MCPcopy Create free account

hub / github.com/AIM-Intelligence/Automated-Multi-Turn-Jailbreaks / functions

Functions21 in github.com/AIM-Intelligence/Automated-Multi-Turn-Jailbreaks

⨍ Functions 21◇ Types & classes 0
↓ 6 callersFunctiongenerate
(messages, client, model, json_format=False, temperature=0.7)
utils/generate.py:2
↓ 5 callersFunctionevaluate_with_rubric
(response, task, rubric, generate)
utils/evaluate_with_rubric.py:1
↓ 4 callersFunctiongenerate_score_rubric
(topic, generate, success_condition=None)
utils/generate_score_rubric.py:32
↓ 3 callersFunctioncheck_refusal
(prompt, response, generate)
utils/check_refusal.py:57
↓ 1 callersFunctionattack_single
(output_file_path, target_system, data, rubric, target_generate, attacker_generate)
actor_attack/run.py:130
↓ 1 callersFunctioncall_multi
(target_system, output_file_path, instruction, queries, query_details, rubric, target_generate, attacker_gener
actor_attack/run.py:75
↓ 1 callersFunctioncheck_disclaimer
(prompt, response, generate)
utils/check_disclaimer.py:22
↓ 1 callersFunctionextract_harm_target
(org_query, attacker_generate)
actor_attack/run.py:6
↓ 1 callersFunctiongenerate_acronym_step
(round_num, task, history_a, last_response, attacker_generate)
acronym/run.py:7
↓ 1 callersFunctiongenerate_crescendo_step
(round_num, task, history_a, last_response, attacker_generate)
crescendomation/run.py:8
↓ 1 callersFunctiongenerate_opposite_day_step
(round_num, task, history_a, last_response, attacker_generate)
opposite_day/run.py:7
↓ 1 callersFunctionget_actors
(harm_target, actor_num, attacker_generate)
actor_attack/run.py:12
↓ 1 callersFunctionget_init_queries
(harm_target, actor, attacker_generate)
actor_attack/run.py:34
↓ 1 callersFunctioninfer_single
(org_query: str, actor_num: int, attacker_generate)
actor_attack/run.py:48
↓ 1 callersFunctionrewrite_query
(query, last_response, queries, query_details, attacker_generate)
actor_attack/run.py:66
↓ 1 callersFunctionrun_acronym
(test_case, output_file_path, target_generate, attacker_generate)
acronym/run.py:21
↓ 1 callersFunctionrun_actor_attack
(test_case, output_file_path, target_generate, attacker_generate)
actor_attack/run.py:146
↓ 1 callersFunctionrun_crescendomation
(test_case, output_file_path, target_generate, attacker_generate)
crescendomation/run.py:22
↓ 1 callersFunctionrun_opposite_day
(test_case, output_file_path, target_generate, attacker_generate)
opposite_day/run.py:21
↓ 1 callersFunctionstep_judge
Attack Step Judge has three types of responses: successful, rejective, and unknown.
actor_attack/run.py:58
↓ 1 callersFunctionsummary
(output_file_path, instruction, query_details, dialog_hist, actor, rubric, target_generate, attacker_generate)
actor_attack/run.py:105