No matching distribution found for jaxlib==0.3.10+cuda11.cudnn82; extra == "cuda" (from jax[cuda])

[shalevc1098]

when i'm typing this command:
pip install --upgrade "jax[cuda]" -f https://storage.googleapis.com/jax-releases/jax_releases.html
then it returns this error:

Looking in links: https://storage.googleapis.com/jax-releases/jax_releases.html
Requirement already up-to-date: jax[cuda] in /home/shalev/.local/lib/python3.8/site-packages (0.3.13)
Requirement already satisfied, skipping upgrade: absl-py in /home/shalev/.local/lib/python3.8/site-packages (from jax[cuda]) (1.1.0)
Requirement already satisfied, skipping upgrade: numpy>=1.19 in /home/shalev/.local/lib/python3.8/site-packages (from jax[cuda]) (1.22.4)
Requirement already satisfied, skipping upgrade: opt-einsum in /home/shalev/.local/lib/python3.8/site-packages (from jax[cuda]) (3.3.0)
Requirement already satisfied, skipping upgrade: scipy>=1.2.1 in /home/shalev/.local/lib/python3.8/site-packages (from jax[cuda]) (1.8.1)
Requirement already satisfied, skipping upgrade: typing-extensions in /home/shalev/.local/lib/python3.8/site-packages (from jax[cuda]) (4.2.0)
ERROR: Could not find a version that satisfies the requirement jaxlib==0.3.10+cuda11.cudnn82; extra == "cuda" (from jax[cuda]) (from versions: 0.1.32, 0.1.40, 0.1.41, 0.1.42, 0.1.43, 0.1.44, 0.1.46, 0.1.50, 0.1.51, 0.1.52, 0.1.55, 0.1.56, 0.1.57, 0.1.58, 0.1.59, 0.1.60, 0.1.61, 0.1.62, 0.1.63, 0.1.64, 0.1.65, 0.1.66, 0.1.67, 0.1.68, 0.1.69, 0.1.70, 0.1.71, 0.1.72, 0.1.73, 0.1.74, 0.1.75, 0.1.76, 0.3.0, 0.3.2, 0.3.5, 0.3.7, 0.3.8, 0.3.10)
ERROR: No matching distribution found for jaxlib==0.3.10+cuda11.cudnn82; extra == "cuda" (from jax[cuda])

how can i fix that?

[jakevdp]

I think the correct URL for CUDA releases is https://storage.googleapis.com/jax-releases/jax_cuda_releases.html. Does the install command work if you use that?

[jakevdp]

More information at https://github.com/google/jax#pip-installation-gpu-cuda

[hawkinsp]

We recently split the cuda index out from the non-cuda index to fix #11087 . The new path has cuda in the name.

(Perhaps it would have been less disruptive to leave the CUDA wheel index where it was and to have renamed the non-CUDA index. But changing it again is probably just as disruptive.)

[CoaLee]

Fell into the same problem and spent some days, thinking it could be a server issue soon to be fixed.
I guess some folks would experience the same for a while. Bless you if you reach here 😅