| { | |
| "best_metric": 0.031979888677597046, | |
| "best_model_checkpoint": "./results/answerdotai/ModernBERT-base/trial-2/checkpoint-6022", | |
| "epoch": 2.0, | |
| "eval_steps": 500, | |
| "global_step": 6022, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.016605778811026237, | |
| "grad_norm": 21.788597106933594, | |
| "learning_rate": 5.429575351871404e-06, | |
| "loss": 0.5789, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.033211557622052475, | |
| "grad_norm": 20.038349151611328, | |
| "learning_rate": 5.416664391316233e-06, | |
| "loss": 0.37, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.04981733643307871, | |
| "grad_norm": 23.927526473999023, | |
| "learning_rate": 5.403753430761063e-06, | |
| "loss": 0.25, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.06642311524410495, | |
| "grad_norm": 4.1712799072265625, | |
| "learning_rate": 5.390842470205893e-06, | |
| "loss": 0.1921, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.08302889405513118, | |
| "grad_norm": 6.138601303100586, | |
| "learning_rate": 5.3779315096507225e-06, | |
| "loss": 0.1365, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.09963467286615742, | |
| "grad_norm": 0.9431160092353821, | |
| "learning_rate": 5.3650205490955514e-06, | |
| "loss": 0.1473, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.11624045167718366, | |
| "grad_norm": 25.303245544433594, | |
| "learning_rate": 5.352109588540381e-06, | |
| "loss": 0.0875, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.1328462304882099, | |
| "grad_norm": 14.83379077911377, | |
| "learning_rate": 5.33919862798521e-06, | |
| "loss": 0.111, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.14945200929923613, | |
| "grad_norm": 0.2346535325050354, | |
| "learning_rate": 5.32628766743004e-06, | |
| "loss": 0.0722, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.16605778811026237, | |
| "grad_norm": 19.045169830322266, | |
| "learning_rate": 5.31337670687487e-06, | |
| "loss": 0.1236, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.1826635669212886, | |
| "grad_norm": 10.871609687805176, | |
| "learning_rate": 5.300465746319699e-06, | |
| "loss": 0.1018, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.19926934573231483, | |
| "grad_norm": 8.278830528259277, | |
| "learning_rate": 5.287554785764528e-06, | |
| "loss": 0.0608, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.2158751245433411, | |
| "grad_norm": 3.4486818313598633, | |
| "learning_rate": 5.274643825209358e-06, | |
| "loss": 0.0684, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.23248090335436733, | |
| "grad_norm": 9.789453506469727, | |
| "learning_rate": 5.261732864654187e-06, | |
| "loss": 0.0826, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.24908668216539356, | |
| "grad_norm": 0.013454285450279713, | |
| "learning_rate": 5.248821904099017e-06, | |
| "loss": 0.0672, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.2656924609764198, | |
| "grad_norm": 0.8878294825553894, | |
| "learning_rate": 5.2359109435438465e-06, | |
| "loss": 0.0472, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.282298239787446, | |
| "grad_norm": 15.41006088256836, | |
| "learning_rate": 5.222999982988676e-06, | |
| "loss": 0.0616, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.29890401859847227, | |
| "grad_norm": 0.04324938729405403, | |
| "learning_rate": 5.210089022433506e-06, | |
| "loss": 0.0215, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.3155097974094985, | |
| "grad_norm": 0.011849366128444672, | |
| "learning_rate": 5.197178061878335e-06, | |
| "loss": 0.0398, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.33211557622052473, | |
| "grad_norm": 0.0020897299982607365, | |
| "learning_rate": 5.184267101323165e-06, | |
| "loss": 0.0294, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.348721355031551, | |
| "grad_norm": 0.00038467388367280364, | |
| "learning_rate": 5.171356140767994e-06, | |
| "loss": 0.0328, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.3653271338425772, | |
| "grad_norm": 0.0022064056247472763, | |
| "learning_rate": 5.158445180212823e-06, | |
| "loss": 0.0216, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.38193291265360346, | |
| "grad_norm": 0.012603014707565308, | |
| "learning_rate": 5.145534219657653e-06, | |
| "loss": 0.0293, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.39853869146462967, | |
| "grad_norm": 0.002970542525872588, | |
| "learning_rate": 5.132623259102483e-06, | |
| "loss": 0.0133, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.41514447027565593, | |
