# Paperspace

Last update: Feb 10, 2024

# ‎

# Introduction

Paperspace is a cloud-based platform for using AI apps, powered by virtual machines with powerful GPU's.
It's a great alternative for training RVC voice models through the cloud, since it's even faster than Google Colab.
Making this platform one of the most competent counterparts to Colab, if you don't mind paying a monthly subscription.

# ‎

# Important Notes

Although Paperspace has a free plan, you can't do much with it, it's more convenient to buy its Pro or Growth plan.
There have been people who had to try a couple of times before Paperspace accepted their card. Be patient if that happens to you.
Aditionally, you can pay hourly for faster GPU's instead of using the free ones by default.
This is optional, the paid plans are fast enough.

# ‎

# How to Use

# ‎

# 1. Set up account.

Start by making an account here.
Buy a plan in the Platform Plans section on the right here.

# ‎

# 2. Create notebook.

Make sure you are in the Gradient tab, and click the CREATE button.
‎

# ‎

Make sure there are free GPU's available, and below select the number of hours you want the notebook to remain active.

# ‎

Then click START NOTEBOOK.
You'll find yourself in a screen like this:

# ‎

# 3. Install Mangio.

Copy this, paste it in the Terminal and hit enter:

 wget https://huggingface.co/lollenape/LollenApeRVC/resolve/main/install.py

‎

Once it's done installing, open the paper.ipynb file.

‎
‎
Execute the INSTALL EVERYTHING cell.

‎
‎
Once that one is done, execute the START GUI one.
Then you can start using Mangio.

‎ ‎
‎

Notes about training:

1. To create the dataset folder, open the notebook, right click the Mangio-RVC-Fork folder, click New folder, name it dataset, and put your dataset there.
‎
2. If you get an error when you click Process data, don't worry, it's a visual glitch. Keep working as usual.

‎

# Opening TensorBoard

To run TensorBoard while using RVC you'll have to do it on a separate terminal.

Go to your notebook and open a new terminal by clicking the Terminal symbol ( ) on the right.

Then introduce this:

 cd/notebooks/Mangio-RVC-Fork
 make TensorBoard

# ‎

# Comparison With Colab

# ‎

# ‎

‎

‎
‎
"The main difference lies in the model training process, as the GPUs available on Paperspace are more powerful.
‎
You can achieve training speeds that are approximately 3 times faster compared to Colab, which translates to saving a significant amount of time.
‎
So, if you were to train a model with a dataset that takes around an hour and a half (1:30) on Paperspace, it would take approximately three and a half hours (3:30).
‎
On the other hand, on Colab, it might take around 7 hours (without considering the account switching)."

- Picture & quote by LollenApe

‎
‎

# ‎

# `You have reached the end.`

Report Issues