the way he said "work of art" to every beauty of machine learning depicts that he loves machine learning so much. the whole video is "work of art" by the way best ever video i have seen on youtube
I feel like you missed the part about transposing the weight matrix before multiplying with inputs, this kind of threw me off a bit and I had to research about how the nn.Linear actually works. Also in general the way nn.Linear works is very different from how neural networks are taught in classes so it was a bit hard to grasp right off the bat. Thanks for the explanations!
same issue bro , 1x2 matrix can be multiplies with 2x8 matrix to yield 1x2 matrix ... I did not get it why is he doing the second step , although in his first step is he trying to perform 1x2 matrix multiplication with 8x2 matrix??
aren't you supposed to transpose the matrix before multiplying? 1,2 array cannot be multiplied to 8,2 matrix, but it can with matrix transpose. Am I wrong?
yeah i was confused by this too, I think what we are actually doing is xA_1A_2 = y where A_1 is 2x8 and A_2 8x1. Notice the notation for nn.Linear(2,8) and nn.Linear(8,1). Could be wrong though
Anyone understands why the NN got "stuck" on a "wrong" output even though it reached a minimum loss function? My understanding is it got to a local minimum point and cant get out of there. If so - is there any other way to release it? other than setting new random values to the matrices and start over.
Fantastic tutorial, as always. It really cleared up the concepts behind neural networks for me. Can't wait for next parts. Keep on going.
More of this please! Great tutorial as usual
the way he said "work of art" to every beauty of machine learning depicts that he loves machine learning so much. the whole video is "work of art" by the way best ever video i have seen on youtube
Hands down a great tutorial! Helps me a bunch on my current classes. If only the lectures at school were as good as these...
I got reccomended this video, and I really enjoyed it. Very intuitive and simple explanation. Can't wait for the next part!
Really useful and clear breakdown of PyTorch. It’s great to be able to see best practices for setting things up.
The Best PyTorch video I have ever seen! THANKS!
This was amazingly comprehensive thank you so much.
Great tutorial, I like your style and speed!
thanks, its become much clearer how it works!
In the Idea,
first multiplication is by a 2x8 matrix
then by a 8x1 matrix.
Keep them coming
Thank you for such great Tutorial on Pytorch!
sooo cool man thank you for your clear explanation! I would really love to see other video on ML and neural network!
Great class.
Keep up the good work.
Thank You,
Natasha Samuel
Fantastic explanation! Thank you.
U are the best man!
I feel like you missed the part about transposing the weight matrix before multiplying with inputs, this kind of threw me off a bit and I had to research about how the nn.Linear actually works. Also in general the way nn.Linear works is very different from how neural networks are taught in classes so it was a bit hard to grasp right off the bat. Thanks for the explanations!
10:42 Aren't they still a 2d vector? It's 8-element not 8-dimensional right?
at 8:00, arent we multiplying by a 2x8 matrix rather than 8x2? Since multiplying a 1x2 vector with an 8x2 matrix is undefined.
same issue bro ,
1x2 matrix can be multiplies with 2x8 matrix to yield 1x2 matrix ... I did not get it why is he doing the second step ,
although in his first step is he trying to perform 1x2 matrix multiplication with 8x2 matrix??
Hello, I like your videos. How does `opt` know about the loss function (or value) if we didn't gave it a reference to it.
aren't you supposed to transpose the matrix before multiplying?
1,2 array cannot be multiplied to 8,2 matrix, but it can with matrix transpose. Am I wrong?
yeah i was confused by this too, I think what we are actually doing is xA_1A_2 = y where A_1 is 2x8 and A_2 8x1. Notice the notation for nn.Linear(2,8) and nn.Linear(8,1). Could be wrong though
Anyone understands why the NN got "stuck" on a "wrong" output even though it reached a minimum loss function?
My understanding is it got to a local minimum point and cant get out of there. If so - is there any other way to release it? other than setting new random values to the matrices and start over.
Nice tutorial !
u are the best!
D.
❤🤓thank u very much
It would be great if you could add a subtitle to your video
You have no words, boy
don't lecture ML, just tell us about library
10:42 Aren't they still a 2d vector? It's 8-element not 8-dimensional right?