OneOpen ML Studio
Open-source, local-first ML data platform — from raw data to training-ready datasets and models.
Computer vision is the deepest path today (boxes, polygons, SAM, Label Assist, versions, local train). Text, tabular, audio, video, and LLM project types ship with sample label UIs and packaging adapters. Team collaboration (orgs, workspaces, members, /setup) ships in 1.0; deeper enterprise (full SSO handshake, realtime cursors) is planned — see Roadmap.
Docs |
|
Version |
1.0.0 |
PyPI |
|
Source |
|
License |
Apache-2.0 |
Local UI |
http://127.0.0.1:8765 |
Quick install
pip install "oneopen-ml-studio[yolo]"
oneopen init
oneopen start
Open http://127.0.0.1:8765 — no cloud account required (Local mode).
Available today
Capability |
Description |
|---|---|
Projects |
CV, text, tabular, audio, video, and LLM workspace types |
CV annotate |
Boxes, polygons, whole-image class labels; SAM + YOLO Label Assist |
Sample annotate |
Spans, transcripts, tabular values, LLM instruction/preference fields |
Versions |
Preprocess, augment, splits (image tasks) |
Train / Export |
Deepest for CV (YOLO / Torchvision); sklearn / HF prepare / modality packages |
Quality & cleaning |
Readiness gate + remediating clean actions per modality |
Pipelines |
Visual DAG editor + YAML/JSON runner |
Collaboration |
Orgs → workspaces → projects; Team/Enterprise |
Plugins |
On-demand install of training extras |
Target platform
See Product vision — annotation, quality, versioning, training readiness, experiments, model registry, active learning, and multi-user collaboration under one local-first / self-hosted roof.
Documentation contents
Product
Getting started
User guide
Project