Lattice AI Research

Research focus

Code Generation & Program Synthesis (2)Eval Frameworks & Benchmarks (1)Architecture Design (Transformers, SSMs, MoE) (1)Data Curation & Synthetic Data (1)

Frequent co-authors

Xueyi Chen (2)Lingcheng Kong (2)Xingze Zou (1)Jing Wang (1)

Papers (2)

Mar 12, 2026

Xingze Zou +124d ago

MobileKernelBench: Can LLMs Write Efficient Kernels for Mobile Devices?

LLMs can now generate mobile kernels that outperform native libraries 27.4% of the time, thanks to a new multi-agent system that tackles compilation failures and hallucinations.

Xingze Zou, Jing Wang, Yuhua Zheng +10

Code Generation & Program Synthesis Eval Frameworks & Benchmarks

Feb 12, 2026

Haolei Bai +4Feb 12, 2026

DICE: Diffusion Large Language Models Excel at Generating CUDA Kernels

Diffusion LMs can now beat autoregressive models at CUDA kernel generation, thanks to a new dataset and reinforcement learning approach.

Haolei Bai, Lingcheng Kong, Xueyi Chen +2

Architecture Design (Transformers, SSMs, MoE)Code Generation & Program Synthesis Data Curation & Synthetic Data

Search

Haolei Bai

Publication activitypapers/week, last 8 weeks

Research focus

Frequent co-authors

Papers (2)