| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 14.0, | |
| "eval_steps": 500, | |
| "global_step": 97538, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 0.00014849999999999998, | |
| "loss": 16.3517, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 0.0002985, | |
| "loss": 4.5739, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.00029856528670112555, | |
| "loss": 2.3946, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "learning_rate": 0.00029711608134872706, | |
| "loss": 1.7079, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "learning_rate": 0.0002956668759963287, | |
| "loss": 1.5216, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 0.0002942176706439302, | |
| "loss": 1.4303, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 0.00029277136370223656, | |
| "loss": 1.343, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 0.0002913221583498382, | |
| "loss": 1.2865, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 0.0002898729529974397, | |
| "loss": 1.2574, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 0.00028842374764504125, | |
| "loss": 1.218, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 0.0002869745422926428, | |
| "loss": 1.1774, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 0.0002855253369402444, | |
| "loss": 1.1657, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 0.00028407613158784595, | |
| "loss": 1.1279, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 0.0002826269262354475, | |
| "loss": 1.1115, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 0.00028117772088304913, | |
| "loss": 1.0697, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 0.00027972851553065064, | |
| "loss": 1.047, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 0.00027827931017825226, | |
| "loss": 1.0408, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 0.00027683010482585377, | |
| "loss": 1.0267, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 1.36, | |
| "learning_rate": 0.0002753808994734554, | |
| "loss": 1.0195, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 0.0002739374909424665, | |
| "loss": 1.0126, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 1.51, | |
| "learning_rate": 0.00027248828559006807, | |
| "loss": 1.01, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 0.0002710390802376697, | |
| "loss": 0.9987, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 0.0002695898748852712, | |
| "loss": 0.9852, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 0.0002681406695328728, | |
| "loss": 0.9706, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 0.00026669146418047433, | |
| "loss": 0.9753, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 0.00026524225882807595, | |
| "loss": 0.9916, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 0.00026379305347567746, | |
| "loss": 0.9812, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 2.01, | |
| "learning_rate": 0.0002623467465339838, | |
| "loss": 0.9579, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 0.0002608975411815854, | |
| "loss": 0.9153, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 0.00025944833582918696, | |
| "loss": 0.9142, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 0.0002579991304767885, | |
| "loss": 0.9152, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 0.0002565499251243901, | |
| "loss": 0.8964, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 0.00025510361818269645, | |
| "loss": 0.9001, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 0.000253654412830298, | |
| "loss": 0.8901, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 0.0002522081058886044, | |
| "loss": 0.8914, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 0.00025075890053620595, | |
| "loss": 0.8872, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 0.0002493096951838075, | |
| "loss": 0.8726, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 0.0002478604898314091, | |
| "loss": 0.8719, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 0.00024641418288971545, | |
| "loss": 0.8746, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 0.000244964977537317, | |
| "loss": 0.8657, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 0.00024351577218491858, | |
| "loss": 0.8712, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 0.00024206656683252017, | |
| "loss": 0.8406, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 0.0002406173614801217, | |
| "loss": 0.8118, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 0.0002391681561277233, | |
| "loss": 0.8327, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 3.23, | |
| "learning_rate": 0.00023771895077532483, | |
| "loss": 0.8094, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 0.0002362726438336312, | |
