initial performance: -3
episode: 0 training return: tensor(1.8457e-07, device='cuda:0', grad_fn=<AddBackward0>)
episode: 1 training return: tensor(9.7790e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 2 training return: tensor(1.1851e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 3 training return: tensor(5.7851e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 1 test_true_pfm: -625 test_simulate_pfm tensor(2.0617e-10, device='cuda:0', grad_fn=<DivBackward0>)
episode: 4 training return: tensor(9.1035e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 5 training return: tensor(2.0996e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 6 training return: tensor(6.7412e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 7 training return: tensor(4.8374e-08, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 2 test_true_pfm: -722 test_simulate_pfm tensor(2.5513e-07, device='cuda:0', grad_fn=<DivBackward0>)
episode: 8 training return: tensor(5.8402e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 9 training return: tensor(9.6725e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 10 training return: tensor(9.8532e-05, device='cuda:0', grad_fn=<AddBackward0>)
episode: 11 training return: tensor(0.0010, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 3 test_true_pfm: -565 test_simulate_pfm tensor(0.0014, device='cuda:0', grad_fn=<DivBackward0>)
episode: 12 training return: tensor(0.0012, device='cuda:0', grad_fn=<AddBackward0>)
episode: 13 training return: tensor(1.5636e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 14 training return: tensor(5.6689e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 15 training return: tensor(6.5437e-06, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 4 test_true_pfm: -389 test_simulate_pfm tensor(3.2054e-08, device='cuda:0', grad_fn=<DivBackward0>)
episode: 16 training return: tensor(3.4903e-06, device='cuda:0', grad_fn=<AddBackward0>)
episode: 17 training return: tensor(4.9671e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 18 training return: tensor(8.0468e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 19 training return: tensor(8.9035e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 5 test_true_pfm: -625 test_simulate_pfm tensor(1.1044e-16, device='cuda:0', grad_fn=<DivBackward0>)
episode: 20 training return: tensor(4.3876e-16, device='cuda:0', grad_fn=<AddBackward0>)
episode: 21 training return: tensor(3.5506e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 22 training return: tensor(4.1957e-06, device='cuda:0', grad_fn=<AddBackward0>)
episode: 23 training return: tensor(9.4411e-07, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 6 test_true_pfm: -396 test_simulate_pfm tensor(1.1351e-07, device='cuda:0', grad_fn=<DivBackward0>)
episode: 24 training return: tensor(2.2547e-06, device='cuda:0', grad_fn=<AddBackward0>)
episode: 25 training return: tensor(4.4824e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 26 training return: tensor(6.6614e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 27 training return: tensor(1.3439e-13, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 7 test_true_pfm: -674 test_simulate_pfm tensor(7.4261e-17, device='cuda:0', grad_fn=<DivBackward0>)
episode: 28 training return: tensor(1.5292e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 29 training return: tensor(2.3909e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 30 training return: tensor(1.4940e-15, device='cuda:0', grad_fn=<AddBackward0>)
episode: 31 training return: tensor(6.3128e-16, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 8 test_true_pfm: -594 test_simulate_pfm tensor(5.9062e-21, device='cuda:0', grad_fn=<DivBackward0>)
episode: 32 training return: tensor(1.0071e-16, device='cuda:0', grad_fn=<AddBackward0>)
episode: 33 training return: tensor(5.6757e-21, device='cuda:0', grad_fn=<AddBackward0>)
episode: 34 training return: tensor(8.6014e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 35 training return: tensor(2.5033e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 9 test_true_pfm: 246 test_simulate_pfm tensor(1.5737e-10, device='cuda:0', grad_fn=<DivBackward0>)
episode: 36 training return: tensor(4.0809e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 37 training return: tensor(4.8192e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 38 training return: tensor(1.6525e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 39 training return: tensor(1.1454e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 10 test_true_pfm: -124 test_simulate_pfm tensor(8.3729e-12, device='cuda:0', grad_fn=<DivBackward0>)
episode: 40 training return: tensor(8.5965e-15, device='cuda:0', grad_fn=<AddBackward0>)
episode: 41 training return: tensor(8.3152e-15, device='cuda:0', grad_fn=<AddBackward0>)
episode: 42 training return: tensor(6.9810e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 43 training return: tensor(4.7625e-12, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 11 test_true_pfm: -575 test_simulate_pfm tensor(1.4384e-14, device='cuda:0', grad_fn=<DivBackward0>)
episode: 44 training return: tensor(6.7115e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 45 training return: tensor(4.9491e-14, device='cuda:0', grad_fn=<AddBackward0>)
episode: 46 training return: tensor(1.5281e-15, device='cuda:0', grad_fn=<AddBackward0>)
episode: 47 training return: tensor(1.1873e-22, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 12 test_true_pfm: -1086 test_simulate_pfm tensor(2.7878e-24, device='cuda:0', grad_fn=<DivBackward0>)
episode: 48 training return: tensor(8.2203e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 49 training return: tensor(1.9804e-17, device='cuda:0', grad_fn=<AddBackward0>)
episode: 50 training return: tensor(8.4682e-18, device='cuda:0', grad_fn=<AddBackward0>)
episode: 51 training return: tensor(5.2892e-14, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 13 test_true_pfm: -572 test_simulate_pfm tensor(6.2787e-14, device='cuda:0', grad_fn=<DivBackward0>)
episode: 52 training return: tensor(7.1251e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 53 training return: tensor(5.3398e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 54 training return: tensor(9.8557e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 55 training return: tensor(5.5941e-13, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 14 test_true_pfm: -36 test_simulate_pfm tensor(6.8606e-13, device='cuda:0', grad_fn=<DivBackward0>)
episode: 56 training return: tensor(2.1154e-14, device='cuda:0', grad_fn=<AddBackward0>)
episode: 57 training return: tensor(1.7751e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 58 training return: tensor(1.0570e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 59 training return: tensor(6.3556e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 15 test_true_pfm: -375 test_simulate_pfm tensor(4.1478e-11, device='cuda:0', grad_fn=<DivBackward0>)
episode: 60 training return: tensor(1.0286e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 61 training return: tensor(2.9443e-16, device='cuda:0', grad_fn=<AddBackward0>)
episode: 62 training return: tensor(2.8056e-14, device='cuda:0', grad_fn=<AddBackward0>)
episode: 63 training return: tensor(4.4867e-12, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 16 test_true_pfm: -512 test_simulate_pfm tensor(2.8743e-12, device='cuda:0', grad_fn=<DivBackward0>)
episode: 64 training return: tensor(1.3056e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 65 training return: tensor(1.0327e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 66 training return: tensor(5.0106e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 67 training return: tensor(1.5940e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 17 test_true_pfm: -419 test_simulate_pfm tensor(8.1446e-15, device='cuda:0', grad_fn=<DivBackward0>)
episode: 68 training return: tensor(4.7292e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 69 training return: tensor(7.8100e-14, device='cuda:0', grad_fn=<AddBackward0>)
episode: 70 training return: tensor(2.7847e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 71 training return: tensor(2.1589e-15, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 18 test_true_pfm: -540 test_simulate_pfm tensor(1.2511e-13, device='cuda:0', grad_fn=<DivBackward0>)
episode: 72 training return: tensor(1.6710e-16, device='cuda:0', grad_fn=<AddBackward0>)
episode: 73 training return: tensor(1.6922e-15, device='cuda:0', grad_fn=<AddBackward0>)
episode: 74 training return: tensor(2.6458e-14, device='cuda:0', grad_fn=<AddBackward0>)
episode: 75 training return: tensor(1.1172e-14, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 19 test_true_pfm: -396 test_simulate_pfm tensor(1.0277e-12, device='cuda:0', grad_fn=<DivBackward0>)
