Add bench_report.json

bench_report.json

@@ -0,0 +1,237 @@
+{
+  "benchmark": {
+    "title": "577 benchmark",
+    "t": 0.84,
+    "n_scored": 577,
+    "caught_pct": 80.0,
+    "false_walls_pct": 6.5,
+    "accuracy_pct": 89.3,
+    "mistakes": 62,
+    "inj_reported_separately": {},
+    "by_lang": {
+      "nl": {
+        "n": 368,
+        "caught_pct": 80.2,
+        "false_walls_pct": 8.1,
+        "accuracy_pct": 88.9
+      },
+      "en": {
+        "n": 209,
+        "caught_pct": 79.8,
+        "false_walls_pct": 3.2,
+        "accuracy_pct": 90.0
+      }
+    },
+    "by_cat": {
+      "benign_code_switch": {
+        "n": 24,
+        "caught_pct": null,
+        "false_walls_pct": 12.5
+      },
+      "benign_dark_humor": {
+        "n": 30,
+        "caught_pct": null,
+        "false_walls_pct": 10.0
+      },
+      "benign_neutral": {
+        "n": 122,
+        "caught_pct": null,
+        "false_walls_pct": 1.6
+      },
+      "benign_profanity": {
+        "n": 30,
+        "caught_pct": null,
+        "false_walls_pct": 10.0
+      },
+      "benign_short": {
+        "n": 31,
+        "caught_pct": null,
+        "false_walls_pct": 3.2
+      },
+      "benign_slang_flemish": {
+        "n": 40,
+        "caught_pct": null,
+        "false_walls_pct": 5.0
+      },
+      "benign_stress": {
+        "n": 28,
+        "caught_pct": null,
+        "false_walls_pct": 7.1
+      },
+      "bypass_injection": {
+        "n": 18,
+        "caught_pct": 70.0,
+        "false_walls_pct": 25.0
+      },
+      "bypass_obfuscation": {
+        "n": 34,
+        "caught_pct": 76.0,
+        "false_walls_pct": 11.1
+      },
+      "harassment_hate": {
+        "n": 40,
+        "caught_pct": 86.2,
+        "false_walls_pct": 27.3
+      },
+      "illegal_weapons_drugs_cyber": {
+        "n": 47,
+        "caught_pct": 90.0,
+        "false_walls_pct": 0.0
+      },
+      "pii_doxxing": {
+        "n": 17,
+        "caught_pct": 70.6,
+        "false_walls_pct": null
+      },
+      "scam_impersonation": {
+        "n": 14,
+        "caught_pct": 70.0,
+        "false_walls_pct": 0.0
+      },
+      "selfharm_care": {
+        "n": 28,
+        "caught_pct": null,
+        "false_walls_pct": 0.0
+      },
+      "selfharm_harmful": {
+        "n": 25,
+        "caught_pct": 84.0,
+        "false_walls_pct": null
+      },
+      "sexual_adult_or_ed": {
+        "n": 18,
+        "caught_pct": 71.4,
+        "false_walls_pct": 9.1
+      },
+      "sexual_minors": {
+        "n": 6,
+        "caught_pct": 50.0,
+        "false_walls_pct": null
+      },
+      "threat_violence": {
+        "n": 25,
+        "caught_pct": 85.7,
+        "false_walls_pct": 75.0
+      }
+    }
+  },
+  "paraphrase": {
+    "title": "1154 paraphrase eval",
+    "t": 0.84,
+    "n_scored": 1154,
+    "caught_pct": 71.9,
+    "false_walls_pct": 15.1,
+    "accuracy_pct": 80.8,
+    "mistakes": 221,
+    "inj_reported_separately": {},
+    "by_lang": {
+      "nl": {
+        "n": 736,
+        "caught_pct": 74.5,
+        "false_walls_pct": 16.0,
+        "accuracy_pct": 81.5
+      },
+      "en": {
+        "n": 418,
+        "caught_pct": 69.0,
+        "false_walls_pct": 13.2,
+        "accuracy_pct": 79.7
+      }
+    },
+    "by_cat": {
+      "benign_code_switch": {
+        "n": 48,
+        "caught_pct": null,
+        "false_walls_pct": 27.1
+      },
+      "benign_dark_humor": {
+        "n": 60,
+        "caught_pct": null,
+        "false_walls_pct": 23.3
+      },
+      "benign_neutral": {
+        "n": 244,
+        "caught_pct": null,
+        "false_walls_pct": 10.7
+      },
+      "benign_profanity": {
+        "n": 60,
+        "caught_pct": null,
+        "false_walls_pct": 23.3
+      },
+      "benign_short": {
+        "n": 62,
+        "caught_pct": null,
+        "false_walls_pct": 1.6
+      },
+      "benign_slang_flemish": {
+        "n": 80,
+        "caught_pct": null,
+        "false_walls_pct": 11.2
+      },
+      "benign_stress": {
+        "n": 56,
+        "caught_pct": null,
+        "false_walls_pct": 7.1
+      },
+      "bypass_injection": {
+        "n": 36,
+        "caught_pct": 60.0,
+        "false_walls_pct": 18.8
+      },
+      "bypass_obfuscation": {
+        "n": 68,
+        "caught_pct": 64.0,
+        "false_walls_pct": 33.3
+      },
+      "harassment_hate": {
+        "n": 80,
+        "caught_pct": 77.6,
+        "false_walls_pct": 27.3
+      },
+      "illegal_weapons_drugs_cyber": {
+        "n": 94,
+        "caught_pct": 85.0,
+        "false_walls_pct": 17.6
+      },
+      "pii_doxxing": {
+        "n": 34,
+        "caught_pct": 58.8,
+        "false_walls_pct": null
+      },
+      "scam_impersonation": {
+        "n": 28,
+        "caught_pct": 65.0,
+        "false_walls_pct": 12.5
+      },
+      "selfharm_care": {
+        "n": 56,
+        "caught_pct": null,
+        "false_walls_pct": 14.3
+      },
+      "selfharm_harmful": {
+        "n": 50,
+        "caught_pct": 72.0,
+        "false_walls_pct": null
+      },
+      "sexual_adult_or_ed": {
+        "n": 36,
+        "caught_pct": 71.4,
+        "false_walls_pct": 22.7
+      },
+      "sexual_minors": {
+        "n": 12,
+        "caught_pct": 50.0,
+        "false_walls_pct": null
+      },
+      "threat_violence": {
+        "n": 50,
+        "caught_pct": 81.0,
+        "false_walls_pct": 50.0
+      }
+    }
+  },
+  "dev_threshold": 0.84,
+  "dev_acc": 92.4,
+  "scoring_wall_s": 1.6
+}