diff --git "a/trainer_state.json" "b/trainer_state.json" new file mode 100644--- /dev/null +++ "b/trainer_state.json" @@ -0,0 +1,22534 @@ +{ + "best_global_step": 23254, + "best_metric": 0.8739344728917919, + "best_model_checkpoint": "output/QA-DeBERTa-v3-base-binary/checkpoint-23254", + "epoch": 3.750887154009936, + "eval_steps": 2114, + "global_step": 31710, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.0011828720132481666, + "grad_norm": 1.452476978302002, + "learning_rate": 5.4e-08, + "loss": 0.6892, + "step": 10 + }, + { + "epoch": 0.0023657440264963333, + "grad_norm": 1.3107129335403442, + "learning_rate": 1.14e-07, + "loss": 0.6895, + "step": 20 + }, + { + "epoch": 0.0035486160397444995, + "grad_norm": 1.5076279640197754, + "learning_rate": 1.7400000000000002e-07, + "loss": 0.6912, + "step": 30 + }, + { + "epoch": 0.0047314880529926665, + "grad_norm": 0.7335036993026733, + "learning_rate": 2.34e-07, + "loss": 0.6838, + "step": 40 + }, + { + "epoch": 0.005914360066240833, + "grad_norm": 1.0573631525039673, + "learning_rate": 2.94e-07, + "loss": 0.692, + "step": 50 + }, + { + "epoch": 0.007097232079488999, + "grad_norm": 0.5848931670188904, + "learning_rate": 3.5399999999999997e-07, + "loss": 0.6922, + "step": 60 + }, + { + "epoch": 0.008280104092737165, + "grad_norm": 0.5945519804954529, + "learning_rate": 4.1400000000000003e-07, + "loss": 0.691, + "step": 70 + }, + { + "epoch": 0.009462976105985333, + "grad_norm": 0.7829030156135559, + "learning_rate": 4.7400000000000004e-07, + "loss": 0.6774, + "step": 80 + }, + { + "epoch": 0.0106458481192335, + "grad_norm": 0.6574153304100037, + "learning_rate": 5.34e-07, + "loss": 0.6908, + "step": 90 + }, + { + "epoch": 0.011828720132481665, + "grad_norm": 0.9069286584854126, + "learning_rate": 5.94e-07, + "loss": 0.6826, + "step": 100 + }, + { + "epoch": 0.013011592145729832, + "grad_norm": 0.7912967801094055, + "learning_rate": 6.54e-07, + "loss": 0.6919, + "step": 110 + }, + { + "epoch": 0.014194464158977998, + "grad_norm": 1.0637558698654175, + "learning_rate": 7.14e-07, + "loss": 0.6882, + "step": 120 + }, + { + "epoch": 0.015377336172226166, + "grad_norm": 0.7037789821624756, + "learning_rate": 7.74e-07, + "loss": 0.6886, + "step": 130 + }, + { + "epoch": 0.01656020818547433, + "grad_norm": 0.830253005027771, + "learning_rate": 8.340000000000001e-07, + "loss": 0.6949, + "step": 140 + }, + { + "epoch": 0.017743080198722498, + "grad_norm": 1.0448371171951294, + "learning_rate": 8.939999999999999e-07, + "loss": 0.6897, + "step": 150 + }, + { + "epoch": 0.018925952211970666, + "grad_norm": 0.7149848341941833, + "learning_rate": 9.54e-07, + "loss": 0.6944, + "step": 160 + }, + { + "epoch": 0.02010882422521883, + "grad_norm": 0.8854143023490906, + "learning_rate": 1.0140000000000002e-06, + "loss": 0.6905, + "step": 170 + }, + { + "epoch": 0.021291696238467, + "grad_norm": 0.5680384039878845, + "learning_rate": 1.074e-06, + "loss": 0.6829, + "step": 180 + }, + { + "epoch": 0.022474568251715163, + "grad_norm": 0.9411191940307617, + "learning_rate": 1.134e-06, + "loss": 0.6861, + "step": 190 + }, + { + "epoch": 0.02365744026496333, + "grad_norm": 0.7921398878097534, + "learning_rate": 1.1940000000000001e-06, + "loss": 0.6854, + "step": 200 + }, + { + "epoch": 0.0248403122782115, + "grad_norm": 0.7178440093994141, + "learning_rate": 1.254e-06, + "loss": 0.6839, + "step": 210 + }, + { + "epoch": 0.026023184291459663, + "grad_norm": 0.8898746371269226, + "learning_rate": 1.314e-06, + "loss": 0.6703, + "step": 220 + }, + { + "epoch": 0.02720605630470783, + "grad_norm": 0.9438018202781677, + "learning_rate": 1.374e-06, + "loss": 0.6892, + "step": 230 + }, + { + "epoch": 0.028388928317955996, + "grad_norm": 1.8930295705795288, + "learning_rate": 1.434e-06, + "loss": 0.6877, + "step": 240 + }, + { + "epoch": 0.029571800331204164, + "grad_norm": 0.9319143891334534, + "learning_rate": 1.494e-06, + "loss": 0.69, + "step": 250 + }, + { + "epoch": 0.03075467234445233, + "grad_norm": 0.824596107006073, + "learning_rate": 1.554e-06, + "loss": 0.6712, + "step": 260 + }, + { + "epoch": 0.0319375443577005, + "grad_norm": 1.2915641069412231, + "learning_rate": 1.6140000000000001e-06, + "loss": 0.6642, + "step": 270 + }, + { + "epoch": 0.03312041637094866, + "grad_norm": 1.0681672096252441, + "learning_rate": 1.6740000000000002e-06, + "loss": 0.6626, + "step": 280 + }, + { + "epoch": 0.03430328838419683, + "grad_norm": 0.9478822946548462, + "learning_rate": 1.7339999999999998e-06, + "loss": 0.6621, + "step": 290 + }, + { + "epoch": 0.035486160397444996, + "grad_norm": 1.407970905303955, + "learning_rate": 1.7939999999999999e-06, + "loss": 0.6333, + "step": 300 + }, + { + "epoch": 0.036669032410693164, + "grad_norm": 1.8239948749542236, + "learning_rate": 1.854e-06, + "loss": 0.6293, + "step": 310 + }, + { + "epoch": 0.03785190442394133, + "grad_norm": 1.1743783950805664, + "learning_rate": 1.9140000000000002e-06, + "loss": 0.5745, + "step": 320 + }, + { + "epoch": 0.03903477643718949, + "grad_norm": 2.4400320053100586, + "learning_rate": 1.974e-06, + "loss": 0.6148, + "step": 330 + }, + { + "epoch": 0.04021764845043766, + "grad_norm": 2.1994576454162598, + "learning_rate": 2.0340000000000003e-06, + "loss": 0.5938, + "step": 340 + }, + { + "epoch": 0.04140052046368583, + "grad_norm": 6.897472858428955, + "learning_rate": 2.0939999999999998e-06, + "loss": 0.5736, + "step": 350 + }, + { + "epoch": 0.042583392476934, + "grad_norm": 2.3344879150390625, + "learning_rate": 2.154e-06, + "loss": 0.5793, + "step": 360 + }, + { + "epoch": 0.043766264490182165, + "grad_norm": 1.583277702331543, + "learning_rate": 2.214e-06, + "loss": 0.5565, + "step": 370 + }, + { + "epoch": 0.044949136503430326, + "grad_norm": 1.6724634170532227, + "learning_rate": 2.274e-06, + "loss": 0.556, + "step": 380 + }, + { + "epoch": 0.046132008516678494, + "grad_norm": 2.3909051418304443, + "learning_rate": 2.334e-06, + "loss": 0.5604, + "step": 390 + }, + { + "epoch": 0.04731488052992666, + "grad_norm": 1.8554753065109253, + "learning_rate": 2.3940000000000003e-06, + "loss": 0.5495, + "step": 400 + }, + { + "epoch": 0.04849775254317483, + "grad_norm": 3.570477247238159, + "learning_rate": 2.4539999999999997e-06, + "loss": 0.5379, + "step": 410 + }, + { + "epoch": 0.049680624556423, + "grad_norm": 3.0133328437805176, + "learning_rate": 2.514e-06, + "loss": 0.5462, + "step": 420 + }, + { + "epoch": 0.05086349656967116, + "grad_norm": 2.053877830505371, + "learning_rate": 2.574e-06, + "loss": 0.5224, + "step": 430 + }, + { + "epoch": 0.052046368582919326, + "grad_norm": 3.332078218460083, + "learning_rate": 2.634e-06, + "loss": 0.5472, + "step": 440 + }, + { + "epoch": 0.053229240596167494, + "grad_norm": 3.0517325401306152, + "learning_rate": 2.694e-06, + "loss": 0.5352, + "step": 450 + }, + { + "epoch": 0.05441211260941566, + "grad_norm": 3.1004769802093506, + "learning_rate": 2.7540000000000002e-06, + "loss": 0.5301, + "step": 460 + }, + { + "epoch": 0.05559498462266383, + "grad_norm": 2.9489076137542725, + "learning_rate": 2.814e-06, + "loss": 0.5352, + "step": 470 + }, + { + "epoch": 0.05677785663591199, + "grad_norm": 2.8148906230926514, + "learning_rate": 2.874e-06, + "loss": 0.518, + "step": 480 + }, + { + "epoch": 0.05796072864916016, + "grad_norm": 1.5071579217910767, + "learning_rate": 2.934e-06, + "loss": 0.5416, + "step": 490 + }, + { + "epoch": 0.05914360066240833, + "grad_norm": 3.4842145442962646, + "learning_rate": 2.994e-06, + "loss": 0.5383, + "step": 500 + }, + { + "epoch": 0.060326472675656495, + "grad_norm": 3.9251625537872314, + "learning_rate": 3.0540000000000003e-06, + "loss": 0.4893, + "step": 510 + }, + { + "epoch": 0.06150934468890466, + "grad_norm": 3.9749224185943604, + "learning_rate": 3.114e-06, + "loss": 0.4967, + "step": 520 + }, + { + "epoch": 0.06269221670215283, + "grad_norm": 2.711362361907959, + "learning_rate": 3.1740000000000004e-06, + "loss": 0.4955, + "step": 530 + }, + { + "epoch": 0.063875088715401, + "grad_norm": 10.314888954162598, + "learning_rate": 3.2340000000000003e-06, + "loss": 0.493, + "step": 540 + }, + { + "epoch": 0.06505796072864917, + "grad_norm": 4.265482425689697, + "learning_rate": 3.294e-06, + "loss": 0.4981, + "step": 550 + }, + { + "epoch": 0.06624083274189732, + "grad_norm": 4.920806407928467, + "learning_rate": 3.3540000000000004e-06, + "loss": 0.496, + "step": 560 + }, + { + "epoch": 0.06742370475514549, + "grad_norm": 4.876100063323975, + "learning_rate": 3.414e-06, + "loss": 0.512, + "step": 570 + }, + { + "epoch": 0.06860657676839366, + "grad_norm": 2.3723716735839844, + "learning_rate": 3.4739999999999997e-06, + "loss": 0.5073, + "step": 580 + }, + { + "epoch": 0.06978944878164182, + "grad_norm": 2.6958115100860596, + "learning_rate": 3.534e-06, + "loss": 0.5293, + "step": 590 + }, + { + "epoch": 0.07097232079488999, + "grad_norm": 3.673887014389038, + "learning_rate": 3.594e-06, + "loss": 0.4858, + "step": 600 + }, + { + "epoch": 0.07215519280813816, + "grad_norm": 3.069746255874634, + "learning_rate": 3.654e-06, + "loss": 0.4787, + "step": 610 + }, + { + "epoch": 0.07333806482138633, + "grad_norm": 1.7422595024108887, + "learning_rate": 3.714e-06, + "loss": 0.4961, + "step": 620 + }, + { + "epoch": 0.0745209368346345, + "grad_norm": 3.383603572845459, + "learning_rate": 3.774e-06, + "loss": 0.5233, + "step": 630 + }, + { + "epoch": 0.07570380884788266, + "grad_norm": 6.0858869552612305, + "learning_rate": 3.834e-06, + "loss": 0.4821, + "step": 640 + }, + { + "epoch": 0.07688668086113083, + "grad_norm": 2.8031091690063477, + "learning_rate": 3.894e-06, + "loss": 0.5011, + "step": 650 + }, + { + "epoch": 0.07806955287437899, + "grad_norm": 2.313732147216797, + "learning_rate": 3.954000000000001e-06, + "loss": 0.4887, + "step": 660 + }, + { + "epoch": 0.07925242488762715, + "grad_norm": 3.6443533897399902, + "learning_rate": 4.014e-06, + "loss": 0.459, + "step": 670 + }, + { + "epoch": 0.08043529690087532, + "grad_norm": 4.758196830749512, + "learning_rate": 4.074e-06, + "loss": 0.4168, + "step": 680 + }, + { + "epoch": 0.08161816891412349, + "grad_norm": 2.776780366897583, + "learning_rate": 4.134e-06, + "loss": 0.4426, + "step": 690 + }, + { + "epoch": 0.08280104092737166, + "grad_norm": 2.4684460163116455, + "learning_rate": 4.194e-06, + "loss": 0.4241, + "step": 700 + }, + { + "epoch": 0.08398391294061983, + "grad_norm": 2.3552892208099365, + "learning_rate": 4.254e-06, + "loss": 0.4329, + "step": 710 + }, + { + "epoch": 0.085166784953868, + "grad_norm": 3.4647741317749023, + "learning_rate": 4.314e-06, + "loss": 0.4323, + "step": 720 + }, + { + "epoch": 0.08634965696711616, + "grad_norm": 2.714560031890869, + "learning_rate": 4.374e-06, + "loss": 0.4453, + "step": 730 + }, + { + "epoch": 0.08753252898036433, + "grad_norm": 3.3246028423309326, + "learning_rate": 4.434e-06, + "loss": 0.4311, + "step": 740 + }, + { + "epoch": 0.0887154009936125, + "grad_norm": 2.8064894676208496, + "learning_rate": 4.4940000000000005e-06, + "loss": 0.3956, + "step": 750 + }, + { + "epoch": 0.08989827300686065, + "grad_norm": 4.829779624938965, + "learning_rate": 4.554e-06, + "loss": 0.4217, + "step": 760 + }, + { + "epoch": 0.09108114502010882, + "grad_norm": 4.982199668884277, + "learning_rate": 4.614e-06, + "loss": 0.4535, + "step": 770 + }, + { + "epoch": 0.09226401703335699, + "grad_norm": 4.116279125213623, + "learning_rate": 4.6740000000000005e-06, + "loss": 0.4447, + "step": 780 + }, + { + "epoch": 0.09344688904660516, + "grad_norm": 3.007605791091919, + "learning_rate": 4.734e-06, + "loss": 0.4308, + "step": 790 + }, + { + "epoch": 0.09462976105985332, + "grad_norm": 6.277472972869873, + "learning_rate": 4.794e-06, + "loss": 0.4705, + "step": 800 + }, + { + "epoch": 0.09581263307310149, + "grad_norm": 2.7225468158721924, + "learning_rate": 4.8540000000000005e-06, + "loss": 0.4687, + "step": 810 + }, + { + "epoch": 0.09699550508634966, + "grad_norm": 3.167908191680908, + "learning_rate": 4.914e-06, + "loss": 0.4203, + "step": 820 + }, + { + "epoch": 0.09817837709959783, + "grad_norm": 3.6750435829162598, + "learning_rate": 4.974e-06, + "loss": 0.424, + "step": 830 + }, + { + "epoch": 0.099361249112846, + "grad_norm": 5.332252025604248, + "learning_rate": 5.0339999999999996e-06, + "loss": 0.4211, + "step": 840 + }, + { + "epoch": 0.10054412112609416, + "grad_norm": 6.136302471160889, + "learning_rate": 5.094e-06, + "loss": 0.4367, + "step": 850 + }, + { + "epoch": 0.10172699313934232, + "grad_norm": 4.148458003997803, + "learning_rate": 5.154e-06, + "loss": 0.4384, + "step": 860 + }, + { + "epoch": 0.10290986515259049, + "grad_norm": 4.16589879989624, + "learning_rate": 5.214e-06, + "loss": 0.4513, + "step": 870 + }, + { + "epoch": 0.10409273716583865, + "grad_norm": 3.297203302383423, + "learning_rate": 5.274e-06, + "loss": 0.4404, + "step": 880 + }, + { + "epoch": 0.10527560917908682, + "grad_norm": 4.277662754058838, + "learning_rate": 5.334e-06, + "loss": 0.4283, + "step": 890 + }, + { + "epoch": 0.10645848119233499, + "grad_norm": 2.4507994651794434, + "learning_rate": 5.394e-06, + "loss": 0.4375, + "step": 900 + }, + { + "epoch": 0.10764135320558316, + "grad_norm": 2.6106863021850586, + "learning_rate": 5.454000000000001e-06, + "loss": 0.4293, + "step": 910 + }, + { + "epoch": 0.10882422521883132, + "grad_norm": 2.76210618019104, + "learning_rate": 5.514e-06, + "loss": 0.4118, + "step": 920 + }, + { + "epoch": 0.11000709723207949, + "grad_norm": 6.881728649139404, + "learning_rate": 5.574e-06, + "loss": 0.3949, + "step": 930 + }, + { + "epoch": 0.11118996924532766, + "grad_norm": 4.583062171936035, + "learning_rate": 5.634e-06, + "loss": 0.4289, + "step": 940 + }, + { + "epoch": 0.11237284125857583, + "grad_norm": 3.283390522003174, + "learning_rate": 5.694e-06, + "loss": 0.4056, + "step": 950 + }, + { + "epoch": 0.11355571327182398, + "grad_norm": 5.579883575439453, + "learning_rate": 5.754e-06, + "loss": 0.4345, + "step": 960 + }, + { + "epoch": 0.11473858528507215, + "grad_norm": 2.8114514350891113, + "learning_rate": 5.814e-06, + "loss": 0.4077, + "step": 970 + }, + { + "epoch": 0.11592145729832032, + "grad_norm": 2.615882396697998, + "learning_rate": 5.874e-06, + "loss": 0.3869, + "step": 980 + }, + { + "epoch": 0.11710432931156849, + "grad_norm": 2.7312803268432617, + "learning_rate": 5.934e-06, + "loss": 0.3984, + "step": 990 + }, + { + "epoch": 0.11828720132481665, + "grad_norm": 2.6970901489257812, + "learning_rate": 5.9940000000000005e-06, + "loss": 0.4326, + "step": 1000 + }, + { + "epoch": 0.11947007333806482, + "grad_norm": 3.9762089252471924, + "learning_rate": 5.999353603064401e-06, + "loss": 0.4402, + "step": 1010 + }, + { + "epoch": 0.12065294535131299, + "grad_norm": 3.597429037094116, + "learning_rate": 5.998635384247068e-06, + "loss": 0.4223, + "step": 1020 + }, + { + "epoch": 0.12183581736456116, + "grad_norm": 4.2126617431640625, + "learning_rate": 5.997917165429735e-06, + "loss": 0.3894, + "step": 1030 + }, + { + "epoch": 0.12301868937780933, + "grad_norm": 3.482478618621826, + "learning_rate": 5.9971989466124016e-06, + "loss": 0.416, + "step": 1040 + }, + { + "epoch": 0.1242015613910575, + "grad_norm": 4.943940162658691, + "learning_rate": 5.9964807277950685e-06, + "loss": 0.4042, + "step": 1050 + }, + { + "epoch": 0.12538443340430566, + "grad_norm": 3.6237728595733643, + "learning_rate": 5.995762508977735e-06, + "loss": 0.4751, + "step": 1060 + }, + { + "epoch": 0.12656730541755382, + "grad_norm": 3.268051862716675, + "learning_rate": 5.995044290160402e-06, + "loss": 0.4226, + "step": 1070 + }, + { + "epoch": 0.127750177430802, + "grad_norm": 3.9961740970611572, + "learning_rate": 5.994326071343069e-06, + "loss": 0.4644, + "step": 1080 + }, + { + "epoch": 0.12893304944405015, + "grad_norm": 4.443755149841309, + "learning_rate": 5.993607852525736e-06, + "loss": 0.4504, + "step": 1090 + }, + { + "epoch": 0.13011592145729833, + "grad_norm": 5.0579657554626465, + "learning_rate": 5.992889633708403e-06, + "loss": 0.416, + "step": 1100 + }, + { + "epoch": 0.1312987934705465, + "grad_norm": 2.979658365249634, + "learning_rate": 5.99217141489107e-06, + "loss": 0.4218, + "step": 1110 + }, + { + "epoch": 0.13248166548379464, + "grad_norm": 3.2593445777893066, + "learning_rate": 5.991453196073737e-06, + "loss": 0.4365, + "step": 1120 + }, + { + "epoch": 0.13366453749704282, + "grad_norm": 3.9376895427703857, + "learning_rate": 5.990734977256405e-06, + "loss": 0.426, + "step": 1130 + }, + { + "epoch": 0.13484740951029098, + "grad_norm": 3.19195818901062, + "learning_rate": 5.990016758439071e-06, + "loss": 0.3864, + "step": 1140 + }, + { + "epoch": 0.13603028152353916, + "grad_norm": 3.565159320831299, + "learning_rate": 5.9892985396217386e-06, + "loss": 0.3965, + "step": 1150 + }, + { + "epoch": 0.1372131535367873, + "grad_norm": 3.654405355453491, + "learning_rate": 5.988580320804405e-06, + "loss": 0.4216, + "step": 1160 + }, + { + "epoch": 0.1383960255500355, + "grad_norm": 2.9028587341308594, + "learning_rate": 5.987862101987072e-06, + "loss": 0.4401, + "step": 1170 + }, + { + "epoch": 0.13957889756328365, + "grad_norm": 3.0010299682617188, + "learning_rate": 5.987143883169739e-06, + "loss": 0.4155, + "step": 1180 + }, + { + "epoch": 0.14076176957653183, + "grad_norm": 3.2944576740264893, + "learning_rate": 5.986425664352406e-06, + "loss": 0.4355, + "step": 1190 + }, + { + "epoch": 0.14194464158977999, + "grad_norm": 2.1551876068115234, + "learning_rate": 5.985707445535073e-06, + "loss": 0.3769, + "step": 1200 + }, + { + "epoch": 0.14312751360302814, + "grad_norm": 2.915942668914795, + "learning_rate": 5.98498922671774e-06, + "loss": 0.4033, + "step": 1210 + }, + { + "epoch": 0.14431038561627632, + "grad_norm": 7.247549533843994, + "learning_rate": 5.984271007900407e-06, + "loss": 0.4371, + "step": 1220 + }, + { + "epoch": 0.14549325762952448, + "grad_norm": 6.543298721313477, + "learning_rate": 5.983552789083074e-06, + "loss": 0.4329, + "step": 1230 + }, + { + "epoch": 0.14667612964277266, + "grad_norm": 2.8402793407440186, + "learning_rate": 5.982834570265742e-06, + "loss": 0.4176, + "step": 1240 + }, + { + "epoch": 0.1478590016560208, + "grad_norm": 3.3635027408599854, + "learning_rate": 5.982116351448408e-06, + "loss": 0.3844, + "step": 1250 + }, + { + "epoch": 0.149041873669269, + "grad_norm": 4.37659215927124, + "learning_rate": 5.9813981326310756e-06, + "loss": 0.4131, + "step": 1260 + }, + { + "epoch": 0.15022474568251715, + "grad_norm": 2.964122772216797, + "learning_rate": 5.980679913813742e-06, + "loss": 0.3912, + "step": 1270 + }, + { + "epoch": 0.15140761769576533, + "grad_norm": 3.0633158683776855, + "learning_rate": 5.979961694996409e-06, + "loss": 0.4181, + "step": 1280 + }, + { + "epoch": 0.15259048970901348, + "grad_norm": 2.4021260738372803, + "learning_rate": 5.979243476179076e-06, + "loss": 0.3976, + "step": 1290 + }, + { + "epoch": 0.15377336172226166, + "grad_norm": 5.8716301918029785, + "learning_rate": 5.978525257361743e-06, + "loss": 0.3847, + "step": 1300 + }, + { + "epoch": 0.15495623373550982, + "grad_norm": 3.9792373180389404, + "learning_rate": 5.97780703854441e-06, + "loss": 0.4625, + "step": 1310 + }, + { + "epoch": 0.15613910574875797, + "grad_norm": 2.3841233253479004, + "learning_rate": 5.977088819727077e-06, + "loss": 0.3979, + "step": 1320 + }, + { + "epoch": 0.15732197776200615, + "grad_norm": 3.4258790016174316, + "learning_rate": 5.976370600909744e-06, + "loss": 0.4252, + "step": 1330 + }, + { + "epoch": 0.1585048497752543, + "grad_norm": 3.380558729171753, + "learning_rate": 5.975652382092411e-06, + "loss": 0.3861, + "step": 1340 + }, + { + "epoch": 0.1596877217885025, + "grad_norm": 3.1744539737701416, + "learning_rate": 5.974934163275078e-06, + "loss": 0.4341, + "step": 1350 + }, + { + "epoch": 0.16087059380175064, + "grad_norm": 4.253352165222168, + "learning_rate": 5.974215944457745e-06, + "loss": 0.4012, + "step": 1360 + }, + { + "epoch": 0.16205346581499883, + "grad_norm": 3.2548880577087402, + "learning_rate": 5.973497725640412e-06, + "loss": 0.4238, + "step": 1370 + }, + { + "epoch": 0.16323633782824698, + "grad_norm": 3.095524549484253, + "learning_rate": 5.972779506823079e-06, + "loss": 0.4042, + "step": 1380 + }, + { + "epoch": 0.16441920984149516, + "grad_norm": 3.0992841720581055, + "learning_rate": 5.9720612880057456e-06, + "loss": 0.3896, + "step": 1390 + }, + { + "epoch": 0.16560208185474332, + "grad_norm": 2.9010965824127197, + "learning_rate": 5.971343069188413e-06, + "loss": 0.375, + "step": 1400 + }, + { + "epoch": 0.16678495386799147, + "grad_norm": 3.7394485473632812, + "learning_rate": 5.970624850371079e-06, + "loss": 0.3699, + "step": 1410 + }, + { + "epoch": 0.16796782588123965, + "grad_norm": 2.4592206478118896, + "learning_rate": 5.969906631553747e-06, + "loss": 0.4302, + "step": 1420 + }, + { + "epoch": 0.1691506978944878, + "grad_norm": 3.285252332687378, + "learning_rate": 5.969188412736413e-06, + "loss": 0.4148, + "step": 1430 + }, + { + "epoch": 0.170333569907736, + "grad_norm": 2.6802074909210205, + "learning_rate": 5.968470193919081e-06, + "loss": 0.3806, + "step": 1440 + }, + { + "epoch": 0.17151644192098414, + "grad_norm": 4.508257865905762, + "learning_rate": 5.967751975101747e-06, + "loss": 0.4043, + "step": 1450 + }, + { + "epoch": 0.17269931393423232, + "grad_norm": 2.8117973804473877, + "learning_rate": 5.967033756284415e-06, + "loss": 0.3803, + "step": 1460 + }, + { + "epoch": 0.17388218594748048, + "grad_norm": 2.628289222717285, + "learning_rate": 5.966315537467082e-06, + "loss": 0.3646, + "step": 1470 + }, + { + "epoch": 0.17506505796072866, + "grad_norm": 2.9634876251220703, + "learning_rate": 5.965597318649749e-06, + "loss": 0.3857, + "step": 1480 + }, + { + "epoch": 0.1762479299739768, + "grad_norm": 6.016693115234375, + "learning_rate": 5.9648790998324165e-06, + "loss": 0.4257, + "step": 1490 + }, + { + "epoch": 0.177430801987225, + "grad_norm": 3.9821629524230957, + "learning_rate": 5.9641608810150826e-06, + "loss": 0.408, + "step": 1500 + }, + { + "epoch": 0.17861367400047315, + "grad_norm": 3.5072484016418457, + "learning_rate": 5.96344266219775e-06, + "loss": 0.372, + "step": 1510 + }, + { + "epoch": 0.1797965460137213, + "grad_norm": 5.004875659942627, + "learning_rate": 5.962724443380416e-06, + "loss": 0.3937, + "step": 1520 + }, + { + "epoch": 0.18097941802696949, + "grad_norm": 2.6555981636047363, + "learning_rate": 5.962006224563084e-06, + "loss": 0.3813, + "step": 1530 + }, + { + "epoch": 0.18216229004021764, + "grad_norm": 2.96158504486084, + "learning_rate": 5.96128800574575e-06, + "loss": 0.3965, + "step": 1540 + }, + { + "epoch": 0.18334516205346582, + "grad_norm": 5.002642631530762, + "learning_rate": 5.960569786928418e-06, + "loss": 0.4257, + "step": 1550 + }, + { + "epoch": 0.18452803406671398, + "grad_norm": 3.1771128177642822, + "learning_rate": 5.959851568111085e-06, + "loss": 0.3539, + "step": 1560 + }, + { + "epoch": 0.18571090607996216, + "grad_norm": 3.9393372535705566, + "learning_rate": 5.959133349293752e-06, + "loss": 0.4115, + "step": 1570 + }, + { + "epoch": 0.1868937780932103, + "grad_norm": 3.128077268600464, + "learning_rate": 5.958415130476419e-06, + "loss": 0.4221, + "step": 1580 + }, + { + "epoch": 0.1880766501064585, + "grad_norm": 2.337475061416626, + "learning_rate": 5.957696911659086e-06, + "loss": 0.3709, + "step": 1590 + }, + { + "epoch": 0.18925952211970665, + "grad_norm": 3.5397274494171143, + "learning_rate": 5.956978692841753e-06, + "loss": 0.4058, + "step": 1600 + }, + { + "epoch": 0.1904423941329548, + "grad_norm": 2.622213363647461, + "learning_rate": 5.9562604740244196e-06, + "loss": 0.4133, + "step": 1610 + }, + { + "epoch": 0.19162526614620298, + "grad_norm": 3.6349985599517822, + "learning_rate": 5.9555422552070865e-06, + "loss": 0.4253, + "step": 1620 + }, + { + "epoch": 0.19280813815945114, + "grad_norm": 4.266642093658447, + "learning_rate": 5.954824036389753e-06, + "loss": 0.388, + "step": 1630 + }, + { + "epoch": 0.19399101017269932, + "grad_norm": 3.642580986022949, + "learning_rate": 5.95410581757242e-06, + "loss": 0.3962, + "step": 1640 + }, + { + "epoch": 0.19517388218594747, + "grad_norm": 2.2537295818328857, + "learning_rate": 5.953387598755087e-06, + "loss": 0.3534, + "step": 1650 + }, + { + "epoch": 0.19635675419919565, + "grad_norm": 2.525167942047119, + "learning_rate": 5.952669379937754e-06, + "loss": 0.3834, + "step": 1660 + }, + { + "epoch": 0.1975396262124438, + "grad_norm": 2.454159736633301, + "learning_rate": 5.951951161120422e-06, + "loss": 0.3894, + "step": 1670 + }, + { + "epoch": 0.198722498225692, + "grad_norm": 1.8713299036026, + "learning_rate": 5.951232942303088e-06, + "loss": 0.3597, + "step": 1680 + }, + { + "epoch": 0.19990537023894014, + "grad_norm": 2.86362886428833, + "learning_rate": 5.950514723485756e-06, + "loss": 0.3814, + "step": 1690 + }, + { + "epoch": 0.20108824225218833, + "grad_norm": 2.5308420658111572, + "learning_rate": 5.949796504668422e-06, + "loss": 0.3884, + "step": 1700 + }, + { + "epoch": 0.20227111426543648, + "grad_norm": 3.108450412750244, + "learning_rate": 5.94907828585109e-06, + "loss": 0.3671, + "step": 1710 + }, + { + "epoch": 0.20345398627868463, + "grad_norm": 3.3343942165374756, + "learning_rate": 5.948360067033756e-06, + "loss": 0.3887, + "step": 1720 + }, + { + "epoch": 0.20463685829193282, + "grad_norm": 4.274463653564453, + "learning_rate": 5.9476418482164235e-06, + "loss": 0.4278, + "step": 1730 + }, + { + "epoch": 0.20581973030518097, + "grad_norm": 3.6236939430236816, + "learning_rate": 5.94692362939909e-06, + "loss": 0.3704, + "step": 1740 + }, + { + "epoch": 0.20700260231842915, + "grad_norm": 2.8490209579467773, + "learning_rate": 5.946205410581757e-06, + "loss": 0.3971, + "step": 1750 + }, + { + "epoch": 0.2081854743316773, + "grad_norm": 4.853827953338623, + "learning_rate": 5.945487191764425e-06, + "loss": 0.3956, + "step": 1760 + }, + { + "epoch": 0.2093683463449255, + "grad_norm": 6.677058219909668, + "learning_rate": 5.944768972947091e-06, + "loss": 0.3551, + "step": 1770 + }, + { + "epoch": 0.21055121835817364, + "grad_norm": 4.976045608520508, + "learning_rate": 5.944050754129759e-06, + "loss": 0.3735, + "step": 1780 + }, + { + "epoch": 0.21173409037142182, + "grad_norm": 2.5858116149902344, + "learning_rate": 5.943332535312425e-06, + "loss": 0.3669, + "step": 1790 + }, + { + "epoch": 0.21291696238466998, + "grad_norm": 3.8134396076202393, + "learning_rate": 5.942614316495093e-06, + "loss": 0.3608, + "step": 1800 + }, + { + "epoch": 0.21409983439791813, + "grad_norm": 2.15547251701355, + "learning_rate": 5.941896097677759e-06, + "loss": 0.3849, + "step": 1810 + }, + { + "epoch": 0.2152827064111663, + "grad_norm": 4.892831325531006, + "learning_rate": 5.941177878860427e-06, + "loss": 0.3866, + "step": 1820 + }, + { + "epoch": 0.21646557842441447, + "grad_norm": 2.8356385231018066, + "learning_rate": 5.9404596600430936e-06, + "loss": 0.3836, + "step": 1830 + }, + { + "epoch": 0.21764845043766265, + "grad_norm": 2.7952866554260254, + "learning_rate": 5.9397414412257605e-06, + "loss": 0.4047, + "step": 1840 + }, + { + "epoch": 0.2188313224509108, + "grad_norm": 2.3543381690979004, + "learning_rate": 5.939023222408427e-06, + "loss": 0.3844, + "step": 1850 + }, + { + "epoch": 0.22001419446415899, + "grad_norm": 2.464482307434082, + "learning_rate": 5.938305003591094e-06, + "loss": 0.4003, + "step": 1860 + }, + { + "epoch": 0.22119706647740714, + "grad_norm": 3.2464239597320557, + "learning_rate": 5.937586784773761e-06, + "loss": 0.3808, + "step": 1870 + }, + { + "epoch": 0.22237993849065532, + "grad_norm": 1.8649559020996094, + "learning_rate": 5.936868565956428e-06, + "loss": 0.3898, + "step": 1880 + }, + { + "epoch": 0.22356281050390348, + "grad_norm": 3.750849723815918, + "learning_rate": 5.936150347139095e-06, + "loss": 0.3423, + "step": 1890 + }, + { + "epoch": 0.22474568251715166, + "grad_norm": 4.348412036895752, + "learning_rate": 5.935432128321762e-06, + "loss": 0.3948, + "step": 1900 + }, + { + "epoch": 0.2259285545303998, + "grad_norm": 3.9699575901031494, + "learning_rate": 5.934713909504429e-06, + "loss": 0.3497, + "step": 1910 + }, + { + "epoch": 0.22711142654364797, + "grad_norm": 3.1678178310394287, + "learning_rate": 5.933995690687096e-06, + "loss": 0.4167, + "step": 1920 + }, + { + "epoch": 0.22829429855689615, + "grad_norm": 2.6658902168273926, + "learning_rate": 5.933277471869763e-06, + "loss": 0.3526, + "step": 1930 + }, + { + "epoch": 0.2294771705701443, + "grad_norm": 4.206330299377441, + "learning_rate": 5.9325592530524306e-06, + "loss": 0.4197, + "step": 1940 + }, + { + "epoch": 0.23066004258339248, + "grad_norm": 2.8765106201171875, + "learning_rate": 5.931841034235097e-06, + "loss": 0.3971, + "step": 1950 + }, + { + "epoch": 0.23184291459664064, + "grad_norm": 4.623599052429199, + "learning_rate": 5.931122815417764e-06, + "loss": 0.3781, + "step": 1960 + }, + { + "epoch": 0.23302578660988882, + "grad_norm": 2.4479308128356934, + "learning_rate": 5.9304045966004305e-06, + "loss": 0.3967, + "step": 1970 + }, + { + "epoch": 0.23420865862313697, + "grad_norm": 3.326092481613159, + "learning_rate": 5.929686377783098e-06, + "loss": 0.382, + "step": 1980 + }, + { + "epoch": 0.23539153063638515, + "grad_norm": 2.986705780029297, + "learning_rate": 5.928968158965765e-06, + "loss": 0.4075, + "step": 1990 + }, + { + "epoch": 0.2365744026496333, + "grad_norm": 3.2481510639190674, + "learning_rate": 5.928249940148432e-06, + "loss": 0.3972, + "step": 2000 + }, + { + "epoch": 0.23775727466288146, + "grad_norm": 2.377636194229126, + "learning_rate": 5.927531721331099e-06, + "loss": 0.3621, + "step": 2010 + }, + { + "epoch": 0.23894014667612964, + "grad_norm": 2.9155020713806152, + "learning_rate": 5.926813502513766e-06, + "loss": 0.3469, + "step": 2020 + }, + { + "epoch": 0.2401230186893778, + "grad_norm": 4.382843017578125, + "learning_rate": 5.926095283696433e-06, + "loss": 0.3988, + "step": 2030 + }, + { + "epoch": 0.24130589070262598, + "grad_norm": 2.594512462615967, + "learning_rate": 5.9253770648791e-06, + "loss": 0.3548, + "step": 2040 + }, + { + "epoch": 0.24248876271587413, + "grad_norm": 2.9058191776275635, + "learning_rate": 5.9246588460617676e-06, + "loss": 0.3762, + "step": 2050 + }, + { + "epoch": 0.24367163472912232, + "grad_norm": 2.5409326553344727, + "learning_rate": 5.923940627244434e-06, + "loss": 0.3736, + "step": 2060 + }, + { + "epoch": 0.24485450674237047, + "grad_norm": 3.0224828720092773, + "learning_rate": 5.923222408427101e-06, + "loss": 0.3646, + "step": 2070 + }, + { + "epoch": 0.24603737875561865, + "grad_norm": 4.444828987121582, + "learning_rate": 5.9225041896097675e-06, + "loss": 0.4032, + "step": 2080 + }, + { + "epoch": 0.2472202507688668, + "grad_norm": 2.7398457527160645, + "learning_rate": 5.921785970792435e-06, + "loss": 0.3379, + "step": 2090 + }, + { + "epoch": 0.248403122782115, + "grad_norm": 3.4850106239318848, + "learning_rate": 5.921067751975102e-06, + "loss": 0.3529, + "step": 2100 + }, + { + "epoch": 0.24958599479536314, + "grad_norm": 4.8330464363098145, + "learning_rate": 5.920349533157769e-06, + "loss": 0.3224, + "step": 2110 + }, + { + "epoch": 0.2500591436006624, + "eval_accuracy": 0.8331669827328076, + "eval_loss": 0.37664809823036194, + "eval_runtime": 77.8219, + "eval_safe_aucpr": 0.880340223163281, + "eval_safe_f1": 0.8234735007832714, + "eval_safe_fpr": 0.2020268452363152, + "eval_safe_precision": 0.7758540630182421, + "eval_safe_recall": 0.8773206315868431, + "eval_samples_per_second": 772.456, + "eval_steps_per_second": 12.079, + "eval_unsafe_aucpr": 0.9350832296948681, + "eval_unsafe_f1": 0.841851297011748, + "eval_unsafe_fpr": 0.12267936841315635, + "eval_unsafe_precision": 0.8908356694700307, + "eval_unsafe_recall": 0.7979731547636842, + "step": 2114 + }, + { + "epoch": 0.2507688668086113, + "grad_norm": 2.590071201324463, + "learning_rate": 5.919631314340436e-06, + "loss": 0.4043, + "step": 2120 + }, + { + "epoch": 0.25195173882185945, + "grad_norm": 3.865354299545288, + "learning_rate": 5.918913095523103e-06, + "loss": 0.4099, + "step": 2130 + }, + { + "epoch": 0.25313461083510763, + "grad_norm": 2.53672456741333, + "learning_rate": 5.91819487670577e-06, + "loss": 0.3427, + "step": 2140 + }, + { + "epoch": 0.2543174828483558, + "grad_norm": 6.158299446105957, + "learning_rate": 5.917476657888437e-06, + "loss": 0.3877, + "step": 2150 + }, + { + "epoch": 0.255500354861604, + "grad_norm": 4.042518138885498, + "learning_rate": 5.916758439071104e-06, + "loss": 0.3818, + "step": 2160 + }, + { + "epoch": 0.2566832268748521, + "grad_norm": 3.8671555519104004, + "learning_rate": 5.916040220253771e-06, + "loss": 0.3413, + "step": 2170 + }, + { + "epoch": 0.2578660988881003, + "grad_norm": 5.481204986572266, + "learning_rate": 5.9153220014364375e-06, + "loss": 0.3718, + "step": 2180 + }, + { + "epoch": 0.2590489709013485, + "grad_norm": 2.517652750015259, + "learning_rate": 5.9146037826191045e-06, + "loss": 0.3354, + "step": 2190 + }, + { + "epoch": 0.26023184291459667, + "grad_norm": 5.2531046867370605, + "learning_rate": 5.913885563801771e-06, + "loss": 0.4276, + "step": 2200 + }, + { + "epoch": 0.2614147149278448, + "grad_norm": 2.3096024990081787, + "learning_rate": 5.913167344984439e-06, + "loss": 0.412, + "step": 2210 + }, + { + "epoch": 0.262597586941093, + "grad_norm": 3.40287446975708, + "learning_rate": 5.912449126167105e-06, + "loss": 0.3723, + "step": 2220 + }, + { + "epoch": 0.26378045895434116, + "grad_norm": 3.98356556892395, + "learning_rate": 5.911730907349773e-06, + "loss": 0.3836, + "step": 2230 + }, + { + "epoch": 0.2649633309675893, + "grad_norm": 3.054837703704834, + "learning_rate": 5.911012688532439e-06, + "loss": 0.3816, + "step": 2240 + }, + { + "epoch": 0.26614620298083747, + "grad_norm": 3.8618481159210205, + "learning_rate": 5.910294469715107e-06, + "loss": 0.3949, + "step": 2250 + }, + { + "epoch": 0.26732907499408565, + "grad_norm": 5.447487831115723, + "learning_rate": 5.909576250897774e-06, + "loss": 0.3752, + "step": 2260 + }, + { + "epoch": 0.26851194700733383, + "grad_norm": 2.9153223037719727, + "learning_rate": 5.908858032080441e-06, + "loss": 0.3893, + "step": 2270 + }, + { + "epoch": 0.26969481902058196, + "grad_norm": 2.5316803455352783, + "learning_rate": 5.908139813263108e-06, + "loss": 0.3725, + "step": 2280 + }, + { + "epoch": 0.27087769103383014, + "grad_norm": 4.825555324554443, + "learning_rate": 5.9074215944457745e-06, + "loss": 0.3761, + "step": 2290 + }, + { + "epoch": 0.2720605630470783, + "grad_norm": 4.993202209472656, + "learning_rate": 5.9067033756284415e-06, + "loss": 0.3532, + "step": 2300 + }, + { + "epoch": 0.27324343506032645, + "grad_norm": 4.285012245178223, + "learning_rate": 5.905985156811108e-06, + "loss": 0.3674, + "step": 2310 + }, + { + "epoch": 0.2744263070735746, + "grad_norm": 2.8119378089904785, + "learning_rate": 5.905266937993776e-06, + "loss": 0.3785, + "step": 2320 + }, + { + "epoch": 0.2756091790868228, + "grad_norm": 2.9245150089263916, + "learning_rate": 5.904548719176442e-06, + "loss": 0.3995, + "step": 2330 + }, + { + "epoch": 0.276792051100071, + "grad_norm": 3.624386787414551, + "learning_rate": 5.90383050035911e-06, + "loss": 0.394, + "step": 2340 + }, + { + "epoch": 0.2779749231133191, + "grad_norm": 2.3951256275177, + "learning_rate": 5.903112281541776e-06, + "loss": 0.3655, + "step": 2350 + }, + { + "epoch": 0.2791577951265673, + "grad_norm": 4.64055061340332, + "learning_rate": 5.902394062724444e-06, + "loss": 0.3347, + "step": 2360 + }, + { + "epoch": 0.2803406671398155, + "grad_norm": 7.313331604003906, + "learning_rate": 5.90167584390711e-06, + "loss": 0.4117, + "step": 2370 + }, + { + "epoch": 0.28152353915306366, + "grad_norm": 3.114755868911743, + "learning_rate": 5.900957625089778e-06, + "loss": 0.3871, + "step": 2380 + }, + { + "epoch": 0.2827064111663118, + "grad_norm": 2.2430553436279297, + "learning_rate": 5.900239406272445e-06, + "loss": 0.3798, + "step": 2390 + }, + { + "epoch": 0.28388928317955997, + "grad_norm": 2.886373519897461, + "learning_rate": 5.8995211874551115e-06, + "loss": 0.3363, + "step": 2400 + }, + { + "epoch": 0.28507215519280815, + "grad_norm": 2.8287220001220703, + "learning_rate": 5.8988029686377785e-06, + "loss": 0.3655, + "step": 2410 + }, + { + "epoch": 0.2862550272060563, + "grad_norm": 2.4173007011413574, + "learning_rate": 5.898084749820445e-06, + "loss": 0.3791, + "step": 2420 + }, + { + "epoch": 0.28743789921930446, + "grad_norm": 2.5527825355529785, + "learning_rate": 5.897366531003112e-06, + "loss": 0.3881, + "step": 2430 + }, + { + "epoch": 0.28862077123255264, + "grad_norm": 2.515178918838501, + "learning_rate": 5.896648312185779e-06, + "loss": 0.3542, + "step": 2440 + }, + { + "epoch": 0.2898036432458008, + "grad_norm": 4.246175765991211, + "learning_rate": 5.895930093368446e-06, + "loss": 0.3882, + "step": 2450 + }, + { + "epoch": 0.29098651525904895, + "grad_norm": 4.691845893859863, + "learning_rate": 5.895211874551113e-06, + "loss": 0.391, + "step": 2460 + }, + { + "epoch": 0.29216938727229713, + "grad_norm": 4.220634460449219, + "learning_rate": 5.89449365573378e-06, + "loss": 0.3567, + "step": 2470 + }, + { + "epoch": 0.2933522592855453, + "grad_norm": 4.378153324127197, + "learning_rate": 5.893775436916448e-06, + "loss": 0.3652, + "step": 2480 + }, + { + "epoch": 0.2945351312987935, + "grad_norm": 3.2689096927642822, + "learning_rate": 5.893057218099114e-06, + "loss": 0.326, + "step": 2490 + }, + { + "epoch": 0.2957180033120416, + "grad_norm": 2.6781718730926514, + "learning_rate": 5.892338999281782e-06, + "loss": 0.4234, + "step": 2500 + }, + { + "epoch": 0.2969008753252898, + "grad_norm": 2.7543728351593018, + "learning_rate": 5.8916207804644485e-06, + "loss": 0.3553, + "step": 2510 + }, + { + "epoch": 0.298083747338538, + "grad_norm": 2.7876243591308594, + "learning_rate": 5.8909025616471155e-06, + "loss": 0.3719, + "step": 2520 + }, + { + "epoch": 0.2992666193517861, + "grad_norm": 2.540703058242798, + "learning_rate": 5.890184342829782e-06, + "loss": 0.3428, + "step": 2530 + }, + { + "epoch": 0.3004494913650343, + "grad_norm": 2.270244598388672, + "learning_rate": 5.889466124012449e-06, + "loss": 0.3873, + "step": 2540 + }, + { + "epoch": 0.3016323633782825, + "grad_norm": 2.754493236541748, + "learning_rate": 5.888747905195116e-06, + "loss": 0.3708, + "step": 2550 + }, + { + "epoch": 0.30281523539153066, + "grad_norm": 3.41325306892395, + "learning_rate": 5.888029686377783e-06, + "loss": 0.4017, + "step": 2560 + }, + { + "epoch": 0.3039981074047788, + "grad_norm": 3.529622793197632, + "learning_rate": 5.88731146756045e-06, + "loss": 0.3569, + "step": 2570 + }, + { + "epoch": 0.30518097941802697, + "grad_norm": 4.984379291534424, + "learning_rate": 5.886593248743117e-06, + "loss": 0.3821, + "step": 2580 + }, + { + "epoch": 0.30636385143127515, + "grad_norm": 2.7920539379119873, + "learning_rate": 5.885875029925785e-06, + "loss": 0.3681, + "step": 2590 + }, + { + "epoch": 0.30754672344452333, + "grad_norm": 2.944453716278076, + "learning_rate": 5.885156811108451e-06, + "loss": 0.3617, + "step": 2600 + }, + { + "epoch": 0.30872959545777146, + "grad_norm": 3.739870309829712, + "learning_rate": 5.884438592291119e-06, + "loss": 0.3568, + "step": 2610 + }, + { + "epoch": 0.30991246747101964, + "grad_norm": 3.3336400985717773, + "learning_rate": 5.883720373473785e-06, + "loss": 0.3694, + "step": 2620 + }, + { + "epoch": 0.3110953394842678, + "grad_norm": 5.322179794311523, + "learning_rate": 5.8830021546564525e-06, + "loss": 0.3779, + "step": 2630 + }, + { + "epoch": 0.31227821149751595, + "grad_norm": 2.317539930343628, + "learning_rate": 5.8822839358391185e-06, + "loss": 0.3868, + "step": 2640 + }, + { + "epoch": 0.3134610835107641, + "grad_norm": 5.140389919281006, + "learning_rate": 5.881565717021786e-06, + "loss": 0.3613, + "step": 2650 + }, + { + "epoch": 0.3146439555240123, + "grad_norm": 2.92518949508667, + "learning_rate": 5.880847498204453e-06, + "loss": 0.3851, + "step": 2660 + }, + { + "epoch": 0.3158268275372605, + "grad_norm": 3.6105711460113525, + "learning_rate": 5.88012927938712e-06, + "loss": 0.3477, + "step": 2670 + }, + { + "epoch": 0.3170096995505086, + "grad_norm": 5.269661903381348, + "learning_rate": 5.879411060569787e-06, + "loss": 0.3607, + "step": 2680 + }, + { + "epoch": 0.3181925715637568, + "grad_norm": 4.06647253036499, + "learning_rate": 5.878692841752454e-06, + "loss": 0.4068, + "step": 2690 + }, + { + "epoch": 0.319375443577005, + "grad_norm": 2.253429412841797, + "learning_rate": 5.877974622935121e-06, + "loss": 0.3937, + "step": 2700 + }, + { + "epoch": 0.32055831559025316, + "grad_norm": 3.5654208660125732, + "learning_rate": 5.877256404117788e-06, + "loss": 0.3438, + "step": 2710 + }, + { + "epoch": 0.3217411876035013, + "grad_norm": 3.4543228149414062, + "learning_rate": 5.876538185300455e-06, + "loss": 0.4375, + "step": 2720 + }, + { + "epoch": 0.32292405961674947, + "grad_norm": 4.9412522315979, + "learning_rate": 5.875819966483122e-06, + "loss": 0.3522, + "step": 2730 + }, + { + "epoch": 0.32410693162999765, + "grad_norm": 2.722914457321167, + "learning_rate": 5.875101747665789e-06, + "loss": 0.386, + "step": 2740 + }, + { + "epoch": 0.3252898036432458, + "grad_norm": 4.104313850402832, + "learning_rate": 5.874383528848456e-06, + "loss": 0.3676, + "step": 2750 + }, + { + "epoch": 0.32647267565649396, + "grad_norm": 2.5414774417877197, + "learning_rate": 5.873665310031123e-06, + "loss": 0.3586, + "step": 2760 + }, + { + "epoch": 0.32765554766974214, + "grad_norm": 4.0231218338012695, + "learning_rate": 5.87294709121379e-06, + "loss": 0.3902, + "step": 2770 + }, + { + "epoch": 0.3288384196829903, + "grad_norm": 2.965622901916504, + "learning_rate": 5.872228872396457e-06, + "loss": 0.4128, + "step": 2780 + }, + { + "epoch": 0.33002129169623845, + "grad_norm": 3.556389331817627, + "learning_rate": 5.871510653579124e-06, + "loss": 0.3614, + "step": 2790 + }, + { + "epoch": 0.33120416370948663, + "grad_norm": 3.5115513801574707, + "learning_rate": 5.870792434761791e-06, + "loss": 0.3352, + "step": 2800 + }, + { + "epoch": 0.3323870357227348, + "grad_norm": 5.717298984527588, + "learning_rate": 5.870074215944458e-06, + "loss": 0.4149, + "step": 2810 + }, + { + "epoch": 0.33356990773598294, + "grad_norm": 3.6197214126586914, + "learning_rate": 5.869355997127125e-06, + "loss": 0.3499, + "step": 2820 + }, + { + "epoch": 0.3347527797492311, + "grad_norm": 2.577221155166626, + "learning_rate": 5.868637778309792e-06, + "loss": 0.401, + "step": 2830 + }, + { + "epoch": 0.3359356517624793, + "grad_norm": 3.8798322677612305, + "learning_rate": 5.867919559492459e-06, + "loss": 0.4022, + "step": 2840 + }, + { + "epoch": 0.3371185237757275, + "grad_norm": 2.9696784019470215, + "learning_rate": 5.867201340675126e-06, + "loss": 0.3903, + "step": 2850 + }, + { + "epoch": 0.3383013957889756, + "grad_norm": 2.462855577468872, + "learning_rate": 5.866483121857793e-06, + "loss": 0.3487, + "step": 2860 + }, + { + "epoch": 0.3394842678022238, + "grad_norm": 4.781073570251465, + "learning_rate": 5.8657649030404595e-06, + "loss": 0.3934, + "step": 2870 + }, + { + "epoch": 0.340667139815472, + "grad_norm": 2.6815013885498047, + "learning_rate": 5.865046684223127e-06, + "loss": 0.3409, + "step": 2880 + }, + { + "epoch": 0.34185001182872016, + "grad_norm": 2.956439971923828, + "learning_rate": 5.864328465405793e-06, + "loss": 0.3573, + "step": 2890 + }, + { + "epoch": 0.3430328838419683, + "grad_norm": 2.7877068519592285, + "learning_rate": 5.863610246588461e-06, + "loss": 0.358, + "step": 2900 + }, + { + "epoch": 0.34421575585521647, + "grad_norm": 3.0387110710144043, + "learning_rate": 5.862892027771127e-06, + "loss": 0.4111, + "step": 2910 + }, + { + "epoch": 0.34539862786846465, + "grad_norm": 2.6195383071899414, + "learning_rate": 5.862173808953795e-06, + "loss": 0.37, + "step": 2920 + }, + { + "epoch": 0.3465814998817128, + "grad_norm": 2.8747026920318604, + "learning_rate": 5.861455590136462e-06, + "loss": 0.3414, + "step": 2930 + }, + { + "epoch": 0.34776437189496096, + "grad_norm": 2.943514585494995, + "learning_rate": 5.860737371319129e-06, + "loss": 0.3947, + "step": 2940 + }, + { + "epoch": 0.34894724390820914, + "grad_norm": 3.779690980911255, + "learning_rate": 5.860019152501796e-06, + "loss": 0.3654, + "step": 2950 + }, + { + "epoch": 0.3501301159214573, + "grad_norm": 5.39602518081665, + "learning_rate": 5.859300933684463e-06, + "loss": 0.3774, + "step": 2960 + }, + { + "epoch": 0.35131298793470545, + "grad_norm": 2.6064445972442627, + "learning_rate": 5.8585827148671295e-06, + "loss": 0.3975, + "step": 2970 + }, + { + "epoch": 0.3524958599479536, + "grad_norm": 2.7651381492614746, + "learning_rate": 5.8578644960497965e-06, + "loss": 0.3907, + "step": 2980 + }, + { + "epoch": 0.3536787319612018, + "grad_norm": 5.168848037719727, + "learning_rate": 5.857146277232463e-06, + "loss": 0.4008, + "step": 2990 + }, + { + "epoch": 0.35486160397445, + "grad_norm": 3.5402283668518066, + "learning_rate": 5.85642805841513e-06, + "loss": 0.3633, + "step": 3000 + }, + { + "epoch": 0.3560444759876981, + "grad_norm": 4.439717769622803, + "learning_rate": 5.855709839597798e-06, + "loss": 0.3586, + "step": 3010 + }, + { + "epoch": 0.3572273480009463, + "grad_norm": 4.7232160568237305, + "learning_rate": 5.854991620780465e-06, + "loss": 0.4107, + "step": 3020 + }, + { + "epoch": 0.3584102200141945, + "grad_norm": 4.003586292266846, + "learning_rate": 5.854273401963132e-06, + "loss": 0.3556, + "step": 3030 + }, + { + "epoch": 0.3595930920274426, + "grad_norm": 3.4900431632995605, + "learning_rate": 5.853555183145799e-06, + "loss": 0.4113, + "step": 3040 + }, + { + "epoch": 0.3607759640406908, + "grad_norm": 4.203525066375732, + "learning_rate": 5.852836964328466e-06, + "loss": 0.3604, + "step": 3050 + }, + { + "epoch": 0.36195883605393897, + "grad_norm": 4.817180156707764, + "learning_rate": 5.852118745511133e-06, + "loss": 0.3702, + "step": 3060 + }, + { + "epoch": 0.36314170806718715, + "grad_norm": 3.0071041584014893, + "learning_rate": 5.8514005266938e-06, + "loss": 0.3406, + "step": 3070 + }, + { + "epoch": 0.3643245800804353, + "grad_norm": 2.2780613899230957, + "learning_rate": 5.8506823078764665e-06, + "loss": 0.3781, + "step": 3080 + }, + { + "epoch": 0.36550745209368346, + "grad_norm": 2.8988680839538574, + "learning_rate": 5.8499640890591335e-06, + "loss": 0.3819, + "step": 3090 + }, + { + "epoch": 0.36669032410693164, + "grad_norm": 5.551490783691406, + "learning_rate": 5.8492458702418e-06, + "loss": 0.3656, + "step": 3100 + }, + { + "epoch": 0.3678731961201798, + "grad_norm": 2.538804769515991, + "learning_rate": 5.848527651424467e-06, + "loss": 0.4029, + "step": 3110 + }, + { + "epoch": 0.36905606813342795, + "grad_norm": 2.9220805168151855, + "learning_rate": 5.847809432607134e-06, + "loss": 0.3703, + "step": 3120 + }, + { + "epoch": 0.37023894014667613, + "grad_norm": 4.952260971069336, + "learning_rate": 5.847091213789802e-06, + "loss": 0.401, + "step": 3130 + }, + { + "epoch": 0.3714218121599243, + "grad_norm": 2.877650260925293, + "learning_rate": 5.846372994972468e-06, + "loss": 0.3581, + "step": 3140 + }, + { + "epoch": 0.37260468417317244, + "grad_norm": 2.4547641277313232, + "learning_rate": 5.845654776155136e-06, + "loss": 0.3879, + "step": 3150 + }, + { + "epoch": 0.3737875561864206, + "grad_norm": 2.919623613357544, + "learning_rate": 5.844936557337802e-06, + "loss": 0.4104, + "step": 3160 + }, + { + "epoch": 0.3749704281996688, + "grad_norm": 3.500019073486328, + "learning_rate": 5.84421833852047e-06, + "loss": 0.4147, + "step": 3170 + }, + { + "epoch": 0.376153300212917, + "grad_norm": 2.103543758392334, + "learning_rate": 5.843500119703136e-06, + "loss": 0.4031, + "step": 3180 + }, + { + "epoch": 0.3773361722261651, + "grad_norm": 2.6760292053222656, + "learning_rate": 5.8427819008858035e-06, + "loss": 0.3965, + "step": 3190 + }, + { + "epoch": 0.3785190442394133, + "grad_norm": 4.392319679260254, + "learning_rate": 5.8420636820684704e-06, + "loss": 0.344, + "step": 3200 + }, + { + "epoch": 0.3797019162526615, + "grad_norm": 2.9038119316101074, + "learning_rate": 5.841345463251137e-06, + "loss": 0.3519, + "step": 3210 + }, + { + "epoch": 0.3808847882659096, + "grad_norm": 2.3231678009033203, + "learning_rate": 5.840627244433804e-06, + "loss": 0.3642, + "step": 3220 + }, + { + "epoch": 0.3820676602791578, + "grad_norm": 2.466710090637207, + "learning_rate": 5.839909025616471e-06, + "loss": 0.4116, + "step": 3230 + }, + { + "epoch": 0.38325053229240597, + "grad_norm": 3.0128297805786133, + "learning_rate": 5.839190806799138e-06, + "loss": 0.4079, + "step": 3240 + }, + { + "epoch": 0.38443340430565415, + "grad_norm": 2.3803048133850098, + "learning_rate": 5.838472587981805e-06, + "loss": 0.3639, + "step": 3250 + }, + { + "epoch": 0.3856162763189023, + "grad_norm": 4.2432541847229, + "learning_rate": 5.837754369164473e-06, + "loss": 0.3898, + "step": 3260 + }, + { + "epoch": 0.38679914833215046, + "grad_norm": 3.1298515796661377, + "learning_rate": 5.837036150347139e-06, + "loss": 0.3861, + "step": 3270 + }, + { + "epoch": 0.38798202034539864, + "grad_norm": 3.26191782951355, + "learning_rate": 5.836317931529807e-06, + "loss": 0.4084, + "step": 3280 + }, + { + "epoch": 0.3891648923586468, + "grad_norm": 2.265130043029785, + "learning_rate": 5.835599712712473e-06, + "loss": 0.3743, + "step": 3290 + }, + { + "epoch": 0.39034776437189495, + "grad_norm": 5.0080060958862305, + "learning_rate": 5.8348814938951405e-06, + "loss": 0.3658, + "step": 3300 + }, + { + "epoch": 0.3915306363851431, + "grad_norm": 2.9180374145507812, + "learning_rate": 5.8341632750778074e-06, + "loss": 0.3638, + "step": 3310 + }, + { + "epoch": 0.3927135083983913, + "grad_norm": 4.007022380828857, + "learning_rate": 5.833445056260474e-06, + "loss": 0.4106, + "step": 3320 + }, + { + "epoch": 0.39389638041163944, + "grad_norm": 2.20994234085083, + "learning_rate": 5.832726837443141e-06, + "loss": 0.3307, + "step": 3330 + }, + { + "epoch": 0.3950792524248876, + "grad_norm": 3.977005958557129, + "learning_rate": 5.832008618625808e-06, + "loss": 0.3331, + "step": 3340 + }, + { + "epoch": 0.3962621244381358, + "grad_norm": 5.412336826324463, + "learning_rate": 5.831290399808475e-06, + "loss": 0.3944, + "step": 3350 + }, + { + "epoch": 0.397444996451384, + "grad_norm": 3.2714927196502686, + "learning_rate": 5.830572180991142e-06, + "loss": 0.3718, + "step": 3360 + }, + { + "epoch": 0.3986278684646321, + "grad_norm": 3.78078031539917, + "learning_rate": 5.829853962173809e-06, + "loss": 0.398, + "step": 3370 + }, + { + "epoch": 0.3998107404778803, + "grad_norm": 3.3845877647399902, + "learning_rate": 5.829135743356476e-06, + "loss": 0.3729, + "step": 3380 + }, + { + "epoch": 0.40099361249112847, + "grad_norm": 2.473590850830078, + "learning_rate": 5.828417524539143e-06, + "loss": 0.3623, + "step": 3390 + }, + { + "epoch": 0.40217648450437665, + "grad_norm": 4.318367958068848, + "learning_rate": 5.827699305721811e-06, + "loss": 0.378, + "step": 3400 + }, + { + "epoch": 0.4033593565176248, + "grad_norm": 2.945312261581421, + "learning_rate": 5.826981086904477e-06, + "loss": 0.3766, + "step": 3410 + }, + { + "epoch": 0.40454222853087296, + "grad_norm": 2.770726203918457, + "learning_rate": 5.8262628680871444e-06, + "loss": 0.3723, + "step": 3420 + }, + { + "epoch": 0.40572510054412114, + "grad_norm": 3.8866899013519287, + "learning_rate": 5.8255446492698105e-06, + "loss": 0.3853, + "step": 3430 + }, + { + "epoch": 0.40690797255736927, + "grad_norm": 2.1387736797332764, + "learning_rate": 5.824826430452478e-06, + "loss": 0.3305, + "step": 3440 + }, + { + "epoch": 0.40809084457061745, + "grad_norm": 1.9608129262924194, + "learning_rate": 5.824108211635144e-06, + "loss": 0.3479, + "step": 3450 + }, + { + "epoch": 0.40927371658386563, + "grad_norm": 2.608383893966675, + "learning_rate": 5.823389992817812e-06, + "loss": 0.3677, + "step": 3460 + }, + { + "epoch": 0.4104565885971138, + "grad_norm": 2.4288978576660156, + "learning_rate": 5.822671774000479e-06, + "loss": 0.3472, + "step": 3470 + }, + { + "epoch": 0.41163946061036194, + "grad_norm": 4.163777828216553, + "learning_rate": 5.821953555183146e-06, + "loss": 0.3938, + "step": 3480 + }, + { + "epoch": 0.4128223326236101, + "grad_norm": 3.4677999019622803, + "learning_rate": 5.821235336365813e-06, + "loss": 0.333, + "step": 3490 + }, + { + "epoch": 0.4140052046368583, + "grad_norm": 3.1297054290771484, + "learning_rate": 5.82051711754848e-06, + "loss": 0.317, + "step": 3500 + }, + { + "epoch": 0.4151880766501065, + "grad_norm": 3.443112373352051, + "learning_rate": 5.819798898731148e-06, + "loss": 0.3727, + "step": 3510 + }, + { + "epoch": 0.4163709486633546, + "grad_norm": 2.942333459854126, + "learning_rate": 5.819080679913814e-06, + "loss": 0.3459, + "step": 3520 + }, + { + "epoch": 0.4175538206766028, + "grad_norm": 2.94356107711792, + "learning_rate": 5.8183624610964814e-06, + "loss": 0.3232, + "step": 3530 + }, + { + "epoch": 0.418736692689851, + "grad_norm": 2.3919339179992676, + "learning_rate": 5.8176442422791475e-06, + "loss": 0.366, + "step": 3540 + }, + { + "epoch": 0.4199195647030991, + "grad_norm": 5.207398414611816, + "learning_rate": 5.816926023461815e-06, + "loss": 0.3714, + "step": 3550 + }, + { + "epoch": 0.4211024367163473, + "grad_norm": 3.081705331802368, + "learning_rate": 5.816207804644481e-06, + "loss": 0.373, + "step": 3560 + }, + { + "epoch": 0.42228530872959547, + "grad_norm": 1.8942437171936035, + "learning_rate": 5.815489585827149e-06, + "loss": 0.3721, + "step": 3570 + }, + { + "epoch": 0.42346818074284365, + "grad_norm": 2.5785560607910156, + "learning_rate": 5.814771367009816e-06, + "loss": 0.3455, + "step": 3580 + }, + { + "epoch": 0.4246510527560918, + "grad_norm": 3.5257067680358887, + "learning_rate": 5.814053148192483e-06, + "loss": 0.3375, + "step": 3590 + }, + { + "epoch": 0.42583392476933996, + "grad_norm": 3.3724899291992188, + "learning_rate": 5.81333492937515e-06, + "loss": 0.3863, + "step": 3600 + }, + { + "epoch": 0.42701679678258814, + "grad_norm": 2.5944249629974365, + "learning_rate": 5.812616710557817e-06, + "loss": 0.358, + "step": 3610 + }, + { + "epoch": 0.42819966879583626, + "grad_norm": 2.341181755065918, + "learning_rate": 5.811898491740484e-06, + "loss": 0.4097, + "step": 3620 + }, + { + "epoch": 0.42938254080908445, + "grad_norm": 2.9497246742248535, + "learning_rate": 5.811180272923151e-06, + "loss": 0.3858, + "step": 3630 + }, + { + "epoch": 0.4305654128223326, + "grad_norm": 2.6940529346466064, + "learning_rate": 5.810462054105818e-06, + "loss": 0.3517, + "step": 3640 + }, + { + "epoch": 0.4317482848355808, + "grad_norm": 3.812220811843872, + "learning_rate": 5.8097438352884845e-06, + "loss": 0.3358, + "step": 3650 + }, + { + "epoch": 0.43293115684882894, + "grad_norm": 3.533381700515747, + "learning_rate": 5.8090256164711514e-06, + "loss": 0.3807, + "step": 3660 + }, + { + "epoch": 0.4341140288620771, + "grad_norm": 2.998255968093872, + "learning_rate": 5.808307397653819e-06, + "loss": 0.3836, + "step": 3670 + }, + { + "epoch": 0.4352969008753253, + "grad_norm": 2.8560478687286377, + "learning_rate": 5.807589178836485e-06, + "loss": 0.3445, + "step": 3680 + }, + { + "epoch": 0.4364797728885735, + "grad_norm": 2.7697792053222656, + "learning_rate": 5.806870960019153e-06, + "loss": 0.3202, + "step": 3690 + }, + { + "epoch": 0.4376626449018216, + "grad_norm": 2.462221384048462, + "learning_rate": 5.806152741201819e-06, + "loss": 0.3565, + "step": 3700 + }, + { + "epoch": 0.4388455169150698, + "grad_norm": 2.524683952331543, + "learning_rate": 5.805434522384487e-06, + "loss": 0.374, + "step": 3710 + }, + { + "epoch": 0.44002838892831797, + "grad_norm": 2.5711159706115723, + "learning_rate": 5.804716303567153e-06, + "loss": 0.3755, + "step": 3720 + }, + { + "epoch": 0.4412112609415661, + "grad_norm": 2.726163148880005, + "learning_rate": 5.803998084749821e-06, + "loss": 0.3834, + "step": 3730 + }, + { + "epoch": 0.4423941329548143, + "grad_norm": 2.688717842102051, + "learning_rate": 5.803279865932488e-06, + "loss": 0.3675, + "step": 3740 + }, + { + "epoch": 0.44357700496806246, + "grad_norm": 3.8020083904266357, + "learning_rate": 5.802561647115155e-06, + "loss": 0.3921, + "step": 3750 + }, + { + "epoch": 0.44475987698131064, + "grad_norm": 2.9287753105163574, + "learning_rate": 5.8018434282978215e-06, + "loss": 0.3394, + "step": 3760 + }, + { + "epoch": 0.44594274899455877, + "grad_norm": 2.516939878463745, + "learning_rate": 5.8011252094804884e-06, + "loss": 0.398, + "step": 3770 + }, + { + "epoch": 0.44712562100780695, + "grad_norm": 2.728116273880005, + "learning_rate": 5.800406990663156e-06, + "loss": 0.3568, + "step": 3780 + }, + { + "epoch": 0.44830849302105513, + "grad_norm": 4.527690887451172, + "learning_rate": 5.799688771845822e-06, + "loss": 0.3663, + "step": 3790 + }, + { + "epoch": 0.4494913650343033, + "grad_norm": 3.5766899585723877, + "learning_rate": 5.79897055302849e-06, + "loss": 0.3889, + "step": 3800 + }, + { + "epoch": 0.45067423704755144, + "grad_norm": 3.0356013774871826, + "learning_rate": 5.798252334211156e-06, + "loss": 0.3635, + "step": 3810 + }, + { + "epoch": 0.4518571090607996, + "grad_norm": 4.886003017425537, + "learning_rate": 5.797534115393824e-06, + "loss": 0.3476, + "step": 3820 + }, + { + "epoch": 0.4530399810740478, + "grad_norm": 2.641568899154663, + "learning_rate": 5.79681589657649e-06, + "loss": 0.3702, + "step": 3830 + }, + { + "epoch": 0.45422285308729593, + "grad_norm": 3.4170846939086914, + "learning_rate": 5.796097677759158e-06, + "loss": 0.3623, + "step": 3840 + }, + { + "epoch": 0.4554057251005441, + "grad_norm": 3.7190732955932617, + "learning_rate": 5.795379458941825e-06, + "loss": 0.4053, + "step": 3850 + }, + { + "epoch": 0.4565885971137923, + "grad_norm": 3.826817035675049, + "learning_rate": 5.794661240124492e-06, + "loss": 0.3523, + "step": 3860 + }, + { + "epoch": 0.4577714691270405, + "grad_norm": 2.791008949279785, + "learning_rate": 5.7939430213071585e-06, + "loss": 0.3953, + "step": 3870 + }, + { + "epoch": 0.4589543411402886, + "grad_norm": 2.7187814712524414, + "learning_rate": 5.7932248024898254e-06, + "loss": 0.3532, + "step": 3880 + }, + { + "epoch": 0.4601372131535368, + "grad_norm": 3.3384859561920166, + "learning_rate": 5.792506583672492e-06, + "loss": 0.3645, + "step": 3890 + }, + { + "epoch": 0.46132008516678497, + "grad_norm": 3.3910651206970215, + "learning_rate": 5.791788364855159e-06, + "loss": 0.3393, + "step": 3900 + }, + { + "epoch": 0.46250295718003315, + "grad_norm": 2.875401258468628, + "learning_rate": 5.791070146037826e-06, + "loss": 0.3337, + "step": 3910 + }, + { + "epoch": 0.4636858291932813, + "grad_norm": 2.77256178855896, + "learning_rate": 5.790351927220493e-06, + "loss": 0.3857, + "step": 3920 + }, + { + "epoch": 0.46486870120652946, + "grad_norm": 2.3083810806274414, + "learning_rate": 5.78963370840316e-06, + "loss": 0.3648, + "step": 3930 + }, + { + "epoch": 0.46605157321977764, + "grad_norm": 2.5307555198669434, + "learning_rate": 5.788915489585827e-06, + "loss": 0.3605, + "step": 3940 + }, + { + "epoch": 0.46723444523302576, + "grad_norm": 4.298022747039795, + "learning_rate": 5.788197270768494e-06, + "loss": 0.3479, + "step": 3950 + }, + { + "epoch": 0.46841731724627395, + "grad_norm": 3.032050848007202, + "learning_rate": 5.787479051951162e-06, + "loss": 0.3622, + "step": 3960 + }, + { + "epoch": 0.4696001892595221, + "grad_norm": 3.9290056228637695, + "learning_rate": 5.786760833133828e-06, + "loss": 0.3834, + "step": 3970 + }, + { + "epoch": 0.4707830612727703, + "grad_norm": 2.1643033027648926, + "learning_rate": 5.7860426143164955e-06, + "loss": 0.3634, + "step": 3980 + }, + { + "epoch": 0.47196593328601844, + "grad_norm": 4.665809154510498, + "learning_rate": 5.785324395499162e-06, + "loss": 0.3849, + "step": 3990 + }, + { + "epoch": 0.4731488052992666, + "grad_norm": 