| {"step": 20, "loss": 0.076785, "loss_mean50": 0.08647, "lr": 5e-07, "grad_norm": 0.1727, "samples": 640, "wall_s": 44.6, "cuda_gb": 16.86} |
| {"step": 40, "loss": 0.094949, "loss_mean50": 0.086559, "lr": 1e-06, "grad_norm": 0.2535, "samples": 1280, "wall_s": 83.6, "cuda_gb": 16.86} |
| {"step": 60, "loss": 0.059598, "loss_mean50": 0.083484, "lr": 1.5e-06, "grad_norm": 0.242, "samples": 1920, "wall_s": 122.7, "cuda_gb": 16.86} |
| {"step": 80, "loss": 0.072547, "loss_mean50": 0.079185, "lr": 2e-06, "grad_norm": 0.2276, "samples": 2560, "wall_s": 161.8, "cuda_gb": 16.86} |
| {"step": 100, "loss": 0.046494, "loss_mean50": 0.067039, "lr": 2.5e-06, "grad_norm": 0.1453, "samples": 3200, "wall_s": 200.7, "cuda_gb": 16.86} |
| {"step": 120, "loss": 0.020664, "loss_mean50": 0.047178, "lr": 3e-06, "grad_norm": 0.0869, "samples": 3840, "wall_s": 239.6, "cuda_gb": 16.86} |
| {"step": 140, "loss": 0.012562, "loss_mean50": 0.031318, "lr": 3.5e-06, "grad_norm": 0.0652, "samples": 4480, "wall_s": 278.5, "cuda_gb": 16.86} |
| {"step": 160, "loss": 0.017317, "loss_mean50": 0.020424, "lr": 4e-06, "grad_norm": 0.0794, "samples": 5120, "wall_s": 317.4, "cuda_gb": 16.86} |
| {"step": 180, "loss": 0.016111, "loss_mean50": 0.017232, "lr": 4.5e-06, "grad_norm": 0.0625, "samples": 5760, "wall_s": 356.3, "cuda_gb": 16.86} |
| {"step": 200, "loss": 0.01164, "loss_mean50": 0.01591, "lr": 5e-06, "grad_norm": 0.0753, "samples": 6400, "wall_s": 395.2, "cuda_gb": 16.86} |
| {"step": 220, "loss": 0.009393, "loss_mean50": 0.014188, "lr": 5.500000000000001e-06, "grad_norm": 0.0544, "samples": 7040, "wall_s": 434.1, "cuda_gb": 16.86} |
| {"step": 240, "loss": 0.009482, "loss_mean50": 0.012267, "lr": 6e-06, "grad_norm": 0.0666, "samples": 7680, "wall_s": 473.0, "cuda_gb": 16.86} |
| {"step": 260, "loss": 0.008879, "loss_mean50": 0.011488, "lr": 6.5000000000000004e-06, "grad_norm": 0.1056, "samples": 8320, "wall_s": 511.9, "cuda_gb": 16.86} |
| {"step": 280, "loss": 0.012476, "loss_mean50": 0.010738, "lr": 7e-06, "grad_norm": 0.1048, "samples": 8960, "wall_s": 550.8, "cuda_gb": 16.86} |
| {"step": 300, "loss": 0.007725, "loss_mean50": 0.009592, "lr": 7.5e-06, "grad_norm": 0.1095, "samples": 9600, "wall_s": 589.7, "cuda_gb": 16.86} |
| {"step": 320, "loss": 0.009558, "loss_mean50": 0.009292, "lr": 8e-06, "grad_norm": 0.0548, "samples": 10240, "wall_s": 628.7, "cuda_gb": 16.86} |
| {"step": 340, "loss": 0.010042, "loss_mean50": 0.009207, "lr": 8.5e-06, "grad_norm": 0.0893, "samples": 10880, "wall_s": 667.6, "cuda_gb": 16.86} |
| {"step": 360, "loss": 0.010355, "loss_mean50": 0.008665, "lr": 9e-06, "grad_norm": 0.051, "samples": 11520, "wall_s": 706.6, "cuda_gb": 16.86} |
| {"step": 380, "loss": 0.007398, "loss_mean50": 0.008359, "lr": 9.5e-06, "grad_norm": 0.0565, "samples": 12160, "wall_s": 745.6, "cuda_gb": 16.86} |
| {"step": 400, "loss": 0.006706, "loss_mean50": 0.007918, "lr": 1e-05, "grad_norm": 0.0644, "samples": 12800, "wall_s": 784.4, "cuda_gb": 16.86} |
| {"step": 420, "loss": 0.009885, "loss_mean50": 0.00787, "lr": 1.0500000000000001e-05, "grad_norm": 0.1199, "samples": 13440, "wall_s": 823.4, "cuda_gb": 16.86} |
| {"step": 440, "loss": 0.006058, "loss_mean50": 0.007932, "lr": 1.1000000000000001e-05, "grad_norm": 0.0644, "samples": 14080, "wall_s": 862.3, "cuda_gb": 16.86} |
| {"step": 460, "loss": 0.006025, "loss_mean50": 0.007699, "lr": 1.15e-05, "grad_norm": 0.0647, "samples": 14720, "wall_s": 901.1, "cuda_gb": 16.86} |
| {"step": 480, "loss": 0.008659, "loss_mean50": 0.007399, "lr": 1.2e-05, "grad_norm": 0.0966, "samples": 15360, "wall_s": 939.7, "cuda_gb": 16.86} |
| {"step": 500, "loss": 0.00731, "loss_mean50": 0.00753, "lr": 1.25e-05, "grad_norm": 0.0458, "samples": 16000, "wall_s": 978.5, "cuda_gb": 16.86} |
| {"step": 520, "loss": 0.006649, "loss_mean50": 0.007057, "lr": 1.3000000000000001e-05, "grad_norm": 0.0601, "samples": 16640, "wall_s": 1017.1, "cuda_gb": 16.86} |
| {"step": 540, "loss": 0.007518, "loss_mean50": 0.006591, "lr": 1.35e-05, "grad_norm": 0.0495, "samples": 17280, "wall_s": 1056.0, "cuda_gb": 16.86} |
| {"step": 560, "loss": 0.005119, "loss_mean50": 0.006603, "lr": 1.4e-05, "grad_norm": 0.0499, "samples": 17920, "wall_s": 1094.7, "cuda_gb": 16.86} |
| {"step": 580, "loss": 0.003765, "loss_mean50": 0.006763, "lr": 1.45e-05, "grad_norm": 0.0325, "samples": 18560, "wall_s": 1133.6, "cuda_gb": 16.86} |
| {"step": 600, "loss": 0.006886, "loss_mean50": 0.006957, "lr": 1.5e-05, "grad_norm": 0.1021, "samples": 19200, "wall_s": 1172.6, "cuda_gb": 16.86} |
| {"step": 620, "loss": 0.007789, "loss_mean50": 0.006666, "lr": 1.55e-05, "grad_norm": 0.0526, "samples": 19840, "wall_s": 1211.6, "cuda_gb": 16.86} |
| {"step": 640, "loss": 0.007178, "loss_mean50": 0.006754, "lr": 1.6e-05, "grad_norm": 0.0599, "samples": 20480, "wall_s": 1250.6, "cuda_gb": 16.86} |
| {"step": 660, "loss": 0.008412, "loss_mean50": 0.006598, "lr": 1.65e-05, "grad_norm": 0.0652, "samples": 21120, "wall_s": 1289.5, "cuda_gb": 16.86} |
| {"step": 680, "loss": 0.005446, "loss_mean50": 0.006546, "lr": 1.7e-05, "grad_norm": 0.0442, "samples": 21760, "wall_s": 1328.3, "cuda_gb": 16.86} |
| {"step": 700, "loss": 0.006141, "loss_mean50": 0.005877, "lr": 1.7500000000000002e-05, "grad_norm": 0.0412, "samples": 22400, "wall_s": 1367.1, "cuda_gb": 16.86} |
| {"step": 720, "loss": 0.003133, "loss_mean50": 0.005897, "lr": 1.8e-05, "grad_norm": 0.0329, "samples": 23040, "wall_s": 1405.9, "cuda_gb": 16.86} |
| {"step": 740, "loss": 0.009962, "loss_mean50": 0.006007, "lr": 1.8500000000000002e-05, "grad_norm": 0.0925, "samples": 23680, "wall_s": 1444.7, "cuda_gb": 16.86} |
| {"step": 760, "loss": 0.009819, "loss_mean50": 0.006079, "lr": 1.9e-05, "grad_norm": 0.0789, "samples": 24320, "wall_s": 1483.4, "cuda_gb": 16.86} |
| {"step": 780, "loss": 0.00779, "loss_mean50": 0.005991, "lr": 1.95e-05, "grad_norm": 0.0457, "samples": 24960, "wall_s": 1522.2, "cuda_gb": 16.86} |
| {"step": 800, "loss": 0.005397, "loss_mean50": 0.006016, "lr": 2e-05, "grad_norm": 0.0415, "samples": 25600, "wall_s": 1561.0, "cuda_gb": 16.86} |
| {"step": 820, "loss": 0.007909, "loss_mean50": 0.005882, "lr": 2.05e-05, "grad_norm": 0.071, "samples": 26240, "wall_s": 1599.9, "cuda_gb": 16.86} |
| {"step": 840, "loss": 0.005987, "loss_mean50": 0.006018, "lr": 2.1000000000000002e-05, "grad_norm": 0.0648, "samples": 26880, "wall_s": 1638.8, "cuda_gb": 16.86} |
| {"step": 860, "loss": 0.012687, "loss_mean50": 0.006552, "lr": 2.15e-05, "grad_norm": 0.0986, "samples": 27520, "wall_s": 1677.7, "cuda_gb": 16.86} |
| {"step": 880, "loss": 0.004422, "loss_mean50": 0.006321, "lr": 2.2000000000000003e-05, "grad_norm": 0.0398, "samples": 28160, "wall_s": 1716.6, "cuda_gb": 16.86} |
| {"step": 900, "loss": 0.004478, "loss_mean50": 0.00603, "lr": 2.25e-05, "grad_norm": 0.0473, "samples": 28800, "wall_s": 1755.5, "cuda_gb": 16.86} |
| {"step": 920, "loss": 0.006735, "loss_mean50": 0.00553, "lr": 2.3e-05, "grad_norm": 0.0496, "samples": 29440, "wall_s": 1794.3, "cuda_gb": 16.86} |
| {"step": 940, "loss": 0.004186, "loss_mean50": 0.005677, "lr": 2.35e-05, "grad_norm": 0.05, "samples": 30080, "wall_s": 1833.1, "cuda_gb": 16.86} |
| {"step": 960, "loss": 0.003931, "loss_mean50": 0.005436, "lr": 2.4e-05, "grad_norm": 0.0414, "samples": 30720, "wall_s": 1871.8, "cuda_gb": 16.86} |
| {"step": 980, "loss": 0.006049, "loss_mean50": 0.005149, "lr": 2.4500000000000003e-05, "grad_norm": 0.0933, "samples": 31360, "wall_s": 1910.5, "cuda_gb": 16.86} |
| {"step": 1000, "loss": 0.006489, "loss_mean50": 0.005741, "lr": 2.5e-05, "grad_norm": 0.0675, "samples": 32000, "wall_s": 1949.3, "cuda_gb": 16.86} |
| {"step": 1000, "val_flow_mse": 0.005433} |
| {"step": 1020, "loss": 0.003515, "loss_mean50": 0.00623, "lr": 2.499989774778022e-05, "grad_norm": 0.0391, "samples": 32640, "wall_s": 2017.0, "cuda_gb": 16.86} |
| {"step": 1040, "loss": 0.004742, "loss_mean50": 0.006372, "lr": 2.4999569183187692e-05, "grad_norm": 0.0596, "samples": 33280, "wall_s": 2055.9, "cuda_gb": 16.86} |
| {"step": 1060, "loss": 0.003542, "loss_mean50": 0.005793, "lr": 2.4999014029592804e-05, "grad_norm": 0.0363, "samples": 33920, "wall_s": 2094.8, "cuda_gb": 16.86} |
| {"step": 1080, "loss": 0.00501, "loss_mean50": 0.00526, "lr": 2.4998232298177463e-05, "grad_norm": 0.0298, "samples": 34560, "wall_s": 2133.7, "cuda_gb": 16.86} |
| {"step": 1100, "loss": 0.005088, "loss_mean50": 0.004945, "lr": 2.499722400468732e-05, "grad_norm": 0.0506, "samples": 35200, "wall_s": 2172.6, "cuda_gb": 16.86} |
| {"step": 1120, "loss": 0.007739, "loss_mean50": 0.00505, "lr": 2.4995989169431427e-05, "grad_norm": 0.0896, "samples": 35840, "wall_s": 2211.4, "cuda_gb": 16.86} |
| {"step": 1140, "loss": 0.003745, "loss_mean50": 0.005639, "lr": 2.499452781728187e-05, "grad_norm": 0.0328, "samples": 36480, "wall_s": 2250.2, "cuda_gb": 16.86} |
| {"step": 1160, "loss": 0.008058, "loss_mean50": 0.00568, "lr": 2.4992839977673223e-05, "grad_norm": 0.0446, "samples": 37120, "wall_s": 2289.1, "cuda_gb": 16.86} |
| {"step": 1180, "loss": 0.006507, "loss_mean50": 0.005769, "lr": 2.4990925684601972e-05, "grad_norm": 0.0572, "samples": 37760, "wall_s": 2327.9, "cuda_gb": 16.86} |
| {"step": 1200, "loss": 0.005752, "loss_mean50": 0.005539, "lr": 2.4988784976625844e-05, "grad_norm": 0.0428, "samples": 38400, "wall_s": 2366.8, "cuda_gb": 16.86} |
| {"step": 1220, "loss": 0.008113, "loss_mean50": 0.005165, "lr": 2.4986417896863013e-05, "grad_norm": 0.0649, "samples": 39040, "wall_s": 2405.6, "cuda_gb": 16.86} |
| {"step": 1240, "loss": 0.003727, "loss_mean50": 0.005097, "lr": 2.4983824492991236e-05, "grad_norm": 0.0547, "samples": 39680, "wall_s": 2444.5, "cuda_gb": 16.86} |
| {"step": 1260, "loss": 0.008039, "loss_mean50": 0.005016, "lr": 2.4981004817246896e-05, "grad_norm": 0.0702, "samples": 40320, "wall_s": 2483.3, "cuda_gb": 16.86} |
| {"step": 1280, "loss": 0.003151, "loss_mean50": 0.005131, "lr": 2.497795892642395e-05, "grad_norm": 0.0414, "samples": 40960, "wall_s": 2522.2, "cuda_gb": 16.86} |
| {"step": 1300, "loss": 0.006934, "loss_mean50": 0.005215, "lr": 2.4974686881872776e-05, "grad_norm": 0.0557, "samples": 41600, "wall_s": 2561.1, "cuda_gb": 16.86} |
| {"step": 1320, "loss": 0.004084, "loss_mean50": 0.004847, "lr": 2.4971188749498953e-05, "grad_norm": 0.0314, "samples": 42240, "wall_s": 2599.9, "cuda_gb": 16.86} |
| {"step": 1340, "loss": 0.005012, "loss_mean50": 0.00488, "lr": 2.4967464599761908e-05, "grad_norm": 0.05, "samples": 42880, "wall_s": 2638.8, "cuda_gb": 16.86} |
| {"step": 1360, "loss": 0.002359, "loss_mean50": 0.00489, "lr": 2.4963514507673534e-05, "grad_norm": 0.0359, "samples": 43520, "wall_s": 2677.5, "cuda_gb": 16.86} |
| {"step": 1380, "loss": 0.004094, "loss_mean50": 0.00477, "lr": 2.4959338552796643e-05, "grad_norm": 0.0403, "samples": 44160, "wall_s": 2716.3, "cuda_gb": 16.86} |
| {"step": 1400, "loss": 0.002653, "loss_mean50": 0.004649, "lr": 2.495493681924339e-05, "grad_norm": 0.0374, "samples": 44800, "wall_s": 2755.1, "cuda_gb": 16.86} |
| {"step": 1420, "loss": 0.003311, "loss_mean50": 0.004941, "lr": 2.4950309395673564e-05, "grad_norm": 0.0357, "samples": 45440, "wall_s": 2793.9, "cuda_gb": 16.86} |
| {"step": 1440, "loss": 0.003138, "loss_mean50": 0.004834, "lr": 2.4945456375292797e-05, "grad_norm": 0.0364, "samples": 46080, "wall_s": 2832.8, "cuda_gb": 16.86} |
| {"step": 1460, "loss": 0.004418, "loss_mean50": 0.00462, "lr": 2.494037785585071e-05, "grad_norm": 0.0819, "samples": 46720, "wall_s": 2871.8, "cuda_gb": 16.86} |
| {"step": 1480, "loss": 0.006247, "loss_mean50": 0.004648, "lr": 2.493507393963892e-05, "grad_norm": 0.0496, "samples": 47360, "wall_s": 2910.7, "cuda_gb": 16.86} |
| {"step": 1500, "loss": 0.00509, "loss_mean50": 0.004747, "lr": 2.492954473348899e-05, "grad_norm": 0.0549, "samples": 48000, "wall_s": 2949.5, "cuda_gb": 16.86} |
| {"step": 1520, "loss": 0.001968, "loss_mean50": 0.004983, "lr": 2.4923790348770284e-05, "grad_norm": 0.0271, "samples": 48640, "wall_s": 2988.3, "cuda_gb": 16.86} |
| {"step": 1540, "loss": 0.004922, "loss_mean50": 0.004941, "lr": 2.4917810901387707e-05, "grad_norm": 0.0292, "samples": 49280, "wall_s": 3027.1, "cuda_gb": 16.86} |
| {"step": 1560, "loss": 0.002284, "loss_mean50": 0.004629, "lr": 2.4911606511779384e-05, "grad_norm": 0.0302, "samples": 49920, "wall_s": 3065.9, "cuda_gb": 16.86} |
| {"step": 1580, "loss": 0.006077, "loss_mean50": 0.004647, "lr": 2.4905177304914233e-05, "grad_norm": 0.047, "samples": 50560, "wall_s": 3104.7, "cuda_gb": 16.86} |
| {"step": 1600, "loss": 0.004436, "loss_mean50": 0.004698, "lr": 2.4898523410289436e-05, "grad_norm": 0.0315, "samples": 51200, "wall_s": 3143.5, "cuda_gb": 16.86} |
| {"step": 1620, "loss": 0.003109, "loss_mean50": 0.004639, "lr": 2.4891644961927848e-05, "grad_norm": 0.0406, "samples": 51840, "wall_s": 3182.3, "cuda_gb": 16.86} |
| {"step": 1640, "loss": 0.004242, "loss_mean50": 0.004517, "lr": 2.488454209837529e-05, "grad_norm": 0.0359, "samples": 52480, "wall_s": 3221.1, "cuda_gb": 16.86} |
| {"step": 1660, "loss": 0.003916, "loss_mean50": 0.004414, "lr": 2.4877214962697744e-05, "grad_norm": 0.0463, "samples": 53120, "wall_s": 3259.9, "cuda_gb": 16.86} |
| {"step": 1680, "loss": 0.006083, "loss_mean50": 0.004551, "lr": 2.4869663702478503e-05, "grad_norm": 0.0403, "samples": 53760, "wall_s": 3298.7, "cuda_gb": 16.86} |
| {"step": 1700, "loss": 0.006253, "loss_mean50": 0.004637, "lr": 2.4861888469815163e-05, "grad_norm": 0.0505, "samples": 54400, "wall_s": 3337.5, "cuda_gb": 16.86} |
| {"step": 1720, "loss": 0.004217, "loss_mean50": 0.00466, "lr": 2.4853889421316596e-05, "grad_norm": 0.0389, "samples": 55040, "wall_s": 3376.4, "cuda_gb": 16.86} |
| {"step": 1740, "loss": 0.009574, "loss_mean50": 0.004765, "lr": 2.484566671809975e-05, "grad_norm": 0.0931, "samples": 55680, "wall_s": 3415.2, "cuda_gb": 16.86} |
| {"step": 1760, "loss": 0.004512, "loss_mean50": 0.004877, "lr": 2.4837220525786454e-05, "grad_norm": 0.0368, "samples": 56320, "wall_s": 3454.1, "cuda_gb": 16.86} |
| {"step": 1780, "loss": 0.005115, "loss_mean50": 0.004864, "lr": 2.4828551014500046e-05, "grad_norm": 0.0449, "samples": 56960, "wall_s": 3492.9, "cuda_gb": 16.86} |
| {"step": 1800, "loss": 0.004481, "loss_mean50": 0.004535, "lr": 2.4819658358861965e-05, "grad_norm": 0.064, "samples": 57600, "wall_s": 3531.7, "cuda_gb": 16.86} |
| {"step": 1820, "loss": 0.004727, "loss_mean50": 0.004585, "lr": 2.4810542737988217e-05, "grad_norm": 0.0412, "samples": 58240, "wall_s": 3570.6, "cuda_gb": 16.86} |
| {"step": 1840, "loss": 0.00418, "loss_mean50": 0.004491, "lr": 2.480120433548579e-05, "grad_norm": 0.0319, "samples": 58880, "wall_s": 3609.3, "cuda_gb": 16.86} |
| {"step": 1860, "loss": 0.00481, "loss_mean50": 0.004756, "lr": 2.479164333944893e-05, "grad_norm": 0.0557, "samples": 59520, "wall_s": 3648.2, "cuda_gb": 16.86} |
| {"step": 1880, "loss": 0.003528, "loss_mean50": 0.004663, "lr": 2.4781859942455375e-05, "grad_norm": 0.0578, "samples": 60160, "wall_s": 3687.0, "cuda_gb": 16.86} |
| {"step": 1900, "loss": 0.003457, "loss_mean50": 0.00484, "lr": 2.4771854341562465e-05, "grad_norm": 0.0328, "samples": 60800, "wall_s": 3725.9, "cuda_gb": 16.86} |
| {"step": 1920, "loss": 0.003457, "loss_mean50": 0.004482, "lr": 2.476162673830317e-05, "grad_norm": 0.0332, "samples": 61440, "wall_s": 3764.7, "cuda_gb": 16.86} |
| {"step": 1940, "loss": 0.003694, "loss_mean50": 0.004257, "lr": 2.475117733868204e-05, "grad_norm": 0.035, "samples": 62080, "wall_s": 3803.5, "cuda_gb": 16.86} |
| {"step": 1960, "loss": 0.002911, "loss_mean50": 0.004106, "lr": 2.474050635317105e-05, "grad_norm": 0.0201, "samples": 62720, "wall_s": 3842.3, "cuda_gb": 16.86} |
| {"step": 1980, "loss": 0.005369, "loss_mean50": 0.004522, "lr": 2.472961399670536e-05, "grad_norm": 0.0445, "samples": 63360, "wall_s": 3881.0, "cuda_gb": 16.86} |
