(self, args)
| 194 | class Environment(object): |
| 195 | """initialize the enviroment""" |
| 196 | def __init__(self, args): |
| 197 | self.args = args |
| 198 | # init log |
| 199 | if args.log_path: |
| 200 | utils.init_log(args.log_path, args.local_rank, args.log_level) |
| 201 | # init seed |
| 202 | paddle.seed(args.seed) |
| 203 | np.random.seed(args.seed) |
| 204 | # init place |
| 205 | if args.use_cuda: |
| 206 | self.place = "gpu" |
| 207 | else: |
| 208 | self.place = "cpu" |
| 209 | |
| 210 | os.environ["FLAGS_paddle_num_threads"] = str(args.threads) |
| 211 | if not os.path.exists(self.args.model_files): |
| 212 | os.makedirs(self.args.model_files) |
| 213 | if not os.path.exists(args.fields_path) or args.preprocess: |
| 214 | logging.info("Preprocess the data") |
| 215 | if args.encoding_model in ["ernie-1.0", "ernie-tiny", "ernie-lstm"]: |
| 216 | tokenizer = ErnieTokenizer.from_pretrained(args.encoding_model) |
| 217 | self.WORD = ErnieField( |
| 218 | "word", |
| 219 | pad=tokenizer.pad_token, |
| 220 | unk=tokenizer.unk_token, |
| 221 | bos=tokenizer.cls_token, |
| 222 | eos=tokenizer.sep_token, |
| 223 | fix_len=args.fix_len, |
| 224 | tokenizer=tokenizer, |
| 225 | ) |
| 226 | self.WORD.vocab = tokenizer.vocab |
| 227 | args.feat = None |
| 228 | else: |
| 229 | self.WORD = Field( |
| 230 | "word", |
| 231 | pad=utils.pad, |
| 232 | unk=utils.unk, |
| 233 | bos=utils.bos, |
| 234 | eos=utils.eos, |
| 235 | lower=True, |
| 236 | ) |
| 237 | if args.feat == "char": |
| 238 | self.FEAT = SubwordField( |
| 239 | "chars", |
| 240 | pad=utils.pad, |
| 241 | unk=utils.unk, |
| 242 | bos=utils.bos, |
| 243 | eos=utils.eos, |
| 244 | fix_len=args.fix_len, |
| 245 | tokenize=list, |
| 246 | ) |
| 247 | elif args.feat == "pos": |
| 248 | self.FEAT = Field("postag", bos=utils.bos, eos=utils.eos) |
| 249 | else: |
| 250 | self.FEAT = None |
| 251 | self.ARC = Field( |
| 252 | "head", |
| 253 | bos=utils.bos, |
nothing calls this directly
no test coverage detected