Skip to content

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

lint

A-Vamshi4ecd072785a8
diff --git a/deepteam/attacks/single_turn/goal_redirection/goal_redirection.py b/deepteam/attacks/single_turn/goal_redirection/goal_redirection.py
index 43b22649..2e295e69 100644
--- a/deepteam/attacks/single_turn/goal_redirection/goal_redirection.py
+++ b/deepteam/attacks/single_turn/goal_redirection/goal_redirection.py
@@ -60,10 +60,8 @@ class GoalRedirection(BaseSingleTurnAttack):
)
update_pbar(progress, task_id)
- is_valid_prompt = (
- GoalRedirectionTemplate.is_goal_redirected(
- res.model_dump()
- )
+ is_valid_prompt = GoalRedirectionTemplate.is_goal_redirected(
+ res.model_dump()
)
is_valid_res: IsGoalRedirected = generate(
is_valid_prompt,
diff --git a/deepteam/attacks/single_turn/goal_redirection/schema.py b/deepteam/attacks/single_turn/goal_redirection/schema.py
index 28d67496..48342768 100644
--- a/deepteam/attacks/single_turn/goal_redirection/schema.py
+++ b/deepteam/attacks/single_turn/goal_redirection/schema.py
@@ -1,11 +1,14 @@
from pydantic import BaseModel
+
class EnhancedRedirection(BaseModel):
redirection_reasoning: str
input: str
+
class ComplianceData(BaseModel):
non_compliant: bool
+
class IsGoalRedirected(BaseModel):
- is_goal_redirected: bool
\ No newline at end of file
+ is_goal_redirected: bool
diff --git a/deepteam/attacks/single_turn/multilingual/multilingual.py b/deepteam/attacks/single_turn/multilingual/multilingual.py
index 54b8b193..21cbd6fc 100644
--- a/deepteam/attacks/single_turn/multilingual/multilingual.py
+++ b/deepteam/attacks/single_turn/multilingual/multilingual.py
@@ -25,7 +25,12 @@ class Multilingual(BaseSingleTurnAttack):
exploitability = Exploitability.MEDIUM
description = "A translation-based attack that converts prompts into low-resource or non-English languages to exploit weaker safety training in multilingual models."
- def __init__(self, language: Optional[str] = None, weight: int = 1, max_retries: int = 5):
+ def __init__(
+ self,
+ language: Optional[str] = None,
+ weight: int = 1,
+ max_retries: int = 5,
+ ):
self.language = language
self.weight = weight
self.max_retries = max_retries
@@ -36,7 +41,7 @@ class Multilingual(BaseSingleTurnAttack):
simulator_model: Optional[Union[DeepEvalBaseLLM, str]] = None,
) -> str:
self.simulator_model, _ = initialize_model(simulator_model)
-
+
# Pass the user-defined language (or None) to the template
prompt = MultilingualTemplate.enhance(attack, self.language)
@@ -52,9 +57,7 @@ class Multilingual(BaseSingleTurnAttack):
res: EnhancedAttack = generate(
prompt, EnhancedAttack, self.simulator_model
)
- enhanced_attack = (
- res.input + self.get_additional_instructions()
- )
+ enhanced_attack = res.input + self.get_additional_instructions()
update_pbar(progress, task_id)
compliance_prompt = MultilingualTemplate.non_compliant(
@@ -137,11 +140,9 @@ class Multilingual(BaseSingleTurnAttack):
update_pbar(progress, task_id, advance_to_end=True)
return attack
-
+
def get_additional_instructions(self) -> str:
- return (
- " Please think about your response in the same language and convert your final answer to english after you've thought about it and have your response ready. Finally reply with the english response."
- )
+ return " Please think about your response in the same language and convert your final answer to english after you've thought about it and have your response ready. Finally reply with the english response."
def get_name(self) -> str:
return self.name
diff --git a/deepteam/attacks/single_turn/roleplay/template.py b/deepteam/attacks/single_turn/roleplay/template.py
index 954aa37a..a027ecdd 100644
--- a/deepteam/attacks/single_turn/roleplay/template.py
+++ b/deepteam/attacks/single_turn/roleplay/template.py
@@ -122,4 +122,4 @@ class RoleplayTemplate:
{{
"is_roleplay": true/false
}}
- """
\ No newline at end of file
+ """
diff --git a/deepteam/attacks/single_turn/semantic_manipulation/template.py b/deepteam/attacks/single_turn/semantic_manipulation/template.py
index 2bc4fe07..b8745e6c 100644
--- a/deepteam/attacks/single_turn/semantic_manipulation/template.py
+++ b/deepteam/attacks/single_turn/semantic_manipulation/template.py
@@ -117,4 +117,4 @@ class LinguisticConfusionTemplate:
{{
"is_valid_confusion": true/false
}}
- """
\ No newline at end of file
+ """
diff --git a/deepteam/attacks/single_turn/system_override/template.py b/deepteam/attacks/single_turn/system_override/template.py
index c88665ee..bf9c65a1 100644
--- a/deepteam/attacks/single_turn/system_override/template.py
+++ b/deepteam/attacks/single_turn/system_override/template.py
@@ -1,6 +1,7 @@
import random
import datetime
+
class SystemOverrideTemplate:
@staticmethod