| "grad_norm": 0.09289965778589249, | |
| "learning_rate": 5.119712298547312e-06, | |
| "loss": 0.0189, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.4317502490866822, | |
| "grad_norm": 0.030116688460111618, | |
| "learning_rate": 5.106801337992142e-06, | |
| "loss": 0.0266, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.4483560278977084, | |
| "grad_norm": 23.291847229003906, | |
| "learning_rate": 5.0938903774369705e-06, | |
| "loss": 0.0378, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.46496180670873466, | |
| "grad_norm": 0.00580954784527421, | |
| "learning_rate": 5.0809794168818e-06, | |
| "loss": 0.0002, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.48156758551976087, | |
| "grad_norm": 0.0036250711418688297, | |
| "learning_rate": 5.06806845632663e-06, | |
| "loss": 0.0297, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.4981733643307871, | |
| "grad_norm": 0.0013630707981064916, | |
| "learning_rate": 5.05515749577146e-06, | |
| "loss": 0.0114, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.5147791431418134, | |
| "grad_norm": 0.025447094812989235, | |
| "learning_rate": 5.042246535216289e-06, | |
| "loss": 0.0019, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.5313849219528396, | |
| "grad_norm": 18.81841468811035, | |
| "learning_rate": 5.0293355746611185e-06, | |
| "loss": 0.0286, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.5479907007638658, | |
| "grad_norm": 0.0033424277789890766, | |
| "learning_rate": 5.016424614105948e-06, | |
| "loss": 0.0393, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.564596479574892, | |
| "grad_norm": 0.039123374968767166, | |
| "learning_rate": 5.003513653550777e-06, | |
| "loss": 0.0186, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.5812022583859183, | |
| "grad_norm": 0.0005275913863442838, | |
| "learning_rate": 4.990602692995607e-06, | |
| "loss": 0.0003, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.5978080371969445, | |
| "grad_norm": 0.005070064682513475, | |
| "learning_rate": 4.977691732440437e-06, | |
| "loss": 0.01, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.6144138160079707, | |
| "grad_norm": 0.003932475112378597, | |
| "learning_rate": 4.9647807718852664e-06, | |
| "loss": 0.0222, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.631019594818997, | |
| "grad_norm": 0.6544032692909241, | |
| "learning_rate": 4.951869811330095e-06, | |
| "loss": 0.0138, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.6476253736300233, | |
| "grad_norm": 0.008768323808908463, | |
| "learning_rate": 4.938958850774925e-06, | |
| "loss": 0.0056, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.6642311524410495, | |
| "grad_norm": 0.0021180976182222366, | |
| "learning_rate": 4.926047890219754e-06, | |
| "loss": 0.0049, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.6808369312520757, | |
| "grad_norm": 0.002039346843957901, | |
| "learning_rate": 4.913136929664584e-06, | |
| "loss": 0.0142, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.697442710063102, | |
| "grad_norm": 0.012900142930448055, | |
| "learning_rate": 4.9002259691094136e-06, | |
| "loss": 0.0105, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.7140484888741282, | |
| "grad_norm": 0.0022153747268021107, | |
| "learning_rate": 4.887315008554243e-06, | |
| "loss": 0.0142, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.7306542676851544, | |
| "grad_norm": 0.001426122267730534, | |
| "learning_rate": 4.874404047999072e-06, | |
| "loss": 0.0068, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.7472600464961807, | |
| "grad_norm": 0.0008603449095971882, | |
| "learning_rate": 4.861493087443902e-06, | |
| "loss": 0.0119, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.7638658253072069, | |
| "grad_norm": 0.0006780526018701494, | |
| "learning_rate": 4.848582126888731e-06, | |
| "loss": 0.0108, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.7804716041182331, | |
| "grad_norm": 0.014527379535138607, | |
| "learning_rate": 4.835671166333561e-06, | |
| "loss": 0.0002, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.7970773829292593, | |
| "grad_norm": 0.00022624376288149506, | |
| "learning_rate": 4.8227602057783904e-06, | |
| "loss": 0.0092, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.8136831617402857, | |
| "grad_norm": 0.0044932495802640915, | |
| "learning_rate": 4.80984924522322e-06, | |
| "loss": 0.0001, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.8302889405513119, | |
| "grad_norm": 0.0009355309884995222, | |
| "learning_rate": 4.79693828466805e-06, | |
| "loss": 0.0002, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.8468947193623381, | |
| "grad_norm": 0.12550997734069824, | |
| "learning_rate": 4.784027324112879e-06, | |
| "loss": 0.0024, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.8635004981733644, | |
| "grad_norm": 0.02399071305990219, | |
| "learning_rate": 4.771116363557709e-06, | |
| "loss": 0.0099, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.8801062769843906, | |
| "grad_norm": 0.008470265194773674, | |
| "learning_rate": 4.7582054030025375e-06, | |
| "loss": 0.0157, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.8967120557954168, | |
| "grad_norm": 3.967735028709285e-05, | |
| "learning_rate": 4.745294442447367e-06, | |
| "loss": 0.0013, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.913317834606443, | |
| "grad_norm": 0.0005532742943614721, | |
| "learning_rate": 4.732383481892197e-06, | |
| "loss": 0.0025, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.9299236134174693, | |
| "grad_norm": 9.227233022102155e-06, | |
| "learning_rate": 4.719472521337027e-06, | |
| "loss": 0.0028, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.9465293922284955, | |
| "grad_norm": 0.280258446931839, | |
| "learning_rate": 4.706561560781856e-06, | |
| "loss": 0.0004, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.9631351710395217, | |
| "grad_norm": 27.427757263183594, | |
| "learning_rate": 4.6936506002266855e-06, | |
| "loss": 0.0127, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.9797409498505479, | |
| "grad_norm": 176.85423278808594, | |
| "learning_rate": 4.680739639671514e-06, | |
| "loss": 0.0298, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.9963467286615743, | |
| "grad_norm": 0.00011263355554547161, | |
| "learning_rate": 4.667828679116344e-06, | |
| "loss": 0.001, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.9963024809160306, | |
| "eval_f1": 0.9962431632227496, | |
| "eval_loss": 0.04071500524878502, | |
| "eval_precision": 0.9962693439313673, | |
| "eval_recall": 0.9963024809160306, | |
| "eval_runtime": 38.0003, | |
| "eval_samples_per_second": 220.63, | |
| "eval_steps_per_second": 13.789, | |
| "step": 3011 | |
| }, | |
| { | |
| "epoch": 1.0129525074726005, | |
| "grad_norm": 0.05092976614832878, | |
| "learning_rate": 4.654917718561174e-06, | |
| "loss": 0.018, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 1.0295582862836268, | |
| "grad_norm": 3.4633874747669324e-05, | |
| "learning_rate": 4.642006758006004e-06, | |
| "loss": 0.0, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 1.0461640650946529, | |
| "grad_norm": 8.058391540544108e-05, | |
| "learning_rate": 4.629095797450833e-06, | |
| "loss": 0.0, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 1.0627698439056792, | |
| "grad_norm": 0.00043129033292643726, | |
| "learning_rate": 4.616184836895662e-06, | |
| "loss": 0.0, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 1.0793756227167055, | |
| "grad_norm": 0.012417804449796677, | |
| "learning_rate": 4.603273876340492e-06, | |
| "loss": 0.0204, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 1.0959814015277316, | |
| "grad_norm": 0.07707448303699493, | |
| "learning_rate": 4.590362915785321e-06, | |
| "loss": 0.0089, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 1.112587180338758, | |
| "grad_norm": 0.0019856118597090244, | |
| "learning_rate": 4.577451955230151e-06, | |
| "loss": 0.0003, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 1.1291929591497842, | |
| "grad_norm": 0.0003844090970233083, | |
| "learning_rate": 4.564540994674981e-06, | |
| "loss": 0.0, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 1.1457987379608103, | |
| "grad_norm": 0.004796341527253389, | |
| "learning_rate": 4.55163003411981e-06, | |
| "loss": 0.0054, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 1.1624045167718366, | |
| "grad_norm": 0.0021394495852291584, | |
| "learning_rate": 4.538719073564639e-06, | |
| "loss": 0.0001, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.1790102955828627, | |
| "grad_norm": 0.00016287445032503456, | |
| "learning_rate": 4.525808113009469e-06, | |
| "loss": 0.0017, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 1.195616074393889, | |
| "grad_norm": 0.005753168836236, | |
| "learning_rate": 4.512897152454298e-06, | |
| "loss": 0.0132, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 1.2122218532049154, | |
| "grad_norm": 0.00012519631127361208, | |
| "learning_rate": 4.499986191899128e-06, | |
| "loss": 0.0, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 1.2288276320159415, | |
| "grad_norm": 0.0009526669164188206, | |
| "learning_rate": 4.487075231343957e-06, | |
| "loss": 0.0083, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 1.2454334108269678, | |
| "grad_norm": 6.90124070388265e-05, | |
| "learning_rate": 4.474164270788787e-06, | |
| "loss": 0.0114, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 1.2620391896379939, | |
| "grad_norm": 0.0029422417283058167, | |
| "learning_rate": 4.461253310233616e-06, | |
| "loss": 0.0001, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 1.2786449684490202, | |
| "grad_norm": 1.6564589738845825, | |
| "learning_rate": 4.448342349678446e-06, | |
| "loss": 0.0065, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 1.2952507472600465, | |
| "grad_norm": 4.6906425268389285e-05, | |
| "learning_rate": 4.435431389123275e-06, | |
| "loss": 0.0, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 1.3118565260710726, | |
| "grad_norm": 1.4456440112553537e-05, | |
| "learning_rate": 4.4225204285681046e-06, | |
| "loss": 0.0, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 1.328462304882099, | |
| "grad_norm": 4.6707005822099745e-05, | |
| "learning_rate": 4.409609468012934e-06, | |
| "loss": 0.0227, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.3450680836931252, | |
| "grad_norm": 4.7155015636235476e-05, | |
| "learning_rate": 4.396698507457763e-06, | |
| "loss": 0.0002, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 1.3616738625041513, | |
| "grad_norm": 0.01696430891752243, | |
| "learning_rate": 4.383787546902593e-06, | |
| "loss": 0.0188, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 1.3782796413151777, | |
| "grad_norm": 0.0008329456904903054, | |
| "learning_rate": 4.370876586347423e-06, | |
| "loss": 0.0178, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 1.394885420126204, | |
| "grad_norm": 9.179511835100129e-05, | |
| "learning_rate": 4.3579656257922525e-06, | |
| "loss": 0.0, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 1.41149119893723, | |
| "grad_norm": 2.924172622442711e-05, | |
| "learning_rate": 4.3450546652370814e-06, | |
| "loss": 0.0013, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 1.4280969777482564, | |
| "grad_norm": 0.015076125971972942, | |
| "learning_rate": 4.332143704681911e-06, | |
| "loss": 0.0104, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 1.4447027565592827, | |
| "grad_norm": 5.385762415244244e-05, | |
| "learning_rate": 4.31923274412674e-06, | |
| "loss": 0.014, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 1.4613085353703088, | |
| "grad_norm": 0.0007110639126040041, | |
| "learning_rate": 4.30632178357157e-06, | |
| "loss": 0.0126, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 1.4779143141813351, | |
| "grad_norm": 0.00014339391782414168, | |
| "learning_rate": 4.2934108230164e-06, | |
| "loss": 0.0003, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 1.4945200929923614, | |
| "grad_norm": 0.0006024091853760183, | |
| "learning_rate": 4.280499862461229e-06, | |
| "loss": 0.0118, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.5111258718033875, | |
| "grad_norm": 0.0002353072923142463, | |
| "learning_rate": 4.267588901906058e-06, | |
| "loss": 0.0086, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 1.5277316506144138, | |
| "grad_norm": 0.0008946498855948448, | |
| "learning_rate": 4.254677941350888e-06, | |
| "loss": 0.0, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 1.5443374294254402, | |
| "grad_norm": 7.315174298128113e-05, | |
| "learning_rate": 4.241766980795717e-06, | |
| "loss": 0.0003, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 1.5609432082364663, | |
| "grad_norm": 9.232313459506258e-05, | |
| "learning_rate": 4.228856020240547e-06, | |
| "loss": 0.0001, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 1.5775489870474926, | |
| "grad_norm": 1.4020029084349517e-05, | |
| "learning_rate": 4.2159450596853765e-06, | |
| "loss": 0.0, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 1.594154765858519, | |
| "grad_norm": 4.0607475966680795e-05, | |
| "learning_rate": 4.203034099130206e-06, | |
| "loss": 0.0, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 1.610760544669545, | |
| "grad_norm": 4.69290571345482e-05, | |
| "learning_rate": 4.190123138575036e-06, | |
| "loss": 0.0177, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 1.627366323480571, | |
| "grad_norm": 0.14096687734127045, | |
| "learning_rate": 4.177212178019865e-06, | |
| "loss": 0.0115, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 1.6439721022915976, | |
| "grad_norm": 0.00020342542848084122, | |
| "learning_rate": 4.164301217464695e-06, | |
| "loss": 0.0001, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 1.6605778811026237, | |
| "grad_norm": 0.0002786288969218731, | |
| "learning_rate": 4.151390256909524e-06, | |
| "loss": 0.0, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.6771836599136498, | |
| "grad_norm": 2.8438846129574813e-05, | |
| "learning_rate": 4.138479296354353e-06, | |
| "loss": 0.0032, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 1.6937894387246761, | |
| "grad_norm": 5.944320037087891e-06, | |
| "learning_rate": 4.125568335799183e-06, | |
| "loss": 0.0001, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 1.7103952175357025, | |
| "grad_norm": 0.005958211608231068, | |
| "learning_rate": 4.112657375244013e-06, | |
| "loss": 0.0, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 1.7270009963467285, | |
| "grad_norm": 0.002004456939175725, | |
| "learning_rate": 4.099746414688842e-06, | |
| "loss": 0.0106, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 1.7436067751577549, | |
| "grad_norm": 0.0008562383009120822, | |
| "learning_rate": 4.086835454133672e-06, | |
| "loss": 0.0081, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 1.7602125539687812, | |
| "grad_norm": 0.03570560738444328, | |
| "learning_rate": 4.0739244935785005e-06, | |
| "loss": 0.025, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 1.7768183327798073, | |
| "grad_norm": 0.001486024702899158, | |
| "learning_rate": 4.06101353302333e-06, | |
| "loss": 0.0145, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 1.7934241115908336, | |
| "grad_norm": 0.0015331929316744208, | |
| "learning_rate": 4.04810257246816e-06, | |
| "loss": 0.0001, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 1.81002989040186, | |
| "grad_norm": 0.004162834957242012, | |
| "learning_rate": 4.03519161191299e-06, | |
| "loss": 0.0005, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 1.826635669212886, | |
| "grad_norm": 0.0003064811462536454, | |
| "learning_rate": 4.022280651357819e-06, | |
| "loss": 0.0, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 1.8432414480239123, | |
| "grad_norm": 0.000830256671179086, | |
| "learning_rate": 4.0093696908026485e-06, | |
| "loss": 0.0034, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 1.8598472268349386, | |
| "grad_norm": 0.001540405093692243, | |
| "learning_rate": 3.996458730247478e-06, | |
| "loss": 0.0, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 1.8764530056459647, | |
| "grad_norm": 0.011221639811992645, | |
| "learning_rate": 3.983547769692307e-06, | |
| "loss": 0.0116, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 1.893058784456991, | |
| "grad_norm": 0.0031693174969404936, | |
| "learning_rate": 3.970636809137137e-06, | |
| "loss": 0.0061, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 1.9096645632680174, | |
| "grad_norm": 7.828649540897459e-05, | |
| "learning_rate": 3.957725848581967e-06, | |
| "loss": 0.0, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 1.9262703420790435, | |
| "grad_norm": 0.00892726145684719, | |
| "learning_rate": 3.9448148880267964e-06, | |
| "loss": 0.0003, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 1.9428761208900698, | |
| "grad_norm": 0.0033830904867500067, | |
| "learning_rate": 3.931903927471625e-06, | |
| "loss": 0.0007, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 1.959481899701096, | |
| "grad_norm": 0.017441514879465103, | |
| "learning_rate": 3.918992966916455e-06, | |
| "loss": 0.0109, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 1.9760876785121222, | |
| "grad_norm": 0.006790176033973694, | |
| "learning_rate": 3.906082006361284e-06, | |
| "loss": 0.0101, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 1.9926934573231485, | |
| "grad_norm": 0.0004248483164701611, | |
| "learning_rate": 3.893171045806114e-06, | |
| "loss": 0.0103, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.9959446564885496, | |
| "eval_f1": 0.9958827988724177, | |
| "eval_loss": 0.031979888677597046, | |
| "eval_precision": 0.9958978797187497, | |
| "eval_recall": 0.9959446564885496, | |
| "eval_runtime": 37.4063, | |
| "eval_samples_per_second": 224.134, | |
| "eval_steps_per_second": 14.008, | |
| "step": 6022 | |
| } | |
| ], | |
| "logging_steps": 50, | |
| "max_steps": 21077, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 7, | |
| "save_steps": 500, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 3.282861088518144e+16, | |
| "train_batch_size": 16, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |