Move default predictors (#31)

* Move decomposable attention predictor

* Move default predictor

* Remove temporary shim

allennlp_models/coref/coref_model.py

@@ -871,3 +871,5 @@ def _compute_coreference_scores(
         # Shape: (batch_size, num_spans_to_keep, max_antecedents + 1)
         coreference_scores = torch.cat([dummy_scores, antecedent_scores], -1)
         return coreference_scores
+
+    default_predictor = "coreference-resolution"

allennlp_models/ner/crf_tagger.py

@@ -296,3 +296,5 @@ def get_metrics(self, reset: bool = False) -> Dict[str, float]:
             else:
                 metrics_to_return.update({x: y for x, y in f1_dict.items() if "overall" in x})
         return metrics_to_return
+
+    default_predictor = "sentence-tagger"

allennlp_models/nli/__init__.py

@@ -1,5 +1,6 @@
 from allennlp_models.nli.snli_reader import SnliReader
 from allennlp_models.nli.decomposable_attention_model import DecomposableAttention
+from allennlp_models.nli.decomposable_attention_predictor import DecomposableAttentionPredictor
 from allennlp_models.nli.bimpm_model import BiMpm
 from allennlp_models.nli.esim_model import ESIM
 from allennlp_models.nli.quora_paraphrase_reader import QuoraParaphraseDatasetReader

allennlp_models/nli/bimpm_model.py

@@ -243,3 +243,5 @@ def get_metrics(self, reset: bool = False) -> Dict[str, float]:
         return {
             metric_name: metric.get_metric(reset) for metric_name, metric in self.metrics.items()
         }
+
+    default_predictor = "textual-entailment"

allennlp_models/nli/decomposable_attention_model.py

@@ -199,3 +199,5 @@ def forward(  # type: ignore
 
     def get_metrics(self, reset: bool = False) -> Dict[str, float]:
         return {"accuracy": self._accuracy.get_metric(reset)}
+
+    default_predictor = "textual-entailment"

allennlp_models/nli/decomposable_attention_predictor.py

@@ -0,0 +1,58 @@
+from typing import List, Dict
+from copy import deepcopy
+
+import numpy
+from overrides import overrides
+
+from allennlp.common.util import JsonDict
+from allennlp.data import Instance
+from allennlp.predictors.predictor import Predictor
+from allennlp.data.fields import LabelField
+
+
+@Predictor.register("textual-entailment")
+class DecomposableAttentionPredictor(Predictor):
+    """
+    Predictor for the [`DecomposableAttention`](../models/decomposable_attention.md) model.
+
+    Registered as a `Predictor` with name "textual-entailment".
+    """
+
+    def predict(self, premise: str, hypothesis: str) -> JsonDict:
+        """
+        Predicts whether the hypothesis is entailed by the premise text.
+
+        # Parameters
+
+        premise : `str`
+            A passage representing what is assumed to be true.
+
+        hypothesis : `str`
+            A sentence that may be entailed by the premise.
+
+        # Returns
+
+        `JsonDict`
+            A dictionary where the key "label_probs" determines the probabilities of each of
+            [entailment, contradiction, neutral].
+        """
+        return self.predict_json({"premise": premise, "hypothesis": hypothesis})
+
+    @overrides
+    def _json_to_instance(self, json_dict: JsonDict) -> Instance:
+        """
+        Expects JSON that looks like `{"premise": "...", "hypothesis": "..."}`.
+        """
+        premise_text = json_dict["premise"]
+        hypothesis_text = json_dict["hypothesis"]
+        return self._dataset_reader.text_to_instance(premise_text, hypothesis_text)
+
+    @overrides
+    def predictions_to_labeled_instances(
+        self, instance: Instance, outputs: Dict[str, numpy.ndarray]
+    ) -> List[Instance]:
+        new_instance = deepcopy(instance)
+        label = numpy.argmax(outputs["label_logits"])
+        # Skip indexing, we have integer representations of the strings "entailment", etc.
+        new_instance.add_field("label", LabelField(int(label), skip_indexing=True))
+        return [new_instance]

allennlp_models/rc/bidaf/bidaf_model.py

@@ -366,3 +366,5 @@ def get_best_span(
         span_start_indices = best_spans // passage_length
         span_end_indices = best_spans % passage_length
         return torch.stack([span_start_indices, span_end_indices], dim=-1)
+
+    default_predictor = "reading-comprehension"

allennlp_models/syntax/biaffine_dependency_parser/biaffine_dependency_parser.py

@@ -28,8 +28,8 @@
 
 POS_TO_IGNORE = {"`", "''", ":", ",", ".", "PU", "PUNCT", "SYM"}
 
-# exist_ok has to be true until we remove this from the core library
-@Model.register("biaffine_parser", exist_ok=True)
+
+@Model.register("biaffine_parser")
 class BiaffineDependencyParser(Model):
     """
     This dependency parser follows the model of
@@ -681,3 +681,5 @@ def _get_mask_for_eval(
     @overrides
     def get_metrics(self, reset: bool = False) -> Dict[str, float]:
         return self._attachment_scores.get_metric(reset)
+
+    default_predictor = "biaffine-dependency-parser"

allennlp_models/syntax/constituency_parser/constituency_parser_model.py

@@ -494,3 +494,5 @@ def get_metrics(self, reset: bool = False) -> Dict[str, float]:
             evalb_metrics = self._evalb_score.get_metric(reset=reset)
             all_metrics.update(evalb_metrics)
         return all_metrics
+
+    default_predictor = "constituency-parser"

allennlp_models/syntax/srl/srl_bert_model.py

@@ -277,3 +277,5 @@ def get_start_transitions(self):
                 start_transitions[i] = float("-inf")
 
         return start_transitions
+
+    default_predictor = "semantic-role-labeling"

allennlp_models/syntax/srl/srl_model.py

@@ -297,6 +297,8 @@ def get_start_transitions(self):
 
         return start_transitions
 
+    default_predictor = "semantic-role-labeling"
+
 
 def write_to_conll_eval_file(
     prediction_file: TextIO,