Awesome GitHub RepositoriesRegularization Techniques

Transformers is a comprehensive library for machine learning that provides a unified interface for training, fine-tuning, and deploying transformer-based models. It supports a wide range of tasks, including text classification, language modeling, question answering, and sequence-to-sequence translation, while offering specialized architectures for both text and vision processing. The framework includes tools for managing the entire model lifecycle, from data preprocessing and tokenization to distributed training and inference. The library features extensive support for model optimization and

This project is an open-source, interactive educational platform designed to teach deep learning through a comprehensive, code-first curriculum. It provides a structured learning path that covers foundational mathematics, modern neural network architectures, and practical optimization techniques, enabling practitioners to master complex artificial intelligence concepts through hands-on experimentation. The platform distinguishes itself by integrating technical explanations with executable Jupyter notebooks. This design allows readers to modify code and hyperparameters in real-time, facilitati

This project is an educational platform and research toolkit designed to teach deep learning through a combination of mathematical theory, visual diagrams, and executable code. It provides a comprehensive environment for building, training, and evaluating neural networks, grounding complex concepts in interactive computational notebooks that allow for hands-on experimentation. The framework distinguishes itself by interleaving theoretical foundations—including linear algebra, calculus, and probability—with practical implementations across multiple industry-standard libraries. It supports flex

Fastai is a high-level deep learning library built on PyTorch that provides a unified interface for managing the entire machine learning lifecycle. It functions as a comprehensive training toolkit, abstracting hardware management and automating complex training loops to simplify the construction and execution of neural network models. The framework is distinguished by its notebook-centric development environment and a type-dispatching data pipeline that automatically applies transformations based on input data formats. It emphasizes transfer learning through discriminative layer-wise optimiza

This project provides a collection of machine learning algorithms implemented from scratch in Python. It serves as an educational resource using interactive notebooks that combine code with mathematical explanations to demonstrate the first principles of data science. The repository includes reference implementations for neural networks, such as multilayer perceptrons with backpropagation, and supervised learning models including linear and logistic regression. It also covers unsupervised learning through k-means clustering and Gaussian anomaly detection. The codebase covers a broad range of

This repository is an educational collection of deep learning implementations designed to demonstrate the fundamental principles of neural network architecture and optimization. It provides a comprehensive resource for understanding machine learning through hands-on code examples, ranging from basic multilayer perceptrons to complex generative models. The project distinguishes itself by emphasizing the manual construction of models, including the implementation of backpropagation from scratch to illustrate core mathematical mechanics. It covers a wide array of architectural design patterns, s

DINOv2 is a self-supervised vision transformer foundation model designed to generate high-quality visual representations from raw image data. By leveraging large-scale unlabelled datasets, the framework learns to extract robust numerical embeddings that serve as inputs for various machine learning and analysis workflows. The model distinguishes itself through a teacher-student training framework that utilizes centered and sharpened soft probability distributions to align feature maps across multiple image crops. It incorporates a masking strategy that forces the model to reconstruct missing i

This repository collects illustrated single-page cheat sheets that compress the core topics of Stanford's CS 230 deep learning course into visual reference summaries. The collection covers convolutional neural networks, recurrent neural networks, and practical training techniques, pairing schematic diagrams with mathematical notation to bridge intuition and formal understanding. The cheat sheets are organized by subject area and link related concepts across topics, such as connecting vanishing gradients to LSTM gates, to reinforce the full deep learning workflow. Practical training advice on

PyTorch Metric Learning is an open-source library for training neural networks to produce similarity-preserving embedding spaces. It provides a modular framework where interchangeable loss functions, mining strategies, and evaluation tools can be composed to learn representations that map similar items to nearby points and dissimilar items to distant points in the embedding space. The library distinguishes itself through a highly configurable architecture that separates concerns across several interchangeable components. Users can assemble custom loss functions from pluggable distance metrics

GluonTS ist ein Framework für probabilistische Zeitreihenprognosen, das darauf ausgelegt ist, zukünftige Werte als Wahrscheinlichkeitsverteilungen mit Konfidenzintervallen vorherzusagen. Es unterstützt sowohl das traditionelle Modelltraining als auch Zero-Shot-Forecasting, bei dem vortrainierte Modelle Vorhersagen für neue Serien ohne zusätzliches Training generieren. Das Projekt zeichnet sich durch die Integration einer Vielzahl von Prognoseansätzen in einen einheitlichen Workflow aus. Dies umfasst Deep-Learning-Architekturen wie rekurrente neuronale Netze und kausale Konvolutionen sowie die Integration externer statistischer Modelle, der Prophet-Bibliothek und R-Paketen. Das Toolkit bietet eine umfassende Oberfläche für das Zeitreihen-Data-Engineering, die Datensatzskalierung, -aufteilung und die Transformation roher Zeitdaten in Tensoren abdeckt. Es enthält zudem eine Suite von Evaluierungstools zur Messung von Prognosegenauigkeit und Unsicherheitsintervallen sowie Hilfsmittel zur Datensatzpersistenz unter Verwendung von Formaten wie Arrow und Parquet. Das Framework unterstützt die Bereitstellung von Prognosemodellen innerhalb der Cloud-Infrastruktur.

GluonTS ist eine probabilistische Zeitreihenbibliothek und ein Deep-Learning-Prognose-Framework. Es bietet ein Toolkit zum Aufbau, Training und zur Evaluierung neuronaler Netzwerkarchitekturen, die zukünftige Werte als Wahrscheinlichkeitsverteilungen vorhersagen, um Unsicherheit zu quantifizieren. Das Projekt zeichnet sich durch die Unterstützung von Zero-Shot-Forecasting und die Integration diverser Modellierungsansätze aus, einschließlich tiefer probabilistischer neuronaler Netze und Wrapper für externe statistische Bibliotheken wie Prophet und R forecast. Es implementiert spezialisierte architektonische Primitiven wie kausale Konvolutionen und invertierbare Residual-Netzwerke, um Informationslecks zu verhindern und latente Repräsentationen in gültige Wahrscheinlichkeitsverteilungen abzubilden. Das Framework deckt eine umfassende Data-Engineering-Oberfläche ab, einschließlich Zeitreihenskalierung, bijektiver Transformationen und hierarchischer Modellierung. Es nutzt Apache Arrow und Parquet für hochperformantes Datensatz-Streaming und Random-Access-Management. Zur Modellbewertung enthält es eine Evaluierungssuite zur Messung von Prognosegenauigkeit und probabilistischer Abdeckung unter Verwendung von Metriken wie Quantile Loss und Continuous Rank Probability Scores. Die Bibliothek unterstützt die Modellbereitstellung durch Integration mit Amazon SageMaker.

This project is a collection of comprehensive guides and reference materials designed for technical interviews, machine learning system design, and professional development. It serves as a technical knowledge base and a career coaching manual, providing structured resources to help candidates navigate the machine learning hiring landscape. The resource distinguishes itself by offering detailed frameworks for comparing industry roles, analyzing company types, and planning long-term career progression. It provides specific guidance on evaluating employer organizational health, identifying resea

Dieses Projekt ist eine umfassende Bibliothek für Transfer Learning und Domain Adaptation im Bereich Computer Vision. Sie dient als Framework für die Angleichung von Feature-Verteilungen zwischen Quell- und Zieldatensätzen, als Toolkit für Domain Generalization und als Bibliothek für semi-überwachtes Lernen unter Verwendung kleiner gelabelter Datensätze und großer ungelabelter Mengen. Die Bibliothek bietet spezialisierte Funktionen für unüberwachte Domain Adaptation, einschließlich der Verwendung von Adversarial Networks, Diskrepanz-basierten Architekturen und Image-to-Image-Translation, um Verteilungs-Mismatch zu reduzieren. Sie enthält zudem Tools für Domain Generalization, um die Modellzuverlässigkeit über ungesehene Ziel-Domains hinweg durch Style-Mixing und Invariant Risk Minimization sicherzustellen. Das Projekt deckt ein breites Funktionsspektrum ab, einschließlich Task Adaptation und Fine-Tuning mit spezialisierter Regularisierung, semi-überwachtem Training durch Pseudo-Labeling und Consistency Learning sowie der Auswahl von Transfer-Learning-Modellen unter Verwendung von Transferability-Metriken. Es enthält zudem einen Datensatz-Manager zur Automatisierung der Akquise und Vorbereitung standardisierter Vision-Benchmarks. Die Bibliothek enthält Dienstprogramme für Monitoring und Observability, wie t-SNE-Visualisierungen und A-Distanz-Metriken, um Feature-Verteilungen und Domain-Diskrepanzen zu analysieren.

This project is a computer vision pipeline and volumetric rendering system used to transform photos and videos into high-fidelity 3D models. It implements a deformable neural radiance field framework that optimizes deformation fields to represent non-rigid moving subjects in three dimensions. The system utilizes volumetric deformation fields to map 3D coordinates from a static canonical space to a deformed state. This allows for the reconstruction of photorealistic scenes and the synthesis of high-fidelity images from camera perspectives not present in the original input data. The framework

DeepLearnToolbox is a research-oriented framework for constructing, training, and optimizing hierarchical neural networks within the Matlab and Octave environments. It provides a modular set of tools for building diverse network topologies, including feedforward, convolutional, and deep belief architectures, using native matrix-based numerical computation. The library distinguishes itself through its support for layer-wise unsupervised pre-training, which establishes initial weights for deep models before supervised fine-tuning. It incorporates stochastic gradient descent and backpropagation

This project is a machine learning educational archive and technical documentation collection. It serves as a deep learning tutorial series and implementation guide, providing theoretical explanations and practical walkthroughs for constructing and optimizing neural networks. The content focuses on the design and construction of diverse model architectures, including convolutional neural networks, Long Short-Term Memory networks, and generative adversarial networks. It details specific implementation patterns for autoencoders, sentiment analysis models, and various classification approaches.