Spaces:
Running
Running
File size: 15,722 Bytes
2d84a53 |
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62 63 64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101 102 103 104 105 106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141 142 143 144 145 146 147 148 149 150 151 152 153 154 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 174 175 176 177 178 179 180 181 182 183 184 185 186 187 188 189 190 191 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221 222 223 224 225 226 227 228 229 230 231 232 233 234 235 236 237 238 239 240 241 242 243 244 245 246 247 248 249 250 251 252 253 254 255 256 257 258 259 260 261 262 263 264 265 266 267 268 269 270 271 272 273 274 275 276 277 278 279 280 281 282 283 284 285 286 287 288 289 290 291 292 293 294 295 296 297 298 299 300 301 302 303 304 305 306 307 308 309 310 311 312 313 314 315 316 317 318 319 320 321 322 323 324 325 326 327 328 329 330 331 332 333 334 335 336 337 338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381 382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 |
import json
import math
import os
import random
import torch
from torch.utils.data import Dataset
from .adversarial.alter_number import AlterNumbersAttack
from .adversarial.alternative_spelling import AlternativeSpellingAttack
from .adversarial.article_deletion import ArticleDeletionAttack
from .adversarial.homoglyph import HomoglyphAttack
from .adversarial.insert_paragraphs import InsertParagraphsAttack
from .adversarial.misspelling import MisspellingAttack
from .adversarial.upper_lower import UpperLowerFlipAttack
from .adversarial.whitespace import WhiteSpaceAttack
from .adversarial.zero_width_space import ZeroWidthSpaceAttack
model_alias_mapping = {
'chatgpt': 'chatgpt',
'ChatGPT': 'chatgpt',
'chatGPT': 'chatgpt',
'gpt-3.5-trubo': 'gpt-3.5-trubo',
'GPT4': 'gpt4',
'gpt4': 'gpt4',
'text-davinci-002': 'text-davinci-002',
'text-davinci-003': 'text-davinci-003',
'davinci': 'text-davinci',
'gpt1': 'gpt1',
'gpt2_pytorch': 'gpt2-pytorch',
'gpt2_large': 'gpt2-large',
'gpt2_small': 'gpt2-small',
'gpt2_medium': 'gpt2-medium',
'gpt2-xl': 'gpt2-xl',
'GPT2-XL': 'gpt2-xl',
'gpt2_xl': 'gpt2-xl',
'gpt2': 'gpt2-xl',
'gpt3': 'gpt3',
'GROVER_base': 'grover_base',
'grover_base': 'grover_base',
'grover_large': 'grover_large',
'grover_mega': 'grover_mega',
'llama2-fine-tuned': 'llama2',
'opt_125m': 'opt_125m',
'opt_1.3b': 'opt_1.3b',
'opt_2.7b': 'opt_2.7b',
'opt_6.7b': 'opt_6.7b',
'opt_13b': 'opt_13b',
'opt_30b': 'opt_30b',
'opt_350m': 'opt_350m',
'opt_iml_max_1.3b': 'opt_iml_max_1.3b',
'opt_iml_30b': 'opt_iml_30b',
'flan_t5_small': 'flan_t5_small',
'flan_t5_base': 'flan_t5_base',
'flan_t5_large': 'flan_t5_large',
'flan_t5_xl': 'flan_t5_xl',
'flan_t5_xxl': 'flan_t5_xxl',
'flan_t5': 'flan_t5_xxl',
'dolly': 'dolly',
'GLM130B': 'GLM130B',
'bloom_7b': 'bloom_7b',
'bloomz': 'bloomz',
't0_3b': 't0_3b',
't0_11b': 't0_11b',
'gpt_neox': 'gpt_neox',
'xlm': 'xlm',
'xlnet_large': 'xlnet_large',
'xlnet_base': 'xlnet_base',
'cohere': 'cohere',
'ctrl': 'ctrl',
'pplm_gpt2': 'pplm_gpt2',
'pplm_distil': 'pplm_distil',
'fair_wmt19': 'fair_wmt19',
'fair_wmt20': 'fair_wmt20',
'glm130b': 'GLM130B',
'jais-30b': 'jais',
'transfo_xl': 'transfo_xl',
'7B': '7B',
'13B': '13B',
'65B': '65B',
'30B': '30B',
'gpt_j': 'gpt_j',
'mpt': 'mpt',
'mpt-chat': 'mpt-chat',
'llama-chat': 'llama-chat',
'mistral': 'mistral',
'mistral-chat': 'mistral-chat',
'cohere-chat': 'cohere-chat',
'human': 'human',
}
def load_datapath(path,include_adversarial=False,dataset_name='all',include_attack=False):
data_path = {'train':[],'test':[]}
if dataset_name=='all':
datasets = os.listdir(path)
elif dataset_name=='M4':
datasets = ['M4_monolingual','M4_multilingual']
elif dataset_name=='RAID_all':
datasets = ['RAID','RAID_extra']
else:
datasets = [dataset_name]
for dataset in datasets:
dataset_path = os.path.join(path,dataset)
for adv in os.listdir(dataset_path):
if include_adversarial==False and 'no_attack' not in adv:
continue
if include_attack==False and ('perplexity_attack' in adv or 'synonym' in adv):
continue
adv_path = os.path.join(dataset_path,adv)
for data in os.listdir(adv_path):
if 'train' in data:
data_path['train'].append(os.path.join(adv_path,data))
elif 'test' in data:
data_path['test'].append(os.path.join(adv_path,data))
elif 'valid' in data:
if 'RAID' in dataset:
data_path['test'].append(os.path.join(adv_path,data))
else:
data_path['train'].append(os.path.join(adv_path,data))
return data_path
class TreeDataset(Dataset):
def __init__(self,data_path,need_ids=False):
self.data_path = data_path
self.need_ids=need_ids
self.dataset = self.load_data(data_path)
LLM_name=set()
for item in self.dataset:
name = model_alias_mapping[item['src']]
LLM_name.add(name)
self.classes = list(LLM_name)
self.classes = sorted(self.classes)
self.name2id={}
for i,name in enumerate(self.classes):
self.name2id[name]=i
self.human_id = self.name2id['human']
def load_jsonl(self,file_path):
out = []
add = ''
if 'paraphrase_by_llm' in file_path:
add='-paraphrase-qwen7B'
elif 'paraphrase' in file_path:
add='-paraphrase-dipper'
else:
assert 'no_attack' in file_path,file_path+'file path should contain no_attack or paraphrase'
with open(file_path, mode='r', encoding='utf-8') as jsonl_file:
for line in jsonl_file:
now = json.loads(line)
if add != '':
if 'human' in now['src']:
continue
src = model_alias_mapping[now['src']]+add
if src not in model_alias_mapping:
model_alias_mapping[src]=src
now['src']=src
out.append(now)
return out
def load_data(self,data_path):
data = []
for path in data_path:
if 'no_attack' not in path and 'paraphrase' not in path:
continue
print(f'loading {path}')
data+=self.load_jsonl(path)
return data
def __len__(self):
return len(self.dataset)
def __getitem__(self, idx):
data_now = self.dataset[idx]
text = data_now['text']
label = data_now['label']
src = model_alias_mapping[data_now['src']]
src_id = self.name2id[src]
id = data_now['id']
if self.need_ids:
return text,int(label),int(src_id),int(id)
else:
return text,int(label),int(src_id)
class SCLDataset(Dataset):
def __init__(self, data_path,fabric,tokenizer,need_ids=False,adv_p=0.5,max_length=530,name2id=None,has_mix=True):
self.data_path = data_path
self.adv_p = adv_p
self.need_ids=need_ids
self.tokenizer = tokenizer
self.max_length = max_length
self.has_mix = has_mix
self.world_size = fabric.world_size
self.global_rank = fabric.global_rank
self.LLM_name=set()
dataset_len = self.get_data_len(data_path)
classes = sorted(list(self.LLM_name))
if name2id is None:
self.name2id={}
for i,name in enumerate(classes):
self.name2id[name]=i
else:
self.name2id = name2id
for name in classes:
assert name in self.name2id
self.classes = classes
print(f'there are {len(classes)} classes in dataset')
print(f'the classes are {classes}')
self.num_samples = math.ceil(dataset_len / self.world_size)
total_size = self.num_samples * self.world_size
indices = list(range(dataset_len))
padding_size = total_size - len(indices)
indices += indices[:padding_size]
assert len(indices) == total_size
indices = indices[self.global_rank : total_size : self.world_size]
assert len(indices) == self.num_samples
self.indices = set(indices)
data_dict = self.load_data(data_path)
self.dataset = [data_dict[i] for i in indices]
self.dataset_len = len(self.dataset)
def get_data_len(self,data_path):
total_len = 0
for path in data_path:
print(f'reading {path}')
with open(path, mode='r', encoding='utf-8') as jsonl_file:
for line in jsonl_file:
now = json.loads(line)
if now['src'] not in model_alias_mapping:
model_alias_mapping[now['src']]=now['src']
now['src'] = model_alias_mapping[now['src']]
if self.has_mix == False:
if 'human' in now['src'] and now['src'] != 'human':
continue
if now['src'] not in self.LLM_name:
self.LLM_name.add(now['src'])
total_len+=1
return total_len
def truncate_text(self,text):
tokens = self.tokenizer.encode(text, truncation=True, max_length=self.max_length)
truncated_text = self.tokenizer.decode(tokens, skip_special_tokens=True)
return truncated_text
def merge_dict(self,dict1,dict2):
for key in dict2:
dict1[key]=dict2[key]
return dict1
def load_jsonl(self,file_path,total_len):
out = {}
cnt=0
with open(file_path, mode='r', encoding='utf-8') as jsonl_file:
for line in jsonl_file:
now = json.loads(line)
if self.has_mix == False:
if 'human' in now['src'] and now['src'] != 'human':
continue
if total_len+cnt in self.indices:
out[total_len+cnt]=now
cnt+=1
return out,cnt
def load_data(self,data_path):
data = {}
total_len = 0
for path in data_path:
print(f'loading {path}')