episode: 76 training return: tensor(1.4844e-15, device='cuda:0', grad_fn=<AddBackward0>)
episode: 77 training return: tensor(1.5378e-15, device='cuda:0', grad_fn=<AddBackward0>)
episode: 78 training return: tensor(1.5963e-16, device='cuda:0', grad_fn=<AddBackward0>)
episode: 79 training return: tensor(1.7907e-18, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 20 test_true_pfm: -521 test_simulate_pfm tensor(5.2432e-22, device='cuda:0', grad_fn=<DivBackward0>)
episode: 80 training return: tensor(5.6842e-21, device='cuda:0', grad_fn=<AddBackward0>)
episode: 81 training return: tensor(7.1132e-23, device='cuda:0', grad_fn=<AddBackward0>)
episode: 82 training return: tensor(5.0175e-23, device='cuda:0', grad_fn=<AddBackward0>)
episode: 83 training return: tensor(3.9759e-23, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 21 test_true_pfm: -543 test_simulate_pfm tensor(1.0622e-16, device='cuda:0', grad_fn=<DivBackward0>)
episode: 84 training return: tensor(7.0809e-18, device='cuda:0', grad_fn=<AddBackward0>)
episode: 85 training return: tensor(2.0218e-16, device='cuda:0', grad_fn=<AddBackward0>)
episode: 86 training return: tensor(3.8635e-15, device='cuda:0', grad_fn=<AddBackward0>)
episode: 87 training return: tensor(1.0009e-18, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 22 test_true_pfm: -495 test_simulate_pfm tensor(7.8200e-15, device='cuda:0', grad_fn=<DivBackward0>)
episode: 88 training return: tensor(7.8755e-15, device='cuda:0', grad_fn=<AddBackward0>)
episode: 89 training return: tensor(2.0661e-16, device='cuda:0', grad_fn=<AddBackward0>)
episode: 90 training return: tensor(8.8855e-17, device='cuda:0', grad_fn=<AddBackward0>)
episode: 91 training return: tensor(2.0176e-19, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 23 test_true_pfm: -592 test_simulate_pfm tensor(1.5421e-18, device='cuda:0', grad_fn=<DivBackward0>)
episode: 92 training return: tensor(1.0054e-17, device='cuda:0', grad_fn=<AddBackward0>)
episode: 93 training return: tensor(1.3273e-18, device='cuda:0', grad_fn=<AddBackward0>)
episode: 94 training return: tensor(1.9500e-18, device='cuda:0', grad_fn=<AddBackward0>)
episode: 95 training return: tensor(9.6229e-18, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 24 test_true_pfm: -571 test_simulate_pfm tensor(1.2496e-13, device='cuda:0', grad_fn=<DivBackward0>)
episode: 96 training return: tensor(1.0396e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 97 training return: tensor(1.1540e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 98 training return: tensor(6.1585e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 99 training return: tensor(4.3035e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 25 test_true_pfm: -639 test_simulate_pfm tensor(2.9015e-08, device='cuda:0', grad_fn=<DivBackward0>)
episode: 100 training return: tensor(1.8949e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 101 training return: tensor(3.5280e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 102 training return: tensor(2.4475e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 103 training return: tensor(1.3095e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 26 test_true_pfm: -662 test_simulate_pfm tensor(5.9006e-12, device='cuda:0', grad_fn=<DivBackward0>)
episode: 104 training return: tensor(1.0745e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 105 training return: tensor(2.4982e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 106 training return: tensor(9.9168e-14, device='cuda:0', grad_fn=<AddBackward0>)
episode: 107 training return: tensor(6.5242e-19, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 27 test_true_pfm: -557 test_simulate_pfm tensor(9.7145e-14, device='cuda:0', grad_fn=<DivBackward0>)
episode: 108 training return: tensor(8.5846e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 109 training return: tensor(9.7417e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 110 training return: tensor(1.5125e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 111 training return: tensor(2.5000e-06, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 28 test_true_pfm: -608 test_simulate_pfm tensor(5.3413e-08, device='cuda:0', grad_fn=<DivBackward0>)
episode: 112 training return: tensor(7.6887e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 113 training return: tensor(2.3972e-05, device='cuda:0', grad_fn=<AddBackward0>)
episode: 114 training return: tensor(2.1997e-07, device='cuda:0', grad_fn=<AddBackward0>)
episode: 115 training return: tensor(1.9052e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 29 test_true_pfm: -597 test_simulate_pfm tensor(6.3951e-12, device='cuda:0', grad_fn=<DivBackward0>)
episode: 116 training return: tensor(2.1458e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 117 training return: tensor(5.8070e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 118 training return: tensor(1.6807e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 119 training return: tensor(2.3234e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 30 test_true_pfm: -600 test_simulate_pfm tensor(5.7520e-12, device='cuda:0', grad_fn=<DivBackward0>)
episode: 120 training return: tensor(1.1207e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 121 training return: tensor(3.1655e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 122 training return: tensor(1.9370e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 123 training return: tensor(6.9938e-12, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 31 test_true_pfm: -560 test_simulate_pfm tensor(3.3443e-13, device='cuda:0', grad_fn=<DivBackward0>)
episode: 124 training return: tensor(3.2108e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 125 training return: tensor(5.8957e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 126 training return: tensor(9.0038e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 127 training return: tensor(7.7437e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 32 test_true_pfm: -603 test_simulate_pfm tensor(7.9057e-09, device='cuda:0', grad_fn=<DivBackward0>)
episode: 128 training return: tensor(8.0792e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 129 training return: tensor(7.9776e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 130 training return: tensor(3.5901e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 131 training return: tensor(2.1823e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 33 test_true_pfm: -669 test_simulate_pfm tensor(7.5951e-09, device='cuda:0', grad_fn=<DivBackward0>)
episode: 132 training return: tensor(8.1213e-14, device='cuda:0', grad_fn=<AddBackward0>)
episode: 133 training return: tensor(2.5270e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 134 training return: tensor(4.0594e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 135 training return: tensor(1.8601e-08, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 34 test_true_pfm: -598 test_simulate_pfm tensor(1.1458e-10, device='cuda:0', grad_fn=<DivBackward0>)
episode: 136 training return: tensor(1.2737e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 137 training return: tensor(9.3417e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 138 training return: tensor(1.9515e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 139 training return: tensor(2.6212e-12, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 35 test_true_pfm: -601 test_simulate_pfm tensor(2.6427e-13, device='cuda:0', grad_fn=<DivBackward0>)
episode: 140 training return: tensor(2.5000e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 141 training return: tensor(8.9369e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 142 training return: tensor(8.8648e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 143 training return: tensor(5.0007e-13, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 36 test_true_pfm: -498 test_simulate_pfm tensor(1.3993e-12, device='cuda:0', grad_fn=<DivBackward0>)
episode: 144 training return: tensor(5.6436e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 145 training return: tensor(3.1905e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 146 training return: tensor(3.5879e-14, device='cuda:0', grad_fn=<AddBackward0>)
episode: 147 training return: tensor(1.0216e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 37 test_true_pfm: -554 test_simulate_pfm tensor(3.3455e-14, device='cuda:0', grad_fn=<DivBackward0>)
episode: 148 training return: tensor(1.8010e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 149 training return: tensor(2.8000e-06, device='cuda:0', grad_fn=<AddBackward0>)
episode: 150 training return: tensor(1.5662e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 151 training return: tensor(5.0555e-13, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 38 test_true_pfm: -483 test_simulate_pfm tensor(7.4240e-14, device='cuda:0', grad_fn=<DivBackward0>)
episode: 152 training return: tensor(2.1589e-14, device='cuda:0', grad_fn=<AddBackward0>)
episode: 153 training return: tensor(1.5672e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 154 training return: tensor(2.5627e-15, device='cuda:0', grad_fn=<AddBackward0>)
episode: 155 training return: tensor(1.4998e-16, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 39 test_true_pfm: -556 test_simulate_pfm tensor(1.6473e-17, device='cuda:0', grad_fn=<DivBackward0>)
episode: 156 training return: tensor(1.5298e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 157 training return: tensor(9.1958e-17, device='cuda:0', grad_fn=<AddBackward0>)
episode: 158 training return: tensor(1.5865e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 159 training return: tensor(4.1357e-18, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 40 test_true_pfm: -597 test_simulate_pfm tensor(6.6556e-16, device='cuda:0', grad_fn=<DivBackward0>)
episode: 160 training return: tensor(7.7183e-16, device='cuda:0', grad_fn=<AddBackward0>)
episode: 161 training return: tensor(6.5920e-14, device='cuda:0', grad_fn=<AddBackward0>)
episode: 162 training return: tensor(1.3275e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 163 training return: tensor(4.9270e-13, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 41 test_true_pfm: -510 test_simulate_pfm tensor(2.4080e-11, device='cuda:0', grad_fn=<DivBackward0>)
episode: 164 training return: tensor(1.4039e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 165 training return: tensor(1.5208e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 166 training return: tensor(1.1847e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 167 training return: tensor(1.3821e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 42 test_true_pfm: -618 test_simulate_pfm tensor(1.4634e-10, device='cuda:0', grad_fn=<DivBackward0>)
episode: 168 training return: tensor(1.2951e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 169 training return: tensor(8.3439e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 170 training return: tensor(2.4103e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 171 training return: tensor(2.1513e-12, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 43 test_true_pfm: -583 test_simulate_pfm tensor(1.2783e-15, device='cuda:0', grad_fn=<DivBackward0>)
episode: 172 training return: tensor(3.8383e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 173 training return: tensor(6.5545e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 174 training return: tensor(1.1903e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 175 training return: tensor(6.8159e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 44 test_true_pfm: -454 test_simulate_pfm tensor(3.5864e-16, device='cuda:0', grad_fn=<DivBackward0>)
episode: 176 training return: tensor(1.7793e-18, device='cuda:0', grad_fn=<AddBackward0>)
episode: 177 training return: tensor(7.7378e-16, device='cuda:0', grad_fn=<AddBackward0>)
episode: 178 training return: tensor(3.1008e-15, device='cuda:0', grad_fn=<AddBackward0>)
episode: 179 training return: tensor(1.0861e-17, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 45 test_true_pfm: -369 test_simulate_pfm tensor(3.9554e-15, device='cuda:0', grad_fn=<DivBackward0>)
episode: 180 training return: tensor(1.3734e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 181 training return: tensor(7.7140e-16, device='cuda:0', grad_fn=<AddBackward0>)
episode: 182 training return: tensor(6.5154e-15, device='cuda:0', grad_fn=<AddBackward0>)
episode: 183 training return: tensor(7.1294e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 46 test_true_pfm: -319 test_simulate_pfm tensor(7.7559e-12, device='cuda:0', grad_fn=<DivBackward0>)
episode: 184 training return: tensor(4.3932e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 185 training return: tensor(1.8700e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 186 training return: tensor(6.4970e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 187 training return: tensor(1.1870e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 47 test_true_pfm: -496 test_simulate_pfm tensor(1.2325e-11, device='cuda:0', grad_fn=<DivBackward0>)
episode: 188 training return: tensor(9.0803e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 189 training return: tensor(2.2372e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 190 training return: tensor(2.6134e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 191 training return: tensor(2.3180e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 48 test_true_pfm: -541 test_simulate_pfm tensor(2.4860e-09, device='cuda:0', grad_fn=<DivBackward0>)
episode: 192 training return: tensor(9.3711e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 193 training return: tensor(1.3207e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 194 training return: tensor(3.8219e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 195 training return: tensor(2.0122e-15, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 49 test_true_pfm: -505 test_simulate_pfm tensor(2.2027e-14, device='cuda:0', grad_fn=<DivBackward0>)
episode: 196 training return: tensor(7.3096e-14, device='cuda:0', grad_fn=<AddBackward0>)
episode: 197 training return: tensor(5.9933e-15, device='cuda:0', grad_fn=<AddBackward0>)
episode: 198 training return: tensor(6.1403e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 199 training return: tensor(6.2666e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 50 test_true_pfm: 0 test_simulate_pfm tensor(2.8871e-08, device='cuda:0', grad_fn=<DivBackward0>)
episode: 200 training return: tensor(2.1699e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 201 training return: tensor(2.4299e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 202 training return: tensor(3.5790e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 203 training return: tensor(4.1896e-07, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 51 test_true_pfm: -279 test_simulate_pfm tensor(2.1620e-06, device='cuda:0', grad_fn=<DivBackward0>)
episode: 204 training return: tensor(2.1260e-07, device='cuda:0', grad_fn=<AddBackward0>)
episode: 205 training return: tensor(3.4890e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 206 training return: tensor(1.3367e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 207 training return: tensor(8.5040e-13, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 52 test_true_pfm: -522 test_simulate_pfm tensor(2.9443e-12, device='cuda:0', grad_fn=<DivBackward0>)
episode: 208 training return: tensor(4.6302e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 209 training return: tensor(4.6584e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 210 training return: tensor(8.7712e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 211 training return: tensor(2.2765e-13, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 53 test_true_pfm: -548 test_simulate_pfm tensor(1.4099e-10, device='cuda:0', grad_fn=<DivBackward0>)
episode: 212 training return: tensor(2.1750e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 213 training return: tensor(5.4477e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 214 training return: tensor(2.3254e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 215 training return: tensor(1.7064e-14, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 54 test_true_pfm: -281 test_simulate_pfm tensor(7.3499e-09, device='cuda:0', grad_fn=<DivBackward0>)
episode: 216 training return: tensor(2.0167e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 217 training return: tensor(1.8908e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 218 training return: tensor(1.5421e-15, device='cuda:0', grad_fn=<AddBackward0>)
episode: 219 training return: tensor(3.5128e-12, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 55 test_true_pfm: -422 test_simulate_pfm tensor(1.0135e-10, device='cuda:0', grad_fn=<DivBackward0>)
episode: 220 training return: tensor(1.1264e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 221 training return: tensor(8.0156e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 222 training return: tensor(1.5787e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 223 training return: tensor(3.5820e-12, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 56 test_true_pfm: -329 test_simulate_pfm tensor(1.9894e-15, device='cuda:0', grad_fn=<DivBackward0>)
episode: 224 training return: tensor(4.1899e-15, device='cuda:0', grad_fn=<AddBackward0>)
episode: 225 training return: tensor(5.1731e-15, device='cuda:0', grad_fn=<AddBackward0>)
episode: 226 training return: tensor(1.7984e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 227 training return: tensor(1.5806e-12, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 57 test_true_pfm: -176 test_simulate_pfm tensor(6.4016e-14, device='cuda:0', grad_fn=<DivBackward0>)
episode: 228 training return: tensor(1.2931e-14, device='cuda:0', grad_fn=<AddBackward0>)
episode: 229 training return: tensor(7.3260e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 230 training return: tensor(1.2011e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 231 training return: tensor(1.3719e-12, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 58 test_true_pfm: -283 test_simulate_pfm tensor(5.4647e-16, device='cuda:0', grad_fn=<DivBackward0>)
episode: 232 training return: tensor(1.3211e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 233 training return: tensor(2.5758e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 234 training return: tensor(2.1098e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 235 training return: tensor(6.7224e-12, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 59 test_true_pfm: -551 test_simulate_pfm tensor(4.9256e-13, device='cuda:0', grad_fn=<DivBackward0>)
episode: 236 training return: tensor(2.6288e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 237 training return: tensor(3.6395e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 238 training return: tensor(2.6519e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 239 training return: tensor(1.5040e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 60 test_true_pfm: -674 test_simulate_pfm tensor(1.1488e-11, device='cuda:0', grad_fn=<DivBackward0>)
episode: 240 training return: tensor(2.3309e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 241 training return: tensor(2.2355e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 242 training return: tensor(8.5293e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 243 training return: tensor(1.2167e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 61 test_true_pfm: -490 test_simulate_pfm tensor(8.2698e-10, device='cuda:0', grad_fn=<DivBackward0>)
episode: 244 training return: tensor(1.0427e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 245 training return: tensor(1.2119e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 246 training return: tensor(5.2113e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 247 training return: tensor(1.9091e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 62 test_true_pfm: -916 test_simulate_pfm tensor(6.6787e-11, device='cuda:0', grad_fn=<DivBackward0>)
episode: 248 training return: tensor(5.6480e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 249 training return: tensor(6.6320e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 250 training return: tensor(8.8800e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 251 training return: tensor(4.3481e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 63 test_true_pfm: -802 test_simulate_pfm tensor(3.7889e-11, device='cuda:0', grad_fn=<DivBackward0>)
episode: 252 training return: tensor(5.2684e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 253 training return: tensor(1.8238e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 254 training return: tensor(6.4891e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 255 training return: tensor(1.0116e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 64 test_true_pfm: -774 test_simulate_pfm tensor(1.2025e-10, device='cuda:0', grad_fn=<DivBackward0>)
episode: 256 training return: tensor(1.1497e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 257 training return: tensor(1.0601e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 258 training return: tensor(3.5822e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 259 training return: tensor(3.4049e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 65 test_true_pfm: -677 test_simulate_pfm tensor(2.4572e-10, device='cuda:0', grad_fn=<DivBackward0>)
episode: 260 training return: tensor(8.9978e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 261 training return: tensor(3.5461e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 262 training return: tensor(2.2125e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 263 training return: tensor(4.4213e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 66 test_true_pfm: -410 test_simulate_pfm tensor(6.4341e-09, device='cuda:0', grad_fn=<DivBackward0>)
episode: 264 training return: tensor(1.2256e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 265 training return: tensor(1.1263e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 266 training return: tensor(5.7655e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 267 training return: tensor(8.5524e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 67 test_true_pfm: -913 test_simulate_pfm tensor(1.5481e-10, device='cuda:0', grad_fn=<DivBackward0>)
episode: 268 training return: tensor(7.4637e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 269 training return: tensor(8.0328e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 270 training return: tensor(2.1214e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 271 training return: tensor(3.1241e-08, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 68 test_true_pfm: -483 test_simulate_pfm tensor(1.8616e-10, device='cuda:0', grad_fn=<DivBackward0>)
episode: 272 training return: tensor(6.3961e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 273 training return: tensor(4.4526e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 274 training return: tensor(1.3925e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 275 training return: tensor(1.5733e-07, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 69 test_true_pfm: -376 test_simulate_pfm tensor(2.7173e-08, device='cuda:0', grad_fn=<DivBackward0>)
episode: 276 training return: tensor(1.4191e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 277 training return: tensor(1.1844e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 278 training return: tensor(1.5924e-06, device='cuda:0', grad_fn=<AddBackward0>)
episode: 279 training return: tensor(1.2017e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 70 test_true_pfm: -420 test_simulate_pfm tensor(5.3572e-05, device='cuda:0', grad_fn=<DivBackward0>)
episode: 280 training return: tensor(5.3614e-06, device='cuda:0', grad_fn=<AddBackward0>)
episode: 281 training return: tensor(6.4034e-07, device='cuda:0', grad_fn=<AddBackward0>)
episode: 282 training return: tensor(5.8679e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 283 training return: tensor(1.5529e-08, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 71 test_true_pfm: -498 test_simulate_pfm tensor(7.6584e-06, device='cuda:0', grad_fn=<DivBackward0>)
episode: 284 training return: tensor(1.9703e-06, device='cuda:0', grad_fn=<AddBackward0>)
episode: 285 training return: tensor(3.9550e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 286 training return: tensor(6.1664e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 287 training return: tensor(4.2888e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 72 test_true_pfm: -494 test_simulate_pfm tensor(9.4465e-10, device='cuda:0', grad_fn=<DivBackward0>)
episode: 288 training return: tensor(3.7166e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 289 training return: tensor(1.6122e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 290 training return: tensor(7.0385e-06, device='cuda:0', grad_fn=<AddBackward0>)
episode: 291 training return: tensor(1.8576e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 73 test_true_pfm: -657 test_simulate_pfm tensor(7.7315e-08, device='cuda:0', grad_fn=<DivBackward0>)
episode: 292 training return: tensor(5.3253e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 293 training return: tensor(2.2092e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 294 training return: tensor(4.6816e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 295 training return: tensor(3.3677e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 74 test_true_pfm: -487 test_simulate_pfm tensor(4.5269e-08, device='cuda:0', grad_fn=<DivBackward0>)
episode: 296 training return: tensor(1.4337e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 297 training return: tensor(1.1682e-07, device='cuda:0', grad_fn=<AddBackward0>)
episode: 298 training return: tensor(4.2558e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 299 training return: tensor(1.7819e-08, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 75 test_true_pfm: -525 test_simulate_pfm tensor(2.9152e-10, device='cuda:0', grad_fn=<DivBackward0>)
