metadata

base_model: distilbert/distilbert-base-uncased
datasets:
  - arize-ai/ecommerce_reviews_with_language_drift
language:
  - en
library_name: transformers
pipeline_tag: text-classification

Model Description

This is a text classification model based on DistilBERT. It has been fine-tuned on the ecommerce_reviews_with_language_drift dataset.

Intended Use

The model is used for classifying product reviews in text format. The probable outputs are 'positive', 'negative' and 'neutral'.

Training Data

The arize-ai/ecommerce_reviews_with_language_drift dataset was used for training. Only the 'text' and 'label' columns were used. The training dataset contains 8k rows out of which 34.1% are labeled 'positive', 33.4 % are labeled 'negative' and 32.5% are labeled 'neutral'. So it is a balanced dataset.

Evaluation

The model was fine tuned based on the F1 score for 50 epochs. The best score obtained was 0.67.

Example Usage

from transformers import pipeline

classifier = pipeline("text-classification", model="your-model-identifier")

result = classifier("Your example text here")
print(result)