Hugging Face's logo

quim-motger
/
reviewBERT-large

Fill-Mask
Transformers
PyTorch
bert
Inference Endpoints
Model card Files Community
reviewBERT-large / README.md
quim-motger's picture
quim-motger
Update README
ea0973f verified
preview code
|
raw
history blame contribute delete
No virus
2.35 kB
metadata
license: gpl-3.0

reviewBERT-large

This model is a fine-tuned version of bert-large-uncased on a large dataset of mobile app reviews. The model is designed to understand and process text from mobile app reviews, providing enhanced performance for tasks such as feature extraction, sentiment analysis and review summarization from app reviews.

Model Details

  • Model Architecture: BERT (Bidirectional Encoder Representations from Transformers)
  • Base Model: bert-large-uncased
  • Pre-training Extension: Mobile app reviews dataset
  • Language: English

Dataset

The extended pre-training was performed using a diverse dataset of mobile app reviews collected from various app stores. The dataset includes reviews of different lengths, sentiments, and topics, providing a robust foundation for understanding the nuances of mobile app user feedback.

Training Procedure

The model was fine-tuned using the following parameters:

  • Batch Size: 16
  • Learning Rate: 2e-5
  • Epochs: 2

Usage

Load the model 

from transformers import BertTokenizer, BertForSequenceClassification

tokenizer = BertTokenizer.from_pretrained('quim-motger/reviewBERT-large')
model = BertForSequenceClassification.from_pretrained('quim-motger/reviewBERT-large')

Example: Sentiment Analysis 

from transformers import pipeline

nlp = pipeline('sentiment-analysis', model=model, tokenizer=tokenizer)

review = "This app is fantastic! I love the user-friendly interface and features."
result = nlp(review)

print(result)
# Output: [{'label': 'POSITIVE', 'score': 0.98}]

Example: Review Summarization 

from transformers import pipeline

summarizer = pipeline('summarization', model=model, tokenizer=tokenizer)

long_review = "I have been using this app for a while and it has significantly improved my productivity.
The range of features is excellent, and the user interface is intuitive. However, there are occasional
bugs that need fixing."
summary = summarizer(long_review, max_length=50, min_length=25, do_sample=False)

print(summary)
# Output: [{'summary_text': 'The app has significantly improved my productivity with its excellent features and intuitive user interface. However, occasional bugs need fixing.'}]