| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 29.99962510778151, | |
| "eval_steps": 500, | |
| "global_step": 1200330, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.07, | |
| "learning_rate": 7.497500833055648e-05, | |
| "loss": 11.5654, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "learning_rate": 0.00014995001666111296, | |
| "loss": 1.9313, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "learning_rate": 0.0002249250249916694, | |
| "loss": 1.3886, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "learning_rate": 0.0002999000333222259, | |
| "loss": 1.3023, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "learning_rate": 0.0002992436418962473, | |
| "loss": 1.2746, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "learning_rate": 0.0002984862739685911, | |
| "loss": 1.2224, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "learning_rate": 0.00029772890604093486, | |
| "loss": 1.1809, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "learning_rate": 0.00029697153811327864, | |
| "loss": 1.1405, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "learning_rate": 0.0002962141701856224, | |
| "loss": 1.1178, | |
| "step": 27000 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "learning_rate": 0.0002954568022579662, | |
| "loss": 1.0984, | |
| "step": 30000 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "learning_rate": 0.00029469943433031, | |
| "loss": 1.0659, | |
| "step": 33000 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "learning_rate": 0.0002939420664026538, | |
| "loss": 1.0466, | |
| "step": 36000 | |
| }, | |
| { | |
| "epoch": 0.97, | |
| "learning_rate": 0.0002931846984749976, | |
| "loss": 1.0377, | |
| "step": 39000 | |
| }, | |
| { | |
| "epoch": 1.05, | |
| "learning_rate": 0.00029242733054734134, | |
| "loss": 1.005, | |
| "step": 42000 | |
| }, | |
| { | |
| "epoch": 1.12, | |
| "learning_rate": 0.0002916699626196851, | |
| "loss": 0.9926, | |
| "step": 45000 | |
| }, | |
| { | |
| "epoch": 1.2, | |
| "learning_rate": 0.00029091259469202896, | |
| "loss": 0.9863, | |
| "step": 48000 | |
| }, | |
| { | |
| "epoch": 1.27, | |
| "learning_rate": 0.0002901552267643727, | |
| "loss": 0.9723, | |
| "step": 51000 | |
| }, | |
| { | |
| "epoch": 1.35, | |
| "learning_rate": 0.00028939785883671653, | |
| "loss": 0.9571, | |
| "step": 54000 | |
| }, | |
| { | |
| "epoch": 1.42, | |
| "learning_rate": 0.0002886404909090603, | |
| "loss": 0.933, | |
| "step": 57000 | |
| }, | |
| { | |
| "epoch": 1.5, | |
| "learning_rate": 0.00028788312298140404, | |
| "loss": 0.9296, | |
| "step": 60000 | |
| }, | |
| { | |
| "epoch": 1.57, | |
| "learning_rate": 0.0002871257550537479, | |
| "loss": 0.9255, | |
| "step": 63000 | |
| }, | |
| { | |
| "epoch": 1.65, | |
| "learning_rate": 0.00028636838712609166, | |
| "loss": 0.9173, | |
| "step": 66000 | |
| }, | |
| { | |
| "epoch": 1.72, | |
| "learning_rate": 0.0002856110191984354, | |
| "loss": 0.9063, | |
| "step": 69000 | |
| }, | |
| { | |
| "epoch": 1.8, | |
| "learning_rate": 0.00028485365127077923, | |
| "loss": 0.9004, | |
| "step": 72000 | |
| }, | |
| { | |
| "epoch": 1.87, | |
| "learning_rate": 0.000284096283343123, | |
| "loss": 0.8997, | |
| "step": 75000 | |
| }, | |
| { | |
| "epoch": 1.95, | |
| "learning_rate": 0.00028333891541546674, | |
| "loss": 0.8943, | |
| "step": 78000 | |
| }, | |
| { | |
| "epoch": 2.02, | |
| "learning_rate": 0.0002825815474878106, | |
| "loss": 0.8797, | |
| "step": 81000 | |
| }, | |
| { | |
| "epoch": 2.1, | |
| "learning_rate": 0.00028182417956015436, | |
| "loss": 0.8501, | |
| "step": 84000 | |
| }, | |
| { | |
| "epoch": 2.17, | |
| "learning_rate": 0.0002810668116324981, | |
| "loss": 0.8479, | |
| "step": 87000 | |
| }, | |
| { | |
| "epoch": 2.25, | |
| "learning_rate": 0.00028030944370484193, | |
| "loss": 0.8451, | |
| "step": 90000 | |
| }, | |
| { | |
| "epoch": 2.32, | |
| "learning_rate": 0.0002795520757771857, | |
| "loss": 0.8315, | |
| "step": 93000 | |
| }, | |
| { | |
| "epoch": 2.4, | |
| "learning_rate": 0.0002787947078495295, | |
| "loss": 0.8353, | |
| "step": 96000 | |
| }, | |
| { | |
| "epoch": 2.47, | |
| "learning_rate": 0.0002780373399218733, | |
| "loss": 0.8271, | |
| "step": 99000 | |
| }, | |
| { | |
| "epoch": 2.55, | |
| "learning_rate": 0.00027727997199421706, | |
| "loss": 0.8129, | |
| "step": 102000 | |
| }, | |
| { | |
| "epoch": 2.62, | |
| "learning_rate": 0.00027652260406656084, | |
| "loss": 0.8125, | |
| "step": 105000 | |
| }, | |
| { | |
| "epoch": 2.7, | |
| "learning_rate": 0.0002757652361389046, | |
| "loss": 0.8119, | |
| "step": 108000 | |
| }, | |
| { | |
| "epoch": 2.77, | |
| "learning_rate": 0.0002750078682112484, | |
| "loss": 0.8, | |
| "step": 111000 | |
| }, | |
| { | |
| "epoch": 2.85, | |
| "learning_rate": 0.0002742505002835922, | |
| "loss": 0.7979, | |
| "step": 114000 | |
| }, | |
| { | |
| "epoch": 2.92, | |
| "learning_rate": 0.000273493132355936, | |
| "loss": 0.7914, | |
| "step": 117000 | |
| }, | |
| { | |
| "epoch": 3.0, | |
| "learning_rate": 0.00027273576442827976, | |
| "loss": 0.7916, | |
| "step": 120000 | |
| }, | |
| { | |
| "epoch": 3.07, | |
| "learning_rate": 0.00027197839650062354, | |
| "loss": 0.7542, | |
| "step": 123000 | |
| }, | |
| { | |
| "epoch": 3.15, | |
| "learning_rate": 0.0002712210285729673, | |
| "loss": 0.7643, | |
| "step": 126000 | |
| }, | |
| { | |
| "epoch": 3.22, | |
| "learning_rate": 0.0002704636606453111, | |
| "loss": 0.7576, | |
| "step": 129000 | |
| }, | |
| { | |
| "epoch": 3.3, | |
| "learning_rate": 0.0002697062927176549, | |
| "loss": 0.7551, | |
| "step": 132000 | |
| }, | |
| { | |
| "epoch": 3.37, | |
| "learning_rate": 0.0002689489247899987, | |
| "loss": 0.7537, | |
| "step": 135000 | |
| }, | |
| { | |
| "epoch": 3.45, | |
| "learning_rate": 0.00026819155686234246, | |
| "loss": 0.7536, | |
| "step": 138000 | |
| }, | |
| { | |
| "epoch": 3.52, | |
| "learning_rate": 0.00026743418893468624, | |
| "loss": 0.7466, | |
| "step": 141000 | |
| }, | |
| { | |
| "epoch": 3.6, | |
| "learning_rate": 0.00026667682100703, | |
| "loss": 0.7488, | |
| "step": 144000 | |
| }, | |
| { | |
| "epoch": 3.67, | |
| "learning_rate": 0.0002659194530793738, | |
| "loss": 0.7445, | |
| "step": 147000 | |
| }, | |
| { | |
| "epoch": 3.75, | |
| "learning_rate": 0.0002651620851517176, | |
| "loss": 0.7396, | |
| "step": 150000 | |
| }, | |
| { | |
| "epoch": 3.82, | |
| "learning_rate": 0.0002644047172240614, | |
