Official hub of Tencent Advertising Technology, sharing the team's latest cutting-edge achievements and advertising technology applications.
This article details QueryFormer, the winning solution for KDD Cup 2026 Tencent UniRec Challenge, which introduces a cross-attention query generation mechanism to unify feature interaction and behavior sequence modeling, achieving 0.83254 AUC on the official test set.
Tencent's ad platform evolved from human-led refactoring with AI Copilot to Loop Engineering—a closed-loop system using data cut planes for observable verification and four specialized agents to automate refactoring of a 300k-line service, cutting cycle from months to weeks and manual diffs from 9 to near zero.
Two senior undergraduates won the Scaling Law Original Breakthrough Award and 10th place in the academic track at the 2026 Tencent Advertising Algorithm Competition by running nearly 1000 experiments, discovering that increasing model width alone reduces AUC unless accompanied by new informative tokens, leading to their CoSFormer architecture that scales capacity and information jointly.
At the KDD Cup 2026 in Jeju, South Korea, Chinese teams lozyyeah and sunshot captured the academic and industrial titles, respectively, sharing a $3 million prize pool by pioneering feature‑transport and QueryFormer techniques, leveraging massive real‑world ad data and AI assistants like DeepSeek and ChatGPT.
The author details how, without any API access, they leveraged DeepSeek’s web interface to iteratively develop and refine the QueryFormer model—solving multi‑GPU batch issues, enhancing query generation, and ultimately achieving the TAAC × KDD Cup 2026 industrial track championship.
The TAAC × KDD Cup 2026 workshop in Jeju Island gathered researchers and engineers from academia and leading tech firms to dissect real‑world recommendation challenges, present the competition’s unified sequence‑modeling and feature‑interaction task, and showcase winning solutions that balance effectiveness, efficiency, and scalability.
The paper identifies a structural mismatch between probability‑driven beam search and reward‑driven RL fine‑tuning in generative recommendation, proposes V-STAR with value‑guided efficient decoding (VED) and sibling‑wise GRPO to align decoding and optimization, and demonstrates superior offline and online performance through extensive experiments and ablations.
The article presents Tencent’s GRB (Generative Reinforcement Bidding) framework, a multi‑slot online advertising solution that combines a Decision‑Transformer‑based generative model with an Offline‑to‑On‑Policy (O2P) training paradigm and an adaptive constraint‑exploration module, and demonstrates statistically significant gains over strong baselines in both offline and large‑scale online A/B tests.
AdPilot, the first end‑to‑end autonomous advertising agent, reformulates ad delivery as a Markov decision process and combines structured memory, LLM‑enhanced reasoning, and a GRPO‑based reinforcement‑learning engine, while the newly released AdBench benchmark evaluates its superior performance across 38 scenarios and 7,600 instances, outperforming strong baselines by up to 11.76%.
A comprehensive dialogue between Hong Kong Chinese University’s vice president and Tencent’s advertising AI experts explores AI frontiers, trustworthy and embodied AI, the four‑generation evolution of recommendation systems, the role of the Tencent Advertising Algorithm Competition in talent development, and practical advice for young professionals navigating the AI‑driven advertising industry.
