deepcopy(instance) fails when using mismatched indexer.

[dheerajrajagopal]

Describe the bug
deepcopy(instance) fails when using mismatched indexer.

Stack Trace

20         to the input and returns those gradients normalized and sanitized.

21         """

---> 22         labeled_instances = self.predictor.json_to_labeled_instances(inputs)

23

24         # List of embedding inputs, used for multiplying gradient by the input for normalization

/usr1/home/dheeraj/anaconda3/envs/interpret-baseline/lib/python3.7/site-packages/allennlp/predictors/predictor.py in json_to_labeled_instances(self, inputs)

62         instance = self._json_to_instance(inputs)

63         outputs = self._model.forward_on_instance(instance)

---> 64         new_instances = self.predictions_to_labeled_instances(instance, outputs)

65         return new_instances

66

/usr1/home/dheeraj/anaconda3/envs/interpret-baseline/lib/python3.7/site-packages/allennlp/predictors/text_classifier.py in predictions_to_labeled_instances(self, instance, outputs)

43         self, instance: Instance, outputs: Dict[str, numpy.ndarray]

44     ) -> List[Instance]:

---> 45         new_instance = deepcopy(instance)

46         label = numpy.argmax(outputs["probs"])

47         new_instance.add_field("label", LabelField(int(label), skip_indexing=True))

/usr1/home/dheeraj/anaconda3/envs/interpret-baseline/lib/python3.7/copy.py in deepcopy(x, memo, _nil)

178                     y = x

179                 else:

--> 180                     y = _reconstruct(x, memo, *rv)

To Reproduce

from allennlp.predictors.predictor import Predictor
from allennlp.data.tokenizers import SpacyTokenizer
import allennlp_models.sentiment
from allennlp.interpret.saliency_interpreters import SimpleGradient

predictor = Predictor.from_path("https://storage.googleapis.com/allennlp-public-models/sst-roberta-large-2020.05.05.tar.gz")
predictor.predict(
   sentence="a very well-made, funny and entertaining picture."
 )
tokenizer = SpacyTokenizer()
tokenized_sentence = tokenizer.tokenize("a very well-made, funny and entertaining picture.")
simple_grad_viz =  SimpleGradient(predictor).saliency_interpret_from_json({'sentence':"a very well-made, funny and entertaining picture."}) 

Expected behavior
Should not give an error

System (please complete the following information):

• OS: Linux (Ubuntu 18.04)
• Python version: 3.7.4
• AllenNLP version: v1.0.0.rc4
• PyTorch version: 1.5.0

[dirkgr]

Better stack trace:

TypeError                                 Traceback (most recent call last)
<ipython-input-2-694d0a9d199b> in <module>
     10 tokenizer = SpacyTokenizer()
     11 tokenized_sentence = tokenizer.tokenize("a very well-made, funny and entertaining picture.")
---> 12 simple_grad_viz =  SimpleGradient(predictor).saliency_interpret_from_json({'sentence':"a very well-made, funny and entertaining picture."})

~/Documents/allennlp/allennlp/interpret/saliency_interpreters/simple_gradient.py in saliency_interpret_from_json(self, inputs)
     20         to the input and returns those gradients normalized and sanitized.
     21         """
---> 22         labeled_instances = self.predictor.json_to_labeled_instances(inputs)
     23
     24         # List of embedding inputs, used for multiplying gradient by the input for normalization

~/Documents/allennlp/allennlp/predictors/predictor.py in json_to_labeled_instances(self, inputs)
     62         instance = self._json_to_instance(inputs)
     63         outputs = self._model.forward_on_instance(instance)
---> 64         new_instances = self.predictions_to_labeled_instances(instance, outputs)
     65         return new_instances
     66

~/Documents/allennlp/allennlp/predictors/text_classifier.py in predictions_to_labeled_instances(self, instance, outputs)
     43         self, instance: Instance, outputs: Dict[str, numpy.ndarray]
     44     ) -> List[Instance]:
---> 45         new_instance = deepcopy(instance)
     46         label = numpy.argmax(outputs["probs"])
     47         new_instance.add_field("label", LabelField(int(label), skip_indexing=True))

~/anaconda3/envs/allennlp-models/lib/python3.7/copy.py in deepcopy(x, memo, _nil)
    178                     y = x
    179                 else:
--> 180                     y = _reconstruct(x, memo, *rv)
    181
    182     # If is its own copy, don't memoize.

~/anaconda3/envs/allennlp-models/lib/python3.7/copy.py in _reconstruct(x, memo, func, args, state, listiter, dictiter, deepcopy)
    279     if state is not None:
    280         if deep:
--> 281             state = deepcopy(state, memo)
    282         if hasattr(y, '__setstate__'):
    283             y.__setstate__(state)

~/anaconda3/envs/allennlp-models/lib/python3.7/copy.py in deepcopy(x, memo, _nil)
    148     copier = _deepcopy_dispatch.get(cls)
    149     if copier:
--> 150         y = copier(x, memo)
    151     else:
    152         try:

~/anaconda3/envs/allennlp-models/lib/python3.7/copy.py in _deepcopy_dict(x, memo, deepcopy)
    239     memo[id(x)] = y
    240     for key, value in x.items():
--> 241         y[deepcopy(key, memo)] = deepcopy(value, memo)
    242     return y
    243 d[dict] = _deepcopy_dict

~/anaconda3/envs/allennlp-models/lib/python3.7/copy.py in deepcopy(x, memo, _nil)
    148     copier = _deepcopy_dispatch.get(cls)
    149     if copier:
--> 150         y = copier(x, memo)
    151     else:
    152         try:

~/anaconda3/envs/allennlp-models/lib/python3.7/copy.py in _deepcopy_dict(x, memo, deepcopy)
    239     memo[id(x)] = y
    240     for key, value in x.items():
--> 241         y[deepcopy(key, memo)] = deepcopy(value, memo)
    242     return y
    243 d[dict] = _deepcopy_dict

~/anaconda3/envs/allennlp-models/lib/python3.7/copy.py in deepcopy(x, memo, _nil)
    178                     y = x
    179                 else:
--> 180                     y = _reconstruct(x, memo, *rv)
    181
    182     # If is its own copy, don't memoize.

~/anaconda3/envs/allennlp-models/lib/python3.7/copy.py in _reconstruct(x, memo, func, args, state, listiter, dictiter, deepcopy)
    279     if state is not None:
    280         if deep:
--> 281             state = deepcopy(state, memo)
    282         if hasattr(y, '__setstate__'):
    283             y.__setstate__(state)

~/anaconda3/envs/allennlp-models/lib/python3.7/copy.py in deepcopy(x, memo, _nil)
    148     copier = _deepcopy_dispatch.get(cls)
    149     if copier:
--> 150         y = copier(x, memo)
    151     else:
    152         try:

~/anaconda3/envs/allennlp-models/lib/python3.7/copy.py in _deepcopy_dict(x, memo, deepcopy)
    239     memo[id(x)] = y
    240     for key, value in x.items():
--> 241         y[deepcopy(key, memo)] = deepcopy(value, memo)
    242     return y
    243 d[dict] = _deepcopy_dict

~/anaconda3/envs/allennlp-models/lib/python3.7/copy.py in deepcopy(x, memo, _nil)
    148     copier = _deepcopy_dispatch.get(cls)
    149     if copier:
--> 150         y = copier(x, memo)
    151     else:
    152         try:

~/anaconda3/envs/allennlp-models/lib/python3.7/copy.py in _deepcopy_dict(x, memo, deepcopy)
    239     memo[id(x)] = y
    240     for key, value in x.items():
--> 241         y[deepcopy(key, memo)] = deepcopy(value, memo)
    242     return y
    243 d[dict] = _deepcopy_dict

~/anaconda3/envs/allennlp-models/lib/python3.7/copy.py in deepcopy(x, memo, _nil)
    178                     y = x
    179                 else:
--> 180                     y = _reconstruct(x, memo, *rv)
    181
    182     # If is its own copy, don't memoize.

~/anaconda3/envs/allennlp-models/lib/python3.7/copy.py in _reconstruct(x, memo, func, args, state, listiter, dictiter, deepcopy)
    279     if state is not None:
    280         if deep:
--> 281             state = deepcopy(state, memo)
    282         if hasattr(y, '__setstate__'):
    283             y.__setstate__(state)

~/anaconda3/envs/allennlp-models/lib/python3.7/copy.py in deepcopy(x, memo, _nil)
    148     copier = _deepcopy_dispatch.get(cls)
    149     if copier:
--> 150         y = copier(x, memo)
    151     else:
    152         try:

~/anaconda3/envs/allennlp-models/lib/python3.7/copy.py in _deepcopy_dict(x, memo, deepcopy)
    239     memo[id(x)] = y
    240     for key, value in x.items():
--> 241         y[deepcopy(key, memo)] = deepcopy(value, memo)
    242     return y
    243 d[dict] = _deepcopy_dict

~/anaconda3/envs/allennlp-models/lib/python3.7/copy.py in deepcopy(x, memo, _nil)
    178                     y = x
    179                 else:
--> 180                     y = _reconstruct(x, memo, *rv)
    181
    182     # If is its own copy, don't memoize.

