Awesome GitHub RepositoriesLayer-Wise Model Assembly

GPT-Neo is an open-source distributed training framework designed for scaling GPT-2 and GPT-3-style language models across multiple devices using mesh-tensorflow for model parallelism. It provides the infrastructure to train transformer-based language models with billions of parameters across distributed computing environments, making large-scale language model research accessible outside of proprietary systems. The framework supports training both autoregressive GPT-style models and masked language models like BERT or RoBERTa, with configurable masking strategies and token handling. It inclu

MergeKit is a toolkit for combining multiple pre-trained large language models into a single entity using algorithmic blending. It provides a specialized system for parameter interpolation and weight extraction to unify model capabilities. The project distinguishes itself through an evolutionary merge optimizer that tunes parameters based on quantitative evaluation metrics. It also features a mixture of experts orchestrator capable of converting dense models into sparse architectures and a tokenizer alignment tool for transplanting embeddings between different models. The toolkit covers a br

Mergekit is a toolkit for combining multiple pretrained large language models into a single model. It functions as an architecture assembler and merging system that transfers capabilities between models using weighted algorithms and layer-wise assembly without requiring additional training. The project provides specialized utilities for extracting low-rank approximations from fine-tuned models to create portable parameter updates. It also includes a framework for converting dense language models into a mixture of experts architecture by constructing gating mechanisms to route inputs to specia

Liger-Kernel is a collection of pre-built fused Triton kernels and patching utilities designed to accelerate large language model training. It provides drop-in kernel replacements for common LLM operations such as RMSNorm, cross-entropy loss, and attention, enabling increased throughput and reduced memory usage while preserving bitwise-exact gradients. The project serves as a toolkit for composing custom model architectures from individual optimized kernels and for patching pre-existing models with minimal code changes. The project distinguishes itself through its ability to perform runtime m

Ce projet est une ressource pédagogique complète et un manuel de tutoriels pour construire, entraîner et déployer des modèles de machine learning avec TensorFlow 2. Il sert de guide d'apprentissage structuré couvrant les concepts fondamentaux du deep learning, notamment les architectures de réseaux de neurones, la différenciation automatique et les opérations sur les tenseurs. Le manuel fournit des conseils techniques pour optimiser l'efficacité de l'exécution via la gestion de la mémoire GPU, l'entraînement distribué et la quantification de modèles. Il inclut également des guides détaillés pour construire des pipelines de données haute performance et exporter des modèles vers des serveurs de production, des appareils mobiles et des navigateurs web. Le contenu couvre un large éventail de capacités, incluant le développement de modèles avec des réseaux convolutifs et récurrents, l'implémentation de fonctions de perte et de couches personnalisées, ainsi que l'utilisation de modèles pré-entraînés pour le transfer learning. Il aborde également les stratégies de déploiement pour les appareils edge et l'utilisation d'environnements d'exécution cloud pour l'accélération matérielle. La ressource est implémentée sous forme d'une collection de Jupyter Notebooks.