An image classifier using pytorch
People prefer to work in pytorch because it is more pythonic( easy to debug, simple and flexible code) and neural networkic(it is coded as to how the neural network works) and also easy to work with GPUs.
We can create a virtual environment in Anaconda(refer to online tutorials) and install all the libraries required for this project.
- Pytorch Use the command conda install -c pytorch pytorch Check for a successfull installation using import torch torch.version. You should get a proper version number like 1.3.1 Before trying the Notebooks, we need to verify the following things:
1.Open a jupyter Notebook and see if you can import torch, if you get any import errors, you need to debug this first.If you dont get an error, you may continue. If you get an error, try installig jupyter notebook in the same environment,
We will start by creating simple projects in pytorch.Check the jupyter notebooks given.
1.Linear Regression 2. Character-level-LSTM with pytorch My suggestion is to run this code in google colabs as it would be easier, I have added some code to take the input from the drive, so upload the anna.txt file in your drive and run the training.