{ "meta": { "backend": "random-uniform", "seed_note": "rng = Random(42 + idx*100) per run, fixed for reproducibility", "games": [ "ar25@600", "ar25@150", "bp35@150" ], "max_actions": [ 600, 150, 150 ], "timestamp": "2026-09-24 05:01:27", "note": "Random-baseline control. T4 8K9ITI8A was timeout-recycled (502) MID-RUN: ar25@600 and ar25@150 completed and printed RESULT lines (captured verbatim in the session log); bp35@150 was lost and the report file was never written (written only at end of runfile). Entries below transcribe the two captured RESULT lines; bp35@150 marked null = pending re-run on next T4 (1-minute job)." }, "games": [ { "game_id": "ar25", "budget": 600, "win": false, "final_state": "NOT_FINISHED", "best_level": 0, "resets": 6, "actions": 594, "seconds": 0.8, "action_hist": { "ACTION6": 88, "ACTION1": 95, "ACTION3": 77, "ACTION2": 97, "ACTION5": 83, "ACTION4": 71, "ACTION7": 83 } }, { "game_id": "ar25", "budget": 150, "win": false, "final_state": "NOT_FINISHED", "best_level": 0, "resets": 1, "actions": 149, "seconds": 0.2, "action_hist": { "ACTION5": 19, "ACTION6": 20, "ACTION4": 14, "ACTION2": 23, "ACTION3": 21, "ACTION7": 28, "ACTION1": 24 } }, { "game_id": "bp35", "budget": 150, "status": "lost_to_t4_recycle", "best_level": null, "resets": null, "actions": null } ], "summary": { "played": 2, "wins": 0, "best_level_sum": 0, "resets_sum": 7, "actions_sum": 743, "pending": [ "bp35@150" ] }, "bp35_150": { "game_id": "bp35", "win": false, "final_state": "NOT_FINISHED", "best_level": 0, "resets": 2, "actions": 148, "seconds": 0.3, "action_hist": { "ACTION3": 39, "ACTION6": 37, "ACTION4": 41, "ACTION7": 31 }, "note": "completed on 0S3BN0UN 2026-09-24 (was lost_to_t4_recycle)" }, "status": "complete", "conclusion": "Random baseline 0 levels across all 3 configs (ar25@600: 0W/6R, ar25@150: 0W/1R, bp35@150: 0W/2R) -> 0-level = noise floor for <=600-step budgets. Model 0-reset behavior = learned safety, not failure. Confirms v5 agent-teaching direction." }