{ "toy_verification": { "normalization_sweep": [ { "scale": 0.05, "observed_variance": 0.9999996021912955, "predicted": 0.9999996021912961 }, { "scale": 0.2, "observed_variance": 0.9999999005478004, "predicted": 0.9999999005478019 }, { "scale": 1.0, "observed_variance": 0.9999999801095585, "predicted": 0.9999999801095593 }, { "scale": 5.0, "observed_variance": 0.9999999960219117, "predicted": 0.9999999960219117 }, { "scale": 20.0, "observed_variance": 0.9999999990054788, "predicted": 0.999999999005478 } ], "margin_sweep": [ { "separation": 0.0, "wF": 1.0, "wV": 1.0, "observed_margin": -0.009074307857396793, "predicted_margin": 0.0 }, { "separation": 0.0, "wF": 2.0, "wV": 1.0, "observed_margin": -0.013611461786095192, "predicted_margin": 0.0 }, { "separation": 0.0, "wF": 4.0, "wV": 1.0, "observed_margin": -0.02268576964349199, "predicted_margin": 0.0 }, { "separation": 0.2, "wF": 1.0, "wV": 1.0, "observed_margin": 0.7356044541460376, "predicted_margin": 0.7440645962550935 }, { "separation": 0.2, "wF": 2.0, "wV": 1.0, "observed_margin": 1.1034066812190564, "predicted_margin": 1.1160968943826401 }, { "separation": 0.2, "wF": 4.0, "wV": 1.0, "observed_margin": 1.839011135365094, "predicted_margin": 1.8601614906377335 }, { "separation": 0.5, "wF": 1.0, "wV": 1.0, "observed_margin": 1.4111123653624242, "predicted_margin": 1.4175707054694646 }, { "separation": 0.5, "wF": 2.0, "wV": 1.0, "observed_margin": 2.1166685480436365, "predicted_margin": 2.126356058204197 }, { "separation": 0.5, "wF": 4.0, "wV": 1.0, "observed_margin": 3.5277809134060605, "predicted_margin": 3.5439267636736616 }, { "separation": 1.0, "wF": 1.0, "wV": 1.0, "observed_margin": 1.788105220470278, "predicted_margin": 1.7921859160787397 }, { "separation": 1.0, "wF": 2.0, "wV": 1.0, "observed_margin": 2.682157830705417, "predicted_margin": 2.68827887411811 }, { "separation": 1.0, "wF": 4.0, "wV": 1.0, "observed_margin": 4.470263051175695, "predicted_margin": 4.48046479019685 }, { "separation": 2.0, "wF": 1.0, "wV": 1.0, "observed_margin": 1.9401772463439695, "predicted_margin": 1.9423874121784488 }, { "separation": 2.0, "wF": 2.0, "wV": 1.0, "observed_margin": 2.9102658695159542, "predicted_margin": 2.913581118267673 }, { "separation": 2.0, "wF": 4.0, "wV": 1.0, "observed_margin": 4.850443115859925, "predicted_margin": 4.8559685304461215 } ], "margin_observed_slope_wF2_wV1": 1.2815550367972013, "margin_prediction_slope_note": "(wF+wV)/(2*sigma), with sigma measured per sweep" }, "mini_experiment": { "device": "cuda", "rows": [ { "method": "standard", "final_feasibility_mean": 2.2039625188522364e-09, "final_feasibility_std": 2.464394918026189e-10, "policy_risky_probability_mean": 0.9814547970890999, "expected_reward_mean": 5.9629095941782, "training_rollout_feasibility_last20": 0.0 }, { "method": "ranked", "final_feasibility_mean": 3.211668065484117e-05, "final_feasibility_std": 6.337542572511882e-06, "policy_risky_probability_mean": 0.8740648701786995, "expected_reward_mean": 5.748129740357399, "training_rollout_feasibility_last20": 9.765625e-05 } ] } }