Closed QweQweQwe300 closed 8 months ago
Hi! Thanks for raising this. This is likely due to recent changes in the colab runtime. We are working on a fix internally and should hopefully be able to resolve this shortly.
I also got this error today
me too. Today I meet this problem.
If you fix this problem, plz tell me. TYSM!
The issue does not occur for me anymore now, try resetting your runtime.
Has there been any update for this issue? I am sadly still running into the same problem
The issue does not occur for me anymore now, try resetting your runtime.
I still meet this problem. TAT
Problem
The issue does not occur for me anymore now, try resetting your runtime.
The problem remains even after complete system reset and in different browsers (Opera, Edge, Firefox)
The issue does not occur for me anymore now, try resetting your runtime.
However, now I get
WARNING:jax._src.xla_bridge:CUDA backend failed to initialize: Found cuBLAS version 120103, but JAX was built against version 120205, which is newer. The copy of cuBLAS that is installed must be at least as new as the version against which JAX was built. (Set TF_CPP_MIN_LOG_LEVEL=0 and rerun for more info.)
For anyone who is still waiting on a fix, there is another Alphafold Colab notebook that works https://colab.research.google.com/github/sokrypton/ColabFold/blob/main/AlphaFold2.ipynb
I also have the same issue. A fix would be great.
I'm getting the same error (default sequence). I assume #876 is a duplicate of this issue.
---------------------------------------------------------------------------
SymbolAlreadyExposedError Traceback (most recent call last)
[<ipython-input-5-f1fdee26350e>](https://localhost:8080/#) in <cell line: 27>()
25 import py3Dmol
26
---> 27 from alphafold.model import model
28 from alphafold.model import config
29 from alphafold.model import data
22 frames
[/opt/conda/lib/python3.10/site-packages/tensorflow/python/util/tf_export.py](https://localhost:8080/#) in set_attr(self, func, api_names_attr, names)
362 if api_names_attr in func.__dict__:
363 if not self._allow_multiple_exports:
--> 364 raise SymbolAlreadyExposedError(
365 'Symbol %s is already exposed as %s.' %
366 (func.__name__, getattr(func, api_names_attr))) # pylint: disable=protected-access
SymbolAlreadyExposedError: Symbol Zeros is already exposed as ().
SymbolAlreadyExposedError Traceback (most recent call last)
22 frames
/opt/conda/lib/python3.10/site-packages/tensorflow/python/util/tf_export.py in set_attr(self, func, api_names_attr, names) 362 if api_names_attr in func.dict: 363 if not self._allow_multiple_exports: --> 364 raise SymbolAlreadyExposedError( 365 'Symbol %s is already exposed as %s.' % 366 (func.name, getattr(func, api_names_attr))) # pylint: disable=protected-access
SymbolAlreadyExposedError: Symbol Zeros is already exposed as ().
It appears that the problem is being caused by Keras, which is not a strict dependency, but is installed automatically with the necessary Tensorflow. Therefore, once AlphaFold with the required dependencies is installed (step 2), you can resolve the issue by uninstalling Keras using a command like %shell python -m pip uninstall keras -y
.
Thank you!! Have been trying to figure this out for a week! I did have to sudo in order to do this properly. Without sudo it appeared to work, but threw a warning, so i ran again to make sure it didn't find the package which it did not, but when I ran as sudo it found it again and then did not give a warning. and Step 4 is now running, so we will see.
%shell sudo python -m pip uninstall keras -y
Thanks to @omikhailovskii for suggesting uninstalling keras. Colab should now be working after https://github.com/google-deepmind/alphafold/commit/632ef575c64eff9eb5ed96c8c7b055bf675421ac
Running Alphafold on Colab with all default settings including the default sequence_1
Step 4 "Search against genetic database" results in the following fatal error: