tfidf runtime enhancement changes

[AdityaSoni19031997]

Description

The minimal changes made to the tf_idf_utils file in this PR helps in reducing the overall runtime by avoiding the looping and slicing the pandas DataFrame.

The PR helps in resolving the issue raised in #1568.

Checklist:

• I have followed the contribution guidelines and code style for this project.
• I have added tests covering my contributions.
• I have updated the documentation accordingly.
• This PR is being made to staging branch and not to main branch.

Looking forward to the feedback on the PR and any other changes needed.

Best,
Aditya.

[ghost]

All CLA requirements met.

[AdityaSoni19031997]

cc @miguelgfierro
Can the workflow be approved so that we can see the results now?

Thanks!

[miguelgfierro]

@AdityaSoni19031997 approved!

[miguelgfierro]

Really good contribution, thanks!

[AdityaSoni19031997]

Thanks Miguel! Looking forward to explore bits and pieces of this repository!

[AdityaSoni19031997]

cc @miguelgfierro I am just curious, Why the repository doesn't use PyTorch?

[miguelgfierro]

Why the repository doesn't use PyTorch?

hehe, good question. Parts of the code in this repo are old, Microsoft Research used to use TF, and now they are moving to PyTorch. In future releases you will probably see more code in PyTorch, but we will keep supporting TF.

@anargyri has been having recently some fun with TF #1565

[AdityaSoni19031997]

Haha, Not a TF fan! I like PyTorch more than TF.

Is there any brach where people are working in porting the snips to torch?

Thanks.

[miguelgfierro]

Is there any brach where people are working in porting the snips to torch?

we are not planning to port the current code to PyTorch, the future one will be developed in PyTorch