| { |
| "best_metric": null, |
| "best_model_checkpoint": null, |
| "epoch": 1.0, |
| "eval_steps": 500, |
| "global_step": 1205, |
| "is_hyper_param_search": false, |
| "is_local_process_zero": true, |
| "is_world_process_zero": true, |
| "log_history": [ |
| { |
| "epoch": 0.008298755186721992, |
| "grad_norm": 0.01952634189028208, |
| "learning_rate": 8.196721311475409e-07, |
| "loss": 0.0002, |
| "step": 10 |
| }, |
| { |
| "epoch": 0.016597510373443983, |
| "grad_norm": 0.2545911846047496, |
| "learning_rate": 1.6393442622950819e-06, |
| "loss": 0.0004, |
| "step": 20 |
| }, |
| { |
| "epoch": 0.024896265560165973, |
| "grad_norm": 0.013501883610251985, |
| "learning_rate": 2.459016393442623e-06, |
| "loss": 0.0002, |
| "step": 30 |
| }, |
| { |
| "epoch": 0.03319502074688797, |
| "grad_norm": 0.01295761896054431, |
| "learning_rate": 3.2786885245901638e-06, |
| "loss": 0.0004, |
| "step": 40 |
| }, |
| { |
| "epoch": 0.04149377593360996, |
| "grad_norm": 0.3035265968896415, |
| "learning_rate": 4.098360655737705e-06, |
| "loss": 0.0003, |
| "step": 50 |
| }, |
| { |
| "epoch": 0.04979253112033195, |
| "grad_norm": 0.0017646037422956603, |
| "learning_rate": 4.918032786885246e-06, |
| "loss": 0.0012, |
| "step": 60 |
| }, |
| { |
| "epoch": 0.058091286307053944, |
| "grad_norm": 0.017035062957617197, |
| "learning_rate": 4.999236479642602e-06, |
| "loss": 0.0025, |
| "step": 70 |
| }, |
| { |
| "epoch": 0.06639004149377593, |
| "grad_norm": 0.004209789578646451, |
| "learning_rate": 4.996597748756039e-06, |
| "loss": 0.0003, |
| "step": 80 |
| }, |
| { |
| "epoch": 0.07468879668049792, |
| "grad_norm": 0.016889199255074427, |
| "learning_rate": 4.992076370516985e-06, |
| "loss": 0.0029, |
| "step": 90 |
| }, |
| { |
| "epoch": 0.08298755186721991, |
| "grad_norm": 0.49972929296558655, |
| "learning_rate": 4.985675754429744e-06, |
| "loss": 0.0012, |
| "step": 100 |
| }, |
| { |
| "epoch": 0.0912863070539419, |
| "grad_norm": 0.05679961395868942, |
| "learning_rate": 4.977400727104024e-06, |
| "loss": 0.0013, |
| "step": 110 |
| }, |
| { |
| "epoch": 0.0995850622406639, |
| "grad_norm": 0.3163904422156924, |
| "learning_rate": 4.967257528615261e-06, |
| "loss": 0.001, |
| "step": 120 |
| }, |
| { |
| "epoch": 0.1078838174273859, |
| "grad_norm": 0.02845684920996449, |
| "learning_rate": 4.955253807799073e-06, |
| "loss": 0.0059, |
| "step": 130 |
| }, |
| { |
| "epoch": 0.11618257261410789, |
| "grad_norm": 0.1724673205403095, |
| "learning_rate": 4.941398616483382e-06, |
| "loss": 0.0016, |
| "step": 140 |
| }, |
| { |
| "epoch": 0.12448132780082988, |
| "grad_norm": 0.0447393567131259, |
| "learning_rate": 4.925702402662568e-06, |
| "loss": 0.001, |
| "step": 150 |
| }, |
| { |
| "epoch": 0.13278008298755187, |
| "grad_norm": 0.10461493774492425, |
| "learning_rate": 4.9081770026187915e-06, |
| "loss": 0.0012, |
| "step": 160 |
| }, |
| { |
| "epoch": 0.14107883817427386, |
| "grad_norm": 0.1340115340660757, |
