Cccccz's picture
Add files using upload-large-folder tool
2437753 verified
Raw
History Blame Contribute Delete
63.4 kB
{"avg_micro_time_s": 7.782912929877057, "epoch": 0, "flow_mse": 1.728674739599228, "global_step": 1, "grad_norm_max": 1.224261999130249, "hidden_mse": 0.34973957017064095, "loss": 1.7636487036943436, "lr_blocks": 0.0, "lr_fusion": 2.0000000000000003e-06, "peak_memory_gib": 59.944087982177734}
{"avg_micro_time_s": 1.852714978063078, "epoch": 0, "flow_mse": 1.7325291699833338, "global_step": 10, "grad_norm_max": 1.5914783477783203, "hidden_mse": 0.36041637013355887, "loss": 1.7685708122120962, "lr_blocks": 0.0, "lr_fusion": 1.1000000000000001e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.846399891503097, "epoch": 0, "flow_mse": 1.699481675028801, "global_step": 20, "grad_norm_max": 1.9605422019958496, "hidden_mse": 0.38648081049323085, "loss": 1.7381297543644907, "lr_blocks": 0.0, "lr_fusion": 2.1e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 2.076498315717618, "epoch": 0, "flow_mse": 1.5910302087664605, "global_step": 30, "grad_norm_max": 1.6393431425094604, "hidden_mse": 0.6486873880028725, "loss": 1.6558989509940147, "lr_blocks": 0.0, "lr_fusion": 3.1e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8261107150639875, "epoch": 0, "flow_mse": 1.4074327528476716, "global_step": 40, "grad_norm_max": 1.6300629377365112, "hidden_mse": 1.5101009473204614, "loss": 1.5584428489208222, "lr_blocks": 0.0, "lr_fusion": 4.1e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.7557343540451258, "epoch": 0, "flow_mse": 1.3375679016113282, "global_step": 50, "grad_norm_max": 1.179707407951355, "hidden_mse": 1.7423540219664575, "loss": 1.5118033051490785, "lr_blocks": 0.0, "lr_fusion": 5.1000000000000006e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.9954431485108217, "epoch": 0, "flow_mse": 1.288495972752571, "global_step": 60, "grad_norm_max": 0.9528158903121948, "hidden_mse": 1.6803042054176331, "loss": 1.4565263986587524, "lr_blocks": 0.0, "lr_fusion": 6.1e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 2.073848069201631, "epoch": 1, "flow_mse": 1.2329863071441651, "global_step": 70, "grad_norm_max": 2.1396095752716064, "hidden_mse": 1.468385949730873, "loss": 1.3798249036073686, "lr_blocks": 0.0, "lr_fusion": 7.1e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.9025867714255584, "epoch": 1, "flow_mse": 1.1653944417834283, "global_step": 80, "grad_norm_max": 3.283475875854492, "hidden_mse": 1.3008124545216562, "loss": 1.2954756870865822, "lr_blocks": 0.0, "lr_fusion": 8.1e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8846276455253248, "epoch": 1, "flow_mse": 1.0928176268935204, "global_step": 90, "grad_norm_max": 2.976677179336548, "hidden_mse": 1.1887190774083138, "loss": 1.211689528822899, "lr_blocks": 0.0, "lr_fusion": 9.1e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 2.0524098566529574, "epoch": 1, "flow_mse": 1.0091206416487695, "global_step": 100, "grad_norm_max": 2.0720155239105225, "hidden_mse": 1.1053067579865457, "loss": 1.119651308655739, "lr_blocks": 1.0000000000000001e-07, "lr_fusion": 0.0001, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8157492470301806, "epoch": 1, "flow_mse": 0.9281245060265064, "global_step": 110, "grad_norm_max": 3.5477075576782227, "hidden_mse": 1.0355910658836365, "loss": 1.031683611869812, "lr_blocks": 1.1e-06, "lr_fusion": 9.999316524962345e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.7559494119937882, "epoch": 1, "flow_mse": 0.8380454987287522, "global_step": 120, "grad_norm_max": 3.569852828979492, "hidden_mse": 1.0150623381137849, "loss": 0.9395517356693746, "lr_blocks": 2.1000000000000002e-06, "lr_fusion": 9.997266286704631e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.9936403229308781, "epoch": 1, "flow_mse": 0.7462111793458462, "global_step": 130, "grad_norm_max": 2.9960243701934814, "hidden_mse": 0.9869170427322388, "loss": 0.8449028812348843, "lr_blocks": 3.1000000000000004e-06, "lr_fusion": 9.993849845741524e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8761029311470339, "epoch": 2, "flow_mse": 0.6552131280303002, "global_step": 140, "grad_norm_max": 3.033362627029419, "hidden_mse": 0.9761701941490174, "loss": 0.7528301522135735, "lr_blocks": 4.1e-06, "lr_fusion": 9.989068136093873e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.6636652560366203, "epoch": 2, "flow_mse": 0.5641707576811313, "global_step": 150, "grad_norm_max": 3.3895790576934814, "hidden_mse": 0.9558152981102467, "loss": 0.6597522847354412, "lr_blocks": 5.1e-06, "lr_fusion": 9.98292246503335e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8629546527619825, "epoch": 2, "flow_mse": 0.49576797895133495, "global_step": 160, "grad_norm_max": 4.9101152420043945, "hidden_mse": 0.9376181185245515, "loss": 0.5895297914743424, "lr_blocks": 6.1e-06, "lr_fusion": 9.975414512725057e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.865570171797299, "epoch": 2, "flow_mse": 0.44515644647181035, "global_step": 170, "grad_norm_max": 3.2292189598083496, "hidden_mse": 0.9383152291178704, "loss": 0.5389879673719407, "lr_blocks": 7.100000000000001e-06, "lr_fusion": 9.966546331768191e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 2.0450479036619074, "epoch": 2, "flow_mse": 0.38247467912733557, "global_step": 180, "grad_norm_max": 2.7824766635894775, "hidden_mse": 0.9554134920239449, "loss": 0.47801603078842164, "lr_blocks": 8.1e-06, "lr_fusion": 9.956320346634876e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 2.016261681757169, "epoch": 2, "flow_mse": 0.3393749196082354, "global_step": 190, "grad_norm_max": 2.3986711502075195, "hidden_mse": 0.9583184562623501, "loss": 0.4352067667990923, "lr_blocks": 9.100000000000001e-06, "lr_fusion": 9.944739353007344e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.955125979850709, "epoch": 3, "flow_mse": 0.2995503291487694, "global_step": 200, "grad_norm_max": 2.834411144256592, "hidden_mse": 0.9311102323234082, "loss": 0.39266135394573215, "lr_blocks": 1e-05, "lr_fusion": 9.931806517013612e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.9450525438660407, "epoch": 3, "flow_mse": 0.26215458251535895, "global_step": 210, "grad_norm_max": 2.5127832889556885, "hidden_mse": 0.9175012007355691, "loss": 0.35390470623970033, "lr_blocks": 9.999238475781957e-06, "lr_fusion": 9.917525374361912e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.794988799588464, "epoch": 3, "flow_mse": 0.24451670069247486, "global_step": 220, "grad_norm_max": 3.86234712600708, "hidden_mse": 0.8481401547789574, "loss": 0.32933071814477444, "lr_blocks": 9.99695413509548e-06, "lr_fusion": 9.901899829374047e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.821383213158697, "epoch": 3, "flow_mse": 0.21927626319229604, "global_step": 230, "grad_norm_max": 2.894408702850342, "hidden_mse": 0.7950924828648568, "loss": 0.2987855151295662, "lr_blocks": 9.993147673772869e-06, "lr_fusion": 9.884934153917997e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.7578791950494634, "epoch": 3, "flow_mse": 0.21038782224059105, "global_step": 240, "grad_norm_max": 4.587767124176025, "hidden_mse": 0.7239097714424134, "loss": 0.2827788006514311, "lr_blocks": 9.987820251299121e-06, "lr_fusion": 9.86663298624003e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.9880367784600823, "epoch": 3, "flow_mse": 0.18143612183630467, "global_step": 250, "grad_norm_max": 2.522402048110962, "hidden_mse": 0.6568384729325771, "loss": 0.24711996987462045, "lr_blocks": 9.980973490458728e-06, "lr_fusion": 9.847001329696653e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 2.1574960700992962, "epoch": 3, "flow_mse": 0.17069688346236944, "global_step": 260, "grad_norm_max": 2.6453685760498047, "hidden_mse": 0.5946614742279053, "loss": 0.23016303274780514, "lr_blocks": 9.972609476841368e-06, "lr_fusion": 9.826044551386744e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.7216765362391016, "epoch": 4, "flow_mse": 0.16511701364070178, "global_step": 270, "grad_norm_max": 4.101485252380371, "hidden_mse": 0.5324683826416731, "loss": 0.21836385373026135, "lr_blocks": 9.962730758206612e-06, "lr_fusion": 9.803768380684242e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 2.036255760942004, "epoch": 4, "flow_mse": 0.15243881475180387, "global_step": 280, "grad_norm_max": 3.777620792388916, "hidden_mse": 0.49607212990522387, "loss": 0.20204602926969528, "lr_blocks": 9.951340343707852e-06, "lr_fusion": 9.780178907671789e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 2.026448777728365, "epoch": 4, "flow_mse": 0.13280713185667992, "global_step": 290, "grad_norm_max": 2.864901542663574, "hidden_mse": 0.45720995105803014, "loss": 0.1785281287506223, "lr_blocks": 9.938441702975689e-06, "lr_fusion": 9.755282581475769e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.7816235546153623, "epoch": 4, "flow_mse": 0.1316212204284966, "global_step": 300, "grad_norm_max": 3.557399272918701, "hidden_mse": 0.4236030507832766, "loss": 0.17398152723908425, "lr_blocks": 9.924038765061042e-06, "lr_fusion": 9.729086208503174e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.7509075272682821, "epoch": 4, "flow_mse": 0.12680668467655778, "global_step": 310, "grad_norm_max": 2.9080326557159424, "hidden_mse": 0.3970677562057972, "loss": 0.16651346255093813, "lr_blocks": 9.908135917238321e-06, "lr_fusion": 9.701596950580806e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.94211900010705, "epoch": 4, "flow_mse": 0.1265168657526374, "global_step": 320, "grad_norm_max": 3.0853631496429443, "hidden_mse": 0.3776782535016537, "loss": 0.16428469233214857, "lr_blocks": 9.890738003669029e-06, "lr_fusion": 9.672822322997305e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.9201412342023105, "epoch": 4, "flow_mse": 0.11589663326740265, "global_step": 330, "grad_norm_max": 3.0462610721588135, "hidden_mse": 0.3677100505679846, "loss": 0.1526676405221224, "lr_blocks": 9.871850323926178e-06, "lr_fusion": 9.642770192448536e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8802110167060164, "epoch": 5, "flow_mse": 0.10986252082511783, "global_step": 340, "grad_norm_max": 2.5844221115112305, "hidden_mse": 0.3474765542894602, "loss": 0.14461017716675997, "lr_blocks": 9.851478631379982e-06, "lr_fusion": 9.611448774886924e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.970779871637933, "epoch": 5, "flow_mse": 0.10572945503517986, "global_step": 350, "grad_norm_max": 2.6753783226013184, "hidden_mse": 0.33289663605391984, "loss": 0.13901912048459053, "lr_blocks": 9.829629131445342e-06, "lr_fusion": 9.578866633275288e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8511785630384112, "epoch": 5, "flow_mse": 0.09914825893938542, "global_step": 360, "grad_norm_max": 2.757209539413452, "hidden_mse": 0.3192139830440283, "loss": 0.13106965916231275, "lr_blocks": 9.806308479691595e-06, "lr_fusion": 9.545032675245813e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.9199401149497135, "epoch": 5, "flow_mse": 0.0996528490446508, "global_step": 370, "grad_norm_max": 2.9345085620880127, "hidden_mse": 0.3086499966681004, "loss": 0.1305178500711918, "lr_blocks": 9.781523779815178e-06, "lr_fusion": 9.509956150664796e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.7542048408882693, "epoch": 5, "flow_mse": 0.11152176447212697, "global_step": 380, "grad_norm_max": 3.106746196746826, "hidden_mse": 0.30627303943037987, "loss": 0.1421490689739585, "lr_blocks": 9.755282581475769e-06, "lr_fusion": 9.473646649103818e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.9700474178520382, "epoch": 5, "flow_mse": 0.10044981110841036, "global_step": 390, "grad_norm_max": 2.560649871826172, "hidden_mse": 0.2952960841357708, "loss": 0.1299794196151197, "lr_blocks": 9.727592877996585e-06, "lr_fusion": 9.43611409721806e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.6663271824698314, "epoch": 6, "flow_mse": 0.09705260461196304, "global_step": 400, "grad_norm_max": 2.532449722290039, "hidden_mse": 0.29162609074264767, "loss": 0.1262152137234807, "lr_blocks": 9.698463103929542e-06, "lr_fusion": 9.397368756032445e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.904068420661497, "epoch": 6, "flow_mse": 0.09287562295794488, "global_step": 410, "grad_norm_max": 2.16715145111084, "hidden_mse": 0.28314792811870576, "loss": 0.121190416906029, "lr_blocks": 9.667902132486009e-06, "lr_fusion": 9.357421218136386e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.744727985771897, "epoch": 6, "flow_mse": 0.09119896441698075, "global_step": 420, "grad_norm_max": 2.299543619155884, "hidden_mse": 0.275063331797719, "loss": 0.1187052981927991, "lr_blocks": 9.635919272833938e-06, "lr_fusion": 9.316282404787871e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.7956668802158675, "epoch": 6, "flow_mse": 0.08759928597137333, "global_step": 430, "grad_norm_max": 2.317425489425659, "hidden_mse": 0.2642714731395245, "loss": 0.11402643444016576, "lr_blocks": 9.602524267262202e-06, "lr_fusion": 9.273963562927695e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.9157415576424683, "epoch": 6, "flow_mse": 0.08734634332358837, "global_step": 440, "grad_norm_max": 2.760629415512085, "hidden_mse": 0.25739369336515666, "loss": 0.11308571305125953, "lr_blocks": 9.567727288213005e-06, "lr_fusion": 9.230476262104677e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8787613131338732, "epoch": 6, "flow_mse": 0.08993547437712551, "global_step": 450, "grad_norm_max": 3.014104127883911, "hidden_mse": 0.25076083578169345, "loss": 0.11501155886799097, "lr_blocks": 9.531538935183252e-06, "lr_fusion": 9.185832391312644e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8499236255913276, "epoch": 6, "flow_mse": 0.0833338218741119, "global_step": 460, "grad_norm_max": 2.0357143878936768, "hidden_mse": 0.24471132159233094, "loss": 0.10780495461076499, "lr_blocks": 9.493970231495836e-06, "lr_fusion": 9.140044155740101e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8277902282425202, "epoch": 7, "flow_mse": 0.08892108229920269, "global_step": 470, "grad_norm_max": 2.870374917984009, "hidden_mse": 0.24110151771456004, "loss": 0.11303123468533159, "lr_blocks": 9.45503262094184e-06, "lr_fusion": 9.093124073433463e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8054274757610984, "epoch": 7, "flow_mse": 0.08921196823939681, "global_step": 480, "grad_norm_max": 2.4050748348236084, "hidden_mse": 0.24109693542122843, "loss": 0.113321662414819, "lr_blocks": 9.414737964294636e-06, "lr_fusion": 9.045084971874738e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.969536912314652, "epoch": 7, "flow_mse": 0.08481648499146105, "global_step": 490, "grad_norm_max": 4.0341668128967285, "hidden_mse": 0.23553607985377312, "loss": 0.10837009344249965, "lr_blocks": 9.37309853569698e-06, "lr_fusion": 8.995939984474624e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.7191957566945348, "epoch": 7, "flow_mse": 0.08171517765149475, "global_step": 500, "grad_norm_max": 2.7221522331237793, "hidden_mse": 0.22536225505173207, "loss": 0.10425140429288149, "lr_blocks": 9.330127018922195e-06, "lr_fusion": 8.945702546981969e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8753900891068043, "epoch": 7, "flow_mse": 0.08210981907323003, "global_step": 510, "grad_norm_max": 1.9840606451034546, "hidden_mse": 0.22030116170644762, "loss": 0.10413993652909995, "lr_blocks": 9.285836503510562e-06, "lr_fusion": 8.894386393810563e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.9803101174184121, "epoch": 7, "flow_mse": 0.08212232678197325, "global_step": 520, "grad_norm_max": 2.06720232963562, "hidden_mse": 0.21767251621931794, "loss": 0.10388957895338535, "lr_blocks": 9.24024048078213e-06, "lr_fusion": 8.842005554284296e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.7816847848371253, "epoch": 8, "flow_mse": 0.07924596662633121, "global_step": 530, "grad_norm_max": 2.3816592693328857, "hidden_mse": 0.21423672661185267, "loss": 0.10066964011639357, "lr_blocks": 9.193352839727122e-06, "lr_fusion": 8.788574348801675e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.862245862412965, "epoch": 8, "flow_mse": 0.07841471419669688, "global_step": 540, "grad_norm_max": 2.3286426067352295, "hidden_mse": 0.2106100643053651, "loss": 0.09947572173550726, "lr_blocks": 9.145187862775208e-06, "lr_fusion": 8.73410738492077e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 2.0419705960011925, "epoch": 8, "flow_mse": 0.07392775514163077, "global_step": 550, "grad_norm_max": 1.3650901317596436, "hidden_mse": 0.2003183823078871, "loss": 0.09395959414541721, "lr_blocks": 9.09576022144496e-06, "lr_fusion": 8.678619553365659e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.6736296482136823, "epoch": 8, "flow_mse": 0.07821477656252683, "global_step": 560, "grad_norm_max": 2.5323493480682373, "hidden_mse": 0.19945396259427073, "loss": 0.09816017346456647, "lr_blocks": 9.045084971874738e-06, "lr_fusion": 8.622126023955446e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 2.003066141484305, "epoch": 8, "flow_mse": 0.08001610483042897, "global_step": 570, "grad_norm_max": 2.0001752376556396, "hidden_mse": 0.19836203493177892, "loss": 0.09985230900347233, "lr_blocks": 8.993177550236464e-06, "lr_fusion": 8.564642241456986e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.9908761504004362, "epoch": 8, "flow_mse": 0.07824829332530499, "global_step": 580, "grad_norm_max": 2.1289918422698975, "hidden_mse": 0.19441615901887419, "loss": 0.09768990930169821, "lr_blocks": 8.94005376803361e-06, "lr_fusion": 8.506183921362443e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.722032713617955, "epoch": 8, "flow_mse": 0.07608461757190527, "global_step": 590, "grad_norm_max": 2.0547220706939697, "hidden_mse": 0.19045738708227875, "loss": 0.09513035686686636, "lr_blocks": 8.885729807284855e-06, "lr_fusion": 8.44676704559283e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.5744449160105434, "epoch": 9, "flow_mse": 0.07611198453232647, "global_step": 600, "grad_norm_max": 1.646276831626892, "hidden_mse": 0.1878601059317589, "loss": 0.09489799588918686, "lr_blocks": 8.83022221559489e-06, "lr_fusion": 8.386407858128706e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.9617489955402563, "epoch": 9, "flow_mse": 0.07234360408037901, "global_step": 610, "grad_norm_max": 2.1270275115966797, "hidden_mse": 0.18017836939543486, "loss": 0.09036144157871605, "lr_blocks": 8.773547901113862e-06, "lr_fusion": 8.32512286056924e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.899630580819212, "epoch": 9, "flow_mse": 0.07547258702106774, "global_step": 620, "grad_norm_max": 2.535357713699341, "hidden_mse": 0.17847571074962618, "loss": 0.09332015877589583, "lr_blocks": 8.715724127386971e-06, "lr_fusion": 8.262928807620843e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.9009491362783593, "epoch": 9, "flow_mse": 0.07615209170617164, "global_step": 630, "grad_norm_max": 2.885998487472534, "hidden_mse": 0.17413493599742652, "loss": 0.09356558537110687, "lr_blocks": 8.656768508095853e-06, "lr_fusion": 8.199842702516583e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 2.033017601488973, "epoch": 9, "flow_mse": 0.07332893046550452, "global_step": 640, "grad_norm_max": 1.3615809679031372, "hidden_mse": 0.17425470147281885, "loss": 0.09075440168380738, "lr_blocks": 8.596699001693257e-06, "lr_fusion": 8.135881792367686e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.6646606730137137, "epoch": 9, "flow_mse": 0.07165623712353408, "global_step": 650, "grad_norm_max": 2.018874406814575, "hidden_mse": 0.17005239613354206, "loss": 0.08866147724911572, "lr_blocks": 8.535533905932739e-06, "lr_fusion": 8.07106356344834e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 2.01382767639152, "epoch": 9, "flow_mse": 0.07549470965750515, "global_step": 660, "grad_norm_max": 1.9090533256530762, "hidden_mse": 0.1704067310318351, "loss": 0.0925353833474219, "lr_blocks": 8.473291852294986e-06, "lr_fusion": 8.005405736415126e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 2.044479904185573, "epoch": 10, "flow_mse": 0.07384380707517267, "global_step": 670, "grad_norm_max": 1.6318471431732178, "hidden_mse": 0.16633677743375302, "loss": 0.09047748530283571, "lr_blocks": 8.409991800312493e-06, "lr_fusion": 7.938926261462366e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.994344039564021, "epoch": 10, "flow_mse": 0.07081504138186574, "global_step": 680, "grad_norm_max": 2.8086788654327393, "hidden_mse": 0.16047870106995107, "loss": 0.08686291193589568, "lr_blocks": 8.345653031794292e-06, "lr_fusion": 7.871643313414718e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.7797691442654469, "epoch": 10, "flow_mse": 0.07098656282760203, "global_step": 690, "grad_norm_max": 1.6218206882476807, "hidden_mse": 0.15934279719367625, "loss": 0.086920843180269, "lr_blocks": 8.280295144952537e-06, "lr_fusion": 7.803575286758364e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8434631714044372, "epoch": 10, "flow_mse": 0.07692104973830283, "global_step": 700, "grad_norm_max": 1.9783955812454224, "hidden_mse": 0.16042201202362777, "loss": 0.0929632511921227, "lr_blocks": 8.213938048432697e-06, "lr_fusion": 7.734740790612136e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8933369300531924, "epoch": 10, "flow_mse": 0.06859853272326291, "global_step": 710, "grad_norm_max": 1.2764592170715332, "hidden_mse": 0.15258000474423172, "loss": 0.08385653356090189, "lr_blocks": 8.146601955249187e-06, "lr_fusion": 7.66515864363997e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.863453762442805, "epoch": 10, "flow_mse": 0.0677291950210929, "global_step": 720, "grad_norm_max": 2.756103992462158, "hidden_mse": 0.15030786208808422, "loss": 0.08275998155586423, "lr_blocks": 8.078307376628292e-06, "lr_fusion": 7.594847868906076e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.5886019194789698, "epoch": 11, "flow_mse": 0.06930862842127682, "global_step": 730, "grad_norm_max": 2.032028913497925, "hidden_mse": 0.14793342733755707, "loss": 0.08410197151824833, "lr_blocks": 8.009075115760243e-06, "lr_fusion": 7.52382768867422e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.7562155689141947, "epoch": 11, "flow_mse": 0.0670456328894943, "global_step": 740, "grad_norm_max": 1.5815403461456299, "hidden_mse": 0.14644069485366346, "loss": 0.08168970262631775, "lr_blocks": 7.938926261462366e-06, "lr_fusion": 7.452117519152542e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.9935554676223548, "epoch": 11, "flow_mse": 0.06759789921343327, "global_step": 750, "grad_norm_max": 1.5702470541000366, "hidden_mse": 0.1456328984349966, "loss": 0.08216118961572648, "lr_blocks": 7.86788218175523e-06, "lr_fusion": 7.379736965185368e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.7392390850000083, "epoch": 11, "flow_mse": 0.0679298588540405, "global_step": 760, "grad_norm_max": 1.7798969745635986, "hidden_mse": 0.145010856539011, "loss": 0.0824309445451945, "lr_blocks": 7.795964517353734e-06, "lr_fusion": 7.30670581489344e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.883487955624878, "epoch": 11, "flow_mse": 0.06899964096955956, "global_step": 770, "grad_norm_max": 2.0383424758911133, "hidden_mse": 0.14355416363105178, "loss": 0.08335505775175989, "lr_blocks": 7.723195175075136e-06, "lr_fusion": 7.233044034264034e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.902795129803417, "epoch": 11, "flow_mse": 0.06839722348377109, "global_step": 780, "grad_norm_max": 1.341465950012207, "hidden_mse": 0.14199498109519482, "loss": 0.08259672177955509, "lr_blocks": 7.649596321166024e-06, "lr_fusion": 7.158771761692464e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.899291706715303, "epoch": 11, "flow_mse": 0.06829775697551668, "global_step": 790, "grad_norm_max": 1.6974772214889526, "hidden_mse": 0.14167611030861735, "loss": 0.08246536822989584, "lr_blocks": 7.575190374550272e-06, "lr_fusion": 7.083909302476453e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.9210877921635985, "epoch": 12, "flow_mse": 0.06724138637073339, "global_step": 800, "grad_norm_max": 1.6519050598144531, "hidden_mse": 0.13736099740490318, "loss": 0.08097748612053693, "lr_blocks": 7.500000000000001e-06, "lr_fusion": 7.008477123264848e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.839930806370103, "epoch": 12, "flow_mse": 0.06486202948726714, "global_step": 810, "grad_norm_max": 1.9736137390136719, "hidden_mse": 0.13331869533285498, "loss": 0.07819389933720232, "lr_blocks": 7.424048101231687e-06, "lr_fusion": 6.932495846462261e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.7203275968320668, "epoch": 12, "flow_mse": 0.06997489500790835, "global_step": 820, "grad_norm_max": 1.8273866176605225, "hidden_mse": 0.1383979432284832, "loss": 0.0838146897032857, "lr_blocks": 7.347357813929455e-06, "lr_fusion": 6.855986244591104e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8308625894540456, "epoch": 12, "flow_mse": 0.06103807487525046, "global_step": 830, "grad_norm_max": 1.24065363407135, "hidden_mse": 0.12885685674846173, "loss": 0.07392376055940986, "lr_blocks": 7.269952498697734e-06, "lr_fusion": 6.778969234612584e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.951346460581408, "epoch": 12, "flow_mse": 0.0672075655311346, "global_step": 840, "grad_norm_max": 1.4299647808074951, "hidden_mse": 0.1341157573275268, "loss": 0.08061914164572954, "lr_blocks": 7.191855733945388e-06, "lr_fusion": 6.701465872208216e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.9852554764482193, "epoch": 12, "flow_mse": 0.06508104703389109, "global_step": 850, "grad_norm_max": 1.3047852516174316, "hidden_mse": 0.13088483419269323, "loss": 0.07816953058354557, "lr_blocks": 7.113091308703498e-06, "lr_fusion": 6.623497346023418e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8463190050082632, "epoch": 13, "flow_mse": 0.06611472489312291, "global_step": 860, "grad_norm_max": 1.3949657678604126, "hidden_mse": 0.13323927726596593, "loss": 0.07943865223787726, "lr_blocks": 7.033683215379002e-06, "lr_fusion": 6.545084971874738e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8538904657267268, "epoch": 13, "flow_mse": 0.0621907118242234, "global_step": 870, "grad_norm_max": 1.0381734371185303, "hidden_mse": 0.1276529028080404, "loss": 0.07495600231923163, "lr_blocks": 6.953655642446368e-06, "lr_fusion": 6.466250186922325e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 2.0335181808739438, "epoch": 13, "flow_mse": 0.06287266602739693, "global_step": 880, "grad_norm_max": 1.0403690338134766, "hidden_mse": 0.1264244217425585, "loss": 0.07551510813646019, "lr_blocks": 6.873032967079562e-06, "lr_fusion": 6.387014543809223e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.6593307598377578, "epoch": 13, "flow_mse": 0.0616169007960707, "global_step": 890, "grad_norm_max": 1.445999264717102, "hidden_mse": 0.12434439333155752, "loss": 0.07405133983120323, "lr_blocks": 6.7918397477265e-06, "lr_fusion": 6.307399704769099e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.6712814146303572, "epoch": 13, "flow_mse": 0.0653102396056056, "global_step": 900, "grad_norm_max": 2.982330322265625, "hidden_mse": 0.12486500488594175, "loss": 0.07779674059711397, "lr_blocks": 6.710100716628345e-06, "lr_fusion": 6.227427435703997e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.9577245332402526, "epoch": 13, "flow_mse": 0.06775598619133234, "global_step": 910, "grad_norm_max": 1.3592380285263062, "hidden_mse": 0.1261634855531156, "loss": 0.08037233473733069, "lr_blocks": 6.627840772285784e-06, "lr_fusion": 6.147119600233758e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.87712880110048, "epoch": 13, "flow_mse": 0.06632841904647649, "global_step": 920, "grad_norm_max": 1.6363564729690552, "hidden_mse": 0.12368027167394757, "loss": 0.07869644630700351, "lr_blocks": 6.545084971874738e-06, "lr_fusion": 6.066498153718735e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 2.105622898535512, "epoch": 14, "flow_mse": 0.0650685330387205, "global_step": 930, "grad_norm_max": 0.9703700542449951, "hidden_mse": 0.12554309023544194, "loss": 0.07762284209020437, "lr_blocks": 6.461858523613684e-06, "lr_fusion": 5.985585137257401e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.9442461250553607, "epoch": 14, "flow_mse": 0.06523986770771444, "global_step": 940, "grad_norm_max": 1.2431607246398926, "hidden_mse": 0.12500393418595196, "loss": 0.07774026063270867, "lr_blocks": 6.378186779084996e-06, "lr_fusion": 5.90440267166055e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.828690172272036, "epoch": 14, "flow_mse": 0.05963015616871417, "global_step": 950, "grad_norm_max": 1.5364301204681396, "hidden_mse": 0.11931315260007978, "loss": 0.07156147151254118, "lr_blocks": 6.294095225512604e-06, "lr_fusion": 5.8229729514036705e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8559695239790017, "epoch": 14, "flow_mse": 0.05979857188649476, "global_step": 960, "grad_norm_max": 1.4135713577270508, "hidden_mse": 0.11980592980980874, "loss": 0.07177916495129466, "lr_blocks": 6.209609477998339e-06, "lr_fusion": 5.74131823855921e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.7981327558882185, "epoch": 14, "flow_mse": 0.06091945511288941, "global_step": 970, "grad_norm_max": 1.1532989740371704, "hidden_mse": 0.1190660865046084, "loss": 0.0728260634932667, "lr_blocks": 6.124755271719326e-06, "lr_fusion": 5.6594608567103456e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8538059903803514, "epoch": 14, "flow_mse": 0.05959429889917374, "global_step": 980, "grad_norm_max": 1.2059638500213623, "hidden_mse": 0.11662983177229763, "loss": 0.07125728223472834, "lr_blocks": 6.039558454088796e-06, "lr_fusion": 5.577423184847932e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.7332944313558984, "epoch": 14, "flow_mse": 0.06722243609838188, "global_step": 990, "grad_norm_max": 1.3793468475341797, "hidden_mse": 0.12186120152473451, "loss": 0.07940855622291565, "lr_blocks": 5.954044976882725e-06, "lr_fusion": 5.495227651252315e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.7073448181181448, "epoch": 15, "flow_mse": 0.05920077268965543, "global_step": 1000, "grad_norm_max": 1.3150396347045898, "hidden_mse": 0.11587047157809138, "loss": 0.07078781956806779, "lr_blocks": 5.8682408883346535e-06, "lr_fusion": 5.4128967273616625e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8634853277268122, "epoch": 15, "flow_mse": 0.06008068108931184, "global_step": 1010, "grad_norm_max": 1.0294694900512695, "hidden_mse": 0.11746087549254298, "loss": 0.07182676899246872, "lr_blocks": 5.782172325201155e-06, "lr_fusion": 5.330452921628497e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8640984378071153, "epoch": 15, "flow_mse": 0.06041769445873797, "global_step": 1020, "grad_norm_max": 1.2436527013778687, "hidden_mse": 0.11662168577313424, "loss": 0.0720798633992672, "lr_blocks": 5.695865504800328e-06, "lr_fusion": 5.247918773366112e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 2.044428478881309, "epoch": 15, "flow_mse": 0.059971056133508685, "global_step": 1030, "grad_norm_max": 0.8764917850494385, "hidden_mse": 0.11762538189068437, "loss": 0.07173359449952842, "lr_blocks": 5.609346717025738e-06, "lr_fusion": 5.165316846586541e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.7723727050048184, "epoch": 15, "flow_mse": 0.058173052780330185, "global_step": 1040, "grad_norm_max": 1.1820777654647827, "hidden_mse": 0.11490522287786008, "loss": 0.06966357519850135, "lr_blocks": 5.522642316338268e-06, "lr_fusion": 5.0826697238317935e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.9333946266793647, "epoch": 15, "flow_mse": 0.06083732731640339, "global_step": 1050, "grad_norm_max": 0.8588114380836487, "hidden_mse": 0.11753584127873183, "loss": 0.07259091124869883, "lr_blocks": 5.435778713738292e-06, "lr_fusion": 5e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.9301579607854364, "epoch": 16, "flow_mse": 0.05984115526080132, "global_step": 1060, "grad_norm_max": 0.7564420104026794, "hidden_mse": 0.116309258248657, "loss": 0.07147208084352315, "lr_blocks": 5.348782368720627e-06, "lr_fusion": 4.917330276168208e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.6499796213785887, "epoch": 16, "flow_mse": 0.05933821597136557, "global_step": 1070, "grad_norm_max": 1.2188807725906372, "hidden_mse": 0.11562003139406443, "loss": 0.07090021898038686, "lr_blocks": 5.2616797812147205e-06, "lr_fusion": 4.834683153413459e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.7709604913383374, "epoch": 16, "flow_mse": 0.05792279005981982, "global_step": 1080, "grad_norm_max": 0.9248772263526917, "hidden_mse": 0.11402276260778309, "loss": 0.0693250666372478, "lr_blocks": 5.174497483512506e-06, "lr_fusion": 4.7520812266338885e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8961538497242145, "epoch": 16, "flow_mse": 0.05921817822381854, "global_step": 1090, "grad_norm_max": 0.9744265675544739, "hidden_mse": 0.11493033999577165, "loss": 0.07071121269837022, "lr_blocks": 5.087262032186418e-06, "lr_fusion": 4.669547078371504e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8375497608751177, "epoch": 16, "flow_mse": 0.05829775258898735, "global_step": 1100, "grad_norm_max": 0.9396805763244629, "hidden_mse": 0.11368074417114259, "loss": 0.06966582736931741, "lr_blocks": 5e-06, "lr_fusion": 4.5871032726383386e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8890359286408058, "epoch": 16, "flow_mse": 0.056764368247240786, "global_step": 1110, "grad_norm_max": 1.4962784051895142, "hidden_mse": 0.11246938547119499, "loss": 0.06801130706444383, "lr_blocks": 4.9127379678135825e-06, "lr_fusion": 4.504772348747687e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.9199464306249867, "epoch": 16, "flow_mse": 0.05812495080754161, "global_step": 1120, "grad_norm_max": 1.0150479078292847, "hidden_mse": 0.11339842304587365, "loss": 0.06946479328908027, "lr_blocks": 4.825502516487497e-06, "lr_fusion": 4.4225768151520694e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.9528670382205746, "epoch": 17, "flow_mse": 0.0582504840567708, "global_step": 1130, "grad_norm_max": 0.5666173100471497, "hidden_mse": 0.11305413423106075, "loss": 0.0695558977779001, "lr_blocks": 4.738320218785281e-06, "lr_fusion": 4.3405391432896555e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8691815458922065, "epoch": 17, "flow_mse": 0.056232606945559385, "global_step": 1140, "grad_norm_max": 1.0409326553344727, "hidden_mse": 0.11095626167953015, "loss": 0.06732823327183723, "lr_blocks": 4.651217631279374e-06, "lr_fusion": 4.2586817614407895e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.9381392417169991, "epoch": 17, "flow_mse": 0.05920435013249517, "global_step": 1150, "grad_norm_max": 0.8891173005104065, "hidden_mse": 0.11471081618219614, "loss": 0.07067543230950833, "lr_blocks": 4.564221286261709e-06, "lr_fusion": 4.17702704859633e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.7007469822870918, "epoch": 17, "flow_mse": 0.05892735007219017, "global_step": 1160, "grad_norm_max": 0.9542156457901001, "hidden_mse": 0.11405815165489913, "loss": 0.07033316562883556, "lr_blocks": 4.477357683661734e-06, "lr_fusion": 4.095597328339452e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.813334083685186, "epoch": 17, "flow_mse": 0.05461530075408519, "global_step": 1170, "grad_norm_max": 0.8323894143104553, "hidden_mse": 0.10710006449371577, "loss": 0.0653253072872758, "lr_blocks": 4.390653282974264e-06, "lr_fusion": 4.0144148627425993e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.7458943290723257, "epoch": 17, "flow_mse": 0.058256217977032065, "global_step": 1180, "grad_norm_max": 1.1402466297149658, "hidden_mse": 0.11414804682135582, "loss": 0.06967102284543217, "lr_blocks": 4.304134495199675e-06, "lr_fusion": 3.933501846281267e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.915744307746354, "epoch": 18, "flow_mse": 0.055332719627767805, "global_step": 1190, "grad_norm_max": 0.8607915639877319, "hidden_mse": 0.10889494800940157, "loss": 0.06622221460565925, "lr_blocks": 4.217827674798845e-06, "lr_fusion": 3.852880399766243e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.9392635020019953, "epoch": 18, "flow_mse": 0.05797289633192122, "global_step": 1200, "grad_norm_max": 1.8679325580596924, "hidden_mse": 0.11137517839670182, "loss": 0.06911041443236172, "lr_blocks": 4.131759111665349e-06, "lr_fusion": 3.772572564296005e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.7942509921369494, "epoch": 18, "flow_mse": 0.056557841086760166, "global_step": 1210, "grad_norm_max": 0.794858992099762, "hidden_mse": 0.11070305658504367, "loss": 0.06762814656831324, "lr_blocks": 4.045955023117276e-06, "lr_fusion": 3.6926002952309016e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.6715436912636505, "epoch": 18, "flow_mse": 0.0571835795417428, "global_step": 1220, "grad_norm_max": 0.9790372848510742, "hidden_mse": 0.11226286897435785, "loss": 0.06840986674651504, "lr_blocks": 3.960441545911205e-06, "lr_fusion": 3.612985456190778e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 2.107250372883573, "epoch": 18, "flow_mse": 0.0564295522402972, "global_step": 1230, "grad_norm_max": 0.7570212483406067, "hidden_mse": 0.10993242673575879, "loss": 0.06742279478348792, "lr_blocks": 3.875244728280676e-06, "lr_fusion": 3.533749813077677e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 2.047389568622748, "epoch": 18, "flow_mse": 0.05746471900492907, "global_step": 1240, "grad_norm_max": 0.8607873320579529, "hidden_mse": 0.11112378425896169, "loss": 0.06857709758915008, "lr_blocks": 3.790390522001662e-06, "lr_fusion": 3.4549150281252636e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.5449788220197662, "epoch": 18, "flow_mse": 0.05702267340384424, "global_step": 1250, "grad_norm_max": 0.6349190473556519, "hidden_mse": 0.11191629096865655, "loss": 0.06821430274285377, "lr_blocks": 3.705904774487396e-06, "lr_fusion": 3.3765026539765834e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8875254256010523, "epoch": 19, "flow_mse": 0.05772090326063335, "global_step": 1260, "grad_norm_max": 0.5843414664268494, "hidden_mse": 0.11196697549894452, "loss": 0.06891760104335845, "lr_blocks": 3.6218132209150047e-06, "lr_fusion": 3.298534127791785e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8776245697605192, "epoch": 19, "flow_mse": 0.05313115566968918, "global_step": 1270, "grad_norm_max": 0.44912803173065186, "hidden_mse": 0.1067539899609983, "loss": 0.06380655500106514, "lr_blocks": 3.538141476386317e-06, "lr_fusion": 3.221030765387417e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.9916057762879065, "epoch": 19, "flow_mse": 0.05610430384986103, "global_step": 1280, "grad_norm_max": 0.7865238785743713, "hidden_mse": 0.10909662246704102, "loss": 0.06701396652497352, "lr_blocks": 3.4549150281252635e-06, "lr_fusion": 3.144013755408895e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.819823783946049, "epoch": 19, "flow_mse": 0.05514522660523653, "global_step": 1290, "grad_norm_max": 0.9301948547363281, "hidden_mse": 0.10944940159097315, "loss": 0.06609016689471901, "lr_blocks": 3.372159227714218e-06, "lr_fusion": 3.0675041535377405e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 2.057221127783123, "epoch": 19, "flow_mse": 0.056626258650794625, "global_step": 1300, "grad_norm_max": 0.5852092504501343, "hidden_mse": 0.1101443249732256, "loss": 0.06764069120399654, "lr_blocks": 3.289899283371657e-06, "lr_fusion": 2.991522876735154e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.6946213100774914, "epoch": 19, "flow_mse": 0.05420284080319107, "global_step": 1310, "grad_norm_max": 0.49697157740592957, "hidden_mse": 0.10914511308073999, "loss": 0.06511735212989152, "lr_blocks": 3.2081602522734987e-06, "lr_fusion": 2.916090697523549e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.6702450912911446, "epoch": 19, "flow_mse": 0.05620301379822195, "global_step": 1320, "grad_norm_max": 0.8781030178070068, "hidden_mse": 0.11186566464602948, "loss": 0.067389580514282, "lr_blocks": 3.12696703292044e-06, "lr_fusion": 2.8412282383075363e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8671580449008616, "epoch": 20, "flow_mse": 0.057774351164698604, "global_step": 1330, "grad_norm_max": 0.6999236941337585, "hidden_mse": 0.1124896040186286, "loss": 0.06902331141754985, "lr_blocks": 3.0463443575536324e-06, "lr_fusion": 2.766955965735968e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.7934378478326836, "epoch": 20, "flow_mse": 0.05308823841623962, "global_step": 1340, "grad_norm_max": 0.6592912673950195, "hidden_mse": 0.10645070374011995, "loss": 0.06373330899514258, "lr_blocks": 2.966316784621e-06, "lr_fusion": 2.693294185106562e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.983334714593366, "epoch": 20, "flow_mse": 0.05545833441428841, "global_step": 1350, "grad_norm_max": 0.619492769241333, "hidden_mse": 0.10996899055317044, "loss": 0.06645523337647319, "lr_blocks": 2.886908691296504e-06, "lr_fusion": 2.6202630348146324e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.707897415009211, "epoch": 20, "flow_mse": 0.05316436188295484, "global_step": 1360, "grad_norm_max": 0.7670770883560181, "hidden_mse": 0.10698874769732357, "loss": 0.06386323706246913, "lr_blocks": 2.8081442660546126e-06, "lr_fusion": 2.547882480847461e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8228236074690358, "epoch": 20, "flow_mse": 0.05519023141823709, "global_step": 1370, "grad_norm_max": 0.5514799356460571, "hidden_mse": 0.10873732827603817, "loss": 0.06606396441347898, "lr_blocks": 2.7300475013022666e-06, "lr_fusion": 2.476172311325783e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8020589397929143, "epoch": 20, "flow_mse": 0.05520334360189736, "global_step": 1380, "grad_norm_max": 0.7075578570365906, "hidden_mse": 0.10943609643727542, "loss": 0.0661469531711191, "lr_blocks": 2.6526421860705474e-06, "lr_fusion": 2.405152131093926e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 2.1456252824849797, "epoch": 21, "flow_mse": 0.055343919666484, "global_step": 1390, "grad_norm_max": 0.4367266297340393, "hidden_mse": 0.10879806661978364, "loss": 0.0662237268872559, "lr_blocks": 2.5759518987683154e-06, "lr_fusion": 2.3348413563600325e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.9225107472884702, "epoch": 21, "flow_mse": 0.05379870082251728, "global_step": 1400, "grad_norm_max": 0.45527708530426025, "hidden_mse": 0.10747067201882601, "loss": 0.0645457681734115, "lr_blocks": 2.5000000000000015e-06, "lr_fusion": 2.2652592093878666e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8355080222332618, "epoch": 21, "flow_mse": 0.05405601593665779, "global_step": 1410, "grad_norm_max": 0.578701376914978, "hidden_mse": 0.10813164105638862, "loss": 0.06486918060109019, "lr_blocks": 2.424809625449729e-06, "lr_fusion": 2.196424713241637e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8781015649496113, "epoch": 21, "flow_mse": 0.05536241387017071, "global_step": 1420, "grad_norm_max": 0.38821011781692505, "hidden_mse": 0.11099598519504071, "loss": 0.06646201252005994, "lr_blocks": 2.3504036788339763e-06, "lr_fusion": 2.128356686585282e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.7020483079002589, "epoch": 21, "flow_mse": 0.05298192100599408, "global_step": 1430, "grad_norm_max": 0.5125593543052673, "hidden_mse": 0.10700325099751354, "loss": 0.06368224625475705, "lr_blocks": 2.2768048249248648e-06, "lr_fusion": 2.061073738537635e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8793439952991322, "epoch": 21, "flow_mse": 0.05324047850444913, "global_step": 1440, "grad_norm_max": 0.36654290556907654, "hidden_mse": 0.10668059764429927, "loss": 0.06390853836201131, "lr_blocks": 2.204035482646267e-06, "lr_fusion": 1.9945942635848748e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.697708303979016, "epoch": 21, "flow_mse": 0.055548487138003116, "global_step": 1450, "grad_norm_max": 0.5792702436447144, "hidden_mse": 0.11123776510357858, "loss": 0.06667226417921483, "lr_blocks": 2.132117818244771e-06, "lr_fusion": 1.928936436551661e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8742496912454953, "epoch": 22, "flow_mse": 0.05278900931589306, "global_step": 1460, "grad_norm_max": 0.41618072986602783, "hidden_mse": 0.10622980045154691, "loss": 0.06341198971495032, "lr_blocks": 2.061073738537635e-06, "lr_fusion": 1.8641182076323148e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 2.0268068581106493, "epoch": 22, "flow_mse": 0.05427227606996894, "global_step": 1470, "grad_norm_max": 0.38854727149009705, "hidden_mse": 0.10826945109292865, "loss": 0.06509922100231051, "lr_blocks": 1.990924884239758e-06, "lr_fusion": 1.800157297483417e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.6323719802472625, "epoch": 22, "flow_mse": 0.05504520414397121, "global_step": 1480, "grad_norm_max": 0.42108166217803955, "hidden_mse": 0.11108364090323448, "loss": 0.06615356872789561, "lr_blocks": 1.9216926233717087e-06, "lr_fusion": 1.7370711923791567e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.7635257252361045, "epoch": 22, "flow_mse": 0.05449590594507754, "global_step": 1490, "grad_norm_max": 0.2980191111564636, "hidden_mse": 0.10923452377319337, "loss": 0.06541935848072171, "lr_blocks": 1.8533980447508138e-06, "lr_fusion": 1.6748771394307585e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.7368722179773615, "epoch": 22, "flow_mse": 0.05059092580340803, "global_step": 1500, "grad_norm_max": 0.3388434052467346, "hidden_mse": 0.10435767890885472, "loss": 0.06102669364772737, "lr_blocks": 1.7860619515673034e-06, "lr_fusion": 1.6135921418712956e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 2.083392089595145, "epoch": 22, "flow_mse": 0.05451004351489246, "global_step": 1510, "grad_norm_max": 0.31937962770462036, "hidden_mse": 0.10862588789314032, "loss": 0.0653726323042065, "lr_blocks": 1.7197048550474643e-06, "lr_fusion": 1.553232954407171e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.7850786256254652, "epoch": 23, "flow_mse": 0.05411349283531308, "global_step": 1520, "grad_norm_max": 0.43102946877479553, "hidden_mse": 0.1085999496281147, "loss": 0.06497348803095519, "lr_blocks": 1.6543469682057105e-06, "lr_fusion": 1.4938160786375572e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.6930420300239348, "epoch": 23, "flow_mse": 0.05289990794844926, "global_step": 1530, "grad_norm_max": 0.3967125415802002, "hidden_mse": 0.10759261148050428, "loss": 0.06365916887298227, "lr_blocks": 1.5900081996875083e-06, "lr_fusion": 1.435357758543015e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8721173314916086, "epoch": 23, "flow_mse": 0.05183863551355899, "global_step": 1540, "grad_norm_max": 0.65227872133255, "hidden_mse": 0.10619828049093485, "loss": 0.062458463944494724, "lr_blocks": 1.5267081477050132e-06, "lr_fusion": 1.3778739760445552e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.9260684134074837, "epoch": 23, "flow_mse": 0.05517242061905563, "global_step": 1550, "grad_norm_max": 0.3173617720603943, "hidden_mse": 0.11037433873862029, "loss": 0.06620985479094088, "lr_blocks": 1.4644660940672628e-06, "lr_fusion": 1.3213804466343421e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 2.00732960163441, "epoch": 23, "flow_mse": 0.052165593765676026, "global_step": 1560, "grad_norm_max": 0.48366814851760864, "hidden_mse": 0.10552244568243624, "loss": 0.06271783891133964, "lr_blocks": 1.4033009983067454e-06, "lr_fusion": 1.2658926150792322e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 2.0028102155803937, "epoch": 23, "flow_mse": 0.054785998072475195, "global_step": 1570, "grad_norm_max": 0.32774579524993896, "hidden_mse": 0.1094754220917821, "loss": 0.0657335405703634, "lr_blocks": 1.3432314919041478e-06, "lr_fusion": 1.2114256511983274e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8898658655904002, "epoch": 23, "flow_mse": 0.05355181619524956, "global_step": 1580, "grad_norm_max": 0.29386377334594727, "hidden_mse": 0.10809278432279826, "loss": 0.06436109468340874, "lr_blocks": 1.2842758726130283e-06, "lr_fusion": 1.157994445715706e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8055248979391765, "epoch": 24, "flow_mse": 0.05223282501101494, "global_step": 1590, "grad_norm_max": 0.30623888969421387, "hidden_mse": 0.10633350862190127, "loss": 0.06286617605946958, "lr_blocks": 1.22645209888614e-06, "lr_fusion": 1.1056136061894384e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.697861864778679, "epoch": 24, "flow_mse": 0.05407982333563269, "global_step": 1600, "grad_norm_max": 0.41349685192108154, "hidden_mse": 0.1090579808689654, "loss": 0.06498562130145728, "lr_blocks": 1.1697777844051105e-06, "lr_fusion": 1.0542974530180327e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.858836678104126, "epoch": 24, "flow_mse": 0.0515467718243599, "global_step": 1610, "grad_norm_max": 0.2459108531475067, "hidden_mse": 0.10587860029190779, "loss": 0.06213463232852519, "lr_blocks": 1.1142701927151456e-06, "lr_fusion": 1.0040600155253765e-05, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.7442898534587585, "epoch": 24, "flow_mse": 0.05539359417743981, "global_step": 1620, "grad_norm_max": 0.4176762104034424, "hidden_mse": 0.11120789768174291, "loss": 0.06651438400149345, "lr_blocks": 1.0599462319663906e-06, "lr_fusion": 9.549150281252633e-06, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.9312700220412808, "epoch": 24, "flow_mse": 0.05194714558310807, "global_step": 1630, "grad_norm_max": 0.27055931091308594, "hidden_mse": 0.10536178648471833, "loss": 0.06248332438990474, "lr_blocks": 1.006822449763537e-06, "lr_fusion": 9.068759265665384e-06, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.9848020562625608, "epoch": 24, "flow_mse": 0.05159918488934637, "global_step": 1640, "grad_norm_max": 0.2410484105348587, "hidden_mse": 0.10558492876589298, "loss": 0.06215767813846469, "lr_blocks": 9.549150281252633e-07, "lr_fusion": 8.599558442598998e-06, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.9862558376597008, "epoch": 24, "flow_mse": 0.051943618524819615, "global_step": 1650, "grad_norm_max": 0.19621193408966064, "hidden_mse": 0.10532341394573451, "loss": 0.062475959723815325, "lr_blocks": 9.042397785550405e-07, "lr_fusion": 8.141676086873572e-06, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.90772508844384, "epoch": 25, "flow_mse": 0.05241604868788272, "global_step": 1660, "grad_norm_max": 0.24477998912334442, "hidden_mse": 0.10615330189466476, "loss": 0.06303137941285968, "lr_blocks": 8.54812137224792e-07, "lr_fusion": 7.695237378953223e-06, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 2.0734705874245267, "epoch": 25, "flow_mse": 0.052382499817758804, "global_step": 1670, "grad_norm_max": 0.18695969879627228, "hidden_mse": 0.10536184627562761, "loss": 0.06291868472471833, "lr_blocks": 8.066471602728804e-07, "lr_fusion": 7.260364370723044e-06, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.784898155926203, "epoch": 25, "flow_mse": 0.05266823512502015, "global_step": 1680, "grad_norm_max": 0.30488482117652893, "hidden_mse": 0.10819421242922544, "loss": 0.0634876563679427, "lr_blocks": 7.597595192178702e-07, "lr_fusion": 6.837175952121306e-06, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.7039467210503063, "epoch": 25, "flow_mse": 0.053952160244807604, "global_step": 1690, "grad_norm_max": 0.21721898019313812, "hidden_mse": 0.10940732667222619, "loss": 0.06489289305172861, "lr_blocks": 7.141634964894389e-07, "lr_fusion": 6.425787818636131e-06, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.7925323831252173, "epoch": 25, "flow_mse": 0.054602593369781974, "global_step": 1700, "grad_norm_max": 0.2606789171695709, "hidden_mse": 0.11060061100870372, "loss": 0.06566265453584492, "lr_blocks": 6.698729810778065e-07, "lr_fusion": 6.026312439675552e-06, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8506990019581282, "epoch": 25, "flow_mse": 0.05084327077493072, "global_step": 1710, "grad_norm_max": 0.18723498284816742, "hidden_mse": 0.10409033391624689, "loss": 0.06125230453908444, "lr_blocks": 6.269014643030214e-07, "lr_fusion": 5.6388590278194096e-06, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.696594527331763, "epoch": 26, "flow_mse": 0.04985265554860235, "global_step": 1720, "grad_norm_max": 0.16147896647453308, "hidden_mse": 0.10344723165035248, "loss": 0.060197379020974044, "lr_blocks": 5.852620357053651e-07, "lr_fusion": 5.263533508961827e-06, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.7064527680224275, "epoch": 26, "flow_mse": 0.052878752723336225, "global_step": 1730, "grad_norm_max": 0.21210241317749023, "hidden_mse": 0.10743851037696005, "loss": 0.0636226034257561, "lr_blocks": 5.449673790581611e-07, "lr_fusion": 4.900438493352055e-06, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.9300253918670933, "epoch": 26, "flow_mse": 0.05500989542342723, "global_step": 1740, "grad_norm_max": 0.18679021298885345, "hidden_mse": 0.11223258469253779, "loss": 0.06623315447941423, "lr_blocks": 5.06029768504166e-07, "lr_fusion": 4.549673247541875e-06, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.6863267055538018, "epoch": 26, "flow_mse": 0.05216321973130107, "global_step": 1750, "grad_norm_max": 0.1739465296268463, "hidden_mse": 0.10685156406834723, "loss": 0.06284837638959288, "lr_blocks": 4.6846106481675035e-07, "lr_fusion": 4.2113336672471245e-06, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.9015966636099622, "epoch": 26, "flow_mse": 0.05238161133602262, "global_step": 1760, "grad_norm_max": 0.16062317788600922, "hidden_mse": 0.1064825899899006, "loss": 0.06302987090311944, "lr_blocks": 4.322727117869951e-07, "lr_fusion": 3.885512251130763e-06, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.89506446889136, "epoch": 26, "flow_mse": 0.05142894261516631, "global_step": 1770, "grad_norm_max": 0.19871976971626282, "hidden_mse": 0.10540924789384008, "loss": 0.06196986767463386, "lr_blocks": 3.9747573273779816e-07, "lr_fusion": 3.5722980755146517e-06, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 2.140559560073598, "epoch": 26, "flow_mse": 0.052294114232063295, "global_step": 1780, "grad_norm_max": 0.13666212558746338, "hidden_mse": 0.10540376659482718, "loss": 0.06283449097536505, "lr_blocks": 3.6408072716606346e-07, "lr_fusion": 3.271776770026963e-06, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.9869108784609126, "epoch": 27, "flow_mse": 0.052665405767038466, "global_step": 1790, "grad_norm_max": 0.1539895087480545, "hidden_mse": 0.1076177385635674, "loss": 0.06342717967927457, "lr_blocks": 3.320978675139919e-07, "lr_fusion": 2.9840304941919415e-06, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8483484975920874, "epoch": 27, "flow_mse": 0.05281953318044544, "global_step": 1800, "grad_norm_max": 0.1717279851436615, "hidden_mse": 0.10747445961460472, "loss": 0.06356697916053236, "lr_blocks": 3.015368960704584e-07, "lr_fusion": 2.7091379149682685e-06, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.7725311637754204, "epoch": 27, "flow_mse": 0.053238363936543465, "global_step": 1810, "grad_norm_max": 0.129792258143425, "hidden_mse": 0.10810153055936099, "loss": 0.06404851726256311, "lr_blocks": 2.724071220034158e-07, "lr_fusion": 2.4471741852423237e-06, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.9046508581523085, "epoch": 27, "flow_mse": 0.05027513452805579, "global_step": 1820, "grad_norm_max": 0.13140571117401123, "hidden_mse": 0.1037200196646154, "loss": 0.06064713676460087, "lr_blocks": 2.447174185242324e-07, "lr_fusion": 2.1982109232821178e-06, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.7696023706230335, "epoch": 27, "flow_mse": 0.053843728452920916, "global_step": 1830, "grad_norm_max": 0.10490740090608597, "hidden_mse": 0.10948641048744322, "loss": 0.06479236944578588, "lr_blocks": 2.1847622018482283e-07, "lr_fusion": 1.962316193157593e-06, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8017618429686992, "epoch": 27, "flow_mse": 0.05273311133496464, "global_step": 1840, "grad_norm_max": 0.1325991004705429, "hidden_mse": 0.10784135181456805, "loss": 0.06351724686101079, "lr_blocks": 1.9369152030840553e-07, "lr_fusion": 1.7395544861325718e-06, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8663546495808987, "epoch": 28, "flow_mse": 0.05010635161306709, "global_step": 1850, "grad_norm_max": 0.13209085166454315, "hidden_mse": 0.10319266710430385, "loss": 0.06042561884969473, "lr_blocks": 1.7037086855465902e-07, "lr_fusion": 1.5299867030334814e-06, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 2.023102572160133, "epoch": 28, "flow_mse": 0.0542366745416075, "global_step": 1860, "grad_norm_max": 0.09557513147592545, "hidden_mse": 0.10798764936625958, "loss": 0.06503543937578797, "lr_blocks": 1.4852136862001766e-07, "lr_fusion": 1.333670137599713e-06, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.9713299808965532, "epoch": 28, "flow_mse": 0.050487085804343225, "global_step": 1870, "grad_norm_max": 0.09084595739841461, "hidden_mse": 0.1046027010306716, "loss": 0.06094735627993941, "lr_blocks": 1.2814967607382433e-07, "lr_fusion": 1.1506584608200367e-06, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.765306223012158, "epoch": 28, "flow_mse": 0.052992650866508485, "global_step": 1880, "grad_norm_max": 0.1068035289645195, "hidden_mse": 0.10854229601100088, "loss": 0.06384688057005405, "lr_blocks": 1.0926199633097156e-07, "lr_fusion": 9.810017062595322e-07, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8488479020117667, "epoch": 28, "flow_mse": 0.05185321508906782, "global_step": 1890, "grad_norm_max": 0.0908748134970665, "hidden_mse": 0.10693074595183134, "loss": 0.06254628994502127, "lr_blocks": 9.186408276168012e-08, "lr_fusion": 8.247462563808817e-07, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.835118718787271, "epoch": 28, "flow_mse": 0.050304826558567585, "global_step": 1900, "grad_norm_max": 0.07472742348909378, "hidden_mse": 0.10459219636395574, "loss": 0.06076404661871493, "lr_blocks": 7.59612349389599e-08, "lr_fusion": 6.819348298638839e-07, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8683127601994785, "epoch": 28, "flow_mse": 0.05247240727767349, "global_step": 1910, "grad_norm_max": 0.0885327011346817, "hidden_mse": 0.1072016378864646, "loss": 0.06319257123395801, "lr_blocks": 6.15582970243117e-08, "lr_fusion": 5.526064699265753e-07, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.6088743058397086, "epoch": 29, "flow_mse": 0.05225406056270004, "global_step": 1920, "grad_norm_max": 0.06508878618478775, "hidden_mse": 0.10751305781304837, "loss": 0.06300536664202809, "lr_blocks": 4.865965629214819e-08, "lr_fusion": 4.367965336512403e-07, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.6788693653565134, "epoch": 29, "flow_mse": 0.05310070822015405, "global_step": 1930, "grad_norm_max": 0.09013048559427261, "hidden_mse": 0.10843602363020183, "loss": 0.0639443107880652, "lr_blocks": 3.726924179339009e-08, "lr_fusion": 3.3453668231809286e-07, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.8483045230590507, "epoch": 29, "flow_mse": 0.05225380118936301, "global_step": 1940, "grad_norm_max": 0.05455300584435463, "hidden_mse": 0.10687568383291364, "loss": 0.06294136978685856, "lr_blocks": 2.7390523158633552e-08, "lr_fusion": 2.458548727494292e-07, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 2.07668071983353, "epoch": 29, "flow_mse": 0.052197744604200125, "global_step": 1950, "grad_norm_max": 0.06954554468393326, "hidden_mse": 0.10710220159962774, "loss": 0.06290796506218613, "lr_blocks": 1.9026509541272276e-08, "lr_fusion": 1.7077534966650766e-07, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.890728423825931, "epoch": 29, "flow_mse": 0.052322149462997916, "global_step": 1960, "grad_norm_max": 0.06443354487419128, "hidden_mse": 0.10821447782218457, "loss": 0.06314359758980573, "lr_blocks": 1.2179748700879013e-08, "lr_fusion": 1.0931863906127327e-07, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.7953088903203025, "epoch": 29, "flow_mse": 0.05093962624669075, "global_step": 1970, "grad_norm_max": 0.07561659812927246, "hidden_mse": 0.10430180076509715, "loss": 0.06136980666778982, "lr_blocks": 6.852326227130835e-09, "lr_fusion": 6.150154258476315e-08, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.842160684411647, "epoch": 29, "flow_mse": 0.053311998723074794, "global_step": 1980, "grad_norm_max": 0.04818754270672798, "hidden_mse": 0.10754476888105274, "loss": 0.06406647614203394, "lr_blocks": 3.0458649045211897e-09, "lr_fusion": 2.7337132953697554e-08, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.7273795908869944, "epoch": 30, "flow_mse": 0.051652479777112605, "global_step": 1990, "grad_norm_max": 0.061520446091890335, "hidden_mse": 0.10536492271348835, "loss": 0.06218897206708789, "lr_blocks": 7.615242180436521e-10, "lr_fusion": 6.834750376549792e-09, "peak_memory_gib": 68.46298551559448}
{"avg_micro_time_s": 1.9548408255825054, "epoch": 30, "flow_mse": 0.05280004227533937, "global_step": 2000, "grad_norm_max": 0.054393548518419266, "hidden_mse": 0.10706793488934636, "loss": 0.06350683607161045, "lr_blocks": 0.0, "lr_fusion": 0.0, "peak_memory_gib": 68.46298551559448}