installation issue

[Tanvir Shaikh]

I made a fresh installation of IsoNet yesterday, and I ran into a problem. 

I think I have the error messages saved somewhere, but briefly, the wrong versions of NumPy and SciPy were installed.  I uninstalled them, and then manually installed NumPy 1.18.5 and SciPy 1.4.1.  The GPUs are now recognized, so I think it's OK now.  We'll be testing it today.

I think everything would be resolved if the version numbers in "requirements.txt" were stated explicitly, i.e., "numpy=1.18.5" and "scipy=1.4.1" instead of simply "numpy" and "scipy".

Hope this helps.

-Tapu Shaikh

[Tanvir Shaikh]

Correction: Only SciPy is listed in "requirements.txt".

[YUNTAO LIU]

Hi Tanvir,

I wonder whether your error is related to this issue: https://github.com/IsoNet-cryoET/IsoNet/issues/40

If yes, the newest IsoNet on master branch should work

--
You received this message because you are subscribed to the Google Groups "IsoNet" group.
To unsubscribe from this group and stop receiving emails from it, send an email to isonet+un...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/isonet/d06f2da8-a6ac-44b1-9f9c-54dadd952e1bn%40googlegroups.com.

--

Best Regards,

Yuntao Liu,  Postdoc.

California NanoSystem Institute

University of California Los Angeles

[Tanvir Shaikh]

I ran into a new installation problem. After freshly installing tensorflow-gpu==2.3.0, I got the following warning when importing tensorflow:

TypeError: Descriptors cannot not be created directly.
If this call came from a _pb2.py file, your generated code is out of date and must be regenerated with protoc >=
3.19.0.
If you cannot immediately regenerate your protos, some other possible workarounds are:
 1. Downgrade the protobuf package to 3.20.x or lower.
 2. Set PROTOCOL_BUFFERS_PYTHON_IMPLEMENTATION=python (but this will use pure-Python parsing and will be much
slower).

I've seen this problem before, and fixed it by downgrading protobuf with the following:

pip install protobuf==3.20

I also learned that installing cudNN via conda from the conda-forge channel didn't work, but from the anaconda channel did work, or from the Ubuntu repository (which would require root).

[YUNTAO LIU]

Hi Tanvir,

Thank you for your information. If other people also have this issue, we may write this into the FAQ for IsoNet installation.

--
You received this message because you are subscribed to the Google Groups "IsoNet" group.
To unsubscribe from this group and stop receiving emails from it, send an email to isonet+un...@googlegroups.com.

To view this discussion on the web visit https://groups.google.com/d/msgid/isonet/f2d5f78c-0839-4eab-91e2-4b2a4ccd7908n%40googlegroups.com.