Long sequences breaking _greedy_decode

[ZhaofengWu]

When I'm parsing a sentence that is too long, I get the following error (this stack trace is for SDP parsing, but it also breaks EDS and UCCA):

Traceback (most recent call last):
  File "~/miniconda3/envs/parsing/bin/allennlp", line 8, in <module>
    sys.exit(run())
  File "~/miniconda3/envs/parsing/lib/python3.7/site-packages/allennlp/run.py", line 18, in run
    main(prog="allennlp")
  File "~/miniconda3/envs/parsing/lib/python3.7/site-packages/allennlp/commands/__init__.py", line 102, in main
    args.func(args)
  File "~/miniconda3/envs/parsing/lib/python3.7/site-packages/allennlp/commands/predict.py", line 227, in _predict
    manager.run()
  File "~/miniconda3/envs/parsing/lib/python3.7/site-packages/allennlp/commands/predict.py", line 206, in run
    for model_input_json, result in zip(batch_json, self._predict_json(batch_json)):
  File "~/miniconda3/envs/parsing/lib/python3.7/site-packages/allennlp/commands/predict.py", line 151, in _predict_json
    results = [self._predictor.predict_json(batch_data[0])]
  File "~/miniconda3/envs/parsing/lib/python3.7/site-packages/allennlp/predictors/predictor.py", line 65, in predict_json
    return self.predict_instance(instance)
  File "./utils/transition_sdp_predictor.py", line 43, in predict_instance
    outputs = self._model.forward_on_instance(instance)
  File "~/miniconda3/envs/parsing/lib/python3.7/site-packages/allennlp/models/model.py", line 124, in forward_on_instance
    return self.forward_on_instances([instance])[0]
  File "~/miniconda3/envs/parsing/lib/python3.7/site-packages/allennlp/models/model.py", line 153, in forward_on_instances
    outputs = self.decode(self(**model_input))
  File "~/miniconda3/envs/parsing/lib/python3.7/site-packages/torch/nn/modules/module.py", line 550, in __call__
    result = self.forward(*input, **kwargs)
  File "./modules/transition_parser_sdp2015.py", line 331, in forward
    embedded_text_input=embedded_text_input)
  File "./modules/transition_parser_sdp2015.py", line 150, in _greedy_decode
    input=embedded_text_input[sent_idx][sent_len[sent_idx] - 1 - token_idx],
IndexError: index 428 is out of bounds for dimension 0 with size 390

This is from https://github.com/DreamerDeo/HIT-SCIR-CoNLL2019/blob/9bc3c0727c13aac72e71ae7bf48cc195f481bffa/modules/transition_parser_sdp2015.py#L146-L151

Does this package support generating a "partial" graph for long sequences, or are long sequences not supported at all? If I change the if statement to the following, would the resulting graph be a valid partial graph?

if sent_len[sent_idx] > token_idx and sent_len[sent_idx] - 1 - token_idx < embedded_text_input.shape[1]: