{ "best_metric": null, "best_model_checkpoint": null, "epoch": 4.0, "eval_steps": 500, "global_step": 20000, "is_hyper_param_search": false, "is_local_process_zero": true, "is_world_process_zero": true, "log_history": [ { "epoch": 0.01, "grad_norm": 24.59290313720703, "learning_rate": 9.976e-07, "loss": 2.6229, "step": 50 }, { "epoch": 0.02, "grad_norm": 44.6561279296875, "learning_rate": 9.951e-07, "loss": 2.3333, "step": 100 }, { "epoch": 0.03, "grad_norm": 31.2443790435791, "learning_rate": 9.926e-07, "loss": 2.1022, "step": 150 }, { "epoch": 0.04, "grad_norm": 30.142139434814453, "learning_rate": 9.900999999999998e-07, "loss": 2.1899, "step": 200 }, { "epoch": 0.05, "grad_norm": 40.865047454833984, "learning_rate": 9.876e-07, "loss": 2.0581, "step": 250 }, { "epoch": 0.06, "grad_norm": 49.21326446533203, "learning_rate": 9.850999999999999e-07, "loss": 2.0981, "step": 300 }, { "epoch": 0.07, "grad_norm": 34.270729064941406, "learning_rate": 9.826e-07, "loss": 2.1487, "step": 350 }, { "epoch": 0.08, "grad_norm": 37.85762405395508, "learning_rate": 9.801e-07, "loss": 2.0944, "step": 400 }, { "epoch": 0.09, "grad_norm": 38.32168197631836, "learning_rate": 9.776e-07, "loss": 2.0425, "step": 450 }, { "epoch": 0.1, "grad_norm": 40.687808990478516, "learning_rate": 9.751e-07, "loss": 1.9918, "step": 500 }, { "epoch": 0.11, "grad_norm": 41.6737174987793, "learning_rate": 9.726e-07, "loss": 1.9946, "step": 550 }, { "epoch": 0.12, "grad_norm": 32.36153030395508, "learning_rate": 9.700999999999998e-07, "loss": 2.0164, "step": 600 }, { "epoch": 0.13, "grad_norm": 30.314594268798828, "learning_rate": 9.676e-07, "loss": 1.9766, "step": 650 }, { "epoch": 0.14, "grad_norm": 32.16353988647461, "learning_rate": 9.650999999999999e-07, "loss": 1.9656, "step": 700 }, { "epoch": 0.15, "grad_norm": 53.54747772216797, "learning_rate": 9.626e-07, "loss": 1.9925, "step": 750 }, { "epoch": 0.16, "grad_norm": 31.085399627685547, "learning_rate": 9.601e-07, "loss": 1.9141, "step": 800 }, { "epoch": 0.17, "grad_norm": 44.15593719482422, "learning_rate": 9.576e-07, "loss": 2.0047, "step": 850 }, { "epoch": 0.18, "grad_norm": 40.597999572753906, "learning_rate": 9.551e-07, "loss": 1.9799, "step": 900 }, { "epoch": 0.19, "grad_norm": 40.38375473022461, "learning_rate": 9.526e-07, "loss": 1.9333, "step": 950 }, { "epoch": 0.2, "grad_norm": 31.839799880981445, "learning_rate": 9.500999999999999e-07, "loss": 2.0041, "step": 1000 }, { "epoch": 0.21, "grad_norm": 43.41249084472656, "learning_rate": 9.475999999999999e-07, "loss": 1.9008, "step": 1050 }, { "epoch": 0.22, "grad_norm": 39.27376937866211, "learning_rate": 9.451e-07, "loss": 1.8454, "step": 1100 }, { "epoch": 0.23, "grad_norm": 37.960269927978516, "learning_rate": 9.426e-07, "loss": 1.8152, "step": 1150 }, { "epoch": 0.24, "grad_norm": 44.50017547607422, "learning_rate": 9.401e-07, "loss": 1.8737, "step": 1200 }, { "epoch": 0.25, "grad_norm": 37.4691162109375, "learning_rate": 9.375999999999999e-07, "loss": 1.808, "step": 1250 }, { "epoch": 0.26, "grad_norm": 37.40227127075195, "learning_rate": 9.351e-07, "loss": 1.832, "step": 1300 }, { "epoch": 0.27, "grad_norm": 39.435123443603516, "learning_rate": 9.326e-07, "loss": 1.9672, "step": 1350 }, { "epoch": 0.28, "grad_norm": 41.97803497314453, "learning_rate": 9.301e-07, "loss": 1.869, "step": 1400 }, { "epoch": 0.29, "grad_norm": 42.29490661621094, "learning_rate": 9.275999999999999e-07, "loss": 1.8844, "step": 1450 }, { "epoch": 0.3, "grad_norm": 39.2964973449707, "learning_rate": 9.250999999999999e-07, "loss": 1.8546, "step": 1500 }, { "epoch": 0.31, "grad_norm": 46.49869155883789, "learning_rate": 9.226e-07, "loss": 1.7998, "step": 1550 }, { "epoch": 0.32, "grad_norm": 35.85550308227539, "learning_rate": 9.201e-07, "loss": 1.8493, "step": 1600 }, { "epoch": 0.33, "grad_norm": 36.285282135009766, "learning_rate": 9.175999999999999e-07, "loss": 1.8918, "step": 1650 }, { "epoch": 0.34, "grad_norm": 36.704795837402344, "learning_rate": 9.151e-07, "loss": 1.8412, "step": 1700 }, { "epoch": 0.35, "grad_norm": 39.69667434692383, "learning_rate": 9.126e-07, "loss": 1.809, "step": 1750 }, { "epoch": 0.36, "grad_norm": 36.81133270263672, "learning_rate": 9.101e-07, "loss": 1.8425, "step": 1800 }, { "epoch": 0.37, "grad_norm": 40.80613708496094, "learning_rate": 9.075999999999999e-07, "loss": 1.9628, "step": 1850 }, { "epoch": 0.38, "grad_norm": 39.36043930053711, "learning_rate": 9.051e-07, "loss": 1.7546, "step": 1900 }, { "epoch": 0.39, "grad_norm": 40.22469711303711, "learning_rate": 9.025999999999999e-07, "loss": 1.8055, "step": 1950 }, { "epoch": 0.4, "grad_norm": 44.26631546020508, "learning_rate": 9.001e-07, "loss": 1.8203, "step": 2000 }, { "epoch": 0.41, "grad_norm": 50.78614807128906, "learning_rate": 8.976499999999999e-07, "loss": 1.8012, "step": 2050 }, { "epoch": 0.42, "grad_norm": 63.06776809692383, "learning_rate": 8.951499999999999e-07, "loss": 1.8745, "step": 2100 }, { "epoch": 0.43, "grad_norm": 48.32567596435547, "learning_rate": 8.9265e-07, "loss": 1.7794, "step": 2150 }, { "epoch": 0.44, "grad_norm": 48.12654495239258, "learning_rate": 8.9015e-07, "loss": 1.9107, "step": 2200 }, { "epoch": 0.45, "grad_norm": 38.266197204589844, "learning_rate": 8.8765e-07, "loss": 1.7588, "step": 2250 }, { "epoch": 0.46, "grad_norm": 39.421504974365234, "learning_rate": 8.851499999999999e-07, "loss": 1.8511, "step": 2300 }, { "epoch": 0.47, "grad_norm": 39.87221908569336, "learning_rate": 8.8265e-07, "loss": 1.6939, "step": 2350 }, { "epoch": 0.48, "grad_norm": 43.736900329589844, "learning_rate": 8.8015e-07, "loss": 1.7321, "step": 2400 }, { "epoch": 0.49, "grad_norm": 43.2308235168457, "learning_rate": 8.7765e-07, "loss": 1.8803, "step": 2450 }, { "epoch": 0.5, "grad_norm": 40.887939453125, "learning_rate": 8.751499999999999e-07, "loss": 1.731, "step": 2500 }, { "epoch": 0.51, "grad_norm": 43.211727142333984, "learning_rate": 8.7265e-07, "loss": 1.8046, "step": 2550 }, { "epoch": 0.52, "grad_norm": 51.66404342651367, "learning_rate": 8.701499999999999e-07, "loss": 1.774, "step": 2600 }, { "epoch": 0.53, "grad_norm": 39.11322021484375, "learning_rate": 8.6765e-07, "loss": 1.8792, "step": 2650 }, { "epoch": 0.54, "grad_norm": 48.62511444091797, "learning_rate": 8.6515e-07, "loss": 1.7991, "step": 2700 }, { "epoch": 0.55, "grad_norm": 41.84499740600586, "learning_rate": 8.6265e-07, "loss": 1.8712, "step": 2750 }, { "epoch": 0.56, "grad_norm": 51.82795715332031, "learning_rate": 8.601499999999999e-07, "loss": 1.7662, "step": 2800 }, { "epoch": 0.57, "grad_norm": 54.27309799194336, "learning_rate": 8.5765e-07, "loss": 1.7643, "step": 2850 }, { "epoch": 0.58, "grad_norm": 39.44385528564453, "learning_rate": 8.5515e-07, "loss": 1.804, "step": 2900 }, { "epoch": 0.59, "grad_norm": 37.73311233520508, "learning_rate": 8.5265e-07, "loss": 1.7039, "step": 2950 }, { "epoch": 0.6, "grad_norm": 43.82680892944336, "learning_rate": 8.501499999999999e-07, "loss": 1.7717, "step": 3000 }, { "epoch": 0.61, "grad_norm": 44.10558319091797, "learning_rate": 8.476499999999999e-07, "loss": 1.7109, "step": 3050 }, { "epoch": 0.62, "grad_norm": 43.69202423095703, "learning_rate": 8.4515e-07, "loss": 1.8569, "step": 3100 }, { "epoch": 0.63, "grad_norm": 46.728538513183594, "learning_rate": 8.4265e-07, "loss": 1.8468, "step": 3150 }, { "epoch": 0.64, "grad_norm": 38.28073501586914, "learning_rate": 8.401499999999999e-07, "loss": 1.6194, "step": 3200 }, { "epoch": 0.65, "grad_norm": 48.87205123901367, "learning_rate": 8.376499999999999e-07, "loss": 1.7792, "step": 3250 }, { "epoch": 0.66, "grad_norm": 39.80657196044922, "learning_rate": 8.3515e-07, "loss": 1.7198, "step": 3300 }, { "epoch": 0.67, "grad_norm": 37.74007034301758, "learning_rate": 8.3265e-07, "loss": 1.7539, "step": 3350 }, { "epoch": 0.68, "grad_norm": 47.30875015258789, "learning_rate": 8.3015e-07, "loss": 1.725, "step": 3400 }, { "epoch": 0.69, "grad_norm": 44.54092025756836, "learning_rate": 8.276499999999999e-07, "loss": 1.7863, "step": 3450 }, { "epoch": 0.7, "grad_norm": 49.025169372558594, "learning_rate": 8.251500000000001e-07, "loss": 1.7914, "step": 3500 }, { "epoch": 0.71, "grad_norm": 45.531864166259766, "learning_rate": 8.2265e-07, "loss": 1.8326, "step": 3550 }, { "epoch": 0.72, "grad_norm": 37.38317108154297, "learning_rate": 8.2015e-07, "loss": 1.8113, "step": 3600 }, { "epoch": 0.73, "grad_norm": 42.720829010009766, "learning_rate": 8.176499999999999e-07, "loss": 1.7043, "step": 3650 }, { "epoch": 0.74, "grad_norm": 42.389339447021484, "learning_rate": 8.1515e-07, "loss": 1.6669, "step": 3700 }, { "epoch": 0.75, "grad_norm": 50.965545654296875, "learning_rate": 8.1265e-07, "loss": 1.6581, "step": 3750 }, { "epoch": 0.76, "grad_norm": 33.738948822021484, "learning_rate": 8.1015e-07, "loss": 1.7835, "step": 3800 }, { "epoch": 0.77, "grad_norm": 47.6729736328125, "learning_rate": 8.076499999999999e-07, "loss": 1.8784, "step": 3850 }, { "epoch": 0.78, "grad_norm": 40.22808837890625, "learning_rate": 8.0515e-07, "loss": 1.6733, "step": 3900 }, { "epoch": 0.79, "grad_norm": 43.46096420288086, "learning_rate": 8.0265e-07, "loss": 1.7327, "step": 3950 }, { "epoch": 0.8, "grad_norm": 39.8518180847168, "learning_rate": 8.0015e-07, "loss": 1.6765, "step": 4000 }, { "epoch": 0.81, "grad_norm": 43.99556350708008, "learning_rate": 7.976999999999999e-07, "loss": 1.6858, "step": 4050 }, { "epoch": 0.82, "grad_norm": 39.69054412841797, "learning_rate": 7.952e-07, "loss": 1.7666, "step": 4100 }, { "epoch": 0.83, "grad_norm": 37.67417907714844, "learning_rate": 7.926999999999999e-07, "loss": 1.7942, "step": 4150 }, { "epoch": 0.84, "grad_norm": 39.24260711669922, "learning_rate": 7.902e-07, "loss": 1.7449, "step": 4200 }, { "epoch": 0.85, "grad_norm": 44.0001220703125, "learning_rate": 7.876999999999999e-07, "loss": 1.7139, "step": 4250 }, { "epoch": 0.86, "grad_norm": 41.069236755371094, "learning_rate": 7.852e-07, "loss": 1.6626, "step": 4300 }, { "epoch": 0.87, "grad_norm": 40.02455139160156, "learning_rate": 7.826999999999999e-07, "loss": 1.6679, "step": 4350 }, { "epoch": 0.88, "grad_norm": 38.73442077636719, "learning_rate": 7.802e-07, "loss": 1.7561, "step": 4400 }, { "epoch": 0.89, "grad_norm": 44.03123474121094, "learning_rate": 7.776999999999999e-07, "loss": 1.7519, "step": 4450 }, { "epoch": 0.9, "grad_norm": 38.86705017089844, "learning_rate": 7.752e-07, "loss": 1.7051, "step": 4500 }, { "epoch": 0.91, "grad_norm": 42.8055534362793, "learning_rate": 7.727e-07, "loss": 1.7045, "step": 4550 }, { "epoch": 0.92, "grad_norm": 50.77599334716797, "learning_rate": 7.702e-07, "loss": 1.7706, "step": 4600 }, { "epoch": 0.93, "grad_norm": 39.55324935913086, "learning_rate": 7.677e-07, "loss": 1.7337, "step": 4650 }, { "epoch": 0.94, "grad_norm": 48.40109634399414, "learning_rate": 7.652e-07, "loss": 1.6519, "step": 4700 }, { "epoch": 0.95, "grad_norm": 47.08484649658203, "learning_rate": 7.627e-07, "loss": 1.7528, "step": 4750 }, { "epoch": 0.96, "grad_norm": 41.425846099853516, "learning_rate": 7.601999999999999e-07, "loss": 1.6663, "step": 4800 }, { "epoch": 0.97, "grad_norm": 42.3223991394043, "learning_rate": 7.577e-07, "loss": 1.6747, "step": 4850 }, { "epoch": 0.98, "grad_norm": 44.03189468383789, "learning_rate": 7.552e-07, "loss": 1.6487, "step": 4900 }, { "epoch": 0.99, "grad_norm": 41.170108795166016, "learning_rate": 7.527e-07, "loss": 1.7019, "step": 4950 }, { "epoch": 1.0, "grad_norm": 42.10016632080078, "learning_rate": 7.501999999999999e-07, "loss": 1.6696, "step": 5000 }, { "epoch": 1.01, "grad_norm": 34.27055358886719, "learning_rate": 7.477e-07, "loss": 1.62, "step": 5050 }, { "epoch": 1.02, "grad_norm": 41.19538116455078, "learning_rate": 