| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 87356, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0011447410595723248, | |
| "grad_norm": 0.08282724022865295, | |
| "learning_rate": 2.8332665560070978e-08, | |
| "loss": 2.747, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.0022894821191446497, | |
| "grad_norm": 0.09139522910118103, | |
| "learning_rate": 5.6665331120141957e-08, | |
| "loss": 2.7153, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.003434223178716974, | |
| "grad_norm": 0.14882275462150574, | |
| "learning_rate": 8.528418522122374e-08, | |
| "loss": 2.7207, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.004578964238289299, | |
| "grad_norm": 0.15320613980293274, | |
| "learning_rate": 1.1390303932230555e-07, | |
| "loss": 2.7451, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.005723705297861623, | |
| "grad_norm": 0.10282430797815323, | |
| "learning_rate": 1.4223570488237652e-07, | |
| "loss": 2.7131, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.006868446357433948, | |
| "grad_norm": 0.14666467905044556, | |
| "learning_rate": 1.7085455898345834e-07, | |
| "loss": 2.7244, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.008013187417006273, | |
| "grad_norm": 0.13914579153060913, | |
| "learning_rate": 1.994734130845401e-07, | |
| "loss": 2.7248, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.009157928476578599, | |
| "grad_norm": 0.13069643080234528, | |
| "learning_rate": 2.278060786446111e-07, | |
| "loss": 2.7237, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.010302669536150923, | |
| "grad_norm": 0.11404985189437866, | |
| "learning_rate": 2.564249327456929e-07, | |
| "loss": 2.7372, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.011447410595723247, | |
| "grad_norm": 0.2278125137090683, | |
| "learning_rate": 2.8504378684677464e-07, | |
| "loss": 2.7491, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.012592151655295572, | |
| "grad_norm": 0.19849324226379395, | |
| "learning_rate": 3.1366264094785646e-07, | |
| "loss": 2.7059, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.013736892714867896, | |
| "grad_norm": 0.21892410516738892, | |
| "learning_rate": 3.422814950489383e-07, | |
| "loss": 2.7291, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.014881633774440222, | |
| "grad_norm": 0.09724584966897964, | |
| "learning_rate": 3.709003491500201e-07, | |
| "loss": 2.7131, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.016026374834012546, | |
| "grad_norm": 0.13547281920909882, | |
| "learning_rate": 3.9923301471009105e-07, | |
| "loss": 2.6872, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.01717111589358487, | |
| "grad_norm": 0.21319937705993652, | |
| "learning_rate": 4.278518688111728e-07, | |
| "loss": 2.7306, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.018315856953157197, | |
| "grad_norm": 0.15332601964473724, | |
| "learning_rate": 4.564707229122546e-07, | |
| "loss": 2.6786, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.01946059801272952, | |
| "grad_norm": 0.13409346342086792, | |
| "learning_rate": 4.850895770133364e-07, | |
| "loss": 2.7021, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.020605339072301845, | |
| "grad_norm": 0.27683642506599426, | |
| "learning_rate": 5.137084311144182e-07, | |
| "loss": 2.7098, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.02175008013187417, | |
| "grad_norm": 0.2598477900028229, | |
| "learning_rate": 5.423272852155001e-07, | |
| "loss": 2.6628, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.022894821191446493, | |
| "grad_norm": 0.2136494219303131, | |
| "learning_rate": 5.709461393165818e-07, | |
| "loss": 2.6882, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.02403956225101882, | |
| "grad_norm": 0.1683170348405838, | |
| "learning_rate": 5.995649934176636e-07, | |
| "loss": 2.6787, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.025184303310591145, | |
| "grad_norm": 0.12898527085781097, | |
| "learning_rate": 6.281838475187455e-07, | |
| "loss": 2.6507, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.02632904437016347, | |
| "grad_norm": 0.25819605588912964, | |
| "learning_rate": 6.568027016198272e-07, | |
| "loss": 2.6485, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.027473785429735793, | |
| "grad_norm": 0.26418036222457886, | |
| "learning_rate": 6.85421555720909e-07, | |
| "loss": 2.665, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.02861852648930812, | |
| "grad_norm": 0.24633832275867462, | |
| "learning_rate": 7.140404098219908e-07, | |
| "loss": 2.6122, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.029763267548880444, | |
| "grad_norm": 0.11409879475831985, | |
| "learning_rate": 7.426592639230726e-07, | |
| "loss": 2.61, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.030908008608452768, | |
| "grad_norm": 0.18866164982318878, | |
| "learning_rate": 7.712781180241544e-07, | |
| "loss": 2.6109, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.03205274966802509, | |
| "grad_norm": 0.2056051343679428, | |
| "learning_rate": 7.998969721252361e-07, | |
| "loss": 2.6274, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.033197490727597416, | |
| "grad_norm": 0.18269231915473938, | |
| "learning_rate": 8.285158262263179e-07, | |
| "loss": 2.6012, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.03434223178716974, | |
| "grad_norm": 0.17360204458236694, | |
| "learning_rate": 8.571346803273998e-07, | |
| "loss": 2.6238, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.035486972846742064, | |
| "grad_norm": 0.1858297735452652, | |
| "learning_rate": 8.857535344284815e-07, | |
| "loss": 2.5836, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.036631713906314395, | |
| "grad_norm": 0.18415091931819916, | |
| "learning_rate": 9.143723885295633e-07, | |
| "loss": 2.5919, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.03777645496588672, | |
| "grad_norm": 0.19720780849456787, | |
| "learning_rate": 9.429912426306452e-07, | |
| "loss": 2.5742, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.03892119602545904, | |
| "grad_norm": 0.21820122003555298, | |
| "learning_rate": 9.71323908190716e-07, | |
| "loss": 2.5653, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.04006593708503137, | |
| "grad_norm": 0.2850651741027832, | |
| "learning_rate": 9.999427622917978e-07, | |
| "loss": 2.5936, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.04121067814460369, | |
| "grad_norm": 0.19742882251739502, | |
| "learning_rate": 1.028275427851869e-06, | |
| "loss": 2.5751, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.042355419204176015, | |
| "grad_norm": 0.1881546527147293, | |
| "learning_rate": 1.0568942819529507e-06, | |
| "loss": 2.569, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.04350016026374834, | |
| "grad_norm": 0.16125181317329407, | |
| "learning_rate": 1.0855131360540325e-06, | |
| "loss": 2.5522, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.04464490132332066, | |
| "grad_norm": 0.17413508892059326, | |
| "learning_rate": 1.1141319901551142e-06, | |
| "loss": 2.5577, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.04578964238289299, | |
| "grad_norm": 0.36808159947395325, | |
| "learning_rate": 1.142750844256196e-06, | |
| "loss": 2.541, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.04693438344246532, | |
| "grad_norm": 0.2387821078300476, | |
| "learning_rate": 1.1713696983572778e-06, | |
| "loss": 2.5481, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 0.04807912450203764, | |
| "grad_norm": 0.28203991055488586, | |
| "learning_rate": 1.1999885524583595e-06, | |
| "loss": 2.5099, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.049223865561609965, | |
| "grad_norm": 0.36418649554252625, | |
| "learning_rate": 1.2286074065594415e-06, | |
| "loss": 2.5158, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 0.05036860662118229, | |
| "grad_norm": 0.24819067120552063, | |
| "learning_rate": 1.2572262606605233e-06, | |
| "loss": 2.5189, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.05151334768075461, | |
| "grad_norm": 0.24554955959320068, | |
| "learning_rate": 1.285845114761605e-06, | |
| "loss": 2.5275, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.05265808874032694, | |
| "grad_norm": 0.3186652660369873, | |
| "learning_rate": 1.3144639688626868e-06, | |
| "loss": 2.5156, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.05380282979989926, | |
| "grad_norm": 0.2992977797985077, | |
| "learning_rate": 1.3430828229637685e-06, | |
| "loss": 2.5009, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 0.054947570859471585, | |
| "grad_norm": 0.3505355715751648, | |
| "learning_rate": 1.3717016770648503e-06, | |
| "loss": 2.4923, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.05609231191904391, | |
| "grad_norm": 0.17197385430335999, | |
| "learning_rate": 1.4003205311659323e-06, | |
| "loss": 2.4825, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 0.05723705297861624, | |
| "grad_norm": 0.21533966064453125, | |
| "learning_rate": 1.4289393852670142e-06, | |
| "loss": 2.484, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.058381794038188564, | |
| "grad_norm": 0.17859208583831787, | |
| "learning_rate": 1.457558239368096e-06, | |
| "loss": 2.4734, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 0.05952653509776089, | |
| "grad_norm": 0.36880823969841003, | |
| "learning_rate": 1.4861770934691778e-06, | |
| "loss": 2.4747, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 0.06067127615733321, | |
| "grad_norm": 0.26152077317237854, | |
| "learning_rate": 1.5147959475702595e-06, | |
| "loss": 2.4696, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 0.061816017216905536, | |
| "grad_norm": 0.37325313687324524, | |
| "learning_rate": 1.5434148016713413e-06, | |
| "loss": 2.4908, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 0.06296075827647786, | |
| "grad_norm": 0.20671696960926056, | |
| "learning_rate": 1.572033655772423e-06, | |
| "loss": 2.4929, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.06410549933605018, | |
| "grad_norm": 0.17978721857070923, | |
| "learning_rate": 1.6006525098735048e-06, | |
| "loss": 2.4557, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 0.06525024039562251, | |
| "grad_norm": 0.17210538685321808, | |
| "learning_rate": 1.6292713639745866e-06, | |
| "loss": 2.4646, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 0.06639498145519483, | |
| "grad_norm": 0.17783130705356598, | |
| "learning_rate": 1.6578902180756683e-06, | |
| "loss": 2.4563, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 0.06753972251476716, | |
| "grad_norm": 0.16413679718971252, | |
| "learning_rate": 1.68650907217675e-06, | |
| "loss": 2.4823, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 0.06868446357433948, | |
| "grad_norm": 0.1958412379026413, | |
| "learning_rate": 1.715127926277832e-06, | |
| "loss": 2.4733, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.0698292046339118, | |
| "grad_norm": 0.20609253644943237, | |
| "learning_rate": 1.7437467803789138e-06, | |
| "loss": 2.4613, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 0.07097394569348413, | |
| "grad_norm": 0.292453795671463, | |
| "learning_rate": 1.7723656344799956e-06, | |
| "loss": 2.4585, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 0.07211868675305645, | |
| "grad_norm": 0.22299447655677795, | |
| "learning_rate": 1.8009844885810774e-06, | |
| "loss": 2.4572, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 0.07326342781262879, | |
| "grad_norm": 0.18281777203083038, | |
| "learning_rate": 1.8296033426821591e-06, | |
| "loss": 2.4511, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 0.07440816887220111, | |
| "grad_norm": 0.19673572480678558, | |
| "learning_rate": 1.8582221967832409e-06, | |
| "loss": 2.4377, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.07555290993177344, | |
| "grad_norm": 0.2068740874528885, | |
| "learning_rate": 1.8868410508843226e-06, | |
| "loss": 2.4483, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 0.07669765099134576, | |
| "grad_norm": 0.22345593571662903, | |
| "learning_rate": 1.9154599049854046e-06, | |
| "loss": 2.4708, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 0.07784239205091809, | |
| "grad_norm": 0.37216469645500183, | |
| "learning_rate": 1.944078759086486e-06, | |
| "loss": 2.453, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 0.07898713311049041, | |
| "grad_norm": 0.20201215147972107, | |
| "learning_rate": 1.972697613187568e-06, | |
| "loss": 2.4536, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 0.08013187417006273, | |
| "grad_norm": 0.32299327850341797, | |
| "learning_rate": 2.0013164672886497e-06, | |
| "loss": 2.4464, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.08127661522963506, | |
| "grad_norm": 0.2131340205669403, | |
| "learning_rate": 2.0299353213897317e-06, | |
| "loss": 2.4699, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 0.08242135628920738, | |
| "grad_norm": 0.2809932231903076, | |
| "learning_rate": 2.0585541754908132e-06, | |
| "loss": 2.4494, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 0.0835660973487797, | |
| "grad_norm": 0.19964805245399475, | |
| "learning_rate": 2.0871730295918956e-06, | |
| "loss": 2.4575, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 0.08471083840835203, | |
| "grad_norm": 0.2731245160102844, | |
| "learning_rate": 2.115791883692977e-06, | |
| "loss": 2.4513, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 0.08585557946792435, | |
| "grad_norm": 0.22060319781303406, | |
| "learning_rate": 2.144410737794059e-06, | |
| "loss": 2.4459, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.08700032052749668, | |
| "grad_norm": 0.23072516918182373, | |
| "learning_rate": 2.1730295918951407e-06, | |
| "loss": 2.4469, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 0.088145061587069, | |
| "grad_norm": 0.2869388163089752, | |
| "learning_rate": 2.2016484459962227e-06, | |
| "loss": 2.426, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 0.08928980264664133, | |
| "grad_norm": 0.4073362648487091, | |
| "learning_rate": 2.230267300097304e-06, | |
| "loss": 2.4399, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 0.09043454370621365, | |
| "grad_norm": 0.3338267207145691, | |
| "learning_rate": 2.258886154198386e-06, | |
| "loss": 2.4442, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 0.09157928476578597, | |
| "grad_norm": 0.32599982619285583, | |
| "learning_rate": 2.2875050082994677e-06, | |
| "loss": 2.448, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.0927240258253583, | |
| "grad_norm": 0.38292455673217773, | |
| "learning_rate": 2.3161238624005497e-06, | |
| "loss": 2.4501, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 0.09386876688493064, | |
| "grad_norm": 0.2262214571237564, | |
| "learning_rate": 2.3447427165016317e-06, | |
| "loss": 2.4306, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 0.09501350794450296, | |
| "grad_norm": 0.23351378738880157, | |
| "learning_rate": 2.3733615706027132e-06, | |
| "loss": 2.4542, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 0.09615824900407528, | |
| "grad_norm": 0.24008594453334808, | |
| "learning_rate": 2.401980424703795e-06, | |
| "loss": 2.4491, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 0.09730299006364761, | |
| "grad_norm": 0.3131836950778961, | |
| "learning_rate": 2.4305992788048767e-06, | |
| "loss": 2.4402, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.09844773112321993, | |
| "grad_norm": 0.3088011145591736, | |
| "learning_rate": 2.4592181329059587e-06, | |
| "loss": 2.4236, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 0.09959247218279225, | |
| "grad_norm": 0.31873244047164917, | |
| "learning_rate": 2.4878369870070403e-06, | |
| "loss": 2.4425, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 0.10073721324236458, | |
| "grad_norm": 0.19671718776226044, | |
| "learning_rate": 2.5164558411081227e-06, | |
| "loss": 2.4277, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 0.1018819543019369, | |
| "grad_norm": 0.3244757056236267, | |
| "learning_rate": 2.5450746952092042e-06, | |
| "loss": 2.4328, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 0.10302669536150923, | |
| "grad_norm": 0.32051828503608704, | |
| "learning_rate": 2.573693549310286e-06, | |
| "loss": 2.4303, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.10417143642108155, | |
| "grad_norm": 0.2557377517223358, | |
| "learning_rate": 2.6023124034113677e-06, | |
| "loss": 2.4426, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 0.10531617748065387, | |
| "grad_norm": 0.29920217394828796, | |
| "learning_rate": 2.6309312575124497e-06, | |
| "loss": 2.4247, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 0.1064609185402262, | |
| "grad_norm": 0.2528887987136841, | |
| "learning_rate": 2.6595501116135313e-06, | |
| "loss": 2.4337, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 0.10760565959979852, | |
| "grad_norm": 0.2672332525253296, | |
| "learning_rate": 2.6881689657146132e-06, | |
| "loss": 2.4327, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 0.10875040065937085, | |
| "grad_norm": 0.26545771956443787, | |
| "learning_rate": 2.7167878198156948e-06, | |
| "loss": 2.411, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.10989514171894317, | |
| "grad_norm": 0.2251209020614624, | |
| "learning_rate": 2.7454066739167768e-06, | |
| "loss": 2.4346, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 0.1110398827785155, | |
| "grad_norm": 0.25982141494750977, | |
| "learning_rate": 2.7740255280178583e-06, | |
| "loss": 2.4155, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 0.11218462383808782, | |
| "grad_norm": 0.2948269844055176, | |
| "learning_rate": 2.8026443821189403e-06, | |
| "loss": 2.4056, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 0.11332936489766014, | |
| "grad_norm": 0.2271622121334076, | |
| "learning_rate": 2.8312632362200223e-06, | |
| "loss": 2.4266, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 0.11447410595723248, | |
| "grad_norm": 0.25277993083000183, | |
| "learning_rate": 2.859882090321104e-06, | |
| "loss": 2.4267, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.1156188470168048, | |
| "grad_norm": 0.2651556730270386, | |
| "learning_rate": 2.8885009444221858e-06, | |
| "loss": 2.4423, | |
| "step": 10100 | |
| }, | |
| { | |
| "epoch": 0.11676358807637713, | |
| "grad_norm": 0.22295983135700226, | |
| "learning_rate": 2.9171197985232673e-06, | |
| "loss": 2.4227, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 0.11790832913594945, | |
| "grad_norm": 0.2351614385843277, | |
| "learning_rate": 2.9457386526243493e-06, | |
| "loss": 2.4237, | |
| "step": 10300 | |
| }, | |
| { | |
| "epoch": 0.11905307019552178, | |
| "grad_norm": 0.327232301235199, | |
| "learning_rate": 2.974357506725431e-06, | |
| "loss": 2.4195, | |
| "step": 10400 | |
| }, | |
| { | |
| "epoch": 0.1201978112550941, | |
| "grad_norm": 0.234052836894989, | |
| "learning_rate": 3.002976360826513e-06, | |
| "loss": 2.4265, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.12134255231466642, | |
| "grad_norm": 0.29197776317596436, | |
| "learning_rate": 3.0315952149275944e-06, | |
| "loss": 2.4175, | |
| "step": 10600 | |
| }, | |
| { | |
| "epoch": 0.12248729337423875, | |
| "grad_norm": 0.3510327935218811, | |
| "learning_rate": 3.0602140690286763e-06, | |
| "loss": 2.4018, | |
| "step": 10700 | |
| }, | |
| { | |
| "epoch": 0.12363203443381107, | |
| "grad_norm": 0.24532395601272583, | |
| "learning_rate": 3.0888329231297583e-06, | |
| "loss": 2.4081, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 0.1247767754933834, | |
| "grad_norm": 0.29377228021621704, | |
| "learning_rate": 3.11745177723084e-06, | |
| "loss": 2.4152, | |
| "step": 10900 | |
| }, | |
| { | |
| "epoch": 0.12592151655295572, | |
| "grad_norm": 0.3314598798751831, | |
| "learning_rate": 3.146070631331922e-06, | |
| "loss": 2.4224, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.12706625761252804, | |
| "grad_norm": 0.31806275248527527, | |
| "learning_rate": 3.1746894854330034e-06, | |
| "loss": 2.4087, | |
| "step": 11100 | |
| }, | |
| { | |
| "epoch": 0.12821099867210037, | |
| "grad_norm": 0.29323023557662964, | |
| "learning_rate": 3.2033083395340854e-06, | |
| "loss": 2.3941, | |
| "step": 11200 | |
| }, | |
| { | |
| "epoch": 0.1293557397316727, | |
| "grad_norm": 0.230011448264122, | |
| "learning_rate": 3.231927193635167e-06, | |
| "loss": 2.4154, | |
| "step": 11300 | |
| }, | |
| { | |
| "epoch": 0.13050048079124502, | |
| "grad_norm": 0.36185070872306824, | |
| "learning_rate": 3.260546047736249e-06, | |
| "loss": 2.4163, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 0.13164522185081734, | |
| "grad_norm": 0.35968175530433655, | |
| "learning_rate": 3.2891649018373304e-06, | |
| "loss": 2.4148, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.13278996291038966, | |
| "grad_norm": 0.3145340383052826, | |
| "learning_rate": 3.3177837559384124e-06, | |
| "loss": 2.4102, | |
| "step": 11600 | |
| }, | |
| { | |
| "epoch": 0.133934703969962, | |
| "grad_norm": 0.26297980546951294, | |
| "learning_rate": 3.3464026100394944e-06, | |
| "loss": 2.4047, | |
| "step": 11700 | |
| }, | |
| { | |
| "epoch": 0.1350794450295343, | |
| "grad_norm": 0.24281686544418335, | |
| "learning_rate": 3.375021464140576e-06, | |
| "loss": 2.3916, | |
| "step": 11800 | |
| }, | |
| { | |
| "epoch": 0.13622418608910664, | |
| "grad_norm": 0.2922670245170593, | |
| "learning_rate": 3.403640318241658e-06, | |
| "loss": 2.4226, | |
| "step": 11900 | |
| }, | |
| { | |
| "epoch": 0.13736892714867896, | |
| "grad_norm": 0.28737780451774597, | |
| "learning_rate": 3.4322591723427395e-06, | |
| "loss": 2.4099, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.13851366820825128, | |
| "grad_norm": 0.7136600613594055, | |
| "learning_rate": 3.4608780264438214e-06, | |
| "loss": 2.4025, | |
| "step": 12100 | |
| }, | |
| { | |
| "epoch": 0.1396584092678236, | |
| "grad_norm": 0.2455575317144394, | |
| "learning_rate": 3.489496880544903e-06, | |
| "loss": 2.403, | |
| "step": 12200 | |
| }, | |
| { | |
| "epoch": 0.14080315032739593, | |
| "grad_norm": 0.24030736088752747, | |
| "learning_rate": 3.518115734645985e-06, | |
| "loss": 2.3824, | |
| "step": 12300 | |
| }, | |
| { | |
| "epoch": 0.14194789138696826, | |
| "grad_norm": 0.28610554337501526, | |
| "learning_rate": 3.5467345887470665e-06, | |
| "loss": 2.414, | |
| "step": 12400 | |
| }, | |
| { | |
| "epoch": 0.14309263244654058, | |
| "grad_norm": 0.28286024928092957, | |
| "learning_rate": 3.5753534428481485e-06, | |
| "loss": 2.403, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.1442373735061129, | |
| "grad_norm": 0.27423399686813354, | |
| "learning_rate": 3.60397229694923e-06, | |
| "loss": 2.3912, | |
| "step": 12600 | |
| }, | |
| { | |
| "epoch": 0.14538211456568526, | |
| "grad_norm": 0.3900642991065979, | |
| "learning_rate": 3.632591151050312e-06, | |
| "loss": 2.391, | |
| "step": 12700 | |
| }, | |
| { | |
| "epoch": 0.14652685562525758, | |
| "grad_norm": 0.3418841063976288, | |
| "learning_rate": 3.661210005151394e-06, | |
| "loss": 2.3931, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 0.1476715966848299, | |
| "grad_norm": 0.22310563921928406, | |
| "learning_rate": 3.6898288592524755e-06, | |
| "loss": 2.4003, | |
| "step": 12900 | |
| }, | |
| { | |
| "epoch": 0.14881633774440223, | |
| "grad_norm": 0.3633168935775757, | |
| "learning_rate": 3.7184477133535575e-06, | |
| "loss": 2.388, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.14996107880397455, | |
| "grad_norm": 0.24249403178691864, | |
| "learning_rate": 3.747066567454639e-06, | |
| "loss": 2.3798, | |
| "step": 13100 | |
| }, | |
| { | |
| "epoch": 0.15110581986354688, | |
| "grad_norm": 0.3075302243232727, | |
| "learning_rate": 3.7756854215557214e-06, | |
| "loss": 2.379, | |
| "step": 13200 | |
| }, | |
| { | |
| "epoch": 0.1522505609231192, | |
| "grad_norm": 0.2638431787490845, | |
| "learning_rate": 3.8043042756568034e-06, | |
| "loss": 2.3826, | |
| "step": 13300 | |
| }, | |
| { | |
| "epoch": 0.15339530198269152, | |
| "grad_norm": 0.26029646396636963, | |
| "learning_rate": 3.832923129757885e-06, | |
| "loss": 2.3876, | |
| "step": 13400 | |
| }, | |
| { | |
| "epoch": 0.15454004304226385, | |
| "grad_norm": 0.297446072101593, | |
| "learning_rate": 3.8615419838589665e-06, | |
| "loss": 2.3791, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.15568478410183617, | |
| "grad_norm": 0.34628915786743164, | |
| "learning_rate": 3.890160837960049e-06, | |
| "loss": 2.4032, | |
| "step": 13600 | |
| }, | |
| { | |
| "epoch": 0.1568295251614085, | |
| "grad_norm": 0.2669197916984558, | |
| "learning_rate": 3.9187796920611305e-06, | |
| "loss": 2.4, | |
| "step": 13700 | |
| }, | |
| { | |
| "epoch": 0.15797426622098082, | |
| "grad_norm": 0.3261224627494812, | |
| "learning_rate": 3.947398546162212e-06, | |
| "loss": 2.3816, | |
| "step": 13800 | |
| }, | |
| { | |
| "epoch": 0.15911900728055314, | |
| "grad_norm": 0.3039107024669647, | |
| "learning_rate": 3.9760174002632936e-06, | |
| "loss": 2.3783, | |
| "step": 13900 | |
| }, | |
| { | |
| "epoch": 0.16026374834012547, | |
| "grad_norm": 0.26345106959342957, | |
| "learning_rate": 4.004636254364376e-06, | |
| "loss": 2.382, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.1614084893996978, | |
| "grad_norm": 0.3465179204940796, | |
| "learning_rate": 4.0332551084654575e-06, | |
| "loss": 2.3873, | |
| "step": 14100 | |
| }, | |
| { | |
| "epoch": 0.16255323045927011, | |
| "grad_norm": 0.2453349232673645, | |
| "learning_rate": 4.061873962566539e-06, | |
| "loss": 2.3881, | |
| "step": 14200 | |
| }, | |
| { | |
| "epoch": 0.16369797151884244, | |
| "grad_norm": 0.24121074378490448, | |
| "learning_rate": 4.0904928166676215e-06, | |
| "loss": 2.3895, | |
| "step": 14300 | |
| }, | |
| { | |
| "epoch": 0.16484271257841476, | |
| "grad_norm": 0.3282526135444641, | |
| "learning_rate": 4.119111670768703e-06, | |
| "loss": 2.391, | |
| "step": 14400 | |
| }, | |
| { | |
| "epoch": 0.1659874536379871, | |
| "grad_norm": 0.32719773054122925, | |
| "learning_rate": 4.1477305248697846e-06, | |
| "loss": 2.3729, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.1671321946975594, | |
| "grad_norm": 0.2743726074695587, | |
| "learning_rate": 4.176349378970866e-06, | |
| "loss": 2.3967, | |
| "step": 14600 | |
| }, | |
| { | |
| "epoch": 0.16827693575713173, | |
| "grad_norm": 0.2472705990076065, | |
| "learning_rate": 4.2049682330719485e-06, | |
| "loss": 2.3871, | |
| "step": 14700 | |
| }, | |
| { | |
| "epoch": 0.16942167681670406, | |
| "grad_norm": 0.23346185684204102, | |
| "learning_rate": 4.23358708717303e-06, | |
| "loss": 2.3762, | |
| "step": 14800 | |
| }, | |
| { | |
| "epoch": 0.17056641787627638, | |
| "grad_norm": 0.261417031288147, | |
| "learning_rate": 4.262205941274112e-06, | |
| "loss": 2.3947, | |
| "step": 14900 | |
| }, | |
| { | |
| "epoch": 0.1717111589358487, | |
| "grad_norm": 0.3324854373931885, | |
| "learning_rate": 4.290824795375194e-06, | |
| "loss": 2.381, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.17285589999542103, | |
| "grad_norm": 0.3383265733718872, | |
| "learning_rate": 4.3194436494762755e-06, | |
| "loss": 2.3716, | |
| "step": 15100 | |
| }, | |
| { | |
| "epoch": 0.17400064105499335, | |
| "grad_norm": 0.34629401564598083, | |
| "learning_rate": 4.348062503577357e-06, | |
| "loss": 2.369, | |
| "step": 15200 | |
| }, | |
| { | |
| "epoch": 0.17514538211456568, | |
| "grad_norm": 0.24686865508556366, | |
| "learning_rate": 4.376681357678439e-06, | |
| "loss": 2.3883, | |
| "step": 15300 | |
| }, | |
| { | |
| "epoch": 0.176290123174138, | |
| "grad_norm": 0.38007158041000366, | |
| "learning_rate": 4.405300211779521e-06, | |
| "loss": 2.3869, | |
| "step": 15400 | |
| }, | |
| { | |
| "epoch": 0.17743486423371033, | |
| "grad_norm": 0.313494473695755, | |
| "learning_rate": 4.433919065880603e-06, | |
| "loss": 2.3859, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.17857960529328265, | |
| "grad_norm": 0.2624611556529999, | |
| "learning_rate": 4.462537919981684e-06, | |
| "loss": 2.3865, | |
| "step": 15600 | |
| }, | |
| { | |
| "epoch": 0.17972434635285497, | |
| "grad_norm": 0.2802521884441376, | |
| "learning_rate": 4.491156774082766e-06, | |
| "loss": 2.3939, | |
| "step": 15700 | |
| }, | |
| { | |
| "epoch": 0.1808690874124273, | |
| "grad_norm": 0.3011086881160736, | |
| "learning_rate": 4.519775628183848e-06, | |
| "loss": 2.385, | |
| "step": 15800 | |
| }, | |
| { | |
| "epoch": 0.18201382847199962, | |
| "grad_norm": 0.28954678773880005, | |
| "learning_rate": 4.54839448228493e-06, | |
| "loss": 2.3897, | |
| "step": 15900 | |
| }, | |
| { | |
| "epoch": 0.18315856953157195, | |
| "grad_norm": 0.2933329939842224, | |
| "learning_rate": 4.577013336386011e-06, | |
| "loss": 2.3761, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.18430331059114427, | |
| "grad_norm": 0.2496791034936905, | |
| "learning_rate": 4.605632190487094e-06, | |
| "loss": 2.3801, | |
| "step": 16100 | |
| }, | |
| { | |
| "epoch": 0.1854480516507166, | |
| "grad_norm": 0.28926581144332886, | |
| "learning_rate": 4.634251044588175e-06, | |
| "loss": 2.3754, | |
| "step": 16200 | |
| }, | |
| { | |
| "epoch": 0.18659279271028895, | |
| "grad_norm": 0.3181098997592926, | |
| "learning_rate": 4.662869898689257e-06, | |
| "loss": 2.3625, | |
| "step": 16300 | |
| }, | |
| { | |
| "epoch": 0.18773753376986127, | |
| "grad_norm": 0.3494364321231842, | |
| "learning_rate": 4.691488752790338e-06, | |
| "loss": 2.3688, | |
| "step": 16400 | |
| }, | |
| { | |
| "epoch": 0.1888822748294336, | |
| "grad_norm": 0.4044504463672638, | |
| "learning_rate": 4.720107606891421e-06, | |
| "loss": 2.379, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.19002701588900592, | |
| "grad_norm": 0.34538954496383667, | |
| "learning_rate": 4.748726460992502e-06, | |
| "loss": 2.3746, | |
| "step": 16600 | |
| }, | |
| { | |
| "epoch": 0.19117175694857824, | |
| "grad_norm": 0.2930959165096283, | |
| "learning_rate": 4.777345315093584e-06, | |
| "loss": 2.3628, | |
| "step": 16700 | |
| }, | |
| { | |
| "epoch": 0.19231649800815057, | |
| "grad_norm": 0.28704413771629333, | |
| "learning_rate": 4.805964169194665e-06, | |
| "loss": 2.3803, | |
| "step": 16800 | |
| }, | |
| { | |
| "epoch": 0.1934612390677229, | |
| "grad_norm": 0.34928804636001587, | |
| "learning_rate": 4.834583023295748e-06, | |
| "loss": 2.3745, | |
| "step": 16900 | |
| }, | |
| { | |
| "epoch": 0.19460598012729521, | |
| "grad_norm": 0.4369732439517975, | |
| "learning_rate": 4.863201877396829e-06, | |
| "loss": 2.3732, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.19575072118686754, | |
| "grad_norm": 0.31494560837745667, | |
| "learning_rate": 4.891820731497911e-06, | |
| "loss": 2.3747, | |
| "step": 17100 | |
| }, | |
| { | |
| "epoch": 0.19689546224643986, | |
| "grad_norm": 0.6197758913040161, | |
| "learning_rate": 4.920439585598993e-06, | |
| "loss": 2.3824, | |
| "step": 17200 | |
| }, | |
| { | |
| "epoch": 0.19804020330601219, | |
| "grad_norm": 0.4105755388736725, | |
| "learning_rate": 4.949058439700075e-06, | |
| "loss": 2.3705, | |
| "step": 17300 | |
| }, | |
| { | |
| "epoch": 0.1991849443655845, | |
| "grad_norm": 0.49081698060035706, | |
| "learning_rate": 4.977677293801156e-06, | |
| "loss": 2.3591, | |
| "step": 17400 | |
| }, | |
| { | |
| "epoch": 0.20032968542515683, | |
| "grad_norm": 0.2819342017173767, | |
| "learning_rate": 5.006296147902239e-06, | |
| "loss": 2.3844, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.20147442648472916, | |
| "grad_norm": 0.30305904150009155, | |
| "learning_rate": 5.03491500200332e-06, | |
| "loss": 2.3749, | |
| "step": 17600 | |
| }, | |
| { | |
| "epoch": 0.20261916754430148, | |
| "grad_norm": 0.36479583382606506, | |
| "learning_rate": 5.063533856104403e-06, | |
| "loss": 2.3765, | |
| "step": 17700 | |
| }, | |
| { | |
| "epoch": 0.2037639086038738, | |
| "grad_norm": 0.3018398880958557, | |
| "learning_rate": 5.092152710205483e-06, | |
| "loss": 2.3673, | |
| "step": 17800 | |
| }, | |
| { | |
| "epoch": 0.20490864966344613, | |
| "grad_norm": 0.32054489850997925, | |
| "learning_rate": 5.120771564306566e-06, | |
| "loss": 2.3658, | |
| "step": 17900 | |
| }, | |
| { | |
| "epoch": 0.20605339072301845, | |
| "grad_norm": 0.3198222219944, | |
| "learning_rate": 5.149390418407647e-06, | |
| "loss": 2.3693, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.20719813178259078, | |
| "grad_norm": 0.405718594789505, | |
| "learning_rate": 5.17800927250873e-06, | |
| "loss": 2.3586, | |
| "step": 18100 | |
| }, | |
| { | |
| "epoch": 0.2083428728421631, | |
| "grad_norm": 0.2990506887435913, | |
| "learning_rate": 5.20662812660981e-06, | |
| "loss": 2.3493, | |
| "step": 18200 | |
| }, | |
| { | |
| "epoch": 0.20948761390173543, | |
| "grad_norm": 0.3617069721221924, | |
| "learning_rate": 5.235246980710893e-06, | |
| "loss": 2.3556, | |
| "step": 18300 | |
| }, | |
| { | |
| "epoch": 0.21063235496130775, | |
| "grad_norm": 0.3361080586910248, | |
| "learning_rate": 5.263865834811974e-06, | |
| "loss": 2.3538, | |
| "step": 18400 | |
| }, | |
| { | |
| "epoch": 0.21177709602088007, | |
| "grad_norm": 0.37158912420272827, | |
| "learning_rate": 5.292484688913057e-06, | |
| "loss": 2.3538, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.2129218370804524, | |
| "grad_norm": 0.3937898576259613, | |
| "learning_rate": 5.321103543014137e-06, | |
| "loss": 2.368, | |
| "step": 18600 | |
| }, | |
| { | |
| "epoch": 0.21406657814002472, | |
| "grad_norm": 0.42322373390197754, | |
| "learning_rate": 5.34972239711522e-06, | |
| "loss": 2.3646, | |
| "step": 18700 | |
| }, | |
| { | |
| "epoch": 0.21521131919959705, | |
| "grad_norm": 0.411285400390625, | |
| "learning_rate": 5.378341251216301e-06, | |
| "loss": 2.3604, | |
| "step": 18800 | |
| }, | |
| { | |
| "epoch": 0.21635606025916937, | |
| "grad_norm": 0.30224665999412537, | |
| "learning_rate": 5.406960105317384e-06, | |
| "loss": 2.3579, | |
| "step": 18900 | |
| }, | |
| { | |
| "epoch": 0.2175008013187417, | |
| "grad_norm": 0.339236319065094, | |
| "learning_rate": 5.435578959418465e-06, | |
| "loss": 2.3741, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.21864554237831402, | |
| "grad_norm": 0.37957480549812317, | |
| "learning_rate": 5.464197813519547e-06, | |
| "loss": 2.3549, | |
| "step": 19100 | |
| }, | |
| { | |
| "epoch": 0.21979028343788634, | |
| "grad_norm": 0.2912836968898773, | |
| "learning_rate": 5.492816667620628e-06, | |
| "loss": 2.352, | |
| "step": 19200 | |
| }, | |
| { | |
| "epoch": 0.22093502449745867, | |
| "grad_norm": 0.2871094346046448, | |
| "learning_rate": 5.521435521721711e-06, | |
| "loss": 2.3602, | |
| "step": 19300 | |
| }, | |
| { | |
| "epoch": 0.222079765557031, | |
| "grad_norm": 0.33945441246032715, | |
| "learning_rate": 5.550054375822792e-06, | |
| "loss": 2.3715, | |
| "step": 19400 | |
| }, | |
| { | |
| "epoch": 0.2232245066166033, | |
| "grad_norm": 0.36321476101875305, | |
| "learning_rate": 5.578673229923875e-06, | |
| "loss": 2.3601, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.22436924767617564, | |
| "grad_norm": 0.27080637216567993, | |
| "learning_rate": 5.6072920840249555e-06, | |
| "loss": 2.3533, | |
| "step": 19600 | |
| }, | |
| { | |
| "epoch": 0.22551398873574796, | |
| "grad_norm": 0.2903904318809509, | |
| "learning_rate": 5.635910938126038e-06, | |
| "loss": 2.3658, | |
| "step": 19700 | |
| }, | |
| { | |
| "epoch": 0.22665872979532029, | |
| "grad_norm": 0.3127708435058594, | |
| "learning_rate": 5.664529792227119e-06, | |
| "loss": 2.3484, | |
| "step": 19800 | |
| }, | |
| { | |
| "epoch": 0.22780347085489264, | |
| "grad_norm": 0.37779003381729126, | |
| "learning_rate": 5.693148646328202e-06, | |
| "loss": 2.3635, | |
| "step": 19900 | |
| }, | |
| { | |
| "epoch": 0.22894821191446496, | |
| "grad_norm": 0.28905001282691956, | |
| "learning_rate": 5.7217675004292825e-06, | |
| "loss": 2.3645, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.23009295297403728, | |
| "grad_norm": 0.3213961720466614, | |
| "learning_rate": 5.750386354530365e-06, | |
| "loss": 2.3584, | |
| "step": 20100 | |
| }, | |
| { | |
| "epoch": 0.2312376940336096, | |
| "grad_norm": 0.28322041034698486, | |
| "learning_rate": 5.7790052086314464e-06, | |
| "loss": 2.3616, | |
| "step": 20200 | |
| }, | |
| { | |
| "epoch": 0.23238243509318193, | |
| "grad_norm": 0.3427826762199402, | |
| "learning_rate": 5.807624062732529e-06, | |
| "loss": 2.366, | |
| "step": 20300 | |
| }, | |
| { | |
| "epoch": 0.23352717615275426, | |
| "grad_norm": 0.35720208287239075, | |
| "learning_rate": 5.836242916833611e-06, | |
| "loss": 2.3604, | |
| "step": 20400 | |
| }, | |
| { | |
| "epoch": 0.23467191721232658, | |
| "grad_norm": 0.3247853219509125, | |
| "learning_rate": 5.864861770934692e-06, | |
| "loss": 2.3528, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.2358166582718989, | |
| "grad_norm": 0.31638818979263306, | |
| "learning_rate": 5.893480625035774e-06, | |
| "loss": 2.3659, | |
| "step": 20600 | |
| }, | |
| { | |
| "epoch": 0.23696139933147123, | |
| "grad_norm": 0.3150993287563324, | |
| "learning_rate": 5.922099479136856e-06, | |
| "loss": 2.3542, | |
| "step": 20700 | |
| }, | |
| { | |
| "epoch": 0.23810614039104355, | |
| "grad_norm": 0.28134357929229736, | |
| "learning_rate": 5.950718333237938e-06, | |
| "loss": 2.3645, | |
| "step": 20800 | |
| }, | |
| { | |
| "epoch": 0.23925088145061588, | |
| "grad_norm": 0.344279408454895, | |
| "learning_rate": 5.979337187339019e-06, | |
| "loss": 2.3553, | |
| "step": 20900 | |
| }, | |
| { | |
| "epoch": 0.2403956225101882, | |
| "grad_norm": 0.3156017065048218, | |
| "learning_rate": 6.007956041440101e-06, | |
| "loss": 2.3399, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.24154036356976052, | |
| "grad_norm": 0.31625500321388245, | |
| "learning_rate": 6.036574895541183e-06, | |
| "loss": 2.3458, | |
| "step": 21100 | |
| }, | |
| { | |
| "epoch": 0.24268510462933285, | |
| "grad_norm": 0.3604189455509186, | |
| "learning_rate": 6.065193749642265e-06, | |
| "loss": 2.3426, | |
| "step": 21200 | |
| }, | |
| { | |
| "epoch": 0.24382984568890517, | |
| "grad_norm": 0.3052213490009308, | |
| "learning_rate": 6.093526415202336e-06, | |
| "loss": 2.3498, | |
| "step": 21300 | |
| }, | |
| { | |
| "epoch": 0.2449745867484775, | |
| "grad_norm": 0.40419507026672363, | |
| "learning_rate": 6.122145269303418e-06, | |
| "loss": 2.3524, | |
| "step": 21400 | |
| }, | |
| { | |
| "epoch": 0.24611932780804982, | |
| "grad_norm": 0.3688088357448578, | |
| "learning_rate": 6.150764123404499e-06, | |
| "loss": 2.3434, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.24726406886762214, | |
| "grad_norm": 0.3363403379917145, | |
| "learning_rate": 6.179382977505581e-06, | |
| "loss": 2.3524, | |
| "step": 21600 | |
| }, | |
| { | |
| "epoch": 0.24840880992719447, | |
| "grad_norm": 0.516409695148468, | |
| "learning_rate": 6.208001831606663e-06, | |
| "loss": 2.3424, | |
| "step": 21700 | |
| }, | |
| { | |
| "epoch": 0.2495535509867668, | |
| "grad_norm": 0.3187144100666046, | |
| "learning_rate": 6.236620685707745e-06, | |
| "loss": 2.3427, | |
| "step": 21800 | |
