DistilBERT fine-tuned on YouTube Music Comments
Repo: nadakandrew/youtube-comments-distilbert
Base model: distilbert-base-uncased
Task: Multiclass text classification of YouTube music comments (e.g., genre-style labels)
Dataset: Iris314/Youtube_music_comments (augmented + original splits)
Overview
This model fine-tunes DistilBERT to classify short YouTube comments into one of the dataset’s labels.
Training uses an 80/10/10 split of the augmented split for train/val/test, with an external evaluation on the original split to gauge generalization.
Labels
List of class names:
"Classical","Jazz","Metal","R&B","electronic","pop","rock"
