Evgeneus/Label-Smoothing-in-Text-Classification
Soft Target and Label Smoothing in Text Classification for Probability Calibration of Output Distributions.
This project helps machine learning practitioners improve the reliability and accuracy of text classification models, especially when using data gathered from crowdsourcing. It takes raw text data and associated labels (including potentially noisy crowd-sourced labels) and outputs a better-performing and more confidently calibrated text classification model. Anyone building or evaluating text classification systems that rely on human-annotated data would benefit from this.
No commits in the last 6 months.
Use this if you are developing text classification models, particularly with datasets annotated by multiple crowd workers, and need to improve your model's prediction accuracy and the trustworthiness of its probability scores.
Not ideal if your text classification models do not rely on crowd-annotated data or if you are not concerned with the calibration of your model's probability outputs.
Stars
9
Forks
—
Language
Python
License
—
Category
Last pushed
Sep 09, 2020
Commits (30d)
0
Get this data via API
curl "https://pt-edge.onrender.com/api/v1/quality/nlp/Evgeneus/Label-Smoothing-in-Text-Classification"
Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.
Higher-rated alternatives
fidelity/textwiser
[AAAI 2021] TextWiser: Text Featurization Library
RandolphVI/Multi-Label-Text-Classification
About Muti-Label Text Classification Based on Neural Network.
ThilinaRajapakse/pytorch-transformers-classification
Based on the Pytorch-Transformers library by HuggingFace. To be used as a starting point for...
ntumlgroup/LibMultiLabel
A library for multi-class and multi-label classification
xuyige/BERT4doc-Classification
Code and source for paper ``How to Fine-Tune BERT for Text Classification?``