TypeError: list indices must be integers or slices, not str

[AshishKumarB]

The below code was working perfectly in pytorch 1.4 and python3.6

import nlpaug.augmenter.sentence as nas
aug = nas.ContextualWordEmbsForSentenceAug(model_path='xlnet-base-cased')
text= "Please find below some recent comparative data between the below product and ibrance"
augmented_texts = aug.augment(text, n=3)

---------------------------------------------- Error-----------------------------------------------------------

TypeError Traceback (most recent call last)
in
1 aug = nas.ContextualWordEmbsForSentenceAug(model_path='xlnet-base-cased')
2 text= "Please find below some recent comparative data between the below product and ibrance"
----> 3 augmented_texts = aug.augment(text, n=3)

~/anaconda3/envs/pytorch_p36/lib/python3.6/site-packages/nlpaug/base_augmenter.py in augment(self, data, n, num_thread)
93 elif self.class.name in ['AbstSummAug', 'BackTranslationAug', 'ContextualWordEmbsAug', 'ContextualWordEmbsForSentenceAug']:
94 for _ in range(aug_num):
---> 95 result = action_fx(clean_data)
96 if isinstance(result, list):
97 augmented_results.extend(result)

~/anaconda3/envs/pytorch_p36/lib/python3.6/site-packages/nlpaug/augmenter/sentence/context_word_embs_sentence.py in insert(self, data)
96 all_data = [data]
97
---> 98 return self.model.predict(all_data)
99
100 @classmethod

~/anaconda3/envs/pytorch_p36/lib/python3.6/site-packages/nlpaug/model/lang_models/text_generation_transformers.py in predict(self, texts, target_words, n)
88 results.extend(predict_result)
89
---> 90 return [r['generated_text'] for r in results]

~/anaconda3/envs/pytorch_p36/lib/python3.6/site-packages/nlpaug/model/lang_models/text_generation_transformers.py in (.0)
88 results.extend(predict_result)
89
---> 90 return [r['generated_text'] for r in results]

TypeError: list indices must be integers or slices, not str

[makcedward]

transformers changed interface from 4.11. You may either use transformers 4.10 or upgrading nlpaug to 1.1.8

[makcedward]

Fixed in 1.1.8 version

[TimbusCalin]

This still doesn't work for me. I tried with transformers==4.10 and nlp==1.1.8 with Python 3.8 and I still get this error thrown.

[TimbusCalin]

If anyone is interested, I made it work with:

transformers==4.11.3
sentencepiece==0.1.96
nlpaug==1.1.8