update caption

Author: lxasqjc

index.html

@@ -276,7 +276,7 @@ <h2 class="title is-3">Key Insights and Figures</h2>
                      style="width: 100%; vertical-align: middle;">
                 <figcaption>
                   <small>
-                    <em><strong>Lavender: Diffusion Instruction Tuning.</strong>
+                    <em>Figure 3. <strong>Lavender: Diffusion Instruction Tuning.</strong>
                       Lavender uses the text-vision attention maps of a Stable Diffusion Model, \(Attention_{SDM}\), as a guiding objective for the attention of the target vision-language model (VLM), \(Attention_{VLM}\). The \textit{Attention Alignment} module employs a 3-Layer ConvNet to transform \(Attention_{VLM}\) to match \(Attention_{SDM}\) via an MSE loss, acting as a regularisation term during supervised fine-tuning.</em>
                   </small>
                 </figcaption>