{"budget_mae": 0.4255277216434479, "budget_mse": 0.29855668544769287, "gradient_norm": 0.8909550309181213, "loss": 2.0634093284606934, "state_acc": 0.3076923191547394, "state_ce": 1.764852523803711, "step": 1, "step_elapsed_sec": 47.49213929194957, "total_elapsed_sec": 47.49214332085103, "train_sample_count": 66, "val_budget_mae": 0.4243839681148529, "val_budget_mse": 0.2999334931373596, "val_loss": 2.085024833679199, "val_sample_count": 11, "val_state_acc": 0.0, "val_state_ce": 1.7850914001464844}
{"budget_mae": 0.3827791213989258, "budget_mse": 0.2534998953342438, "gradient_norm": 0.9616190195083618, "loss": 1.9279193878173828, "state_acc": 0.692307710647583, "state_ce": 1.6744195222854614, "step": 10, "step_elapsed_sec": 0.0031248820014297962, "total_elapsed_sec": 47.526363929733634, "train_sample_count": 66, "val_budget_mae": 0.41653719544410706, "val_budget_mse": 0.29431939125061035, "val_loss": 2.08248233795166, "val_sample_count": 11, "val_state_acc": 0.1818181872367859, "val_state_ce": 1.7881628274917603}
{"budget_mae": 0.33535200357437134, "budget_mse": 0.2089511901140213, "gradient_norm": 0.9492255449295044, "loss": 1.786813735961914, "state_acc": 0.6538461446762085, "state_ce": 1.5778625011444092, "step": 20, "step_elapsed_sec": 0.0030813319608569145, "total_elapsed_sec": 47.54955251608044, "train_sample_count": 66, "val_budget_mae": 0.4108409285545349, "val_budget_mse": 0.2891269326210022, "val_loss": 2.081315040588379, "val_sample_count": 11, "val_state_acc": 0.1818181872367859, "val_state_ce": 1.792188048362732}
{"budget_mae": 0.29686954617500305, "budget_mse": 0.165657639503479, "gradient_norm": 1.0180567502975464, "loss": 1.6360582113265991, "state_acc": 0.6538461446762085, "state_ce": 1.4704005718231201, "step": 30, "step_elapsed_sec": 0.0030339830555021763, "total_elapsed_sec": 47.57231603609398, "train_sample_count": 66, "val_budget_mae": 0.40941867232322693, "val_budget_mse": 0.28554731607437134, "val_loss": 2.0855414867401123, "val_sample_count": 11, "val_state_acc": 0.1818181872367859, "val_state_ce": 1.7999941110610962}
{"budget_mae": 0.265657514333725, "budget_mse": 0.12912486493587494, "gradient_norm": 1.1516008377075195, "loss": 1.4855971336364746, "state_acc": 0.6538461446762085, "state_ce": 1.3564722537994385, "step": 40, "step_elapsed_sec": 0.0029810741543769836, "total_elapsed_sec": 47.59501026710495, "train_sample_count": 66, "val_budget_mae": 0.4111120104789734, "val_budget_mse": 0.2833913266658783, "val_loss": 2.09466552734375, "val_sample_count": 11, "val_state_acc": 0.1818181872367859, "val_state_ce": 1.8112741708755493}
{"budget_mae": 0.24003669619560242, "budget_mse": 0.10103082656860352, "gradient_norm": 1.1429387331008911, "loss": 1.326805591583252, "state_acc": 0.6538461446762085, "state_ce": 1.2257747650146484, "step": 50, "step_elapsed_sec": 0.0029983436688780785, "total_elapsed_sec": 47.61794251482934, "train_sample_count": 66, "val_budget_mae": 0.42112573981285095, "val_budget_mse": 0.29116374254226685, "val_loss": 2.1275925636291504, "val_sample_count": 11, "val_state_acc": 0.1818181872367859, "val_state_ce": 1.8364288806915283}
{"budget_mae": 0.2131318300962448, "budget_mse": 0.0796784907579422, "gradient_norm": 1.0677598714828491, "loss": 1.1761895418167114, "state_acc": 0.6538461446762085, "state_ce": 1.0965110063552856, "step": 60, "step_elapsed_sec": 0.003019853960722685, "total_elapsed_sec": 47.64118039095774, "train_sample_count": 66, "val_budget_mae": 0.43580636382102966, "val_budget_mse": 0.30765682458877563, "val_loss": 2.1857314109802246, "val_sample_count": 11, "val_state_acc": 0.1818181872367859, "val_state_ce": 1.8780745267868042}
{"budget_mae": 0.18586303293704987, "budget_mse": 0.05965156480669975, "gradient_norm": 1.120399832725525, "loss": 1.0208079814910889, "state_acc": 0.6538461446762085, "state_ce": 0.9611563682556152, "step": 70, "step_elapsed_sec": 0.002998574171215296, "total_elapsed_sec": 47.66426231805235, "train_sample_count": 66, "val_budget_mae": 0.45015397667884827, "val_budget_mse": 0.32655179500579834, "val_loss": 2.266745090484619, "val_sample_count": 11, "val_state_acc": 0.1818181872367859, "val_state_ce": 1.9401931762695312}
{"budget_mae": 0.15157262980937958, "budget_mse": 0.04360909387469292, "gradient_norm": 0.9075548648834229, "loss": 0.8589676022529602, "state_acc": 0.692307710647583, "state_ce": 0.8153585195541382, "step": 80, "step_elapsed_sec": 0.003009183332324028, "total_elapsed_sec": 47.6865761410445, "train_sample_count": 66, "val_budget_mae": 0.4578384757041931, "val_budget_mse": 0.33956509828567505, "val_loss": 2.3527088165283203, "val_sample_count": 11, "val_state_acc": 0.1818181872367859, "val_state_ce": 2.01314377784729}
{"budget_mae": 0.13783852756023407, "budget_mse": 0.03725862503051758, "gradient_norm": 1.176679253578186, "loss": 0.7258786559104919, "state_acc": 0.692307710647583, "state_ce": 0.6886200308799744, "step": 90, "step_elapsed_sec": 0.0029762843623757362, "total_elapsed_sec": 47.7089786240831, "train_sample_count": 66, "val_budget_mae": 0.45566239953041077, "val_budget_mse": 0.34146469831466675, "val_loss": 2.4131171703338623, "val_sample_count": 11, "val_state_acc": 0.1818181872367859, "val_state_ce": 2.071652412414551}
{"budget_mae": 0.12898240983486176, "budget_mse": 0.0319182425737381, "gradient_norm": 0.6047582626342773, "loss": 0.5865507125854492, "state_acc": 0.7692307829856873, "state_ce": 0.5546324849128723, "step": 100, "step_elapsed_sec": 0.0030416841618716717, "total_elapsed_sec": 47.73202531179413, "train_sample_count": 66, "val_budget_mae": 0.4516168236732483, "val_budget_mse": 0.340259313583374, "val_loss": 2.462308168411255, "val_sample_count": 11, "val_state_acc": 0.1818181872367859, "val_state_ce": 2.122048854827881}
{"budget_mae": 0.11717724055051804, "budget_mse": 0.029132496565580368, "gradient_norm": 0.7285044193267822, "loss": 0.49436068534851074, "state_acc": 0.8461538553237915, "state_ce": 0.46522819995880127, "step": 110, "step_elapsed_sec": 0.003010762855410576, "total_elapsed_sec": 47.7559005911462, "train_sample_count": 66, "val_budget_mae": 0.44337695837020874, "val_budget_mse": 0.33511587977409363, "val_loss": 2.5064568519592285, "val_sample_count": 11, "val_state_acc": 0.1818181872367859, "val_state_ce": 2.1713409423828125}
{"budget_mae": 0.10478130728006363, "budget_mse": 0.026042824611067772, "gradient_norm": 1.1421325206756592, "loss": 0.41387948393821716, "state_acc": 0.9230769276618958, "state_ce": 0.38783666491508484, "step": 120, "step_elapsed_sec": 0.002973784226924181, "total_elapsed_sec": 47.77843944309279, "train_sample_count": 66, "val_budget_mae": 0.43447285890579224, "val_budget_mse": 0.3266820013523102, "val_loss": 2.547940969467163, "val_sample_count": 11, "val_state_acc": 0.1818181872367859, "val_state_ce": 2.2212588787078857}
{"budget_mae": 0.09833146631717682, "budget_mse": 0.02315005473792553, "gradient_norm": 0.4625439941883087, "loss": 0.35223808884620667, "state_acc": 0.9615384340286255, "state_ce": 0.329088032245636, "step": 130, "step_elapsed_sec": 0.003010203130543232, "total_elapsed_sec": 47.80173162790015, "train_sample_count": 66, "val_budget_mae": 0.42831993103027344, "val_budget_mse": 0.3214476406574249, "val_loss": 2.5909781455993652, "val_sample_count": 11, "val_state_acc": 0.1818181872367859, "val_state_ce": 2.2695305347442627}
{"budget_mae": 0.09005923569202423, "budget_mse": 0.020270096138119698, "gradient_norm": 0.8840328454971313, "loss": 0.30223727226257324, "state_acc": 0.9615384340286255, "state_ce": 0.2819671630859375, "step": 140, "step_elapsed_sec": 0.0029921340756118298, "total_elapsed_sec": 47.82456156797707, "train_sample_count": 66, "val_budget_mae": 0.4246237874031067, "val_budget_mse": 0.319164901971817, "val_loss": 2.630950689315796, "val_sample_count": 11, "val_state_acc": 0.1818181872367859, "val_state_ce": 2.3117856979370117}
{"budget_mae": 0.09021179378032684, "budget_mse": 0.02272844687104225, "gradient_norm": 1.5829627513885498, "loss": 0.3100954294204712, "state_acc": 0.9230769276618958, "state_ce": 0.28736698627471924, "step": 150, "step_elapsed_sec": 0.003059742972254753, "total_elapsed_sec": 47.84768993407488, "train_sample_count": 66, "val_budget_mae": 0.4238142669200897, "val_budget_mse": 0.31980961561203003, "val_loss": 2.688524007797241, "val_sample_count": 11, "val_state_acc": 0.1818181872367859, "val_state_ce": 2.3687143325805664}
{"budget_mae": 0.08204175531864166, "budget_mse": 0.018640633672475815, "gradient_norm": 0.8156708478927612, "loss": 0.2800027132034302, "state_acc": 0.9230769276618958, "state_ce": 0.26136207580566406, "step": 160, "step_elapsed_sec": 0.0030798232182860374, "total_elapsed_sec": 47.87083057081327, "train_sample_count": 66, "val_budget_mae": 0.4202950894832611, "val_budget_mse": 0.3164873719215393, "val_loss": 2.730607509613037, "val_sample_count": 11, "val_state_acc": 0.1818181872367859, "val_state_ce": 2.4141201972961426}
{"budget_mae": 0.07463950663805008, "budget_mse": 0.016610227525234222, "gradient_norm": 0.9519374370574951, "loss": 0.20072108507156372, "state_acc": 0.9615384340286255, "state_ce": 0.1841108500957489, "step": 170, "step_elapsed_sec": 0.0030513941310346127, "total_elapsed_sec": 47.89380058879033, "train_sample_count": 66, "val_budget_mae": 0.41928884387016296, "val_budget_mse": 0.31602025032043457, "val_loss": 2.756617307662964, "val_sample_count": 11, "val_state_acc": 0.1818181872367859, "val_state_ce": 2.4405970573425293}
{"budget_mae": 0.07407543063163757, "budget_mse": 0.016475534066557884, "gradient_norm": 1.0333036184310913, "loss": 0.16613025963306427, "state_acc": 1.0, "state_ce": 0.14965473115444183, "step": 180, "step_elapsed_sec": 0.0030206041410565376, "total_elapsed_sec": 47.917197942733765, "train_sample_count": 66, "val_budget_mae": 0.4211028814315796, "val_budget_mse": 0.3204299509525299, "val_loss": 2.79980206489563, "val_sample_count": 11, "val_state_acc": 0.1818181872367859, "val_state_ce": 2.479372024536133}
{"budget_mae": 0.07336845248937607, "budget_mse": 0.01645343191921711, "gradient_norm": 0.7130651473999023, "loss": 0.17002946138381958, "state_acc": 1.0, "state_ce": 0.15357603132724762, "step": 190, "step_elapsed_sec": 0.0030532130040228367, "total_elapsed_sec": 47.94031562888995, "train_sample_count": 66, "val_budget_mae": 0.419605553150177, "val_budget_mse": 0.31814485788345337, "val_loss": 2.8513081073760986, "val_sample_count": 11, "val_state_acc": 0.1818181872367859, "val_state_ce": 2.53316330909729}
{"budget_mae": 0.072269968688488, "budget_mse": 0.016645926982164383, "gradient_norm": 0.996533215045929, "loss": 0.15895600616931915, "state_acc": 0.9615384340286255, "state_ce": 0.14231008291244507, "step": 200, "step_elapsed_sec": 0.00309173297137022, "total_elapsed_sec": 47.963351977057755, "train_sample_count": 66, "val_budget_mae": 0.41442999243736267, "val_budget_mse": 0.3108381927013397, "val_loss": 2.8909666538238525, "val_sample_count": 11, "val_state_acc": 0.1818181872367859, "val_state_ce": 2.5801284313201904}
{"budget_mae": 0.06967762112617493, "budget_mse": 0.015462537296116352, "gradient_norm": 1.3049638271331787, "loss": 0.13064314424991608, "state_acc": 0.9615384340286255, "state_ce": 0.115180604159832, "step": 210, "step_elapsed_sec": 0.0030745631083846092, "total_elapsed_sec": 47.98742601508275, "train_sample_count": 66, "val_budget_mae": 0.41286665201187134, "val_budget_mse": 0.3122458755970001, "val_loss": 2.9282355308532715, "val_sample_count": 11, "val_state_acc": 0.1818181872367859, "val_state_ce": 2.6159896850585938}
{"budget_mae": 0.0718250423669815, "budget_mse": 0.014970225282013416, "gradient_norm": 0.546826958656311, "loss": 0.10183162987232208, "state_acc": 1.0, "state_ce": 0.08686140179634094, "step": 220, "step_elapsed_sec": 0.0030293227173388004, "total_elapsed_sec": 48.01062946114689, "train_sample_count": 66, "val_budget_mae": 0.40866342186927795, "val_budget_mse": 0.31039291620254517, "val_loss": 2.9542245864868164, "val_sample_count": 11, "val_state_acc": 0.1818181872367859, "val_state_ce": 2.643831729888916}
{"budget_mae": 0.06625562161207199, "budget_mse": 0.015259500592947006, "gradient_norm": 0.6374114155769348, "loss": 0.09717357158660889, "state_acc": 1.0, "state_ce": 0.08191407471895218, "step": 230, "step_elapsed_sec": 0.003024884033948183, "total_elapsed_sec": 48.03409460512921, "train_sample_count": 66, "val_budget_mae": 0.40297383069992065, "val_budget_mse": 0.30067428946495056, "val_loss": 2.994717597961426, "val_sample_count": 11, "val_state_acc": 0.1818181872367859, "val_state_ce": 2.6940433979034424}
{"budget_mae": 0.0651903823018074, "budget_mse": 0.013966568745672703, "gradient_norm": 1.0323323011398315, "loss": 0.09623391181230545, "state_acc": 0.9615384340286255, "state_ce": 0.08226734399795532, "step": 240, "step_elapsed_sec": 0.0030424930155277252, "total_elapsed_sec": 48.05718545103446, "train_sample_count": 66, "val_budget_mae": 0.4034954607486725, "val_budget_mse": 0.30172598361968994, "val_loss": 3.045167922973633, "val_sample_count": 11, "val_state_acc": 0.1818181872367859, "val_state_ce": 2.7434420585632324}
{"budget_mae": 0.06353217363357544, "budget_mse": 0.012686487287282944, "gradient_norm": 0.4719961881637573, "loss": 0.06827211380004883, "state_acc": 1.0, "state_ce": 0.05558563023805618, "step": 250, "step_elapsed_sec": 0.0034290500916540623, "total_elapsed_sec": 48.085814259946346, "train_sample_count": 66, "val_budget_mae": 0.39377880096435547, "val_budget_mse": 0.29063060879707336, "val_loss": 3.0516977310180664, "val_sample_count": 11, "val_state_acc": 0.1818181872367859, "val_state_ce": 2.7610671520233154}
{"budget_mae": 0.06310553103685379, "budget_mse": 0.01321922242641449, "gradient_norm": 0.85733962059021, "loss": 0.07316455245018005, "state_acc": 1.0, "state_ce": 0.059945330023765564, "step": 260, "step_elapsed_sec": 0.003346070647239685, "total_elapsed_sec": 48.11282736202702, "train_sample_count": 66, "val_budget_mae": 0.39169782400131226, "val_budget_mse": 0.2890923321247101, "val_loss": 3.104436159133911, "val_sample_count": 11, "val_state_acc": 0.1818181872367859, "val_state_ce": 2.8153438568115234}
{"budget_mae": 0.05738447979092598, "budget_mse": 0.012219946831464767, "gradient_norm": 0.3324139714241028, "loss": 0.06523439288139343, "state_acc": 1.0, "state_ce": 0.053014446049928665, "step": 270, "step_elapsed_sec": 0.0029671438969671726, "total_elapsed_sec": 48.13822168903425, "train_sample_count": 66, "val_budget_mae": 0.39063066244125366, "val_budget_mse": 0.28674811124801636, "val_loss": 3.1403372287750244, "val_sample_count": 11, "val_state_acc": 0.1818181872367859, "val_state_ce": 2.8535890579223633}
{"budget_mae": 0.06036568433046341, "budget_mse": 0.012317025102674961, "gradient_norm": 1.1827818155288696, "loss": 0.07630619406700134, "state_acc": 0.9615384340286255, "state_ce": 0.06398916989564896, "step": 280, "step_elapsed_sec": 0.0033760610967874527, "total_elapsed_sec": 48.16370562510565, "train_sample_count": 66, "val_budget_mae": 0.38578876852989197, "val_budget_mse": 0.28587883710861206, "val_loss": 3.200936794281006, "val_sample_count": 11, "val_state_acc": 0.27272728085517883, "val_state_ce": 2.915057897567749}
{"budget_mae": 0.05474644899368286, "budget_mse": 0.011494267731904984, "gradient_norm": 0.28931090235710144, "loss": 0.049986228346824646, "state_acc": 1.0, "state_ce": 0.03849196061491966, "step": 290, "step_elapsed_sec": 0.0052551343105733395, "total_elapsed_sec": 48.20559961581603, "train_sample_count": 66, "val_budget_mae": 0.38459452986717224, "val_budget_mse": 0.2842617630958557, "val_loss": 3.2308340072631836, "val_sample_count": 11, "val_state_acc": 0.27272728085517883, "val_state_ce": 2.9465723037719727}
{"budget_mae": 0.05405930429697037, "budget_mse": 0.011000405065715313, "gradient_norm": 0.2669156789779663, "loss": 0.04073794186115265, "state_acc": 1.0, "state_ce": 0.02973753772675991, "step": 300, "step_elapsed_sec": 0.0030171829275786877, "total_elapsed_sec": 48.23737612599507, "train_sample_count": 66, "val_budget_mae": 0.38401249051094055, "val_budget_mse": 0.28252968192100525, "val_loss": 3.2696735858917236, "val_sample_count": 11, "val_state_acc": 0.27272728085517883, "val_state_ce": 2.9871439933776855}
{"budget_mae": 0.05258719623088837, "budget_mse": 0.010654639452695847, "gradient_norm": 0.5153996348381042, "loss": 0.0381925031542778, "state_acc": 1.0, "state_ce": 0.027537861838936806, "step": 310, "step_elapsed_sec": 0.003822856117039919, "total_elapsed_sec": 48.26241883588955, "train_sample_count": 66, "val_budget_mae": 0.37977880239486694, "val_budget_mse": 0.27715522050857544, "val_loss": 3.3186728954315186, "val_sample_count": 11, "val_state_acc": 0.27272728085517883, "val_state_ce": 3.041517734527588}
{"budget_mae": 0.05259399488568306, "budget_mse": 0.0100801857188344, "gradient_norm": 0.22333082556724548, "loss": 0.03557455167174339, "state_acc": 1.0, "state_ce": 0.025494365021586418, "step": 320, "step_elapsed_sec": 0.0031280224211513996, "total_elapsed_sec": 48.287732942961156, "train_sample_count": 66, "val_budget_mae": 0.3789427876472473, "val_budget_mse": 0.2781027555465698, "val_loss": 3.3775248527526855, "val_sample_count": 11, "val_state_acc": 0.27272728085517883, "val_state_ce": 3.099421977996826}
{"budget_mae": 0.04890512675046921, "budget_mse": 0.010186421684920788, "gradient_norm": 0.2906658351421356, "loss": 0.032129645347595215, "state_acc": 1.0, "state_ce": 0.021943224593997, "step": 330, "step_elapsed_sec": 0.0036324081011116505, "total_elapsed_sec": 48.31588646583259, "train_sample_count": 66, "val_budget_mae": 0.3750700354576111, "val_budget_mse": 0.27351096272468567, "val_loss": 3.4403998851776123, "val_sample_count": 11, "val_state_acc": 0.27272728085517883, "val_state_ce": 3.166888952255249}
{"budget_mae": 0.052361782640218735, "budget_mse": 0.010320654138922691, "gradient_norm": 0.22304773330688477, "loss": 0.033323708921670914, "state_acc": 1.0, "state_ce": 0.023003054782748222, "step": 340, "step_elapsed_sec": 0.0033972500823438168, "total_elapsed_sec": 48.34479501191527, "train_sample_count": 66, "val_budget_mae": 0.37700769305229187, "val_budget_mse": 0.2740420699119568, "val_loss": 3.5121781826019287, "val_sample_count": 11, "val_state_acc": 0.27272728085517883, "val_state_ce": 3.238136053085327}
{"budget_mae": 0.047340694814920425, "budget_mse": 0.009351963177323341, "gradient_norm": 0.3183414340019226, "loss": 0.02503371797502041, "state_acc": 1.0, "state_ce": 0.015681754797697067, "step": 350, "step_elapsed_sec": 0.0031888517551124096, "total_elapsed_sec": 48.372732125688344, "train_sample_count": 66, "val_budget_mae": 0.3720158040523529, "val_budget_mse": 0.2697282135486603, "val_loss": 3.5461790561676025, "val_sample_count": 11, "val_state_acc": 0.27272728085517883, "val_state_ce": 3.2764508724212646}
{"budget_mae": 0.04946057125926018, "budget_mse": 0.009459968656301498, "gradient_norm": 0.48625218868255615, "loss": 0.024831600487232208, "state_acc": 1.0, "state_ce": 0.01537163183093071, "step": 360, "step_elapsed_sec": 0.003340871073305607, "total_elapsed_sec": 48.400690329726785, "train_sample_count": 66, "val_budget_mae": 0.37368521094322205, "val_budget_mse": 0.26943767070770264, "val_loss": 3.616868019104004, "val_sample_count": 11, "val_state_acc": 0.27272728085517883, "val_state_ce": 3.3474302291870117}
{"budget_mae": 0.0480189248919487, "budget_mse": 0.009176241233944893, "gradient_norm": 0.22478654980659485, "loss": 0.020406823605298996, "state_acc": 1.0, "state_ce": 0.011230581440031528, "step": 370, "step_elapsed_sec": 0.0032586511224508286, "total_elapsed_sec": 48.4270740179345, "train_sample_count": 66, "val_budget_mae": 0.37254905700683594, "val_budget_mse": 0.2697407901287079, "val_loss": 3.6926934719085693, "val_sample_count": 11, "val_state_acc": 0.27272728085517883, "val_state_ce": 3.422952651977539}
{"budget_mae": 0.043373677879571915, "budget_mse": 0.008703899569809437, "gradient_norm": 0.14052993059158325, "loss": 0.01959844119846821, "state_acc": 1.0, "state_ce": 0.010894541628658772, "step": 380, "step_elapsed_sec": 0.0031366320326924324, "total_elapsed_sec": 48.4519257186912, "train_sample_count": 66, "val_budget_mae": 0.37116900086402893, "val_budget_mse": 0.2666594982147217, "val_loss": 3.757439374923706, "val_sample_count": 11, "val_state_acc": 0.27272728085517883, "val_state_ce": 3.4907798767089844}
{"budget_mae": 0.043208133429288864, "budget_mse": 0.008179049007594585, "gradient_norm": 0.25986576080322266, "loss": 0.018892109394073486, "state_acc": 1.0, "state_ce": 0.010713059455156326, "step": 390, "step_elapsed_sec": 0.003135331906378269, "total_elapsed_sec": 48.47589112771675, "train_sample_count": 66, "val_budget_mae": 0.37149131298065186, "val_budget_mse": 0.26489728689193726, "val_loss": 3.804652690887451, "val_sample_count": 11, "val_state_acc": 0.27272728085517883, "val_state_ce": 3.539755344390869}
{"budget_mae": 0.04245690628886223, "budget_mse": 0.00787554681301117, "gradient_norm": 0.14593514800071716, "loss": 0.014589671045541763, "state_acc": 1.0, "state_ce": 0.006714124232530594, "step": 400, "step_elapsed_sec": 0.0032866611145436764, "total_elapsed_sec": 48.501134955789894, "train_sample_count": 66, "val_budget_mae": 0.3715783953666687, "val_budget_mse": 0.2655692994594574, "val_loss": 3.848928451538086, "val_sample_count": 11, "val_state_acc": 0.27272728085517883, "val_state_ce": 3.5833592414855957}
{"budget_mae": 0.04256272688508034, "budget_mse": 0.0072626941837370396, "gradient_norm": 0.22804950177669525, "loss": 0.016588516533374786, "state_acc": 1.0, "state_ce": 0.009325822815299034, "step": 410, "step_elapsed_sec": 0.003069682978093624, "total_elapsed_sec": 48.52596798771992, "train_sample_count": 66, "val_budget_mae": 0.3724565804004669, "val_budget_mse": 0.2648444175720215, "val_loss": 3.944016456604004, "val_sample_count": 11, "val_state_acc": 0.27272728085517883, "val_state_ce": 3.6791720390319824}
{"budget_mae": 0.04551519826054573, "budget_mse": 0.007620672695338726, "gradient_norm": 0.7166860699653625, "loss": 0.012833717279136181, "state_acc": 1.0, "state_ce": 0.005213044583797455, "step": 420, "step_elapsed_sec": 0.0035156398080289364, "total_elapsed_sec": 48.55194361973554, "train_sample_count": 66, "val_budget_mae": 0.37314262986183167, "val_budget_mse": 0.2653397023677826, "val_loss": 3.981052875518799, "val_sample_count": 11, "val_state_acc": 0.27272728085517883, "val_state_ce": 3.7157132625579834}
{"budget_mae": 0.041333071887493134, "budget_mse": 0.007494995836168528, "gradient_norm": 0.3296487629413605, "loss": 0.014655902981758118, "state_acc": 1.0, "state_ce": 0.007160907611250877, "step": 430, "step_elapsed_sec": 0.0031222919933497906, "total_elapsed_sec": 48.5780035299249, "train_sample_count": 66, "val_budget_mae": 0.37128227949142456, "val_budget_mse": 0.2636193335056305, "val_loss": 4.058364391326904, "val_sample_count": 11, "val_state_acc": 0.27272728085517883, "val_state_ce": 3.7947452068328857}
{"budget_mae": 0.03957062214612961, "budget_mse": 0.0070515903644263744, "gradient_norm": 0.1393749862909317, "loss": 0.012180164456367493, "state_acc": 1.0, "state_ce": 0.0051285745576024055, "step": 440, "step_elapsed_sec": 0.005184913985431194, "total_elapsed_sec": 48.62042381707579, "train_sample_count": 66, "val_budget_mae": 0.3716779351234436, "val_budget_mse": 0.26160359382629395, "val_loss": 4.113404750823975, "val_sample_count": 11, "val_state_acc": 0.27272728085517883, "val_state_ce": 3.8518011569976807}
{"budget_mae": 0.045468349009752274, "budget_mse": 0.0072785778902471066, "gradient_norm": 0.32421955466270447, "loss": 0.01084614172577858, "state_acc": 1.0, "state_ce": 0.0035675636027008295, "step": 450, "step_elapsed_sec": 0.005312493070960045, "total_elapsed_sec": 48.660929910838604, "train_sample_count": 66, "val_budget_mae": 0.37237557768821716, "val_budget_mse": 0.2625282406806946, "val_loss": 4.176968574523926, "val_sample_count": 11, "val_state_acc": 0.27272728085517883, "val_state_ce": 3.914440155029297}
{"budget_mae": 0.03960411250591278, "budget_mse": 0.0070035080425441265, "gradient_norm": 0.18624837696552277, "loss": 0.010252727195620537, "state_acc": 1.0, "state_ce": 0.0032492196187376976, "step": 460, "step_elapsed_sec": 0.0031293323263525963, "total_elapsed_sec": 48.687954483088106, "train_sample_count": 66, "val_budget_mae": 0.36996152997016907, "val_budget_mse": 0.26039305329322815, "val_loss": 4.242122650146484, "val_sample_count": 11, "val_state_acc": 0.27272728085517883, "val_state_ce": 3.981729745864868}
{"budget_mae": 0.04102334752678871, "budget_mse": 0.006218007765710354, "gradient_norm": 0.30382242798805237, "loss": 0.009852125309407711, "state_acc": 1.0, "state_ce": 0.003634117776528001, "step": 470, "step_elapsed_sec": 0.0031097927130758762, "total_elapsed_sec": 48.713133892044425, "train_sample_count": 66, "val_budget_mae": 0.37262284755706787, "val_budget_mse": 0.2601635158061981, "val_loss": 4.290213108062744, "val_sample_count": 11, "val_state_acc": 0.27272728085517883, "val_state_ce": 4.030049800872803}
{"budget_mae": 0.03605636954307556, "budget_mse": 0.006120773032307625, "gradient_norm": 0.1336098313331604, "loss": 0.008830911479890347, "state_acc": 1.0, "state_ce": 0.002710138214752078, "step": 480, "step_elapsed_sec": 0.003052772954106331, "total_elapsed_sec": 48.74073546892032, "train_sample_count": 66, "val_budget_mae": 0.3707212209701538, "val_budget_mse": 0.25900423526763916, "val_loss": 4.343898296356201, "val_sample_count": 11, "val_state_acc": 0.27272728085517883, "val_state_ce": 4.084894180297852}
{"budget_mae": 0.038101211190223694, "budget_mse": 0.006209174171090126, "gradient_norm": 0.17512889206409454, "loss": 0.008348844945430756, "state_acc": 1.0, "state_ce": 0.0021396707743406296, "step": 490, "step_elapsed_sec": 0.003031332977116108, "total_elapsed_sec": 48.76400463376194, "train_sample_count": 66, "val_budget_mae": 0.375817209482193, "val_budget_mse": 0.2640281915664673, "val_loss": 4.432578086853027, "val_sample_count": 11, "val_state_acc": 0.27272728085517883, "val_state_ce": 4.16855001449585}
{"budget_mae": 0.03770504146814346, "budget_mse": 0.005806234665215015, "gradient_norm": 0.21749278903007507, "loss": 0.00822984054684639, "state_acc": 1.0, "state_ce": 0.002423605415970087, "step": 500, "step_elapsed_sec": 0.0031720218248665333, "total_elapsed_sec": 48.78971529798582, "train_sample_count": 66, "val_budget_mae": 0.3736957609653473, "val_budget_mse": 0.2621137797832489, "val_loss": 4.499844074249268, "val_sample_count": 11, "val_state_acc": 0.27272728085517883, "val_state_ce": 4.237730503082275}
