searchindex.js

Search.setIndex({"docnames": ["Seed", "Selector", "attacker", "constraint", "datasets", "evaluator", "index", "metrics"], "filenames": ["Seed.rst", "Selector.rst", "attacker.rst", "constraint.rst", "datasets.rst", "evaluator.rst", "index.rst", "metrics.rst"], "titles": ["Seed Module", "Selector Module", "Attacker Module", "Constraint Module", "Datasets Module", "Evaluator Module", "Attacker", "Metric Module"], "terms": {"\u8fd9\u90e8\u5206\u6587\u6863\u4ecb\u7ecd\u4e86": [], "easyjailbreak": [0, 1, 2, 3, 4, 5, 6, 7], "\u4e2d\u7684\u5404\u4e2a\u5b50\u6a21\u5757": [], "\u6a21\u5757": [], "thi": [0, 1, 2, 4, 5, 6, 7], "provid": [0, 1, 2, 4, 6], "jailbreakdataset": [1, 2, 3, 4, 5, 6, 7], "class": [0, 7], "which": [0, 1, 3, 4, 5, 6, 7], "i": [1, 2, 3, 4, 5, 6, 7], "design": [1, 4, 5, 6, 7], "manag": [4, 6], "manipul": [4, 6], "easi": [1, 4, 6], "jailbreak": [0, 1, 2, 3, 4, 5, 6, 7], "applic": [4, 6], "It": [1, 2, 3, 4, 5, 6, 7], "capabl": [4, 6], "handl": [4, 6], "structur": [4, 6], "promptnod": [4, 6], "offer": [1, 4, 5, 6], "function": [0, 4, 6], "shuffl": [4, 6], "access": [2, 4, 6], "process": [0, 1, 2, 3, 4, 6], "data": [4, 6], "point": [4, 6], "an": [1, 2, 3, 4, 5, 6], "organ": [4, 6], "wai": [4, 6], "machin": [4, 6], "learn": [3, 4, 6], "task": [1, 4, 5, 6], "relat": [4, 6], "list": [0, 1, 3, 4, 5, 6, 7], "str": [0, 2, 3, 4, 5, 6], "bool": [2, 3, 4, 5, 6], "fals": [1, 2, 3, 4, 5, 6], "local_file_typ": [4, 6], "json": [4, 6], "specif": [1, 2, 4, 5, 6], "allow": [4, 5, 6], "represent": [4, 6], "form": [4, 6], "essenti": [4, 6], "format": [4, 5, 6], "us": [0, 1, 2, 3, 4, 5, 6, 7], "model": [0, 1, 2, 3, 4, 5, 6, 7], "add": [4, 6], "new": [0, 1, 4, 6], "paramet": [0, 1, 2, 3, 4, 5, 6, 7], "The": [0, 1, 2, 3, 4, 5, 6, 7], "ad": [4, 6], "group_bi": [4, 6], "kei": [4, 6], "group": [2, 4, 6], "base": [0, 1, 2, 3, 4, 5, 6], "specifi": [3, 4, 5, 6], "A": [0, 1, 2, 3, 4, 5, 6, 7], "take": [4, 6], "return": [0, 1, 2, 3, 4, 5, 6, 7], "hashabl": [4, 6], "object": [2, 4, 6, 7], "where": [4, 5, 6], "each": [1, 4, 6, 7], "sublist": [4, 6], "contain": [1, 2, 3, 4, 5, 6, 7], "group_by_par": [4, 6], "parent": [1, 4, 6], "node": [1, 2, 3, 4, 6], "static": [4, 6], "load_csv": [4, 6], "path": [4, 5, 6], "csv": [4, 6], "header": [4, 6], "int": [0, 2, 3, 4, 5, 6], "none": [0, 1, 2, 3, 4, 5, 6], "load": [4, 5, 6], "file": [0, 1, 4, 6], "column": [4, 6], "name": [0, 4, 5, 6], "default": [4, 5, 6], "load_jsonl": [4, 6], "jsonl": [2, 4, 6], "classmethod": [4, 6], "merg": [2, 4, 6], "dataset_list": [4, 6], "multipl": [4, 6], "singl": [2, 4, 6], "from": [0, 1, 2, 3, 4, 5, 6, 7], "save_to_csv": [4, 6], "save": [4, 6], "save_to_jsonl": [2, 4, 6], "jsonlin": [4, 6], "librari": [4, 6], "place": [4, 6], "method": [0, 1, 2, 4, 6], "random": [1, 4, 6], "order": [4, 6], "": [0, 2, 3, 4, 5, 6], "element": [4, 6], "updat": [1, 2, 4, 5, 6], "attribut": [4, 5, 6], "true": [2, 3, 4, 5, 6], "metric": 5, "\u6b22\u8fce\u6765\u5230": [], "\u6ce8\u91ca\u6587\u6863\u8bf4\u660e": [], "iter": [1, 2, 6], "optim": [2, 6], "section": [0, 1, 2, 6], "prompt": [0, 1, 2, 3, 6, 