7.452e-07, "loss": 1.6309, "step": 5100 }, { "epoch": 1.03, "grad_norm": 45.550575256347656, "learning_rate": 7.427e-07, "loss": 1.6535, "step": 5150 }, { "epoch": 1.04, "grad_norm": 45.184539794921875, "learning_rate": 7.401999999999999e-07, "loss": 1.7223, "step": 5200 }, { "epoch": 1.05, "grad_norm": 43.35379409790039, "learning_rate": 7.376999999999999e-07, "loss": 1.6844, "step": 5250 }, { "epoch": 1.06, "grad_norm": 40.49492645263672, "learning_rate": 7.352e-07, "loss": 1.689, "step": 5300 }, { "epoch": 1.07, "grad_norm": 38.820106506347656, "learning_rate": 7.327e-07, "loss": 1.7506, "step": 5350 }, { "epoch": 1.08, "grad_norm": 38.87533187866211, "learning_rate": 7.301999999999999e-07, "loss": 1.5689, "step": 5400 }, { "epoch": 1.09, "grad_norm": 34.46385192871094, "learning_rate": 7.276999999999999e-07, "loss": 1.6876, "step": 5450 }, { "epoch": 1.1, "grad_norm": 39.77639389038086, "learning_rate": 7.252e-07, "loss": 1.6515, "step": 5500 }, { "epoch": 1.11, "grad_norm": 42.202735900878906, "learning_rate": 7.227e-07, "loss": 1.5767, "step": 5550 }, { "epoch": 1.12, "grad_norm": 36.47328186035156, "learning_rate": 7.201999999999999e-07, "loss": 1.7376, "step": 5600 }, { "epoch": 1.13, "grad_norm": 40.54157257080078, "learning_rate": 7.176999999999999e-07, "loss": 1.768, "step": 5650 }, { "epoch": 1.1400000000000001, "grad_norm": 36.865177154541016, "learning_rate": 7.151999999999999e-07, "loss": 1.5431, "step": 5700 }, { "epoch": 1.15, "grad_norm": 32.530216217041016, "learning_rate": 7.127e-07, "loss": 1.6509, "step": 5750 }, { "epoch": 1.16, "grad_norm": 38.9181022644043, "learning_rate": 7.102e-07, "loss": 1.6686, "step": 5800 }, { "epoch": 1.17, "grad_norm": 47.71958541870117, "learning_rate": 7.076999999999999e-07, "loss": 1.6332, "step": 5850 }, { "epoch": 1.18, "grad_norm": 33.65346908569336, "learning_rate": 7.052e-07, "loss": 1.6374, "step": 5900 }, { "epoch": 1.19, "grad_norm": 50.740474700927734, "learning_rate": 7.027e-07, "loss": 1.6919, "step": 5950 }, { "epoch": 1.2, "grad_norm": 35.024147033691406, "learning_rate": 7.002e-07, "loss": 1.5119, "step": 6000 }, { "epoch": 1.21, "grad_norm": 46.42051315307617, "learning_rate": 6.977499999999999e-07, "loss": 1.7009, "step": 6050 }, { "epoch": 1.22, "grad_norm": 40.99652099609375, "learning_rate": 6.9525e-07, "loss": 1.671, "step": 6100 }, { "epoch": 1.23, "grad_norm": 48.398765563964844, "learning_rate": 6.9275e-07, "loss": 1.5888, "step": 6150 }, { "epoch": 1.24, "grad_norm": 37.039058685302734, "learning_rate": 6.9025e-07, "loss": 1.7995, "step": 6200 }, { "epoch": 1.25, "grad_norm": 39.58694839477539, "learning_rate": 6.877499999999999e-07, "loss": 1.6646, "step": 6250 }, { "epoch": 1.26, "grad_norm": 43.098045349121094, "learning_rate": 6.8525e-07, "loss": 1.7114, "step": 6300 }, { "epoch": 1.27, "grad_norm": 38.67097091674805, "learning_rate": 6.827499999999999e-07, "loss": 1.5854, "step": 6350 }, { "epoch": 1.28, "grad_norm": 43.86952590942383, "learning_rate": 6.8025e-07, "loss": 1.6068, "step": 6400 }, { "epoch": 1.29, "grad_norm": 43.158714294433594, "learning_rate": 6.777499999999999e-07, "loss": 1.7206, "step": 6450 }, { "epoch": 1.3, "grad_norm": 44.74959945678711, "learning_rate": 6.7525e-07, "loss": 1.656, "step": 6500 }, { "epoch": 1.31, "grad_norm": 34.64699172973633, "learning_rate": 6.727499999999999e-07, "loss": 1.6306, "step": 6550 }, { "epoch": 1.32, "grad_norm": 51.52064895629883, "learning_rate": 6.7025e-07, "loss": 1.7831, "step": 6600 }, { "epoch": 1.33, "grad_norm": 55.97758102416992, "learning_rate": 6.677499999999999e-07, "loss": 1.5366, "step": 6650 }, { "epoch": 1.34, "grad_norm": 40.172523498535156, "learning_rate": 6.6525e-07, "loss": 1.7331, "step": 6700 }, { "epoch": 1.35, "grad_norm": 42.77679443359375, "learning_rate": 6.627499999999999e-07, "loss": 1.6112, "step": 6750 }, { "epoch": 1.3599999999999999, "grad_norm": 36.01056671142578, "learning_rate": 6.602499999999999e-07, "loss": 1.5688, "step": 6800 }, { "epoch": 1.37, "grad_norm": 40.94023895263672, "learning_rate": 6.577499999999999e-07, "loss": 1.5867, "step": 6850 }, { "epoch": 1.38, "grad_norm": 45.53089141845703, "learning_rate": 6.5525e-07, "loss": 1.5666, "step": 6900 }, { "epoch": 1.3900000000000001, "grad_norm": 37.20089340209961, "learning_rate": 6.5275e-07, "loss": 1.5804, "step": 6950 }, { "epoch": 1.4, "grad_norm": 33.455230712890625, "learning_rate": 6.502499999999999e-07, "loss": 1.5933, "step": 7000 }, { "epoch": 1.41, "grad_norm": 39.31380844116211, "learning_rate": 6.477500000000001e-07, "loss": 1.6599, "step": 7050 }, { "epoch": 1.42, "grad_norm": 40.90291976928711, "learning_rate": 6.4525e-07, "loss": 1.7809, "step": 7100 }, { "epoch": 1.43, "grad_norm": 38.44248580932617, "learning_rate": 6.4275e-07, "loss": 1.554, "step": 7150 }, { "epoch": 1.44, "grad_norm": 41.40769958496094, "learning_rate": 6.402499999999999e-07, "loss": 1.5848, "step": 7200 }, { "epoch": 1.45, "grad_norm": 34.839202880859375, "learning_rate": 6.377500000000001e-07, "loss": 1.5506, "step": 7250 }, { "epoch": 1.46, "grad_norm": 39.07792663574219, "learning_rate": 6.3525e-07, "loss": 1.6936, "step": 7300 }, { "epoch": 1.47, "grad_norm": 47.81125259399414, "learning_rate": 6.3275e-07, "loss": 1.7948, "step": 7350 }, { "epoch": 1.48, "grad_norm": 33.074684143066406, "learning_rate": 6.302499999999999e-07, "loss": 1.5964, "step": 7400 }, { "epoch": 1.49, "grad_norm": 35.54746627807617, "learning_rate": 6.2775e-07, "loss": 1.5494, "step": 7450 }, { "epoch": 1.5, "grad_norm": 47.26594543457031, "learning_rate": 6.2525e-07, "loss": 1.6013, "step": 7500 }, { "epoch": 1.51, "grad_norm": 47.13039779663086, "learning_rate": 6.2275e-07, "loss": 1.5403, "step": 7550 }, { "epoch": 1.52, "grad_norm": 42.410614013671875, "learning_rate": 6.202499999999999e-07, "loss": 1.5956, "step": 7600 }, { "epoch": 