Open-source alternatives to NVIDIA NeMo RL

Dopamine is a reinforcement learning research framework designed for prototyping and testing algorithms across diverse simulated environments. It provides an agent development toolkit that utilizes a flat class hierarchy to facilitate the creation and extension of learning agents. The framework includes a standardization layer via environment wrappers that connect agents to various physics simulations and gaming environments. It also features a high-performance experience replay buffer for storing and sampling transition data to improve training stability, alongside a dedicated hyperparameter

Habitat-Lab is an open-source platform for training and evaluating embodied AI agents in photorealistic 3D indoor environments. It functions as a high-performance 3D indoor environment simulator that supports physics-based interaction, enabling research into navigation and manipulation tasks. The platform provides a modular task-environment abstraction that separates task logic from environment simulation, using configuration-driven pipeline assembly to compose simulation and training pipelines. It includes a hierarchical sensor-actuator architecture for mixing and matching perception and act

tflearn is a deep learning framework and high-level API wrapper for TensorFlow. It provides a toolkit for designing neural network architectures and a system for executing training loops and optimizing model weights across CPUs and GPUs. The project simplifies the process of building and training models through a modular interface and a high-level API for prototyping. It includes specialized utilities for deep learning visualization, allowing for the generation of graphical diagrams to analyze network structures, weights, gradients, and activations. The framework covers a broad range of capa

Revolutionizing Reinforcement Learning Framework for Diffusion Large Language Models Most comprehensive framework for dLLM's and multimodal dLLM's post-training

EasyR1 is a distributed model training system and reinforcement learning framework for large language and vision-language models. It functions as a multimodal trainer and an implementation of a Proximal Policy Optimization pipeline designed to refine the reasoning and perception capabilities of models that process both text and images. The system specializes in distributing reinforcement learning workloads across multiple compute nodes to manage high memory requirements. It optimizes hardware utilization through padding-free training and fine-tuning to fit large models onto available graphics

simpleRL-reason is a training framework designed to improve mathematical and logical deduction in large language models. It utilizes reinforcement learning and policy optimization to enhance the accuracy and transparency of step-by-step deduction chains. The project implements a pipeline that establishes baseline capabilities through supervised fine-tuning before applying reinforcement learning to maximize deductive accuracy. It features a reward modeling toolkit that calculates scalar feedback by comparing generated reasoning steps against verified mathematical ground truths. The framework

This project is a transformer-based framework for generating dense and sparse vector embeddings of text and multimodal data. It serves as a library for fine-tuning models to perform semantic similarity tasks, retrieval, and reranking. The system is distinguished by its support for diverse architectural patterns, including bi-encoders for fast similarity search and cross-encoders for high-precision reranking. It provides dedicated pipelines for multimodal embeddings, mapping text and images into a shared vector space, and implements knowledge distillation to compress large models into smaller,

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

🕹️ A diverse suite of scalable reinforcement learning environments in JAX

TinyZero is a reinforcement learning framework and implementation designed to train language models to develop reasoning and self-verification abilities. It provides a training pipeline to optimize model performance on mathematical and logical tasks. The project serves as a minimal reproduction of the DeepSeek R1 architectural and training approach. It focuses on creating reasoning models that can solve structured problems through autonomous chain-of-thought discovery. The framework incorporates group relative policy optimization and reward-based self-correction to improve accuracy on logica

Julia code for the book Reinforcement Learning An Introduction

Persistent advantage learning dueling double DQN for the Arcade Learning Environment

Kiln is an LLM development workbench and evaluation framework designed for designing, testing, and optimizing prompts and AI agents. It functions as a multi-agent orchestrator and a RAG optimization tool, providing a visual interface for the iterative development of AI systems. The project distinguishes itself through a comprehensive fine-tuning pipeline that supports zero-code model training and reasoning distillation. It enables the creation of hierarchical multi-agent systems where specialized actors coordinate via tool calling, and it implements a Model Context Protocol server to expose t

Solutions of Reinforcement Learning, An Introduction

LLM Reasoners is a library to enable LLMs to conduct complex reasoning, with advanced reasoning algorithms. It approaches multi-step reasoning as planning and searches for the optimal reasoning chain, which achieves the best balance of exploration vs exploitation with the idea of "World Model"…

An e2e framework for creating, deploying and using isolated execution environments for agentic RL training, built using Gymnasium style simple APIs.

Agent Lightning is an optimization framework designed to refine the performance of individual AI agents within complex multi-agent systems. It provides a platform for improving decision-making and task execution by applying reinforcement learning, supervised fine-tuning, and automated prompt optimization. The framework distinguishes itself through its ability to isolate specific agents for targeted tuning, allowing developers to enhance individual behaviors while maintaining the stability of the broader system architecture. By utilizing a modular interface, it integrates with diverse agent fr

MM-EUREKA: Exploring the Frontiers of Multimodal Reasoning with Rule-based Reinforcement Learning

A deep Q learning demonstration using Google Tensorflow

Requirements • Quick Start • Available Environments • Documentation & Resources • Community & Support • Citations

VLM-R1 is a reasoning vision-language model and embodied AI framework designed to map visual inputs and language instructions into physical navigation waypoints and robotic actions. It functions as a multimodal policy optimizer and an open vocabulary detector capable of locating objects based on arbitrary natural language descriptions. The system distinguishes itself through the use of chain-of-thought reasoning and reinforcement learning to solve complex visual and spatial tasks. It utilizes a video semantic memory system, which employs a visual cache to maintain a history of live video for

An Open Source Approach to Scaling Up Reinforcement Learning on the Base Model

Baselines is a comprehensive suite of frameworks for reinforcement learning algorithm implementation, imitation learning, and training orchestration. It provides a library of standardized learning algorithms used to benchmark and replicate research results, alongside a deep learning policy framework for constructing neural network architectures such as multi-layer perceptrons, convolutional networks, and long short-term memory networks. The project includes a specialized imitation learning toolkit that enables agents to mimic expert behavior through behavior cloning and generative adversarial

Heretic is a specialized toolkit for removing safety alignment and refusal constraints from transformer-based language models. It utilizes directional ablation to suppress model refusals and restore unrestricted output capabilities. The project provides a framework for quantifying the effectiveness of these modifications by measuring refusal rates and evaluating divergence from the original model behavior. It also includes a suite for residual vector analysis, allowing for the calculation of geometric relationships between prompts and the visualization of hidden states across model layers. A

Agentic RL Training at Scale