7], "guidanc": [2, 6], "token": [2, 5, 6], "gradient": [2, 6], "ensur": [2, 6], "produc": [2, 6], "desir": [2, 6], "text": [2, 3, 5, 6], "paper": [0, 1, 2, 3, 6], "titl": [1, 2, 3, 6], "univers": [2, 6], "transfer": [2, 6], "adversari": [1, 2, 6], "align": [2, 6], "languag": [0, 2, 3, 6], "arxiv": [1, 2, 3, 6], "link": [1, 2, 3, 6], "http": [1, 2, 3, 6], "org": [1, 2, 3, 6], "ab": [1, 2, 3, 6], "2307": [2, 6], "15043": [2, 6], "sourc": [1, 2, 3, 6], "repositori": [1, 2, 3, 6], "github": [1, 2, 3, 6], "com": [1, 2, 3, 6], "llm": [1, 2, 3, 6], "gcg": [2, 6], "attack_model": [2, 6], "whiteboxmodelbas": [1, 2, 6, 7], "target_model": [2, 6], "modelbas": [2, 6], "jailbreak_dataset": [1, 2, 3, 7], "jailbreak_prompt_length": [2, 6], "20": [2, 6], "num_turb_sampl": [2, 6], "512": [2, 3, 6, 7], "batchsiz": [2, 6], "top_k": [2, 6], "256": [2, 6], "max_num_it": [2, 6], "500": [2, 3, 6], "is_univers": [1, 2, 6], "abstract": [1, 2, 5, 6], "perform": [1, 2, 5, 6], "single_attack": [2, 6], "instanc": [1, 2, 3, 5, 7], "common": [2, 6], "case": [2, 6], "result": [2, 5, 6, 7], "dataset": [1, 2, 3, 5, 7], "modifi": [2, 6], "type": [2, 6], "achiev": [2, 6], "describ": [2, 6], "below": [2, 3, 6], "part": [2, 6, 7], "code": [2, 6], "red": [0, 2, 6], "team": [0, 2, 6], "larg": [2, 6], "auto": [2, 6], "gener": [0, 1, 2, 3, 5, 6, 7], "pdf": [2, 6], "2309": [2, 6], "10253": [2, 6], "sherdencoop": [2, 6], "gptfuzz": [2, 6], "eval_model": [2, 3, 5, 6], "energi": [1, 2, 6], "1": [0, 1, 2, 3, 5, 6], "max_queri": [2, 6], "100": [0, 2, 6], "max_jailbreak": [2, 6], "max_reject": [2, 6], "max_iter": [2, 6], "seeds_num": [0, 2, 6], "76": [2, 6], "fuzz": [2, 6], "util": [2, 3, 6], "mutat": [1, 2, 3, 6], "select": [1, 2, 6], "polici": [1, 2, 6], "aim": [2, 3, 6], "find": [2, 6], "vulner": [2, 6], "target": [1, 2, 4, 5, 6], "main": [2, 6], "loop": [2, 6], "repeatedli": [2, 6], "evalu": [1, 2, 3, 7], "is_stop": [2, 6], "check": [2, 5, 6], "stop": [2, 5, 6], "criteria": [1, 2, 6], "ar": [0, 1, 2, 3, 5, 6], "met": [2, 6], "ani": [1, 2, 3, 5, 6], "otherwis": [2, 3, 5, 6], "log": [2, 6], "current": [2, 6], "statu": [2, 6], "displai": [2, 6], "queri": [0, 2, 3, 4, 5, 6], "param": [0, 2, 6], "string": [5, 6], "respons": [0, 2, 4, 5, 6], "state": [1, 2, 6], "fuzzer": [2, 6], "prompt_nod": [1, 2, 6], "have": [1, 2, 3, 6], "been": [1, 2, 6], "execut": [2, 6], "In": [0, 1, 2, 6], "context": [0, 1, 2, 6], "algorithm": [1, 2, 6], "guard": [2, 6], "onli": [2, 3, 6], "few": [2, 6], "demonstr": [2, 6], "2310": [2, 6], "06387": [2, 6], "prompt_num": [2, 6], "5": [1, 2, 6], "user_input": [2, 6], "pattern_dict": [2, 5, 6], "craft": [2, 3, 6], "malici": [2, 6], "guid": [2, 6], "harm": [0, 2, 3, 4, 5, 6], "output": [0, 2, 3, 6], "through": [2, 3, 6], "report": [2, 6], "sampl": [1, 2, 6], "conduct": [2, 6], "n": [2, 6], "shot": [2, 6], "black": [1, 2, 3, 6], "box": [1, 2, 3, 6], "twenti": [2, 6], "08419": [2, 6], "patrickrchao": [2, 6], "jailbreakingllm": [2, 6], "pair": [2, 6], "template_fil": [0, 2, 6], "attack_max_n_token": [2, 6], "max_n_attack_attempt": [2, 6], "attack_temperatur": [2, 