pi05_arx_decomposed / train_log.jsonl
KasLoot's picture
Upload folder using huggingface_hub
faef969 verified
Raw
History Blame Contribute Delete
123 kB
{"step": 20, "loss": 0.067503, "loss_mean50": 0.080653, "lr": 5e-07, "grad_norm": 0.2523, "samples": 640, "wall_s": 44.2, "cuda_gb": 16.86}
{"step": 40, "loss": 0.081014, "loss_mean50": 0.080018, "lr": 1e-06, "grad_norm": 0.219, "samples": 1280, "wall_s": 82.9, "cuda_gb": 16.86}
{"step": 60, "loss": 0.061559, "loss_mean50": 0.07785, "lr": 1.5e-06, "grad_norm": 0.2039, "samples": 1920, "wall_s": 121.8, "cuda_gb": 16.86}
{"step": 80, "loss": 0.071491, "loss_mean50": 0.073838, "lr": 2e-06, "grad_norm": 0.1861, "samples": 2560, "wall_s": 160.7, "cuda_gb": 16.86}
{"step": 100, "loss": 0.048766, "loss_mean50": 0.061855, "lr": 2.5e-06, "grad_norm": 0.1442, "samples": 3200, "wall_s": 199.6, "cuda_gb": 16.86}
{"step": 120, "loss": 0.018467, "loss_mean50": 0.043621, "lr": 3e-06, "grad_norm": 0.0884, "samples": 3840, "wall_s": 238.5, "cuda_gb": 16.86}
{"step": 140, "loss": 0.020463, "loss_mean50": 0.029294, "lr": 3.5e-06, "grad_norm": 0.1017, "samples": 4480, "wall_s": 277.3, "cuda_gb": 16.86}
{"step": 160, "loss": 0.013608, "loss_mean50": 0.019526, "lr": 4e-06, "grad_norm": 0.0674, "samples": 5120, "wall_s": 316.3, "cuda_gb": 16.86}
{"step": 180, "loss": 0.015079, "loss_mean50": 0.016001, "lr": 4.5e-06, "grad_norm": 0.0793, "samples": 5760, "wall_s": 355.1, "cuda_gb": 16.86}
{"step": 200, "loss": 0.010489, "loss_mean50": 0.013991, "lr": 5e-06, "grad_norm": 0.0616, "samples": 6400, "wall_s": 394.1, "cuda_gb": 16.86}
{"step": 220, "loss": 0.011847, "loss_mean50": 0.012306, "lr": 5.500000000000001e-06, "grad_norm": 0.0566, "samples": 7040, "wall_s": 433.0, "cuda_gb": 16.86}
{"step": 240, "loss": 0.011088, "loss_mean50": 0.010886, "lr": 6e-06, "grad_norm": 0.0649, "samples": 7680, "wall_s": 471.9, "cuda_gb": 16.86}
{"step": 260, "loss": 0.013353, "loss_mean50": 0.01, "lr": 6.5000000000000004e-06, "grad_norm": 0.1569, "samples": 8320, "wall_s": 510.8, "cuda_gb": 16.86}
{"step": 280, "loss": 0.010539, "loss_mean50": 0.009471, "lr": 7e-06, "grad_norm": 0.1201, "samples": 8960, "wall_s": 549.7, "cuda_gb": 16.86}
{"step": 300, "loss": 0.011455, "loss_mean50": 0.008992, "lr": 7.5e-06, "grad_norm": 0.1192, "samples": 9600, "wall_s": 588.6, "cuda_gb": 16.86}
{"step": 320, "loss": 0.00633, "loss_mean50": 0.008748, "lr": 8e-06, "grad_norm": 0.0593, "samples": 10240, "wall_s": 627.5, "cuda_gb": 16.86}
{"step": 340, "loss": 0.008583, "loss_mean50": 0.00864, "lr": 8.5e-06, "grad_norm": 0.0643, "samples": 10880, "wall_s": 666.3, "cuda_gb": 16.86}
{"step": 360, "loss": 0.004483, "loss_mean50": 0.008362, "lr": 9e-06, "grad_norm": 0.0385, "samples": 11520, "wall_s": 705.2, "cuda_gb": 16.86}
{"step": 380, "loss": 0.009801, "loss_mean50": 0.007987, "lr": 9.5e-06, "grad_norm": 0.0589, "samples": 12160, "wall_s": 744.1, "cuda_gb": 16.86}
{"step": 400, "loss": 0.008249, "loss_mean50": 0.007399, "lr": 1e-05, "grad_norm": 0.0744, "samples": 12800, "wall_s": 783.0, "cuda_gb": 16.86}
{"step": 420, "loss": 0.011665, "loss_mean50": 0.00739, "lr": 1.0500000000000001e-05, "grad_norm": 0.0978, "samples": 13440, "wall_s": 822.0, "cuda_gb": 16.86}
{"step": 440, "loss": 0.010001, "loss_mean50": 0.007491, "lr": 1.1000000000000001e-05, "grad_norm": 0.0651, "samples": 14080, "wall_s": 861.0, "cuda_gb": 16.86}
{"step": 460, "loss": 0.00636, "loss_mean50": 0.007032, "lr": 1.15e-05, "grad_norm": 0.0579, "samples": 14720, "wall_s": 900.0, "cuda_gb": 16.86}
{"step": 480, "loss": 0.006683, "loss_mean50": 0.006433, "lr": 1.2e-05, "grad_norm": 0.0587, "samples": 15360, "wall_s": 939.0, "cuda_gb": 16.86}
{"step": 500, "loss": 0.007948, "loss_mean50": 0.006454, "lr": 1.25e-05, "grad_norm": 0.0661, "samples": 16000, "wall_s": 977.9, "cuda_gb": 16.86}
{"step": 520, "loss": 0.005933, "loss_mean50": 0.006517, "lr": 1.3000000000000001e-05, "grad_norm": 0.0498, "samples": 16640, "wall_s": 1016.7, "cuda_gb": 16.86}
{"step": 540, "loss": 0.007616, "loss_mean50": 0.006493, "lr": 1.35e-05, "grad_norm": 0.0508, "samples": 17280, "wall_s": 1055.5, "cuda_gb": 16.86}
{"step": 560, "loss": 0.005586, "loss_mean50": 0.006581, "lr": 1.4e-05, "grad_norm": 0.0491, "samples": 17920, "wall_s": 1094.4, "cuda_gb": 16.86}
{"step": 580, "loss": 0.005017, "loss_mean50": 0.006491, "lr": 1.45e-05, "grad_norm": 0.0391, "samples": 18560, "wall_s": 1133.1, "cuda_gb": 16.86}
{"step": 600, "loss": 0.004483, "loss_mean50": 0.00613, "lr": 1.5e-05, "grad_norm": 0.0498, "samples": 19200, "wall_s": 1172.0, "cuda_gb": 16.86}
{"step": 620, "loss": 0.007059, "loss_mean50": 0.005998, "lr": 1.55e-05, "grad_norm": 0.0593, "samples": 19840, "wall_s": 1210.9, "cuda_gb": 16.86}
{"step": 640, "loss": 0.005429, "loss_mean50": 0.005962, "lr": 1.6e-05, "grad_norm": 0.0456, "samples": 20480, "wall_s": 1250.0, "cuda_gb": 16.86}
{"step": 660, "loss": 0.00738, "loss_mean50": 0.005974, "lr": 1.65e-05, "grad_norm": 0.0602, "samples": 21120, "wall_s": 1288.8, "cuda_gb": 16.86}
{"step": 680, "loss": 0.00534, "loss_mean50": 0.005623, "lr": 1.7e-05, "grad_norm": 0.0434, "samples": 21760, "wall_s": 1327.7, "cuda_gb": 16.86}
{"step": 700, "loss": 0.006618, "loss_mean50": 0.005717, "lr": 1.7500000000000002e-05, "grad_norm": 0.0493, "samples": 22400, "wall_s": 1366.5, "cuda_gb": 16.86}
{"step": 720, "loss": 0.005566, "loss_mean50": 0.006031, "lr": 1.8e-05, "grad_norm": 0.0425, "samples": 23040, "wall_s": 1405.3, "cuda_gb": 16.86}
{"step": 740, "loss": 0.006953, "loss_mean50": 0.006321, "lr": 1.8500000000000002e-05, "grad_norm": 0.0608, "samples": 23680, "wall_s": 1444.3, "cuda_gb": 16.86}
{"step": 760, "loss": 0.008775, "loss_mean50": 0.006323, "lr": 1.9e-05, "grad_norm": 0.0483, "samples": 24320, "wall_s": 1483.1, "cuda_gb": 16.86}
{"step": 780, "loss": 0.009467, "loss_mean50": 0.005928, "lr": 1.95e-05, "grad_norm": 0.0612, "samples": 24960, "wall_s": 1522.0, "cuda_gb": 16.86}
{"step": 800, "loss": 0.006501, "loss_mean50": 0.005481, "lr": 2e-05, "grad_norm": 0.0558, "samples": 25600, "wall_s": 1560.9, "cuda_gb": 16.86}
{"step": 820, "loss": 0.004765, "loss_mean50": 0.005427, "lr": 2.05e-05, "grad_norm": 0.0455, "samples": 26240, "wall_s": 1599.8, "cuda_gb": 16.86}
{"step": 840, "loss": 0.005527, "loss_mean50": 0.005432, "lr": 2.1000000000000002e-05, "grad_norm": 0.0491, "samples": 26880, "wall_s": 1638.7, "cuda_gb": 16.86}
{"step": 860, "loss": 0.00829, "loss_mean50": 0.005668, "lr": 2.15e-05, "grad_norm": 0.0701, "samples": 27520, "wall_s": 1677.7, "cuda_gb": 16.86}
{"step": 880, "loss": 0.008139, "loss_mean50": 0.005821, "lr": 2.2000000000000003e-05, "grad_norm": 0.0468, "samples": 28160, "wall_s": 1716.7, "cuda_gb": 16.86}
{"step": 900, "loss": 0.00303, "loss_mean50": 0.005726, "lr": 2.25e-05, "grad_norm": 0.0374, "samples": 28800, "wall_s": 1755.5, "cuda_gb": 16.86}
{"step": 920, "loss": 0.007042, "loss_mean50": 0.005609, "lr": 2.3e-05, "grad_norm": 0.0421, "samples": 29440, "wall_s": 1794.3, "cuda_gb": 16.86}
{"step": 940, "loss": 0.005449, "loss_mean50": 0.005749, "lr": 2.35e-05, "grad_norm": 0.0545, "samples": 30080, "wall_s": 1833.3, "cuda_gb": 16.86}
{"step": 960, "loss": 0.004219, "loss_mean50": 0.005917, "lr": 2.4e-05, "grad_norm": 0.0386, "samples": 30720, "wall_s": 1872.2, "cuda_gb": 16.86}
{"step": 980, "loss": 0.005837, "loss_mean50": 0.005709, "lr": 2.4500000000000003e-05, "grad_norm": 0.0716, "samples": 31360, "wall_s": 1911.1, "cuda_gb": 16.86}
{"step": 1000, "loss": 0.007222, "loss_mean50": 0.005298, "lr": 2.5e-05, "grad_norm": 0.0639, "samples": 32000, "wall_s": 1950.0, "cuda_gb": 16.86}
{"step": 1000, "val_flow_mse": 0.004683}
{"step": 1020, "loss": 0.006777, "loss_mean50": 0.005261, "lr": 2.499989774778022e-05, "grad_norm": 0.0779, "samples": 32640, "wall_s": 2017.8, "cuda_gb": 16.86}
{"step": 1040, "loss": 0.007995, "loss_mean50": 0.005455, "lr": 2.4999569183187692e-05, "grad_norm": 0.0434, "samples": 33280, "wall_s": 2056.8, "cuda_gb": 16.86}
{"step": 1060, "loss": 0.004304, "loss_mean50": 0.005595, "lr": 2.4999014029592804e-05, "grad_norm": 0.0279, "samples": 33920, "wall_s": 2095.6, "cuda_gb": 16.86}
{"step": 1080, "loss": 0.003341, "loss_mean50": 0.005214, "lr": 2.4998232298177463e-05, "grad_norm": 0.0363, "samples": 34560, "wall_s": 2134.6, "cuda_gb": 16.86}
{"step": 1100, "loss": 0.004847, "loss_mean50": 0.004693, "lr": 2.499722400468732e-05, "grad_norm": 0.0607, "samples": 35200, "wall_s": 2173.5, "cuda_gb": 16.86}
{"step": 1120, "loss": 0.005469, "loss_mean50": 0.005451, "lr": 2.4995989169431427e-05, "grad_norm": 0.072, "samples": 35840, "wall_s": 2212.4, "cuda_gb": 16.86}
{"step": 1140, "loss": 0.00426, "loss_mean50": 0.005986, "lr": 2.499452781728187e-05, "grad_norm": 0.0344, "samples": 36480, "wall_s": 2251.4, "cuda_gb": 16.86}
{"step": 1160, "loss": 0.003419, "loss_mean50": 0.005582, "lr": 2.4992839977673223e-05, "grad_norm": 0.032, "samples": 37120, "wall_s": 2290.2, "cuda_gb": 16.86}
{"step": 1180, "loss": 0.005224, "loss_mean50": 0.005473, "lr": 2.4990925684601972e-05, "grad_norm": 0.0437, "samples": 37760, "wall_s": 2329.1, "cuda_gb": 16.86}
{"step": 1200, "loss": 0.004348, "loss_mean50": 0.005206, "lr": 2.4988784976625844e-05, "grad_norm": 0.0457, "samples": 38400, "wall_s": 2368.0, "cuda_gb": 16.86}
{"step": 1220, "loss": 0.003945, "loss_mean50": 0.004849, "lr": 2.4986417896863013e-05, "grad_norm": 0.0526, "samples": 39040, "wall_s": 2407.0, "cuda_gb": 16.86}
{"step": 1240, "loss": 0.003237, "loss_mean50": 0.004682, "lr": 2.4983824492991236e-05, "grad_norm": 0.039, "samples": 39680, "wall_s": 2446.0, "cuda_gb": 16.86}
{"step": 1260, "loss": 0.003509, "loss_mean50": 0.004743, "lr": 2.4981004817246896e-05, "grad_norm": 0.0729, "samples": 40320, "wall_s": 2485.1, "cuda_gb": 16.86}
{"step": 1280, "loss": 0.003847, "loss_mean50": 0.004833, "lr": 2.497795892642395e-05, "grad_norm": 0.0363, "samples": 40960, "wall_s": 2524.1, "cuda_gb": 16.86}
{"step": 1300, "loss": 0.005111, "loss_mean50": 0.004922, "lr": 2.4974686881872776e-05, "grad_norm": 0.05, "samples": 41600, "wall_s": 2563.0, "cuda_gb": 16.86}
{"step": 1320, "loss": 0.003294, "loss_mean50": 0.004829, "lr": 2.4971188749498953e-05, "grad_norm": 0.0289, "samples": 42240, "wall_s": 2602.0, "cuda_gb": 16.86}
{"step": 1340, "loss": 0.005642, "loss_mean50": 0.004738, "lr": 2.4967464599761908e-05, "grad_norm": 0.0399, "samples": 42880, "wall_s": 2641.0, "cuda_gb": 16.86}
{"step": 1360, "loss": 0.003057, "loss_mean50": 0.005039, "lr": 2.4963514507673534e-05, "grad_norm": 0.0324, "samples": 43520, "wall_s": 2679.9, "cuda_gb": 16.86}
{"step": 1380, "loss": 0.004624, "loss_mean50": 0.005523, "lr": 2.4959338552796643e-05, "grad_norm": 0.0377, "samples": 44160, "wall_s": 2718.8, "cuda_gb": 16.86}
{"step": 1400, "loss": 0.003622, "loss_mean50": 0.005516, "lr": 2.495493681924339e-05, "grad_norm": 0.0376, "samples": 44800, "wall_s": 2757.6, "cuda_gb": 16.86}
{"step": 1420, "loss": 0.005002, "loss_mean50": 0.004989, "lr": 2.4950309395673564e-05, "grad_norm": 0.0539, "samples": 45440, "wall_s": 2796.5, "cuda_gb": 16.86}
{"step": 1440, "loss": 0.013161, "loss_mean50": 0.004966, "lr": 2.4945456375292797e-05, "grad_norm": 0.0661, "samples": 46080, "wall_s": 2835.3, "cuda_gb": 16.86}
{"step": 1460, "loss": 0.003706, "loss_mean50": 0.004717, "lr": 2.494037785585071e-05, "grad_norm": 0.0395, "samples": 46720, "wall_s": 2874.4, "cuda_gb": 16.86}
{"step": 1480, "loss": 0.004116, "loss_mean50": 0.004735, "lr": 2.493507393963892e-05, "grad_norm": 0.0358, "samples": 47360, "wall_s": 2913.3, "cuda_gb": 16.86}
{"step": 1500, "loss": 0.004611, "loss_mean50": 0.004357, "lr": 2.492954473348899e-05, "grad_norm": 0.0607, "samples": 48000, "wall_s": 2952.2, "cuda_gb": 16.86}
{"step": 1520, "loss": 0.004425, "loss_mean50": 0.004273, "lr": 2.4923790348770284e-05, "grad_norm": 0.0337, "samples": 48640, "wall_s": 2991.1, "cuda_gb": 16.86}
{"step": 1540, "loss": 0.00303, "loss_mean50": 0.004293, "lr": 2.4917810901387707e-05, "grad_norm": 0.0341, "samples": 49280, "wall_s": 3030.0, "cuda_gb": 16.86}
{"step": 1560, "loss": 0.004908, "loss_mean50": 0.004452, "lr": 2.4911606511779384e-05, "grad_norm": 0.038, "samples": 49920, "wall_s": 3068.7, "cuda_gb": 16.86}
{"step": 1580, "loss": 0.003968, "loss_mean50": 0.004519, "lr": 2.4905177304914233e-05, "grad_norm": 0.0324, "samples": 50560, "wall_s": 3107.7, "cuda_gb": 16.86}
{"step": 1600, "loss": 0.004094, "loss_mean50": 0.004515, "lr": 2.4898523410289436e-05, "grad_norm": 0.031, "samples": 51200, "wall_s": 3146.5, "cuda_gb": 16.86}
{"step": 1620, "loss": 0.002783, "loss_mean50": 0.004553, "lr": 2.4891644961927848e-05, "grad_norm": 0.0316, "samples": 51840, "wall_s": 3185.2, "cuda_gb": 16.86}
{"step": 1640, "loss": 0.004534, "loss_mean50": 0.004391, "lr": 2.488454209837529e-05, "grad_norm": 0.0391, "samples": 52480, "wall_s": 3224.1, "cuda_gb": 16.86}
{"step": 1660, "loss": 0.005913, "loss_mean50": 0.004475, "lr": 2.4877214962697744e-05, "grad_norm": 0.0721, "samples": 53120, "wall_s": 3263.1, "cuda_gb": 16.86}
{"step": 1680, "loss": 0.002896, "loss_mean50": 0.004528, "lr": 2.4869663702478503e-05, "grad_norm": 0.0277, "samples": 53760, "wall_s": 3302.1, "cuda_gb": 16.86}
{"step": 1700, "loss": 0.002509, "loss_mean50": 0.004203, "lr": 2.4861888469815163e-05, "grad_norm": 0.0284, "samples": 54400, "wall_s": 3340.9, "cuda_gb": 16.86}
{"step": 1720, "loss": 0.003323, "loss_mean50": 0.003964, "lr": 2.4853889421316596e-05, "grad_norm": 0.0278, "samples": 55040, "wall_s": 3379.7, "cuda_gb": 16.86}
{"step": 1740, "loss": 0.003087, "loss_mean50": 0.003875, "lr": 2.484566671809975e-05, "grad_norm": 0.0444, "samples": 55680, "wall_s": 3418.6, "cuda_gb": 16.86}
{"step": 1760, "loss": 0.004327, "loss_mean50": 0.003933, "lr": 2.4837220525786454e-05, "grad_norm": 0.0394, "samples": 56320, "wall_s": 3457.4, "cuda_gb": 16.86}
{"step": 1780, "loss": 0.006524, "loss_mean50": 0.004323, "lr": 2.4828551014500046e-05, "grad_norm": 0.0507, "samples": 56960, "wall_s": 3496.3, "cuda_gb": 16.86}
{"step": 1800, "loss": 0.004405, "loss_mean50": 0.004708, "lr": 2.4819658358861965e-05, "grad_norm": 0.0714, "samples": 57600, "wall_s": 3535.2, "cuda_gb": 16.86}
{"step": 1820, "loss": 0.005893, "loss_mean50": 0.004519, "lr": 2.4810542737988217e-05, "grad_norm": 0.0424, "samples": 58240, "wall_s": 3574.1, "cuda_gb": 16.86}
{"step": 1840, "loss": 0.002143, "loss_mean50": 0.004205, "lr": 2.480120433548579e-05, "grad_norm": 0.0408, "samples": 58880, "wall_s": 3612.9, "cuda_gb": 16.86}
{"step": 1860, "loss": 0.005955, "loss_mean50": 0.004393, "lr": 2.479164333944893e-05, "grad_norm": 0.0534, "samples": 59520, "wall_s": 3651.9, "cuda_gb": 16.86}
{"step": 1880, "loss": 0.004634, "loss_mean50": 0.004579, "lr": 2.4781859942455375e-05, "grad_norm": 0.0586, "samples": 60160, "wall_s": 3690.9, "cuda_gb": 16.86}
{"step": 1900, "loss": 0.004876, "loss_mean50": 0.004509, "lr": 2.4771854341562465e-05, "grad_norm": 0.0355, "samples": 60800, "wall_s": 3729.7, "cuda_gb": 16.86}
{"step": 1920, "loss": 0.004343, "loss_mean50": 0.004267, "lr": 2.476162673830317e-05, "grad_norm": 0.0431, "samples": 61440, "wall_s": 3768.6, "cuda_gb": 16.86}
{"step": 1940, "loss": 0.002759, "loss_mean50": 0.003953, "lr": 2.475117733868204e-05, "grad_norm": 0.0243, "samples": 62080, "wall_s": 3807.6, "cuda_gb": 16.86}
{"step": 1960, "loss": 0.004722, "loss_mean50": 0.004086, "lr": 2.474050635317105e-05, "grad_norm": 0.0368, "samples": 62720, "wall_s": 3846.4, "cuda_gb": 16.86}
{"step": 1980, "loss": 0.003505, "loss_mean50": 0.004169, "lr": 2.472961399670536e-05, "grad_norm": 0.0417, "samples": 63360, "wall_s": 3885.2, "cuda_gb": 16.86}
{"step": 2000, "loss": 0.003226, "loss_mean50": 0.004223, "lr": 2.471850048867899e-05, "grad_norm": 0.0311, "samples": 64000, "wall_s": 3924.0, "cuda_gb": 16.86}
{"step": 2000, "val_flow_mse": 0.003806}
{"step": 2020, "loss": 0.003172, "loss_mean50": 0.00418, "lr": 2.470716605294039e-05, "grad_norm": 0.03, "samples": 64640, "wall_s": 3991.7, "cuda_gb": 16.86}
{"step": 2040, "loss": 0.005057, "loss_mean50": 0.004175, "lr": 2.4695610917787952e-05, "grad_norm": 0.0447, "samples": 65280, "wall_s": 4030.6, "cuda_gb": 16.86}
{"step": 2060, "loss": 0.005209, "loss_mean50": 0.00426, "lr": 2.468383531596539e-05, "grad_norm": 0.0381, "samples": 65920, "wall_s": 4069.4, "cuda_gb": 16.86}
{"step": 2080, "loss": 0.00397, "loss_mean50": 0.004013, "lr": 2.467183948465707e-05, "grad_norm": 0.0271, "samples": 66560, "wall_s": 4108.3, "cuda_gb": 16.86}
{"step": 2100, "loss": 0.002423, "loss_mean50": 0.004134, "lr": 2.4659623665483212e-05, "grad_norm": 0.0274, "samples": 67200, "wall_s": 4147.2, "cuda_gb": 16.86}
{"step": 2120, "loss": 0.005674, "loss_mean50": 0.004185, "lr": 2.4647188104495043e-05, "grad_norm": 0.0817, "samples": 67840, "wall_s": 4186.1, "cuda_gb": 16.86}
{"step": 2140, "loss": 0.004641, "loss_mean50": 0.004112, "lr": 2.463453305216983e-05, "grad_norm": 0.0545, "samples": 68480, "wall_s": 4224.9, "cuda_gb": 16.86}
{"step": 2160, "loss": 0.003984, "loss_mean50": 0.004078, "lr": 2.4621658763405842e-05, "grad_norm": 0.0345, "samples": 69120, "wall_s": 4263.9, "cuda_gb": 16.86}
{"step": 2180, "loss": 0.006615, "loss_mean50": 0.004066, "lr": 2.4608565497517207e-05, "grad_norm": 0.0633, "samples": 69760, "wall_s": 4302.8, "cuda_gb": 16.86}
{"step": 2200, "loss": 0.00752, "loss_mean50": 0.004352, "lr": 2.4595253518228692e-05, "grad_norm": 0.05, "samples": 70400, "wall_s": 4341.8, "cuda_gb": 16.86}
{"step": 2220, "loss": 0.003655, "loss_mean50": 0.0045, "lr": 2.4581723093670407e-05, "grad_norm": 0.0527, "samples": 71040, "wall_s": 4380.7, "cuda_gb": 16.86}
{"step": 2240, "loss": 0.002887, "loss_mean50": 0.004367, "lr": 2.456797449637237e-05, "grad_norm": 0.0306, "samples": 71680, "wall_s": 4419.6, "cuda_gb": 16.86}
{"step": 2260, "loss": 0.002589, "loss_mean50": 0.00407, "lr": 2.455400800325905e-05, "grad_norm": 0.0514, "samples": 72320, "wall_s": 4458.5, "cuda_gb": 16.86}
{"step": 2280, "loss": 0.006895, "loss_mean50": 0.004023, "lr": 2.4539823895643783e-05, "grad_norm": 0.0465, "samples": 72960, "wall_s": 4497.4, "cuda_gb": 16.86}
{"step": 2300, "loss": 0.002769, "loss_mean50": 0.003898, "lr": 2.4525422459223077e-05, "grad_norm": 0.0379, "samples": 73600, "wall_s": 4536.3, "cuda_gb": 16.86}
{"step": 2320, "loss": 0.003626, "loss_mean50": 0.003745, "lr": 2.451080398407091e-05, "grad_norm": 0.0307, "samples": 74240, "wall_s": 4575.3, "cuda_gb": 16.86}
{"step": 2340, "loss": 0.003993, "loss_mean50": 0.003691, "lr": 2.4495968764632826e-05, "grad_norm": 0.0332, "samples": 74880, "wall_s": 4614.1, "cuda_gb": 16.86}
{"step": 2360, "loss": 0.003211, "loss_mean50": 0.003939, "lr": 2.4480917099720065e-05, "grad_norm": 0.043, "samples": 75520, "wall_s": 4653.0, "cuda_gb": 16.86}
{"step": 2380, "loss": 0.003958, "loss_mean50": 0.003945, "lr": 2.4465649292503495e-05, "grad_norm": 0.0316, "samples": 76160, "wall_s": 4691.9, "cuda_gb": 16.86}
{"step": 2400, "loss": 0.004608, "loss_mean50": 0.003982, "lr": 2.4450165650507544e-05, "grad_norm": 0.0426, "samples": 76800, "wall_s": 4730.8, "cuda_gb": 16.86}
{"step": 2420, "loss": 0.003716, "loss_mean50": 0.003898, "lr": 2.443446648560396e-05, "grad_norm": 0.0307, "samples": 77440, "wall_s": 4769.8, "cuda_gb": 16.86}
{"step": 2440, "loss": 0.005193, "loss_mean50": 0.003763, "lr": 2.4418552114005585e-05, "grad_norm": 0.0478, "samples": 78080, "wall_s": 4808.7, "cuda_gb": 16.86}
{"step": 2460, "loss": 0.004165, "loss_mean50": 0.00396, "lr": 2.4402422856259944e-05, "grad_norm": 0.0389, "samples": 78720, "wall_s": 4847.7, "cuda_gb": 16.86}
{"step": 2480, "loss": 0.003884, "loss_mean50": 0.004072, "lr": 2.43860790372428e-05, "grad_norm": 0.0316, "samples": 79360, "wall_s": 4886.5, "cuda_gb": 16.86}
{"step": 2500, "loss": 0.002905, "loss_mean50": 0.003949, "lr": 2.4369520986151623e-05, "grad_norm": 0.0209, "samples": 80000, "wall_s": 4925.3, "cuda_gb": 16.86}
{"step": 2520, "loss": 0.003092, "loss_mean50": 0.003984, "lr": 2.435274903649894e-05, "grad_norm": 0.0246, "samples": 80640, "wall_s": 4997.9, "cuda_gb": 16.86}
{"step": 2540, "loss": 0.002855, "loss_mean50": 0.003649, "lr": 2.4335763526105632e-05, "grad_norm": 0.0329, "samples": 81280, "wall_s": 5037.0, "cuda_gb": 16.86}
{"step": 2560, "loss": 0.004275, "loss_mean50": 0.003562, "lr": 2.4318564797094113e-05, "grad_norm": 0.0311, "samples": 81920, "wall_s": 5076.1, "cuda_gb": 16.86}
{"step": 2580, "loss": 0.003234, "loss_mean50": 0.003772, "lr": 2.4301153195881467e-05, "grad_norm": 0.0332, "samples": 82560, "wall_s": 5115.0, "cuda_gb": 16.86}
{"step": 2600, "loss": 0.002902, "loss_mean50": 0.003897, "lr": 2.4283529073172444e-05, "grad_norm": 0.0265, "samples": 83200, "wall_s": 5154.2, "cuda_gb": 16.86}
{"step": 2620, "loss": 0.003816, "loss_mean50": 0.003698, "lr": 2.42656927839524e-05, "grad_norm": 0.0394, "samples": 83840, "wall_s": 5193.4, "cuda_gb": 16.86}
{"step": 2640, "loss": 0.001855, "loss_mean50": 0.003456, "lr": 2.4247644687480155e-05, "grad_norm": 0.0239, "samples": 84480, "wall_s": 5232.5, "cuda_gb": 16.86}
{"step": 2660, "loss": 0.002827, "loss_mean50": 0.003587, "lr": 2.4229385147280765e-05, "grad_norm": 0.0291, "samples": 85120, "wall_s": 5271.6, "cuda_gb": 16.86}
{"step": 2680, "loss": 0.002293, "loss_mean50": 0.003671, "lr": 2.4210914531138177e-05, "grad_norm": 0.0353, "samples": 85760, "wall_s": 5310.5, "cuda_gb": 16.86}
{"step": 2700, "loss": 0.004166, "loss_mean50": 0.00372, "lr": 2.419223321108783e-05, "grad_norm": 0.0328, "samples": 86400, "wall_s": 5349.6, "cuda_gb": 16.86}
{"step": 2720, "loss": 0.00292, "loss_mean50": 0.003643, "lr": 2.4173341563409186e-05, "grad_norm": 0.0396, "samples": 87040, "wall_s": 5388.7, "cuda_gb": 16.86}
{"step": 2740, "loss": 0.003284, "loss_mean50": 0.003592, "lr": 2.41542399686181e-05, "grad_norm": 0.034, "samples": 87680, "wall_s": 5427.6, "cuda_gb": 16.86}
{"step": 2760, "loss": 0.003666, "loss_mean50": 0.003353, "lr": 2.413492881145921e-05, "grad_norm": 0.0434, "samples": 88320, "wall_s": 5466.7, "cuda_gb": 16.86}
{"step": 2780, "loss": 0.003238, "loss_mean50": 0.003578, "lr": 2.4115408480898146e-05, "grad_norm": 0.0441, "samples": 88960, "wall_s": 5505.7, "cuda_gb": 16.86}
{"step": 2800, "loss": 0.002789, "loss_mean50": 0.003639, "lr": 2.4095679370113723e-05, "grad_norm": 0.0278, "samples": 89600, "wall_s": 5544.9, "cuda_gb": 16.86}
{"step": 2820, "loss": 0.00337, "loss_mean50": 0.003671, "lr": 2.407574187649e-05, "grad_norm": 0.0331, "samples": 90240, "wall_s": 5583.9, "cuda_gb": 16.86}
{"step": 2840, "loss": 0.004437, "loss_mean50": 0.003413, "lr": 2.4055596401608298e-05, "grad_norm": 0.0257, "samples": 90880, "wall_s": 5623.0, "cuda_gb": 16.86}
{"step": 2860, "loss": 0.004476, "loss_mean50": 0.003451, "lr": 2.4035243351239095e-05, "grad_norm": 0.0345, "samples": 91520, "wall_s": 5662.0, "cuda_gb": 16.86}
{"step": 2880, "loss": 0.003875, "loss_mean50": 0.003446, "lr": 2.4014683135333848e-05, "grad_norm": 0.0379, "samples": 92160, "wall_s": 5701.0, "cuda_gb": 16.86}
{"step": 2900, "loss": 0.004099, "loss_mean50": 0.003507, "lr": 2.3993916168016762e-05, "grad_norm": 0.0445, "samples": 92800, "wall_s": 5740.2, "cuda_gb": 16.86}
{"step": 2920, "loss": 0.001313, "loss_mean50": 0.003373, "lr": 2.3972942867576424e-05, "grad_norm": 0.0182, "samples": 93440, "wall_s": 5779.3, "cuda_gb": 16.86}
{"step": 2940, "loss": 0.003076, "loss_mean50": 0.00348, "lr": 2.3951763656457387e-05, "grad_norm": 0.0325, "samples": 94080, "wall_s": 5818.4, "cuda_gb": 16.86}
{"step": 2960, "loss": 0.005723, "loss_mean50": 0.003593, "lr": 2.3930378961251658e-05, "grad_norm": 0.0519, "samples": 94720, "wall_s": 5857.4, "cuda_gb": 16.86}
{"step": 2980, "loss": 0.003309, "loss_mean50": 0.003771, "lr": 2.3908789212690112e-05, "grad_norm": 0.0442, "samples": 95360, "wall_s": 5896.5, "cuda_gb": 16.86}
{"step": 3000, "loss": 0.003056, "loss_mean50": 0.003568, "lr": 2.3886994845633805e-05, "grad_norm": 0.0487, "samples": 96000, "wall_s": 5935.7, "cuda_gb": 16.86}
{"step": 3000, "val_flow_mse": 0.003161}
{"step": 3020, "loss": 0.002863, "loss_mean50": 0.003362, "lr": 2.3864996299065235e-05, "grad_norm": 0.0289, "samples": 96640, "wall_s": 6003.9, "cuda_gb": 16.86}
{"step": 3040, "loss": 0.00247, "loss_mean50": 0.003322, "lr": 2.384279401607947e-05, "grad_norm": 0.0199, "samples": 97280, "wall_s": 6042.9, "cuda_gb": 16.86}
{"step": 3060, "loss": 0.004347, "loss_mean50": 0.00348, "lr": 2.3820388443875256e-05, "grad_norm": 0.0351, "samples": 97920, "wall_s": 6082.0, "cuda_gb": 16.86}
{"step": 3080, "loss": 0.002534, "loss_mean50": 0.003687, "lr": 2.379778003374598e-05, "grad_norm": 0.0278, "samples": 98560, "wall_s": 6121.0, "cuda_gb": 16.86}
{"step": 3100, "loss": 0.003201, "loss_mean50": 0.003955, "lr": 2.3774969241070605e-05, "grad_norm": 0.0498, "samples": 99200, "wall_s": 6160.2, "cuda_gb": 16.86}
{"step": 3120, "loss": 0.006163, "loss_mean50": 0.003757, "lr": 2.375195652530448e-05, "grad_norm": 0.0382, "samples": 99840, "wall_s": 6199.2, "cuda_gb": 16.86}
{"step": 3140, "loss": 0.004194, "loss_mean50": 0.003632, "lr": 2.3728742349970097e-05, "grad_norm": 0.0436, "samples": 100480, "wall_s": 6238.2, "cuda_gb": 16.86}
{"step": 3160, "loss": 0.002945, "loss_mean50": 0.003348, "lr": 2.3705327182647743e-05, "grad_norm": 0.0324, "samples": 101120, "wall_s": 6277.3, "cuda_gb": 16.86}
{"step": 3180, "loss": 0.002846, "loss_mean50": 0.003162, "lr": 2.3681711494966095e-05, "grad_norm": 0.0271, "samples": 101760, "wall_s": 6316.3, "cuda_gb": 16.86}
{"step": 3200, "loss": 0.005161, "loss_mean50": 0.003276, "lr": 2.365789576259271e-05, "grad_norm": 0.0468, "samples": 102400, "wall_s": 6355.5, "cuda_gb": 16.86}
{"step": 3220, "loss": 0.003351, "loss_mean50": 0.003344, "lr": 2.363388046522445e-05, "grad_norm": 0.0588, "samples": 103040, "wall_s": 6394.6, "cuda_gb": 16.86}
{"step": 3240, "loss": 0.002934, "loss_mean50": 0.003744, "lr": 2.3609666086577818e-05, "grad_norm": 0.0234, "samples": 103680, "wall_s": 6433.6, "cuda_gb": 16.86}
{"step": 3260, "loss": 0.003354, "loss_mean50": 0.00372, "lr": 2.358525311437922e-05, "grad_norm": 0.0368, "samples": 104320, "wall_s": 6472.7, "cuda_gb": 16.86}
{"step": 3280, "loss": 0.002015, "loss_mean50": 0.003539, "lr": 2.356064204035513e-05, "grad_norm": 0.0286, "samples": 104960, "wall_s": 6511.8, "cuda_gb": 16.86}
{"step": 3300, "loss": 0.006753, "loss_mean50": 0.003442, "lr": 2.3535833360222197e-05, "grad_norm": 0.0544, "samples": 105600, "wall_s": 6550.9, "cuda_gb": 16.86}
{"step": 3320, "loss": 0.002701, "loss_mean50": 0.003368, "lr": 2.351082757367725e-05, "grad_norm": 0.0229, "samples": 106240, "wall_s": 6590.0, "cuda_gb": 16.86}
{"step": 3340, "loss": 0.003714, "loss_mean50": 0.003394, "lr": 2.3485625184387237e-05, "grad_norm": 0.0289, "samples": 106880, "wall_s": 6629.0, "cuda_gb": 16.86}
{"step": 3360, "loss": 0.003343, "loss_mean50": 0.003457, "lr": 2.3460226699979094e-05, "grad_norm": 0.033, "samples": 107520, "wall_s": 6668.0, "cuda_gb": 16.86}
{"step": 3380, "loss": 0.001935, "loss_mean50": 0.003329, "lr": 2.3434632632029495e-05, "grad_norm": 0.0254, "samples": 108160, "wall_s": 6707.1, "cuda_gb": 16.86}
{"step": 3400, "loss": 0.004246, "loss_mean50": 0.003306, "lr": 2.340884349605457e-05, "grad_norm": 0.0369, "samples": 108800, "wall_s": 6746.1, "cuda_gb": 16.86}
{"step": 3420, "loss": 0.002968, "loss_mean50": 0.003254, "lr": 2.3382859811499506e-05, "grad_norm": 0.0312, "samples": 109440, "wall_s": 6785.3, "cuda_gb": 16.86}
{"step": 3440, "loss": 0.002689, "loss_mean50": 0.003261, "lr": 2.3356682101728088e-05, "grad_norm": 0.0327, "samples": 110080, "wall_s": 6824.3, "cuda_gb": 16.86}
{"step": 3460, "loss": 0.003148, "loss_mean50": 0.003288, "lr": 2.3330310894012176e-05, "grad_norm": 0.0294, "samples": 110720, "wall_s": 6863.3, "cuda_gb": 16.86}
{"step": 3480, "loss": 0.002615, "loss_mean50": 0.003562, "lr": 2.3303746719521045e-05, "grad_norm": 0.0229, "samples": 111360, "wall_s": 6902.4, "cuda_gb": 16.86}
{"step": 3500, "loss": 0.002561, "loss_mean50": 0.003611, "lr": 2.3276990113310735e-05, "grad_norm": 0.0252, "samples": 112000, "wall_s": 6941.4, "cuda_gb": 16.86}
{"step": 3520, "loss": 0.004331, "loss_mean50": 0.003602, "lr": 2.3250041614313225e-05, "grad_norm": 0.0603, "samples": 112640, "wall_s": 6980.6, "cuda_gb": 16.86}
{"step": 3540, "loss": 0.002576, "loss_mean50": 0.00348, "lr": 2.3222901765325615e-05, "grad_norm": 0.0285, "samples": 113280, "wall_s": 7019.6, "cuda_gb": 16.86}
{"step": 3560, "loss": 0.003941, "loss_mean50": 0.003184, "lr": 2.3195571112999185e-05, "grad_norm": 0.0516, "samples": 113920, "wall_s": 7058.7, "cuda_gb": 16.86}
{"step": 3580, "loss": 0.002963, "loss_mean50": 0.003172, "lr": 2.3168050207828366e-05, "grad_norm": 0.0449, "samples": 114560, "wall_s": 7097.8, "cuda_gb": 16.86}
{"step": 3600, "loss": 0.003467, "loss_mean50": 0.003558, "lr": 2.3140339604139666e-05, "grad_norm": 0.0364, "samples": 115200, "wall_s": 7137.0, "cuda_gb": 16.86}
{"step": 3620, "loss": 0.002585, "loss_mean50": 0.00336, "lr": 2.3112439860080513e-05, "grad_norm": 0.0205, "samples": 115840, "wall_s": 7176.1, "cuda_gb": 16.86}
{"step": 3640, "loss": 0.003226, "loss_mean50": 0.00332, "lr": 2.308435153760799e-05, "grad_norm": 0.0321, "samples": 116480, "wall_s": 7215.2, "cuda_gb": 16.86}
{"step": 3660, "loss": 0.002493, "loss_mean50": 0.003165, "lr": 2.305607520247755e-05, "grad_norm": 0.0295, "samples": 117120, "wall_s": 7254.2, "cuda_gb": 16.86}
{"step": 3680, "loss": 0.003995, "loss_mean50": 0.003325, "lr": 2.302761142423158e-05, "grad_norm": 0.0313, "samples": 117760, "wall_s": 7293.4, "cuda_gb": 16.86}
{"step": 3700, "loss": 0.003086, "loss_mean50": 0.00317, "lr": 2.2998960776187957e-05, "grad_norm": 0.0337, "samples": 118400, "wall_s": 7332.4, "cuda_gb": 16.86}
{"step": 3720, "loss": 0.004277, "loss_mean50": 0.003346, "lr": 2.2970123835428493e-05, "grad_norm": 0.027, "samples": 119040, "wall_s": 7371.6, "cuda_gb": 16.86}
{"step": 3740, "loss": 0.002169, "loss_mean50": 0.00342, "lr": 2.2941101182787307e-05, "grad_norm": 0.0227, "samples": 119680, "wall_s": 7410.6, "cuda_gb": 16.86}
{"step": 3760, "loss": 0.002332, "loss_mean50": 0.003108, "lr": 2.2911893402839136e-05, "grad_norm": 0.0302, "samples": 120320, "wall_s": 7449.7, "cuda_gb": 16.86}
{"step": 3780, "loss": 0.00359, "loss_mean50": 0.003154, "lr": 2.288250108388755e-05, "grad_norm": 0.0391, "samples": 120960, "wall_s": 7488.8, "cuda_gb": 16.86}
{"step": 3800, "loss": 0.001907, "loss_mean50": 0.003194, "lr": 2.2852924817953105e-05, "grad_norm": 0.0207, "samples": 121600, "wall_s": 7527.8, "cuda_gb": 16.86}
{"step": 3820, "loss": 0.003192, "loss_mean50": 0.003201, "lr": 2.2823165200761428e-05, "grad_norm": 0.0251, "samples": 122240, "wall_s": 7566.9, "cuda_gb": 16.86}
{"step": 3840, "loss": 0.004049, "loss_mean50": 0.00311, "lr": 2.27932228317312e-05, "grad_norm": 0.0789, "samples": 122880, "wall_s": 7606.0, "cuda_gb": 16.86}
