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

Fix UnboundLocalError on the report-reuse path

Browse Source 
The progress-reporter setup lived inside the generation branch; reuse
skipped it but the success path still advanced the overall bar ->
'cannot access local variable progress_reporter'. Setup now runs for
every path (reuse included).

Co-Authored-By: Claude <noreply@anthropic.com>
This commit is contained in:
sora 2026-09-14 06:24:10 +00:00
parent 7eb7b26786
commit 6f19719df2
1 changed files with 30 additions and 26 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

56
evalharness/cli.py
View File

@ -755,35 +755,39 @@ def _cmd_eval_run(args) -> int:
except Exception:
pass # unreadable/stale report: score normally
# progress reporter setup runs for EVERY path (report reuse
# included): the success path advances the overall bar even when
# nothing was generated -- previously this block lived inside the
# generation branch and reuse blew up with UnboundLocalError
progress_reporter = None
if args.progress and model_spec:
from evalharness.progress import PROGRESS_REGISTRY
_pname = getattr(args, 'progress_plugin', 'rich') \
if args.progress is True else args.progress
try:
_P = PROGRESS_REGISTRY.get(_pname)
except KeyError:
raise SystemExit(f'unknown progress plugin {_pname!r}; '
f'available: {", ".join(PROGRESS_REGISTRY.names())}')
if _P is not None:
# share ONE console: phase lines printed by another
# writer during the live bar interleave incorrectly.
# One reporter for the WHOLE run: overall bar (which
# benchmark) + sample bar (which sample), reused per
# benchmark via reset_samples().
if _shared_reporter is None and (
_pname == 'plain' or console.is_terminal):
# live bars only on a real terminal: through pipes
# (| grep, > log) rich's refresh thread misbehaves
# and stalls the run -- plain phases instead
_shared_reporter = _P(console=console)
_shared_reporter.owned_externally = True
progress_reporter = _shared_reporter
if report is None and model_spec: # generate + score in one go
from evalharness.model import run_eval
progress_reporter = None
if args.progress:
from evalharness.progress import PROGRESS_REGISTRY
_pname = getattr(args, 'progress_plugin', 'rich') \
if args.progress is True else args.progress
try:
_P = PROGRESS_REGISTRY.get(_pname)
except KeyError:
raise SystemExit(f'unknown progress plugin {_pname!r}; '
f'available: {", ".join(PROGRESS_REGISTRY.names())}')
if _P is not None:
# share ONE console: phase lines printed by another
# writer during the live bar interleave incorrectly.
# One reporter for the WHOLE run: overall bar (which
# benchmark) + sample bar (which sample), reused per
# benchmark via reset_samples().
if _shared_reporter is None and (
_pname == 'plain' or console.is_terminal):
# live bars only on a real terminal: through pipes
# (| grep, > log) rich's refresh thread misbehaves
# and stalls the run -- plain phases instead
_shared_reporter = _P(console=console)
_shared_reporter.owned_externally = True
progress_reporter = _shared_reporter
def status_callback(msg, _idx=i + 1, _name=name,
_reporter=progress_reporter,