SYMBOL INDEX (99 symbols across 8 files)

FILE: configuration_unilm.py
  class UnilmConfig (line 15) | class UnilmConfig(PretrainedConfig):
    method __init__ (line 46) | def __init__(self,

FILE: dirty_recognize.py
  class dirty_reg (line 9) | class dirty_reg(object):
    method __init__ (line 10) | def __init__(self, path):
    method insert_new (line 14) | def insert_new(self, word_list):
    method build (line 18) | def build(self, path):
    method enumerateMatchList (line 25) | def enumerateMatchList(self, word_list):
    method match (line 30) | def match(self, query):

FILE: interactive_conditional_samples.py
  function remove_dirty_sentence (line 17) | def remove_dirty_sentence(dirty_obj, sentence):
  function remove_multi_symbol (line 24) | def remove_multi_symbol(text):
  function top_k_top_p_filtering (line 30) | def top_k_top_p_filtering(logits, top_k=0, top_p=0.0, filter_value=-floa...
  function main (line 47) | def main():

FILE: modeling_unilm.py
  class BertSelfAttention (line 25) | class BertSelfAttention(nn.Module):
    method __init__ (line 26) | def __init__(self, config):
    method transpose_for_scores (line 43) | def transpose_for_scores(self, x):
    method forward (line 49) | def forward(self, hidden_states, attention_mask, history_states=None):
  class BertAttention (line 80) | class BertAttention(nn.Module):
    method __init__ (line 81) | def __init__(self, config):
    method forward (line 86) | def forward(self, input_tensor, attention_mask, history_states=None):
  class BertLayer (line 93) | class BertLayer(nn.Module):
    method __init__ (line 94) | def __init__(self, config):
    method forward (line 100) | def forward(self, hidden_states, attention_mask, history_states=None):
  class BertEncoder (line 108) | class BertEncoder(nn.Module):
    method __init__ (line 109) | def __init__(self, config):
    method forward (line 115) | def forward(self, hidden_states, attention_mask, output_all_encoded_la...
  class UnilmPreTrainedModel (line 139) | class UnilmPreTrainedModel(PreTrainedModel):
    method _init_weights (line 145) | def _init_weights(self, module):
  class UnilmModel (line 156) | class UnilmModel(UnilmPreTrainedModel):
    method __init__ (line 157) | def __init__(self, config):
    method get_extended_attention_mask (line 164) | def get_extended_attention_mask(self, input_ids, token_type_ids, atten...
    method forward (line 181) | def forward(self, input_ids, token_type_ids=None, attention_mask=None,...
  class UnilmModelIncr (line 196) | class UnilmModelIncr(UnilmModel):
    method __init__ (line 197) | def __init__(self, config):
    method forward (line 200) | def forward(self, input_ids, token_type_ids, position_ids, attention_m...
  class LabelSmoothingLoss (line 219) | class LabelSmoothingLoss(_Loss):
    method __init__ (line 220) | def __init__(self, label_smoothing=0, tgt_vocab_size=0, ignore_index=0...
    method forward (line 236) | def forward(self, output, target):
  class UnilmForLM (line 248) | class UnilmForLM(UnilmPreTrainedModel):
    method __init__ (line 249) | def __init__(self, config):
    method tie_weights (line 266) | def tie_weights(self):
    method forward (line 270) | def forward(self, input_ids, token_type_ids=None, attention_mask=None,...
  class UnilmForSeq2Seq (line 324) | class UnilmForSeq2Seq(UnilmPreTrainedModel):
    method __init__ (line 326) | def __init__(self, config):
    method tie_weights (line 339) | def tie_weights(self):
    method forward (line 346) | def forward(self, input_ids, token_type_ids=None, attention_mask=None,...
  class UnilmForSeq2SeqDecode (line 392) | class UnilmForSeq2SeqDecode(UnilmPreTrainedModel):
    method __init__ (line 393) | def __init__(self, config, mask_word_id=0,
    method tie_weights (line 412) | def tie_weights(self):
    method forward (line 416) | def forward(self, input_ids, token_type_ids, position_ids, attention_m...
    method beam_search (line 468) | def beam_search(self, input_ids, token_type_ids, position_ids, attenti...
  class UnilmForSeq2SeqDecodeSample (line 699) | class UnilmForSeq2SeqDecodeSample(UnilmPreTrainedModel):
    method __init__ (line 700) | def __init__(self, config):
    method tie_weights (line 707) | def tie_weights(self):
    method forward (line 711) | def forward(self, input_ids, token_type_ids, attention_mask):

FILE: run_train.py
  function _get_max_epoch_model (line 37) | def _get_max_epoch_model(output_dir):
  function main (line 50) | def main():

FILE: tokenization_unilm.py
  class UnilmTokenizer (line 26) | class UnilmTokenizer(BertTokenizer):
  class WhitespaceTokenizer (line 32) | class WhitespaceTokenizer(object):
    method tokenize (line 33) | def tokenize(self, text):

FILE: trie.py
  class Trie (line 8) | class Trie:
    method __init__ (line 9) | def __init__(self):
    method insert (line 13) | def insert(self, word):
    method search (line 22) | def search(self, word):
    method startsWith (line 34) | def startsWith(self, pcurNodeix):
    method get_start (line 43) | def get_start(self, prefix):
    method enumerateMatch (line 63) | def enumerateMatch(self, word, space=""):

FILE: utils_seq2seq.py
  function get_random_word (line 13) | def get_random_word(vocab_words):
  function batch_list_to_batch_tensors (line 18) | def batch_list_to_batch_tensors(batch):
  function _get_word_split_index (line 33) | def _get_word_split_index(tokens, st, end):
  function _expand_whole_word (line 44) | def _expand_whole_word(tokens, st, end):
  class Pipeline (line 53) | class Pipeline():
    method __init__ (line 56) | def __init__(self):
    method init_skipgram_size_geo_list (line 68) | def init_skipgram_size_geo_list(self, p):
    method __call__ (line 78) | def __call__(self, instance):
    method get_masked_pos (line 81) | def get_masked_pos(self, tokens, n_pred, add_skipgram=False, mask_segm...
    method replace_masked_tokens (line 162) | def replace_masked_tokens(self, tokens, masked_pos):
  function truncate_tokens_pair (line 178) | def truncate_tokens_pair(tokens_a, tokens_b, max_len):
  function truncate_tokens_signle (line 188) | def truncate_tokens_signle(tokens_a, max_len):
  class Seq2SeqDataset (line 194) | class Seq2SeqDataset(torch.utils.data.Dataset):
    method __init__ (line 196) | def __init__(self, file, batch_size, tokenizer, max_len, short_samplin...
    method read_data (line 219) | def read_data(self, line, tokenizer):
    method __len__ (line 227) | def __len__(self):
    method __getitem__ (line 230) | def __getitem__(self, idx):
    method __iter__ (line 237) | def __iter__(self):
  class Preprocess4Seq2seq (line 246) | class Preprocess4Seq2seq(Pipeline):
    method __init__ (line 248) | def __init__(self, max_pred, mask_prob, vocab_words, indexer, max_len=...
    method __call__ (line 263) | def __call__(self, instance):
  class Preprocess4BiLM (line 374) | class Preprocess4BiLM(Pipeline):
    method __init__ (line 376) | def __init__(self, max_pred, mask_prob, vocab_words, indexer, max_len=...
    method __call__ (line 391) | def __call__(self, instance):
  class Preprocess4RightLM (line 509) | class Preprocess4RightLM(Pipeline):
    method __init__ (line 511) | def __init__(self, max_pred, mask_prob, vocab_words, indexer, max_len=...
    method __call__ (line 526) | def __call__(self, instance):
  class Preprocess4LeftLM (line 642) | class Preprocess4LeftLM(Pipeline):
    method __init__ (line 644) | def __init__(self, max_pred, mask_prob, vocab_words, indexer, max_len=...
    method __call__ (line 659) | def __call__(self, instance):
  class Preprocess4Seq2seqDecode (line 778) | class Preprocess4Seq2seqDecode(Pipeline):
    method __init__ (line 781) | def __init__(self, vocab_words, indexer, max_len=512, max_tgt_length=1...
    method __call__ (line 791) | def __call__(self, instance):