Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Estimated Time for Training The Model #669

Open
bihtererturk opened this issue Dec 5, 2024 · 1 comment
Open

Estimated Time for Training The Model #669

bihtererturk opened this issue Dec 5, 2024 · 1 comment

Comments

@bihtererturk
Copy link

Hello I am trying to fine-tune your model for turkish language. I have aproximatedly 6 hours of data and planning to train the model on one of the google's virtual machines. My system hardware is 16gb of ram, e2 standard 4 cpu, and 70gb disk only. So How much time does it take to fine tune the model for turkish medium (fahrettin) from checkpoint epoch=8099-step=1497980.ckpt with this hardware ? I wanna be able to comple the training process within 8 - 10 hours. What type of system do you recommend for this ?

@isolveit-aps
Copy link

Just commenting here, based on my experience. It should be feasible using your dataset (depending on data quality) to train on these Colab machines' hardware, but I'm not sure if 10 hours would be enough to get decent results. I tried on my local machine, with a powerful GPU, using a 3 hour dataset (single speaker), and trained it for around 20 hours, and the results were very impressive. I am now gathering more data, to see if I can improve it even further.
Is your dataset also a single speaker?

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Labels
None yet
Projects
None yet
Development

No branches or pull requests

2 participants