awesome-repositories.com
Blog
awesome-repositories.com

Découvrez les meilleurs dépôts open-source grâce à notre recherche par IA.

ExplorerRecherches sélectionnéesAlternatives open sourceLogiciels auto-hébergésBlogPlan du site
ProjetÀ proposNotre méthodologiePresseServeur MCP
Mentions légalesConfidentialitéConditions d'utilisation
© 2026 Bringes Technology SRL·VAT RO45896025·hello@awesome-repositories.com
·

16 dépôts

Awesome GitHub RepositoriesRegularization Techniques

Methods for preventing model overfitting by constraining parameter values.

Distinguishing note: Focuses on the general class of regularization methods.

Explore 16 awesome GitHub repositories matching artificial intelligence & ml · Regularization Techniques. Refine with filters or upvote what's useful.

Awesome Regularization Techniques GitHub Repositories

Trouvez les meilleurs dépôts grâce à l'IA.Nous recherchons les dépôts les plus pertinents grâce à l'IA.
  • huggingface/transformersAvatar de huggingface

    huggingface/transformers

    161,630Voir sur GitHub↗

    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

    Injects random noise into token embeddings during the forward pass to enhance model robustness during instruction fine-tuning.

    Pythonaudiodeep-learningdeepseek
    Voir sur GitHub↗161,630
  • d2l-ai/d2l-zhAvatar de d2l-ai

    d2l-ai/d2l-zh

    78,493Voir sur GitHub↗

    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

    Mitigates overfitting in deep neural networks by teaching weight decay and dropout techniques.

    Pythonbookchinesecomputer-vision
    Voir sur GitHub↗78,493
  • d2l-ai/d2l-enAvatar de d2l-ai

    d2l-ai/d2l-en

    29,001Voir sur GitHub↗

    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

    Implements various regularization techniques to prevent overfitting in high-capacity neural networks.

    Pythonbookcomputer-visiondata-science
    Voir sur GitHub↗29,001
  • fastai/fastaiAvatar de fastai

    fastai/fastai

    27,862Voir sur GitHub↗

    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

    Adds activation and temporal regularization to recurrent neural networks to prevent overfitting.

    Jupyter Notebookcolabdeep-learningfastai
    Voir sur GitHub↗27,862
  • trekhleb/homemade-machine-learningAvatar de trekhleb

    trekhleb/homemade-machine-learning

    24,608Voir sur GitHub↗

    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

    Implements L2 regularization to penalize large coefficients and prevent model overfitting.

    Jupyter Notebook
    Voir sur GitHub↗24,608
  • rasbt/deeplearning-modelsAvatar de rasbt

    rasbt/deeplearning-models

    17,427Voir sur GitHub↗

    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

    Integrates dropout and batch normalization to improve training stability and prevent overfitting.

    Jupyter Notebook
    Voir sur GitHub↗17,427
  • facebookresearch/dinov2Avatar de facebookresearch

    facebookresearch/dinov2

    12,987Voir sur GitHub↗

    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

    Applies regularization to encourage uniform distribution of feature embeddings and prevent representation collapse.

    Jupyter Notebook
    Voir sur GitHub↗12,987
  • afshinea/stanford-cs-230-deep-learningAvatar de afshinea

    afshinea/stanford-cs-230-deep-learning

    7,028Voir sur GitHub↗

    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

    Documents dropout, weight penalties, and batch normalization to prevent overfitting in neural networks.

    cheatsheetconvolutional-neural-networksdata-science
    Voir sur GitHub↗7,028
  • kevinmusgrave/pytorch-metric-learningAvatar de KevinMusgrave

    KevinMusgrave/pytorch-metric-learning

    6,328Voir sur GitHub↗

    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

    Provides norm-based penalties to shrink embedding magnitudes during metric learning training.

    Pythoncomputer-visioncontrastive-learningdeep-learning
    Voir sur GitHub↗6,328
  • awslabs/gluon-tsAvatar de awslabs

    awslabs/gluon-ts

    5,200Voir sur GitHub↗

    GluonTS est un framework pour la prévision probabiliste de séries temporelles, conçu pour prédire les valeurs futures sous forme de distributions de probabilité avec des intervalles de confiance. Il prend en charge à la fois l'entraînement de modèle traditionnel et la prévision zero-shot, où des modèles pré-entraînés génèrent des prédictions pour de nouvelles séries sans entraînement supplémentaire. Le projet se distingue en intégrant une grande variété d'approches de prévision dans un flux de travail unifié. Cela inclut des architectures de deep learning telles que les réseaux neuronaux récurrents et les convolutions causales, ainsi que l'intégration de modèles statistiques externes, la bibliothèque Prophet et les packages R. La boîte à outils fournit une surface complète pour l'ingénierie des données de séries temporelles, couvrant la mise à l'échelle des ensembles de données, la division et la transformation des données temporelles brutes en tenseurs. Elle inclut également une suite d'outils d'évaluation pour mesurer la précision des prévisions et les intervalles d'incertitude, ainsi que des utilitaires pour la persistance des ensembles de données utilisant des formats comme Arrow et Parquet. Le framework prend en charge le déploiement de modèles de prévision au sein de l'infrastructure cloud.

    Calculates loss based on activation magnitudes at each timestep to constrain model complexity and prevent overfitting.

    Python
    Voir sur GitHub↗5,200
  • awslabs/gluontsAvatar de awslabs

    awslabs/gluonts

    5,199Voir sur GitHub↗

    GluonTS est une bibliothèque de séries temporelles probabilistes et un framework de prévision par deep learning. Il fournit une boîte à outils pour construire, entraîner et évaluer des architectures de réseau neuronal qui prédisent les valeurs futures sous forme de distributions de probabilité pour quantifier l'incertitude. Le projet se distingue en prenant en charge la prévision zero-shot et en intégrant diverses approches de modélisation, y compris les réseaux neuronaux probabilistes profonds et des wrappers pour des bibliothèques statistiques externes telles que Prophet et R forecast. Il implémente des primitives architecturales spécialisées comme les convolutions causales et les réseaux résiduels inversibles pour empêcher la fuite d'informations et mapper les représentations latentes en distributions de probabilité valides. Le framework couvre une surface d'ingénierie de données complète, y compris la mise à l'échelle des séries temporelles, les transformations bijectives et la modélisation hiérarchique. Il utilise Apache Arrow et Parquet pour la diffusion d'ensembles de données haute performance et la gestion de l'accès aléatoire. Pour l'évaluation des modèles, il inclut une suite d'évaluation pour mesurer la précision des prévisions et la couverture probabiliste en utilisant des métriques comme la perte quantile et les scores de probabilité de rang continu. La bibliothèque prend en charge le déploiement de modèles via l'intégration avec Amazon SageMaker.

    Implements the zoneout regularization mechanism for recurrent neural network cells to prevent overfitting.

    Pythonartificial-intelligenceawsdata-science
    Voir sur GitHub↗5,199
  • chiphuyen/ml-interviews-bookAvatar de chiphuyen

    chiphuyen/ml-interviews-book

    4,523Voir sur GitHub↗

    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

    Compares train and test splits to verify they come from the same underlying distribution.

