initial performance: -4
episode: 0 training return: tensor(1.2208e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 1 training return: tensor(1.7056e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 2 training return: tensor(1.3573e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 3 training return: tensor(1.5248e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 1 test_pfm: -565
episode: 4 training return: tensor(4.2376e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 5 training return: tensor(3.7663e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 6 training return: tensor(4.8815e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 7 training return: tensor(7.0235e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 2 test_pfm: -455
episode: 8 training return: tensor(3.7356e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 9 training return: tensor(4.5532e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 10 training return: tensor(1.7042e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 11 training return: tensor(2.1357e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 3 test_pfm: -512
episode: 12 training return: tensor(4.2014e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 13 training return: tensor(1.2612e-05, device='cuda:0', grad_fn=<AddBackward0>)
episode: 14 training return: tensor(1.6043e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 15 training return: tensor(2.1601e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 4 test_pfm: -442
episode: 16 training return: tensor(7.8145e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 17 training return: tensor(4.8822e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 18 training return: tensor(3.6313e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 19 training return: tensor(7.2784e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 5 test_pfm: -304
episode: 20 training return: tensor(1.4170e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 21 training return: tensor(2.2099e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 22 training return: tensor(1.3005e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 23 training return: tensor(9.8321e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 6 test_pfm: -467
episode: 24 training return: tensor(3.2293e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 25 training return: tensor(2.0730e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 26 training return: tensor(1.5810e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 27 training return: tensor(1.3058e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 7 test_pfm: -600
episode: 28 training return: tensor(1.1973e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 29 training return: tensor(1.3463e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 30 training return: tensor(3.2626e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 31 training return: tensor(8.7132e-05, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 8 test_pfm: -622
episode: 32 training return: tensor(7.5684e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 33 training return: tensor(3.8082e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 34 training return: tensor(2.7848e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 35 training return: tensor(3.2742e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 9 test_pfm: -891
episode: 36 training return: tensor(1.5169e-14, device='cuda:0', grad_fn=<AddBackward0>)
episode: 37 training return: tensor(7.4945e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 38 training return: tensor(1.3321e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 39 training return: tensor(1.7064e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 10 test_pfm: -1181
episode: 40 training return: tensor(1.3744e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 41 training return: tensor(2.6787e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 42 training return: tensor(2.8581e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 43 training return: tensor(8.5011e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 11 test_pfm: -595
episode: 44 training return: tensor(2.6682e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 45 training return: tensor(3.7239e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 46 training return: tensor(9.0699e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 47 training return: tensor(1.1513e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 12 test_pfm: -73
episode: 48 training return: tensor(6.5307e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 49 training return: tensor(6.1771e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 50 training return: tensor(4.9062e-07, device='cuda:0', grad_fn=<AddBackward0>)
episode: 51 training return: tensor(4.9211e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 13 test_pfm: -467
episode: 52 training return: tensor(1.3656e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 53 training return: tensor(1.3095e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 54 training return: tensor(1.2916e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 55 training return: tensor(1.3528e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 14 test_pfm: -605
episode: 56 training return: tensor(1.3136e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 57 training return: tensor(1.2244e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 58 training return: tensor(1.2336e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 59 training return: tensor(1.3330e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 15 test_pfm: -598
episode: 60 training return: tensor(8.1889e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 61 training return: tensor(1.0069e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 62 training return: tensor(8.5437e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 63 training return: tensor(9.8874e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 16 test_pfm: -569
episode: 64 training return: tensor(1.0148e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 65 training return: tensor(9.7845e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 66 training return: tensor(8.7130e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 67 training return: tensor(2.3861e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 17 test_pfm: -785
episode: 68 training return: tensor(8.4210e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 69 training return: tensor(6.8502e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 70 training return: tensor(2.4090e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 71 training return: tensor(1.7129e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 18 test_pfm: -728
episode: 72 training return: tensor(3.3087e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 73 training return: tensor(2.7049e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 74 training return: tensor(2.8600e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 75 training return: tensor(1.3998e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 19 test_pfm: -360