2.8869147300720215, + "learning_rate": 5.784606176681829e-06, + "loss": 0.3884, + "step": 4000 + }, + { + "epoch": 0.4743316773125148, + "grad_norm": 2.3516039848327637, + "learning_rate": 5.783887957864496e-06, + "loss": 0.3515, + "step": 4010 + }, + { + "epoch": 0.4755145493257629, + "grad_norm": 2.363905191421509, + "learning_rate": 5.783169739047163e-06, + "loss": 0.3157, + "step": 4020 + }, + { + "epoch": 0.4766974213390111, + "grad_norm": 3.9049179553985596, + "learning_rate": 5.78245152022983e-06, + "loss": 0.3558, + "step": 4030 + }, + { + "epoch": 0.4778802933522593, + "grad_norm": 3.5855188369750977, + "learning_rate": 5.781733301412497e-06, + "loss": 0.3629, + "step": 4040 + }, + { + "epoch": 0.47906316536550747, + "grad_norm": 2.6367034912109375, + "learning_rate": 5.781015082595165e-06, + "loss": 0.3628, + "step": 4050 + }, + { + "epoch": 0.4802460373787556, + "grad_norm": 2.409397602081299, + "learning_rate": 5.780296863777831e-06, + "loss": 0.3406, + "step": 4060 + }, + { + "epoch": 0.4814289093920038, + "grad_norm": 2.5876379013061523, + "learning_rate": 5.779578644960499e-06, + "loss": 0.3215, + "step": 4070 + }, + { + "epoch": 0.48261178140525196, + "grad_norm": 2.785447597503662, + "learning_rate": 5.778860426143165e-06, + "loss": 0.3363, + "step": 4080 + }, + { + "epoch": 0.48379465341850014, + "grad_norm": 2.5361294746398926, + "learning_rate": 5.7781422073258325e-06, + "loss": 0.3546, + "step": 4090 + }, + { + "epoch": 0.48497752543174827, + "grad_norm": 2.828770637512207, + "learning_rate": 5.777423988508499e-06, + "loss": 0.368, + "step": 4100 + }, + { + "epoch": 0.48616039744499645, + "grad_norm": 4.09275484085083, + "learning_rate": 5.776705769691166e-06, + "loss": 0.3641, + "step": 4110 + }, + { + "epoch": 0.48734326945824463, + "grad_norm": 2.1944663524627686, + "learning_rate": 5.775987550873833e-06, + "loss": 0.4035, + "step": 4120 + }, + { + "epoch": 0.48852614147149276, + "grad_norm": 2.6475987434387207, + "learning_rate": 5.7752693320565e-06, + "loss": 0.3765, + "step": 4130 + }, + { + "epoch": 0.48970901348474094, + "grad_norm": 3.410083770751953, + "learning_rate": 5.774551113239167e-06, + "loss": 0.3852, + "step": 4140 + }, + { + "epoch": 0.4908918854979891, + "grad_norm": 2.8198800086975098, + "learning_rate": 5.773832894421834e-06, + "loss": 0.4084, + "step": 4150 + }, + { + "epoch": 0.4920747575112373, + "grad_norm": 2.296844482421875, + "learning_rate": 5.773114675604501e-06, + "loss": 0.3593, + "step": 4160 + }, + { + "epoch": 0.49325762952448543, + "grad_norm": 2.5193898677825928, + "learning_rate": 5.772396456787168e-06, + "loss": 0.3598, + "step": 4170 + }, + { + "epoch": 0.4944405015377336, + "grad_norm": 3.2671520709991455, + "learning_rate": 5.771678237969835e-06, + "loss": 0.3588, + "step": 4180 + }, + { + "epoch": 0.4956233735509818, + "grad_norm": 2.365128517150879, + "learning_rate": 5.770960019152502e-06, + "loss": 0.3568, + "step": 4190 + }, + { + "epoch": 0.49680624556423, + "grad_norm": 4.848801136016846, + "learning_rate": 5.770241800335169e-06, + "loss": 0.348, + "step": 4200 + }, + { + "epoch": 0.4979891175774781, + "grad_norm": 5.710461139678955, + "learning_rate": 5.769523581517836e-06, + "loss": 0.3588, + "step": 4210 + }, + { + "epoch": 0.4991719895907263, + "grad_norm": 3.037342071533203, + "learning_rate": 5.7688053627005025e-06, + "loss": 0.3601, + "step": 4220 + }, + { + "epoch": 0.5001182872013248, + "eval_accuracy": 0.8440795821272915, + "eval_loss": 0.34948381781578064, + "eval_runtime": 77.7073, + "eval_safe_aucpr": 0.8954198959966346, + "eval_safe_f1": 0.823160953153595, + "eval_safe_fpr": 0.13527248811694675, + "eval_safe_precision": 0.8282080485952923, + "eval_safe_recall": 0.8181749990623711, + "eval_samples_per_second": 773.595, + "eval_steps_per_second": 12.097, + "eval_unsafe_aucpr": 0.9433171137892656, + "eval_unsafe_f1": 0.860572703607289, + "eval_unsafe_fpr": 0.18182500093762827, + "eval_unsafe_precision": 0.8564576301296856, + "eval_unsafe_recall": 0.8647275118830529, + "step": 4228 + }, + { + "epoch": 0.5003548616039745, + "grad_norm": 2.9329936504364014, + "learning_rate": 5.76808714388317e-06, + "loss": 0.3792, + "step": 4230 + }, + { + "epoch": 0.5015377336172226, + "grad_norm": 2.630964756011963, + "learning_rate": 5.767368925065836e-06, + "loss": 0.3469, + "step": 4240 + }, + { + "epoch": 0.5027206056304708, + "grad_norm": 3.4824013710021973, + "learning_rate": 5.766650706248504e-06, + "loss": 0.3968, + "step": 4250 + }, + { + "epoch": 0.5039034776437189, + "grad_norm": 2.520982503890991, + "learning_rate": 5.76593248743117e-06, + "loss": 0.3671, + "step": 4260 + }, + { + "epoch": 0.5050863496569671, + "grad_norm": 2.3600871562957764, + "learning_rate": 5.765214268613838e-06, + "loss": 0.3573, + "step": 4270 + }, + { + "epoch": 0.5062692216702153, + "grad_norm": 2.865267515182495, + "learning_rate": 5.764496049796505e-06, + "loss": 0.3421, + "step": 4280 + }, + { + "epoch": 0.5074520936834634, + "grad_norm": 2.9639503955841064, + "learning_rate": 5.763777830979172e-06, + "loss": 0.361, + "step": 4290 + }, + { + "epoch": 0.5086349656967116, + "grad_norm": 2.9893205165863037, + "learning_rate": 5.763059612161839e-06, + "loss": 0.3452, + "step": 4300 + }, + { + "epoch": 0.5098178377099598, + "grad_norm": 4.492689609527588, + "learning_rate": 5.762341393344506e-06, + "loss": 0.3961, + "step": 4310 + }, + { + "epoch": 0.511000709723208, + "grad_norm": 3.260425329208374, + "learning_rate": 5.7616231745271734e-06, + "loss": 0.4085, + "step": 4320 + }, + { + "epoch": 0.5121835817364561, + "grad_norm": 2.229388475418091, + "learning_rate": 5.7609049557098395e-06, + "loss": 0.3584, + "step": 4330 + }, + { + "epoch": 0.5133664537497042, + "grad_norm": 2.2329540252685547, + "learning_rate": 5.760186736892507e-06, + "loss": 0.3875, + "step": 4340 + }, + { + "epoch": 0.5145493257629524, + "grad_norm": 2.5038681030273438, + "learning_rate": 5.759468518075173e-06, + "loss": 0.3654, + "step": 4350 + }, + { + "epoch": 0.5157321977762006, + "grad_norm": 2.483992338180542, + "learning_rate": 5.758750299257841e-06, + "loss": 0.3613, + "step": 4360 + }, + { + "epoch": 0.5169150697894488, + "grad_norm": 2.7301363945007324, + "learning_rate": 5.758032080440507e-06, + "loss": 0.3873, + "step": 4370 + }, + { + "epoch": 0.518097941802697, + "grad_norm": 1.9997198581695557, + "learning_rate": 5.757313861623175e-06, + "loss": 0.3657, + "step": 4380 + }, + { + "epoch": 0.5192808138159452, + "grad_norm": 3.300851345062256, + "learning_rate": 5.756595642805842e-06, + "loss": 0.378, + "step": 4390 + }, + { + "epoch": 0.5204636858291933, + "grad_norm": 4.093571186065674, + "learning_rate": 5.755877423988509e-06, + "loss": 0.3517, + "step": 4400 + }, + { + "epoch": 0.5216465578424414, + "grad_norm": 3.943171977996826, + "learning_rate": 5.755159205171176e-06, + "loss": 0.3609, + "step": 4410 + }, + { + "epoch": 0.5228294298556896, + "grad_norm": 5.173525810241699, + "learning_rate": 5.754440986353843e-06, + "loss": 0.3513, + "step": 4420 + }, + { + "epoch": 0.5240123018689378, + "grad_norm": 3.548471450805664, + "learning_rate": 5.7537227675365096e-06, + "loss": 0.3501, + "step": 4430 + }, + { + "epoch": 0.525195173882186, + "grad_norm": 3.0566258430480957, + "learning_rate": 5.7530045487191765e-06, + "loss": 0.3577, + "step": 4440 + }, + { + "epoch": 0.5263780458954341, + "grad_norm": 3.547243595123291, + "learning_rate": 5.752286329901843e-06, + "loss": 0.359, + "step": 4450 + }, + { + "epoch": 0.5275609179086823, + "grad_norm": 3.283626079559326, + "learning_rate": 5.75156811108451e-06, + "loss": 0.376, + "step": 4460 + }, + { + "epoch": 0.5287437899219305, + "grad_norm": 2.571317672729492, + "learning_rate": 5.750849892267177e-06, + "loss": 0.4127, + "step": 4470 + }, + { + "epoch": 0.5299266619351786, + "grad_norm": 2.866814374923706, + "learning_rate": 5.750131673449844e-06, + "loss": 0.3591, + "step": 4480 + }, + { + "epoch": 0.5311095339484267, + "grad_norm": 2.8994390964508057, + "learning_rate": 5.749413454632511e-06, + "loss": 0.4139, + "step": 4490 + }, + { + "epoch": 0.5322924059616749, + "grad_norm": 2.784627676010132, + "learning_rate": 5.748695235815179e-06, + "loss": 0.352, + "step": 4500 + }, + { + "epoch": 0.5334752779749231, + "grad_norm": 2.3118021488189697, + "learning_rate": 5.747977016997845e-06, + "loss": 0.3363, + "step": 4510 + }, + { + "epoch": 0.5346581499881713, + "grad_norm": 2.584204912185669, + "learning_rate": 5.747258798180513e-06, + "loss": 0.3569, + "step": 4520 + }, + { + "epoch": 0.5358410220014195, + "grad_norm": 2.7564291954040527, + "learning_rate": 5.74654057936318e-06, + "loss": 0.3062, + "step": 4530 + }, + { + "epoch": 0.5370238940146677, + "grad_norm": 3.0223724842071533, + "learning_rate": 5.7458223605458466e-06, + "loss": 0.3547, + "step": 4540 + }, + { + "epoch": 0.5382067660279157, + "grad_norm": 2.4846091270446777, + "learning_rate": 5.7451041417285135e-06, + "loss": 0.3431, + "step": 4550 + }, + { + "epoch": 0.5393896380411639, + "grad_norm": 3.637694835662842, + "learning_rate": 5.74438592291118e-06, + "loss": 0.364, + "step": 4560 + }, + { + "epoch": 0.5405725100544121, + "grad_norm": 2.5037472248077393, + "learning_rate": 5.743667704093847e-06, + "loss": 0.3858, + "step": 4570 + }, + { + "epoch": 0.5417553820676603, + "grad_norm": 2.9102423191070557, + "learning_rate": 5.742949485276514e-06, + "loss": 0.3944, + "step": 4580 + }, + { + "epoch": 0.5429382540809085, + "grad_norm": 3.7652969360351562, + "learning_rate": 5.742231266459182e-06, + "loss": 0.3244, + "step": 4590 + }, + { + "epoch": 0.5441211260941566, + "grad_norm": 2.990288257598877, + "learning_rate": 5.741513047641848e-06, + "loss": 0.3314, + "step": 4600 + }, + { + "epoch": 0.5453039981074048, + "grad_norm": 2.4400131702423096, + "learning_rate": 5.740794828824516e-06, + "loss": 0.3781, + "step": 4610 + }, + { + "epoch": 0.5464868701206529, + "grad_norm": 2.8999412059783936, + "learning_rate": 5.740076610007182e-06, + "loss": 0.3951, + "step": 4620 + }, + { + "epoch": 0.5476697421339011, + "grad_norm": 3.390578508377075, + "learning_rate": 5.73935839118985e-06, + "loss": 0.3489, + "step": 4630 + }, + { + "epoch": 0.5488526141471493, + "grad_norm": 2.4869701862335205, + "learning_rate": 5.738640172372516e-06, + "loss": 0.3451, + "step": 4640 + }, + { + "epoch": 0.5500354861603974, + "grad_norm": 2.432013988494873, + "learning_rate": 5.7379219535551836e-06, + "loss": 0.3418, + "step": 4650 + }, + { + "epoch": 0.5512183581736456, + "grad_norm": 3.0324032306671143, + "learning_rate": 5.7372037347378505e-06, + "loss": 0.3946, + "step": 4660 + }, + { + "epoch": 0.5524012301868938, + "grad_norm": 2.967174530029297, + "learning_rate": 5.736485515920517e-06, + "loss": 0.3793, + "step": 4670 + }, + { + "epoch": 0.553584102200142, + "grad_norm": 2.171663999557495, + "learning_rate": 5.735767297103184e-06, + "loss": 0.3787, + "step": 4680 + }, + { + "epoch": 0.5547669742133902, + "grad_norm": 2.6231937408447266, + "learning_rate": 5.735049078285851e-06, + "loss": 0.3866, + "step": 4690 + }, + { + "epoch": 0.5559498462266382, + "grad_norm": 2.7934768199920654, + "learning_rate": 5.734330859468518e-06, + "loss": 0.3324, + "step": 4700 + }, + { + "epoch": 0.5571327182398864, + "grad_norm": 3.470262289047241, + "learning_rate": 5.733612640651185e-06, + "loss": 0.3173, + "step": 4710 + }, + { + "epoch": 0.5583155902531346, + "grad_norm": 3.697437047958374, + "learning_rate": 5.732894421833852e-06, + "loss": 0.391, + "step": 4720 + }, + { + "epoch": 0.5594984622663828, + "grad_norm": 2.727518320083618, + "learning_rate": 5.732176203016519e-06, + "loss": 0.3807, + "step": 4730 + }, + { + "epoch": 0.560681334279631, + "grad_norm": 2.2123401165008545, + "learning_rate": 5.731457984199186e-06, + "loss": 0.365, + "step": 4740 + }, + { + "epoch": 0.5618642062928791, + "grad_norm": 2.3912441730499268, + "learning_rate": 5.730739765381853e-06, + "loss": 0.3316, + "step": 4750 + }, + { + "epoch": 0.5630470783061273, + "grad_norm": 2.8068788051605225, + "learning_rate": 5.73002154656452e-06, + "loss": 0.3761, + "step": 4760 + }, + { + "epoch": 0.5642299503193754, + "grad_norm": 3.748828887939453, + "learning_rate": 5.7293033277471875e-06, + "loss": 0.3867, + "step": 4770 + }, + { + "epoch": 0.5654128223326236, + "grad_norm": 3.5966951847076416, + "learning_rate": 5.728585108929854e-06, + "loss": 0.3144, + "step": 4780 + }, + { + "epoch": 0.5665956943458718, + "grad_norm": 2.2783448696136475, + "learning_rate": 5.727866890112521e-06, + "loss": 0.3135, + "step": 4790 + }, + { + "epoch": 0.5677785663591199, + "grad_norm": 4.028083801269531, + "learning_rate": 5.727148671295188e-06, + "loss": 0.3491, + "step": 4800 + }, + { + "epoch": 0.5689614383723681, + "grad_norm": 2.8650357723236084, + "learning_rate": 5.726430452477855e-06, + "loss": 0.413, + "step": 4810 + }, + { + "epoch": 0.5701443103856163, + "grad_norm": 2.58620285987854, + "learning_rate": 5.725712233660522e-06, + "loss": 0.3532, + "step": 4820 + }, + { + "epoch": 0.5713271823988645, + "grad_norm": 2.4366037845611572, + "learning_rate": 5.724994014843189e-06, + "loss": 0.3804, + "step": 4830 + }, + { + "epoch": 0.5725100544121126, + "grad_norm": 2.0628104209899902, + "learning_rate": 5.724275796025856e-06, + "loss": 0.3129, + "step": 4840 + }, + { + "epoch": 0.5736929264253607, + "grad_norm": 2.460841417312622, + "learning_rate": 5.723557577208523e-06, + "loss": 0.3626, + "step": 4850 + }, + { + "epoch": 0.5748757984386089, + "grad_norm": 2.232182025909424, + "learning_rate": 5.72283935839119e-06, + "loss": 0.3221, + "step": 4860 + }, + { + "epoch": 0.5760586704518571, + "grad_norm": 3.2975473403930664, + "learning_rate": 5.722121139573857e-06, + "loss": 0.3529, + "step": 4870 + }, + { + "epoch": 0.5772415424651053, + "grad_norm": 3.717776298522949, + "learning_rate": 5.7214029207565245e-06, + "loss": 0.3185, + "step": 4880 + }, + { + "epoch": 0.5784244144783535, + "grad_norm": 2.8947947025299072, + "learning_rate": 5.7206847019391906e-06, + "loss": 0.3667, + "step": 4890 + }, + { + "epoch": 0.5796072864916016, + "grad_norm": 2.6210103034973145, + "learning_rate": 5.719966483121858e-06, + "loss": 0.345, + "step": 4900 + }, + { + "epoch": 0.5807901585048498, + "grad_norm": 2.359241008758545, + "learning_rate": 5.719248264304524e-06, + "loss": 0.3065, + "step": 4910 + }, + { + "epoch": 0.5819730305180979, + "grad_norm": 3.0238962173461914, + "learning_rate": 5.718530045487192e-06, + "loss": 0.3423, + "step": 4920 + }, + { + "epoch": 0.5831559025313461, + "grad_norm": 2.6816389560699463, + "learning_rate": 5.717811826669859e-06, + "loss": 0.34, + "step": 4930 + }, + { + "epoch": 0.5843387745445943, + "grad_norm": 2.968125820159912, + "learning_rate": 5.717093607852526e-06, + "loss": 0.3532, + "step": 4940 + }, + { + "epoch": 0.5855216465578424, + "grad_norm": 2.4294793605804443, + "learning_rate": 5.716375389035193e-06, + "loss": 0.3613, + "step": 4950 + }, + { + "epoch": 0.5867045185710906, + "grad_norm": 3.200100898742676, + "learning_rate": 5.71565717021786e-06, + "loss": 0.3516, + "step": 4960 + }, + { + "epoch": 0.5878873905843388, + "grad_norm": 2.0361745357513428, + "learning_rate": 5.714938951400527e-06, + "loss": 0.3764, + "step": 4970 + }, + { + "epoch": 0.589070262597587, + "grad_norm": 2.789825916290283, + "learning_rate": 5.714220732583194e-06, + "loss": 0.3775, + "step": 4980 + }, + { + "epoch": 0.5902531346108351, + "grad_norm": 2.6486117839813232, + "learning_rate": 5.713502513765861e-06, + "loss": 0.3044, + "step": 4990 + }, + { + "epoch": 0.5914360066240832, + "grad_norm": 3.0227205753326416, + "learning_rate": 5.7127842949485276e-06, + "loss": 0.3552, + "step": 5000 + }, + { + "epoch": 0.5926188786373314, + "grad_norm": 3.0813512802124023, + "learning_rate": 5.7120660761311945e-06, + "loss": 0.349, + "step": 5010 + }, + { + "epoch": 0.5938017506505796, + "grad_norm": 3.0543744564056396, + "learning_rate": 5.711347857313861e-06, + "loss": 0.3271, + "step": 5020 + }, + { + "epoch": 0.5949846226638278, + "grad_norm": 2.3086960315704346, + "learning_rate": 5.710629638496529e-06, + "loss": 0.3666, + "step": 5030 + }, + { + "epoch": 0.596167494677076, + "grad_norm": 3.788231134414673, + "learning_rate": 5.709911419679196e-06, + "loss": 0.3399, + "step": 5040 + }, + { + "epoch": 0.5973503666903242, + "grad_norm": 2.0678153038024902, + "learning_rate": 5.709193200861863e-06, + "loss": 0.3751, + "step": 5050 + }, + { + "epoch": 0.5985332387035722, + "grad_norm": 2.638007879257202, + "learning_rate": 5.70847498204453e-06, + "loss": 0.395, + "step": 5060 + }, + { + "epoch": 0.5997161107168204, + "grad_norm": 3.0321481227874756, + "learning_rate": 5.707756763227197e-06, + "loss": 0.327, + "step": 5070 + }, + { + "epoch": 0.6008989827300686, + "grad_norm": 2.9551167488098145, + "learning_rate": 5.707038544409864e-06, + "loss": 0.3802, + "step": 5080 + }, + { + "epoch": 0.6020818547433168, + "grad_norm": 3.5318825244903564, + "learning_rate": 5.706320325592531e-06, + "loss": 0.3315, + "step": 5090 + }, + { + "epoch": 0.603264726756565, + "grad_norm": 4.272298812866211, + "learning_rate": 5.705602106775198e-06, + "loss": 0.3295, + "step": 5100 + }, + { + "epoch": 0.6044475987698131, + "grad_norm": 3.083944082260132, + "learning_rate": 5.7048838879578646e-06, + "loss": 0.3564, + "step": 5110 + }, + { + "epoch": 0.6056304707830613, + "grad_norm": 3.0958378314971924, + "learning_rate": 5.7041656691405315e-06, + "loss": 0.3427, + "step": 5120 + }, + { + "epoch": 0.6068133427963094, + "grad_norm": 2.100768804550171, + "learning_rate": 5.703447450323198e-06, + "loss": 0.3449, + "step": 5130 + }, + { + "epoch": 0.6079962148095576, + "grad_norm": 2.492600679397583, + "learning_rate": 5.702729231505865e-06, + "loss": 0.3812, + "step": 5140 + }, + { + "epoch": 0.6091790868228057, + "grad_norm": 2.495830774307251, + "learning_rate": 5.702011012688533e-06, + "loss": 0.3339, + "step": 5150 + }, + { + "epoch": 0.6103619588360539, + "grad_norm": 3.3077280521392822, + "learning_rate": 5.701292793871199e-06, + "loss": 0.3747, + "step": 5160 + }, + { + "epoch": 0.6115448308493021, + "grad_norm": 2.5646684169769287, + "learning_rate": 5.700574575053867e-06, + "loss": 0.347, + "step": 5170 + }, + { + "epoch": 0.6127277028625503, + "grad_norm": 4.435232162475586, + "learning_rate": 5.699856356236533e-06, + "loss": 0.3447, + "step": 5180 + }, + { + "epoch": 0.6139105748757985, + "grad_norm": 4.579883098602295, + "learning_rate": 5.699138137419201e-06, + "loss": 0.3769, + "step": 5190 + }, + { + "epoch": 0.6150934468890467, + "grad_norm": 3.25585675239563, + "learning_rate": 5.698419918601867e-06, + "loss": 0.3722, + "step": 5200 + }, + { + "epoch": 0.6162763189022947, + "grad_norm": 3.463557243347168, + "learning_rate": 5.697701699784535e-06, + "loss": 0.3351, + "step": 5210 + }, + { + "epoch": 0.6174591909155429, + "grad_norm": 2.9261481761932373, + "learning_rate": 5.6969834809672016e-06, + "loss": 0.382, + "step": 5220 + }, + { + "epoch": 0.6186420629287911, + "grad_norm": 4.577706813812256, + "learning_rate": 5.6962652621498685e-06, + "loss": 0.3918, + "step": 5230 + }, + { + "epoch": 0.6198249349420393, + "grad_norm": 2.9830856323242188, + "learning_rate": 5.695547043332535e-06, + "loss": 0.3715, + "step": 5240 + }, + { + "epoch": 0.6210078069552875, + "grad_norm": 4.0253071784973145, + "learning_rate": 5.694828824515202e-06, + "loss": 0.4072, + "step": 5250 + }, + { + "epoch": 0.6221906789685356, + "grad_norm": 2.062934637069702, + "learning_rate": 5.694110605697869e-06, + "loss": 0.3296, + "step": 5260 + }, + { + "epoch": 0.6233735509817838, + "grad_norm": 2.803471565246582, + "learning_rate": 5.693392386880536e-06, + "loss": 0.3698, + "step": 5270 + }, + { + "epoch": 0.6245564229950319, + "grad_norm": 3.8620738983154297, + "learning_rate": 5.692674168063204e-06, + "loss": 0.3858, + "step": 5280 + }, + { + "epoch": 0.6257392950082801, + "grad_norm": 2.846827983856201, + "learning_rate": 5.69195594924587e-06, + "loss": 0.3495, + "step": 5290 + }, + { + "epoch": 0.6269221670215283, + "grad_norm": 2.0253992080688477, + "learning_rate": 5.691237730428538e-06, + "loss": 0.3363, + "step": 5300 + }, + { + "epoch": 0.6281050390347764, + "grad_norm": 2.903935432434082, + "learning_rate": 5.690519511611205e-06, + "loss": 0.3236, + "step": 5310 + }, + { + "epoch": 0.6292879110480246, + "grad_norm": 2.9494948387145996, + "learning_rate": 5.689801292793872e-06, + "loss": 0.3778, + "step": 5320 + }, + { + "epoch": 0.6304707830612728, + "grad_norm": 2.0794708728790283, + "learning_rate": 5.6890830739765386e-06, + "loss": 0.33, + "step": 5330 + }, + { + "epoch": 0.631653655074521, + "grad_norm": 2.6170406341552734, + "learning_rate": 5.6883648551592055e-06, + "loss": 0.3391, + "step": 5340 + }, + { + "epoch": 0.632836527087769, + "grad_norm": 2.3067479133605957, + "learning_rate": 5.687646636341872e-06, + "loss": 0.3172, + "step": 5350 + }, + { + "epoch": 0.6340193991010172, + "grad_norm": 3.9792582988739014, + "learning_rate": 5.686928417524539e-06, + "loss": 0.3137, + "step": 5360 + }, + { + "epoch": 0.6352022711142654, + "grad_norm": 3.4487500190734863, + "learning_rate": 5.686210198707206e-06, + "loss": 0.342, + "step": 5370 + }, + { + "epoch": 0.6363851431275136, + "grad_norm": 2.9459004402160645, + "learning_rate": 5.685491979889873e-06, + "loss": 0.3605, + "step": 5380 + }, + { + "epoch": 0.6375680151407618, + "grad_norm": 3.2506394386291504, + "learning_rate": 5.68477376107254e-06, + "loss": 0.3207, + "step": 5390 + }, + { + "epoch": 0.63875088715401, + "grad_norm": 2.6361186504364014, + "learning_rate": 5.684055542255207e-06, + "loss": 0.3793, + "step": 5400 + }, + { + "epoch": 0.6399337591672581, + "grad_norm": 3.1658058166503906, + "learning_rate": 5.683337323437874e-06, + "loss": 0.3764, + "step": 5410 + }, + { + "epoch": 0.6411166311805063, + "grad_norm": 2.703413486480713, + "learning_rate": 5.682619104620542e-06, + "loss": 0.3297, + "step": 5420 + }, + { + "epoch": 0.6422995031937544, + "grad_norm": 3.6086349487304688, + "learning_rate": 5.681900885803208e-06, + "loss": 0.3371, + "step": 5430 + }, + { + "epoch": 0.6434823752070026, + "grad_norm": 3.9716713428497314, + "learning_rate": 5.6811826669858756e-06, + "loss": 0.3965, + "step": 5440 + }, + { + "epoch": 0.6446652472202508, + "grad_norm": 2.501960515975952, + "learning_rate": 5.680464448168542e-06, + "loss": 0.3049, + "step": 5450 + }, + { + "epoch": 0.6458481192334989, + "grad_norm": 3.220345973968506, + "learning_rate": 5.679746229351209e-06, + "loss": 0.3382, + "step": 5460 + }, + { + "epoch": 0.6470309912467471, + "grad_norm": 3.021064043045044, + "learning_rate": 5.6790280105338755e-06, + "loss": 0.3277, + "step": 5470 + }, + { + "epoch": 0.6482138632599953, + "grad_norm": 2.709476947784424, + "learning_rate": 5.678309791716543e-06, + "loss": 0.3224, + "step": 5480 + }, + { + "epoch": 0.6493967352732435, + "grad_norm": 2.4916460514068604, + "learning_rate": 5.67759157289921e-06, + "loss": 0.3742, + "step": 5490 + }, + { + "epoch": 0.6505796072864916, + "grad_norm": 3.362166404724121, + "learning_rate": 5.676873354081877e-06, + "loss": 0.3745, + "step": 5500 + }, + { + "epoch": 0.6517624792997397, + "grad_norm": 2.5697948932647705, + "learning_rate": 5.676155135264544e-06, + "loss": 0.4016, + "step": 5510 + }, + { + "epoch": 0.6529453513129879, + "grad_norm": 2.7755134105682373, + "learning_rate": 5.675436916447211e-06, + "loss": 0.3634, + "step": 5520 + }, + { + "epoch": 0.6541282233262361, + "grad_norm": 3.1397430896759033, + "learning_rate": 5.674718697629878e-06, + "loss": 0.3518, + "step": 5530 + }, + { + "epoch": 0.6553110953394843, + "grad_norm": 3.169858694076538, + "learning_rate": 5.674000478812545e-06, + "loss": 0.3747, + "step": 5540 + }, + { + "epoch": 0.6564939673527325, + "grad_norm": 3.622579574584961, + "learning_rate": 5.6732822599952126e-06, + "loss": 0.3737, + "step": 5550 + }, + { + "epoch": 0.6576768393659806, + "grad_norm": 3.7488350868225098, + "learning_rate": 5.672564041177879e-06, + "loss": 0.347, + "step": 5560 + }, + { + "epoch": 0.6588597113792287, + "grad_norm": 3.859649419784546, + "learning_rate": 5.671845822360546e-06, + "loss": 0.3324, + "step": 5570 + }, + { + "epoch": 0.6600425833924769, + "grad_norm": 2.853318214416504, + "learning_rate": 5.671127603543213e-06, + "loss": 0.3608, + "step": 5580 + }, + { + "epoch": 0.6612254554057251, + "grad_norm": 2.2855758666992188, + "learning_rate": 5.67040938472588e-06, + "loss": 0.3387, + "step": 5590 + }, + { + "epoch": 0.6624083274189733, + "grad_norm": 2.5921871662139893, + "learning_rate": 5.669691165908547e-06, + "loss": 0.3591, + "step": 5600 + }, + { + "epoch": 0.6635911994322214, + "grad_norm": 2.95080304145813, + "learning_rate": 5.668972947091214e-06, + "loss": 0.3631, + "step": 5610 + }, + { + "epoch": 0.6647740714454696, + "grad_norm": 2.3368730545043945, + "learning_rate": 5.668254728273881e-06, + "loss": 0.3485, + "step": 5620 + }, + { + "epoch": 0.6659569434587178, + "grad_norm": 3.513460397720337, + "learning_rate": 5.667536509456548e-06, + "loss": 0.367, + "step": 5630 + }, + { + "epoch": 0.6671398154719659, + "grad_norm": 2.4163262844085693, + "learning_rate": 5.666818290639215e-06, + "loss": 0.3482, + "step": 5640 + }, + { + "epoch": 0.6683226874852141, + "grad_norm": 4.090065956115723, + "learning_rate": 5.666100071821882e-06, + "loss": 0.3526, + "step": 5650 + }, + { + "epoch": 0.6695055594984622, + "grad_norm": 3.317941427230835, + "learning_rate": 5.665381853004549e-06, + "loss": 0.3667, + "step": 5660 + }, + { + "epoch": 0.6706884315117104, + "grad_norm": 2.652280569076538, + "learning_rate": 5.664663634187216e-06, + "loss": 0.3699, + "step": 5670 + }, + { + "epoch": 0.6718713035249586, + "grad_norm": 2.5559115409851074, + "learning_rate": 5.6639454153698825e-06, + "loss": 0.3333, + "step": 5680 + }, + { + "epoch": 0.6730541755382068, + "grad_norm": 2.700054883956909, + "learning_rate": 5.66322719655255e-06, + "loss": 0.3706, + "step": 5690 + }, + { + "epoch": 0.674237047551455, + "grad_norm": 2.8341550827026367, + "learning_rate": 5.662508977735216e-06, + "loss": 0.3469, + "step": 5700 + }, + { + "epoch": 0.6754199195647032, + "grad_norm": 3.4807088375091553, + "learning_rate": 5.661790758917884e-06, + "loss": 0.3222, + "step": 5710 + }, + { + "epoch": 0.6766027915779512, + "grad_norm": 3.6275129318237305, + "learning_rate": 5.66107254010055e-06, + "loss": 0.3334, + "step": 5720 + }, + { + "epoch": 0.6777856635911994, + "grad_norm": 2.0586740970611572, + "learning_rate": 5.660354321283218e-06, + "loss": 0.3323, + "step": 5730 + }, + { + "epoch": 0.6789685356044476, + "grad_norm": 2.88397479057312, + "learning_rate": 5.659636102465884e-06, + "loss": 0.2893, + "step": 5740 + }, + { + "epoch": 0.6801514076176958, + "grad_norm": 3.2640860080718994, + "learning_rate": 5.658917883648552e-06, + "loss": 0.3097, + "step": 5750 + }, + { + "epoch": 0.681334279630944, + "grad_norm": 3.416465997695923, + "learning_rate": 5.658199664831219e-06, + "loss": 0.3649, + "step": 5760 + }, + { + "epoch": 0.6825171516441921, + "grad_norm": 1.7977756261825562, + "learning_rate": 5.657481446013886e-06, + "loss": 0.3342, + "step": 5770 + }, + { + "epoch": 0.6837000236574403, + "grad_norm": 2.6693594455718994, + "learning_rate": 5.656763227196553e-06, + "loss": 0.351, + "step": 5780 + }, + { + "epoch": 0.6848828956706884, + "grad_norm": 2.2618138790130615, + "learning_rate": 5.6560450083792195e-06, + "loss": 0.3289, + "step": 5790 + }, + { + "epoch": 0.6860657676839366, + "grad_norm": 2.292304754257202, + "learning_rate": 5.655326789561887e-06, + "loss": 0.3386, + "step": 5800 + }, + { + "epoch": 0.6872486396971847, + "grad_norm": 2.810405731201172, + "learning_rate": 5.654608570744553e-06, + "loss": 0.3391, + "step": 5810 + }, + { + "epoch": 0.6884315117104329, + "grad_norm": 2.73765230178833, + "learning_rate": 5.653890351927221e-06, + "loss": 0.3935, + "step": 5820 + }, + { + "epoch": 0.6896143837236811, + "grad_norm": 2.1042165756225586, + "learning_rate": 5.653172133109887e-06, + "loss": 0.3523, + "step": 5830 + }, + { + "epoch": 0.6907972557369293, + "grad_norm": 2.7655107975006104, + "learning_rate": 5.652453914292555e-06, + "loss": 0.3465, + "step": 5840 + }, + { + "epoch": 0.6919801277501775, + "grad_norm": 2.2958226203918457, + "learning_rate": 5.651735695475222e-06, + "loss": 0.365, + "step": 5850 + }, + { + "epoch": 0.6931629997634255, + "grad_norm": 2.7822420597076416, + "learning_rate": 5.651017476657889e-06, + "loss": 0.3341, + "step": 5860 + }, + { + "epoch": 0.6943458717766737, + "grad_norm": 3.8763067722320557, + "learning_rate": 5.650299257840556e-06, + "loss": 0.3503, + "step": 5870 + }, + { + "epoch": 0.6955287437899219, + "grad_norm": 4.107594966888428, + "learning_rate": 5.649581039023223e-06, + "loss": 0.3743, + "step": 5880 + }, + { + "epoch": 0.6967116158031701, + "grad_norm": 2.4280354976654053, + "learning_rate": 5.64886282020589e-06, + "loss": 0.3681, + "step": 5890 + }, + { + "epoch": 0.6978944878164183, + "grad_norm": 2.016907215118408, + "learning_rate": 5.6481446013885565e-06, + "loss": 0.3443, + "step": 5900 + }, + { + "epoch": 0.6990773598296665, + "grad_norm": 2.1594183444976807, + "learning_rate": 5.6474263825712235e-06, + "loss": 0.352, + "step": 5910 + }, + { + "epoch": 0.7002602318429146, + "grad_norm": 2.6787216663360596, + "learning_rate": 5.64670816375389e-06, + "loss": 0.3404, + "step": 5920 + }, + { + "epoch": 0.7014431038561627, + "grad_norm": 4.282622814178467, + "learning_rate": 5.645989944936557e-06, + "loss": 0.3229, + "step": 5930 + }, + { + "epoch": 0.7026259758694109, + "grad_norm": 3.5148532390594482, + "learning_rate": 5.645271726119224e-06, + "loss": 0.3727, + "step": 5940 + }, + { + "epoch": 0.7038088478826591, + "grad_norm": 2.5899875164031982, + "learning_rate": 5.644553507301891e-06, + "loss": 0.3066, + "step": 5950 + }, + { + "epoch": 0.7049917198959073, + "grad_norm": 1.9410797357559204, + "learning_rate": 5.643835288484559e-06, + "loss": 0.3376, + "step": 5960 + }, + { + "epoch": 0.7061745919091554, + "grad_norm": 2.4812545776367188, + "learning_rate": 5.643117069667225e-06, + "loss": 0.3315, + "step": 5970 + }, + { + "epoch": 0.7073574639224036, + "grad_norm": 2.838182210922241, + "learning_rate": 5.642398850849893e-06, + "loss": 0.3412, + "step": 5980 + }, + { + "epoch": 0.7085403359356518, + "grad_norm": 3.1450905799865723, + "learning_rate": 5.641680632032559e-06, + "loss": 0.3058, + "step": 5990 + }, + { + "epoch": 0.7097232079489, + "grad_norm": 3.805575132369995, + "learning_rate": 5.640962413215227e-06, + "loss": 0.3116, + "step": 6000 + }, + { + "epoch": 0.710906079962148, + "grad_norm": 2.4876396656036377, + "learning_rate": 5.640244194397893e-06, + "loss": 0.326, + "step": 6010 + }, + { + "epoch": 0.7120889519753962, + "grad_norm": 2.369574546813965, + "learning_rate": 5.6395259755805605e-06, + "loss": 0.335, + "step": 6020 + }, + { + "epoch": 0.7132718239886444, + "grad_norm": 2.075327157974243, + "learning_rate": 5.638807756763227e-06, + "loss": 0.3385, + "step": 6030 + }, + { + "epoch": 0.7144546960018926, + "grad_norm": 1.7152106761932373, + "learning_rate": 5.638089537945894e-06, + "loss": 0.312, + "step": 6040 + }, + { + "epoch": 0.7156375680151408, + "grad_norm": 2.547874689102173, + "learning_rate": 5.637371319128561e-06, + "loss": 0.3252, + "step": 6050 + }, + { + "epoch": 0.716820440028389, + "grad_norm": 3.207555055618286, + "learning_rate": 5.636653100311228e-06, + "loss": 0.3626, + "step": 6060 + }, + { + "epoch": 0.7180033120416371, + "grad_norm": 3.471651315689087, + "learning_rate": 5.635934881493896e-06, + "loss": 0.4018, + "step": 6070 + }, + { + "epoch": 0.7191861840548852, + "grad_norm": 2.3718714714050293, + "learning_rate": 5.635216662676562e-06, + "loss": 0.3157, + "step": 6080 + }, + { + "epoch": 0.7203690560681334, + "grad_norm": 2.1830906867980957, + "learning_rate": 5.63449844385923e-06, + "loss": 0.3814, + "step": 6090 + }, + { + "epoch": 0.7215519280813816, + "grad_norm": 2.37980318069458, + "learning_rate": 5.633780225041896e-06, + "loss": 0.2789, + "step": 6100 + }, + { + "epoch": 0.7227348000946298, + "grad_norm": 3.602928638458252, + "learning_rate": 5.633062006224564e-06, + "loss": 0.3802, + "step": 6110 + }, + { + "epoch": 0.7239176721078779, + "grad_norm": 2.9294888973236084, + "learning_rate": 5.63234378740723e-06, + "loss": 0.3705, + "step": 6120 + }, + { + "epoch": 0.7251005441211261, + "grad_norm": 2.5093209743499756, + "learning_rate": 5.6316255685898975e-06, + "loss": 0.3333, + "step": 6130 + }, + { + "epoch": 0.7262834161343743, + "grad_norm": 1.8491617441177368, + "learning_rate": 5.630907349772564e-06, + "loss": 0.3774, + "step": 6140 + }, + { + "epoch": 0.7274662881476224, + "grad_norm": 3.2130048274993896, + "learning_rate": 5.630189130955231e-06, + "loss": 0.3286, + "step": 6150 + }, + { + "epoch": 0.7286491601608706, + "grad_norm": 2.143611192703247, + "learning_rate": 5.629470912137898e-06, + "loss": 0.3252, + "step": 6160 + }, + { + "epoch": 0.7298320321741187, + "grad_norm": 3.1197423934936523, + "learning_rate": 5.628752693320565e-06, + "loss": 0.3415, + "step": 6170 + }, + { + "epoch": 0.7310149041873669, + "grad_norm": 4.243905067443848, + "learning_rate": 5.628034474503232e-06, + "loss": 0.3856, + "step": 6180 + }, + { + "epoch": 0.7321977762006151, + "grad_norm": 2.5275042057037354, + "learning_rate": 5.627316255685899e-06, + "loss": 0.3272, + "step": 6190 + }, + { + "epoch": 0.7333806482138633, + "grad_norm": 3.6818149089813232, + "learning_rate": 5.626598036868566e-06, + "loss": 0.3799, + "step": 6200 + }, + { + "epoch": 0.7345635202271115, + "grad_norm": 2.4534871578216553, + "learning_rate": 5.625879818051233e-06, + "loss": 0.3836, + "step": 6210 + }, + { + "epoch": 0.7357463922403596, + "grad_norm": 4.3578200340271, + "learning_rate": 5.6251615992339e-06, + "loss": 0.3487, + "step": 6220 + }, + { + "epoch": 0.7369292642536077, + "grad_norm": 3.0013763904571533, + "learning_rate": 5.6244433804165675e-06, + "loss": 0.3369, + "step": 6230 + }, + { + "epoch": 0.7381121362668559, + "grad_norm": 3.0948095321655273, + "learning_rate": 5.623725161599234e-06, + "loss": 0.3457, + "step": 6240 + }, + { + "epoch": 0.7392950082801041, + "grad_norm": 2.9563987255096436, + "learning_rate": 5.623006942781901e-06, + "loss": 0.3261, + "step": 6250 + }, + { + "epoch": 0.7404778802933523, + "grad_norm": 2.799682855606079, + "learning_rate": 5.6222887239645675e-06, + "loss": 0.3734, + "step": 6260 + }, + { + "epoch": 0.7416607523066004, + "grad_norm": 2.5684351921081543, + "learning_rate": 5.621570505147235e-06, + "loss": 0.3747, + "step": 6270 + }, + { + "epoch": 0.7428436243198486, + "grad_norm": 2.4596593379974365, + "learning_rate": 5.620852286329901e-06, + "loss": 0.3313, + "step": 6280 + }, + { + "epoch": 0.7440264963330968, + "grad_norm": 2.4302868843078613, + "learning_rate": 5.620134067512569e-06, + "loss": 0.34, + "step": 6290 + }, + { + "epoch": 0.7452093683463449, + "grad_norm": 4.000497341156006, + "learning_rate": 5.619415848695236e-06, + "loss": 0.3304, + "step": 6300 + }, + { + "epoch": 0.7463922403595931, + "grad_norm": 3.440857172012329, + "learning_rate": 5.618697629877903e-06, + "loss": 0.3036, + "step": 6310 + }, + { + "epoch": 0.7475751123728412, + "grad_norm": 2.37349796295166, + "learning_rate": 5.61797941106057e-06, + "loss": 0.3056, + "step": 6320 + }, + { + "epoch": 0.7487579843860894, + "grad_norm": 3.1543045043945312, + "learning_rate": 5.617261192243237e-06, + "loss": 0.3822, + "step": 6330 + }, + { + "epoch": 0.7499408563993376, + "grad_norm": 2.3295421600341797, + "learning_rate": 5.6165429734259045e-06, + "loss": 0.3229, + "step": 6340 + }, + { + "epoch": 0.7501774308019872, + "eval_accuracy": 0.8502678244668463, + "eval_loss": 0.3409457206726074, + "eval_runtime": 77.6735, + "eval_safe_aucpr": 0.9011741121226409, + "eval_safe_f1": 0.8374008707119244, + "eval_safe_fpr": 0.16489791037637092, + "eval_safe_precision": 0.8077646894821217, + "eval_safe_recall": 0.8692945279975997, + "eval_samples_per_second": 773.932, + "eval_steps_per_second": 12.102, + "eval_unsafe_aucpr": 0.9463641974187076, + "eval_unsafe_f1": 0.861247706987714, + "eval_unsafe_fpr": 0.13070547200239985, + "eval_unsafe_precision": 0.8890833863781031, + "eval_unsafe_recall": 0.8351020896236286, + "step": 6342 + }, + { + "epoch": 0.7511237284125858, + "grad_norm": 2.7695460319519043, + "learning_rate": 5.615824754608571e-06, + "loss": 0.3108, + "step": 6350 + }, + { + "epoch": 0.752306600425834, + "grad_norm": 3.2920620441436768, + "learning_rate": 5.615106535791238e-06, + "loss": 0.3713, + "step": 6360 + }, + { + "epoch": 0.753489472439082, + "grad_norm": 2.614239454269409, + "learning_rate": 5.6143883169739045e-06, + "loss": 0.3636, + "step": 6370 + }, + { + "epoch": 0.7546723444523302, + "grad_norm": 2.2651712894439697, + "learning_rate": 5.613670098156572e-06, + "loss": 0.3549, + "step": 6380 + }, + { + "epoch": 0.7558552164655784, + "grad_norm": 2.1967551708221436, + "learning_rate": 5.612951879339238e-06, + "loss": 0.3479, + "step": 6390 + }, + { + "epoch": 0.7570380884788266, + "grad_norm": 2.7348268032073975, + "learning_rate": 5.612233660521906e-06, + "loss": 0.3288, + "step": 6400 + }, + { + "epoch": 0.7582209604920748, + "grad_norm": 2.1394784450531006, + "learning_rate": 5.611515441704573e-06, + "loss": 0.3259, + "step": 6410 + }, + { + "epoch": 0.759403832505323, + "grad_norm": 2.4483065605163574, + "learning_rate": 5.61079722288724e-06, + "loss": 0.3101, + "step": 6420 + }, + { + "epoch": 0.7605867045185711, + "grad_norm": 2.4361226558685303, + "learning_rate": 5.610079004069907e-06, + "loss": 0.3997, + "step": 6430 + }, + { + "epoch": 0.7617695765318192, + "grad_norm": 2.7097060680389404, + "learning_rate": 5.609360785252574e-06, + "loss": 0.3409, + "step": 6440 + }, + { + "epoch": 0.7629524485450674, + "grad_norm": 2.4710869789123535, + "learning_rate": 5.608642566435241e-06, + "loss": 0.3373, + "step": 6450 + }, + { + "epoch": 0.7641353205583156, + "grad_norm": 2.2179853916168213, + "learning_rate": 5.607924347617908e-06, + "loss": 0.317, + "step": 6460 + }, + { + "epoch": 0.7653181925715637, + "grad_norm": 2.3045737743377686, + "learning_rate": 5.6072061288005745e-06, + "loss": 0.367, + "step": 6470 + }, + { + "epoch": 0.7665010645848119, + "grad_norm": 2.5970547199249268, + "learning_rate": 5.6064879099832415e-06, + "loss": 0.3172, + "step": 6480 + }, + { + "epoch": 0.7676839365980601, + "grad_norm": 2.2353339195251465, + "learning_rate": 5.605769691165908e-06, + "loss": 0.371, + "step": 6490 + }, + { + "epoch": 0.7688668086113083, + "grad_norm": 3.267253875732422, + "learning_rate": 5.605051472348576e-06, + "loss": 0.3598, + "step": 6500 + }, + { + "epoch": 0.7700496806245565, + "grad_norm": 3.300684928894043, + "learning_rate": 5.604333253531242e-06, + "loss": 0.3908, + "step": 6510 + }, + { + "epoch": 0.7712325526378045, + "grad_norm": 2.7900242805480957, + "learning_rate": 5.60361503471391e-06, + "loss": 0.317, + "step": 6520 + }, + { + "epoch": 0.7724154246510527, + "grad_norm": 3.9156737327575684, + "learning_rate": 5.602896815896576e-06, + "loss": 0.3443, + "step": 6530 + }, + { + "epoch": 0.7735982966643009, + "grad_norm": 2.0143091678619385, + "learning_rate": 5.602178597079244e-06, + "loss": 0.3617, + "step": 6540 + }, + { + "epoch": 0.7747811686775491, + "grad_norm": 3.1424200534820557, + "learning_rate": 5.601460378261911e-06, + "loss": 0.3308, + "step": 6550 + }, + { + "epoch": 0.7759640406907973, + "grad_norm": 2.1877145767211914, + "learning_rate": 5.600742159444578e-06, + "loss": 0.346, + "step": 6560 + }, + { + "epoch": 0.7771469127040455, + "grad_norm": 3.5627498626708984, + "learning_rate": 5.600023940627245e-06, + "loss": 0.3614, + "step": 6570 + }, + { + "epoch": 0.7783297847172936, + "grad_norm": 2.8966684341430664, + "learning_rate": 5.5993057218099115e-06, + "loss": 0.3911, + "step": 6580 + }, + { + "epoch": 0.7795126567305417, + "grad_norm": 2.6046037673950195, + "learning_rate": 5.5985875029925785e-06, + "loss": 0.3354, + "step": 6590 + }, + { + "epoch": 0.7806955287437899, + "grad_norm": 4.150275230407715, + "learning_rate": 5.597869284175245e-06, + "loss": 0.3374, + "step": 6600 + }, + { + "epoch": 0.7818784007570381, + "grad_norm": 3.0789692401885986, + "learning_rate": 5.597151065357913e-06, + "loss": 0.3208, + "step": 6610 + }, + { + "epoch": 0.7830612727702863, + "grad_norm": 2.3965251445770264, + "learning_rate": 5.596432846540579e-06, + "loss": 0.3108, + "step": 6620 + }, + { + "epoch": 0.7842441447835344, + "grad_norm": 2.5840611457824707, + "learning_rate": 5.595714627723247e-06, + "loss": 0.3603, + "step": 6630 + }, + { + "epoch": 0.7854270167967826, + "grad_norm": 3.554948329925537, + "learning_rate": 5.594996408905913e-06, + "loss": 0.3952, + "step": 6640 + }, + { + "epoch": 0.7866098888100308, + "grad_norm": 2.754347324371338, + "learning_rate": 5.594278190088581e-06, + "loss": 0.3722, + "step": 6650 + }, + { + "epoch": 0.7877927608232789, + "grad_norm": 2.548170566558838, + "learning_rate": 5.593559971271247e-06, + "loss": 0.3622, + "step": 6660 + }, + { + "epoch": 0.788975632836527, + "grad_norm": 2.657484769821167, + "learning_rate": 5.592841752453915e-06, + "loss": 0.3553, + "step": 6670 + }, + { + "epoch": 0.7901585048497752, + "grad_norm": 3.3273067474365234, + "learning_rate": 5.592123533636582e-06, + "loss": 0.3271, + "step": 6680 + }, + { + "epoch": 0.7913413768630234, + "grad_norm": 3.056093692779541, + "learning_rate": 5.5914053148192485e-06, + "loss": 0.3669, + "step": 6690 + }, + { + "epoch": 0.7925242488762716, + "grad_norm": 2.1021077632904053, + "learning_rate": 5.5906870960019154e-06, + "loss": 0.3486, + "step": 6700 + }, + { + "epoch": 0.7937071208895198, + "grad_norm": 2.397303819656372, + "learning_rate": 5.589968877184582e-06, + "loss": 0.3442, + "step": 6710 + }, + { + "epoch": 0.794889992902768, + "grad_norm": 3.0344865322113037, + "learning_rate": 5.589250658367249e-06, + "loss": 0.3759, + "step": 6720 + }, + { + "epoch": 0.796072864916016, + "grad_norm": 2.3328771591186523, + "learning_rate": 5.588532439549916e-06, + "loss": 0.3658, + "step": 6730 + }, + { + "epoch": 0.7972557369292642, + "grad_norm": 2.2357239723205566, + "learning_rate": 5.587814220732583e-06, + "loss": 0.328, + "step": 6740 + }, + { + "epoch": 0.7984386089425124, + "grad_norm": 2.4008595943450928, + "learning_rate": 5.58709600191525e-06, + "loss": 0.3679, + "step": 6750 + }, + { + "epoch": 0.7996214809557606, + "grad_norm": 2.082193613052368, + "learning_rate": 5.586377783097917e-06, + "loss": 0.3653, + "step": 6760 + }, + { + "epoch": 0.8008043529690088, + "grad_norm": 2.2875049114227295, + "learning_rate": 5.585659564280584e-06, + "loss": 0.3993, + "step": 6770 + }, + { + "epoch": 0.8019872249822569, + "grad_norm": 4.390810012817383, + "learning_rate": 5.584941345463251e-06, + "loss": 0.3251, + "step": 6780 + }, + { + "epoch": 0.8031700969955051, + "grad_norm": 2.708571672439575, + "learning_rate": 5.584223126645919e-06, + "loss": 0.3497, + "step": 6790 + }, + { + "epoch": 0.8043529690087533, + "grad_norm": 3.668095827102661, + "learning_rate": 5.5835049078285855e-06, + "loss": 0.4234, + "step": 6800 + }, + { + "epoch": 0.8055358410220014, + "grad_norm": 2.2173752784729004, + "learning_rate": 5.5827866890112524e-06, + "loss": 0.3288, + "step": 6810 + }, + { + "epoch": 0.8067187130352496, + "grad_norm": 2.5242574214935303, + "learning_rate": 5.582068470193919e-06, + "loss": 0.3371, + "step": 6820 + }, + { + "epoch": 0.8079015850484977, + "grad_norm": 2.4785118103027344, + "learning_rate": 5.581350251376586e-06, + "loss": 0.3406, + "step": 6830 + }, + { + "epoch": 0.8090844570617459, + "grad_norm": 2.005140781402588, + "learning_rate": 5.580632032559253e-06, + "loss": 0.3422, + "step": 6840 + }, + { + "epoch": 0.8102673290749941, + "grad_norm": 4.3739728927612305, + "learning_rate": 5.57991381374192e-06, + "loss": 0.3839, + "step": 6850 + }, + { + "epoch": 0.8114502010882423, + "grad_norm": 2.7078897953033447, + "learning_rate": 5.579195594924587e-06, + "loss": 0.3326, + "step": 6860 + }, + { + "epoch": 0.8126330731014905, + "grad_norm": 2.3173391819000244, + "learning_rate": 5.578477376107254e-06, + "loss": 0.3626, + "step": 6870 + }, + { + "epoch": 0.8138159451147385, + "grad_norm": 2.316404342651367, + "learning_rate": 5.577759157289922e-06, + "loss": 0.3743, + "step": 6880 + }, + { + "epoch": 0.8149988171279867, + "grad_norm": 2.4765071868896484, + "learning_rate": 5.577040938472588e-06, + "loss": 0.3304, + "step": 6890 + }, + { + "epoch": 0.8161816891412349, + "grad_norm": 2.8449864387512207, + "learning_rate": 5.576322719655256e-06, + "loss": 0.3603, + "step": 6900 + }, + { + "epoch": 0.8173645611544831, + "grad_norm": 2.9987564086914062, + "learning_rate": 5.575604500837922e-06, + "loss": 0.348, + "step": 6910 + }, + { + "epoch": 0.8185474331677313, + "grad_norm": 2.6204235553741455, + "learning_rate": 5.5748862820205894e-06, + "loss": 0.3072, + "step": 6920 + }, + { + "epoch": 0.8197303051809794, + "grad_norm": 2.3653435707092285, + "learning_rate": 5.5741680632032555e-06, + "loss": 0.3398, + "step": 6930 + }, + { + "epoch": 0.8209131771942276, + "grad_norm": 3.261428117752075, + "learning_rate": 5.573449844385923e-06, + "loss": 0.3434, + "step": 6940 + }, + { + "epoch": 0.8220960492074757, + "grad_norm": 4.365106582641602, + "learning_rate": 5.57273162556859e-06, + "loss": 0.3566, + "step": 6950 + }, + { + "epoch": 0.8232789212207239, + "grad_norm": 3.3341214656829834, + "learning_rate": 5.572013406751257e-06, + "loss": 0.3665, + "step": 6960 + }, + { + "epoch": 0.8244617932339721, + "grad_norm": 2.0983595848083496, + "learning_rate": 5.571295187933924e-06, + "loss": 0.3646, + "step": 6970 + }, + { + "epoch": 0.8256446652472202, + "grad_norm": 3.0781376361846924, + "learning_rate": 5.570576969116591e-06, + "loss": 0.3294, + "step": 6980 + }, + { + "epoch": 0.8268275372604684, + "grad_norm": 2.7174408435821533, + "learning_rate": 5.569858750299258e-06, + "loss": 0.3606, + "step": 6990 + }, + { + "epoch": 0.8280104092737166, + "grad_norm": 2.8317363262176514, + "learning_rate": 5.569140531481925e-06, + "loss": 0.3627, + "step": 7000 + }, + { + "epoch": 0.8291932812869648, + "grad_norm": 2.5212972164154053, + "learning_rate": 5.568422312664592e-06, + "loss": 0.3358, + "step": 7010 + }, + { + "epoch": 0.830376153300213, + "grad_norm": 2.977804660797119, + "learning_rate": 5.567704093847259e-06, + "loss": 0.3542, + "step": 7020 + }, + { + "epoch": 0.831559025313461, + "grad_norm": 2.880445957183838, + "learning_rate": 5.566985875029926e-06, + "loss": 0.3508, + "step": 7030 + }, + { + "epoch": 0.8327418973267092, + "grad_norm": 3.8674228191375732, + "learning_rate": 5.5662676562125925e-06, + "loss": 0.3337, + "step": 7040 + }, + { + "epoch": 0.8339247693399574, + "grad_norm": 3.2273359298706055, + "learning_rate": 5.5655494373952594e-06, + "loss": 0.3638, + "step": 7050 + }, + { + "epoch": 0.8351076413532056, + "grad_norm": 2.6208560466766357, + "learning_rate": 5.564831218577927e-06, + "loss": 0.3764, + "step": 7060 + }, + { + "epoch": 0.8362905133664538, + "grad_norm": 4.46844482421875, + "learning_rate": 5.564112999760594e-06, + "loss": 0.3223, + "step": 7070 + }, + { + "epoch": 0.837473385379702, + "grad_norm": 2.8332033157348633, + "learning_rate": 5.563394780943261e-06, + "loss": 0.3439, + "step": 7080 + }, + { + "epoch": 0.8386562573929501, + "grad_norm": 1.84604811668396, + "learning_rate": 5.562676562125928e-06, + "loss": 0.3599, + "step": 7090 + }, + { + "epoch": 0.8398391294061982, + "grad_norm": 3.16796875, + "learning_rate": 5.561958343308595e-06, + "loss": 0.3226, + "step": 7100 + }, + { + "epoch": 0.8410220014194464, + "grad_norm": 2.880558729171753, + "learning_rate": 5.561240124491262e-06, + "loss": 0.3564, + "step": 7110 + }, + { + "epoch": 0.8422048734326946, + "grad_norm": 3.119356632232666, + "learning_rate": 5.560521905673929e-06, + "loss": 0.3507, + "step": 7120 + }, + { + "epoch": 0.8433877454459427, + "grad_norm": 2.4311368465423584, + "learning_rate": 5.559803686856596e-06, + "loss": 0.3507, + "step": 7130 + }, + { + "epoch": 0.8445706174591909, + "grad_norm": 2.9251863956451416, + "learning_rate": 5.559085468039263e-06, + "loss": 0.3467, + "step": 7140 + }, + { + "epoch": 0.8457534894724391, + "grad_norm": 3.1497440338134766, + "learning_rate": 5.55836724922193e-06, + "loss": 0.3782, + "step": 7150 + }, + { + "epoch": 0.8469363614856873, + "grad_norm": 2.303025960922241, + "learning_rate": 5.5576490304045964e-06, + "loss": 0.3522, + "step": 7160 + }, + { + "epoch": 0.8481192334989354, + "grad_norm": 2.60343074798584, + "learning_rate": 5.556930811587264e-06, + "loss": 0.3398, + "step": 7170 + }, + { + "epoch": 0.8493021055121835, + "grad_norm": 2.61942982673645, + "learning_rate": 5.55621259276993e-06, + "loss": 0.3504, + "step": 7180 + }, + { + "epoch": 0.8504849775254317, + "grad_norm": 3.136608600616455, + "learning_rate": 5.555494373952598e-06, + "loss": 0.3493, + "step": 7190 + }, + { + "epoch": 0.8516678495386799, + "grad_norm": 2.6096303462982178, + "learning_rate": 5.554776155135264e-06, + "loss": 0.3217, + "step": 7200 + }, + { + "epoch": 0.8528507215519281, + "grad_norm": 3.7467339038848877, + "learning_rate": 5.554057936317932e-06, + "loss": 0.3334, + "step": 7210 + }, + { + "epoch": 0.8540335935651763, + "grad_norm": 2.4341225624084473, + "learning_rate": 5.553339717500599e-06, + "loss": 0.3638, + "step": 7220 + }, + { + "epoch": 0.8552164655784245, + "grad_norm": 2.3212246894836426, + "learning_rate": 5.552621498683266e-06, + "loss": 0.3515, + "step": 7230 + }, + { + "epoch": 0.8563993375916725, + "grad_norm": 2.513843059539795, + "learning_rate": 5.551903279865933e-06, + "loss": 0.3225, + "step": 7240 + }, + { + "epoch": 0.8575822096049207, + "grad_norm": 2.3628575801849365, + "learning_rate": 5.5511850610486e-06, + "loss": 0.3138, + "step": 7250 + }, + { + "epoch": 0.8587650816181689, + "grad_norm": 3.140106678009033, + "learning_rate": 5.5504668422312665e-06, + "loss": 0.3321, + "step": 7260 + }, + { + "epoch": 0.8599479536314171, + "grad_norm": 2.6459336280822754, + "learning_rate": 5.5497486234139334e-06, + "loss": 0.3633, + "step": 7270 + }, + { + "epoch": 0.8611308256446653, + "grad_norm": 2.208141326904297, + "learning_rate": 5.5490304045966e-06, + "loss": 0.3335, + "step": 7280 + }, + { + "epoch": 0.8623136976579134, + "grad_norm": 2.6104226112365723, + "learning_rate": 5.548312185779267e-06, + "loss": 0.352, + "step": 7290 + }, + { + "epoch": 0.8634965696711616, + "grad_norm": 2.234896183013916, + "learning_rate": 5.547593966961934e-06, + "loss": 0.3889, + "step": 7300 + }, + { + "epoch": 0.8646794416844098, + "grad_norm": 2.363894462585449, + "learning_rate": 5.546875748144601e-06, + "loss": 0.3238, + "step": 7310 + }, + { + "epoch": 0.8658623136976579, + "grad_norm": 2.8927001953125, + "learning_rate": 5.546157529327269e-06, + "loss": 0.357, + "step": 7320 + }, + { + "epoch": 0.867045185710906, + "grad_norm": 3.3064918518066406, + "learning_rate": 5.545439310509936e-06, + "loss": 0.3878, + "step": 7330 + }, + { + "epoch": 0.8682280577241542, + "grad_norm": 3.236659526824951, + "learning_rate": 5.544721091692603e-06, + "loss": 0.3851, + "step": 7340 + }, + { + "epoch": 0.8694109297374024, + "grad_norm": 2.66019344329834, + "learning_rate": 5.54400287287527e-06, + "loss": 0.3599, + "step": 7350 + }, + { + "epoch": 0.8705938017506506, + "grad_norm": 1.7286221981048584, + "learning_rate": 5.543284654057937e-06, + "loss": 0.3258, + "step": 7360 + }, + { + "epoch": 0.8717766737638988, + "grad_norm": 2.7896177768707275, + "learning_rate": 5.5425664352406035e-06, + "loss": 0.3575, + "step": 7370 + }, + { + "epoch": 0.872959545777147, + "grad_norm": 3.1597604751586914, + "learning_rate": 5.5418482164232704e-06, + "loss": 0.3931, + "step": 7380 + }, + { + "epoch": 0.874142417790395, + "grad_norm": 1.8662405014038086, + "learning_rate": 5.541129997605937e-06, + "loss": 0.3439, + "step": 7390 + }, + { + "epoch": 0.8753252898036432, + "grad_norm": 2.0772244930267334, + "learning_rate": 5.540411778788604e-06, + "loss": 0.3351, + "step": 7400 + }, + { + "epoch": 0.8765081618168914, + "grad_norm": 2.4114174842834473, + "learning_rate": 5.539693559971271e-06, + "loss": 0.3593, + "step": 7410 + }, + { + "epoch": 0.8776910338301396, + "grad_norm": 3.092982053756714, + "learning_rate": 5.538975341153939e-06, + "loss": 0.3258, + "step": 7420 + }, + { + "epoch": 0.8788739058433878, + "grad_norm": 1.9947749376296997, + "learning_rate": 5.538257122336605e-06, + "loss": 0.3298, + "step": 7430 + }, + { + "epoch": 0.8800567778566359, + "grad_norm": 4.120667457580566, + "learning_rate": 5.537538903519273e-06, + "loss": 0.3927, + "step": 7440 + }, + { + "epoch": 0.8812396498698841, + "grad_norm": 2.6464829444885254, + "learning_rate": 5.536820684701939e-06, + "loss": 0.3306, + "step": 7450 + }, + { + "epoch": 0.8824225218831322, + "grad_norm": 2.8128652572631836, + "learning_rate": 5.536102465884607e-06, + "loss": 0.3556, + "step": 7460 + }, + { + "epoch": 0.8836053938963804, + "grad_norm": 2.0512959957122803, + "learning_rate": 5.535384247067273e-06, + "loss": 0.3366, + "step": 7470 + }, + { + "epoch": 0.8847882659096286, + "grad_norm": 2.7022056579589844, + "learning_rate": 5.5346660282499405e-06, + "loss": 0.3759, + "step": 7480 + }, + { + "epoch": 0.8859711379228767, + "grad_norm": 1.867337942123413, + "learning_rate": 5.5339478094326074e-06, + "loss": 0.3218, + "step": 7490 + }, + { + "epoch": 0.8871540099361249, + "grad_norm": 3.7488763332366943, + "learning_rate": 5.533229590615274e-06, + "loss": 0.3551, + "step": 7500 + }, + { + "epoch": 0.8883368819493731, + "grad_norm": 4.215018272399902, + "learning_rate": 5.532511371797941e-06, + "loss": 0.3316, + "step": 7510 + }, + { + "epoch": 0.8895197539626213, + "grad_norm": 2.2023723125457764, + "learning_rate": 5.531793152980608e-06, + "loss": 0.3306, + "step": 7520 + }, + { + "epoch": 0.8907026259758695, + "grad_norm": 2.64302659034729, + "learning_rate": 5.531074934163275e-06, + "loss": 0.3486, + "step": 7530 + }, + { + "epoch": 0.8918854979891175, + "grad_norm": 3.5505239963531494, + "learning_rate": 5.530356715345942e-06, + "loss": 0.3361, + "step": 7540 + }, + { + "epoch": 0.8930683700023657, + "grad_norm": 2.461738348007202, + "learning_rate": 5.529638496528609e-06, + "loss": 0.3531, + "step": 7550 + }, + { + "epoch": 0.8942512420156139, + "grad_norm": 2.906794548034668, + "learning_rate": 5.528920277711276e-06, + "loss": 0.3305, + "step": 7560 + }, + { + "epoch": 0.8954341140288621, + "grad_norm": 3.0541648864746094, + "learning_rate": 5.528202058893944e-06, + "loss": 0.3634, + "step": 7570 + }, + { + "epoch": 0.8966169860421103, + "grad_norm": 2.518399715423584, + "learning_rate": 5.52748384007661e-06, + "loss": 0.3543, + "step": 7580 + }, + { + "epoch": 0.8977998580553584, + "grad_norm": 3.4840755462646484, + "learning_rate": 5.5267656212592775e-06, + "loss": 0.366, + "step": 7590 + }, + { + "epoch": 0.8989827300686066, + "grad_norm": 2.303093194961548, + "learning_rate": 5.5260474024419444e-06, + "loss": 0.3602, + "step": 7600 + }, + { + "epoch": 0.9001656020818547, + "grad_norm": 3.2364866733551025, + "learning_rate": 5.525329183624611e-06, + "loss": 0.3562, + "step": 7610 + }, + { + "epoch": 0.9013484740951029, + "grad_norm": 2.9462339878082275, + "learning_rate": 5.524610964807278e-06, + "loss": 0.3609, + "step": 7620 + }, + { + "epoch": 0.9025313461083511, + "grad_norm": 2.3655786514282227, + "learning_rate": 5.523892745989945e-06, + "loss": 0.3436, + "step": 7630 + }, + { + "epoch": 0.9037142181215992, + "grad_norm": 2.2203845977783203, + "learning_rate": 5.523174527172612e-06, + "loss": 0.3728, + "step": 7640 + }, + { + "epoch": 0.9048970901348474, + "grad_norm": 2.290207862854004, + "learning_rate": 5.522456308355279e-06, + "loss": 0.3182, + "step": 7650 + }, + { + "epoch": 0.9060799621480956, + "grad_norm": 3.579007863998413, + "learning_rate": 5.521738089537946e-06, + "loss": 0.3684, + "step": 7660 + }, + { + "epoch": 0.9072628341613438, + "grad_norm": 2.4544854164123535, + "learning_rate": 5.521019870720613e-06, + "loss": 0.2881, + "step": 7670 + }, + { + "epoch": 0.9084457061745919, + "grad_norm": 2.399073839187622, + "learning_rate": 5.52030165190328e-06, + "loss": 0.3262, + "step": 7680 + }, + { + "epoch": 0.90962857818784, + "grad_norm": 3.2292399406433105, + "learning_rate": 5.519583433085947e-06, + "loss": 0.3417, + "step": 7690 + }, + { + "epoch": 0.9108114502010882, + "grad_norm": 2.4320437908172607, + "learning_rate": 5.518865214268614e-06, + "loss": 0.3597, + "step": 7700 + }, + { + "epoch": 0.9119943222143364, + "grad_norm": 3.3153836727142334, + "learning_rate": 5.5181469954512814e-06, + "loss": 0.331, + "step": 7710 + }, + { + "epoch": 0.9131771942275846, + "grad_norm": 3.12115478515625, + "learning_rate": 5.5174287766339475e-06, + "loss": 0.3033, + "step": 7720 + }, + { + "epoch": 0.9143600662408328, + "grad_norm": 1.7770037651062012, + "learning_rate": 5.516710557816615e-06, + "loss": 0.3451, + "step": 7730 + }, + { + "epoch": 0.915542938254081, + "grad_norm": 2.761713743209839, + "learning_rate": 5.515992338999281e-06, + "loss": 0.3295, + "step": 7740 + }, + { + "epoch": 0.916725810267329, + "grad_norm": 3.4372050762176514, + "learning_rate": 5.515274120181949e-06, + "loss": 0.3115, + "step": 7750 + }, + { + "epoch": 0.9179086822805772, + "grad_norm": 2.6007134914398193, + "learning_rate": 5.514555901364616e-06, + "loss": 0.3574, + "step": 7760 + }, + { + "epoch": 0.9190915542938254, + "grad_norm": 1.9798741340637207, + "learning_rate": 5.513837682547283e-06, + "loss": 0.355, + "step": 7770 + }, + { + "epoch": 0.9202744263070736, + "grad_norm": 2.710803270339966, + "learning_rate": 5.51311946372995e-06, + "loss": 0.3608, + "step": 7780 + }, + { + "epoch": 0.9214572983203217, + "grad_norm": 2.823716163635254, + "learning_rate": 5.512401244912617e-06, + "loss": 0.3676, + "step": 7790 + }, + { + "epoch": 0.9226401703335699, + "grad_norm": 2.5856072902679443, + "learning_rate": 5.511683026095284e-06, + "loss": 0.3661, + "step": 7800 + }, + { + "epoch": 0.9238230423468181, + "grad_norm": 2.7194437980651855, + "learning_rate": 5.510964807277951e-06, + "loss": 0.3507, + "step": 7810 + }, + { + "epoch": 0.9250059143600663, + "grad_norm": 2.136730909347534, + "learning_rate": 5.5102465884606184e-06, + "loss": 0.3346, + "step": 7820 + }, + { + "epoch": 0.9261887863733144, + "grad_norm": 3.40328049659729, + "learning_rate": 5.5095283696432845e-06, + "loss": 0.3275, + "step": 7830 + }, + { + "epoch": 0.9273716583865625, + "grad_norm": 5.959146022796631, + "learning_rate": 5.508810150825952e-06, + "loss": 0.3761, + "step": 7840 + }, + { + "epoch": 0.9285545303998107, + "grad_norm": 2.9853756427764893, + "learning_rate": 5.508091932008618e-06, + "loss": 0.3588, + "step": 7850 + }, + { + "epoch": 0.9297374024130589, + "grad_norm": 2.2519445419311523, + "learning_rate": 5.507373713191286e-06, + "loss": 0.3017, + "step": 7860 + }, + { + "epoch": 0.9309202744263071, + "grad_norm": 3.8326361179351807, + "learning_rate": 5.506655494373953e-06, + "loss": 0.3472, + "step": 7870 + }, + { + "epoch": 0.9321031464395553, + "grad_norm": 3.3324077129364014, + "learning_rate": 5.50593727555662e-06, + "loss": 0.3803, + "step": 7880 + }, + { + "epoch": 0.9332860184528035, + "grad_norm": 2.6756067276000977, + "learning_rate": 5.505219056739287e-06, + "loss": 0.3185, + "step": 7890 + }, + { + "epoch": 0.9344688904660515, + "grad_norm": 2.8437869548797607, + "learning_rate": 5.504500837921954e-06, + "loss": 0.3579, + "step": 7900 + }, + { + "epoch": 0.9356517624792997, + "grad_norm": 3.054748773574829, + "learning_rate": 5.503782619104621e-06, + "loss": 0.3564, + "step": 7910 + }, + { + "epoch": 0.9368346344925479, + "grad_norm": 2.167482852935791, + "learning_rate": 5.503064400287288e-06, + "loss": 0.3494, + "step": 7920 + }, + { + "epoch": 0.9380175065057961, + "grad_norm": 