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

Promptsource

Promptsource is a Python toolkit and web-based GUI for creating, sharing and applying Jinja natural-language prompts to dataset examples. It includes reusable prompt templates and Python interfaces for retrieving and applying them.

Explore the practical guide ↓

Independent OpenViST project profile, based on linked sources.

Promptsource — 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 Promptsource when a language-processing experiment needs reusable natural-language prompt templates tied to dataset examples, whether for authoring in its GUI or applying templates through Python.

Who it is for

  • NLP researchers preparing or comparing prompt formulations for dataset tasks.
  • Developers who want to organize reusable prompt templates or apply them programmatically to dataset examples.

Documented uses

  • Write and organize prompts as Jinja templates with dataset fields and target outputs.
  • Review authored or existing prompts against examples across a dataset.
  • Retrieve templates for a dataset and apply one to an example in Python.
  • Browse and reuse prompts from the P3 collection.
Before you choose

Requirements & limits

The repository is public and identifies an Apache-2.0 license. The README describes a hosted prompt browser and local installation paths.

  • Prompt templates are written in Jinja and map dataset examples to natural-language input and target output.
  • Using prompts programmatically is distinct from creating them through the GUI; the README gives separate setup guidance for these paths.
  • The local editable-install instructions specify Python 3.7 for running the app.
  • The README's P3 size figures are a snapshot dated January 20, 2022.

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

Apply a prompt to a dataset example

Load a Hugging Face dataset example, choose one of its Promptsource templates and produce the prompted input and target.

Before you start

  • Python environment with the Promptsource package installed using pip install promptsource.
  • Hugging Face Datasets library and a dataset available through it.

Steps

  1. Load a dataset and select an example.

  2. Create a DatasetTemplates collection for the dataset name, or dataset and subset name.

  3. Select a template by its name and call apply(example).

  4. Read the resulting input and target.

Expected outcome

A natural-language input and target generated from the selected dataset example and prompt template.

Guide limitations

  • The guide uses the Python package; it does not cover creating prompts in the web interface.
  • The instructions describe applying a prompt, not running a model.
Documentation 1 ↗
Evidence and transparency

Sources & review

Official repository ↗
Source record details
Source availability
Active
Sources checked
15 October 2026
Original website title
bigscience-workshop/promptsource · Official homepage title · retained in its source language
How we review sourcesSuggest a correction