Crate lc_chains 

Chain system for composing operations.

Chain is LangChain’s core abstraction, representing a sequence of operations.

Core Concepts

• BaseChain: Base trait for chains.
• LLMChain: Most basic chain (Prompt + LLM).
• SequentialChain: Execute multiple chains sequentially.

Example

use lc_chains::{LLMChain, SequentialChain};

let chain1 = LLMChain::new(llm.clone(), "Generate a word about {topic}");
let chain2 = LLMChain::new(llm, "Make a sentence with word: {word}");

let seq_chain = SequentialChain::new()
    .add_chain(Arc::new(chain1), vec!["topic"], vec!["word"])
    .add_chain(Arc::new(chain2), vec!["word"], vec!["sentence"]);

let inputs = HashMap::from([("topic".into(), "programming".into())]);
let result = seq_chain.invoke(inputs).await?;

Re-exports

pub use adapter::ChainRunnable;

pub use base::BaseChain;

pub use base::ChainError;

pub use base::ChainResult;

pub use base::ChainStream;

pub use base::StreamToken;

pub use conversation_chain::ConversationChain;

pub use conversation_chain::ConversationChainBuilder;

pub use conversation_retrieval::ConversationRetrievalChain;

pub use document_chains::MapReduceDocumentsChain;

pub use document_chains::MapRerankDocumentsChain;

pub use document_chains::RefineDocumentsChain;

pub use document_chains::StuffDocumentsChain;

pub use llm_chain::LLMChain;

pub use llm_chain::LLMChainBuilder;

pub use retrieval_qa::RetrievalQA;

pub use router_chain::LLMRouterChain;

pub use router_chain::RouteDestination;

pub use router_chain::RouterChain;

pub use sequential_chain::SequentialChain;

Modules

adapter

ChainRunnable adapter - bridges BaseChain to the Runnable trait.

base

Chain base trait.

conversation_chain

Conversation Chain

conversation_retrieval

ConversationRetrieval Chain

document_chains

Document processing chains.

llm_chain

LLM Chain

retrieval_qa

RetrievalQA Chain

router_chain

Router Chain

sequential_chain

Sequential Chain