1<!doctype html> 2<html lang="en" class="no-js"> 3 <head> 4 <meta charset="utf-8"> 5 6<!-- begin SEO --> 7 8 9 10 11 12<title>Yuyang Song's Homepage - Homepage</title> 13 14 15 16 17 18 19 20<meta property="og:locale" content="en"> 21<meta property="og:site_name" content="Yuyang Song's Homepage"> 22<meta property="og:title" content="Yuyang Song's Homepage"> 23 24 25 <link rel="canonical" href="https://yuyang-song.github.io/"> 26 <meta property="og:url" content="https://yuyang-song.github.io/"> 27 28 29 30 <meta property="og:description" content=""> 31 32 33 34 35 36 37 38 39 40<!-- end SEO --> 41 42 43<meta name="HandheldFriendly" content="True"> 44<meta name="MobileOptimized" content="320"> 45<meta name="viewport" content="width=device-width, initial-scale=1.0"> 46
47<script> 48 document.documentElement.className = document.documentElement.className.replace(/\bno-js\b/g, '') + ' js '; 49</script>
49 50 51<!-- For all browsers --> 52<link rel="stylesheet" href="assets/css/main.css"> 53 54<meta http-equiv="cleartype" content="on"> 55<head> 56 <base target="_blank"> 57</head> 58 <link rel="apple-touch-icon" sizes="180x180" href="images/apple-touch-icon.png"> 59<link rel="icon" type="image/png" sizes="32x32" href="images/favicon-32x32.png"> 60<link rel="icon" type="image/png" sizes="16x16" href="images/favicon-16x16.png"> 61<link rel="manifest" href="images/site.webmanifest"> 62 63<meta name="msapplication-TileColor" content="#000000"> 64<meta name="msapplication-TileImage" content="images/mstile-144x144.png?v=M44lzPylqQ"> 65<meta name="msapplication-config" content="images/browserconfig.xml?v=M44lzPylqQ"> 66<meta name="theme-color" content="#ffffff"> 67<link rel="stylesheet" href="assets/css/academicons.css"/> 68
69<script type="text/x-mathjax-config"> MathJax.Hub.Config({ TeX: { equationNumbers: { autoNumber: "all" } } }); </script>
vendor: 1 bytes, line 69
69
70<script type="text/x-mathjax-config"> 71 MathJax.Hub.Config({ 72 tex2jax: { 73 inlineMath: [ ['$','$'], ["\\(","\\)"] ], 74 processEscapes: true 75 } 76 }); 77</script>
vendor: 1 bytes, line 77
77
78<script src='https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.4/latest.js?config=TeX-MML-AM_CHTML' async></script>
78 79 80<!-- end custom head snippets --> 81 82 </head> 83 84 <body> 85 86 <!--[if lt IE 9]> 87<div class="notice--danger align-center" style="margin: 0;">You are using an <strong>outdated</strong> browser. Please <a href="http://browsehappy.com/">upgrade your browser</a> to improve your experience.</div> 88<![endif]--> 89 <div class="masthead"> 90 <div class="masthead__inner-wrap"> 91 <div class="masthead__menu"> 92 <nav id="site-nav" class="greedy-nav"> 93 <button><div class="navicon"></div></button> 94 <ul class="visible-links"> 95 <li class="masthead__menu-item masthead__menu-item--lg masthead__menu-home-item"><a href="#about-me">Homepage</a></li> 96 97 <li class="masthead__menu-item"><a href="/#about-me">About Me</a></li> 98 99 <li class="masthead__menu-item"><a href="/#-publications">Publications</a></li> 100 101 <li class="masthead__menu-item"><a href="/#-honors-and-awards">Honors and Awards</a></li> 102 103 <li class="masthead__menu-item"><a href="/#-educations">Educations</a></li> 104 105 <li class="masthead__menu-item"><a href="/#-internships">Internships</a></li> 106 107 </ul> 108 <ul class="hidden-links hidden"></ul> 109 </nav> 110 </div> 111 </div> 112</div> 113 114 <div id="main" role="main"> 115 116 <div class="sidebar sticky"> 117 118 119<div itemscope itemtype="http://schema.org/Person" class="profile_box"> 120 121 <div class="author__avatar"> 122 <img src="images/android-chrome-512x512.png" class="author__avatar" alt="Yuyang Song"> 123 </div> 124 125 <div class="author__content"> 126 <h3 class="author__name">Yuyang Song</h3> 127 <p class="author__bio">Ph.D. @ UT Austin</p> 128 </div> 129 130 <div class="author__urls-wrapper"> 131 <!-- <button class="btn btn--inverse">More Info & Contact</button> --> 132 <ul class="author__urls social-icons"> 133 134 135 <li><i class="fa fa-fw fa-map-marker" aria-hidden="true"></i> Austin, TX, USA</li> 136 137 138 139 140 <li><a href="mailto:[email protected]"><i class="fas fa-fw fa-envelope" aria-hidden="true"></i> [email protected]</a></li> 141 142 143 144 145 146 147 148 149 150 151 152 153 154 <li><a href="https://github.com/Yuyang-Song"><i class="fab fa-fw fa-github" aria-hidden="true"></i> Github</a></li> 155 156 157 158 159 160 161 162 163 164 165 166 167 168 169 170 171 172 173 </ul> 174 <div class="author__urls_sm"> 175 176 177 <a href="mailto:[email protected]"><i class="fas fa-fw fa-envelope" aria-hidden="true"></i></a> 178 179 180 181 182 183 184 185 186 187 188 189 190 191 <a href="https://github.com/Yuyang-Song"><i class="fab fa-fw fa-github" aria-hidden="true"></i></a> 192 193 194 195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 </div> 211 </div> 212</div> 213 214 215 </div> 216 217 218 <article class="page" itemscope itemtype="http://schema.org/CreativeWork"> 219 <meta itemprop="headline" content=""> 220 <div class="page__inner-wrap"> 221 <section class="page__content" itemprop="text"> 222 223<p><span class="anchor" id="about-me"></span></p> 224 225<!-- Yuyang Song is an incoming PhD student in Computer Science at The University of Texas at Austin, advised by Prof. [Yizhong Wang](https://yizhong-wang.com/). He receives his B.S. in Computer Science from the Wu Yuzhang Honors College at Sichuan University in June 2026, under the supervision of Prof. [Mingjie Tang](https://merlintang.github.io/). Previously, he was a research intern at [Purdue University](https://www.purdue.edu/) advised by Prof. [Jianguo Wang](https://www.cs.purdue.edu/homes/csjgwang/). His research interest lies in **Efficient LLM Reasoning** and **System Optimization**. --> 226 227<!-- Hi! I'm Yuyang Song, an incoming PhD student in Computer Science at [University of Texas at Austin](https://www.cs.utexas.edu/), where I'll be working with Prof. [Yizhong Wang](https://yizhong-wang.com/). I wrap up my B.S. in Computer Science at the Wu Yuzhang Honors College, [Sichuan University](http://en.scu.edu.cn/), advised by Prof. [Mingjie Tang](https://merlintang.github.io/). Before that, I was fortunate to be advised by Prof. [Jianguo Wang](https://www.cs.purdue.edu/homes/csjgwang/) at [Purdue University](https://www.purdue.edu/). --> 228 229<p>Hi! Iâm Yuyang Song, a first-year PhD student in Computer Science at <
229a href="https://www.cs.utexas.edu/">University of Texas at Austin</a>, where I feel fortunate to be advised by Prof. <a href="https://yizhongw.github.io/">Yizhong Wang</a>. I earned my B.S. in Computer Science at the Wu Yuzhang Honors College, <a href="http://en.scu.edu.cn/">Sichuan University</a>, under the supervision of Prof. <a href="https://merlintang.github.io/">Mingjie Tang</a>. From October 2025 to July 2026, I was a member of the Foundation Model Group at IQuest AI Lab, Ubiquant.</p> 230 231<p>Iâm broadly interested in <strong>how data shapes a model as it flows through training, and how architecture and training should be designed around that flow</strong>, with efficiency as a guiding principle. Rather than treating the model, its data, and its training as separate problems, I see them as forces that continually shape one another, and I want to understand how to harness that interplay to build capable models more efficiently.</p> 232 233<h1 id="-selected-publications">ð Selected Publications</h1> 234<div class="paper-box"><div class="paper-box-image"><div><div class="badge">Tech Report</div><img src="images/iquest-q1.png" alt="sym" width="100%" /></div></div> 235<div class="paper-box-text"> 236 237 <p><a href="https://iquestlab.github.io/">IQuest-Q1: Advancing Foundation Capabilities for Agentic CLI Systems 238</a> | <a href="https://huggingface.co/IQuestLab/IQuest-Q1"><strong>Code</strong></a></p> 239 240 <p><strong><span class="show_paper_citations" data="DhtAFkwAAAAJ:ALROH1vI_8AC"></span></strong>
241IQuest-Q1 is an open-source agentic foundation model for CLI systems. A sparse Mixture-of-Experts model with 320B total parameters and 15B activated, it combines general and coding agent capabilities with efficient compute for medium- and long-horizon tasks. During development, it also takes part in its own improvement under human supervision.</p> 242 </div> 243</div> 244 245<div class="paper-box"><div class="paper-box-image"><div><div class="badge">ICML 2026 (Spotlight)</div><img src="images/terminal.png" alt="sym" width="100%" /></div></div> 246<div class="paper-box-text"> 247 248 <p><a href="https://openreview.net/pdf?id=PeFSCRulgy">Large-Scale Terminal Agentic Trajectory Generation 249from Dockerized Environments</a> | <a href="https://github.com/multimodal-art-projection/TerminalTraj"><strong>Code</strong></a></p> 250 251 <p>Siwei Wu*, Yizhi Li*, <strong>Yuyang Song</strong>*, Wei Zhang, Riza Batista-Navarro, Xian yang, Mingjie Tang, Bryan Dai, Jian Yang, Chenghua Lin.</p> 252 253 <p><strong><span class="show_paper_citations" data="DhtAFkwAAAAJ:ALROH1vI_8AC"></span></strong> 254We study how large-scale, execution-grounded terminal trajectories reshape the agentic capabilities of code LLMs. Trained on this data, TerminalTraj-32B achieves state-of-the-art results among models under 100B parameters and matches models more than 10à larger, while showing markedly steeper test-time scalingâevidence that the right training data, not just scale, drives terminal agent capability.</p> 255 </div> 256</div> 257 258<div class="paper-box"><div class="paper-box-image"><div><div class="badge">Tech Report</div><img src="images/iquest_result.png" alt="sym" width="100%" /></div></div> 259<div class="paper-box-text"> 260 261 <p><a href="https://arxiv.org/pdf/2603.16733">IQuest-Coder-V1 Technical Report</a> | <a href="https://huggingface.co/IQuestLab"><strong>Code</strong></a></p> 262 263 <p>Jian Yang, Wei Zhang, Shawn Guo, Zhengmao Ye, Lin Jing, Shark Liu, Yizhi Li, Jiajun Wu, Cening Liu, X. Ma, <strong>Yuyang Song</strong>, et al.</p> 264 265 <p><strong><span class="show_paper_citations" data="DhtAFkwAAAAJ:ALROH1vI_8AC"></span></strong> 266IQuest-Coder-V1 is a new family of code large language models (LLMs) designed to advance autonomous software engineering and code intelligence. Built on the innovative code-flow multi-stage training paradigm, IQuest-Coder-V1 captures the dynamic evolution of software logic, delivering state-of-the-art performance across critical dimensions.</p> 267 </div> 268</div> 269 270<div class="paper-box"><div class="paper-box-image"><div><div class="badge">arXiv'25</div><img src="images/quite.png" alt="sym" width="100%" /></div></div> 271<div class="paper-box-text"> 272 273 <p><a href="https://arxiv.org/pdf/2506.07675">QUITE: A Query Rewrite System Beyond Rules with LLM Agents</a> | <a href="https://github.com/Yuyang-Song/QUITE"><strong>Code</strong></a></p> 274 275 <p><strong>Yuyang Song</strong>, Hanxu Yan, Jiale Lao, Yibo Wang, Yufei Li, Yuanchun Zhou, Jianguo Wang, Mingjie Tang</p> 276 277 <p><strong><span class="show_paper_citations" data="DhtAFkwAAAAJ:ALROH1vI_8AC"></span></strong> 278QUITE is a training-free and feedback-aware system based on LLM agents that rewrites SQL queries into semantically equivalent forms with significantly better performance, covering a broader range of query patterns and rewrite strategies compared to rule-based methods.</p> 279 </div> 280</div> 281 282<!-- - [Lorem ipsum dolor sit amet, consectetur adipiscing elit. Vivamus ornare aliquet ipsum, ac tempus justo dapibus sit amet](https://github.com), A, B, C, **CVPR 2020** --> 283 284<!-- # ð¬ Projects 285### Query Rewrite via Large Language Model for DBMS | 2024.07 - 2024.08 286- Leverage the analytical power of Large Language Model to propose the relational rules of SQL optimization. 287- Repeat a LLM-based optimization pipline to reduce the hallucination issues in Large Language Model. 288- Construct a rewrite rule knowledge pool to further gain the learn ability. 