| "loss": 0.7387, | |
| "step": 153000 | |
| }, | |
| { | |
| "epoch": 3.9, | |
| "learning_rate": 0.00026364734929640516, | |
| "loss": 0.7324, | |
| "step": 156000 | |
| }, | |
| { | |
| "epoch": 3.97, | |
| "learning_rate": 0.00026288998136874894, | |
| "loss": 0.7297, | |
| "step": 159000 | |
| }, | |
| { | |
| "epoch": 4.05, | |
| "learning_rate": 0.0002621326134410928, | |
| "loss": 0.7069, | |
| "step": 162000 | |
| }, | |
| { | |
| "epoch": 4.12, | |
| "learning_rate": 0.0002613752455134365, | |
| "loss": 0.7001, | |
| "step": 165000 | |
| }, | |
| { | |
| "epoch": 4.2, | |
| "learning_rate": 0.0002606178775857803, | |
| "loss": 0.6945, | |
| "step": 168000 | |
| }, | |
| { | |
| "epoch": 4.27, | |
| "learning_rate": 0.00025986050965812413, | |
| "loss": 0.7017, | |
| "step": 171000 | |
| }, | |
| { | |
| "epoch": 4.35, | |
| "learning_rate": 0.00025910314173046786, | |
| "loss": 0.7018, | |
| "step": 174000 | |
| }, | |
| { | |
| "epoch": 4.42, | |
| "learning_rate": 0.00025834577380281164, | |
| "loss": 0.6977, | |
| "step": 177000 | |
| }, | |
| { | |
| "epoch": 4.5, | |
| "learning_rate": 0.0002575884058751555, | |
| "loss": 0.6923, | |
| "step": 180000 | |
| }, | |
| { | |
| "epoch": 4.57, | |
| "learning_rate": 0.0002568310379474992, | |
| "loss": 0.6972, | |
| "step": 183000 | |
| }, | |
| { | |
| "epoch": 4.65, | |
| "learning_rate": 0.000256073670019843, | |
| "loss": 0.685, | |
| "step": 186000 | |
| }, | |
| { | |
| "epoch": 4.72, | |
| "learning_rate": 0.00025531630209218683, | |
| "loss": 0.6857, | |
| "step": 189000 | |
| }, | |
| { | |
| "epoch": 4.8, | |
| "learning_rate": 0.00025455893416453056, | |
| "loss": 0.6846, | |
| "step": 192000 | |
| }, | |
| { | |
| "epoch": 4.87, | |
| "learning_rate": 0.00025380156623687434, | |
| "loss": 0.6848, | |
| "step": 195000 | |
| }, | |
| { | |
| "epoch": 4.95, | |
| "learning_rate": 0.0002530441983092182, | |
| "loss": 0.6783, | |
| "step": 198000 | |
| }, | |
| { | |
| "epoch": 5.02, | |
| "learning_rate": 0.00025228683038156196, | |
| "loss": 0.6676, | |
| "step": 201000 | |
| }, | |
| { | |
| "epoch": 5.1, | |
| "learning_rate": 0.0002515294624539057, | |
| "loss": 0.6514, | |
| "step": 204000 | |
| }, | |
| { | |
| "epoch": 5.17, | |
| "learning_rate": 0.00025077209452624953, | |
| "loss": 0.6469, | |
| "step": 207000 | |
| }, | |
| { | |
| "epoch": 5.25, | |
| "learning_rate": 0.0002500147265985933, | |
| "loss": 0.6483, | |
| "step": 210000 | |
| }, | |
| { | |
| "epoch": 5.32, | |
| "learning_rate": 0.00024925735867093704, | |
| "loss": 0.6525, | |
| "step": 213000 | |
| }, | |
| { | |
| "epoch": 5.4, | |
| "learning_rate": 0.0002484999907432809, | |
| "loss": 0.6487, | |
| "step": 216000 | |
| }, | |
| { | |
| "epoch": 5.47, | |
| "learning_rate": 0.00024774262281562466, | |
| "loss": 0.6495, | |
| "step": 219000 | |
| }, | |
| { | |
| "epoch": 5.55, | |
| "learning_rate": 0.00024698525488796845, | |
| "loss": 0.645, | |
| "step": 222000 | |
| }, | |
| { | |
| "epoch": 5.62, | |
| "learning_rate": 0.00024622788696031223, | |
| "loss": 0.6517, | |
| "step": 225000 | |
| }, | |
| { | |
| "epoch": 5.7, | |
| "learning_rate": 0.000245470519032656, | |
| "loss": 0.6459, | |
| "step": 228000 | |
| }, | |
| { | |
| "epoch": 5.77, | |
| "learning_rate": 0.0002447131511049998, | |
| "loss": 0.6451, | |
| "step": 231000 | |
| }, | |
| { | |
| "epoch": 5.85, | |
| "learning_rate": 0.00024395578317734358, | |
| "loss": 0.6412, | |
| "step": 234000 | |
| }, | |
| { | |
| "epoch": 5.92, | |
| "learning_rate": 0.00024319841524968736, | |
| "loss": 0.6457, | |
| "step": 237000 | |
| }, | |
| { | |
| "epoch": 6.0, | |
| "learning_rate": 0.00024244104732203112, | |
| "loss": 0.64, | |
| "step": 240000 | |
| }, | |
| { | |
| "epoch": 6.07, | |
| "learning_rate": 0.00024168367939437493, | |
| "loss": 0.6137, | |
| "step": 243000 | |
| }, | |
| { | |
| "epoch": 6.15, | |
| "learning_rate": 0.0002409263114667187, | |
| "loss": 0.6131, | |
| "step": 246000 | |
| }, | |
| { | |
| "epoch": 6.22, | |
| "learning_rate": 0.0002401689435390625, | |
| "loss": 0.6153, | |
| "step": 249000 | |
| }, | |
| { | |
| "epoch": 6.3, | |
| "learning_rate": 0.00023941157561140628, | |
| "loss": 0.6193, | |
| "step": 252000 | |
| }, | |
| { | |
| "epoch": 6.37, | |
| "learning_rate": 0.00023865420768375006, | |
| "loss": 0.6161, | |
| "step": 255000 | |
| }, | |
| { | |
| "epoch": 6.45, | |
| "learning_rate": 0.00023789683975609387, | |
| "loss": 0.614, | |
| "step": 258000 | |
| }, | |
| { | |
| "epoch": 6.52, | |
| "learning_rate": 0.00023713947182843763, | |
| "loss": 0.6142, | |
| "step": 261000 | |
| }, | |
| { | |
| "epoch": 6.6, | |
| "learning_rate": 0.0002363821039007814, | |
| "loss": 0.6181, | |
| "step": 264000 | |
| }, | |
| { | |
| "epoch": 6.67, | |
| "learning_rate": 0.00023562473597312522, | |
| "loss": 0.6098, | |
| "step": 267000 | |
| }, | |
| { | |
| "epoch": 6.75, | |
| "learning_rate": 0.00023486736804546898, | |
| "loss": 0.6165, | |
| "step": 270000 | |
| }, | |
| { | |
| "epoch": 6.82, | |
| "learning_rate": 0.00023411000011781276, | |
| "loss": 0.612, | |
| "step": 273000 | |
| }, | |
| { | |
| "epoch": 6.9, | |
| "learning_rate": 0.00023335263219015657, | |
| "loss": 0.6109, | |
| "step": 276000 | |
| }, | |
| { | |
| "epoch": 6.97, | |
| "learning_rate": 0.00023259526426250033, | |
| "loss": 0.6141, | |
| "step": 279000 | |
| }, | |
| { | |
| "epoch": 7.05, | |
| "learning_rate": 0.0002318378963348441, | |
| "loss": 0.5912, | |
| "step": 282000 | |
| }, | |
| { | |
| "epoch": 7.12, | |
| "learning_rate": 0.00023108052840718792, | |
| "loss": 0.5838, | |
| "step": 285000 | |
| }, | |
| { | |
| "epoch": 7.2, | |
| "learning_rate": 0.00023032316047953168, | |
| "loss": 0.5823, | |
| "step": 288000 | |
| }, | |
| { | |
| "epoch": 7.27, | |
| "learning_rate": 0.00022956579255187546, | |
| "loss": 0.5836, | |
| "step": 291000 | |
| }, | |
| { | |
| "epoch": 7.35, | |
| "learning_rate": 0.00022880842462421927, | |
| "loss": 0.5857, | |
| "step": 294000 | |
| }, | |
| { | |
| "epoch": 7.42, | |
| "learning_rate": 0.00022805105669656303, | |
| "loss": 0.5872, | |
| "step": 297000 | |
| }, | |
| { | |
| "epoch": 7.5, | |
| "learning_rate": 0.0002272936887689068, | |
| "loss": 0.59, | |
| "step": 300000 | |
| }, | |
| { | |
| "epoch": 7.57, | |
| "learning_rate": 0.00022653632084125062, | |
| "loss": 0.5808, | |
| "step": 303000 | |
| }, | |
| { | |
| "epoch": 7.65, | |
