Skip to content

DeepTeam is a framework to red team LLMs and AI agents.

added to_df methods

A-Vamshi196559552f00
diff --git a/deepteam/red_teamer/risk_assessment.py b/deepteam/red_teamer/risk_assessment.py
index 48473b43..1ecd1b31 100644
--- a/deepteam/red_teamer/risk_assessment.py
+++ b/deepteam/red_teamer/risk_assessment.py
@@ -37,6 +37,41 @@ class TestCasesList(list):
case_data.update(case.metadata)
data.append(case_data)
return pd.DataFrame(data)
+
+class VulnerabilityTypeResultsList(list):
+
+ def to_df(self) -> "pd.DataFrame":
+ import pandas as pd
+
+ data = []
+ for case in self:
+ case_data = {
+ "Vulnerability": case.vulnerability,
+ "Vulnerability Type": str(case.vulnerability_type.value),
+ "Pass Rate": case.pass_rate,
+ "Passing": case.passing,
+ "Failing": case.failing,
+ "Errored": case.errored,
+ }
+ data.append(case_data)
+ return pd.DataFrame(data)
+
+class AttackMethodResultList(list):
+
+ def to_df(self) -> "pd.DataFrame":
+ import pandas as pd
+
+ data = []
+ for case in self:
+ case_data = {
+ "Attack Method": case.attack_method,
+ "Pass Rate": case.pass_rate,
+ "Passing": case.passing,
+ "Failing": case.failing,
+ "Errored": case.errored,
+ }
+ data.append(case_data)
+ return pd.DataFrame(data)
class VulnerabilityTypeResult(BaseModel):
@@ -96,6 +131,8 @@ class RiskAssessment(BaseModel):
self.test_cases: TestCasesList = TestCasesList[RTTestCase](
self.test_cases
)
+ self.overview.vulnerability_type_results: VulnerabilityTypeResultsList = VulnerabilityTypeResultsList[VulnerabilityTypeResult](self.overview.vulnerability_type_results)
+ self.overview.attack_method_results: AttackMethodResultList = AttackMethodResultList[AttackMethodResult](self.overview.attack_method_results)
def save(self, to: str) -> str:
try: