SNORKEL: THE SYSTEM FOR
Labeling and managing training datasets by hand is one of the biggest bottlenecks in machine learning.
In Snorkel, write heuristic functions to do this programmatically instead!
Programmatic or weak supervision sources can be noisy and correlated.
Snorkel uses novel, theoretically-grounded unsupervised modeling techniques to automatically clean and
Snorkel outputs clean, confidence-weighted training datasets that easily plug into any modern machine
start in minutes
# For pip userspip install snorkel
# For conda usersconda install snorkel -c conda-forge
Labeling data for spam classification
Data augmentation for spam classification
Monitoring critical data subsets for spam classification
We introduce a programming model for improving performance on application-critical data subsets, or slices.
Introducing our biggest update to Snorkel yet, version 0.9.
Recap of June 2019 Snorkel workshop.
Snorkel achieves state-of-the-art result on the SuperGLUE NLP benchmark.