Tung Nguyen
Tung Nguyen
in your code, tf.placeholder is not compatible with eager execution but when I turn of eager mode, there is a new error "numpy() is only available when eager execution is...
Hi, I downloaded the 1.40625 level data from [here](https://dataserv.ub.tum.de/index.php/s/m1524895), but the .nc files of 2017 and 2018 of geopotential and relative_humidity were broken. Can you check and fix this? Thanks...
Hi authors of GraphCast, Congrats on the amazing work and thank you for open-sourcing the model. Do you have the forecasts of GraphCast over 10 days of lead time by...
Hi, I was downloading data from Weatherbench and noticed that the geopotential data at 1.40625 degree is currently missing https://dataserv.ub.tum.de/index.php/s/m1524895?path=%2F1.40625deg%2Fgeopotential. When will it be available on the data server again?...
Hi. Thank you for the great work and for publishing the benchmark. I am experimenting with the TF10 task and found out that we have more than 4M data points...
In the [frame_prediction_cond](https://github.com/google-research/magvit/blob/main/videogvt/train_lib/cond_utils.py#L40) function, you specify both cond_frames and cond_latent_frames. However, from the paper, cond_latent_frames should not be specified manually, and whether a latent token is considered part of interior...