6], "attack_top_p": [2, 6], "0": [0, 1, 2, 3, 5, 6], "9": [2, 6], "target_max_n_token": [2, 6], "150": [2, 6], "target_temperatur": [2, 6], "target_top_p": [2, 6], "judge_max_n_token": [2, 6], "10": [2, 3, 6], "judge_temperatur": [2, 6], "n_stream": [2, 6], "keep_last_n": [2, 6], "3": [2, 5, 6], "n_iter": [2, 6], "save_path": [2, 6], "pair_attack_result": [2, 6], "implement": [1, 2, 5, 6, 7], "automat": [1, 2, 3, 5, 6], "2312": [1, 2, 3, 6], "02119": [1, 2, 3, 6], "ricommun": [1, 2, 3, 6], "tap": [1, 2, 3, 6], "author": [], "fukang": [], "zhu": [], "tree_width": [1, 2, 3, 6], "tree_depth": [2, 6], "root_num": [2, 6], "branching_factor": [2, 6], "4": [2, 6], "extens": [2, 6], "phase": [2, 6], "branch": [2, 6], "2": [1, 2, 5, 6, 7], "prune": [2, 6], "rubric": [], "exampl": [5, 6], "import": [1, 2, 3, 5, 6], "huggingface_model": [2, 5, 6], "from_pretrain": [2, 5, 6], "model_path_1": [2, 6], "model_path_2": [2, 6], "model_path_3": [2, 6], "advbench": [2, 6], "tap_result": [2, 6], "our": [2, 6], "wolf": [2, 6], "sheep": [2, 6], "cloth": [2, 6], "nest": [2, 6], "can": [0, 1, 2, 3, 5, 6], "fool": [2, 6], "easili": [2, 6], "2311": [2, 6], "08268": [2, 6], "njunlp": [2, 6], "evo_max": [2, 6], "integr": [2, 6], "strategi": [1, 2, 5, 6], "exploit": [1, 2, 6], "weak": [2, 6], "prompt_pattern": [3, 5, 6], "attr_nam": [3, 5, 6], "filter": [3, 6], "out": [3, 6], "consid": [3, 6], "harmless": [3, 6], "determin": [3, 5, 6, 7], "given": [3, 5, 6, 7], "potenti": [1, 3, 6], "neg": [3, 6], "danger": [3, 6], "uneth": [3, 6], "illeg": [3, 5, 6], "content": [0, 3], "identifi": [3, 6], "retain": [3, 6], "judg": [0, 3, 5, 6], "seed": [1, 3, 5], "delet": [3, 6], "off": [3, 6], "topic": [3, 6], "tree": [1, 3], "attack": [0, 1, 3, 7], "detail": [1, 3, 6], "principl": [3, 6], "found": [1, 3, 6], "follow": [1, 3, 6], "constrain": [3, 6], "instance_list": [3, 6], "write": [0, 3, 6], "tutori": [3, 5, 6], "make": [0, 3, 5, 6], "bomb": [3, 5, 6], "jailbreak_prompt": [2, 3, 6], "you": [0, 3, 6], "chemist": [3, 6], "want": [0, 3, 6], "how": [0, 2, 3, 6], "pleas": [3, 6], "hello": [3, 6], "todai": [3, 6], "new_dataset_on_top": [3, 6], "threshold": [3, 6], "max_length": [3, 6, 7], "stride": [3, 6, 7], "perplex": 3, "score": [1, 2, 3, 5, 6, 7], "comput": [1, 3, 6, 7], "indic": [0, 3, 5, 6], "non": [2, 3, 6], "serv": [5, 6], "templat": [0, 5, 6], "creat": [5, 6], "custom": [5, 6], "divers": [5, 6], "special": [5, 6], "tailor": [1, 5, 6], "differ": [2, 5, 6], "system_prompt": [5, 6], "attempt": [5, 6], "success": [1, 5, 6, 7], "assign": [5, 6], "judgment": [5, 6], "higher": [5, 6], "get_llamagurad": [5, 6], "model_path": [5, 6], "llamaguard": [5, 6], "If": [5, 6], "get_scor": [5, 6], "score_format": [5, 6], "kwarg": [0, 5, 6], "evalatorclassifcaton": [5, 6], "inherit": [5, 6], "classif": [5, 6], "call": [5, 6], "__call__": [5, 6], "test": [5, 6], "_evalu": [5, 6], "both": [5, 6], "support": [4, 5, 6], "variabl": [5, 6], "respect": [5, 6], "replac": [0, 2, 5, 6], "same": [3, 5, 6], "input": [0, 3, 5, 6], "classifi": [5, 6], "set_model": [5, 6], "set": [1, 5, 6], "roberta": [5, 