Author

Kuaishou Tech

Official Kuaishou tech account, providing real-time updates on the latest Kuaishou technology practices.

225

Articles

Likes

574

Views

Comments

Latest from Kuaishou Tech

100 recent articles max

Kuaishou Tech

Nov 5, 2025 · Artificial Intelligence

How HiPO Gives LLMs a Smart Thinking Switch to Cut Costs and Boost Accuracy

This article explains the overthinking problem of large language models, introduces the HiPO framework with hybrid data cold‑start and reinforcement‑learning reward mechanisms that let models decide when to think deeply or answer directly, and shows experimental results demonstrating significant efficiency gains and accuracy improvements across multiple benchmarks.

Hybrid Policy OptimizationLLMadaptive inference

0 likes · 13 min read

How HiPO Gives LLMs a Smart Thinking Switch to Cut Costs and Boost Accuracy

Kuaishou Tech

Oct 30, 2025 · Artificial Intelligence

How EMER Revolutionizes Short‑Video Ranking with End‑to‑End Multi‑Objective Learning

This article details the EMER framework—a Transformer‑based, end‑to‑end multi‑objective ranking system that replaces handcrafted formulas with a learnable AI model, introduces relative‑satisfaction signals and dynamic loss weighting, and demonstrates significant offline and online performance gains in Kuaishou's short‑video recommendation pipeline.

AImulti-objective learningonline experiments

0 likes · 16 min read

How EMER Revolutionizes Short‑Video Ranking with End‑to‑End Multi‑Objective Learning

Kuaishou Tech

Oct 11, 2025 · Artificial Intelligence

How KAT-Dev-72B-Exp Sets a New Record in Large‑Scale RL for Code Generation

The KAT‑Dev‑72B‑Exp model, an experimental reinforcement‑learning version of KAT‑Coder, achieves a 74.6% performance boost on the SWE‑Bench Verified benchmark, introduces Trie Packing and entropy‑aware advantage scaling, and showcases a decoupled training architecture that dramatically speeds up large‑scale agentic RL training.

AICode Generationagentic training

0 likes · 9 min read

How KAT-Dev-72B-Exp Sets a New Record in Large‑Scale RL for Code Generation

Kuaishou Tech

Sep 23, 2025 · Artificial Intelligence

How Generative Reinforcement Learning is Revolutionizing Real-Time Bidding

This article explains the core challenges of real‑time bidding, traces the evolution from PID to MPC to reinforcement learning, and details how generative reinforcement‑learning techniques such as GAVE and CBD combine diffusion models, value‑guided exploration, and score‑based return‑to‑go to dramatically improve ad‑bid efficiency and revenue.

CBDGAVEadvertising algorithms

0 likes · 15 min read

How Generative Reinforcement Learning is Revolutionizing Real-Time Bidding

Kuaishou Tech

Sep 17, 2025 · Artificial Intelligence

How MIDAS Achieves Real‑Time Multimodal Digital‑Human Video Generation

The MIDAS framework introduced by the Kling Team combines autoregressive video generation with a lightweight diffusion denoising head to deliver real‑time, high‑quality digital‑human synthesis under multimodal control, achieving sub‑500 ms latency, 64× compression, and robust performance across multilingual dialogue, singing, and interactive world modeling tasks.

AIReal-time Videoautoregressive

0 likes · 6 min read

How MIDAS Achieves Real‑Time Multimodal Digital‑Human Video Generation

Kuaishou Tech

Sep 16, 2025 · Artificial Intelligence

How Kling-Avatar Generates Long, Emotionally Rich Digital Human Videos with Multimodal LLMs

Kuaishou's Kling-Avatar leverages a multimodal large‑language‑model‑driven two‑stage generation framework to produce minute‑long digital‑human videos that synchronize lip movements, facial expressions, and body gestures with audio, achieving high visual quality, identity consistency, and controllable storytelling across diverse scenarios.

AI Avatardigital humanlong video synthesis

0 likes · 9 min read

Kuaishou Tech

Sep 5, 2025 · Artificial Intelligence

How Keye‑VL‑1.5‑8B Sets New Benchmarks in Multimodal AI

Fast‑search platform Kwai has open‑sourced the 8‑billion‑parameter multimodal LLM Keye‑VL‑1.5, which introduces a slow‑fast frame encoding, a progressive four‑stage pre‑training pipeline, and an automated data construction workflow, achieving state‑of‑the‑art results on video and vision‑language benchmarks and surpassing many closed‑source models.

benchmark performancelarge language modelmultimodal AI

0 likes · 12 min read

How Keye‑VL‑1.5‑8B Sets New Benchmarks in Multimodal AI

Kuaishou Tech

Sep 4, 2025 · Fundamentals

How LingXi Achieves User‑Level QoE Optimization in Large‑Scale Adaptive Video Streaming

The paper “Towards User‑level QoE: Large‑scale Personalized Optimization of Adaptive Video Streaming” introduces LingXi, the first production‑grade system that deploys per‑user Bayesian optimization and Monte‑Carlo simulation to reduce video stalls and boost both QoE and QoS across millions of viewers, with especially strong gains for low‑bandwidth users.

Monte Carlo SimulationQoE optimizationadaptive streaming

0 likes · 15 min read

How LingXi Achieves User‑Level QoE Optimization in Large‑Scale Adaptive Video Streaming

Kuaishou Tech

Aug 28, 2025 · Big Data

Auron Joins Apache Incubator: High‑Performance Vectorized Engine Accelerates Big Data Workloads

The Auron project, originally the Blaze engine from Kuaishou, has entered the Apache Software Foundation incubator, offering a Rust‑based native vectorized execution engine that integrates with Spark, delivers over two‑fold performance gains on TPC‑DS benchmarks, and is supported by a growing open‑source community.

Apache IncubatorAuronBig Data

0 likes · 6 min read

Auron Joins Apache Incubator: High‑Performance Vectorized Engine Accelerates Big Data Workloads

Kuaishou Tech

Aug 25, 2025 · Artificial Intelligence

How Context-as-Memory Enables Scene‑Consistent Long Video Generation

This article introduces the Context-as-Memory approach, which treats previously generated video frames as memory to achieve scene‑consistent interactive long video generation, and details a camera‑trajectory‑based memory retrieval mechanism that dramatically improves efficiency and performance over existing state‑of‑the‑art methods.

AIMemory retrievalcontext memory

0 likes · 7 min read

How Context-as-Memory Enables Scene‑Consistent Long Video Generation