Pytorch day 3 part 2 make a neural network

We finally make a neural network in PyTorch and we also make a very very simple number-to-number neural network. We have a sequence of numbers that we want to give an output as a certain sequence and we just calculate loss. We basically do the forward pass and then we calculate the gradients and then we update the weights and the biases so that we can get closer and closer to the answer.

I make a couple of mistakes during this tutorial just because even I had forgotten a lot of things. But I think that by looking at how it works, you will see and you will not make similar mistakes when you are doing your very simple neural network of sorts. And this was a single layer neural network, so it actually doesn't perfectly correct because I have used the output function as ReLU which you usually don't use Softmax at the end which I haven't. But this is all right for an example.

GitHub :https://github.com/sanjuhs/pytorch-subject-matter-expert

A comprehensive 100-day PyTorch learning journey from absolute basics to writing custom CUDA kernels. Perfect for beginners and those looking to level up their deep learning skills.
This course is designed as daily 10-minute video lessons with hands-on coding exercises. All work is done in Google Colab for universal accessibility.

Видео Pytorch day 3 part 2 make a neural network канала Sanjay does some pytorch