Agent frameworks & infrastructure

Models, fine-tuning & training

Models, fine-tuning and training projects handle the model layer beneath an agent: running local checkpoints, fine-tuning them, building training datasets, and running reinforcement learning. Unsloth shows the practical side, a local UI for training and running LLMs and diffusion models, while easy-dataset covers dataset creation for fine-tuning, RAG, and evaluation.

2,504 agents · most starred first

Top 100 of 2,504

unsloth★ 79,826

Local UI to run and train LLMs and diffusion models. Supports GGUF, MLX, Qwen3.8, DeepSeek-V4.1, MiniMax-H3, Gemma 4, FLUX and more.

grok-1★ 52,237

Grok open release

Qwen★ 22,063

The official repo of Qwen (通义千问) chat & pretrained large language model proposed by Alibaba Cloud.

agent-lightning★ 18,986

The absolute trainer to light up AI agents.

easy-dataset★ 15,130

A powerful tool for creating datasets for LLM fine-tuning 、RAG and Eval

OpenMythos★ 14,932

A theoretical reconstruction of the Claude Mythos architecture, built from first principles using the available research literature.

stable-baselines3★ 14,038

PyTorch version of Stable Baselines, reliable implementations of reinforcement learning algorithms.

train-llm-from-scratch★ 13,279

A straightforward method for training your LLM, from downloading data to generating text.

gorilla★ 13,082

Gorilla: Training and Evaluating LLMs for Function Calls (Tool Calls)

Gymnasium★ 12,841

A standard API for single-agent reinforcement learning environments, with popular reference environments and related utilities (formerly Gym)

ludwig★ 11,792

Low-code framework for building custom LLMs, neural networks, and other AI models

Kimi-K2★ 11,121

Kimi K2 is the large language model series developed by Moonshot AI team

ART★ 10,866

Agent Reinforcement Trainer: train multi-step agents for real-world tasks using GRPO. Give your agents on-the-job training. Reinforcement learning for Qwen3.6, GPT-OSS, Llama, and more!

dolly★ 10,806

Databricks’ Dolly, a large language model trained on the Databricks Machine Learning Platform

OpenRLHF★ 10,171

An Easy-to-use, Scalable and High-performance Agentic RL Framework based on Ray (PPO & DAPO & REINFORCE++ & VLM & TIS & vLLM & Ray & Async RL)

oumi★ 9,441

Easily fine-tune (SFT/RL), evaluate, and deploy Qwen, Gemma, or any open agentic LLM/VLM!

reef★ 7,660

Infrastructure for continually self‑improving agents

mergekit★ 7,481

Tools for merging pretrained large language models.

TensorLayer★ 7,385

Deep Learning and Reinforcement Learning Library for Scientists and Engineers

InternLM★ 7,276

Official release of InternLM series (InternLM, InternLM2, InternLM2.5, InternLM3).

rllm★ 5,927

Democratizing Reinforcement Learning for LLMs

AReaL★ 5,922

The RL Bridge for LLM-based Agent Applications. Made Simple & Flexible.

OpenClaw-RL★ 5,749

OpenClaw-RL: Train any agent simply by talking

RLinf★ 5,684

RLinf: Reinforcement Learning Infrastructure for Embodied and Agentic AI

h2o-llmstudio★ 5,199

H2O LLM Studio - a framework and no-code GUI for fine-tuning LLMs. Documentation: https://docs.h2o.ai/h2o-llmstudio/

Llama-2-7b-chat-hf★ 4,971

transformers pytorch safetensors llama text-generation facebook

hands-on-modern-rl★ 4,931

🚀 An open-source, hands-on curriculum bridging the gap from basic RL concepts to LLM alignment, RLVR, and advanced Agentic systems.

GLM-4.5★ 4,426

GLM-4.5: Agentic, Reasoning, and Coding (ARC) Foundation Models

OpenManus-RL★ 4,178

A live stream development of RL tunning for LLM agents

acme★ 4,086

A library of reinforcement learning components and agents

nixtla★ 4,049

TimeGPT-2.1: production ready pre-trained Time Series Foundation Model for forecasting and anomaly detection. Generative pretrained transformer for time series trained on over 100B data points. It's capable of accurately predicting various domains such as retail, electricity, finance, and IoT with just a few lines of code 🚀.

FlagAI★ 3,864

FlagAI (Fast LArge-scale General AI models) is a fast, easy-to-use and extensible toolkit for large-scale model.

PettingZoo★ 3,555

A standard API for multi-agent reinforcement learning environments, with popular reference environments and related utilities

MetaClaw★ 3,470

🦞 Just talk to your agent — it learns and EVOLVES 🧬.

distilabel★ 3,465

Distilabel is a framework for synthetic data and AI feedback for engineers who need fast, reliable and scalable pipelines based on verified research papers.

ROLL★ 3,456

An Efficient and User-Friendly Scaling Library for Reinforcement Learning with Large Language Models

gpt-2-simple★ 3,405

Python package to easily retrain OpenAI's GPT-2 text-generating model on new texts

TensorFlow.NET★ 3,387

.NET Standard bindings for Google's TensorFlow for developing, training and deploying Machine Learning models in C# and F#.

Pearl★ 3,041

A Production-ready Reinforcement Learning AI Agent Library brought by the Applied Reinforcement Learning team at Meta.

Linly★ 3,036

Chinese-LLaMA 1&2、Chinese-Falcon 基础模型;ChatFlow中文对话模型;中文OpenLLaMA模型;NLP预训练/指令微调数据集

agents★ 3,028

TF-Agents: A reliable, scalable and easy to use TensorFlow library for Contextual Bandits and Reinforcement Learning.

gemma-4-12B-coder-fable5-composer2.5-v1-GGUF★ 3,021

gguf gemma4 coding code reasoning thinking

UltraChat★ 2,922

Large-scale, Informative, and Diverse Multi-round Chat Data (and Models)

rl-baselines3-zoo★ 2,914

A training framework for Stable Baselines3 reinforcement learning agents, with hyperparameter optimization and pre-trained agents included.

RAGEN★ 2,846

Agent RL framework for LLM agents: multi-turn reinforcement learning with StarPO and reasoning-collapse diagnostics

Memento★ 2,712

Official Code of Memento: Fine-tuning LLM Agents without Fine-tuning LLMs

plexe★ 2,653

✨ Build a machine learning model from a prompt

MiniMax-M2★ 2,603

MiniMax-M2, a model built for Max coding & agentic workflows.

mPLUG-Owl★ 2,538

mPLUG-Owl: The Powerful Multi-modal Large Language Model Family

EasyLM★ 2,514

Large language models (LLMs) made easy, EasyLM is a one stop solution for pre-training, finetuning, evaluating and serving LLMs in JAX/Flax.

verl-agent★ 2,456

verl-agent is an extension of veRL, designed for training LLM/VLM agents via RL. verl-agent is also the official code for paper "Group-in-Group Policy Optimization for LLM Agent Training"

DeepRL-Agents★ 2,275

A set of Deep Reinforcement Learning Agents implemented in Tensorflow.

prime-rl★ 2,265

Agentic RL Training at Scale

Visual-RFT★ 2,255

Official repository of 'Visual-RFT: Visual Reinforcement Fine-Tuning' & 'Visual-ARFT: Visual Agentic Reinforcement Fine-Tuning'’

Qwen2.5-Coder-32B-Instruct★ 2,216

transformers safetensors qwen2 text-generation code codeqwen

Llama-2-70b-chat-hf★ 2,210

transformers pytorch safetensors llama text-generation facebook

sia★ 2,168

SIA is a Self Improving AI framework to autonomously improve the performance of any AI system (Model / Agent) on a benchmark task.

dgm★ 2,160

Darwin Gödel Machine: Open-Ended Evolution of Self-Improving Agents

on-policy★ 2,139

This is the official implementation of Multi-Agent PPO (MAPPO).

dataclaw★ 2,106

Agent harness to publish your history from Claude Code et al. as Huggingface datasets.

Step-3.5-Flash★ 2,091

Fast, Sharp & Reliable Agentic Intelligence

diamond★ 2,073

DIAMOND (DIffusion As a Model Of eNvironment Dreams) is a reinforcement learning agent trained in a diffusion world model. NeurIPS 2024 Spotlight.

Dialog_Corpus★ 2,057

用于训练中英文对话系统的语料库 Datasets for Training Chatbot System

all-rl-algorithms★ 2,004

Implementation of all RL algorithms in a simpler way

Magma★ 1,947

[CVPR 2025] Magma: A Foundation Model for Multimodal AI Agents

Awesome-AgenticLLM-RL-Papers★ 1,922
TinyLlama-1.1B-Chat-v1.0★ 1,917

transformers safetensors llama text-generation conversational en

Awesome-AgenticLLM-RL-Papers★ 1,816
gemma-4-12B-agentic-fable5-composer2.5-v2-3.5x-tau2-GGUF★ 1,792

gguf gemma4 coding agentic terminal tool-use

RL-Factory★ 1,789

Train your Agent model via our easy and efficient framework

NLP-Models-Tensorflow★ 1,777

Gathers machine learning and Tensorflow deep learning models for NLP problems, 1.13 < Tensorflow < 2.0

Jackrong-llm-finetuning-guide★ 1,760
Qwen3-Coder-Next★ 1,749

transformers safetensors qwen3_next text-generation conversational license:apache-2.0

ChatLM-mini-Chinese★ 1,738

中文对话0.2B小模型(ChatLM-Chinese-0.2B),开源所有数据集来源、数据清洗、tokenizer训练、模型预训练、SFT指令微调、RLHF优化等流程的全部代码。支持下游任务sft微调,给出三元组信息抽取微调示例。

Agent-R1★ 1,724

Agent-R1: Training Powerful LLM Agents with End-to-End Reinforcement Learning

Adala★ 1,651

Adala: Autonomous DAta (Labeling) Agent framework

Phi-4-multimodal-instruct★ 1,629

transformers safetensors phi4mm text-generation nlp code

aideml★ 1,627

AIDE: an LLM agent for machine learning engineering - the research Weco grew out of. Referenced in OpenAI MLE-bench.

agentic-rl-lab★ 1,539

Reproducing and studying RL algorithms for LLM agents, including GRPO, GSPO, DAPO, OPD, Search-R1, ReTool, ALFWorld and beyond.

Ovis★ 1,536

A novel Multimodal Large Language Model (MLLM) architecture, designed to structurally align visual and textual embeddings.

pytorch-openai-transformer-lm★ 1,524

🐥A PyTorch implementation of OpenAI's finetuned transformer language model with a script to import the weights pre-trained by OpenAI

AgentTuning★ 1,509

AgentTuning: Enabling Generalized Agent Abilities for LLMs

Hermes-Function-Calling★ 1,504
Ovis★ 1,452

A novel Multimodal Large Language Model (MLLM) architecture, designed to structurally align visual and textual embeddings.

ReCall★ 1,449

ReSearch: Learning to Reason with Search for LLMs via Reinforcement Learning & ReCall: Learning to Reason with Tool Call for LLMs via Reinforcement Learning

evolutionary-model-merge★ 1,439

Official repository of Evolutionary Optimization of Model Merging Recipes

spacy-transformers★ 1,411

🛸 Use pretrained transformers like BERT, XLNet and GPT-2 in spaCy

Qwen3-Coder-480B-A35B-Instruct★ 1,390

transformers safetensors qwen3_moe text-generation conversational arxiv:2505.09388

labs-molt★ 1,385

A scalable, agentic-first, and HuggingFace-native RL framework for research (9k lines).

Qwen3-Coder-30B-A3B-Instruct★ 1,347

transformers safetensors qwen3_moe text-generation conversational arxiv:2505.09388

EvoSkill★ 1,309

EvoSkill — An open-source framework that automatically discovers and synthesizes reusable agent skills from failed trajectories to improve coding agent performance.

LlamaGym★ 1,255

Fine-tune LLM agents with online reinforcement learning

Evolutionary-Algorithm★ 1,232

Evolutionary Algorithm using Python, 莫烦Python 中文AI教学

Qwen3-Coder-30B-A3B-Instruct-GGUF★ 1,232

transformers gguf unsloth qwen3 qwen text-generation

rl-baselines-zoo★ 1,204

A collection of 100+ pre-trained RL agents using Stable Baselines, training and hyperparameter optimization included.

RSI-Anything★ 1,189

RSI-Anything: Unlocking Recursive Self-Improvement with World Models

Llama-2-13b-chat-hf★ 1,138

transformers pytorch safetensors llama text-generation facebook

DataDreamer★ 1,125

DataDreamer: Prompt. Generate Synthetic Data. Train & Align Models.   🤖💤

skrl★ 1,103

Modular Reinforcement Learning (RL) library (implemented in PyTorch, JAX, and NVIDIA Warp) with support for Gymnasium/Gym, NVIDIA Isaac Lab, MuJoCo Playground and other environments

Automodel★ 1,094

🚀 Pytorch Distributed native training library for LLMs/VLMs with OOTB Hugging Face support

More in Agent frameworks & infrastructure