| "learning_rate": 4.888835631996426e-06, |
| "loss": 0.0008, |
| "step": 170 |
| }, |
| { |
| "epoch": 0.14937759336099585, |
| "grad_norm": 0.025774190164302806, |
| "learning_rate": 4.86769287583634e-06, |
| "loss": 0.0018, |
| "step": 180 |
| }, |
| { |
| "epoch": 0.15767634854771784, |
| "grad_norm": 0.6750469917729486, |
| "learning_rate": 4.844764677577531e-06, |
| "loss": 0.0024, |
| "step": 190 |
| }, |
| { |
| "epoch": 0.16597510373443983, |
| "grad_norm": 0.5121699739264448, |
| "learning_rate": 4.820068327034416e-06, |
| "loss": 0.0025, |
| "step": 200 |
| }, |
| { |
| "epoch": 0.17427385892116182, |
| "grad_norm": 0.005162235007018351, |
| "learning_rate": 4.793622447358837e-06, |
| "loss": 0.0013, |
| "step": 210 |
| }, |
| { |
| "epoch": 0.1825726141078838, |
| "grad_norm": 0.0065204757813555, |
| "learning_rate": 4.765446980996622e-06, |
| "loss": 0.0025, |
| "step": 220 |
| }, |
| { |
| "epoch": 0.1908713692946058, |
| "grad_norm": 0.1834554236512134, |
| "learning_rate": 4.735563174649278e-06, |
| "loss": 0.0015, |
| "step": 230 |
| }, |
| { |
| "epoch": 0.1991701244813278, |
| "grad_norm": 1.0765680707456238, |
| "learning_rate": 4.703993563252172e-06, |
| "loss": 0.0034, |
| "step": 240 |
| }, |
| { |
| "epoch": 0.2074688796680498, |
| "grad_norm": 0.34800167205451604, |
| "learning_rate": 4.670761952981264e-06, |
| "loss": 0.0016, |
| "step": 250 |
| }, |
| { |
| "epoch": 0.2157676348547718, |
| "grad_norm": 0.12237972347896162, |
| "learning_rate": 4.635893403301233e-06, |
| "loss": 0.0014, |
| "step": 260 |
| }, |
| { |
| "epoch": 0.22406639004149378, |
| "grad_norm": 0.2737166694487476, |
| "learning_rate": 4.5994142080684956e-06, |
| "loss": 0.0025, |
| "step": 270 |
| }, |
| { |
| "epoch": 0.23236514522821577, |
| "grad_norm": 0.04068916670988713, |
| "learning_rate": 4.561351875703406e-06, |
| "loss": 0.0014, |
| "step": 280 |
| }, |
| { |
| "epoch": 0.24066390041493776, |
| "grad_norm": 0.3245341216908413, |
| "learning_rate": 4.52173510844656e-06, |
| "loss": 0.0012, |
| "step": 290 |
| }, |
| { |
| "epoch": 0.24896265560165975, |
| "grad_norm": 0.03783916464770265, |
| "learning_rate": 4.480593780714863e-06, |
| "loss": 0.0026, |
| "step": 300 |
| }, |
| { |
| "epoch": 0.2572614107883817, |
| "grad_norm": 0.4859003212418197, |
| "learning_rate": 4.4379589165736736e-06, |
| "loss": 0.0035, |
| "step": 310 |
| }, |
| { |
| "epoch": 0.26556016597510373, |
| "grad_norm": 0.43809699541809793, |
| "learning_rate": 4.393862666342017e-06, |
| "loss": 0.0006, |
| "step": 320 |
| }, |
| { |
| "epoch": 0.27385892116182575, |
| "grad_norm": 0.18772242388214733, |
| "learning_rate": 4.348338282348504e-06, |
| "loss": 0.0023, |
| "step": 330 |
| }, |
| { |
| "epoch": 0.2821576763485477, |
| "grad_norm": 0.0908936575905787, |
| "learning_rate": 4.301420093856247e-06, |
| "loss": 0.0011, |
| "step": 340 |
| }, |
| { |
| "epoch": 0.29045643153526973, |
| "grad_norm": 0.4647945053304386, |
| "learning_rate": 4.2531434811756675e-06, |
| "loss": 0.0008, |
| "step": 350 |
| }, |
| { |
| "epoch": 0.2987551867219917, |
| "grad_norm": 0.009093547777615585, |
| "learning_rate": 4.203544848984729e-06, |
| "loss": 0.0024, |
| "step": 360 |
| }, |
| { |
| "epoch": 0.3070539419087137, |
| "grad_norm": 0.06847326021423142, |
| "learning_rate": 4.15266159887671e-06, |
| "loss": 0.002, |
| "step": 370 |
| }, |
| { |
| "epoch": 0.3153526970954357, |
| "grad_norm": 0.09182889879413814, |
| "learning_rate": 4.10053210115622e-06, |
| "loss": 0.0016, |
| "step": 380 |
| }, |
| { |
| "epoch": 0.3236514522821577, |
| "grad_norm": 0.13339020629841652, |
| "learning_rate": 4.0471956659047166e-06, |
| "loss": 0.001, |
| "step": 390 |
| }, |
| { |
| "epoch": 0.33195020746887965, |
| "grad_norm": 0.008092316624748167, |
| "learning_rate": 3.9926925133373565e-06, |
| "loss": 0.0023, |
| "step": 400 |
| }, |
| { |
| "epoch": 0.34024896265560167, |
| "grad_norm": 0.023997101221404513, |
| "learning_rate": 3.93706374347353e-06, |
| "loss": 0.0007, |
| "step": 410 |
| }, |
| { |
| "epoch": 0.34854771784232363, |
| "grad_norm": 0.02702823530323454, |
| "learning_rate": 3.880351305143945e-06, |
| "loss": 0.0016, |
| "step": 420 |
| }, |
| { |
| "epoch": 0.35684647302904565, |
| "grad_norm": 0.14873455961870954, |
| "learning_rate": 3.822597964357632e-06, |
| "loss": 0.0014, |
| "step": 430 |
| }, |
| { |
| "epoch": 0.3651452282157676, |
| "grad_norm": 0.030315999061406942, |
| "learning_rate": 3.76384727205274e-06, |
| "loss": 0.0012, |
| "step": 440 |
| }, |
| { |
| "epoch": 0.37344398340248963, |
| "grad_norm": 0.011605294695610434, |
| "learning_rate": 3.7041435312554053e-06, |
| "loss": 0.0004, |
| "step": 450 |
| }, |
| { |
| "epoch": 0.3817427385892116, |
| "grad_norm": 0.009516457718031966, |
| "learning_rate": 3.6435317636715104e-06, |
| "loss": 0.001, |
| "step": 460 |
| }, |
| { |
| "epoch": 0.3900414937759336, |
| "grad_norm": 0.24625371958398262, |
| "learning_rate": 3.5820576757364743e-06, |
| "loss": 0.0005, |
| "step": 470 |
| }, |
| { |
| "epoch": 0.3983402489626556, |
| "grad_norm": 0.011967039603122181, |
| "learning_rate": 3.51976762414871e-06, |
| "loss": 0.0016, |
| "step": 480 |
| }, |
| { |
| "epoch": 0.4066390041493776, |
| "grad_norm": 0.03929185536463483, |
| "learning_rate": 3.4567085809127247e-06, |
| "loss": 0.0009, |
| "step": 490 |
| }, |
| { |
| "epoch": 0.4149377593360996, |
| "grad_norm": 0.1491795016675303, |
| "learning_rate": 3.392928097918229e-06, |
| "loss": 0.0011, |
| "step": 500 |
| }, |
| { |
| "epoch": 0.42323651452282157, |
| "grad_norm": 0.030419957346452066, |
| "learning_rate": 3.328474271081959e-06, |
| "loss": 0.0007, |
| "step": 510 |
| }, |
| { |
| "epoch": 0.4315352697095436, |
| "grad_norm": 0.011487360993853078, |
| "learning_rate": 3.2633957040792624e-06, |
| "loss": 0.0027, |
| "step": 520 |
| }, |
| { |
| "epoch": 0.43983402489626555, |
| "grad_norm": 0.10771152282716176, |
| "learning_rate": 3.1977414716927837e-06, |
