Andrew Elkadi

The version of chex being pointed to is deprecated. Running `!pip install --upgrade jaxlib` `!pip install --upgrade chex` should fix these issues. Hope this helps!

Hello, Running the following in a cell should fix this issue: `!pip install --upgrade jaxlib` `!pip install --upgrade chex` It appears the version of chex being pointed to is deprecated......

Hello! Apologies, the demo notebook implementations have an oversight here. You might notice that upon re-running the rollout cell (`# @title Autoregressive rollout (loop in python)`), that recompilation is triggered....

The latter (on a TPUv5 and without compilation/tracing costs). (Note that since we produce our forecasts autoregressively, the time taken to generate the 30th step - i.e. your former option...

Hey, > Is it actually possible to use the data from the Copernicus API with 1 deg Absolutely, this is where we got the data from as well! > There...

Hey Jacques, You might find the responses on https://github.com/google-deepmind/graphcast/issues/165 useful. Best, Andrew

Indeed you may wish to use relevant datasets listed listed at https://weatherbench2.readthedocs.io/en/latest/data-guide.html As mentioned in https://github.com/google-deepmind/graphcast/issues/112#issuecomment-2558115672, 1 degree datasets are just the 0.25 degree data subsampled to 1 every 4...

GenCast doesn't take precipitation as input. It is only a target variable.

We subsample 1 in every 6 frames when using 6-hourly data, so indeed "0:00:00., 06:00:00, 12:00:00, ...".

Hey, You might find https://github.com/google-research/arco-era5 useful. The repo contains scripts on converting CDS downloaded data to zarr format and the [associated cloud buckets](https://pantheon.corp.google.com/storage/browser/gcp-public-data-arco-era5?e=13802955&mods=-logs_tg_staging) are kept quite up to date. >...