{"id":15127475,"url":"https://github.com/nvidia/openseq2seq","last_synced_at":"2025-09-28T14:31:18.021Z","repository":{"id":45022407,"uuid":"102903563","full_name":"NVIDIA/OpenSeq2Seq","owner":"NVIDIA","description":"Toolkit for efficient experimentation with Speech Recognition, Text2Speech and NLP","archived":true,"fork":false,"pushed_at":"2021-05-11T15:50:05.000Z","size":60230,"stargazers_count":1550,"open_issues_count":87,"forks_count":369,"subscribers_count":93,"default_branch":"master","last_synced_at":"2024-11-27T03:34:51.642Z","etag":null,"topics":["deep-learning","float16","language-model","mixed-precision","multi-gpu","multi-node","neural-machine-translation","seq2seq","sequence-to-sequence","speech-recognition","speech-synthesis","speech-to-text","tensorflow","text-to-speech"],"latest_commit_sha":null,"homepage":"https://nvidia.github.io/OpenSeq2Seq","language":"Python","has_issues":true,"has_wiki":null,"has_pages":null,"mirror_url":null,"source_name":null,"license":"apache-2.0","status":null,"scm":"git","pull_requests_enabled":true,"icon_url":"https://github.com/NVIDIA.png","metadata":{"files":{"readme":"README.md","changelog":null,"contributing":"CONTRIBUTING.md","funding":null,"license":"LICENSE","code_of_conduct":null,"threat_model":null,"audit":null,"citation":null,"codeowners":null,"security":null,"support":null}},"created_at":"2017-09-08T20:53:07.000Z","updated_at":"2024-11-18T19:13:18.000Z","dependencies_parsed_at":"2022-08-12T11:40:48.811Z","dependency_job_id":null,"html_url":"https://github.com/NVIDIA/OpenSeq2Seq","commit_stats":null,"previous_names":[],"tags_count":8,"template":false,"template_full_name":null,"repository_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/NVIDIA%2FOpenSeq2Seq","tags_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/NVIDIA%2FOpenSeq2Seq/tags","releases_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/NVIDIA%2FOpenSeq2Seq/releases","manifests_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories/NVIDIA%2FOpenSeq2Seq/manifests","owner_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners/NVIDIA","download_url":"https://codeload.github.com/NVIDIA/OpenSeq2Seq/tar.gz/refs/heads/master","host":{"name":"GitHub","url":"https://github.com","kind":"github","repositories_count":234525631,"owners_count":18846937,"icon_url":"https://github.com/github.png","version":null,"created_at":"2022-05-30T11:31:42.601Z","updated_at":"2022-07-04T15:15:14.044Z","host_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub","repositories_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repositories","repository_names_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/repository_names","owners_url":"https://repos.ecosyste.ms/api/v1/hosts/GitHub/owners"}},"keywords":["deep-learning","float16","language-model","mixed-precision","multi-gpu","multi-node","neural-machine-translation","seq2seq","sequence-to-sequence","speech-recognition","speech-synthesis","speech-to-text","tensorflow","text-to-speech"],"created_at":"2024-09-26T02:03:27.984Z","updated_at":"2025-09-28T14:31:09.405Z","avatar_url":"https://github.com/NVIDIA.png","language":"Python","funding_links":[],"categories":[],"sub_categories":[],"readme":"[![License](https://img.shields.io/badge/License-Apache%202.0-brightgreen.svg)](https://opensource.org/licenses/Apache-2.0)\n[![Documentation](https://img.shields.io/badge/documentation-github.io-blue.svg)](https://nvidia.github.io/OpenSeq2Seq/html/index.html)\n\u003cdiv align=\"center\"\u003e\n  \u003cimg src=\"./docs/logo-shadow.png\" alt=\"OpenSeq2Seq\" width=\"250px\"\u003e\n  \u003cbr\u003e\n\u003c/div\u003e\n\n# OpenSeq2Seq: toolkit for distributed and mixed precision training of sequence-to-sequence models\n\nOpenSeq2Seq main goal is to allow researchers to most effectively explore various\nsequence-to-sequence models. The efficiency is achieved by fully supporting\ndistributed and mixed-precision training.\nOpenSeq2Seq is built using TensorFlow and provides all the necessary\nbuilding blocks for training encoder-decoder models for neural machine translation, automatic speech recognition, speech synthesis, and language modeling.\n\n## Documentation and installation instructions \nhttps://nvidia.github.io/OpenSeq2Seq/\n\n## Features\n1. Models for:\n   1. Neural Machine Translation\n   2. Automatic Speech Recognition\n   3. Speech Synthesis\n   4. Language Modeling\n   5. NLP tasks (sentiment analysis)\n2. Data-parallel distributed training\n   1. Multi-GPU\n   2. Multi-node\n3. Mixed precision training for NVIDIA Volta/Turing GPUs\n\n## Software Requirements\n1. Python \u003e= 3.5\n2. TensorFlow \u003e= 1.10\n3. CUDA \u003e= 9.0, cuDNN \u003e= 7.0 \n4. Horovod \u003e= 0.13 (using Horovod is not required, but is highly recommended for multi-GPU setup)\n\n## Acknowledgments\nSpeech-to-text workflow uses some parts of [Mozilla DeepSpeech](https://github.com/Mozilla/DeepSpeech) project.\n\nBeam search decoder with language model re-scoring implementation (in `decoders`) is based on [Baidu DeepSpeech](https://github.com/PaddlePaddle/DeepSpeech).\n\nText-to-text workflow uses some functions from [Tensor2Tensor](https://github.com/tensorflow/tensor2tensor) and [Neural Machine Translation (seq2seq) Tutorial](https://github.com/tensorflow/nmt).\n\n## Disclaimer\nThis is a research project, not an official NVIDIA product.\n\n## Related resources\n* [Tensor2Tensor](https://github.com/tensorflow/tensor2tensor)\n* [Neural Machine Translation (seq2seq) Tutorial](https://github.com/tensorflow/nmt)\n* [OpenNMT](http://opennmt.net/)\n* [Neural Monkey](https://github.com/ufal/neuralmonkey)\n* [Sockeye](https://github.com/awslabs/sockeye)\n* [TF-seq2seq](https://github.com/google/seq2seq)\n* [Moses](http://www.statmt.org/moses/)\n\n## Paper\nIf you use OpenSeq2Seq, please cite [this paper](https://arxiv.org/abs/1805.10387)\n```\n@misc{openseq2seq,\n    title={Mixed-Precision Training for NLP and Speech Recognition with OpenSeq2Seq},\n    author={Oleksii Kuchaiev and Boris Ginsburg and Igor Gitman and Vitaly Lavrukhin and Jason Li and Huyen Nguyen and Carl Case and Paulius Micikevicius},\n    year={2018},\n    eprint={1805.10387},\n    archivePrefix={arXiv},\n    primaryClass={cs.CL}\n}\n```\n","project_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fnvidia%2Fopenseq2seq","html_url":"https://awesome.ecosyste.ms/projects/github.com%2Fnvidia%2Fopenseq2seq","lists_url":"https://awesome.ecosyste.ms/api/v1/projects/github.com%2Fnvidia%2Fopenseq2seq/lists"}