6], "pretrain": [5, 6], "special_method": [5, 6], "whether": [0, 5, 6], "subclass": [1, 5, 6], "direct": [5, 6], "match": [5, 6], "between": [1, 5, 6], "refer": [0, 1, 5, 6], "binari": [5, 6], "exactli": [5, 6], "evalatorclassifcaton\u7ee7\u627f\u4e86evaluator\u7c7b": [], "\u7528\u4e8e\u8bc4\u4f30\u5206\u7c7b\u4efb\u52a1\u7684\u6a21\u578b": [], "\u53ef\u4ee5\u901a\u8fc7__call__\u6765\u8c03\u7528": [], "\u4e5f\u53ef\u4ee5\u901a\u8fc7_evaluate\u5bf9\u4e8e\u4e00\u6761instance\u8fdb\u884c\u6d4b\u8bd5": [], "\u8fd9\u91cc\u652f\u6301\u4f20\u5165prompt_pattern\u548cattr_name\u4e24\u4e2a\u53c2\u6570": [], "\u5206\u522b\u7528\u4e8e\u6307\u5b9a\u6a21\u7248\u548c\u6a21\u7248\u4e2d\u7684\u53d8\u91cf\u540d": [], "\u6a21\u7248\u4e2d\u7684\u53d8\u91cf\u540d\u4f1a\u88abinstance\u4e2d\u7684\u540c\u540d\u5c5e\u6027\u66ff\u6362": [], "pattern": [5, 6], "against": [5, 6], "unsuccess": [5, 6], "failur": [5, 6], "set_pattern": [5, 6], "dictionari": [2, 5, 6, 7], "dict": [0, 2, 5, 6, 7], "effect": [1, 6, 7], "calcul": [1, 2, 6, 7], "number": [6, 7], "fail": [2, 6, 7], "overal": [6, 7], "rate": [6, 7], "within": [6, 7], "averag": [6, 7], "final": [6, 7], "pre": [6, 7], "train": [2, 6, 7], "small": [6, 7], "gpt": [2, 6, 7], "evaluatorprefixexactmatch": [5, 6], "extend": [1, 5, 6], "start": [4, 5, 6], "prefix": [5, 6], "exact": [5, 6], "stealthi": [2, 6], "ON": [2, 6], "04451": [2, 6], "sheltonliu": [2, 6], "git": [2, 6], "devic": [2, 6], "cuda": [2, 6], "num_step": [2, 6], "sentence_level_step": [2, 6], "word_dict_s": [2, 6], "30": [2, 6], "batch_siz": [1, 2, 6], "64": [2, 6], "num_elit": [2, 6], "float": [1, 2, 6], "crossover_r": [2, 6], "mutation_r": [2, 6], "01": [2, 6], "num_point": [2, 6], "model_nam": [2, 6], "llama2": [2, 6], "low_memori": [2, 6], "hierarch": [2, 6], "genet": [2, 6], "construct_momentum_word_dictionari": [2, 6], "word_dict": [2, 6], "individu": [2, 6], "score_list": [2, 6], "momentum": [2, 6], "maintain": [2, 6], "evaluate_candidate_prompt": [2, 6], "prefix_manag": [2, 6], "candid": [2, 6], "get": [2, 3, 6], "best": [2, 6], "correspond": [2, 6], "get_score_autodan": [2, 6], "conv_templ": [2, 6], "instruct": [2, 6], "test_control": [2, 6], "crit": [2, 6], "convert": [2, 6], "all": [2, 6], "id": [2, 6], "max": [2, 6], "length": [1, 2, 6], "get_score_autodan_low_memori": [2, 6], "when": [1, 2, 5, 6], "memori": [2, 6], "low": [2, 6], "replace_with_synonym": [2, 6], "sentenc": [2, 6], "num": [2, 6], "word": [2, 6], "synonym": [2, 6], "roulette_wheel_select": [2, 6], "data_list": [2, 6], "num_select": [2, 6], "appli": [2, 6], "hga": [2, 6], "too": [2, 6], "smart": [2, 6], "To": [2, 6], "Be": [2, 6], "safe": [0, 2, 6], "chat": [2, 5, 6], "via": [2, 6], "cipher": [], "2308": [2, 6], "06463": [2, 6], "enabl": [2, 6], "human": [2, 6], "top": [2, 6], "system": [2, 6], "role": [2, 6], "descript": [2, 6], "enciph": [2, 6], "should": [4, 6], "four": [2, 6], "entir": [2, 6], "attack_mehtod": [2, 6], "includ": [], "attack_result": [2, 6], "deepincept": [], "hypnot": [2, 6], "03191": [2, 6], "unlock": [2, 6], "its": [1, 2, 6], "misus": [2, 6], "risk": [2, 6], "leverag": [], "personif": [], "abil": [], "construct": [2, 6], "novel": [], "scene": [2, 6], "behav": [], "realiz": [], "adapt": [], "escap": [], "usag": [], "control": [], "normal": [], "scenario": [1, 2, 6], "possibl": [0, 6], "further": [], "character_numb": [2, 6], "layer_numb": [2, 6], "valu": [], "scienc": [], "fiction": [], "charact": [], "layer": [], "jailbroken": [], "doe": [2, 6], "safeti": [2, 6], "02483": [2, 6], "jailbrooken": [], "compet": [2, 6], "mismatch": [2, 6], "mode": [2, 6], "29": [2, 6], "artifici": [2, 6], "challeng": [2, 6], "multilingu": [], "IN": [2, 6], "06474": [2, 6], "translat": [2, 6], "english": [2, 6], "nine": [2, 6], "vari": [2, 6], "level": [2, 6], "resourc": [1, 2, 6], "intent": [2, 6], "user": [0, 1, 2, 3, 4, 5, 6], "deliber": [2, 6], "combin": [2, 6], "translate_to_en": [2, 6], "src_lang": [2, 6], "seedbas": [0, 6], "store": [0, 4, 6], "new_se": [0, 6], "old": [0, 6], "batch": [0, 4, 6], "keyword": [0, 6], "seedllm": [0, 6], "insert_valu": [0, 6], "help": [0, 6], "assist": [0, 6], "need": [0, 4, 5, 6], "examin": [0, 6], "alwai": [0, 6], "now": [0, 6], "me": [0, 6], "ha": [0, 6], "highest": [0, 6], "chanc": [0, 6], "answer": [0, 6], "input_field_nam": [0, 6], "messag": [0, 6], "show": [0, 6], "what": [0, 6], "insert": [0, 6], "e": [0, 3, 4, 6], "g": [0, 3, 4, 6], "mani": [0, 6], "field": [0, 6], "mai": [0, 3, 4, 5, 6], "temperatur": [0, 6], "seedrandom": [0, 6], "posible_token": [0, 6], "seeds_max_length": [0, 6], "early_stop_poss": [0, 6], "randomli": [0, 1, 6], "seedtempl": [0, 6], "prompt_usag": [0, 6], "method_list": [0, 6], "origin": [0, 2, 6], "defin": [1, 6], "concret": [1, 6], "variou": [0, 1, 6], "repres": [1, 6], "framework": [1, 6], "initi": [0, 1, 2, 6], "reset": [1, 6], "intern": [1, 6], "necessari": [1, 6], "must": [1, 6], "explore_coeff": [1, 6], "upper": [1, 6], "confid": [1, 6], "bound": [1, 6], "ucb": [1, 6], "balanc": [1, 6], "explor": [1, 6], "formula": [1, 6], "either": [1, 6], "high": [1, 3, 6], "reward": [1, 6], "much": [1, 6], "last": [1, 6], "those": [1, 6], "whose": [1, 6], "extent": [1, 6], "inform": [1, 4, 6], "subset": [1, 6], "relev": [1, 6], "round": [1, 6], "robin": [1, 6], "manner": [1, 6], "over": [1, 6], "turn": [1, 6], "repeat": [1, 6], "next": [1, 6], "approach": [1, 6], "increment": [1, 6], "visit": [1, 6], "count": [1, 6], "index": [1, 6], "Not": [1, 6], "inital_prompt_pool": [1, 6], "question": [1, 6], "ratio": [1, 6], "alpha": [1, 6], "beta": [1, 6], "mont": [1, 6], "carlo": [1, 6], "search": [1, 6], "mct": [1, 6], "weight": [1, 6], "gamma": [1, 6], "05": [1, 6], "25": [1, 6], "exponenti": [1, 6], "exp3": [1, 6], "environ": [1, 6], "known": [1, 6], "probabl": [1, 6], "chosen": [1, 6], "referencelossselector": [], "loss": [1, 6], "minimum": [1, 6], "discard": [1, 6], "other": [0, 1, 4, 6], "after": [2, 6], "get_evaluator_prompt_on_top": [3, 6], "attack_prompt": [3, 6], "process_output_on_topic_scor": [3, 6], "raw_output": [3, 6], "eval": [3, 6], "ye": [3, 6], "els": [3, 6], "huggingfacemodel": [5, 6], "internlm": [5, 6], "20b": [5, 6], "7b": [5, 6], "m": [5, 6], "sorri": [5, 6], "t": [5, 6], "print": [5, 6], "robustnlp": [2, 6], "cipherchat": [2, 6], "tmlr": [2, 6], "damo": [2, 6], "nlp": [2, 6], "sg": [2, 6], "document": [0, 1, 2, 6, 7], "introduc": [0, 1, 6, 7], "submodul": [0, 1, 2, 6], "fit": [0, 6], "most": [0, 1, 6], "former": [0, 6], "research": [0, 6], "some": [0, 1, 3, 6], "might": [0, 6], "partial": [0, 6], "chang": [0, 6], "while": [0, 6], "whole": [0, 6], "histori": [0, 6], "three": [0, 6], "suitabl": [1, 6], "circumst": [1, 6], "innumer": [1, 6], "therefor": [1, 6], "great": [1, 6], "later": [1, 5, 6], "especi": [1, 6], "limit": [1, 6], "sever": [1, 3, 5, 6], "kind": [1, 3, 6], "pick": [1, 6], "here": [2, 6], "propos": [2, 6], "As": [3, 6], "refut": [3, 6], "emploi": [3, 6], "rule": [3, 6], "deni": [3, 6], "abandon": [3, 6], "useless": [3, 6], "furnish": [3, 6], "choos": [3, 6], "befor": [4, 6], "prepar": [4, 6], "respond": [4, 6], "meanwhil": [4, 6], "gather": [4, 6], "up": [4, 6], "oper": [4, 6], "sign": [5, 6], "earli": [5, 6], "just": [5, 6], "quantifi": [5, 6], "scheme": [5, 6], "welcom": 6, "annot": 6, "split": [2, 6], "them": [2, 6], "conversation_templ": [2, 6], "tap_attack_result": [2, 6]}, "objects": {}, "objtypes": {}, "objnames": {}, "titleterms": {"attack": [2, 6], "modul": [0, 1, 2, 3, 4, 5, 6, 7], "autodan_liu_2023": [2, 6], "cipher_yuan_2023": [2, 6], "deepinception_li_2023": [2, 6], "gcg_zou_2023": [2, 6], "gptfuzzer_yu_2023": [2, 6], "ica_wei_2023": [2, 6], "jailbroken_wei_2023": [2, 6], "multilingual_deng_2023": [2, 6], "pair_chao_2023": [2, 6], "tap_mehrotra_2023": [2, 6], "renellm_ding_2023": [2, 6], "constraint": [3, 6], "deleteharmless": [3, 6], "deleteofftop": [3, 6], "perplexityconstraint": [3, 6], "dataset": [4, 6], "instanc": [4, 6], "jailbreak_dataset": [4, 6], "evalu": [5, 6], "evaluator_classificationgetscor": [5, 6], "evaluator_classificationjudg": [5, 6], "evaluator_generativegetscor": [5, 6], "evaluator_generativejudg": [5, 6], "evaluator_match": [5, 6], "evaluator_patternjudg": [5, 6], "evaluatorgreedyexactmatch": [], "content": 6, "metric": [6, 7], "metric_asr": [6, 7], "metric_perplexit": [6, 7], "gptfuzzer": [2, 6], "class": [1, 2, 3, 4, 5, 6], "ica": [2, 6], "catastroph": [2, 6], "tree": [2, 6], "recip": [2, 6], "renellm": [2, 6], "evaluatorgenerativegetscor": [5, 6], "evaluatorclassificatonjudg": [5, 6], "evaluatorgenerativejudg": [5, 6], "evalatormatch": [5, 6], "evaluatorpatternjudg": [5, 6], "attacksuccessr": [6, 7], "perplex": [6, 7], "evaluator_prefixexactmatch": [5, 6], "autodan": [2, 6], "seed": [0, 6], "seed_bas": [0, 6], "seed_llm": [0, 6], "seed_random": [0, 6], "seed_templ": [0, 6], "selector": [1, 6], "selectpolici": [1, 6], "ucbselectpolici": [1, 6], "selectbasedonscor": [1, 6], "roundrobinselectpolici": [1, 6], "randomselector": [1, 6], "randomselectpolici": [1, 6], "mctsexploreselectpolici": [1, 6], "exp3selectpolici": [1, 6], "loss_selector": [], "cipher": [2, 6], "deepincept": [2, 6], "jailbroken": [2, 6], "multilingu": [2, 6], "selecotr": [], "referencelossselector": [1, 6]}, "envversion": {"sphinx.domains.c": 