checkpoints/shallow001_nas_8b_SAM/train_log.txt

[1] loss: 3.922, train acc: 9.710 test acc: 16.210  19.120 s
[2] loss: 3.306, train acc: 19.734 test acc: 24.700  19.272 s
[3] loss: 2.890, train acc: 27.350 test acc: 30.530  19.156 s
[4] loss: 2.572, train acc: 33.976 test acc: 34.630  18.599 s
[5] loss: 2.351, train acc: 38.444 test acc: 39.340  19.358 s
[6] loss: 2.182, train acc: 42.156 test acc: 41.620  19.322 s
[7] loss: 2.060, train acc: 44.980 test acc: 43.830  18.912 s
[8] loss: 1.946, train acc: 47.628 test acc: 45.620  19.201 s
[9] loss: 1.861, train acc: 49.356 test acc: 47.350  18.739 s
[10] loss: 1.778, train acc: 51.572 test acc: 47.440  19.149 s
[11] loss: 1.720, train acc: 52.758 test acc: 48.690  18.959 s
[12] loss: 1.665, train acc: 54.102 test acc: 50.070  18.651 s
[13] loss: 1.611, train acc: 55.504 test acc: 51.010  18.681 s
[14] loss: 1.565, train acc: 56.742 test acc: 51.310  18.636 s
[15] loss: 1.523, train acc: 57.590 test acc: 50.750  19.178 s
[16] loss: 1.493, train acc: 58.122 test acc: 52.760  18.726 s
[17] loss: 1.456, train acc: 59.148 test acc: 53.310  19.150 s
[18] loss: 1.425, train acc: 60.064 test acc: 53.020  18.625 s
[19] loss: 1.395, train acc: 60.686 test acc: 53.310  18.945 s
[20] loss: 1.366, train acc: 61.512 test acc: 54.200  20.388 s
[21] loss: 1.337, train acc: 62.098 test acc: 54.400  18.636 s
[22] loss: 1.317, train acc: 62.850 test acc: 54.450  18.698 s
[23] loss: 1.288, train acc: 63.556 test acc: 54.980  24.444 s
[24] loss: 1.270, train acc: 63.970 test acc: 54.640  19.223 s
[25] loss: 1.242, train acc: 64.418 test acc: 55.670  19.068 s
[26] loss: 1.228, train acc: 65.022 test acc: 55.390  18.723 s
[27] loss: 1.212, train acc: 65.308 test acc: 56.070  18.621 s
[28] loss: 1.192, train acc: 65.950 test acc: 55.740  18.721 s
[29] loss: 1.172, train acc: 66.610 test acc: 56.360  18.999 s
[30] loss: 1.162, train acc: 66.744 test acc: 56.040  19.265 s
[31] loss: 1.139, train acc: 67.142 test acc: 56.610  18.620 s
[32] loss: 1.127, train acc: 67.530 test acc: 56.350  18.952 s
[33] loss: 1.113, train acc: 67.938 test acc: 56.930  19.421 s
[34] loss: 1.103, train acc: 68.186 test acc: 56.610  19.007 s
[35] loss: 1.081, train acc: 68.868 test acc: 56.850  19.002 s
[36] loss: 1.077, train acc: 68.798 test acc: 57.090  18.931 s
[37] loss: 1.063, train acc: 69.366 test acc: 57.010  18.142 s
[38] loss: 1.048, train acc: 69.726 test acc: 57.600  18.577 s
[39] loss: 1.034, train acc: 70.048 test acc: 57.630  19.337 s
[40] loss: 1.021, train acc: 70.398 test acc: 58.170  18.606 s
[41] loss: 1.013, train acc: 70.720 test acc: 57.340  19.218 s
[42] loss: 1.001, train acc: 71.000 test acc: 58.030  18.656 s
[43] loss: 0.991, train acc: 71.130 test acc: 58.170  18.731 s
[44] loss: 0.982, train acc: 71.388 test acc: 58.150  18.939 s
[45] loss: 0.972, train acc: 71.786 test acc: 57.920  20.176 s
[46] loss: 0.959, train acc: 72.054 test acc: 58.770  19.481 s
[47] loss: 0.946, train acc: 72.474 test acc: 57.930  19.065 s
[48] loss: 0.935, train acc: 72.638 test acc: 57.890  19.334 s
[49] loss: 0.928, train acc: 72.724 test acc: 58.370  18.734 s
[50] loss: 0.925, train acc: 72.930 test acc: 58.690  18.609 s
[51] loss: 0.911, train acc: 73.478 test acc: 58.120  19.188 s
[52] loss: 0.906, train acc: 73.406 test acc: 57.950  18.921 s
[53] loss: 0.896, train acc: 73.732 test acc: 58.300  18.764 s
[54] loss: 0.891, train acc: 73.804 test acc: 58.070  18.855 s
[55] loss: 0.881, train acc: 74.204 test acc: 57.960  18.914 s
[56] loss: 0.873, train acc: 74.446 test acc: 58.690  18.841 s
[57] loss: 0.865, train acc: 74.332 test acc: 58.390  19.063 s
[58] loss: 0.856, train acc: 74.850 test acc: 58.630  19.052 s
[59] loss: 0.849, train acc: 75.136 test acc: 59.100  18.923 s
[60] loss: 0.851, train acc: 74.982 test acc: 58.100  18.426 s
[61] loss: 0.839, train acc: 75.072 test acc: 57.940  19.223 s
[62] loss: 0.828, train acc: 75.610 test acc: 58.210  19.462 s
[63] loss: 0.821, train acc: 75.916 test acc: 57.980  18.999 s
[64] loss: 0.816, train acc: 75.868 test acc: 59.340  18.477 s
[65] loss: 0.806, train acc: 76.154 test acc: 58.640  19.336 s
[66] loss: 0.802, train acc: 76.380 test acc: 59.180  19.209 s
[67] loss: 0.794, train acc: 76.694 test acc: 59.110  18.478 s
[68] loss: 0.792, train acc: 76.544 test acc: 59.230  18.842 s
[69] loss: 0.781, train acc: 77.010 test acc: 58.640  18.791 s
[70] loss: 0.777, train acc: 77.002 test acc: 59.170  19.276 s
[71] loss: 0.773, train acc: 77.146 test acc: 59.250  19.578 s
[72] loss: 0.767, train acc: 77.232 test acc: 59.000  19.281 s
[73] loss: 0.760, train acc: 77.390 test acc: 59.020  18.526 s
[74] loss: 0.762, train acc: 77.430 test acc: 58.650  18.691 s
[75] loss: 0.755, train acc: 77.836 test acc: 59.310  20.628 s
[76] loss: 0.750, train acc: 77.732 test acc: 59.170  18.904 s
[77] loss: 0.745, train acc: 77.560 test acc: 58.820  19.015 s
[78] loss: 0.738, train acc: 78.148 test acc: 58.990  19.101 s
[79] loss: 0.729, train acc: 78.210 test acc: 58.660  18.940 s
[80] loss: 0.728, train acc: 78.240 test acc: 58.870  18.424 s
[81] loss: 0.723, train acc: 78.442 test acc: 58.510  19.399 s
[82] loss: 0.718, train acc: 78.706 test acc: 58.610  18.937 s
[83] loss: 0.712, train acc: 78.724 test acc: 58.560  19.048 s
[84] loss: 0.705, train acc: 78.776 test acc: 58.810  18.905 s
[85] loss: 0.704, train acc: 78.982 test acc: 58.250  19.172 s
[86] loss: 0.698, train acc: 79.308 test acc: 58.380  19.347 s
[87] loss: 0.693, train acc: 79.318 test acc: 58.450  19.214 s
[88] loss: 0.686, train acc: 79.432 test acc: 59.050  19.092 s
[89] loss: 0.683, train acc: 79.574 test acc: 59.140  18.626 s
[90] loss: 0.679, train acc: 79.708 test acc: 58.440  19.234 s
[91] loss: 0.672, train acc: 79.968 test acc: 58.560  18.429 s
[92] loss: 0.669, train acc: 80.088 test acc: 58.820  18.924 s
[93] loss: 0.660, train acc: 80.174 test acc: 58.480  18.966 s
[94] loss: 0.664, train acc: 80.024 test acc: 58.970  18.989 s
[95] loss: 0.656, train acc: 80.338 test acc: 59.070  18.756 s
[96] loss: 0.654, train acc: 80.278 test acc: 59.270  19.369 s
[97] loss: 0.648, train acc: 80.548 test acc: 59.050  19.416 s
[98] loss: 0.641, train acc: 80.714 test acc: 59.120  18.987 s
[99] loss: 0.646, train acc: 80.624 test acc: 58.520  18.932 s
[100] loss: 0.638, train acc: 80.954 test acc: 59.050  19.094 s
[1] loss: 0.580, train acc: 82.956 test acc: 60.010  18.612 s
[2] loss: 0.557, train acc: 83.868 test acc: 59.950  18.785 s
[3] loss: 0.552, train acc: 83.906 test acc: 60.080  19.294 s
[4] loss: 0.546, train acc: 84.102 test acc: 60.190  19.067 s
[5] loss: 0.539, train acc: 84.412 test acc: 59.960  18.777 s
[6] loss: 0.539, train acc: 84.556 test acc: 60.070  18.761 s
[7] loss: 0.536, train acc: 84.534 test acc: 60.050  18.752 s
[8] loss: 0.530, train acc: 84.778 test acc: 59.820  18.836 s
[9] loss: 0.533, train acc: 84.568 test acc: 60.220  19.284 s
[10] loss: 0.528, train acc: 84.792 test acc: 59.970  18.962 s
[11] loss: 0.528, train acc: 84.710 test acc: 60.090  18.949 s
[12] loss: 0.527, train acc: 84.716 test acc: 60.050  18.657 s
[13] loss: 0.525, train acc: 84.716 test acc: 60.180  18.807 s
[14] loss: 0.521, train acc: 84.866 test acc: 59.980  18.586 s
[15] loss: 0.522, train acc: 84.864 test acc: 60.010  19.012 s
[16] loss: 0.517, train acc: 85.004 test acc: 59.850  19.005 s
[17] loss: 0.520, train acc: 84.860 test acc: 60.080  19.120 s
[18] loss: 0.511, train acc: 85.258 test acc: 60.210  18.975 s
[19] loss: 0.513, train acc: 85.128 test acc: 60.210  19.032 s
[20] loss: 0.507, train acc: 85.348 test acc: 59.940  18.446 s
[1] loss: 0.501, train acc: 85.592 test acc: 60.100  18.988 s
[2] loss: 0.490, train acc: 86.018 test acc: 60.070  18.917 s
[3] loss: 0.488, train acc: 85.992 test acc: 59.990  18.860 s
[4] loss: 0.493, train acc: 86.016 test acc: 59.870  18.987 s
[5] loss: 0.485, train acc: 86.248 test acc: 60.040  18.584 s
[6] loss: 0.487, train acc: 86.264 test acc: 60.130  18.601 s
[7] loss: 0.486, train acc: 86.110 test acc: 60.160  18.754 s
[8] loss: 0.486, train acc: 86.056 test acc: 60.070  18.997 s
[9] loss: 0.485, train acc: 86.114 test acc: 60.190  18.654 s
[10] loss: 0.484, train acc: 86.144 test acc: 60.130  18.356 s
[11] loss: 0.482, train acc: 86.410 test acc: 59.970  18.743 s
[12] loss: 0.484, train acc: 86.180 test acc: 60.030  19.216 s
[13] loss: 0.482, train acc: 86.230 test acc: 60.250  20.355 s
[14] loss: 0.483, train acc: 86.010 test acc: 60.300  19.104 s
[15] loss: 0.482, train acc: 86.146 test acc: 59.910  18.860 s
[16] loss: 0.484, train acc: 86.202 test acc: 60.070  18.826 s
[17] loss: 0.480, train acc: 86.304 test acc: 60.060  18.555 s
[18] loss: 0.482, train acc: 86.260 test acc: 60.280  19.010 s
[19] loss: 0.481, train acc: 86.156 test acc: 60.300  18.804 s
[20] loss: 0.479, train acc: 86.360 test acc: 60.310  18.998 s
[1] loss: 0.479, train acc: 86.142 test acc: 60.280  18.646 s
[2] loss: 0.476, train acc: 86.300 test acc: 60.320  18.658 s
