Use markdown table

app.py

@@ -648,9 +648,24 @@ Observe the [Speeding up Inference](https://sbert.net/docs/sentence_transformer/
 
 <details><summary>Click to see performance benchmarks</summary>
 
-| GPU | CPU |
-| --- | --- |
-| ![](https://sbert.net/_images/backends_benchmark_gpu.png) | ![](https://sbert.net/_images/backends_benchmark_cpu.png) |
+<table>
+  <thead>
+    <tr>
+      <th>GPU</th>
+      <th>CPU</th>
+    </tr>
+  </thead>
+  <tbody>
+    <tr>
+      <td>
+        <img src="https://sbert.net/_images/backends_benchmark_gpu.png" alt="">
+      </td>
+      <td>
+        <img src="https://sbert.net/_images/backends_benchmark_cpu.png" alt="">
+      </td>
+    </tr>
+  </tbody>
+</table>
 
 * `onnx` refers to the ONNX backend
 * `onnx-qint8` refers to ONNX (Dynamic Quantization)