HW 8 error: "Expected tensor for argument #1 'indices' to have scalar type Long; but got torch.IntTensor instead"

[jacyanthis]

When running the "# Prediction on test set" cell in HW 8 (before I get to the first exercise), I get the following error:

RuntimeError: Expected tensor for argument #1 'indices' to have scalar type Long; but got torch.IntTensor instead (while checking arguments for embedding)

A number of people (example) suggest this error has to do with Windows 10 or different versions of huggingface/transformers, and I can successfully run this code on the Google Collab server, just not my local machine.

Any tips?

Full error:

Predicting labels for 516 test sentences...
---------------------------------------------------------------------------
RuntimeError                              Traceback (most recent call last)
<ipython-input-63-6b924c3d5378> in <module>
     21     with torch.no_grad():
     22       # Forward pass, calculate logit predictions
---> 23       outputs = model(b_input_ids, token_type_ids=None, 
     24                       attention_mask=b_input_mask)
     25 

C:\ProgramData\Anaconda3\lib\site-packages\torch\nn\modules\module.py in _call_impl(self, *input, **kwargs)
    725             result = self._slow_forward(*input, **kwargs)
    726         else:
--> 727             result = self.forward(*input, **kwargs)
    728         for hook in itertools.chain(
    729                 _global_forward_hooks.values(),

C:\ProgramData\Anaconda3\lib\site-packages\transformers\models\bert\modeling_bert.py in forward(self, input_ids, attention_mask, token_type_ids, position_ids, head_mask, inputs_embeds, labels, output_attentions, output_hidden_states, return_dict)
   1494         return_dict = return_dict if return_dict is not None else self.config.use_return_dict
   1495 
-> 1496         outputs = self.bert(
   1497             input_ids,
   1498             attention_mask=attention_mask,

C:\ProgramData\Anaconda3\lib\site-packages\torch\nn\modules\module.py in _call_impl(self, *input, **kwargs)
    725             result = self._slow_forward(*input, **kwargs)
    726         else:
--> 727             result = self.forward(*input, **kwargs)
    728         for hook in itertools.chain(
    729                 _global_forward_hooks.values(),

C:\ProgramData\Anaconda3\lib\site-packages\transformers\models\bert\modeling_bert.py in forward(self, input_ids, attention_mask, token_type_ids, position_ids, head_mask, inputs_embeds, encoder_hidden_states, encoder_attention_mask, past_key_values, use_cache, output_attentions, output_hidden_states, return_dict)
    957         head_mask = self.get_head_mask(head_mask, self.config.num_hidden_layers)
    958 
--> 959         embedding_output = self.embeddings(
    960             input_ids=input_ids,
    961             position_ids=position_ids,

C:\ProgramData\Anaconda3\lib\site-packages\torch\nn\modules\module.py in _call_impl(self, *input, **kwargs)
    725             result = self._slow_forward(*input, **kwargs)
    726         else:
--> 727             result = self.forward(*input, **kwargs)
    728         for hook in itertools.chain(
    729                 _global_forward_hooks.values(),

C:\ProgramData\Anaconda3\lib\site-packages\transformers\models\bert\modeling_bert.py in forward(self, input_ids, token_type_ids, position_ids, inputs_embeds, past_key_values_length)
    198 
    199         if inputs_embeds is None:
--> 200             inputs_embeds = self.word_embeddings(input_ids)
    201         token_type_embeddings = self.token_type_embeddings(token_type_ids)
    202 

C:\ProgramData\Anaconda3\lib\site-packages\torch\nn\modules\module.py in _call_impl(self, *input, **kwargs)
    725             result = self._slow_forward(*input, **kwargs)
    726         else:
--> 727             result = self.forward(*input, **kwargs)
    728         for hook in itertools.chain(
    729                 _global_forward_hooks.values(),

C:\ProgramData\Anaconda3\lib\site-packages\torch\nn\modules\sparse.py in forward(self, input)
    122 
    123     def forward(self, input: Tensor) -> Tensor:
--> 124         return F.embedding(
    125             input, self.weight, self.padding_idx, self.max_norm,
    126             self.norm_type, self.scale_grad_by_freq, self.sparse)

C:\ProgramData\Anaconda3\lib\site-packages\torch\nn\functional.py in embedding(input, weight, padding_idx, max_norm, norm_type, scale_grad_by_freq, sparse)
   1850         # remove once script supports set_grad_enabled
   1851         _no_grad_embedding_renorm_(weight, input, max_norm, norm_type)
-> 1852     return torch.embedding(weight, input, padding_idx, scale_grad_by_freq, sparse)
   1853 
   1854 

RuntimeError: Expected tensor for argument #1 'indices' to have scalar type Long; but got torch.IntTensor instead (while checking arguments for embedding)

[bhargavvader]

Hmm, is your data quite different from the sample training data available? Are you using Windows? What happens when you use an older version of Transformers?