Merlin ： GPU 加速的推荐系统框架.pdf
50墨值下载
王泽寰 (Joey Wang), Dec 17th 2020
Merlin: A GPU Accelerated
Recommendation Framework
2
Industrial Recommendation Challenges
Iterating over feature
combinations takes
longer than training.
Tabular data scales
poorly using the
common deep learning
method of item by
item.
Embedding tables
don’t fit easily on GPU
and are hard to scale.
The path to production
from research is
complex and requires
significant engineering
effort.
Feature
Engineering
Dataloading Scaling Training Deployment
of 40
50墨值下载
【版权声明】本文为墨天轮用户原创内容，转载时必须标注文档的来源（墨天轮），文档链接，文档作者等基本信息，否则作者和墨天轮有权追究责任。如果您发现墨天轮中有涉嫌抄袭或者侵权的内容，欢迎发送邮件至：contact@modb.pro进行举报，并提供相关证据，一经查实，墨天轮将立刻删除相关内容。
下载排行榜
文档被以下合辑收录
评论