pi05_arx_ambiguous / train_log.jsonl
KasLoot's picture
Upload folder using huggingface_hub
c61522f verified
Raw
History Blame Contribute Delete
123 kB
{"step": 20, "loss": 0.067208, "loss_mean50": 0.081939, "lr": 5e-07, "grad_norm": 0.2002, "samples": 640, "wall_s": 44.6, "cuda_gb": 16.86}
{"step": 40, "loss": 0.080062, "loss_mean50": 0.082754, "lr": 1e-06, "grad_norm": 0.2529, "samples": 1280, "wall_s": 83.4, "cuda_gb": 16.86}
{"step": 60, "loss": 0.064026, "loss_mean50": 0.079078, "lr": 1.5e-06, "grad_norm": 0.2111, "samples": 1920, "wall_s": 122.4, "cuda_gb": 16.86}
{"step": 80, "loss": 0.067946, "loss_mean50": 0.074302, "lr": 2e-06, "grad_norm": 0.1562, "samples": 2560, "wall_s": 161.4, "cuda_gb": 16.86}
{"step": 100, "loss": 0.038423, "loss_mean50": 0.062938, "lr": 2.5e-06, "grad_norm": 0.1492, "samples": 3200, "wall_s": 200.5, "cuda_gb": 16.86}
{"step": 120, "loss": 0.02301, "loss_mean50": 0.045117, "lr": 3e-06, "grad_norm": 0.1059, "samples": 3840, "wall_s": 239.5, "cuda_gb": 16.86}
{"step": 140, "loss": 0.016224, "loss_mean50": 0.029271, "lr": 3.5e-06, "grad_norm": 0.0653, "samples": 4480, "wall_s": 278.5, "cuda_gb": 16.86}
{"step": 160, "loss": 0.017074, "loss_mean50": 0.019534, "lr": 4e-06, "grad_norm": 0.0768, "samples": 5120, "wall_s": 317.6, "cuda_gb": 16.86}
{"step": 180, "loss": 0.015486, "loss_mean50": 0.016652, "lr": 4.5e-06, "grad_norm": 0.073, "samples": 5760, "wall_s": 356.6, "cuda_gb": 16.86}
{"step": 200, "loss": 0.016944, "loss_mean50": 0.015143, "lr": 5e-06, "grad_norm": 0.1032, "samples": 6400, "wall_s": 395.6, "cuda_gb": 16.86}
{"step": 220, "loss": 0.009363, "loss_mean50": 0.013227, "lr": 5.500000000000001e-06, "grad_norm": 0.0553, "samples": 7040, "wall_s": 434.6, "cuda_gb": 16.86}
{"step": 240, "loss": 0.008781, "loss_mean50": 0.012148, "lr": 6e-06, "grad_norm": 0.0574, "samples": 7680, "wall_s": 473.7, "cuda_gb": 16.86}
{"step": 260, "loss": 0.010875, "loss_mean50": 0.011276, "lr": 6.5000000000000004e-06, "grad_norm": 0.0963, "samples": 8320, "wall_s": 513.0, "cuda_gb": 16.86}
{"step": 280, "loss": 0.009348, "loss_mean50": 0.010277, "lr": 7e-06, "grad_norm": 0.0922, "samples": 8960, "wall_s": 552.2, "cuda_gb": 16.86}
{"step": 300, "loss": 0.011581, "loss_mean50": 0.010123, "lr": 7.5e-06, "grad_norm": 0.0594, "samples": 9600, "wall_s": 591.3, "cuda_gb": 16.86}
{"step": 320, "loss": 0.007272, "loss_mean50": 0.009754, "lr": 8e-06, "grad_norm": 0.0524, "samples": 10240, "wall_s": 630.4, "cuda_gb": 16.86}
{"step": 340, "loss": 0.008242, "loss_mean50": 0.009541, "lr": 8.5e-06, "grad_norm": 0.0543, "samples": 10880, "wall_s": 669.4, "cuda_gb": 16.86}
{"step": 360, "loss": 0.008943, "loss_mean50": 0.009514, "lr": 9e-06, "grad_norm": 0.0599, "samples": 11520, "wall_s": 708.4, "cuda_gb": 16.86}
{"step": 380, "loss": 0.005263, "loss_mean50": 0.008514, "lr": 9.5e-06, "grad_norm": 0.0533, "samples": 12160, "wall_s": 747.6, "cuda_gb": 16.86}
{"step": 400, "loss": 0.006857, "loss_mean50": 0.007888, "lr": 1e-05, "grad_norm": 0.055, "samples": 12800, "wall_s": 786.5, "cuda_gb": 16.86}
{"step": 420, "loss": 0.005418, "loss_mean50": 0.007707, "lr": 1.0500000000000001e-05, "grad_norm": 0.0488, "samples": 13440, "wall_s": 825.5, "cuda_gb": 16.86}
{"step": 440, "loss": 0.007814, "loss_mean50": 0.007764, "lr": 1.1000000000000001e-05, "grad_norm": 0.0632, "samples": 14080, "wall_s": 864.5, "cuda_gb": 16.86}
{"step": 460, "loss": 0.006673, "loss_mean50": 0.007533, "lr": 1.15e-05, "grad_norm": 0.0618, "samples": 14720, "wall_s": 903.5, "cuda_gb": 16.86}
{"step": 480, "loss": 0.005445, "loss_mean50": 0.007416, "lr": 1.2e-05, "grad_norm": 0.0506, "samples": 15360, "wall_s": 942.6, "cuda_gb": 16.86}
{"step": 500, "loss": 0.00549, "loss_mean50": 0.006934, "lr": 1.25e-05, "grad_norm": 0.0444, "samples": 16000, "wall_s": 981.6, "cuda_gb": 16.86}
{"step": 520, "loss": 0.005953, "loss_mean50": 0.006724, "lr": 1.3000000000000001e-05, "grad_norm": 0.0592, "samples": 16640, "wall_s": 1020.4, "cuda_gb": 16.86}
{"step": 540, "loss": 0.00591, "loss_mean50": 0.006687, "lr": 1.35e-05, "grad_norm": 0.0369, "samples": 17280, "wall_s": 1059.4, "cuda_gb": 16.86}
{"step": 560, "loss": 0.007787, "loss_mean50": 0.006948, "lr": 1.4e-05, "grad_norm": 0.0466, "samples": 17920, "wall_s": 1098.3, "cuda_gb": 16.86}
{"step": 580, "loss": 0.005246, "loss_mean50": 0.006995, "lr": 1.45e-05, "grad_norm": 0.0482, "samples": 18560, "wall_s": 1137.3, "cuda_gb": 16.86}
{"step": 600, "loss": 0.005657, "loss_mean50": 0.006473, "lr": 1.5e-05, "grad_norm": 0.0464, "samples": 19200, "wall_s": 1176.3, "cuda_gb": 16.86}
{"step": 620, "loss": 0.006045, "loss_mean50": 0.006346, "lr": 1.55e-05, "grad_norm": 0.0482, "samples": 19840, "wall_s": 1215.1, "cuda_gb": 16.86}
{"step": 640, "loss": 0.005013, "loss_mean50": 0.006256, "lr": 1.6e-05, "grad_norm": 0.0412, "samples": 20480, "wall_s": 1254.0, "cuda_gb": 16.86}
{"step": 660, "loss": 0.006335, "loss_mean50": 0.006403, "lr": 1.65e-05, "grad_norm": 0.0482, "samples": 21120, "wall_s": 1292.9, "cuda_gb": 16.86}
{"step": 680, "loss": 0.003668, "loss_mean50": 0.006101, "lr": 1.7e-05, "grad_norm": 0.0437, "samples": 21760, "wall_s": 1331.8, "cuda_gb": 16.86}
{"step": 700, "loss": 0.005056, "loss_mean50": 0.006104, "lr": 1.7500000000000002e-05, "grad_norm": 0.0393, "samples": 22400, "wall_s": 1370.7, "cuda_gb": 16.86}
{"step": 720, "loss": 0.00607, "loss_mean50": 0.006052, "lr": 1.8e-05, "grad_norm": 0.0415, "samples": 23040, "wall_s": 1409.7, "cuda_gb": 16.86}
{"step": 740, "loss": 0.00934, "loss_mean50": 0.005681, "lr": 1.8500000000000002e-05, "grad_norm": 0.1067, "samples": 23680, "wall_s": 1448.6, "cuda_gb": 16.86}
{"step": 760, "loss": 0.006776, "loss_mean50": 0.005991, "lr": 1.9e-05, "grad_norm": 0.067, "samples": 24320, "wall_s": 1487.5, "cuda_gb": 16.86}
{"step": 780, "loss": 0.00385, "loss_mean50": 0.006442, "lr": 1.95e-05, "grad_norm": 0.0335, "samples": 24960, "wall_s": 1526.3, "cuda_gb": 16.86}
{"step": 800, "loss": 0.005118, "loss_mean50": 0.006298, "lr": 2e-05, "grad_norm": 0.0425, "samples": 25600, "wall_s": 1565.3, "cuda_gb": 16.86}
{"step": 820, "loss": 0.015013, "loss_mean50": 0.006019, "lr": 2.05e-05, "grad_norm": 0.1266, "samples": 26240, "wall_s": 1604.2, "cuda_gb": 16.86}
{"step": 840, "loss": 0.007511, "loss_mean50": 0.006098, "lr": 2.1000000000000002e-05, "grad_norm": 0.0555, "samples": 26880, "wall_s": 1643.1, "cuda_gb": 16.86}
{"step": 860, "loss": 0.013417, "loss_mean50": 0.006196, "lr": 2.15e-05, "grad_norm": 0.142, "samples": 27520, "wall_s": 1682.1, "cuda_gb": 16.86}
{"step": 880, "loss": 0.006691, "loss_mean50": 0.005896, "lr": 2.2000000000000003e-05, "grad_norm": 0.0435, "samples": 28160, "wall_s": 1721.0, "cuda_gb": 16.86}
{"step": 900, "loss": 0.005132, "loss_mean50": 0.006119, "lr": 2.25e-05, "grad_norm": 0.0439, "samples": 28800, "wall_s": 1760.0, "cuda_gb": 16.86}
{"step": 920, "loss": 0.00592, "loss_mean50": 0.006175, "lr": 2.3e-05, "grad_norm": 0.0533, "samples": 29440, "wall_s": 1799.0, "cuda_gb": 16.86}
{"step": 940, "loss": 0.006757, "loss_mean50": 0.006348, "lr": 2.35e-05, "grad_norm": 0.0596, "samples": 30080, "wall_s": 1837.9, "cuda_gb": 16.86}
{"step": 960, "loss": 0.004966, "loss_mean50": 0.006215, "lr": 2.4e-05, "grad_norm": 0.0394, "samples": 30720, "wall_s": 1876.8, "cuda_gb": 16.86}
{"step": 980, "loss": 0.004428, "loss_mean50": 0.006169, "lr": 2.4500000000000003e-05, "grad_norm": 0.0906, "samples": 31360, "wall_s": 1915.8, "cuda_gb": 16.86}
{"step": 1000, "loss": 0.006727, "loss_mean50": 0.006168, "lr": 2.5e-05, "grad_norm": 0.0932, "samples": 32000, "wall_s": 1954.6, "cuda_gb": 16.86}
{"step": 1000, "val_flow_mse": 0.004808}
{"step": 1020, "loss": 0.006983, "loss_mean50": 0.00575, "lr": 2.499989774778022e-05, "grad_norm": 0.0519, "samples": 32640, "wall_s": 2022.3, "cuda_gb": 16.86}
{"step": 1040, "loss": 0.006411, "loss_mean50": 0.005268, "lr": 2.4999569183187692e-05, "grad_norm": 0.0588, "samples": 33280, "wall_s": 2061.2, "cuda_gb": 16.86}
{"step": 1060, "loss": 0.00476, "loss_mean50": 0.005661, "lr": 2.4999014029592804e-05, "grad_norm": 0.047, "samples": 33920, "wall_s": 2100.0, "cuda_gb": 16.86}
{"step": 1080, "loss": 0.003591, "loss_mean50": 0.005869, "lr": 2.4998232298177463e-05, "grad_norm": 0.0349, "samples": 34560, "wall_s": 2138.9, "cuda_gb": 16.86}
{"step": 1100, "loss": 0.006924, "loss_mean50": 0.005714, "lr": 2.499722400468732e-05, "grad_norm": 0.0517, "samples": 35200, "wall_s": 2177.8, "cuda_gb": 16.86}
{"step": 1120, "loss": 0.004813, "loss_mean50": 0.005276, "lr": 2.4995989169431427e-05, "grad_norm": 0.0754, "samples": 35840, "wall_s": 2216.7, "cuda_gb": 16.86}
{"step": 1140, "loss": 0.005158, "loss_mean50": 0.005309, "lr": 2.499452781728187e-05, "grad_norm": 0.0423, "samples": 36480, "wall_s": 2255.6, "cuda_gb": 16.86}
{"step": 1160, "loss": 0.005936, "loss_mean50": 0.005405, "lr": 2.4992839977673223e-05, "grad_norm": 0.0444, "samples": 37120, "wall_s": 2294.5, "cuda_gb": 16.86}
{"step": 1180, "loss": 0.004724, "loss_mean50": 0.005381, "lr": 2.4990925684601972e-05, "grad_norm": 0.0484, "samples": 37760, "wall_s": 2333.5, "cuda_gb": 16.86}
{"step": 1200, "loss": 0.005912, "loss_mean50": 0.005546, "lr": 2.4988784976625844e-05, "grad_norm": 0.0422, "samples": 38400, "wall_s": 2372.4, "cuda_gb": 16.86}
{"step": 1220, "loss": 0.004399, "loss_mean50": 0.005598, "lr": 2.4986417896863013e-05, "grad_norm": 0.037, "samples": 39040, "wall_s": 2411.3, "cuda_gb": 16.86}
{"step": 1240, "loss": 0.004997, "loss_mean50": 0.00511, "lr": 2.4983824492991236e-05, "grad_norm": 0.0432, "samples": 39680, "wall_s": 2450.4, "cuda_gb": 16.86}
{"step": 1260, "loss": 0.007442, "loss_mean50": 0.00468, "lr": 2.4981004817246896e-05, "grad_norm": 0.068, "samples": 40320, "wall_s": 2489.3, "cuda_gb": 16.86}
{"step": 1280, "loss": 0.004612, "loss_mean50": 0.004631, "lr": 2.497795892642395e-05, "grad_norm": 0.0358, "samples": 40960, "wall_s": 2528.3, "cuda_gb": 16.86}
{"step": 1300, "loss": 0.006622, "loss_mean50": 0.004774, "lr": 2.4974686881872776e-05, "grad_norm": 0.0571, "samples": 41600, "wall_s": 2567.3, "cuda_gb": 16.86}
{"step": 1320, "loss": 0.004074, "loss_mean50": 0.004924, "lr": 2.4971188749498953e-05, "grad_norm": 0.036, "samples": 42240, "wall_s": 2606.1, "cuda_gb": 16.86}
{"step": 1340, "loss": 0.003178, "loss_mean50": 0.005103, "lr": 2.4967464599761908e-05, "grad_norm": 0.0315, "samples": 42880, "wall_s": 2645.2, "cuda_gb": 16.86}
{"step": 1360, "loss": 0.004151, "loss_mean50": 0.004992, "lr": 2.4963514507673534e-05, "grad_norm": 0.0404, "samples": 43520, "wall_s": 2684.1, "cuda_gb": 16.86}
{"step": 1380, "loss": 0.005191, "loss_mean50": 0.005357, "lr": 2.4959338552796643e-05, "grad_norm": 0.035, "samples": 44160, "wall_s": 2723.1, "cuda_gb": 16.86}
{"step": 1400, "loss": 0.006314, "loss_mean50": 0.005251, "lr": 2.495493681924339e-05, "grad_norm": 0.0478, "samples": 44800, "wall_s": 2762.1, "cuda_gb": 16.86}
{"step": 1420, "loss": 0.003524, "loss_mean50": 0.005111, "lr": 2.4950309395673564e-05, "grad_norm": 0.0364, "samples": 45440, "wall_s": 2801.1, "cuda_gb": 16.86}
{"step": 1440, "loss": 0.004054, "loss_mean50": 0.004828, "lr": 2.4945456375292797e-05, "grad_norm": 0.0408, "samples": 46080, "wall_s": 2840.1, "cuda_gb": 16.86}
{"step": 1460, "loss": 0.006746, "loss_mean50": 0.005291, "lr": 2.494037785585071e-05, "grad_norm": 0.042, "samples": 46720, "wall_s": 2879.1, "cuda_gb": 16.86}
{"step": 1480, "loss": 0.006359, "loss_mean50": 0.005874, "lr": 2.493507393963892e-05, "grad_norm": 0.0564, "samples": 47360, "wall_s": 2918.0, "cuda_gb": 16.86}
{"step": 1500, "loss": 0.007242, "loss_mean50": 0.005918, "lr": 2.492954473348899e-05, "grad_norm": 0.0629, "samples": 48000, "wall_s": 2957.1, "cuda_gb": 16.86}
{"step": 1520, "loss": 0.007587, "loss_mean50": 0.005328, "lr": 2.4923790348770284e-05, "grad_norm": 0.0593, "samples": 48640, "wall_s": 2996.0, "cuda_gb": 16.86}
{"step": 1540, "loss": 0.006711, "loss_mean50": 0.005051, "lr": 2.4917810901387707e-05, "grad_norm": 0.0445, "samples": 49280, "wall_s": 3035.0, "cuda_gb": 16.86}
{"step": 1560, "loss": 0.002796, "loss_mean50": 0.004634, "lr": 2.4911606511779384e-05, "grad_norm": 0.0293, "samples": 49920, "wall_s": 3073.9, "cuda_gb": 16.86}
{"step": 1580, "loss": 0.003637, "loss_mean50": 0.004693, "lr": 2.4905177304914233e-05, "grad_norm": 0.0443, "samples": 50560, "wall_s": 3112.8, "cuda_gb": 16.86}
{"step": 1600, "loss": 0.005552, "loss_mean50": 0.004835, "lr": 2.4898523410289436e-05, "grad_norm": 0.0417, "samples": 51200, "wall_s": 3151.7, "cuda_gb": 16.86}
{"step": 1620, "loss": 0.003713, "loss_mean50": 0.005046, "lr": 2.4891644961927848e-05, "grad_norm": 0.0385, "samples": 51840, "wall_s": 3190.6, "cuda_gb": 16.86}
{"step": 1640, "loss": 0.003294, "loss_mean50": 0.004496, "lr": 2.488454209837529e-05, "grad_norm": 0.0365, "samples": 52480, "wall_s": 3229.5, "cuda_gb": 16.86}
{"step": 1660, "loss": 0.004554, "loss_mean50": 0.004636, "lr": 2.4877214962697744e-05, "grad_norm": 0.05, "samples": 53120, "wall_s": 3268.5, "cuda_gb": 16.86}
{"step": 1680, "loss": 0.002895, "loss_mean50": 0.004519, "lr": 2.4869663702478503e-05, "grad_norm": 0.0264, "samples": 53760, "wall_s": 3307.5, "cuda_gb": 16.86}
{"step": 1700, "loss": 0.003596, "loss_mean50": 0.00438, "lr": 2.4861888469815163e-05, "grad_norm": 0.0411, "samples": 54400, "wall_s": 3346.5, "cuda_gb": 16.86}
{"step": 1720, "loss": 0.002865, "loss_mean50": 0.004323, "lr": 2.4853889421316596e-05, "grad_norm": 0.0217, "samples": 55040, "wall_s": 3385.5, "cuda_gb": 16.86}
{"step": 1740, "loss": 0.004097, "loss_mean50": 0.004353, "lr": 2.484566671809975e-05, "grad_norm": 0.0612, "samples": 55680, "wall_s": 3424.4, "cuda_gb": 16.86}
{"step": 1760, "loss": 0.002348, "loss_mean50": 0.004253, "lr": 2.4837220525786454e-05, "grad_norm": 0.0289, "samples": 56320, "wall_s": 3463.5, "cuda_gb": 16.86}
{"step": 1780, "loss": 0.004496, "loss_mean50": 0.004597, "lr": 2.4828551014500046e-05, "grad_norm": 0.0441, "samples": 56960, "wall_s": 3502.4, "cuda_gb": 16.86}
{"step": 1800, "loss": 0.006158, "loss_mean50": 0.004896, "lr": 2.4819658358861965e-05, "grad_norm": 0.0719, "samples": 57600, "wall_s": 3541.3, "cuda_gb": 16.86}
{"step": 1820, "loss": 0.006815, "loss_mean50": 0.004839, "lr": 2.4810542737988217e-05, "grad_norm": 0.0337, "samples": 58240, "wall_s": 3580.2, "cuda_gb": 16.86}
{"step": 1840, "loss": 0.002807, "loss_mean50": 0.004678, "lr": 2.480120433548579e-05, "grad_norm": 0.0358, "samples": 58880, "wall_s": 3619.1, "cuda_gb": 16.86}
{"step": 1860, "loss": 0.002973, "loss_mean50": 0.004284, "lr": 2.479164333944893e-05, "grad_norm": 0.0402, "samples": 59520, "wall_s": 3658.1, "cuda_gb": 16.86}
{"step": 1880, "loss": 0.005714, "loss_mean50": 0.004381, "lr": 2.4781859942455375e-05, "grad_norm": 0.0623, "samples": 60160, "wall_s": 3697.0, "cuda_gb": 16.86}
{"step": 1900, "loss": 0.004265, "loss_mean50": 0.004521, "lr": 2.4771854341562465e-05, "grad_norm": 0.0331, "samples": 60800, "wall_s": 3735.9, "cuda_gb": 16.86}
{"step": 1920, "loss": 0.007087, "loss_mean50": 0.004297, "lr": 2.476162673830317e-05, "grad_norm": 0.051, "samples": 61440, "wall_s": 3774.9, "cuda_gb": 16.86}
{"step": 1940, "loss": 0.00238, "loss_mean50": 0.004313, "lr": 2.475117733868204e-05, "grad_norm": 0.0255, "samples": 62080, "wall_s": 3813.8, "cuda_gb": 16.86}
{"step": 1960, "loss": 0.004447, "loss_mean50": 0.004386, "lr": 2.474050635317105e-05, "grad_norm": 0.0269, "samples": 62720, "wall_s": 3852.7, "cuda_gb": 16.86}
{"step": 1980, "loss": 0.003113, "loss_mean50": 0.004336, "lr": 2.472961399670536e-05, "grad_norm": 0.0398, "samples": 63360, "wall_s": 3891.6, "cuda_gb": 16.86}
{"step": 2000, "loss": 0.006345, "loss_mean50": 0.004456, "lr": 2.471850048867899e-05, "grad_norm": 0.0476, "samples": 64000, "wall_s": 3930.6, "cuda_gb": 16.86}
{"step": 2000, "val_flow_mse": 0.004137}
{"step": 2020, "loss": 0.003463, "loss_mean50": 0.004279, "lr": 2.470716605294039e-05, "grad_norm": 0.0324, "samples": 64640, "wall_s": 3998.4, "cuda_gb": 16.86}
{"step": 2040, "loss": 0.003255, "loss_mean50": 0.00432, "lr": 2.4695610917787952e-05, "grad_norm": 0.0313, "samples": 65280, "wall_s": 4037.3, "cuda_gb": 16.86}
{"step": 2060, "loss": 0.005121, "loss_mean50": 0.004472, "lr": 2.468383531596539e-05, "grad_norm": 0.0438, "samples": 65920, "wall_s": 4076.3, "cuda_gb": 16.86}
{"step": 2080, "loss": 0.003028, "loss_mean50": 0.004216, "lr": 2.467183948465707e-05, "grad_norm": 0.0263, "samples": 66560, "wall_s": 4115.3, "cuda_gb": 16.86}
{"step": 2100, "loss": 0.004821, "loss_mean50": 0.004236, "lr": 2.4659623665483212e-05, "grad_norm": 0.045, "samples": 67200, "wall_s": 4154.2, "cuda_gb": 16.86}
{"step": 2120, "loss": 0.004735, "loss_mean50": 0.004049, "lr": 2.4647188104495043e-05, "grad_norm": 0.0754, "samples": 67840, "wall_s": 4193.3, "cuda_gb": 16.86}
{"step": 2140, "loss": 0.004925, "loss_mean50": 0.004118, "lr": 2.463453305216983e-05, "grad_norm": 0.0531, "samples": 68480, "wall_s": 4232.1, "cuda_gb": 16.86}
{"step": 2160, "loss": 0.004791, "loss_mean50": 0.004082, "lr": 2.4621658763405842e-05, "grad_norm": 0.0338, "samples": 69120, "wall_s": 4271.0, "cuda_gb": 16.86}
{"step": 2180, "loss": 0.007092, "loss_mean50": 0.004024, "lr": 2.4608565497517207e-05, "grad_norm": 0.0974, "samples": 69760, "wall_s": 4309.8, "cuda_gb": 16.86}
{"step": 2200, "loss": 0.006453, "loss_mean50": 0.003974, "lr": 2.4595253518228692e-05, "grad_norm": 0.0467, "samples": 70400, "wall_s": 4348.7, "cuda_gb": 16.86}
{"step": 2220, "loss": 0.003187, "loss_mean50": 0.004184, "lr": 2.4581723093670407e-05, "grad_norm": 0.0467, "samples": 71040, "wall_s": 4387.6, "cuda_gb": 16.86}
{"step": 2240, "loss": 0.004276, "loss_mean50": 0.003981, "lr": 2.456797449637237e-05, "grad_norm": 0.0376, "samples": 71680, "wall_s": 4426.5, "cuda_gb": 16.86}
{"step": 2260, "loss": 0.00413, "loss_mean50": 0.00401, "lr": 2.455400800325905e-05, "grad_norm": 0.0531, "samples": 72320, "wall_s": 4465.4, "cuda_gb": 16.86}
{"step": 2280, "loss": 0.003081, "loss_mean50": 0.004325, "lr": 2.4539823895643783e-05, "grad_norm": 0.0351, "samples": 72960, "wall_s": 4504.4, "cuda_gb": 16.86}
{"step": 2300, "loss": 0.003893, "loss_mean50": 0.004124, "lr": 2.4525422459223077e-05, "grad_norm": 0.0378, "samples": 73600, "wall_s": 4543.4, "cuda_gb": 16.86}
{"step": 2320, "loss": 0.001856, "loss_mean50": 0.003831, "lr": 2.451080398407091e-05, "grad_norm": 0.0219, "samples": 74240, "wall_s": 4582.3, "cuda_gb": 16.86}
{"step": 2340, "loss": 0.003374, "loss_mean50": 0.00376, "lr": 2.4495968764632826e-05, "grad_norm": 0.028, "samples": 74880, "wall_s": 4621.3, "cuda_gb": 16.86}
{"step": 2360, "loss": 0.00352, "loss_mean50": 0.003898, "lr": 2.4480917099720065e-05, "grad_norm": 0.0453, "samples": 75520, "wall_s": 4660.1, "cuda_gb": 16.86}
{"step": 2380, "loss": 0.003663, "loss_mean50": 0.00433, "lr": 2.4465649292503495e-05, "grad_norm": 0.0409, "samples": 76160, "wall_s": 4699.1, "cuda_gb": 16.86}
{"step": 2400, "loss": 0.004063, "loss_mean50": 0.004304, "lr": 2.4450165650507544e-05, "grad_norm": 0.0509, "samples": 76800, "wall_s": 4738.0, "cuda_gb": 16.86}
{"step": 2420, "loss": 0.002751, "loss_mean50": 0.004081, "lr": 2.443446648560396e-05, "grad_norm": 0.0331, "samples": 77440, "wall_s": 4777.0, "cuda_gb": 16.86}
{"step": 2440, "loss": 0.002822, "loss_mean50": 0.003976, "lr": 2.4418552114005585e-05, "grad_norm": 0.0216, "samples": 78080, "wall_s": 4815.8, "cuda_gb": 16.86}
{"step": 2460, "loss": 0.004551, "loss_mean50": 0.004007, "lr": 2.4402422856259944e-05, "grad_norm": 0.0588, "samples": 78720, "wall_s": 4854.7, "cuda_gb": 16.86}
{"step": 2480, "loss": 0.003567, "loss_mean50": 0.003907, "lr": 2.43860790372428e-05, "grad_norm": 0.0312, "samples": 79360, "wall_s": 4893.6, "cuda_gb": 16.86}
{"step": 2500, "loss": 0.004749, "loss_mean50": 0.003761, "lr": 2.4369520986151623e-05, "grad_norm": 0.038, "samples": 80000, "wall_s": 4932.5, "cuda_gb": 16.86}
{"step": 2520, "loss": 0.003018, "loss_mean50": 0.00407, "lr": 2.435274903649894e-05, "grad_norm": 0.0247, "samples": 80640, "wall_s": 5001.0, "cuda_gb": 16.86}
{"step": 2540, "loss": 0.005485, "loss_mean50": 0.004323, "lr": 2.4335763526105632e-05, "grad_norm": 0.05, "samples": 81280, "wall_s": 5039.9, "cuda_gb": 16.86}
{"step": 2560, "loss": 0.002315, "loss_mean50": 0.004158, "lr": 2.4318564797094113e-05, "grad_norm": 0.0311, "samples": 81920, "wall_s": 5078.8, "cuda_gb": 16.86}
{"step": 2580, "loss": 0.003313, "loss_mean50": 0.004052, "lr": 2.4301153195881467e-05, "grad_norm": 0.0303, "samples": 82560, "wall_s": 5117.8, "cuda_gb": 16.86}
{"step": 2600, "loss": 0.003855, "loss_mean50": 0.00417, "lr": 2.4283529073172444e-05, "grad_norm": 0.0438, "samples": 83200, "wall_s": 5156.8, "cuda_gb": 16.86}
{"step": 2620, "loss": 0.003848, "loss_mean50": 0.003911, "lr": 2.42656927839524e-05, "grad_norm": 0.0332, "samples": 83840, "wall_s": 5195.8, "cuda_gb": 16.86}
{"step": 2640, "loss": 0.002829, "loss_mean50": 0.003706, "lr": 2.4247644687480155e-05, "grad_norm": 0.0308, "samples": 84480, "wall_s": 5234.8, "cuda_gb": 16.86}
{"step": 2660, "loss": 0.00579, "loss_mean50": 0.003754, "lr": 2.4229385147280765e-05, "grad_norm": 0.0471, "samples": 85120, "wall_s": 5273.7, "cuda_gb": 16.86}
{"step": 2680, "loss": 0.003081, "loss_mean50": 0.003997, "lr": 2.4210914531138177e-05, "grad_norm": 0.0395, "samples": 85760, "wall_s": 5312.7, "cuda_gb": 16.86}
{"step": 2700, "loss": 0.002611, "loss_mean50": 0.004, "lr": 2.419223321108783e-05, "grad_norm": 0.0292, "samples": 86400, "wall_s": 5351.7, "cuda_gb": 16.86}
{"step": 2720, "loss": 0.004724, "loss_mean50": 0.004059, "lr": 2.4173341563409186e-05, "grad_norm": 0.0487, "samples": 87040, "wall_s": 5390.7, "cuda_gb": 16.86}
{"step": 2740, "loss": 0.003303, "loss_mean50": 0.003858, "lr": 2.41542399686181e-05, "grad_norm": 0.0356, "samples": 87680, "wall_s": 5429.6, "cuda_gb": 16.86}
{"step": 2760, "loss": 0.006121, "loss_mean50": 0.003786, "lr": 2.413492881145921e-05, "grad_norm": 0.0307, "samples": 88320, "wall_s": 5468.6, "cuda_gb": 16.86}
{"step": 2780, "loss": 0.004206, "loss_mean50": 0.003802, "lr": 2.4115408480898146e-05, "grad_norm": 0.0606, "samples": 88960, "wall_s": 5507.6, "cuda_gb": 16.86}
{"step": 2800, "loss": 0.002875, "loss_mean50": 0.003911, "lr": 2.4095679370113723e-05, "grad_norm": 0.0323, "samples": 89600, "wall_s": 5546.4, "cuda_gb": 16.86}
{"step": 2820, "loss": 0.003914, "loss_mean50": 0.003769, "lr": 2.407574187649e-05, "grad_norm": 0.0377, "samples": 90240, "wall_s": 5585.4, "cuda_gb": 16.86}
{"step": 2840, "loss": 0.001896, "loss_mean50": 0.00358, "lr": 2.4055596401608298e-05, "grad_norm": 0.0218, "samples": 90880, "wall_s": 5624.3, "cuda_gb": 16.86}
{"step": 2860, "loss": 0.002692, "loss_mean50": 0.003785, "lr": 2.4035243351239095e-05, "grad_norm": 0.0275, "samples": 91520, "wall_s": 5663.2, "cuda_gb": 16.86}
{"step": 2880, "loss": 0.002454, "loss_mean50": 0.003736, "lr": 2.4014683135333848e-05, "grad_norm": 0.0297, "samples": 92160, "wall_s": 5702.2, "cuda_gb": 16.86}
{"step": 2900, "loss": 0.003523, "loss_mean50": 0.003774, "lr": 2.3993916168016762e-05, "grad_norm": 0.0313, "samples": 92800, "wall_s": 5741.1, "cuda_gb": 16.86}
{"step": 2920, "loss": 0.004505, "loss_mean50": 0.003966, "lr": 2.3972942867576424e-05, "grad_norm": 0.0342, "samples": 93440, "wall_s": 5780.1, "cuda_gb": 16.86}
{"step": 2940, "loss": 0.003473, "loss_mean50": 0.004066, "lr": 2.3951763656457387e-05, "grad_norm": 0.0303, "samples": 94080, "wall_s": 5819.0, "cuda_gb": 16.86}
{"step": 2960, "loss": 0.009001, "loss_mean50": 0.004126, "lr": 2.3930378961251658e-05, "grad_norm": 0.0492, "samples": 94720, "wall_s": 5857.9, "cuda_gb": 16.86}
{"step": 2980, "loss": 0.005774, "loss_mean50": 0.003831, "lr": 2.3908789212690112e-05, "grad_norm": 0.0611, "samples": 95360, "wall_s": 5896.9, "cuda_gb": 16.86}
{"step": 3000, "loss": 0.003047, "loss_mean50": 0.003742, "lr": 2.3886994845633805e-05, "grad_norm": 0.0347, "samples": 96000, "wall_s": 5935.8, "cuda_gb": 16.86}
{"step": 3000, "val_flow_mse": 0.003363}
{"step": 3020, "loss": 0.002911, "loss_mean50": 0.003584, "lr": 2.3864996299065235e-05, "grad_norm": 0.0337, "samples": 96640, "wall_s": 6003.5, "cuda_gb": 16.86}
{"step": 3040, "loss": 0.002706, "loss_mean50": 0.003498, "lr": 2.384279401607947e-05, "grad_norm": 0.0287, "samples": 97280, "wall_s": 6042.5, "cuda_gb": 16.86}
{"step": 3060, "loss": 0.003773, "loss_mean50": 0.003621, "lr": 2.3820388443875256e-05, "grad_norm": 0.038, "samples": 97920, "wall_s": 6081.4, "cuda_gb": 16.86}
{"step": 3080, "loss": 0.002516, "loss_mean50": 0.003792, "lr": 2.379778003374598e-05, "grad_norm": 0.0348, "samples": 98560, "wall_s": 6120.4, "cuda_gb": 16.86}
{"step": 3100, "loss": 0.003004, "loss_mean50": 0.003991, "lr": 2.3774969241070605e-05, "grad_norm": 0.0435, "samples": 99200, "wall_s": 6159.3, "cuda_gb": 16.86}
{"step": 3120, "loss": 0.005384, "loss_mean50": 0.003789, "lr": 2.375195652530448e-05, "grad_norm": 0.0386, "samples": 99840, "wall_s": 6198.3, "cuda_gb": 16.86}
{"step": 3140, "loss": 0.003139, "loss_mean50": 0.003747, "lr": 2.3728742349970097e-05, "grad_norm": 0.0394, "samples": 100480, "wall_s": 6237.3, "cuda_gb": 16.86}
{"step": 3160, "loss": 0.002419, "loss_mean50": 0.003615, "lr": 2.3705327182647743e-05, "grad_norm": 0.0288, "samples": 101120, "wall_s": 6276.3, "cuda_gb": 16.86}
{"step": 3180, "loss": 0.002553, "loss_mean50": 0.003527, "lr": 2.3681711494966095e-05, "grad_norm": 0.0319, "samples": 101760, "wall_s": 6315.3, "cuda_gb": 16.86}
{"step": 3200, "loss": 0.002725, "loss_mean50": 0.003555, "lr": 2.365789576259271e-05, "grad_norm": 0.0226, "samples": 102400, "wall_s": 6354.2, "cuda_gb": 16.86}
{"step": 3220, "loss": 0.005478, "loss_mean50": 0.003551, "lr": 2.363388046522445e-05, "grad_norm": 0.0678, "samples": 103040, "wall_s": 6393.2, "cuda_gb": 16.86}
{"step": 3240, "loss": 0.003961, "loss_mean50": 0.003695, "lr": 2.3609666086577818e-05, "grad_norm": 0.0285, "samples": 103680, "wall_s": 6432.1, "cuda_gb": 16.86}
{"step": 3260, "loss": 0.005625, "loss_mean50": 0.0038, "lr": 2.358525311437922e-05, "grad_norm": 0.0408, "samples": 104320, "wall_s": 6471.0, "cuda_gb": 16.86}
{"step": 3280, "loss": 0.003267, "loss_mean50": 0.003848, "lr": 2.356064204035513e-05, "grad_norm": 0.0362, "samples": 104960, "wall_s": 6509.8, "cuda_gb": 16.86}
{"step": 3300, "loss": 0.00503, "loss_mean50": 0.003578, "lr": 2.3535833360222197e-05, "grad_norm": 0.0352, "samples": 105600, "wall_s": 6548.7, "cuda_gb": 16.86}
{"step": 3320, "loss": 0.003468, "loss_mean50": 0.003621, "lr": 2.351082757367725e-05, "grad_norm": 0.026, "samples": 106240, "wall_s": 6587.7, "cuda_gb": 16.86}
{"step": 3340, "loss": 0.001705, "loss_mean50": 0.003746, "lr": 2.3485625184387237e-05, "grad_norm": 0.0219, "samples": 106880, "wall_s": 6626.5, "cuda_gb": 16.86}
{"step": 3360, "loss": 0.002321, "loss_mean50": 0.003612, "lr": 2.3460226699979094e-05, "grad_norm": 0.0278, "samples": 107520, "wall_s": 6665.4, "cuda_gb": 16.86}
{"step": 3380, "loss": 0.001977, "loss_mean50": 0.003457, "lr": 2.3434632632029495e-05, "grad_norm": 0.0322, "samples": 108160, "wall_s": 6704.4, "cuda_gb": 16.86}
{"step": 3400, "loss": 0.003019, "loss_mean50": 0.003711, "lr": 2.340884349605457e-05, "grad_norm": 0.0232, "samples": 108800, "wall_s": 6743.4, "cuda_gb": 16.86}
{"step": 3420, "loss": 0.002972, "loss_mean50": 0.003586, "lr": 2.3382859811499506e-05, "grad_norm": 0.0289, "samples": 109440, "wall_s": 6782.4, "cuda_gb": 16.86}
{"step": 3440, "loss": 0.003089, "loss_mean50": 0.003338, "lr": 2.3356682101728088e-05, "grad_norm": 0.0244, "samples": 110080, "wall_s": 6821.4, "cuda_gb": 16.86}
{"step": 3460, "loss": 0.002596, "loss_mean50": 0.003273, "lr": 2.3330310894012176e-05, "grad_norm": 0.0265, "samples": 110720, "wall_s": 6860.3, "cuda_gb": 16.86}
{"step": 3480, "loss": 0.001711, "loss_mean50": 0.003481, "lr": 2.3303746719521045e-05, "grad_norm": 0.0251, "samples": 111360, "wall_s": 6899.3, "cuda_gb": 16.86}
{"step": 3500, "loss": 0.00303, "loss_mean50": 0.00341, "lr": 2.3276990113310735e-05, "grad_norm": 0.0278, "samples": 112000, "wall_s": 6938.1, "cuda_gb": 16.86}
{"step": 3520, "loss": 0.003366, "loss_mean50": 0.003428, "lr": 2.3250041614313225e-05, "grad_norm": 0.0505, "samples": 112640, "wall_s": 6977.0, "cuda_gb": 16.86}
{"step": 3540, "loss": 0.002918, "loss_mean50": 0.00347, "lr": 2.3222901765325615e-05, "grad_norm": 0.028, "samples": 113280, "wall_s": 7016.0, "cuda_gb": 16.86}
{"step": 3560, "loss": 0.004699, "loss_mean50": 0.003531, "lr": 2.3195571112999185e-05, "grad_norm": 0.0425, "samples": 113920, "wall_s": 7055.1, "cuda_gb": 16.86}
{"step": 3580, "loss": 0.00173, "loss_mean50": 0.003288, "lr": 2.3168050207828366e-05, "grad_norm": 0.0471, "samples": 114560, "wall_s": 7094.0, "cuda_gb": 16.86}
{"step": 3600, "loss": 0.002528, "loss_mean50": 0.003368, "lr": 2.3140339604139666e-05, "grad_norm": 0.0383, "samples": 115200, "wall_s": 7132.9, "cuda_gb": 16.86}
{"step": 3620, "loss": 0.002792, "loss_mean50": 0.003503, "lr": 2.3112439860080513e-05, "grad_norm": 0.0298, "samples": 115840, "wall_s": 7171.9, "cuda_gb": 16.86}
{"step": 3640, "loss": 0.002118, "loss_mean50": 0.003672, "lr": 2.308435153760799e-05, "grad_norm": 0.0249, "samples": 116480, "wall_s": 7211.0, "cuda_gb": 16.86}
{"step": 3660, "loss": 0.005279, "loss_mean50": 0.003693, "lr": 2.305607520247755e-05, "grad_norm": 0.035, "samples": 117120, "wall_s": 7249.9, "cuda_gb": 16.86}
{"step": 3680, "loss": 0.002461, "loss_mean50": 0.003563, "lr": 2.302761142423158e-05, "grad_norm": 0.0263, "samples": 117760, "wall_s": 7288.8, "cuda_gb": 16.86}
{"step": 3700, "loss": 0.002473, "loss_mean50": 0.003566, "lr": 2.2998960776187957e-05, "grad_norm": 0.0243, "samples": 118400, "wall_s": 7327.8, "cuda_gb": 16.86}
{"step": 3720, "loss": 0.001823, "loss_mean50": 0.003311, "lr": 2.2970123835428493e-05, "grad_norm": 0.0212, "samples": 119040, "wall_s": 7366.8, "cuda_gb": 16.86}
{"step": 3740, "loss": 0.003177, "loss_mean50": 0.003135, "lr": 2.2941101182787307e-05, "grad_norm": 0.0256, "samples": 119680, "wall_s": 7405.8, "cuda_gb": 16.86}
{"step": 3760, "loss": 0.004733, "loss_mean50": 0.003227, "lr": 2.2911893402839136e-05, "grad_norm": 0.0414, "samples": 120320, "wall_s": 7444.8, "cuda_gb": 16.86}
{"step": 3780, "loss": 0.00418, "loss_mean50": 0.003378, "lr": 2.288250108388755e-05, "grad_norm": 0.0523, "samples": 120960, "wall_s": 7483.8, "cuda_gb": 16.86}
{"step": 3800, "loss": 0.003053, "loss_mean50": 0.003406, "lr": 2.2852924817953105e-05, "grad_norm": 0.0268, "samples": 121600, "wall_s": 7522.7, "cuda_gb": 16.86}
{"step": 3820, "loss": 0.00237, "loss_mean50": 0.003347, "lr": 2.2823165200761428e-05, "grad_norm": 0.0229, "samples": 122240, "wall_s": 7561.6, "cuda_gb": 16.86}
{"step": 3840, "loss": 0.004166, "loss_mean50": 0.003431, "lr": 2.27932228317312e-05, "grad_norm": 0.0563, "samples": 122880, "wall_s": 7600.6, "cuda_gb": 16.86}
{"step": 3860, "loss": 0.002353, "loss_mean50": 0.003344, "lr": 2.2763098313962093e-05, "grad_norm": 0.0302, "samples": 123520, "wall_s": 7639.4, "cuda_gb": 16.86}
{"step": 3880, "loss": 0.003833, "loss_mean50": 0.003402, "lr": 2.2732792254222633e-05, "grad_norm": 0.0322, "samples": 124160, "wall_s": 7678.3, "cuda_gb": 16.86}
{"step": 3900, "loss": 0.001806, "loss_mean50": 0.003357, "lr": 2.2702305262937958e-05, "grad_norm": 0.0259, "samples": 124800, "wall_s": 7717.3, "cuda_gb": 16.86}
{"step": 3920, "loss": 0.002572, "loss_mean50": 0.003392, "lr": 2.2671637954177526e-05, "grad_norm": 0.0224, "samples": 125440, "wall_s": 7756.2, "cuda_gb": 16.86}
{"step": 3940, "loss": 0.00376, "loss_mean50": 0.003482, "lr": 2.2640790945642772e-05, "grad_norm": 0.0388, "samples": 126080, "wall_s": 7795.2, "cuda_gb": 16.86}
{"step": 3960, "loss": 0.003453, "loss_mean50": 0.003963, "lr": 2.2609764858654632e-05, "grad_norm": 0.06, "samples": 126720, "wall_s": 7834.2, "cuda_gb": 16.86}
{"step": 3980, "loss": 0.001369, "loss_mean50": 0.003745, "lr": 2.2578560318141043e-05, "grad_norm": 0.021, "samples": 127360, "wall_s": 7873.0, "cuda_gb": 16.86}
{"step": 4000, "loss": 0.003773, "loss_mean50": 0.003288, "lr": 2.254717795262436e-05, "grad_norm": 0.039, "samples": 128000, "wall_s": 7912.0, "cuda_gb": 16.86}
{"step": 4000, "val_flow_mse": 0.002934}
{"step": 4020, "loss": 0.003889, "loss_mean50": 0.003228, "lr": 2.2515618394208694e-05, "grad_norm": 0.0397, "samples": 128640, "wall_s": 7979.8, "cuda_gb": 16.86}
{"step": 4040, "loss": 0.002547, "loss_mean50": 0.003292, "lr": 2.248388227856717e-05, "grad_norm": 0.0256, "samples": 129280, "wall_s": 8018.8, "cuda_gb": 16.86}
{"step": 4060, "loss": 0.002079, "loss_mean50": 0.003307, "lr": 2.2451970244929137e-05, "grad_norm": 0.0325, "samples": 129920, "wall_s": 8057.7, "cuda_gb": 16.86}
{"step": 4080, "loss": 0.002484, "loss_mean50": 0.003183, "lr": 2.2419882936067294e-05, "grad_norm": 0.0408, "samples": 130560, "wall_s": 8096.7, "cuda_gb": 16.86}
{"step": 4100, "loss": 0.002137, "loss_mean50": 0.002931, "lr": 2.238762099828472e-05, "grad_norm": 0.0325, "samples": 131200, "wall_s": 8135.7, "cuda_gb": 16.86}
{"step": 4120, "loss": 0.002939, "loss_mean50": 0.003165, "lr": 2.235518508140189e-05, "grad_norm": 0.0325, "samples": 131840, "wall_s": 8174.7, "cuda_gb": 16.86}
{"step": 4140, "loss": 0.002078, "loss_mean50": 0.003294, "lr": 2.232257583874355e-05, "grad_norm": 0.0297, "samples": 132480, "wall_s": 8213.6, "cuda_gb": 16.86}
{"step": 4160, "loss": 0.003074, "loss_mean50": 0.003587, "lr": 2.2289793927125603e-05, "grad_norm": 0.0458, "samples": 133120, "wall_s": 8252.5, "cuda_gb": 16.86}
{"step": 4180, "loss": 0.00386, "loss_mean50": 0.003477, "lr": 2.225684000684182e-05, "grad_norm": 0.049, "samples": 133760, "wall_s": 8291.4, "cuda_gb": 16.86}
{"step": 4200, "loss": 0.003808, "loss_mean50": 0.003476, "lr": 2.22237147416506e-05, "grad_norm": 0.0519, "samples": 134400, "wall_s": 8330.4, "cuda_gb": 16.86}
{"step": 4220, "loss": 0.002709, "loss_mean50": 0.00331, "lr": 2.2190418798761562e-05, "grad_norm": 0.0214, "samples": 135040, "wall_s": 8369.2, "cuda_gb": 16.86}
{"step": 4240, "loss": 0.001883, "loss_mean50": 0.003211, "lr": 2.2156952848822125e-05, "grad_norm": 0.0247, "samples": 135680, "wall_s": 8408.2, "cuda_gb": 16.86}
{"step": 4260, "loss": 0.003442, "loss_mean50": 0.003253, "lr": 2.212331756590397e-05, "grad_norm": 0.0795, "samples": 136320, "wall_s": 8447.2, "cuda_gb": 16.86}
{"step": 4280, "loss": 0.002729, "loss_mean50": 0.003019, "lr": 2.208951362748953e-05, "grad_norm": 0.0265, "samples": 136960, "wall_s": 8486.1, "cuda_gb": 16.86}
{"step": 4300, "loss": 0.001946, "loss_mean50": 0.002813, "lr": 2.2055541714458252e-05, "grad_norm": 0.0413, "samples": 137600, "wall_s": 8525.0, "cuda_gb": 16.86}
{"step": 4320, "loss": 0.002772, "loss_mean50": 0.002824, "lr": 2.2021402511072963e-05, "grad_norm": 0.0525, "samples": 138240, "wall_s": 8563.8, "cuda_gb": 16.86}
{"step": 4340, "loss": 0.003534, "loss_mean50": 0.002948, "lr": 2.1987096704966034e-05, "grad_norm": 0.0437, "samples": 138880, "wall_s": 8602.7, "cuda_gb": 16.86}
{"step": 4360, "loss": 0.003643, "loss_mean50": 0.003114, "lr": 2.1952624987125566e-05, "grad_norm": 0.0445, "samples": 139520, "wall_s": 8641.7, "cuda_gb": 16.86}
{"step": 4380, "loss": 0.00283, "loss_mean50": 0.003155, "lr": 2.1917988051881456e-05, "grad_norm": 0.0292, "samples": 140160, "wall_s": 8680.4, "cuda_gb": 16.86}
{"step": 4400, "loss": 0.002409, "loss_mean50": 0.003017, "lr": 2.188318659689139e-05, "grad_norm": 0.0304, "samples": 140800, "wall_s": 8719.3, "cuda_gb": 16.86}
{"step": 4420, "loss": 0.002893, "loss_mean50": 0.003159, "lr": 2.1848221323126857e-05, "grad_norm": 0.0372, "samples": 141440, "wall_s": 8758.1, "cuda_gb": 16.86}
{"step": 4440, "loss": 0.003159, "loss_mean50": 0.003089, "lr": 2.1813092934858935e-05, "grad_norm": 0.0285, "samples": 142080, "wall_s": 8797.1, "cuda_gb": 16.86}
{"step": 4460, "loss": 0.002001, "loss_mean50": 0.003122, "lr": 2.17778021396442e-05, "grad_norm": 0.0292, "samples": 142720, "wall_s": 8836.1, "cuda_gb": 16.86}
{"step": 4480, "loss": 0.003324, "loss_mean50": 0.003256, "lr": 2.1742349648310403e-05, "grad_norm": 0.0724, "samples": 143360, "wall_s": 8875.1, "cuda_gb": 16.86}
{"step": 4500, "loss": 0.00157, "loss_mean50": 0.003227, "lr": 2.1706736174942196e-05, "grad_norm": 0.0216, "samples": 144000, "wall_s": 8914.0, "cuda_gb": 16.86}
{"step": 4520, "loss": 0.006828, "loss_mean50": 0.003214, "lr": 2.1670962436866725e-05, "grad_norm": 0.0413, "samples": 144640, "wall_s": 8953.1, "cuda_gb": 16.86}
{"step": 4540, "loss": 0.002612, "loss_mean50": 0.003178, "lr": 2.16350291546392e-05, "grad_norm": 0.0358, "samples": 145280, "wall_s": 8991.9, "cuda_gb": 16.86}
{"step": 4560, "loss": 0.003402, "loss_mean50": 0.00373, "lr": 2.159893705202836e-05, "grad_norm": 0.035, "samples": 145920, "wall_s": 9030.9, "cuda_gb": 16.86}
{"step": 4580, "loss": 0.001653, "loss_mean50": 0.003598, "lr": 2.1562686856001923e-05, "grad_norm": 0.0325, "samples": 146560, "wall_s": 9069.8, "cuda_gb": 16.86}
{"step": 4600, "loss": 0.003194, "loss_mean50": 0.003227, "lr": 2.1526279296711908e-05, "grad_norm": 0.042, "samples": 147200, "wall_s": 9108.7, "cuda_gb": 16.86}
{"step": 4620, "loss": 0.004131, "loss_mean50": 0.003002, "lr": 2.1489715107479956e-05, "grad_norm": 0.0407, "samples": 147840, "wall_s": 9147.4, "cuda_gb": 16.86}
{"step": 4640, "loss": 0.001112, "loss_mean50": 0.002788, "lr": 2.1452995024782553e-05, "grad_norm": 0.0208, "samples": 148480, "wall_s": 9186.2, "cuda_gb": 16.86}
{"step": 4660, "loss": 0.002377, "loss_mean50": 0.00281, "lr": 2.1416119788236198e-05, "grad_norm": 0.022, "samples": 149120, "wall_s": 9225.0, "cuda_gb": 16.86}
{"step": 4680, "loss": 0.002865, "loss_mean50": 0.002988, "lr": 2.1379090140582484e-05, "grad_norm": 0.0263, "samples": 149760, "wall_s": 9263.7, "cuda_gb": 16.86}
{"step": 4700, "loss": 0.002079, "loss_mean50": 0.003203, "lr": 2.1341906827673175e-05, "grad_norm": 0.0291, "samples": 150400, "wall_s": 9302.4, "cuda_gb": 16.86}
{"step": 4720, "loss": 0.000817, "loss_mean50": 0.003116, "lr": 2.130457059845515e-05, "grad_norm": 0.0157, "samples": 151040, "wall_s": 9341.2, "cuda_gb": 16.86}
{"step": 4740, "loss": 0.004543, "loss_mean50": 0.00323, "lr": 2.1267082204955335e-05, "grad_norm": 0.0367, "samples": 151680, "wall_s": 9380.0, "cuda_gb": 16.86}
{"step": 4760, "loss": 0.002686, "loss_mean50": 0.003144, "lr": 2.1229442402265554e-05, "grad_norm": 0.0461, "samples": 152320, "wall_s": 9418.8, "cuda_gb": 16.86}
{"step": 4780, "loss": 0.001632, "loss_mean50": 0.003151, "lr": 2.1191651948527313e-05, "grad_norm": 0.0254, "samples": 152960, "wall_s": 9457.6, "cuda_gb": 16.86}
{"step": 4800, "loss": 0.002579, "loss_mean50": 0.003061, "lr": 2.1153711604916535e-05, "grad_norm": 0.03, "samples": 153600, "wall_s": 9496.3, "cuda_gb": 16.86}
{"step": 4820, "loss": 0.001617, "loss_mean50": 0.003228, "lr": 2.111562213562823e-05, "grad_norm": 0.0252, "samples": 154240, "wall_s": 9535.2, "cuda_gb": 16.86}
{"step": 4840, "loss": 0.001801, "loss_mean50": 0.00295, "lr": 2.1077384307861093e-05, "grad_norm": 0.0201, "samples": 154880, "wall_s": 9574.1, "cuda_gb": 16.86}
{"step": 4860, "loss": 0.004706, "loss_mean50": 0.002867, "lr": 2.103899889180206e-05, "grad_norm": 0.0551, "samples": 155520, "wall_s": 9612.9, "cuda_gb": 16.86}
{"step": 4880, "loss": 0.002154, "loss_mean50": 0.002974, "lr": 2.100046666061081e-05, "grad_norm": 0.0239, "samples": 156160, "wall_s": 9651.7, "cuda_gb": 16.86}
{"step": 4900, "loss": 0.004092, "loss_mean50": 0.003168, "lr": 2.096178839040414e-05, "grad_norm": 0.0287, "samples": 156800, "wall_s": 9690.5, "cuda_gb": 16.86}
{"step": 4920, "loss": 0.004579, "loss_mean50": 0.003082, "lr": 2.0922964860240404e-05, "grad_norm": 0.0419, "samples": 157440, "wall_s": 9729.2, "cuda_gb": 16.86}
{"step": 4940, "loss": 0.002621, "loss_mean50": 0.00279, "lr": 2.088399685210376e-05, "grad_norm": 0.0427, "samples": 158080, "wall_s": 9768.0, "cuda_gb": 16.86}
{"step": 4960, "loss": 0.002992, "loss_mean50": 0.002921, "lr": 2.0844885150888443e-05, "grad_norm": 0.0595, "samples": 158720, "wall_s": 9806.9, "cuda_gb": 16.86}
{"step": 4980, "loss": 0.001362, "loss_mean50": 0.002969, "lr": 2.0805630544382978e-05, "grad_norm": 0.0185, "samples": 159360, "wall_s": 9845.7, "cuda_gb": 16.86}
{"step": 5000, "loss": 0.001735, "loss_mean50": 0.00293, "lr": 2.076623382325427e-05, "grad_norm": 0.0239, "samples": 160000, "wall_s": 9884.5, "cuda_gb": 16.86}
{"step": 5000, "val_flow_mse": 0.002648, "val_action_mae": 0.02156, "best": true}
{"step": 5020, "loss": 0.002887, "loss_mean50": 0.002908, "lr": 2.0726695781031705e-05, "grad_norm": 0.0348, "samples": 160640, "wall_s": 10006.7, "cuda_gb": 16.86}
{"step": 5040, "loss": 0.003124, "loss_mean50": 0.003036, "lr": 2.0687017214091166e-05, "grad_norm": 0.0689, "samples": 161280, "wall_s": 10045.5, "cuda_gb": 16.86}
{"step": 5060, "loss": 0.002735, "loss_mean50": 0.002919, "lr": 2.064719892163898e-05, "grad_norm": 0.0308, "samples": 161920, "wall_s": 10084.3, "cuda_gb": 16.86}
{"step": 5080, "loss": 0.003412, "loss_mean50": 0.002932, "lr": 2.0607241705695827e-05, "grad_norm": 0.0264, "samples": 162560, "wall_s": 10123.2, "cuda_gb": 16.86}
{"step": 5100, "loss": 0.003165, "loss_mean50": 0.002932, "lr": 2.05671463710806e-05, "grad_norm": 0.0344, "samples": 163200, "wall_s": 10162.0, "cuda_gb": 16.86}
{"step": 5120, "loss": 0.001726, "loss_mean50": 0.002793, "lr": 2.0526913725394155e-05, "grad_norm": 0.0291, "samples": 163840, "wall_s": 10200.9, "cuda_gb": 16.86}
{"step": 5140, "loss": 0.002288, "loss_mean50": 0.002804, "lr": 2.0486544579003098e-05, "grad_norm": 0.0452, "samples": 164480, "wall_s": 10239.9, "cuda_gb": 16.86}
{"step": 5160, "loss": 0.003208, "loss_mean50": 0.002875, "lr": 2.044603974502341e-05, "grad_norm": 0.0419, "samples": 165120, "wall_s": 10278.7, "cuda_gb": 16.86}
{"step": 5180, "loss": 0.002254, "loss_mean50": 0.00276, "lr": 2.0405400039304117e-05, "grad_norm": 0.0325, "samples": 165760, "wall_s": 10317.6, "cuda_gb": 16.86}
{"step": 5200, "loss": 0.003947, "loss_mean50": 0.002905, "lr": 2.0364626280410812e-05, "grad_norm": 0.0469, "samples": 166400, "wall_s": 10356.5, "cuda_gb": 16.86}
{"step": 5220, "loss": 0.004643, "loss_mean50": 0.002904, "lr": 2.03237192896092e-05, "grad_norm": 0.0405, "samples": 167040, "wall_s": 10395.5, "cuda_gb": 16.86}
{"step": 5240, "loss": 0.001919, "loss_mean50": 0.002878, "lr": 2.028267989084855e-05, "grad_norm": 0.0242, "samples": 167680, "wall_s": 10434.4, "cuda_gb": 16.86}
{"step": 5260, "loss": 0.003055, "loss_mean50": 0.0029, "lr": 2.024150891074508e-05, "grad_norm": 0.0273, "samples": 168320, "wall_s": 10473.3, "cuda_gb": 16.86}
{"step": 5280, "loss": 0.001235, "loss_mean50": 0.002948, "lr": 2.0200207178565332e-05, "grad_norm": 0.0256, "samples": 168960, "wall_s": 10512.2, "cuda_gb": 16.86}
{"step": 5300, "loss": 0.00271, "loss_mean50": 0.002994, "lr": 2.015877552620946e-05, "grad_norm": 0.0336, "samples": 169600, "wall_s": 10551.3, "cuda_gb": 16.86}
{"step": 5320, "loss": 0.001997, "loss_mean50": 0.002842, "lr": 2.011721478819446e-05, "grad_norm": 0.0332, "samples": 170240, "wall_s": 10590.2, "cuda_gb": 16.86}
{"step": 5340, "loss": 0.003367, "loss_mean50": 0.002819, "lr": 2.007552580163739e-05, "grad_norm": 0.0299, "samples": 170880, "wall_s": 10629.1, "cuda_gb": 16.86}
{"step": 5360, "loss": 0.002093, "loss_mean50": 0.00278, "lr": 2.003370940623848e-05, "grad_norm": 0.0282, "samples": 171520, "wall_s": 10668.0, "cuda_gb": 16.86}
{"step": 5380, "loss": 0.001454, "loss_mean50": 0.00284, "lr": 1.9991766444264238e-05, "grad_norm": 0.0281, "samples": 172160, "wall_s": 10707.0, "cuda_gb": 16.86}
{"step": 5400, "loss": 0.002939, "loss_mean50": 0.002888, "lr": 1.9949697760530468e-05, "grad_norm": 0.0477, "samples": 172800, "wall_s": 10745.9, "cuda_gb": 16.86}
{"step": 5420, "loss": 0.003758, "loss_mean50": 0.003011, "lr": 1.9907504202385275e-05, "grad_norm": 0.0789, "samples": 173440, "wall_s": 10784.8, "cuda_gb": 16.86}
{"step": 5440, "loss": 0.003238, "loss_mean50": 0.003043, "lr": 1.9865186619691983e-05, "grad_norm": 0.0532, "samples": 174080, "wall_s": 10823.7, "cuda_gb": 16.86}
{"step": 5460, "loss": 0.001803, "loss_mean50": 0.003132, "lr": 1.9822745864812015e-05, "grad_norm": 0.0412, "samples": 174720, "wall_s": 10862.6, "cuda_gb": 16.86}
{"step": 5480, "loss": 0.00412, "loss_mean50": 0.003084, "lr": 1.9780182792587738e-05, "grad_norm": 0.0597, "samples": 175360, "wall_s": 10901.6, "cuda_gb": 16.86}
{"step": 5500, "loss": 0.00277, "loss_mean50": 0.00308, "lr": 1.9737498260325232e-05, "grad_norm": 0.0309, "samples": 176000, "wall_s": 10940.6, "cuda_gb": 16.86}
{"step": 5520, "loss": 0.006163, "loss_mean50": 0.002987, "lr": 1.969469312777703e-05, "grad_norm": 0.1234, "samples": 176640, "wall_s": 10979.5, "cuda_gb": 16.86}
{"step": 5540, "loss": 0.002729, "loss_mean50": 0.002745, "lr": 1.9651768257124793e-05, "grad_norm": 0.03, "samples": 177280, "wall_s": 11018.4, "cuda_gb": 16.86}
{"step": 5560, "loss": 0.003225, "loss_mean50": 0.002709, "lr": 1.9608724512961953e-05, "grad_norm": 0.0409, "samples": 177920, "wall_s": 11057.3, "cuda_gb": 16.86}
{"step": 5580, "loss": 0.004459, "loss_mean50": 0.002968, "lr": 1.9565562762276297e-05, "grad_norm": 0.0363, "samples": 178560, "wall_s": 11096.3, "cuda_gb": 16.86}
{"step": 5600, "loss": 0.00282, "loss_mean50": 0.002841, "lr": 1.9522283874432492e-05, "grad_norm": 0.0493, "samples": 179200, "wall_s": 11135.1, "cuda_gb": 16.86}
{"step": 5620, "loss": 0.003416, "loss_mean50": 0.002853, "lr": 1.9478888721154596e-05, "grad_norm": 0.0597, "samples": 179840, "wall_s": 11174.1, "cuda_gb": 16.86}
{"step": 5640, "loss": 0.004228, "loss_mean50": 0.00282, "lr": 1.9435378176508466e-05, "grad_norm": 0.0547, "samples": 180480, "wall_s": 11213.0, "cuda_gb": 16.86}
{"step": 5660, "loss": 0.002678, "loss_mean50": 0.003024, "lr": 1.9391753116884205e-05, "grad_norm": 0.0233, "samples": 181120, "wall_s": 11251.9, "cuda_gb": 16.86}
{"step": 5680, "loss": 0.002407, "loss_mean50": 0.00295, "lr": 1.934801442097845e-05, "grad_norm": 0.0528, "samples": 181760, "wall_s": 11290.8, "cuda_gb": 16.86}
{"step": 5700, "loss": 0.00438, "loss_mean50": 0.00277, "lr": 1.9304162969776722e-05, "grad_norm": 0.0653, "samples": 182400, "wall_s": 11329.8, "cuda_gb": 16.86}
{"step": 5720, "loss": 0.002778, "loss_mean50": 0.002828, "lr": 1.9260199646535644e-05, "grad_norm": 0.0262, "samples": 183040, "wall_s": 11368.8, "cuda_gb": 16.86}
{"step": 5740, "loss": 0.00217, "loss_mean50": 0.002835, "lr": 1.921612533676518e-05, "grad_norm": 0.0445, "samples": 183680, "wall_s": 11408.2, "cuda_gb": 16.86}
{"step": 5760, "loss": 0.001443, "loss_mean50": 0.00273, "lr": 1.9171940928210784e-05, "grad_norm": 0.0263, "samples": 184320, "wall_s": 11447.2, "cuda_gb": 16.86}
{"step": 5780, "loss": 0.002872, "loss_mean50": 0.00259, "lr": 1.9127647310835513e-05, "grad_norm": 0.0355, "samples": 184960, "wall_s": 11486.1, "cuda_gb": 16.86}
{"step": 5800, "loss": 0.002885, "loss_mean50": 0.002718, "lr": 1.9083245376802126e-05, "grad_norm": 0.0266, "samples": 185600, "wall_s": 11525.0, "cuda_gb": 16.86}
{"step": 5820, "loss": 0.005979, "loss_mean50": 0.002737, "lr": 1.903873602045508e-05, "grad_norm": 0.0477, "samples": 186240, "wall_s": 11563.9, "cuda_gb": 16.86}
{"step": 5840, "loss": 0.003231, "loss_mean50": 0.002856, "lr": 1.8994120138302543e-05, "grad_norm": 0.0557, "samples": 186880, "wall_s": 11602.9, "cuda_gb": 16.86}
{"step": 5860, "loss": 0.002298, "loss_mean50": 0.003162, "lr": 1.894939862899833e-05, "grad_norm": 0.0201, "samples": 187520, "wall_s": 11641.9, "cuda_gb": 16.86}
{"step": 5880, "loss": 0.00245, "loss_mean50": 0.002985, "lr": 1.8904572393323796e-05, "grad_norm": 0.0406, "samples": 188160, "wall_s": 11680.7, "cuda_gb": 16.86}
{"step": 5900, "loss": 0.001527, "loss_mean50": 0.002755, "lr": 1.8859642334169693e-05, "grad_norm": 0.022, "samples": 188800, "wall_s": 11719.6, "cuda_gb": 16.86}
{"step": 5920, "loss": 0.003556, "loss_mean50": 0.002614, "lr": 1.8814609356517996e-05, "grad_norm": 0.0803, "samples": 189440, "wall_s": 11758.6, "cuda_gb": 16.86}
{"step": 5940, "loss": 0.002083, "loss_mean50": 0.002633, "lr": 1.8769474367423657e-05, "grad_norm": 0.0251, "samples": 190080, "wall_s": 11797.6, "cuda_gb": 16.86}
{"step": 5960, "loss": 0.002788, "loss_mean50": 0.002837, "lr": 1.872423827599635e-05, "grad_norm": 0.0314, "samples": 190720, "wall_s": 11836.5, "cuda_gb": 16.86}
{"step": 5980, "loss": 0.001172, "loss_mean50": 0.002695, "lr": 1.8678901993382156e-05, "grad_norm": 0.0165, "samples": 191360, "wall_s": 11875.4, "cuda_gb": 16.86}
{"step": 6000, "loss": 0.001674, "loss_mean50": 0.002881, "lr": 1.8633466432745193e-05, "grad_norm": 0.022, "samples": 192000, "wall_s": 11914.3, "cuda_gb": 16.86}
{"step": 6000, "val_flow_mse": 0.002398}
{"step": 6020, "loss": 0.002418, "loss_mean50": 0.002664, "lr": 1.8587932509249265e-05, "grad_norm": 0.0566, "samples": 192640, "wall_s": 11982.0, "cuda_gb": 16.86}
{"step": 6040, "loss": 0.003099, "loss_mean50": 0.002892, "lr": 1.8542301140039374e-05, "grad_norm": 0.0282, "samples": 193280, "wall_s": 12020.9, "cuda_gb": 16.86}
{"step": 6060, "loss": 0.004903, "loss_mean50": 0.003073, "lr": 1.849657324422329e-05, "grad_norm": 0.0436, "samples": 193920, "wall_s": 12059.8, "cuda_gb": 16.86}
{"step": 6080, "loss": 0.002246, "loss_mean50": 0.003143, "lr": 1.845074974285303e-05, "grad_norm": 0.0483, "samples": 194560, "wall_s": 12098.6, "cuda_gb": 16.86}
{"step": 6100, "loss": 0.002816, "loss_mean50": 0.003001, "lr": 1.8404831558906283e-05, "grad_norm": 0.0273, "samples": 195200, "wall_s": 12137.5, "cuda_gb": 16.86}
{"step": 6120, "loss": 0.004002, "loss_mean50": 0.002855, "lr": 1.8358819617267847e-05, "grad_norm": 0.0723, "samples": 195840, "wall_s": 12176.3, "cuda_gb": 16.86}
{"step": 6140, "loss": 0.001328, "loss_mean50": 0.002709, "lr": 1.831271484471099e-05, "grad_norm": 0.0219, "samples": 196480, "wall_s": 12215.2, "cuda_gb": 16.86}
{"step": 6160, "loss": 0.001938, "loss_mean50": 0.002628, "lr": 1.8266518169878774e-05, "grad_norm": 0.05, "samples": 197120, "wall_s": 12254.2, "cuda_gb": 16.86}
{"step": 6180, "loss": 0.003556, "loss_mean50": 0.00245, "lr": 1.822023052326536e-05, "grad_norm": 0.0355, "samples": 197760, "wall_s": 12293.0, "cuda_gb": 16.86}
{"step": 6200, "loss": 0.002542, "loss_mean50": 0.002728, "lr": 1.8173852837197273e-05, "grad_norm": 0.0712, "samples": 198400, "wall_s": 12331.9, "cuda_gb": 16.86}
{"step": 6220, "loss": 0.001949, "loss_mean50": 0.00269, "lr": 1.81273860458146e-05, "grad_norm": 0.054, "samples": 199040, "wall_s": 12370.9, "cuda_gb": 16.86}
{"step": 6240, "loss": 0.003624, "loss_mean50": 0.002906, "lr": 1.8080831085052197e-05, "grad_norm": 0.1604, "samples": 199680, "wall_s": 12409.8, "cuda_gb": 16.86}
{"step": 6260, "loss": 0.001974, "loss_mean50": 0.002562, "lr": 1.8034188892620834e-05, "grad_norm": 0.0272, "samples": 200320, "wall_s": 12448.8, "cuda_gb": 16.86}
{"step": 6280, "loss": 0.002317, "loss_mean50": 0.002468, "lr": 1.7987460407988294e-05, "grad_norm": 0.0425, "samples": 200960, "wall_s": 12487.7, "cuda_gb": 16.86}
{"step": 6300, "loss": 0.003541, "loss_mean50": 0.002463, "lr": 1.794064657236046e-05, "grad_norm": 0.0773, "samples": 201600, "wall_s": 12526.7, "cuda_gb": 16.86}
{"step": 6320, "loss": 0.001567, "loss_mean50": 0.002541, "lr": 1.7893748328662355e-05, "grad_norm": 0.0258, "samples": 202240, "wall_s": 12565.6, "cuda_gb": 16.86}
{"step": 6340, "loss": 0.001657, "loss_mean50": 0.002549, "lr": 1.7846766621519163e-05, "grad_norm": 0.0311, "samples": 202880, "wall_s": 12604.5, "cuda_gb": 16.86}
{"step": 6360, "loss": 0.002417, "loss_mean50": 0.002428, "lr": 1.7799702397237172e-05, "grad_norm": 0.0332, "samples": 203520, "wall_s": 12643.4, "cuda_gb": 16.86}
{"step": 6380, "loss": 0.002812, "loss_mean50": 0.002518, "lr": 1.775255660378474e-05, "grad_norm": 0.0615, "samples": 204160, "wall_s": 12682.3, "cuda_gb": 16.86}
{"step": 6400, "loss": 0.00271, "loss_mean50": 0.002483, "lr": 1.77053301907732e-05, "grad_norm": 0.0824, "samples": 204800, "wall_s": 12721.2, "cuda_gb": 16.86}
{"step": 6420, "loss": 0.001927, "loss_mean50": 0.002653, "lr": 1.765802410943771e-05, "grad_norm": 0.0284, "samples": 205440, "wall_s": 12760.1, "cuda_gb": 16.86}
{"step": 6440, "loss": 0.002587, "loss_mean50": 0.002669, "lr": 1.761063931261812e-05, "grad_norm": 0.0305, "samples": 206080, "wall_s": 12799.1, "cuda_gb": 16.86}
{"step": 6460, "loss": 0.002152, "loss_mean50": 0.002597, "lr": 1.7563176754739762e-05, "grad_norm": 0.0239, "samples": 206720, "wall_s": 12838.0, "cuda_gb": 16.86}
{"step": 6480, "loss": 0.00335, "loss_mean50": 0.002516, "lr": 1.751563739179424e-05, "grad_norm": 0.0535, "samples": 207360, "wall_s": 12877.0, "cuda_gb": 16.86}
{"step": 6500, "loss": 0.003672, "loss_mean50": 0.002454, "lr": 1.7468022181320165e-05, "grad_norm": 0.0312, "samples": 208000, "wall_s": 12916.0, "cuda_gb": 16.86}
{"step": 6520, "loss": 0.004623, "loss_mean50": 0.002574, "lr": 1.742033208238386e-05, "grad_norm": 0.0808, "samples": 208640, "wall_s": 12954.8, "cuda_gb": 16.86}
{"step": 6540, "loss": 0.002787, "loss_mean50": 0.002598, "lr": 1.737256805556007e-05, "grad_norm": 0.0216, "samples": 209280, "wall_s": 12993.8, "cuda_gb": 16.86}
{"step": 6560, "loss": 0.001601, "loss_mean50": 0.002489, "lr": 1.7324731062912574e-05, "grad_norm": 0.0249, "samples": 209920, "wall_s": 13032.6, "cuda_gb": 16.86}
{"step": 6580, "loss": 0.004019, "loss_mean50": 0.002525, "lr": 1.7276822067974855e-05, "grad_norm": 0.0422, "samples": 210560, "wall_s": 13071.6, "cuda_gb": 16.86}
{"step": 6600, "loss": 0.002115, "loss_mean50": 0.002471, "lr": 1.722884203573065e-05, "grad_norm": 0.0345, "samples": 211200, "wall_s": 13110.5, "cuda_gb": 16.86}
{"step": 6620, "loss": 0.001998, "loss_mean50": 0.002637, "lr": 1.7180791932594536e-05, "grad_norm": 0.0306, "samples": 211840, "wall_s": 13149.4, "cuda_gb": 16.86}
{"step": 6640, "loss": 0.002185, "loss_mean50": 0.002735, "lr": 1.7132672726392453e-05, "grad_norm": 0.0226, "samples": 212480, "wall_s": 13188.4, "cuda_gb": 16.86}
{"step": 6660, "loss": 0.00523, "loss_mean50": 0.002947, "lr": 1.708448538634223e-05, "grad_norm": 0.1114, "samples": 213120, "wall_s": 13227.2, "cuda_gb": 16.86}
{"step": 6680, "loss": 0.00207, "loss_mean50": 0.002707, "lr": 1.703623088303404e-05, "grad_norm": 0.021, "samples": 213760, "wall_s": 13266.0, "cuda_gb": 16.86}
{"step": 6700, "loss": 0.002942, "loss_mean50": 0.002541, "lr": 1.698791018841086e-05, "grad_norm": 0.0303, "samples": 214400, "wall_s": 13305.0, "cuda_gb": 16.86}
{"step": 6720, "loss": 0.00207, "loss_mean50": 0.002677, "lr": 1.693952427574889e-05, "grad_norm": 0.0263, "samples": 215040, "wall_s": 13344.0, "cuda_gb": 16.86}
{"step": 6740, "loss": 0.002237, "loss_mean50": 0.002694, "lr": 1.6891074119637956e-05, "grad_norm": 0.0255, "samples": 215680, "wall_s": 13382.9, "cuda_gb": 16.86}
{"step": 6760, "loss": 0.002468, "loss_mean50": 0.002715, "lr": 1.6842560695961884e-05, "grad_norm": 0.0351, "samples": 216320, "wall_s": 13421.8, "cuda_gb": 16.86}
{"step": 6780, "loss": 0.003442, "loss_mean50": 0.002527, "lr": 1.6793984981878822e-05, "grad_norm": 0.0491, "samples": 216960, "wall_s": 13460.8, "cuda_gb": 16.86}
{"step": 6800, "loss": 0.00203, "loss_mean50": 0.002581, "lr": 1.6745347955801585e-05, "grad_norm": 0.0323, "samples": 217600, "wall_s": 13499.8, "cuda_gb": 16.86}
{"step": 6820, "loss": 0.004218, "loss_mean50": 0.002834, "lr": 1.6696650597377923e-05, "grad_norm": 0.031, "samples": 218240, "wall_s": 13538.8, "cuda_gb": 16.86}
{"step": 6840, "loss": 0.001671, "loss_mean50": 0.00282, "lr": 1.6647893887470816e-05, "grad_norm": 0.0312, "samples": 218880, "wall_s": 13577.8, "cuda_gb": 16.86}
{"step": 6860, "loss": 0.002639, "loss_mean50": 0.002568, "lr": 1.6599078808138693e-05, "grad_norm": 0.0256, "samples": 219520, "wall_s": 13616.8, "cuda_gb": 16.86}
{"step": 6880, "loss": 0.001536, "loss_mean50": 0.002492, "lr": 1.6550206342615658e-05, "grad_norm": 0.0195, "samples": 220160, "wall_s": 13655.9, "cuda_gb": 16.86}
{"step": 6900, "loss": 0.0008, "loss_mean50": 0.002373, "lr": 1.6501277475291696e-05, "grad_norm": 0.018, "samples": 220800, "wall_s": 13694.8, "cuda_gb": 16.86}
{"step": 6920, "loss": 0.002915, "loss_mean50": 0.002411, "lr": 1.645229319169284e-05, "grad_norm": 0.0382, "samples": 221440, "wall_s": 13733.8, "cuda_gb": 16.86}
{"step": 6940, "loss": 0.002988, "loss_mean50": 0.002613, "lr": 1.64032544784613e-05, "grad_norm": 0.0269, "samples": 222080, "wall_s": 13772.9, "cuda_gb": 16.86}
{"step": 6960, "loss": 0.00183, "loss_mean50": 0.002481, "lr": 1.6354162323335633e-05, "grad_norm": 0.0282, "samples": 222720, "wall_s": 13811.8, "cuda_gb": 16.86}
{"step": 6980, "loss": 0.003163, "loss_mean50": 0.002362, "lr": 1.6305017715130812e-05, "grad_norm": 0.0582, "samples": 223360, "wall_s": 13850.8, "cuda_gb": 16.86}
{"step": 7000, "loss": 0.001925, "loss_mean50": 0.002471, "lr": 1.6255821643718313e-05, "grad_norm": 0.0325, "samples": 224000, "wall_s": 13889.7, "cuda_gb": 16.86}
{"step": 7000, "val_flow_mse": 0.002234}
{"step": 7020, "loss": 0.002809, "loss_mean50": 0.002548, "lr": 1.6206575100006205e-05, "grad_norm": 0.0383, "samples": 224640, "wall_s": 13957.5, "cuda_gb": 16.86}
{"step": 7040, "loss": 0.002029, "loss_mean50": 0.002524, "lr": 1.615727907591915e-05, "grad_norm": 0.0334, "samples": 225280, "wall_s": 13996.5, "cuda_gb": 16.86}
{"step": 7060, "loss": 0.001539, "loss_mean50": 0.002483, "lr": 1.610793456437846e-05, "grad_norm": 0.0258, "samples": 225920, "wall_s": 14035.3, "cuda_gb": 16.86}
{"step": 7080, "loss": 0.004501, "loss_mean50": 0.002432, "lr": 1.6058542559282066e-05, "grad_norm": 0.0555, "samples": 226560, "wall_s": 14074.4, "cuda_gb": 16.86}
{"step": 7100, "loss": 0.000753, "loss_mean50": 0.002301, "lr": 1.6009104055484537e-05, "grad_norm": 0.0181, "samples": 227200, "wall_s": 14113.3, "cuda_gb": 16.86}
{"step": 7120, "loss": 0.002475, "loss_mean50": 0.002342, "lr": 1.5959620048776996e-05, "grad_norm": 0.043, "samples": 227840, "wall_s": 14152.2, "cuda_gb": 16.86}
{"step": 7140, "loss": 0.001425, "loss_mean50": 0.002206, "lr": 1.59100915358671e-05, "grad_norm": 0.0246, "samples": 228480, "wall_s": 14191.1, "cuda_gb": 16.86}
{"step": 7160, "loss": 0.002447, "loss_mean50": 0.00235, "lr": 1.586051951435895e-05, "grad_norm": 0.0216, "samples": 229120, "wall_s": 14230.0, "cuda_gb": 16.86}
{"step": 7180, "loss": 0.004131, "loss_mean50": 0.002611, "lr": 1.5810904982733e-05, "grad_norm": 0.0362, "samples": 229760, "wall_s": 14269.0, "cuda_gb": 16.86}
{"step": 7200, "loss": 0.001542, "loss_mean50": 0.002663, "lr": 1.576124894032593e-05, "grad_norm": 0.0236, "samples": 230400, "wall_s": 14307.9, "cuda_gb": 16.86}
{"step": 7220, "loss": 0.002455, "loss_mean50": 0.002646, "lr": 1.5711552387310548e-05, "grad_norm": 0.0257, "samples": 231040, "wall_s": 14346.9, "cuda_gb": 16.86}
{"step": 7240, "loss": 0.002147, "loss_mean50": 0.002381, "lr": 1.5661816324675613e-05, "grad_norm": 0.0285, "samples": 231680, "wall_s": 14385.8, "cuda_gb": 16.86}
{"step": 7260, "loss": 0.001656, "loss_mean50": 0.002212, "lr": 1.56120417542057e-05, "grad_norm": 0.022, "samples": 232320, "wall_s": 14424.8, "cuda_gb": 16.86}
{"step": 7280, "loss": 0.003527, "loss_mean50": 0.00246, "lr": 1.5562229678460998e-05, "grad_norm": 0.0466, "samples": 232960, "wall_s": 14463.9, "cuda_gb": 16.86}
{"step": 7300, "loss": 0.003038, "loss_mean50": 0.002535, "lr": 1.5512381100757136e-05, "grad_norm": 0.0584, "samples": 233600, "wall_s": 14503.0, "cuda_gb": 16.86}
{"step": 7320, "loss": 0.002766, "loss_mean50": 0.002639, "lr": 1.5462497025144967e-05, "grad_norm": 0.0334, "samples": 234240, "wall_s": 14542.0, "cuda_gb": 16.86}
{"step": 7340, "loss": 0.001539, "loss_mean50": 0.002396, "lr": 1.541257845639033e-05, "grad_norm": 0.0229, "samples": 234880, "wall_s": 14581.0, "cuda_gb": 16.86}
{"step": 7360, "loss": 0.001583, "loss_mean50": 0.002235, "lr": 1.536262639995384e-05, "grad_norm": 0.0239, "samples": 235520, "wall_s": 14620.1, "cuda_gb": 16.86}
{"step": 7380, "loss": 0.003329, "loss_mean50": 0.002347, "lr": 1.531264186197062e-05, "grad_norm": 0.0812, "samples": 236160, "wall_s": 14659.3, "cuda_gb": 16.86}
{"step": 7400, "loss": 0.003966, "loss_mean50": 0.002468, "lr": 1.526262584923003e-05, "grad_norm": 0.0365, "samples": 236800, "wall_s": 14698.4, "cuda_gb": 16.86}
{"step": 7420, "loss": 0.002023, "loss_mean50": 0.002402, "lr": 1.5212579369155395e-05, "grad_norm": 0.0285, "samples": 237440, "wall_s": 14737.4, "cuda_gb": 16.86}
{"step": 7440, "loss": 0.001897, "loss_mean50": 0.002176, "lr": 1.5162503429783722e-05, "grad_norm": 0.0499, "samples": 238080, "wall_s": 14776.5, "cuda_gb": 16.86}
{"step": 7460, "loss": 0.002563, "loss_mean50": 0.002184, "lr": 1.5112399039745381e-05, "grad_norm": 0.0303, "samples": 238720, "wall_s": 14815.5, "cuda_gb": 16.86}
{"step": 7480, "loss": 0.002795, "loss_mean50": 0.002463, "lr": 1.5062267208243797e-05, "grad_norm": 0.0304, "samples": 239360, "wall_s": 14854.5, "cuda_gb": 16.86}
{"step": 7500, "loss": 0.001745, "loss_mean50": 0.002564, "lr": 1.5012108945035122e-05, "grad_norm": 0.0227, "samples": 240000, "wall_s": 14893.5, "cuda_gb": 16.86}
{"step": 7520, "loss": 0.001454, "loss_mean50": 0.002362, "lr": 1.49619252604079e-05, "grad_norm": 0.0184, "samples": 240640, "wall_s": 14968.9, "cuda_gb": 16.86}
{"step": 7540, "loss": 0.001862, "loss_mean50": 0.002289, "lr": 1.49117171651627e-05, "grad_norm": 0.0189, "samples": 241280, "wall_s": 15007.7, "cuda_gb": 16.86}
{"step": 7560, "loss": 0.001168, "loss_mean50": 0.002371, "lr": 1.4861485670591782e-05, "grad_norm": 0.0197, "samples": 241920, "wall_s": 15046.7, "cuda_gb": 16.86}
{"step": 7580, "loss": 0.001969, "loss_mean50": 0.002346, "lr": 1.4811231788458716e-05, "grad_norm": 0.0403, "samples": 242560, "wall_s": 15085.8, "cuda_gb": 16.86}
{"step": 7600, "loss": 0.001753, "loss_mean50": 0.002347, "lr": 1.4760956530978001e-05, "grad_norm": 0.0235, "samples": 243200, "wall_s": 15124.8, "cuda_gb": 16.86}
{"step": 7620, "loss": 0.003379, "loss_mean50": 0.002447, "lr": 1.471066091079467e-05, "grad_norm": 0.0326, "samples": 243840, "wall_s": 15163.7, "cuda_gb": 16.86}
{"step": 7640, "loss": 0.002967, "loss_mean50": 0.002522, "lr": 1.4660345940963918e-05, "grad_norm": 0.0331, "samples": 244480, "wall_s": 15202.8, "cuda_gb": 16.86}
{"step": 7660, "loss": 0.00168, "loss_mean50": 0.002492, "lr": 1.4610012634930665e-05, "grad_norm": 0.0396, "samples": 245120, "wall_s": 15241.8, "cuda_gb": 16.86}
{"step": 7680, "loss": 0.002986, "loss_mean50": 0.002525, "lr": 1.4559662006509167e-05, "grad_norm": 0.033, "samples": 245760, "wall_s": 15280.8, "cuda_gb": 16.86}
{"step": 7700, "loss": 0.004386, "loss_mean50": 0.002624, "lr": 1.4509295069862588e-05, "grad_norm": 0.1112, "samples": 246400, "wall_s": 15319.9, "cuda_gb": 16.86}
{"step": 7720, "loss": 0.001847, "loss_mean50": 0.002674, "lr": 1.4458912839482571e-05, "grad_norm": 0.0406, "samples": 247040, "wall_s": 15359.0, "cuda_gb": 16.86}
{"step": 7740, "loss": 0.002899, "loss_mean50": 0.002476, "lr": 1.4408516330168806e-05, "grad_norm": 0.0714, "samples": 247680, "wall_s": 15398.0, "cuda_gb": 16.86}
{"step": 7760, "loss": 0.001788, "loss_mean50": 0.002465, "lr": 1.4358106557008589e-05, "grad_norm": 0.0241, "samples": 248320, "wall_s": 15436.9, "cuda_gb": 16.86}
{"step": 7780, "loss": 0.003142, "loss_mean50": 0.002469, "lr": 1.4307684535356378e-05, "grad_norm": 0.0422, "samples": 248960, "wall_s": 15475.9, "cuda_gb": 16.86}
{"step": 7800, "loss": 0.00243, "loss_mean50": 0.002103, "lr": 1.4257251280813339e-05, "grad_norm": 0.0251, "samples": 249600, "wall_s": 15515.0, "cuda_gb": 16.86}
{"step": 7820, "loss": 0.00142, "loss_mean50": 0.002048, "lr": 1.420680780920689e-05, "grad_norm": 0.0175, "samples": 250240, "wall_s": 15553.9, "cuda_gb": 16.86}
{"step": 7840, "loss": 0.00228, "loss_mean50": 0.002094, "lr": 1.4156355136570243e-05, "grad_norm": 0.0357, "samples": 250880, "wall_s": 15593.0, "cuda_gb": 16.86}
{"step": 7860, "loss": 0.001864, "loss_mean50": 0.001989, "lr": 1.4105894279121939e-05, "grad_norm": 0.0397, "samples": 251520, "wall_s": 15632.1, "cuda_gb": 16.86}
{"step": 7880, "loss": 0.003618, "loss_mean50": 0.002164, "lr": 1.4055426253245374e-05, "grad_norm": 0.0389, "samples": 252160, "wall_s": 15671.1, "cuda_gb": 16.86}
{"step": 7900, "loss": 0.00122, "loss_mean50": 0.002232, "lr": 1.4004952075468333e-05, "grad_norm": 0.0275, "samples": 252800, "wall_s": 15710.1, "cuda_gb": 16.86}
{"step": 7920, "loss": 0.003606, "loss_mean50": 0.002467, "lr": 1.3954472762442508e-05, "grad_norm": 0.0644, "samples": 253440, "wall_s": 15749.1, "cuda_gb": 16.86}
{"step": 7940, "loss": 0.00534, "loss_mean50": 0.002443, "lr": 1.3903989330923037e-05, "grad_norm": 0.1265, "samples": 254080, "wall_s": 15788.2, "cuda_gb": 16.86}
{"step": 7960, "loss": 0.001792, "loss_mean50": 0.002365, "lr": 1.3853502797747997e-05, "grad_norm": 0.0251, "samples": 254720, "wall_s": 15827.2, "cuda_gb": 16.86}
{"step": 7980, "loss": 0.002984, "loss_mean50": 0.002181, "lr": 1.3803014179817949e-05, "grad_norm": 0.0366, "samples": 255360, "wall_s": 15866.2, "cuda_gb": 16.86}
{"step": 8000, "loss": 0.002771, "loss_mean50": 0.002266, "lr": 1.375252449407545e-05, "grad_norm": 0.0412, "samples": 256000, "wall_s": 15905.1, "cuda_gb": 16.86}
{"step": 8000, "val_flow_mse": 0.002004}
{"step": 8020, "loss": 0.001854, "loss_mean50": 0.002471, "lr": 1.3702034757484548e-05, "grad_norm": 0.0371, "samples": 256640, "wall_s": 15972.9, "cuda_gb": 16.86}
{"step": 8040, "loss": 0.001384, "loss_mean50": 0.002549, "lr": 1.3651545987010325e-05, "grad_norm": 0.0229, "samples": 257280, "wall_s": 16012.0, "cuda_gb": 16.86}
{"step": 8060, "loss": 0.002763, "loss_mean50": 0.002333, "lr": 1.3601059199598407e-05, "grad_norm": 0.0287, "samples": 257920, "wall_s": 16051.0, "cuda_gb": 16.86}
{"step": 8080, "loss": 0.001138, "loss_mean50": 0.002208, "lr": 1.355057541215447e-05, "grad_norm": 0.0194, "samples": 258560, "wall_s": 16090.0, "cuda_gb": 16.86}
{"step": 8100, "loss": 0.001341, "loss_mean50": 0.00213, "lr": 1.3500095641523772e-05, "grad_norm": 0.024, "samples": 259200, "wall_s": 16129.0, "cuda_gb": 16.86}
{"step": 8120, "loss": 0.002396, "loss_mean50": 0.002036, "lr": 1.3449620904470664e-05, "grad_norm": 0.0296, "samples": 259840, "wall_s": 16167.9, "cuda_gb": 16.86}
{"step": 8140, "loss": 0.002686, "loss_mean50": 0.002187, "lr": 1.3399152217658095e-05, "grad_norm": 0.0291, "samples": 260480, "wall_s": 16206.9, "cuda_gb": 16.86}
{"step": 8160, "loss": 0.003644, "loss_mean50": 0.002402, "lr": 1.3348690597627166e-05, "grad_norm": 0.0395, "samples": 261120, "wall_s": 16246.0, "cuda_gb": 16.86}
{"step": 8180, "loss": 0.001895, "loss_mean50": 0.002215, "lr": 1.3298237060776639e-05, "grad_norm": 0.039, "samples": 261760, "wall_s": 16285.1, "cuda_gb": 16.86}
{"step": 8200, "loss": 0.002789, "loss_mean50": 0.002174, "lr": 1.324779262334246e-05, "grad_norm": 0.0247, "samples": 262400, "wall_s": 16324.2, "cuda_gb": 16.86}
{"step": 8220, "loss": 0.004392, "loss_mean50": 0.002234, "lr": 1.3197358301377289e-05, "grad_norm": 0.0672, "samples": 263040, "wall_s": 16363.2, "cuda_gb": 16.86}
{"step": 8240, "loss": 0.004547, "loss_mean50": 0.00227, "lr": 1.3146935110730046e-05, "grad_norm": 0.078, "samples": 263680, "wall_s": 16402.2, "cuda_gb": 16.86}
{"step": 8260, "loss": 0.00205, "loss_mean50": 0.002212, "lr": 1.3096524067025451e-05, "grad_norm": 0.0371, "samples": 264320, "wall_s": 16441.3, "cuda_gb": 16.86}
{"step": 8280, "loss": 0.002153, "loss_mean50": 0.002478, "lr": 1.3046126185643545e-05, "grad_norm": 0.0275, "samples": 264960, "wall_s": 16480.3, "cuda_gb": 16.86}
{"step": 8300, "loss": 0.001839, "loss_mean50": 0.00267, "lr": 1.2995742481699261e-05, "grad_norm": 0.0203, "samples": 265600, "wall_s": 16519.3, "cuda_gb": 16.86}
{"step": 8320, "loss": 0.001709, "loss_mean50": 0.002552, "lr": 1.2945373970021971e-05, "grad_norm": 0.0426, "samples": 266240, "wall_s": 16558.4, "cuda_gb": 16.86}
{"step": 8340, "loss": 0.000855, "loss_mean50": 0.002367, "lr": 1.2895021665135043e-05, "grad_norm": 0.0164, "samples": 266880, "wall_s": 16597.4, "cuda_gb": 16.86}
{"step": 8360, "loss": 0.001528, "loss_mean50": 0.002234, "lr": 1.2844686581235405e-05, "grad_norm": 0.0201, "samples": 267520, "wall_s": 16636.5, "cuda_gb": 16.86}
{"step": 8380, "loss": 0.002075, "loss_mean50": 0.002292, "lr": 1.2794369732173127e-05, "grad_norm": 0.0582, "samples": 268160, "wall_s": 16675.5, "cuda_gb": 16.86}
{"step": 8400, "loss": 0.001928, "loss_mean50": 0.002377, "lr": 1.2744072131430981e-05, "grad_norm": 0.0779, "samples": 268800, "wall_s": 16714.5, "cuda_gb": 16.86}
{"step": 8420, "loss": 0.001737, "loss_mean50": 0.002254, "lr": 1.2693794792104049e-05, "grad_norm": 0.0358, "samples": 269440, "wall_s": 16753.5, "cuda_gb": 16.86}
{"step": 8440, "loss": 0.00235, "loss_mean50": 0.002104, "lr": 1.2643538726879303e-05, "grad_norm": 0.0359, "samples": 270080, "wall_s": 16792.5, "cuda_gb": 16.86}
{"step": 8460, "loss": 0.003262, "loss_mean50": 0.002372, "lr": 1.2593304948015212e-05, "grad_norm": 0.0321, "samples": 270720, "wall_s": 16831.4, "cuda_gb": 16.86}
{"step": 8480, "loss": 0.002598, "loss_mean50": 0.002372, "lr": 1.254309446732135e-05, "grad_norm": 0.0302, "samples": 271360, "wall_s": 16870.5, "cuda_gb": 16.86}
{"step": 8500, "loss": 0.00332, "loss_mean50": 0.002306, "lr": 1.2492908296138028e-05, "grad_norm": 0.0333, "samples": 272000, "wall_s": 16909.5, "cuda_gb": 16.86}
{"step": 8520, "loss": 0.001321, "loss_mean50": 0.002231, "lr": 1.2442747445315909e-05, "grad_norm": 0.0173, "samples": 272640, "wall_s": 16948.5, "cuda_gb": 16.86}
{"step": 8540, "loss": 0.001442, "loss_mean50": 0.0022, "lr": 1.2392612925195645e-05, "grad_norm": 0.0289, "samples": 273280, "wall_s": 16987.4, "cuda_gb": 16.86}
{"step": 8560, "loss": 0.001656, "loss_mean50": 0.002426, "lr": 1.2342505745587546e-05, "grad_norm": 0.0261, "samples": 273920, "wall_s": 17026.5, "cuda_gb": 16.86}
{"step": 8580, "loss": 0.004018, "loss_mean50": 0.002391, "lr": 1.2292426915751223e-05, "grad_norm": 0.0387, "samples": 274560, "wall_s": 17065.4, "cuda_gb": 16.86}
{"step": 8600, "loss": 0.001818, "loss_mean50": 0.002128, "lr": 1.2242377444375272e-05, "grad_norm": 0.0389, "samples": 275200, "wall_s": 17104.5, "cuda_gb": 16.86}
{"step": 8620, "loss": 0.001814, "loss_mean50": 0.001978, "lr": 1.2192358339556945e-05, "grad_norm": 0.0326, "samples": 275840, "wall_s": 17143.4, "cuda_gb": 16.86}
{"step": 8640, "loss": 0.001401, "loss_mean50": 0.001958, "lr": 1.2142370608781848e-05, "grad_norm": 0.0247, "samples": 276480, "wall_s": 17182.4, "cuda_gb": 16.86}
{"step": 8660, "loss": 0.002207, "loss_mean50": 0.001941, "lr": 1.209241525890365e-05, "grad_norm": 0.0205, "samples": 277120, "wall_s": 17221.3, "cuda_gb": 16.86}
{"step": 8680, "loss": 0.001215, "loss_mean50": 0.002019, "lr": 1.204249329612381e-05, "grad_norm": 0.0191, "samples": 277760, "wall_s": 17260.4, "cuda_gb": 16.86}
{"step": 8700, "loss": 0.002147, "loss_mean50": 0.00215, "lr": 1.1992605725971308e-05, "grad_norm": 0.0261, "samples": 278400, "wall_s": 17299.5, "cuda_gb": 16.86}
{"step": 8720, "loss": 0.000839, "loss_mean50": 0.00208, "lr": 1.194275355328237e-05, "grad_norm": 0.0208, "samples": 279040, "wall_s": 17338.6, "cuda_gb": 16.86}
{"step": 8740, "loss": 0.002022, "loss_mean50": 0.002134, "lr": 1.1892937782180261e-05, "grad_norm": 0.0227, "samples": 279680, "wall_s": 17377.7, "cuda_gb": 16.86}
{"step": 8760, "loss": 0.003016, "loss_mean50": 0.002265, "lr": 1.1843159416055037e-05, "grad_norm": 0.0339, "samples": 280320, "wall_s": 17416.8, "cuda_gb": 16.86}
{"step": 8780, "loss": 0.001373, "loss_mean50": 0.002213, "lr": 1.1793419457543346e-05, "grad_norm": 0.0173, "samples": 280960, "wall_s": 17456.0, "cuda_gb": 16.86}
{"step": 8800, "loss": 0.002548, "loss_mean50": 0.002194, "lr": 1.1743718908508222e-05, "grad_norm": 0.0307, "samples": 281600, "wall_s": 17495.0, "cuda_gb": 16.86}
{"step": 8820, "loss": 0.001064, "loss_mean50": 0.002243, "lr": 1.1694058770018925e-05, "grad_norm": 0.017, "samples": 282240, "wall_s": 17534.1, "cuda_gb": 16.86}
{"step": 8840, "loss": 0.002114, "loss_mean50": 0.002349, "lr": 1.1644440042330754e-05, "grad_norm": 0.0313, "samples": 282880, "wall_s": 17573.2, "cuda_gb": 16.86}
{"step": 8860, "loss": 0.001722, "loss_mean50": 0.002334, "lr": 1.1594863724864918e-05, "grad_norm": 0.0336, "samples": 283520, "wall_s": 17612.1, "cuda_gb": 16.86}
{"step": 8880, "loss": 0.000848, "loss_mean50": 0.00217, "lr": 1.1545330816188398e-05, "grad_norm": 0.0212, "samples": 284160, "wall_s": 17651.1, "cuda_gb": 16.86}
{"step": 8900, "loss": 0.003849, "loss_mean50": 0.002125, "lr": 1.1495842313993832e-05, "grad_norm": 0.0416, "samples": 284800, "wall_s": 17690.2, "cuda_gb": 16.86}
{"step": 8920, "loss": 0.001866, "loss_mean50": 0.002005, "lr": 1.1446399215079417e-05, "grad_norm": 0.0293, "samples": 285440, "wall_s": 17729.3, "cuda_gb": 16.86}
{"step": 8940, "loss": 0.001751, "loss_mean50": 0.0019, "lr": 1.1397002515328845e-05, "grad_norm": 0.0445, "samples": 286080, "wall_s": 17768.3, "cuda_gb": 16.86}
{"step": 8960, "loss": 0.001626, "loss_mean50": 0.001888, "lr": 1.1347653209691237e-05, "grad_norm": 0.0395, "samples": 286720, "wall_s": 17807.3, "cuda_gb": 16.86}
{"step": 8980, "loss": 0.0012, "loss_mean50": 0.002001, "lr": 1.1298352292161085e-05, "grad_norm": 0.019, "samples": 287360, "wall_s": 17846.2, "cuda_gb": 16.86}
{"step": 9000, "loss": 0.001539, "loss_mean50": 0.002244, "lr": 1.1249100755758268e-05, "grad_norm": 0.0229, "samples": 288000, "wall_s": 17885.2, "cuda_gb": 16.86}
{"step": 9000, "val_flow_mse": 0.001993}
{"step": 9020, "loss": 0.00266, "loss_mean50": 0.002169, "lr": 1.1199899592508024e-05, "grad_norm": 0.0268, "samples": 288640, "wall_s": 17953.1, "cuda_gb": 16.86}
{"step": 9040, "loss": 0.002791, "loss_mean50": 0.002143, "lr": 1.1150749793420958e-05, "grad_norm": 0.0377, "samples": 289280, "wall_s": 17992.1, "cuda_gb": 16.86}
{"step": 9060, "loss": 0.002714, "loss_mean50": 0.002127, "lr": 1.1101652348473113e-05, "grad_norm": 0.0459, "samples": 289920, "wall_s": 18031.2, "cuda_gb": 16.86}
{"step": 9080, "loss": 0.001395, "loss_mean50": 0.002127, "lr": 1.1052608246586007e-05, "grad_norm": 0.0448, "samples": 290560, "wall_s": 18070.1, "cuda_gb": 16.86}
{"step": 9100, "loss": 0.001854, "loss_mean50": 0.002092, "lr": 1.1003618475606737e-05, "grad_norm": 0.0366, "samples": 291200, "wall_s": 18109.1, "cuda_gb": 16.86}
{"step": 9120, "loss": 0.001447, "loss_mean50": 0.002053, "lr": 1.0954684022288039e-05, "grad_norm": 0.0213, "samples": 291840, "wall_s": 18148.1, "cuda_gb": 16.86}
{"step": 9140, "loss": 0.001391, "loss_mean50": 0.00204, "lr": 1.0905805872268463e-05, "grad_norm": 0.0236, "samples": 292480, "wall_s": 18187.1, "cuda_gb": 16.86}
{"step": 9160, "loss": 0.003422, "loss_mean50": 0.001902, "lr": 1.0856985010052487e-05, "grad_norm": 0.0408, "samples": 293120, "wall_s": 18226.1, "cuda_gb": 16.86}
{"step": 9180, "loss": 0.002145, "loss_mean50": 0.002044, "lr": 1.0808222418990693e-05, "grad_norm": 0.0416, "samples": 293760, "wall_s": 18265.1, "cuda_gb": 16.86}
{"step": 9200, "loss": 0.002169, "loss_mean50": 0.002123, "lr": 1.0759519081259975e-05, "grad_norm": 0.0251, "samples": 294400, "wall_s": 18304.1, "cuda_gb": 16.86}
{"step": 9220, "loss": 0.00271, "loss_mean50": 0.002169, "lr": 1.071087597784374e-05, "grad_norm": 0.0359, "samples": 295040, "wall_s": 18343.1, "cuda_gb": 16.86}
{"step": 9240, "loss": 0.001545, "loss_mean50": 0.002134, "lr": 1.0662294088512154e-05, "grad_norm": 0.0277, "samples": 295680, "wall_s": 18382.2, "cuda_gb": 16.86}
{"step": 9260, "loss": 0.001076, "loss_mean50": 0.002006, "lr": 1.0613774391802405e-05, "grad_norm": 0.0389, "samples": 296320, "wall_s": 18421.1, "cuda_gb": 16.86}
{"step": 9280, "loss": 0.001886, "loss_mean50": 0.002162, "lr": 1.0565317864999011e-05, "grad_norm": 0.0728, "samples": 296960, "wall_s": 18460.2, "cuda_gb": 16.86}
{"step": 9300, "loss": 0.001312, "loss_mean50": 0.002039, "lr": 1.0516925484114096e-05, "grad_norm": 0.0195, "samples": 297600, "wall_s": 18499.3, "cuda_gb": 16.86}
{"step": 9320, "loss": 0.00251, "loss_mean50": 0.001936, "lr": 1.0468598223867779e-05, "grad_norm": 0.0187, "samples": 298240, "wall_s": 18538.2, "cuda_gb": 16.86}
{"step": 9340, "loss": 0.001777, "loss_mean50": 0.001915, "lr": 1.04203370576685e-05, "grad_norm": 0.0652, "samples": 298880, "wall_s": 18577.4, "cuda_gb": 16.86}
{"step": 9360, "loss": 0.002297, "loss_mean50": 0.002052, "lr": 1.037214295759345e-05, "grad_norm": 0.0302, "samples": 299520, "wall_s": 18616.4, "cuda_gb": 16.86}
{"step": 9380, "loss": 0.002713, "loss_mean50": 0.002015, "lr": 1.0324016894368957e-05, "grad_norm": 0.0287, "samples": 300160, "wall_s": 18655.3, "cuda_gb": 16.86}
{"step": 9400, "loss": 0.001146, "loss_mean50": 0.00208, "lr": 1.0275959837350957e-05, "grad_norm": 0.0252, "samples": 300800, "wall_s": 18694.4, "cuda_gb": 16.86}
{"step": 9420, "loss": 0.002628, "loss_mean50": 0.002315, "lr": 1.0227972754505468e-05, "grad_norm": 0.0404, "samples": 301440, "wall_s": 18733.5, "cuda_gb": 16.86}
{"step": 9440, "loss": 0.002677, "loss_mean50": 0.00234, "lr": 1.0180056612389072e-05, "grad_norm": 0.0324, "samples": 302080, "wall_s": 18772.5, "cuda_gb": 16.86}
{"step": 9460, "loss": 0.001623, "loss_mean50": 0.002323, "lr": 1.0132212376129465e-05, "grad_norm": 0.0239, "samples": 302720, "wall_s": 18811.6, "cuda_gb": 16.86}
{"step": 9480, "loss": 0.001618, "loss_mean50": 0.002112, "lr": 1.0084441009406028e-05, "grad_norm": 0.0228, "samples": 303360, "wall_s": 18850.6, "cuda_gb": 16.86}
{"step": 9500, "loss": 0.001283, "loss_mean50": 0.001877, "lr": 1.0036743474430393e-05, "grad_norm": 0.0246, "samples": 304000, "wall_s": 18889.6, "cuda_gb": 16.86}
{"step": 9520, "loss": 0.002863, "loss_mean50": 0.002062, "lr": 9.989120731927065e-06, "grad_norm": 0.0442, "samples": 304640, "wall_s": 18928.7, "cuda_gb": 16.86}
{"step": 9540, "loss": 0.001537, "loss_mean50": 0.002038, "lr": 9.941573741114099e-06, "grad_norm": 0.0253, "samples": 305280, "wall_s": 18967.8, "cuda_gb": 16.86}
{"step": 9560, "loss": 0.001736, "loss_mean50": 0.002048, "lr": 9.894103459683726e-06, "grad_norm": 0.0204, "samples": 305920, "wall_s": 19006.9, "cuda_gb": 16.86}
{"step": 9580, "loss": 0.002433, "loss_mean50": 0.001971, "lr": 9.846710843783125e-06, "grad_norm": 0.0352, "samples": 306560, "wall_s": 19045.9, "cuda_gb": 16.86}
{"step": 9600, "loss": 0.001555, "loss_mean50": 0.002051, "lr": 9.799396847995121e-06, "grad_norm": 0.0251, "samples": 307200, "wall_s": 19085.0, "cuda_gb": 16.86}
{"step": 9620, "loss": 0.005813, "loss_mean50": 0.002193, "lr": 9.752162425318972e-06, "grad_norm": 0.0484, "samples": 307840, "wall_s": 19124.3, "cuda_gb": 16.86}
{"step": 9640, "loss": 0.001849, "loss_mean50": 0.001945, "lr": 9.705008527151176e-06, "grad_norm": 0.0317, "samples": 308480, "wall_s": 19163.5, "cuda_gb": 16.86}
{"step": 9660, "loss": 0.003702, "loss_mean50": 0.002051, "lr": 9.657936103266308e-06, "grad_norm": 0.0419, "samples": 309120, "wall_s": 19202.6, "cuda_gb": 16.86}
{"step": 9680, "loss": 0.002075, "loss_mean50": 0.001894, "lr": 9.61094610179788e-06, "grad_norm": 0.0348, "samples": 309760, "wall_s": 19241.8, "cuda_gb": 16.86}
{"step": 9700, "loss": 0.002045, "loss_mean50": 0.00199, "lr": 9.564039469219244e-06, "grad_norm": 0.0343, "samples": 310400, "wall_s": 19280.8, "cuda_gb": 16.86}
{"step": 9720, "loss": 0.001008, "loss_mean50": 0.002021, "lr": 9.517217150324556e-06, "grad_norm": 0.0248, "samples": 311040, "wall_s": 19319.9, "cuda_gb": 16.86}
{"step": 9740, "loss": 0.002574, "loss_mean50": 0.001894, "lr": 9.470480088209707e-06, "grad_norm": 0.0365, "samples": 311680, "wall_s": 19359.0, "cuda_gb": 16.86}
{"step": 9760, "loss": 0.001175, "loss_mean50": 0.001745, "lr": 9.423829224253347e-06, "grad_norm": 0.0221, "samples": 312320, "wall_s": 19397.9, "cuda_gb": 16.86}
{"step": 9780, "loss": 0.001864, "loss_mean50": 0.001865, "lr": 9.377265498097926e-06, "grad_norm": 0.0289, "samples": 312960, "wall_s": 19437.0, "cuda_gb": 16.86}
{"step": 9800, "loss": 0.002321, "loss_mean50": 0.001837, "lr": 9.330789847630767e-06, "grad_norm": 0.0309, "samples": 313600, "wall_s": 19475.9, "cuda_gb": 16.86}
{"step": 9820, "loss": 0.002665, "loss_mean50": 0.001826, "lr": 9.284403208965151e-06, "grad_norm": 0.0292, "samples": 314240, "wall_s": 19514.9, "cuda_gb": 16.86}
{"step": 9840, "loss": 0.000594, "loss_mean50": 0.00182, "lr": 9.238106516421504e-06, "grad_norm": 0.0168, "samples": 314880, "wall_s": 19553.9, "cuda_gb": 16.86}
{"step": 9860, "loss": 0.002535, "loss_mean50": 0.002049, "lr": 9.191900702508551e-06, "grad_norm": 0.0367, "samples": 315520, "wall_s": 19593.0, "cuda_gb": 16.86}
{"step": 9880, "loss": 0.002347, "loss_mean50": 0.002146, "lr": 9.145786697904526e-06, "grad_norm": 0.0292, "samples": 316160, "wall_s": 19632.1, "cuda_gb": 16.86}
{"step": 9900, "loss": 0.001727, "loss_mean50": 0.002104, "lr": 9.099765431438454e-06, "grad_norm": 0.0291, "samples": 316800, "wall_s": 19671.1, "cuda_gb": 16.86}
{"step": 9920, "loss": 0.002089, "loss_mean50": 0.002055, "lr": 9.053837830071418e-06, "grad_norm": 0.0312, "samples": 317440, "wall_s": 19710.1, "cuda_gb": 16.86}
{"step": 9940, "loss": 0.003025, "loss_mean50": 0.002135, "lr": 9.008004818877899e-06, "grad_norm": 0.0368, "samples": 318080, "wall_s": 19749.2, "cuda_gb": 16.86}
{"step": 9960, "loss": 0.00128, "loss_mean50": 0.001974, "lr": 8.96226732102714e-06, "grad_norm": 0.0207, "samples": 318720, "wall_s": 19788.3, "cuda_gb": 16.86}
{"step": 9980, "loss": 0.001204, "loss_mean50": 0.002047, "lr": 8.91662625776456e-06, "grad_norm": 0.0217, "samples": 319360, "wall_s": 19827.2, "cuda_gb": 16.86}
{"step": 10000, "loss": 0.001862, "loss_mean50": 0.001893, "lr": 8.87108254839317e-06, "grad_norm": 0.0463, "samples": 320000, "wall_s": 19866.2, "cuda_gb": 16.86}
{"step": 10000, "val_flow_mse": 0.001747, "val_action_mae": 0.01365, "best": true}
{"step": 10020, "loss": 0.003076, "loss_mean50": 0.001847, "lr": 8.825637110255108e-06, "grad_norm": 0.0489, "samples": 320640, "wall_s": 19997.5, "cuda_gb": 16.86}
{"step": 10040, "loss": 0.001706, "loss_mean50": 0.001967, "lr": 8.78029085871311e-06, "grad_norm": 0.0316, "samples": 321280, "wall_s": 20036.5, "cuda_gb": 16.86}
{"step": 10060, "loss": 0.001411, "loss_mean50": 0.001953, "lr": 8.7350447071321e-06, "grad_norm": 0.0224, "samples": 321920, "wall_s": 20075.5, "cuda_gb": 16.86}
{"step": 10080, "loss": 0.003891, "loss_mean50": 0.00203, "lr": 8.68989956686079e-06, "grad_norm": 0.0541, "samples": 322560, "wall_s": 20114.5, "cuda_gb": 16.86}
{"step": 10100, "loss": 0.00164, "loss_mean50": 0.00204, "lr": 8.644856347213308e-06, "grad_norm": 0.0297, "samples": 323200, "wall_s": 20153.5, "cuda_gb": 16.86}
{"step": 10120, "loss": 0.001531, "loss_mean50": 0.002017, "lr": 8.59991595545091e-06, "grad_norm": 0.0216, "samples": 323840, "wall_s": 20192.6, "cuda_gb": 16.86}
{"step": 10140, "loss": 0.002367, "loss_mean50": 0.001965, "lr": 8.555079296763687e-06, "grad_norm": 0.0288, "samples": 324480, "wall_s": 20231.5, "cuda_gb": 16.86}
{"step": 10160, "loss": 0.002645, "loss_mean50": 0.001879, "lr": 8.510347274252333e-06, "grad_norm": 0.0242, "samples": 325120, "wall_s": 20270.4, "cuda_gb": 16.86}
{"step": 10180, "loss": 0.001791, "loss_mean50": 0.002127, "lr": 8.465720788909959e-06, "grad_norm": 0.03, "samples": 325760, "wall_s": 20309.4, "cuda_gb": 16.86}
{"step": 10200, "loss": 0.001679, "loss_mean50": 0.002033, "lr": 8.421200739603955e-06, "grad_norm": 0.0354, "samples": 326400, "wall_s": 20348.4, "cuda_gb": 16.86}
{"step": 10220, "loss": 0.001674, "loss_mean50": 0.001993, "lr": 8.376788023057854e-06, "grad_norm": 0.0223, "samples": 327040, "wall_s": 20387.3, "cuda_gb": 16.86}
{"step": 10240, "loss": 0.001578, "loss_mean50": 0.001861, "lr": 8.33248353383331e-06, "grad_norm": 0.0419, "samples": 327680, "wall_s": 20426.2, "cuda_gb": 16.86}
{"step": 10260, "loss": 0.00141, "loss_mean50": 0.001876, "lr": 8.288288164312053e-06, "grad_norm": 0.0224, "samples": 328320, "wall_s": 20465.2, "cuda_gb": 16.86}
{"step": 10280, "loss": 0.002209, "loss_mean50": 0.001782, "lr": 8.244202804677917e-06, "grad_norm": 0.0212, "samples": 328960, "wall_s": 20504.2, "cuda_gb": 16.86}
{"step": 10300, "loss": 0.001943, "loss_mean50": 0.001706, "lr": 8.200228342898927e-06, "grad_norm": 0.0271, "samples": 329600, "wall_s": 20543.2, "cuda_gb": 16.86}
{"step": 10320, "loss": 0.00177, "loss_mean50": 0.001859, "lr": 8.1563656647094e-06, "grad_norm": 0.0259, "samples": 330240, "wall_s": 20582.4, "cuda_gb": 16.86}
{"step": 10340, "loss": 0.001452, "loss_mean50": 0.001904, "lr": 8.112615653592086e-06, "grad_norm": 0.0303, "samples": 330880, "wall_s": 20621.5, "cuda_gb": 16.86}
{"step": 10360, "loss": 0.001454, "loss_mean50": 0.001892, "lr": 8.06897919076042e-06, "grad_norm": 0.0268, "samples": 331520, "wall_s": 20660.5, "cuda_gb": 16.86}
{"step": 10380, "loss": 0.001517, "loss_mean50": 0.001869, "lr": 8.025457155140734e-06, "grad_norm": 0.0245, "samples": 332160, "wall_s": 20699.5, "cuda_gb": 16.86}
{"step": 10400, "loss": 0.001029, "loss_mean50": 0.001962, "lr": 7.982050423354554e-06, "grad_norm": 0.025, "samples": 332800, "wall_s": 20738.5, "cuda_gb": 16.86}
{"step": 10420, "loss": 0.001831, "loss_mean50": 0.002164, "lr": 7.938759869700983e-06, "grad_norm": 0.0214, "samples": 333440, "wall_s": 20777.4, "cuda_gb": 16.86}
{"step": 10440, "loss": 0.005248, "loss_mean50": 0.002103, "lr": 7.895586366139032e-06, "grad_norm": 0.041, "samples": 334080, "wall_s": 20816.4, "cuda_gb": 16.86}
{"step": 10460, "loss": 0.002594, "loss_mean50": 0.00194, "lr": 7.852530782270106e-06, "grad_norm": 0.0345, "samples": 334720, "wall_s": 20855.5, "cuda_gb": 16.86}
{"step": 10480, "loss": 0.00226, "loss_mean50": 0.001895, "lr": 7.809593985320462e-06, "grad_norm": 0.0477, "samples": 335360, "wall_s": 20894.3, "cuda_gb": 16.86}
{"step": 10500, "loss": 0.001833, "loss_mean50": 0.001845, "lr": 7.766776840123753e-06, "grad_norm": 0.0381, "samples": 336000, "wall_s": 20933.4, "cuda_gb": 16.86}
{"step": 10520, "loss": 0.002649, "loss_mean50": 0.001843, "lr": 7.724080209103593e-06, "grad_norm": 0.088, "samples": 336640, "wall_s": 20972.4, "cuda_gb": 16.86}
{"step": 10540, "loss": 0.002178, "loss_mean50": 0.001906, "lr": 7.681504952256218e-06, "grad_norm": 0.0294, "samples": 337280, "wall_s": 21011.4, "cuda_gb": 16.86}
{"step": 10560, "loss": 0.000879, "loss_mean50": 0.001862, "lr": 7.639051927133128e-06, "grad_norm": 0.0237, "samples": 337920, "wall_s": 21050.4, "cuda_gb": 16.86}
{"step": 10580, "loss": 0.001436, "loss_mean50": 0.001665, "lr": 7.596721988823835e-06, "grad_norm": 0.0325, "samples": 338560, "wall_s": 21089.4, "cuda_gb": 16.86}
{"step": 10600, "loss": 0.001499, "loss_mean50": 0.001766, "lr": 7.554515989938631e-06, "grad_norm": 0.0246, "samples": 339200, "wall_s": 21128.4, "cuda_gb": 16.86}
{"step": 10620, "loss": 0.001494, "loss_mean50": 0.001928, "lr": 7.512434780591415e-06, "grad_norm": 0.029, "samples": 339840, "wall_s": 21167.4, "cuda_gb": 16.86}
{"step": 10640, "loss": 0.003764, "loss_mean50": 0.001898, "lr": 7.470479208382591e-06, "grad_norm": 0.0438, "samples": 340480, "wall_s": 21206.4, "cuda_gb": 16.86}
{"step": 10660, "loss": 0.000558, "loss_mean50": 0.001694, "lr": 7.4286501183819604e-06, "grad_norm": 0.0172, "samples": 341120, "wall_s": 21245.4, "cuda_gb": 16.86}
{"step": 10680, "loss": 0.002726, "loss_mean50": 0.001796, "lr": 7.386948353111717e-06, "grad_norm": 0.0625, "samples": 341760, "wall_s": 21284.3, "cuda_gb": 16.86}
{"step": 10700, "loss": 0.001278, "loss_mean50": 0.001721, "lr": 7.345374752529491e-06, "grad_norm": 0.0235, "samples": 342400, "wall_s": 21323.2, "cuda_gb": 16.86}
{"step": 10720, "loss": 0.001121, "loss_mean50": 0.001818, "lr": 7.30393015401141e-06, "grad_norm": 0.0145, "samples": 343040, "wall_s": 21362.1, "cuda_gb": 16.86}
{"step": 10740, "loss": 0.001884, "loss_mean50": 0.001795, "lr": 7.2626153923352316e-06, "grad_norm": 0.036, "samples": 343680, "wall_s": 21401.2, "cuda_gb": 16.86}
{"step": 10760, "loss": 0.00109, "loss_mean50": 0.001711, "lr": 7.221431299663555e-06, "grad_norm": 0.0607, "samples": 344320, "wall_s": 21440.2, "cuda_gb": 16.86}
{"step": 10780, "loss": 0.001035, "loss_mean50": 0.001753, "lr": 7.18037870552703e-06, "grad_norm": 0.0303, "samples": 344960, "wall_s": 21479.1, "cuda_gb": 16.86}
{"step": 10800, "loss": 0.003039, "loss_mean50": 0.001684, "lr": 7.139458436807661e-06, "grad_norm": 0.0282, "samples": 345600, "wall_s": 21518.0, "cuda_gb": 16.86}
{"step": 10820, "loss": 0.001677, "loss_mean50": 0.001862, "lr": 7.098671317722161e-06, "grad_norm": 0.0819, "samples": 346240, "wall_s": 21557.0, "cuda_gb": 16.86}
{"step": 10840, "loss": 0.001906, "loss_mean50": 0.001926, "lr": 7.05801816980533e-06, "grad_norm": 0.0359, "samples": 346880, "wall_s": 21596.0, "cuda_gb": 16.86}
{"step": 10860, "loss": 0.000486, "loss_mean50": 0.002022, "lr": 7.0174998118935295e-06, "grad_norm": 0.0144, "samples": 347520, "wall_s": 21635.1, "cuda_gb": 16.86}
{"step": 10880, "loss": 0.001562, "loss_mean50": 0.001977, "lr": 6.977117060108165e-06, "grad_norm": 0.0516, "samples": 348160, "wall_s": 21674.1, "cuda_gb": 16.86}
{"step": 10900, "loss": 0.002643, "loss_mean50": 0.001927, "lr": 6.9368707278392706e-06, "grad_norm": 0.0264, "samples": 348800, "wall_s": 21713.1, "cuda_gb": 16.86}
{"step": 10920, "loss": 0.003109, "loss_mean50": 0.001855, "lr": 6.896761625729117e-06, "grad_norm": 0.0301, "samples": 349440, "wall_s": 21752.2, "cuda_gb": 16.86}
{"step": 10940, "loss": 0.000814, "loss_mean50": 0.001706, "lr": 6.856790561655879e-06, "grad_norm": 0.0219, "samples": 350080, "wall_s": 21791.1, "cuda_gb": 16.86}
{"step": 10960, "loss": 0.00356, "loss_mean50": 0.001908, "lr": 6.816958340717365e-06, "grad_norm": 0.1168, "samples": 350720, "wall_s": 21830.1, "cuda_gb": 16.86}
{"step": 10980, "loss": 0.002751, "loss_mean50": 0.001852, "lr": 6.777265765214818e-06, "grad_norm": 0.0702, "samples": 351360, "wall_s": 21869.0, "cuda_gb": 16.86}
{"step": 11000, "loss": 0.00123, "loss_mean50": 0.001928, "lr": 6.737713634636714e-06, "grad_norm": 0.066, "samples": 352000, "wall_s": 21908.1, "cuda_gb": 16.86}
{"step": 11000, "val_flow_mse": 0.001681}
{"step": 11020, "loss": 0.000936, "loss_mean50": 0.001856, "lr": 6.698302745642702e-06, "grad_norm": 0.022, "samples": 352640, "wall_s": 21975.9, "cuda_gb": 16.86}
{"step": 11040, "loss": 0.001893, "loss_mean50": 0.001728, "lr": 6.659033892047546e-06, "grad_norm": 0.0416, "samples": 353280, "wall_s": 22014.8, "cuda_gb": 16.86}
{"step": 11060, "loss": 0.001228, "loss_mean50": 0.001607, "lr": 6.619907864805116e-06, "grad_norm": 0.0457, "samples": 353920, "wall_s": 22053.9, "cuda_gb": 16.86}
{"step": 11080, "loss": 0.001761, "loss_mean50": 0.00176, "lr": 6.580925451992481e-06, "grad_norm": 0.0498, "samples": 354560, "wall_s": 22092.8, "cuda_gb": 16.86}
{"step": 11100, "loss": 0.00095, "loss_mean50": 0.001929, "lr": 6.542087438794027e-06, "grad_norm": 0.0146, "samples": 355200, "wall_s": 22131.7, "cuda_gb": 16.86}
{"step": 11120, "loss": 0.00145, "loss_mean50": 0.001837, "lr": 6.503394607485626e-06, "grad_norm": 0.0296, "samples": 355840, "wall_s": 22170.7, "cuda_gb": 16.86}
{"step": 11140, "loss": 0.001085, "loss_mean50": 0.00174, "lr": 6.464847737418916e-06, "grad_norm": 0.0169, "samples": 356480, "wall_s": 22209.6, "cuda_gb": 16.86}
{"step": 11160, "loss": 0.001037, "loss_mean50": 0.001673, "lr": 6.426447605005567e-06, "grad_norm": 0.0271, "samples": 357120, "wall_s": 22248.6, "cuda_gb": 16.86}
{"step": 11180, "loss": 0.00258, "loss_mean50": 0.001916, "lr": 6.388194983701658e-06, "grad_norm": 0.0808, "samples": 357760, "wall_s": 22287.6, "cuda_gb": 16.86}
{"step": 11200, "loss": 0.002073, "loss_mean50": 0.001981, "lr": 6.350090643992096e-06, "grad_norm": 0.0273, "samples": 358400, "wall_s": 22326.5, "cuda_gb": 16.86}
{"step": 11220, "loss": 0.001743, "loss_mean50": 0.001938, "lr": 6.312135353375108e-06, "grad_norm": 0.0455, "samples": 359040, "wall_s": 22365.6, "cuda_gb": 16.86}
{"step": 11240, "loss": 0.002496, "loss_mean50": 0.001905, "lr": 6.274329876346763e-06, "grad_norm": 0.0288, "samples": 359680, "wall_s": 22404.7, "cuda_gb": 16.86}
{"step": 11260, "loss": 0.001268, "loss_mean50": 0.001945, "lr": 6.2366749743855796e-06, "grad_norm": 0.02, "samples": 360320, "wall_s": 22443.7, "cuda_gb": 16.86}
{"step": 11280, "loss": 0.0024, "loss_mean50": 0.001753, "lr": 6.199171405937199e-06, "grad_norm": 0.0624, "samples": 360960, "wall_s": 22482.7, "cuda_gb": 16.86}
{"step": 11300, "loss": 0.002314, "loss_mean50": 0.001933, "lr": 6.161819926399091e-06, "grad_norm": 0.0385, "samples": 361600, "wall_s": 22521.7, "cuda_gb": 16.86}
{"step": 11320, "loss": 0.001541, "loss_mean50": 0.001877, "lr": 6.124621288105361e-06, "grad_norm": 0.0328, "samples": 362240, "wall_s": 22560.7, "cuda_gb": 16.86}
{"step": 11340, "loss": 0.001182, "loss_mean50": 0.001921, "lr": 6.0875762403115745e-06, "grad_norm": 0.0186, "samples": 362880, "wall_s": 22599.7, "cuda_gb": 16.86}
{"step": 11360, "loss": 0.001477, "loss_mean50": 0.001803, "lr": 6.050685529179673e-06, "grad_norm": 0.0213, "samples": 363520, "wall_s": 22638.7, "cuda_gb": 16.86}
{"step": 11380, "loss": 0.00129, "loss_mean50": 0.001738, "lr": 6.013949897762954e-06, "grad_norm": 0.0191, "samples": 364160, "wall_s": 22677.7, "cuda_gb": 16.86}
{"step": 11400, "loss": 0.001575, "loss_mean50": 0.001645, "lr": 5.977370085991092e-06, "grad_norm": 0.0369, "samples": 364800, "wall_s": 22716.6, "cuda_gb": 16.86}
{"step": 11420, "loss": 0.001631, "loss_mean50": 0.001662, "lr": 5.9409468306552325e-06, "grad_norm": 0.0378, "samples": 365440, "wall_s": 22755.6, "cuda_gb": 16.86}
{"step": 11440, "loss": 0.002029, "loss_mean50": 0.001707, "lr": 5.904680865393181e-06, "grad_norm": 0.027, "samples": 366080, "wall_s": 22794.6, "cuda_gb": 16.86}
{"step": 11460, "loss": 0.001084, "loss_mean50": 0.001853, "lr": 5.868572920674581e-06, "grad_norm": 0.0224, "samples": 366720, "wall_s": 22833.6, "cuda_gb": 16.86}
{"step": 11480, "loss": 0.001513, "loss_mean50": 0.001744, "lr": 5.83262372378623e-06, "grad_norm": 0.0189, "samples": 367360, "wall_s": 22873.4, "cuda_gb": 16.86}
{"step": 11500, "loss": 0.000696, "loss_mean50": 0.001712, "lr": 5.7968339988174385e-06, "grad_norm": 0.0204, "samples": 368000, "wall_s": 22912.6, "cuda_gb": 16.86}
{"step": 11520, "loss": 0.001439, "loss_mean50": 0.001779, "lr": 5.7612044666454034e-06, "grad_norm": 0.0252, "samples": 368640, "wall_s": 22952.1, "cuda_gb": 16.86}
{"step": 11540, "loss": 0.001763, "loss_mean50": 0.001731, "lr": 5.7257358449207435e-06, "grad_norm": 0.025, "samples": 369280, "wall_s": 22991.3, "cuda_gb": 16.86}
{"step": 11560, "loss": 0.000818, "loss_mean50": 0.001791, "lr": 5.690428848053e-06, "grad_norm": 0.0264, "samples": 369920, "wall_s": 23030.5, "cuda_gb": 16.86}
{"step": 11580, "loss": 0.001613, "loss_mean50": 0.001781, "lr": 5.655284187196265e-06, "grad_norm": 0.0192, "samples": 370560, "wall_s": 23069.4, "cuda_gb": 16.86}
{"step": 11600, "loss": 0.002449, "loss_mean50": 0.00183, "lr": 5.620302570234867e-06, "grad_norm": 0.0407, "samples": 371200, "wall_s": 23108.4, "cuda_gb": 16.86}
{"step": 11620, "loss": 0.002682, "loss_mean50": 0.001636, "lr": 5.5854847017690945e-06, "grad_norm": 0.0233, "samples": 371840, "wall_s": 23147.4, "cuda_gb": 16.86}
{"step": 11640, "loss": 0.001283, "loss_mean50": 0.001711, "lr": 5.550831283100999e-06, "grad_norm": 0.021, "samples": 372480, "wall_s": 23186.2, "cuda_gb": 16.86}
{"step": 11660, "loss": 0.002823, "loss_mean50": 0.001695, "lr": 5.516343012220304e-06, "grad_norm": 0.0973, "samples": 373120, "wall_s": 23225.1, "cuda_gb": 16.86}
{"step": 11680, "loss": 0.000862, "loss_mean50": 0.001802, "lr": 5.482020583790311e-06, "grad_norm": 0.0179, "samples": 373760, "wall_s": 23264.1, "cuda_gb": 16.86}
{"step": 11700, "loss": 0.00201, "loss_mean50": 0.001821, "lr": 5.447864689133917e-06, "grad_norm": 0.0271, "samples": 374400, "wall_s": 23303.0, "cuda_gb": 16.86}
{"step": 11720, "loss": 0.001462, "loss_mean50": 0.001745, "lr": 5.413876016219707e-06, "grad_norm": 0.0285, "samples": 375040, "wall_s": 23342.0, "cuda_gb": 16.86}
{"step": 11740, "loss": 0.001304, "loss_mean50": 0.001887, "lr": 5.380055249648071e-06, "grad_norm": 0.0311, "samples": 375680, "wall_s": 23380.9, "cuda_gb": 16.86}
{"step": 11760, "loss": 0.001863, "loss_mean50": 0.001755, "lr": 5.346403070637427e-06, "grad_norm": 0.0316, "samples": 376320, "wall_s": 23419.8, "cuda_gb": 16.86}
{"step": 11780, "loss": 0.001959, "loss_mean50": 0.001734, "lr": 5.312920157010506e-06, "grad_norm": 0.0258, "samples": 376960, "wall_s": 23458.6, "cuda_gb": 16.86}
{"step": 11800, "loss": 0.000778, "loss_mean50": 0.001788, "lr": 5.2796071831806846e-06, "grad_norm": 0.0256, "samples": 377600, "wall_s": 23497.6, "cuda_gb": 16.86}
{"step": 11820, "loss": 0.001136, "loss_mean50": 0.001765, "lr": 5.2464648201384225e-06, "grad_norm": 0.0179, "samples": 378240, "wall_s": 23536.5, "cuda_gb": 16.86}
{"step": 11840, "loss": 0.002717, "loss_mean50": 0.001808, "lr": 5.213493735437722e-06, "grad_norm": 0.0271, "samples": 378880, "wall_s": 23575.3, "cuda_gb": 16.86}
{"step": 11860, "loss": 0.002446, "loss_mean50": 0.001791, "lr": 5.180694593182696e-06, "grad_norm": 0.0462, "samples": 379520, "wall_s": 23614.0, "cuda_gb": 16.86}
{"step": 11880, "loss": 0.001741, "loss_mean50": 0.00185, "lr": 5.148068054014195e-06, "grad_norm": 0.0818, "samples": 380160, "wall_s": 23652.9, "cuda_gb": 16.86}
{"step": 11900, "loss": 0.001641, "loss_mean50": 0.001807, "lr": 5.115614775096492e-06, "grad_norm": 0.0249, "samples": 380800, "wall_s": 23691.8, "cuda_gb": 16.86}
{"step": 11920, "loss": 0.001031, "loss_mean50": 0.001816, "lr": 5.083335410104047e-06, "grad_norm": 0.0147, "samples": 381440, "wall_s": 23730.6, "cuda_gb": 16.86}
{"step": 11940, "loss": 0.003377, "loss_mean50": 0.001904, "lr": 5.051230609208352e-06, "grad_norm": 0.0675, "samples": 382080, "wall_s": 23769.5, "cuda_gb": 16.86}
{"step": 11960, "loss": 0.001638, "loss_mean50": 0.00164, "lr": 5.019301019064816e-06, "grad_norm": 0.0285, "samples": 382720, "wall_s": 23808.5, "cuda_gb": 16.86}
{"step": 11980, "loss": 0.001029, "loss_mean50": 0.001514, "lr": 4.987547282799753e-06, "grad_norm": 0.0158, "samples": 383360, "wall_s": 23847.6, "cuda_gb": 16.86}
{"step": 12000, "loss": 0.001504, "loss_mean50": 0.001528, "lr": 4.955970039997429e-06, "grad_norm": 0.0182, "samples": 384000, "wall_s": 23886.5, "cuda_gb": 16.86}
{"step": 12000, "val_flow_mse": 0.001672}
{"step": 12020, "loss": 0.001853, "loss_mean50": 0.001645, "lr": 4.924569926687178e-06, "grad_norm": 0.0409, "samples": 384640, "wall_s": 23954.3, "cuda_gb": 16.86}
{"step": 12040, "loss": 0.001672, "loss_mean50": 0.001626, "lr": 4.893347575330571e-06, "grad_norm": 0.0318, "samples": 385280, "wall_s": 23993.3, "cuda_gb": 16.86}
{"step": 12060, "loss": 0.002027, "loss_mean50": 0.001686, "lr": 4.862303614808722e-06, "grad_norm": 0.0467, "samples": 385920, "wall_s": 24032.2, "cuda_gb": 16.86}
{"step": 12080, "loss": 0.003174, "loss_mean50": 0.00171, "lr": 4.831438670409574e-06, "grad_norm": 0.0315, "samples": 386560, "wall_s": 24071.2, "cuda_gb": 16.86}
{"step": 12100, "loss": 0.001439, "loss_mean50": 0.0017, "lr": 4.800753363815327e-06, "grad_norm": 0.0228, "samples": 387200, "wall_s": 24110.1, "cuda_gb": 16.86}
{"step": 12120, "loss": 0.002734, "loss_mean50": 0.001532, "lr": 4.770248313089926e-06, "grad_norm": 0.046, "samples": 387840, "wall_s": 24149.0, "cuda_gb": 16.86}
{"step": 12140, "loss": 0.001507, "loss_mean50": 0.001594, "lr": 4.739924132666585e-06, "grad_norm": 0.0194, "samples": 388480, "wall_s": 24187.9, "cuda_gb": 16.86}
{"step": 12160, "loss": 0.003845, "loss_mean50": 0.001716, "lr": 4.709781433335432e-06, "grad_norm": 0.0472, "samples": 389120, "wall_s": 24226.8, "cuda_gb": 16.86}
{"step": 12180, "loss": 0.001009, "loss_mean50": 0.001766, "lr": 4.6798208222311985e-06, "grad_norm": 0.0218, "samples": 389760, "wall_s": 24265.8, "cuda_gb": 16.86}
{"step": 12200, "loss": 0.000793, "loss_mean50": 0.001778, "lr": 4.6500429028209904e-06, "grad_norm": 0.0146, "samples": 390400, "wall_s": 24304.7, "cuda_gb": 16.86}
{"step": 12220, "loss": 0.002066, "loss_mean50": 0.001688, "lr": 4.620448274892142e-06, "grad_norm": 0.0334, "samples": 391040, "wall_s": 24343.7, "cuda_gb": 16.86}
{"step": 12240, "loss": 0.004108, "loss_mean50": 0.001599, "lr": 4.591037534540117e-06, "grad_norm": 0.0326, "samples": 391680, "wall_s": 24382.7, "cuda_gb": 16.86}
{"step": 12260, "loss": 0.002178, "loss_mean50": 0.001635, "lr": 4.56181127415652e-06, "grad_norm": 0.0299, "samples": 392320, "wall_s": 24421.7, "cuda_gb": 16.86}
{"step": 12280, "loss": 0.001359, "loss_mean50": 0.001767, "lr": 4.5327700824171575e-06, "grad_norm": 0.0379, "samples": 392960, "wall_s": 24460.7, "cuda_gb": 16.86}
{"step": 12300, "loss": 0.000963, "loss_mean50": 0.001692, "lr": 4.5039145442701735e-06, "grad_norm": 0.0172, "samples": 393600, "wall_s": 24499.7, "cuda_gb": 16.86}
{"step": 12320, "loss": 0.0014, "loss_mean50": 0.001686, "lr": 4.475245240924278e-06, "grad_norm": 0.022, "samples": 394240, "wall_s": 24538.7, "cuda_gb": 16.86}
{"step": 12340, "loss": 0.001526, "loss_mean50": 0.001691, "lr": 4.446762749837041e-06, "grad_norm": 0.03, "samples": 394880, "wall_s": 24577.7, "cuda_gb": 16.86}
{"step": 12360, "loss": 0.001905, "loss_mean50": 0.001748, "lr": 4.418467644703254e-06, "grad_norm": 0.0389, "samples": 395520, "wall_s": 24616.8, "cuda_gb": 16.86}
{"step": 12380, "loss": 0.001422, "loss_mean50": 0.001572, "lr": 4.3903604954433725e-06, "grad_norm": 0.0217, "samples": 396160, "wall_s": 24655.7, "cuda_gb": 16.86}
{"step": 12400, "loss": 0.000967, "loss_mean50": 0.001589, "lr": 4.362441868192056e-06, "grad_norm": 0.0169, "samples": 396800, "wall_s": 24694.7, "cuda_gb": 16.86}
{"step": 12420, "loss": 0.001969, "loss_mean50": 0.001592, "lr": 4.334712325286726e-06, "grad_norm": 0.0261, "samples": 397440, "wall_s": 24733.7, "cuda_gb": 16.86}
{"step": 12440, "loss": 0.001071, "loss_mean50": 0.001576, "lr": 4.307172425256291e-06, "grad_norm": 0.0383, "samples": 398080, "wall_s": 24772.7, "cuda_gb": 16.86}
{"step": 12460, "loss": 0.001183, "loss_mean50": 0.001529, "lr": 4.279822722809855e-06, "grad_norm": 0.0217, "samples": 398720, "wall_s": 24811.6, "cuda_gb": 16.86}
{"step": 12480, "loss": 0.002273, "loss_mean50": 0.001653, "lr": 4.25266376882556e-06, "grad_norm": 0.0293, "samples": 399360, "wall_s": 24850.6, "cuda_gb": 16.86}
{"step": 12500, "loss": 0.001514, "loss_mean50": 0.001722, "lr": 4.2256961103394905e-06, "grad_norm": 0.0573, "samples": 400000, "wall_s": 24889.6, "cuda_gb": 16.86}
{"step": 12520, "loss": 0.001169, "loss_mean50": 0.001719, "lr": 4.198920290534661e-06, "grad_norm": 0.0473, "samples": 400640, "wall_s": 24968.8, "cuda_gb": 16.86}
{"step": 12540, "loss": 0.000697, "loss_mean50": 0.001728, "lr": 4.1723368487300514e-06, "grad_norm": 0.0168, "samples": 401280, "wall_s": 25007.9, "cuda_gb": 16.86}
{"step": 12560, "loss": 0.001202, "loss_mean50": 0.001832, "lr": 4.145946320369784e-06, "grad_norm": 0.0214, "samples": 401920, "wall_s": 25046.9, "cuda_gb": 16.86}
{"step": 12580, "loss": 0.001678, "loss_mean50": 0.001772, "lr": 4.119749237012297e-06, "grad_norm": 0.0398, "samples": 402560, "wall_s": 25086.0, "cuda_gb": 16.86}
{"step": 12600, "loss": 0.001422, "loss_mean50": 0.001669, "lr": 4.093746126319657e-06, "grad_norm": 0.0196, "samples": 403200, "wall_s": 25125.1, "cuda_gb": 16.86}
{"step": 12620, "loss": 0.002485, "loss_mean50": 0.001691, "lr": 4.067937512046945e-06, "grad_norm": 0.0381, "samples": 403840, "wall_s": 25164.3, "cuda_gb": 16.86}
{"step": 12640, "loss": 0.001587, "loss_mean50": 0.001935, "lr": 4.042323914031678e-06, "grad_norm": 0.0378, "samples": 404480, "wall_s": 25203.4, "cuda_gb": 16.86}
{"step": 12660, "loss": 0.000545, "loss_mean50": 0.001889, "lr": 4.016905848183357e-06, "grad_norm": 0.0181, "samples": 405120, "wall_s": 25242.5, "cuda_gb": 16.86}
{"step": 12680, "loss": 0.001363, "loss_mean50": 0.001905, "lr": 3.991683826473071e-06, "grad_norm": 0.0487, "samples": 405760, "wall_s": 25281.7, "cuda_gb": 16.86}
{"step": 12700, "loss": 0.002117, "loss_mean50": 0.00182, "lr": 3.9666583569231835e-06, "grad_norm": 0.0472, "samples": 406400, "wall_s": 25320.8, "cuda_gb": 16.86}
{"step": 12720, "loss": 0.003973, "loss_mean50": 0.001918, "lr": 3.9418299435971e-06, "grad_norm": 0.0399, "samples": 407040, "wall_s": 25359.8, "cuda_gb": 16.86}
{"step": 12740, "loss": 0.001955, "loss_mean50": 0.001922, "lr": 3.9171990865891265e-06, "grad_norm": 0.0375, "samples": 407680, "wall_s": 25398.9, "cuda_gb": 16.86}
{"step": 12760, "loss": 0.001491, "loss_mean50": 0.001832, "lr": 3.892766282014372e-06, "grad_norm": 0.0182, "samples": 408320, "wall_s": 25437.9, "cuda_gb": 16.86}
{"step": 12780, "loss": 0.001014, "loss_mean50": 0.001828, "lr": 3.868532021998779e-06, "grad_norm": 0.019, "samples": 408960, "wall_s": 25477.0, "cuda_gb": 16.86}
{"step": 12800, "loss": 0.000961, "loss_mean50": 0.001775, "lr": 3.844496794669201e-06, "grad_norm": 0.0201, "samples": 409600, "wall_s": 25516.1, "cuda_gb": 16.86}
{"step": 12820, "loss": 0.001121, "loss_mean50": 0.00168, "lr": 3.820661084143568e-06, "grad_norm": 0.022, "samples": 410240, "wall_s": 25555.2, "cuda_gb": 16.86}
{"step": 12840, "loss": 0.001855, "loss_mean50": 0.001573, "lr": 3.797025370521147e-06, "grad_norm": 0.03, "samples": 410880, "wall_s": 25594.1, "cuda_gb": 16.86}
{"step": 12860, "loss": 0.001562, "loss_mean50": 0.001645, "lr": 3.7735901298728575e-06, "grad_norm": 0.0264, "samples": 411520, "wall_s": 25633.1, "cuda_gb": 16.86}
{"step": 12880, "loss": 0.001269, "loss_mean50": 0.001637, "lr": 3.7503558342316883e-06, "grad_norm": 0.0273, "samples": 412160, "wall_s": 25672.3, "cuda_gb": 16.86}
{"step": 12900, "loss": 0.002477, "loss_mean50": 0.001632, "lr": 3.7273229515831993e-06, "grad_norm": 0.0414, "samples": 412800, "wall_s": 25711.4, "cuda_gb": 16.86}
{"step": 12920, "loss": 0.001418, "loss_mean50": 0.001726, "lr": 3.704491945856082e-06, "grad_norm": 0.0242, "samples": 413440, "wall_s": 25750.6, "cuda_gb": 16.86}
{"step": 12940, "loss": 0.001925, "loss_mean50": 0.001658, "lr": 3.6818632769128106e-06, "grad_norm": 0.0465, "samples": 414080, "wall_s": 25789.7, "cuda_gb": 16.86}
{"step": 12960, "loss": 0.000998, "loss_mean50": 0.001538, "lr": 3.6594374005404026e-06, "grad_norm": 0.0384, "samples": 414720, "wall_s": 25828.7, "cuda_gb": 16.86}
{"step": 12980, "loss": 0.001157, "loss_mean50": 0.001488, "lr": 3.637214768441219e-06, "grad_norm": 0.0188, "samples": 415360, "wall_s": 25867.8, "cuda_gb": 16.86}
{"step": 13000, "loss": 0.001192, "loss_mean50": 0.001555, "lr": 3.6151958282238664e-06, "grad_norm": 0.0333, "samples": 416000, "wall_s": 25906.8, "cuda_gb": 16.86}
{"step": 13000, "val_flow_mse": 0.001652}
{"step": 13020, "loss": 0.001823, "loss_mean50": 0.001678, "lr": 3.593381023394196e-06, "grad_norm": 0.026, "samples": 416640, "wall_s": 25974.7, "cuda_gb": 16.86}
{"step": 13040, "loss": 0.001333, "loss_mean50": 0.001674, "lr": 3.571770793346353e-06, "grad_norm": 0.027, "samples": 417280, "wall_s": 26013.8, "cuda_gb": 16.86}
{"step": 13060, "loss": 0.00171, "loss_mean50": 0.00166, "lr": 3.5503655733539372e-06, "grad_norm": 0.0327, "samples": 417920, "wall_s": 26052.9, "cuda_gb": 16.86}
{"step": 13080, "loss": 0.001894, "loss_mean50": 0.001614, "lr": 3.5291657945612303e-06, "grad_norm": 0.0472, "samples": 418560, "wall_s": 26092.0, "cuda_gb": 16.86}
{"step": 13100, "loss": 0.000864, "loss_mean50": 0.001515, "lr": 3.5081718839745164e-06, "grad_norm": 0.0243, "samples": 419200, "wall_s": 26131.0, "cuda_gb": 16.86}
{"step": 13120, "loss": 0.001785, "loss_mean50": 0.001566, "lr": 3.4873842644534798e-06, "grad_norm": 0.0363, "samples": 419840, "wall_s": 26170.0, "cuda_gb": 16.86}
{"step": 13140, "loss": 0.000731, "loss_mean50": 0.001593, "lr": 3.466803354702685e-06, "grad_norm": 0.0173, "samples": 420480, "wall_s": 26209.1, "cuda_gb": 16.86}
{"step": 13160, "loss": 0.00136, "loss_mean50": 0.001645, "lr": 3.4464295692631432e-06, "grad_norm": 0.0228, "samples": 421120, "wall_s": 26248.2, "cuda_gb": 16.86}
{"step": 13180, "loss": 0.001574, "loss_mean50": 0.001698, "lr": 3.4262633185039696e-06, "grad_norm": 0.0256, "samples": 421760, "wall_s": 26287.3, "cuda_gb": 16.86}
{"step": 13200, "loss": 0.001033, "loss_mean50": 0.001578, "lr": 3.406305008614106e-06, "grad_norm": 0.0302, "samples": 422400, "wall_s": 26326.3, "cuda_gb": 16.86}
{"step": 13220, "loss": 0.001578, "loss_mean50": 0.001495, "lr": 3.3865550415941497e-06, "grad_norm": 0.0399, "samples": 423040, "wall_s": 26365.5, "cuda_gb": 16.86}
{"step": 13240, "loss": 0.001372, "loss_mean50": 0.001522, "lr": 3.367013815248255e-06, "grad_norm": 0.0344, "samples": 423680, "wall_s": 26404.6, "cuda_gb": 16.86}
{"step": 13260, "loss": 0.001036, "loss_mean50": 0.001725, "lr": 3.347681723176115e-06, "grad_norm": 0.0514, "samples": 424320, "wall_s": 26443.7, "cuda_gb": 16.86}
{"step": 13280, "loss": 0.000758, "loss_mean50": 0.001743, "lr": 3.3285591547650326e-06, "grad_norm": 0.0447, "samples": 424960, "wall_s": 26482.8, "cuda_gb": 16.86}
{"step": 13300, "loss": 0.00094, "loss_mean50": 0.001743, "lr": 3.3096464951820954e-06, "grad_norm": 0.0161, "samples": 425600, "wall_s": 26521.8, "cuda_gb": 16.86}
{"step": 13320, "loss": 0.001874, "loss_mean50": 0.001479, "lr": 3.290944125366388e-06, "grad_norm": 0.0311, "samples": 426240, "wall_s": 26560.8, "cuda_gb": 16.86}
{"step": 13340, "loss": 0.003514, "loss_mean50": 0.001618, "lr": 3.272452422021341e-06, "grad_norm": 0.0617, "samples": 426880, "wall_s": 26600.0, "cuda_gb": 16.86}
{"step": 13360, "loss": 0.002827, "loss_mean50": 0.001786, "lr": 3.254171757607145e-06, "grad_norm": 0.032, "samples": 427520, "wall_s": 26639.1, "cuda_gb": 16.86}
{"step": 13380, "loss": 0.002033, "loss_mean50": 0.001868, "lr": 3.236102500333231e-06, "grad_norm": 0.0205, "samples": 428160, "wall_s": 26678.2, "cuda_gb": 16.86}
{"step": 13400, "loss": 0.001273, "loss_mean50": 0.001638, "lr": 3.2182450141508667e-06, "grad_norm": 0.0181, "samples": 428800, "wall_s": 26717.2, "cuda_gb": 16.86}
{"step": 13420, "loss": 0.003504, "loss_mean50": 0.001496, "lr": 3.2005996587458287e-06, "grad_norm": 0.0725, "samples": 429440, "wall_s": 26756.3, "cuda_gb": 16.86}
{"step": 13440, "loss": 0.001712, "loss_mean50": 0.001555, "lr": 3.183166789531146e-06, "grad_norm": 0.0183, "samples": 430080, "wall_s": 26795.4, "cuda_gb": 16.86}
{"step": 13460, "loss": 0.000868, "loss_mean50": 0.001589, "lr": 3.1659467576399488e-06, "grad_norm": 0.0403, "samples": 430720, "wall_s": 26834.5, "cuda_gb": 16.86}
{"step": 13480, "loss": 0.001443, "loss_mean50": 0.001553, "lr": 3.1489399099183936e-06, "grad_norm": 0.0342, "samples": 431360, "wall_s": 26873.6, "cuda_gb": 16.86}
{"step": 13500, "loss": 0.000784, "loss_mean50": 0.001517, "lr": 3.132146588918679e-06, "grad_norm": 0.0209, "samples": 432000, "wall_s": 26912.6, "cuda_gb": 16.86}
{"step": 13520, "loss": 0.002083, "loss_mean50": 0.00148, "lr": 3.1155671328921473e-06, "grad_norm": 0.0215, "samples": 432640, "wall_s": 26951.6, "cuda_gb": 16.86}
{"step": 13540, "loss": 0.003808, "loss_mean50": 0.001606, "lr": 3.0992018757824658e-06, "grad_norm": 0.0501, "samples": 433280, "wall_s": 26990.6, "cuda_gb": 16.86}
{"step": 13560, "loss": 0.00128, "loss_mean50": 0.001584, "lr": 3.0830511472189056e-06, "grad_norm": 0.0367, "samples": 433920, "wall_s": 27029.6, "cuda_gb": 16.86}
{"step": 13580, "loss": 0.001019, "loss_mean50": 0.001519, "lr": 3.0671152725096976e-06, "grad_norm": 0.0194, "samples": 434560, "wall_s": 27068.6, "cuda_gb": 16.86}
{"step": 13600, "loss": 0.001252, "loss_mean50": 0.001553, "lr": 3.0513945726354877e-06, "grad_norm": 0.0187, "samples": 435200, "wall_s": 27107.7, "cuda_gb": 16.86}
{"step": 13620, "loss": 0.001843, "loss_mean50": 0.001626, "lr": 3.0358893642428604e-06, "grad_norm": 0.0221, "samples": 435840, "wall_s": 27146.7, "cuda_gb": 16.86}
{"step": 13640, "loss": 0.000985, "loss_mean50": 0.001599, "lr": 3.020599959637975e-06, "grad_norm": 0.0177, "samples": 436480, "wall_s": 27185.7, "cuda_gb": 16.86}
{"step": 13660, "loss": 0.001502, "loss_mean50": 0.001657, "lr": 3.005526666780261e-06, "grad_norm": 0.0463, "samples": 437120, "wall_s": 27224.6, "cuda_gb": 16.86}
{"step": 13680, "loss": 0.001898, "loss_mean50": 0.001682, "lr": 2.990669789276225e-06, "grad_norm": 0.0226, "samples": 437760, "wall_s": 27263.7, "cuda_gb": 16.86}
{"step": 13700, "loss": 0.001657, "loss_mean50": 0.001656, "lr": 2.9760296263733332e-06, "grad_norm": 0.0402, "samples": 438400, "wall_s": 27302.7, "cuda_gb": 16.86}
{"step": 13720, "loss": 0.001067, "loss_mean50": 0.0015, "lr": 2.961606472953977e-06, "grad_norm": 0.0241, "samples": 439040, "wall_s": 27341.8, "cuda_gb": 16.86}
{"step": 13740, "loss": 0.001404, "loss_mean50": 0.001507, "lr": 2.947400619529546e-06, "grad_norm": 0.0467, "samples": 439680, "wall_s": 27380.9, "cuda_gb": 16.86}
{"step": 13760, "loss": 0.000842, "loss_mean50": 0.001474, "lr": 2.933412352234567e-06, "grad_norm": 0.0218, "samples": 440320, "wall_s": 27420.0, "cuda_gb": 16.86}
{"step": 13780, "loss": 0.000997, "loss_mean50": 0.001498, "lr": 2.919641952820941e-06, "grad_norm": 0.0159, "samples": 440960, "wall_s": 27459.1, "cuda_gb": 16.86}
{"step": 13800, "loss": 0.001131, "loss_mean50": 0.001454, "lr": 2.906089698652277e-06, "grad_norm": 0.0282, "samples": 441600, "wall_s": 27498.1, "cuda_gb": 16.86}
{"step": 13820, "loss": 0.001426, "loss_mean50": 0.001537, "lr": 2.892755862698292e-06, "grad_norm": 0.0178, "samples": 442240, "wall_s": 27537.2, "cuda_gb": 16.86}
{"step": 13840, "loss": 0.001267, "loss_mean50": 0.001475, "lr": 2.8796407135293233e-06, "grad_norm": 0.0305, "samples": 442880, "wall_s": 27576.2, "cuda_gb": 16.86}
{"step": 13860, "loss": 0.000868, "loss_mean50": 0.001465, "lr": 2.866744515310917e-06, "grad_norm": 0.0091, "samples": 443520, "wall_s": 27615.1, "cuda_gb": 16.86}
{"step": 13880, "loss": 0.004188, "loss_mean50": 0.001547, "lr": 2.854067527798503e-06, "grad_norm": 0.0889, "samples": 444160, "wall_s": 27654.1, "cuda_gb": 16.86}
{"step": 13900, "loss": 0.001592, "loss_mean50": 0.001494, "lr": 2.8416100063321654e-06, "grad_norm": 0.0424, "samples": 444800, "wall_s": 27693.1, "cuda_gb": 16.86}
{"step": 13920, "loss": 0.001423, "loss_mean50": 0.001503, "lr": 2.829372201831502e-06, "grad_norm": 0.0263, "samples": 445440, "wall_s": 27732.1, "cuda_gb": 16.86}
{"step": 13940, "loss": 0.001757, "loss_mean50": 0.001545, "lr": 2.817354360790566e-06, "grad_norm": 0.0286, "samples": 446080, "wall_s": 27771.0, "cuda_gb": 16.86}
{"step": 13960, "loss": 0.00127, "loss_mean50": 0.001575, "lr": 2.8055567252729046e-06, "grad_norm": 0.0329, "samples": 446720, "wall_s": 27809.9, "cuda_gb": 16.86}
{"step": 13980, "loss": 0.002197, "loss_mean50": 0.001629, "lr": 2.7939795329066803e-06, "grad_norm": 0.1178, "samples": 447360, "wall_s": 27848.9, "cuda_gb": 16.86}
{"step": 14000, "loss": 0.001105, "loss_mean50": 0.001621, "lr": 2.7826230168798876e-06, "grad_norm": 0.023, "samples": 448000, "wall_s": 27887.8, "cuda_gb": 16.86}
{"step": 14000, "val_flow_mse": 0.001525}
{"step": 14020, "loss": 0.000914, "loss_mean50": 0.001595, "lr": 2.7714874059356517e-06, "grad_norm": 0.0165, "samples": 448640, "wall_s": 27955.6, "cuda_gb": 16.86}
{"step": 14040, "loss": 0.001434, "loss_mean50": 0.001418, "lr": 2.76057292436763e-06, "grad_norm": 0.0239, "samples": 449280, "wall_s": 27994.6, "cuda_gb": 16.86}
{"step": 14060, "loss": 0.001984, "loss_mean50": 0.001451, "lr": 2.7498797920154817e-06, "grad_norm": 0.0275, "samples": 449920, "wall_s": 28033.6, "cuda_gb": 16.86}
{"step": 14080, "loss": 0.000677, "loss_mean50": 0.001498, "lr": 2.7394082242604514e-06, "grad_norm": 0.0108, "samples": 450560, "wall_s": 28072.5, "cuda_gb": 16.86}
{"step": 14100, "loss": 0.00112, "loss_mean50": 0.001483, "lr": 2.7291584320210242e-06, "grad_norm": 0.0301, "samples": 451200, "wall_s": 28111.3, "cuda_gb": 16.86}
{"step": 14120, "loss": 0.000812, "loss_mean50": 0.001384, "lr": 2.7191306217486763e-06, "grad_norm": 0.0178, "samples": 451840, "wall_s": 28150.2, "cuda_gb": 16.86}
{"step": 14140, "loss": 0.001144, "loss_mean50": 0.001402, "lr": 2.709324995423726e-06, "grad_norm": 0.0195, "samples": 452480, "wall_s": 28189.1, "cuda_gb": 16.86}
{"step": 14160, "loss": 0.00179, "loss_mean50": 0.001633, "lr": 2.6997417505512548e-06, "grad_norm": 0.0504, "samples": 453120, "wall_s": 28228.0, "cuda_gb": 16.86}
{"step": 14180, "loss": 0.002228, "loss_mean50": 0.00174, "lr": 2.690381080157134e-06, "grad_norm": 0.0309, "samples": 453760, "wall_s": 28267.0, "cuda_gb": 16.86}
{"step": 14200, "loss": 0.000876, "loss_mean50": 0.001559, "lr": 2.681243172784135e-06, "grad_norm": 0.0107, "samples": 454400, "wall_s": 28305.9, "cuda_gb": 16.86}
{"step": 14220, "loss": 0.00091, "loss_mean50": 0.001525, "lr": 2.6723282124881367e-06, "grad_norm": 0.0156, "samples": 455040, "wall_s": 28344.8, "cuda_gb": 16.86}
{"step": 14240, "loss": 0.001792, "loss_mean50": 0.001463, "lr": 2.663636378834409e-06, "grad_norm": 0.028, "samples": 455680, "wall_s": 28383.9, "cuda_gb": 16.86}
{"step": 14260, "loss": 0.000991, "loss_mean50": 0.001555, "lr": 2.6551678468940077e-06, "grad_norm": 0.0207, "samples": 456320, "wall_s": 28422.9, "cuda_gb": 16.86}
{"step": 14280, "loss": 0.001741, "loss_mean50": 0.00162, "lr": 2.6469227872402357e-06, "grad_norm": 0.101, "samples": 456960, "wall_s": 28462.0, "cuda_gb": 16.86}
{"step": 14300, "loss": 0.00088, "loss_mean50": 0.001435, "lr": 2.638901365945217e-06, "grad_norm": 0.019, "samples": 457600, "wall_s": 28501.0, "cuda_gb": 16.86}
{"step": 14320, "loss": 0.001242, "loss_mean50": 0.001491, "lr": 2.6311037445765488e-06, "grad_norm": 0.0214, "samples": 458240, "wall_s": 28540.0, "cuda_gb": 16.86}
{"step": 14340, "loss": 0.00079, "loss_mean50": 0.001499, "lr": 2.623530080194044e-06, "grad_norm": 0.0187, "samples": 458880, "wall_s": 28578.9, "cuda_gb": 16.86}
{"step": 14360, "loss": 0.001109, "loss_mean50": 0.001479, "lr": 2.6161805253465737e-06, "grad_norm": 0.0246, "samples": 459520, "wall_s": 28617.8, "cuda_gb": 16.86}
{"step": 14380, "loss": 0.002129, "loss_mean50": 0.001502, "lr": 2.609055228068987e-06, "grad_norm": 0.0369, "samples": 460160, "wall_s": 28656.8, "cuda_gb": 16.86}
{"step": 14400, "loss": 0.00098, "loss_mean50": 0.001565, "lr": 2.6021543318791354e-06, "grad_norm": 0.019, "samples": 460800, "wall_s": 28695.8, "cuda_gb": 16.86}
{"step": 14420, "loss": 0.00097, "loss_mean50": 0.001643, "lr": 2.595477975774981e-06, "grad_norm": 0.0315, "samples": 461440, "wall_s": 28734.7, "cuda_gb": 16.86}
{"step": 14440, "loss": 0.002477, "loss_mean50": 0.001531, "lr": 2.589026294231797e-06, "grad_norm": 0.0929, "samples": 462080, "wall_s": 28773.7, "cuda_gb": 16.86}
{"step": 14460, "loss": 0.001876, "loss_mean50": 0.001568, "lr": 2.582799417199454e-06, "grad_norm": 0.0505, "samples": 462720, "wall_s": 28812.7, "cuda_gb": 16.86}
{"step": 14480, "loss": 0.001273, "loss_mean50": 0.00162, "lr": 2.5767974700998093e-06, "grad_norm": 0.0242, "samples": 463360, "wall_s": 28851.7, "cuda_gb": 16.86}
{"step": 14500, "loss": 0.001798, "loss_mean50": 0.001589, "lr": 2.5710205738241775e-06, "grad_norm": 0.0274, "samples": 464000, "wall_s": 28890.7, "cuda_gb": 16.86}
{"step": 14520, "loss": 0.00182, "loss_mean50": 0.001568, "lr": 2.565468844730895e-06, "grad_norm": 0.0365, "samples": 464640, "wall_s": 28929.6, "cuda_gb": 16.86}
{"step": 14540, "loss": 0.003005, "loss_mean50": 0.001617, "lr": 2.5601423946429807e-06, "grad_norm": 0.0788, "samples": 465280, "wall_s": 28968.4, "cuda_gb": 16.86}
{"step": 14560, "loss": 0.00094, "loss_mean50": 0.001516, "lr": 2.555041330845874e-06, "grad_norm": 0.0163, "samples": 465920, "wall_s": 29007.4, "cuda_gb": 16.86}
{"step": 14580, "loss": 0.001575, "loss_mean50": 0.001605, "lr": 2.550165756085284e-06, "grad_norm": 0.0271, "samples": 466560, "wall_s": 29046.3, "cuda_gb": 16.86}
{"step": 14600, "loss": 0.001329, "loss_mean50": 0.001754, "lr": 2.5455157685651184e-06, "grad_norm": 0.0224, "samples": 467200, "wall_s": 29085.2, "cuda_gb": 16.86}
{"step": 14620, "loss": 0.00091, "loss_mean50": 0.001772, "lr": 2.5410914619454962e-06, "grad_norm": 0.0225, "samples": 467840, "wall_s": 29124.2, "cuda_gb": 16.86}
{"step": 14640, "loss": 0.001202, "loss_mean50": 0.001646, "lr": 2.5368929253408747e-06, "grad_norm": 0.024, "samples": 468480, "wall_s": 29163.1, "cuda_gb": 16.86}
{"step": 14660, "loss": 0.001441, "loss_mean50": 0.001459, "lr": 2.5329202433182453e-06, "grad_norm": 0.0175, "samples": 469120, "wall_s": 29202.1, "cuda_gb": 16.86}
{"step": 14680, "loss": 0.001365, "loss_mean50": 0.001542, "lr": 2.529173495895432e-06, "grad_norm": 0.0366, "samples": 469760, "wall_s": 29241.0, "cuda_gb": 16.86}
{"step": 14700, "loss": 0.001593, "loss_mean50": 0.001612, "lr": 2.5256527585394805e-06, "grad_norm": 0.0251, "samples": 470400, "wall_s": 29279.9, "cuda_gb": 16.86}
{"step": 14720, "loss": 0.000689, "loss_mean50": 0.001667, "lr": 2.5223581021651405e-06, "grad_norm": 0.0144, "samples": 471040, "wall_s": 29318.8, "cuda_gb": 16.86}
{"step": 14740, "loss": 0.001544, "loss_mean50": 0.00153, "lr": 2.5192895931334335e-06, "grad_norm": 0.0201, "samples": 471680, "wall_s": 29357.7, "cuda_gb": 16.86}
{"step": 14760, "loss": 0.001805, "loss_mean50": 0.001477, "lr": 2.516447293250316e-06, "grad_norm": 0.0264, "samples": 472320, "wall_s": 29396.6, "cuda_gb": 16.86}
{"step": 14780, "loss": 0.002091, "loss_mean50": 0.001501, "lr": 2.5138312597654382e-06, "grad_norm": 0.0385, "samples": 472960, "wall_s": 29435.6, "cuda_gb": 16.86}
{"step": 14800, "loss": 0.000893, "loss_mean50": 0.001624, "lr": 2.5114415453709886e-06, "grad_norm": 0.0152, "samples": 473600, "wall_s": 29474.5, "cuda_gb": 16.86}
{"step": 14820, "loss": 0.001256, "loss_mean50": 0.001637, "lr": 2.509278198200633e-06, "grad_norm": 0.0253, "samples": 474240, "wall_s": 29513.5, "cuda_gb": 16.86}
{"step": 14840, "loss": 0.000892, "loss_mean50": 0.001505, "lr": 2.5073412618285445e-06, "grad_norm": 0.014, "samples": 474880, "wall_s": 29552.4, "cuda_gb": 16.86}
{"step": 14860, "loss": 0.001317, "loss_mean50": 0.00143, "lr": 2.505630775268527e-06, "grad_norm": 0.0405, "samples": 475520, "wall_s": 29591.3, "cuda_gb": 16.86}
{"step": 14880, "loss": 0.001315, "loss_mean50": 0.001468, "lr": 2.5041467729732266e-06, "grad_norm": 0.021, "samples": 476160, "wall_s": 29630.2, "cuda_gb": 16.86}
{"step": 14900, "loss": 0.000922, "loss_mean50": 0.00144, "lr": 2.502889284833443e-06, "grad_norm": 0.0156, "samples": 476800, "wall_s": 29669.1, "cuda_gb": 16.86}
{"step": 14920, "loss": 0.001913, "loss_mean50": 0.001442, "lr": 2.5018583361775205e-06, "grad_norm": 0.0319, "samples": 477440, "wall_s": 29708.1, "cuda_gb": 16.86}
{"step": 14940, "loss": 0.000967, "loss_mean50": 0.001487, "lr": 2.501053947770846e-06, "grad_norm": 0.0243, "samples": 478080, "wall_s": 29747.0, "cuda_gb": 16.86}
{"step": 14960, "loss": 0.001677, "loss_mean50": 0.001583, "lr": 2.5004761358154212e-06, "grad_norm": 0.0329, "samples": 478720, "wall_s": 29785.9, "cuda_gb": 16.86}
{"step": 14980, "loss": 0.002625, "loss_mean50": 0.001594, "lr": 2.5001249119495454e-06, "grad_norm": 0.047, "samples": 479360, "wall_s": 29824.9, "cuda_gb": 16.86}
{"step": 15000, "loss": 0.001234, "loss_mean50": 0.001561, "lr": 2.5000002832475742e-06, "grad_norm": 0.0229, "samples": 480000, "wall_s": 29863.7, "cuda_gb": 16.86}
{"step": 15000, "val_flow_mse": 0.001588, "val_action_mae": 0.011025, "best": true}