| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 0.96, | |
| "eval_steps": 500, | |
| "global_step": 1000, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0, | |
| "grad_norm": 0.8561736941337585, | |
| "learning_rate": 1.9999873507471375e-05, | |
| "loss": 1.0407, | |
| "step": 5 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 0.9348568320274353, | |
| "learning_rate": 1.9999494033085565e-05, | |
| "loss": 1.021, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.01, | |
| "grad_norm": 1.0065371990203857, | |
| "learning_rate": 1.999886158644271e-05, | |
| "loss": 0.9471, | |
| "step": 15 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 1.0246785879135132, | |
| "learning_rate": 1.999797618354276e-05, | |
| "loss": 0.9153, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.02, | |
| "grad_norm": 1.1248774528503418, | |
| "learning_rate": 1.9996837846785094e-05, | |
| "loss": 0.8506, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 0.9718716740608215, | |
| "learning_rate": 1.999544660496792e-05, | |
| "loss": 0.805, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.03, | |
| "grad_norm": 0.670078456401825, | |
| "learning_rate": 1.9993802493287588e-05, | |
| "loss": 0.756, | |
| "step": 35 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 0.5407480597496033, | |
| "learning_rate": 1.9991905553337657e-05, | |
| "loss": 0.7278, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.04, | |
| "grad_norm": 0.5028639435768127, | |
| "learning_rate": 1.9989755833107875e-05, | |
| "loss": 0.7153, | |
| "step": 45 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 0.4602053761482239, | |
| "learning_rate": 1.998735338698296e-05, | |
| "loss": 0.6868, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.05, | |
| "grad_norm": 0.47076714038848877, | |
| "learning_rate": 1.9984698275741198e-05, | |
| "loss": 0.6744, | |
| "step": 55 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 0.38986819982528687, | |
| "learning_rate": 1.9981790566552944e-05, | |
| "loss": 0.6422, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.06, | |
| "grad_norm": 0.36573711037635803, | |
| "learning_rate": 1.997863033297889e-05, | |
| "loss": 0.6242, | |
| "step": 65 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 0.31471285223960876, | |
| "learning_rate": 1.9975217654968228e-05, | |
| "loss": 0.6036, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.07, | |
| "grad_norm": 0.34020736813545227, | |
| "learning_rate": 1.9971552618856612e-05, | |
| "loss": 0.6003, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 0.3019075393676758, | |
| "learning_rate": 1.996763531736397e-05, | |
| "loss": 0.5907, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.08, | |
| "grad_norm": 0.2804803252220154, | |
| "learning_rate": 1.996346584959219e-05, | |
| "loss": 0.5733, | |
| "step": 85 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 0.26424163579940796, | |
| "learning_rate": 1.9959044321022563e-05, | |
| "loss": 0.5826, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.09, | |
| "grad_norm": 0.27084848284721375, | |
| "learning_rate": 1.9954370843513167e-05, | |
| "loss": 0.5781, | |
| "step": 95 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 0.25091850757598877, | |
| "learning_rate": 1.9949445535295992e-05, | |
| "loss": 0.5667, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.1, | |
| "grad_norm": 0.2530795633792877, | |
| "learning_rate": 1.9944268520973978e-05, | |
| "loss": 0.5635, | |
| "step": 105 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 0.23881025612354279, | |
| "learning_rate": 1.993883993151785e-05, | |
| "loss": 0.5643, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.11, | |
| "grad_norm": 0.22561447322368622, | |
| "learning_rate": 1.9933159904262812e-05, | |
| "loss": 0.5257, | |
| "step": 115 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 0.21529339253902435, | |
| "learning_rate": 1.9927228582905063e-05, | |
| "loss": 0.5387, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 0.21853545308113098, | |
| "learning_rate": 1.9921046117498175e-05, | |
| "loss": 0.5363, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.12, | |
| "grad_norm": 0.20658166706562042, | |
| "learning_rate": 1.991461266444928e-05, | |
| "loss": 0.5447, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "grad_norm": 0.20857058465480804, | |
| "learning_rate": 1.9907928386515126e-05, | |
| "loss": 0.5226, | |
| "step": 135 | |
| }, | |
| { | |
| "epoch": 0.13, | |
| "grad_norm": 0.21309202909469604, | |
| "learning_rate": 1.9900993452797964e-05, | |
| "loss": 0.5234, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "grad_norm": 0.1945490837097168, | |
| "learning_rate": 1.989380803874125e-05, | |
| "loss": 0.5073, | |
| "step": 145 | |
| }, | |
| { | |
| "epoch": 0.14, | |
| "grad_norm": 0.21293672919273376, | |
| "learning_rate": 1.9886372326125215e-05, | |
| "loss": 0.508, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "grad_norm": 0.19945946335792542, | |
| "learning_rate": 1.987868650306229e-05, | |
| "loss": 0.5119, | |
| "step": 155 | |
| }, | |
| { | |
| "epoch": 0.15, | |
| "grad_norm": 0.19202299416065216, | |
| "learning_rate": 1.9870750763992306e-05, | |
| "loss": 0.5139, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 0.18136940896511078, | |
| "learning_rate": 1.9862565309677606e-05, | |
| "loss": 0.4982, | |
| "step": 165 | |
| }, | |
| { | |
| "epoch": 0.16, | |
| "grad_norm": 0.20753854513168335, | |
| "learning_rate": 1.9854130347197954e-05, | |
| "loss": 0.4979, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "grad_norm": 0.19880987703800201, | |
| "learning_rate": 1.9845446089945294e-05, | |
| "loss": 0.5069, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.17, | |
| "grad_norm": 0.18631042540073395, | |
| "learning_rate": 1.9836512757618355e-05, | |
| "loss": 0.4827, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "grad_norm": 0.17570152878761292, | |
| "learning_rate": 1.9827330576217105e-05, | |
| "loss": 0.4638, | |
| "step": 185 | |
| }, | |
| { | |
| "epoch": 0.18, | |
| "grad_norm": 0.1938246786594391, | |
| "learning_rate": 1.9817899778037e-05, | |
| "loss": 0.4866, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "grad_norm": 0.17978984117507935, | |
| "learning_rate": 1.9808220601663155e-05, | |
| "loss": 0.4911, | |
| "step": 195 | |
| }, | |
| { | |
| "epoch": 0.19, | |
| "grad_norm": 0.17453525960445404, | |
| "learning_rate": 1.9798293291964265e-05, | |
| "loss": 0.4841, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 0.19657620787620544, | |
| "learning_rate": 1.9788118100086423e-05, | |
| "loss": 0.5002, | |
| "step": 205 | |
| }, | |
| { | |
| "epoch": 0.2, | |
| "grad_norm": 0.18103839457035065, | |
| "learning_rate": 1.9777695283446787e-05, | |
| "loss": 0.5118, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "grad_norm": 0.17228654026985168, | |
| "learning_rate": 1.9767025105727043e-05, | |
| "loss": 0.4707, | |
| "step": 215 | |
| }, | |
| { | |
| "epoch": 0.21, | |
| "grad_norm": 0.17088425159454346, | |
| "learning_rate": 1.975610783686674e-05, | |
| "loss": 0.4884, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "grad_norm": 0.19012100994586945, | |
| "learning_rate": 1.974494375305647e-05, | |
| "loss": 0.4865, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.22, | |
| "grad_norm": 0.1744072586297989, | |
| "learning_rate": 1.973353313673087e-05, | |
| "loss": 0.4884, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "grad_norm": 0.17286820709705353, | |
| "learning_rate": 1.9721876276561485e-05, | |
| "loss": 0.4828, | |
| "step": 235 | |
| }, | |
| { | |
| "epoch": 0.23, | |
| "grad_norm": 0.1775512546300888, | |
| "learning_rate": 1.970997346744945e-05, | |
| "loss": 0.4811, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "grad_norm": 0.16784435510635376, | |
| "learning_rate": 1.969782501051806e-05, | |
| "loss": 0.4736, | |
| "step": 245 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "grad_norm": 0.1838577836751938, | |
| "learning_rate": 1.9685431213105115e-05, | |
| "loss": 0.4844, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.24, | |
| "grad_norm": 0.17393770813941956, | |
| "learning_rate": 1.9672792388755174e-05, | |
| "loss": 0.4763, | |
| "step": 255 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "grad_norm": 0.1882689893245697, | |
| "learning_rate": 1.9659908857211606e-05, | |
| "loss": 0.4756, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "grad_norm": 0.16501489281654358, | |
| "learning_rate": 1.9646780944408507e-05, | |
| "loss": 0.4573, | |
| "step": 265 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "grad_norm": 0.16918818652629852, | |
| "learning_rate": 1.9633408982462453e-05, | |
| "loss": 0.4636, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.26, | |
| "grad_norm": 0.17074143886566162, | |
| "learning_rate": 1.9619793309664097e-05, | |
| "loss": 0.4709, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "grad_norm": 0.18831667304039001, | |
| "learning_rate": 1.9605934270469625e-05, | |
| "loss": 0.4631, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.27, | |
| "grad_norm": 0.19625447690486908, | |
| "learning_rate": 1.9591832215492018e-05, | |
| "loss": 0.4719, | |
| "step": 285 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "grad_norm": 0.1936866044998169, | |
| "learning_rate": 1.9577487501492183e-05, | |
| "loss": 0.468, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.28, | |
| "grad_norm": 0.16921108961105347, | |
| "learning_rate": 1.956290049136996e-05, | |
| "loss": 0.4516, | |
| "step": 295 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "grad_norm": 0.18268905580043793, | |
| "learning_rate": 1.9548071554154905e-05, | |
| "loss": 0.4558, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.29, | |
| "grad_norm": 0.16877137124538422, | |
| "learning_rate": 1.953300106499697e-05, | |
| "loss": 0.4764, | |
| "step": 305 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "grad_norm": 0.17508965730667114, | |
| "learning_rate": 1.951768940515701e-05, | |
| "loss": 0.4485, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.3, | |
| "grad_norm": 0.16913452744483948, | |
| "learning_rate": 1.9502136961997144e-05, | |
| "loss": 0.4635, | |
| "step": 315 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "grad_norm": 0.1623164862394333, | |
| "learning_rate": 1.948634412897094e-05, | |
| "loss": 0.4592, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.31, | |
| "grad_norm": 0.19096989929676056, | |
| "learning_rate": 1.9470311305613478e-05, | |
| "loss": 0.4768, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 0.19626876711845398, | |
| "learning_rate": 1.9454038897531226e-05, | |
| "loss": 0.4615, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.32, | |
| "grad_norm": 0.1713666319847107, | |
| "learning_rate": 1.9437527316391802e-05, | |
| "loss": 0.4583, | |
| "step": 335 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "grad_norm": 0.17599208652973175, | |
| "learning_rate": 1.9420776979913528e-05, | |
| "loss": 0.4552, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.33, | |
| "grad_norm": 0.18346691131591797, | |
| "learning_rate": 1.940378831185489e-05, | |
| "loss": 0.4605, | |
| "step": 345 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "grad_norm": 0.2010510265827179, | |
| "learning_rate": 1.9386561742003804e-05, | |
| "loss": 0.48, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.34, | |
| "grad_norm": 0.16948485374450684, | |
| "learning_rate": 1.936909770616675e-05, | |
| "loss": 0.4501, | |
| "step": 355 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "grad_norm": 0.18418210744857788, | |
| "learning_rate": 1.935139664615773e-05, | |
| "loss": 0.4604, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.35, | |
| "grad_norm": 0.1840277910232544, | |
| "learning_rate": 1.933345900978712e-05, | |
| "loss": 0.4586, | |
| "step": 365 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 0.1805981695652008, | |
| "learning_rate": 1.9315285250850312e-05, | |
| "loss": 0.452, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 0.1838221549987793, | |
| "learning_rate": 1.9296875829116252e-05, | |
| "loss": 0.4571, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.36, | |
| "grad_norm": 0.17758053541183472, | |
| "learning_rate": 1.92782312103158e-05, | |
| "loss": 0.4927, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "grad_norm": 0.16368024051189423, | |
| "learning_rate": 1.9259351866129955e-05, | |
| "loss": 0.4504, | |
| "step": 385 | |
| }, | |
| { | |
| "epoch": 0.37, | |
| "grad_norm": 0.17798829078674316, | |
| "learning_rate": 1.9240238274177907e-05, | |
| "loss": 0.4422, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "grad_norm": 0.16979840397834778, | |
| "learning_rate": 1.9220890918004978e-05, | |
| "loss": 0.4333, | |
| "step": 395 | |
| }, | |
| { | |
| "epoch": 0.38, | |
| "grad_norm": 0.1931125670671463, | |
| "learning_rate": 1.9201310287070368e-05, | |
| "loss": 0.4312, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "grad_norm": 0.18460631370544434, | |
| "learning_rate": 1.918149687673478e-05, | |
| "loss": 0.4634, | |
| "step": 405 | |
| }, | |
| { | |
| "epoch": 0.39, | |
| "grad_norm": 0.16709978878498077, | |
| "learning_rate": 1.9161451188247887e-05, | |
| "loss": 0.4451, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 0.19406840205192566, | |
| "learning_rate": 1.914117372873565e-05, | |
| "loss": 0.4542, | |
| "step": 415 | |
| }, | |
| { | |
| "epoch": 0.4, | |
| "grad_norm": 0.16695013642311096, | |
| "learning_rate": 1.912066501118751e-05, | |
| "loss": 0.4343, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "grad_norm": 0.18530474603176117, | |
| "learning_rate": 1.9099925554443357e-05, | |
| "loss": 0.444, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.41, | |
| "grad_norm": 0.18910014629364014, | |
| "learning_rate": 1.9078955883180466e-05, | |
| "loss": 0.4575, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "grad_norm": 0.19215096533298492, | |
| "learning_rate": 1.9057756527900188e-05, | |
| "loss": 0.4587, | |
| "step": 435 | |
| }, | |
| { | |
| "epoch": 0.42, | |
| "grad_norm": 0.18122592568397522, | |
| "learning_rate": 1.9036328024914525e-05, | |
| "loss": 0.4381, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "grad_norm": 0.17227929830551147, | |
| "learning_rate": 1.9014670916332593e-05, | |
| "loss": 0.4473, | |
| "step": 445 | |
| }, | |
| { | |
| "epoch": 0.43, | |
| "grad_norm": 0.17411768436431885, | |
| "learning_rate": 1.8992785750046866e-05, | |
| "loss": 0.4408, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "grad_norm": 0.18275217711925507, | |
| "learning_rate": 1.8970673079719357e-05, | |
| "loss": 0.4493, | |
| "step": 455 | |
| }, | |
| { | |
| "epoch": 0.44, | |
| "grad_norm": 0.1775893270969391, | |
| "learning_rate": 1.8948333464767578e-05, | |
| "loss": 0.4654, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "grad_norm": 0.16952849924564362, | |
| "learning_rate": 1.892576747035041e-05, | |
| "loss": 0.4403, | |
| "step": 465 | |
| }, | |
| { | |
| "epoch": 0.45, | |
| "grad_norm": 0.1598004549741745, | |
| "learning_rate": 1.890297566735379e-05, | |
| "loss": 0.4313, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "grad_norm": 0.17070458829402924, | |
| "learning_rate": 1.8879958632376276e-05, | |
| "loss": 0.4362, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.46, | |
| "grad_norm": 0.18404103815555573, | |
| "learning_rate": 1.885671694771446e-05, | |
| "loss": 0.441, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "grad_norm": 0.20091940462589264, | |
| "learning_rate": 1.8833251201348233e-05, | |
| "loss": 0.4354, | |
| "step": 485 | |
| }, | |
| { | |
| "epoch": 0.47, | |
| "grad_norm": 0.1788869947195053, | |
| "learning_rate": 1.8809561986925917e-05, | |
| "loss": 0.4617, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 0.20356979966163635, | |
| "learning_rate": 1.8785649903749236e-05, | |
| "loss": 0.4338, | |
| "step": 495 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 0.17706063389778137, | |
| "learning_rate": 1.8761515556758162e-05, | |
| "loss": 0.4566, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.48, | |
| "grad_norm": 0.17191752791404724, | |
| "learning_rate": 1.8737159556515614e-05, | |
| "loss": 0.4416, | |
| "step": 505 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "grad_norm": 0.19822804629802704, | |
| "learning_rate": 1.8712582519192e-05, | |
| "loss": 0.4592, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.49, | |
| "grad_norm": 0.19628900289535522, | |
| "learning_rate": 1.868778506654964e-05, | |
| "loss": 0.4463, | |
| "step": 515 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 0.20023687183856964, | |
| "learning_rate": 1.8662767825927038e-05, | |
| "loss": 0.4313, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 0.1745893806219101, | |
| "learning_rate": 1.8637531430222993e-05, | |
| "loss": 0.4451, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "grad_norm": 0.18421079218387604, | |
| "learning_rate": 1.8612076517880605e-05, | |
| "loss": 0.4569, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.51, | |
| "grad_norm": 0.1991654932498932, | |
| "learning_rate": 1.8586403732871124e-05, | |
| "loss": 0.4381, | |
| "step": 535 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "grad_norm": 0.18618810176849365, | |
| "learning_rate": 1.8560513724677644e-05, | |
| "loss": 0.447, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.52, | |
| "grad_norm": 0.19722780585289001, | |
| "learning_rate": 1.853440714827869e-05, | |
| "loss": 0.4499, | |
| "step": 545 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "grad_norm": 0.19388031959533691, | |
| "learning_rate": 1.850808466413163e-05, | |
| "loss": 0.4504, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.53, | |
| "grad_norm": 0.1990053355693817, | |
| "learning_rate": 1.8481546938155988e-05, | |
| "loss": 0.4414, | |
| "step": 555 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "grad_norm": 0.2028382122516632, | |
| "learning_rate": 1.8454794641716567e-05, | |
| "loss": 0.4319, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.54, | |
| "grad_norm": 0.21612244844436646, | |
| "learning_rate": 1.84278284516065e-05, | |
| "loss": 0.4538, | |
| "step": 565 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "grad_norm": 0.18235105276107788, | |
| "learning_rate": 1.840064905003009e-05, | |
| "loss": 0.4421, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.55, | |
| "grad_norm": 0.21971595287322998, | |
| "learning_rate": 1.8373257124585592e-05, | |
| "loss": 0.4354, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "grad_norm": 0.1827440857887268, | |
| "learning_rate": 1.834565336824779e-05, | |
| "loss": 0.4234, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.56, | |
| "grad_norm": 0.18092140555381775, | |
| "learning_rate": 1.8317838479350473e-05, | |
| "loss": 0.439, | |
| "step": 585 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "grad_norm": 0.1922396719455719, | |
| "learning_rate": 1.8289813161568758e-05, | |
| "loss": 0.4387, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.57, | |
| "grad_norm": 0.18677982687950134, | |
| "learning_rate": 1.8261578123901314e-05, | |
| "loss": 0.4346, | |
| "step": 595 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "grad_norm": 0.1832067370414734, | |
| "learning_rate": 1.82331340806524e-05, | |
| "loss": 0.434, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.58, | |
| "grad_norm": 0.19588692486286163, | |
| "learning_rate": 1.8204481751413813e-05, | |
| "loss": 0.4298, | |
| "step": 605 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "grad_norm": 0.19335758686065674, | |
| "learning_rate": 1.817562186104666e-05, | |
| "loss": 0.4388, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.59, | |
| "grad_norm": 0.18936721980571747, | |
| "learning_rate": 1.814655513966305e-05, | |
| "loss": 0.4418, | |
| "step": 615 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 0.18840767443180084, | |
| "learning_rate": 1.8117282322607596e-05, | |
| "loss": 0.418, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 0.19814783334732056, | |
| "learning_rate": 1.808780415043883e-05, | |
| "loss": 0.4368, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.6, | |
| "grad_norm": 0.19724276661872864, | |
| "learning_rate": 1.805812136891046e-05, | |
