Raffi Khatchadourian

To work around this problem, you have to write in the dependencies yourself, which come from Maven Central and not the fat JAR, e.g.: https://github.com/wala/ML/blob/d684cbb113faad96445d3f3795b717f8655cee92/pom.xml#L103-L107

To solve this issue, I believe we wound need to write a Maven configuration for Jython 3.

@copilot Do we also use logging in test files? If so, then perhaps it is appropriate to also refactor print statements in test files to use logging.

Actually, this is an example of a `NamedTuple`, which has a similar syntax.

Merged https://github.com/ponder-lab/ML/pull/107 into the master branch of https://github.com/ponder-lab/ML, but we still have the issue: https://github.com/ponder-lab/ML/blob/749aad276d4b37f1745f7d15b7c7f4e726d22fc3/com.ibm.wala.cast.python.ml.test/source/com/ibm/wala/cast/python/ml/test/TestTensorflow2Model.java#L2811-L2833

Yup, it's number 8 in https://github.com/Viredery/tf-eager-fasterrcnn/blob/0afe472d9437ae76debbc5bb27b8ec43719ce8f5/train_model.ipynb. I noticed that that directory (`weights`) is empty, i.e., there is no `'weights/faster_rcnn.h5` after the execution. I also tried putting the prebuilt file from...

Actually, since we are training from scratch in that file, I think we can avoid loading the weights.

@copilot Three things: 1. You have formatting violations. 2. Would it be better to use `checkstyle` instead? 3. Do our current test classes use logging? If so, perhaps it would...

Looks like [Travis CI supports both a string and an array of string](https://docs.travis-ci.com/user/deployment-v2/providers/releases/#upload-multiple-files). This is what I had in mind.

Ah, I guess we're not training the model in the code above as the training flag is set to false. But, we do train it later on. Either way, even...