AI repos for Fine-tuning
Open-source AI projects built around Fine-tuning or genuinely useful if you work with it, each with what it does for you. 12 repos, most relevant first.
-
minimind
Runs pretraining, SFT and reinforcement learning stages you can read and rerun on one GPU.
-
dive-into-llms
Chinese notebooks and slides walk through fine-tuning LLMs with hands-on practice.
-
transformers
Fine-tune pretrained models using its consistent model and training APIs.
-
LLMs-from-scratch
Chapters and notebooks on finetuning a small GPT for classification and instruction following.
-
Kronos
Follow a fine-tuning pipeline to adapt small downloadable models to your own market data.
-
reasoning-from-scratch
Walks through GRPO reinforcement learning and distillation to add reasoning to a small model.
-
GLiNER
Includes training code to fine-tune the entity extraction model on your own labels.
-
airllm
Train large open models with small VRAM by keeping one layer on the GPU at a time.
-
needle
Includes fine-tuning and deployment support to adapt the small model to your own tools.
-
ACE-Step-1.5
Train LoRAs to adapt the music model to a particular style.
-
GLM-5
Lists fine-tuning support so you can adapt the open-weights model yourself.
-
HyperQwen
Includes a requantization pipeline that prepares the model for a single consumer GPU.