SDK: Update model alias resolution (#249)

* don't hardcode ep list

* update best model logic

* clean

* put device in front

* update versions to 0.5.0, js sdk

* update c#

* python integ test

* open ai dev req

* skip service stops, fix cs download publisher

* add js integration test

* update c# version to 0.3.0

Author: jambayk

.gitignore

@@ -1,3 +1,5 @@
+test.ipynb
+
 # Visual Studio and Visual Studio Code
 bin/
 obj/

sdk/cs/samples/TestApp/Program.cs

@@ -17,41 +17,41 @@ public static async Task Main(string[] args)
     {
         var app = new TestApp(); // Create an instance of TestApp
 
-        Console.WriteLine(new string('=', 80)); // Separator for clarity
+        Console.WriteLine(new string('=', 80));
         Console.WriteLine("Testing catalog integration...");
         await app.TestCatalog(); // Call the instance method
 
-        Console.WriteLine(new string('=', 80)); // Separator for clarity
+        Console.WriteLine(new string('=', 80));
         Console.WriteLine("Testing cache operations...");
         await app.TestCacheOperations(); // Call the instance method
 
-        Console.WriteLine(new string('=', 80)); // Separator for clarity
-        Console.WriteLine("Testing OpenAI integration (from stopped service)...");
-        using var manager = new FoundryLocalManager();
-        if (manager != null)
+        string[] aliasesOrModelIds = new[] { "qwen2.5-0.5b", "qwen2.5-0.5b-instruct-generic-cpu:3" };
+
+        foreach (var aliasOrModelId in aliasesOrModelIds)
         {
-            await manager.StopServiceAsync();
+            Console.WriteLine(new string('=', 80));
+            // don't stop for now. the catalog api doesn't return register EP models afer stop, start and model list
+            // Console.WriteLine($"Testing OpenAI integration (from stopped service) with {aliasOrModelId}...");
+            // using var manager = new FoundryLocalManager();
+            // await manager.StopServiceAsync();
+            // await app.TestOpenAIIntegration(aliasOrModelId);
+
+            Console.WriteLine(new string('=', 80));
+            Console.WriteLine($"Testing OpenAI integration (service running) with {aliasOrModelId}...");
+            await app.TestOpenAIIntegration(aliasOrModelId);
+
+            Console.WriteLine(new string('=', 80));
+            Console.WriteLine($"Testing service operations with {aliasOrModelId}...");
+            await app.TestService();
+
+            Console.WriteLine(new string('=', 80));
+            Console.WriteLine($"Testing model (un)loading with {aliasOrModelId}...");
+            await app.TestModelLoadUnload(aliasOrModelId);
+
+            // Console.WriteLine(new string('=', 80));
+            // Console.WriteLine($"Testing force downloading with {aliasOrModelId}...");
+            // await app.TestDownload(aliasOrModelId);
         }
-        await app.TestOpenAIIntegration("qwen2.5-0.5b");
-
-        Console.WriteLine(new string('=', 80)); // Separator for clarity
-        Console.WriteLine("Testing OpenAI integration (test again service is started)...");
-        await app.TestOpenAIIntegration("qwen2.5-0.5b");
-
-        Console.WriteLine(new string('=', 80)); // Separator for clarity
-        Console.WriteLine("Testing service operations");
-        await app.TestService(); // Call the instance method
-
-        Console.WriteLine(new string('=', 80)); // Separator for clarity
-        Console.WriteLine("Testing model (un)loading");
-        await app.TestModelLoadUnload("qwen2.5-0.5b"); // Call the instance method
-
-        Console.WriteLine(new string('=', 80)); // Separator for clarity
-        Console.WriteLine("Testing downloading");
-        await app.TestDownload("qwen2.5-0.5b"); // Call the instance method
-
-        Console.WriteLine("Press any key to exit...");
-        Console.ReadKey(true);
     }
 
     private async Task TestCacheOperations()
@@ -77,10 +77,11 @@ private async Task TestService()
         Console.WriteLine($"Service Uri: {manager.ServiceUri}");
         Console.WriteLine($"Endpoint {manager.Endpoint}");
         Console.WriteLine($"ApiKey: {manager.ApiKey}");
-        // stop the service
-        await manager.StopServiceAsync();
-        Console.WriteLine($"Service stopped");
-        Console.WriteLine($"Service running (should be false): {manager.IsServiceRunning}");
+        // don't stop for now. the catalog api doesn't return register EP models afer stop, start and model list
+        // // stop the service
+        // await manager.StopServiceAsync();
+        // Console.WriteLine($"Service stopped");
+        // Console.WriteLine($"Service running (should be false): {manager.IsServiceRunning}");
     }
 
     private async Task TestCatalog()
@@ -106,7 +107,7 @@ private async Task TestOpenAIIntegration(string aliasOrModelId)
 
         var chatClient = client.GetChatClient(model?.ModelId);
 
-        CollectionResult<StreamingChatCompletionUpdate> completionUpdates = chatClient.CompleteChatStreaming("Why is the sky blue'");
+        CollectionResult<StreamingChatCompletionUpdate> completionUpdates = chatClient.CompleteChatStreaming("Why is the sky blue?");
 
         Console.Write($"[ASSISTANT]: ");
         foreach (StreamingChatCompletionUpdate completionUpdate in completionUpdates)
@@ -121,14 +122,25 @@ private async Task TestOpenAIIntegration(string aliasOrModelId)
     private async Task TestModelLoadUnload(string aliasOrModelId)
     {
         using var manager = new FoundryLocalManager();
-        // Load a model
+
+        // Load the model
         var model = await manager.LoadModelAsync(aliasOrModelId);
         Console.WriteLine($"Loaded model: {model.Alias} ({model.ModelId})");
-        // Unload the model
+
+        // Attempt to unload without forcing
         await manager.UnloadModelAsync(aliasOrModelId);
-        Console.WriteLine($"Unloaded model: {model.Alias} ({model.ModelId})");
+        var stillLoaded = (await manager.ListLoadedModelsAsync())
+            .Any(m => m.ModelId == model.ModelId);
+        Console.WriteLine($"Model still loaded (expected: True): {stillLoaded}");
+
+        // Force unload
+        await manager.UnloadModelAsync(aliasOrModelId, force: true);
+        var unloaded = (await manager.ListLoadedModelsAsync())
+            .Any(m => m.ModelId == model.ModelId);
+        Console.WriteLine($"Model unloaded (expected: True): {!unloaded}");
     }
 
+
     private async Task TestDownload(string aliasOrModelId)
     {
         using var manager = new FoundryLocalManager();