{"step": 3860, "loss": 0.003403, "loss_mean50": 0.003198, "lr": 2.2763098313962093e-05, "grad_norm": 0.0217, "samples": 123520, "wall_s": 7645.0, "cuda_gb": 16.86}
{"step": 3880, "loss": 0.002925, "loss_mean50": 0.003211, "lr": 2.2732792254222633e-05, "grad_norm": 0.0203, "samples": 124160, "wall_s": 7684.1, "cuda_gb": 16.86}
{"step": 3900, "loss": 0.003258, "loss_mean50": 0.003373, "lr": 2.2702305262937958e-05, "grad_norm": 0.0225, "samples": 124800, "wall_s": 7723.2, "cuda_gb": 16.86}
{"step": 3920, "loss": 0.001825, "loss_mean50": 0.003376, "lr": 2.2671637954177526e-05, "grad_norm": 0.0263, "samples": 125440, "wall_s": 7762.5, "cuda_gb": 16.86}
{"step": 3940, "loss": 0.001681, "loss_mean50": 0.003437, "lr": 2.2640790945642772e-05, "grad_norm": 0.0229, "samples": 126080, "wall_s": 7801.7, "cuda_gb": 16.86}
{"step": 3960, "loss": 0.00345, "loss_mean50": 0.003236, "lr": 2.2609764858654632e-05, "grad_norm": 0.0525, "samples": 126720, "wall_s": 7840.7, "cuda_gb": 16.86}
{"step": 3980, "loss": 0.004608, "loss_mean50": 0.003134, "lr": 2.2578560318141043e-05, "grad_norm": 0.0368, "samples": 127360, "wall_s": 7879.8, "cuda_gb": 16.86}
{"step": 4000, "loss": 0.001899, "loss_mean50": 0.003182, "lr": 2.254717795262436e-05, "grad_norm": 0.0453, "samples": 128000, "wall_s": 7918.9, "cuda_gb": 16.86}
{"step": 4000, "val_flow_mse": 0.002894}
{"step": 4020, "loss": 0.003213, "loss_mean50": 0.003386, "lr": 2.2515618394208694e-05, "grad_norm": 0.0269, "samples": 128640, "wall_s": 7986.9, "cuda_gb": 16.86}
{"step": 4040, "loss": 0.001878, "loss_mean50": 0.003192, "lr": 2.248388227856717e-05, "grad_norm": 0.0242, "samples": 129280, "wall_s": 8026.0, "cuda_gb": 16.86}
{"step": 4060, "loss": 0.003555, "loss_mean50": 0.003031, "lr": 2.2451970244929137e-05, "grad_norm": 0.0448, "samples": 129920, "wall_s": 8065.1, "cuda_gb": 16.86}
{"step": 4080, "loss": 0.002319, "loss_mean50": 0.002925, "lr": 2.2419882936067294e-05, "grad_norm": 0.0392, "samples": 130560, "wall_s": 8104.4, "cuda_gb": 16.86}
{"step": 4100, "loss": 0.009083, "loss_mean50": 0.0031, "lr": 2.238762099828472e-05, "grad_norm": 0.0702, "samples": 131200, "wall_s": 8143.5, "cuda_gb": 16.86}
{"step": 4120, "loss": 0.00451, "loss_mean50": 0.003153, "lr": 2.235518508140189e-05, "grad_norm": 0.042, "samples": 131840, "wall_s": 8182.7, "cuda_gb": 16.86}
{"step": 4140, "loss": 0.002726, "loss_mean50": 0.003235, "lr": 2.232257583874355e-05, "grad_norm": 0.0324, "samples": 132480, "wall_s": 8221.9, "cuda_gb": 16.86}
{"step": 4160, "loss": 0.004389, "loss_mean50": 0.003078, "lr": 2.2289793927125603e-05, "grad_norm": 0.0415, "samples": 133120, "wall_s": 8260.9, "cuda_gb": 16.86}
{"step": 4180, "loss": 0.003022, "loss_mean50": 0.00311, "lr": 2.225684000684182e-05, "grad_norm": 0.0434, "samples": 133760, "wall_s": 8299.9, "cuda_gb": 16.86}
{"step": 4200, "loss": 0.001885, "loss_mean50": 0.003211, "lr": 2.22237147416506e-05, "grad_norm": 0.0305, "samples": 134400, "wall_s": 8338.9, "cuda_gb": 16.86}
{"step": 4220, "loss": 0.003343, "loss_mean50": 0.003116, "lr": 2.2190418798761562e-05, "grad_norm": 0.0234, "samples": 135040, "wall_s": 8378.0, "cuda_gb": 16.86}
{"step": 4240, "loss": 0.004352, "loss_mean50": 0.002911, "lr": 2.2156952848822125e-05, "grad_norm": 0.0524, "samples": 135680, "wall_s": 8417.0, "cuda_gb": 16.86}
{"step": 4260, "loss": 0.002917, "loss_mean50": 0.00305, "lr": 2.212331756590397e-05, "grad_norm": 0.0722, "samples": 136320, "wall_s": 8456.0, "cuda_gb": 16.86}
{"step": 4280, "loss": 0.002972, "loss_mean50": 0.002908, "lr": 2.208951362748953e-05, "grad_norm": 0.036, "samples": 136960, "wall_s": 8495.1, "cuda_gb": 16.86}
{"step": 4300, "loss": 0.003171, "loss_mean50": 0.003041, "lr": 2.2055541714458252e-05, "grad_norm": 0.0297, "samples": 137600, "wall_s": 8534.1, "cuda_gb": 16.86}
{"step": 4320, "loss": 0.00273, "loss_mean50": 0.003174, "lr": 2.2021402511072963e-05, "grad_norm": 0.0532, "samples": 138240, "wall_s": 8573.2, "cuda_gb": 16.86}
{"step": 4340, "loss": 0.004613, "loss_mean50": 0.003307, "lr": 2.1987096704966034e-05, "grad_norm": 0.0375, "samples": 138880, "wall_s": 8612.2, "cuda_gb": 16.86}
{"step": 4360, "loss": 0.00138, "loss_mean50": 0.003116, "lr": 2.1952624987125566e-05, "grad_norm": 0.0322, "samples": 139520, "wall_s": 8651.2, "cuda_gb": 16.86}
{"step": 4380, "loss": 0.002955, "loss_mean50": 0.003161, "lr": 2.1917988051881456e-05, "grad_norm": 0.0327, "samples": 140160, "wall_s": 8690.2, "cuda_gb": 16.86}
{"step": 4400, "loss": 0.002084, "loss_mean50": 0.002978, "lr": 2.188318659689139e-05, "grad_norm": 0.024, "samples": 140800, "wall_s": 8729.2, "cuda_gb": 16.86}
{"step": 4420, "loss": 0.00227, "loss_mean50": 0.003096, "lr": 2.1848221323126857e-05, "grad_norm": 0.0175, "samples": 141440, "wall_s": 8768.2, "cuda_gb": 16.86}
{"step": 4440, "loss": 0.001878, "loss_mean50": 0.003185, "lr": 2.1813092934858935e-05, "grad_norm": 0.0354, "samples": 142080, "wall_s": 8807.2, "cuda_gb": 16.86}
{"step": 4460, "loss": 0.002796, "loss_mean50": 0.002976, "lr": 2.17778021396442e-05, "grad_norm": 0.0307, "samples": 142720, "wall_s": 8846.3, "cuda_gb": 16.86}
{"step": 4480, "loss": 0.002693, "loss_mean50": 0.002857, "lr": 2.1742349648310403e-05, "grad_norm": 0.0663, "samples": 143360, "wall_s": 8885.5, "cuda_gb": 16.86}
{"step": 4500, "loss": 0.001437, "loss_mean50": 0.003031, "lr": 2.1706736174942196e-05, "grad_norm": 0.0234, "samples": 144000, "wall_s": 8924.6, "cuda_gb": 16.86}
{"step": 4520, "loss": 0.004832, "loss_mean50": 0.003023, "lr": 2.1670962436866725e-05, "grad_norm": 0.0305, "samples": 144640, "wall_s": 8963.8, "cuda_gb": 16.86}
{"step": 4540, "loss": 0.00514, "loss_mean50": 0.003013, "lr": 2.16350291546392e-05, "grad_norm": 0.0318, "samples": 145280, "wall_s": 9002.8, "cuda_gb": 16.86}
{"step": 4560, "loss": 0.001726, "loss_mean50": 0.003013, "lr": 2.159893705202836e-05, "grad_norm": 0.0321, "samples": 145920, "wall_s": 9042.0, "cuda_gb": 16.86}
{"step": 4580, "loss": 0.003825, "loss_mean50": 0.002968, "lr": 2.1562686856001923e-05, "grad_norm": 0.0318, "samples": 146560, "wall_s": 9081.1, "cuda_gb": 16.86}
{"step": 4600, "loss": 0.003647, "loss_mean50": 0.002878, "lr": 2.1526279296711908e-05, "grad_norm": 0.0262, "samples": 147200, "wall_s": 9120.3, "cuda_gb": 16.86}
{"step": 4620, "loss": 0.00374, "loss_mean50": 0.002945, "lr": 2.1489715107479956e-05, "grad_norm": 0.0455, "samples": 147840, "wall_s": 9159.4, "cuda_gb": 16.86}
{"step": 4640, "loss": 0.002574, "loss_mean50": 0.002858, "lr": 2.1452995024782553e-05, "grad_norm": 0.022, "samples": 148480, "wall_s": 9198.6, "cuda_gb": 16.86}
{"step": 4660, "loss": 0.003107, "loss_mean50": 0.003027, "lr": 2.1416119788236198e-05, "grad_norm": 0.0214, "samples": 149120, "wall_s": 9237.8, "cuda_gb": 16.86}
{"step": 4680, "loss": 0.002005, "loss_mean50": 0.002918, "lr": 2.1379090140582484e-05, "grad_norm": 0.0216, "samples": 149760, "wall_s": 9276.9, "cuda_gb": 16.86}
{"step": 4700, "loss": 0.002245, "loss_mean50": 0.002939, "lr": 2.1341906827673175e-05, "grad_norm": 0.033, "samples": 150400, "wall_s": 9315.9, "cuda_gb": 16.86}
{"step": 4720, "loss": 0.002582, "loss_mean50": 0.002837, "lr": 2.130457059845515e-05, "grad_norm": 0.0255, "samples": 151040, "wall_s": 9355.0, "cuda_gb": 16.86}
{"step": 4740, "loss": 0.001638, "loss_mean50": 0.002857, "lr": 2.1267082204955335e-05, "grad_norm": 0.0306, "samples": 151680, "wall_s": 9394.1, "cuda_gb": 16.86}
{"step": 4760, "loss": 0.003595, "loss_mean50": 0.002886, "lr": 2.1229442402265554e-05, "grad_norm": 0.0525, "samples": 152320, "wall_s": 9433.1, "cuda_gb": 16.86}
{"step": 4780, "loss": 0.005197, "loss_mean50": 0.002887, "lr": 2.1191651948527313e-05, "grad_norm": 0.0324, "samples": 152960, "wall_s": 9472.1, "cuda_gb": 16.86}
{"step": 4800, "loss": 0.004448, "loss_mean50": 0.002859, "lr": 2.1153711604916535e-05, "grad_norm": 0.0351, "samples": 153600, "wall_s": 9511.1, "cuda_gb": 16.86}
{"step": 4820, "loss": 0.003179, "loss_mean50": 0.003019, "lr": 2.111562213562823e-05, "grad_norm": 0.0283, "samples": 154240, "wall_s": 9550.1, "cuda_gb": 16.86}
{"step": 4840, "loss": 0.001973, "loss_mean50": 0.002774, "lr": 2.1077384307861093e-05, "grad_norm": 0.0286, "samples": 154880, "wall_s": 9589.2, "cuda_gb": 16.86}
{"step": 4860, "loss": 0.003341, "loss_mean50": 0.002643, "lr": 2.103899889180206e-05, "grad_norm": 0.0374, "samples": 155520, "wall_s": 9628.4, "cuda_gb": 16.86}
{"step": 4880, "loss": 0.00585, "loss_mean50": 0.002573, "lr": 2.100046666061081e-05, "grad_norm": 0.0369, "samples": 156160, "wall_s": 9667.4, "cuda_gb": 16.86}
{"step": 4900, "loss": 0.002354, "loss_mean50": 0.002778, "lr": 2.096178839040414e-05, "grad_norm": 0.0231, "samples": 156800, "wall_s": 9706.5, "cuda_gb": 16.86}
{"step": 4920, "loss": 0.004332, "loss_mean50": 0.002966, "lr": 2.0922964860240404e-05, "grad_norm": 0.0425, "samples": 157440, "wall_s": 9745.6, "cuda_gb": 16.86}
{"step": 4940, "loss": 0.001768, "loss_mean50": 0.002847, "lr": 2.088399685210376e-05, "grad_norm": 0.0371, "samples": 158080, "wall_s": 9784.6, "cuda_gb": 16.86}
{"step": 4960, "loss": 0.002559, "loss_mean50": 0.002783, "lr": 2.0844885150888443e-05, "grad_norm": 0.0619, "samples": 158720, "wall_s": 9823.6, "cuda_gb": 16.86}
{"step": 4980, "loss": 0.001236, "loss_mean50": 0.00275, "lr": 2.0805630544382978e-05, "grad_norm": 0.025, "samples": 159360, "wall_s": 9862.6, "cuda_gb": 16.86}
{"step": 5000, "loss": 0.001796, "loss_mean50": 0.002944, "lr": 2.076623382325427e-05, "grad_norm": 0.0301, "samples": 160000, "wall_s": 9901.6, "cuda_gb": 16.86}
{"step": 5000, "val_flow_mse": 0.002436, "val_action_mae": 0.019881, "best": true}
{"step": 5020, "loss": 0.003273, "loss_mean50": 0.002922, "lr": 2.0726695781031705e-05, "grad_norm": 0.0363, "samples": 160640, "wall_s": 10036.3, "cuda_gb": 16.86}
{"step": 5040, "loss": 0.002959, "loss_mean50": 0.00291, "lr": 2.0687017214091166e-05, "grad_norm": 0.0713, "samples": 161280, "wall_s": 10075.2, "cuda_gb": 16.86}
{"step": 5060, "loss": 0.003963, "loss_mean50": 0.002845, "lr": 2.064719892163898e-05, "grad_norm": 0.0347, "samples": 161920, "wall_s": 10114.2, "cuda_gb": 16.86}
{"step": 5080, "loss": 0.001777, "loss_mean50": 0.002749, "lr": 2.0607241705695827e-05, "grad_norm": 0.0209, "samples": 162560, "wall_s": 10153.0, "cuda_gb": 16.86}
{"step": 5100, "loss": 0.001998, "loss_mean50": 0.002686, "lr": 2.05671463710806e-05, "grad_norm": 0.0218, "samples": 163200, "wall_s": 10191.8, "cuda_gb": 16.86}
{"step": 5120, "loss": 0.002062, "loss_mean50": 0.002713, "lr": 2.0526913725394155e-05, "grad_norm": 0.0307, "samples": 163840, "wall_s": 10230.5, "cuda_gb": 16.86}
{"step": 5140, "loss": 0.005038, "loss_mean50": 0.002721, "lr": 2.0486544579003098e-05, "grad_norm": 0.0739, "samples": 164480, "wall_s": 10269.4, "cuda_gb": 16.86}
{"step": 5160, "loss": 0.003789, "loss_mean50": 0.002608, "lr": 2.044603974502341e-05, "grad_norm": 0.0366, "samples": 165120, "wall_s": 10308.3, "cuda_gb": 16.86}
{"step": 5180, "loss": 0.003223, "loss_mean50": 0.00274, "lr": 2.0405400039304117e-05, "grad_norm": 0.0303, "samples": 165760, "wall_s": 10347.2, "cuda_gb": 16.86}
{"step": 5200, "loss": 0.003343, "loss_mean50": 0.002938, "lr": 2.0364626280410812e-05, "grad_norm": 0.0342, "samples": 166400, "wall_s": 10386.1, "cuda_gb": 16.86}
{"step": 5220, "loss": 0.002414, "loss_mean50": 0.002912, "lr": 2.03237192896092e-05, "grad_norm": 0.0423, "samples": 167040, "wall_s": 10425.0, "cuda_gb": 16.86}
{"step": 5240, "loss": 0.001398, "loss_mean50": 0.002714, "lr": 2.028267989084855e-05, "grad_norm": 0.023, "samples": 167680, "wall_s": 10463.9, "cuda_gb": 16.86}
{"step": 5260, "loss": 0.001926, "loss_mean50": 0.002707, "lr": 2.024150891074508e-05, "grad_norm": 0.0242, "samples": 168320, "wall_s": 10502.7, "cuda_gb": 16.86}
{"step": 5280, "loss": 0.004688, "loss_mean50": 0.002886, "lr": 2.0200207178565332e-05, "grad_norm": 0.0324, "samples": 168960, "wall_s": 10541.7, "cuda_gb": 16.86}
{"step": 5300, "loss": 0.002247, "loss_mean50": 0.002752, "lr": 2.015877552620946e-05, "grad_norm": 0.0215, "samples": 169600, "wall_s": 10580.7, "cuda_gb": 16.86}
{"step": 5320, "loss": 0.001931, "loss_mean50": 0.002753, "lr": 2.011721478819446e-05, "grad_norm": 0.0332, "samples": 170240, "wall_s": 10619.7, "cuda_gb": 16.86}
{"step": 5340, "loss": 0.00183, "loss_mean50": 0.002671, "lr": 2.007552580163739e-05, "grad_norm": 0.0196, "samples": 170880, "wall_s": 10658.6, "cuda_gb": 16.86}
{"step": 5360, "loss": 0.001886, "loss_mean50": 0.002709, "lr": 2.003370940623848e-05, "grad_norm": 0.0269, "samples": 171520, "wall_s": 10697.6, "cuda_gb": 16.86}
{"step": 5380, "loss": 0.001962, "loss_mean50": 0.002825, "lr": 1.9991766444264238e-05, "grad_norm": 0.0321, "samples": 172160, "wall_s": 10736.6, "cuda_gb": 16.86}
{"step": 5400, "loss": 0.004168, "loss_mean50": 0.002964, "lr": 1.9949697760530468e-05, "grad_norm": 0.053, "samples": 172800, "wall_s": 10775.6, "cuda_gb": 16.86}
{"step": 5420, "loss": 0.003326, "loss_mean50": 0.002934, "lr": 1.9907504202385275e-05, "grad_norm": 0.0768, "samples": 173440, "wall_s": 10814.5, "cuda_gb": 16.86}
{"step": 5440, "loss": 0.002783, "loss_mean50": 0.002826, "lr": 1.9865186619691983e-05, "grad_norm": 0.0439, "samples": 174080, "wall_s": 10853.5, "cuda_gb": 16.86}
{"step": 5460, "loss": 0.004136, "loss_mean50": 0.002704, "lr": 1.9822745864812015e-05, "grad_norm": 0.0427, "samples": 174720, "wall_s": 10892.3, "cuda_gb": 16.86}
{"step": 5480, "loss": 0.002351, "loss_mean50": 0.002768, "lr": 1.9780182792587738e-05, "grad_norm": 0.0442, "samples": 175360, "wall_s": 10931.2, "cuda_gb": 16.86}
{"step": 5500, "loss": 0.002547, "loss_mean50": 0.002796, "lr": 1.9737498260325232e-05, "grad_norm": 0.0238, "samples": 176000, "wall_s": 10970.1, "cuda_gb": 16.86}
{"step": 5520, "loss": 0.003439, "loss_mean50": 0.002673, "lr": 1.969469312777703e-05, "grad_norm": 0.0945, "samples": 176640, "wall_s": 11008.9, "cuda_gb": 16.86}
{"step": 5540, "loss": 0.005075, "loss_mean50": 0.002628, "lr": 1.9651768257124793e-05, "grad_norm": 0.0387, "samples": 177280, "wall_s": 11047.7, "cuda_gb": 16.86}
{"step": 5560, "loss": 0.001814, "loss_mean50": 0.002656, "lr": 1.9608724512961953e-05, "grad_norm": 0.0291, "samples": 177920, "wall_s": 11086.7, "cuda_gb": 16.86}
{"step": 5580, "loss": 0.000768, "loss_mean50": 0.002811, "lr": 1.9565562762276297e-05, "grad_norm": 0.0171, "samples": 178560, "wall_s": 11125.6, "cuda_gb": 16.86}
{"step": 5600, "loss": 0.002535, "loss_mean50": 0.002554, "lr": 1.9522283874432492e-05, "grad_norm": 0.0409, "samples": 179200, "wall_s": 11164.4, "cuda_gb": 16.86}
{"step": 5620, "loss": 0.002339, "loss_mean50": 0.002583, "lr": 1.9478888721154596e-05, "grad_norm": 0.0651, "samples": 179840, "wall_s": 11203.3, "cuda_gb": 16.86}
{"step": 5640, "loss": 0.002032, "loss_mean50": 0.002624, "lr": 1.9435378176508466e-05, "grad_norm": 0.0473, "samples": 180480, "wall_s": 11242.3, "cuda_gb": 16.86}
{"step": 5660, "loss": 0.003597, "loss_mean50": 0.002739, "lr": 1.9391753116884205e-05, "grad_norm": 0.0605, "samples": 181120, "wall_s": 11281.2, "cuda_gb": 16.86}
{"step": 5680, "loss": 0.002206, "loss_mean50": 0.002704, "lr": 1.934801442097845e-05, "grad_norm": 0.0323, "samples": 181760, "wall_s": 11320.1, "cuda_gb": 16.86}
{"step": 5700, "loss": 0.003807, "loss_mean50": 0.002703, "lr": 1.9304162969776722e-05, "grad_norm": 0.0664, "samples": 182400, "wall_s": 11359.0, "cuda_gb": 16.86}
{"step": 5720, "loss": 0.00382, "loss_mean50": 0.002579, "lr": 1.9260199646535644e-05, "grad_norm": 0.0383, "samples": 183040, "wall_s": 11397.8, "cuda_gb": 16.86}
{"step": 5740, "loss": 0.002654, "loss_mean50": 0.00249, "lr": 1.921612533676518e-05, "grad_norm": 0.0302, "samples": 183680, "wall_s": 11436.7, "cuda_gb": 16.86}
{"step": 5760, "loss": 0.002548, "loss_mean50": 0.002321, "lr": 1.9171940928210784e-05, "grad_norm": 0.0332, "samples": 184320, "wall_s": 11475.4, "cuda_gb": 16.86}
{"step": 5780, "loss": 0.001231, "loss_mean50": 0.002342, "lr": 1.9127647310835513e-05, "grad_norm": 0.0248, "samples": 184960, "wall_s": 11514.4, "cuda_gb": 16.86}
{"step": 5800, "loss": 0.00294, "loss_mean50": 0.002323, "lr": 1.9083245376802126e-05, "grad_norm": 0.0297, "samples": 185600, "wall_s": 11553.3, "cuda_gb": 16.86}
{"step": 5820, "loss": 0.003079, "loss_mean50": 0.002468, "lr": 1.903873602045508e-05, "grad_norm": 0.0374, "samples": 186240, "wall_s": 11592.3, "cuda_gb": 16.86}
{"step": 5840, "loss": 0.003894, "loss_mean50": 0.00249, "lr": 1.8994120138302543e-05, "grad_norm": 0.0762, "samples": 186880, "wall_s": 11631.2, "cuda_gb": 16.86}
{"step": 5860, "loss": 0.002756, "loss_mean50": 0.002687, "lr": 1.894939862899833e-05, "grad_norm": 0.0298, "samples": 187520, "wall_s": 11670.0, "cuda_gb": 16.86}
{"step": 5880, "loss": 0.002032, "loss_mean50": 0.00277, "lr": 1.8904572393323796e-05, "grad_norm": 0.0476, "samples": 188160, "wall_s": 11708.9, "cuda_gb": 16.86}
{"step": 5900, "loss": 0.001721, "loss_mean50": 0.002623, "lr": 1.8859642334169693e-05, "grad_norm": 0.0263, "samples": 188800, "wall_s": 11747.9, "cuda_gb": 16.86}
{"step": 5920, "loss": 0.004459, "loss_mean50": 0.002581, "lr": 1.8814609356517996e-05, "grad_norm": 0.0659, "samples": 189440, "wall_s": 11786.8, "cuda_gb": 16.86}
{"step": 5940, "loss": 0.001589, "loss_mean50": 0.002404, "lr": 1.8769474367423657e-05, "grad_norm": 0.0193, "samples": 190080, "wall_s": 11825.8, "cuda_gb": 16.86}
{"step": 5960, "loss": 0.002662, "loss_mean50": 0.002428, "lr": 1.872423827599635e-05, "grad_norm": 0.0239, "samples": 190720, "wall_s": 11864.6, "cuda_gb": 16.86}
{"step": 5980, "loss": 0.003121, "loss_mean50": 0.002537, "lr": 1.8678901993382156e-05, "grad_norm": 0.0285, "samples": 191360, "wall_s": 11903.4, "cuda_gb": 16.86}
{"step": 6000, "loss": 0.002273, "loss_mean50": 0.002571, "lr": 1.8633466432745193e-05, "grad_norm": 0.0287, "samples": 192000, "wall_s": 11942.4, "cuda_gb": 16.86}
{"step": 6000, "val_flow_mse": 0.00234}
{"step": 6020, "loss": 0.00475, "loss_mean50": 0.002503, "lr": 1.8587932509249265e-05, "grad_norm": 0.0698, "samples": 192640, "wall_s": 12010.2, "cuda_gb": 16.86}
{"step": 6040, "loss": 0.003032, "loss_mean50": 0.002656, "lr": 1.8542301140039374e-05, "grad_norm": 0.0286, "samples": 193280, "wall_s": 12049.9, "cuda_gb": 16.86}
{"step": 6060, "loss": 0.001471, "loss_mean50": 0.002499, "lr": 1.849657324422329e-05, "grad_norm": 0.0248, "samples": 193920, "wall_s": 12089.0, "cuda_gb": 16.86}
{"step": 6080, "loss": 0.002622, "loss_mean50": 0.002522, "lr": 1.845074974285303e-05, "grad_norm": 0.0606, "samples": 194560, "wall_s": 12128.0, "cuda_gb": 16.86}
{"step": 6100, "loss": 0.002805, "loss_mean50": 0.002484, "lr": 1.8404831558906283e-05, "grad_norm": 0.0256, "samples": 195200, "wall_s": 12167.0, "cuda_gb": 16.86}
{"step": 6120, "loss": 0.002007, "loss_mean50": 0.002699, "lr": 1.8358819617267847e-05, "grad_norm": 0.0342, "samples": 195840, "wall_s": 12206.0, "cuda_gb": 16.86}
{"step": 6140, "loss": 0.001483, "loss_mean50": 0.00256, "lr": 1.831271484471099e-05, "grad_norm": 0.0214, "samples": 196480, "wall_s": 12244.9, "cuda_gb": 16.86}
{"step": 6160, "loss": 0.00219, "loss_mean50": 0.002679, "lr": 1.8266518169878774e-05, "grad_norm": 0.062, "samples": 197120, "wall_s": 12283.8, "cuda_gb": 16.86}
{"step": 6180, "loss": 0.003283, "loss_mean50": 0.002379, "lr": 1.822023052326536e-05, "grad_norm": 0.037, "samples": 197760, "wall_s": 12322.7, "cuda_gb": 16.86}
{"step": 6200, "loss": 0.00393, "loss_mean50": 0.002249, "lr": 1.8173852837197273e-05, "grad_norm": 0.0589, "samples": 198400, "wall_s": 12361.7, "cuda_gb": 16.86}
{"step": 6220, "loss": 0.002823, "loss_mean50": 0.002333, "lr": 1.81273860458146e-05, "grad_norm": 0.0779, "samples": 199040, "wall_s": 12400.5, "cuda_gb": 16.86}
{"step": 6240, "loss": 0.004328, "loss_mean50": 0.002563, "lr": 1.8080831085052197e-05, "grad_norm": 0.1364, "samples": 199680, "wall_s": 12439.3, "cuda_gb": 16.86}
{"step": 6260, "loss": 0.004069, "loss_mean50": 0.002491, "lr": 1.8034188892620834e-05, "grad_norm": 0.0306, "samples": 200320, "wall_s": 12478.4, "cuda_gb": 16.86}
{"step": 6280, "loss": 0.001872, "loss_mean50": 0.002513, "lr": 1.7987460407988294e-05, "grad_norm": 0.0339, "samples": 200960, "wall_s": 12517.4, "cuda_gb": 16.86}
{"step": 6300, "loss": 0.004148, "loss_mean50": 0.002434, "lr": 1.794064657236046e-05, "grad_norm": 0.0854, "samples": 201600, "wall_s": 12556.3, "cuda_gb": 16.86}
{"step": 6320, "loss": 0.001442, "loss_mean50": 0.002273, "lr": 1.7893748328662355e-05, "grad_norm": 0.0295, "samples": 202240, "wall_s": 12595.3, "cuda_gb": 16.86}
{"step": 6340, "loss": 0.002379, "loss_mean50": 0.002543, "lr": 1.7846766621519163e-05, "grad_norm": 0.0231, "samples": 202880, "wall_s": 12634.3, "cuda_gb": 16.86}
{"step": 6360, "loss": 0.001765, "loss_mean50": 0.00242, "lr": 1.7799702397237172e-05, "grad_norm": 0.0271, "samples": 203520, "wall_s": 12673.2, "cuda_gb": 16.86}
{"step": 6380, "loss": 0.002624, "loss_mean50": 0.002457, "lr": 1.775255660378474e-05, "grad_norm": 0.0574, "samples": 204160, "wall_s": 12712.0, "cuda_gb": 16.86}
{"step": 6400, "loss": 0.003121, "loss_mean50": 0.002492, "lr": 1.77053301907732e-05, "grad_norm": 0.0554, "samples": 204800, "wall_s": 12750.9, "cuda_gb": 16.86}
{"step": 6420, "loss": 0.002345, "loss_mean50": 0.002494, "lr": 1.765802410943771e-05, "grad_norm": 0.0288, "samples": 205440, "wall_s": 12789.9, "cuda_gb": 16.86}
{"step": 6440, "loss": 0.002139, "loss_mean50": 0.002483, "lr": 1.761063931261812e-05, "grad_norm": 0.0212, "samples": 206080, "wall_s": 12828.8, "cuda_gb": 16.86}
{"step": 6460, "loss": 0.004847, "loss_mean50": 0.002393, "lr": 1.7563176754739762e-05, "grad_norm": 0.036, "samples": 206720, "wall_s": 12867.8, "cuda_gb": 16.86}
{"step": 6480, "loss": 0.001529, "loss_mean50": 0.002227, "lr": 1.751563739179424e-05, "grad_norm": 0.0458, "samples": 207360, "wall_s": 12906.7, "cuda_gb": 16.86}
{"step": 6500, "loss": 0.00158, "loss_mean50": 0.002129, "lr": 1.7468022181320165e-05, "grad_norm": 0.0157, "samples": 208000, "wall_s": 12945.6, "cuda_gb": 16.86}
{"step": 6520, "loss": 0.003444, "loss_mean50": 0.002072, "lr": 1.742033208238386e-05, "grad_norm": 0.0556, "samples": 208640, "wall_s": 12984.4, "cuda_gb": 16.86}
{"step": 6540, "loss": 0.002893, "loss_mean50": 0.002152, "lr": 1.737256805556007e-05, "grad_norm": 0.025, "samples": 209280, "wall_s": 13023.2, "cuda_gb": 16.86}
{"step": 6560, "loss": 0.00252, "loss_mean50": 0.002409, "lr": 1.7324731062912574e-05, "grad_norm": 0.0215, "samples": 209920, "wall_s": 13062.1, "cuda_gb": 16.86}
{"step": 6580, "loss": 0.004378, "loss_mean50": 0.002602, "lr": 1.7276822067974855e-05, "grad_norm": 0.0586, "samples": 210560, "wall_s": 13100.9, "cuda_gb": 16.86}
{"step": 6600, "loss": 0.002644, "loss_mean50": 0.00247, "lr": 1.722884203573065e-05, "grad_norm": 0.0284, "samples": 211200, "wall_s": 13139.8, "cuda_gb": 16.86}
{"step": 6620, "loss": 0.001359, "loss_mean50": 0.002342, "lr": 1.7180791932594536e-05, "grad_norm": 0.026, "samples": 211840, "wall_s": 13178.5, "cuda_gb": 16.86}
{"step": 6640, "loss": 0.004117, "loss_mean50": 0.002268, "lr": 1.7132672726392453e-05, "grad_norm": 0.0384, "samples": 212480, "wall_s": 13217.4, "cuda_gb": 16.86}
{"step": 6660, "loss": 0.006358, "loss_mean50": 0.002268, "lr": 1.708448538634223e-05, "grad_norm": 0.1786, "samples": 213120, "wall_s": 13256.2, "cuda_gb": 16.86}
{"step": 6680, "loss": 0.003285, "loss_mean50": 0.002343, "lr": 1.703623088303404e-05, "grad_norm": 0.0329, "samples": 213760, "wall_s": 13294.9, "cuda_gb": 16.86}
{"step": 6700, "loss": 0.003113, "loss_mean50": 0.002393, "lr": 1.698791018841086e-05, "grad_norm": 0.027, "samples": 214400, "wall_s": 13333.8, "cuda_gb": 16.86}
{"step": 6720, "loss": 0.001296, "loss_mean50": 0.002392, "lr": 1.693952427574889e-05, "grad_norm": 0.0232, "samples": 215040, "wall_s": 13372.6, "cuda_gb": 16.86}
{"step": 6740, "loss": 0.001694, "loss_mean50": 0.002342, "lr": 1.6891074119637956e-05, "grad_norm": 0.0266, "samples": 215680, "wall_s": 13411.2, "cuda_gb": 16.86}
{"step": 6760, "loss": 0.001459, "loss_mean50": 0.002281, "lr": 1.6842560695961884e-05, "grad_norm": 0.0449, "samples": 216320, "wall_s": 13450.0, "cuda_gb": 16.86}
{"step": 6780, "loss": 0.002, "loss_mean50": 0.00215, "lr": 1.6793984981878822e-05, "grad_norm": 0.0465, "samples": 216960, "wall_s": 13488.7, "cuda_gb": 16.86}
{"step": 6800, "loss": 0.001134, "loss_mean50": 0.002397, "lr": 1.6745347955801585e-05, "grad_norm": 0.0167, "samples": 217600, "wall_s": 13527.5, "cuda_gb": 16.86}
{"step": 6820, "loss": 0.003088, "loss_mean50": 0.002463, "lr": 1.6696650597377923e-05, "grad_norm": 0.0336, "samples": 218240, "wall_s": 13566.2, "cuda_gb": 16.86}
{"step": 6840, "loss": 0.003243, "loss_mean50": 0.00247, "lr": 1.6647893887470816e-05, "grad_norm": 0.0314, "samples": 218880, "wall_s": 13605.0, "cuda_gb": 16.86}
{"step": 6860, "loss": 0.001658, "loss_mean50": 0.002469, "lr": 1.6599078808138693e-05, "grad_norm": 0.0192, "samples": 219520, "wall_s": 13643.7, "cuda_gb": 16.86}
{"step": 6880, "loss": 0.00107, "loss_mean50": 0.00241, "lr": 1.6550206342615658e-05, "grad_norm": 0.0229, "samples": 220160, "wall_s": 13682.7, "cuda_gb": 16.86}
{"step": 6900, "loss": 0.002073, "loss_mean50": 0.002335, "lr": 1.6501277475291696e-05, "grad_norm": 0.0201, "samples": 220800, "wall_s": 13721.5, "cuda_gb": 16.86}
{"step": 6920, "loss": 0.002906, "loss_mean50": 0.002248, "lr": 1.645229319169284e-05, "grad_norm": 0.0297, "samples": 221440, "wall_s": 13760.3, "cuda_gb": 16.86}
{"step": 6940, "loss": 0.003047, "loss_mean50": 0.002397, "lr": 1.64032544784613e-05, "grad_norm": 0.0277, "samples": 222080, "wall_s": 13799.1, "cuda_gb": 16.86}
{"step": 6960, "loss": 0.00133, "loss_mean50": 0.002383, "lr": 1.6354162323335633e-05, "grad_norm": 0.0234, "samples": 222720, "wall_s": 13837.8, "cuda_gb": 16.86}
{"step": 6980, "loss": 0.002951, "loss_mean50": 0.002209, "lr": 1.6305017715130812e-05, "grad_norm": 0.0252, "samples": 223360, "wall_s": 13876.6, "cuda_gb": 16.86}
{"step": 7000, "loss": 0.001849, "loss_mean50": 0.002303, "lr": 1.6255821643718313e-05, "grad_norm": 0.0151, "samples": 224000, "wall_s": 13915.5, "cuda_gb": 16.86}
{"step": 7000, "val_flow_mse": 0.002075}
{"step": 7020, "loss": 0.001378, "loss_mean50": 0.002222, "lr": 1.6206575100006205e-05, "grad_norm": 0.0272, "samples": 224640, "wall_s": 13983.6, "cuda_gb": 16.86}
{"step": 7040, "loss": 0.002416, "loss_mean50": 0.002191, "lr": 1.615727907591915e-05, "grad_norm": 0.0414, "samples": 225280, "wall_s": 14022.7, "cuda_gb": 16.86}
{"step": 7060, "loss": 0.001638, "loss_mean50": 0.00209, "lr": 1.610793456437846e-05, "grad_norm": 0.0374, "samples": 225920, "wall_s": 14061.7, "cuda_gb": 16.86}
{"step": 7080, "loss": 0.003333, "loss_mean50": 0.0023, "lr": 1.6058542559282066e-05, "grad_norm": 0.0458, "samples": 226560, "wall_s": 14100.8, "cuda_gb": 16.86}
{"step": 7100, "loss": 0.00126, "loss_mean50": 0.002237, "lr": 1.6009104055484537e-05, "grad_norm": 0.0143, "samples": 227200, "wall_s": 14140.1, "cuda_gb": 16.86}
{"step": 7120, "loss": 0.002819, "loss_mean50": 0.002172, "lr": 1.5959620048776996e-05, "grad_norm": 0.0511, "samples": 227840, "wall_s": 14179.4, "cuda_gb": 16.86}
{"step": 7140, "loss": 0.00178, "loss_mean50": 0.002064, "lr": 1.59100915358671e-05, "grad_norm": 0.0216, "samples": 228480, "wall_s": 14219.9, "cuda_gb": 16.86}
{"step": 7160, "loss": 0.003195, "loss_mean50": 0.002096, "lr": 1.586051951435895e-05, "grad_norm": 0.0498, "samples": 229120, "wall_s": 14259.1, "cuda_gb": 16.86}
{"step": 7180, "loss": 0.003677, "loss_mean50": 0.002279, "lr": 1.5810904982733e-05, "grad_norm": 0.0317, "samples": 229760, "wall_s": 14298.4, "cuda_gb": 16.86}
{"step": 7200, "loss": 0.002182, "loss_mean50": 0.002275, "lr": 1.576124894032593e-05, "grad_norm": 0.035, "samples": 230400, "wall_s": 14337.5, "cuda_gb": 16.86}
{"step": 7220, "loss": 0.002416, "loss_mean50": 0.0021, "lr": 1.5711552387310548e-05, "grad_norm": 0.0282, "samples": 231040, "wall_s": 14378.0, "cuda_gb": 16.86}
{"step": 7240, "loss": 0.001159, "loss_mean50": 0.002091, "lr": 1.5661816324675613e-05, "grad_norm": 0.0198, "samples": 231680, "wall_s": 14418.8, "cuda_gb": 16.86}
{"step": 7260, "loss": 0.001566, "loss_mean50": 0.002008, "lr": 1.56120417542057e-05, "grad_norm": 0.0243, "samples": 232320, "wall_s": 14458.5, "cuda_gb": 16.86}
{"step": 7280, "loss": 0.002285, "loss_mean50": 0.001905, "lr": 1.5562229678460998e-05, "grad_norm": 0.0371, "samples": 232960, "wall_s": 14497.2, "cuda_gb": 16.86}
{"step": 7300, "loss": 0.002597, "loss_mean50": 0.002065, "lr": 1.5512381100757136e-05, "grad_norm": 0.0714, "samples": 233600, "wall_s": 14536.1, "cuda_gb": 16.86}
{"step": 7320, "loss": 0.000952, "loss_mean50": 0.00223, "lr": 1.5462497025144967e-05, "grad_norm": 0.0221, "samples": 234240, "wall_s": 14575.0, "cuda_gb": 16.86}
{"step": 7340, "loss": 0.001918, "loss_mean50": 0.002226, "lr": 1.541257845639033e-05, "grad_norm": 0.0266, "samples": 234880, "wall_s": 14614.0, "cuda_gb": 16.86}
{"step": 7360, "loss": 0.002371, "loss_mean50": 0.002044, "lr": 1.536262639995384e-05, "grad_norm": 0.0219, "samples": 235520, "wall_s": 14652.8, "cuda_gb": 16.86}
{"step": 7380, "loss": 0.004627, "loss_mean50": 0.002228, "lr": 1.531264186197062e-05, "grad_norm": 0.0791, "samples": 236160, "wall_s": 14691.8, "cuda_gb": 16.86}
{"step": 7400, "loss": 0.002541, "loss_mean50": 0.002349, "lr": 1.526262584923003e-05, "grad_norm": 0.029, "samples": 236800, "wall_s": 14730.7, "cuda_gb": 16.86}
{"step": 7420, "loss": 0.001295, "loss_mean50": 0.00242, "lr": 1.5212579369155395e-05, "grad_norm": 0.0296, "samples": 237440, "wall_s": 14769.7, "cuda_gb": 16.86}
{"step": 7440, "loss": 0.003913, "loss_mean50": 0.002453, "lr": 1.5162503429783722e-05, "grad_norm": 0.0468, "samples": 238080, "wall_s": 14808.7, "cuda_gb": 16.86}
{"step": 7460, "loss": 0.002061, "loss_mean50": 0.002437, "lr": 1.5112399039745381e-05, "grad_norm": 0.0326, "samples": 238720, "wall_s": 14848.8, "cuda_gb": 16.86}
{"step": 7480, "loss": 0.001361, "loss_mean50": 0.002396, "lr": 1.5062267208243797e-05, "grad_norm": 0.0205, "samples": 239360, "wall_s": 14892.0, "cuda_gb": 16.86}
{"step": 7500, "loss": 0.003203, "loss_mean50": 0.002252, "lr": 1.5012108945035122e-05, "grad_norm": 0.0307, "samples": 240000, "wall_s": 14931.0, "cuda_gb": 16.86}
{"step": 7520, "loss": 0.00225, "loss_mean50": 0.002155, "lr": 1.49619252604079e-05, "grad_norm": 0.0359, "samples": 240640, "wall_s": 15003.8, "cuda_gb": 16.86}
{"step": 7540, "loss": 0.001383, "loss_mean50": 0.002197, "lr": 1.49117171651627e-05, "grad_norm": 0.0174, "samples": 241280, "wall_s": 15042.6, "cuda_gb": 16.86}
{"step": 7560, "loss": 0.002827, "loss_mean50": 0.002085, "lr": 1.4861485670591782e-05, "grad_norm": 0.0692, "samples": 241920, "wall_s": 15081.4, "cuda_gb": 16.86}
{"step": 7580, "loss": 0.0026, "loss_mean50": 0.002054, "lr": 1.4811231788458716e-05, "grad_norm": 0.0554, "samples": 242560, "wall_s": 15120.4, "cuda_gb": 16.86}
{"step": 7600, "loss": 0.000876, "loss_mean50": 0.002059, "lr": 1.4760956530978001e-05, "grad_norm": 0.016, "samples": 243200, "wall_s": 15159.2, "cuda_gb": 16.86}
{"step": 7620, "loss": 0.001315, "loss_mean50": 0.00198, "lr": 1.471066091079467e-05, "grad_norm": 0.0337, "samples": 243840, "wall_s": 15198.0, "cuda_gb": 16.86}
{"step": 7640, "loss": 0.001767, "loss_mean50": 0.00207, "lr": 1.4660345940963918e-05, "grad_norm": 0.0279, "samples": 244480, "wall_s": 15236.7, "cuda_gb": 16.86}
{"step": 7660, "loss": 0.001165, "loss_mean50": 0.002115, "lr": 1.4610012634930665e-05, "grad_norm": 0.0257, "samples": 245120, "wall_s": 15275.5, "cuda_gb": 16.86}
{"step": 7680, "loss": 0.00151, "loss_mean50": 0.002278, "lr": 1.4559662006509167e-05, "grad_norm": 0.0264, "samples": 245760, "wall_s": 15314.3, "cuda_gb": 16.86}
{"step": 7700, "loss": 0.002214, "loss_mean50": 0.002269, "lr": 1.4509295069862588e-05, "grad_norm": 0.0959, "samples": 246400, "wall_s": 15353.0, "cuda_gb": 16.86}
{"step": 7720, "loss": 0.002355, "loss_mean50": 0.002267, "lr": 1.4458912839482571e-05, "grad_norm": 0.0368, "samples": 247040, "wall_s": 15391.7, "cuda_gb": 16.86}
{"step": 7740, "loss": 0.004036, "loss_mean50": 0.002116, "lr": 1.4408516330168806e-05, "grad_norm": 0.0784, "samples": 247680, "wall_s": 15430.5, "cuda_gb": 16.86}
{"step": 7760, "loss": 0.001447, "loss_mean50": 0.002158, "lr": 1.4358106557008589e-05, "grad_norm": 0.0252, "samples": 248320, "wall_s": 15469.4, "cuda_gb": 16.86}
{"step": 7780, "loss": 0.002433, "loss_mean50": 0.002272, "lr": 1.4307684535356378e-05, "grad_norm": 0.0253, "samples": 248960, "wall_s": 15508.3, "cuda_gb": 16.86}
{"step": 7800, "loss": 0.001118, "loss_mean50": 0.002096, "lr": 1.4257251280813339e-05, "grad_norm": 0.0197, "samples": 249600, "wall_s": 15547.1, "cuda_gb": 16.86}
{"step": 7820, "loss": 0.00221, "loss_mean50": 0.002157, "lr": 1.420680780920689e-05, "grad_norm": 0.0206, "samples": 250240, "wall_s": 15586.0, "cuda_gb": 16.86}
{"step": 7840, "loss": 0.001945, "loss_mean50": 0.002144, "lr": 1.4156355136570243e-05, "grad_norm": 0.0232, "samples": 250880, "wall_s": 15624.9, "cuda_gb": 16.86}
{"step": 7860, "loss": 0.001882, "loss_mean50": 0.001962, "lr": 1.4105894279121939e-05, "grad_norm": 0.0352, "samples": 251520, "wall_s": 15663.7, "cuda_gb": 16.86}
{"step": 7880, "loss": 0.001086, "loss_mean50": 0.001858, "lr": 1.4055426253245374e-05, "grad_norm": 0.0208, "samples": 252160, "wall_s": 15702.6, "cuda_gb": 16.86}
{"step": 7900, "loss": 0.001803, "loss_mean50": 0.001937, "lr": 1.4004952075468333e-05, "grad_norm": 0.0234, "samples": 252800, "wall_s": 15741.3, "cuda_gb": 16.86}
{"step": 7920, "loss": 0.00167, "loss_mean50": 0.002084, "lr": 1.3954472762442508e-05, "grad_norm": 0.0365, "samples": 253440, "wall_s": 15780.1, "cuda_gb": 16.86}
{"step": 7940, "loss": 0.003828, "loss_mean50": 0.00226, "lr": 1.3903989330923037e-05, "grad_norm": 0.0754, "samples": 254080, "wall_s": 15818.9, "cuda_gb": 16.86}
{"step": 7960, "loss": 0.000884, "loss_mean50": 0.002125, "lr": 1.3853502797747997e-05, "grad_norm": 0.0191, "samples": 254720, "wall_s": 15857.8, "cuda_gb": 16.86}
{"step": 7980, "loss": 0.002224, "loss_mean50": 0.002151, "lr": 1.3803014179817949e-05, "grad_norm": 0.0221, "samples": 255360, "wall_s": 15896.6, "cuda_gb": 16.86}
{"step": 8000, "loss": 0.001454, "loss_mean50": 0.002105, "lr": 1.375252449407545e-05, "grad_norm": 0.0268, "samples": 256000, "wall_s": 15935.5, "cuda_gb": 16.86}
{"step": 8000, "val_flow_mse": 0.001931}
{"step": 8020, "loss": 0.001422, "loss_mean50": 0.002133, "lr": 1.3702034757484548e-05, "grad_norm": 0.0299, "samples": 256640, "wall_s": 16003.2, "cuda_gb": 16.86}
{"step": 8040, "loss": 0.001009, "loss_mean50": 0.002302, "lr": 1.3651545987010325e-05, "grad_norm": 0.018, "samples": 257280, "wall_s": 16042.0, "cuda_gb": 16.86}
{"step": 8060, "loss": 0.001866, "loss_mean50": 0.002105, "lr": 1.3601059199598407e-05, "grad_norm": 0.0287, "samples": 257920, "wall_s": 16080.8, "cuda_gb": 16.86}
{"step": 8080, "loss": 0.002818, "loss_mean50": 0.001992, "lr": 1.355057541215447e-05, "grad_norm": 0.0459, "samples": 258560, "wall_s": 16119.8, "cuda_gb": 16.86}
{"step": 8100, "loss": 0.005084, "loss_mean50": 0.002055, "lr": 1.3500095641523772e-05, "grad_norm": 0.0569, "samples": 259200, "wall_s": 16158.6, "cuda_gb": 16.86}
{"step": 8120, "loss": 0.001558, "loss_mean50": 0.002075, "lr": 1.3449620904470664e-05, "grad_norm": 0.0215, "samples": 259840, "wall_s": 16197.4, "cuda_gb": 16.86}
{"step": 8140, "loss": 0.001491, "loss_mean50": 0.002048, "lr": 1.3399152217658095e-05, "grad_norm": 0.0209, "samples": 260480, "wall_s": 16236.2, "cuda_gb": 16.86}
{"step": 8160, "loss": 0.004726, "loss_mean50": 0.002147, "lr": 1.3348690597627166e-05, "grad_norm": 0.0379, "samples": 261120, "wall_s": 16275.0, "cuda_gb": 16.86}
{"step": 8180, "loss": 0.002523, "loss_mean50": 0.002162, "lr": 1.3298237060776639e-05, "grad_norm": 0.0372, "samples": 261760, "wall_s": 16313.8, "cuda_gb": 16.86}
{"step": 8200, "loss": 0.00182, "loss_mean50": 0.002181, "lr": 1.324779262334246e-05, "grad_norm": 0.0189, "samples": 262400, "wall_s": 16352.6, "cuda_gb": 16.86}
{"step": 8220, "loss": 0.002673, "loss_mean50": 0.002204, "lr": 1.3197358301377289e-05, "grad_norm": 0.0655, "samples": 263040, "wall_s": 16391.5, "cuda_gb": 16.86}
{"step": 8240, "loss": 0.004296, "loss_mean50": 0.002133, "lr": 1.3146935110730046e-05, "grad_norm": 0.1179, "samples": 263680, "wall_s": 16430.4, "cuda_gb": 16.86}
{"step": 8260, "loss": 0.002016, "loss_mean50": 0.002225, "lr": 1.3096524067025451e-05, "grad_norm": 0.0347, "samples": 264320, "wall_s": 16469.3, "cuda_gb": 16.86}
{"step": 8280, "loss": 0.003131, "loss_mean50": 0.002359, "lr": 1.3046126185643545e-05, "grad_norm": 0.0271, "samples": 264960, "wall_s": 16508.1, "cuda_gb": 16.86}
{"step": 8300, "loss": 0.000557, "loss_mean50": 0.002176, "lr": 1.2995742481699261e-05, "grad_norm": 0.0141, "samples": 265600, "wall_s": 16546.9, "cuda_gb": 16.86}
{"step": 8320, "loss": 0.002376, "loss_mean50": 0.002194, "lr": 1.2945373970021971e-05, "grad_norm": 0.0419, "samples": 266240, "wall_s": 16585.7, "cuda_gb": 16.86}
{"step": 8340, "loss": 0.001162, "loss_mean50": 0.002108, "lr": 1.2895021665135043e-05, "grad_norm": 0.0208, "samples": 266880, "wall_s": 16624.5, "cuda_gb": 16.86}
{"step": 8360, "loss": 0.001333, "loss_mean50": 0.002123, "lr": 1.2844686581235405e-05, "grad_norm": 0.0149, "samples": 267520, "wall_s": 16663.3, "cuda_gb": 16.86}
{"step": 8380, "loss": 0.002557, "loss_mean50": 0.001958, "lr": 1.2794369732173127e-05, "grad_norm": 0.0418, "samples": 268160, "wall_s": 16702.1, "cuda_gb": 16.86}
{"step": 8400, "loss": 0.001838, "loss_mean50": 0.001938, "lr": 1.2744072131430981e-05, "grad_norm": 0.0491, "samples": 268800, "wall_s": 16740.9, "cuda_gb": 16.86}
{"step": 8420, "loss": 0.001973, "loss_mean50": 0.001942, "lr": 1.2693794792104049e-05, "grad_norm": 0.0332, "samples": 269440, "wall_s": 16779.7, "cuda_gb": 16.86}
{"step": 8440, "loss": 0.00084, "loss_mean50": 0.002084, "lr": 1.2643538726879303e-05, "grad_norm": 0.014, "samples": 270080, "wall_s": 16818.6, "cuda_gb": 16.86}
{"step": 8460, "loss": 0.003639, "loss_mean50": 0.002051, "lr": 1.2593304948015212e-05, "grad_norm": 0.0311, "samples": 270720, "wall_s": 16857.4, "cuda_gb": 16.86}
{"step": 8480, "loss": 0.001653, "loss_mean50": 0.001837, "lr": 1.254309446732135e-05, "grad_norm": 0.0258, "samples": 271360, "wall_s": 16896.2, "cuda_gb": 16.86}
{"step": 8500, "loss": 0.001758, "loss_mean50": 0.001961, "lr": 1.2492908296138028e-05, "grad_norm": 0.0202, "samples": 272000, "wall_s": 16935.0, "cuda_gb": 16.86}
{"step": 8520, "loss": 0.002719, "loss_mean50": 0.002052, "lr": 1.2442747445315909e-05, "grad_norm": 0.0361, "samples": 272640, "wall_s": 16973.9, "cuda_gb": 16.86}
{"step": 8540, "loss": 0.001083, "loss_mean50": 0.002088, "lr": 1.2392612925195645e-05, "grad_norm": 0.0332, "samples": 273280, "wall_s": 17012.7, "cuda_gb": 16.86}
{"step": 8560, "loss": 0.001777, "loss_mean50": 0.002116, "lr": 1.2342505745587546e-05, "grad_norm": 0.0256, "samples": 273920, "wall_s": 17051.6, "cuda_gb": 16.86}
{"step": 8580, "loss": 0.00233, "loss_mean50": 0.001977, "lr": 1.2292426915751223e-05, "grad_norm": 0.0246, "samples": 274560, "wall_s": 17090.4, "cuda_gb": 16.86}
{"step": 8600, "loss": 0.002967, "loss_mean50": 0.001904, "lr": 1.2242377444375272e-05, "grad_norm": 0.0297, "samples": 275200, "wall_s": 17129.3, "cuda_gb": 16.86}
{"step": 8620, "loss": 0.002033, "loss_mean50": 0.001906, "lr": 1.2192358339556945e-05, "grad_norm": 0.0337, "samples": 275840, "wall_s": 17168.1, "cuda_gb": 16.86}
{"step": 8640, "loss": 0.002123, "loss_mean50": 0.001964, "lr": 1.2142370608781848e-05, "grad_norm": 0.0361, "samples": 276480, "wall_s": 17207.0, "cuda_gb": 16.86}
{"step": 8660, "loss": 0.0016, "loss_mean50": 0.001855, "lr": 1.209241525890365e-05, "grad_norm": 0.0331, "samples": 277120, "wall_s": 17246.0, "cuda_gb": 16.86}
{"step": 8680, "loss": 0.002611, "loss_mean50": 0.00193, "lr": 1.204249329612381e-05, "grad_norm": 0.0312, "samples": 277760, "wall_s": 17284.8, "cuda_gb": 16.86}
{"step": 8700, "loss": 0.001334, "loss_mean50": 0.001871, "lr": 1.1992605725971308e-05, "grad_norm": 0.0195, "samples": 278400, "wall_s": 17323.6, "cuda_gb": 16.86}
{"step": 8720, "loss": 0.002091, "loss_mean50": 0.002018, "lr": 1.194275355328237e-05, "grad_norm": 0.0243, "samples": 279040, "wall_s": 17362.4, "cuda_gb": 16.86}
{"step": 8740, "loss": 0.002506, "loss_mean50": 0.002063, "lr": 1.1892937782180261e-05, "grad_norm": 0.0371, "samples": 279680, "wall_s": 17401.3, "cuda_gb": 16.86}
{"step": 8760, "loss": 0.00368, "loss_mean50": 0.002049, "lr": 1.1843159416055037e-05, "grad_norm": 0.0401, "samples": 280320, "wall_s": 17440.3, "cuda_gb": 16.86}
{"step": 8780, "loss": 0.003044, "loss_mean50": 0.002033, "lr": 1.1793419457543346e-05, "grad_norm": 0.0297, "samples": 280960, "wall_s": 17479.2, "cuda_gb": 16.86}
{"step": 8800, "loss": 0.002793, "loss_mean50": 0.002099, "lr": 1.1743718908508222e-05, "grad_norm": 0.047, "samples": 281600, "wall_s": 17518.2, "cuda_gb": 16.86}
{"step": 8820, "loss": 0.001343, "loss_mean50": 0.002137, "lr": 1.1694058770018925e-05, "grad_norm": 0.0232, "samples": 282240, "wall_s": 17557.0, "cuda_gb": 16.86}
{"step": 8840, "loss": 0.001959, "loss_mean50": 0.002221, "lr": 1.1644440042330754e-05, "grad_norm": 0.0234, "samples": 282880, "wall_s": 17595.8, "cuda_gb": 16.86}
{"step": 8860, "loss": 0.001443, "loss_mean50": 0.002113, "lr": 1.1594863724864918e-05, "grad_norm": 0.0229, "samples": 283520, "wall_s": 17634.6, "cuda_gb": 16.86}
{"step": 8880, "loss": 0.001505, "loss_mean50": 0.002085, "lr": 1.1545330816188398e-05, "grad_norm": 0.0187, "samples": 284160, "wall_s": 17673.4, "cuda_gb": 16.86}
{"step": 8900, "loss": 0.001976, "loss_mean50": 0.001945, "lr": 1.1495842313993832e-05, "grad_norm": 0.0398, "samples": 284800, "wall_s": 17712.2, "cuda_gb": 16.86}
{"step": 8920, "loss": 0.001893, "loss_mean50": 0.002006, "lr": 1.1446399215079417e-05, "grad_norm": 0.0253, "samples": 285440, "wall_s": 17751.0, "cuda_gb": 16.86}
{"step": 8940, "loss": 0.00139, "loss_mean50": 0.001967, "lr": 1.1397002515328845e-05, "grad_norm": 0.0603, "samples": 286080, "wall_s": 17789.7, "cuda_gb": 16.86}
{"step": 8960, "loss": 0.001389, "loss_mean50": 0.002059, "lr": 1.1347653209691237e-05, "grad_norm": 0.0331, "samples": 286720, "wall_s": 17828.5, "cuda_gb": 16.86}
{"step": 8980, "loss": 0.006349, "loss_mean50": 0.002192, "lr": 1.1298352292161085e-05, "grad_norm": 0.0333, "samples": 287360, "wall_s": 17867.3, "cuda_gb": 16.86}
{"step": 9000, "loss": 0.001557, "loss_mean50": 0.002167, "lr": 1.1249100755758268e-05, "grad_norm": 0.0214, "samples": 288000, "wall_s": 17906.1, "cuda_gb": 16.86}
{"step": 9000, "val_flow_mse": 0.001988}
{"step": 9020, "loss": 0.001292, "loss_mean50": 0.002087, "lr": 1.1199899592508024e-05, "grad_norm": 0.0144, "samples": 288640, "wall_s": 17973.6, "cuda_gb": 16.86}
{"step": 9040, "loss": 0.002019, "loss_mean50": 0.001963, "lr": 1.1150749793420958e-05, "grad_norm": 0.0294, "samples": 289280, "wall_s": 18012.5, "cuda_gb": 16.86}
{"step": 9060, "loss": 0.000829, "loss_mean50": 0.001994, "lr": 1.1101652348473113e-05, "grad_norm": 0.0241, "samples": 289920, "wall_s": 18051.4, "cuda_gb": 16.86}
{"step": 9080, "loss": 0.001696, "loss_mean50": 0.001965, "lr": 1.1052608246586007e-05, "grad_norm": 0.0292, "samples": 290560, "wall_s": 18090.2, "cuda_gb": 16.86}
{"step": 9100, "loss": 0.002544, "loss_mean50": 0.002069, "lr": 1.1003618475606737e-05, "grad_norm": 0.0246, "samples": 291200, "wall_s": 18129.1, "cuda_gb": 16.86}
{"step": 9120, "loss": 0.001347, "loss_mean50": 0.001942, "lr": 1.0954684022288039e-05, "grad_norm": 0.0246, "samples": 291840, "wall_s": 18167.9, "cuda_gb": 16.86}
{"step": 9140, "loss": 0.003761, "loss_mean50": 0.001912, "lr": 1.0905805872268463e-05, "grad_norm": 0.0514, "samples": 292480, "wall_s": 18206.8, "cuda_gb": 16.86}
{"step": 9160, "loss": 0.003069, "loss_mean50": 0.001975, "lr": 1.0856985010052487e-05, "grad_norm": 0.0232, "samples": 293120, "wall_s": 18245.8, "cuda_gb": 16.86}
{"step": 9180, "loss": 0.001952, "loss_mean50": 0.002026, "lr": 1.0808222418990693e-05, "grad_norm": 0.0321, "samples": 293760, "wall_s": 18284.6, "cuda_gb": 16.86}
{"step": 9200, "loss": 0.001906, "loss_mean50": 0.001918, "lr": 1.0759519081259975e-05, "grad_norm": 0.0189, "samples": 294400, "wall_s": 18323.6, "cuda_gb": 16.86}
{"step": 9220, "loss": 0.003069, "loss_mean50": 0.001767, "lr": 1.071087597784374e-05, "grad_norm": 0.034, "samples": 295040, "wall_s": 18362.4, "cuda_gb": 16.86}
{"step": 9240, "loss": 0.001364, "loss_mean50": 0.001874, "lr": 1.0662294088512154e-05, "grad_norm": 0.0264, "samples": 295680, "wall_s": 18401.3, "cuda_gb": 16.86}
{"step": 9260, "loss": 0.002274, "loss_mean50": 0.001944, "lr": 1.0613774391802405e-05, "grad_norm": 0.0364, "samples": 296320, "wall_s": 18440.2, "cuda_gb": 16.86}
{"step": 9280, "loss": 0.003842, "loss_mean50": 0.001929, "lr": 1.0565317864999011e-05, "grad_norm": 0.0621, "samples": 296960, "wall_s": 18479.1, "cuda_gb": 16.86}
{"step": 9300, "loss": 0.000934, "loss_mean50": 0.001839, "lr": 1.0516925484114096e-05, "grad_norm": 0.0223, "samples": 297600, "wall_s": 18517.9, "cuda_gb": 16.86}
{"step": 9320, "loss": 0.002822, "loss_mean50": 0.001897, "lr": 1.0468598223867779e-05, "grad_norm": 0.0309, "samples": 298240, "wall_s": 18556.7, "cuda_gb": 16.86}
{"step": 9340, "loss": 0.001726, "loss_mean50": 0.001881, "lr": 1.04203370576685e-05, "grad_norm": 0.0397, "samples": 298880, "wall_s": 18595.6, "cuda_gb": 16.86}
{"step": 9360, "loss": 0.001799, "loss_mean50": 0.001776, "lr": 1.037214295759345e-05, "grad_norm": 0.0213, "samples": 299520, "wall_s": 18634.4, "cuda_gb": 16.86}
{"step": 9380, "loss": 0.001547, "loss_mean50": 0.001656, "lr": 1.0324016894368957e-05, "grad_norm": 0.0304, "samples": 300160, "wall_s": 18673.2, "cuda_gb": 16.86}
{"step": 9400, "loss": 0.001599, "loss_mean50": 0.001888, "lr": 1.0275959837350957e-05, "grad_norm": 0.0186, "samples": 300800, "wall_s": 18712.1, "cuda_gb": 16.86}
{"step": 9420, "loss": 0.001107, "loss_mean50": 0.001895, "lr": 1.0227972754505468e-05, "grad_norm": 0.0215, "samples": 301440, "wall_s": 18750.9, "cuda_gb": 16.86}
{"step": 9440, "loss": 0.001099, "loss_mean50": 0.00175, "lr": 1.0180056612389072e-05, "grad_norm": 0.0226, "samples": 302080, "wall_s": 18789.7, "cuda_gb": 16.86}
{"step": 9460, "loss": 0.001745, "loss_mean50": 0.001792, "lr": 1.0132212376129465e-05, "grad_norm": 0.0302, "samples": 302720, "wall_s": 18828.5, "cuda_gb": 16.86}
{"step": 9480, "loss": 0.0011, "loss_mean50": 0.002002, "lr": 1.0084441009406028e-05, "grad_norm": 0.0175, "samples": 303360, "wall_s": 18867.3, "cuda_gb": 16.86}
{"step": 9500, "loss": 0.001257, "loss_mean50": 0.002123, "lr": 1.0036743474430393e-05, "grad_norm": 0.0191, "samples": 304000, "wall_s": 18906.1, "cuda_gb": 16.86}
{"step": 9520, "loss": 0.002697, "loss_mean50": 0.002084, "lr": 9.989120731927065e-06, "grad_norm": 0.0377, "samples": 304640, "wall_s": 18945.0, "cuda_gb": 16.86}
{"step": 9540, "loss": 0.002078, "loss_mean50": 0.001868, "lr": 9.941573741114099e-06, "grad_norm": 0.0265, "samples": 305280, "wall_s": 18983.9, "cuda_gb": 16.86}
{"step": 9560, "loss": 0.001976, "loss_mean50": 0.001759, "lr": 9.894103459683726e-06, "grad_norm": 0.0298, "samples": 305920, "wall_s": 19022.7, "cuda_gb": 16.86}
{"step": 9580, "loss": 0.00282, "loss_mean50": 0.001781, "lr": 9.846710843783125e-06, "grad_norm": 0.0361, "samples": 306560, "wall_s": 19061.5, "cuda_gb": 16.86}
{"step": 9600, "loss": 0.001471, "loss_mean50": 0.002038, "lr": 9.799396847995121e-06, "grad_norm": 0.025, "samples": 307200, "wall_s": 19100.3, "cuda_gb": 16.86}
{"step": 9620, "loss": 0.001285, "loss_mean50": 0.002101, "lr": 9.752162425318972e-06, "grad_norm": 0.0159, "samples": 307840, "wall_s": 19139.1, "cuda_gb": 16.86}
{"step": 9640, "loss": 0.001333, "loss_mean50": 0.001869, "lr": 9.705008527151176e-06, "grad_norm": 0.0256, "samples": 308480, "wall_s": 19178.0, "cuda_gb": 16.86}
{"step": 9660, "loss": 0.001847, "loss_mean50": 0.001796, "lr": 9.657936103266308e-06, "grad_norm": 0.019, "samples": 309120, "wall_s": 19216.9, "cuda_gb": 16.86}
{"step": 9680, "loss": 0.00223, "loss_mean50": 0.001758, "lr": 9.61094610179788e-06, "grad_norm": 0.0412, "samples": 309760, "wall_s": 19255.9, "cuda_gb": 16.86}
{"step": 9700, "loss": 0.001978, "loss_mean50": 0.001749, "lr": 9.564039469219244e-06, "grad_norm": 0.0341, "samples": 310400, "wall_s": 19294.9, "cuda_gb": 16.86}
{"step": 9720, "loss": 0.001939, "loss_mean50": 0.001902, "lr": 9.517217150324556e-06, "grad_norm": 0.0263, "samples": 311040, "wall_s": 19333.9, "cuda_gb": 16.86}
{"step": 9740, "loss": 0.001671, "loss_mean50": 0.001893, "lr": 9.470480088209707e-06, "grad_norm": 0.0347, "samples": 311680, "wall_s": 19372.8, "cuda_gb": 16.86}
{"step": 9760, "loss": 0.00168, "loss_mean50": 0.001905, "lr": 9.423829224253347e-06, "grad_norm": 0.0183, "samples": 312320, "wall_s": 19411.7, "cuda_gb": 16.86}
{"step": 9780, "loss": 0.002552, "loss_mean50": 0.001771, "lr": 9.377265498097926e-06, "grad_norm": 0.0316, "samples": 312960, "wall_s": 19450.6, "cuda_gb": 16.86}
{"step": 9800, "loss": 0.004426, "loss_mean50": 0.001893, "lr": 9.330789847630767e-06, "grad_norm": 0.0464, "samples": 313600, "wall_s": 19489.4, "cuda_gb": 16.86}
{"step": 9820, "loss": 0.002243, "loss_mean50": 0.002005, "lr": 9.284403208965151e-06, "grad_norm": 0.0319, "samples": 314240, "wall_s": 19528.2, "cuda_gb": 16.86}
{"step": 9840, "loss": 0.000946, "loss_mean50": 0.002138, "lr": 9.238106516421504e-06, "grad_norm": 0.0204, "samples": 314880, "wall_s": 19566.9, "cuda_gb": 16.86}
{"step": 9860, "loss": 0.001632, "loss_mean50": 0.001905, "lr": 9.191900702508551e-06, "grad_norm": 0.0245, "samples": 315520, "wall_s": 19605.7, "cuda_gb": 16.86}
{"step": 9880, "loss": 0.001529, "loss_mean50": 0.001874, "lr": 9.145786697904526e-06, "grad_norm": 0.0286, "samples": 316160, "wall_s": 19644.5, "cuda_gb": 16.86}
{"step": 9900, "loss": 0.000841, "loss_mean50": 0.001779, "lr": 9.099765431438454e-06, "grad_norm": 0.0182, "samples": 316800, "wall_s": 19683.3, "cuda_gb": 16.86}
{"step": 9920, "loss": 0.002295, "loss_mean50": 0.001826, "lr": 9.053837830071418e-06, "grad_norm": 0.0225, "samples": 317440, "wall_s": 19722.2, "cuda_gb": 16.86}
{"step": 9940, "loss": 0.001912, "loss_mean50": 0.001864, "lr": 9.008004818877899e-06, "grad_norm": 0.0224, "samples": 318080, "wall_s": 19761.0, "cuda_gb": 16.86}
{"step": 9960, "loss": 0.002968, "loss_mean50": 0.001972, "lr": 8.96226732102714e-06, "grad_norm": 0.0312, "samples": 318720, "wall_s": 19799.9, "cuda_gb": 16.86}
{"step": 9980, "loss": 0.001535, "loss_mean50": 0.001829, "lr": 8.91662625776456e-06, "grad_norm": 0.026, "samples": 319360, "wall_s": 19838.7, "cuda_gb": 16.86}
{"step": 10000, "loss": 0.002237, "loss_mean50": 0.001723, "lr": 8.87108254839317e-06, "grad_norm": 0.046, "samples": 320000, "wall_s": 19877.6, "cuda_gb": 16.86}
{"step": 10000, "val_flow_mse": 0.001804, "val_action_mae": 0.014031, "best": true}
{"step": 10020, "loss": 0.001025, "loss_mean50": 0.001663, "lr": 8.825637110255108e-06, "grad_norm": 0.0387, "samples": 320640, "wall_s": 19996.3, "cuda_gb": 16.86}
{"step": 10040, "loss": 0.001415, "loss_mean50": 0.00184, "lr": 8.78029085871311e-06, "grad_norm": 0.0301, "samples": 321280, "wall_s": 20035.2, "cuda_gb": 16.86}
{"step": 10060, "loss": 0.000592, "loss_mean50": 0.001783, "lr": 8.7350447071321e-06, "grad_norm": 0.0181, "samples": 321920, "wall_s": 20074.0, "cuda_gb": 16.86}
{"step": 10080, "loss": 0.00131, "loss_mean50": 0.001728, "lr": 8.68989956686079e-06, "grad_norm": 0.0258, "samples": 322560, "wall_s": 20112.8, "cuda_gb": 16.86}
{"step": 10100, "loss": 0.001634, "loss_mean50": 0.001838, "lr": 8.644856347213308e-06, "grad_norm": 0.0204, "samples": 323200, "wall_s": 20151.5, "cuda_gb": 16.86}
{"step": 10120, "loss": 0.00164, "loss_mean50": 0.001946, "lr": 8.59991595545091e-06, "grad_norm": 0.0208, "samples": 323840, "wall_s": 20190.4, "cuda_gb": 16.86}
{"step": 10140, "loss": 0.002019, "loss_mean50": 0.001773, "lr": 8.555079296763687e-06, "grad_norm": 0.0276, "samples": 324480, "wall_s": 20229.2, "cuda_gb": 16.86}
{"step": 10160, "loss": 0.001405, "loss_mean50": 0.001505, "lr": 8.510347274252333e-06, "grad_norm": 0.0213, "samples": 325120, "wall_s": 20268.0, "cuda_gb": 16.86}
{"step": 10180, "loss": 0.002601, "loss_mean50": 0.001628, "lr": 8.465720788909959e-06, "grad_norm": 0.047, "samples": 325760, "wall_s": 20306.7, "cuda_gb": 16.86}
{"step": 10200, "loss": 0.003239, "loss_mean50": 0.001802, "lr": 8.421200739603955e-06, "grad_norm": 0.0593, "samples": 326400, "wall_s": 20345.5, "cuda_gb": 16.86}
{"step": 10220, "loss": 0.002822, "loss_mean50": 0.00201, "lr": 8.376788023057854e-06, "grad_norm": 0.0188, "samples": 327040, "wall_s": 20384.2, "cuda_gb": 16.86}
{"step": 10240, "loss": 0.003968, "loss_mean50": 0.001983, "lr": 8.33248353383331e-06, "grad_norm": 0.0591, "samples": 327680, "wall_s": 20423.0, "cuda_gb": 16.86}
{"step": 10260, "loss": 0.001797, "loss_mean50": 0.001893, "lr": 8.288288164312053e-06, "grad_norm": 0.026, "samples": 328320, "wall_s": 20461.8, "cuda_gb": 16.86}
{"step": 10280, "loss": 0.002466, "loss_mean50": 0.00169, "lr": 8.244202804677917e-06, "grad_norm": 0.0262, "samples": 328960, "wall_s": 20500.8, "cuda_gb": 16.86}
{"step": 10300, "loss": 0.000995, "loss_mean50": 0.00177, "lr": 8.200228342898927e-06, "grad_norm": 0.0218, "samples": 329600, "wall_s": 20539.8, "cuda_gb": 16.86}
{"step": 10320, "loss": 0.001229, "loss_mean50": 0.001829, "lr": 8.1563656647094e-06, "grad_norm": 0.0192, "samples": 330240, "wall_s": 20578.7, "cuda_gb": 16.86}
{"step": 10340, "loss": 0.003228, "loss_mean50": 0.00182, "lr": 8.112615653592086e-06, "grad_norm": 0.0389, "samples": 330880, "wall_s": 20617.4, "cuda_gb": 16.86}
{"step": 10360, "loss": 0.002405, "loss_mean50": 0.001924, "lr": 8.06897919076042e-06, "grad_norm": 0.0191, "samples": 331520, "wall_s": 20656.2, "cuda_gb": 16.86}
{"step": 10380, "loss": 0.001681, "loss_mean50": 0.001795, "lr": 8.025457155140734e-06, "grad_norm": 0.032, "samples": 332160, "wall_s": 20695.0, "cuda_gb": 16.86}
{"step": 10400, "loss": 0.001233, "loss_mean50": 0.001788, "lr": 7.982050423354554e-06, "grad_norm": 0.035, "samples": 332800, "wall_s": 20733.8, "cuda_gb": 16.86}
{"step": 10420, "loss": 0.001842, "loss_mean50": 0.001686, "lr": 7.938759869700983e-06, "grad_norm": 0.0298, "samples": 333440, "wall_s": 20772.6, "cuda_gb": 16.86}
{"step": 10440, "loss": 0.001646, "loss_mean50": 0.001669, "lr": 7.895586366139032e-06, "grad_norm": 0.036, "samples": 334080, "wall_s": 20811.4, "cuda_gb": 16.86}
{"step": 10460, "loss": 0.002367, "loss_mean50": 0.001643, "lr": 7.852530782270106e-06, "grad_norm": 0.0382, "samples": 334720, "wall_s": 20850.2, "cuda_gb": 16.86}
{"step": 10480, "loss": 0.001604, "loss_mean50": 0.001732, "lr": 7.809593985320462e-06, "grad_norm": 0.0433, "samples": 335360, "wall_s": 20889.1, "cuda_gb": 16.86}
{"step": 10500, "loss": 0.001474, "loss_mean50": 0.001563, "lr": 7.766776840123753e-06, "grad_norm": 0.0281, "samples": 336000, "wall_s": 20928.0, "cuda_gb": 16.86}
{"step": 10520, "loss": 0.0023, "loss_mean50": 0.00167, "lr": 7.724080209103593e-06, "grad_norm": 0.0761, "samples": 336640, "wall_s": 20967.0, "cuda_gb": 16.86}
{"step": 10540, "loss": 0.000708, "loss_mean50": 0.00168, "lr": 7.681504952256218e-06, "grad_norm": 0.0251, "samples": 337280, "wall_s": 21006.0, "cuda_gb": 16.86}
{"step": 10560, "loss": 0.001286, "loss_mean50": 0.001657, "lr": 7.639051927133128e-06, "grad_norm": 0.0255, "samples": 337920, "wall_s": 21044.7, "cuda_gb": 16.86}
{"step": 10580, "loss": 0.001834, "loss_mean50": 0.00165, "lr": 7.596721988823835e-06, "grad_norm": 0.037, "samples": 338560, "wall_s": 21083.5, "cuda_gb": 16.86}
{"step": 10600, "loss": 0.001363, "loss_mean50": 0.001686, "lr": 7.554515989938631e-06, "grad_norm": 0.022, "samples": 339200, "wall_s": 21122.3, "cuda_gb": 16.86}
{"step": 10620, "loss": 0.001079, "loss_mean50": 0.001721, "lr": 7.512434780591415e-06, "grad_norm": 0.0275, "samples": 339840, "wall_s": 21161.1, "cuda_gb": 16.86}
{"step": 10640, "loss": 0.000841, "loss_mean50": 0.001825, "lr": 7.470479208382591e-06, "grad_norm": 0.0255, "samples": 340480, "wall_s": 21199.9, "cuda_gb": 16.86}
{"step": 10660, "loss": 0.001057, "loss_mean50": 0.001674, "lr": 7.4286501183819604e-06, "grad_norm": 0.0164, "samples": 341120, "wall_s": 21238.7, "cuda_gb": 16.86}
{"step": 10680, "loss": 0.000901, "loss_mean50": 0.001596, "lr": 7.386948353111717e-06, "grad_norm": 0.0437, "samples": 341760, "wall_s": 21277.6, "cuda_gb": 16.86}
{"step": 10700, "loss": 0.001424, "loss_mean50": 0.001639, "lr": 7.345374752529491e-06, "grad_norm": 0.029, "samples": 342400, "wall_s": 21316.3, "cuda_gb": 16.86}
{"step": 10720, "loss": 0.000916, "loss_mean50": 0.001682, "lr": 7.30393015401141e-06, "grad_norm": 0.0174, "samples": 343040, "wall_s": 21355.1, "cuda_gb": 16.86}
{"step": 10740, "loss": 0.002036, "loss_mean50": 0.001699, "lr": 7.2626153923352316e-06, "grad_norm": 0.0381, "samples": 343680, "wall_s": 21393.9, "cuda_gb": 16.86}
{"step": 10760, "loss": 0.002804, "loss_mean50": 0.001822, "lr": 7.221431299663555e-06, "grad_norm": 0.0703, "samples": 344320, "wall_s": 21432.8, "cuda_gb": 16.86}
{"step": 10780, "loss": 0.001361, "loss_mean50": 0.001824, "lr": 7.18037870552703e-06, "grad_norm": 0.027, "samples": 344960, "wall_s": 21471.6, "cuda_gb": 16.86}
{"step": 10800, "loss": 0.000941, "loss_mean50": 0.001768, "lr": 7.139458436807661e-06, "grad_norm": 0.0158, "samples": 345600, "wall_s": 21510.3, "cuda_gb": 16.86}
{"step": 10820, "loss": 0.003554, "loss_mean50": 0.001869, "lr": 7.098671317722161e-06, "grad_norm": 0.0944, "samples": 346240, "wall_s": 21549.1, "cuda_gb": 16.86}
{"step": 10840, "loss": 0.001988, "loss_mean50": 0.00187, "lr": 7.05801816980533e-06, "grad_norm": 0.0334, "samples": 346880, "wall_s": 21587.9, "cuda_gb": 16.86}
{"step": 10860, "loss": 0.001581, "loss_mean50": 0.001795, "lr": 7.0174998118935295e-06, "grad_norm": 0.0229, "samples": 347520, "wall_s": 21626.7, "cuda_gb": 16.86}
{"step": 10880, "loss": 0.002179, "loss_mean50": 0.001668, "lr": 6.977117060108165e-06, "grad_norm": 0.0655, "samples": 348160, "wall_s": 21665.4, "cuda_gb": 16.86}
{"step": 10900, "loss": 0.001641, "loss_mean50": 0.001771, "lr": 6.9368707278392706e-06, "grad_norm": 0.0243, "samples": 348800, "wall_s": 21704.1, "cuda_gb": 16.86}
{"step": 10920, "loss": 0.002075, "loss_mean50": 0.001706, "lr": 6.896761625729117e-06, "grad_norm": 0.0271, "samples": 349440, "wall_s": 21742.9, "cuda_gb": 16.86}
{"step": 10940, "loss": 0.000738, "loss_mean50": 0.001644, "lr": 6.856790561655879e-06, "grad_norm": 0.0232, "samples": 350080, "wall_s": 21781.8, "cuda_gb": 16.86}
{"step": 10960, "loss": 0.002331, "loss_mean50": 0.001857, "lr": 6.816958340717365e-06, "grad_norm": 0.0962, "samples": 350720, "wall_s": 21820.6, "cuda_gb": 16.86}
{"step": 10980, "loss": 0.001702, "loss_mean50": 0.001802, "lr": 6.777265765214818e-06, "grad_norm": 0.0572, "samples": 351360, "wall_s": 21859.3, "cuda_gb": 16.86}
{"step": 11000, "loss": 0.001792, "loss_mean50": 0.001728, "lr": 6.737713634636714e-06, "grad_norm": 0.0519, "samples": 352000, "wall_s": 21898.0, "cuda_gb": 16.86}
{"step": 11000, "val_flow_mse": 0.001679}
{"step": 11020, "loss": 0.002297, "loss_mean50": 0.001639, "lr": 6.698302745642702e-06, "grad_norm": 0.0318, "samples": 352640, "wall_s": 21965.4, "cuda_gb": 16.86}
{"step": 11040, "loss": 0.001447, "loss_mean50": 0.001628, "lr": 6.659033892047546e-06, "grad_norm": 0.0256, "samples": 353280, "wall_s": 22004.2, "cuda_gb": 16.86}
{"step": 11060, "loss": 0.00214, "loss_mean50": 0.001623, "lr": 6.619907864805116e-06, "grad_norm": 0.0446, "samples": 353920, "wall_s": 22043.1, "cuda_gb": 16.86}
{"step": 11080, "loss": 0.001097, "loss_mean50": 0.001771, "lr": 6.580925451992481e-06, "grad_norm": 0.0512, "samples": 354560, "wall_s": 22081.9, "cuda_gb": 16.86}
{"step": 11100, "loss": 0.000888, "loss_mean50": 0.001819, "lr": 6.542087438794027e-06, "grad_norm": 0.0187, "samples": 355200, "wall_s": 22120.9, "cuda_gb": 16.86}
{"step": 11120, "loss": 0.001211, "loss_mean50": 0.001666, "lr": 6.503394607485626e-06, "grad_norm": 0.0233, "samples": 355840, "wall_s": 22159.7, "cuda_gb": 16.86}
{"step": 11140, "loss": 0.00136, "loss_mean50": 0.001734, "lr": 6.464847737418916e-06, "grad_norm": 0.0336, "samples": 356480, "wall_s": 22198.5, "cuda_gb": 16.86}
{"step": 11160, "loss": 0.001862, "loss_mean50": 0.00174, "lr": 6.426447605005567e-06, "grad_norm": 0.0346, "samples": 357120, "wall_s": 22237.2, "cuda_gb": 16.86}
{"step": 11180, "loss": 0.002682, "loss_mean50": 0.001893, "lr": 6.388194983701658e-06, "grad_norm": 0.0765, "samples": 357760, "wall_s": 22276.0, "cuda_gb": 16.86}
{"step": 11200, "loss": 0.000747, "loss_mean50": 0.00176, "lr": 6.350090643992096e-06, "grad_norm": 0.0149, "samples": 358400, "wall_s": 22314.9, "cuda_gb": 16.86}
{"step": 11220, "loss": 0.001675, "loss_mean50": 0.001797, "lr": 6.312135353375108e-06, "grad_norm": 0.0418, "samples": 359040, "wall_s": 22353.6, "cuda_gb": 16.86}
{"step": 11240, "loss": 0.00199, "loss_mean50": 0.001819, "lr": 6.274329876346763e-06, "grad_norm": 0.0142, "samples": 359680, "wall_s": 22392.2, "cuda_gb": 16.86}
{"step": 11260, "loss": 0.001059, "loss_mean50": 0.001769, "lr": 6.2366749743855796e-06, "grad_norm": 0.016, "samples": 360320, "wall_s": 22431.0, "cuda_gb": 16.86}
{"step": 11280, "loss": 0.00279, "loss_mean50": 0.001606, "lr": 6.199171405937199e-06, "grad_norm": 0.0615, "samples": 360960, "wall_s": 22469.8, "cuda_gb": 16.86}
{"step": 11300, "loss": 0.001087, "loss_mean50": 0.001648, "lr": 6.161819926399091e-06, "grad_norm": 0.0215, "samples": 361600, "wall_s": 22508.6, "cuda_gb": 16.86}
{"step": 11320, "loss": 0.001443, "loss_mean50": 0.001657, "lr": 6.124621288105361e-06, "grad_norm": 0.0218, "samples": 362240, "wall_s": 22547.3, "cuda_gb": 16.86}
{"step": 11340, "loss": 0.001751, "loss_mean50": 0.00173, "lr": 6.0875762403115745e-06, "grad_norm": 0.0318, "samples": 362880, "wall_s": 22586.1, "cuda_gb": 16.86}
{"step": 11360, "loss": 0.002102, "loss_mean50": 0.001627, "lr": 6.050685529179673e-06, "grad_norm": 0.0305, "samples": 363520, "wall_s": 22624.9, "cuda_gb": 16.86}
{"step": 11380, "loss": 0.001989, "loss_mean50": 0.001754, "lr": 6.013949897762954e-06, "grad_norm": 0.0198, "samples": 364160, "wall_s": 22663.6, "cuda_gb": 16.86}
{"step": 11400, "loss": 0.001205, "loss_mean50": 0.001683, "lr": 5.977370085991092e-06, "grad_norm": 0.0309, "samples": 364800, "wall_s": 22702.4, "cuda_gb": 16.86}
{"step": 11420, "loss": 0.002001, "loss_mean50": 0.001724, "lr": 5.9409468306552325e-06, "grad_norm": 0.0482, "samples": 365440, "wall_s": 22741.1, "cuda_gb": 16.86}
{"step": 11440, "loss": 0.002183, "loss_mean50": 0.001563, "lr": 5.904680865393181e-06, "grad_norm": 0.0234, "samples": 366080, "wall_s": 22779.8, "cuda_gb": 16.86}
{"step": 11460, "loss": 0.000814, "loss_mean50": 0.001597, "lr": 5.868572920674581e-06, "grad_norm": 0.0146, "samples": 366720, "wall_s": 22818.6, "cuda_gb": 16.86}
{"step": 11480, "loss": 0.001091, "loss_mean50": 0.00161, "lr": 5.83262372378623e-06, "grad_norm": 0.0252, "samples": 367360, "wall_s": 22857.3, "cuda_gb": 16.86}
{"step": 11500, "loss": 0.001718, "loss_mean50": 0.001683, "lr": 5.7968339988174385e-06, "grad_norm": 0.027, "samples": 368000, "wall_s": 22896.1, "cuda_gb": 16.86}
{"step": 11520, "loss": 0.002314, "loss_mean50": 0.001725, "lr": 5.7612044666454034e-06, "grad_norm": 0.0335, "samples": 368640, "wall_s": 22934.9, "cuda_gb": 16.86}
{"step": 11540, "loss": 0.001329, "loss_mean50": 0.001656, "lr": 5.7257358449207435e-06, "grad_norm": 0.0202, "samples": 369280, "wall_s": 22973.7, "cuda_gb": 16.86}
{"step": 11560, "loss": 0.001027, "loss_mean50": 0.001547, "lr": 5.690428848053e-06, "grad_norm": 0.0197, "samples": 369920, "wall_s": 23012.5, "cuda_gb": 16.86}
{"step": 11580, "loss": 0.001766, "loss_mean50": 0.001574, "lr": 5.655284187196265e-06, "grad_norm": 0.0257, "samples": 370560, "wall_s": 23051.3, "cuda_gb": 16.86}
{"step": 11600, "loss": 0.001193, "loss_mean50": 0.001588, "lr": 5.620302570234867e-06, "grad_norm": 0.0214, "samples": 371200, "wall_s": 23090.1, "cuda_gb": 16.86}
{"step": 11620, "loss": 0.000677, "loss_mean50": 0.001628, "lr": 5.5854847017690945e-06, "grad_norm": 0.0168, "samples": 371840, "wall_s": 23128.8, "cuda_gb": 16.86}
{"step": 11640, "loss": 0.000881, "loss_mean50": 0.001715, "lr": 5.550831283100999e-06, "grad_norm": 0.0146, "samples": 372480, "wall_s": 23167.6, "cuda_gb": 16.86}
{"step": 11660, "loss": 0.001693, "loss_mean50": 0.001613, "lr": 5.516343012220304e-06, "grad_norm": 0.0732, "samples": 373120, "wall_s": 23206.3, "cuda_gb": 16.86}
{"step": 11680, "loss": 0.001289, "loss_mean50": 0.001636, "lr": 5.482020583790311e-06, "grad_norm": 0.0272, "samples": 373760, "wall_s": 23245.2, "cuda_gb": 16.86}
{"step": 11700, "loss": 0.000811, "loss_mean50": 0.001538, "lr": 5.447864689133917e-06, "grad_norm": 0.016, "samples": 374400, "wall_s": 23284.0, "cuda_gb": 16.86}
{"step": 11720, "loss": 0.001527, "loss_mean50": 0.00158, "lr": 5.413876016219707e-06, "grad_norm": 0.0258, "samples": 375040, "wall_s": 23322.8, "cuda_gb": 16.86}
{"step": 11740, "loss": 0.003672, "loss_mean50": 0.001604, "lr": 5.380055249648071e-06, "grad_norm": 0.043, "samples": 375680, "wall_s": 23361.6, "cuda_gb": 16.86}
{"step": 11760, "loss": 0.002349, "loss_mean50": 0.00161, "lr": 5.346403070637427e-06, "grad_norm": 0.0382, "samples": 376320, "wall_s": 23400.4, "cuda_gb": 16.86}
{"step": 11780, "loss": 0.001569, "loss_mean50": 0.001706, "lr": 5.312920157010506e-06, "grad_norm": 0.0359, "samples": 376960, "wall_s": 23439.2, "cuda_gb": 16.86}
{"step": 11800, "loss": 0.001647, "loss_mean50": 0.001769, "lr": 5.2796071831806846e-06, "grad_norm": 0.0252, "samples": 377600, "wall_s": 23478.1, "cuda_gb": 16.86}
{"step": 11820, "loss": 0.000513, "loss_mean50": 0.00162, "lr": 5.2464648201384225e-06, "grad_norm": 0.0126, "samples": 378240, "wall_s": 23516.8, "cuda_gb": 16.86}
{"step": 11840, "loss": 0.00061, "loss_mean50": 0.001423, "lr": 5.213493735437722e-06, "grad_norm": 0.0111, "samples": 378880, "wall_s": 23555.6, "cuda_gb": 16.86}
{"step": 11860, "loss": 0.001722, "loss_mean50": 0.001657, "lr": 5.180694593182696e-06, "grad_norm": 0.0313, "samples": 379520, "wall_s": 23594.5, "cuda_gb": 16.86}
{"step": 11880, "loss": 0.003446, "loss_mean50": 0.001665, "lr": 5.148068054014195e-06, "grad_norm": 0.1157, "samples": 380160, "wall_s": 23633.3, "cuda_gb": 16.86}
{"step": 11900, "loss": 0.001416, "loss_mean50": 0.001574, "lr": 5.115614775096492e-06, "grad_norm": 0.0163, "samples": 380800, "wall_s": 23672.1, "cuda_gb": 16.86}
{"step": 11920, "loss": 0.001878, "loss_mean50": 0.001531, "lr": 5.083335410104047e-06, "grad_norm": 0.0241, "samples": 381440, "wall_s": 23711.0, "cuda_gb": 16.86}
{"step": 11940, "loss": 0.002139, "loss_mean50": 0.001543, "lr": 5.051230609208352e-06, "grad_norm": 0.0544, "samples": 382080, "wall_s": 23750.0, "cuda_gb": 16.86}
{"step": 11960, "loss": 0.002004, "loss_mean50": 0.001574, "lr": 5.019301019064816e-06, "grad_norm": 0.039, "samples": 382720, "wall_s": 23789.1, "cuda_gb": 16.86}
{"step": 11980, "loss": 0.000693, "loss_mean50": 0.001559, "lr": 4.987547282799753e-06, "grad_norm": 0.0148, "samples": 383360, "wall_s": 23827.9, "cuda_gb": 16.86}
{"step": 12000, "loss": 0.001034, "loss_mean50": 0.001583, "lr": 4.955970039997429e-06, "grad_norm": 0.0209, "samples": 384000, "wall_s": 23866.9, "cuda_gb": 16.86}
{"step": 12000, "val_flow_mse": 0.001542}
{"step": 12020, "loss": 0.000731, "loss_mean50": 0.001472, "lr": 4.924569926687178e-06, "grad_norm": 0.0367, "samples": 384640, "wall_s": 23934.5, "cuda_gb": 16.86}
{"step": 12040, "loss": 0.000737, "loss_mean50": 0.001543, "lr": 4.893347575330571e-06, "grad_norm": 0.0368, "samples": 385280, "wall_s": 23973.4, "cuda_gb": 16.86}
{"step": 12060, "loss": 0.001291, "loss_mean50": 0.001555, "lr": 4.862303614808722e-06, "grad_norm": 0.0263, "samples": 385920, "wall_s": 24012.1, "cuda_gb": 16.86}
{"step": 12080, "loss": 0.00401, "loss_mean50": 0.001571, "lr": 4.831438670409574e-06, "grad_norm": 0.0729, "samples": 386560, "wall_s": 24051.8, "cuda_gb": 16.86}
{"step": 12100, "loss": 0.001632, "loss_mean50": 0.001499, "lr": 4.800753363815327e-06, "grad_norm": 0.0219, "samples": 387200, "wall_s": 24091.0, "cuda_gb": 16.86}
{"step": 12120, "loss": 0.001384, "loss_mean50": 0.001536, "lr": 4.770248313089926e-06, "grad_norm": 0.0232, "samples": 387840, "wall_s": 24130.2, "cuda_gb": 16.86}
{"step": 12140, "loss": 0.000738, "loss_mean50": 0.001539, "lr": 4.739924132666585e-06, "grad_norm": 0.0203, "samples": 388480, "wall_s": 24169.4, "cuda_gb": 16.86}
{"step": 12160, "loss": 0.001473, "loss_mean50": 0.001497, "lr": 4.709781433335432e-06, "grad_norm": 0.0228, "samples": 389120, "wall_s": 24208.5, "cuda_gb": 16.86}
{"step": 12180, "loss": 0.001023, "loss_mean50": 0.001448, "lr": 4.6798208222311985e-06, "grad_norm": 0.02, "samples": 389760, "wall_s": 24247.4, "cuda_gb": 16.86}
{"step": 12200, "loss": 0.001015, "loss_mean50": 0.001548, "lr": 4.6500429028209904e-06, "grad_norm": 0.0167, "samples": 390400, "wall_s": 24286.3, "cuda_gb": 16.86}
{"step": 12220, "loss": 0.001413, "loss_mean50": 0.001593, "lr": 4.620448274892142e-06, "grad_norm": 0.0156, "samples": 391040, "wall_s": 24325.1, "cuda_gb": 16.86}
{"step": 12240, "loss": 0.0014, "loss_mean50": 0.00151, "lr": 4.591037534540117e-06, "grad_norm": 0.0194, "samples": 391680, "wall_s": 24363.9, "cuda_gb": 16.86}
{"step": 12260, "loss": 0.001, "loss_mean50": 0.001455, "lr": 4.56181127415652e-06, "grad_norm": 0.0228, "samples": 392320, "wall_s": 24402.8, "cuda_gb": 16.86}
{"step": 12280, "loss": 0.00101, "loss_mean50": 0.00149, "lr": 4.5327700824171575e-06, "grad_norm": 0.0252, "samples": 392960, "wall_s": 24441.6, "cuda_gb": 16.86}
{"step": 12300, "loss": 0.001534, "loss_mean50": 0.001439, "lr": 4.5039145442701735e-06, "grad_norm": 0.0173, "samples": 393600, "wall_s": 24480.5, "cuda_gb": 16.86}
{"step": 12320, "loss": 0.00102, "loss_mean50": 0.00154, "lr": 4.475245240924278e-06, "grad_norm": 0.016, "samples": 394240, "wall_s": 24519.4, "cuda_gb": 16.86}
{"step": 12340, "loss": 0.001196, "loss_mean50": 0.001578, "lr": 4.446762749837041e-06, "grad_norm": 0.0252, "samples": 394880, "wall_s": 24558.2, "cuda_gb": 16.86}
{"step": 12360, "loss": 0.001098, "loss_mean50": 0.001465, "lr": 4.418467644703254e-06, "grad_norm": 0.025, "samples": 395520, "wall_s": 24597.1, "cuda_gb": 16.86}
{"step": 12380, "loss": 0.000861, "loss_mean50": 0.001367, "lr": 4.3903604954433725e-06, "grad_norm": 0.0215, "samples": 396160, "wall_s": 24636.0, "cuda_gb": 16.86}
{"step": 12400, "loss": 0.000691, "loss_mean50": 0.001303, "lr": 4.362441868192056e-06, "grad_norm": 0.0159, "samples": 396800, "wall_s": 24674.8, "cuda_gb": 16.86}
{"step": 12420, "loss": 0.001771, "loss_mean50": 0.001345, "lr": 4.334712325286726e-06, "grad_norm": 0.0263, "samples": 397440, "wall_s": 24713.6, "cuda_gb": 16.86}
{"step": 12440, "loss": 0.002502, "loss_mean50": 0.001403, "lr": 4.307172425256291e-06, "grad_norm": 0.069, "samples": 398080, "wall_s": 24752.5, "cuda_gb": 16.86}
{"step": 12460, "loss": 0.002138, "loss_mean50": 0.001426, "lr": 4.279822722809855e-06, "grad_norm": 0.0441, "samples": 398720, "wall_s": 24791.3, "cuda_gb": 16.86}
{"step": 12480, "loss": 0.001487, "loss_mean50": 0.001488, "lr": 4.25266376882556e-06, "grad_norm": 0.0295, "samples": 399360, "wall_s": 24830.1, "cuda_gb": 16.86}
{"step": 12500, "loss": 0.00082, "loss_mean50": 0.001553, "lr": 4.2256961103394905e-06, "grad_norm": 0.0164, "samples": 400000, "wall_s": 24868.9, "cuda_gb": 16.86}
{"step": 12520, "loss": 0.002443, "loss_mean50": 0.001558, "lr": 4.198920290534661e-06, "grad_norm": 0.0493, "samples": 400640, "wall_s": 24938.6, "cuda_gb": 16.86}
{"step": 12540, "loss": 0.001833, "loss_mean50": 0.001568, "lr": 4.1723368487300514e-06, "grad_norm": 0.0307, "samples": 401280, "wall_s": 24977.4, "cuda_gb": 16.86}
{"step": 12560, "loss": 0.001037, "loss_mean50": 0.00143, "lr": 4.145946320369784e-06, "grad_norm": 0.0152, "samples": 401920, "wall_s": 25016.4, "cuda_gb": 16.86}
{"step": 12580, "loss": 0.000741, "loss_mean50": 0.001435, "lr": 4.119749237012297e-06, "grad_norm": 0.0132, "samples": 402560, "wall_s": 25055.4, "cuda_gb": 16.86}
{"step": 12600, "loss": 0.000897, "loss_mean50": 0.001365, "lr": 4.093746126319657e-06, "grad_norm": 0.0172, "samples": 403200, "wall_s": 25094.3, "cuda_gb": 16.86}
{"step": 12620, "loss": 0.001747, "loss_mean50": 0.00139, "lr": 4.067937512046945e-06, "grad_norm": 0.0412, "samples": 403840, "wall_s": 25133.2, "cuda_gb": 16.86}
{"step": 12640, "loss": 0.001942, "loss_mean50": 0.001406, "lr": 4.042323914031678e-06, "grad_norm": 0.0401, "samples": 404480, "wall_s": 25172.1, "cuda_gb": 16.86}
{"step": 12660, "loss": 0.00181, "loss_mean50": 0.001483, "lr": 4.016905848183357e-06, "grad_norm": 0.0375, "samples": 405120, "wall_s": 25211.0, "cuda_gb": 16.86}
{"step": 12680, "loss": 0.00146, "loss_mean50": 0.001645, "lr": 3.991683826473071e-06, "grad_norm": 0.0267, "samples": 405760, "wall_s": 25249.9, "cuda_gb": 16.86}
{"step": 12700, "loss": 0.001199, "loss_mean50": 0.001511, "lr": 3.9666583569231835e-06, "grad_norm": 0.025, "samples": 406400, "wall_s": 25288.8, "cuda_gb": 16.86}
{"step": 12720, "loss": 0.001931, "loss_mean50": 0.001779, "lr": 3.9418299435971e-06, "grad_norm": 0.0269, "samples": 407040, "wall_s": 25327.8, "cuda_gb": 16.86}
{"step": 12740, "loss": 0.001029, "loss_mean50": 0.001748, "lr": 3.9171990865891265e-06, "grad_norm": 0.0318, "samples": 407680, "wall_s": 25366.9, "cuda_gb": 16.86}
{"step": 12760, "loss": 0.00152, "loss_mean50": 0.001534, "lr": 3.892766282014372e-06, "grad_norm": 0.0155, "samples": 408320, "wall_s": 25405.9, "cuda_gb": 16.86}
{"step": 12780, "loss": 0.003319, "loss_mean50": 0.001412, "lr": 3.868532021998779e-06, "grad_norm": 0.077, "samples": 408960, "wall_s": 25444.9, "cuda_gb": 16.86}
{"step": 12800, "loss": 0.001009, "loss_mean50": 0.001512, "lr": 3.844496794669201e-06, "grad_norm": 0.0196, "samples": 409600, "wall_s": 25483.8, "cuda_gb": 16.86}
{"step": 12820, "loss": 0.00099, "loss_mean50": 0.001549, "lr": 3.820661084143568e-06, "grad_norm": 0.0192, "samples": 410240, "wall_s": 25522.9, "cuda_gb": 16.86}
{"step": 12840, "loss": 0.001859, "loss_mean50": 0.001577, "lr": 3.797025370521147e-06, "grad_norm": 0.0258, "samples": 410880, "wall_s": 25561.7, "cuda_gb": 16.86}
{"step": 12860, "loss": 0.000884, "loss_mean50": 0.001526, "lr": 3.7735901298728575e-06, "grad_norm": 0.0149, "samples": 411520, "wall_s": 25600.6, "cuda_gb": 16.86}
{"step": 12880, "loss": 0.001352, "loss_mean50": 0.001531, "lr": 3.7503558342316883e-06, "grad_norm": 0.027, "samples": 412160, "wall_s": 25639.5, "cuda_gb": 16.86}
{"step": 12900, "loss": 0.002201, "loss_mean50": 0.001588, "lr": 3.7273229515831993e-06, "grad_norm": 0.0346, "samples": 412800, "wall_s": 25678.4, "cuda_gb": 16.86}
{"step": 12920, "loss": 0.001286, "loss_mean50": 0.001588, "lr": 3.704491945856082e-06, "grad_norm": 0.0179, "samples": 413440, "wall_s": 25717.4, "cuda_gb": 16.86}
{"step": 12940, "loss": 0.000731, "loss_mean50": 0.001541, "lr": 3.6818632769128106e-06, "grad_norm": 0.0103, "samples": 414080, "wall_s": 25756.3, "cuda_gb": 16.86}
{"step": 12960, "loss": 0.000709, "loss_mean50": 0.001505, "lr": 3.6594374005404026e-06, "grad_norm": 0.0335, "samples": 414720, "wall_s": 25795.2, "cuda_gb": 16.86}
{"step": 12980, "loss": 0.000578, "loss_mean50": 0.001464, "lr": 3.637214768441219e-06, "grad_norm": 0.0145, "samples": 415360, "wall_s": 25834.1, "cuda_gb": 16.86}
{"step": 13000, "loss": 0.002302, "loss_mean50": 0.001484, "lr": 3.6151958282238664e-06, "grad_norm": 0.0391, "samples": 416000, "wall_s": 25873.0, "cuda_gb": 16.86}
{"step": 13000, "val_flow_mse": 0.00148}
{"step": 13020, "loss": 0.00176, "loss_mean50": 0.00165, "lr": 3.593381023394196e-06, "grad_norm": 0.0279, "samples": 416640, "wall_s": 25940.7, "cuda_gb": 16.86}
{"step": 13040, "loss": 0.002676, "loss_mean50": 0.001627, "lr": 3.571770793346353e-06, "grad_norm": 0.0394, "samples": 417280, "wall_s": 25979.6, "cuda_gb": 16.86}
{"step": 13060, "loss": 0.001029, "loss_mean50": 0.001525, "lr": 3.5503655733539372e-06, "grad_norm": 0.022, "samples": 417920, "wall_s": 26018.5, "cuda_gb": 16.86}
{"step": 13080, "loss": 0.000876, "loss_mean50": 0.00129, "lr": 3.5291657945612303e-06, "grad_norm": 0.0262, "samples": 418560, "wall_s": 26057.3, "cuda_gb": 16.86}
{"step": 13100, "loss": 0.00225, "loss_mean50": 0.00124, "lr": 3.5081718839745164e-06, "grad_norm": 0.0292, "samples": 419200, "wall_s": 26096.2, "cuda_gb": 16.86}
{"step": 13120, "loss": 0.000806, "loss_mean50": 0.001324, "lr": 3.4873842644534798e-06, "grad_norm": 0.0324, "samples": 419840, "wall_s": 26135.3, "cuda_gb": 16.86}
{"step": 13140, "loss": 0.001034, "loss_mean50": 0.001352, "lr": 3.466803354702685e-06, "grad_norm": 0.0212, "samples": 420480, "wall_s": 26174.2, "cuda_gb": 16.86}
{"step": 13160, "loss": 0.001174, "loss_mean50": 0.00136, "lr": 3.4464295692631432e-06, "grad_norm": 0.0243, "samples": 421120, "wall_s": 26213.1, "cuda_gb": 16.86}
{"step": 13180, "loss": 0.001084, "loss_mean50": 0.001418, "lr": 3.4262633185039696e-06, "grad_norm": 0.0216, "samples": 421760, "wall_s": 26251.9, "cuda_gb": 16.86}
{"step": 13200, "loss": 0.00193, "loss_mean50": 0.001421, "lr": 3.406305008614106e-06, "grad_norm": 0.0294, "samples": 422400, "wall_s": 26290.7, "cuda_gb": 16.86}
{"step": 13220, "loss": 0.000664, "loss_mean50": 0.001391, "lr": 3.3865550415941497e-06, "grad_norm": 0.0396, "samples": 423040, "wall_s": 26329.5, "cuda_gb": 16.86}
{"step": 13240, "loss": 0.001278, "loss_mean50": 0.001393, "lr": 3.367013815248255e-06, "grad_norm": 0.0236, "samples": 423680, "wall_s": 26368.2, "cuda_gb": 16.86}
{"step": 13260, "loss": 0.000897, "loss_mean50": 0.001383, "lr": 3.347681723176115e-06, "grad_norm": 0.0292, "samples": 424320, "wall_s": 26406.9, "cuda_gb": 16.86}
{"step": 13280, "loss": 0.001496, "loss_mean50": 0.001421, "lr": 3.3285591547650326e-06, "grad_norm": 0.0414, "samples": 424960, "wall_s": 26445.6, "cuda_gb": 16.86}
{"step": 13300, "loss": 0.00076, "loss_mean50": 0.00142, "lr": 3.3096464951820954e-06, "grad_norm": 0.0172, "samples": 425600, "wall_s": 26484.4, "cuda_gb": 16.86}
{"step": 13320, "loss": 0.001494, "loss_mean50": 0.00138, "lr": 3.290944125366388e-06, "grad_norm": 0.0357, "samples": 426240, "wall_s": 26523.2, "cuda_gb": 16.86}
{"step": 13340, "loss": 0.001564, "loss_mean50": 0.00144, "lr": 3.272452422021341e-06, "grad_norm": 0.0276, "samples": 426880, "wall_s": 26562.0, "cuda_gb": 16.86}
{"step": 13360, "loss": 0.001473, "loss_mean50": 0.001452, "lr": 3.254171757607145e-06, "grad_norm": 0.0259, "samples": 427520, "wall_s": 26600.8, "cuda_gb": 16.86}
{"step": 13380, "loss": 0.00188, "loss_mean50": 0.001478, "lr": 3.236102500333231e-06, "grad_norm": 0.0333, "samples": 428160, "wall_s": 26639.7, "cuda_gb": 16.86}
{"step": 13400, "loss": 0.000596, "loss_mean50": 0.001409, "lr": 3.2182450141508667e-06, "grad_norm": 0.0139, "samples": 428800, "wall_s": 26678.6, "cuda_gb": 16.86}
{"step": 13420, "loss": 0.00332, "loss_mean50": 0.00142, "lr": 3.2005996587458287e-06, "grad_norm": 0.0475, "samples": 429440, "wall_s": 26717.3, "cuda_gb": 16.86}
{"step": 13440, "loss": 0.002532, "loss_mean50": 0.001559, "lr": 3.183166789531146e-06, "grad_norm": 0.0355, "samples": 430080, "wall_s": 26756.0, "cuda_gb": 16.86}
{"step": 13460, "loss": 0.001416, "loss_mean50": 0.001596, "lr": 3.1659467576399488e-06, "grad_norm": 0.0483, "samples": 430720, "wall_s": 26794.8, "cuda_gb": 16.86}
{"step": 13480, "loss": 0.001709, "loss_mean50": 0.001453, "lr": 3.1489399099183936e-06, "grad_norm": 0.0491, "samples": 431360, "wall_s": 26833.5, "cuda_gb": 16.86}
{"step": 13500, "loss": 0.001052, "loss_mean50": 0.001451, "lr": 3.132146588918679e-06, "grad_norm": 0.0178, "samples": 432000, "wall_s": 26872.2, "cuda_gb": 16.86}
{"step": 13520, "loss": 0.000885, "loss_mean50": 0.001459, "lr": 3.1155671328921473e-06, "grad_norm": 0.0174, "samples": 432640, "wall_s": 26910.9, "cuda_gb": 16.86}
{"step": 13540, "loss": 0.002835, "loss_mean50": 0.001518, "lr": 3.0992018757824658e-06, "grad_norm": 0.0463, "samples": 433280, "wall_s": 26949.7, "cuda_gb": 16.86}
{"step": 13560, "loss": 0.001692, "loss_mean50": 0.001491, "lr": 3.0830511472189056e-06, "grad_norm": 0.0513, "samples": 433920, "wall_s": 26988.5, "cuda_gb": 16.86}
{"step": 13580, "loss": 0.00281, "loss_mean50": 0.001553, "lr": 3.0671152725096976e-06, "grad_norm": 0.0257, "samples": 434560, "wall_s": 27027.3, "cuda_gb": 16.86}
{"step": 13600, "loss": 0.001159, "loss_mean50": 0.00152, "lr": 3.0513945726354877e-06, "grad_norm": 0.0221, "samples": 435200, "wall_s": 27066.1, "cuda_gb": 16.86}
{"step": 13620, "loss": 0.001603, "loss_mean50": 0.001608, "lr": 3.0358893642428604e-06, "grad_norm": 0.0264, "samples": 435840, "wall_s": 27105.0, "cuda_gb": 16.86}
{"step": 13640, "loss": 0.001427, "loss_mean50": 0.001446, "lr": 3.020599959637975e-06, "grad_norm": 0.027, "samples": 436480, "wall_s": 27143.8, "cuda_gb": 16.86}
{"step": 13660, "loss": 0.001528, "loss_mean50": 0.001319, "lr": 3.005526666780261e-06, "grad_norm": 0.035, "samples": 437120, "wall_s": 27182.6, "cuda_gb": 16.86}
{"step": 13680, "loss": 0.000626, "loss_mean50": 0.001367, "lr": 2.990669789276225e-06, "grad_norm": 0.0115, "samples": 437760, "wall_s": 27221.4, "cuda_gb": 16.86}
{"step": 13700, "loss": 0.001181, "loss_mean50": 0.001498, "lr": 2.9760296263733332e-06, "grad_norm": 0.0506, "samples": 438400, "wall_s": 27260.2, "cuda_gb": 16.86}
{"step": 13720, "loss": 0.001597, "loss_mean50": 0.001481, "lr": 2.961606472953977e-06, "grad_norm": 0.0231, "samples": 439040, "wall_s": 27298.9, "cuda_gb": 16.86}
{"step": 13740, "loss": 0.000903, "loss_mean50": 0.001465, "lr": 2.947400619529546e-06, "grad_norm": 0.0283, "samples": 439680, "wall_s": 27337.7, "cuda_gb": 16.86}
{"step": 13760, "loss": 0.001895, "loss_mean50": 0.001507, "lr": 2.933412352234567e-06, "grad_norm": 0.0253, "samples": 440320, "wall_s": 27376.5, "cuda_gb": 16.86}
{"step": 13780, "loss": 0.001731, "loss_mean50": 0.001357, "lr": 2.919641952820941e-06, "grad_norm": 0.0334, "samples": 440960, "wall_s": 27415.3, "cuda_gb": 16.86}
{"step": 13800, "loss": 0.000749, "loss_mean50": 0.0013, "lr": 2.906089698652277e-06, "grad_norm": 0.0196, "samples": 441600, "wall_s": 27454.2, "cuda_gb": 16.86}
{"step": 13820, "loss": 0.00133, "loss_mean50": 0.001371, "lr": 2.892755862698292e-06, "grad_norm": 0.0205, "samples": 442240, "wall_s": 27492.9, "cuda_gb": 16.86}
{"step": 13840, "loss": 0.002054, "loss_mean50": 0.001486, "lr": 2.8796407135293233e-06, "grad_norm": 0.03, "samples": 442880, "wall_s": 27531.7, "cuda_gb": 16.86}
{"step": 13860, "loss": 0.000723, "loss_mean50": 0.001431, "lr": 2.866744515310917e-06, "grad_norm": 0.0125, "samples": 443520, "wall_s": 27570.5, "cuda_gb": 16.86}
{"step": 13880, "loss": 0.001317, "loss_mean50": 0.001428, "lr": 2.854067527798503e-06, "grad_norm": 0.0632, "samples": 444160, "wall_s": 27609.2, "cuda_gb": 16.86}
{"step": 13900, "loss": 0.001004, "loss_mean50": 0.001462, "lr": 2.8416100063321654e-06, "grad_norm": 0.0296, "samples": 444800, "wall_s": 27648.0, "cuda_gb": 16.86}
{"step": 13920, "loss": 0.001331, "loss_mean50": 0.001559, "lr": 2.829372201831502e-06, "grad_norm": 0.0257, "samples": 445440, "wall_s": 27686.9, "cuda_gb": 16.86}
{"step": 13940, "loss": 0.001136, "loss_mean50": 0.001447, "lr": 2.817354360790566e-06, "grad_norm": 0.0211, "samples": 446080, "wall_s": 27725.8, "cuda_gb": 16.86}
{"step": 13960, "loss": 0.001135, "loss_mean50": 0.001397, "lr": 2.8055567252729046e-06, "grad_norm": 0.0249, "samples": 446720, "wall_s": 27764.7, "cuda_gb": 16.86}
{"step": 13980, "loss": 0.002487, "loss_mean50": 0.001532, "lr": 2.7939795329066803e-06, "grad_norm": 0.1275, "samples": 447360, "wall_s": 27803.6, "cuda_gb": 16.86}
{"step": 14000, "loss": 0.001429, "loss_mean50": 0.001556, "lr": 2.7826230168798876e-06, "grad_norm": 0.0308, "samples": 448000, "wall_s": 27842.6, "cuda_gb": 16.86}
{"step": 14000, "val_flow_mse": 0.00152}
{"step": 14020, "loss": 0.000712, "loss_mean50": 0.001473, "lr": 2.7714874059356517e-06, "grad_norm": 0.0112, "samples": 448640, "wall_s": 27910.2, "cuda_gb": 16.86}
{"step": 14040, "loss": 0.001608, "loss_mean50": 0.001422, "lr": 2.76057292436763e-06, "grad_norm": 0.0235, "samples": 449280, "wall_s": 27949.0, "cuda_gb": 16.86}
{"step": 14060, "loss": 0.000344, "loss_mean50": 0.001428, "lr": 2.7498797920154817e-06, "grad_norm": 0.0177, "samples": 449920, "wall_s": 27987.9, "cuda_gb": 16.86}
{"step": 14080, "loss": 0.0021, "loss_mean50": 0.001516, "lr": 2.7394082242604514e-06, "grad_norm": 0.0331, "samples": 450560, "wall_s": 28026.8, "cuda_gb": 16.86}
{"step": 14100, "loss": 0.000961, "loss_mean50": 0.001346, "lr": 2.7291584320210242e-06, "grad_norm": 0.0223, "samples": 451200, "wall_s": 28065.8, "cuda_gb": 16.86}
{"step": 14120, "loss": 0.002161, "loss_mean50": 0.001248, "lr": 2.7191306217486763e-06, "grad_norm": 0.041, "samples": 451840, "wall_s": 28104.8, "cuda_gb": 16.86}
{"step": 14140, "loss": 0.001363, "loss_mean50": 0.001195, "lr": 2.709324995423726e-06, "grad_norm": 0.0181, "samples": 452480, "wall_s": 28143.7, "cuda_gb": 16.86}
{"step": 14160, "loss": 0.001861, "loss_mean50": 0.001263, "lr": 2.6997417505512548e-06, "grad_norm": 0.062, "samples": 453120, "wall_s": 28182.6, "cuda_gb": 16.86}
{"step": 14180, "loss": 0.000587, "loss_mean50": 0.001391, "lr": 2.690381080157134e-06, "grad_norm": 0.0194, "samples": 453760, "wall_s": 28221.5, "cuda_gb": 16.86}
{"step": 14200, "loss": 0.002142, "loss_mean50": 0.001418, "lr": 2.681243172784135e-06, "grad_norm": 0.0243, "samples": 454400, "wall_s": 28260.4, "cuda_gb": 16.86}
{"step": 14220, "loss": 0.001072, "loss_mean50": 0.001499, "lr": 2.6723282124881367e-06, "grad_norm": 0.0236, "samples": 455040, "wall_s": 28299.2, "cuda_gb": 16.86}
{"step": 14240, "loss": 0.001988, "loss_mean50": 0.001333, "lr": 2.663636378834409e-06, "grad_norm": 0.0301, "samples": 455680, "wall_s": 28338.1, "cuda_gb": 16.86}
{"step": 14260, "loss": 0.000794, "loss_mean50": 0.00138, "lr": 2.6551678468940077e-06, "grad_norm": 0.0293, "samples": 456320, "wall_s": 28376.9, "cuda_gb": 16.86}
{"step": 14280, "loss": 0.001828, "loss_mean50": 0.001481, "lr": 2.6469227872402357e-06, "grad_norm": 0.061, "samples": 456960, "wall_s": 28415.8, "cuda_gb": 16.86}
{"step": 14300, "loss": 0.001831, "loss_mean50": 0.001536, "lr": 2.638901365945217e-06, "grad_norm": 0.026, "samples": 457600, "wall_s": 28454.8, "cuda_gb": 16.86}
{"step": 14320, "loss": 0.000501, "loss_mean50": 0.001412, "lr": 2.6311037445765488e-06, "grad_norm": 0.0132, "samples": 458240, "wall_s": 28493.7, "cuda_gb": 16.86}
{"step": 14340, "loss": 0.003406, "loss_mean50": 0.001583, "lr": 2.623530080194044e-06, "grad_norm": 0.0363, "samples": 458880, "wall_s": 28532.6, "cuda_gb": 16.86}
{"step": 14360, "loss": 0.002705, "loss_mean50": 0.001645, "lr": 2.6161805253465737e-06, "grad_norm": 0.0509, "samples": 459520, "wall_s": 28571.4, "cuda_gb": 16.86}
{"step": 14380, "loss": 0.00097, "loss_mean50": 0.001597, "lr": 2.609055228068987e-06, "grad_norm": 0.0303, "samples": 460160, "wall_s": 28610.3, "cuda_gb": 16.86}
{"step": 14400, "loss": 0.001162, "loss_mean50": 0.00151, "lr": 2.6021543318791354e-06, "grad_norm": 0.0166, "samples": 460800, "wall_s": 28649.1, "cuda_gb": 16.86}
{"step": 14420, "loss": 0.001299, "loss_mean50": 0.001422, "lr": 2.595477975774981e-06, "grad_norm": 0.0308, "samples": 461440, "wall_s": 28688.0, "cuda_gb": 16.86}
{"step": 14440, "loss": 0.002007, "loss_mean50": 0.001429, "lr": 2.589026294231797e-06, "grad_norm": 0.0822, "samples": 462080, "wall_s": 28726.8, "cuda_gb": 16.86}
{"step": 14460, "loss": 0.000851, "loss_mean50": 0.001452, "lr": 2.582799417199454e-06, "grad_norm": 0.0185, "samples": 462720, "wall_s": 28765.5, "cuda_gb": 16.86}
{"step": 14480, "loss": 0.001304, "loss_mean50": 0.001423, "lr": 2.5767974700998093e-06, "grad_norm": 0.0161, "samples": 463360, "wall_s": 28804.2, "cuda_gb": 16.86}
{"step": 14500, "loss": 0.000579, "loss_mean50": 0.001273, "lr": 2.5710205738241775e-06, "grad_norm": 0.0168, "samples": 464000, "wall_s": 28843.1, "cuda_gb": 16.86}
{"step": 14520, "loss": 0.001077, "loss_mean50": 0.001321, "lr": 2.565468844730895e-06, "grad_norm": 0.0199, "samples": 464640, "wall_s": 28881.8, "cuda_gb": 16.86}
{"step": 14540, "loss": 0.001548, "loss_mean50": 0.001334, "lr": 2.5601423946429807e-06, "grad_norm": 0.0555, "samples": 465280, "wall_s": 28920.6, "cuda_gb": 16.86}
{"step": 14560, "loss": 0.00188, "loss_mean50": 0.001391, "lr": 2.555041330845874e-06, "grad_norm": 0.0277, "samples": 465920, "wall_s": 28959.4, "cuda_gb": 16.86}
{"step": 14580, "loss": 0.001171, "loss_mean50": 0.001509, "lr": 2.550165756085284e-06, "grad_norm": 0.0273, "samples": 466560, "wall_s": 28998.2, "cuda_gb": 16.86}
{"step": 14600, "loss": 0.000883, "loss_mean50": 0.001468, "lr": 2.5455157685651184e-06, "grad_norm": 0.0166, "samples": 467200, "wall_s": 29036.9, "cuda_gb": 16.86}
{"step": 14620, "loss": 0.000775, "loss_mean50": 0.001466, "lr": 2.5410914619454962e-06, "grad_norm": 0.0225, "samples": 467840, "wall_s": 29075.7, "cuda_gb": 16.86}
{"step": 14640, "loss": 0.001347, "loss_mean50": 0.001366, "lr": 2.5368929253408747e-06, "grad_norm": 0.0356, "samples": 468480, "wall_s": 29114.5, "cuda_gb": 16.86}
{"step": 14660, "loss": 0.001482, "loss_mean50": 0.001435, "lr": 2.5329202433182453e-06, "grad_norm": 0.0163, "samples": 469120, "wall_s": 29153.3, "cuda_gb": 16.86}
{"step": 14680, "loss": 0.000748, "loss_mean50": 0.001497, "lr": 2.529173495895432e-06, "grad_norm": 0.018, "samples": 469760, "wall_s": 29192.1, "cuda_gb": 16.86}
{"step": 14700, "loss": 0.001246, "loss_mean50": 0.001595, "lr": 2.5256527585394805e-06, "grad_norm": 0.0286, "samples": 470400, "wall_s": 29231.0, "cuda_gb": 16.86}
{"step": 14720, "loss": 0.000665, "loss_mean50": 0.001564, "lr": 2.5223581021651405e-06, "grad_norm": 0.0145, "samples": 471040, "wall_s": 29270.0, "cuda_gb": 16.86}
{"step": 14740, "loss": 0.000905, "loss_mean50": 0.001463, "lr": 2.5192895931334335e-06, "grad_norm": 0.0149, "samples": 471680, "wall_s": 29308.8, "cuda_gb": 16.86}
{"step": 14760, "loss": 0.001352, "loss_mean50": 0.00149, "lr": 2.516447293250316e-06, "grad_norm": 0.0329, "samples": 472320, "wall_s": 29347.7, "cuda_gb": 16.86}
{"step": 14780, "loss": 0.000859, "loss_mean50": 0.001483, "lr": 2.5138312597654382e-06, "grad_norm": 0.0215, "samples": 472960, "wall_s": 29386.5, "cuda_gb": 16.86}
{"step": 14800, "loss": 0.002127, "loss_mean50": 0.001578, "lr": 2.5114415453709886e-06, "grad_norm": 0.0284, "samples": 473600, "wall_s": 29425.5, "cuda_gb": 16.86}
{"step": 14820, "loss": 0.001984, "loss_mean50": 0.001463, "lr": 2.509278198200633e-06, "grad_norm": 0.0458, "samples": 474240, "wall_s": 29464.4, "cuda_gb": 16.86}
{"step": 14840, "loss": 0.000832, "loss_mean50": 0.00129, "lr": 2.5073412618285445e-06, "grad_norm": 0.0157, "samples": 474880, "wall_s": 29503.3, "cuda_gb": 16.86}
{"step": 14860, "loss": 0.001242, "loss_mean50": 0.00118, "lr": 2.505630775268527e-06, "grad_norm": 0.0384, "samples": 475520, "wall_s": 29542.2, "cuda_gb": 16.86}
{"step": 14880, "loss": 0.001692, "loss_mean50": 0.001221, "lr": 2.5041467729732266e-06, "grad_norm": 0.0333, "samples": 476160, "wall_s": 29581.2, "cuda_gb": 16.86}
{"step": 14900, "loss": 0.002699, "loss_mean50": 0.001351, "lr": 2.502889284833443e-06, "grad_norm": 0.0454, "samples": 476800, "wall_s": 29620.1, "cuda_gb": 16.86}
{"step": 14920, "loss": 0.001361, "loss_mean50": 0.001471, "lr": 2.5018583361775205e-06, "grad_norm": 0.025, "samples": 477440, "wall_s": 29658.9, "cuda_gb": 16.86}
{"step": 14940, "loss": 0.000621, "loss_mean50": 0.00149, "lr": 2.501053947770846e-06, "grad_norm": 0.013, "samples": 478080, "wall_s": 29697.9, "cuda_gb": 16.86}
{"step": 14960, "loss": 0.000741, "loss_mean50": 0.001342, "lr": 2.5004761358154212e-06, "grad_norm": 0.0191, "samples": 478720, "wall_s": 29736.8, "cuda_gb": 16.86}
{"step": 14980, "loss": 0.002246, "loss_mean50": 0.001181, "lr": 2.5001249119495454e-06, "grad_norm": 0.0517, "samples": 479360, "wall_s": 29775.7, "cuda_gb": 16.86}
{"step": 15000, "loss": 0.001596, "loss_mean50": 0.001193, "lr": 2.5000002832475742e-06, "grad_norm": 0.0185, "samples": 480000, "wall_s": 29814.6, "cuda_gb": 16.86}
{"step": 15000, "val_flow_mse": 0.001561, "val_action_mae": 0.012072, "best": true}