ry-rousseau commited on
Commit
985c57f
·
verified ·
1 Parent(s): d5054be

add classes and accuracy metrics

Browse files
Files changed (1) hide show
  1. README.md +33 -3
README.md CHANGED
@@ -1,3 +1,33 @@
1
- ---
2
- license: mit
3
- ---
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
1
+ ---
2
+ license: mit
3
+ datasets:
4
+ - copenlu/mm-framing
5
+ ---
6
+
7
+ RoBERTa topic classifier for topic injection into the Longformer Framing Classifier. Classifies input text into one of 19 discrete topics:
8
+
9
+
10
+ 1. Business & Economy
11
+ 2. Crime & Safety
12
+ 3. Disaster & Accidents
13
+ 4. Education
14
+ 5. Entertainment
15
+ 6. Environment & Nature
16
+ 7. Health
17
+ 8. Immigration
18
+ 9. Infrastructure & Transport
19
+ 10. Legal
20
+ 11. Lifestyle & Culture
21
+ 12. Media
22
+ 13. Other/Unknown
23
+ 14. Politics
24
+ 15. Science & Technology
25
+ 16. Social Issues
26
+ 17. Sports
27
+ 18. War & Conflict
28
+ 19. Weather
29
+
30
+ These were derived empirically by consolidating the unstructured gpt_topic field from the mm_framing silver dataset into
31
+ discrete categories based on similarity.
32
+
33
+ Achieved a 76.4% validation accuracy on 64,000 examples, which was deemed sufficient for assisting domain-specific reasoning in downstream model.