~/anaconda3/envs/allennlp-models/lib/python3.7/copy.py in _reconstruct(x, memo, func, args, state, listiter, dictiter, deepcopy)
    279     if state is not None:
    280         if deep:
--> 281             state = deepcopy(state, memo)
    282         if hasattr(y, '__setstate__'):
    283             y.__setstate__(state)

~/anaconda3/envs/allennlp-models/lib/python3.7/copy.py in deepcopy(x, memo, _nil)
    148     copier = _deepcopy_dispatch.get(cls)
    149     if copier:
--> 150         y = copier(x, memo)
    151     else:
    152         try:

~/anaconda3/envs/allennlp-models/lib/python3.7/copy.py in _deepcopy_dict(x, memo, deepcopy)
    239     memo[id(x)] = y
    240     for key, value in x.items():
--> 241         y[deepcopy(key, memo)] = deepcopy(value, memo)
    242     return y
    243 d[dict] = _deepcopy_dict

~/anaconda3/envs/allennlp-models/lib/python3.7/copy.py in deepcopy(x, memo, _nil)
    178                     y = x
    179                 else:
--> 180                     y = _reconstruct(x, memo, *rv)
    181
    182     # If is its own copy, don't memoize.

~/anaconda3/envs/allennlp-models/lib/python3.7/copy.py in _reconstruct(x, memo, func, args, state, listiter, dictiter, deepcopy)
    279     if state is not None:
    280         if deep:
--> 281             state = deepcopy(state, memo)
    282         if hasattr(y, '__setstate__'):
    283             y.__setstate__(state)

~/anaconda3/envs/allennlp-models/lib/python3.7/copy.py in deepcopy(x, memo, _nil)
    148     copier = _deepcopy_dispatch.get(cls)
    149     if copier:
--> 150         y = copier(x, memo)
    151     else:
    152         try:

~/anaconda3/envs/allennlp-models/lib/python3.7/copy.py in _deepcopy_dict(x, memo, deepcopy)
    239     memo[id(x)] = y
    240     for key, value in x.items():
--> 241         y[deepcopy(key, memo)] = deepcopy(value, memo)
    242     return y
    243 d[dict] = _deepcopy_dict

~/anaconda3/envs/allennlp-models/lib/python3.7/copy.py in deepcopy(x, memo, _nil)
    178                     y = x
    179                 else:
--> 180                     y = _reconstruct(x, memo, *rv)
    181
    182     # If is its own copy, don't memoize.

~/anaconda3/envs/allennlp-models/lib/python3.7/copy.py in _reconstruct(x, memo, func, args, state, listiter, dictiter, deepcopy)
    279     if state is not None:
    280         if deep:
--> 281             state = deepcopy(state, memo)
    282         if hasattr(y, '__setstate__'):
    283             y.__setstate__(state)

~/anaconda3/envs/allennlp-models/lib/python3.7/copy.py in deepcopy(x, memo, _nil)
    148     copier = _deepcopy_dispatch.get(cls)
    149     if copier:
--> 150         y = copier(x, memo)
    151     else:
    152         try:

~/anaconda3/envs/allennlp-models/lib/python3.7/copy.py in _deepcopy_dict(x, memo, deepcopy)
    239     memo[id(x)] = y
    240     for key, value in x.items():
--> 241         y[deepcopy(key, memo)] = deepcopy(value, memo)
    242     return y
    243 d[dict] = _deepcopy_dict

~/anaconda3/envs/allennlp-models/lib/python3.7/copy.py in deepcopy(x, memo, _nil)
    178                     y = x
    179                 else:
--> 180                     y = _reconstruct(x, memo, *rv)
    181
    182     # If is its own copy, don't memoize.

~/anaconda3/envs/allennlp-models/lib/python3.7/copy.py in _reconstruct(x, memo, func, args, state, listiter, dictiter, deepcopy)
    279     if state is not None:
    280         if deep:
--> 281             state = deepcopy(state, memo)
    282         if hasattr(y, '__setstate__'):
    283             y.__setstate__(state)

~/anaconda3/envs/allennlp-models/lib/python3.7/copy.py in deepcopy(x, memo, _nil)
    148     copier = _deepcopy_dispatch.get(cls)
    149     if copier:
--> 150         y = copier(x, memo)
    151     else:
    152         try:

~/anaconda3/envs/allennlp-models/lib/python3.7/copy.py in _deepcopy_dict(x, memo, deepcopy)
    239     memo[id(x)] = y
    240     for key, value in x.items():
--> 241         y[deepcopy(key, memo)] = deepcopy(value, memo)
    242     return y
    243 d[dict] = _deepcopy_dict

~/anaconda3/envs/allennlp-models/lib/python3.7/copy.py in deepcopy(x, memo, _nil)
    167                     reductor = getattr(x, "__reduce_ex__", None)
    168                     if reductor:
--> 169                         rv = reductor(4)
    170                     else:
    171                         reductor = getattr(x, "__reduce__", None)

TypeError: can't pickle Tokenizer objects

[dirkgr]

We're thinking we have to either prevent deep copy during interpret, or figure out a way to prevent it from trying to make a copy of the tokenizer.