core-labs/core
1
0
Fork 0
Code Issues Pull requests 1 Projects Releases Packages Wiki Activity Actions

chore(gsm8k): polish monotonic baseline test hygiene

Browse source
This commit is contained in:
Shay 2026-06-17 13:13:20 -07:00
parent 2349a06fe8
commit ea81e7c49d
2 changed files with 1 additions and 4 deletions
Show stats Download patch file Download diff file Expand all files Collapse all files

2
tests/gsm8k_train_sample_baseline.py
View file

@ -28,4 +28,4 @@ def assert_monotonic_serving_counts(counts: dict[str, int]) -> None:
total = counts["correct"] + counts["wrong"] + counts["refused"]
assert total == TRAIN_SAMPLE_COUNT, (
f"expected {TRAIN_SAMPLE_COUNT} scored cases, got {total}"
)
)

3
tests/test_gsm8k_frontier_report.py
View file

@ -60,9 +60,6 @@ def test_committed_frontier_fixture_is_6_44_0_until_rebaselined():
def test_live_serving_meets_monotonic_capability_contract():
"""Live train_sample: wrong=0 hard; counts monotonic; refusals carry reasons."""
import re
from collections import Counter
from evals.gsm8k_math.train_sample.v1.runner import build_report
cases_path = _REPO_ROOT / "evals/gsm8k_math/train_sample/v1/cases.jsonl"