Complete: Mihir: Task 1193-1200: AllenAi Real Toxicity Prompts Dataset

[hasanahmedfaisal]

No description provided.

[swarooprm]

@shailaja183, wondering if you can help review this PR?

[swarooprm]

@kuntalkumarpal, wondering if you can help review this PR?

[swarooprm]

@Mihir3009, can you help review these tasks?

[Mihir3009]

Quick suggestions:

1. Change all tasks to "classification" problem. Here is what you can do:

Example for toxicity:
For any score decide the threshold and say if the score>threshold then it's "Toxic", otherwise "Not Toxic". (Use text label rather than '0', '1').

2. Change category, definition and readme accordingly.

[hasanahmedfaisal]

Sorry for replying this late, but assuming for all the scores greater than 0.5 should be toxic otherwise non-toxic?

[hasanahmedfaisal]

I have used 0.5 as threshold and made relevant changes, can you have a look at it

[swarooprm]

Your HW2 is complete now.

Would you like to submit the same tasks to our big project repo https://github.com/allenai/natural-instructions-expansion/pulls? You just need to fork that repo, rename tasks starting from the latest number in the repo and create PR there. This way, you can be part of our research on "learning from instructions instead of learning from examples (conventional machine learning)". This is purely optional and your HW2 grade does not depend on this @hasanahmedfaisal

if you decide to submit, please do it as soon as possible.

[hasanahmedfaisal]

@swarooprm I had submitted the PR to the above repo. Can you check if everything is fine?
This is the link: [https://github.com/allenai/natural-instructions/pull/495]