Skip to content

Commit b5027f9

Browse files
asudehasudeh
committed
--
1 parent 8eba83a commit b5027f9

10 files changed

Lines changed: 34 additions & 8 deletions

File tree

efficient-on-device-llm-inference/.DS_Store renamed to efficient-llm-inference/.DS_Store

File renamed without changes.

efficient-on-device-llm-inference/imgs/dehghankar25a.jpg renamed to efficient-llm-inference/imgs/dehghankar25a.jpg

File renamed without changes.

efficient-on-device-llm-inference/imgs/edbt25.jpg renamed to efficient-llm-inference/imgs/edbt25.jpg

File renamed without changes.

efficient-on-device-llm-inference/imgs/rsr-core.jpg renamed to efficient-llm-inference/imgs/rsr-core.jpg

File renamed without changes.

efficient-on-device-llm-inference/index.html renamed to efficient-llm-inference/index.html

Lines changed: 5 additions & 5 deletions
Original file line numberDiff line numberDiff line change
@@ -3,14 +3,14 @@
33
<head>
44
<meta charset="UTF-8" />
55
<meta name="viewport" content="width=device-width, initial-scale=1.0" />
6-
<title>Efficient and Private On-device LLM Inference</title>
6+
<title>Efficient LLM Inference</title>
77
<meta name="description" content="Project webpage for Efficient On-device LLM Inference." />
88
<meta name="robots" content="index,follow,max-image-preview:large" />
9-
<link rel="canonical" href="https://uic-indexlab.github.io/efficient-on-device-llm-inference/" />
9+
<link rel="canonical" href="https://uic-indexlab.github.io/efficient-llm-inference/" />
1010
<meta property="og:title" content="Efficient On-device LLM Inference" />
1111
<meta property="og:description" content="Project webpage for Efficient On-device LLM Inference." />
1212
<meta property="og:type" content="website" />
13-
<meta property="og:url" content="https://uic-indexlab.github.io/efficient-on-device-llm-inference/" />
13+
<meta property="og:url" content="https://uic-indexlab.github.io/efficient-llm-inference/" />
1414
<link rel="preconnect" href="https://fonts.googleapis.com">
1515
<link rel="preconnect" href="https://fonts.gstatic.com" crossorigin>
1616
<link href="https://fonts.googleapis.com/css2?family=Inter:wght@400;500;600;700;800&display=swap" rel="stylesheet">
@@ -22,9 +22,9 @@
2222
<div class="hero-layout">
2323
<div class="hero-main">
2424
<div class="eyebrow">Project Website</div>
25-
<h1>Efficient and Private On-device LLM Inference</h1>
25+
<h1>Efficient LLM Inference</h1>
2626
<p class="subtitle">
27-
Privacy-preserving and efficient inference on personal devices
27+
Efficient (and provacy-preserving) inference on personal devices
2828
</p>
2929
<p class="hero-text">
3030
This project advances systems and algorithmic foundations for running large language models efficiently on personal devices, reducing dependence on cloud-only deployment while improving privacy, accessibility, and cost efficiency.

efficient-on-device-llm-inference/script.js renamed to efficient-llm-inference/script.js

File renamed without changes.

efficient-on-device-llm-inference/style.css renamed to efficient-llm-inference/style.css

File renamed without changes.

‎js/indexstart.js‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -19,7 +19,7 @@ st+= ' Active Projects';
1919
st+= ' </a>';
2020
st+= ' <div class="dropdown-menu" aria-labelledby="fundedProjectsDropdown">';
2121
st+= ' <a class="dropdown-item" target="_blank" href="'+path+'nsf-iis-2348919/index.html">NSF IIS-2348919 | Fairness-aware Data Structures</a>';
22-
st+= ' <a class="dropdown-item" target="_blank" href="'+path+'efficient-on-device-llm-inference/index.html">Efficient and Private On-device LLM Inference</a>';
22+
st+= ' <a class="dropdown-item" target="_blank" href="'+path+'efficient-llm-inference/index.html">Efficient and Private On-device LLM Inference</a>';
2323
st+= ' <a class="dropdown-item" target="_blank" href="'+path+'trustworthy-llms-and-llm-agents/index.html">Trustworthy LLMs and LLM Agents</a>';
2424
st+= ' </div>';
2525
st += ' </li> ';

‎sitemap.xml‎

Lines changed: 1 addition & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -19,6 +19,6 @@
1919
<loc>https://uic-indexlab.github.io/trustworthy-llms-and-llm-agents/</loc>
2020
</url>
2121
<url>
22-
<loc>https://uic-indexlab.github.io/efficient-on-device-llm-inference/</loc>
22+
<loc>https://uic-indexlab.github.io/efficient-llm-inference/</loc>
2323
</url>
2424
</urlset>

‎trustworthy-llms-and-llm-agents/index.html‎

Lines changed: 27 additions & 1 deletion
Original file line numberDiff line numberDiff line change
@@ -54,6 +54,7 @@ <h1>Trustworthy LLMs and LLM Agents</h1>
5454
<p class="section-label">Table of Content</p>
5555
<nav class="toc-inner">
5656
<a href="#overview">Overview</a>
57+
<a href="#research-directions">Research Directions</a>
5758
<a href="#publications">Project Outcomes</a>
5859
<a href="#team">Team</a>
5960
</nav>
@@ -101,7 +102,32 @@ <h2>Why this project matters</h2>
101102
</div>
102103
</section>
103104

104-
<section class="section alt" id="publications">
105+
<section class="section alt" id="research-directions">
106+
<div class="container two-col">
107+
<div>
108+
<p class="section-label">Research Directions</p>
109+
<h2>What we do</h2>
110+
</div>
111+
<div>
112+
<div class="card">
113+
<p class="section-intro">
114+
Our research asks a broader set of questions about how large
115+
language models and multi-agent LLM systems can become genuinely
116+
trustworthy in practice. These questions include:
117+
</p>
118+
<ul class="paper-points">
119+
<li>How can agentic LLM systems remain reliable when multiple agents interact, disagree, or behave adversarially, and what mechanisms such as credibility scoring should govern collaboration?</li>
120+
<li>How should responsibility in LLM-based multi-agent systems be formalized across reliability, transparency, accountability, and fairness, so these properties can be evaluated rather than merely claimed?</li>
121+
<li>How can we make LLM decisions more robust to prompt and input sensitivity, including in symmetric settings where answer quality should not depend on arbitrary ordering?</li>
122+
<li>How can fairness and debiasing interventions be made practical for real deployments, including lightweight methods such as prompt rewriting and assisted self-correction that work with black-box models?</li>
123+
<li>How can trustworthy behavior be supported by principled algorithmic foundations, drawing on ranking and top-k optimization, relevance learning, reinforcement learning, computational geometry, and query rewriting?</li>
124+
</ul>
125+
</div>
126+
</div>
127+
</div>
128+
</section>
129+
130+
<section class="section" id="publications">
105131
<div class="container">
106132
<p class="section-label">Project Outcomes</p>
107133
<h2>Publication</h2>

0 commit comments

Comments
 (0)