Datasets:
Formats:
csv
Languages:
English
Size:
1K - 10K
Tags:
arxiv-artifact
reproducibility
research-artifact
computer-science
computer-logic
formal-methods
License:
| set -euo pipefail | |
| export MPLBACKEND="${MPLBACKEND:-Agg}" | |
| export MPLCONFIGDIR="${MPLCONFIGDIR:-tmp/mpl}" | |
| mkdir -p "$MPLCONFIGDIR" | |
| python3 -m pcmt.cli figures --input runs/run-small/traces.csv --output figures | |
| if [[ -f data/real_videos/sample-20s-360p.mp4 && -f runs/detector-video/traces.csv ]]; then | |
| python3 -m pcmt.cli real-figures \ | |
| --video data/real_videos/sample-20s-360p.mp4 \ | |
| --traces runs/detector-video/traces.csv \ | |
| --output figures \ | |
| --use-gpu auto | |
| fi | |
| best_sweep="$( | |
| python3 - <<'PY' | |
| from pathlib import Path | |
| import csv | |
| candidates = [ | |
| Path("runs/remote_gpu/benchmark-sweep-gpu"), | |
| Path("runs/remote_gpu/youcook2-scale2-sweep-gpu"), | |
| Path("runs/remote_gpu/youcook2-batched-sweep-gpu"), | |
| Path("runs/remote_gpu/ave-scale1-sweep-gpu"), | |
| Path("runs/remote_gpu/youcook2-sweep"), | |
| Path("runs/remote_gpu/sweep-detector-gpu-scale1"), | |
| Path("runs/remote_gpu/sweep-detector-gpu"), | |
| Path("runs/sweep-detector"), | |
| ] | |
| def rows(path: Path) -> int: | |
| with (path / "sweep_traces.csv").open(newline="") as handle: | |
| return sum(1 for _ in csv.DictReader(handle)) | |
| valid = [path for path in candidates if (path / "sweep_traces.csv").exists() and (path / "sweep_metrics.json").exists()] | |
| if valid: | |
| print(max(valid, key=rows)) | |
| PY | |
| )" | |
| if [[ -n "$best_sweep" ]]; then | |
| python3 -m pcmt.cli sweep-figures \ | |
| --traces "$best_sweep/sweep_traces.csv" \ | |
| --metrics "$best_sweep/sweep_metrics.json" \ | |
| --output figures | |
| echo "publication_sweep_source=$best_sweep" | |
| fi | |
| if [[ -z "$best_sweep" ]]; then | |
| echo "publication_sweep_source=run-small-fallback" | |
| fi | |
| best_robustness="$( | |
| python3 - <<'PY' | |
| from pathlib import Path | |
| import csv | |
| candidates = [ | |
| Path("runs/remote_gpu/robustness-atlas-gpu"), | |
| Path("runs/robustness-atlas"), | |
| ] | |
| def rows(path: Path) -> int: | |
| with (path / "stability_phase_diagram_values.csv").open(newline="") as handle: | |
| return sum(1 for _ in csv.DictReader(handle)) | |
| valid = [path for path in candidates if (path / "stability_phase_diagram_values.csv").exists()] | |
| if valid: | |
| print(max(valid, key=rows)) | |
| PY | |
| )" | |
| if [[ -n "$best_robustness" ]]; then | |
| python3 -m pcmt.cli robustness-figures --input "$best_robustness" --output figures | |
| echo "publication_robustness_source=$best_robustness" | |
| fi | |