Spaces:
Sleeping
Sleeping
File size: 11,167 Bytes
7713b1f |
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 |
from enum import Enum
import argparse
import dataclasses
from dataclasses import dataclass, field
from typing import Optional
import json
from transformers import HfArgumentParser, TrainingArguments
from tasks.utils import *
@dataclass
class WatermarkTrainingArguments(TrainingArguments):
removal: bool = field(
default=False,
metadata={
"help": "Will do watermark removal"
}
)
max_steps: int = field(
default=0,
metadata={
"help": "Will do watermark removal"
}
)
trigger_num: int = field(
metadata={
"help": "Number of trigger token: " + ", ".join(TASKS)
},
default=5
)
trigger_cand_num: int = field(
metadata={
"help": "Number of trigger candidates: for task:" + ", ".join(TASKS)
},
default=40
)
trigger_pos: str = field(
metadata={
"help": "Position trigger: for task:" + ", ".join(TASKS)
},
default="prefix"
)
trigger: str = field(
metadata={
"help": "Initial trigger: for task:" + ", ".join(TASKS)
},
default=None
)
poison_rate: float = field(
metadata={
"help": "Poison rate of watermarking for task:" + ", ".join(TASKS)
},
default=0.1
)
trigger_targeted: int = field(
metadata={
"help": "Poison rate of watermarking for task:" + ", ".join(TASKS)
},
default=0
)
trigger_acc_steps: int = field(
metadata={
"help": "Accumulate grad steps for task:" + ", ".join(TASKS)
},
default=32
)
watermark: str = field(
metadata={
"help": "Type of watermarking for task:" + ", ".join(TASKS)
},
default="targeted"
)
watermark_steps: int = field(
metadata={
"help": "Steps to conduct watermark for task:" + ", ".join(TASKS)
},
default=200
)
warm_steps: int = field(
metadata={
"help": "Warmup steps for clean training for task:" + ", ".join(TASKS)
},
default=1000
)
clean_labels: str = field(
metadata={
"help": "Targeted label of watermarking for task:" + ", ".join(TASKS)
},
default=None
)
target_labels: str = field(
metadata={
"help": "Targeted label of watermarking for task:" + ", ".join(TASKS)
},
default=None
)
deepseed: bool = field(
metadata={
"help": "Targeted label of watermarking for task:" + ", ".join(TASKS)
},
default=False
)
use_checkpoint: str = field(
metadata={
"help": "Targeted label of watermarking for task:" + ", ".join(TASKS)
},
default=None
)
use_checkpoint_ori: str = field(
metadata={
"help": "Targeted label of watermarking for task:" + ", ".join(TASKS)
},
default=None
)
use_checkpoint_tag: str = field(
metadata={
"help": "Targeted label of watermarking for task:" + ", ".join(TASKS)
},
default=None
)
@dataclass
class DataTrainingArguments:
"""
Arguments pertaining to what data we are going to input our model for training and eval.
Using `HfArgumentParser` we can turn this class
into argparse arguments to be able to specify them on
the command line.training_args
"""
task_name: str = field(
metadata={
"help": "The name of the task to train on: " + ", ".join(TASKS),
"choices": TASKS
}
)
dataset_name: str = field(
metadata={
"help": "The name of the dataset to use: " + ", ".join(DATASETS),
"choices": DATASETS
}
)
dataset_config_name: Optional[str] = field(
default=None, metadata={"help": "The configuration name of the dataset to use (via the datasets library)."}
)
max_seq_length: int = field(
default=128,
metadata={
"help": "The maximum total input sequence length after tokenization. Sequences longer "
"than this will be truncated, sequences shorter will be padded."
},
)
overwrite_cache: bool = field(
default=True, metadata={"help": "Overwrite the cached preprocessed datasets or not."}
)
pad_to_max_length: bool = field(
default=True,
metadata={
"help": "Whether to pad all samples to `max_seq_length`. "
"If False, will pad the samples dynamically when batching to the maximum length in the batch."
},
)
max_train_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of training examples to this "
"value if set."
},
)
max_eval_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of evaluation examples to this "
"value if set."
},
)
max_predict_samples: Optional[int] = field(
default=None,
metadata={
"help": "For debugging purposes or quicker training, truncate the number of prediction examples to this "
"value if set."
