Projects sharing features with Agent Lightning

AReaL is a system for agent orchestration, distributed model training, and parameter-efficient tuning. It provides a framework for developing multi-turn reasoning agents and training large models using reinforcement learning from human feedback. The project implements a toolkit for improving the visual reasoning and geometry problem solving capabilities of vision-language models. It utilizes a memory-efficient tuning system to optimize mathematical and reasoning models across different inference backends. The infrastructure supports large-scale training through tensor, pipeline, and expert p

Composer is a PyTorch distributed training framework designed for scaling large-scale models across multi-node GPU clusters. It functions as a large language model trainer, a distributed model optimizer, and a training lifecycle manager. The project differentiates itself as a deep learning regularization library, providing specialized optimization techniques such as Sharpness Aware Minimization, MixUp, and CutMix to improve model generalization. It further distinguishes its training flow through the use of sequence length warmup, progressive layer freezing, and sharded-state checkpointing for

This project is an educational curriculum and architectural framework for building autonomous AI agents and multi-agent systems. It provides a structured learning path focused on the development of independent software components capable of planning, executing tasks, and utilizing external tools to achieve high-level goals. The framework emphasizes multi-agent system orchestration through distributed architectures where specialized agents collaborate using standardized communication protocols. It details specific design patterns such as dual-memory systems for maintaining short-term plans and

This project is a deep learning framework designed for constructing, training, and deploying neural networks across diverse hardware environments. It functions as a high-performance tensor computation library that provides both imperative and symbolic programming interfaces, allowing developers to balance flexible, step-by-step model building with the efficiency of compiled computation graphs. The framework distinguishes itself through a hybrid execution engine that integrates declarative graph compilation with imperative runtime logic. It supports scalable, distributed training across multip

ZenML is an orchestration platform designed for building, deploying, and monitoring reproducible machine learning pipelines and agentic workflows. It provides a unified framework that manages the entire lifecycle of machine learning assets, from data processing and model training to the deployment of persistent inference services. By decoupling pipeline logic from underlying compute and storage, the platform enables teams to transition workflows seamlessly from local development environments to production-grade cloud infrastructure. The platform distinguishes itself through a service-oriented

ZenML is an extensible machine learning orchestration framework designed to manage the end-to-end lifecycle of data pipelines and AI agent workflows. It functions as a durable orchestrator that executes machine learning tasks as directed acyclic graphs, ensuring that every step is containerized for consistent performance across local, cloud, and hybrid infrastructure. By decoupling pipeline code from underlying compute and storage backends, the platform allows developers to define infrastructure-agnostic stacks that remain portable across diverse environments. The project distinguishes itself

This framework provides a development environment for building collaborative systems where autonomous agents interact to solve complex tasks through conversational workflows. It functions as a conversational workflow engine and event-driven runtime, coordinating multi-step processes by translating high-level goals into structured dialogue sequences between specialized agents. The system distinguishes itself through its message-passing orchestration, which manages state transitions and task delegation between independent participants. It supports dynamic conversation state management to provid

Agentscope is a comprehensive toolkit for developing and orchestrating autonomous multi-agent systems. It provides a unified framework for building agents that can reason, execute tools, and manage memory, enabling the creation of complex, collaborative workflows where multiple specialized agents interact to solve multi-step objectives. The platform distinguishes itself through a robust orchestration engine that supports both sequential and concurrent agent pipelines. It utilizes a centralized event bus for real-time telemetry, allowing developers to track agent reasoning, tool usage, and sys

Swarm is a framework for building conversational systems that coordinate multi-agent workflows. It functions as an orchestration engine that manages persistent, multi-turn dialogues by routing tasks between specialized agents and executing local functions. The system is designed to handle complex, multi-step processes by maintaining shared state and context across agent interactions. The framework distinguishes itself through its approach to dynamic task delegation and execution control. It enables agents to hand off tasks to one another by returning agent objects, allowing for modular, domai

Mastra is an orchestration framework designed for building, deploying, and managing autonomous AI agents and multi-agent systems. It provides a comprehensive suite of primitives for creating resilient AI applications, including durable workflow orchestration, event-driven agent loops, and semantic memory management. By integrating these core components, the platform enables developers to build complex, multi-step processes that can reason about goals and execute tasks without manual intervention. The framework distinguishes itself through its focus on observability and secure, isolated execut

SLIME is a distributed reinforcement learning framework for large language model post-training that bridges Megatron training with SGLang inference servers. It orchestrates scalable RL loops across GPU clusters, decoupling training and inference into independent processes that communicate over HTTP and NCCL for independent scaling and fault tolerance. The system supports multi-agent reinforcement learning workflows with parallel agent instances, customizable rollout strategies, and personalized agent serving that improves models from prior conversations without disrupting API serving. The fra

CrewAI is a multi-agent orchestration framework designed for building autonomous systems that execute complex, multi-step workflows. It provides a development platform where specialized agents are defined with specific roles, goals, and tool sets to perform tasks collaboratively. By leveraging a declarative workflow engine, the system manages task dependencies, state transitions, and execution logic, allowing for the creation of structured, stateful sequences of operations. The framework distinguishes itself through its hierarchical management capabilities, which utilize manager agents to coo

This project is a game AI training framework designed to develop and monitor reinforcement learning agents within a legacy game environment. It functions as a training and monitoring system that optimizes autonomous agents to complete game objectives through exploration and reward-based learning. The framework includes tools for game memory mapping and real-time trajectory visualization. These capabilities translate raw game memory addresses into visual coordinates, allowing agent movements and session data to be streamed to a map for the analysis of navigation patterns and area exploration.

AISystem is a comprehensive AI full-stack infrastructure project covering the entire pipeline from AI chip architecture to high-level training frameworks. It encompasses the development of AI compiler frameworks, inference engines, and distributed training orchestrators designed to coordinate workloads across a heterogeneous compute stack of CPUs, GPUs, and NPUs. The project focuses on the deep integration of software and hardware, employing software-hardware co-design to align tensor layouts with physical memory structures. It provides specialized capabilities for accelerating Transformer mo

Flashlight is a C++ machine learning library and deep learning framework designed for building and training neural networks. It functions as a tensor manipulation library and an automatic differentiation engine that tracks operations to calculate gradients via backpropagation for model optimization. The project is distinguished by its role as a distributed training framework, utilizing all-reduce gradient synchronization and distributed environments to scale machine learning workloads across multiple nodes and devices. It features a backend-agnostic memory interface and RAII-based management

Torchtitan is a reference implementation for distributed deep learning built within the PyTorch ecosystem. It provides a framework for training large neural network models across multiple GPUs and nodes by combining several parallelism techniques, including fully sharded data parallelism (FSDP), tensor parallelism, and pipeline parallelism, making it possible to train models that exceed the memory capacity of a single device. The system distinguishes itself through asynchronous checkpointing, which saves model and optimizer state to persistent storage without pausing the training loop, enabli

This project provides a framework for managing multi-agent systems, designed to automate complex software development, infrastructure, and business workflows. It functions as a multi-agent workflow orchestrator that routes tasks to domain-specific workers while maintaining state persistence and infrastructure automation. By leveraging large language models, the system decomposes high-level objectives into actionable plans, ensuring that complex operations are executed with consistency and reliability. The framework distinguishes itself through its hierarchical agent registry and policy-driven

Torchtune is a PyTorch-native library for fine-tuning, aligning, and quantizing large language models. It provides a configurable training pipeline orchestrated through YAML recipes, with CLI overrides and component swapping, distributed training via FSDP2, memory optimizations, and parameter-efficient fine-tuning methods like LoRA, DoRA, and QLoRA. The library distinguishes itself through its YAML-driven configuration system that defines all training parameters and instantiates components from config files, with full CLI override capability for any field or component at launch time. It suppo

Search-R1 is a distributed training system and reinforcement learning framework designed to create search-augmented language models. It provides an architecture for scaling model workloads across head and worker nodes while optimizing how models interleave internal reasoning with external tool calls. The system focuses on refining model behavior through custom reward signals and reinforcement learning to improve tool-use formatting and information retrieval. It implements an interleaved reasoning-search loop that allows models to alternate between internal thought generation and external data

PyTorch Lightning is a high-level deep learning framework for PyTorch that automates training loops and removes repetitive engineering boilerplate. It functions as a structured pipeline for managing machine learning experiments, providing a distributed training orchestrator and tools for mixed-precision training. The framework decouples scientific model architecture from the engineering required for infrastructure and scaling. This separation allows the same model code to execute across CPUs, GPUs, or TPUs through a hardware-agnostic execution engine and a centralized trainer that manages the

This project is a PyTorch reinforcement learning library and agent training framework. It provides a suite of deep reinforcement learning algorithms, including DQN, PPO, and SAC, to facilitate the development of autonomous agents that optimize behavior through trial and error. The library focuses on the implementation of various actor-critic methods and deep learning architectures for research into autonomous decision making. It enables the training of intelligent agents within diverse environments by leveraging PyTorch-based model implementations. The codebase covers core reinforcement lear

AIOS is an LLM agent operating system and orchestration kernel designed to manage memory, resource scheduling, and tool execution for multiple autonomous AI agents. It serves as a comprehensive framework for developing and deploying agents, featuring a dedicated resource manager that coordinates model backends, GPU memory, and isolated kernel instances. The system distinguishes itself through a semantic memory engine that uses vector search and autonomous clustering for long-term knowledge management, and a semantic file system that allows users to control computer files and system operations

GenAI_Agents is a development framework and orchestration engine designed for building autonomous, multi-agent systems. It provides the infrastructure to construct complex, state-managed workflows where specialized agents collaborate to execute multi-step tasks, manage long-term memory, and perform iterative reasoning. The platform distinguishes itself through its graph-based orchestration model, which allows developers to define intricate agentic processes with explicit state transitions. It supports advanced control mechanisms such as human-in-the-loop intervention for manual oversight and

This project provides a comprehensive framework for building, deploying, and orchestrating autonomous agents within a decentralized network. It serves as a collection of patterns and examples for developing intelligent software entities capable of performing complex tasks, making decisions, and interacting with other agents to achieve shared goals. The framework distinguishes itself through its focus on multi-agent orchestration and decentralized communication. It enables the coordination of specialized agent teams that collaborate on workflows through structured messaging protocols, allowing

This project is a collection of TensorFlow machine learning examples providing reference implementations for various neural network paradigms. It covers supervised, unsupervised, reinforcement, and sequential learning models. The repository includes implementations for convolutional neural networks focused on image classification and ranking, as well as recurrent neural networks for time-series forecasting and sequence-to-sequence translation. It further provides examples of reinforcement learning agents trained via reward optimization and unsupervised learning techniques such as autoencoders

Acme is a reinforcement learning framework and execution engine designed for developing and benchmarking learning algorithms. It provides a library of modular components and reference implementations used to construct agents and establish performance baselines. The system enables the scaling of agent architectures from single-stream execution to large distributed environments. This allows for the transition from initial prototyping to distributed execution for training and evaluation. The framework covers reinforcement learning development and agent architecture prototyping, providing the bu