episode: 76 training return: tensor(1.8829e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 77 training return: tensor(2.3735e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 78 training return: tensor(1.6763e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 79 training return: tensor(7.7742e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 20 test_pfm: -519
episode: 80 training return: tensor(2.7117e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 81 training return: tensor(1.4930e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 82 training return: tensor(1.8531e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 83 training return: tensor(1.7962e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 21 test_pfm: -816
episode: 84 training return: tensor(2.4310e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 85 training return: tensor(4.1488e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 86 training return: tensor(1.9167e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 87 training return: tensor(3.5743e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 22 test_pfm: -895
episode: 88 training return: tensor(1.3319e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 89 training return: tensor(6.5879e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 90 training return: tensor(1.0515e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 91 training return: tensor(6.3321e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 23 test_pfm: -901
episode: 92 training return: tensor(6.1579e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 93 training return: tensor(1.3513e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 94 training return: tensor(3.2519e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 95 training return: tensor(1.1690e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 24 test_pfm: -937
episode: 96 training return: tensor(4.3617e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 97 training return: tensor(1.0489e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 98 training return: tensor(3.6187e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 99 training return: tensor(6.7236e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 25 test_pfm: -771
episode: 100 training return: tensor(1.0282e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 101 training return: tensor(3.1393e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 102 training return: tensor(5.9149e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 103 training return: tensor(4.8352e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 26 test_pfm: -778
episode: 104 training return: tensor(4.9963e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 105 training return: tensor(4.0926e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 106 training return: tensor(1.3584e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 107 training return: tensor(6.3344e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 27 test_pfm: -759
episode: 108 training return: tensor(1.6642e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 109 training return: tensor(3.0111e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 110 training return: tensor(5.0125e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 111 training return: tensor(8.9912e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 28 test_pfm: -732
episode: 112 training return: tensor(3.7520e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 113 training return: tensor(6.6739e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 114 training return: tensor(8.4853e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 115 training return: tensor(8.2569e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 29 test_pfm: -555
episode: 116 training return: tensor(8.4128e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 117 training return: tensor(9.9801e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 118 training return: tensor(9.5007e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 119 training return: tensor(9.8708e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 30 test_pfm: -548
episode: 120 training return: tensor(1.0669e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 121 training return: tensor(1.1998e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 122 training return: tensor(1.2125e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 123 training return: tensor(1.0377e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 31 test_pfm: -598
episode: 124 training return: tensor(1.3744e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 125 training return: tensor(1.3317e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 126 training return: tensor(8.1929e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 127 training return: tensor(8.9190e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 32 test_pfm: -555
episode: 128 training return: tensor(6.5184e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 129 training return: tensor(5.7179e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 130 training return: tensor(5.1657e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 131 training return: tensor(6.3743e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 33 test_pfm: -549
episode: 132 training return: tensor(6.2258e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 133 training return: tensor(5.8747e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 134 training return: tensor(6.2960e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 135 training return: tensor(6.1219e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 34 test_pfm: -536
episode: 136 training return: tensor(7.5695e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 137 training return: tensor(6.6283e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 138 training return: tensor(8.3066e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 139 training return: tensor(7.0292e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 35 test_pfm: -532
episode: 140 training return: tensor(8.0087e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 141 training return: tensor(7.4778e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 142 training return: tensor(7.5386e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 143 training return: tensor(7.0261e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 36 test_pfm: -550
episode: 144 training return: tensor(4.7542e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 145 training return: tensor(5.3374e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 146 training return: tensor(6.2165e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 147 training return: tensor(5.9122e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 37 test_pfm: -550
episode: 148 training return: tensor(5.0670e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 149 training return: tensor(4.8231e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 150 training return: tensor(4.5501e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 151 training return: tensor(3.0654e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 38 test_pfm: -534
