# facebookresearch/ParlAI

A framework for training and evaluating AI models on a variety of openly available dialogue datasets.

Repository: https://github.com/facebookresearch/ParlAI
Canonical: https://ross.abutalabs.com/products/parlai
Homepage: https://parl.ai
Language: Python
License: MIT
License Family: permissive
Archived: true
Last push: 2026-07-30T21:27:52+00:00

## Health v2 (maintenance only)
Score: 10/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 95, release rhythm 8, longevity 100
- inputs: {"age_days": 3418, "days_push": 34, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: archived
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 10621, forks 2052 (observed 2026-08-28T04:10:42.590323+00:00)

## What it is
ParlAI is a Python framework from Facebook AI Research for sharing, training, and evaluating dialogue models across many openly available datasets, from open-domain chitchat to task-oriented dialogue and visual question answering. It unifies 100+ datasets behind a single API, provides reference and pretrained models, and integrates with Mechanical Turk and chat services for data collection and human evaluation.

## Use cases
- train and evaluate dialogue models on datasets like PersonaChat or Wizard of Wikipedia
- benchmark chatbots across 100+ NLP datasets with one API
- run human evaluations via Amazon Mechanical Turk for dialogue agents
- deploy a pretrained conversational model in a chat interface
- train transformer models on question answering tasks like SQuAD or HotpotQA
- collect new dialogue training data through crowdsourcing
- experiment with retrieval and generative dialogue baselines

## When to choose
- you are a researcher training or benchmarking dialogue or QA models across many datasets
- you want a unified API over dozens of standard NLP dialogue datasets
- you need built-in crowdsourcing or chat-service integration for human evaluation
- you want off-the-shelf pretrained conversational models to fine-tune

## When to avoid
- you need a production chatbot framework rather than a research platform
- you require Windows support, which is not officially supported
- you are building modern LLM applications where newer frameworks like Hugging Face Transformers or TRL may be better maintained
- you need lightweight inference-only deployment without the research tooling

## Facets
- artifact type: framework
- maturity: maintenance
- function: machine-learning, llm-training, chatbot, cli, data-science
- domain: machine-learning, chatbots, artificial-intelligence
- platform: python
- tags: dialogue-systems, nlp-research, datasets, pytorch, crowdsourcing, pretrained-models, natural-language-processing, research, linux, macos

## Member repositories
- facebookresearch/ParlAI (main) score 10

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:10:42.590323+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-29T17:18:27.789647+00:00, confidence not recorded.
  - readme: https://github.com/facebookresearch/ParlAI (fetched 2026-08-28T04:10:42.590323+00:00, sha 12ee329b4cca)
  - homepage: https://parl.ai (fetched 2026-08-29T08:17:32.980353+00:00, sha d87b86e15fa9)
  - site_page: https://parl.ai/about (fetched 2026-08-29T08:17:32.992268+00:00, sha c3ff88a32773)
  - site_page: https://parl.ai/docs/index.html (fetched 2026-08-29T08:17:32.990001+00:00, sha 04de7f594356)
- Data as of 2026-08-30T08:39:29.467469+00:00.