Meta-Eval/EvalHarness
3
0
Fork 0
Code Issues Pull Requests Actions Packages Projects Releases Wiki Activity

Per-benchmark Excel: each <bench>/<bench>.xlsx (4 sheets) alongside report.jsonl

Browse Source 
Co-Authored-By: Claude <noreply@anthropic.com>
This commit is contained in:
sora 2026-09-11 09:58:57 +00:00
parent f118f4092b
commit 52872bcbc7
1 changed files with 8 additions and 0 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

8
evalharness/cli.py
View File

@ -690,6 +690,14 @@ def _cmd_eval_run(args) -> int:
bench_dir = _P(out_dir) / name bench_dir = _P(out_dir) / name
bench_dir.mkdir(parents=True, exist_ok=True) bench_dir.mkdir(parents=True, exist_ok=True)
report.save(str(bench_dir / 'report.jsonl')) report.save(str(bench_dir / 'report.jsonl'))
# per-benchmark Excel (Summary/Perf/Categories/Samples sheets)
try:
from evalharness.viz import render as _r2
_r2([report], style='excel',
out=str(bench_dir / f'{name}.xlsx'))
except Exception:
pass # excel is a nice-to-have, never block the run
if args.verbose: if args.verbose:
if not _print_result_panel(console, report, _time.time() - t0): if not _print_result_panel(console, report, _time.time() - t0):
print(render(report, style=args.style)) print(render(report, style=args.style))