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

Jointbert

Jointbert is an unofficial Python and PyTorch implementation of a BERT-based model that jointly predicts intent labels and semantic slot labels from text. It provides training, evaluation and file-based prediction scripts, supports several BERT-family model variants, and offers optional CRF decoding for slot sequences.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Jointbert — 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 Jointbert when you need a codebase for joint intent classification and slot filling, and your workflow can work with its Python/PyTorch training scripts or file-based prediction.

Who it is for

  • Developers building intent-and-slot understanding into task-oriented language systems.
  • Researchers comparing joint intent classification and slot filling on ATIS or Snips.
  • Python and PyTorch practitioners who want a training and file-prediction codebase to adapt.

Documented uses

  • Train a model to classify an utterance's intent while assigning semantic labels to its slots.
  • Evaluate joint predictions on the included ATIS or Snips task configurations.
  • Generate predictions from sentences supplied through an input file.
  • Compare documented BERT-family model variants, with or without CRF decoding.
Before you choose

Requirements & limits

The public GitHub repository is licensed under Apache-2.0.

  • The README specifies Python 3.6 or later, PyTorch 1.6.0 and Transformers 3.0.2, alongside seqeval and pytorch-crf dependencies.
  • Training requires intent and slot labels; the README calls for UNK handling for labels absent from training and PAD for slot labels.
  • The repository reports different outcomes across model variants and datasets, and flags possible convergence difficulty for ALBERT xxlarge slot prediction.
  • Prediction is documented as input-file to output-file processing, not as a hosted service or API.

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 evaluate an ATIS or Snips model

Use the documented training and evaluation workflow for one of the repository's supported task configurations.

Before you start

  • Python 3.6 or later
  • PyTorch 1.6.0
  • Transformers 3.0.2
  • seqeval 0.0.12
  • pytorch-crf 0.7.2 when using the CRF option

Steps

  1. Prepare the task labels, including UNK labels for labels appearing only in development or test data and PAD for slot labels.

  2. Run main.py with the selected task, model type and model directory, enabling --do_train and --do_eval.

  3. Add --use_crf if CRF decoding is wanted.

Expected outcome

A trained model is evaluated for the selected task.

Guide limitations

  • The README reports results only for uncased models.
  • The project notes that ALBERT xxlarge may not converge well for slot prediction.
Documentation 1 ↗
Evidence and transparency

Sources & review

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