A computational model for TensorFlow: an introduction

By Martín Abadi, Michael Isard, Derek Gordon Murray. 2017

89 citations recorded by the source corpus.

TensorFlow is a powerful, programmable system for machine learning. This paper aims to provide the basics of a conceptual framework for understanding the behavior of TensorFlow models during training and inference: it describes an operational semantics, of the kind common in the literature on programming languages. More broadly, the paper suggests that a programming-language perspective is fruitful in designing and in explaining systems such as TensorFlow.

Subfields: Hardware and Architecture, Information Systems and Management, ML Systems & Hardware

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Hardware and Architecture

In Information Systems and Management

Shane Legg

In ML Systems & Hardware

Citations in the corpus

IMPALA: Scalable Distributed Deep-RL with Importance Weighted Actor-Learner Architectures (2018, methodology): “Finally, we also make use of several off the shelf optimisations available in TensorFlow (Abadi et al., 2017) such as preparing the next batch of data for the learner while still performing computation, compiling parts of the computational graph with XLA (a TensorFlow Just-In-Time compiler) and opti”
Relay: a new IR for machine learning frameworks (2018, background): “The semantics provided by these languages have been sketched in previous work [5].”
TFLMS: Large Model Support in TensorFlow by Graph Rewriting (2018, methodology): “Our definition of a computational graph is inspired by that in TensorFlow [1], a popular framework for deep learning.”
LightSeq: A High Performance Inference Library for Transformers (2021, methodology): “Second, current optimizing compilers for deep learning such as TensorFlow XLA (Abadi et al., 2017), TVM (Chen et al.”