[3] loss: 0.475, train acc: 86.410 test acc: 60.240  19.096 s
[4] loss: 0.475, train acc: 86.532 test acc: 60.260  18.890 s
[5] loss: 0.476, train acc: 86.228 test acc: 60.250  19.536 s
[6] loss: 0.473, train acc: 86.540 test acc: 60.290  18.323 s
[7] loss: 0.476, train acc: 86.352 test acc: 60.230  19.586 s
[8] loss: 0.473, train acc: 86.520 test acc: 60.230  19.256 s
[9] loss: 0.472, train acc: 86.624 test acc: 60.310  18.598 s
[10] loss: 0.475, train acc: 86.556 test acc: 60.350  18.936 s
[11] loss: 0.475, train acc: 86.476 test acc: 60.380  18.681 s
[12] loss: 0.471, train acc: 86.486 test acc: 60.340  20.621 s
[13] loss: 0.474, train acc: 86.558 test acc: 60.310  18.922 s
[14] loss: 0.470, train acc: 86.620 test acc: 60.290  19.109 s
[15] loss: 0.473, train acc: 86.634 test acc: 60.170  19.187 s
[16] loss: 0.474, train acc: 86.436 test acc: 60.270  18.899 s
[17] loss: 0.471, train acc: 86.656 test acc: 60.280  19.279 s
[18] loss: 0.474, train acc: 86.480 test acc: 60.150  19.134 s
[19] loss: 0.471, train acc: 86.580 test acc: 60.200  18.532 s
[20] loss: 0.473, train acc: 86.662 test acc: 60.170  18.995 s
[1] loss: 1.106, train acc: 76.134 test acc: 62.780  38.125 s
[2] loss: 0.874, train acc: 80.666 test acc: 63.290  39.722 s
[3] loss: 0.838, train acc: 80.908 test acc: 63.320  38.934 s
[4] loss: 0.819, train acc: 81.398 test acc: 63.560  38.463 s
[5] loss: 0.810, train acc: 81.292 test acc: 63.210  38.697 s
[6] loss: 0.803, train acc: 81.268 test acc: 63.530  38.476 s
[7] loss: 0.793, train acc: 81.176 test acc: 63.700  38.083 s
[8] loss: 0.790, train acc: 81.434 test acc: 63.320  38.817 s
[9] loss: 0.787, train acc: 81.242 test acc: 63.570  38.433 s
[10] loss: 0.782, train acc: 81.380 test acc: 63.710  38.234 s
[11] loss: 0.778, train acc: 81.572 test acc: 63.640  39.205 s
[12] loss: 0.773, train acc: 81.422 test acc: 63.700  38.101 s
[13] loss: 0.767, train acc: 81.550 test acc: 63.580  38.276 s
[14] loss: 0.762, train acc: 81.648 test acc: 63.680  38.218 s
[15] loss: 0.766, train acc: 81.220 test acc: 63.710  38.191 s
[16] loss: 0.759, train acc: 81.704 test acc: 63.640  37.920 s
[17] loss: 0.756, train acc: 81.480 test acc: 63.790  38.715 s
[18] loss: 0.758, train acc: 81.528 test acc: 63.760  38.157 s
[19] loss: 0.756, train acc: 81.654 test acc: 63.840  38.704 s
[20] loss: 0.756, train acc: 81.532 test acc: 63.800  38.097 s
[21] loss: 0.752, train acc: 81.542 test acc: 63.900  38.504 s
[22] loss: 0.746, train acc: 81.598 test acc: 63.830  38.281 s
[23] loss: 0.747, train acc: 81.616 test acc: 63.760  38.159 s

restarting with half the learning rate, zero optimizer state

[1] loss: 0.742, train acc: 81.706 test acc: 63.920  36.892 s
[2] loss: 0.743, train acc: 81.778 test acc: 63.970  36.748 s
[3] loss: 0.739, train acc: 81.960 test acc: 63.890  36.376 s
[4] loss: 0.737, train acc: 81.954 test acc: 63.770  35.944 s
[5] loss: 0.735, train acc: 81.996 test acc: 64.210  36.866 s