| "loss": 0.8134, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 0.00023482343848123277, | |
| "loss": 0.7993, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 0.00023337423312883433, | |
| "loss": 0.8074, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 0.0002319250277764359, | |
| "loss": 0.7971, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 0.0002304758224240375, | |
| "loss": 0.7938, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 0.00022902661707163902, | |
| "loss": 0.8041, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 0.0002275803101299454, | |
| "loss": 0.7918, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 3.8, | |
| "learning_rate": 0.00022613110477754696, | |
| "loss": 0.7848, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 3.88, | |
| "learning_rate": 0.00022468189942514852, | |
| "loss": 0.7887, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 3.95, | |
| "learning_rate": 0.00022323559248345486, | |
| "loss": 0.7707, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 0.00022178638713105645, | |
| "loss": 0.7574, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 0.000220337181778658, | |
| "loss": 0.7468, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 4.16, | |
| "learning_rate": 0.00021888797642625958, | |
| "loss": 0.7387, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 4.23, | |
| "learning_rate": 0.00021743877107386117, | |
| "loss": 0.7381, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 4.31, | |
| "learning_rate": 0.00021599246413216752, | |
| "loss": 0.733, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 4.38, | |
| "learning_rate": 0.00021454325877976908, | |
| "loss": 0.7271, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 4.45, | |
| "learning_rate": 0.00021309405342737064, | |
| "loss": 0.7362, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 4.52, | |
| "learning_rate": 0.00021164484807497218, | |
| "loss": 0.7337, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 4.59, | |
| "learning_rate": 0.00021019564272257377, | |
| "loss": 0.727, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 4.66, | |
| "learning_rate": 0.0002087464373701753, | |
| "loss": 0.7262, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "learning_rate": 0.0002072972320177769, | |
| "loss": 0.7257, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 4.81, | |
| "learning_rate": 0.00020584802666537844, | |
| "loss": 0.7228, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 4.88, | |
| "learning_rate": 0.00020439882131298003, | |
| "loss": 0.7281, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 4.95, | |
| "learning_rate": 0.0002029525143712864, | |
| "loss": 0.7263, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 0.00020150330901888796, | |
| "loss": 0.7053, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 5.1, | |
| "learning_rate": 0.00020005410366648953, | |
| "loss": 0.681, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 5.17, | |
| "learning_rate": 0.00019860779672479587, | |
| "loss": 0.6896, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 5.24, | |
| "learning_rate": 0.00019715859137239746, | |
| "loss": 0.6886, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 5.31, | |
| "learning_rate": 0.000195709386019999, | |
| "loss": 0.6883, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 5.38, | |
| "learning_rate": 0.0001942601806676006, | |
| "loss": 0.6835, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 5.45, | |
| "learning_rate": 0.00019281097531520213, | |
| "loss": 0.694, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 5.53, | |
| "learning_rate": 0.00019136176996280372, | |
| "loss": 0.6928, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 5.6, | |
| "learning_rate": 0.00018991256461040528, | |
| "loss": 0.689, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 5.67, | |
| "learning_rate": 0.00018846625766871165, | |
| "loss": 0.6876, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 5.74, | |
| "learning_rate": 0.0001870170523163132, | |
| "loss": 0.6777, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 5.81, | |
| "learning_rate": 0.00018556784696391478, | |
| "loss": 0.6777, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 5.88, | |
| "learning_rate": 0.00018411864161151632, | |
| "loss": 0.6744, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 5.96, | |
| "learning_rate": 0.0001826694362591179, | |
| "loss": 0.667, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 6.03, | |
| "learning_rate": 0.00018122023090671945, | |
| "loss": 0.6599, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 6.1, | |
| "learning_rate": 0.00017977102555432104, | |
| "loss": 0.63, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 6.17, | |
| "learning_rate": 0.00017832182020192258, | |
| "loss": 0.6447, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 6.24, | |
| "learning_rate": 0.00017687261484952417, | |
| "loss": 0.6461, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 6.32, | |
| "learning_rate": 0.0001754234094971257, | |
| "loss": 0.6407, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 6.39, | |
| "learning_rate": 0.0001739771025554321, | |
| "loss": 0.6391, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 6.46, | |
| "learning_rate": 0.00017252789720303364, | |
| "loss": 0.6352, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 6.53, | |
| "learning_rate": 0.00017107869185063523, | |
| "loss": 0.6365, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 6.6, | |
| "learning_rate": 0.00016962948649823677, | |
| "loss": 0.6407, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 6.67, | |
| "learning_rate": 0.00016818028114583836, | |
| "loss": 0.6432, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 6.75, | |
| "learning_rate": 0.0001667310757934399, | |
| "loss": 0.6268, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 6.82, | |
| "learning_rate": 0.0001652818704410415, | |
| "loss": 0.6371, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 6.89, | |
| "learning_rate": 0.00016383266508864303, | |
| "loss": 0.6351, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 6.96, | |
| "learning_rate": 0.00016238635814694942, | |
| "loss": 0.6332, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 7.03, | |
| "learning_rate": 0.0001609400512052558, | |
| "loss": 0.6173, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 7.1, | |
| "learning_rate": 0.00015949084585285733, | |
| "loss": 0.5989, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 7.18, | |
| "learning_rate": 0.00015804164050045892, | |
| "loss": 0.5929, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 7.25, | |
| "learning_rate": 0.00015659243514806046, | |
| "loss": 0.6017, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 7.32, | |
| "learning_rate": 0.00015514322979566205, | |
| "loss": 0.5986, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 7.39, | |
| "learning_rate": 0.00015369402444326358, | |
| "loss": 0.5993, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 7.46, | |
| "learning_rate": 0.00015224481909086518, | |
| "loss": 0.5997, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 7.54, | |
| "learning_rate": 0.00015079561373846671, | |
| "loss": 0.5846, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 7.61, | |
| "learning_rate": 0.0001493464083860683, | |
| "loss": 0.6038, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 7.68, | |
| "learning_rate": 0.00014789720303366987, | |
| "loss": 0.5991, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 7.75, | |
| "learning_rate": 0.00014644799768127143, | |
| "loss": 0.5947, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 7.82, | |
| "learning_rate": 0.000144998792328873, | |
| "loss": 0.5889, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 7.89, | |
| "learning_rate": 0.00014354958697647456, | |
| "loss": 0.5855, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 7.97, | |
| "learning_rate": 0.0001421032800347809, | |
| "loss": 0.5902, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 8.04, | |
| "learning_rate": 0.00014065697309308727, | |
| "loss": 0.5805, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 8.11, | |
| "learning_rate": 0.00013920776774068884, | |
| "loss": 0.5627, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 8.18, | |
| "learning_rate": 0.0001377585623882904, | |
| "loss": 0.5566, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 8.25, | |
| "learning_rate": 0.00013630935703589197, | |
| "loss": 0.5597, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 8.32, | |
| "learning_rate": 0.00013486015168349353, | |
| "loss": 0.5553, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 8.4, | |
| "learning_rate": 0.0001334167431525047, | |
| "loss": 0.5587, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 8.47, | |
| "learning_rate": 0.00013196753780010627, | |
| "loss": 0.5561, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 8.54, | |
| "learning_rate": 0.00013051833244770783, | |
| "loss": 0.5624, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 8.61, | |
| "learning_rate": 0.0001290691270953094, | |
| "loss": 0.5525, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 8.68, | |
| "learning_rate": 0.00012761992174291096, | |
| "loss": 0.5619, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 8.76, | |
| "learning_rate": 0.00012617071639051252, | |
| "loss": 0.5552, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 8.83, | |
| "learning_rate": 0.0001247215110381141, | |
| "loss": 0.5523, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 8.9, | |
| "learning_rate": 0.00012327230568571565, | |
| "loss": 0.5527, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 8.97, | |
| "learning_rate": 0.00012182310033331722, | |
| "loss": 0.553, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 9.04, | |
| "learning_rate": 0.00012037679339162357, | |
| "loss": 0.5306, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 9.11, | |
| "learning_rate": 0.00011892758803922514, | |
| "loss": 0.5233, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 9.19, | |
| "learning_rate": 0.0001174783826868267, | |
| "loss": 0.5203, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 9.26, | |
| "learning_rate": 0.00011602917733442828, | |
| "loss": 0.531, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 9.33, | |
| "learning_rate": 0.00011458287039273465, | |
| "loss": 0.5203, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 9.4, | |
| "learning_rate": 0.00011313366504033621, | |
| "loss": 0.5253, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 9.47, | |
| "learning_rate": 0.00011168445968793778, | |
| "loss": 0.5311, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 9.54, | |
| "learning_rate": 0.00011023525433553934, | |
| "loss": 0.5246, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 9.62, | |
| "learning_rate": 0.0001087860489831409, | |
| "loss": 0.5228, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 9.69, | |
| "learning_rate": 0.00010733684363074247, | |
| "loss": 0.5264, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 9.76, | |
| "learning_rate": 0.00010589053668904882, | |
| "loss": 0.529, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 9.83, | |
| "learning_rate": 0.0001044413313366504, | |
| "loss": 0.5248, | |
| "step": 68500 | |
| }, | |
| { | |
| "epoch": 9.9, | |
| "learning_rate": 0.00010299502439495676, | |
| "loss": 0.523, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 9.98, | |
| "learning_rate": 0.00010154581904255832, | |
| "loss": 0.5269, | |
| "step": 69500 | |
| }, | |
| { | |
| "epoch": 10.05, | |
| "learning_rate": 0.00010009661369015989, | |
| "loss": 0.5044, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 10.12, | |
| "learning_rate": 9.865030674846624e-05, | |
| "loss": 0.4907, | |
| "step": 70500 | |
| }, | |
| { | |
| "epoch": 10.19, | |
| "learning_rate": 9.72011013960678e-05, | |
| "loss": 0.4982, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 10.26, | |
| "learning_rate": 9.575189604366937e-05, | |
| "loss": 0.4995, | |
| "step": 71500 | |
| }, | |
| { | |
| "epoch": 10.33, | |
| "learning_rate": 9.430269069127093e-05, | |
| "loss": 0.494, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 10.41, | |
| "learning_rate": 9.28534853388725e-05, | |
| "loss": 0.4928, | |
| "step": 72500 | |
| }, | |
| { | |
| "epoch": 10.48, | |
| "learning_rate": 9.140427998647408e-05, | |
| "loss": 0.4902, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 10.55, | |
| "learning_rate": 8.995507463407564e-05, | |
| "loss": 0.4932, | |
| "step": 73500 | |
| }, | |
| { | |
| "epoch": 10.62, | |
| "learning_rate": 8.85058692816772e-05, | |
| "loss": 0.4932, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 10.69, | |
| "learning_rate": 8.705666392927877e-05, | |
| "loss": 0.4905, | |
| "step": 74500 | |
| }, | |
| { | |
| "epoch": 10.77, | |
| "learning_rate": 8.560745857688034e-05, | |
| "loss": 0.4927, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 10.84, | |
| "learning_rate": 8.41582532244819e-05, | |
| "loss": 0.4782, | |
| "step": 75500 | |
| }, | |
| { | |
| "epoch": 10.91, | |
| "learning_rate": 8.270904787208346e-05, | |
| "loss": 0.483, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 10.98, | |
| "learning_rate": 8.126274093038982e-05, | |
| "loss": 0.4826, | |
| "step": 76500 | |
| }, | |
| { | |
| "epoch": 11.05, | |
| "learning_rate": 7.981353557799138e-05, | |
| "loss": 0.4733, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 11.12, | |
| "learning_rate": 7.836433022559295e-05, | |
| "loss": 0.4609, | |
| "step": 77500 | |
| }, | |
| { | |
| "epoch": 11.2, | |
| "learning_rate": 7.691512487319451e-05, | |
| "loss": 0.4652, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 11.27, | |
| "learning_rate": 7.54659195207961e-05, | |
| "loss": 0.4594, | |
| "step": 78500 | |
| }, | |
| { | |
| "epoch": 11.34, | |
| "learning_rate": 7.401671416839766e-05, | |
| "loss": 0.4688, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 11.41, | |
| "learning_rate": 7.257040722670402e-05, | |
| "loss": 0.4633, | |
| "step": 79500 | |
| }, | |
| { | |
| "epoch": 11.48, | |
| "learning_rate": 7.112120187430559e-05, | |
| "loss": 0.4654, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 11.55, | |
| "learning_rate": 6.967199652190715e-05, | |
| "loss": 0.4637, | |
| "step": 80500 | |
| }, | |
| { | |
| "epoch": 11.63, | |
| "learning_rate": 6.822568958021352e-05, | |
| "loss": 0.4599, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 11.7, | |
| "learning_rate": 6.677648422781507e-05, | |
| "loss": 0.4608, | |
| "step": 81500 | |
| }, | |
| { | |
| "epoch": 11.77, | |
| "learning_rate": 6.532727887541664e-05, | |
| "loss": 0.4536, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 11.84, | |
| "learning_rate": 6.38780735230182e-05, | |
| "loss": 0.4576, | |
| "step": 82500 | |
| }, | |
| { | |
| "epoch": 11.91, | |
| "learning_rate": 6.242886817061976e-05, | |
| "loss": 0.4555, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 11.99, | |
| "learning_rate": 6.0979662818221336e-05, | |
| "loss": 0.4545, | |
| "step": 83500 | |
| }, | |
| { | |
| "epoch": 12.06, | |
| "learning_rate": 5.95304574658229e-05, | |
| "loss": 0.4332, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 12.13, | |
| "learning_rate": 5.808125211342447e-05, | |
| "loss": 0.4359, | |
| "step": 84500 | |
| }, | |
| { | |
| "epoch": 12.2, | |
| "learning_rate": 5.663204676102604e-05, | |
| "loss": 0.4358, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 12.27, | |
| "learning_rate": 5.51828414086276e-05, | |
| "loss": 0.4365, | |
| "step": 85500 | |
| }, | |
| { | |
| "epoch": 12.34, | |
| "learning_rate": 5.373363605622916e-05, | |
| "loss": 0.4411, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 12.42, | |
| "learning_rate": 5.228732911453553e-05, | |
| "loss": 0.4354, | |
| "step": 86500 | |
| }, | |
| { | |
| "epoch": 12.49, | |
| "learning_rate": 5.083812376213709e-05, | |
| "loss": 0.4366, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 12.56, | |
| "learning_rate": 4.9388918409738656e-05, | |
| "loss": 0.4313, | |
| "step": 87500 | |
| }, | |
| { | |
| "epoch": 12.63, | |
| "learning_rate": 4.793971305734022e-05, | |
| "loss": 0.4342, | |
| "step": 88000 | |
| }, | |
| { | |
| "epoch": 12.7, | |
| "learning_rate": 4.6490507704941785e-05, | |
| "loss": 0.4328, | |
| "step": 88500 | |
| }, | |
| { | |
| "epoch": 12.77, | |
| "learning_rate": 4.504130235254335e-05, | |
| "loss": 0.4327, | |
| "step": 89000 | |
| }, | |
| { | |
| "epoch": 12.85, | |
| "learning_rate": 4.359499541084972e-05, | |
| "loss": 0.4329, | |
| "step": 89500 | |
| }, | |
| { | |
| "epoch": 12.92, | |
| "learning_rate": 4.2145790058451275e-05, | |
| "loss": 0.4343, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 12.99, | |
| "learning_rate": 4.069658470605284e-05, | |
| "loss": 0.4249, | |
| "step": 90500 | |
| }, | |
| { | |
| "epoch": 13.06, | |
| "learning_rate": 3.9247379353654405e-05, | |
| "loss": 0.4166, | |
| "step": 91000 | |
| }, | |
| { | |
| "epoch": 13.13, | |
| "learning_rate": 3.779817400125597e-05, | |
| "loss": 0.4145, | |
| "step": 91500 | |
| }, | |
| { | |
| "epoch": 13.21, | |
| "learning_rate": 3.635186705956234e-05, | |
| "loss": 0.4156, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 13.28, | |
| "learning_rate": 3.49026617071639e-05, | |
| "loss": 0.4132, | |
| "step": 92500 | |
| }, | |
| { | |
| "epoch": 13.35, | |
| "learning_rate": 3.345635476547026e-05, | |
| "loss": 0.4062, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 13.42, | |
| "learning_rate": 3.2007149413071834e-05, | |
| "loss": 0.4141, | |
| "step": 93500 | |
| }, | |
| { | |
| "epoch": 13.49, | |
| "learning_rate": 3.05579440606734e-05, | |
| "loss": 0.4101, | |
| "step": 94000 | |
| }, | |
| { | |
| "epoch": 13.56, | |
| "learning_rate": 2.910873870827496e-05, | |
| "loss": 0.4153, | |
| "step": 94500 | |
| }, | |
| { | |
| "epoch": 13.64, | |
| "learning_rate": 2.7659533355876528e-05, | |
| "loss": 0.4096, | |
| "step": 95000 | |
| }, | |
| { | |
| "epoch": 13.71, | |
| "learning_rate": 2.6210328003478092e-05, | |
| "loss": 0.4118, | |
| "step": 95500 | |
| }, | |
| { | |
| "epoch": 13.78, | |
| "learning_rate": 2.4761122651079657e-05, | |
| "loss": 0.4113, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 13.85, | |
| "learning_rate": 2.3314815709386018e-05, | |
| "loss": 0.4117, | |
| "step": 96500 | |
| }, | |
| { | |
| "epoch": 13.92, | |
| "learning_rate": 2.1865610356987582e-05, | |
| "loss": 0.408, | |
| "step": 97000 | |
| }, | |
| { | |
| "epoch": 13.99, | |
| "learning_rate": 2.041640500458915e-05, | |
| "loss": 0.4088, | |
| "step": 97500 | |
| } | |
| ], | |
| "logging_steps": 500, | |
| "max_steps": 104505, | |
| "num_train_epochs": 15, | |
| "save_steps": 500, | |
| "total_flos": 3.788962021131011e+20, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |