real-time training

DataFunSummit

Jul 25, 2023 · Artificial Intelligence

Real‑Time Deep Learning Training with PAI‑ODL: Architecture, Pipeline, and Key Technologies

This article introduces PAI‑ODL, a real‑time deep‑learning training platform that supports online model updates for search, advertising, and recommendation scenarios, detailing its pipeline modules, system architecture, large‑scale sparse model techniques, incremental model export, embedding store design, and performance optimizations that together enable low‑latency, high‑throughput serving.

PAI ODLembedding storelarge-scale sparse models

0 likes · 19 min read

Real‑Time Deep Learning Training with PAI‑ODL: Architecture, Pipeline, and Key Technologies

Xiaohongshu Tech REDtech

Mar 21, 2023 · Artificial Intelligence

From Daily to Minute-Level Updates: Real-Time Recommendation System Enhancements at Xiaohongshu

Xiaohongshu transformed its recommendation pipeline from daily to minute‑level updates by redesigning recall, ranking and feature‑joining components, deploying a base‑plus‑incremental training scheme, migrating Spark to Flink, rewriting services in C++, and optimizing RocksDB, which yielded over 10% longer dwell time, 15% more interactions and roughly 50% higher new‑note efficiency.

Vector Searchlarge-scale systemsmachine learning

0 likes · 20 min read

From Daily to Minute-Level Updates: Real-Time Recommendation System Enhancements at Xiaohongshu

Architecture Digest

Mar 2, 2020 · Artificial Intelligence

Recommendation System Architecture and Practices at Toutiao

This article provides a comprehensive overview of Toutiao's recommendation system, covering its three-dimensional modeling of content, user, and environment features, various algorithmic approaches, feature extraction, real‑time training pipelines, recall strategies, user‑tag engineering, evaluation methods, and content‑safety measures.

A/B testingFeature Engineeringcontent safety

0 likes · 18 min read

Recommendation System Architecture and Practices at Toutiao

iQIYI Technical Product Team

Oct 31, 2019 · Artificial Intelligence

Online Learning for Large‑Scale DNN Ranking Models in iQIYI Feed Recommendation

iQIYI’s feed recommendation system adopts an online‑learning framework that continuously trains a massive Wide‑and‑Deep DNN on billions of streaming samples, handling dynamic user interests, OOV embeddings, delayed labels, and non‑convex optimization, enabling hourly model refreshes and delivering up to 3.8 % higher consumption versus offline baselines.

DNNdeep learningiQIYI

0 likes · 17 min read