Implementation:Norrrrrrr lyn WAInjectBench Conda Environment Setup

Knowledge Sources	WAInjectBench Conda Environments
Domains	Environment_Management, GPU_Computing
Last Updated	2026-02-14 16:00 GMT

Overview

Concrete tool for provisioning a reproducible Python environment and selecting GPU devices, provided by conda and the os standard library.

Description

The WAInjectBench project uses a single environment.yml file (315 lines) that specifies all conda and pip dependencies needed across the entire benchmark. At runtime, both main_text.py and main_image.py accept a --gpu CLI argument and set os.environ["CUDA_VISIBLE_DEVICES"] before any CUDA-dependent library is imported.

Usage

Run conda environment creation once before first use. The GPU selection happens automatically each time an experiment script is invoked via the --gpu argument.

Code Reference

Source Location

• Repository: WAInjectBench
• File: environment.yml (L1-315), main_text.py (L50), main_image.py (L57)

Signature

# Environment creation (CLI)
conda env create -f environment.yml
conda activate wainjectbench

# GPU selection (runtime)
os.environ["CUDA_VISIBLE_DEVICES"] = gpu  # gpu: str, e.g. "0" or "0,1"

Import

import os

I/O Contract

Inputs

Name	Type	Required	Description
environment.yml	File	Yes	Conda environment specification with pinned dependencies
gpu	str	Yes	CUDA device ID(s), passed via --gpu CLI argument (default "9" for text, "0" for image)

Outputs

Name	Type	Description
Conda environment	Virtual environment	Activated environment with all packages installed
CUDA_VISIBLE_DEVICES	Environment variable	Restricts GPU visibility for the current process

Usage Examples

Full Setup

# 1. Create the conda environment from the spec file
conda env create -f environment.yml

# 2. Activate the environment
conda activate wainjectbench

# 3. Run a text detection experiment on GPU 0
python main_text.py --detector promptguard --gpu 0

# 4. Run an image detection experiment on GPU 1
python main_image.py --detector embedding-i --gpu 1

Related Pages

Implements Principle

• Principle:Norrrrrrr_lyn_WAInjectBench_Environment_Setup

Requires Environment

• Environment:Norrrrrrr_lyn_WAInjectBench_Conda_Python_39_CUDA_Environment