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

Multi Task Nlp

Multi Task Nlp is a Python toolkit for defining, training and running a shared transformer-based model across multiple natural-language understanding tasks. Developers configure tasks in YAML, prepare their data, train the model and use an inference pipeline to request predictions.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Multi Task Nlp — 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 it when developing an NLU pipeline that combines several tasks, such as intent detection and NER, and you want to configure and train them through one shared-model workflow.

Who it is for

  • NLP developers building conversational AI systems with multiple language tasks.
  • Teams that need intent classification, named-entity recognition or other supported NLU tasks in a shared-model workflow.

Documented uses

  • Define multiple tasks and their data and model settings in a YAML file.
  • Prepare task data and train a shared transformer model.
  • Request predictions for selected tasks from samples using a trained model.
Before you choose

Requirements & limits

The project is an open-source library under the Apache-2.0 license, distributed through its public GitHub repository.

  • The README documents a Python 3.7.3 development and test environment and recommends installation in a virtual environment.
  • The workflow requires task-specific YAML configuration and data files before data preparation and training.
  • Inference uses a trained model and explicit task names for the predictions requested.

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

Prepare data and train a multi-task model

Train a shared transformer model for tasks defined in a YAML task file.

Before you start

  • The repository cloned locally with its requirements installed; Python 3.7.3 is the version stated as built and tested.
  • Training and development data files named in the YAML task file, and any referenced label-map file.

Steps

  1. Create a YAML task file defining each task's model type, configuration, labels, metrics, loss, task type, and data filenames.

  2. Prepare the data with the command below.

    Documented command
    python data_preparation.py --task_file 'sample_task_file.yml' --data_dir 'data' --max_seq_len 50

  3. Train with `python train.py --data_dir 'data/bert-base-uncased_prepared_data' --task_file 'sample_task_file.yml' --out_dir 'sample_out' --epochs 5 --train_batch_size 4 --eval_batch_size 8 --grad_accumulation_steps 2 --log_per_updates 25 --save_per_updates 1000 --eval_while_train True --test_while_train True --max_seq_l

Expected outcome

The training command is configured to write model output to `sample_out`; the README describes training a single model across the tasks defined in the file.

Guide limitations

  • The documented task-file example uses BERT and specifies a particular prepared-data directory; adjust parameters to match the task configuration and data.
  • The repository recommends installing requirements in a virtual environment.
  • The documented training workflow does not establish performance or successful execution for a particular dataset or environment.
Documentation 1 ↗Documentation 2 ↗
Source-based practical guide

Run inference with a trained multi-task model

Request predictions for named tasks using a trained model and input samples.

Before you start

  • A trained model file, such as sample_out_dir/multi_task_model.pt.
  • Input samples and task names corresponding to tasks in the trained model.

Steps

  1. Import the inference class with from infer_pipeline import inferPipeline.

  2. Create a pipeline with pipe = inferPipeline(modelPath='sample_out_dir/multi_task_model.pt', maxSeqLen=50).

  3. Provide samples and task names, then call pipe.infer(samples, tasks), for example with samples = [['sample_sentence_1'], ['sample_sentence_2']] and tasks = ['TaskA', 'TaskB'].

Expected outcome

The inference pipeline returns predictions for the supplied samples and requested task names.

Guide limitations

  • This workflow assumes a trained model and task names available in that model.
  • The documented example shows sample sentences and two tasks; it does not specify output interpretation or guarantees for other inputs.
Documentation 1 ↗Documentation 2 ↗
COMPARE OPTIONS

Compare options for Model inference and serving

These projects document a common task and share a project role. This is a source-based comparison, not an execution benchmark or a guarantee of interchangeability.

ProjectDocumented purposePlatformAccess and requirements
Multi Task NlpDefine multiple tasks and their data and model settings in a YAML file.Linux, macOS, WindowsThe project is an open-source library under the Apache-2.0 license, distributed through its public GitHub repository.
HyperposeCustomize pose-estimation architectures and datasets.Docker, Linux, macOS, Windows
Evidence and transparency

Sources & review

Official repository ↗Guide documentation ↗
Source record details
Source availability
Active
Sources checked
17 October 2026
Original website title
hellohaptik/multi-task-NLP · Official homepage title · retained in its source language
How we review sourcesSuggest a correction