OpenViSTAI · Agents · Blockchain · Crypto
Submit a project
Menu
AI Models

Rebel

REBEL is a sequence-to-sequence relation-extraction model that generates structured triplets linking entities in natural-language text. The Babelscape repository provides Python training and testing code, dataset scripts, configuration files, and a demo, alongside instructions for using a Hugging Face model.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Rebel — OpenViST project profile in AI Models.
OpenViST project illustration. See the linked sources for product details.
Understand the project

Practical overview

When it fits

Use REBEL when a workflow needs relation triplets extracted from text and can accommodate generated-token decoding and parsing before consuming structured output.

Who it is for

  • NLP practitioners building relation-extraction pipelines.
  • Teams converting text into structured entity relationships for knowledge-base or downstream information-extraction work.
  • Researchers exploring sequence-to-sequence relation extraction and its multilingual direction.

Documented uses

  • Extract subject–relation–object triplets from natural-language passages.
  • Prepare structured relationships for downstream knowledge-graph or knowledge-base workflows.
  • Train or test the model using the repository's Python code and dataset scripts.
Before you choose

Requirements & limits

  • The demonstrated inference path uses Hugging Face Transformers with Babelscape/rebel-large.
  • Generated special tokens need decoding and parsing to produce triplets.
  • The repository's setup script uses Conda and installs PyTorch plus requirements-file dependencies.
  • The README announces mREBEL and multilingual datasets, but the supplied material does not specify supported languages.

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

Load REBEL and parse generated triplets

Use the documented Transformers example to generate relation text and parse it into subject, relation, and object fields.

Before you start

  • Python environment with the Transformers package and dependencies installed.
  • Access to the Hugging Face model Babelscape/rebel-large.

Steps

  1. Create a Transformers text2text-generation pipeline using Babelscape/rebel-large as both model and tokenizer.

  2. Pass text to the pipeline to generate token IDs.

  3. Decode the generated token IDs with the tokenizer to preserve the special tokens.

  4. Parse the decoded sequence using the documented triplet markers to produce head, type, and tail fields.

Expected outcome

Generated relation output is converted into structured entity-relation triplets.

Guide limitations

  • The README example does not specify hardware or runtime requirements.
  • The generated format requires manual token-aware decoding and parsing.
  • The repository evidence does not establish an API or hosted inference service.
Documentation 1 ↗
Evidence and transparency

Sources & review

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