Week 2 Assignment 2 Alpaca Model: No improvement after fine-tuning

Hello Deeplearning.AI,

I have completed the assignment 2 of week 2 (100/100) and have a doubt regarding fine-tuning step. It says

Ahhh, quite an improvement! A little fine-tuning can really go a long way.

But I didn’t notice any improvement in accuracy. Is this expected or am I missing something as I passed the assignment with 100/100 grade ?

Before fine tuning, these are the results:

After fine tuning, it should have improved to this:

This seems like a useful improvement.

@TMosh Thanks for the response. Actually I missed defining weights from imagenet. That property was set to None and I was getting 0.4466 accuracy before and after fine-tuning. When I fixed this, everything worked fine.

An interesting thing to note here is that I got 100 points even with incorrect implementation.

Thanks for the explanation.

I’ll submit an issue to the course staff about there being no test for the weight values.