| [ | |
| { | |
| "loss": 1.3916, | |
| "grad_norm": 7.633892059326172, | |
| "learning_rate": 1.9831190798376187e-05, | |
| "epoch": 0.08457374830852503, | |
| "step": 500 | |
| }, | |
| { | |
| "loss": 1.2408, | |
| "grad_norm": 4.5788655281066895, | |
| "learning_rate": 1.9662043301759137e-05, | |
| "epoch": 0.16914749661705006, | |
| "step": 1000 | |
| }, | |
| { | |
| "loss": 1.2028, | |
| "grad_norm": 5.243614673614502, | |
| "learning_rate": 1.9492895805142083e-05, | |
| "epoch": 0.25372124492557513, | |
| "step": 1500 | |
| }, | |
| { | |
| "loss": 1.1534, | |
| "grad_norm": 3.415882110595703, | |
| "learning_rate": 1.9323748308525033e-05, | |
| "epoch": 0.3382949932341001, | |
| "step": 2000 | |
| }, | |
| { | |
| "loss": 1.1093, | |
| "grad_norm": 3.2337677478790283, | |
| "learning_rate": 1.9154600811907986e-05, | |
| "epoch": 0.42286874154262516, | |
| "step": 2500 | |
| }, | |
| { | |
| "loss": 1.0754, | |
| "grad_norm": 3.4433956146240234, | |
| "learning_rate": 1.8985453315290936e-05, | |
| "epoch": 0.5074424898511503, | |
| "step": 3000 | |
| }, | |
| { | |
| "loss": 1.0467, | |
| "grad_norm": 4.641908168792725, | |
| "learning_rate": 1.8816305818673886e-05, | |
| "epoch": 0.5920162381596752, | |
| "step": 3500 | |
| }, | |
| { | |
| "loss": 1.0355, | |
| "grad_norm": 3.7774765491485596, | |
| "learning_rate": 1.8647158322056836e-05, | |
| "epoch": 0.6765899864682002, | |
| "step": 4000 | |
| }, | |
| { | |
| "loss": 1.0421, | |
| "grad_norm": 3.130302906036377, | |
| "learning_rate": 1.8478349120433018e-05, | |
| "epoch": 0.7611637347767253, | |
| "step": 4500 | |
| }, | |
| { | |
| "loss": 1.0158, | |
| "grad_norm": 2.876955509185791, | |
| "learning_rate": 1.8309201623815968e-05, | |
| "epoch": 0.8457374830852503, | |
| "step": 5000 | |
| }, | |
| { | |
| "loss": 1.0, | |
| "grad_norm": 3.218794107437134, | |
| "learning_rate": 1.8140054127198918e-05, | |
| "epoch": 0.9303112313937754, | |
| "step": 5500 | |
| }, | |
| { | |
| "loss": 0.9665, | |
| "grad_norm": 2.9098243713378906, | |
| "learning_rate": 1.7971244925575103e-05, | |
| "epoch": 1.0148849797023005, | |
| "step": 6000 | |
| }, | |
| { | |
| "loss": 0.8815, | |
| "grad_norm": 2.203686475753784, | |
| "learning_rate": 1.7802097428958052e-05, | |
| "epoch": 1.0994587280108254, | |
| "step": 6500 | |
| }, | |
| { | |
| "loss": 0.9049, | |
| "grad_norm": 3.389420986175537, | |
| "learning_rate": 1.7632949932341002e-05, | |
| "epoch": 1.1840324763193504, | |
| "step": 7000 | |
| }, | |
| { | |
| "loss": 0.9059, | |
| "grad_norm": 2.778923749923706, | |
| "learning_rate": 1.7463802435723952e-05, | |
| "epoch": 1.2686062246278755, | |
| "step": 7500 | |
| }, | |
| { | |
| "loss": 0.8858, | |
| "grad_norm": 5.216675758361816, | |
| "learning_rate": 1.7294993234100137e-05, | |
| "epoch": 1.3531799729364005, | |
| "step": 8000 | |
| }, | |
| { | |
| "loss": 0.8881, | |
| "grad_norm": 3.9376771450042725, | |
| "learning_rate": 1.7125845737483087e-05, | |
| "epoch": 1.4377537212449256, | |
| "step": 8500 | |
| }, | |
| { | |
| "loss": 0.8885, | |
| "grad_norm": 3.8835389614105225, | |
| "learning_rate": 1.6956698240866037e-05, | |
| "epoch": 1.5223274695534506, | |
| "step": 9000 | |
| }, | |
| { | |
| "loss": 0.8774, | |
| "grad_norm": 3.470211982727051, | |
| "learning_rate": 1.6787550744248987e-05, | |
| "epoch": 1.6069012178619757, | |
| "step": 9500 | |
| }, | |
| { | |
| "loss": 0.872, | |
| "grad_norm": 3.03437876701355, | |
| "learning_rate": 1.6618403247631937e-05, | |
| "epoch": 1.6914749661705006, | |
| "step": 10000 | |
| }, | |
| { | |
| "loss": 0.8883, | |
| "grad_norm": 3.5217363834381104, | |
| "learning_rate": 1.6449255751014887e-05, | |
| "epoch": 1.7760487144790256, | |
| "step": 10500 | |
| }, | |
| { | |
| "loss": 0.8614, | |
| "grad_norm": 3.965338706970215, | |
| "learning_rate": 1.6280446549391072e-05, | |
| "epoch": 1.8606224627875507, | |
| "step": 11000 | |
| }, | |
| { | |
| "loss": 0.8682, | |
| "grad_norm": 2.316436767578125, | |
| "learning_rate": 1.6111299052774022e-05, | |
| "epoch": 1.9451962110960759, | |
| "step": 11500 | |
| }, | |
| { | |
| "loss": 0.8286, | |
| "grad_norm": 3.9571282863616943, | |
| "learning_rate": 1.5942151556156972e-05, | |
| "epoch": 2.029769959404601, | |
| "step": 12000 | |
| }, | |
| { | |
| "loss": 0.7808, | |
| "grad_norm": 3.676339864730835, | |
| "learning_rate": 1.577300405953992e-05, | |
| "epoch": 2.1143437077131257, | |
| "step": 12500 | |
| }, | |
| { | |
| "loss": 0.7811, | |
| "grad_norm": 2.8704166412353516, | |
| "learning_rate": 1.560385656292287e-05, | |
| "epoch": 2.198917456021651, | |
| "step": 13000 | |
| }, | |
| { | |
| "loss": 0.8, | |
| "grad_norm": 4.323342800140381, | |
| "learning_rate": 1.5434709066305818e-05, | |
| "epoch": 2.283491204330176, | |
| "step": 13500 | |
| }, | |
| { | |
| "loss": 0.79, | |
| "grad_norm": 2.5053164958953857, | |
| "learning_rate": 1.5265561569688768e-05, | |
| "epoch": 2.3680649526387008, | |
| "step": 14000 | |
| }, | |
| { | |
| "loss": 0.7849, | |
| "grad_norm": 3.5477893352508545, | |
| "learning_rate": 1.509641407307172e-05, | |
| "epoch": 2.452638700947226, | |
| "step": 14500 | |
| }, | |
| { | |
| "loss": 0.7839, | |
| "grad_norm": 3.699144124984741, | |
| "learning_rate": 1.4927604871447903e-05, | |
| "epoch": 2.537212449255751, | |
| "step": 15000 | |
| }, | |
| { | |
| "loss": 0.799, | |
| "grad_norm": 2.969682455062866, | |
| "learning_rate": 1.4758795669824088e-05, | |
| "epoch": 2.621786197564276, | |
| "step": 15500 | |
| }, | |
| { | |
| "loss": 0.7799, | |
| "grad_norm": 4.236715316772461, | |
| "learning_rate": 1.4589648173207038e-05, | |
| "epoch": 2.706359945872801, | |
| "step": 16000 | |
| }, | |
| { | |
| "loss": 0.7756, | |
| "grad_norm": 3.2528302669525146, | |
| "learning_rate": 1.4420500676589988e-05, | |
| "epoch": 2.790933694181326, | |
| "step": 16500 | |
| }, | |
| { | |
| "loss": 0.7837, | |
| "grad_norm": 4.050159454345703, | |
| "learning_rate": 1.4251353179972938e-05, | |
| "epoch": 2.8755074424898512, | |
| "step": 17000 | |
| }, | |
| { | |
| "loss": 0.7907, | |
| "grad_norm": 3.53711199760437, | |
| "learning_rate": 1.4082543978349121e-05, | |
| "epoch": 2.960081190798376, | |
| "step": 17500 | |
| }, | |
| { | |
| "loss": 0.7533, | |
| "grad_norm": 4.171680927276611, | |
| "learning_rate": 1.3913396481732071e-05, | |
| "epoch": 3.044654939106901, | |
| "step": 18000 | |
| }, | |
| { | |
| "loss": 0.724, | |
| "grad_norm": 3.026613712310791, | |
| "learning_rate": 1.3744248985115021e-05, | |
| "epoch": 3.1292286874154263, | |
| "step": 18500 | |
| }, | |
| { | |
| "loss": 0.7289, | |
| "grad_norm": 3.7827978134155273, | |
| "learning_rate": 1.357510148849797e-05, | |
| "epoch": 3.2138024357239514, | |
| "step": 19000 | |
| }, | |
| { | |
| "loss": 0.7078, | |
| "grad_norm": 2.7544713020324707, | |
| "learning_rate": 1.3405953991880922e-05, | |
| "epoch": 3.2983761840324766, | |
| "step": 19500 | |
| }, | |
| { | |
| "loss": 0.7154, | |
| "grad_norm": 2.6823747158050537, | |
| "learning_rate": 1.3236806495263872e-05, | |
| "epoch": 3.3829499323410013, | |
| "step": 20000 | |
| }, | |
| { | |
| "loss": 0.7314, | |
| "grad_norm": 3.3478825092315674, | |
| "learning_rate": 1.3067658998646822e-05, | |
| "epoch": 3.4675236806495264, | |
| "step": 20500 | |
| }, | |
| { | |
| "loss": 0.726, | |
| "grad_norm": 3.5908212661743164, | |
| "learning_rate": 1.289851150202977e-05, | |
| "epoch": 3.5520974289580516, | |
| "step": 21000 | |
| }, | |
| { | |
| "loss": 0.7285, | |
| "grad_norm": 3.1159133911132812, | |
| "learning_rate": 1.2729702300405956e-05, | |
| "epoch": 3.6366711772665763, | |
| "step": 21500 | |
| }, | |
| { | |
| "loss": 0.7162, | |
| "grad_norm": 3.3147387504577637, | |
| "learning_rate": 1.2560554803788905e-05, | |
| "epoch": 3.7212449255751014, | |
| "step": 22000 | |
| }, | |
| { | |
| "loss": 0.7233, | |
| "grad_norm": 3.7307050228118896, | |
| "learning_rate": 1.2391407307171854e-05, | |
| "epoch": 3.8058186738836266, | |
| "step": 22500 | |
| }, | |
| { | |
| "loss": 0.7159, | |
| "grad_norm": 2.382382392883301, | |
| "learning_rate": 1.2222259810554804e-05, | |
| "epoch": 3.8903924221921518, | |
| "step": 23000 | |
| }, | |
| { | |
| "loss": 0.722, | |
| "grad_norm": 4.039222717285156, | |
| "learning_rate": 1.2053112313937754e-05, | |
| "epoch": 3.9749661705006765, | |
| "step": 23500 | |
| }, | |
| { | |
| "loss": 0.6786, | |
| "grad_norm": 3.7908201217651367, | |
| "learning_rate": 1.188430311231394e-05, | |
| "epoch": 4.059539918809202, | |
| "step": 24000 | |
| }, | |
| { | |
| "loss": 0.6735, | |
| "grad_norm": 3.7970995903015137, | |
| "learning_rate": 1.171515561569689e-05, | |
| "epoch": 4.144113667117726, | |
| "step": 24500 | |
| }, | |
| { | |
| "loss": 0.6736, | |
| "grad_norm": 4.007111549377441, | |
| "learning_rate": 1.1546008119079838e-05, | |
| "epoch": 4.2286874154262515, | |
| "step": 25000 | |
| }, | |
| { | |
| "loss": 0.664, | |
| "grad_norm": 3.4510090351104736, | |
| "learning_rate": 1.1376860622462788e-05, | |
| "epoch": 4.313261163734777, | |
| "step": 25500 | |
| }, | |
| { | |
| "loss": 0.6724, | |
| "grad_norm": 3.279106378555298, | |
| "learning_rate": 1.1207713125845738e-05, | |
| "epoch": 4.397834912043302, | |
| "step": 26000 | |
| }, | |
| { | |
| "loss": 0.6696, | |
| "grad_norm": 2.7026331424713135, | |
| "learning_rate": 1.1038565629228688e-05, | |
| "epoch": 4.482408660351827, | |
| "step": 26500 | |
| }, | |
| { | |
| "loss": 0.6738, | |
| "grad_norm": 3.247185230255127, | |
| "learning_rate": 1.0869418132611638e-05, | |
| "epoch": 4.566982408660352, | |
| "step": 27000 | |
| }, | |
| { | |
| "loss": 0.6908, | |
| "grad_norm": 3.6047909259796143, | |
| "learning_rate": 1.0700608930987821e-05, | |
| "epoch": 4.651556156968876, | |
| "step": 27500 | |
| }, | |
| { | |
| "loss": 0.6687, | |
| "grad_norm": 4.114670753479004, | |
| "learning_rate": 1.0531461434370771e-05, | |
| "epoch": 4.7361299052774015, | |
| "step": 28000 | |
| }, | |
| { | |
| "loss": 0.6632, | |
| "grad_norm": 2.717122793197632, | |
| "learning_rate": 1.0362313937753723e-05, | |
| "epoch": 4.820703653585927, | |
| "step": 28500 | |
| }, | |
| { | |
| "loss": 0.6842, | |
| "grad_norm": 2.6075901985168457, | |
| "learning_rate": 1.0193166441136673e-05, | |
| "epoch": 4.905277401894452, | |
| "step": 29000 | |
| }, | |
| { | |
| "loss": 0.6767, | |
| "grad_norm": 3.6151123046875, | |
| "learning_rate": 1.0024018944519623e-05, | |
| "epoch": 4.989851150202977, | |
| "step": 29500 | |
| }, | |
| { | |
| "loss": 0.6441, | |
| "grad_norm": 3.7179670333862305, | |
| "learning_rate": 9.855209742895806e-06, | |
| "epoch": 5.074424898511502, | |
| "step": 30000 | |
| }, | |
| { | |
| "loss": 0.6274, | |
| "grad_norm": 3.9154396057128906, | |
| "learning_rate": 9.686062246278756e-06, | |
| "epoch": 5.158998646820027, | |
| "step": 30500 | |
| }, | |
| { | |
| "loss": 0.6214, | |
| "grad_norm": 2.7849080562591553, | |
| "learning_rate": 9.516914749661706e-06, | |
| "epoch": 5.243572395128552, | |
| "step": 31000 | |
| }, | |
| { | |
| "loss": 0.6362, | |
| "grad_norm": 3.1513593196868896, | |
| "learning_rate": 9.347767253044656e-06, | |
| "epoch": 5.328146143437078, | |
| "step": 31500 | |
| }, | |
| { | |
| "loss": 0.6366, | |
| "grad_norm": 3.0636239051818848, | |
| "learning_rate": 9.178619756427606e-06, | |
| "epoch": 5.412719891745602, | |
| "step": 32000 | |
| }, | |
| { | |
| "loss": 0.635, | |
| "grad_norm": 3.542881727218628, | |
| "learning_rate": 9.009810554803789e-06, | |
| "epoch": 5.497293640054127, | |
| "step": 32500 | |
| }, | |
| { | |
| "loss": 0.629, | |
| "grad_norm": 2.9938108921051025, | |
| "learning_rate": 8.840663058186739e-06, | |
| "epoch": 5.581867388362652, | |
| "step": 33000 | |
| }, | |
| { | |
| "loss": 0.6424, | |
| "grad_norm": 3.608818769454956, | |
| "learning_rate": 8.67151556156969e-06, | |
| "epoch": 5.666441136671177, | |
| "step": 33500 | |
| }, | |
| { | |
| "loss": 0.6354, | |
| "grad_norm": 4.858671188354492, | |
| "learning_rate": 8.50236806495264e-06, | |
| "epoch": 5.7510148849797025, | |
| "step": 34000 | |
| }, | |
| { | |
| "loss": 0.6365, | |
| "grad_norm": 3.254009246826172, | |
| "learning_rate": 8.333220568335589e-06, | |
| "epoch": 5.835588633288228, | |
| "step": 34500 | |
| }, | |
| { | |
| "loss": 0.6343, | |
| "grad_norm": 2.389611005783081, | |
| "learning_rate": 8.164073071718539e-06, | |
| "epoch": 5.920162381596752, | |
| "step": 35000 | |
| }, | |
| { | |
| "loss": 0.6305, | |
| "grad_norm": 3.2198381423950195, | |
| "learning_rate": 7.994925575101489e-06, | |
| "epoch": 6.004736129905277, | |
| "step": 35500 | |
| }, | |
| { | |
| "loss": 0.597, | |
| "grad_norm": 2.834723711013794, | |
| "learning_rate": 7.82577807848444e-06, | |
| "epoch": 6.089309878213802, | |
| "step": 36000 | |
| }, | |
| { | |
| "loss": 0.5914, | |
| "grad_norm": 2.7054672241210938, | |
| "learning_rate": 7.656630581867388e-06, | |
| "epoch": 6.173883626522327, | |
| "step": 36500 | |
| }, | |
| { | |
| "loss": 0.6084, | |
| "grad_norm": 2.8164889812469482, | |
| "learning_rate": 7.487483085250339e-06, | |
| "epoch": 6.2584573748308525, | |
| "step": 37000 | |
| }, | |
| { | |
| "loss": 0.5947, | |
| "grad_norm": 3.42501163482666, | |
| "learning_rate": 7.3186738836265225e-06, | |
| "epoch": 6.343031123139378, | |
| "step": 37500 | |
| }, | |
| { | |
| "loss": 0.604, | |
| "grad_norm": 3.881469249725342, | |
| "learning_rate": 7.149526387009473e-06, | |
| "epoch": 6.427604871447903, | |
| "step": 38000 | |
| }, | |
| { | |
| "loss": 0.5896, | |
| "grad_norm": 3.2387328147888184, | |
| "learning_rate": 6.980378890392423e-06, | |
| "epoch": 6.512178619756428, | |
| "step": 38500 | |
| }, | |
| { | |
| "loss": 0.5985, | |
| "grad_norm": 3.245598316192627, | |
| "learning_rate": 6.811231393775373e-06, | |
| "epoch": 6.596752368064953, | |
| "step": 39000 | |
| }, | |
| { | |
| "loss": 0.6207, | |
| "grad_norm": 4.7686448097229, | |
| "learning_rate": 6.642422192151556e-06, | |
| "epoch": 6.681326116373477, | |
| "step": 39500 | |
| }, | |
| { | |
| "loss": 0.6164, | |
| "grad_norm": 3.3545920848846436, | |
| "learning_rate": 6.473274695534507e-06, | |
| "epoch": 6.7658998646820026, | |
| "step": 40000 | |
| }, | |
| { | |
| "loss": 0.5994, | |
| "grad_norm": 3.037534713745117, | |
| "learning_rate": 6.304127198917457e-06, | |
| "epoch": 6.850473612990528, | |
| "step": 40500 | |
| }, | |
| { | |
| "loss": 0.6249, | |
| "grad_norm": 4.4626784324646, | |
| "learning_rate": 6.134979702300406e-06, | |
| "epoch": 6.935047361299053, | |
| "step": 41000 | |
| }, | |
| { | |
| "loss": 0.5985, | |
| "grad_norm": 2.8611857891082764, | |
| "learning_rate": 5.96617050067659e-06, | |
| "epoch": 7.019621109607578, | |
| "step": 41500 | |
| }, | |
| { | |
| "loss": 0.5841, | |
| "grad_norm": 3.028613805770874, | |
| "learning_rate": 5.79702300405954e-06, | |
| "epoch": 7.104194857916103, | |
| "step": 42000 | |
| }, | |
| { | |
| "loss": 0.586, | |
| "grad_norm": 2.902698040008545, | |
| "learning_rate": 5.627875507442491e-06, | |
| "epoch": 7.188768606224627, | |
| "step": 42500 | |
| }, | |
| { | |
| "loss": 0.5869, | |
| "grad_norm": 2.2707433700561523, | |
| "learning_rate": 5.45872801082544e-06, | |
| "epoch": 7.273342354533153, | |
| "step": 43000 | |
| }, | |
| { | |
| "loss": 0.5888, | |
| "grad_norm": 3.4720981121063232, | |
| "learning_rate": 5.289918809201624e-06, | |
| "epoch": 7.357916102841678, | |
| "step": 43500 | |
| }, | |
| { | |
| "loss": 0.576, | |
| "grad_norm": 2.8364577293395996, | |
| "learning_rate": 5.120771312584573e-06, | |
| "epoch": 7.442489851150203, | |
| "step": 44000 | |
| }, | |
| { | |
| "loss": 0.5759, | |
| "grad_norm": 3.178103446960449, | |
| "learning_rate": 4.951623815967524e-06, | |
| "epoch": 7.527063599458728, | |
| "step": 44500 | |
| }, | |
| { | |
| "loss": 0.5742, | |
| "grad_norm": 3.5063467025756836, | |
| "learning_rate": 4.782476319350474e-06, | |
| "epoch": 7.611637347767253, | |
| "step": 45000 | |
| }, | |
| { | |
| "loss": 0.5912, | |
| "grad_norm": 2.37205171585083, | |
| "learning_rate": 4.613328822733424e-06, | |
| "epoch": 7.696211096075778, | |
| "step": 45500 | |
| }, | |
| { | |
| "loss": 0.575, | |
| "grad_norm": 3.2511661052703857, | |
| "learning_rate": 4.444181326116374e-06, | |
| "epoch": 7.7807848443843035, | |
| "step": 46000 | |
| }, | |
| { | |
| "loss": 0.5724, | |
| "grad_norm": 3.2974693775177, | |
| "learning_rate": 4.275033829499324e-06, | |
| "epoch": 7.865358592692828, | |
| "step": 46500 | |
| }, | |
| { | |
| "loss": 0.5745, | |
| "grad_norm": 3.180819511413574, | |
| "learning_rate": 4.105886332882274e-06, | |
| "epoch": 7.949932341001353, | |
| "step": 47000 | |
| }, | |
| { | |
| "loss": 0.5697, | |
| "grad_norm": 2.4791033267974854, | |
| "learning_rate": 3.937077131258458e-06, | |
| "epoch": 8.034506089309879, | |
| "step": 47500 | |
| }, | |
| { | |
| "loss": 0.556, | |
| "grad_norm": 3.5898892879486084, | |
| "learning_rate": 3.7679296346414073e-06, | |
| "epoch": 8.119079837618404, | |
| "step": 48000 | |
| }, | |
| { | |
| "loss": 0.5726, | |
| "grad_norm": 2.7320892810821533, | |
| "learning_rate": 3.5987821380243577e-06, | |
| "epoch": 8.203653585926928, | |
| "step": 48500 | |
| }, | |
| { | |
| "loss": 0.5603, | |
| "grad_norm": 3.3177103996276855, | |
| "learning_rate": 3.429634641407307e-06, | |
| "epoch": 8.288227334235453, | |
| "step": 49000 | |
| }, | |
| { | |
| "loss": 0.558, | |
| "grad_norm": 2.1732170581817627, | |
| "learning_rate": 3.2604871447902575e-06, | |
| "epoch": 8.372801082543978, | |
| "step": 49500 | |
| }, | |
| { | |
| "loss": 0.5631, | |
| "grad_norm": 3.1351735591888428, | |
| "learning_rate": 3.091339648173207e-06, | |
| "epoch": 8.457374830852503, | |
| "step": 50000 | |
| }, | |
| { | |
| "loss": 0.5486, | |
| "grad_norm": 2.601547956466675, | |
| "learning_rate": 2.9221921515561573e-06, | |
| "epoch": 8.541948579161028, | |
| "step": 50500 | |
| }, | |
| { | |
| "loss": 0.5706, | |
| "grad_norm": 4.160942554473877, | |
| "learning_rate": 2.753382949932341e-06, | |
| "epoch": 8.626522327469553, | |
| "step": 51000 | |
| }, | |
| { | |
| "loss": 0.5683, | |
| "grad_norm": 3.348295211791992, | |
| "learning_rate": 2.5842354533152914e-06, | |
| "epoch": 8.711096075778078, | |
| "step": 51500 | |
| }, | |
| { | |
| "loss": 0.564, | |
| "grad_norm": 3.148343563079834, | |
| "learning_rate": 2.415087956698241e-06, | |
| "epoch": 8.795669824086604, | |
| "step": 52000 | |
| }, | |
| { | |
| "loss": 0.5725, | |
| "grad_norm": 3.285578489303589, | |
| "learning_rate": 2.246278755074425e-06, | |
| "epoch": 8.880243572395129, | |
| "step": 52500 | |
| }, | |
| { | |
| "loss": 0.567, | |
| "grad_norm": 3.201730251312256, | |
| "learning_rate": 2.077131258457375e-06, | |
| "epoch": 8.964817320703654, | |
| "step": 53000 | |
| }, | |
| { | |
| "loss": 0.553, | |
| "grad_norm": 3.166001796722412, | |
| "learning_rate": 1.907983761840325e-06, | |
| "epoch": 9.049391069012179, | |
| "step": 53500 | |
| }, | |
| { | |
| "loss": 0.5614, | |
| "grad_norm": 3.105032444000244, | |
| "learning_rate": 1.7388362652232748e-06, | |
| "epoch": 9.133964817320704, | |
| "step": 54000 | |
| }, | |
| { | |
| "loss": 0.5553, | |
| "grad_norm": 4.7028937339782715, | |
| "learning_rate": 1.5696887686062248e-06, | |
| "epoch": 9.21853856562923, | |
| "step": 54500 | |
| }, | |
| { | |
| "loss": 0.552, | |
| "grad_norm": 3.5488646030426025, | |
| "learning_rate": 1.4005412719891747e-06, | |
| "epoch": 9.303112313937755, | |
| "step": 55000 | |
| }, | |
| { | |
| "loss": 0.5483, | |
| "grad_norm": 2.8257858753204346, | |
| "learning_rate": 1.2313937753721246e-06, | |
| "epoch": 9.387686062246278, | |
| "step": 55500 | |
| }, | |
| { | |
| "loss": 0.5476, | |
| "grad_norm": 4.221645355224609, | |
| "learning_rate": 1.0622462787550745e-06, | |
| "epoch": 9.472259810554803, | |
| "step": 56000 | |
| }, | |
| { | |
| "loss": 0.5494, | |
| "grad_norm": 3.1773674488067627, | |
| "learning_rate": 8.930987821380243e-07, | |
| "epoch": 9.556833558863328, | |
| "step": 56500 | |
| }, | |
| { | |
| "loss": 0.5477, | |
| "grad_norm": 2.9824230670928955, | |
| "learning_rate": 7.242895805142085e-07, | |
| "epoch": 9.641407307171853, | |
| "step": 57000 | |
| }, | |
| { | |
| "loss": 0.5579, | |
| "grad_norm": 2.51481294631958, | |
| "learning_rate": 5.551420838971583e-07, | |
| "epoch": 9.725981055480379, | |
| "step": 57500 | |
| }, | |
| { | |
| "loss": 0.5392, | |
| "grad_norm": 3.0290215015411377, | |
| "learning_rate": 3.8599458728010834e-07, | |
| "epoch": 9.810554803788904, | |
| "step": 58000 | |
| }, | |
| { | |
| "loss": 0.5458, | |
| "grad_norm": 2.9967031478881836, | |
| "learning_rate": 2.168470906630582e-07, | |
| "epoch": 9.895128552097429, | |
| "step": 58500 | |
| }, | |
| { | |
| "loss": 0.5434, | |
| "grad_norm": 2.6248016357421875, | |
| "learning_rate": 4.769959404600812e-08, | |
| "epoch": 9.979702300405954, | |
| "step": 59000 | |
| }, | |
| { | |
| "train_runtime": 8128.9827, | |
| "train_samples_per_second": 232.692, | |
| "train_steps_per_second": 7.273, | |
| "total_flos": 4.123593985189478e+16, | |
| "train_loss": 0.7104815463735867, | |
| "epoch": 10.0, | |
| "step": 59120 | |
| }, | |
| { | |
| "eval_loss": 0.7965446710586548, | |
| "eval_model_preparation_time": 0.002, | |
| "eval_bleu": 54.95663610516514, | |
| "eval_runtime": 6012.2223, | |
| "eval_samples_per_second": 3.496, | |
| "eval_steps_per_second": 0.055, | |
| "epoch": 10.0, | |
| "step": 59120 | |
| } | |
| ] |