#amazon-s3 #machine-learning #deep-learning #mlops #local-filesystem #light-gbm

jams-core

jams-core provides thin abstraction around common machine learning and deep learning models and model stores like AWS S3, Azure Blob Storage, MinIO, Local Filesystem. You can think of each component as a LEGO block which can be used to build a system depending on the requirements

41 releases

0.2.22 Dec 3, 2024
0.2.21 Dec 2, 2024
0.2.19 Nov 25, 2024
0.2.15 Oct 30, 2024
0.1.19 Jun 9, 2024

#706 in Machine learning

Download history 2/week @ 2024-09-09 10/week @ 2024-09-16 29/week @ 2024-09-23 3/week @ 2024-09-30 495/week @ 2024-10-14 1014/week @ 2024-10-21 345/week @ 2024-10-28 128/week @ 2024-11-04 135/week @ 2024-11-11 11/week @ 2024-11-18 243/week @ 2024-11-25 837/week @ 2024-12-02 971/week @ 2024-12-09

2,067 downloads per month
Used in 2 crates

Apache-2.0

1MB
4K SLoC

Contains (Zip file, 15KB) pytorch-my_awesome_californiahousing_model.pt, (Zip file, 8KB) torch-my_awesome_penguin_model.pt

JAMS-CORE

This library crate is part of a wider project called J.A.M.S - Just Another Model Server. Please refer here.

Features

  • Async
  • Multiple Model Frameworks Supported
    • Tensorflow
    • Torch
    • Catboost
    • LightGBM
  • Multiple Model Store Backends Supported
    • Local File System
    • AWS S3
    • Azure Blob Storage
    • MinIO
  • Model Store Polling

The following features are in progress 🚧

  • Support XGBoost framework
  • ModelSpec artefacts - Single source of information about models. This will assist in input validations

Overview

Below diagram provides a high level overview of the crate

Alt text

Dependencies

~87MB
~1.5M SLoC