OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Developer Tools & SDKs

Openue

Openue is a Python toolkit for building, training, testing and deploying pretrained-language-model-based NLP systems that extract structured information from text. It includes workflows for entity and relation extraction, event extraction, and intent and slot extraction, with model, training and data modules.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Openue — OpenViST project profile in AI Developer Tools & SDKs.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use Openue when you need a Python codebase and documented training workflow for structured text extraction, or a path from trained extraction models to TorchServe packaging.

Who it is for

  • NLP developers building knowledge-graph extraction pipelines from text.
  • Researchers training or validating entity recognition and relation-classification models.
  • Teams packaging extraction models for TorchServe.

Documented uses

  • Prepare text and structured relation examples for extraction models.
  • Train named-entity recognition and sentence-level relation-classification modules.
  • Validate the two modules together through a pipeline inference configuration.
  • Package trained model files and submit them to TorchServe.
Before you choose

Requirements & limits

The repository is public and licensed under MIT. The README specifies Python 3.8 and requirements.txt; it also documents installation through pip or from the repository.

  • The documented pipeline requires separate training of NER and relation modules before unified validation.
  • The codebase uses Transformers tokenizers and PyTorch Lightning-based training interfaces.
  • The README includes an English-data configuration note: change model_name_or_path from the Chinese pretrained model to an English model such as bert-base-uncased.
  • The deployment example depends on TorchServe, Docker and model artifacts with their configuration and tokenizer files.

Check current requirements in the official documentation. Source-based guidance does not mean OpenViST has run or independently tested this project.

Source-based practical guide

Train and validate entity and relation extraction models

Use the documented scripts to train the NER and sentence-level relation modules, then run pipeline validation with their saved weights.

Before you start

  • Python 3.8 and packages listed in requirements.txt.
  • JSON training data, with dataset files placed in ./dataset/.
  • Network access if the dataset directory is empty, so the scripts can download datasets and pretrained models.

Steps

  1. Install the requirements and Openue using the repository's setup instructions.

  2. Place the dataset in ./dataset/, or allow the scripts to download data and pretrained models.

  3. Run ./scripts/run_ner.sh and ./scripts/run_seq.sh to train the two modules.

  4. Set the resulting output/ner/${dataset} and output/seq/${dataset} weight directories as ner_model_name_or_path and seq_model_name_or_path in run_infer.yaml or run_infer.sh, then run the validation script.

Expected outcome

Separate trained NER and relation-classification weights are available for pipeline validation.

Guide limitations

  • The README says the pipeline modules cannot be trained jointly and must be trained separately before unified validation.
  • The README warns that downloads require a network connection.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
17 October 2026
Original website title
zjunlp/OpenUE · Official homepage title · retained in its source language
How we review sourcesSuggest a correction