episode: 152 training return: tensor(3.6611e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 153 training return: tensor(2.7635e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 154 training return: tensor(4.2792e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 155 training return: tensor(5.8443e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 39 test_pfm: -539
episode: 156 training return: tensor(6.7871e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 157 training return: tensor(7.9298e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 158 training return: tensor(7.5674e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 159 training return: tensor(7.1886e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 40 test_pfm: -540
episode: 160 training return: tensor(9.1837e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 161 training return: tensor(6.9155e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 162 training return: tensor(7.3177e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 163 training return: tensor(7.5534e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 41 test_pfm: -540
episode: 164 training return: tensor(7.8565e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 165 training return: tensor(7.5844e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 166 training return: tensor(7.1861e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 167 training return: tensor(6.1496e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 42 test_pfm: -561
episode: 168 training return: tensor(6.1714e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 169 training return: tensor(9.2687e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 170 training return: tensor(1.4243e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 171 training return: tensor(1.1564e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 43 test_pfm: -522
episode: 172 training return: tensor(6.2329e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 173 training return: tensor(1.1934e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 174 training return: tensor(1.4045e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 175 training return: tensor(7.8662e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 44 test_pfm: -565
episode: 176 training return: tensor(1.6511e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 177 training return: tensor(6.2123e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 178 training return: tensor(1.9353e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 179 training return: tensor(2.8184e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 45 test_pfm: -653
episode: 180 training return: tensor(9.1634e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 181 training return: tensor(8.1391e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 182 training return: tensor(1.0508e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 183 training return: tensor(3.7156e-12, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 46 test_pfm: -877
episode: 184 training return: tensor(3.1527e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 185 training return: tensor(2.7669e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 186 training return: tensor(4.9703e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 187 training return: tensor(4.4782e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 47 test_pfm: -878
episode: 188 training return: tensor(5.2051e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 189 training return: tensor(4.7865e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 190 training return: tensor(3.5618e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 191 training return: tensor(9.1566e-07, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 48 test_pfm: -648
episode: 192 training return: tensor(1.7640e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 193 training return: tensor(8.2046e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 194 training return: tensor(1.1357e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 195 training return: tensor(3.8033e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 49 test_pfm: -646
episode: 196 training return: tensor(1.1809e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 197 training return: tensor(8.4059e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 198 training return: tensor(4.8713e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 199 training return: tensor(1.0105e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 50 test_pfm: -722
episode: 200 training return: tensor(5.1322e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 201 training return: tensor(1.7697e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 202 training return: tensor(5.1807e-07, device='cuda:0', grad_fn=<AddBackward0>)
episode: 203 training return: tensor(3.5788e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 51 test_pfm: -943
episode: 204 training return: tensor(1.7564e-07, device='cuda:0', grad_fn=<AddBackward0>)
episode: 205 training return: tensor(7.9009e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 206 training return: tensor(1.0695e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 207 training return: tensor(1.3629e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 52 test_pfm: -609
episode: 208 training return: tensor(1.3655e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 209 training return: tensor(7.8286e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 210 training return: tensor(1.0113e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 211 training return: tensor(5.1659e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 53 test_pfm: -802
episode: 212 training return: tensor(2.8078e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 213 training return: tensor(2.5238e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 214 training return: tensor(1.3532e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 215 training return: tensor(1.2232e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 54 test_pfm: -747
episode: 216 training return: tensor(1.2643e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 217 training return: tensor(2.0268e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 218 training return: tensor(1.1392e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 219 training return: tensor(2.8101e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 55 test_pfm: -814
episode: 220 training return: tensor(8.3555e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 221 training return: tensor(3.7200e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 222 training return: tensor(1.7947e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 223 training return: tensor(1.2123e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 56 test_pfm: -819
episode: 224 training return: tensor(8.8440e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 225 training return: tensor(3.6491e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 226 training return: tensor(2.2242e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 227 training return: tensor(1.6448e-12, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 57 test_pfm: -754
episode: 228 training return: tensor(3.8769e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 229 training return: tensor(2.3335e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 230 training return: tensor(1.4040e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 231 training return: tensor(1.3325e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 58 test_pfm: -840
episode: 232 training return: tensor(5.0963e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 233 training return: tensor(3.9064e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 234 training return: tensor(1.6427e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 235 training return: tensor(2.5849e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 59 test_pfm: -793
episode: 236 training return: tensor(6.1845e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 237 training return: tensor(1.5305e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 238 training return: tensor(4.8555e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 239 training return: tensor(6.5244e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 60 test_pfm: -988
episode: 240 training return: tensor(3.0334e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 241 training return: tensor(4.8864e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 242 training return: tensor(2.3236e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 243 training return: tensor(1.8704e-12, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 61 test_pfm: -840
episode: 244 training return: tensor(3.8969e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 245 training return: tensor(1.8491e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 246 training return: tensor(3.7704e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 247 training return: tensor(7.9368e-12, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 62 test_pfm: -748
episode: 248 training return: tensor(3.4727e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 249 training return: tensor(3.2385e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 250 training return: tensor(8.7173e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 251 training return: tensor(2.5500e-12, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 63 test_pfm: -688
episode: 252 training return: tensor(1.1130e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 253 training return: tensor(4.3192e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 254 training return: tensor(4.2438e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 255 training return: tensor(5.1458e-12, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 64 test_pfm: -811
episode: 256 training return: tensor(1.6473e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 257 training return: tensor(4.4301e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 258 training return: tensor(6.9337e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 259 training return: tensor(1.4096e-12, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 65 test_pfm: -756
episode: 260 training return: tensor(4.8871e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 261 training return: tensor(5.5542e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 262 training return: tensor(1.7826e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 263 training return: tensor(2.0247e-12, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 66 test_pfm: -790
episode: 264 training return: tensor(2.1686e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 265 training return: tensor(2.3888e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 266 training return: tensor(1.5423e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 267 training return: tensor(6.0379e-13, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 67 test_pfm: -800
episode: 268 training return: tensor(1.4199e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 269 training return: tensor(1.8753e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 270 training return: tensor(1.2602e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 271 training return: tensor(6.0394e-12, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 68 test_pfm: -869
episode: 272 training return: tensor(6.6627e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 273 training return: tensor(4.4598e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 274 training return: tensor(4.2850e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 275 training return: tensor(3.6540e-12, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 69 test_pfm: -809
episode: 276 training return: tensor(7.1354e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 277 training return: tensor(3.6445e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 278 training return: tensor(2.7018e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 279 training return: tensor(3.7375e-12, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 70 test_pfm: -803
episode: 280 training return: tensor(1.9440e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 281 training return: tensor(8.0323e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 282 training return: tensor(1.0551e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 283 training return: tensor(1.6778e-13, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 71 test_pfm: -787
episode: 284 training return: tensor(4.4726e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 285 training return: tensor(7.3952e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 286 training return: tensor(3.2794e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 287 training return: tensor(2.0022e-13, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 72 test_pfm: -786
episode: 288 training return: tensor(4.8590e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 289 training return: tensor(7.7533e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 290 training return: tensor(3.5326e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 291 training return: tensor(6.8243e-12, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 73 test_pfm: -803
episode: 292 training return: tensor(3.1060e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 293 training return: tensor(1.5110e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 294 training return: tensor(2.0782e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 295 training return: tensor(8.4057e-12, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 74 test_pfm: -808
episode: 296 training return: tensor(1.9293e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 297 training return: tensor(6.2646e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 298 training return: tensor(7.2177e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 299 training return: tensor(4.4196e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 75 test_pfm: -622
episode: 300 training return: tensor(5.6227e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 301 training return: tensor(6.7608e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 302 training return: tensor(1.4238e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 303 training return: tensor(1.5042e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 76 test_pfm: -612
episode: 304 training return: tensor(1.2375e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 305 training return: tensor(1.4000e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 306 training return: tensor(1.1922e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 307 training return: tensor(1.1776e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 77 test_pfm: -605
episode: 308 training return: tensor(1.3527e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 309 training return: tensor(1.2508e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 310 training return: tensor(1.2225e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 311 training return: tensor(1.2104e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 78 test_pfm: -611
episode: 312 training return: tensor(1.2223e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 313 training return: tensor(1.2175e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 314 training return: tensor(1.2325e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 315 training return: tensor(1.0801e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 79 test_pfm: -595
episode: 316 training return: tensor(1.3556e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 317 training return: tensor(1.2100e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 318 training return: tensor(1.0416e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 319 training return: tensor(9.8430e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 80 test_pfm: -606
episode: 320 training return: tensor(7.1827e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 321 training return: tensor(5.7133e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 322 training return: tensor(7.6587e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 323 training return: tensor(1.1153e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 81 test_pfm: -584
episode: 324 training return: tensor(5.6126e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 325 training return: tensor(5.0384e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 326 training return: tensor(3.3522e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 327 training return: tensor(4.1046e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 82 test_pfm: -598
episode: 328 training return: tensor(3.9913e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 329 training return: tensor(7.0573e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 330 training return: tensor(8.6430e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 331 training return: tensor(6.8575e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 83 test_pfm: -586
episode: 332 training return: tensor(9.9431e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 333 training return: tensor(8.3483e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 334 training return: tensor(8.5981e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 335 training return: tensor(1.0054e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 84 test_pfm: -597
episode: 336 training return: tensor(7.5333e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 337 training return: tensor(9.3236e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 338 training return: tensor(1.1210e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 339 training return: tensor(1.0142e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 85 test_pfm: -611
episode: 340 training return: tensor(9.6063e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 341 training return: tensor(9.1406e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 342 training return: tensor(1.1317e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 343 training return: tensor(1.0291e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 86 test_pfm: -586
episode: 344 training return: tensor(1.1547e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 345 training return: tensor(1.3686e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 346 training return: tensor(1.0697e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 347 training return: tensor(1.1840e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 87 test_pfm: -625
episode: 348 training return: tensor(1.1270e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 349 training return: tensor(1.0786e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 350 training return: tensor(1.0040e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 351 training return: tensor(1.1874e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 88 test_pfm: -617
episode: 352 training return: tensor(9.5472e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 353 training return: tensor(8.8780e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 354 training return: tensor(7.5397e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 355 training return: tensor(6.7700e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 89 test_pfm: -607
episode: 356 training return: tensor(8.1654e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 357 training return: tensor(9.6826e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 358 training return: tensor(9.2182e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 359 training return: tensor(8.1820e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 90 test_pfm: -628
episode: 360 training return: tensor(9.1402e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 361 training return: tensor(1.0484e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 362 training return: tensor(1.3588e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 363 training return: tensor(1.2181e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 91 test_pfm: -575
episode: 364 training return: tensor(1.1566e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 365 training return: tensor(8.4173e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 366 training return: tensor(9.3354e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 367 training return: tensor(8.1937e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 92 test_pfm: -647
episode: 368 training return: tensor(9.6846e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 369 training return: tensor(6.9006e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 370 training return: tensor(1.0632e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 371 training return: tensor(5.9092e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 93 test_pfm: -647
episode: 372 training return: tensor(1.6805e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 373 training return: tensor(7.7690e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 374 training return: tensor(9.5304e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 375 training return: tensor(3.3723e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 94 test_pfm: 294
episode: 376 training return: tensor(8.8774e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 377 training return: tensor(1.0686e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 378 training return: tensor(9.8496e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 379 training return: tensor(1.7218e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 95 test_pfm: 251
episode: 380 training return: tensor(1.1515e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 381 training return: tensor(7.7246e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 382 training return: tensor(8.8074e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 383 training return: tensor(8.5387e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 96 test_pfm: 28
episode: 384 training return: tensor(1.7727e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 385 training return: tensor(1.1250e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 386 training return: tensor(7.7330e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 387 training return: tensor(9.1662e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 97 test_pfm: 57
episode: 388 training return: tensor(8.1119e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 389 training return: tensor(1.6324e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 390 training return: tensor(1.4267e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 391 training return: tensor(7.7140e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 98 test_pfm: -265
episode: 392 training return: tensor(4.3059e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 393 training return: tensor(8.8308e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 394 training return: tensor(8.6991e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 395 training return: tensor(1.2183e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 99 test_pfm: -509
episode: 396 training return: tensor(2.5420e-14, device='cuda:0', grad_fn=<AddBackward0>)
episode: 397 training return: tensor(2.3050e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 398 training return: tensor(2.5629e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 399 training return: tensor(8.7740e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 100 test_pfm: -161
episode: 400 training return: tensor(1.3764e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 401 training return: tensor(8.5936e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 402 training return: tensor(2.8614e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 403 training return: tensor(3.2547e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 101 test_pfm: -465
episode: 404 training return: tensor(3.8793e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 405 training return: tensor(4.7637e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 406 training return: tensor(2.7707e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 407 training return: tensor(3.3801e-13, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 102 test_pfm: -527
episode: 408 training return: tensor(2.5215e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 409 training return: tensor(5.4531e-15, device='cuda:0', grad_fn=<AddBackward0>)
episode: 410 training return: tensor(1.5750e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 411 training return: tensor(2.4800e-12, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 103 test_pfm: -562
episode: 412 training return: tensor(4.7768e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 413 training return: tensor(6.9408e-14, device='cuda:0', grad_fn=<AddBackward0>)
episode: 414 training return: tensor(3.1098e-14, device='cuda:0', grad_fn=<AddBackward0>)
episode: 415 training return: tensor(5.5645e-14, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 104 test_pfm: -646
episode: 416 training return: tensor(1.0381e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 417 training return: tensor(3.0943e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 418 training return: tensor(6.3690e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 419 training return: tensor(1.7046e-12, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 105 test_pfm: -694
episode: 420 training return: tensor(8.7262e-17, device='cuda:0', grad_fn=<AddBackward0>)
episode: 421 training return: tensor(2.4993e-15, device='cuda:0', grad_fn=<AddBackward0>)
episode: 422 training return: tensor(3.3791e-17, device='cuda:0', grad_fn=<AddBackward0>)
episode: 423 training return: tensor(1.4725e-15, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 106 test_pfm: -734
episode: 424 training return: tensor(2.9362e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 425 training return: tensor(4.4342e-14, device='cuda:0', grad_fn=<AddBackward0>)
episode: 426 training return: tensor(8.2725e-14, device='cuda:0', grad_fn=<AddBackward0>)
episode: 427 training return: tensor(7.4750e-13, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 107 test_pfm: -597
episode: 428 training return: tensor(8.3294e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 429 training return: tensor(6.4291e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 430 training return: tensor(1.2435e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 431 training return: tensor(1.5938e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 108 test_pfm: -618
episode: 432 training return: tensor(2.7838e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 433 training return: tensor(1.5259e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 434 training return: tensor(2.3891e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 435 training return: tensor(5.1254e-12, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 109 test_pfm: -598
episode: 436 training return: tensor(2.1836e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 437 training return: tensor(9.0977e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 438 training return: tensor(1.0784e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 439 training return: tensor(2.9952e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 110 test_pfm: -656
episode: 440 training return: tensor(1.2321e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 441 training return: tensor(7.5106e-15, device='cuda:0', grad_fn=<AddBackward0>)
episode: 442 training return: tensor(3.7570e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 443 training return: tensor(2.6253e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 111 test_pfm: -759
episode: 444 training return: tensor(5.5164e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 445 training return: tensor(1.4900e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 446 training return: tensor(2.3065e-07, device='cuda:0', grad_fn=<AddBackward0>)
episode: 447 training return: tensor(1.1177e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 112 test_pfm: -919
episode: 448 training return: tensor(8.6718e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 449 training return: tensor(3.1600e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 450 training return: tensor(4.0590e-15, device='cuda:0', grad_fn=<AddBackward0>)
