Langlin Huang

Langlin Huang黄浪林

Ph.D. Student in Computer Science · Washington University in St. Louis

I am a third-year PhD student at Washington University in St. Louis (WUSTL), luckily advised by Prof. Jiaxin Huang.

Before that, I received my Master degree at Institute of Computing Technology, Chinese Academy of Sciences (ICT/CAS) in Jul. 2024, where I studied at ICT Natural Language Processing (ICTNLP) group and worked on machine translation and multilingual NLP.

CV Google Scholar DBLP GitHub
Photo of Langlin Huang

Research

My research interests mainly lie in Natural Language Processing and Large Language Models. Particularly, I am interested in:

News

Show older newsHide older news
  • Our paper Training Data Efficiency in Multimodal Process Reward Models is accepted by ICML 2026. Thanks to all collaborators!
  • Our paper CaTS: Calibrated Test-Time Scaling for Efficient LLM Reasoning is accepted by ICLR 2026. Thanks to all collaborators!
  • We released RelayLLM: Efficient Reasoning via Collaborative Decoding.
  • Our CrossWordBench is accepted by COLM 2025. Thanks to all collaborators!
  • We released PosS, an effective Speculative Decoding method that reaches the highest draft prediction accuracy.
  • We released CrossWordBench, a challenging multimodal Benchmark that requires continuous reasoning ability.
  • One paper about Adaptive Heterogeneous MoE for flexible tokenization is accepted by NAACL 2025 as oral presentation. Thanks to all collaborators!
  • We released CaTS: Calibrated Test-Time Scaling for Efficient LLM Reasoning, finding self-calibration efficiently improves test-time scaling method which saves up to 94% sampling.
  • We released LARA, A logit aggregation approach for In-Context Learning!
  • My new journey set sail on Washington University in St. Louis, I love the peaceful campus and welcoming people!
  • One paper about Byte-based Language Model is accepted to ACL 2024! Thanks to all collaborators!
  • I successfully defended my dissertation for Master degree! My dissertation name is "A Study of Neural Machine Translation on Multiple Semantic Granularities (基于多语义粒度的神经机器翻译研究)".
  • One paper about Compositional Phrase Representation is accepted to EMNLP 2023! Thanks to all collaborators!
  • Our LLM BayLing (百聆) is released! BayLing is an instruction-following LLM with advanced language alignment and multi-turn interaction capability. Read our paper and try our online demo! Thanks to all collaborators!
  • We won first place in the Chinese-Thai track and second place in the Mongolian-Chinese track in the 18th China Conference on Machine Translation (CCMT 2022)! Thanks to all collaborators! The technical report is accessible to public, check here for Chinese-Thai track and here for Mongolian-Chinese track.
  • I start my M.S. life at University of Chinese Academy of Sciences (UCAS).
  • I obtain my B.E. degree from University of International Business and Economics (UIBE).
  • One paper about automatic construction of depression-domain lexicon is published on JMIR Medical Informatics! Thanks to all collaborators!

Publications

LLM Reasoning & Efficiency

NeurIPS 2026Workshops: MATH-AI and Pre-to-Post

On the Off-Policy Teacher in On-Policy Distillation

Langlin Huang*, Hao Liu*, Mononito Goswami, Xinyu Li, Prithwish Jana, Nikos Kanakaris, Patrick Blöbaum, Purak Jain (* equal contribution)

PaperProject Page
EMNLP 2026

Nonsense Helps: Prompt Space Perturbation Broadens Reasoning Exploration

Langlin Huang, Chengsong Huang, Jinyuan Li, Donghong Cai, Yuyi Yang, Jiaxin Huang

PaperCode
COLM 2026

SPEED: Specialized Position Experts for Efficient Speculative Decoding

Langlin Huang, Chengsong Huang, Jixuan Leng, Di Huang, Jiaxin Huang

PaperCode
NeurIPS 2026Spotlight

G-Zero: Self-Play for Open-Ended Generation from Zero Data

Chengsong Huang, Haolin Liu, Tong Zheng, Runpeng Dai, Langlin Huang, Jinyuan Li, Zongxia Li, Zhepei Wei, Yu Meng, Jiaxin Huang

PaperCode
ICML 2026

Training Data Efficiency in Multimodal Process Reward Models

Jinyuan Li, Chengsong Huang, Langlin Huang, Shaoyang Xu, Haolin Liu, Wenxuan Zhang, Jiaxin Huang

PaperCode
ICLR 2026

CaTS: Calibrated Test-Time Scaling for Efficient LLM Reasoning

Chengsong Huang, Langlin Huang, Jixuan Leng, Jiacheng Liu, Jiaxin Huang

PaperCode
EACL 2026Oral

Divide, Reweight, and Conquer: A Logit Arithmetic Approach for In-Context Learning

Chengsong Huang, Langlin Huang, Jiaxin Huang

PaperCode
EMNLP 2026

RelayLLM: Efficient Reasoning via Collaborative Decoding

Chengsong Huang, Tong Zheng, Langlin Huang, Jinyuan Li, Haolin Liu, Jiaxin Huang

PaperCode
COLM 2025

CrossWordBench: Evaluating the Reasoning Capabilities of LLMs and LVLMs with Controllable Puzzle Generation

Jixuan Leng, Chengsong Huang, Langlin Huang, Bill Yuchen Lin, William W. Cohen, Haohan Wang, Jiaxin Huang

PaperCodeDataset

Multilingual

NAACL 2025Oral

MoCE: Adaptive Mixture of Contextualization Experts for Byte-based Neural Machine Translation

Langlin Huang, Mengyu Bu, Yang Feng

PaperCode
ACL 2024

Integrating Multi-scale Contextualized Information for Byte-based Neural Machine Translation

Langlin Huang, Yang Feng

PaperCode
EMNLP 2023

Enhancing Neural Machine Translation with Semantic Units

Langlin Huang, Shuhao Gu, Zhuocheng Zhang, Yang Feng

PaperCode
Technical Report

BayLing: Bridging Cross-lingual Alignment and Instruction Following through Interactive Translation for Large Language Models

Shaolei Zhang, Qingkai Fang, Zhuocheng Zhang, Zhengrui Ma, Yan Zhou, Langlin Huang, Mengyu Bu, Shangtong Gui, Yunji Chen, Xilin Chen, Yang Feng

PaperCodeDemoProject Page

Others

JMIR Med. Inform.2020

Automatic Construction of a Depression-Domain Lexicon Based on Microblogs: Text Mining Study

Genghao Li, Bing Li, Langlin Huang, Sibing Hou

Paper

Awards

Services

Reviewer: NeurIPS 2025, COLM 2025, ICLR 2025, ACL Rolling Review, ACL 2023, TALLIP