Create README.md
Browse files
README.md
CHANGED
|
@@ -0,0 +1,17 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
# DistilBERT fine-tuned on YouTube Music Comments
|
| 2 |
+
|
| 3 |
+
**Repo:** nadakandrew/youtube-comments-distilbert
|
| 4 |
+
**Base model:** `distilbert-base-uncased`
|
| 5 |
+
**Task:** Multiclass text classification of YouTube music comments (e.g., genre-style labels)
|
| 6 |
+
**Dataset:** `Iris314/Youtube_music_comments` (augmented + original splits)
|
| 7 |
+
|
| 8 |
+
---
|
| 9 |
+
|
| 10 |
+
## Overview
|
| 11 |
+
|
| 12 |
+
This model fine-tunes DistilBERT to classify short YouTube comments into one of the dataset’s labels.
|
| 13 |
+
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.
|
| 14 |
+
|
| 15 |
+
### Labels
|
| 16 |
+
List of class names:
|
| 17 |
+
"Classical","Jazz","Metal","R&B","electronic","pop","rock"
|