3, "sphinx.domains.changeset": 1, "sphinx.domains.citation": 1, "sphinx.domains.cpp": 9, "sphinx.domains.index": 1, "sphinx.domains.javascript": 3, "sphinx.domains.math": 2, "sphinx.domains.python": 4, "sphinx.domains.rst": 2, "sphinx.domains.std": 2, "sphinx": 58}, "alltitles": {"Seed Module": [[0, "seed-module"], [6, "seed-module"]], "seed_base": [[0, "seed-base"], [6, "seed-base"]], "seed_llm": [[0, "seed-llm"], [6, "seed-llm"]], "seed_random": [[0, "seed-random"], [6, "seed-random"]], "seed_template": [[0, "seed-template"], [6, "seed-template"]], "Constraint Module": [[3, "constraint-module"], [6, "constraint-module"]], "DeleteHarmLess": [[3, "deleteharmless"], [6, "deleteharmless"]], "DeleteHarmLess class": [[3, "deleteharmless-class"], [6, "deleteharmless-class"]], "DeleteOffTopic": [[3, "deleteofftopic"], [6, "deleteofftopic"]], "PerplexityConstraint": [[3, "perplexityconstraint"], [6, "perplexityconstraint"]], "PerplexityConstraint class": [[3, "perplexityconstraint-class"], [6, "perplexityconstraint-class"]], "Datasets Module": [[4, "datasets-module"], [6, "datasets-module"]], "instance": [[4, "instance"], [6, "instance"]], "Instance class": [[4, "instance-class"], [6, "instance-class"]], "jailbreak_datasets": [[4, "jailbreak-datasets"], [6, "jailbreak-datasets"]], "Jailbreak_Dataset Module": [[4, "jailbreak-dataset-module"], [6, "jailbreak-dataset-module"]], "Evaluator Module": [[5, "evaluator-module"], [6, "evaluator-module"]], "Evaluator": [[5, "evaluator"], [6, "evaluator"], [6, "id1"]], "Evaluator class": [[5, "evaluator-class"], [6, "evaluator-class"]], "Evaluator_ClassificationGetScore": [[5, "evaluator-classificationgetscore"], [6, "evaluator-classificationgetscore"]], "EvaluatorGenerativeGetScore class": [[5, "evaluatorgenerativegetscore-class"], [5, "id1"], [6, "evaluatorgenerativegetscore-class"], [6, "id2"]], "Evaluator_ClassificationJudge": [[5, "evaluator-classificationjudge"], [6, "evaluator-classificationjudge"]], "EvaluatorClassificatonJudge class": [[5, "evaluatorclassificatonjudge-class"], [6, "evaluatorclassificatonjudge-class"]], "Evaluator_GenerativeGetScore": [[5, "evaluator-generativegetscore"], [6, "evaluator-generativegetscore"]], "Evaluator_GenerativeJudge": [[5, "evaluator-generativejudge"], [6, "evaluator-generativejudge"]], "EvaluatorGenerativeJudge class": [[5, "evaluatorgenerativejudge-class"], [6, "evaluatorgenerativejudge-class"]], "Evaluator_Match": [[5, "evaluator-match"], [6, "evaluator-match"]], "EvalatorMatch class": [[5, "evalatormatch-class"], [6, "evalatormatch-class"]], "Evaluator_PatternJudge": [[5, "evaluator-patternjudge"], [6, "evaluator-patternjudge"]], "EvaluatorPatternJudge class": [[5, "evaluatorpatternjudge-class"], [6, "evaluatorpatternjudge-class"]], "Evaluator_PrefixExactMatch": [[5, "evaluator-prefixexactmatch"], [6, "evaluator-prefixexactmatch"]], "Metric Module": [[7, "metric-module"], [6, "metric-module"]], "metric_ASR": [[7, "metric-asr"], [6, "metric-asr"]], "Metrics on AttackSuccessRate": [[7, "metrics-on-attacksuccessrate"], [6, "metrics-on-attacksuccessrate"]], "metric_perplexit": [[7, "metric-perplexit"], [6, "metric-perplexit"]], "Perplexity