| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 5.0, | |
| "global_step": 184555, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.01, | |
| "learning_rate": 4.9865351792148686e-05, | |
| "loss": 8.7575, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "learning_rate": 4.9729890818455207e-05, | |
| "loss": 8.269, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 4.959442984476173e-05, | |
| "loss": 8.1258, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "learning_rate": 4.945923979301563e-05, | |
| "loss": 7.9945, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 4.932404974126954e-05, | |
| "loss": 7.8585, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "learning_rate": 4.9188588767576065e-05, | |
| "loss": 7.791, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 4.9053127793882585e-05, | |
| "loss": 7.6988, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 4.8917666820189106e-05, | |
| "loss": 7.6292, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "learning_rate": 4.8782205846495626e-05, | |
| "loss": 7.5104, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "learning_rate": 4.8646744872802146e-05, | |
| "loss": 7.4403, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 4.851155482105605e-05, | |
| "loss": 7.3613, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "learning_rate": 4.837609384736257e-05, | |
| "loss": 7.2621, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "learning_rate": 4.82406328736691e-05, | |
| "loss": 7.2195, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 4.810517189997562e-05, | |
| "loss": 7.1283, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "learning_rate": 4.796971092628214e-05, | |
| "loss": 7.0981, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 4.783424995258866e-05, | |
| "loss": 7.0347, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "learning_rate": 4.769878897889518e-05, | |
| "loss": 6.9573, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 4.75633280052017e-05, | |
| "loss": 6.9229, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 4.7427867031508226e-05, | |
| "loss": 6.8601, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "learning_rate": 4.7292406057814747e-05, | |
| "loss": 6.7912, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 4.715721600606865e-05, | |
| "loss": 6.7661, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 4.702175503237517e-05, | |
| "loss": 6.7346, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "learning_rate": 4.688629405868169e-05, | |
| "loss": 6.6528, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "learning_rate": 4.675083308498821e-05, | |
| "loss": 6.6245, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 4.661537211129473e-05, | |
| "loss": 6.6023, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "learning_rate": 4.6480182059548646e-05, | |
| "loss": 6.5243, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 4.634499200780256e-05, | |
| "loss": 6.533, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "learning_rate": 4.620953103410908e-05, | |
| "loss": 6.4781, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 4.60740700604156e-05, | |
| "loss": 6.4819, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 4.593860908672212e-05, | |
| "loss": 6.4321, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "learning_rate": 4.580314811302864e-05, | |
| "loss": 6.3922, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 4.566768713933516e-05, | |
| "loss": 6.368, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 4.553222616564168e-05, | |
| "loss": 6.3346, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "learning_rate": 4.5396765191948206e-05, | |
| "loss": 6.3076, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 4.52618460621495e-05, | |
| "loss": 6.2569, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 4.512638508845602e-05, | |
| "loss": 6.258, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "learning_rate": 4.499092411476254e-05, | |
| "loss": 6.245, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "learning_rate": 4.485546314106906e-05, | |
| "loss": 6.2077, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "learning_rate": 4.472000216737558e-05, | |
| "loss": 6.1827, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 4.45845411936821e-05, | |
| "loss": 6.143, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 4.4449080219988625e-05, | |
| "loss": 6.1349, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "learning_rate": 4.4313619246295145e-05, | |
| "loss": 6.1154, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 4.4178158272601666e-05, | |
| "loss": 6.0943, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 4.4042697298908186e-05, | |
| "loss": 6.062, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "learning_rate": 4.3907236325214706e-05, | |
| "loss": 6.0494, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 4.377204627346861e-05, | |
| "loss": 6.0567, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 4.363658529977513e-05, | |
| "loss": 6.0257, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "learning_rate": 4.350112432608166e-05, | |
| "loss": 5.9735, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "learning_rate": 4.336566335238818e-05, | |
| "loss": 5.9651, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "learning_rate": 4.32302023786947e-05, | |
| "loss": 5.9506, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 4.3095283248896e-05, | |
| "loss": 5.9435, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "learning_rate": 4.295982227520252e-05, | |
| "loss": 5.9315, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "learning_rate": 4.282436130150904e-05, | |
| "loss": 5.9151, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 4.268890032781556e-05, | |
| "loss": 5.8711, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 4.2553439354122085e-05, | |
| "loss": 5.8413, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "learning_rate": 4.2417978380428605e-05, | |
| "loss": 5.8496, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 4.2282517406735125e-05, | |
| "loss": 5.8372, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 4.214732735498903e-05, | |
| "loss": 5.8329, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "learning_rate": 4.201186638129555e-05, | |
| "loss": 5.7871, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "learning_rate": 4.187640540760207e-05, | |
| "loss": 5.803, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "learning_rate": 4.174094443390859e-05, | |
| "loss": 5.7591, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 4.160548346021512e-05, | |
| "loss": 5.7674, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "learning_rate": 4.147002248652164e-05, | |
| "loss": 5.7595, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "learning_rate": 4.133456151282816e-05, | |
| "loss": 5.7644, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 4.119910053913468e-05, | |
| "loss": 5.7161, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "learning_rate": 4.10636395654412e-05, | |
| "loss": 5.6858, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "learning_rate": 4.0928449513695105e-05, | |
| "loss": 5.7075, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 4.0792988540001625e-05, | |
| "loss": 5.6809, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "learning_rate": 4.065752756630815e-05, | |
| "loss": 5.6655, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "learning_rate": 4.052206659261467e-05, | |
| "loss": 5.6607, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "learning_rate": 4.038660561892119e-05, | |
| "loss": 5.6546, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 0.98, | |
| "learning_rate": 4.025114464522771e-05, | |
| "loss": 5.6226, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 4.011568367153423e-05, | |
| "loss": 5.6151, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "learning_rate": 3.998049361978814e-05, | |
| "loss": 5.5884, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 1.02, | |
| "learning_rate": 3.984503264609466e-05, | |
| "loss": 5.5605, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 3.9709571672401185e-05, | |
| "loss": 5.5594, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 1.04, | |
| "learning_rate": 3.9574110698707705e-05, | |
| "loss": 5.5597, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 1.06, | |
| "learning_rate": 3.9438649725014225e-05, | |
| "loss": 5.5272, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 3.9303188751320746e-05, | |
| "loss": 5.5298, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 1.08, | |
| "learning_rate": 3.9167727777627266e-05, | |
| "loss": 5.5227, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 3.9032266803933786e-05, | |
| "loss": 5.5018, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 1.11, | |
| "learning_rate": 3.8896805830240306e-05, | |
| "loss": 5.4736, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 3.876134485654683e-05, | |
| "loss": 5.4921, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 3.862669664869552e-05, | |
| "loss": 5.48, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 1.15, | |
| "learning_rate": 3.849123567500204e-05, | |
| "loss": 5.4851, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 3.835604562325594e-05, | |
| "loss": 5.4579, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 3.822058464956246e-05, | |
| "loss": 5.431, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 1.19, | |
| "learning_rate": 3.808512367586898e-05, | |
| "loss": 5.4482, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 1.21, | |
| "learning_rate": 3.79496627021755e-05, | |
| "loss": 5.4345, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 3.7814201728482023e-05, | |
| "loss": 5.433, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 1.23, | |
| "learning_rate": 3.767874075478855e-05, | |
| "loss": 5.4085, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 3.754327978109507e-05, | |
| "loss": 5.372, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 1.26, | |
| "learning_rate": 3.740781880740159e-05, | |
| "loss": 5.3914, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 3.727235783370811e-05, | |
| "loss": 5.3619, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 3.713689686001463e-05, | |
| "loss": 5.3477, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 1.3, | |
| "learning_rate": 3.700143588632115e-05, | |
| "loss": 5.3665, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 3.686597491262767e-05, | |
| "loss": 5.3525, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 3.67305139389342e-05, | |
| "loss": 5.3414, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 1.34, | |
| "learning_rate": 3.6595323887188104e-05, | |
| "loss": 5.3219, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 3.6459862913494624e-05, | |
| "loss": 5.348, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 3.632467286174853e-05, | |
| "loss": 5.3005, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 1.38, | |
| "learning_rate": 3.618921188805505e-05, | |
