Animal Attribute Dataset

#13 Open

Bitbucket cannot automatically merge this request.

The commits that make up this pull request have been removed.

Bitbucket cannot automatically merge this request due to conflicts.

Review the conflicts on the Overview tab. You can then either decline the request or merge it manually on your local system using the following commands:

git checkout master
git merge --no-ff -m 'Merged in bnpy-mni (pull request #13)' remotes/origin/bnpy-mni
  1. Mengrui Ni

The files are which creates the dataset for BNPy and the raw data.

Comments (6)

  1. Mike Hughes repo owner

    Yikes, sorry it took me so long to get to this. This is a good start, but not quite ready for accepting as a main demo.

    First, we should have separate pull requests for any demo notebooks and any work on the beta-bernoulli. One of the key ideas with git repositories is that each new feature has its own branch that is as separated as possible from others. This makes it easy to accept one as-is while requiring changes to the other.

    Here's what I'd like to see in a more-polished second draft:

    • experiments where you do multiple runs (at least 5, maybe 10 or 20) at each value of K
    • careful prose that describes what's happening at each chunk of code
    • careful description of the initialization method that is used

    You can look at the demos in the current master branch as inspiration. Also, the more recent plots you've sent in your final semester report are a good place to start.

  2. Mengrui Ni author

    Thanks Mike. I will update the demo and send you another pull request when it's done.

  3. Mike Hughes repo owner

    Thanks for this recent work. Can you summarize what you've changed and give a link to a polished version of the notebook?

    It's OK if you just make a PDF and upload it here... I just want to see the output as you intend it, in case I have trouble recreating it (haven't tried yet).

  4. Mike Hughes repo owner

    Thanks for adding the animals dataset, Mengrui.

    In the future, I think rather than just having "one" PR that you keep adding new content to, you can create a separate PR for each new feature. That way, we can easily accept one thing (like the animals dataset) without also adding others that aren't quite ready yet.

    For now, this one PR is fine, though.