cookbook/environments/_13_saved_baselines/TEST_LOG.md
Tested 2026-07-20 with OpenAIResponses(id="gpt-5.5", reasoning_effort="low").
Status: PASS
Description: Saved a live twelve-attempt result as a plain JSON baseline.
Result: product-a passed 3/6 (0.50) and product-d passed 3/6 (0.50).
All twelve attempts were written to the generated baseline artifact.
Calibration: The first task set (product-a, product-b, k=4) saturated at
4/4 on both rows. product-b was replaced with product-d and k was raised to
six before this PASS was recorded.
Status: PASS
Description: Saved and reloaded a baseline, then compared its complete
summary() result with the live object.
Result: product-a passed 2/4 (0.50) and product-c passed 4/4 (1.00).
The aggregate pass rate was 0.75 and both fingerprints survived the round trip.
Status: PASS
Description: Used arun_rollouts, asave, and aload inside one event loop.
Result: product-a passed 3/4 (0.75) and product-b passed 4/4 (1.00).
The async round trip preserved all eight attempts and the complete summary.