Lattice AI Research

Research focus

Code Generation & Program Synthesis (2)Eval Frameworks & Benchmarks (2)Tool Use & Agents (1)

Frequent co-authors

Tencent WorkBuddy Bench Team (1)Tencent WorkBuddy Bench Team Siqi Cai (1)Shaopeng Chen (1)Siqi Cai (1)

Papers (2)

Jul 23, 2026

Tencent WorkBuddy Bench Team +443d ago

Tencent WorkBuddy Bench: A Multi-Domain Coding-Agent Benchmark with Contamination-Resistant Task Construction

Real-world coding tasks can be evaluated without risk of contamination, thanks to a unique reverse-engineering approach that ensures task prompts are untraceable.

Tencent WorkBuddy Bench Team, Tencent WorkBuddy Bench Team Siqi Cai, Shaopeng Chen +42

Code Generation & Program Synthesis Eval Frameworks & Benchmarks

Mar 17, 2026

Mar 17, 2026·also Corresponding Author, NJU, Waterloo

SWE-QA-Pro: A Representative Benchmark and Scalable Training Recipe for Repository-Level Code Understanding

A Qwen3-8B model, trained with a new SFT+RLAIF recipe on a challenging new benchmark, SWE-QA-Pro, beats GPT-4o in repository-level code understanding.

Songcheng Cai, Z. Lyu, Yuansheng Ni +14

Code Generation & Program Synthesis Eval Frameworks & Benchmarks Tool Use & Agents

Search

Z. Lyu

Publication activitypapers/week, last 8 weeks

Research focus

Frequent co-authors

Papers (2)