Sft projects on GitHub

Supervised fine-tuning of language models on curated instruction datasets.

As of October 17, 2026, the most-starred Sft projects on GitHub are happy-llm (34.4K stars), open-r1 (26.5K stars) and ms-swift (15.9K stars). GitHub.Locker tracks 14 Sft repositories with 136.4K stars combined. The fastest-growing on October 16 was Soup, with +79 stars in a day.

14Projects

136.4KStars in total

+189Stars in a day

Browse in the catalog →github.com/topics/sft ↗

Hot right now

Most stars gained in a day · Oct 16

+79

Soup

MakazhanAlpamys/Soup · ★ 9.2K

Soup is a command-line tool for fine-tuning and post-training LLMs from a single YAML config file, automating GPU detection, batching and quantization so you just write a config and run one command.

+37

AgentGuide

adongwanai/AgentGuide · ★ 11K

AgentGuide is a free, open-source Chinese-language knowledge base for AI Agent engineering and job hunting, modeled on JavaGuide. It stitches scattered courses, tutorials, papers and projects into one systematic path from beginner to a job offer.

+20

train-llm-from-scratch

FareedKhan-dev/train-llm-from-scratch · ★ 12.2K

A from-scratch tutorial and codebase for training your own LLM end to end: tokenizing raw text, pretraining a GPT-style Transformer, and then post-training it with SFT, a reward model, DPO/ORPO/KTO, PPO and GRPO into an aligned, reasoning-style assistant.

+18

happy-llm

datawhalechina/happy-llm · ★ 34.4K

Happy-LLM is Datawhale's free open-source LLM course: a systematic tutorial that goes from NLP basics and the Transformer architecture to building and training your first large language model, then applying it with RAG and agents.

The all-time top

14 projects · by stars

#1 · datawhalechina/happy-llm
happy-llm
Happy-LLM is Datawhale's free open-source LLM course: a systematic tutorial that goes from NLP basics and the Transformer architecture to building and training your first large language model, then applying it with RAG and agents.
34.4K
#2 · huggingface/open-r1
open-r1
Open R1 is a fully open effort to reproduce the training pipeline behind DeepSeek-R1, providing scripts to run SFT and GRPO training, generate synthetic reasoning data, and evaluate the resulting reasoning models.
26.5K
#3 · modelscope/ms-swift
ms-swift
ms-swift is the ModelScope community's fine-tuning and deployment framework for LLMs and multimodal models, covering training, inference, evaluation, quantization and deployment in one pipeline.
15.9K

4
train-llm-from-scratch
12.2KFareedKhan-dev/train-llm-from-scratch · Python
A from-scratch tutorial and codebase for training your own LLM end to end: tokenizing raw text, pretraining a GPT-style Transformer, and then post-training it with SFT, a reward model, DPO/ORPO/KTO, PPO and GRPO into an aligned, reasoning-style assistant.
5
bisheng
12.1Kdataelement/bisheng · Python
BISHENG is an open-source LLM application devops platform built for enterprise AI, covering GenAI workflow building, RAG, agents, model management, evaluation, SFT, dataset management and observability.
6
AgentGuide
11Kadongwanai/AgentGuide · MDX
AgentGuide is a free, open-source Chinese-language knowledge base for AI Agent engineering and job hunting, modeled on JavaGuide. It stitches scattered courses, tutorials, papers and projects into one systematic path from beginner to a job offer.
7
Soup
9.2KMakazhanAlpamys/Soup · Python
Soup is a command-line tool for fine-tuning and post-training LLMs from a single YAML config file, automating GPU detection, batching and quantization so you just write a config and run one command.
8
fara
6.2Kmicrosoft/fara · Python
A family of open computer-use agent models — Fara1.5-4B, 9B and 27B, built on Qwen3.5 — that perform multi-step web tasks for you by controlling the mouse and keyboard like a human would.
9
hands-on-modern-rl
4.7Kwalkinglabs/hands-on-modern-rl · Python
Hands-On Modern RL is an open-source course and book that walks learners from basic reinforcement learning concepts through LLM alignment (RLHF, DPO, GRPO, RLVR) and into advanced agentic systems, combining theory with runnable code.
- rl
- dpo
- ppo
- grpo
10
NanoJev
2.6KTianyuCodings/NanoJev · Python
A nano replica of Jev: a 0.6B-parameter parallel decision model that takes states and questions in and returns complete probability distributions — no output tokens are ever decoded.
11
LoongForge
687baidu-baige/LoongForge · Python
An open-source training framework from Baidu's Baige team for LLMs, VLMs, diffusion and embodied models, built for faster training and lower cost on NVIDIA GPUs and Kunlun XPUs.
12
Qwen3.8-Flash-Next
433QwenLM/Qwen3.8-Flash-Next
Qwen3.8-Flash-Next is an open-weight multimodal Mixture-of-Experts foundation model from Alibaba's Qwen Team, also serving as an early preview of the architecture Alibaba plans to use for the upcoming Qwen4 model family.
13
jeeves
425PostHog/jeeves · Python
Jeeves is a reasoning-enhanced 'Jev-like' decision model built on Qwen3.5-9B with LoRA. It thinks through a reasoning chain before giving calibrated yes/no, multiple-choice, or rating answers, and ships with a diffusion drafter for faster inference.
14
EvoTrace
292jinzijian/EvoTrace · Python
EvoTrace mines the Claude Code and Codex coding-agent sessions already on your machine, finds the ones worth keeping, and compiles them into verified, reusable post-training assets: preference data, replayable coding tasks, RL environments, and execution-reward candidates.