### NativeTensorArray Usage Example

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.NativeTensorArray.html

Demonstrates allocating native memory, setting and getting values, copying to a managed array, and disposing of the native memory. Ensure Dispose() is called to release resources.

```csharp
// Allocate native memory for 1024 float elements
var tensorArray = new NativeTensorArray(1024);
// Set and get values
tensorArray.Set(0, 1.0f);
float value = tensorArray.Get<float>(0);
// Copy to managed array
float[] floatArray = tensorArray.ToArray<float>(1024, 0);
// Dispose of the native memory
tensorArray.Dispose();
```

--------------------------------

### Get Tokens (Output Collection)

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tokenization.IEncoding.html

Populates a provided collection with tokens. Returns the number of available tokens.

```csharp
int GetTokens(ICollection<Token> output)
```

--------------------------------

### Create Runtime Model from Asset

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/manual/import-a-model-file.html

Use `ModelLoader.Load` to create a runtime `Model` object from an imported `ModelAsset`. This is typically done in `Start()` or `Awake()`.

```csharp
using UnityEngine;
using Unity.InferenceEngine;

public class CreateRuntimeModel : MonoBehaviour
{
    public ModelAsset modelAsset;
    Model runtimeModel;

    void Start()
    {
        runtimeModel = ModelLoader.Load(modelAsset);
    }
}
```

--------------------------------

### Implementing IONNXMetadataImportCallbackReceiver

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.IONNXMetadataImportCallbackReceiver.html

Example of a custom class that implements IONNXMetadataImportCallbackReceiver to capture and store ONNX model metadata during the import process.

```csharp
public class MyMetadataCallbackHandler : IONNXMetadataImportCallbackReceiver
{
    public ONNXModelMetadata Metadata { get; private set; }

    // Callback for metadata import. Member Metadata is of type ONNXModelMetadata.
    public void OnMetadataImported(AssetImportContext ctx, ONNXModelMetadata metadata)
    {
        Metadata = metadata;
    }
}
```

--------------------------------

### Implement Custom Normalizer

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/manual/tokenizer.html

Implement the INormalizer interface to create a custom normalization logic. This example reverses input characters and adds a prefix.

```csharp
using System.Text;
using Unity.InferenceEngine.Tokenization;
using Unity.InferenceEngine.Tokenization.Normalizers;

class ReversePrefixNormalizer : INormalizer
{
    readonly string m_Prefix;

    public ReversePrefixNormalizer(string prefix = null)
    {
        m_Prefix = prefix ?? string.Empty;
    }

    public SubString Normalize(SubString input)
    {
        var sb = new StringBuilder()
            .Append(m_Prefix);

        for(var i = input.Length - 1; i >= 0; --i)
            sb.Append(input[i]);

        return sb.ToString();
    }
}
```

--------------------------------

### Get Encodings (Sequential)

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tokenization.IEncoding.html

Provides an enumerable sequence of IEncoding instances, starting with the main encoding and followed by any overflowing sequences.

```csharp
IEnumerable<IEncoding> GetEncodings()
```

--------------------------------

### Manual Tokenizer Initialization Sample

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/manual/tokenizer.html

Demonstrates the manual creation and configuration of a tokenizer, including normalization, pre-tokenization, truncation, post-processing, and decoding. This is useful for understanding the underlying components of the tokenization pipeline.

```csharp
using System;
using System.Collections.Generic;
using System.Linq;
using Unity.InferenceEngine;
using Unity.InferenceEngine.Tokenization;
using Unity.InferenceEngine.Tokenization.Decoders;
using Unity.InferenceEngine.Tokenization.Mappers;
using Unity.InferenceEngine.Tokenization.Normalizers;
using Unity.InferenceEngine.Tokenization.Padding;
using Unity.InferenceEngine.Tokenization.PostProcessors;
using Unity.InferenceEngine.Tokenization.PostProcessors.Templating;
using Unity.InferenceEngine.Tokenization.PreTokenizers;
using Unity.InferenceEngine.Tokenization.Truncators;
using UnityEngine;

class TokenizerSample : MonoBehaviour
{
    static Tensor<int> Encode(ITokenizer tokenizer, string input)
    {
        // Generates the sequence
        var encoding = tokenizer.Encode(input);

        // Then you can use the encoding to generate your tensors.

        // Gets this ids
        // Other masks or available, like:
        // - attention
        // - type ids
        // - special mask.
        int[] ids = encoding.GetIds().ToArray();

        // Create a 3D tensor shape
        TensorShape shape = new TensorShape(1, 1, ids.Length);

        // Create a new tensor from the array
        return new Tensor<int>(shape, ids);
    }

    static string Decode(ITokenizer tokenizer, Tensor<int> tensor)
    {
        var ids = tensor.DownloadToArray();
        return tokenizer.Decode(ids);
    }

    static Dictionary<string, int> BuildVocabulary()
    {
        // This stub method returns a legitimate string to id mapping for the tokenizer.
        // It is usually built from a large configuration JSON file.
        return new Dictionary<string, int>();
    }

    static TokenConfiguration[] GetAddedTokens()
    {
        // This stub method returns a legitimate collection of token configuration.
        // Token configuration is the Hugging Face equivalent of added token.
        return Array.Empty<TokenConfiguration>();
    }

    /// This sample initializes a tokenizer based on All MiniLM L6 v2.
    public ITokenizer CreateTokenizer()
    {
        var vocabulary = BuildVocabulary();
        var addedTokens = GetAddedTokens();

        // Central step of the tokenizer
        var mapper = new WordPieceMapper(vocabulary, "[UNK]", "##", 100);


        // Preliminary steps of the tokenization:
        // - normalization (transforms the input string)
        // - pre-tokenization (splits the input string)

        var normalizer = new BertNormalizer(
            cleanText: true,
            handleCjkChars: true,
            stripAccents: null,
            lowerCase: true);

        var preTokenizer = new BertPreTokenizer();


        // Final steps of tokenization:
        // - truncation (splits the token sequences)
        // - post-processing (decorates the token sequences)
        // - padding (adds tokens to match a sequence size).

        var truncator = new LongestFirstTruncator(new RightDirectionRangeGenerator(), 128, 0);

        var clsId = addedTokens.Where(tc => tc.Value == "[CLS]").Select(tc => tc.Id).FirstOrDefault();
        var sepId = addedTokens.Where(tc => tc.Value == "[SEP]").Select(tc => tc.Id).FirstOrDefault();
        var padId = addedTokens.Where(tc => tc.Value == "[PAD]").Select(tc => tc.Id).FirstOrDefault();

        var postProcessor = new TemplatePostProcessor(
          new(Template.Parse("[CLS]:0 $A:0 [SEP]:0")),
          new(Template.Parse("[CLS]:0 $A:0 [SEP]:0 $B:1 [SEP]:1")),
          new (string, int)[] { ("[CLS]", clsId), ("[SEP]", sepId) });

        var padding = new RightPadding(
          new FixedPaddingSizeProvider(128),
          new Token(padId, "[PAD]"));


        // Decoding.

        var decoder = new WordPieceDecoder("##", true);


        // Creates the tokenizer from all the components
        // initialized above.

        return new Tokenizer(
            mapper,
            normalizer: normalizer,
            preTokenizer: preTokenizer,
            truncator: truncator,
            postProcessor: postProcessor,
            paddingProcessor: padding,
            decoder: decoder,
            addedVocabulary: addedTokens);
    }
}
```

--------------------------------

### Create and Manage Tensor<T> Instance

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tensor-1.html

Demonstrates creating a Tensor<float> with specified dimensions and data, and how to obtain a CPU-accessible copy using ReadbackAndClone(). It also shows proper resource management by disposing of tensors.

```csharp
 // Create a tensor
         m_Tensor = new Tensor<float>(new TensorShape(2, 3), new[] { 1.0f, 2.0f, 3.0f, 4.0f, 5.0f, 6.0f });
         // Alternatively with `using` so you don't need to call `Dispose()` when you are done with the tensor
         using var m_OtherTensor = new Tensor<float>(new TensorShape(2, 3), new[] { 1.0f, 2.0f, 3.0f, 4.0f, 5.0f, 6.0f });

         // Get a CPU-accessible clone of the tensor. This copy owns its own data.
         Tensor<float> cpuCopyTensor = m_Tensor.ReadbackAndClone() as Tensor<float>;

         // Release memory.
         cpuCopyTensor.Dispose();
         m_Tensor.Dispose();
```

--------------------------------

### Get Rank of TensorShape

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.TensorShape.html

Illustrates how to get the number of dimensions (rank) of a TensorShape. The rank indicates the dimensionality of the tensor.

```csharp
var shape = new TensorShape(4, 3, 2, 4);
shape.rank;  // Returns 4
```

--------------------------------

### Completing Pending Jobs and Downloading Data

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.ITensorData.html

This example shows how to ensure all pending jobs and GPU operations are completed before downloading tensor data. This is crucial for accurate CPU readback.

```csharp
cpuData.fence = job.Schedule(count, 64);
worker.Schedule(inputTensor);
cpuData.CompleteAllPendingOperations();
var data = cpuData.Download<float>(count);
```

--------------------------------

### Length(int)

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.TensorShape.html

Returns the number of elements from a given start axis to the end. Calculates the product of dimensions from the start axis through the last axis.

```APIDOC
## Length(int start)

### Description
Returns the number of elements from a given `start` axis to the end.

### Method
`public int Length(int start)`

### Parameters
- **start** (int) - The first axis to count from. Negative values count from the end.

### Returns
- **int**: The product of dimensions from `start` to the last axis.

### Remarks
This method calculates the product of dimensions from the `start` axis through the last axis. Negative `start` values count backwards from the innermost dimension. If `start` is beyond the shape's rank, returns `1`.

### Examples
Calculate the number of elements from a given axis to the end
```csharp
var shape = new TensorShape(2, 3, 4, 5);
shape.Length(1);   // Returns 60 (3 * 4 * 5)
shape.Length(-2);  // Returns 20 (4 * 5)
```
```

--------------------------------

### Create a Functional Graph from Scratch

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.FunctionalGraph.html

Build a computation graph with inputs, operations, and outputs, then compile it into an optimized runtime model.

```csharp
var graph = new FunctionalGraph();
var x = graph.AddInput<float>(new TensorShape(6), "input_x");
var y = graph.AddInput<float>(new TensorShape(6), "input_y");
var prod = x * y;
var reduce = Functional.ReduceSum(prod, dim: 0, keepdim: false);
graph.AddOutput(reduce, "output");
Model model = graph.Compile();
```

--------------------------------

### Create TensorShapes and Access Properties

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.TensorShape.html

Demonstrates how to create TensorShape instances using constructors or arrays and how to access properties like rank, length, and individual dimension values.

```csharp
using Unity.InferenceEngine;
// Create TensorShape
var shape = new TensorShape(3, 4);       // Shape: (3, 4), rank: 2
// Create shape from an array
var shape2 = new TensorShape(new[] { 2, 3, 4, 5 });
// Shape: (2, 3, 4, 5), rank: 4
// Shape properties
shape.rank;    // Returns 2
shape.length;  // Returns 12 (3 * 4)
shape[0];      // Returns 3
shape[-1];     // Returns 4 (last dimension)
```

--------------------------------

### Calculate Length from Axis to End

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.TensorShape.html

Compute the number of elements from a specified axis to the end of the shape. Negative values for 'start' count from the end. If 'start' is out of bounds, it returns 1.

```csharp
var shape = new TensorShape(2, 3, 4, 5);
shape.Length(1);   // Returns 60 (3 * 4 * 5)
shape.Length(-2);  // Returns 20 (4 * 5)
```

