huggingface/notebooks
Notebooks
This is a collection of Jupyter notebooks that serve as educational guides for training, fine-tuning, and deploying machine learning models within the Hugging Face ecosystem. The notebooks cover the full lifecycle of model development, from loading and configuring pre-trained transformers to packaging trained models for real-time inference via scalable endpoints.
The notebooks demonstrate a range of capabilities including diffusion model training and fine-tuning for image generation and editing, transformer model adaptation for natural language processing tasks, and parameter-efficient fine-tuning techniques that reduce computational cost. They also cover multi-GPU training orchestration, hardware accelerator utilisation, and the deployment of models as production inference endpoints.
Beyond core training workflows, the collection includes guides for image generation tasks such as text-to-image synthesis, inpainting, super-resolution, and instruction-based editing. Additional notebooks cover robot policy training from demonstration data and long-form question answering systems using retrieval-augmented approaches. The repository also provides tooling for converting static documentation into executable notebooks for interactive learning.
Features
- Tutorials and Notebooks - Organises executable code examples and explanations into self-contained Jupyter notebooks for interactive learning.
- Diffusion Model Training - Provides notebooks that teach training and fine-tuning of diffusion models for image generation and editing.
- Text-to-Image Generators - Produces visual content from textual descriptions using diffusion models for creative and guided image synthesis.
- Large Language Model Fine-Tuning - Demonstrates adapting large language models to custom datasets using parameter-efficient techniques.
- Model Inference and Serving - Provides notebooks demonstrating how to load trained models and run inference in production-like settings.
- Parameter Efficient Fine-Tuning - Demonstrates adapting large language models by updating only a small subset of parameters to reduce computational cost.
- Model Serving & Deployment - Ships guides for packaging trained models and deploying them as scalable inference endpoints.
- Pretrained Checkpoint Fine-Tuning - Provides notebooks for adjusting pre-existing models on custom datasets by modifying training configurations.
- Transformers Tutorials - Ships guided notebooks for implementing transformer models for NLP tasks like question answering and text generation.
- Hugging Face Implementations - Provides guided notebooks for implementing transformer models for NLP tasks using the Hugging Face ecosystem.
- Machine Learning Tutorials - Provides executable Jupyter notebooks demonstrating how to train, fine-tune, and deploy ML models using the Hugging Face ecosystem.
- Image Inpainting - Fills in missing or selected parts of an image with new content that blends into the surrounding area.
- Distributed Deep Learning - Runs distributed training across multiple GPUs and optimizes compute costs with spot instances for large-scale experiments.
- Image Variation and Mixing - Creates new image variants from a source image, using text prompts to guide the changes.
- Text-Instruction Editors - Modifies existing images by giving natural language instructions to change content or style.
- Image Super Resolution Models - Increases the resolution of input images to produce higher-quality versions with more detail.
- Concept-Specific Adaptations - Includes notebooks for adapting pre-trained diffusion models to specific concepts using a small set of example images.
- Hardware Acceleration - Demonstrates leveraging GPUs and hardware accelerators to speed up model training and inference.
- Multi-GPU Training Utilities - Covers multi-GPU training orchestration with distributed strategies and spot instance optimisation.
- Structural Image Generation - Generates images guided by structural inputs such as depth maps or edge detections.
- Transformer Training Walkthroughs - Provides walkthrough notebooks for loading datasets, configuring models, and running training loops for NLP tasks.
- Model Endpoint Deployments - Packages trained models into scalable inference endpoints for real-time predictions via REST APIs.
