Struct LLMChain 

pub struct LLMChain { /* private fields */ }

LLM Chain

Combines a Prompt template and an LLM. The most basic Chain.

Examples

use lc_chains::LLMChain;

let chain = LLMChain::new(llm, "{question}");

let inputs = HashMap::from([("question".to_string(), "What is Rust?".into())]);
let result = chain.invoke(inputs).await?;

Implementations

impl LLMChain

pub fn new<L>(llm: L, prompt_template: impl Into<String>) -> Self

where L: BaseChatModel + Send + Sync + 'static, L::Error: Into<ProviderError>,

Create a new LLMChain.

Arguments

• llm - LLM client (any type implementing BaseChatModel)
• prompt_template - Prompt template string with {variable} placeholders

pub fn with_input_key(self, key: impl Into<String>) -> Self

Set input key name.

pub fn with_output_key(self, key: impl Into<String>) -> Self

Set output key name.

pub fn with_name(self, name: impl Into<String>) -> Self

Set chain name.

Trait Implementations

impl BaseChain for LLMChain

fn invoke_with_config<'life0, 'async_trait>( &'life0 self, inputs: HashMap<String, Value>, config: Option<RunnableConfig>, ) -> Pin<Box<dyn Future<Output = Result<ChainResult, ChainError>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait,

Execute the Chain with callback propagation.

Fires on_chain_start → on_llm_start → LLM call → on_llm_end → on_chain_end. On error, fires on_llm_error / on_chain_error instead.

fn stream<'life0, 'async_trait>( &'life0 self, inputs: HashMap<String, Value>, ) -> Pin<Box<dyn Future<Output = Result<ChainStream, ChainError>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait,

Stream execution for LLMChain – token by token output.

fn stream_with_config<'life0, 'async_trait>( &'life0 self, inputs: HashMap<String, Value>, config: Option<RunnableConfig>, ) -> Pin<Box<dyn Future<Output = Result<ChainStream, ChainError>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait,

Stream with config propagation.

0.22.0 audit fix (H-C3): the chain’s RunnableConfig is threaded into stream_chat, so sampling overrides / cancellation token / callbacks reach the provider (providers consume config via apply_overrides) instead of every streaming call being hardwired to None.

fn input_keys(&self) -> Vec<&str>

Get input keys.

fn output_keys(&self) -> Vec<&str>

Get output keys.

fn invoke<'life0, 'async_trait>( &'life0 self, inputs: HashMap<String, Value>, ) -> Pin<Box<dyn Future<Output = Result<ChainResult, ChainError>> + Send + 'async_trait>>

where Self: 'async_trait, 'life0: 'async_trait,

Execute the Chain.

fn name(&self) -> &str

Get Chain name.

fn validate_inputs( &self, inputs: &HashMap<String, Value>, ) -> Result<(), ChainError>

Validate inputs.