[6] loss: 0.734, train acc: 82.072 test acc: 63.930  36.578 s
[7] loss: 0.734, train acc: 81.916 test acc: 63.930  37.215 s
[8] loss: 0.729, train acc: 81.992 test acc: 63.880  36.817 s
[9] loss: 0.732, train acc: 82.108 test acc: 64.080  36.487 s
[10] loss: 0.728, train acc: 82.142 test acc: 64.070  36.806 s
[11] loss: 0.733, train acc: 81.934 test acc: 63.990  36.853 s
[1] loss: 0.781, train acc: 81.422 test acc: 63.790  37.518 s
[2] loss: 0.821, train acc: 80.904 test acc: 63.350  37.203 s
[3] loss: 0.841, train acc: 80.668 test acc: 63.400  37.730 s
[4] loss: 0.856, train acc: 80.196 test acc: 63.190  37.715 s
[5] loss: 0.866, train acc: 80.016 test acc: 63.070  37.500 s
[6] loss: 0.874, train acc: 79.680 test acc: 63.050  38.076 s
[7] loss: 0.881, train acc: 79.606 test acc: 63.030  37.768 s
[8] loss: 0.882, train acc: 79.624 test acc: 62.860  38.120 s
[9] loss: 0.884, train acc: 79.590 test acc: 62.980  37.331 s
[1] loss: 0.737, train acc: 81.764 test acc: 63.780  39.241 s
[2] loss: 0.706, train acc: 81.852 test acc: 64.160  38.618 s
[3] loss: 0.691, train acc: 82.076 test acc: 64.070  39.309 s
[4] loss: 0.686, train acc: 82.174 test acc: 64.260  38.344 s
[5] loss: 0.674, train acc: 82.528 test acc: 64.100  38.361 s
[6] loss: 0.673, train acc: 82.422 test acc: 64.480  38.350 s
[7] loss: 0.667, train acc: 82.700 test acc: 64.370  38.942 s
[8] loss: 0.665, train acc: 82.792 test acc: 64.400  38.189 s
[9] loss: 0.662, train acc: 82.726 test acc: 64.440  38.667 s
[10] loss: 0.660, train acc: 82.766 test acc: 64.370  39.073 s
[11] loss: 0.660, train acc: 82.808 test acc: 64.400  38.822 s
[12] loss: 0.653, train acc: 83.032 test acc: 64.430  38.702 s
[1] loss: 0.678, train acc: 82.660 test acc: 64.240  37.287 s
[2] loss: 0.688, train acc: 82.854 test acc: 64.310  37.077 s
[3] loss: 0.694, train acc: 82.710 test acc: 64.230  36.969 s
[4] loss: 0.701, train acc: 82.636 test acc: 64.210  36.958 s
[5] loss: 0.702, train acc: 82.640 test acc: 64.300  36.997 s
[6] loss: 0.704, train acc: 82.408 test acc: 64.180  37.049 s
[7] loss: 0.703, train acc: 82.806 test acc: 64.160  37.687 s
[8] loss: 0.710, train acc: 82.334 test acc: 63.980  37.277 s
[9] loss: 0.709, train acc: 82.544 test acc: 64.290  37.380 s
[10] loss: 0.706, train acc: 82.538 test acc: 64.070  37.523 s
[11] loss: 0.712, train acc: 82.400 test acc: 64.020  37.281 s
[12] loss: 0.708, train acc: 82.548 test acc: 63.950  36.890 s
[13] loss: 0.710, train acc: 82.606 test acc: 64.150  36.889 s
[14] loss: 0.709, train acc: 82.514 test acc: 64.210  38.943 s
[15] loss: 0.710, train acc: 82.704 test acc: 64.310  37.126 s
[16] loss: 0.710, train acc: 82.650 test acc: 64.090  36.937 s
[17] loss: 0.712, train acc: 82.526 test acc: 64.180  37.442 s
[18] loss: 0.710, train acc: 82.840 test acc: 64.070  37.089 s
[19] loss: 0.711, train acc: 82.582 test acc: 64.220  37.877 s
[20] loss: 0.710, train acc: 82.668 test acc: 64.150  37.814 s
[21] loss: 0.709, train acc: 82.544 test acc: 64.150  37.165 s