Implementation:Microsoft Semantic kernel HuggingFace TextGeneration TestData

Knowledge Sources	Microsoft_Semantic_kernel
Domains	HuggingFace, Text_Generation, Unit_Testing, Test_Data
Last Updated	2026-02-11 00:00 GMT

Overview

Concrete mock HuggingFace text generation API response JSON file used in unit tests, provided by the Connectors.HuggingFace.UnitTests project.

Description

textgeneration_test_response.json is a test fixture file containing a mock response from the HuggingFace Inference API for the text generation task. At 913 lines, the file consists of a JSON array with a single object containing:

• generated_text -- A string with the generated text continuation (in this case, an essay about the difference between Data Science and AI Engineering, generated for 150 tokens).
• details -- A metadata object containing:
  • finish_reason -- "length" (indicating generation stopped at the token limit)
  • generated_tokens -- 150 (the number of tokens produced)
  • seed -- null (no seed was specified)
  • prefill -- An empty array (no prefill tokens)
  • tokens -- A large array of token objects, each containing id (integer token ID), text (string token text), logprob (float log probability), and special (boolean indicating if this is a special token)

The tokens array makes up the bulk of the file (913 lines), providing detailed per-token metadata that enables unit tests to verify the connector's token-level parsing capabilities.

Usage

This file is loaded at test time through the HuggingFaceTestHelper.GetTestResponse() helper method. It is used by HuggingFaceTextGenerationTests to mock HTTP responses for text generation endpoint testing. Developers modify this file when the HuggingFace text generation API response format changes or when new token-level test scenarios are needed.

Code Reference

Source Location

• Repository: Microsoft_Semantic_kernel
• File: dotnet/src/Connectors/Connectors.HuggingFace.UnitTests/TestData/textgeneration_test_response.json
• Lines: 1-913

Signature

[
  {
    "generated_text": "Write about the difference between Data Science and AI Engineering.\n\nData Science and AI Engineering are two interconnected fields...",
    "details": {
      "finish_reason": "length",
      "generated_tokens": 150,
      "seed": null,
      "prefill": [],
      "tokens": [
        {
          "id": 13,
          "text": "\n",
          "logprob": -0.11578369,
          "special": false
        },
        {
          "id": 13,
          "text": "\n",
          "logprob": -0.15930176,
          "special": false
        },
        {
          "id": 1333,
          "text": "Data",
          "logprob": -0.25341797,
          "special": false
        }
      ]
    }
  }
]

Import

// Via the HuggingFaceTestHelper utility class:
using SemanticKernel.Connectors.HuggingFace.UnitTests;

// In test methods:
string responseJson = HuggingFaceTestHelper.GetTestResponse("textgeneration_test_response.json");

// Used to populate mock HTTP response:
this._messageHandlerStub.ResponseToReturn.Content =
    new StringContent(responseJson);

I/O Contract

Inputs

Name	Type	Required	Description
N/A (static file)	JSON file	N/A	This is a static test fixture; it has no runtime inputs

Outputs

Name	Type	Description
[].generated_text	string	The full generated text continuation from the model
[].details.finish_reason	string	Reason generation stopped ("length", "eos_token", or "stop_sequence")
[].details.generated_tokens	integer	Total number of tokens generated
[].details.seed	integer or null	Random seed used for generation, null if not specified
[].details.prefill	array	Array of prefill tokens (empty in this test data)
[].details.tokens	array	Array of token objects with id, text, logprob, and special fields
[].details.tokens[].id	integer	Token ID from the model's vocabulary
[].details.tokens[].text	string	Decoded text representation of the token
[].details.tokens[].logprob	float	Log probability of the token
[].details.tokens[].special	boolean	Whether this is a special token (BOS, EOS, PAD, etc.)

Usage Examples

Loading in HuggingFace Text Generation Tests

using System.Net.Http;
using Xunit;

public class HuggingFaceTextGenerationTests
{
    [Fact]
    public async Task ItReturnsTextGenerationCorrectly()
    {
        // Arrange
        this._messageHandlerStub.ResponseToReturn.Content =
            new StringContent(
                HuggingFaceTestHelper.GetTestResponse(
                    "textgeneration_test_response.json"));

        // Act
        var results = await this._textGeneration
            .GetTextContentsAsync("Write about the difference between Data Science and AI Engineering.");

        // Assert
        Assert.NotNull(results);
        Assert.Single(results);
        Assert.Contains("Data Science", results[0].Text);
    }
}

Related Pages

• Environment:Microsoft_Semantic_kernel_DotNet_SDK_Environment
• Microsoft_Semantic_kernel_HuggingFace_Embeddings_TestData -- Mock HuggingFace embeddings response in the same test project
• Microsoft_Semantic_kernel_Google_Embeddings_TestData -- Mock Google AI API test data (different service, similar testing pattern)
• Microsoft_Semantic_kernel_MistralAI_Embeddings_TestData -- Mock MistralAI API test data (different service, similar testing pattern)