Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

train error I can't handle out #3

Open
TingFree opened this issue Sep 14, 2019 · 10 comments
Open

train error I can't handle out #3

TingFree opened this issue Sep 14, 2019 · 10 comments

Comments

@TingFree
Copy link

I run your code in google colab, but I meet a crucial error and I can't deal with it by google, maybe you can help me, thanks.
ps: all packages you mention is installed.

[2019-09-14 02:57:15,574 INFO]  * src vocab size = 49766
[2019-09-14 02:57:15,574 INFO]  * history vocab size = 49766
[2019-09-14 02:57:15,574 INFO]  * tgt vocab size = 49766
[2019-09-14 02:57:15,574 INFO] Building model...
Traceback (most recent call last):
  File "/content/drive/My Drive/ReDR/train.py", line 109, in <module>
    main(opt)
  File "/content/drive/My Drive/ReDR/train.py", line 41, in main
    single_main(opt, -1)
  File "/content/drive/My Drive/ReDR/onmt/train_single.py", line 86, in main
    model = build_model(model_opt, opt, fields, checkpoint)
  File "/content/drive/My Drive/ReDR/onmt/model_builder.py", line 235, in build_model
    vocab = torch.load(opt.drqa_vocab_path)
  File "/usr/local/lib/python3.6/dist-packages/torch/serialization.py", line 387, in load
    return _load(f, map_location, pickle_module, **pickle_load_args)
  File "/usr/local/lib/python3.6/dist-packages/torch/serialization.py", line 574, in _load
    result = unpickler.load()
AttributeError: Can't get attribute '_default_unk_index' on <module 'torchtext.vocab' from '/usr/local/lib/python3.6/dist-packages/torchtext/vocab.py'>
@TingFree
Copy link
Author

in fact， when I torch.load the 'drqa_param/vocab.pt', the same exception will be reported. Is the pt file wrong? The test code as follow:

import torch
vocab = torch.load('./drqa_param/vocab.pt')

Traceback (most recent call last):
  File "G:/NLP/code/cqg_ReDR/just_test.py", line 2, in <module>
    vocab = torch.load('./drqa_param/vocab.pt')
  File "D:\Python\lib\site-packages\torch\serialization.py", line 368, in load
    return _load(f, map_location, pickle_module)
  File "D:\Python\lib\site-packages\torch\serialization.py", line 542, in _load
    result = unpickler.load()
AttributeError: Can't get attribute '_default_unk_index' on <module 'torchtext.vocab' from 'D:\\Python\\lib\\site-packages\\torchtext\\vocab.py'>

@kouhonglady
Copy link

Have you solved this problem？ I have encountered the same problem as you, and I don't know how to solve it. help~

@TingFree
Copy link
Author

@kouhonglady when I change the torchtext version to 0.3.1, another error was reported. So I guess it maybe the version's problem. After that, I was too busy participating in other competitions to continue my investigation. If you solve this problem, please tell me in time, or I will share you my method when I run this code perfect in the near future. I hope so~

@kouhonglady
Copy link

kouhonglady commented Sep 30, 2019 •
edited
Loading

I solved this problem using torchtext version to 0.3.1 and python 2.7
image

@AlbertChen1991
Copy link

@kouhonglady @dingjiajie Hi, have you corrected the errors?

@AlbertChen1991
Copy link

Hello, @kouhonglady @dingjiajie . I have fixed the error in python 3.7. Just install the torchtext==0.4.0, and add the following code into the model_builder.py file.

from torchtext import vocab
try:
vocab._default_unk_index
except AttributeError:
def _default_unk_index():
return 0
vocab._default_unk_index = _default_unk_index

@ZhiliWang
Copy link

Hello, @kouhonglady @dingjiajie . I have fixed the error in python 3.7. Just install the torchtext==0.4.0, and add the following code into the model_builder.py file.

from torchtext import vocab
try:
vocab._default_unk_index
except AttributeError:
def _default_unk_index():
return 0
vocab._default_unk_index = _default_unk_index

@AlbertChen1991

This solves OP's issue but triggered another runtime error for me:

/opt/anaconda3/envs/py36/lib/python3.6/site-packages/torchtext/data/field.py:359: UserWarning: To copy construct from a tensor, it is recommended to use sourceTensor.clone().detach() or sourceTensor.clone().detach().requires_grad_(True), rather than torch.tensor(sourceTensor).
var = torch.tensor(arr, dtype=self.dtype, device=device)
Traceback (most recent call last):
File "train.py", line 109, in
main(opt)
File "train.py", line 41, in main
single_main(opt, -1)
File "/Users/zhiliwang/Documents/nlp/ReDR/onmt/train_single.py", line 134, in main
valid_steps=opt.valid_steps)
File "/Users/zhiliwang/Documents/nlp/ReDR/onmt/trainer.py", line 217, in train
report_stats, local_step)
File "/Users/zhiliwang/Documents/nlp/ReDR/onmt/trainer.py", line 348, in _gradient_accumulation
outputs, attns, results = self.model(batch, src, history, tgt, src_lengths, history_lengths, bptt=bptt)
File "/opt/anaconda3/envs/py36/lib/python3.6/site-packages/torch/nn/modules/module.py", line 532, in call
result = self.forward(*input, **kwargs)
File "/Users/zhiliwang/Documents/nlp/ReDR/onmt/models/model.py", line 70, in forward
memory_lengths=src_lengths)
File "/opt/anaconda3/envs/py36/lib/python3.6/site-packages/torch/nn/modules/module.py", line 532, in call
result = self.forward(*input, **kwargs)
File "/Users/zhiliwang/Documents/nlp/ReDR/onmt/decoders/decoder.py", line 210, in forward
tgt, memory_bank, memory_lengths=memory_lengths)
File "/Users/zhiliwang/Documents/nlp/ReDR/onmt/decoders/decoder.py", line 388, in run_forward_pass
memory_lengths=memory_lengths)
File "/opt/anaconda3/envs/py36/lib/python3.6/site-packages/torch/nn/modules/module.py", line 532, in call
result = self.forward(*input, **kwargs)
File "/Users/zhiliwang/Documents/nlp/ReDR/onmt/modules/global_attention.py", line 183, in forward
align.masked_fill(1 - mask, -float('inf'))
File "/opt/anaconda3/envs/py36/lib/python3.6/site-packages/torch/tensor.py", line 394, in rsub
return _C._VariableFunctions.rsub(self, other)
RuntimeError: Subtraction, the - operator, with a bool tensor is not supported. If you are trying to invert a mask, use the ~ or logical_not() operator instead.

@AlbertChen1991
Copy link

AlbertChen1991 commented Feb 29, 2020 via email

@AlbertChen1991
Copy link

AlbertChen1991 commented Feb 29, 2020 via email

@dxlong2000
Copy link

@AlbertChen1991 Thanks for your support.

After fixing the 1 - mask I got another error as below. Could you suggest any solution? Thank you very much!

Traceback (most recent call last):
  File "train.py", line 109, in <module>
    main(opt)
  File "train.py", line 41, in main
    single_main(opt, -1)
  File "/content/gdrive/MyDrive/A/factoid_one_focus/CoQA/baselines/ReDR/onmt/train_single.py", line 134, in main
    valid_steps=opt.valid_steps)
  File "/content/gdrive/MyDrive/A/factoid_one_focus/CoQA/baselines/ReDR/onmt/trainer.py", line 217, in train
    report_stats, local_step)
  File "/content/gdrive/MyDrive/A/factoid_one_focus/CoQA/baselines/ReDR/onmt/trainer.py", line 348, in _gradient_accumulation
    outputs, attns, results = self.model(batch, src, history, tgt, src_lengths, history_lengths, bptt=bptt)
  File "/usr/local/lib/python3.7/dist-packages/torch/nn/modules/module.py", line 1110, in _call_impl
    return forward_call(*input, **kwargs)
  File "/content/gdrive/MyDrive/A/factoid_one_focus/CoQA/baselines/ReDR/onmt/models/model.py", line 72, in forward
    results = self.translator.translate_batch(batch, vocabs, False, training=self.train())
  File "/content/gdrive/MyDrive/A/factoid_one_focus/CoQA/baselines/ReDR/onmt/translate/translator.py", line 539, in translate_batch
    training=training)
  File "/content/gdrive/MyDrive/A/factoid_one_focus/CoQA/baselines/ReDR/onmt/translate/translator.py", line 695, in _translate_batch
    beam.advance(log_probs, attn, dec_out_attn)
  File "/content/gdrive/MyDrive/A/factoid_one_focus/CoQA/baselines/ReDR/onmt/translate/beam_search.py", line 182, in advance
    [self.alive_seq.index_select(0, self.select_indices),
RuntimeError: index_select(): Expected dtype int32 or int64 for index

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Assignees
No one assigned
Labels
None yet
Projects
None yet
Milestone
No milestone
Development

No branches or pull requests
5 participants
@AlbertChen1991 @ZhiliWang @kouhonglady @TingFree @dxlong2000