minor

Author: f14-bertolotti

posts/02-09-25-torch-distributed/index.html

@@ -175,7 +175,8 @@ <h3> Initialization </h3>
             Finally, once your distributed job is complete, you need to terminate the process group by calling:
             </p>
 
-            <pre><code>torch.distributed.destroy_process_group()</code></pre>
+            <pre><code class="language-python">torch.distributed.destroy_process_group()
+            </code></pre>
 
             <p>
             This cleans up the communication resources and ensures that all processes exit gracefully.
@@ -499,6 +500,7 @@ <h3> Broadcasting Weights </h3>
 if torch.distributed.get_rank() == 0: 
     print(tokenizer.batch_decode(completions, skip_special_tokens=False))
 
+
 torch.distributed.destroy_process_group()
 
 </code></pre>