289- [Query Rewrite via Large Language Model for DBMS](https://github.com/Orderheart/GenRewrite) 290 291### Construction of targeted anti-caries polypeptides based on machine learning | 2023.11 - 2024.03 292- Characterization of anti-caries antimicrobial peptide protein sequence and construction of data set. 293- Feature extraction and feature scoring index system were constructed by machine learning algorithm. 294- The optimal antimicrobial peptides were found by regression analysis of MIC value. 295- [Construction of targeted anti-caries polypeptides based on machine learning](https://github.com/Orderheart/Machine_Learning_AMP_Analysis) --> 296 297<!-- <font color=red>Nationcal Scholarship</font> --> 298 299<h1 id="-honors-and-awards">ð Honors and Awards</h1> 300<ul> 301 <li><em>2026.05</em>, Outstanding Undergraduate Graduate, Sichuan Province.</li> 302 <li><em>2023.11</em>, First Prize, China Undergraduate Mathematical Contest in Modeling.</li> 303 <li><em>2023.10</em>, National Scholarship, Ministry of Education, China (Top 1%).</li> 304 <li><em>2023.10</em>, First-Class Scholarship, Sichuan University (Top 1%).</li> 305 <li><em>2023.08</em>, First Prize, China Undergraduate Internet of Things Design Competition.</li> 306</ul> 307
308<h1 id="-educations">ð Educations</h1> 309<div class="school-list"> 310<div class="school-item"> 311 <div class="school-logo"><img src="images/utaustin.png" alt="UT Austin" /></div> 312 <div class="school-text"> 313 <p><em>2026.08 - Present</em>, Ph.D. in Computer Science, University of Texas at Austin.</p> 314 </div> 315</div> 316<div class="school-item"> 317 <div class="school-logo"><img src="images/scu.png" alt="Sichuan University" /></div> 318 <div class="school-text"> 319 <p><em>2022.09 - 2026.06</em>, B.S. in Computer Science, Wuyuzhang College (Honours Programme), Sichuan University.</p> 320 </div> 321</div> 322</div> 323 324<!-- # ð¬ Invited Talks 325- *2021.06*, Lorem ipsum dolor sit amet, consectetur adipiscing elit. Vivamus ornare aliquet ipsum, ac tempus justo dapibus sit amet. 326- *2021.03*, Lorem ipsum dolor sit amet, consectetur adipiscing elit. Vivamus ornare aliquet ipsum, ac tempus justo dapibus sit amet. \| [\[video\]](https://github.com/) --> 327 328<h1 id="-internships">ð» Internships</h1> 329<ul> 330 <li><em>2025.10 - 2026.07</em>, Research Internship, IQuest AI Lab, Ubiquant, China. 331<br /></li> 332</ul> 333 334<div style="width: 300px; height: 200px; overflow: hidden; margin: 0 auto;"> 335
335<script type="text/javascript" id="mapmyvisitors" src="https://mapmyvisitors.com/map.js?cl=ffffff&w=a&t=tt&d=BcrM0fkNxoe20Vnth_-J_CaOsYoubXBEWPLVc2YWMeI"></script>
335 336</div> 337 338 </section> 339 </div> 340 </article> 341 </div> 342 343
343<script src="assets/js/main.min.js"></script>
343 344 345 346 347<!-- Global site tag (gtag.js) - Google Analytics -->
348<script async src="https://www.googletagmanager.com/gtag/js?id="></script>
vendor: 1 bytes, line 348
348
349<script> 350 window.dataLayer = window.dataLayer || []; 351 function gtag(){dataLayer.push(arguments);} 352 gtag('js', new Date()); 353 354 gtag('config', ""); 355</script>
355 356 357
358<script> 359 $(document).ready(function () { 360 361 var gsDataBaseUrl = 'https://cdn.jsdelivr.net/gh/Yuyang-Song/Yuyang-Song.github.io@' 362 363 $.getJSON(gsDataBaseUrl + "google-scholar-stats/gs_data.json", function (data) { 364 var totalCitation = data['citedby'] 365 document.getElementById('total_cit').innerHTML = totalCitation; 366 var citationEles = document.getElementsByClassName('show_paper_citations') 367 Array.prototype.forEach.call(citationEles, element => { 368 var paperId = element.getAttribute('data') 369 var numCitations = data['publications'][paperId]['num_citations'] 370 element.innerHTML = '| Citations: ' + numCitations; 371 }); 372 }); 373 }) 374</script>
374 375 376 377 </body> 378</html>
Line numbers count LF bytes from the start of the resource, as the search results do. Vendor segments are library code the classifier recognised; they are stored but not indexed. Bytes are shown as Latin1 characters, one per byte.