# NVIDIA/OpenSeq2Seq

Toolkit for efficient experimentation with Speech Recognition, Text2Speech and NLP

Repository: https://github.com/NVIDIA/OpenSeq2Seq
Canonical: https://ross.abutalabs.com/products/openseq2seq
Homepage: https://nvidia.github.io/OpenSeq2Seq
Language: Python
License: Apache-2.0
License Family: permissive
Topics: neural-machine-translation, multi-gpu, deep-learning, sequence-to-sequence, seq2seq, multi-node, speech-recognition, speech-to-text, mixed-precision, float16, speech-synthesis, text-to-speech, language-model, tensorflow
Archived: true
Last push: 2021-05-11T15:50:05+00:00

## Health v2 (maintenance only)
Score: 10/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 8, longevity 100
- inputs: {"age_days": 3281, "days_push": 1940, "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 1558, forks 374 (observed 2026-08-28T04:05:03.487662+00:00)

## What it is
OpenSeq2Seq is a TensorFlow-based toolkit for building and training sequence-to-sequence models for neural machine translation, speech recognition, speech synthesis, and language modeling. It emphasizes efficient experimentation through distributed multi-GPU/multi-node training and mixed-precision (FP16) support on NVIDIA GPUs.

## Use cases
- train a neural machine translation model with multi-GPU distributed training
- build an automatic speech recognition system with seq2seq models
- experiment with text-to-speech / speech synthesis models
- train language models with mixed-precision FP16 on Volta/Turing GPUs
- run sentiment analysis NLP experiments with encoder-decoder models
- scale seq2seq training across multiple nodes with Horovod

## When to choose
- you need a research toolkit for seq2seq models in TensorFlow 1.x
- you want mixed-precision and multi-GPU/multi-node training for NLP or speech models
- you are reproducing experiments from the OpenSeq2Seq paper

## When to avoid
- you use TensorFlow 2.x or PyTorch — the project is unmaintained and tied to TF 1.x
- you need production-ready speech or translation systems rather than research experimentation
- you want actively maintained tooling — consider NeMo or Fairseq instead

## Facets
- artifact type: library
- maturity: abandoned
- function: machine-learning, deep-learning, speech-recognition, tts, nlp, llm-training
- domain: deep-learning, speech-processing, machine-learning
- platform: python
- tags: seq2seq, tensorflow, mixed-precision, multi-gpu, distributed-training, neural-machine-translation, speech-synthesis, language-modeling, horovod, natural-language-processing, linux, gpu

## Member repositories
- NVIDIA/OpenSeq2Seq (main) score 10

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:05:03.487662+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 2026-08-30T04:29:52.631276+00:00, confidence not recorded.
  - readme: https://github.com/NVIDIA/OpenSeq2Seq (fetched 2026-08-28T04:05:03.487662+00:00, sha 65c11108f872)
  - homepage: https://nvidia.github.io/OpenSeq2Seq (fetched 2026-08-29T11:29:48.550786+00:00, sha 7c643ca2e1c7)
- Data as of 2026-08-30T08:39:29.467469+00:00.