Class CompressionOptions

Namespace

AiDotNet.PromptEngineering.Compression

Assembly

AiDotNet.dll

Options for controlling prompt compression behavior.

public class CompressionOptions

Inheritance

object

CompressionOptions

Inherited Members

object.Equals(object)

object.Equals(object, object)

object.GetHashCode()

object.GetType()

object.MemberwiseClone()

object.ReferenceEquals(object, object)

object.ToString()

Remarks

Configures how aggressive compression should be, what techniques to use, and constraints on the output.

For Beginners: Settings for how to compress your prompt.

Example:

var options = new CompressionOptions
{
    TargetReduction = 0.3,  // Try to reduce by 30%
    PreserveVariables = true,  // Don't remove {placeholders}
    MinTokenCount = 50  // Don't compress below 50 tokens
};

Properties

Aggressive

Gets aggressive compression options for maximum reduction.

public static CompressionOptions Aggressive { get; }

Property Value

CompressionOptions

Conservative

Gets conservative compression options that preserve more content.

public static CompressionOptions Conservative { get; }

Property Value

CompressionOptions

Default

Gets default compression options with moderate settings.

public static CompressionOptions Default { get; }

Property Value

CompressionOptions

MaxTokens

Gets or sets the maximum number of tokens in the output.

public int? MaxTokens { get; set; }

Property Value

int?

Remarks

If set, compression will try to get the output below this token count. Useful when you need to fit within a specific context window.

For Beginners: The maximum tokens allowed after compression. Set this if you have a hard limit to fit within.

MinTokenCount

Gets or sets the minimum number of tokens in the output.

public int MinTokenCount { get; set; }

Property Value

int

Remarks

A safety limit to prevent over-compression that would lose essential content.

For Beginners: Don't compress below this many tokens. Prevents the prompt from becoming too short to be useful.

ModelName

Gets or sets the model name for accurate token counting.

public string ModelName { get; set; }

Property Value

string

Remarks

Different models tokenize differently. Specifying the target model ensures accurate token counts for that model.

For Beginners: Which AI model you'll use this prompt with. Helps count tokens accurately for your specific model.

PreserveCodeBlocks

Gets or sets whether to preserve code blocks during compression.

public bool PreserveCodeBlocks { get; set; }

Property Value

bool

Remarks

When true, content within code blocks (```) will not be modified. Important for prompts that include code examples.

For Beginners: Keep code examples unchanged. You don't want compression to break your code!

PreserveVariables

Gets or sets whether to preserve template variables during compression.

public bool PreserveVariables { get; set; }

Property Value

bool

Remarks

When true, placeholders like {variable_name} will not be modified or removed.

For Beginners: Keep {placeholders} intact. You don't want compression to break your template!

TargetReduction

Gets or sets the target reduction ratio (0.0 to 1.0).

public double TargetReduction { get; set; }

Property Value

double

Remarks

The desired percentage of tokens to remove. A value of 0.3 means try to remove 30% of tokens. The compressor will try to achieve this target but may not always succeed.

For Beginners: How much smaller you want the prompt. - 0.2 = Try to make it 20% smaller - 0.5 = Try to make it 50% smaller Higher values = more aggressive compression