Tried Running the model with simple instructions posted in readme file. Threw error

[1shoaibazhar]

Error: Traceback (most recent call last): File "/home/shoaib/.local/lib/python3.10/site-packages/pyabsa/tasks/AspectPolarityClassification/dataset_utils/lcf/apc_utils.py", line 385, in configure_spacy_model nlp = spacy.load(config.spacy_model) File "/home/shoaib/.local/lib/python3.10/site-packages/spacy/init.py", line 54, in load return util.load_model( File "/home/shoaib/.local/lib/python3.10/site-packages/spacy/util.py", line 449, in load_model raise IOError(Errors.E050.format(name=name)) OSError: [E050] Can't find model 'en_core_web_sm'. It doesn't seem to be a Python package or a valid path to a data directory.

During handling of the above exception, another exception occurred:

Traceback (most recent call last): File "/home/shoaib/.local/lib/python3.10/site-packages/pyabsa/tasks/AspectPolarityClassification/dataset_utils/lcf/apc_utils.py", line 397, in configure_spacy_model nlp = spacy.load(config.spacy_model) File "/home/shoaib/.local/lib/python3.10/site-packages/spacy/init.py", line 54, in load return util.load_model( File "/home/shoaib/.local/lib/python3.10/site-packages/spacy/util.py", line 449, in load_model raise IOError(Errors.E050.format(name=name)) OSError: [E050] Can't find model 'en_core_web_sm'. It doesn't seem to be a Python package or a valid path to a data directory.

During handling of the above exception, another exception occurred:

Traceback (most recent call last): File "/home/shoaib/Personal Learning/PyAbsa/Test Codes/aspectTermExtractor.py", line 12, in aspect_extractor.predict(['I love this movie, it is so great!'], File "/home/shoaib/.local/lib/python3.10/site-packages/pyabsa/tasks/AspectTermExtraction/prediction/aspect_extractor.py", line 264, in predict return self.batch_predict( File "/home/shoaib/.local/lib/python3.10/site-packages/pyabsa/tasks/AspectTermExtraction/prediction/aspect_extractor.py", line 304, in batch_predict extraction_res, sentence_res = self._extract(target_file) File "/home/shoaib/.local/lib/python3.10/site-packages/pyabsa/tasks/AspectTermExtraction/prediction/aspect_extractor.py", line 374, in _extract infer_features = convert_ate_examples_to_features( File "/home/shoaib/.local/lib/python3.10/site-packages/pyabsa/tasks/AspectTermExtraction/dataset_utils/lcf/data_utils_for_inference.py", line 168, in convert_ate_examples_to_features configure_spacy_model(config) File "/home/shoaib/.local/lib/python3.10/site-packages/pyabsa/tasks/AspectPolarityClassification/dataset_utils/lcf/apc_utils.py", line 399, in configure_spacy_model raise RuntimeError( RuntimeError: Download failed, you can download en_core_web_sm manually.

[yangheng95]

Please check if you can download the en_core_web_sm: pip install spacy python -m spacy download en_core_web_sm

[chen-bowen]

doesn't work for me. I tried the above suggested steps but getting this error now

--------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
Cell In[1], line 12
      6 aspect_extractor = ATEPC.AspectExtractor('english',
      7                                          auto_device=True,  # False means load model on CPU
      8                                          cal_perplexity=True,
      9                                          )
     11 # instance inference
---> 12 aspect_extractor.predict(['I love this movie, it is so great!'],
     13                          save_result=True,
     14                          print_result=True,  # print the result
     15                          ignore_error=True,  # ignore the error when the model cannot predict the input
     16                          )

File ~/opt/anaconda3/envs/mlqu/lib/python3.10/site-packages/pyabsa/tasks/AspectTermExtraction/prediction/aspect_extractor.py:264, in AspectExtractor.predict(self, text, save_result, print_result, pred_sentiment, **kwargs)
    260     return self.batch_predict(
    261         [text], save_result, print_result, pred_sentiment, **kwargs
    262     )[0]
    263 elif isinstance(text, list):
--> 264     return self.batch_predict(
    265         text, save_result, print_result, pred_sentiment, **kwargs
    266     )

