// the find
roboflow/notebooks
A collection of tutorials on state-of-the-art computer vision models and techniques. Explore everything from foundational architectures like ResNet to cutting-edge models like RF-DETR, YOLO11, SAM 3, and Qwen3-VL.
A big, continuously updated collection of Jupyter notebooks for fine-tuning and running SOTA computer vision and VLM models (YOLO26, RF-DETR, SAM3, Qwen3-VL, PaliGemma2, etc.) on Colab/Kaggle/SageMaker. It's for developers who want a working training/inference script for a specific model without writing the boilerplate themselves.
Genuinely current — notebooks for SAM3 and Qwen3-VL exist within weeks of those models shipping, and the last push is from this month. Each entry links out to the paper, source repo, a blog post, and often a YouTube walkthrough, so you're not left guessing what the notebook is doing. One-click Colab/Kaggle/SageMaker badges mean zero local setup to try a model. The notebook table is autogenerated from a CSV via a GitHub Action, so the index doesn't silently drift out of sync with the actual files like most tutorial repos.
There's no CI that actually executes the notebooks, so nothing stops one from silently breaking when an upstream dependency (SAM3, Qwen3-VL, etc.) changes its API — and these are exactly the fast-moving repos most likely to do that. It's a flat pile of 100+ notebooks with inconsistent naming (train-x vs how-to-x vs zero-shot-x), so the generated table is the only real navigation; browsing the repo directly is unpleasant. A meaningful chunk of the tutorials funnel into Roboflow's own stack (inference, hosted datasets, API keys), so it's not fully vendor-neutral despite covering third-party models. There's no versioning or pinning guidance per notebook, so running an older one against current library versions is a coin flip.