1.53, "grad_norm": 38.69276428222656, "learning_rate": 6.1775e-07, "loss": 1.6829, "step": 7650 }, { "epoch": 1.54, "grad_norm": 41.635135650634766, "learning_rate": 6.1525e-07, "loss": 1.5228, "step": 7700 }, { "epoch": 1.55, "grad_norm": 38.301753997802734, "learning_rate": 6.1275e-07, "loss": 1.5039, "step": 7750 }, { "epoch": 1.56, "grad_norm": 40.455101013183594, "learning_rate": 6.102499999999999e-07, "loss": 1.5666, "step": 7800 }, { "epoch": 1.5699999999999998, "grad_norm": 49.50397491455078, "learning_rate": 6.0775e-07, "loss": 1.5666, "step": 7850 }, { "epoch": 1.58, "grad_norm": 51.95878601074219, "learning_rate": 6.052499999999999e-07, "loss": 1.5874, "step": 7900 }, { "epoch": 1.5899999999999999, "grad_norm": 36.6114501953125, "learning_rate": 6.0275e-07, "loss": 1.63, "step": 7950 }, { "epoch": 1.6, "grad_norm": 41.69790267944336, "learning_rate": 6.002499999999999e-07, "loss": 1.5995, "step": 8000 }, { "epoch": 1.6099999999999999, "grad_norm": 38.66043472290039, "learning_rate": 5.978e-07, "loss": 1.5629, "step": 8050 }, { "epoch": 1.62, "grad_norm": 59.15949249267578, "learning_rate": 5.953e-07, "loss": 1.6076, "step": 8100 }, { "epoch": 1.63, "grad_norm": 37.75910186767578, "learning_rate": 5.928e-07, "loss": 1.5531, "step": 8150 }, { "epoch": 1.6400000000000001, "grad_norm": 40.374855041503906, "learning_rate": 5.903e-07, "loss": 1.5393, "step": 8200 }, { "epoch": 1.65, "grad_norm": 47.41810989379883, "learning_rate": 5.878e-07, "loss": 1.5062, "step": 8250 }, { "epoch": 1.6600000000000001, "grad_norm": 46.51014709472656, "learning_rate": 5.853e-07, "loss": 1.6144, "step": 8300 }, { "epoch": 1.67, "grad_norm": 49.626834869384766, "learning_rate": 5.828e-07, "loss": 1.6439, "step": 8350 }, { "epoch": 1.6800000000000002, "grad_norm": 32.59950256347656, "learning_rate": 5.803e-07, "loss": 1.5362, "step": 8400 }, { "epoch": 1.69, "grad_norm": 49.02924346923828, "learning_rate": 5.778e-07, "loss": 1.7244, "step": 8450 }, { "epoch": 1.7, "grad_norm": 41.86074447631836, "learning_rate": 5.753e-07, "loss": 1.5701, "step": 8500 }, { "epoch": 1.71, "grad_norm": 30.711210250854492, "learning_rate": 5.727999999999999e-07, "loss": 1.6219, "step": 8550 }, { "epoch": 1.72, "grad_norm": 47.8115234375, "learning_rate": 5.703e-07, "loss": 1.5723, "step": 8600 }, { "epoch": 1.73, "grad_norm": 39.29584884643555, "learning_rate": 5.678e-07, "loss": 1.5869, "step": 8650 }, { "epoch": 1.74, "grad_norm": 31.378740310668945, "learning_rate": 5.653e-07, "loss": 1.635, "step": 8700 }, { "epoch": 1.75, "grad_norm": 38.850608825683594, "learning_rate": 5.627999999999999e-07, "loss": 1.6046, "step": 8750 }, { "epoch": 1.76, "grad_norm": 26.495405197143555, "learning_rate": 5.603e-07, "loss": 1.6276, "step": 8800 }, { "epoch": 1.77, "grad_norm": 44.871849060058594, "learning_rate": 5.578e-07, "loss": 1.6889, "step": 8850 }, { "epoch": 1.78, "grad_norm": 47.68773651123047, "learning_rate": 5.553e-07, "loss": 1.7151, "step": 8900 }, { "epoch": 1.79, "grad_norm": 31.6007080078125, "learning_rate": 5.527999999999999e-07, "loss": 1.5286, "step": 8950 }, { "epoch": 1.8, "grad_norm": 30.19587516784668, "learning_rate": 5.502999999999999e-07, "loss": 1.6597, "step": 9000 }, { "epoch": 1.81, "grad_norm": 36.03053665161133, "learning_rate": 5.478e-07, "loss": 1.6422, "step": 9050 }, { "epoch": 1.8199999999999998, "grad_norm": 41.28793716430664, "learning_rate": 5.453e-07, "loss": 1.6893, "step": 9100 }, { "epoch": 1.83, "grad_norm": 32.481689453125, "learning_rate": 5.427999999999999e-07, "loss": 1.5683, "step": 9150 }, { "epoch": 1.8399999999999999, "grad_norm": 40.114654541015625, "learning_rate": 5.402999999999999e-07, "loss": 1.6021, "step": 9200 }, { "epoch": 1.85, "grad_norm": 44.78034210205078, "learning_rate": 5.378e-07, "loss": 1.6241, "step": 9250 }, { "epoch": 1.8599999999999999, "grad_norm": 32.96756362915039, "learning_rate": 5.353e-07, "loss": 1.4664, "step": 9300 }, { "epoch": 1.87, "grad_norm": 35.44109344482422, "learning_rate": 5.328e-07, "loss": 1.6045, "step": 9350 }, { "epoch": 1.88, "grad_norm": 47.10539245605469, "learning_rate": 5.302999999999999e-07, "loss": 1.6529, "step": 9400 }, { "epoch": 1.8900000000000001, "grad_norm": 35.667083740234375, "learning_rate": 5.278000000000001e-07, "loss": 1.6878, "step": 9450 }, { "epoch": 1.9, "grad_norm": 42.152652740478516, "learning_rate": 5.253e-07, "loss": 1.6694, "step": 9500 }, { "epoch": 1.9100000000000001, "grad_norm": 44.23321533203125, "learning_rate": 5.228e-07, "loss": 1.61, "step": 9550 }, { "epoch": 1.92, "grad_norm": 37.32196807861328, "learning_rate": 5.202999999999999e-07, "loss": 1.5521, "step": 9600 }, { "epoch": 1.9300000000000002, "grad_norm": 51.841514587402344, "learning_rate": 5.178e-07, "loss": 1.682, "step": 9650 }, { "epoch": 1.94, "grad_norm": 41.69253158569336, "learning_rate": 5.153e-07, "loss": 1.6228, "step": 9700 }, { "epoch": 1.95, "grad_norm": 32.16023635864258, "learning_rate": 5.128e-07, "loss": 1.6022, "step": 9750 }, { "epoch": 1.96, "grad_norm": 40.430206298828125, "learning_rate": 5.102999999999999e-07, "loss": 1.6262, "step": 9800 }, { "epoch": 1.97, "grad_norm": 47.571449279785156, "learning_rate": 5.078e-07, "loss": 1.6431, "step": 9850 }, { "epoch": 1.98, "grad_norm": 40.32009506225586, "learning_rate": 5.053e-07, "loss": 1.4864, "step": 9900 }, { "epoch": 1.99, "grad_norm": 40.59520721435547, "learning_rate": 5.028e-07, "loss": 1.595, "step": 9950 }, { "epoch": 2.0, "grad_norm": 44.62666702270508, "learning_rate": 5.002999999999999e-07, "loss": 1.5604, "step": 10000 }, { "epoch": 2.01, "grad_norm": 37.61200714111328, "learning_rate": 4.9785e-07, "loss": 1.6435, "step": 10050 }, { "epoch": 2.02, "grad_norm": 40.34099197387695, "learning_rate": 4.953499999999999e-07, "loss": 1.7094, "step": 10100 }, { "epoch": 2.03, "grad_norm": 47.07732009887695, "learning_rate": 4.9285e-07, "loss": 1.5482, "step": 10150 }, { "epoch": 2.04, "grad_norm": 23.103185653686523, "learning_rate": 4.9035e-07, "loss": 1.6298, "step": 10200 }, { "epoch": 2.05, "grad_norm": 44.33336639404297, "learning_rate": 4.8785e-07, "loss": 1.5512, "step": 10250 }, { "epoch": 2.06, "grad_norm": 41.70211410522461, "learning_rate": 4.853499999999999e-07, "loss": 1.4399, "step": 10300 }, { "epoch": 2.07, "grad_norm": 36.45768737792969, "learning_rate": 4.8285e-07, "loss": 1.5103, "step": 10350 }, { "epoch": 2.08, "grad_norm": 30.36566162109375, "learning_rate": 4.8035e-07, "loss": 1.4481, "step": 10400 }, { "epoch": 2.09, "grad_norm": 35.793094635009766, "learning_rate": 4.7785e-07, "loss": 1.5339, "step": 10450 }, { "epoch": 2.1, "grad_norm": 52.48720932006836, "learning_rate": 4.7535e-07, "loss": 1.6079, "step": 10500 }, { "epoch": 2.11, "grad_norm": 41.160335540771484, "learning_rate": 4.7284999999999995e-07, "loss": 1.4992, "step": 10550 }, { "epoch": 2.12, "grad_norm": 40.25874710083008, "learning_rate": 4.7034999999999997e-07, "loss": 1.601, "step": 10600 }, { "epoch": 2.13, "grad_norm": 35.40253829956055, "learning_rate": 4.6784999999999994e-07, "loss": 1.5437, "step": 10650 }, { "epoch": 2.14, "grad_norm": 42.93209457397461, "learning_rate": 4.6534999999999997e-07, "loss": 1.5923, "step": 10700 }, { "epoch": 2.15, "grad_norm": 43.273956298828125, "learning_rate": 4.6284999999999994e-07, "loss": 1.5925, "step": 10750 }, { "epoch": 2.16, "grad_norm": 49.30186080932617, "learning_rate": 4.6034999999999996e-07, "loss": 1.538, "step": 10800 }, { "epoch": 2.17, "grad_norm": 40.10365295410156, "learning_rate": 4.5784999999999993e-07, "loss": 1.545, "step": 10850 }, { "epoch": 2.18, "grad_norm": 40.76726531982422, "learning_rate": 4.5534999999999996e-07, "loss": 1.4837, "step": 10900 }, { "epoch": 2.19, "grad_norm": 36.3805046081543, "learning_rate": 4.5284999999999993e-07, "loss": 1.5463, "step": 10950 }, { "epoch": 2.2, "grad_norm": 25.484167098999023, "learning_rate": 4.5034999999999995e-07, "loss": 1.5916, "step": 11000 }, { "epoch": 2.21, "grad_norm": 33.473121643066406, "learning_rate": 4.4785000000000003e-07, "loss": 1.5033, "step": 11050 }, { "epoch": 2.22, "grad_norm": 36.8259391784668, "learning_rate": 4.4535e-07, "loss": 1.552, "step": 11100 }, { "epoch": 2.23, "grad_norm": 32.61063766479492, "learning_rate": 4.4285e-07, "loss": 1.5759, "step": 11150 }, { "epoch": 2.24, "grad_norm": 51.49959945678711, "learning_rate": 4.4035e-07, "loss": 1.556, "step": 11200 }, { "epoch": 2.25, "grad_norm": 44.21879959106445, "learning_rate": 4.3785e-07, "loss": 1.4935, "step": 11250 }, { "epoch": 2.26, "grad_norm": 36.884010314941406, "learning_rate": 4.3535e-07, "loss": 1.4284, "step": 11300 }, { "epoch": 2.27, "grad_norm": 38.094329833984375, "learning_rate": 4.3285e-07, "loss": 1.6868, "step": 11350 }, { "epoch": 2.2800000000000002, "grad_norm": 41.53178024291992, "learning_rate": 4.3035e-07, "loss": 1.6034, "step": 11400 }, { "epoch": 2.29, "grad_norm": 46.737213134765625, "learning_rate": 4.2785e-07, "loss": 1.5557, "step": 11450 }, { "epoch": 2.3, "grad_norm": 50.67173385620117, "learning_rate": 4.2535e-07, "loss": 1.6998, "step": 11500 }, { "epoch": 2.31, "grad_norm": 52.222557067871094, "learning_rate": 4.2285e-07, "loss": 1.5495, "step": 11550 }, { "epoch": 2.32, "grad_norm": 48.568050384521484, "learning_rate": 4.2035e-07, "loss": 1.5441, "step": 11600 }, { "epoch": 2.33, "grad_norm": 55.280521392822266, "learning_rate": 4.1785e-07, "loss": 1.68, "step": 11650 }, { "epoch": 2.34, "grad_norm": 42.30193328857422, "learning_rate": 4.1534999999999997e-07, "loss": 1.6276, "step": 11700 }, { "epoch": 2.35, "grad_norm": 54.417327880859375, "learning_rate": 4.1285e-07, "loss": 1.5603, "step": 11750 }, { "epoch": 2.36, "grad_norm": 33.117340087890625, "learning_rate": 4.1034999999999997e-07, "loss": 1.5169, "step": 11800 }, { "epoch": 2.37, "grad_norm": 43.10250473022461, "learning_rate": 4.0785e-07, "loss": 1.4491, "step": 11850 }, { "epoch": 2.38, "grad_norm": 48.663917541503906, "learning_rate": 4.0534999999999996e-07, "loss": 1.5617, "step": 11900 }, { "epoch": 2.39, "grad_norm": 41.02192687988281, "learning_rate": 4.0285e-07, "loss": 1.6111, "step": 11950 }, { "epoch": 2.4, "grad_norm": 42.699256896972656, "learning_rate": 4.0034999999999996e-07, "loss": 1.4217, "step": 12000 }, { "epoch": 2.41, "grad_norm": 49.665855407714844, "learning_rate": 3.979e-07, "loss": 1.4223, "step": 12050 }, { "epoch": 2.42, "grad_norm": 43.366397857666016, "learning_rate": 3.9539999999999995e-07, "loss": 1.5452, "step": 12100 }, { "epoch": 2.43, "grad_norm": 35.55706024169922, "learning_rate": 3.9290000000000003e-07, "loss": 1.545, "step": 12150 }, { "epoch": 2.44, "grad_norm": 42.98080062866211, "learning_rate": 3.904e-07, "loss": 1.5648, "step": 12200 }, { "epoch": 2.45, "grad_norm": 36.39263916015625, "learning_rate": 3.879e-07, "loss": 1.5724, "step": 12250 }, { "epoch": 2.46, "grad_norm": 42.547401428222656, "learning_rate": 3.854e-07, "loss": 1.7179, "step": 12300 }, { "epoch": 2.4699999999999998, "grad_norm": 45.591087341308594, "learning_rate": 3.829e-07, "loss": 1.5122, "step": 12350 }, { "epoch": 2.48, "grad_norm": 43.481258392333984, "learning_rate": 3.804e-07, "loss": 1.4843, "step": 12400 }, { "epoch": 2.49, "grad_norm": 44.69862747192383, "learning_rate": 3.779e-07, "loss": 1.5978, "step": 12450 }, { "epoch": 2.5, "grad_norm": 43.04928970336914, "learning_rate": 3.754e-07, "loss": 1.512, "step": 12500 }, { "epoch": 2.51, "grad_norm": 35.41480255126953, "learning_rate": 3.729e-07, "loss": 1.6957, "step": 12550 }, { "epoch": 2.52, "grad_norm": 42.25669860839844, "learning_rate": 3.704e-07, "loss": 1.4797, "step": 12600 }, { "epoch": 2.5300000000000002, "grad_norm": 23.68979835510254, "learning_rate": 3.679e-07, "loss": 1.515, "step": 12650 }, { "epoch": 2.54, "grad_norm": 39.010196685791016, "learning_rate": 3.654e-07, "loss": 1.5484, "step": 12700 }, { "epoch": 2.55, "grad_norm": 40.15721130371094, "learning_rate": 3.629e-07, "loss": 1.595, "step": 12750 }, { "epoch": 2.56, "grad_norm": 32.100643157958984, "learning_rate": 3.6039999999999997e-07, "loss": 1.5125, "step": 12800 }, { "epoch": 2.57, "grad_norm": 51.687583923339844, "learning_rate": 3.579e-07, "loss": 1.6924, "step": 12850 }, { "epoch": 2.58, "grad_norm": 40.378807067871094, "learning_rate": 3.5539999999999997e-07, "loss": 1.5349, "step": 12900 }, { "epoch": 2.59, "grad_norm": 46.35411834716797, "learning_rate": 3.529e-07, "loss": 1.4451, "step": 12950 }, { "epoch": 2.6, "grad_norm": 39.091644287109375, "learning_rate": 3.5039999999999996e-07, "loss": 1.5871, "step": 13000 }, { "epoch": 2.61, "grad_norm": 46.588539123535156, "learning_rate": 3.479e-07, "loss": 1.5514, "step": 13050 }, { "epoch": 2.62, "grad_norm": 40.58433151245117, "learning_rate": 3.4539999999999996e-07, "loss": 1.6805, "step": 13100 }, { "epoch": 2.63, "grad_norm": 44.80520248413086, "learning_rate": 3.429e-07, "loss": 1.4672, "step": 13150 }, { "epoch": 2.64, "grad_norm": 42.662078857421875, "learning_rate": 3.4039999999999995e-07, "loss": 1.5712, "step": 13200 }, { "epoch": 2.65, "grad_norm": 40.33576583862305, "learning_rate": 3.379e-07, "loss": 1.4617, "step": 13250 }, { "epoch": 2.66, "grad_norm": 44.45260238647461, "learning_rate": 3.3539999999999995e-07, "loss": 1.5336, "step": 13300 }, { "epoch": 2.67, "grad_norm": 48.2208251953125, "learning_rate": 3.3289999999999997e-07, "loss": 1.5241, "step": 13350 }, { "epoch": 2.68, "grad_norm": 37.904762268066406, "learning_rate": 3.304e-07, "loss": 1.5679, "step": 13400 }, { "epoch": 2.69, "grad_norm": 39.32900619506836, "learning_rate": 3.279e-07, "loss": 1.5121, "step": 13450 }, { "epoch": 2.7, "grad_norm": 36.95504379272461, "learning_rate": 3.254e-07, "loss": 1.4855, "step": 13500 }, { "epoch": 2.71, "grad_norm": 29.145931243896484, "learning_rate": 3.229e-07, "loss": 1.5645, "step": 13550 }, { "epoch": 2.7199999999999998, "grad_norm": 41.01043701171875, "learning_rate": 3.204e-07, "loss": 1.4902, "step": 13600 }, { "epoch": 2.73, "grad_norm": 42.938472747802734, "learning_rate": 3.179e-07, "loss": 1.421, "step": 13650 }, { "epoch": 2.74, "grad_norm": 42.611812591552734, "learning_rate": 3.154e-07, "loss": 1.6238, "step": 13700 }, { "epoch": 2.75, "grad_norm": 40.880706787109375, "learning_rate": 3.129e-07, "loss": 1.554, "step": 13750 }, { "epoch": 2.76, "grad_norm": 44.605411529541016, "learning_rate": 3.104e-07, "loss": 1.631, "step": 13800 }, { "epoch": 2.77, "grad_norm": 35.732723236083984, "learning_rate": 3.079e-07, "loss": 1.5479, "step": 13850 }, { "epoch": 2.7800000000000002, "grad_norm": 38.93905258178711, "learning_rate": 3.0539999999999997e-07, "loss": 1.5417, "step": 13900 }, { "epoch": 2.79, "grad_norm": 50.35736083984375, "learning_rate": 3.029e-07, "loss": 1.6767, "step": 13950 }, { "epoch": 2.8, "grad_norm": 35.24892807006836, "learning_rate": 3.0039999999999996e-07, "loss": 1.5639, "step": 14000 }, { "epoch": 2.81, "grad_norm": 41.46996307373047, "learning_rate": 2.9795e-07, "loss": 1.4937, "step": 14050 }, { "epoch": 2.82, "grad_norm": 39.519866943359375, "learning_rate": 2.9544999999999996e-07, "loss": 1.6308, "step": 14100 }, { "epoch": 2.83, "grad_norm": 44.51643753051758, "learning_rate": 2.9295e-07, "loss": 1.4379, "step": 14150 }, { "epoch": 2.84, "grad_norm": 35.87133026123047, "learning_rate": 2.9044999999999996e-07, "loss": 1.4668, "step": 14200 }, { "epoch": 2.85, "grad_norm": 45.94535446166992, "learning_rate": 2.8795e-07, "loss": 1.6009, "step": 14250 }, { "epoch": 2.86, "grad_norm": 44.09996032714844, "learning_rate": 2.8544999999999995e-07, "loss": 1.5932, "step": 14300 }, { "epoch": 2.87, "grad_norm": 48.415313720703125, "learning_rate": 2.8295e-07, "loss": 1.5851, "step": 14350 }, { "epoch": 2.88, "grad_norm": 33.39976501464844, "learning_rate": 2.8044999999999995e-07, "loss": 1.6999, "step": 14400 }, { "epoch": 2.89, "grad_norm": 40.66514587402344, "learning_rate": 2.7794999999999997e-07, "loss": 1.4782, "step": 14450 }, { "epoch": 2.9, "grad_norm": 45.65864944458008, "learning_rate": 2.7544999999999994e-07, "loss": 1.6338, "step": 14500 }, { "epoch": 2.91, "grad_norm": 39.462677001953125, "learning_rate": 2.7295e-07, "loss": 1.5968, "step": 14550 }, { "epoch": 2.92, "grad_norm": 41.13777542114258, "learning_rate": 2.7045e-07, "loss": 1.4837, "step": 14600 }, { "epoch": 2.93, "grad_norm": 38.25054168701172, "learning_rate": 2.6795e-07, "loss": 1.5727, "step": 14650 }, { "epoch": 2.94, "grad_norm": 37.76020431518555, "learning_rate": 2.6545e-07, "loss": 1.642, "step": 14700 }, { "epoch": 2.95, "grad_norm": 50.80886459350586, "learning_rate": 2.6295e-07, "loss": 1.605, "step": 14750 }, { "epoch": 2.96, "grad_norm": 52.12263870239258, "learning_rate": 2.6045e-07, "loss": 1.5843, "step": 14800 }, { "epoch": 2.9699999999999998, "grad_norm": 40.298282623291016, "learning_rate": 2.5795e-07, "loss": 1.4729, "step": 14850 }, { "epoch": 2.98, "grad_norm": 30.732044219970703, "learning_rate": 2.5545e-07, "loss": 1.5218, "step": 14900 }, { "epoch": 2.99, "grad_norm": 43.94182586669922, "learning_rate": 2.5295e-07, "loss": 1.5147, "step": 14950 }, { "epoch": 3.0, "grad_norm": 60.4287223815918, "learning_rate": 2.5044999999999997e-07, "loss": 1.446, "step": 15000 }, { "epoch": 3.01, "grad_norm": 35.2944221496582, "learning_rate": 2.4795e-07, "loss": 1.5543, "step": 15050 }, { "epoch": 3.02, "grad_norm": 53.06623077392578, "learning_rate": 2.4544999999999996e-07, "loss": 1.5674, "step": 15100 }, { "epoch": 3.03, "grad_norm": 56.413692474365234, "learning_rate": 2.4295e-07, "loss": 1.4601, "step": 15150 }, { "epoch": 3.04, "grad_norm": 44.59477233886719, "learning_rate": 2.4045e-07, "loss": 1.5439, "step": 15200 }, { "epoch": 3.05, "grad_norm": 46.42093276977539, "learning_rate": 2.3794999999999998e-07, "loss": 1.4917, "step": 15250 }, { "epoch": 3.06, "grad_norm": 31.83967399597168, "learning_rate": 2.3544999999999998e-07, "loss": 1.5199, "step": 15300 }, { "epoch": 3.07, "grad_norm": 52.99776077270508, "learning_rate": 2.3294999999999998e-07, "loss": 1.509, "step": 15350 }, { "epoch": 3.08, "grad_norm": 47.901885986328125, "learning_rate": 2.3044999999999998e-07, "loss": 1.4072, "step": 15400 }, { "epoch": 3.09, "grad_norm": 35.03591537475586, "learning_rate": 2.2795e-07, "loss": 1.4489, "step": 15450 }, { "epoch": 3.1, "grad_norm": 40.465789794921875, "learning_rate": 2.2545e-07, "loss": 1.4475, "step": 15500 }, { "epoch": 3.11, "grad_norm": 38.67084503173828, "learning_rate": 2.2295e-07, "loss": 1.4501, "step": 15550 }, { "epoch": 3.12, "grad_norm": 44.53215026855469, "learning_rate": 2.2045e-07, "loss": 1.5509, "step": 15600 }, { "epoch": 3.13, "grad_norm": 50.45119857788086, "learning_rate": 2.1795e-07, "loss": 1.5862, "step": 15650 }, { "epoch": 3.14, "grad_norm": 42.37192153930664, "learning_rate": 2.1545e-07, "loss": 1.6511, "step": 15700 }, { "epoch": 3.15, "grad_norm": 39.11933898925781, "learning_rate": 2.1294999999999998e-07, "loss": 1.5206, "step": 15750 }, { "epoch": 3.16, "grad_norm": 39.057823181152344, "learning_rate": 2.1044999999999998e-07, "loss": 1.4265, "step": 15800 }, { "epoch": 3.17, "grad_norm": 32.509098052978516, "learning_rate": 2.0794999999999998e-07, "loss": 1.5199, "step": 15850 }, { "epoch": 3.18, "grad_norm": 34.1364860534668, "learning_rate": 2.0544999999999998e-07, "loss": 1.5162, "step": 15900 }, { "epoch": 3.19, "grad_norm": 48.204994201660156, "learning_rate": 2.0294999999999997e-07, "loss": 1.4383, "step": 15950 }, { "epoch": 3.2, "grad_norm": 53.505615234375, "learning_rate": 2.0044999999999997e-07, "loss": 1.5337, "step": 16000 }, { "epoch": 3.21, "grad_norm": 48.69506072998047, "learning_rate": 1.98e-07, "loss": 1.4313, "step": 16050 }, { "epoch": 3.22, "grad_norm": 45.98126220703125, "learning_rate": 1.955e-07, "loss": 1.5255, "step": 16100 }, { "epoch": 3.23, "grad_norm": 39.528587341308594, "learning_rate": 1.93e-07, "loss": 1.5407, "step": 16150 }, { "epoch": 3.24, "grad_norm": 41.41926956176758, "learning_rate": 1.905e-07, "loss": 1.4167, "step": 16200 }, { "epoch": 3.25, "grad_norm": 50.91017150878906, "learning_rate": 1.88e-07, "loss": 1.5819, "step": 16250 }, { "epoch": 3.26, "grad_norm": 51.679622650146484, "learning_rate": 1.855e-07, "loss": 1.573, "step": 16300 }, { "epoch": 3.27, "grad_norm": 39.97464370727539, "learning_rate": 1.8299999999999998e-07, "loss": 1.4209, "step": 16350 }, { "epoch": 3.2800000000000002, "grad_norm": 37.9398307800293, "learning_rate": 1.8049999999999998e-07, "loss": 1.4962, "step": 16400 }, { "epoch": 3.29, "grad_norm": 41.62467956542969, "learning_rate": 1.7799999999999998e-07, "loss": 1.5011, "step": 16450 }, { "epoch": 3.3, "grad_norm": 39.8488655090332, "learning_rate": 1.7549999999999998e-07, "loss": 1.5504, "step": 16500 }, { "epoch": 3.31, "grad_norm": 39.093360900878906, "learning_rate": 1.7299999999999997e-07, "loss": 1.4504, "step": 16550 }, { "epoch": 3.32, "grad_norm": 46.60807800292969, "learning_rate": 1.705e-07, "loss": 1.5712, "step": 16600 }, { "epoch": 3.33, "grad_norm": 37.40671920776367, "learning_rate": 1.68e-07, "loss": 1.5793, "step": 16650 }, { "epoch": 3.34, "grad_norm": 45.785953521728516, "learning_rate": 1.655e-07, "loss": 1.4975, "step": 16700 }, { "epoch": 3.35, "grad_norm": 42.13127517700195, "learning_rate": 1.63e-07, "loss": 1.5418, "step": 16750 }, { "epoch": 3.36, "grad_norm": 51.313777923583984, "learning_rate": 1.605e-07, "loss": 1.6146, "step": 16800 }, { "epoch": 3.37, "grad_norm": 38.90798568725586, "learning_rate": 1.5799999999999999e-07, "loss": 1.5328, "step": 16850 }, { "epoch": 3.38, "grad_norm": 42.80254364013672, "learning_rate": 1.5549999999999998e-07, "loss": 1.4382, "step": 16900 }, { "epoch": 3.39, "grad_norm": 38.583885192871094, "learning_rate": 1.5299999999999998e-07, "loss": 1.421, "step": 16950 }, { "epoch": 3.4, "grad_norm": 45.506473541259766, "learning_rate": 1.5049999999999998e-07, "loss": 1.4643, "step": 17000 }, { "epoch": 3.41, "grad_norm": 37.714202880859375, "learning_rate": 1.4799999999999998e-07, "loss": 1.4807, "step": 17050 }, { "epoch": 3.42, "grad_norm": 52.356605529785156, "learning_rate": 1.4549999999999997e-07, "loss": 1.4663, "step": 17100 }, { "epoch": 3.43, "grad_norm": 40.869590759277344, "learning_rate": 1.4299999999999997e-07, "loss": 1.542, "step": 17150 }, { "epoch": 3.44, "grad_norm": 55.49686813354492, "learning_rate": 1.4050000000000002e-07, "loss": 1.5719, "step": 17200 }, { "epoch": 3.45, "grad_norm": 41.89445114135742, "learning_rate": 1.3800000000000002e-07, "loss": 1.4327, "step": 17250 }, { "epoch": 3.46, "grad_norm": 41.772361755371094, "learning_rate": 1.3550000000000002e-07, "loss": 1.5065, "step": 17300 }, { "epoch": 3.4699999999999998, "grad_norm": 45.570396423339844, "learning_rate": 1.33e-07, "loss": 1.4851, "step": 17350 }, { "epoch": 3.48, "grad_norm": 45.119693756103516, "learning_rate": 1.305e-07, "loss": 1.4926, "step": 17400 }, { "epoch": 3.49, "grad_norm": 39.28087615966797, "learning_rate": 1.28e-07, "loss": 1.5135, "step": 17450 }, { "epoch": 3.5, "grad_norm": 49.64841079711914, "learning_rate": 1.255e-07, "loss": 1.5662, "step": 17500 }, { "epoch": 3.51, "grad_norm": 39.4202995300293, "learning_rate": 1.23e-07, "loss": 1.5854, "step": 17550 }, { "epoch": 3.52, "grad_norm": 39.55594253540039, "learning_rate": 1.205e-07, "loss": 1.5577, "step": 17600 }, { "epoch": 3.5300000000000002, "grad_norm": 38.05757522583008, "learning_rate": 1.1799999999999998e-07, "loss": 1.4087, "step": 17650 }, { "epoch": 3.54, "grad_norm": 35.86211013793945, "learning_rate": 1.155e-07, "loss": 1.4881, "step": 17700 }, { "epoch": 3.55, "grad_norm": 39.4666862487793, "learning_rate": 1.1299999999999999e-07, "loss": 1.5927, "step": 17750 }, { "epoch": 3.56, "grad_norm": 42.395320892333984, "learning_rate": 1.1049999999999999e-07, "loss": 1.5176, "step": 17800 }, { "epoch": 3.57, "grad_norm": 34.1979866027832, "learning_rate": 1.0799999999999999e-07, "loss": 1.545, "step": 17850 }, { "epoch": 3.58, "grad_norm": 41.41266632080078, "learning_rate": 1.0549999999999999e-07, "loss": 1.5388, "step": 17900 }, { "epoch": 3.59, "grad_norm": 40.88737869262695, "learning_rate": 1.03e-07, "loss": 1.4851, "step": 17950 }, { "epoch": 3.6, "grad_norm": 44.86513137817383, "learning_rate": 1.005e-07, "loss": 1.5413, "step": 18000 }, { "epoch": 3.61, "grad_norm": 38.92195510864258, "learning_rate": 9.805e-08, "loss": 1.5463, "step": 18050 }, { "epoch": 3.62, "grad_norm": 46.53792953491211, "learning_rate": 9.554999999999999e-08, "loss": 1.5022, "step": 18100 }, { "epoch": 3.63, "grad_norm": 35.3355598449707, "learning_rate": 9.304999999999999e-08, "loss": 1.5245, "step": 18150 }, { "epoch": 3.64, "grad_norm": 55.84636306762695, "learning_rate": 9.055e-08, "loss": 1.6236, "step": 18200 }, { "epoch": 3.65, "grad_norm": 36.84891128540039, "learning_rate": 8.805e-08, "loss": 1.3286, "step": 18250 }, { "epoch": 3.66, "grad_norm": 51.336795806884766, "learning_rate": 8.555e-08, "loss": 1.5488, "step": 18300 }, { "epoch": 3.67, "grad_norm": 38.55815887451172, "learning_rate": 8.304999999999999e-08, "loss": 1.4718, "step": 18350 }, { "epoch": 3.68, "grad_norm": 47.52061462402344, "learning_rate": 8.054999999999999e-08, "loss": 1.5148, "step": 18400 }, { "epoch": 3.69, "grad_norm": 42.93059539794922, "learning_rate": 7.804999999999999e-08, "loss": 1.5243, "step": 18450 }, { "epoch": 3.7, "grad_norm": 40.72081756591797, "learning_rate": 7.555e-08, "loss": 1.4389, "step": 18500 }, { "epoch": 3.71, "grad_norm": 42.075836181640625, "learning_rate": 7.305e-08, "loss": 1.5487, "step": 18550 }, { "epoch": 3.7199999999999998, "grad_norm": 50.6606559753418, "learning_rate": 7.054999999999999e-08, "loss": 1.6045, "step": 18600 }, { "epoch": 3.73, "grad_norm": 39.99626541137695, "learning_rate": 6.804999999999999e-08, "loss": 1.5214, "step": 18650 }, { "epoch": 3.74, "grad_norm": 48.29393005371094, "learning_rate": 6.554999999999999e-08, "loss": 1.471, "step": 18700 }, { "epoch": 3.75, "grad_norm": 41.546470642089844, "learning_rate": 6.304999999999999e-08, "loss": 1.4242, "step": 18750 }, { "epoch": 3.76, "grad_norm": 43.29277420043945, "learning_rate": 6.055e-08, "loss": 1.3872, "step": 18800 }, { "epoch": 3.77, "grad_norm": 60.80516052246094, "learning_rate": 5.8049999999999994e-08, "loss": 1.5309, "step": 18850 }, { "epoch": 3.7800000000000002, "grad_norm": 37.9664421081543, "learning_rate": 5.555e-08, "loss": 1.4366, "step": 18900 }, { "epoch": 3.79, "grad_norm": 41.17642593383789, "learning_rate": 5.3049999999999995e-08, "loss": 1.5797, "step": 18950 }, { "epoch": 3.8, "grad_norm": 36.21015167236328, "learning_rate": 5.054999999999999e-08, "loss": 1.5096, "step": 19000 }, { "epoch": 3.81, "grad_norm": 47.9669075012207, "learning_rate": 4.8050000000000003e-08, "loss": 1.5302, "step": 19050 }, { "epoch": 3.82, "grad_norm": 28.955060958862305, "learning_rate": 4.555e-08, "loss": 1.498, "step": 19100 }, { "epoch": 3.83, "grad_norm": 52.464290618896484, "learning_rate": 4.305e-08, "loss": 1.4575, "step": 19150 }, { "epoch": 3.84, "grad_norm": 29.458721160888672, "learning_rate": 4.055e-08, "loss": 1.484, "step": 19200 }, { "epoch": 3.85, "grad_norm": 41.32457733154297, "learning_rate": 3.805e-08, "loss": 1.4677, "step": 19250 }, { "epoch": 3.86, "grad_norm": 48.74892807006836, "learning_rate": 3.555e-08, "loss": 1.4268, "step": 19300 }, { "epoch": 3.87, "grad_norm": 30.619657516479492, "learning_rate": 3.305e-08, "loss": 1.5416, "step": 19350 }, { "epoch": 3.88, "grad_norm": 48.18694305419922, "learning_rate": 3.055e-08, "loss": 1.4068, "step": 19400 }, { "epoch": 3.89, "grad_norm": 35.75336456298828, "learning_rate": 2.8049999999999996e-08, "loss": 1.3868, "step": 19450 }, { "epoch": 3.9, "grad_norm": 46.65244674682617, "learning_rate": 2.5549999999999997e-08, "loss": 1.5598, "step": 19500 }, { "epoch": 3.91, "grad_norm": 48.187679290771484, "learning_rate": 2.305e-08, "loss": 1.5666, "step": 19550 }, { "epoch": 3.92, "grad_norm": 36.0362663269043, "learning_rate": 2.055e-08, "loss": 1.6636, "step": 19600 }, { "epoch": 3.93, "grad_norm": 55.22818374633789, "learning_rate": 1.805e-08, "loss": 1.5572, "step": 19650 }, { "epoch": 3.94, "grad_norm": 32.651451110839844, "learning_rate": 1.555e-08, "loss": 1.5158, "step": 19700 }, { "epoch": 3.95, "grad_norm": 42.34391403198242, "learning_rate": 1.3050000000000001e-08, "loss": 1.4648, "step": 19750 }, { "epoch": 3.96, "grad_norm": 48.5484733581543, "learning_rate": 1.055e-08, "loss": 1.5808, "step": 19800 }, { "epoch": 3.9699999999999998, "grad_norm": 51.64519500732422, "learning_rate": 8.05e-09, "loss": 1.4912, "step": 19850 }, { "epoch": 3.98, "grad_norm": 44.73149490356445, "learning_rate": 5.55e-09, "loss": 1.694, "step": 19900 }, { "epoch": 3.99, "grad_norm": 50.322174072265625, "learning_rate": 3.05e-09, "loss": 1.5025, "step": 19950 }, { "epoch": 4.0, "grad_norm": 51.53582763671875, "learning_rate": 5.5e-10, "loss": 1.5558, "step": 20000 } ], "logging_steps": 50, "max_steps": 20000, "num_input_tokens_seen": 0, "num_train_epochs": 4, "save_steps": 1000, "stateful_callbacks": { "TrainerControl": { "args": { "should_epoch_stop": false, "should_evaluate": false, "should_log": false, "should_save": true, "should_training_stop": true }, "attributes": {} } }, "total_flos": 0.0, "train_batch_size": 8, "trial_name": null, "trial_params": null }