Update index.html

index.html

@@ -164,20 +164,7 @@
 
           <div class="publication-item">
             <h4>
-              [3] ViLegalBERT & ViLegalQwen: Lightweight Domain-Adaptive
-              Language Models for Vietnamese Legal Text Processing
-            </h4>
-            <p class="publication-meta">
-              <strong>Authors:</strong> <strong>Truong-Phuc Nguyen</strong>,
-              Quy-Nhan Nguyen, Manh-Cuong Phan, Tien-Manh Tran, Huy-The Vu &
-              Minh-Tien Nguyen<br />
-              <strong>Status:</strong> Under Writing
-            </p>
-          </div>
-
-          <div class="publication-item">
-            <h4>
-              [4] Application of Machine Learning in Image Recognition to Detect
+              [3] Application of Machine Learning in Image Recognition to Detect
               Some Abnormalities in the Examination Rooms
             </h4>
             <p class="publication-meta">
@@ -196,7 +183,19 @@
 
           <div class="publication-item">
             <h4>
-              [1] UTEHY-NLU@ALQAC 2025: Dynamic Weighted Ensemble and Adaptive
+              [1] ViLegalLM: Language Models for Vietnamese Legal Text
+            </h4>
+            <p class="publication-meta">
+              <strong>Authors:</strong> <strong>Truong-Phuc Nguyen</strong>,
+              Quy-Nhan Nguyen, Van-Quyet Nguyen &
+              Minh-Tien Nguyen<br />
+              <strong>Status:</strong> Under Writing
+            </p>
+          </div>
+          
+          <div class="publication-item">
+            <h4>
+              [2] UTEHY-NLU@ALQAC 2025: Dynamic Weighted Ensemble and Adaptive
               Reasoning for Vietnamese Legal Text Processing
             </h4>
             <p class="publication-meta">
@@ -211,7 +210,7 @@
 
           <div class="publication-item">
             <h4>
-              [2] ViEduQA: A New Vietnamese Dataset for Question Answer
+              [3] ViEduQA: A New Vietnamese Dataset for Question Answer
               Generation in Education
             </h4>
             <p class="publication-meta">
@@ -232,7 +231,7 @@
 
           <div class="publication-item">
             <h4>
-              [3] Vietnamese Legal Question Answering: An Experimental Study
+              [4] Vietnamese Legal Question Answering: An Experimental Study
             </h4>
             <p class="publication-meta">
               <strong>Authors:</strong> Thu-Ha Nguyen,
@@ -305,22 +304,25 @@
               <div class="timeline-item">
                 <div class="timeline-date">September 2024 – Present</div>
                 <h3>
-                  ViLegalBERT & ViLegalQwen - Domain-specific Language Models
+                  ViLegalLM: Language Models for Vietnamese Legal Text
                 </h3>
                 <p style="color: var(--accent); font-weight: 600">
                   NLU Laboratory, Hung Yen University of Technology and
                   Education
                 </p>
                 <p class="timeline-content">
-                  Developing representation and generation models specifically
-                  for the legal domain in Vietnam through continual pretraining
-                  of language models on large datasets from four sources of
-                  authoritative legal documents in Vietnam. Legal pretrained
-                  models are trained on high-quality large-scale synthetic
-                  datasets, compared with base models and Vietnamese-specific
-                  models of the same size on the problems of Question Answering
-                  (True/False, Multiple-choice), Natural Language Inference,
-                  Text Classification.
+                  Developing one representation (135M) and two generation (1.54B, 
+                  1.72B) models specifically for the legal domain in Vietnam 
+                  through continual pretraining of language models on large 
+                  datasets from four sources of authoritative legal documents 
+                  in Vietnam. Legal pretrained models are trained on high-quality
+                  large-scale synthetic datasets, compared with 7 state-of-the-art 
+                  Vietnamese general and legal LMs of the same size across 10 
+                  benchmarks spanning 4 main tasks: Information Retrieval, 
+                  Question Answering, Natural Language Inference, and Syllogism Reasoning. 
+                  ViLegalLM achieves state-of-the-art performance on 10 benchmarks, 
+                  establishes the newest strong baselines for Vietnamese Legal text 
+                  processing.
                 </p>
                 <p
                   style="