XReyRobert's picture
Publish Smoke24 agentic benchmark dataset
4f24bd8 verified
|
Raw
History Blame Contribute Delete
1.07 kB
# Migration From `qwopus-gptq-pro-runpod`
The first extraction copied only aggregate, dashboard-ready artifacts:
- `benchmarks/terminal_bench/derived/*with_external_public*` into `data/public/derived`
- selected public metric CSVs into `reports/public`
- selected public PNGs into `reports/public/images`
- the Smoke24 task list into `data/public/task-selection`
- the dynamic dashboard template into `dashboard/templates`
Not migrated:
- `benchmarks/terminal_bench/runs`
- raw trial logs, trajectories, recordings, model outputs, or prompt/response payloads
- IBM Bob internal-only reports
- temporary quantization probe artifacts
- private LiteLLM endpoint configuration
Next migration steps:
1. Move raw-run extractors behind `data/raw` inputs with explicit opt-in.
2. Add a sanitized runner manifest format for future public benchmark launches.
3. Split internal-only report generation into a separate private branch or private repo if needed.
4. Keep the HF Dataset as the canonical data/report package and the HF Space as the canonical interactive view.