[meetup preview] openmldb + ONEFLOW: link feature engineering to model training to accelerate machine learning model development

2022 year 7 month 31 Japan （ Sunday ） Afternoon 14:00-16:30, Open source machine learning database OpenMLDB Issue 5 Meetup It will be launched in the form of online live broadcast .

Activity background

An open source learning database that provides production level real-time data and feature development full stack solutions OpenMLDB Invite static compilation and streaming parallel deep learning framework OneFlow Cooperation brings OpenMLDB Meetup Issue 5 .

This online sharing will lead you to an in-depth understanding of the... After the iteration and upgrading OpenMLDB and OneFlow, Analyze the architecture idea and hard core technology behind the product , And demonstrate how to pass OpenMLDB Easily calculate features , combination OneFlow Fluid training model , Accelerate machine learning model development , help AI Low threshold and low cost ！

Brief introduction

OpenMLDB PMC core member Lu Mian , From low cost 、 High performance online and offline consistency feature platform open source solution , Introduce the latest version of OpenMLDB And its performance improvement 、 Cost reduction 、 New features for increased flexibility .

OneFlow PMC core member Chengcheng , Will be with OneFlow —— Making large-scale distributed deep learning more convenient is a topic , Introduce ease of use and completeness to the audience to further improve 、 Model migration is more convenient and fast 、 Large models support more efficient OneFlow v0.8.0 And other highly available and extensible solutions and components .

OpenMLDB system architect Huang Wei , Will show how to pass OpenMLDB Calculate features and how to use OneFlow Load characteristic data and train , Show how to combine with practice OpenMLDB and OneFlow Easily realize feature calculation and model training .

OpenMLDB Platform architect Deng long , Will analyze in depth OpenMLDB Hard core technology behind architecture design , Guide you to understand OpenMLDB The internal implementation of millisecond real-time online feature computing engine .

See the poster for the specific schedule , The live information will be OpenMLDB Synchronization in technical communication group , Welcome friends who are not in the group to watch ~

Share and watch first

OpenMLDB： Consistent production level feature platform online and offline

【 Outline of the speech 】

1. Data and feature challenges of artificial intelligence engineering landing
2. OpenMLDB： Online and offline consistent production level feature computing platform
3. v0.5.0 Introduction to new features ： Performance improvements 、 Cost reduction 、 Increased flexibility

【 Audience benefits 】

1. Understand the pain points of data and characteristics encountered by enterprises in the process of artificial intelligence engineering implementation
2. Understand the low-cost and high-performance online and offline consistency feature platform open source solutions ：OpenMLDB
3. understand OpenMLDB The concept of online and offline consistent design architecture and enterprise level product features
4. understand OpenMLDB v0.5.0 characteristic , Performance improvements 、 Cost reduction 、 Increased flexibility

OneFlow —— Make large-scale distributed deep learning more convenient

【 Outline of the speech 】

1. OneFlow v0.8.0 New version interpretation
2. Global Tensor：OneFlow Distributed implementation and easy-to-use solutions for the community
3. Graph： Efficient and fast dynamic and static conversion scheme , Provide easy-to-use advanced distributed parallel optimization configuration
4. LiBai： be based on OneFlow Develop an efficient and scalable large-scale distributed pre training code base
5. OneEmbedding： Efficient and flexible extension components designed for large-scale recommendation systems

【 Audience benefits 】

1. understand OneFlow The distributed execution and easy-to-use solution provided
2. Understand the high-order optimization technology in large-scale distributed parallel training and the essence of distributed parallel
3. understand LiBai The characteristics of large-scale pre training model base and its advantages over other solutions in the industry
4. understand OneEmbedding Features and advantages in solving the problem of large-scale recommendation system

OpenMLDB+OneFlow, Teach you how to quickly link feature engineering to model training

【 Outline of the speech 】

1. Demonstrate the use of OpenMLDB Calculation characteristics and OneFlow Load characteristic data for training

【 Audience benefits 】

1. Learn how to use OpenMLDB Calculating characteristics
2. Learn how to use OneFlow Loading characteristic data and training
3. understand OpenMLDB and OneFlow How to combine

In depth disclosure OpenMLDB Millisecond level real-time online feature computing engine

【 Outline of the speech 】

1. OpenMLDB Online architecture
2. Design and implementation of storage engine
3. The principle of building a highly available database

【 Audience benefits 】

1. understand OpenMLDB Overall architecture design
2. Understand the implementation path of millisecond real-time online feature computing engine