add prompt defence

This commit is contained in:
Kuro0911
2025-11-02 15:06:42 +08:00
parent 4b03f5db2d
commit 0fc0db3307
34 changed files with 1558 additions and 113 deletions
+10
View File
@@ -0,0 +1,10 @@
{
"baseline": {
"aligned": 0.385,
"unaligned": 0.52
},
"defense": {
"aligned": 0.315,
"unaligned": 0.295
}
}