Why my test_loss and test_acc are not improving per epoch

[SailSabnis]

Hi all.

Struggling with this one. I am sure this seems pretty basic and would appreciate any help.
For the FoodVisionMini model using 3 classes (pizza, sushi, steak) example - my train_loss and test_acc seem to remain constant across all epochs. See Screenshot 1 -
Screenshot 1 -

Here's my train_step and test_step function along with acc function and a train_func that combines both train and test step-

Screenshot 2 - (Defining TinyVGG model)

Screenshot 3 - (accuracy function, train_step and test_step function)

Screenshot 4 - (train function combining train_step and test_step and running 5 epochs)

[SailSabnis]

@mrdbourke Please don't kill me for tagging you directly but scrambling for help. Scratching my head over this for the last 3 days and couldn't figure what's wrong with my test_loss and test_Acc. Can you (or any other expert if reading this) please help. Many thanks!

[SailSabnis]

Hi fellow learners, can someone please help out here?

[mrdbourke]

Hey @SailSabnis ,

Great question!

Your code looks pretty good to me.

And it looks like your results are similar to what the results are in the example notebook:

 0%|          | 0/5 [00:00<?, ?it/s]
Epoch: 1 | train_loss: 1.1078 | train_acc: 0.2578 | test_loss: 1.1360 | test_acc: 0.2604
Epoch: 2 | train_loss: 1.0847 | train_acc: 0.4258 | test_loss: 1.1620 | test_acc: 0.1979
Epoch: 3 | train_loss: 1.1157 | train_acc: 0.2930 | test_loss: 1.1697 | test_acc: 0.1979
Epoch: 4 | train_loss: 1.0956 | train_acc: 0.4141 | test_loss: 1.1384 | test_acc: 0.1979
Epoch: 5 | train_loss: 1.0985 | train_acc: 0.2930 | test_loss: 1.1426 | test_acc: 0.1979
Total training time: 4.935 seconds

See here: https://www.learnpytorch.io/04_pytorch_custom_datasets/#77-train-and-evaluate-model-0

It seems that the TinyVGG model isn't capable of forming a great representation with the current data setup/training setup we have (this is okay, sometimes models don't have the parameters/data to learn, we'll improve results in the Transfer Learning section: https://www.learnpytorch.io/06_pytorch_transfer_learning/).

Your results look similar to the example results too:

---

If I've misunderstood your question, please feel free to add more information/link to a video/timestamp you're working through so I can see the full code (e.g. what data is being used).

Best,

Daniel