| "loss": 0.0009, |
| "step": 530 |
| }, |
| { |
| "epoch": 0.44813278008298757, |
| "grad_norm": 0.011123547297225648, |
| "learning_rate": 3.131561082805898e-06, |
| "loss": 0.0012, |
| "step": 540 |
| }, |
| { |
| "epoch": 0.45643153526970953, |
| "grad_norm": 0.04158374445810553, |
| "learning_rate": 3.0649044430688017e-06, |
| "loss": 0.0005, |
| "step": 550 |
| }, |
| { |
| "epoch": 0.46473029045643155, |
| "grad_norm": 0.20382791628540436, |
| "learning_rate": 2.997821817265394e-06, |
| "loss": 0.0015, |
| "step": 560 |
| }, |
| { |
| "epoch": 0.4730290456431535, |
| "grad_norm": 0.10253541306876046, |
| "learning_rate": 2.9303637914093532e-06, |
| "loss": 0.0002, |
| "step": 570 |
| }, |
| { |
| "epoch": 0.48132780082987553, |
| "grad_norm": 0.0368908683192571, |
| "learning_rate": 2.8625812345979716e-06, |
| "loss": 0.0004, |
| "step": 580 |
| }, |
| { |
| "epoch": 0.4896265560165975, |
| "grad_norm": 0.4132164259019655, |
| "learning_rate": 2.7945252606525244e-06, |
| "loss": 0.0009, |
| "step": 590 |
| }, |
| { |
| "epoch": 0.4979253112033195, |
| "grad_norm": 0.02180157304973387, |
| "learning_rate": 2.726247189574095e-06, |
| "loss": 0.0007, |
| "step": 600 |
| }, |
| { |
| "epoch": 0.5062240663900415, |
| "grad_norm": 0.023273145386899886, |
| "learning_rate": 2.6577985088439212e-06, |
| "loss": 0.0013, |
| "step": 610 |
| }, |
| { |
| "epoch": 0.5145228215767634, |
| "grad_norm": 0.03091350937162301, |
| "learning_rate": 2.5892308345974517e-06, |
| "loss": 0.0003, |
| "step": 620 |
| }, |
| { |
| "epoch": 0.5228215767634855, |
| "grad_norm": 0.002613622106717658, |
| "learning_rate": 2.520595872701379e-06, |
| "loss": 0.0001, |
| "step": 630 |
| }, |
| { |
| "epoch": 0.5311203319502075, |
| "grad_norm": 0.06951997008434906, |
| "learning_rate": 2.451945379763019e-06, |
| "loss": 0.0002, |
| "step": 640 |
| }, |
| { |
| "epoch": 0.5394190871369294, |
| "grad_norm": 0.0012440442823388085, |
| "learning_rate": 2.383331124101412e-06, |
| "loss": 0.0004, |
| "step": 650 |
| }, |
| { |
| "epoch": 0.5477178423236515, |
| "grad_norm": 0.017965233689354338, |
| "learning_rate": 2.3148048467096076e-06, |
| "loss": 0.0005, |
| "step": 660 |
| }, |
| { |
| "epoch": 0.5560165975103735, |
| "grad_norm": 0.0017474955006528351, |
| "learning_rate": 2.2464182222375466e-06, |
| "loss": 0.0009, |
| "step": 670 |
| }, |
| { |
| "epoch": 0.5643153526970954, |
| "grad_norm": 0.08853303234184502, |
| "learning_rate": 2.1782228200249705e-06, |
| "loss": 0.0002, |
| "step": 680 |
| }, |
| { |
| "epoch": 0.5726141078838174, |
| "grad_norm": 0.055082873747902855, |
| "learning_rate": 2.1102700652137527e-06, |
| "loss": 0.0004, |
| "step": 690 |
| }, |
| { |
| "epoch": 0.5809128630705395, |
| "grad_norm": 0.0989162087632671, |
| "learning_rate": 2.0426111999689568e-06, |
| "loss": 0.001, |
| "step": 700 |
| }, |
| { |
| "epoch": 0.5892116182572614, |
| "grad_norm": 0.002211178671875013, |
| "learning_rate": 1.9752972448378817e-06, |
| "loss": 0.0004, |
| "step": 710 |
| }, |
| { |
| "epoch": 0.5975103734439834, |
| "grad_norm": 0.05298706307817462, |
| "learning_rate": 1.908378960276221e-06, |
| "loss": 0.0003, |
| "step": 720 |
| }, |
| { |
| "epoch": 0.6058091286307054, |
| "grad_norm": 0.0037145084040453163, |
| "learning_rate": 1.84190680837036e-06, |
| "loss": 0.0003, |
| "step": 730 |
| }, |
| { |
| "epoch": 0.6141078838174274, |
| "grad_norm": 0.0020591082549731407, |
| "learning_rate": 1.775930914784659e-06, |
| "loss": 0.0001, |
| "step": 740 |
| }, |
| { |
| "epoch": 0.6224066390041494, |
| "grad_norm": 0.002395583907814531, |
| "learning_rate": 1.7105010309624381e-06, |
| "loss": 0.001, |
| "step": 750 |
| }, |
| { |
| "epoch": 0.6307053941908713, |
| "grad_norm": 0.010281137035125171, |
| "learning_rate": 1.6456664966091463e-06, |
| "loss": 0.0001, |
| "step": 760 |
| }, |
| { |
| "epoch": 0.6390041493775933, |
| "grad_norm": 0.005408958318286851, |
| "learning_rate": 1.5814762024860261e-06, |
| "loss": 0.001, |
| "step": 770 |
| }, |
| { |
| "epoch": 0.6473029045643154, |
| "grad_norm": 0.016316274316565883, |
| "learning_rate": 1.5179785535423109e-06, |
| "loss": 0.0002, |
| "step": 780 |
| }, |
| { |
| "epoch": 0.6556016597510373, |
| "grad_norm": 0.025131773659064247, |
| "learning_rate": 1.4552214324137744e-06, |
| "loss": 0.0003, |
| "step": 790 |
| }, |
| { |
| "epoch": 0.6639004149377593, |
| "grad_norm": 0.005562868034328769, |
| "learning_rate": 1.393252163315138e-06, |
| "loss": 0.0003, |
| "step": 800 |
| }, |
| { |
| "epoch": 0.6721991701244814, |
| "grad_norm": 0.014346694616478282, |
| "learning_rate": 1.3321174763535926e-06, |
| "loss": 0.0005, |
| "step": 810 |
| }, |
| { |
| "epoch": 0.6804979253112033, |
| "grad_norm": 0.009721758111534043, |
| "learning_rate": 1.2718634722903073e-06, |
| "loss": 0.0011, |
| "step": 820 |
| }, |
| { |
| "epoch": 0.6887966804979253, |
| "grad_norm": 0.06104884728923139, |
| "learning_rate": 1.2125355877765336e-06, |
| "loss": 0.0003, |
| "step": 830 |
| }, |
| { |
| "epoch": 0.6970954356846473, |
| "grad_norm": 0.008343792515062282, |
| "learning_rate": 1.1541785610904995e-06, |
| "loss": 0.0, |
| "step": 840 |
| }, |
| { |
| "epoch": 0.7053941908713693, |
| "grad_norm": 0.7020599008202609, |
| "learning_rate": 1.0968363984009402e-06, |
| "loss": 0.001, |
| "step": 850 |
| }, |
| { |
| "epoch": 0.7136929460580913, |
| "grad_norm": 0.0019620049947002543, |
| "learning_rate": 1.0405523405826917e-06, |
| "loss": 0.0005, |
| "step": 860 |
| }, |
| { |
| "epoch": 0.7219917012448133, |
| "grad_norm": 0.0018741347835914191, |
| "learning_rate": 9.853688306093935e-07, |
| "loss": 0.0001, |
| "step": 870 |
| }, |
| { |
| "epoch": 0.7302904564315352, |
| "grad_norm": 0.0030367972735562128, |
| "learning_rate": 9.313274815478698e-07, |
| "loss": 0.0001, |
| "step": 880 |
| }, |
| { |
| "epoch": 0.7385892116182573, |
| "grad_norm": 0.06025337850459377, |
| "learning_rate": 8.784690451783337e-07, |
| "loss": 0.0012, |
| "step": 890 |
| }, |
| { |
| "epoch": 0.7468879668049793, |
| "grad_norm": 0.0008554758999917974, |
| "learning_rate": 8.268333812640717e-07, |
| "loss": 0.0013, |
| "step": 900 |
| }, |
| { |
| "epoch": 0.7551867219917012, |
| "grad_norm": 0.156929704775561, |
| "learning_rate": 7.764594274937884e-07, |
| "loss": 0.0006, |
| "step": 910 |
| }, |
| { |
| "epoch": 0.7634854771784232, |
| "grad_norm": 0.0015978578008091497, |
| "learning_rate": 7.27385170119273e-07, |
| "loss": 0.0001, |
| "step": 920 |
| }, |
| { |
| "epoch": 0.7717842323651453, |
| "grad_norm": 0.04001419658127684, |
| "learning_rate": 6.796476153105294e-07, |
| "loss": 0.0002, |
| "step": 930 |
| }, |
| { |
| "epoch": 0.7800829875518672, |
| "grad_norm": 0.0031656530089931665, |
| "learning_rate": 6.332827612499718e-07, |
| "loss": 0.0001, |
| "step": 940 |
| }, |
| { |
| "epoch": 0.7883817427385892, |
| "grad_norm": 0.0136750474203667, |
| "learning_rate": 5.883255709867288e-07, |
| "loss": 0.0001, |
| "step": 950 |
| }, |
| { |
| "epoch": 0.7966804979253111, |
| "grad_norm": 0.006491577024871877, |
| "learning_rate": 5.448099460715289e-07, |
| "loss": 0.0003, |
| "step": 960 |
| }, |
| { |
| "epoch": 0.8049792531120332, |
| "grad_norm": 0.03313707798793249, |
| "learning_rate": 5.027687009920393e-07, |
| "loss": 0.0003, |
| "step": 970 |
| }, |
| { |
| "epoch": 0.8132780082987552, |
| "grad_norm": 0.011464450089201293, |
| "learning_rate": 4.6223353842795053e-07, |
| "loss": 0.0003, |
| "step": 980 |
| }, |
| { |
| "epoch": 0.8215767634854771, |
| "grad_norm": 0.001030709037873557, |
| "learning_rate": 4.232350253444539e-07, |
| "loss": 0.0003, |
| "step": 990 |
| }, |
| { |
| "epoch": 0.8298755186721992, |
| "grad_norm": 0.01249193689623948, |
| "learning_rate": 3.8580256994214805e-07, |
| "loss": 0.0003, |
| "step": 1000 |
| }, |
| { |
| "epoch": 0.8381742738589212, |
| "grad_norm": 0.003310230202843625, |
| "learning_rate": 3.499643994807486e-07, |
| "loss": 0.0003, |
| "step": 1010 |
| }, |
| { |
| "epoch": 0.8464730290456431, |
| "grad_norm": 0.007029766220764641, |
| "learning_rate": 3.1574753899333176e-07, |
| "loss": 0.0001, |
| "step": 1020 |
| }, |
| { |
| "epoch": 0.8547717842323651, |
| "grad_norm": 0.015014047974484341, |
| "learning_rate": 2.8317779090715593e-07, |
| "loss": 0.0015, |
| "step": 1030 |
| }, |
| { |
| "epoch": 0.8630705394190872, |
| "grad_norm": 0.028295865852545128, |
| "learning_rate": 2.522797155864354e-07, |
| "loss": 0.0001, |
| "step": 1040 |
| }, |
| { |
| "epoch": 0.8713692946058091, |
| "grad_norm": 0.031063791170280682, |
| "learning_rate": 2.2307661281173314e-07, |
| "loss": 0.0001, |
| "step": 1050 |
| }, |
| { |
| "epoch": 0.8796680497925311, |
| "grad_norm": 0.016605299497541786, |
| "learning_rate": 1.9559050420994124e-07, |
| "loss": 0.0001, |
| "step": 1060 |
| }, |
| { |
| "epoch": 0.8879668049792531, |
| "grad_norm": 0.02594585316898821, |
| "learning_rate": 1.6984211664809957e-07, |
| "loss": 0.0002, |
| "step": 1070 |
| }, |
| { |
| "epoch": 0.8962655601659751, |
| "grad_norm": 0.0010463749820481666, |
| "learning_rate": 1.4585086660357155e-07, |
| "loss": 0.0, |
| "step": 1080 |
| }, |
| { |
| "epoch": 0.9045643153526971, |
| "grad_norm": 0.014847962234540494, |
| "learning_rate": 1.2363484552236654e-07, |
| "loss": 0.0005, |
| "step": 1090 |
| }, |
| { |
| "epoch": 0.9128630705394191, |
| "grad_norm": 0.0031223430965961747, |
| "learning_rate": 1.0321080617664846e-07, |
| "loss": 0.0001, |
| "step": 1100 |
| }, |
| { |
| "epoch": 0.921161825726141, |
| "grad_norm": 0.0014872131531329376, |
| "learning_rate": 8.4594150031721e-08, |
| "loss": 0.0001, |
| "step": 1110 |
| }, |
| { |
| "epoch": 0.9294605809128631, |
| "grad_norm": 0.0008679571388583524, |
| "learning_rate": 6.779891563200664e-08, |
| "loss": 0.0001, |
| "step": 1120 |
| }, |
| { |
| "epoch": 0.9377593360995851, |
| "grad_norm": 0.0028433062789473456, |
| "learning_rate": 5.2837768014792066e-08, |
| "loss": 0.0001, |
| "step": 1130 |
| }, |
| { |
| "epoch": 0.946058091286307, |
| "grad_norm": 0.01390940083993496, |
| "learning_rate": 3.972198915970976e-08, |
| "loss": 0.0004, |
| "step": 1140 |
| }, |
| { |
| "epoch": 0.9543568464730291, |
| "grad_norm": 0.013627677425191088, |
| "learning_rate": 2.8461469481164682e-08, |
| "loss": 0.0002, |
| "step": 1150 |
| }, |
| { |
| "epoch": 0.9626556016597511, |
| "grad_norm": 0.0006225638406747092, |
| "learning_rate": 1.9064700370118593e-08, |
| "loss": 0.001, |
| "step": 1160 |
| }, |
| { |
| "epoch": 0.970954356846473, |
| "grad_norm": 0.06472085081508153, |
| "learning_rate": 1.1538767790859885e-08, |
| "loss": 0.0009, |
| "step": 1170 |
| }, |
| { |
| "epoch": 0.979253112033195, |
| "grad_norm": 0.03204457354498749, |
| "learning_rate": 5.889346937581475e-09, |
| "loss": 0.0002, |
| "step": 1180 |
| }, |
| { |
| "epoch": 0.9875518672199171, |
| "grad_norm": 0.19770717466243257, |
| "learning_rate": 2.120697954800799e-09, |
| "loss": 0.0003, |
| "step": 1190 |
| }, |
| { |
| "epoch": 0.995850622406639, |
| "grad_norm": 0.0013634677194990677, |
| "learning_rate": 2.35662724848984e-10, |
| "loss": 0.0001, |
| "step": 1200 |
| }, |
| { |
| "epoch": 1.0, |
| "step": 1205, |
| "total_flos": 901340604923904.0, |
| "train_loss": 0.0009180365409025791, |
| "train_runtime": 38806.1809, |
| "train_samples_per_second": 5.96, |
| "train_steps_per_second": 0.031 |
| } |
| ], |
| "logging_steps": 10, |
| "max_steps": 1205, |
| "num_input_tokens_seen": 0, |
| "num_train_epochs": 1, |
| "save_steps": 800, |
| "stateful_callbacks": { |
| "TrainerControl": { |
| "args": { |
| "should_epoch_stop": false, |
| "should_evaluate": false, |
| "should_log": false, |
| "should_save": true, |
| "should_training_stop": true |
| }, |
| "attributes": {} |
| } |
| }, |
| "total_flos": 901340604923904.0, |
| "train_batch_size": 4, |
| "trial_name": null, |
| "trial_params": null |
| } |
|
|
