// Set up a rank_llm development environment. Use when someone is onboarding, setting up a fresh clone, choosing extras such as cloud, api, local, or pyserini, or troubleshooting whether the packaged rank-llm CLI is ready.
| name | rank-llm-install |
| description | Set up a rank_llm development environment. Use when someone is onboarding, setting up a fresh clone, choosing extras such as cloud, api, local, or pyserini, or troubleshooting whether the packaged rank-llm CLI is ready. |
Development environment setup for rank_llm and its packaged rank-llm CLI.
pyserinipython3 --version
command -v uv
If uv is on PATH, use it silently. If not, ask once whether to install uv or continue with a fallback pip or conda path.
If no pyproject.toml is present:
git clone git@github.com:castorini/rank_llm.git && cd rank_llm
uv python install 3.11
uv venv --python 3.11
source .venv/bin/activate
uv sync --group dev --extra cloud --extra api
python3 -m venv .venv
source .venv/bin/activate
pip install -e ".[cloud,api]"
pip install pre-commit
conda create -n rankllm python=3.11 -c conda-forge -y
conda activate rankllm
pip install -e ".[cloud,api]"
uv pip install rank-llm
rank-llm --output json doctor
rank-llm --help
uv run pre-commit install --install-hooks --hook-type pre-commit --hook-type pre-push
references/extras.md - Optional extras and when to add themrank-llm is the canonical CLI entry point. The legacy scripts under src/rank_llm/scripts/ remain available, but they are compatibility wrappers.cloud is the default hosted-provider stack and api is the lightweight HTTP server stack. This pair is the best default for most contributor workflows.pyserini requires Java 21 and is only needed for retrieval or evaluation workflows.local, vllm, and sglang pull in heavier inference stacks. Do not install them by default.Use when working with the rank-llm CLI: rerank, evaluate, analyze, retrieve-cache, serve, validate, prompt, view, describe, schema, or doctor. Covers entry points, common flags, JSONL and TREC artifacts, and end-to-end retrieval plus reranking workflows.
Use when analyzing rank_llm evaluation outputs across runs or models. Covers aggregated trec_eval JSONL files, response-analysis metrics, retrieval-cache handoff files, and side-by-side comparison of stored evaluation artifacts.
Use when validating rank_llm artifacts after rerank, retrieve-cache, or related CLI workflows. Checks JSONL integrity, TREC formatting, candidate shape, invocation-history structure, and duplicate query identifiers. Wraps rank-llm view plus custom assertions.