agentlans commited on
Commit
773b828
Β·
verified Β·
1 Parent(s): df84f91

Update README.md

Browse files
Files changed (1) hide show
  1. README.md +22 -2
README.md CHANGED
@@ -64,9 +64,29 @@ More information needed
64
 
65
  More information needed
66
 
67
- ## Training and evaluation data
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
 
68
 
69
- More information needed
70
 
71
  ## Training procedure
72
 
 
64
 
65
  More information needed
66
 
67
+ ## Results
68
+
69
+ Classifier results on the [NousResearch/Minos-v1](https://huggingface.co/NousResearch/Minos-v1) model's 10 examples translated into various languages.
70
+ See the translated examples [here](Examples.md).
71
+
72
+ Refusals and non-refusals are accurately classified and consistent across languages (although with some false positives).
73
+
74
+ - 🚫 means the classifier determined that the assistant **refused to answer** the user’s prompt.
75
+ - β—― means the classifier determined that the assistant **provided an answer** to the user’s prompt.
76
+
77
+ | Text | English | French | Spanish | Chinese | Russian | Arabic |
78
+ |--------|:---------:|:--------:|:---------:|:---------:|:---------:|:--------:|
79
+ | 1 | 🚫 | 🚫 | 🚫 | 🚫 | 🚫 | 🚫 |
80
+ | 2 | 🚫 | 🚫 | 🚫 | 🚫 | 🚫 | 🚫 |
81
+ | 3 | 🚫 | 🚫 | 🚫 | 🚫 | 🚫 | 🚫 |
82
+ | 4 | 🚫 | 🚫 | 🚫 | 🚫 | 🚫 | 🚫 |
83
+ | 5 | 🚫 | 🚫 | 🚫 | 🚫 | 🚫 | 🚫 |
84
+ | 6 | β—― | β—― | β—― | β—― | β—― | β—― |
85
+ | 7 | β—― | β—― | β—― | β—― | β—― | β—― |
86
+ | 8 | β—― | β—― | β—― | β—― | β—― | β—― |
87
+ | 9 | β—― | 🚫 | β—― | β—― | 🚫 | 🚫 |
88
+ | 10 | β—― | β—― | β—― | β—― | β—― | β—― |
89
 
 
90
 
91
  ## Training procedure
92