| }, | |
| { | |
| "epoch": 0.25069829204633914, | |
| "grad_norm": 0.3366081416606903, | |
| "learning_rate": 6.265239539808826e-06, | |
| "loss": 2.3461, | |
| "step": 21900 | |
| }, | |
| { | |
| "epoch": 0.25184303310591144, | |
| "grad_norm": 0.3357242941856384, | |
| "learning_rate": 6.293858393909908e-06, | |
| "loss": 2.3423, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.2529877741654838, | |
| "grad_norm": 0.3461867570877075, | |
| "learning_rate": 6.32247724801099e-06, | |
| "loss": 2.3448, | |
| "step": 22100 | |
| }, | |
| { | |
| "epoch": 0.2541325152250561, | |
| "grad_norm": 0.3421408236026764, | |
| "learning_rate": 6.351096102112072e-06, | |
| "loss": 2.3513, | |
| "step": 22200 | |
| }, | |
| { | |
| "epoch": 0.25527725628462844, | |
| "grad_norm": 0.3435458242893219, | |
| "learning_rate": 6.379714956213153e-06, | |
| "loss": 2.359, | |
| "step": 22300 | |
| }, | |
| { | |
| "epoch": 0.25642199734420074, | |
| "grad_norm": 0.31256601214408875, | |
| "learning_rate": 6.408333810314235e-06, | |
| "loss": 2.3417, | |
| "step": 22400 | |
| }, | |
| { | |
| "epoch": 0.2575667384037731, | |
| "grad_norm": 0.4700869023799896, | |
| "learning_rate": 6.436952664415317e-06, | |
| "loss": 2.3609, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 0.2587114794633454, | |
| "grad_norm": 0.3140374422073364, | |
| "learning_rate": 6.465571518516399e-06, | |
| "loss": 2.3749, | |
| "step": 22600 | |
| }, | |
| { | |
| "epoch": 0.25985622052291774, | |
| "grad_norm": 0.35399436950683594, | |
| "learning_rate": 6.49419037261748e-06, | |
| "loss": 2.35, | |
| "step": 22700 | |
| }, | |
| { | |
| "epoch": 0.26100096158249003, | |
| "grad_norm": 0.3978697657585144, | |
| "learning_rate": 6.5228092267185625e-06, | |
| "loss": 2.3486, | |
| "step": 22800 | |
| }, | |
| { | |
| "epoch": 0.2621457026420624, | |
| "grad_norm": 0.4412658214569092, | |
| "learning_rate": 6.551428080819644e-06, | |
| "loss": 2.3555, | |
| "step": 22900 | |
| }, | |
| { | |
| "epoch": 0.2632904437016347, | |
| "grad_norm": 0.3189774453639984, | |
| "learning_rate": 6.580046934920726e-06, | |
| "loss": 2.3482, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.26443518476120703, | |
| "grad_norm": 0.40347644686698914, | |
| "learning_rate": 6.608665789021808e-06, | |
| "loss": 2.3528, | |
| "step": 23100 | |
| }, | |
| { | |
| "epoch": 0.26557992582077933, | |
| "grad_norm": 0.3658533990383148, | |
| "learning_rate": 6.6372846431228895e-06, | |
| "loss": 2.3328, | |
| "step": 23200 | |
| }, | |
| { | |
| "epoch": 0.2667246668803517, | |
| "grad_norm": 0.31631380319595337, | |
| "learning_rate": 6.665903497223972e-06, | |
| "loss": 2.3554, | |
| "step": 23300 | |
| }, | |
| { | |
| "epoch": 0.267869407939924, | |
| "grad_norm": 0.3406871259212494, | |
| "learning_rate": 6.6945223513250534e-06, | |
| "loss": 2.358, | |
| "step": 23400 | |
| }, | |
| { | |
| "epoch": 0.2690141489994963, | |
| "grad_norm": 0.34881141781806946, | |
| "learning_rate": 6.723141205426136e-06, | |
| "loss": 2.343, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 0.2701588900590686, | |
| "grad_norm": 0.5699728727340698, | |
| "learning_rate": 6.751760059527217e-06, | |
| "loss": 2.353, | |
| "step": 23600 | |
| }, | |
| { | |
| "epoch": 0.271303631118641, | |
| "grad_norm": 0.36337733268737793, | |
| "learning_rate": 6.780378913628299e-06, | |
| "loss": 2.3463, | |
| "step": 23700 | |
| }, | |
| { | |
| "epoch": 0.27244837217821327, | |
| "grad_norm": 0.34283825755119324, | |
| "learning_rate": 6.8089977677293805e-06, | |
| "loss": 2.3343, | |
| "step": 23800 | |
| }, | |
| { | |
| "epoch": 0.2735931132377856, | |
| "grad_norm": 0.3663531243801117, | |
| "learning_rate": 6.837616621830463e-06, | |
| "loss": 2.3477, | |
| "step": 23900 | |
| }, | |
| { | |
| "epoch": 0.2747378542973579, | |
| "grad_norm": 0.4608674943447113, | |
| "learning_rate": 6.8662354759315444e-06, | |
| "loss": 2.3396, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.27588259535693027, | |
| "grad_norm": 0.32811376452445984, | |
| "learning_rate": 6.894854330032627e-06, | |
| "loss": 2.3409, | |
| "step": 24100 | |
| }, | |
| { | |
| "epoch": 0.27702733641650257, | |
| "grad_norm": 0.3832899034023285, | |
| "learning_rate": 6.9231869955926965e-06, | |
| "loss": 2.3573, | |
| "step": 24200 | |
| }, | |
| { | |
| "epoch": 0.2781720774760749, | |
| "grad_norm": 0.3628441095352173, | |
| "learning_rate": 6.951805849693779e-06, | |
| "loss": 2.3515, | |
| "step": 24300 | |
| }, | |
| { | |
| "epoch": 0.2793168185356472, | |
| "grad_norm": 0.35530924797058105, | |
| "learning_rate": 6.98042470379486e-06, | |
| "loss": 2.3608, | |
| "step": 24400 | |
| }, | |
| { | |
| "epoch": 0.28046155959521957, | |
| "grad_norm": 0.33606886863708496, | |
| "learning_rate": 7.009043557895943e-06, | |
| "loss": 2.3445, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 0.28160630065479186, | |
| "grad_norm": 0.3475317358970642, | |
| "learning_rate": 7.0376624119970235e-06, | |
| "loss": 2.3527, | |
| "step": 24600 | |
| }, | |
| { | |
| "epoch": 0.2827510417143642, | |
| "grad_norm": 0.3616209030151367, | |
| "learning_rate": 7.066281266098106e-06, | |
| "loss": 2.3456, | |
| "step": 24700 | |
| }, | |
| { | |
| "epoch": 0.2838957827739365, | |
| "grad_norm": 0.4273635745048523, | |
| "learning_rate": 7.0949001201991875e-06, | |
| "loss": 2.3356, | |
| "step": 24800 | |
| }, | |
| { | |
| "epoch": 0.28504052383350886, | |
| "grad_norm": 0.32688596844673157, | |
| "learning_rate": 7.12351897430027e-06, | |
| "loss": 2.3435, | |
| "step": 24900 | |
| }, | |
| { | |
| "epoch": 0.28618526489308116, | |
| "grad_norm": 0.3584359288215637, | |
| "learning_rate": 7.1521378284013506e-06, | |
| "loss": 2.3431, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.2873300059526535, | |
| "grad_norm": 0.35065096616744995, | |
| "learning_rate": 7.180756682502433e-06, | |
| "loss": 2.3316, | |
| "step": 25100 | |
| }, | |
| { | |
| "epoch": 0.2884747470122258, | |
| "grad_norm": 0.3658662736415863, | |
| "learning_rate": 7.2093755366035145e-06, | |
| "loss": 2.342, | |
| "step": 25200 | |
| }, | |
| { | |
| "epoch": 0.28961948807179816, | |
| "grad_norm": 0.3615448772907257, | |
| "learning_rate": 7.237994390704597e-06, | |
| "loss": 2.3413, | |
| "step": 25300 | |
| }, | |
| { | |
| "epoch": 0.2907642291313705, | |
| "grad_norm": 0.37493348121643066, | |
| "learning_rate": 7.2666132448056785e-06, | |
| "loss": 2.3261, | |
| "step": 25400 | |
| }, | |
| { | |
| "epoch": 0.2919089701909428, | |
| "grad_norm": 0.43712013959884644, | |
| "learning_rate": 7.29523209890676e-06, | |
| "loss": 2.3367, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 0.29305371125051516, | |
| "grad_norm": 0.37490740418434143, | |
| "learning_rate": 7.3238509530078416e-06, | |
| "loss": 2.323, | |
| "step": 25600 | |
| }, | |
| { | |
| "epoch": 0.29419845231008745, | |
| "grad_norm": 0.33274149894714355, | |
| "learning_rate": 7.352469807108924e-06, | |
| "loss": 2.3333, | |
| "step": 25700 | |
| }, | |
| { | |
| "epoch": 0.2953431933696598, | |
| "grad_norm": 0.41356754302978516, | |
| "learning_rate": 7.3810886612100055e-06, | |
| "loss": 2.3253, | |
| "step": 25800 | |
| }, | |
| { | |
| "epoch": 0.2964879344292321, | |
| "grad_norm": 0.3717619478702545, | |
| "learning_rate": 7.409707515311088e-06, | |
| "loss": 2.3483, | |
| "step": 25900 | |
| }, | |
| { | |
| "epoch": 0.29763267548880445, | |
| "grad_norm": 0.3571007251739502, | |
| "learning_rate": 7.438326369412169e-06, | |
| "loss": 2.3429, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.29877741654837675, | |
| "grad_norm": 0.33724385499954224, | |
| "learning_rate": 7.466945223513251e-06, | |
| "loss": 2.3387, | |
| "step": 26100 | |
| }, | |
| { | |
| "epoch": 0.2999221576079491, | |
| "grad_norm": 0.4527655243873596, | |
| "learning_rate": 7.495564077614333e-06, | |
| "loss": 2.3387, | |
| "step": 26200 | |
| }, | |
| { | |
| "epoch": 0.3010668986675214, | |
| "grad_norm": 0.5135347247123718, | |
| "learning_rate": 7.524182931715415e-06, | |
| "loss": 2.3476, | |
| "step": 26300 | |
| }, | |
| { | |
| "epoch": 0.30221163972709375, | |
| "grad_norm": 0.36313098669052124, | |
| "learning_rate": 7.552801785816497e-06, | |
| "loss": 2.3419, | |
| "step": 26400 | |
| }, | |
| { | |
| "epoch": 0.30335638078666605, | |
| "grad_norm": 0.34982365369796753, | |
| "learning_rate": 7.581420639917578e-06, | |
| "loss": 2.3333, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 0.3045011218462384, | |
| "grad_norm": 0.5387922525405884, | |
| "learning_rate": 7.6100394940186604e-06, | |
| "loss": 2.3267, | |
| "step": 26600 | |
| }, | |
| { | |
| "epoch": 0.3056458629058107, | |
| "grad_norm": 0.3917737305164337, | |
| "learning_rate": 7.638658348119742e-06, | |
| "loss": 2.3441, | |
| "step": 26700 | |
| }, | |
| { | |
| "epoch": 0.30679060396538305, | |
| "grad_norm": 0.362425297498703, | |
| "learning_rate": 7.667277202220824e-06, | |
| "loss": 2.3279, | |
| "step": 26800 | |
| }, | |
| { | |
| "epoch": 0.30793534502495534, | |
| "grad_norm": 0.3922022581100464, | |
| "learning_rate": 7.695896056321905e-06, | |
| "loss": 2.328, | |
| "step": 26900 | |
| }, | |
| { | |
| "epoch": 0.3090800860845277, | |
| "grad_norm": 0.3867265582084656, | |
| "learning_rate": 7.724514910422987e-06, | |
| "loss": 2.3343, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.3102248271441, | |
| "grad_norm": 0.3767886757850647, | |
| "learning_rate": 7.753133764524068e-06, | |
| "loss": 2.3209, | |
| "step": 27100 | |
| }, | |
| { | |
| "epoch": 0.31136956820367234, | |
| "grad_norm": 0.3708842694759369, | |
| "learning_rate": 7.78175261862515e-06, | |
| "loss": 2.3406, | |
| "step": 27200 | |
| }, | |
| { | |
| "epoch": 0.31251430926324464, | |
| "grad_norm": 0.387917697429657, | |
| "learning_rate": 7.810085284185221e-06, | |
| "loss": 2.3314, | |
| "step": 27300 | |
| }, | |
| { | |
| "epoch": 0.313659050322817, | |
| "grad_norm": 0.38667747378349304, | |
| "learning_rate": 7.838704138286303e-06, | |
| "loss": 2.3534, | |
| "step": 27400 | |
| }, | |
| { | |
| "epoch": 0.3148037913823893, | |
| "grad_norm": 0.3695323169231415, | |
| "learning_rate": 7.867322992387386e-06, | |
| "loss": 2.3217, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 0.31594853244196164, | |
| "grad_norm": 0.4034270644187927, | |
| "learning_rate": 7.895941846488467e-06, | |
| "loss": 2.3202, | |
| "step": 27600 | |
| }, | |
| { | |
| "epoch": 0.31709327350153393, | |
| "grad_norm": 0.3806785047054291, | |
| "learning_rate": 7.924560700589549e-06, | |
| "loss": 2.3181, | |
| "step": 27700 | |
| }, | |
| { | |
| "epoch": 0.3182380145611063, | |
| "grad_norm": 0.4768499433994293, | |
| "learning_rate": 7.953179554690631e-06, | |
| "loss": 2.3294, | |
| "step": 27800 | |
| }, | |
| { | |
| "epoch": 0.3193827556206786, | |
| "grad_norm": 0.35020050406455994, | |
| "learning_rate": 7.981798408791712e-06, | |
| "loss": 2.348, | |
| "step": 27900 | |
| }, | |
| { | |
| "epoch": 0.32052749668025093, | |
| "grad_norm": 0.37575003504753113, | |
| "learning_rate": 8.010417262892794e-06, | |
| "loss": 2.3327, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.32167223773982323, | |
| "grad_norm": 0.38812920451164246, | |
| "learning_rate": 8.039036116993875e-06, | |
| "loss": 2.3488, | |
| "step": 28100 | |
| }, | |
| { | |
| "epoch": 0.3228169787993956, | |
| "grad_norm": 0.4457553029060364, | |
| "learning_rate": 8.067654971094958e-06, | |
| "loss": 2.3239, | |
| "step": 28200 | |
| }, | |
| { | |
| "epoch": 0.3239617198589679, | |
| "grad_norm": 0.4105226695537567, | |
| "learning_rate": 8.09627382519604e-06, | |
| "loss": 2.337, | |
| "step": 28300 | |
| }, | |
| { | |
| "epoch": 0.32510646091854023, | |
| "grad_norm": 0.3988581597805023, | |
| "learning_rate": 8.124892679297122e-06, | |
| "loss": 2.3405, | |
| "step": 28400 | |
| }, | |
| { | |
| "epoch": 0.3262512019781125, | |
| "grad_norm": 0.5257512927055359, | |
| "learning_rate": 8.153511533398203e-06, | |
| "loss": 2.3093, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 0.3273959430376849, | |
| "grad_norm": 0.4554663896560669, | |
| "learning_rate": 8.182130387499285e-06, | |
| "loss": 2.3376, | |
| "step": 28600 | |
| }, | |
| { | |
| "epoch": 0.3285406840972572, | |
| "grad_norm": 0.5744247436523438, | |
| "learning_rate": 8.210749241600366e-06, | |
| "loss": 2.3282, | |
| "step": 28700 | |
| }, | |
| { | |
| "epoch": 0.3296854251568295, | |
| "grad_norm": 0.4724181592464447, | |
| "learning_rate": 8.239368095701449e-06, | |
| "loss": 2.3158, | |
| "step": 28800 | |
| }, | |
| { | |
| "epoch": 0.3308301662164019, | |
| "grad_norm": 0.37950775027275085, | |
| "learning_rate": 8.267986949802531e-06, | |
| "loss": 2.33, | |
| "step": 28900 | |
| }, | |
| { | |
| "epoch": 0.3319749072759742, | |
| "grad_norm": 0.3992998003959656, | |
| "learning_rate": 8.296605803903612e-06, | |
| "loss": 2.3445, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.3331196483355465, | |
| "grad_norm": 0.3856017291545868, | |
| "learning_rate": 8.325224658004694e-06, | |
| "loss": 2.3205, | |
| "step": 29100 | |
| }, | |
| { | |
| "epoch": 0.3342643893951188, | |
| "grad_norm": 0.4017845690250397, | |
| "learning_rate": 8.353843512105776e-06, | |
| "loss": 2.3236, | |
| "step": 29200 | |
| }, | |
| { | |
| "epoch": 0.3354091304546912, | |
| "grad_norm": 0.557650625705719, | |
| "learning_rate": 8.382462366206859e-06, | |
| "loss": 2.3188, | |
| "step": 29300 | |
| }, | |
| { | |
| "epoch": 0.33655387151426347, | |
| "grad_norm": 0.5162211060523987, | |
| "learning_rate": 8.410795031766928e-06, | |
| "loss": 2.317, | |
| "step": 29400 | |
| }, | |
| { | |
| "epoch": 0.3376986125738358, | |
| "grad_norm": 0.4619830846786499, | |
| "learning_rate": 8.43941388586801e-06, | |
| "loss": 2.325, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 0.3388433536334081, | |
| "grad_norm": 0.36212870478630066, | |
| "learning_rate": 8.468032739969092e-06, | |
| "loss": 2.3266, | |
| "step": 29600 | |
| }, | |
| { | |
| "epoch": 0.33998809469298047, | |
| "grad_norm": 0.3987511098384857, | |
| "learning_rate": 8.496651594070175e-06, | |
| "loss": 2.3352, | |
| "step": 29700 | |
| }, | |
| { | |
| "epoch": 0.34113283575255277, | |
| "grad_norm": 0.49319741129875183, | |
| "learning_rate": 8.525270448171256e-06, | |
| "loss": 2.331, | |
| "step": 29800 | |
| }, | |
| { | |
| "epoch": 0.3422775768121251, | |
| "grad_norm": 0.3750767111778259, | |
| "learning_rate": 8.553889302272338e-06, | |
| "loss": 2.3197, | |
| "step": 29900 | |
| }, | |
| { | |
| "epoch": 0.3434223178716974, | |
| "grad_norm": 0.42684081196784973, | |
| "learning_rate": 8.582508156373419e-06, | |
| "loss": 2.3196, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.34456705893126977, | |
| "grad_norm": 0.4298862814903259, | |
| "learning_rate": 8.611127010474501e-06, | |
| "loss": 2.309, | |
| "step": 30100 | |
| }, | |
| { | |
| "epoch": 0.34571179999084206, | |
| "grad_norm": 0.4002019762992859, | |
| "learning_rate": 8.639745864575583e-06, | |
| "loss": 2.312, | |
| "step": 30200 | |
| }, | |
| { | |
| "epoch": 0.3468565410504144, | |
| "grad_norm": 0.3674704134464264, | |
| "learning_rate": 8.668364718676666e-06, | |
| "loss": 2.3147, | |
| "step": 30300 | |
| }, | |
| { | |
| "epoch": 0.3480012821099867, | |
| "grad_norm": 0.5854533314704895, | |
| "learning_rate": 8.696983572777747e-06, | |
| "loss": 2.3236, | |
| "step": 30400 | |
| }, | |
| { | |
| "epoch": 0.34914602316955906, | |
| "grad_norm": 0.4132590889930725, | |
| "learning_rate": 8.725602426878829e-06, | |
| "loss": 2.3359, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 0.35029076422913136, | |
| "grad_norm": 0.39070025086402893, | |
| "learning_rate": 8.75422128097991e-06, | |
| "loss": 2.3223, | |
| "step": 30600 | |
| }, | |
| { | |
| "epoch": 0.3514355052887037, | |
| "grad_norm": 0.40703412890434265, | |
| "learning_rate": 8.782840135080992e-06, | |
| "loss": 2.3243, | |
| "step": 30700 | |
| }, | |
| { | |
| "epoch": 0.352580246348276, | |
| "grad_norm": 0.3701010048389435, | |
| "learning_rate": 8.811458989182073e-06, | |
| "loss": 2.3297, | |
| "step": 30800 | |
| }, | |
| { | |
| "epoch": 0.35372498740784836, | |
| "grad_norm": 0.5442121028900146, | |
| "learning_rate": 8.840077843283155e-06, | |
| "loss": 2.3256, | |
| "step": 30900 | |
| }, | |
| { | |
| "epoch": 0.35486972846742065, | |
| "grad_norm": 0.4204414486885071, | |
| "learning_rate": 8.868696697384238e-06, | |
| "loss": 2.3178, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 0.356014469526993, | |
| "grad_norm": 0.4551771283149719, | |
| "learning_rate": 8.89731555148532e-06, | |
| "loss": 2.3299, | |
| "step": 31100 | |
| }, | |
| { | |
| "epoch": 0.3571592105865653, | |
| "grad_norm": 0.4303077757358551, | |
| "learning_rate": 8.9259344055864e-06, | |
| "loss": 2.3251, | |
| "step": 31200 | |
| }, | |
| { | |
| "epoch": 0.35830395164613765, | |
| "grad_norm": 0.41671931743621826, | |
| "learning_rate": 8.954553259687483e-06, | |
| "loss": 2.3279, | |
| "step": 31300 | |
| }, | |
| { | |
| "epoch": 0.35944869270570995, | |
| "grad_norm": 0.4018367528915405, | |
| "learning_rate": 8.983172113788564e-06, | |
| "loss": 2.3069, | |
| "step": 31400 | |
| }, | |
| { | |
| "epoch": 0.3605934337652823, | |
| "grad_norm": 0.5130965113639832, | |
| "learning_rate": 9.011790967889646e-06, | |
| "loss": 2.3315, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 0.3617381748248546, | |
| "grad_norm": 0.4209829270839691, | |
| "learning_rate": 9.040409821990729e-06, | |
| "loss": 2.324, | |
| "step": 31600 | |
| }, | |
| { | |
| "epoch": 0.36288291588442695, | |
| "grad_norm": 0.41159483790397644, | |
| "learning_rate": 9.06902867609181e-06, | |
| "loss": 2.321, | |
| "step": 31700 | |
| }, | |
| { | |
| "epoch": 0.36402765694399924, | |
| "grad_norm": 0.46957677602767944, | |
| "learning_rate": 9.097647530192892e-06, | |
| "loss": 2.3249, | |
| "step": 31800 | |
| }, | |
| { | |
| "epoch": 0.3651723980035716, | |
| "grad_norm": 0.39695438742637634, | |
| "learning_rate": 9.126266384293974e-06, | |
| "loss": 2.3112, | |
| "step": 31900 | |
| }, | |
| { | |
| "epoch": 0.3663171390631439, | |
| "grad_norm": 0.3984281122684479, | |
| "learning_rate": 9.154885238395056e-06, | |
| "loss": 2.3269, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.36746188012271624, | |
| "grad_norm": 0.40746673941612244, | |
| "learning_rate": 9.183504092496137e-06, | |
| "loss": 2.333, | |
| "step": 32100 | |
| }, | |
| { | |
| "epoch": 0.36860662118228854, | |
| "grad_norm": 0.39766183495521545, | |
| "learning_rate": 9.21212294659722e-06, | |
| "loss": 2.3354, | |
| "step": 32200 | |
| }, | |
| { | |
| "epoch": 0.3697513622418609, | |
| "grad_norm": 0.42040807008743286, | |
| "learning_rate": 9.2407418006983e-06, | |
| "loss": 2.3165, | |
| "step": 32300 | |
| }, | |
| { | |
| "epoch": 0.3708961033014332, | |
| "grad_norm": 0.43911924958229065, | |
| "learning_rate": 9.269360654799383e-06, | |
| "loss": 2.3386, | |
| "step": 32400 | |
| }, | |
| { | |
| "epoch": 0.37204084436100554, | |
| "grad_norm": 0.46791312098503113, | |
| "learning_rate": 9.297979508900465e-06, | |
| "loss": 2.3116, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 0.3731855854205779, | |
| "grad_norm": 0.41716283559799194, | |
| "learning_rate": 9.326598363001547e-06, | |
| "loss": 2.3298, | |
| "step": 32600 | |
| }, | |
| { | |
| "epoch": 0.3743303264801502, | |
| "grad_norm": 0.38628315925598145, | |
| "learning_rate": 9.355217217102628e-06, | |
| "loss": 2.3093, | |
| "step": 32700 | |
| }, | |
| { | |
| "epoch": 0.37547506753972254, | |
| "grad_norm": 0.4715399146080017, | |
| "learning_rate": 9.38383607120371e-06, | |
| "loss": 2.3353, | |
| "step": 32800 | |
| }, | |
| { | |
| "epoch": 0.37661980859929484, | |
| "grad_norm": 0.4824056923389435, | |
| "learning_rate": 9.412454925304791e-06, | |
| "loss": 2.3165, | |
| "step": 32900 | |
| }, | |
| { | |
| "epoch": 0.3777645496588672, | |
| "grad_norm": 0.3745131194591522, | |
| "learning_rate": 9.441073779405874e-06, | |
| "loss": 2.31, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 0.3789092907184395, | |
| "grad_norm": 0.4431547522544861, | |
| "learning_rate": 9.469406444965944e-06, | |
| "loss": 2.3158, | |
| "step": 33100 | |
| }, | |
| { | |
| "epoch": 0.38005403177801184, | |
| "grad_norm": 0.43701136112213135, | |
| "learning_rate": 9.498025299067026e-06, | |
| "loss": 2.3184, | |
| "step": 33200 | |
| }, | |
| { | |
| "epoch": 0.38119877283758413, | |
| "grad_norm": 0.40076711773872375, | |
| "learning_rate": 9.526644153168107e-06, | |
| "loss": 2.3247, | |
| "step": 33300 | |
| }, | |
| { | |
| "epoch": 0.3823435138971565, | |
| "grad_norm": Infinity, | |
| "learning_rate": 9.55497681872818e-06, | |
| "loss": 2.3038, | |
| "step": 33400 | |
| }, | |
| { | |
| "epoch": 0.3834882549567288, | |
| "grad_norm": 0.4818211793899536, | |
| "learning_rate": 9.583595672829262e-06, | |
| "loss": 2.3125, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 0.38463299601630113, | |
| "grad_norm": 0.44447603821754456, | |
| "learning_rate": 9.612214526930342e-06, | |
| "loss": 2.3056, | |
| "step": 33600 | |
| }, | |
| { | |
| "epoch": 0.38577773707587343, | |
| "grad_norm": 0.42127880454063416, | |
| "learning_rate": 9.640833381031425e-06, | |
| "loss": 2.318, | |
| "step": 33700 | |
| }, | |
| { | |
| "epoch": 0.3869224781354458, | |
| "grad_norm": 0.42268356680870056, | |
| "learning_rate": 9.669452235132506e-06, | |
| "loss": 2.3247, | |
| "step": 33800 | |
| }, | |
| { | |
| "epoch": 0.3880672191950181, | |
| "grad_norm": 0.42822974920272827, | |
| "learning_rate": 9.698071089233588e-06, | |
| "loss": 2.3164, | |
| "step": 33900 | |
| }, | |
| { | |
| "epoch": 0.38921196025459043, | |
| "grad_norm": 0.4107573926448822, | |
| "learning_rate": 9.72668994333467e-06, | |
| "loss": 2.3187, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 0.3903567013141627, | |
| "grad_norm": 0.40011221170425415, | |
| "learning_rate": 9.755308797435751e-06, | |
| "loss": 2.3197, | |
| "step": 34100 | |
| }, | |
| { | |
| "epoch": 0.3915014423737351, | |
| "grad_norm": 0.4174409508705139, | |
| "learning_rate": 9.783927651536833e-06, | |
| "loss": 2.3082, | |
| "step": 34200 | |
| }, | |
| { | |
| "epoch": 0.39264618343330737, | |
| "grad_norm": 0.4117099344730377, | |
| "learning_rate": 9.812546505637916e-06, | |
| "loss": 2.3197, | |
| "step": 34300 | |
| }, | |
| { | |
| "epoch": 0.3937909244928797, | |
| "grad_norm": 0.4379769563674927, | |
| "learning_rate": 9.841165359738997e-06, | |
| "loss": 2.3151, | |
| "step": 34400 | |
| }, | |
| { | |
| "epoch": 0.394935665552452, | |
| "grad_norm": 0.5240621566772461, | |
| "learning_rate": 9.869784213840079e-06, | |
| "loss": 2.309, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 0.39608040661202437, | |
| "grad_norm": 0.38917648792266846, | |
| "learning_rate": 9.89840306794116e-06, | |
| "loss": 2.3122, | |
| "step": 34600 | |
| }, | |
| { | |
| "epoch": 0.39722514767159667, | |
| "grad_norm": 0.47040241956710815, | |
| "learning_rate": 9.927021922042242e-06, | |
| "loss": 2.3053, | |
| "step": 34700 | |
| }, | |
| { | |
| "epoch": 0.398369888731169, | |
| "grad_norm": 0.42958155274391174, | |
| "learning_rate": 9.955640776143324e-06, | |
| "loss": 2.3038, | |
| "step": 34800 | |
| }, | |
| { | |
| "epoch": 0.3995146297907413, | |
| "grad_norm": 0.4274247884750366, | |
| "learning_rate": 9.984259630244407e-06, | |
| "loss": 2.3104, | |
| "step": 34900 | |
| }, | |
| { | |
| "epoch": 0.40065937085031367, | |
| "grad_norm": 0.4401596188545227, | |
| "learning_rate": 1.0012878484345488e-05, | |
| "loss": 2.2919, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 0.40180411190988596, | |
| "grad_norm": 0.4685971438884735, | |
| "learning_rate": 1.0041497338446568e-05, | |
| "loss": 2.323, | |
| "step": 35100 | |
| }, | |
| { | |
| "epoch": 0.4029488529694583, | |
| "grad_norm": 0.4304906725883484, | |
| "learning_rate": 1.0070116192547652e-05, | |
| "loss": 2.3182, | |
| "step": 35200 | |
| }, | |
| { | |
| "epoch": 0.4040935940290306, | |
| "grad_norm": 0.44299623370170593, | |
| "learning_rate": 1.0098735046648733e-05, | |
| "loss": 2.3041, | |
| "step": 35300 | |
| }, | |
| { | |
| "epoch": 0.40523833508860296, | |
| "grad_norm": 0.42946410179138184, | |
| "learning_rate": 1.0127353900749814e-05, | |
| "loss": 2.3094, | |
| "step": 35400 | |
| }, | |
| { | |
| "epoch": 0.40638307614817526, | |
| "grad_norm": 0.4753871262073517, | |
| "learning_rate": 1.0155972754850896e-05, | |
| "loss": 2.3141, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 0.4075278172077476, | |
| "grad_norm": 0.4177212417125702, | |
| "learning_rate": 1.0184591608951979e-05, | |
| "loss": 2.3255, | |
| "step": 35600 | |
| }, | |
| { | |
| "epoch": 0.4086725582673199, | |
| "grad_norm": 0.4225813150405884, | |
| "learning_rate": 1.0213210463053061e-05, | |
| "loss": 2.3238, | |
| "step": 35700 | |
| }, | |
| { | |
| "epoch": 0.40981729932689226, | |
| "grad_norm": 0.39542898535728455, | |
| "learning_rate": 1.0241829317154142e-05, | |
| "loss": 2.3013, | |
| "step": 35800 | |
| }, | |
| { | |
| "epoch": 0.41096204038646456, | |
| "grad_norm": 0.44745051860809326, | |
| "learning_rate": 1.0270448171255222e-05, | |
| "loss": 2.3142, | |
| "step": 35900 | |
| }, | |
| { | |
| "epoch": 0.4121067814460369, | |
| "grad_norm": 0.46079352498054504, | |
| "learning_rate": 1.0299067025356306e-05, | |
| "loss": 2.3153, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.41325152250560926, | |
| "grad_norm": 0.4723173975944519, | |
| "learning_rate": 1.0327685879457387e-05, | |
| "loss": 2.3053, | |
| "step": 36100 | |
| }, | |
| { | |
| "epoch": 0.41439626356518156, | |
| "grad_norm": 0.4553997218608856, | |
| "learning_rate": 1.035630473355847e-05, | |
| "loss": 2.3143, | |
| "step": 36200 | |
| }, | |
| { | |
| "epoch": 0.4155410046247539, | |
| "grad_norm": 0.43542203307151794, | |
| "learning_rate": 1.038492358765955e-05, | |
| "loss": 2.32, | |
| "step": 36300 | |
| }, | |
| { | |
| "epoch": 0.4166857456843262, | |
| "grad_norm": 0.5056464076042175, | |
| "learning_rate": 1.0413542441760633e-05, | |
| "loss": 2.3055, | |
| "step": 36400 | |
| }, | |
| { | |
| "epoch": 0.41783048674389855, | |
| "grad_norm": 0.424907922744751, | |
| "learning_rate": 1.0442161295861715e-05, | |
| "loss": 2.312, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 0.41897522780347085, | |
| "grad_norm": 0.4522388279438019, | |
| "learning_rate": 1.0470780149962796e-05, | |
| "loss": 2.304, | |
| "step": 36600 | |
| }, | |
| { | |
| "epoch": 0.4201199688630432, | |
| "grad_norm": 0.42723220586776733, | |
| "learning_rate": 1.0499399004063876e-05, | |
| "loss": 2.3128, | |
| "step": 36700 | |
| }, | |
| { | |
| "epoch": 0.4212647099226155, | |
| "grad_norm": 0.4533430337905884, | |
| "learning_rate": 1.052801785816496e-05, | |
| "loss": 2.3273, | |
| "step": 36800 | |
| }, | |
| { | |
| "epoch": 0.42240945098218785, | |
| "grad_norm": 0.4739341735839844, | |
| "learning_rate": 1.0556636712266041e-05, | |
| "loss": 2.3114, | |
| "step": 36900 | |
| }, | |
| { | |
| "epoch": 0.42355419204176015, | |
| "grad_norm": 0.43692803382873535, | |
| "learning_rate": 1.0585255566367124e-05, | |
| "loss": 2.3093, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 0.4246989331013325, | |
| "grad_norm": 0.437219500541687, | |
| "learning_rate": 1.0613874420468204e-05, | |
| "loss": 2.2975, | |
| "step": 37100 | |
| }, | |
| { | |
| "epoch": 0.4258436741609048, | |
| "grad_norm": 0.5384771227836609, | |
| "learning_rate": 1.0642493274569288e-05, | |
| "loss": 2.3272, | |
| "step": 37200 | |
| }, | |
| { | |
| "epoch": 0.42698841522047715, | |
| "grad_norm": 0.3731122314929962, | |
| "learning_rate": 1.0671112128670369e-05, | |
| "loss": 2.3239, | |
| "step": 37300 | |
| }, | |
| { | |
| "epoch": 0.42813315628004944, | |
| "grad_norm": 0.4134446978569031, | |
| "learning_rate": 1.069973098277145e-05, | |
| "loss": 2.2992, | |
| "step": 37400 | |
| }, | |
| { | |
| "epoch": 0.4292778973396218, | |
| "grad_norm": 0.3970800042152405, | |
| "learning_rate": 1.0728349836872532e-05, | |
| "loss": 2.3166, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 0.4304226383991941, | |
| "grad_norm": 0.4207904040813446, | |
| "learning_rate": 1.0756968690973615e-05, | |
| "loss": 2.2956, | |
| "step": 37600 | |
| }, | |
| { | |
| "epoch": 0.43156737945876644, | |
| "grad_norm": 0.43248116970062256, | |
| "learning_rate": 1.0785587545074695e-05, | |
| "loss": 2.2988, | |
| "step": 37700 | |
| }, | |
| { | |
| "epoch": 0.43271212051833874, | |
| "grad_norm": 0.5825778245925903, | |
| "learning_rate": 1.0814206399175778e-05, | |
| "loss": 2.3124, | |
| "step": 37800 | |
| }, | |
| { | |
| "epoch": 0.4338568615779111, | |
| "grad_norm": 0.4129347503185272, | |
| "learning_rate": 1.084282525327686e-05, | |
| "loss": 2.3045, | |
| "step": 37900 | |
| }, | |
| { | |
| "epoch": 0.4350016026374834, | |
| "grad_norm": 0.5687834024429321, | |
| "learning_rate": 1.0871444107377943e-05, | |
| "loss": 2.3107, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 0.43614634369705574, | |
| "grad_norm": 0.4236217737197876, | |
| "learning_rate": 1.0900062961479023e-05, | |
| "loss": 2.3019, | |
| "step": 38100 | |
| }, | |
| { | |
| "epoch": 0.43729108475662803, | |
| "grad_norm": 0.4377936124801636, | |
| "learning_rate": 1.0928681815580104e-05, | |
| "loss": 2.3093, | |
| "step": 38200 | |
| }, | |
| { | |
| "epoch": 0.4384358258162004, | |
| "grad_norm": 0.7092427611351013, | |
| "learning_rate": 1.0957300669681188e-05, | |
| "loss": 2.3122, | |
| "step": 38300 | |
| }, | |
| { | |
| "epoch": 0.4395805668757727, | |
| "grad_norm": 0.5125077366828918, | |
| "learning_rate": 1.0985919523782269e-05, | |
| "loss": 2.3101, | |
| "step": 38400 | |
| }, | |
| { | |
| "epoch": 0.44072530793534503, | |
| "grad_norm": 0.5460866093635559, | |
| "learning_rate": 1.1014252189342341e-05, | |
| "loss": 2.3059, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 0.44187004899491733, | |
| "grad_norm": 0.5048667788505554, | |
| "learning_rate": 1.1042871043443422e-05, | |
| "loss": 2.2934, | |
| "step": 38600 | |
| }, | |
| { | |
| "epoch": 0.4430147900544897, | |
| "grad_norm": 0.505024790763855, | |
| "learning_rate": 1.1071489897544502e-05, | |
| "loss": 2.3106, | |
| "step": 38700 | |
| }, | |
| { | |
| "epoch": 0.444159531114062, | |
| "grad_norm": 0.4877238869667053, | |
| "learning_rate": 1.1100108751645585e-05, | |
| "loss": 2.3108, | |
| "step": 38800 | |
| }, | |
| { | |
| "epoch": 0.44530427217363433, | |
| "grad_norm": 0.43280166387557983, | |
| "learning_rate": 1.1128727605746667e-05, | |
| "loss": 2.2922, | |
| "step": 38900 | |
| }, | |
| { | |
| "epoch": 0.4464490132332066, | |
| "grad_norm": 0.47969332337379456, | |
| "learning_rate": 1.115734645984775e-05, | |
| "loss": 2.3093, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 0.447593754292779, | |
| "grad_norm": 0.45905813574790955, | |
| "learning_rate": 1.118596531394883e-05, | |
| "loss": 2.302, | |
| "step": 39100 | |
| }, | |
| { | |
| "epoch": 0.4487384953523513, | |
| "grad_norm": 0.4151560962200165, | |
| "learning_rate": 1.1214584168049911e-05, | |
| "loss": 2.3278, | |
| "step": 39200 | |
| }, | |
| { | |
| "epoch": 0.4498832364119236, | |
| "grad_norm": 0.47377634048461914, | |
| "learning_rate": 1.1243203022150995e-05, | |
| "loss": 2.3144, | |
| "step": 39300 | |
| }, | |
| { | |
| "epoch": 0.4510279774714959, | |
| "grad_norm": 0.5002289414405823, | |
| "learning_rate": 1.1271821876252076e-05, | |
| "loss": 2.3065, | |
| "step": 39400 | |
| }, | |
| { | |
| "epoch": 0.4521727185310683, | |
| "grad_norm": 0.48427700996398926, | |
| "learning_rate": 1.1300440730353156e-05, | |
| "loss": 2.3175, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 0.45331745959064057, | |
| "grad_norm": 0.41482630372047424, | |
| "learning_rate": 1.1329059584454239e-05, | |
| "loss": 2.3115, | |
| "step": 39600 | |
| }, | |
| { | |
| "epoch": 0.4544622006502129, | |
| "grad_norm": 0.45701828598976135, | |
| "learning_rate": 1.1357678438555321e-05, | |
| "loss": 2.3166, | |
| "step": 39700 | |
| }, | |
| { | |
| "epoch": 0.4556069417097853, | |
| "grad_norm": 0.4916311502456665, | |
| "learning_rate": 1.1386297292656404e-05, | |
| "loss": 2.3017, | |
| "step": 39800 | |
| }, | |
| { | |
| "epoch": 0.45675168276935757, | |
| "grad_norm": 0.4516671299934387, | |
| "learning_rate": 1.1414916146757484e-05, | |
| "loss": 2.2997, | |
| "step": 39900 | |
| }, | |
| { | |
| "epoch": 0.4578964238289299, | |
| "grad_norm": 0.46147215366363525, | |
| "learning_rate": 1.1443535000858565e-05, | |
| "loss": 2.3052, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 0.4590411648885022, | |
| "grad_norm": 0.48540788888931274, | |
| "learning_rate": 1.1472153854959649e-05, | |
| "loss": 2.3032, | |
| "step": 40100 | |
| }, | |
| { | |
| "epoch": 0.46018590594807457, | |
| "grad_norm": 0.47240906953811646, | |
| "learning_rate": 1.150077270906073e-05, | |
| "loss": 2.2986, | |
| "step": 40200 | |
| }, | |
| { | |
| "epoch": 0.46133064700764687, | |
| "grad_norm": 0.5069533586502075, | |
| "learning_rate": 1.1529391563161812e-05, | |
| "loss": 2.3019, | |
| "step": 40300 | |
| }, | |
| { | |
| "epoch": 0.4624753880672192, | |
| "grad_norm": 0.5581790208816528, | |
| "learning_rate": 1.1558010417262893e-05, | |
| "loss": 2.297, | |
| "step": 40400 | |
| }, | |
| { | |
| "epoch": 0.4636201291267915, | |
| "grad_norm": 0.42001545429229736, | |
| "learning_rate": 1.1586629271363975e-05, | |
| "loss": 2.3023, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 0.46476487018636387, | |
| "grad_norm": 0.43199682235717773, | |
| "learning_rate": 1.1615248125465058e-05, | |
| "loss": 2.307, | |
| "step": 40600 | |
| }, | |
| { | |
| "epoch": 0.46590961124593616, | |
| "grad_norm": 0.39316660165786743, | |
| "learning_rate": 1.1643580791025128e-05, | |
| "loss": 2.3006, | |
| "step": 40700 | |
| }, | |
| { | |
| "epoch": 0.4670543523055085, | |
| "grad_norm": 0.4300936162471771, | |
| "learning_rate": 1.167219964512621e-05, | |
| "loss": 2.2991, | |
| "step": 40800 | |
| }, | |
| { | |
| "epoch": 0.4681990933650808, | |
| "grad_norm": 0.3927803933620453, | |
| "learning_rate": 1.1700818499227291e-05, | |
| "loss": 2.302, | |
| "step": 40900 | |
| }, | |
| { | |
| "epoch": 0.46934383442465316, | |
| "grad_norm": 0.4072405993938446, | |
| "learning_rate": 1.1729437353328375e-05, | |
| "loss": 2.3233, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 0.47048857548422546, | |
| "grad_norm": 0.5310955047607422, | |
| "learning_rate": 1.1758056207429456e-05, | |
| "loss": 2.2919, | |
| "step": 41100 | |
| }, | |
| { | |
| "epoch": 0.4716333165437978, | |
| "grad_norm": 0.48148131370544434, | |
| "learning_rate": 1.1786675061530537e-05, | |
| "loss": 2.2854, | |
| "step": 41200 | |
| }, | |
| { | |
| "epoch": 0.4727780576033701, | |
| "grad_norm": 0.4595455527305603, | |
| "learning_rate": 1.1815293915631617e-05, | |
| "loss": 2.2966, | |
| "step": 41300 | |
| }, | |
| { | |
| "epoch": 0.47392279866294246, | |
| "grad_norm": 0.41068974137306213, | |
| "learning_rate": 1.1843912769732702e-05, | |
| "loss": 2.3041, | |
| "step": 41400 | |
| }, | |
| { | |
| "epoch": 0.47506753972251475, | |
| "grad_norm": 0.48817092180252075, | |
| "learning_rate": 1.1872531623833782e-05, | |
| "loss": 2.3031, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 0.4762122807820871, | |
| "grad_norm": 0.4430725872516632, | |
| "learning_rate": 1.1901150477934865e-05, | |
| "loss": 2.306, | |
| "step": 41600 | |
| }, | |
| { | |
| "epoch": 0.4773570218416594, | |
| "grad_norm": 0.41381534934043884, | |
| "learning_rate": 1.1929769332035945e-05, | |
| "loss": 2.2938, | |
| "step": 41700 | |
| }, | |
| { | |
| "epoch": 0.47850176290123175, | |
| "grad_norm": 0.5231103897094727, | |
| "learning_rate": 1.195838818613703e-05, | |
| "loss": 2.287, | |
| "step": 41800 | |
| }, | |
| { | |
| "epoch": 0.47964650396080405, | |
| "grad_norm": 0.45451679825782776, | |
| "learning_rate": 1.198700704023811e-05, | |
| "loss": 2.3014, | |
| "step": 41900 | |
| }, | |
| { | |
| "epoch": 0.4807912450203764, | |
| "grad_norm": 0.4158308804035187, | |
| "learning_rate": 1.2015625894339191e-05, | |
| "loss": 2.2863, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 0.4819359860799487, | |
| "grad_norm": 0.4745527505874634, | |
| "learning_rate": 1.2044244748440273e-05, | |
| "loss": 2.3069, | |
| "step": 42100 | |
| }, | |
| { | |
| "epoch": 0.48308072713952105, | |
| "grad_norm": 0.4103641211986542, | |
| "learning_rate": 1.2072863602541356e-05, | |
| "loss": 2.3, | |
| "step": 42200 | |
| }, | |
| { | |
| "epoch": 0.48422546819909335, | |
| "grad_norm": 0.4857043921947479, | |
| "learning_rate": 1.2101482456642436e-05, | |
| "loss": 2.2829, | |
| "step": 42300 | |
| }, | |
| { | |
| "epoch": 0.4853702092586657, | |
| "grad_norm": 0.4224902093410492, | |
| "learning_rate": 1.2130101310743519e-05, | |
| "loss": 2.3012, | |
| "step": 42400 | |
| }, | |
| { | |
| "epoch": 0.486514950318238, | |
| "grad_norm": 0.4603799283504486, | |
| "learning_rate": 1.21587201648446e-05, | |
| "loss": 2.29, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 0.48765969137781034, | |
| "grad_norm": 0.43657830357551575, | |
| "learning_rate": 1.2187339018945684e-05, | |
| "loss": 2.3105, | |
| "step": 42600 | |
| }, | |
| { | |
| "epoch": 0.48880443243738264, | |
| "grad_norm": 0.411188006401062, | |
| "learning_rate": 1.2215957873046764e-05, | |
| "loss": 2.2911, | |
| "step": 42700 | |
| }, | |
| { | |
| "epoch": 0.489949173496955, | |
| "grad_norm": 0.4367277920246124, | |
| "learning_rate": 1.2244576727147845e-05, | |
| "loss": 2.3071, | |
| "step": 42800 | |
| }, | |
| { | |
| "epoch": 0.4910939145565273, | |
| "grad_norm": 0.4958134889602661, | |
| "learning_rate": 1.2272909392707917e-05, | |
| "loss": 2.3065, | |
| "step": 42900 | |
| }, | |
| { | |
| "epoch": 0.49223865561609964, | |
| "grad_norm": 0.4951634705066681, | |
| "learning_rate": 1.2301528246808998e-05, | |
| "loss": 2.2884, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 0.49338339667567194, | |
| "grad_norm": 0.44968706369400024, | |
| "learning_rate": 1.2330147100910082e-05, | |
| "loss": 2.2957, | |
| "step": 43100 | |
| }, | |
| { | |
| "epoch": 0.4945281377352443, | |
| "grad_norm": 0.45405635237693787, | |
| "learning_rate": 1.2358765955011163e-05, | |
| "loss": 2.2915, | |
| "step": 43200 | |
| }, | |
| { | |
| "epoch": 0.4956728787948166, | |
| "grad_norm": 0.5005086660385132, | |
| "learning_rate": 1.2387384809112243e-05, | |
| "loss": 2.3055, | |
| "step": 43300 | |
| }, | |
| { | |
| "epoch": 0.49681761985438894, | |
| "grad_norm": 0.5079677104949951, | |
| "learning_rate": 1.2416003663213326e-05, | |
| "loss": 2.3047, | |
| "step": 43400 | |
| }, | |
| { | |
| "epoch": 0.4979623609139613, | |
| "grad_norm": 0.47394371032714844, | |
| "learning_rate": 1.2444336328773396e-05, | |
| "loss": 2.2976, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 0.4991071019735336, | |
| "grad_norm": 0.5010650157928467, | |
| "learning_rate": 1.2472955182874477e-05, | |
| "loss": 2.2843, | |
| "step": 43600 | |
| }, | |
| { | |
| "epoch": 0.5002518430331059, | |
| "grad_norm": 0.443935751914978, | |
| "learning_rate": 1.2501574036975561e-05, | |
| "loss": 2.3022, | |
| "step": 43700 | |
| }, | |
| { | |
