A Hierarchical Model for Device Placement

2018-01-01 · ICLR 2018 1 · Azalia Mirhoseini, Anna Goldie, Hieu Pham, Benoit Steiner, Quoc V. Le, Jeff Dean

We introduce a hierarchical model for efficient placement of computational graphs onto hardware devices, especially in heterogeneous environments with a mixture of CPUs, GPUs, and other computational devices. Our method learns to assign graph operations to groups and to allocate those groups to available devices. The grouping and device allocations are learned jointly. The proposed method is trained with policy gradient and requires no human intervention. Experiments with widely-used computer vision and natural language models show that our algorithm can find optimized, non-trivial placements for TensorFlow computational graphs with over 80,000 operations. In addition, our approach outperforms placements by human experts as well as a previous state-of-the-art placement method based on deep reinforcement learning. Our method achieves runtime reductions of up to 60.6% per training step when applied to models such as Neural Machine Translation.

📄 PDF Abstract BibTeX

Code (0)

등록된 구현이 없습니다.

Tasks

Deep Reinforcement LearningMachine TranslationmodelReinforcement Learning Reinforcement Learning (RL)Translation

Accelerate Model Parallel Training by Using Efficient Graph Traversal Order in Device Placement

2022-01-21 · Tianze Wang, Amir H. Payberah, Desta Haileselassie Hagos, Vladimir Vlassov

Modern neural networks require long training to reach decent performance on massive datasets. One common approach to speed up training is model parallelization, where large neural networks are split across multiple devic…

Decision Making Sequential Decision Making

Location- and Orientation-aware Millimeter Wave Beam Selection for Multi-Panel Antenna Devices

2022-03-22 · Sajad Rezaie, João Morais, Elisabeth de Carvalho, Ahmed Alkhateeb 외

While initial beam alignment (BA) in millimeter-wave networks has been thoroughly investigated, most research assumes a simplified terminal model based on uniform linear/planar arrays with isotropic antennas. Devices wit…

GiPH: Generalizable Placement Learning for Adaptive Heterogeneous Computing

2023-05-23 · Yi Hu, Chaoran Zhang, Edward Andert, Harshul Singh 외

Careful placement of a computational application within a target device cluster is critical for achieving low application completion time. The problem is challenging due to its NP-hardness and combinatorial nature. In re…

Edge-computing Graph Neural Network

Learning Generalizable Device Placement Algorithms for Distributed Machine Learning

2019-12-01 · NeurIPS 2019 12 · Ravichandra Addanki, Shaileshh Bojja Venkatakrishnan, Shreyan Gupta, Hongzi Mao 외

We present Placeto, a reinforcement learning (RL) approach to efficiently find device placements for distributed neural network training. Unlike prior approaches that only find a device placement for a specific computat…

BIG-bench Machine Learning Reinforcement Learning Reinforcement Learning (RL)

Placeto: Learning Generalizable Device Placement Algorithms for Distributed Machine Learning

2019-06-20 · Ravichandra Addanki, Shaileshh Bojja Venkatakrishnan, Shreyan Gupta, Hongzi Mao 외