2.3468101024627686, + "learning_rate": 5.5023461814699546e-06, + "loss": 0.3285, + "step": 7930 + }, + { + "epoch": 0.9392003785190443, + "grad_norm": 2.8190131187438965, + "learning_rate": 5.5016279626526215e-06, + "loss": 0.2858, + "step": 7940 + }, + { + "epoch": 0.9403832505322924, + "grad_norm": 2.783245325088501, + "learning_rate": 5.500909743835288e-06, + "loss": 0.3725, + "step": 7950 + }, + { + "epoch": 0.9415661225455406, + "grad_norm": 2.941138505935669, + "learning_rate": 5.500191525017955e-06, + "loss": 0.3345, + "step": 7960 + }, + { + "epoch": 0.9427489945587887, + "grad_norm": 2.27093505859375, + "learning_rate": 5.499473306200622e-06, + "loss": 0.3384, + "step": 7970 + }, + { + "epoch": 0.9439318665720369, + "grad_norm": 2.350109338760376, + "learning_rate": 5.49875508738329e-06, + "loss": 0.381, + "step": 7980 + }, + { + "epoch": 0.945114738585285, + "grad_norm": 2.4800150394439697, + "learning_rate": 5.498036868565956e-06, + "loss": 0.3243, + "step": 7990 + }, + { + "epoch": 0.9462976105985332, + "grad_norm": 2.5707695484161377, + "learning_rate": 5.497318649748624e-06, + "loss": 0.3237, + "step": 8000 + }, + { + "epoch": 0.9474804826117814, + "grad_norm": 5.119116306304932, + "learning_rate": 5.49660043093129e-06, + "loss": 0.3766, + "step": 8010 + }, + { + "epoch": 0.9486633546250296, + "grad_norm": 2.0871899127960205, + "learning_rate": 5.495882212113958e-06, + "loss": 0.3399, + "step": 8020 + }, + { + "epoch": 0.9498462266382778, + "grad_norm": 2.1982295513153076, + "learning_rate": 5.495163993296624e-06, + "loss": 0.3437, + "step": 8030 + }, + { + "epoch": 0.9510290986515259, + "grad_norm": 2.7046656608581543, + "learning_rate": 5.4944457744792916e-06, + "loss": 0.3225, + "step": 8040 + }, + { + "epoch": 0.952211970664774, + "grad_norm": 2.497591495513916, + "learning_rate": 5.4937275556619585e-06, + "loss": 0.369, + "step": 8050 + }, + { + "epoch": 0.9533948426780222, + "grad_norm": 2.4972617626190186, + "learning_rate": 5.493009336844625e-06, + "loss": 0.3459, + "step": 8060 + }, + { + "epoch": 0.9545777146912704, + "grad_norm": 2.939979314804077, + "learning_rate": 5.492291118027293e-06, + "loss": 0.3471, + "step": 8070 + }, + { + "epoch": 0.9557605867045186, + "grad_norm": 2.3696694374084473, + "learning_rate": 5.491572899209959e-06, + "loss": 0.3112, + "step": 8080 + }, + { + "epoch": 0.9569434587177668, + "grad_norm": 2.358957290649414, + "learning_rate": 5.490854680392627e-06, + "loss": 0.329, + "step": 8090 + }, + { + "epoch": 0.9581263307310149, + "grad_norm": 2.4226739406585693, + "learning_rate": 5.490136461575293e-06, + "loss": 0.3186, + "step": 8100 + }, + { + "epoch": 0.9593092027442631, + "grad_norm": 2.8412845134735107, + "learning_rate": 5.489418242757961e-06, + "loss": 0.3168, + "step": 8110 + }, + { + "epoch": 0.9604920747575112, + "grad_norm": 2.624074697494507, + "learning_rate": 5.488700023940627e-06, + "loss": 0.3293, + "step": 8120 + }, + { + "epoch": 0.9616749467707594, + "grad_norm": 3.3990612030029297, + "learning_rate": 5.487981805123295e-06, + "loss": 0.3583, + "step": 8130 + }, + { + "epoch": 0.9628578187840076, + "grad_norm": 2.37146258354187, + "learning_rate": 5.487263586305962e-06, + "loss": 0.3207, + "step": 8140 + }, + { + "epoch": 0.9640406907972557, + "grad_norm": 2.7736618518829346, + "learning_rate": 5.4865453674886286e-06, + "loss": 0.3393, + "step": 8150 + }, + { + "epoch": 0.9652235628105039, + "grad_norm": 2.7632699012756348, + "learning_rate": 5.4858271486712955e-06, + "loss": 0.3341, + "step": 8160 + }, + { + "epoch": 0.9664064348237521, + "grad_norm": 2.1342551708221436, + "learning_rate": 5.485108929853962e-06, + "loss": 0.377, + "step": 8170 + }, + { + "epoch": 0.9675893068370003, + "grad_norm": 2.6788430213928223, + "learning_rate": 5.484390711036629e-06, + "loss": 0.3033, + "step": 8180 + }, + { + "epoch": 0.9687721788502484, + "grad_norm": 2.574594020843506, + "learning_rate": 5.483672492219296e-06, + "loss": 0.3315, + "step": 8190 + }, + { + "epoch": 0.9699550508634965, + "grad_norm": 2.56632399559021, + "learning_rate": 5.482954273401963e-06, + "loss": 0.3317, + "step": 8200 + }, + { + "epoch": 0.9711379228767447, + "grad_norm": 4.020124435424805, + "learning_rate": 5.48223605458463e-06, + "loss": 0.3644, + "step": 8210 + }, + { + "epoch": 0.9723207948899929, + "grad_norm": 2.2321345806121826, + "learning_rate": 5.481517835767297e-06, + "loss": 0.3244, + "step": 8220 + }, + { + "epoch": 0.9735036669032411, + "grad_norm": 2.39266037940979, + "learning_rate": 5.480799616949964e-06, + "loss": 0.3584, + "step": 8230 + }, + { + "epoch": 0.9746865389164893, + "grad_norm": 2.82624888420105, + "learning_rate": 5.480081398132631e-06, + "loss": 0.335, + "step": 8240 + }, + { + "epoch": 0.9758694109297374, + "grad_norm": 2.847728729248047, + "learning_rate": 5.479363179315299e-06, + "loss": 0.3797, + "step": 8250 + }, + { + "epoch": 0.9770522829429855, + "grad_norm": 2.8681814670562744, + "learning_rate": 5.478644960497965e-06, + "loss": 0.3158, + "step": 8260 + }, + { + "epoch": 0.9782351549562337, + "grad_norm": 2.9551432132720947, + "learning_rate": 5.4779267416806325e-06, + "loss": 0.3126, + "step": 8270 + }, + { + "epoch": 0.9794180269694819, + "grad_norm": 3.640418529510498, + "learning_rate": 5.4772085228632986e-06, + "loss": 0.3419, + "step": 8280 + }, + { + "epoch": 0.9806008989827301, + "grad_norm": 3.0809133052825928, + "learning_rate": 5.476490304045966e-06, + "loss": 0.3261, + "step": 8290 + }, + { + "epoch": 0.9817837709959782, + "grad_norm": 2.4587645530700684, + "learning_rate": 5.475772085228632e-06, + "loss": 0.3296, + "step": 8300 + }, + { + "epoch": 0.9829666430092264, + "grad_norm": 2.6227495670318604, + "learning_rate": 5.4750538664113e-06, + "loss": 0.3633, + "step": 8310 + }, + { + "epoch": 0.9841495150224746, + "grad_norm": 2.2411069869995117, + "learning_rate": 5.474335647593967e-06, + "loss": 0.3587, + "step": 8320 + }, + { + "epoch": 0.9853323870357228, + "grad_norm": 2.9606146812438965, + "learning_rate": 5.473617428776634e-06, + "loss": 0.3344, + "step": 8330 + }, + { + "epoch": 0.9865152590489709, + "grad_norm": 2.1389524936676025, + "learning_rate": 5.472899209959302e-06, + "loss": 0.3239, + "step": 8340 + }, + { + "epoch": 0.987698131062219, + "grad_norm": 2.7911806106567383, + "learning_rate": 5.472180991141968e-06, + "loss": 0.3391, + "step": 8350 + }, + { + "epoch": 0.9888810030754672, + "grad_norm": 2.3505120277404785, + "learning_rate": 5.471462772324636e-06, + "loss": 0.3235, + "step": 8360 + }, + { + "epoch": 0.9900638750887154, + "grad_norm": 2.090029001235962, + "learning_rate": 5.470744553507302e-06, + "loss": 0.3564, + "step": 8370 + }, + { + "epoch": 0.9912467471019636, + "grad_norm": 3.6726338863372803, + "learning_rate": 5.4700263346899695e-06, + "loss": 0.3693, + "step": 8380 + }, + { + "epoch": 0.9924296191152118, + "grad_norm": 3.1619338989257812, + "learning_rate": 5.4693081158726356e-06, + "loss": 0.3692, + "step": 8390 + }, + { + "epoch": 0.99361249112846, + "grad_norm": 2.480262517929077, + "learning_rate": 5.468589897055303e-06, + "loss": 0.3565, + "step": 8400 + }, + { + "epoch": 0.994795363141708, + "grad_norm": 4.622408866882324, + "learning_rate": 5.46787167823797e-06, + "loss": 0.358, + "step": 8410 + }, + { + "epoch": 0.9959782351549562, + "grad_norm": 2.087460994720459, + "learning_rate": 5.467153459420637e-06, + "loss": 0.3632, + "step": 8420 + }, + { + "epoch": 0.9971611071682044, + "grad_norm": 2.0399322509765625, + "learning_rate": 5.466435240603304e-06, + "loss": 0.3306, + "step": 8430 + }, + { + "epoch": 0.9983439791814526, + "grad_norm": 2.897136926651001, + "learning_rate": 5.465717021785971e-06, + "loss": 0.3553, + "step": 8440 + }, + { + "epoch": 0.9995268511947007, + "grad_norm": 3.2266910076141357, + "learning_rate": 5.464998802968638e-06, + "loss": 0.3597, + "step": 8450 + }, + { + "epoch": 1.0002365744026496, + "eval_accuracy": 0.8519812356522607, + "eval_loss": 0.33436116576194763, + "eval_runtime": 79.1167, + "eval_safe_aucpr": 0.9059820300762905, + "eval_safe_f1": 0.8338313289011727, + "eval_safe_fpr": 0.1363187946548679, + "eval_safe_precision": 0.8303886925795053, + "eval_safe_recall": 0.8373026291115029, + "eval_samples_per_second": 759.815, + "eval_steps_per_second": 11.881, + "eval_unsafe_aucpr": 0.9481555247484019, + "eval_unsafe_f1": 0.8665566886622675, + "eval_unsafe_fpr": 0.16269737088849656, + "eval_unsafe_precision": 0.8694513828282524, + "eval_unsafe_recall": 0.8636812053451317, + "step": 8456 + }, + { + "epoch": 1.000709723207949, + "grad_norm": 3.032815456390381, + "learning_rate": 5.464280584151305e-06, + "loss": 0.3003, + "step": 8460 + }, + { + "epoch": 1.001892595221197, + "grad_norm": 3.270686149597168, + "learning_rate": 5.463562365333972e-06, + "loss": 0.3169, + "step": 8470 + }, + { + "epoch": 1.0030754672344453, + "grad_norm": 3.1649651527404785, + "learning_rate": 5.462844146516639e-06, + "loss": 0.3107, + "step": 8480 + }, + { + "epoch": 1.0042583392476934, + "grad_norm": 4.533438205718994, + "learning_rate": 5.462125927699306e-06, + "loss": 0.3696, + "step": 8490 + }, + { + "epoch": 1.0054412112609417, + "grad_norm": 2.518247604370117, + "learning_rate": 5.4614077088819726e-06, + "loss": 0.3226, + "step": 8500 + }, + { + "epoch": 1.0066240832741897, + "grad_norm": 2.138230562210083, + "learning_rate": 5.4606894900646395e-06, + "loss": 0.3334, + "step": 8510 + }, + { + "epoch": 1.0078069552874378, + "grad_norm": 2.659940719604492, + "learning_rate": 5.459971271247307e-06, + "loss": 0.3236, + "step": 8520 + }, + { + "epoch": 1.008989827300686, + "grad_norm": 2.521111011505127, + "learning_rate": 5.459253052429973e-06, + "loss": 0.3356, + "step": 8530 + }, + { + "epoch": 1.0101726993139342, + "grad_norm": 2.4494261741638184, + "learning_rate": 5.458534833612641e-06, + "loss": 0.3562, + "step": 8540 + }, + { + "epoch": 1.0113555713271825, + "grad_norm": 2.304314374923706, + "learning_rate": 5.457816614795307e-06, + "loss": 0.3342, + "step": 8550 + }, + { + "epoch": 1.0125384433404305, + "grad_norm": 2.936882734298706, + "learning_rate": 5.457098395977975e-06, + "loss": 0.3332, + "step": 8560 + }, + { + "epoch": 1.0137213153536788, + "grad_norm": 2.1796059608459473, + "learning_rate": 5.456380177160641e-06, + "loss": 0.2911, + "step": 8570 + }, + { + "epoch": 1.014904187366927, + "grad_norm": 3.1883201599121094, + "learning_rate": 5.455661958343309e-06, + "loss": 0.342, + "step": 8580 + }, + { + "epoch": 1.016087059380175, + "grad_norm": 2.9558184146881104, + "learning_rate": 5.454943739525976e-06, + "loss": 0.3225, + "step": 8590 + }, + { + "epoch": 1.0172699313934233, + "grad_norm": 2.6654281616210938, + "learning_rate": 5.454225520708643e-06, + "loss": 0.2589, + "step": 8600 + }, + { + "epoch": 1.0184528034066713, + "grad_norm": 3.352184534072876, + "learning_rate": 5.4535073018913096e-06, + "loss": 0.3118, + "step": 8610 + }, + { + "epoch": 1.0196356754199196, + "grad_norm": 2.928210496902466, + "learning_rate": 5.4527890830739765e-06, + "loss": 0.3515, + "step": 8620 + }, + { + "epoch": 1.0208185474331677, + "grad_norm": 2.8811118602752686, + "learning_rate": 5.452070864256644e-06, + "loss": 0.3125, + "step": 8630 + }, + { + "epoch": 1.022001419446416, + "grad_norm": 2.919084072113037, + "learning_rate": 5.45135264543931e-06, + "loss": 0.3126, + "step": 8640 + }, + { + "epoch": 1.023184291459664, + "grad_norm": 2.5222673416137695, + "learning_rate": 5.450634426621978e-06, + "loss": 0.336, + "step": 8650 + }, + { + "epoch": 1.0243671634729121, + "grad_norm": 3.217428684234619, + "learning_rate": 5.449916207804644e-06, + "loss": 0.3497, + "step": 8660 + }, + { + "epoch": 1.0255500354861604, + "grad_norm": 2.234304189682007, + "learning_rate": 5.449197988987312e-06, + "loss": 0.3029, + "step": 8670 + }, + { + "epoch": 1.0267329074994085, + "grad_norm": 2.950589418411255, + "learning_rate": 5.448479770169979e-06, + "loss": 0.3024, + "step": 8680 + }, + { + "epoch": 1.0279157795126568, + "grad_norm": 3.1446995735168457, + "learning_rate": 5.447761551352646e-06, + "loss": 0.2645, + "step": 8690 + }, + { + "epoch": 1.0290986515259049, + "grad_norm": 4.240466117858887, + "learning_rate": 5.447043332535313e-06, + "loss": 0.3296, + "step": 8700 + }, + { + "epoch": 1.0302815235391531, + "grad_norm": 2.519169330596924, + "learning_rate": 5.44632511371798e-06, + "loss": 0.2848, + "step": 8710 + }, + { + "epoch": 1.0314643955524012, + "grad_norm": 2.583394765853882, + "learning_rate": 5.4456068949006466e-06, + "loss": 0.3419, + "step": 8720 + }, + { + "epoch": 1.0326472675656495, + "grad_norm": 3.5348169803619385, + "learning_rate": 5.4448886760833135e-06, + "loss": 0.3245, + "step": 8730 + }, + { + "epoch": 1.0338301395788976, + "grad_norm": 2.2338781356811523, + "learning_rate": 5.44417045726598e-06, + "loss": 0.3585, + "step": 8740 + }, + { + "epoch": 1.0350130115921456, + "grad_norm": 2.400956630706787, + "learning_rate": 5.443452238448647e-06, + "loss": 0.3284, + "step": 8750 + }, + { + "epoch": 1.036195883605394, + "grad_norm": 2.292003870010376, + "learning_rate": 5.442734019631314e-06, + "loss": 0.3532, + "step": 8760 + }, + { + "epoch": 1.037378755618642, + "grad_norm": 2.004927158355713, + "learning_rate": 5.442015800813981e-06, + "loss": 0.3454, + "step": 8770 + }, + { + "epoch": 1.0385616276318903, + "grad_norm": 2.7153475284576416, + "learning_rate": 5.441297581996648e-06, + "loss": 0.3152, + "step": 8780 + }, + { + "epoch": 1.0397444996451384, + "grad_norm": 2.201547861099243, + "learning_rate": 5.440579363179316e-06, + "loss": 0.3092, + "step": 8790 + }, + { + "epoch": 1.0409273716583867, + "grad_norm": 2.537216901779175, + "learning_rate": 5.439861144361982e-06, + "loss": 0.323, + "step": 8800 + }, + { + "epoch": 1.0421102436716347, + "grad_norm": 1.951098918914795, + "learning_rate": 5.43914292554465e-06, + "loss": 0.3439, + "step": 8810 + }, + { + "epoch": 1.0432931156848828, + "grad_norm": 2.515214443206787, + "learning_rate": 5.438424706727316e-06, + "loss": 0.3453, + "step": 8820 + }, + { + "epoch": 1.044475987698131, + "grad_norm": 2.52101731300354, + "learning_rate": 5.4377064879099836e-06, + "loss": 0.2679, + "step": 8830 + }, + { + "epoch": 1.0456588597113792, + "grad_norm": 2.983762502670288, + "learning_rate": 5.4369882690926505e-06, + "loss": 0.3119, + "step": 8840 + }, + { + "epoch": 1.0468417317246275, + "grad_norm": 3.678133726119995, + "learning_rate": 5.436270050275317e-06, + "loss": 0.3321, + "step": 8850 + }, + { + "epoch": 1.0480246037378755, + "grad_norm": 2.3628365993499756, + "learning_rate": 5.435551831457984e-06, + "loss": 0.3077, + "step": 8860 + }, + { + "epoch": 1.0492074757511238, + "grad_norm": 2.0662803649902344, + "learning_rate": 5.434833612640651e-06, + "loss": 0.3505, + "step": 8870 + }, + { + "epoch": 1.050390347764372, + "grad_norm": 3.0162265300750732, + "learning_rate": 5.434115393823318e-06, + "loss": 0.3485, + "step": 8880 + }, + { + "epoch": 1.05157321977762, + "grad_norm": 2.200587034225464, + "learning_rate": 5.433397175005985e-06, + "loss": 0.327, + "step": 8890 + }, + { + "epoch": 1.0527560917908683, + "grad_norm": 2.201749324798584, + "learning_rate": 5.432678956188653e-06, + "loss": 0.3306, + "step": 8900 + }, + { + "epoch": 1.0539389638041163, + "grad_norm": 3.2741239070892334, + "learning_rate": 5.431960737371319e-06, + "loss": 0.3378, + "step": 8910 + }, + { + "epoch": 1.0551218358173646, + "grad_norm": 2.7297585010528564, + "learning_rate": 5.431242518553987e-06, + "loss": 0.3071, + "step": 8920 + }, + { + "epoch": 1.0563047078306127, + "grad_norm": 2.7042174339294434, + "learning_rate": 5.430524299736653e-06, + "loss": 0.3246, + "step": 8930 + }, + { + "epoch": 1.057487579843861, + "grad_norm": 2.48810076713562, + "learning_rate": 5.4298060809193206e-06, + "loss": 0.3364, + "step": 8940 + }, + { + "epoch": 1.058670451857109, + "grad_norm": 3.0956532955169678, + "learning_rate": 5.429087862101987e-06, + "loss": 0.2946, + "step": 8950 + }, + { + "epoch": 1.0598533238703571, + "grad_norm": 2.7961843013763428, + "learning_rate": 5.428369643284654e-06, + "loss": 0.4025, + "step": 8960 + }, + { + "epoch": 1.0610361958836054, + "grad_norm": 2.458209753036499, + "learning_rate": 5.427651424467321e-06, + "loss": 0.3359, + "step": 8970 + }, + { + "epoch": 1.0622190678968535, + "grad_norm": 2.2479140758514404, + "learning_rate": 5.426933205649988e-06, + "loss": 0.3544, + "step": 8980 + }, + { + "epoch": 1.0634019399101018, + "grad_norm": 2.5327320098876953, + "learning_rate": 5.426214986832655e-06, + "loss": 0.2918, + "step": 8990 + }, + { + "epoch": 1.0645848119233499, + "grad_norm": 2.4770946502685547, + "learning_rate": 5.425496768015322e-06, + "loss": 0.337, + "step": 9000 + }, + { + "epoch": 1.0657676839365982, + "grad_norm": 2.397608995437622, + "learning_rate": 5.424778549197989e-06, + "loss": 0.3459, + "step": 9010 + }, + { + "epoch": 1.0669505559498462, + "grad_norm": 3.4776039123535156, + "learning_rate": 5.424060330380656e-06, + "loss": 0.3269, + "step": 9020 + }, + { + "epoch": 1.0681334279630943, + "grad_norm": 2.9039711952209473, + "learning_rate": 5.423342111563323e-06, + "loss": 0.3209, + "step": 9030 + }, + { + "epoch": 1.0693162999763426, + "grad_norm": 2.191887140274048, + "learning_rate": 5.42262389274599e-06, + "loss": 0.3018, + "step": 9040 + }, + { + "epoch": 1.0704991719895907, + "grad_norm": 2.682880163192749, + "learning_rate": 5.421905673928657e-06, + "loss": 0.3186, + "step": 9050 + }, + { + "epoch": 1.071682044002839, + "grad_norm": 3.5908334255218506, + "learning_rate": 5.4211874551113245e-06, + "loss": 0.3186, + "step": 9060 + }, + { + "epoch": 1.072864916016087, + "grad_norm": 2.4704625606536865, + "learning_rate": 5.4204692362939906e-06, + "loss": 0.3082, + "step": 9070 + }, + { + "epoch": 1.0740477880293353, + "grad_norm": 2.317227840423584, + "learning_rate": 5.419751017476658e-06, + "loss": 0.3457, + "step": 9080 + }, + { + "epoch": 1.0752306600425834, + "grad_norm": 2.4439492225646973, + "learning_rate": 5.419032798659325e-06, + "loss": 0.3069, + "step": 9090 + }, + { + "epoch": 1.0764135320558315, + "grad_norm": 4.290738582611084, + "learning_rate": 5.418314579841992e-06, + "loss": 0.3373, + "step": 9100 + }, + { + "epoch": 1.0775964040690797, + "grad_norm": 3.5892770290374756, + "learning_rate": 5.417596361024659e-06, + "loss": 0.3549, + "step": 9110 + }, + { + "epoch": 1.0787792760823278, + "grad_norm": 2.733102798461914, + "learning_rate": 5.416878142207326e-06, + "loss": 0.3389, + "step": 9120 + }, + { + "epoch": 1.0799621480955761, + "grad_norm": 3.0099611282348633, + "learning_rate": 5.416159923389993e-06, + "loss": 0.3304, + "step": 9130 + }, + { + "epoch": 1.0811450201088242, + "grad_norm": 3.5245816707611084, + "learning_rate": 5.41544170457266e-06, + "loss": 0.3359, + "step": 9140 + }, + { + "epoch": 1.0823278921220725, + "grad_norm": 3.3787643909454346, + "learning_rate": 5.414723485755327e-06, + "loss": 0.2982, + "step": 9150 + }, + { + "epoch": 1.0835107641353205, + "grad_norm": 2.499584197998047, + "learning_rate": 5.414005266937994e-06, + "loss": 0.3199, + "step": 9160 + }, + { + "epoch": 1.0846936361485686, + "grad_norm": 2.0115272998809814, + "learning_rate": 5.4132870481206615e-06, + "loss": 0.3543, + "step": 9170 + }, + { + "epoch": 1.085876508161817, + "grad_norm": 2.129696846008301, + "learning_rate": 5.4125688293033275e-06, + "loss": 0.3323, + "step": 9180 + }, + { + "epoch": 1.087059380175065, + "grad_norm": 2.24391508102417, + "learning_rate": 5.411850610485995e-06, + "loss": 0.3399, + "step": 9190 + }, + { + "epoch": 1.0882422521883133, + "grad_norm": 3.175449848175049, + "learning_rate": 5.411132391668661e-06, + "loss": 0.3249, + "step": 9200 + }, + { + "epoch": 1.0894251242015613, + "grad_norm": 2.4642605781555176, + "learning_rate": 5.410414172851329e-06, + "loss": 0.3292, + "step": 9210 + }, + { + "epoch": 1.0906079962148096, + "grad_norm": 2.4592621326446533, + "learning_rate": 5.409695954033995e-06, + "loss": 0.3159, + "step": 9220 + }, + { + "epoch": 1.0917908682280577, + "grad_norm": 2.652513027191162, + "learning_rate": 5.408977735216663e-06, + "loss": 0.3197, + "step": 9230 + }, + { + "epoch": 1.0929737402413058, + "grad_norm": 3.861598253250122, + "learning_rate": 5.40825951639933e-06, + "loss": 0.3261, + "step": 9240 + }, + { + "epoch": 1.094156612254554, + "grad_norm": 3.9381461143493652, + "learning_rate": 5.407541297581997e-06, + "loss": 0.328, + "step": 9250 + }, + { + "epoch": 1.0953394842678021, + "grad_norm": 2.2220299243927, + "learning_rate": 5.406823078764664e-06, + "loss": 0.3508, + "step": 9260 + }, + { + "epoch": 1.0965223562810504, + "grad_norm": 2.5837607383728027, + "learning_rate": 5.406104859947331e-06, + "loss": 0.3117, + "step": 9270 + }, + { + "epoch": 1.0977052282942985, + "grad_norm": 2.157978057861328, + "learning_rate": 5.405386641129998e-06, + "loss": 0.3027, + "step": 9280 + }, + { + "epoch": 1.0988881003075468, + "grad_norm": 2.818922281265259, + "learning_rate": 5.4046684223126645e-06, + "loss": 0.32, + "step": 9290 + }, + { + "epoch": 1.1000709723207949, + "grad_norm": 3.150458812713623, + "learning_rate": 5.4039502034953315e-06, + "loss": 0.3058, + "step": 9300 + }, + { + "epoch": 1.101253844334043, + "grad_norm": 2.7295992374420166, + "learning_rate": 5.403231984677998e-06, + "loss": 0.3101, + "step": 9310 + }, + { + "epoch": 1.1024367163472912, + "grad_norm": 3.992326259613037, + "learning_rate": 5.402513765860665e-06, + "loss": 0.3732, + "step": 9320 + }, + { + "epoch": 1.1036195883605393, + "grad_norm": 2.365307331085205, + "learning_rate": 5.401795547043333e-06, + "loss": 0.332, + "step": 9330 + }, + { + "epoch": 1.1048024603737876, + "grad_norm": 1.9618815183639526, + "learning_rate": 5.401077328226e-06, + "loss": 0.3279, + "step": 9340 + }, + { + "epoch": 1.1059853323870357, + "grad_norm": 2.1657872200012207, + "learning_rate": 5.400359109408667e-06, + "loss": 0.3234, + "step": 9350 + }, + { + "epoch": 1.107168204400284, + "grad_norm": 2.3764379024505615, + "learning_rate": 5.399640890591334e-06, + "loss": 0.3213, + "step": 9360 + }, + { + "epoch": 1.108351076413532, + "grad_norm": 2.755964994430542, + "learning_rate": 5.398922671774001e-06, + "loss": 0.3053, + "step": 9370 + }, + { + "epoch": 1.1095339484267803, + "grad_norm": 2.9501616954803467, + "learning_rate": 5.398204452956668e-06, + "loss": 0.3158, + "step": 9380 + }, + { + "epoch": 1.1107168204400284, + "grad_norm": 2.6937944889068604, + "learning_rate": 5.397486234139335e-06, + "loss": 0.347, + "step": 9390 + }, + { + "epoch": 1.1118996924532765, + "grad_norm": 3.0070104598999023, + "learning_rate": 5.3967680153220015e-06, + "loss": 0.3224, + "step": 9400 + }, + { + "epoch": 1.1130825644665248, + "grad_norm": 2.28945255279541, + "learning_rate": 5.3960497965046685e-06, + "loss": 0.3066, + "step": 9410 + }, + { + "epoch": 1.1142654364797728, + "grad_norm": 3.1918983459472656, + "learning_rate": 5.395331577687335e-06, + "loss": 0.3332, + "step": 9420 + }, + { + "epoch": 1.1154483084930211, + "grad_norm": 1.994618535041809, + "learning_rate": 5.394613358870002e-06, + "loss": 0.3331, + "step": 9430 + }, + { + "epoch": 1.1166311805062692, + "grad_norm": 3.330002546310425, + "learning_rate": 5.39389514005267e-06, + "loss": 0.2817, + "step": 9440 + }, + { + "epoch": 1.1178140525195175, + "grad_norm": 3.884190320968628, + "learning_rate": 5.393176921235336e-06, + "loss": 0.3136, + "step": 9450 + }, + { + "epoch": 1.1189969245327656, + "grad_norm": 3.1539881229400635, + "learning_rate": 5.392458702418004e-06, + "loss": 0.3238, + "step": 9460 + }, + { + "epoch": 1.1201797965460136, + "grad_norm": 2.9703457355499268, + "learning_rate": 5.39174048360067e-06, + "loss": 0.3234, + "step": 9470 + }, + { + "epoch": 1.121362668559262, + "grad_norm": 1.9814268350601196, + "learning_rate": 5.391022264783338e-06, + "loss": 0.3297, + "step": 9480 + }, + { + "epoch": 1.12254554057251, + "grad_norm": 3.241884231567383, + "learning_rate": 5.390304045966004e-06, + "loss": 0.3699, + "step": 9490 + }, + { + "epoch": 1.1237284125857583, + "grad_norm": 2.133580446243286, + "learning_rate": 5.389585827148672e-06, + "loss": 0.2978, + "step": 9500 + }, + { + "epoch": 1.1249112845990064, + "grad_norm": 2.891195774078369, + "learning_rate": 5.3888676083313385e-06, + "loss": 0.3122, + "step": 9510 + }, + { + "epoch": 1.1260941566122546, + "grad_norm": 3.0090155601501465, + "learning_rate": 5.3881493895140055e-06, + "loss": 0.3151, + "step": 9520 + }, + { + "epoch": 1.1272770286255027, + "grad_norm": 3.4183151721954346, + "learning_rate": 5.387431170696672e-06, + "loss": 0.3306, + "step": 9530 + }, + { + "epoch": 1.1284599006387508, + "grad_norm": 2.636528968811035, + "learning_rate": 5.386712951879339e-06, + "loss": 0.3244, + "step": 9540 + }, + { + "epoch": 1.129642772651999, + "grad_norm": 2.98867130279541, + "learning_rate": 5.385994733062006e-06, + "loss": 0.3545, + "step": 9550 + }, + { + "epoch": 1.1308256446652472, + "grad_norm": 3.331969976425171, + "learning_rate": 5.385276514244673e-06, + "loss": 0.3422, + "step": 9560 + }, + { + "epoch": 1.1320085166784954, + "grad_norm": 3.2384042739868164, + "learning_rate": 5.38455829542734e-06, + "loss": 0.3648, + "step": 9570 + }, + { + "epoch": 1.1331913886917435, + "grad_norm": 2.898747205734253, + "learning_rate": 5.383840076610007e-06, + "loss": 0.2721, + "step": 9580 + }, + { + "epoch": 1.1343742607049918, + "grad_norm": 3.3465516567230225, + "learning_rate": 5.383121857792675e-06, + "loss": 0.3728, + "step": 9590 + }, + { + "epoch": 1.1355571327182399, + "grad_norm": 2.604771137237549, + "learning_rate": 5.382403638975341e-06, + "loss": 0.3116, + "step": 9600 + }, + { + "epoch": 1.1367400047314882, + "grad_norm": 4.8374786376953125, + "learning_rate": 5.381685420158009e-06, + "loss": 0.3038, + "step": 9610 + }, + { + "epoch": 1.1379228767447362, + "grad_norm": 3.862927198410034, + "learning_rate": 5.3809672013406755e-06, + "loss": 0.3708, + "step": 9620 + }, + { + "epoch": 1.1391057487579843, + "grad_norm": 2.758249282836914, + "learning_rate": 5.3802489825233425e-06, + "loss": 0.3599, + "step": 9630 + }, + { + "epoch": 1.1402886207712326, + "grad_norm": 2.8053455352783203, + "learning_rate": 5.379530763706009e-06, + "loss": 0.3608, + "step": 9640 + }, + { + "epoch": 1.1414714927844807, + "grad_norm": 2.310523509979248, + "learning_rate": 5.378812544888676e-06, + "loss": 0.3451, + "step": 9650 + }, + { + "epoch": 1.142654364797729, + "grad_norm": 1.741196632385254, + "learning_rate": 5.378094326071343e-06, + "loss": 0.387, + "step": 9660 + }, + { + "epoch": 1.143837236810977, + "grad_norm": 3.097195863723755, + "learning_rate": 5.37737610725401e-06, + "loss": 0.3098, + "step": 9670 + }, + { + "epoch": 1.1450201088242253, + "grad_norm": 4.80789041519165, + "learning_rate": 5.376657888436677e-06, + "loss": 0.3382, + "step": 9680 + }, + { + "epoch": 1.1462029808374734, + "grad_norm": 2.501476526260376, + "learning_rate": 5.375939669619344e-06, + "loss": 0.3259, + "step": 9690 + }, + { + "epoch": 1.1473858528507215, + "grad_norm": 3.66853404045105, + "learning_rate": 5.375221450802011e-06, + "loss": 0.334, + "step": 9700 + }, + { + "epoch": 1.1485687248639698, + "grad_norm": 1.9440921545028687, + "learning_rate": 5.374503231984679e-06, + "loss": 0.383, + "step": 9710 + }, + { + "epoch": 1.1497515968772178, + "grad_norm": 2.009800434112549, + "learning_rate": 5.373785013167345e-06, + "loss": 0.2965, + "step": 9720 + }, + { + "epoch": 1.1509344688904661, + "grad_norm": 3.147747755050659, + "learning_rate": 5.3730667943500125e-06, + "loss": 0.3293, + "step": 9730 + }, + { + "epoch": 1.1521173409037142, + "grad_norm": 2.8316309452056885, + "learning_rate": 5.372348575532679e-06, + "loss": 0.2886, + "step": 9740 + }, + { + "epoch": 1.1533002129169625, + "grad_norm": 2.775974988937378, + "learning_rate": 5.371630356715346e-06, + "loss": 0.3287, + "step": 9750 + }, + { + "epoch": 1.1544830849302106, + "grad_norm": 3.234229803085327, + "learning_rate": 5.3709121378980125e-06, + "loss": 0.3466, + "step": 9760 + }, + { + "epoch": 1.1556659569434586, + "grad_norm": 2.839085102081299, + "learning_rate": 5.37019391908068e-06, + "loss": 0.2998, + "step": 9770 + }, + { + "epoch": 1.156848828956707, + "grad_norm": 2.5250282287597656, + "learning_rate": 5.369475700263347e-06, + "loss": 0.2971, + "step": 9780 + }, + { + "epoch": 1.158031700969955, + "grad_norm": 2.6362595558166504, + "learning_rate": 5.368757481446014e-06, + "loss": 0.3456, + "step": 9790 + }, + { + "epoch": 1.1592145729832033, + "grad_norm": 2.554025650024414, + "learning_rate": 5.368039262628681e-06, + "loss": 0.3319, + "step": 9800 + }, + { + "epoch": 1.1603974449964514, + "grad_norm": 3.0926215648651123, + "learning_rate": 5.367321043811348e-06, + "loss": 0.3508, + "step": 9810 + }, + { + "epoch": 1.1615803170096997, + "grad_norm": 3.5291664600372314, + "learning_rate": 5.366602824994015e-06, + "loss": 0.3262, + "step": 9820 + }, + { + "epoch": 1.1627631890229477, + "grad_norm": 2.089956283569336, + "learning_rate": 5.365884606176682e-06, + "loss": 0.3305, + "step": 9830 + }, + { + "epoch": 1.1639460610361958, + "grad_norm": 2.995192289352417, + "learning_rate": 5.3651663873593495e-06, + "loss": 0.3556, + "step": 9840 + }, + { + "epoch": 1.165128933049444, + "grad_norm": 2.551785707473755, + "learning_rate": 5.364448168542016e-06, + "loss": 0.3554, + "step": 9850 + }, + { + "epoch": 1.1663118050626922, + "grad_norm": 2.8216392993927, + "learning_rate": 5.363729949724683e-06, + "loss": 0.3189, + "step": 9860 + }, + { + "epoch": 1.1674946770759405, + "grad_norm": 2.8255176544189453, + "learning_rate": 5.3630117309073495e-06, + "loss": 0.3238, + "step": 9870 + }, + { + "epoch": 1.1686775490891885, + "grad_norm": 2.8966610431671143, + "learning_rate": 5.362293512090017e-06, + "loss": 0.3263, + "step": 9880 + }, + { + "epoch": 1.1698604211024368, + "grad_norm": 2.827005386352539, + "learning_rate": 5.361575293272684e-06, + "loss": 0.3291, + "step": 9890 + }, + { + "epoch": 1.171043293115685, + "grad_norm": 3.3333427906036377, + "learning_rate": 5.360857074455351e-06, + "loss": 0.3327, + "step": 9900 + }, + { + "epoch": 1.172226165128933, + "grad_norm": 6.229929447174072, + "learning_rate": 5.360138855638018e-06, + "loss": 0.2978, + "step": 9910 + }, + { + "epoch": 1.1734090371421813, + "grad_norm": 3.740514039993286, + "learning_rate": 5.359420636820685e-06, + "loss": 0.3287, + "step": 9920 + }, + { + "epoch": 1.1745919091554293, + "grad_norm": 3.0174272060394287, + "learning_rate": 5.358702418003352e-06, + "loss": 0.3065, + "step": 9930 + }, + { + "epoch": 1.1757747811686776, + "grad_norm": 2.7568047046661377, + "learning_rate": 5.357984199186019e-06, + "loss": 0.3337, + "step": 9940 + }, + { + "epoch": 1.1769576531819257, + "grad_norm": 2.7569351196289062, + "learning_rate": 5.357265980368686e-06, + "loss": 0.2926, + "step": 9950 + }, + { + "epoch": 1.178140525195174, + "grad_norm": 3.5561347007751465, + "learning_rate": 5.356547761551353e-06, + "loss": 0.323, + "step": 9960 + }, + { + "epoch": 1.179323397208422, + "grad_norm": 3.757917642593384, + "learning_rate": 5.3558295427340195e-06, + "loss": 0.329, + "step": 9970 + }, + { + "epoch": 1.1805062692216701, + "grad_norm": 2.378509044647217, + "learning_rate": 5.355111323916687e-06, + "loss": 0.3016, + "step": 9980 + }, + { + "epoch": 1.1816891412349184, + "grad_norm": 3.223623752593994, + "learning_rate": 5.354393105099353e-06, + "loss": 0.2803, + "step": 9990 + }, + { + "epoch": 1.1828720132481665, + "grad_norm": 3.2154698371887207, + "learning_rate": 5.353674886282021e-06, + "loss": 0.3432, + "step": 10000 + }, + { + "epoch": 1.1840548852614148, + "grad_norm": 3.9279282093048096, + "learning_rate": 5.352956667464687e-06, + "loss": 0.3255, + "step": 10010 + }, + { + "epoch": 1.1852377572746629, + "grad_norm": 2.274242401123047, + "learning_rate": 5.352238448647355e-06, + "loss": 0.2996, + "step": 10020 + }, + { + "epoch": 1.1864206292879111, + "grad_norm": 2.1344165802001953, + "learning_rate": 5.351520229830021e-06, + "loss": 0.3347, + "step": 10030 + }, + { + "epoch": 1.1876035013011592, + "grad_norm": 2.5024712085723877, + "learning_rate": 5.350802011012689e-06, + "loss": 0.3257, + "step": 10040 + }, + { + "epoch": 1.1887863733144073, + "grad_norm": 3.5476930141448975, + "learning_rate": 5.350083792195356e-06, + "loss": 0.3462, + "step": 10050 + }, + { + "epoch": 1.1899692453276556, + "grad_norm": 2.743263006210327, + "learning_rate": 5.349365573378023e-06, + "loss": 0.3142, + "step": 10060 + }, + { + "epoch": 1.1911521173409036, + "grad_norm": 2.665689468383789, + "learning_rate": 5.34864735456069e-06, + "loss": 0.3255, + "step": 10070 + }, + { + "epoch": 1.192334989354152, + "grad_norm": 2.821195363998413, + "learning_rate": 5.3479291357433565e-06, + "loss": 0.3485, + "step": 10080 + }, + { + "epoch": 1.1935178613674, + "grad_norm": 3.750598669052124, + "learning_rate": 5.347210916926024e-06, + "loss": 0.3565, + "step": 10090 + }, + { + "epoch": 1.1947007333806483, + "grad_norm": 3.2968881130218506, + "learning_rate": 5.34649269810869e-06, + "loss": 0.3316, + "step": 10100 + }, + { + "epoch": 1.1958836053938964, + "grad_norm": 2.7333712577819824, + "learning_rate": 5.345774479291358e-06, + "loss": 0.3552, + "step": 10110 + }, + { + "epoch": 1.1970664774071444, + "grad_norm": 3.273253917694092, + "learning_rate": 5.345056260474024e-06, + "loss": 0.3052, + "step": 10120 + }, + { + "epoch": 1.1982493494203927, + "grad_norm": 3.5871641635894775, + "learning_rate": 5.344338041656692e-06, + "loss": 0.3539, + "step": 10130 + }, + { + "epoch": 1.1994322214336408, + "grad_norm": 2.6012492179870605, + "learning_rate": 5.343619822839358e-06, + "loss": 0.3048, + "step": 10140 + }, + { + "epoch": 1.200615093446889, + "grad_norm": 2.8074467182159424, + "learning_rate": 5.342901604022026e-06, + "loss": 0.3208, + "step": 10150 + }, + { + "epoch": 1.2017979654601372, + "grad_norm": 2.587714433670044, + "learning_rate": 5.342183385204693e-06, + "loss": 0.364, + "step": 10160 + }, + { + "epoch": 1.2029808374733855, + "grad_norm": 2.8514552116394043, + "learning_rate": 5.34146516638736e-06, + "loss": 0.374, + "step": 10170 + }, + { + "epoch": 1.2041637094866335, + "grad_norm": 2.119905471801758, + "learning_rate": 5.340746947570027e-06, + "loss": 0.3108, + "step": 10180 + }, + { + "epoch": 1.2053465814998816, + "grad_norm": 2.2122278213500977, + "learning_rate": 5.3400287287526935e-06, + "loss": 0.3663, + "step": 10190 + }, + { + "epoch": 1.20652945351313, + "grad_norm": 2.449744462966919, + "learning_rate": 5.3393105099353604e-06, + "loss": 0.316, + "step": 10200 + }, + { + "epoch": 1.207712325526378, + "grad_norm": 2.322458028793335, + "learning_rate": 5.338592291118027e-06, + "loss": 0.3294, + "step": 10210 + }, + { + "epoch": 1.2088951975396263, + "grad_norm": 3.058540105819702, + "learning_rate": 5.337874072300694e-06, + "loss": 0.3217, + "step": 10220 + }, + { + "epoch": 1.2100780695528743, + "grad_norm": 2.831397771835327, + "learning_rate": 5.337155853483361e-06, + "loss": 0.3483, + "step": 10230 + }, + { + "epoch": 1.2112609415661226, + "grad_norm": 3.7924118041992188, + "learning_rate": 5.336437634666028e-06, + "loss": 0.3268, + "step": 10240 + }, + { + "epoch": 1.2124438135793707, + "grad_norm": 2.430237293243408, + "learning_rate": 5.335719415848696e-06, + "loss": 0.3177, + "step": 10250 + }, + { + "epoch": 1.2136266855926188, + "grad_norm": 2.777331829071045, + "learning_rate": 5.335001197031362e-06, + "loss": 0.3226, + "step": 10260 + }, + { + "epoch": 1.214809557605867, + "grad_norm": 4.125950336456299, + "learning_rate": 5.33428297821403e-06, + "loss": 0.3387, + "step": 10270 + }, + { + "epoch": 1.2159924296191151, + "grad_norm": 4.0093865394592285, + "learning_rate": 5.333564759396696e-06, + "loss": 0.3114, + "step": 10280 + }, + { + "epoch": 1.2171753016323634, + "grad_norm": 2.7753725051879883, + "learning_rate": 5.332846540579364e-06, + "loss": 0.3477, + "step": 10290 + }, + { + "epoch": 1.2183581736456115, + "grad_norm": 2.7971582412719727, + "learning_rate": 5.33212832176203e-06, + "loss": 0.2958, + "step": 10300 + }, + { + "epoch": 1.2195410456588598, + "grad_norm": 4.457556247711182, + "learning_rate": 5.3314101029446974e-06, + "loss": 0.3467, + "step": 10310 + }, + { + "epoch": 1.2207239176721079, + "grad_norm": 2.207031726837158, + "learning_rate": 5.330691884127364e-06, + "loss": 0.3495, + "step": 10320 + }, + { + "epoch": 1.221906789685356, + "grad_norm": 2.6760246753692627, + "learning_rate": 5.329973665310031e-06, + "loss": 0.2982, + "step": 10330 + }, + { + "epoch": 1.2230896616986042, + "grad_norm": 2.5255050659179688, + "learning_rate": 5.329255446492698e-06, + "loss": 0.3041, + "step": 10340 + }, + { + "epoch": 1.2242725337118523, + "grad_norm": 3.199742555618286, + "learning_rate": 5.328537227675365e-06, + "loss": 0.3206, + "step": 10350 + }, + { + "epoch": 1.2254554057251006, + "grad_norm": 2.4137117862701416, + "learning_rate": 5.327819008858033e-06, + "loss": 0.3294, + "step": 10360 + }, + { + "epoch": 1.2266382777383487, + "grad_norm": 2.261509895324707, + "learning_rate": 5.327100790040699e-06, + "loss": 0.3637, + "step": 10370 + }, + { + "epoch": 1.227821149751597, + "grad_norm": 2.8423104286193848, + "learning_rate": 5.326382571223367e-06, + "loss": 0.3163, + "step": 10380 + }, + { + "epoch": 1.229004021764845, + "grad_norm": 2.7373669147491455, + "learning_rate": 5.325664352406033e-06, + "loss": 0.3, + "step": 10390 + }, + { + "epoch": 1.230186893778093, + "grad_norm": 2.387768268585205, + "learning_rate": 5.324946133588701e-06, + "loss": 0.2863, + "step": 10400 + }, + { + "epoch": 1.2313697657913414, + "grad_norm": 4.27761697769165, + "learning_rate": 5.324227914771367e-06, + "loss": 0.3159, + "step": 10410 + }, + { + "epoch": 1.2325526378045895, + "grad_norm": 3.5470752716064453, + "learning_rate": 5.3235096959540344e-06, + "loss": 0.3164, + "step": 10420 + }, + { + "epoch": 1.2337355098178377, + "grad_norm": 3.484423875808716, + "learning_rate": 5.322791477136701e-06, + "loss": 0.3537, + "step": 10430 + }, + { + "epoch": 1.2349183818310858, + "grad_norm": 2.320530414581299, + "learning_rate": 5.322073258319368e-06, + "loss": 0.3151, + "step": 10440 + }, + { + "epoch": 1.2361012538443341, + "grad_norm": 2.4881606101989746, + "learning_rate": 5.321355039502035e-06, + "loss": 0.343, + "step": 10450 + }, + { + "epoch": 1.2372841258575822, + "grad_norm": 2.5587258338928223, + "learning_rate": 5.320636820684702e-06, + "loss": 0.3331, + "step": 10460 + }, + { + "epoch": 1.2384669978708303, + "grad_norm": 3.2363245487213135, + "learning_rate": 5.319918601867369e-06, + "loss": 0.3409, + "step": 10470 + }, + { + "epoch": 1.2396498698840785, + "grad_norm": 3.1542718410491943, + "learning_rate": 5.319200383050036e-06, + "loss": 0.336, + "step": 10480 + }, + { + "epoch": 1.2408327418973266, + "grad_norm": 2.3025355339050293, + "learning_rate": 5.318482164232703e-06, + "loss": 0.3123, + "step": 10490 + }, + { + "epoch": 1.242015613910575, + "grad_norm": 2.1847636699676514, + "learning_rate": 5.31776394541537e-06, + "loss": 0.3192, + "step": 10500 + }, + { + "epoch": 1.243198485923823, + "grad_norm": 3.5734481811523438, + "learning_rate": 5.317045726598037e-06, + "loss": 0.3639, + "step": 10510 + }, + { + "epoch": 1.2443813579370713, + "grad_norm": 3.351245880126953, + "learning_rate": 5.316327507780704e-06, + "loss": 0.3508, + "step": 10520 + }, + { + "epoch": 1.2455642299503193, + "grad_norm": 2.6809604167938232, + "learning_rate": 5.315609288963371e-06, + "loss": 0.3249, + "step": 10530 + }, + { + "epoch": 1.2467471019635676, + "grad_norm": 2.790921688079834, + "learning_rate": 5.314891070146038e-06, + "loss": 0.3244, + "step": 10540 + }, + { + "epoch": 1.2479299739768157, + "grad_norm": 3.341811418533325, + "learning_rate": 5.3141728513287044e-06, + "loss": 0.3594, + "step": 10550 + }, + { + "epoch": 1.2491128459900638, + "grad_norm": 2.7346391677856445, + "learning_rate": 5.313454632511372e-06, + "loss": 0.3232, + "step": 10560 + }, + { + "epoch": 1.250295718003312, + "grad_norm": 3.2251312732696533, + "learning_rate": 5.312736413694038e-06, + "loss": 0.3295, + "step": 10570 + }, + { + "epoch": 1.250295718003312, + "eval_accuracy": 0.8541105233389893, + "eval_loss": 0.3356674313545227, + "eval_runtime": 82.1894, + "eval_safe_aucpr": 0.9073804423875799, + "eval_safe_f1": 0.8369646044021416, + "eval_safe_fpr": 0.1380526740605658, + "eval_safe_precision": 0.8297762541929301, + "eval_safe_recall": 0.8442785883058921, + "eval_samples_per_second": 731.408, + "eval_steps_per_second": 11.437, + "eval_unsafe_aucpr": 0.9491187573387495, + "eval_unsafe_f1": 0.8679932566680715, + "eval_unsafe_fpr": 0.15572141169410736, + "eval_unsafe_precision": 0.8741246020918599, + "eval_unsafe_recall": 0.8619473259394338, + "step": 10570 + }, + { + "epoch": 1.2514785900165601, + "grad_norm": 3.807976484298706, + "learning_rate": 5.312018194876706e-06, + "loss": 0.3023, + "step": 10580 + }, + { + "epoch": 1.2526614620298084, + "grad_norm": 2.064093589782715, + "learning_rate": 5.311299976059373e-06, + "loss": 0.2914, + "step": 10590 + }, + { + "epoch": 1.2538443340430565, + "grad_norm": 3.07285475730896, + "learning_rate": 5.31058175724204e-06, + "loss": 0.2944, + "step": 10600 + }, + { + "epoch": 1.2550272060563046, + "grad_norm": 2.0549986362457275, + "learning_rate": 5.309863538424707e-06, + "loss": 0.3066, + "step": 10610 + }, + { + "epoch": 1.2562100780695529, + "grad_norm": 3.30754017829895, + "learning_rate": 5.309145319607374e-06, + "loss": 0.3404, + "step": 10620 + }, + { + "epoch": 1.2573929500828012, + "grad_norm": 2.483982563018799, + "learning_rate": 5.3084271007900415e-06, + "loss": 0.3406, + "step": 10630 + }, + { + "epoch": 1.2585758220960492, + "grad_norm": 2.510258436203003, + "learning_rate": 5.307708881972708e-06, + "loss": 0.3272, + "step": 10640 + }, + { + "epoch": 1.2597586941092973, + "grad_norm": 2.251511335372925, + "learning_rate": 5.306990663155375e-06, + "loss": 0.3063, + "step": 10650 + }, + { + "epoch": 1.2609415661225456, + "grad_norm": 2.3225739002227783, + "learning_rate": 5.3062724443380414e-06, + "loss": 0.3289, + "step": 10660 + }, + { + "epoch": 1.2621244381357937, + "grad_norm": 2.219557046890259, + "learning_rate": 5.305554225520709e-06, + "loss": 0.3263, + "step": 10670 + }, + { + "epoch": 1.2633073101490417, + "grad_norm": 2.8224964141845703, + "learning_rate": 5.304836006703375e-06, + "loss": 0.2961, + "step": 10680 + }, + { + "epoch": 1.26449018216229, + "grad_norm": 2.6272690296173096, + "learning_rate": 5.304117787886043e-06, + "loss": 0.3388, + "step": 10690 + }, + { + "epoch": 1.2656730541755383, + "grad_norm": 3.0494840145111084, + "learning_rate": 5.30339956906871e-06, + "loss": 0.3588, + "step": 10700 + }, + { + "epoch": 1.2668559261887864, + "grad_norm": 2.026481866836548, + "learning_rate": 5.302681350251377e-06, + "loss": 0.3125, + "step": 10710 + }, + { + "epoch": 1.2680387982020345, + "grad_norm": 2.3703815937042236, + "learning_rate": 5.301963131434044e-06, + "loss": 0.2892, + "step": 10720 + }, + { + "epoch": 1.2692216702152828, + "grad_norm": 3.1509850025177, + "learning_rate": 5.301244912616711e-06, + "loss": 0.3643, + "step": 10730 + }, + { + "epoch": 1.2704045422285308, + "grad_norm": 2.430760145187378, + "learning_rate": 5.300526693799378e-06, + "loss": 0.3583, + "step": 10740 + }, + { + "epoch": 1.271587414241779, + "grad_norm": 3.22849702835083, + "learning_rate": 5.299808474982045e-06, + "loss": 0.3373, + "step": 10750 + }, + { + "epoch": 1.2727702862550272, + "grad_norm": 2.1429686546325684, + "learning_rate": 5.2990902561647115e-06, + "loss": 0.3179, + "step": 10760 + }, + { + "epoch": 1.2739531582682755, + "grad_norm": 3.37874174118042, + "learning_rate": 5.2983720373473784e-06, + "loss": 0.3002, + "step": 10770 + }, + { + "epoch": 1.2751360302815236, + "grad_norm": 2.9585418701171875, + "learning_rate": 5.297653818530045e-06, + "loss": 0.302, + "step": 10780 + }, + { + "epoch": 1.2763189022947716, + "grad_norm": 2.9790499210357666, + "learning_rate": 5.296935599712712e-06, + "loss": 0.3347, + "step": 10790 + }, + { + "epoch": 1.27750177430802, + "grad_norm": 3.7681424617767334, + "learning_rate": 5.296217380895379e-06, + "loss": 0.2844, + "step": 10800 + }, + { + "epoch": 1.278684646321268, + "grad_norm": 3.4820749759674072, + "learning_rate": 5.295499162078047e-06, + "loss": 0.3105, + "step": 10810 + }, + { + "epoch": 1.2798675183345163, + "grad_norm": 3.544236421585083, + "learning_rate": 5.294780943260713e-06, + "loss": 0.3314, + "step": 10820 + }, + { + "epoch": 1.2810503903477644, + "grad_norm": 2.3502871990203857, + "learning_rate": 5.294062724443381e-06, + "loss": 0.3441, + "step": 10830 + }, + { + "epoch": 1.2822332623610126, + "grad_norm": 3.532820701599121, + "learning_rate": 5.293344505626047e-06, + "loss": 0.3264, + "step": 10840 + }, + { + "epoch": 1.2834161343742607, + "grad_norm": 2.5632216930389404, + "learning_rate": 5.292626286808715e-06, + "loss": 0.3387, + "step": 10850 + }, + { + "epoch": 1.2845990063875088, + "grad_norm": 2.079050064086914, + "learning_rate": 5.291908067991382e-06, + "loss": 0.3391, + "step": 10860 + }, + { + "epoch": 1.285781878400757, + "grad_norm": 2.3422341346740723, + "learning_rate": 5.2911898491740485e-06, + "loss": 0.355, + "step": 10870 + }, + { + "epoch": 1.2869647504140052, + "grad_norm": 3.7250454425811768, + "learning_rate": 5.2904716303567154e-06, + "loss": 0.3188, + "step": 10880 + }, + { + "epoch": 1.2881476224272534, + "grad_norm": 2.586632013320923, + "learning_rate": 5.289753411539382e-06, + "loss": 0.2941, + "step": 10890 + }, + { + "epoch": 1.2893304944405015, + "grad_norm": 3.148038387298584, + "learning_rate": 5.28903519272205e-06, + "loss": 0.3393, + "step": 10900 + }, + { + "epoch": 1.2905133664537498, + "grad_norm": 2.8730502128601074, + "learning_rate": 5.288316973904716e-06, + "loss": 0.3557, + "step": 10910 + }, + { + "epoch": 1.2916962384669979, + "grad_norm": 2.5956265926361084, + "learning_rate": 5.287598755087384e-06, + "loss": 0.3339, + "step": 10920 + }, + { + "epoch": 1.292879110480246, + "grad_norm": 3.057295083999634, + "learning_rate": 5.28688053627005e-06, + "loss": 0.2943, + "step": 10930 + }, + { + "epoch": 1.2940619824934942, + "grad_norm": 2.643946647644043, + "learning_rate": 5.286162317452718e-06, + "loss": 0.3291, + "step": 10940 + }, + { + "epoch": 1.2952448545067423, + "grad_norm": 2.0065228939056396, + "learning_rate": 5.285444098635384e-06, + "loss": 0.3167, + "step": 10950 + }, + { + "epoch": 1.2964277265199906, + "grad_norm": 2.5850989818573, + "learning_rate": 5.284725879818052e-06, + "loss": 0.3454, + "step": 10960 + }, + { + "epoch": 1.2976105985332387, + "grad_norm": 2.7382800579071045, + "learning_rate": 5.284007661000719e-06, + "loss": 0.3218, + "step": 10970 + }, + { + "epoch": 1.298793470546487, + "grad_norm": 3.3061835765838623, + "learning_rate": 5.2832894421833855e-06, + "loss": 0.2986, + "step": 10980 + }, + { + "epoch": 1.299976342559735, + "grad_norm": 3.0424976348876953, + "learning_rate": 5.2825712233660524e-06, + "loss": 0.3172, + "step": 10990 + }, + { + "epoch": 1.3011592145729831, + "grad_norm": 3.1832244396209717, + "learning_rate": 5.281853004548719e-06, + "loss": 0.3074, + "step": 11000 + }, + { + "epoch": 1.3023420865862314, + "grad_norm": 2.583782434463501, + "learning_rate": 5.281134785731386e-06, + "loss": 0.3497, + "step": 11010 + }, + { + "epoch": 1.3035249585994795, + "grad_norm": 2.6923210620880127, + "learning_rate": 5.280416566914053e-06, + "loss": 0.314, + "step": 11020 + }, + { + "epoch": 1.3047078306127278, + "grad_norm": 2.3665506839752197, + "learning_rate": 5.27969834809672e-06, + "loss": 0.351, + "step": 11030 + }, + { + "epoch": 1.3058907026259758, + "grad_norm": 2.606867790222168, + "learning_rate": 5.278980129279387e-06, + "loss": 0.3156, + "step": 11040 + }, + { + "epoch": 1.3070735746392241, + "grad_norm": 2.3391175270080566, + "learning_rate": 5.278261910462054e-06, + "loss": 0.3196, + "step": 11050 + }, + { + "epoch": 1.3082564466524722, + "grad_norm": 2.604893207550049, + "learning_rate": 5.277543691644721e-06, + "loss": 0.3428, + "step": 11060 + }, + { + "epoch": 1.3094393186657203, + "grad_norm": 2.3523573875427246, + "learning_rate": 5.276825472827388e-06, + "loss": 0.3045, + "step": 11070 + }, + { + "epoch": 1.3106221906789686, + "grad_norm": 2.6932084560394287, + "learning_rate": 5.276107254010056e-06, + "loss": 0.3248, + "step": 11080 + }, + { + "epoch": 1.3118050626922166, + "grad_norm": 3.2833149433135986, + "learning_rate": 5.275389035192722e-06, + "loss": 0.2873, + "step": 11090 + }, + { + "epoch": 1.312987934705465, + "grad_norm": 2.871731996536255, + "learning_rate": 5.2746708163753894e-06, + "loss": 0.3365, + "step": 11100 + }, + { + "epoch": 1.314170806718713, + "grad_norm": 2.786803722381592, + "learning_rate": 5.273952597558056e-06, + "loss": 0.3591, + "step": 11110 + }, + { + "epoch": 1.3153536787319613, + "grad_norm": 1.9333808422088623, + "learning_rate": 5.273234378740723e-06, + "loss": 0.3163, + "step": 11120 + }, + { + "epoch": 1.3165365507452094, + "grad_norm": 2.6212830543518066, + "learning_rate": 5.27251615992339e-06, + "loss": 0.3183, + "step": 11130 + }, + { + "epoch": 1.3177194227584574, + "grad_norm": 2.1172983646392822, + "learning_rate": 5.271797941106057e-06, + "loss": 0.3007, + "step": 11140 + }, + { + "epoch": 1.3189022947717057, + "grad_norm": 2.256084680557251, + "learning_rate": 5.271079722288724e-06, + "loss": 0.3298, + "step": 11150 + }, + { + "epoch": 1.3200851667849538, + "grad_norm": 2.8654685020446777, + "learning_rate": 5.270361503471391e-06, + "loss": 0.2992, + "step": 11160 + }, + { + "epoch": 1.321268038798202, + "grad_norm": 2.6788671016693115, + "learning_rate": 5.269643284654059e-06, + "loss": 0.357, + "step": 11170 + }, + { + "epoch": 1.3224509108114502, + "grad_norm": 3.450791358947754, + "learning_rate": 5.268925065836725e-06, + "loss": 0.3608, + "step": 11180 + }, + { + "epoch": 1.3236337828246985, + "grad_norm": 3.0075740814208984, + "learning_rate": 5.268206847019393e-06, + "loss": 0.3562, + "step": 11190 + }, + { + "epoch": 1.3248166548379465, + "grad_norm": 2.720026969909668, + "learning_rate": 5.267488628202059e-06, + "loss": 0.3092, + "step": 11200 + }, + { + "epoch": 1.3259995268511946, + "grad_norm": 2.3526370525360107, + "learning_rate": 5.2667704093847264e-06, + "loss": 0.3435, + "step": 11210 + }, + { + "epoch": 1.327182398864443, + "grad_norm": 2.9320015907287598, + "learning_rate": 5.2660521905673925e-06, + "loss": 0.3487, + "step": 11220 + }, + { + "epoch": 1.328365270877691, + "grad_norm": 4.365499973297119, + "learning_rate": 5.26533397175006e-06, + "loss": 0.3665, + "step": 11230 + }, + { + "epoch": 1.3295481428909393, + "grad_norm": 2.805607795715332, + "learning_rate": 5.264615752932727e-06, + "loss": 0.3094, + "step": 11240 + }, + { + "epoch": 1.3307310149041873, + "grad_norm": 2.4460062980651855, + "learning_rate": 5.263897534115394e-06, + "loss": 0.3369, + "step": 11250 + }, + { + "epoch": 1.3319138869174356, + "grad_norm": 2.7052359580993652, + "learning_rate": 5.263179315298061e-06, + "loss": 0.2975, + "step": 11260 + }, + { + "epoch": 1.3330967589306837, + "grad_norm": 2.7273125648498535, + "learning_rate": 5.262461096480728e-06, + "loss": 0.3063, + "step": 11270 + }, + { + "epoch": 1.3342796309439318, + "grad_norm": 3.5072457790374756, + "learning_rate": 5.261742877663395e-06, + "loss": 0.3081, + "step": 11280 + }, + { + "epoch": 1.33546250295718, + "grad_norm": 2.7681143283843994, + "learning_rate": 5.261024658846062e-06, + "loss": 0.2928, + "step": 11290 + }, + { + "epoch": 1.3366453749704281, + "grad_norm": 3.2708802223205566, + "learning_rate": 5.260306440028729e-06, + "loss": 0.394, + "step": 11300 + }, + { + "epoch": 1.3378282469836764, + "grad_norm": 2.7023983001708984, + "learning_rate": 5.259588221211396e-06, + "loss": 0.2943, + "step": 11310 + }, + { + "epoch": 1.3390111189969245, + "grad_norm": 4.062198638916016, + "learning_rate": 5.258870002394063e-06, + "loss": 0.3312, + "step": 11320 + }, + { + "epoch": 1.3401939910101728, + "grad_norm": 2.1799333095550537, + "learning_rate": 5.2581517835767295e-06, + "loss": 0.351, + "step": 11330 + }, + { + "epoch": 1.3413768630234209, + "grad_norm": 2.283543109893799, + "learning_rate": 5.257433564759396e-06, + "loss": 0.3142, + "step": 11340 + }, + { + "epoch": 1.342559735036669, + "grad_norm": 2.688551664352417, + "learning_rate": 5.256715345942064e-06, + "loss": 0.3193, + "step": 11350 + }, + { + "epoch": 1.3437426070499172, + "grad_norm": 2.31280517578125, + "learning_rate": 5.255997127124731e-06, + "loss": 0.364, + "step": 11360 + }, + { + "epoch": 1.3449254790631653, + "grad_norm": 1.948833703994751, + "learning_rate": 5.255278908307398e-06, + "loss": 0.3257, + "step": 11370 + }, + { + "epoch": 1.3461083510764136, + "grad_norm": 2.9267709255218506, + "learning_rate": 5.254560689490065e-06, + "loss": 0.3508, + "step": 11380 + }, + { + "epoch": 1.3472912230896616, + "grad_norm": 2.355447292327881, + "learning_rate": 5.253842470672732e-06, + "loss": 0.2633, + "step": 11390 + }, + { + "epoch": 1.34847409510291, + "grad_norm": 3.1869235038757324, + "learning_rate": 5.253124251855399e-06, + "loss": 0.3702, + "step": 11400 + }, + { + "epoch": 1.349656967116158, + "grad_norm": 2.661611557006836, + "learning_rate": 5.252406033038066e-06, + "loss": 0.3112, + "step": 11410 + }, + { + "epoch": 1.350839839129406, + "grad_norm": 2.0842931270599365, + "learning_rate": 5.251687814220733e-06, + "loss": 0.3578, + "step": 11420 + }, + { + "epoch": 1.3520227111426544, + "grad_norm": 2.38382887840271, + "learning_rate": 5.2509695954033996e-06, + "loss": 0.2996, + "step": 11430 + }, + { + "epoch": 1.3532055831559027, + "grad_norm": 4.086627960205078, + "learning_rate": 5.2502513765860665e-06, + "loss": 0.3499, + "step": 11440 + }, + { + "epoch": 1.3543884551691507, + "grad_norm": 2.342360496520996, + "learning_rate": 5.249533157768733e-06, + "loss": 0.3639, + "step": 11450 + }, + { + "epoch": 1.3555713271823988, + "grad_norm": 2.4554357528686523, + "learning_rate": 5.248814938951401e-06, + "loss": 0.3128, + "step": 11460 + }, + { + "epoch": 1.356754199195647, + "grad_norm": 3.7021262645721436, + "learning_rate": 5.248096720134067e-06, + "loss": 0.2959, + "step": 11470 + }, + { + "epoch": 1.3579370712088952, + "grad_norm": 2.7638843059539795, + "learning_rate": 5.247378501316735e-06, + "loss": 0.3535, + "step": 11480 + }, + { + "epoch": 1.3591199432221432, + "grad_norm": 2.3986899852752686, + "learning_rate": 5.246660282499401e-06, + "loss": 0.3337, + "step": 11490 + }, + { + "epoch": 1.3603028152353915, + "grad_norm": 2.0446488857269287, + "learning_rate": 5.245942063682069e-06, + "loss": 0.2912, + "step": 11500 + }, + { + "epoch": 1.3614856872486398, + "grad_norm": 3.6219053268432617, + "learning_rate": 5.245223844864736e-06, + "loss": 0.3002, + "step": 11510 + }, + { + "epoch": 1.362668559261888, + "grad_norm": 3.7377607822418213, + "learning_rate": 5.244505626047403e-06, + "loss": 0.3342, + "step": 11520 + }, + { + "epoch": 1.363851431275136, + "grad_norm": 2.9352986812591553, + "learning_rate": 5.24378740723007e-06, + "loss": 0.332, + "step": 11530 + }, + { + "epoch": 1.3650343032883843, + "grad_norm": 2.2465410232543945, + "learning_rate": 5.2430691884127366e-06, + "loss": 0.3389, + "step": 11540 + }, + { + "epoch": 1.3662171753016323, + "grad_norm": 2.2292497158050537, + "learning_rate": 5.2423509695954035e-06, + "loss": 0.3494, + "step": 11550 + }, + { + "epoch": 1.3674000473148804, + "grad_norm": 3.6063504219055176, + "learning_rate": 5.24163275077807e-06, + "loss": 0.3452, + "step": 11560 + }, + { + "epoch": 1.3685829193281287, + "grad_norm": 4.247971057891846, + "learning_rate": 5.240914531960737e-06, + "loss": 0.3217, + "step": 11570 + }, + { + "epoch": 1.369765791341377, + "grad_norm": 2.7763800621032715, + "learning_rate": 5.240196313143404e-06, + "loss": 0.3203, + "step": 11580 + }, + { + "epoch": 1.370948663354625, + "grad_norm": 1.9633818864822388, + "learning_rate": 5.239478094326071e-06, + "loss": 0.3002, + "step": 11590 + }, + { + "epoch": 1.3721315353678731, + "grad_norm": 2.645674467086792, + "learning_rate": 5.238759875508738e-06, + "loss": 0.3495, + "step": 11600 + }, + { + "epoch": 1.3733144073811214, + "grad_norm": 2.3986480236053467, + "learning_rate": 5.238041656691406e-06, + "loss": 0.2729, + "step": 11610 + }, + { + "epoch": 1.3744972793943695, + "grad_norm": 2.614625930786133, + "learning_rate": 5.237323437874073e-06, + "loss": 0.3146, + "step": 11620 + }, + { + "epoch": 1.3756801514076176, + "grad_norm": 2.432405948638916, + "learning_rate": 5.23660521905674e-06, + "loss": 0.3236, + "step": 11630 + }, + { + "epoch": 1.3768630234208659, + "grad_norm": 2.2257299423217773, + "learning_rate": 5.235887000239407e-06, + "loss": 0.2882, + "step": 11640 + }, + { + "epoch": 1.3780458954341142, + "grad_norm": 3.6342475414276123, + "learning_rate": 5.2351687814220736e-06, + "loss": 0.3301, + "step": 11650 + }, + { + "epoch": 1.3792287674473622, + "grad_norm": 2.5768871307373047, + "learning_rate": 5.2344505626047405e-06, + "loss": 0.3673, + "step": 11660 + }, + { + "epoch": 1.3804116394606103, + "grad_norm": 3.101595640182495, + "learning_rate": 5.233732343787407e-06, + "loss": 0.3189, + "step": 11670 + }, + { + "epoch": 1.3815945114738586, + "grad_norm": 2.9230449199676514, + "learning_rate": 5.233014124970074e-06, + "loss": 0.3122, + "step": 11680 + }, + { + "epoch": 1.3827773834871067, + "grad_norm": 3.3076393604278564, + "learning_rate": 5.232295906152741e-06, + "loss": 0.3312, + "step": 11690 + }, + { + "epoch": 1.3839602555003547, + "grad_norm": 2.8266141414642334, + "learning_rate": 5.231577687335408e-06, + "loss": 0.3375, + "step": 11700 + }, + { + "epoch": 1.385143127513603, + "grad_norm": 3.0489702224731445, + "learning_rate": 5.230859468518075e-06, + "loss": 0.3299, + "step": 11710 + }, + { + "epoch": 1.3863259995268513, + "grad_norm": 2.535065174102783, + "learning_rate": 5.230141249700742e-06, + "loss": 0.2962, + "step": 11720 + }, + { + "epoch": 1.3875088715400994, + "grad_norm": 2.4045450687408447, + "learning_rate": 5.22942303088341e-06, + "loss": 0.3162, + "step": 11730 + }, + { + "epoch": 1.3886917435533475, + "grad_norm": 3.010890483856201, + "learning_rate": 5.228704812066076e-06, + "loss": 0.346, + "step": 11740 + }, + { + "epoch": 1.3898746155665958, + "grad_norm": 2.096226215362549, + "learning_rate": 5.227986593248744e-06, + "loss": 0.3457, + "step": 11750 + }, + { + "epoch": 1.3910574875798438, + "grad_norm": 2.902386426925659, + "learning_rate": 5.22726837443141e-06, + "loss": 0.361, + "step": 11760 + }, + { + "epoch": 1.392240359593092, + "grad_norm": 2.158539056777954, + "learning_rate": 5.2265501556140775e-06, + "loss": 0.3379, + "step": 11770 + }, + { + "epoch": 1.3934232316063402, + "grad_norm": 2.7760837078094482, + "learning_rate": 5.2258319367967436e-06, + "loss": 0.346, + "step": 11780 + }, + { + "epoch": 1.3946061036195885, + "grad_norm": 2.0591464042663574, + "learning_rate": 5.225113717979411e-06, + "loss": 0.3663, + "step": 11790 + }, + { + "epoch": 1.3957889756328365, + "grad_norm": 2.9406681060791016, + "learning_rate": 5.224395499162078e-06, + "loss": 0.3514, + "step": 11800 + }, + { + "epoch": 1.3969718476460846, + "grad_norm": 2.612901210784912, + "learning_rate": 5.223677280344745e-06, + "loss": 0.3659, + "step": 11810 + }, + { + "epoch": 1.398154719659333, + "grad_norm": 3.072481870651245, + "learning_rate": 5.222959061527412e-06, + "loss": 0.3547, + "step": 11820 + }, + { + "epoch": 1.399337591672581, + "grad_norm": 2.3625690937042236, + "learning_rate": 5.222240842710079e-06, + "loss": 0.3135, + "step": 11830 + }, + { + "epoch": 1.4005204636858293, + "grad_norm": 2.9312098026275635, + "learning_rate": 5.221522623892746e-06, + "loss": 0.3182, + "step": 11840 + }, + { + "epoch": 1.4017033356990773, + "grad_norm": 3.0753791332244873, + "learning_rate": 5.220804405075413e-06, + "loss": 