<p align="center">
<img width="150px" src="https://user-images.githubusercontent.com/18649508/141604792-b256023d-c751-46d8-bab5-29a207d714ba.png"/>
</p>
<hr/>
<p align="center">
<a href="https://persiaml-tutorials.pages.dev" rel="nofollow"><img src="https://camo.githubusercontent.com/5f2d7c7e08b25fa4f95ce11be89982f25bb81bdef3e15f90b765aa91db371ff6/68747470733a2f2f696d672e736869656c64732e696f2f62616467652f7475746f7269616c732d70617373696e672d677265656e" alt="tutorials" data-canonical-src="https://img.shields.io/badge/tutorials-passing-green" style="max-width: 100%;"></a>
<a href="https://persiaml.pages.dev" rel="nofollow"><img src="https://camo.githubusercontent.com/bf535e4ed96252a7731419446fe108bb36ce4b91e4960630ceccf31558329193/68747470733a2f2f696d672e736869656c64732e696f2f62616467652f646f63756d656e746174696f6e2d70617373696e672d677265656e" alt="Documentation Status" data-canonical-src="https://img.shields.io/badge/documentation-passing-green" style="max-width: 100%;"></a>
<a href="https://badge.fury.io/py/persia" rel="nofollow"><img src="https://camo.githubusercontent.com/d6f9832aba04a67bbb6643eb81bd6be2173e12e51150553f30f16b629f40dc73/68747470733a2f2f62616467652e667572792e696f2f70792f7065727369612e737667" alt="PyPI version" data-canonical-src="https://badge.fury.io/py/persia.svg" style="max-width: 100%;"></a>
<a href="https://pypi.org/project/persia/"><img src="https://pepy.tech/badge/persia" alt="PyPI downloads"></a>
<a href="https://hub.docker.com/u/persiaml"><img alt="Docker Pulls" src="https://img.shields.io/docker/pulls/persiaml/persia-cuda-runtime"></a>
<a href="https://github.com/PersiaML/Persia/blob/main/LICENSE" rel="nofollow"><img src="https://img.shields.io/github/license/PersiaML/Persia" alt="license" style="max-width: 100%;"></a>
</p>
<div align="center">
<a href="https://github.com/PersiaML/Persia/stargazers"><img src="https://reporoster.com/stars/PersiaML/Persia" /><a/>
</div>
*WARNING: THIS PROJECT IS CURRENTLY IN MAINTENANCE MODE, DUE TO COMPANY REORGANIZATION.*
**PERSIA** (**P**arallel r**E**commendation t**R**aining **S**ystem with hybr**I**d **A**cceleration) is developed by [AI platform@Kuaishou Technology](https://www.kuaishou.com/en), collaborating with ETH. It is a PyTorch-based (the first public one to our best knowledge) system for training large scale deep learning recommendation models on commodity hardwares. It is capable of training recommendation models with up to 100 trillion parameters. To the best of our knowledge, this is the largest model size in recommendation systems so far. Empirical study on public datasets indicate PERSIA's significant advantage over several other existing training systems in recommendation [1]. Its efficiency and robustness have also been validated by multiple applications with 100 million level DAU at Kuaishou.
*Disclaimer: The program is usable and has served several important businesses. However, the official English documentation and tutorials are still under heavy construction and they are a bit raw now. We encourage adventurers to try out PERSIA and contribute!*
## News
* [Training Deep Learning-based recommender models of 100 trillion parameters over Google Cloud](https://archive.ph/8ay0C)
* [突破百万亿参数规模,追求极致的效率和性价比:华人团队开源首个异构并行推荐系统训练框架 PERSIA](https://archive.ph/Mixk0) (In Chinese. Title: Breaking through the trillion parameter scale in pursuit of ultimate efficiency and cost effectiveness: Chinese team open source PERSIA, the first heterogeneous parallel recommendation system)
* [参数量卷到一百万亿!华人团队开源史上最大的推荐训练系统 PERSIA](https://archive.md/FbocB) (In Chinese. Title: PERSIA, the Largest Recommended Training System in the History of Open Source by Far)
* AI Engines in the "Short-video" Era: Eating 100 Trillion Parameters, Invited talk, Facebook, 2021.
* 单机训练速度提升 640 倍!独家解读快手商业广告模型 GPU 训练平台 PERSIA (In Chinese. Title: 640x Faster GPU Based Learning System for Ad Recommendation)
* [[AI Front]](https://archive.is/2ii2L) [[中国日报]](https://archive.is/N8fK2) [[InfoQ]](https://archive.is/JESDU) [[CSDN]](https://archive.is/tpvkN) [[Tencent Cloud News]](https://archive.is/kLuaT) [[AcFun]](https://archive.md/vuPmb)
* 创新、平衡与大格局:快手商业化的慢与快 (In Chinese. Title: Innovation, Balance, and Big Picture: The Speed of Kwai Commercialization)
* [[TechSir]](https://archive.is/EOQ18) [[China Daily]](https://archive.is/L2VJE) [[Sohu]](https://archive.is/aY66U)
## Links
* [GitHub Repository](https://github.com/PersiaML/PERSIA)
* [Tutorials](https://persiaml-tutorials.pages.dev/)
* [API documentation](https://persiaml.pages.dev/) (Under Construction)
## Discussion
Feel free to join our [Telegram Group](https://t.me/joinchat/fLlD66VX8PQxMmJh) for discussion!
## References
1. Xiangru Lian, Binhang Yuan, Xuefeng Zhu, Yulong Wang, Yongjun He, Honghuan Wu, Lei Sun, Haodong Lyu, Chengjun Liu, Xing Dong, Yiqiao Liao, Mingnan Luo, Congfei Zhang, Jingru Xie, Haonan Li, Lei Chen, Renjie Huang, Jianying Lin, Chengchun Shu, Xuezhong Qiu, Zhishan Liu, Dongying Kong, Lei Yuan, Hai Yu, Sen Yang, Ce Zhang, & Ji Liu. (2021). [Persia: A Hybrid System Scaling Deep Learning Based Recommenders up to 100 Trillion Parameters.](https://arxiv.org/abs/2111.05897)
2. Ji Liu & Ce Zhang. (2021). [Distributed Learning Systems with First-order Methods](https://arxiv.org/pdf/2104.05245).
## License
This source code is licensed under the MIT license found in the LICENSE file in the root directory of this source tree.
没有合适的资源?快使用搜索试试~ 我知道了~
基于PyTorch的深度学习推荐模型训练高性能分布式框架。_Rust_Python_下载.zip
共171个文件
rs:59个
py:34个
toml:21个
1.该资源内容由用户上传,如若侵权请联系客服进行举报
2.虚拟产品一经售出概不退款(资源遇到问题,请及时私信上传者)
2.虚拟产品一经售出概不退款(资源遇到问题,请及时私信上传者)
版权申诉
0 下载量 66 浏览量
2023-04-23
09:50:41
上传
评论
收藏 260KB ZIP 举报
温馨提示
基于PyTorch的深度学习推荐模型训练高性能分布式框架。_Rust_Python_下载.zip
资源推荐
资源详情
资源评论
收起资源包目录
基于PyTorch的深度学习推荐模型训练高性能分布式框架。_Rust_Python_下载.zip (171个子文件)
CITATION.cff 2KB
setup.cfg 234B
Dockerfile 4KB
.ci.env 564B
.docker.env 229B
.honcho.env 52B
.gitignore 2KB
.gitmodules 109B
versions.html 806B
redirect-index.html 267B
perisa-training.json 23KB
LICENSE 1KB
Cargo.lock 82KB
Cargo.lock 74KB
Makefile 895B
Makefile 708B
README.md 6KB
README.md 845B
README.md 607B
Procfile 565B
ts_config.properties 43B
inference.proto 828B
ctx.py 41KB
distributed.py 15KB
helper.py 15KB
data.py 13KB
data.py 10KB
launcher.py 7KB
train.py 5KB
test_ctx.py 5KB
conf.py 4KB
env.py 4KB
logger.py 4KB
data_generator.py 3KB
optim.py 3KB
setup.py 3KB
data_preprocess.py 2KB
utils.py 2KB
serve_client.py 2KB
test_data.py 2KB
model.py 1KB
prelude.py 1KB
setup.py 1KB
serve_handler.py 1KB
__init__.py 956B
k8s_utils.py 945B
data_loader.py 812B
test_env.py 810B
test_service.py 582B
service.py 313B
__init__.py 284B
utils.py 116B
__init__.py 84B
version.py 44B
version.py 26B
__init__.py 0B
mod.rs 64KB
array_linked_list.rs 35KB
forward.rs 33KB
mod.rs 24KB
crd.rs 19KB
lib.rs 19KB
nats.rs 15KB
lib.rs 15KB
lib.rs 15KB
backward.rs 13KB
lib.rs 13KB
optim.rs 12KB
lib.rs 11KB
serialize_inf_request.rs 10KB
lib.rs 10KB
tensor.rs 10KB
rpc.rs 9KB
lib.rs 9KB
data.rs 8KB
lib.rs 8KB
e2e.rs 8KB
lib.rs 7KB
lib.rs 6KB
server.rs 6KB
lib.rs 5KB
message_queue.rs 5KB
emb_entry.rs 5KB
lib.rs 5KB
persia-embedding-worker.rs 5KB
eviction_map.rs 4KB
persia-embedding-parameter-server.rs 4KB
monitor.rs 4KB
utils.rs 4KB
operator.rs 3KB
lib.rs 3KB
lib.rs 3KB
mod.rs 3KB
dlpack.rs 2KB
resource_pool.rs 2KB
metrics.rs 2KB
service.rs 2KB
optim.rs 2KB
pinned_memory_pool.rs 2KB
cuda_memory_pool.rs 2KB
共 171 条
- 1
- 2
资源评论
快撑死的鱼
- 粉丝: 1w+
- 资源: 9154
上传资源 快速赚钱
- 我的内容管理 展开
- 我的资源 快来上传第一个资源
- 我的收益 登录查看自己的收益
- 我的积分 登录查看自己的积分
- 我的C币 登录后查看C币余额
- 我的收藏
- 我的下载
- 下载帮助
安全验证
文档复制为VIP权益,开通VIP直接复制
信息提交成功