# google-research/multilingual-t5

Repository: https://github.com/google-research/multilingual-t5
Canonical: https://ross.abutalabs.com/products/multilingual-t5
Language: Python
License: Apache-2.0
License Family: permissive
Archived: true
Last push: 2022-12-15T16:32:35+00:00

## Health v2 (maintenance only)
Score: 10/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 35, longevity 100
- inputs: {"age_days": 2141, "days_push": 1357, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, archived
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1294, forks 132 (observed 2026-08-28T04:04:16.380334+00:00)

## What it is
Code and resources for mT5, a massively multilingual text-to-text transformer pretrained on the mC4 corpus covering 101 languages. It reproduces the mT5 paper experiments and relies on the t5 library for training and fine-tuning.

## Use cases
- fine-tune a multilingual text-to-text transformer on my own NLP task
- reproduce the mT5 paper experiments
- train a seq2seq model that works across 101 languages
- get state-of-the-art zero-shot cross-lingual classification and QA
- download pretrained mT5 checkpoints of various sizes
- do cross-lingual question answering or NER with a pretrained model

## When to choose
- you need a pretrained multilingual encoder-decoder model covering many languages
- you want to reproduce or extend the mT5 paper using the t5 ecosystem
- you need cross-lingual transfer for classification, structured prediction, or QA

## When to avoid
- you only need English-language models
- you want a maintained library with frequent updates and broad community support
- you need inference-only usage without the t5 library dependency
- you want a lightweight model for resource-constrained environments

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, deep-learning, nlp, llm-training, transformers
- domain: machine-learning, deep-learning, large-language-models
- platform: python
- tags: mt5, multilingual, text-to-text, pretrained-models, t5, mc4, research-code, natural-language-processing

## Member repositories
- google-research/multilingual-t5 (main) score 10

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:04:16.380334+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 2026-08-30T04:54:04.793193+00:00, confidence not recorded.
  - readme: https://github.com/google-research/multilingual-t5 (fetched 2026-08-28T04:04:16.380334+00:00, sha dcfa59ad7074)
- Data as of 2026-08-30T08:39:29.467469+00:00.