Metric:": [[7, "perplexity-metric"], [6, "perplexity-metric"]], "Selector Module": [[1, "selector-module"], [6, "selector-module"]], "selector": [[1, "selector"], [6, "id3"]], "SelectPolicy class": [[1, "selectpolicy-class"], [6, "selectpolicy-class"]], "UCBSelectPolicy": [[1, "ucbselectpolicy"], [6, "ucbselectpolicy"]], "UCBSelectPolicy class": [[1, "ucbselectpolicy-class"], [6, "ucbselectpolicy-class"]], "SelectBasedOnScores": [[1, "selectbasedonscores"], [6, "selectbasedonscores"]], "RoundRobinSelectPolicy": [[1, "roundrobinselectpolicy"], [6, "roundrobinselectpolicy"]], "RandomSelector": [[1, "randomselector"], [6, "randomselector"]], "RandomSelectPolicy class": [[1, "randomselectpolicy-class"], [6, "randomselectpolicy-class"]], "MCTSExploreSelectPolicy": [[1, "mctsexploreselectpolicy"], [6, "mctsexploreselectpolicy"]], "MCTSExploreSelectPolicy class": [[1, "mctsexploreselectpolicy-class"], [6, "mctsexploreselectpolicy-class"]], "EXP3SelectPolicy": [[1, "exp3selectpolicy"], [6, "exp3selectpolicy"]], "EXP3SelectPolicy class": [[1, "exp3selectpolicy-class"], [6, "exp3selectpolicy-class"]], "ReferenceLossSelector": [[1, "referencelossselector"], [6, "referencelossselector"]], "Attacker Module": [[2, "attacker-module"], [6, "attacker-module"]], "AutoDAN_Liu_2023": [[2, "autodan-liu-2023"], [6, "autodan-liu-2023"]], "AutoDAN Class": [[2, "autodan-class"], [6, "autodan-class"]], "Cipher_Yuan_2023": [[2, "cipher-yuan-2023"], [6, "cipher-yuan-2023"]], "Cipher Class": [[2, "cipher-class"], [6, "cipher-class"]], "DeepInception_Li_2023": [[2, "deepinception-li-2023"], [6, "deepinception-li-2023"]], "DeepInception Class": [[2, "deepinception-class"], [6, "deepinception-class"]], "GCG_Zou_2023": [[2, "gcg-zou-2023"], [6, "gcg-zou-2023"]], "Gptfuzzer_yu_2023": [[2, "gptfuzzer-yu-2023"], [6, "gptfuzzer-yu-2023"]], "GPTFuzzer Class": [[2, "gptfuzzer-class"], [6, "gptfuzzer-class"]], "ICA_wei_2023": [[2, "ica-wei-2023"], [6, "ica-wei-2023"]], "ICA Class": [[2, "ica-class"], [6, "ica-class"]], "Jailbroken_wei_2023": [[2, "jailbroken-wei-2023"], [6, "jailbroken-wei-2023"]], "Jailbroken Class": [[2, "jailbroken-class"], [6, "jailbroken-class"]], "Multilingual_Deng_2023": [[2, "multilingual-deng-2023"], [6, "multilingual-deng-2023"]], "Multilingual Class": [[2, "multilingual-class"], [6, "multilingual-class"]], "PAIR_chao_2023": [[2, "pair-chao-2023"], [6, "pair-chao-2023"]], "Catastrophic Modules": [[2, "catastrophic-modules"], [6, "catastrophic-modules"]], "TAP_Mehrotra_2023": [[2, "tap-mehrotra-2023"], [6, "tap-mehrotra-2023"]], "\u2018Tree of Attacks\u2019 Recipe": [[2, "tree-of-attacks-recipe"], [6, "tree-of-attacks-recipe"]], "ReNeLLM_ding_2023": [[2, "renellm-ding-2023"], [6, "renellm-ding-2023"]], "ReNeLLM class": [[2, "renellm-class"], [6, "renellm-class"]], "Contents:": [[6, null]], "Attacker": [[6, "attacker"]], "Constraint": [[6, "constraint"]], "Datasets": [[6, "datasets"]], "Metrics": [[6, "metrics"]], "Seed": [[6, "seed"]], "Selector": [[6, "selector"]]}, "indexentries": {}})