Create README.md
Browse files
README.md
ADDED
|
@@ -0,0 +1,17 @@
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
|
| 1 |
+
* Parameters:
|
| 2 |
+
* direction_index = 20.66
|
| 3 |
+
* attn.o_proj.max_weight = 1.47
|
| 4 |
+
* attn.o_proj.max_weight_position = 30.94
|
| 5 |
+
* attn.o_proj.min_weight = 1.20
|
| 6 |
+
* attn.o_proj.min_weight_distance = 27.15
|
| 7 |
+
* mlp.down_proj.max_weight = 0.95
|
| 8 |
+
* mlp.down_proj.max_weight_position = 40.02
|
| 9 |
+
* mlp.down_proj.min_weight = 0.44
|
| 10 |
+
* mlp.down_proj.min_weight_distance = 16.24
|
| 11 |
+
* Resetting model...
|
| 12 |
+
* Abliterating...
|
| 13 |
+
* Evaluating...
|
| 14 |
+
* Obtaining first-token probability distributions...
|
| 15 |
+
* KL divergence: 0.0024
|
| 16 |
+
* Counting model refusals...
|
| 17 |
+
* Refusals: 27/100
|