makcedward / nlpaug

Data augmentation for NLP
https://makcedward.github.io/
MIT License
4.46k stars 463 forks source link

Numpy Error #320

Open akhileshrai003 opened 1 year ago

akhileshrai003 commented 1 year ago

import nlpaug.augmenter.word as naw aug = naw.ContextualWordEmbsAug() text="Transformers are the most popular toys" print(f"original text:{text}") print(f"Augmented text: {aug.augment(text)}")

original text:Transformers are the most popular toys

RuntimeError Traceback (most recent call last) Input In [28], in <cell line: 3>() 1 text="Transformers are the most popular toys" 2 print(f"original text:{text}") ----> 3 print(f"Augmented text: {aug.augment(text)}")

File ~/.local/lib/python3.9/site-packages/nlpaug/base_augmenter.py:98, in Augmenter.augment(self, data, n, numthread) 96 elif self.class.name in ['AbstSummAug', 'BackTranslationAug', 'ContextualWordEmbsAug', 'ContextualWordEmbsForSentenceAug']: 97 for in range(aug_num): ---> 98 result = action_fx(clean_data) 99 if isinstance(result, list): 100 augmented_results.extend(result)

File ~/.local/lib/python3.9/site-packages/nlpaug/augmenter/word/context_word_embs.py:471, in ContextualWordEmbsAug.substitute(self, data) 468 if not len(masked_texts): 469 continue --> 471 outputs = self.model.predict(masked_texts, target_words=original_tokens, n=2) 473 # Update doc 474 for original_token, aug_input_pos, output, masked_text in zip(original_tokens, aug_input_poses, outputs, masked_texts):

File ~/.local/lib/python3.9/site-packages/nlpaug/model/lang_models/bert.py:113, in Bert.predict(self, texts, target_words, n) 111 seed = {'temperature': self.temperature, 'top_k': self.top_k, 'top_p': self.top_p} 112 target_token_logits = self.control_randomness(target_token_logits, seed) --> 113 target_token_logits, target_token_idxes = self.filtering(target_token_logits, seed) 114 if len(target_token_idxes) != 0: 115 new_tokens = self.pick(target_token_logits, target_token_idxes, target_word=target_token, n=10)

File ~/.local/lib/python3.9/site-packages/nlpaug/model/lang_models/language_models.py:146, in LanguageModels.filtering(self, logits, seed) 144 if 'cuda' in self.device: 145 idxes = idxes.cpu() --> 146 idxes = idxes.detach().numpy().tolist() 147 else: 148 idxes = np.arange(len(logits)).tolist()

RuntimeError: Numpy is not available