20 Repos
Tools for managing and tuning model parameters to optimize training and inference performance.
Distinguishing note: Focuses on the configuration of model parameters rather than the architecture definition itself.
Explore 20 awesome GitHub repositories matching artificial intelligence & ml · Hyperparameter Configurations. Refine with filters or upvote what's useful.
This project is a comprehensive framework for the entire lifecycle of transformer-based language models, supporting everything from foundational pretraining to specialized deployment. It provides a modular toolkit for defining neural network architectures, managing data preparation pipelines, and executing training routines across various scales. The framework is designed to handle the full model development process, including supervised fine-tuning, behavioral alignment, and the integration of agentic capabilities. What distinguishes this framework is its focus on efficient training and adva
The framework allows for the configuration of model hyperparameters, such as embedding dimensions and layer counts, to balance training stability, inference speed, and performance.
Tensor2Tensor is a deep learning library built on TensorFlow designed for training and evaluating complex machine learning models. It provides a unified framework for managing the entire model lifecycle, including data ingestion, training execution, and performance evaluation across diverse hardware environments. The library distinguishes itself through a modular architecture that supports multimodal data processing, allowing for the simultaneous analysis of text, audio, and image inputs. It features a central registry system that enables developers to extend the framework with custom models,
Uses structured configuration files to define and manage model hyperparameters for automated parallel search.
SpeechBrain is an all-in-one deep learning toolkit designed for speech and audio processing. Built as a modular library, it provides a structured environment for developing, training, and deploying neural network models across a wide range of tasks, including automatic speech recognition, speaker identification, and audio enhancement. The framework distinguishes itself through a configuration-driven approach that separates model architecture and training hyperparameters from application logic. By utilizing externalized configuration files and standardized recipes, it enables reproducible rese
Organizes training experiments and model settings using a structured configuration language to streamline development workflows.
Dopamine is a reinforcement learning research framework designed for prototyping and testing algorithms across diverse simulated environments. It provides an agent development toolkit that utilizes a flat class hierarchy to facilitate the creation and extension of learning agents. The framework includes a standardization layer via environment wrappers that connect agents to various physics simulations and gaming environments. It also features a high-performance experience replay buffer for storing and sampling transition data to improve training stability, alongside a dedicated hyperparameter
Provides tools for managing and tuning model hyperparameters via external files for reproducible research.
This project is a comprehensive collection of educational examples and reference implementations for building vision and language models using PyTorch. It serves as a deep learning tutorial covering the end-to-end process of developing neural networks, from initial architecture definition to final production deployment. The repository provides detailed guides on implementing a wide range of domain-specific models, including convolutional neural networks for object detection and segmentation, as well as transformer and recurrent architectures for natural language processing. It emphasizes gene
Manages model hyperparameters through a command-line argument parser to facilitate tuning without code changes.
This project is a deep learning research toolkit and generative model library providing implementations of Variational Autoencoders using the PyTorch framework. It serves as a framework for training and evaluating autoencoder architectures to learn latent representations for data reconstruction and the generation of synthetic data samples. The toolkit focuses on unsupervised feature learning and generative model training, featuring a system for mapping external configuration files to model hyperparameters to ensure reproducible experimental runs. It includes mechanisms for tracking training p
Provides a system for managing model hyperparameters via external configuration files to ensure experimental reproducibility.
gpt-neox is a distributed training system and framework for building large-scale autoregressive language models. It implements the transformer architecture and provides a toolkit for training models with billions of parameters by distributing weights across compute clusters. The framework distinguishes itself through extensive support for distributed model parallelism, including pipeline and sequence parallelism, to overcome single-device memory limits. It further supports sparse model architectures using a mixture of experts system with Sinkhorn-based routing. The project covers a broad ran
Enables the definition of model hyperparameters such as layer count, hidden size, and attention head configurations.
This project is a collection of interactive graphical tools designed for monitoring neural network training, latent space mappings, and the internal mechanisms of transformers. It functions as a visual learning environment for understanding how large language models process tokens and an educational tool for analyzing the interactions between generators and discriminators within adversarial networks. The system provides a browser-based transformer architecture visualizer to show the mathematical operations used for token prediction in real time. It also includes a generative adversarial netwo
Allows updating model configuration variables in memory during runtime to observe immediate changes in learning behavior.
MMDetection3D is an open-source toolbox for 3D perception, providing a unified framework for detecting and segmenting objects in three-dimensional environments. It supports a range of core tasks including monocular 3D object detection from single camera images, LiDAR-based 3D object detection from raw point clouds, and multi-modal fusion that combines camera images with LiDAR data. The toolbox also covers point cloud semantic segmentation, assigning class labels to every point in a scan for scene understanding. The project distinguishes itself through a config-driven pipeline that orchestrate
Enables adapting 3D detection pipelines to new datasets through configuration files.
Dieses Projekt ist ein Framework für Natural Language Processing, das sich auf einen generalisierten autoregressiven Pretrainer für unüberwachte Sprachrepräsentation konzentriert. Es implementiert ein Sprachmodell, das Permutations-basiertes Training mit einem Transformer-XL-Backbone kombiniert, um als Langkontext-Textprozessor zu fungieren. Das System zeichnet sich durch die Fähigkeit aus, Textsequenzen zu verarbeiten, die Standardlängenlimits überschreiten, indem es Segment-Level-Recurrence und relative positionale Kodierung verwendet. Es skaliert High-Performance-Pretraining über mehrere GPUs und TPU-Cluster mittels verteilter Trainingsimplementierungen. Die Codebasis deckt den gesamten Machine-Learning-Workflow ab, einschließlich Textbereinigung und Subword-Tokenisierung für die Datenvorverarbeitung sowie aufgabenspezifisches Fine-Tuning für Question Answering, Leseverständnis und Textklassifizierung. Es enthält Utilities für Parameteroptimierung, Learning-Rate-Scheduling und die Evaluierung von Antwortwahrscheinlichkeiten mittels Precision-Recall-Metriken. Das Projekt bietet Konfigurationen für die Verwaltung von Modell-Hyperparametern und hardwarebeschleunigtem Training über mehrere Hosts hinweg.
Provides configurations for layer counts, attention heads, and hidden sizes to ensure consistency across training phases.
MMF is a modular framework for building, training, and evaluating vision-and-language models. It provides a configuration-driven experiment system where model, dataset, and training parameters are defined through composable YAML files, alongside a curated model zoo of pretrained checkpoints for state-of-the-art multimodal architectures. The framework includes a multimodal dataset loader that downloads, processes, and batches vision-and-language data, and a vision-language model trainer supporting distributed training, mixed precision, and checkpoint-based resumption. The framework distinguish
Defines encoder types, dimensions, and classifier settings through structured configuration files.
Dieses Projekt ist ein standardisiertes Boilerplate für Machine-Learning-Experimente und ein Projekt-Template, das PyTorch Lightning mit dem Hydra-Konfigurations-Framework kombiniert. Es bietet eine strukturierte Codebasis zur Organisation von Deep-Learning-Workflows, die speziell darauf ausgelegt ist, hierarchisches Konfigurationsmanagement mit verteiltem Training zu integrieren. Das Template bietet einen spezialisierten Workflow für Hyperparameter-Optimierung und Batch-Experiment-Ausführung, was automatisierte Parameter-Sweeps ohne Änderung des Quellcodes ermöglicht. Es verwendet ein hierarchisches System zur Verwaltung von Einstellungen über YAML-Dateien und Kommandozeilen-Overrides, um reproduzierbare Ergebnisse über verschiedene Experimentläufe hinweg sicherzustellen. Das Projekt deckt breite Funktionsbereiche ab, einschließlich verteiltem Deep-Learning-Training über mehrere Hardware-Beschleuniger, Kapselung von Datenpipelines und Multi-Backend-Experiment-Logging. Es integriert zudem Code-Qualitätsautomatisierung durch Pre-Commit-Hooks, Linter und Formatierer, neben Tools für das Management und die Evaluierung von Modell-Checkpoints.
Tracks sets of hyperparameters through dedicated configurations to maintain a history of optimal settings.
Dieses Projekt ist ein PyTorch-Projekt-Boilerplate und Trainings-Framework, das darauf ausgelegt ist, die Entwicklung von Deep-Learning-Experimenten zu standardisieren. Es bietet ein strukturiertes Verzeichnis-Layout und eine Reihe von Basisklassen, um neue Projekte zu bootstrappen und einen konsistenten Workflow von der Daten-Pipeline-Konstruktion bis zur Modellausführung sicherzustellen. Das Framework zeichnet sich durch einen zentralisierten Konfigurationsmanager für Hyperparameter aus, der Befehlszeilen-Overrides unterstützt, sowie durch eine Hardware-Beschleunigungsschicht zur Verteilung von Rechenaufgaben auf mehrere Grafikprozessoren. Zudem implementiert es eine Basisklassen-Orchestrierungsschicht, um das Mischen von Datensätzen, die Batch-Generierung und die Validierungs-Splits zu automatisieren. Das System deckt ein breites Spektrum an Trainingsfunktionen ab, einschließlich automatisiertem Metrik-Logging, Checkpoint-basierter Status-Serialisierung zum Fortsetzen des Trainings und Ergebnis-Determinismus durch Seed-Synchronisation. Es enthält zudem Tools zur Überwachung des Trainingsfortschritts und zur Implementierung von Early-Stopping basierend auf Performance-Benchmarks.
Provides tools for managing and tuning model hyperparameters via configuration files and command-line flags.
