{"step": 20, "loss": 0.076324, "loss_mean50": 0.074406, "lr": 5e-07, "grad_norm": 0.2141, "samples": 640, "wall_s": 18.7, "cuda_gb": 17.68} {"step": 40, "loss": 0.067783, "loss_mean50": 0.076254, "lr": 1e-06, "grad_norm": 0.2103, "samples": 1280, "wall_s": 30.7, "cuda_gb": 17.68} {"step": 60, "loss": 0.051624, "loss_mean50": 0.07296, "lr": 1.5e-06, "grad_norm": 0.152, "samples": 1920, "wall_s": 42.7, "cuda_gb": 17.68} {"step": 80, "loss": 0.059354, "loss_mean50": 0.067195, "lr": 2e-06, "grad_norm": 0.1924, "samples": 2560, "wall_s": 55.0, "cuda_gb": 17.68} {"step": 100, "loss": 0.033683, "loss_mean50": 0.054211, "lr": 2.5e-06, "grad_norm": 0.1458, "samples": 3200, "wall_s": 66.9, "cuda_gb": 17.68} {"step": 120, "loss": 0.014558, "loss_mean50": 0.035422, "lr": 3e-06, "grad_norm": 0.0803, "samples": 3840, "wall_s": 78.9, "cuda_gb": 17.68} {"step": 140, "loss": 0.01536, "loss_mean50": 0.022164, "lr": 3.5e-06, "grad_norm": 0.0834, "samples": 4480, "wall_s": 90.9, "cuda_gb": 17.68} {"step": 160, "loss": 0.012314, "loss_mean50": 0.014388, "lr": 4e-06, "grad_norm": 0.067, "samples": 5120, "wall_s": 102.9, "cuda_gb": 17.68} {"step": 180, "loss": 0.012732, "loss_mean50": 0.01213, "lr": 4.5e-06, "grad_norm": 0.0807, "samples": 5760, "wall_s": 115.0, "cuda_gb": 17.68} {"step": 200, "loss": 0.007973, "loss_mean50": 0.010466, "lr": 5e-06, "grad_norm": 0.0624, "samples": 6400, "wall_s": 126.7, "cuda_gb": 17.68} {"step": 220, "loss": 0.006935, "loss_mean50": 0.008947, "lr": 5.500000000000001e-06, "grad_norm": 0.0479, "samples": 7040, "wall_s": 138.7, "cuda_gb": 17.68} {"step": 240, "loss": 0.008611, "loss_mean50": 0.007723, "lr": 6e-06, "grad_norm": 0.0927, "samples": 7680, "wall_s": 150.6, "cuda_gb": 17.68} {"step": 260, "loss": 0.005738, "loss_mean50": 0.007243, "lr": 6.5000000000000004e-06, "grad_norm": 0.0832, "samples": 8320, "wall_s": 162.5, "cuda_gb": 17.68} {"step": 280, "loss": 0.0068, "loss_mean50": 0.007041, "lr": 7e-06, "grad_norm": 0.0805, "samples": 8960, "wall_s": 174.5, "cuda_gb": 17.68} {"step": 300, "loss": 0.007046, "loss_mean50": 0.00681, "lr": 7.5e-06, "grad_norm": 0.0563, "samples": 9600, "wall_s": 186.6, "cuda_gb": 17.68} {"step": 320, "loss": 0.005467, "loss_mean50": 0.006357, "lr": 8e-06, "grad_norm": 0.0537, "samples": 10240, "wall_s": 198.5, "cuda_gb": 17.68} {"step": 340, "loss": 0.006259, "loss_mean50": 0.006299, "lr": 8.5e-06, "grad_norm": 0.0523, "samples": 10880, "wall_s": 210.5, "cuda_gb": 17.68} {"step": 360, "loss": 0.005633, "loss_mean50": 0.005921, "lr": 9e-06, "grad_norm": 0.0434, "samples": 11520, "wall_s": 222.5, "cuda_gb": 17.68} {"step": 380, "loss": 0.004595, "loss_mean50": 0.005644, "lr": 9.5e-06, "grad_norm": 0.0389, "samples": 12160, "wall_s": 234.5, "cuda_gb": 17.68} {"step": 400, "loss": 0.005283, "loss_mean50": 0.005507, "lr": 1e-05, "grad_norm": 0.0585, "samples": 12800, "wall_s": 246.4, "cuda_gb": 17.68} {"step": 420, "loss": 0.004135, "loss_mean50": 0.005252, "lr": 1.0500000000000001e-05, "grad_norm": 0.036, "samples": 13440, "wall_s": 258.4, "cuda_gb": 17.68} {"step": 440, "loss": 0.006399, "loss_mean50": 0.00516, "lr": 1.1000000000000001e-05, "grad_norm": 0.089, "samples": 14080, "wall_s": 270.4, "cuda_gb": 17.68} {"step": 460, "loss": 0.006417, "loss_mean50": 0.004792, "lr": 1.15e-05, "grad_norm": 0.0619, "samples": 14720, "wall_s": 282.4, "cuda_gb": 17.68} {"step": 480, "loss": 0.003574, "loss_mean50": 0.004804, "lr": 1.2e-05, "grad_norm": 0.0418, "samples": 15360, "wall_s": 294.3, "cuda_gb": 17.68} {"step": 500, "loss": 0.006275, "loss_mean50": 0.004745, "lr": 1.25e-05, "grad_norm": 0.0441, "samples": 16000, "wall_s": 306.2, "cuda_gb": 17.68} {"step": 520, "loss": 0.007093, "loss_mean50": 0.004667, "lr": 1.3000000000000001e-05, "grad_norm": 0.0689, "samples": 16640, "wall_s": 318.2, "cuda_gb": 17.68} {"step": 540, "loss": 0.002886, "loss_mean50": 0.004801, "lr": 1.35e-05, "grad_norm": 0.0276, "samples": 17280, "wall_s": 330.2, "cuda_gb": 17.68} {"step": 560, "loss": 0.004365, "loss_mean50": 0.004669, "lr": 1.4e-05, "grad_norm": 0.0434, "samples": 17920, "wall_s": 342.1, "cuda_gb": 17.68} {"step": 580, "loss": 0.00331, "loss_mean50": 0.00458, "lr": 1.45e-05, "grad_norm": 0.0391, "samples": 18560, "wall_s": 354.1, "cuda_gb": 17.68} {"step": 600, "loss": 0.005972, "loss_mean50": 0.004657, "lr": 1.5e-05, "grad_norm": 0.0537, "samples": 19200, "wall_s": 366.2, "cuda_gb": 17.68} {"step": 620, "loss": 0.003374, "loss_mean50": 0.004624, "lr": 1.55e-05, "grad_norm": 0.0323, "samples": 19840, "wall_s": 378.1, "cuda_gb": 17.68} {"step": 640, "loss": 0.005363, "loss_mean50": 0.004746, "lr": 1.6e-05, "grad_norm": 0.0439, "samples": 20480, "wall_s": 390.1, "cuda_gb": 17.68} {"step": 660, "loss": 0.003932, "loss_mean50": 0.004453, "lr": 1.65e-05, "grad_norm": 0.0383, "samples": 21120, "wall_s": 402.0, "cuda_gb": 17.68} {"step": 680, "loss": 0.003174, "loss_mean50": 0.004298, "lr": 1.7e-05, "grad_norm": 0.0331, "samples": 21760, "wall_s": 414.0, "cuda_gb": 17.68} {"step": 700, "loss": 0.004957, "loss_mean50": 0.004186, "lr": 1.7500000000000002e-05, "grad_norm": 0.0532, "samples": 22400, "wall_s": 426.1, "cuda_gb": 17.68} {"step": 720, "loss": 0.00485, "loss_mean50": 0.004338, "lr": 1.8e-05, "grad_norm": 0.0369, "samples": 23040, "wall_s": 438.0, "cuda_gb": 17.68} {"step": 740, "loss": 0.003773, "loss_mean50": 0.004292, "lr": 1.8500000000000002e-05, "grad_norm": 0.1041, "samples": 23680, "wall_s": 449.9, "cuda_gb": 17.68} {"step": 760, "loss": 0.003001, "loss_mean50": 0.004229, "lr": 1.9e-05, "grad_norm": 0.0531, "samples": 24320, "wall_s": 462.0, "cuda_gb": 17.68} {"step": 780, "loss": 0.00506, "loss_mean50": 0.004447, "lr": 1.95e-05, "grad_norm": 0.0461, "samples": 24960, "wall_s": 473.8, "cuda_gb": 17.68} {"step": 800, "loss": 0.004628, "loss_mean50": 0.004294, "lr": 2e-05, "grad_norm": 0.0434, "samples": 25600, "wall_s": 485.9, "cuda_gb": 17.68} {"step": 820, "loss": 0.010712, "loss_mean50": 0.004052, "lr": 2.05e-05, "grad_norm": 0.0771, "samples": 26240, "wall_s": 497.8, "cuda_gb": 17.68} {"step": 840, "loss": 0.005735, "loss_mean50": 0.004176, "lr": 2.1000000000000002e-05, "grad_norm": 0.0457, "samples": 26880, "wall_s": 509.8, "cuda_gb": 17.68} {"step": 860, "loss": 0.009287, "loss_mean50": 0.004431, "lr": 2.15e-05, "grad_norm": 0.0777, "samples": 27520, "wall_s": 521.8, "cuda_gb": 17.68} {"step": 880, "loss": 0.003835, "loss_mean50": 0.004286, "lr": 2.2000000000000003e-05, "grad_norm": 0.0401, "samples": 28160, "wall_s": 533.8, "cuda_gb": 17.68} {"step": 900, "loss": 0.003843, "loss_mean50": 0.003911, "lr": 2.25e-05, "grad_norm": 0.0427, "samples": 28800, "wall_s": 545.8, "cuda_gb": 17.68} {"step": 920, "loss": 0.004381, "loss_mean50": 0.003962, "lr": 2.3e-05, "grad_norm": 0.0501, "samples": 29440, "wall_s": 557.8, "cuda_gb": 17.68} {"step": 940, "loss": 0.004285, "loss_mean50": 0.004054, "lr": 2.35e-05, "grad_norm": 0.049, "samples": 30080, "wall_s": 569.8, "cuda_gb": 17.68} {"step": 960, "loss": 0.003468, "loss_mean50": 0.004234, "lr": 2.4e-05, "grad_norm": 0.0349, "samples": 30720, "wall_s": 581.8, "cuda_gb": 17.68} {"step": 980, "loss": 0.004907, "loss_mean50": 0.004179, "lr": 2.4500000000000003e-05, "grad_norm": 0.072, "samples": 31360, "wall_s": 593.7, "cuda_gb": 17.68} {"step": 1000, "loss": 0.005331, "loss_mean50": 0.004037, "lr": 2.5e-05, "grad_norm": 0.0725, "samples": 32000, "wall_s": 605.6, "cuda_gb": 17.68} {"step": 1000, "val_flow_mse": 0.004514} {"step": 1020, "loss": 0.0036, "loss_mean50": 0.003976, "lr": 2.499989774778022e-05, "grad_norm": 0.0466, "samples": 32640, "wall_s": 625.5, "cuda_gb": 17.68} {"step": 1040, "loss": 0.004163, "loss_mean50": 0.004168, "lr": 2.4999569183187692e-05, "grad_norm": 0.0453, "samples": 33280, "wall_s": 637.5, "cuda_gb": 17.68} {"step": 1060, "loss": 0.001625, "loss_mean50": 0.004275, "lr": 2.4999014029592804e-05, "grad_norm": 0.0295, "samples": 33920, "wall_s": 649.5, "cuda_gb": 17.68} {"step": 1080, "loss": 0.004148, "loss_mean50": 0.004073, "lr": 2.4998232298177463e-05, "grad_norm": 0.0425, "samples": 34560, "wall_s": 661.5, "cuda_gb": 17.68} {"step": 1100, "loss": 0.003566, "loss_mean50": 0.003627, "lr": 2.499722400468732e-05, "grad_norm": 0.0424, "samples": 35200, "wall_s": 673.6, "cuda_gb": 17.68} {"step": 1120, "loss": 0.002636, "loss_mean50": 0.00337, "lr": 2.4995989169431427e-05, "grad_norm": 0.0696, "samples": 35840, "wall_s": 685.5, "cuda_gb": 17.68} {"step": 1140, "loss": 0.00503, "loss_mean50": 0.003652, "lr": 2.499452781728187e-05, "grad_norm": 0.0459, "samples": 36480, "wall_s": 697.5, "cuda_gb": 17.68} {"step": 1160, "loss": 0.003354, "loss_mean50": 0.003778, "lr": 2.4992839977673223e-05, "grad_norm": 0.0526, "samples": 37120, "wall_s": 709.4, "cuda_gb": 17.68} {"step": 1180, "loss": 0.005674, "loss_mean50": 0.00363, "lr": 2.4990925684601972e-05, "grad_norm": 0.063, "samples": 37760, "wall_s": 721.4, "cuda_gb": 17.68} {"step": 1200, "loss": 0.002833, "loss_mean50": 0.003627, "lr": 2.4988784976625844e-05, "grad_norm": 0.0275, "samples": 38400, "wall_s": 733.4, "cuda_gb": 17.68} {"step": 1220, "loss": 0.005132, "loss_mean50": 0.003713, "lr": 2.4986417896863013e-05, "grad_norm": 0.0484, "samples": 39040, "wall_s": 745.4, "cuda_gb": 17.68} {"step": 1240, "loss": 0.004662, "loss_mean50": 0.00355, "lr": 2.4983824492991236e-05, "grad_norm": 0.047, "samples": 39680, "wall_s": 757.5, "cuda_gb": 17.68} {"step": 1260, "loss": 0.00334, "loss_mean50": 0.003281, "lr": 2.4981004817246896e-05, "grad_norm": 0.0611, "samples": 40320, "wall_s": 769.4, "cuda_gb": 17.68} {"step": 1280, "loss": 0.003235, "loss_mean50": 0.003501, "lr": 2.497795892642395e-05, "grad_norm": 0.0402, "samples": 40960, "wall_s": 781.3, "cuda_gb": 17.68} {"step": 1300, "loss": 0.004159, "loss_mean50": 0.003352, "lr": 2.4974686881872776e-05, "grad_norm": 0.0463, "samples": 41600, "wall_s": 793.3, "cuda_gb": 17.68} {"step": 1320, "loss": 0.001613, "loss_mean50": 0.003271, "lr": 2.4971188749498953e-05, "grad_norm": 0.027, "samples": 42240, "wall_s": 805.3, "cuda_gb": 17.68} {"step": 1340, "loss": 0.003543, "loss_mean50": 0.003194, "lr": 2.4967464599761908e-05, "grad_norm": 0.0386, "samples": 42880, "wall_s": 817.2, "cuda_gb": 17.68} {"step": 1360, "loss": 0.001748, "loss_mean50": 0.003328, "lr": 2.4963514507673534e-05, "grad_norm": 0.0254, "samples": 43520, "wall_s": 829.2, "cuda_gb": 17.68} {"step": 1380, "loss": 0.002778, "loss_mean50": 0.003622, "lr": 2.4959338552796643e-05, "grad_norm": 0.0309, "samples": 44160, "wall_s": 841.2, "cuda_gb": 17.68} {"step": 1400, "loss": 0.003229, "loss_mean50": 0.003569, "lr": 2.495493681924339e-05, "grad_norm": 0.0413, "samples": 44800, "wall_s": 853.1, "cuda_gb": 17.68} {"step": 1420, "loss": 0.002251, "loss_mean50": 0.003317, "lr": 2.4950309395673564e-05, "grad_norm": 0.0386, "samples": 45440, "wall_s": 865.1, "cuda_gb": 17.68} {"step": 1440, "loss": 0.003345, "loss_mean50": 0.003078, "lr": 2.4945456375292797e-05, "grad_norm": 0.0368, "samples": 46080, "wall_s": 877.2, "cuda_gb": 17.68} {"step": 1460, "loss": 0.005684, "loss_mean50": 0.003372, "lr": 2.494037785585071e-05, "grad_norm": 0.0435, "samples": 46720, "wall_s": 889.0, "cuda_gb": 17.68} {"step": 1480, "loss": 0.003035, "loss_mean50": 0.003385, "lr": 2.493507393963892e-05, "grad_norm": 0.0292, "samples": 47360, "wall_s": 901.0, "cuda_gb": 17.68} {"step": 1500, "loss": 0.002119, "loss_mean50": 0.003587, "lr": 2.492954473348899e-05, "grad_norm": 0.0499, "samples": 48000, "wall_s": 913.0, "cuda_gb": 17.68} {"step": 1520, "loss": 0.001699, "loss_mean50": 0.003505, "lr": 2.4923790348770284e-05, "grad_norm": 0.0253, "samples": 48640, "wall_s": 925.0, "cuda_gb": 17.68} {"step": 1540, "loss": 0.00201, "loss_mean50": 0.003198, "lr": 2.4917810901387707e-05, "grad_norm": 0.0257, "samples": 49280, "wall_s": 937.0, "cuda_gb": 17.68} {"step": 1560, "loss": 0.002084, "loss_mean50": 0.003226, "lr": 2.4911606511779384e-05, "grad_norm": 0.0374, "samples": 49920, "wall_s": 949.0, "cuda_gb": 17.68} {"step": 1580, "loss": 0.003488, "loss_mean50": 0.003503, "lr": 2.4905177304914233e-05, "grad_norm": 0.0408, "samples": 50560, "wall_s": 960.8, "cuda_gb": 17.68} {"step": 1600, "loss": 0.001403, "loss_mean50": 0.003498, "lr": 2.4898523410289436e-05, "grad_norm": 0.0224, "samples": 51200, "wall_s": 972.8, "cuda_gb": 17.68} {"step": 1620, "loss": 0.003787, "loss_mean50": 0.003187, "lr": 2.4891644961927848e-05, "grad_norm": 0.0458, "samples": 51840, "wall_s": 984.8, "cuda_gb": 17.68} {"step": 1640, "loss": 0.004211, "loss_mean50": 0.003041, "lr": 2.488454209837529e-05, "grad_norm": 0.0423, "samples": 52480, "wall_s": 996.8, "cuda_gb": 17.68} {"step": 1660, "loss": 0.004909, "loss_mean50": 0.003267, "lr": 2.4877214962697744e-05, "grad_norm": 0.0534, "samples": 53120, "wall_s": 1008.8, "cuda_gb": 17.68} {"step": 1680, "loss": 0.004871, "loss_mean50": 0.003488, "lr": 2.4869663702478503e-05, "grad_norm": 0.0391, "samples": 53760, "wall_s": 1020.6, "cuda_gb": 17.68} {"step": 1700, "loss": 0.005478, "loss_mean50": 0.003393, "lr": 2.4861888469815163e-05, "grad_norm": 0.0543, "samples": 54400, "wall_s": 1032.6, "cuda_gb": 17.68} {"step": 1720, "loss": 0.001568, "loss_mean50": 0.003369, "lr": 2.4853889421316596e-05, "grad_norm": 0.0256, "samples": 55040, "wall_s": 1044.6, "cuda_gb": 17.68} {"step": 1740, "loss": 0.011224, "loss_mean50": 0.003347, "lr": 2.484566671809975e-05, "grad_norm": 0.056, "samples": 55680, "wall_s": 1056.6, "cuda_gb": 17.68} {"step": 1760, "loss": 0.005183, "loss_mean50": 0.003346, "lr": 2.4837220525786454e-05, "grad_norm": 0.0301, "samples": 56320, "wall_s": 1068.6, "cuda_gb": 17.68} {"step": 1780, "loss": 0.002015, "loss_mean50": 0.003656, "lr": 2.4828551014500046e-05, "grad_norm": 0.0341, "samples": 56960, "wall_s": 1080.6, "cuda_gb": 17.68} {"step": 1800, "loss": 0.002844, "loss_mean50": 0.003479, "lr": 2.4819658358861965e-05, "grad_norm": 0.0622, "samples": 57600, "wall_s": 1092.6, "cuda_gb": 17.68} {"step": 1820, "loss": 0.001809, "loss_mean50": 0.003235, "lr": 2.4810542737988217e-05, "grad_norm": 0.0334, "samples": 58240, "wall_s": 1104.6, "cuda_gb": 17.68} {"step": 1840, "loss": 0.002826, "loss_mean50": 0.003135, "lr": 2.480120433548579e-05, "grad_norm": 0.0256, "samples": 58880, "wall_s": 1116.7, "cuda_gb": 17.68} {"step": 1860, "loss": 0.001946, "loss_mean50": 0.00303, "lr": 2.479164333944893e-05, "grad_norm": 0.0478, "samples": 59520, "wall_s": 1128.6, "cuda_gb": 17.68} {"step": 1880, "loss": 0.006061, "loss_mean50": 0.003178, "lr": 2.4781859942455375e-05, "grad_norm": 0.0702, "samples": 60160, "wall_s": 1140.4, "cuda_gb": 17.68} {"step": 1900, "loss": 0.002259, "loss_mean50": 0.003321, "lr": 2.4771854341562465e-05, "grad_norm": 0.0318, "samples": 60800, "wall_s": 1152.4, "cuda_gb": 17.68} {"step": 1920, "loss": 0.001525, "loss_mean50": 0.003344, "lr": 2.476162673830317e-05, "grad_norm": 0.0291, "samples": 61440, "wall_s": 1164.4, "cuda_gb": 17.68} {"step": 1940, "loss": 0.002431, "loss_mean50": 0.00335, "lr": 2.475117733868204e-05, "grad_norm": 0.0268, "samples": 62080, "wall_s": 1176.4, "cuda_gb": 17.68} {"step": 1960, "loss": 0.003243, "loss_mean50": 0.003012, "lr": 2.474050635317105e-05, "grad_norm": 0.0269, "samples": 62720, "wall_s": 1188.4, "cuda_gb": 17.68} {"step": 1980, "loss": 0.001643, "loss_mean50": 0.003021, "lr": 2.472961399670536e-05, "grad_norm": 0.0435, "samples": 63360, "wall_s": 1200.4, "cuda_gb": 17.68} {"step": 2000, "loss": 0.003795, "loss_mean50": 0.003242, "lr": 2.471850048867899e-05, "grad_norm": 0.0348, "samples": 64000, "wall_s": 1212.4, "cuda_gb": 17.68} {"step": 2000, "val_flow_mse": 0.003646} {"step": 2020, "loss": 0.002427, "loss_mean50": 0.003429, "lr": 2.470716605294039e-05, "grad_norm": 0.0409, "samples": 64640, "wall_s": 1232.3, "cuda_gb": 17.68} {"step": 2040, "loss": 0.002578, "loss_mean50": 0.003238, "lr": 2.4695610917787952e-05, "grad_norm": 0.032, "samples": 65280, "wall_s": 1244.2, "cuda_gb": 17.68} {"step": 2060, "loss": 0.001101, "loss_mean50": 0.003115, "lr": 2.468383531596539e-05, "grad_norm": 0.0211, "samples": 65920, "wall_s": 1256.1, "cuda_gb": 17.68} {"step": 2080, "loss": 0.001881, "loss_mean50": 0.002968, "lr": 2.467183948465707e-05, "grad_norm": 0.0226, "samples": 66560, "wall_s": 1268.1, "cuda_gb": 17.68} {"step": 2100, "loss": 0.001202, "loss_mean50": 0.002961, "lr": 2.4659623665483212e-05, "grad_norm": 0.025, "samples": 67200, "wall_s": 1280.0, "cuda_gb": 17.68} {"step": 2120, "loss": 0.004015, "loss_mean50": 0.003309, "lr": 2.4647188104495043e-05, "grad_norm": 0.0696, "samples": 67840, "wall_s": 1292.0, "cuda_gb": 17.68} {"step": 2140, "loss": 0.003866, "loss_mean50": 0.003326, "lr": 2.463453305216983e-05, "grad_norm": 0.054, "samples": 68480, "wall_s": 1304.0, "cuda_gb": 17.68} {"step": 2160, "loss": 0.002366, "loss_mean50": 0.003254, "lr": 2.4621658763405842e-05, "grad_norm": 0.0269, "samples": 69120, "wall_s": 1316.0, "cuda_gb": 17.68} {"step": 2180, "loss": 0.004712, "loss_mean50": 0.003184, "lr": 2.4608565497517207e-05, "grad_norm": 0.0803, "samples": 69760, "wall_s": 1327.9, "cuda_gb": 17.68} {"step": 2200, "loss": 0.003502, "loss_mean50": 0.003304, "lr": 2.4595253518228692e-05, "grad_norm": 0.0271, "samples": 70400, "wall_s": 1339.9, "cuda_gb": 17.68} {"step": 2220, "loss": 0.006339, "loss_mean50": 0.003588, "lr": 2.4581723093670407e-05, "grad_norm": 0.0507, "samples": 71040, "wall_s": 1351.9, "cuda_gb": 17.68} {"step": 2240, "loss": 0.001649, "loss_mean50": 0.00302, "lr": 2.456797449637237e-05, "grad_norm": 0.0377, "samples": 71680, "wall_s": 1363.9, "cuda_gb": 17.68} {"step": 2260, "loss": 0.003196, "loss_mean50": 0.002921, "lr": 2.455400800325905e-05, "grad_norm": 0.0543, "samples": 72320, "wall_s": 1375.9, "cuda_gb": 17.68} {"step": 2280, "loss": 0.002023, "loss_mean50": 0.003056, "lr": 2.4539823895643783e-05, "grad_norm": 0.0363, "samples": 72960, "wall_s": 1387.8, "cuda_gb": 17.68} {"step": 2300, "loss": 0.001494, "loss_mean50": 0.003253, "lr": 2.4525422459223077e-05, "grad_norm": 0.0329, "samples": 73600, "wall_s": 1399.8, "cuda_gb": 17.68} {"step": 2320, "loss": 0.001398, "loss_mean50": 0.00313, "lr": 2.451080398407091e-05, "grad_norm": 0.0243, "samples": 74240, "wall_s": 1411.8, "cuda_gb": 17.68} {"step": 2340, "loss": 0.004482, "loss_mean50": 0.003109, "lr": 2.4495968764632826e-05, "grad_norm": 0.0301, "samples": 74880, "wall_s": 1423.8, "cuda_gb": 17.68} {"step": 2360, "loss": 0.004225, "loss_mean50": 0.002838, "lr": 2.4480917099720065e-05, "grad_norm": 0.0604, "samples": 75520, "wall_s": 1435.8, "cuda_gb": 17.68} {"step": 2380, "loss": 0.004758, "loss_mean50": 0.002881, "lr": 2.4465649292503495e-05, "grad_norm": 0.0523, "samples": 76160, "wall_s": 1447.8, "cuda_gb": 17.68} {"step": 2400, "loss": 0.002047, "loss_mean50": 0.002981, "lr": 2.4450165650507544e-05, "grad_norm": 0.0254, "samples": 76800, "wall_s": 1459.8, "cuda_gb": 17.68} {"step": 2420, "loss": 0.002417, "loss_mean50": 0.00299, "lr": 2.443446648560396e-05, "grad_norm": 0.0372, "samples": 77440, "wall_s": 1471.9, "cuda_gb": 17.68} {"step": 2440, "loss": 0.00136, "loss_mean50": 0.003074, "lr": 2.4418552114005585e-05, "grad_norm": 0.0207, "samples": 78080, "wall_s": 1483.8, "cuda_gb": 17.68} {"step": 2460, "loss": 0.004266, "loss_mean50": 0.003314, "lr": 2.4402422856259944e-05, "grad_norm": 0.0433, "samples": 78720, "wall_s": 1495.7, "cuda_gb": 17.68} {"step": 2480, "loss": 0.000909, "loss_mean50": 0.003014, "lr": 2.43860790372428e-05, "grad_norm": 0.0197, "samples": 79360, "wall_s": 1507.7, "cuda_gb": 17.68} {"step": 2500, "loss": 0.000947, "loss_mean50": 0.002793, "lr": 2.4369520986151623e-05, "grad_norm": 0.0207, "samples": 80000, "wall_s": 1519.7, "cuda_gb": 17.68} {"step": 2520, "loss": 0.004104, "loss_mean50": 0.002894, "lr": 2.435274903649894e-05, "grad_norm": 0.0333, "samples": 80640, "wall_s": 1571.2, "cuda_gb": 17.68} {"step": 2540, "loss": 0.001968, "loss_mean50": 0.003217, "lr": 2.4335763526105632e-05, "grad_norm": 0.0311, "samples": 81280, "wall_s": 1583.2, "cuda_gb": 17.68} {"step": 2560, "loss": 0.001815, "loss_mean50": 0.003177, "lr": 2.4318564797094113e-05, "grad_norm": 0.0395, "samples": 81920, "wall_s": 1595.2, "cuda_gb": 17.68} {"step": 2580, "loss": 0.002976, "loss_mean50": 0.003124, "lr": 2.4301153195881467e-05, "grad_norm": 0.0321, "samples": 82560, "wall_s": 1607.2, "cuda_gb": 17.68} {"step": 2600, "loss": 0.002167, "loss_mean50": 0.003133, "lr": 2.4283529073172444e-05, "grad_norm": 0.0261, "samples": 83200, "wall_s": 1619.1, "cuda_gb": 17.68} {"step": 2620, "loss": 0.002959, "loss_mean50": 0.003009, "lr": 2.42656927839524e-05, "grad_norm": 0.027, "samples": 83840, "wall_s": 1631.1, "cuda_gb": 17.68} {"step": 2640, "loss": 0.002832, "loss_mean50": 0.002982, "lr": 2.4247644687480155e-05, "grad_norm": 0.024, "samples": 84480, "wall_s": 1643.1, "cuda_gb": 17.68} {"step": 2660, "loss": 0.004261, "loss_mean50": 0.003001, "lr": 2.4229385147280765e-05, "grad_norm": 0.034, "samples": 85120, "wall_s": 1655.1, "cuda_gb": 17.68} {"step": 2680, "loss": 0.002482, "loss_mean50": 0.003118, "lr": 2.4210914531138177e-05, "grad_norm": 0.0545, "samples": 85760, "wall_s": 1667.1, "cuda_gb": 17.68} {"step": 2700, "loss": 0.002987, "loss_mean50": 0.002928, "lr": 2.419223321108783e-05, "grad_norm": 0.0244, "samples": 86400, "wall_s": 1679.0, "cuda_gb": 17.68} {"step": 2720, "loss": 0.001752, "loss_mean50": 0.002855, "lr": 2.4173341563409186e-05, "grad_norm": 0.04, "samples": 87040, "wall_s": 1691.0, "cuda_gb": 17.68} {"step": 2740, "loss": 0.001116, "loss_mean50": 0.0029, "lr": 2.41542399686181e-05, "grad_norm": 0.033, "samples": 87680, "wall_s": 1703.0, "cuda_gb": 17.68} {"step": 2760, "loss": 0.003996, "loss_mean50": 0.002697, "lr": 2.413492881145921e-05, "grad_norm": 0.048, "samples": 88320, "wall_s": 1715.0, "cuda_gb": 17.68} {"step": 2780, "loss": 0.001838, "loss_mean50": 0.002665, "lr": 2.4115408480898146e-05, "grad_norm": 0.0252, "samples": 88960, "wall_s": 1727.0, "cuda_gb": 17.68} {"step": 2800, "loss": 0.00143, "loss_mean50": 0.002788, "lr": 2.4095679370113723e-05, "grad_norm": 0.0303, "samples": 89600, "wall_s": 1739.0, "cuda_gb": 17.68} {"step": 2820, "loss": 0.002957, "loss_mean50": 0.003023, "lr": 2.407574187649e-05, "grad_norm": 0.0471, "samples": 90240, "wall_s": 1751.0, "cuda_gb": 17.68} {"step": 2840, "loss": 0.001349, "loss_mean50": 0.002829, "lr": 2.4055596401608298e-05, "grad_norm": 0.0265, "samples": 90880, "wall_s": 1763.0, "cuda_gb": 17.68} {"step": 2860, "loss": 0.001176, "loss_mean50": 0.002789, "lr": 2.4035243351239095e-05, "grad_norm": 0.0225, "samples": 91520, "wall_s": 1775.1, "cuda_gb": 17.68} {"step": 2880, "loss": 0.003024, "loss_mean50": 0.002997, "lr": 2.4014683135333848e-05, "grad_norm": 0.0395, "samples": 92160, "wall_s": 1787.1, "cuda_gb": 17.68} {"step": 2900, "loss": 0.003429, "loss_mean50": 0.003105, "lr": 2.3993916168016762e-05, "grad_norm": 0.0316, "samples": 92800, "wall_s": 1799.0, "cuda_gb": 17.68} {"step": 2920, "loss": 0.002555, "loss_mean50": 0.003112, "lr": 2.3972942867576424e-05, "grad_norm": 0.0258, "samples": 93440, "wall_s": 1811.0, "cuda_gb": 17.68} {"step": 2940, "loss": 0.004385, "loss_mean50": 0.003084, "lr": 2.3951763656457387e-05, "grad_norm": 0.0388, "samples": 94080, "wall_s": 1823.0, "cuda_gb": 17.68} {"step": 2960, "loss": 0.002642, "loss_mean50": 0.003152, "lr": 2.3930378961251658e-05, "grad_norm": 0.0314, "samples": 94720, "wall_s": 1835.0, "cuda_gb": 17.68} {"step": 2980, "loss": 0.003168, "loss_mean50": 0.003281, "lr": 2.3908789212690112e-05, "grad_norm": 0.0358, "samples": 95360, "wall_s": 1847.0, "cuda_gb": 17.68} {"step": 3000, "loss": 0.002118, "loss_mean50": 0.002606, "lr": 2.3886994845633805e-05, "grad_norm": 0.0389, "samples": 96000, "wall_s": 1859.0, "cuda_gb": 17.68} {"step": 3000, "val_flow_mse": 0.002984} {"step": 3020, "loss": 0.004613, "loss_mean50": 0.002832, "lr": 2.3864996299065235e-05, "grad_norm": 0.033, "samples": 96640, "wall_s": 1878.9, "cuda_gb": 17.68} {"step": 3040, "loss": 0.00119, "loss_mean50": 0.00298, "lr": 2.384279401607947e-05, "grad_norm": 0.0201, "samples": 97280, "wall_s": 1890.9, "cuda_gb": 17.68} {"step": 3060, "loss": 0.002502, "loss_mean50": 0.003169, "lr": 2.3820388443875256e-05, "grad_norm": 0.0395, "samples": 97920, "wall_s": 1902.9, "cuda_gb": 17.68} {"step": 3080, "loss": 0.002829, "loss_mean50": 0.002656, "lr": 2.379778003374598e-05, "grad_norm": 0.0296, "samples": 98560, "wall_s": 1914.9, "cuda_gb": 17.68} {"step": 3100, "loss": 0.001617, "loss_mean50": 0.002775, "lr": 2.3774969241070605e-05, "grad_norm": 0.0428, "samples": 99200, "wall_s": 1926.9, "cuda_gb": 17.68} {"step": 3120, "loss": 0.003427, "loss_mean50": 0.00273, "lr": 2.375195652530448e-05, "grad_norm": 0.0321, "samples": 99840, "wall_s": 1938.9, "cuda_gb": 17.68} {"step": 3140, "loss": 0.003956, "loss_mean50": 0.00286, "lr": 2.3728742349970097e-05, "grad_norm": 0.0411, "samples": 100480, "wall_s": 1950.9, "cuda_gb": 17.68} {"step": 3160, "loss": 0.002034, "loss_mean50": 0.002573, "lr": 2.3705327182647743e-05, "grad_norm": 0.0287, "samples": 101120, "wall_s": 1962.9, "cuda_gb": 17.68} {"step": 3180, "loss": 0.002378, "loss_mean50": 0.002823, "lr": 2.3681711494966095e-05, "grad_norm": 0.0291, "samples": 101760, "wall_s": 1974.8, "cuda_gb": 17.68} {"step": 3200, "loss": 0.001103, "loss_mean50": 0.002863, "lr": 2.365789576259271e-05, "grad_norm": 0.0287, "samples": 102400, "wall_s": 1986.7, "cuda_gb": 17.68} {"step": 3220, "loss": 0.003387, "loss_mean50": 0.003068, "lr": 2.363388046522445e-05, "grad_norm": 0.0477, "samples": 103040, "wall_s": 1998.7, "cuda_gb": 17.68} {"step": 3240, "loss": 0.00159, "loss_mean50": 0.002933, "lr": 2.3609666086577818e-05, "grad_norm": 0.0186, "samples": 103680, "wall_s": 2010.7, "cuda_gb": 17.68} {"step": 3260, "loss": 0.003384, "loss_mean50": 0.002814, "lr": 2.358525311437922e-05, "grad_norm": 0.0364, "samples": 104320, "wall_s": 2022.6, "cuda_gb": 17.68} {"step": 3280, "loss": 0.00233, "loss_mean50": 0.002942, "lr": 2.356064204035513e-05, "grad_norm": 0.0233, "samples": 104960, "wall_s": 2034.6, "cuda_gb": 17.68} {"step": 3300, "loss": 0.003418, "loss_mean50": 0.003041, "lr": 2.3535833360222197e-05, "grad_norm": 0.0262, "samples": 105600, "wall_s": 2046.5, "cuda_gb": 17.68} {"step": 3320, "loss": 0.004654, "loss_mean50": 0.003232, "lr": 2.351082757367725e-05, "grad_norm": 0.0285, "samples": 106240, "wall_s": 2058.4, "cuda_gb": 17.68} {"step": 3340, "loss": 0.001187, "loss_mean50": 0.002836, "lr": 2.3485625184387237e-05, "grad_norm": 0.0194, "samples": 106880, "wall_s": 2070.4, "cuda_gb": 17.68} {"step": 3360, "loss": 0.002392, "loss_mean50": 0.002883, "lr": 2.3460226699979094e-05, "grad_norm": 0.0303, "samples": 107520, "wall_s": 2082.4, "cuda_gb": 17.68} {"step": 3380, "loss": 0.0011, "loss_mean50": 0.002632, "lr": 2.3434632632029495e-05, "grad_norm": 0.0246, "samples": 108160, "wall_s": 2094.3, "cuda_gb": 17.68} {"step": 3400, "loss": 0.002527, "loss_mean50": 0.002547, "lr": 2.340884349605457e-05, "grad_norm": 0.0344, "samples": 108800, "wall_s": 2106.3, "cuda_gb": 17.68} {"step": 3420, "loss": 0.000981, "loss_mean50": 0.002856, "lr": 2.3382859811499506e-05, "grad_norm": 0.0195, "samples": 109440, "wall_s": 2118.3, "cuda_gb": 17.68} {"step": 3440, "loss": 0.00146, "loss_mean50": 0.003177, "lr": 2.3356682101728088e-05, "grad_norm": 0.0254, "samples": 110080, "wall_s": 2130.3, "cuda_gb": 17.68} {"step": 3460, "loss": 0.001727, "loss_mean50": 0.003109, "lr": 2.3330310894012176e-05, "grad_norm": 0.0189, "samples": 110720, "wall_s": 2142.3, "cuda_gb": 17.68} {"step": 3480, "loss": 0.002267, "loss_mean50": 0.003024, "lr": 2.3303746719521045e-05, "grad_norm": 0.0257, "samples": 111360, "wall_s": 2154.3, "cuda_gb": 17.68} {"step": 3500, "loss": 0.00127, "loss_mean50": 0.002787, "lr": 2.3276990113310735e-05, "grad_norm": 0.0277, "samples": 112000, "wall_s": 2166.2, "cuda_gb": 17.68} {"step": 3520, "loss": 0.003381, "loss_mean50": 0.002505, "lr": 2.3250041614313225e-05, "grad_norm": 0.0422, "samples": 112640, "wall_s": 2178.2, "cuda_gb": 17.68} {"step": 3540, "loss": 0.002547, "loss_mean50": 0.002315, "lr": 2.3222901765325615e-05, "grad_norm": 0.0492, "samples": 113280, "wall_s": 2190.1, "cuda_gb": 17.68} {"step": 3560, "loss": 0.001962, "loss_mean50": 0.002624, "lr": 2.3195571112999185e-05, "grad_norm": 0.0309, "samples": 113920, "wall_s": 2202.1, "cuda_gb": 17.68} {"step": 3580, "loss": 0.005164, "loss_mean50": 0.002826, "lr": 2.3168050207828366e-05, "grad_norm": 0.0521, "samples": 114560, "wall_s": 2214.1, "cuda_gb": 17.68} {"step": 3600, "loss": 0.002889, "loss_mean50": 0.002922, "lr": 2.3140339604139666e-05, "grad_norm": 0.0411, "samples": 115200, "wall_s": 2226.0, "cuda_gb": 17.68} {"step": 3620, "loss": 0.003226, "loss_mean50": 0.003022, "lr": 2.3112439860080513e-05, "grad_norm": 0.0213, "samples": 115840, "wall_s": 2238.0, "cuda_gb": 17.68} {"step": 3640, "loss": 0.001055, "loss_mean50": 0.002719, "lr": 2.308435153760799e-05, "grad_norm": 0.0214, "samples": 116480, "wall_s": 2250.0, "cuda_gb": 17.68} {"step": 3660, "loss": 0.0038, "loss_mean50": 0.00252, "lr": 2.305607520247755e-05, "grad_norm": 0.0329, "samples": 117120, "wall_s": 2262.0, "cuda_gb": 17.68} {"step": 3680, "loss": 0.003694, "loss_mean50": 0.002573, "lr": 2.302761142423158e-05, "grad_norm": 0.0269, "samples": 117760, "wall_s": 2274.0, "cuda_gb": 17.68} {"step": 3700, "loss": 0.003064, "loss_mean50": 0.002816, "lr": 2.2998960776187957e-05, "grad_norm": 0.035, "samples": 118400, "wall_s": 2285.9, "cuda_gb": 17.68} {"step": 3720, "loss": 0.002984, "loss_mean50": 0.00257, "lr": 2.2970123835428493e-05, "grad_norm": 0.0235, "samples": 119040, "wall_s": 2297.9, "cuda_gb": 17.68} {"step": 3740, "loss": 0.001502, "loss_mean50": 0.002332, "lr": 2.2941101182787307e-05, "grad_norm": 0.0215, "samples": 119680, "wall_s": 2309.9, "cuda_gb": 17.68} {"step": 3760, "loss": 0.003137, "loss_mean50": 0.002327, "lr": 2.2911893402839136e-05, "grad_norm": 0.0332, "samples": 120320, "wall_s": 2321.9, "cuda_gb": 17.68} {"step": 3780, "loss": 0.003636, "loss_mean50": 0.002713, "lr": 2.288250108388755e-05, "grad_norm": 0.0432, "samples": 120960, "wall_s": 2333.9, "cuda_gb": 17.68} {"step": 3800, "loss": 0.002544, "loss_mean50": 0.003046, "lr": 2.2852924817953105e-05, "grad_norm": 0.0225, "samples": 121600, "wall_s": 2345.9, "cuda_gb": 17.68} {"step": 3820, "loss": 0.002258, "loss_mean50": 0.003054, "lr": 2.2823165200761428e-05, "grad_norm": 0.0284, "samples": 122240, "wall_s": 2357.9, "cuda_gb": 17.68} {"step": 3840, "loss": 0.001517, "loss_mean50": 0.002739, "lr": 2.27932228317312e-05, "grad_norm": 0.0587, "samples": 122880, "wall_s": 2370.0, "cuda_gb": 17.68} {"step": 3860, "loss": 0.002875, "loss_mean50": 0.002635, "lr": 2.2763098313962093e-05, "grad_norm": 0.0329, "samples": 123520, "wall_s": 2381.9, "cuda_gb": 17.68} {"step": 3880, "loss": 0.001573, "loss_mean50": 0.002397, "lr": 2.2732792254222633e-05, "grad_norm": 0.0299, "samples": 124160, "wall_s": 2393.9, "cuda_gb": 17.68} {"step": 3900, "loss": 0.001623, "loss_mean50": 0.002655, "lr": 2.2702305262937958e-05, "grad_norm": 0.035, "samples": 124800, "wall_s": 2405.9, "cuda_gb": 17.68} {"step": 3920, "loss": 0.001286, "loss_mean50": 0.002438, "lr": 2.2671637954177526e-05, "grad_norm": 0.019, "samples": 125440, "wall_s": 2417.8, "cuda_gb": 17.68} {"step": 3940, "loss": 0.003396, "loss_mean50": 0.002379, "lr": 2.2640790945642772e-05, "grad_norm": 0.0348, "samples": 126080, "wall_s": 2429.8, "cuda_gb": 17.68} {"step": 3960, "loss": 0.005895, "loss_mean50": 0.002955, "lr": 2.2609764858654632e-05, "grad_norm": 0.0499, "samples": 126720, "wall_s": 2441.8, "cuda_gb": 17.68} {"step": 3980, "loss": 0.003353, "loss_mean50": 0.003137, "lr": 2.2578560318141043e-05, "grad_norm": 0.0244, "samples": 127360, "wall_s": 2453.8, "cuda_gb": 17.68} {"step": 4000, "loss": 0.001603, "loss_mean50": 0.003039, "lr": 2.254717795262436e-05, "grad_norm": 0.0275, "samples": 128000, "wall_s": 2465.7, "cuda_gb": 17.68} {"step": 4000, "val_flow_mse": 0.002935} {"step": 4020, "loss": 0.002067, "loss_mean50": 0.002932, "lr": 2.2515618394208694e-05, "grad_norm": 0.0336, "samples": 128640, "wall_s": 2485.5, "cuda_gb": 17.68} {"step": 4040, "loss": 0.002138, "loss_mean50": 0.002392, "lr": 2.248388227856717e-05, "grad_norm": 0.0225, "samples": 129280, "wall_s": 2497.5, "cuda_gb": 17.68} {"step": 4060, "loss": 0.002341, "loss_mean50": 0.002218, "lr": 2.2451970244929137e-05, "grad_norm": 0.0488, "samples": 129920, "wall_s": 2509.5, "cuda_gb": 17.68} {"step": 4080, "loss": 0.001244, "loss_mean50": 0.002341, "lr": 2.2419882936067294e-05, "grad_norm": 0.0276, "samples": 130560, "wall_s": 2521.5, "cuda_gb": 17.68} {"step": 4100, "loss": 0.005202, "loss_mean50": 0.002653, "lr": 2.238762099828472e-05, "grad_norm": 0.0498, "samples": 131200, "wall_s": 2533.5, "cuda_gb": 17.68} {"step": 4120, "loss": 0.004059, "loss_mean50": 0.002744, "lr": 2.235518508140189e-05, "grad_norm": 0.0336, "samples": 131840, "wall_s": 2545.5, "cuda_gb": 17.68} {"step": 4140, "loss": 0.002467, "loss_mean50": 0.002707, "lr": 2.232257583874355e-05, "grad_norm": 0.0274, "samples": 132480, "wall_s": 2557.5, "cuda_gb": 17.68} {"step": 4160, "loss": 0.005133, "loss_mean50": 0.00266, "lr": 2.2289793927125603e-05, "grad_norm": 0.0534, "samples": 133120, "wall_s": 2569.5, "cuda_gb": 17.68} {"step": 4180, "loss": 0.002609, "loss_mean50": 0.002627, "lr": 2.225684000684182e-05, "grad_norm": 0.0479, "samples": 133760, "wall_s": 2581.5, "cuda_gb": 17.68} {"step": 4200, "loss": 0.001146, "loss_mean50": 0.002723, "lr": 2.22237147416506e-05, "grad_norm": 0.0262, "samples": 134400, "wall_s": 2593.5, "cuda_gb": 17.68} {"step": 4220, "loss": 0.003151, "loss_mean50": 0.002639, "lr": 2.2190418798761562e-05, "grad_norm": 0.0273, "samples": 135040, "wall_s": 2605.5, "cuda_gb": 17.68} {"step": 4240, "loss": 0.00143, "loss_mean50": 0.002772, "lr": 2.2156952848822125e-05, "grad_norm": 0.0256, "samples": 135680, "wall_s": 2617.5, "cuda_gb": 17.68} {"step": 4260, "loss": 0.003298, "loss_mean50": 0.002621, "lr": 2.212331756590397e-05, "grad_norm": 0.0751, "samples": 136320, "wall_s": 2629.5, "cuda_gb": 17.68} {"step": 4280, "loss": 0.003357, "loss_mean50": 0.002645, "lr": 2.208951362748953e-05, "grad_norm": 0.0311, "samples": 136960, "wall_s": 2641.5, "cuda_gb": 17.68} {"step": 4300, "loss": 0.001493, "loss_mean50": 0.002725, "lr": 2.2055541714458252e-05, "grad_norm": 0.0337, "samples": 137600, "wall_s": 2653.6, "cuda_gb": 17.68} {"step": 4320, "loss": 0.005301, "loss_mean50": 0.002618, "lr": 2.2021402511072963e-05, "grad_norm": 0.0646, "samples": 138240, "wall_s": 2665.5, "cuda_gb": 17.68} {"step": 4340, "loss": 0.00151, "loss_mean50": 0.002666, "lr": 2.1987096704966034e-05, "grad_norm": 0.04, "samples": 138880, "wall_s": 2677.5, "cuda_gb": 17.68} {"step": 4360, "loss": 0.002659, "loss_mean50": 0.002615, "lr": 2.1952624987125566e-05, "grad_norm": 0.0432, "samples": 139520, "wall_s": 2689.5, "cuda_gb": 17.68} {"step": 4380, "loss": 0.003232, "loss_mean50": 0.002463, "lr": 2.1917988051881456e-05, "grad_norm": 0.0284, "samples": 140160, "wall_s": 2701.4, "cuda_gb": 17.68} {"step": 4400, "loss": 0.003779, "loss_mean50": 0.002409, "lr": 2.188318659689139e-05, "grad_norm": 0.0268, "samples": 140800, "wall_s": 2713.4, "cuda_gb": 17.68} {"step": 4420, "loss": 0.003683, "loss_mean50": 0.002189, "lr": 2.1848221323126857e-05, "grad_norm": 0.0224, "samples": 141440, "wall_s": 2725.4, "cuda_gb": 17.68} {"step": 4440, "loss": 0.0016, "loss_mean50": 0.002373, "lr": 2.1813092934858935e-05, "grad_norm": 0.0283, "samples": 142080, "wall_s": 2737.4, "cuda_gb": 17.68} {"step": 4460, "loss": 0.002177, "loss_mean50": 0.002403, "lr": 2.17778021396442e-05, "grad_norm": 0.0346, "samples": 142720, "wall_s": 2749.4, "cuda_gb": 17.68} {"step": 4480, "loss": 0.002576, "loss_mean50": 0.002607, "lr": 2.1742349648310403e-05, "grad_norm": 0.0619, "samples": 143360, "wall_s": 2761.3, "cuda_gb": 17.68} {"step": 4500, "loss": 0.002779, "loss_mean50": 0.002559, "lr": 2.1706736174942196e-05, "grad_norm": 0.0263, "samples": 144000, "wall_s": 2773.3, "cuda_gb": 17.68} {"step": 4520, "loss": 0.002167, "loss_mean50": 0.002316, "lr": 2.1670962436866725e-05, "grad_norm": 0.0232, "samples": 144640, "wall_s": 2785.3, "cuda_gb": 17.68} {"step": 4540, "loss": 0.002363, "loss_mean50": 0.002365, "lr": 2.16350291546392e-05, "grad_norm": 0.0269, "samples": 145280, "wall_s": 2797.3, "cuda_gb": 17.68} {"step": 4560, "loss": 0.003647, "loss_mean50": 0.002662, "lr": 2.159893705202836e-05, "grad_norm": 0.0304, "samples": 145920, "wall_s": 2809.3, "cuda_gb": 17.68} {"step": 4580, "loss": 0.003806, "loss_mean50": 0.002597, "lr": 2.1562686856001923e-05, "grad_norm": 0.0352, "samples": 146560, "wall_s": 2821.3, "cuda_gb": 17.68} {"step": 4600, "loss": 0.001699, "loss_mean50": 0.002338, "lr": 2.1526279296711908e-05, "grad_norm": 0.0225, "samples": 147200, "wall_s": 2833.3, "cuda_gb": 17.68} {"step": 4620, "loss": 0.004531, "loss_mean50": 0.002275, "lr": 2.1489715107479956e-05, "grad_norm": 0.0463, "samples": 147840, "wall_s": 2845.3, "cuda_gb": 17.68} {"step": 4640, "loss": 0.003146, "loss_mean50": 0.002178, "lr": 2.1452995024782553e-05, "grad_norm": 0.0241, "samples": 148480, "wall_s": 2857.3, "cuda_gb": 17.68} {"step": 4660, "loss": 0.00089, "loss_mean50": 0.002292, "lr": 2.1416119788236198e-05, "grad_norm": 0.0137, "samples": 149120, "wall_s": 2869.3, "cuda_gb": 17.68} {"step": 4680, "loss": 0.002862, "loss_mean50": 0.002285, "lr": 2.1379090140582484e-05, "grad_norm": 0.0246, "samples": 149760, "wall_s": 2881.3, "cuda_gb": 17.68} {"step": 4700, "loss": 0.002547, "loss_mean50": 0.002508, "lr": 2.1341906827673175e-05, "grad_norm": 0.031, "samples": 150400, "wall_s": 2893.3, "cuda_gb": 17.68} {"step": 4720, "loss": 0.001485, "loss_mean50": 0.002465, "lr": 2.130457059845515e-05, "grad_norm": 0.0159, "samples": 151040, "wall_s": 2905.4, "cuda_gb": 17.68} {"step": 4740, "loss": 0.001108, "loss_mean50": 0.002743, "lr": 2.1267082204955335e-05, "grad_norm": 0.0201, "samples": 151680, "wall_s": 2917.3, "cuda_gb": 17.68} {"step": 4760, "loss": 0.001715, "loss_mean50": 0.002621, "lr": 2.1229442402265554e-05, "grad_norm": 0.049, "samples": 152320, "wall_s": 2929.3, "cuda_gb": 17.68} {"step": 4780, "loss": 0.001392, "loss_mean50": 0.002437, "lr": 2.1191651948527313e-05, "grad_norm": 0.0242, "samples": 152960, "wall_s": 2941.3, "cuda_gb": 17.68} {"step": 4800, "loss": 0.002641, "loss_mean50": 0.002415, "lr": 2.1153711604916535e-05, "grad_norm": 0.0222, "samples": 153600, "wall_s": 2953.3, "cuda_gb": 17.68} {"step": 4820, "loss": 0.001132, "loss_mean50": 0.002754, "lr": 2.111562213562823e-05, "grad_norm": 0.021, "samples": 154240, "wall_s": 2965.4, "cuda_gb": 17.68} {"step": 4840, "loss": 0.0019, "loss_mean50": 0.002708, "lr": 2.1077384307861093e-05, "grad_norm": 0.0258, "samples": 154880, "wall_s": 2977.2, "cuda_gb": 17.68} {"step": 4860, "loss": 0.002924, "loss_mean50": 0.002755, "lr": 2.103899889180206e-05, "grad_norm": 0.0405, "samples": 155520, "wall_s": 2989.2, "cuda_gb": 17.68} {"step": 4880, "loss": 0.002379, "loss_mean50": 0.002669, "lr": 2.100046666061081e-05, "grad_norm": 0.0249, "samples": 156160, "wall_s": 3001.2, "cuda_gb": 17.68} {"step": 4900, "loss": 0.005318, "loss_mean50": 0.002435, "lr": 2.096178839040414e-05, "grad_norm": 0.0277, "samples": 156800, "wall_s": 3013.2, "cuda_gb": 17.68} {"step": 4920, "loss": 0.002845, "loss_mean50": 0.002391, "lr": 2.0922964860240404e-05, "grad_norm": 0.0435, "samples": 157440, "wall_s": 3025.2, "cuda_gb": 17.68} {"step": 4940, "loss": 0.002667, "loss_mean50": 0.002534, "lr": 2.088399685210376e-05, "grad_norm": 0.0364, "samples": 158080, "wall_s": 3037.2, "cuda_gb": 17.68} {"step": 4960, "loss": 0.002898, "loss_mean50": 0.002646, "lr": 2.0844885150888443e-05, "grad_norm": 0.0553, "samples": 158720, "wall_s": 3049.2, "cuda_gb": 17.68} {"step": 4980, "loss": 0.003455, "loss_mean50": 0.002449, "lr": 2.0805630544382978e-05, "grad_norm": 0.0338, "samples": 159360, "wall_s": 3061.2, "cuda_gb": 17.68} {"step": 5000, "loss": 0.007103, "loss_mean50": 0.002692, "lr": 2.076623382325427e-05, "grad_norm": 0.0492, "samples": 160000, "wall_s": 3073.2, "cuda_gb": 17.68} {"step": 5000, "val_flow_mse": 0.002709, "val_action_mae": 0.014198, "best": true} {"step": 5020, "loss": 0.004521, "loss_mean50": 0.002794, "lr": 2.0726695781031705e-05, "grad_norm": 0.041, "samples": 160640, "wall_s": 3158.7, "cuda_gb": 17.68} {"step": 5040, "loss": 0.002122, "loss_mean50": 0.002892, "lr": 2.0687017214091166e-05, "grad_norm": 0.0704, "samples": 161280, "wall_s": 3170.7, "cuda_gb": 17.68} {"step": 5060, "loss": 0.001735, "loss_mean50": 0.002461, "lr": 2.064719892163898e-05, "grad_norm": 0.0303, "samples": 161920, "wall_s": 3182.5, "cuda_gb": 17.68} {"step": 5080, "loss": 0.001976, "loss_mean50": 0.002261, "lr": 2.0607241705695827e-05, "grad_norm": 0.0287, "samples": 162560, "wall_s": 3194.5, "cuda_gb": 17.68} {"step": 5100, "loss": 0.002236, "loss_mean50": 0.002332, "lr": 2.05671463710806e-05, "grad_norm": 0.0233, "samples": 163200, "wall_s": 3206.5, "cuda_gb": 17.68} {"step": 5120, "loss": 0.001708, "loss_mean50": 0.002086, "lr": 2.0526913725394155e-05, "grad_norm": 0.0313, "samples": 163840, "wall_s": 3218.5, "cuda_gb": 17.68} {"step": 5140, "loss": 0.002773, "loss_mean50": 0.002338, "lr": 2.0486544579003098e-05, "grad_norm": 0.0445, "samples": 164480, "wall_s": 3230.6, "cuda_gb": 17.68} {"step": 5160, "loss": 0.002588, "loss_mean50": 0.002276, "lr": 2.044603974502341e-05, "grad_norm": 0.0357, "samples": 165120, "wall_s": 3242.5, "cuda_gb": 17.68} {"step": 5180, "loss": 0.002859, "loss_mean50": 0.002245, "lr": 2.0405400039304117e-05, "grad_norm": 0.0337, "samples": 165760, "wall_s": 3254.5, "cuda_gb": 17.68} {"step": 5200, "loss": 0.001325, "loss_mean50": 0.002237, "lr": 2.0364626280410812e-05, "grad_norm": 0.0258, "samples": 166400, "wall_s": 3266.5, "cuda_gb": 17.68} {"step": 5220, "loss": 0.002114, "loss_mean50": 0.002345, "lr": 2.03237192896092e-05, "grad_norm": 0.0265, "samples": 167040, "wall_s": 3278.5, "cuda_gb": 17.68} {"step": 5240, "loss": 0.001789, "loss_mean50": 0.002499, "lr": 2.028267989084855e-05, "grad_norm": 0.0218, "samples": 167680, "wall_s": 3290.5, "cuda_gb": 17.68} {"step": 5260, "loss": 0.002287, "loss_mean50": 0.002717, "lr": 2.024150891074508e-05, "grad_norm": 0.0237, "samples": 168320, "wall_s": 3302.2, "cuda_gb": 17.68} {"step": 5280, "loss": 0.003568, "loss_mean50": 0.00258, "lr": 2.0200207178565332e-05, "grad_norm": 0.0308, "samples": 168960, "wall_s": 3314.2, "cuda_gb": 17.68} {"step": 5300, "loss": 0.002112, "loss_mean50": 0.00266, "lr": 2.015877552620946e-05, "grad_norm": 0.0221, "samples": 169600, "wall_s": 3326.1, "cuda_gb": 17.68} {"step": 5320, "loss": 0.002127, "loss_mean50": 0.00267, "lr": 2.011721478819446e-05, "grad_norm": 0.0258, "samples": 170240, "wall_s": 3338.1, "cuda_gb": 17.68} {"step": 5340, "loss": 0.001821, "loss_mean50": 0.00212, "lr": 2.007552580163739e-05, "grad_norm": 0.0205, "samples": 170880, "wall_s": 3350.1, "cuda_gb": 17.68} {"step": 5360, "loss": 0.001266, "loss_mean50": 0.002217, "lr": 2.003370940623848e-05, "grad_norm": 0.0255, "samples": 171520, "wall_s": 3362.1, "cuda_gb": 17.68} {"step": 5380, "loss": 0.003754, "loss_mean50": 0.00241, "lr": 1.9991766444264238e-05, "grad_norm": 0.0364, "samples": 172160, "wall_s": 3374.1, "cuda_gb": 17.68} {"step": 5400, "loss": 0.003413, "loss_mean50": 0.002621, "lr": 1.9949697760530468e-05, "grad_norm": 0.0568, "samples": 172800, "wall_s": 3386.0, "cuda_gb": 17.68} {"step": 5420, "loss": 0.001547, "loss_mean50": 0.002611, "lr": 1.9907504202385275e-05, "grad_norm": 0.0529, "samples": 173440, "wall_s": 3398.0, "cuda_gb": 17.68} {"step": 5440, "loss": 0.002396, "loss_mean50": 0.002401, "lr": 1.9865186619691983e-05, "grad_norm": 0.0436, "samples": 174080, "wall_s": 3410.0, "cuda_gb": 17.68} {"step": 5460, "loss": 0.00202, "loss_mean50": 0.002177, "lr": 1.9822745864812015e-05, "grad_norm": 0.0434, "samples": 174720, "wall_s": 3421.9, "cuda_gb": 17.68} {"step": 5480, "loss": 0.002744, "loss_mean50": 0.002213, "lr": 1.9780182792587738e-05, "grad_norm": 0.0434, "samples": 175360, "wall_s": 3433.9, "cuda_gb": 17.68} {"step": 5500, "loss": 0.002099, "loss_mean50": 0.00233, "lr": 1.9737498260325232e-05, "grad_norm": 0.0272, "samples": 176000, "wall_s": 3445.9, "cuda_gb": 17.68} {"step": 5520, "loss": 0.005478, "loss_mean50": 0.002229, "lr": 1.969469312777703e-05, "grad_norm": 0.0836, "samples": 176640, "wall_s": 3457.9, "cuda_gb": 17.68} {"step": 5540, "loss": 0.00349, "loss_mean50": 0.002644, "lr": 1.9651768257124793e-05, "grad_norm": 0.0319, "samples": 177280, "wall_s": 3469.8, "cuda_gb": 17.68} {"step": 5560, "loss": 0.001053, "loss_mean50": 0.002617, "lr": 1.9608724512961953e-05, "grad_norm": 0.0287, "samples": 177920, "wall_s": 3481.8, "cuda_gb": 17.68} {"step": 5580, "loss": 0.002833, "loss_mean50": 0.002628, "lr": 1.9565562762276297e-05, "grad_norm": 0.0293, "samples": 178560, "wall_s": 3493.8, "cuda_gb": 17.68} {"step": 5600, "loss": 0.001942, "loss_mean50": 0.002354, "lr": 1.9522283874432492e-05, "grad_norm": 0.0459, "samples": 179200, "wall_s": 3505.8, "cuda_gb": 17.68} {"step": 5620, "loss": 0.001759, "loss_mean50": 0.002398, "lr": 1.9478888721154596e-05, "grad_norm": 0.0567, "samples": 179840, "wall_s": 3517.8, "cuda_gb": 17.68} {"step": 5640, "loss": 0.001587, "loss_mean50": 0.002437, "lr": 1.9435378176508466e-05, "grad_norm": 0.0266, "samples": 180480, "wall_s": 3529.8, "cuda_gb": 17.68} {"step": 5660, "loss": 0.001995, "loss_mean50": 0.002481, "lr": 1.9391753116884205e-05, "grad_norm": 0.0198, "samples": 181120, "wall_s": 3541.8, "cuda_gb": 17.68} {"step": 5680, "loss": 0.001959, "loss_mean50": 0.002429, "lr": 1.934801442097845e-05, "grad_norm": 0.0332, "samples": 181760, "wall_s": 3553.7, "cuda_gb": 17.68} {"step": 5700, "loss": 0.007714, "loss_mean50": 0.0023, "lr": 1.9304162969776722e-05, "grad_norm": 0.085, "samples": 182400, "wall_s": 3565.7, "cuda_gb": 17.68} {"step": 5720, "loss": 0.002324, "loss_mean50": 0.002508, "lr": 1.9260199646535644e-05, "grad_norm": 0.0335, "samples": 183040, "wall_s": 3577.7, "cuda_gb": 17.68} {"step": 5740, "loss": 0.002306, "loss_mean50": 0.002411, "lr": 1.921612533676518e-05, "grad_norm": 0.0221, "samples": 183680, "wall_s": 3589.7, "cuda_gb": 17.68} {"step": 5760, "loss": 0.002057, "loss_mean50": 0.002056, "lr": 1.9171940928210784e-05, "grad_norm": 0.0246, "samples": 184320, "wall_s": 3601.6, "cuda_gb": 17.68} {"step": 5780, "loss": 0.002869, "loss_mean50": 0.002046, "lr": 1.9127647310835513e-05, "grad_norm": 0.0363, "samples": 184960, "wall_s": 3613.5, "cuda_gb": 17.68} {"step": 5800, "loss": 0.0013, "loss_mean50": 0.002256, "lr": 1.9083245376802126e-05, "grad_norm": 0.0209, "samples": 185600, "wall_s": 3625.5, "cuda_gb": 17.68} {"step": 5820, "loss": 0.00132, "loss_mean50": 0.002491, "lr": 1.903873602045508e-05, "grad_norm": 0.031, "samples": 186240, "wall_s": 3637.6, "cuda_gb": 17.68} {"step": 5840, "loss": 0.002559, "loss_mean50": 0.002394, "lr": 1.8994120138302543e-05, "grad_norm": 0.0609, "samples": 186880, "wall_s": 3649.5, "cuda_gb": 17.68} {"step": 5860, "loss": 0.001196, "loss_mean50": 0.002239, "lr": 1.894939862899833e-05, "grad_norm": 0.021, "samples": 187520, "wall_s": 3661.5, "cuda_gb": 17.68} {"step": 5880, "loss": 0.002683, "loss_mean50": 0.002413, "lr": 1.8904572393323796e-05, "grad_norm": 0.0355, "samples": 188160, "wall_s": 3673.4, "cuda_gb": 17.68} {"step": 5900, "loss": 0.002742, "loss_mean50": 0.002446, "lr": 1.8859642334169693e-05, "grad_norm": 0.032, "samples": 188800, "wall_s": 3685.4, "cuda_gb": 17.68} {"step": 5920, "loss": 0.002815, "loss_mean50": 0.002496, "lr": 1.8814609356517996e-05, "grad_norm": 0.0508, "samples": 189440, "wall_s": 3697.4, "cuda_gb": 17.68} {"step": 5940, "loss": 0.002298, "loss_mean50": 0.00237, "lr": 1.8769474367423657e-05, "grad_norm": 0.0221, "samples": 190080, "wall_s": 3709.4, "cuda_gb": 17.68} {"step": 5960, "loss": 0.001233, "loss_mean50": 0.002199, "lr": 1.872423827599635e-05, "grad_norm": 0.0199, "samples": 190720, "wall_s": 3721.4, "cuda_gb": 17.68} {"step": 5980, "loss": 0.002019, "loss_mean50": 0.002279, "lr": 1.8678901993382156e-05, "grad_norm": 0.0275, "samples": 191360, "wall_s": 3733.4, "cuda_gb": 17.68} {"step": 6000, "loss": 0.003418, "loss_mean50": 0.002303, "lr": 1.8633466432745193e-05, "grad_norm": 0.0318, "samples": 192000, "wall_s": 3745.4, "cuda_gb": 17.68} {"step": 6000, "val_flow_mse": 0.002349} {"step": 6020, "loss": 0.001961, "loss_mean50": 0.002404, "lr": 1.8587932509249265e-05, "grad_norm": 0.0382, "samples": 192640, "wall_s": 3765.3, "cuda_gb": 17.68} {"step": 6040, "loss": 0.002864, "loss_mean50": 0.002418, "lr": 1.8542301140039374e-05, "grad_norm": 0.0258, "samples": 193280, "wall_s": 3777.3, "cuda_gb": 17.68} {"step": 6060, "loss": 0.004992, "loss_mean50": 0.002341, "lr": 1.849657324422329e-05, "grad_norm": 0.0319, "samples": 193920, "wall_s": 3789.4, "cuda_gb": 17.68} {"step": 6080, "loss": 0.001873, "loss_mean50": 0.002333, "lr": 1.845074974285303e-05, "grad_norm": 0.0441, "samples": 194560, "wall_s": 3801.3, "cuda_gb": 17.68} {"step": 6100, "loss": 0.002037, "loss_mean50": 0.002307, "lr": 1.8404831558906283e-05, "grad_norm": 0.0185, "samples": 195200, "wall_s": 3813.3, "cuda_gb": 17.68} {"step": 6120, "loss": 0.003626, "loss_mean50": 0.00242, "lr": 1.8358819617267847e-05, "grad_norm": 0.0417, "samples": 195840, "wall_s": 3825.3, "cuda_gb": 17.68} {"step": 6140, "loss": 0.002079, "loss_mean50": 0.00241, "lr": 1.831271484471099e-05, "grad_norm": 0.0217, "samples": 196480, "wall_s": 3837.2, "cuda_gb": 17.68} {"step": 6160, "loss": 0.002208, "loss_mean50": 0.002135, "lr": 1.8266518169878774e-05, "grad_norm": 0.0715, "samples": 197120, "wall_s": 3849.2, "cuda_gb": 17.68} {"step": 6180, "loss": 0.002612, "loss_mean50": 0.001971, "lr": 1.822023052326536e-05, "grad_norm": 0.0394, "samples": 197760, "wall_s": 3861.2, "cuda_gb": 17.68} {"step": 6200, "loss": 0.002193, "loss_mean50": 0.002072, "lr": 1.8173852837197273e-05, "grad_norm": 0.0699, "samples": 198400, "wall_s": 3873.2, "cuda_gb": 17.68} {"step": 6220, "loss": 0.001512, "loss_mean50": 0.002024, "lr": 1.81273860458146e-05, "grad_norm": 0.0938, "samples": 199040, "wall_s": 3885.2, "cuda_gb": 17.68} {"step": 6240, "loss": 0.006021, "loss_mean50": 0.00239, "lr": 1.8080831085052197e-05, "grad_norm": 0.099, "samples": 199680, "wall_s": 3897.2, "cuda_gb": 17.68} {"step": 6260, "loss": 0.001203, "loss_mean50": 0.002248, "lr": 1.8034188892620834e-05, "grad_norm": 0.0245, "samples": 200320, "wall_s": 3909.3, "cuda_gb": 17.68} {"step": 6280, "loss": 0.002105, "loss_mean50": 0.002059, "lr": 1.7987460407988294e-05, "grad_norm": 0.0508, "samples": 200960, "wall_s": 3921.3, "cuda_gb": 17.68} {"step": 6300, "loss": 0.002608, "loss_mean50": 0.001926, "lr": 1.794064657236046e-05, "grad_norm": 0.0799, "samples": 201600, "wall_s": 3933.3, "cuda_gb": 17.68} {"step": 6320, "loss": 0.000597, "loss_mean50": 0.002083, "lr": 1.7893748328662355e-05, "grad_norm": 0.0218, "samples": 202240, "wall_s": 3945.2, "cuda_gb": 17.68} {"step": 6340, "loss": 0.001181, "loss_mean50": 0.002189, "lr": 1.7846766621519163e-05, "grad_norm": 0.0201, "samples": 202880, "wall_s": 3957.2, "cuda_gb": 17.68} {"step": 6360, "loss": 0.003276, "loss_mean50": 0.002318, "lr": 1.7799702397237172e-05, "grad_norm": 0.0358, "samples": 203520, "wall_s": 3969.2, "cuda_gb": 17.68} {"step": 6380, "loss": 0.001678, "loss_mean50": 0.00239, "lr": 1.775255660378474e-05, "grad_norm": 0.0659, "samples": 204160, "wall_s": 3981.2, "cuda_gb": 17.68} {"step": 6400, "loss": 0.003938, "loss_mean50": 0.002298, "lr": 1.77053301907732e-05, "grad_norm": 0.0615, "samples": 204800, "wall_s": 3993.2, "cuda_gb": 17.68} {"step": 6420, "loss": 0.002264, "loss_mean50": 0.002202, "lr": 1.765802410943771e-05, "grad_norm": 0.0314, "samples": 205440, "wall_s": 4005.1, "cuda_gb": 17.68} {"step": 6440, "loss": 0.002581, "loss_mean50": 0.002205, "lr": 1.761063931261812e-05, "grad_norm": 0.0308, "samples": 206080, "wall_s": 4017.2, "cuda_gb": 17.68} {"step": 6460, "loss": 0.001015, "loss_mean50": 0.001837, "lr": 1.7563176754739762e-05, "grad_norm": 0.0193, "samples": 206720, "wall_s": 4029.0, "cuda_gb": 17.68} {"step": 6480, "loss": 0.003945, "loss_mean50": 0.001935, "lr": 1.751563739179424e-05, "grad_norm": 0.0536, "samples": 207360, "wall_s": 4040.9, "cuda_gb": 17.68} {"step": 6500, "loss": 0.001979, "loss_mean50": 0.001994, "lr": 1.7468022181320165e-05, "grad_norm": 0.0167, "samples": 208000, "wall_s": 4052.9, "cuda_gb": 17.68} {"step": 6520, "loss": 0.00188, "loss_mean50": 0.001868, "lr": 1.742033208238386e-05, "grad_norm": 0.0811, "samples": 208640, "wall_s": 4064.9, "cuda_gb": 17.68} {"step": 6540, "loss": 0.003507, "loss_mean50": 0.002159, "lr": 1.737256805556007e-05, "grad_norm": 0.0234, "samples": 209280, "wall_s": 4076.9, "cuda_gb": 17.68} {"step": 6560, "loss": 0.001607, "loss_mean50": 0.002232, "lr": 1.7324731062912574e-05, "grad_norm": 0.021, "samples": 209920, "wall_s": 4088.9, "cuda_gb": 17.68} {"step": 6580, "loss": 0.002099, "loss_mean50": 0.002239, "lr": 1.7276822067974855e-05, "grad_norm": 0.0355, "samples": 210560, "wall_s": 4100.9, "cuda_gb": 17.68} {"step": 6600, "loss": 0.00183, "loss_mean50": 0.00232, "lr": 1.722884203573065e-05, "grad_norm": 0.0348, "samples": 211200, "wall_s": 4112.9, "cuda_gb": 17.68} {"step": 6620, "loss": 0.000936, "loss_mean50": 0.002376, "lr": 1.7180791932594536e-05, "grad_norm": 0.0308, "samples": 211840, "wall_s": 4125.0, "cuda_gb": 17.68} {"step": 6640, "loss": 0.002375, "loss_mean50": 0.002311, "lr": 1.7132672726392453e-05, "grad_norm": 0.0255, "samples": 212480, "wall_s": 4136.9, "cuda_gb": 17.68} {"step": 6660, "loss": 0.007137, "loss_mean50": 0.002332, "lr": 1.708448538634223e-05, "grad_norm": 0.1343, "samples": 213120, "wall_s": 4148.9, "cuda_gb": 17.68} {"step": 6680, "loss": 0.001378, "loss_mean50": 0.002326, "lr": 1.703623088303404e-05, "grad_norm": 0.0229, "samples": 213760, "wall_s": 4160.9, "cuda_gb": 17.68} {"step": 6700, "loss": 0.003687, "loss_mean50": 0.002264, "lr": 1.698791018841086e-05, "grad_norm": 0.0501, "samples": 214400, "wall_s": 4172.9, "cuda_gb": 17.68} {"step": 6720, "loss": 0.004321, "loss_mean50": 0.002118, "lr": 1.693952427574889e-05, "grad_norm": 0.0224, "samples": 215040, "wall_s": 4184.9, "cuda_gb": 17.68} {"step": 6740, "loss": 0.001461, "loss_mean50": 0.002062, "lr": 1.6891074119637956e-05, "grad_norm": 0.019, "samples": 215680, "wall_s": 4196.9, "cuda_gb": 17.68} {"step": 6760, "loss": 0.001509, "loss_mean50": 0.002125, "lr": 1.6842560695961884e-05, "grad_norm": 0.0417, "samples": 216320, "wall_s": 4208.8, "cuda_gb": 17.68} {"step": 6780, "loss": 0.001744, "loss_mean50": 0.00205, "lr": 1.6793984981878822e-05, "grad_norm": 0.04, "samples": 216960, "wall_s": 4220.8, "cuda_gb": 17.68} {"step": 6800, "loss": 0.003297, "loss_mean50": 0.002059, "lr": 1.6745347955801585e-05, "grad_norm": 0.0245, "samples": 217600, "wall_s": 4232.8, "cuda_gb": 17.68} {"step": 6820, "loss": 0.003027, "loss_mean50": 0.002151, "lr": 1.6696650597377923e-05, "grad_norm": 0.0283, "samples": 218240, "wall_s": 4244.7, "cuda_gb": 17.68} {"step": 6840, "loss": 0.001615, "loss_mean50": 0.002024, "lr": 1.6647893887470816e-05, "grad_norm": 0.0268, "samples": 218880, "wall_s": 4256.7, "cuda_gb": 17.68} {"step": 6860, "loss": 0.003597, "loss_mean50": 0.002338, "lr": 1.6599078808138693e-05, "grad_norm": 0.0337, "samples": 219520, "wall_s": 4268.6, "cuda_gb": 17.68} {"step": 6880, "loss": 0.001094, "loss_mean50": 0.002539, "lr": 1.6550206342615658e-05, "grad_norm": 0.0148, "samples": 220160, "wall_s": 4280.6, "cuda_gb": 17.68} {"step": 6900, "loss": 0.002241, "loss_mean50": 0.002388, "lr": 1.6501277475291696e-05, "grad_norm": 0.0194, "samples": 220800, "wall_s": 4292.7, "cuda_gb": 17.68} {"step": 6920, "loss": 0.001048, "loss_mean50": 0.002491, "lr": 1.645229319169284e-05, "grad_norm": 0.0207, "samples": 221440, "wall_s": 4304.5, "cuda_gb": 17.68} {"step": 6940, "loss": 0.001502, "loss_mean50": 0.002425, "lr": 1.64032544784613e-05, "grad_norm": 0.0234, "samples": 222080, "wall_s": 4316.5, "cuda_gb": 17.68} {"step": 6960, "loss": 0.001308, "loss_mean50": 0.002192, "lr": 1.6354162323335633e-05, "grad_norm": 0.0209, "samples": 222720, "wall_s": 4328.5, "cuda_gb": 17.68} {"step": 6980, "loss": 0.001123, "loss_mean50": 0.002131, "lr": 1.6305017715130812e-05, "grad_norm": 0.0255, "samples": 223360, "wall_s": 4340.5, "cuda_gb": 17.68} {"step": 7000, "loss": 0.001538, "loss_mean50": 0.002142, "lr": 1.6255821643718313e-05, "grad_norm": 0.0158, "samples": 224000, "wall_s": 4352.5, "cuda_gb": 17.68} {"step": 7000, "val_flow_mse": 0.002328} {"step": 7020, "loss": 0.001418, "loss_mean50": 0.001977, "lr": 1.6206575100006205e-05, "grad_norm": 0.0285, "samples": 224640, "wall_s": 4372.4, "cuda_gb": 17.68} {"step": 7040, "loss": 0.000836, "loss_mean50": 0.002045, "lr": 1.615727907591915e-05, "grad_norm": 0.0302, "samples": 225280, "wall_s": 4384.4, "cuda_gb": 17.68} {"step": 7060, "loss": 0.001268, "loss_mean50": 0.001922, "lr": 1.610793456437846e-05, "grad_norm": 0.016, "samples": 225920, "wall_s": 4396.4, "cuda_gb": 17.68} {"step": 7080, "loss": 0.001526, "loss_mean50": 0.001947, "lr": 1.6058542559282066e-05, "grad_norm": 0.0451, "samples": 226560, "wall_s": 4408.4, "cuda_gb": 17.68} {"step": 7100, "loss": 0.002, "loss_mean50": 0.002096, "lr": 1.6009104055484537e-05, "grad_norm": 0.0225, "samples": 227200, "wall_s": 4420.4, "cuda_gb": 17.68} {"step": 7120, "loss": 0.001681, "loss_mean50": 0.001939, "lr": 1.5959620048776996e-05, "grad_norm": 0.03, "samples": 227840, "wall_s": 4432.4, "cuda_gb": 17.68} {"step": 7140, "loss": 0.001735, "loss_mean50": 0.001873, "lr": 1.59100915358671e-05, "grad_norm": 0.041, "samples": 228480, "wall_s": 4444.4, "cuda_gb": 17.68} {"step": 7160, "loss": 0.001651, "loss_mean50": 0.002022, "lr": 1.586051951435895e-05, "grad_norm": 0.0196, "samples": 229120, "wall_s": 4456.4, "cuda_gb": 17.68} {"step": 7180, "loss": 0.002689, "loss_mean50": 0.002124, "lr": 1.5810904982733e-05, "grad_norm": 0.0224, "samples": 229760, "wall_s": 4468.4, "cuda_gb": 17.68} {"step": 7200, "loss": 0.00125, "loss_mean50": 0.002072, "lr": 1.576124894032593e-05, "grad_norm": 0.0244, "samples": 230400, "wall_s": 4480.4, "cuda_gb": 17.68} {"step": 7220, "loss": 0.001168, "loss_mean50": 0.002005, "lr": 1.5711552387310548e-05, "grad_norm": 0.0191, "samples": 231040, "wall_s": 4492.4, "cuda_gb": 17.68} {"step": 7240, "loss": 0.001385, "loss_mean50": 0.002032, "lr": 1.5661816324675613e-05, "grad_norm": 0.0257, "samples": 231680, "wall_s": 4504.3, "cuda_gb": 17.68} {"step": 7260, "loss": 0.001268, "loss_mean50": 0.001703, "lr": 1.56120417542057e-05, "grad_norm": 0.0173, "samples": 232320, "wall_s": 4516.2, "cuda_gb": 17.68} {"step": 7280, "loss": 0.003274, "loss_mean50": 0.001656, "lr": 1.5562229678460998e-05, "grad_norm": 0.0465, "samples": 232960, "wall_s": 4528.2, "cuda_gb": 17.68} {"step": 7300, "loss": 0.004688, "loss_mean50": 0.001896, "lr": 1.5512381100757136e-05, "grad_norm": 0.0714, "samples": 233600, "wall_s": 4540.2, "cuda_gb": 17.68} {"step": 7320, "loss": 0.002874, "loss_mean50": 0.002225, "lr": 1.5462497025144967e-05, "grad_norm": 0.0322, "samples": 234240, "wall_s": 4552.2, "cuda_gb": 17.68} {"step": 7340, "loss": 0.004852, "loss_mean50": 0.002252, "lr": 1.541257845639033e-05, "grad_norm": 0.0512, "samples": 234880, "wall_s": 4564.2, "cuda_gb": 17.68} {"step": 7360, "loss": 0.001, "loss_mean50": 0.00199, "lr": 1.536262639995384e-05, "grad_norm": 0.0171, "samples": 235520, "wall_s": 4576.2, "cuda_gb": 17.68} {"step": 7380, "loss": 0.003915, "loss_mean50": 0.002214, "lr": 1.531264186197062e-05, "grad_norm": 0.0531, "samples": 236160, "wall_s": 4588.2, "cuda_gb": 17.68} {"step": 7400, "loss": 0.00152, "loss_mean50": 0.002263, "lr": 1.526262584923003e-05, "grad_norm": 0.0199, "samples": 236800, "wall_s": 4600.2, "cuda_gb": 17.68} {"step": 7420, "loss": 0.002569, "loss_mean50": 0.002205, "lr": 1.5212579369155395e-05, "grad_norm": 0.0383, "samples": 237440, "wall_s": 4612.2, "cuda_gb": 17.68} {"step": 7440, "loss": 0.003433, "loss_mean50": 0.002135, "lr": 1.5162503429783722e-05, "grad_norm": 0.0819, "samples": 238080, "wall_s": 4624.2, "cuda_gb": 17.68} {"step": 7460, "loss": 0.000956, "loss_mean50": 0.002079, "lr": 1.5112399039745381e-05, "grad_norm": 0.0271, "samples": 238720, "wall_s": 4636.2, "cuda_gb": 17.68} {"step": 7480, "loss": 0.001014, "loss_mean50": 0.002063, "lr": 1.5062267208243797e-05, "grad_norm": 0.0182, "samples": 239360, "wall_s": 4648.2, "cuda_gb": 17.68} {"step": 7500, "loss": 0.001944, "loss_mean50": 0.002, "lr": 1.5012108945035122e-05, "grad_norm": 0.0195, "samples": 240000, "wall_s": 4660.2, "cuda_gb": 17.68} {"step": 7520, "loss": 0.000901, "loss_mean50": 0.001944, "lr": 1.49619252604079e-05, "grad_norm": 0.0176, "samples": 240640, "wall_s": 4714.6, "cuda_gb": 17.68} {"step": 7540, "loss": 0.001129, "loss_mean50": 0.001904, "lr": 1.49117171651627e-05, "grad_norm": 0.0213, "samples": 241280, "wall_s": 4726.6, "cuda_gb": 17.68} {"step": 7560, "loss": 0.000443, "loss_mean50": 0.001798, "lr": 1.4861485670591782e-05, "grad_norm": 0.0135, "samples": 241920, "wall_s": 4738.6, "cuda_gb": 17.68} {"step": 7580, "loss": 0.001693, "loss_mean50": 0.001942, "lr": 1.4811231788458716e-05, "grad_norm": 0.0493, "samples": 242560, "wall_s": 4750.5, "cuda_gb": 17.68} {"step": 7600, "loss": 0.000537, "loss_mean50": 0.001982, "lr": 1.4760956530978001e-05, "grad_norm": 0.0169, "samples": 243200, "wall_s": 4762.5, "cuda_gb": 17.68} {"step": 7620, "loss": 0.001929, "loss_mean50": 0.00226, "lr": 1.471066091079467e-05, "grad_norm": 0.0247, "samples": 243840, "wall_s": 4774.5, "cuda_gb": 17.68} {"step": 7640, "loss": 0.00177, "loss_mean50": 0.002406, "lr": 1.4660345940963918e-05, "grad_norm": 0.0261, "samples": 244480, "wall_s": 4786.5, "cuda_gb": 17.68} {"step": 7660, "loss": 0.000581, "loss_mean50": 0.002396, "lr": 1.4610012634930665e-05, "grad_norm": 0.0268, "samples": 245120, "wall_s": 4798.5, "cuda_gb": 17.68} {"step": 7680, "loss": 0.001571, "loss_mean50": 0.002182, "lr": 1.4559662006509167e-05, "grad_norm": 0.0204, "samples": 245760, "wall_s": 4810.2, "cuda_gb": 17.68} {"step": 7700, "loss": 0.003657, "loss_mean50": 0.002195, "lr": 1.4509295069862588e-05, "grad_norm": 0.1333, "samples": 246400, "wall_s": 4822.1, "cuda_gb": 17.68} {"step": 7720, "loss": 0.001547, "loss_mean50": 0.002017, "lr": 1.4458912839482571e-05, "grad_norm": 0.0245, "samples": 247040, "wall_s": 4834.1, "cuda_gb": 17.68} {"step": 7740, "loss": 0.003691, "loss_mean50": 0.001949, "lr": 1.4408516330168806e-05, "grad_norm": 0.0826, "samples": 247680, "wall_s": 4846.0, "cuda_gb": 17.68} {"step": 7760, "loss": 0.002558, "loss_mean50": 0.002013, "lr": 1.4358106557008589e-05, "grad_norm": 0.0379, "samples": 248320, "wall_s": 4858.0, "cuda_gb": 17.68} {"step": 7780, "loss": 0.001983, "loss_mean50": 0.00218, "lr": 1.4307684535356378e-05, "grad_norm": 0.0282, "samples": 248960, "wall_s": 4870.0, "cuda_gb": 17.68} {"step": 7800, "loss": 0.002167, "loss_mean50": 0.001946, "lr": 1.4257251280813339e-05, "grad_norm": 0.0156, "samples": 249600, "wall_s": 4882.0, "cuda_gb": 17.68} {"step": 7820, "loss": 0.000799, "loss_mean50": 0.001898, "lr": 1.420680780920689e-05, "grad_norm": 0.0168, "samples": 250240, "wall_s": 4894.0, "cuda_gb": 17.68} {"step": 7840, "loss": 0.003908, "loss_mean50": 0.001906, "lr": 1.4156355136570243e-05, "grad_norm": 0.0327, "samples": 250880, "wall_s": 4906.0, "cuda_gb": 17.68} {"step": 7860, "loss": 0.001353, "loss_mean50": 0.001883, "lr": 1.4105894279121939e-05, "grad_norm": 0.0321, "samples": 251520, "wall_s": 4918.0, "cuda_gb": 17.68} {"step": 7880, "loss": 0.001286, "loss_mean50": 0.001995, "lr": 1.4055426253245374e-05, "grad_norm": 0.0178, "samples": 252160, "wall_s": 4930.0, "cuda_gb": 17.68} {"step": 7900, "loss": 0.001807, "loss_mean50": 0.001943, "lr": 1.4004952075468333e-05, "grad_norm": 0.0314, "samples": 252800, "wall_s": 4942.0, "cuda_gb": 17.68} {"step": 7920, "loss": 0.002415, "loss_mean50": 0.002142, "lr": 1.3954472762442508e-05, "grad_norm": 0.0545, "samples": 253440, "wall_s": 4954.0, "cuda_gb": 17.68} {"step": 7940, "loss": 0.001834, "loss_mean50": 0.002211, "lr": 1.3903989330923037e-05, "grad_norm": 0.072, "samples": 254080, "wall_s": 4966.0, "cuda_gb": 17.68} {"step": 7960, "loss": 0.000535, "loss_mean50": 0.002068, "lr": 1.3853502797747997e-05, "grad_norm": 0.0114, "samples": 254720, "wall_s": 4978.0, "cuda_gb": 17.68} {"step": 7980, "loss": 0.001925, "loss_mean50": 0.001936, "lr": 1.3803014179817949e-05, "grad_norm": 0.0189, "samples": 255360, "wall_s": 4990.0, "cuda_gb": 17.68} {"step": 8000, "loss": 0.001369, "loss_mean50": 0.001935, "lr": 1.375252449407545e-05, "grad_norm": 0.0244, "samples": 256000, "wall_s": 5002.0, "cuda_gb": 17.68} {"step": 8000, "val_flow_mse": 0.002297} {"step": 8020, "loss": 0.00303, "loss_mean50": 0.002124, "lr": 1.3702034757484548e-05, "grad_norm": 0.0257, "samples": 256640, "wall_s": 5021.9, "cuda_gb": 17.68} {"step": 8040, "loss": 0.004307, "loss_mean50": 0.002343, "lr": 1.3651545987010325e-05, "grad_norm": 0.0239, "samples": 257280, "wall_s": 5033.9, "cuda_gb": 17.68} {"step": 8060, "loss": 0.002895, "loss_mean50": 0.002656, "lr": 1.3601059199598407e-05, "grad_norm": 0.0249, "samples": 257920, "wall_s": 5045.9, "cuda_gb": 17.68} {"step": 8080, "loss": 0.003548, "loss_mean50": 0.002189, "lr": 1.355057541215447e-05, "grad_norm": 0.0343, "samples": 258560, "wall_s": 5057.9, "cuda_gb": 17.68} {"step": 8100, "loss": 0.001801, "loss_mean50": 0.001926, "lr": 1.3500095641523772e-05, "grad_norm": 0.0284, "samples": 259200, "wall_s": 5069.9, "cuda_gb": 17.68} {"step": 8120, "loss": 0.00071, "loss_mean50": 0.001627, "lr": 1.3449620904470664e-05, "grad_norm": 0.0187, "samples": 259840, "wall_s": 5081.9, "cuda_gb": 17.68} {"step": 8140, "loss": 0.002995, "loss_mean50": 0.001914, "lr": 1.3399152217658095e-05, "grad_norm": 0.0239, "samples": 260480, "wall_s": 5093.9, "cuda_gb": 17.68} {"step": 8160, "loss": 0.002443, "loss_mean50": 0.001824, "lr": 1.3348690597627166e-05, "grad_norm": 0.0425, "samples": 261120, "wall_s": 5105.9, "cuda_gb": 17.68} {"step": 8180, "loss": 0.00215, "loss_mean50": 0.001814, "lr": 1.3298237060776639e-05, "grad_norm": 0.0378, "samples": 261760, "wall_s": 5117.9, "cuda_gb": 17.68} {"step": 8200, "loss": 0.001742, "loss_mean50": 0.001911, "lr": 1.324779262334246e-05, "grad_norm": 0.0183, "samples": 262400, "wall_s": 5129.9, "cuda_gb": 17.68} {"step": 8220, "loss": 0.002039, "loss_mean50": 0.001831, "lr": 1.3197358301377289e-05, "grad_norm": 0.0706, "samples": 263040, "wall_s": 5141.9, "cuda_gb": 17.68} {"step": 8240, "loss": 0.004531, "loss_mean50": 0.001892, "lr": 1.3146935110730046e-05, "grad_norm": 0.0777, "samples": 263680, "wall_s": 5153.9, "cuda_gb": 17.68} {"step": 8260, "loss": 0.001131, "loss_mean50": 0.002019, "lr": 1.3096524067025451e-05, "grad_norm": 0.0249, "samples": 264320, "wall_s": 5165.9, "cuda_gb": 17.68} {"step": 8280, "loss": 0.000817, "loss_mean50": 0.002187, "lr": 1.3046126185643545e-05, "grad_norm": 0.0197, "samples": 264960, "wall_s": 5177.9, "cuda_gb": 17.68} {"step": 8300, "loss": 0.003003, "loss_mean50": 0.002143, "lr": 1.2995742481699261e-05, "grad_norm": 0.0332, "samples": 265600, "wall_s": 5190.0, "cuda_gb": 17.68} {"step": 8320, "loss": 0.001648, "loss_mean50": 0.002102, "lr": 1.2945373970021971e-05, "grad_norm": 0.0349, "samples": 266240, "wall_s": 5201.9, "cuda_gb": 17.68} {"step": 8340, "loss": 0.001557, "loss_mean50": 0.001972, "lr": 1.2895021665135043e-05, "grad_norm": 0.021, "samples": 266880, "wall_s": 5213.9, "cuda_gb": 17.68} {"step": 8360, "loss": 0.001754, "loss_mean50": 0.001815, "lr": 1.2844686581235405e-05, "grad_norm": 0.0164, "samples": 267520, "wall_s": 5225.9, "cuda_gb": 17.68} {"step": 8380, "loss": 0.001463, "loss_mean50": 0.001538, "lr": 1.2794369732173127e-05, "grad_norm": 0.0298, "samples": 268160, "wall_s": 5237.9, "cuda_gb": 17.68} {"step": 8400, "loss": 0.001409, "loss_mean50": 0.001884, "lr": 1.2744072131430981e-05, "grad_norm": 0.0613, "samples": 268800, "wall_s": 5249.9, "cuda_gb": 17.68} {"step": 8420, "loss": 0.00142, "loss_mean50": 0.001958, "lr": 1.2693794792104049e-05, "grad_norm": 0.0394, "samples": 269440, "wall_s": 5261.9, "cuda_gb": 17.68} {"step": 8440, "loss": 0.001641, "loss_mean50": 0.001831, "lr": 1.2643538726879303e-05, "grad_norm": 0.0328, "samples": 270080, "wall_s": 5273.9, "cuda_gb": 17.68} {"step": 8460, "loss": 0.001376, "loss_mean50": 0.001844, "lr": 1.2593304948015212e-05, "grad_norm": 0.0303, "samples": 270720, "wall_s": 5285.9, "cuda_gb": 17.68} {"step": 8480, "loss": 0.000841, "loss_mean50": 0.001793, "lr": 1.254309446732135e-05, "grad_norm": 0.0286, "samples": 271360, "wall_s": 5297.9, "cuda_gb": 17.68} {"step": 8500, "loss": 0.001162, "loss_mean50": 0.001817, "lr": 1.2492908296138028e-05, "grad_norm": 0.025, "samples": 272000, "wall_s": 5309.9, "cuda_gb": 17.68} {"step": 8520, "loss": 0.002147, "loss_mean50": 0.001857, "lr": 1.2442747445315909e-05, "grad_norm": 0.0258, "samples": 272640, "wall_s": 5321.9, "cuda_gb": 17.68} {"step": 8540, "loss": 0.004053, "loss_mean50": 0.001919, "lr": 1.2392612925195645e-05, "grad_norm": 0.036, "samples": 273280, "wall_s": 5333.9, "cuda_gb": 17.68} {"step": 8560, "loss": 0.003135, "loss_mean50": 0.002169, "lr": 1.2342505745587546e-05, "grad_norm": 0.0181, "samples": 273920, "wall_s": 5346.0, "cuda_gb": 17.68} {"step": 8580, "loss": 0.001751, "loss_mean50": 0.002078, "lr": 1.2292426915751223e-05, "grad_norm": 0.0221, "samples": 274560, "wall_s": 5357.9, "cuda_gb": 17.68} {"step": 8600, "loss": 0.003423, "loss_mean50": 0.002058, "lr": 1.2242377444375272e-05, "grad_norm": 0.046, "samples": 275200, "wall_s": 5369.9, "cuda_gb": 17.68} {"step": 8620, "loss": 0.001711, "loss_mean50": 0.001806, "lr": 1.2192358339556945e-05, "grad_norm": 0.0316, "samples": 275840, "wall_s": 5381.6, "cuda_gb": 17.68} {"step": 8640, "loss": 0.004367, "loss_mean50": 0.001651, "lr": 1.2142370608781848e-05, "grad_norm": 0.041, "samples": 276480, "wall_s": 5393.6, "cuda_gb": 17.68} {"step": 8660, "loss": 0.000928, "loss_mean50": 0.001787, "lr": 1.209241525890365e-05, "grad_norm": 0.0282, "samples": 277120, "wall_s": 5405.6, "cuda_gb": 17.68} {"step": 8680, "loss": 0.001003, "loss_mean50": 0.001982, "lr": 1.204249329612381e-05, "grad_norm": 0.0165, "samples": 277760, "wall_s": 5417.6, "cuda_gb": 17.68} {"step": 8700, "loss": 0.000338, "loss_mean50": 0.001936, "lr": 1.1992605725971308e-05, "grad_norm": 0.0135, "samples": 278400, "wall_s": 5429.5, "cuda_gb": 17.68} {"step": 8720, "loss": 0.000853, "loss_mean50": 0.001953, "lr": 1.194275355328237e-05, "grad_norm": 0.0208, "samples": 279040, "wall_s": 5441.5, "cuda_gb": 17.68} {"step": 8740, "loss": 0.001111, "loss_mean50": 0.00183, "lr": 1.1892937782180261e-05, "grad_norm": 0.0175, "samples": 279680, "wall_s": 5453.5, "cuda_gb": 17.68} {"step": 8760, "loss": 0.001587, "loss_mean50": 0.001844, "lr": 1.1843159416055037e-05, "grad_norm": 0.0335, "samples": 280320, "wall_s": 5465.4, "cuda_gb": 17.68} {"step": 8780, "loss": 0.004778, "loss_mean50": 0.001722, "lr": 1.1793419457543346e-05, "grad_norm": 0.0269, "samples": 280960, "wall_s": 5477.4, "cuda_gb": 17.68} {"step": 8800, "loss": 0.000415, "loss_mean50": 0.001622, "lr": 1.1743718908508222e-05, "grad_norm": 0.0118, "samples": 281600, "wall_s": 5489.4, "cuda_gb": 17.68} {"step": 8820, "loss": 0.001871, "loss_mean50": 0.001681, "lr": 1.1694058770018925e-05, "grad_norm": 0.0345, "samples": 282240, "wall_s": 5501.5, "cuda_gb": 17.68} {"step": 8840, "loss": 0.002364, "loss_mean50": 0.001976, "lr": 1.1644440042330754e-05, "grad_norm": 0.0265, "samples": 282880, "wall_s": 5513.5, "cuda_gb": 17.68} {"step": 8860, "loss": 0.002428, "loss_mean50": 0.002224, "lr": 1.1594863724864918e-05, "grad_norm": 0.0353, "samples": 283520, "wall_s": 5525.4, "cuda_gb": 17.68} {"step": 8880, "loss": 0.001295, "loss_mean50": 0.002288, "lr": 1.1545330816188398e-05, "grad_norm": 0.027, "samples": 284160, "wall_s": 5537.4, "cuda_gb": 17.68} {"step": 8900, "loss": 0.000288, "loss_mean50": 0.002028, "lr": 1.1495842313993832e-05, "grad_norm": 0.0126, "samples": 284800, "wall_s": 5549.4, "cuda_gb": 17.68} {"step": 8920, "loss": 0.003179, "loss_mean50": 0.00164, "lr": 1.1446399215079417e-05, "grad_norm": 0.0282, "samples": 285440, "wall_s": 5561.4, "cuda_gb": 17.68} {"step": 8940, "loss": 0.002834, "loss_mean50": 0.001839, "lr": 1.1397002515328845e-05, "grad_norm": 0.0333, "samples": 286080, "wall_s": 5573.4, "cuda_gb": 17.68} {"step": 8960, "loss": 0.001579, "loss_mean50": 0.001956, "lr": 1.1347653209691237e-05, "grad_norm": 0.0427, "samples": 286720, "wall_s": 5585.3, "cuda_gb": 17.68} {"step": 8980, "loss": 0.001204, "loss_mean50": 0.001965, "lr": 1.1298352292161085e-05, "grad_norm": 0.0168, "samples": 287360, "wall_s": 5597.3, "cuda_gb": 17.68} {"step": 9000, "loss": 0.002249, "loss_mean50": 0.001974, "lr": 1.1249100755758268e-05, "grad_norm": 0.0211, "samples": 288000, "wall_s": 5609.3, "cuda_gb": 17.68} {"step": 9000, "val_flow_mse": 0.00226} {"step": 9020, "loss": 0.003941, "loss_mean50": 0.00192, "lr": 1.1199899592508024e-05, "grad_norm": 0.0317, "samples": 288640, "wall_s": 5629.2, "cuda_gb": 17.68} {"step": 9040, "loss": 0.000887, "loss_mean50": 0.001976, "lr": 1.1150749793420958e-05, "grad_norm": 0.0239, "samples": 289280, "wall_s": 5641.0, "cuda_gb": 17.68} {"step": 9060, "loss": 0.000684, "loss_mean50": 0.001673, "lr": 1.1101652348473113e-05, "grad_norm": 0.0182, "samples": 289920, "wall_s": 5652.9, "cuda_gb": 17.68} {"step": 9080, "loss": 0.001908, "loss_mean50": 0.001588, "lr": 1.1052608246586007e-05, "grad_norm": 0.0282, "samples": 290560, "wall_s": 5664.9, "cuda_gb": 17.68} {"step": 9100, "loss": 0.004542, "loss_mean50": 0.001894, "lr": 1.1003618475606737e-05, "grad_norm": 0.04, "samples": 291200, "wall_s": 5676.9, "cuda_gb": 17.68} {"step": 9120, "loss": 0.000751, "loss_mean50": 0.00203, "lr": 1.0954684022288039e-05, "grad_norm": 0.0212, "samples": 291840, "wall_s": 5688.9, "cuda_gb": 17.68} {"step": 9140, "loss": 0.002224, "loss_mean50": 0.001928, "lr": 1.0905805872268463e-05, "grad_norm": 0.029, "samples": 292480, "wall_s": 5700.9, "cuda_gb": 17.68} {"step": 9160, "loss": 0.000535, "loss_mean50": 0.001986, "lr": 1.0856985010052487e-05, "grad_norm": 0.0163, "samples": 293120, "wall_s": 5712.8, "cuda_gb": 17.68} {"step": 9180, "loss": 0.001232, "loss_mean50": 0.001905, "lr": 1.0808222418990693e-05, "grad_norm": 0.0305, "samples": 293760, "wall_s": 5724.8, "cuda_gb": 17.68} {"step": 9200, "loss": 0.002555, "loss_mean50": 0.001858, "lr": 1.0759519081259975e-05, "grad_norm": 0.024, "samples": 294400, "wall_s": 5736.8, "cuda_gb": 17.68} {"step": 9220, "loss": 0.002128, "loss_mean50": 0.001907, "lr": 1.071087597784374e-05, "grad_norm": 0.0308, "samples": 295040, "wall_s": 5748.8, "cuda_gb": 17.68} {"step": 9240, "loss": 0.001775, "loss_mean50": 0.00206, "lr": 1.0662294088512154e-05, "grad_norm": 0.0262, "samples": 295680, "wall_s": 5760.8, "cuda_gb": 17.68} {"step": 9260, "loss": 0.001177, "loss_mean50": 0.001851, "lr": 1.0613774391802405e-05, "grad_norm": 0.0288, "samples": 296320, "wall_s": 5772.8, "cuda_gb": 17.68} {"step": 9280, "loss": 0.00335, "loss_mean50": 0.00169, "lr": 1.0565317864999011e-05, "grad_norm": 0.0675, "samples": 296960, "wall_s": 5784.8, "cuda_gb": 17.68} {"step": 9300, "loss": 0.003068, "loss_mean50": 0.001745, "lr": 1.0516925484114096e-05, "grad_norm": 0.0627, "samples": 297600, "wall_s": 5796.8, "cuda_gb": 17.68} {"step": 9320, "loss": 0.001896, "loss_mean50": 0.001929, "lr": 1.0468598223867779e-05, "grad_norm": 0.0273, "samples": 298240, "wall_s": 5808.8, "cuda_gb": 17.68} {"step": 9340, "loss": 0.003048, "loss_mean50": 0.002069, "lr": 1.04203370576685e-05, "grad_norm": 0.0532, "samples": 298880, "wall_s": 5820.8, "cuda_gb": 17.68} {"step": 9360, "loss": 0.001684, "loss_mean50": 0.001877, "lr": 1.037214295759345e-05, "grad_norm": 0.022, "samples": 299520, "wall_s": 5832.8, "cuda_gb": 17.68} {"step": 9380, "loss": 0.001352, "loss_mean50": 0.001776, "lr": 1.0324016894368957e-05, "grad_norm": 0.0218, "samples": 300160, "wall_s": 5844.8, "cuda_gb": 17.68} {"step": 9400, "loss": 0.001642, "loss_mean50": 0.001863, "lr": 1.0275959837350957e-05, "grad_norm": 0.0232, "samples": 300800, "wall_s": 5856.7, "cuda_gb": 17.68} {"step": 9420, "loss": 0.001708, "loss_mean50": 0.001954, "lr": 1.0227972754505468e-05, "grad_norm": 0.0195, "samples": 301440, "wall_s": 5868.7, "cuda_gb": 17.68} {"step": 9440, "loss": 0.001098, "loss_mean50": 0.001968, "lr": 1.0180056612389072e-05, "grad_norm": 0.0203, "samples": 302080, "wall_s": 5880.6, "cuda_gb": 17.68} {"step": 9460, "loss": 0.001919, "loss_mean50": 0.00206, "lr": 1.0132212376129465e-05, "grad_norm": 0.0276, "samples": 302720, "wall_s": 5892.4, "cuda_gb": 17.68} {"step": 9480, "loss": 0.001435, "loss_mean50": 0.001973, "lr": 1.0084441009406028e-05, "grad_norm": 0.0124, "samples": 303360, "wall_s": 5904.4, "cuda_gb": 17.68} {"step": 9500, "loss": 0.002137, "loss_mean50": 0.001808, "lr": 1.0036743474430393e-05, "grad_norm": 0.0222, "samples": 304000, "wall_s": 5916.4, "cuda_gb": 17.68} {"step": 9520, "loss": 0.002185, "loss_mean50": 0.001845, "lr": 9.989120731927065e-06, "grad_norm": 0.0352, "samples": 304640, "wall_s": 5928.4, "cuda_gb": 17.68} {"step": 9540, "loss": 0.000426, "loss_mean50": 0.001808, "lr": 9.941573741114099e-06, "grad_norm": 0.0199, "samples": 305280, "wall_s": 5940.4, "cuda_gb": 17.68} {"step": 9560, "loss": 0.00171, "loss_mean50": 0.001627, "lr": 9.894103459683726e-06, "grad_norm": 0.0161, "samples": 305920, "wall_s": 5952.4, "cuda_gb": 17.68} {"step": 9580, "loss": 0.000648, "loss_mean50": 0.00146, "lr": 9.846710843783125e-06, "grad_norm": 0.025, "samples": 306560, "wall_s": 5964.4, "cuda_gb": 17.68} {"step": 9600, "loss": 0.001599, "loss_mean50": 0.001758, "lr": 9.799396847995121e-06, "grad_norm": 0.0297, "samples": 307200, "wall_s": 5976.4, "cuda_gb": 17.68} {"step": 9620, "loss": 0.001433, "loss_mean50": 0.001721, "lr": 9.752162425318972e-06, "grad_norm": 0.021, "samples": 307840, "wall_s": 5988.4, "cuda_gb": 17.68} {"step": 9640, "loss": 0.003749, "loss_mean50": 0.001798, "lr": 9.705008527151176e-06, "grad_norm": 0.0389, "samples": 308480, "wall_s": 6000.4, "cuda_gb": 17.68} {"step": 9660, "loss": 0.002605, "loss_mean50": 0.001787, "lr": 9.657936103266308e-06, "grad_norm": 0.0247, "samples": 309120, "wall_s": 6012.4, "cuda_gb": 17.68} {"step": 9680, "loss": 0.001451, "loss_mean50": 0.00188, "lr": 9.61094610179788e-06, "grad_norm": 0.0305, "samples": 309760, "wall_s": 6024.4, "cuda_gb": 17.68} {"step": 9700, "loss": 0.002194, "loss_mean50": 0.001714, "lr": 9.564039469219244e-06, "grad_norm": 0.0281, "samples": 310400, "wall_s": 6036.4, "cuda_gb": 17.68} {"step": 9720, "loss": 0.001258, "loss_mean50": 0.001723, "lr": 9.517217150324556e-06, "grad_norm": 0.0243, "samples": 311040, "wall_s": 6048.4, "cuda_gb": 17.68} {"step": 9740, "loss": 0.001469, "loss_mean50": 0.001593, "lr": 9.470480088209707e-06, "grad_norm": 0.0278, "samples": 311680, "wall_s": 6060.3, "cuda_gb": 17.68} {"step": 9760, "loss": 0.00166, "loss_mean50": 0.001696, "lr": 9.423829224253347e-06, "grad_norm": 0.021, "samples": 312320, "wall_s": 6072.3, "cuda_gb": 17.68} {"step": 9780, "loss": 0.002283, "loss_mean50": 0.001919, "lr": 9.377265498097926e-06, "grad_norm": 0.0381, "samples": 312960, "wall_s": 6084.3, "cuda_gb": 17.68} {"step": 9800, "loss": 0.003862, "loss_mean50": 0.001891, "lr": 9.330789847630767e-06, "grad_norm": 0.0348, "samples": 313600, "wall_s": 6096.3, "cuda_gb": 17.68} {"step": 9820, "loss": 0.00127, "loss_mean50": 0.001974, "lr": 9.284403208965151e-06, "grad_norm": 0.0133, "samples": 314240, "wall_s": 6108.3, "cuda_gb": 17.68} {"step": 9840, "loss": 0.001663, "loss_mean50": 0.002089, "lr": 9.238106516421504e-06, "grad_norm": 0.0283, "samples": 314880, "wall_s": 6120.2, "cuda_gb": 17.68} {"step": 9860, "loss": 0.002138, "loss_mean50": 0.00203, "lr": 9.191900702508551e-06, "grad_norm": 0.0206, "samples": 315520, "wall_s": 6132.2, "cuda_gb": 17.68} {"step": 9880, "loss": 0.001166, "loss_mean50": 0.001864, "lr": 9.145786697904526e-06, "grad_norm": 0.0304, "samples": 316160, "wall_s": 6144.2, "cuda_gb": 17.68} {"step": 9900, "loss": 0.002306, "loss_mean50": 0.001746, "lr": 9.099765431438454e-06, "grad_norm": 0.0365, "samples": 316800, "wall_s": 6156.2, "cuda_gb": 17.68} {"step": 9920, "loss": 0.000618, "loss_mean50": 0.001748, "lr": 9.053837830071418e-06, "grad_norm": 0.0111, "samples": 317440, "wall_s": 6168.2, "cuda_gb": 17.68} {"step": 9940, "loss": 0.000932, "loss_mean50": 0.001887, "lr": 9.008004818877899e-06, "grad_norm": 0.0181, "samples": 318080, "wall_s": 6180.2, "cuda_gb": 17.68} {"step": 9960, "loss": 0.000704, "loss_mean50": 0.001856, "lr": 8.96226732102714e-06, "grad_norm": 0.0189, "samples": 318720, "wall_s": 6192.0, "cuda_gb": 17.68} {"step": 9980, "loss": 0.002154, "loss_mean50": 0.001851, "lr": 8.91662625776456e-06, "grad_norm": 0.0279, "samples": 319360, "wall_s": 6204.0, "cuda_gb": 17.68} {"step": 10000, "loss": 0.000785, "loss_mean50": 0.001747, "lr": 8.87108254839317e-06, "grad_norm": 0.0342, "samples": 320000, "wall_s": 6216.1, "cuda_gb": 17.68} {"step": 10000, "val_flow_mse": 0.002025, "val_action_mae": 0.010183, "best": true} {"step": 10020, "loss": 0.001874, "loss_mean50": 0.001707, "lr": 8.825637110255108e-06, "grad_norm": 0.0386, "samples": 320640, "wall_s": 6303.0, "cuda_gb": 17.68} {"step": 10040, "loss": 0.00089, "loss_mean50": 0.001705, "lr": 8.78029085871311e-06, "grad_norm": 0.0216, "samples": 321280, "wall_s": 6314.9, "cuda_gb": 17.68} {"step": 10060, "loss": 0.000957, "loss_mean50": 0.001572, "lr": 8.7350447071321e-06, "grad_norm": 0.0186, "samples": 321920, "wall_s": 6326.9, "cuda_gb": 17.68} {"step": 10080, "loss": 0.002322, "loss_mean50": 0.001747, "lr": 8.68989956686079e-06, "grad_norm": 0.0385, "samples": 322560, "wall_s": 6338.9, "cuda_gb": 17.68} {"step": 10100, "loss": 0.002385, "loss_mean50": 0.001713, "lr": 8.644856347213308e-06, "grad_norm": 0.0284, "samples": 323200, "wall_s": 6350.9, "cuda_gb": 17.68} {"step": 10120, "loss": 0.001965, "loss_mean50": 0.001729, "lr": 8.59991595545091e-06, "grad_norm": 0.0229, "samples": 323840, "wall_s": 6362.9, "cuda_gb": 17.68} {"step": 10140, "loss": 0.001243, "loss_mean50": 0.001667, "lr": 8.555079296763687e-06, "grad_norm": 0.0234, "samples": 324480, "wall_s": 6374.9, "cuda_gb": 17.68} {"step": 10160, "loss": 0.001122, "loss_mean50": 0.00157, "lr": 8.510347274252333e-06, "grad_norm": 0.0295, "samples": 325120, "wall_s": 6386.9, "cuda_gb": 17.68} {"step": 10180, "loss": 0.000997, "loss_mean50": 0.001515, "lr": 8.465720788909959e-06, "grad_norm": 0.0347, "samples": 325760, "wall_s": 6398.9, "cuda_gb": 17.68} {"step": 10200, "loss": 0.001977, "loss_mean50": 0.001689, "lr": 8.421200739603955e-06, "grad_norm": 0.0425, "samples": 326400, "wall_s": 6410.9, "cuda_gb": 17.68} {"step": 10220, "loss": 0.000458, "loss_mean50": 0.001532, "lr": 8.376788023057854e-06, "grad_norm": 0.0091, "samples": 327040, "wall_s": 6422.8, "cuda_gb": 17.68} {"step": 10240, "loss": 0.002178, "loss_mean50": 0.00157, "lr": 8.33248353383331e-06, "grad_norm": 0.0452, "samples": 327680, "wall_s": 6434.8, "cuda_gb": 17.68} {"step": 10260, "loss": 0.001004, "loss_mean50": 0.001493, "lr": 8.288288164312053e-06, "grad_norm": 0.0141, "samples": 328320, "wall_s": 6446.8, "cuda_gb": 17.68} {"step": 10280, "loss": 0.001043, "loss_mean50": 0.001655, "lr": 8.244202804677917e-06, "grad_norm": 0.0176, "samples": 328960, "wall_s": 6458.8, "cuda_gb": 17.68} {"step": 10300, "loss": 0.002508, "loss_mean50": 0.001573, "lr": 8.200228342898927e-06, "grad_norm": 0.0252, "samples": 329600, "wall_s": 6471.9, "cuda_gb": 17.68} {"step": 10320, "loss": 0.000624, "loss_mean50": 0.001716, "lr": 8.1563656647094e-06, "grad_norm": 0.0128, "samples": 330240, "wall_s": 6483.9, "cuda_gb": 17.68} {"step": 10340, "loss": 0.001565, "loss_mean50": 0.001595, "lr": 8.112615653592086e-06, "grad_norm": 0.0385, "samples": 330880, "wall_s": 6496.0, "cuda_gb": 17.68} {"step": 10360, "loss": 0.000842, "loss_mean50": 0.001404, "lr": 8.06897919076042e-06, "grad_norm": 0.0133, "samples": 331520, "wall_s": 6508.0, "cuda_gb": 17.68} {"step": 10380, "loss": 0.001976, "loss_mean50": 0.001463, "lr": 8.025457155140734e-06, "grad_norm": 0.0223, "samples": 332160, "wall_s": 6520.0, "cuda_gb": 17.68} {"step": 10400, "loss": 0.000679, "loss_mean50": 0.001701, "lr": 7.982050423354554e-06, "grad_norm": 0.022, "samples": 332800, "wall_s": 6532.0, "cuda_gb": 17.68} {"step": 10420, "loss": 0.002811, "loss_mean50": 0.00175, "lr": 7.938759869700983e-06, "grad_norm": 0.0222, "samples": 333440, "wall_s": 6544.0, "cuda_gb": 17.68} {"step": 10440, "loss": 0.00046, "loss_mean50": 0.00199, "lr": 7.895586366139032e-06, "grad_norm": 0.0242, "samples": 334080, "wall_s": 6556.0, "cuda_gb": 17.68} {"step": 10460, "loss": 0.001014, "loss_mean50": 0.001904, "lr": 7.852530782270106e-06, "grad_norm": 0.0249, "samples": 334720, "wall_s": 6568.0, "cuda_gb": 17.68} {"step": 10480, "loss": 0.003425, "loss_mean50": 0.001822, "lr": 7.809593985320462e-06, "grad_norm": 0.0414, "samples": 335360, "wall_s": 6580.1, "cuda_gb": 17.68} {"step": 10500, "loss": 0.003121, "loss_mean50": 0.001712, "lr": 7.766776840123753e-06, "grad_norm": 0.0327, "samples": 336000, "wall_s": 6592.2, "cuda_gb": 17.68} {"step": 10520, "loss": 0.001913, "loss_mean50": 0.001549, "lr": 7.724080209103593e-06, "grad_norm": 0.0833, "samples": 336640, "wall_s": 6604.1, "cuda_gb": 17.68} {"step": 10540, "loss": 0.00095, "loss_mean50": 0.001659, "lr": 7.681504952256218e-06, "grad_norm": 0.0216, "samples": 337280, "wall_s": 6616.2, "cuda_gb": 17.68} {"step": 10560, "loss": 0.001153, "loss_mean50": 0.001679, "lr": 7.639051927133128e-06, "grad_norm": 0.0225, "samples": 337920, "wall_s": 6628.2, "cuda_gb": 17.68} {"step": 10580, "loss": 0.000524, "loss_mean50": 0.001679, "lr": 7.596721988823835e-06, "grad_norm": 0.0261, "samples": 338560, "wall_s": 6640.2, "cuda_gb": 17.68} {"step": 10600, "loss": 0.00076, "loss_mean50": 0.001689, "lr": 7.554515989938631e-06, "grad_norm": 0.0183, "samples": 339200, "wall_s": 6652.2, "cuda_gb": 17.68} {"step": 10620, "loss": 0.002623, "loss_mean50": 0.001788, "lr": 7.512434780591415e-06, "grad_norm": 0.0289, "samples": 339840, "wall_s": 6664.2, "cuda_gb": 17.68} {"step": 10640, "loss": 0.00383, "loss_mean50": 0.001697, "lr": 7.470479208382591e-06, "grad_norm": 0.0346, "samples": 340480, "wall_s": 6676.1, "cuda_gb": 17.68} {"step": 10660, "loss": 0.000739, "loss_mean50": 0.001665, "lr": 7.4286501183819604e-06, "grad_norm": 0.0192, "samples": 341120, "wall_s": 6688.3, "cuda_gb": 17.68} {"step": 10680, "loss": 0.00127, "loss_mean50": 0.001512, "lr": 7.386948353111717e-06, "grad_norm": 0.0849, "samples": 341760, "wall_s": 6700.2, "cuda_gb": 17.68} {"step": 10700, "loss": 0.002947, "loss_mean50": 0.00158, "lr": 7.345374752529491e-06, "grad_norm": 0.0414, "samples": 342400, "wall_s": 6712.3, "cuda_gb": 17.68} {"step": 10720, "loss": 0.002194, "loss_mean50": 0.001683, "lr": 7.30393015401141e-06, "grad_norm": 0.0129, "samples": 343040, "wall_s": 6724.4, "cuda_gb": 17.68} {"step": 10740, "loss": 0.001155, "loss_mean50": 0.001794, "lr": 7.2626153923352316e-06, "grad_norm": 0.0295, "samples": 343680, "wall_s": 6736.4, "cuda_gb": 17.68} {"step": 10760, "loss": 0.003522, "loss_mean50": 0.001962, "lr": 7.221431299663555e-06, "grad_norm": 0.1063, "samples": 344320, "wall_s": 6748.4, "cuda_gb": 17.68} {"step": 10780, "loss": 0.000872, "loss_mean50": 0.00183, "lr": 7.18037870552703e-06, "grad_norm": 0.028, "samples": 344960, "wall_s": 6760.5, "cuda_gb": 17.68} {"step": 10800, "loss": 0.003233, "loss_mean50": 0.001821, "lr": 7.139458436807661e-06, "grad_norm": 0.0393, "samples": 345600, "wall_s": 6772.5, "cuda_gb": 17.68} {"step": 10820, "loss": 0.001928, "loss_mean50": 0.001738, "lr": 7.098671317722161e-06, "grad_norm": 0.0725, "samples": 346240, "wall_s": 6784.4, "cuda_gb": 17.68} {"step": 10840, "loss": 0.001013, "loss_mean50": 0.00179, "lr": 7.05801816980533e-06, "grad_norm": 0.0316, "samples": 346880, "wall_s": 6796.4, "cuda_gb": 17.68} {"step": 10860, "loss": 0.002077, "loss_mean50": 0.001841, "lr": 7.0174998118935295e-06, "grad_norm": 0.0214, "samples": 347520, "wall_s": 6808.5, "cuda_gb": 17.68} {"step": 10880, "loss": 0.001841, "loss_mean50": 0.001844, "lr": 6.977117060108165e-06, "grad_norm": 0.0509, "samples": 348160, "wall_s": 6820.6, "cuda_gb": 17.68} {"step": 10900, "loss": 0.00125, "loss_mean50": 0.001701, "lr": 6.9368707278392706e-06, "grad_norm": 0.0122, "samples": 348800, "wall_s": 6832.5, "cuda_gb": 17.68} {"step": 10920, "loss": 0.000752, "loss_mean50": 0.00169, "lr": 6.896761625729117e-06, "grad_norm": 0.0166, "samples": 349440, "wall_s": 6844.6, "cuda_gb": 17.68} {"step": 10940, "loss": 0.001794, "loss_mean50": 0.001601, "lr": 6.856790561655879e-06, "grad_norm": 0.0242, "samples": 350080, "wall_s": 6856.5, "cuda_gb": 17.68} {"step": 10960, "loss": 0.001679, "loss_mean50": 0.001498, "lr": 6.816958340717365e-06, "grad_norm": 0.0992, "samples": 350720, "wall_s": 6868.5, "cuda_gb": 17.68} {"step": 10980, "loss": 0.002031, "loss_mean50": 0.001747, "lr": 6.777265765214818e-06, "grad_norm": 0.0504, "samples": 351360, "wall_s": 6880.7, "cuda_gb": 17.68} {"step": 11000, "loss": 0.001054, "loss_mean50": 0.001931, "lr": 6.737713634636714e-06, "grad_norm": 0.0281, "samples": 352000, "wall_s": 6892.6, "cuda_gb": 17.68} {"step": 11000, "val_flow_mse": 0.001962} {"step": 11020, "loss": 0.002578, "loss_mean50": 0.001952, "lr": 6.698302745642702e-06, "grad_norm": 0.0176, "samples": 352640, "wall_s": 6912.6, "cuda_gb": 17.68} {"step": 11040, "loss": 0.001462, "loss_mean50": 0.001819, "lr": 6.659033892047546e-06, "grad_norm": 0.0237, "samples": 353280, "wall_s": 6924.6, "cuda_gb": 17.68} {"step": 11060, "loss": 0.002431, "loss_mean50": 0.001883, "lr": 6.619907864805116e-06, "grad_norm": 0.0409, "samples": 353920, "wall_s": 6936.1, "cuda_gb": 17.68} {"step": 11080, "loss": 0.00321, "loss_mean50": 0.001962, "lr": 6.580925451992481e-06, "grad_norm": 0.0492, "samples": 354560, "wall_s": 6947.7, "cuda_gb": 17.68} {"step": 11100, "loss": 0.001183, "loss_mean50": 0.001732, "lr": 6.542087438794027e-06, "grad_norm": 0.0161, "samples": 355200, "wall_s": 6959.6, "cuda_gb": 17.68} {"step": 11120, "loss": 0.001382, "loss_mean50": 0.001717, "lr": 6.503394607485626e-06, "grad_norm": 0.0221, "samples": 355840, "wall_s": 6971.6, "cuda_gb": 17.68} {"step": 11140, "loss": 0.000912, "loss_mean50": 0.001675, "lr": 6.464847737418916e-06, "grad_norm": 0.0174, "samples": 356480, "wall_s": 6983.6, "cuda_gb": 17.68} {"step": 11160, "loss": 0.002287, "loss_mean50": 0.001558, "lr": 6.426447605005567e-06, "grad_norm": 0.0254, "samples": 357120, "wall_s": 6995.6, "cuda_gb": 17.68} {"step": 11180, "loss": 0.002523, "loss_mean50": 0.00184, "lr": 6.388194983701658e-06, "grad_norm": 0.1025, "samples": 357760, "wall_s": 7007.7, "cuda_gb": 17.68} {"step": 11200, "loss": 0.002494, "loss_mean50": 0.001785, "lr": 6.350090643992096e-06, "grad_norm": 0.0238, "samples": 358400, "wall_s": 7019.6, "cuda_gb": 17.68} {"step": 11220, "loss": 0.001461, "loss_mean50": 0.001737, "lr": 6.312135353375108e-06, "grad_norm": 0.0462, "samples": 359040, "wall_s": 7031.7, "cuda_gb": 17.68} {"step": 11240, "loss": 0.002173, "loss_mean50": 0.001629, "lr": 6.274329876346763e-06, "grad_norm": 0.018, "samples": 359680, "wall_s": 7043.5, "cuda_gb": 17.68} {"step": 11260, "loss": 0.002258, "loss_mean50": 0.001574, "lr": 6.2366749743855796e-06, "grad_norm": 0.0186, "samples": 360320, "wall_s": 7055.5, "cuda_gb": 17.68} {"step": 11280, "loss": 0.001732, "loss_mean50": 0.001603, "lr": 6.199171405937199e-06, "grad_norm": 0.0485, "samples": 360960, "wall_s": 7067.5, "cuda_gb": 17.68} {"step": 11300, "loss": 0.003445, "loss_mean50": 0.001822, "lr": 6.161819926399091e-06, "grad_norm": 0.0543, "samples": 361600, "wall_s": 7079.5, "cuda_gb": 17.68} {"step": 11320, "loss": 0.001369, "loss_mean50": 0.002044, "lr": 6.124621288105361e-06, "grad_norm": 0.016, "samples": 362240, "wall_s": 7091.4, "cuda_gb": 17.68} {"step": 11340, "loss": 0.001815, "loss_mean50": 0.001847, "lr": 6.0875762403115745e-06, "grad_norm": 0.0205, "samples": 362880, "wall_s": 7103.5, "cuda_gb": 17.68} {"step": 11360, "loss": 0.001555, "loss_mean50": 0.001584, "lr": 6.050685529179673e-06, "grad_norm": 0.0191, "samples": 363520, "wall_s": 7115.4, "cuda_gb": 17.68} {"step": 11380, "loss": 0.002117, "loss_mean50": 0.001703, "lr": 6.013949897762954e-06, "grad_norm": 0.042, "samples": 364160, "wall_s": 7127.4, "cuda_gb": 17.68} {"step": 11400, "loss": 0.000894, "loss_mean50": 0.001605, "lr": 5.977370085991092e-06, "grad_norm": 0.0251, "samples": 364800, "wall_s": 7139.5, "cuda_gb": 17.68} {"step": 11420, "loss": 0.002543, "loss_mean50": 0.001816, "lr": 5.9409468306552325e-06, "grad_norm": 0.0385, "samples": 365440, "wall_s": 7151.4, "cuda_gb": 17.68} {"step": 11440, "loss": 0.001965, "loss_mean50": 0.001656, "lr": 5.904680865393181e-06, "grad_norm": 0.0274, "samples": 366080, "wall_s": 7163.4, "cuda_gb": 17.68} {"step": 11460, "loss": 0.000311, "loss_mean50": 0.001546, "lr": 5.868572920674581e-06, "grad_norm": 0.0113, "samples": 366720, "wall_s": 7175.5, "cuda_gb": 17.68} {"step": 11480, "loss": 0.000828, "loss_mean50": 0.001608, "lr": 5.83262372378623e-06, "grad_norm": 0.0139, "samples": 367360, "wall_s": 7187.4, "cuda_gb": 17.68} {"step": 11500, "loss": 0.005306, "loss_mean50": 0.001694, "lr": 5.7968339988174385e-06, "grad_norm": 0.0747, "samples": 368000, "wall_s": 7199.4, "cuda_gb": 17.68} {"step": 11520, "loss": 0.001139, "loss_mean50": 0.001771, "lr": 5.7612044666454034e-06, "grad_norm": 0.0231, "samples": 368640, "wall_s": 7211.4, "cuda_gb": 17.68} {"step": 11540, "loss": 0.002489, "loss_mean50": 0.00179, "lr": 5.7257358449207435e-06, "grad_norm": 0.0183, "samples": 369280, "wall_s": 7223.5, "cuda_gb": 17.68} {"step": 11560, "loss": 0.00081, "loss_mean50": 0.001638, "lr": 5.690428848053e-06, "grad_norm": 0.0163, "samples": 369920, "wall_s": 7235.5, "cuda_gb": 17.68} {"step": 11580, "loss": 0.002036, "loss_mean50": 0.001699, "lr": 5.655284187196265e-06, "grad_norm": 0.0255, "samples": 370560, "wall_s": 7247.4, "cuda_gb": 17.68} {"step": 11600, "loss": 0.002114, "loss_mean50": 0.00168, "lr": 5.620302570234867e-06, "grad_norm": 0.0481, "samples": 371200, "wall_s": 7259.5, "cuda_gb": 17.68} {"step": 11620, "loss": 0.002151, "loss_mean50": 0.001609, "lr": 5.5854847017690945e-06, "grad_norm": 0.0248, "samples": 371840, "wall_s": 7271.3, "cuda_gb": 17.68} {"step": 11640, "loss": 0.001272, "loss_mean50": 0.001721, "lr": 5.550831283100999e-06, "grad_norm": 0.0218, "samples": 372480, "wall_s": 7283.4, "cuda_gb": 17.68} {"step": 11660, "loss": 0.001717, "loss_mean50": 0.001754, "lr": 5.516343012220304e-06, "grad_norm": 0.102, "samples": 373120, "wall_s": 7295.3, "cuda_gb": 17.68} {"step": 11680, "loss": 0.000762, "loss_mean50": 0.001714, "lr": 5.482020583790311e-06, "grad_norm": 0.0165, "samples": 373760, "wall_s": 7307.4, "cuda_gb": 17.68} {"step": 11700, "loss": 0.000543, "loss_mean50": 0.001648, "lr": 5.447864689133917e-06, "grad_norm": 0.0146, "samples": 374400, "wall_s": 7319.5, "cuda_gb": 17.68} {"step": 11720, "loss": 0.003085, "loss_mean50": 0.001536, "lr": 5.413876016219707e-06, "grad_norm": 0.0409, "samples": 375040, "wall_s": 7331.5, "cuda_gb": 17.68} {"step": 11740, "loss": 0.001082, "loss_mean50": 0.00162, "lr": 5.380055249648071e-06, "grad_norm": 0.0137, "samples": 375680, "wall_s": 7343.4, "cuda_gb": 17.68} {"step": 11760, "loss": 0.000704, "loss_mean50": 0.001522, "lr": 5.346403070637427e-06, "grad_norm": 0.0197, "samples": 376320, "wall_s": 7355.5, "cuda_gb": 17.68} {"step": 11780, "loss": 0.001265, "loss_mean50": 0.00151, "lr": 5.312920157010506e-06, "grad_norm": 0.0195, "samples": 376960, "wall_s": 7367.5, "cuda_gb": 17.68} {"step": 11800, "loss": 0.00124, "loss_mean50": 0.001397, "lr": 5.2796071831806846e-06, "grad_norm": 0.024, "samples": 377600, "wall_s": 7379.4, "cuda_gb": 17.68} {"step": 11820, "loss": 0.002432, "loss_mean50": 0.001528, "lr": 5.2464648201384225e-06, "grad_norm": 0.0169, "samples": 378240, "wall_s": 7391.5, "cuda_gb": 17.68} {"step": 11840, "loss": 0.000482, "loss_mean50": 0.001592, "lr": 5.213493735437722e-06, "grad_norm": 0.0116, "samples": 378880, "wall_s": 7403.4, "cuda_gb": 17.68} {"step": 11860, "loss": 0.00118, "loss_mean50": 0.001387, "lr": 5.180694593182696e-06, "grad_norm": 0.0226, "samples": 379520, "wall_s": 7415.5, "cuda_gb": 17.68} {"step": 11880, "loss": 0.002337, "loss_mean50": 0.001354, "lr": 5.148068054014195e-06, "grad_norm": 0.0533, "samples": 380160, "wall_s": 7427.5, "cuda_gb": 17.68} {"step": 11900, "loss": 0.000696, "loss_mean50": 0.001307, "lr": 5.115614775096492e-06, "grad_norm": 0.0129, "samples": 380800, "wall_s": 7439.6, "cuda_gb": 17.68} {"step": 11920, "loss": 0.000385, "loss_mean50": 0.001361, "lr": 5.083335410104047e-06, "grad_norm": 0.0117, "samples": 381440, "wall_s": 7451.5, "cuda_gb": 17.68} {"step": 11940, "loss": 0.002683, "loss_mean50": 0.001475, "lr": 5.051230609208352e-06, "grad_norm": 0.0519, "samples": 382080, "wall_s": 7463.6, "cuda_gb": 17.68} {"step": 11960, "loss": 0.001285, "loss_mean50": 0.001547, "lr": 5.019301019064816e-06, "grad_norm": 0.0333, "samples": 382720, "wall_s": 7475.6, "cuda_gb": 17.68} {"step": 11980, "loss": 0.002039, "loss_mean50": 0.001764, "lr": 4.987547282799753e-06, "grad_norm": 0.0239, "samples": 383360, "wall_s": 7487.5, "cuda_gb": 17.68} {"step": 12000, "loss": 0.001864, "loss_mean50": 0.001638, "lr": 4.955970039997429e-06, "grad_norm": 0.0188, "samples": 384000, "wall_s": 7499.6, "cuda_gb": 17.68} {"step": 12000, "val_flow_mse": 0.001966} {"step": 12020, "loss": 0.000784, "loss_mean50": 0.001488, "lr": 4.924569926687178e-06, "grad_norm": 0.0408, "samples": 384640, "wall_s": 7519.4, "cuda_gb": 17.68} {"step": 12040, "loss": 0.001889, "loss_mean50": 0.001548, "lr": 4.893347575330571e-06, "grad_norm": 0.0251, "samples": 385280, "wall_s": 7531.5, "cuda_gb": 17.68} {"step": 12060, "loss": 0.00265, "loss_mean50": 0.001609, "lr": 4.862303614808722e-06, "grad_norm": 0.0326, "samples": 385920, "wall_s": 7543.6, "cuda_gb": 17.68} {"step": 12080, "loss": 0.002407, "loss_mean50": 0.00183, "lr": 4.831438670409574e-06, "grad_norm": 0.0215, "samples": 386560, "wall_s": 7555.6, "cuda_gb": 17.68} {"step": 12100, "loss": 0.003758, "loss_mean50": 0.001862, "lr": 4.800753363815327e-06, "grad_norm": 0.0248, "samples": 387200, "wall_s": 7567.6, "cuda_gb": 17.68} {"step": 12120, "loss": 0.000896, "loss_mean50": 0.001688, "lr": 4.770248313089926e-06, "grad_norm": 0.0185, "samples": 387840, "wall_s": 7579.7, "cuda_gb": 17.68} {"step": 12140, "loss": 0.00205, "loss_mean50": 0.001562, "lr": 4.739924132666585e-06, "grad_norm": 0.021, "samples": 388480, "wall_s": 7591.9, "cuda_gb": 17.68} {"step": 12160, "loss": 0.000633, "loss_mean50": 0.00156, "lr": 4.709781433335432e-06, "grad_norm": 0.0164, "samples": 389120, "wall_s": 7603.9, "cuda_gb": 17.68} {"step": 12180, "loss": 0.003695, "loss_mean50": 0.001743, "lr": 4.6798208222311985e-06, "grad_norm": 0.032, "samples": 389760, "wall_s": 7616.0, "cuda_gb": 17.68} {"step": 12200, "loss": 0.001093, "loss_mean50": 0.001809, "lr": 4.6500429028209904e-06, "grad_norm": 0.0227, "samples": 390400, "wall_s": 7627.9, "cuda_gb": 17.68} {"step": 12220, "loss": 0.000689, "loss_mean50": 0.001582, "lr": 4.620448274892142e-06, "grad_norm": 0.0137, "samples": 391040, "wall_s": 7639.9, "cuda_gb": 17.68} {"step": 12240, "loss": 0.002276, "loss_mean50": 0.001463, "lr": 4.591037534540117e-06, "grad_norm": 0.0217, "samples": 391680, "wall_s": 7652.1, "cuda_gb": 17.68} {"step": 12260, "loss": 0.002598, "loss_mean50": 0.001514, "lr": 4.56181127415652e-06, "grad_norm": 0.0306, "samples": 392320, "wall_s": 7664.1, "cuda_gb": 17.68} {"step": 12280, "loss": 0.00159, "loss_mean50": 0.001482, "lr": 4.5327700824171575e-06, "grad_norm": 0.0299, "samples": 392960, "wall_s": 7676.1, "cuda_gb": 17.68} {"step": 12300, "loss": 0.002433, "loss_mean50": 0.001469, "lr": 4.5039145442701735e-06, "grad_norm": 0.0282, "samples": 393600, "wall_s": 7687.6, "cuda_gb": 17.68} {"step": 12320, "loss": 0.002177, "loss_mean50": 0.001424, "lr": 4.475245240924278e-06, "grad_norm": 0.0141, "samples": 394240, "wall_s": 7699.1, "cuda_gb": 17.68} {"step": 12340, "loss": 0.00192, "loss_mean50": 0.001451, "lr": 4.446762749837041e-06, "grad_norm": 0.0377, "samples": 394880, "wall_s": 7711.2, "cuda_gb": 17.68} {"step": 12360, "loss": 0.001382, "loss_mean50": 0.001408, "lr": 4.418467644703254e-06, "grad_norm": 0.0304, "samples": 395520, "wall_s": 7723.2, "cuda_gb": 17.68} {"step": 12380, "loss": 0.000767, "loss_mean50": 0.001414, "lr": 4.3903604954433725e-06, "grad_norm": 0.0185, "samples": 396160, "wall_s": 7735.2, "cuda_gb": 17.68} {"step": 12400, "loss": 0.004727, "loss_mean50": 0.001563, "lr": 4.362441868192056e-06, "grad_norm": 0.0495, "samples": 396800, "wall_s": 7747.2, "cuda_gb": 17.68} {"step": 12420, "loss": 0.001493, "loss_mean50": 0.001517, "lr": 4.334712325286726e-06, "grad_norm": 0.0283, "samples": 397440, "wall_s": 7759.2, "cuda_gb": 17.68} {"step": 12440, "loss": 0.001714, "loss_mean50": 0.00153, "lr": 4.307172425256291e-06, "grad_norm": 0.0367, "samples": 398080, "wall_s": 7771.3, "cuda_gb": 17.68} {"step": 12460, "loss": 0.000872, "loss_mean50": 0.001543, "lr": 4.279822722809855e-06, "grad_norm": 0.0173, "samples": 398720, "wall_s": 7783.4, "cuda_gb": 17.68} {"step": 12480, "loss": 0.001395, "loss_mean50": 0.001543, "lr": 4.25266376882556e-06, "grad_norm": 0.0255, "samples": 399360, "wall_s": 7795.4, "cuda_gb": 17.68} {"step": 12500, "loss": 0.003511, "loss_mean50": 0.001596, "lr": 4.2256961103394905e-06, "grad_norm": 0.0379, "samples": 400000, "wall_s": 7807.4, "cuda_gb": 17.68} {"step": 12520, "loss": 0.000644, "loss_mean50": 0.00161, "lr": 4.198920290534661e-06, "grad_norm": 0.0303, "samples": 400640, "wall_s": 7859.8, "cuda_gb": 17.68} {"step": 12540, "loss": 0.001314, "loss_mean50": 0.001607, "lr": 4.1723368487300514e-06, "grad_norm": 0.034, "samples": 401280, "wall_s": 7871.8, "cuda_gb": 17.68} {"step": 12560, "loss": 0.002559, "loss_mean50": 0.001565, "lr": 4.145946320369784e-06, "grad_norm": 0.0702, "samples": 401920, "wall_s": 7883.9, "cuda_gb": 17.68} {"step": 12580, "loss": 0.002516, "loss_mean50": 0.001638, "lr": 4.119749237012297e-06, "grad_norm": 0.0433, "samples": 402560, "wall_s": 7896.0, "cuda_gb": 17.68} {"step": 12600, "loss": 0.00082, "loss_mean50": 0.001554, "lr": 4.093746126319657e-06, "grad_norm": 0.0136, "samples": 403200, "wall_s": 7908.0, "cuda_gb": 17.68} {"step": 12620, "loss": 0.002229, "loss_mean50": 0.001558, "lr": 4.067937512046945e-06, "grad_norm": 0.0225, "samples": 403840, "wall_s": 7919.9, "cuda_gb": 17.68} {"step": 12640, "loss": 0.00171, "loss_mean50": 0.001426, "lr": 4.042323914031678e-06, "grad_norm": 0.016, "samples": 404480, "wall_s": 7931.9, "cuda_gb": 17.68} {"step": 12660, "loss": 0.00086, "loss_mean50": 0.001438, "lr": 4.016905848183357e-06, "grad_norm": 0.026, "samples": 405120, "wall_s": 7943.9, "cuda_gb": 17.68} {"step": 12680, "loss": 0.000942, "loss_mean50": 0.001428, "lr": 3.991683826473071e-06, "grad_norm": 0.0144, "samples": 405760, "wall_s": 7955.9, "cuda_gb": 17.68} {"step": 12700, "loss": 0.001918, "loss_mean50": 0.001557, "lr": 3.9666583569231835e-06, "grad_norm": 0.0328, "samples": 406400, "wall_s": 7968.0, "cuda_gb": 17.68} {"step": 12720, "loss": 0.001342, "loss_mean50": 0.001801, "lr": 3.9418299435971e-06, "grad_norm": 0.0238, "samples": 407040, "wall_s": 7979.8, "cuda_gb": 17.68} {"step": 12740, "loss": 0.002851, "loss_mean50": 0.001896, "lr": 3.9171990865891265e-06, "grad_norm": 0.0436, "samples": 407680, "wall_s": 7991.7, "cuda_gb": 17.68} {"step": 12760, "loss": 0.000902, "loss_mean50": 0.001826, "lr": 3.892766282014372e-06, "grad_norm": 0.0127, "samples": 408320, "wall_s": 8003.7, "cuda_gb": 17.68} {"step": 12780, "loss": 0.00239, "loss_mean50": 0.001849, "lr": 3.868532021998779e-06, "grad_norm": 0.0287, "samples": 408960, "wall_s": 8015.6, "cuda_gb": 17.68} {"step": 12800, "loss": 0.001028, "loss_mean50": 0.001745, "lr": 3.844496794669201e-06, "grad_norm": 0.0341, "samples": 409600, "wall_s": 8027.7, "cuda_gb": 17.68} {"step": 12820, "loss": 0.002383, "loss_mean50": 0.00159, "lr": 3.820661084143568e-06, "grad_norm": 0.0192, "samples": 410240, "wall_s": 8039.6, "cuda_gb": 17.68} {"step": 12840, "loss": 0.001827, "loss_mean50": 0.00146, "lr": 3.797025370521147e-06, "grad_norm": 0.0218, "samples": 410880, "wall_s": 8051.7, "cuda_gb": 17.68} {"step": 12860, "loss": 0.001404, "loss_mean50": 0.001416, "lr": 3.7735901298728575e-06, "grad_norm": 0.0238, "samples": 411520, "wall_s": 8063.6, "cuda_gb": 17.68} {"step": 12880, "loss": 0.002753, "loss_mean50": 0.00141, "lr": 3.7503558342316883e-06, "grad_norm": 0.0197, "samples": 412160, "wall_s": 8075.6, "cuda_gb": 17.68} {"step": 12900, "loss": 0.003207, "loss_mean50": 0.001627, "lr": 3.7273229515831993e-06, "grad_norm": 0.0506, "samples": 412800, "wall_s": 8087.9, "cuda_gb": 17.68} {"step": 12920, "loss": 0.003182, "loss_mean50": 0.001763, "lr": 3.704491945856082e-06, "grad_norm": 0.0701, "samples": 413440, "wall_s": 8099.9, "cuda_gb": 17.68} {"step": 12940, "loss": 0.002207, "loss_mean50": 0.001839, "lr": 3.6818632769128106e-06, "grad_norm": 0.0282, "samples": 414080, "wall_s": 8112.0, "cuda_gb": 17.68} {"step": 12960, "loss": 0.002355, "loss_mean50": 0.001611, "lr": 3.6594374005404026e-06, "grad_norm": 0.0353, "samples": 414720, "wall_s": 8124.0, "cuda_gb": 17.68} {"step": 12980, "loss": 0.000881, "loss_mean50": 0.001441, "lr": 3.637214768441219e-06, "grad_norm": 0.0126, "samples": 415360, "wall_s": 8136.1, "cuda_gb": 17.68} {"step": 13000, "loss": 0.001039, "loss_mean50": 0.001526, "lr": 3.6151958282238664e-06, "grad_norm": 0.0284, "samples": 416000, "wall_s": 8148.0, "cuda_gb": 17.68} {"step": 13000, "val_flow_mse": 0.001775} {"step": 13020, "loss": 0.000389, "loss_mean50": 0.001662, "lr": 3.593381023394196e-06, "grad_norm": 0.02, "samples": 416640, "wall_s": 8168.1, "cuda_gb": 17.68} {"step": 13040, "loss": 0.003926, "loss_mean50": 0.001538, "lr": 3.571770793346353e-06, "grad_norm": 0.0318, "samples": 417280, "wall_s": 8180.0, "cuda_gb": 17.68} {"step": 13060, "loss": 0.000751, "loss_mean50": 0.001388, "lr": 3.5503655733539372e-06, "grad_norm": 0.0141, "samples": 417920, "wall_s": 8192.0, "cuda_gb": 17.68} {"step": 13080, "loss": 0.001073, "loss_mean50": 0.001411, "lr": 3.5291657945612303e-06, "grad_norm": 0.0362, "samples": 418560, "wall_s": 8204.1, "cuda_gb": 17.68} {"step": 13100, "loss": 0.000984, "loss_mean50": 0.001404, "lr": 3.5081718839745164e-06, "grad_norm": 0.0125, "samples": 419200, "wall_s": 8216.1, "cuda_gb": 17.68} {"step": 13120, "loss": 0.000999, "loss_mean50": 0.001532, "lr": 3.4873842644534798e-06, "grad_norm": 0.0343, "samples": 419840, "wall_s": 8228.1, "cuda_gb": 17.68} {"step": 13140, "loss": 0.00086, "loss_mean50": 0.001624, "lr": 3.466803354702685e-06, "grad_norm": 0.024, "samples": 420480, "wall_s": 8240.2, "cuda_gb": 17.68} {"step": 13160, "loss": 0.001083, "loss_mean50": 0.001711, "lr": 3.4464295692631432e-06, "grad_norm": 0.0222, "samples": 421120, "wall_s": 8252.2, "cuda_gb": 17.68} {"step": 13180, "loss": 0.001601, "loss_mean50": 0.001594, "lr": 3.4262633185039696e-06, "grad_norm": 0.0253, "samples": 421760, "wall_s": 8264.2, "cuda_gb": 17.68} {"step": 13200, "loss": 0.003264, "loss_mean50": 0.001709, "lr": 3.406305008614106e-06, "grad_norm": 0.0414, "samples": 422400, "wall_s": 8276.2, "cuda_gb": 17.68} {"step": 13220, "loss": 0.001257, "loss_mean50": 0.001584, "lr": 3.3865550415941497e-06, "grad_norm": 0.0265, "samples": 423040, "wall_s": 8288.2, "cuda_gb": 17.68} {"step": 13240, "loss": 0.000401, "loss_mean50": 0.00167, "lr": 3.367013815248255e-06, "grad_norm": 0.0263, "samples": 423680, "wall_s": 8300.3, "cuda_gb": 17.68} {"step": 13260, "loss": 0.000517, "loss_mean50": 0.001432, "lr": 3.347681723176115e-06, "grad_norm": 0.0364, "samples": 424320, "wall_s": 8312.3, "cuda_gb": 17.68} {"step": 13280, "loss": 0.002739, "loss_mean50": 0.001338, "lr": 3.3285591547650326e-06, "grad_norm": 0.0271, "samples": 424960, "wall_s": 8324.3, "cuda_gb": 17.68} {"step": 13300, "loss": 0.001695, "loss_mean50": 0.00134, "lr": 3.3096464951820954e-06, "grad_norm": 0.0212, "samples": 425600, "wall_s": 8336.2, "cuda_gb": 17.68} {"step": 13320, "loss": 0.002634, "loss_mean50": 0.001438, "lr": 3.290944125366388e-06, "grad_norm": 0.0421, "samples": 426240, "wall_s": 8348.4, "cuda_gb": 17.68} {"step": 13340, "loss": 0.000595, "loss_mean50": 0.001544, "lr": 3.272452422021341e-06, "grad_norm": 0.0166, "samples": 426880, "wall_s": 8360.4, "cuda_gb": 17.68} {"step": 13360, "loss": 0.002696, "loss_mean50": 0.001713, "lr": 3.254171757607145e-06, "grad_norm": 0.0542, "samples": 427520, "wall_s": 8372.3, "cuda_gb": 17.68} {"step": 13380, "loss": 0.001105, "loss_mean50": 0.001465, "lr": 3.236102500333231e-06, "grad_norm": 0.0114, "samples": 428160, "wall_s": 8384.4, "cuda_gb": 17.68} {"step": 13400, "loss": 0.004065, "loss_mean50": 0.001631, "lr": 3.2182450141508667e-06, "grad_norm": 0.0376, "samples": 428800, "wall_s": 8396.3, "cuda_gb": 17.68} {"step": 13420, "loss": 0.002308, "loss_mean50": 0.001576, "lr": 3.2005996587458287e-06, "grad_norm": 0.0378, "samples": 429440, "wall_s": 8408.4, "cuda_gb": 17.68} {"step": 13440, "loss": 0.002267, "loss_mean50": 0.001558, "lr": 3.183166789531146e-06, "grad_norm": 0.0161, "samples": 430080, "wall_s": 8420.4, "cuda_gb": 17.68} {"step": 13460, "loss": 0.001686, "loss_mean50": 0.001422, "lr": 3.1659467576399488e-06, "grad_norm": 0.0553, "samples": 430720, "wall_s": 8432.3, "cuda_gb": 17.68} {"step": 13480, "loss": 0.00085, "loss_mean50": 0.001419, "lr": 3.1489399099183936e-06, "grad_norm": 0.0387, "samples": 431360, "wall_s": 8444.4, "cuda_gb": 17.68} {"step": 13500, "loss": 0.000841, "loss_mean50": 0.001475, "lr": 3.132146588918679e-06, "grad_norm": 0.0119, "samples": 432000, "wall_s": 8456.5, "cuda_gb": 17.68} {"step": 13520, "loss": 0.000306, "loss_mean50": 0.001476, "lr": 3.1155671328921473e-06, "grad_norm": 0.0071, "samples": 432640, "wall_s": 8468.4, "cuda_gb": 17.68} {"step": 13540, "loss": 0.000461, "loss_mean50": 0.001523, "lr": 3.0992018757824658e-06, "grad_norm": 0.0134, "samples": 433280, "wall_s": 8480.5, "cuda_gb": 17.68} {"step": 13560, "loss": 0.002415, "loss_mean50": 0.001419, "lr": 3.0830511472189056e-06, "grad_norm": 0.0395, "samples": 433920, "wall_s": 8492.5, "cuda_gb": 17.68} {"step": 13580, "loss": 0.001185, "loss_mean50": 0.001374, "lr": 3.0671152725096976e-06, "grad_norm": 0.0295, "samples": 434560, "wall_s": 8504.5, "cuda_gb": 17.68} {"step": 13600, "loss": 0.001842, "loss_mean50": 0.001406, "lr": 3.0513945726354877e-06, "grad_norm": 0.0349, "samples": 435200, "wall_s": 8516.5, "cuda_gb": 17.68} {"step": 13620, "loss": 0.002111, "loss_mean50": 0.001563, "lr": 3.0358893642428604e-06, "grad_norm": 0.0276, "samples": 435840, "wall_s": 8528.6, "cuda_gb": 17.68} {"step": 13640, "loss": 0.000315, "loss_mean50": 0.001536, "lr": 3.020599959637975e-06, "grad_norm": 0.0082, "samples": 436480, "wall_s": 8540.5, "cuda_gb": 17.68} {"step": 13660, "loss": 0.000723, "loss_mean50": 0.001492, "lr": 3.005526666780261e-06, "grad_norm": 0.0148, "samples": 437120, "wall_s": 8552.6, "cuda_gb": 17.68} {"step": 13680, "loss": 0.001507, "loss_mean50": 0.001491, "lr": 2.990669789276225e-06, "grad_norm": 0.016, "samples": 437760, "wall_s": 8564.6, "cuda_gb": 17.68} {"step": 13700, "loss": 0.000793, "loss_mean50": 0.001333, "lr": 2.9760296263733332e-06, "grad_norm": 0.0378, "samples": 438400, "wall_s": 8576.6, "cuda_gb": 17.68} {"step": 13720, "loss": 0.000589, "loss_mean50": 0.001474, "lr": 2.961606472953977e-06, "grad_norm": 0.0267, "samples": 439040, "wall_s": 8588.7, "cuda_gb": 17.68} {"step": 13740, "loss": 0.003455, "loss_mean50": 0.001576, "lr": 2.947400619529546e-06, "grad_norm": 0.045, "samples": 439680, "wall_s": 8600.8, "cuda_gb": 17.68} {"step": 13760, "loss": 0.002159, "loss_mean50": 0.001657, "lr": 2.933412352234567e-06, "grad_norm": 0.0208, "samples": 440320, "wall_s": 8612.7, "cuda_gb": 17.68} {"step": 13780, "loss": 0.000951, "loss_mean50": 0.00168, "lr": 2.919641952820941e-06, "grad_norm": 0.018, "samples": 440960, "wall_s": 8624.8, "cuda_gb": 17.68} {"step": 13800, "loss": 0.000675, "loss_mean50": 0.001345, "lr": 2.906089698652277e-06, "grad_norm": 0.0166, "samples": 441600, "wall_s": 8636.8, "cuda_gb": 17.68} {"step": 13820, "loss": 0.00205, "loss_mean50": 0.001386, "lr": 2.892755862698292e-06, "grad_norm": 0.0311, "samples": 442240, "wall_s": 8648.7, "cuda_gb": 17.68} {"step": 13840, "loss": 0.001414, "loss_mean50": 0.00149, "lr": 2.8796407135293233e-06, "grad_norm": 0.0204, "samples": 442880, "wall_s": 8660.8, "cuda_gb": 17.68} {"step": 13860, "loss": 0.002901, "loss_mean50": 0.001557, "lr": 2.866744515310917e-06, "grad_norm": 0.0271, "samples": 443520, "wall_s": 8672.9, "cuda_gb": 17.68} {"step": 13880, "loss": 0.001714, "loss_mean50": 0.001261, "lr": 2.854067527798503e-06, "grad_norm": 0.0685, "samples": 444160, "wall_s": 8684.9, "cuda_gb": 17.68} {"step": 13900, "loss": 0.000879, "loss_mean50": 0.001374, "lr": 2.8416100063321654e-06, "grad_norm": 0.0319, "samples": 444800, "wall_s": 8697.0, "cuda_gb": 17.68} {"step": 13920, "loss": 0.000535, "loss_mean50": 0.001302, "lr": 2.829372201831502e-06, "grad_norm": 0.0267, "samples": 445440, "wall_s": 8709.0, "cuda_gb": 17.68} {"step": 13940, "loss": 0.00053, "loss_mean50": 0.001322, "lr": 2.817354360790566e-06, "grad_norm": 0.018, "samples": 446080, "wall_s": 8721.1, "cuda_gb": 17.68} {"step": 13960, "loss": 0.001451, "loss_mean50": 0.001302, "lr": 2.8055567252729046e-06, "grad_norm": 0.0214, "samples": 446720, "wall_s": 8733.2, "cuda_gb": 17.68} {"step": 13980, "loss": 0.004787, "loss_mean50": 0.001443, "lr": 2.7939795329066803e-06, "grad_norm": 0.1183, "samples": 447360, "wall_s": 8745.2, "cuda_gb": 17.68} {"step": 14000, "loss": 0.000664, "loss_mean50": 0.001466, "lr": 2.7826230168798876e-06, "grad_norm": 0.0309, "samples": 448000, "wall_s": 8757.3, "cuda_gb": 17.68} {"step": 14000, "val_flow_mse": 0.001887} {"step": 14020, "loss": 0.000742, "loss_mean50": 0.00142, "lr": 2.7714874059356517e-06, "grad_norm": 0.0133, "samples": 448640, "wall_s": 8777.0, "cuda_gb": 17.68} {"step": 14040, "loss": 0.000535, "loss_mean50": 0.001347, "lr": 2.76057292436763e-06, "grad_norm": 0.0124, "samples": 449280, "wall_s": 8789.1, "cuda_gb": 17.68} {"step": 14060, "loss": 0.000744, "loss_mean50": 0.001359, "lr": 2.7498797920154817e-06, "grad_norm": 0.0148, "samples": 449920, "wall_s": 8801.1, "cuda_gb": 17.68} {"step": 14080, "loss": 0.004797, "loss_mean50": 0.001509, "lr": 2.7394082242604514e-06, "grad_norm": 0.0264, "samples": 450560, "wall_s": 8813.2, "cuda_gb": 17.68} {"step": 14100, "loss": 0.00063, "loss_mean50": 0.001658, "lr": 2.7291584320210242e-06, "grad_norm": 0.0228, "samples": 451200, "wall_s": 8825.2, "cuda_gb": 17.68} {"step": 14120, "loss": 0.000393, "loss_mean50": 0.001497, "lr": 2.7191306217486763e-06, "grad_norm": 0.0248, "samples": 451840, "wall_s": 8837.2, "cuda_gb": 17.68} {"step": 14140, "loss": 0.001398, "loss_mean50": 0.001453, "lr": 2.709324995423726e-06, "grad_norm": 0.0198, "samples": 452480, "wall_s": 8849.2, "cuda_gb": 17.68} {"step": 14160, "loss": 0.002232, "loss_mean50": 0.001434, "lr": 2.6997417505512548e-06, "grad_norm": 0.0511, "samples": 453120, "wall_s": 8861.3, "cuda_gb": 17.68} {"step": 14180, "loss": 0.000632, "loss_mean50": 0.001431, "lr": 2.690381080157134e-06, "grad_norm": 0.0215, "samples": 453760, "wall_s": 8873.5, "cuda_gb": 17.68} {"step": 14200, "loss": 0.000254, "loss_mean50": 0.001196, "lr": 2.681243172784135e-06, "grad_norm": 0.0078, "samples": 454400, "wall_s": 8885.5, "cuda_gb": 17.68} {"step": 14220, "loss": 0.00223, "loss_mean50": 0.001378, "lr": 2.6723282124881367e-06, "grad_norm": 0.0268, "samples": 455040, "wall_s": 8897.8, "cuda_gb": 17.68} {"step": 14240, "loss": 0.001908, "loss_mean50": 0.001617, "lr": 2.663636378834409e-06, "grad_norm": 0.0223, "samples": 455680, "wall_s": 8909.9, "cuda_gb": 17.68} {"step": 14260, "loss": 0.000915, "loss_mean50": 0.001702, "lr": 2.6551678468940077e-06, "grad_norm": 0.0236, "samples": 456320, "wall_s": 8921.9, "cuda_gb": 17.68} {"step": 14280, "loss": 0.000663, "loss_mean50": 0.001607, "lr": 2.6469227872402357e-06, "grad_norm": 0.0828, "samples": 456960, "wall_s": 8934.1, "cuda_gb": 17.68} {"step": 14300, "loss": 0.000452, "loss_mean50": 0.001383, "lr": 2.638901365945217e-06, "grad_norm": 0.0201, "samples": 457600, "wall_s": 8946.0, "cuda_gb": 17.68} {"step": 14320, "loss": 0.001405, "loss_mean50": 0.001337, "lr": 2.6311037445765488e-06, "grad_norm": 0.0124, "samples": 458240, "wall_s": 8958.1, "cuda_gb": 17.68} {"step": 14340, "loss": 0.004499, "loss_mean50": 0.00142, "lr": 2.623530080194044e-06, "grad_norm": 0.0457, "samples": 458880, "wall_s": 8970.2, "cuda_gb": 17.68} {"step": 14360, "loss": 0.00281, "loss_mean50": 0.001713, "lr": 2.6161805253465737e-06, "grad_norm": 0.0223, "samples": 459520, "wall_s": 8982.4, "cuda_gb": 17.68} {"step": 14380, "loss": 0.000725, "loss_mean50": 0.001829, "lr": 2.609055228068987e-06, "grad_norm": 0.0313, "samples": 460160, "wall_s": 8994.4, "cuda_gb": 17.68} {"step": 14400, "loss": 0.000932, "loss_mean50": 0.001772, "lr": 2.6021543318791354e-06, "grad_norm": 0.0216, "samples": 460800, "wall_s": 9006.5, "cuda_gb": 17.68} {"step": 14420, "loss": 0.000534, "loss_mean50": 0.001349, "lr": 2.595477975774981e-06, "grad_norm": 0.0232, "samples": 461440, "wall_s": 9018.6, "cuda_gb": 17.68} {"step": 14440, "loss": 0.001209, "loss_mean50": 0.001373, "lr": 2.589026294231797e-06, "grad_norm": 0.097, "samples": 462080, "wall_s": 9030.7, "cuda_gb": 17.68} {"step": 14460, "loss": 0.001633, "loss_mean50": 0.001325, "lr": 2.582799417199454e-06, "grad_norm": 0.0264, "samples": 462720, "wall_s": 9042.8, "cuda_gb": 17.68} {"step": 14480, "loss": 0.000582, "loss_mean50": 0.001334, "lr": 2.5767974700998093e-06, "grad_norm": 0.0093, "samples": 463360, "wall_s": 9054.8, "cuda_gb": 17.68} {"step": 14500, "loss": 0.001595, "loss_mean50": 0.00146, "lr": 2.5710205738241775e-06, "grad_norm": 0.0253, "samples": 464000, "wall_s": 9066.8, "cuda_gb": 17.68} {"step": 14520, "loss": 0.000914, "loss_mean50": 0.001648, "lr": 2.565468844730895e-06, "grad_norm": 0.0163, "samples": 464640, "wall_s": 9078.8, "cuda_gb": 17.68} {"step": 14540, "loss": 0.001451, "loss_mean50": 0.001776, "lr": 2.5601423946429807e-06, "grad_norm": 0.0635, "samples": 465280, "wall_s": 9090.9, "cuda_gb": 17.68} {"step": 14560, "loss": 0.000463, "loss_mean50": 0.001483, "lr": 2.555041330845874e-06, "grad_norm": 0.0121, "samples": 465920, "wall_s": 9102.8, "cuda_gb": 17.68} {"step": 14580, "loss": 0.001378, "loss_mean50": 0.001586, "lr": 2.550165756085284e-06, "grad_norm": 0.0278, "samples": 466560, "wall_s": 9114.8, "cuda_gb": 17.68} {"step": 14600, "loss": 0.002761, "loss_mean50": 0.001861, "lr": 2.5455157685651184e-06, "grad_norm": 0.0207, "samples": 467200, "wall_s": 9127.0, "cuda_gb": 17.68} {"step": 14620, "loss": 0.001227, "loss_mean50": 0.001864, "lr": 2.5410914619454962e-06, "grad_norm": 0.0224, "samples": 467840, "wall_s": 9138.9, "cuda_gb": 17.68} {"step": 14640, "loss": 0.000859, "loss_mean50": 0.001658, "lr": 2.5368929253408747e-06, "grad_norm": 0.0237, "samples": 468480, "wall_s": 9150.9, "cuda_gb": 17.68} {"step": 14660, "loss": 0.000439, "loss_mean50": 0.001434, "lr": 2.5329202433182453e-06, "grad_norm": 0.0074, "samples": 469120, "wall_s": 9163.0, "cuda_gb": 17.68} {"step": 14680, "loss": 0.001644, "loss_mean50": 0.001574, "lr": 2.529173495895432e-06, "grad_norm": 0.029, "samples": 469760, "wall_s": 9175.0, "cuda_gb": 17.68} {"step": 14700, "loss": 0.000265, "loss_mean50": 0.00166, "lr": 2.5256527585394805e-06, "grad_norm": 0.01, "samples": 470400, "wall_s": 9187.0, "cuda_gb": 17.68} {"step": 14720, "loss": 0.000507, "loss_mean50": 0.001577, "lr": 2.5223581021651405e-06, "grad_norm": 0.009, "samples": 471040, "wall_s": 9199.1, "cuda_gb": 17.68} {"step": 14740, "loss": 0.003859, "loss_mean50": 0.001499, "lr": 2.5192895931334335e-06, "grad_norm": 0.0487, "samples": 471680, "wall_s": 9211.0, "cuda_gb": 17.68} {"step": 14760, "loss": 0.000198, "loss_mean50": 0.001328, "lr": 2.516447293250316e-06, "grad_norm": 0.0089, "samples": 472320, "wall_s": 9223.0, "cuda_gb": 17.68} {"step": 14780, "loss": 0.000351, "loss_mean50": 0.001337, "lr": 2.5138312597654382e-06, "grad_norm": 0.0106, "samples": 472960, "wall_s": 9235.2, "cuda_gb": 17.68} {"step": 14800, "loss": 0.000927, "loss_mean50": 0.001315, "lr": 2.5114415453709886e-06, "grad_norm": 0.0134, "samples": 473600, "wall_s": 9247.4, "cuda_gb": 17.68} {"step": 14820, "loss": 0.000505, "loss_mean50": 0.001573, "lr": 2.509278198200633e-06, "grad_norm": 0.0212, "samples": 474240, "wall_s": 9259.3, "cuda_gb": 17.68} {"step": 14840, "loss": 0.000243, "loss_mean50": 0.001505, "lr": 2.5073412618285445e-06, "grad_norm": 0.0072, "samples": 474880, "wall_s": 9271.4, "cuda_gb": 17.68} {"step": 14860, "loss": 0.004072, "loss_mean50": 0.001472, "lr": 2.505630775268527e-06, "grad_norm": 0.0442, "samples": 475520, "wall_s": 9283.4, "cuda_gb": 17.68} {"step": 14880, "loss": 0.000781, "loss_mean50": 0.00144, "lr": 2.5041467729732266e-06, "grad_norm": 0.0124, "samples": 476160, "wall_s": 9295.5, "cuda_gb": 17.68} {"step": 14900, "loss": 0.000497, "loss_mean50": 0.001595, "lr": 2.502889284833443e-06, "grad_norm": 0.0121, "samples": 476800, "wall_s": 9307.6, "cuda_gb": 17.68} {"step": 14920, "loss": 0.000599, "loss_mean50": 0.001599, "lr": 2.5018583361775205e-06, "grad_norm": 0.0234, "samples": 477440, "wall_s": 9319.7, "cuda_gb": 17.68} {"step": 14940, "loss": 0.002269, "loss_mean50": 0.001687, "lr": 2.501053947770846e-06, "grad_norm": 0.0182, "samples": 478080, "wall_s": 9331.7, "cuda_gb": 17.68} {"step": 14960, "loss": 0.001183, "loss_mean50": 0.001657, "lr": 2.5004761358154212e-06, "grad_norm": 0.0247, "samples": 478720, "wall_s": 9343.7, "cuda_gb": 17.68} {"step": 14980, "loss": 0.001595, "loss_mean50": 0.001739, "lr": 2.5001249119495454e-06, "grad_norm": 0.0304, "samples": 479360, "wall_s": 9355.6, "cuda_gb": 17.68} {"step": 15000, "loss": 0.001958, "loss_mean50": 0.001473, "lr": 2.5000002832475742e-06, "grad_norm": 0.0155, "samples": 480000, "wall_s": 9367.7, "cuda_gb": 17.68} {"step": 15000, "val_flow_mse": 0.001764, "val_action_mae": 0.008264, "best": true}