OpenSpace/benchmarks/gdpval/README.md
2026-07-17 11:43:42 +08:00

78 lines
3 KiB
Markdown

# GDPVal Benchmark
Benchmark for evaluating OpenSpace on [GDPVal](https://huggingface.co/datasets/openai/gdpval) (220 occupational tasks across 44 occupations and 9 sectors). Measures token savings from skill accumulation by running each task twice:
- **Phase 1** — Cold start. Skills accumulate as tasks run sequentially.
- **Phase 2** — Warm start. Re-run all tasks with the full Phase 1 skill library.
Evaluation uses [ClawWork](https://github.com/HKUDS/ClawWork)'s LLM evaluator (same rubrics, same 0.6 payment cliff).
## Project Layout
```
parent/
├── OpenSpace/ ← this repo
│ └── benchmarks/gdpval/ ← this directory
└── ClawWork/ ← required
├── eval/meta_prompts/ ← evaluation rubrics
└── livebench/data/agent_data/ ← ClawWork agent results (for leaderboard)
```
## Setup
```bash
pip install -e . && pip install datasets
git clone https://github.com/HKUDS/ClawWork.git ../ClawWork
pip install -r benchmarks/gdpval/requirements-eval.txt
export OPENROUTER_API_KEY="sk-or-..."
export EVALUATION_API_KEY="sk-..."
```
## Run
```bash
python -u -m benchmarks.gdpval.run_benchmark \
--task-list benchmarks/gdpval/tasks_50.json \
--model openrouter/qwen/qwen3.5-plus-02-15 \
--use-clawwork-productivity \
--clawwork-root ../ClawWork \
--resume
```
Key flags: `--phase1-only`, `--phase2-only`, `--no-eval`, `--concurrency N`, `--max-tasks N`, `--prefetch-only`, `--dry-run`.
## Included Data
```
skills/ # evolved skills
.openspace/openspace.db # skill & tool quality DB (auto-generated during evolution)
```
`skills/` contains the full skill library produced by evolution — each subdirectory holds a `SKILL.md`. `.openspace/openspace.db` tracks skill lineage, tool quality records, and execution analyses accumulated across benchmark runs.
## Output
```
results/<run_name>/
├── phase1_results.jsonl # per-task Phase 1
├── phase2_results.jsonl # per-task Phase 2
├── comparison.jsonl # Phase 1 vs Phase 2 deltas
├── summary.json # aggregate statistics
├── skills_snapshot.json # skills after Phase 1
├── config.json # run config
├── workspace/ # agent working directories
└── recordings/ # execution trajectories
```
## Analyze
```bash
python -m benchmarks.gdpval.calc_subset_performance
```
This command produces leaderboard (OpenSpace vs ClawWork agents), head-to-head comparison, and token savings breakdown.
## Task List
- **[`tasks_50.json`](tasks_50.json)** — 50 task IDs (deterministic subset of GDPVal-220).
- **[`tasks_50_full.jsonl`](tasks_50_full.jsonl)** — Full task data downloaded from [HuggingFace](https://huggingface.co/datasets/openai/gdpval). One JSON object per line with all original fields (`task_id`, `sector`, `occupation`, `prompt`, `reference_files`, `reference_file_urls`, `rubric_json`, etc.). Covers all 9 sectors and 44 occupations.