s/step - loss: 0.3250 - accuracy: 0.8397 - val_loss: 0.2224 - val_accuracy: 0.9231
Epoch 7/10
1/9 [==>…] - ETA: 0s - loss: 0.1379 - accuracy: 0.9688

I only have this notebook running, have tried to re-start the kernel several times. Re-start my computer and run again. Got same problem.

Your help is much appreciated.

Cindy

2

Perhaps your code has an infinite loop. To guard against that, the kernel has a time-out that detects when your code is stuck.

3

Thanks for input TMosh.
When I reduced the (initial_epochs = 5) to 3
and (fine_tune_epochs = 5) to 2
The program runs well.

It seems that it runs out of memory with more epochs?
I did not add any additional loops to the code

4

There is an error in your code. It’s eating memory on every iteration.

Perhaps you deleted something from one of the code loops that you shouldn’t have.

5

Thank you so much for your help, Tmosh.

Found the error. I forgot to re-freeze the layers.

Working now. :slight_smile: