Benchmarks

June 21, 2026 · View on GitHub

Paper-scale, dataset-backed reproductions, grouped by the two LEVI entry points. New to LEVI? Start with ../quickstart/ instead.

Install uv first (https://docs.astral.sh/uv/getting-started/installation/), then uv sync (or uv sync --extra examples for all example dependencies).

code/ — evolve_code

The paper-scale circle packing plus the seven ADRS Leaderboard problems (LLM-guided algorithm discovery across systems-optimization tasks).

code/circle_packing/ packs n=26 circles in a unit square — self-contained, no dataset. run.py defaults to OpenRouter + a local Qwen at http://localhost:8000/v1; run_claude.py runs it via the Claude Code CLI.

The ADRS problems share one entrypoint (uv run python run.py) and mostly need the ADRS dataset:

git clone https://github.com/cmu-db/ADRS-Leaderboard.git
export ADRS_EXAMPLE_DATA_ROOT=/path/to/ADRS-Leaderboard
ProblemOptimizesBudgetADRS data
cant_be_late/Spot vs on-demand scheduling under deadlines$4.50Yes
cant_be_late_multi/…same, switching across regions$5.00Yes
cloudcast/Broadcast topology across AWS/Azure/GCP$3.00Yes
eplb/Place 64 MoE experts across 288 GPU slots$4.50Yes
llm_sql/Reorder CSV columns for prefix hit rate$4.50Yes¹
prism/Assign ML models to GPUs to cut KV pressure$4.50No
txn_scheduling/Order 100 transactions to minimize makespan$13.00No

¹ run problems/llm_sql/download_datasets.sh in your ADRS-Leaderboard checkout.

Extra deps: uv sync --extra example-eplb (torch), uv sync --extra example-llm-sql (pandas).

Suggested picks: prism/ (no dataset, OpenRouter throughout), cloudcast/ (small budget), cant_be_late/ (exercises prompt_opt), cant_be_late_multi/ (richest: custom init, behavior features, punctuated equilibrium, prompt_opt).

prompts/ — evolve_prompts

Larger prompt-evolution setups, close in shape to GEPA's harness. These pull HuggingFace datasets and assume an OpenRouter key for the proposer/judge models.

Model configuration

Examples use LiteLLM model ids. The published configs split paradigm models (creative exploration) from mutation models (cheap iteration):

  • Paradigm: openrouter/google/gemini-3-flash-preview
  • Mutation: openrouter/xiaomi/mimo-v2-flash, local Qwen 30B (assumed at http://localhost:8000/v1)
export OPENROUTER_API_KEY=your-key-here
export OPENAI_API_KEY=your-key-here   # if using OpenAI models