| "loss": 0.4474, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "grad_norm": 0.21561075747013092, | |
| "learning_rate": 1.8028234728952496e-05, | |
| "loss": 0.4333, | |
| "step": 635 | |
| }, | |
| { | |
| "epoch": 0.61, | |
| "grad_norm": 0.21301332116127014, | |
| "learning_rate": 1.799814498665228e-05, | |
| "loss": 0.4186, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "grad_norm": 0.21463420987129211, | |
| "learning_rate": 1.7967852903235328e-05, | |
| "loss": 0.4549, | |
| "step": 645 | |
| }, | |
| { | |
| "epoch": 0.62, | |
| "grad_norm": 0.19322441518306732, | |
| "learning_rate": 1.7937359245046083e-05, | |
| "loss": 0.4399, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "grad_norm": 0.1851872205734253, | |
| "learning_rate": 1.7906664783528533e-05, | |
| "loss": 0.4314, | |
| "step": 655 | |
| }, | |
| { | |
| "epoch": 0.63, | |
| "grad_norm": 0.2039497047662735, | |
| "learning_rate": 1.787577029520669e-05, | |
| "loss": 0.4268, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "grad_norm": 0.17417435348033905, | |
| "learning_rate": 1.784467656166494e-05, | |
| "loss": 0.448, | |
| "step": 665 | |
| }, | |
| { | |
| "epoch": 0.64, | |
| "grad_norm": 0.1900678277015686, | |
| "learning_rate": 1.7813384369528285e-05, | |
| "loss": 0.4182, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "grad_norm": 0.1949600726366043, | |
| "learning_rate": 1.778189451044242e-05, | |
| "loss": 0.433, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.65, | |
| "grad_norm": 0.19561733305454254, | |
| "learning_rate": 1.7750207781053727e-05, | |
| "loss": 0.4251, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "grad_norm": 0.20206551253795624, | |
| "learning_rate": 1.7718324982989113e-05, | |
| "loss": 0.4379, | |
| "step": 685 | |
| }, | |
| { | |
| "epoch": 0.66, | |
| "grad_norm": 0.18985505402088165, | |
| "learning_rate": 1.7686246922835728e-05, | |
| "loss": 0.4164, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "grad_norm": 0.19377106428146362, | |
| "learning_rate": 1.7653974412120556e-05, | |
| "loss": 0.4318, | |
| "step": 695 | |
| }, | |
| { | |
| "epoch": 0.67, | |
| "grad_norm": 0.20084458589553833, | |
| "learning_rate": 1.76215082672899e-05, | |
| "loss": 0.4331, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 0.20488500595092773, | |
| "learning_rate": 1.7588849309688704e-05, | |
| "loss": 0.4377, | |
| "step": 705 | |
| }, | |
| { | |
| "epoch": 0.68, | |
| "grad_norm": 0.17935730516910553, | |
| "learning_rate": 1.75559983655398e-05, | |
| "loss": 0.4246, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "grad_norm": 0.1966232806444168, | |
| "learning_rate": 1.7522956265922983e-05, | |
| "loss": 0.439, | |
| "step": 715 | |
| }, | |
| { | |
| "epoch": 0.69, | |
| "grad_norm": 0.19671481847763062, | |
| "learning_rate": 1.7489723846754e-05, | |
| "loss": 0.4371, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "grad_norm": 0.19839002192020416, | |
| "learning_rate": 1.7456301948763407e-05, | |
| "loss": 0.4181, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.7, | |
| "grad_norm": 0.218344584107399, | |
| "learning_rate": 1.7422691417475266e-05, | |
| "loss": 0.4417, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "grad_norm": 0.19715192914009094, | |
| "learning_rate": 1.73888931031858e-05, | |
| "loss": 0.4296, | |
| "step": 735 | |
| }, | |
| { | |
| "epoch": 0.71, | |
| "grad_norm": 0.2075146734714508, | |
| "learning_rate": 1.7354907860941863e-05, | |
| "loss": 0.4353, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "grad_norm": 0.21111871302127838, | |
| "learning_rate": 1.7320736550519296e-05, | |
| "loss": 0.4236, | |
| "step": 745 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "grad_norm": 0.2017849087715149, | |
| "learning_rate": 1.728638003640119e-05, | |
| "loss": 0.4254, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.72, | |
| "grad_norm": 0.2018367499113083, | |
| "learning_rate": 1.7251839187756016e-05, | |
| "loss": 0.4323, | |
| "step": 755 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "grad_norm": 0.2039439082145691, | |
| "learning_rate": 1.7217114878415633e-05, | |
| "loss": 0.4423, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.73, | |
| "grad_norm": 0.2052575796842575, | |
| "learning_rate": 1.7182207986853176e-05, | |
| "loss": 0.4436, | |
| "step": 765 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "grad_norm": 0.1971895545721054, | |
| "learning_rate": 1.714711939616084e-05, | |
| "loss": 0.4492, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.74, | |
| "grad_norm": 0.20616604387760162, | |
| "learning_rate": 1.7111849994027545e-05, | |
| "loss": 0.4619, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "grad_norm": 0.20350460708141327, | |
| "learning_rate": 1.7076400672716455e-05, | |
| "loss": 0.4497, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "grad_norm": 0.21535155177116394, | |
| "learning_rate": 1.7040772329042435e-05, | |
| "loss": 0.43, | |
| "step": 785 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "grad_norm": 0.22234632074832916, | |
| "learning_rate": 1.7004965864349335e-05, | |
| "loss": 0.4264, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.76, | |
| "grad_norm": 0.20027673244476318, | |
| "learning_rate": 1.6968982184487205e-05, | |
| "loss": 0.4315, | |
| "step": 795 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "grad_norm": 0.22603030502796173, | |
| "learning_rate": 1.693282219978939e-05, | |
| "loss": 0.4398, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.77, | |
| "grad_norm": 0.21727794408798218, | |
| "learning_rate": 1.6896486825049456e-05, | |
| "loss": 0.4226, | |
| "step": 805 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "grad_norm": 0.21654784679412842, | |
| "learning_rate": 1.685997697949809e-05, | |
| "loss": 0.4131, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.78, | |
| "grad_norm": 0.21828626096248627, | |
| "learning_rate": 1.6823293586779838e-05, | |
| "loss": 0.4267, | |
| "step": 815 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "grad_norm": 0.20646880567073822, | |
| "learning_rate": 1.6786437574929712e-05, | |
| "loss": 0.4535, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.79, | |
| "grad_norm": 0.21312721073627472, | |
| "learning_rate": 1.674940987634974e-05, | |
| "loss": 0.4561, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 0.2161865383386612, | |
| "learning_rate": 1.6712211427785373e-05, | |
| "loss": 0.4367, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.8, | |
| "grad_norm": 0.24069428443908691, | |
| "learning_rate": 1.6674843170301764e-05, | |
| "loss": 0.4312, | |
| "step": 835 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "grad_norm": 0.21740484237670898, | |
| "learning_rate": 1.6637306049259998e-05, | |
| "loss": 0.4213, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.81, | |
| "grad_norm": 0.20397590100765228, | |
| "learning_rate": 1.6599601014293144e-05, | |
| "loss": 0.4311, | |
| "step": 845 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "grad_norm": 0.21462039649486542, | |
| "learning_rate": 1.6561729019282245e-05, | |
| "loss": 0.4413, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.82, | |
| "grad_norm": 0.19624510407447815, | |
| "learning_rate": 1.6523691022332184e-05, | |
| "loss": 0.4304, | |
| "step": 855 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "grad_norm": 0.19382067024707794, | |
| "learning_rate": 1.648548798574744e-05, | |
| "loss": 0.4249, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.83, | |
| "grad_norm": 0.20487557351589203, | |
| "learning_rate": 1.6447120876007764e-05, | |
| "loss": 0.4257, | |
| "step": 865 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "grad_norm": 0.22991390526294708, | |
| "learning_rate": 1.6408590663743688e-05, | |
| "loss": 0.4377, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "grad_norm": 0.19853994250297546, | |
| "learning_rate": 1.6369898323712016e-05, | |
| "loss": 0.4225, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.84, | |
| "grad_norm": 0.2055709809064865, | |
| "learning_rate": 1.633104483477113e-05, | |
| "loss": 0.4413, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "grad_norm": 0.20579014718532562, | |
| "learning_rate": 1.6292031179856246e-05, | |
| "loss": 0.4219, | |
| "step": 885 | |
| }, | |
| { | |
| "epoch": 0.85, | |
| "grad_norm": 0.20581454038619995, | |
| "learning_rate": 1.625285834595453e-05, | |
| "loss": 0.4398, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "grad_norm": 0.19869303703308105, | |
| "learning_rate": 1.6213527324080152e-05, | |
| "loss": 0.4399, | |
| "step": 895 | |
| }, | |
| { | |
| "epoch": 0.86, | |
| "grad_norm": 0.20192164182662964, | |
| "learning_rate": 1.617403910924919e-05, | |
| "loss": 0.4246, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "grad_norm": 0.2064668983221054, | |
| "learning_rate": 1.6134394700454478e-05, | |
| "loss": 0.4064, | |
| "step": 905 | |
| }, | |
| { | |
| "epoch": 0.87, | |
| "grad_norm": 0.19731996953487396, | |
| "learning_rate": 1.6094595100640307e-05, | |
| "loss": 0.4355, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "grad_norm": 0.21017956733703613, | |
| "learning_rate": 1.6054641316677094e-05, | |
| "loss": 0.4211, | |
| "step": 915 | |
| }, | |
| { | |
| "epoch": 0.88, | |
| "grad_norm": 0.21060003340244293, | |
| "learning_rate": 1.6014534359335866e-05, | |
| "loss": 0.4234, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "grad_norm": 0.23994727432727814, | |
| "learning_rate": 1.597427524326271e-05, | |
| "loss": 0.4358, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.89, | |
| "grad_norm": 0.234989732503891, | |
| "learning_rate": 1.593386498695311e-05, | |
| "loss": 0.4239, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "grad_norm": 0.23091141879558563, | |
| "learning_rate": 1.589330461272616e-05, | |
| "loss": 0.4217, | |
| "step": 935 | |
| }, | |
| { | |
| "epoch": 0.9, | |
| "grad_norm": 0.2015177458524704, | |
| "learning_rate": 1.5852595146698728e-05, | |
| "loss": 0.4165, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "grad_norm": 0.21807970106601715, | |
| "learning_rate": 1.581173761875947e-05, | |
| "loss": 0.4324, | |
| "step": 945 | |
| }, | |
| { | |
| "epoch": 0.91, | |
| "grad_norm": 0.22525441646575928, | |
| "learning_rate": 1.5770733062542783e-05, | |
| "loss": 0.4281, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "grad_norm": 0.208124041557312, | |
| "learning_rate": 1.5729582515402676e-05, | |
| "loss": 0.426, | |
| "step": 955 | |
| }, | |
| { | |
| "epoch": 0.92, | |
| "grad_norm": 0.20913442969322205, | |
| "learning_rate": 1.5688287018386505e-05, | |
| "loss": 0.435, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "grad_norm": 0.2537720501422882, | |
| "learning_rate": 1.5646847616208625e-05, | |
| "loss": 0.4152, | |
| "step": 965 | |
| }, | |
| { | |
| "epoch": 0.93, | |
| "grad_norm": 0.23365460336208344, | |
| "learning_rate": 1.5605265357223998e-05, | |
| "loss": 0.4335, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "grad_norm": 0.2154242992401123, | |
| "learning_rate": 1.556354129340164e-05, | |
| "loss": 0.4368, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.94, | |
| "grad_norm": 0.19884240627288818, | |
| "learning_rate": 1.5521676480298014e-05, | |
| "loss": 0.4181, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "grad_norm": 0.1923339068889618, | |
| "learning_rate": 1.547967197703034e-05, | |
| "loss": 0.4266, | |
| "step": 985 | |
| }, | |
| { | |
| "epoch": 0.95, | |
| "grad_norm": 0.21156719326972961, | |
| "learning_rate": 1.5437528846249783e-05, | |
| "loss": 0.4298, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 0.21929724514484406, | |
| "learning_rate": 1.5395248154114574e-05, | |
| "loss": 0.4258, | |
| "step": 995 | |
| }, | |
| { | |
| "epoch": 0.96, | |
| "grad_norm": 0.22225961089134216, | |
| "learning_rate": 1.535283097026305e-05, | |
| "loss": 0.4345, | |
| "step": 1000 | |
| } | |
| ], | |
| "logging_steps": 5, | |
| "max_steps": 3123, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "save_steps": 500, | |
| "total_flos": 3.53274538981589e+18, | |
| "train_batch_size": 8, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |