I’m trying to implement the SimpleQuadratic model (x^2a + xb + c); however, although my model follows all the instructions, its accuracy doesn’t get better than 86% whereas the assignment expects it to be more than 97%. I don’t get what’s the problem really, I don’t think the problem is from my solution… It does exactly what it was told. Can someone help please?
C1W3 Assignment --- Low Accuracy of the model
Course Q&A
TensorFlow: Advanced Techniques Specialization
Custom Models, Layers and Loss Functions with TF
Hi @mehradj
Please pay attention to the following points:
- Are you normalizing or scaling the input features?
- Did you implement the training loop correctly?
- Double-check that your forward pass really computes
a*x^2 + b*x + cfor each input in the batch, not just for a scalar.
If these didn’t work, you can share your code so we can fine the issue together!
hi @mehradj
an usual mistake can happen if your input shape for a_init_val is incorrect. Can you please share a screenshot of the codes you wrote by personal DM.
Regards
DP
Can we review the code please? I think I have done the things you mentioned. Where can I send the code though? is it OK to paste it here?