--------------------------------

### Build and Execute Command Buffer Per Frame

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/manual/use-command-buffer.html

Build the command buffer once and update inputs every frame for reduced scheduling time. This example demonstrates setting up a worker, scheduling it, and executing the command buffer in Update.

```csharp
Tensor<float> input0 = new Tensor<float>(new TensorShape(1));
Model model;
CommandBuffer cb;
Worker worker;

void Start()
{
    worker = new Worker(model, BackendType.GPUCompute);

    worker.SetInput("input0", input0);

    cb = new CommandBuffer();
    cb.ScheduleWorker(worker);
}

void Update()
{
    // modify input0
    input0.Upload(new float[] { Time.deltaTime });
    Graphics.ExecuteCommandBuffer(cb);
}

```

--------------------------------

### Get Tensor Output with PeekOutput

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/manual/get-the-output.html

Use PeekOutput to get a reference to the output tensor. The Sentis worker owns this reference, so you don't need to dispose of it. Be aware that the output can be overwritten if Schedule is called again.

```csharp
worker.Schedule(inputTensor);
Tensor<float> outputTensor = worker.PeekOutput() as Tensor<float>;
```

--------------------------------

### SplitDelimiterIsolate.Instance

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tokenization.SplitDelimiterBehaviors.SplitDelimiterIsolate.html

Gets a shared instance of the SplitDelimiterIsolate class.

```APIDOC
## Property Instance

Gets a shared instance of the SplitDelimiterIsolate.

### Declaration
```csharp
public static SplitDelimiterIsolate Instance { get; }
```

### Property Value
Type | Description
---|---
SplitDelimiterIsolate |
```

--------------------------------

### BpeMapper Constructor

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tokenization.Mappers.BpeMapper.html

Initializes a new instance of the BpeMapper class with a vocabulary, optional merge pairs, and options.

```APIDOC
## BpeMapper(IReadOnlyDictionary<string, int>, IEnumerable<MergePair>, BpeMapperOptions)

### Description
Converts a substring into a sequence of Token instances using the Byte-Pair Encoding strategy.

### Parameters
- **vocabulary** (IReadOnlyDictionary<string, int>) - Required - The map associating token string representation with their ids.
- **merges** (IEnumerable<MergePair>) - Optional - The list of mergeable token pairs, ordered by priority.
- **options** (BpeMapperOptions) - Optional - See BpeMapperOptions
```

--------------------------------

### TensorShape Constructors and Properties

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.TensorShape.html

Demonstrates how to create TensorShape instances using constructors and arrays, and how to access its rank, length, and individual dimension sizes.

```APIDOC
## Struct TensorShape

Represents the shape of a tensor.

### Remarks
A `TensorShape` describes the shape of a Tensor or FunctionalTensor.
`TensorShape` supports rank up to maxRank (`8`).
Create a `TensorShape` using one of the constructors, or by passing an array of integers.
`TensorShape` provides methods for common shape transformations such as Squeeze() and Squeeze(int), Unsqueeze(int), Flatten(), Broadcast(TensorShape), and Reduce(int, bool). These methods return new `TensorShapes` without modifying the original.
The rank property returns the number of dimensions, and the length property returns the total number of elements (the product of all dimensions).

### Examples

Create `TensorShapes` and get their properties
```csharp
using Unity.InferenceEngine;

// Create TensorShape
var shape = new TensorShape(3, 4);       // Shape: (3, 4), rank: 2

// Create shape from an array
var shape2 = new TensorShape(new[] { 2, 3, 4, 5 });
// Shape: (2, 3, 4, 5), rank: 4

// Shape properties
shape.rank;    // Returns 2
shape.length;  // Returns 12 (3 * 4)
shape[0];      // Returns 3
shape[-1];     // Returns 4 (last dimension)
```
```

--------------------------------

### TokenToId Method

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tokenization.Mappers.UnigramMapper.html

Gets the ID of the specified token.

```APIDOC
## TokenToId(string, out int)

### Description
Gets the ID of the specified `token`.

### Parameters
- **token** (string) - Required - The token we want to get the ID of.
- **id** (out int) - Required - The ID of the specified `token`.

### Returns
- **bool**: Whether the token exists.

### Exceptions
- **ArgumentNullException**: Thrown when `token` is null.
```

--------------------------------

### ByteLevelPostProcessor Constructor

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tokenization.PostProcessors.ByteLevelPostProcessor.html

Initializes a new instance of the ByteLevelPostProcessor. The trimOffsets parameter is available but not yet implemented.

```csharp
public ByteLevelPostProcessor(bool trimOffsets = false)
```

--------------------------------

### Instance Property

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tokenization.SplitDelimiterBehaviors.SplitDelimiterMergeWithNext.html

Gets a shared instance of the SplitDelimiterMergeWithNext class.

```APIDOC
### Properties 
#### Instance
Gets a shared instance of the SplitDelimiterMergeWithNext.
##### Declaration
```csharp
public static SplitDelimiterMergeWithNext Instance { get; }
```

##### Property Value
Type | Description  
---|---
SplitDelimiterMergeWithNext |

```

--------------------------------

### Instance Property

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tokenization.SplitDelimiterBehaviors.SplitDelimiterRemove.html

Gets a shared instance of the SplitDelimiterRemove class.

```APIDOC
### Properties

#### Instance

Gets a shared instance of the SplitDelimiterRemove.

##### Declaration
```csharp
public static SplitDelimiterRemove Instance { get; }
```

##### Property Value
Type | Description
---|---
SplitDelimiterRemove |
```

--------------------------------

### Initialize and Copy Tensor Output

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/manual/get-the-output.html

Demonstrates initializing a tensor with a specific shape and then using CopyOutput to populate it with model results. CopyOutputInto can also be used with tensors of sufficient capacity but different shapes.

```csharp
// The model outputs a tensor of shape (1, 10)

// CopyOutput works on empty tensors, i.e. tensors without a tensor data.
myOutputTensor = new Tensor<float>(new TensorShape(1, 10), data: null);
worker.CopyOutput("output", ref myOutputTensor);

// CopyOutputInto works on tensors of different shape as long as the dataOnBackend has large enough capacity
myOutputTensor = new Tensor<float>(new TensorShape(152));
worker.CopyOutput("output", ref myOutputTensor);
// myOutputTensor now has shape (1, 10) but still has dataOnBackend.maxCapacity == 152
```

--------------------------------

### Sequence.GetPieceHashCode()

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tokenization.PostProcessors.Templating.Sequence.html

Gets the hash code for this Piece, which is part of the Sequence.

```APIDOC
## GetPieceHashCode()

### Description
Gets the hash code of this Piece.

### Returns
- **int** - The hash code of this Piece.

### Overrides
Piece.GetPieceHashCode()
```

--------------------------------

### SequencePostProcessor Constructor

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tokenization.PostProcessors.SequencePostProcessor.html

Initializes a new instance of the SequencePostProcessor class with a collection of IPostProcessor instances.

```APIDOC
## SequencePostProcessor(params IPostProcessor[])

### Description
Initializes a new instance of the SequencePostProcessor type.

### Parameters
- **processors** (IPostProcessor[]) - Required - The IPostProcessor instances to call in sequence.
```

--------------------------------

### SpecialToken.Value Property

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tokenization.PostProcessors.Templating.SpecialToken.html

Gets the string value of the special token.

```APIDOC
## Value

### Description
The value of the token.

### Property Value
- **string** - The value of the token.
```

--------------------------------

### Copy Model Output to Screen with Style Transfer

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/manual/use-model-output.html

This script demonstrates how to use a model to transform a texture and then copy the result to the screen. Ensure `modelAsset` is set to a style transfer model and `inputImage` to a suitable texture. The texture import settings must match the model's input requirements.

```csharp
using UnityEngine;
using Unity.InferenceEngine;

public class StyleTransfer : MonoBehaviour
{
    public ModelAsset modelAsset;
    public Model runtimeModel;
    public Texture2D inputImage;
    public RenderTexture outputTexture;

    Worker worker;
    Tensor<float> inputTensor;

    void Start()
    {
        var sourceModel = ModelLoader.Load(modelAsset);
        var graph = new FunctionalGraph();
        var input = graph.AddInput(sourceModel, 0);
        var output = Functional.Forward(sourceModel, input)[0];
        // rescale output of source model
        output /= 255f;
        graph.AddOutput(output);
        var runtimeModel = graph.Compile();

        worker = new Worker(runtimeModel, BackendType.GPUCompute);
        inputTensor = new Tensor<float>(new TensorShape(1, 3, 256, 256));
    }

    void OnRenderImage(RenderTexture source, RenderTexture destination)
    {
        // Create the input tensor from the texture
        TextureConverter.ToTensor(inputImage, inputTensor, new TextureTransform());

        // Run the model and get the output as a tensor
        worker.Schedule(inputTensor);
        Tensor<float> outputTensor = worker.PeekOutput() as Tensor<float>;

        // Copy the rescaled tensor to the screen as a texture
        TextureConverter.RenderToScreen(outputTensor);
    }

    void OnDisable()
    {
        worker.Dispose();
    }
}
```

--------------------------------

### Piece.SequenceId Property

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tokenization.PostProcessors.Templating.Piece.html

Gets the sequence ID associated with this piece.

```APIDOC
### Properties
#### SequenceId
The type id of the sequence.

##### Declaration
```csharp
public int SequenceId { get; }
```

##### Property Value
Type | Description  
---|---  
int |
```

--------------------------------

### DirectionalPaddingBase Constructor (Protected)

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tokenization.Padding.DirectionalPaddingBase.html

Protected constructor for initializing DirectionalPaddingBase. It requires an IPaddingSizeProvider to determine the final padded sequence size and a Token for padding. An ArgumentNullException is thrown if paddingSizeProvider is null.

```csharp
protected DirectionalPaddingBase(IPaddingSizeProvider paddingSizeProvider, Token padToken)
{
}
```

--------------------------------

### TokenConfiguration.Equals(TokenConfiguration) Method

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tokenization.TokenConfiguration.html

Compares the current TokenConfiguration instance with another TokenConfiguration for equality.

```csharp
public bool Equals(TokenConfiguration other)
```

--------------------------------

### IdToToken Method

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tokenization.Mappers.UnigramMapper.html

Gets the token value from the specified ID.

```APIDOC
## IdToToken(int)

### Description
Gets the token value from the specified `id`.

### Parameters
- **id** (int) - Required - The ID of the requested token.

### Returns
- **string**: The token value.

### Exceptions
- **ArgumentOutOfRangeException**: Thrown when `id` is outside the vocabulary range.
```

--------------------------------

### LongestFirstStrategy.Instance

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tokenization.Truncators.Strategies.LongestFirstStrategy.html

Gets the singleton instance of the LongestFirstStrategy. This strategy is used for token truncation.

```APIDOC
## Property Instance

Gets the singleton instance of the LongestFirstStrategy.

### Declaration
```csharp
public static ITruncationStrategy Instance { get; }
```

### Property Value
Type | Description
---|---
ITruncationStrategy | 
```

--------------------------------

### Worker Constructor with Model and DeviceType

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Worker.html

Initializes a new Worker instance with a specified model and device type. The fastest available backend for the given device will be automatically chosen.

```csharp
public Worker(Model model, DeviceType deviceType)
```

--------------------------------

### HuggingFaceParser.GetDefault()

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tokenization.Parsers.HuggingFace.HuggingFaceParser.html

Gets a pre-configured HuggingFaceParser instance with built-in and HfAttribute-decorated component builders.

```APIDOC
## HuggingFaceParser.GetDefault()

### Description
Gets a parser configured with all the built-in components builders, with the additional builders decorated with HfAttribute (with priority to the built-ins.

### Returns
- **HuggingFaceParser** - A pre-configured parser.
```

--------------------------------

### RightPadding Constructors

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tokenization.Padding.RightPadding.html

Provides information on how to initialize the RightPadding class, with options for specifying padding size and token, and an optional multiple for padding length.

```APIDOC
## RightPadding Constructors

### RightPadding(IPaddingSizeProvider, Token)
Initializes a new instance of the RightPadding type.

#### Parameters
* **paddingSizeProvider** (IPaddingSizeProvider) - Required - When applying the padding, this object provides the final size of the padded sequence.
* **padToken** (Token) - Required - The token to use to pad a sequence of tokens.

### RightPadding(IPaddingSizeProvider, Token, int)
Initializes a new instance of the RightPadding type.

#### Parameters
* **paddingSizeProvider** (IPaddingSizeProvider) - Required - When applying the padding, this object provides the final size of the padded sequence.
* **padToken** (Token) - Required - The token to use to pad a sequence of tokens.
* **padToMultipleOf** (int) - Optional - Sets the pad length to the upper multiple of this value. Defaults to 1.
```

--------------------------------

### Initialize TextureTransform

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.TextureTransform.html

Create an instance of TextureTransform and set dimensions and tensor layout.

```csharp
TextureTransform settings = new TextureTransform().SetDimensions(256, 256, 4).SetTensorLayout(TensorLayout.NHWC);
```

--------------------------------

### Get<T>(int)

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.NativeTensorArray.html

Retrieves the value of an element at a specific index from the backing data.

```APIDOC
## Get<T>(int)

### Description
Returns the value of the backing data at a given index.

### Method Signature
```csharp
public T Get<T>(int index) where T : unmanaged
```

### Parameters
#### Path Parameters
- **index** (int) - The index of the element.

### Returns
- **T** - The value of the element at `index`, read as type `T`.

### Type Parameters
- **T** - The type of the element.

### Remarks
Uses `sizeof(T)` for byte-offset indexing. For correct behavior, use `T` such that `sizeof(T) == k_DataItemSize` (for example, `float` or `int`), since the backing buffer stores elements as 32-bit floats.

### Examples
```csharp
using (var tensorArray = new NativeTensorArray(64))
{
    tensorArray.Set(0, 42.0f);
    float value = tensorArray.Get<float>(0); // Returns 42.0f
}
```
```

--------------------------------

### Loading ModelAsset and Creating a Worker

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.ModelAsset.html

This script demonstrates loading a ModelAsset into a Model and then creating a Worker for inference using the GPUCompute backend. Ensure the ModelAsset is assigned in the Inspector.

```csharp
public class MyScript : MonoBehaviour
{
    public ModelAsset modelAsset;
    Model m_Model;
    Worker m_Worker;
    ...
void Start()
{
    // Load the binary asset and create a Worker
    m_Model = ModelLoader.Load(modelAsset);
    m_Worker = new Worker(m_Model, BackendType.GPUCompute);
    ...
}

}
```

--------------------------------

### UtfSub(Range)

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tokenization.SubString.html

Gets a portion of this instance, considering the unicode characters instead of chars.

```APIDOC
## UtfSub(Range)

### Description
Gets a portion of this instance, considering the unicode characters instead of chars.

### Method
```
public SubString UtfSub(Range offsets)
```

### Parameters
#### Path Parameters
- **offsets** (Range) - Required - The bounds of the subpart to extract.

### Returns
- **SubString** - A new SubString instance.
```

--------------------------------

### SubString Indexer for Character Access

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tokenization.SubString.html

Gets the character at a specific index within the SubString.

```csharp
public char this[int index] { get; }
```

--------------------------------

### Creating Static and Dynamic Tensor Shapes

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.DynamicTensorShape.html

Demonstrates how to create a static tensor shape and a dynamic tensor shape with an unknown batch size. Use -1 to indicate a dynamic dimension.

```csharp
// Static shape: (1, 3, 224, 224)
var staticShape = new DynamicTensorShape(1, 3, 224, 224);

// Dynamic shape with unknown dimension 0
var dynamicBatch = new DynamicTensorShape(-1, 3, 224, 224);

// All dimensions dynamic with rank 2
var shape = DynamicTensorShape.DynamicOfRank(2);
```

--------------------------------

### Load Model Asset and Runtime Model

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/manual/understand-sentis-workflow.html

Load a model from Unity's Assets folder and create a runtime model for inference. Ensure the model file is correctly placed in the Assets folder.

```csharp
ModelAsset modelAsset = Resources.Load("model-file-in-assets-folder") as ModelAsset;
var runtimeModel = ModelLoader.Load(modelAsset);

```

--------------------------------

### Get Tokens (ReadOnly)

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tokenization.IEncoding.html

Retrieves the list of tokens. Returns the number of available tokens.

```csharp
IReadOnlyList<Token> GetTokens()
```

--------------------------------

### Create and Access Tensor Values (CPU)

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/manual/do-basic-tensor-operations.html

Demonstrates creating a float tensor on the CPU, setting a specific value using indexers, and retrieving that value. Ensure the tensor is on the CPU and computation is complete before accessing values directly.

```csharp
var tensor = new Tensor<float>(new TensorShape(1, 2, 3));
tensor[0, 1, 2] = 5.2f; // set value at index 0 of dim0 = 1, index 1 of dim1 = 2 and index 2 of dim2 = 3

float value = tensor[0, 1, 2];
Assert.AreEqual(5.2f, value);
```

--------------------------------

### Get

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.DynamicTensorShape.html

Returns the dimension at a given axis as an integer. Dynamic dimensions return -1.

```APIDOC
## Get(int)

### Description
Returns the dimension at a given axis as an integer.

### Method
`public int Get(int axis)`

### Parameters
- **axis** (int) - The axis index (0-based, or negative for reverse indexing).

### Returns
- **int** - The dimension value at the axis, or `-1` for dynamic dimensions.

### Remarks
Dynamic dimensions return `-1`. Supports negative axis indexing (e.g. `-1` for the last axis).

### Examples
```csharp
var shape = new DynamicTensorShape(-1, 3, 224);
int batch = shape.Get(0);   // -1 (dynamic)
int channels = shape.Get(1); // 3
int last = shape.Get(-1);    // 224 (last axis)
```

### Exceptions
- **AssertionException** - Thrown when the axis is out of bounds.
```

--------------------------------

### implicit operator string(SubString)

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tokenization.SubString.html

Gets a string value from the portion of the source string of this SubString.

```APIDOC
## implicit operator string(SubString)

### Description
Gets a string value from the portion of the source string of this SubString.

### Method
```
public static implicit operator string(SubString input)
```

### Parameters
#### Path Parameters
- **input** (SubString) - Required - The SubString value to convert to a string value.

### Returns
- **string** - The string representing the value of this SubString.
```

--------------------------------

### Upload Tensor Data and Wait for Completion

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.ComputeTensorData.html

Demonstrates how to upload data to backing storage and ensure all pending operations are completed. This is useful for synchronizing data transfers before reading.

```csharp
var data = new NativeArray<float>(256, Allocator.Temp);
// Fill data
tensorData.Upload(data, 256);
tensorData.CompleteAllPendingOperations();

```

--------------------------------

### FixedPaddingSizeProvider Constructor

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tokenization.Padding.FixedPaddingSizeProvider.html

Initializes a new instance of the FixedPaddingSizeProvider class with a specified padding size.

```APIDOC
## FixedPaddingSizeProvider(int)

### Description
Initializes a new instance of the FixedPaddingSizeProvider type.

### Parameters
- **size** (int) - The target padding size.
```

--------------------------------

### SubString Indexer for Range Access

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tokenization.SubString.html

Gets a portion of the current SubString as a new SubString instance.

```csharp
public SubString this[Range offsets] { get; }
```

--------------------------------

### GetRanges Method

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tokenization.Truncators.StrategicTruncator.html

Gets the ranges to truncate sequences of tokens. Returns the number of ranges generated.

```csharp
protected int GetRanges(int length, int rangeMaxLength, Output<Range> output)
```

--------------------------------

### Create and Dispose Tensor

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tensor.html

Demonstrates creating a Tensor with specific dimensions and data, and how to properly dispose of it. It also shows an alternative using a 'using' statement for automatic disposal.

```csharp
// Create a tensor
m_Tensor = new Tensor<float>(new TensorShape(2, 3), new[] { 1.0f, 2.0f, 3.0f, 4.0f, 5.0f, 6.0f });
// Alternatively with `using` so you don't need to call `Dispose()` when you are done with the tensor
using var m_OtherTensor = new Tensor<float>(new TensorShape(2, 3), new[] { 1.0f, 2.0f, 3.0f, 4.0f, 5.0f, 6.0f });
// Get a CPU-accessible clone of the tensor. This copy owns its own data.
Tensor<float> cpuCopyTensor = m_Tensor.ReadbackAndClone() as Tensor<float>;
// Release memory.
cpuCopyTensor.Dispose();
m_Tensor.Dispose();
```

--------------------------------

### Tensor<T> Constructor and Usage

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tensor-1.html

Demonstrates how to create a Tensor<T> object, access its data, and manage its lifecycle with Dispose().

```APIDOC
## Tensor<T> Constructor and Usage

### Description
This example shows how to instantiate a `Tensor<float>` with specific dimensions and data, and how to obtain a CPU-accessible copy using `ReadbackAndClone()`. It also illustrates proper resource management by calling `Dispose()` on tensors when they are no longer needed, or by using a `using` statement for automatic disposal.

### Method
`Tensor<T>(TensorShape shape, T[] data)`
`Tensor<T>.ReadbackAndClone()`
`Dispose()`

### Endpoint
N/A (SDK Method)

### Parameters
#### Constructor Parameters
- **shape** (TensorShape) - The dimensions of the tensor.
- **data** (T[]) - An array containing the tensor's data.

#### ReadbackAndClone Parameters
None

### Request Example
```csharp
// Create a tensor
Tensor<float> m_Tensor = new Tensor<float>(new TensorShape(2, 3), new[] { 1.0f, 2.0f, 3.0f, 4.0f, 5.0f, 6.0f });

// Alternatively with `using` so you don't need to call `Dispose()` when you are done with the tensor
using var m_OtherTensor = new Tensor<float>(new TensorShape(2, 3), new[] { 1.0f, 2.0f, 3.0f, 4.0f, 5.0f, 6.0f });

// Get a CPU-accessible clone of the tensor. This copy owns its own data.
Tensor<float> cpuCopyTensor = m_Tensor.ReadbackAndClone() as Tensor<float>;

// Release memory.
cpuCopyTensor.Dispose();
m_Tensor.Dispose();
```

### Response
#### Success Response (N/A for constructors/methods)
N/A

#### Response Example
N/A
```

--------------------------------

### OnlyFirstStrategy.Instance

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tokenization.Truncators.Strategies.OnlyFirstStrategy.html

