{"t": 1790410587.0015318, "kind": "start", "update": 0, "max_updates": 8000, "config": "configs/lora_10h.yaml"} {"t": 1790410777.7644396, "kind": "train", "update": 1, "max_updates": 8000, "epoch": 0, "loss": 0.8147109150886536, "grad_norm": 0.03254738077521324, "lr": 2.5000000000000004e-07, "sec_per_update": 190.76245737075806, "speed": 1.6692633531795942, "vram_gb": 19.837255477905273} {"t": 1790411115.9362626, "kind": "train", "update": 25, "max_updates": 8000, "epoch": 0, "loss": 0.6900548040866852, "grad_norm": 0.04316746070981026, "lr": 3.2500000000000002e-06, "sec_per_update": 14.090316772460938, "speed": 13.711149313110907, "vram_gb": 24.763202667236328} {"t": 1790411125.5531483, "kind": "train", "update": 50, "max_updates": 8000, "epoch": 0, "loss": 0.6957867789268494, "grad_norm": 0.04564041644334793, "lr": 6.375000000000001e-06, "sec_per_update": 0.3848441982269287, "speed": 27.061381168050065, "vram_gb": 25.51632595062256} {"t": 1790411133.3164616, "kind": "train", "update": 75, "max_updates": 8000, "epoch": 0, "loss": 0.712929904460907, "grad_norm": 0.07226406782865524, "lr": 9.5e-06, "sec_per_update": 0.3105123233795166, "speed": 39.42442562790116, "vram_gb": 25.51632595062256} {"t": 1790411143.105903, "kind": "train", "update": 100, "max_updates": 8000, "epoch": 0, "loss": 0.6927954936027527, "grad_norm": 0.031036319211125374, "lr": 1.2625e-05, "sec_per_update": 0.3915961456298828, "speed": 52.991891661303676, "vram_gb": 25.51632595062256} {"t": 1790411153.647305, "kind": "train", "update": 125, "max_updates": 8000, "epoch": 1, "loss": 0.6638682508468627, "grad_norm": 0.0399162732064724, "lr": 1.575e-05, "sec_per_update": 0.4216579723358154, "speed": 65.24358255885112, "vram_gb": 25.51632595062256} {"t": 1790411162.951946, "kind": "train", "update": 150, "max_updates": 8000, "epoch": 1, "loss": 0.6627382373809815, "grad_norm": 0.0956173986196518, "lr": 1.8875e-05, "sec_per_update": 0.3720043468475342, "speed": 77.08104102321728, "vram_gb": 25.519293308258057} {"t": 1790411171.1445417, "kind": "train", "update": 175, "max_updates": 8000, "epoch": 1, "loss": 0.6384814608097077, "grad_norm": 0.06326854228973389, "lr": 2.2000000000000003e-05, "sec_per_update": 0.32788442611694335, "speed": 87.77931187477978, "vram_gb": 25.519293308258057} {"t": 1790411180.3593836, "kind": "train", "update": 200, "max_updates": 8000, "epoch": 1, "loss": 0.6590252161026001, "grad_norm": 0.04942508786916733, "lr": 2.5124999999999997e-05, "sec_per_update": 0.3685938358306885, "speed": 98.80565829292175, "vram_gb": 25.519293308258057} {"t": 1790411189.5941608, "kind": "train", "update": 225, "max_updates": 8000, "epoch": 1, "loss": 0.6332191872596741, "grad_norm": 0.03064320981502533, "lr": 2.825e-05, "sec_per_update": 0.3693943405151367, "speed": 109.52877997531571, "vram_gb": 25.519293308258057} {"t": 1790411199.0784101, "kind": "train", "update": 250, "max_updates": 8000, "epoch": 2, "loss": 0.6932531356811523, "grad_norm": 0.05592277646064758, "lr": 3.1375e-05, "sec_per_update": 0.37918968200683595, "speed": 120.25776062783828, "vram_gb": 25.519293308258057} {"t": 1790411207.5265272, "kind": "train", "update": 275, "max_updates": 8000, "epoch": 2, "loss": 0.645858952999115, "grad_norm": 0.020656878128647804, "lr": 3.45e-05, "sec_per_update": 0.3381047821044922, "speed": 129.69961418875351, "vram_gb": 25.519293308258057} {"t": 1790411216.3903213, "kind": "train", "update": 300, "max_updates": 8000, "epoch": 2, "loss": 0.6447462773323059, "grad_norm": 0.016667921096086502, "lr": 3.7625e-05, "sec_per_update": 0.354534330368042, "speed": 139.1427656908657, "vram_gb": 25.519293308258057} {"t": 1790411225.4823885, "kind": "train", "update": 325, "max_updates": 8000, "epoch": 2, "loss": 0.6399024057388306, "grad_norm": 0.03350725769996643, "lr": 4.075e-05, "sec_per_update": 0.3636774158477783, "speed": 149.02450717911697, "vram_gb": 25.519293308258057} {"t": 1790411235.3522267, "kind": "train", "update": 350, "max_updates": 8000, "epoch": 3, "loss": 0.6258823323249817, "grad_norm": 0.07967661321163177, "lr": 4.3875e-05, "sec_per_update": 0.394813814163208, "speed": 158.19636219956243, "vram_gb": 25.519293308258057} {"t": 1790411244.7756026, "kind": "train", "update": 375, "max_updates": 8000, "epoch": 3, "loss": 0.6540462350845337, "grad_norm": 0.06419770419597626, "lr": 4.7e-05, "sec_per_update": 0.37687871932983397, "speed": 167.7353816623255, "vram_gb": 25.519293308258057} {"t": 1790411253.9853234, "kind": "train", "update": 400, "max_updates": 8000, "epoch": 3, "loss": 0.6355213642120361, "grad_norm": 0.0332409106194973, "lr": 5.0125e-05, "sec_per_update": 0.3684427261352539, "speed": 176.68135064727403, "vram_gb": 25.519293308258057} {"t": 1790411263.7139518, "kind": "train", "update": 425, "max_updates": 8000, "epoch": 3, "loss": 0.6056721210479736, "grad_norm": 0.08861125260591507, "lr": 5.325e-05, "sec_per_update": 0.38912190437316896, "speed": 185.97065420266273, "vram_gb": 25.519293308258057} {"t": 1790411272.0173247, "kind": "train", "update": 450, "max_updates": 8000, "epoch": 3, "loss": 0.6364820170402526, "grad_norm": 0.08056347072124481, "lr": 5.6375e-05, "sec_per_update": 0.33215970039367676, "speed": 193.75607126328643, "vram_gb": 25.519293308258057} {"t": 1790411281.5454483, "kind": "train", "update": 475, "max_updates": 8000, "epoch": 4, "loss": 0.6128824663162231, "grad_norm": 0.05432860925793648, "lr": 5.95e-05, "sec_per_update": 0.38112611770629884, "speed": 200.7485901551532, "vram_gb": 25.519293308258057} {"t": 1790411289.9681623, "kind": "train", "update": 500, "max_updates": 8000, "epoch": 4, "loss": 0.6355342483520507, "grad_norm": 0.04555418714880943, "lr": 6.2625e-05, "sec_per_update": 0.33690221786499025, "speed": 208.8278251441527, "vram_gb": 25.519293308258057} {"t": 1790411590.363129, "kind": "val", "update": 500, "val_loss": 0.6770535508791605} {"t": 1790411599.1309488, "kind": "train", "update": 525, "max_updates": 8000, "epoch": 4, "loss": 0.6438445544242859, "grad_norm": 0.07761651277542114, "lr": 6.575e-05, "sec_per_update": 12.366520586013793, "speed": 152.39446829583864, "vram_gb": 25.519293308258057} {"t": 1790411608.029572, "kind": "train", "update": 550, "max_updates": 8000, "epoch": 4, "loss": 0.6367286634445191, "grad_norm": 0.05241548269987106, "lr": 6.887500000000001e-05, "sec_per_update": 0.35594823837280276, "speed": 158.15257602423753, "vram_gb": 25.519293308258057} {"t": 1790411616.9261878, "kind": "train", "update": 575, "max_updates": 8000, "epoch": 4, "loss": 0.6073742151260376, "grad_norm": 0.04746076464653015, "lr": 7.2e-05, "sec_per_update": 0.35585577964782716, "speed": 163.63521877998957, "vram_gb": 25.519293308258057} {"t": 1790411626.084809, "kind": "train", "update": 600, "max_updates": 8000, "epoch": 5, "loss": 0.6413608241081238, "grad_norm": 0.0875387191772461, "lr": 7.5125e-05, "sec_per_update": 0.3663290309906006, "speed": 169.19447348271243, "vram_gb": 25.519293308258057} {"t": 1790411635.3083029, "kind": "train", "update": 625, "max_updates": 8000, "epoch": 5, "loss": 0.6233300113677979, "grad_norm": 0.07536232471466064, "lr": 7.825e-05, "sec_per_update": 0.3689345645904541, "speed": 174.9338086301633, "vram_gb": 25.52091121673584} {"t": 1790411644.508861, "kind": "train", "update": 650, "max_updates": 8000, "epoch": 5, "loss": 0.6274115920066834, "grad_norm": 0.077048659324646, "lr": 8.1375e-05, "sec_per_update": 0.368048152923584, "speed": 180.41777237893257, "vram_gb": 25.52091121673584} {"t": 1790411653.4170122, "kind": "train", "update": 675, "max_updates": 8000, "epoch": 5, "loss": 0.6380851459503174, "grad_norm": 0.03976396843791008, "lr": 8.450000000000001e-05, "sec_per_update": 0.35632102966308593, "speed": 185.90365645861237, "vram_gb": 25.52091121673584} {"t": 1790411662.797197, "kind": "train", "update": 700, "max_updates": 8000, "epoch": 6, "loss": 0.6219055938720703, "grad_norm": 0.10034383088350296, "lr": 8.7625e-05, "sec_per_update": 0.37520891189575195, "speed": 190.59712452065548, "vram_gb": 25.52091121673584} {"t": 1790411672.494654, "kind": "train", "update": 725, "max_updates": 8000, "epoch": 6, "loss": 0.6337138676643371, "grad_norm": 0.07603931427001953, "lr": 9.075e-05, "sec_per_update": 0.3877535152435303, "speed": 196.0285321320122, "vram_gb": 25.52091121673584} {"t": 1790411681.5036964, "kind": "train", "update": 750, "max_updates": 8000, "epoch": 6, "loss": 0.6037895154953002, "grad_norm": 0.031188853085041046, "lr": 9.3875e-05, "sec_per_update": 0.3605054759979248, "speed": 201.4493323396476, "vram_gb": 25.52091121673584} {"t": 1790411689.8566349, "kind": "train", "update": 775, "max_updates": 8000, "epoch": 6, "loss": 0.604896445274353, "grad_norm": 0.13935531675815582, "lr": 9.7e-05, "sec_per_update": 0.3341220855712891, "speed": 205.80037334735636, "vram_gb": 25.52091121673584} {"t": 1790411698.8673675, "kind": "train", "update": 800, "max_updates": 8000, "epoch": 6, "loss": 0.6195506846904755, "grad_norm": 0.05091392621397972, "lr": 0.0001, "sec_per_update": 0.36042964935302735, "speed": 210.5244810027428, "vram_gb": 25.52091121673584} {"t": 1790411708.1514935, "kind": "train", "update": 825, "max_updates": 8000, "epoch": 7, "loss": 0.6131996154785156, "grad_norm": 0.0616089291870594, "lr": 9.965277777777778e-05, "sec_per_update": 0.37119767189025876, "speed": 215.4413811291366, "vram_gb": 25.52091121673584} {"t": 1790411717.0535917, "kind": "train", "update": 850, "max_updates": 8000, "epoch": 7, "loss": 0.6398351454734802, "grad_norm": 0.07702547311782837, "lr": 9.930555555555556e-05, "sec_per_update": 0.35621827125549316, "speed": 219.96719369393367, "vram_gb": 25.52091121673584} {"t": 1790411726.3605187, "kind": "train", "update": 875, "max_updates": 8000, "epoch": 7, "loss": 0.5968553447723388, "grad_norm": 0.22644950449466705, "lr": 9.895833333333334e-05, "sec_per_update": 0.37228453636169434, "speed": 224.78963295114264, "vram_gb": 25.52091121673584} {"t": 1790411735.7909477, "kind": "train", "update": 900, "max_updates": 8000, "epoch": 7, "loss": 0.6087638711929322, "grad_norm": 0.11926290392875671, "lr": 9.861111111111112e-05, "sec_per_update": 0.37724252700805666, "speed": 229.38117956593118, "vram_gb": 25.52091121673584} {"t": 1790411744.6338084, "kind": "train", "update": 925, "max_updates": 8000, "epoch": 7, "loss": 0.6117314183712006, "grad_norm": 0.03765645623207092, "lr": 9.826388888888889e-05, "sec_per_update": 0.3535502624511719, "speed": 234.20640412976744, "vram_gb": 25.52091121673584} {"t": 1790411754.3400764, "kind": "train", "update": 950, "max_updates": 8000, "epoch": 8, "loss": 0.5908283352851867, "grad_norm": 0.030001092702150345, "lr": 9.791666666666667e-05, "sec_per_update": 0.38839470863342285, "speed": 238.2729825408108, "vram_gb": 25.52091121673584} {"t": 1790411762.6801274, "kind": "train", "update": 975, "max_updates": 8000, "epoch": 8, "loss": 0.6187149810791016, "grad_norm": 0.1522456854581833, "lr": 9.756944444444445e-05, "sec_per_update": 0.3336167812347412, "speed": 242.4235524628157, "vram_gb": 25.52091121673584} {"t": 1790411771.759659, "kind": "train", "update": 1000, "max_updates": 8000, "epoch": 8, "loss": 0.5986440348625183, "grad_norm": 0.0952797532081604, "lr": 9.722222222222223e-05, "sec_per_update": 0.3631863784790039, "speed": 246.7691223622964, "vram_gb": 25.52091121673584} {"t": 1790411772.4299948, "kind": "val", "update": 1000, "val_loss": 0.6559972167015076} {"t": 1790411927.1655514, "kind": "checkpoint", "update": 1000} {"t": 1790411936.6465528, "kind": "train", "update": 1025, "max_updates": 8000, "epoch": 8, "loss": 0.6116156220436096, "grad_norm": 0.24003736674785614, "lr": 9.687500000000001e-05, "sec_per_update": 6.595474481582642, "speed": 222.37043823513838, "vram_gb": 25.570744514465332} {"t": 1790411945.8092294, "kind": "train", "update": 1050, "max_updates": 8000, "epoch": 9, "loss": 0.5928326964378356, "grad_norm": 0.04466141015291214, "lr": 9.652777777777779e-05, "sec_per_update": 0.36636914253234865, "speed": 226.45111414436442, "vram_gb": 25.570744514465332} {"t": 1790411954.9364665, "kind": "train", "update": 1075, "max_updates": 8000, "epoch": 9, "loss": 0.5983987665176391, "grad_norm": 0.062056224793195724, "lr": 9.618055555555556e-05, "sec_per_update": 0.3652078914642334, "speed": 230.12369753984635, "vram_gb": 25.570744514465332} {"t": 1790411964.515635, "kind": "train", "update": 1100, "max_updates": 8000, "epoch": 9, "loss": 0.5983208775520324, "grad_norm": 0.09924066066741943, "lr": 9.583333333333334e-05, "sec_per_update": 0.3831838893890381, "speed": 234.26546958284987, "vram_gb": 25.570744514465332} {"t": 1790411973.1600878, "kind": "train", "update": 1125, "max_updates": 8000, "epoch": 9, "loss": 0.6319615292549133, "grad_norm": 0.1527726799249649, "lr": 9.548611111111112e-05, "sec_per_update": 0.34577948570251466, "speed": 237.75905461212542, "vram_gb": 25.570744514465332} {"t": 1790411981.7826915, "kind": "train", "update": 1150, "max_updates": 8000, "epoch": 9, "loss": 0.611964361667633, "grad_norm": 0.03011351265013218, "lr": 9.513888888888888e-05, "sec_per_update": 0.3449059867858887, "speed": 241.46308074107966, "vram_gb": 25.572148323059082} {"t": 1790411991.3039262, "kind": "train", "update": 1175, "max_updates": 8000, "epoch": 10, "loss": 0.6069459342956542, "grad_norm": 0.03489772975444794, "lr": 9.479166666666666e-05, "sec_per_update": 0.38082887649536135, "speed": 245.04558725422754, "vram_gb": 25.572148323059082} {"t": 1790411999.8141282, "kind": "train", "update": 1200, "max_updates": 8000, "epoch": 10, "loss": 0.6160257244110108, "grad_norm": 0.092926025390625, "lr": 9.444444444444444e-05, "sec_per_update": 0.340430850982666, "speed": 248.87571131682878, "vram_gb": 25.572148323059082} {"t": 1790412008.1299186, "kind": "train", "update": 1225, "max_updates": 8000, "epoch": 10, "loss": 0.6080887198448182, "grad_norm": 0.11004412919282913, "lr": 9.409722222222222e-05, "sec_per_update": 0.33262969970703127, "speed": 252.6041226002254, "vram_gb": 25.572148323059082} {"t": 1790412016.5600765, "kind": "train", "update": 1250, "max_updates": 8000, "epoch": 10, "loss": 0.5935288095474243, "grad_norm": 0.0369139201939106, "lr": 9.375e-05, "sec_per_update": 0.3372048854827881, "speed": 256.3102633149319, "vram_gb": 25.572148323059082} {"t": 1790412026.210349, "kind": "train", "update": 1275, "max_updates": 8000, "epoch": 10, "loss": 0.5936650490760803, "grad_norm": 0.06895715743303299, "lr": 9.340277777777778e-05, "sec_per_update": 0.3860089683532715, "speed": 259.5536968357441, "vram_gb": 25.572148323059082} {"t": 1790412034.7176225, "kind": "train", "update": 1300, "max_updates": 8000, "epoch": 11, "loss": 0.6000477361679077, "grad_norm": 0.12470611184835434, "lr": 9.305555555555556e-05, "sec_per_update": 0.34029236793518064, "speed": 262.54116098767275, "vram_gb": 25.572148323059082} {"t": 1790412043.2679162, "kind": "train", "update": 1325, "max_updates": 8000, "epoch": 11, "loss": 0.5992231750488282, "grad_norm": 0.06886265426874161, "lr": 9.270833333333334e-05, "sec_per_update": 0.34201680183410643, "speed": 266.19059844945644, "vram_gb": 25.572148323059082} {"t": 1790412052.658246, "kind": "train", "update": 1350, "max_updates": 8000, "epoch": 11, "loss": 0.6561801290512085, "grad_norm": 0.08127732574939728, "lr": 9.236111111111112e-05, "sec_per_update": 0.37561387062072754, "speed": 269.56985497610265, "vram_gb": 25.572148323059082} {"t": 1790412061.7775805, "kind": "train", "update": 1375, "max_updates": 8000, "epoch": 11, "loss": 0.6145827937126159, "grad_norm": 0.20869454741477966, "lr": 9.201388888888889e-05, "sec_per_update": 0.36476662635803225, "speed": 272.8450366021773, "vram_gb": 25.572148323059082} {"t": 1790412071.1545463, "kind": "train", "update": 1400, "max_updates": 8000, "epoch": 12, "loss": 0.6073225724697113, "grad_norm": 0.22331194579601288, "lr": 9.166666666666667e-05, "sec_per_update": 0.37507704734802244, "speed": 276.3454986098793, "vram_gb": 25.572148323059082} {"t": 1790412081.038087, "kind": "train", "update": 1425, "max_updates": 8000, "epoch": 12, "loss": 0.5873915100097656, "grad_norm": 0.24128590524196625, "lr": 9.131944444444445e-05, "sec_per_update": 0.39534462928771974, "speed": 279.3770506368674, "vram_gb": 25.572148323059082} {"t": 1790412090.12905, "kind": "train", "update": 1450, "max_updates": 8000, "epoch": 12, "loss": 0.6113316774368286, "grad_norm": 0.08730830997228622, "lr": 9.097222222222223e-05, "sec_per_update": 0.36364402770996096, "speed": 283.0031136568683, "vram_gb": 25.572148323059082} {"t": 1790412098.617869, "kind": "train", "update": 1475, "max_updates": 8000, "epoch": 12, "loss": 0.6136069536209107, "grad_norm": 0.048539720475673676, "lr": 9.062500000000001e-05, "sec_per_update": 0.33955146789550783, "speed": 285.9479743312251, "vram_gb": 25.572148323059082} {"t": 1790412106.8852549, "kind": "train", "update": 1500, "max_updates": 8000, "epoch": 12, "loss": 0.5940157651901246, "grad_norm": 0.06604445725679398, "lr": 9.027777777777779e-05, "sec_per_update": 0.3306924343109131, "speed": 289.2140697139797, "vram_gb": 25.572148323059082} {"t": 1790412107.6936157, "kind": "val", "update": 1500, "val_loss": 0.6392743786176046} {"t": 1790412116.456561, "kind": "train", "update": 1525, "max_updates": 8000, "epoch": 13, "loss": 0.6136642050743103, "grad_norm": 0.04097522422671318, "lr": 8.993055555555557e-05, "sec_per_update": 0.3828464984893799, "speed": 291.8796146414726, "vram_gb": 25.572148323059082} {"t": 1790412126.2213783, "kind": "train", "update": 1550, "max_updates": 8000, "epoch": 13, "loss": 0.6172984075546265, "grad_norm": 0.057243719696998596, "lr": 8.958333333333335e-05, "sec_per_update": 0.39059239387512207, "speed": 294.8631167794671, "vram_gb": 25.572148323059082} {"t": 1790412134.7348197, "kind": "train", "update": 1575, "max_updates": 8000, "epoch": 13, "loss": 0.6363663792610168, "grad_norm": 0.06264521181583405, "lr": 8.923611111111113e-05, "sec_per_update": 0.34054083824157716, "speed": 298.07405746782115, "vram_gb": 25.572148323059082} {"t": 1790412143.8657322, "kind": "train", "update": 1600, "max_updates": 8000, "epoch": 13, "loss": 0.6302765822410583, "grad_norm": 0.10923252999782562, "lr": 8.888888888888889e-05, "sec_per_update": 0.36523651123046874, "speed": 300.994993459776, "vram_gb": 25.572148323059082} {"t": 1790412153.7220669, "kind": "train", "update": 1625, "max_updates": 8000, "epoch": 14, "loss": 0.6258451318740845, "grad_norm": 0.07289490103721619, "lr": 8.854166666666667e-05, "sec_per_update": 0.39423317909240724, "speed": 303.78192803816603, "vram_gb": 25.572148323059082} {"t": 1790412162.9319618, "kind": "train", "update": 1650, "max_updates": 8000, "epoch": 14, "loss": 0.5881187951564789, "grad_norm": 0.06668366491794586, "lr": 8.819444444444445e-05, "sec_per_update": 0.36826338768005373, "speed": 306.9671865877534, "vram_gb": 25.572148323059082} {"t": 1790412171.7580953, "kind": "train", "update": 1675, "max_updates": 8000, "epoch": 14, "loss": 0.6161750125885009, "grad_norm": 0.037623774260282516, "lr": 8.784722222222222e-05, "sec_per_update": 0.35319104194641116, "speed": 309.8234309935084, "vram_gb": 25.572148323059082} {"t": 1790412181.2765207, "kind": "train", "update": 1700, "max_updates": 8000, "epoch": 14, "loss": 0.6176591467857361, "grad_norm": 0.07449880242347717, "lr": 8.75e-05, "sec_per_update": 0.3807420253753662, "speed": 312.50137034605285, "vram_gb": 25.572148323059082} {"t": 1790412189.8618255, "kind": "train", "update": 1725, "max_updates": 8000, "epoch": 14, "loss": 0.5942883515357971, "grad_norm": 0.04003840312361717, "lr": 8.715277777777778e-05, "sec_per_update": 0.34341654777526853, "speed": 315.18251998221274, "vram_gb": 25.572148323059082} {"t": 1790412199.2540343, "kind": "train", "update": 1750, "max_updates": 8000, "epoch": 15, "loss": 0.6283674240112305, "grad_norm": 0.08846288174390793, "lr": 8.680555555555556e-05, "sec_per_update": 0.3756880760192871, "speed": 317.8509632266863, "vram_gb": 25.572148323059082} {"t": 1790412207.4403367, "kind": "train", "update": 1775, "max_updates": 8000, "epoch": 15, "loss": 0.6151638913154602, "grad_norm": 0.0677659809589386, "lr": 8.645833333333334e-05, "sec_per_update": 0.3273362922668457, "speed": 320.461614700868, "vram_gb": 25.572148323059082} {"t": 1790412216.9009478, "kind": "train", "update": 1800, "max_updates": 8000, "epoch": 15, "loss": 0.6203151965141296, "grad_norm": 0.023910799995064735, "lr": 8.611111111111112e-05, "sec_per_update": 0.3785425758361816, "speed": 323.1015213444947, "vram_gb": 25.572148323059082} {"t": 1790412226.2293332, "kind": "train", "update": 1825, "max_updates": 8000, "epoch": 15, "loss": 0.6013158535957337, "grad_norm": 0.09319061040878296, "lr": 8.576388888888888e-05, "sec_per_update": 0.37298978805541994, "speed": 325.8587050502445, "vram_gb": 25.572148323059082} {"t": 1790412235.6259987, "kind": "train", "update": 1850, "max_updates": 8000, "epoch": 15, "loss": 0.6152528667449951, "grad_norm": 0.11999815702438354, "lr": 8.541666666666666e-05, "sec_per_update": 0.37586737632751466, "speed": 328.8855062853095, "vram_gb": 25.5728120803833} {"t": 1790412245.1896164, "kind": "train", "update": 1875, "max_updates": 8000, "epoch": 16, "loss": 0.6106893682479858, "grad_norm": 0.08346477150917053, "lr": 8.506944444444444e-05, "sec_per_update": 0.3825601291656494, "speed": 331.2801138377586, "vram_gb": 25.5728120803833} {"t": 1790412255.0413232, "kind": "train", "update": 1900, "max_updates": 8000, "epoch": 16, "loss": 0.6044126665592193, "grad_norm": 0.07640587538480759, "lr": 8.472222222222222e-05, "sec_per_update": 0.39419137001037596, "speed": 334.03897790028805, "vram_gb": 25.5728120803833} {"t": 1790412264.1142218, "kind": "train", "update": 1925, "max_updates": 8000, "epoch": 16, "loss": 0.5961460590362548, "grad_norm": 0.15933020412921906, "lr": 8.4375e-05, "sec_per_update": 0.36291721343994143, "speed": 336.6610411410123, "vram_gb": 25.5728120803833} {"t": 1790412271.8587132, "kind": "train", "update": 1950, "max_updates": 8000, "epoch": 16, "loss": 0.6208109498023987, "grad_norm": 0.11287906020879745, "lr": 8.402777777777778e-05, "sec_per_update": 0.30977964401245117, "speed": 339.168296469585, "vram_gb": 25.5728120803833} {"t": 1790412280.7948382, "kind": "train", "update": 1975, "max_updates": 8000, "epoch": 17, "loss": 0.6021999788284301, "grad_norm": 0.053345054388046265, "lr": 8.368055555555556e-05, "sec_per_update": 0.3574475574493408, "speed": 341.5820887615562, "vram_gb": 25.5728120803833} {"t": 1790412290.6112661, "kind": "train", "update": 2000, "max_updates": 8000, "epoch": 17, "loss": 0.611911747455597, "grad_norm": 0.13564984500408173, "lr": 8.333333333333334e-05, "sec_per_update": 0.39265836715698244, "speed": 343.9407452139933, "vram_gb": 25.5728120803833} {"t": 1790412291.2743082, "kind": "val", "update": 2000, "val_loss": 0.6291637023289999} {"t": 1790412305.0400848, "kind": "checkpoint", "update": 2000} {"t": 1790412314.006631, "kind": "train", "update": 2025, "max_updates": 8000, "epoch": 17, "loss": 0.593350602388382, "grad_norm": 0.07953763753175735, "lr": 8.298611111111112e-05, "sec_per_update": 0.9357731914520264, "speed": 343.6115191922767, "vram_gb": 25.5728120803833} {"t": 1790412322.5226088, "kind": "train", "update": 2050, "max_updates": 8000, "epoch": 17, "loss": 0.5876725721359253, "grad_norm": 0.03976723179221153, "lr": 8.263888888888889e-05, "sec_per_update": 0.3406157112121582, "speed": 346.0168162097549, "vram_gb": 25.5728120803833} {"t": 1790412332.1575842, "kind": "train", "update": 2075, "max_updates": 8000, "epoch": 17, "loss": 0.6028019571304322, "grad_norm": 0.10748686641454697, "lr": 8.229166666666667e-05, "sec_per_update": 0.3854615020751953, "speed": 348.3608784310716, "vram_gb": 25.5728120803833} {"t": 1790412341.7454686, "kind": "train", "update": 2100, "max_updates": 8000, "epoch": 18, "loss": 0.5978861999511719, "grad_norm": 0.019819770008325577, "lr": 8.194444444444445e-05, "sec_per_update": 0.38351943969726565, "speed": 350.59215214415525, "vram_gb": 25.5728120803833} {"t": 1790412349.5886939, "kind": "train", "update": 2125, "max_updates": 8000, "epoch": 18, "loss": 0.6206836915016174, "grad_norm": 0.04733414947986603, "lr": 8.159722222222223e-05, "sec_per_update": 0.3137138271331787, "speed": 352.9051286291385, "vram_gb": 25.573785305023193} {"t": 1790412359.7909112, "kind": "train", "update": 2150, "max_updates": 8000, "epoch": 18, "loss": 0.6050532817840576, "grad_norm": 0.06429401785135269, "lr": 8.125000000000001e-05, "sec_per_update": 0.40810270309448243, "speed": 355.4358425744753, "vram_gb": 25.573785305023193} {"t": 1790412368.9238074, "kind": "train", "update": 2175, "max_updates": 8000, "epoch": 18, "loss": 0.5932930600643158, "grad_norm": 0.03528021275997162, "lr": 8.090277777777778e-05, "sec_per_update": 0.3652998638153076, "speed": 357.7434828316276, "vram_gb": 25.573785305023193} {"t": 1790412377.059424, "kind": "train", "update": 2200, "max_updates": 8000, "epoch": 18, "loss": 0.6095753276348114, "grad_norm": 0.1166008785367012, "lr": 8.055555555555556e-05, "sec_per_update": 0.32544243812561036, "speed": 360.10263203828714, "vram_gb": 25.573785305023193} {"t": 1790412387.0631244, "kind": "train", "update": 2225, "max_updates": 8000, "epoch": 19, "loss": 0.6121022152900696, "grad_norm": 0.10619866847991943, "lr": 8.020833333333334e-05, "sec_per_update": 0.40014214515686036, "speed": 362.4386221131929, "vram_gb": 25.573785305023193} {"t": 1790412395.9103556, "kind": "train", "update": 2250, "max_updates": 8000, "epoch": 19, "loss": 0.618574035167694, "grad_norm": 0.19257771968841553, "lr": 7.986111111111112e-05, "sec_per_update": 0.353891019821167, "speed": 364.5697787729827, "vram_gb": 25.573785305023193} {"t": 1790412404.4905758, "kind": "train", "update": 2275, "max_updates": 8000, "epoch": 19, "loss": 0.6354413688182831, "grad_norm": 0.02698569744825363, "lr": 7.951388888888888e-05, "sec_per_update": 0.3432076358795166, "speed": 366.5610425600569, "vram_gb": 25.573785305023193} {"t": 1790412413.6702397, "kind": "train", "update": 2300, "max_updates": 8000, "epoch": 19, "loss": 0.585665762424469, "grad_norm": 0.07496923953294754, "lr": 7.916666666666666e-05, "sec_per_update": 0.36704575538635253, "speed": 368.8062482109911, "vram_gb": 25.573785305023193} {"t": 1790412422.554089, "kind": "train", "update": 2325, "max_updates": 8000, "epoch": 20, "loss": 0.6340105485916138, "grad_norm": 0.4254879653453827, "lr": 7.881944444444444e-05, "sec_per_update": 0.3554848766326904, "speed": 371.056022098587, "vram_gb": 25.573785305023193} {"t": 1790412431.3684633, "kind": "train", "update": 2350, "max_updates": 8000, "epoch": 20, "loss": 0.6406313896179199, "grad_norm": 0.08221424371004105, "lr": 7.847222222222222e-05, "sec_per_update": 0.3525737380981445, "speed": 373.117692034346, "vram_gb": 25.573785305023193} {"t": 1790412440.603084, "kind": "train", "update": 2375, "max_updates": 8000, "epoch": 20, "loss": 0.5933948171138763, "grad_norm": 0.019189830869436264, "lr": 7.8125e-05, "sec_per_update": 0.36925912857055665, "speed": 375.3247347911745, "vram_gb": 25.573785305023193} {"t": 1790412451.121053, "kind": "train", "update": 2400, "max_updates": 8000, "epoch": 20, "loss": 0.5975787663459777, "grad_norm": 0.133104607462883, "lr": 7.777777777777778e-05, "sec_per_update": 0.4208596134185791, "speed": 377.3278054415624, "vram_gb": 25.573785305023193} {"t": 1790412459.2410488, "kind": "train", "update": 2425, "max_updates": 8000, "epoch": 20, "loss": 0.590526990890503, "grad_norm": 0.08265983313322067, "lr": 7.743055555555556e-05, "sec_per_update": 0.3247993946075439, "speed": 379.5095801675699, "vram_gb": 25.573785305023193} {"t": 1790412467.4739976, "kind": "train", "update": 2450, "max_updates": 8000, "epoch": 21, "loss": 0.5756406188011169, "grad_norm": 0.1206614300608635, "lr": 7.708333333333334e-05, "sec_per_update": 0.3293089294433594, "speed": 381.53443633765966, "vram_gb": 25.573785305023193} {"t": 1790412476.8990824, "kind": "train", "update": 2475, "max_updates": 8000, "epoch": 21, "loss": 0.6078348815441131, "grad_norm": 0.06329820305109024, "lr": 7.673611111111112e-05, "sec_per_update": 0.37699782371520996, "speed": 383.7519343951299, "vram_gb": 25.573785305023193} {"t": 1790412486.8012652, "kind": "train", "update": 2500, "max_updates": 8000, "epoch": 21, "loss": 0.6114926826953888, "grad_norm": 0.06979589909315109, "lr": 7.638888888888889e-05, "sec_per_update": 0.3961019420623779, "speed": 385.94944888063344, "vram_gb": 25.573785305023193} {"t": 1790412487.4657784, "kind": "val", "update": 2500, "val_loss": 0.6239073276519775} {"t": 1790412496.0345478, "kind": "train", "update": 2525, "max_updates": 8000, "epoch": 21, "loss": 0.5873431944847107, "grad_norm": 0.02880818583071232, "lr": 7.604166666666667e-05, "sec_per_update": 0.3693308639526367, "speed": 387.59973473682305, "vram_gb": 25.573785305023193} {"t": 1790412505.3312173, "kind": "train", "update": 2550, "max_updates": 8000, "epoch": 21, "loss": 0.6120765447616577, "grad_norm": 0.08350429683923721, "lr": 7.569444444444445e-05, "sec_per_update": 0.37170641899108886, "speed": 389.5097153595931, "vram_gb": 25.573785305023193} {"t": 1790412514.762788, "kind": "train", "update": 2575, "max_updates": 8000, "epoch": 22, "loss": 0.5985669994354248, "grad_norm": 0.039109762758016586, "lr": 7.534722222222223e-05, "sec_per_update": 0.3774218368530273, "speed": 391.3750285954407, "vram_gb": 25.573785305023193} {"t": 1790412523.9055293, "kind": "train", "update": 2600, "max_updates": 8000, "epoch": 22, "loss": 0.5751598227024078, "grad_norm": 0.07480902969837189, "lr": 7.500000000000001e-05, "sec_per_update": 0.3657072925567627, "speed": 393.26235533712037, "vram_gb": 25.573785305023193} {"t": 1790412532.788453, "kind": "train", "update": 2625, "max_updates": 8000, "epoch": 22, "loss": 0.5910657358169555, "grad_norm": 0.07261669635772705, "lr": 7.465277777777779e-05, "sec_per_update": 0.3553196620941162, "speed": 395.1192562262271, "vram_gb": 25.573785305023193} {"t": 1790412542.138567, "kind": "train", "update": 2650, "max_updates": 8000, "epoch": 22, "loss": 0.6202205526828766, "grad_norm": 0.2694619297981262, "lr": 7.430555555555557e-05, "sec_per_update": 0.37400580406188966, "speed": 396.82085224061893, "vram_gb": 25.573785305023193} {"t": 1790412551.3236744, "kind": "train", "update": 2675, "max_updates": 8000, "epoch": 23, "loss": 0.5887118339538574, "grad_norm": 0.04676234722137451, "lr": 7.395833333333335e-05, "sec_per_update": 0.36740702629089356, "speed": 398.98319361864344, "vram_gb": 25.573785305023193} {"t": 1790412560.877704, "kind": "train", "update": 2700, "max_updates": 8000, "epoch": 23, "loss": 0.6173994445800781, "grad_norm": 0.06225204095244408, "lr": 7.361111111111111e-05, "sec_per_update": 0.3820726108551025, "speed": 400.83619729257657, "vram_gb": 25.573785305023193} {"t": 1790412569.9027321, "kind": "train", "update": 2725, "max_updates": 8000, "epoch": 23, "loss": 0.6171201539039611, "grad_norm": 0.27008697390556335, "lr": 7.326388888888889e-05, "sec_per_update": 0.3610873031616211, "speed": 402.7940402558112, "vram_gb": 25.573785305023193} {"t": 1790412578.7620294, "kind": "train", "update": 2750, "max_updates": 8000, "epoch": 23, "loss": 0.5930681073665619, "grad_norm": 0.08380739390850067, "lr": 7.291666666666667e-05, "sec_per_update": 0.3543680000305176, "speed": 404.43894175712944, "vram_gb": 25.573785305023193} {"t": 1790412588.0602157, "kind": "train", "update": 2775, "max_updates": 8000, "epoch": 23, "loss": 0.5999419188499451, "grad_norm": 0.03762417659163475, "lr": 7.256944444444444e-05, "sec_per_update": 0.3717661762237549, "speed": 406.16626210552226, "vram_gb": 25.573785305023193} {"t": 1790412597.6571915, "kind": "train", "update": 2800, "max_updates": 8000, "epoch": 24, "loss": 0.6404625272750855, "grad_norm": 0.15995994210243225, "lr": 7.222222222222222e-05, "sec_per_update": 0.3838991737365723, "speed": 408.1346235307192, "vram_gb": 25.573785305023193} {"t": 1790412606.4364078, "kind": "train", "update": 2825, "max_updates": 8000, "epoch": 24, "loss": 0.5844697046279907, "grad_norm": 0.08519211411476135, "lr": 7.1875e-05, "sec_per_update": 0.3513099956512451, "speed": 409.9660345357144, "vram_gb": 25.573785305023193} {"t": 1790412615.3730972, "kind": "train", "update": 2850, "max_updates": 8000, "epoch": 24, "loss": 0.6050732564926148, "grad_norm": 0.04065209627151489, "lr": 7.152777777777778e-05, "sec_per_update": 0.35745662689208985, "speed": 411.7287085836634, "vram_gb": 25.573785305023193} {"t": 1790412624.8206713, "kind": "train", "update": 2875, "max_updates": 8000, "epoch": 24, "loss": 0.5971733784675598, "grad_norm": 0.029275383800268173, "lr": 7.118055555555556e-05, "sec_per_update": 0.3778972053527832, "speed": 413.30395049733437, "vram_gb": 25.573785305023193} {"t": 1790412633.0475385, "kind": "train", "update": 2900, "max_updates": 8000, "epoch": 24, "loss": 0.6091199505329132, "grad_norm": 0.05222068727016449, "lr": 7.083333333333334e-05, "sec_per_update": 0.32909090995788576, "speed": 415.20971653774933, "vram_gb": 25.573785305023193} {"t": 1790412642.8105247, "kind": "train", "update": 2925, "max_updates": 8000, "epoch": 25, "loss": 0.6185024952888489, "grad_norm": 0.0951673686504364, "lr": 7.048611111111112e-05, "sec_per_update": 0.39052720069885255, "speed": 416.74391632506007, "vram_gb": 25.573785305023193} {"t": 1790412652.5330913, "kind": "train", "update": 2950, "max_updates": 8000, "epoch": 25, "loss": 0.5965577793121338, "grad_norm": 0.09663140773773193, "lr": 7.013888888888888e-05, "sec_per_update": 0.38890006065368654, "speed": 418.6501860625804, "vram_gb": 25.573785305023193} {"t": 1790412661.7344162, "kind": "train", "update": 2975, "max_updates": 8000, "epoch": 25, "loss": 0.6057767403125763, "grad_norm": 0.05103851854801178, "lr": 6.979166666666666e-05, "sec_per_update": 0.36805357933044436, "speed": 420.4533450610596, "vram_gb": 25.573785305023193} {"t": 1790412670.1225717, "kind": "train", "update": 3000, "max_updates": 8000, "epoch": 25, "loss": 0.610289306640625, "grad_norm": 0.14849938452243805, "lr": 6.944444444444444e-05, "sec_per_update": 0.3355226993560791, "speed": 421.9591450683051, "vram_gb": 25.573785305023193} {"t": 1790412670.7455115, "kind": "val", "update": 3000, "val_loss": 0.6217572391033173} {"t": 1790412684.7018301, "kind": "checkpoint", "update": 3000} {"t": 1790412693.2949684, "kind": "train", "update": 3025, "max_updates": 8000, "epoch": 26, "loss": 0.6235876226425171, "grad_norm": 0.0925745964050293, "lr": 6.909722222222222e-05, "sec_per_update": 0.9267119979858398, "speed": 420.7527364865991, "vram_gb": 25.573785305023193} {"t": 1790412702.0904713, "kind": "train", "update": 3050, "max_updates": 8000, "epoch": 26, "loss": 0.6021445000171661, "grad_norm": 0.07504400610923767, "lr": 6.875e-05, "sec_per_update": 0.3519827079772949, "speed": 422.37846474999884, "vram_gb": 25.573785305023193} {"t": 1790412711.7184443, "kind": "train", "update": 3075, "max_updates": 8000, "epoch": 26, "loss": 0.6188790845870972, "grad_norm": 0.0732988566160202, "lr": 6.840277777777778e-05, "sec_per_update": 0.38512799263000486, "speed": 424.26520189959047, "vram_gb": 25.573785305023193} {"t": 1790412720.1586232, "kind": "train", "update": 3100, "max_updates": 8000, "epoch": 26, "loss": 0.6038633203506469, "grad_norm": 0.05727212131023407, "lr": 6.805555555555556e-05, "sec_per_update": 0.3376193904876709, "speed": 425.7030816996551, "vram_gb": 25.573785305023193} {"t": 1790412728.9576159, "kind": "train", "update": 3125, "max_updates": 8000, "epoch": 26, "loss": 0.5895130276679993, "grad_norm": 0.03733319789171219, "lr": 6.770833333333334e-05, "sec_per_update": 0.3518227958679199, "speed": 427.24728441839403, "vram_gb": 25.573785305023193} {"t": 1790412738.0803678, "kind": "train", "update": 3150, "max_updates": 8000, "epoch": 27, "loss": 0.5959513783454895, "grad_norm": 0.07048115134239197, "lr": 6.736111111111112e-05, "sec_per_update": 0.36504789352416994, "speed": 428.8029509553771, "vram_gb": 25.573785305023193} {"t": 1790412747.3211417, "kind": "train", "update": 3175, "max_updates": 8000, "epoch": 27, "loss": 0.6104988503456116, "grad_norm": 0.10413394868373871, "lr": 6.701388888888889e-05, "sec_per_update": 0.3696280479431152, "speed": 430.5136853413513, "vram_gb": 25.573785305023193} {"t": 1790412756.3535538, "kind": "train", "update": 3200, "max_updates": 8000, "epoch": 27, "loss": 0.6002311873435974, "grad_norm": 0.11660690605640411, "lr": 6.666666666666667e-05, "sec_per_update": 0.361294641494751, "speed": 432.04414910889716, "vram_gb": 25.573785305023193} {"t": 1790412764.9867995, "kind": "train", "update": 3225, "max_updates": 8000, "epoch": 27, "loss": 0.6218644940853119, "grad_norm": 0.25245803594589233, "lr": 6.631944444444445e-05, "sec_per_update": 0.34533612251281737, "speed": 433.61284354871333, "vram_gb": 25.573785305023193} {"t": 1790412774.4208436, "kind": "train", "update": 3250, "max_updates": 8000, "epoch": 28, "loss": 0.6046915483474732, "grad_norm": 0.11426277458667755, "lr": 6.597222222222223e-05, "sec_per_update": 0.377209415435791, "speed": 435.2583631613258, "vram_gb": 25.573785305023193} {"t": 1790412784.2289164, "kind": "train", "update": 3275, "max_updates": 8000, "epoch": 28, "loss": 0.5966799747943878, "grad_norm": 0.07965132594108582, "lr": 6.562500000000001e-05, "sec_per_update": 0.39246819496154783, "speed": 436.5981073063706, "vram_gb": 25.573785305023193} {"t": 1790412793.9630663, "kind": "train", "update": 3300, "max_updates": 8000, "epoch": 28, "loss": 0.6135600864887237, "grad_norm": 0.09350656718015671, "lr": 6.527777777777778e-05, "sec_per_update": 0.38937753677368164, "speed": 438.347263039156, "vram_gb": 25.573785305023193} {"t": 1790412802.9236386, "kind": "train", "update": 3325, "max_updates": 8000, "epoch": 28, "loss": 0.5990207695960998, "grad_norm": 0.19556014239788055, "lr": 6.493055555555556e-05, "sec_per_update": 0.3584246921539307, "speed": 439.7674716858677, "vram_gb": 25.573785305023193} {"t": 1790412811.11451, "kind": "train", "update": 3350, "max_updates": 8000, "epoch": 28, "loss": 0.5793376863002777, "grad_norm": 0.04210825264453888, "lr": 6.458333333333334e-05, "sec_per_update": 0.32753790855407716, "speed": 441.27390406771013, "vram_gb": 25.573785305023193} {"t": 1790412819.5736802, "kind": "train", "update": 3375, "max_updates": 8000, "epoch": 29, "loss": 0.5991100239753723, "grad_norm": 0.12776269018650055, "lr": 6.423611111111112e-05, "sec_per_update": 0.3384576892852783, "speed": 442.74810834353457, "vram_gb": 25.573785305023193} {"t": 1790412829.3745525, "kind": "train", "update": 3400, "max_updates": 8000, "epoch": 29, "loss": 0.6092597281932831, "grad_norm": 0.05053310841321945, "lr": 6.388888888888888e-05, "sec_per_update": 0.39203170776367186, "speed": 444.2377084921231, "vram_gb": 25.573785305023193} {"t": 1790412838.5090246, "kind": "train", "update": 3425, "max_updates": 8000, "epoch": 29, "loss": 0.5850168240070343, "grad_norm": 0.10888852179050446, "lr": 6.354166666666666e-05, "sec_per_update": 0.3653771209716797, "speed": 445.55812565340466, "vram_gb": 25.573785305023193} {"t": 1790412847.7790997, "kind": "train", "update": 3450, "max_updates": 8000, "epoch": 29, "loss": 0.601176552772522, "grad_norm": 0.20579655468463898, "lr": 6.319444444444444e-05, "sec_per_update": 0.37080739974975585, "speed": 447.0687017260026, "vram_gb": 25.573785305023193} {"t": 1790412856.0877855, "kind": "train", "update": 3475, "max_updates": 8000, "epoch": 29, "loss": 0.6071528244018555, "grad_norm": 0.0823894664645195, "lr": 6.284722222222222e-05, "sec_per_update": 0.3323490047454834, "speed": 448.62903593549015, "vram_gb": 25.573785305023193} {"t": 1790412865.9039104, "kind": "train", "update": 3500, "max_updates": 8000, "epoch": 30, "loss": 0.5810014426708221, "grad_norm": 0.1567849963903427, "lr": 6.25e-05, "sec_per_update": 0.39264239311218263, "speed": 449.8786236118882, "vram_gb": 25.573785305023193} {"t": 1790412866.5880485, "kind": "val", "update": 3500, "val_loss": 0.6204530696074168} {"t": 1790412875.7602599, "kind": "train", "update": 3525, "max_updates": 8000, "epoch": 30, "loss": 0.5894006156921386, "grad_norm": 0.05821894481778145, "lr": 6.215277777777778e-05, "sec_per_update": 0.39412174224853513, "speed": 451.16625395397, "vram_gb": 25.573785305023193} {"t": 1790412885.3424692, "kind": "train", "update": 3550, "max_updates": 8000, "epoch": 30, "loss": 0.6038748359680176, "grad_norm": 0.05562679097056389, "lr": 6.180555555555556e-05, "sec_per_update": 0.38341996192932126, "speed": 452.4037341788402, "vram_gb": 25.573785305023193} {"t": 1790412893.268543, "kind": "train", "update": 3575, "max_updates": 8000, "epoch": 30, "loss": 0.5850949966907502, "grad_norm": 0.1150360107421875, "lr": 6.145833333333334e-05, "sec_per_update": 0.31702930450439454, "speed": 453.86958770019834, "vram_gb": 25.573785305023193} {"t": 1790412903.2375293, "kind": "train", "update": 3600, "max_updates": 8000, "epoch": 31, "loss": 0.5789397335052491, "grad_norm": 0.06296088546514511, "lr": 6.111111111111112e-05, "sec_per_update": 0.3987717914581299, "speed": 455.32596812469194, "vram_gb": 25.573785305023193} {"t": 1790412912.3510525, "kind": "train", "update": 3625, "max_updates": 8000, "epoch": 31, "loss": 0.5752247965335846, "grad_norm": 0.26390236616134644, "lr": 6.076388888888889e-05, "sec_per_update": 0.36453770637512206, "speed": 456.7264546753579, "vram_gb": 25.573785305023193} {"t": 1790412921.4542751, "kind": "train", "update": 3650, "max_updates": 8000, "epoch": 31, "loss": 0.6006597876548767, "grad_norm": 0.05855653062462807, "lr": 6.041666666666667e-05, "sec_per_update": 0.36399600028991697, "speed": 458.3276730124315, "vram_gb": 25.573785305023193} {"t": 1790412929.7898762, "kind": "train", "update": 3675, "max_updates": 8000, "epoch": 31, "loss": 0.6192888140678405, "grad_norm": 0.054948434233665466, "lr": 6.006944444444445e-05, "sec_per_update": 0.3335647392272949, "speed": 459.6935978793382, "vram_gb": 25.573785305023193} {"t": 1790412938.7555518, "kind": "train", "update": 3700, "max_updates": 8000, "epoch": 31, "loss": 0.5733065319061279, "grad_norm": 0.0490972176194191, "lr": 5.972222222222223e-05, "sec_per_update": 0.3586249828338623, "speed": 460.99496670708925, "vram_gb": 25.573785305023193} {"t": 1790412948.459947, "kind": "train", "update": 3725, "max_updates": 8000, "epoch": 32, "loss": 0.6040151596069336, "grad_norm": 0.09008781611919403, "lr": 5.9375e-05, "sec_per_update": 0.3881632614135742, "speed": 462.3281945503874, "vram_gb": 25.573785305023193} {"t": 1790412957.8873541, "kind": "train", "update": 3750, "max_updates": 8000, "epoch": 32, "loss": 0.6144152808189393, "grad_norm": 0.12616343796253204, "lr": 5.902777777777778e-05, "sec_per_update": 0.3771080493927002, "speed": 463.69189841229604, "vram_gb": 25.573785305023193} {"t": 1790412966.6456597, "kind": "train", "update": 3775, "max_updates": 8000, "epoch": 32, "loss": 0.6081431174278259, "grad_norm": 0.1944907009601593, "lr": 5.868055555555556e-05, "sec_per_update": 0.35033326148986815, "speed": 464.96941965883923, "vram_gb": 25.573785305023193} {"t": 1790412976.2026105, "kind": "train", "update": 3800, "max_updates": 8000, "epoch": 32, "loss": 0.6003656196594238, "grad_norm": 0.12933480739593506, "lr": 5.833333333333334e-05, "sec_per_update": 0.3822756862640381, "speed": 466.01416684635694, "vram_gb": 25.573785305023193} {"t": 1790412984.130079, "kind": "train", "update": 3825, "max_updates": 8000, "epoch": 32, "loss": 0.6051198267936706, "grad_norm": 0.06774280965328217, "lr": 5.798611111111112e-05, "sec_per_update": 0.3170987415313721, "speed": 467.36386371025634, "vram_gb": 25.573785305023193} {"t": 1790412994.6595643, "kind": "train", "update": 3850, "max_updates": 8000, "epoch": 33, "loss": 0.612715311050415, "grad_norm": 0.10358993709087372, "lr": 5.7638888888888886e-05, "sec_per_update": 0.42117897033691404, "speed": 468.6059402011905, "vram_gb": 25.573785305023193} {"t": 1790413003.2487304, "kind": "train", "update": 3875, "max_updates": 8000, "epoch": 33, "loss": 0.5925045144557953, "grad_norm": 0.07310069352388382, "lr": 5.7291666666666666e-05, "sec_per_update": 0.3435703754425049, "speed": 469.6674827784788, "vram_gb": 25.573785305023193} {"t": 1790413011.3065834, "kind": "train", "update": 3900, "max_updates": 8000, "epoch": 33, "loss": 0.5971087980270385, "grad_norm": 0.19519862532615662, "lr": 5.6944444444444445e-05, "sec_per_update": 0.32230994224548337, "speed": 471.01232648193013, "vram_gb": 25.573785305023193} {"t": 1790413021.22855, "kind": "train", "update": 3925, "max_updates": 8000, "epoch": 33, "loss": 0.5960041344165802, "grad_norm": 0.06449916958808899, "lr": 5.6597222222222225e-05, "sec_per_update": 0.396879825592041, "speed": 472.1730156825643, "vram_gb": 25.573785305023193} {"t": 1790413030.83352, "kind": "train", "update": 3950, "max_updates": 8000, "epoch": 34, "loss": 0.6035065722465515, "grad_norm": 0.27689027786254883, "lr": 5.6250000000000005e-05, "sec_per_update": 0.3842006492614746, "speed": 473.4345440279446, "vram_gb": 25.573785305023193} {"t": 1790413039.6503856, "kind": "train", "update": 3975, "max_updates": 8000, "epoch": 34, "loss": 0.604899092912674, "grad_norm": 0.023644229397177696, "lr": 5.590277777777778e-05, "sec_per_update": 0.35253349304199216, "speed": 474.77451998627663, "vram_gb": 25.573785305023193} {"t": 1790413048.1733356, "kind": "train", "update": 4000, "max_updates": 8000, "epoch": 34, "loss": 0.594733612537384, "grad_norm": 0.14059986174106598, "lr": 5.555555555555556e-05, "sec_per_update": 0.3410596656799316, "speed": 476.1313530043228, "vram_gb": 25.573785305023193} {"t": 1790413048.8187945, "kind": "val", "update": 4000, "val_loss": 0.6199431916077932} {"t": 1790413063.0753407, "kind": "checkpoint", "update": 4000} {"t": 