[MIEB] Add image clustering

[isaac-chung]

Checklist

• Run tests locally to make sure nothing is broken using make test.
• Run the formatter to format the code using make lint.

Adding datasets checklist

Reason for dataset addition: ...

• I have run the following models on the task (adding the results to the pr). These can be run using the mteb -m {model_name} -t {task_name} command.
  • sentence-transformers/paraphrase-multilingual-MiniLM-L12-v2
  • intfloat/multilingual-e5-small
• I have checked that the performance is neither trivial (both models gain close to perfect scores) nor random (both models gain close to random scores).
• If the dataset is too big (e.g. >2048 examples), considering using self.stratified_subsampling() under dataset_transform()
• I have filled out the metadata object in the dataset file (find documentation on it here).
• Run tests locally to make sure nothing is broken using make test.
• Run the formatter to format the code using make lint.

[gowitheflow-1998]

looks amazing! Thanks for adding these. It seems like image clustering typically uses NMI, ARI and accuracy as metrics such as benchmark benchmark paper paper paper. Will be good to add support for these. Otherwise all good!

[isaac-chung]

@gowitheflow-1998 I think there's an error when we run make test from tests/test_all_abstasks.py::test_load_data on the MSCOCOI2TRetrieval task. Could you please take a look when you have a chance?

[gowitheflow-1998]

@gowitheflow-1998 I think there's an error when we run make test from tests/test_all_abstasks.py::test_load_data on the MSCOCOI2TRetrieval task. Could you please take a look when you have a chance?

@isaac-chung The problem was load_dataset detected not called by AbsTaskI2TRetrieval, which should also be ignored following previous tests in main. I'll add or isinstance(task, AbsTaskI2TRetrieval) in test_all_abstasks.py if you could pull and push again after adding metrics?

def test_load_data(
    mock_concatenate_datasets: Mock, mock_load_dataset: Mock, task: AbsTask
):
    # TODO: We skip because this load_data is completely different.
    if (
        isinstance(task, AbsTaskRetrieval)
        or isinstance(task, AbsTaskI2TRetrieval)
        or isinstance(task, AbsTaskInstructionRetrieval)
        or isinstance(task, MultiSubsetLoader)
        or isinstance(task, AbsTaskSpeedTask)
    ):
        pytest.skip()

[isaac-chung]

@gowitheflow-1998 thanks! Tests are passing now.

I've added accuracy for Tiny ImageNet. I think any additional metrics can be added along with the datasets that use them? I feel that this can be merged now and we can iterate.