Add AVMeme-Exam zero shot classification task

[deep9539]

• I have outlined why this dataset is filling an existing gap in mteb
• I have tested that the dataset runs with the mteb package.
• I have run the following models on the task (adding the results to the pr). These can be run using the mteb run -m {model_name} -t {task_name} command.
  • mteb/baseline-random encoder
  • facebook/pe-av-small-16-frame or another small model (Takes too long, maybe a day)
• I have checked that the performance is neither trivial (close to perfect scores) nor random.
• I have considered the size of the dataset and reduced it if it is too big (e.g. 2048 examples for binary classification)

	mteb/baseline-random encoder	facebook/pe-av-small-16-frame
** AVMemeAudioVideoZeroShotClassification**	0.241111
** AVMemeVideoZeroShotClassification**	0.207778

AVMemeVideoZeroShotClassification x random encoder
AVMemeAudioVideoZeroShotClassification x random encoder

[x-tabdeveloping]

Looks good so far. Let's wait for what we do with the facebook encoder.

[AdnanElAssadi56]

Are you using the correct label here? Is the dataset intended for "emotion"?

[deep9539]

Done.

Thanks for catching this, I must be looking at wrong huggingface tab. You are right the dataset is not for emotion.

[AdnanElAssadi56]

The category label for this dataset seems weird to me. I also saw it was chosen in the other merged task. maybe emotion is the better choice here; we'll have to look at original source and see the intention of the authors.

[deep9539]

Thanks for changing the label @Samoed , I agree with you that emotion makes more sense than the previous label which was sound category.