episode: 451 training return: tensor(5.1852e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 113 test_pfm: -697
episode: 452 training return: tensor(1.5343e-07, device='cuda:0', grad_fn=<AddBackward0>)
episode: 453 training return: tensor(1.2187e-06, device='cuda:0', grad_fn=<AddBackward0>)
episode: 454 training return: tensor(1.2019e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 455 training return: tensor(8.3172e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 114 test_pfm: -788
episode: 456 training return: tensor(4.5620e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 457 training return: tensor(2.4531e-15, device='cuda:0', grad_fn=<AddBackward0>)
episode: 458 training return: tensor(2.8884e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 459 training return: tensor(5.4695e-08, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 115 test_pfm: -832
episode: 460 training return: tensor(3.7688e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 461 training return: tensor(5.4300e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 462 training return: tensor(3.2069e-14, device='cuda:0', grad_fn=<AddBackward0>)
episode: 463 training return: tensor(5.9310e-12, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 116 test_pfm: -756
episode: 464 training return: tensor(1.2025e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 465 training return: tensor(1.1880e-16, device='cuda:0', grad_fn=<AddBackward0>)
episode: 466 training return: tensor(6.6268e-15, device='cuda:0', grad_fn=<AddBackward0>)
episode: 467 training return: tensor(2.2920e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 117 test_pfm: -522
episode: 468 training return: tensor(2.4614e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 469 training return: tensor(1.1971e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 470 training return: tensor(2.4574e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 471 training return: tensor(1.1122e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 118 test_pfm: -584
episode: 472 training return: tensor(2.0517e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 473 training return: tensor(2.6023e-14, device='cuda:0', grad_fn=<AddBackward0>)
episode: 474 training return: tensor(3.9622e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 475 training return: tensor(5.9638e-12, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 119 test_pfm: -583
episode: 476 training return: tensor(2.2071e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 477 training return: tensor(1.3718e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 478 training return: tensor(4.4309e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 479 training return: tensor(3.7195e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 120 test_pfm: -587
episode: 480 training return: tensor(1.8709e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 481 training return: tensor(2.5492e-07, device='cuda:0', grad_fn=<AddBackward0>)
episode: 482 training return: tensor(7.9254e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 483 training return: tensor(6.9418e-08, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 121 test_pfm: -558
episode: 484 training return: tensor(6.0008e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 485 training return: tensor(3.6704e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 486 training return: tensor(1.7589e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 487 training return: tensor(3.8176e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 122 test_pfm: -564
episode: 488 training return: tensor(1.0691e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 489 training return: tensor(9.1085e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 490 training return: tensor(3.6395e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 491 training return: tensor(5.0738e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 123 test_pfm: -576
episode: 492 training return: tensor(6.8668e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 493 training return: tensor(8.6012e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 494 training return: tensor(8.8703e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 495 training return: tensor(1.2781e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 124 test_pfm: -579
episode: 496 training return: tensor(2.9055e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 497 training return: tensor(3.2726e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 498 training return: tensor(3.9096e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 499 training return: tensor(4.4435e-12, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 125 test_pfm: -566
episode: 500 training return: tensor(6.4933e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 501 training return: tensor(3.9998e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 502 training return: tensor(9.8926e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 503 training return: tensor(2.3734e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 126 test_pfm: -539
episode: 504 training return: tensor(5.0741e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 505 training return: tensor(8.7573e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 506 training return: tensor(3.6614e-15, device='cuda:0', grad_fn=<AddBackward0>)
episode: 507 training return: tensor(2.0368e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 127 test_pfm: -564
episode: 508 training return: tensor(3.9582e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 509 training return: tensor(1.4665e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 510 training return: tensor(4.6925e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 511 training return: tensor(2.5959e-06, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 128 test_pfm: -591
episode: 512 training return: tensor(1.7509e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 513 training return: tensor(8.0968e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 514 training return: tensor(1.7404e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 515 training return: tensor(2.3935e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 129 test_pfm: -563
episode: 516 training return: tensor(3.0363e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 517 training return: tensor(1.4076e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 518 training return: tensor(5.0960e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 519 training return: tensor(4.7127e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 130 test_pfm: -560
episode: 520 training return: tensor(5.2640e-14, device='cuda:0', grad_fn=<AddBackward0>)
episode: 521 training return: tensor(3.4387e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 522 training return: tensor(4.3417e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 523 training return: tensor(1.5167e-12, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 131 test_pfm: -543
