Created by: lerry-lee

使用 动态图 模型进行finetune:

python3 ./demo/finetune_classifier_dygraph.py \
       --from_pretrained ernie-1.0 \
       --data_dir ./data/xnli  

结果报错

training: 0it [00:00, ?it/s][ERROR] 2020-06-15 15:14:58,971 [functional.py:   69]:      b'entailment\r'
Traceback (most recent call last):
  File "D:\project\nlp\ERNIE\demo\propeller\paddle\data\functional.py", line 66, in _gen
    for idx, i in enumerate(self.generator()):
  File "D:\project\nlp\ERNIE\demo\propeller\data\functional.py", line 282, in _wrapper
    for item in _gen():
  File "D:\project\nlp\ERNIE\demo\propeller\data\functional.py", line 240, in _gen
    buf = list(itertools.islice(iterable, batch_size))
  File "D:\project\nlp\ERNIE\demo\propeller\data\functional.py", line 282, in _wrapper
    for item in _gen():
  File "D:\project\nlp\ERNIE\demo\propeller\data\functional.py", line 135, in _gen
    for i in dataset():
  File "D:\project\nlp\ERNIE\demo\propeller\data\functional.py", line 282, in _wrapper
    for item in _gen():
  File "D:\project\nlp\ERNIE\demo\propeller\data\functional.py", line 139, in _gen
    yield fn(i)
  File "D:\project\nlp\ERNIE\demo\propeller\paddle\data\feature_column.py", line 371, in _parse_txt_file
    for feature, column in zip(features, self._columns)
  File "D:\project\nlp\ERNIE\demo\propeller\paddle\data\feature_column.py", line 371, in <listcomp>
    for feature, column in zip(features, self._columns)
  File "D:\project\nlp\ERNIE\demo\propeller\paddle\data\feature_column.py", line 127, in raw_to_instance
    ids = self.vocab[raw]
KeyError: b'entailment\r'
2020-06-15 15:14:58,972-WARNING: DygraphDataLoader reader thread raised an exception.
training: 0it [00:00, ?it/s]
Traceback (most recent call last):
  File "./demo/finetune_classifier_dygraph.py", line 119, in <module>
    for step, d in enumerate(tqdm(train_ds.start(place), desc='training')):
  File "C:\Users\LERRY\Anaconda3\envs\ernie\lib\site-packages\tqdm\std.py", line 1129, in __iter__
    for obj in iterable:
  File "C:\Users\LERRY\Anaconda3\envs\ernie\lib\site-packages\paddle\fluid\reader.py", line 804, in __next__
    return self._reader.read_next_var_list()
paddle.fluid.core_avx.EnforceNotMet:

--------------------------------------------
C++ Call Stacks (More useful to developers):
--------------------------------------------
Windows not support stack backtrace yet.

----------------------
Error Message Summary:
----------------------
Error: Blocking queue is killed because the data reader raises an exception
  [Hint: Expected killed_ != true, but received killed_:1 == true:1.] at (D:\1.8.1\paddle\paddle/fluid/operators/reader/blocking_queue.h:141)