#4545th of 105 in ML frameworks and libraries
vowpal_wabbit
Online machine learning library with contextual bandits and hashed features
- Stars
- 8.7k
- License
- custom license
- Last commit
- Sep 2026
- Last release
- Sep 2026
- Language
- C++
8,625 on 1 Apr 20268,731 on 21 Oct 2026
From GitHub's daily star counts.
Checked on GitHub.
Overview
Vowpal Wabbit is a C++ online learning system that trains with sparse gradient descent and other optimizers, streaming data so memory stays bounded regardless of training set size. It uses the hashing trick for features and supports reductions, allreduce, active learning and several contextual bandit algorithms. It is a classic ML tool, not an LLM project.
Who it is for: ML engineers needing fast online learning or contextual bandits
Strengths
- Memory footprint is bounded; training data is not loaded into RAM
- Hashing trick bounds feature set size independent of data volume
- Several contextual bandit algorithms for reinforcement learning
- Runs on Windows, macOS and Linux; Python notebooks and CLI demos included
Weaknesses
- Not an LLM or deep learning tool; no neural model serving described
- README gives no hardware requirements or benchmark numbers
- License is detected as NOASSERTION; check the repository license file
- Most documentation lives in the wiki, not the README
What it needs
- Models: sparse gradient descent, contextual bandits, bag-of-words text features
Also in ML frameworks and libraries
See all 105| Rank | Project | Score |
|---|---|---|
| 1 | transformersPython library defining open models for text, vision, audio and multimodal tasks | 94 out of 100 |
| 2 | pytorchPython tensor library with GPU support and tape-based autograd for deep learning | 94 out of 100 |
| 3 | ultralyticsPython library and CLI for training and running YOLO vision models | 93 out of 100 |
| #4 | pytorch-image-modelsPyTorch library of image models, pretrained weights and reference training scripts | 92 out of 100 |
| #5 | supervisionPython toolkit for loading, annotating and counting computer vision detections | 91 out of 100 |