KungFu: Making Training in Distributed Machine Learning Adaptive

Luo Mai, Guo Li, Marcel Wagenländer, Konstantinos Fertakis, Andrei-Octavian Brabete, Peter Pietzuch

Research output: Chapter in Book/Report/Conference proceedingConference contribution


When using distributed machine learning (ML) systems to train models on a cluster of worker machines, users must configure a large number of parameters: hyper-parameters (e.g. the batch size and the learning rate) affect model convergence; system parameters (e.g. the number of workers and their communication topology) impact training performance. In current systems, adapting such parameters during training is ill-supported. Users must set system parameters at deployment time, and provide fixed adaptation schedules for hyper-parameters in the training program.

We describe KungFu, a distributed ML library for Tensor-Flow that is designed to enable adaptive training. KungFu allows users to express high-level Adaptation Policies (APs) that describe how to change hyper- and system parameters during training. APs take real-time monitored metrics (e.g. signal-to-noise ratios and noise scale) as input and trigger control actions (e.g. cluster rescaling or synchronisation strategy updates). For execution, APs are translated into monitoring and control operators, which are embedded in the data flowgraph. APs exploit an efficient asynchronous collective communication layer, which ensures concurrency and consistency of monitoring and adaptation operations.
Original languageEnglish
Title of host publication14th USENIX Symposium on Operating Systems Design and Implementation (OSDI 20)
PublisherUSENIX Association
Number of pages18
ISBN (Print)978-1-939133-19-9
Publication statusPublished - 4 Nov 2020
Event14th USENIX Symposium on Operating Systems Design and Implementation - Banff, Canada
Duration: 4 Nov 20206 Nov 2020


Symposium14th USENIX Symposium on Operating Systems Design and Implementation
Abbreviated titleOSDI 2020
Internet address


Dive into the research topics of 'KungFu: Making Training in Distributed Machine Learning Adaptive'. Together they form a unique fingerprint.

Cite this