please go through the screenshot below . I am trying to get a smooth curve but nit able to achieve so using mini. batch gradient descent. I have tried various hit and trail for different parameters but this is the curve I am getting . What should I do to get a smooth curve
https://drive.google.com/file/d/1peZJsKVmaByWlKKsAAHlGN7jxpyBMgRX/view?usp=sharing
Final parameters:
batch_size=400,learning_rate=0.001,max_itrs=500
hey @Par1hsharma ,
parameters like batch_size , learning_rate and max_itrs are not the only things needed.
The model architecture also plays a very big role in this.
hence try a bit on that , i guess that would improve a bit.
What is model architecture and please suggest me some ways around what can de done to improve model architecture
Model architecture is how your build those deep learning models, how you stack those layers upon each other and use them.
like for example 3 dense layers ,one after the other , and then a output layer.
Now , the number of nodes on these layers , is what your need to work upon.
See which is the best combination and all.
I used Logistic Regression here . I haven’t reached deep learning in the course yet
okay.
Then it might work or might not.
You need to work on data, complete logistic regression assumptions and then try again once.
I did it again but no success what to do next?
what did you did ? can you tell me