Liuyihan Song — accepted AI-conference papers

2025

OpenCoder: The Open Cookbook for Top-Tier Code Large Language Models

Siming Huang, Tianhao Cheng, Jason Klein Liu, Weidi Xu, Jiaran Hao, Liuyihan Song +20

ACL 2025long

Code LLMs have been widely used in various domains, including code generation, logical reasoning, and agent systems. However, open-access code LLMs mostly only release weights, lacking key features such as reproducible data pipelines and transparent training protocols, which are crucial for advancin…

Cited by 0 Source ↗PDF ↗code ↗Scholar ↗

2021

Communication Efficient SGD via Gradient Sampling With Bayes Prior

Liuyihan Song, Kang Zhao, Pan Pan, Yu Liu, Yingya Zhang, Yinghui Xu +1

CVPR 2021poster

Gradient compression has been widely adopted in data-parallel distributed training of deep neural networks to reduce communication overhead. Some literatures have demonstrated that large gradients are more important than small ones because they contain more information, such as Top-k compressor. Oth…

Cited by 13 PDF ↗code ↗Scholar ↗