| "learning_rate": 0.0002257789529135944, | |
| "loss": 0.5826, | |
| "step": 306000 | |
| }, | |
| { | |
| "epoch": 7.72, | |
| "learning_rate": 0.00022502158498593816, | |
| "loss": 0.5813, | |
| "step": 309000 | |
| }, | |
| { | |
| "epoch": 7.8, | |
| "learning_rate": 0.00022426421705828197, | |
| "loss": 0.5883, | |
| "step": 312000 | |
| }, | |
| { | |
| "epoch": 7.87, | |
| "learning_rate": 0.00022350684913062576, | |
| "loss": 0.5852, | |
| "step": 315000 | |
| }, | |
| { | |
| "epoch": 7.95, | |
| "learning_rate": 0.00022274948120296954, | |
| "loss": 0.5824, | |
| "step": 318000 | |
| }, | |
| { | |
| "epoch": 8.02, | |
| "learning_rate": 0.00022199211327531332, | |
| "loss": 0.575, | |
| "step": 321000 | |
| }, | |
| { | |
| "epoch": 8.1, | |
| "learning_rate": 0.0002212347453476571, | |
| "loss": 0.5561, | |
| "step": 324000 | |
| }, | |
| { | |
| "epoch": 8.17, | |
| "learning_rate": 0.0002204773774200009, | |
| "loss": 0.5596, | |
| "step": 327000 | |
| }, | |
| { | |
| "epoch": 8.25, | |
| "learning_rate": 0.00021972000949234467, | |
| "loss": 0.5603, | |
| "step": 330000 | |
| }, | |
| { | |
| "epoch": 8.32, | |
| "learning_rate": 0.00021896264156468846, | |
| "loss": 0.5575, | |
| "step": 333000 | |
| }, | |
| { | |
| "epoch": 8.4, | |
| "learning_rate": 0.00021820527363703224, | |
| "loss": 0.5545, | |
| "step": 336000 | |
| }, | |
| { | |
| "epoch": 8.47, | |
| "learning_rate": 0.00021744790570937602, | |
| "loss": 0.5629, | |
| "step": 339000 | |
| }, | |
| { | |
| "epoch": 8.55, | |
| "learning_rate": 0.00021669053778171983, | |
| "loss": 0.5572, | |
| "step": 342000 | |
| }, | |
| { | |
| "epoch": 8.62, | |
| "learning_rate": 0.0002159331698540636, | |
| "loss": 0.5602, | |
| "step": 345000 | |
| }, | |
| { | |
| "epoch": 8.7, | |
| "learning_rate": 0.00021517580192640737, | |
| "loss": 0.5521, | |
| "step": 348000 | |
| }, | |
| { | |
| "epoch": 8.77, | |
| "learning_rate": 0.00021441843399875118, | |
| "loss": 0.5582, | |
| "step": 351000 | |
| }, | |
| { | |
| "epoch": 8.85, | |
| "learning_rate": 0.00021366106607109497, | |
| "loss": 0.5539, | |
| "step": 354000 | |
| }, | |
| { | |
| "epoch": 8.92, | |
| "learning_rate": 0.00021290369814343872, | |
| "loss": 0.5589, | |
| "step": 357000 | |
| }, | |
| { | |
| "epoch": 9.0, | |
| "learning_rate": 0.00021214633021578253, | |
| "loss": 0.5549, | |
| "step": 360000 | |
| }, | |
| { | |
| "epoch": 9.07, | |
| "learning_rate": 0.00021138896228812632, | |
| "loss": 0.525, | |
| "step": 363000 | |
| }, | |
| { | |
| "epoch": 9.15, | |
| "learning_rate": 0.00021063159436047007, | |
| "loss": 0.5304, | |
| "step": 366000 | |
| }, | |
| { | |
| "epoch": 9.22, | |
| "learning_rate": 0.00020987422643281388, | |
| "loss": 0.5331, | |
| "step": 369000 | |
| }, | |
| { | |
| "epoch": 9.3, | |
| "learning_rate": 0.00020911685850515766, | |
| "loss": 0.5341, | |
| "step": 372000 | |
| }, | |
| { | |
| "epoch": 9.37, | |
| "learning_rate": 0.00020835949057750142, | |
| "loss": 0.5319, | |
| "step": 375000 | |
| }, | |
| { | |
| "epoch": 9.45, | |
| "learning_rate": 0.00020760212264984523, | |
| "loss": 0.5318, | |
| "step": 378000 | |
| }, | |
| { | |
| "epoch": 9.52, | |
| "learning_rate": 0.00020684475472218901, | |
| "loss": 0.5368, | |
| "step": 381000 | |
| }, | |
| { | |
| "epoch": 9.6, | |
| "learning_rate": 0.00020608738679453277, | |
| "loss": 0.532, | |
| "step": 384000 | |
| }, | |
| { | |
| "epoch": 9.67, | |
| "learning_rate": 0.00020533001886687658, | |
| "loss": 0.5353, | |
| "step": 387000 | |
| }, | |
| { | |
| "epoch": 9.75, | |
| "learning_rate": 0.00020457265093922036, | |
| "loss": 0.5329, | |
| "step": 390000 | |
| }, | |
| { | |
| "epoch": 9.82, | |
| "learning_rate": 0.00020381528301156415, | |
| "loss": 0.5354, | |
| "step": 393000 | |
| }, | |
| { | |
| "epoch": 9.9, | |
| "learning_rate": 0.00020305791508390793, | |
| "loss": 0.5375, | |
| "step": 396000 | |
| }, | |
| { | |
| "epoch": 9.97, | |
| "learning_rate": 0.00020230054715625171, | |
| "loss": 0.5377, | |
| "step": 399000 | |
| }, | |
| { | |
| "epoch": 10.05, | |
| "learning_rate": 0.0002015431792285955, | |
| "loss": 0.5191, | |
| "step": 402000 | |
| }, | |
| { | |
| "epoch": 10.12, | |
| "learning_rate": 0.00020078581130093928, | |
| "loss": 0.5057, | |
| "step": 405000 | |
| }, | |
| { | |
| "epoch": 10.2, | |
| "learning_rate": 0.00020002844337328306, | |
| "loss": 0.508, | |
| "step": 408000 | |
| }, | |
| { | |
| "epoch": 10.27, | |
| "learning_rate": 0.00019927107544562687, | |
| "loss": 0.5117, | |
| "step": 411000 | |
| }, | |
| { | |
| "epoch": 10.35, | |
| "learning_rate": 0.00019851370751797063, | |
| "loss": 0.5135, | |
| "step": 414000 | |
| }, | |
| { | |
| "epoch": 10.42, | |
| "learning_rate": 0.00019775633959031441, | |
| "loss": 0.5129, | |
| "step": 417000 | |
| }, | |
| { | |
| "epoch": 10.5, | |
| "learning_rate": 0.00019699897166265822, | |
| "loss": 0.511, | |
| "step": 420000 | |
| }, | |
| { | |
| "epoch": 10.57, | |
| "learning_rate": 0.00019624160373500198, | |
| "loss": 0.5095, | |
| "step": 423000 | |
| }, | |
| { | |
| "epoch": 10.65, | |
| "learning_rate": 0.0001954842358073458, | |
| "loss": 0.5129, | |
| "step": 426000 | |
| }, | |
| { | |
| "epoch": 10.72, | |
| "learning_rate": 0.00019472686787968957, | |
| "loss": 0.5124, | |
| "step": 429000 | |
| }, | |
| { | |
| "epoch": 10.8, | |
| "learning_rate": 0.00019396949995203333, | |
| "loss": 0.5138, | |
| "step": 432000 | |
| }, | |
| { | |
| "epoch": 10.87, | |
| "learning_rate": 0.00019321213202437714, | |
| "loss": 0.5125, | |
| "step": 435000 | |
| }, | |
| { | |
| "epoch": 10.95, | |
| "learning_rate": 0.00019245476409672092, | |
| "loss": 0.5153, | |
| "step": 438000 | |
| }, | |
| { | |
| "epoch": 11.02, | |
| "learning_rate": 0.00019169739616906468, | |
| "loss": 0.5022, | |
| "step": 441000 | |
| }, | |
| { | |
| "epoch": 11.1, | |
| "learning_rate": 0.0001909400282414085, | |
| "loss": 0.4908, | |
| "step": 444000 | |
| }, | |
| { | |
| "epoch": 11.17, | |
| "learning_rate": 0.00019018266031375227, | |
| "loss": 0.4908, | |
| "step": 447000 | |
| }, | |
| { | |
| "epoch": 11.25, | |
| "learning_rate": 0.00018942529238609603, | |
| "loss": 0.4907, | |
| "step": 450000 | |
| }, | |
| { | |
| "epoch": 11.32, | |
| "learning_rate": 0.00018866792445843984, | |
| "loss": 0.4895, | |
| "step": 453000 | |
| }, | |
| { | |
| "epoch": 11.4, | |
| "learning_rate": 0.00018791055653078362, | |
| "loss": 0.4948, | |
| "step": 456000 | |
| }, | |
| { | |
| "epoch": 11.47, | |
| "learning_rate": 0.00018715318860312743, | |
| "loss": 0.4908, | |
| "step": 459000 | |
| }, | |
| { | |
| "epoch": 11.55, | |
| "learning_rate": 0.0001863958206754712, | |
| "loss": 0.4931, | |
| "step": 462000 | |
| }, | |
| { | |
| "epoch": 11.62, | |
| "learning_rate": 0.00018563845274781497, | |
| "loss": 0.4945, | |
| "step": 465000 | |
| }, | |
| { | |
| "epoch": 11.7, | |
| "learning_rate": 0.00018488108482015878, | |
| "loss": 0.5005, | |
| "step": 468000 | |
| }, | |
| { | |
| "epoch": 11.77, | |
| "learning_rate": 0.00018412371689250254, | |
| "loss": 0.4974, | |
| "step": 471000 | |
| }, | |
| { | |
| "epoch": 11.85, | |
| "learning_rate": 0.00018336634896484632, | |
| "loss": 0.4997, | |
| "step": 474000 | |
| }, | |
| { | |
| "epoch": 11.92, | |
| "learning_rate": 0.00018260898103719013, | |
| "loss": 0.5027, | |
| "step": 477000 | |
| }, | |
| { | |
| "epoch": 12.0, | |
| "learning_rate": 0.0001818516131095339, | |
| "loss": 0.4962, | |
| "step": 480000 | |
| }, | |
| { | |
| "epoch": 12.07, | |
| "learning_rate": 0.00018109424518187767, | |
| "loss": 0.4766, | |
| "step": 483000 | |
| }, | |
| { | |
| "epoch": 12.15, | |
| "learning_rate": 0.00018033687725422148, | |
| "loss": 0.4782, | |
| "step": 486000 | |
| }, | |
| { | |
| "epoch": 12.22, | |
| "learning_rate": 0.00017957950932656524, | |
| "loss": 0.4793, | |
| "step": 489000 | |
| }, | |
| { | |
| "epoch": 12.3, | |
| "learning_rate": 0.00017882214139890902, | |
| "loss": 0.4769, | |
| "step": 492000 | |
| }, | |
| { | |
| "epoch": 12.37, | |
| "learning_rate": 0.00017806477347125283, | |
| "loss": 0.4801, | |
| "step": 495000 | |
| }, | |
| { | |
| "epoch": 12.45, | |
| "learning_rate": 0.0001773074055435966, | |
| "loss": 0.4783, | |
| "step": 498000 | |
| }, | |
| { | |
| "epoch": 12.52, | |
| "learning_rate": 0.0001765500376159404, | |
| "loss": 0.4804, | |
| "step": 501000 | |
| }, | |
| { | |
| "epoch": 12.6, | |
| "learning_rate": 0.00017579266968828418, | |
| "loss": 0.4758, | |
| "step": 504000 | |
| }, | |
| { | |
| "epoch": 12.67, | |
| "learning_rate": 0.00017503530176062794, | |
| "loss": 0.4761, | |
| "step": 507000 | |
| }, | |
| { | |
| "epoch": 12.75, | |
| "learning_rate": 0.00017427793383297175, | |
| "loss": 0.473, | |
| "step": 510000 | |
| }, | |
| { | |
| "epoch": 12.82, | |
| "learning_rate": 0.00017352056590531553, | |
| "loss": 0.4793, | |
| "step": 513000 | |
| }, | |
| { | |
| "epoch": 12.9, | |
| "learning_rate": 0.00017276319797765932, | |
| "loss": 0.4772, | |
| "step": 516000 | |
| }, | |
| { | |
| "epoch": 12.97, | |
| "learning_rate": 0.0001720058300500031, | |
| "loss": 0.4752, | |
| "step": 519000 | |
| }, | |
| { | |
| "epoch": 13.05, | |
| "learning_rate": 0.00017124846212234688, | |
| "loss": 0.4611, | |
| "step": 522000 | |
| }, | |
| { | |
| "epoch": 13.12, | |
| "learning_rate": 0.0001704910941946907, | |
| "loss": 0.4549, | |
| "step": 525000 | |
| }, | |
| { | |
| "epoch": 13.2, | |
| "learning_rate": 0.00016973372626703445, | |
| "loss": 0.456, | |
| "step": 528000 | |
| }, | |
| { | |
| "epoch": 13.27, | |
| "learning_rate": 0.00016897635833937823, | |
| "loss": 0.4596, | |
| "step": 531000 | |
| }, | |
| { | |
| "epoch": 13.35, | |
| "learning_rate": 0.00016821899041172204, | |
| "loss": 0.4629, | |
| "step": 534000 | |
| }, | |
| { | |
| "epoch": 13.42, | |
| "learning_rate": 0.0001674616224840658, | |
| "loss": 0.463, | |
| "step": 537000 | |
| }, | |
| { | |
| "epoch": 13.5, | |
| "learning_rate": 0.00016670425455640958, | |
| "loss": 0.461, | |
| "step": 540000 | |
| }, | |
| { | |
| "epoch": 13.57, | |
| "learning_rate": 0.0001659468866287534, | |
| "loss": 0.4631, | |
| "step": 543000 | |
| }, | |
| { | |
| "epoch": 13.65, | |
| "learning_rate": 0.00016518951870109715, | |
| "loss": 0.4627, | |
| "step": 546000 | |
| }, | |
| { | |
| "epoch": 13.72, | |
| "learning_rate": 0.00016443215077344093, | |
| "loss": 0.4639, | |
| "step": 549000 | |
| }, | |
| { | |
| "epoch": 13.8, | |
| "learning_rate": 0.00016367478284578474, | |
| "loss": 0.4663, | |
| "step": 552000 | |
| }, | |
| { | |
| "epoch": 13.87, | |
| "learning_rate": 0.0001629174149181285, | |
| "loss": 0.46, | |
| "step": 555000 | |
| }, | |
| { | |
| "epoch": 13.95, | |
| "learning_rate": 0.00016216004699047228, | |
| "loss": 0.4661, | |
| "step": 558000 | |
| }, | |
| { | |
| "epoch": 14.02, | |
| "learning_rate": 0.0001614026790628161, | |
| "loss": 0.4564, | |
| "step": 561000 | |
| }, | |
| { | |
| "epoch": 14.1, | |
| "learning_rate": 0.00016064531113515988, | |
| "loss": 0.4391, | |
| "step": 564000 | |
| }, | |
| { | |
| "epoch": 14.17, | |
| "learning_rate": 0.00015988794320750363, | |
| "loss": 0.4479, | |
| "step": 567000 | |
| }, | |
| { | |
| "epoch": 14.25, | |
| "learning_rate": 0.00015913057527984744, | |
| "loss": 0.4427, | |
| "step": 570000 | |
| }, | |
| { | |
| "epoch": 14.32, | |
| "learning_rate": 0.00015837320735219123, | |
| "loss": 0.4425, | |
| "step": 573000 | |
| }, | |
| { | |
| "epoch": 14.4, | |
| "learning_rate": 0.00015761583942453498, | |
| "loss": 0.4412, | |
| "step": 576000 | |
| }, | |
| { | |
| "epoch": 14.47, | |
| "learning_rate": 0.0001568584714968788, | |
| "loss": 0.4488, | |
| "step": 579000 | |
| }, | |
| { | |
| "epoch": 14.55, | |
| "learning_rate": 0.00015610110356922258, | |
| "loss": 0.4464, | |
| "step": 582000 | |
| }, | |
| { | |
| "epoch": 14.62, | |
| "learning_rate": 0.00015534373564156636, | |
| "loss": 0.4507, | |
| "step": 585000 | |
| }, | |
| { | |
| "epoch": 14.7, | |
| "learning_rate": 0.00015458636771391014, | |
| "loss": 0.4476, | |
| "step": 588000 | |
| }, | |
| { | |
| "epoch": 14.77, | |
| "learning_rate": 0.00015382899978625393, | |
| "loss": 0.4462, | |
| "step": 591000 | |
| }, | |
| { | |
| "epoch": 14.85, | |
| "learning_rate": 0.0001530716318585977, | |
| "loss": 0.4493, | |
| "step": 594000 | |
| }, | |
| { | |
| "epoch": 14.92, | |
| "learning_rate": 0.0001523142639309415, | |
| "loss": 0.4478, | |
| "step": 597000 | |
| }, | |
| { | |
| "epoch": 15.0, | |
| "learning_rate": 0.00015155689600328528, | |
| "loss": 0.4483, | |
| "step": 600000 | |
| }, | |
| { | |
| "epoch": 15.07, | |
| "learning_rate": 0.00015079952807562906, | |
| "loss": 0.4276, | |
| "step": 603000 | |
| }, | |
| { | |
| "epoch": 15.15, | |
| "learning_rate": 0.00015004216014797284, | |
| "loss": 0.4259, | |
| "step": 606000 | |
| }, | |
| { | |
| "epoch": 15.22, | |
| "learning_rate": 0.00014928479222031665, | |
| "loss": 0.4263, | |
| "step": 609000 | |
| }, | |
| { | |
| "epoch": 15.3, | |
| "learning_rate": 0.0001485274242926604, | |
| "loss": 0.4302, | |
| "step": 612000 | |
| }, | |
| { | |
| "epoch": 15.37, | |
| "learning_rate": 0.0001477700563650042, | |
| "loss": 0.4302, | |
| "step": 615000 | |
| }, | |
| { | |
| "epoch": 15.45, | |
| "learning_rate": 0.000147012688437348, | |
| "loss": 0.4289, | |
| "step": 618000 | |
| }, | |
| { | |
| "epoch": 15.52, | |
| "learning_rate": 0.00014625532050969176, | |
| "loss": 0.4311, | |
| "step": 621000 | |
| }, | |
| { | |
| "epoch": 15.6, | |
| "learning_rate": 0.00014549795258203554, | |
| "loss": 0.4327, | |
| "step": 624000 | |
| }, | |
| { | |
| "epoch": 15.67, | |
| "learning_rate": 0.00014474058465437935, | |
| "loss": 0.4315, | |
| "step": 627000 | |
| }, | |
| { | |
| "epoch": 15.75, | |
| "learning_rate": 0.0001439832167267231, | |
| "loss": 0.4305, | |
| "step": 630000 | |
| }, | |
| { | |
| "epoch": 15.82, | |
| "learning_rate": 0.00014322584879906692, | |
| "loss": 0.429, | |
| "step": 633000 | |
| }, | |
| { | |
| "epoch": 15.9, | |
| "learning_rate": 0.0001424684808714107, | |
| "loss": 0.4288, | |
| "step": 636000 | |
| }, | |
| { | |
| "epoch": 15.97, | |
| "learning_rate": 0.00014171111294375449, | |
| "loss": 0.4309, | |
| "step": 639000 | |
| }, | |
| { | |
| "epoch": 16.05, | |
| "learning_rate": 0.00014095374501609827, | |
| "loss": 0.4179, | |
| "step": 642000 | |
| }, | |
| { | |
| "epoch": 16.12, | |
| "learning_rate": 0.00014019637708844205, | |
| "loss": 0.4098, | |
| "step": 645000 | |
| }, | |
| { | |
| "epoch": 16.2, | |
| "learning_rate": 0.00013943900916078584, | |
| "loss": 0.4136, | |
| "step": 648000 | |
| }, | |
| { | |
| "epoch": 16.27, | |
| "learning_rate": 0.00013868164123312962, | |
| "loss": 0.4115, | |
| "step": 651000 | |
| }, | |
| { | |
| "epoch": 16.35, | |
| "learning_rate": 0.0001379242733054734, | |
| "loss": 0.4192, | |
| "step": 654000 | |
| }, | |
| { | |
| "epoch": 16.42, | |
| "learning_rate": 0.00013716690537781719, | |
| "loss": 0.4159, | |
| "step": 657000 | |
| }, | |
| { | |
| "epoch": 16.5, | |
| "learning_rate": 0.00013640953745016097, | |
| "loss": 0.4176, | |
| "step": 660000 | |
| }, | |
| { | |
| "epoch": 16.57, | |
| "learning_rate": 0.00013565216952250475, | |
| "loss": 0.4166, | |
| "step": 663000 | |
| }, | |
| { | |
| "epoch": 16.65, | |
| "learning_rate": 0.00013489480159484854, | |
| "loss": 0.4214, | |
| "step": 666000 | |
| }, | |
| { | |
| "epoch": 16.72, | |
| "learning_rate": 0.00013413743366719232, | |
| "loss": 0.4121, | |
| "step": 669000 | |
| }, | |
| { | |
| "epoch": 16.8, | |
| "learning_rate": 0.0001333800657395361, | |
| "loss": 0.414, | |
| "step": 672000 | |
| }, | |
| { | |
| "epoch": 16.87, | |
| "learning_rate": 0.00013262269781187989, | |
| "loss": 0.4159, | |
| "step": 675000 | |
| }, | |
| { | |
| "epoch": 16.95, | |
| "learning_rate": 0.00013186532988422367, | |
| "loss": 0.4135, | |
| "step": 678000 | |
| }, | |
| { | |
| "epoch": 17.02, | |
| "learning_rate": 0.00013110796195656748, | |
| "loss": 0.4098, | |
| "step": 681000 | |
| }, | |
| { | |
| "epoch": 17.1, | |
| "learning_rate": 0.00013035059402891124, | |
| "loss": 0.3947, | |
| "step": 684000 | |
| }, | |
| { | |
| "epoch": 17.17, | |
| "learning_rate": 0.00012959322610125502, | |
| "loss": 0.4026, | |
| "step": 687000 | |
| }, | |
| { | |
| "epoch": 17.25, | |
| "learning_rate": 0.00012883585817359883, | |
| "loss": 0.397, | |
| "step": 690000 | |
| }, | |
| { | |
| "epoch": 17.32, | |
| "learning_rate": 0.0001280784902459426, | |
| "loss": 0.3998, | |
| "step": 693000 | |
| }, | |
| { | |
| "epoch": 17.39, | |
| "learning_rate": 0.00012732112231828637, | |
| "loss": 0.3988, | |
| "step": 696000 | |
| }, | |
| { | |
| "epoch": 17.47, | |
| "learning_rate": 0.00012656375439063018, | |
| "loss": 0.4015, | |
| "step": 699000 | |
| }, | |
| { | |
| "epoch": 17.54, | |
| "learning_rate": 0.00012580638646297396, | |
| "loss": 0.402, | |
| "step": 702000 | |
| }, | |
| { | |
| "epoch": 17.62, | |
| "learning_rate": 0.00012504901853531775, | |
| "loss": 0.4018, | |
| "step": 705000 | |
| }, | |
| { | |
| "epoch": 17.69, | |
| "learning_rate": 0.00012429165060766153, | |
| "loss": 0.3987, | |
| "step": 708000 | |
| }, | |
| { | |
| "epoch": 17.77, | |
| "learning_rate": 0.0001235342826800053, | |
| "loss": 0.3974, | |
| "step": 711000 | |
| }, | |
| { | |
| "epoch": 17.84, | |
| "learning_rate": 0.0001227769147523491, | |
| "loss": 0.4006, | |
| "step": 714000 | |
| }, | |
| { | |
| "epoch": 17.92, | |
| "learning_rate": 0.00012201954682469288, | |
| "loss": 0.4028, | |
| "step": 717000 | |
| }, | |
| { | |
| "epoch": 17.99, | |
| "learning_rate": 0.00012126217889703665, | |
| "loss": 0.4019, | |
| "step": 720000 | |
| }, | |
| { | |
| "epoch": 18.07, | |
| "learning_rate": 0.00012050481096938045, | |
| "loss": 0.3853, | |
| "step": 723000 | |
| }, | |
| { | |
| "epoch": 18.14, | |
| "learning_rate": 0.00011974744304172423, | |
| "loss": 0.387, | |
| "step": 726000 | |
| }, | |
| { | |
| "epoch": 18.22, | |
| "learning_rate": 0.00011899007511406801, | |
| "loss": 0.3822, | |
| "step": 729000 | |
| }, | |
| { | |
| "epoch": 18.29, | |
| "learning_rate": 0.0001182327071864118, | |
| "loss": 0.3861, | |
| "step": 732000 | |
| }, | |
| { | |
| "epoch": 18.37, | |
| "learning_rate": 0.00011747533925875558, | |
| "loss": 0.3877, | |
| "step": 735000 | |
| }, | |
| { | |
| "epoch": 18.44, | |
| "learning_rate": 0.00011671797133109938, | |
| "loss": 0.3867, | |
| "step": 738000 | |
| }, | |
| { | |
| "epoch": 18.52, | |
| "learning_rate": 0.00011596060340344316, | |
| "loss": 0.388, | |
| "step": 741000 | |
| }, | |
| { | |
| "epoch": 18.59, | |
| "learning_rate": 0.00011520323547578693, | |
| "loss": 0.3889, | |
| "step": 744000 | |
| }, | |
| { | |
| "epoch": 18.67, | |
| "learning_rate": 0.00011444586754813073, | |
| "loss": 0.3867, | |
| "step": 747000 | |
| }, | |
| { | |
| "epoch": 18.74, | |
| "learning_rate": 0.00011368849962047451, | |
| "loss": 0.3897, | |
| "step": 750000 | |
| }, | |
| { | |
| "epoch": 18.82, | |
| "learning_rate": 0.00011293113169281828, | |
| "loss": 0.3873, | |
| "step": 753000 | |
| }, | |
| { | |
| "epoch": 18.89, | |
| "learning_rate": 0.00011217376376516208, | |
| "loss": 0.3881, | |
| "step": 756000 | |
| }, | |
| { | |
| "epoch": 18.97, | |
| "learning_rate": 0.00011141639583750586, | |
| "loss": 0.3915, | |
| "step": 759000 | |
| }, | |
| { | |
| "epoch": 19.04, | |
| "learning_rate": 0.00011065902790984966, | |
| "loss": 0.3766, | |
| "step": 762000 | |
| }, | |
| { | |
| "epoch": 19.12, | |
| "learning_rate": 0.00010990165998219343, | |
| "loss": 0.3757, | |
| "step": 765000 | |
| }, | |
| { | |
| "epoch": 19.19, | |
| "learning_rate": 0.00010914429205453721, | |
| "loss": 0.3705, | |
| "step": 768000 | |
| }, | |
| { | |
| "epoch": 19.27, | |
| "learning_rate": 0.000108386924126881, | |
| "loss": 0.3728, | |
| "step": 771000 | |
| }, | |
| { | |
| "epoch": 19.34, | |
| "learning_rate": 0.00010762955619922477, | |
| "loss": 0.3756, | |
| "step": 774000 | |
| }, | |
| { | |
| "epoch": 19.42, | |
| "learning_rate": 0.00010687218827156856, | |
| "loss": 0.3772, | |
| "step": 777000 | |
| }, | |
| { | |
| "epoch": 19.49, | |
| "learning_rate": 0.00010611482034391236, | |
| "loss": 0.3761, | |
| "step": 780000 | |
| }, | |
| { | |
| "epoch": 19.57, | |
| "learning_rate": 0.00010535745241625614, | |
| "loss": 0.3739, | |
| "step": 783000 | |
| }, | |
| { | |
| "epoch": 19.64, | |
| "learning_rate": 0.00010460008448859992, | |
| "loss": 0.371, | |
| "step": 786000 | |
| }, | |
| { | |
| "epoch": 19.72, | |
| "learning_rate": 0.0001038427165609437, | |
| "loss": 0.3721, | |
| "step": 789000 | |
| }, | |
| { | |
| "epoch": 19.79, | |
| "learning_rate": 0.00010308534863328749, | |
| "loss": 0.3733, | |
| "step": 792000 | |
| }, | |
| { | |
| "epoch": 19.87, | |
| "learning_rate": 0.00010232798070563129, | |
| "loss": 0.3742, | |
| "step": 795000 | |
| }, | |
| { | |
| "epoch": 19.94, | |
| "learning_rate": 0.00010157061277797505, | |
| "loss": 0.3732, | |
| "step": 798000 | |
| }, | |
| { | |
| "epoch": 20.02, | |
| "learning_rate": 0.00010081324485031884, | |
| "loss": 0.3703, | |
| "step": 801000 | |
| }, | |
| { | |
| "epoch": 20.09, | |
| "learning_rate": 0.00010005587692266263, | |
| "loss": 0.3609, | |
| "step": 804000 | |
| }, | |
| { | |
| "epoch": 20.17, | |
| "learning_rate": 9.92985089950064e-05, | |
| "loss": 0.3604, | |
| "step": 807000 | |
| }, | |
| { | |
| "epoch": 20.24, | |
| "learning_rate": 9.85411410673502e-05, | |
| "loss": 0.3614, | |
| "step": 810000 | |
| }, | |
| { | |
| "epoch": 20.32, | |
| "learning_rate": 9.778377313969398e-05, | |
| "loss": 0.3617, | |
| "step": 813000 | |
| }, | |
| { | |
| "epoch": 20.39, | |
| "learning_rate": 9.702640521203775e-05, | |
| "loss": 0.3611, | |
| "step": 816000 | |
| }, | |
| { | |
| "epoch": 20.47, | |
| "learning_rate": 9.626903728438155e-05, | |
| "loss": 0.3607, | |
| "step": 819000 | |
| }, | |
| { | |
| "epoch": 20.54, | |
| "learning_rate": 9.551166935672533e-05, | |
| "loss": 0.3605, | |
| "step": 822000 | |
| }, | |
| { | |
| "epoch": 20.62, | |
| "learning_rate": 9.475430142906912e-05, | |
| "loss": 0.3589, | |
| "step": 825000 | |
| }, | |
| { | |
| "epoch": 20.69, | |
| "learning_rate": 9.39969335014129e-05, | |
| "loss": 0.3586, | |
| "step": 828000 | |
| }, | |
| { | |
| "epoch": 20.77, | |
| "learning_rate": 9.323956557375668e-05, | |
| "loss": 0.3626, | |
| "step": 831000 | |
| }, | |
| { | |
| "epoch": 20.84, | |
| "learning_rate": 9.248219764610048e-05, | |
| "loss": 0.3616, | |
| "step": 834000 | |
| }, | |
| { | |
| "epoch": 20.92, | |
| "learning_rate": 9.172482971844426e-05, | |
| "loss": 0.3595, | |
| "step": 837000 | |
| }, | |
| { | |
| "epoch": 20.99, | |
| "learning_rate": 9.096746179078803e-05, | |
| "loss": 0.3624, | |
| "step": 840000 | |
| }, | |
| { | |
| "epoch": 21.07, | |
| "learning_rate": 9.021009386313183e-05, | |
| "loss": 0.3468, | |
| "step": 843000 | |
| }, | |
| { | |
| "epoch": 21.14, | |
| "learning_rate": 8.945272593547561e-05, | |
| "loss": 0.3475, | |
| "step": 846000 | |
| }, | |
| { | |
| "epoch": 21.22, | |
| "learning_rate": 8.869535800781938e-05, | |
| "loss": 0.3486, | |
| "step": 849000 | |
| }, | |
| { | |
| "epoch": 21.29, | |
| "learning_rate": 8.793799008016318e-05, | |
| "loss": 0.3454, | |
| "step": 852000 | |
| }, | |
| { | |
| "epoch": 21.37, | |
| "learning_rate": 8.718062215250696e-05, | |
| "loss": 0.3496, | |
| "step": 855000 | |
| }, | |
| { | |
| "epoch": 21.44, | |
| "learning_rate": 8.642325422485073e-05, | |
| "loss": 0.3478, | |
| "step": 858000 | |
| }, | |
| { | |
| "epoch": 21.52, | |
| "learning_rate": 8.566588629719453e-05, | |
| "loss": 0.346, | |
| "step": 861000 | |
| }, | |
| { | |
| "epoch": 21.59, | |
| "learning_rate": 8.490851836953831e-05, | |
| "loss": 0.3513, | |
| "step": 864000 | |
| }, | |
| { | |
| "epoch": 21.67, | |
| "learning_rate": 8.415115044188211e-05, | |
| "loss": 0.3524, | |
| "step": 867000 | |
| }, | |
| { | |
| "epoch": 21.74, | |
| "learning_rate": 8.339378251422588e-05, | |
| "loss": 0.3494, | |
| "step": 870000 | |
| }, | |
| { | |
| "epoch": 21.82, | |
| "learning_rate": 8.263641458656966e-05, | |
| "loss": 0.3521, | |
| "step": 873000 | |
| }, | |
| { | |
| "epoch": 21.89, | |
| "learning_rate": 8.187904665891346e-05, | |
| "loss": 0.3509, | |
| "step": 876000 | |
| }, | |
| { | |
| "epoch": 21.97, | |
| "learning_rate": 8.112167873125724e-05, | |
| "loss": 0.3501, | |
| "step": 879000 | |
| }, | |
| { | |
| "epoch": 22.04, | |
| "learning_rate": 8.036431080360101e-05, | |
| "loss": 0.3394, | |
| "step": 882000 | |
| }, | |
| { | |
| "epoch": 22.12, | |
| "learning_rate": 7.960694287594481e-05, | |
| "loss": 0.3357, | |
| "step": 885000 | |
| }, | |
| { | |
| "epoch": 22.19, | |
| "learning_rate": 7.88495749482886e-05, | |
| "loss": 0.3378, | |
| "step": 888000 | |
| }, | |
| { | |
| "epoch": 22.27, | |
| "learning_rate": 7.809220702063239e-05, | |
| "loss": 0.3362, | |
| "step": 891000 | |
| }, | |
| { | |
| "epoch": 22.34, | |
| "learning_rate": 7.733483909297616e-05, | |
| "loss": 0.3382, | |
| "step": 894000 | |
| }, | |
| { | |
| "epoch": 22.42, | |
| "learning_rate": 7.657747116531994e-05, | |
| "loss": 0.3407, | |
| "step": 897000 | |
| }, | |
| { | |
| "epoch": 22.49, | |
| "learning_rate": 7.582010323766374e-05, | |
| "loss": 0.3381, | |
| "step": 900000 | |
| }, | |
| { | |
| "epoch": 22.57, | |
| "learning_rate": 7.506273531000751e-05, | |
| "loss": 0.3378, | |
| "step": 903000 | |
| }, | |
| { | |
| "epoch": 22.64, | |
| "learning_rate": 7.430536738235131e-05, | |
| "loss": 0.3368, | |
| "step": 906000 | |
| }, | |
| { | |
| "epoch": 22.72, | |
| "learning_rate": 7.354799945469509e-05, | |
| "loss": 0.3384, | |
| "step": 909000 | |
| }, | |
| { | |
| "epoch": 22.79, | |
| "learning_rate": 7.279063152703887e-05, | |
| "loss": 0.3351, | |
| "step": 912000 | |
| }, | |
| { | |
| "epoch": 22.87, | |
| "learning_rate": 7.203326359938266e-05, | |
| "loss": 0.3399, | |
| "step": 915000 | |
| }, | |
| { | |
| "epoch": 22.94, | |
| "learning_rate": 7.127589567172644e-05, | |
| "loss": 0.3366, | |
| "step": 918000 | |
| }, | |
| { | |
| "epoch": 23.02, | |
| "learning_rate": 7.051852774407022e-05, | |
| "loss": 0.3313, | |
| "step": 921000 | |
| }, | |
| { | |
| "epoch": 23.09, | |
| "learning_rate": 6.976115981641401e-05, | |
| "loss": 0.3291, | |
| "step": 924000 | |
| }, | |
| { | |
| "epoch": 23.17, | |
| "learning_rate": 6.900379188875779e-05, | |
| "loss": 0.3269, | |
| "step": 927000 | |
| }, | |
| { | |
| "epoch": 23.24, | |
| "learning_rate": 6.824642396110157e-05, | |
| "loss": 0.324, | |
| "step": 930000 | |
| }, | |
| { | |
| "epoch": 23.32, | |
| "learning_rate": 6.748905603344537e-05, | |
| "loss": 0.3249, | |
| "step": 933000 | |
| }, | |
| { | |
| "epoch": 23.39, | |
| "learning_rate": 6.673168810578914e-05, | |
| "loss": 0.3257, | |
| "step": 936000 | |
| }, | |
| { | |
| "epoch": 23.47, | |
| "learning_rate": 6.597432017813294e-05, | |
| "loss": 0.3266, | |
| "step": 939000 | |
| }, | |
| { | |
| "epoch": 23.54, | |
| "learning_rate": 6.521695225047672e-05, | |
| "loss": 0.3275, | |
| "step": 942000 | |
| }, | |
| { | |
| "epoch": 23.62, | |
| "learning_rate": 6.44595843228205e-05, | |
| "loss": 0.3279, | |
| "step": 945000 | |
| }, | |
| { | |
| "epoch": 23.69, | |
| "learning_rate": 6.370221639516429e-05, | |
| "loss": 0.3255, | |
| "step": 948000 | |
| }, | |
| { | |
| "epoch": 23.77, | |
| "learning_rate": 6.294484846750807e-05, | |
| "loss": 0.3255, | |
| "step": 951000 | |
| }, | |
| { | |
| "epoch": 23.84, | |
| "learning_rate": 6.218748053985185e-05, | |
| "loss": 0.3233, | |
| "step": 954000 | |
| }, | |
| { | |
| "epoch": 23.92, | |
| "learning_rate": 6.143011261219564e-05, | |
| "loss": 0.3267, | |
| "step": 957000 | |
| }, | |
| { | |
| "epoch": 23.99, | |
| "learning_rate": 6.067274468453942e-05, | |
| "loss": 0.3234, | |
| "step": 960000 | |
| }, | |
| { | |
| "epoch": 24.07, | |
| "learning_rate": 5.991537675688321e-05, | |
| "loss": 0.3174, | |
| "step": 963000 | |
| }, | |
| { | |
| "epoch": 24.14, | |
| "learning_rate": 5.9158008829226994e-05, | |
| "loss": 0.3149, | |
| "step": 966000 | |
| }, | |
| { | |
| "epoch": 24.22, | |
| "learning_rate": 5.840064090157077e-05, | |
| "loss": 0.314, | |
| "step": 969000 | |
| }, | |
| { | |
| "epoch": 24.29, | |
| "learning_rate": 5.764327297391456e-05, | |
| "loss": 0.318, | |
| "step": 972000 | |
| }, | |
| { | |
| "epoch": 24.37, | |
| "learning_rate": 5.6885905046258344e-05, | |
| "loss": 0.3147, | |
| "step": 975000 | |
| }, | |
| { | |
| "epoch": 24.44, | |
| "learning_rate": 5.6128537118602134e-05, | |
| "loss": 0.3158, | |
| "step": 978000 | |
| }, | |
| { | |
| "epoch": 24.52, | |
| "learning_rate": 5.537116919094591e-05, | |
| "loss": 0.3164, | |
| "step": 981000 | |
| }, | |
| { | |
| "epoch": 24.59, | |
| "learning_rate": 5.46138012632897e-05, | |
| "loss": 0.3162, | |
| "step": 984000 | |
| }, | |
| { | |
| "epoch": 24.67, | |
| "learning_rate": 5.3856433335633483e-05, | |
| "loss": 0.3153, | |
| "step": 987000 | |
| }, | |
| { | |
| "epoch": 24.74, | |
| "learning_rate": 5.3099065407977274e-05, | |
| "loss": 0.3165, | |
| "step": 990000 | |
| }, | |
| { | |
| "epoch": 24.82, | |
| "learning_rate": 5.234169748032105e-05, | |
| "loss": 0.315, | |
| "step": 993000 | |
| }, | |
| { | |
| "epoch": 24.89, | |
| "learning_rate": 5.158432955266483e-05, | |
| "loss": 0.3132, | |
| "step": 996000 | |
| }, | |
| { | |
| "epoch": 24.97, | |
| "learning_rate": 5.0826961625008623e-05, | |
| "loss": 0.3163, | |
| "step": 999000 | |
| }, | |
| { | |
| "epoch": 25.04, | |
| "learning_rate": 5.006959369735241e-05, | |
| "loss": 0.3107, | |
| "step": 1002000 | |
| }, | |
| { | |
| "epoch": 25.12, | |
| "learning_rate": 4.931222576969619e-05, | |
| "loss": 0.3062, | |
| "step": 1005000 | |
| }, | |
| { | |
| "epoch": 25.19, | |
| "learning_rate": 4.855485784203997e-05, | |
| "loss": 0.3083, | |
| "step": 1008000 | |
| }, | |
| { | |
| "epoch": 25.27, | |
| "learning_rate": 4.779748991438376e-05, | |
| "loss": 0.3096, | |
| "step": 1011000 | |
| }, | |
| { | |
| "epoch": 25.34, | |
| "learning_rate": 4.704012198672755e-05, | |
| "loss": 0.3056, | |
| "step": 1014000 | |
| }, | |
| { | |
| "epoch": 25.42, | |
| "learning_rate": 4.628275405907132e-05, | |
| "loss": 0.3042, | |
| "step": 1017000 | |
| }, | |
| { | |
| "epoch": 25.49, | |
| "learning_rate": 4.552538613141511e-05, | |
| "loss": 0.3063, | |
| "step": 1020000 | |
| }, | |
| { | |
| "epoch": 25.57, | |
| "learning_rate": 4.4768018203758897e-05, | |
| "loss": 0.3025, | |
| "step": 1023000 | |
| }, | |
| { | |
| "epoch": 25.64, | |
| "learning_rate": 4.4010650276102687e-05, | |
| "loss": 0.3056, | |
| "step": 1026000 | |
| }, | |
| { | |
| "epoch": 25.72, | |
| "learning_rate": 4.325328234844646e-05, | |
| "loss": 0.3081, | |
| "step": 1029000 | |
| }, | |
| { | |
| "epoch": 25.79, | |
| "learning_rate": 4.249591442079025e-05, | |
| "loss": 0.3082, | |
| "step": 1032000 | |
| }, | |
| { | |
| "epoch": 25.87, | |
| "learning_rate": 4.1738546493134036e-05, | |
| "loss": 0.3054, | |
| "step": 1035000 | |
| }, | |
| { | |
| "epoch": 25.94, | |
| "learning_rate": 4.0981178565477827e-05, | |
| "loss": 0.3052, | |
| "step": 1038000 | |
| }, | |
| { | |
| "epoch": 26.02, | |
| "learning_rate": 4.02238106378216e-05, | |
| "loss": 0.3013, | |
| "step": 1041000 | |
| }, | |
| { | |
| "epoch": 26.09, | |
| "learning_rate": 3.9466442710165386e-05, | |
| "loss": 0.2982, | |
| "step": 1044000 | |
| }, | |
| { | |
| "epoch": 26.17, | |
| "learning_rate": 3.8709074782509176e-05, | |
| "loss": 0.2962, | |
| "step": 1047000 | |
| }, | |
| { | |
| "epoch": 26.24, | |
| "learning_rate": 3.795170685485296e-05, | |
| "loss": 0.2961, | |
| "step": 1050000 | |
| }, | |
| { | |
| "epoch": 26.32, | |
| "learning_rate": 3.719433892719674e-05, | |
| "loss": 0.2993, | |
| "step": 1053000 | |
| }, | |
| { | |
| "epoch": 26.39, | |
| "learning_rate": 3.6436970999540526e-05, | |
| "loss": 0.2941, | |
| "step": 1056000 | |
| }, | |
| { | |
| "epoch": 26.47, | |
| "learning_rate": 3.5679603071884316e-05, | |
| "loss": 0.2948, | |
| "step": 1059000 | |
| }, | |
| { | |
| "epoch": 26.54, | |
| "learning_rate": 3.492223514422809e-05, | |
| "loss": 0.2966, | |
| "step": 1062000 | |
| }, | |
| { | |
| "epoch": 26.62, | |
| "learning_rate": 3.4164867216571876e-05, | |
| "loss": 0.2964, | |
| "step": 1065000 | |
| }, | |
| { | |
| "epoch": 26.69, | |
| "learning_rate": 3.3407499288915666e-05, | |
| "loss": 0.2971, | |
| "step": 1068000 | |
| }, | |
| { | |
| "epoch": 26.77, | |
| "learning_rate": 3.265013136125945e-05, | |
| "loss": 0.2974, | |
| "step": 1071000 | |
| }, | |
| { | |
| "epoch": 26.84, | |
| "learning_rate": 3.189276343360323e-05, | |
| "loss": 0.2963, | |
| "step": 1074000 | |
| }, | |
| { | |
| "epoch": 26.92, | |
| "learning_rate": 3.1135395505947016e-05, | |
| "loss": 0.2954, | |
| "step": 1077000 | |
| }, | |
| { | |
| "epoch": 26.99, | |
| "learning_rate": 3.0378027578290803e-05, | |
| "loss": 0.2941, | |
| "step": 1080000 | |
| }, | |
| { | |
| "epoch": 27.07, | |
| "learning_rate": 2.9620659650634586e-05, | |
| "loss": 0.2896, | |
| "step": 1083000 | |
| }, | |
| { | |
| "epoch": 27.14, | |
| "learning_rate": 2.8863291722978373e-05, | |
| "loss": 0.289, | |
| "step": 1086000 | |
| }, | |
| { | |
| "epoch": 27.22, | |
| "learning_rate": 2.8105923795322156e-05, | |
| "loss": 0.2899, | |
| "step": 1089000 | |
| }, | |
| { | |
| "epoch": 27.29, | |
| "learning_rate": 2.7348555867665943e-05, | |
| "loss": 0.2878, | |
| "step": 1092000 | |
| }, | |
| { | |
| "epoch": 27.37, | |
| "learning_rate": 2.6591187940009723e-05, | |
| "loss": 0.2858, | |
| "step": 1095000 | |
| }, | |
| { | |
| "epoch": 27.44, | |
| "learning_rate": 2.583382001235351e-05, | |
| "loss": 0.2901, | |
| "step": 1098000 | |
| }, | |
| { | |
| "epoch": 27.52, | |
| "learning_rate": 2.5076452084697293e-05, | |
| "loss": 0.2885, | |
| "step": 1101000 | |
| }, | |
| { | |
| "epoch": 27.59, | |
| "learning_rate": 2.431908415704108e-05, | |
| "loss": 0.2903, | |
| "step": 1104000 | |
| }, | |
| { | |
| "epoch": 27.67, | |
| "learning_rate": 2.3561716229384863e-05, | |
| "loss": 0.2904, | |
| "step": 1107000 | |
| }, | |
| { | |
| "epoch": 27.74, | |
| "learning_rate": 2.280434830172865e-05, | |
| "loss": 0.2869, | |
| "step": 1110000 | |
| }, | |
| { | |
| "epoch": 27.82, | |
| "learning_rate": 2.2046980374072432e-05, | |
| "loss": 0.2902, | |
| "step": 1113000 | |
| }, | |
| { | |
| "epoch": 27.89, | |
| "learning_rate": 2.128961244641622e-05, | |
| "loss": 0.2867, | |
| "step": 1116000 | |
| }, | |
| { | |
| "epoch": 27.97, | |
| "learning_rate": 2.053224451876e-05, | |
| "loss": 0.2869, | |
| "step": 1119000 | |
| }, | |
| { | |
| "epoch": 28.04, | |
| "learning_rate": 1.9774876591103786e-05, | |
| "loss": 0.2844, | |
| "step": 1122000 | |
| }, | |
| { | |
| "epoch": 28.12, | |
| "learning_rate": 1.901750866344757e-05, | |
| "loss": 0.283, | |
| "step": 1125000 | |
| }, | |
| { | |
| "epoch": 28.19, | |
| "learning_rate": 1.8260140735791356e-05, | |
| "loss": 0.2816, | |
| "step": 1128000 | |
| }, | |
| { | |
| "epoch": 28.27, | |
| "learning_rate": 1.750277280813514e-05, | |
| "loss": 0.2848, | |
| "step": 1131000 | |
| }, | |
| { | |
| "epoch": 28.34, | |
| "learning_rate": 1.6745404880478922e-05, | |
| "loss": 0.2811, | |
| "step": 1134000 | |
| }, | |
| { | |
| "epoch": 28.42, | |
| "learning_rate": 1.598803695282271e-05, | |
| "loss": 0.283, | |
| "step": 1137000 | |
| }, | |
| { | |
| "epoch": 28.49, | |
| "learning_rate": 1.5230669025166492e-05, | |
| "loss": 0.2814, | |
| "step": 1140000 | |
| }, | |
| { | |
| "epoch": 28.57, | |
| "learning_rate": 1.4473301097510277e-05, | |
| "loss": 0.2793, | |
| "step": 1143000 | |
| }, | |
| { | |
| "epoch": 28.64, | |
| "learning_rate": 1.3715933169854062e-05, | |
| "loss": 0.2834, | |
| "step": 1146000 | |
| }, | |
| { | |
| "epoch": 28.72, | |
| "learning_rate": 1.2958565242197847e-05, | |
| "loss": 0.2798, | |
| "step": 1149000 | |
| }, | |
| { | |
| "epoch": 28.79, | |
| "learning_rate": 1.220119731454163e-05, | |
| "loss": 0.2807, | |
| "step": 1152000 | |
| }, | |
| { | |
| "epoch": 28.87, | |
| "learning_rate": 1.1443829386885416e-05, | |
| "loss": 0.28, | |
| "step": 1155000 | |
| }, | |
| { | |
| "epoch": 28.94, | |
| "learning_rate": 1.06864614592292e-05, | |
| "loss": 0.2762, | |
| "step": 1158000 | |
| }, | |
| { | |
| "epoch": 29.02, | |
| "learning_rate": 9.929093531572985e-06, | |
| "loss": 0.2802, | |
| "step": 1161000 | |
| }, | |
| { | |
| "epoch": 29.09, | |
| "learning_rate": 9.171725603916769e-06, | |
| "loss": 0.2761, | |
| "step": 1164000 | |
| }, | |
| { | |
| "epoch": 29.17, | |
| "learning_rate": 8.414357676260554e-06, | |
| "loss": 0.2771, | |
| "step": 1167000 | |
| }, | |
| { | |
| "epoch": 29.24, | |
| "learning_rate": 7.656989748604339e-06, | |
| "loss": 0.2777, | |
| "step": 1170000 | |
| }, | |
| { | |
| "epoch": 29.32, | |
| "learning_rate": 6.899621820948123e-06, | |
| "loss": 0.2767, | |
| "step": 1173000 | |
| }, | |
| { | |
| "epoch": 29.39, | |
| "learning_rate": 6.142253893291908e-06, | |
| "loss": 0.2773, | |
| "step": 1176000 | |
| }, | |
| { | |
| "epoch": 29.47, | |
| "learning_rate": 5.384885965635692e-06, | |
| "loss": 0.2765, | |
| "step": 1179000 | |
| }, | |
| { | |
| "epoch": 29.54, | |
| "learning_rate": 4.627518037979477e-06, | |
| "loss": 0.2786, | |
| "step": 1182000 | |
| }, | |
| { | |
| "epoch": 29.62, | |
| "learning_rate": 3.870150110323261e-06, | |
| "loss": 0.2768, | |
| "step": 1185000 | |
| }, | |
| { | |
| "epoch": 29.69, | |
| "learning_rate": 3.1127821826670457e-06, | |
| "loss": 0.2771, | |
| "step": 1188000 | |
| }, | |
| { | |
| "epoch": 29.77, | |
| "learning_rate": 2.35541425501083e-06, | |
| "loss": 0.2737, | |
| "step": 1191000 | |
| }, | |
| { | |
| "epoch": 29.84, | |
| "learning_rate": 1.5980463273546148e-06, | |
| "loss": 0.2737, | |
| "step": 1194000 | |
| }, | |
| { | |
| "epoch": 29.92, | |
| "learning_rate": 8.406783996983991e-07, | |
| "loss": 0.2781, | |
| "step": 1197000 | |
| }, | |
| { | |
| "epoch": 29.99, | |
| "learning_rate": 8.33104720421837e-08, | |
| "loss": 0.2745, | |
| "step": 1200000 | |
| }, | |
| { | |
| "epoch": 30.0, | |
| "step": 1200330, | |
| "total_flos": 2.468458772043806e+21, | |
| "train_loss": 0.5222103027782066, | |
| "train_runtime": 708527.1323, | |
| "train_samples_per_second": 27.106, | |
| "train_steps_per_second": 1.694 | |
| } | |
| ], | |
| "logging_steps": 3000, | |
| "max_steps": 1200330, | |
| "num_train_epochs": 30, | |
| "save_steps": 500, | |
| "total_flos": 2.468458772043806e+21, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |