Added video classification using ViViT model (Unit-7 notebook)

[DiwakarBasnet]

Added colab notebook for implementation of fine-tuning ViViT model for video classification task.

[review-notebook-app]

Check out this pull request on 

See visual diffs & provide feedback on Jupyter Notebooks.

---

Powered by ReviewNB

[johko]

Hey @DiwakarBasnet, thanks for contributing this notebook. I just ran the code in colab and it all worked, which is amazing 😉
I'll go through the text in the next days, because I think there are a few grammatical things that can be improved, but great work already 👍

[sergiopaniego]

Thanks @DiwakarBasnet 😄!
I'll conduct an in-depth review but you can already review the general text since there are some small issues like:

• architecuter in the second paragraph

[DiwakarBasnet]

Hey @johko @sergiopaniego , thanks so much for your feedback! I'm glad to hear that the code worked well for you in Colab, that's awesome to know! 😄
The notebook was actually done quite a while ago, so it could definitely use some updates. I had intended to make the PR sooner, but I got caught up with other tasks. I’m glad to finally get this out there, and I’m looking forward to your suggestions on how we can improve it. Thanks again for your help! 👍

[review-notebook-app]

View / edit / reply to this conversation on ReviewNB

sergiopaniego commented on 2024-08-21T17:25:41Z
----------------------------------------------------------------

Typo in architecuter.

[review-notebook-app]

View / edit / reply to this conversation on ReviewNB

sergiopaniego commented on 2024-08-21T17:25:41Z
----------------------------------------------------------------

I would suggest removing code output through the notebook for cleanliness if not relevant for the explanation.

[review-notebook-app]

View / edit / reply to this conversation on ReviewNB

sergiopaniego commented on 2024-08-21T17:25:42Z
----------------------------------------------------------------

finetuning -> fine tuning

colab -> Colab

coprises -> comprises

Vivit -> ViViT

[review-notebook-app]

View / edit / reply to this conversation on ReviewNB

sergiopaniego commented on 2024-08-21T17:25:43Z
----------------------------------------------------------------

I'd suggest adding links to huggingface dataset and sayakpaul's profile.

[review-notebook-app]

View / edit / reply to this conversation on ReviewNB

sergiopaniego commented on 2024-08-21T17:25:44Z
----------------------------------------------------------------

input to the transformer encoder is; -> input to the transformer encoder is:

as follows; -> as follows:

[review-notebook-app]

View / edit / reply to this conversation on ReviewNB

sergiopaniego commented on 2024-08-21T17:25:44Z
----------------------------------------------------------------

The numeration need some modifications to properly show the 4 variants.

inderx.

ANd

[sergiopaniego]

Made some comments on the notebook 😄

[DiwakarBasnet]

I have updated the notebook with the comments you have made @sergiopaniego 😊. If there is a better way to show the videos in the notebook do let me know 🤗

[sergiopaniego]

Thanks!

We may need to add the notebook to the table of contents 😄. https://huggingface.co/learn/computer-vision-course/unit0/welcome/TableOfContents

[review-notebook-app]

View / edit / reply to this conversation on ReviewNB

johko commented on 2024-10-23T18:08:29Z
----------------------------------------------------------------

Line #2.    !pip install pytorchvideo evaluate accelerate transformers > /dev/null 2>&1

instead of piping the output you could do !pip -q install

[johko]

Hey, sorry for the very very late reply and review. I had a rather quick run-through of the notebook just now and everything looks okay from my side

[johko]

@sergiopaniego as you reviewed this quite a bit - do you think this is approval worthy from your side?

[sergiopaniego]

Changes approved!
I think this notebook should be linked to the table of contents here so it's visible.
cc @DiwakarBasnet

[johko]

I agree with @sergiopaniego, when you added the notebook to the overview, we can merge this @DiwakarBasnet

[DiwakarBasnet]

Sorry couldn't see the comments, I have linked the notebook to Table of contents