nielsrogge/transformers-tutorials

★ 11,626Jupyter NotebookAudience · researcherComplexity · 3/5Setup · easy

Mindmap

mindmap
  root((repo))
    What it does
      Tutorial notebooks
      Pre-trained AI models
      Google Colab ready
    Task areas
      Text NER with BERT
      Image classification
      Object detection DETR
      Document understanding
    Tech used
      Python and PyTorch
      HuggingFace Transformers
      Jupyter notebooks
    Audience
      Python learners
      ML practitioners
      Researchers

mindmap root((repo)) What it does Tutorial notebooks Pre-trained AI models Google Colab ready Task areas Text NER with BERT Image classification Object detection DETR Document understanding Tech used Python and PyTorch HuggingFace Transformers Jupyter notebooks Audience Python learners ML practitioners Researchers

Click or tap to explore — scroll the page freely

Things people build with this

USE CASE 1

Run a named entity recognition notebook in Colab to label people, places, and organizations in text using BERT.

USE CASE 2

Apply DETR for object detection on your own images without writing model code from scratch.

USE CASE 3

Use LayoutLM or Donut to extract structured data from scanned PDFs or document images.

USE CASE 4

Estimate depth from a single photo using a pre-trained model via a self-contained notebook.

Tech stack

PythonPyTorchJupyter NotebookHuggingFace TransformersGoogle Colab

Getting it running

Difficulty · easy Time to first run · 30min

Requires a Google account to run notebooks in Colab, GPU runtime is recommended for faster model inference.

No license information was mentioned in the explanation.

In plain English

This repository is a personal collection of tutorial notebooks created by a HuggingFace researcher using the Transformers library. The Transformers library, made by HuggingFace, provides pre-trained AI models for tasks like understanding text, classifying images, detecting objects in photos, and working with audio. These notebooks show how to use those models in practice. Each notebook focuses on a specific model and a specific task. There are notebooks for recognizing named entities in text (identifying names of people, places, and organizations), classifying images into categories, detecting objects in photos, segmenting images into regions, estimating depth from a single photo, and reading text from scanned documents. Most notebooks run in Google Colab, a free browser-based tool that lets you run Python code without installing anything on your own computer. The models covered span a wide range of modern AI approaches. For text tasks there is BERT and related models. For images there are Vision Transformer, DETR, and several object detection variants. For documents there are LayoutLM and Donut, which are designed to understand the layout and content of scanned pages or PDFs. Each notebook is self-contained: it loads a pre-trained model, prepares input data, runs the model, and shows the output. This collection is aimed at people who already have some Python experience and want to see how to apply AI models to real problems without building those models from scratch. It is not a beginner-to-programming tutorial, but it does explain concepts as it goes. The author recommends the free HuggingFace course as background reading for anyone unfamiliar with these model types. All notebooks are implemented in PyTorch. The full README is longer than what was shown.

Copy-paste prompts

Prompt 1

I'm in a Google Colab notebook from nielsrogge/transformers-tutorials using DETR. Help me adapt the object detection code to run on my own JPG images stored in Google Drive.

Prompt 2

I want to use the LayoutLM notebook from transformers-tutorials to extract invoice fields from a PDF. Walk me through preparing data in the expected format.

Prompt 3

I'm using the BERT NER notebook from nielsrogge/transformers-tutorials. How do I change it to recognize custom entity types like product names and prices?

Prompt 4

The Vision Transformer image classification notebook is giving wrong results on my dataset. Help me fine-tune the model on my own labeled images using the existing notebook structure.

Open on GitHub → Explain another repo

← nielsrogge on gitmyhub — every repo by this author, as a profile.

Verify against the repo before relying on details.