0.3232, + "step": 11850 + }, + { + "epoch": 1.4028862077123256, + "grad_norm": 3.2985544204711914, + "learning_rate": 5.22008618625808e-06, + "loss": 0.3542, + "step": 11860 + }, + { + "epoch": 1.4040690797255737, + "grad_norm": 3.0760841369628906, + "learning_rate": 5.219367967440747e-06, + "loss": 0.3421, + "step": 11870 + }, + { + "epoch": 1.4052519517388218, + "grad_norm": 2.818378210067749, + "learning_rate": 5.2186497486234145e-06, + "loss": 0.324, + "step": 11880 + }, + { + "epoch": 1.40643482375207, + "grad_norm": 2.093960762023926, + "learning_rate": 5.217931529806081e-06, + "loss": 0.324, + "step": 11890 + }, + { + "epoch": 1.4076176957653181, + "grad_norm": 2.4740841388702393, + "learning_rate": 5.217213310988748e-06, + "loss": 0.3533, + "step": 11900 + }, + { + "epoch": 1.4088005677785664, + "grad_norm": 2.261019229888916, + "learning_rate": 5.216495092171415e-06, + "loss": 0.3465, + "step": 11910 + }, + { + "epoch": 1.4099834397918145, + "grad_norm": 2.8680613040924072, + "learning_rate": 5.215776873354082e-06, + "loss": 0.3562, + "step": 11920 + }, + { + "epoch": 1.4111663118050628, + "grad_norm": 2.969210386276245, + "learning_rate": 5.215058654536749e-06, + "loss": 0.3316, + "step": 11930 + }, + { + "epoch": 1.4123491838183109, + "grad_norm": 2.5013744831085205, + "learning_rate": 5.214340435719416e-06, + "loss": 0.3386, + "step": 11940 + }, + { + "epoch": 1.413532055831559, + "grad_norm": 3.086883068084717, + "learning_rate": 5.213622216902083e-06, + "loss": 0.3355, + "step": 11950 + }, + { + "epoch": 1.4147149278448072, + "grad_norm": 2.6564736366271973, + "learning_rate": 5.21290399808475e-06, + "loss": 0.2641, + "step": 11960 + }, + { + "epoch": 1.4158977998580553, + "grad_norm": 3.8855438232421875, + "learning_rate": 5.212185779267417e-06, + "loss": 0.3573, + "step": 11970 + }, + { + "epoch": 1.4170806718713036, + "grad_norm": 3.215693473815918, + "learning_rate": 5.211467560450084e-06, + "loss": 0.3343, + "step": 11980 + }, + { + "epoch": 1.4182635438845517, + "grad_norm": 2.6905269622802734, + "learning_rate": 5.210749341632751e-06, + "loss": 0.3521, + "step": 11990 + }, + { + "epoch": 1.4194464158978, + "grad_norm": 2.1683082580566406, + "learning_rate": 5.210031122815418e-06, + "loss": 0.3104, + "step": 12000 + }, + { + "epoch": 1.420629287911048, + "grad_norm": 1.7524878978729248, + "learning_rate": 5.2093129039980845e-06, + "loss": 0.3374, + "step": 12010 + }, + { + "epoch": 1.421812159924296, + "grad_norm": 2.7673189640045166, + "learning_rate": 5.208594685180752e-06, + "loss": 0.2807, + "step": 12020 + }, + { + "epoch": 1.4229950319375444, + "grad_norm": 3.531315565109253, + "learning_rate": 5.207876466363418e-06, + "loss": 0.2941, + "step": 12030 + }, + { + "epoch": 1.4241779039507925, + "grad_norm": 3.0989954471588135, + "learning_rate": 5.207158247546086e-06, + "loss": 0.3545, + "step": 12040 + }, + { + "epoch": 1.4253607759640408, + "grad_norm": 3.6251354217529297, + "learning_rate": 5.206440028728752e-06, + "loss": 0.3375, + "step": 12050 + }, + { + "epoch": 1.4265436479772888, + "grad_norm": 2.1570992469787598, + "learning_rate": 5.20572180991142e-06, + "loss": 0.3477, + "step": 12060 + }, + { + "epoch": 1.4277265199905371, + "grad_norm": 2.2915897369384766, + "learning_rate": 5.205003591094087e-06, + "loss": 0.3349, + "step": 12070 + }, + { + "epoch": 1.4289093920037852, + "grad_norm": 2.6954102516174316, + "learning_rate": 5.204285372276754e-06, + "loss": 0.3117, + "step": 12080 + }, + { + "epoch": 1.4300922640170333, + "grad_norm": 2.97685170173645, + "learning_rate": 5.203567153459421e-06, + "loss": 0.2804, + "step": 12090 + }, + { + "epoch": 1.4312751360302816, + "grad_norm": 2.3667688369750977, + "learning_rate": 5.202848934642088e-06, + "loss": 0.3374, + "step": 12100 + }, + { + "epoch": 1.4324580080435296, + "grad_norm": 2.4757254123687744, + "learning_rate": 5.2021307158247546e-06, + "loss": 0.299, + "step": 12110 + }, + { + "epoch": 1.433640880056778, + "grad_norm": 2.718502998352051, + "learning_rate": 5.2014124970074215e-06, + "loss": 0.3317, + "step": 12120 + }, + { + "epoch": 1.434823752070026, + "grad_norm": 2.0895771980285645, + "learning_rate": 5.200694278190089e-06, + "loss": 0.3412, + "step": 12130 + }, + { + "epoch": 1.4360066240832743, + "grad_norm": 3.8666303157806396, + "learning_rate": 5.199976059372755e-06, + "loss": 0.3479, + "step": 12140 + }, + { + "epoch": 1.4371894960965224, + "grad_norm": 2.6922268867492676, + "learning_rate": 5.199257840555423e-06, + "loss": 0.3274, + "step": 12150 + }, + { + "epoch": 1.4383723681097704, + "grad_norm": 1.681287407875061, + "learning_rate": 5.19853962173809e-06, + "loss": 0.3032, + "step": 12160 + }, + { + "epoch": 1.4395552401230187, + "grad_norm": 3.0996005535125732, + "learning_rate": 5.197821402920757e-06, + "loss": 0.3212, + "step": 12170 + }, + { + "epoch": 1.4407381121362668, + "grad_norm": 3.6387147903442383, + "learning_rate": 5.197103184103424e-06, + "loss": 0.3263, + "step": 12180 + }, + { + "epoch": 1.441920984149515, + "grad_norm": 2.525237798690796, + "learning_rate": 5.196384965286091e-06, + "loss": 0.3255, + "step": 12190 + }, + { + "epoch": 1.4431038561627632, + "grad_norm": 2.6674628257751465, + "learning_rate": 5.195666746468758e-06, + "loss": 0.3566, + "step": 12200 + }, + { + "epoch": 1.4442867281760114, + "grad_norm": 3.596350908279419, + "learning_rate": 5.194948527651425e-06, + "loss": 0.3259, + "step": 12210 + }, + { + "epoch": 1.4454696001892595, + "grad_norm": 2.30674409866333, + "learning_rate": 5.1942303088340916e-06, + "loss": 0.3329, + "step": 12220 + }, + { + "epoch": 1.4466524722025076, + "grad_norm": 1.9230536222457886, + "learning_rate": 5.1935120900167585e-06, + "loss": 0.3326, + "step": 12230 + }, + { + "epoch": 1.4478353442157559, + "grad_norm": 2.4507272243499756, + "learning_rate": 5.192793871199425e-06, + "loss": 0.31, + "step": 12240 + }, + { + "epoch": 1.449018216229004, + "grad_norm": 2.7487950325012207, + "learning_rate": 5.192075652382092e-06, + "loss": 0.3455, + "step": 12250 + }, + { + "epoch": 1.4502010882422522, + "grad_norm": 2.4836816787719727, + "learning_rate": 5.191357433564759e-06, + "loss": 0.3169, + "step": 12260 + }, + { + "epoch": 1.4513839602555003, + "grad_norm": 3.292184591293335, + "learning_rate": 5.190639214747427e-06, + "loss": 0.3226, + "step": 12270 + }, + { + "epoch": 1.4525668322687486, + "grad_norm": 2.257387638092041, + "learning_rate": 5.189920995930093e-06, + "loss": 0.2974, + "step": 12280 + }, + { + "epoch": 1.4537497042819967, + "grad_norm": 3.23683762550354, + "learning_rate": 5.189202777112761e-06, + "loss": 0.3179, + "step": 12290 + }, + { + "epoch": 1.4549325762952448, + "grad_norm": 3.1413047313690186, + "learning_rate": 5.188484558295427e-06, + "loss": 0.3107, + "step": 12300 + }, + { + "epoch": 1.456115448308493, + "grad_norm": 3.582163095474243, + "learning_rate": 5.187766339478095e-06, + "loss": 0.3294, + "step": 12310 + }, + { + "epoch": 1.4572983203217411, + "grad_norm": 2.904231548309326, + "learning_rate": 5.187048120660761e-06, + "loss": 0.3329, + "step": 12320 + }, + { + "epoch": 1.4584811923349894, + "grad_norm": 3.519880533218384, + "learning_rate": 5.1863299018434286e-06, + "loss": 0.3343, + "step": 12330 + }, + { + "epoch": 1.4596640643482375, + "grad_norm": 3.1488707065582275, + "learning_rate": 5.1856116830260955e-06, + "loss": 0.3015, + "step": 12340 + }, + { + "epoch": 1.4608469363614858, + "grad_norm": 2.2017030715942383, + "learning_rate": 5.184893464208762e-06, + "loss": 0.3118, + "step": 12350 + }, + { + "epoch": 1.4620298083747338, + "grad_norm": 4.503572940826416, + "learning_rate": 5.184175245391429e-06, + "loss": 0.3325, + "step": 12360 + }, + { + "epoch": 1.463212680387982, + "grad_norm": 2.683884859085083, + "learning_rate": 5.183457026574096e-06, + "loss": 0.3451, + "step": 12370 + }, + { + "epoch": 1.4643955524012302, + "grad_norm": 2.986691951751709, + "learning_rate": 5.182738807756764e-06, + "loss": 0.3588, + "step": 12380 + }, + { + "epoch": 1.4655784244144783, + "grad_norm": 2.1415863037109375, + "learning_rate": 5.18202058893943e-06, + "loss": 0.3279, + "step": 12390 + }, + { + "epoch": 1.4667612964277266, + "grad_norm": 2.3410966396331787, + "learning_rate": 5.181302370122098e-06, + "loss": 0.3175, + "step": 12400 + }, + { + "epoch": 1.4679441684409746, + "grad_norm": 3.037393808364868, + "learning_rate": 5.180584151304764e-06, + "loss": 0.3485, + "step": 12410 + }, + { + "epoch": 1.469127040454223, + "grad_norm": 2.986999034881592, + "learning_rate": 5.179865932487432e-06, + "loss": 0.3268, + "step": 12420 + }, + { + "epoch": 1.470309912467471, + "grad_norm": 2.3629612922668457, + "learning_rate": 5.179147713670098e-06, + "loss": 0.2918, + "step": 12430 + }, + { + "epoch": 1.471492784480719, + "grad_norm": 3.546999931335449, + "learning_rate": 5.1784294948527656e-06, + "loss": 0.3504, + "step": 12440 + }, + { + "epoch": 1.4726756564939674, + "grad_norm": 2.512006998062134, + "learning_rate": 5.1777112760354325e-06, + "loss": 0.3147, + "step": 12450 + }, + { + "epoch": 1.4738585285072154, + "grad_norm": 2.3637118339538574, + "learning_rate": 5.176993057218099e-06, + "loss": 0.3577, + "step": 12460 + }, + { + "epoch": 1.4750414005204637, + "grad_norm": 3.7535767555236816, + "learning_rate": 5.176274838400766e-06, + "loss": 0.3255, + "step": 12470 + }, + { + "epoch": 1.4762242725337118, + "grad_norm": 2.303182363510132, + "learning_rate": 5.175556619583433e-06, + "loss": 0.3162, + "step": 12480 + }, + { + "epoch": 1.47740714454696, + "grad_norm": 2.876225471496582, + "learning_rate": 5.1748384007661e-06, + "loss": 0.2976, + "step": 12490 + }, + { + "epoch": 1.4785900165602082, + "grad_norm": 2.6297693252563477, + "learning_rate": 5.174120181948767e-06, + "loss": 0.3224, + "step": 12500 + }, + { + "epoch": 1.4797728885734562, + "grad_norm": 3.692824125289917, + "learning_rate": 5.173401963131434e-06, + "loss": 0.3632, + "step": 12510 + }, + { + "epoch": 1.4809557605867045, + "grad_norm": 3.0064969062805176, + "learning_rate": 5.172683744314101e-06, + "loss": 0.3408, + "step": 12520 + }, + { + "epoch": 1.4821386325999528, + "grad_norm": 3.15277099609375, + "learning_rate": 5.171965525496768e-06, + "loss": 0.3292, + "step": 12530 + }, + { + "epoch": 1.483321504613201, + "grad_norm": 3.2995247840881348, + "learning_rate": 5.171247306679436e-06, + "loss": 0.3373, + "step": 12540 + }, + { + "epoch": 1.484504376626449, + "grad_norm": 2.506042957305908, + "learning_rate": 5.170529087862102e-06, + "loss": 0.2838, + "step": 12550 + }, + { + "epoch": 1.4856872486396973, + "grad_norm": 4.044424533843994, + "learning_rate": 5.1698108690447695e-06, + "loss": 0.3801, + "step": 12560 + }, + { + "epoch": 1.4868701206529453, + "grad_norm": 2.122138261795044, + "learning_rate": 5.1690926502274355e-06, + "loss": 0.3225, + "step": 12570 + }, + { + "epoch": 1.4880529926661934, + "grad_norm": 2.4794294834136963, + "learning_rate": 5.168374431410103e-06, + "loss": 0.3351, + "step": 12580 + }, + { + "epoch": 1.4892358646794417, + "grad_norm": 2.462641716003418, + "learning_rate": 5.167656212592769e-06, + "loss": 0.3448, + "step": 12590 + }, + { + "epoch": 1.49041873669269, + "grad_norm": 2.709441900253296, + "learning_rate": 5.166937993775437e-06, + "loss": 0.3474, + "step": 12600 + }, + { + "epoch": 1.491601608705938, + "grad_norm": 1.9150044918060303, + "learning_rate": 5.166219774958104e-06, + "loss": 0.302, + "step": 12610 + }, + { + "epoch": 1.4927844807191861, + "grad_norm": 3.5195353031158447, + "learning_rate": 5.165501556140771e-06, + "loss": 0.2885, + "step": 12620 + }, + { + "epoch": 1.4939673527324344, + "grad_norm": 2.907759666442871, + "learning_rate": 5.164783337323438e-06, + "loss": 0.3514, + "step": 12630 + }, + { + "epoch": 1.4951502247456825, + "grad_norm": 2.093947410583496, + "learning_rate": 5.164065118506105e-06, + "loss": 0.3439, + "step": 12640 + }, + { + "epoch": 1.4963330967589306, + "grad_norm": 2.4467952251434326, + "learning_rate": 5.163346899688773e-06, + "loss": 0.3049, + "step": 12650 + }, + { + "epoch": 1.4975159687721789, + "grad_norm": 3.3428568840026855, + "learning_rate": 5.162628680871439e-06, + "loss": 0.3278, + "step": 12660 + }, + { + "epoch": 1.4986988407854271, + "grad_norm": 2.2584049701690674, + "learning_rate": 5.1619104620541065e-06, + "loss": 0.2973, + "step": 12670 + }, + { + "epoch": 1.4998817127986752, + "grad_norm": 2.3084425926208496, + "learning_rate": 5.1611922432367725e-06, + "loss": 0.2894, + "step": 12680 + }, + { + "epoch": 1.5003548616039746, + "eval_accuracy": 0.8551918022424061, + "eval_loss": 0.3492177426815033, + "eval_runtime": 78.9912, + "eval_safe_aucpr": 0.9095762579248807, + "eval_safe_f1": 0.8370795979861878, + "eval_safe_fpr": 0.13168515141550288, + "eval_safe_precision": 0.835437836222355, + "eval_safe_recall": 0.8387278250759479, + "eval_samples_per_second": 761.022, + "eval_steps_per_second": 11.9, + "eval_unsafe_aucpr": 0.9499714959856256, + "eval_unsafe_f1": 0.869679776037846, + "eval_unsafe_fpr": 0.16127217492405146, + "eval_unsafe_precision": 0.8710490013794758, + "eval_unsafe_recall": 0.8683148485844967, + "step": 12684 + }, + { + "epoch": 1.5010645848119233, + "grad_norm": 3.8595757484436035, + "learning_rate": 5.16047402441944e-06, + "loss": 0.3577, + "step": 12690 + }, + { + "epoch": 1.5022474568251716, + "grad_norm": 2.9607419967651367, + "learning_rate": 5.159755805602106e-06, + "loss": 0.3321, + "step": 12700 + }, + { + "epoch": 1.5034303288384196, + "grad_norm": 2.4730141162872314, + "learning_rate": 5.159037586784774e-06, + "loss": 0.3155, + "step": 12710 + }, + { + "epoch": 1.5046132008516677, + "grad_norm": 2.097149133682251, + "learning_rate": 5.158319367967441e-06, + "loss": 0.3253, + "step": 12720 + }, + { + "epoch": 1.505796072864916, + "grad_norm": 3.887446641921997, + "learning_rate": 5.157601149150108e-06, + "loss": 0.3647, + "step": 12730 + }, + { + "epoch": 1.5069789448781643, + "grad_norm": 2.82692289352417, + "learning_rate": 5.156882930332775e-06, + "loss": 0.3596, + "step": 12740 + }, + { + "epoch": 1.5081618168914124, + "grad_norm": 2.2878870964050293, + "learning_rate": 5.156164711515442e-06, + "loss": 0.3325, + "step": 12750 + }, + { + "epoch": 1.5093446889046604, + "grad_norm": 1.9933377504348755, + "learning_rate": 5.155446492698109e-06, + "loss": 0.3292, + "step": 12760 + }, + { + "epoch": 1.5105275609179087, + "grad_norm": 2.473442792892456, + "learning_rate": 5.154728273880776e-06, + "loss": 0.3369, + "step": 12770 + }, + { + "epoch": 1.5117104329311568, + "grad_norm": 2.962991237640381, + "learning_rate": 5.154010055063443e-06, + "loss": 0.3111, + "step": 12780 + }, + { + "epoch": 1.5128933049444049, + "grad_norm": 3.5352776050567627, + "learning_rate": 5.1532918362461095e-06, + "loss": 0.3317, + "step": 12790 + }, + { + "epoch": 1.5140761769576532, + "grad_norm": 2.941756010055542, + "learning_rate": 5.1525736174287765e-06, + "loss": 0.3628, + "step": 12800 + }, + { + "epoch": 1.5152590489709015, + "grad_norm": 2.2009263038635254, + "learning_rate": 5.151855398611444e-06, + "loss": 0.3233, + "step": 12810 + }, + { + "epoch": 1.5164419209841495, + "grad_norm": 1.9298968315124512, + "learning_rate": 5.15113717979411e-06, + "loss": 0.3409, + "step": 12820 + }, + { + "epoch": 1.5176247929973976, + "grad_norm": 3.332143545150757, + "learning_rate": 5.150418960976778e-06, + "loss": 0.3158, + "step": 12830 + }, + { + "epoch": 1.518807665010646, + "grad_norm": 3.445125102996826, + "learning_rate": 5.149700742159444e-06, + "loss": 0.3261, + "step": 12840 + }, + { + "epoch": 1.519990537023894, + "grad_norm": 4.30973482131958, + "learning_rate": 5.148982523342112e-06, + "loss": 0.3142, + "step": 12850 + }, + { + "epoch": 1.521173409037142, + "grad_norm": 3.637707471847534, + "learning_rate": 5.148264304524778e-06, + "loss": 0.3362, + "step": 12860 + }, + { + "epoch": 1.5223562810503903, + "grad_norm": 1.9568496942520142, + "learning_rate": 5.147546085707446e-06, + "loss": 0.3156, + "step": 12870 + }, + { + "epoch": 1.5235391530636386, + "grad_norm": 3.361077308654785, + "learning_rate": 5.146827866890113e-06, + "loss": 0.3081, + "step": 12880 + }, + { + "epoch": 1.5247220250768867, + "grad_norm": 2.1092593669891357, + "learning_rate": 5.14610964807278e-06, + "loss": 0.3118, + "step": 12890 + }, + { + "epoch": 1.5259048970901348, + "grad_norm": 2.4282894134521484, + "learning_rate": 5.1453914292554465e-06, + "loss": 0.3418, + "step": 12900 + }, + { + "epoch": 1.527087769103383, + "grad_norm": 1.910859227180481, + "learning_rate": 5.1446732104381135e-06, + "loss": 0.3249, + "step": 12910 + }, + { + "epoch": 1.5282706411166311, + "grad_norm": 1.8701509237289429, + "learning_rate": 5.143954991620781e-06, + "loss": 0.321, + "step": 12920 + }, + { + "epoch": 1.5294535131298792, + "grad_norm": 1.975669503211975, + "learning_rate": 5.143236772803447e-06, + "loss": 0.3111, + "step": 12930 + }, + { + "epoch": 1.5306363851431275, + "grad_norm": 2.1121163368225098, + "learning_rate": 5.142518553986115e-06, + "loss": 0.366, + "step": 12940 + }, + { + "epoch": 1.5318192571563758, + "grad_norm": 2.2763612270355225, + "learning_rate": 5.141800335168781e-06, + "loss": 0.3514, + "step": 12950 + }, + { + "epoch": 1.5330021291696239, + "grad_norm": 3.066706895828247, + "learning_rate": 5.141082116351449e-06, + "loss": 0.2817, + "step": 12960 + }, + { + "epoch": 1.534185001182872, + "grad_norm": 3.723223924636841, + "learning_rate": 5.140363897534115e-06, + "loss": 0.311, + "step": 12970 + }, + { + "epoch": 1.5353678731961202, + "grad_norm": 3.5070247650146484, + "learning_rate": 5.139645678716783e-06, + "loss": 0.2961, + "step": 12980 + }, + { + "epoch": 1.5365507452093683, + "grad_norm": 2.911402463912964, + "learning_rate": 5.13892745989945e-06, + "loss": 0.293, + "step": 12990 + }, + { + "epoch": 1.5377336172226164, + "grad_norm": 3.4748942852020264, + "learning_rate": 5.138209241082117e-06, + "loss": 0.3386, + "step": 13000 + }, + { + "epoch": 1.5389164892358647, + "grad_norm": 3.109605550765991, + "learning_rate": 5.1374910222647835e-06, + "loss": 0.3143, + "step": 13010 + }, + { + "epoch": 1.540099361249113, + "grad_norm": 2.4909555912017822, + "learning_rate": 5.1367728034474505e-06, + "loss": 0.3147, + "step": 13020 + }, + { + "epoch": 1.541282233262361, + "grad_norm": 4.059962272644043, + "learning_rate": 5.136054584630117e-06, + "loss": 0.3662, + "step": 13030 + }, + { + "epoch": 1.542465105275609, + "grad_norm": 2.515399694442749, + "learning_rate": 5.135336365812784e-06, + "loss": 0.3076, + "step": 13040 + }, + { + "epoch": 1.5436479772888574, + "grad_norm": 2.5504794120788574, + "learning_rate": 5.134618146995451e-06, + "loss": 0.3077, + "step": 13050 + }, + { + "epoch": 1.5448308493021055, + "grad_norm": 3.5667295455932617, + "learning_rate": 5.133899928178118e-06, + "loss": 0.3484, + "step": 13060 + }, + { + "epoch": 1.5460137213153535, + "grad_norm": 2.1200897693634033, + "learning_rate": 5.133181709360785e-06, + "loss": 0.3031, + "step": 13070 + }, + { + "epoch": 1.5471965933286018, + "grad_norm": 3.417478084564209, + "learning_rate": 5.132463490543453e-06, + "loss": 0.2872, + "step": 13080 + }, + { + "epoch": 1.5483794653418501, + "grad_norm": 3.6542742252349854, + "learning_rate": 5.131745271726119e-06, + "loss": 0.3273, + "step": 13090 + }, + { + "epoch": 1.5495623373550982, + "grad_norm": 2.444547414779663, + "learning_rate": 5.131027052908787e-06, + "loss": 0.334, + "step": 13100 + }, + { + "epoch": 1.5507452093683463, + "grad_norm": 3.357236385345459, + "learning_rate": 5.130308834091453e-06, + "loss": 0.319, + "step": 13110 + }, + { + "epoch": 1.5519280813815945, + "grad_norm": 2.1662325859069824, + "learning_rate": 5.1295906152741205e-06, + "loss": 0.2919, + "step": 13120 + }, + { + "epoch": 1.5531109533948428, + "grad_norm": 3.5623395442962646, + "learning_rate": 5.1288723964567875e-06, + "loss": 0.2937, + "step": 13130 + }, + { + "epoch": 1.5542938254080907, + "grad_norm": 2.6335065364837646, + "learning_rate": 5.128154177639454e-06, + "loss": 0.3072, + "step": 13140 + }, + { + "epoch": 1.555476697421339, + "grad_norm": 2.833822727203369, + "learning_rate": 5.127435958822121e-06, + "loss": 0.3179, + "step": 13150 + }, + { + "epoch": 1.5566595694345873, + "grad_norm": 2.881520986557007, + "learning_rate": 5.126717740004788e-06, + "loss": 0.3229, + "step": 13160 + }, + { + "epoch": 1.5578424414478353, + "grad_norm": 3.164757251739502, + "learning_rate": 5.125999521187455e-06, + "loss": 0.3208, + "step": 13170 + }, + { + "epoch": 1.5590253134610834, + "grad_norm": 2.3359062671661377, + "learning_rate": 5.125281302370122e-06, + "loss": 0.2989, + "step": 13180 + }, + { + "epoch": 1.5602081854743317, + "grad_norm": 2.1890311241149902, + "learning_rate": 5.12456308355279e-06, + "loss": 0.3499, + "step": 13190 + }, + { + "epoch": 1.56139105748758, + "grad_norm": 5.257014751434326, + "learning_rate": 5.123844864735456e-06, + "loss": 0.3021, + "step": 13200 + }, + { + "epoch": 1.5625739295008279, + "grad_norm": 3.5936100482940674, + "learning_rate": 5.123126645918124e-06, + "loss": 0.317, + "step": 13210 + }, + { + "epoch": 1.5637568015140761, + "grad_norm": 2.7884414196014404, + "learning_rate": 5.12240842710079e-06, + "loss": 0.2951, + "step": 13220 + }, + { + "epoch": 1.5649396735273244, + "grad_norm": 3.281930685043335, + "learning_rate": 5.1216902082834575e-06, + "loss": 0.3239, + "step": 13230 + }, + { + "epoch": 1.5661225455405725, + "grad_norm": 3.519055128097534, + "learning_rate": 5.120971989466124e-06, + "loss": 0.3244, + "step": 13240 + }, + { + "epoch": 1.5673054175538206, + "grad_norm": 3.2428009510040283, + "learning_rate": 5.120253770648791e-06, + "loss": 0.3556, + "step": 13250 + }, + { + "epoch": 1.5684882895670689, + "grad_norm": 2.028970956802368, + "learning_rate": 5.119535551831458e-06, + "loss": 0.3077, + "step": 13260 + }, + { + "epoch": 1.5696711615803172, + "grad_norm": 2.781470537185669, + "learning_rate": 5.118817333014125e-06, + "loss": 0.3255, + "step": 13270 + }, + { + "epoch": 1.570854033593565, + "grad_norm": 2.2510530948638916, + "learning_rate": 5.118099114196792e-06, + "loss": 0.3051, + "step": 13280 + }, + { + "epoch": 1.5720369056068133, + "grad_norm": 2.658205986022949, + "learning_rate": 5.117380895379459e-06, + "loss": 0.3221, + "step": 13290 + }, + { + "epoch": 1.5732197776200616, + "grad_norm": 2.8818469047546387, + "learning_rate": 5.116662676562126e-06, + "loss": 0.3675, + "step": 13300 + }, + { + "epoch": 1.5744026496333097, + "grad_norm": 2.4344351291656494, + "learning_rate": 5.115944457744793e-06, + "loss": 0.3007, + "step": 13310 + }, + { + "epoch": 1.5755855216465577, + "grad_norm": 3.1090714931488037, + "learning_rate": 5.11522623892746e-06, + "loss": 0.3178, + "step": 13320 + }, + { + "epoch": 1.576768393659806, + "grad_norm": 2.475126266479492, + "learning_rate": 5.114508020110127e-06, + "loss": 0.3233, + "step": 13330 + }, + { + "epoch": 1.5779512656730543, + "grad_norm": 2.812148094177246, + "learning_rate": 5.113789801292794e-06, + "loss": 0.3157, + "step": 13340 + }, + { + "epoch": 1.5791341376863024, + "grad_norm": 2.5128300189971924, + "learning_rate": 5.113071582475461e-06, + "loss": 0.3839, + "step": 13350 + }, + { + "epoch": 1.5803170096995505, + "grad_norm": 2.6277406215667725, + "learning_rate": 5.1123533636581275e-06, + "loss": 0.2921, + "step": 13360 + }, + { + "epoch": 1.5814998817127988, + "grad_norm": 2.604107141494751, + "learning_rate": 5.111635144840795e-06, + "loss": 0.3278, + "step": 13370 + }, + { + "epoch": 1.5826827537260468, + "grad_norm": 2.4044811725616455, + "learning_rate": 5.110916926023461e-06, + "loss": 0.3081, + "step": 13380 + }, + { + "epoch": 1.583865625739295, + "grad_norm": 3.4405109882354736, + "learning_rate": 5.110198707206129e-06, + "loss": 0.3289, + "step": 13390 + }, + { + "epoch": 1.5850484977525432, + "grad_norm": 2.8589260578155518, + "learning_rate": 5.109480488388796e-06, + "loss": 0.3478, + "step": 13400 + }, + { + "epoch": 1.5862313697657915, + "grad_norm": 2.8309414386749268, + "learning_rate": 5.108762269571463e-06, + "loss": 0.3041, + "step": 13410 + }, + { + "epoch": 1.5874142417790396, + "grad_norm": 3.194594383239746, + "learning_rate": 5.10804405075413e-06, + "loss": 0.3462, + "step": 13420 + }, + { + "epoch": 1.5885971137922876, + "grad_norm": 2.5025646686553955, + "learning_rate": 5.107325831936797e-06, + "loss": 0.3289, + "step": 13430 + }, + { + "epoch": 1.589779985805536, + "grad_norm": 2.4323976039886475, + "learning_rate": 5.106607613119464e-06, + "loss": 0.3008, + "step": 13440 + }, + { + "epoch": 1.590962857818784, + "grad_norm": 2.7668864727020264, + "learning_rate": 5.105889394302131e-06, + "loss": 0.3194, + "step": 13450 + }, + { + "epoch": 1.592145729832032, + "grad_norm": 2.426161289215088, + "learning_rate": 5.1051711754847985e-06, + "loss": 0.3612, + "step": 13460 + }, + { + "epoch": 1.5933286018452804, + "grad_norm": 2.4547650814056396, + "learning_rate": 5.1044529566674645e-06, + "loss": 0.2981, + "step": 13470 + }, + { + "epoch": 1.5945114738585286, + "grad_norm": 1.7698646783828735, + "learning_rate": 5.103734737850132e-06, + "loss": 0.3075, + "step": 13480 + }, + { + "epoch": 1.5956943458717767, + "grad_norm": 2.407890558242798, + "learning_rate": 5.103016519032798e-06, + "loss": 0.3229, + "step": 13490 + }, + { + "epoch": 1.5968772178850248, + "grad_norm": 3.103731393814087, + "learning_rate": 5.102298300215466e-06, + "loss": 0.3252, + "step": 13500 + }, + { + "epoch": 1.598060089898273, + "grad_norm": 2.5034029483795166, + "learning_rate": 5.101580081398132e-06, + "loss": 0.3042, + "step": 13510 + }, + { + "epoch": 1.5992429619115212, + "grad_norm": 3.7358813285827637, + "learning_rate": 5.1008618625808e-06, + "loss": 0.3194, + "step": 13520 + }, + { + "epoch": 1.6004258339247692, + "grad_norm": 2.2295784950256348, + "learning_rate": 5.100143643763467e-06, + "loss": 0.3506, + "step": 13530 + }, + { + "epoch": 1.6016087059380175, + "grad_norm": 2.52673077583313, + "learning_rate": 5.099425424946134e-06, + "loss": 0.3234, + "step": 13540 + }, + { + "epoch": 1.6027915779512658, + "grad_norm": 2.1552894115448, + "learning_rate": 5.098707206128801e-06, + "loss": 0.2655, + "step": 13550 + }, + { + "epoch": 1.6039744499645139, + "grad_norm": 2.6268768310546875, + "learning_rate": 5.097988987311468e-06, + "loss": 0.2887, + "step": 13560 + }, + { + "epoch": 1.605157321977762, + "grad_norm": 2.4877986907958984, + "learning_rate": 5.097270768494135e-06, + "loss": 0.3372, + "step": 13570 + }, + { + "epoch": 1.6063401939910102, + "grad_norm": 4.144022464752197, + "learning_rate": 5.0965525496768015e-06, + "loss": 0.3158, + "step": 13580 + }, + { + "epoch": 1.6075230660042583, + "grad_norm": 2.7924280166625977, + "learning_rate": 5.0958343308594685e-06, + "loss": 0.2999, + "step": 13590 + }, + { + "epoch": 1.6087059380175064, + "grad_norm": 2.384228229522705, + "learning_rate": 5.095116112042135e-06, + "loss": 0.3614, + "step": 13600 + }, + { + "epoch": 1.6098888100307547, + "grad_norm": 2.802626371383667, + "learning_rate": 5.094397893224802e-06, + "loss": 0.324, + "step": 13610 + }, + { + "epoch": 1.611071682044003, + "grad_norm": 2.819253921508789, + "learning_rate": 5.093679674407469e-06, + "loss": 0.3243, + "step": 13620 + }, + { + "epoch": 1.612254554057251, + "grad_norm": 3.2491488456726074, + "learning_rate": 5.092961455590136e-06, + "loss": 0.3314, + "step": 13630 + }, + { + "epoch": 1.6134374260704991, + "grad_norm": 3.445171356201172, + "learning_rate": 5.092243236772804e-06, + "loss": 0.3056, + "step": 13640 + }, + { + "epoch": 1.6146202980837474, + "grad_norm": 3.236532688140869, + "learning_rate": 5.091525017955471e-06, + "loss": 0.3026, + "step": 13650 + }, + { + "epoch": 1.6158031700969955, + "grad_norm": 2.369788646697998, + "learning_rate": 5.090806799138138e-06, + "loss": 0.3126, + "step": 13660 + }, + { + "epoch": 1.6169860421102435, + "grad_norm": 3.9594063758850098, + "learning_rate": 5.090088580320805e-06, + "loss": 0.3252, + "step": 13670 + }, + { + "epoch": 1.6181689141234918, + "grad_norm": 2.6396729946136475, + "learning_rate": 5.089370361503472e-06, + "loss": 0.3687, + "step": 13680 + }, + { + "epoch": 1.6193517861367401, + "grad_norm": 2.9766297340393066, + "learning_rate": 5.0886521426861385e-06, + "loss": 0.3366, + "step": 13690 + }, + { + "epoch": 1.6205346581499882, + "grad_norm": 2.8285045623779297, + "learning_rate": 5.0879339238688054e-06, + "loss": 0.3369, + "step": 13700 + }, + { + "epoch": 1.6217175301632363, + "grad_norm": 2.51643967628479, + "learning_rate": 5.087215705051472e-06, + "loss": 0.3449, + "step": 13710 + }, + { + "epoch": 1.6229004021764846, + "grad_norm": 2.7198190689086914, + "learning_rate": 5.086497486234139e-06, + "loss": 0.3007, + "step": 13720 + }, + { + "epoch": 1.6240832741897326, + "grad_norm": 2.0131969451904297, + "learning_rate": 5.085779267416807e-06, + "loss": 0.3598, + "step": 13730 + }, + { + "epoch": 1.6252661462029807, + "grad_norm": 2.2213804721832275, + "learning_rate": 5.085061048599473e-06, + "loss": 0.3138, + "step": 13740 + }, + { + "epoch": 1.626449018216229, + "grad_norm": 2.639836311340332, + "learning_rate": 5.084342829782141e-06, + "loss": 0.3667, + "step": 13750 + }, + { + "epoch": 1.6276318902294773, + "grad_norm": 2.830441474914551, + "learning_rate": 5.083624610964807e-06, + "loss": 0.3059, + "step": 13760 + }, + { + "epoch": 1.6288147622427254, + "grad_norm": 3.0879757404327393, + "learning_rate": 5.082906392147475e-06, + "loss": 0.338, + "step": 13770 + }, + { + "epoch": 1.6299976342559734, + "grad_norm": 3.0475282669067383, + "learning_rate": 5.082188173330141e-06, + "loss": 0.3497, + "step": 13780 + }, + { + "epoch": 1.6311805062692217, + "grad_norm": 2.055182695388794, + "learning_rate": 5.081469954512809e-06, + "loss": 0.3156, + "step": 13790 + }, + { + "epoch": 1.6323633782824698, + "grad_norm": 2.2830474376678467, + "learning_rate": 5.0807517356954755e-06, + "loss": 0.3428, + "step": 13800 + }, + { + "epoch": 1.6335462502957179, + "grad_norm": 3.2028403282165527, + "learning_rate": 5.0800335168781424e-06, + "loss": 0.2927, + "step": 13810 + }, + { + "epoch": 1.6347291223089662, + "grad_norm": 3.4929287433624268, + "learning_rate": 5.079315298060809e-06, + "loss": 0.3376, + "step": 13820 + }, + { + "epoch": 1.6359119943222145, + "grad_norm": 2.8479928970336914, + "learning_rate": 5.078597079243476e-06, + "loss": 0.3134, + "step": 13830 + }, + { + "epoch": 1.6370948663354625, + "grad_norm": 1.9470689296722412, + "learning_rate": 5.077878860426143e-06, + "loss": 0.3183, + "step": 13840 + }, + { + "epoch": 1.6382777383487106, + "grad_norm": 2.4081637859344482, + "learning_rate": 5.07716064160881e-06, + "loss": 0.3391, + "step": 13850 + }, + { + "epoch": 1.639460610361959, + "grad_norm": 2.22745943069458, + "learning_rate": 5.076442422791477e-06, + "loss": 0.3205, + "step": 13860 + }, + { + "epoch": 1.640643482375207, + "grad_norm": 2.6729397773742676, + "learning_rate": 5.075724203974144e-06, + "loss": 0.324, + "step": 13870 + }, + { + "epoch": 1.641826354388455, + "grad_norm": 2.435185432434082, + "learning_rate": 5.075005985156811e-06, + "loss": 0.3658, + "step": 13880 + }, + { + "epoch": 1.6430092264017033, + "grad_norm": 2.6831552982330322, + "learning_rate": 5.074287766339478e-06, + "loss": 0.2849, + "step": 13890 + }, + { + "epoch": 1.6441920984149516, + "grad_norm": 2.8662123680114746, + "learning_rate": 5.073569547522146e-06, + "loss": 0.3481, + "step": 13900 + }, + { + "epoch": 1.6453749704281997, + "grad_norm": 2.2142205238342285, + "learning_rate": 5.0728513287048125e-06, + "loss": 0.3413, + "step": 13910 + }, + { + "epoch": 1.6465578424414478, + "grad_norm": 3.125593423843384, + "learning_rate": 5.0721331098874794e-06, + "loss": 0.3057, + "step": 13920 + }, + { + "epoch": 1.647740714454696, + "grad_norm": 3.180004835128784, + "learning_rate": 5.071414891070146e-06, + "loss": 0.3383, + "step": 13930 + }, + { + "epoch": 1.6489235864679441, + "grad_norm": 3.923089027404785, + "learning_rate": 5.070696672252813e-06, + "loss": 0.3122, + "step": 13940 + }, + { + "epoch": 1.6501064584811922, + "grad_norm": 2.646047592163086, + "learning_rate": 5.06997845343548e-06, + "loss": 0.3103, + "step": 13950 + }, + { + "epoch": 1.6512893304944405, + "grad_norm": 3.303234100341797, + "learning_rate": 5.069260234618147e-06, + "loss": 0.3356, + "step": 13960 + }, + { + "epoch": 1.6524722025076888, + "grad_norm": 1.9866607189178467, + "learning_rate": 5.068542015800814e-06, + "loss": 0.3274, + "step": 13970 + }, + { + "epoch": 1.6536550745209369, + "grad_norm": 2.8178138732910156, + "learning_rate": 5.067823796983481e-06, + "loss": 0.3697, + "step": 13980 + }, + { + "epoch": 1.654837946534185, + "grad_norm": 1.9609490633010864, + "learning_rate": 5.067105578166148e-06, + "loss": 0.3062, + "step": 13990 + }, + { + "epoch": 1.6560208185474332, + "grad_norm": 2.2423834800720215, + "learning_rate": 5.066387359348816e-06, + "loss": 0.3148, + "step": 14000 + }, + { + "epoch": 1.6572036905606813, + "grad_norm": 3.6977040767669678, + "learning_rate": 5.065669140531482e-06, + "loss": 0.2901, + "step": 14010 + }, + { + "epoch": 1.6583865625739294, + "grad_norm": 3.108161449432373, + "learning_rate": 5.0649509217141495e-06, + "loss": 0.3094, + "step": 14020 + }, + { + "epoch": 1.6595694345871776, + "grad_norm": 3.174072027206421, + "learning_rate": 5.064232702896816e-06, + "loss": 0.336, + "step": 14030 + }, + { + "epoch": 1.660752306600426, + "grad_norm": 3.211223602294922, + "learning_rate": 5.063514484079483e-06, + "loss": 0.284, + "step": 14040 + }, + { + "epoch": 1.661935178613674, + "grad_norm": 2.340867757797241, + "learning_rate": 5.0627962652621494e-06, + "loss": 0.3352, + "step": 14050 + }, + { + "epoch": 1.663118050626922, + "grad_norm": 2.3574111461639404, + "learning_rate": 5.062078046444817e-06, + "loss": 0.3324, + "step": 14060 + }, + { + "epoch": 1.6643009226401704, + "grad_norm": 3.4371180534362793, + "learning_rate": 5.061359827627484e-06, + "loss": 0.3469, + "step": 14070 + }, + { + "epoch": 1.6654837946534184, + "grad_norm": 2.4418962001800537, + "learning_rate": 5.060641608810151e-06, + "loss": 0.3366, + "step": 14080 + }, + { + "epoch": 1.6666666666666665, + "grad_norm": 3.1625897884368896, + "learning_rate": 5.059923389992818e-06, + "loss": 0.309, + "step": 14090 + }, + { + "epoch": 1.6678495386799148, + "grad_norm": 2.799833297729492, + "learning_rate": 5.059205171175485e-06, + "loss": 0.3117, + "step": 14100 + }, + { + "epoch": 1.669032410693163, + "grad_norm": 2.8320069313049316, + "learning_rate": 5.058486952358152e-06, + "loss": 0.329, + "step": 14110 + }, + { + "epoch": 1.6702152827064112, + "grad_norm": 2.614542007446289, + "learning_rate": 5.057768733540819e-06, + "loss": 0.322, + "step": 14120 + }, + { + "epoch": 1.6713981547196592, + "grad_norm": 3.401003122329712, + "learning_rate": 5.057050514723486e-06, + "loss": 0.2921, + "step": 14130 + }, + { + "epoch": 1.6725810267329075, + "grad_norm": 2.30155086517334, + "learning_rate": 5.056332295906153e-06, + "loss": 0.331, + "step": 14140 + }, + { + "epoch": 1.6737638987461558, + "grad_norm": 2.7106785774230957, + "learning_rate": 5.05561407708882e-06, + "loss": 0.3345, + "step": 14150 + }, + { + "epoch": 1.6749467707594037, + "grad_norm": 2.9260377883911133, + "learning_rate": 5.0548958582714864e-06, + "loss": 0.3113, + "step": 14160 + }, + { + "epoch": 1.676129642772652, + "grad_norm": 2.626962423324585, + "learning_rate": 5.054177639454154e-06, + "loss": 0.3409, + "step": 14170 + }, + { + "epoch": 1.6773125147859003, + "grad_norm": 2.456526517868042, + "learning_rate": 5.053459420636821e-06, + "loss": 0.3169, + "step": 14180 + }, + { + "epoch": 1.6784953867991483, + "grad_norm": 2.4258885383605957, + "learning_rate": 5.052741201819488e-06, + "loss": 0.3114, + "step": 14190 + }, + { + "epoch": 1.6796782588123964, + "grad_norm": 3.0785810947418213, + "learning_rate": 5.052022983002155e-06, + "loss": 0.2927, + "step": 14200 + }, + { + "epoch": 1.6808611308256447, + "grad_norm": 2.3288445472717285, + "learning_rate": 5.051304764184822e-06, + "loss": 0.3244, + "step": 14210 + }, + { + "epoch": 1.682044002838893, + "grad_norm": 2.9143729209899902, + "learning_rate": 5.050586545367489e-06, + "loss": 0.3313, + "step": 14220 + }, + { + "epoch": 1.6832268748521408, + "grad_norm": 2.184540033340454, + "learning_rate": 5.049868326550156e-06, + "loss": 0.3088, + "step": 14230 + }, + { + "epoch": 1.6844097468653891, + "grad_norm": 2.156137228012085, + "learning_rate": 5.049150107732823e-06, + "loss": 0.3172, + "step": 14240 + }, + { + "epoch": 1.6855926188786374, + "grad_norm": 2.418489694595337, + "learning_rate": 5.04843188891549e-06, + "loss": 0.3047, + "step": 14250 + }, + { + "epoch": 1.6867754908918855, + "grad_norm": 3.1165213584899902, + "learning_rate": 5.0477136700981565e-06, + "loss": 0.2754, + "step": 14260 + }, + { + "epoch": 1.6879583629051336, + "grad_norm": 2.4814577102661133, + "learning_rate": 5.0469954512808234e-06, + "loss": 0.3371, + "step": 14270 + }, + { + "epoch": 1.6891412349183819, + "grad_norm": 3.270131826400757, + "learning_rate": 5.04627723246349e-06, + "loss": 0.2998, + "step": 14280 + }, + { + "epoch": 1.6903241069316302, + "grad_norm": 2.3310813903808594, + "learning_rate": 5.045559013646158e-06, + "loss": 0.2948, + "step": 14290 + }, + { + "epoch": 1.691506978944878, + "grad_norm": 3.6687893867492676, + "learning_rate": 5.044840794828824e-06, + "loss": 0.3071, + "step": 14300 + }, + { + "epoch": 1.6926898509581263, + "grad_norm": 2.5189130306243896, + "learning_rate": 5.044122576011492e-06, + "loss": 0.2991, + "step": 14310 + }, + { + "epoch": 1.6938727229713746, + "grad_norm": 2.2292873859405518, + "learning_rate": 5.043404357194158e-06, + "loss": 0.3645, + "step": 14320 + }, + { + "epoch": 1.6950555949846227, + "grad_norm": 2.4766671657562256, + "learning_rate": 5.042686138376826e-06, + "loss": 0.3062, + "step": 14330 + }, + { + "epoch": 1.6962384669978707, + "grad_norm": 3.1414923667907715, + "learning_rate": 5.041967919559493e-06, + "loss": 0.3216, + "step": 14340 + }, + { + "epoch": 1.697421339011119, + "grad_norm": 2.6457090377807617, + "learning_rate": 5.04124970074216e-06, + "loss": 0.3048, + "step": 14350 + }, + { + "epoch": 1.6986042110243673, + "grad_norm": 1.8806536197662354, + "learning_rate": 5.040531481924827e-06, + "loss": 0.3906, + "step": 14360 + }, + { + "epoch": 1.6997870830376152, + "grad_norm": 2.4284303188323975, + "learning_rate": 5.0398132631074935e-06, + "loss": 0.3135, + "step": 14370 + }, + { + "epoch": 1.7009699550508635, + "grad_norm": 2.1625561714172363, + "learning_rate": 5.0390950442901604e-06, + "loss": 0.3359, + "step": 14380 + }, + { + "epoch": 1.7021528270641118, + "grad_norm": 2.284396171569824, + "learning_rate": 5.038376825472827e-06, + "loss": 0.3306, + "step": 14390 + }, + { + "epoch": 1.7033356990773598, + "grad_norm": 2.887921094894409, + "learning_rate": 5.037658606655495e-06, + "loss": 0.3091, + "step": 14400 + }, + { + "epoch": 1.704518571090608, + "grad_norm": 3.903743267059326, + "learning_rate": 5.036940387838161e-06, + "loss": 0.3783, + "step": 14410 + }, + { + "epoch": 1.7057014431038562, + "grad_norm": 3.2908666133880615, + "learning_rate": 5.036222169020829e-06, + "loss": 0.3208, + "step": 14420 + }, + { + "epoch": 1.7068843151171045, + "grad_norm": 2.6162149906158447, + "learning_rate": 5.035503950203495e-06, + "loss": 0.3251, + "step": 14430 + }, + { + "epoch": 1.7080671871303525, + "grad_norm": 2.951775074005127, + "learning_rate": 5.034785731386163e-06, + "loss": 0.3457, + "step": 14440 + }, + { + "epoch": 1.7092500591436006, + "grad_norm": 2.26690411567688, + "learning_rate": 5.03406751256883e-06, + "loss": 0.3251, + "step": 14450 + }, + { + "epoch": 1.710432931156849, + "grad_norm": 3.271415948867798, + "learning_rate": 5.033349293751497e-06, + "loss": 0.3411, + "step": 14460 + }, + { + "epoch": 1.711615803170097, + "grad_norm": 2.0259809494018555, + "learning_rate": 5.032631074934164e-06, + "loss": 0.3225, + "step": 14470 + }, + { + "epoch": 1.712798675183345, + "grad_norm": 2.0512285232543945, + "learning_rate": 5.0319128561168305e-06, + "loss": 0.2946, + "step": 14480 + }, + { + "epoch": 1.7139815471965933, + "grad_norm": 2.5374696254730225, + "learning_rate": 5.0311946372994974e-06, + "loss": 0.3411, + "step": 14490 + }, + { + "epoch": 1.7151644192098416, + "grad_norm": 2.3170814514160156, + "learning_rate": 5.030476418482164e-06, + "loss": 0.3025, + "step": 14500 + }, + { + "epoch": 1.7163472912230897, + "grad_norm": 2.375981330871582, + "learning_rate": 5.029758199664831e-06, + "loss": 0.2748, + "step": 14510 + }, + { + "epoch": 1.7175301632363378, + "grad_norm": 3.1739301681518555, + "learning_rate": 5.029039980847498e-06, + "loss": 0.312, + "step": 14520 + }, + { + "epoch": 1.718713035249586, + "grad_norm": 2.581071138381958, + "learning_rate": 5.028321762030165e-06, + "loss": 0.3039, + "step": 14530 + }, + { + "epoch": 1.7198959072628341, + "grad_norm": 2.506803274154663, + "learning_rate": 5.027603543212832e-06, + "loss": 0.3452, + "step": 14540 + }, + { + "epoch": 1.7210787792760822, + "grad_norm": 3.3117330074310303, + "learning_rate": 5.026885324395499e-06, + "loss": 0.3663, + "step": 14550 + }, + { + "epoch": 1.7222616512893305, + "grad_norm": 2.346388816833496, + "learning_rate": 5.026167105578167e-06, + "loss": 0.354, + "step": 14560 + }, + { + "epoch": 1.7234445233025788, + "grad_norm": 2.311544418334961, + "learning_rate": 5.025448886760833e-06, + "loss": 0.3132, + "step": 14570 + }, + { + "epoch": 1.7246273953158269, + "grad_norm": 2.3980321884155273, + "learning_rate": 5.024730667943501e-06, + "loss": 0.2823, + "step": 14580 + }, + { + "epoch": 1.725810267329075, + "grad_norm": 2.5401647090911865, + "learning_rate": 5.024012449126167e-06, + "loss": 0.2922, + "step": 14590 + }, + { + "epoch": 1.7269931393423232, + "grad_norm": 2.3093924522399902, + "learning_rate": 5.0232942303088344e-06, + "loss": 0.2924, + "step": 14600 + }, + { + "epoch": 1.7281760113555713, + "grad_norm": 2.8203935623168945, + "learning_rate": 5.0225760114915005e-06, + "loss": 0.3306, + "step": 14610 + }, + { + "epoch": 1.7293588833688194, + "grad_norm": 3.4530975818634033, + "learning_rate": 5.021857792674168e-06, + "loss": 0.3344, + "step": 14620 + }, + { + "epoch": 1.7305417553820677, + "grad_norm": 3.3713724613189697, + "learning_rate": 5.021139573856835e-06, + "loss": 0.3436, + "step": 14630 + }, + { + "epoch": 1.731724627395316, + "grad_norm": 2.5409672260284424, + "learning_rate": 5.020421355039502e-06, + "loss": 0.3749, + "step": 14640 + }, + { + "epoch": 1.732907499408564, + "grad_norm": 4.277678966522217, + "learning_rate": 5.01970313622217e-06, + "loss": 0.3052, + "step": 14650 + }, + { + "epoch": 1.734090371421812, + "grad_norm": 3.7035672664642334, + "learning_rate": 5.018984917404836e-06, + "loss": 0.3081, + "step": 14660 + }, + { + "epoch": 1.7352732434350604, + "grad_norm": 2.3813202381134033, + "learning_rate": 5.018266698587504e-06, + "loss": 0.3127, + "step": 14670 + }, + { + "epoch": 1.7364561154483085, + "grad_norm": 2.580883741378784, + "learning_rate": 5.01754847977017e-06, + "loss": 0.2905, + "step": 14680 + }, + { + "epoch": 1.7376389874615565, + "grad_norm": 2.821316957473755, + "learning_rate": 5.016830260952838e-06, + "loss": 0.3214, + "step": 14690 + }, + { + "epoch": 1.7388218594748048, + "grad_norm": 3.987694025039673, + "learning_rate": 5.016112042135504e-06, + "loss": 0.2954, + "step": 14700 + }, + { + "epoch": 1.7400047314880531, + "grad_norm": 3.289832830429077, + "learning_rate": 5.0153938233181714e-06, + "loss": 0.329, + "step": 14710 + }, + { + "epoch": 1.7411876035013012, + "grad_norm": 2.1757147312164307, + "learning_rate": 5.014675604500838e-06, + "loss": 0.3071, + "step": 14720 + }, + { + "epoch": 1.7423704755145493, + "grad_norm": 2.166836977005005, + "learning_rate": 5.013957385683505e-06, + "loss": 0.3247, + "step": 14730 + }, + { + "epoch": 1.7435533475277976, + "grad_norm": 3.105241298675537, + "learning_rate": 5.013239166866172e-06, + "loss": 0.2866, + "step": 14740 + }, + { + "epoch": 1.7447362195410456, + "grad_norm": 2.3274075984954834, + "learning_rate": 5.012520948048839e-06, + "loss": 0.3615, + "step": 14750 + }, + { + "epoch": 1.7459190915542937, + "grad_norm": 2.7024152278900146, + "learning_rate": 5.011802729231506e-06, + "loss": 0.3349, + "step": 14760 + }, + { + "epoch": 1.747101963567542, + "grad_norm": 2.497018575668335, + "learning_rate": 5.011084510414173e-06, + "loss": 0.2949, + "step": 14770 + }, + { + "epoch": 1.7482848355807903, + "grad_norm": 3.0676562786102295, + "learning_rate": 5.01036629159684e-06, + "loss": 0.3364, + "step": 14780 + }, + { + "epoch": 1.7494677075940384, + "grad_norm": 3.283435583114624, + "learning_rate": 5.009648072779507e-06, + "loss": 0.2902, + "step": 14790 + }, + { + "epoch": 1.7504140052046369, + "eval_accuracy": 0.8563562564460858, + "eval_loss": 0.3321095407009125, + "eval_runtime": 77.8043, + "eval_safe_aucpr": 0.912319127206396, + "eval_safe_f1": 0.8371891321153158, + "eval_safe_fpr": 0.12474963379271135, + "eval_safe_precision": 0.841775991506787, + "eval_safe_recall": 0.8326519896485767, + "eval_samples_per_second": 772.631, + "eval_steps_per_second": 12.082, + "eval_unsafe_aucpr": 0.9509520176742567, + "eval_unsafe_f1": 0.8714857644625024, + "eval_unsafe_fpr": 0.16734801035142272, + "eval_unsafe_precision": 0.8677534084173089, + "eval_unsafe_recall": 0.8752503662072882, + "step": 14798 + }, + { + "epoch": 1.7506505796072864, + "grad_norm": 2.357008695602417, + "learning_rate": 5.008929853962174e-06, + "loss": 0.2962, + "step": 14800 + }, + { + "epoch": 1.7518334516205347, + "grad_norm": 3.520799160003662, + "learning_rate": 5.008211635144841e-06, + "loss": 0.3438, + "step": 14810 + }, + { + "epoch": 1.7530163236337828, + "grad_norm": 2.4718270301818848, + "learning_rate": 5.007493416327508e-06, + "loss": 0.2888, + "step": 14820 + }, + { + "epoch": 1.7541991956470309, + "grad_norm": 2.4665021896362305, + "learning_rate": 5.006775197510175e-06, + "loss": 0.3219, + "step": 14830 + }, + { + "epoch": 1.7553820676602792, + "grad_norm": 3.32997727394104, + "learning_rate": 5.006056978692841e-06, + "loss": 0.3331, + "step": 14840 + }, + { + "epoch": 1.7565649396735274, + "grad_norm": 2.663910388946533, + "learning_rate": 5.005338759875509e-06, + "loss": 0.3391, + "step": 14850 + }, + { + "epoch": 1.7577478116867755, + "grad_norm": 2.1464967727661133, + "learning_rate": 5.004620541058175e-06, + "loss": 0.2876, + "step": 14860 + }, + { + "epoch": 1.7589306837000236, + "grad_norm": 2.893955945968628, + "learning_rate": 5.003902322240843e-06, + "loss": 0.333, + "step": 14870 + }, + { + "epoch": 1.7601135557132719, + "grad_norm": 2.531080722808838, + "learning_rate": 5.003184103423509e-06, + "loss": 0.318, + "step": 14880 + }, + { + "epoch": 1.76129642772652, + "grad_norm": 2.4850759506225586, + "learning_rate": 5.002465884606177e-06, + "loss": 0.3414, + "step": 14890 + }, + { + "epoch": 1.762479299739768, + "grad_norm": 3.054048776626587, + "learning_rate": 5.001747665788844e-06, + "loss": 0.3491, + "step": 14900 + }, + { + "epoch": 1.7636621717530163, + "grad_norm": 2.300985097885132, + "learning_rate": 5.001029446971511e-06, + "loss": 0.3134, + "step": 14910 + }, + { + "epoch": 1.7648450437662646, + "grad_norm": 2.196075677871704, + "learning_rate": 5.000311228154178e-06, + "loss": 0.3487, + "step": 14920 + }, + { + "epoch": 1.7660279157795127, + "grad_norm": 2.2579915523529053, + "learning_rate": 4.9995930093368446e-06, + "loss": 0.2823, + "step": 14930 + }, + { + "epoch": 1.7672107877927608, + "grad_norm": 3.092942953109741, + "learning_rate": 4.998874790519512e-06, + "loss": 0.3153, + "step": 14940 + }, + { + "epoch": 1.768393659806009, + "grad_norm": 4.024479866027832, + "learning_rate": 4.998156571702178e-06, + "loss": 0.3484, + "step": 14950 + }, + { + "epoch": 1.7695765318192571, + "grad_norm": 2.4754703044891357, + "learning_rate": 4.997438352884846e-06, + "loss": 0.3691, + "step": 14960 + }, + { + "epoch": 1.7707594038325052, + "grad_norm": 2.749239444732666, + "learning_rate": 4.996720134067512e-06, + "loss": 0.3407, + "step": 14970 + }, + { + "epoch": 1.7719422758457535, + "grad_norm": 1.894676923751831, + "learning_rate": 4.99600191525018e-06, + "loss": 0.312, + "step": 14980 + }, + { + "epoch": 1.7731251478590018, + "grad_norm": 3.5794310569763184, + "learning_rate": 4.995283696432847e-06, + "loss": 0.3488, + "step": 14990 + }, + { + "epoch": 1.7743080198722498, + "grad_norm": 3.6035690307617188, + "learning_rate": 4.994565477615514e-06, + "loss": 0.3257, + "step": 15000 + }, + { + "epoch": 1.775490891885498, + "grad_norm": 1.712687373161316, + "learning_rate": 4.993847258798181e-06, + "loss": 0.2978, + "step": 15010 + }, + { + "epoch": 1.7766737638987462, + "grad_norm": 2.74722957611084, + "learning_rate": 4.993129039980848e-06, + "loss": 0.3155, + "step": 15020 + }, + { + "epoch": 1.7778566359119943, + "grad_norm": 2.21646785736084, + "learning_rate": 4.992410821163515e-06, + "loss": 0.3277, + "step": 15030 + }, + { + "epoch": 1.7790395079252423, + "grad_norm": 2.4206202030181885, + "learning_rate": 4.9916926023461816e-06, + "loss": 0.2708, + "step": 15040 + }, + { + "epoch": 1.7802223799384906, + "grad_norm": 3.1641170978546143, + "learning_rate": 4.9909743835288485e-06, + "loss": 0.3207, + "step": 15050 + }, + { + "epoch": 1.781405251951739, + "grad_norm": 2.7150776386260986, + "learning_rate": 4.990256164711515e-06, + "loss": 0.3354, + "step": 15060 + }, + { + "epoch": 1.782588123964987, + "grad_norm": 2.523766040802002, + "learning_rate": 4.989537945894182e-06, + "loss": 0.3017, + "step": 15070 + }, + { + "epoch": 1.783770995978235, + "grad_norm": 2.2823326587677, + "learning_rate": 4.988819727076849e-06, + "loss": 0.3233, + "step": 15080 + }, + { + "epoch": 1.7849538679914834, + "grad_norm": 4.041593074798584, + "learning_rate": 4.988101508259516e-06, + "loss": 0.3148, + "step": 15090 + }, + { + "epoch": 1.7861367400047314, + "grad_norm": 2.364863395690918, + "learning_rate": 4.987383289442184e-06, + "loss": 0.2999, + "step": 15100 + }, + { + "epoch": 1.7873196120179795, + "grad_norm": 4.033505916595459, + "learning_rate": 4.98666507062485e-06, + "loss": 0.3229, + "step": 15110 + }, + { + "epoch": 1.7885024840312278, + "grad_norm": 2.4584884643554688, + "learning_rate": 4.985946851807518e-06, + "loss": 0.3095, + "step": 15120 + }, + { + "epoch": 1.789685356044476, + "grad_norm": 1.9042985439300537, + "learning_rate": 4.985228632990184e-06, + "loss": 0.2744, + "step": 15130 + }, + { + "epoch": 1.7908682280577242, + "grad_norm": 3.170943021774292, + "learning_rate": 4.984510414172852e-06, + "loss": 0.3133, + "step": 15140 + }, + { + "epoch": 1.7920511000709722, + "grad_norm": 2.462503671646118, + "learning_rate": 4.983792195355518e-06, + "loss": 0.3562, + "step": 15150 + }, + { + "epoch": 1.7932339720842205, + "grad_norm": 2.148439407348633, + "learning_rate": 4.9830739765381855e-06, + "loss": 0.3529, + "step": 15160 + }, + { + "epoch": 1.7944168440974686, + "grad_norm": 2.2401418685913086, + "learning_rate": 4.982355757720852e-06, + "loss": 0.3314, + "step": 15170 + }, + { + "epoch": 1.7955997161107167, + "grad_norm": 2.526993989944458, + "learning_rate": 4.981637538903519e-06, + "loss": 0.3091, + "step": 15180 + }, + { + "epoch": 1.796782588123965, + "grad_norm": 4.488979816436768, + "learning_rate": 4.980919320086186e-06, + "loss": 0.3199, + "step": 15190 + }, + { + "epoch": 1.7979654601372133, + "grad_norm": 2.7945244312286377, + "learning_rate": 4.980201101268853e-06, + "loss": 0.3415, + "step": 15200 + }, + { + "epoch": 1.7991483321504613, + "grad_norm": 2.6339950561523438, + "learning_rate": 4.979482882451521e-06, + "loss": 0.3692, + "step": 15210 + }, + { + "epoch": 1.8003312041637094, + "grad_norm": 3.3200294971466064, + "learning_rate": 4.978764663634187e-06, + "loss": 0.3222, + "step": 15220 + }, + { + "epoch": 1.8015140761769577, + "grad_norm": 2.7311697006225586, + "learning_rate": 4.978046444816855e-06, + "loss": 0.3199, + "step": 15230 + }, + { + "epoch": 1.802696948190206, + "grad_norm": 2.4543228149414062, + "learning_rate": 4.977328225999521e-06, + "loss": 0.2966, + "step": 15240 + }, + { + "epoch": 1.8038798202034538, + "grad_norm": 2.3450186252593994, + "learning_rate": 4.976610007182189e-06, + "loss": 0.3515, + "step": 15250 + }, + { + "epoch": 1.8050626922167021, + "grad_norm": 3.66555118560791, + "learning_rate": 4.975891788364855e-06, + "loss": 0.324, + "step": 15260 + }, + { + "epoch": 1.8062455642299504, + "grad_norm": 2.4101505279541016, + "learning_rate": 4.9751735695475225e-06, + "loss": 0.2979, + "step": 15270 + }, + { + "epoch": 1.8074284362431985, + "grad_norm": 2.7355422973632812, + "learning_rate": 4.974455350730189e-06, + "loss": 0.3147, + "step": 15280 + }, + { + "epoch": 1.8086113082564466, + "grad_norm": 3.208298444747925, + "learning_rate": 4.973737131912856e-06, + "loss": 0.3688, + "step": 15290 + }, + { + "epoch": 1.8097941802696949, + "grad_norm": 2.7905843257904053, + "learning_rate": 4.973018913095523e-06, + "loss": 0.3597, + "step": 15300 + }, + { + "epoch": 1.8109770522829431, + "grad_norm": 2.308422565460205, + "learning_rate": 4.97230069427819e-06, + "loss": 0.3251, + "step": 15310 + }, + { + "epoch": 1.812159924296191, + "grad_norm": 3.4839212894439697, + "learning_rate": 4.971582475460857e-06, + "loss": 0.3373, + "step": 15320 + }, + { + "epoch": 1.8133427963094393, + "grad_norm": 2.911137819290161, + "learning_rate": 4.970864256643524e-06, + "loss": 0.3159, + "step": 15330 + }, + { + "epoch": 1.8145256683226876, + "grad_norm": 2.1877129077911377, + "learning_rate": 4.970146037826191e-06, + "loss": 0.2728, + "step": 15340 + }, + { + "epoch": 1.8157085403359357, + "grad_norm": 2.5772745609283447, + "learning_rate": 4.969427819008858e-06, + "loss": 0.295, + "step": 15350 + }, + { + "epoch": 1.8168914123491837, + "grad_norm": 2.590778112411499, + "learning_rate": 4.968709600191525e-06, + "loss": 0.3209, + "step": 15360 + }, + { + "epoch": 1.818074284362432, + "grad_norm": 2.6555583477020264, + "learning_rate": 4.9679913813741926e-06, + "loss": 0.294, + "step": 15370 + }, + { + "epoch": 1.8192571563756803, + "grad_norm": 2.4802496433258057, + "learning_rate": 4.967273162556859e-06, + "loss": 0.3685, + "step": 15380 + }, + { + "epoch": 1.8204400283889282, + "grad_norm": 2.965207815170288, + "learning_rate": 4.966554943739526e-06, + "loss": 0.3094, + "step": 15390 + }, + { + "epoch": 1.8216229004021764, + "grad_norm": 2.8850719928741455, + "learning_rate": 4.9658367249221925e-06, + "loss": 0.2854, + "step": 15400 + }, + { + "epoch": 1.8228057724154247, + "grad_norm": 2.8029391765594482, + "learning_rate": 4.96511850610486e-06, + "loss": 0.3411, + "step": 15410 + }, + { + "epoch": 1.8239886444286728, + "grad_norm": 2.1750428676605225, + "learning_rate": 4.964400287287527e-06, + "loss": 0.3108, + "step": 15420 + }, + { + "epoch": 1.8251715164419209, + "grad_norm": 1.994527816772461, + "learning_rate": 4.963682068470194e-06, + "loss": 0.3013, + "step": 15430 + }, + { + "epoch": 1.8263543884551692, + "grad_norm": 2.478555917739868, + "learning_rate": 4.962963849652861e-06, + "loss": 0.2966, + "step": 15440 + }, + { + "epoch": 1.8275372604684175, + "grad_norm": 2.7759952545166016, + "learning_rate": 4.962245630835528e-06, + "loss": 0.3379, + "step": 15450 + }, + { + "epoch": 1.8287201324816655, + "grad_norm": 2.4554290771484375, + "learning_rate": 4.961527412018195e-06, + "loss": 0.2662, + "step": 15460 + }, + { + "epoch": 1.8299030044949136, + "grad_norm": 2.1369402408599854, + "learning_rate": 4.960809193200862e-06, + "loss": 0.2674, + "step": 15470 + }, + { + "epoch": 1.831085876508162, + "grad_norm": 3.9057397842407227, + "learning_rate": 4.9600909743835296e-06, + "loss": 0.3578, + "step": 15480 + }, + { + "epoch": 1.83226874852141, + "grad_norm": 2.922919988632202, + "learning_rate": 4.959372755566196e-06, + "loss": 0.3491, + "step": 15490 + }, + { + "epoch": 1.833451620534658, + "grad_norm": 3.4353342056274414, + "learning_rate": 4.958654536748863e-06, + "loss": 0.3066, + "step": 15500 + }, + { + "epoch": 1.8346344925479063, + "grad_norm": 3.313187837600708, + "learning_rate": 4.9579363179315295e-06, + "loss": 0.3208, + "step": 15510 + }, + { + "epoch": 1.8358173645611546, + "grad_norm": 3.00732159614563, + "learning_rate": 4.957218099114197e-06, + "loss": 0.3246, + "step": 15520 + }, + { + "epoch": 1.8370002365744027, + "grad_norm": 1.9314966201782227, + "learning_rate": 4.956499880296863e-06, + "loss": 0.2884, + "step": 15530 + }, + { + "epoch": 1.8381831085876508, + "grad_norm": 1.8576322793960571, + "learning_rate": 4.955781661479531e-06, + "loss": 0.3116, + "step": 15540 + }, + { + "epoch": 1.839365980600899, + "grad_norm": 2.75687313079834, + "learning_rate": 4.955063442662198e-06, + "loss": 0.3189, + "step": 15550 + }, + { + "epoch": 1.8405488526141471, + "grad_norm": 2.6360726356506348, + "learning_rate": 4.954345223844865e-06, + "loss": 0.3422, + "step": 15560 + }, + { + "epoch": 1.8417317246273952, + "grad_norm": 3.248277425765991, + "learning_rate": 4.953627005027532e-06, + "loss": 0.3313, + "step": 15570 + }, + { + "epoch": 1.8429145966406435, + "grad_norm": 2.726261615753174, + "learning_rate": 4.952908786210199e-06, + "loss": 0.3268, + "step": 15580 + }, + { + "epoch": 1.8440974686538918, + "grad_norm": 2.659207344055176, + "learning_rate": 4.952190567392866e-06, + "loss": 0.302, + "step": 15590 + }, + { + "epoch": 1.8452803406671399, + "grad_norm": 2.1712844371795654, + "learning_rate": 4.951472348575533e-06, + "loss": 0.3303, + "step": 15600 + }, + { + "epoch": 1.846463212680388, + "grad_norm": 2.3552823066711426, + "learning_rate": 4.9507541297581996e-06, + "loss": 0.3412, + "step": 15610 + }, + { + "epoch": 1.8476460846936362, + "grad_norm": 2.0031118392944336, + "learning_rate": 4.9500359109408665e-06, + "loss": 0.3161, + "step": 15620 + }, + { + "epoch": 1.8488289567068843, + "grad_norm": 2.548931360244751, + "learning_rate": 4.949317692123533e-06, + "loss": 0.3335, + "step": 15630 + }, + { + "epoch": 1.8500118287201324, + "grad_norm": 3.401005983352661, + "learning_rate": 4.948599473306201e-06, + "loss": 0.3518, + "step": 15640 + }, + { + "epoch": 1.8511947007333807, + "grad_norm": 2.6640453338623047, + "learning_rate": 4.947881254488867e-06, + "loss": 0.3158, + "step": 15650 + }, + { + "epoch": 1.852377572746629, + "grad_norm": 1.7123150825500488, + "learning_rate": 4.947163035671535e-06, + "loss": 0.297, + "step": 15660 + }, + { + "epoch": 1.853560444759877, + "grad_norm": 2.112438201904297, + "learning_rate": 4.946444816854202e-06, + "loss": 0.3326, + "step": 15670 + }, + { + "epoch": 1.854743316773125, + "grad_norm": 2.7215986251831055, + "learning_rate": 4.945726598036869e-06, + "loss": 0.3217, + "step": 15680 + }, + { + "epoch": 1.8559261887863734, + "grad_norm": 3.511922597885132, + "learning_rate": 4.945008379219536e-06, + "loss": 0.3307, + "step": 15690 + }, + { + "epoch": 1.8571090607996215, + "grad_norm": 2.523838996887207, + "learning_rate": 4.944290160402203e-06, + "loss": 0.3054, + "step": 15700 + }, + { + "epoch": 1.8582919328128695, + "grad_norm": 3.113474130630493, + "learning_rate": 4.94357194158487e-06, + "loss": 0.3372, + "step": 15710 + }, + { + "epoch": 1.8594748048261178, + "grad_norm": 2.440829277038574, + "learning_rate": 4.9428537227675366e-06, + "loss": 0.3174, + "step": 15720 + }, + { + "epoch": 1.8606576768393661, + "grad_norm": 2.771348714828491, + "learning_rate": 4.9421355039502035e-06, + "loss": 0.325, + "step": 15730 + }, + { + "epoch": 1.8618405488526142, + "grad_norm": 2.479052782058716, + "learning_rate": 4.94141728513287e-06, + "loss": 0.3264, + "step": 15740 + }, + { + "epoch": 1.8630234208658623, + "grad_norm": 3.5350215435028076, + "learning_rate": 4.940699066315538e-06, + "loss": 0.2986, + "step": 15750 + }, + { + "epoch": 1.8642062928791105, + "grad_norm": 2.6968698501586914, + "learning_rate": 