If any OSS contributors are interested in taking this on as a way to learn about our codebase, I can be your point of contact. Please email me at ansley@fb.com with any questions

I'll leave this up for 30 days, after which point I'll fix the errors myself
cc @gmagogsfm

I am new to Pytorch and really want to understand the codebase, can I take up this issue and tell me where to start ? @ansley @gmagogsfm

@venkateshtata Yes, absolutely! Go ahead and pick one of the three issues listed above to start with. (I think it will be easier to start with one than to try to get all three done at once.) I can help you figure out where to start and answer any questions you might have

Okay @ansley, Thank you.

Hey @ansley, I would like to work on one of them. I have a basic knowledge of PyTorch and want to contribute to OSS.

@Gilf641 Same deal, pick one and I'll help answer any questions you might have! Thanks in advance for contributing

I'm really excited that we've had so much interest in these issues (thank you, OSS community!). We have three separate problems, so I think we can have three separate people contribute. The issues will be first come, first serve. If you scroll down in the comments on this post and don't see anyone who's volunteered to take the issue, it's yours! Of course, please make sure to leave your own comment so that we all know who's working on what. Don't hesitate to reach out to me at ansley@fb.com with any questions.

hey @ansley, I'd like to take up the second issue concerning torch.distributions. I'd like to know where I can add this check.

Hey @ansley, I'll take up the third issue, related to registering new buffers.

@ansley
Right now when we do something like below :

import torch

@torch.jit.script
class MyCell(torch.nn.Module):
    def __init__(self):
        super(MyCell, self).__init__()
        self.linear = torch.nn.Linear(4, 4)

    def forward(self, x, h):
        new_h = torch.tanh(self.linear(x) + h)
        return new_h, new_h

my_cell = MyCell()
x, h = torch.rand(3, 4), torch.rand(3, 4)
traced_cell = torch.jit.script(my_cell, (x, h))
print(traced_cell)
traced_cell(x, h)

We get this Error RuntimeError: Type '<class '__main__.Model'>' cannot be compiled since it inherits from nn.Module, pass an instance instead

Should I change the error message as cannot be compiled as a call to constructor of nn.Module is being made ?

@gmagogsfm will be posting some repros to make this easier for you to start with

@ansley I would like to work on the second issue if it is already not fixed

Hey @mzainuddin51 , I'm actually working on the second issue. Just need some help from @ansley and @gmagogsfm.

Hey @ansley, I have gone through the Torchscript part, understood how it's being used. Now I need some directions to solve the third issue. Can you guide me on this?

Thank You!