|
from transformers import pipeline |
|
import gradio as gr |
|
|
|
|
|
model_checkpoint = 'zinoubm/e-comerce-category-classification' |
|
model = pipeline( |
|
"text-classification", model=model_checkpoint, |
|
) |
|
|
|
def predict(input): |
|
predictions = model(input) |
|
predictions = [prediction['label'] for prediction in predictions] |
|
return ' '.join(predictions) |
|
|
|
|
|
title = 'E-Commerce Category Prediction.' |
|
|
|
description = ''' |
|
This is a classification model that predicts the category of an input. |
|
We have 4 Categories, Electronics, Household, Books and Clothing & Accessories. |
|
''' |
|
|
|
article = ''' |
|
# How to use this interface |
|
Using the interface is straight forward, just type some text that falls in one of these 4 categories: **Electronics**, **Household**, **Books** or **Clothing & Accessories**. |
|
and then hit **Submit**. the results will be in the output cell. You can also try one of the provided examples. |
|
Here is the [notebook](https://colab.research.google.com/drive/1MGlDaJXcjECrSRmXgHzUgFBTDuKEwPV0?usp=share_link) used to train the model. |
|
''' |
|
|
|
examples = [ |
|
['I want to sell a laptop'], |
|
['This is a beatiful T-shirt'], |
|
['Save 50% on detergent powder'] |
|
] |
|
|
|
|
|
gr.Interface(fn=predict, |
|
inputs="text", |
|
title=title, |
|
description=description, |
|
article=article, |
|
outputs="text", |
|
examples = examples, |
|
theme='default', |
|
).launch() |
|
|