def predict_json()

in curiosity/predictors.py [0:0]

• 16 lines of code
• 6 McCabe index (conditional complexity)

    def predict_json(self, inputs: JsonDict) -> JsonDict:

        dialogs = inputs['dialogs']
        out = []

        for i, d in enumerate(dialogs):
            if i == 30:
                # Early termination to save time
                break

            instance = self._dataset_reader.text_to_instance(d)
            prediction = self.predict_instance(instance)

            # Label predictions for this dialog
            label_prediction = {
                'dialog_id': d['dialog_id']
            }

            for k, v in prediction.items():
                if k != 'loss':
                    label_prediction[k] = np.argmax(v, axis=1).tolist()

            out.append(label_prediction)

        return out