We encountered some performance issues and implemented the following …

…fix. We updated the TokenUtils class within the Skills Web API to address these issues. A newly-introduced tokenizer has been implemented for encoding text, which significantly improves the efficiency of the TokenCount method. As a result of this change, the performance of the token counting operations has been enhanced, ensuring more accurate and faster responses for users.

webapi/Skills/Utils/TokenUtils.cs

@@ -17,6 +17,8 @@ namespace CopilotChat.WebApi.Skills.Utils;
 /// </summary>
 public static class TokenUtils
 {
+    private static SharpToken.GptEncoding tokenizer = SharpToken.GptEncoding.GetEncoding("cl100k_base");
+
     /// <summary>
     /// Semantic dependencies of ChatSkill.
     ///  If you add a new semantic dependency, please add it here.
@@ -98,7 +100,6 @@ internal static void GetFunctionTokenUsage(SKContext result, SKContext chatConte
     /// <param name="text">The string to calculate the number of tokens in.</param>
     internal static int TokenCount(string text)
     {
-        var tokenizer = SharpToken.GptEncoding.GetEncoding("cl100k_base");
         var tokens = tokenizer.Encode(text);
         return tokens.Count;
     }