INNER CODE UNIT · Python
__getitem__
IBM/transition-amr-parser · src/ibm_neural_aligner/main.py:540
def __getitem__(self, idx):
if idx in self.cached:
return self.cached[idx]
amr = self.corpus[idx]
item = {}
# text
tokenizer_output = self.text_tokenizer.tokenize(amr.tokens)
item['text_original_tokens'] = tokenizer_output['original_input']
item['text_tokens'] = tokenizer_output['input_ids']
# amr
tokenizer_output = self.amr_tokenizer.tokenize(amr)
item['linearized_parse'] = tokenizer_output['original_input']
item['amr_tokens'] = tokenizer_output['input_ids']
item['amr_node_ids'] = tokenizer_output['token_type_ids']
item['amr_node_mask'] = tokenizer_output['node_mask']