Gets the singleton instance of the OnlyFirstStrategy. This strategy is used to truncate only the first sequence of tokens.

```APIDOC
## Property: Instance

### Declaration
```csharp
public static ITruncationStrategy Instance { get; }
```

### Property Value
- **Type**: ITruncationStrategy
- **Description**: The singleton instance of the OnlyFirstStrategy.
```

--------------------------------

### DefaultTruncator.Instance

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tokenization.Truncators.DefaultTruncator.html

Gets a singleton instance of the default truncator. This provides access to the default truncation logic.

```APIDOC
## Property Instance

Gets a singleton instance of the default truncator.

### Declaration
```csharp
public static ITruncator Instance { get; }
```

### Property Value
Type | Description
---|---
ITruncator | The singleton instance of the DefaultTruncator.
```

--------------------------------

### Include Unity.InferenceEngine Namespace

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/manual/understand-sentis-workflow.html

Add this using statement at the top of your script to access Sentis functionalities.

```csharp
using Unity.InferenceEngine;

```

--------------------------------

### Sequence.Identifier Property

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tokenization.PostProcessors.Templating.Sequence.html

Gets the identifier for the sequence, which is either 'A' for the primary sequence or 'B' for the secondary sequence.

```APIDOC
## Identifier

### Description
Identifies the sequence: A for the primary sequence, B for the secondary sequence.

### Property Value
- **SequenceIdentifier** - The identifier of the sequence.
```

--------------------------------

### DefaultPostProcessor Constructor

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tokenization.PostProcessors.DefaultPostProcessor.html

Initializes a new instance of the DefaultPostProcessor class.

```APIDOC
## DefaultPostProcessor()

### Description
Initializes a new instance of the DefaultPostProcessor type.

### Declaration
```csharp
public DefaultPostProcessor()
```
```

--------------------------------

### RightDirectionRangeGenerator

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tokenization.Truncators.html

Generates a sequence of ranges starting from the right (upper bound) of the source token sequence.

```APIDOC
## Class: RightDirectionRangeGenerator

### Description
Generates a sequence of Range starting from the right (the upper bound of the source).

### Usage
This class is used to define truncation ranges that prioritize keeping tokens from the end of the sequence.
```

--------------------------------

### PreTokenize Method

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tokenization.PreTokenizers.MetaspacePreTokenizer.html

Performs the pre-tokenization on the input substring. The results are added to the provided output collection.

```csharp
public void PreTokenize(SubString input, Output<SubString> output)
{
}
```

--------------------------------

### LeftDirectionRangeGenerator

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tokenization.Truncators.html

Generates a sequence of ranges starting from the left (index 0) of the source token sequence.

```APIDOC
## Class: LeftDirectionRangeGenerator

### Description
Generates a sequence of Range starting from the left (`0`, the lower bound of the source).

### Usage
This class is used to define truncation ranges that prioritize keeping tokens from the beginning of the sequence.
```

--------------------------------

### Download Tensor to CPU

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/manual/use-model-output.html

Use `ReadbackAndClone` to move a tensor from the GPU to the CPU. This creates a read-writable copy for further processing.

```csharp
var outputTensor = worker.Schedule(inputTensor).PeekOutput() as Tensor<float>;
var cpuTensor = outputTensor.ReadbackAndClone();
```

--------------------------------

### Piece.GetPieceHashCode() Method

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tokenization.PostProcessors.Templating.Piece.html

Gets the hash code of this Piece. This is an abstract method that must be implemented by derived classes.

```APIDOC
#### GetPieceHashCode()
Gets the hash code of this Piece.

##### Declaration
```csharp
protected abstract int GetPieceHashCode()
```

##### Returns
Type | Description  
---|---  
int | The hash code of this Piece.
```

--------------------------------

### ByteLevelPreTokenizer Constructors

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tokenization.PreTokenizers.ByteLevelPreTokenizer.html

Initializes a new instance of the ByteLevelPreTokenizer class. You can control whether a prefix space is added and if GPT-2's regex is used for splitting.

```APIDOC
## ByteLevelPreTokenizer(bool addPrefixSpace = true, bool gpt2Regex = true)

### Description
Initializes a new instance of the ByteLevelPreTokenizer type.

### Parameters
- **addPrefixSpace** (bool) - Optional - Adds a whitespace at the beginning of the input if it doesn't start with one.
- **gpt2Regex** (bool) - Optional - Uses the GPT2 regex to split the input into smaller SubStrings.
```

--------------------------------

### Get Type IDs (ReadOnly)

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tokenization.IEncoding.html

Retrieves the list of type IDs. Returns the number of available tokens.

```csharp
IReadOnlyList<int> GetTypeIds()
```

--------------------------------

### LeftPadding Constructor with Padding Size Provider and Pad Token

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tokenization.Padding.LeftPadding.html

Initializes a new instance of the LeftPadding class. Requires an IPaddingSizeProvider to determine the final padded sequence size and a Token to use for padding.

```csharp
public LeftPadding(IPaddingSizeProvider paddingSizeProvider, Token padToken)
```

--------------------------------

### MetaspacePreTokenizer Constructor

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tokenization.PreTokenizers.MetaspacePreTokenizer.html

Initializes a new instance of the MetaspacePreTokenizer. You can customize the replacement character, the prepending scheme, and whether to split the input at metaspace boundaries.

```csharp
public MetaspacePreTokenizer(char replacement = '▁', PrependScheme prependScheme = PrependScheme.Always, bool split = true)
{
}
```

--------------------------------

### Get Token IDs (ReadOnly)

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tokenization.IEncoding.html

Retrieves the list of token IDs. Returns the number of available tokens.

```csharp
IReadOnlyList<int> GetIds()
```

--------------------------------

### TemplatePostProcessor Constructor

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tokenization.PostProcessors.TemplatePostProcessor.html

