Rename TextContent to GeneratedTextContent; Undo package upgrades; Fi…

…x typos

examples/104-dotnet-custom-LLM/Program.cs

@@ -3,7 +3,6 @@
 using System.Runtime.CompilerServices;
 using Microsoft.KernelMemory;
 using Microsoft.KernelMemory.AI;
-using Microsoft.KernelMemory.Models;
 
 public static class Program
 {
@@ -69,7 +68,7 @@ public IReadOnlyList<string> GetTokens(string text)
     }
 
     /// <inheritdoc />
-    public async IAsyncEnumerable<TextContent> GenerateTextAsync(
+    public async IAsyncEnumerable<GeneratedTextContent> GenerateTextAsync(
         string prompt,
         TextGenerationOptions options,
         [EnumeratorCancellation] CancellationToken cancellationToken = default)
@@ -79,6 +78,6 @@ public async IAsyncEnumerable<TextContent> GenerateTextAsync(
         // Remove this
         await Task.Delay(0, cancellationToken).ConfigureAwait(false);
 
-        yield return new("some text");
+        yield return "some text";
     }
 }

extensions/Anthropic/AnthropicTextGeneration.cs

@@ -10,7 +10,6 @@
 using Microsoft.KernelMemory.AI.Anthropic.Client;
 using Microsoft.KernelMemory.Context;
 using Microsoft.KernelMemory.Diagnostics;
-using Microsoft.KernelMemory.Models;
 
 namespace Microsoft.KernelMemory.AI.Anthropic;
 
@@ -98,7 +97,7 @@ public IReadOnlyList<string> GetTokens(string text)
     }
 
     /// <inheritdoc />
-    public async IAsyncEnumerable<TextContent> GenerateTextAsync(
+    public async IAsyncEnumerable<GeneratedTextContent> GenerateTextAsync(
         string prompt,
         TextGenerationOptions options,
         [EnumeratorCancellation] CancellationToken cancellationToken = default)
@@ -121,7 +120,7 @@ public async IAsyncEnumerable<TextContent> GenerateTextAsync(
             switch (response)
             {
                 case ContentBlockDelta blockDelta:
-                    yield return new(blockDelta.Delta.Text);
+                    yield return blockDelta.Delta.Text;
                     break;
 
                 default:

extensions/AzureOpenAI/AzureOpenAI/AzureOpenAITextGenerator.cs

@@ -11,7 +11,6 @@
 using Microsoft.Extensions.Logging;
 using Microsoft.KernelMemory.AI.AzureOpenAI.Internals;
 using Microsoft.KernelMemory.Diagnostics;
-using Microsoft.KernelMemory.Models;
 using Microsoft.SemanticKernel;
 using Microsoft.SemanticKernel.Connectors.AzureOpenAI;
 using OpenAI.Chat;
@@ -120,7 +119,7 @@ public IReadOnlyList<string> GetTokens(string text)
     }
 
     /// <inheritdoc/>
-    public async IAsyncEnumerable<Models.TextContent> GenerateTextAsync(
+    public async IAsyncEnumerable<GeneratedTextContent> GenerateTextAsync(
         string prompt,
         TextGenerationOptions options,
         [EnumeratorCancellation] CancellationToken cancellationToken = default)
@@ -157,15 +156,15 @@ public IReadOnlyList<string> GetTokens(string text)
             throw new AzureOpenAIException(e.Message, e, isTransient: e.StatusCode.IsTransientError());
         }
 
-        await foreach (var x in result.WithCancellation(cancellationToken))
+        await foreach (StreamingTextContent x in result.WithCancellation(cancellationToken))
         {
             TokenUsage? tokenUsage = null;
 
-            // The last message in the chunk has the usage metadata.
+            // The last message includes tokens usage metadata.
             // https://platform.openai.com/docs/api-reference/chat/create#chat-create-stream_options
-            if (x.Metadata?["Usage"] is ChatTokenUsage { } usage)
+            if (x.Metadata?["Usage"] is ChatTokenUsage usage)
             {
-                this._log.LogTrace("Usage report: input tokens {0}, output tokens {1}, output reasoning tokens {2}",
+                this._log.LogTrace("Usage report: input tokens: {InputTokenCount}, output tokens: {OutputTokenCount}, output reasoning tokens: {ReasoningTokenCount}",
                     usage.InputTokenCount, usage.OutputTokenCount, usage.OutputTokenDetails?.ReasoningTokenCount ?? 0);
 
                 tokenUsage = new TokenUsage
@@ -181,7 +180,7 @@ public IReadOnlyList<string> GetTokens(string text)
             }
 
