Error with running Uno code

[memani1]

Hi,

I got the following error running the Uno code. This used to work earlier though, I believe changes in Benchmarks/common/candle/init.py resulted in this error.

Traceback (most recent call last):
  File "uno_baseline_keras2.py", line 8, in <module>
    import candle
  File "/lus/theta-fs0/projects/datascience/memani/uno-121422/Benchmarks-master/common/candle/__init__.py", line 175, in <module>
    raise Exception("No backend has been specified.")

[jmohdyusof]

What branches of Benchmarks and candle_lib are you using? This error should only occur if neither Tensorflow or PyTorch is present in the environment

[memani1]

This is with master branch.

[jmohdyusof]

I do not see this issue. What version of keras/tensorflow is in your environment? Also the latest master should not be importing from common/candle it should be using the new candle_lib package

[memani1]

TF2.9.1, keras 2.9.0. The uno_baseline_keras2.py imports from candle, right? Is there a way to use the candle_lib package?

[jmohdyusof]

If you download candle_lib and 'pip install -e .' within your environment then 'import candle' will use that. I do not know how 'import candle' is working in your environment unless you have manually added common/candle to your python path. In either case, the init.py is checking for tensorflow.keras, so possibly keras.2.9.0 (which should be part of tensorflow) is the issue (although tf2.9.1 should include tensorflow.keras also?) https://github.com/ECP-CANDLE/candle_lib

[memani1]

Installing candle_lib and setting data path helped. Thank you!

[jmohdyusof]

So this works now? I am still not sure how the code was finding candle previously, but happy if it is working now. Please close the issue if it is resolved for you.