HiQ -- A Declarative, Non-intrusive, Dynamic and Transparent Observability and Optimization System

Fuheng Wu, Ivan Davchev, Jun Qian
4/26/2023
cs.DCcs.AIcs.LG

Official code: oracle/hiq (73 stars, last commit 2026-02-20).

Abstract

This paper proposes a non-intrusive, declarative, dynamic and transparent system called `HiQ` to track Python program runtime information without compromising on the run-time system performance and losing insight. HiQ can be used for monolithic and distributed systems, offline and online applications. HiQ is developed when we optimize our large deep neural network (DNN) models which are written in Python, but it can be generalized to any Python program or distributed system, or even other languages like Java. We have implemented the system and adopted it in our deep learning model life cycle management system to catch the bottleneck while keeping our production code clean and highly performant. The implementation is open-sourced at: [https://github.com/oracle/hiq](https://github.com/oracle/hiq).

View on arXivView PDF

Code Implementations(1)

oracle/hiqOfficial100%

HiQ - Observability And Optimization In Modern AI Era

7311CSS, HTMLMar 4, 20228 months agoNOASSERTION
loggingmonitoringobservabilitypythontracing

Similar Papers

Python Wrapper for Simulating Multi-Fidelity Optimization on HPO Benchmarks without Any Wait

80% match1 repo72023

Fast Benchmarking of Asynchronous Multi-Fidelity Optimization on Zero-Cost Benchmarks

76% match1 repo72024

DeepContext: A Context-aware, Cross-platform, and Cross-framework Tool for Performance Profiling and Analysis of Deep Learning Workloads

75% match2024

ART: Actually Robust Training

75% match1 repo442024

Enabling Performant and Flexible Model-Internal Observability for LLM Inference

75% match1 repo212026

Asynchronous Decentralized Bayesian Optimization for Large Scale Hyperparameter Optimization

75% match2022

Cite this paper

@article{wu2023declarative,
  title  = {HiQ -- A Declarative, Non-intrusive, Dynamic and Transparent Observability and Optimization System},
  author = {Fuheng Wu and Ivan Davchev and Jun Qian},
  year   = {2023},
  eprint = {2304.13302},
  archivePrefix = {arXiv},
  url    = {http://arxiv.org/abs/2304.13302v1}
}

Add a badge to your README

Shows whether official code is linked for this paper.

Code of Paper badge
[![Code of Paper](https://codeofpaper.com/badge/2304.13302.svg)](https://codeofpaper.com/paper/2304.13302)

Discussion