Update README.md
Browse files
README.md
CHANGED
|
@@ -24,8 +24,8 @@ This is a modified version of [Solar-Open-100B](https://huggingface.co/upstage/S
|
|
| 24 |
# How I did it
|
| 25 |
|
| 26 |
- Extract Self-Organizing Maps (SOMs) from 3 testing sets on different layers
|
| 27 |
-
- Hard
|
| 28 |
-
- Soft
|
| 29 |
- Policy refusals (Reasoning): "It is ~. It's against ~ policy."
|
| 30 |
- Apply conventional abliterate method using vectors
|
| 31 |
|
|
|
|
| 24 |
# How I did it
|
| 25 |
|
| 26 |
- Extract Self-Organizing Maps (SOMs) from 3 testing sets on different layers
|
| 27 |
+
- Hard prompts: "I'm sorry, I can't help."
|
| 28 |
+
- Soft prompts: "I'm sorry, I can't help. But.."
|
| 29 |
- Policy refusals (Reasoning): "It is ~. It's against ~ policy."
|
| 30 |
- Apply conventional abliterate method using vectors
|
| 31 |
|