from summarizer import Summarizer it's not running in Colab

Question

I'm trying to use the BERT Text Summarizer inside Colab but I'm getting the following error

from summarizer import Summarizer

I am getting the error as below,

    ---------------------------------------------------------------------------
ImportError                               Traceback (most recent call last)
<ipython-input-12-ebcc505c0b7d> in <module>()
----> 1 from summarizer import Summarizer

1 frames
/usr/local/lib/python3.7/dist-packages/summarizer/model_processors.py in <module>()
      2 
      3 import numpy as np
----> 4 from transformers import (AlbertModel, AlbertTokenizer, BartModel, BigBirdModel, BigBirdTokenizer,
      5                           BartTokenizer, BertModel, BertTokenizer,
      6                           CamembertModel, CamembertTokenizer, CTRLModel,

ImportError: cannot import name 'BartModel' from 'transformers' (/usr/local/lib/python3.7/dist-packages/transformers/__init__.py)

Answer 1

According to the Huggingface documentation, using summarization pipeline in the easiest way can be implemented like this:

from transformers import pipeline 
# use bart in pytorch
summarizer = pipeline("summarization")
ptorch = summarizer("An apple a day, keeps the doctor away", min_length=5, max_length=20)

# use t5 in tf
summarizer = pipeline("summarization", model="t5-base", tokenizer="t5-base", framework="tf")
tflow = summarizer("An apple a day, keeps the doctor away", min_length=5, max_length=20)

print(ptorch, tflow)

output for pyTorch and tensorflow:

[{'summary_text': ' An apple a day, keeps the doctor away from your doctor away, says Dr.'}] [{'summary_text': 'an apple a day, keeps the doctor away from the doctor .'}]

but if you are willing to use BertModel as the model you would have two options:

1. Break-down pipeline into tokenization ==> encoding ==> decoding (do the work manually)
2. Pass a Bert-based model like bert-base-cased as the model argument to the pipeline.