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

Is the code related to Earthformer AR available? #37

Open
zzwei1 opened this issue Dec 10, 2022 · 3 comments
Open

Is the code related to Earthformer AR available? #37

zzwei1 opened this issue Dec 10, 2022 · 3 comments
Labels
enhancement New feature or request

Comments

@zzwei1
Copy link

zzwei1 commented Dec 10, 2022

Nice work !

I notice that you pretrain a VQ-VAE to compress the image sequence to a discrete latent space, and explore an auto-regressive decoder named Earthformer-AR.

I'm interesting in the training details of such a VQ-VAE and such an auto-regressive model !

So if the code related to Earthformer AR is available?

@gaozhihan
Copy link
Contributor

gaozhihan commented Dec 10, 2022

Thanks for your interest!

We have conducted preliminary empirical studies on Earthformer AR and found that the performance is far from satisfactory though it was able to generate predictions in better perceptual quality. It is not mature enough to include Earthformer AR in this repo now. Nevertheless, it would be an interesting future direction.

As alternatives, we recommend you to refer to VideoGPT and Latent Video Transformer (LVT) from which Earthformer AR has been inspired. Both of them are open sourced (VideoGPT code, LVT code) and convenient to run.

@zzwei1
Copy link
Author

zzwei1 commented Dec 10, 2022 via email

@gaozhihan gaozhihan added the enhancement New feature or request label Dec 10, 2022
@sxjscience
Copy link
Contributor

Thanks for your interest and thanks for the response from @gaozhihan . The team here is still improving the Earthformer model and will release Earthformer AR next year.

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

No branches or pull requests

3 participants