TypeError: CorefResolver.make_output_human_readable: return type `None` is not a `typing.Dict[str, torch.Tensor]`.

[dave-shore]

Dear all, running your notebook for the integration of DyGIE in SpaCy, I get the following error:

TypeError Traceback (most recent call last) Input In [20], in <cell line: 2>() 1 import spacy ----> 2 from dygie.spacy_interface.spacy_interface import DygieppPipe 3 nlp = spacy.load('en_core_web_sm')

File ~\LAZY\dygie\spacy_interface\spacy_interface.py:3, in 1 from typing import Dict, List 2 from allennlp.data import Batch ----> 3 from dygie.models.dygie import DyGIE 4 from dygie.data.dataset_readers.dygie import DyGIEReader 5 from allennlp.models.archival import load_archive

File ~\LAZY\dygie\models__init__.py:1, in ----> 1 from dygie.models.dygie import DyGIE

File ~\LAZY\dygie\models\dygie.py:17, in 14 from allennlp.nn import util, InitializerApplicator, RegularizerApplicator 16 # Import submodules. ---> 17 from dygie.models.coref import CorefResolver 18 from dygie.models.ner import NERTagger 19 from dygie.models.relation import RelationExtractor

File ~\LAZY\dygie\models\coref.py:24, in 19 from dygie.models.entity_beam_pruner import Pruner 21 logger = logging.getLogger(name) # pylint: disable=invalid-name ---> 24 class CorefResolver(Model): 25 """ 26 TODO(dwadden) document correctly. 27 (...) 46 If provided, will be used to calculate the regularization penalty during training. 47 """ 48 def init(self, 49 vocab: Vocabulary, 50 make_feedforward: Callable, (...) 56 coref_prop_dropout_f: float = 0.0, 57 regularizer: Optional[RegularizerApplicator] = None) -> None:

File ~\LAZY\dygie\models\coref.py:388, in CorefResolver() 383 output_dict["loss"] = negative_marginal_log_likelihood 385 return output_dict 387 @overrides --> 388 def make_output_human_readable(self, output_dict: Dict[str, torch.Tensor]): 389 """ 390 Converts the list of spans and predicted antecedent indices into clusters 391 of spans for each element in the batch. (...) 405 original document. 406 """ 408 # A tensor of shape (batch_size, num_spans_to_keep, 2), representing 409 # the start and end indices of each span.

File ~\anaconda3\lib\site-packages\overrides\overrides.py:88, in overrides(method, check_signature, check_at_runtime) 58 """Decorator to indicate that the decorated method overrides a method in 59 superclass. 60 The decorator code is executed while loading class. Using this method (...) 85 docstring from super class 86 """ 87 if method is not None: ---> 88 return _overrides(method, check_signature, check_at_runtime) 89 else: 90 return functools.partial( 91 overrides, 92 check_signature=check_signature, 93 check_at_runtime=check_at_runtime, 94 )

File ~\anaconda3\lib\site-packages\overrides\overrides.py:114, in _overrides(method, check_signature, check_at_runtime) 112 return wrapper # type: ignore 113 else: --> 114 _validate_method(method, super_class, check_signature) 115 return method 116 raise TypeError(f"{method.qualname}: No super class method found")

File ~\anaconda3\lib\site-packages\overrides\overrides.py:135, in _validate_method(method, super_class, check_signature) 129 method.doc = super_method.doc 130 if ( 131 check_signature 132 and not method.name.startswith("__") 133 and not isinstance(super_method, property) 134 ): --> 135 ensure_signature_is_compatible(super_method, method, is_static)

File ~\anaconda3\lib\site-packages\overrides\signature.py:93, in ensure_signature_is_compatible(super_callable, sub_callable, is_static) 90 same_main_module = _is_same_module(sub_callable, super_callable) 92 if super_type_hints is not None and sub_type_hints is not None: ---> 93 ensure_return_type_compatibility(super_type_hints, sub_type_hints, method_name) 94 ensure_all_kwargs_defined_in_sub( 95 super_sig, sub_sig, super_type_hints, sub_type_hints, is_static, method_name 96 ) 97 ensure_all_positional_args_defined_in_sub( 98 super_sig, 99 sub_sig, (...) 104 method_name, 105 )

File ~\anaconda3\lib\site-packages\overrides\signature.py:287, in ensure_return_type_compatibility(super_type_hints, sub_type_hints, method_name) 285 sub_return = sub_type_hints.get("return", None) 286 if not _issubtype(sub_return, super_return) and super_return is not None: --> 287 raise TypeError( 288 f"{method_name}: return type {sub_return} is not a {super_return}." 289 )

TypeError: CorefResolver.make_output_human_readable: return type None is not a typing.Dict[str, torch.Tensor].

Do you have any suggestions?

[dwadden]

Hi,

The DyGIE / Spacy interface was added by @e3oroush. I unfortunately can't offer support, but maybe Ebi can help.

Dave

[e3oroush]

Hi, I'm sorry I haven't checked my notifications. I will try to take a look at it over the weekend.

[dwadden]

Thanks!

[e3oroush]

I tried to reproduce the problem. but I couldn't. Can you send me the information about your Python environment?
Make sure to have sapcy==2.3.7 You can get the list of installed packages using pip freeze, explained here

[dwadden]

I'm gonna close this for lack of activity. Feel free to reopen if this is still an issue.

[nat-btlr]

Hello! I am facing the same issue. I followed the readme instructions and I followed the instructions regarding having spacy==2.3.7. My Python env. info: python 3.7.16. Could you suggest what the problem could be, please? Thanks

[dwadden]

Hi, I'm traveling for the next two weeks but will try to take a look at this when I get back!

[dwadden]

Taking a look now. Unfortunately as I mentioned earlier in the thread, I can't offer support for this but maybe @e3oroush can help?

[Alienboyplus]

Hi, I've just faced this TypeError. I used allennlp>2 and allennlp-models>2, which is not recommend in https://github.com/dwadden/dygiepp/issues/58

So I set

allennlp==1.1.0
allennlp-models==1.1.0
overrides==3.1.0

and fixed this problem !

[Alienboyplus]

Hi, I've just faced this TypeError. I used allennlp>2 and allennlp-models>2, which is not recommend in #58

So I set

allennlp==1.1.0
allennlp-models==1.1.0
overrides==3.1.0

and fixed this problem !

but it gives another error when i use overrides==7.7.0:

TypeError: ArrayField.empty_field: return type `None` is not a `<class 'allennlp.data.fields.field.Field'>`.

Also, the cluster field is missing