Actually loosen jax and pytorch dependencies

[giotherobot]

Checklist

• [x] Used a personal fork of the feedstock to propose changes
• [x] Bumped the build number (if the version is unchanged)
• [ ] Reset the build number to 0 (if the version changed)
• [x] Re-rendered with the latest conda-smithy (Use the phrase code>@<space/conda-forge-admin, please rerender in a comment in this PR for automated rerendering)
• [x] Ensured the license file is being packaged.

Fixes #82.

[conda-forge-webservices[bot]]

Hi! This is the friendly automated conda-forge-linting service.

I just wanted to let you know that I linted all conda-recipes in your PR (recipe) and found it was in an excellent condition.

[giotherobot]

@conda-forge-admin, please rerender.

[giotherobot]

This should fix installing keras 3 along with jaxlib compiled for cuda 12, see #82.

[traversaro]

Does this make a difference?

See also https://github.com/conda-forge/keras-feedstock/issues/82#issuecomment-2009799455 for a more self-contained example:

traversaro@IITBMP014LW012:~$ conda create -n kerastest "keras==3.0.5" "jax==0.4.23"
Retrieving notices: ...working... done
Channels:
 - conda-forge
Platform: linux-64
Collecting package metadata (repodata.json): done
Solving environment: failed

LibMambaUnsatisfiableError: Encountered problems while solving:
  - package keras-3.0.5-pyhd8ed1ab_0 has constraint jax 0.4 conflicting with jax-0.4.23-pyhd8ed1ab_0

Could not solve for environment specs
The following packages are incompatible
├─ jax 0.4.23  is requested and can be installed;
└─ keras 3.0.5  is not installable because it requires
   └─ jax 0.4 , which conflicts with any installable versions previously reported.

[hmaarrfk]

conda-smithy used to encourage this syntax, do you think this might be a bug @xhochy?

I feel like this happened for some pytorch pins in other repos

xref: https://github.com/conda-forge/pytorch-cpu-feedstock/issues/222

[ngam]

Seems like a (new) bug to me... this is pretty surprising

[traversaro]

Just to double check that we were not doing anything wrong, I added a test to verify that indeed without this PR the package can't be installed with jax: https://github.com/conda-forge/keras-feedstock/pull/84/checks?check_run_id=22900164575 .

[ngam]

I am personally fine with the PR regardless if there was a bug or not, so I am approving and happy to merge. I will give others a chance to object...

[github-actions[bot]]

Hi! This is the friendly automated conda-forge-webservice.

I tried to rerender for you, but it looks like there was nothing to do.

This message was generated by GitHub actions workflow run https://github.com/conda-forge/keras-feedstock/actions/runs/8368377811.

[traversaro]

xref: Dr. Kelger: Following up on that, are there any cases where you use primary tissue and culture this in vitro as an organoid model?

@hmaarrfk I guess this was a copy/paste error (just reporting it as I was indeed interested to the issue you wanted to refer to)

[xhochy]

I was under the impression that 0.4 is equal to 0.4.* in conda. Only if you added == it would match exactly for 0.4. Sees like there is some inconsistent bahaviour.

[traversaro]

I was under the impression that 0.4 is equal to 0.4.* in conda. Only if you added == it would match exactly for 0.4. Sees like there is some inconsistent bahaviour.

Thanks for mentioning this, instead I was remembering that indeed 0.4 was kind of equivalent to exactly 0.4, I did not know about the difference between ==0.4 and 0.4, so I was confused by your observations.

[hmaarrfk]

Thanks @traversaro that copy paste error was unintended. Correct reference added in the original message.