TextClassificationJsonReader and Basic_Classifier Bug With Segment Sentences

[zhoujs93]

OS: Windows Python version: 3.6.5 AllenNLP version: 0.8.5 PyTorch version: 1.1.0

Question:

I get the following error when running TextClassificationJsonReader with segment sentences = True.

RuntimeError: input must have 2 dimensions, got 3

The configuration for my model is as follows:

{
    "dataset_reader": {
        "lazy": true,
        "type": "text_classification_json",
        "tokenizer": {
            "word_splitter": "spacy"
        },
        "token_indexers": {
            "tokens": {
                "type": "single_id",
                "namespace": "tokens",
                "lowercase_tokens": true
            }
        },
        "segment_sentences" : true
    },
  "train_data_path": "./tests/fixtures/sample_data.jsonl",
  "validation_data_path": "./tests/fixtures/valid_sample_data.jsonl",
    "model": {
      "type": "basic_classifier",
      "text_field_embedder": {
          "token_embedders": {
              "tokens": {
                  "type": "embedding",
                  "embedding_dim": 300,
                  "trainable": true
              }
          }
      },
      "seq2seq_encoder": {
          "type": "lstm",
          "num_layers": 1,
          "bidirectional": true,
          "input_size": 300,
          "hidden_size": 50
      },
      "seq2vec_encoder": {
           "type": "cnn",
           "num_filters": 8,
           "embedding_dim": 100,
           "output_dim": 50
        }
    },

I know that the dataset_reader is correct based on a text, and that there is no issue with my data since it is similar to what is specified in allennlp/tests/data/dataset_readers/text_classification_json_test.py

So I am not sure what I should do. Am I specifying the text_field_embedder correctly ?

The goal is for me to build a model that can perform document classification based on the sentence embeddings since my document length are quite long (similar to the one mentioned here and referenced in the dataset_readers/text_classification_json documention: hierarchical-attention-networks) [https://www.cs.cmu.edu/~hovy/papers/16HLT-hierarchical-attention-networks.pdf]

Thanks

[brendan-ai2]

@putsandcalls, could you please include the full stack trace of that error? I can't tell where it's coming from for certain with just the exception text. Thanks.

[brendan-ai2]

It would also be useful to have a sample input that causes the failure.

[zhoujs93]

@brendan-ai2

I attached the sample input with the json configuration as follows, and the error log is as follows:

  File "c:\users\xbbns8w\appdata\local\continuum\anaconda3\envs\python36\lib\runpy.py", line 193, in _run_module_as_main
    "__main__", mod_spec)
  File "c:\users\xbbns8w\appdata\local\continuum\anaconda3\envs\python36\lib\runpy.py", line 85, in _run_code
    exec(code, run_globals)
  File "C:\Users\xbbns8w\AppData\Local\Continuum\anaconda3\envs\python36\Scripts\allennlp.exe\__main__.py", line 9, in <module>
  File "c:\users\xbbns8w\appdata\local\continuum\anaconda3\envs\python36\lib\site-packages\allennlp\run.py", line 18, in run
    main(prog="allennlp")
  File "c:\users\xbbns8w\appdata\local\continuum\anaconda3\envs\python36\lib\site-packages\allennlp\commands\__init__.py", line 102, in main
    args.func(args)
  File "c:\users\xbbns8w\appdata\local\continuum\anaconda3\envs\python36\lib\site-packages\allennlp\commands\train.py", line 124, in train_model_from_args
    args.cache_prefix)
  File "c:\users\xbbns8w\appdata\local\continuum\anaconda3\envs\python36\lib\site-packages\allennlp\commands\train.py", line 168, in train_model_from_file
    cache_directory, cache_prefix)
  File "c:\users\xbbns8w\appdata\local\continuum\anaconda3\envs\python36\lib\site-packages\allennlp\commands\train.py", line 252, in train_model
    metrics = trainer.train()
  File "c:\users\xbbns8w\appdata\local\continuum\anaconda3\envs\python36\lib\site-packages\allennlp\training\trainer.py", line 478, in train
    train_metrics = self._train_epoch(epoch)
  File "c:\users\xbbns8w\appdata\local\continuum\anaconda3\envs\python36\lib\site-packages\allennlp\training\trainer.py", line 320, in _train_epoch
    loss = self.batch_loss(batch_group, for_training=True)
  File "c:\users\xbbns8w\appdata\local\continuum\anaconda3\envs\python36\lib\site-packages\allennlp\training\trainer.py", line 261, in batch_loss
    output_dict = self.model(**batch)
  File "c:\users\xbbns8w\appdata\local\continuum\anaconda3\envs\python36\lib\site-packages\torch\nn\modules\module.py", line 493, in __call__
    result = self.forward(*input, **kwargs)
  File "c:\users\xbbns8w\appdata\local\continuum\anaconda3\envs\python36\lib\site-packages\allennlp\models\basic_classifier.py", line 114, in forward
    embedded_text = self._seq2seq_encoder(embedded_text, mask=mask)
  File "c:\users\xbbns8w\appdata\local\continuum\anaconda3\envs\python36\lib\site-packages\torch\nn\modules\module.py", line 493, in __call__
    result = self.forward(*input, **kwargs)
  File "c:\users\xbbns8w\appdata\local\continuum\anaconda3\envs\python36\lib\site-packages\allennlp\modules\seq2seq_encoders\pytorch_seq2seq_wrapper.py", line 83, in forward
    self.sort_and_run_forward(self._module, inputs, mask, hidden_state)
  File "c:\users\xbbns8w\appdata\local\continuum\anaconda3\envs\python36\lib\site-packages\allennlp\modules\encoder_base.py", line 116, in sort_and_run_forward
    module_output, final_states = module(packed_sequence_input, initial_states)
  File "c:\users\xbbns8w\appdata\local\continuum\anaconda3\envs\python36\lib\site-packages\torch\nn\modules\module.py", line 493, in __call__
    result = self.forward(*input, **kwargs)
  File "c:\users\xbbns8w\appdata\local\continuum\anaconda3\envs\python36\lib\site-packages\torch\nn\modules\rnn.py", line 557, in forward
    return self.forward_packed(input, hx)
  File "c:\users\xbbns8w\appdata\local\continuum\anaconda3\envs\python36\lib\site-packages\torch\nn\modules\rnn.py", line 550, in forward_packed
    output, hidden = self.forward_impl(input, hx, batch_sizes, max_batch_size, sorted_indices)
  File "c:\users\xbbns8w\appdata\local\continuum\anaconda3\envs\python36\lib\site-packages\torch\nn\modules\rnn.py", line 519, in forward_impl
    self.check_forward_args(input, hx, batch_sizes)
  File "c:\users\xbbns8w\appdata\local\continuum\anaconda3\envs\python36\lib\site-packages\torch\nn\modules\rnn.py", line 490, in check_forward_args
    self.check_input(input, batch_sizes)
  File "c:\users\xbbns8w\appdata\local\continuum\anaconda3\envs\python36\lib\site-packages\torch\nn\modules\rnn.py", line 149, in check_input
    expected_input_dim, input.dim()))
RuntimeError: input must have 2 dimensions, got 3

sample.zip

[DeNeutoy]

Closing due to inactivity