| {"step": 2000, "loss": 0.00414, "loss_mean50": 0.004677, "lr": 2.471850048867899e-05, "grad_norm": 0.0432, "samples": 64000, "wall_s": 3919.8, "cuda_gb": 16.86} |
| {"step": 2000, "val_flow_mse": 0.00438} |
| {"step": 2020, "loss": 0.008193, "loss_mean50": 0.004682, "lr": 2.470716605294039e-05, "grad_norm": 0.045, "samples": 64640, "wall_s": 3987.4, "cuda_gb": 16.86} |
| {"step": 2040, "loss": 0.003519, "loss_mean50": 0.004353, "lr": 2.4695610917787952e-05, "grad_norm": 0.0327, "samples": 65280, "wall_s": 4026.1, "cuda_gb": 16.86} |
| {"step": 2060, "loss": 0.003442, "loss_mean50": 0.004474, "lr": 2.468383531596539e-05, "grad_norm": 0.0397, "samples": 65920, "wall_s": 4064.8, "cuda_gb": 16.86} |
| {"step": 2080, "loss": 0.003777, "loss_mean50": 0.004547, "lr": 2.467183948465707e-05, "grad_norm": 0.0329, "samples": 66560, "wall_s": 4103.6, "cuda_gb": 16.86} |
| {"step": 2100, "loss": 0.003108, "loss_mean50": 0.00431, "lr": 2.4659623665483212e-05, "grad_norm": 0.0318, "samples": 67200, "wall_s": 4142.4, "cuda_gb": 16.86} |
| {"step": 2120, "loss": 0.00305, "loss_mean50": 0.00419, "lr": 2.4647188104495043e-05, "grad_norm": 0.0618, "samples": 67840, "wall_s": 4181.2, "cuda_gb": 16.86} |
| {"step": 2140, "loss": 0.003254, "loss_mean50": 0.004272, "lr": 2.463453305216983e-05, "grad_norm": 0.0472, "samples": 68480, "wall_s": 4220.0, "cuda_gb": 16.86} |
| {"step": 2160, "loss": 0.005482, "loss_mean50": 0.004401, "lr": 2.4621658763405842e-05, "grad_norm": 0.0393, "samples": 69120, "wall_s": 4258.8, "cuda_gb": 16.86} |
| {"step": 2180, "loss": 0.004781, "loss_mean50": 0.004529, "lr": 2.4608565497517207e-05, "grad_norm": 0.0731, "samples": 69760, "wall_s": 4297.5, "cuda_gb": 16.86} |
| {"step": 2200, "loss": 0.003903, "loss_mean50": 0.00441, "lr": 2.4595253518228692e-05, "grad_norm": 0.0328, "samples": 70400, "wall_s": 4336.3, "cuda_gb": 16.86} |
| {"step": 2220, "loss": 0.003748, "loss_mean50": 0.004269, "lr": 2.4581723093670407e-05, "grad_norm": 0.0527, "samples": 71040, "wall_s": 4375.2, "cuda_gb": 16.86} |
| {"step": 2240, "loss": 0.004003, "loss_mean50": 0.004113, "lr": 2.456797449637237e-05, "grad_norm": 0.0458, "samples": 71680, "wall_s": 4413.9, "cuda_gb": 16.86} |
| {"step": 2260, "loss": 0.005726, "loss_mean50": 0.004191, "lr": 2.455400800325905e-05, "grad_norm": 0.0768, "samples": 72320, "wall_s": 4452.7, "cuda_gb": 16.86} |
| {"step": 2280, "loss": 0.004988, "loss_mean50": 0.004462, "lr": 2.4539823895643783e-05, "grad_norm": 0.0505, "samples": 72960, "wall_s": 4491.5, "cuda_gb": 16.86} |
| {"step": 2300, "loss": 0.002699, "loss_mean50": 0.004672, "lr": 2.4525422459223077e-05, "grad_norm": 0.0267, "samples": 73600, "wall_s": 4530.3, "cuda_gb": 16.86} |
| {"step": 2320, "loss": 0.00222, "loss_mean50": 0.004422, "lr": 2.451080398407091e-05, "grad_norm": 0.0236, "samples": 74240, "wall_s": 4569.1, "cuda_gb": 16.86} |
| {"step": 2340, "loss": 0.00757, "loss_mean50": 0.00421, "lr": 2.4495968764632826e-05, "grad_norm": 0.0726, "samples": 74880, "wall_s": 4608.0, "cuda_gb": 16.86} |
| {"step": 2360, "loss": 0.002517, "loss_mean50": 0.004162, "lr": 2.4480917099720065e-05, "grad_norm": 0.0395, "samples": 75520, "wall_s": 4646.7, "cuda_gb": 16.86} |
| {"step": 2380, "loss": 0.00532, "loss_mean50": 0.004288, "lr": 2.4465649292503495e-05, "grad_norm": 0.0475, "samples": 76160, "wall_s": 4685.5, "cuda_gb": 16.86} |
| {"step": 2400, "loss": 0.003062, "loss_mean50": 0.004158, "lr": 2.4450165650507544e-05, "grad_norm": 0.0338, "samples": 76800, "wall_s": 4724.3, "cuda_gb": 16.86} |
| {"step": 2420, "loss": 0.002526, "loss_mean50": 0.00405, "lr": 2.443446648560396e-05, "grad_norm": 0.0317, "samples": 77440, "wall_s": 4763.1, "cuda_gb": 16.86} |
| {"step": 2440, "loss": 0.003149, "loss_mean50": 0.003901, "lr": 2.4418552114005585e-05, "grad_norm": 0.0303, "samples": 78080, "wall_s": 4801.8, "cuda_gb": 16.86} |
| {"step": 2460, "loss": 0.004051, "loss_mean50": 0.003901, "lr": 2.4402422856259944e-05, "grad_norm": 0.0399, "samples": 78720, "wall_s": 4840.8, "cuda_gb": 16.86} |
| {"step": 2480, "loss": 0.003464, "loss_mean50": 0.00389, "lr": 2.43860790372428e-05, "grad_norm": 0.0381, "samples": 79360, "wall_s": 4879.6, "cuda_gb": 16.86} |
| {"step": 2500, "loss": 0.003728, "loss_mean50": 0.003801, "lr": 2.4369520986151623e-05, "grad_norm": 0.03, "samples": 80000, "wall_s": 4918.5, "cuda_gb": 16.86} |
| {"step": 2520, "loss": 0.002743, "loss_mean50": 0.003867, "lr": 2.435274903649894e-05, "grad_norm": 0.0276, "samples": 80640, "wall_s": 4986.0, "cuda_gb": 16.86} |
| {"step": 2540, "loss": 0.002614, "loss_mean50": 0.003818, "lr": 2.4335763526105632e-05, "grad_norm": 0.0373, "samples": 81280, "wall_s": 5024.8, "cuda_gb": 16.86} |
| {"step": 2560, "loss": 0.002781, "loss_mean50": 0.003966, "lr": 2.4318564797094113e-05, "grad_norm": 0.0341, "samples": 81920, "wall_s": 5063.6, "cuda_gb": 16.86} |
| {"step": 2580, "loss": 0.002755, "loss_mean50": 0.004007, "lr": 2.4301153195881467e-05, "grad_norm": 0.0357, "samples": 82560, "wall_s": 5102.4, "cuda_gb": 16.86} |
| {"step": 2600, "loss": 0.004515, "loss_mean50": 0.004128, "lr": 2.4283529073172444e-05, "grad_norm": 0.0374, "samples": 83200, "wall_s": 5141.3, "cuda_gb": 16.86} |
| {"step": 2620, "loss": 0.002331, "loss_mean50": 0.004195, "lr": 2.42656927839524e-05, "grad_norm": 0.0246, "samples": 83840, "wall_s": 5180.2, "cuda_gb": 16.86} |
| {"step": 2640, "loss": 0.001871, "loss_mean50": 0.00402, "lr": 2.4247644687480155e-05, "grad_norm": 0.0313, "samples": 84480, "wall_s": 5219.1, "cuda_gb": 16.86} |
| {"step": 2660, "loss": 0.00383, "loss_mean50": 0.003963, "lr": 2.4229385147280765e-05, "grad_norm": 0.0362, "samples": 85120, "wall_s": 5257.9, "cuda_gb": 16.86} |
| {"step": 2680, "loss": 0.004455, "loss_mean50": 0.003792, "lr": 2.4210914531138177e-05, "grad_norm": 0.0448, "samples": 85760, "wall_s": 5296.7, "cuda_gb": 16.86} |
| {"step": 2700, "loss": 0.00693, "loss_mean50": 0.003884, "lr": 2.419223321108783e-05, "grad_norm": 0.058, "samples": 86400, "wall_s": 5335.6, "cuda_gb": 16.86} |
| {"step": 2720, "loss": 0.005898, "loss_mean50": 0.00418, "lr": 2.4173341563409186e-05, "grad_norm": 0.0605, "samples": 87040, "wall_s": 5374.4, "cuda_gb": 16.86} |
| {"step": 2740, "loss": 0.004664, "loss_mean50": 0.004267, "lr": 2.41542399686181e-05, "grad_norm": 0.049, "samples": 87680, "wall_s": 5413.3, "cuda_gb": 16.86} |
| {"step": 2760, "loss": 0.002756, "loss_mean50": 0.004026, "lr": 2.413492881145921e-05, "grad_norm": 0.0292, "samples": 88320, "wall_s": 5452.1, "cuda_gb": 16.86} |
| {"step": 2780, "loss": 0.002532, "loss_mean50": 0.003684, "lr": 2.4115408480898146e-05, "grad_norm": 0.0263, "samples": 88960, "wall_s": 5490.8, "cuda_gb": 16.86} |
| {"step": 2800, "loss": 0.003931, "loss_mean50": 0.003494, "lr": 2.4095679370113723e-05, "grad_norm": 0.035, "samples": 89600, "wall_s": 5529.7, "cuda_gb": 16.86} |
| {"step": 2820, "loss": 0.003607, "loss_mean50": 0.003542, "lr": 2.407574187649e-05, "grad_norm": 0.0361, "samples": 90240, "wall_s": 5568.4, "cuda_gb": 16.86} |
| {"step": 2840, "loss": 0.002552, "loss_mean50": 0.003717, "lr": 2.4055596401608298e-05, "grad_norm": 0.0236, "samples": 90880, "wall_s": 5607.3, "cuda_gb": 16.86} |
| {"step": 2860, "loss": 0.002596, "loss_mean50": 0.003857, "lr": 2.4035243351239095e-05, "grad_norm": 0.0291, "samples": 91520, "wall_s": 5646.0, "cuda_gb": 16.86} |
| {"step": 2880, "loss": 0.004721, "loss_mean50": 0.003871, "lr": 2.4014683135333848e-05, "grad_norm": 0.0557, "samples": 92160, "wall_s": 5684.8, "cuda_gb": 16.86} |
| {"step": 2900, "loss": 0.002337, "loss_mean50": 0.00357, "lr": 2.3993916168016762e-05, "grad_norm": 0.0296, "samples": 92800, "wall_s": 5723.6, "cuda_gb": 16.86} |
| {"step": 2920, "loss": 0.003542, "loss_mean50": 0.003801, "lr": 2.3972942867576424e-05, "grad_norm": 0.0359, "samples": 93440, "wall_s": 5762.4, "cuda_gb": 16.86} |
| {"step": 2940, "loss": 0.003904, "loss_mean50": 0.003902, "lr": 2.3951763656457387e-05, "grad_norm": 0.0382, "samples": 94080, "wall_s": 5801.3, "cuda_gb": 16.86} |
| {"step": 2960, "loss": 0.007087, "loss_mean50": 0.003945, "lr": 2.3930378961251658e-05, "grad_norm": 0.0417, "samples": 94720, "wall_s": 5840.1, "cuda_gb": 16.86} |
| {"step": 2980, "loss": 0.003405, "loss_mean50": 0.003846, "lr": 2.3908789212690112e-05, "grad_norm": 0.0396, "samples": 95360, "wall_s": 5878.9, "cuda_gb": 16.86} |
| {"step": 3000, "loss": 0.002657, "loss_mean50": 0.003707, "lr": 2.3886994845633805e-05, "grad_norm": 0.0487, "samples": 96000, "wall_s": 5917.7, "cuda_gb": 16.86} |
| {"step": 3000, "val_flow_mse": 0.003085} |
| {"step": 3020, "loss": 0.003534, "loss_mean50": 0.003528, "lr": 2.3864996299065235e-05, "grad_norm": 0.0343, "samples": 96640, "wall_s": 5985.3, "cuda_gb": 16.86} |
| {"step": 3040, "loss": 0.003788, "loss_mean50": 0.003577, "lr": 2.384279401607947e-05, "grad_norm": 0.0319, "samples": 97280, "wall_s": 6024.2, "cuda_gb": 16.86} |
| {"step": 3060, "loss": 0.00416, "loss_mean50": 0.003894, "lr": 2.3820388443875256e-05, "grad_norm": 0.0329, "samples": 97920, "wall_s": 6062.9, "cuda_gb": 16.86} |
| {"step": 3080, "loss": 0.004086, "loss_mean50": 0.004092, "lr": 2.379778003374598e-05, "grad_norm": 0.0374, "samples": 98560, "wall_s": 6101.8, "cuda_gb": 16.86} |
| {"step": 3100, "loss": 0.005012, "loss_mean50": 0.003923, "lr": 2.3774969241070605e-05, "grad_norm": 0.05, "samples": 99200, "wall_s": 6140.5, "cuda_gb": 16.86} |
| {"step": 3120, "loss": 0.003306, "loss_mean50": 0.003803, "lr": 2.375195652530448e-05, "grad_norm": 0.0326, "samples": 99840, "wall_s": 6179.3, "cuda_gb": 16.86} |
| {"step": 3140, "loss": 0.003322, "loss_mean50": 0.003751, "lr": 2.3728742349970097e-05, "grad_norm": 0.0345, "samples": 100480, "wall_s": 6218.1, "cuda_gb": 16.86} |
| {"step": 3160, "loss": 0.005428, "loss_mean50": 0.00372, "lr": 2.3705327182647743e-05, "grad_norm": 0.0421, "samples": 101120, "wall_s": 6256.9, "cuda_gb": 16.86} |
| {"step": 3180, "loss": 0.003036, "loss_mean50": 0.003664, "lr": 2.3681711494966095e-05, "grad_norm": 0.0337, "samples": 101760, "wall_s": 6295.7, "cuda_gb": 16.86} |
| {"step": 3200, "loss": 0.002507, "loss_mean50": 0.003434, "lr": 2.365789576259271e-05, "grad_norm": 0.0244, "samples": 102400, "wall_s": 6334.4, "cuda_gb": 16.86} |
| {"step": 3220, "loss": 0.003814, "loss_mean50": 0.003443, "lr": 2.363388046522445e-05, "grad_norm": 0.0604, "samples": 103040, "wall_s": 6373.2, "cuda_gb": 16.86} |
| {"step": 3240, "loss": 0.003837, "loss_mean50": 0.003493, "lr": 2.3609666086577818e-05, "grad_norm": 0.0413, "samples": 103680, "wall_s": 6412.0, "cuda_gb": 16.86} |
| {"step": 3260, "loss": 0.00357, "loss_mean50": 0.003686, "lr": 2.358525311437922e-05, "grad_norm": 0.0375, "samples": 104320, "wall_s": 6450.9, "cuda_gb": 16.86} |
| {"step": 3280, "loss": 0.002541, "loss_mean50": 0.003646, "lr": 2.356064204035513e-05, "grad_norm": 0.0344, "samples": 104960, "wall_s": 6489.7, "cuda_gb": 16.86} |
| {"step": 3300, "loss": 0.005259, "loss_mean50": 0.003617, "lr": 2.3535833360222197e-05, "grad_norm": 0.0622, "samples": 105600, "wall_s": 6528.5, "cuda_gb": 16.86} |
| {"step": 3320, "loss": 0.004229, "loss_mean50": 0.00362, "lr": 2.351082757367725e-05, "grad_norm": 0.0392, "samples": 106240, "wall_s": 6567.3, "cuda_gb": 16.86} |
| {"step": 3340, "loss": 0.003208, "loss_mean50": 0.003469, "lr": 2.3485625184387237e-05, "grad_norm": 0.0267, "samples": 106880, "wall_s": 6606.1, "cuda_gb": 16.86} |
| {"step": 3360, "loss": 0.003009, "loss_mean50": 0.003471, "lr": 2.3460226699979094e-05, "grad_norm": 0.0307, "samples": 107520, "wall_s": 6644.8, "cuda_gb": 16.86} |
| {"step": 3380, "loss": 0.002373, "loss_mean50": 0.003362, "lr": 2.3434632632029495e-05, "grad_norm": 0.0271, "samples": 108160, "wall_s": 6683.6, "cuda_gb": 16.86} |
| {"step": 3400, "loss": 0.003661, "loss_mean50": 0.00365, "lr": 2.340884349605457e-05, "grad_norm": 0.0306, "samples": 108800, "wall_s": 6722.4, "cuda_gb": 16.86} |
| {"step": 3420, "loss": 0.002939, "loss_mean50": 0.003781, "lr": 2.3382859811499506e-05, "grad_norm": 0.0277, "samples": 109440, "wall_s": 6761.2, "cuda_gb": 16.86} |
| {"step": 3440, "loss": 0.002923, "loss_mean50": 0.003756, "lr": 2.3356682101728088e-05, "grad_norm": 0.0264, "samples": 110080, "wall_s": 6800.0, "cuda_gb": 16.86} |
| {"step": 3460, "loss": 0.003306, "loss_mean50": 0.003554, "lr": 2.3330310894012176e-05, "grad_norm": 0.0309, "samples": 110720, "wall_s": 6838.8, "cuda_gb": 16.86} |
| {"step": 3480, "loss": 0.004556, "loss_mean50": 0.003376, "lr": 2.3303746719521045e-05, "grad_norm": 0.0307, "samples": 111360, "wall_s": 6877.7, "cuda_gb": 16.86} |
| {"step": 3500, "loss": 0.002277, "loss_mean50": 0.003141, "lr": 2.3276990113310735e-05, "grad_norm": 0.0293, "samples": 112000, "wall_s": 6916.6, "cuda_gb": 16.86} |
| {"step": 3520, "loss": 0.003063, "loss_mean50": 0.003278, "lr": 2.3250041614313225e-05, "grad_norm": 0.0518, "samples": 112640, "wall_s": 6955.3, "cuda_gb": 16.86} |
| {"step": 3540, "loss": 0.003204, "loss_mean50": 0.003462, "lr": 2.3222901765325615e-05, "grad_norm": 0.0321, "samples": 113280, "wall_s": 6994.0, "cuda_gb": 16.86} |
| {"step": 3560, "loss": 0.004175, "loss_mean50": 0.003528, "lr": 2.3195571112999185e-05, "grad_norm": 0.0578, "samples": 113920, "wall_s": 7032.7, "cuda_gb": 16.86} |
| {"step": 3580, "loss": 0.003158, "loss_mean50": 0.003425, "lr": 2.3168050207828366e-05, "grad_norm": 0.0462, "samples": 114560, "wall_s": 7071.5, "cuda_gb": 16.86} |
| {"step": 3600, "loss": 0.004456, "loss_mean50": 0.003496, "lr": 2.3140339604139666e-05, "grad_norm": 0.0405, "samples": 115200, "wall_s": 7110.2, "cuda_gb": 16.86} |
| {"step": 3620, "loss": 0.002685, "loss_mean50": 0.003528, "lr": 2.3112439860080513e-05, "grad_norm": 0.0259, "samples": 115840, "wall_s": 7149.0, "cuda_gb": 16.86} |
| {"step": 3640, "loss": 0.00152, "loss_mean50": 0.003437, "lr": 2.308435153760799e-05, "grad_norm": 0.0205, "samples": 116480, "wall_s": 7187.8, "cuda_gb": 16.86} |
| {"step": 3660, "loss": 0.002592, "loss_mean50": 0.003461, "lr": 2.305607520247755e-05, "grad_norm": 0.0348, "samples": 117120, "wall_s": 7226.6, "cuda_gb": 16.86} |
| {"step": 3680, "loss": 0.002383, "loss_mean50": 0.003603, "lr": 2.302761142423158e-05, "grad_norm": 0.0308, "samples": 117760, "wall_s": 7265.4, "cuda_gb": 16.86} |
| {"step": 3700, "loss": 0.00293, "loss_mean50": 0.003505, "lr": 2.2998960776187957e-05, "grad_norm": 0.0546, "samples": 118400, "wall_s": 7304.2, "cuda_gb": 16.86} |
| {"step": 3720, "loss": 0.00357, "loss_mean50": 0.003502, "lr": 2.2970123835428493e-05, "grad_norm": 0.0367, "samples": 119040, "wall_s": 7343.1, "cuda_gb": 16.86} |
| {"step": 3740, "loss": 0.002799, "loss_mean50": 0.003642, "lr": 2.2941101182787307e-05, "grad_norm": 0.025, "samples": 119680, "wall_s": 7381.9, "cuda_gb": 16.86} |
| {"step": 3760, "loss": 0.003203, "loss_mean50": 0.003672, "lr": 2.2911893402839136e-05, "grad_norm": 0.039, "samples": 120320, "wall_s": 7420.8, "cuda_gb": 16.86} |
| {"step": 3780, "loss": 0.002837, "loss_mean50": 0.003631, "lr": 2.288250108388755e-05, "grad_norm": 0.0361, "samples": 120960, "wall_s": 7459.5, "cuda_gb": 16.86} |
| {"step": 3800, "loss": 0.001825, "loss_mean50": 0.003704, "lr": 2.2852924817953105e-05, "grad_norm": 0.0238, "samples": 121600, "wall_s": 7498.3, "cuda_gb": 16.86} |
| {"step": 3820, "loss": 0.002231, "loss_mean50": 0.003453, "lr": 2.2823165200761428e-05, "grad_norm": 0.0256, "samples": 122240, "wall_s": 7537.1, "cuda_gb": 16.86} |
| {"step": 3840, "loss": 0.00278, "loss_mean50": 0.003245, "lr": 2.27932228317312e-05, "grad_norm": 0.0702, "samples": 122880, "wall_s": 7575.8, "cuda_gb": 16.86} |
| {"step": 3860, "loss": 0.003522, "loss_mean50": 0.003128, "lr": 2.2763098313962093e-05, "grad_norm": 0.0311, "samples": 123520, "wall_s": 7614.6, "cuda_gb": 16.86} |
| {"step": 3880, "loss": 0.002572, "loss_mean50": 0.003283, "lr": 2.2732792254222633e-05, "grad_norm": 0.0256, "samples": 124160, "wall_s": 7653.4, "cuda_gb": 16.86} |
| {"step": 3900, "loss": 0.002149, "loss_mean50": 0.003127, "lr": 2.2702305262937958e-05, "grad_norm": 0.0302, "samples": 124800, "wall_s": 7692.2, "cuda_gb": 16.86} |
| {"step": 3920, "loss": 0.002681, "loss_mean50": 0.003217, "lr": 2.2671637954177526e-05, "grad_norm": 0.0243, "samples": 125440, "wall_s": 7731.1, "cuda_gb": 16.86} |
| {"step": 3940, "loss": 0.001681, "loss_mean50": 0.003263, "lr": 2.2640790945642772e-05, "grad_norm": 0.0404, "samples": 126080, "wall_s": 7769.9, "cuda_gb": 16.86} |
| {"step": 3960, "loss": 0.004127, "loss_mean50": 0.003439, "lr": 2.2609764858654632e-05, "grad_norm": 0.0658, "samples": 126720, "wall_s": 7808.8, "cuda_gb": 16.86} |
| {"step": 3980, "loss": 0.003119, "loss_mean50": 0.003423, "lr": 2.2578560318141043e-05, "grad_norm": 0.0351, "samples": 127360, "wall_s": 7847.6, "cuda_gb": 16.86} |
| {"step": 4000, "loss": 0.006203, "loss_mean50": 0.003511, "lr": 2.254717795262436e-05, "grad_norm": 0.0482, "samples": 128000, "wall_s": 7886.4, "cuda_gb": 16.86} |
| {"step": 4000, "val_flow_mse": 0.002829} |
| {"step": 4020, "loss": 0.003558, "loss_mean50": 0.003533, "lr": 2.2515618394208694e-05, "grad_norm": 0.0315, "samples": 128640, "wall_s": 7954.0, "cuda_gb": 16.86} |
| {"step": 4040, "loss": 0.0022, "loss_mean50": 0.003263, "lr": 2.248388227856717e-05, "grad_norm": 0.0245, "samples": 129280, "wall_s": 7992.8, "cuda_gb": 16.86} |
| {"step": 4060, "loss": 0.002106, "loss_mean50": 0.003061, "lr": 2.2451970244929137e-05, "grad_norm": 0.035, "samples": 129920, "wall_s": 8031.6, "cuda_gb": 16.86} |
| {"step": 4080, "loss": 0.003493, "loss_mean50": 0.002948, "lr": 2.2419882936067294e-05, "grad_norm": 0.0318, "samples": 130560, "wall_s": 8070.4, "cuda_gb": 16.86} |
| {"step": 4100, "loss": 0.003395, "loss_mean50": 0.00307, "lr": 2.238762099828472e-05, "grad_norm": 0.038, "samples": 131200, "wall_s": 8109.3, "cuda_gb": 16.86} |
| {"step": 4120, "loss": 0.00268, "loss_mean50": 0.003055, "lr": 2.235518508140189e-05, "grad_norm": 0.0341, "samples": 131840, "wall_s": 8148.2, "cuda_gb": 16.86} |
| {"step": 4140, "loss": 0.002226, "loss_mean50": 0.003214, "lr": 2.232257583874355e-05, "grad_norm": 0.0287, "samples": 132480, "wall_s": 8187.0, "cuda_gb": 16.86} |
| {"step": 4160, "loss": 0.002028, "loss_mean50": 0.003231, "lr": 2.2289793927125603e-05, "grad_norm": 0.0358, "samples": 133120, "wall_s": 8226.0, "cuda_gb": 16.86} |
| {"step": 4180, "loss": 0.002479, "loss_mean50": 0.003011, "lr": 2.225684000684182e-05, "grad_norm": 0.0437, "samples": 133760, "wall_s": 8264.8, "cuda_gb": 16.86} |
| {"step": 4200, "loss": 0.005411, "loss_mean50": 0.003178, "lr": 2.22237147416506e-05, "grad_norm": 0.0404, "samples": 134400, "wall_s": 8303.7, "cuda_gb": 16.86} |
| {"step": 4220, "loss": 0.00234, "loss_mean50": 0.003114, "lr": 2.2190418798761562e-05, "grad_norm": 0.0227, "samples": 135040, "wall_s": 8342.6, "cuda_gb": 16.86} |
| {"step": 4240, "loss": 0.003831, "loss_mean50": 0.003307, "lr": 2.2156952848822125e-05, "grad_norm": 0.035, "samples": 135680, "wall_s": 8381.5, "cuda_gb": 16.86} |
| {"step": 4260, "loss": 0.003997, "loss_mean50": 0.003158, "lr": 2.212331756590397e-05, "grad_norm": 0.0804, "samples": 136320, "wall_s": 8420.3, "cuda_gb": 16.86} |
| {"step": 4280, "loss": 0.002618, "loss_mean50": 0.003229, "lr": 2.208951362748953e-05, "grad_norm": 0.0263, "samples": 136960, "wall_s": 8459.1, "cuda_gb": 16.86} |
| {"step": 4300, "loss": 0.003202, "loss_mean50": 0.003163, "lr": 2.2055541714458252e-05, "grad_norm": 0.0331, "samples": 137600, "wall_s": 8498.0, "cuda_gb": 16.86} |
| {"step": 4320, "loss": 0.002391, "loss_mean50": 0.002933, "lr": 2.2021402511072963e-05, "grad_norm": 0.0513, "samples": 138240, "wall_s": 8536.8, "cuda_gb": 16.86} |
| {"step": 4340, "loss": 0.003518, "loss_mean50": 0.003066, "lr": 2.1987096704966034e-05, "grad_norm": 0.0296, "samples": 138880, "wall_s": 8575.6, "cuda_gb": 16.86} |
| {"step": 4360, "loss": 0.004299, "loss_mean50": 0.003148, "lr": 2.1952624987125566e-05, "grad_norm": 0.0446, "samples": 139520, "wall_s": 8614.5, "cuda_gb": 16.86} |
| {"step": 4380, "loss": 0.001683, "loss_mean50": 0.003229, "lr": 2.1917988051881456e-05, "grad_norm": 0.022, "samples": 140160, "wall_s": 8653.3, "cuda_gb": 16.86} |
| {"step": 4400, "loss": 0.003349, "loss_mean50": 0.003095, "lr": 2.188318659689139e-05, "grad_norm": 0.0302, "samples": 140800, "wall_s": 8692.0, "cuda_gb": 16.86} |
| {"step": 4420, "loss": 0.00377, "loss_mean50": 0.002878, "lr": 2.1848221323126857e-05, "grad_norm": 0.0447, "samples": 141440, "wall_s": 8730.8, "cuda_gb": 16.86} |
| {"step": 4440, "loss": 0.004346, "loss_mean50": 0.002823, "lr": 2.1813092934858935e-05, "grad_norm": 0.0344, "samples": 142080, "wall_s": 8769.6, "cuda_gb": 16.86} |
| {"step": 4460, "loss": 0.003051, "loss_mean50": 0.002821, "lr": 2.17778021396442e-05, "grad_norm": 0.0384, "samples": 142720, "wall_s": 8808.3, "cuda_gb": 16.86} |
| {"step": 4480, "loss": 0.003501, "loss_mean50": 0.002872, "lr": 2.1742349648310403e-05, "grad_norm": 0.0854, "samples": 143360, "wall_s": 8847.1, "cuda_gb": 16.86} |
| {"step": 4500, "loss": 0.003207, "loss_mean50": 0.002831, "lr": 2.1706736174942196e-05, "grad_norm": 0.0386, "samples": 144000, "wall_s": 8885.9, "cuda_gb": 16.86} |
| {"step": 4520, "loss": 0.003039, "loss_mean50": 0.003013, "lr": 2.1670962436866725e-05, "grad_norm": 0.0249, "samples": 144640, "wall_s": 8924.7, "cuda_gb": 16.86} |
| {"step": 4540, "loss": 0.003363, "loss_mean50": 0.003253, "lr": 2.16350291546392e-05, "grad_norm": 0.0454, "samples": 145280, "wall_s": 8963.5, "cuda_gb": 16.86} |
| {"step": 4560, "loss": 0.002271, "loss_mean50": 0.003484, "lr": 2.159893705202836e-05, "grad_norm": 0.0364, "samples": 145920, "wall_s": 9002.2, "cuda_gb": 16.86} |
| {"step": 4580, "loss": 0.003364, "loss_mean50": 0.003347, "lr": 2.1562686856001923e-05, "grad_norm": 0.038, "samples": 146560, "wall_s": 9040.9, "cuda_gb": 16.86} |
| {"step": 4600, "loss": 0.003255, "loss_mean50": 0.003187, "lr": 2.1526279296711908e-05, "grad_norm": 0.0402, "samples": 147200, "wall_s": 9079.6, "cuda_gb": 16.86} |
| {"step": 4620, "loss": 0.003383, "loss_mean50": 0.003057, "lr": 2.1489715107479956e-05, "grad_norm": 0.0424, "samples": 147840, "wall_s": 9118.3, "cuda_gb": 16.86} |
| {"step": 4640, "loss": 0.002631, "loss_mean50": 0.003046, "lr": 2.1452995024782553e-05, "grad_norm": 0.0311, "samples": 148480, "wall_s": 9157.1, "cuda_gb": 16.86} |
| {"step": 4660, "loss": 0.002166, "loss_mean50": 0.002857, "lr": 2.1416119788236198e-05, "grad_norm": 0.0257, "samples": 149120, "wall_s": 9195.9, "cuda_gb": 16.86} |
| {"step": 4680, "loss": 0.002221, "loss_mean50": 0.003036, "lr": 2.1379090140582484e-05, "grad_norm": 0.0242, "samples": 149760, "wall_s": 9234.7, "cuda_gb": 16.86} |
| {"step": 4700, "loss": 0.001757, "loss_mean50": 0.003152, "lr": 2.1341906827673175e-05, "grad_norm": 0.0246, "samples": 150400, "wall_s": 9273.5, "cuda_gb": 16.86} |
| {"step": 4720, "loss": 0.004281, "loss_mean50": 0.003078, "lr": 2.130457059845515e-05, "grad_norm": 0.0265, "samples": 151040, "wall_s": 9312.3, "cuda_gb": 16.86} |
| {"step": 4740, "loss": 0.002234, "loss_mean50": 0.002876, "lr": 2.1267082204955335e-05, "grad_norm": 0.0288, "samples": 151680, "wall_s": 9351.0, "cuda_gb": 16.86} |
| {"step": 4760, "loss": 0.004462, "loss_mean50": 0.002996, "lr": 2.1229442402265554e-05, "grad_norm": 0.0526, "samples": 152320, "wall_s": 9389.8, "cuda_gb": 16.86} |
| {"step": 4780, "loss": 0.001817, "loss_mean50": 0.002877, "lr": 2.1191651948527313e-05, "grad_norm": 0.0183, "samples": 152960, "wall_s": 9428.5, "cuda_gb": 16.86} |
| {"step": 4800, "loss": 0.003229, "loss_mean50": 0.002842, "lr": 2.1153711604916535e-05, "grad_norm": 0.0299, "samples": 153600, "wall_s": 9467.2, "cuda_gb": 16.86} |
| {"step": 4820, "loss": 0.002039, "loss_mean50": 0.002745, "lr": 2.111562213562823e-05, "grad_norm": 0.0237, "samples": 154240, "wall_s": 9506.0, "cuda_gb": 16.86} |
| {"step": 4840, "loss": 0.00547, "loss_mean50": 0.002786, "lr": 2.1077384307861093e-05, "grad_norm": 0.0474, "samples": 154880, "wall_s": 9544.7, "cuda_gb": 16.86} |
| {"step": 4860, "loss": 0.002214, "loss_mean50": 0.00282, "lr": 2.103899889180206e-05, "grad_norm": 0.0372, "samples": 155520, "wall_s": 9583.5, "cuda_gb": 16.86} |
| {"step": 4880, "loss": 0.001556, "loss_mean50": 0.002789, "lr": 2.100046666061081e-05, "grad_norm": 0.0207, "samples": 156160, "wall_s": 9622.4, "cuda_gb": 16.86} |
| {"step": 4900, "loss": 0.001535, "loss_mean50": 0.002777, "lr": 2.096178839040414e-05, "grad_norm": 0.0147, "samples": 156800, "wall_s": 9661.2, "cuda_gb": 16.86} |
| {"step": 4920, "loss": 0.002268, "loss_mean50": 0.00277, "lr": 2.0922964860240404e-05, "grad_norm": 0.0463, "samples": 157440, "wall_s": 9699.9, "cuda_gb": 16.86} |
| {"step": 4940, "loss": 0.003021, "loss_mean50": 0.002663, "lr": 2.088399685210376e-05, "grad_norm": 0.0422, "samples": 158080, "wall_s": 9738.7, "cuda_gb": 16.86} |
| {"step": 4960, "loss": 0.00563, "loss_mean50": 0.002937, "lr": 2.0844885150888443e-05, "grad_norm": 0.1024, "samples": 158720, "wall_s": 9777.5, "cuda_gb": 16.86} |
| {"step": 4980, "loss": 0.001859, "loss_mean50": 0.002953, "lr": 2.0805630544382978e-05, "grad_norm": 0.0195, "samples": 159360, "wall_s": 9816.2, "cuda_gb": 16.86} |
| {"step": 5000, "loss": 0.003203, "loss_mean50": 0.002929, "lr": 2.076623382325427e-05, "grad_norm": 0.0362, "samples": 160000, "wall_s": 9855.0, "cuda_gb": 16.86} |
| {"step": 5000, "val_flow_mse": 0.002425, "val_action_mae": 0.020015, "best": true} |
| {"step": 5020, "loss": 0.002364, "loss_mean50": 0.002858, "lr": 2.0726695781031705e-05, "grad_norm": 0.0279, "samples": 160640, "wall_s": 9975.4, "cuda_gb": 16.86} |
| {"step": 5040, "loss": 0.00525, "loss_mean50": 0.003048, "lr": 2.0687017214091166e-05, "grad_norm": 0.1016, "samples": 161280, "wall_s": 10014.2, "cuda_gb": 16.86} |
| {"step": 5060, "loss": 0.00261, "loss_mean50": 0.002822, "lr": 2.064719892163898e-05, "grad_norm": 0.0384, "samples": 161920, "wall_s": 10053.0, "cuda_gb": 16.86} |
| {"step": 5080, "loss": 0.00206, "loss_mean50": 0.003011, "lr": 2.0607241705695827e-05, "grad_norm": 0.0255, "samples": 162560, "wall_s": 10091.8, "cuda_gb": 16.86} |
| {"step": 5100, "loss": 0.003248, "loss_mean50": 0.003163, "lr": 2.05671463710806e-05, "grad_norm": 0.0271, "samples": 163200, "wall_s": 10130.5, "cuda_gb": 16.86} |
| {"step": 5120, "loss": 0.003423, "loss_mean50": 0.003067, "lr": 2.0526913725394155e-05, "grad_norm": 0.0323, "samples": 163840, "wall_s": 10169.2, "cuda_gb": 16.86} |
| {"step": 5140, "loss": 0.003089, "loss_mean50": 0.002774, "lr": 2.0486544579003098e-05, "grad_norm": 0.0584, "samples": 164480, "wall_s": 10207.9, "cuda_gb": 16.86} |
| {"step": 5160, "loss": 0.002655, "loss_mean50": 0.002794, "lr": 2.044603974502341e-05, "grad_norm": 0.0271, "samples": 165120, "wall_s": 10246.7, "cuda_gb": 16.86} |
| {"step": 5180, "loss": 0.001697, "loss_mean50": 0.002805, "lr": 2.0405400039304117e-05, "grad_norm": 0.0367, "samples": 165760, "wall_s": 10285.4, "cuda_gb": 16.86} |
| {"step": 5200, "loss": 0.005638, "loss_mean50": 0.002778, "lr": 2.0364626280410812e-05, "grad_norm": 0.0549, "samples": 166400, "wall_s": 10324.2, "cuda_gb": 16.86} |
| {"step": 5220, "loss": 0.00216, "loss_mean50": 0.002868, "lr": 2.03237192896092e-05, "grad_norm": 0.0305, "samples": 167040, "wall_s": 10363.0, "cuda_gb": 16.86} |
| {"step": 5240, "loss": 0.003929, "loss_mean50": 0.003053, "lr": 2.028267989084855e-05, "grad_norm": 0.032, "samples": 167680, "wall_s": 10401.7, "cuda_gb": 16.86} |
| {"step": 5260, "loss": 0.001537, "loss_mean50": 0.002898, "lr": 2.024150891074508e-05, "grad_norm": 0.019, "samples": 168320, "wall_s": 10440.5, "cuda_gb": 16.86} |
| {"step": 5280, "loss": 0.003028, "loss_mean50": 0.003018, "lr": 2.0200207178565332e-05, "grad_norm": 0.0338, "samples": 168960, "wall_s": 10479.2, "cuda_gb": 16.86} |
| {"step": 5300, "loss": 0.001567, "loss_mean50": 0.002939, "lr": 2.015877552620946e-05, "grad_norm": 0.0204, "samples": 169600, "wall_s": 10518.0, "cuda_gb": 16.86} |
| {"step": 5320, "loss": 0.001758, "loss_mean50": 0.002798, "lr": 2.011721478819446e-05, "grad_norm": 0.0244, "samples": 170240, "wall_s": 10556.9, "cuda_gb": 16.86} |
| {"step": 5340, "loss": 0.002265, "loss_mean50": 0.002447, "lr": 2.007552580163739e-05, "grad_norm": 0.0214, "samples": 170880, "wall_s": 10595.6, "cuda_gb": 16.86} |
| {"step": 5360, "loss": 0.002338, "loss_mean50": 0.002561, "lr": 2.003370940623848e-05, "grad_norm": 0.0311, "samples": 171520, "wall_s": 10634.4, "cuda_gb": 16.86} |
| {"step": 5380, "loss": 0.007892, "loss_mean50": 0.002799, "lr": 1.9991766444264238e-05, "grad_norm": 0.0589, "samples": 172160, "wall_s": 10673.2, "cuda_gb": 16.86} |
| {"step": 5400, "loss": 0.003146, "loss_mean50": 0.002947, "lr": 1.9949697760530468e-05, "grad_norm": 0.0507, "samples": 172800, "wall_s": 10712.0, "cuda_gb": 16.86} |
| {"step": 5420, "loss": 0.00416, "loss_mean50": 0.002999, "lr": 1.9907504202385275e-05, "grad_norm": 0.091, "samples": 173440, "wall_s": 10750.8, "cuda_gb": 16.86} |
| {"step": 5440, "loss": 0.003324, "loss_mean50": 0.002865, "lr": 1.9865186619691983e-05, "grad_norm": 0.0483, "samples": 174080, "wall_s": 10789.6, "cuda_gb": 16.86} |
| {"step": 5460, "loss": 0.002857, "loss_mean50": 0.002708, "lr": 1.9822745864812015e-05, "grad_norm": 0.0445, "samples": 174720, "wall_s": 10828.4, "cuda_gb": 16.86} |
| {"step": 5480, "loss": 0.001979, "loss_mean50": 0.002791, "lr": 1.9780182792587738e-05, "grad_norm": 0.0442, "samples": 175360, "wall_s": 10867.2, "cuda_gb": 16.86} |
| {"step": 5500, "loss": 0.001799, "loss_mean50": 0.002776, "lr": 1.9737498260325232e-05, "grad_norm": 0.0252, "samples": 176000, "wall_s": 10906.1, "cuda_gb": 16.86} |
| {"step": 5520, "loss": 0.005635, "loss_mean50": 0.002661, "lr": 1.969469312777703e-05, "grad_norm": 0.0937, "samples": 176640, "wall_s": 10944.9, "cuda_gb": 16.86} |
| {"step": 5540, "loss": 0.002646, "loss_mean50": 0.002766, "lr": 1.9651768257124793e-05, "grad_norm": 0.0232, "samples": 177280, "wall_s": 10983.7, "cuda_gb": 16.86} |
| {"step": 5560, "loss": 0.002355, "loss_mean50": 0.002787, "lr": 1.9608724512961953e-05, "grad_norm": 0.032, "samples": 177920, "wall_s": 11022.5, "cuda_gb": 16.86} |
| {"step": 5580, "loss": 0.002233, "loss_mean50": 0.002797, "lr": 1.9565562762276297e-05, "grad_norm": 0.0263, "samples": 178560, "wall_s": 11061.4, "cuda_gb": 16.86} |
| {"step": 5600, "loss": 0.00203, "loss_mean50": 0.00274, "lr": 1.9522283874432492e-05, "grad_norm": 0.0509, "samples": 179200, "wall_s": 11100.2, "cuda_gb": 16.86} |
| {"step": 5620, "loss": 0.00357, "loss_mean50": 0.0028, "lr": 1.9478888721154596e-05, "grad_norm": 0.0748, "samples": 179840, "wall_s": 11139.0, "cuda_gb": 16.86} |
| {"step": 5640, "loss": 0.001885, "loss_mean50": 0.002707, "lr": 1.9435378176508466e-05, "grad_norm": 0.0344, "samples": 180480, "wall_s": 11177.8, "cuda_gb": 16.86} |
| {"step": 5660, "loss": 0.001158, "loss_mean50": 0.002713, "lr": 1.9391753116884205e-05, "grad_norm": 0.0153, "samples": 181120, "wall_s": 11216.8, "cuda_gb": 16.86} |
| {"step": 5680, "loss": 0.001981, "loss_mean50": 0.002761, "lr": 1.934801442097845e-05, "grad_norm": 0.0343, "samples": 181760, "wall_s": 11255.7, "cuda_gb": 16.86} |
| {"step": 5700, "loss": 0.006324, "loss_mean50": 0.002883, "lr": 1.9304162969776722e-05, "grad_norm": 0.0665, "samples": 182400, "wall_s": 11294.6, "cuda_gb": 16.86} |
| {"step": 5720, "loss": 0.002285, "loss_mean50": 0.002759, "lr": 1.9260199646535644e-05, "grad_norm": 0.0248, "samples": 183040, "wall_s": 11333.4, "cuda_gb": 16.86} |
| {"step": 5740, "loss": 0.003411, "loss_mean50": 0.002585, "lr": 1.921612533676518e-05, "grad_norm": 0.032, "samples": 183680, "wall_s": 11372.3, "cuda_gb": 16.86} |
| {"step": 5760, "loss": 0.00318, "loss_mean50": 0.002425, "lr": 1.9171940928210784e-05, "grad_norm": 0.049, "samples": 184320, "wall_s": 11411.7, "cuda_gb": 16.86} |
| {"step": 5780, "loss": 0.001864, "loss_mean50": 0.002815, "lr": 1.9127647310835513e-05, "grad_norm": 0.0233, "samples": 184960, "wall_s": 11450.4, "cuda_gb": 16.86} |
| {"step": 5800, "loss": 0.003762, "loss_mean50": 0.002809, "lr": 1.9083245376802126e-05, "grad_norm": 0.0298, "samples": 185600, "wall_s": 11489.4, "cuda_gb": 16.86} |
| {"step": 5820, "loss": 0.003034, "loss_mean50": 0.002664, "lr": 1.903873602045508e-05, "grad_norm": 0.0342, "samples": 186240, "wall_s": 11528.1, "cuda_gb": 16.86} |
| {"step": 5840, "loss": 0.00311, "loss_mean50": 0.002894, "lr": 1.8994120138302543e-05, "grad_norm": 0.0528, "samples": 186880, "wall_s": 11566.9, "cuda_gb": 16.86} |
| {"step": 5860, "loss": 0.00135, "loss_mean50": 0.002909, "lr": 1.894939862899833e-05, "grad_norm": 0.0208, "samples": 187520, "wall_s": 11605.9, "cuda_gb": 16.86} |
| {"step": 5880, "loss": 0.001841, "loss_mean50": 0.002922, "lr": 1.8904572393323796e-05, "grad_norm": 0.0314, "samples": 188160, "wall_s": 11644.6, "cuda_gb": 16.86} |
| {"step": 5900, "loss": 0.001547, "loss_mean50": 0.002677, "lr": 1.8859642334169693e-05, "grad_norm": 0.025, "samples": 188800, "wall_s": 11683.4, "cuda_gb": 16.86} |
| {"step": 5920, "loss": 0.004377, "loss_mean50": 0.002969, "lr": 1.8814609356517996e-05, "grad_norm": 0.0817, "samples": 189440, "wall_s": 11722.2, "cuda_gb": 16.86} |
| {"step": 5940, "loss": 0.004946, "loss_mean50": 0.002892, "lr": 1.8769474367423657e-05, "grad_norm": 0.0352, "samples": 190080, "wall_s": 11761.0, "cuda_gb": 16.86} |
| {"step": 5960, "loss": 0.001936, "loss_mean50": 0.002745, "lr": 1.872423827599635e-05, "grad_norm": 0.0172, "samples": 190720, "wall_s": 11799.8, "cuda_gb": 16.86} |
| {"step": 5980, "loss": 0.002521, "loss_mean50": 0.002662, "lr": 1.8678901993382156e-05, "grad_norm": 0.0369, "samples": 191360, "wall_s": 11838.5, "cuda_gb": 16.86} |
| {"step": 6000, "loss": 0.003034, "loss_mean50": 0.002558, "lr": 1.8633466432745193e-05, "grad_norm": 0.0265, "samples": 192000, "wall_s": 11877.3, "cuda_gb": 16.86} |
| {"step": 6000, "val_flow_mse": 0.002111} |
| {"step": 6020, "loss": 0.003802, "loss_mean50": 0.002541, "lr": 1.8587932509249265e-05, "grad_norm": 0.049, "samples": 192640, "wall_s": 11944.8, "cuda_gb": 16.86} |
| {"step": 6040, "loss": 0.008713, "loss_mean50": 0.002607, "lr": 1.8542301140039374e-05, "grad_norm": 0.0464, "samples": 193280, "wall_s": 11983.6, "cuda_gb": 16.86} |
| {"step": 6060, "loss": 0.001889, "loss_mean50": 0.002619, "lr": 1.849657324422329e-05, "grad_norm": 0.0191, "samples": 193920, "wall_s": 12022.4, "cuda_gb": 16.86} |
| {"step": 6080, "loss": 0.004152, "loss_mean50": 0.002633, "lr": 1.845074974285303e-05, "grad_norm": 0.0532, "samples": 194560, "wall_s": 12061.3, "cuda_gb": 16.86} |
| {"step": 6100, "loss": 0.004023, "loss_mean50": 0.002634, "lr": 1.8404831558906283e-05, "grad_norm": 0.0482, "samples": 195200, "wall_s": 12100.1, "cuda_gb": 16.86} |
| {"step": 6120, "loss": 0.003712, "loss_mean50": 0.002698, "lr": 1.8358819617267847e-05, "grad_norm": 0.0465, "samples": 195840, "wall_s": 12138.8, "cuda_gb": 16.86} |
| {"step": 6140, "loss": 0.002566, "loss_mean50": 0.002587, "lr": 1.831271484471099e-05, "grad_norm": 0.0329, "samples": 196480, "wall_s": 12177.7, "cuda_gb": 16.86} |
| {"step": 6160, "loss": 0.003213, "loss_mean50": 0.002722, "lr": 1.8266518169878774e-05, "grad_norm": 0.0635, "samples": 197120, "wall_s": 12216.6, "cuda_gb": 16.86} |
| {"step": 6180, "loss": 0.002498, "loss_mean50": 0.002674, "lr": 1.822023052326536e-05, "grad_norm": 0.0256, "samples": 197760, "wall_s": 12255.5, "cuda_gb": 16.86} |
| {"step": 6200, "loss": 0.002727, "loss_mean50": 0.002821, "lr": 1.8173852837197273e-05, "grad_norm": 0.0575, "samples": 198400, "wall_s": 12294.4, "cuda_gb": 16.86} |
| {"step": 6220, "loss": 0.002843, "loss_mean50": 0.002928, "lr": 1.81273860458146e-05, "grad_norm": 0.0893, "samples": 199040, "wall_s": 12333.3, "cuda_gb": 16.86} |
| {"step": 6240, "loss": 0.00339, "loss_mean50": 0.002776, "lr": 1.8080831085052197e-05, "grad_norm": 0.1139, "samples": 199680, "wall_s": 12372.2, "cuda_gb": 16.86} |
| {"step": 6260, "loss": 0.001208, "loss_mean50": 0.00252, "lr": 1.8034188892620834e-05, "grad_norm": 0.0214, "samples": 200320, "wall_s": 12411.1, "cuda_gb": 16.86} |
| {"step": 6280, "loss": 0.003755, "loss_mean50": 0.002709, "lr": 1.7987460407988294e-05, "grad_norm": 0.0479, "samples": 200960, "wall_s": 12450.1, "cuda_gb": 16.86} |
| {"step": 6300, "loss": 0.00272, "loss_mean50": 0.00248, "lr": 1.794064657236046e-05, "grad_norm": 0.0825, "samples": 201600, "wall_s": 12489.1, "cuda_gb": 16.86} |
| {"step": 6320, "loss": 0.003252, "loss_mean50": 0.00237, "lr": 1.7893748328662355e-05, "grad_norm": 0.0384, "samples": 202240, "wall_s": 12527.9, "cuda_gb": 16.86} |
| {"step": 6340, "loss": 0.001776, "loss_mean50": 0.002323, "lr": 1.7846766621519163e-05, "grad_norm": 0.0237, "samples": 202880, "wall_s": 12566.6, "cuda_gb": 16.86} |
| {"step": 6360, "loss": 0.001817, "loss_mean50": 0.002322, "lr": 1.7799702397237172e-05, "grad_norm": 0.0211, "samples": 203520, "wall_s": 12605.5, "cuda_gb": 16.86} |
| {"step": 6380, "loss": 0.002515, "loss_mean50": 0.002502, "lr": 1.775255660378474e-05, "grad_norm": 0.0484, "samples": 204160, "wall_s": 12644.4, "cuda_gb": 16.86} |
| {"step": 6400, "loss": 0.001913, "loss_mean50": 0.002366, "lr": 1.77053301907732e-05, "grad_norm": 0.0535, "samples": 204800, "wall_s": 12683.3, "cuda_gb": 16.86} |
| {"step": 6420, "loss": 0.002473, "loss_mean50": 0.002381, "lr": 1.765802410943771e-05, "grad_norm": 0.0365, "samples": 205440, "wall_s": 12722.3, "cuda_gb": 16.86} |
| {"step": 6440, "loss": 0.003558, "loss_mean50": 0.002527, "lr": 1.761063931261812e-05, "grad_norm": 0.0286, "samples": 206080, "wall_s": 12761.2, "cuda_gb": 16.86} |
| {"step": 6460, "loss": 0.003161, "loss_mean50": 0.002661, "lr": 1.7563176754739762e-05, "grad_norm": 0.0263, "samples": 206720, "wall_s": 12800.1, "cuda_gb": 16.86} |
| {"step": 6480, "loss": 0.002977, "loss_mean50": 0.002499, "lr": 1.751563739179424e-05, "grad_norm": 0.0813, "samples": 207360, "wall_s": 12839.1, "cuda_gb": 16.86} |
| {"step": 6500, "loss": 0.002479, "loss_mean50": 0.002485, "lr": 1.7468022181320165e-05, "grad_norm": 0.0314, "samples": 208000, "wall_s": 12878.0, "cuda_gb": 16.86} |
| {"step": 6520, "loss": 0.002245, "loss_mean50": 0.00257, "lr": 1.742033208238386e-05, "grad_norm": 0.0495, "samples": 208640, "wall_s": 12916.8, "cuda_gb": 16.86} |
| {"step": 6540, "loss": 0.001414, "loss_mean50": 0.002573, "lr": 1.737256805556007e-05, "grad_norm": 0.021, "samples": 209280, "wall_s": 12955.6, "cuda_gb": 16.86} |
| {"step": 6560, "loss": 0.001515, "loss_mean50": 0.002576, "lr": 1.7324731062912574e-05, "grad_norm": 0.0221, "samples": 209920, "wall_s": 12994.6, "cuda_gb": 16.86} |
| {"step": 6580, "loss": 0.0038, "loss_mean50": 0.002588, "lr": 1.7276822067974855e-05, "grad_norm": 0.0365, "samples": 210560, "wall_s": 13033.6, "cuda_gb": 16.86} |
| {"step": 6600, "loss": 0.00339, "loss_mean50": 0.002531, "lr": 1.722884203573065e-05, "grad_norm": 0.0417, "samples": 211200, "wall_s": 13072.6, "cuda_gb": 16.86} |
| {"step": 6620, "loss": 0.001454, "loss_mean50": 0.002456, "lr": 1.7180791932594536e-05, "grad_norm": 0.03, "samples": 211840, "wall_s": 13111.6, "cuda_gb": 16.86} |
| {"step": 6640, "loss": 0.002322, "loss_mean50": 0.002425, "lr": 1.7132672726392453e-05, "grad_norm": 0.0256, "samples": 212480, "wall_s": 13150.6, "cuda_gb": 16.86} |
| {"step": 6660, "loss": 0.007091, "loss_mean50": 0.002531, "lr": 1.708448538634223e-05, "grad_norm": 0.1327, "samples": 213120, "wall_s": 13189.5, "cuda_gb": 16.86} |
| {"step": 6680, "loss": 0.002018, "loss_mean50": 0.002548, "lr": 1.703623088303404e-05, "grad_norm": 0.0406, "samples": 213760, "wall_s": 13228.5, "cuda_gb": 16.86} |
| {"step": 6700, "loss": 0.00135, "loss_mean50": 0.002699, "lr": 1.698791018841086e-05, "grad_norm": 0.0237, "samples": 214400, "wall_s": 13267.4, "cuda_gb": 16.86} |
| {"step": 6720, "loss": 0.003559, "loss_mean50": 0.002552, "lr": 1.693952427574889e-05, "grad_norm": 0.043, "samples": 215040, "wall_s": 13306.4, "cuda_gb": 16.86} |
| {"step": 6740, "loss": 0.001537, "loss_mean50": 0.002367, "lr": 1.6891074119637956e-05, "grad_norm": 0.0216, "samples": 215680, "wall_s": 13345.4, "cuda_gb": 16.86} |
| {"step": 6760, "loss": 0.003062, "loss_mean50": 0.002376, "lr": 1.6842560695961884e-05, "grad_norm": 0.0322, "samples": 216320, "wall_s": 13384.4, "cuda_gb": 16.86} |
| {"step": 6780, "loss": 0.002052, "loss_mean50": 0.002467, "lr": 1.6793984981878822e-05, "grad_norm": 0.0509, "samples": 216960, "wall_s": 13423.4, "cuda_gb": 16.86} |
| {"step": 6800, "loss": 0.001136, "loss_mean50": 0.002463, "lr": 1.6745347955801585e-05, "grad_norm": 0.0177, "samples": 217600, "wall_s": 13462.3, "cuda_gb": 16.86} |
| {"step": 6820, "loss": 0.002584, "loss_mean50": 0.002279, "lr": 1.6696650597377923e-05, "grad_norm": 0.0267, "samples": 218240, "wall_s": 13501.2, "cuda_gb": 16.86} |
| {"step": 6840, "loss": 0.003376, "loss_mean50": 0.002435, "lr": 1.6647893887470816e-05, "grad_norm": 0.0331, "samples": 218880, "wall_s": 13540.2, "cuda_gb": 16.86} |
| {"step": 6860, "loss": 0.001567, "loss_mean50": 0.002517, "lr": 1.6599078808138693e-05, "grad_norm": 0.0213, "samples": 219520, "wall_s": 13579.2, "cuda_gb": 16.86} |
| {"step": 6880, "loss": 0.002267, "loss_mean50": 0.002673, "lr": 1.6550206342615658e-05, "grad_norm": 0.0243, "samples": 220160, "wall_s": 13618.1, "cuda_gb": 16.86} |
| {"step": 6900, "loss": 0.00179, "loss_mean50": 0.002519, "lr": 1.6501277475291696e-05, "grad_norm": 0.0183, "samples": 220800, "wall_s": 13657.1, "cuda_gb": 16.86} |
| {"step": 6920, "loss": 0.002676, "loss_mean50": 0.002587, "lr": 1.645229319169284e-05, "grad_norm": 0.0319, "samples": 221440, "wall_s": 13696.1, "cuda_gb": 16.86} |
| {"step": 6940, "loss": 0.001164, "loss_mean50": 0.002434, "lr": 1.64032544784613e-05, "grad_norm": 0.0249, "samples": 222080, "wall_s": 13735.0, "cuda_gb": 16.86} |
| {"step": 6960, "loss": 0.001766, "loss_mean50": 0.002406, "lr": 1.6354162323335633e-05, "grad_norm": 0.0299, "samples": 222720, "wall_s": 13774.0, "cuda_gb": 16.86} |
| {"step": 6980, "loss": 0.001862, "loss_mean50": 0.002331, "lr": 1.6305017715130812e-05, "grad_norm": 0.0323, "samples": 223360, "wall_s": 13813.0, "cuda_gb": 16.86} |
| {"step": 7000, "loss": 0.002183, "loss_mean50": 0.002179, "lr": 1.6255821643718313e-05, "grad_norm": 0.0277, "samples": 224000, "wall_s": 13851.8, "cuda_gb": 16.86} |
| {"step": 7000, "val_flow_mse": 0.002} |
| {"step": 7020, "loss": 0.001221, "loss_mean50": 0.002243, "lr": 1.6206575100006205e-05, "grad_norm": 0.0281, "samples": 224640, "wall_s": 13919.5, "cuda_gb": 16.86} |
| {"step": 7040, "loss": 0.005387, "loss_mean50": 0.002387, "lr": 1.615727907591915e-05, "grad_norm": 0.0645, "samples": 225280, "wall_s": 13958.3, "cuda_gb": 16.86} |
| {"step": 7060, "loss": 0.001143, "loss_mean50": 0.002539, "lr": 1.610793456437846e-05, "grad_norm": 0.0146, "samples": 225920, "wall_s": 13997.3, "cuda_gb": 16.86} |
| {"step": 7080, "loss": 0.003774, "loss_mean50": 0.002516, "lr": 1.6058542559282066e-05, "grad_norm": 0.0398, "samples": 226560, "wall_s": 14036.2, "cuda_gb": 16.86} |
| {"step": 7100, "loss": 0.001975, "loss_mean50": 0.002368, "lr": 1.6009104055484537e-05, "grad_norm": 0.0238, "samples": 227200, "wall_s": 14074.9, "cuda_gb": 16.86} |
| {"step": 7120, "loss": 0.002276, "loss_mean50": 0.002222, "lr": 1.5959620048776996e-05, "grad_norm": 0.0402, "samples": 227840, "wall_s": 14113.8, "cuda_gb": 16.86} |
| {"step": 7140, "loss": 0.003726, "loss_mean50": 0.002374, "lr": 1.59100915358671e-05, "grad_norm": 0.0313, "samples": 228480, "wall_s": 14152.8, "cuda_gb": 16.86} |
| {"step": 7160, "loss": 0.002538, "loss_mean50": 0.0026, "lr": 1.586051951435895e-05, "grad_norm": 0.0215, "samples": 229120, "wall_s": 14191.7, "cuda_gb": 16.86} |
| {"step": 7180, "loss": 0.001782, "loss_mean50": 0.002577, "lr": 1.5810904982733e-05, "grad_norm": 0.0252, "samples": 229760, "wall_s": 14230.6, "cuda_gb": 16.86} |
| {"step": 7200, "loss": 0.001824, "loss_mean50": 0.002362, "lr": 1.576124894032593e-05, "grad_norm": 0.0255, "samples": 230400, "wall_s": 14269.6, "cuda_gb": 16.86} |
| {"step": 7220, "loss": 0.002963, "loss_mean50": 0.002279, "lr": 1.5711552387310548e-05, "grad_norm": 0.0311, "samples": 231040, "wall_s": 14308.5, "cuda_gb": 16.86} |
| {"step": 7240, "loss": 0.00226, "loss_mean50": 0.002254, "lr": 1.5661816324675613e-05, "grad_norm": 0.0263, "samples": 231680, "wall_s": 14347.4, "cuda_gb": 16.86} |
| {"step": 7260, "loss": 0.002333, "loss_mean50": 0.00226, "lr": 1.56120417542057e-05, "grad_norm": 0.0297, "samples": 232320, "wall_s": 14386.4, "cuda_gb": 16.86} |
| {"step": 7280, "loss": 0.002289, "loss_mean50": 0.002203, "lr": 1.5562229678460998e-05, "grad_norm": 0.0277, "samples": 232960, "wall_s": 14425.3, "cuda_gb": 16.86} |
| {"step": 7300, "loss": 0.003027, "loss_mean50": 0.002257, "lr": 1.5512381100757136e-05, "grad_norm": 0.0838, "samples": 233600, "wall_s": 14464.2, "cuda_gb": 16.86} |
| {"step": 7320, "loss": 0.002356, "loss_mean50": 0.002375, "lr": 1.5462497025144967e-05, "grad_norm": 0.0267, "samples": 234240, "wall_s": 14503.2, "cuda_gb": 16.86} |
| {"step": 7340, "loss": 0.002322, "loss_mean50": 0.002275, "lr": 1.541257845639033e-05, "grad_norm": 0.0306, "samples": 234880, "wall_s": 14542.1, "cuda_gb": 16.86} |
| {"step": 7360, "loss": 0.002074, "loss_mean50": 0.002266, "lr": 1.536262639995384e-05, "grad_norm": 0.025, "samples": 235520, "wall_s": 14580.9, "cuda_gb": 16.86} |
| {"step": 7380, "loss": 0.002644, "loss_mean50": 0.00245, "lr": 1.531264186197062e-05, "grad_norm": 0.0846, "samples": 236160, "wall_s": 14619.8, "cuda_gb": 16.86} |
| {"step": 7400, "loss": 0.002823, "loss_mean50": 0.002594, "lr": 1.526262584923003e-05, "grad_norm": 0.0274, "samples": 236800, "wall_s": 14658.7, "cuda_gb": 16.86} |
| {"step": 7420, "loss": 0.002289, "loss_mean50": 0.002603, "lr": 1.5212579369155395e-05, "grad_norm": 0.0365, "samples": 237440, "wall_s": 14697.4, "cuda_gb": 16.86} |
| {"step": 7440, "loss": 0.002702, "loss_mean50": 0.002305, "lr": 1.5162503429783722e-05, "grad_norm": 0.0517, "samples": 238080, "wall_s": 14736.2, "cuda_gb": 16.86} |
| {"step": 7460, "loss": 0.001955, "loss_mean50": 0.002368, "lr": 1.5112399039745381e-05, "grad_norm": 0.0335, "samples": 238720, "wall_s": 14775.0, "cuda_gb": 16.86} |
| {"step": 7480, "loss": 0.001347, "loss_mean50": 0.002323, "lr": 1.5062267208243797e-05, "grad_norm": 0.0216, "samples": 239360, "wall_s": 14813.9, "cuda_gb": 16.86} |
| {"step": 7500, "loss": 0.002797, "loss_mean50": 0.002351, "lr": 1.5012108945035122e-05, "grad_norm": 0.023, "samples": 240000, "wall_s": 14852.8, "cuda_gb": 16.86} |
| {"step": 7520, "loss": 0.002334, "loss_mean50": 0.002428, "lr": 1.49619252604079e-05, "grad_norm": 0.0262, "samples": 240640, "wall_s": 14923.7, "cuda_gb": 16.86} |
| {"step": 7540, "loss": 0.003142, "loss_mean50": 0.002495, "lr": 1.49117171651627e-05, "grad_norm": 0.0299, "samples": 241280, "wall_s": 14962.6, "cuda_gb": 16.86} |
| {"step": 7560, "loss": 0.001165, "loss_mean50": 0.002397, "lr": 1.4861485670591782e-05, "grad_norm": 0.0165, "samples": 241920, "wall_s": 15001.6, "cuda_gb": 16.86} |
| {"step": 7580, "loss": 0.001434, "loss_mean50": 0.002126, "lr": 1.4811231788458716e-05, "grad_norm": 0.0459, "samples": 242560, "wall_s": 15040.7, "cuda_gb": 16.86} |
| {"step": 7600, "loss": 0.001966, "loss_mean50": 0.002126, "lr": 1.4760956530978001e-05, "grad_norm": 0.0246, "samples": 243200, "wall_s": 15079.7, "cuda_gb": 16.86} |
| {"step": 7620, "loss": 0.001114, "loss_mean50": 0.002268, "lr": 1.471066091079467e-05, "grad_norm": 0.025, "samples": 243840, "wall_s": 15118.6, "cuda_gb": 16.86} |
| {"step": 7640, "loss": 0.001375, "loss_mean50": 0.002237, "lr": 1.4660345940963918e-05, "grad_norm": 0.0277, "samples": 244480, "wall_s": 15157.6, "cuda_gb": 16.86} |
| {"step": 7660, "loss": 0.001449, "loss_mean50": 0.002039, "lr": 1.4610012634930665e-05, "grad_norm": 0.0463, "samples": 245120, "wall_s": 15196.6, "cuda_gb": 16.86} |
| {"step": 7680, "loss": 0.002418, "loss_mean50": 0.002069, "lr": 1.4559662006509167e-05, "grad_norm": 0.0306, "samples": 245760, "wall_s": 15235.6, "cuda_gb": 16.86} |
| {"step": 7700, "loss": 0.002786, "loss_mean50": 0.002185, "lr": 1.4509295069862588e-05, "grad_norm": 0.1553, "samples": 246400, "wall_s": 15274.5, "cuda_gb": 16.86} |
| {"step": 7720, "loss": 0.00223, "loss_mean50": 0.002139, "lr": 1.4458912839482571e-05, "grad_norm": 0.0279, "samples": 247040, "wall_s": 15313.4, "cuda_gb": 16.86} |
| {"step": 7740, "loss": 0.002138, "loss_mean50": 0.002102, "lr": 1.4408516330168806e-05, "grad_norm": 0.0684, "samples": 247680, "wall_s": 15352.3, "cuda_gb": 16.86} |
| {"step": 7760, "loss": 0.001508, "loss_mean50": 0.002363, "lr": 1.4358106557008589e-05, "grad_norm": 0.0305, "samples": 248320, "wall_s": 15391.1, "cuda_gb": 16.86} |
| {"step": 7780, "loss": 0.002209, "loss_mean50": 0.002409, "lr": 1.4307684535356378e-05, "grad_norm": 0.024, "samples": 248960, "wall_s": 15430.0, "cuda_gb": 16.86} |
| {"step": 7800, "loss": 0.002069, "loss_mean50": 0.002501, "lr": 1.4257251280813339e-05, "grad_norm": 0.0266, "samples": 249600, "wall_s": 15468.8, "cuda_gb": 16.86} |
| {"step": 7820, "loss": 0.003588, "loss_mean50": 0.002511, "lr": 1.420680780920689e-05, "grad_norm": 0.0302, "samples": 250240, "wall_s": 15507.7, "cuda_gb": 16.86} |
| {"step": 7840, "loss": 0.003451, "loss_mean50": 0.00243, "lr": 1.4156355136570243e-05, "grad_norm": 0.0305, "samples": 250880, "wall_s": 15546.7, "cuda_gb": 16.86} |
| {"step": 7860, "loss": 0.001588, "loss_mean50": 0.002364, "lr": 1.4105894279121939e-05, "grad_norm": 0.0359, "samples": 251520, "wall_s": 15585.7, "cuda_gb": 16.86} |
| {"step": 7880, "loss": 0.002507, "loss_mean50": 0.002511, "lr": 1.4055426253245374e-05, "grad_norm": 0.0278, "samples": 252160, "wall_s": 15624.6, "cuda_gb": 16.86} |
| {"step": 7900, "loss": 0.001366, "loss_mean50": 0.002515, "lr": 1.4004952075468333e-05, "grad_norm": 0.0288, "samples": 252800, "wall_s": 15663.5, "cuda_gb": 16.86} |
| {"step": 7920, "loss": 0.001933, "loss_mean50": 0.002485, "lr": 1.3954472762442508e-05, "grad_norm": 0.0437, "samples": 253440, "wall_s": 15702.5, "cuda_gb": 16.86} |
| {"step": 7940, "loss": 0.002905, "loss_mean50": 0.002509, "lr": 1.3903989330923037e-05, "grad_norm": 0.103, "samples": 254080, "wall_s": 15741.4, "cuda_gb": 16.86} |
| {"step": 7960, "loss": 0.002224, "loss_mean50": 0.002371, "lr": 1.3853502797747997e-05, "grad_norm": 0.026, "samples": 254720, "wall_s": 15780.5, "cuda_gb": 16.86} |
| {"step": 7980, "loss": 0.001176, "loss_mean50": 0.002406, "lr": 1.3803014179817949e-05, "grad_norm": 0.0231, "samples": 255360, "wall_s": 15819.5, "cuda_gb": 16.86} |
| {"step": 8000, "loss": 0.003055, "loss_mean50": 0.002455, "lr": 1.375252449407545e-05, "grad_norm": 0.0414, "samples": 256000, "wall_s": 15858.4, "cuda_gb": 16.86} |
| {"step": 8000, "val_flow_mse": 0.002003} |
| {"step": 8020, "loss": 0.00172, "loss_mean50": 0.002454, "lr": 1.3702034757484548e-05, "grad_norm": 0.0413, "samples": 256640, "wall_s": 15927.0, "cuda_gb": 16.86} |
| {"step": 8040, "loss": 0.004739, "loss_mean50": 0.002335, "lr": 1.3651545987010325e-05, "grad_norm": 0.039, "samples": 257280, "wall_s": 15966.3, "cuda_gb": 16.86} |
| {"step": 8060, "loss": 0.002524, "loss_mean50": 0.002265, "lr": 1.3601059199598407e-05, "grad_norm": 0.0292, "samples": 257920, "wall_s": 16005.9, "cuda_gb": 16.86} |
| {"step": 8080, "loss": 0.003509, "loss_mean50": 0.002266, "lr": 1.355057541215447e-05, "grad_norm": 0.0268, "samples": 258560, "wall_s": 16045.3, "cuda_gb": 16.86} |
| {"step": 8100, "loss": 0.002238, "loss_mean50": 0.002251, "lr": 1.3500095641523772e-05, "grad_norm": 0.0415, "samples": 259200, "wall_s": 16084.8, "cuda_gb": 16.86} |
| {"step": 8120, "loss": 0.001217, "loss_mean50": 0.002154, "lr": 1.3449620904470664e-05, "grad_norm": 0.0196, "samples": 259840, "wall_s": 16124.0, "cuda_gb": 16.86} |
| {"step": 8140, "loss": 0.001815, "loss_mean50": 0.002063, "lr": 1.3399152217658095e-05, "grad_norm": 0.0245, "samples": 260480, "wall_s": 16163.0, "cuda_gb": 16.86} |
| {"step": 8160, "loss": 0.003644, "loss_mean50": 0.002143, "lr": 1.3348690597627166e-05, "grad_norm": 0.0432, "samples": 261120, "wall_s": 16201.9, "cuda_gb": 16.86} |
| {"step": 8180, "loss": 0.002831, "loss_mean50": 0.002317, "lr": 1.3298237060776639e-05, "grad_norm": 0.0411, "samples": 261760, "wall_s": 16240.9, "cuda_gb": 16.86} |
| {"step": 8200, "loss": 0.002553, "loss_mean50": 0.002378, "lr": 1.324779262334246e-05, "grad_norm": 0.0355, "samples": 262400, "wall_s": 16279.7, "cuda_gb": 16.86} |
| {"step": 8220, "loss": 0.001914, "loss_mean50": 0.002187, "lr": 1.3197358301377289e-05, "grad_norm": 0.0629, "samples": 263040, "wall_s": 16318.6, "cuda_gb": 16.86} |
| {"step": 8240, "loss": 0.003147, "loss_mean50": 0.002241, "lr": 1.3146935110730046e-05, "grad_norm": 0.0626, "samples": 263680, "wall_s": 16357.4, "cuda_gb": 16.86} |
| {"step": 8260, "loss": 0.003112, "loss_mean50": 0.002415, "lr": 1.3096524067025451e-05, "grad_norm": 0.0513, "samples": 264320, "wall_s": 16396.7, "cuda_gb": 16.86} |
| {"step": 8280, "loss": 0.003859, "loss_mean50": 0.002488, "lr": 1.3046126185643545e-05, "grad_norm": 0.0323, "samples": 264960, "wall_s": 16436.1, "cuda_gb": 16.86} |
| {"step": 8300, "loss": 0.001584, "loss_mean50": 0.002356, "lr": 1.2995742481699261e-05, "grad_norm": 0.0205, "samples": 265600, "wall_s": 16475.5, "cuda_gb": 16.86} |
| {"step": 8320, "loss": 0.003899, "loss_mean50": 0.002299, "lr": 1.2945373970021971e-05, "grad_norm": 0.0541, "samples": 266240, "wall_s": 16515.0, "cuda_gb": 16.86} |
| {"step": 8340, "loss": 0.00332, "loss_mean50": 0.002306, "lr": 1.2895021665135043e-05, "grad_norm": 0.0273, "samples": 266880, "wall_s": 16554.6, "cuda_gb": 16.86} |
| {"step": 8360, "loss": 0.00147, "loss_mean50": 0.002261, "lr": 1.2844686581235405e-05, "grad_norm": 0.0199, "samples": 267520, "wall_s": 16594.2, "cuda_gb": 16.86} |
| {"step": 8380, "loss": 0.001568, "loss_mean50": 0.00216, "lr": 1.2794369732173127e-05, "grad_norm": 0.0359, "samples": 268160, "wall_s": 16633.7, "cuda_gb": 16.86} |
| {"step": 8400, "loss": 0.001802, "loss_mean50": 0.002262, "lr": 1.2744072131430981e-05, "grad_norm": 0.0456, "samples": 268800, "wall_s": 16673.3, "cuda_gb": 16.86} |
| {"step": 8420, "loss": 0.003712, "loss_mean50": 0.002242, "lr": 1.2693794792104049e-05, "grad_norm": 0.051, "samples": 269440, "wall_s": 16712.7, "cuda_gb": 16.86} |
| {"step": 8440, "loss": 0.002014, "loss_mean50": 0.002294, "lr": 1.2643538726879303e-05, "grad_norm": 0.0243, "samples": 270080, "wall_s": 16752.2, "cuda_gb": 16.86} |
| {"step": 8460, "loss": 0.001395, "loss_mean50": 0.002165, "lr": 1.2593304948015212e-05, "grad_norm": 0.0248, "samples": 270720, "wall_s": 16791.7, "cuda_gb": 16.86} |
| {"step": 8480, "loss": 0.002235, "loss_mean50": 0.00215, "lr": 1.254309446732135e-05, "grad_norm": 0.0275, "samples": 271360, "wall_s": 16830.9, "cuda_gb": 16.86} |
| {"step": 8500, "loss": 0.00185, "loss_mean50": 0.002153, "lr": 1.2492908296138028e-05, "grad_norm": 0.0236, "samples": 272000, "wall_s": 16869.8, "cuda_gb": 16.86} |
| {"step": 8520, "loss": 0.001335, "loss_mean50": 0.002128, "lr": 1.2442747445315909e-05, "grad_norm": 0.0287, "samples": 272640, "wall_s": 16908.7, "cuda_gb": 16.86} |
| {"step": 8540, "loss": 0.002042, "loss_mean50": 0.002035, "lr": 1.2392612925195645e-05, "grad_norm": 0.0386, "samples": 273280, "wall_s": 16947.7, "cuda_gb": 16.86} |
| {"step": 8560, "loss": 0.001558, "loss_mean50": 0.002072, "lr": 1.2342505745587546e-05, "grad_norm": 0.0225, "samples": 273920, "wall_s": 16986.7, "cuda_gb": 16.86} |
| {"step": 8580, "loss": 0.001189, "loss_mean50": 0.00205, "lr": 1.2292426915751223e-05, "grad_norm": 0.0208, "samples": 274560, "wall_s": 17025.4, "cuda_gb": 16.86} |
| {"step": 8600, "loss": 0.00207, "loss_mean50": 0.002173, "lr": 1.2242377444375272e-05, "grad_norm": 0.0551, "samples": 275200, "wall_s": 17064.1, "cuda_gb": 16.86} |
| {"step": 8620, "loss": 0.00204, "loss_mean50": 0.002216, "lr": 1.2192358339556945e-05, "grad_norm": 0.0355, "samples": 275840, "wall_s": 17103.3, "cuda_gb": 16.86} |
| {"step": 8640, "loss": 0.001472, "loss_mean50": 0.002074, "lr": 1.2142370608781848e-05, "grad_norm": 0.0306, "samples": 276480, "wall_s": 17142.8, "cuda_gb": 16.86} |
| {"step": 8660, "loss": 0.00239, "loss_mean50": 0.001982, "lr": 1.209241525890365e-05, "grad_norm": 0.0252, "samples": 277120, "wall_s": 17181.9, "cuda_gb": 16.86} |
| {"step": 8680, "loss": 0.001866, "loss_mean50": 0.001929, "lr": 1.204249329612381e-05, "grad_norm": 0.026, "samples": 277760, "wall_s": 17221.4, "cuda_gb": 16.86} |
| {"step": 8700, "loss": 0.001442, "loss_mean50": 0.00185, "lr": 1.1992605725971308e-05, "grad_norm": 0.021, "samples": 278400, "wall_s": 17260.4, "cuda_gb": 16.86} |
| {"step": 8720, "loss": 0.002424, "loss_mean50": 0.002014, "lr": 1.194275355328237e-05, "grad_norm": 0.0284, "samples": 279040, "wall_s": 17299.3, "cuda_gb": 16.86} |
| {"step": 8740, "loss": 0.000927, "loss_mean50": 0.001998, "lr": 1.1892937782180261e-05, "grad_norm": 0.0201, "samples": 279680, "wall_s": 17338.2, "cuda_gb": 16.86} |
| {"step": 8760, "loss": 0.001502, "loss_mean50": 0.002038, "lr": 1.1843159416055037e-05, "grad_norm": 0.0322, "samples": 280320, "wall_s": 17377.0, "cuda_gb": 16.86} |
| {"step": 8780, "loss": 0.001067, "loss_mean50": 0.001896, "lr": 1.1793419457543346e-05, "grad_norm": 0.018, "samples": 280960, "wall_s": 17415.8, "cuda_gb": 16.86} |
| {"step": 8800, "loss": 0.00184, "loss_mean50": 0.001975, "lr": 1.1743718908508222e-05, "grad_norm": 0.0286, "samples": 281600, "wall_s": 17454.6, "cuda_gb": 16.86} |
| {"step": 8820, "loss": 0.001582, "loss_mean50": 0.002153, "lr": 1.1694058770018925e-05, "grad_norm": 0.0225, "samples": 282240, "wall_s": 17493.3, "cuda_gb": 16.86} |
| {"step": 8840, "loss": 0.001512, "loss_mean50": 0.00222, "lr": 1.1644440042330754e-05, "grad_norm": 0.023, "samples": 282880, "wall_s": 17532.1, "cuda_gb": 16.86} |
| {"step": 8860, "loss": 0.001166, "loss_mean50": 0.002239, "lr": 1.1594863724864918e-05, "grad_norm": 0.039, "samples": 283520, "wall_s": 17571.0, "cuda_gb": 16.86} |
| {"step": 8880, "loss": 0.00196, "loss_mean50": 0.002251, "lr": 1.1545330816188398e-05, "grad_norm": 0.0217, "samples": 284160, "wall_s": 17609.9, "cuda_gb": 16.86} |
| {"step": 8900, "loss": 0.002536, "loss_mean50": 0.002199, "lr": 1.1495842313993832e-05, "grad_norm": 0.0252, "samples": 284800, "wall_s": 17648.8, "cuda_gb": 16.86} |
| {"step": 8920, "loss": 0.001908, "loss_mean50": 0.001954, "lr": 1.1446399215079417e-05, "grad_norm": 0.025, "samples": 285440, "wall_s": 17687.6, "cuda_gb": 16.86} |
| {"step": 8940, "loss": 0.001247, "loss_mean50": 0.001798, "lr": 1.1397002515328845e-05, "grad_norm": 0.0362, "samples": 286080, "wall_s": 17726.4, "cuda_gb": 16.86} |
| {"step": 8960, "loss": 0.003202, "loss_mean50": 0.00189, "lr": 1.1347653209691237e-05, "grad_norm": 0.0452, "samples": 286720, "wall_s": 17765.2, "cuda_gb": 16.86} |
| {"step": 8980, "loss": 0.003032, "loss_mean50": 0.002222, "lr": 1.1298352292161085e-05, "grad_norm": 0.0277, "samples": 287360, "wall_s": 17804.1, "cuda_gb": 16.86} |
| {"step": 9000, "loss": 0.001186, "loss_mean50": 0.002371, "lr": 1.1249100755758268e-05, "grad_norm": 0.0227, "samples": 288000, "wall_s": 17842.8, "cuda_gb": 16.86} |
| {"step": 9000, "val_flow_mse": 0.001819} |
| {"step": 9020, "loss": 0.000861, "loss_mean50": 0.002057, "lr": 1.1199899592508024e-05, "grad_norm": 0.0195, "samples": 288640, "wall_s": 17910.5, "cuda_gb": 16.86} |
| {"step": 9040, "loss": 0.001691, "loss_mean50": 0.001978, "lr": 1.1150749793420958e-05, "grad_norm": 0.023, "samples": 289280, "wall_s": 17949.4, "cuda_gb": 16.86} |
| {"step": 9060, "loss": 0.009274, "loss_mean50": 0.002005, "lr": 1.1101652348473113e-05, "grad_norm": 0.0719, "samples": 289920, "wall_s": 17988.2, "cuda_gb": 16.86} |
| {"step": 9080, "loss": 0.00206, "loss_mean50": 0.002048, "lr": 1.1052608246586007e-05, "grad_norm": 0.0281, "samples": 290560, "wall_s": 18027.0, "cuda_gb": 16.86} |
| {"step": 9100, "loss": 0.00148, "loss_mean50": 0.001997, "lr": 1.1003618475606737e-05, "grad_norm": 0.0216, "samples": 291200, "wall_s": 18065.8, "cuda_gb": 16.86} |
| {"step": 9120, "loss": 0.002729, "loss_mean50": 0.001851, "lr": 1.0954684022288039e-05, "grad_norm": 0.0584, "samples": 291840, "wall_s": 18104.7, "cuda_gb": 16.86} |
| {"step": 9140, "loss": 0.002828, "loss_mean50": 0.001915, "lr": 1.0905805872268463e-05, "grad_norm": 0.0271, "samples": 292480, "wall_s": 18143.5, "cuda_gb": 16.86} |
| {"step": 9160, "loss": 0.002794, "loss_mean50": 0.001888, "lr": 1.0856985010052487e-05, "grad_norm": 0.0319, "samples": 293120, "wall_s": 18182.4, "cuda_gb": 16.86} |
| {"step": 9180, "loss": 0.002482, "loss_mean50": 0.001914, "lr": 1.0808222418990693e-05, "grad_norm": 0.0445, "samples": 293760, "wall_s": 18221.2, "cuda_gb": 16.86} |
| {"step": 9200, "loss": 0.001495, "loss_mean50": 0.00192, "lr": 1.0759519081259975e-05, "grad_norm": 0.021, "samples": 294400, "wall_s": 18260.1, "cuda_gb": 16.86} |
| {"step": 9220, "loss": 0.001955, "loss_mean50": 0.00212, "lr": 1.071087597784374e-05, "grad_norm": 0.0307, "samples": 295040, "wall_s": 18299.1, "cuda_gb": 16.86} |
| {"step": 9240, "loss": 0.001786, "loss_mean50": 0.002007, "lr": 1.0662294088512154e-05, "grad_norm": 0.0223, "samples": 295680, "wall_s": 18338.0, "cuda_gb": 16.86} |
| {"step": 9260, "loss": 0.001567, "loss_mean50": 0.002119, "lr": 1.0613774391802405e-05, "grad_norm": 0.0274, "samples": 296320, "wall_s": 18376.8, "cuda_gb": 16.86} |
| {"step": 9280, "loss": 0.002396, "loss_mean50": 0.002116, "lr": 1.0565317864999011e-05, "grad_norm": 0.0834, "samples": 296960, "wall_s": 18415.7, "cuda_gb": 16.86} |
| {"step": 9300, "loss": 0.002298, "loss_mean50": 0.002085, "lr": 1.0516925484114096e-05, "grad_norm": 0.0359, "samples": 297600, "wall_s": 18454.7, "cuda_gb": 16.86} |
| {"step": 9320, "loss": 0.002129, "loss_mean50": 0.002019, "lr": 1.0468598223867779e-05, "grad_norm": 0.0233, "samples": 298240, "wall_s": 18493.5, "cuda_gb": 16.86} |
| {"step": 9340, "loss": 0.002078, "loss_mean50": 0.001981, "lr": 1.04203370576685e-05, "grad_norm": 0.0487, "samples": 298880, "wall_s": 18532.4, "cuda_gb": 16.86} |
| {"step": 9360, "loss": 0.001998, "loss_mean50": 0.00198, "lr": 1.037214295759345e-05, "grad_norm": 0.0206, "samples": 299520, "wall_s": 18571.3, "cuda_gb": 16.86} |
| {"step": 9380, "loss": 0.002026, "loss_mean50": 0.002047, "lr": 1.0324016894368957e-05, "grad_norm": 0.0277, "samples": 300160, "wall_s": 18610.0, "cuda_gb": 16.86} |
| {"step": 9400, "loss": 0.002856, "loss_mean50": 0.002049, "lr": 1.0275959837350957e-05, "grad_norm": 0.0317, "samples": 300800, "wall_s": 18648.9, "cuda_gb": 16.86} |
| {"step": 9420, "loss": 0.001669, "loss_mean50": 0.001984, "lr": 1.0227972754505468e-05, "grad_norm": 0.0273, "samples": 301440, "wall_s": 18687.8, "cuda_gb": 16.86} |
| {"step": 9440, "loss": 0.002827, "loss_mean50": 0.001951, "lr": 1.0180056612389072e-05, "grad_norm": 0.0269, "samples": 302080, "wall_s": 18726.6, "cuda_gb": 16.86} |
| {"step": 9460, "loss": 0.001631, "loss_mean50": 0.001922, "lr": 1.0132212376129465e-05, "grad_norm": 0.0225, "samples": 302720, "wall_s": 18765.5, "cuda_gb": 16.86} |
| {"step": 9480, "loss": 0.001567, "loss_mean50": 0.001891, "lr": 1.0084441009406028e-05, "grad_norm": 0.0227, "samples": 303360, "wall_s": 18804.3, "cuda_gb": 16.86} |
| {"step": 9500, "loss": 0.000954, "loss_mean50": 0.001795, "lr": 1.0036743474430393e-05, "grad_norm": 0.0201, "samples": 304000, "wall_s": 18843.3, "cuda_gb": 16.86} |
| {"step": 9520, "loss": 0.002419, "loss_mean50": 0.001819, "lr": 9.989120731927065e-06, "grad_norm": 0.0465, "samples": 304640, "wall_s": 18882.2, "cuda_gb": 16.86} |
| {"step": 9540, "loss": 0.001102, "loss_mean50": 0.001818, "lr": 9.941573741114099e-06, "grad_norm": 0.0326, "samples": 305280, "wall_s": 18921.1, "cuda_gb": 16.86} |
| {"step": 9560, "loss": 0.009296, "loss_mean50": 0.001971, "lr": 9.894103459683726e-06, "grad_norm": 0.062, "samples": 305920, "wall_s": 18960.1, "cuda_gb": 16.86} |
| {"step": 9580, "loss": 0.001528, "loss_mean50": 0.00188, "lr": 9.846710843783125e-06, "grad_norm": 0.0276, "samples": 306560, "wall_s": 18999.1, "cuda_gb": 16.86} |
| {"step": 9600, "loss": 0.000982, "loss_mean50": 0.002113, "lr": 9.799396847995121e-06, "grad_norm": 0.0198, "samples": 307200, "wall_s": 19038.0, "cuda_gb": 16.86} |
| {"step": 9620, "loss": 0.000774, "loss_mean50": 0.002113, "lr": 9.752162425318972e-06, "grad_norm": 0.0158, "samples": 307840, "wall_s": 19077.1, "cuda_gb": 16.86} |
| {"step": 9640, "loss": 0.001453, "loss_mean50": 0.001985, "lr": 9.705008527151176e-06, "grad_norm": 0.0304, "samples": 308480, "wall_s": 19116.1, "cuda_gb": 16.86} |
| {"step": 9660, "loss": 0.001734, "loss_mean50": 0.001946, "lr": 9.657936103266308e-06, "grad_norm": 0.0334, "samples": 309120, "wall_s": 19154.9, "cuda_gb": 16.86} |
| {"step": 9680, "loss": 0.002493, "loss_mean50": 0.001919, "lr": 9.61094610179788e-06, "grad_norm": 0.0373, "samples": 309760, "wall_s": 19193.8, "cuda_gb": 16.86} |
| {"step": 9700, "loss": 0.001169, "loss_mean50": 0.002005, "lr": 9.564039469219244e-06, "grad_norm": 0.0268, "samples": 310400, "wall_s": 19232.8, "cuda_gb": 16.86} |
| {"step": 9720, "loss": 0.00134, "loss_mean50": 0.001977, "lr": 9.517217150324556e-06, "grad_norm": 0.0189, "samples": 311040, "wall_s": 19271.7, "cuda_gb": 16.86} |
| {"step": 9740, "loss": 0.002203, "loss_mean50": 0.001853, "lr": 9.470480088209707e-06, "grad_norm": 0.0329, "samples": 311680, "wall_s": 19310.6, "cuda_gb": 16.86} |
| {"step": 9760, "loss": 0.001607, "loss_mean50": 0.001749, "lr": 9.423829224253347e-06, "grad_norm": 0.024, "samples": 312320, "wall_s": 19349.4, "cuda_gb": 16.86} |
| {"step": 9780, "loss": 0.00112, "loss_mean50": 0.001662, "lr": 9.377265498097926e-06, "grad_norm": 0.0251, "samples": 312960, "wall_s": 19388.4, "cuda_gb": 16.86} |
| {"step": 9800, "loss": 0.001556, "loss_mean50": 0.001813, "lr": 9.330789847630767e-06, "grad_norm": 0.03, "samples": 313600, "wall_s": 19427.5, "cuda_gb": 16.86} |
| {"step": 9820, "loss": 0.001341, "loss_mean50": 0.001871, "lr": 9.284403208965151e-06, "grad_norm": 0.0158, "samples": 314240, "wall_s": 19466.5, "cuda_gb": 16.86} |
| {"step": 9840, "loss": 0.002146, "loss_mean50": 0.002079, "lr": 9.238106516421504e-06, "grad_norm": 0.0269, "samples": 314880, "wall_s": 19505.5, "cuda_gb": 16.86} |
| {"step": 9860, "loss": 0.004569, "loss_mean50": 0.002159, "lr": 9.191900702508551e-06, "grad_norm": 0.032, "samples": 315520, "wall_s": 19544.4, "cuda_gb": 16.86} |
| {"step": 9880, "loss": 0.000743, "loss_mean50": 0.002117, "lr": 9.145786697904526e-06, "grad_norm": 0.022, "samples": 316160, "wall_s": 19583.4, "cuda_gb": 16.86} |
| {"step": 9900, "loss": 0.001945, "loss_mean50": 0.001967, "lr": 9.099765431438454e-06, "grad_norm": 0.0261, "samples": 316800, "wall_s": 19622.3, "cuda_gb": 16.86} |
| {"step": 9920, "loss": 0.001382, "loss_mean50": 0.001916, "lr": 9.053837830071418e-06, "grad_norm": 0.0208, "samples": 317440, "wall_s": 19661.3, "cuda_gb": 16.86} |
| {"step": 9940, "loss": 0.001685, "loss_mean50": 0.001926, "lr": 9.008004818877899e-06, "grad_norm": 0.026, "samples": 318080, "wall_s": 19700.2, "cuda_gb": 16.86} |
| {"step": 9960, "loss": 0.001116, "loss_mean50": 0.001923, "lr": 8.96226732102714e-06, "grad_norm": 0.0295, "samples": 318720, "wall_s": 19739.2, "cuda_gb": 16.86} |
| {"step": 9980, "loss": 0.001536, "loss_mean50": 0.001946, "lr": 8.91662625776456e-06, "grad_norm": 0.0224, "samples": 319360, "wall_s": 19778.3, "cuda_gb": 16.86} |
| {"step": 10000, "loss": 0.001441, "loss_mean50": 0.001916, "lr": 8.87108254839317e-06, "grad_norm": 0.0394, "samples": 320000, "wall_s": 19817.2, "cuda_gb": 16.86} |
| {"step": 10000, "val_flow_mse": 0.001662, "val_action_mae": 0.013208, "best": true} |
| {"step": 10020, "loss": 0.002259, "loss_mean50": 0.001859, "lr": 8.825637110255108e-06, "grad_norm": 0.0304, "samples": 320640, "wall_s": 19940.3, "cuda_gb": 16.86} |
| {"step": 10040, "loss": 0.002173, "loss_mean50": 0.001991, "lr": 8.78029085871311e-06, "grad_norm": 0.0321, "samples": 321280, "wall_s": 19979.2, "cuda_gb": 16.86} |
| {"step": 10060, "loss": 0.001108, "loss_mean50": 0.001792, "lr": 8.7350447071321e-06, "grad_norm": 0.0207, "samples": 321920, "wall_s": 20018.0, "cuda_gb": 16.86} |
| {"step": 10080, "loss": 0.001314, "loss_mean50": 0.001772, "lr": 8.68989956686079e-06, "grad_norm": 0.0196, "samples": 322560, "wall_s": 20056.8, "cuda_gb": 16.86} |
| {"step": 10100, "loss": 0.002752, "loss_mean50": 0.002056, "lr": 8.644856347213308e-06, "grad_norm": 0.0462, "samples": 323200, "wall_s": 20095.6, "cuda_gb": 16.86} |
| {"step": 10120, "loss": 0.000986, "loss_mean50": 0.00207, "lr": 8.59991595545091e-06, "grad_norm": 0.0233, "samples": 323840, "wall_s": 20134.6, "cuda_gb": 16.86} |
| {"step": 10140, "loss": 0.001685, "loss_mean50": 0.001916, "lr": 8.555079296763687e-06, "grad_norm": 0.0324, "samples": 324480, "wall_s": 20173.4, "cuda_gb": 16.86} |
| {"step": 10160, "loss": 0.001215, "loss_mean50": 0.001775, "lr": 8.510347274252333e-06, "grad_norm": 0.0299, "samples": 325120, "wall_s": 20212.3, "cuda_gb": 16.86} |
| {"step": 10180, "loss": 0.001533, "loss_mean50": 0.001892, "lr": 8.465720788909959e-06, "grad_norm": 0.0417, "samples": 325760, "wall_s": 20251.1, "cuda_gb": 16.86} |
| {"step": 10200, "loss": 0.001957, "loss_mean50": 0.002023, "lr": 8.421200739603955e-06, "grad_norm": 0.0522, "samples": 326400, "wall_s": 20290.1, "cuda_gb": 16.86} |
| {"step": 10220, "loss": 0.002084, "loss_mean50": 0.002018, "lr": 8.376788023057854e-06, "grad_norm": 0.0221, "samples": 327040, "wall_s": 20329.0, "cuda_gb": 16.86} |
| {"step": 10240, "loss": 0.00132, "loss_mean50": 0.001978, "lr": 8.33248353383331e-06, "grad_norm": 0.0458, "samples": 327680, "wall_s": 20367.9, "cuda_gb": 16.86} |
| {"step": 10260, "loss": 0.002151, "loss_mean50": 0.001641, "lr": 8.288288164312053e-06, "grad_norm": 0.0562, "samples": 328320, "wall_s": 20406.8, "cuda_gb": 16.86} |
| {"step": 10280, "loss": 0.000789, "loss_mean50": 0.001732, "lr": 8.244202804677917e-06, "grad_norm": 0.0146, "samples": 328960, "wall_s": 20446.2, "cuda_gb": 16.86} |
| {"step": 10300, "loss": 0.002478, "loss_mean50": 0.001843, "lr": 8.200228342898927e-06, "grad_norm": 0.0319, "samples": 329600, "wall_s": 20485.9, "cuda_gb": 16.86} |
| {"step": 10320, "loss": 0.001045, "loss_mean50": 0.001973, "lr": 8.1563656647094e-06, "grad_norm": 0.0277, "samples": 330240, "wall_s": 20524.7, "cuda_gb": 16.86} |
| {"step": 10340, "loss": 0.001428, "loss_mean50": 0.002006, "lr": 8.112615653592086e-06, "grad_norm": 0.0264, "samples": 330880, "wall_s": 20563.6, "cuda_gb": 16.86} |
| {"step": 10360, "loss": 0.001501, "loss_mean50": 0.001765, "lr": 8.06897919076042e-06, "grad_norm": 0.0305, "samples": 331520, "wall_s": 20602.4, "cuda_gb": 16.86} |
| {"step": 10380, "loss": 0.003594, "loss_mean50": 0.001816, "lr": 8.025457155140734e-06, "grad_norm": 0.04, "samples": 332160, "wall_s": 20641.3, "cuda_gb": 16.86} |
| {"step": 10400, "loss": 0.002374, "loss_mean50": 0.001809, "lr": 7.982050423354554e-06, "grad_norm": 0.0299, "samples": 332800, "wall_s": 20680.3, "cuda_gb": 16.86} |
| {"step": 10420, "loss": 0.001285, "loss_mean50": 0.001793, "lr": 7.938759869700983e-06, "grad_norm": 0.0223, "samples": 333440, "wall_s": 20719.1, "cuda_gb": 16.86} |
| {"step": 10440, "loss": 0.001644, "loss_mean50": 0.001687, "lr": 7.895586366139032e-06, "grad_norm": 0.0279, "samples": 334080, "wall_s": 20758.0, "cuda_gb": 16.86} |
| {"step": 10460, "loss": 0.002487, "loss_mean50": 0.001813, "lr": 7.852530782270106e-06, "grad_norm": 0.0321, "samples": 334720, "wall_s": 20796.9, "cuda_gb": 16.86} |
| {"step": 10480, "loss": 0.002126, "loss_mean50": 0.001975, "lr": 7.809593985320462e-06, "grad_norm": 0.0374, "samples": 335360, "wall_s": 20835.8, "cuda_gb": 16.86} |
| {"step": 10500, "loss": 0.001848, "loss_mean50": 0.001942, "lr": 7.766776840123753e-06, "grad_norm": 0.0341, "samples": 336000, "wall_s": 20874.6, "cuda_gb": 16.86} |
| {"step": 10520, "loss": 0.002184, "loss_mean50": 0.00188, "lr": 7.724080209103593e-06, "grad_norm": 0.0769, "samples": 336640, "wall_s": 20913.5, "cuda_gb": 16.86} |
| {"step": 10540, "loss": 0.001535, "loss_mean50": 0.001808, "lr": 7.681504952256218e-06, "grad_norm": 0.0269, "samples": 337280, "wall_s": 20952.5, "cuda_gb": 16.86} |
| {"step": 10560, "loss": 0.001085, "loss_mean50": 0.00183, "lr": 7.639051927133128e-06, "grad_norm": 0.0272, "samples": 337920, "wall_s": 20991.4, "cuda_gb": 16.86} |
| {"step": 10580, "loss": 0.002168, "loss_mean50": 0.001846, "lr": 7.596721988823835e-06, "grad_norm": 0.0349, "samples": 338560, "wall_s": 21030.3, "cuda_gb": 16.86} |
| {"step": 10600, "loss": 0.001589, "loss_mean50": 0.001692, "lr": 7.554515989938631e-06, "grad_norm": 0.0232, "samples": 339200, "wall_s": 21069.3, "cuda_gb": 16.86} |
| {"step": 10620, "loss": 0.00159, "loss_mean50": 0.001671, "lr": 7.512434780591415e-06, "grad_norm": 0.026, "samples": 339840, "wall_s": 21108.3, "cuda_gb": 16.86} |
| {"step": 10640, "loss": 0.001427, "loss_mean50": 0.001687, "lr": 7.470479208382591e-06, "grad_norm": 0.0278, "samples": 340480, "wall_s": 21147.2, "cuda_gb": 16.86} |
| {"step": 10660, "loss": 0.003649, "loss_mean50": 0.00188, "lr": 7.4286501183819604e-06, "grad_norm": 0.0464, "samples": 341120, "wall_s": 21186.1, "cuda_gb": 16.86} |
| {"step": 10680, "loss": 0.003739, "loss_mean50": 0.002154, "lr": 7.386948353111717e-06, "grad_norm": 0.0931, "samples": 341760, "wall_s": 21225.0, "cuda_gb": 16.86} |
| {"step": 10700, "loss": 0.002286, "loss_mean50": 0.00215, "lr": 7.345374752529491e-06, "grad_norm": 0.0361, "samples": 342400, "wall_s": 21263.9, "cuda_gb": 16.86} |
| {"step": 10720, "loss": 0.000926, "loss_mean50": 0.001868, "lr": 7.30393015401141e-06, "grad_norm": 0.014, "samples": 343040, "wall_s": 21302.9, "cuda_gb": 16.86} |
| {"step": 10740, "loss": 0.002456, "loss_mean50": 0.001959, "lr": 7.2626153923352316e-06, "grad_norm": 0.0511, "samples": 343680, "wall_s": 21341.8, "cuda_gb": 16.86} |
| {"step": 10760, "loss": 0.002636, "loss_mean50": 0.001999, "lr": 7.221431299663555e-06, "grad_norm": 0.064, "samples": 344320, "wall_s": 21380.7, "cuda_gb": 16.86} |
| {"step": 10780, "loss": 0.002091, "loss_mean50": 0.001994, "lr": 7.18037870552703e-06, "grad_norm": 0.0393, "samples": 344960, "wall_s": 21419.6, "cuda_gb": 16.86} |
| {"step": 10800, "loss": 0.001623, "loss_mean50": 0.001852, "lr": 7.139458436807661e-06, "grad_norm": 0.0232, "samples": 345600, "wall_s": 21458.6, "cuda_gb": 16.86} |
| {"step": 10820, "loss": 0.00217, "loss_mean50": 0.002015, "lr": 7.098671317722161e-06, "grad_norm": 0.106, "samples": 346240, "wall_s": 21497.5, "cuda_gb": 16.86} |
| {"step": 10840, "loss": 0.003289, "loss_mean50": 0.002117, "lr": 7.05801816980533e-06, "grad_norm": 0.0331, "samples": 346880, "wall_s": 21536.5, "cuda_gb": 16.86} |
| {"step": 10860, "loss": 0.001094, "loss_mean50": 0.002008, "lr": 7.0174998118935295e-06, "grad_norm": 0.0215, "samples": 347520, "wall_s": 21575.5, "cuda_gb": 16.86} |
| {"step": 10880, "loss": 0.003622, "loss_mean50": 0.001772, "lr": 6.977117060108165e-06, "grad_norm": 0.0751, "samples": 348160, "wall_s": 21614.5, "cuda_gb": 16.86} |
| {"step": 10900, "loss": 0.000713, "loss_mean50": 0.001814, "lr": 6.9368707278392706e-06, "grad_norm": 0.0198, "samples": 348800, "wall_s": 21653.5, "cuda_gb": 16.86} |
| {"step": 10920, "loss": 0.002231, "loss_mean50": 0.00198, "lr": 6.896761625729117e-06, "grad_norm": 0.0224, "samples": 349440, "wall_s": 21692.5, "cuda_gb": 16.86} |
| {"step": 10940, "loss": 0.00138, "loss_mean50": 0.001763, "lr": 6.856790561655879e-06, "grad_norm": 0.0323, "samples": 350080, "wall_s": 21731.5, "cuda_gb": 16.86} |
| {"step": 10960, "loss": 0.003329, "loss_mean50": 0.001789, "lr": 6.816958340717365e-06, "grad_norm": 0.1167, "samples": 350720, "wall_s": 21770.3, "cuda_gb": 16.86} |
| {"step": 10980, "loss": 0.0015, "loss_mean50": 0.001668, "lr": 6.777265765214818e-06, "grad_norm": 0.063, "samples": 351360, "wall_s": 21809.1, "cuda_gb": 16.86} |
| {"step": 11000, "loss": 0.002448, "loss_mean50": 0.001786, "lr": 6.737713634636714e-06, "grad_norm": 0.0478, "samples": 352000, "wall_s": 21848.0, "cuda_gb": 16.86} |
| {"step": 11000, "val_flow_mse": 0.001598} |
| {"step": 11020, "loss": 0.00232, "loss_mean50": 0.001811, "lr": 6.698302745642702e-06, "grad_norm": 0.0222, "samples": 352640, "wall_s": 21915.8, "cuda_gb": 16.86} |
| {"step": 11040, "loss": 0.002247, "loss_mean50": 0.001859, "lr": 6.659033892047546e-06, "grad_norm": 0.0393, "samples": 353280, "wall_s": 21954.8, "cuda_gb": 16.86} |
| {"step": 11060, "loss": 0.00123, "loss_mean50": 0.001741, "lr": 6.619907864805116e-06, "grad_norm": 0.0346, "samples": 353920, "wall_s": 21993.7, "cuda_gb": 16.86} |
| {"step": 11080, "loss": 0.002285, "loss_mean50": 0.001796, "lr": 6.580925451992481e-06, "grad_norm": 0.0482, "samples": 354560, "wall_s": 22032.6, "cuda_gb": 16.86} |
| {"step": 11100, "loss": 0.001579, "loss_mean50": 0.001978, "lr": 6.542087438794027e-06, "grad_norm": 0.021, "samples": 355200, "wall_s": 22071.4, "cuda_gb": 16.86} |
| {"step": 11120, "loss": 0.001737, "loss_mean50": 0.001845, "lr": 6.503394607485626e-06, "grad_norm": 0.0209, "samples": 355840, "wall_s": 22110.4, "cuda_gb": 16.86} |
| {"step": 11140, "loss": 0.001058, "loss_mean50": 0.001762, "lr": 6.464847737418916e-06, "grad_norm": 0.0229, "samples": 356480, "wall_s": 22149.3, "cuda_gb": 16.86} |
| {"step": 11160, "loss": 0.003287, "loss_mean50": 0.001849, "lr": 6.426447605005567e-06, "grad_norm": 0.0407, "samples": 357120, "wall_s": 22188.2, "cuda_gb": 16.86} |
| {"step": 11180, "loss": 0.003345, "loss_mean50": 0.00202, "lr": 6.388194983701658e-06, "grad_norm": 0.0814, "samples": 357760, "wall_s": 22227.1, "cuda_gb": 16.86} |
| {"step": 11200, "loss": 0.002595, "loss_mean50": 0.001924, "lr": 6.350090643992096e-06, "grad_norm": 0.0353, "samples": 358400, "wall_s": 22266.0, "cuda_gb": 16.86} |
| {"step": 11220, "loss": 0.002528, "loss_mean50": 0.00191, "lr": 6.312135353375108e-06, "grad_norm": 0.041, "samples": 359040, "wall_s": 22304.8, "cuda_gb": 16.86} |
| {"step": 11240, "loss": 0.001931, "loss_mean50": 0.001852, "lr": 6.274329876346763e-06, "grad_norm": 0.0309, "samples": 359680, "wall_s": 22343.8, "cuda_gb": 16.86} |
| {"step": 11260, "loss": 0.001626, "loss_mean50": 0.00195, "lr": 6.2366749743855796e-06, "grad_norm": 0.021, "samples": 360320, "wall_s": 22382.7, "cuda_gb": 16.86} |
| {"step": 11280, "loss": 0.002153, "loss_mean50": 0.001852, "lr": 6.199171405937199e-06, "grad_norm": 0.0438, "samples": 360960, "wall_s": 22421.6, "cuda_gb": 16.86} |
| {"step": 11300, "loss": 0.002178, "loss_mean50": 0.001672, "lr": 6.161819926399091e-06, "grad_norm": 0.0255, "samples": 361600, "wall_s": 22460.5, "cuda_gb": 16.86} |
| {"step": 11320, "loss": 0.001291, "loss_mean50": 0.00173, "lr": 6.124621288105361e-06, "grad_norm": 0.0185, "samples": 362240, "wall_s": 22499.4, "cuda_gb": 16.86} |
| {"step": 11340, "loss": 0.00184, "loss_mean50": 0.001728, "lr": 6.0875762403115745e-06, "grad_norm": 0.028, "samples": 362880, "wall_s": 22538.2, "cuda_gb": 16.86} |
| {"step": 11360, "loss": 0.001678, "loss_mean50": 0.001622, "lr": 6.050685529179673e-06, "grad_norm": 0.0269, "samples": 363520, "wall_s": 22577.1, "cuda_gb": 16.86} |
| {"step": 11380, "loss": 0.001176, "loss_mean50": 0.001625, "lr": 6.013949897762954e-06, "grad_norm": 0.0229, "samples": 364160, "wall_s": 22616.0, "cuda_gb": 16.86} |
| {"step": 11400, "loss": 0.0016, "loss_mean50": 0.001615, "lr": 5.977370085991092e-06, "grad_norm": 0.0357, "samples": 364800, "wall_s": 22654.8, "cuda_gb": 16.86} |
| {"step": 11420, "loss": 0.001675, "loss_mean50": 0.001646, "lr": 5.9409468306552325e-06, "grad_norm": 0.0337, "samples": 365440, "wall_s": 22693.7, "cuda_gb": 16.86} |
| {"step": 11440, "loss": 0.005054, "loss_mean50": 0.001935, "lr": 5.904680865393181e-06, "grad_norm": 0.0507, "samples": 366080, "wall_s": 22732.6, "cuda_gb": 16.86} |
| {"step": 11460, "loss": 0.000783, "loss_mean50": 0.001821, "lr": 5.868572920674581e-06, "grad_norm": 0.0124, "samples": 366720, "wall_s": 22771.5, "cuda_gb": 16.86} |
| {"step": 11480, "loss": 0.001596, "loss_mean50": 0.001803, "lr": 5.83262372378623e-06, "grad_norm": 0.0216, "samples": 367360, "wall_s": 22810.4, "cuda_gb": 16.86} |
| {"step": 11500, "loss": 0.002244, "loss_mean50": 0.001768, "lr": 5.7968339988174385e-06, "grad_norm": 0.0343, "samples": 368000, "wall_s": 22850.1, "cuda_gb": 16.86} |
| {"step": 11520, "loss": 0.001379, "loss_mean50": 0.001948, "lr": 5.7612044666454034e-06, "grad_norm": 0.0223, "samples": 368640, "wall_s": 22888.9, "cuda_gb": 16.86} |
| {"step": 11540, "loss": 0.001287, "loss_mean50": 0.001849, "lr": 5.7257358449207435e-06, "grad_norm": 0.0221, "samples": 369280, "wall_s": 22927.8, "cuda_gb": 16.86} |
| {"step": 11560, "loss": 0.001522, "loss_mean50": 0.001798, "lr": 5.690428848053e-06, "grad_norm": 0.0264, "samples": 369920, "wall_s": 22966.7, "cuda_gb": 16.86} |
| {"step": 11580, "loss": 0.001498, "loss_mean50": 0.001669, "lr": 5.655284187196265e-06, "grad_norm": 0.0262, "samples": 370560, "wall_s": 23005.6, "cuda_gb": 16.86} |
| {"step": 11600, "loss": 0.001934, "loss_mean50": 0.001692, "lr": 5.620302570234867e-06, "grad_norm": 0.0398, "samples": 371200, "wall_s": 23044.6, "cuda_gb": 16.86} |
| {"step": 11620, "loss": 0.001742, "loss_mean50": 0.001847, "lr": 5.5854847017690945e-06, "grad_norm": 0.046, "samples": 371840, "wall_s": 23083.5, "cuda_gb": 16.86} |
| {"step": 11640, "loss": 0.001709, "loss_mean50": 0.001936, "lr": 5.550831283100999e-06, "grad_norm": 0.0288, "samples": 372480, "wall_s": 23122.5, "cuda_gb": 16.86} |
| {"step": 11660, "loss": 0.002371, "loss_mean50": 0.001793, "lr": 5.516343012220304e-06, "grad_norm": 0.0728, "samples": 373120, "wall_s": 23161.5, "cuda_gb": 16.86} |
| {"step": 11680, "loss": 0.001888, "loss_mean50": 0.001895, "lr": 5.482020583790311e-06, "grad_norm": 0.036, "samples": 373760, "wall_s": 23200.3, "cuda_gb": 16.86} |
| {"step": 11700, "loss": 0.00136, "loss_mean50": 0.001881, "lr": 5.447864689133917e-06, "grad_norm": 0.0231, "samples": 374400, "wall_s": 23239.3, "cuda_gb": 16.86} |
| {"step": 11720, "loss": 0.001899, "loss_mean50": 0.001785, "lr": 5.413876016219707e-06, "grad_norm": 0.0335, "samples": 375040, "wall_s": 23278.2, "cuda_gb": 16.86} |
| {"step": 11740, "loss": 0.001848, "loss_mean50": 0.001626, "lr": 5.380055249648071e-06, "grad_norm": 0.0351, "samples": 375680, "wall_s": 23317.2, "cuda_gb": 16.86} |
| {"step": 11760, "loss": 0.001892, "loss_mean50": 0.001569, "lr": 5.346403070637427e-06, "grad_norm": 0.0243, "samples": 376320, "wall_s": 23356.1, "cuda_gb": 16.86} |
| {"step": 11780, "loss": 0.000761, "loss_mean50": 0.001611, "lr": 5.312920157010506e-06, "grad_norm": 0.0178, "samples": 376960, "wall_s": 23395.0, "cuda_gb": 16.86} |
| {"step": 11800, "loss": 0.001502, "loss_mean50": 0.001586, "lr": 5.2796071831806846e-06, "grad_norm": 0.0175, "samples": 377600, "wall_s": 23433.9, "cuda_gb": 16.86} |
| {"step": 11820, "loss": 0.000852, "loss_mean50": 0.001648, "lr": 5.2464648201384225e-06, "grad_norm": 0.0182, "samples": 378240, "wall_s": 23472.8, "cuda_gb": 16.86} |
| {"step": 11840, "loss": 0.001569, "loss_mean50": 0.001693, "lr": 5.213493735437722e-06, "grad_norm": 0.0282, "samples": 378880, "wall_s": 23511.7, "cuda_gb": 16.86} |
| {"step": 11860, "loss": 0.002044, "loss_mean50": 0.001804, "lr": 5.180694593182696e-06, "grad_norm": 0.0365, "samples": 379520, "wall_s": 23550.6, "cuda_gb": 16.86} |
| {"step": 11880, "loss": 0.001418, "loss_mean50": 0.001831, "lr": 5.148068054014195e-06, "grad_norm": 0.0533, "samples": 380160, "wall_s": 23589.4, "cuda_gb": 16.86} |
| {"step": 11900, "loss": 0.004738, "loss_mean50": 0.001893, "lr": 5.115614775096492e-06, "grad_norm": 0.0343, "samples": 380800, "wall_s": 23628.4, "cuda_gb": 16.86} |
| {"step": 11920, "loss": 0.001811, "loss_mean50": 0.002031, "lr": 5.083335410104047e-06, "grad_norm": 0.0286, "samples": 381440, "wall_s": 23667.3, "cuda_gb": 16.86} |
| {"step": 11940, "loss": 0.002899, "loss_mean50": 0.001923, "lr": 5.051230609208352e-06, "grad_norm": 0.0671, "samples": 382080, "wall_s": 23706.2, "cuda_gb": 16.86} |
| {"step": 11960, "loss": 0.0015, "loss_mean50": 0.001614, "lr": 5.019301019064816e-06, "grad_norm": 0.0265, "samples": 382720, "wall_s": 23745.3, "cuda_gb": 16.86} |
| {"step": 11980, "loss": 0.000773, "loss_mean50": 0.001737, "lr": 4.987547282799753e-06, "grad_norm": 0.0164, "samples": 383360, "wall_s": 23784.3, "cuda_gb": 16.86} |
| {"step": 12000, "loss": 0.001718, "loss_mean50": 0.001921, "lr": 4.955970039997429e-06, "grad_norm": 0.0337, "samples": 384000, "wall_s": 23823.3, "cuda_gb": 16.86} |
| {"step": 12000, "val_flow_mse": 0.001532} |
| {"step": 12020, "loss": 0.002013, "loss_mean50": 0.001863, "lr": 4.924569926687178e-06, "grad_norm": 0.0398, "samples": 384640, "wall_s": 23891.2, "cuda_gb": 16.86} |
| {"step": 12040, "loss": 0.003023, "loss_mean50": 0.001644, "lr": 4.893347575330571e-06, "grad_norm": 0.0389, "samples": 385280, "wall_s": 23930.1, "cuda_gb": 16.86} |
| {"step": 12060, "loss": 0.002547, "loss_mean50": 0.001652, "lr": 4.862303614808722e-06, "grad_norm": 0.045, "samples": 385920, "wall_s": 23969.1, "cuda_gb": 16.86} |
| {"step": 12080, "loss": 0.003241, "loss_mean50": 0.001606, "lr": 4.831438670409574e-06, "grad_norm": 0.0336, "samples": 386560, "wall_s": 24008.0, "cuda_gb": 16.86} |
| {"step": 12100, "loss": 0.001498, "loss_mean50": 0.001676, "lr": 4.800753363815327e-06, "grad_norm": 0.0202, "samples": 387200, "wall_s": 24047.0, "cuda_gb": 16.86} |
| {"step": 12120, "loss": 0.001814, "loss_mean50": 0.0018, "lr": 4.770248313089926e-06, "grad_norm": 0.0282, "samples": 387840, "wall_s": 24085.9, "cuda_gb": 16.86} |
| {"step": 12140, "loss": 0.002635, "loss_mean50": 0.001631, "lr": 4.739924132666585e-06, "grad_norm": 0.0274, "samples": 388480, "wall_s": 24124.7, "cuda_gb": 16.86} |
| {"step": 12160, "loss": 0.000834, "loss_mean50": 0.001651, "lr": 4.709781433335432e-06, "grad_norm": 0.0186, "samples": 389120, "wall_s": 24163.6, "cuda_gb": 16.86} |
| {"step": 12180, "loss": 0.001318, "loss_mean50": 0.001543, "lr": 4.6798208222311985e-06, "grad_norm": 0.0225, "samples": 389760, "wall_s": 24202.7, "cuda_gb": 16.86} |
| {"step": 12200, "loss": 0.000694, "loss_mean50": 0.001586, "lr": 4.6500429028209904e-06, "grad_norm": 0.0153, "samples": 390400, "wall_s": 24241.7, "cuda_gb": 16.86} |
| {"step": 12220, "loss": 0.000514, "loss_mean50": 0.001524, "lr": 4.620448274892142e-06, "grad_norm": 0.015, "samples": 391040, "wall_s": 24280.6, "cuda_gb": 16.86} |
| {"step": 12240, "loss": 0.000956, "loss_mean50": 0.001476, "lr": 4.591037534540117e-06, "grad_norm": 0.0198, "samples": 391680, "wall_s": 24319.5, "cuda_gb": 16.86} |
| {"step": 12260, "loss": 0.001926, "loss_mean50": 0.001506, "lr": 4.56181127415652e-06, "grad_norm": 0.0281, "samples": 392320, "wall_s": 24358.5, "cuda_gb": 16.86} |
| {"step": 12280, "loss": 0.002858, "loss_mean50": 0.001518, "lr": 4.5327700824171575e-06, "grad_norm": 0.0493, "samples": 392960, "wall_s": 24397.5, "cuda_gb": 16.86} |
| {"step": 12300, "loss": 0.001911, "loss_mean50": 0.001465, "lr": 4.5039145442701735e-06, "grad_norm": 0.0228, "samples": 393600, "wall_s": 24436.3, "cuda_gb": 16.86} |
| {"step": 12320, "loss": 0.001791, "loss_mean50": 0.00157, "lr": 4.475245240924278e-06, "grad_norm": 0.0231, "samples": 394240, "wall_s": 24475.3, "cuda_gb": 16.86} |
| {"step": 12340, "loss": 0.001026, "loss_mean50": 0.00157, "lr": 4.446762749837041e-06, "grad_norm": 0.0161, "samples": 394880, "wall_s": 24514.4, "cuda_gb": 16.86} |
| {"step": 12360, "loss": 0.001512, "loss_mean50": 0.001685, "lr": 4.418467644703254e-06, "grad_norm": 0.0238, "samples": 395520, "wall_s": 24553.4, "cuda_gb": 16.86} |
| {"step": 12380, "loss": 0.001721, "loss_mean50": 0.001713, "lr": 4.3903604954433725e-06, "grad_norm": 0.0247, "samples": 396160, "wall_s": 24592.5, "cuda_gb": 16.86} |
| {"step": 12400, "loss": 0.000787, "loss_mean50": 0.00163, "lr": 4.362441868192056e-06, "grad_norm": 0.0162, "samples": 396800, "wall_s": 24631.5, "cuda_gb": 16.86} |
| {"step": 12420, "loss": 0.000801, "loss_mean50": 0.001702, "lr": 4.334712325286726e-06, "grad_norm": 0.0217, "samples": 397440, "wall_s": 24670.4, "cuda_gb": 16.86} |
| {"step": 12440, "loss": 0.00112, "loss_mean50": 0.001606, "lr": 4.307172425256291e-06, "grad_norm": 0.0358, "samples": 398080, "wall_s": 24709.3, "cuda_gb": 16.86} |
| {"step": 12460, "loss": 0.001305, "loss_mean50": 0.001445, "lr": 4.279822722809855e-06, "grad_norm": 0.0257, "samples": 398720, "wall_s": 24748.2, "cuda_gb": 16.86} |
| {"step": 12480, "loss": 0.001924, "loss_mean50": 0.001441, "lr": 4.25266376882556e-06, "grad_norm": 0.0335, "samples": 399360, "wall_s": 24787.1, "cuda_gb": 16.86} |
| {"step": 12500, "loss": 0.001329, "loss_mean50": 0.00164, "lr": 4.2256961103394905e-06, "grad_norm": 0.0173, "samples": 400000, "wall_s": 24826.0, "cuda_gb": 16.86} |
| {"step": 12520, "loss": 0.002313, "loss_mean50": 0.001629, "lr": 4.198920290534661e-06, "grad_norm": 0.0601, "samples": 400640, "wall_s": 24900.2, "cuda_gb": 16.86} |
| {"step": 12540, "loss": 0.001371, "loss_mean50": 0.001604, "lr": 4.1723368487300514e-06, "grad_norm": 0.019, "samples": 401280, "wall_s": 24939.2, "cuda_gb": 16.86} |
| {"step": 12560, "loss": 0.001485, "loss_mean50": 0.00161, "lr": 4.145946320369784e-06, "grad_norm": 0.0416, "samples": 401920, "wall_s": 24978.2, "cuda_gb": 16.86} |
| {"step": 12580, "loss": 0.002745, "loss_mean50": 0.001617, "lr": 4.119749237012297e-06, "grad_norm": 0.0425, "samples": 402560, "wall_s": 25017.1, "cuda_gb": 16.86} |
| {"step": 12600, "loss": 0.001831, "loss_mean50": 0.001653, "lr": 4.093746126319657e-06, "grad_norm": 0.0358, "samples": 403200, "wall_s": 25056.0, "cuda_gb": 16.86} |
| {"step": 12620, "loss": 0.002688, "loss_mean50": 0.001682, "lr": 4.067937512046945e-06, "grad_norm": 0.0362, "samples": 403840, "wall_s": 25095.1, "cuda_gb": 16.86} |
| {"step": 12640, "loss": 0.001517, "loss_mean50": 0.001642, "lr": 4.042323914031678e-06, "grad_norm": 0.0216, "samples": 404480, "wall_s": 25134.1, "cuda_gb": 16.86} |
| {"step": 12660, "loss": 0.001031, "loss_mean50": 0.001549, "lr": 4.016905848183357e-06, "grad_norm": 0.0212, "samples": 405120, "wall_s": 25173.0, "cuda_gb": 16.86} |
| {"step": 12680, "loss": 0.003315, "loss_mean50": 0.00153, "lr": 3.991683826473071e-06, "grad_norm": 0.0403, "samples": 405760, "wall_s": 25212.0, "cuda_gb": 16.86} |
| {"step": 12700, "loss": 0.001284, "loss_mean50": 0.001645, "lr": 3.9666583569231835e-06, "grad_norm": 0.0331, "samples": 406400, "wall_s": 25250.9, "cuda_gb": 16.86} |
| {"step": 12720, "loss": 0.002404, "loss_mean50": 0.001856, "lr": 3.9418299435971e-06, "grad_norm": 0.0309, "samples": 407040, "wall_s": 25289.8, "cuda_gb": 16.86} |
| {"step": 12740, "loss": 0.003183, "loss_mean50": 0.00182, "lr": 3.9171990865891265e-06, "grad_norm": 0.0863, "samples": 407680, "wall_s": 25328.7, "cuda_gb": 16.86} |
| {"step": 12760, "loss": 0.000687, "loss_mean50": 0.001773, "lr": 3.892766282014372e-06, "grad_norm": 0.013, "samples": 408320, "wall_s": 25367.6, "cuda_gb": 16.86} |
| {"step": 12780, "loss": 0.00092, "loss_mean50": 0.00164, "lr": 3.868532021998779e-06, "grad_norm": 0.0165, "samples": 408960, "wall_s": 25406.5, "cuda_gb": 16.86} |
| {"step": 12800, "loss": 0.001129, "loss_mean50": 0.001616, "lr": 3.844496794669201e-06, "grad_norm": 0.0285, "samples": 409600, "wall_s": 25445.6, "cuda_gb": 16.86} |
| {"step": 12820, "loss": 0.000939, "loss_mean50": 0.001512, "lr": 3.820661084143568e-06, "grad_norm": 0.0172, "samples": 410240, "wall_s": 25484.6, "cuda_gb": 16.86} |
| {"step": 12840, "loss": 0.001394, "loss_mean50": 0.001616, "lr": 3.797025370521147e-06, "grad_norm": 0.0266, "samples": 410880, "wall_s": 25523.6, "cuda_gb": 16.86} |
| {"step": 12860, "loss": 0.003553, "loss_mean50": 0.001784, "lr": 3.7735901298728575e-06, "grad_norm": 0.0599, "samples": 411520, "wall_s": 25562.6, "cuda_gb": 16.86} |
| {"step": 12880, "loss": 0.001051, "loss_mean50": 0.00163, "lr": 3.7503558342316883e-06, "grad_norm": 0.0193, "samples": 412160, "wall_s": 25601.5, "cuda_gb": 16.86} |
| {"step": 12900, "loss": 0.001435, "loss_mean50": 0.001584, "lr": 3.7273229515831993e-06, "grad_norm": 0.0275, "samples": 412800, "wall_s": 25640.4, "cuda_gb": 16.86} |
| {"step": 12920, "loss": 0.002297, "loss_mean50": 0.001701, "lr": 3.704491945856082e-06, "grad_norm": 0.0251, "samples": 413440, "wall_s": 25679.4, "cuda_gb": 16.86} |
| {"step": 12940, "loss": 0.002076, "loss_mean50": 0.001719, "lr": 3.6818632769128106e-06, "grad_norm": 0.0249, "samples": 414080, "wall_s": 25718.2, "cuda_gb": 16.86} |
| {"step": 12960, "loss": 0.001726, "loss_mean50": 0.001646, "lr": 3.6594374005404026e-06, "grad_norm": 0.028, "samples": 414720, "wall_s": 25757.2, "cuda_gb": 16.86} |
| {"step": 12980, "loss": 0.001089, "loss_mean50": 0.001635, "lr": 3.637214768441219e-06, "grad_norm": 0.0261, "samples": 415360, "wall_s": 25796.1, "cuda_gb": 16.86} |
| {"step": 13000, "loss": 0.00213, "loss_mean50": 0.001631, "lr": 3.6151958282238664e-06, "grad_norm": 0.0327, "samples": 416000, "wall_s": 25835.1, "cuda_gb": 16.86} |
| {"step": 13000, "val_flow_mse": 0.00146} |
| {"step": 13020, "loss": 0.000896, "loss_mean50": 0.001598, "lr": 3.593381023394196e-06, "grad_norm": 0.0253, "samples": 416640, "wall_s": 25902.8, "cuda_gb": 16.86} |
| {"step": 13040, "loss": 0.001202, "loss_mean50": 0.001687, "lr": 3.571770793346353e-06, "grad_norm": 0.0272, "samples": 417280, "wall_s": 25941.9, "cuda_gb": 16.86} |
| {"step": 13060, "loss": 0.001803, "loss_mean50": 0.001724, "lr": 3.5503655733539372e-06, "grad_norm": 0.0208, "samples": 417920, "wall_s": 25980.8, "cuda_gb": 16.86} |
| {"step": 13080, "loss": 0.001173, "loss_mean50": 0.001645, "lr": 3.5291657945612303e-06, "grad_norm": 0.0218, "samples": 418560, "wall_s": 26019.7, "cuda_gb": 16.86} |
| {"step": 13100, "loss": 0.001278, "loss_mean50": 0.001655, "lr": 3.5081718839745164e-06, "grad_norm": 0.029, "samples": 419200, "wall_s": 26058.8, "cuda_gb": 16.86} |
| {"step": 13120, "loss": 0.001631, "loss_mean50": 0.001546, "lr": 3.4873842644534798e-06, "grad_norm": 0.0348, "samples": 419840, "wall_s": 26097.7, "cuda_gb": 16.86} |
| {"step": 13140, "loss": 0.000493, "loss_mean50": 0.001511, "lr": 3.466803354702685e-06, "grad_norm": 0.0116, "samples": 420480, "wall_s": 26136.6, "cuda_gb": 16.86} |
| {"step": 13160, "loss": 0.00219, "loss_mean50": 0.001653, "lr": 3.4464295692631432e-06, "grad_norm": 0.035, "samples": 421120, "wall_s": 26175.5, "cuda_gb": 16.86} |
| {"step": 13180, "loss": 0.001067, "loss_mean50": 0.001595, "lr": 3.4262633185039696e-06, "grad_norm": 0.0224, "samples": 421760, "wall_s": 26214.5, "cuda_gb": 16.86} |
| {"step": 13200, "loss": 0.000584, "loss_mean50": 0.001633, "lr": 3.406305008614106e-06, "grad_norm": 0.0152, "samples": 422400, "wall_s": 26253.3, "cuda_gb": 16.86} |
| {"step": 13220, "loss": 0.00128, "loss_mean50": 0.00159, "lr": 3.3865550415941497e-06, "grad_norm": 0.0352, "samples": 423040, "wall_s": 26292.3, "cuda_gb": 16.86} |
| {"step": 13240, "loss": 0.002589, "loss_mean50": 0.001617, "lr": 3.367013815248255e-06, "grad_norm": 0.0265, "samples": 423680, "wall_s": 26331.2, "cuda_gb": 16.86} |
| {"step": 13260, "loss": 0.000887, "loss_mean50": 0.001536, "lr": 3.347681723176115e-06, "grad_norm": 0.0327, "samples": 424320, "wall_s": 26370.3, "cuda_gb": 16.86} |
| {"step": 13280, "loss": 0.00122, "loss_mean50": 0.001598, "lr": 3.3285591547650326e-06, "grad_norm": 0.0328, "samples": 424960, "wall_s": 26409.2, "cuda_gb": 16.86} |
| {"step": 13300, "loss": 0.001268, "loss_mean50": 0.001479, "lr": 3.3096464951820954e-06, "grad_norm": 0.0249, "samples": 425600, "wall_s": 26448.1, "cuda_gb": 16.86} |
| {"step": 13320, "loss": 0.001176, "loss_mean50": 0.001473, "lr": 3.290944125366388e-06, "grad_norm": 0.0176, "samples": 426240, "wall_s": 26487.0, "cuda_gb": 16.86} |
| {"step": 13340, "loss": 0.004432, "loss_mean50": 0.001382, "lr": 3.272452422021341e-06, "grad_norm": 0.0384, "samples": 426880, "wall_s": 26525.8, "cuda_gb": 16.86} |
| {"step": 13360, "loss": 0.000944, "loss_mean50": 0.001431, "lr": 3.254171757607145e-06, "grad_norm": 0.0182, "samples": 427520, "wall_s": 26564.7, "cuda_gb": 16.86} |
| {"step": 13380, "loss": 0.001329, "loss_mean50": 0.001516, "lr": 3.236102500333231e-06, "grad_norm": 0.0175, "samples": 428160, "wall_s": 26603.7, "cuda_gb": 16.86} |
| {"step": 13400, "loss": 0.001258, "loss_mean50": 0.001451, "lr": 3.2182450141508667e-06, "grad_norm": 0.0187, "samples": 428800, "wall_s": 26642.6, "cuda_gb": 16.86} |
| {"step": 13420, "loss": 0.001508, "loss_mean50": 0.00145, "lr": 3.2005996587458287e-06, "grad_norm": 0.0471, "samples": 429440, "wall_s": 26681.5, "cuda_gb": 16.86} |
| {"step": 13440, "loss": 0.001478, "loss_mean50": 0.001458, "lr": 3.183166789531146e-06, "grad_norm": 0.03, "samples": 430080, "wall_s": 26720.3, "cuda_gb": 16.86} |
| {"step": 13460, "loss": 0.001455, "loss_mean50": 0.001507, "lr": 3.1659467576399488e-06, "grad_norm": 0.0449, "samples": 430720, "wall_s": 26759.2, "cuda_gb": 16.86} |
| {"step": 13480, "loss": 0.001488, "loss_mean50": 0.001501, "lr": 3.1489399099183936e-06, "grad_norm": 0.0447, "samples": 431360, "wall_s": 26798.0, "cuda_gb": 16.86} |
| {"step": 13500, "loss": 0.001848, "loss_mean50": 0.001658, "lr": 3.132146588918679e-06, "grad_norm": 0.0204, "samples": 432000, "wall_s": 26836.9, "cuda_gb": 16.86} |
| {"step": 13520, "loss": 0.001518, "loss_mean50": 0.001628, "lr": 3.1155671328921473e-06, "grad_norm": 0.0186, "samples": 432640, "wall_s": 26875.9, "cuda_gb": 16.86} |
| {"step": 13540, "loss": 0.001745, "loss_mean50": 0.001527, "lr": 3.0992018757824658e-06, "grad_norm": 0.0274, "samples": 433280, "wall_s": 26914.8, "cuda_gb": 16.86} |
| {"step": 13560, "loss": 0.001156, "loss_mean50": 0.001462, "lr": 3.0830511472189056e-06, "grad_norm": 0.0319, "samples": 433920, "wall_s": 26953.7, "cuda_gb": 16.86} |
| {"step": 13580, "loss": 0.00219, "loss_mean50": 0.00149, "lr": 3.0671152725096976e-06, "grad_norm": 0.0323, "samples": 434560, "wall_s": 26992.7, "cuda_gb": 16.86} |
| {"step": 13600, "loss": 0.001828, "loss_mean50": 0.001449, "lr": 3.0513945726354877e-06, "grad_norm": 0.0435, "samples": 435200, "wall_s": 27031.5, "cuda_gb": 16.86} |
| {"step": 13620, "loss": 0.000496, "loss_mean50": 0.001462, "lr": 3.0358893642428604e-06, "grad_norm": 0.0161, "samples": 435840, "wall_s": 27070.5, "cuda_gb": 16.86} |
| {"step": 13640, "loss": 0.001424, "loss_mean50": 0.001417, "lr": 3.020599959637975e-06, "grad_norm": 0.0221, "samples": 436480, "wall_s": 27109.4, "cuda_gb": 16.86} |
| {"step": 13660, "loss": 0.001088, "loss_mean50": 0.001399, "lr": 3.005526666780261e-06, "grad_norm": 0.0183, "samples": 437120, "wall_s": 27148.4, "cuda_gb": 16.86} |
| {"step": 13680, "loss": 0.002135, "loss_mean50": 0.001505, "lr": 2.990669789276225e-06, "grad_norm": 0.0277, "samples": 437760, "wall_s": 27187.3, "cuda_gb": 16.86} |
| {"step": 13700, "loss": 0.00198, "loss_mean50": 0.001602, "lr": 2.9760296263733332e-06, "grad_norm": 0.0485, "samples": 438400, "wall_s": 27226.4, "cuda_gb": 16.86} |
| {"step": 13720, "loss": 0.001598, "loss_mean50": 0.001607, "lr": 2.961606472953977e-06, "grad_norm": 0.0276, "samples": 439040, "wall_s": 27265.4, "cuda_gb": 16.86} |
| {"step": 13740, "loss": 0.002248, "loss_mean50": 0.001639, "lr": 2.947400619529546e-06, "grad_norm": 0.0379, "samples": 439680, "wall_s": 27304.2, "cuda_gb": 16.86} |
| {"step": 13760, "loss": 0.001782, "loss_mean50": 0.001673, "lr": 2.933412352234567e-06, "grad_norm": 0.0227, "samples": 440320, "wall_s": 27343.1, "cuda_gb": 16.86} |
| {"step": 13780, "loss": 0.000951, "loss_mean50": 0.001679, "lr": 2.919641952820941e-06, "grad_norm": 0.0159, "samples": 440960, "wall_s": 27382.0, "cuda_gb": 16.86} |
| {"step": 13800, "loss": 0.002289, "loss_mean50": 0.001581, "lr": 2.906089698652277e-06, "grad_norm": 0.0397, "samples": 441600, "wall_s": 27420.9, "cuda_gb": 16.86} |
| {"step": 13820, "loss": 0.002437, "loss_mean50": 0.001519, "lr": 2.892755862698292e-06, "grad_norm": 0.0274, "samples": 442240, "wall_s": 27459.7, "cuda_gb": 16.86} |
| {"step": 13840, "loss": 0.001072, "loss_mean50": 0.001517, "lr": 2.8796407135293233e-06, "grad_norm": 0.0228, "samples": 442880, "wall_s": 27498.7, "cuda_gb": 16.86} |
| {"step": 13860, "loss": 0.00122, "loss_mean50": 0.00141, "lr": 2.866744515310917e-06, "grad_norm": 0.0234, "samples": 443520, "wall_s": 27537.6, "cuda_gb": 16.86} |
| {"step": 13880, "loss": 0.001711, "loss_mean50": 0.001392, "lr": 2.854067527798503e-06, "grad_norm": 0.1036, "samples": 444160, "wall_s": 27576.6, "cuda_gb": 16.86} |
| {"step": 13900, "loss": 0.001355, "loss_mean50": 0.001532, "lr": 2.8416100063321654e-06, "grad_norm": 0.0419, "samples": 444800, "wall_s": 27615.5, "cuda_gb": 16.86} |
| {"step": 13920, "loss": 0.000859, "loss_mean50": 0.001576, "lr": 2.829372201831502e-06, "grad_norm": 0.0199, "samples": 445440, "wall_s": 27654.5, "cuda_gb": 16.86} |
| {"step": 13940, "loss": 0.002144, "loss_mean50": 0.001528, "lr": 2.817354360790566e-06, "grad_norm": 0.0431, "samples": 446080, "wall_s": 27693.4, "cuda_gb": 16.86} |
| {"step": 13960, "loss": 0.001577, "loss_mean50": 0.001619, "lr": 2.8055567252729046e-06, "grad_norm": 0.0244, "samples": 446720, "wall_s": 27732.3, "cuda_gb": 16.86} |
| {"step": 13980, "loss": 0.002896, "loss_mean50": 0.001609, "lr": 2.7939795329066803e-06, "grad_norm": 0.1256, "samples": 447360, "wall_s": 27771.1, "cuda_gb": 16.86} |
| {"step": 14000, "loss": 0.001226, "loss_mean50": 0.001707, "lr": 2.7826230168798876e-06, "grad_norm": 0.0268, "samples": 448000, "wall_s": 27810.0, "cuda_gb": 16.86} |
| {"step": 14000, "val_flow_mse": 0.00147} |
| {"step": 14020, "loss": 0.001313, "loss_mean50": 0.001628, "lr": 2.7714874059356517e-06, "grad_norm": 0.0252, "samples": 448640, "wall_s": 27877.7, "cuda_gb": 16.86} |
| {"step": 14040, "loss": 0.001959, "loss_mean50": 0.001511, "lr": 2.76057292436763e-06, "grad_norm": 0.0384, "samples": 449280, "wall_s": 27916.7, "cuda_gb": 16.86} |
| {"step": 14060, "loss": 0.001818, "loss_mean50": 0.001558, "lr": 2.7498797920154817e-06, "grad_norm": 0.0244, "samples": 449920, "wall_s": 27955.6, "cuda_gb": 16.86} |
| {"step": 14080, "loss": 0.000791, "loss_mean50": 0.001509, "lr": 2.7394082242604514e-06, "grad_norm": 0.024, "samples": 450560, "wall_s": 27994.5, "cuda_gb": 16.86} |
| {"step": 14100, "loss": 0.000517, "loss_mean50": 0.0015, "lr": 2.7291584320210242e-06, "grad_norm": 0.0192, "samples": 451200, "wall_s": 28033.5, "cuda_gb": 16.86} |
| {"step": 14120, "loss": 0.003129, "loss_mean50": 0.001523, "lr": 2.7191306217486763e-06, "grad_norm": 0.0414, "samples": 451840, "wall_s": 28072.4, "cuda_gb": 16.86} |
| {"step": 14140, "loss": 0.001425, "loss_mean50": 0.001537, "lr": 2.709324995423726e-06, "grad_norm": 0.0283, "samples": 452480, "wall_s": 28111.3, "cuda_gb": 16.86} |
| {"step": 14160, "loss": 0.002592, "loss_mean50": 0.00159, "lr": 2.6997417505512548e-06, "grad_norm": 0.0509, "samples": 453120, "wall_s": 28150.1, "cuda_gb": 16.86} |
| {"step": 14180, "loss": 0.002673, "loss_mean50": 0.001629, "lr": 2.690381080157134e-06, "grad_norm": 0.032, "samples": 453760, "wall_s": 28189.0, "cuda_gb": 16.86} |
| {"step": 14200, "loss": 0.003077, "loss_mean50": 0.001713, "lr": 2.681243172784135e-06, "grad_norm": 0.0458, "samples": 454400, "wall_s": 28228.0, "cuda_gb": 16.86} |
| {"step": 14220, "loss": 0.001121, "loss_mean50": 0.001729, "lr": 2.6723282124881367e-06, "grad_norm": 0.0246, "samples": 455040, "wall_s": 28266.9, "cuda_gb": 16.86} |
| {"step": 14240, "loss": 0.001258, "loss_mean50": 0.001733, "lr": 2.663636378834409e-06, "grad_norm": 0.019, "samples": 455680, "wall_s": 28305.9, "cuda_gb": 16.86} |
| {"step": 14260, "loss": 0.000865, "loss_mean50": 0.001575, "lr": 2.6551678468940077e-06, "grad_norm": 0.021, "samples": 456320, "wall_s": 28344.8, "cuda_gb": 16.86} |
| {"step": 14280, "loss": 0.001346, "loss_mean50": 0.001581, "lr": 2.6469227872402357e-06, "grad_norm": 0.0607, "samples": 456960, "wall_s": 28383.8, "cuda_gb": 16.86} |
| {"step": 14300, "loss": 0.000785, "loss_mean50": 0.00153, "lr": 2.638901365945217e-06, "grad_norm": 0.0156, "samples": 457600, "wall_s": 28422.7, "cuda_gb": 16.86} |
| {"step": 14320, "loss": 0.001457, "loss_mean50": 0.001483, "lr": 2.6311037445765488e-06, "grad_norm": 0.0231, "samples": 458240, "wall_s": 28461.7, "cuda_gb": 16.86} |
| {"step": 14340, "loss": 0.001365, "loss_mean50": 0.001403, "lr": 2.623530080194044e-06, "grad_norm": 0.0214, "samples": 458880, "wall_s": 28500.5, "cuda_gb": 16.86} |
| {"step": 14360, "loss": 0.000596, "loss_mean50": 0.001348, "lr": 2.6161805253465737e-06, "grad_norm": 0.0146, "samples": 459520, "wall_s": 28539.4, "cuda_gb": 16.86} |
| {"step": 14380, "loss": 0.002447, "loss_mean50": 0.001428, "lr": 2.609055228068987e-06, "grad_norm": 0.0501, "samples": 460160, "wall_s": 28578.3, "cuda_gb": 16.86} |
| {"step": 14400, "loss": 0.001515, "loss_mean50": 0.001437, "lr": 2.6021543318791354e-06, "grad_norm": 0.034, "samples": 460800, "wall_s": 28617.1, "cuda_gb": 16.86} |
| {"step": 14420, "loss": 0.001425, "loss_mean50": 0.001399, "lr": 2.595477975774981e-06, "grad_norm": 0.0275, "samples": 461440, "wall_s": 28655.8, "cuda_gb": 16.86} |
| {"step": 14440, "loss": 0.001787, "loss_mean50": 0.001417, "lr": 2.589026294231797e-06, "grad_norm": 0.067, "samples": 462080, "wall_s": 28694.8, "cuda_gb": 16.86} |
| {"step": 14460, "loss": 0.000948, "loss_mean50": 0.001393, "lr": 2.582799417199454e-06, "grad_norm": 0.0264, "samples": 462720, "wall_s": 28733.7, "cuda_gb": 16.86} |
| {"step": 14480, "loss": 0.001359, "loss_mean50": 0.001499, "lr": 2.5767974700998093e-06, "grad_norm": 0.0206, "samples": 463360, "wall_s": 28772.6, "cuda_gb": 16.86} |
| {"step": 14500, "loss": 0.001061, "loss_mean50": 0.001546, "lr": 2.5710205738241775e-06, "grad_norm": 0.0209, "samples": 464000, "wall_s": 28811.5, "cuda_gb": 16.86} |
| {"step": 14520, "loss": 0.001135, "loss_mean50": 0.001505, "lr": 2.565468844730895e-06, "grad_norm": 0.0231, "samples": 464640, "wall_s": 28850.2, "cuda_gb": 16.86} |
| {"step": 14540, "loss": 0.001753, "loss_mean50": 0.001403, "lr": 2.5601423946429807e-06, "grad_norm": 0.0751, "samples": 465280, "wall_s": 28888.9, "cuda_gb": 16.86} |
| {"step": 14560, "loss": 0.00061, "loss_mean50": 0.001351, "lr": 2.555041330845874e-06, "grad_norm": 0.0129, "samples": 465920, "wall_s": 28927.7, "cuda_gb": 16.86} |
| {"step": 14580, "loss": 0.002258, "loss_mean50": 0.001462, "lr": 2.550165756085284e-06, "grad_norm": 0.0304, "samples": 466560, "wall_s": 28966.3, "cuda_gb": 16.86} |
| {"step": 14600, "loss": 0.002991, "loss_mean50": 0.001419, "lr": 2.5455157685651184e-06, "grad_norm": 0.028, "samples": 467200, "wall_s": 29005.1, "cuda_gb": 16.86} |
| {"step": 14620, "loss": 0.001289, "loss_mean50": 0.001382, "lr": 2.5410914619454962e-06, "grad_norm": 0.0264, "samples": 467840, "wall_s": 29043.8, "cuda_gb": 16.86} |
| {"step": 14640, "loss": 0.001222, "loss_mean50": 0.00134, "lr": 2.5368929253408747e-06, "grad_norm": 0.0245, "samples": 468480, "wall_s": 29082.5, "cuda_gb": 16.86} |
| {"step": 14660, "loss": 0.000859, "loss_mean50": 0.001306, "lr": 2.5329202433182453e-06, "grad_norm": 0.014, "samples": 469120, "wall_s": 29121.2, "cuda_gb": 16.86} |
| {"step": 14680, "loss": 0.001755, "loss_mean50": 0.001533, "lr": 2.529173495895432e-06, "grad_norm": 0.0346, "samples": 469760, "wall_s": 29160.1, "cuda_gb": 16.86} |
| {"step": 14700, "loss": 0.000998, "loss_mean50": 0.001583, "lr": 2.5256527585394805e-06, "grad_norm": 0.0162, "samples": 470400, "wall_s": 29198.8, "cuda_gb": 16.86} |
| {"step": 14720, "loss": 0.002697, "loss_mean50": 0.001716, "lr": 2.5223581021651405e-06, "grad_norm": 0.0401, "samples": 471040, "wall_s": 29237.5, "cuda_gb": 16.86} |
| {"step": 14740, "loss": 0.000419, "loss_mean50": 0.00156, "lr": 2.5192895931334335e-06, "grad_norm": 0.0119, "samples": 471680, "wall_s": 29276.2, "cuda_gb": 16.86} |
| {"step": 14760, "loss": 0.000658, "loss_mean50": 0.001505, "lr": 2.516447293250316e-06, "grad_norm": 0.0174, "samples": 472320, "wall_s": 29315.0, "cuda_gb": 16.86} |
| {"step": 14780, "loss": 0.000545, "loss_mean50": 0.001371, "lr": 2.5138312597654382e-06, "grad_norm": 0.0126, "samples": 472960, "wall_s": 29353.8, "cuda_gb": 16.86} |
| {"step": 14800, "loss": 0.002305, "loss_mean50": 0.001286, "lr": 2.5114415453709886e-06, "grad_norm": 0.0519, "samples": 473600, "wall_s": 29392.5, "cuda_gb": 16.86} |
| {"step": 14820, "loss": 0.00434, "loss_mean50": 0.001287, "lr": 2.509278198200633e-06, "grad_norm": 0.0547, "samples": 474240, "wall_s": 29431.3, "cuda_gb": 16.86} |
| {"step": 14840, "loss": 0.000743, "loss_mean50": 0.001352, "lr": 2.5073412618285445e-06, "grad_norm": 0.0144, "samples": 474880, "wall_s": 29470.0, "cuda_gb": 16.86} |
| {"step": 14860, "loss": 0.002102, "loss_mean50": 0.001418, "lr": 2.505630775268527e-06, "grad_norm": 0.0312, "samples": 475520, "wall_s": 29508.8, "cuda_gb": 16.86} |
| {"step": 14880, "loss": 0.001726, "loss_mean50": 0.001454, "lr": 2.5041467729732266e-06, "grad_norm": 0.0206, "samples": 476160, "wall_s": 29547.7, "cuda_gb": 16.86} |
| {"step": 14900, "loss": 0.001238, "loss_mean50": 0.00151, "lr": 2.502889284833443e-06, "grad_norm": 0.0242, "samples": 476800, "wall_s": 29586.8, "cuda_gb": 16.86} |
| {"step": 14920, "loss": 0.000825, "loss_mean50": 0.001543, "lr": 2.5018583361775205e-06, "grad_norm": 0.0284, "samples": 477440, "wall_s": 29625.7, "cuda_gb": 16.86} |
| {"step": 14940, "loss": 0.00137, "loss_mean50": 0.001417, "lr": 2.501053947770846e-06, "grad_norm": 0.0231, "samples": 478080, "wall_s": 29664.5, "cuda_gb": 16.86} |
| {"step": 14960, "loss": 0.001844, "loss_mean50": 0.001285, "lr": 2.5004761358154212e-06, "grad_norm": 0.0325, "samples": 478720, "wall_s": 29703.3, "cuda_gb": 16.86} |
| {"step": 14980, "loss": 0.002142, "loss_mean50": 0.00139, "lr": 2.5001249119495454e-06, "grad_norm": 0.0638, "samples": 479360, "wall_s": 29742.1, "cuda_gb": 16.86} |
| {"step": 15000, "loss": 0.002672, "loss_mean50": 0.001498, "lr": 2.5000002832475742e-06, "grad_norm": 0.04, "samples": 480000, "wall_s": 29780.8, "cuda_gb": 16.86} |
| {"step": 15000, "val_flow_mse": 0.00136, "val_action_mae": 0.011581, "best": true} |
|
|