Edit model card

What is YODA

YODA is a series of models for Google Feed product optimization. We aim to increase the market reach for ecommerce by augmenting and improving certain metadata like short titles, colors, measures and more. YODA is being used in production by +300 companies with +3.5M products.

What we use NER for

We have trained a NER model for product feature extraction. We retrieve data like colors, sizes, brands and energy labels. Trained with +3M lines of product metadata, the model returns the next scores:

Results:

  • F-score (micro) 0.972
  • F-score (macro) 0.9692
  • Accuracy 0.9461

By class:

precision recall f1-score support
size 0.9734 0.9793 0.9764 26707
brand 0.9618 0.9788 0.9702 15621
color 0.9566 0.9612 0.9589 6785
energy 0.9444 1.0000 0.9714 119
precision recall f1-score support
micro avg 0.9673 0.9767 0.9720 49232
macro avg 0.9591 0.9798 0.9692 49232
weighted avg 0.9674 0.9767 0.9720 49232

Demo: How to use in Flair

Requires:

  • Flair (pip install flair)
from flair.data import Sentence
from flair.models import SequenceTagger

# load tagger
tagger = SequenceTagger.load("lighthousefeed/yoda-ner")

# make example sentence
sentence = Sentence("Jean Paul Gaultier Classique - 50 ML Eau de Parfum  Damen Parfum.")

# predict NER tags
tagger.predict(sentence)

# print sentence
print(sentence)

# print predicted NER spans
print('The following NER tags are found:')

# iterate over entities and print
for entity in sentence.get_spans('ner'):
    print(entity)

Contact

Contact the lead ML developer Iván R. Gázquez for any inquiry. We love hearing what you used this model for!

Downloads last month
49
Inference Examples
This model does not have enough activity to be deployed to Inference API (serverless) yet. Increase its social visibility and check back later, or deploy to Inference Endpoints (dedicated) instead.