    HTML
    Voir sur GitHub↗4,523
  • thuml/transfer-learning-libraryAvatar de thuml

    thuml/Transfer-Learning-Library

    3,917Voir sur GitHub↗

    Ce projet est une bibliothèque complète pour le transfer learning et l'adaptation de domaine en vision par ordinateur. Il sert de framework pour aligner les distributions de caractéristiques entre des jeux de données source et cible, une boîte à outils pour la généralisation de domaine, et une bibliothèque pour l'apprentissage semi-supervisé utilisant de petits jeux de données étiquetés et de grands ensembles non étiquetés. La bibliothèque fournit des capacités spécialisées pour l'adaptation de domaine non supervisée, incluant l'utilisation de réseaux adverses, d'architectures basées sur la divergence et de traduction image-à-image pour réduire le décalage de distribution. Elle inclut également des outils pour la généralisation de domaine afin d'assurer la fiabilité du modèle à travers des domaines cibles non vus via le mélange de styles et la minimisation du risque invariant. Le projet couvre une large surface de capacités incluant l'adaptation de tâche et le fine-tuning avec régularisation spécialisée, l'entraînement semi-supervisé via le pseudo-étiquetage et l'apprentissage par cohérence, et la sélection de modèles de transfer learning utilisant des métriques de transférabilité. Il inclut également un gestionnaire de jeux de données pour automatiser l'acquisition et la préparation de benchmarks de vision standardisés. La bibliothèque inclut des utilitaires pour la surveillance et l'observabilité, tels que des visualisations t-SNE et des métriques de distance A pour analyser les distributions de caractéristiques et la divergence de domaine.

    Provides a dual-classifier regularization technique to enhance image classification task adaptation.

    Python
    Voir sur GitHub↗3,917
  • nerfies/nerfies.github.ioAvatar de nerfies

    nerfies/nerfies.github.io

    3,966Voir sur GitHub↗

    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

    Constrains deformation fields to ensure the reconstructed object's movements remain smooth and physically realistic.

    JavaScript
    Voir sur GitHub↗3,966
  • rasmusbergpalm/deeplearntoolboxAvatar de rasmusbergpalm

    rasmusbergpalm/DeepLearnToolbox

    3,868Voir sur GitHub↗

    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

    Implements regularization techniques like weight decay and dropout to improve generalization.

    Matlab
    Voir sur GitHub↗3,868
  • christianversloot/machine-learning-articlesAvatar de christianversloot

    christianversloot/machine-learning-articles

    3,683Voir sur GitHub↗

    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.

    Demonstrates practical ways to integrate dropout and other constraints to prevent model overfitting.

    albertbertclustering
    Voir sur GitHub↗3,683
  1. Home
  2. Artificial Intelligence & ML
  3. Regularization Techniques

Explorer les sous-tags

  • Activation Magnitude PenaltiesRegularization that penalizes the magnitude of network activations at each timestep. **Distinct from Weight Decay Regularization:** Specifically targets the output activations, unlike weight decay which targets model parameters.
  • Distribution ShiftsChanges in data distribution between training and inference.
  • Dual-Classifier RegularizationsRegularization methods utilizing dual-classifier architectures to improve task adaptation. **Distinct from Regularization Techniques:** Specifically addresses dual-classifier penalty mechanisms rather than general parameter constraints like L1/L2.
  • Elastic RegularizationConstraints applied to deformation fields to ensure smooth and physically plausible movements. **Distinct from Regularization Techniques:** Distinct from general parameter regularization as it specifically constrains the geometric plausibility of volumetric deformations.
  • Embedding Regularization4 sous-tagsNoise injection or constraints applied specifically to embedding layers.
  • Hypersphere Embedding RegularizationLoss functions that constrain feature embeddings to a uniform distribution on a hypersphere to prevent representation collapse. **Distinct from Regularization Techniques:** Distinct from general regularization: focuses specifically on embedding uniformity on hyperspheres during training.
  • Normalization and Regularization CombinationsTechniques combining dropout, weight normalization, layer norm, batch norm, and L2 regularization to prevent overfitting and stabilize training. **Distinct from Regularization Techniques:** Distinct from Regularization Techniques: covers the combined application of normalization layers and regularization methods, not just weight penalties.
  • Recurrent Regularization Utilities1 sous-tagApplies activation and temporal regularization specifically to recurrent neural network layers. **Distinct from Regularization Techniques:** Distinct from general regularization: focuses specifically on temporal activation constraints for recurrent architectures.