# microsoft/MASS

MASS: Masked Sequence to Sequence Pre-training for Language Generation

Repository: https://github.com/microsoft/MASS
Canonical: https://ross.abutalabs.com/products/mass
Homepage: https://arxiv.org/pdf/1905.02450.pdf
Language: Python
License: NOASSERTION
License Family: other
Archived: true
Last push: 2022-11-28T19:09:50+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": 2655, "days_push": 1374, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: no_releases, archived, no_license
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 1115, forks 203 (observed 2026-08-28T04:03:38.324705+00:00)

## What it is
MASS is Microsoft's PyTorch implementation of Masked Sequence to Sequence Pre-training for language generation tasks. It provides pre-trained models and training code for sequence-to-sequence tasks like machine translation, summarization, and conversational response generation.

## Use cases
- pre-train a sequence-to-sequence model for text generation
- fine-tune a pre-trained model for machine translation
- run unsupervised machine translation experiments
- generate abstractive text summaries
- reproduce the MASS paper results on WMT benchmarks

## When to choose
- you need a seq2seq pre-training baseline for NLP research
- you want to reproduce or extend the MASS paper
- you need pre-trained encoder-decoder models for translation or summarization

## When to avoid
- you need a modern, actively maintained LLM training framework
- you want production-ready translation services
- you need the latest transformer architectures beyond the MASS method

## Facets
- artifact type: library
- maturity: maintenance
- function: machine-learning, llm-training, nlp
- domain: machine-learning, deep-learning
- platform: python
- tags: pre-training, seq2seq, masked-language-model, machine-translation, text-summarization, research-code, natural-language-processing, linux, gpu

## Member repositories
- microsoft/MASS (main) score 10

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:03:38.324705+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 2026-08-30T06:42:11.341369+00:00, confidence not recorded.
  - readme: https://github.com/microsoft/MASS (fetched 2026-08-28T04:03:38.324705+00:00, sha 332d95ac813e)
- Data as of 2026-08-30T08:39:29.467469+00:00.