episode: 300 training return: tensor(1.3372e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 301 training return: tensor(5.9991e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 302 training return: tensor(4.0451e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 303 training return: tensor(1.8971e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 76 test_true_pfm: -541 test_simulate_pfm tensor(8.7541e-10, device='cuda:0', grad_fn=<DivBackward0>)
episode: 304 training return: tensor(7.6019e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 305 training return: tensor(3.6777e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 306 training return: tensor(2.3668e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 307 training return: tensor(2.7284e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 77 test_true_pfm: -532 test_simulate_pfm tensor(1.7715e-09, device='cuda:0', grad_fn=<DivBackward0>)
episode: 308 training return: tensor(2.6600e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 309 training return: tensor(2.7441e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 310 training return: tensor(6.0501e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 311 training return: tensor(5.7872e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 78 test_true_pfm: -651 test_simulate_pfm tensor(1.7902e-09, device='cuda:0', grad_fn=<DivBackward0>)
episode: 312 training return: tensor(6.1530e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 313 training return: tensor(5.6127e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 314 training return: tensor(7.6572e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 315 training return: tensor(2.5264e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 79 test_true_pfm: -656 test_simulate_pfm tensor(1.9996e-08, device='cuda:0', grad_fn=<DivBackward0>)
episode: 316 training return: tensor(7.8122e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 317 training return: tensor(3.3989e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 318 training return: tensor(6.0158e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 319 training return: tensor(2.1861e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 80 test_true_pfm: -567 test_simulate_pfm tensor(7.4287e-10, device='cuda:0', grad_fn=<DivBackward0>)
episode: 320 training return: tensor(5.7723e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 321 training return: tensor(1.6102e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 322 training return: tensor(9.6701e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 323 training return: tensor(8.3611e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 81 test_true_pfm: -556 test_simulate_pfm tensor(7.5234e-08, device='cuda:0', grad_fn=<DivBackward0>)
episode: 324 training return: tensor(2.1185e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 325 training return: tensor(1.2247e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 326 training return: tensor(7.2621e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 327 training return: tensor(3.6648e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 82 test_true_pfm: -539 test_simulate_pfm tensor(4.8706e-10, device='cuda:0', grad_fn=<DivBackward0>)
episode: 328 training return: tensor(4.9082e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 329 training return: tensor(1.0191e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 330 training return: tensor(1.7171e-07, device='cuda:0', grad_fn=<AddBackward0>)
episode: 331 training return: tensor(2.3807e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 83 test_true_pfm: -567 test_simulate_pfm tensor(1.1161e-07, device='cuda:0', grad_fn=<DivBackward0>)
episode: 332 training return: tensor(5.4949e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 333 training return: tensor(3.0987e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 334 training return: tensor(2.4752e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 335 training return: tensor(5.3100e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 84 test_true_pfm: -527 test_simulate_pfm tensor(2.8890e-06, device='cuda:0', grad_fn=<DivBackward0>)
episode: 336 training return: tensor(4.3154e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 337 training return: tensor(2.5576e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 338 training return: tensor(1.9591e-06, device='cuda:0', grad_fn=<AddBackward0>)
episode: 339 training return: tensor(1.7859e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 85 test_true_pfm: -508 test_simulate_pfm tensor(5.7396e-09, device='cuda:0', grad_fn=<DivBackward0>)
episode: 340 training return: tensor(2.6695e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 341 training return: tensor(1.8412e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 342 training return: tensor(3.3478e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 343 training return: tensor(1.0744e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 86 test_true_pfm: -553 test_simulate_pfm tensor(5.6680e-07, device='cuda:0', grad_fn=<DivBackward0>)
episode: 344 training return: tensor(4.1228e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 345 training return: tensor(1.6445e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 346 training return: tensor(3.9376e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 347 training return: tensor(1.2866e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 87 test_true_pfm: -463 test_simulate_pfm tensor(4.0896e-10, device='cuda:0', grad_fn=<DivBackward0>)
episode: 348 training return: tensor(7.3923e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 349 training return: tensor(7.6935e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 350 training return: tensor(5.5542e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 351 training return: tensor(1.0981e-12, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 88 test_true_pfm: -613 test_simulate_pfm tensor(2.8307e-06, device='cuda:0', grad_fn=<DivBackward0>)
episode: 352 training return: tensor(3.1811e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 353 training return: tensor(1.1270e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 354 training return: tensor(4.4809e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 355 training return: tensor(1.2648e-07, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 89 test_true_pfm: -495 test_simulate_pfm tensor(1.2429e-07, device='cuda:0', grad_fn=<DivBackward0>)
episode: 356 training return: tensor(2.7945e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 357 training return: tensor(3.7103e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 358 training return: tensor(6.7581e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 359 training return: tensor(1.9935e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 90 test_true_pfm: -607 test_simulate_pfm tensor(1.0039e-08, device='cuda:0', grad_fn=<DivBackward0>)
episode: 360 training return: tensor(6.5319e-07, device='cuda:0', grad_fn=<AddBackward0>)
episode: 361 training return: tensor(5.0998e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 362 training return: tensor(5.2986e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 363 training return: tensor(3.3427e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 91 test_true_pfm: -619 test_simulate_pfm tensor(8.6072e-11, device='cuda:0', grad_fn=<DivBackward0>)
episode: 364 training return: tensor(5.4161e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 365 training return: tensor(7.3252e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 366 training return: tensor(2.6103e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 367 training return: tensor(6.9719e-12, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 92 test_true_pfm: -461 test_simulate_pfm tensor(4.3518e-07, device='cuda:0', grad_fn=<DivBackward0>)
episode: 368 training return: tensor(7.6526e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 369 training return: tensor(1.0123e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 370 training return: tensor(1.3940e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 371 training return: tensor(3.0862e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 93 test_true_pfm: -556 test_simulate_pfm tensor(1.6550e-10, device='cuda:0', grad_fn=<DivBackward0>)
episode: 372 training return: tensor(2.5000e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 373 training return: tensor(1.4165e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 374 training return: tensor(1.3709e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 375 training return: tensor(1.2507e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 94 test_true_pfm: -542 test_simulate_pfm tensor(7.7322e-07, device='cuda:0', grad_fn=<DivBackward0>)
episode: 376 training return: tensor(2.7671e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 377 training return: tensor(2.2605e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 378 training return: tensor(1.3295e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 379 training return: tensor(1.6019e-06, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 95 test_true_pfm: -574 test_simulate_pfm tensor(1.3396e-11, device='cuda:0', grad_fn=<DivBackward0>)
episode: 380 training return: tensor(4.2296e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 381 training return: tensor(4.3353e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 382 training return: tensor(7.8084e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 383 training return: tensor(2.5313e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 96 test_true_pfm: -454 test_simulate_pfm tensor(2.6441e-10, device='cuda:0', grad_fn=<DivBackward0>)
episode: 384 training return: tensor(4.5009e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 385 training return: tensor(7.6788e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 386 training return: tensor(2.3704e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 387 training return: tensor(2.7079e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 97 test_true_pfm: -542 test_simulate_pfm tensor(1.8442e-11, device='cuda:0', grad_fn=<DivBackward0>)
episode: 388 training return: tensor(3.6028e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 389 training return: tensor(4.3579e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 390 training return: tensor(1.0764e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 391 training return: tensor(2.7373e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 98 test_true_pfm: -429 test_simulate_pfm tensor(2.2150e-10, device='cuda:0', grad_fn=<DivBackward0>)
episode: 392 training return: tensor(7.4265e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 393 training return: tensor(6.4469e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 394 training return: tensor(6.4119e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 395 training return: tensor(4.3885e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 99 test_true_pfm: -497 test_simulate_pfm tensor(4.3628e-11, device='cuda:0', grad_fn=<DivBackward0>)
episode: 396 training return: tensor(7.4290e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 397 training return: tensor(7.7643e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 398 training return: tensor(1.1163e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 399 training return: tensor(1.0420e-13, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 100 test_true_pfm: -555 test_simulate_pfm tensor(8.7210e-11, device='cuda:0', grad_fn=<DivBackward0>)
episode: 400 training return: tensor(2.7584e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 401 training return: tensor(1.3386e-07, device='cuda:0', grad_fn=<AddBackward0>)
episode: 402 training return: tensor(2.9078e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 403 training return: tensor(1.5958e-12, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 101 test_true_pfm: -468 test_simulate_pfm tensor(1.2670e-07, device='cuda:0', grad_fn=<DivBackward0>)
episode: 404 training return: tensor(2.7874e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 405 training return: tensor(2.6235e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 406 training return: tensor(2.8999e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 407 training return: tensor(1.9766e-13, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 102 test_true_pfm: -430 test_simulate_pfm tensor(1.3853e-10, device='cuda:0', grad_fn=<DivBackward0>)
episode: 408 training return: tensor(1.9166e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 409 training return: tensor(2.8386e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 410 training return: tensor(2.0351e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 411 training return: tensor(1.9500e-13, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 103 test_true_pfm: -492 test_simulate_pfm tensor(4.0022e-12, device='cuda:0', grad_fn=<DivBackward0>)
episode: 412 training return: tensor(8.5115e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 413 training return: tensor(5.1907e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 414 training return: tensor(8.7997e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 415 training return: tensor(3.8499e-12, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 104 test_true_pfm: -466 test_simulate_pfm tensor(3.4405e-10, device='cuda:0', grad_fn=<DivBackward0>)
episode: 416 training return: tensor(9.3940e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 417 training return: tensor(9.1629e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 418 training return: tensor(1.0391e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 419 training return: tensor(8.4939e-12, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 105 test_true_pfm: -453 test_simulate_pfm tensor(2.8460e-11, device='cuda:0', grad_fn=<DivBackward0>)
episode: 420 training return: tensor(4.2730e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 421 training return: tensor(3.5138e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 422 training return: tensor(2.0051e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 423 training return: tensor(2.1041e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 106 test_true_pfm: -455 test_simulate_pfm tensor(1.2079e-09, device='cuda:0', grad_fn=<DivBackward0>)
episode: 424 training return: tensor(1.0777e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 425 training return: tensor(1.9448e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 426 training return: tensor(2.6806e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 427 training return: tensor(1.7708e-13, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 107 test_true_pfm: -413 test_simulate_pfm tensor(8.9053e-13, device='cuda:0', grad_fn=<DivBackward0>)
episode: 428 training return: tensor(5.4460e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 429 training return: tensor(4.3112e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 430 training return: tensor(3.6193e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 431 training return: tensor(1.4185e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 108 test_true_pfm: -377 test_simulate_pfm tensor(5.7232e-13, device='cuda:0', grad_fn=<DivBackward0>)
episode: 432 training return: tensor(2.0356e-14, device='cuda:0', grad_fn=<AddBackward0>)
episode: 433 training return: tensor(1.0609e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 434 training return: tensor(2.3200e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 435 training return: tensor(1.1791e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 109 test_true_pfm: -408 test_simulate_pfm tensor(1.1682e-13, device='cuda:0', grad_fn=<DivBackward0>)
episode: 436 training return: tensor(1.7522e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 437 training return: tensor(3.7627e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 438 training return: tensor(1.8043e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 439 training return: tensor(3.3326e-12, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 110 test_true_pfm: -371 test_simulate_pfm tensor(5.5186e-11, device='cuda:0', grad_fn=<DivBackward0>)
episode: 440 training return: tensor(3.4081e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 441 training return: tensor(1.4665e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 442 training return: tensor(1.1350e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 443 training return: tensor(1.3070e-08, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 111 test_true_pfm: -460 test_simulate_pfm tensor(4.4181e-11, device='cuda:0', grad_fn=<DivBackward0>)
episode: 444 training return: tensor(2.3049e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 445 training return: tensor(2.4570e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 446 training return: tensor(1.1863e-06, device='cuda:0', grad_fn=<AddBackward0>)
episode: 447 training return: tensor(1.4439e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 112 test_true_pfm: -427 test_simulate_pfm tensor(1.2346e-09, device='cuda:0', grad_fn=<DivBackward0>)
episode: 448 training return: tensor(1.7725e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 449 training return: tensor(2.0442e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 450 training return: tensor(4.1666e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 451 training return: tensor(8.2699e-13, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 113 test_true_pfm: -466 test_simulate_pfm tensor(9.1333e-10, device='cuda:0', grad_fn=<DivBackward0>)
episode: 452 training return: tensor(8.3694e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 453 training return: tensor(7.1123e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 454 training return: tensor(1.0464e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 455 training return: tensor(3.0827e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 114 test_true_pfm: -503 test_simulate_pfm tensor(2.6779e-09, device='cuda:0', grad_fn=<DivBackward0>)
episode: 456 training return: tensor(4.8827e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 457 training return: tensor(9.4305e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 458 training return: tensor(3.6537e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 459 training return: tensor(3.7374e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 115 test_true_pfm: -451 test_simulate_pfm tensor(2.9391e-09, device='cuda:0', grad_fn=<DivBackward0>)
episode: 460 training return: tensor(1.3598e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 461 training return: tensor(1.2790e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 462 training return: tensor(4.2784e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 463 training return: tensor(1.1880e-08, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 116 test_true_pfm: -490 test_simulate_pfm tensor(3.5157e-11, device='cuda:0', grad_fn=<DivBackward0>)
episode: 464 training return: tensor(1.5462e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 465 training return: tensor(5.8409e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 466 training return: tensor(1.3068e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 467 training return: tensor(2.6181e-12, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 117 test_true_pfm: -454 test_simulate_pfm tensor(1.7120e-09, device='cuda:0', grad_fn=<DivBackward0>)
episode: 468 training return: tensor(7.8112e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 469 training return: tensor(3.4578e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 470 training return: tensor(7.4084e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 471 training return: tensor(4.1509e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 118 test_true_pfm: -447 test_simulate_pfm tensor(1.4677e-07, device='cuda:0', grad_fn=<DivBackward0>)
episode: 472 training return: tensor(4.5035e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 473 training return: tensor(7.3263e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 474 training return: tensor(2.0525e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 475 training return: tensor(9.5680e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 119 test_true_pfm: -439 test_simulate_pfm tensor(3.4519e-09, device='cuda:0', grad_fn=<DivBackward0>)
episode: 476 training return: tensor(9.0904e-07, device='cuda:0', grad_fn=<AddBackward0>)
episode: 477 training return: tensor(6.5225e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 478 training return: tensor(2.8957e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 479 training return: tensor(2.8323e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 120 test_true_pfm: -549 test_simulate_pfm tensor(1.0804e-08, device='cuda:0', grad_fn=<DivBackward0>)
episode: 480 training return: tensor(1.3880e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 481 training return: tensor(3.8153e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 482 training return: tensor(5.1743e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 483 training return: tensor(4.4970e-13, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 121 test_true_pfm: -445 test_simulate_pfm tensor(7.4325e-09, device='cuda:0', grad_fn=<DivBackward0>)
episode: 484 training return: tensor(1.3216e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 485 training return: tensor(2.4989e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 486 training return: tensor(1.9146e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 487 training return: tensor(6.0999e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 122 test_true_pfm: -420 test_simulate_pfm tensor(2.0235e-10, device='cuda:0', grad_fn=<DivBackward0>)
episode: 488 training return: tensor(7.5458e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 489 training return: tensor(4.0709e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 490 training return: tensor(2.7803e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 491 training return: tensor(4.0517e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 123 test_true_pfm: -449 test_simulate_pfm tensor(1.9394e-09, device='cuda:0', grad_fn=<DivBackward0>)
episode: 492 training return: tensor(7.7056e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 493 training return: tensor(4.0239e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 494 training return: tensor(2.9883e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 495 training return: tensor(2.3016e-14, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 124 test_true_pfm: -477 test_simulate_pfm tensor(8.0636e-09, device='cuda:0', grad_fn=<DivBackward0>)
episode: 496 training return: tensor(6.1287e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 497 training return: tensor(2.3494e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 498 training return: tensor(1.8671e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 499 training return: tensor(1.3414e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 125 test_true_pfm: -558 test_simulate_pfm tensor(4.3343e-08, device='cuda:0', grad_fn=<DivBackward0>)
episode: 500 training return: tensor(4.2950e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 501 training return: tensor(1.0641e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 502 training return: tensor(8.7001e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 503 training return: tensor(4.3611e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 126 test_true_pfm: -422 test_simulate_pfm tensor(6.8733e-09, device='cuda:0', grad_fn=<DivBackward0>)
episode: 504 training return: tensor(1.7033e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 505 training return: tensor(1.4340e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 506 training return: tensor(4.2260e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 507 training return: tensor(1.6212e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 127 test_true_pfm: -501 test_simulate_pfm tensor(4.3098e-11, device='cuda:0', grad_fn=<DivBackward0>)
episode: 508 training return: tensor(7.2634e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 509 training return: tensor(8.7939e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 510 training return: tensor(3.4032e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 511 training return: tensor(5.2269e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 128 test_true_pfm: -486 test_simulate_pfm tensor(1.8249e-10, device='cuda:0', grad_fn=<DivBackward0>)
episode: 512 training return: tensor(9.6880e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 513 training return: tensor(2.7871e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 514 training return: tensor(1.9409e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 515 training return: tensor(2.3742e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 129 test_true_pfm: -491 test_simulate_pfm tensor(1.2659e-09, device='cuda:0', grad_fn=<DivBackward0>)
episode: 516 training return: tensor(2.3485e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 517 training return: tensor(2.4483e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 518 training return: tensor(4.7091e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 519 training return: tensor(2.7241e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 130 test_true_pfm: -541 test_simulate_pfm tensor(4.9788e-10, device='cuda:0', grad_fn=<DivBackward0>)
episode: 520 training return: tensor(7.4357e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 521 training return: tensor(1.7410e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 522 training return: tensor(4.9806e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 523 training return: tensor(2.2862e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 131 test_true_pfm: -556 test_simulate_pfm tensor(3.9393e-07, device='cuda:0', grad_fn=<DivBackward0>)
episode: 524 training return: tensor(7.2927e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 525 training return: tensor(1.3765e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 526 training return: tensor(7.0304e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 527 training return: tensor(8.4262e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 132 test_true_pfm: -532 test_simulate_pfm tensor(4.2695e-10, device='cuda:0', grad_fn=<DivBackward0>)
episode: 528 training return: tensor(5.7007e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 529 training return: tensor(2.3255e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 530 training return: tensor(6.1866e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 531 training return: tensor(1.0007e-12, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 133 test_true_pfm: -493 test_simulate_pfm tensor(7.3026e-11, device='cuda:0', grad_fn=<DivBackward0>)
episode: 532 training return: tensor(2.6758e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 533 training return: tensor(5.2129e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 534 training return: tensor(4.0421e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 535 training return: tensor(1.5995e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 134 test_true_pfm: -574 test_simulate_pfm tensor(1.1267e-07, device='cuda:0', grad_fn=<DivBackward0>)
episode: 536 training return: tensor(3.1926e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 537 training return: tensor(1.9131e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 538 training return: tensor(3.5482e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 539 training return: tensor(1.2080e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 135 test_true_pfm: -547 test_simulate_pfm tensor(4.6430e-09, device='cuda:0', grad_fn=<DivBackward0>)
episode: 540 training return: tensor(3.0808e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 541 training return: tensor(7.8411e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 542 training return: tensor(4.5628e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 543 training return: tensor(2.8368e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 136 test_true_pfm: -525 test_simulate_pfm tensor(1.2140e-10, device='cuda:0', grad_fn=<DivBackward0>)
episode: 544 training return: tensor(9.3865e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 545 training return: tensor(1.3071e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 546 training return: tensor(5.4569e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 547 training return: tensor(4.9176e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 137 test_true_pfm: -550 test_simulate_pfm tensor(4.3445e-10, device='cuda:0', grad_fn=<DivBackward0>)
episode: 548 training return: tensor(1.0879e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 549 training return: tensor(4.9976e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 550 training return: tensor(1.3094e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 551 training return: tensor(2.6161e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 138 test_true_pfm: -490 test_simulate_pfm tensor(1.5365e-09, device='cuda:0', grad_fn=<DivBackward0>)
episode: 552 training return: tensor(3.3810e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 553 training return: tensor(1.8635e-07, device='cuda:0', grad_fn=<AddBackward0>)
episode: 554 training return: tensor(1.6856e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 555 training return: tensor(1.6771e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 139 test_true_pfm: -481 test_simulate_pfm tensor(1.2588e-08, device='cuda:0', grad_fn=<DivBackward0>)
episode: 556 training return: tensor(3.3313e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 557 training return: tensor(2.3730e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 558 training return: tensor(1.2039e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 559 training return: tensor(1.0967e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 140 test_true_pfm: -382 test_simulate_pfm tensor(3.8087e-09, device='cuda:0', grad_fn=<DivBackward0>)
episode: 560 training return: tensor(1.2334e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 561 training return: tensor(5.8816e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 562 training return: tensor(4.2802e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 563 training return: tensor(6.0983e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 141 test_true_pfm: -494 test_simulate_pfm tensor(3.4582e-10, device='cuda:0', grad_fn=<DivBackward0>)
episode: 564 training return: tensor(4.4251e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 565 training return: tensor(1.3024e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 566 training return: tensor(4.1177e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 567 training return: tensor(1.7645e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 142 test_true_pfm: -438 test_simulate_pfm tensor(1.8951e-09, device='cuda:0', grad_fn=<DivBackward0>)
episode: 568 training return: tensor(6.8846e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 569 training return: tensor(1.4083e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 570 training return: tensor(1.7592e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 571 training return: tensor(2.5285e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 143 test_true_pfm: -493 test_simulate_pfm tensor(3.5763e-10, device='cuda:0', grad_fn=<DivBackward0>)
episode: 572 training return: tensor(1.5519e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 573 training return: tensor(4.5635e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 574 training return: tensor(3.9950e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 575 training return: tensor(2.0826e-12, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 144 test_true_pfm: -585 test_simulate_pfm tensor(6.9064e-09, device='cuda:0', grad_fn=<DivBackward0>)
episode: 576 training return: tensor(1.4269e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 577 training return: tensor(1.8030e-07, device='cuda:0', grad_fn=<AddBackward0>)
episode: 578 training return: tensor(2.3250e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 579 training return: tensor(3.9522e-13, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 145 test_true_pfm: -399 test_simulate_pfm tensor(2.3423e-10, device='cuda:0', grad_fn=<DivBackward0>)
episode: 580 training return: tensor(1.5637e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 581 training return: tensor(3.0434e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 582 training return: tensor(1.7501e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 583 training return: tensor(6.1485e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 146 test_true_pfm: -508 test_simulate_pfm tensor(2.1296e-09, device='cuda:0', grad_fn=<DivBackward0>)
episode: 584 training return: tensor(4.8183e-14, device='cuda:0', grad_fn=<AddBackward0>)
episode: 585 training return: tensor(1.7000e-14, device='cuda:0', grad_fn=<AddBackward0>)
episode: 586 training return: tensor(1.0636e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 587 training return: tensor(1.8965e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 147 test_true_pfm: -513 test_simulate_pfm tensor(2.9936e-10, device='cuda:0', grad_fn=<DivBackward0>)
episode: 588 training return: tensor(3.5858e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 589 training return: tensor(1.6936e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 590 training return: tensor(2.3106e-14, device='cuda:0', grad_fn=<AddBackward0>)
episode: 591 training return: tensor(1.4425e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 148 test_true_pfm: -563 test_simulate_pfm tensor(1.0506e-11, device='cuda:0', grad_fn=<DivBackward0>)
episode: 592 training return: tensor(3.2982e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 593 training return: tensor(4.3004e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 594 training return: tensor(2.6050e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 595 training return: tensor(1.0643e-12, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 149 test_true_pfm: -514 test_simulate_pfm tensor(1.6432e-10, device='cuda:0', grad_fn=<DivBackward0>)
episode: 596 training return: tensor(5.8289e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 597 training return: tensor(3.1801e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 598 training return: tensor(3.1100e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 599 training return: tensor(4.6353e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 150 test_true_pfm: -521 test_simulate_pfm tensor(1.4166e-08, device='cuda:0', grad_fn=<DivBackward0>)
