Advanced Deep Learning with Modern C++: Architecting, Training, and Deploying Neural Systems Using PyTorch C++, Flashlight, and ONNX

Jae-Lin, Min

/ ISBN 13: 9798273859401

Editore: Independently published, 2025

Codice articolo 51868327-n

Riassunto:

Volume II – Advanced Deep Learning with Modern C++
Architecting, Training, and Deploying Neural Systems with PyTorch C++, Flashlight, and ONNX

Master the cutting edge of high-performance deep learning engineering with Modern C++. This advanced volume takes you far beyond Python workflows, showing you how to design, optimize, and deploy neural systems directly in C++ using PyTorch’s C++ API, Facebook’s Flashlight framework, and ONNX Runtime.

From classic architectures like CNNs and RNNs to state-of-the-art Transformers, you’ll learn how to implement, train, and fine-tune models with full control over memory, performance, and execution. The book explores mixed-precision training, model quantization, distributed training strategies, and advanced optimization techniques tailored for production-grade systems.

You’ll also build complete inference pipelines, mastering ONNX export, runtime integration, CUDA acceleration, cuDNN optimizations, and TensorRT deployment for real-time, low-latency applications. Every chapter is designed to help experienced developers engineer deep learning systems that are fast, scalable, and production ready.

Keywords: deep learning with C++, PyTorch C++ frontend, neural network engineering, ONNX Runtime, Flashlight AI, CUDA optimization, Transformer models, CNNs, RNNs, model quantization, TensorRT deployment, GPU computing, high-performance AI systems.

Dati bibliografici

Titolo: Advanced Deep Learning with Modern C++: ...
Casa editrice: Independently published
Data di pubblicazione: 2025
Legatura: Brossura
Condizione: New