# nyu-mll/jiant

jiant is an nlp toolkit

Repository: https://github.com/nyu-mll/jiant
Canonical: https://ross.abutalabs.com/products/jiant
Homepage: https://jiant.info
Language: Python
License: MIT
License Family: permissive
Topics: nlp, sentence-representation, bert, multitask-learning, transformers, transfer-learning
Last push: 2023-07-06T22:00:38+00:00

## Health v2 (maintenance only)
Score: 23/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 8, longevity 100
- inputs: {"age_days": 2998, "days_push": 1154, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1673, forks 296 (observed 2026-08-28T04:05:20.303440+00:00)

## What it is
jiant is a PyTorch-based NLP research toolkit for multitask and transfer learning, supporting 50+ natural language understanding tasks and benchmarks like GLUE, SuperGLUE, and XTREME. It is configuration-file driven and integrates with Hugging Face's transformers and datasets libraries.

## Use cases
- run multitask learning experiments across NLU tasks
- apply transfer learning with STILTs to NLP tasks
- evaluate models on GLUE, SuperGLUE, and XTREME benchmarks
- fine-tune BERT and other transformer models on NLP tasks
- train sentence representation models
- benchmark pretrained language models across many tasks

## When to choose
- conducting NLP research on multitask or transfer learning with PyTorch
- reproducing GLUE/SuperGLUE/XTREME benchmark experiments
- needing a configuration-driven harness for many NLU tasks

## When to avoid
- you need active maintenance, new models, or updated library support (project unmaintained since October 2021)
- building production NLP applications rather than research experiments
- you want the latest transformer architectures or task support

## Facets
- artifact type: library
- maturity: abandoned
- function: machine-learning, nlp, llm-training
- domain: machine-learning, deep-learning
- platform: python
- tags: multitask-learning, transfer-learning, pytorch, transformers, bert, glue-benchmark, superglue, xtreme, research-toolkit, sentence-representations, natural-language-processing, linux, macos

## Member repositories
- nyu-mll/jiant (main) score 23

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:20.303440+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 2026-08-30T03:42:35.764260+00:00, confidence not recorded.
  - readme: https://github.com/nyu-mll/jiant (fetched 2026-08-28T04:05:20.303440+00:00, sha 651d7b711b84)
  - registry_pypi: https://pypi.org/pypi/jiant/json (fetched 2026-08-29T11:16:01.359949+00:00, sha 6f1d4a449667)
- Data as of 2026-08-30T08:39:29.467469+00:00.