4.939980847498204e-06, + "loss": 0.2951, + "step": 15760 + }, + { + "epoch": 1.8653891648923586, + "grad_norm": 4.0666890144348145, + "learning_rate": 4.939262628680872e-06, + "loss": 0.2839, + "step": 15770 + }, + { + "epoch": 1.8665720369056067, + "grad_norm": 2.678946018218994, + "learning_rate": 4.938544409863538e-06, + "loss": 0.3036, + "step": 15780 + }, + { + "epoch": 1.867754908918855, + "grad_norm": 3.919001340866089, + "learning_rate": 4.937826191046206e-06, + "loss": 0.3149, + "step": 15790 + }, + { + "epoch": 1.8689377809321033, + "grad_norm": 3.258880138397217, + "learning_rate": 4.937107972228872e-06, + "loss": 0.295, + "step": 15800 + }, + { + "epoch": 1.8701206529453513, + "grad_norm": 2.228896141052246, + "learning_rate": 4.93638975341154e-06, + "loss": 0.3111, + "step": 15810 + }, + { + "epoch": 1.8713035249585994, + "grad_norm": 4.7691144943237305, + "learning_rate": 4.935671534594207e-06, + "loss": 0.3496, + "step": 15820 + }, + { + "epoch": 1.8724863969718477, + "grad_norm": 2.3389461040496826, + "learning_rate": 4.9349533157768736e-06, + "loss": 0.3096, + "step": 15830 + }, + { + "epoch": 1.8736692689850958, + "grad_norm": 2.6259357929229736, + "learning_rate": 4.9342350969595405e-06, + "loss": 0.3102, + "step": 15840 + }, + { + "epoch": 1.8748521409983439, + "grad_norm": 2.053807258605957, + "learning_rate": 4.933516878142207e-06, + "loss": 0.3245, + "step": 15850 + }, + { + "epoch": 1.8760350130115921, + "grad_norm": 2.1412627696990967, + "learning_rate": 4.932798659324874e-06, + "loss": 0.3178, + "step": 15860 + }, + { + "epoch": 1.8772178850248404, + "grad_norm": 2.4492874145507812, + "learning_rate": 4.932080440507541e-06, + "loss": 0.3419, + "step": 15870 + }, + { + "epoch": 1.8784007570380885, + "grad_norm": 2.0245025157928467, + "learning_rate": 4.931362221690208e-06, + "loss": 0.2854, + "step": 15880 + }, + { + "epoch": 1.8795836290513366, + "grad_norm": 2.60402774810791, + "learning_rate": 4.930644002872875e-06, + "loss": 0.3392, + "step": 15890 + }, + { + "epoch": 1.8807665010645849, + "grad_norm": 2.6295523643493652, + "learning_rate": 4.929925784055542e-06, + "loss": 0.3452, + "step": 15900 + }, + { + "epoch": 1.881949373077833, + "grad_norm": 2.099587917327881, + "learning_rate": 4.92920756523821e-06, + "loss": 0.3254, + "step": 15910 + }, + { + "epoch": 1.883132245091081, + "grad_norm": 2.967813491821289, + "learning_rate": 4.928489346420877e-06, + "loss": 0.294, + "step": 15920 + }, + { + "epoch": 1.8843151171043293, + "grad_norm": 2.3281209468841553, + "learning_rate": 4.927771127603544e-06, + "loss": 0.2939, + "step": 15930 + }, + { + "epoch": 1.8854979891175776, + "grad_norm": 2.6038031578063965, + "learning_rate": 4.9270529087862106e-06, + "loss": 0.2921, + "step": 15940 + }, + { + "epoch": 1.8866808611308257, + "grad_norm": 2.3744089603424072, + "learning_rate": 4.9263346899688775e-06, + "loss": 0.3583, + "step": 15950 + }, + { + "epoch": 1.8878637331440737, + "grad_norm": 2.654546022415161, + "learning_rate": 4.925616471151544e-06, + "loss": 0.2888, + "step": 15960 + }, + { + "epoch": 1.889046605157322, + "grad_norm": 2.916886329650879, + "learning_rate": 4.924898252334211e-06, + "loss": 0.3246, + "step": 15970 + }, + { + "epoch": 1.89022947717057, + "grad_norm": 3.3064868450164795, + "learning_rate": 4.924180033516878e-06, + "loss": 0.3232, + "step": 15980 + }, + { + "epoch": 1.8914123491838182, + "grad_norm": 3.14512038230896, + "learning_rate": 4.923461814699545e-06, + "loss": 0.3583, + "step": 15990 + }, + { + "epoch": 1.8925952211970665, + "grad_norm": 2.039816379547119, + "learning_rate": 4.922743595882212e-06, + "loss": 0.3007, + "step": 16000 + }, + { + "epoch": 1.8937780932103148, + "grad_norm": 2.053156614303589, + "learning_rate": 4.922025377064879e-06, + "loss": 0.3105, + "step": 16010 + }, + { + "epoch": 1.8949609652235628, + "grad_norm": 3.142106771469116, + "learning_rate": 4.921307158247547e-06, + "loss": 0.2969, + "step": 16020 + }, + { + "epoch": 1.896143837236811, + "grad_norm": 3.165627956390381, + "learning_rate": 4.920588939430213e-06, + "loss": 0.3458, + "step": 16030 + }, + { + "epoch": 1.8973267092500592, + "grad_norm": 3.0727994441986084, + "learning_rate": 4.919870720612881e-06, + "loss": 0.346, + "step": 16040 + }, + { + "epoch": 1.8985095812633073, + "grad_norm": 2.516692876815796, + "learning_rate": 4.919152501795547e-06, + "loss": 0.3155, + "step": 16050 + }, + { + "epoch": 1.8996924532765553, + "grad_norm": 2.974064588546753, + "learning_rate": 4.9184342829782145e-06, + "loss": 0.3093, + "step": 16060 + }, + { + "epoch": 1.9008753252898036, + "grad_norm": 3.889378309249878, + "learning_rate": 4.9177160641608805e-06, + "loss": 0.3042, + "step": 16070 + }, + { + "epoch": 1.902058197303052, + "grad_norm": 3.5090787410736084, + "learning_rate": 4.916997845343548e-06, + "loss": 0.2987, + "step": 16080 + }, + { + "epoch": 1.9032410693163, + "grad_norm": 2.5992817878723145, + "learning_rate": 4.916279626526215e-06, + "loss": 0.3221, + "step": 16090 + }, + { + "epoch": 1.904423941329548, + "grad_norm": 2.4082863330841064, + "learning_rate": 4.915561407708882e-06, + "loss": 0.32, + "step": 16100 + }, + { + "epoch": 1.9056068133427964, + "grad_norm": 2.10164737701416, + "learning_rate": 4.914843188891549e-06, + "loss": 0.3309, + "step": 16110 + }, + { + "epoch": 1.9067896853560444, + "grad_norm": 3.4077935218811035, + "learning_rate": 4.914124970074216e-06, + "loss": 0.2901, + "step": 16120 + }, + { + "epoch": 1.9079725573692925, + "grad_norm": 3.679767608642578, + "learning_rate": 4.913406751256883e-06, + "loss": 0.2932, + "step": 16130 + }, + { + "epoch": 1.9091554293825408, + "grad_norm": 2.589367389678955, + "learning_rate": 4.91268853243955e-06, + "loss": 0.3221, + "step": 16140 + }, + { + "epoch": 1.910338301395789, + "grad_norm": 2.4946177005767822, + "learning_rate": 4.911970313622217e-06, + "loss": 0.3345, + "step": 16150 + }, + { + "epoch": 1.9115211734090372, + "grad_norm": 2.770401954650879, + "learning_rate": 4.911252094804884e-06, + "loss": 0.3289, + "step": 16160 + }, + { + "epoch": 1.9127040454222852, + "grad_norm": 1.7573366165161133, + "learning_rate": 4.9105338759875515e-06, + "loss": 0.3577, + "step": 16170 + }, + { + "epoch": 1.9138869174355335, + "grad_norm": 2.5026323795318604, + "learning_rate": 4.9098156571702175e-06, + "loss": 0.2945, + "step": 16180 + }, + { + "epoch": 1.9150697894487816, + "grad_norm": 2.5507638454437256, + "learning_rate": 4.909097438352885e-06, + "loss": 0.3074, + "step": 16190 + }, + { + "epoch": 1.9162526614620297, + "grad_norm": 2.670314073562622, + "learning_rate": 4.908379219535552e-06, + "loss": 0.364, + "step": 16200 + }, + { + "epoch": 1.917435533475278, + "grad_norm": 2.6045992374420166, + "learning_rate": 4.907661000718219e-06, + "loss": 0.2892, + "step": 16210 + }, + { + "epoch": 1.9186184054885262, + "grad_norm": 3.1325511932373047, + "learning_rate": 4.906942781900886e-06, + "loss": 0.3123, + "step": 16220 + }, + { + "epoch": 1.9198012775017743, + "grad_norm": 3.4678025245666504, + "learning_rate": 4.906224563083553e-06, + "loss": 0.3233, + "step": 16230 + }, + { + "epoch": 1.9209841495150224, + "grad_norm": 2.2840728759765625, + "learning_rate": 4.90550634426622e-06, + "loss": 0.3126, + "step": 16240 + }, + { + "epoch": 1.9221670215282707, + "grad_norm": 2.3752148151397705, + "learning_rate": 4.904788125448887e-06, + "loss": 0.3123, + "step": 16250 + }, + { + "epoch": 1.923349893541519, + "grad_norm": 3.966721296310425, + "learning_rate": 4.904069906631554e-06, + "loss": 0.3001, + "step": 16260 + }, + { + "epoch": 1.9245327655547668, + "grad_norm": 3.2313265800476074, + "learning_rate": 4.903351687814221e-06, + "loss": 0.3092, + "step": 16270 + }, + { + "epoch": 1.9257156375680151, + "grad_norm": 3.653590679168701, + "learning_rate": 4.902633468996888e-06, + "loss": 0.3049, + "step": 16280 + }, + { + "epoch": 1.9268985095812634, + "grad_norm": 2.0597825050354004, + "learning_rate": 4.901915250179555e-06, + "loss": 0.2974, + "step": 16290 + }, + { + "epoch": 1.9280813815945115, + "grad_norm": 4.069886207580566, + "learning_rate": 4.9011970313622215e-06, + "loss": 0.3612, + "step": 16300 + }, + { + "epoch": 1.9292642536077595, + "grad_norm": 2.4084854125976562, + "learning_rate": 4.900478812544889e-06, + "loss": 0.3309, + "step": 16310 + }, + { + "epoch": 1.9304471256210078, + "grad_norm": 2.6260111331939697, + "learning_rate": 4.899760593727555e-06, + "loss": 0.3455, + "step": 16320 + }, + { + "epoch": 1.9316299976342561, + "grad_norm": 2.047407865524292, + "learning_rate": 4.899042374910223e-06, + "loss": 0.3119, + "step": 16330 + }, + { + "epoch": 1.932812869647504, + "grad_norm": 3.172539472579956, + "learning_rate": 4.898324156092889e-06, + "loss": 0.3302, + "step": 16340 + }, + { + "epoch": 1.9339957416607523, + "grad_norm": 2.5367207527160645, + "learning_rate": 4.897605937275557e-06, + "loss": 0.3523, + "step": 16350 + }, + { + "epoch": 1.9351786136740006, + "grad_norm": 2.7896671295166016, + "learning_rate": 4.896887718458224e-06, + "loss": 0.2973, + "step": 16360 + }, + { + "epoch": 1.9363614856872486, + "grad_norm": 2.9528543949127197, + "learning_rate": 4.896169499640891e-06, + "loss": 0.3202, + "step": 16370 + }, + { + "epoch": 1.9375443577004967, + "grad_norm": 2.150696277618408, + "learning_rate": 4.895451280823558e-06, + "loss": 0.3064, + "step": 16380 + }, + { + "epoch": 1.938727229713745, + "grad_norm": 2.2622129917144775, + "learning_rate": 4.894733062006225e-06, + "loss": 0.2953, + "step": 16390 + }, + { + "epoch": 1.9399101017269933, + "grad_norm": 2.1877410411834717, + "learning_rate": 4.8940148431888915e-06, + "loss": 0.3107, + "step": 16400 + }, + { + "epoch": 1.9410929737402411, + "grad_norm": 4.725532531738281, + "learning_rate": 4.8932966243715585e-06, + "loss": 0.2905, + "step": 16410 + }, + { + "epoch": 1.9422758457534894, + "grad_norm": 3.707885980606079, + "learning_rate": 4.892578405554226e-06, + "loss": 0.2985, + "step": 16420 + }, + { + "epoch": 1.9434587177667377, + "grad_norm": 2.2911036014556885, + "learning_rate": 4.891860186736892e-06, + "loss": 0.2885, + "step": 16430 + }, + { + "epoch": 1.9446415897799858, + "grad_norm": 2.9711685180664062, + "learning_rate": 4.89114196791956e-06, + "loss": 0.3159, + "step": 16440 + }, + { + "epoch": 1.9458244617932339, + "grad_norm": 3.330817461013794, + "learning_rate": 4.890423749102226e-06, + "loss": 0.3232, + "step": 16450 + }, + { + "epoch": 1.9470073338064822, + "grad_norm": 2.778247117996216, + "learning_rate": 4.889705530284894e-06, + "loss": 0.3191, + "step": 16460 + }, + { + "epoch": 1.9481902058197305, + "grad_norm": 3.1285018920898438, + "learning_rate": 4.888987311467561e-06, + "loss": 0.3045, + "step": 16470 + }, + { + "epoch": 1.9493730778329783, + "grad_norm": 2.4701411724090576, + "learning_rate": 4.888269092650228e-06, + "loss": 0.3315, + "step": 16480 + }, + { + "epoch": 1.9505559498462266, + "grad_norm": 2.2952330112457275, + "learning_rate": 4.887550873832895e-06, + "loss": 0.2978, + "step": 16490 + }, + { + "epoch": 1.951738821859475, + "grad_norm": 2.485168218612671, + "learning_rate": 4.886832655015562e-06, + "loss": 0.327, + "step": 16500 + }, + { + "epoch": 1.952921693872723, + "grad_norm": 2.294721841812134, + "learning_rate": 4.8861144361982285e-06, + "loss": 0.3108, + "step": 16510 + }, + { + "epoch": 1.954104565885971, + "grad_norm": 2.7091660499572754, + "learning_rate": 4.8853962173808955e-06, + "loss": 0.3095, + "step": 16520 + }, + { + "epoch": 1.9552874378992193, + "grad_norm": 3.023503541946411, + "learning_rate": 4.884677998563562e-06, + "loss": 0.329, + "step": 16530 + }, + { + "epoch": 1.9564703099124676, + "grad_norm": 2.9048104286193848, + "learning_rate": 4.883959779746229e-06, + "loss": 0.3134, + "step": 16540 + }, + { + "epoch": 1.9576531819257157, + "grad_norm": 2.192518949508667, + "learning_rate": 4.883241560928896e-06, + "loss": 0.3424, + "step": 16550 + }, + { + "epoch": 1.9588360539389638, + "grad_norm": 2.448760986328125, + "learning_rate": 4.882523342111564e-06, + "loss": 0.3148, + "step": 16560 + }, + { + "epoch": 1.960018925952212, + "grad_norm": 2.6753344535827637, + "learning_rate": 4.88180512329423e-06, + "loss": 0.3368, + "step": 16570 + }, + { + "epoch": 1.9612017979654601, + "grad_norm": 2.6930882930755615, + "learning_rate": 4.881086904476898e-06, + "loss": 0.3326, + "step": 16580 + }, + { + "epoch": 1.9623846699787082, + "grad_norm": 3.712798833847046, + "learning_rate": 4.880368685659564e-06, + "loss": 0.3191, + "step": 16590 + }, + { + "epoch": 1.9635675419919565, + "grad_norm": 2.318413496017456, + "learning_rate": 4.879650466842232e-06, + "loss": 0.313, + "step": 16600 + }, + { + "epoch": 1.9647504140052048, + "grad_norm": 3.4627861976623535, + "learning_rate": 4.878932248024898e-06, + "loss": 0.3249, + "step": 16610 + }, + { + "epoch": 1.9659332860184529, + "grad_norm": 2.2681097984313965, + "learning_rate": 4.8782140292075655e-06, + "loss": 0.3827, + "step": 16620 + }, + { + "epoch": 1.967116158031701, + "grad_norm": 2.2495839595794678, + "learning_rate": 4.8774958103902325e-06, + "loss": 0.3035, + "step": 16630 + }, + { + "epoch": 1.9682990300449492, + "grad_norm": 3.3230841159820557, + "learning_rate": 4.876777591572899e-06, + "loss": 0.2947, + "step": 16640 + }, + { + "epoch": 1.9694819020581973, + "grad_norm": 3.0023653507232666, + "learning_rate": 4.876059372755566e-06, + "loss": 0.3282, + "step": 16650 + }, + { + "epoch": 1.9706647740714454, + "grad_norm": 3.0990710258483887, + "learning_rate": 4.875341153938233e-06, + "loss": 0.3423, + "step": 16660 + }, + { + "epoch": 1.9718476460846937, + "grad_norm": 4.0534586906433105, + "learning_rate": 4.8746229351209e-06, + "loss": 0.3292, + "step": 16670 + }, + { + "epoch": 1.973030518097942, + "grad_norm": 2.5327095985412598, + "learning_rate": 4.873904716303567e-06, + "loss": 0.3534, + "step": 16680 + }, + { + "epoch": 1.97421339011119, + "grad_norm": 2.820181131362915, + "learning_rate": 4.873186497486235e-06, + "loss": 0.3455, + "step": 16690 + }, + { + "epoch": 1.975396262124438, + "grad_norm": 2.5984952449798584, + "learning_rate": 4.872468278668901e-06, + "loss": 0.3098, + "step": 16700 + }, + { + "epoch": 1.9765791341376864, + "grad_norm": 2.8330180644989014, + "learning_rate": 4.871750059851569e-06, + "loss": 0.2865, + "step": 16710 + }, + { + "epoch": 1.9777620061509344, + "grad_norm": 3.222081422805786, + "learning_rate": 4.871031841034235e-06, + "loss": 0.3366, + "step": 16720 + }, + { + "epoch": 1.9789448781641825, + "grad_norm": 2.0557262897491455, + "learning_rate": 4.8703136222169025e-06, + "loss": 0.298, + "step": 16730 + }, + { + "epoch": 1.9801277501774308, + "grad_norm": 2.3653440475463867, + "learning_rate": 4.8695954033995695e-06, + "loss": 0.3525, + "step": 16740 + }, + { + "epoch": 1.981310622190679, + "grad_norm": 1.8951219320297241, + "learning_rate": 4.868877184582236e-06, + "loss": 0.2949, + "step": 16750 + }, + { + "epoch": 1.9824934942039272, + "grad_norm": 3.2149484157562256, + "learning_rate": 4.868158965764903e-06, + "loss": 0.2989, + "step": 16760 + }, + { + "epoch": 1.9836763662171752, + "grad_norm": 2.8718087673187256, + "learning_rate": 4.86744074694757e-06, + "loss": 0.3223, + "step": 16770 + }, + { + "epoch": 1.9848592382304235, + "grad_norm": 2.507594585418701, + "learning_rate": 4.866722528130237e-06, + "loss": 0.2927, + "step": 16780 + }, + { + "epoch": 1.9860421102436716, + "grad_norm": 2.338623523712158, + "learning_rate": 4.866004309312904e-06, + "loss": 0.3463, + "step": 16790 + }, + { + "epoch": 1.9872249822569197, + "grad_norm": 1.9228732585906982, + "learning_rate": 4.865286090495571e-06, + "loss": 0.3451, + "step": 16800 + }, + { + "epoch": 1.988407854270168, + "grad_norm": 1.9833261966705322, + "learning_rate": 4.864567871678238e-06, + "loss": 0.3127, + "step": 16810 + }, + { + "epoch": 1.9895907262834163, + "grad_norm": 2.5209105014801025, + "learning_rate": 4.863849652860905e-06, + "loss": 0.3199, + "step": 16820 + }, + { + "epoch": 1.9907735982966643, + "grad_norm": 3.8026509284973145, + "learning_rate": 4.863131434043573e-06, + "loss": 0.3096, + "step": 16830 + }, + { + "epoch": 1.9919564703099124, + "grad_norm": 1.6769155263900757, + "learning_rate": 4.862413215226239e-06, + "loss": 0.3013, + "step": 16840 + }, + { + "epoch": 1.9931393423231607, + "grad_norm": 2.8461453914642334, + "learning_rate": 4.8616949964089065e-06, + "loss": 0.2846, + "step": 16850 + }, + { + "epoch": 1.9943222143364088, + "grad_norm": 2.2293806076049805, + "learning_rate": 4.8609767775915725e-06, + "loss": 0.3239, + "step": 16860 + }, + { + "epoch": 1.9955050863496568, + "grad_norm": 3.478363275527954, + "learning_rate": 4.86025855877424e-06, + "loss": 0.3146, + "step": 16870 + }, + { + "epoch": 1.9966879583629051, + "grad_norm": 2.097290277481079, + "learning_rate": 4.859540339956906e-06, + "loss": 0.3078, + "step": 16880 + }, + { + "epoch": 1.9978708303761534, + "grad_norm": 3.2661945819854736, + "learning_rate": 4.858822121139574e-06, + "loss": 0.2909, + "step": 16890 + }, + { + "epoch": 1.9990537023894015, + "grad_norm": 2.188856840133667, + "learning_rate": 4.858103902322241e-06, + "loss": 0.354, + "step": 16900 + }, + { + "epoch": 2.0002365744026496, + "grad_norm": 2.378852128982544, + "learning_rate": 4.857385683504908e-06, + "loss": 0.3441, + "step": 16910 + }, + { + "epoch": 2.000473148805299, + "eval_accuracy": 0.8596999035166517, + "eval_loss": 0.32458868622779846, + "eval_runtime": 77.9165, + "eval_safe_aucpr": 0.9144648316642549, + "eval_safe_f1": 0.8435192400460129, + "eval_safe_fpr": 0.13461480972168205, + "eval_safe_precision": 0.8346612814393244, + "eval_safe_recall": 0.8525672279938492, + "eval_samples_per_second": 771.518, + "eval_steps_per_second": 12.064, + "eval_unsafe_aucpr": 0.9524256073577522, + "eval_unsafe_f1": 0.8728478818031057, + "eval_unsafe_fpr": 0.1474327720061503, + "eval_unsafe_precision": 0.8804404026886462, + "eval_unsafe_recall": 0.8653851902783175, + "step": 16912 + }, + { + "epoch": 2.001419446415898, + "grad_norm": 2.5568602085113525, + "learning_rate": 4.856667464687575e-06, + "loss": 0.3011, + "step": 16920 + }, + { + "epoch": 2.002602318429146, + "grad_norm": 2.720656156539917, + "learning_rate": 4.855949245870242e-06, + "loss": 0.3187, + "step": 16930 + }, + { + "epoch": 2.003785190442394, + "grad_norm": 2.677154779434204, + "learning_rate": 4.85523102705291e-06, + "loss": 0.2824, + "step": 16940 + }, + { + "epoch": 2.0049680624556423, + "grad_norm": 2.2328920364379883, + "learning_rate": 4.854512808235576e-06, + "loss": 0.3051, + "step": 16950 + }, + { + "epoch": 2.0061509344688906, + "grad_norm": 2.8281667232513428, + "learning_rate": 4.8537945894182435e-06, + "loss": 0.2961, + "step": 16960 + }, + { + "epoch": 2.0073338064821384, + "grad_norm": 2.5169615745544434, + "learning_rate": 4.8530763706009095e-06, + "loss": 0.2839, + "step": 16970 + }, + { + "epoch": 2.0085166784953867, + "grad_norm": 2.4663641452789307, + "learning_rate": 4.852358151783577e-06, + "loss": 0.2746, + "step": 16980 + }, + { + "epoch": 2.009699550508635, + "grad_norm": 2.8655757904052734, + "learning_rate": 4.851639932966243e-06, + "loss": 0.2983, + "step": 16990 + }, + { + "epoch": 2.0108824225218833, + "grad_norm": 2.2925379276275635, + "learning_rate": 4.850921714148911e-06, + "loss": 0.2793, + "step": 17000 + }, + { + "epoch": 2.012065294535131, + "grad_norm": 2.860703945159912, + "learning_rate": 4.850203495331578e-06, + "loss": 0.2945, + "step": 17010 + }, + { + "epoch": 2.0132481665483795, + "grad_norm": 1.9645798206329346, + "learning_rate": 4.849485276514245e-06, + "loss": 0.3051, + "step": 17020 + }, + { + "epoch": 2.0144310385616278, + "grad_norm": 3.2021446228027344, + "learning_rate": 4.848767057696912e-06, + "loss": 0.3046, + "step": 17030 + }, + { + "epoch": 2.0156139105748756, + "grad_norm": 2.800635576248169, + "learning_rate": 4.848048838879579e-06, + "loss": 0.2976, + "step": 17040 + }, + { + "epoch": 2.016796782588124, + "grad_norm": 2.4295828342437744, + "learning_rate": 4.847330620062246e-06, + "loss": 0.2766, + "step": 17050 + }, + { + "epoch": 2.017979654601372, + "grad_norm": 3.77390718460083, + "learning_rate": 4.846612401244913e-06, + "loss": 0.327, + "step": 17060 + }, + { + "epoch": 2.0191625266146205, + "grad_norm": 2.4511208534240723, + "learning_rate": 4.84589418242758e-06, + "loss": 0.2611, + "step": 17070 + }, + { + "epoch": 2.0203453986278683, + "grad_norm": 1.8200575113296509, + "learning_rate": 4.8451759636102465e-06, + "loss": 0.2896, + "step": 17080 + }, + { + "epoch": 2.0215282706411166, + "grad_norm": 3.8342087268829346, + "learning_rate": 4.8444577447929135e-06, + "loss": 0.3108, + "step": 17090 + }, + { + "epoch": 2.022711142654365, + "grad_norm": 2.991922616958618, + "learning_rate": 4.84373952597558e-06, + "loss": 0.2787, + "step": 17100 + }, + { + "epoch": 2.0238940146676128, + "grad_norm": 2.300657033920288, + "learning_rate": 4.843021307158247e-06, + "loss": 0.2909, + "step": 17110 + }, + { + "epoch": 2.025076886680861, + "grad_norm": 2.628596782684326, + "learning_rate": 4.842303088340915e-06, + "loss": 0.2577, + "step": 17120 + }, + { + "epoch": 2.0262597586941093, + "grad_norm": 4.038423538208008, + "learning_rate": 4.841584869523581e-06, + "loss": 0.2979, + "step": 17130 + }, + { + "epoch": 2.0274426307073576, + "grad_norm": 3.1706392765045166, + "learning_rate": 4.840866650706249e-06, + "loss": 0.2929, + "step": 17140 + }, + { + "epoch": 2.0286255027206055, + "grad_norm": 3.052032232284546, + "learning_rate": 4.840148431888915e-06, + "loss": 0.3322, + "step": 17150 + }, + { + "epoch": 2.029808374733854, + "grad_norm": 2.10490083694458, + "learning_rate": 4.839430213071583e-06, + "loss": 0.2729, + "step": 17160 + }, + { + "epoch": 2.030991246747102, + "grad_norm": 3.376361608505249, + "learning_rate": 4.83871199425425e-06, + "loss": 0.2934, + "step": 17170 + }, + { + "epoch": 2.03217411876035, + "grad_norm": 2.8784351348876953, + "learning_rate": 4.837993775436917e-06, + "loss": 0.3, + "step": 17180 + }, + { + "epoch": 2.033356990773598, + "grad_norm": 2.5840699672698975, + "learning_rate": 4.8372755566195835e-06, + "loss": 0.2863, + "step": 17190 + }, + { + "epoch": 2.0345398627868465, + "grad_norm": 3.836290121078491, + "learning_rate": 4.8365573378022504e-06, + "loss": 0.2774, + "step": 17200 + }, + { + "epoch": 2.035722734800095, + "grad_norm": 2.947138547897339, + "learning_rate": 4.835839118984918e-06, + "loss": 0.3377, + "step": 17210 + }, + { + "epoch": 2.0369056068133427, + "grad_norm": 3.6667375564575195, + "learning_rate": 4.835120900167584e-06, + "loss": 0.3329, + "step": 17220 + }, + { + "epoch": 2.038088478826591, + "grad_norm": 2.7514986991882324, + "learning_rate": 4.834402681350252e-06, + "loss": 0.3048, + "step": 17230 + }, + { + "epoch": 2.0392713508398392, + "grad_norm": 2.772559642791748, + "learning_rate": 4.833684462532918e-06, + "loss": 0.3044, + "step": 17240 + }, + { + "epoch": 2.040454222853087, + "grad_norm": 3.929675340652466, + "learning_rate": 4.832966243715586e-06, + "loss": 0.3011, + "step": 17250 + }, + { + "epoch": 2.0416370948663354, + "grad_norm": 3.0461666584014893, + "learning_rate": 4.832248024898252e-06, + "loss": 0.3615, + "step": 17260 + }, + { + "epoch": 2.0428199668795837, + "grad_norm": 3.056796073913574, + "learning_rate": 4.83152980608092e-06, + "loss": 0.3246, + "step": 17270 + }, + { + "epoch": 2.044002838892832, + "grad_norm": 3.44435453414917, + "learning_rate": 4.830811587263587e-06, + "loss": 0.3283, + "step": 17280 + }, + { + "epoch": 2.04518571090608, + "grad_norm": 2.164142370223999, + "learning_rate": 4.830093368446254e-06, + "loss": 0.2769, + "step": 17290 + }, + { + "epoch": 2.046368582919328, + "grad_norm": 2.2050135135650635, + "learning_rate": 4.8293751496289205e-06, + "loss": 0.2849, + "step": 17300 + }, + { + "epoch": 2.0475514549325764, + "grad_norm": 1.9663429260253906, + "learning_rate": 4.8286569308115874e-06, + "loss": 0.3121, + "step": 17310 + }, + { + "epoch": 2.0487343269458242, + "grad_norm": 1.7338041067123413, + "learning_rate": 4.827938711994254e-06, + "loss": 0.3041, + "step": 17320 + }, + { + "epoch": 2.0499171989590725, + "grad_norm": 2.510803699493408, + "learning_rate": 4.827220493176921e-06, + "loss": 0.3124, + "step": 17330 + }, + { + "epoch": 2.051100070972321, + "grad_norm": 2.887864112854004, + "learning_rate": 4.826502274359588e-06, + "loss": 0.3033, + "step": 17340 + }, + { + "epoch": 2.052282942985569, + "grad_norm": 2.86704158782959, + "learning_rate": 4.825784055542255e-06, + "loss": 0.2912, + "step": 17350 + }, + { + "epoch": 2.053465814998817, + "grad_norm": 3.1380202770233154, + "learning_rate": 4.825065836724922e-06, + "loss": 0.3342, + "step": 17360 + }, + { + "epoch": 2.0546486870120653, + "grad_norm": 2.568093776702881, + "learning_rate": 4.824347617907589e-06, + "loss": 0.3281, + "step": 17370 + }, + { + "epoch": 2.0558315590253136, + "grad_norm": 2.2394888401031494, + "learning_rate": 4.823629399090256e-06, + "loss": 0.3013, + "step": 17380 + }, + { + "epoch": 2.057014431038562, + "grad_norm": 2.0911223888397217, + "learning_rate": 4.822911180272924e-06, + "loss": 0.3067, + "step": 17390 + }, + { + "epoch": 2.0581973030518097, + "grad_norm": 2.685105800628662, + "learning_rate": 4.82219296145559e-06, + "loss": 0.328, + "step": 17400 + }, + { + "epoch": 2.059380175065058, + "grad_norm": 2.4722275733947754, + "learning_rate": 4.8214747426382575e-06, + "loss": 0.2925, + "step": 17410 + }, + { + "epoch": 2.0605630470783063, + "grad_norm": 3.348921537399292, + "learning_rate": 4.820756523820924e-06, + "loss": 0.3009, + "step": 17420 + }, + { + "epoch": 2.061745919091554, + "grad_norm": 2.4126553535461426, + "learning_rate": 4.820038305003591e-06, + "loss": 0.3329, + "step": 17430 + }, + { + "epoch": 2.0629287911048024, + "grad_norm": 3.171835422515869, + "learning_rate": 4.819320086186258e-06, + "loss": 0.2819, + "step": 17440 + }, + { + "epoch": 2.0641116631180507, + "grad_norm": 3.453155279159546, + "learning_rate": 4.818601867368925e-06, + "loss": 0.3032, + "step": 17450 + }, + { + "epoch": 2.065294535131299, + "grad_norm": 3.60829758644104, + "learning_rate": 4.817883648551592e-06, + "loss": 0.311, + "step": 17460 + }, + { + "epoch": 2.066477407144547, + "grad_norm": 2.449158191680908, + "learning_rate": 4.817165429734259e-06, + "loss": 0.2647, + "step": 17470 + }, + { + "epoch": 2.067660279157795, + "grad_norm": 4.3593220710754395, + "learning_rate": 4.816447210916927e-06, + "loss": 0.3355, + "step": 17480 + }, + { + "epoch": 2.0688431511710434, + "grad_norm": 3.0162782669067383, + "learning_rate": 4.815728992099593e-06, + "loss": 0.2849, + "step": 17490 + }, + { + "epoch": 2.0700260231842913, + "grad_norm": 2.3126392364501953, + "learning_rate": 4.815010773282261e-06, + "loss": 0.3154, + "step": 17500 + }, + { + "epoch": 2.0712088951975396, + "grad_norm": 2.1091065406799316, + "learning_rate": 4.814292554464927e-06, + "loss": 0.2635, + "step": 17510 + }, + { + "epoch": 2.072391767210788, + "grad_norm": 2.544283628463745, + "learning_rate": 4.8135743356475945e-06, + "loss": 0.3101, + "step": 17520 + }, + { + "epoch": 2.073574639224036, + "grad_norm": 3.6290197372436523, + "learning_rate": 4.812856116830261e-06, + "loss": 0.2831, + "step": 17530 + }, + { + "epoch": 2.074757511237284, + "grad_norm": 3.472325086593628, + "learning_rate": 4.812137898012928e-06, + "loss": 0.3088, + "step": 17540 + }, + { + "epoch": 2.0759403832505323, + "grad_norm": 3.375314950942993, + "learning_rate": 4.811419679195595e-06, + "loss": 0.305, + "step": 17550 + }, + { + "epoch": 2.0771232552637806, + "grad_norm": 1.766257882118225, + "learning_rate": 4.810701460378262e-06, + "loss": 0.2758, + "step": 17560 + }, + { + "epoch": 2.0783061272770285, + "grad_norm": 3.06916880607605, + "learning_rate": 4.809983241560929e-06, + "loss": 0.2686, + "step": 17570 + }, + { + "epoch": 2.0794889992902768, + "grad_norm": 2.8144593238830566, + "learning_rate": 4.809265022743596e-06, + "loss": 0.3432, + "step": 17580 + }, + { + "epoch": 2.080671871303525, + "grad_norm": 2.9351181983947754, + "learning_rate": 4.808546803926263e-06, + "loss": 0.2899, + "step": 17590 + }, + { + "epoch": 2.0818547433167733, + "grad_norm": 3.8633251190185547, + "learning_rate": 4.80782858510893e-06, + "loss": 0.3236, + "step": 17600 + }, + { + "epoch": 2.083037615330021, + "grad_norm": 2.226646900177002, + "learning_rate": 4.807110366291597e-06, + "loss": 0.3458, + "step": 17610 + }, + { + "epoch": 2.0842204873432695, + "grad_norm": 3.1733455657958984, + "learning_rate": 4.806392147474264e-06, + "loss": 0.3047, + "step": 17620 + }, + { + "epoch": 2.0854033593565178, + "grad_norm": 2.6069700717926025, + "learning_rate": 4.805673928656931e-06, + "loss": 0.3481, + "step": 17630 + }, + { + "epoch": 2.0865862313697656, + "grad_norm": 2.2039647102355957, + "learning_rate": 4.804955709839598e-06, + "loss": 0.3056, + "step": 17640 + }, + { + "epoch": 2.087769103383014, + "grad_norm": 2.762326717376709, + "learning_rate": 4.8042374910222645e-06, + "loss": 0.2749, + "step": 17650 + }, + { + "epoch": 2.088951975396262, + "grad_norm": 3.3768422603607178, + "learning_rate": 4.803519272204932e-06, + "loss": 0.3152, + "step": 17660 + }, + { + "epoch": 2.0901348474095105, + "grad_norm": 3.592968463897705, + "learning_rate": 4.802801053387598e-06, + "loss": 0.322, + "step": 17670 + }, + { + "epoch": 2.0913177194227583, + "grad_norm": 3.6042120456695557, + "learning_rate": 4.802082834570266e-06, + "loss": 0.3365, + "step": 17680 + }, + { + "epoch": 2.0925005914360066, + "grad_norm": 3.1354405879974365, + "learning_rate": 4.801364615752933e-06, + "loss": 0.3007, + "step": 17690 + }, + { + "epoch": 2.093683463449255, + "grad_norm": 2.3326992988586426, + "learning_rate": 4.8006463969356e-06, + "loss": 0.262, + "step": 17700 + }, + { + "epoch": 2.094866335462503, + "grad_norm": 2.435826063156128, + "learning_rate": 4.799928178118267e-06, + "loss": 0.2746, + "step": 17710 + }, + { + "epoch": 2.096049207475751, + "grad_norm": 2.6682724952697754, + "learning_rate": 4.799209959300934e-06, + "loss": 0.2686, + "step": 17720 + }, + { + "epoch": 2.0972320794889994, + "grad_norm": 3.148589611053467, + "learning_rate": 4.798491740483601e-06, + "loss": 0.3076, + "step": 17730 + }, + { + "epoch": 2.0984149515022477, + "grad_norm": 3.278606414794922, + "learning_rate": 4.797773521666268e-06, + "loss": 0.2969, + "step": 17740 + }, + { + "epoch": 2.0995978235154955, + "grad_norm": 2.676833391189575, + "learning_rate": 4.797055302848935e-06, + "loss": 0.335, + "step": 17750 + }, + { + "epoch": 2.100780695528744, + "grad_norm": 2.208115339279175, + "learning_rate": 4.7963370840316015e-06, + "loss": 0.3225, + "step": 17760 + }, + { + "epoch": 2.101963567541992, + "grad_norm": 2.8714101314544678, + "learning_rate": 4.795618865214269e-06, + "loss": 0.3168, + "step": 17770 + }, + { + "epoch": 2.10314643955524, + "grad_norm": 3.031416416168213, + "learning_rate": 4.794900646396935e-06, + "loss": 0.322, + "step": 17780 + }, + { + "epoch": 2.1043293115684882, + "grad_norm": 3.5075860023498535, + "learning_rate": 4.794182427579603e-06, + "loss": 0.263, + "step": 17790 + }, + { + "epoch": 2.1055121835817365, + "grad_norm": 2.9896793365478516, + "learning_rate": 4.793464208762269e-06, + "loss": 0.3071, + "step": 17800 + }, + { + "epoch": 2.106695055594985, + "grad_norm": 2.4877758026123047, + "learning_rate": 4.792745989944937e-06, + "loss": 0.3474, + "step": 17810 + }, + { + "epoch": 2.1078779276082327, + "grad_norm": 2.5066843032836914, + "learning_rate": 4.792027771127604e-06, + "loss": 0.2792, + "step": 17820 + }, + { + "epoch": 2.109060799621481, + "grad_norm": 4.5107879638671875, + "learning_rate": 4.791309552310271e-06, + "loss": 0.3341, + "step": 17830 + }, + { + "epoch": 2.1102436716347293, + "grad_norm": 2.9022371768951416, + "learning_rate": 4.790591333492938e-06, + "loss": 0.3126, + "step": 17840 + }, + { + "epoch": 2.111426543647977, + "grad_norm": 2.792145252227783, + "learning_rate": 4.789873114675605e-06, + "loss": 0.2626, + "step": 17850 + }, + { + "epoch": 2.1126094156612254, + "grad_norm": 2.527423143386841, + "learning_rate": 4.789154895858272e-06, + "loss": 0.254, + "step": 17860 + }, + { + "epoch": 2.1137922876744737, + "grad_norm": 2.630368232727051, + "learning_rate": 4.7884366770409385e-06, + "loss": 0.2923, + "step": 17870 + }, + { + "epoch": 2.114975159687722, + "grad_norm": 3.8062503337860107, + "learning_rate": 4.7877184582236054e-06, + "loss": 0.2795, + "step": 17880 + }, + { + "epoch": 2.11615803170097, + "grad_norm": 2.3464980125427246, + "learning_rate": 4.787000239406272e-06, + "loss": 0.2899, + "step": 17890 + }, + { + "epoch": 2.117340903714218, + "grad_norm": 3.2609386444091797, + "learning_rate": 4.786282020588939e-06, + "loss": 0.3082, + "step": 17900 + }, + { + "epoch": 2.1185237757274664, + "grad_norm": 3.002751111984253, + "learning_rate": 4.785563801771606e-06, + "loss": 0.3, + "step": 17910 + }, + { + "epoch": 2.1197066477407143, + "grad_norm": 4.029548645019531, + "learning_rate": 4.784845582954273e-06, + "loss": 0.2626, + "step": 17920 + }, + { + "epoch": 2.1208895197539626, + "grad_norm": 3.2029058933258057, + "learning_rate": 4.784127364136941e-06, + "loss": 0.2857, + "step": 17930 + }, + { + "epoch": 2.122072391767211, + "grad_norm": 3.222637176513672, + "learning_rate": 4.783409145319608e-06, + "loss": 0.2907, + "step": 17940 + }, + { + "epoch": 2.123255263780459, + "grad_norm": 3.828500509262085, + "learning_rate": 4.782690926502275e-06, + "loss": 0.317, + "step": 17950 + }, + { + "epoch": 2.124438135793707, + "grad_norm": 2.0794427394866943, + "learning_rate": 4.781972707684942e-06, + "loss": 0.2811, + "step": 17960 + }, + { + "epoch": 2.1256210078069553, + "grad_norm": 3.8097901344299316, + "learning_rate": 4.781254488867609e-06, + "loss": 0.3, + "step": 17970 + }, + { + "epoch": 2.1268038798202036, + "grad_norm": 3.452726125717163, + "learning_rate": 4.7805362700502755e-06, + "loss": 0.3051, + "step": 17980 + }, + { + "epoch": 2.1279867518334514, + "grad_norm": 2.9992544651031494, + "learning_rate": 4.7798180512329424e-06, + "loss": 0.3314, + "step": 17990 + }, + { + "epoch": 2.1291696238466997, + "grad_norm": 3.1494879722595215, + "learning_rate": 4.779099832415609e-06, + "loss": 0.2847, + "step": 18000 + }, + { + "epoch": 2.130352495859948, + "grad_norm": 2.6865220069885254, + "learning_rate": 4.778381613598276e-06, + "loss": 0.3348, + "step": 18010 + }, + { + "epoch": 2.1315353678731963, + "grad_norm": 2.88586163520813, + "learning_rate": 4.777663394780943e-06, + "loss": 0.315, + "step": 18020 + }, + { + "epoch": 2.132718239886444, + "grad_norm": 1.7792019844055176, + "learning_rate": 4.77694517596361e-06, + "loss": 0.297, + "step": 18030 + }, + { + "epoch": 2.1339011118996924, + "grad_norm": 2.483880043029785, + "learning_rate": 4.776226957146278e-06, + "loss": 0.2953, + "step": 18040 + }, + { + "epoch": 2.1350839839129407, + "grad_norm": 2.2779011726379395, + "learning_rate": 4.775508738328944e-06, + "loss": 0.2728, + "step": 18050 + }, + { + "epoch": 2.1362668559261886, + "grad_norm": 2.3991661071777344, + "learning_rate": 4.774790519511612e-06, + "loss": 0.2754, + "step": 18060 + }, + { + "epoch": 2.137449727939437, + "grad_norm": 2.5661368370056152, + "learning_rate": 4.774072300694278e-06, + "loss": 0.3089, + "step": 18070 + }, + { + "epoch": 2.138632599952685, + "grad_norm": 3.713167667388916, + "learning_rate": 4.773354081876946e-06, + "loss": 0.3264, + "step": 18080 + }, + { + "epoch": 2.1398154719659335, + "grad_norm": 3.65191388130188, + "learning_rate": 4.772635863059612e-06, + "loss": 0.3124, + "step": 18090 + }, + { + "epoch": 2.1409983439791813, + "grad_norm": 3.8617939949035645, + "learning_rate": 4.7719176442422794e-06, + "loss": 0.2983, + "step": 18100 + }, + { + "epoch": 2.1421812159924296, + "grad_norm": 3.0634047985076904, + "learning_rate": 4.771199425424946e-06, + "loss": 0.3095, + "step": 18110 + }, + { + "epoch": 2.143364088005678, + "grad_norm": 2.234780788421631, + "learning_rate": 4.770481206607613e-06, + "loss": 0.3135, + "step": 18120 + }, + { + "epoch": 2.1445469600189258, + "grad_norm": 4.0136308670043945, + "learning_rate": 4.76976298779028e-06, + "loss": 0.3775, + "step": 18130 + }, + { + "epoch": 2.145729832032174, + "grad_norm": 2.7620391845703125, + "learning_rate": 4.769044768972947e-06, + "loss": 0.2839, + "step": 18140 + }, + { + "epoch": 2.1469127040454223, + "grad_norm": 2.1845767498016357, + "learning_rate": 4.768326550155614e-06, + "loss": 0.2854, + "step": 18150 + }, + { + "epoch": 2.1480955760586706, + "grad_norm": 3.211034059524536, + "learning_rate": 4.767608331338281e-06, + "loss": 0.3039, + "step": 18160 + }, + { + "epoch": 2.1492784480719185, + "grad_norm": 3.8874318599700928, + "learning_rate": 4.766890112520948e-06, + "loss": 0.3248, + "step": 18170 + }, + { + "epoch": 2.1504613200851668, + "grad_norm": 2.706275224685669, + "learning_rate": 4.766171893703615e-06, + "loss": 0.3081, + "step": 18180 + }, + { + "epoch": 2.151644192098415, + "grad_norm": 2.9191946983337402, + "learning_rate": 4.765453674886283e-06, + "loss": 0.2904, + "step": 18190 + }, + { + "epoch": 2.152827064111663, + "grad_norm": 2.4864437580108643, + "learning_rate": 4.7647354560689495e-06, + "loss": 0.3024, + "step": 18200 + }, + { + "epoch": 2.154009936124911, + "grad_norm": 2.5812764167785645, + "learning_rate": 4.7640172372516164e-06, + "loss": 0.3345, + "step": 18210 + }, + { + "epoch": 2.1551928081381595, + "grad_norm": 3.733234167098999, + "learning_rate": 4.763299018434283e-06, + "loss": 0.2855, + "step": 18220 + }, + { + "epoch": 2.156375680151408, + "grad_norm": 2.7658438682556152, + "learning_rate": 4.76258079961695e-06, + "loss": 0.2872, + "step": 18230 + }, + { + "epoch": 2.1575585521646556, + "grad_norm": 3.477675437927246, + "learning_rate": 4.761862580799617e-06, + "loss": 0.3414, + "step": 18240 + }, + { + "epoch": 2.158741424177904, + "grad_norm": 2.810882806777954, + "learning_rate": 4.761144361982284e-06, + "loss": 0.3015, + "step": 18250 + }, + { + "epoch": 2.1599242961911522, + "grad_norm": 2.0806736946105957, + "learning_rate": 4.760426143164951e-06, + "loss": 0.3025, + "step": 18260 + }, + { + "epoch": 2.1611071682044, + "grad_norm": 3.2939205169677734, + "learning_rate": 4.759707924347618e-06, + "loss": 0.2662, + "step": 18270 + }, + { + "epoch": 2.1622900402176484, + "grad_norm": 3.723284959793091, + "learning_rate": 4.758989705530285e-06, + "loss": 0.314, + "step": 18280 + }, + { + "epoch": 2.1634729122308967, + "grad_norm": 3.262580394744873, + "learning_rate": 4.758271486712952e-06, + "loss": 0.3444, + "step": 18290 + }, + { + "epoch": 2.164655784244145, + "grad_norm": 3.2462034225463867, + "learning_rate": 4.757553267895619e-06, + "loss": 0.3254, + "step": 18300 + }, + { + "epoch": 2.165838656257393, + "grad_norm": 1.9273457527160645, + "learning_rate": 4.7568350490782865e-06, + "loss": 0.3244, + "step": 18310 + }, + { + "epoch": 2.167021528270641, + "grad_norm": 2.0826480388641357, + "learning_rate": 4.756116830260953e-06, + "loss": 0.2756, + "step": 18320 + }, + { + "epoch": 2.1682044002838894, + "grad_norm": 2.599860906600952, + "learning_rate": 4.75539861144362e-06, + "loss": 0.3436, + "step": 18330 + }, + { + "epoch": 2.1693872722971372, + "grad_norm": 3.9145467281341553, + "learning_rate": 4.754680392626286e-06, + "loss": 0.3371, + "step": 18340 + }, + { + "epoch": 2.1705701443103855, + "grad_norm": 2.7741270065307617, + "learning_rate": 4.753962173808954e-06, + "loss": 0.2983, + "step": 18350 + }, + { + "epoch": 2.171753016323634, + "grad_norm": 3.1485610008239746, + "learning_rate": 4.75324395499162e-06, + "loss": 0.3037, + "step": 18360 + }, + { + "epoch": 2.172935888336882, + "grad_norm": 2.2255494594573975, + "learning_rate": 4.752525736174288e-06, + "loss": 0.303, + "step": 18370 + }, + { + "epoch": 2.17411876035013, + "grad_norm": 2.101179599761963, + "learning_rate": 4.751807517356955e-06, + "loss": 0.3207, + "step": 18380 + }, + { + "epoch": 2.1753016323633783, + "grad_norm": 2.9513349533081055, + "learning_rate": 4.751089298539622e-06, + "loss": 0.2832, + "step": 18390 + }, + { + "epoch": 2.1764845043766265, + "grad_norm": 1.9937074184417725, + "learning_rate": 4.750371079722289e-06, + "loss": 0.2832, + "step": 18400 + }, + { + "epoch": 2.1776673763898744, + "grad_norm": 2.811944007873535, + "learning_rate": 4.749652860904956e-06, + "loss": 0.2897, + "step": 18410 + }, + { + "epoch": 2.1788502484031227, + "grad_norm": 2.1398425102233887, + "learning_rate": 4.748934642087623e-06, + "loss": 0.3283, + "step": 18420 + }, + { + "epoch": 2.180033120416371, + "grad_norm": 2.717740058898926, + "learning_rate": 4.7482164232702896e-06, + "loss": 0.2815, + "step": 18430 + }, + { + "epoch": 2.1812159924296193, + "grad_norm": 3.368557929992676, + "learning_rate": 4.7474982044529565e-06, + "loss": 0.2796, + "step": 18440 + }, + { + "epoch": 2.182398864442867, + "grad_norm": 3.1988909244537354, + "learning_rate": 4.746779985635623e-06, + "loss": 0.3179, + "step": 18450 + }, + { + "epoch": 2.1835817364561154, + "grad_norm": 2.1743054389953613, + "learning_rate": 4.746061766818291e-06, + "loss": 0.3072, + "step": 18460 + }, + { + "epoch": 2.1847646084693637, + "grad_norm": 2.70023250579834, + "learning_rate": 4.745343548000958e-06, + "loss": 0.318, + "step": 18470 + }, + { + "epoch": 2.1859474804826116, + "grad_norm": 2.7275190353393555, + "learning_rate": 4.744625329183625e-06, + "loss": 0.3574, + "step": 18480 + }, + { + "epoch": 2.18713035249586, + "grad_norm": 2.8033287525177, + "learning_rate": 4.743907110366292e-06, + "loss": 0.3234, + "step": 18490 + }, + { + "epoch": 2.188313224509108, + "grad_norm": 2.5095129013061523, + "learning_rate": 4.743188891548959e-06, + "loss": 0.3088, + "step": 18500 + }, + { + "epoch": 2.1894960965223564, + "grad_norm": 2.745701551437378, + "learning_rate": 4.742470672731626e-06, + "loss": 0.3002, + "step": 18510 + }, + { + "epoch": 2.1906789685356043, + "grad_norm": 4.642023086547852, + "learning_rate": 4.741752453914293e-06, + "loss": 0.3005, + "step": 18520 + }, + { + "epoch": 2.1918618405488526, + "grad_norm": 3.1611831188201904, + "learning_rate": 4.74103423509696e-06, + "loss": 0.2878, + "step": 18530 + }, + { + "epoch": 2.193044712562101, + "grad_norm": 2.574026107788086, + "learning_rate": 4.7403160162796266e-06, + "loss": 0.3123, + "step": 18540 + }, + { + "epoch": 2.1942275845753487, + "grad_norm": 3.551522970199585, + "learning_rate": 4.7395977974622935e-06, + "loss": 0.3249, + "step": 18550 + }, + { + "epoch": 2.195410456588597, + "grad_norm": 3.1419999599456787, + "learning_rate": 4.73887957864496e-06, + "loss": 0.3533, + "step": 18560 + }, + { + "epoch": 2.1965933286018453, + "grad_norm": 2.939589500427246, + "learning_rate": 4.738161359827627e-06, + "loss": 0.3039, + "step": 18570 + }, + { + "epoch": 2.1977762006150936, + "grad_norm": 2.616450786590576, + "learning_rate": 4.737443141010295e-06, + "loss": 0.3401, + "step": 18580 + }, + { + "epoch": 2.1989590726283414, + "grad_norm": 2.348172187805176, + "learning_rate": 4.736724922192961e-06, + "loss": 0.3196, + "step": 18590 + }, + { + "epoch": 2.2001419446415897, + "grad_norm": 3.9113495349884033, + "learning_rate": 4.736006703375629e-06, + "loss": 0.3107, + "step": 18600 + }, + { + "epoch": 2.201324816654838, + "grad_norm": 3.003316640853882, + "learning_rate": 4.735288484558295e-06, + "loss": 0.274, + "step": 18610 + }, + { + "epoch": 2.202507688668086, + "grad_norm": 2.2264556884765625, + "learning_rate": 4.734570265740963e-06, + "loss": 0.2746, + "step": 18620 + }, + { + "epoch": 2.203690560681334, + "grad_norm": 3.4186272621154785, + "learning_rate": 4.733852046923629e-06, + "loss": 0.3261, + "step": 18630 + }, + { + "epoch": 2.2048734326945825, + "grad_norm": 2.3731203079223633, + "learning_rate": 4.733133828106297e-06, + "loss": 0.2993, + "step": 18640 + }, + { + "epoch": 2.2060563047078308, + "grad_norm": 2.3863048553466797, + "learning_rate": 4.7324156092889636e-06, + "loss": 0.3394, + "step": 18650 + }, + { + "epoch": 2.2072391767210786, + "grad_norm": 2.5173656940460205, + "learning_rate": 4.7316973904716305e-06, + "loss": 0.3133, + "step": 18660 + }, + { + "epoch": 2.208422048734327, + "grad_norm": 3.6146798133850098, + "learning_rate": 4.730979171654297e-06, + "loss": 0.3189, + "step": 18670 + }, + { + "epoch": 2.209604920747575, + "grad_norm": 5.136168003082275, + "learning_rate": 4.730260952836964e-06, + "loss": 0.2865, + "step": 18680 + }, + { + "epoch": 2.210787792760823, + "grad_norm": 3.4043924808502197, + "learning_rate": 4.729542734019631e-06, + "loss": 0.2866, + "step": 18690 + }, + { + "epoch": 2.2119706647740713, + "grad_norm": 3.6937315464019775, + "learning_rate": 4.728824515202298e-06, + "loss": 0.3092, + "step": 18700 + }, + { + "epoch": 2.2131535367873196, + "grad_norm": 2.4753952026367188, + "learning_rate": 4.728106296384966e-06, + "loss": 0.2671, + "step": 18710 + }, + { + "epoch": 2.214336408800568, + "grad_norm": 3.532831907272339, + "learning_rate": 4.727388077567632e-06, + "loss": 0.3366, + "step": 18720 + }, + { + "epoch": 2.2155192808138158, + "grad_norm": 2.442025899887085, + "learning_rate": 4.7266698587503e-06, + "loss": 0.2818, + "step": 18730 + }, + { + "epoch": 2.216702152827064, + "grad_norm": 2.913315773010254, + "learning_rate": 4.725951639932967e-06, + "loss": 0.3276, + "step": 18740 + }, + { + "epoch": 2.2178850248403124, + "grad_norm": 3.2740283012390137, + "learning_rate": 4.725233421115634e-06, + "loss": 0.3276, + "step": 18750 + }, + { + "epoch": 2.2190678968535607, + "grad_norm": 2.6545071601867676, + "learning_rate": 4.7245152022983006e-06, + "loss": 0.2847, + "step": 18760 + }, + { + "epoch": 2.2202507688668085, + "grad_norm": 3.57932186126709, + "learning_rate": 4.7237969834809675e-06, + "loss": 0.2928, + "step": 18770 + }, + { + "epoch": 2.221433640880057, + "grad_norm": 2.417505979537964, + "learning_rate": 4.723078764663634e-06, + "loss": 0.2684, + "step": 18780 + }, + { + "epoch": 2.222616512893305, + "grad_norm": 2.3136589527130127, + "learning_rate": 4.722360545846301e-06, + "loss": 0.3005, + "step": 18790 + }, + { + "epoch": 2.223799384906553, + "grad_norm": 2.154871940612793, + "learning_rate": 4.721642327028968e-06, + "loss": 0.3537, + "step": 18800 + }, + { + "epoch": 2.2249822569198012, + "grad_norm": 3.6316823959350586, + "learning_rate": 4.720924108211635e-06, + "loss": 0.2946, + "step": 18810 + }, + { + "epoch": 2.2261651289330495, + "grad_norm": 3.7252657413482666, + "learning_rate": 4.720205889394302e-06, + "loss": 0.291, + "step": 18820 + }, + { + "epoch": 2.227348000946298, + "grad_norm": 3.5270731449127197, + "learning_rate": 4.719487670576969e-06, + "loss": 0.2811, + "step": 18830 + }, + { + "epoch": 2.2285308729595457, + "grad_norm": 3.1668782234191895, + "learning_rate": 4.718769451759636e-06, + "loss": 0.2607, + "step": 18840 + }, + { + "epoch": 2.229713744972794, + "grad_norm": 4.512148380279541, + "learning_rate": 4.718051232942304e-06, + "loss": 0.3083, + "step": 18850 + }, + { + "epoch": 2.2308966169860422, + "grad_norm": 3.8025619983673096, + "learning_rate": 4.71733301412497e-06, + "loss": 0.2977, + "step": 18860 + }, + { + "epoch": 2.23207948899929, + "grad_norm": 2.41748309135437, + "learning_rate": 4.7166147953076376e-06, + "loss": 0.2873, + "step": 18870 + }, + { + "epoch": 2.2332623610125384, + "grad_norm": 3.397749662399292, + "learning_rate": 4.715896576490304e-06, + "loss": 0.2936, + "step": 18880 + }, + { + "epoch": 2.2344452330257867, + "grad_norm": 2.466710090637207, + "learning_rate": 4.715178357672971e-06, + "loss": 0.3118, + "step": 18890 + }, + { + "epoch": 2.235628105039035, + "grad_norm": 3.05507493019104, + "learning_rate": 4.7144601388556375e-06, + "loss": 0.3089, + "step": 18900 + }, + { + "epoch": 2.236810977052283, + "grad_norm": 3.3372766971588135, + "learning_rate": 4.713741920038305e-06, + "loss": 0.3141, + "step": 18910 + }, + { + "epoch": 2.237993849065531, + "grad_norm": 3.158478260040283, + "learning_rate": 4.713023701220972e-06, + "loss": 0.3711, + "step": 18920 + }, + { + "epoch": 2.2391767210787794, + "grad_norm": 2.5206143856048584, + "learning_rate": 4.712305482403639e-06, + "loss": 0.2905, + "step": 18930 + }, + { + "epoch": 2.2403595930920273, + "grad_norm": 1.9004238843917847, + "learning_rate": 4.711587263586306e-06, + "loss": 0.3032, + "step": 18940 + }, + { + "epoch": 2.2415424651052756, + "grad_norm": 2.576606512069702, + "learning_rate": 4.710869044768973e-06, + "loss": 0.3181, + "step": 18950 + }, + { + "epoch": 2.242725337118524, + "grad_norm": 3.5385725498199463, + "learning_rate": 4.710150825951641e-06, + "loss": 0.3029, + "step": 18960 + }, + { + "epoch": 2.243908209131772, + "grad_norm": 3.0396499633789062, + "learning_rate": 4.709432607134307e-06, + "loss": 0.3044, + "step": 18970 + }, + { + "epoch": 2.24509108114502, + "grad_norm": 1.7584151029586792, + "learning_rate": 4.7087143883169746e-06, + "loss": 0.3075, + "step": 18980 + }, + { + "epoch": 2.2462739531582683, + "grad_norm": 2.714115858078003, + "learning_rate": 4.707996169499641e-06, + "loss": 0.3015, + "step": 18990 + }, + { + "epoch": 2.2474568251715166, + "grad_norm": 2.1493730545043945, + "learning_rate": 4.707277950682308e-06, + "loss": 0.3111, + "step": 19000 + }, + { + "epoch": 2.2486396971847644, + "grad_norm": 3.415705680847168, + "learning_rate": 4.7065597318649745e-06, + "loss": 0.3232, + "step": 19010 + }, + { + "epoch": 2.2498225691980127, + "grad_norm": 3.356984853744507, + "learning_rate": 4.705841513047642e-06, + "loss": 0.3268, + "step": 19020 + }, + { + "epoch": 2.250532292405962, + "eval_accuracy": 0.8579532222111321, + "eval_loss": 0.33104264736175537, + "eval_runtime": 77.9569, + "eval_safe_aucpr": 0.9135082414503558, + "eval_safe_f1": 0.8382121677182213, + "eval_safe_fpr": 0.1194583121580816, + "eval_safe_precision": 0.8469903507428397, + "eval_safe_recall": 0.829614071934891, + "eval_samples_per_second": 771.118, + "eval_steps_per_second": 12.058, + "eval_unsafe_aucpr": 0.9516369514480087, + "eval_unsafe_f1": 0.8734006434491245, + "eval_unsafe_fpr": 0.17038592806510833, + "eval_unsafe_precision": 0.8663744926172128, + "eval_unsafe_recall": 0.880541687841918, + "step": 19026 + }, + { + "epoch": 2.251005441211261, + "grad_norm": 4.034279823303223, + "learning_rate": 4.705123294230309e-06, + "loss": 0.2868, + "step": 19030 + }, + { + "epoch": 2.2521883132245093, + "grad_norm": 2.953068733215332, + "learning_rate": 4.704405075412976e-06, + "loss": 0.3178, + "step": 19040 + }, + { + "epoch": 2.253371185237757, + "grad_norm": 3.3599815368652344, + "learning_rate": 4.703686856595643e-06, + "loss": 0.321, + "step": 19050 + }, + { + "epoch": 2.2545540572510054, + "grad_norm": 3.519252300262451, + "learning_rate": 4.70296863777831e-06, + "loss": 0.3394, + "step": 19060 + }, + { + "epoch": 2.2557369292642537, + "grad_norm": 2.8658931255340576, + "learning_rate": 4.702250418960977e-06, + "loss": 0.3005, + "step": 19070 + }, + { + "epoch": 2.2569198012775016, + "grad_norm": 3.4949440956115723, + "learning_rate": 4.701532200143644e-06, + "loss": 0.2953, + "step": 19080 + }, + { + "epoch": 2.25810267329075, + "grad_norm": 3.6394600868225098, + "learning_rate": 4.700813981326311e-06, + "loss": 0.3103, + "step": 19090 + }, + { + "epoch": 2.259285545303998, + "grad_norm": 2.5448102951049805, + "learning_rate": 4.700095762508978e-06, + "loss": 0.3538, + "step": 19100 + }, + { + "epoch": 2.2604684173172465, + "grad_norm": 3.217200517654419, + "learning_rate": 4.6993775436916446e-06, + "loss": 0.3279, + "step": 19110 + }, + { + "epoch": 2.2616512893304943, + "grad_norm": 3.0768990516662598, + "learning_rate": 4.698659324874312e-06, + "loss": 0.2874, + "step": 19120 + }, + { + "epoch": 2.2628341613437426, + "grad_norm": 2.182265281677246, + "learning_rate": 4.697941106056978e-06, + "loss": 0.3065, + "step": 19130 + }, + { + "epoch": 2.264017033356991, + "grad_norm": 2.897308111190796, + "learning_rate": 4.697222887239646e-06, + "loss": 0.322, + "step": 19140 + }, + { + "epoch": 2.265199905370239, + "grad_norm": 3.513573408126831, + "learning_rate": 4.696504668422312e-06, + "loss": 0.3455, + "step": 19150 + }, + { + "epoch": 2.266382777383487, + "grad_norm": 2.9148173332214355, + "learning_rate": 4.69578644960498e-06, + "loss": 0.3234, + "step": 19160 + }, + { + "epoch": 2.2675656493967353, + "grad_norm": 2.3201258182525635, + "learning_rate": 4.695068230787646e-06, + "loss": 0.3156, + "step": 19170 + }, + { + "epoch": 2.2687485214099836, + "grad_norm": 2.708378791809082, + "learning_rate": 4.694350011970314e-06, + "loss": 0.3013, + "step": 19180 + }, + { + "epoch": 2.2699313934232315, + "grad_norm": 3.0447473526000977, + "learning_rate": 4.693631793152981e-06, + "loss": 0.3114, + "step": 19190 + }, + { + "epoch": 2.2711142654364798, + "grad_norm": 3.66853666305542, + "learning_rate": 4.692913574335648e-06, + "loss": 0.2915, + "step": 19200 + }, + { + "epoch": 2.272297137449728, + "grad_norm": 3.050086498260498, + "learning_rate": 4.692195355518315e-06, + "loss": 0.3295, + "step": 19210 + }, + { + "epoch": 2.2734800094629763, + "grad_norm": 3.328956365585327, + "learning_rate": 4.6914771367009816e-06, + "loss": 0.3121, + "step": 19220 + }, + { + "epoch": 2.274662881476224, + "grad_norm": 1.9344512224197388, + "learning_rate": 4.690758917883649e-06, + "loss": 0.2827, + "step": 19230 + }, + { + "epoch": 2.2758457534894725, + "grad_norm": 3.0727407932281494, + "learning_rate": 4.690040699066315e-06, + "loss": 0.2722, + "step": 19240 + }, + { + "epoch": 2.277028625502721, + "grad_norm": 2.5586421489715576, + "learning_rate": 4.689322480248983e-06, + "loss": 0.2848, + "step": 19250 + }, + { + "epoch": 2.2782114975159686, + "grad_norm": 2.8167285919189453, + "learning_rate": 4.688604261431649e-06, + "loss": 0.3229, + "step": 19260 + }, + { + "epoch": 2.279394369529217, + "grad_norm": 3.2754266262054443, + "learning_rate": 4.687886042614317e-06, + "loss": 0.3083, + "step": 19270 + }, + { + "epoch": 2.280577241542465, + "grad_norm": 2.7070839405059814, + "learning_rate": 4.687167823796983e-06, + "loss": 0.3037, + "step": 19280 + }, + { + "epoch": 2.2817601135557135, + "grad_norm": 3.1131722927093506, + "learning_rate": 4.686449604979651e-06, + "loss": 0.2822, + "step": 19290 + }, + { + "epoch": 2.2829429855689614, + "grad_norm": 3.0373032093048096, + "learning_rate": 4.685731386162318e-06, + "loss": 0.3162, + "step": 19300 + }, + { + "epoch": 2.2841258575822097, + "grad_norm": 2.733614444732666, + "learning_rate": 4.685013167344985e-06, + "loss": 0.3181, + "step": 19310 + }, + { + "epoch": 2.285308729595458, + "grad_norm": 2.5604472160339355, + "learning_rate": 4.684294948527652e-06, + "loss": 0.319, + "step": 19320 + }, + { + "epoch": 2.286491601608706, + "grad_norm": 2.2983901500701904, + "learning_rate": 4.6835767297103186e-06, + "loss": 0.2795, + "step": 19330 + }, + { + "epoch": 2.287674473621954, + "grad_norm": 3.369814872741699, + "learning_rate": 4.6828585108929855e-06, + "loss": 0.2908, + "step": 19340 + }, + { + "epoch": 2.2888573456352024, + "grad_norm": 2.4933390617370605, + "learning_rate": 4.682140292075652e-06, + "loss": 0.2893, + "step": 19350 + }, + { + "epoch": 2.2900402176484507, + "grad_norm": 3.431755304336548, + "learning_rate": 4.681422073258319e-06, + "loss": 0.2856, + "step": 19360 + }, + { + "epoch": 2.2912230896616985, + "grad_norm": 3.2312777042388916, + "learning_rate": 4.680703854440986e-06, + "loss": 0.2634, + "step": 19370 + }, + { + "epoch": 2.292405961674947, + "grad_norm": 2.5952422618865967, + "learning_rate": 4.679985635623653e-06, + "loss": 0.3406, + "step": 19380 + }, + { + "epoch": 2.293588833688195, + "grad_norm": 2.603264093399048, + "learning_rate": 4.679267416806321e-06, + "loss": 0.3124, + "step": 19390 + }, + { + "epoch": 2.294771705701443, + "grad_norm": 2.6363089084625244, + "learning_rate": 4.678549197988987e-06, + "loss": 0.2636, + "step": 19400 + }, + { + "epoch": 2.2959545777146912, + "grad_norm": 3.3012826442718506, + "learning_rate": 4.677830979171655e-06, + "loss": 0.2939, + "step": 19410 + }, + { + "epoch": 2.2971374497279395, + "grad_norm": 3.1667308807373047, + "learning_rate": 4.677112760354321e-06, + "loss": 0.285, + "step": 19420 + }, + { + "epoch": 2.298320321741188, + "grad_norm": 2.783691167831421, + "learning_rate": 4.676394541536989e-06, + "loss": 0.336, + "step": 19430 + }, + { + "epoch": 2.2995031937544357, + "grad_norm": 2.593380928039551, + "learning_rate": 4.675676322719655e-06, + "loss": 0.3245, + "step": 19440 + }, + { + "epoch": 2.300686065767684, + "grad_norm": 2.234391450881958, + "learning_rate": 4.6749581039023225e-06, + "loss": 0.322, + "step": 19450 + }, + { + "epoch": 2.3018689377809323, + "grad_norm": 2.6867315769195557, + "learning_rate": 4.674239885084989e-06, + "loss": 0.2845, + "step": 19460 + }, + { + "epoch": 2.30305180979418, + "grad_norm": 3.1070406436920166, + "learning_rate": 4.673521666267656e-06, + "loss": 0.2986, + "step": 19470 + }, + { + "epoch": 2.3042346818074284, + "grad_norm": 2.7052433490753174, + "learning_rate": 4.672803447450323e-06, + "loss": 0.3074, + "step": 19480 + }, + { + "epoch": 2.3054175538206767, + "grad_norm": 4.0416460037231445, + "learning_rate": 4.67208522863299e-06, + "loss": 0.3245, + "step": 19490 + }, + { + "epoch": 2.306600425833925, + "grad_norm": 2.0998635292053223, + "learning_rate": 4.671367009815658e-06, + "loss": 0.3209, + "step": 19500 + }, + { + "epoch": 2.307783297847173, + "grad_norm": 2.429802179336548, + "learning_rate": 4.670648790998324e-06, + "loss": 0.309, + "step": 19510 + }, + { + "epoch": 2.308966169860421, + "grad_norm": 2.9419801235198975, + "learning_rate": 4.669930572180992e-06, + "loss": 0.2895, + "step": 19520 + }, + { + "epoch": 2.3101490418736694, + "grad_norm": 2.927443027496338, + "learning_rate": 4.669212353363658e-06, + "loss": 0.334, + "step": 19530 + }, + { + "epoch": 2.3113319138869173, + "grad_norm": 3.3526933193206787, + "learning_rate": 4.668494134546326e-06, + "loss": 0.3067, + "step": 19540 + }, + { + "epoch": 2.3125147859001656, + "grad_norm": 2.4565911293029785, + "learning_rate": 4.667775915728992e-06, + "loss": 0.3019, + "step": 19550 + }, + { + "epoch": 2.313697657913414, + "grad_norm": 2.520803928375244, + "learning_rate": 4.6670576969116595e-06, + "loss": 0.3434, + "step": 19560 + }, + { + "epoch": 2.314880529926662, + "grad_norm": 3.4721646308898926, + "learning_rate": 4.666339478094326e-06, + "loss": 0.2899, + "step": 19570 + }, + { + "epoch": 2.31606340193991, + "grad_norm": 2.2394354343414307, + "learning_rate": 4.665621259276993e-06, + "loss": 0.326, + "step": 19580 + }, + { + "epoch": 2.3172462739531583, + "grad_norm": 2.9581689834594727, + "learning_rate": 4.66490304045966e-06, + "loss": 0.328, + "step": 19590 + }, + { + "epoch": 2.3184291459664066, + "grad_norm": 3.223263740539551, + "learning_rate": 4.664184821642327e-06, + "loss": 0.2757, + "step": 19600 + }, + { + "epoch": 2.3196120179796544, + "grad_norm": 3.566864490509033, + "learning_rate": 4.663466602824994e-06, + "loss": 0.2946, + "step": 19610 + }, + { + "epoch": 2.3207948899929027, + "grad_norm": 3.2584853172302246, + "learning_rate": 4.662748384007661e-06, + "loss": 0.2535, + "step": 19620 + }, + { + "epoch": 2.321977762006151, + "grad_norm": 2.983661413192749, + "learning_rate": 4.662030165190328e-06, + "loss": 0.3433, + "step": 19630 + }, + { + "epoch": 2.3231606340193993, + "grad_norm": 2.6936511993408203, + "learning_rate": 4.661311946372995e-06, + "loss": 0.2743, + "step": 19640 + }, + { + "epoch": 2.324343506032647, + "grad_norm": 2.610374927520752, + "learning_rate": 4.660593727555662e-06, + "loss": 0.2931, + "step": 19650 + }, + { + "epoch": 2.3255263780458955, + "grad_norm": 2.68072247505188, + "learning_rate": 4.6598755087383295e-06, + "loss": 0.2751, + "step": 19660 + }, + { + "epoch": 2.3267092500591438, + "grad_norm": 2.336444139480591, + "learning_rate": 4.659157289920996e-06, + "loss": 0.2942, + "step": 19670 + }, + { + "epoch": 2.3278921220723916, + "grad_norm": 2.8506205081939697, + "learning_rate": 4.658439071103663e-06, + "loss": 0.275, + "step": 19680 + }, + { + "epoch": 2.32907499408564, + "grad_norm": 2.9952659606933594, + "learning_rate": 4.6577208522863295e-06, + "loss": 0.3054, + "step": 19690 + }, + { + "epoch": 2.330257866098888, + "grad_norm": 1.9596766233444214, + "learning_rate": 4.657002633468997e-06, + "loss": 0.2583, + "step": 19700 + }, + { + "epoch": 2.3314407381121365, + "grad_norm": 2.6753013134002686, + "learning_rate": 4.656284414651664e-06, + "loss": 0.2724, + "step": 19710 + }, + { + "epoch": 2.3326236101253843, + "grad_norm": 4.986375331878662, + "learning_rate": 4.655566195834331e-06, + "loss": 0.2966, + "step": 19720 + }, + { + "epoch": 2.3338064821386326, + "grad_norm": 2.780099630355835, + "learning_rate": 4.654847977016998e-06, + "loss": 0.3073, + "step": 19730 + }, + { + "epoch": 2.334989354151881, + "grad_norm": 3.2119815349578857, + "learning_rate": 4.654129758199665e-06, + "loss": 0.3336, + "step": 19740 + }, + { + "epoch": 2.3361722261651288, + "grad_norm": 3.1022114753723145, + "learning_rate": 4.653411539382332e-06, + "loss": 0.2799, + "step": 19750 + }, + { + "epoch": 2.337355098178377, + "grad_norm": 3.45047664642334, + "learning_rate": 4.652693320564999e-06, + "loss": 0.2875, + "step": 19760 + }, + { + "epoch": 2.3385379701916253, + "grad_norm": 2.6528635025024414, + "learning_rate": 4.6519751017476665e-06, + "loss": 0.3256, + "step": 19770 + }, + { + "epoch": 2.3397208422048736, + "grad_norm": 2.416339159011841, + "learning_rate": 4.651256882930333e-06, + "loss": 0.304, + "step": 19780 + }, + { + "epoch": 2.3409037142181215, + "grad_norm": 2.61565899848938, + "learning_rate": 4.650538664113e-06, + "loss": 0.318, + "step": 19790 + }, + { + "epoch": 2.34208658623137, + "grad_norm": 2.7060976028442383, + "learning_rate": 4.6498204452956665e-06, + "loss": 0.2995, + "step": 19800 + }, + { + "epoch": 2.343269458244618, + "grad_norm": 3.9703264236450195, + "learning_rate": 4.649102226478334e-06, + "loss": 0.3036, + "step": 19810 + }, + { + "epoch": 2.344452330257866, + "grad_norm": 2.4769108295440674, + "learning_rate": 4.648384007661e-06, + "loss": 0.3118, + "step": 19820 + }, + { + "epoch": 2.345635202271114, + "grad_norm": 3.06601881980896, + "learning_rate": 4.647665788843668e-06, + "loss": 0.2861, + "step": 19830 + }, + { + "epoch": 2.3468180742843625, + "grad_norm": 2.9873290061950684, + "learning_rate": 4.646947570026335e-06, + "loss": 0.2802, + "step": 19840 + }, + { + "epoch": 2.348000946297611, + "grad_norm": 3.300034284591675, + "learning_rate": 4.646229351209002e-06, + "loss": 0.2915, + "step": 19850 + }, + { + "epoch": 2.3491838183108587, + "grad_norm": 2.580639362335205, + "learning_rate": 4.645511132391669e-06, + "loss": 0.2945, + "step": 19860 + }, + { + "epoch": 2.350366690324107, + "grad_norm": 3.396291494369507, + "learning_rate": 4.644792913574336e-06, + "loss": 0.2461, + "step": 19870 + }, + { + "epoch": 2.3515495623373552, + "grad_norm": 3.3407390117645264, + "learning_rate": 4.644074694757003e-06, + "loss": 0.339, + "step": 19880 + }, + { + "epoch": 2.352732434350603, + "grad_norm": 2.8220138549804688, + "learning_rate": 4.64335647593967e-06, + "loss": 0.34, + "step": 19890 + }, + { + "epoch": 2.3539153063638514, + "grad_norm": 2.042041778564453, + "learning_rate": 4.6426382571223365e-06, + "loss": 0.2851, + "step": 19900 + }, + { + "epoch": 2.3550981783770997, + "grad_norm": 3.4644346237182617, + "learning_rate": 4.6419200383050035e-06, + "loss": 0.3093, + "step": 19910 + }, + { + "epoch": 2.356281050390348, + "grad_norm": 2.6001265048980713, + "learning_rate": 4.64120181948767e-06, + "loss": 0.2824, + "step": 19920 + }, + { + "epoch": 2.357463922403596, + "grad_norm": 3.2763378620147705, + "learning_rate": 4.640483600670337e-06, + "loss": 0.3283, + "step": 19930 + }, + { + "epoch": 2.358646794416844, + "grad_norm": 2.4971261024475098, + "learning_rate": 4.639765381853004e-06, + "loss": 0.2916, + "step": 19940 + }, + { + "epoch": 2.3598296664300924, + "grad_norm": 2.6270179748535156, + "learning_rate": 4.639047163035672e-06, + "loss": 0.3105, + "step": 19950 + }, + { + "epoch": 2.3610125384433402, + "grad_norm": 3.1729302406311035, + "learning_rate": 4.638328944218338e-06, + "loss": 0.3487, + "step": 19960 + }, + { + "epoch": 2.3621954104565885, + "grad_norm": 3.2375006675720215, + "learning_rate": 4.637610725401006e-06, + "loss": 0.2896, + "step": 19970 + }, + { + "epoch": 2.363378282469837, + "grad_norm": 2.603255033493042, + "learning_rate": 4.636892506583673e-06, + "loss": 0.2954, + "step": 19980 + }, + { + "epoch": 2.364561154483085, + "grad_norm": 3.0251717567443848, + "learning_rate": 4.63617428776634e-06, + "loss": 0.308, + "step": 19990 + }, + { + "epoch": 2.365744026496333, + "grad_norm": 3.3653202056884766, + "learning_rate": 4.635456068949007e-06, + "loss": 0.3014, + "step": 20000 + }, + { + "epoch": 2.3669268985095813, + "grad_norm": 3.3685343265533447, + "learning_rate": 4.6347378501316735e-06, + "loss": 0.3402, + "step": 20010 + }, + { + "epoch": 2.3681097705228296, + "grad_norm": 3.4178988933563232, + "learning_rate": 4.6340196313143405e-06, + "loss": 0.303, + "step": 20020 + }, + { + "epoch": 2.3692926425360774, + "grad_norm": 2.6259653568267822, + "learning_rate": 4.633301412497007e-06, + "loss": 0.3156, + "step": 20030 + }, + { + "epoch": 2.3704755145493257, + "grad_norm": 2.317650318145752, + "learning_rate": 4.632583193679675e-06, + "loss": 0.346, + "step": 20040 + }, + { + "epoch": 2.371658386562574, + "grad_norm": 2.2929182052612305, + "learning_rate": 4.631864974862341e-06, + "loss": 0.2802, + "step": 20050 + }, + { + "epoch": 2.3728412585758223, + "grad_norm": 2.9007229804992676, + "learning_rate": 4.631146756045009e-06, + "loss": 0.3227, + "step": 20060 + }, + { + "epoch": 2.37402413058907, + "grad_norm": 2.8625431060791016, + "learning_rate": 4.630428537227675e-06, + "loss": 0.3436, + "step": 20070 + }, + { + "epoch": 2.3752070026023184, + "grad_norm": 2.63838791847229, + "learning_rate": 4.629710318410343e-06, + "loss": 0.2974, + "step": 20080 + }, + { + "epoch": 2.3763898746155667, + "grad_norm": 2.5006182193756104, + "learning_rate": 4.628992099593009e-06, + "loss": 0.2927, + "step": 20090 + }, + { + "epoch": 2.3775727466288146, + "grad_norm": 3.098212718963623, + "learning_rate": 4.628273880775677e-06, + "loss": 0.3077, + "step": 20100 + }, + { + "epoch": 2.378755618642063, + "grad_norm": 3.923412799835205, + "learning_rate": 4.627555661958344e-06, + "loss": 0.303, + "step": 20110 + }, + { + "epoch": 2.379938490655311, + "grad_norm": 3.1200942993164062, + "learning_rate": 4.6268374431410105e-06, + "loss": 0.3419, + "step": 20120 + }, + { + "epoch": 2.3811213626685594, + "grad_norm": 1.6198961734771729, + "learning_rate": 4.6261192243236775e-06, + "loss": 0.286, + "step": 20130 + }, + { + "epoch": 2.3823042346818073, + "grad_norm": 2.5041675567626953, + "learning_rate": 4.625401005506344e-06, + "loss": 0.3103, + "step": 20140 + }, + { + "epoch": 2.3834871066950556, + "grad_norm": 2.900592565536499, + "learning_rate": 4.624682786689011e-06, + "loss": 0.331, + "step": 20150 + }, + { + "epoch": 2.384669978708304, + "grad_norm": 2.689657688140869, + "learning_rate": 4.623964567871678e-06, + "loss": 0.3322, + "step": 20160 + }, + { + "epoch": 2.3858528507215517, + "grad_norm": 2.471005916595459, + "learning_rate": 4.623246349054345e-06, + "loss": 0.3073, + "step": 20170 + }, + { + "epoch": 2.3870357227348, + "grad_norm": 2.933124542236328, + "learning_rate": 4.622528130237012e-06, + "loss": 0.3033, + "step": 20180 + }, + { + "epoch": 2.3882185947480483, + "grad_norm": 2.789827585220337, + "learning_rate": 4.621809911419679e-06, + "loss": 0.2927, + "step": 20190 + }, + { + "epoch": 2.3894014667612966, + "grad_norm": 2.2694129943847656, + "learning_rate": 4.621091692602346e-06, + "loss": 0.3027, + "step": 20200 + }, + { + "epoch": 2.3905843387745445, + "grad_norm": 3.2975857257843018, + "learning_rate": 4.620373473785013e-06, + "loss": 0.3143, + "step": 20210 + }, + { + "epoch": 2.3917672107877928, + "grad_norm": 3.478466272354126, + "learning_rate": 4.619655254967681e-06, + "loss": 0.3402, + "step": 20220 + }, + { + "epoch": 2.392950082801041, + "grad_norm": 2.7437846660614014, + "learning_rate": 4.6189370361503475e-06, + "loss": 0.3057, + "step": 20230 + }, + { + "epoch": 2.394132954814289, + "grad_norm": 2.510840892791748, + "learning_rate": 4.6182188173330145e-06, + "loss": 0.3097, + "step": 20240 + }, + { + "epoch": 2.395315826827537, + "grad_norm": 3.0884249210357666, + "learning_rate": 4.617500598515681e-06, + "loss": 0.3152, + "step": 20250 + }, + { + "epoch": 2.3964986988407855, + "grad_norm": 2.712763786315918, + "learning_rate": 4.616782379698348e-06, + "loss": 0.3074, + "step": 20260 + }, + { + "epoch": 2.3976815708540338, + "grad_norm": 2.518413543701172, + "learning_rate": 4.616064160881015e-06, + "loss": 0.3171, + "step": 20270 + }, + { + "epoch": 2.3988644428672816, + "grad_norm": 3.805736541748047, + "learning_rate": 4.615345942063682e-06, + "loss": 0.2897, + "step": 20280 + }, + { + "epoch": 2.40004731488053, + "grad_norm": 3.109661817550659, + "learning_rate": 4.614627723246349e-06, + "loss": 0.3072, + "step": 20290 + }, + { + "epoch": 2.401230186893778, + "grad_norm": 4.400880336761475, + "learning_rate": 4.613909504429016e-06, + "loss": 0.3297, + "step": 20300 + }, + { + "epoch": 2.402413058907026, + "grad_norm": 2.9064347743988037, + "learning_rate": 4.613191285611684e-06, + "loss": 0.2985, + "step": 20310 + }, + { + "epoch": 2.4035959309202743, + "grad_norm": 3.0515923500061035, + "learning_rate": 4.61247306679435e-06, + "loss": 0.2738, + "step": 20320 + }, + { + "epoch": 2.4047788029335226, + "grad_norm": 3.078885793685913, + "learning_rate": 4.611754847977018e-06, + "loss": 0.3325, + "step": 20330 + }, + { + "epoch": 2.405961674946771, + "grad_norm": 2.3880527019500732, + "learning_rate": 4.611036629159684e-06, + "loss": 0.3024, + "step": 20340 + }, + { + "epoch": 2.407144546960019, + "grad_norm": 2.6147024631500244, + "learning_rate": 4.6103184103423515e-06, + "loss": 0.288, + "step": 20350 + }, + { + "epoch": 2.408327418973267, + "grad_norm": 2.7234206199645996, + "learning_rate": 4.6096001915250175e-06, + "loss": 0.3391, + "step": 20360 + }, + { + "epoch": 2.4095102909865154, + "grad_norm": 2.5554802417755127, + "learning_rate": 4.608881972707685e-06, + "loss": 0.2922, + "step": 20370 + }, + { + "epoch": 2.410693162999763, + "grad_norm": 2.6028528213500977, + "learning_rate": 4.608163753890352e-06, + "loss": 0.3206, + "step": 20380 + }, + { + "epoch": 2.4118760350130115, + "grad_norm": 2.4608428478240967, + "learning_rate": 4.607445535073019e-06, + "loss": 0.2791, + "step": 20390 + }, + { + "epoch": 2.41305890702626, + "grad_norm": 2.69881272315979, + "learning_rate": 4.606727316255686e-06, + "loss": 0.3024, + "step": 20400 + }, + { + "epoch": 2.414241779039508, + "grad_norm": 3.897646903991699, + "learning_rate": 4.606009097438353e-06, + "loss": 0.2985, + "step": 20410 + }, + { + "epoch": 2.415424651052756, + "grad_norm": 3.009213924407959, + "learning_rate": 4.60529087862102e-06, + "loss": 0.2716, + "step": 20420 + }, + { + "epoch": 2.4166075230660042, + "grad_norm": 3.926177740097046, + "learning_rate": 4.604572659803687e-06, + "loss": 0.2428, + "step": 20430 + }, + { + "epoch": 2.4177903950792525, + "grad_norm": 4.079202175140381, + "learning_rate": 4.603854440986354e-06, + "loss": 0.306, + "step": 20440 + }, + { + "epoch": 2.4189732670925004, + "grad_norm": 2.471909523010254, + "learning_rate": 4.603136222169021e-06, + "loss": 0.2836, + "step": 20450 + }, + { + "epoch": 2.4201561391057487, + "grad_norm": 2.6632494926452637, + "learning_rate": 4.602418003351688e-06, + "loss": 0.3114, + "step": 20460 + }, + { + "epoch": 2.421339011118997, + "grad_norm": 2.557344675064087, + "learning_rate": 4.6016997845343545e-06, + "loss": 0.253, + "step": 20470 + }, + { + "epoch": 2.4225218831322453, + "grad_norm": 3.323770046234131, + "learning_rate": 4.600981565717022e-06, + "loss": 0.3351, + "step": 20480 + }, + { + "epoch": 2.423704755145493, + "grad_norm": 2.374248504638672, + "learning_rate": 4.600263346899689e-06, + "loss": 0.3271, + "step": 20490 + }, + { + "epoch": 2.4248876271587414, + "grad_norm": 2.0508761405944824, + "learning_rate": 4.599545128082356e-06, + "loss": 0.2808, + "step": 20500 + }, + { + "epoch": 2.4260704991719897, + "grad_norm": 2.739738702774048, + "learning_rate": 4.598826909265023e-06, + "loss": 0.2765, + "step": 20510 + }, + { + "epoch": 2.4272533711852375, + "grad_norm": 1.862618327140808, + "learning_rate": 4.59810869044769e-06, + "loss": 0.3155, + "step": 20520 + }, + { + "epoch": 2.428436243198486, + "grad_norm": 3.4794418811798096, + "learning_rate": 4.597390471630357e-06, + "loss": 0.3207, + "step": 20530 + }, + { + "epoch": 2.429619115211734, + "grad_norm": 2.0952260494232178, + "learning_rate": 4.596672252813024e-06, + "loss": 0.2776, + "step": 20540 + }, + { + "epoch": 2.4308019872249824, + "grad_norm": 2.8477554321289062, + "learning_rate": 4.595954033995691e-06, + "loss": 0.3107, + "step": 20550 + }, + { + "epoch": 2.4319848592382303, + "grad_norm": 3.1266961097717285, + "learning_rate": 4.595235815178358e-06, + "loss": 0.2608, + "step": 20560 + }, + { + "epoch": 2.4331677312514786, + "grad_norm": 2.8391010761260986, + "learning_rate": 4.594517596361025e-06, + "loss": 0.2696, + "step": 20570 + }, + { + "epoch": 2.434350603264727, + "grad_norm": 1.6859503984451294, + "learning_rate": 4.5937993775436915e-06, + "loss": 0.3014, + "step": 20580 + }, + { + "epoch": 2.4355334752779747, + "grad_norm": 2.543356418609619, + "learning_rate": 4.5930811587263584e-06, + "loss": 0.3163, + "step": 20590 + }, + { + "epoch": 2.436716347291223, + "grad_norm": 2.9169318675994873, + "learning_rate": 4.592362939909026e-06, + "loss": 0.2907, + "step": 20600 + }, + { + "epoch": 2.4378992193044713, + "grad_norm": 3.6094706058502197, + "learning_rate": 4.591644721091692e-06, + "loss": 0.3256, + "step": 20610 + }, + { + "epoch": 2.4390820913177196, + "grad_norm": 3.3926591873168945, + "learning_rate": 4.59092650227436e-06, + "loss": 0.3021, + "step": 20620 + }, + { + "epoch": 2.4402649633309674, + "grad_norm": 3.370145797729492, + "learning_rate": 4.590208283457026e-06, + "loss": 0.3536, + "step": 20630 + }, + { + "epoch": 2.4414478353442157, + "grad_norm": 2.698267936706543, + "learning_rate": 4.589490064639694e-06, + "loss": 0.2693, + "step": 20640 + }, + { + "epoch": 2.442630707357464, + "grad_norm": 2.918368339538574, + "learning_rate": 4.588771845822361e-06, + "loss": 0.301, + "step": 20650 + }, + { + "epoch": 2.443813579370712, + "grad_norm": 3.023146867752075, + "learning_rate": 4.588053627005028e-06, + "loss": 0.3137, + "step": 20660 + }, + { + "epoch": 2.44499645138396, + "grad_norm": 3.6508171558380127, + "learning_rate": 4.587335408187695e-06, + "loss": 0.3164, + "step": 20670 + }, + { + "epoch": 2.4461793233972084, + "grad_norm": 4.299176216125488, + "learning_rate": 4.586617189370362e-06, + "loss": 0.2992, + "step": 20680 + }, + { + "epoch": 2.4473621954104567, + "grad_norm": 3.4763195514678955, + "learning_rate": 4.5858989705530285e-06, + "loss": 0.3068, + "step": 20690 + }, + { + "epoch": 2.4485450674237046, + "grad_norm": 4.034111022949219, + "learning_rate": 4.5851807517356954e-06, + "loss": 0.3909, + "step": 20700 + }, + { + "epoch": 2.449727939436953, + "grad_norm": 2.4070658683776855, + "learning_rate": 4.584462532918362e-06, + "loss": 0.2944, + "step": 20710 + }, + { + "epoch": 2.450910811450201, + "grad_norm": 3.4822850227355957, + "learning_rate": 4.583744314101029e-06, + "loss": 0.2817, + "step": 20720 + }, + { + "epoch": 2.452093683463449, + "grad_norm": 2.9878408908843994, + "learning_rate": 4.583026095283697e-06, + "loss": 0.2915, + "step": 20730 + }, + { + "epoch": 2.4532765554766973, + "grad_norm": 3.455247163772583, + "learning_rate": 4.582307876466363e-06, + "loss": 0.301, + "step": 20740 + }, + { + "epoch": 2.4544594274899456, + "grad_norm": 2.6818161010742188, + "learning_rate": 4.581589657649031e-06, + "loss": 0.286, + "step": 20750 + }, + { + "epoch": 2.455642299503194, + "grad_norm": 3.109696388244629, + "learning_rate": 4.580871438831698e-06, + "loss": 0.3193, + "step": 20760 + }, + { + "epoch": 2.4568251715164418, + "grad_norm": 2.9084906578063965, + "learning_rate": 4.580153220014365e-06, + "loss": 0.3013, + "step": 20770 + }, + { + "epoch": 2.45800804352969, + "grad_norm": 3.4805383682250977, + "learning_rate": 4.579435001197032e-06, + "loss": 0.2984, + "step": 20780 + }, + { + "epoch": 2.4591909155429383, + "grad_norm": 2.6101019382476807, + "learning_rate": 4.578716782379699e-06, + "loss": 0.3048, + "step": 20790 + }, + { + "epoch": 2.460373787556186, + "grad_norm": 2.232529640197754, + "learning_rate": 4.5779985635623655e-06, + "loss": 0.2833, + "step": 20800 + }, + { + "epoch": 2.4615566595694345, + "grad_norm": 3.5824155807495117, + "learning_rate": 4.5772803447450324e-06, + "loss": 0.3059, + "step": 20810 + }, + { + "epoch": 2.4627395315826828, + "grad_norm": 2.1113779544830322, + "learning_rate": 4.576562125927699e-06, + "loss": 0.3131, + "step": 20820 + }, + { + "epoch": 2.463922403595931, + "grad_norm": 3.2497670650482178, + "learning_rate": 4.575843907110366e-06, + "loss": 0.3125, + "step": 20830 + }, + { + "epoch": 2.465105275609179, + "grad_norm": 2.5704400539398193, + "learning_rate": 4.575125688293033e-06, + "loss": 0.3283, + "step": 20840 + }, + { + "epoch": 2.466288147622427, + "grad_norm": 2.7244441509246826, + "learning_rate": 4.5744074694757e-06, + "loss": 0.3156, + "step": 20850 + }, + { + "epoch": 2.4674710196356755, + "grad_norm": 2.3715591430664062, + "learning_rate": 4.573689250658367e-06, + "loss": 0.313, + "step": 20860 + }, + { + "epoch": 2.4686538916489233, + "grad_norm": 4.268498420715332, + "learning_rate": 4.572971031841035e-06, + "loss": 0.3363, + "step": 20870 + }, + { + "epoch": 2.4698367636621716, + "grad_norm": 3.0117766857147217, + "learning_rate": 4.572252813023701e-06, + "loss": 0.335, + "step": 20880 + }, + { + "epoch": 2.47101963567542, + "grad_norm": 3.2900314331054688, + "learning_rate": 4.571534594206369e-06, + "loss": 0.3027, + "step": 20890 + }, + { + "epoch": 2.4722025076886682, + "grad_norm": 2.577617883682251, + "learning_rate": 4.570816375389035e-06, + "loss": 0.2658, + "step": 20900 + }, + { + "epoch": 2.473385379701916, + "grad_norm": 3.5049071311950684, + "learning_rate": 4.5700981565717025e-06, + "loss": 0.2969, + "step": 20910 + }, + { + "epoch": 2.4745682517151644, + "grad_norm": 3.682277202606201, + "learning_rate": 4.569379937754369e-06, + "loss": 0.2825, + "step": 20920 + }, + { + "epoch": 2.4757511237284127, + "grad_norm": 2.5858194828033447, + "learning_rate": 4.568661718937036e-06, + "loss": 0.332, + "step": 20930 + }, + { + "epoch": 2.4769339957416605, + "grad_norm": 2.7255759239196777, + "learning_rate": 4.567943500119703e-06, + "loss": 0.3069, + "step": 20940 + }, + { + "epoch": 2.478116867754909, + "grad_norm": 2.588320016860962, + "learning_rate": 4.56722528130237e-06, + "loss": 0.3146, + "step": 20950 + }, + { + "epoch": 2.479299739768157, + "grad_norm": 3.010643482208252, + "learning_rate": 4.566507062485037e-06, + "loss": 0.2928, + "step": 20960 + }, + { + "epoch": 2.4804826117814054, + "grad_norm": 3.469820499420166, + "learning_rate": 4.565788843667704e-06, + "loss": 0.2786, + "step": 20970 + }, + { + "epoch": 2.4816654837946532, + "grad_norm": 2.7268991470336914, + "learning_rate": 4.565070624850372e-06, + "loss": 0.2685, + "step": 20980 + }, + { + "epoch": 2.4828483558079015, + "grad_norm": 4.038430690765381, + "learning_rate": 4.564352406033038e-06, + "loss": 0.2477, + "step": 20990 + }, + { + "epoch": 2.48403122782115, + "grad_norm": 2.6853692531585693, + "learning_rate": 4.563634187215706e-06, + "loss": 0.3062, + "step": 21000 + }, + { + "epoch": 2.4852140998343977, + "grad_norm": 2.2105298042297363, + "learning_rate": 4.562915968398372e-06, + "loss": 0.3348, + "step": 21010 + }, + { + "epoch": 2.486396971847646, + "grad_norm": 2.810724973678589, + "learning_rate": 4.5621977495810395e-06, + "loss": 0.2911, + "step": 21020 + }, + { + "epoch": 2.4875798438608943, + "grad_norm": 4.207290172576904, + "learning_rate": 4.5614795307637064e-06, + "loss": 0.2766, + "step": 21030 + }, + { + "epoch": 2.4887627158741426, + "grad_norm": 2.644639492034912, + "learning_rate": 4.560761311946373e-06, + "loss": 0.2967, + "step": 21040 + }, + { + "epoch": 2.4899455878873904, + "grad_norm": 3.2864556312561035, + "learning_rate": 4.56004309312904e-06, + "loss": 0.3415, + "step": 21050 + }, + { + "epoch": 2.4911284599006387, + "grad_norm": 3.14288330078125, + "learning_rate": 4.559324874311707e-06, + "loss": 0.2855, + "step": 21060 + }, + { + "epoch": 2.492311331913887, + "grad_norm": 2.1075081825256348, + "learning_rate": 4.558606655494374e-06, + "loss": 0.2802, + "step": 21070 + }, + { + "epoch": 2.4934942039271353, + "grad_norm": 3.4036238193511963, + "learning_rate": 4.557888436677041e-06, + "loss": 0.2902, + "step": 21080 + }, + { + "epoch": 2.494677075940383, + "grad_norm": 2.650482654571533, + "learning_rate": 4.557170217859708e-06, + "loss": 0.2826, + "step": 21090 + }, + { + "epoch": 2.4958599479536314, + "grad_norm": 3.0203163623809814, + "learning_rate": 4.556451999042375e-06, + "loss": 0.2931, + "step": 21100 + }, + { + "epoch": 2.4970428199668797, + "grad_norm": 3.077556848526001, + "learning_rate": 4.555733780225042e-06, + "loss": 0.3441, + "step": 21110 + }, + { + "epoch": 2.4982256919801276, + "grad_norm": 1.740944504737854, + "learning_rate": 4.555015561407709e-06, + "loss": 0.2761, + "step": 21120 + }, + { + "epoch": 2.499408563993376, + "grad_norm": 3.0141775608062744, + "learning_rate": 4.554297342590376e-06, + "loss": 0.3299, + "step": 21130 + }, + { + "epoch": 2.500591436006624, + "grad_norm": 3.199282646179199, + "learning_rate": 4.5535791237730434e-06, + "loss": 0.2752, + "step": 21140 + }, + { + "epoch": 2.500591436006624, + "eval_accuracy": 0.8580031273912899, + "eval_loss": 0.3318200707435608, + "eval_runtime": 77.8696, + "eval_safe_aucpr": 0.9148747208601448, + "eval_safe_f1": 0.8377741457295982, + "eval_safe_fpr": 0.11700696541209495, + "eval_safe_precision": 0.8492005393951069, + "eval_safe_recall": 0.8266511645351236, + "eval_samples_per_second": 771.983, + "eval_steps_per_second": 12.071, + "eval_unsafe_aucpr": 0.95208993937996, + "eval_unsafe_f1": 0.8737464872060346, + "eval_unsafe_fpr": 0.17334883546487578, + "eval_unsafe_precision": 0.8646915893322404, + "eval_unsafe_recall": 0.8829930345879047, + "step": 21140 + }, + { + "epoch": 2.501774308019872, + "grad_norm": 3.0372202396392822, + "learning_rate": 4.5528609049557095e-06, + "loss": 0.3057, + "step": 21150 + }, + { + "epoch": 2.5029571800331203, + "grad_norm": 2.6270434856414795, + "learning_rate": 4.552142686138377e-06, + "loss": 0.2973, + "step": 21160 + }, + { + "epoch": 2.5041400520463686, + "grad_norm": 3.112229108810425, + "learning_rate": 4.551424467321043e-06, + "loss": 0.3133, + "step": 21170 + }, + { + "epoch": 2.505322924059617, + "grad_norm": 2.4659974575042725, + "learning_rate": 4.550706248503711e-06, + "loss": 0.3142, + "step": 21180 + }, + { + "epoch": 2.506505796072865, + "grad_norm": 3.3613669872283936, + "learning_rate": 4.549988029686377e-06, + "loss": 0.3077, + "step": 21190 + }, + { + "epoch": 2.507688668086113, + "grad_norm": 4.1130781173706055, + "learning_rate": 4.549269810869045e-06, + "loss": 0.3078, + "step": 21200 + }, + { + "epoch": 2.5088715400993613, + "grad_norm": 2.4725160598754883, + "learning_rate": 4.548551592051712e-06, + "loss": 0.2686, + "step": 21210 + }, + { + "epoch": 2.510054412112609, + "grad_norm": 4.3177103996276855, + "learning_rate": 4.547833373234379e-06, + "loss": 0.3091, + "step": 21220 + }, + { + "epoch": 2.5112372841258574, + "grad_norm": 3.7701759338378906, + "learning_rate": 4.547115154417047e-06, + "loss": 0.3035, + "step": 21230 + }, + { + "epoch": 2.5124201561391057, + "grad_norm": 2.6487650871276855, + "learning_rate": 4.546396935599713e-06, + "loss": 0.2616, + "step": 21240 + }, + { + "epoch": 2.513603028152354, + "grad_norm": 2.873460292816162, + "learning_rate": 4.5456787167823804e-06, + "loss": 0.3269, + "step": 21250 + }, + { + "epoch": 2.5147859001656023, + "grad_norm": 3.469243288040161, + "learning_rate": 4.5449604979650465e-06, + "loss": 0.3474, + "step": 21260 + }, + { + "epoch": 2.51596877217885, + "grad_norm": 2.8477437496185303, + "learning_rate": 4.544242279147714e-06, + "loss": 0.3172, + "step": 21270 + }, + { + "epoch": 2.5171516441920985, + "grad_norm": 3.201421022415161, + "learning_rate": 4.54352406033038e-06, + "loss": 0.3075, + "step": 21280 + }, + { + "epoch": 2.5183345162053463, + "grad_norm": 2.6932663917541504, + "learning_rate": 4.542805841513048e-06, + "loss": 0.2892, + "step": 21290 + }, + { + "epoch": 2.5195173882185946, + "grad_norm": 2.7699522972106934, + "learning_rate": 4.542087622695715e-06, + "loss": 0.3032, + "step": 21300 + }, + { + "epoch": 2.520700260231843, + "grad_norm": 3.174916982650757, + "learning_rate": 4.541369403878382e-06, + "loss": 0.2635, + "step": 21310 + }, + { + "epoch": 2.521883132245091, + "grad_norm": 2.373048782348633, + "learning_rate": 4.540651185061049e-06, + "loss": 0.2538, + "step": 21320 + }, + { + "epoch": 2.5230660042583395, + "grad_norm": 2.9634103775024414, + "learning_rate": 4.539932966243716e-06, + "loss": 0.3142, + "step": 21330 + }, + { + "epoch": 2.5242488762715873, + "grad_norm": 3.273416042327881, + "learning_rate": 4.539214747426383e-06, + "loss": 0.3097, + "step": 21340 + }, + { + "epoch": 2.5254317482848356, + "grad_norm": 2.718534231185913, + "learning_rate": 4.53849652860905e-06, + "loss": 0.2889, + "step": 21350 + }, + { + "epoch": 2.5266146202980835, + "grad_norm": 3.2437613010406494, + "learning_rate": 4.537778309791717e-06, + "loss": 0.3551, + "step": 21360 + }, + { + "epoch": 2.5277974923113318, + "grad_norm": 3.0339436531066895, + "learning_rate": 4.5370600909743835e-06, + "loss": 0.292, + "step": 21370 + }, + { + "epoch": 2.52898036432458, + "grad_norm": 2.994718074798584, + "learning_rate": 4.5363418721570504e-06, + "loss": 0.3065, + "step": 21380 + }, + { + "epoch": 2.5301632363378284, + "grad_norm": 3.1579229831695557, + "learning_rate": 4.535623653339717e-06, + "loss": 0.2907, + "step": 21390 + }, + { + "epoch": 2.5313461083510767, + "grad_norm": 3.506098985671997, + "learning_rate": 4.534905434522384e-06, + "loss": 0.2769, + "step": 21400 + }, + { + "epoch": 2.5325289803643245, + "grad_norm": 3.2988998889923096, + "learning_rate": 4.534187215705052e-06, + "loss": 0.3311, + "step": 21410 + }, + { + "epoch": 2.533711852377573, + "grad_norm": 3.10424542427063, + "learning_rate": 4.533468996887718e-06, + "loss": 0.3184, + "step": 21420 + }, + { + "epoch": 2.5348947243908206, + "grad_norm": 2.427213430404663, + "learning_rate": 4.532750778070386e-06, + "loss": 0.2945, + "step": 21430 + }, + { + "epoch": 2.536077596404069, + "grad_norm": 2.4065308570861816, + "learning_rate": 4.532032559253052e-06, + "loss": 0.3029, + "step": 21440 + }, + { + "epoch": 2.5372604684173172, + "grad_norm": 2.4313242435455322, + "learning_rate": 4.53131434043572e-06, + "loss": 0.3207, + "step": 21450 + }, + { + "epoch": 2.5384433404305655, + "grad_norm": 3.308527946472168, + "learning_rate": 4.530596121618386e-06, + "loss": 0.3129, + "step": 21460 + }, + { + "epoch": 2.539626212443814, + "grad_norm": 3.2686848640441895, + "learning_rate": 4.529877902801054e-06, + "loss": 0.3103, + "step": 21470 + }, + { + "epoch": 2.5408090844570617, + "grad_norm": 3.570476531982422, + "learning_rate": 4.5291596839837205e-06, + "loss": 0.3136, + "step": 21480 + }, + { + "epoch": 2.54199195647031, + "grad_norm": 2.1805944442749023, + "learning_rate": 4.5284414651663874e-06, + "loss": 0.2934, + "step": 21490 + }, + { + "epoch": 2.543174828483558, + "grad_norm": 2.063673496246338, + "learning_rate": 4.527723246349054e-06, + "loss": 0.2818, + "step": 21500 + }, + { + "epoch": 2.544357700496806, + "grad_norm": 2.700390100479126, + "learning_rate": 4.527005027531721e-06, + "loss": 0.2917, + "step": 21510 + }, + { + "epoch": 2.5455405725100544, + "grad_norm": 2.2416744232177734, + "learning_rate": 4.526286808714389e-06, + "loss": 0.2994, + "step": 21520 + }, + { + "epoch": 2.5467234445233027, + "grad_norm": 3.3600687980651855, + "learning_rate": 4.525568589897055e-06, + "loss": 0.3007, + "step": 21530 + }, + { + "epoch": 2.547906316536551, + "grad_norm": 3.249323844909668, + "learning_rate": 4.524850371079723e-06, + "loss": 0.3158, + "step": 21540 + }, + { + "epoch": 2.549089188549799, + "grad_norm": 4.444723129272461, + "learning_rate": 4.524132152262389e-06, + "loss": 0.3219, + "step": 21550 + }, + { + "epoch": 2.550272060563047, + "grad_norm": 4.1026434898376465, + "learning_rate": 4.523413933445057e-06, + "loss": 0.3285, + "step": 21560 + }, + { + "epoch": 2.5514549325762954, + "grad_norm": 3.7367851734161377, + "learning_rate": 4.522695714627724e-06, + "loss": 0.2852, + "step": 21570 + }, + { + "epoch": 2.5526378045895433, + "grad_norm": 3.6800010204315186, + "learning_rate": 4.521977495810391e-06, + "loss": 0.3259, + "step": 21580 + }, + { + "epoch": 2.5538206766027916, + "grad_norm": 2.178433418273926, + "learning_rate": 4.5212592769930575e-06, + "loss": 0.311, + "step": 21590 + }, + { + "epoch": 2.55500354861604, + "grad_norm": 2.957578420639038, + "learning_rate": 4.5205410581757244e-06, + "loss": 0.3059, + "step": 21600 + }, + { + "epoch": 2.556186420629288, + "grad_norm": 2.5414247512817383, + "learning_rate": 4.519822839358391e-06, + "loss": 0.2645, + "step": 21610 + }, + { + "epoch": 2.557369292642536, + "grad_norm": 3.454031467437744, + "learning_rate": 4.519104620541058e-06, + "loss": 0.3247, + "step": 21620 + }, + { + "epoch": 2.5585521646557843, + "grad_norm": 2.417210817337036, + "learning_rate": 4.518386401723725e-06, + "loss": 0.266, + "step": 21630 + }, + { + "epoch": 2.5597350366690326, + "grad_norm": 2.803642988204956, + "learning_rate": 4.517668182906392e-06, + "loss": 0.3107, + "step": 21640 + }, + { + "epoch": 2.5609179086822804, + "grad_norm": 4.093634128570557, + "learning_rate": 4.516949964089059e-06, + "loss": 0.3397, + "step": 21650 + }, + { + "epoch": 2.5621007806955287, + "grad_norm": 2.1972320079803467, + "learning_rate": 4.516231745271726e-06, + "loss": 0.3163, + "step": 21660 + }, + { + "epoch": 2.563283652708777, + "grad_norm": 2.724701166152954, + "learning_rate": 4.515513526454393e-06, + "loss": 0.3146, + "step": 21670 + }, + { + "epoch": 2.5644665247220253, + "grad_norm": 2.9636809825897217, + "learning_rate": 4.514795307637061e-06, + "loss": 0.2948, + "step": 21680 + }, + { + "epoch": 2.565649396735273, + "grad_norm": 1.956942081451416, + "learning_rate": 4.514077088819727e-06, + "loss": 0.3168, + "step": 21690 + }, + { + "epoch": 2.5668322687485214, + "grad_norm": 3.7892258167266846, + "learning_rate": 4.5133588700023945e-06, + "loss": 0.293, + "step": 21700 + }, + { + "epoch": 2.5680151407617697, + "grad_norm": 3.2520649433135986, + "learning_rate": 4.512640651185061e-06, + "loss": 0.3285, + "step": 21710 + }, + { + "epoch": 2.5691980127750176, + "grad_norm": 2.8439433574676514, + "learning_rate": 4.511922432367728e-06, + "loss": 0.2665, + "step": 21720 + }, + { + "epoch": 2.570380884788266, + "grad_norm": 2.6865758895874023, + "learning_rate": 4.5112042135503944e-06, + "loss": 0.2898, + "step": 21730 + }, + { + "epoch": 2.571563756801514, + "grad_norm": 3.938616991043091, + "learning_rate": 4.510485994733062e-06, + "loss": 0.3113, + "step": 21740 + }, + { + "epoch": 2.5727466288147625, + "grad_norm": 3.2345218658447266, + "learning_rate": 4.509767775915729e-06, + "loss": 0.3455, + "step": 21750 + }, + { + "epoch": 2.5739295008280103, + "grad_norm": 2.0615406036376953, + "learning_rate": 4.509049557098396e-06, + "loss": 0.2942, + "step": 21760 + }, + { + "epoch": 2.5751123728412586, + "grad_norm": 2.9529225826263428, + "learning_rate": 4.508331338281063e-06, + "loss": 0.279, + "step": 21770 + }, + { + "epoch": 2.576295244854507, + "grad_norm": 3.24729061126709, + "learning_rate": 4.50761311946373e-06, + "loss": 0.3042, + "step": 21780 + }, + { + "epoch": 2.5774781168677547, + "grad_norm": 4.246401786804199, + "learning_rate": 4.506894900646398e-06, + "loss": 0.2974, + "step": 21790 + }, + { + "epoch": 2.578660988881003, + "grad_norm": 2.7710227966308594, + "learning_rate": 4.506176681829064e-06, + "loss": 0.3063, + "step": 21800 + }, + { + "epoch": 2.5798438608942513, + "grad_norm": 3.360891342163086, + "learning_rate": 4.5054584630117315e-06, + "loss": 0.3089, + "step": 21810 + }, + { + "epoch": 2.5810267329074996, + "grad_norm": 3.049873113632202, + "learning_rate": 4.504740244194398e-06, + "loss": 0.2715, + "step": 21820 + }, + { + "epoch": 2.5822096049207475, + "grad_norm": 2.9670944213867188, + "learning_rate": 4.504022025377065e-06, + "loss": 0.3164, + "step": 21830 + }, + { + "epoch": 2.5833924769339958, + "grad_norm": 3.935147762298584, + "learning_rate": 4.503303806559731e-06, + "loss": 0.3048, + "step": 21840 + }, + { + "epoch": 2.584575348947244, + "grad_norm": 2.235919713973999, + "learning_rate": 4.502585587742399e-06, + "loss": 0.2899, + "step": 21850 + }, + { + "epoch": 2.585758220960492, + "grad_norm": 2.5635204315185547, + "learning_rate": 4.501867368925066e-06, + "loss": 0.3051, + "step": 21860 + }, + { + "epoch": 2.58694109297374, + "grad_norm": 2.865705966949463, + "learning_rate": 4.501149150107733e-06, + "loss": 0.3228, + "step": 21870 + }, + { + "epoch": 2.5881239649869885, + "grad_norm": 2.436107873916626, + "learning_rate": 4.5004309312904e-06, + "loss": 0.293, + "step": 21880 + }, + { + "epoch": 2.589306837000237, + "grad_norm": 3.265331506729126, + "learning_rate": 4.499712712473067e-06, + "loss": 0.3014, + "step": 21890 + }, + { + "epoch": 2.5904897090134846, + "grad_norm": 2.8012731075286865, + "learning_rate": 4.498994493655734e-06, + "loss": 0.3133, + "step": 21900 + }, + { + "epoch": 2.591672581026733, + "grad_norm": 2.5874252319335938, + "learning_rate": 4.498276274838401e-06, + "loss": 0.3176, + "step": 21910 + }, + { + "epoch": 2.592855453039981, + "grad_norm": 2.8500173091888428, + "learning_rate": 4.497558056021068e-06, + "loss": 0.2909, + "step": 21920 + }, + { + "epoch": 2.594038325053229, + "grad_norm": 2.56345272064209, + "learning_rate": 4.4968398372037346e-06, + "loss": 0.3019, + "step": 21930 + }, + { + "epoch": 2.5952211970664774, + "grad_norm": 3.251715660095215, + "learning_rate": 4.4961216183864015e-06, + "loss": 0.2899, + "step": 21940 + }, + { + "epoch": 2.5964040690797257, + "grad_norm": 2.881084442138672, + "learning_rate": 4.495403399569069e-06, + "loss": 0.29, + "step": 21950 + }, + { + "epoch": 2.597586941092974, + "grad_norm": 2.4817402362823486, + "learning_rate": 4.494685180751735e-06, + "loss": 0.3291, + "step": 21960 + }, + { + "epoch": 2.598769813106222, + "grad_norm": 3.5438082218170166, + "learning_rate": 4.493966961934403e-06, + "loss": 0.2523, + "step": 21970 + }, + { + "epoch": 2.59995268511947, + "grad_norm": 2.795470714569092, + "learning_rate": 4.493248743117069e-06, + "loss": 0.3352, + "step": 21980 + }, + { + "epoch": 2.6011355571327184, + "grad_norm": 2.066418170928955, + "learning_rate": 4.492530524299737e-06, + "loss": 0.3098, + "step": 21990 + }, + { + "epoch": 2.6023184291459662, + "grad_norm": 3.746610403060913, + "learning_rate": 4.491812305482404e-06, + "loss": 0.323, + "step": 22000 + }, + { + "epoch": 2.6035013011592145, + "grad_norm": 3.1087605953216553, + "learning_rate": 4.491094086665071e-06, + "loss": 0.3441, + "step": 22010 + }, + { + "epoch": 2.604684173172463, + "grad_norm": 2.8416616916656494, + "learning_rate": 4.490375867847738e-06, + "loss": 0.2848, + "step": 22020 + }, + { + "epoch": 2.605867045185711, + "grad_norm": 2.4401543140411377, + "learning_rate": 4.489657649030405e-06, + "loss": 0.2817, + "step": 22030 + }, + { + "epoch": 2.607049917198959, + "grad_norm": 2.596233367919922, + "learning_rate": 4.4889394302130716e-06, + "loss": 0.2915, + "step": 22040 + }, + { + "epoch": 2.6082327892122072, + "grad_norm": 2.7294564247131348, + "learning_rate": 4.4882212113957385e-06, + "loss": 0.3363, + "step": 22050 + }, + { + "epoch": 2.6094156612254555, + "grad_norm": 4.349059104919434, + "learning_rate": 4.487502992578406e-06, + "loss": 0.2844, + "step": 22060 + }, + { + "epoch": 2.6105985332387034, + "grad_norm": 3.3832664489746094, + "learning_rate": 4.486784773761072e-06, + "loss": 0.3283, + "step": 22070 + }, + { + "epoch": 2.6117814052519517, + "grad_norm": 3.112755537033081, + "learning_rate": 4.48606655494374e-06, + "loss": 0.3205, + "step": 22080 + }, + { + "epoch": 2.6129642772652, + "grad_norm": 2.8589746952056885, + "learning_rate": 4.485348336126406e-06, + "loss": 0.2879, + "step": 22090 + }, + { + "epoch": 2.6141471492784483, + "grad_norm": 2.2620599269866943, + "learning_rate": 4.484630117309074e-06, + "loss": 0.2856, + "step": 22100 + }, + { + "epoch": 2.615330021291696, + "grad_norm": 3.2509114742279053, + "learning_rate": 4.48391189849174e-06, + "loss": 0.3067, + "step": 22110 + }, + { + "epoch": 2.6165128933049444, + "grad_norm": 2.9454476833343506, + "learning_rate": 4.483193679674408e-06, + "loss": 0.3149, + "step": 22120 + }, + { + "epoch": 2.6176957653181927, + "grad_norm": 3.732296943664551, + "learning_rate": 4.482475460857075e-06, + "loss": 0.3113, + "step": 22130 + }, + { + "epoch": 2.6188786373314406, + "grad_norm": 2.731187343597412, + "learning_rate": 4.481757242039742e-06, + "loss": 0.3033, + "step": 22140 + }, + { + "epoch": 2.620061509344689, + "grad_norm": 2.3635094165802, + "learning_rate": 4.4810390232224086e-06, + "loss": 0.2897, + "step": 22150 + }, + { + "epoch": 2.621244381357937, + "grad_norm": 2.8644402027130127, + "learning_rate": 4.4803208044050755e-06, + "loss": 0.285, + "step": 22160 + }, + { + "epoch": 2.6224272533711854, + "grad_norm": 3.13881254196167, + "learning_rate": 4.479602585587742e-06, + "loss": 0.3308, + "step": 22170 + }, + { + "epoch": 2.6236101253844333, + "grad_norm": 2.368556022644043, + "learning_rate": 4.478884366770409e-06, + "loss": 0.2808, + "step": 22180 + }, + { + "epoch": 2.6247929973976816, + "grad_norm": 3.4502034187316895, + "learning_rate": 4.478166147953076e-06, + "loss": 0.3235, + "step": 22190 + }, + { + "epoch": 2.62597586941093, + "grad_norm": 2.3046300411224365, + "learning_rate": 4.477447929135743e-06, + "loss": 0.3126, + "step": 22200 + }, + { + "epoch": 2.6271587414241777, + "grad_norm": 2.2939493656158447, + "learning_rate": 4.47672971031841e-06, + "loss": 0.3189, + "step": 22210 + }, + { + "epoch": 2.628341613437426, + "grad_norm": 2.177417039871216, + "learning_rate": 4.476011491501078e-06, + "loss": 0.2869, + "step": 22220 + }, + { + "epoch": 2.6295244854506743, + "grad_norm": 3.3147706985473633, + "learning_rate": 4.475293272683744e-06, + "loss": 0.2878, + "step": 22230 + }, + { + "epoch": 2.6307073574639226, + "grad_norm": 3.703477382659912, + "learning_rate": 4.474575053866412e-06, + "loss": 0.3166, + "step": 22240 + }, + { + "epoch": 2.6318902294771704, + "grad_norm": 2.5452001094818115, + "learning_rate": 4.473856835049079e-06, + "loss": 0.2937, + "step": 22250 + }, + { + "epoch": 2.6330731014904187, + "grad_norm": 5.185401439666748, + "learning_rate": 4.4731386162317456e-06, + "loss": 0.354, + "step": 22260 + }, + { + "epoch": 2.634255973503667, + "grad_norm": 4.242445945739746, + "learning_rate": 4.4724203974144125e-06, + "loss": 0.3036, + "step": 22270 + }, + { + "epoch": 2.635438845516915, + "grad_norm": 2.7605392932891846, + "learning_rate": 4.471702178597079e-06, + "loss": 0.2871, + "step": 22280 + }, + { + "epoch": 2.636621717530163, + "grad_norm": 2.5750670433044434, + "learning_rate": 4.470983959779746e-06, + "loss": 0.3005, + "step": 22290 + }, + { + "epoch": 2.6378045895434115, + "grad_norm": 2.137120008468628, + "learning_rate": 4.470265740962413e-06, + "loss": 0.2761, + "step": 22300 + }, + { + "epoch": 2.6389874615566598, + "grad_norm": 2.8040382862091064, + "learning_rate": 4.46954752214508e-06, + "loss": 0.3107, + "step": 22310 + }, + { + "epoch": 2.6401703335699076, + "grad_norm": 2.520559787750244, + "learning_rate": 4.468829303327747e-06, + "loss": 0.3228, + "step": 22320 + }, + { + "epoch": 2.641353205583156, + "grad_norm": 2.437270402908325, + "learning_rate": 4.468111084510415e-06, + "loss": 0.3162, + "step": 22330 + }, + { + "epoch": 2.642536077596404, + "grad_norm": 2.3186631202697754, + "learning_rate": 4.467392865693081e-06, + "loss": 0.2796, + "step": 22340 + }, + { + "epoch": 2.643718949609652, + "grad_norm": 3.4087679386138916, + "learning_rate": 4.466674646875749e-06, + "loss": 0.3037, + "step": 22350 + }, + { + "epoch": 2.6449018216229003, + "grad_norm": 2.9389445781707764, + "learning_rate": 4.465956428058415e-06, + "loss": 0.3005, + "step": 22360 + }, + { + "epoch": 2.6460846936361486, + "grad_norm": 2.935872793197632, + "learning_rate": 4.4652382092410826e-06, + "loss": 0.3152, + "step": 22370 + }, + { + "epoch": 2.647267565649397, + "grad_norm": 3.1000871658325195, + "learning_rate": 4.464519990423749e-06, + "loss": 0.3329, + "step": 22380 + }, + { + "epoch": 2.6484504376626448, + "grad_norm": 2.442791700363159, + "learning_rate": 4.463801771606416e-06, + "loss": 0.3109, + "step": 22390 + }, + { + "epoch": 2.649633309675893, + "grad_norm": 2.769963026046753, + "learning_rate": 4.463083552789083e-06, + "loss": 0.3012, + "step": 22400 + }, + { + "epoch": 2.6508161816891413, + "grad_norm": 2.4669077396392822, + "learning_rate": 4.46236533397175e-06, + "loss": 0.3252, + "step": 22410 + }, + { + "epoch": 2.651999053702389, + "grad_norm": 3.6878228187561035, + "learning_rate": 4.461647115154417e-06, + "loss": 0.3205, + "step": 22420 + }, + { + "epoch": 2.6531819257156375, + "grad_norm": 2.7062792778015137, + "learning_rate": 4.460928896337084e-06, + "loss": 0.3093, + "step": 22430 + }, + { + "epoch": 2.654364797728886, + "grad_norm": 3.205240249633789, + "learning_rate": 4.460210677519751e-06, + "loss": 0.3015, + "step": 22440 + }, + { + "epoch": 2.655547669742134, + "grad_norm": 3.016710042953491, + "learning_rate": 4.459492458702418e-06, + "loss": 0.3214, + "step": 22450 + }, + { + "epoch": 2.656730541755382, + "grad_norm": 3.4308531284332275, + "learning_rate": 4.458774239885085e-06, + "loss": 0.2982, + "step": 22460 + }, + { + "epoch": 2.65791341376863, + "grad_norm": 5.198909759521484, + "learning_rate": 4.458056021067752e-06, + "loss": 0.3001, + "step": 22470 + }, + { + "epoch": 2.6590962857818785, + "grad_norm": 3.1551146507263184, + "learning_rate": 4.457337802250419e-06, + "loss": 0.2969, + "step": 22480 + }, + { + "epoch": 2.6602791577951264, + "grad_norm": 3.4289937019348145, + "learning_rate": 4.4566195834330865e-06, + "loss": 0.3356, + "step": 22490 + }, + { + "epoch": 2.6614620298083747, + "grad_norm": 2.881150007247925, + "learning_rate": 4.455901364615753e-06, + "loss": 0.3053, + "step": 22500 + }, + { + "epoch": 2.662644901821623, + "grad_norm": 2.57087779045105, + "learning_rate": 4.45518314579842e-06, + "loss": 0.2908, + "step": 22510 + }, + { + "epoch": 2.6638277738348712, + "grad_norm": 3.0298633575439453, + "learning_rate": 4.454464926981087e-06, + "loss": 0.3385, + "step": 22520 + }, + { + "epoch": 2.665010645848119, + "grad_norm": 3.2643322944641113, + "learning_rate": 4.453746708163754e-06, + "loss": 0.3455, + "step": 22530 + }, + { + "epoch": 2.6661935178613674, + "grad_norm": 3.6884989738464355, + "learning_rate": 4.453028489346421e-06, + "loss": 0.2844, + "step": 22540 + }, + { + "epoch": 2.6673763898746157, + "grad_norm": 2.1394948959350586, + "learning_rate": 4.452310270529088e-06, + "loss": 0.2886, + "step": 22550 + }, + { + "epoch": 2.6685592618878635, + "grad_norm": 2.022263765335083, + "learning_rate": 4.451592051711755e-06, + "loss": 0.3082, + "step": 22560 + }, + { + "epoch": 2.669742133901112, + "grad_norm": 2.3887062072753906, + "learning_rate": 4.450873832894422e-06, + "loss": 0.3268, + "step": 22570 + }, + { + "epoch": 2.67092500591436, + "grad_norm": 3.5192759037017822, + "learning_rate": 4.450155614077089e-06, + "loss": 0.3252, + "step": 22580 + }, + { + "epoch": 2.6721078779276084, + "grad_norm": 3.427703380584717, + "learning_rate": 4.449437395259756e-06, + "loss": 0.3138, + "step": 22590 + }, + { + "epoch": 2.6732907499408562, + "grad_norm": 2.6806910037994385, + "learning_rate": 4.4487191764424235e-06, + "loss": 0.3115, + "step": 22600 + }, + { + "epoch": 2.6744736219541045, + "grad_norm": 3.288033962249756, + "learning_rate": 4.4480009576250896e-06, + "loss": 0.2962, + "step": 22610 + }, + { + "epoch": 2.675656493967353, + "grad_norm": 3.175037145614624, + "learning_rate": 4.447282738807757e-06, + "loss": 0.2638, + "step": 22620 + }, + { + "epoch": 2.6768393659806007, + "grad_norm": 3.9674432277679443, + "learning_rate": 4.446564519990423e-06, + "loss": 0.3351, + "step": 22630 + }, + { + "epoch": 2.678022237993849, + "grad_norm": 2.132941484451294, + "learning_rate": 4.445846301173091e-06, + "loss": 0.264, + "step": 22640 + }, + { + "epoch": 2.6792051100070973, + "grad_norm": 2.6347246170043945, + "learning_rate": 4.445128082355757e-06, + "loss": 0.2353, + "step": 22650 + }, + { + "epoch": 2.6803879820203456, + "grad_norm": 3.1588587760925293, + "learning_rate": 4.444409863538425e-06, + "loss": 0.2735, + "step": 22660 + }, + { + "epoch": 2.6815708540335934, + "grad_norm": 2.4475293159484863, + "learning_rate": 4.443691644721092e-06, + "loss": 0.2938, + "step": 22670 + }, + { + "epoch": 2.6827537260468417, + "grad_norm": 2.680957317352295, + "learning_rate": 4.442973425903759e-06, + "loss": 0.3135, + "step": 22680 + }, + { + "epoch": 2.68393659806009, + "grad_norm": 2.5454654693603516, + "learning_rate": 4.442255207086426e-06, + "loss": 0.2888, + "step": 22690 + }, + { + "epoch": 2.685119470073338, + "grad_norm": 2.876617193222046, + "learning_rate": 4.441536988269093e-06, + "loss": 0.3378, + "step": 22700 + }, + { + "epoch": 2.686302342086586, + "grad_norm": 2.1972174644470215, + "learning_rate": 4.44081876945176e-06, + "loss": 0.2835, + "step": 22710 + }, + { + "epoch": 2.6874852140998344, + "grad_norm": 2.8111729621887207, + "learning_rate": 4.4401005506344266e-06, + "loss": 0.3071, + "step": 22720 + }, + { + "epoch": 2.6886680861130827, + "grad_norm": 2.6271204948425293, + "learning_rate": 4.4393823318170935e-06, + "loss": 0.3073, + "step": 22730 + }, + { + "epoch": 2.6898509581263306, + "grad_norm": 2.106633186340332, + "learning_rate": 4.43866411299976e-06, + "loss": 0.2769, + "step": 22740 + }, + { + "epoch": 2.691033830139579, + "grad_norm": 1.852852463722229, + "learning_rate": 4.437945894182428e-06, + "loss": 0.36, + "step": 22750 + }, + { + "epoch": 2.692216702152827, + "grad_norm": 3.6838326454162598, + "learning_rate": 4.437227675365094e-06, + "loss": 0.2906, + "step": 22760 + }, + { + "epoch": 2.693399574166075, + "grad_norm": 3.3854281902313232, + "learning_rate": 4.436509456547762e-06, + "loss": 0.3291, + "step": 22770 + }, + { + "epoch": 2.6945824461793233, + "grad_norm": 1.9606446027755737, + "learning_rate": 4.435791237730429e-06, + "loss": 0.3019, + "step": 22780 + }, + { + "epoch": 2.6957653181925716, + "grad_norm": 2.7873902320861816, + "learning_rate": 4.435073018913096e-06, + "loss": 0.2771, + "step": 22790 + }, + { + "epoch": 2.69694819020582, + "grad_norm": 2.504549026489258, + "learning_rate": 4.434354800095763e-06, + "loss": 0.3005, + "step": 22800 + }, + { + "epoch": 2.6981310622190677, + "grad_norm": 3.122368812561035, + "learning_rate": 4.43363658127843e-06, + "loss": 0.3112, + "step": 22810 + }, + { + "epoch": 2.699313934232316, + "grad_norm": 2.7788901329040527, + "learning_rate": 4.432918362461097e-06, + "loss": 0.3093, + "step": 22820 + }, + { + "epoch": 2.7004968062455643, + "grad_norm": 2.8516581058502197, + "learning_rate": 4.4322001436437636e-06, + "loss": 0.2746, + "step": 22830 + }, + { + "epoch": 2.701679678258812, + "grad_norm": 2.4541964530944824, + "learning_rate": 4.4314819248264305e-06, + "loss": 0.3131, + "step": 22840 + }, + { + "epoch": 2.7028625502720605, + "grad_norm": 3.6100246906280518, + "learning_rate": 4.430763706009097e-06, + "loss": 0.3273, + "step": 22850 + }, + { + "epoch": 2.7040454222853088, + "grad_norm": 3.8532838821411133, + "learning_rate": 4.430045487191764e-06, + "loss": 0.3644, + "step": 22860 + }, + { + "epoch": 2.705228294298557, + "grad_norm": 1.9744536876678467, + "learning_rate": 4.429327268374432e-06, + "loss": 0.296, + "step": 22870 + }, + { + "epoch": 2.7064111663118053, + "grad_norm": 2.1615397930145264, + "learning_rate": 4.428609049557098e-06, + "loss": 0.289, + "step": 22880 + }, + { + "epoch": 2.707594038325053, + "grad_norm": 3.225660800933838, + "learning_rate": 4.427890830739766e-06, + "loss": 0.322, + "step": 22890 + }, + { + "epoch": 2.7087769103383015, + "grad_norm": 2.590725898742676, + "learning_rate": 4.427172611922432e-06, + "loss": 0.2718, + "step": 22900 + }, + { + "epoch": 2.7099597823515493, + "grad_norm": 2.792945384979248, + "learning_rate": 4.4264543931051e-06, + "loss": 0.325, + "step": 22910 + }, + { + "epoch": 2.7111426543647976, + "grad_norm": 2.9607298374176025, + "learning_rate": 4.425736174287766e-06, + "loss": 0.342, + "step": 22920 + }, + { + "epoch": 2.712325526378046, + "grad_norm": 3.1209685802459717, + "learning_rate": 4.425017955470434e-06, + "loss": 0.3042, + "step": 22930 + }, + { + "epoch": 2.713508398391294, + "grad_norm": 1.8263630867004395, + "learning_rate": 4.4242997366531006e-06, + "loss": 0.2865, + "step": 22940 + }, + { + "epoch": 2.7146912704045425, + "grad_norm": 3.2093710899353027, + "learning_rate": 4.4235815178357675e-06, + "loss": 0.3028, + "step": 22950 + }, + { + "epoch": 2.7158741424177903, + "grad_norm": 2.9916210174560547, + "learning_rate": 4.422863299018434e-06, + "loss": 0.2778, + "step": 22960 + }, + { + "epoch": 2.7170570144310386, + "grad_norm": 3.09580135345459, + "learning_rate": 4.422145080201101e-06, + "loss": 0.325, + "step": 22970 + }, + { + "epoch": 2.7182398864442865, + "grad_norm": 3.132143259048462, + "learning_rate": 4.421426861383768e-06, + "loss": 0.3074, + "step": 22980 + }, + { + "epoch": 2.719422758457535, + "grad_norm": 2.373967170715332, + "learning_rate": 4.420708642566435e-06, + "loss": 0.2816, + "step": 22990 + }, + { + "epoch": 2.720605630470783, + "grad_norm": 2.1527655124664307, + "learning_rate": 4.419990423749103e-06, + "loss": 0.2915, + "step": 23000 + }, + { + "epoch": 2.7217885024840314, + "grad_norm": 2.5220601558685303, + "learning_rate": 4.419272204931769e-06, + "loss": 0.2757, + "step": 23010 + }, + { + "epoch": 2.7229713744972797, + "grad_norm": 3.4006309509277344, + "learning_rate": 4.418553986114437e-06, + "loss": 0.3054, + "step": 23020 + }, + { + "epoch": 2.7241542465105275, + "grad_norm": 2.8180737495422363, + "learning_rate": 4.417835767297103e-06, + "loss": 0.2854, + "step": 23030 + }, + { + "epoch": 2.725337118523776, + "grad_norm": 3.7637462615966797, + "learning_rate": 4.417117548479771e-06, + "loss": 0.3404, + "step": 23040 + }, + { + "epoch": 2.7265199905370237, + "grad_norm": 2.473134756088257, + "learning_rate": 4.4163993296624376e-06, + "loss": 0.3056, + "step": 23050 + }, + { + "epoch": 2.727702862550272, + "grad_norm": 3.0845396518707275, + "learning_rate": 4.4156811108451045e-06, + "loss": 0.2885, + "step": 23060 + }, + { + "epoch": 2.7288857345635202, + "grad_norm": 2.06636118888855, + "learning_rate": 4.414962892027771e-06, + "loss": 0.3223, + "step": 23070 + }, + { + "epoch": 2.7300686065767685, + "grad_norm": 2.638631582260132, + "learning_rate": 4.414244673210438e-06, + "loss": 0.3525, + "step": 23080 + }, + { + "epoch": 2.731251478590017, + "grad_norm": 2.579003095626831, + "learning_rate": 4.413526454393105e-06, + "loss": 0.316, + "step": 23090 + }, + { + "epoch": 2.7324343506032647, + "grad_norm": 3.0096957683563232, + "learning_rate": 4.412808235575772e-06, + "loss": 0.3101, + "step": 23100 + }, + { + "epoch": 2.733617222616513, + "grad_norm": 2.992541790008545, + "learning_rate": 4.412090016758439e-06, + "loss": 0.2963, + "step": 23110 + }, + { + "epoch": 2.734800094629761, + "grad_norm": 4.029899597167969, + "learning_rate": 4.411371797941106e-06, + "loss": 0.3535, + "step": 23120 + }, + { + "epoch": 2.735982966643009, + "grad_norm": 2.9068877696990967, + "learning_rate": 4.410653579123773e-06, + "loss": 0.2761, + "step": 23130 + }, + { + "epoch": 2.7371658386562574, + "grad_norm": 2.3513588905334473, + "learning_rate": 4.409935360306441e-06, + "loss": 0.3317, + "step": 23140 + }, + { + "epoch": 2.7383487106695057, + "grad_norm": 2.666271448135376, + "learning_rate": 4.409217141489107e-06, + "loss": 0.2716, + "step": 23150 + }, + { + "epoch": 2.739531582682754, + "grad_norm": 2.6380770206451416, + "learning_rate": 4.4084989226717745e-06, + "loss": 0.3471, + "step": 23160 + }, + { + "epoch": 2.740714454696002, + "grad_norm": 2.5514445304870605, + "learning_rate": 4.407780703854441e-06, + "loss": 0.2789, + "step": 23170 + }, + { + "epoch": 2.74189732670925, + "grad_norm": 2.63517165184021, + "learning_rate": 4.407062485037108e-06, + "loss": 0.3273, + "step": 23180 + }, + { + "epoch": 2.743080198722498, + "grad_norm": 2.637009620666504, + "learning_rate": 4.4063442662197745e-06, + "loss": 0.2867, + "step": 23190 + }, + { + "epoch": 2.7442630707357463, + "grad_norm": 2.422926187515259, + "learning_rate": 4.405626047402442e-06, + "loss": 0.3315, + "step": 23200 + }, + { + "epoch": 2.7454459427489946, + "grad_norm": 3.1612865924835205, + "learning_rate": 4.404907828585109e-06, + "loss": 0.3239, + "step": 23210 + }, + { + "epoch": 2.746628814762243, + "grad_norm": 4.27061653137207, + "learning_rate": 4.404189609767776e-06, + "loss": 0.3408, + "step": 23220 + }, + { + "epoch": 2.747811686775491, + "grad_norm": 2.0837695598602295, + "learning_rate": 4.403471390950443e-06, + "loss": 0.2802, + "step": 23230 + }, + { + "epoch": 2.748994558788739, + "grad_norm": 2.7004873752593994, + "learning_rate": 4.40275317213311e-06, + "loss": 0.3214, + "step": 23240 + }, + { + "epoch": 2.7501774308019873, + "grad_norm": 2.703378438949585, + "learning_rate": 4.402034953315777e-06, + "loss": 0.2938, + "step": 23250 + }, + { + "epoch": 2.7506505796072864, + "eval_accuracy": 0.8600159696576505, + "eval_loss": 0.3236427307128906, + "eval_runtime": 78.0091, + "eval_safe_aucpr": 0.915810949681183, + "eval_safe_f1": 0.8426426314116349, + "eval_safe_fpr": 0.12803802576903492, + "eval_safe_precision": 0.8402700082046692, + "eval_safe_recall": 0.8450286914450736, + "eval_samples_per_second": 770.602, + "eval_steps_per_second": 12.05, + "eval_unsafe_aucpr": 0.9529369806974803, + "eval_unsafe_f1": 0.8739344728917919, + "eval_unsafe_fpr": 0.15497130855492572, + "eval_unsafe_precision": 0.8759159159159159, + "eval_unsafe_recall": 0.8719619742309647, + "step": 23254 + }, + { + "epoch": 2.751360302815235, + "grad_norm": 3.1155622005462646, + "learning_rate": 4.401316734498444e-06, + "loss": 0.2745, + "step": 23260 + }, + { + "epoch": 2.7525431748284834, + "grad_norm": 2.831904888153076, + "learning_rate": 4.4005985156811115e-06, + "loss": 0.318, + "step": 23270 + }, + { + "epoch": 2.7537260468417317, + "grad_norm": 3.1426374912261963, + "learning_rate": 4.399880296863778e-06, + "loss": 0.3276, + "step": 23280 + }, + { + "epoch": 2.75490891885498, + "grad_norm": 2.297433614730835, + "learning_rate": 4.399162078046445e-06, + "loss": 0.3027, + "step": 23290 + }, + { + "epoch": 2.7560917908682283, + "grad_norm": 2.904245615005493, + "learning_rate": 4.3984438592291115e-06, + "loss": 0.3362, + "step": 23300 + }, + { + "epoch": 2.757274662881476, + "grad_norm": 2.9260730743408203, + "learning_rate": 4.397725640411779e-06, + "loss": 0.2942, + "step": 23310 + }, + { + "epoch": 2.7584575348947244, + "grad_norm": 2.8044800758361816, + "learning_rate": 4.397007421594446e-06, + "loss": 0.319, + "step": 23320 + }, + { + "epoch": 2.7596404069079723, + "grad_norm": 3.2318618297576904, + "learning_rate": 4.396289202777113e-06, + "loss": 0.3323, + "step": 23330 + }, + { + "epoch": 2.7608232789212206, + "grad_norm": 2.8827571868896484, + "learning_rate": 4.39557098395978e-06, + "loss": 0.3319, + "step": 23340 + }, + { + "epoch": 2.762006150934469, + "grad_norm": 2.591367721557617, + "learning_rate": 4.394852765142447e-06, + "loss": 0.3233, + "step": 23350 + }, + { + "epoch": 2.763189022947717, + "grad_norm": 2.1080996990203857, + "learning_rate": 4.394134546325114e-06, + "loss": 0.3121, + "step": 23360 + }, + { + "epoch": 2.7643718949609655, + "grad_norm": 2.24174427986145, + "learning_rate": 4.393416327507781e-06, + "loss": 0.2933, + "step": 23370 + }, + { + "epoch": 2.7655547669742133, + "grad_norm": 2.059626579284668, + "learning_rate": 4.392698108690448e-06, + "loss": 0.288, + "step": 23380 + }, + { + "epoch": 2.7667376389874616, + "grad_norm": 3.090696334838867, + "learning_rate": 4.391979889873115e-06, + "loss": 0.3132, + "step": 23390 + }, + { + "epoch": 2.7679205110007095, + "grad_norm": 2.660612106323242, + "learning_rate": 4.3912616710557815e-06, + "loss": 0.3008, + "step": 23400 + }, + { + "epoch": 2.7691033830139578, + "grad_norm": 2.2376856803894043, + "learning_rate": 4.3905434522384485e-06, + "loss": 0.3195, + "step": 23410 + }, + { + "epoch": 2.770286255027206, + "grad_norm": 2.5992982387542725, + "learning_rate": 4.389825233421115e-06, + "loss": 0.3193, + "step": 23420 + }, + { + "epoch": 2.7714691270404543, + "grad_norm": 2.2484004497528076, + "learning_rate": 4.389107014603783e-06, + "loss": 0.2877, + "step": 23430 + }, + { + "epoch": 2.7726519990537026, + "grad_norm": 2.537644386291504, + "learning_rate": 4.388388795786449e-06, + "loss": 0.3035, + "step": 23440 + }, + { + "epoch": 2.7738348710669505, + "grad_norm": 2.8209824562072754, + "learning_rate": 4.387670576969117e-06, + "loss": 0.3389, + "step": 23450 + }, + { + "epoch": 2.7750177430801988, + "grad_norm": 2.1926157474517822, + "learning_rate": 4.386952358151783e-06, + "loss": 0.2674, + "step": 23460 + }, + { + "epoch": 2.7762006150934466, + "grad_norm": 3.265162706375122, + "learning_rate": 4.386234139334451e-06, + "loss": 0.3192, + "step": 23470 + }, + { + "epoch": 2.777383487106695, + "grad_norm": 5.2143707275390625, + "learning_rate": 4.385515920517118e-06, + "loss": 0.2838, + "step": 23480 + }, + { + "epoch": 2.778566359119943, + "grad_norm": 2.268308162689209, + "learning_rate": 4.384797701699785e-06, + "loss": 0.3102, + "step": 23490 + }, + { + "epoch": 2.7797492311331915, + "grad_norm": 1.8075358867645264, + "learning_rate": 4.384079482882452e-06, + "loss": 0.3567, + "step": 23500 + }, + { + "epoch": 2.78093210314644, + "grad_norm": 1.7917972803115845, + "learning_rate": 4.3833612640651185e-06, + "loss": 0.3117, + "step": 23510 + }, + { + "epoch": 2.7821149751596876, + "grad_norm": 2.8360867500305176, + "learning_rate": 4.382643045247786e-06, + "loss": 0.2782, + "step": 23520 + }, + { + "epoch": 2.783297847172936, + "grad_norm": 4.134777545928955, + "learning_rate": 4.381924826430452e-06, + "loss": 0.3221, + "step": 23530 + }, + { + "epoch": 2.784480719186184, + "grad_norm": 3.0642571449279785, + "learning_rate": 4.38120660761312e-06, + "loss": 0.3283, + "step": 23540 + }, + { + "epoch": 2.785663591199432, + "grad_norm": 3.0967137813568115, + "learning_rate": 4.380488388795786e-06, + "loss": 0.3269, + "step": 23550 + }, + { + "epoch": 2.7868464632126804, + "grad_norm": 3.2665817737579346, + "learning_rate": 4.379770169978454e-06, + "loss": 0.3101, + "step": 23560 + }, + { + "epoch": 2.7880293352259287, + "grad_norm": 3.0433547496795654, + "learning_rate": 4.37905195116112e-06, + "loss": 0.3173, + "step": 23570 + }, + { + "epoch": 2.789212207239177, + "grad_norm": 2.697173595428467, + "learning_rate": 4.378333732343788e-06, + "loss": 0.3025, + "step": 23580 + }, + { + "epoch": 2.790395079252425, + "grad_norm": 2.210649013519287, + "learning_rate": 4.377615513526455e-06, + "loss": 0.3011, + "step": 23590 + }, + { + "epoch": 2.791577951265673, + "grad_norm": 3.6849350929260254, + "learning_rate": 4.376897294709122e-06, + "loss": 0.2838, + "step": 23600 + }, + { + "epoch": 2.792760823278921, + "grad_norm": 3.105637788772583, + "learning_rate": 4.376179075891789e-06, + "loss": 0.3174, + "step": 23610 + }, + { + "epoch": 2.7939436952921692, + "grad_norm": 2.7724554538726807, + "learning_rate": 4.3754608570744555e-06, + "loss": 0.355, + "step": 23620 + }, + { + "epoch": 2.7951265673054175, + "grad_norm": 2.5946154594421387, + "learning_rate": 4.3747426382571225e-06, + "loss": 0.2961, + "step": 23630 + }, + { + "epoch": 2.796309439318666, + "grad_norm": 2.994760274887085, + "learning_rate": 4.374024419439789e-06, + "loss": 0.281, + "step": 23640 + }, + { + "epoch": 2.797492311331914, + "grad_norm": 2.9595296382904053, + "learning_rate": 4.373306200622456e-06, + "loss": 0.3369, + "step": 23650 + }, + { + "epoch": 2.798675183345162, + "grad_norm": 3.436298131942749, + "learning_rate": 4.372587981805123e-06, + "loss": 0.2829, + "step": 23660 + }, + { + "epoch": 2.7998580553584103, + "grad_norm": 3.4422378540039062, + "learning_rate": 4.37186976298779e-06, + "loss": 0.3076, + "step": 23670 + }, + { + "epoch": 2.8010409273716586, + "grad_norm": 2.474370002746582, + "learning_rate": 4.371151544170457e-06, + "loss": 0.3513, + "step": 23680 + }, + { + "epoch": 2.8022237993849064, + "grad_norm": 2.483954906463623, + "learning_rate": 4.370433325353124e-06, + "loss": 0.2953, + "step": 23690 + }, + { + "epoch": 2.8034066713981547, + "grad_norm": 2.6980764865875244, + "learning_rate": 4.369715106535792e-06, + "loss": 0.3324, + "step": 23700 + }, + { + "epoch": 2.804589543411403, + "grad_norm": 3.0941340923309326, + "learning_rate": 4.368996887718458e-06, + "loss": 0.3381, + "step": 23710 + }, + { + "epoch": 2.8057724154246513, + "grad_norm": 3.7191810607910156, + "learning_rate": 4.368278668901126e-06, + "loss": 0.3103, + "step": 23720 + }, + { + "epoch": 2.806955287437899, + "grad_norm": 2.45870041847229, + "learning_rate": 4.367560450083792e-06, + "loss": 0.2724, + "step": 23730 + }, + { + "epoch": 2.8081381594511474, + "grad_norm": 2.5216989517211914, + "learning_rate": 4.3668422312664595e-06, + "loss": 0.2965, + "step": 23740 + }, + { + "epoch": 2.8093210314643957, + "grad_norm": 3.067351818084717, + "learning_rate": 4.3661240124491255e-06, + "loss": 0.3239, + "step": 23750 + }, + { + "epoch": 2.8105039034776436, + "grad_norm": 2.456937551498413, + "learning_rate": 4.365405793631793e-06, + "loss": 0.2802, + "step": 23760 + }, + { + "epoch": 2.811686775490892, + "grad_norm": 2.1812660694122314, + "learning_rate": 4.36468757481446e-06, + "loss": 0.2657, + "step": 23770 + }, + { + "epoch": 2.81286964750414, + "grad_norm": 2.795524835586548, + "learning_rate": 4.363969355997127e-06, + "loss": 0.2958, + "step": 23780 + }, + { + "epoch": 2.8140525195173884, + "grad_norm": 3.1462368965148926, + "learning_rate": 4.363251137179795e-06, + "loss": 0.3151, + "step": 23790 + }, + { + "epoch": 2.8152353915306363, + "grad_norm": 3.474961996078491, + "learning_rate": 4.362532918362461e-06, + "loss": 0.3123, + "step": 23800 + }, + { + "epoch": 2.8164182635438846, + "grad_norm": 2.382246732711792, + "learning_rate": 4.361814699545129e-06, + "loss": 0.303, + "step": 23810 + }, + { + "epoch": 2.817601135557133, + "grad_norm": 3.0838286876678467, + "learning_rate": 4.361096480727795e-06, + "loss": 0.3176, + "step": 23820 + }, + { + "epoch": 2.8187840075703807, + "grad_norm": 3.0229458808898926, + "learning_rate": 4.360378261910463e-06, + "loss": 0.2883, + "step": 23830 + }, + { + "epoch": 2.819966879583629, + "grad_norm": 2.019643783569336, + "learning_rate": 4.359660043093129e-06, + "loss": 0.2997, + "step": 23840 + }, + { + "epoch": 2.8211497515968773, + "grad_norm": 2.7812955379486084, + "learning_rate": 4.3589418242757965e-06, + "loss": 0.2914, + "step": 23850 + }, + { + "epoch": 2.8223326236101256, + "grad_norm": 2.832305908203125, + "learning_rate": 4.358223605458463e-06, + "loss": 0.2701, + "step": 23860 + }, + { + "epoch": 2.8235154956233735, + "grad_norm": 2.620081663131714, + "learning_rate": 4.35750538664113e-06, + "loss": 0.2716, + "step": 23870 + }, + { + "epoch": 2.8246983676366217, + "grad_norm": 3.746365547180176, + "learning_rate": 4.356787167823797e-06, + "loss": 0.3167, + "step": 23880 + }, + { + "epoch": 2.82588123964987, + "grad_norm": 3.218179941177368, + "learning_rate": 4.356068949006464e-06, + "loss": 0.2752, + "step": 23890 + }, + { + "epoch": 2.827064111663118, + "grad_norm": 2.8903679847717285, + "learning_rate": 4.355350730189131e-06, + "loss": 0.3341, + "step": 23900 + }, + { + "epoch": 2.828246983676366, + "grad_norm": 3.1378731727600098, + "learning_rate": 4.354632511371798e-06, + "loss": 0.3288, + "step": 23910 + }, + { + "epoch": 2.8294298556896145, + "grad_norm": 3.520890951156616, + "learning_rate": 4.353914292554465e-06, + "loss": 0.3031, + "step": 23920 + }, + { + "epoch": 2.8306127277028628, + "grad_norm": 3.1894991397857666, + "learning_rate": 4.353196073737132e-06, + "loss": 0.2866, + "step": 23930 + }, + { + "epoch": 2.8317955997161106, + "grad_norm": 3.105095863342285, + "learning_rate": 4.352477854919799e-06, + "loss": 0.3252, + "step": 23940 + }, + { + "epoch": 2.832978471729359, + "grad_norm": 3.4214463233947754, + "learning_rate": 4.351759636102466e-06, + "loss": 0.3235, + "step": 23950 + }, + { + "epoch": 2.834161343742607, + "grad_norm": 2.883387327194214, + "learning_rate": 4.351041417285133e-06, + "loss": 0.2861, + "step": 23960 + }, + { + "epoch": 2.835344215755855, + "grad_norm": 3.905980348587036, + "learning_rate": 4.3503231984678e-06, + "loss": 0.3208, + "step": 23970 + }, + { + "epoch": 2.8365270877691033, + "grad_norm": 2.9791083335876465, + "learning_rate": 4.3496049796504665e-06, + "loss": 0.3083, + "step": 23980 + }, + { + "epoch": 2.8377099597823516, + "grad_norm": 2.9695980548858643, + "learning_rate": 4.348886760833134e-06, + "loss": 0.2938, + "step": 23990 + }, + { + "epoch": 2.8388928317956, + "grad_norm": 2.961609125137329, + "learning_rate": 4.3481685420158e-06, + "loss": 0.2956, + "step": 24000 + }, + { + "epoch": 2.8400757038088478, + "grad_norm": 4.365261554718018, + "learning_rate": 4.347450323198468e-06, + "loss": 0.3352, + "step": 24010 + }, + { + "epoch": 2.841258575822096, + "grad_norm": 2.174947738647461, + "learning_rate": 4.346732104381135e-06, + "loss": 0.3087, + "step": 24020 + }, + { + "epoch": 2.8424414478353444, + "grad_norm": 2.263009548187256, + "learning_rate": 4.346013885563802e-06, + "loss": 0.3029, + "step": 24030 + }, + { + "epoch": 2.843624319848592, + "grad_norm": 2.728564739227295, + "learning_rate": 4.345295666746469e-06, + "loss": 0.2667, + "step": 24040 + }, + { + "epoch": 2.8448071918618405, + "grad_norm": 2.0310652256011963, + "learning_rate": 4.344577447929136e-06, + "loss": 0.2938, + "step": 24050 + }, + { + "epoch": 2.845990063875089, + "grad_norm": 3.1241891384124756, + "learning_rate": 4.3438592291118035e-06, + "loss": 0.302, + "step": 24060 + }, + { + "epoch": 2.847172935888337, + "grad_norm": 1.7154382467269897, + "learning_rate": 4.34314101029447e-06, + "loss": 0.3165, + "step": 24070 + }, + { + "epoch": 2.848355807901585, + "grad_norm": 2.4480531215667725, + "learning_rate": 4.342422791477137e-06, + "loss": 0.3132, + "step": 24080 + }, + { + "epoch": 2.8495386799148332, + "grad_norm": 2.5914530754089355, + "learning_rate": 4.3417045726598034e-06, + "loss": 0.3099, + "step": 24090 + }, + { + "epoch": 2.8507215519280815, + "grad_norm": 2.5049376487731934, + "learning_rate": 4.340986353842471e-06, + "loss": 0.2598, + "step": 24100 + }, + { + "epoch": 2.8519044239413294, + "grad_norm": 2.4667346477508545, + "learning_rate": 4.340268135025137e-06, + "loss": 0.3143, + "step": 24110 + }, + { + "epoch": 2.8530872959545777, + "grad_norm": 3.8974785804748535, + "learning_rate": 4.339549916207805e-06, + "loss": 0.3016, + "step": 24120 + }, + { + "epoch": 2.854270167967826, + "grad_norm": 3.359978199005127, + "learning_rate": 4.338831697390472e-06, + "loss": 0.3351, + "step": 24130 + }, + { + "epoch": 2.8554530399810742, + "grad_norm": 2.733987808227539, + "learning_rate": 4.338113478573139e-06, + "loss": 0.3091, + "step": 24140 + }, + { + "epoch": 2.856635911994322, + "grad_norm": 2.608760356903076, + "learning_rate": 4.337395259755806e-06, + "loss": 0.3083, + "step": 24150 + }, + { + "epoch": 2.8578187840075704, + "grad_norm": 2.0584022998809814, + "learning_rate": 4.336677040938473e-06, + "loss": 0.2611, + "step": 24160 + }, + { + "epoch": 2.8590016560208187, + "grad_norm": 3.2685577869415283, + "learning_rate": 4.33595882212114e-06, + "loss": 0.3026, + "step": 24170 + }, + { + "epoch": 2.8601845280340665, + "grad_norm": 2.4171926975250244, + "learning_rate": 4.335240603303807e-06, + "loss": 0.2818, + "step": 24180 + }, + { + "epoch": 2.861367400047315, + "grad_norm": 4.359286308288574, + "learning_rate": 4.3345223844864735e-06, + "loss": 0.3372, + "step": 24190 + }, + { + "epoch": 2.862550272060563, + "grad_norm": 2.5716099739074707, + "learning_rate": 4.3338041656691404e-06, + "loss": 0.2796, + "step": 24200 + }, + { + "epoch": 2.8637331440738114, + "grad_norm": 3.2029051780700684, + "learning_rate": 4.333085946851807e-06, + "loss": 0.3038, + "step": 24210 + }, + { + "epoch": 2.8649160160870593, + "grad_norm": 3.3147220611572266, + "learning_rate": 4.332367728034474e-06, + "loss": 0.3271, + "step": 24220 + }, + { + "epoch": 2.8660988881003076, + "grad_norm": 2.245283365249634, + "learning_rate": 4.331649509217141e-06, + "loss": 0.2911, + "step": 24230 + }, + { + "epoch": 2.867281760113556, + "grad_norm": 2.960596799850464, + "learning_rate": 4.330931290399809e-06, + "loss": 0.2754, + "step": 24240 + }, + { + "epoch": 2.8684646321268037, + "grad_norm": 2.703169822692871, + "learning_rate": 4.330213071582475e-06, + "loss": 0.3137, + "step": 24250 + }, + { + "epoch": 2.869647504140052, + "grad_norm": 2.8850743770599365, + "learning_rate": 4.329494852765143e-06, + "loss": 0.3061, + "step": 24260 + }, + { + "epoch": 2.8708303761533003, + "grad_norm": 2.7234113216400146, + "learning_rate": 4.32877663394781e-06, + "loss": 0.3161, + "step": 24270 + }, + { + "epoch": 2.8720132481665486, + "grad_norm": 2.654510021209717, + "learning_rate": 4.328058415130477e-06, + "loss": 0.3112, + "step": 24280 + }, + { + "epoch": 2.8731961201797964, + "grad_norm": 2.526492118835449, + "learning_rate": 4.327340196313144e-06, + "loss": 0.3162, + "step": 24290 + }, + { + "epoch": 2.8743789921930447, + "grad_norm": 2.7180864810943604, + "learning_rate": 4.3266219774958105e-06, + "loss": 0.2565, + "step": 24300 + }, + { + "epoch": 2.875561864206293, + "grad_norm": 2.5691475868225098, + "learning_rate": 4.3259037586784774e-06, + "loss": 0.27, + "step": 24310 + }, + { + "epoch": 2.876744736219541, + "grad_norm": 3.05277681350708, + "learning_rate": 4.325185539861144e-06, + "loss": 0.3074, + "step": 24320 + }, + { + "epoch": 2.877927608232789, + "grad_norm": 3.2249338626861572, + "learning_rate": 4.324467321043811e-06, + "loss": 0.2773, + "step": 24330 + }, + { + "epoch": 2.8791104802460374, + "grad_norm": 1.9042539596557617, + "learning_rate": 4.323749102226478e-06, + "loss": 0.3033, + "step": 24340 + }, + { + "epoch": 2.8802933522592857, + "grad_norm": 2.6295275688171387, + "learning_rate": 4.323030883409146e-06, + "loss": 0.3131, + "step": 24350 + }, + { + "epoch": 2.8814762242725336, + "grad_norm": 4.007050514221191, + "learning_rate": 4.322312664591812e-06, + "loss": 0.303, + "step": 24360 + }, + { + "epoch": 2.882659096285782, + "grad_norm": 3.637464761734009, + "learning_rate": 4.32159444577448e-06, + "loss": 0.2933, + "step": 24370 + }, + { + "epoch": 2.88384196829903, + "grad_norm": 3.6937456130981445, + "learning_rate": 4.320876226957146e-06, + "loss": 0.3288, + "step": 24380 + }, + { + "epoch": 2.885024840312278, + "grad_norm": 2.2494723796844482, + "learning_rate": 4.320158008139814e-06, + "loss": 0.3043, + "step": 24390 + }, + { + "epoch": 2.8862077123255263, + "grad_norm": 2.7955732345581055, + "learning_rate": 4.319439789322481e-06, + "loss": 0.2888, + "step": 24400 + }, + { + "epoch": 2.8873905843387746, + "grad_norm": 3.020750045776367, + "learning_rate": 4.3187215705051475e-06, + "loss": 0.3021, + "step": 24410 + }, + { + "epoch": 2.888573456352023, + "grad_norm": 3.558333158493042, + "learning_rate": 4.3180033516878144e-06, + "loss": 0.3243, + "step": 24420 + }, + { + "epoch": 2.8897563283652707, + "grad_norm": 2.969136953353882, + "learning_rate": 4.317285132870481e-06, + "loss": 0.2831, + "step": 24430 + }, + { + "epoch": 2.890939200378519, + "grad_norm": 2.7450790405273438, + "learning_rate": 4.316566914053148e-06, + "loss": 0.2479, + "step": 24440 + }, + { + "epoch": 2.8921220723917673, + "grad_norm": 2.771754026412964, + "learning_rate": 4.315848695235815e-06, + "loss": 0.3102, + "step": 24450 + }, + { + "epoch": 2.893304944405015, + "grad_norm": 3.246392250061035, + "learning_rate": 4.315130476418482e-06, + "loss": 0.3173, + "step": 24460 + }, + { + "epoch": 2.8944878164182635, + "grad_norm": 2.8471264839172363, + "learning_rate": 4.314412257601149e-06, + "loss": 0.3438, + "step": 24470 + }, + { + "epoch": 2.8956706884315118, + "grad_norm": 2.771209478378296, + "learning_rate": 4.313694038783816e-06, + "loss": 0.3138, + "step": 24480 + }, + { + "epoch": 2.89685356044476, + "grad_norm": 2.642932176589966, + "learning_rate": 4.312975819966483e-06, + "loss": 0.3052, + "step": 24490 + }, + { + "epoch": 2.898036432458008, + "grad_norm": 2.9823434352874756, + "learning_rate": 4.31225760114915e-06, + "loss": 0.3557, + "step": 24500 + }, + { + "epoch": 2.899219304471256, + "grad_norm": 2.5010318756103516, + "learning_rate": 4.311539382331818e-06, + "loss": 0.2704, + "step": 24510 + }, + { + "epoch": 2.9004021764845045, + "grad_norm": 2.6502225399017334, + "learning_rate": 4.3108211635144845e-06, + "loss": 0.3315, + "step": 24520 + }, + { + "epoch": 2.9015850484977523, + "grad_norm": 3.0236613750457764, + "learning_rate": 4.3101029446971514e-06, + "loss": 0.2636, + "step": 24530 + }, + { + "epoch": 2.9027679205110006, + "grad_norm": 2.8792481422424316, + "learning_rate": 4.309384725879818e-06, + "loss": 0.325, + "step": 24540 + }, + { + "epoch": 2.903950792524249, + "grad_norm": 4.091868877410889, + "learning_rate": 4.308666507062485e-06, + "loss": 0.3304, + "step": 24550 + }, + { + "epoch": 2.905133664537497, + "grad_norm": 2.891231060028076, + "learning_rate": 4.307948288245152e-06, + "loss": 0.3346, + "step": 24560 + }, + { + "epoch": 2.906316536550745, + "grad_norm": 2.7249867916107178, + "learning_rate": 4.307230069427819e-06, + "loss": 0.32, + "step": 24570 + }, + { + "epoch": 2.9074994085639934, + "grad_norm": 2.7951316833496094, + "learning_rate": 4.306511850610486e-06, + "loss": 0.3018, + "step": 24580 + }, + { + "epoch": 2.9086822805772417, + "grad_norm": 3.3010826110839844, + "learning_rate": 4.305793631793153e-06, + "loss": 0.2936, + "step": 24590 + }, + { + "epoch": 2.9098651525904895, + "grad_norm": 2.40899920463562, + "learning_rate": 4.30507541297582e-06, + "loss": 0.2715, + "step": 24600 + }, + { + "epoch": 2.911048024603738, + "grad_norm": 2.509979724884033, + "learning_rate": 4.304357194158487e-06, + "loss": 0.3051, + "step": 24610 + }, + { + "epoch": 2.912230896616986, + "grad_norm": 3.967857599258423, + "learning_rate": 4.303638975341155e-06, + "loss": 0.3077, + "step": 24620 + }, + { + "epoch": 2.9134137686302344, + "grad_norm": 2.6073169708251953, + "learning_rate": 4.302920756523821e-06, + "loss": 0.3067, + "step": 24630 + }, + { + "epoch": 2.9145966406434822, + "grad_norm": 2.43867564201355, + "learning_rate": 4.3022025377064884e-06, + "loss": 0.3202, + "step": 24640 + }, + { + "epoch": 2.9157795126567305, + "grad_norm": 3.090456247329712, + "learning_rate": 4.3014843188891545e-06, + "loss": 0.2951, + "step": 24650 + }, + { + "epoch": 2.916962384669979, + "grad_norm": 2.836073160171509, + "learning_rate": 4.300766100071822e-06, + "loss": 0.3012, + "step": 24660 + }, + { + "epoch": 2.9181452566832267, + "grad_norm": 2.6881330013275146, + "learning_rate": 4.300047881254488e-06, + "loss": 0.3224, + "step": 24670 + }, + { + "epoch": 2.919328128696475, + "grad_norm": 2.8682494163513184, + "learning_rate": 4.299329662437156e-06, + "loss": 0.3178, + "step": 24680 + }, + { + "epoch": 2.9205110007097232, + "grad_norm": 2.399474620819092, + "learning_rate": 4.298611443619823e-06, + "loss": 0.2934, + "step": 24690 + }, + { + "epoch": 2.9216938727229715, + "grad_norm": 2.8152213096618652, + "learning_rate": 4.29789322480249e-06, + "loss": 0.3258, + "step": 24700 + }, + { + "epoch": 2.9228767447362194, + "grad_norm": 2.896705389022827, + "learning_rate": 4.297175005985157e-06, + "loss": 0.3083, + "step": 24710 + }, + { + "epoch": 2.9240596167494677, + "grad_norm": 3.403679132461548, + "learning_rate": 4.296456787167824e-06, + "loss": 0.309, + "step": 24720 + }, + { + "epoch": 2.925242488762716, + "grad_norm": 2.7205591201782227, + "learning_rate": 4.295738568350491e-06, + "loss": 0.3258, + "step": 24730 + }, + { + "epoch": 2.926425360775964, + "grad_norm": 2.4609246253967285, + "learning_rate": 4.295020349533158e-06, + "loss": 0.2887, + "step": 24740 + }, + { + "epoch": 2.927608232789212, + "grad_norm": 2.5012879371643066, + "learning_rate": 4.294302130715825e-06, + "loss": 0.3044, + "step": 24750 + }, + { + "epoch": 2.9287911048024604, + "grad_norm": 3.2999508380889893, + "learning_rate": 4.2935839118984915e-06, + "loss": 0.2661, + "step": 24760 + }, + { + "epoch": 2.9299739768157087, + "grad_norm": 2.513847827911377, + "learning_rate": 4.2928656930811584e-06, + "loss": 0.2638, + "step": 24770 + }, + { + "epoch": 2.9311568488289566, + "grad_norm": 2.975158452987671, + "learning_rate": 4.292147474263826e-06, + "loss": 0.3524, + "step": 24780 + }, + { + "epoch": 2.932339720842205, + "grad_norm": 2.611801862716675, + "learning_rate": 4.291429255446493e-06, + "loss": 0.3385, + "step": 24790 + }, + { + "epoch": 2.933522592855453, + "grad_norm": 1.6549113988876343, + "learning_rate": 4.29071103662916e-06, + "loss": 0.2907, + "step": 24800 + }, + { + "epoch": 2.934705464868701, + "grad_norm": 2.966315984725952, + "learning_rate": 4.289992817811827e-06, + "loss": 0.2909, + "step": 24810 + }, + { + "epoch": 2.9358883368819493, + "grad_norm": 2.4530463218688965, + "learning_rate": 4.289274598994494e-06, + "loss": 0.3181, + "step": 24820 + }, + { + "epoch": 2.9370712088951976, + "grad_norm": 2.3788347244262695, + "learning_rate": 4.288556380177161e-06, + "loss": 0.3199, + "step": 24830 + }, + { + "epoch": 2.938254080908446, + "grad_norm": 2.2570769786834717, + "learning_rate": 4.287838161359828e-06, + "loss": 0.2976, + "step": 24840 + }, + { + "epoch": 2.9394369529216937, + "grad_norm": 2.7231757640838623, + "learning_rate": 4.287119942542495e-06, + "loss": 0.3131, + "step": 24850 + }, + { + "epoch": 2.940619824934942, + "grad_norm": 2.869208812713623, + "learning_rate": 4.286401723725162e-06, + "loss": 0.2664, + "step": 24860 + }, + { + "epoch": 2.9418026969481903, + "grad_norm": 2.2201197147369385, + "learning_rate": 4.2856835049078285e-06, + "loss": 0.2978, + "step": 24870 + }, + { + "epoch": 2.942985568961438, + "grad_norm": 3.213529109954834, + "learning_rate": 4.2849652860904954e-06, + "loss": 0.3149, + "step": 24880 + }, + { + "epoch": 2.9441684409746864, + "grad_norm": 3.057131052017212, + "learning_rate": 4.284247067273163e-06, + "loss": 0.308, + "step": 24890 + }, + { + "epoch": 2.9453513129879347, + "grad_norm": 2.3544867038726807, + "learning_rate": 4.283528848455829e-06, + "loss": 0.3129, + "step": 24900 + }, + { + "epoch": 2.946534185001183, + "grad_norm": 2.861154794692993, + "learning_rate": 4.282810629638497e-06, + "loss": 0.2907, + "step": 24910 + }, + { + "epoch": 2.947717057014431, + "grad_norm": 2.8775789737701416, + "learning_rate": 4.282092410821163e-06, + "loss": 0.2641, + "step": 24920 + }, + { + "epoch": 2.948899929027679, + "grad_norm": 5.674015998840332, + "learning_rate": 4.281374192003831e-06, + "loss": 0.3215, + "step": 24930 + }, + { + "epoch": 2.9500828010409275, + "grad_norm": 2.2902441024780273, + "learning_rate": 4.280655973186497e-06, + "loss": 0.2831, + "step": 24940 + }, + { + "epoch": 2.9512656730541753, + "grad_norm": 3.575136423110962, + "learning_rate": 4.279937754369165e-06, + "loss": 0.3087, + "step": 24950 + }, + { + "epoch": 2.9524485450674236, + "grad_norm": 2.246788263320923, + "learning_rate": 4.279219535551832e-06, + "loss": 0.3228, + "step": 24960 + }, + { + "epoch": 2.953631417080672, + "grad_norm": 2.2626841068267822, + "learning_rate": 4.278501316734499e-06, + "loss": 0.3177, + "step": 24970 + }, + { + "epoch": 2.95481428909392, + "grad_norm": 2.23987078666687, + "learning_rate": 4.2777830979171655e-06, + "loss": 0.3079, + "step": 24980 + }, + { + "epoch": 2.9559971611071685, + "grad_norm": 2.4407312870025635, + "learning_rate": 4.2770648790998324e-06, + "loss": 0.2809, + "step": 24990 + }, + { + "epoch": 2.9571800331204163, + "grad_norm": 2.185276508331299, + "learning_rate": 4.276346660282499e-06, + "loss": 0.2932, + "step": 25000 + }, + { + "epoch": 2.9583629051336646, + "grad_norm": 2.1709935665130615, + "learning_rate": 4.275628441465166e-06, + "loss": 0.3016, + "step": 25010 + }, + { + "epoch": 2.9595457771469125, + "grad_norm": 3.0491926670074463, + "learning_rate": 4.274910222647833e-06, + "loss": 0.2597, + "step": 25020 + }, + { + "epoch": 2.9607286491601608, + "grad_norm": 3.196272134780884, + "learning_rate": 4.2741920038305e-06, + "loss": 0.3248, + "step": 25030 + }, + { + "epoch": 2.961911521173409, + "grad_norm": 2.5127756595611572, + "learning_rate": 4.273473785013168e-06, + "loss": 0.3259, + "step": 25040 + }, + { + "epoch": 2.9630943931866573, + "grad_norm": 2.741196393966675, + "learning_rate": 4.272755566195835e-06, + "loss": 0.3141, + "step": 25050 + }, + { + "epoch": 2.9642772651999056, + "grad_norm": 2.100416421890259, + "learning_rate": 4.272037347378502e-06, + "loss": 0.2968, + "step": 25060 + }, + { + "epoch": 2.9654601372131535, + "grad_norm": 3.8466129302978516, + "learning_rate": 4.271319128561169e-06, + "loss": 0.2745, + "step": 25070 + }, + { + "epoch": 2.966643009226402, + "grad_norm": 2.1351864337921143, + "learning_rate": 4.270600909743836e-06, + "loss": 0.2583, + "step": 25080 + }, + { + "epoch": 2.9678258812396496, + "grad_norm": 2.9576733112335205, + "learning_rate": 4.2698826909265025e-06, + "loss": 0.3536, + "step": 25090 + }, + { + "epoch": 2.969008753252898, + "grad_norm": 2.7691287994384766, + "learning_rate": 4.2691644721091694e-06, + "loss": 0.3374, + "step": 25100 + }, + { + "epoch": 2.970191625266146, + "grad_norm": 1.9637224674224854, + "learning_rate": 4.268446253291836e-06, + "loss": 0.3056, + "step": 25110 + }, + { + "epoch": 2.9713744972793945, + "grad_norm": 3.046872138977051, + "learning_rate": 4.267728034474503e-06, + "loss": 0.3292, + "step": 25120 + }, + { + "epoch": 2.972557369292643, + "grad_norm": 1.9840370416641235, + "learning_rate": 4.26700981565717e-06, + "loss": 0.3321, + "step": 25130 + }, + { + "epoch": 2.9737402413058907, + "grad_norm": 2.9284114837646484, + "learning_rate": 4.266291596839837e-06, + "loss": 0.2942, + "step": 25140 + }, + { + "epoch": 2.974923113319139, + "grad_norm": 2.532626152038574, + "learning_rate": 4.265573378022504e-06, + "loss": 0.345, + "step": 25150 + }, + { + "epoch": 2.976105985332387, + "grad_norm": 4.168957233428955, + "learning_rate": 4.264855159205172e-06, + "loss": 0.3143, + "step": 25160 + }, + { + "epoch": 2.977288857345635, + "grad_norm": 2.9889402389526367, + "learning_rate": 4.264136940387838e-06, + "loss": 0.3262, + "step": 25170 + }, + { + "epoch": 2.9784717293588834, + "grad_norm": 2.7924888134002686, + "learning_rate": 4.263418721570506e-06, + "loss": 0.3165, + "step": 25180 + }, + { + "epoch": 2.9796546013721317, + "grad_norm": 2.8963141441345215, + "learning_rate": 4.262700502753172e-06, + "loss": 0.2907, + "step": 25190 + }, + { + "epoch": 2.98083747338538, + "grad_norm": 2.3187415599823, + "learning_rate": 4.2619822839358395e-06, + "loss": 0.3271, + "step": 25200 + }, + { + "epoch": 2.982020345398628, + "grad_norm": 2.196073532104492, + "learning_rate": 4.261264065118506e-06, + "loss": 0.2723, + "step": 25210 + }, + { + "epoch": 2.983203217411876, + "grad_norm": 2.0752670764923096, + "learning_rate": 4.260545846301173e-06, + "loss": 0.2533, + "step": 25220 + }, + { + "epoch": 2.984386089425124, + "grad_norm": 2.282978057861328, + "learning_rate": 4.25982762748384e-06, + "loss": 0.2677, + "step": 25230 + }, + { + "epoch": 2.9855689614383722, + "grad_norm": 3.7450320720672607, + "learning_rate": 4.259109408666507e-06, + "loss": 0.281, + "step": 25240 + }, + { + "epoch": 2.9867518334516205, + "grad_norm": 3.024444103240967, + "learning_rate": 4.258391189849174e-06, + "loss": 0.3004, + "step": 25250 + }, + { + "epoch": 2.987934705464869, + "grad_norm": 1.9286129474639893, + "learning_rate": 4.257672971031841e-06, + "loss": 0.2872, + "step": 25260 + }, + { + "epoch": 2.989117577478117, + "grad_norm": 2.5781302452087402, + "learning_rate": 4.256954752214508e-06, + "loss": 0.2969, + "step": 25270 + }, + { + "epoch": 2.990300449491365, + "grad_norm": 3.5776491165161133, + "learning_rate": 4.256236533397175e-06, + "loss": 0.2907, + "step": 25280 + }, + { + "epoch": 2.9914833215046133, + "grad_norm": 2.300605297088623, + "learning_rate": 4.255518314579843e-06, + "loss": 0.3402, + "step": 25290 + }, + { + "epoch": 2.992666193517861, + "grad_norm": 3.1672215461730957, + "learning_rate": 4.254800095762509e-06, + "loss": 0.2997, + "step": 25300 + }, + { + "epoch": 2.9938490655311094, + "grad_norm": 1.8890864849090576, + "learning_rate": 4.2540818769451765e-06, + "loss": 0.2888, + "step": 25310 + }, + { + "epoch": 2.9950319375443577, + "grad_norm": 2.8092312812805176, + "learning_rate": 4.253363658127843e-06, + "loss": 0.3448, + "step": 25320 + }, + { + "epoch": 2.996214809557606, + "grad_norm": 3.5105812549591064, + "learning_rate": 4.25264543931051e-06, + "loss": 0.3037, + "step": 25330 + }, + { + "epoch": 2.9973976815708543, + "grad_norm": 2.169095277786255, + "learning_rate": 4.251927220493177e-06, + "loss": 0.3121, + "step": 25340 + }, + { + "epoch": 2.998580553584102, + "grad_norm": 3.7993338108062744, + "learning_rate": 4.251209001675844e-06, + "loss": 0.3087, + "step": 25350 + }, + { + "epoch": 2.9997634255973504, + "grad_norm": 3.136486768722534, + "learning_rate": 4.250490782858511e-06, + "loss": 0.2993, + "step": 25360 + }, + { + "epoch": 3.0007097232079487, + "eval_accuracy": 0.8604651162790697, + "eval_loss": 0.3228709399700165, + "eval_runtime": 80.8873, + "eval_safe_aucpr": 0.9165642380986325, + "eval_safe_f1": 0.8447241762310256, + "eval_safe_fpr": 0.1357507996771393, + "eval_safe_precision": 0.8340095770735095, + "eval_safe_recall": 0.8557176611784121, + "eval_samples_per_second": 743.182, + "eval_steps_per_second": 11.621, + "eval_unsafe_aucpr": 0.95316312759668, + "eval_unsafe_f1": 0.8733083615273078, + "eval_unsafe_fpr": 0.14428233882158745, + "eval_unsafe_precision": 0.8825594529413561, + "eval_unsafe_recall": 0.8642492003228603, + "step": 25368 + }, + { + "epoch": 3.0009462976105987, + "grad_norm": 2.4251770973205566, + "learning_rate": 4.249772564041178e-06, + "loss": 0.324, + "step": 25370 + }, + { + "epoch": 3.0021291696238466, + "grad_norm": 2.4300856590270996, + "learning_rate": 4.249054345223845e-06, + "loss": 0.2973, + "step": 25380 + }, + { + "epoch": 3.003312041637095, + "grad_norm": 3.1689414978027344, + "learning_rate": 4.248336126406512e-06, + "loss": 0.2707, + "step": 25390 + }, + { + "epoch": 3.004494913650343, + "grad_norm": 3.675501585006714, + "learning_rate": 4.247617907589179e-06, + "loss": 0.2787, + "step": 25400 + }, + { + "epoch": 3.005677785663591, + "grad_norm": 2.5266361236572266, + "learning_rate": 4.246899688771846e-06, + "loss": 0.3087, + "step": 25410 + }, + { + "epoch": 3.0068606576768393, + "grad_norm": 2.952092170715332, + "learning_rate": 4.246181469954513e-06, + "loss": 0.2941, + "step": 25420 + }, + { + "epoch": 3.0080435296900876, + "grad_norm": 3.5902297496795654, + "learning_rate": 4.24546325113718e-06, + "loss": 0.3034, + "step": 25430 + }, + { + "epoch": 3.009226401703336, + "grad_norm": 3.6910159587860107, + "learning_rate": 4.2447450323198465e-06, + "loss": 0.3116, + "step": 25440 + }, + { + "epoch": 3.0104092737165837, + "grad_norm": 2.407165288925171, + "learning_rate": 4.244026813502514e-06, + "loss": 0.2696, + "step": 25450 + }, + { + "epoch": 3.011592145729832, + "grad_norm": 3.68829345703125, + "learning_rate": 4.24330859468518e-06, + "loss": 0.2955, + "step": 25460 + }, + { + "epoch": 3.0127750177430803, + "grad_norm": 4.646860122680664, + "learning_rate": 4.242590375867848e-06, + "loss": 0.3156, + "step": 25470 + }, + { + "epoch": 3.013957889756328, + "grad_norm": 2.9903318881988525, + "learning_rate": 4.241872157050514e-06, + "loss": 0.3033, + "step": 25480 + }, + { + "epoch": 3.0151407617695765, + "grad_norm": 2.9443302154541016, + "learning_rate": 4.241153938233182e-06, + "loss": 0.2607, + "step": 25490 + }, + { + "epoch": 3.0163236337828248, + "grad_norm": 3.1221108436584473, + "learning_rate": 4.240435719415849e-06, + "loss": 0.2323, + "step": 25500 + }, + { + "epoch": 3.017506505796073, + "grad_norm": 4.149600028991699, + "learning_rate": 4.239717500598516e-06, + "loss": 0.3212, + "step": 25510 + }, + { + "epoch": 3.018689377809321, + "grad_norm": 4.904343605041504, + "learning_rate": 4.238999281781183e-06, + "loss": 0.3559, + "step": 25520 + }, + { + "epoch": 3.019872249822569, + "grad_norm": 2.3706648349761963, + "learning_rate": 4.23828106296385e-06, + "loss": 0.2591, + "step": 25530 + }, + { + "epoch": 3.0210551218358175, + "grad_norm": 3.659315586090088, + "learning_rate": 4.237562844146517e-06, + "loss": 0.2776, + "step": 25540 + }, + { + "epoch": 3.0222379938490653, + "grad_norm": 3.6037652492523193, + "learning_rate": 4.2368446253291835e-06, + "loss": 0.2918, + "step": 25550 + }, + { + "epoch": 3.0234208658623136, + "grad_norm": 3.219311237335205, + "learning_rate": 4.236126406511851e-06, + "loss": 0.2757, + "step": 25560 + }, + { + "epoch": 3.024603737875562, + "grad_norm": 3.2861392498016357, + "learning_rate": 4.235408187694517e-06, + "loss": 0.3014, + "step": 25570 + }, + { + "epoch": 3.02578660988881, + "grad_norm": 2.407735586166382, + "learning_rate": 4.234689968877185e-06, + "loss": 0.3028, + "step": 25580 + }, + { + "epoch": 3.026969481902058, + "grad_norm": 2.3614001274108887, + "learning_rate": 4.233971750059851e-06, + "loss": 0.2979, + "step": 25590 + }, + { + "epoch": 3.0281523539153063, + "grad_norm": 2.4324471950531006, + "learning_rate": 4.233253531242519e-06, + "loss": 0.2938, + "step": 25600 + }, + { + "epoch": 3.0293352259285546, + "grad_norm": 3.4657468795776367, + "learning_rate": 4.232535312425186e-06, + "loss": 0.3027, + "step": 25610 + }, + { + "epoch": 3.0305180979418025, + "grad_norm": 2.8468799591064453, + "learning_rate": 4.231817093607853e-06, + "loss": 0.2888, + "step": 25620 + }, + { + "epoch": 3.031700969955051, + "grad_norm": 3.5469186305999756, + "learning_rate": 4.23109887479052e-06, + "loss": 0.2841, + "step": 25630 + }, + { + "epoch": 3.032883841968299, + "grad_norm": 2.9140517711639404, + "learning_rate": 4.230380655973187e-06, + "loss": 0.3127, + "step": 25640 + }, + { + "epoch": 3.0340667139815474, + "grad_norm": 3.262416124343872, + "learning_rate": 4.2296624371558536e-06, + "loss": 0.2479, + "step": 25650 + }, + { + "epoch": 3.035249585994795, + "grad_norm": 3.1434836387634277, + "learning_rate": 4.2289442183385205e-06, + "loss": 0.2631, + "step": 25660 + }, + { + "epoch": 3.0364324580080435, + "grad_norm": 2.56406569480896, + "learning_rate": 4.228225999521187e-06, + "loss": 0.2893, + "step": 25670 + }, + { + "epoch": 3.037615330021292, + "grad_norm": 4.192211627960205, + "learning_rate": 4.227507780703854e-06, + "loss": 0.2978, + "step": 25680 + }, + { + "epoch": 3.0387982020345397, + "grad_norm": 2.9986400604248047, + "learning_rate": 4.226789561886521e-06, + "loss": 0.299, + "step": 25690 + }, + { + "epoch": 3.039981074047788, + "grad_norm": 2.8809149265289307, + "learning_rate": 4.226071343069189e-06, + "loss": 0.2407, + "step": 25700 + }, + { + "epoch": 3.0411639460610362, + "grad_norm": 2.47594952583313, + "learning_rate": 4.225353124251855e-06, + "loss": 0.252, + "step": 25710 + }, + { + "epoch": 3.0423468180742845, + "grad_norm": 2.641633987426758, + "learning_rate": 4.224634905434523e-06, + "loss": 0.2982, + "step": 25720 + }, + { + "epoch": 3.0435296900875324, + "grad_norm": 2.83441424369812, + "learning_rate": 4.223916686617189e-06, + "loss": 0.2832, + "step": 25730 + }, + { + "epoch": 3.0447125621007807, + "grad_norm": 4.333640098571777, + "learning_rate": 4.223198467799857e-06, + "loss": 0.2938, + "step": 25740 + }, + { + "epoch": 3.045895434114029, + "grad_norm": 3.4308552742004395, + "learning_rate": 4.222480248982523e-06, + "loss": 0.3003, + "step": 25750 + }, + { + "epoch": 3.047078306127277, + "grad_norm": 2.4360756874084473, + "learning_rate": 4.2217620301651906e-06, + "loss": 0.3109, + "step": 25760 + }, + { + "epoch": 3.048261178140525, + "grad_norm": 2.7177233695983887, + "learning_rate": 4.2210438113478575e-06, + "loss": 0.2916, + "step": 25770 + }, + { + "epoch": 3.0494440501537734, + "grad_norm": 2.774137020111084, + "learning_rate": 4.220325592530524e-06, + "loss": 0.2899, + "step": 25780 + }, + { + "epoch": 3.0506269221670217, + "grad_norm": 3.293748378753662, + "learning_rate": 4.219607373713191e-06, + "loss": 0.2763, + "step": 25790 + }, + { + "epoch": 3.0518097941802695, + "grad_norm": 2.5247225761413574, + "learning_rate": 4.218889154895858e-06, + "loss": 0.2886, + "step": 25800 + }, + { + "epoch": 3.052992666193518, + "grad_norm": 2.5937037467956543, + "learning_rate": 4.218170936078526e-06, + "loss": 0.3018, + "step": 25810 + }, + { + "epoch": 3.054175538206766, + "grad_norm": 3.2484025955200195, + "learning_rate": 4.217452717261192e-06, + "loss": 0.3125, + "step": 25820 + }, + { + "epoch": 3.055358410220014, + "grad_norm": 3.243189573287964, + "learning_rate": 4.21673449844386e-06, + "loss": 0.2932, + "step": 25830 + }, + { + "epoch": 3.0565412822332623, + "grad_norm": 2.8589673042297363, + "learning_rate": 4.216016279626526e-06, + "loss": 0.295, + "step": 25840 + }, + { + "epoch": 3.0577241542465106, + "grad_norm": 2.360625743865967, + "learning_rate": 4.215298060809194e-06, + "loss": 0.2603, + "step": 25850 + }, + { + "epoch": 3.058907026259759, + "grad_norm": 2.2223730087280273, + "learning_rate": 4.21457984199186e-06, + "loss": 0.2855, + "step": 25860 + }, + { + "epoch": 3.0600898982730067, + "grad_norm": 3.6078221797943115, + "learning_rate": 4.2138616231745276e-06, + "loss": 0.2577, + "step": 25870 + }, + { + "epoch": 3.061272770286255, + "grad_norm": 4.722879886627197, + "learning_rate": 4.2131434043571945e-06, + "loss": 0.2839, + "step": 25880 + }, + { + "epoch": 3.0624556422995033, + "grad_norm": 4.164831161499023, + "learning_rate": 4.212425185539861e-06, + "loss": 0.3016, + "step": 25890 + }, + { + "epoch": 3.0636385143127516, + "grad_norm": 3.193854808807373, + "learning_rate": 4.211706966722528e-06, + "loss": 0.2676, + "step": 25900 + }, + { + "epoch": 3.0648213863259994, + "grad_norm": 2.405898332595825, + "learning_rate": 4.210988747905195e-06, + "loss": 0.3062, + "step": 25910 + }, + { + "epoch": 3.0660042583392477, + "grad_norm": 3.8819572925567627, + "learning_rate": 4.210270529087862e-06, + "loss": 0.3073, + "step": 25920 + }, + { + "epoch": 3.067187130352496, + "grad_norm": 2.2638633251190186, + "learning_rate": 4.209552310270529e-06, + "loss": 0.3137, + "step": 25930 + }, + { + "epoch": 3.068370002365744, + "grad_norm": 3.4960858821868896, + "learning_rate": 4.208834091453196e-06, + "loss": 0.2656, + "step": 25940 + }, + { + "epoch": 3.069552874378992, + "grad_norm": 2.579909086227417, + "learning_rate": 4.208115872635863e-06, + "loss": 0.2826, + "step": 25950 + }, + { + "epoch": 3.0707357463922405, + "grad_norm": 2.192331552505493, + "learning_rate": 4.20739765381853e-06, + "loss": 0.2903, + "step": 25960 + }, + { + "epoch": 3.0719186184054887, + "grad_norm": 2.671562910079956, + "learning_rate": 4.206679435001198e-06, + "loss": 0.3281, + "step": 25970 + }, + { + "epoch": 3.0731014904187366, + "grad_norm": 2.543046474456787, + "learning_rate": 4.205961216183864e-06, + "loss": 0.2938, + "step": 25980 + }, + { + "epoch": 3.074284362431985, + "grad_norm": 2.1825907230377197, + "learning_rate": 4.2052429973665315e-06, + "loss": 0.2729, + "step": 25990 + }, + { + "epoch": 3.075467234445233, + "grad_norm": 3.01511812210083, + "learning_rate": 4.2045247785491976e-06, + "loss": 0.3047, + "step": 26000 + }, + { + "epoch": 3.076650106458481, + "grad_norm": 2.9775047302246094, + "learning_rate": 4.203806559731865e-06, + "loss": 0.31, + "step": 26010 + }, + { + "epoch": 3.0778329784717293, + "grad_norm": 2.236973762512207, + "learning_rate": 4.203088340914531e-06, + "loss": 0.2602, + "step": 26020 + }, + { + "epoch": 3.0790158504849776, + "grad_norm": 3.7626545429229736, + "learning_rate": 4.202370122097199e-06, + "loss": 0.2554, + "step": 26030 + }, + { + "epoch": 3.080198722498226, + "grad_norm": 3.7447025775909424, + "learning_rate": 4.201651903279866e-06, + "loss": 0.2911, + "step": 26040 + }, + { + "epoch": 3.0813815945114738, + "grad_norm": 3.142407178878784, + "learning_rate": 4.200933684462533e-06, + "loss": 0.3039, + "step": 26050 + }, + { + "epoch": 3.082564466524722, + "grad_norm": 2.8024702072143555, + "learning_rate": 4.2002154656452e-06, + "loss": 0.264, + "step": 26060 + }, + { + "epoch": 3.0837473385379703, + "grad_norm": 3.014652967453003, + "learning_rate": 4.199497246827867e-06, + "loss": 0.2737, + "step": 26070 + }, + { + "epoch": 3.084930210551218, + "grad_norm": 3.9171454906463623, + "learning_rate": 4.198779028010535e-06, + "loss": 0.306, + "step": 26080 + }, + { + "epoch": 3.0861130825644665, + "grad_norm": 3.677460193634033, + "learning_rate": 4.198060809193201e-06, + "loss": 0.3058, + "step": 26090 + }, + { + "epoch": 3.0872959545777148, + "grad_norm": 3.0797815322875977, + "learning_rate": 4.1973425903758685e-06, + "loss": 0.2664, + "step": 26100 + }, + { + "epoch": 3.088478826590963, + "grad_norm": 3.137662649154663, + "learning_rate": 4.1966243715585346e-06, + "loss": 0.292, + "step": 26110 + }, + { + "epoch": 3.089661698604211, + "grad_norm": 3.9400734901428223, + "learning_rate": 4.195906152741202e-06, + "loss": 0.2802, + "step": 26120 + }, + { + "epoch": 3.090844570617459, + "grad_norm": 2.6781139373779297, + "learning_rate": 4.195187933923868e-06, + "loss": 0.2936, + "step": 26130 + }, + { + "epoch": 3.0920274426307075, + "grad_norm": 4.493440628051758, + "learning_rate": 4.194469715106536e-06, + "loss": 0.2962, + "step": 26140 + }, + { + "epoch": 3.0932103146439553, + "grad_norm": 2.9951131343841553, + "learning_rate": 4.193751496289203e-06, + "loss": 0.2792, + "step": 26150 + }, + { + "epoch": 3.0943931866572036, + "grad_norm": 3.6866304874420166, + "learning_rate": 4.19303327747187e-06, + "loss": 0.2699, + "step": 26160 + }, + { + "epoch": 3.095576058670452, + "grad_norm": 3.072524070739746, + "learning_rate": 4.192315058654537e-06, + "loss": 0.3185, + "step": 26170 + }, + { + "epoch": 3.0967589306837002, + "grad_norm": 3.688554525375366, + "learning_rate": 4.191596839837204e-06, + "loss": 0.341, + "step": 26180 + }, + { + "epoch": 3.097941802696948, + "grad_norm": 2.4094104766845703, + "learning_rate": 4.190878621019871e-06, + "loss": 0.2707, + "step": 26190 + }, + { + "epoch": 3.0991246747101964, + "grad_norm": 2.873302459716797, + "learning_rate": 4.190160402202538e-06, + "loss": 0.3242, + "step": 26200 + }, + { + "epoch": 3.1003075467234447, + "grad_norm": 2.9964234828948975, + "learning_rate": 4.189442183385205e-06, + "loss": 0.2686, + "step": 26210 + }, + { + "epoch": 3.1014904187366925, + "grad_norm": 2.406423330307007, + "learning_rate": 4.1887239645678716e-06, + "loss": 0.2489, + "step": 26220 + }, + { + "epoch": 3.102673290749941, + "grad_norm": 2.8716514110565186, + "learning_rate": 4.1880057457505385e-06, + "loss": 0.2898, + "step": 26230 + }, + { + "epoch": 3.103856162763189, + "grad_norm": 3.505645990371704, + "learning_rate": 4.187287526933205e-06, + "loss": 0.2902, + "step": 26240 + }, + { + "epoch": 3.1050390347764374, + "grad_norm": 3.3556034564971924, + "learning_rate": 4.186569308115872e-06, + "loss": 0.2803, + "step": 26250 + }, + { + "epoch": 3.1062219067896852, + "grad_norm": 2.842924118041992, + "learning_rate": 4.18585108929854e-06, + "loss": 0.2709, + "step": 26260 + }, + { + "epoch": 3.1074047788029335, + "grad_norm": 3.348853588104248, + "learning_rate": 4.185132870481206e-06, + "loss": 0.3035, + "step": 26270 + }, + { + "epoch": 3.108587650816182, + "grad_norm": 3.3913230895996094, + "learning_rate": 4.184414651663874e-06, + "loss": 0.2929, + "step": 26280 + }, + { + "epoch": 3.1097705228294297, + "grad_norm": 2.457864761352539, + "learning_rate": 4.18369643284654e-06, + "loss": 0.326, + "step": 26290 + }, + { + "epoch": 3.110953394842678, + "grad_norm": 2.600060224533081, + "learning_rate": 4.182978214029208e-06, + "loss": 0.2937, + "step": 26300 + }, + { + "epoch": 3.1121362668559263, + "grad_norm": 2.278590440750122, + "learning_rate": 4.182259995211875e-06, + "loss": 0.2695, + "step": 26310 + }, + { + "epoch": 3.1133191388691746, + "grad_norm": 3.3055567741394043, + "learning_rate": 4.181541776394542e-06, + "loss": 0.3319, + "step": 26320 + }, + { + "epoch": 3.1145020108824224, + "grad_norm": 2.4813735485076904, + "learning_rate": 4.1808235575772086e-06, + "loss": 0.2872, + "step": 26330 + }, + { + "epoch": 3.1156848828956707, + "grad_norm": 2.511098861694336, + "learning_rate": 4.1801053387598755e-06, + "loss": 0.2558, + "step": 26340 + }, + { + "epoch": 3.116867754908919, + "grad_norm": 3.455292224884033, + "learning_rate": 4.179387119942543e-06, + "loss": 0.2826, + "step": 26350 + }, + { + "epoch": 3.118050626922167, + "grad_norm": 4.4167680740356445, + "learning_rate": 4.178668901125209e-06, + "loss": 0.2787, + "step": 26360 + }, + { + "epoch": 3.119233498935415, + "grad_norm": 3.423621892929077, + "learning_rate": 4.177950682307877e-06, + "loss": 0.2755, + "step": 26370 + }, + { + "epoch": 3.1204163709486634, + "grad_norm": 2.4179739952087402, + "learning_rate": 4.177232463490543e-06, + "loss": 0.287, + "step": 26380 + }, + { + "epoch": 3.1215992429619117, + "grad_norm": 3.2276782989501953, + "learning_rate": 4.176514244673211e-06, + "loss": 0.2641, + "step": 26390 + }, + { + "epoch": 3.1227821149751596, + "grad_norm": 2.348263740539551, + "learning_rate": 4.175796025855877e-06, + "loss": 0.2567, + "step": 26400 + }, + { + "epoch": 3.123964986988408, + "grad_norm": 3.7675414085388184, + "learning_rate": 4.175077807038545e-06, + "loss": 0.2954, + "step": 26410 + }, + { + "epoch": 3.125147859001656, + "grad_norm": 2.0599918365478516, + "learning_rate": 4.174359588221212e-06, + "loss": 0.2725, + "step": 26420 + }, + { + "epoch": 3.126330731014904, + "grad_norm": 2.903104305267334, + "learning_rate": 4.173641369403879e-06, + "loss": 0.2488, + "step": 26430 + }, + { + "epoch": 3.1275136030281523, + "grad_norm": 3.985673189163208, + "learning_rate": 4.1729231505865456e-06, + "loss": 0.3004, + "step": 26440 + }, + { + "epoch": 3.1286964750414006, + "grad_norm": 3.22426700592041, + "learning_rate": 4.1722049317692125e-06, + "loss": 0.2582, + "step": 26450 + }, + { + "epoch": 3.129879347054649, + "grad_norm": 3.736717700958252, + "learning_rate": 4.171486712951879e-06, + "loss": 0.3004, + "step": 26460 + }, + { + "epoch": 3.1310622190678967, + "grad_norm": 3.241243362426758, + "learning_rate": 4.170768494134546e-06, + "loss": 0.296, + "step": 26470 + }, + { + "epoch": 3.132245091081145, + "grad_norm": 3.2949328422546387, + "learning_rate": 4.170050275317213e-06, + "loss": 0.3042, + "step": 26480 + }, + { + "epoch": 3.1334279630943933, + "grad_norm": 2.447862148284912, + "learning_rate": 4.16933205649988e-06, + "loss": 0.2656, + "step": 26490 + }, + { + "epoch": 3.134610835107641, + "grad_norm": 3.1463232040405273, + "learning_rate": 4.168613837682547e-06, + "loss": 0.3263, + "step": 26500 + }, + { + "epoch": 3.1357937071208895, + "grad_norm": 3.263399362564087, + "learning_rate": 4.167895618865214e-06, + "loss": 0.3124, + "step": 26510 + }, + { + "epoch": 3.1369765791341377, + "grad_norm": 3.49643611907959, + "learning_rate": 4.167177400047881e-06, + "loss": 0.2748, + "step": 26520 + }, + { + "epoch": 3.138159451147386, + "grad_norm": 2.467164993286133, + "learning_rate": 4.166459181230549e-06, + "loss": 0.2554, + "step": 26530 + }, + { + "epoch": 3.139342323160634, + "grad_norm": 3.2034528255462646, + "learning_rate": 4.165740962413215e-06, + "loss": 0.2237, + "step": 26540 + }, + { + "epoch": 3.140525195173882, + "grad_norm": 3.184513807296753, + "learning_rate": 4.1650227435958825e-06, + "loss": 0.2879, + "step": 26550 + }, + { + "epoch": 3.1417080671871305, + "grad_norm": 3.386791467666626, + "learning_rate": 4.1643045247785495e-06, + "loss": 0.2779, + "step": 26560 + }, + { + "epoch": 3.1428909392003783, + "grad_norm": 2.694204330444336, + "learning_rate": 4.163586305961216e-06, + "loss": 0.2807, + "step": 26570 + }, + { + "epoch": 3.1440738112136266, + "grad_norm": 3.4444072246551514, + "learning_rate": 4.162868087143883e-06, + "loss": 0.2786, + "step": 26580 + }, + { + "epoch": 3.145256683226875, + "grad_norm": 2.3471930027008057, + "learning_rate": 4.16214986832655e-06, + "loss": 0.2867, + "step": 26590 + }, + { + "epoch": 3.146439555240123, + "grad_norm": 2.7064785957336426, + "learning_rate": 4.161431649509217e-06, + "loss": 0.294, + "step": 26600 + }, + { + "epoch": 3.147622427253371, + "grad_norm": 2.492715358734131, + "learning_rate": 4.160713430691884e-06, + "loss": 0.3076, + "step": 26610 + }, + { + "epoch": 3.1488052992666193, + "grad_norm": 2.777384042739868, + "learning_rate": 4.159995211874552e-06, + "loss": 0.3291, + "step": 26620 + }, + { + "epoch": 3.1499881712798676, + "grad_norm": 2.460726261138916, + "learning_rate": 4.159276993057218e-06, + "loss": 0.2631, + "step": 26630 + }, + { + "epoch": 3.1511710432931155, + "grad_norm": 2.6271657943725586, + "learning_rate": 4.158558774239886e-06, + "loss": 0.3205, + "step": 26640 + }, + { + "epoch": 3.1523539153063638, + "grad_norm": 2.0604748725891113, + "learning_rate": 4.157840555422552e-06, + "loss": 0.234, + "step": 26650 + }, + { + "epoch": 3.153536787319612, + "grad_norm": 3.011331558227539, + "learning_rate": 4.1571223366052195e-06, + "loss": 0.3137, + "step": 26660 + }, + { + "epoch": 3.1547196593328604, + "grad_norm": 3.1126832962036133, + "learning_rate": 4.156404117787886e-06, + "loss": 0.2921, + "step": 26670 + }, + { + "epoch": 3.155902531346108, + "grad_norm": 2.7079503536224365, + "learning_rate": 4.155685898970553e-06, + "loss": 0.2412, + "step": 26680 + }, + { + "epoch": 3.1570854033593565, + "grad_norm": 2.268372058868408, + "learning_rate": 4.15496768015322e-06, + "loss": 0.2539, + "step": 26690 + }, + { + "epoch": 3.158268275372605, + "grad_norm": 2.8086812496185303, + "learning_rate": 4.154249461335887e-06, + "loss": 0.2623, + "step": 26700 + }, + { + "epoch": 3.1594511473858526, + "grad_norm": 4.629752159118652, + "learning_rate": 4.153531242518554e-06, + "loss": 0.302, + "step": 26710 + }, + { + "epoch": 3.160634019399101, + "grad_norm": 4.051839828491211, + "learning_rate": 4.152813023701221e-06, + "loss": 0.3064, + "step": 26720 + }, + { + "epoch": 3.1618168914123492, + "grad_norm": 4.20828104019165, + "learning_rate": 4.152094804883888e-06, + "loss": 0.297, + "step": 26730 + }, + { + "epoch": 3.1629997634255975, + "grad_norm": 3.0412421226501465, + "learning_rate": 4.151376586066555e-06, + "loss": 0.2532, + "step": 26740 + }, + { + "epoch": 3.1641826354388454, + "grad_norm": 3.1068429946899414, + "learning_rate": 4.150658367249222e-06, + "loss": 0.3012, + "step": 26750 + }, + { + "epoch": 3.1653655074520937, + "grad_norm": 2.692471742630005, + "learning_rate": 4.149940148431889e-06, + "loss": 0.2613, + "step": 26760 + }, + { + "epoch": 3.166548379465342, + "grad_norm": 4.510305881500244, + "learning_rate": 4.149221929614556e-06, + "loss": 0.2919, + "step": 26770 + }, + { + "epoch": 3.16773125147859, + "grad_norm": 3.4577081203460693, + "learning_rate": 4.148503710797223e-06, + "loss": 0.2821, + "step": 26780 + }, + { + "epoch": 3.168914123491838, + "grad_norm": 3.4594273567199707, + "learning_rate": 4.1477854919798895e-06, + "loss": 0.2767, + "step": 26790 + }, + { + "epoch": 3.1700969955050864, + "grad_norm": 3.4486966133117676, + "learning_rate": 4.147067273162557e-06, + "loss": 0.3115, + "step": 26800 + }, + { + "epoch": 3.1712798675183347, + "grad_norm": 3.1208348274230957, + "learning_rate": 4.146349054345224e-06, + "loss": 0.2534, + "step": 26810 + }, + { + "epoch": 3.1724627395315825, + "grad_norm": 3.5404696464538574, + "learning_rate": 4.145630835527891e-06, + "loss": 0.3081, + "step": 26820 + }, + { + "epoch": 3.173645611544831, + "grad_norm": 3.9507696628570557, + "learning_rate": 4.144912616710558e-06, + "loss": 0.3049, + "step": 26830 + }, + { + "epoch": 3.174828483558079, + "grad_norm": 2.977548599243164, + "learning_rate": 4.144194397893225e-06, + "loss": 0.3269, + "step": 26840 + }, + { + "epoch": 3.176011355571327, + "grad_norm": 2.3489491939544678, + "learning_rate": 4.143476179075892e-06, + "loss": 0.2973, + "step": 26850 + }, + { + "epoch": 3.1771942275845753, + "grad_norm": 2.961366653442383, + "learning_rate": 4.142757960258559e-06, + "loss": 0.2739, + "step": 26860 + }, + { + "epoch": 3.1783770995978236, + "grad_norm": 3.6704015731811523, + "learning_rate": 4.142039741441226e-06, + "loss": 0.2234, + "step": 26870 + }, + { + "epoch": 3.179559971611072, + "grad_norm": 3.6285295486450195, + "learning_rate": 4.141321522623893e-06, + "loss": 0.2764, + "step": 26880 + }, + { + "epoch": 3.1807428436243197, + "grad_norm": 3.0299291610717773, + "learning_rate": 4.1406033038065605e-06, + "loss": 0.3245, + "step": 26890 + }, + { + "epoch": 3.181925715637568, + "grad_norm": 2.3116037845611572, + "learning_rate": 4.1398850849892265e-06, + "loss": 0.2951, + "step": 26900 + }, + { + "epoch": 3.1831085876508163, + "grad_norm": 3.053914785385132, + "learning_rate": 4.139166866171894e-06, + "loss": 0.2644, + "step": 26910 + }, + { + "epoch": 3.184291459664064, + "grad_norm": 2.29020094871521, + "learning_rate": 4.13844864735456e-06, + "loss": 0.2855, + "step": 26920 + }, + { + "epoch": 3.1854743316773124, + "grad_norm": 2.8435728549957275, + "learning_rate": 4.137730428537228e-06, + "loss": 0.2589, + "step": 26930 + }, + { + "epoch": 3.1866572036905607, + "grad_norm": 1.9491233825683594, + "learning_rate": 4.137012209719894e-06, + "loss": 0.2356, + "step": 26940 + }, + { + "epoch": 3.187840075703809, + "grad_norm": 5.043102741241455, + "learning_rate": 4.136293990902562e-06, + "loss": 0.2842, + "step": 26950 + }, + { + "epoch": 3.189022947717057, + "grad_norm": 3.5162951946258545, + "learning_rate": 4.135575772085229e-06, + "loss": 0.2923, + "step": 26960 + }, + { + "epoch": 3.190205819730305, + "grad_norm": 3.5482230186462402, + "learning_rate": 4.134857553267896e-06, + "loss": 0.2906, + "step": 26970 + }, + { + "epoch": 3.1913886917435534, + "grad_norm": 2.1520400047302246, + "learning_rate": 4.134139334450563e-06, + "loss": 0.2535, + "step": 26980 + }, + { + "epoch": 3.1925715637568013, + "grad_norm": 3.102755308151245, + "learning_rate": 4.13342111563323e-06, + "loss": 0.3323, + "step": 26990 + }, + { + "epoch": 3.1937544357700496, + "grad_norm": 3.3286232948303223, + "learning_rate": 4.132702896815897e-06, + "loss": 0.3122, + "step": 27000 + }, + { + "epoch": 3.194937307783298, + "grad_norm": 2.536963701248169, + "learning_rate": 4.1319846779985635e-06, + "loss": 0.3061, + "step": 27010 + }, + { + "epoch": 3.196120179796546, + "grad_norm": 2.525052309036255, + "learning_rate": 4.1312664591812305e-06, + "loss": 0.2897, + "step": 27020 + }, + { + "epoch": 3.197303051809794, + "grad_norm": 3.134566307067871, + "learning_rate": 4.130548240363897e-06, + "loss": 0.2569, + "step": 27030 + }, + { + "epoch": 3.1984859238230423, + "grad_norm": 2.9788222312927246, + "learning_rate": 4.129830021546564e-06, + "loss": 0.3115, + "step": 27040 + }, + { + "epoch": 3.1996687958362906, + "grad_norm": 3.5071332454681396, + "learning_rate": 4.129111802729231e-06, + "loss": 0.2822, + "step": 27050 + }, + { + "epoch": 3.2008516678495385, + "grad_norm": 3.399627685546875, + "learning_rate": 4.128393583911899e-06, + "loss": 0.3212, + "step": 27060 + }, + { + "epoch": 3.2020345398627867, + "grad_norm": 3.182009696960449, + "learning_rate": 4.127675365094566e-06, + "loss": 0.3061, + "step": 27070 + }, + { + "epoch": 3.203217411876035, + "grad_norm": 3.0417139530181885, + "learning_rate": 4.126957146277233e-06, + "loss": 0.2725, + "step": 27080 + }, + { + "epoch": 3.2044002838892833, + "grad_norm": 3.618405818939209, + "learning_rate": 4.1262389274599e-06, + "loss": 0.284, + "step": 27090 + }, + { + "epoch": 3.205583155902531, + "grad_norm": 3.0297694206237793, + "learning_rate": 4.125520708642567e-06, + "loss": 0.2716, + "step": 27100 + }, + { + "epoch": 3.2067660279157795, + "grad_norm": 3.3863964080810547, + "learning_rate": 4.124802489825234e-06, + "loss": 0.3105, + "step": 27110 + }, + { + "epoch": 3.2079488999290278, + "grad_norm": 3.787440776824951, + "learning_rate": 4.1240842710079005e-06, + "loss": 0.2648, + "step": 27120 + }, + { + "epoch": 3.2091317719422756, + "grad_norm": 2.421241521835327, + "learning_rate": 4.1233660521905675e-06, + "loss": 0.2739, + "step": 27130 + }, + { + "epoch": 3.210314643955524, + "grad_norm": 2.030874729156494, + "learning_rate": 4.122647833373234e-06, + "loss": 0.2796, + "step": 27140 + }, + { + "epoch": 3.211497515968772, + "grad_norm": 2.9440155029296875, + "learning_rate": 4.121929614555901e-06, + "loss": 0.294, + "step": 27150 + }, + { + "epoch": 3.2126803879820205, + "grad_norm": 2.9077959060668945, + "learning_rate": 4.121211395738568e-06, + "loss": 0.3183, + "step": 27160 + }, + { + "epoch": 3.2138632599952683, + "grad_norm": 1.959438681602478, + "learning_rate": 4.120493176921235e-06, + "loss": 0.2543, + "step": 27170 + }, + { + "epoch": 3.2150461320085166, + "grad_norm": 2.29925537109375, + "learning_rate": 4.119774958103903e-06, + "loss": 0.3062, + "step": 27180 + }, + { + "epoch": 3.216229004021765, + "grad_norm": 2.8926730155944824, + "learning_rate": 4.119056739286569e-06, + "loss": 0.299, + "step": 27190 + }, + { + "epoch": 3.217411876035013, + "grad_norm": 3.610867738723755, + "learning_rate": 4.118338520469237e-06, + "loss": 0.2981, + "step": 27200 + }, + { + "epoch": 3.218594748048261, + "grad_norm": 2.610407829284668, + "learning_rate": 4.117620301651903e-06, + "loss": 0.2397, + "step": 27210 + }, + { + "epoch": 3.2197776200615094, + "grad_norm": 3.3427069187164307, + "learning_rate": 4.116902082834571e-06, + "loss": 0.2785, + "step": 27220 + }, + { + "epoch": 3.2209604920747577, + "grad_norm": 4.121489524841309, + "learning_rate": 4.1161838640172375e-06, + "loss": 0.255, + "step": 27230 + }, + { + "epoch": 3.2221433640880055, + "grad_norm": 2.9172403812408447, + "learning_rate": 4.1154656451999045e-06, + "loss": 0.3029, + "step": 27240 + }, + { + "epoch": 3.223326236101254, + "grad_norm": 3.5640225410461426, + "learning_rate": 4.114747426382571e-06, + "loss": 0.2919, + "step": 27250 + }, + { + "epoch": 3.224509108114502, + "grad_norm": 2.6899254322052, + "learning_rate": 4.114029207565238e-06, + "loss": 0.2827, + "step": 27260 + }, + { + "epoch": 3.2256919801277504, + "grad_norm": 2.448078155517578, + "learning_rate": 4.113310988747905e-06, + "loss": 0.268, + "step": 27270 + }, + { + "epoch": 3.2268748521409982, + "grad_norm": 3.2071781158447266, + "learning_rate": 4.112592769930572e-06, + "loss": 0.3381, + "step": 27280 + }, + { + "epoch": 3.2280577241542465, + "grad_norm": 4.201130390167236, + "learning_rate": 4.111874551113239e-06, + "loss": 0.2953, + "step": 27290 + }, + { + "epoch": 3.229240596167495, + "grad_norm": 3.7503464221954346, + "learning_rate": 4.111156332295906e-06, + "loss": 0.2597, + "step": 27300 + }, + { + "epoch": 3.2304234681807427, + "grad_norm": 2.761509418487549, + "learning_rate": 4.110438113478574e-06, + "loss": 0.2491, + "step": 27310 + }, + { + "epoch": 3.231606340193991, + "grad_norm": 2.505431652069092, + "learning_rate": 4.10971989466124e-06, + "loss": 0.3237, + "step": 27320 + }, + { + "epoch": 3.2327892122072392, + "grad_norm": 3.6810736656188965, + "learning_rate": 4.109001675843908e-06, + "loss": 0.2835, + "step": 27330 + }, + { + "epoch": 3.2339720842204875, + "grad_norm": 3.3416645526885986, + "learning_rate": 4.1082834570265745e-06, + "loss": 0.2891, + "step": 27340 + }, + { + "epoch": 3.2351549562337354, + "grad_norm": 3.9127213954925537, + "learning_rate": 4.1075652382092415e-06, + "loss": 0.3115, + "step": 27350 + }, + { + "epoch": 3.2363378282469837, + "grad_norm": 2.8421289920806885, + "learning_rate": 4.106847019391908e-06, + "loss": 0.2562, + "step": 27360 + }, + { + "epoch": 3.237520700260232, + "grad_norm": 3.3126332759857178, + "learning_rate": 4.106128800574575e-06, + "loss": 0.2883, + "step": 27370 + }, + { + "epoch": 3.23870357227348, + "grad_norm": 2.2606143951416016, + "learning_rate": 4.105410581757242e-06, + "loss": 0.2951, + "step": 27380 + }, + { + "epoch": 3.239886444286728, + "grad_norm": 3.2302465438842773, + "learning_rate": 4.104692362939909e-06, + "loss": 0.2649, + "step": 27390 + }, + { + "epoch": 3.2410693162999764, + "grad_norm": 2.675093412399292, + "learning_rate": 4.103974144122576e-06, + "loss": 0.2912, + "step": 27400 + }, + { + "epoch": 3.2422521883132247, + "grad_norm": 2.608977794647217, + "learning_rate": 4.103255925305243e-06, + "loss": 0.2848, + "step": 27410 + }, + { + "epoch": 3.2434350603264726, + "grad_norm": 2.509690523147583, + "learning_rate": 4.10253770648791e-06, + "loss": 0.2799, + "step": 27420 + }, + { + "epoch": 3.244617932339721, + "grad_norm": 3.0851919651031494, + "learning_rate": 4.101819487670577e-06, + "loss": 0.2864, + "step": 27430 + }, + { + "epoch": 3.245800804352969, + "grad_norm": 5.2798051834106445, + "learning_rate": 4.101101268853244e-06, + "loss": 0.3175, + "step": 27440 + }, + { + "epoch": 3.246983676366217, + "grad_norm": 2.9658639430999756, + "learning_rate": 4.1003830500359115e-06, + "loss": 0.2932, + "step": 27450 + }, + { + "epoch": 3.2481665483794653, + "grad_norm": 2.701160192489624, + "learning_rate": 4.099664831218578e-06, + "loss": 0.2835, + "step": 27460 + }, + { + "epoch": 3.2493494203927136, + "grad_norm": 2.572829246520996, + "learning_rate": 4.098946612401245e-06, + "loss": 0.2978, + "step": 27470 + }, + { + "epoch": 3.250532292405962, + "grad_norm": 3.43685245513916, + "learning_rate": 4.0982283935839115e-06, + "loss": 0.2973, + "step": 27480 + }, + { + "epoch": 3.2507688668086114, + "eval_accuracy": 0.8603985760388595, + "eval_loss": 0.3283090889453888, + "eval_runtime": 81.9879, + "eval_safe_aucpr": 0.9154902059248551, + "eval_safe_f1": 0.8448052668565299, + "eval_safe_fpr": 0.13661773937998822, + "eval_safe_precision": 0.8332786107766955, + "eval_safe_recall": 0.856655290102389, + "eval_samples_per_second": 733.206, + "eval_steps_per_second": 11.465, + "eval_unsafe_aucpr": 0.9529076569503385, + "eval_unsafe_f1": 0.8731444810593464, + "eval_unsafe_fpr": 0.1433447098976104, + "eval_unsafe_precision": 0.8831299880744886, + "eval_unsafe_recall": 0.8633822606200113, + "step": 27482 + }, + { + "epoch": 3.2517151644192097, + "grad_norm": 2.9928042888641357, + "learning_rate": 4.097510174766579e-06, + "loss": 0.3163, + "step": 27490 + }, + { + "epoch": 3.252898036432458, + "grad_norm": 3.460139274597168, + "learning_rate": 4.096791955949245e-06, + "loss": 0.3174, + "step": 27500 + }, + { + "epoch": 