Tiny Universe is an educational monorepo that delivers multiple independent implementations of core AI subsystems as self-contained Jupyter notebooks. It provides from-scratch constructions of foundational architectures including a complete Transformer model built from the original paper specification, a denoising diffusion probabilistic model for image generation, and a ReAct-style autonomous agent framework that equips an LLM with tools for planning and multi-step task execution. The project distinguishes itself by covering the full lifecycle of modern AI systems through hands-on implementa
Configures model hyperparameters like vocabulary size, hidden dimensions, and layer counts.
Dieses Projekt ist ein Framework für die Implementierung und Erforschung von Generative Adversarial Networks (GANs). Es stellt die notwendigen Tools und Hyperparameter bereit, um generative Modelle über verschiedene Datensätze hinweg zu trainieren und zu evaluieren, speziell mit dem Ziel, Ergebnisse aus der akademischen Forschung zu reproduzieren. Das Framework enthält einen Parzen-Dichte-Likelihood-Schätzer zur Berechnung der Log-Likelihood des Modells. Dies ermöglicht die quantitative Bewertung generativer Verteilungen und die Messung der Gesamtleistung des Modells. Die Codebasis deckt Machine-Learning-Forschungsfähigkeiten ab, mit Fokus auf das Training von Adversarial Networks und die Evaluierung synthetischer Datenverteilungen.
Provides tools for managing and tuning hyperparameters via external configuration files for research reproducibility.
PromptWizard ist ein automatisiertes Framework zur systematischen Optimierung von Anweisungen für Large Language Models (LLMs). Es bietet eine strukturierte Pipeline, die iterative Feedback-Schleifen nutzt, um natürliche Sprach-Prompts zu bewerten, zu kritisieren und zu verfeinern, wodurch eine konsistente Leistung bei verschiedenen generativen Aufgaben sichergestellt wird. Das System zeichnet sich durch selbstreflexive Optimierung aus, bei der ein Modell beauftragt wird, seine eigenen Anweisungen basierend auf automatisierten Leistungsmetriken umzuschreiben. Es verbessert die Prompt-Qualität weiter durch die Integration logischer Schlussfolgerungsketten und Experten-Personas, während gleichzeitig Few-Shot-Beispiele kuratiert werden, um die Ausgabe des Modells zu steuern. Über die grundlegende Verfeinerung hinaus enthält das Framework Hilfsprogramme für die Generierung synthetischer Daten, um spezifische Leistungslücken zu schließen und Trainingsdatensätze zu erweitern. Benutzer können die Intensität und den Umfang des Evolutionsprozesses über konfigurierbare Hyperparameter steuern, einschließlich Einstellungen für Mutationsrunden, Batch-Größen und Iterationsanzahlen.
Implements configurable settings to control the scope and batching of the prompt evolution process.
Triangula is a genetic algorithm image stylizer and renderer that transforms raster images into stylized polygonal artwork. It functions as an image-to-SVG converter that optimizes point placement to recreate the shapes and colors of a source image using triangulated polygons. The project utilizes a fitness-based point selection process and genetic algorithm optimization to iteratively evolve vertex positions. This approach minimizes the difference between the generated polygons and the original image through crossover, mutation, and iterative polygon refinement. The system covers the full p
Provides parametric controls for mutation rates and population size to balance generation speed and artistic detail.
This project is a Python software development kit and framework for building applications that integrate with large language models. It serves as a multimodal content generator and vector embedding library, enabling the production and editing of text, images, audio, and video. The toolkit provides specialized capabilities for adapting base models through supervised and reinforcement training. It further distinguishes itself by offering tools for orchestrating complex workflows, including stateful chat sessions, the enforcement of structured output via schemas, and the integration of external
Provides tools to manage and tune model hyperparameters like temperature, presence penalty, and frequency penalty.
This project is a collection of pretrained reinforcement learning agents and training scripts built on Stable Baselines3 and Gymnasium. It provides a framework for training agents to solve specific tasks, managing experiment reproducibility, and deploying pretrained models. The system includes a specialized benchmarking suite and optimization tools for tuning agent settings. It utilizes automated search spaces and distributed trials to maximize performance, while employing bootstrap sampling to generate statistically robust performance metrics and confidence intervals. Broad capabilities cov
Enables defining learning rates and custom policies via external configuration files to tune agent behavior.
This project is an educational resource and step-by-step guide for implementing end-to-end machine learning workflows. It provides a structured walkthrough for managing the entire lifecycle of a predictive modeling project, from initial data cleaning and feature engineering to final model training and performance assessment. The repository utilizes interactive documents to interleave code, data visualizations, and narrative explanations, facilitating a reproducible approach to data science. By following this guided sequence, users can construct and orchestrate pipelines that transform raw dat
Provides declarative configuration of model parameters to optimize predictive outcomes.