Agent frameworks & infrastructure

Models, fine-tuning & training

2,504 agents · most starred first

Models, fine-tuning & training, page 25 of 26

thaillm

Thai large language model(LLM)

thebestllmever

andromeda - Pytorch

tictactoe-gpt-finetuning

Python tic tac toe state generator and GPT fine tuning

tilearn-llm

TILEARN for LLM

tinker-agent

Intelligent agent for Anthropic's Tinker fine-tuning platform

tinyllama

Model classes and pre-training utilities for a tiny version of Llama in PyTorch.

token-shift-gpt

Token Shift GPT - Pytorch

torch-agents

Easy Deep Reinforcement Learning for PyTorch

torch-gpt-2

GPT-2 implemented in PyTorch

torchrl-agents

Object-oriented RL agents based on torchrl.

toy-gpt-train-100-unigram

Toy GPT next-token prediction using a unigram model.

toy-gpt-train-200-bigram

Toy GPT next-token prediction using a bigram model.

toy-gpt-train-300-context-2

Toy GPT next-token prediction using a 2-token context window.

toy-gpt-train-400-context-3

Toy GPT next-token prediction using a 3-token context window.

toyllm

Toy LLM

trainllm

Fine-tuning LLMs for instruction using QLoRA.

transfer-llm

A modular PyTorch framework for fine-tuning Hugging Face models.

transnormerllm

transnormerllm

tsagentkit-patchtst-fm

Granite TSFM toolkit, model wrappers, and demos for time-series forecasting workflows.

tsagentkit-timesfm

TimesFM fork distribution for tsagentkit.

tsagentkit-uni2ts

Unified Training of Universal Time Series Forecasting Transformers

twoNeuron-pypi-chittariraghu

its an implimentation of Perceptron

types-boto3-bedrock

Type annotations for boto3 Bedrock 1.44.0 service generated with mypy-boto3-builder 8.12.0

ullme

ULLME: A Unified Framework for Large Language Model Embeddings with Generation-Augmented Learning

ultragrad

An automatic gradient engine for training your all in one deep learning model

unit-averaging

Unit averaging in Python: efficient ensemble estimation of unit-specific parameters

verbatim-llm

Library to mitigate verbatim or near-verbatim memorization in LLMs

vibhu-llm
virt-llm

Short description of your package

vocab-coverage

语言模型中文识字率分析

weather-swarm

Weather Swarm - Pytorch

willminds

Explore to building the intelligence easily! It's based on Torch, Transformers, etc.

xagents

Implementations of deep reinforcement learning algorithms in tensorflow 2.5

xllm

Simple & Cutting Edge LLM Finetuning

yieldthought-agents

Automation for TTNN model bringup.

PlexeAI

Create AI models instantly with plain English, no code, no fuss.

@ark7/model

Ark7 model used for both backend and frontend

OpenPipe AI

Build, fine-tune, and deploy custom AI models for better performance and lower costs.

agent2model

Turn your LangGraph agent into a small fine-tuned model that runs with no orchestrator - near-frontier quality at a fraction of the inference cost.

agentcamp

AgentCamp SDK — where agents train their own experts. Observe, train, shadow, and activate cheaper expert models behind your existing agents.

openai-harmony-js

TypeScript/JavaScript utilities for the GPT‑OSS Harmony format: renderers, parsers, tokenizers, and streaming helpers

cambium-llm

Advanced LLM architecture augmentation library - surgical model expansion with minimal compute

seren-llm

Seren: A Self-Refining Decoder-Only LLM with GQA, RoPE, Flash Attention, and a novel two-pass self-refinement mechanism.

cleanllm

Streaming JSONL cleaner for LLM fine-tuning datasets.

loom-gpt

A local toolkit for training tiny GPT models on your own data.

llm-from-scratch

Building Language Models from Scratch — tokenizer, Transformer, training, inference, and fine-tuning in clean, tested PyTorch.

villm-tok-fast

ViLLM Fast Tokenizer — Rust backend with embedded SentencePiece for Vietnamese-English code-switching

tokenizers-villm
villm-tok-rs
trigram-llm

Fast C-backed trigram language model for word prediction and sentence completion

agent-curriculum

Curriculum learning for agent training — difficulty-scored stages

umbrellm

A lightweight, trainable language model SDK

iconoclast-llm

ICONOCLAST — Discriminative representation editing for open-weight LLMs. Beats HERETIC baseline across all tested models.

semantic-llm-short-ans-generator

Sample responses and log-probabilities from HuggingFace causal LMs

Nous Research API

Nous Research is an AI research team focused on advancing open-source AI.

gpt-simple-lm

A clean, efficient framework for pretraining language models from scratch

llmtune-local

Fine-tune LLMs (LoRA/QLoRA) and image classifiers locally — browser UI + optional TUI

ossllms

Pull open-weight AI models from the ossllms torrent network. An HF-cache-compatible download shim for common huggingface_hub public model pulls.

skillmodels

Estimators for skill formation models

m23-llm

Mathieu Group M23 Weight Initialization for PyTorch Transformers and LLMs

@metaharness/weight-eft

Fine-tune cheap open-source LLMs (GLM, Qwen, DeepSeek) on your AI coding agent's successful runs with LoRA (SFT + DPO) so your model cascade escalates to expensive frontier models (GPT, Claude) less often — cutting cost-per-resolved. Turns run history int

@aihu/compiler-native-linux-x64-gnu

aihu compiler native addon (napi) — linux-x64-gnu platform binary.

@aihu/compiler-native-win32-x64-msvc

aihu compiler native addon (napi) — win32-x64-msvc platform binary.

@aihu/compiler-native-darwin-arm64

aihu compiler native addon (napi) — darwin-arm64 platform binary.

ablate-llm

Open-source, local chat with unrestricted small language models (ablate chat), plus a directional ablation (abliteration) toolkit to remove refusal from open LLMs.

agentdescent

Gradient descent, but the parameters are agents — a parallel, asynchronous framework for self-evolving agents (skills, prompts, harnesses)

gemma4-pt-claude

Canonical PyTorch implementation of Gemma 4

llmsurgery

LLM Surgery

pyautollm

AI-Native LLM Training & Infrastructure Abstraction Library

rayllm-orchestrator

Fine-tune and serve small LLMs on your laptop

tacl-llm

Ternary Additive Contrastive LLM — CPU-friendly integer neural network

ultragraph-1bit

A pure-Python (+numpy) byte-graph that is a 1-bit (ternary) LLM: dunder API, autograd, ultra-edges, visualization.

zahidgpt

Modular Transformer & Multi-Corpus (English, Arabic, Code) LLM Library

bpe-lite

Offline BPE tokenizer for OpenAI, Anthropic, and Gemini — zero dependencies

agents-learning-sdk

Native, in-process reinforcement learning SDK for AI agents — in-memory and local-file storage by default.

@deepseek-ai/dsh-agent-default-model

Default model selection shared by Agent entry points

tuisko-llm

Exact tokenizer and chat-template frontend for TuiskoLLM

agentevolve-optimizer

A drop-in NSGA-II replacement whose authored machinery installs only when measurement says it beats the classical path

automl-llm

Easy data distillation of LLMs for text classification, information extraction, and open-ended tasks

gpuhive-agent

RL-Split decision plane: masked PPO/TD3/bandit agents over the decision-parameter contract

llm-distillery

Capture LLM traffic through a drop-in proxy, curate it into a dataset, and distill a small model that replaces the frontier one.

agent-flywheel

Self-improving agents through continual learning: in-process agentic RL that turns your own agent's work into training data. Observed tool outcomes, outbound-only, one runtime dependency.

llm-decision-engine

Structured probabilistic decision inference for local causal language models using direct candidate scoring.

llm-typesafe

Use TypeSafe classification and scoring models with LLM

scalellm-ai

A simple high-level AI toolkit with scratch-trained LLM, image, video, and TTS models plus easy pretrained wrappers.

turbo-dllm

Highly optimized distributed training for diffusion language models.

agenttune

AgentTune: build agentic workflows, then train, evaluate, distill, and self-heal them through one trajectory schema. A TRL-backed alignment + fine-tuning library (full RL coverage — DPO, PPO, GRPO) with an agentic integration spine on top: agent-design / memory-design / harness-design pillars, a Project lifecycle, and a FastAPI service.

baby-llm

A developer-first PyTorch engine for explicit, modular LLM training. Zero magic.

glassllm

A small Transformer written from scratch in NumPy, built to be looked at.

jepa-agent

Name reserved for jepa-agent. jepa-agent: training and planning with action-conditioned JEPA world models.

llamafactory-kt

Unified Efficient Fine-Tuning of 100+ LLMs

llm-finetuner-pro

Advanced LLM Fine-Tuner — SFT, DPO, ORPO, KTO, GRPO, reward modelling and evaluation

llm-hpo-optimizer

General-purpose, multi-objective hyperparameter optimizer for LLM training, with a pluggable model adapter interface.

llm-zipper

Open-source framework for cheaper, faster, energy-efficient LLMs. Know before you compress.

loragression

LoRA fine-tuning, but more aggressively

rumi-llm

RUMI-8M small decoder-only Transformer language model

tabagent

Distill an LLM agent's closed-set decisions into cheap tabular classifiers: TabSchema, TabSynth, TabHead.

bedrock-tricore

Bedrock-TriCore: An Attention-Free, Constant-State O(1) Sequence Operator (17.00 KiB) with Causal Delta-Rule Fast Weights and Latent Multi-Hop Chaining

harmonic-llm

Training-free amplitude outlier suppression for Transformer Q/O projections, with a measured cost.

recursive-llm

Recursive Language Models for efficient long-context processing

More in Agent frameworks & infrastructure