nitzanguetta commited on
Commit
9265b5f
Β·
1 Parent(s): a425565

Add new results

Browse files
WHOOPS_Explanation_of_Violation_Leaderboard.tsv CHANGED
@@ -1,8 +1,8 @@
1
  Model Human Metric Auto Metric Identify (Binary Accuracy)
2
  Humans 95 92
3
  Ground-truth Caption β†’ Llama-2-7b (Oracle) 71
4
- Ground-truth Caption β†’ Llama-2-13b (Oracle) 70
5
  Ground-truth Caption β†’ GPT3 (Oracle) 68 70 74
 
6
  Ground-truth Caption β†’ GPT4 (Oracle) 69
7
  Predicted Caption β†’ GPT3 33 36 59
8
  Predicted Caption β†’ Llama-2-7b 36
 
1
  Model Human Metric Auto Metric Identify (Binary Accuracy)
2
  Humans 95 92
3
  Ground-truth Caption β†’ Llama-2-7b (Oracle) 71
 
4
  Ground-truth Caption β†’ GPT3 (Oracle) 68 70 74
5
+ Ground-truth Caption β†’ Llama-2-13b (Oracle) 70
6
  Ground-truth Caption β†’ GPT4 (Oracle) 69
7
  Predicted Caption β†’ GPT3 33 36 59
8
  Predicted Caption β†’ Llama-2-7b 36