| "loss": 5.3209, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 3.605375091436157e-05, | |
| "loss": 5.3151, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 1.41, | |
| "learning_rate": 3.591828994066809e-05, | |
| "loss": 5.29, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 3.578282896697462e-05, | |
| "loss": 5.2857, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 3.564736799328114e-05, | |
| "loss": 5.2799, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 1.45, | |
| "learning_rate": 3.551217794153504e-05, | |
| "loss": 5.2552, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 3.5376716967841564e-05, | |
| "loss": 5.2445, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 3.5241255994148084e-05, | |
| "loss": 5.2395, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 1.49, | |
| "learning_rate": 3.5105795020454604e-05, | |
| "loss": 5.2328, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 3.4970604968708516e-05, | |
| "loss": 5.2544, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 3.483514399501504e-05, | |
| "loss": 5.2206, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 1.53, | |
| "learning_rate": 3.4699683021321563e-05, | |
| "loss": 5.2362, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 1.54, | |
| "learning_rate": 3.4564222047628084e-05, | |
| "loss": 5.2165, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 1.56, | |
| "learning_rate": 3.4428761073934604e-05, | |
| "loss": 5.194, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 3.4293300100241124e-05, | |
| "loss": 5.1808, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 1.58, | |
| "learning_rate": 3.4157839126547644e-05, | |
| "loss": 5.1896, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 1.6, | |
| "learning_rate": 3.402264907480155e-05, | |
| "loss": 5.2052, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 3.388718810110808e-05, | |
| "loss": 5.1603, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 3.37517271274146e-05, | |
| "loss": 5.1743, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 1.64, | |
| "learning_rate": 3.361626615372112e-05, | |
| "loss": 5.1778, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 3.348107610197502e-05, | |
| "loss": 5.1594, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 3.334561512828154e-05, | |
| "loss": 5.1613, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 1.68, | |
| "learning_rate": 3.321015415458806e-05, | |
| "loss": 5.1426, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 1.69, | |
| "learning_rate": 3.3074693180894583e-05, | |
| "loss": 5.1318, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 1.71, | |
| "learning_rate": 3.293923220720111e-05, | |
| "loss": 5.1315, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 3.280377123350763e-05, | |
| "loss": 5.1152, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 1.73, | |
| "learning_rate": 3.266831025981415e-05, | |
| "loss": 5.1054, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 1.75, | |
| "learning_rate": 3.2533120208068056e-05, | |
| "loss": 5.1067, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 3.2397659234374577e-05, | |
| "loss": 5.097, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 1.77, | |
| "learning_rate": 3.22621982606811e-05, | |
| "loss": 5.0843, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 1.79, | |
| "learning_rate": 3.212673728698762e-05, | |
| "loss": 5.0832, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 3.1991276313294144e-05, | |
| "loss": 5.088, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 1.82, | |
| "learning_rate": 3.185608626154805e-05, | |
| "loss": 5.0687, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 3.172062528785457e-05, | |
| "loss": 5.0503, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 1.84, | |
| "learning_rate": 3.158516431416109e-05, | |
| "loss": 5.0483, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 1.86, | |
| "learning_rate": 3.144970334046761e-05, | |
| "loss": 5.0509, | |
| "step": 68500 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 3.131424236677413e-05, | |
| "loss": 5.0699, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 1.88, | |
| "learning_rate": 3.117905231502804e-05, | |
| "loss": 5.0364, | |
| "step": 69500 | |
| }, | |
| { | |
| "epoch": 1.9, | |
| "learning_rate": 3.104359134133456e-05, | |
| "loss": 5.0406, | |
| "step": 70000 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 3.090813036764108e-05, | |
| "loss": 5.036, | |
| "step": 70500 | |
| }, | |
| { | |
| "epoch": 1.92, | |
| "learning_rate": 3.07726693939476e-05, | |
| "loss": 5.0205, | |
| "step": 71000 | |
| }, | |
| { | |
| "epoch": 1.94, | |
| "learning_rate": 3.0637479342201516e-05, | |
| "loss": 5.0466, | |
| "step": 71500 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 3.0502018368508033e-05, | |
| "loss": 4.9909, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 1.96, | |
| "learning_rate": 3.0366557394814553e-05, | |
| "loss": 5.0072, | |
| "step": 72500 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 3.0231096421121073e-05, | |
| "loss": 4.999, | |
| "step": 73000 | |
| }, | |
| { | |
| "epoch": 1.99, | |
| "learning_rate": 3.0095906369374982e-05, | |
| "loss": 5.0315, | |
| "step": 73500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 2.996044539568151e-05, | |
| "loss": 5.0116, | |
| "step": 74000 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 2.982498442198803e-05, | |
| "loss": 4.9781, | |
| "step": 74500 | |
| }, | |
| { | |
| "epoch": 2.03, | |
| "learning_rate": 2.968952344829455e-05, | |
| "loss": 4.9791, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 2.05, | |
| "learning_rate": 2.955406247460107e-05, | |
| "loss": 4.9612, | |
| "step": 75500 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 2.941860150090759e-05, | |
| "loss": 4.9777, | |
| "step": 76000 | |
| }, | |
| { | |
| "epoch": 2.07, | |
| "learning_rate": 2.928314052721411e-05, | |
| "loss": 4.9613, | |
| "step": 76500 | |
| }, | |
| { | |
| "epoch": 2.09, | |
| "learning_rate": 2.9147679553520633e-05, | |
| "loss": 4.9391, | |
| "step": 77000 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 2.9012218579827153e-05, | |
| "loss": 4.9471, | |
| "step": 77500 | |
| }, | |
| { | |
| "epoch": 2.11, | |
| "learning_rate": 2.8877028528081063e-05, | |
| "loss": 4.9495, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 2.8741567554387583e-05, | |
| "loss": 4.9406, | |
| "step": 78500 | |
| }, | |
| { | |
| "epoch": 2.14, | |
| "learning_rate": 2.8606106580694103e-05, | |
| "loss": 4.9162, | |
| "step": 79000 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 2.8470645607000623e-05, | |
| "loss": 4.942, | |
| "step": 79500 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 2.8335184633307143e-05, | |
| "loss": 4.9158, | |
| "step": 80000 | |
| }, | |
| { | |
| "epoch": 2.18, | |
| "learning_rate": 2.8199994581561053e-05, | |
| "loss": 4.9133, | |
| "step": 80500 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 2.8064533607867576e-05, | |
| "loss": 4.9137, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 2.7929072634174096e-05, | |
| "loss": 4.8945, | |
| "step": 81500 | |
| }, | |
| { | |
| "epoch": 2.22, | |
| "learning_rate": 2.7793611660480616e-05, | |
| "loss": 4.9128, | |
| "step": 82000 | |
| }, | |
| { | |
| "epoch": 2.24, | |
| "learning_rate": 2.7658150686787136e-05, | |
| "loss": 4.922, | |
| "step": 82500 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 2.7522689713093657e-05, | |
| "loss": 4.8827, | |
| "step": 83000 | |
| }, | |
| { | |
| "epoch": 2.26, | |
| "learning_rate": 2.7387228739400177e-05, | |
| "loss": 4.889, | |
| "step": 83500 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 2.7252038687654086e-05, | |
| "loss": 4.8505, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 2.29, | |
| "learning_rate": 2.7116577713960613e-05, | |
| "loss": 4.8842, | |
| "step": 84500 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 2.6981116740267133e-05, | |
| "loss": 4.851, | |
| "step": 85000 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 2.6845655766573653e-05, | |
| "loss": 4.858, | |
| "step": 85500 | |
| }, | |
| { | |
| "epoch": 2.33, | |
| "learning_rate": 2.6710194792880173e-05, | |
| "loss": 4.8586, | |
| "step": 86000 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 2.6574733819186694e-05, | |
| "loss": 4.8802, | |
| "step": 86500 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 2.6439272845493214e-05, | |
| "loss": 4.8456, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 2.37, | |
| "learning_rate": 2.6303811871799737e-05, | |
| "loss": 4.8277, | |
| "step": 87500 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 2.6168350898106257e-05, | |
| "loss": 4.8182, | |
| "step": 88000 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 2.6033160846360167e-05, | |
| "loss": 4.8282, | |
| "step": 88500 | |
| }, | |
| { | |
| "epoch": 2.41, | |
| "learning_rate": 2.5897699872666687e-05, | |
| "loss": 4.8361, | |
| "step": 89000 | |
| }, | |
| { | |
| "epoch": 2.42, | |
| "learning_rate": 2.5762238898973207e-05, | |
| "loss": 4.8205, | |
| "step": 89500 | |
| }, | |
| { | |
| "epoch": 2.44, | |
| "learning_rate": 2.5626777925279727e-05, | |
| "loss": 4.8186, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 2.5491316951586247e-05, | |
| "loss": 4.7993, | |
| "step": 90500 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 2.5355855977892774e-05, | |
| "loss": 4.8049, | |
| "step": 91000 | |
| }, | |
| { | |
| "epoch": 2.48, | |
| "learning_rate": 2.5220395004199294e-05, | |
| "loss": 4.8183, | |
| "step": 91500 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 2.5084934030505814e-05, | |
| "loss": 4.7909, | |
| "step": 92000 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 2.4949473056812334e-05, | |
| "loss": 4.7821, | |
| "step": 92500 | |
| }, | |
| { | |
| "epoch": 2.52, | |
| "learning_rate": 2.4814012083118855e-05, | |
| "loss": 4.8002, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 2.467882203137276e-05, | |
| "loss": 4.8025, | |
| "step": 93500 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 2.4543361057679284e-05, | |
| "loss": 4.7971, | |
| "step": 94000 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 2.4407900083985804e-05, | |
| "loss": 4.8041, | |
| "step": 94500 | |
| }, | |
| { | |
| "epoch": 2.57, | |
| "learning_rate": 2.4272439110292324e-05, | |
| "loss": 4.7919, | |
| "step": 95000 | |
| }, | |
| { | |
| "epoch": 2.59, | |
| "learning_rate": 2.4136978136598848e-05, | |
| "loss": 4.7786, | |
| "step": 95500 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 2.4001788084852757e-05, | |
| "loss": 4.7784, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 2.61, | |
| "learning_rate": 2.3866327111159277e-05, | |
| "loss": 4.7562, | |
| "step": 96500 | |
| }, | |
| { | |
| "epoch": 2.63, | |
| "learning_rate": 2.3730866137465797e-05, | |
| "loss": 4.7655, | |
| "step": 97000 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 2.3595676085719707e-05, | |
| "loss": 4.7558, | |
| "step": 97500 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 2.3460215112026227e-05, | |
| "loss": 4.7599, | |
| "step": 98000 | |
| }, | |
| { | |
| "epoch": 2.67, | |
| "learning_rate": 2.3324754138332747e-05, | |
| "loss": 4.7534, | |
| "step": 98500 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 2.318929316463927e-05, | |
| "loss": 4.7538, | |
| "step": 99000 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 2.305383219094579e-05, | |
| "loss": 4.7441, | |
| "step": 99500 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 2.291837121725231e-05, | |
| "loss": 4.7245, | |
| "step": 100000 | |
| }, | |
| { | |
| "epoch": 2.72, | |
| "learning_rate": 2.278291024355883e-05, | |
| "loss": 4.7254, | |
| "step": 100500 | |
| }, | |
| { | |
| "epoch": 2.74, | |
| "learning_rate": 2.2647449269865354e-05, | |
| "loss": 4.758, | |
| "step": 101000 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 2.2511988296171875e-05, | |
| "loss": 4.7233, | |
| "step": 101500 | |
| }, | |
| { | |
| "epoch": 2.76, | |
| "learning_rate": 2.237679824442578e-05, | |
| "loss": 4.7253, | |
| "step": 102000 | |
| }, | |
| { | |
| "epoch": 2.78, | |
| "learning_rate": 2.2241337270732304e-05, | |
| "loss": 4.7053, | |
| "step": 102500 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 2.2105876297038824e-05, | |
| "loss": 4.744, | |
| "step": 103000 | |
| }, | |
| { | |
| "epoch": 2.8, | |
| "learning_rate": 2.1970686245292733e-05, | |
| "loss": 4.7202, | |
| "step": 103500 | |
| }, | |
| { | |
| "epoch": 2.82, | |
| "learning_rate": 2.1835225271599253e-05, | |
| "loss": 4.7183, | |
| "step": 104000 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 2.1699764297905777e-05, | |
| "loss": 4.6929, | |
| "step": 104500 | |
| }, | |
| { | |
| "epoch": 2.84, | |
| "learning_rate": 2.1564303324212297e-05, | |
| "loss": 4.6945, | |
| "step": 105000 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 2.1428842350518817e-05, | |
| "loss": 4.6947, | |
| "step": 105500 | |
| }, | |
| { | |
| "epoch": 2.87, | |
| "learning_rate": 2.1293381376825337e-05, | |
| "loss": 4.7301, | |
| "step": 106000 | |
| }, | |
| { | |
| "epoch": 2.89, | |
| "learning_rate": 2.1157920403131858e-05, | |
| "loss": 4.6732, | |
| "step": 106500 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 2.1022459429438378e-05, | |
| "loss": 4.6876, | |
| "step": 107000 | |
| }, | |
| { | |
| "epoch": 2.91, | |
| "learning_rate": 2.08869984557449e-05, | |
| "loss": 4.6905, | |
| "step": 107500 | |
| }, | |
| { | |
| "epoch": 2.93, | |
| "learning_rate": 2.075180840399881e-05, | |
| "loss": 4.6483, | |
| "step": 108000 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 2.061634743030533e-05, | |
| "loss": 4.6874, | |
| "step": 108500 | |
| }, | |
| { | |
| "epoch": 2.95, | |
| "learning_rate": 2.048088645661185e-05, | |
| "loss": 4.6589, | |
| "step": 109000 | |
| }, | |
| { | |
| "epoch": 2.97, | |
| "learning_rate": 2.0345425482918374e-05, | |
| "loss": 4.6826, | |
| "step": 109500 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 2.021023543117228e-05, | |
| "loss": 4.6928, | |
| "step": 110000 | |
| }, | |
| { | |
| "epoch": 2.99, | |
| "learning_rate": 2.007504537942619e-05, | |
| "loss": 4.6559, | |
| "step": 110500 | |
| }, | |
| { | |
| "epoch": 3.01, | |
| "learning_rate": 1.993958440573271e-05, | |
| "loss": 4.6833, | |
| "step": 111000 | |
| }, | |
| { | |
| "epoch": 3.02, | |
| "learning_rate": 1.9804123432039233e-05, | |
| "loss": 4.6348, | |
| "step": 111500 | |
| }, | |
| { | |
| "epoch": 3.03, | |
| "learning_rate": 1.9668662458345753e-05, | |
| "loss": 4.6374, | |
| "step": 112000 | |
| }, | |
| { | |
| "epoch": 3.05, | |
| "learning_rate": 1.9533201484652273e-05, | |
| "loss": 4.6329, | |
| "step": 112500 | |
| }, | |
| { | |
| "epoch": 3.06, | |
| "learning_rate": 1.9397740510958793e-05, | |
| "loss": 4.6247, | |
| "step": 113000 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 1.9262279537265317e-05, | |
| "loss": 4.6435, | |
| "step": 113500 | |
| }, | |
| { | |
| "epoch": 3.09, | |
| "learning_rate": 1.9126818563571837e-05, | |
| "loss": 4.6287, | |
| "step": 114000 | |
| }, | |
| { | |
| "epoch": 3.1, | |
| "learning_rate": 1.8991357589878357e-05, | |
| "loss": 4.6161, | |
| "step": 114500 | |
| }, | |
| { | |
| "epoch": 3.12, | |
| "learning_rate": 1.8855896616184877e-05, | |
| "loss": 4.6376, | |
| "step": 115000 | |
| }, | |
| { | |
| "epoch": 3.13, | |
| "learning_rate": 1.8720706564438787e-05, | |
| "loss": 4.6203, | |
| "step": 115500 | |
| }, | |
| { | |
| "epoch": 3.14, | |
| "learning_rate": 1.8585245590745307e-05, | |
| "loss": 4.61, | |
| "step": 116000 | |
| }, | |
| { | |
| "epoch": 3.16, | |
| "learning_rate": 1.844978461705183e-05, | |
| "loss": 4.6313, | |
| "step": 116500 | |
| }, | |
| { | |
| "epoch": 3.17, | |
| "learning_rate": 1.8314594565305736e-05, | |
| "loss": 4.6221, | |
| "step": 117000 | |
| }, | |
| { | |
| "epoch": 3.18, | |
| "learning_rate": 1.8179133591612256e-05, | |
| "loss": 4.5978, | |
| "step": 117500 | |
| }, | |
| { | |
| "epoch": 3.2, | |
| "learning_rate": 1.8043672617918776e-05, | |
| "loss": 4.6247, | |
| "step": 118000 | |
| }, | |
| { | |
| "epoch": 3.21, | |
| "learning_rate": 1.79082116442253e-05, | |
| "loss": 4.6023, | |
| "step": 118500 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 1.777275067053182e-05, | |
| "loss": 4.613, | |
| "step": 119000 | |
| }, | |
| { | |
| "epoch": 3.24, | |
| "learning_rate": 1.763728969683834e-05, | |
| "loss": 4.585, | |
| "step": 119500 | |
| }, | |
| { | |
| "epoch": 3.25, | |
| "learning_rate": 1.7501828723144864e-05, | |
| "loss": 4.5845, | |
| "step": 120000 | |
| }, | |
| { | |
| "epoch": 3.26, | |
| "learning_rate": 1.7366367749451384e-05, | |
| "loss": 4.6044, | |
| "step": 120500 | |
| }, | |
| { | |
| "epoch": 3.28, | |
| "learning_rate": 1.7230906775757904e-05, | |
| "loss": 4.5841, | |
| "step": 121000 | |
| }, | |
| { | |
| "epoch": 3.29, | |
| "learning_rate": 1.7095445802064424e-05, | |
| "loss": 4.5852, | |
| "step": 121500 | |
| }, | |
| { | |
| "epoch": 3.31, | |
| "learning_rate": 1.6959984828370948e-05, | |
| "loss": 4.5812, | |
| "step": 122000 | |
| }, | |
| { | |
| "epoch": 3.32, | |
| "learning_rate": 1.6824523854677468e-05, | |
| "loss": 4.5725, | |
| "step": 122500 | |
| }, | |
| { | |
| "epoch": 3.33, | |
| "learning_rate": 1.6689062880983988e-05, | |
| "loss": 4.6056, | |
| "step": 123000 | |
| }, | |
| { | |
| "epoch": 3.35, | |
| "learning_rate": 1.655360190729051e-05, | |
| "loss": 4.5775, | |
| "step": 123500 | |
| }, | |
| { | |
| "epoch": 3.36, | |
| "learning_rate": 1.6418140933597032e-05, | |
| "loss": 4.5549, | |
| "step": 124000 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 1.6282679959903552e-05, | |
| "loss": 4.563, | |
| "step": 124500 | |
| }, | |
| { | |
| "epoch": 3.39, | |
| "learning_rate": 1.6147218986210075e-05, | |
| "loss": 4.5744, | |
| "step": 125000 | |
| }, | |
| { | |
| "epoch": 3.4, | |
| "learning_rate": 1.601202893446398e-05, | |
| "loss": 4.5789, | |
| "step": 125500 | |
| }, | |
| { | |
| "epoch": 3.41, | |
| "learning_rate": 1.58765679607705e-05, | |
| "loss": 4.5893, | |
| "step": 126000 | |
| }, | |
| { | |
| "epoch": 3.43, | |
| "learning_rate": 1.574110698707702e-05, | |
| "loss": 4.5927, | |
| "step": 126500 | |
| }, | |
| { | |
| "epoch": 3.44, | |
| "learning_rate": 1.5605646013383545e-05, | |
| "loss": 4.5773, | |
| "step": 127000 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 1.5470455961637454e-05, | |
| "loss": 4.5735, | |
| "step": 127500 | |
| }, | |
| { | |
| "epoch": 3.47, | |
| "learning_rate": 1.533526590989136e-05, | |
| "loss": 4.5505, | |
| "step": 128000 | |
| }, | |
| { | |
| "epoch": 3.48, | |
| "learning_rate": 1.5199804936197882e-05, | |
| "loss": 4.5526, | |
| "step": 128500 | |
| }, | |
| { | |
| "epoch": 3.49, | |
| "learning_rate": 1.5064343962504404e-05, | |
| "loss": 4.5524, | |
| "step": 129000 | |
| }, | |
| { | |
| "epoch": 3.51, | |
| "learning_rate": 1.4928882988810924e-05, | |
| "loss": 4.5399, | |
| "step": 129500 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 1.4793422015117444e-05, | |
| "loss": 4.5601, | |
| "step": 130000 | |
| }, | |
| { | |
| "epoch": 3.54, | |
| "learning_rate": 1.4658231963371352e-05, | |
| "loss": 4.5481, | |
| "step": 130500 | |
| }, | |
| { | |
| "epoch": 3.55, | |
| "learning_rate": 1.4522770989677875e-05, | |
| "loss": 4.5509, | |
| "step": 131000 | |
| }, | |
| { | |
| "epoch": 3.56, | |
| "learning_rate": 1.4387310015984395e-05, | |
| "loss": 4.5436, | |
| "step": 131500 | |
| }, | |
| { | |
| "epoch": 3.58, | |
| "learning_rate": 1.4251849042290915e-05, | |
| "loss": 4.5246, | |
| "step": 132000 | |
| }, | |
| { | |
| "epoch": 3.59, | |
| "learning_rate": 1.4116658990544826e-05, | |
| "loss": 4.5459, | |
| "step": 132500 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 1.3981198016851347e-05, | |
| "loss": 4.5395, | |
| "step": 133000 | |
| }, | |
| { | |
| "epoch": 3.62, | |
| "learning_rate": 1.3846278887052641e-05, | |
| "loss": 4.5183, | |
| "step": 133500 | |
| }, | |
| { | |
| "epoch": 3.63, | |
| "learning_rate": 1.3710817913359162e-05, | |
| "loss": 4.545, | |
| "step": 134000 | |
| }, | |
| { | |
| "epoch": 3.64, | |
| "learning_rate": 1.3575356939665685e-05, | |
| "loss": 4.5206, | |
| "step": 134500 | |
| }, | |
| { | |
| "epoch": 3.66, | |
| "learning_rate": 1.3439895965972205e-05, | |
| "loss": 4.5522, | |
| "step": 135000 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 1.3304434992278725e-05, | |
| "loss": 4.5318, | |
| "step": 135500 | |
| }, | |
| { | |
| "epoch": 3.68, | |
| "learning_rate": 1.3168974018585246e-05, | |
| "loss": 4.5134, | |
| "step": 136000 | |
| }, | |
| { | |
| "epoch": 3.7, | |
| "learning_rate": 1.3033513044891767e-05, | |
| "loss": 4.4964, | |
| "step": 136500 | |
| }, | |
| { | |
| "epoch": 3.71, | |
| "learning_rate": 1.2898052071198288e-05, | |
| "loss": 4.5127, | |
| "step": 137000 | |
| }, | |
| { | |
| "epoch": 3.73, | |
| "learning_rate": 1.2762591097504808e-05, | |
| "loss": 4.4998, | |
| "step": 137500 | |
| }, | |
| { | |
| "epoch": 3.74, | |
| "learning_rate": 1.2627401045758719e-05, | |
| "loss": 4.5086, | |
| "step": 138000 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 1.2491940072065239e-05, | |
| "loss": 4.4935, | |
| "step": 138500 | |
| }, | |
| { | |
| "epoch": 3.77, | |
| "learning_rate": 1.2356479098371759e-05, | |
| "loss": 4.5101, | |
| "step": 139000 | |
| }, | |
| { | |
| "epoch": 3.78, | |
| "learning_rate": 1.222101812467828e-05, | |
| "loss": 4.4978, | |
| "step": 139500 | |
| }, | |
| { | |
| "epoch": 3.79, | |
| "learning_rate": 1.2085557150984801e-05, | |
| "loss": 4.4964, | |
| "step": 140000 | |
| }, | |
| { | |
| "epoch": 3.81, | |
| "learning_rate": 1.1950096177291323e-05, | |
| "loss": 4.4983, | |
| "step": 140500 | |
| }, | |
| { | |
| "epoch": 3.82, | |
| "learning_rate": 1.1814635203597845e-05, | |
| "loss": 4.5057, | |
| "step": 141000 | |
| }, | |
| { | |
| "epoch": 3.83, | |
| "learning_rate": 1.1679174229904365e-05, | |
| "loss": 4.5075, | |
| "step": 141500 | |
| }, | |
| { | |
| "epoch": 3.85, | |
| "learning_rate": 1.1543713256210887e-05, | |
| "loss": 4.491, | |
| "step": 142000 | |
| }, | |
| { | |
| "epoch": 3.86, | |
| "learning_rate": 1.1408252282517408e-05, | |
| "loss": 4.4984, | |
| "step": 142500 | |
| }, | |
| { | |
| "epoch": 3.87, | |
| "learning_rate": 1.1272791308823929e-05, | |
| "loss": 4.4833, | |
| "step": 143000 | |
| }, | |
| { | |
| "epoch": 3.89, | |
| "learning_rate": 1.113733033513045e-05, | |
| "loss": 4.4912, | |
| "step": 143500 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 1.100186936143697e-05, | |
| "loss": 4.4761, | |
| "step": 144000 | |
| }, | |
| { | |
| "epoch": 3.91, | |
| "learning_rate": 1.086667930969088e-05, | |
| "loss": 4.5069, | |
| "step": 144500 | |
| }, | |
| { | |
| "epoch": 3.93, | |
| "learning_rate": 1.07312183359974e-05, | |
| "loss": 4.4795, | |
| "step": 145000 | |
| }, | |
| { | |
| "epoch": 3.94, | |
| "learning_rate": 1.059575736230392e-05, | |
| "loss": 4.4809, | |
| "step": 145500 | |
| }, | |
| { | |
| "epoch": 3.96, | |
| "learning_rate": 1.0460567310557828e-05, | |
| "loss": 4.4631, | |
| "step": 146000 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 1.0325377258811737e-05, | |
| "loss": 4.4903, | |
| "step": 146500 | |
| }, | |
| { | |
| "epoch": 3.98, | |
| "learning_rate": 1.0189916285118257e-05, | |
| "loss": 4.4633, | |
| "step": 147000 | |
| }, | |
| { | |
| "epoch": 4.0, | |
| "learning_rate": 1.0054455311424779e-05, | |
| "loss": 4.4753, | |
| "step": 147500 | |
| }, | |
| { | |
| "epoch": 4.01, | |
| "learning_rate": 9.9189943377313e-06, | |
| "loss": 4.4397, | |
| "step": 148000 | |
| }, | |
| { | |
| "epoch": 4.02, | |
| "learning_rate": 9.78353336403782e-06, | |
| "loss": 4.4516, | |
| "step": 148500 | |
| }, | |
| { | |
| "epoch": 4.04, | |
| "learning_rate": 9.648343312291728e-06, | |
| "loss": 4.4489, | |
| "step": 149000 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 9.51288233859825e-06, | |
| "loss": 4.4331, | |
| "step": 149500 | |
| }, | |
| { | |
| "epoch": 4.06, | |
| "learning_rate": 9.377692286852158e-06, | |
| "loss": 4.4521, | |
| "step": 150000 | |
| }, | |
| { | |
| "epoch": 4.08, | |
| "learning_rate": 9.24223131315868e-06, | |
| "loss": 4.4598, | |
| "step": 150500 | |
| }, | |
| { | |
| "epoch": 4.09, | |
| "learning_rate": 9.106770339465201e-06, | |
| "loss": 4.4427, | |
| "step": 151000 | |
| }, | |
| { | |
| "epoch": 4.1, | |
| "learning_rate": 8.971309365771722e-06, | |
| "loss": 4.4481, | |
| "step": 151500 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 8.835848392078243e-06, | |
| "loss": 4.4459, | |
| "step": 152000 | |
| }, | |
| { | |
| "epoch": 4.13, | |
| "learning_rate": 8.700387418384765e-06, | |
| "loss": 4.4573, | |
| "step": 152500 | |
| }, | |
| { | |
| "epoch": 4.15, | |
| "learning_rate": 8.564926444691285e-06, | |
| "loss": 4.4291, | |
| "step": 153000 | |
| }, | |
| { | |
| "epoch": 4.16, | |
| "learning_rate": 8.429465470997805e-06, | |
| "loss": 4.4504, | |
| "step": 153500 | |
| }, | |
| { | |
| "epoch": 4.17, | |
| "learning_rate": 8.294004497304327e-06, | |
| "loss": 4.4393, | |
| "step": 154000 | |
| }, | |
| { | |
| "epoch": 4.19, | |
| "learning_rate": 8.158543523610847e-06, | |
| "loss": 4.4256, | |
| "step": 154500 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "learning_rate": 8.02308254991737e-06, | |
| "loss": 4.4317, | |
| "step": 155000 | |
| }, | |
| { | |
| "epoch": 4.21, | |
| "learning_rate": 7.88762157622389e-06, | |
| "loss": 4.4313, | |
| "step": 155500 | |
| }, | |
| { | |
| "epoch": 4.23, | |
| "learning_rate": 7.752160602530411e-06, | |
| "loss": 4.4242, | |
| "step": 156000 | |
| }, | |
| { | |
| "epoch": 4.24, | |
| "learning_rate": 7.616699628836932e-06, | |
| "loss": 4.4395, | |
| "step": 156500 | |
| }, | |
| { | |
| "epoch": 4.25, | |
| "learning_rate": 7.481509577090841e-06, | |
| "loss": 4.4611, | |
| "step": 157000 | |
| }, | |
| { | |
| "epoch": 4.27, | |
| "learning_rate": 7.346048603397361e-06, | |
| "loss": 4.4614, | |
| "step": 157500 | |
| }, | |
| { | |
| "epoch": 4.28, | |
| "learning_rate": 7.21085855165127e-06, | |
| "loss": 4.4374, | |
| "step": 158000 | |
| }, | |
| { | |
| "epoch": 4.29, | |
| "learning_rate": 7.07539757795779e-06, | |
| "loss": 4.4461, | |
| "step": 158500 | |
| }, | |
| { | |
| "epoch": 4.31, | |
| "learning_rate": 6.939936604264312e-06, | |
| "loss": 4.423, | |
| "step": 159000 | |
| }, | |
| { | |
| "epoch": 4.32, | |
| "learning_rate": 6.804475630570834e-06, | |
| "loss": 4.4385, | |
| "step": 159500 | |
| }, | |
| { | |
| "epoch": 4.33, | |
| "learning_rate": 6.669014656877354e-06, | |
| "loss": 4.4149, | |
| "step": 160000 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "learning_rate": 6.533553683183875e-06, | |
| "loss": 4.4289, | |
| "step": 160500 | |
| }, | |
| { | |
| "epoch": 4.36, | |
| "learning_rate": 6.398092709490396e-06, | |
| "loss": 4.4138, | |
| "step": 161000 | |
| }, | |
| { | |
| "epoch": 4.38, | |
| "learning_rate": 6.262631735796917e-06, | |
| "loss": 4.4308, | |
| "step": 161500 | |
| }, | |
| { | |
| "epoch": 4.39, | |
| "learning_rate": 6.127170762103438e-06, | |
| "loss": 4.4329, | |
| "step": 162000 | |
| }, | |
| { | |
| "epoch": 4.4, | |
| "learning_rate": 5.991980710357346e-06, | |
| "loss": 4.4126, | |
| "step": 162500 | |
| }, | |
| { | |
| "epoch": 4.42, | |
| "learning_rate": 5.856519736663867e-06, | |
| "loss": 4.4112, | |
| "step": 163000 | |
| }, | |
| { | |
| "epoch": 4.43, | |
| "learning_rate": 5.721058762970388e-06, | |
| "loss": 4.4055, | |
| "step": 163500 | |
| }, | |
| { | |
| "epoch": 4.44, | |
| "learning_rate": 5.58559778927691e-06, | |
| "loss": 4.4062, | |
| "step": 164000 | |
| }, | |
| { | |
| "epoch": 4.46, | |
| "learning_rate": 5.450407737530818e-06, | |
| "loss": 4.41, | |
| "step": 164500 | |
| }, | |
| { | |
| "epoch": 4.47, | |
| "learning_rate": 5.314946763837339e-06, | |
| "loss": 4.4198, | |
| "step": 165000 | |
| }, | |
| { | |
| "epoch": 4.48, | |
| "learning_rate": 5.17948579014386e-06, | |
| "loss": 4.3997, | |
| "step": 165500 | |
| }, | |
| { | |
| "epoch": 4.5, | |
| "learning_rate": 5.044024816450381e-06, | |
| "loss": 4.4254, | |
| "step": 166000 | |
| }, | |
| { | |
| "epoch": 4.51, | |
| "learning_rate": 4.908563842756902e-06, | |
| "loss": 4.4098, | |
| "step": 166500 | |
| }, | |
| { | |
| "epoch": 4.52, | |
| "learning_rate": 4.773102869063423e-06, | |
| "loss": 4.4132, | |
| "step": 167000 | |
| }, | |
| { | |
| "epoch": 4.54, | |
| "learning_rate": 4.6376418953699445e-06, | |
| "loss": 4.4106, | |
| "step": 167500 | |
| }, | |
| { | |
| "epoch": 4.55, | |
| "learning_rate": 4.502451843623852e-06, | |
| "loss": 4.4227, | |
| "step": 168000 | |
| }, | |
| { | |
| "epoch": 4.57, | |
| "learning_rate": 4.366990869930374e-06, | |
| "loss": 4.3846, | |
| "step": 168500 | |
| }, | |
| { | |
| "epoch": 4.58, | |
| "learning_rate": 4.231529896236895e-06, | |
| "loss": 4.4244, | |
| "step": 169000 | |
| }, | |
| { | |
| "epoch": 4.59, | |
| "learning_rate": 4.096068922543415e-06, | |
| "loss": 4.3934, | |
| "step": 169500 | |
| }, | |
| { | |
| "epoch": 4.61, | |
| "learning_rate": 3.960607948849936e-06, | |
| "loss": 4.4047, | |
| "step": 170000 | |
| }, | |
| { | |
| "epoch": 4.62, | |
| "learning_rate": 3.825146975156458e-06, | |
| "loss": 4.3781, | |
| "step": 170500 | |
| }, | |
| { | |
| "epoch": 4.63, | |
| "learning_rate": 3.689686001462979e-06, | |
| "loss": 4.4025, | |
| "step": 171000 | |
| }, | |
| { | |
| "epoch": 4.65, | |
| "learning_rate": 3.5542250277695e-06, | |
| "loss": 4.395, | |
| "step": 171500 | |
| }, | |
| { | |
| "epoch": 4.66, | |
| "learning_rate": 3.4190349760234082e-06, | |
| "loss": 4.3907, | |
| "step": 172000 | |
| }, | |
| { | |
| "epoch": 4.67, | |
| "learning_rate": 3.2835740023299288e-06, | |
| "loss": 4.406, | |
| "step": 172500 | |
| }, | |
| { | |
| "epoch": 4.69, | |
| "learning_rate": 3.1481130286364498e-06, | |
| "loss": 4.4056, | |
| "step": 173000 | |
| }, | |
| { | |
| "epoch": 4.7, | |
| "learning_rate": 3.012652054942971e-06, | |
| "loss": 4.4171, | |
| "step": 173500 | |
| }, | |
| { | |
| "epoch": 4.71, | |
| "learning_rate": 2.877191081249492e-06, | |
| "loss": 4.3648, | |
| "step": 174000 | |
| }, | |
| { | |
| "epoch": 4.73, | |
| "learning_rate": 2.7417301075560136e-06, | |
| "loss": 4.361, | |
| "step": 174500 | |
| }, | |
| { | |
| "epoch": 4.74, | |
| "learning_rate": 2.606269133862534e-06, | |
| "loss": 4.3864, | |
| "step": 175000 | |
| }, | |
| { | |
| "epoch": 4.75, | |
| "learning_rate": 2.470808160169055e-06, | |
| "loss": 4.3895, | |
| "step": 175500 | |
| }, | |
| { | |
| "epoch": 4.77, | |
| "learning_rate": 2.3353471864755766e-06, | |
| "loss": 4.393, | |
| "step": 176000 | |
| }, | |
| { | |
| "epoch": 4.78, | |
| "learning_rate": 2.2001571347294845e-06, | |
| "loss": 4.3817, | |
| "step": 176500 | |
| }, | |
| { | |
| "epoch": 4.8, | |
| "learning_rate": 2.0646961610360055e-06, | |
| "loss": 4.3793, | |
| "step": 177000 | |
| }, | |
| { | |
| "epoch": 4.81, | |
| "learning_rate": 1.9295061092899135e-06, | |
| "loss": 4.3629, | |
| "step": 177500 | |
| }, | |
| { | |
| "epoch": 4.82, | |
| "learning_rate": 1.794045135596435e-06, | |
| "loss": 4.3904, | |
| "step": 178000 | |
| }, | |
| { | |
| "epoch": 4.84, | |
| "learning_rate": 1.6588550838503429e-06, | |
| "loss": 4.3818, | |
| "step": 178500 | |
| }, | |
| { | |
| "epoch": 4.85, | |
| "learning_rate": 1.5236650321042508e-06, | |
| "loss": 4.3815, | |
| "step": 179000 | |
| }, | |
| { | |
| "epoch": 4.86, | |
| "learning_rate": 1.388204058410772e-06, | |
| "loss": 4.3918, | |
| "step": 179500 | |
| }, | |
| { | |
| "epoch": 4.88, | |
| "learning_rate": 1.252743084717293e-06, | |
| "loss": 4.384, | |
| "step": 180000 | |
| }, | |
| { | |
| "epoch": 4.89, | |
| "learning_rate": 1.1172821110238142e-06, | |
| "loss": 4.3801, | |
| "step": 180500 | |
| }, | |
| { | |
| "epoch": 4.9, | |
| "learning_rate": 9.818211373303352e-07, | |
| "loss": 4.3809, | |
| "step": 181000 | |
| }, | |
| { | |
| "epoch": 4.92, | |
| "learning_rate": 8.463601636368562e-07, | |
| "loss": 4.3948, | |
| "step": 181500 | |
| }, | |
| { | |
| "epoch": 4.93, | |
| "learning_rate": 7.108991899433774e-07, | |
| "loss": 4.3838, | |
| "step": 182000 | |
| }, | |
| { | |
| "epoch": 4.94, | |
| "learning_rate": 5.754382162498984e-07, | |
| "loss": 4.3895, | |
| "step": 182500 | |
| }, | |
| { | |
| "epoch": 4.96, | |
| "learning_rate": 4.399772425564195e-07, | |
| "loss": 4.3856, | |
| "step": 183000 | |
| }, | |
| { | |
| "epoch": 4.97, | |
| "learning_rate": 3.045162688629406e-07, | |
| "loss": 4.3909, | |
| "step": 183500 | |
| }, | |
| { | |
| "epoch": 4.98, | |
| "learning_rate": 1.6905529516946169e-07, | |
| "loss": 4.3851, | |
| "step": 184000 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "learning_rate": 3.3594321475982776e-08, | |
| "loss": 4.3734, | |
| "step": 184500 | |
| }, | |
| { | |
| "epoch": 5.0, | |
| "step": 184555, | |
| "total_flos": 1.7213246560027238e+18, | |
| "train_loss": 5.094274029172764, | |
| "train_runtime": 57370.2102, | |
| "train_samples_per_second": 32.169, | |
| "train_steps_per_second": 3.217 | |
| } | |
| ], | |
| "max_steps": 184555, | |
| "num_train_epochs": 5, | |
| "total_flos": 1.7213246560027238e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |