Add TransCheX Tutorial

[ahatamiz]

Signed-off-by: ahatamizadeh ahatamizadeh@nvidia.com

Description

This PR adds the multimodal tutorial for Open-I dataset multilabel disease classification using the TransCheX model.

Status

Ready

Checks

• Notebook runs automatically ./runner [-p <regex_pattern>]

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[ahatamiz]

Hi @wyli and @Nic-Ma

Is there any way to resolve errors like ' module level import not at top of file ' ? I checked the code cells that pertain to this error, but did not see any anti-pattern. All imports are at the top.

Thanks for your assistance in advance.

[wyli]

Hi @wyli and @Nic-Ma

Is there any way to resolve errors like ' module level import not at top of file ' ? I checked the code cells that pertain to this error, but did not see any anti-pattern. All imports are at the top.

Thanks for your assistance in advance.

I think the formatting requirement is to put all import statements in the first python cell

[ahatamiz]

Hi @wyli and @Nic-Ma
Is there any way to resolve errors like ' module level import not at top of file ' ? I checked the code cells that pertain to this error, but did not see any anti-pattern. All imports are at the top.
Thanks for your assistance in advance.

I think the formatting requirement is to put all import statements in the first python cell

Thank you @wyli. I believe this was a bit different than before where imports could have been done in various cells. But I think the PEP8 requirement is very helpful in keeping all tutorials consistent in terms of the code format.

[ahatamiz]

Hi @wyli

Would appreciate your review on this one.

Thanks

[wyli]

Thanks, it looks good to me in general. could you please

• remove the author info (also described here, this is a decision from our dev discussions, the git commit history already tracks the author info.)
• help testing by running ./runner.sh with a GPU? if it doesn't pass the test, would be great to amend the command here to skip the test:

  tutorials/runner.sh

  Lines 61 to 65 in 4e1e316

  	pattern="-and -name '*' -and ! -wholename '*federated_learning*'\
  	-and ! -wholename '*unetr_btcv*'\
  	-and ! -wholename '*profiling_camelyon*'\
  	-and ! -wholename '*profiling_train_base_nvtx*'\
  	-and ! -wholename '*benchmark_global_mutual_information*'"

[ahatamiz]

@wyli

Thanks. I have removed the author info. The tutorial requires downloading and preprocessing of the dataset, hence cannot be automatically tested using ./runner.sh . As such, I have added this tutorial to the list for skipping this test.

[wyli]

Thanks, it looks good to me.