ModelPreferences

Namespace: ``

Configuration class for model sampling preferences in the MCP system.

This class encapsulates various parameters that control how language models generate responses. It allows fine-tuning of the sampling process to achieve different types of outputs, from deterministic to highly creative.

Preferences can control: - Temperature (randomness) - Top-p sampling - Maximum tokens - Stop sequences - Frequency penalties - Presence penalties

@package LaravelMCP\MCP\Sampling

Methods

__construct

Create a new model preferences instance.

@param float|null $temperature Controls randomness in generation (0.0 to 2.0) @param float|null $top_p Controls diversity via nucleus sampling (0.0 to 1.0) @param int|null $max_tokens Maximum tokens to generate, null for no limit @param array|null $stop Sequences that will stop generation @param float|null $frequency_penalty Penalty for frequent tokens (0.0 to 2.0) @param float|null $presence_penalty Penalty for token presence (0.0 to 2.0) @param float|null $cost_priority Priority for cost optimization (0.0 to 1.0) @param float|null $intelligence_priority Priority for intelligence optimization (0.0 to 1.0) @param float|null $speed_priority Priority for speed optimization (0.0 to 1.0) @param array|null $hints Hints for the model

getTemperature

Get the sampling temperature.

Higher values make the output more random, while lower values make it more deterministic.

@return float The temperature value (0.0 to 2.0)

getTopP

Get the nucleus sampling threshold.

Controls diversity via nucleus sampling. A value of 1.0 considers all tokens, while lower values restrict to more likely tokens.

@return float The top-p value (0.0 to 1.0)

getMaxTokens

Get the maximum number of tokens to generate.

@return int|null The maximum tokens, or null for no limit

getStop

Get the stop sequences.

These sequences will cause the model to stop generating when encountered.

@return array<string> The stop sequences

getFrequencyPenalty

Get the frequency penalty.

Higher values reduce the likelihood of the model repeating the same tokens frequently.

@return float The frequency penalty (0.0 to 2.0)

getPresencePenalty

Get the presence penalty.

Higher values reduce the likelihood of the model repeating any token that has appeared in the text so far.

@return float The presence penalty (0.0 to 2.0)

getHints

Get the hints for the model.

@return array<string, mixed> The hints for the model

setHints

Set the hints for the model.

@param array<string, mixed> $hints The new hints for the model

getCostPriority

Get the cost priority.

@return float The cost priority (0.0 to 1.0)

getIntelligencePriority

Get the intelligence priority.

@return float The intelligence priority (0.0 to 1.0)

getSpeedPriority

Get the speed priority.

@return float The speed priority (0.0 to 1.0)

toArray

Convert the preferences to an array format.

@return array The preferences as a key-value array

create

Create a new instance from an array of data.

@param array $values The values to create the instance from @return static A new ModelPreferences instance