This paper investigates how the configuration of on-device hardware affects energy consumption for neural network inference with regular fine-tuning.
This paper is available on arxiv under CC BY-NC-ND 4.0 DEED license. Authors: Minghao Yan, University of Wisconsin-Madison; Hongyi Wang, Carnegie Mellon University; Shivaram Venkataraman, myan@cs.wisc.edu. Table of Links Abstract & Introduction Motivation Opportunities Architecture Overview Proble Formulation: Two-Phase Tuning Modeling Workload Interference Experiments Conclusion & References A. Hardware Details B. Experimental Results C. Arithmetic Intensity D.
The overhead of performing CBO is also minimal. As shown in Figure 5, CBO only requires around 15 samples to find a near-optimal solution and the optimization procedure can be completed in a few minutes. In cases where a new model is deployed, only a few minutes of overhead are needed to find optimal configurations for the new model.
Nigeria Latest News, Nigeria Headlines
Similar News:You can also read news stories similar to this one that we have collected from other news sources.
PolyThrottle: Energy-efficient Neural Network Inference on Edge Devices: Experimental ResultsThis paper investigates how the configuration of on-device hardware affects energy consumption for neural network inference with regular fine-tuning.
Read more »
PolyThrottle: Energy-efficient Neural Network Inference on Edge Devices: OpportunitiesThis paper investigates how the configuration of on-device hardware affects energy consumption for neural network inference with regular fine-tuning.
Read more »
PolyThrottle: Energy-efficient Neural Network Inference on Edge Devices: MotivationThis paper investigates how the configuration of on-device hardware affects energy consumption for neural network inference with regular fine-tuning.
Read more »
PolyThrottle: Energy-efficient Neural Network Inference on Edge Devices: Arithmetic IntensityThis paper investigates how the configuration of on-device hardware affects energy consumption for neural network inference with regular fine-tuning.
Read more »
ANYmal robot excels in parkour feats thanks to neural network trainingDog-like robot ANYmal's agility is boosted by a new framework, allowing it to tackle a basic parkour course at up to 6 feet per second.
Read more »
Using Machine Learning & Neural Networks to Maximize Solar Energy GloballyOther ways to generate solar energy that just trying to make solar cells efficient potential of luminescent solar concentration
Read more »