Learn Hugging Face Bootcamp

Set up Hugging Face in a notebook and manage transformers and diffusers versions, then access GPU notebooks via Colab or SageMaker Studio Lab.

Explore text models with the transformers library from Hugging Face, loading models and datasets to run tasks like sentiment analysis, translation, classification, and text generation using large language models.

Learn Hugging Face basics by loading a Rotten Tomatoes dataset with datasets, using a transformers sentiment analysis pipeline, and exploring pandas conversion and DistilBert options.

Explore how Transformers pipelines unlock a range of natural language processing tasks—from sentiment analysis and named entity recognition to Q&A and translation—using Hugging Face models and the diffusers ecosystem.

Explore how large language models process text from tokenization to vector embeddings, use transformers for inference, and sample next tokens with temperature, top k, and top p.

Explore how language models tokenize text into tokens and generate input IDs with GPT-2 and tokenizers. Understand how logits yield next-token probabilities, how decoding and sampling (top-k) drive text generation.

Explore generating text with gpt-2 by sampling from the model’s output, using repetition penalty, top-k, top-p, and temperature, via a transformers pipeline.

Explore the diffusers library from Hugging Face to work with diffusion-based text-to-image models, learn image data handling with OpenCV and NumPy in Python, and study auto pipelines.

Explore how computers represent image data, from grayscale 28x28 arrays to RGB with three color channels and optional alpha, and learn normalization from 0–255 to 0–1 for practical image processing.

Learn to load, display, and transform image data in Python using PIL and NumPy, convert images to arrays, manage RGB/BGR channels with Matplotlib and OpenCV, and resize and grayscale images.

Explore how text-to-image diffusion models convert captions into images via a two-stage pipeline: a text embedding with clip and a diffusion decoder. Learn how the prior and decoder stages, including clip, unclip, cross-attention, and diffusion reversal of Gaussian noise, drive image generation and higher-resolution upscaling.

Learn to load a pre-trained diffusion model with the diffusers library, run a cuda-enabled DPM scheduler on a U-Net 2D model, and generate a fake celebrity image via denoising.

Explore how to bring diffusion-based models into pipelines with the diffusers library. Load from Hugging Face, use auto pipelines for text-to-image, image-to-image, and inpainting, and run on CUDA.

Explore how stable video diffusion generates video sequences from text or images, using latent space, temporal layers, and attention, with text-to-image pre-training on large video data sets and high-quality fine-tuning.

Demonstrate stable video diffusion by running the stability AI stable video diffusion XT model on GPUs, installing torch and diffusers, and generating a short video from a resized NASA image.

Discover the i2 v gen xl image-to-video model that combines an image with a text prompt to generate frames and export a video using a diffusers pipeline.

Explore audio data and learn how to process it with Python, then examine Hugging Face audio models for classification, transcription, and generation.

Explore audio data fundamentals, from frequency, amplitude and bit depth to visualization with waveforms, frequency spectrum and spectrograms, and learn practical sampling and resampling concepts.

Learn to load and visualize audio data in python with numpy, matplotlib, and librosa, creating waveforms, spectrograms, and mel spectrograms from an mp3 example for the Hugging Face bootcamp.

Explore audio classification with Hugging Face open source models, using either a feature extractor plus classifier or a simple transformers pipeline, while managing torch, torch audio, and transformers versions.

Combine diarization and transcription models to produce timestamped, speaker-labeled transcripts from raw audio. Use Hugging Face tools to identify speaker segments, then transcribe each segment with a dedicated model.

Explore audio generation with Hugging Face text-to-speech pipelines, using the Bach Small model from Tsuno to turn text into speech and save the output as mp3.

Learn to build graphical user interfaces with Gradio using Python, using input and output components that share a single call, and design simple layouts with interface objects.

Explore Gradio components and interfaces to quickly build graphical user interfaces in Python, using notebooks or scripts, with inputs, outputs, and local hosting.

Delve into Gradio's most common components—text, slider, image, dropdown, json, and label—with hands-on examples and debugging tips to build flexible interfaces.

Learn to design gradient layouts with blocks, rows, and columns, then add tabs and accordions, control sizing, and understand the distinction between code scaffolding and user view.

Learn gradient layouts in Gradio: group components with the gradient group, remove padding, and adjust dimensions with height, width, or CSS, using blocks or interface as needed.

Learn to manually build Gradio interfaces by linking a function, inputs, and outputs with click or change listeners, using blocks, rows, and columns to wire interactions.

Explore Gradio interactions in part two by building multi-output demos with grayscale image processing, Gradio blocks, and multiple functions controlled by separate buttons for add and multiply.

Explore machine learning integrations with Gradio by building a GUI for a Hugging Face ResNet-18 image classifier, loading the processor and model, and mapping logits to labels.

Explore gradient load and pipelines integration to quickly launch a Gradio interface from a Hugging Face model or space, enabling sentiment analysis demos with automatic input and outputs.

Explore Gradio's error, warning, and information components to validate inputs, stop execution with errors, or display warnings and info without breaking the program.

Explore Gradio themes and styling by applying pre-built options like glass, monochrome, and soft, and using the theme builder to customize hundreds of CSS variables for your interfaces.

Trace the origins of git and GitHub from Linus Torvalds' 2005 creation to GitHub's 2007 hosting service. Learn how GitHub's integration with Microsoft and GUI tools complements command line workflows.

Learn how to install git, an open source version control system, across macOS, Linux, and Windows, verify the version with git --version, and set Visual Studio Code as default editor.

Set up your GitHub profile with a matching username and email, sign up for the free tier, install GitHub Desktop and VS Code, and explore GUI and CLI repo workflows.

Configure git locally by setting your username and email with git config, and verify the setup. Prepare to connect GitHub desktop and VS Code to GitHub and create a repository.

Learn to create and clone git repositories, initialize with git init, inspect dot git, use git status, explore public repositories like TensorFlow on GitHub, and clone private repos with tokens.

Learn how to clone private repositories on GitHub using the command line and GitHub desktop. Create a personal access token with repo scope to authorize private clones over https.

Explore the basics of git usage and a basic workflow with git and GitHub for a solo developer on a single branch, including staging, commits, and pushing to GitHub.

Create a git repository from the command line, stage files with git add, and commit with a message. Stage single, multiple, or all files with staging area and VS Code.

Learn how to push local code to a remote GitHub branch using git remote add origin and git push, including handling main vs master and personal access tokens.

Learn how to use git log to view a repository’s commit history, including hash, author, date, and messages. Exit long logs with q.

Learn how to use git fetch and git pull to sync with a remote origin; fetch downloads commits without altering the working directory, while pull updates and merges them.