|
127 | 127 | .stories-gallery img:hover { |
128 | 128 | transform: scale(1.05); |
129 | 129 | } |
| 130 | + |
| 131 | + .news-list { |
| 132 | + margin: 0; |
| 133 | + padding: 0; |
| 134 | + margin-top: 14px; |
| 135 | + } |
| 136 | + |
| 137 | + .news-item { |
| 138 | + display: flex; |
| 139 | + gap: 16px; |
| 140 | + margin-bottom: 8px; |
| 141 | + align-items: baseline; |
| 142 | + } |
| 143 | + |
| 144 | + .news-date { |
| 145 | + min-width: 75px; |
| 146 | + font-weight: bold; |
| 147 | + flex-shrink: 0; |
| 148 | + } |
| 149 | + |
| 150 | + .news-content { |
| 151 | + flex: 1; |
| 152 | + } |
130 | 153 | </style> |
131 | 154 | </head> |
132 | 155 | <body> |
@@ -176,67 +199,54 @@ <h1>Lee Optimization Group</h1> |
176 | 199 | <a href="hiring.html">Hiring</a>. |
177 | 200 | </p> --> |
178 | 201 | <h3>Recent News</h3> |
179 | | - <ul> |
180 | | - |
181 | | - <li> |
182 | | - <strong>Feb 2026 </strong> |
183 | | - 🤝 Our lab has been selected for the "Early-Career Researcher Infrastructure Support Program" (NRF) — one of the most competitive grants in Korea. With <em>500M KRW</em> in funding, we will build a high-performance compute infrastructure to power our research on extreme compression of hyperscale AI foundation models. |
184 | | - </li> |
185 | | - |
186 | | - <li> |
187 | | - <strong>Feb 2026 </strong> |
188 | | - 📃 Our new paper is now available on <a href="https://arxiv.org/abs/2602.18181v1" target="_blank">arXiv</a>! In this work, We introduce SeedFlood, a decentralized LLM training framework that enables model-size–independent communication cost and perfect consensus. |
189 | | - </li> |
190 | | - |
191 | | - <li> |
192 | | - <strong>Feb 2026 </strong> |
193 | | - 📃 Our new paper is now available on <a href="https://www.arxiv.org/abs/2602.03515" target="_blank">arXiv</a>! In this work, we propose a basis rotation approach to effectively address the gradient staleness problem in asynchronous pipeline parallelism. |
194 | | - </li> |
195 | | - |
196 | | - <li> |
197 | | - <strong>Feb 2026 </strong> |
198 | | - ✈️ Kwanhee started as a visiting student researcher in the |
199 | | - <a href="https://daslab.pages.ista.ac.at/" target="_blank"> |
200 | | - Dan Alistarh Group</a>. |
201 | | - </li> |
202 | | - |
203 | | - <li> |
204 | | - <strong>Jan 2026 </strong> |
205 | | - 🎓 Our paper on extreme LLM sparsity |
206 | | - (<a href="https://arxiv.org/abs/2510.01650" target="_blank">ELSA</a>) |
207 | | - has been accepted to ICLR 2026. |
208 | | - </a> |
209 | | - </li> |
210 | | - |
211 | | - <li> |
212 | | - <strong>Nov 2025 </strong> |
213 | | - 🏆 Dongyeop recived the |
214 | | - <a href="https://www.qualcomm.com/research/university-relations/innovation-fellowship/2025-south-korea" target="_blank">Qualcomm Innovation Fellowship Korea 2025</a>. |
215 | | - </a> |
216 | | - </li> |
217 | | - |
218 | | - <!-- <li> |
219 | | - <strong>Nov 2025 </strong> |
220 | | - 🏆 Our new paper on extreme LLM sparsity |
221 | | - (<a href="https://arxiv.org/abs/2510.01650" target="_blank">ELSA</a>) |
222 | | - won Best Paper Award at JKAIA 2025. |
223 | | - </a> |
224 | | - </li> --> |
225 | | - |
226 | | - <!-- <li> |
227 | | - <strong>Nov 2025 </strong> |
228 | | - 🏆 Hyunji recived NRF Fellowship. |
229 | | - </a> |
230 | | - </li> --> |
231 | | - |
232 | | - <!-- <li> |
233 | | - <strong>Nov 2025 </strong> |
234 | | - 🎓 We had 2 papers |
235 | | - (<a href="https://arxiv.org/abs/2510.25798" target="_blank">MemEIC</a> and <a href="https://arxiv.org/abs/2505.16705" target="_blank">An Analysis of Impact of Noisy Annotations on Concept Bottleneck Models</a>) |
236 | | - accepted to NeurIPS 2025. |
237 | | - </li> --> |
238 | | - |
239 | | - </ul> |
| 202 | + <div class="news-list"> |
| 203 | + |
| 204 | + <div class="news-item"> |
| 205 | + <span class="news-date">Feb 2026</span> |
| 206 | + <span class="news-content">🤝 Our lab has been selected for the "Early-Career Researcher Infrastructure Support Program" (NRF) — one of the most competitive grants in Korea. With <em>500M KRW</em> in funding, we will build a high-performance compute infrastructure to power our research on extreme compression of hyperscale AI foundation models.</span> |
| 207 | + </div> |
| 208 | + |
| 209 | + <div class="news-item"> |
| 210 | + <span class="news-date">Feb 2026</span> |
| 211 | + <span class="news-content">📃 Our new paper is now available on <a href="https://arxiv.org/abs/2602.18181v1" target="_blank">arXiv</a>! In this work, We introduce SeedFlood, a decentralized LLM training framework that enables model-size–independent communication cost and perfect consensus.</span> |
| 212 | + </div> |
| 213 | + |
| 214 | + <div class="news-item"> |
| 215 | + <span class="news-date">Feb 2026</span> |
| 216 | + <span class="news-content">📃 Our new paper is now available on <a href="https://www.arxiv.org/abs/2602.03515" target="_blank">arXiv</a>! In this work, we propose a basis rotation approach to effectively address the gradient staleness problem in asynchronous pipeline parallelism.</span> |
| 217 | + </div> |
| 218 | + |
| 219 | + <div class="news-item"> |
| 220 | + <span class="news-date">Feb 2026</span> |
| 221 | + <span class="news-content">✈️ Kwanhee started as a visiting student researcher in the <a href="https://daslab.pages.ista.ac.at/" target="_blank">Dan Alistarh Group</a>.</span> |
| 222 | + </div> |
| 223 | + |
| 224 | + <div class="news-item"> |
| 225 | + <span class="news-date">Jan 2026</span> |
| 226 | + <span class="news-content">🎓 Our paper on extreme LLM sparsity (<a href="https://arxiv.org/abs/2510.01650" target="_blank">ELSA</a>) has been accepted to ICLR 2026.</span> |
| 227 | + </div> |
| 228 | + |
| 229 | + <div class="news-item"> |
| 230 | + <span class="news-date">Nov 2025</span> |
| 231 | + <span class="news-content">🏆 Dongyeop recived the <a href="https://www.qualcomm.com/research/university-relations/innovation-fellowship/2025-south-korea" target="_blank">Qualcomm Innovation Fellowship Korea 2025</a>.</span> |
| 232 | + </div> |
| 233 | + |
| 234 | + <!-- <div class="news-item"> |
| 235 | + <span class="news-date">Nov 2025</span> |
| 236 | + <span class="news-content">🏆 Our new paper on extreme LLM sparsity (<a href="https://arxiv.org/abs/2510.01650" target="_blank">ELSA</a>) won Best Paper Award at JKAIA 2025.</span> |
| 237 | + </div> --> |
| 238 | + |
| 239 | + <!-- <div class="news-item"> |
| 240 | + <span class="news-date">Nov 2025</span> |
| 241 | + <span class="news-content">🏆 Hyunji recived NRF Fellowship.</span> |
| 242 | + </div> --> |
| 243 | + |
| 244 | + <!-- <div class="news-item"> |
| 245 | + <span class="news-date">Nov 2025</span> |
| 246 | + <span class="news-content">🎓 We had 2 papers (<a href="https://arxiv.org/abs/2510.25798" target="_blank">MemEIC</a> and <a href="https://arxiv.org/abs/2505.16705" target="_blank">An Analysis of Impact of Noisy Annotations on Concept Bottleneck Models</a>) accepted to NeurIPS 2025.</span> |
| 247 | + </div> --> |
| 248 | + |
| 249 | + </div> |
240 | 250 |
|
241 | 251 | <!-- <section class="stories-section"> |
242 | 252 |
|
|
0 commit comments