BlackHC/batchbald_redux

Reusable BatchBALD implementation

52
/ 100
Established

This tool helps machine learning engineers efficiently select the most informative data points to label next for their deep learning models. By analyzing the uncertainty and diversity within your unlabeled data (input), it identifies small batches of examples (output) that will most effectively improve your model's performance when labeled. This is ideal for those building and maintaining deep Bayesian active learning systems.

No commits in the last 6 months. Available on PyPI.

Use this if you need to strategically choose which new data to label to reduce the cost and time of training deep learning models, particularly when using Bayesian Neural Networks and Monte Carlo dropout.

Not ideal if your existing dataset is already fully labeled, or if you are not working with active learning or deep Bayesian models.

active-learning deep-learning-training data-labeling-optimization machine-learning-operations model-efficiency
Stale 6m
Maintenance 0 / 25
Adoption 9 / 25
Maturity 25 / 25
Community 18 / 25

How are scores calculated?

Stars

78

Forks

15

Language

Jupyter Notebook

License

Apache-2.0

Last pushed

Feb 28, 2024

Commits (30d)

0

Dependencies

6

Get this data via API

curl "https://pt-edge.onrender.com/api/v1/quality/ml-frameworks/BlackHC/batchbald_redux"

Open to everyone — 100 requests/day, no key needed. Get a free key for 1,000/day.