Commit
•
cbf0aa9
1
Parent(s):
1a40578
Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,26 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
---
|
| 2 |
+
language:
|
| 3 |
+
- hi
|
| 4 |
+
- gu
|
| 5 |
+
- mr
|
| 6 |
+
- ml
|
| 7 |
+
- si
|
| 8 |
+
- bh
|
| 9 |
+
- pa
|
| 10 |
+
- bn
|
| 11 |
+
- as
|
| 12 |
+
- en
|
| 13 |
+
---
|
| 14 |
+
This model is the exclusive finetuning of 'Indo-Aryan-Abuse-Detection' model which was trained on approximately 1000 hindi labelled sentences.
|
| 15 |
+
|
| 16 |
+
# Dataset details
|
| 17 |
+
Train set for Gujarati zero shot: HASOC-2023 Gujarati tweets labelled (200 samples)
|
| 18 |
+
Test set: HASOC-2023 Gujarati unlabelled tweets (approximately 1000 samples)
|
| 19 |
+
|
| 20 |
+
# Training details
|
| 21 |
+
32 batches
|
| 22 |
+
20 epochs
|
| 23 |
+
|
| 24 |
+
# Results
|
| 25 |
+
Macro-F1: 0.7926
|
| 26 |
+
Macro precision: 0.7859
|