File ~/opt/anaconda3/envs/mlqu/lib/python3.10/site-packages/pyabsa/tasks/AspectTermExtraction/prediction/aspect_extractor.py:304, in AspectExtractor.batch_predict(self, target_file, save_result, print_result, pred_sentiment, **kwargs)
    299     raise ValueError(
    300         "Please run inference using examples list or inference dataset path (list)!"
    301     )
    303 if target_file:
--> 304     extraction_res, sentence_res = self._extract(target_file)
    305     results["extraction_res"] = extraction_res
    306     if pred_sentiment:

File ~/opt/anaconda3/envs/mlqu/lib/python3.10/site-packages/pyabsa/tasks/AspectTermExtraction/prediction/aspect_extractor.py:449, in AspectExtractor._extract(self, examples)
    447 l_mask = l_mask.to(self.config.device)
    448 with torch.no_grad():
--> 449     ate_logits, apc_logits = self.model(
    450         input_ids_spc,
    451         token_type_ids=segment_ids,
    452         attention_mask=input_mask,
    453         labels=None,
    454         polarity=polarity,
    455         valid_ids=valid_ids,
    456         attention_mask_label=l_mask,
    457     )
    458 if self.config.use_bert_spc:
    459     label_ids = self.model.get_batch_token_labels_bert_base_indices(
    460         label_ids
    461     )

File ~/opt/anaconda3/envs/mlqu/lib/python3.10/site-packages/torch/nn/modules/module.py:1110, in Module._call_impl(self, *input, **kwargs)
   1106 # If we don't have any hooks, we want to skip the rest of the logic in
   1107 # this function, and just call forward.
   1108 if not (self._backward_hooks or self._forward_hooks or self._forward_pre_hooks or _global_backward_hooks
   1109         or _global_forward_hooks or _global_forward_pre_hooks):
-> 1110     return forward_call(*input, **kwargs)
   1111 # Do not call functions when jit is used
   1112 full_backward_hooks, non_full_backward_hooks = [], []

File ~/opt/anaconda3/envs/mlqu/lib/python3.10/site-packages/pyabsa/tasks/AspectTermExtraction/models/__lcf__/fast_lcf_atepc.py:75, in FAST_LCF_ATEPC.forward(self, input_ids_spc, token_type_ids, attention_mask, labels, polarity, valid_ids, attention_mask_label, lcf_cdm_vec, lcf_cdw_vec)
     73     input_ids = self.get_ids_for_local_context_extractor(input_ids_spc)
     74     labels = self.get_batch_token_labels_bert_base_indices(labels)
---> 75     global_context_out = self.bert4global(
     76         input_ids=input_ids, attention_mask=attention_mask
     77     )["last_hidden_state"]
     78 else:
     79     global_context_out = self.bert4global(
     80         input_ids=input_ids_spc, attention_mask=attention_mask
     81     )["last_hidden_state"]

File ~/opt/anaconda3/envs/mlqu/lib/python3.10/site-packages/torch/nn/modules/module.py:1110, in Module._call_impl(self, *input, **kwargs)
   1106 # If we don't have any hooks, we want to skip the rest of the logic in
   1107 # this function, and just call forward.
   1108 if not (self._backward_hooks or self._forward_hooks or self._forward_pre_hooks or _global_backward_hooks
   1109         or _global_forward_hooks or _global_forward_pre_hooks):
-> 1110     return forward_call(*input, **kwargs)
   1111 # Do not call functions when jit is used
   1112 full_backward_hooks, non_full_backward_hooks = [], []

File ~/opt/anaconda3/envs/mlqu/lib/python3.10/site-packages/transformers/models/deberta_v2/modeling_deberta_v2.py:1049, in DebertaV2Model.forward(self, input_ids, attention_mask, token_type_ids, position_ids, inputs_embeds, output_attentions, output_hidden_states, return_dict)
   1039     token_type_ids = torch.zeros(input_shape, dtype=torch.long, device=device)
   1041 embedding_output = self.embeddings(
   1042     input_ids=input_ids,
   1043     token_type_ids=token_type_ids,
   (...)
   1046     inputs_embeds=inputs_embeds,
   1047 )
-> 1049 encoder_outputs = self.encoder(
   1050     embedding_output,
   1051     attention_mask,
   1052     output_hidden_states=True,
   1053     output_attentions=output_attentions,
   1054     return_dict=return_dict,
   1055 )
   1056 encoded_layers = encoder_outputs[1]
   1058 if self.z_steps > 1:

File ~/opt/anaconda3/envs/mlqu/lib/python3.10/site-packages/torch/nn/modules/module.py:1110, in Module._call_impl(self, *input, **kwargs)
   1106 # If we don't have any hooks, we want to skip the rest of the logic in
   1107 # this function, and just call forward.
   1108 if not (self._backward_hooks or self._forward_hooks or self._forward_pre_hooks or _global_backward_hooks
   1109         or _global_forward_hooks or _global_forward_pre_hooks):
-> 1110     return forward_call(*input, **kwargs)
   1111 # Do not call functions when jit is used
   1112 full_backward_hooks, non_full_backward_hooks = [], []

File ~/opt/anaconda3/envs/mlqu/lib/python3.10/site-packages/transformers/models/deberta_v2/modeling_deberta_v2.py:465, in DebertaV2Encoder.forward(self, hidden_states, attention_mask, output_hidden_states, output_attentions, query_states, relative_pos, return_dict)
    463     input_mask = (attention_mask.sum(-2) > 0).byte()
    464 attention_mask = self.get_attention_mask(attention_mask)
--> 465 relative_pos = self.get_rel_pos(hidden_states, query_states, relative_pos)
    467 all_hidden_states = () if output_hidden_states else None
    468 all_attentions = () if output_attentions else None

File ~/opt/anaconda3/envs/mlqu/lib/python3.10/site-packages/transformers/models/deberta_v2/modeling_deberta_v2.py:445, in DebertaV2Encoder.get_rel_pos(self, hidden_states, query_states, relative_pos)
    443 if self.relative_attention and relative_pos is None:
    444     q = query_states.size(-2) if query_states is not None else hidden_states.size(-2)
--> 445     relative_pos = build_relative_position(
    446         q, hidden_states.size(-2), bucket_size=self.position_buckets, max_position=self.max_relative_positions
    447     )
    448 return relative_pos

File ~/opt/anaconda3/envs/mlqu/lib/python3.10/site-packages/transformers/models/deberta_v2/modeling_deberta_v2.py:564, in build_relative_position(query_size, key_size, bucket_size, max_position)
    562 rel_pos_ids = q_ids[:, None] - np.tile(k_ids, (q_ids.shape[0], 1))
    563 if bucket_size > 0 and max_position > 0:
--> 564     rel_pos_ids = make_log_bucket_position(rel_pos_ids, bucket_size, max_position)
    565 rel_pos_ids = torch.tensor(rel_pos_ids, dtype=torch.long)
    566 rel_pos_ids = rel_pos_ids[:query_size, :]

File ~/opt/anaconda3/envs/mlqu/lib/python3.10/site-packages/transformers/models/deberta_v2/modeling_deberta_v2.py:538, in make_log_bucket_position(relative_pos, bucket_size, max_position)
    536 abs_pos = np.where((relative_pos < mid) & (relative_pos > -mid), mid - 1, np.abs(relative_pos))
    537 log_pos = np.ceil(np.log(abs_pos / mid) / np.log((max_position - 1) / mid) * (mid - 1)) + mid
--> 538 bucket_pos = np.where(abs_pos <= mid, relative_pos, log_pos * sign).astype(np.int)
    539 return bucket_pos

File ~/opt/anaconda3/envs/mlqu/lib/python3.10/site-packages/numpy/__init__.py:305, in __getattr__(attr)
    300     warnings.warn(
    301         f"In the future `np.{attr}` will be defined as the "
    302         "corresponding NumPy scalar.", FutureWarning, stacklevel=2)
    304 if attr in __former_attrs__:
--> 305     raise AttributeError(__former_attrs__[attr])
    307 # Importing Tester requires importing all of UnitTest which is not a
    308 # cheap import Since it is mainly used in test suits, we lazy import it
    309 # here to save on the order of 10 ms of import time for most users
    310 #
    311 # The previous way Tester was imported also had a side effect of adding
    312 # the full `numpy.testing` namespace
    313 if attr == 'testing':

AttributeError: module 'numpy' has no attribute 'int'.
`np.int` was a deprecated alias for the builtin `int`. To avoid this error in existing code, use `int` by itself. Doing this will not modify any behavior and is safe. When replacing `np.int`, you may wish to use e.g. `np.int64` or `np.int32` to specify the precision. If you wish to review your current use, check the release note link for additional information.
The aliases was originally deprecated in NumPy 1.20; for more details and guidance see the original release note at:
    https://numpy.org/devdocs/release/1.20.0-notes.html#deprecations

/Users/bowen.chen/opt/anaconda3/envs/mlqu/lib/python3.10/site-packages/prompt_toolkit/application/application.py:955: DeprecationWarning: There is no current event loop
  loop = asyncio.get_event_loop()

Any ideas?

[yangheng95]

doesn't work for me. I tried the above suggested steps but getting this error now

--------------------------------------------------------------------------
AttributeError                            Traceback (most recent call last)
Cell In[1], line 12
      6 aspect_extractor = ATEPC.AspectExtractor('english',
      7                                          auto_device=True,  # False means load model on CPU
      8                                          cal_perplexity=True,
      9                                          )
     11 # instance inference
---> 12 aspect_extractor.predict(['I love this movie, it is so great!'],
     13                          save_result=True,
     14                          print_result=True,  # print the result
     15                          ignore_error=True,  # ignore the error when the model cannot predict the input
     16                          )

File ~/opt/anaconda3/envs/mlqu/lib/python3.10/site-packages/pyabsa/tasks/AspectTermExtraction/prediction/aspect_extractor.py:264, in AspectExtractor.predict(self, text, save_result, print_result, pred_sentiment, **kwargs)
    260     return self.batch_predict(
    261         [text], save_result, print_result, pred_sentiment, **kwargs
    262     )[0]
    263 elif isinstance(text, list):
--> 264     return self.batch_predict(
    265         text, save_result, print_result, pred_sentiment, **kwargs
    266     )

File ~/opt/anaconda3/envs/mlqu/lib/python3.10/site-packages/pyabsa/tasks/AspectTermExtraction/prediction/aspect_extractor.py:304, in AspectExtractor.batch_predict(self, target_file, save_result, print_result, pred_sentiment, **kwargs)
    299     raise ValueError(
    300         "Please run inference using examples list or inference dataset path (list)!"
    301     )
    303 if target_file:
--> 304     extraction_res, sentence_res = self._extract(target_file)
    305     results["extraction_res"] = extraction_res
    306     if pred_sentiment:

File ~/opt/anaconda3/envs/mlqu/lib/python3.10/site-packages/pyabsa/tasks/AspectTermExtraction/prediction/aspect_extractor.py:449, in AspectExtractor._extract(self, examples)
    447 l_mask = l_mask.to(self.config.device)
    448 with torch.no_grad():
--> 449     ate_logits, apc_logits = self.model(
    450         input_ids_spc,
    451         token_type_ids=segment_ids,
    452         attention_mask=input_mask,
    453         labels=None,
    454         polarity=polarity,
    455         valid_ids=valid_ids,
    456         attention_mask_label=l_mask,
    457     )
    458 if self.config.use_bert_spc:
    459     label_ids = self.model.get_batch_token_labels_bert_base_indices(
    460         label_ids
    461     )

File ~/opt/anaconda3/envs/mlqu/lib/python3.10/site-packages/torch/nn/modules/module.py:1110, in Module._call_impl(self, *input, **kwargs)
   1106 # If we don't have any hooks, we want to skip the rest of the logic in
   1107 # this function, and just call forward.
   1108 if not (self._backward_hooks or self._forward_hooks or self._forward_pre_hooks or _global_backward_hooks
   1109         or _global_forward_hooks or _global_forward_pre_hooks):
-> 1110     return forward_call(*input, **kwargs)
   1111 # Do not call functions when jit is used
   1112 full_backward_hooks, non_full_backward_hooks = [], []

File ~/opt/anaconda3/envs/mlqu/lib/python3.10/site-packages/pyabsa/tasks/AspectTermExtraction/models/__lcf__/fast_lcf_atepc.py:75, in FAST_LCF_ATEPC.forward(self, input_ids_spc, token_type_ids, attention_mask, labels, polarity, valid_ids, attention_mask_label, lcf_cdm_vec, lcf_cdw_vec)
     73     input_ids = self.get_ids_for_local_context_extractor(input_ids_spc)
     74     labels = self.get_batch_token_labels_bert_base_indices(labels)
---> 75     global_context_out = self.bert4global(
     76         input_ids=input_ids, attention_mask=attention_mask
     77     )["last_hidden_state"]
     78 else:
     79     global_context_out = self.bert4global(
     80         input_ids=input_ids_spc, attention_mask=attention_mask
     81     )["last_hidden_state"]

File ~/opt/anaconda3/envs/mlqu/lib/python3.10/site-packages/torch/nn/modules/module.py:1110, in Module._call_impl(self, *input, **kwargs)
   1106 # If we don't have any hooks, we want to skip the rest of the logic in
   1107 # this function, and just call forward.
   1108 if not (self._backward_hooks or self._forward_hooks or self._forward_pre_hooks or _global_backward_hooks
   1109         or _global_forward_hooks or _global_forward_pre_hooks):
-> 1110     return forward_call(*input, **kwargs)
   1111 # Do not call functions when jit is used
   1112 full_backward_hooks, non_full_backward_hooks = [], []

File ~/opt/anaconda3/envs/mlqu/lib/python3.10/site-packages/transformers/models/deberta_v2/modeling_deberta_v2.py:1049, in DebertaV2Model.forward(self, input_ids, attention_mask, token_type_ids, position_ids, inputs_embeds, output_attentions, output_hidden_states, return_dict)
   1039     token_type_ids = torch.zeros(input_shape, dtype=torch.long, device=device)
   1041 embedding_output = self.embeddings(
   1042     input_ids=input_ids,
   1043     token_type_ids=token_type_ids,
   (...)
   1046     inputs_embeds=inputs_embeds,
   1047 )
-> 1049 encoder_outputs = self.encoder(
   1050     embedding_output,
   1051     attention_mask,
   1052     output_hidden_states=True,
   1053     output_attentions=output_attentions,
   1054     return_dict=return_dict,
   1055 )
   1056 encoded_layers = encoder_outputs[1]
   1058 if self.z_steps > 1:

File ~/opt/anaconda3/envs/mlqu/lib/python3.10/site-packages/torch/nn/modules/module.py:1110, in Module._call_impl(self, *input, **kwargs)
   1106 # If we don't have any hooks, we want to skip the rest of the logic in
   1107 # this function, and just call forward.
   1108 if not (self._backward_hooks or self._forward_hooks or self._forward_pre_hooks or _global_backward_hooks
   1109         or _global_forward_hooks or _global_forward_pre_hooks):
-> 1110     return forward_call(*input, **kwargs)
   1111 # Do not call functions when jit is used
   1112 full_backward_hooks, non_full_backward_hooks = [], []

File ~/opt/anaconda3/envs/mlqu/lib/python3.10/site-packages/transformers/models/deberta_v2/modeling_deberta_v2.py:465, in DebertaV2Encoder.forward(self, hidden_states, attention_mask, output_hidden_states, output_attentions, query_states, relative_pos, return_dict)
    463     input_mask = (attention_mask.sum(-2) > 0).byte()
    464 attention_mask = self.get_attention_mask(attention_mask)
--> 465 relative_pos = self.get_rel_pos(hidden_states, query_states, relative_pos)
    467 all_hidden_states = () if output_hidden_states else None
    468 all_attentions = () if output_attentions else None

File ~/opt/anaconda3/envs/mlqu/lib/python3.10/site-packages/transformers/models/deberta_v2/modeling_deberta_v2.py:445, in DebertaV2Encoder.get_rel_pos(self, hidden_states, query_states, relative_pos)
    443 if self.relative_attention and relative_pos is None:
    444     q = query_states.size(-2) if query_states is not None else hidden_states.size(-2)
--> 445     relative_pos = build_relative_position(
    446         q, hidden_states.size(-2), bucket_size=self.position_buckets, max_position=self.max_relative_positions
    447     )
    448 return relative_pos

