{ "base": { "label": "Base Qwen 7B", "focal_mode": "base", "mean_reward": 0.5285364583333333, "std_reward": 0.1366888070980726, "bankruptcy_rate": 0.0, "rubric_means": { "economic": 0.68115, "diplomatic": 0.31531132185545574, "military": 0.46458333333333335, "stability": 0.7702083333333332, "hidden": 0.33424479166666665 }, "per_objective": { "KINGMAKER": 0.4363375, "SABOTEUR": 0.47791666666666666, "SURVIVOR": 0.7575458333333334, "TRADE_HEGEMON": 0.4207375, "COALITION_BUILDER": 0.5899625, "LONE_WOLF": 0.7169791666666667, "SPY_MASTER": 0.4815125, "PEACEKEEPER": 0.3473 }, "per_objective_hidden": { "KINGMAKER": 0.20000000000000004, "SABOTEUR": 0.265625, "SURVIVOR": 1.0, "TRADE_HEGEMON": 0.0, "COALITION_BUILDER": 0.20833333333333334, "LONE_WOLF": 1.0, "SPY_MASTER": 0.0, "PEACEKEEPER": 0.0 }, "rollouts": [ { "objective": "KINGMAKER", "seed": 1000, "reward": 0.4363375, "final_grade": 0.4567, "dense_reward": 0.415975, "components": { "economic": 0.6455000000000001, "diplomatic": 0.25, "military": 0.44999999999999996, "stability": 0.7291666666666666, "hidden": 0.20000000000000004 }, "bankrupt": false, "n_turns": 12 }, { "objective": "SABOTEUR", "seed": 1001, "reward": 0.47791666666666666, "final_grade": 0.4782, "dense_reward": 0.47763333333333335, "components": { "economic": 0.6785, "diplomatic": 0.3749441322589204, "military": 0.5, "stability": 0.7483333333333334, "hidden": 0.265625 }, "bankrupt": false, "n_turns": 12 }, { "objective": "SURVIVOR", "seed": 1002, "reward": 0.7575458333333334, "final_grade": 0.8254, "dense_reward": 0.6896916666666666, "components": { "economic": 0.7025, "diplomatic": 0.28508358016534163, "military": 0.48333333333333334, "stability": 0.7979166666666666, "hidden": 1.0 }, "bankrupt": false, "n_turns": 12 }, { "objective": "TRADE_HEGEMON", "seed": 1003, "reward": 0.4207375, "final_grade": 0.4545, "dense_reward": 0.386975, "components": { "economic": 0.722, "diplomatic": 0.25, "military": 0.5, "stability": 0.78375, "hidden": 0.0 }, "bankrupt": false, "n_turns": 12 }, { "objective": "COALITION_BUILDER", "seed": 1004, "reward": 0.5899625, "final_grade": 0.6949, "dense_reward": 0.485025, "components": { "economic": 0.707, "diplomatic": 0.41364197731973334, "military": 0.48333333333333334, "stability": 0.8391666666666665, "hidden": 0.20833333333333334 }, "bankrupt": false, "n_turns": 12 }, { "objective": "LONE_WOLF", "seed": 1005, "reward": 0.7169791666666667, "final_grade": 0.753, "dense_reward": 0.6809583333333333, "components": { "economic": 0.6951999999999999, "diplomatic": 0.25, "military": 0.5, "stability": 0.7795833333333332, "hidden": 1.0 }, "bankrupt": false, "n_turns": 12 }, { "objective": "SPY_MASTER", "seed": 1006, "reward": 0.4815125, "final_grade": 0.5242, "dense_reward": 0.43882499999999997, "components": { "economic": 0.7075, "diplomatic": 0.44882088509965046, "military": 0.5, "stability": 0.8837499999999999, "hidden": 0.0 }, "bankrupt": false, "n_turns": 12 }, { "objective": "PEACEKEEPER", "seed": 1007, "reward": 0.3473, "final_grade": 0.3914, "dense_reward": 0.30319999999999997, "components": { "economic": 0.5910000000000001, "diplomatic": 0.25, "military": 0.29999999999999993, "stability": 0.5999999999999999, "hidden": 0.0 }, "bankrupt": false, "n_turns": 12 } ] }, "sft_only": { "label": "+ SFT only", "focal_mode": "frozen_opp", "mean_reward": 0.5007453125000001, "std_reward": 0.15565937267436883, "bankruptcy_rate": 0.0, "rubric_means": { "economic": 0.56209, "diplomatic": 0.33704327491204017, "military": 0.30885416666666665, "stability": 0.6252083333333334, "hidden": 0.4708333333333334 }, "per_objective": { "KINGMAKER": 0.3601, "SABOTEUR": 0.43814583333333335, "SURVIVOR": 0.6456000000000001, "TRADE_HEGEMON": 0.3506625, "COALITION_BUILDER": 0.78195, "LONE_WOLF": 0.43613749999999996, "SPY_MASTER": 0.6439041666666667, "PEACEKEEPER": 0.3494625 }, "per_objective_hidden": { "KINGMAKER": 0.10000000000000002, "SABOTEUR": 0.25, "SURVIVOR": 1.0, "TRADE_HEGEMON": 0.0, "COALITION_BUILDER": 0.9583333333333334, "LONE_WOLF": 0.5833333333333334, "SPY_MASTER": 0.875, "PEACEKEEPER": 0.0 }, "rollouts": [ { "objective": "KINGMAKER", "seed": 1000, "reward": 0.3601, "final_grade": 0.3936, "dense_reward": 0.3266, "components": { "economic": 0.5419999999999999, "diplomatic": 0.25, "military": 0.2833333333333334, "stability": 0.6379166666666666, "hidden": 0.10000000000000002 }, "bankrupt": false, "n_turns": 12 }, { "objective": "SABOTEUR", "seed": 1001, "reward": 0.43814583333333335, "final_grade": 0.4695, "dense_reward": 0.4067916666666667, "components": { "economic": 0.537, "diplomatic": 0.384097174508573, "military": 0.32500000000000007, "stability": 0.6587500000000001, "hidden": 0.25 }, "bankrupt": false, "n_turns": 12 }, { "objective": "SURVIVOR", "seed": 1002, "reward": 0.6456000000000001, "final_grade": 0.6886, "dense_reward": 0.6026000000000001, "components": { "economic": 0.5880000000000001, "diplomatic": 0.25, "military": 0.29999999999999993, "stability": 0.5999999999999999, "hidden": 1.0 }, "bankrupt": false, "n_turns": 12 }, { "objective": "TRADE_HEGEMON", "seed": 1003, "reward": 0.3506625, "final_grade": 0.421, "dense_reward": 0.280325, "components": { "economic": 0.5449999999999999, "diplomatic": 0.25, "military": 0.28750000000000003, "stability": 0.5212500000000001, "hidden": 0.0 }, "bankrupt": false, "n_turns": 12 }, { "objective": "COALITION_BUILDER", "seed": 1004, "reward": 0.78195, "final_grade": 0.8406, "dense_reward": 0.7233, "components": { "economic": 0.5965000000000001, "diplomatic": 0.8122490247877483, "military": 0.36666666666666664, "stability": 0.6604166666666665, "hidden": 0.9583333333333334 }, "bankrupt": false, "n_turns": 12 }, { "objective": "LONE_WOLF", "seed": 1005, "reward": 0.43613749999999996, "final_grade": 0.3811, "dense_reward": 0.491175, "components": { "economic": 0.5949399999999999, "diplomatic": 0.25, "military": 0.33333333333333326, "stability": 0.6479166666666666, "hidden": 0.5833333333333334 }, "bankrupt": false, "n_turns": 12 }, { "objective": "SPY_MASTER", "seed": 1006, "reward": 0.6439041666666667, "final_grade": 0.7232, "dense_reward": 0.5646083333333334, "components": { "economic": 0.543, "diplomatic": 0.25, "military": 0.28750000000000003, "stability": 0.6691666666666668, "hidden": 0.875 }, "bankrupt": false, "n_turns": 12 }, { "objective": "PEACEKEEPER", "seed": 1007, "reward": 0.3494625, "final_grade": 0.4048, "dense_reward": 0.294125, "components": { "economic": 0.5502800000000002, "diplomatic": 0.25, "military": 0.2874999999999999, "stability": 0.6062499999999998, "hidden": 0.0 }, "bankrupt": false, "n_turns": 12 } ] }, "sft_grpo": { "label": "+ SFT + v3 GRPO", "focal_mode": "default", "mean_reward": 0.5272281249999999, "std_reward": 0.17839166418196054, "bankruptcy_rate": 0.125, "rubric_means": { "economic": 0.5851999999999999, "diplomatic": 0.5622153227401966, "military": 0.375, "stability": 0.6490625000000001, "hidden": 0.45729166666666665 }, "per_objective": { "KINGMAKER": 0.6204958333333334, "SABOTEUR": 0.421575, "SURVIVOR": 0.7224416666666666, "TRADE_HEGEMON": 0.3705708333333333, "COALITION_BUILDER": 0.8253625, "LONE_WOLF": 0.4734083333333333, "SPY_MASTER": 0.24414166666666662, "PEACEKEEPER": 0.5398291666666667 }, "per_objective_hidden": { "KINGMAKER": 0.4166666666666667, "SABOTEUR": 0.325, "SURVIVOR": 1.0, "TRADE_HEGEMON": 0.0, "COALITION_BUILDER": 0.9583333333333334, "LONE_WOLF": 0.08333333333333333, "SPY_MASTER": 0.875, "PEACEKEEPER": 0.0 }, "rollouts": [ { "objective": "KINGMAKER", "seed": 1000, "reward": 0.6204958333333334, "final_grade": 0.6218, "dense_reward": 0.6191916666666667, "components": { "economic": 0.656, "diplomatic": 0.8340047760738459, "military": 0.5, "stability": 0.8079166666666665, "hidden": 0.4166666666666667 }, "bankrupt": false, "n_turns": 12 }, { "objective": "SABOTEUR", "seed": 1001, "reward": 0.421575, "final_grade": 0.4386, "dense_reward": 0.40454999999999997, "components": { "economic": 0.549, "diplomatic": 0.25, "military": 0.28750000000000003, "stability": 0.6941666666666668, "hidden": 0.325 }, "bankrupt": false, "n_turns": 12 }, { "objective": "SURVIVOR", "seed": 1002, "reward": 0.7224416666666666, "final_grade": 0.7897, "dense_reward": 0.6551833333333335, "components": { "economic": 0.6230000000000001, "diplomatic": 0.382526789099765, "military": 0.3166666666666666, "stability": 0.7104166666666666, "hidden": 1.0 }, "bankrupt": false, "n_turns": 12 }, { "objective": "TRADE_HEGEMON", "seed": 1003, "reward": 0.3705708333333333, "final_grade": 0.4352, "dense_reward": 0.3059416666666666, "components": { "economic": 0.5435, "diplomatic": 0.25, "military": 0.28750000000000003, "stability": 0.6941666666666668, "hidden": 0.0 }, "bankrupt": false, "n_turns": 12 }, { "objective": "COALITION_BUILDER", "seed": 1004, "reward": 0.8253625, "final_grade": 0.8842, "dense_reward": 0.766525, "components": { "economic": 0.62, "diplomatic": 0.9420390347738513, "military": 0.39999999999999997, "stability": 0.7108333333333334, "hidden": 0.9583333333333334 }, "bankrupt": false, "n_turns": 12 }, { "objective": "LONE_WOLF", "seed": 1005, "reward": 0.4734083333333333, "final_grade": 0.5061, "dense_reward": 0.4407166666666667, "components": { "economic": 0.5555, "diplomatic": 0.7058862000438966, "military": 0.44999999999999996, "stability": 0.6395833333333334, "hidden": 0.08333333333333333 }, "bankrupt": false, "n_turns": 12 }, { "objective": "SPY_MASTER", "seed": 1006, "reward": 0.24414166666666662, "final_grade": 0.0, "dense_reward": 0.48828333333333324, "components": { "economic": 0.5226, "diplomatic": 0.25, "military": 0.2750000000000001, "stability": 0.19999999999999998, "hidden": 0.875 }, "bankrupt": true, "n_turns": 12 }, { "objective": "PEACEKEEPER", "seed": 1007, "reward": 0.5398291666666667, "final_grade": 0.5978, "dense_reward": 0.48185833333333333, "components": { "economic": 0.612, "diplomatic": 0.883265781930214, "military": 0.48333333333333334, "stability": 0.7354166666666666, "hidden": 0.0 }, "bankrupt": false, "n_turns": 12 } ] } }