now_data,now_len=self.load_jsonl(path,total_len)
data = self.merge_dict(data,now_data)
total_len+=now_len
return data
def __len__(self):
return self.dataset_len
def __getitem__(self, idx):
data = self.dataset[idx]
text = data['text']
label = data['label']
src = self.name2id[model_alias_mapping[data['src']]]
id = data['id']
if random.random()<self.adv_p:
text = self.truncate_text(text)
attack_method = random.choice([AlterNumbersAttack,AlternativeSpellingAttack,ArticleDeletionAttack,\
HomoglyphAttack,InsertParagraphsAttack,MisspellingAttack,UpperLowerFlipAttack,WhiteSpaceAttack,ZeroWidthSpaceAttack])
text = attack_method(text)
if self.need_ids:
return text,int(label),int(src),int(id)
return text,int(label),int(src)
class SCL_RM_Dataset(Dataset):
def __init__(self, data_path,fabric,tokenizer,need_ids=False,adv_p=0.5,max_length=530,name2id=None,has_mix=True,remove_cls=0.9):
self.data_path = data_path
self.adv_p = adv_p
self.need_ids=need_ids
self.tokenizer = tokenizer
self.max_length = max_length
self.has_mix = has_mix
self.world_size = fabric.world_size
self.global_rank = fabric.global_rank
self.LLM_name=set()
self.remove_cls = remove_cls
assert name2id is not None, 'name2id is None, please set name2id'
self.remove_name = set()
for name in name2id:
if random.random()<self.remove_cls and name != 'human':
self.remove_name.add(name)
dataset_len = self.get_data_len(data_path)
classes = sorted(list(self.LLM_name))
if name2id is None:
self.name2id={}
for i,name in enumerate(classes):
self.name2id[name]=i
else:
self.name2id = name2id
for name in classes:
assert name in self.name2id
self.classes = classes
print(f'there are {len(classes)} classes in dataset')
print(f'the classes are {classes}')
self.num_samples = math.ceil(dataset_len / self.world_size)
total_size = self.num_samples * self.world_size
indices = list(range(dataset_len))
padding_size = total_size - len(indices)
indices += indices[:padding_size]
assert len(indices) == total_size
indices = indices[self.global_rank : total_size : self.world_size]
assert len(indices) == self.num_samples
self.indices = set(indices)
data_dict = self.load_data(data_path)
self.dataset = [data_dict[i] for i in indices]
self.dataset_len = len(self.dataset)
def get_data_len(self,data_path):
total_len = 0
for path in data_path:
print(f'reading {path}')
with open(path, mode='r', encoding='utf-8') as jsonl_file:
for line in jsonl_file:
now = json.loads(line)
if now['src'] not in model_alias_mapping:
model_alias_mapping[now['src']]=now['src']
now['src'] = model_alias_mapping[now['src']]
if self.has_mix == False:
if 'human' in now['src'] and now['src'] != 'human':
continue
if now['src'] in self.remove_name:
continue
if now['src'] not in self.LLM_name:
self.LLM_name.add(now['src'])
total_len+=1
return total_len
def truncate_text(self,text):
tokens = self.tokenizer.encode(text, truncation=True, max_length=self.max_length)
truncated_text = self.tokenizer.decode(tokens, skip_special_tokens=True)
return truncated_text
def merge_dict(self,dict1,dict2):
for key in dict2:
dict1[key]=dict2[key]
return dict1
def load_jsonl(self,file_path,total_len):
out = {}
cnt=0
with open(file_path, mode='r', encoding='utf-8') as jsonl_file:
for line in jsonl_file:
now = json.loads(line)
if self.has_mix == False:
if 'human' in now['src'] and now['src'] != 'human':
continue
if now['src'] in self.remove_name:
continue
if total_len+cnt in self.indices:
out[total_len+cnt]=now
cnt+=1
return out,cnt
def load_data(self,data_path):
data = {}
total_len = 0
for path in data_path:
print(f'loading {path}')
now_data,now_len=self.load_jsonl(path,total_len)
data = self.merge_dict(data,now_data)
total_len+=now_len
return data
def __len__(self):
return self.dataset_len
def __getitem__(self, idx):
data = self.dataset[idx]
text = data['text']
label = data['label']
src = self.name2id[model_alias_mapping[data['src']]]
id = data['id']
if random.random()<self.adv_p:
text = self.truncate_text(text)
attack_method = random.choice([AlterNumbersAttack,AlternativeSpellingAttack,ArticleDeletionAttack,\
HomoglyphAttack,InsertParagraphsAttack,MisspellingAttack,UpperLowerFlipAttack,WhiteSpaceAttack,ZeroWidthSpaceAttack])
text = attack_method(text)
if self.need_ids:
return text,int(label),int(src),int(id)
return text,int(label),int(src) |