File ~/opt/anaconda3/envs/mlqu/lib/python3.10/site-packages/transformers/models/deberta_v2/modeling_deberta_v2.py:564, in build_relative_position(query_size, key_size, bucket_size, max_position)
    562 rel_pos_ids = q_ids[:, None] - np.tile(k_ids, (q_ids.shape[0], 1))
    563 if bucket_size > 0 and max_position > 0:
--> 564     rel_pos_ids = make_log_bucket_position(rel_pos_ids, bucket_size, max_position)
    565 rel_pos_ids = torch.tensor(rel_pos_ids, dtype=torch.long)
    566 rel_pos_ids = rel_pos_ids[:query_size, :]

File ~/opt/anaconda3/envs/mlqu/lib/python3.10/site-packages/transformers/models/deberta_v2/modeling_deberta_v2.py:538, in make_log_bucket_position(relative_pos, bucket_size, max_position)
    536 abs_pos = np.where((relative_pos < mid) & (relative_pos > -mid), mid - 1, np.abs(relative_pos))
    537 log_pos = np.ceil(np.log(abs_pos / mid) / np.log((max_position - 1) / mid) * (mid - 1)) + mid
--> 538 bucket_pos = np.where(abs_pos <= mid, relative_pos, log_pos * sign).astype(np.int)
    539 return bucket_pos

File ~/opt/anaconda3/envs/mlqu/lib/python3.10/site-packages/numpy/__init__.py:305, in __getattr__(attr)
    300     warnings.warn(
    301         f"In the future `np.{attr}` will be defined as the "
    302         "corresponding NumPy scalar.", FutureWarning, stacklevel=2)
    304 if attr in __former_attrs__:
--> 305     raise AttributeError(__former_attrs__[attr])
    307 # Importing Tester requires importing all of UnitTest which is not a
    308 # cheap import Since it is mainly used in test suits, we lazy import it
    309 # here to save on the order of 10 ms of import time for most users
    310 #
    311 # The previous way Tester was imported also had a side effect of adding
    312 # the full `numpy.testing` namespace
    313 if attr == 'testing':

AttributeError: module 'numpy' has no attribute 'int'.
`np.int` was a deprecated alias for the builtin `int`. To avoid this error in existing code, use `int` by itself. Doing this will not modify any behavior and is safe. When replacing `np.int`, you may wish to use e.g. `np.int64` or `np.int32` to specify the precision. If you wish to review your current use, check the release note link for additional information.
The aliases was originally deprecated in NumPy 1.20; for more details and guidance see the original release note at:
    https://numpy.org/devdocs/release/1.20.0-notes.html#deprecations

/Users/bowen.chen/opt/anaconda3/envs/mlqu/lib/python3.10/site-packages/prompt_toolkit/application/application.py:955: DeprecationWarning: There is no current event loop
  loop = asyncio.get_event_loop()

Any ideas?

try pip uninstall numpy and then pip install numpy

[chen-bowen]

Same error. Looks like this model is requiring an really old version of numpy <1.20, but pip install would install numpy 1.22.4, which would not solve the issue. I was using this package on python 3.10.6. Looks like python 3.10.9 fix this behavior, but unfortunately I can't use python 3.10.6 due to a product requirement. Would it be possible if you fix this line bucket_pos = np.where(abs_pos <= mid, relative_pos, log_pos * sign).astype(np.int) in modeling_deberta_v2.py? (Would probably just need a retrain) The base deberta_v2 model on huggingface has been updated according to here https://github.com/huggingface/transformers/blob/main/src/transformers/models/deberta_v2/modeling_deberta_v2.py.

[yangheng95]

Then you can revise the code, try np.int_ instead in this code: bucket_pos = np.where(abs_pos <= mid, relative_pos, log_pos * sign).astype(np.int)

[yangheng95]

I notice that this error has been fixed in transfomers, so I supposed it would work if you update the transformers version

[chen-bowen]

That's another product requirement I have to use transformers version 4.18.0, is there any other way this could work?

[yangheng95]

You can clone transformers repo and switch to 4.18.0 version, then move the src/transformers to your project path to make it an local package. Then revise the modeling_deberta_v2.py.

[chen-bowen]

thank you, going to give it a try

[chen-bowen]

I'm a little confused. If transformers are modified, do we need to modify pyabsa too? How does pyabsa know which transformers to use?

[yangheng95]

No need to revise PyABSA at this issue. If you place the transformers package in the current working dir, it would overload the transformers package installed by pip or conda, etc.