Author

Kuaishou Tech

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

246

Articles

Likes

2.0k

Views

Comments

Latest from Kuaishou Tech

100 recent articles max

Kuaishou Tech

Jul 22, 2025 · Artificial Intelligence

How Orthus Achieves Lossless Multimodal Generation with a Unified Autoregressive Transformer

Orthus, a new unified multimodal model presented at ICML 2025, leverages an autoregressive Transformer backbone with separate language and diffusion heads to enable lossless image‑text interleaved generation, outperforming existing models on both understanding and generation benchmarks while remaining computationally efficient.

AI researchautoregressive transformerdiffusion models

0 likes · 11 min read

How Orthus Achieves Lossless Multimodal Generation with a Unified Autoregressive Transformer

Kuaishou Tech

Jul 21, 2025 · Artificial Intelligence

Can AI Models Think on Demand? Inside KAT‑V1 AutoThink’s Dynamic Reasoning

The article introduces KAT‑V1 AutoThink, a dual‑mode large language model that automatically switches between thinking and non‑thinking modes based on problem difficulty, details its novel training paradigm, reinforcement‑learning enhancements, performance benchmarks against leading open‑source models, and provides open‑source resources for further research.

Knowledge Distillationauto-thinklarge language model

0 likes · 14 min read

Can AI Models Think on Demand? Inside KAT‑V1 AutoThink’s Dynamic Reasoning

Kuaishou Tech

Jul 17, 2025 · Artificial Intelligence

How DHPS Boosted Online Inference Throughput by 270% with RDMA

This article details the design and evolution of DHPS, Kuaishou's load‑balanced, RDMA‑based high‑performance service architecture, explaining its network, storage, and traffic‑scheduling innovations that deliver over 270% query‑throughput improvement, lower latency, reduced CPU usage, and near‑five‑nine availability for large‑scale AI inference workloads.

RDMAdistributed systemshigh-performance computing

0 likes · 17 min read

How DHPS Boosted Online Inference Throughput by 270% with RDMA

Kuaishou Tech

Jul 16, 2025 · Artificial Intelligence

How KuaiMM Conversation Revolutionizes Multimodal Dialogue on Short‑Video Platforms

The KuaiMM Conversation project introduces a multimodal large‑model‑driven dialogue system for Kuaishou, featuring the world‑first short‑video mixed‑dialogue dataset, a Chain‑of‑Thought interaction framework, and large‑scale industrial deployments that dramatically improve live‑stream comments and intelligent customer service.

Kuaishouchain-of-thoughtconversation AI

0 likes · 11 min read

How KuaiMM Conversation Revolutionizes Multimodal Dialogue on Short‑Video Platforms

Kuaishou Tech

Jul 11, 2025 · Artificial Intelligence

How VARSR Redefines Image Super‑Resolution with Autoregressive Modeling

The VARSR algorithm introduces autoregressive modeling to image super‑resolution, leveraging prefix tokens, scale‑aligned rotary positional encodings, quantization error correction, and image‑quality‑guided diffusion to achieve faster inference and superior visual fidelity, as demonstrated by extensive ICML‑2025 experiments.

ICML 2025VARSRautoregressive modeling

0 likes · 11 min read

How VARSR Redefines Image Super‑Resolution with Autoregressive Modeling

Kuaishou Tech

Jul 10, 2025 · Artificial Intelligence

How MODA’s Modular Duplex Attention Solves Multimodal Attention Imbalance and Boosts Emotion Understanding

The paper introduces MODA, a modular duplex attention multimodal model that addresses severe cross‑modal attention imbalance in existing large multimodal models, proposes a novel attention paradigm and masking scheme, and demonstrates significant performance gains across 21 benchmarks in perception, cognition, and emotion tasks, earning a Spotlight paper at ICML 2025.

Emotion RecognitionLarge Language ModelsMoDA

0 likes · 13 min read

How MODA’s Modular Duplex Attention Solves Multimodal Attention Imbalance and Boosts Emotion Understanding

Kuaishou Tech

Jul 9, 2025 · Artificial Intelligence

How ResULIC Achieves Ultra‑Low‑Rate Image Compression with Semantic Residual Coding and Diffusion

The paper introduces ResULIC, a residual‑guided ultra‑low‑bitrate image compression framework that combines semantic residual coding, a compression‑aware diffusion model, and perceptual fidelity optimization to dramatically improve visual quality and outperform prior diffusion‑based methods on standard benchmarks.

Machine LearningResULICdiffusion model

0 likes · 12 min read

How ResULIC Achieves Ultra‑Low‑Rate Image Compression with Semantic Residual Coding and Diffusion

Kuaishou Tech

Jul 7, 2025 · Artificial Intelligence

8 Kuaishou Papers Spotlighted at ICML 2025: Multimodal AI, Causal Inference and More

Kuaishou has had eight cutting‑edge papers accepted at the International Conference on Machine Learning 2025, covering breakthroughs in multimodal emotion modeling, monotonic probability learning, causal effect generalization, cascade ranking, multimodal LLM alignment, ultra‑low‑rate image compression, and visual autoregressive super‑resolution, with links to each work and accompanying code repositories.

AIMachine Learningcausal inference

0 likes · 13 min read

8 Kuaishou Papers Spotlighted at ICML 2025: Multimodal AI, Causal Inference and More

Kuaishou Tech

Jul 2, 2025 · Artificial Intelligence

How EvoSearch Supercharges Image and Video Generation with Test‑Time Evolutionary Search

EvoSearch, a test‑time evolutionary search method, dramatically improves image and video generation by increasing inference compute without extra training, outperforming existing scaling techniques on diffusion and flow models while maintaining robustness and diversity across multiple benchmarks.

AI researchEvolutionary Searchdiffusion models

0 likes · 8 min read

How EvoSearch Supercharges Image and Video Generation with Test‑Time Evolutionary Search

Kuaishou Tech

Jun 20, 2025 · Artificial Intelligence

How OneRec Redefines Recommendation with End‑to‑End Generative Modeling and RL Alignment

The OneRec system from Kuaishou replaces traditional cascade recommendation pipelines with an encoder‑decoder architecture, leverages reward‑based preference alignment via reinforcement learning, achieves ten‑fold FLOPs gains, cuts operational costs by 90%, and delivers significant user‑engagement improvements across short‑video and local‑service scenarios.

KuaishouOneRecgenerative modeling

0 likes · 17 min read

How OneRec Redefines Recommendation with End‑to‑End Generative Modeling and RL Alignment