             // NOTE: as stated at https://platform.openai.com/docs/api-reference/chat/streaming#chat/streaming-choices,
-            // The Choice can also be empty for the last chunk if we set stream_options: { "include_usage": true} to get token counts, so it is possible that
+            // the Choice can also be empty for the last chunk if we set stream_options: { "include_usage": true} to get token counts, so it is possible that
             // x.Text is null, but tokenUsage is not (token usage statistics for the entire request are included in the last chunk).
             if (x.Text is null && tokenUsage is null) { continue; }
 

extensions/LlamaSharp/LlamaSharp/LlamaSharpTextGenerator.cs

@@ -4,16 +4,13 @@
 using System.Collections.Generic;
 using System.Diagnostics.CodeAnalysis;
 using System.Linq;
-using System.Runtime.CompilerServices;
 using System.Threading;
-using System.Threading.Tasks;
 using LLama;
 using LLama.Common;
 using LLama.Native;
 using LLama.Sampling;
 using Microsoft.Extensions.Logging;
 using Microsoft.KernelMemory.Diagnostics;
-using Microsoft.KernelMemory.Models;
 
 namespace Microsoft.KernelMemory.AI.LlamaSharp;
 
@@ -77,18 +74,18 @@ public IReadOnlyList<string> GetTokens(string text)
     }
 
     /// <inheritdoc/>
-    public async IAsyncEnumerable<TextContent> GenerateTextAsync(
+    public IAsyncEnumerable<GeneratedTextContent> GenerateTextAsync(
         string prompt,
         TextGenerationOptions options,
-        [EnumeratorCancellation] CancellationToken cancellationToken = default)
+        CancellationToken cancellationToken = default)
     {
         var executor = new InteractiveExecutor(this._context);
 
         var logitBias = options.TokenSelectionBiases.Count > 0
             ? options.TokenSelectionBiases.ToDictionary(pair => (LLamaToken)pair.Key, pair => pair.Value)
             : [];
 
-        var samplingPipeline = new DefaultSamplingPipeline()
+        var samplingPipeline = new DefaultSamplingPipeline
         {
             Temperature = (float)options.Temperature,
             TopP = (float)options.NucleusSampling,
@@ -106,12 +103,7 @@ public async IAsyncEnumerable<TextContent> GenerateTextAsync(
         };
 
         this._log.LogTrace("Generating text, temperature {0}, max tokens {1}", samplingPipeline.Temperature, settings.MaxTokens);
-
-        IAsyncEnumerable<string> streamingResponse = executor.InferAsync(prompt, settings, cancellationToken);
-        await foreach (var x in streamingResponse.ConfigureAwait(false))
-        {
-            yield return new(x);
-        }
+        return executor.InferAsync(prompt, settings, cancellationToken).Select(x => new GeneratedTextContent(x));
     }
 
     /// <inheritdoc/>

extensions/ONNX/Onnx/OnnxTextGenerator.cs

@@ -10,7 +10,6 @@
 using System.Threading.Tasks;
 using Microsoft.Extensions.Logging;
 using Microsoft.KernelMemory.Diagnostics;
-using Microsoft.KernelMemory.Models;
 using Microsoft.ML.OnnxRuntimeGenAI;
 using static Microsoft.KernelMemory.OnnxConfig;
 
@@ -86,7 +85,7 @@ public OnnxTextGenerator(
     }
 
     /// <inheritdoc/>
-    public async IAsyncEnumerable<TextContent> GenerateTextAsync(
+    public async IAsyncEnumerable<GeneratedTextContent> GenerateTextAsync(
         string prompt,
         TextGenerationOptions? options = null,
         [EnumeratorCancellation] CancellationToken cancellationToken = default)
@@ -163,7 +162,7 @@ public async IAsyncEnumerable<TextContent> GenerateTextAsync(
                 if (outputTokens.Count > 0 && this._tokenizer != null)
                 {
                     var newToken = outputTokens[^1];
-                    yield return new(this._tokenizer.Decode([newToken]));
+                    yield return this._tokenizer.Decode([newToken]);
                 }
             }
         }

extensions/Ollama/Ollama/OllamaTextGenerator.cs

@@ -9,7 +9,6 @@
 using Microsoft.Extensions.Logging;
 using Microsoft.KernelMemory.Context;
 using Microsoft.KernelMemory.Diagnostics;
-using Microsoft.KernelMemory.Models;
 using OllamaSharp;
 using OllamaSharp.Models;
 
