A system for quickly generating training data with weak supervision

A PyTorch-based open-source framework that provides methods for improving the weakly annotated data and allows researchers to efficiently develop and compare their own methods.

Retrieve and process PubTator annotations

Big Old Heuristic Repository

Process flow to generate labels on Text data using Snorkel and maintain DB to repurpose unlabelled data

Code for the KDD-2023 paper: Neural-Hidden-CRF: A Robust Weakly-Supervised Sequence Labeler

Snorkel MeTaL: A framework for training models with multi-task weak supervision

Utilizing the snorkel machine learning model to label biomimicry papers. Snorkel uses weak supervision to label large amounts of training data using programmatic labeling functions based on keyword rules.

Mongolian Polarity Detection in Weakly Supervised manner

Code accompanying the TOP paper "Predicting the Demographics of Twitter Users with Programmatic Weak Supervision".

This is a POC project of a system that aim to label unlabeled streaming data using Snorkel

"Snorkel with Turtle" is a realtime labeling system based on Snorkel

Weak Labeling of Fake News Articles with Snorkel and Snuba

Interactive Label Generation Tool for Time Series Data

Sentiment analysis web app for non-technical users.