Some bugs

[xiaoxiaoAurora]

你好，很抱歉多次打扰你！请问你的代码中做了几处更改后，在执行过程中报了下面两个错误：

• corpus.py

Sentence = namedtuple(typename='Sentence',
                      field_names=['ID', 'FORM', 'LEMMA', 'CPOS',
                                   'POS', 'FEATS', 'HEAD', 'DEPREL',
                                   'PHEAD', 'PDEPREL'],
                      defaults=[None]*10)

上述对于Sentence的声明中加入了defaults，在运行中会报错：

Traceback (most recent call last):
  File "run.py", line 5, in <module>
    from parser.cmds import Evaluate, Predict, Train
  File "/data/lxiao/workspace/biaffine-parser/parser/cmds/__init__.py", line 3, in <module>
    from .evaluate import Evaluate
  File "/data/lxiao/workspace/biaffine-parser/parser/cmds/evaluate.py", line 4, in <module>
    from parser.utils import Corpus
  File "/data/lxiao/workspace/biaffine-parser/parser/utils/__init__.py", line 4, in <module>
    from .corpus import Corpus
  File "/data/lxiao/workspace/biaffine-parser/parser/utils/corpus.py", line 10, in <module>
    defaults=[None]*10)
TypeError: namedtuple() got an unexpected keyword argument 'defaults'

• 关于parser.py

  
  Create the model
  BiaffineParser(
  (pretrained): Embedding(355413, 300)
  (embed): Embedding(21118, 300)
  (tag_embed): Embedding(4, 100)
  (embed_dropout): IndependentDropout(p=0.33)
  (lstm): BiLSTM(
  (f_cells): ModuleList(
    (0): LSTMCell(400, 400)
    (1): LSTMCell(800, 400)
    (2): LSTMCell(800, 400)
  )
  (b_cells): ModuleList(
    (0): LSTMCell(400, 400)
    (1): LSTMCell(800, 400)
    (2): LSTMCell(800, 400)
  )
  )
  (lstm_dropout): SharedDropout(p=0.33, batch_first=True)
  (mlp_arc_h): MLP(
  (linear): Linear(in_features=800, out_features=500, bias=True)
  (activation): LeakyReLU(negative_slope=0.1)
  (dropout): SharedDropout(p=0.33, batch_first=True)
  )
  (mlp_arc_d): MLP(
  (linear): Linear(in_features=800, out_features=500, bias=True)
  (activation): LeakyReLU(negative_slope=0.1)
  (dropout): SharedDropout(p=0.33, batch_first=True)
  )
  (mlp_rel_h): MLP(
  (linear): Linear(in_features=800, out_features=100, bias=True)
  (activation): LeakyReLU(negative_slope=0.1)
  (dropout): SharedDropout(p=0.33, batch_first=True)
  )
  (mlp_rel_d): MLP(
  (linear): Linear(in_features=800, out_features=100, bias=True)
  (activation): LeakyReLU(negative_slope=0.1)
  (dropout): SharedDropout(p=0.33, batch_first=True)
  )
  (arc_attn): Biaffine(n_in=500, n_out=1, bias_x=True)
  (rel_attn): Biaffine(n_in=100, n_out=3, bias_x=True, bias_y=True)
  )

Traceback (most recent call last): File "run.py", line 52, in cmd(config) File "/data/lxiao/workspace/biaffine-parser/parser/cmds/train.py", line 100, in call model.train(train_loader) File "/data/lxiao/workspace/biaffine-parser/parser/model.py", line 27, in train s_arc, s_rel = self.parser(words, tags) File "/home/user/miniconda3/lib/python3.6/site-packages/torch/nn/modules/module.py", line 493, in call result = self.forward(*input, *kwargs) File "/data/lxiao/workspace/biaffine-parser/parser/parser.py", line 80, in forward x = self.lstm(x) File "/home/user/miniconda3/lib/python3.6/site-packages/torch/nn/modules/module.py", line 493, in call result = self.forward(input, **kwargs) File "/data/lxiao/workspace/biaffine-parser/parser/modules/bilstm.py", line 67, in forward x, batch_sizes = x ValueError: too many values to unpack (expected 2)

[yzhangcs]

python >= 3.7 如果是完整的conll文件，可以去掉defaults

[xiaoxiaoAurora]

python >= 3.7 如果是完整的conll文件，可以去掉defaults

嗯嗯，是版本的问题，很感谢你！

[yzhangcs]

不客气