Problem with C3W2_Assignment losses increments

I follow the instructions and the examples but the loss increases instead of decrease,

Start fine-tuning!
batch 0 of 100, loss=1.1338427
batch 10 of 100, loss=45.494125
batch 20 of 100, loss=32279.213
batch 30 of 100, loss=45429.24
batch 40 of 100, loss=49855.926
batch 50 of 100, loss=51232.387
batch 60 of 100, loss=51545.312
batch 70 of 100, loss=51607.55
batch 80 of 100, loss=51563.16
batch 90 of 100, loss=51373.293
Done fine-tuning!

I’m getting in Exercise 6.1
but Expected output:

can I share code ?

and i Exercise 10: Define the training step is confuse to me,
is true_shape_tensor the shape of preprocessed_image_tensor.shape ?

or model.provide_groundtruth is after or before with tf.GradientTape() as tape:
because the ### START CODE HERE (Replace instances of None with your code) ###

I hope some can help me

Hello Edwin,

when your loss get increased with each batch_size means there is an issue with your model algorithm.

Please share your notebook via person DM. Click my name and then message. Thank you for not posting the codes and following Code of Conduct.


Hi Edwin, were you able to resolve this?

I get tensorflow.python.checkpoint.checkpoint.Checkpoint instead of

Hello Ola,

His issue was resolved. Please share your detail issue without sharing codes.

If no error, then you can send me your notebook for review with explanation of your issue.

If sending notebook, click on my name and then message.