This article explains the fundamental differences between static and dynamic computation graphs, compares their memory and performance characteristics, shows how automatic differentiation works in each paradigm, and provides a step‑by‑step implementation of a toy static‑graph AutoDiff engine with Python code examples.
The article presents a comprehensive overview of how graph algorithms are applied in Alibaba's e‑commerce risk control system, detailing six sections that include risk scenario introductions, interaction and product content risk methods, dynamic heterogeneous graph practices, a large‑scale competition, and future research directions.
This article introduces OneFlow Coop, a memory‑optimization technique that jointly optimizes dynamic‑graph recomputation strategies and GPU memory allocation by analyzing existing DTR limitations, proposing recomputable in‑place, op‑guided tensor allocation, and layout‑aware eviction modules, and demonstrating superior experimental results.
The article explains how Go's lock‑free sync.Pool can cut garbage‑collection overhead, shows practical FFmpeg thread‑parameter tuning that balances CPU use and latency for video filtering versus encoding, and compares PaddlePaddle's static and dynamic graph modes, including debugging tips and conversion to static.
This article explains how eBay uses graph neural networks and a heterogeneous‑graph fraud detection framework (xFraud) to improve payment risk assessment, overcome the limitations of traditional machine‑learning models, and effectively identify both individual and organized fraud in a large‑scale e‑commerce environment.
This article explains how PyTorch implements its dynamic autograd graph using the core C++ classes Variable, Function, and Engine, detailing their initialization, inheritance hierarchy, code structures, and the execution flow of backward propagation.
The article explores how chatbots require personalized dense knowledge graphs, dynamic temporal graphs, subjective emotion modeling, integration with external services, and multimodal media support, while also promoting a new NLP book and a related giveaway for readers.