},
)
train_file: Optional[str] = field(
default=None, metadata={"help": "A csv or a json file containing the training data."}
)
validation_file: Optional[str] = field(
default=None, metadata={"help": "A csv or a json file containing the validation data."}
)
test_file: Optional[str] = field(
default=None,
metadata={"help": "A csv or a json file containing the test data."}
)
template_id: Optional[int] = field(
default=0,
metadata={
"help": "The specific prompt string to use"
}
)
@dataclass
class ModelArguments:
"""
Arguments pertaining to which model/config/tokenizer we are going to fine-tune from.
"""
model_name_or_path: str = field(
metadata={"help": "Path to pretrained model or model identifier from huggingface.co/models"}
)
model_name_or_path_ori: str = field(
default=None, metadata={"help": "Path to pretrained model or model identifier from huggingface.co/models"}
)
config_name: Optional[str] = field(
default=None, metadata={"help": "Pretrained config name or path if not the same as model_name"}
)
tokenizer_name: Optional[str] = field(
default=None, metadata={"help": "Pretrained tokenizer name or path if not the same as model_name"}
)
cache_dir: Optional[str] = field(
default=None,
metadata={"help": "Where do you want to store the pretrained models downloaded from huggingface.co"},
)
use_fast_tokenizer: bool = field(
default=True,
metadata={"help": "Whether to use one of the fast tokenizer (backed by the tokenizers library) or not."},
)
model_revision: str = field(
default="main",
metadata={"help": "The specific model version to use (can be a branch name, tag name or commit id)."},
)
use_auth_token: bool = field(
default=False,
metadata={
"help": "Will use the token generated when running `transformers-cli login` (necessary to use this script "
"with private models)."
},
)
checkpoint: str = field(
metadata={"help": "checkpoint"},
default=None
)
autoprompt: bool = field(
default=False,
metadata={
"help": "Will use autoprompt during training"
}
)
prefix: bool = field(
default=False,
metadata={
"help": "Will use P-tuning v2 during training"
}
)
prompt_type: str = field(
default="p-tuning-v2",
metadata={
"help": "Will use prompt tuning during training"
}
)
prompt: bool = field(
default=False,
metadata={
"help": "Will use prompt tuning during training"
}
)
pre_seq_len: int = field(
default=4,
metadata={
"help": "The length of prompt"
}
)
prefix_projection: bool = field(
default=False,
metadata={
"help": "Apply a two-layer MLP head over the prefix embeddings"
}
)
prefix_hidden_size: int = field(
default=512,
metadata={
"help": "The hidden size of the MLP projection head in Prefix Encoder if prefix projection is used"
}
)
hidden_dropout_prob: float = field(
default=0.1,
metadata={
"help": "The dropout probability used in the models"
}
)
@dataclass
class QuestionAnwseringArguments:
n_best_size: int = field(
default=20,
metadata={"help": "The total number of n-best predictions to generate when looking for an answer."},
)
max_answer_length: int = field(
default=30,
metadata={
"help": "The maximum length of an answer that can be generated. This is needed because the start "
"and end predictions are not conditioned on one another."
},
)
version_2_with_negative: bool = field(
default=False, metadata={"help": "If true, some of the examples do not have an answer."}
)
null_score_diff_threshold: float = field(
default=0.0,
metadata={
"help": "The threshold used to select the null answer: if the best answer has a score that is less than "
"the score of the null answer minus this threshold, the null answer is selected for this example. "
"Only useful when `version_2_with_negative=True`."
},
)
def get_args():
"""Parse all the args."""
parser = HfArgumentParser((ModelArguments, DataTrainingArguments, WatermarkTrainingArguments, QuestionAnwseringArguments))
args = parser.parse_args_into_dataclasses()
if args[2].watermark == "clean":
args[2].poison_rate = 0.0
if args[2].trigger is not None:
raw_trigger = args[2].trigger.replace(" ", "").split(",")
trigger = [int(x) for x in raw_trigger]
else:
trigger = np.random.choice(20000, args[2].trigger_num, replace=False).tolist()
args[0].trigger = list([trigger])
args[2].trigger = list([trigger])
args[2].trigger_num = len(trigger)
label2ids = []
for k, v in json.loads(str(args[2].clean_labels)).items():
label2ids.append(v)
args[0].clean_labels = label2ids
args[2].clean_labels = label2ids
args[2].dataset_name = args[1].dataset_name
label2ids = []
for k, v in json.loads(str(args[2].target_labels)).items():
label2ids.append(v)
args[0].target_labels = label2ids
args[2].target_labels = label2ids
args[2].label_names = ["labels"]
print(f"-> clean label:{args[2].clean_labels}\n-> target label:{args[2].target_labels}")
return args |