Hepynet

High energy physics, python-based, neural-network assistant framework

Introduction

The goal of the hepynet: perform DNN related high energy physics analysis tasks with simple config files

• for ATLAS Analysis: include supports for various ATLAS analysis jobs

• Config Driven: all tasks defined by a simple config file

• Python Based: codes are written in Python, which is the mainstream language for DNN studies

Installation

pip install hepynet

GPU support

• You can refer to Tensorflow GPU support to set up the environment to use GPU for training

• This is not mandatory, CPU alone is enough to run hepynet

Set Up the Workspace

Please refer to hepynet_example to see how to set up the workspace of hepynet.

Release Note - v0.4.2

• Re-implement auto-tuning functions with Ray (currently, support three hyperparameter-tuning algorithms: Ax, HyperOpt and HEBO)
• Minor issues fixes and improvements