@@ -92,7 +91,7 @@ public IReadOnlyList<string> GetTokens(string text)
         return this._textTokenizer.GetTokens(text);
     }
 
-    public async IAsyncEnumerable<TextContent> GenerateTextAsync(
+    public async IAsyncEnumerable<GeneratedTextContent> GenerateTextAsync(
         string prompt,
         TextGenerationOptions options,
         [EnumeratorCancellation] CancellationToken cancellationToken = default)
@@ -147,7 +146,7 @@ public async IAsyncEnumerable<TextContent> GenerateTextAsync(
         IAsyncEnumerable<string?> stream = chat.SendAsync(prompt, cancellationToken);
         await foreach (string? token in stream.ConfigureAwait(false))
         {
-            if (token != null) { yield return new(token); }
+            if (token != null) { yield return token; }
         }
     }
 }

extensions/OpenAI/OpenAI/OpenAITextGenerator.cs

@@ -10,7 +10,6 @@
 using Microsoft.Extensions.Logging;
 using Microsoft.KernelMemory.AI.OpenAI.Internals;
 using Microsoft.KernelMemory.Diagnostics;
-using Microsoft.KernelMemory.Models;
 using Microsoft.SemanticKernel;
 using Microsoft.SemanticKernel.Connectors.OpenAI;
 using OpenAI;
@@ -125,7 +124,7 @@ public IReadOnlyList<string> GetTokens(string text)
     }
 
     /// <inheritdoc/>
-    public async IAsyncEnumerable<Models.TextContent> GenerateTextAsync(
+    public async IAsyncEnumerable<GeneratedTextContent> GenerateTextAsync(
         string prompt,
         TextGenerationOptions options,
         [EnumeratorCancellation] CancellationToken cancellationToken = default)

service/Abstractions/AI/ITextGenerator.cs

@@ -2,7 +2,6 @@
 
 using System.Collections.Generic;
 using System.Threading;
-using Microsoft.KernelMemory.Models;
 
 namespace Microsoft.KernelMemory.AI;
 
@@ -20,7 +19,7 @@ public interface ITextGenerator : ITextTokenizer
     /// <param name="options">Options for the LLM request</param>
     /// <param name="cancellationToken">Async task cancellation token</param>
     /// <returns>Text generated, returned as a stream of strings/tokens</returns>
-    public IAsyncEnumerable<TextContent> GenerateTextAsync(
+    public IAsyncEnumerable<GeneratedTextContent> GenerateTextAsync(
         string prompt,
         TextGenerationOptions options,
         CancellationToken cancellationToken = default);

service/Abstractions/Models/GeneratedTextContent.cs

@@ -0,0 +1,31 @@
+// Copyright (c) Microsoft. All rights reserved.
+
+namespace Microsoft.KernelMemory;
+
+public class GeneratedTextContent
+{
+    public string Text { get; set; }
+
+    public TokenUsage? TokenUsage { get; set; }
+
+    public GeneratedTextContent(string text, TokenUsage? tokenUsage = null)
+    {
+        this.Text = text;
+        this.TokenUsage = tokenUsage;
+    }
+
+    /// <inheritdoc/>
+    public override string ToString()
+    {
+        return this.Text;
+    }
+
+    /// <summary>
+    /// Convert a string to an instance of GeneratedTextContent
+    /// </summary>
+    /// <param name="text">Text content</param>
+    public static implicit operator GeneratedTextContent(string text)
+    {
+        return new GeneratedTextContent(text);
+    }
+}

service/Abstractions/Models/MemoryAnswer.cs

@@ -6,7 +6,6 @@
 using System.Text;
 using System.Text.Json;
 using System.Text.Json.Serialization;
-using Microsoft.KernelMemory.Models;
 
 namespace Microsoft.KernelMemory;
 
@@ -53,7 +52,7 @@ public class MemoryAnswer
     /// <remarks>Not all the models and text generators return token usage information.</remarks>
     [JsonPropertyName("tokenUsage")]
     [JsonPropertyOrder(11)]
-    public IList<TokenUsage> TokenUsage { get; set; } = [];
+    public List<TokenUsage> TokenUsage { get; set; } = [];
 
     /// <summary>
     /// List of the relevant sources used to produce the answer.

service/Abstractions/Models/TokenUsage.cs

@@ -3,7 +3,7 @@
 using System;
 using System.Text.Json.Serialization;
 