Initializes a new instance of the TemplatePostProcessor. It requires templates for single and paired sequences, along with a list of special tokens.

```csharp
public TemplatePostProcessor(Template single, Template pair, IEnumerable<(string value, int id)> specialTokens)
```

--------------------------------

### Accessing Tensor Elements

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tensor-1.html

Provides methods for getting and setting tensor elements using multi-dimensional indices.

```APIDOC
## this[int d0]

### Description
Returns the tensor element at offset `d0`.

### Method
`get; set;`

### Parameters
#### Path Parameters
- **d0** (int) - Axis 0.

### Property Value
- **T** - The type of the tensor element.
```

```APIDOC
## this[int d1, int d0]

### Description
Returns the tensor element at offset `(d1, d0)`, which is position `d1 * stride0 + d0`.

### Method
`get; set;`

### Parameters
#### Path Parameters
- **d1** (int) - Axis 1.
- **d0** (int) - Axis 0.

### Property Value
- **T** - The type of the tensor element.
```

```APIDOC
## this[int d2, int d1, int d0]

### Description
Returns the tensor element at offset `(d2, d1, d0)`, which is position `d2 * stride1 + d1 * stride0 + d0`.

### Method
`get; set;`

### Parameters
#### Path Parameters
- **d2** (int) - Axis 2.
- **d1** (int) - Axis 1.
- **d0** (int) - Axis 0.

### Property Value
- **T** - The type of the tensor element.
```

```APIDOC
## this[int d3, int d2, int d1, int d0]

### Description
Returns the tensor element at offset `(d3, d2, d1, d0)`, which is position `d3 * stride2 + d2 * stride1 + d1 * stride0 + d0` in this tensor.

### Method
`get; set;`

### Parameters
#### Path Parameters
- **d3** (int) - Axis 3.
- **d2** (int) - Axis 2.
- **d1** (int) - Axis 1.
- **d0** (int) - Axis 0.

### Property Value
- **T** - The type of the tensor element.
```

```APIDOC
## this[int d4, int d3, int d2, int d1, int d0]

### Description
Returns the tensor element at offset `(d4, d3, d2, d1, d0)`, which is position `d4 * stride3 + d3 * stride2 + d2 * stride1 + d1 * stride0 + d0`.

### Method
`get; set;`

### Parameters
#### Path Parameters
- **d4** (int) - Axis 4.
- **d3** (int) - Axis 3.
- **d2** (int) - Axis 2.
- **d1** (int) - Axis 1.
- **d0** (int) - Axis 0.

### Property Value
- **T** - The type of the tensor element.
```

```APIDOC
## this[int d5, int d4, int d3, int d2, int d1, int d0]

### Description
Returns the tensor element at offset `(d5, d4, d3, d2, d1, d0)`, which is position `d5 * stride4 + d4 * stride3 + d3 * stride2 + d2 * stride1 + d1 * stride0 + d0`.

### Method
`get; set;`

### Parameters
#### Path Parameters
- **d5** (int) - Axis 5.
- **d4** (int) - Axis 4.
- **d3** (int) - Axis 3.
- **d2** (int) - Axis 2.
- **d1** (int) - Axis 1.
- **d0** (int) - Axis 0.

### Property Value
- **T** - The type of the tensor element.
```

```APIDOC
## this[int d6, int d5, int d4, int d3, int d2, int d1, int d0]

### Description
Returns the tensor element at offset `(d6, d5, d4, d3, d2, d1, d0)`, which is position `d6 * stride5 + d5 * stride4 + d4 * stride3 + d3 * stride2 + d2 * stride1 + d1 * stride0 + d0`.

### Method
`get; set;`

### Parameters
#### Path Parameters
- **d6** (int) - Axis 6.
- **d5** (int) - Axis 5.
- **d4** (int) - Axis 4.
- **d3** (int) - Axis 3.
- **d2** (int) - Axis 2.
- **d1** (int) - Axis 1.
- **d0** (int) - Axis 0.

### Property Value
- **T** - The type of the tensor element.
```

```APIDOC
## this[int d7, int d6, int d5, int d4, int d3, int d2, int d1, int d0]

### Description
Returns the tensor element at offset `(d7, d6, d5, d4, d3, d2, d1, d0)`, which is position `d7 * stride6 + d6 * stride5 + d5 * stride4 + d4 * stride3 + d3 * stride2 + d2 * stride1 + d1 * stride0 + d0`.

### Method
`get; set;`

### Parameters
#### Path Parameters
- **d7** (int) - Axis 7.
- **d6** (int) - Axis 6.
- **d5** (int) - Axis 5.
- **d4** (int) - Axis 4.
- **d3** (int) - Axis 3.
- **d2** (int) - Axis 2.
- **d1** (int) - Axis 1.
- **d0** (int) - Axis 0.

### Property Value
- **T** - The type of the tensor element.
```

--------------------------------

### TokenConfiguration Constructor

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tokenization.TokenConfiguration.html

Initializes a new TokenConfiguration with specified properties for token behavior.

```csharp
public TokenConfiguration(int id, string value, bool wholeWord, Direction strip, bool normalized, bool special)
```

--------------------------------

### SplitDelimiterMergeWithPrevious.Instance

Source: https://docs.unity3d.com/Packages/com.unity.ai.inference%402.6/api/Unity.InferenceEngine.Tokenization.SplitDelimiterBehaviors.SplitDelimiterMergeWithPrevious.html

Gets a shared instance of the SplitDelimiterMergeWithPrevious class. This is a static property providing access to a singleton instance of the behavior.

```APIDOC
## Property Instance

### Description
Gets a shared instance of the SplitDelimiterMergeWithPrevious.

### Declaration
```csharp
public static SplitDelimiterMergeWithPrevious Instance { get; }
```

### Property Value
* **Type**: SplitDelimiterMergeWithPrevious
* **Description**: A shared instance of the SplitDelimiterMergeWithPrevious class.
```