episode: 524 training return: tensor(6.7400e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 525 training return: tensor(5.0666e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 526 training return: tensor(3.8806e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 527 training return: tensor(2.5979e-08, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 132 test_pfm: -544
episode: 528 training return: tensor(9.7632e-14, device='cuda:0', grad_fn=<AddBackward0>)
episode: 529 training return: tensor(1.5900e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 530 training return: tensor(3.1857e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 531 training return: tensor(1.8543e-07, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 133 test_pfm: -552
episode: 532 training return: tensor(3.0089e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 533 training return: tensor(1.0157e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 534 training return: tensor(1.1946e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 535 training return: tensor(2.5336e-13, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 134 test_pfm: -567
episode: 536 training return: tensor(2.3369e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 537 training return: tensor(1.2218e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 538 training return: tensor(3.8512e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 539 training return: tensor(7.8078e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 135 test_pfm: -559
episode: 540 training return: tensor(9.3760e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 541 training return: tensor(1.2482e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 542 training return: tensor(2.6395e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 543 training return: tensor(6.0452e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 136 test_pfm: -586
episode: 544 training return: tensor(9.6995e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 545 training return: tensor(2.7008e-14, device='cuda:0', grad_fn=<AddBackward0>)
episode: 546 training return: tensor(4.7825e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 547 training return: tensor(2.7804e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 137 test_pfm: -584
episode: 548 training return: tensor(1.1205e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 549 training return: tensor(8.9376e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 550 training return: tensor(2.3776e-06, device='cuda:0', grad_fn=<AddBackward0>)
episode: 551 training return: tensor(1.0323e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 138 test_pfm: -560
episode: 552 training return: tensor(6.0140e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 553 training return: tensor(1.2210e-07, device='cuda:0', grad_fn=<AddBackward0>)
episode: 554 training return: tensor(8.0367e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 555 training return: tensor(1.2818e-07, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 139 test_pfm: -580
episode: 556 training return: tensor(1.6940e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 557 training return: tensor(1.7266e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 558 training return: tensor(1.6261e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 559 training return: tensor(3.8920e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 140 test_pfm: -552
episode: 560 training return: tensor(1.0704e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 561 training return: tensor(2.1006e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 562 training return: tensor(4.1864e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 563 training return: tensor(1.6365e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 141 test_pfm: -560
episode: 564 training return: tensor(1.4932e-06, device='cuda:0', grad_fn=<AddBackward0>)
episode: 565 training return: tensor(8.5186e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 566 training return: tensor(1.2901e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 567 training return: tensor(7.8950e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 142 test_pfm: -563
episode: 568 training return: tensor(1.4187e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 569 training return: tensor(1.4744e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 570 training return: tensor(2.2028e-07, device='cuda:0', grad_fn=<AddBackward0>)
episode: 571 training return: tensor(3.0666e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 143 test_pfm: -581
episode: 572 training return: tensor(1.2170e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 573 training return: tensor(1.2345e-07, device='cuda:0', grad_fn=<AddBackward0>)
episode: 574 training return: tensor(7.4104e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 575 training return: tensor(1.0567e-09, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 144 test_pfm: -579
episode: 576 training return: tensor(1.5527e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 577 training return: tensor(4.6819e-12, device='cuda:0', grad_fn=<AddBackward0>)
episode: 578 training return: tensor(1.0980e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 579 training return: tensor(5.3710e-14, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 145 test_pfm: -558
episode: 580 training return: tensor(5.8805e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 581 training return: tensor(3.7573e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 582 training return: tensor(1.2103e-05, device='cuda:0', grad_fn=<AddBackward0>)
episode: 583 training return: tensor(4.4988e-14, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 146 test_pfm: -581
episode: 584 training return: tensor(9.3754e-08, device='cuda:0', grad_fn=<AddBackward0>)
episode: 585 training return: tensor(1.7689e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 586 training return: tensor(2.1718e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 587 training return: tensor(3.1798e-11, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 147 test_pfm: -582
episode: 588 training return: tensor(1.1326e-07, device='cuda:0', grad_fn=<AddBackward0>)
episode: 589 training return: tensor(8.2446e-13, device='cuda:0', grad_fn=<AddBackward0>)
episode: 590 training return: tensor(1.4009e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 591 training return: tensor(2.2470e-12, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 148 test_pfm: -595
episode: 592 training return: tensor(2.9436e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 593 training return: tensor(9.9136e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 594 training return: tensor(3.2043e-09, device='cuda:0', grad_fn=<AddBackward0>)
episode: 595 training return: tensor(1.0179e-10, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 149 test_pfm: -575
episode: 596 training return: tensor(4.6077e-10, device='cuda:0', grad_fn=<AddBackward0>)
episode: 597 training return: tensor(5.6967e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 598 training return: tensor(2.8042e-11, device='cuda:0', grad_fn=<AddBackward0>)
episode: 599 training return: tensor(7.7185e-14, device='cuda:0', grad_fn=<AddBackward0>)
epoch: 150 test_pfm: -563
