Installation

Requirements

  • Python >=3.10
  • pip (latest version recommended)
  • Internet access for dependency resolution

Recommended:

  • An isolated environment created with venv or virtualenv

Install From PyPI

Use this path when you want the published package and CLI quickly.

bash
python -m venv .venv
source .venv/bin/activate          # Windows: .venv\Scripts\activate
python -m pip install --upgrade pip
pip install nexa-gauge

Optional Hugging Face dataset support:

bash
pip install "nexa-gauge[huggingface]"

Install From Source

Use this path for local development, debugging, and contributions.

bash
git clone https://github.com/Sardhendu/nexa-gauge.git
cd nexa-gauge

python -m venv .venv
source .venv/bin/activate          # Windows: .venv\Scripts\activate
python -m pip install --upgrade pip
pip install -e .

Install contributor dependencies:

bash
pip install -e ".[dev]"

Verify Installation

bash
nexa-gauge --help
nexagauge run --help
nexagauge estimate --help

If these commands print help output, the CLI is installed correctly.

Runtime Configuration

For LLM-backed evaluation, create a local environment file:

bash
cp .env.example .env

Set at least:

  • OPENAI_API_KEY
  • LLM_MODEL (or pass model overrides through CLI flags)

Smoke Test

bash
nexagauge estimate grounding --input sample.json --limit 3
nexagauge run eval --input sample.json --limit 3 --output-dir ./report

Upgrade Or Uninstall

bash
pip install --upgrade nexa-gauge
pip uninstall nexa-gauge

Troubleshooting

  • command not found: nexa-gauge Activate your environment and reinstall the package.

  • Hugging Face dataset errors for hf://... Install extras: pip install "nexa-gauge[huggingface]".

  • API or authentication errors Verify provider keys in .env, including OPENAI_API_KEY.

Sources

Implementation Sources

  • Packaging metadata: pyproject.toml
  • Environment template: .env.example
  • CLI entrypoint: nexa_gauge_api.main:main