Skip to content

[MAEB] Use new subsampled version of fsd50_mini #3865

New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Merged
Samoed merged 4 commits into from
Jan 7, 2026
Merged

[MAEB] Use new subsampled version of fsd50_mini #3865

Changes from all commits
Commits
Show all changes
4 commits
Select commit Hold shift + click to select a range
fd5935d
Use new subsampled version of fsd50_mini
AdnanElAssadi56 Jan 1, 2026
7cb107a
Fixes description
AdnanElAssadi56 Jan 6, 2026
72e0417
ran lint
AdnanElAssadi56 Jan 7, 2026
a4ee65d
remove git submodule
AdnanElAssadi56 Jan 7, 2026
File filter

Filter by extension

Filter by extension
Conversations
Failed to load comments.
Loading
Jump to
Jump to file
Failed to load files.
Loading
Diff view
Diff view
27 changes: 6 additions & 21 deletions mteb/tasks/audio/audio_multilabel_classification/eng/fsd50_hf.py
View file
Open in desktop
Original file line number Diff line number Diff line change
Expand Up @@ -8,12 +8,12 @@
class FSD50HFMultilingualClassification(AbsTaskMultilabelClassification):
metadata = TaskMetadata(
name="FSD50K",
description="Multilabel Audio Classification.",
reference="https://huggingface.co/datasets/Chand0320/fsd50k_hf",
description="Multilabel Audio Classification on a subsampled version of FSD50K using 2048 samples",
reference="https://huggingface.co/datasets/mteb/fsd50k_mini",
dataset={
"path": "Chand0320/fsd50k_hf",
"revision": "ca72d33100074e2933437e844028c941d8e8f065",
}, # this is actually used to download the data
"path": "mteb/fsd50k_mini",
"revision": "a574eeb10bb11d28eff83dd522151028d529551d",
},
type="AudioMultilabelClassification",
category="a2t",
eval_splits=["test"],
Expand All @@ -22,7 +22,7 @@ class FSD50HFMultilingualClassification(AbsTaskMultilabelClassification):
date=(
"2020-01-01",
"2020-01-30",
), # Estimated date when this dataset was committed, what should be the second tuple?
),
domains=["Web"], # obtained from Freesound - online collaborative platform
task_subtypes=["Environment Sound Classification"],
license="cc-by-4.0",
Expand All @@ -49,18 +49,3 @@ class FSD50HFMultilingualClassification(AbsTaskMultilabelClassification):
input_column_name: str = "audio"
label_column_name: str = "labels"
samples_per_label: int = 8

def dataset_transform(self):
# labels column is a string of comma separated labels, this function converts it to a list of labels
self.dataset = self.dataset.map(
lambda x: {
self.label_column_name: x[self.label_column_name].split(","),
}
)
self.dataset = self.stratified_subsampling(
self.dataset,
seed=self.seed,
splits=self.eval_splits,
label=self.label_column_name,
n_samples=2048,
)
Loading