API Reference

Welcome to AdalFlow. The API reference is organized by subdirectories.

Core

All base/abstract classes, core components like generator, embedder, and basic functions are here.

core.component	Base building block for building LLM task pipelines.
core.container	Container component for composing multiple components, such as Sequential.
core.base_data_class	A base class that provides an easy way for data to interact with LLMs.
core.default_prompt_template	This is the default system prompt template used in the LightRAG.
core.model_client	ModelClient is the protocol and base class for all models(either via APIs or local models) to communicate with components.
core.db	LocalDB to perform in-memory storage and data persistence(pickle or any filesystem) for data models like documents and dialogturn.
core.functional	Functional interface.
core.generator	Generator is a user-facing orchestration component with a simple and unified interface for LLM prediction.
core.string_parser	Extract and convert common string to Python objects.
core.embedder	The component that orchestrates model client (Embedding models in particular) and output processors.
core.retriever	The base class for all retrievers who in particular retrieve documents from a given database.
core.prompt_builder	Class prompt builder for LightRAG system prompt.
core.tokenizer	Tokenizer from tiktoken.
core.func_tool	Tool is LLM's extended capability which is one of the core design pattern of Agent.
core.tool_manager	The ToolManager manages a list of tools, context, and all ways to execute functions.
core.types	Functional data classes to support functional components like Generator, Retriever, and Assistant.

Components

Functional components like model client, retriever, agent, local data processing, and output parsers are here.

components.agent.react	Implementation and optimization of React agent.
components.model_client.anthropic_client	Anthropic ModelClient integration.
components.model_client.cohere_client	Cohere ModelClient integration.
components.model_client.google_client	Google GenAI ModelClient integration.
components.model_client.groq_client	Groq ModelClient integration.
components.model_client.openai_client	OpenAI ModelClient integration.
components.model_client.transformers_client	Huggingface transformers ModelClient integration.
components.model_client.utils	Helpers for model client for integrating models and parsing the output.
components.data_process.data_components	Helper components for data transformation such as embeddings and document splitting.
components.data_process.text_splitter	Splitting texts is commonly used as a preprocessing step before embedding and retrieving texts.
components.reasoning.chain_of_thought	Chain of the thought(CoT) is to mimic a step-by-step thought process for arriving at the answer.
components.retriever.bm25_retriever	BM25 retriever implementation.
components.retriever.faiss_retriever	Semantic search/embedding-based retriever using FAISS.
components.retriever.llm_retriever	LLM as retriever module.
components.retriever.postgres_retriever	Leverage a postgres database to store and retrieve documents.
components.retriever.reranker_retriever	Reranking model using modelclient as a retriever.
components.output_parsers.outputs	The most commonly used output parsers for the Generator.

Datasets

datasets.big_bench_hard
datasets.trec
datasets.hotpot_qa
datasets.types

Evaluation

eval.base	Abstract base class for evaluation metrics.
eval.answer_match_acc	This is the metric for QA generation.
eval.retriever_recall	Retriever Recall @k metric.
eval.llm_as_judge	This is the metric to use an LLM as a judge for evaluating the performance of predicted answers.
eval.g_eval	Implementation of G-Eval: G-eval <https://arxiv.org/abs/2303.08774, nlpyang/geval> Instead of getting 1/5 as the score, AdalFlow will use 0.2 as the score, so that we can have a score in range [0, 1] for all metrics.

Optimization

optim.parameter	Parameter is used by Optimizer, Trainers, AdalComponent to auto-optimizations
optim.optimizer	Base Classes for AdalFlow Optimizers, including Optimizer, TextOptimizer, and DemoOptimizer.
optim.grad_component	Base class for Autograd Components that can be called and backpropagated through.
optim.types	All data types used by Parameter, Optimizer, AdalComponent, and Trainer.
optim.function
optim.few_shot.bootstrap_optimizer	Adapted and optimized boostrap fewshot optimizer:
optim.text_grad.text_loss_with_eval_fn	Adapted from text_grad's String Based Function
optim.text_grad.tgd_optimizer	Text-grad optimizer and prompts.
optim.text_grad.llm_text_loss	Implementation of TextGrad: Automatic “Differentiation” via Text
optim.trainer.trainer	Ready to use trainer for LLM task pipeline
optim.trainer.adal	AdalComponent provides an interface to compose different parts, from eval_fn, train_step, loss_step, optimizers, backward engine, teacher generator, etc to work with Trainer.

Tracing

tracing.decorators
tracing.generator_state_logger
tracing.generator_call_logger

Utils

utils.data	Default Dataset, DataLoader similar to utils.data in PyTorch.
utils.logger	This logger file provides easy configurability of the root and named loggers, along with a color print function for console output.
utils.setup_env([dotenv_path])	Load environment variables from .env file.
utils.lazy_import	Lazy import a module and class.
utils.serialization
utils.config	Config helper functions to manage configuration and rebuilt your task pipeline.
utils.registry