Blog • Highlights • Overview • Install • Getting Started • Documentation • Tutorials • Contributing
Koursaros is a distributed cloud platform for developing and deploying neural search and inference applications.
Koursaros leverages a general-purpose microservice architecture to enable low-latency, scalable deep neural network training and can be directly deployed to kubernetes for production.
This is page is a work in progress.
| Benchmark | Label Accuracy | Paper | Models |
|---|---|---|---|
| fever.ai | 0.7396 (2nd) | An Automated Fact Checker in Era of Fake News | coming soon |
You need Python 3.6 or later to run Koursaros.
We recommend installing Koursaros via pip:
pip3 install koursaros
Installation will use Python wheels from PyPI, available for OSX, Linux, and Windows.
You can install the latest version from Git:
pip3 install git+https://[email protected]/koursaros-ai/koursaros.git
kctl deploy app
- Use Koursaros to get SoTA results in dev environment on the fever.ai benchmark using pretrained models.
- Training custom models and deploying them as stubs
- Training Elastic Search BM25 algorithm using Ax Bayesian Optimizer (coming soon)
- Deploying fever.ai pipeline to production (Coming Soon)