Hello,In the first video Akash sir showed us the actual way to linear regression and gradient descent then he showed the linear regression & gradient descent using pytorch. In the first actual way,we can set the learning rate of weight(value of W.grad) after watching the gradients.But when we will use the pytorch for this operation, How can we determine the learning rate?
It’s set up during optimizer initialization. You may still change it (although not so easily) or apply some scheduler to adjust it during the training.
Ya but when using pytorch,Can I see the derivatives of weights?
weight.grad is the thing you probably look for. It is actually derivative (partial
error / partial
weight). No idea what you want to achieve with this info though.
I noticed that some of the
jovian.commit() statements in the notebooks set
environment=None. I checked the docs and found out that it is used to control the “capturing the environment” part. What exactly would be recorded as part of “capturing the environment”? Or what’s ignored when
Am i late for this course, as 3 lessons have almost completed and last date for assignment 1 has already been passed. My assignments will be evaluated and am i eligible for a certification of this course?
No, the date for assignment 1 has been extended to 12 dec, now see if you can manage that.
Will there be up after course is done that you can work at own paste like with Zero to Pandas? I wonder because I am unsure if I will pass and am sure many others are. Would be good to know that even if you fail the course. You can do it again and this time have all the knowledge from before.
Even if you feel you are not learning fast enough. If you keep on you will still learn a bunch if you keep trying.
I Would like to thank your entire team for providing excellent series of lectures on machine learning, Python etc.
In the lectures mostly we have seen image classification problems.
I have a specific problem statement, where inputs are images with varying color intensities and for each image there is output (Number varying 0 to 12000 (this is not a label)) which is dependent on the color intensities of different pixels.
Can experts guide me how this can be done?
Thanks in advance.
Not an expert, but anyway:
Depends on the distribution of the output.
Also a similar task (regression) has been done during second assignment of this course.
My Second Assignment for Pytorch course is not yet evaluated. What can be the reason?
I came across this course quite late. Can I still be a part? I mean are these deadlines flexible? Can anything be done?
After completion of all course assignment and project how long it takes to receive course completion certificate ??
As you have completed the course assignments and project they will be evaluated first you will receive a passing grade on the assignments and the project after that the certificate will be issued within a few days. If by any chance you do not get a passing grade at first, you will get the chance to resubmit it.
RuntimeError Traceback (most recent call last)
----> 1 history = [evaluate(model, valid_dl)]
/usr/local/lib/python3.6/dist-packages/torch/nn/functional.py in linear(input, weight, bias)
1688 if input.dim() == 2 and bias is not None:
1689 # fused op is marginally faster
-> 1690 ret = torch.addmm(bias, input, weight.t())
1692 output = input.matmul(weight.t())
RuntimeError: mat1 dim 1 must match mat2 dim 0
Is it due to a limit on colab or something else?
No, it is because of dimension mismatch somewhere in your model, kindly check the model structure and inputs and targets of your model, for more information share your notebook link.
I have completed Python Zero to GANS course and by all assignments and project have received the pass grade.
So can you give me an estimate that after how much time will I get my certificate?
BTW thanks so much to jovian for providing this opportunity
Hey, Congratulations on completing the course project, Your certificates will be used within a few days have, wait for it to be issued.
Hi, I have completed the course and received the certificate. However at that moment, I had not included my real name in my profile, so the certificate is showing my github username. Is there a way to change it to use my real name? I have already updated my profile with it.
Kindly fill this google form to change your name on the certificate. https://docs.google.com/forms/d/e/1FAIpQLSeq9Oy8HflrHcRAht71am3DfsMr7wjUff5ZU1ku9b9OXgwJQA/viewform