1790413071.4174976, "kind": "train", "update": 4025, "max_updates": 8000, "epoch": 34, "loss": 0.5842022681236267, "grad_norm": 0.09382713586091995, "lr": 5.520833333333334e-05, "sec_per_update": 0.9297682952880859, "speed": 474.5315916131963, "vram_gb": 25.573785305023193} {"t": 1790413081.4158223, "kind": "train", "update": 4050, "max_updates": 8000, "epoch": 34, "loss": 0.5983719384670257, "grad_norm": 0.12704886496067047, "lr": 5.486111111111112e-05, "sec_per_update": 0.3999203395843506, "speed": 475.7370393441914, "vram_gb": 25.573785305023193} {"t": 1790413090.8252492, "kind": "train", "update": 4075, "max_updates": 8000, "epoch": 35, "loss": 0.6208819699287415, "grad_norm": 0.13475938141345978, "lr": 5.4513888888888884e-05, "sec_per_update": 0.3763862991333008, "speed": 476.79022451634927, "vram_gb": 25.573785305023193} {"t": 1790413098.4926157, "kind": "train", "update": 4100, "max_updates": 8000, "epoch": 35, "loss": 0.593897385597229, "grad_norm": 0.09111468493938446, "lr": 5.4166666666666664e-05, "sec_per_update": 0.3065385627746582, "speed": 477.984487446962, "vram_gb": 25.573785305023193} {"t": 1790413108.331647, "kind": "train", "update": 4125, "max_updates": 8000, "epoch": 35, "loss": 0.629542727470398, "grad_norm": 0.19081389904022217, "lr": 5.3819444444444444e-05, "sec_per_update": 0.39371834754943846, "speed": 479.24217116036414, "vram_gb": 25.573785305023193} {"t": 1790413118.0018227, "kind": "train", "update": 4150, "max_updates": 8000, "epoch": 35, "loss": 0.5998423266410827, "grad_norm": 0.14640364050865173, "lr": 5.3472222222222224e-05, "sec_per_update": 0.38678881645202634, "speed": 480.34682877839475, "vram_gb": 25.573785305023193} {"t": 1790413126.5625844, "kind": "train", "update": 4175, "max_updates": 8000, "epoch": 35, "loss": 0.6176993870735168, "grad_norm": 0.20216114819049835, "lr": 5.3125000000000004e-05, "sec_per_update": 0.3422941780090332, "speed": 481.551437806391, "vram_gb": 25.573785305023193} {"t": 1790413136.3752434, "kind": "train", "update": 4200, "max_updates": 8000, "epoch": 36, "loss": 0.6117775499820709, "grad_norm": 0.09594213217496872, "lr": 5.2777777777777784e-05, "sec_per_update": 0.3926590538024902, "speed": 482.7304044359643, "vram_gb": 25.573785305023193} {"t": 1790413145.1099732, "kind": "train", "update": 4225, "max_updates": 8000, "epoch": 36, "loss": 0.5826602423191071, "grad_norm": 0.11004484444856644, "lr": 5.2430555555555564e-05, "sec_per_update": 0.34939722061157225, "speed": 483.7573847350234, "vram_gb": 25.573785305023193} {"t": 1790413154.6428733, "kind": "train", "update": 4250, "max_updates": 8000, "epoch": 36, "loss": 0.6082110428810119, "grad_norm": 0.01750103197991848, "lr": 5.208333333333334e-05, "sec_per_update": 0.38128923416137694, "speed": 484.76972233203117, "vram_gb": 25.573785305023193} {"t": 1790413163.7864473, "kind": "train", "update": 4275, "max_updates": 8000, "epoch": 36, "loss": 0.6123255825042725, "grad_norm": 0.14371344447135925, "lr": 5.1736111111111117e-05, "sec_per_update": 0.3657671546936035, "speed": 485.8752483877081, "vram_gb": 25.573785305023193} {"t": 1790413172.92435, "kind": "train", "update": 4300, "max_updates": 8000, "epoch": 37, "loss": 0.6075144982337952, "grad_norm": 0.1062256321310997, "lr": 5.138888888888889e-05, "sec_per_update": 0.36550827980041506, "speed": 487.1431601885414, "vram_gb": 25.573785305023193} {"t": 1790413182.569289, "kind": "train", "update": 4325, "max_updates": 8000, "epoch": 37, "loss": 0.5889302468299866, "grad_norm": 0.02762414515018463, "lr": 5.104166666666666e-05, "sec_per_update": 0.38579931259155276, "speed": 488.29775767518146, "vram_gb": 25.573785305023193} {"t": 1790413191.442274, "kind": "train", "update": 4350, "max_updates": 8000, "epoch": 37, "loss": 0.5881770932674408, "grad_norm": 0.2007870078086853, "lr": 5.069444444444444e-05, "sec_per_update": 0.3547498035430908, "speed": 489.4728574582414, "vram_gb": 25.573785305023193} {"t": 1790413201.0314562, "kind": "train", "update": 4375, "max_updates": 8000, "epoch": 37, "loss": 0.6000057017803192, "grad_norm": 0.14692822098731995, "lr": 5.034722222222222e-05, "sec_per_update": 0.3837419319152832, "speed": 490.49617052159266, "vram_gb": 25.573785305023193} {"t": 1790413209.6410563, "kind": "train", "update": 4400, "max_updates": 8000, "epoch": 37, "loss": 0.6500714159011841, "grad_norm": 0.03379867598414421, "lr": 5e-05, "sec_per_update": 0.34437952041625974, "speed": 491.53793801718246, "vram_gb": 25.573785305023193} {"t": 1790413218.3176603, "kind": "train", "update": 4425, "max_updates": 8000, "epoch": 38, "loss": 0.6243892884254456, "grad_norm": 0.06656256318092346, "lr": 4.965277777777778e-05, "sec_per_update": 0.3470548152923584, "speed": 492.68234903886656, "vram_gb": 25.573785305023193} {"t": 1790413228.6163094, "kind": "train", "update": 4450, "max_updates": 8000, "epoch": 38, "loss": 0.6018027949333191, "grad_norm": 0.11351517587900162, "lr": 4.930555555555556e-05, "sec_per_update": 0.4119601821899414, "speed": 493.66393274264783, "vram_gb": 25.573785305023193} {"t": 1790413237.3834221, "kind": "train", "update": 4475, "max_updates": 8000, "epoch": 38, "loss": 0.6035371506214142, "grad_norm": 0.04795940965414047, "lr": 4.8958333333333335e-05, "sec_per_update": 0.35068004608154296, "speed": 494.94939954868875, "vram_gb": 25.573785305023193} {"t": 1790413246.1532874, "kind": "train", "update": 4500, "max_updates": 8000, "epoch": 38, "loss": 0.5968300950527191, "grad_norm": 0.21424420177936554, "lr": 4.8611111111111115e-05, "sec_per_update": 0.35078508377075196, "speed": 495.75119914413006, "vram_gb": 25.573785305023193} {"t": 1790413246.7761624, "kind": "val", "update": 4500, "val_loss": 0.619319329659144} {"t": 1790413256.0324845, "kind": "train", "update": 4525, "max_updates": 8000, "epoch": 39, "loss": 0.5854579246044159, "grad_norm": 0.09741132706403732, "lr": 4.8263888888888895e-05, "sec_per_update": 0.3950393009185791, "speed": 496.61010897808126, "vram_gb": 25.573785305023193} {"t": 1790413265.3661847, "kind": "train", "update": 4550, "max_updates": 8000, "epoch": 39, "loss": 0.6049584114551544, "grad_norm": 0.09537078440189362, "lr": 4.791666666666667e-05, "sec_per_update": 0.37346715927124025, "speed": 497.7047037673566, "vram_gb": 25.573785305023193} {"t": 1790413274.8168955, "kind": "train", "update": 4575, "max_updates": 8000, "epoch": 39, "loss": 0.5678646552562714, "grad_norm": 0.05716056004166603, "lr": 4.756944444444444e-05, "sec_per_update": 0.37804662704467773, "speed": 498.6281013489207, "vram_gb": 25.573785305023193} {"t": 1790413283.0526721, "kind": "train", "update": 4600, "max_updates": 8000, "epoch": 39, "loss": 0.6109844517707824, "grad_norm": 0.048038505017757416, "lr": 4.722222222222222e-05, "sec_per_update": 0.3294373607635498, "speed": 499.7926278839506, "vram_gb": 25.573785305023193} {"t": 1790413292.399888, "kind": "train", "update": 4625, "max_updates": 8000, "epoch": 39, "loss": 0.5785321688652039, "grad_norm": 0.05302637815475464, "lr": 4.6875e-05, "sec_per_update": 0.3738795566558838, "speed": 500.7496400150745, "vram_gb": 25.573785305023193} {"t": 1790413302.1030116, "kind": "train", "update": 4650, "max_updates": 8000, "epoch": 40, "loss": 0.5924923205375672, "grad_norm": 0.0693344920873642, "lr": 4.652777777777778e-05, "sec_per_update": 0.38802419662475585, "speed": 501.6187043076361, "vram_gb": 25.573785305023193} {"t": 1790413311.8723657, "kind": "train", "update": 4675, "max_updates": 8000, "epoch": 40, "loss": 0.6130695509910583, "grad_norm": 0.08595722913742065, "lr": 4.618055555555556e-05, "sec_per_update": 0.39088348388671873, "speed": 502.8105955963909, "vram_gb": 25.573785305023193} {"t": 1790413321.222434, "kind": "train", "update": 4700, "max_updates": 8000, "epoch": 40, "loss": 0.577542781829834, "grad_norm": 0.09723711758852005, "lr": 4.5833333333333334e-05, "sec_per_update": 0.3740027904510498, "speed": 503.85229498151267, "vram_gb": 25.573785305023193} {"t": 1790413329.607197, "kind": "train", "update": 4725, "max_updates": 8000, "epoch": 40, "loss": 0.6063918268680573, "grad_norm": 0.023695247247815132, "lr": 4.5486111111111114e-05, "sec_per_update": 0.3353889846801758, "speed": 504.7967494356553, "vram_gb": 25.573785305023193} {"t": 1790413338.4344118, "kind": "train", "update": 4750, "max_updates": 8000, "epoch": 40, "loss": 0.6035262179374695, "grad_norm": 0.05274379998445511, "lr": 4.5138888888888894e-05, "sec_per_update": 0.3530902481079102, "speed": 505.68782968344084, "vram_gb": 25.573785305023193} {"t": 1790413347.8293693, "kind": "train", "update": 4775, "max_updates": 8000, "epoch": 41, "loss": 0.588496390581131, "grad_norm": 0.09346730262041092, "lr": 4.4791666666666673e-05, "sec_per_update": 0.37579712867736814, "speed": 506.8196106570874, "vram_gb": 25.573785305023193} {"t": 1790413357.2986119, "kind": "train", "update": 4800, "max_updates": 8000, "epoch": 41, "loss": 0.5806238842010498, "grad_norm": 0.031255584210157394, "lr": 4.4444444444444447e-05, "sec_per_update": 0.37876879692077636, "speed": 507.77255935966264, "vram_gb": 25.573785305023193} {"t": 1790413367.649416, "kind": "train", "update": 4825, "max_updates": 8000, "epoch": 41, "loss": 0.6129928147792816, "grad_norm": 0.07300570607185364, "lr": 4.4097222222222226e-05, "sec_per_update": 0.41402836799621584, "speed": 508.5945237412939, "vram_gb": 25.573785305023193} {"t": 1790413387.0157113, "kind": "train", "update": 4850, "max_updates": 8000, "epoch": 41, "loss": 0.5850380945205689, "grad_norm": 0.17041513323783875, "lr": 4.375e-05, "sec_per_update": 0.7744358444213867, "speed": 507.6329466950563, "vram_gb": 25.573785305023193} {"t": 1790413395.1746836, "kind": "train", "update": 4875, "max_updates": 8000, "epoch": 42, "loss": 0.608460898399353, "grad_norm": 0.05969937518239021, "lr": 4.340277777777778e-05, "sec_per_update": 0.3265743637084961, "speed": 508.519456743004, "vram_gb": 25.573785305023193} {"t": 1790413404.2442448, "kind": "train", "update": 4900, "max_updates": 8000, "epoch": 42, "loss": 0.5814646458625794, "grad_norm": 0.05729926750063896, "lr": 4.305555555555556e-05, "sec_per_update": 0.3627535915374756, "speed": 509.4486359079439, "vram_gb": 25.573785305023193} {"t": 1790413413.4252818, "kind": "train", "update": 4925, "max_updates": 8000, "epoch": 42, "loss": 0.6151999473571778, "grad_norm": 0.110280342400074, "lr": 4.270833333333333e-05, "sec_per_update": 0.36727420806884764, "speed": 510.453519714258, "vram_gb": 25.573785305023193} {"t": 1790413422.0703974, "kind": "train", "update": 4950, "max_updates": 8000, "epoch": 42, "loss": 0.5974718713760376, "grad_norm": 0.23215334117412567, "lr": 4.236111111111111e-05, "sec_per_update": 0.345796070098877, "speed": 511.35913084901205, "vram_gb": 25.573785305023193} {"t": 1790413430.789873, "kind": "train", "update": 4975, "max_updates": 8000, "epoch": 42, "loss": 0.6360674333572388, "grad_norm": 0.11982952058315277, "lr": 4.201388888888889e-05, "sec_per_update": 0.34878448486328123, "speed": 512.4878656094005, "vram_gb": 25.573785305023193} {"t": 1790413439.9063363, "kind": "train", "update": 5000, "max_updates": 8000, "epoch": 43, "loss": 0.6220068848133087, "grad_norm": 0.0531739741563797, "lr": 4.166666666666667e-05, "sec_per_update": 0.36466660499572756, "speed": 513.3543848349567, "vram_gb": 25.573785305023193} {"t": 1790413440.5968719, "kind": "val", "update": 5000, "val_loss": 0.6194551984469095} {"t": 1790413454.9529605, "kind": "checkpoint", "update": 5000} {"t": 1790413463.718663, "kind": "train", "update": 5025, "max_updates": 8000, "epoch": 43, "loss": 0.5925779926776886, "grad_norm": 0.049669403582811356, "lr": 4.1319444444444445e-05, "sec_per_update": 0.9524534797668457, "speed": 511.5733228264549, "vram_gb": 25.573785305023193} {"t": 1790413472.5651872, "kind": "train", "update": 5050, "max_updates": 8000, "epoch": 43, "loss": 0.5973811888694763, "grad_norm": 0.158927783370018, "lr": 4.0972222222222225e-05, "sec_per_update": 0.3538946628570557, "speed": 512.4591132770736, "vram_gb": 25.573785305023193} {"t": 1790413480.7203982, "kind": "train", "update": 5075, "max_updates": 8000, "epoch": 43, "loss": 0.6121859216690063, "grad_norm": 0.06254684180021286, "lr": 4.0625000000000005e-05, "sec_per_update": 0.3261766624450684, "speed": 513.4846244109859, "vram_gb": 25.573785305023193} {"t": 1790413489.823527, "kind": "train", "update": 5100, "max_updates": 8000, "epoch": 43, "loss": 0.5878397607803345, "grad_norm": 0.19743411242961884, "lr": 4.027777777777778e-05, "sec_per_update": 0.3641558742523193, "speed": 514.6163916843201, "vram_gb": 25.573785305023193} {"t": 1790413500.1427138, "kind": "train", "update": 5125, "max_updates": 8000, "epoch": 44, "loss": 0.6017758142948151, "grad_norm": 0.13934174180030823, "lr": 3.993055555555556e-05, "sec_per_update": 0.41258073806762696, "speed": 515.3818942202606, "vram_gb": 25.573785305023193} {"t": 1790413508.5264175, "kind": "train", "update": 5150, "max_updates": 8000, "epoch": 44, "loss": 0.5870275688171387, "grad_norm": 0.15107734501361847, "lr": 3.958333333333333e-05, "sec_per_update": 0.33552983283996585, "speed": 516.405132615324, "vram_gb": 25.573785305023193} {"t": 1790413517.2310362, "kind": "train", "update": 5175, "max_updates": 8000, "epoch": 44, "loss": 0.6095253217220307, "grad_norm": 0.06326558440923691, "lr": 3.923611111111111e-05, "sec_per_update": 0.34816354751586914, "speed": 517.4536097369812, "vram_gb": 25.573785305023193} {"t": 1790413525.6767974, "kind": "train", "update": 5200, "max_updates": 8000, "epoch": 44, "loss": 0.6067442417144775, "grad_norm": 0.04388001561164856, "lr": 3.888888888888889e-05, "sec_per_update": 0.33785765647888183, "speed": 518.2526325306528, "vram_gb": 25.573785305023193} {"t": 1790413535.945981, "kind": "train", "update": 5225, "max_updates": 8000, "epoch": 45, "loss": 0.5954852426052093, "grad_norm": 0.14576800167560577, "lr": 3.854166666666667e-05, "sec_per_update": 0.4105742359161377, "speed": 519.0928589807494, "vram_gb": 25.573785305023193} {"t": 1790413545.446509, "kind": "train", "update": 5250, "max_updates": 8000, "epoch": 45, "loss": 0.597111701965332, "grad_norm": 0.08855421096086502, "lr": 3.8194444444444444e-05, "sec_per_update": 0.3800361156463623, "speed": 520.1389510310811, "vram_gb": 25.573785305023193} {"t": 1790413553.7210178, "kind": "train", "update": 5275, "max_updates": 8000, "epoch": 45, "loss": 0.5991003632545471, "grad_norm": 0.13555669784545898, "lr": 3.7847222222222224e-05, "sec_per_update": 0.33115444183349607, "speed": 520.9572387248985, "vram_gb": 25.573785305023193} {"t": 1790413562.5021656, "kind": "train", "update": 5300, "max_updates": 8000, "epoch": 45, "loss": 0.5906443953514099, "grad_norm": 0.04442907124757767, "lr": 3.7500000000000003e-05, "sec_per_update": 0.351249418258667, "speed": 521.7693241213699, "vram_gb": 25.573785305023193} {"t": 1790413570.638792, "kind": "train", "update": 5325, "max_updates": 8000, "epoch": 45, "loss": 0.5875586640834808, "grad_norm": 0.10294172912836075, "lr": 3.715277777777778e-05, "sec_per_update": 0.32546319961547854, "speed": 522.7512382624415, "vram_gb": 25.573785305023193} {"t": 1790413581.0993237, "kind": "train", "update": 5350, "max_updates": 8000, "epoch": 46, "loss": 0.5782023906707764, "grad_norm": 0.25585925579071045, "lr": 3.6805555555555556e-05, "sec_per_update": 0.41839161872863767, "speed": 523.3336193196624, "vram_gb": 25.573785305023193} {"t": 1790413589.4108398, "kind": "train", "update": 5375, "max_updates": 8000, "epoch": 46, "loss": 0.5906147480010986, "grad_norm": 0.02619323506951332, "lr": 3.6458333333333336e-05, "sec_per_update": 0.33248827934265135, "speed": 524.3352646091299, "vram_gb": 25.573785305023193} {"t": 1790413597.714331, "kind": "train", "update": 5400, "max_updates": 8000, "epoch": 46, "loss": 0.5972812891006469, "grad_norm": 0.06937909126281738, "lr": 3.611111111111111e-05, "sec_per_update": 0.33214094161987306, "speed": 525.3065444836328, "vram_gb": 25.573785305023193} {"t": 1790413606.3375752, "kind": "train", "update": 5425, "max_updates": 8000, "epoch": 46, "loss": 0.6073799526691437, "grad_norm": 0.10889477282762527, "lr": 3.576388888888889e-05, "sec_per_update": 0.3449258899688721, "speed": 526.3114358478888, "vram_gb": 25.573785305023193} {"t": 1790413616.1247673, "kind": "train", "update": 5450, "max_updates": 8000, "epoch": 46, "loss": 0.6166214799880981, "grad_norm": 0.1759961098432541, "lr": 3.541666666666667e-05, "sec_per_update": 0.3914900588989258, "speed": 527.0934851553321, "vram_gb": 25.573785305023193} {"t": 1790413623.8890867, "kind": "train", "update": 5475, "max_updates": 8000, "epoch": 47, "loss": 0.6075017845630646, "grad_norm": 0.07808016240596771, "lr": 3.506944444444444e-05, "sec_per_update": 0.3105699634552002, "speed": 527.9023862207869, "vram_gb": 25.573785305023193} {"t": 1790413633.2125273, "kind": "train", "update": 5500, "max_updates": 8000, "epoch": 47, "loss": 0.5885541534423828, "grad_norm": 0.13646191358566284, "lr": 3.472222222222222e-05, "sec_per_update": 0.3729435920715332, "speed": 528.8089883267172, "vram_gb": 25.573785305023193} {"t": 1790413633.8647792, "kind": "val", "update": 5500, "val_loss": 0.6190898021062216} {"t": 1790413644.0630484, "kind": "train", "update": 5525, "max_updates": 8000, "epoch": 47, "loss": 0.6175569427013398, "grad_norm": 0.09650113433599472, "lr": 3.4375e-05, "sec_per_update": 0.4338880634307861, "speed": 529.5569901795487, "vram_gb": 25.573785305023193} {"t": 1790413652.9346867, "kind": "train", "update": 5550, "max_updates": 8000, "epoch": 47, "loss": 0.604192658662796, "grad_norm": 0.05057144537568092, "lr": 3.402777777777778e-05, "sec_per_update": 0.3548183822631836, "speed": 530.3538899514725, "vram_gb": 25.573785305023193} {"t": 1790413661.7234108, "kind": "train", "update": 5575, "max_updates": 8000, "epoch": 48, "loss": 0.5946308267116547, "grad_norm": 0.07162676751613617, "lr": 3.368055555555556e-05, "sec_per_update": 0.351710786819458, "speed": 531.1906999563078, "vram_gb": 25.573785305023193} {"t": 1790413670.5298114, "kind": "train", "update": 5600, "max_updates": 8000, "epoch": 48, "loss": 0.5655250537395478, "grad_norm": 0.08094687014818192, "lr": 3.3333333333333335e-05, "sec_per_update": 0.35226988792419434, "speed": 531.9829173682948, "vram_gb": 25.573785305023193} {"t": 1790413680.4161086, "kind": "train", "update": 5625, "max_updates": 8000, "epoch": 48, "loss": 0.5886497378349305, "grad_norm": 0.04057919606566429, "lr": 3.2986111111111115e-05, "sec_per_update": 0.3954527187347412, "speed": 532.8202792742089, "vram_gb": 25.573785305023193} {"t": 1790413689.2018023, "kind": "train", "update": 5650, "max_updates": 8000, "epoch": 48, "loss": 0.5908509039878845, "grad_norm": 0.06724292039871216, "lr": 3.263888888888889e-05, "sec_per_update": 0.3512259292602539, "speed": 533.4999313490349, "vram_gb": 25.573785305023193} {"t": 1790413698.3557668, "kind": "train", "update": 5675, "max_updates": 8000, "epoch": 48, "loss": 0.6174918055534363, "grad_norm": 0.18049748241901398, "lr": 3.229166666666667e-05, "sec_per_update": 0.3663618946075439, "speed": 534.2897616115742, "vram_gb": 25.573785305023193} {"t": 1790413709.0572178, "kind": "train", "update": 5700, "max_updates": 8000, "epoch": 49, "loss": 0.5728487730026245, "grad_norm": 0.04632279649376869, "lr": 3.194444444444444e-05, "sec_per_update": 0.4280611801147461, "speed": 534.8530408272466, "vram_gb": 25.573785305023193} {"t": 1790413719.1677632, "kind": "train", "update": 5725, "max_updates": 8000, "epoch": 49, "loss": 0.5806352746486664, "grad_norm": 0.05672949552536011, "lr": 3.159722222222222e-05, "sec_per_update": 0.40442121505737305, "speed": 535.2997396092796, "vram_gb": 25.573785305023193} {"t": 1790413727.3230422, "kind": "train", "update": 5750, "max_updates": 8000, "epoch": 49, "loss": 0.5730978083610535, "grad_norm": 0.08685867488384247, "lr": 3.125e-05, "sec_per_update": 0.32618959426879884, "speed": 536.1937559395982, "vram_gb": 25.573785305023193} {"t": 1790413736.975029, "kind": "train", "update": 5775, "max_updates": 8000, "epoch": 49, "loss": 0.590560028553009, "grad_norm": 0.14578497409820557, "lr": 3.090277777777778e-05, "sec_per_update": 0.38610058784484863, "speed": 536.9953064126789, "vram_gb": 25.573785305023193} {"t": 1790413746.7996943, "kind": "train", "update": 5800, "max_updates": 8000, "epoch": 49, "loss": 0.5970869016647339, "grad_norm": 0.07363889366388321, "lr": 3.055555555555556e-05, "sec_per_update": 0.3929838562011719, "speed": 537.7166697463532, "vram_gb": 25.573785305023193} {"t": 1790413755.294226, "kind": "train", "update": 5825, "max_updates": 8000, "epoch": 50, "loss": 0.5951080298423768, "grad_norm": 0.12887094914913177, "lr": 3.0208333333333334e-05, "sec_per_update": 0.3397576332092285, "speed": 538.3629730012392, "vram_gb": 25.573785305023193} {"t": 1790413764.199514, "kind": "train", "update": 5850, "max_updates": 8000, "epoch": 50, "loss": 0.6203608894348145, "grad_norm": 0.04077449440956116, "lr": 2.9861111111111113e-05, "sec_per_update": 0.3562100315093994, "speed": 539.3151033447888, "vram_gb": 25.573785305023193} {"t": 1790413772.8971136, "kind": "train", "update": 5875, "max_updates": 8000, "epoch": 50, "loss": 0.6050035643577576, "grad_norm": 0.1123705580830574, "lr": 2.951388888888889e-05, "sec_per_update": 0.34792633056640626, "speed": 540.0154223920315, "vram_gb": 25.573785305023193} {"t": 1790413781.4400742, "kind": "train", "update": 5900, "max_updates": 8000, "epoch": 50, "loss": 0.6060620319843292, "grad_norm": 0.17694929242134094, "lr": 2.916666666666667e-05, "sec_per_update": 0.341701545715332, "speed": 540.7859127983032, "vram_gb": 25.573785305023193} {"t": 1790413791.8517303, "kind": "train", "update": 5925, "max_updates": 8000, "epoch": 51, "loss": 0.6229052114486694, "grad_norm": 0.1097964271903038, "lr": 2.8819444444444443e-05, "sec_per_update": 0.41628570556640626, "speed": 541.5578706129132, "vram_gb": 25.573785305023193} {"t": 1790413800.8946342, "kind": "train", "update": 5950, "max_updates": 8000, "epoch": 51, "loss": 0.626915591955185, "grad_norm": 0.06941532343626022, "lr": 2.8472222222222223e-05, "sec_per_update": 0.3619157600402832, "speed": 542.4689327954832, "vram_gb": 25.573785305023193} {"t": 1790413809.4019158, "kind": "train", "update": 5975, "max_updates": 8000, "epoch": 51, "loss": 0.5957850444316865, "grad_norm": 0.05419963598251343, "lr": 2.8125000000000003e-05, "sec_per_update": 0.34029786109924315, "speed": 543.2340742681662, "vram_gb": 25.573785305023193} {"t": 1790413818.3486385, "kind": "train", "update": 6000, "max_updates": 8000, "epoch": 51, "loss": 0.5815680468082428, "grad_norm": 0.09819923341274261, "lr": 2.777777777777778e-05, "sec_per_update": 0.35767578125, "speed": 543.8231855598261, "vram_gb": 25.573785305023193} {"t": 1790413819.9576788, "kind": "val", "update": 6000, "val_loss": 0.6181407769521078} {"t": 1790413833.5157897, "kind": "checkpoint", "update": 6000} {"t": 1790413843.0716372, "kind": "train", "update": 6025, "max_updates": 8000, "epoch": 51, "loss": 0.5724575781822204, "grad_norm": 0.03383361175656319, "lr": 2.743055555555556e-05, "sec_per_update": 0.9891087627410888, "speed": 542.0518568031788, "vram_gb": 25.573785305023193} {"t": 1790413852.5934336, "kind": "train", "update": 6050, "max_updates": 8000, "epoch": 52, "loss": 0.5697406494617462, "grad_norm": 0.15754467248916626, "lr": 2.7083333333333332e-05, "sec_per_update": 0.3806879425048828, "speed": 542.6493406293928, "vram_gb": 25.573785305023193} {"t": 1790413861.9655495, "kind": "train", "update": 6075, "max_updates": 8000, "epoch": 52, "loss": 0.5791209924221039, "grad_norm": 0.04005102068185806, "lr": 2.6736111111111112e-05, "sec_per_update": 0.3750295352935791, "speed": 543.3338829168133, "vram_gb": 25.573785305023193} {"t": 1790413870.150614, "kind": "train", "update": 6100, "max_updates": 8000, "epoch": 52, "loss": 0.6116561841964722, "grad_norm": 0.04260818660259247, "lr": 2.6388888888888892e-05, "sec_per_update": 0.32744014739990235, "speed": 544.1807822990767, "vram_gb": 25.573785305023193} {"t": 1790413879.2000334, "kind": "train", "update": 6125, "max_updates": 8000, "epoch": 52, "loss": 0.5784101259708404, "grad_norm": 0.023541131988167763, "lr": 2.604166666666667e-05, "sec_per_update": 0.36197873115539553, "speed": 544.9161838047156, "vram_gb": 25.573785305023193} {"t": 1790413889.1737165, "kind": "train", "update": 6150, "max_updates": 8000, "epoch": 53, "loss": 0.5890005469322205, "grad_norm": 0.15977266430854797, "lr": 2.5694444444444445e-05, "sec_per_update": 0.3989235210418701, "speed": 545.5225251358595, "vram_gb": 25.573785305023193} {"t": 1790413897.2497492, "kind": "train", "update": 6175, "max_updates": 8000, "epoch": 53, "loss": 0.5920079040527344, "grad_norm": 0.05769114941358566, "lr": 2.534722222222222e-05, "sec_per_update": 0.32306464195251466, "speed": 546.3458758830662, "vram_gb": 25.573785305023193} {"t": 1790413905.9006352, "kind": "train", "update": 6200, "max_updates": 8000, "epoch": 53, "loss": 0.6087512373924255, "grad_norm": 0.06669894605875015, "lr": 2.5e-05, "sec_per_update": 0.3460280704498291, "speed": 547.058285416257, "vram_gb": 25.573785305023193} {"t": 1790413915.2820194, "kind": "train", "update": 6225, "max_updates": 8000, "epoch": 53, "loss": 0.604623327255249, "grad_norm": 0.04544741287827492, "lr": 2.465277777777778e-05, "sec_per_update": 0.3752622699737549, "speed": 547.8284547832274, "vram_gb": 25.573785305023193} {"t": 1790413924.4029138, "kind": "train", "update": 6250, "max_updates": 8000, "epoch": 53, "loss": 0.5910661685466766, "grad_norm": 0.07145274430513382, "lr": 2.4305555555555558e-05, "sec_per_update": 0.3648351764678955, "speed": 548.6050440198553, "vram_gb": 25.573785305023193} {"t": 1790413935.0503068, "kind": "train", "update": 6275, "max_updates": 8000, "epoch": 54, "loss": 0.5948214769363404, "grad_norm": 0.0768839418888092, "lr": 2.3958333333333334e-05, "sec_per_update": 0.4258955192565918, "speed": 549.0120974614202, "vram_gb": 25.573785305023193} {"t": 1790413944.2234924, "kind": "train", "update": 6300, "max_updates": 8000, "epoch": 54, "loss": 0.5914307594299316, "grad_norm": 0.09771620482206345, "lr": 2.361111111111111e-05, "sec_per_update": 0.36689004898071287, "speed": 549.7755900976282, "vram_gb": 25.573785305023193} {"t": 1790413953.2799869, "kind": "train", "update": 6325, "max_updates": 8000, "epoch": 54, "loss": 0.5928710412979126, "grad_norm": 0.038602713495492935, "lr": 2.326388888888889e-05, "sec_per_update": 0.36209200859069823, "speed": 550.6484237034163, "vram_gb": 25.573785305023193} {"t": 1790413961.9863992, "kind": "train", "update": 6350, "max_updates": 8000, "epoch": 54, "loss": 0.6159613418579102, "grad_norm": 0.13822023570537567, "lr": 2.2916666666666667e-05, "sec_per_update": 0.3482749843597412, "speed": 551.2139050797674, "vram_gb": 25.573785305023193} {"t": 1790413971.308413, "kind": "train", "update": 6375, "max_updates": 8000, "epoch": 54, "loss": 0.6006080341339112, "grad_norm": 0.053588367998600006, "lr": 2.2569444444444447e-05, "sec_per_update": 0.37306803703308106, "speed": 551.8363396311379, "vram_gb": 25.573785305023193} {"t": 1790413980.4501772, "kind": "train", "update": 6400, "max_updates": 8000, "epoch": 55, "loss": 0.6002300095558166, "grad_norm": 0.1501763015985489, "lr": 2.2222222222222223e-05, "sec_per_update": 0.36546396255493163, "speed": 552.5315982769064, "vram_gb": 25.573785305023193} {"t": 1790413988.9473598, "kind": "train", "update": 6425, "max_updates": 8000, "epoch": 55, "loss": 0.6099102139472962, "grad_norm": 0.09373430907726288, "lr": 2.1875e-05, "sec_per_update": 0.3400901985168457, "speed": 553.3256221245799, "vram_gb": 25.573785305023193} {"t": 1790413998.487478, "kind": "train", "update": 6450, "max_updates": 8000, "epoch": 55, "loss": 0.6040300011634827, "grad_norm": 0.07440842688083649, "lr": 2.152777777777778e-05, "sec_per_update": 0.3816138935089111, "speed": 554.0393917377806, "vram_gb": 25.573785305023193} {"t": 1790414007.77454, "kind": "train", "update": 6475, "max_updates": 8000, "epoch": 55, "loss": 0.5807465243339539, "grad_norm": 0.08549696952104568, "lr": 2.1180555555555556e-05, "sec_per_update": 0.3714797115325928, "speed": 554.4317957453337, "vram_gb": 25.573785305023193} {"t": 1790414019.2238846, "kind": "train", "update": 6500, "max_updates": 8000, "epoch": 56, "loss": 0.6052289485931397, "grad_norm": 0.061002302914857864, "lr": 2.0833333333333336e-05, "sec_per_update": 0.45796913146972656, "speed": 554.8702212557956, "vram_gb": 25.573785305023193} {"t": 1790414019.879552, "kind": "val", "update": 6500, "val_loss": 0.617749442656835} {"t": 1790414027.9685724, "kind": "train", "update": 6525, "max_updates": 8000, "epoch": 56, "loss": 0.5927794539928436, "grad_norm": 0.17304271459579468, "lr": 2.0486111111111113e-05, "sec_per_update": 0.3497928237915039, "speed": 555.5346913177352, "vram_gb": 25.573785305023193} {"t": 1790414036.9954932, "kind": "train", "update": 6550, "max_updates": 8000, "epoch": 56, "loss": 0.6060429608821869, "grad_norm": 0.10543566942214966, "lr": 2.013888888888889e-05, "sec_per_update": 0.3610781002044678, "speed": 556.214713279312, "vram_gb": 25.573785305023193} {"t": 1790414046.0276716, "kind": "train", "update": 6575, "max_updates": 8000, "epoch": 56, "loss": 0.5931818103790283, "grad_norm": 0.11246868968009949, "lr": 1.9791666666666665e-05, "sec_per_update": 0.3612857437133789, "speed": 556.9018232380066, "vram_gb": 25.573785305023193} {"t": 1790414055.034404, "kind": "train", "update": 6600, "max_updates": 8000, "epoch": 56, "loss": 0.5933521008491516, "grad_norm": 0.08876486867666245, "lr": 1.9444444444444445e-05, "sec_per_update": 0.3602670478820801, "speed": 557.5234984637766, "vram_gb": 25.573785305023193} {"t": 1790414063.6057863, "kind": "train", "update": 6625, "max_updates": 8000, "epoch": 57, "loss": 0.5854306221008301, "grad_norm": 0.08191549777984619, "lr": 1.9097222222222222e-05, "sec_per_update": 0.3428529930114746, "speed": 558.244005923044, "vram_gb": 25.573785305023193} {"t": 1790414073.3757765, "kind": "train", "update": 6650, "max_updates": 8000, "epoch": 57, "loss": 0.6205316138267517, "grad_norm": 0.05424825847148895, "lr": 1.8750000000000002e-05, "sec_per_update": 0.39061546325683594, "speed": 558.78630819884, "vram_gb": 25.573785305023193} {"t": 1790414081.4872723, "kind": "train", "update": 6675, "max_updates": 8000, "epoch": 57, "loss": 0.6167682433128356, "grad_norm": 0.08006912469863892, "lr": 1.8402777777777778e-05, "sec_per_update": 0.32464414596557617, "speed": 559.50074537321, "vram_gb": 25.573785305023193} {"t": 1790414090.2620285, "kind": "train", "update": 6700, "max_updates": 8000, "epoch": 57, "loss": 0.615933449268341, "grad_norm": 0.021656407043337822, "lr": 1.8055555555555555e-05, "sec_per_update": 0.3509946155548096, "speed": 560.1451334949461, "vram_gb": 25.573785305023193} {"t": 1790414099.6801577, "kind": "train", "update": 6725, "max_updates": 8000, "epoch": 57, "loss": 0.582102255821228, "grad_norm": 0.04404151439666748, "lr": 1.7708333333333335e-05, "sec_per_update": 0.3767180061340332, "speed": 560.8678470152945, "vram_gb": 25.573785305023193} {"t": 1790414109.0040665, "kind": "train", "update": 6750, "max_updates": 8000, "epoch": 58, "loss": 0.5916840302944183, "grad_norm": 0.055403806269168854, "lr": 1.736111111111111e-05, "sec_per_update": 0.3729558944702148, "speed": 561.4023324163717, "vram_gb": 25.573785305023193} {"t": 1790414117.3316607, "kind": "train", "update": 6775, "max_updates": 8000, "epoch": 58, "loss": 0.6031506633758545, "grad_norm": 0.08582417666912079, "lr": 1.701388888888889e-05, "sec_per_update": 0.3329234504699707, "speed": 562.1216511872836, "vram_gb": 25.573785305023193} {"t": 1790414126.53465, "kind": "train", "update": 6800, "max_updates": 8000, "epoch": 58, "loss": 0.5907380270957947, "grad_norm": 0.07936902344226837, "lr": 1.6666666666666667e-05, "sec_per_update": 0.3683056449890137, "speed": 562.8040056232359, "vram_gb": 25.573785305023193} {"t": 1790414136.1135526, "kind": "train", "update": 6825, "max_updates": 8000, "epoch": 58, "loss": 0.5926194119453431, "grad_norm": 0.10760452598333359, "lr": 1.6319444444444444e-05, "sec_per_update": 0.3831519794464111, "speed": 563.2745261850669, "vram_gb": 25.573785305023193} {"t": 1790414145.7121813, "kind": "train", "update": 6850, "max_updates": 8000, "epoch": 59, "loss": 0.6253035020828247, "grad_norm": 0.1888289600610733, "lr": 1.597222222222222e-05, "sec_per_update": 0.3839457893371582, "speed": 563.8524325978606, "vram_gb": 25.573785305023193} {"t": 1790414154.1509593, "kind": "train", "update": 6875, "max_updates": 8000, "epoch": 59, "loss": 0.5920596933364868, "grad_norm": 0.08370193094015121, "lr": 1.5625e-05, "sec_per_update": 0.33735530853271484, "speed": 564.605789527416, "vram_gb": 25.573785305023193} {"t": 1790414163.0516047, "kind": "train", "update": 6900, "max_updates": 8000, "epoch": 59, "loss": 0.5885671877861023, "grad_norm": 0.053709711879491806, "lr": 1.527777777777778e-05, "sec_per_update": 0.356221170425415, "speed": 565.19034848535, "vram_gb": 25.573785305023193} {"t": 1790414172.7238767, "kind": "train", "update": 6925, "max_updates": 8000, "epoch": 59, "loss": 0.6049936628341674, "grad_norm": 0.21436066925525665, "lr": 1.4930555555555557e-05, "sec_per_update": 0.3867088508605957, "speed": 565.7320336657303, "vram_gb": 25.573785305023193} {"t": 1790414182.0840352, "kind": "train", "update": 6950, "max_updates": 8000, "epoch": 59, "loss": 0.6290047430992126, "grad_norm": 0.1304701417684555, "lr": 1.4583333333333335e-05, "sec_per_update": 0.37453564643859866, "speed": 566.3211852379928, "vram_gb": 25.573785305023193} {"t": 1790414190.6994953, "kind": "train", "update": 6975, "max_updates": 8000, "epoch": 60, "loss": 0.5974596107006073, "grad_norm": 0.03376231715083122, "lr": 1.4236111111111111e-05, "sec_per_update": 0.3446777057647705, "speed": 566.9675509858894, "vram_gb": 25.573785305023193} {"t": 1790414199.6853008, "kind": "train", "update": 7000, "max_updates": 8000, "epoch": 60, "loss": 0.601999044418335, "grad_norm": 0.12935268878936768, "lr": 1.388888888888889e-05, "sec_per_update": 0.35943476676940916, "speed": 567.537237915648, "vram_gb": 25.573785305023193} {"t": 1790414200.3719864, "kind": "val", "update": 7000, "val_loss": 0.6174406011899313} {"t": 1790414214.7488823, "kind": "checkpoint", "update": 7000} {"t": 1790414223.8023596, "kind": "train", "update": 7025, "max_updates": 8000, "epoch": 60, "loss": 0.5816021800041199, "grad_norm": 0.0967046469449997, "lr": 1.3541666666666666e-05, "sec_per_update": 0.9646739673614502, "speed": 565.7834800894889, "vram_gb": 25.573785305023193} {"t": 1790414232.6488738, "kind": "train", "update": 7050, "max_updates": 8000, "epoch": 60, "loss": 0.5901495492458344, "grad_norm": 0.16983751952648163, "lr": 1.3194444444444446e-05, "sec_per_update": 0.35386101722717284, "speed": 566.4061586345013, "vram_gb": 25.573785305023193} {"t": 1790414241.3647916, "kind": "train", "update": 7075, "max_updates": 8000, "epoch": 60, "loss": 0.588796923160553, "grad_norm": 0.06506264209747314, "lr": 1.2847222222222222e-05, "sec_per_update": 0.34863672256469724, "speed": 567.1755412592898, "vram_gb": 25.573785305023193} {"t": 1790414251.8574214, "kind": "train", "update": 7100, "max_updates": 8000, "epoch": 61, "loss": 0.5980820488929749, "grad_norm": 0.039547935128211975, "lr": 1.25e-05, "sec_per_update": 0.419704008102417, "speed": 567.6295750767175, "vram_gb": 25.573785305023193} {"t": 1790414260.4607096, "kind": "train", "update": 7125, "max_updates": 8000, "epoch": 61, "loss": 0.5855468082427978, "grad_norm": 0.09741643071174622, "lr": 1.2152777777777779e-05, "sec_per_update": 0.3441163921356201, "speed": 568.1523352251063, "vram_gb": 25.573785305023193} {"t": 1790414268.717143, "kind": "train", "update": 7150, "max_updates": 8000, "epoch": 61, "loss": 0.5886214113235474, "grad_norm": 0.06308828294277191, "lr": 1.1805555555555555e-05, "sec_per_update": 0.33027474403381346, "speed": 568.8466879933212, "vram_gb": 25.573785305023193} {"t": 1790414276.6953118, "kind": "train", "update": 7175, "max_updates": 8000, "epoch": 61, "loss": 0.5926775193214416, "grad_norm": 0.050767503678798676, "lr": 1.1458333333333333e-05, "sec_per_update": 0.3191243362426758, "speed": 569.5170064531782, "vram_gb": 25.573785305023193} {"t": 1790414287.0779293, "kind": "train", "update": 7200, "max_updates": 8000, "epoch": 62, "loss": 0.596670137643814, "grad_norm": 0.13771679997444153, "lr": 1.1111111111111112e-05, "sec_per_update": 0.4153098964691162, "speed": 570.0513774148808, "vram_gb": 25.573785305023193} {"t": 1790414296.5311604, "kind": "train", "update": 7225, "max_updates": 8000, "epoch": 62, "loss": 0.5927868258953094, "grad_norm": 0.07002386450767517, "lr": 1.076388888888889e-05, "sec_per_update": 0.3781271839141846, "speed": 570.6795403167433, "vram_gb": 25.573785305023193} {"t": 1790414304.5854666, "kind": "train", "update": 7250, "max_updates": 8000, "epoch": 62, "loss": 0.602108027935028, "grad_norm": 0.06305194646120071, "lr": 1.0416666666666668e-05, "sec_per_update": 0.3221448516845703, "speed": 571.3505862975264, "vram_gb": 25.573785305023193} {"t": 1790414312.990234, "kind": "train", "update": 7275, "max_updates": 8000, "epoch": 62, "loss": 0.5998900294303894, "grad_norm": 0.0482589453458786, "lr": 1.0069444444444445e-05, "sec_per_update": 0.33603928565979, "speed": 572.0171166670656, "vram_gb": 25.573785305023193} {"t": 1790414322.3423676, "kind": "train", "update": 7300, "max_updates": 8000, "epoch": 62, "loss": 0.629984085559845, "grad_norm": 0.11577985435724258, "lr": 9.722222222222223e-06, "sec_per_update": 0.37425914764404294, "speed": 572.5042976681601, "vram_gb": 25.573785305023193} {"t": 1790414331.464733, "kind": "train", "update": 7325, "max_updates": 8000, "epoch": 63, "loss": 0.5913032174110413, "grad_norm": 0.0598689466714859, "lr": 9.375000000000001e-06, "sec_per_update": 0.36474609375, "speed": 573.0063123815311, "vram_gb": 25.573785305023193} {"t": 1790414339.4401524, "kind": "train", "update": 7350, "max_updates": 8000, "epoch": 63, "loss": 0.6146734309196472, "grad_norm": 0.041429851204156876, "lr": 9.027777777777777e-06, "sec_per_update": 0.31916510581970214, "speed": 573.6624038168494, "vram_gb": 25.573785305023193} {"t": 1790414348.420253, "kind": "train", "update": 7375, "max_updates": 8000, "epoch": 63, "loss": 0.5932638597488403, "grad_norm": 0.028057558462023735, "lr": 8.680555555555556e-06, "sec_per_update": 0.35920086860656736, "speed": 574.2596894232884, "vram_gb": 25.573785305023193} {"t": 1790414358.4067497, "kind": "train", "update": 7400, "max_updates": 8000, "epoch": 63, "loss": 0.5877576875686645, "grad_norm": 0.07689822465181351, "lr": 8.333333333333334e-06, "sec_per_update": 0.3992439556121826, "speed": 574.7386327833158, "vram_gb": 25.573785305023193} {"t": 1790414367.8913376, "kind": "train", "update": 7425, "max_updates": 8000, "epoch": 64, "loss": 0.5852125811576844, "grad_norm": 0.052736274898052216, "lr": 7.98611111111111e-06, "sec_per_update": 0.37957708358764647, "speed": 575.2858874724834, "vram_gb": 25.573785305023193} {"t": 1790414375.454061, "kind": "train", "update": 7450, "max_updates": 8000, "epoch": 64, "loss": 0.5931089556217194, "grad_norm": 0.09108671545982361, "lr": 7.63888888888889e-06, "sec_per_update": 0.30254307746887205, "speed": 575.8909296395315, "vram_gb": 25.573785305023193} {"t": 1790414385.4344695, "kind": "train", "update": 7475, "max_updates": 8000, "epoch": 64, "loss": 0.6149545168876648, "grad_norm": 0.07197306305170059, "lr": 7.2916666666666674e-06, "sec_per_update": 0.3992152500152588, "speed": 576.42525901528, "vram_gb": 25.573785305023193} {"t": 1790414394.7507162, "kind": "train", "update": 7500, "max_updates": 8000, "epoch": 64, "loss": 0.5768891835212707, "grad_norm": 0.1397637128829956, "lr": 6.944444444444445e-06, "sec_per_update": 0.372452974319458, "speed": 576.8541029905093, "vram_gb": 25.573785305023193} {"t": 1790414396.5057309, "kind": "val", "update": 7500, "val_loss": 0.6172564228375753} {"t": 1790414406.2143617, "kind": "train", "update": 7525, "max_updates": 8000, "epoch": 64, "loss": 0.6065577387809753, "grad_norm": 0.18302437663078308, "lr": 6.597222222222223e-06, "sec_per_update": 0.45873820304870605, "speed": 577.1915965633076, "vram_gb": 25.573785305023193} {"t": 1790414416.2546217, "kind": "train", "update": 7550, "max_updates": 8000, "epoch": 65, "loss": 0.6134380960464477, "grad_norm": 0.05556856840848923, "lr": 6.25e-06, "sec_per_update": 0.4016182613372803, "speed": 577.6332984649245, "vram_gb": 25.573785305023193} {"t": 1790414425.042983, "kind": "train", "update": 7575, "max_updates": 8000, "epoch": 65, "loss": 0.5953474617004395, "grad_norm": 0.13713113963603973, "lr": 5.902777777777778e-06, "sec_per_update": 0.351526517868042, "speed": 578.1686911982835, "vram_gb": 25.573785305023193} {"t": 1790414433.846845, "kind": "train", "update": 7600, "max_updates": 8000, "epoch": 65, "loss": 0.5880022764205932, "grad_norm": 0.11069236695766449, "lr": 5.555555555555556e-06, "sec_per_update": 0.352159366607666, "speed": 578.7371193952647, "vram_gb": 25.573785305023193} {"t": 1790414443.0690181, "kind": "train", "update": 7625, "max_updates": 8000, "epoch": 65, "loss": 0.6138472390174866, "grad_norm": 0.10537399351596832, "lr": 5.208333333333334e-06, "sec_per_update": 0.3688806056976318, "speed": 579.213324061693, "vram_gb": 25.573785305023193} {"t": 1790414452.4513383, "kind": "train", "update": 7650, "max_updates": 8000, "epoch": 65, "loss": 0.5813142192363739, "grad_norm": 0.0801844373345375, "lr": 4.861111111111111e-06, "sec_per_update": 0.3752998638153076, "speed": 579.6778604560666, "vram_gb": 25.573785305023193} {"t": 1790414462.6036763, "kind": "train", "update": 7675, "max_updates": 8000, "epoch": 66, "loss": 0.5722175216674805, "grad_norm": 0.16084276139736176, "lr": 4.513888888888889e-06, "sec_per_update": 0.40608853340148926, "speed": 580.1094255367169, "vram_gb": 25.573785305023193} {"t": 1790414471.409319, "kind": "train", "update": 7700, "max_updates": 8000, "epoch": 66, "loss": 0.5725356435775757, "grad_norm": 0.11287423223257065, "lr": 4.166666666666667e-06, "sec_per_update": 0.35222151756286624, "speed": 580.6272286458789, "vram_gb": 25.573785305023193} {"t": 1790414480.2327342, "kind": "train", "update": 7725, "max_updates": 8000, "epoch": 66, "loss": 0.5819069516658782, "grad_norm": 0.05712136626243591, "lr": 3.819444444444445e-06, "sec_per_update": 0.35291252136230467, "speed": 581.1723559698081, "vram_gb": 25.573785305023193} {"t": 1790414489.2419755, "kind": "train", "update": 7750, "max_updates": 8000, "epoch": 66, "loss": 0.6016232562065125, "grad_norm": 0.0383242629468441, "lr": 3.4722222222222224e-06, "sec_per_update": 0.36037339210510255, "speed": 581.7417229961403, "vram_gb": 25.573785305023193} {"t": 1790414497.664311, "kind": "train", "update": 7775, "max_updates": 8000, "epoch": 67, "loss": 0.5867641413211823, "grad_norm": 0.2023821324110031, "lr": 3.125e-06, "sec_per_update": 0.3368945503234863, "speed": 582.3328827649236, "vram_gb": 25.573785305023193} {"t": 1790414507.9751024, "kind": "train", "update": 7800, "max_updates": 8000, "epoch": 67, "loss": 0.6163804376125336, "grad_norm": 0.19195827841758728, "lr": 2.777777777777778e-06, "sec_per_update": 0.41243711471557615, "speed": 582.8282633916963, "vram_gb": 25.573785305023193} {"t": 1790414517.0351179, "kind": "train", "update": 7825, "max_updates": 8000, "epoch": 67, "loss": 0.5933734464645386, "grad_norm": 0.10967369377613068, "lr": 2.4305555555555557e-06, "sec_per_update": 0.36241650581359863, "speed": 583.3887698816478, "vram_gb": 25.573785305023193} {"t": 1790414525.2971659, "kind": "train", "update": 7850, "max_updates": 8000, "epoch": 67, "loss": 0.5872143912315368, "grad_norm": 0.08899804949760437, "lr": 2.0833333333333334e-06, "sec_per_update": 0.33048203468322757, "speed": 584.0113588084362, "vram_gb": 25.573785305023193} {"t": 1790414534.0307782, "kind": "train", "update": 7875, "max_updates": 8000, "epoch": 67, "loss": 0.5968269371986389, "grad_norm": 0.05752637982368469, "lr": 1.7361111111111112e-06, "sec_per_update": 0.3493461513519287, "speed": 584.5061808466398, "vram_gb": 25.573785305023193} {"t": 1790414543.4706988, "kind": "train", "update": 7900, "max_updates": 8000, "epoch": 68, "loss": 0.5943387830257416, "grad_norm": 0.028529690578579903, "lr": 1.388888888888889e-06, "sec_per_update": 0.37758039474487304, "speed": 584.9423312878404, "vram_gb": 25.573785305023193} {"t": 1790414551.622369, "kind": "train", "update": 7925, "max_updates": 8000, "epoch": 68, "loss": 0.5934077835083008, "grad_norm": 0.13988763093948364, "lr": 1.0416666666666667e-06, "sec_per_update": 0.32608202934265135, "speed": 585.5431589333283, "vram_gb": 25.573785305023193} {"t": 1790414560.622802, "kind": "train", "update": 7950, "max_updates": 8000, "epoch": 68, "loss": 0.6221013677120208, "grad_norm": 0.15350466966629028, "lr": 6.944444444444445e-07, "sec_per_update": 0.3599900245666504, "speed": 586.0478294541587, "vram_gb": 25.573785305023193} {"t": 1790414569.9258945, "kind": "train", "update": 7975, "max_updates": 8000, "epoch": 68, "loss": 0.5867332971096039, "grad_norm": 0.02815970405936241, "lr": 3.4722222222222224e-07, "sec_per_update": 0.3721506309509277, "speed": 586.5497616115124, "vram_gb": 25.573785305023193} {"t": 1790414579.3635693, "kind": "train", "update": 8000, "max_updates": 8000, "epoch": 68, "loss": 0.5862796068191528, "grad_norm": 0.1042901948094368, "lr": 0.0, "sec_per_update": 0.3773200225830078, "speed": 586.9475735486755, "vram_gb": 25.573785305023193} {"t": 1790414580.4806445, "kind": "val", "update": 8000, "val_loss": 0.6173214813073477} {"t": 1790414593.9680555, "kind": "checkpoint", "update": 8000} {"t": 1790414593.9683924, "kind": "done", "update": 8000}