I ran it and doesn’t work also. seems like someone requested PR,
and I just plotted without last element.
I have been trying to search on why a uniform distribution would be better or worse than a normal distribution to initialize weights. I have not been able to get a good explanation on the forums. Can someone pls point me to it?
been a while I had the video, but as jeremy explained briefly(course 8), initializing with normal dist doesn’t guarantee mean 0, std1 and even worse when it gets deeper.
xavier/bengio pointed out this, might feel like experimental prove, not mathmatical
The reason may be because a Normal distribution concentrates the probability density around zero, while a Uniform distribution spreads the probability distribution evenly, including values well away from zero.
can you elaborate how that helps?
that is what i am seeing in most of the forums. looks more like emperical
Can someone please explain me the reason why Dataset and Dataloader classes are two separate classes created?
I mean, both can be implemented together, and then we can just create objects of this new class for our training purpose, so why they are two different classes, where Dataloader just inherits Dataset objects as it’s parameters?