chore(llm-tool): add the top_p option in the llm-tool (#41)

* chore: add top p option in llm-tool
* chore: wire up the top p with async generate

mlx-swift-examples.xcodeproj/project.xcworkspace/xcshareddata/swiftpm/Package.resolved

@@ -1,5 +1,4 @@
 {
-  "originHash" : "da53546673b6d05016b6e5640c18814c7dba5b5af8db34715afe6d633037c758",
   "pins" : [
     {
       "identity" : "gzipswift",
@@ -79,9 +78,9 @@
       "location" : "https://github.com/huggingface/swift-transformers",
       "state" : {
         "branch" : "main",
-        "revision" : "3bd02269b7797ade67c15679a575cd5c6f203ce6"
+        "revision" : "9d82e00af680253499f1a9372abb2552a73527fb"
       }
     }
   ],
-  "version" : 3
+  "version" : 2
 }

mlx-swift-examples.xcodeproj/xcshareddata/xcschemes/llm-tool.xcscheme

@@ -55,6 +55,10 @@
             argument = "--model mlx-community/CodeLlama-13b-Instruct-hf-4bit-MLX"
             isEnabled = "NO">
          </CommandLineArgument>
+         <CommandLineArgument
+            argument = "--top-p 0.95"
+            isEnabled = "NO">
+         </CommandLineArgument>
          <CommandLineArgument
             argument = "--model mlx-community/c4ai-command-r-v01-4bit"
             isEnabled = "YES">