FuChong-cyber / label-inference-attacks

Code & supplementary material of the paper Label Inference Attacks Against Federated Learning on Usenix Security 2022.
MIT License
79 stars 16 forks source link

Help about the bert on Yahoo Answers. #9

Open mitthumeng opened 7 months ago

mitthumeng commented 7 months ago

I am trying to run vfl_framework.py based on the dataset Yahoo Answer. I have downloaded the Bert model 'bert-base-uncased' and saved it in models/transformers/. However, something is still wrong. error

judge-x commented 2 weeks ago

i download the bert to local and try to import with: self.bert = BertModel4Mix.from_pretrained('./models/Bert/bert-base-uncased') but its final top-1 accuray is ~37% So it make me confused

mitthumeng commented 2 weeks ago

i download the bert to local and try to import with: self.bert = BertModel4Mix.from_pretrained('./models/Bert/bert-base-uncased') but its final top-1 accuray is ~37% So it make me confused

Hello, I have solved the problem. Maybe the data in Yahoo Answers should be preprocessed.

judge-x commented 2 weeks ago

wow,thank you, any reference code for preprocessing?

mitthumeng commented 2 weeks ago

wow,thank you, any reference code for preprocessing?

https://github.com/SALT-NLP/MixText/blob/master/data/yahoo_answers_csv/Pre-process.ipynb

judge-x commented 2 weeks ago

thank you, i will try. orz