-namespace Microsoft.KernelMemory.Models;
+namespace Microsoft.KernelMemory;
 
 /// <summary>
 /// Represents the usage of tokens in a request and response cycle.
@@ -22,13 +22,13 @@ public class TokenUsage
     /// The number of tokens in the request message input, spanning all message content items, measured by the tokenizer.
     /// </summary>
     [JsonPropertyName("tokenizer_tokens_in")]
-    public int TokeninzerTokensIn { get; set; }
+    public int TokenizerTokensIn { get; set; }
 
     /// <summary>
     /// The combined number of output tokens in the generated completion, measured by the tokenizer.
     /// </summary>
     [JsonPropertyName("tokenizer_tokens_out")]
-    public int TokeninzerTokensOut { get; set; }
+    public int TokenizerTokensOut { get; set; }
 
     /// <summary>
     /// The number of tokens in the request message input, spanning all message content items, measured by the service.

service/Core/AI/NoTextGenerator.cs

@@ -5,7 +5,6 @@
 using System.Threading;
 using Microsoft.Extensions.Logging;
 using Microsoft.KernelMemory.Diagnostics;
-using Microsoft.KernelMemory.Models;
 
 namespace Microsoft.KernelMemory.AI;
 
@@ -38,7 +37,7 @@ public IReadOnlyList<string> GetTokens(string text)
     }
 
     /// <inheritdoc />
-    public IAsyncEnumerable<TextContent> GenerateTextAsync(string prompt, TextGenerationOptions options, CancellationToken cancellationToken = default)
+    public IAsyncEnumerable<GeneratedTextContent> GenerateTextAsync(string prompt, TextGenerationOptions options, CancellationToken cancellationToken = default)
     {
         throw this.Error();
     }

service/Core/Search/AnswerGenerator.cs

@@ -11,7 +11,6 @@
 using Microsoft.KernelMemory.AI;
 using Microsoft.KernelMemory.Context;
 using Microsoft.KernelMemory.Diagnostics;
-using Microsoft.KernelMemory.Models;
 using Microsoft.KernelMemory.Prompts;
 
 namespace Microsoft.KernelMemory.Search;
@@ -76,7 +75,7 @@ internal async IAsyncEnumerable<MemoryAnswer> GenerateAnswerAsync(
         {
             Timestamp = DateTimeOffset.UtcNow,
             ModelType = Constants.ModelType.TextGeneration,
-            TokeninzerTokensIn = this._textGenerator.CountTokens(prompt)
+            TokenizerTokensIn = this._textGenerator.CountTokens(prompt)
         };
 
         result.AskResult.TokenUsage.Add(tokenUsage);
@@ -100,7 +99,7 @@ internal async IAsyncEnumerable<MemoryAnswer> GenerateAnswerAsync(
 
         // Finalize the answer, checking if it's empty
         result.AskResult.Result = completeAnswer.ToString();
-        tokenUsage.TokeninzerTokensOut = this._textGenerator.CountTokens(result.AskResult.Result);
+        tokenUsage.TokenizerTokensOut = this._textGenerator.CountTokens(result.AskResult.Result);
 
         if (string.IsNullOrWhiteSpace(result.AskResult.Result)
             || ValueIsEquivalentTo(result.AskResult.Result, this._config.EmptyAnswer))
@@ -138,7 +137,7 @@ private string CreatePrompt(string question, string facts, IContext? context)
         return prompt;
     }
 
-    private IAsyncEnumerable<TextContent> GenerateAnswerTokensAsync(string prompt, IContext? context, CancellationToken cancellationToken)
+    private IAsyncEnumerable<GeneratedTextContent> GenerateAnswerTokensAsync(string prompt, IContext? context, CancellationToken cancellationToken)
     {
         int maxTokens = context.GetCustomRagMaxTokensOrDefault(this._config.AnswerTokens);
         double temperature = context.GetCustomRagTemperatureOrDefault(this._config.Temperature);

service/Core/SemanticKernel/SemanticKernelTextGenerator.cs

@@ -52,7 +52,7 @@ public SemanticKernelTextGenerator(
     }
 
     /// <inheritdoc />
-    public async IAsyncEnumerable<Models.TextContent> GenerateTextAsync(
+    public async IAsyncEnumerable<GeneratedTextContent> GenerateTextAsync(
         string prompt,
         TextGenerationOptions options,
         [EnumeratorCancellation] CancellationToken cancellationToken = default)