| "epoch": 0.5013965840926783, | |
| "grad_norm": 0.44586101174354553, | |
| "learning_rate": 1.2530192891076642e-05, | |
| "loss": 2.3024, | |
| "step": 43800 | |
| }, | |
| { | |
| "epoch": 0.5025413251522506, | |
| "grad_norm": 0.4664213955402374, | |
| "learning_rate": 1.2558811745177724e-05, | |
| "loss": 2.2865, | |
| "step": 43900 | |
| }, | |
| { | |
| "epoch": 0.5036860662118229, | |
| "grad_norm": 0.4366970956325531, | |
| "learning_rate": 1.2587430599278805e-05, | |
| "loss": 2.3052, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 0.5048308072713952, | |
| "grad_norm": 0.44286203384399414, | |
| "learning_rate": 1.2616049453379889e-05, | |
| "loss": 2.3006, | |
| "step": 44100 | |
| }, | |
| { | |
| "epoch": 0.5059755483309676, | |
| "grad_norm": 0.4843718707561493, | |
| "learning_rate": 1.264466830748097e-05, | |
| "loss": 2.2882, | |
| "step": 44200 | |
| }, | |
| { | |
| "epoch": 0.5071202893905399, | |
| "grad_norm": 0.4327425956726074, | |
| "learning_rate": 1.267328716158205e-05, | |
| "loss": 2.3146, | |
| "step": 44300 | |
| }, | |
| { | |
| "epoch": 0.5082650304501122, | |
| "grad_norm": 0.45917651057243347, | |
| "learning_rate": 1.2701906015683134e-05, | |
| "loss": 2.2763, | |
| "step": 44400 | |
| }, | |
| { | |
| "epoch": 0.5094097715096845, | |
| "grad_norm": 0.4044801890850067, | |
| "learning_rate": 1.2730524869784215e-05, | |
| "loss": 2.2919, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 0.5105545125692569, | |
| "grad_norm": 0.43837985396385193, | |
| "learning_rate": 1.2759143723885296e-05, | |
| "loss": 2.3, | |
| "step": 44600 | |
| }, | |
| { | |
| "epoch": 0.5116992536288292, | |
| "grad_norm": 0.5559438467025757, | |
| "learning_rate": 1.2787762577986378e-05, | |
| "loss": 2.2841, | |
| "step": 44700 | |
| }, | |
| { | |
| "epoch": 0.5128439946884015, | |
| "grad_norm": 0.5006335377693176, | |
| "learning_rate": 1.281638143208746e-05, | |
| "loss": 2.3024, | |
| "step": 44800 | |
| }, | |
| { | |
| "epoch": 0.5139887357479738, | |
| "grad_norm": 0.5347406268119812, | |
| "learning_rate": 1.2845000286188543e-05, | |
| "loss": 2.2915, | |
| "step": 44900 | |
| }, | |
| { | |
| "epoch": 0.5151334768075462, | |
| "grad_norm": 0.49963897466659546, | |
| "learning_rate": 1.2873619140289624e-05, | |
| "loss": 2.3094, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 0.5162782178671185, | |
| "grad_norm": 0.4746800661087036, | |
| "learning_rate": 1.2902237994390704e-05, | |
| "loss": 2.288, | |
| "step": 45100 | |
| }, | |
| { | |
| "epoch": 0.5174229589266908, | |
| "grad_norm": 0.44089171290397644, | |
| "learning_rate": 1.2930856848491788e-05, | |
| "loss": 2.2791, | |
| "step": 45200 | |
| }, | |
| { | |
| "epoch": 0.5185676999862631, | |
| "grad_norm": 0.46968016028404236, | |
| "learning_rate": 1.295947570259287e-05, | |
| "loss": 2.3088, | |
| "step": 45300 | |
| }, | |
| { | |
| "epoch": 0.5197124410458355, | |
| "grad_norm": 0.45375433564186096, | |
| "learning_rate": 1.2988094556693952e-05, | |
| "loss": 2.3083, | |
| "step": 45400 | |
| }, | |
| { | |
| "epoch": 0.5208571821054078, | |
| "grad_norm": 0.5065542459487915, | |
| "learning_rate": 1.3016713410795032e-05, | |
| "loss": 2.2694, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 0.5220019231649801, | |
| "grad_norm": 0.5144473910331726, | |
| "learning_rate": 1.3045332264896116e-05, | |
| "loss": 2.3044, | |
| "step": 45600 | |
| }, | |
| { | |
| "epoch": 0.5231466642245524, | |
| "grad_norm": 0.5982611179351807, | |
| "learning_rate": 1.3073951118997197e-05, | |
| "loss": 2.2964, | |
| "step": 45700 | |
| }, | |
| { | |
| "epoch": 0.5242914052841248, | |
| "grad_norm": 0.42570099234580994, | |
| "learning_rate": 1.3102569973098278e-05, | |
| "loss": 2.2898, | |
| "step": 45800 | |
| }, | |
| { | |
| "epoch": 0.5254361463436971, | |
| "grad_norm": 0.5816085934638977, | |
| "learning_rate": 1.3131188827199358e-05, | |
| "loss": 2.2974, | |
| "step": 45900 | |
| }, | |
| { | |
| "epoch": 0.5265808874032694, | |
| "grad_norm": 0.5254452228546143, | |
| "learning_rate": 1.3159807681300443e-05, | |
| "loss": 2.3149, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 0.5277256284628417, | |
| "grad_norm": 0.43442779779434204, | |
| "learning_rate": 1.3188426535401523e-05, | |
| "loss": 2.2951, | |
| "step": 46100 | |
| }, | |
| { | |
| "epoch": 0.5288703695224141, | |
| "grad_norm": 0.4493260085582733, | |
| "learning_rate": 1.3217045389502606e-05, | |
| "loss": 2.2768, | |
| "step": 46200 | |
| }, | |
| { | |
| "epoch": 0.5300151105819864, | |
| "grad_norm": 0.4513915777206421, | |
| "learning_rate": 1.3245378055062676e-05, | |
| "loss": 2.3075, | |
| "step": 46300 | |
| }, | |
| { | |
| "epoch": 0.5311598516415587, | |
| "grad_norm": 0.45114508271217346, | |
| "learning_rate": 1.3273996909163757e-05, | |
| "loss": 2.2895, | |
| "step": 46400 | |
| }, | |
| { | |
| "epoch": 0.532304592701131, | |
| "grad_norm": 0.43823984265327454, | |
| "learning_rate": 1.330261576326484e-05, | |
| "loss": 2.3087, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 0.5334493337607034, | |
| "grad_norm": 0.453106164932251, | |
| "learning_rate": 1.3331234617365922e-05, | |
| "loss": 2.2851, | |
| "step": 46600 | |
| }, | |
| { | |
| "epoch": 0.5345940748202757, | |
| "grad_norm": 0.46690353751182556, | |
| "learning_rate": 1.3359853471467004e-05, | |
| "loss": 2.2914, | |
| "step": 46700 | |
| }, | |
| { | |
| "epoch": 0.535738815879848, | |
| "grad_norm": 0.46535834670066833, | |
| "learning_rate": 1.3388472325568085e-05, | |
| "loss": 2.2938, | |
| "step": 46800 | |
| }, | |
| { | |
| "epoch": 0.5368835569394202, | |
| "grad_norm": 0.45468568801879883, | |
| "learning_rate": 1.3417091179669165e-05, | |
| "loss": 2.2974, | |
| "step": 46900 | |
| }, | |
| { | |
| "epoch": 0.5380282979989927, | |
| "grad_norm": 0.4835493862628937, | |
| "learning_rate": 1.344571003377025e-05, | |
| "loss": 2.2882, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 0.539173039058565, | |
| "grad_norm": 0.45315301418304443, | |
| "learning_rate": 1.347432888787133e-05, | |
| "loss": 2.2875, | |
| "step": 47100 | |
| }, | |
| { | |
| "epoch": 0.5403177801181372, | |
| "grad_norm": 0.5241557359695435, | |
| "learning_rate": 1.3502947741972413e-05, | |
| "loss": 2.2927, | |
| "step": 47200 | |
| }, | |
| { | |
| "epoch": 0.5414625211777097, | |
| "grad_norm": 0.4486404061317444, | |
| "learning_rate": 1.3531566596073495e-05, | |
| "loss": 2.2898, | |
| "step": 47300 | |
| }, | |
| { | |
| "epoch": 0.542607262237282, | |
| "grad_norm": 0.4849669933319092, | |
| "learning_rate": 1.3560185450174577e-05, | |
| "loss": 2.283, | |
| "step": 47400 | |
| }, | |
| { | |
| "epoch": 0.5437520032968542, | |
| "grad_norm": 0.6526544690132141, | |
| "learning_rate": 1.3588804304275658e-05, | |
| "loss": 2.287, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 0.5448967443564265, | |
| "grad_norm": 0.4628201723098755, | |
| "learning_rate": 1.3617423158376739e-05, | |
| "loss": 2.3099, | |
| "step": 47600 | |
| }, | |
| { | |
| "epoch": 0.546041485415999, | |
| "grad_norm": 0.5132496356964111, | |
| "learning_rate": 1.3646042012477823e-05, | |
| "loss": 2.2969, | |
| "step": 47700 | |
| }, | |
| { | |
| "epoch": 0.5471862264755712, | |
| "grad_norm": 0.545789897441864, | |
| "learning_rate": 1.3674660866578904e-05, | |
| "loss": 2.2948, | |
| "step": 47800 | |
| }, | |
| { | |
| "epoch": 0.5483309675351435, | |
| "grad_norm": 0.5407856106758118, | |
| "learning_rate": 1.3703279720679984e-05, | |
| "loss": 2.2861, | |
| "step": 47900 | |
| }, | |
| { | |
| "epoch": 0.5494757085947158, | |
| "grad_norm": 0.494488000869751, | |
| "learning_rate": 1.3731898574781067e-05, | |
| "loss": 2.2687, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 0.5506204496542882, | |
| "grad_norm": 0.44262704253196716, | |
| "learning_rate": 1.3760517428882149e-05, | |
| "loss": 2.2879, | |
| "step": 48100 | |
| }, | |
| { | |
| "epoch": 0.5517651907138605, | |
| "grad_norm": 0.4556616544723511, | |
| "learning_rate": 1.3789136282983232e-05, | |
| "loss": 2.2733, | |
| "step": 48200 | |
| }, | |
| { | |
| "epoch": 0.5529099317734328, | |
| "grad_norm": 0.5023077130317688, | |
| "learning_rate": 1.3817755137084312e-05, | |
| "loss": 2.284, | |
| "step": 48300 | |
| }, | |
| { | |
| "epoch": 0.5540546728330051, | |
| "grad_norm": 0.44959184527397156, | |
| "learning_rate": 1.3846373991185393e-05, | |
| "loss": 2.2821, | |
| "step": 48400 | |
| }, | |
| { | |
| "epoch": 0.5551994138925775, | |
| "grad_norm": 0.6102599501609802, | |
| "learning_rate": 1.3874992845286477e-05, | |
| "loss": 2.2889, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 0.5563441549521498, | |
| "grad_norm": 0.508794367313385, | |
| "learning_rate": 1.3903611699387558e-05, | |
| "loss": 2.2766, | |
| "step": 48600 | |
| }, | |
| { | |
| "epoch": 0.5574888960117221, | |
| "grad_norm": 0.5081732869148254, | |
| "learning_rate": 1.3932230553488638e-05, | |
| "loss": 2.2843, | |
| "step": 48700 | |
| }, | |
| { | |
| "epoch": 0.5586336370712944, | |
| "grad_norm": 0.4801699221134186, | |
| "learning_rate": 1.396084940758972e-05, | |
| "loss": 2.2749, | |
| "step": 48800 | |
| }, | |
| { | |
| "epoch": 0.5597783781308668, | |
| "grad_norm": 0.5260947346687317, | |
| "learning_rate": 1.3989468261690803e-05, | |
| "loss": 2.2883, | |
| "step": 48900 | |
| }, | |
| { | |
| "epoch": 0.5609231191904391, | |
| "grad_norm": 0.44729700684547424, | |
| "learning_rate": 1.4018087115791886e-05, | |
| "loss": 2.2889, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 0.5620678602500114, | |
| "grad_norm": 0.4468446671962738, | |
| "learning_rate": 1.4046705969892966e-05, | |
| "loss": 2.2892, | |
| "step": 49100 | |
| }, | |
| { | |
| "epoch": 0.5632126013095837, | |
| "grad_norm": 0.4718739688396454, | |
| "learning_rate": 1.4075324823994047e-05, | |
| "loss": 2.2894, | |
| "step": 49200 | |
| }, | |
| { | |
| "epoch": 0.5643573423691561, | |
| "grad_norm": 0.5241585373878479, | |
| "learning_rate": 1.4103943678095131e-05, | |
| "loss": 2.2922, | |
| "step": 49300 | |
| }, | |
| { | |
| "epoch": 0.5655020834287284, | |
| "grad_norm": 0.43622729182243347, | |
| "learning_rate": 1.4132562532196212e-05, | |
| "loss": 2.2816, | |
| "step": 49400 | |
| }, | |
| { | |
| "epoch": 0.5666468244883007, | |
| "grad_norm": 0.4169420599937439, | |
| "learning_rate": 1.4161181386297294e-05, | |
| "loss": 2.3186, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 0.567791565547873, | |
| "grad_norm": 0.4381948411464691, | |
| "learning_rate": 1.4189800240398375e-05, | |
| "loss": 2.2903, | |
| "step": 49600 | |
| }, | |
| { | |
| "epoch": 0.5689363066074454, | |
| "grad_norm": 0.5486495494842529, | |
| "learning_rate": 1.4218419094499459e-05, | |
| "loss": 2.2964, | |
| "step": 49700 | |
| }, | |
| { | |
| "epoch": 0.5700810476670177, | |
| "grad_norm": 0.42850059270858765, | |
| "learning_rate": 1.424703794860054e-05, | |
| "loss": 2.29, | |
| "step": 49800 | |
| }, | |
| { | |
| "epoch": 0.57122578872659, | |
| "grad_norm": 0.4907155930995941, | |
| "learning_rate": 1.427565680270162e-05, | |
| "loss": 2.3085, | |
| "step": 49900 | |
| }, | |
| { | |
| "epoch": 0.5723705297861623, | |
| "grad_norm": 0.43422576785087585, | |
| "learning_rate": 1.4304275656802701e-05, | |
| "loss": 2.2787, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 0.5735152708457347, | |
| "grad_norm": 0.4992702007293701, | |
| "learning_rate": 1.4332894510903785e-05, | |
| "loss": 2.287, | |
| "step": 50100 | |
| }, | |
| { | |
| "epoch": 0.574660011905307, | |
| "grad_norm": 0.4858098030090332, | |
| "learning_rate": 1.4361227176463857e-05, | |
| "loss": 2.2761, | |
| "step": 50200 | |
| }, | |
| { | |
| "epoch": 0.5758047529648793, | |
| "grad_norm": 0.48108112812042236, | |
| "learning_rate": 1.4389846030564938e-05, | |
| "loss": 2.2883, | |
| "step": 50300 | |
| }, | |
| { | |
| "epoch": 0.5769494940244516, | |
| "grad_norm": 0.38939031958580017, | |
| "learning_rate": 1.4418464884666019e-05, | |
| "loss": 2.2837, | |
| "step": 50400 | |
| }, | |
| { | |
| "epoch": 0.578094235084024, | |
| "grad_norm": 0.488679438829422, | |
| "learning_rate": 1.44470837387671e-05, | |
| "loss": 2.298, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 0.5792389761435963, | |
| "grad_norm": 0.5358524918556213, | |
| "learning_rate": 1.4475702592868184e-05, | |
| "loss": 2.2897, | |
| "step": 50600 | |
| }, | |
| { | |
| "epoch": 0.5803837172031686, | |
| "grad_norm": 0.48244425654411316, | |
| "learning_rate": 1.4504321446969264e-05, | |
| "loss": 2.2748, | |
| "step": 50700 | |
| }, | |
| { | |
| "epoch": 0.581528458262741, | |
| "grad_norm": 0.49125197529792786, | |
| "learning_rate": 1.4532940301070347e-05, | |
| "loss": 2.2666, | |
| "step": 50800 | |
| }, | |
| { | |
| "epoch": 0.5826731993223133, | |
| "grad_norm": 0.5043622851371765, | |
| "learning_rate": 1.4561559155171427e-05, | |
| "loss": 2.2948, | |
| "step": 50900 | |
| }, | |
| { | |
| "epoch": 0.5838179403818856, | |
| "grad_norm": 0.4543743431568146, | |
| "learning_rate": 1.4590178009272512e-05, | |
| "loss": 2.2801, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 0.5849626814414579, | |
| "grad_norm": 0.45934557914733887, | |
| "learning_rate": 1.4618796863373592e-05, | |
| "loss": 2.2897, | |
| "step": 51100 | |
| }, | |
| { | |
| "epoch": 0.5861074225010303, | |
| "grad_norm": 0.48373672366142273, | |
| "learning_rate": 1.4647415717474673e-05, | |
| "loss": 2.2808, | |
| "step": 51200 | |
| }, | |
| { | |
| "epoch": 0.5872521635606026, | |
| "grad_norm": 0.42684435844421387, | |
| "learning_rate": 1.4676034571575755e-05, | |
| "loss": 2.2707, | |
| "step": 51300 | |
| }, | |
| { | |
| "epoch": 0.5883969046201749, | |
| "grad_norm": 0.439179927110672, | |
| "learning_rate": 1.4704653425676838e-05, | |
| "loss": 2.2887, | |
| "step": 51400 | |
| }, | |
| { | |
| "epoch": 0.5895416456797472, | |
| "grad_norm": 0.48160520195961, | |
| "learning_rate": 1.473327227977792e-05, | |
| "loss": 2.282, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 0.5906863867393196, | |
| "grad_norm": 0.48224136233329773, | |
| "learning_rate": 1.4761891133879e-05, | |
| "loss": 2.2766, | |
| "step": 51600 | |
| }, | |
| { | |
| "epoch": 0.5918311277988919, | |
| "grad_norm": 0.46199363470077515, | |
| "learning_rate": 1.4790509987980082e-05, | |
| "loss": 2.2836, | |
| "step": 51700 | |
| }, | |
| { | |
| "epoch": 0.5929758688584642, | |
| "grad_norm": 0.4785059690475464, | |
| "learning_rate": 1.4819128842081166e-05, | |
| "loss": 2.2865, | |
| "step": 51800 | |
| }, | |
| { | |
| "epoch": 0.5941206099180365, | |
| "grad_norm": 0.43915683031082153, | |
| "learning_rate": 1.4847747696182246e-05, | |
| "loss": 2.2785, | |
| "step": 51900 | |
| }, | |
| { | |
| "epoch": 0.5952653509776089, | |
| "grad_norm": 0.5053157806396484, | |
| "learning_rate": 1.4876366550283327e-05, | |
| "loss": 2.2903, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 0.5964100920371812, | |
| "grad_norm": 0.4726928174495697, | |
| "learning_rate": 1.490498540438441e-05, | |
| "loss": 2.2818, | |
| "step": 52100 | |
| }, | |
| { | |
| "epoch": 0.5975548330967535, | |
| "grad_norm": 0.430034875869751, | |
| "learning_rate": 1.4933604258485492e-05, | |
| "loss": 2.29, | |
| "step": 52200 | |
| }, | |
| { | |
| "epoch": 0.5986995741563258, | |
| "grad_norm": 0.4643426835536957, | |
| "learning_rate": 1.4962223112586574e-05, | |
| "loss": 2.2897, | |
| "step": 52300 | |
| }, | |
| { | |
| "epoch": 0.5998443152158982, | |
| "grad_norm": 0.5476269125938416, | |
| "learning_rate": 1.4990841966687655e-05, | |
| "loss": 2.2786, | |
| "step": 52400 | |
| }, | |
| { | |
| "epoch": 0.6009890562754705, | |
| "grad_norm": 0.4216204285621643, | |
| "learning_rate": 1.5019174632247725e-05, | |
| "loss": 2.2814, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 0.6021337973350428, | |
| "grad_norm": 0.4980791211128235, | |
| "learning_rate": 1.5047793486348808e-05, | |
| "loss": 2.2648, | |
| "step": 52600 | |
| }, | |
| { | |
| "epoch": 0.6032785383946151, | |
| "grad_norm": 0.48100781440734863, | |
| "learning_rate": 1.5076412340449888e-05, | |
| "loss": 2.2881, | |
| "step": 52700 | |
| }, | |
| { | |
| "epoch": 0.6044232794541875, | |
| "grad_norm": 0.5112878084182739, | |
| "learning_rate": 1.5105031194550973e-05, | |
| "loss": 2.2761, | |
| "step": 52800 | |
| }, | |
| { | |
| "epoch": 0.6055680205137598, | |
| "grad_norm": 0.4899493455886841, | |
| "learning_rate": 1.5133650048652053e-05, | |
| "loss": 2.2852, | |
| "step": 52900 | |
| }, | |
| { | |
| "epoch": 0.6067127615733321, | |
| "grad_norm": 0.486299067735672, | |
| "learning_rate": 1.5162268902753134e-05, | |
| "loss": 2.2958, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 0.6078575026329044, | |
| "grad_norm": 0.580345630645752, | |
| "learning_rate": 1.5190887756854218e-05, | |
| "loss": 2.2716, | |
| "step": 53100 | |
| }, | |
| { | |
| "epoch": 0.6090022436924768, | |
| "grad_norm": 0.4456554055213928, | |
| "learning_rate": 1.5219506610955299e-05, | |
| "loss": 2.285, | |
| "step": 53200 | |
| }, | |
| { | |
| "epoch": 0.6101469847520491, | |
| "grad_norm": 0.4706750512123108, | |
| "learning_rate": 1.5248125465056381e-05, | |
| "loss": 2.2968, | |
| "step": 53300 | |
| }, | |
| { | |
| "epoch": 0.6112917258116214, | |
| "grad_norm": 0.47107580304145813, | |
| "learning_rate": 1.527674431915746e-05, | |
| "loss": 2.2733, | |
| "step": 53400 | |
| }, | |
| { | |
| "epoch": 0.6124364668711937, | |
| "grad_norm": 0.45870354771614075, | |
| "learning_rate": 1.5305363173258546e-05, | |
| "loss": 2.2962, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 0.6135812079307661, | |
| "grad_norm": 0.526592493057251, | |
| "learning_rate": 1.5333982027359625e-05, | |
| "loss": 2.28, | |
| "step": 53600 | |
| }, | |
| { | |
| "epoch": 0.6147259489903384, | |
| "grad_norm": 0.4595036506652832, | |
| "learning_rate": 1.5362600881460707e-05, | |
| "loss": 2.2751, | |
| "step": 53700 | |
| }, | |
| { | |
| "epoch": 0.6158706900499107, | |
| "grad_norm": 0.47698622941970825, | |
| "learning_rate": 1.539121973556179e-05, | |
| "loss": 2.2764, | |
| "step": 53800 | |
| }, | |
| { | |
| "epoch": 0.617015431109483, | |
| "grad_norm": 0.47543615102767944, | |
| "learning_rate": 1.5419838589662872e-05, | |
| "loss": 2.269, | |
| "step": 53900 | |
| }, | |
| { | |
| "epoch": 0.6181601721690554, | |
| "grad_norm": 0.46203306317329407, | |
| "learning_rate": 1.5448457443763955e-05, | |
| "loss": 2.3127, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 0.6193049132286277, | |
| "grad_norm": 0.4622338116168976, | |
| "learning_rate": 1.5477076297865034e-05, | |
| "loss": 2.2795, | |
| "step": 54100 | |
| }, | |
| { | |
| "epoch": 0.6204496542882, | |
| "grad_norm": 0.43615639209747314, | |
| "learning_rate": 1.5505695151966116e-05, | |
| "loss": 2.2751, | |
| "step": 54200 | |
| }, | |
| { | |
| "epoch": 0.6215943953477724, | |
| "grad_norm": 0.4956182837486267, | |
| "learning_rate": 1.55343140060672e-05, | |
| "loss": 2.2888, | |
| "step": 54300 | |
| }, | |
| { | |
| "epoch": 0.6227391364073447, | |
| "grad_norm": 0.44354909658432007, | |
| "learning_rate": 1.556293286016828e-05, | |
| "loss": 2.278, | |
| "step": 54400 | |
| }, | |
| { | |
| "epoch": 0.623883877466917, | |
| "grad_norm": 0.46796587109565735, | |
| "learning_rate": 1.5591551714269363e-05, | |
| "loss": 2.2843, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 0.6250286185264893, | |
| "grad_norm": 0.43353140354156494, | |
| "learning_rate": 1.5620170568370442e-05, | |
| "loss": 2.2814, | |
| "step": 54600 | |
| }, | |
| { | |
| "epoch": 0.6261733595860617, | |
| "grad_norm": 0.47816458344459534, | |
| "learning_rate": 1.5648789422471528e-05, | |
| "loss": 2.2708, | |
| "step": 54700 | |
| }, | |
| { | |
| "epoch": 0.627318100645634, | |
| "grad_norm": 0.4949074387550354, | |
| "learning_rate": 1.5677408276572607e-05, | |
| "loss": 2.2785, | |
| "step": 54800 | |
| }, | |
| { | |
| "epoch": 0.6284628417052063, | |
| "grad_norm": 0.41498610377311707, | |
| "learning_rate": 1.570602713067369e-05, | |
| "loss": 2.2752, | |
| "step": 54900 | |
| }, | |
| { | |
| "epoch": 0.6296075827647786, | |
| "grad_norm": 0.41272154450416565, | |
| "learning_rate": 1.5734645984774772e-05, | |
| "loss": 2.2679, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 0.630752323824351, | |
| "grad_norm": 0.47275310754776, | |
| "learning_rate": 1.5763264838875854e-05, | |
| "loss": 2.2838, | |
| "step": 55100 | |
| }, | |
| { | |
| "epoch": 0.6318970648839233, | |
| "grad_norm": 0.41480526328086853, | |
| "learning_rate": 1.5791883692976933e-05, | |
| "loss": 2.2905, | |
| "step": 55200 | |
| }, | |
| { | |
| "epoch": 0.6330418059434956, | |
| "grad_norm": 0.45607683062553406, | |
| "learning_rate": 1.5820502547078016e-05, | |
| "loss": 2.2793, | |
| "step": 55300 | |
| }, | |
| { | |
| "epoch": 0.6341865470030679, | |
| "grad_norm": 0.4298737645149231, | |
| "learning_rate": 1.5849121401179098e-05, | |
| "loss": 2.2864, | |
| "step": 55400 | |
| }, | |
| { | |
| "epoch": 0.6353312880626403, | |
| "grad_norm": 0.45687663555145264, | |
| "learning_rate": 1.587774025528018e-05, | |
| "loss": 2.2761, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 0.6364760291222126, | |
| "grad_norm": 0.4270581901073456, | |
| "learning_rate": 1.5906359109381263e-05, | |
| "loss": 2.2859, | |
| "step": 55600 | |
| }, | |
| { | |
| "epoch": 0.6376207701817849, | |
| "grad_norm": 0.4622785449028015, | |
| "learning_rate": 1.5934977963482342e-05, | |
| "loss": 2.2751, | |
| "step": 55700 | |
| }, | |
| { | |
| "epoch": 0.6387655112413572, | |
| "grad_norm": 0.4890844523906708, | |
| "learning_rate": 1.5963596817583424e-05, | |
| "loss": 2.2908, | |
| "step": 55800 | |
| }, | |
| { | |
| "epoch": 0.6399102523009296, | |
| "grad_norm": 0.4259001910686493, | |
| "learning_rate": 1.5992215671684507e-05, | |
| "loss": 2.2724, | |
| "step": 55900 | |
| }, | |
| { | |
| "epoch": 0.6410549933605019, | |
| "grad_norm": 0.5524899363517761, | |
| "learning_rate": 1.602083452578559e-05, | |
| "loss": 2.2707, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 0.6421997344200742, | |
| "grad_norm": 0.4600765109062195, | |
| "learning_rate": 1.604945337988667e-05, | |
| "loss": 2.2828, | |
| "step": 56100 | |
| }, | |
| { | |
| "epoch": 0.6433444754796465, | |
| "grad_norm": 0.4435892403125763, | |
| "learning_rate": 1.607807223398775e-05, | |
| "loss": 2.2781, | |
| "step": 56200 | |
| }, | |
| { | |
| "epoch": 0.6444892165392189, | |
| "grad_norm": 0.47321733832359314, | |
| "learning_rate": 1.6106691088088836e-05, | |
| "loss": 2.2494, | |
| "step": 56300 | |
| }, | |
| { | |
| "epoch": 0.6456339575987912, | |
| "grad_norm": 0.530928373336792, | |
| "learning_rate": 1.6135309942189915e-05, | |
| "loss": 2.275, | |
| "step": 56400 | |
| }, | |
| { | |
| "epoch": 0.6467786986583635, | |
| "grad_norm": 0.5238829851150513, | |
| "learning_rate": 1.6163928796290998e-05, | |
| "loss": 2.2823, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 0.6479234397179358, | |
| "grad_norm": 0.5112258791923523, | |
| "learning_rate": 1.619254765039208e-05, | |
| "loss": 2.2724, | |
| "step": 56600 | |
| }, | |
| { | |
| "epoch": 0.6490681807775082, | |
| "grad_norm": 0.5023364424705505, | |
| "learning_rate": 1.6221166504493162e-05, | |
| "loss": 2.2655, | |
| "step": 56700 | |
| }, | |
| { | |
| "epoch": 0.6502129218370805, | |
| "grad_norm": 0.45399200916290283, | |
| "learning_rate": 1.6249785358594245e-05, | |
| "loss": 2.2697, | |
| "step": 56800 | |
| }, | |
| { | |
| "epoch": 0.6513576628966528, | |
| "grad_norm": 0.4420014023780823, | |
| "learning_rate": 1.6278404212695324e-05, | |
| "loss": 2.2772, | |
| "step": 56900 | |
| }, | |
| { | |
| "epoch": 0.652502403956225, | |
| "grad_norm": 0.4523197412490845, | |
| "learning_rate": 1.6307023066796406e-05, | |
| "loss": 2.2743, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 0.6536471450157975, | |
| "grad_norm": 0.5409209728240967, | |
| "learning_rate": 1.633564192089749e-05, | |
| "loss": 2.2727, | |
| "step": 57100 | |
| }, | |
| { | |
| "epoch": 0.6547918860753698, | |
| "grad_norm": 0.5374095439910889, | |
| "learning_rate": 1.636426077499857e-05, | |
| "loss": 2.261, | |
| "step": 57200 | |
| }, | |
| { | |
| "epoch": 0.655936627134942, | |
| "grad_norm": 0.5544825196266174, | |
| "learning_rate": 1.6392879629099653e-05, | |
| "loss": 2.272, | |
| "step": 57300 | |
| }, | |
| { | |
| "epoch": 0.6570813681945143, | |
| "grad_norm": 0.4391005337238312, | |
| "learning_rate": 1.6421498483200732e-05, | |
| "loss": 2.2752, | |
| "step": 57400 | |
| }, | |
| { | |
| "epoch": 0.6582261092540868, | |
| "grad_norm": 0.524519145488739, | |
| "learning_rate": 1.6450117337301815e-05, | |
| "loss": 2.2798, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 0.659370850313659, | |
| "grad_norm": 0.4723650813102722, | |
| "learning_rate": 1.6478736191402897e-05, | |
| "loss": 2.2566, | |
| "step": 57600 | |
| }, | |
| { | |
| "epoch": 0.6605155913732313, | |
| "grad_norm": 0.4081030786037445, | |
| "learning_rate": 1.6507068856962968e-05, | |
| "loss": 2.2812, | |
| "step": 57700 | |
| }, | |
| { | |
| "epoch": 0.6616603324328038, | |
| "grad_norm": 0.4323836863040924, | |
| "learning_rate": 1.653568771106405e-05, | |
| "loss": 2.2651, | |
| "step": 57800 | |
| }, | |
| { | |
| "epoch": 0.662805073492376, | |
| "grad_norm": 0.5252947807312012, | |
| "learning_rate": 1.6564306565165132e-05, | |
| "loss": 2.2754, | |
| "step": 57900 | |
| }, | |
| { | |
| "epoch": 0.6639498145519483, | |
| "grad_norm": 0.4679439663887024, | |
| "learning_rate": 1.6592925419266215e-05, | |
| "loss": 2.2735, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 0.6650945556115206, | |
| "grad_norm": 0.4603148102760315, | |
| "learning_rate": 1.6621544273367297e-05, | |
| "loss": 2.2741, | |
| "step": 58100 | |
| }, | |
| { | |
| "epoch": 0.666239296671093, | |
| "grad_norm": 0.44324785470962524, | |
| "learning_rate": 1.6650163127468376e-05, | |
| "loss": 2.2858, | |
| "step": 58200 | |
| }, | |
| { | |
| "epoch": 0.6673840377306653, | |
| "grad_norm": 0.45351341366767883, | |
| "learning_rate": 1.667878198156946e-05, | |
| "loss": 2.2761, | |
| "step": 58300 | |
| }, | |
| { | |
| "epoch": 0.6685287787902376, | |
| "grad_norm": 0.49566417932510376, | |
| "learning_rate": 1.670740083567054e-05, | |
| "loss": 2.2684, | |
| "step": 58400 | |
| }, | |
| { | |
| "epoch": 0.6696735198498099, | |
| "grad_norm": 0.48700281977653503, | |
| "learning_rate": 1.6736019689771623e-05, | |
| "loss": 2.2683, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 0.6708182609093823, | |
| "grad_norm": 0.47343066334724426, | |
| "learning_rate": 1.6764638543872706e-05, | |
| "loss": 2.2727, | |
| "step": 58600 | |
| }, | |
| { | |
| "epoch": 0.6719630019689546, | |
| "grad_norm": 0.4507409334182739, | |
| "learning_rate": 1.6793257397973785e-05, | |
| "loss": 2.2774, | |
| "step": 58700 | |
| }, | |
| { | |
| "epoch": 0.6731077430285269, | |
| "grad_norm": 0.5125613808631897, | |
| "learning_rate": 1.682187625207487e-05, | |
| "loss": 2.2724, | |
| "step": 58800 | |
| }, | |
| { | |
| "epoch": 0.6742524840880992, | |
| "grad_norm": 0.4266802966594696, | |
| "learning_rate": 1.685049510617595e-05, | |
| "loss": 2.264, | |
| "step": 58900 | |
| }, | |
| { | |
| "epoch": 0.6753972251476716, | |
| "grad_norm": 0.4939129650592804, | |
| "learning_rate": 1.687882777173602e-05, | |
| "loss": 2.2736, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 0.6765419662072439, | |
| "grad_norm": 0.4348323345184326, | |
| "learning_rate": 1.6907446625837102e-05, | |
| "loss": 2.2647, | |
| "step": 59100 | |
| }, | |
| { | |
| "epoch": 0.6776867072668162, | |
| "grad_norm": 0.4148264527320862, | |
| "learning_rate": 1.6936065479938185e-05, | |
| "loss": 2.2731, | |
| "step": 59200 | |
| }, | |
| { | |
| "epoch": 0.6788314483263885, | |
| "grad_norm": 0.5018272399902344, | |
| "learning_rate": 1.6964684334039267e-05, | |
| "loss": 2.2798, | |
| "step": 59300 | |
| }, | |
| { | |
| "epoch": 0.6799761893859609, | |
| "grad_norm": 0.405222624540329, | |
| "learning_rate": 1.699330318814035e-05, | |
| "loss": 2.2716, | |
| "step": 59400 | |
| }, | |
| { | |
| "epoch": 0.6811209304455332, | |
| "grad_norm": 0.43340057134628296, | |
| "learning_rate": 1.702192204224143e-05, | |
| "loss": 2.2616, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 0.6822656715051055, | |
| "grad_norm": 0.4897302985191345, | |
| "learning_rate": 1.705054089634251e-05, | |
| "loss": 2.2937, | |
| "step": 59600 | |
| }, | |
| { | |
| "epoch": 0.6834104125646778, | |
| "grad_norm": 0.45297375321388245, | |
| "learning_rate": 1.7079159750443593e-05, | |
| "loss": 2.2646, | |
| "step": 59700 | |
| }, | |
| { | |
| "epoch": 0.6845551536242502, | |
| "grad_norm": 0.5536375641822815, | |
| "learning_rate": 1.7107778604544676e-05, | |
| "loss": 2.2762, | |
| "step": 59800 | |
| }, | |
| { | |
| "epoch": 0.6856998946838225, | |
| "grad_norm": 0.42413586378097534, | |
| "learning_rate": 1.7136397458645758e-05, | |
| "loss": 2.2578, | |
| "step": 59900 | |
| }, | |
| { | |
| "epoch": 0.6868446357433948, | |
| "grad_norm": 0.45568087697029114, | |
| "learning_rate": 1.7165016312746837e-05, | |
| "loss": 2.2875, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 0.6879893768029671, | |
| "grad_norm": 0.45397791266441345, | |
| "learning_rate": 1.7193635166847923e-05, | |
| "loss": 2.28, | |
| "step": 60100 | |
| }, | |
| { | |
| "epoch": 0.6891341178625395, | |
| "grad_norm": 0.4058510661125183, | |
| "learning_rate": 1.7222254020949002e-05, | |
| "loss": 2.2845, | |
| "step": 60200 | |
| }, | |
| { | |
| "epoch": 0.6902788589221118, | |
| "grad_norm": 0.4380168318748474, | |
| "learning_rate": 1.7250872875050084e-05, | |
| "loss": 2.2574, | |
| "step": 60300 | |
| }, | |
| { | |
| "epoch": 0.6914235999816841, | |
| "grad_norm": 0.4758777320384979, | |
| "learning_rate": 1.7279491729151167e-05, | |
| "loss": 2.2597, | |
| "step": 60400 | |
| }, | |
| { | |
| "epoch": 0.6925683410412564, | |
| "grad_norm": 0.45504075288772583, | |
| "learning_rate": 1.730811058325225e-05, | |
| "loss": 2.2763, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 0.6937130821008288, | |
| "grad_norm": 0.4878067374229431, | |
| "learning_rate": 1.733672943735333e-05, | |
| "loss": 2.2811, | |
| "step": 60600 | |
| }, | |
| { | |
| "epoch": 0.6948578231604011, | |
| "grad_norm": 0.49453550577163696, | |
| "learning_rate": 1.736534829145441e-05, | |
| "loss": 2.2545, | |
| "step": 60700 | |
| }, | |
| { | |
| "epoch": 0.6960025642199734, | |
| "grad_norm": 0.43168744444847107, | |
| "learning_rate": 1.7393967145555493e-05, | |
| "loss": 2.2584, | |
| "step": 60800 | |
| }, | |
| { | |
| "epoch": 0.6971473052795457, | |
| "grad_norm": 0.5180889368057251, | |
| "learning_rate": 1.7422585999656575e-05, | |
| "loss": 2.274, | |
| "step": 60900 | |
| }, | |
| { | |
| "epoch": 0.6982920463391181, | |
| "grad_norm": 0.4250308573246002, | |
| "learning_rate": 1.7451204853757658e-05, | |
| "loss": 2.2531, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 0.6994367873986904, | |
| "grad_norm": 0.40109291672706604, | |
| "learning_rate": 1.7479823707858737e-05, | |
| "loss": 2.2771, | |
| "step": 61100 | |
| }, | |
| { | |
| "epoch": 0.7005815284582627, | |
| "grad_norm": 0.4581041932106018, | |
| "learning_rate": 1.750844256195982e-05, | |
| "loss": 2.2713, | |
| "step": 61200 | |
| }, | |
| { | |
| "epoch": 0.701726269517835, | |
| "grad_norm": 0.44385623931884766, | |
| "learning_rate": 1.75370614160609e-05, | |
| "loss": 2.2559, | |
| "step": 61300 | |
| }, | |
| { | |
| "epoch": 0.7028710105774074, | |
| "grad_norm": 0.44979363679885864, | |
| "learning_rate": 1.7565680270161984e-05, | |
| "loss": 2.2533, | |
| "step": 61400 | |
| }, | |
| { | |
| "epoch": 0.7040157516369797, | |
| "grad_norm": 0.44290637969970703, | |
| "learning_rate": 1.7594299124263066e-05, | |
| "loss": 2.2726, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 0.705160492696552, | |
| "grad_norm": 0.4479668438434601, | |
| "learning_rate": 1.7622917978364145e-05, | |
| "loss": 2.2673, | |
| "step": 61600 | |
| }, | |
| { | |
| "epoch": 0.7063052337561244, | |
| "grad_norm": 0.416456401348114, | |
| "learning_rate": 1.765153683246523e-05, | |
| "loss": 2.266, | |
| "step": 61700 | |
| }, | |
| { | |
| "epoch": 0.7074499748156967, | |
| "grad_norm": 0.45117634534835815, | |
| "learning_rate": 1.768015568656631e-05, | |
| "loss": 2.2791, | |
| "step": 61800 | |
| }, | |
| { | |
| "epoch": 0.708594715875269, | |
| "grad_norm": 0.5188822150230408, | |
| "learning_rate": 1.7708774540667393e-05, | |
| "loss": 2.266, | |
| "step": 61900 | |
| }, | |
| { | |
| "epoch": 0.7097394569348413, | |
| "grad_norm": 0.47650662064552307, | |
| "learning_rate": 1.7737393394768475e-05, | |
| "loss": 2.2695, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 0.7108841979944137, | |
| "grad_norm": 0.5149694681167603, | |
| "learning_rate": 1.7766012248869557e-05, | |
| "loss": 2.2775, | |
| "step": 62100 | |
| }, | |
| { | |
| "epoch": 0.712028939053986, | |
| "grad_norm": 0.4305098354816437, | |
| "learning_rate": 1.779463110297064e-05, | |
| "loss": 2.2722, | |
| "step": 62200 | |
| }, | |
| { | |
| "epoch": 0.7131736801135583, | |
| "grad_norm": 0.48085054755210876, | |
| "learning_rate": 1.782324995707172e-05, | |
| "loss": 2.2712, | |
| "step": 62300 | |
| }, | |
| { | |
| "epoch": 0.7143184211731306, | |
| "grad_norm": 0.4287306070327759, | |
| "learning_rate": 1.78518688111728e-05, | |
| "loss": 2.2629, | |
| "step": 62400 | |
| }, | |
| { | |
| "epoch": 0.715463162232703, | |
| "grad_norm": 0.45178937911987305, | |
| "learning_rate": 1.7880487665273884e-05, | |
| "loss": 2.2787, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 0.7166079032922753, | |
| "grad_norm": 0.5393545031547546, | |
| "learning_rate": 1.7909106519374966e-05, | |
| "loss": 2.2803, | |
| "step": 62600 | |
| }, | |
| { | |
| "epoch": 0.7177526443518476, | |
| "grad_norm": 0.4494490325450897, | |
| "learning_rate": 1.793772537347605e-05, | |
| "loss": 2.2714, | |
| "step": 62700 | |
| }, | |
| { | |
| "epoch": 0.7188973854114199, | |
| "grad_norm": 0.43690425157546997, | |
| "learning_rate": 1.7966344227577127e-05, | |
| "loss": 2.2639, | |
| "step": 62800 | |
| }, | |
| { | |
| "epoch": 0.7200421264709923, | |
| "grad_norm": 0.5241349339485168, | |
| "learning_rate": 1.7994963081678213e-05, | |
| "loss": 2.2639, | |
| "step": 62900 | |
| }, | |
| { | |
| "epoch": 0.7211868675305646, | |
| "grad_norm": 0.5191497206687927, | |
| "learning_rate": 1.8023581935779292e-05, | |
| "loss": 2.269, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 0.7223316085901369, | |
| "grad_norm": 0.4875340461730957, | |
| "learning_rate": 1.8052200789880375e-05, | |
| "loss": 2.262, | |
| "step": 63100 | |
| }, | |
| { | |
| "epoch": 0.7234763496497092, | |
| "grad_norm": 0.47728395462036133, | |
| "learning_rate": 1.8080533455440445e-05, | |
| "loss": 2.2653, | |
| "step": 63200 | |
| }, | |
| { | |
| "epoch": 0.7246210907092816, | |
| "grad_norm": 0.517727792263031, | |
| "learning_rate": 1.8109152309541528e-05, | |
| "loss": 2.2505, | |
| "step": 63300 | |
| }, | |
| { | |
| "epoch": 0.7257658317688539, | |
| "grad_norm": 0.5039493441581726, | |
| "learning_rate": 1.813777116364261e-05, | |
| "loss": 2.2858, | |
| "step": 63400 | |
| }, | |
| { | |
| "epoch": 0.7269105728284262, | |
| "grad_norm": 0.5291385054588318, | |
| "learning_rate": 1.8166390017743692e-05, | |
| "loss": 2.2559, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 0.7280553138879985, | |
| "grad_norm": 0.4564548432826996, | |
| "learning_rate": 1.819500887184477e-05, | |
| "loss": 2.2834, | |
| "step": 63600 | |
| }, | |
| { | |
| "epoch": 0.7292000549475709, | |
| "grad_norm": 0.4725103974342346, | |
| "learning_rate": 1.8223627725945854e-05, | |
| "loss": 2.2542, | |
| "step": 63700 | |
| }, | |
| { | |
| "epoch": 0.7303447960071432, | |
| "grad_norm": 0.475724995136261, | |
| "learning_rate": 1.8252246580046936e-05, | |
| "loss": 2.2672, | |
| "step": 63800 | |
| }, | |
| { | |
| "epoch": 0.7314895370667155, | |
| "grad_norm": 0.46552959084510803, | |
| "learning_rate": 1.828086543414802e-05, | |
| "loss": 2.276, | |
| "step": 63900 | |
| }, | |
| { | |
| "epoch": 0.7326342781262878, | |
| "grad_norm": 0.4661727845668793, | |
| "learning_rate": 1.83094842882491e-05, | |
| "loss": 2.2629, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 0.7337790191858602, | |
| "grad_norm": 0.5402230620384216, | |
| "learning_rate": 1.833810314235018e-05, | |
| "loss": 2.2578, | |
| "step": 64100 | |
| }, | |
| { | |
| "epoch": 0.7349237602454325, | |
| "grad_norm": 0.5088352560997009, | |
| "learning_rate": 1.8366721996451266e-05, | |
| "loss": 2.2584, | |
| "step": 64200 | |
| }, | |
| { | |
| "epoch": 0.7360685013050048, | |
| "grad_norm": 0.4990089535713196, | |
| "learning_rate": 1.8395340850552345e-05, | |
| "loss": 2.2565, | |
| "step": 64300 | |
| }, | |
| { | |
| "epoch": 0.7372132423645771, | |
| "grad_norm": 0.4477214813232422, | |
| "learning_rate": 1.8423959704653427e-05, | |
| "loss": 2.2702, | |
| "step": 64400 | |
| }, | |
| { | |
| "epoch": 0.7383579834241495, | |
| "grad_norm": 0.43167996406555176, | |
| "learning_rate": 1.845257855875451e-05, | |
| "loss": 2.2744, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 0.7395027244837218, | |
| "grad_norm": 0.4676847457885742, | |
| "learning_rate": 1.8481197412855592e-05, | |
| "loss": 2.2624, | |
| "step": 64600 | |
| }, | |
| { | |
| "epoch": 0.7406474655432941, | |
| "grad_norm": 0.5216718912124634, | |
| "learning_rate": 1.8509816266956674e-05, | |
| "loss": 2.2801, | |
| "step": 64700 | |
| }, | |
| { | |
| "epoch": 0.7417922066028664, | |
| "grad_norm": 0.4484277665615082, | |
| "learning_rate": 1.8538435121057753e-05, | |
| "loss": 2.2567, | |
| "step": 64800 | |
| }, | |
| { | |
| "epoch": 0.7429369476624388, | |
| "grad_norm": 0.47648686170578003, | |
| "learning_rate": 1.8567053975158836e-05, | |
| "loss": 2.2605, | |
| "step": 64900 | |
| }, | |
| { | |
| "epoch": 0.7440816887220111, | |
| "grad_norm": 0.46372881531715393, | |
| "learning_rate": 1.8595672829259918e-05, | |
| "loss": 2.2673, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 0.7452264297815834, | |
| "grad_norm": 0.43808409571647644, | |
| "learning_rate": 1.8624291683361e-05, | |
| "loss": 2.263, | |
| "step": 65100 | |
| }, | |
| { | |
| "epoch": 0.7463711708411558, | |
| "grad_norm": 0.4345923364162445, | |
| "learning_rate": 1.865291053746208e-05, | |
| "loss": 2.2638, | |
| "step": 65200 | |
| }, | |
| { | |
| "epoch": 0.7475159119007281, | |
| "grad_norm": 0.49368196725845337, | |
| "learning_rate": 1.8681529391563162e-05, | |
| "loss": 2.271, | |
| "step": 65300 | |
| }, | |
| { | |
| "epoch": 0.7486606529603004, | |
| "grad_norm": 0.46279481053352356, | |
| "learning_rate": 1.8710148245664244e-05, | |
| "loss": 2.2721, | |
| "step": 65400 | |
| }, | |
| { | |
| "epoch": 0.7498053940198727, | |
| "grad_norm": 0.423225998878479, | |
| "learning_rate": 1.8738767099765327e-05, | |
| "loss": 2.2634, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 0.7509501350794451, | |
| "grad_norm": 0.48171648383140564, | |
| "learning_rate": 1.876738595386641e-05, | |
| "loss": 2.2656, | |
| "step": 65600 | |
| }, | |
| { | |
| "epoch": 0.7520948761390174, | |
| "grad_norm": 0.41427454352378845, | |
| "learning_rate": 1.879571861942648e-05, | |
| "loss": 2.2663, | |
| "step": 65700 | |
| }, | |
| { | |
| "epoch": 0.7532396171985897, | |
| "grad_norm": 0.5507281422615051, | |
| "learning_rate": 1.8824337473527562e-05, | |
| "loss": 2.2572, | |
| "step": 65800 | |
| }, | |
| { | |
| "epoch": 0.754384358258162, | |
| "grad_norm": 0.44962623715400696, | |
| "learning_rate": 1.8852956327628644e-05, | |
| "loss": 2.2745, | |
| "step": 65900 | |
| }, | |
| { | |
| "epoch": 0.7555290993177344, | |
| "grad_norm": 0.5174722671508789, | |
| "learning_rate": 1.8881575181729727e-05, | |
| "loss": 2.2387, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 0.7566738403773067, | |
| "grad_norm": 0.5622259974479675, | |
| "learning_rate": 1.8910194035830806e-05, | |
| "loss": 2.2602, | |
| "step": 66100 | |
| }, | |
| { | |
| "epoch": 0.757818581436879, | |
| "grad_norm": 0.5210707187652588, | |
| "learning_rate": 1.8938812889931888e-05, | |
| "loss": 2.2643, | |
| "step": 66200 | |
| }, | |
| { | |
| "epoch": 0.7589633224964513, | |
| "grad_norm": 0.486509770154953, | |
| "learning_rate": 1.896743174403297e-05, | |
| "loss": 2.2707, | |
| "step": 66300 | |
| }, | |
| { | |
| "epoch": 0.7601080635560237, | |
| "grad_norm": 0.47192618250846863, | |
| "learning_rate": 1.8996050598134053e-05, | |
| "loss": 2.2613, | |
| "step": 66400 | |
| }, | |
| { | |
| "epoch": 0.761252804615596, | |
| "grad_norm": 0.5344927906990051, | |
| "learning_rate": 1.9024669452235135e-05, | |
| "loss": 2.2561, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 0.7623975456751683, | |
| "grad_norm": 0.5391865372657776, | |
| "learning_rate": 1.9053288306336214e-05, | |
| "loss": 2.2763, | |
| "step": 66600 | |
| }, | |
| { | |
| "epoch": 0.7635422867347406, | |
| "grad_norm": 0.45489776134490967, | |
| "learning_rate": 1.90819071604373e-05, | |
| "loss": 2.2611, | |
| "step": 66700 | |
| }, | |
| { | |
| "epoch": 0.764687027794313, | |
| "grad_norm": 0.38119086623191833, | |
| "learning_rate": 1.911052601453838e-05, | |
| "loss": 2.2602, | |
| "step": 66800 | |
| }, | |
| { | |
| "epoch": 0.7658317688538853, | |
| "grad_norm": 0.49369150400161743, | |
| "learning_rate": 1.913914486863946e-05, | |
| "loss": 2.2613, | |
| "step": 66900 | |
| }, | |
| { | |
| "epoch": 0.7669765099134576, | |
| "grad_norm": 0.6532511711120605, | |
| "learning_rate": 1.916776372274054e-05, | |
| "loss": 2.268, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 0.7681212509730299, | |
| "grad_norm": 0.4440617263317108, | |
| "learning_rate": 1.9196382576841626e-05, | |
| "loss": 2.2776, | |
| "step": 67100 | |
| }, | |
| { | |
| "epoch": 0.7692659920326023, | |
| "grad_norm": 0.46082597970962524, | |
| "learning_rate": 1.9225001430942705e-05, | |
| "loss": 2.255, | |
| "step": 67200 | |
| }, | |
| { | |
| "epoch": 0.7704107330921746, | |
| "grad_norm": 0.40288957953453064, | |
| "learning_rate": 1.9253620285043788e-05, | |
| "loss": 2.2652, | |
| "step": 67300 | |
| }, | |
| { | |
| "epoch": 0.7715554741517469, | |
| "grad_norm": 0.51495760679245, | |
| "learning_rate": 1.928223913914487e-05, | |
| "loss": 2.2651, | |
| "step": 67400 | |
| }, | |
| { | |
| "epoch": 0.7727002152113192, | |
| "grad_norm": 0.5247004628181458, | |
| "learning_rate": 1.9310857993245953e-05, | |
| "loss": 2.2534, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 0.7738449562708916, | |
| "grad_norm": 0.4820224344730377, | |
| "learning_rate": 1.9339476847347035e-05, | |
| "loss": 2.2811, | |
| "step": 67600 | |
| }, | |
| { | |
| "epoch": 0.7749896973304639, | |
| "grad_norm": 0.43811190128326416, | |
| "learning_rate": 1.9368095701448114e-05, | |
| "loss": 2.2638, | |
| "step": 67700 | |
| }, | |
| { | |
| "epoch": 0.7761344383900362, | |
| "grad_norm": 0.4226974546909332, | |
| "learning_rate": 1.9396714555549196e-05, | |
| "loss": 2.2605, | |
| "step": 67800 | |
| }, | |
| { | |
| "epoch": 0.7772791794496084, | |
| "grad_norm": 0.4748658835887909, | |
| "learning_rate": 1.9425047221109267e-05, | |
| "loss": 2.2617, | |
| "step": 67900 | |
| }, | |
| { | |
| "epoch": 0.7784239205091809, | |
| "grad_norm": 0.4761633574962616, | |
| "learning_rate": 1.9453666075210353e-05, | |
| "loss": 2.2614, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 0.7795686615687532, | |
| "grad_norm": 0.48032355308532715, | |
| "learning_rate": 1.948228492931143e-05, | |
| "loss": 2.2443, | |
| "step": 68100 | |
| }, | |
| { | |
| "epoch": 0.7807134026283254, | |
| "grad_norm": 0.4990929961204529, | |
| "learning_rate": 1.9510903783412514e-05, | |
| "loss": 2.2702, | |
| "step": 68200 | |
| }, | |
| { | |
| "epoch": 0.7818581436878977, | |
| "grad_norm": 0.40114450454711914, | |
| "learning_rate": 1.9539522637513596e-05, | |
| "loss": 2.2482, | |
| "step": 68300 | |
| }, | |
| { | |
| "epoch": 0.7830028847474702, | |
| "grad_norm": 0.4980379343032837, | |
| "learning_rate": 1.9567855303073667e-05, | |
| "loss": 2.2706, | |
| "step": 68400 | |
| }, | |
| { | |
| "epoch": 0.7841476258070424, | |
| "grad_norm": 0.42115017771720886, | |
| "learning_rate": 1.9596474157174746e-05, | |
| "loss": 2.2741, | |
| "step": 68500 | |
| }, | |
| { | |
| "epoch": 0.7852923668666147, | |
| "grad_norm": 0.4727267622947693, | |
| "learning_rate": 1.9625093011275832e-05, | |
| "loss": 2.2649, | |
| "step": 68600 | |
| }, | |
| { | |
| "epoch": 0.7864371079261872, | |
| "grad_norm": 0.47092244029045105, | |
| "learning_rate": 1.965371186537691e-05, | |
| "loss": 2.2482, | |
| "step": 68700 | |
| }, | |
| { | |
| "epoch": 0.7875818489857594, | |
| "grad_norm": 0.4399222433567047, | |
| "learning_rate": 1.9682330719477993e-05, | |
| "loss": 2.2601, | |
| "step": 68800 | |
| }, | |
| { | |
| "epoch": 0.7887265900453317, | |
| "grad_norm": 0.44569170475006104, | |
| "learning_rate": 1.9710949573579076e-05, | |
| "loss": 2.2679, | |
| "step": 68900 | |
| }, | |
| { | |
| "epoch": 0.789871331104904, | |
| "grad_norm": 0.43348217010498047, | |
| "learning_rate": 1.9739568427680158e-05, | |
| "loss": 2.2529, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 0.7910160721644764, | |
| "grad_norm": 0.4533138573169708, | |
| "learning_rate": 1.976818728178124e-05, | |
| "loss": 2.2634, | |
| "step": 69100 | |
| }, | |
| { | |
| "epoch": 0.7921608132240487, | |
| "grad_norm": 0.5940411686897278, | |
| "learning_rate": 1.979680613588232e-05, | |
| "loss": 2.2526, | |
| "step": 69200 | |
| }, | |
| { | |
| "epoch": 0.793305554283621, | |
| "grad_norm": 0.4553944766521454, | |
| "learning_rate": 1.9825424989983402e-05, | |
| "loss": 2.2514, | |
| "step": 69300 | |
| }, | |
| { | |
| "epoch": 0.7944502953431933, | |
| "grad_norm": 0.5016659498214722, | |
| "learning_rate": 1.9854043844084484e-05, | |
| "loss": 2.2452, | |
| "step": 69400 | |
| }, | |
| { | |
| "epoch": 0.7955950364027657, | |
| "grad_norm": 0.45411109924316406, | |
| "learning_rate": 1.9882662698185567e-05, | |
| "loss": 2.2576, | |
| "step": 69500 | |
| }, | |
| { | |
| "epoch": 0.796739777462338, | |
| "grad_norm": 0.3970607817173004, | |
| "learning_rate": 1.991128155228665e-05, | |
| "loss": 2.2403, | |
| "step": 69600 | |
| }, | |
| { | |
| "epoch": 0.7978845185219103, | |
| "grad_norm": 0.47627323865890503, | |
| "learning_rate": 1.9939900406387728e-05, | |
| "loss": 2.2614, | |
| "step": 69700 | |
| }, | |
| { | |
| "epoch": 0.7990292595814826, | |
| "grad_norm": 0.4684958755970001, | |
| "learning_rate": 1.9968519260488814e-05, | |
| "loss": 2.2614, | |
| "step": 69800 | |
| }, | |
| { | |
| "epoch": 0.800174000641055, | |
| "grad_norm": 0.418066143989563, | |
| "learning_rate": 1.9997138114589893e-05, | |
| "loss": 2.2627, | |
| "step": 69900 | |
| }, | |
| { | |
| "epoch": 0.8013187417006273, | |
| "grad_norm": 0.43731051683425903, | |
| "learning_rate": 1.9998690637047088e-05, | |
| "loss": 2.2473, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 0.8024634827601996, | |
| "grad_norm": 0.43731534481048584, | |
| "learning_rate": 1.999416488433588e-05, | |
| "loss": 2.2512, | |
| "step": 70100 | |
| }, | |
| { | |
| "epoch": 0.8036082238197719, | |
| "grad_norm": 0.46741268038749695, | |
| "learning_rate": 1.99864080397093e-05, | |
| "loss": 2.2373, | |
| "step": 70200 | |
| }, | |
| { | |
| "epoch": 0.8047529648793443, | |
| "grad_norm": 0.4025750160217285, | |
| "learning_rate": 1.9975422610938463e-05, | |
| "loss": 2.2711, | |
| "step": 70300 | |
| }, | |
| { | |
| "epoch": 0.8058977059389166, | |
| "grad_norm": 0.5251903533935547, | |
| "learning_rate": 1.996121214958875e-05, | |
| "loss": 2.2627, | |
| "step": 70400 | |
| }, | |
| { | |
| "epoch": 0.8070424469984889, | |
| "grad_norm": 0.5502318143844604, | |
| "learning_rate": 1.9943781249871618e-05, | |
| "loss": 2.2615, | |
| "step": 70500 | |
| }, | |
| { | |
| "epoch": 0.8081871880580612, | |
| "grad_norm": 0.4485688805580139, | |
| "learning_rate": 1.992313554715929e-05, | |
| "loss": 2.2486, | |
| "step": 70600 | |
| }, | |
| { | |
| "epoch": 0.8093319291176336, | |
| "grad_norm": 0.42768144607543945, | |
| "learning_rate": 1.9899281716162846e-05, | |
| "loss": 2.2602, | |
| "step": 70700 | |
| }, | |
| { | |
| "epoch": 0.8104766701772059, | |
| "grad_norm": 0.446748286485672, | |
| "learning_rate": 1.987222746877431e-05, | |
| "loss": 2.2705, | |
| "step": 70800 | |
| }, | |
| { | |
| "epoch": 0.8116214112367782, | |
| "grad_norm": 0.4175347685813904, | |
| "learning_rate": 1.9841981551573424e-05, | |
| "loss": 2.2437, | |
| "step": 70900 | |
| }, | |
| { | |
| "epoch": 0.8127661522963505, | |
| "grad_norm": 0.4600895047187805, | |
| "learning_rate": 1.9808553742999863e-05, | |
| "loss": 2.2592, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 0.8139108933559229, | |
| "grad_norm": 0.4427413046360016, | |
| "learning_rate": 1.9771954850191927e-05, | |
| "loss": 2.2508, | |
| "step": 71100 | |
| }, | |
| { | |
| "epoch": 0.8150556344154952, | |
| "grad_norm": 0.41499242186546326, | |
| "learning_rate": 1.973219670549259e-05, | |
| "loss": 2.2677, | |
| "step": 71200 | |
| }, | |
| { | |
| "epoch": 0.8162003754750675, | |
| "grad_norm": 0.48436206579208374, | |
| "learning_rate": 1.9689292162624135e-05, | |
| "loss": 2.2569, | |
| "step": 71300 | |
| }, | |
| { | |
| "epoch": 0.8173451165346398, | |
| "grad_norm": 0.6114481687545776, | |
| "learning_rate": 1.9643255092532582e-05, | |
| "loss": 2.262, | |
| "step": 71400 | |
| }, | |
| { | |
| "epoch": 0.8184898575942122, | |
| "grad_norm": 0.489258348941803, | |
| "learning_rate": 1.959410037890323e-05, | |
| "loss": 2.2376, | |
| "step": 71500 | |
| }, | |
| { | |
| "epoch": 0.8196345986537845, | |
| "grad_norm": 0.5034488439559937, | |
| "learning_rate": 1.9541843913348804e-05, | |
| "loss": 2.2542, | |
| "step": 71600 | |
| }, | |
| { | |
| "epoch": 0.8207793397133568, | |
| "grad_norm": 0.4198042154312134, | |
| "learning_rate": 1.948650259027172e-05, | |
| "loss": 2.2509, | |
| "step": 71700 | |
| }, | |
| { | |
| "epoch": 0.8219240807729291, | |
| "grad_norm": 0.462247759103775, | |
| "learning_rate": 1.9428094301402164e-05, | |
| "loss": 2.2493, | |
| "step": 71800 | |
| }, | |
| { | |
| "epoch": 0.8230688218325015, | |
| "grad_norm": 0.4464458227157593, | |
| "learning_rate": 1.936663793001374e-05, | |
| "loss": 2.2645, | |
| "step": 71900 | |
| }, | |
| { | |
| "epoch": 0.8242135628920738, | |
| "grad_norm": 0.5643649697303772, | |
| "learning_rate": 1.930215334481855e-05, | |
| "loss": 2.2455, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 0.8253583039516461, | |
| "grad_norm": 0.48486921191215515, | |
| "learning_rate": 1.9234661393543668e-05, | |
| "loss": 2.256, | |
| "step": 72100 | |
| }, | |
| { | |
| "epoch": 0.8265030450112185, | |
| "grad_norm": 0.4897540807723999, | |
| "learning_rate": 1.9164183896191133e-05, | |
| "loss": 2.2453, | |
| "step": 72200 | |
| }, | |
| { | |
| "epoch": 0.8276477860707908, | |
| "grad_norm": 0.47083780169487, | |
| "learning_rate": 1.9090743637983577e-05, | |
| "loss": 2.239, | |
| "step": 72300 | |
| }, | |
| { | |
| "epoch": 0.8287925271303631, | |
| "grad_norm": 0.42039257287979126, | |
| "learning_rate": 1.9014364361997813e-05, | |
| "loss": 2.2483, | |
| "step": 72400 | |
| }, | |
| { | |
| "epoch": 0.8299372681899354, | |
| "grad_norm": 0.5170241594314575, | |
| "learning_rate": 1.8935070761488754e-05, | |
| "loss": 2.2686, | |
| "step": 72500 | |
| }, | |
| { | |
| "epoch": 0.8310820092495078, | |
| "grad_norm": 0.46110501885414124, | |
| "learning_rate": 1.885288847190614e-05, | |
| "loss": 2.2444, | |
| "step": 72600 | |
| }, | |
| { | |
| "epoch": 0.8322267503090801, | |
| "grad_norm": 0.48973456025123596, | |
| "learning_rate": 1.876784406260664e-05, | |
| "loss": 2.2475, | |
| "step": 72700 | |
| }, | |
| { | |
| "epoch": 0.8333714913686524, | |
| "grad_norm": 0.45546847581863403, | |
| "learning_rate": 1.8679965028264055e-05, | |
| "loss": 2.2519, | |
| "step": 72800 | |
| }, | |
| { | |
| "epoch": 0.8345162324282247, | |
| "grad_norm": 0.49703338742256165, | |
| "learning_rate": 1.859020042770291e-05, | |
| "loss": 2.2518, | |
| "step": 72900 | |
| }, | |
| { | |
| "epoch": 0.8356609734877971, | |
| "grad_norm": 0.494150310754776, | |
| "learning_rate": 1.8496765904697226e-05, | |
| "loss": 2.2682, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 0.8368057145473694, | |
| "grad_norm": 0.46213245391845703, | |
| "learning_rate": 1.840058439563126e-05, | |
| "loss": 2.2499, | |
| "step": 73100 | |
| }, | |
| { | |
| "epoch": 0.8379504556069417, | |
| "grad_norm": 0.40609198808670044, | |
| "learning_rate": 1.830168699577909e-05, | |
| "loss": 2.2706, | |
| "step": 73200 | |
| }, | |
| { | |
| "epoch": 0.839095196666514, | |
| "grad_norm": 0.6065017580986023, | |
| "learning_rate": 1.820010567845644e-05, | |
| "loss": 2.2553, | |
| "step": 73300 | |
| }, | |
| { | |
| "epoch": 0.8402399377260864, | |
| "grad_norm": 0.44308215379714966, | |
| "learning_rate": 1.809587328468373e-05, | |
| "loss": 2.2503, | |
| "step": 73400 | |
| }, | |
| { | |
| "epoch": 0.8413846787856587, | |
| "grad_norm": 0.41164156794548035, | |
| "learning_rate": 1.7989023512568686e-05, | |
| "loss": 2.2491, | |
| "step": 73500 | |
| }, | |
| { | |
| "epoch": 0.842529419845231, | |
| "grad_norm": 0.4938010573387146, | |
| "learning_rate": 1.7879590906411786e-05, | |
| "loss": 2.2581, | |
| "step": 73600 | |
| }, | |
| { | |
| "epoch": 0.8436741609048033, | |
| "grad_norm": 0.4634738862514496, | |
| "learning_rate": 1.7767610845538178e-05, | |
| "loss": 2.2665, | |
| "step": 73700 | |
| }, | |
| { | |
| "epoch": 0.8448189019643757, | |
| "grad_norm": 0.43750569224357605, | |
| "learning_rate": 1.7653119532859626e-05, | |
| "loss": 2.2497, | |
| "step": 73800 | |
| }, | |
| { | |
| "epoch": 0.845963643023948, | |
| "grad_norm": 0.5143136978149414, | |
| "learning_rate": 1.7536153983170157e-05, | |
| "loss": 2.2318, | |
| "step": 73900 | |
| }, | |
| { | |
| "epoch": 0.8471083840835203, | |
| "grad_norm": 0.5141619443893433, | |
| "learning_rate": 1.7416752011179294e-05, | |
| "loss": 2.2418, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 0.8482531251430926, | |
| "grad_norm": 0.46615302562713623, | |
| "learning_rate": 1.7296181957754126e-05, | |
| "loss": 2.2496, | |
| "step": 74100 | |
| }, | |
| { | |
| "epoch": 0.849397866202665, | |
| "grad_norm": 0.4517097473144531, | |
| "learning_rate": 1.71720471105587e-05, | |
| "loss": 2.2501, | |
| "step": 74200 | |
| }, | |
| { | |
| "epoch": 0.8505426072622373, | |
| "grad_norm": 0.4446674883365631, | |
| "learning_rate": 1.7045593556027164e-05, | |
| "loss": 2.2487, | |
| "step": 74300 | |
| }, | |
| { | |
| "epoch": 0.8516873483218096, | |
| "grad_norm": 0.5222780108451843, | |
| "learning_rate": 1.691686217632051e-05, | |
| "loss": 2.2495, | |
| "step": 74400 | |
| }, | |
| { | |
| "epoch": 0.8528320893813819, | |
| "grad_norm": 0.4754522442817688, | |
| "learning_rate": 1.678589459001567e-05, | |
| "loss": 2.251, | |
| "step": 74500 | |
| }, | |
| { | |
| "epoch": 0.8539768304409543, | |
| "grad_norm": 0.46791428327560425, | |
| "learning_rate": 1.6652733138650367e-05, | |
| "loss": 2.2502, | |
| "step": 74600 | |
| }, | |
| { | |
| "epoch": 0.8551215715005266, | |
| "grad_norm": 0.5582058429718018, | |
| "learning_rate": 1.651742087303412e-05, | |
| "loss": 2.249, | |
| "step": 74700 | |
| }, | |
| { | |
| "epoch": 0.8562663125600989, | |
| "grad_norm": 0.5812872648239136, | |
| "learning_rate": 1.6380001539330088e-05, | |
| "loss": 2.2402, | |
| "step": 74800 | |
| }, | |
| { | |
| "epoch": 0.8574110536196712, | |
| "grad_norm": 0.4505755305290222, | |
| "learning_rate": 1.624051956491196e-05, | |
| "loss": 2.2445, | |
| "step": 74900 | |
| }, | |
| { | |
| "epoch": 0.8585557946792436, | |
| "grad_norm": 0.46958789229393005, | |
| "learning_rate": 1.609902004400073e-05, | |
| "loss": 2.2729, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 0.8597005357388159, | |
| "grad_norm": 0.44104379415512085, | |
| "learning_rate": 1.5955548723085804e-05, | |
| "loss": 2.2429, | |
| "step": 75100 | |
| }, | |
| { | |
| "epoch": 0.8608452767983882, | |
| "grad_norm": 0.4598877429962158, | |
| "learning_rate": 1.581015198613528e-05, | |
| "loss": 2.2471, | |
| "step": 75200 | |
| }, | |
| { | |
| "epoch": 0.8619900178579605, | |
| "grad_norm": 0.43592801690101624, | |
| "learning_rate": 1.5662876839600084e-05, | |
| "loss": 2.2537, | |
| "step": 75300 | |
| }, | |
| { | |
| "epoch": 0.8631347589175329, | |
| "grad_norm": 0.46111100912094116, | |
| "learning_rate": 1.551377089721692e-05, | |
| "loss": 2.2262, | |
| "step": 75400 | |
| }, | |
| { | |
| "epoch": 0.8642794999771052, | |
| "grad_norm": 0.4764077961444855, | |
| "learning_rate": 1.5362882364614825e-05, | |
| "loss": 2.2556, | |
| "step": 75500 | |
| }, | |
| { | |
| "epoch": 0.8654242410366775, | |
| "grad_norm": 0.5170374512672424, | |
| "learning_rate": 1.5210260023730402e-05, | |
| "loss": 2.2654, | |
| "step": 75600 | |
| }, | |
| { | |
| "epoch": 0.8665689820962498, | |
| "grad_norm": 0.4534991383552551, | |
| "learning_rate": 1.5055953217036735e-05, | |
| "loss": 2.2647, | |
| "step": 75700 | |
| }, | |
| { | |
| "epoch": 0.8677137231558222, | |
| "grad_norm": 0.42923828959465027, | |
| "learning_rate": 1.490001183159105e-05, | |
| "loss": 2.2366, | |
| "step": 75800 | |
| }, | |
| { | |
| "epoch": 0.8688584642153945, | |
| "grad_norm": 0.5356833934783936, | |
| "learning_rate": 1.474248628290637e-05, | |
| "loss": 2.2459, | |
| "step": 75900 | |
| }, | |
| { | |
| "epoch": 0.8700032052749668, | |
| "grad_norm": 0.5153579115867615, | |
| "learning_rate": 1.4583427498652252e-05, | |
| "loss": 2.2416, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 0.8711479463345392, | |
| "grad_norm": 0.48512566089630127, | |
| "learning_rate": 1.4422886902190014e-05, | |
| "loss": 2.2519, | |
| "step": 76100 | |
| }, | |
| { | |
| "epoch": 0.8722926873941115, | |
| "grad_norm": 0.5230854153633118, | |
| "learning_rate": 1.4260916395947657e-05, | |
| "loss": 2.2429, | |
| "step": 76200 | |
| }, | |
| { | |
| "epoch": 0.8734374284536838, | |
| "grad_norm": 0.4530618190765381, | |
| "learning_rate": 1.4097568344639916e-05, | |
| "loss": 2.2675, | |
| "step": 76300 | |
| }, | |
| { | |
| "epoch": 0.8745821695132561, | |
| "grad_norm": 0.4915354251861572, | |
| "learning_rate": 1.3932895558338879e-05, | |
| "loss": 2.2443, | |
| "step": 76400 | |
| }, | |
| { | |
| "epoch": 0.8757269105728285, | |
| "grad_norm": 0.437199205160141, | |
| "learning_rate": 1.3766951275400596e-05, | |
| "loss": 2.2549, | |
| "step": 76500 | |
| }, | |
| { | |
| "epoch": 0.8768716516324008, | |
| "grad_norm": 0.48987364768981934, | |
| "learning_rate": 1.3599789145253226e-05, | |
| "loss": 2.2294, | |
| "step": 76600 | |
| }, | |
| { | |
| "epoch": 0.8780163926919731, | |
| "grad_norm": 0.40682971477508545, | |
| "learning_rate": 1.34314632110523e-05, | |
| "loss": 2.2404, | |
| "step": 76700 | |
| }, | |
| { | |
| "epoch": 0.8791611337515454, | |
| "grad_norm": 0.4797169268131256, | |
| "learning_rate": 1.3262027892208696e-05, | |
| "loss": 2.2613, | |
| "step": 76800 | |
| }, | |
| { | |
| "epoch": 0.8803058748111178, | |
| "grad_norm": 0.5203677415847778, | |
| "learning_rate": 1.3091537966794933e-05, | |
| "loss": 2.2509, | |
| "step": 76900 | |
| }, | |
| { | |
| "epoch": 0.8814506158706901, | |
| "grad_norm": 0.519087553024292, | |
| "learning_rate": 1.2920048553835574e-05, | |
| "loss": 2.2625, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 0.8825953569302624, | |
| "grad_norm": 0.4839503765106201, | |
| "learning_rate": 1.2747615095487331e-05, | |
| "loss": 2.2487, | |
| "step": 77100 | |
| }, | |
| { | |
| "epoch": 0.8837400979898347, | |
| "grad_norm": 0.4150638282299042, | |
| "learning_rate": 1.2574293339114757e-05, | |
| "loss": 2.254, | |
| "step": 77200 | |
| }, | |
| { | |
| "epoch": 0.8848848390494071, | |
| "grad_norm": 0.5131925940513611, | |
| "learning_rate": 1.240013931926724e-05, | |
| "loss": 2.238, | |
| "step": 77300 | |
| }, | |
| { | |
| "epoch": 0.8860295801089794, | |
| "grad_norm": 0.5299201607704163, | |
| "learning_rate": 1.2225209339563144e-05, | |
| "loss": 2.2517, | |
| "step": 77400 | |
| }, | |
| { | |
| "epoch": 0.8871743211685517, | |
| "grad_norm": 0.4446033239364624, | |
| "learning_rate": 1.204955995448699e-05, | |
| "loss": 2.2521, | |
| "step": 77500 | |
| }, | |
| { | |
| "epoch": 0.888319062228124, | |
| "grad_norm": 0.4393691122531891, | |
| "learning_rate": 1.1873247951105489e-05, | |
| "loss": 2.2634, | |
| "step": 77600 | |
| }, | |
| { | |
| "epoch": 0.8894638032876964, | |
| "grad_norm": 0.48134180903434753, | |
| "learning_rate": 1.1696330330708421e-05, | |
| "loss": 2.2478, | |
| "step": 77700 | |
| }, | |
| { | |
| "epoch": 0.8906085443472687, | |
| "grad_norm": 0.4115225076675415, | |
| "learning_rate": 1.1518864290380249e-05, | |
| "loss": 2.2197, | |
| "step": 77800 | |
| }, | |
| { | |
| "epoch": 0.891753285406841, | |
| "grad_norm": 0.47142502665519714, | |
| "learning_rate": 1.1340907204508403e-05, | |
| "loss": 2.2344, | |
| "step": 77900 | |
| }, | |
| { | |
| "epoch": 0.8928980264664133, | |
| "grad_norm": 0.5209864377975464, | |
| "learning_rate": 1.1162516606234276e-05, | |
| "loss": 2.2486, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 0.8940427675259857, | |
| "grad_norm": 0.4119342267513275, | |
| "learning_rate": 1.09837501688529e-05, | |
| "loss": 2.2425, | |
| "step": 78100 | |
| }, | |
| { | |
| "epoch": 0.895187508585558, | |
| "grad_norm": 0.4754474461078644, | |
| "learning_rate": 1.0806457916272542e-05, | |
| "loss": 2.246, | |
| "step": 78200 | |
| }, | |
| { | |
| "epoch": 0.8963322496451303, | |
| "grad_norm": 0.511458694934845, | |
| "learning_rate": 1.0627115602490508e-05, | |
| "loss": 2.2453, | |
| "step": 78300 | |
| }, | |
| { | |
| "epoch": 0.8974769907047025, | |
| "grad_norm": 0.5435287356376648, | |
| "learning_rate": 1.0447570543589034e-05, | |
| "loss": 2.2452, | |
| "step": 78400 | |
| }, | |
| { | |
| "epoch": 0.898621731764275, | |
| "grad_norm": 0.4102800190448761, | |
| "learning_rate": 1.0267880786097762e-05, | |
| "loss": 2.2459, | |
| "step": 78500 | |
| }, | |
| { | |
| "epoch": 0.8997664728238473, | |
| "grad_norm": 0.3983183801174164, | |
| "learning_rate": 1.0088104423327082e-05, | |
| "loss": 2.2274, | |
| "step": 78600 | |
| }, | |
| { | |
| "epoch": 0.9009112138834195, | |
| "grad_norm": 0.44764354825019836, | |
| "learning_rate": 9.908299576586684e-06, | |
| "loss": 2.2512, | |
| "step": 78700 | |
| }, | |
| { | |
| "epoch": 0.9020559549429918, | |
| "grad_norm": 0.5015277862548828, | |
| "learning_rate": 9.728524376395068e-06, | |
| "loss": 2.244, | |
| "step": 78800 | |
| }, | |
| { | |
| "epoch": 0.9032006960025643, | |
| "grad_norm": 0.4522726237773895, | |
| "learning_rate": 9.548836943686055e-06, | |
| "loss": 2.2443, | |
| "step": 78900 | |
| }, | |
| { | |
| "epoch": 0.9043454370621365, | |
| "grad_norm": 0.4915413558483124, | |
| "learning_rate": 9.369295371018442e-06, | |
| "loss": 2.246, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 0.9054901781217088, | |
| "grad_norm": 0.45682036876678467, | |
| "learning_rate": 9.18995770379478e-06, | |
| "loss": 2.2353, | |
| "step": 79100 | |
| }, | |
| { | |
| "epoch": 0.9066349191812811, | |
| "grad_norm": 0.41946882009506226, | |
| "learning_rate": 9.010881921495438e-06, | |
| "loss": 2.2468, | |
| "step": 79200 | |
| }, | |
| { | |
| "epoch": 0.9077796602408535, | |
| "grad_norm": 0.4232065975666046, | |
| "learning_rate": 8.832125918933955e-06, | |
| "loss": 2.2575, | |
| "step": 79300 | |
| }, | |
| { | |
| "epoch": 0.9089244013004258, | |
| "grad_norm": 0.4463779926300049, | |
| "learning_rate": 8.653747487539764e-06, | |
| "loss": 2.2379, | |
| "step": 79400 | |
| }, | |
| { | |
| "epoch": 0.9100691423599981, | |
| "grad_norm": 0.43841028213500977, | |
| "learning_rate": 8.47580429667436e-06, | |
| "loss": 2.2414, | |
| "step": 79500 | |
| }, | |
| { | |
| "epoch": 0.9112138834195705, | |
| "grad_norm": 0.43061941862106323, | |
| "learning_rate": 8.29835387498692e-06, | |
| "loss": 2.2475, | |
| "step": 79600 | |
| }, | |
| { | |
| "epoch": 0.9123586244791428, | |
| "grad_norm": 0.41848114132881165, | |
| "learning_rate": 8.121453591815401e-06, | |
| "loss": 2.2535, | |
| "step": 79700 | |
| }, | |
| { | |
| "epoch": 0.9135033655387151, | |
| "grad_norm": 0.4669823944568634, | |
| "learning_rate": 7.94516063863917e-06, | |
| "loss": 2.2551, | |
| "step": 79800 | |
| }, | |
| { | |
| "epoch": 0.9146481065982874, | |
| "grad_norm": 0.4634804129600525, | |
| "learning_rate": 7.769532010589123e-06, | |
| "loss": 2.2393, | |
| "step": 79900 | |
| }, | |
| { | |
| "epoch": 0.9157928476578598, | |
| "grad_norm": 0.43346285820007324, | |
| "learning_rate": 7.594624488021274e-06, | |
| "loss": 2.2607, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 0.9169375887174321, | |
| "grad_norm": 0.43006187677383423, | |
| "learning_rate": 7.420494618159791e-06, | |
| "loss": 2.2408, | |
| "step": 80100 | |
| }, | |
| { | |
| "epoch": 0.9180823297770044, | |
| "grad_norm": 0.577170729637146, | |
| "learning_rate": 7.2471986968154075e-06, | |
| "loss": 2.2421, | |
| "step": 80200 | |
| }, | |
| { | |
| "epoch": 0.9192270708365767, | |
| "grad_norm": 0.43540096282958984, | |
| "learning_rate": 7.074792750185093e-06, | |
| "loss": 2.2577, | |
| "step": 80300 | |
| }, | |
| { | |
| "epoch": 0.9203718118961491, | |
| "grad_norm": 0.41239428520202637, | |
| "learning_rate": 6.905042255517393e-06, | |
| "loss": 2.2246, | |
| "step": 80400 | |
| }, | |
| { | |
| "epoch": 0.9215165529557214, | |
| "grad_norm": 0.4757671654224396, | |
| "learning_rate": 6.734572883175328e-06, | |
| "loss": 2.243, | |
| "step": 80500 | |
| }, | |
| { | |
| "epoch": 0.9226612940152937, | |
| "grad_norm": 0.5074242353439331, | |
| "learning_rate": 6.5651592163653885e-06, | |
| "loss": 2.243, | |
| "step": 80600 | |
| }, | |
| { | |
| "epoch": 0.923806035074866, | |
| "grad_norm": 0.4626462459564209, | |
| "learning_rate": 6.3968560261600545e-06, | |
| "loss": 2.2369, | |
| "step": 80700 | |
| }, | |
| { | |
| "epoch": 0.9249507761344384, | |
| "grad_norm": 0.4515824317932129, | |
| "learning_rate": 6.229717724617108e-06, | |
| "loss": 2.2479, | |
| "step": 80800 | |
| }, | |
| { | |
| "epoch": 0.9260955171940107, | |
| "grad_norm": 0.43820834159851074, | |
| "learning_rate": 6.063798347188343e-06, | |
| "loss": 2.2437, | |
| "step": 80900 | |
| }, | |
| { | |
| "epoch": 0.927240258253583, | |
| "grad_norm": 0.47702258825302124, | |
| "learning_rate": 5.899151535250031e-06, | |
| "loss": 2.2382, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 0.9283849993131553, | |
| "grad_norm": 0.4041842520236969, | |
| "learning_rate": 5.735830518760757e-06, | |
| "loss": 2.2412, | |
| "step": 81100 | |
| }, | |
| { | |
| "epoch": 0.9295297403727277, | |
| "grad_norm": 0.5503548979759216, | |
| "learning_rate": 5.573888099052307e-06, | |
| "loss": 2.2515, | |
| "step": 81200 | |
| }, | |
| { | |
| "epoch": 0.9306744814323, | |
| "grad_norm": 0.4641313850879669, | |
| "learning_rate": 5.413376631759115e-06, | |
| "loss": 2.2299, | |
| "step": 81300 | |
| }, | |
| { | |
| "epoch": 0.9318192224918723, | |
| "grad_norm": 0.4651617109775543, | |
| "learning_rate": 5.254348009891777e-06, | |
| "loss": 2.2394, | |
| "step": 81400 | |
| }, | |
| { | |
| "epoch": 0.9329639635514446, | |
| "grad_norm": 0.5168155431747437, | |
| "learning_rate": 5.096853647060169e-06, | |
| "loss": 2.2464, | |
| "step": 81500 | |
| }, | |
| { | |
| "epoch": 0.934108704611017, | |
| "grad_norm": 0.48371872305870056, | |
| "learning_rate": 4.940944460851545e-06, | |
| "loss": 2.2488, | |
| "step": 81600 | |
| }, | |
| { | |
| "epoch": 0.9352534456705893, | |
| "grad_norm": 0.4888920187950134, | |
| "learning_rate": 4.7866708563689654e-06, | |
| "loss": 2.2428, | |
| "step": 81700 | |
| }, | |
| { | |
| "epoch": 0.9363981867301616, | |
| "grad_norm": 0.4613405764102936, | |
| "learning_rate": 4.635600085943046e-06, | |
| "loss": 2.2641, | |
| "step": 81800 | |
| }, | |
| { | |
| "epoch": 0.9375429277897339, | |
| "grad_norm": 0.39950716495513916, | |
| "learning_rate": 4.484729138719958e-06, | |
| "loss": 2.2513, | |
| "step": 81900 | |
| }, | |
| { | |
| "epoch": 0.9386876688493063, | |
| "grad_norm": 0.44666656851768494, | |
| "learning_rate": 4.335641266650937e-06, | |
| "loss": 2.2352, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 0.9398324099088786, | |
| "grad_norm": 0.6108360886573792, | |
| "learning_rate": 4.188384669522936e-06, | |
| "loss": 2.2526, | |
| "step": 82100 | |
| }, | |
| { | |
| "epoch": 0.9409771509684509, | |
| "grad_norm": 0.4861396551132202, | |
| "learning_rate": 4.043006955075667e-06, | |
| "loss": 2.2421, | |
| "step": 82200 | |
| }, | |
| { | |
| "epoch": 0.9421218920280232, | |
| "grad_norm": 0.4066333472728729, | |
| "learning_rate": 3.899555123610131e-06, | |
| "loss": 2.2425, | |
| "step": 82300 | |
| }, | |
| { | |
| "epoch": 0.9432666330875956, | |
| "grad_norm": 0.47433343529701233, | |
| "learning_rate": 3.7580755527935232e-06, | |
| "loss": 2.2357, | |
| "step": 82400 | |
| }, | |
| { | |
| "epoch": 0.9444113741471679, | |
| "grad_norm": 0.447336882352829, | |
| "learning_rate": 3.6186139826654253e-06, | |
| "loss": 2.2458, | |
| "step": 82500 | |
| }, | |
| { | |
| "epoch": 0.9455561152067402, | |
| "grad_norm": 0.48044124245643616, | |
| "learning_rate": 3.4812155008501692e-06, | |
| "loss": 2.2451, | |
| "step": 82600 | |
| }, | |
| { | |
| "epoch": 0.9467008562663125, | |
| "grad_norm": 0.43815353512763977, | |
| "learning_rate": 3.3459245279800846e-06, | |
| "loss": 2.2457, | |
| "step": 82700 | |
| }, | |
| { | |
| "epoch": 0.9478455973258849, | |
| "grad_norm": 0.4744986295700073, | |
| "learning_rate": 3.2127848033344124e-06, | |
| "loss": 2.2303, | |
| "step": 82800 | |
| }, | |
| { | |
| "epoch": 0.9489903383854572, | |
| "grad_norm": 0.46926042437553406, | |
| "learning_rate": 3.0818393706984906e-06, | |
| "loss": 2.2346, | |
| "step": 82900 | |
| }, | |
| { | |
| "epoch": 0.9501350794450295, | |
| "grad_norm": 0.423065721988678, | |
| "learning_rate": 2.9531305644477883e-06, | |
| "loss": 2.2323, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 0.9512798205046019, | |
| "grad_norm": 0.4213075339794159, | |
| "learning_rate": 2.8266999958613017e-06, | |
| "loss": 2.2362, | |
| "step": 83100 | |
| }, | |
| { | |
| "epoch": 0.9524245615641742, | |
| "grad_norm": 0.45430952310562134, | |
| "learning_rate": 2.7025885396687145e-06, | |
| "loss": 2.2441, | |
| "step": 83200 | |
| }, | |
| { | |
| "epoch": 0.9535693026237465, | |
| "grad_norm": 0.48012691736221313, | |
| "learning_rate": 2.5808363208356746e-06, | |
| "loss": 2.2623, | |
| "step": 83300 | |
| }, | |
| { | |
| "epoch": 0.9547140436833188, | |
| "grad_norm": 0.4386986494064331, | |
| "learning_rate": 2.461482701591493e-06, | |
| "loss": 2.2329, | |
| "step": 83400 | |
| }, | |
| { | |
| "epoch": 0.9558587847428912, | |
| "grad_norm": 0.4368671476840973, | |
| "learning_rate": 2.3445662687034143e-06, | |
| "loss": 2.2391, | |
| "step": 83500 | |
| }, | |
| { | |
| "epoch": 0.9570035258024635, | |
| "grad_norm": 0.4472872316837311, | |
| "learning_rate": 2.2301248210016024e-06, | |
| "loss": 2.2378, | |
| "step": 83600 | |
| }, | |
| { | |
| "epoch": 0.9581482668620358, | |
| "grad_norm": 0.4506527781486511, | |
| "learning_rate": 2.1181953571588596e-06, | |
| "loss": 2.2362, | |
| "step": 83700 | |
| }, | |
| { | |
| "epoch": 0.9592930079216081, | |
| "grad_norm": 0.4752218425273895, | |
| "learning_rate": 2.0088140637290265e-06, | |
| "loss": 2.2397, | |
| "step": 83800 | |
| }, | |
| { | |
| "epoch": 0.9604377489811805, | |
| "grad_norm": 0.4612346589565277, | |
| "learning_rate": 1.9020163034479567e-06, | |
| "loss": 2.2447, | |
| "step": 83900 | |
| }, | |
| { | |
| "epoch": 0.9615824900407528, | |
| "grad_norm": 0.49490582942962646, | |
| "learning_rate": 1.7978366038008234e-06, | |
| "loss": 2.243, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 0.9627272311003251, | |
| "grad_norm": 0.4314175844192505, | |
| "learning_rate": 1.696308645859447e-06, | |
| "loss": 2.2398, | |
| "step": 84100 | |
| }, | |
| { | |
| "epoch": 0.9638719721598974, | |
| "grad_norm": 0.4729178249835968, | |
| "learning_rate": 1.5974652533932833e-06, | |
| "loss": 2.2528, | |
| "step": 84200 | |
| }, | |
| { | |
| "epoch": 0.9650167132194698, | |
| "grad_norm": 0.45053282380104065, | |
| "learning_rate": 1.5013383822575766e-06, | |
| "loss": 2.2393, | |
| "step": 84300 | |
| }, | |
| { | |
| "epoch": 0.9661614542790421, | |
| "grad_norm": 0.3972727358341217, | |
| "learning_rate": 1.4079591100620837e-06, | |
| "loss": 2.2551, | |
| "step": 84400 | |
| }, | |
| { | |
| "epoch": 0.9673061953386144, | |
| "grad_norm": 0.4374849200248718, | |
| "learning_rate": 1.317357626123772e-06, | |
| "loss": 2.2434, | |
| "step": 84500 | |
| }, | |
| { | |
| "epoch": 0.9684509363981867, | |
| "grad_norm": 0.44703078269958496, | |
| "learning_rate": 1.2295632217066567e-06, | |
| "loss": 2.2446, | |
| "step": 84600 | |
| }, | |
| { | |
| "epoch": 0.9695956774577591, | |
| "grad_norm": 0.49270525574684143, | |
| "learning_rate": 1.1446042805520098e-06, | |
| "loss": 2.2554, | |
| "step": 84700 | |
| }, | |
| { | |
| "epoch": 0.9707404185173314, | |
| "grad_norm": 0.45350950956344604, | |
| "learning_rate": 1.062508269701963e-06, | |
| "loss": 2.2649, | |
| "step": 84800 | |
| }, | |
| { | |
| "epoch": 0.9718851595769037, | |
| "grad_norm": 0.45871463418006897, | |
| "learning_rate": 9.833017306194558e-07, | |
| "loss": 2.2386, | |
| "step": 84900 | |
| }, | |
| { | |
| "epoch": 0.973029900636476, | |
| "grad_norm": 0.4371441900730133, | |
| "learning_rate": 9.07010270607459e-07, | |
| "loss": 2.2543, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 0.9741746416960484, | |
| "grad_norm": 0.47834667563438416, | |
| "learning_rate": 8.33658554530169e-07, | |
| "loss": 2.2385, | |
| "step": 85100 | |
| }, | |
| { | |
| "epoch": 0.9753193827556207, | |
| "grad_norm": 0.5165786743164062, | |
| "learning_rate": 7.632702968389205e-07, | |
| "loss": 2.2423, | |
| "step": 85200 | |
| }, | |
| { | |
| "epoch": 0.976464123815193, | |
| "grad_norm": 0.4625995457172394, | |
| "learning_rate": 6.958682539053563e-07, | |
| "loss": 2.2618, | |
| "step": 85300 | |
| }, | |
| { | |
| "epoch": 0.9776088648747653, | |
| "grad_norm": 0.5300964713096619, | |
| "learning_rate": 6.314742166643406e-07, | |
| "loss": 2.2383, | |
| "step": 85400 | |
| }, | |
| { | |
| "epoch": 0.9787536059343377, | |
| "grad_norm": 0.49308186769485474, | |
| "learning_rate": 5.701090035689949e-07, | |
| "loss": 2.2657, | |
| "step": 85500 | |
| }, | |
| { | |
| "epoch": 0.97989834699391, | |
| "grad_norm": 0.4634435176849365, | |
| "learning_rate": 5.117924538601371e-07, | |
| "loss": 2.2347, | |
| "step": 85600 | |
| }, | |
| { | |
| "epoch": 0.9810430880534823, | |
| "grad_norm": 0.4959612488746643, | |
| "learning_rate": 4.565434211522979e-07, | |
| "loss": 2.2421, | |
| "step": 85700 | |
| }, | |
| { | |
| "epoch": 0.9821878291130546, | |
| "grad_norm": 0.43508780002593994, | |
| "learning_rate": 4.0437976733838757e-07, | |
| "loss": 2.2601, | |
| "step": 85800 | |
| }, | |
| { | |
| "epoch": 0.983332570172627, | |
| "grad_norm": 0.4364226460456848, | |
| "learning_rate": 3.557935619007491e-07, | |
| "loss": 2.2362, | |
| "step": 85900 | |
| }, | |
| { | |
| "epoch": 0.9844773112321993, | |
| "grad_norm": 0.4956010580062866, | |
| "learning_rate": 3.09818999860273e-07, | |
| "loss": 2.2456, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 0.9856220522917716, | |
| "grad_norm": 0.45080068707466125, | |
| "learning_rate": 2.6697725240214076e-07, | |
| "loss": 2.2333, | |
| "step": 86100 | |
| }, | |
| { | |
| "epoch": 0.9867667933513439, | |
| "grad_norm": 0.47665461897850037, | |
| "learning_rate": 2.2728217017075395e-07, | |
| "loss": 2.2434, | |
| "step": 86200 | |
| }, | |
| { | |
| "epoch": 0.9879115344109163, | |
| "grad_norm": 0.48589542508125305, | |
| "learning_rate": 1.9074658650043764e-07, | |
| "loss": 2.2461, | |
| "step": 86300 | |
| }, | |
| { | |
| "epoch": 0.9890562754704886, | |
| "grad_norm": 0.47914570569992065, | |
| "learning_rate": 1.5738231326645758e-07, | |
| "loss": 2.2246, | |
| "step": 86400 | |
| }, | |
| { | |
| "epoch": 0.9902010165300609, | |
| "grad_norm": 0.5974397659301758, | |
| "learning_rate": 1.2720013706627122e-07, | |
| "loss": 2.227, | |
| "step": 86500 | |
| }, | |
| { | |
| "epoch": 0.9913457575896332, | |
| "grad_norm": 0.5031090378761292, | |
| "learning_rate": 1.002098157322362e-07, | |
| "loss": 2.2415, | |
| "step": 86600 | |
| }, | |
| { | |
| "epoch": 0.9924904986492056, | |
| "grad_norm": 0.470460444688797, | |
| "learning_rate": 7.642007517693062e-08, | |
| "loss": 2.2416, | |
| "step": 86700 | |
| }, | |
| { | |
| "epoch": 0.9936352397087779, | |
| "grad_norm": 0.4445749521255493, | |
| "learning_rate": 5.5838606572078404e-08, | |
| "loss": 2.228, | |
| "step": 86800 | |
| }, | |
| { | |
| "epoch": 0.9947799807683502, | |
| "grad_norm": 0.45365962386131287, | |
| "learning_rate": 3.847206386201507e-08, | |
| "loss": 2.2488, | |
| "step": 86900 | |
| }, | |
| { | |
| "epoch": 0.9959247218279226, | |
| "grad_norm": 0.4715825319290161, | |
| "learning_rate": 2.4326061612479633e-08, | |
| "loss": 2.2225, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 0.9970694628874949, | |
| "grad_norm": 0.4650324583053589, | |
| "learning_rate": 1.340517319543877e-08, | |
| "loss": 2.2315, | |
| "step": 87100 | |
| }, | |
| { | |
| "epoch": 0.9982142039470672, | |
| "grad_norm": 0.4570050537586212, | |
| "learning_rate": 5.712929310521809e-09, | |
| "loss": 2.2593, | |
| "step": 87200 | |
| }, | |
| { | |
| "epoch": 0.9993589450066395, | |
| "grad_norm": 0.47539758682250977, | |
| "learning_rate": 1.2518168435593502e-09, | |
| "loss": 2.2317, | |
| "step": 87300 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "step": 87356, | |
| "total_flos": 1.2729340395184456e+19, | |
| "train_loss": 2.3255439768606525, | |
| "train_runtime": 52634.826, | |
| "train_samples_per_second": 26.555, | |
| "train_steps_per_second": 1.66 | |
| } | |
| ], | |
| "logging_steps": 100, | |
| "max_steps": 87356, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.2729340395184456e+19, | |
| "train_batch_size": 2, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |