| { | |
| "best_metric": 0.7949723601341248, | |
| "best_model_checkpoint": "/jmain02/home/J2AD002/jxm09/hrk35-jxm09/hatemoji_r7_data/output/hs_data--prior_rounds--r7--5/deberta123/checkpoint-46574", | |
| "epoch": 3.0, | |
| "global_step": 69861, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.02, | |
| "learning_rate": 1.9856858619258244e-05, | |
| "loss": 0.53, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "learning_rate": 1.9713717238516486e-05, | |
| "loss": 0.308, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "learning_rate": 1.9570575857774725e-05, | |
| "loss": 0.2211, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "learning_rate": 1.9427434477032967e-05, | |
| "loss": 0.1787, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "learning_rate": 1.928429309629121e-05, | |
| "loss": 0.1482, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "learning_rate": 1.914115171554945e-05, | |
| "loss": 0.1392, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 1.899801033480769e-05, | |
| "loss": 0.1246, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "learning_rate": 1.8854868954065932e-05, | |
| "loss": 0.1191, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "learning_rate": 1.8711727573324174e-05, | |
| "loss": 0.115, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "learning_rate": 1.8568586192582416e-05, | |
| "loss": 0.1038, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "learning_rate": 1.8425444811840655e-05, | |
| "loss": 0.098, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "learning_rate": 1.8282303431098897e-05, | |
| "loss": 0.1044, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "learning_rate": 1.813916205035714e-05, | |
| "loss": 0.1011, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 1.799602066961538e-05, | |
| "loss": 0.0935, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "learning_rate": 1.785287928887362e-05, | |
| "loss": 0.0926, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "learning_rate": 1.7709737908131862e-05, | |
| "loss": 0.095, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 1.7566596527390104e-05, | |
| "loss": 0.0863, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "learning_rate": 1.7423455146648346e-05, | |
| "loss": 0.0886, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "learning_rate": 1.7280313765906588e-05, | |
| "loss": 0.0876, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "learning_rate": 1.713717238516483e-05, | |
| "loss": 0.089, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 1.699403100442307e-05, | |
| "loss": 0.0871, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "learning_rate": 1.685088962368131e-05, | |
| "loss": 0.0874, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "learning_rate": 1.6707748242939553e-05, | |
| "loss": 0.0802, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 1.6564606862197795e-05, | |
| "loss": 0.0826, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "learning_rate": 1.6421465481456037e-05, | |
| "loss": 0.08, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "learning_rate": 1.627832410071428e-05, | |
| "loss": 0.081, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "learning_rate": 1.6135182719972518e-05, | |
| "loss": 0.0798, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 1.599204133923076e-05, | |
| "loss": 0.0813, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "learning_rate": 1.5848899958489002e-05, | |
| "loss": 0.0779, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "learning_rate": 1.5705758577747244e-05, | |
| "loss": 0.0769, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 1.5562617197005483e-05, | |
| "loss": 0.0817, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "learning_rate": 1.5419475816263725e-05, | |
| "loss": 0.0765, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "learning_rate": 1.5276334435521967e-05, | |
| "loss": 0.0784, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "learning_rate": 1.5133193054780208e-05, | |
| "loss": 0.0763, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 1.4990051674038448e-05, | |
| "loss": 0.0795, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "learning_rate": 1.484691029329669e-05, | |
| "loss": 0.0758, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "learning_rate": 1.4703768912554932e-05, | |
| "loss": 0.0748, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 1.4560627531813175e-05, | |
| "loss": 0.0741, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "learning_rate": 1.4417486151071413e-05, | |
| "loss": 0.0726, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "learning_rate": 1.4274344770329655e-05, | |
| "loss": 0.0752, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "learning_rate": 1.4131203389587897e-05, | |
| "loss": 0.0746, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 1.398806200884614e-05, | |
| "loss": 0.0733, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "learning_rate": 1.3844920628104378e-05, | |
| "loss": 0.071, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "learning_rate": 1.370177924736262e-05, | |
| "loss": 0.0745, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 1.3558637866620862e-05, | |
| "loss": 0.0751, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 0.99, | |
| "learning_rate": 1.3415496485879105e-05, | |
| "loss": 0.0722, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "eval_accuracy": 0.8151779770851135, | |
| "eval_loss": 0.8366696834564209, | |
| "eval_runtime": 23.897, | |
| "eval_samples_per_second": 196.301, | |
| "eval_steps_per_second": 3.097, | |
| "step": 23287 | |
| }, | |
| { | |
| "epoch": 1.01, | |
| "learning_rate": 1.3272355105137345e-05, | |
| "loss": 0.0662, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 1.03, | |
| "learning_rate": 1.3129213724395587e-05, | |
| "loss": 0.0602, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 1.2986072343653827e-05, | |
| "loss": 0.0619, | |
| "step": 24500 | |
| }, | |
| { | |
| "epoch": 1.07, | |
| "learning_rate": 1.284293096291207e-05, | |
| "loss": 0.0583, | |
| "step": 25000 | |
| }, | |
| { | |
| "epoch": 1.1, | |
| "learning_rate": 1.269978958217031e-05, | |
| "loss": 0.0586, | |
| "step": 25500 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 1.2556648201428552e-05, | |
| "loss": 0.0585, | |
| "step": 26000 | |
| }, | |
| { | |
| "epoch": 1.14, | |
| "learning_rate": 1.2413506820686794e-05, | |
| "loss": 0.059, | |
| "step": 26500 | |
| }, | |
| { | |
| "epoch": 1.16, | |
| "learning_rate": 1.2270365439945036e-05, | |
| "loss": 0.0575, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 1.18, | |
| "learning_rate": 1.2127224059203275e-05, | |
| "loss": 0.0596, | |
| "step": 27500 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 1.1984082678461517e-05, | |
| "loss": 0.0595, | |
| "step": 28000 | |
| }, | |
| { | |
| "epoch": 1.22, | |
| "learning_rate": 1.184094129771976e-05, | |
| "loss": 0.0588, | |
| "step": 28500 | |
| }, | |
| { | |
| "epoch": 1.25, | |
| "learning_rate": 1.1697799916978001e-05, | |
| "loss": 0.0593, | |
| "step": 29000 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 1.155465853623624e-05, | |
| "loss": 0.0612, | |
| "step": 29500 | |
| }, | |
| { | |
| "epoch": 1.29, | |
| "learning_rate": 1.1411517155494482e-05, | |
| "loss": 0.059, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 1.31, | |
| "learning_rate": 1.1268375774752724e-05, | |
| "loss": 0.0566, | |
| "step": 30500 | |
| }, | |
| { | |
| "epoch": 1.33, | |
| "learning_rate": 1.1125234394010966e-05, | |
| "loss": 0.0589, | |
| "step": 31000 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 1.0982093013269207e-05, | |
| "loss": 0.0601, | |
| "step": 31500 | |
| }, | |
| { | |
| "epoch": 1.37, | |
| "learning_rate": 1.0838951632527447e-05, | |
| "loss": 0.0601, | |
| "step": 32000 | |
| }, | |
| { | |
| "epoch": 1.4, | |
| "learning_rate": 1.069581025178569e-05, | |
| "loss": 0.0595, | |
| "step": 32500 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 1.0552668871043931e-05, | |
| "loss": 0.0603, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 1.44, | |
| "learning_rate": 1.0409527490302172e-05, | |
| "loss": 0.0571, | |
| "step": 33500 | |
| }, | |
| { | |
| "epoch": 1.46, | |
| "learning_rate": 1.0266386109560414e-05, | |
| "loss": 0.0587, | |
| "step": 34000 | |
| }, | |
| { | |
| "epoch": 1.48, | |
| "learning_rate": 1.0123244728818656e-05, | |
| "loss": 0.0565, | |
| "step": 34500 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 9.980103348076896e-06, | |
| "loss": 0.0579, | |
| "step": 35000 | |
| }, | |
| { | |
| "epoch": 1.52, | |
| "learning_rate": 9.836961967335139e-06, | |
| "loss": 0.0619, | |
| "step": 35500 | |
| }, | |
| { | |
| "epoch": 1.55, | |
| "learning_rate": 9.693820586593379e-06, | |
| "loss": 0.0563, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 9.550679205851621e-06, | |
| "loss": 0.0548, | |
| "step": 36500 | |
| }, | |
| { | |
| "epoch": 1.59, | |
| "learning_rate": 9.407537825109861e-06, | |
| "loss": 0.0581, | |
| "step": 37000 | |
| }, | |
| { | |
| "epoch": 1.61, | |
| "learning_rate": 9.264396444368104e-06, | |
| "loss": 0.0568, | |
| "step": 37500 | |
| }, | |
| { | |
| "epoch": 1.63, | |
| "learning_rate": 9.121255063626344e-06, | |
| "loss": 0.055, | |
| "step": 38000 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 8.978113682884586e-06, | |
| "loss": 0.0598, | |
| "step": 38500 | |
| }, | |
| { | |
| "epoch": 1.67, | |
| "learning_rate": 8.834972302142826e-06, | |
| "loss": 0.0597, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 1.7, | |
| "learning_rate": 8.691830921401069e-06, | |
| "loss": 0.058, | |
| "step": 39500 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 8.548689540659309e-06, | |
| "loss": 0.0554, | |
| "step": 40000 | |
| }, | |
| { | |
| "epoch": 1.74, | |
| "learning_rate": 8.405548159917551e-06, | |
| "loss": 0.0563, | |
| "step": 40500 | |
| }, | |
| { | |
| "epoch": 1.76, | |
| "learning_rate": 8.262406779175792e-06, | |
| "loss": 0.0571, | |
| "step": 41000 | |
| }, | |
| { | |
| "epoch": 1.78, | |
| "learning_rate": 8.119265398434034e-06, | |
| "loss": 0.0579, | |
| "step": 41500 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 7.976124017692276e-06, | |
| "loss": 0.0564, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 1.83, | |
| "learning_rate": 7.832982636950516e-06, | |
| "loss": 0.0577, | |
| "step": 42500 | |
| }, | |
| { | |
| "epoch": 1.85, | |
| "learning_rate": 7.689841256208758e-06, | |
| "loss": 0.0568, | |
| "step": 43000 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 7.5466998754669995e-06, | |
| "loss": 0.0596, | |
| "step": 43500 | |
| }, | |
| { | |
| "epoch": 1.89, | |
| "learning_rate": 7.403558494725241e-06, | |
| "loss": 0.057, | |
| "step": 44000 | |
| }, | |
| { | |
| "epoch": 1.91, | |
| "learning_rate": 7.260417113983482e-06, | |
| "loss": 0.0579, | |
| "step": 44500 | |
| }, | |
| { | |
| "epoch": 1.93, | |
| "learning_rate": 7.117275733241723e-06, | |
| "loss": 0.0561, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 6.9741343524999645e-06, | |
| "loss": 0.0548, | |
| "step": 45500 | |
| }, | |
| { | |
| "epoch": 1.98, | |
| "learning_rate": 6.830992971758206e-06, | |
| "loss": 0.0571, | |
| "step": 46000 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "learning_rate": 6.687851591016448e-06, | |
| "loss": 0.055, | |
| "step": 46500 | |
| }, | |
| { | |
| "epoch": 2.0, | |
| "eval_accuracy": 0.8311660885810852, | |
| "eval_loss": 0.7949723601341248, | |
| "eval_runtime": 27.7911, | |
| "eval_samples_per_second": 168.795, | |
| "eval_steps_per_second": 2.663, | |
| "step": 46574 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 6.544710210274688e-06, | |
| "loss": 0.0471, | |
| "step": 47000 | |
| }, | |
| { | |
| "epoch": 2.04, | |
| "learning_rate": 6.40156882953293e-06, | |
| "loss": 0.0435, | |
| "step": 47500 | |
| }, | |
| { | |
| "epoch": 2.06, | |
| "learning_rate": 6.258427448791171e-06, | |
| "loss": 0.043, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 2.08, | |
| "learning_rate": 6.115286068049413e-06, | |
| "loss": 0.0431, | |
| "step": 48500 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 5.972144687307654e-06, | |
| "loss": 0.0433, | |
| "step": 49000 | |
| }, | |
| { | |
| "epoch": 2.13, | |
| "learning_rate": 5.8290033065658954e-06, | |
| "loss": 0.0442, | |
| "step": 49500 | |
| }, | |
| { | |
| "epoch": 2.15, | |
| "learning_rate": 5.685861925824137e-06, | |
| "loss": 0.0447, | |
| "step": 50000 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 5.542720545082379e-06, | |
| "loss": 0.0454, | |
| "step": 50500 | |
| }, | |
| { | |
| "epoch": 2.19, | |
| "learning_rate": 5.399579164340619e-06, | |
| "loss": 0.0441, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 2.21, | |
| "learning_rate": 5.256437783598861e-06, | |
| "loss": 0.0441, | |
| "step": 51500 | |
| }, | |
| { | |
| "epoch": 2.23, | |
| "learning_rate": 5.113296402857102e-06, | |
| "loss": 0.0444, | |
| "step": 52000 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 4.970155022115344e-06, | |
| "loss": 0.0435, | |
| "step": 52500 | |
| }, | |
| { | |
| "epoch": 2.28, | |
| "learning_rate": 4.827013641373585e-06, | |
| "loss": 0.0445, | |
| "step": 53000 | |
| }, | |
| { | |
| "epoch": 2.3, | |
| "learning_rate": 4.683872260631826e-06, | |
| "loss": 0.041, | |
| "step": 53500 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 4.540730879890068e-06, | |
| "loss": 0.0461, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 2.34, | |
| "learning_rate": 4.397589499148309e-06, | |
| "loss": 0.0421, | |
| "step": 54500 | |
| }, | |
| { | |
| "epoch": 2.36, | |
| "learning_rate": 4.254448118406551e-06, | |
| "loss": 0.0423, | |
| "step": 55000 | |
| }, | |
| { | |
| "epoch": 2.38, | |
| "learning_rate": 4.111306737664792e-06, | |
| "loss": 0.0448, | |
| "step": 55500 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 3.9681653569230335e-06, | |
| "loss": 0.0447, | |
| "step": 56000 | |
| }, | |
| { | |
| "epoch": 2.43, | |
| "learning_rate": 3.825023976181275e-06, | |
| "loss": 0.0461, | |
| "step": 56500 | |
| }, | |
| { | |
| "epoch": 2.45, | |
| "learning_rate": 3.681882595439516e-06, | |
| "loss": 0.0444, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 3.5387412146977573e-06, | |
| "loss": 0.0431, | |
| "step": 57500 | |
| }, | |
| { | |
| "epoch": 2.49, | |
| "learning_rate": 3.3955998339559985e-06, | |
| "loss": 0.042, | |
| "step": 58000 | |
| }, | |
| { | |
| "epoch": 2.51, | |
| "learning_rate": 3.25245845321424e-06, | |
| "loss": 0.0415, | |
| "step": 58500 | |
| }, | |
| { | |
| "epoch": 2.53, | |
| "learning_rate": 3.1093170724724815e-06, | |
| "loss": 0.0433, | |
| "step": 59000 | |
| }, | |
| { | |
| "epoch": 2.56, | |
| "learning_rate": 2.9661756917307227e-06, | |
| "loss": 0.0422, | |
| "step": 59500 | |
| }, | |
| { | |
| "epoch": 2.58, | |
| "learning_rate": 2.823034310988964e-06, | |
| "loss": 0.0389, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 2.6, | |
| "learning_rate": 2.6798929302472052e-06, | |
| "loss": 0.0424, | |
| "step": 60500 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 2.536751549505447e-06, | |
| "loss": 0.0397, | |
| "step": 61000 | |
| }, | |
| { | |
| "epoch": 2.64, | |
| "learning_rate": 2.393610168763688e-06, | |
| "loss": 0.044, | |
| "step": 61500 | |
| }, | |
| { | |
| "epoch": 2.66, | |
| "learning_rate": 2.2504687880219294e-06, | |
| "loss": 0.0443, | |
| "step": 62000 | |
| }, | |
| { | |
| "epoch": 2.68, | |
| "learning_rate": 2.1073274072801707e-06, | |
| "loss": 0.0414, | |
| "step": 62500 | |
| }, | |
| { | |
| "epoch": 2.71, | |
| "learning_rate": 1.9641860265384124e-06, | |
| "loss": 0.041, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 2.73, | |
| "learning_rate": 1.8210446457966536e-06, | |
| "loss": 0.0389, | |
| "step": 63500 | |
| }, | |
| { | |
| "epoch": 2.75, | |
| "learning_rate": 1.6779032650548949e-06, | |
| "loss": 0.0433, | |
| "step": 64000 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 1.5347618843131361e-06, | |
| "loss": 0.0417, | |
| "step": 64500 | |
| }, | |
| { | |
| "epoch": 2.79, | |
| "learning_rate": 1.3916205035713776e-06, | |
| "loss": 0.0416, | |
| "step": 65000 | |
| }, | |
| { | |
| "epoch": 2.81, | |
| "learning_rate": 1.2484791228296189e-06, | |
| "loss": 0.0424, | |
| "step": 65500 | |
| }, | |
| { | |
| "epoch": 2.83, | |
| "learning_rate": 1.1053377420878603e-06, | |
| "loss": 0.0431, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 2.86, | |
| "learning_rate": 9.621963613461016e-07, | |
| "loss": 0.0417, | |
| "step": 66500 | |
| }, | |
| { | |
| "epoch": 2.88, | |
| "learning_rate": 8.19054980604343e-07, | |
| "loss": 0.0418, | |
| "step": 67000 | |
| }, | |
| { | |
| "epoch": 2.9, | |
| "learning_rate": 6.759135998625843e-07, | |
| "loss": 0.0399, | |
| "step": 67500 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 5.327722191208257e-07, | |
| "loss": 0.0416, | |
| "step": 68000 | |
| }, | |
| { | |
| "epoch": 2.94, | |
| "learning_rate": 3.89630838379067e-07, | |
| "loss": 0.0424, | |
| "step": 68500 | |
| }, | |
| { | |
| "epoch": 2.96, | |
| "learning_rate": 2.464894576373084e-07, | |
| "loss": 0.0418, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 2.98, | |
| "learning_rate": 1.0334807689554974e-07, | |
| "loss": 0.0399, | |
| "step": 69500 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "eval_accuracy": 0.8266894221305847, | |
| "eval_loss": 0.950989842414856, | |
| "eval_runtime": 23.2697, | |
| "eval_samples_per_second": 201.593, | |
| "eval_steps_per_second": 3.18, | |
| "step": 69861 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "step": 69861, | |
| "total_flos": 1.9117803270808904e+18, | |
| "train_runtime": 54665.4444, | |
| "train_samples_per_second": 81.787, | |
| "train_steps_per_second": 1.278 | |
| } | |
| ], | |
| "max_steps": 69861, | |
| "num_train_epochs": 3, | |
| "total_flos": 1.9117803270808904e+18, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |