Struct SimplePromptOptimizer 

pub struct SimplePromptOptimizer { /* private fields */ }

Simple prompt optimizer that uses an LLM to iteratively rewrite instructions.

Process:

1. Score the initial instruction on a training batch
2. For each iteration, generate a candidate instruction using the optimizer LLM
3. Score the candidate — keep it if it improves on the best score
4. Validate the best instruction on the validation set

Implementations

impl SimplePromptOptimizer

pub fn new( optimizer_llm: Arc<dyn BaseLlm>, sampler: Arc<dyn Sampler>, config: SimplePromptOptimizerConfig, ) -> Self

Create a new simple prompt optimizer.

Trait Implementations

impl AgentOptimizer for SimplePromptOptimizer

fn optimize<'life0, 'life1, 'life2, 'async_trait>( &'life0 self, initial_instruction: &'life1 str, model_id: &'life2 str, ) -> Pin<Box<dyn Future<Output = Result<OptimizerResult, OptimizerError>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait, 'life1: 'async_trait, 'life2: 'async_trait,

Run the optimization process.