Agent frameworks & infrastructure

Models, fine-tuning & training

2,504 agents · most starred first

Models, fine-tuning & training, page 8 of 26

DaMo★ 30

The official implement of paper 《DaMo: Data Mixing Optimizer in Fine-tuning Multimodal LLMs for Mobile Phone Agents》

Transformer★ 30

Chatbot using Tensorflow (Model is transformer) ko

Qwen3.5-27B-Claude-4.6-Opus-Reasoning-Distilled★ 30

gradio region:us

guillaumetell-7b★ 30

transformers safetensors mistral text-generation conversational fr

gemma-2-9b-it-russian-function-calling-GGUF★ 30

transformers safetensors gguf gemma2 text-generation conversational

llama-3-8B-Instruct-function-calling★ 30

transformers safetensors llama text-generation conversational en

DeepSeek-V2-Chat-GGUF★ 30

transformers gguf code deepseek bf16 text-generation

Atlas-Chat-9B★ 30

transformers safetensors gemma2 text-generation conversational ar

Qwen1.5-14B-Chat-AWQ★ 30

transformers safetensors qwen2 text-generation chat conversational

Jais-2-70B-Chat★ 30

transformers safetensors jais2 text-generation conversational ar

Toucan-LLM★ 29

Self-trained Large Language Models based on Meta LLaMa

ddrl★ 29

Deep Developmental Reinforcement Learning

atari_multitask★ 29

Atari gauntlet for RL agents

Multi_Agent_Deep_Reinforcement_Learning★ 29

Implement Google Deep Minds DQN for multiple agents for a grid world environment where vehicles must pick up customers.

STeCa★ 29

(ACL2025 Findings) Official code for the paper "STeCa: Step-level Trajectory Calibration for LLM Agent Learning"

Llama3-8B-Chinese-Chat-GGUF-f16★ 29

transformers gguf llama text-generation llama-factory orpo

Huihui-Agents-A1-abliterated★ 29

transformers safetensors qwen3_5_moe image-text-to-text moe vlm

AgenticQwen★ 29

Reproducing AgenticQwen (arXiv:2604.21590) — dual-flywheel data synthesis + GRPO RL training for agentic small LLMs

Local_LLM_Training_Apple_Silicon★ 28

Created and enhanced a local LLM training system on Apple Silicon with MLX and Metal API, overcoming the absence of CUDA support. Fine-tuned the Llama3 model on 16 GPUs for streamlined solution of verbose math word problems. Result: a powerful, privacy-preserving chatbot that runs smoothly on-device.

intro-to-llms★ 28

Short fine tuning demo with custom dataset for training gemma2-2b-it to perform function calling

rStar2-Agent-14B★ 28

transformers safetensors qwen3 text-generation reinforcement-learning agentic-reasoning

gemma-4-E2B-it-qat-q4_0-unquantized-assistant★ 28

transformers safetensors gemma4_assistant text-generation any-to-any arxiv:2607.02770

Yi-34B-Chat-8bits★ 28

transformers safetensors llama text-generation conversational arxiv:2403.04652

Eagle-X5-13B-Chat★ 28

transformers safetensors eagle_llama text-generation Eagle VLM

CodeLlama-13b-Instruct-hf★ 28

transformers pytorch safetensors llama text-generation facebook

CodeLlama-7b-Python-hf★ 28

transformers pytorch safetensors llama text-generation facebook

CoEvolve★ 28

CoEvolve: Training LLM Agents via Agent-Data Mutual Evolution

pytorch-meena★ 27

Open-domain chatbot (Meena-style) with a vanilla Transformer seq2seq in PyTorch.

LiteResearcher★ 27

A Scalable Agentic RL Training Framework for Deep Research Agent

Chatbot-Dialog-Dataset★ 27

Dialogs for training or setting up a chatbot

Model-Augmented-Fine-Tuning★ 27

Fine-tuning black-box OpenAI embedding models

madac★ 27

Official implementation of NeurIPS22 paper “Multi-agent Dynamic Algorithm Configuration”

AgentLadder★ 27

Climb from LLM API to RL Agent with Klara

Yi-1.5-34B-Chat-16K★ 27

transformers safetensors llama text-generation conversational arxiv:2403.04652

Atlas-Chat-2B★ 27

transformers safetensors gemma2 text-generation conversational ar

Llama3-8B-Chinese-Chat-GGUF-4bit★ 27

transformers gguf llama text-generation llama-factory orpo

CAPO★ 27

CAPO: Critic-Guided Action-Aligned Policy Optimization for Advancing LLM Agent Capabilities

AceGPT-13B-chat★ 27

transformers pytorch llama text-generation ar license:apache-2.0

gpt2★ 26

R Interface to OpenAI's GPT-2 model

Meta-LLAMA3-GenAI-UseCases-End-To-End-Implementation-Guides★ 26

META LLAMA3 GENAI Real World UseCases End To End Implementation Guide

MultiAgent-ImitationLearning★ 26

Code for training policies based on paper Coordinated Multi-Agent Imitation Learning

SARNet★ 26

Code repository for SARNet: Learning Multi-Agent Communication through Structured Attentive Reasoning (NeurIPS 2020)

Targeted-Manipulation-and-Deception-in-LLMs★ 26

Codebase for "On Targeted Manipulation and Deception when Optimizing LLMs for User Feedback". This repo implements a generative multi-turn RL environment with support for agent, user, user feedback, transition and veto models. It also implements KTO and expert iteration for training on user preferences.

Qwen1.5-72B-Chat-AWQ★ 26

transformers safetensors qwen2 text-generation chat conversational

internlm2_5-1_8b-chat★ 26

transformers safetensors internlm2 feature-extraction text-generation conversational

jais-30b-chat-v3★ 26

transformers pytorch jais text-generation custom_code arxiv:2308.16149

Llama-2-13B-chat-AWQ★ 26

transformers safetensors llama text-generation facebook meta

gemma-4-26B-A4B-it-qat-q4_0-unquantized-assistant★ 26

transformers safetensors gemma4_assistant text-generation image-text-to-text conversational

mcp-tool-use-quality-ranger-0.6b-GGUF★ 26

gguf qwen3 text-generation-inference mcp text-generation en

Qwen3.8-Distill-35B-A3B-Coder-Abliterated★ 26

gguf qwen3_5_moe_text qwen3 moe a3b mtp

pi-agent-rl-toolkit★ 26

Train AI coding agents with RL (GRPO) on Prime Intellect — works with Pi Agent

open-spark-jev★ 26

Open-source, local decision models inspired by TypeSafe’s Jev and System One - built on Qwen3 for NVIDIA DGX Spark.

Trainee-to-Trainer★ 26

Official repository for “From Trainee to Trainer: LLM-Designed Training Environment for RL with Multi-Agent Reasoning”“

airgpt★ 26

train your own chatbot from scratch on a mac

tokenizer★ 25

C++ implementation of tokenizers, including tiktoken.

Matlab2TorchRL★ 25

This project is a pipeline that connects a Matlab simulation (Simulink) to an OpenAI Gym wrapper for PyTorch Reinforcement Learning using the DQN algorithm.

HyperMARL★ 25

Adaptive Hypernetworks for Multi-Agent RL. NeurIPS 2025.

AGmind-ML★ 25

Locally fine-tuned Russian RAG models: document splitter, query expansion, retrieval embedder. Teacher distillation → GGUF → llama.cpp on AMD Vulkan. Commercial-OK licenses.

glm-4-9b-chat-hf★ 25

transformers safetensors glm text-generation chatglm conversational

Jais-2-70B-Chat★ 25

transformers safetensors jais2 text-generation conversational ar

Agents-A1-FP8★ 25

transformers safetensors qwen3_5_moe image-text-to-text quantized moe

Qwen1.5-7B-Chat-GPTQ-Int4★ 25

transformers safetensors qwen2 text-generation chat conversational

Agents-A1-MTP-APEX-GGUF★ 25

transformers gguf qwen3_5_moe qwen3_5 reasoning agentic

Llama-3-Groq-8B-Tool-Use-GGUF★ 25

transformers gguf mistral quantized 2-bit 3-bit

LongCat-Flash-Chat-FP8★ 25

LongCat-Flash-Chat safetensors text-generation transformers conversational custom_code

mvp-engine★ 25

MVP Engine - The Next-Generation Framework for Multimodal Model Training with Agents

deepseek-coder-33B-instruct-GPTQ★ 25

transformers safetensors llama text-generation conversational base_model:deepseek-ai/deepseek-coder-33b-instruct

distrl-open★ 24

DistRL: An Asynchronous Distributed Reinforcement Learning Framework for On-Device Control Agents

chatbot★ 24

chatbot coded in videos part wise

MountainCar-v0_DeepRL★ 24

OpenAI MountainCar-v0 DeepRL-based solutions (DQN, DuelingDQN, D3QN)

gym-adserver★ 24

An OpenAI gym environment for ad serving algorithms

AdaExplore★ 24

[COLM-LLA 2026] The official implementation for paper "AdaExplore: Failure-Driven Adaptation and Diversity-Preserving Search for Efficient Kernel Generation"

Llama-2-7B-Chat-AWQ★ 24

transformers safetensors llama text-generation facebook meta

Llama-2-70B-Chat-AWQ★ 24

transformers safetensors llama text-generation facebook meta

Qwen3.5-9B-C3SM-SDM-Agentic-Coder-GGUF★ 24

transformers gguf llama.cpp qwen qwen3.5 agent

kgpt★ 23

A GPT-2-class language model trained from scratch on OpenWebText with the intent to augment AI Transformer-model education and reverse engineer GPT models from scratch.

Pixtral-Finetune★ 23

An open-source implementaion for fine-tuning Pixtral by MistralAI.

Reward-Modeling-RLHF-Finetune-and-RAG★ 23

Gemma2(9B), Llama3-8B-Finetune-and-RAG, code base for sample, implemented in Kaggle platform

torchwm★ 23

A modular PyTorch library designed for learning, training, and deploying world models across various environments.

trajectoryRL★ 23

Bittensor Subnet 11: verified agent work on open-weight models. Season 2: fusion policies over the Engy catalog; next: an open protocol for verifiable environments.

AI-Dataset-Generator★ 23

Tool-native dataset generation skill for Codex, Claude Code, and Antigravity with SFT/DPO pipelines, verification, deduplication, and flexible export.

StepPO★ 23

StepPO: Step-Aligned Policy Optimization for Agentic Reinforcement Learning

Llama-3.1-TAIDE-R1-8B-Chat★ 23

transformers safetensors gguf llama text-generation mergekit

Huihui-Agents-A1-abliterated-GGUF★ 23

transformers gguf moe vlm vision agentic

Llama-3-Groq-70B-Tool-Use-GGUF★ 23

gguf facebook meta pytorch llama llama-3

KAT-Coder-V2.5-Dev-OptiQ-4bit★ 23

mlx safetensors qwen3_5_moe optiq code text-generation

gpt-finetune-kit★ 22

Fine-tune GPT models ready-to-go

h2o-danube3-4b-chat-GGUF★ 22

transformers gguf gpt llm large language model h2o-llmstudio

Jais-2-8B-Chat★ 22

transformers safetensors jais2 text-generation conversational ar

llama3_8b_chat_uncensored★ 22

transformers tensorboard safetensors gguf llama text-generation

glm-edge-1.5b-chat★ 22

safetensors glm edge text-generation conversational license:other

Olmo-3.1-7B-RL-Zero-Code★ 22

transformers safetensors olmo3 text-generation conversational en

Qwen3-Coder-480B-A35B-Instruct-NVFP4★ 22

Model Optimizer safetensors qwen3_moe nvidia ModelOpt Qwen3

Llama-3-Groq-8B-Tool-Use-4bit★ 22

mlx safetensors llama facebook meta pytorch

stable-code-instruct-3b-GGUF★ 22

transformers gguf causal-lm code text-generation en

lfm2.5-2.6b-fable5-coding-agent-heretic★ 22

safetensors gguf lfm2 heretic uncensored decensored

MiniCPM5-1B-Agentic-Tooluse-Merged-FP16★ 22

transformers safetensors llama text-generation minicpm minicpm5

agentic-rl-forge★ 22

Local-first Agent RL workbench for Search-R1 rollouts, reward evaluation, GRPO/verl training-data handoff, and an auxiliary private RAG workspace.

FedAgent★ 22

[NeurIPS 2026] FedAgent: A Library for Decentralized Agent Learning. 🏆 Best Paper Award at the AAAI'26 TrustAgent Workshop, 🏆 Outstanding Paper Award in the AAAI'26 PerFM Workshop.

Gymnasium-Colaboratory-RL★ 21

Reinforcement Learning + Open AI Gym/Gymnasium in Google's Colab.

More in Agent frameworks & infrastructure