adjust img sizes

ml-research · May 24, 2024 · a5bcd13 · a5bcd13
1 parent 07df6dd
commit a5bcd13
Showing 1 changed file with 4 additions and 4 deletions.
diff --git a/projects/llavaguard/index.html b/projects/llavaguard/index.html
@@ -152,7 +152,7 @@ <h1 class="title is-1 publication-title">LlavaGuard: VLM-based Safeguards for Vi
     <div class="hero-body">
         <div class="container has-text-centered">
             <div>
-                <img src="/human-centered-genai/static/images/llavaguard/llavaguard_pipe.png" style="max-width: 1200px">
+                <img src="/human-centered-genai/static/images/llavaguard/llavaguard_pipe.png" style="max-width: 1000px">
             </div>
         </div>
     </div>
@@ -184,7 +184,7 @@ <h2 class="title is-3">Setup</h2>
             </div>
             <br/>
             <h2 class="has-text-justified">
-                LlavaGuard's safety taxonomy entailing 9 default categories and one additional NA category.
+                LlavaGuard's safety taxonomy entailing 9 default categories (O1-O9) and one additional NA category.
             </h2>
             <br/>
             <h2 class="title is-4">Policy Example</h2>
@@ -201,7 +201,7 @@ <h2 class="has-text-justified">
         <div class="container has-text-centered">
             <div>
                 <h2 class="title is-3">Results</h2>
-                <embed src="/human-centered-genai/static/images/llavaguard/llavaguard_compass.pdf" style="max-width: 700px">
+                <embed src="/human-centered-genai/static/images/llavaguard/llavaguard_compass.pdf" style="max-width: 1000px">
             </div>
             <br/>
             <h2 class="has-text-justified">
@@ -214,7 +214,7 @@ <h2 class="has-text-justified">
             </h2>
             <br/>
             <h2 class="title is-4">Qualitative</h2>
-            <embed src="/human-centered-genai/static/images/llavaguard/qualitative_eval.pdf" style="max-width: 700px">
+            <embed src="/human-centered-genai/static/images/llavaguard/qualitative_eval.pdf" style="max-width: 1000px">
             <h2 class="has-text-justified">
                 Safety evaluations for 2 test set images with LlavaGuard. The model offers detailed safety assessments including category, rationale, and rating. The first image is classified as safe according to the default policy. For the second image, we provide two evaluations: one with the default policy and another one where category O6 is declared as non-violating. LlavaGuard is able to adjust its rating according to the modified policy, providing well-grounded reasoning that justifies its safety rating.
             </h2>