diff --git a/training/eval_history.json b/training/eval_history.json index af29b6b6..f10284d9 100644 --- a/training/eval_history.json +++ b/training/eval_history.json @@ -387,5 +387,32 @@ "team_1": 42 }, "win_rate_a": 0.41 + }, + { + "timestamp": "2026-08-11T14:12:24+00:00", + "model_a": "/home/jcreek/ai-training/CosmicClash/Game/bots/20260811-0858-gen5-s4-handling-retry4.json", + "model_b": "/home/jcreek/ai-training/CosmicClash/Game/bots/promoted/easy.json", + "seed": 1, + "episodes": 100, + "wins_a": 32, + "wins_b": 33, + "draws": 35, + "side_results": { + "a_team_0": { + "wins_a": 15, + "wins_b": 18, + "draws": 17 + }, + "a_team_1": { + "wins_a": 17, + "wins_b": 15, + "draws": 18 + } + }, + "physical_team_wins": { + "team_0": 30, + "team_1": 35 + }, + "win_rate_a": 0.32 } ] diff --git a/training/generation5_state.json b/training/generation5_state.json index 4aabbf32..b582f609 100644 --- a/training/generation5_state.json +++ b/training/generation5_state.json @@ -1,6 +1,6 @@ { "stage_index": 0, - "attempt": 3, + "attempt": 4, "status": "blocked", "log": [ { @@ -342,6 +342,90 @@ } ], "decision": "fail" + }, + { + "stage_index": 0, + "stage_number": 4, + "stage_name": "handling", + "experiment": "20260811-0858-gen5-s4-handling-retry4", + "attempt": 4, + "telemetry_tail": { + "rollout/air_touch_fraction": 0.0002199999950826168, + "rollout/airborne_fraction": 0.39690066695213316, + "rollout/ep_len_mean": 157.15703988647462, + "rollout/ep_rew_mean": 12.389891688346863, + "rollout/forward_motion_fraction": 0.18413566660881042, + "rollout/goal_rate": 0.36632000091671946, + "rollout/mean_altitude": 4.058596196651458, + "rollout/productive_air_touch_fraction": 9.999999776482581e-05, + "rollout/upright_fraction": 0.6957313330769539, + "rollout/vertical_thrust_mean": -0.25598900008201597 + }, + "telemetry_failures": [ + "rollout/goal_rate=0.3663 < 0.8000", + "rollout/forward_motion_fraction=0.1841 < 0.2500" + ], + "evaluation_goal_failures": [ + "easy.json: goal_rate=0.650 < 0.800" + ], + "side_balance_failures": [], + "eval": { + "timestamp": "2026-08-11T14:12:24+00:00", + "model_a": "/home/jcreek/ai-training/CosmicClash/Game/bots/20260811-0858-gen5-s4-handling-retry4.json", + "model_b": "/home/jcreek/ai-training/CosmicClash/Game/bots/promoted/easy.json", + "seed": 1, + "episodes": 100, + "wins_a": 32, + "wins_b": 33, + "draws": 35, + "side_results": { + "a_team_0": { + "wins_a": 15, + "wins_b": 18, + "draws": 17 + }, + "a_team_1": { + "wins_a": 17, + "wins_b": 15, + "draws": 18 + } + }, + "physical_team_wins": { + "team_0": 30, + "team_1": 35 + }, + "win_rate_a": 0.32 + }, + "evals": [ + { + "timestamp": "2026-08-11T14:12:24+00:00", + "model_a": "/home/jcreek/ai-training/CosmicClash/Game/bots/20260811-0858-gen5-s4-handling-retry4.json", + "model_b": "/home/jcreek/ai-training/CosmicClash/Game/bots/promoted/easy.json", + "seed": 1, + "episodes": 100, + "wins_a": 32, + "wins_b": 33, + "draws": 35, + "side_results": { + "a_team_0": { + "wins_a": 15, + "wins_b": 18, + "draws": 17 + }, + "a_team_1": { + "wins_a": 17, + "wins_b": 15, + "draws": 18 + } + }, + "physical_team_wins": { + "team_0": 30, + "team_1": 35 + }, + "win_rate_a": 0.32 + } + ], + "decision": "fail" } ] }