'use simpletransformers.language_representation to extract embedding from bert

i want to extract embedding from each post using bert and save it as . mat file I use anaconda jupterlab

from simpletransformers.language_representation import RepresentationModel
!pip3 install torch torchvision torchaudio
import torch
from torch import Tensor       


model = RepresentationModel(
        model_type="bert",
        model_name="bert-base-uncased",
        args=model_args,
        use_cuda=False
    ) 
c=1
for index, row in df4.iterrows():
  post= row['new posts']
  label=row['gt']
  word_vector= model.encode_sentences(post, combine_strategy=None)
  scipy.io.savemat(r"/Documents/embedding_bert/post"+ str(c) +".mat", {'myposts': word_vector,'label': label ,'post_no':c})
  c=c+1

I got error in

word_vector= model.encode_sentences(post, combine_strategy=None)

the error is

Unable to convert output to PyTorch tensors format, PyTorch is not installed.

Sources

This article follows the attribution requirements of Stack Overflow and is licensed under CC BY-SA 3.0.

Source: Stack Overflow

Solution	Source

'use simpletransformers.language_representation to extract embedding from bert

Sources

Related Questions