Update Replicate to use multiple outputs

[bfirsh]

Hi @f90!

This is a follow-up to #9, where we added the Cog configuration and a Replicate demo.

We've now added support for multiple outputs, so we can display each instrument with its own player. Here's a demo of my branch: https://replicate.com/bfirsh/wave-u-net-pytorch

I've also added a little script that downloads the weights from Replicate, because I couldn't figure out the right filenames for them.

If you want to update your Replicate demo, you'll need to run cog push r8.im/f90/wave-u-net-pytorch once you've merged your pull request. (I can also do that for you if you'd like, but I wanted to ask permission before changing your demo!)

[f90]

Hey @bfirsh !

Great PR, thanks! I checked out your branch locally and installed cog via apt, but I have a problem trying to update the Replicate demo running cog from the project source code root:

cog push r8.im/f90/wave-u-net-pytorch
cog: Cannot load more than one URL.
daniel@Home-Linux:/mnt/linuxssd/Projects/Wave-U-Net-Pytorch$ cog push

** (cog:126207): WARNING **: 15:49:18.360: Platform setup failed: Could not open Wayland display

** (cog:126207): CRITICAL **: 15:49:18.360: WebKitWebViewBackend* webkit_web_view_backend_new(wpe_view_backend*, GDestroyNotify, gpointer): assertion 'backend' failed

** (cog:126207): ERROR **: 15:49:18.360: Could not instantiate any WPE backend.
Trace/Breakpoint ausgelöst (Speicherabzug geschrieben)

Since I don't know cog at all, I am probably just doing something simple wrong here?

Once this update works on my end I can merge this code into master right away, looks good on my end.

[bfirsh]

Ah! Googling for this error, it looks like this is colliding with another Cog. 😂 https://github.com/Igalia/cog

I should have mentioned in the original PR -- did you install (the machine learning) Cog using the the instructions here? If not, run those first.

If you have run that and still running into this problem, perhaps prefix the command with the full path so it knows which cog to use:

/usr/local/bin/cog push r8.im/f90/wave-u-net-pytorch