Practical overview
When it fits
Use Oneke when you can define the structure of the information you want extracted and need a CLI or Web UI workflow for turning text or files into schema-shaped records.
Who it is for
- Developers setting up document-to-structured-data workflows.
- Researchers extracting entities, relations or events against defined schemas.
- Teams constructing knowledge-graph records from unstructured text.
Documented uses
- Extract named entities from text into a defined schema.
- Convert documents into structured event records.
- Extract relations or triples and use them to construct knowledge graphs.
- Configure model, input source, extraction task and output schema for an example workflow.
Requirements & limits
The project is a public repository licensed under MIT. Its README documents local Conda and Docker setup and also links a basic Gradio demo; local deployment is recommended for further features.
- The output schema is a core part of task configuration.
- Model setup differs between API-served models, which use an API key and base URL, and open-source models.
- The documented local setup recommends GPU support and CUDA 12.4; the Docker route lists Docker and NVIDIA Container Toolkit.
- The linked Gradio demo is described as showing basic capabilities, while local deployment supports further features.
Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.
Prepare a local Oneke environment with Conda
Install the documented Python environment as a starting point for trying Oneke’s examples.
Before you start
- Anaconda
- GPU support is recommended; the README recommends CUDA 12.4.
Steps
Clone the Oneke repository and change into its directory.
Create and activate a Conda environment with Python 3.9.
Install the required packages from requirements.txt.
Expected outcome
A configured Conda environment in the Oneke working directory, ready for the repository’s documented example stage.
Guide limitations
- These steps configure the environment; they do not cover running a particular extraction task.
- The README recommends GPU support and CUDA 12.4.
Configure a model and extraction task in an example YAML file
Set the documented model and extraction parameters in a repository example configuration.
Before you start
- The Oneke repository and its example configuration files.
- For an API-served model, its API key and service base URL; these settings are not needed for open-source models.
Steps
Choose a model category and model name or path in the configuration.
For an API-served model, enter the API key and base URL.
Choose an extraction task such as Base, NER, RE or EE, and set an instruction when the selected task calls for one.
Set whether input comes from a file and, if so, provide its path; choose an output schema and extraction mode.
Expected outcome
An example configuration that specifies model settings, input handling, extraction task, schema and mode.
Guide limitations
- The documented configuration steps do not themselves run extraction or describe the resulting output.
- The README directs users to its examples for subsequent CLI or Web UI use.
Sources & review
Source record details
- Source availability
- Active
- Sources checked
- 15 October 2026
- Original website title
- zjunlp/OneKE · Official homepage title · retained in its source language