sgrvinod / a-pytorch-tutorial-to-sequence-labeling Goto Github PK

Hi,
Thanks for the great explanation.
I am wondering if I want to calculate partial CRF (http://aclweb.org/anthology/C18-1183)where we have some gold tag for some token and all possible tags for other tokens and it needs to calculate the following formula?
log_sum_ex (all socore) - log_sum_ex(partial score)
what would be the changes in the code?

Thanks

As of PyTorch 1.1.0, the inputs to pack_padded_sequence do not need to be sorted by length if enforce_sorted=False:

https://pytorch.org/docs/master/nn.html#torch.nn.utils.rnn.pack_padded_sequence

enforce_sorted (bool, optional) – if True, the input is expected to contain sequences sorted by length in a decreasing order. If False, this condition is not checked. Default: True.

Thanks for the great tutorial!

Thank you for the tutorial, it is very easy to follow.
However, I am unable to replicate the results as mentioned in the tutorial.
I attempted to use the checkpoint for pertained model. The tag_map seems to not include all the tags. Is this on a different dataset?

In inference.py line 60, self.start_tag should be self.end_tag

in models.py line 184
:param cmap_lengths: character sequence lengths, a tensor of dimensions (batch_size, word_pad_len)
the shape of cmap_lengths should be (batch_size)?

Thanks for your code.
Could you please add a section for prediction using the trained weights ?

Getting the following error. Did someone face the same?

Embedding length is 100.
You have elected to include embeddings that are out-of-corpus.

Loading embeddings...
Traceback (most recent call last):
  File "train.py", line 356, in <module>
    main()
  File "train.py", line 79, in main
    expand_vocab)  # load pre-trained embeddings
  File "C:\Users\user\Desktop\NLP\a-PyTorch-Tutorial-to-Sequence-Labeling-master\utils.py", line 230, in load_embeddings
    for line in open(emb_file, 'r', 'utf-8'):
TypeError: an integer is required (got type str)