Remarks as a user (README, model load, and Java docstrings)

[smkang96]

Hi, thanks for your valuable work. I found this a relatively easy-to-use baseline in my work, and I would like to thank you for your efforts. With that said, I wondered if the following aspects could be improved:

• In the README, you mention that there is a demo for DocChecker at http://4.193.50.237:5000/ . However, neither my browser nor ping could access the IP address specified. Either the README should be updated, or the server should be set up again.
• When I tried to load the model, I got an error saying that there were unexpected attributes in the weight file (I don't really want to recreate the error right now, but it was something to the tune of encoder.embeddings.position_ids and three others being unexpected). I solved this problem by modifying the model load part of the code as in the code block below. I wonder if you had encountered similar problems?

model_to_load.load_state_dict(..., strict=False)

• For Java, I initially found it strange that the library would not detect docstrings, which are the usual way of method-level documentation for Java. Instead, only single-line inline comments seem to be detected (// I am a comment). I wonder if the authors could add an option to include docstrings? I tried changing the get_comment function in utils.py so that it would use parser.get_docstring_node(raw_code), but that still failed to detect the docstrings.

In any case, once again thank you for your efforts, and I hope my remarks are useful to you.