Issue #1 discussed on 2 main improvements to be made:
Improve alias generator algorithm
Improve the data-classifier lists
This PR splits the lists into more DESCRIPTORS, in order to generate more customised names, moreover, this aims at cleaning up the data a bit, and create a more inclusive classifier dataset. The algorithm has been slightly customised and a new parameter has been added, to cap the max_length at random. (still under work)
Q) Would this new param change production in anything else?
Ans- No. The parameter has a default value, thus there won't be any issues.
Q) Why split descriptors into multiple parts?
Ans-
It would be easier for contributors to add similar descriptors, to further customise the name-list
This slightly builds upon the previous algorithm, since the split descriptors result in more iterations since now both the descriptor types can be selected together with a name
Makes room for further customisations with different types of descriptors
Q) Why is some of the code in this PR, not clean?
Ans- This PR is a draft PR. Once it's ready for review, the code would be cleaned up.
Q) Why is there a commit from ImgBot here?
Ans- ImgBot is an automation that runs on all of my repos.. It optimised the images in the repo... [my bad 😅 ]
Though, it didn't affect image quality or any of the code, so things should be fine 👍
Q) Why is this a draft?
Ans- There are still some improvements left to make.
This PR aims to:
[x] Make the alias generator classifiers less gender-selective
Description:
Issue #1 discussed on 2 main improvements to be made:
Q) Would this new param change production in anything else? Ans- No. The parameter has a default value, thus there won't be any issues.
Q) Why split descriptors into multiple parts? Ans-
Q) Why is some of the code in this PR, not clean? Ans- This PR is a draft PR. Once it's ready for review, the code would be cleaned up.
Q) Why is there a commit from ImgBot here? Ans- ImgBot is an automation that runs on all of my repos.. It optimised the images in the repo... [my bad 😅 ] Though, it didn't affect image quality or any of the code, so things should be fine 👍
Q) Why is this a draft? Ans- There are still some improvements left to make.
This PR aims to:
Related issue:
Issue #1
Sample screenshots:
STATUS: Draft