def _generate_training_examples_md(self,
training_data: "TrainingData") -> Text:
"""Generates markdown training examples."""
import rasa.shared.nlu.training_data.util as rasa_nlu_training_data_utils
training_examples = OrderedDict()
# Sort by intent while keeping basic intent order
for example in [
e.as_dict_nlu() for e in training_data.training_examples
]:
rasa_nlu_training_data_utils.remove_untrainable_entities_from(
example)
intent = example[INTENT]
training_examples.setdefault(intent, [])
training_examples[intent].append(example)
# Don't prepend newline for first line
prepend_newline = False
lines = []
for intent, examples in training_examples.items():
section_header = self._generate_section_header_md(
INTENT, intent, prepend_newline=prepend_newline)
lines.append(section_header)
prepend_newline = True
lines += [
self.generate_list_item(self.generate_message(example))
for example in examples
]
return "".join(lines)
def prepare_training_examples(
training_data: "TrainingData") -> OrderedDict:
"""Pre-processes training data examples by removing not trainable entities."""
import rasa.shared.nlu.training_data.util as rasa_nlu_training_data_utils
training_examples = OrderedDict()
# Sort by intent while keeping basic intent order
for example in [
e.as_dict_nlu() for e in training_data.training_examples
]:
if not example.get(INTENT):
continue
# bf >
metadata = example.pop("metadata", {})
language = metadata.pop("language", None)
if metadata:
example["metadata"] = metadata
# < bf
rasa_nlu_training_data_utils.remove_untrainable_entities_from(
example)
# bf >
intent = (example[INTENT], language)
# intent = example[INTENT]
# < bf
training_examples.setdefault(intent, [])
training_examples[intent].append(example)
return training_examples
def prepare_training_examples(training_data: "TrainingData") -> OrderedDict:
"""Pre-processes training data examples by removing not trainable entities."""
import rasa.shared.nlu.training_data.util as rasa_nlu_training_data_utils
training_examples = OrderedDict()
# Sort by intent while keeping basic intent order
for example in [e.as_dict_nlu() for e in training_data.training_examples]:
rasa_nlu_training_data_utils.remove_untrainable_entities_from(example)
intent = example[INTENT]
training_examples.setdefault(intent, [])
training_examples[intent].append(example)
return training_examples
def prepare_training_examples(
training_data: "TrainingData",
) -> Dict[Text, List[Union[Dict, Text]]]:
"""Pre-processes training data examples by removing not trainable entities."""
import rasa.shared.nlu.training_data.util as rasa_nlu_training_data_utils
training_examples: Dict[Text, List[Union[Dict, Text]]] = OrderedDict()
# Sort by intent while keeping basic intent order
for example in [e.as_dict_nlu() for e in training_data.training_examples]:
if not example.get(INTENT):
continue
rasa_nlu_training_data_utils.remove_untrainable_entities_from(example)
intent = example[INTENT]
training_examples.setdefault(intent, [])
training_examples[intent].append(example)
return training_examples