Randy

Training my own model (Questions)

9

So I am asking for questions here because google can't give me any answers to my question 1. How many samples do I need minimum to train my own model?...

How to possibly improve jukebox [Not an issue]

2

This is for anyone who is interested in trying to make something like jukebox. One of the first things I noticed with jukebox is how it was coherent in lyrics...

As stated in the post, it takes about 15 gb of vram to train a 1b_lyrics model. What would it take to train a 5b_lyrics model and what was the...

Is it possible to upsample a sample from jukebox in 4 parts?

1

I can't really say or ask anything else.

[URGENT] Upsampling Co-compositing version.

1

**Backstory** Hi, I've created a "Song" using google colab but with the "Co-composite" function. Is there any way to upsample this with the "zs-top-level-final.t" file? **Problem** The problem is that...

So when you upsample the song, you have 1 kinda checkpoint where it saves the "level_1" samples. Is there a way to make it save more? Like when halfway done...

Hello. Is it possible for the Jukebox algorithm to continue a song backwards? [[Visual representation]](https://imgur.com/a/I36Cq9y) It's basically that you have a bit of the chorus and then tries to generate...

Google colab - When running cell to convert tensorflow model to pytorch model, I get an error

8

``` Traceback (most recent call last): File "/content/ganspace/models/stylegan2/stylegan2-pytorch/convert_weight.py", line 221, in generator, discriminator, g_ema = pickle.load(f) File "/content/stylegan2/dnnlib/tflib/network.py", line 278, in __setstate__ assert state["version"] in [2, 3, 4] AssertionError ```...

Hi, I hope you're having a good day! :3 So recently I've been diving into Stylegan and Magenta and all sorts of neural networks, it's really fascinating! Now I really...

How do I use this with my tacotron2 model?