3.2540809084457063, + "grad_norm": 2.9319708347320557, + "learning_rate": 4.096073737131913e-06, + "loss": 0.3242, + "step": 27510 + }, + { + "epoch": 3.255263780458954, + "grad_norm": 2.2399494647979736, + "learning_rate": 4.09535551831458e-06, + "loss": 0.2882, + "step": 27520 + }, + { + "epoch": 3.2564466524722024, + "grad_norm": 3.403815746307373, + "learning_rate": 4.094637299497247e-06, + "loss": 0.2891, + "step": 27530 + }, + { + "epoch": 3.2576295244854507, + "grad_norm": 2.6851112842559814, + "learning_rate": 4.093919080679914e-06, + "loss": 0.2943, + "step": 27540 + }, + { + "epoch": 3.258812396498699, + "grad_norm": 2.8747034072875977, + "learning_rate": 4.093200861862581e-06, + "loss": 0.2988, + "step": 27550 + }, + { + "epoch": 3.259995268511947, + "grad_norm": 3.6352221965789795, + "learning_rate": 4.0924826430452485e-06, + "loss": 0.2984, + "step": 27560 + }, + { + "epoch": 3.261178140525195, + "grad_norm": 1.976767897605896, + "learning_rate": 4.091764424227915e-06, + "loss": 0.2717, + "step": 27570 + }, + { + "epoch": 3.2623610125384435, + "grad_norm": 2.54714035987854, + "learning_rate": 4.091046205410582e-06, + "loss": 0.2882, + "step": 27580 + }, + { + "epoch": 3.2635438845516918, + "grad_norm": 2.0810391902923584, + "learning_rate": 4.0903279865932484e-06, + "loss": 0.2399, + "step": 27590 + }, + { + "epoch": 3.2647267565649396, + "grad_norm": 2.7840323448181152, + "learning_rate": 4.089609767775916e-06, + "loss": 0.2715, + "step": 27600 + }, + { + "epoch": 3.265909628578188, + "grad_norm": 2.911987781524658, + "learning_rate": 4.088891548958583e-06, + "loss": 0.3072, + "step": 27610 + }, + { + "epoch": 3.267092500591436, + "grad_norm": 3.1399085521698, + "learning_rate": 4.08817333014125e-06, + "loss": 0.2501, + "step": 27620 + }, + { + "epoch": 3.268275372604684, + "grad_norm": 3.8143203258514404, + "learning_rate": 4.087455111323917e-06, + "loss": 0.2905, + "step": 27630 + }, + { + "epoch": 3.2694582446179323, + "grad_norm": 2.5066142082214355, + "learning_rate": 4.086736892506584e-06, + "loss": 0.2796, + "step": 27640 + }, + { + "epoch": 3.2706411166311806, + "grad_norm": 4.362666130065918, + "learning_rate": 4.086018673689251e-06, + "loss": 0.2595, + "step": 27650 + }, + { + "epoch": 3.271823988644429, + "grad_norm": 2.412616491317749, + "learning_rate": 4.085300454871918e-06, + "loss": 0.2998, + "step": 27660 + }, + { + "epoch": 3.2730068606576768, + "grad_norm": 2.690500020980835, + "learning_rate": 4.084582236054585e-06, + "loss": 0.2642, + "step": 27670 + }, + { + "epoch": 3.274189732670925, + "grad_norm": 3.0411665439605713, + "learning_rate": 4.083864017237252e-06, + "loss": 0.2956, + "step": 27680 + }, + { + "epoch": 3.2753726046841733, + "grad_norm": 4.12064266204834, + "learning_rate": 4.0831457984199185e-06, + "loss": 0.3109, + "step": 27690 + }, + { + "epoch": 3.276555476697421, + "grad_norm": 3.065070152282715, + "learning_rate": 4.0824275796025854e-06, + "loss": 0.3127, + "step": 27700 + }, + { + "epoch": 3.2777383487106695, + "grad_norm": 2.9523682594299316, + "learning_rate": 4.081709360785252e-06, + "loss": 0.273, + "step": 27710 + }, + { + "epoch": 3.278921220723918, + "grad_norm": 2.6885054111480713, + "learning_rate": 4.08099114196792e-06, + "loss": 0.3113, + "step": 27720 + }, + { + "epoch": 3.280104092737166, + "grad_norm": 5.8971638679504395, + "learning_rate": 4.080272923150586e-06, + "loss": 0.3232, + "step": 27730 + }, + { + "epoch": 3.281286964750414, + "grad_norm": 3.010758876800537, + "learning_rate": 4.079554704333254e-06, + "loss": 0.3236, + "step": 27740 + }, + { + "epoch": 3.282469836763662, + "grad_norm": 2.8600690364837646, + "learning_rate": 4.07883648551592e-06, + "loss": 0.2913, + "step": 27750 + }, + { + "epoch": 3.2836527087769105, + "grad_norm": 2.074984312057495, + "learning_rate": 4.078118266698588e-06, + "loss": 0.2787, + "step": 27760 + }, + { + "epoch": 3.2848355807901584, + "grad_norm": 4.146457672119141, + "learning_rate": 4.077400047881254e-06, + "loss": 0.2468, + "step": 27770 + }, + { + "epoch": 3.2860184528034067, + "grad_norm": 3.0289111137390137, + "learning_rate": 4.076681829063922e-06, + "loss": 0.3381, + "step": 27780 + }, + { + "epoch": 3.287201324816655, + "grad_norm": 2.3751113414764404, + "learning_rate": 4.075963610246589e-06, + "loss": 0.3018, + "step": 27790 + }, + { + "epoch": 3.2883841968299032, + "grad_norm": 2.964958667755127, + "learning_rate": 4.0752453914292555e-06, + "loss": 0.289, + "step": 27800 + }, + { + "epoch": 3.289567068843151, + "grad_norm": 2.8259074687957764, + "learning_rate": 4.074527172611923e-06, + "loss": 0.2733, + "step": 27810 + }, + { + "epoch": 3.2907499408563994, + "grad_norm": 3.40511417388916, + "learning_rate": 4.073808953794589e-06, + "loss": 0.2844, + "step": 27820 + }, + { + "epoch": 3.2919328128696477, + "grad_norm": 5.153759956359863, + "learning_rate": 4.073090734977257e-06, + "loss": 0.3, + "step": 27830 + }, + { + "epoch": 3.2931156848828955, + "grad_norm": 2.7530295848846436, + "learning_rate": 4.072372516159923e-06, + "loss": 0.3119, + "step": 27840 + }, + { + "epoch": 3.294298556896144, + "grad_norm": 3.2891340255737305, + "learning_rate": 4.071654297342591e-06, + "loss": 0.3307, + "step": 27850 + }, + { + "epoch": 3.295481428909392, + "grad_norm": 3.0643022060394287, + "learning_rate": 4.070936078525257e-06, + "loss": 0.2848, + "step": 27860 + }, + { + "epoch": 3.2966643009226404, + "grad_norm": 2.238170862197876, + "learning_rate": 4.070217859707925e-06, + "loss": 0.293, + "step": 27870 + }, + { + "epoch": 3.2978471729358882, + "grad_norm": 2.643157482147217, + "learning_rate": 4.069499640890592e-06, + "loss": 0.2731, + "step": 27880 + }, + { + "epoch": 3.2990300449491365, + "grad_norm": 3.4272263050079346, + "learning_rate": 4.068781422073259e-06, + "loss": 0.2782, + "step": 27890 + }, + { + "epoch": 3.300212916962385, + "grad_norm": 2.283717155456543, + "learning_rate": 4.068063203255926e-06, + "loss": 0.3043, + "step": 27900 + }, + { + "epoch": 3.3013957889756327, + "grad_norm": 3.3664350509643555, + "learning_rate": 4.0673449844385925e-06, + "loss": 0.2904, + "step": 27910 + }, + { + "epoch": 3.302578660988881, + "grad_norm": 2.9079642295837402, + "learning_rate": 4.0666267656212594e-06, + "loss": 0.2865, + "step": 27920 + }, + { + "epoch": 3.3037615330021293, + "grad_norm": 3.117150068283081, + "learning_rate": 4.065908546803926e-06, + "loss": 0.3322, + "step": 27930 + }, + { + "epoch": 3.3049444050153776, + "grad_norm": 2.965116262435913, + "learning_rate": 4.065190327986593e-06, + "loss": 0.2442, + "step": 27940 + }, + { + "epoch": 3.3061272770286254, + "grad_norm": 2.861928939819336, + "learning_rate": 4.06447210916926e-06, + "loss": 0.2987, + "step": 27950 + }, + { + "epoch": 3.3073101490418737, + "grad_norm": 2.784311056137085, + "learning_rate": 4.063753890351927e-06, + "loss": 0.3441, + "step": 27960 + }, + { + "epoch": 3.308493021055122, + "grad_norm": 3.677492380142212, + "learning_rate": 4.063035671534594e-06, + "loss": 0.3065, + "step": 27970 + }, + { + "epoch": 3.30967589306837, + "grad_norm": 2.0552244186401367, + "learning_rate": 4.062317452717261e-06, + "loss": 0.2875, + "step": 27980 + }, + { + "epoch": 3.310858765081618, + "grad_norm": 1.9184958934783936, + "learning_rate": 4.061599233899929e-06, + "loss": 0.2706, + "step": 27990 + }, + { + "epoch": 3.3120416370948664, + "grad_norm": 4.538043975830078, + "learning_rate": 4.060881015082595e-06, + "loss": 0.3078, + "step": 28000 + }, + { + "epoch": 3.3132245091081147, + "grad_norm": 3.2088682651519775, + "learning_rate": 4.060162796265263e-06, + "loss": 0.2863, + "step": 28010 + }, + { + "epoch": 3.3144073811213626, + "grad_norm": 2.603923797607422, + "learning_rate": 4.059444577447929e-06, + "loss": 0.2776, + "step": 28020 + }, + { + "epoch": 3.315590253134611, + "grad_norm": 4.3191423416137695, + "learning_rate": 4.0587263586305964e-06, + "loss": 0.2976, + "step": 28030 + }, + { + "epoch": 3.316773125147859, + "grad_norm": 3.0668866634368896, + "learning_rate": 4.0580081398132625e-06, + "loss": 0.2839, + "step": 28040 + }, + { + "epoch": 3.317955997161107, + "grad_norm": 4.304630756378174, + "learning_rate": 4.05728992099593e-06, + "loss": 0.2949, + "step": 28050 + }, + { + "epoch": 3.3191388691743553, + "grad_norm": 2.4871647357940674, + "learning_rate": 4.056571702178597e-06, + "loss": 0.2625, + "step": 28060 + }, + { + "epoch": 3.3203217411876036, + "grad_norm": 4.306042671203613, + "learning_rate": 4.055853483361264e-06, + "loss": 0.3007, + "step": 28070 + }, + { + "epoch": 3.321504613200852, + "grad_norm": 3.3634445667266846, + "learning_rate": 4.055135264543931e-06, + "loss": 0.2725, + "step": 28080 + }, + { + "epoch": 3.3226874852140997, + "grad_norm": 3.9011049270629883, + "learning_rate": 4.054417045726598e-06, + "loss": 0.2554, + "step": 28090 + }, + { + "epoch": 3.323870357227348, + "grad_norm": 3.2487316131591797, + "learning_rate": 4.053698826909266e-06, + "loss": 0.2659, + "step": 28100 + }, + { + "epoch": 3.3250532292405963, + "grad_norm": 3.0760748386383057, + "learning_rate": 4.052980608091932e-06, + "loss": 0.2734, + "step": 28110 + }, + { + "epoch": 3.326236101253844, + "grad_norm": 3.7253193855285645, + "learning_rate": 4.0522623892746e-06, + "loss": 0.3043, + "step": 28120 + }, + { + "epoch": 3.3274189732670925, + "grad_norm": 2.54416823387146, + "learning_rate": 4.051544170457266e-06, + "loss": 0.2756, + "step": 28130 + }, + { + "epoch": 3.3286018452803408, + "grad_norm": 3.150343894958496, + "learning_rate": 4.0508259516399334e-06, + "loss": 0.276, + "step": 28140 + }, + { + "epoch": 3.329784717293589, + "grad_norm": 3.0068886280059814, + "learning_rate": 4.0501077328226e-06, + "loss": 0.3174, + "step": 28150 + }, + { + "epoch": 3.330967589306837, + "grad_norm": 2.7225942611694336, + "learning_rate": 4.049389514005267e-06, + "loss": 0.2524, + "step": 28160 + }, + { + "epoch": 3.332150461320085, + "grad_norm": 3.4403152465820312, + "learning_rate": 4.048671295187934e-06, + "loss": 0.2884, + "step": 28170 + }, + { + "epoch": 3.3333333333333335, + "grad_norm": 4.012026786804199, + "learning_rate": 4.047953076370601e-06, + "loss": 0.2943, + "step": 28180 + }, + { + "epoch": 3.3345162053465813, + "grad_norm": 2.679258108139038, + "learning_rate": 4.047234857553268e-06, + "loss": 0.3146, + "step": 28190 + }, + { + "epoch": 3.3356990773598296, + "grad_norm": 2.6075501441955566, + "learning_rate": 4.046516638735935e-06, + "loss": 0.2733, + "step": 28200 + }, + { + "epoch": 3.336881949373078, + "grad_norm": 2.1568117141723633, + "learning_rate": 4.045798419918602e-06, + "loss": 0.3105, + "step": 28210 + }, + { + "epoch": 3.338064821386326, + "grad_norm": 3.158740997314453, + "learning_rate": 4.045080201101269e-06, + "loss": 0.2996, + "step": 28220 + }, + { + "epoch": 3.339247693399574, + "grad_norm": 3.0008444786071777, + "learning_rate": 4.044361982283936e-06, + "loss": 0.3032, + "step": 28230 + }, + { + "epoch": 3.3404305654128224, + "grad_norm": 2.9587295055389404, + "learning_rate": 4.043643763466603e-06, + "loss": 0.2997, + "step": 28240 + }, + { + "epoch": 3.3416134374260706, + "grad_norm": 2.94698429107666, + "learning_rate": 4.04292554464927e-06, + "loss": 0.3154, + "step": 28250 + }, + { + "epoch": 3.3427963094393185, + "grad_norm": 2.5821921825408936, + "learning_rate": 4.042207325831937e-06, + "loss": 0.2642, + "step": 28260 + }, + { + "epoch": 3.343979181452567, + "grad_norm": 2.43442702293396, + "learning_rate": 4.0414891070146034e-06, + "loss": 0.2981, + "step": 28270 + }, + { + "epoch": 3.345162053465815, + "grad_norm": 2.9026975631713867, + "learning_rate": 4.040770888197271e-06, + "loss": 0.3275, + "step": 28280 + }, + { + "epoch": 3.3463449254790634, + "grad_norm": 2.9597880840301514, + "learning_rate": 4.040052669379937e-06, + "loss": 0.3208, + "step": 28290 + }, + { + "epoch": 3.347527797492311, + "grad_norm": 2.3282508850097656, + "learning_rate": 4.039334450562605e-06, + "loss": 0.2796, + "step": 28300 + }, + { + "epoch": 3.3487106695055595, + "grad_norm": 3.9105443954467773, + "learning_rate": 4.038616231745271e-06, + "loss": 0.3055, + "step": 28310 + }, + { + "epoch": 3.349893541518808, + "grad_norm": 2.4692790508270264, + "learning_rate": 4.037898012927939e-06, + "loss": 0.2689, + "step": 28320 + }, + { + "epoch": 3.3510764135320557, + "grad_norm": 3.3734171390533447, + "learning_rate": 4.037179794110606e-06, + "loss": 0.2962, + "step": 28330 + }, + { + "epoch": 3.352259285545304, + "grad_norm": 3.8791041374206543, + "learning_rate": 4.036461575293273e-06, + "loss": 0.3038, + "step": 28340 + }, + { + "epoch": 3.3534421575585522, + "grad_norm": 3.6339375972747803, + "learning_rate": 4.03574335647594e-06, + "loss": 0.248, + "step": 28350 + }, + { + "epoch": 3.3546250295718005, + "grad_norm": 3.4359793663024902, + "learning_rate": 4.035025137658607e-06, + "loss": 0.2854, + "step": 28360 + }, + { + "epoch": 3.3558079015850484, + "grad_norm": 2.6512808799743652, + "learning_rate": 4.034306918841274e-06, + "loss": 0.31, + "step": 28370 + }, + { + "epoch": 3.3569907735982967, + "grad_norm": 3.134368896484375, + "learning_rate": 4.0335887000239404e-06, + "loss": 0.2799, + "step": 28380 + }, + { + "epoch": 3.358173645611545, + "grad_norm": 4.1644158363342285, + "learning_rate": 4.032870481206608e-06, + "loss": 0.309, + "step": 28390 + }, + { + "epoch": 3.359356517624793, + "grad_norm": 2.5575039386749268, + "learning_rate": 4.032152262389274e-06, + "loss": 0.2866, + "step": 28400 + }, + { + "epoch": 3.360539389638041, + "grad_norm": 3.720283269882202, + "learning_rate": 4.031434043571942e-06, + "loss": 0.2776, + "step": 28410 + }, + { + "epoch": 3.3617222616512894, + "grad_norm": 2.4713282585144043, + "learning_rate": 4.030715824754608e-06, + "loss": 0.3024, + "step": 28420 + }, + { + "epoch": 3.3629051336645377, + "grad_norm": 3.4714038372039795, + "learning_rate": 4.029997605937276e-06, + "loss": 0.3007, + "step": 28430 + }, + { + "epoch": 3.3640880056777855, + "grad_norm": 3.2137274742126465, + "learning_rate": 4.029279387119943e-06, + "loss": 0.2642, + "step": 28440 + }, + { + "epoch": 3.365270877691034, + "grad_norm": 2.4611263275146484, + "learning_rate": 4.02856116830261e-06, + "loss": 0.2609, + "step": 28450 + }, + { + "epoch": 3.366453749704282, + "grad_norm": 3.8252193927764893, + "learning_rate": 4.027842949485277e-06, + "loss": 0.2761, + "step": 28460 + }, + { + "epoch": 3.36763662171753, + "grad_norm": 2.8619930744171143, + "learning_rate": 4.027124730667944e-06, + "loss": 0.2902, + "step": 28470 + }, + { + "epoch": 3.3688194937307783, + "grad_norm": 3.314082622528076, + "learning_rate": 4.0264065118506105e-06, + "loss": 0.2961, + "step": 28480 + }, + { + "epoch": 3.3700023657440266, + "grad_norm": 3.308055877685547, + "learning_rate": 4.0256882930332774e-06, + "loss": 0.3228, + "step": 28490 + }, + { + "epoch": 3.371185237757275, + "grad_norm": 2.3181183338165283, + "learning_rate": 4.024970074215944e-06, + "loss": 0.3098, + "step": 28500 + }, + { + "epoch": 3.3723681097705227, + "grad_norm": 2.3771865367889404, + "learning_rate": 4.024251855398611e-06, + "loss": 0.3007, + "step": 28510 + }, + { + "epoch": 3.373550981783771, + "grad_norm": 2.6836299896240234, + "learning_rate": 4.023533636581278e-06, + "loss": 0.3389, + "step": 28520 + }, + { + "epoch": 3.3747338537970193, + "grad_norm": 3.9915354251861572, + "learning_rate": 4.022815417763946e-06, + "loss": 0.2986, + "step": 28530 + }, + { + "epoch": 3.375916725810267, + "grad_norm": 2.5483341217041016, + "learning_rate": 4.022097198946612e-06, + "loss": 0.3298, + "step": 28540 + }, + { + "epoch": 3.3770995978235154, + "grad_norm": 2.7282094955444336, + "learning_rate": 4.02137898012928e-06, + "loss": 0.2541, + "step": 28550 + }, + { + "epoch": 3.3782824698367637, + "grad_norm": 2.5950067043304443, + "learning_rate": 4.020660761311946e-06, + "loss": 0.2956, + "step": 28560 + }, + { + "epoch": 3.379465341850012, + "grad_norm": 3.0112249851226807, + "learning_rate": 4.019942542494614e-06, + "loss": 0.2583, + "step": 28570 + }, + { + "epoch": 3.38064821386326, + "grad_norm": 2.200986862182617, + "learning_rate": 4.019224323677281e-06, + "loss": 0.2461, + "step": 28580 + }, + { + "epoch": 3.381831085876508, + "grad_norm": 2.6726768016815186, + "learning_rate": 4.0185061048599475e-06, + "loss": 0.3154, + "step": 28590 + }, + { + "epoch": 3.3830139578897565, + "grad_norm": 3.4485435485839844, + "learning_rate": 4.0177878860426144e-06, + "loss": 0.3149, + "step": 28600 + }, + { + "epoch": 3.3841968299030043, + "grad_norm": 3.5943193435668945, + "learning_rate": 4.017069667225281e-06, + "loss": 0.3536, + "step": 28610 + }, + { + "epoch": 3.3853797019162526, + "grad_norm": 3.0588438510894775, + "learning_rate": 4.016351448407948e-06, + "loss": 0.3039, + "step": 28620 + }, + { + "epoch": 3.386562573929501, + "grad_norm": 3.3482131958007812, + "learning_rate": 4.015633229590615e-06, + "loss": 0.3269, + "step": 28630 + }, + { + "epoch": 3.387745445942749, + "grad_norm": 2.8402023315429688, + "learning_rate": 4.014915010773283e-06, + "loss": 0.3037, + "step": 28640 + }, + { + "epoch": 3.388928317955997, + "grad_norm": 2.354872465133667, + "learning_rate": 4.014196791955949e-06, + "loss": 0.2845, + "step": 28650 + }, + { + "epoch": 3.3901111899692453, + "grad_norm": 2.502476453781128, + "learning_rate": 4.013478573138617e-06, + "loss": 0.2786, + "step": 28660 + }, + { + "epoch": 3.3912940619824936, + "grad_norm": 2.517012119293213, + "learning_rate": 4.012760354321283e-06, + "loss": 0.2767, + "step": 28670 + }, + { + "epoch": 3.3924769339957415, + "grad_norm": 4.204011917114258, + "learning_rate": 4.012042135503951e-06, + "loss": 0.2871, + "step": 28680 + }, + { + "epoch": 3.3936598060089898, + "grad_norm": 2.779322385787964, + "learning_rate": 4.011323916686617e-06, + "loss": 0.243, + "step": 28690 + }, + { + "epoch": 3.394842678022238, + "grad_norm": 4.354095458984375, + "learning_rate": 4.0106056978692845e-06, + "loss": 0.3019, + "step": 28700 + }, + { + "epoch": 3.3960255500354863, + "grad_norm": 2.8547163009643555, + "learning_rate": 4.0098874790519514e-06, + "loss": 0.2686, + "step": 28710 + }, + { + "epoch": 3.397208422048734, + "grad_norm": 2.772738456726074, + "learning_rate": 4.009169260234618e-06, + "loss": 0.2578, + "step": 28720 + }, + { + "epoch": 3.3983912940619825, + "grad_norm": 2.603830337524414, + "learning_rate": 4.008451041417285e-06, + "loss": 0.2592, + "step": 28730 + }, + { + "epoch": 3.3995741660752308, + "grad_norm": 2.9823734760284424, + "learning_rate": 4.007732822599952e-06, + "loss": 0.2849, + "step": 28740 + }, + { + "epoch": 3.4007570380884786, + "grad_norm": 3.963719129562378, + "learning_rate": 4.007014603782619e-06, + "loss": 0.3095, + "step": 28750 + }, + { + "epoch": 3.401939910101727, + "grad_norm": 4.043334484100342, + "learning_rate": 4.006296384965286e-06, + "loss": 0.2659, + "step": 28760 + }, + { + "epoch": 3.403122782114975, + "grad_norm": 2.87335467338562, + "learning_rate": 4.005578166147953e-06, + "loss": 0.2677, + "step": 28770 + }, + { + "epoch": 3.4043056541282235, + "grad_norm": 2.177003860473633, + "learning_rate": 4.00485994733062e-06, + "loss": 0.2967, + "step": 28780 + }, + { + "epoch": 3.4054885261414714, + "grad_norm": 5.445313453674316, + "learning_rate": 4.004141728513287e-06, + "loss": 0.2742, + "step": 28790 + }, + { + "epoch": 3.4066713981547196, + "grad_norm": 2.3107638359069824, + "learning_rate": 4.003423509695955e-06, + "loss": 0.3064, + "step": 28800 + }, + { + "epoch": 3.407854270167968, + "grad_norm": 2.7333712577819824, + "learning_rate": 4.002705290878621e-06, + "loss": 0.3044, + "step": 28810 + }, + { + "epoch": 3.409037142181216, + "grad_norm": 1.6254793405532837, + "learning_rate": 4.001987072061288e-06, + "loss": 0.2496, + "step": 28820 + }, + { + "epoch": 3.410220014194464, + "grad_norm": 4.3058671951293945, + "learning_rate": 4.001268853243955e-06, + "loss": 0.291, + "step": 28830 + }, + { + "epoch": 3.4114028862077124, + "grad_norm": 3.107118606567383, + "learning_rate": 4.000550634426622e-06, + "loss": 0.3196, + "step": 28840 + }, + { + "epoch": 3.4125857582209607, + "grad_norm": 3.680126190185547, + "learning_rate": 3.999832415609289e-06, + "loss": 0.275, + "step": 28850 + }, + { + "epoch": 3.4137686302342085, + "grad_norm": 2.1049747467041016, + "learning_rate": 3.999114196791956e-06, + "loss": 0.3002, + "step": 28860 + }, + { + "epoch": 3.414951502247457, + "grad_norm": 4.366775035858154, + "learning_rate": 3.998395977974623e-06, + "loss": 0.255, + "step": 28870 + }, + { + "epoch": 3.416134374260705, + "grad_norm": 2.862595796585083, + "learning_rate": 3.99767775915729e-06, + "loss": 0.2793, + "step": 28880 + }, + { + "epoch": 3.417317246273953, + "grad_norm": 2.8145334720611572, + "learning_rate": 3.996959540339957e-06, + "loss": 0.2671, + "step": 28890 + }, + { + "epoch": 3.4185001182872012, + "grad_norm": 4.602574825286865, + "learning_rate": 3.996241321522624e-06, + "loss": 0.2923, + "step": 28900 + }, + { + "epoch": 3.4196829903004495, + "grad_norm": 2.1978883743286133, + "learning_rate": 3.9955231027052916e-06, + "loss": 0.2923, + "step": 28910 + }, + { + "epoch": 3.420865862313698, + "grad_norm": 4.055898666381836, + "learning_rate": 3.994804883887958e-06, + "loss": 0.2681, + "step": 28920 + }, + { + "epoch": 3.4220487343269457, + "grad_norm": 4.268734455108643, + "learning_rate": 3.994086665070625e-06, + "loss": 0.29, + "step": 28930 + }, + { + "epoch": 3.423231606340194, + "grad_norm": 3.2915849685668945, + "learning_rate": 3.9933684462532915e-06, + "loss": 0.3109, + "step": 28940 + }, + { + "epoch": 3.4244144783534423, + "grad_norm": 2.6965219974517822, + "learning_rate": 3.992650227435959e-06, + "loss": 0.2631, + "step": 28950 + }, + { + "epoch": 3.42559735036669, + "grad_norm": 2.662470579147339, + "learning_rate": 3.991932008618625e-06, + "loss": 0.2842, + "step": 28960 + }, + { + "epoch": 3.4267802223799384, + "grad_norm": 3.452353000640869, + "learning_rate": 3.991213789801293e-06, + "loss": 0.2976, + "step": 28970 + }, + { + "epoch": 3.4279630943931867, + "grad_norm": 2.6745400428771973, + "learning_rate": 3.99049557098396e-06, + "loss": 0.2956, + "step": 28980 + }, + { + "epoch": 3.429145966406435, + "grad_norm": 3.3390755653381348, + "learning_rate": 3.989777352166627e-06, + "loss": 0.2756, + "step": 28990 + }, + { + "epoch": 3.430328838419683, + "grad_norm": 3.5056352615356445, + "learning_rate": 3.989059133349294e-06, + "loss": 0.2445, + "step": 29000 + }, + { + "epoch": 3.431511710432931, + "grad_norm": 2.393807888031006, + "learning_rate": 3.988340914531961e-06, + "loss": 0.2724, + "step": 29010 + }, + { + "epoch": 3.4326945824461794, + "grad_norm": 3.7909843921661377, + "learning_rate": 3.987622695714628e-06, + "loss": 0.2675, + "step": 29020 + }, + { + "epoch": 3.4338774544594273, + "grad_norm": 2.4259259700775146, + "learning_rate": 3.986904476897295e-06, + "loss": 0.2832, + "step": 29030 + }, + { + "epoch": 3.4350603264726756, + "grad_norm": 3.82837176322937, + "learning_rate": 3.9861862580799616e-06, + "loss": 0.2667, + "step": 29040 + }, + { + "epoch": 3.436243198485924, + "grad_norm": 4.073571681976318, + "learning_rate": 3.9854680392626285e-06, + "loss": 0.3378, + "step": 29050 + }, + { + "epoch": 3.437426070499172, + "grad_norm": 2.9060354232788086, + "learning_rate": 3.984749820445295e-06, + "loss": 0.2834, + "step": 29060 + }, + { + "epoch": 3.43860894251242, + "grad_norm": 1.9702935218811035, + "learning_rate": 3.984031601627962e-06, + "loss": 0.2717, + "step": 29070 + }, + { + "epoch": 3.4397918145256683, + "grad_norm": 3.5200788974761963, + "learning_rate": 3.98331338281063e-06, + "loss": 0.2951, + "step": 29080 + }, + { + "epoch": 3.4409746865389166, + "grad_norm": 3.1678929328918457, + "learning_rate": 3.982595163993297e-06, + "loss": 0.3277, + "step": 29090 + }, + { + "epoch": 3.4421575585521644, + "grad_norm": 3.3186113834381104, + "learning_rate": 3.981876945175964e-06, + "loss": 0.2906, + "step": 29100 + }, + { + "epoch": 3.4433404305654127, + "grad_norm": 3.593663215637207, + "learning_rate": 3.981158726358631e-06, + "loss": 0.3329, + "step": 29110 + }, + { + "epoch": 3.444523302578661, + "grad_norm": 2.9948604106903076, + "learning_rate": 3.980440507541298e-06, + "loss": 0.2916, + "step": 29120 + }, + { + "epoch": 3.4457061745919093, + "grad_norm": 2.2265982627868652, + "learning_rate": 3.979722288723965e-06, + "loss": 0.2569, + "step": 29130 + }, + { + "epoch": 3.446889046605157, + "grad_norm": 3.2792186737060547, + "learning_rate": 3.979004069906632e-06, + "loss": 0.2539, + "step": 29140 + }, + { + "epoch": 3.4480719186184055, + "grad_norm": 1.9254900217056274, + "learning_rate": 3.9782858510892986e-06, + "loss": 0.278, + "step": 29150 + }, + { + "epoch": 3.4492547906316537, + "grad_norm": 3.3341944217681885, + "learning_rate": 3.9775676322719655e-06, + "loss": 0.2713, + "step": 29160 + }, + { + "epoch": 3.4504376626449016, + "grad_norm": 2.7676820755004883, + "learning_rate": 3.976849413454632e-06, + "loss": 0.2993, + "step": 29170 + }, + { + "epoch": 3.45162053465815, + "grad_norm": 2.2772769927978516, + "learning_rate": 3.9761311946373e-06, + "loss": 0.2982, + "step": 29180 + }, + { + "epoch": 3.452803406671398, + "grad_norm": 2.897183895111084, + "learning_rate": 3.975412975819966e-06, + "loss": 0.3238, + "step": 29190 + }, + { + "epoch": 3.4539862786846465, + "grad_norm": 4.300553798675537, + "learning_rate": 3.974694757002634e-06, + "loss": 0.3149, + "step": 29200 + }, + { + "epoch": 3.4551691506978943, + "grad_norm": 2.5118746757507324, + "learning_rate": 3.9739765381853e-06, + "loss": 0.2477, + "step": 29210 + }, + { + "epoch": 3.4563520227111426, + "grad_norm": 3.409400463104248, + "learning_rate": 3.973258319367968e-06, + "loss": 0.304, + "step": 29220 + }, + { + "epoch": 3.457534894724391, + "grad_norm": 3.270049810409546, + "learning_rate": 3.972540100550634e-06, + "loss": 0.2817, + "step": 29230 + }, + { + "epoch": 3.4587177667376388, + "grad_norm": 2.6571133136749268, + "learning_rate": 3.971821881733302e-06, + "loss": 0.299, + "step": 29240 + }, + { + "epoch": 3.459900638750887, + "grad_norm": 2.746952533721924, + "learning_rate": 3.971103662915969e-06, + "loss": 0.2604, + "step": 29250 + }, + { + "epoch": 3.4610835107641353, + "grad_norm": 3.310290575027466, + "learning_rate": 3.9703854440986356e-06, + "loss": 0.2783, + "step": 29260 + }, + { + "epoch": 3.4622663827773836, + "grad_norm": 3.8183412551879883, + "learning_rate": 3.9696672252813025e-06, + "loss": 0.2463, + "step": 29270 + }, + { + "epoch": 3.4634492547906315, + "grad_norm": 2.7475006580352783, + "learning_rate": 3.968949006463969e-06, + "loss": 0.309, + "step": 29280 + }, + { + "epoch": 3.4646321268038798, + "grad_norm": 2.4192402362823486, + "learning_rate": 3.968230787646636e-06, + "loss": 0.2886, + "step": 29290 + }, + { + "epoch": 3.465814998817128, + "grad_norm": 4.498946189880371, + "learning_rate": 3.967512568829303e-06, + "loss": 0.2904, + "step": 29300 + }, + { + "epoch": 3.466997870830376, + "grad_norm": 3.021507978439331, + "learning_rate": 3.96679435001197e-06, + "loss": 0.2949, + "step": 29310 + }, + { + "epoch": 3.468180742843624, + "grad_norm": 2.8634607791900635, + "learning_rate": 3.966076131194637e-06, + "loss": 0.2391, + "step": 29320 + }, + { + "epoch": 3.4693636148568725, + "grad_norm": 3.459980010986328, + "learning_rate": 3.965357912377305e-06, + "loss": 0.2775, + "step": 29330 + }, + { + "epoch": 3.470546486870121, + "grad_norm": 2.320286750793457, + "learning_rate": 3.964639693559971e-06, + "loss": 0.2625, + "step": 29340 + }, + { + "epoch": 3.4717293588833686, + "grad_norm": 3.8461697101593018, + "learning_rate": 3.963921474742639e-06, + "loss": 0.3486, + "step": 29350 + }, + { + "epoch": 3.472912230896617, + "grad_norm": 2.476952075958252, + "learning_rate": 3.963203255925306e-06, + "loss": 0.2703, + "step": 29360 + }, + { + "epoch": 3.4740951029098652, + "grad_norm": 3.1494569778442383, + "learning_rate": 3.9624850371079726e-06, + "loss": 0.2762, + "step": 29370 + }, + { + "epoch": 3.475277974923113, + "grad_norm": 2.9971723556518555, + "learning_rate": 3.9617668182906395e-06, + "loss": 0.2826, + "step": 29380 + }, + { + "epoch": 3.4764608469363614, + "grad_norm": 3.3210978507995605, + "learning_rate": 3.961048599473306e-06, + "loss": 0.3112, + "step": 29390 + }, + { + "epoch": 3.4776437189496097, + "grad_norm": 3.2452621459960938, + "learning_rate": 3.960330380655973e-06, + "loss": 0.294, + "step": 29400 + }, + { + "epoch": 3.478826590962858, + "grad_norm": 2.9135382175445557, + "learning_rate": 3.95961216183864e-06, + "loss": 0.2956, + "step": 29410 + }, + { + "epoch": 3.480009462976106, + "grad_norm": 3.071747064590454, + "learning_rate": 3.958893943021307e-06, + "loss": 0.3052, + "step": 29420 + }, + { + "epoch": 3.481192334989354, + "grad_norm": 3.312474012374878, + "learning_rate": 3.958175724203974e-06, + "loss": 0.2897, + "step": 29430 + }, + { + "epoch": 3.4823752070026024, + "grad_norm": 2.7206039428710938, + "learning_rate": 3.957457505386641e-06, + "loss": 0.2913, + "step": 29440 + }, + { + "epoch": 3.4835580790158502, + "grad_norm": 2.353300094604492, + "learning_rate": 3.956739286569309e-06, + "loss": 0.2414, + "step": 29450 + }, + { + "epoch": 3.4847409510290985, + "grad_norm": 2.789315938949585, + "learning_rate": 3.956021067751975e-06, + "loss": 0.2547, + "step": 29460 + }, + { + "epoch": 3.485923823042347, + "grad_norm": 2.9860832691192627, + "learning_rate": 3.955302848934643e-06, + "loss": 0.293, + "step": 29470 + }, + { + "epoch": 3.487106695055595, + "grad_norm": 2.748837947845459, + "learning_rate": 3.954584630117309e-06, + "loss": 0.3159, + "step": 29480 + }, + { + "epoch": 3.488289567068843, + "grad_norm": 2.4259722232818604, + "learning_rate": 3.9538664112999765e-06, + "loss": 0.266, + "step": 29490 + }, + { + "epoch": 3.4894724390820913, + "grad_norm": 3.2140469551086426, + "learning_rate": 3.9531481924826426e-06, + "loss": 0.3055, + "step": 29500 + }, + { + "epoch": 3.4906553110953396, + "grad_norm": 2.755739212036133, + "learning_rate": 3.95242997366531e-06, + "loss": 0.29, + "step": 29510 + }, + { + "epoch": 3.4918381831085874, + "grad_norm": 2.7468137741088867, + "learning_rate": 3.951711754847977e-06, + "loss": 0.301, + "step": 29520 + }, + { + "epoch": 3.4930210551218357, + "grad_norm": 3.684769630432129, + "learning_rate": 3.950993536030644e-06, + "loss": 0.2755, + "step": 29530 + }, + { + "epoch": 3.494203927135084, + "grad_norm": 2.875627279281616, + "learning_rate": 3.950275317213311e-06, + "loss": 0.3011, + "step": 29540 + }, + { + "epoch": 3.4953867991483323, + "grad_norm": 3.1565229892730713, + "learning_rate": 3.949557098395978e-06, + "loss": 0.3004, + "step": 29550 + }, + { + "epoch": 3.49656967116158, + "grad_norm": 4.146491527557373, + "learning_rate": 3.948838879578645e-06, + "loss": 0.291, + "step": 29560 + }, + { + "epoch": 3.4977525431748284, + "grad_norm": 2.752413749694824, + "learning_rate": 3.948120660761312e-06, + "loss": 0.3018, + "step": 29570 + }, + { + "epoch": 3.4989354151880767, + "grad_norm": 3.5726985931396484, + "learning_rate": 3.947402441943979e-06, + "loss": 0.3009, + "step": 29580 + }, + { + "epoch": 3.5001182872013246, + "grad_norm": 2.2508862018585205, + "learning_rate": 3.946684223126646e-06, + "loss": 0.2741, + "step": 29590 + }, + { + "epoch": 3.5008280104092737, + "eval_accuracy": 0.8600159696576505, + "eval_loss": 0.32884901762008667, + "eval_runtime": 77.7977, + "eval_safe_aucpr": 0.915389373270268, + "eval_safe_f1": 0.8444748368972591, + "eval_safe_fpr": 0.13745478461032512, + "eval_safe_precision": 0.8324588252441335, + "eval_safe_recall": 0.8568428158871845, + "eval_samples_per_second": 772.696, + "eval_steps_per_second": 12.083, + "eval_unsafe_aucpr": 0.9530512352714813, + "eval_unsafe_f1": 0.8727333222425553, + "eval_unsafe_fpr": 0.14315718411281497, + "eval_unsafe_precision": 0.8831649831649832, + "eval_unsafe_recall": 0.8625452153896744, + "step": 29596 + }, + { + "epoch": 3.501301159214573, + "grad_norm": 3.029310941696167, + "learning_rate": 3.9459660043093135e-06, + "loss": 0.2791, + "step": 29600 + }, + { + "epoch": 3.502484031227821, + "grad_norm": 2.4385011196136475, + "learning_rate": 3.9452477854919796e-06, + "loss": 0.2656, + "step": 29610 + }, + { + "epoch": 3.5036669032410694, + "grad_norm": 3.1024386882781982, + "learning_rate": 3.944529566674647e-06, + "loss": 0.272, + "step": 29620 + }, + { + "epoch": 3.5048497752543177, + "grad_norm": 2.7745182514190674, + "learning_rate": 3.943811347857314e-06, + "loss": 0.2835, + "step": 29630 + }, + { + "epoch": 3.5060326472675656, + "grad_norm": 3.7471024990081787, + "learning_rate": 3.943093129039981e-06, + "loss": 0.3007, + "step": 29640 + }, + { + "epoch": 3.507215519280814, + "grad_norm": 3.8284647464752197, + "learning_rate": 3.942374910222648e-06, + "loss": 0.3026, + "step": 29650 + }, + { + "epoch": 3.5083983912940617, + "grad_norm": 2.3284730911254883, + "learning_rate": 3.941656691405315e-06, + "loss": 0.2985, + "step": 29660 + }, + { + "epoch": 3.50958126330731, + "grad_norm": 3.232443332672119, + "learning_rate": 3.940938472587982e-06, + "loss": 0.3027, + "step": 29670 + }, + { + "epoch": 3.5107641353205583, + "grad_norm": 2.2535243034362793, + "learning_rate": 3.940220253770649e-06, + "loss": 0.3141, + "step": 29680 + }, + { + "epoch": 3.5119470073338066, + "grad_norm": 2.911020278930664, + "learning_rate": 3.939502034953316e-06, + "loss": 0.3234, + "step": 29690 + }, + { + "epoch": 3.513129879347055, + "grad_norm": 2.8949711322784424, + "learning_rate": 3.938783816135983e-06, + "loss": 0.3127, + "step": 29700 + }, + { + "epoch": 3.5143127513603027, + "grad_norm": 3.0444653034210205, + "learning_rate": 3.93806559731865e-06, + "loss": 0.2895, + "step": 29710 + }, + { + "epoch": 3.515495623373551, + "grad_norm": 3.122600555419922, + "learning_rate": 3.937347378501317e-06, + "loss": 0.3108, + "step": 29720 + }, + { + "epoch": 3.516678495386799, + "grad_norm": 2.93048095703125, + "learning_rate": 3.9366291596839835e-06, + "loss": 0.2738, + "step": 29730 + }, + { + "epoch": 3.517861367400047, + "grad_norm": 3.568284034729004, + "learning_rate": 3.935910940866651e-06, + "loss": 0.2656, + "step": 29740 + }, + { + "epoch": 3.5190442394132955, + "grad_norm": 3.0430760383605957, + "learning_rate": 3.935192722049317e-06, + "loss": 0.3161, + "step": 29750 + }, + { + "epoch": 3.5202271114265438, + "grad_norm": 3.3558688163757324, + "learning_rate": 3.934474503231985e-06, + "loss": 0.2549, + "step": 29760 + }, + { + "epoch": 3.521409983439792, + "grad_norm": 4.9178972244262695, + "learning_rate": 3.933756284414651e-06, + "loss": 0.2799, + "step": 29770 + }, + { + "epoch": 3.52259285545304, + "grad_norm": 5.353640556335449, + "learning_rate": 3.933038065597319e-06, + "loss": 0.2798, + "step": 29780 + }, + { + "epoch": 3.523775727466288, + "grad_norm": 2.5971882343292236, + "learning_rate": 3.932319846779986e-06, + "loss": 0.2976, + "step": 29790 + }, + { + "epoch": 3.524958599479536, + "grad_norm": 4.564641952514648, + "learning_rate": 3.931601627962653e-06, + "loss": 0.3102, + "step": 29800 + }, + { + "epoch": 3.5261414714927843, + "grad_norm": 3.1244876384735107, + "learning_rate": 3.93088340914532e-06, + "loss": 0.3059, + "step": 29810 + }, + { + "epoch": 3.5273243435060326, + "grad_norm": 2.325773000717163, + "learning_rate": 3.930165190327987e-06, + "loss": 0.2833, + "step": 29820 + }, + { + "epoch": 3.528507215519281, + "grad_norm": 2.4007983207702637, + "learning_rate": 3.9294469715106536e-06, + "loss": 0.2945, + "step": 29830 + }, + { + "epoch": 3.529690087532529, + "grad_norm": 2.8207406997680664, + "learning_rate": 3.9287287526933205e-06, + "loss": 0.314, + "step": 29840 + }, + { + "epoch": 3.530872959545777, + "grad_norm": 3.3562471866607666, + "learning_rate": 3.928010533875988e-06, + "loss": 0.3105, + "step": 29850 + }, + { + "epoch": 3.5320558315590254, + "grad_norm": 1.9103692770004272, + "learning_rate": 3.927292315058654e-06, + "loss": 0.2849, + "step": 29860 + }, + { + "epoch": 3.533238703572273, + "grad_norm": 2.370021104812622, + "learning_rate": 3.926574096241322e-06, + "loss": 0.2933, + "step": 29870 + }, + { + "epoch": 3.5344215755855215, + "grad_norm": 3.22601318359375, + "learning_rate": 3.925855877423988e-06, + "loss": 0.2486, + "step": 29880 + }, + { + "epoch": 3.53560444759877, + "grad_norm": 2.821993589401245, + "learning_rate": 3.925137658606656e-06, + "loss": 0.2691, + "step": 29890 + }, + { + "epoch": 3.536787319612018, + "grad_norm": 3.3346800804138184, + "learning_rate": 3.924419439789323e-06, + "loss": 0.2883, + "step": 29900 + }, + { + "epoch": 3.5379701916252664, + "grad_norm": 3.168386220932007, + "learning_rate": 3.92370122097199e-06, + "loss": 0.3288, + "step": 29910 + }, + { + "epoch": 3.5391530636385142, + "grad_norm": 2.7423057556152344, + "learning_rate": 3.922983002154657e-06, + "loss": 0.2781, + "step": 29920 + }, + { + "epoch": 3.5403359356517625, + "grad_norm": 2.565031051635742, + "learning_rate": 3.922264783337324e-06, + "loss": 0.2943, + "step": 29930 + }, + { + "epoch": 3.5415188076650104, + "grad_norm": 4.034544944763184, + "learning_rate": 3.9215465645199906e-06, + "loss": 0.2915, + "step": 29940 + }, + { + "epoch": 3.5427016796782587, + "grad_norm": 2.539607524871826, + "learning_rate": 3.9208283457026575e-06, + "loss": 0.2865, + "step": 29950 + }, + { + "epoch": 3.543884551691507, + "grad_norm": 3.653226375579834, + "learning_rate": 3.920110126885324e-06, + "loss": 0.262, + "step": 29960 + }, + { + "epoch": 3.5450674237047552, + "grad_norm": 2.855692148208618, + "learning_rate": 3.919391908067991e-06, + "loss": 0.2774, + "step": 29970 + }, + { + "epoch": 3.5462502957180035, + "grad_norm": 2.8344321250915527, + "learning_rate": 3.918673689250658e-06, + "loss": 0.2708, + "step": 29980 + }, + { + "epoch": 3.5474331677312514, + "grad_norm": 3.342066526412964, + "learning_rate": 3.917955470433325e-06, + "loss": 0.3056, + "step": 29990 + }, + { + "epoch": 3.5486160397444997, + "grad_norm": 2.6532962322235107, + "learning_rate": 3.917237251615992e-06, + "loss": 0.2947, + "step": 30000 + }, + { + "epoch": 3.549798911757748, + "grad_norm": 2.5257811546325684, + "learning_rate": 3.91651903279866e-06, + "loss": 0.2967, + "step": 30010 + }, + { + "epoch": 3.550981783770996, + "grad_norm": 2.410673141479492, + "learning_rate": 3.915800813981326e-06, + "loss": 0.272, + "step": 30020 + }, + { + "epoch": 3.552164655784244, + "grad_norm": 1.8771846294403076, + "learning_rate": 3.915082595163994e-06, + "loss": 0.2444, + "step": 30030 + }, + { + "epoch": 3.5533475277974924, + "grad_norm": 3.0626044273376465, + "learning_rate": 3.91436437634666e-06, + "loss": 0.3246, + "step": 30040 + }, + { + "epoch": 3.5545303998107407, + "grad_norm": 4.596074104309082, + "learning_rate": 3.9136461575293275e-06, + "loss": 0.2811, + "step": 30050 + }, + { + "epoch": 3.5557132718239886, + "grad_norm": 3.1455276012420654, + "learning_rate": 3.9129279387119945e-06, + "loss": 0.3007, + "step": 30060 + }, + { + "epoch": 3.556896143837237, + "grad_norm": 2.6878602504730225, + "learning_rate": 3.912209719894661e-06, + "loss": 0.2825, + "step": 30070 + }, + { + "epoch": 3.558079015850485, + "grad_norm": 4.2512664794921875, + "learning_rate": 3.911491501077328e-06, + "loss": 0.2945, + "step": 30080 + }, + { + "epoch": 3.559261887863733, + "grad_norm": 3.9039673805236816, + "learning_rate": 3.910773282259995e-06, + "loss": 0.3123, + "step": 30090 + }, + { + "epoch": 3.5604447598769813, + "grad_norm": 2.679293394088745, + "learning_rate": 3.910055063442663e-06, + "loss": 0.2603, + "step": 30100 + }, + { + "epoch": 3.5616276318902296, + "grad_norm": 2.824568510055542, + "learning_rate": 3.909336844625329e-06, + "loss": 0.3108, + "step": 30110 + }, + { + "epoch": 3.562810503903478, + "grad_norm": 2.5078423023223877, + "learning_rate": 3.908618625807997e-06, + "loss": 0.28, + "step": 30120 + }, + { + "epoch": 3.5639933759167257, + "grad_norm": 3.575838327407837, + "learning_rate": 3.907900406990663e-06, + "loss": 0.3054, + "step": 30130 + }, + { + "epoch": 3.565176247929974, + "grad_norm": 2.186633825302124, + "learning_rate": 3.907182188173331e-06, + "loss": 0.3101, + "step": 30140 + }, + { + "epoch": 3.5663591199432223, + "grad_norm": 3.1343424320220947, + "learning_rate": 3.906463969355997e-06, + "loss": 0.2979, + "step": 30150 + }, + { + "epoch": 3.56754199195647, + "grad_norm": 2.1943230628967285, + "learning_rate": 3.9057457505386645e-06, + "loss": 0.2691, + "step": 30160 + }, + { + "epoch": 3.5687248639697184, + "grad_norm": 2.861821413040161, + "learning_rate": 3.9050275317213315e-06, + "loss": 0.3343, + "step": 30170 + }, + { + "epoch": 3.5699077359829667, + "grad_norm": 3.097752809524536, + "learning_rate": 3.904309312903998e-06, + "loss": 0.2983, + "step": 30180 + }, + { + "epoch": 3.571090607996215, + "grad_norm": 2.444945812225342, + "learning_rate": 3.903591094086665e-06, + "loss": 0.287, + "step": 30190 + }, + { + "epoch": 3.572273480009463, + "grad_norm": 2.8476860523223877, + "learning_rate": 3.902872875269332e-06, + "loss": 0.329, + "step": 30200 + }, + { + "epoch": 3.573456352022711, + "grad_norm": 2.4883365631103516, + "learning_rate": 3.902154656451999e-06, + "loss": 0.2899, + "step": 30210 + }, + { + "epoch": 3.5746392240359595, + "grad_norm": 1.879276990890503, + "learning_rate": 3.901436437634666e-06, + "loss": 0.2965, + "step": 30220 + }, + { + "epoch": 3.5758220960492073, + "grad_norm": 2.9747819900512695, + "learning_rate": 3.900718218817333e-06, + "loss": 0.2782, + "step": 30230 + }, + { + "epoch": 3.5770049680624556, + "grad_norm": 3.0752792358398438, + "learning_rate": 3.9e-06, + "loss": 0.2892, + "step": 30240 + }, + { + "epoch": 3.578187840075704, + "grad_norm": 4.906447410583496, + "learning_rate": 3.899281781182667e-06, + "loss": 0.2656, + "step": 30250 + }, + { + "epoch": 3.579370712088952, + "grad_norm": 2.7060699462890625, + "learning_rate": 3.898563562365334e-06, + "loss": 0.2929, + "step": 30260 + }, + { + "epoch": 3.5805535841022, + "grad_norm": 3.572575330734253, + "learning_rate": 3.897845343548001e-06, + "loss": 0.2854, + "step": 30270 + }, + { + "epoch": 3.5817364561154483, + "grad_norm": 2.5565311908721924, + "learning_rate": 3.8971271247306685e-06, + "loss": 0.3296, + "step": 30280 + }, + { + "epoch": 3.5829193281286966, + "grad_norm": 4.189861297607422, + "learning_rate": 3.8964089059133345e-06, + "loss": 0.2903, + "step": 30290 + }, + { + "epoch": 3.5841022001419445, + "grad_norm": 2.8932878971099854, + "learning_rate": 3.895690687096002e-06, + "loss": 0.2952, + "step": 30300 + }, + { + "epoch": 3.5852850721551928, + "grad_norm": 2.233625888824463, + "learning_rate": 3.894972468278668e-06, + "loss": 0.2778, + "step": 30310 + }, + { + "epoch": 3.586467944168441, + "grad_norm": 3.5226712226867676, + "learning_rate": 3.894254249461336e-06, + "loss": 0.2912, + "step": 30320 + }, + { + "epoch": 3.5876508161816894, + "grad_norm": 2.51426362991333, + "learning_rate": 3.893536030644002e-06, + "loss": 0.2679, + "step": 30330 + }, + { + "epoch": 3.588833688194937, + "grad_norm": 2.8012824058532715, + "learning_rate": 3.89281781182667e-06, + "loss": 0.2974, + "step": 30340 + }, + { + "epoch": 3.5900165602081855, + "grad_norm": 3.285240411758423, + "learning_rate": 3.892099593009337e-06, + "loss": 0.2842, + "step": 30350 + }, + { + "epoch": 3.591199432221434, + "grad_norm": 2.258108377456665, + "learning_rate": 3.891381374192004e-06, + "loss": 0.3086, + "step": 30360 + }, + { + "epoch": 3.5923823042346816, + "grad_norm": 2.47900652885437, + "learning_rate": 3.890663155374672e-06, + "loss": 0.3126, + "step": 30370 + }, + { + "epoch": 3.59356517624793, + "grad_norm": 3.498553514480591, + "learning_rate": 3.889944936557338e-06, + "loss": 0.2909, + "step": 30380 + }, + { + "epoch": 3.594748048261178, + "grad_norm": 3.9336187839508057, + "learning_rate": 3.8892267177400055e-06, + "loss": 0.3375, + "step": 30390 + }, + { + "epoch": 3.5959309202744265, + "grad_norm": 3.1217427253723145, + "learning_rate": 3.8885084989226715e-06, + "loss": 0.3114, + "step": 30400 + }, + { + "epoch": 3.5971137922876744, + "grad_norm": 2.4833168983459473, + "learning_rate": 3.887790280105339e-06, + "loss": 0.257, + "step": 30410 + }, + { + "epoch": 3.5982966643009227, + "grad_norm": 3.588465929031372, + "learning_rate": 3.887072061288005e-06, + "loss": 0.2929, + "step": 30420 + }, + { + "epoch": 3.599479536314171, + "grad_norm": 2.513824939727783, + "learning_rate": 3.886353842470673e-06, + "loss": 0.2817, + "step": 30430 + }, + { + "epoch": 3.600662408327419, + "grad_norm": 2.834028720855713, + "learning_rate": 3.88563562365334e-06, + "loss": 0.2872, + "step": 30440 + }, + { + "epoch": 3.601845280340667, + "grad_norm": 2.060053586959839, + "learning_rate": 3.884917404836007e-06, + "loss": 0.3027, + "step": 30450 + }, + { + "epoch": 3.6030281523539154, + "grad_norm": 2.384411096572876, + "learning_rate": 3.884199186018674e-06, + "loss": 0.2466, + "step": 30460 + }, + { + "epoch": 3.6042110243671637, + "grad_norm": 2.235736131668091, + "learning_rate": 3.883480967201341e-06, + "loss": 0.2809, + "step": 30470 + }, + { + "epoch": 3.6053938963804115, + "grad_norm": 2.988762378692627, + "learning_rate": 3.882762748384008e-06, + "loss": 0.2932, + "step": 30480 + }, + { + "epoch": 3.60657676839366, + "grad_norm": 2.637009620666504, + "learning_rate": 3.882044529566675e-06, + "loss": 0.2987, + "step": 30490 + }, + { + "epoch": 3.607759640406908, + "grad_norm": 2.588085651397705, + "learning_rate": 3.881326310749342e-06, + "loss": 0.2842, + "step": 30500 + }, + { + "epoch": 3.608942512420156, + "grad_norm": 2.797858953475952, + "learning_rate": 3.8806080919320085e-06, + "loss": 0.3066, + "step": 30510 + }, + { + "epoch": 3.6101253844334042, + "grad_norm": 2.368443250656128, + "learning_rate": 3.8798898731146755e-06, + "loss": 0.2947, + "step": 30520 + }, + { + "epoch": 3.6113082564466525, + "grad_norm": 2.7464444637298584, + "learning_rate": 3.879171654297342e-06, + "loss": 0.2945, + "step": 30530 + }, + { + "epoch": 3.612491128459901, + "grad_norm": 2.8854618072509766, + "learning_rate": 3.878453435480009e-06, + "loss": 0.3248, + "step": 30540 + }, + { + "epoch": 3.6136740004731487, + "grad_norm": 2.9559497833251953, + "learning_rate": 3.877735216662677e-06, + "loss": 0.2718, + "step": 30550 + }, + { + "epoch": 3.614856872486397, + "grad_norm": 3.2206692695617676, + "learning_rate": 3.877016997845343e-06, + "loss": 0.3053, + "step": 30560 + }, + { + "epoch": 3.6160397444996453, + "grad_norm": 2.0784428119659424, + "learning_rate": 3.876298779028011e-06, + "loss": 0.3018, + "step": 30570 + }, + { + "epoch": 3.617222616512893, + "grad_norm": 4.011874198913574, + "learning_rate": 3.875580560210677e-06, + "loss": 0.3048, + "step": 30580 + }, + { + "epoch": 3.6184054885261414, + "grad_norm": 3.6103837490081787, + "learning_rate": 3.874862341393345e-06, + "loss": 0.3138, + "step": 30590 + }, + { + "epoch": 3.6195883605393897, + "grad_norm": 2.5846097469329834, + "learning_rate": 3.874144122576012e-06, + "loss": 0.3142, + "step": 30600 + }, + { + "epoch": 3.620771232552638, + "grad_norm": 2.300175428390503, + "learning_rate": 3.873425903758679e-06, + "loss": 0.305, + "step": 30610 + }, + { + "epoch": 3.621954104565886, + "grad_norm": 2.6782939434051514, + "learning_rate": 3.8727076849413455e-06, + "loss": 0.3091, + "step": 30620 + }, + { + "epoch": 3.623136976579134, + "grad_norm": 2.2132413387298584, + "learning_rate": 3.8719894661240125e-06, + "loss": 0.2721, + "step": 30630 + }, + { + "epoch": 3.6243198485923824, + "grad_norm": 3.218013286590576, + "learning_rate": 3.87127124730668e-06, + "loss": 0.317, + "step": 30640 + }, + { + "epoch": 3.6255027206056303, + "grad_norm": 3.503964424133301, + "learning_rate": 3.870553028489346e-06, + "loss": 0.3056, + "step": 30650 + }, + { + "epoch": 3.6266855926188786, + "grad_norm": 2.749220132827759, + "learning_rate": 3.869834809672014e-06, + "loss": 0.3126, + "step": 30660 + }, + { + "epoch": 3.627868464632127, + "grad_norm": 3.0884857177734375, + "learning_rate": 3.86911659085468e-06, + "loss": 0.3057, + "step": 30670 + }, + { + "epoch": 3.629051336645375, + "grad_norm": 4.062482833862305, + "learning_rate": 3.868398372037348e-06, + "loss": 0.3089, + "step": 30680 + }, + { + "epoch": 3.630234208658623, + "grad_norm": 2.715261220932007, + "learning_rate": 3.867680153220014e-06, + "loss": 0.281, + "step": 30690 + }, + { + "epoch": 3.6314170806718713, + "grad_norm": 2.3458049297332764, + "learning_rate": 3.866961934402682e-06, + "loss": 0.2811, + "step": 30700 + }, + { + "epoch": 3.6325999526851196, + "grad_norm": 4.217658996582031, + "learning_rate": 3.866243715585349e-06, + "loss": 0.2974, + "step": 30710 + }, + { + "epoch": 3.6337828246983674, + "grad_norm": 4.3304219245910645, + "learning_rate": 3.865525496768016e-06, + "loss": 0.2866, + "step": 30720 + }, + { + "epoch": 3.6349656967116157, + "grad_norm": 2.932298421859741, + "learning_rate": 3.8648072779506825e-06, + "loss": 0.2598, + "step": 30730 + }, + { + "epoch": 3.636148568724864, + "grad_norm": 2.707641124725342, + "learning_rate": 3.8640890591333495e-06, + "loss": 0.302, + "step": 30740 + }, + { + "epoch": 3.6373314407381123, + "grad_norm": 3.807304859161377, + "learning_rate": 3.863370840316016e-06, + "loss": 0.2939, + "step": 30750 + }, + { + "epoch": 3.63851431275136, + "grad_norm": 3.115504741668701, + "learning_rate": 3.862652621498683e-06, + "loss": 0.2974, + "step": 30760 + }, + { + "epoch": 3.6396971847646085, + "grad_norm": 3.444378614425659, + "learning_rate": 3.86193440268135e-06, + "loss": 0.3166, + "step": 30770 + }, + { + "epoch": 3.6408800567778568, + "grad_norm": 2.917494773864746, + "learning_rate": 3.861216183864017e-06, + "loss": 0.2872, + "step": 30780 + }, + { + "epoch": 3.6420629287911046, + "grad_norm": 2.107465982437134, + "learning_rate": 3.860497965046684e-06, + "loss": 0.2691, + "step": 30790 + }, + { + "epoch": 3.643245800804353, + "grad_norm": 2.2286996841430664, + "learning_rate": 3.859779746229351e-06, + "loss": 0.2554, + "step": 30800 + }, + { + "epoch": 3.644428672817601, + "grad_norm": 2.9824001789093018, + "learning_rate": 3.859061527412018e-06, + "loss": 0.304, + "step": 30810 + }, + { + "epoch": 3.6456115448308495, + "grad_norm": 2.3912899494171143, + "learning_rate": 3.858343308594686e-06, + "loss": 0.2915, + "step": 30820 + }, + { + "epoch": 3.6467944168440973, + "grad_norm": 5.2417192459106445, + "learning_rate": 3.857625089777352e-06, + "loss": 0.2984, + "step": 30830 + }, + { + "epoch": 3.6479772888573456, + "grad_norm": 2.8914151191711426, + "learning_rate": 3.8569068709600195e-06, + "loss": 0.2892, + "step": 30840 + }, + { + "epoch": 3.649160160870594, + "grad_norm": 4.74766206741333, + "learning_rate": 3.8561886521426865e-06, + "loss": 0.2714, + "step": 30850 + }, + { + "epoch": 3.6503430328838418, + "grad_norm": 3.0595216751098633, + "learning_rate": 3.855470433325353e-06, + "loss": 0.2804, + "step": 30860 + }, + { + "epoch": 3.65152590489709, + "grad_norm": 4.202428340911865, + "learning_rate": 3.85475221450802e-06, + "loss": 0.2727, + "step": 30870 + }, + { + "epoch": 3.6527087769103384, + "grad_norm": 4.499485015869141, + "learning_rate": 3.854033995690687e-06, + "loss": 0.2583, + "step": 30880 + }, + { + "epoch": 3.6538916489235866, + "grad_norm": 3.0247628688812256, + "learning_rate": 3.853315776873354e-06, + "loss": 0.31, + "step": 30890 + }, + { + "epoch": 3.6550745209368345, + "grad_norm": 2.4006307125091553, + "learning_rate": 3.852597558056021e-06, + "loss": 0.2623, + "step": 30900 + }, + { + "epoch": 3.656257392950083, + "grad_norm": 2.835519552230835, + "learning_rate": 3.851879339238688e-06, + "loss": 0.2986, + "step": 30910 + }, + { + "epoch": 3.657440264963331, + "grad_norm": 2.091010570526123, + "learning_rate": 3.851161120421355e-06, + "loss": 0.2754, + "step": 30920 + }, + { + "epoch": 3.658623136976579, + "grad_norm": 3.398745059967041, + "learning_rate": 3.850442901604023e-06, + "loss": 0.3174, + "step": 30930 + }, + { + "epoch": 3.659806008989827, + "grad_norm": 3.152393102645874, + "learning_rate": 3.849724682786689e-06, + "loss": 0.3182, + "step": 30940 + }, + { + "epoch": 3.6609888810030755, + "grad_norm": 3.5269439220428467, + "learning_rate": 3.8490064639693565e-06, + "loss": 0.2934, + "step": 30950 + }, + { + "epoch": 3.662171753016324, + "grad_norm": 2.7336370944976807, + "learning_rate": 3.848288245152023e-06, + "loss": 0.292, + "step": 30960 + }, + { + "epoch": 3.6633546250295717, + "grad_norm": 3.7625656127929688, + "learning_rate": 3.84757002633469e-06, + "loss": 0.2588, + "step": 30970 + }, + { + "epoch": 3.66453749704282, + "grad_norm": 2.673926830291748, + "learning_rate": 3.846851807517357e-06, + "loss": 0.3115, + "step": 30980 + }, + { + "epoch": 3.6657203690560682, + "grad_norm": 4.187891006469727, + "learning_rate": 3.846133588700024e-06, + "loss": 0.3174, + "step": 30990 + }, + { + "epoch": 3.666903241069316, + "grad_norm": 1.9319275617599487, + "learning_rate": 3.845415369882691e-06, + "loss": 0.2886, + "step": 31000 + }, + { + "epoch": 3.6680861130825644, + "grad_norm": 3.7331576347351074, + "learning_rate": 3.844697151065358e-06, + "loss": 0.3056, + "step": 31010 + }, + { + "epoch": 3.6692689850958127, + "grad_norm": 2.1255035400390625, + "learning_rate": 3.843978932248025e-06, + "loss": 0.2886, + "step": 31020 + }, + { + "epoch": 3.670451857109061, + "grad_norm": 2.0068323612213135, + "learning_rate": 3.843260713430692e-06, + "loss": 0.268, + "step": 31030 + }, + { + "epoch": 3.671634729122309, + "grad_norm": 3.2803544998168945, + "learning_rate": 3.842542494613359e-06, + "loss": 0.2855, + "step": 31040 + }, + { + "epoch": 3.672817601135557, + "grad_norm": 3.192193031311035, + "learning_rate": 3.841824275796026e-06, + "loss": 0.2921, + "step": 31050 + }, + { + "epoch": 3.6740004731488054, + "grad_norm": 2.564889430999756, + "learning_rate": 3.841106056978693e-06, + "loss": 0.319, + "step": 31060 + }, + { + "epoch": 3.6751833451620533, + "grad_norm": 2.705695152282715, + "learning_rate": 3.84038783816136e-06, + "loss": 0.3019, + "step": 31070 + }, + { + "epoch": 3.6763662171753015, + "grad_norm": 2.6498305797576904, + "learning_rate": 3.8396696193440265e-06, + "loss": 0.3075, + "step": 31080 + }, + { + "epoch": 3.67754908918855, + "grad_norm": 2.5106942653656006, + "learning_rate": 3.838951400526694e-06, + "loss": 0.2847, + "step": 31090 + }, + { + "epoch": 3.678731961201798, + "grad_norm": 2.990410089492798, + "learning_rate": 3.83823318170936e-06, + "loss": 0.2857, + "step": 31100 + }, + { + "epoch": 3.679914833215046, + "grad_norm": 3.141381025314331, + "learning_rate": 3.837514962892028e-06, + "loss": 0.2668, + "step": 31110 + }, + { + "epoch": 3.6810977052282943, + "grad_norm": 3.528865337371826, + "learning_rate": 3.836796744074695e-06, + "loss": 0.277, + "step": 31120 + }, + { + "epoch": 3.6822805772415426, + "grad_norm": 4.061430931091309, + "learning_rate": 3.836078525257362e-06, + "loss": 0.2929, + "step": 31130 + }, + { + "epoch": 3.6834634492547904, + "grad_norm": 2.838484287261963, + "learning_rate": 3.835360306440029e-06, + "loss": 0.2854, + "step": 31140 + }, + { + "epoch": 3.6846463212680387, + "grad_norm": 3.8740718364715576, + "learning_rate": 3.834642087622696e-06, + "loss": 0.3001, + "step": 31150 + }, + { + "epoch": 3.685829193281287, + "grad_norm": 2.8395168781280518, + "learning_rate": 3.833923868805363e-06, + "loss": 0.2852, + "step": 31160 + }, + { + "epoch": 3.6870120652945353, + "grad_norm": 2.3050224781036377, + "learning_rate": 3.83320564998803e-06, + "loss": 0.3339, + "step": 31170 + }, + { + "epoch": 3.688194937307783, + "grad_norm": 2.0187599658966064, + "learning_rate": 3.832487431170697e-06, + "loss": 0.2871, + "step": 31180 + }, + { + "epoch": 3.6893778093210314, + "grad_norm": 2.740751028060913, + "learning_rate": 3.8317692123533635e-06, + "loss": 0.2929, + "step": 31190 + }, + { + "epoch": 3.6905606813342797, + "grad_norm": 3.669309377670288, + "learning_rate": 3.831050993536031e-06, + "loss": 0.2948, + "step": 31200 + }, + { + "epoch": 3.6917435533475276, + "grad_norm": 2.3854892253875732, + "learning_rate": 3.830332774718697e-06, + "loss": 0.2667, + "step": 31210 + }, + { + "epoch": 3.692926425360776, + "grad_norm": 3.5488622188568115, + "learning_rate": 3.829614555901365e-06, + "loss": 0.2893, + "step": 31220 + }, + { + "epoch": 3.694109297374024, + "grad_norm": 2.9817185401916504, + "learning_rate": 3.828896337084031e-06, + "loss": 0.3179, + "step": 31230 + }, + { + "epoch": 3.6952921693872725, + "grad_norm": 3.7668981552124023, + "learning_rate": 3.828178118266699e-06, + "loss": 0.2976, + "step": 31240 + }, + { + "epoch": 3.6964750414005203, + "grad_norm": 3.384310483932495, + "learning_rate": 3.827459899449365e-06, + "loss": 0.2769, + "step": 31250 + }, + { + "epoch": 3.6976579134137686, + "grad_norm": 3.0994839668273926, + "learning_rate": 3.826741680632033e-06, + "loss": 0.2736, + "step": 31260 + }, + { + "epoch": 3.698840785427017, + "grad_norm": 3.4753875732421875, + "learning_rate": 3.8260234618147e-06, + "loss": 0.3087, + "step": 31270 + }, + { + "epoch": 3.7000236574402647, + "grad_norm": 3.772836685180664, + "learning_rate": 3.825305242997367e-06, + "loss": 0.2893, + "step": 31280 + }, + { + "epoch": 3.701206529453513, + "grad_norm": 2.2013442516326904, + "learning_rate": 3.824587024180034e-06, + "loss": 0.3055, + "step": 31290 + }, + { + "epoch": 3.7023894014667613, + "grad_norm": 4.810434341430664, + "learning_rate": 3.8238688053627005e-06, + "loss": 0.276, + "step": 31300 + }, + { + "epoch": 3.7035722734800096, + "grad_norm": 3.599900007247925, + "learning_rate": 3.8231505865453674e-06, + "loss": 0.2991, + "step": 31310 + }, + { + "epoch": 3.704755145493258, + "grad_norm": 2.5832552909851074, + "learning_rate": 3.822432367728034e-06, + "loss": 0.2866, + "step": 31320 + }, + { + "epoch": 3.7059380175065058, + "grad_norm": 3.2084455490112305, + "learning_rate": 3.821714148910701e-06, + "loss": 0.2615, + "step": 31330 + }, + { + "epoch": 3.707120889519754, + "grad_norm": 2.4905202388763428, + "learning_rate": 3.820995930093368e-06, + "loss": 0.2596, + "step": 31340 + }, + { + "epoch": 3.708303761533002, + "grad_norm": 3.578376293182373, + "learning_rate": 3.820277711276035e-06, + "loss": 0.3025, + "step": 31350 + }, + { + "epoch": 3.70948663354625, + "grad_norm": 2.4837441444396973, + "learning_rate": 3.819559492458703e-06, + "loss": 0.2811, + "step": 31360 + }, + { + "epoch": 3.7106695055594985, + "grad_norm": 2.794616222381592, + "learning_rate": 3.81884127364137e-06, + "loss": 0.3052, + "step": 31370 + }, + { + "epoch": 3.7118523775727468, + "grad_norm": 3.169340133666992, + "learning_rate": 3.818123054824037e-06, + "loss": 0.3129, + "step": 31380 + }, + { + "epoch": 3.713035249585995, + "grad_norm": 2.6523597240448, + "learning_rate": 3.817404836006704e-06, + "loss": 0.2436, + "step": 31390 + }, + { + "epoch": 3.714218121599243, + "grad_norm": 3.1492745876312256, + "learning_rate": 3.816686617189371e-06, + "loss": 0.307, + "step": 31400 + }, + { + "epoch": 3.715400993612491, + "grad_norm": 2.586974859237671, + "learning_rate": 3.8159683983720375e-06, + "loss": 0.315, + "step": 31410 + }, + { + "epoch": 3.716583865625739, + "grad_norm": 2.3166165351867676, + "learning_rate": 3.8152501795547044e-06, + "loss": 0.2497, + "step": 31420 + }, + { + "epoch": 3.7177667376389874, + "grad_norm": 3.380039930343628, + "learning_rate": 3.814531960737372e-06, + "loss": 0.3091, + "step": 31430 + }, + { + "epoch": 3.7189496096522356, + "grad_norm": 3.345799446105957, + "learning_rate": 3.8138137419200383e-06, + "loss": 0.3012, + "step": 31440 + }, + { + "epoch": 3.720132481665484, + "grad_norm": 2.7278225421905518, + "learning_rate": 3.8130955231027056e-06, + "loss": 0.2915, + "step": 31450 + }, + { + "epoch": 3.7213153536787322, + "grad_norm": 2.6513779163360596, + "learning_rate": 3.812377304285372e-06, + "loss": 0.2837, + "step": 31460 + }, + { + "epoch": 3.72249822569198, + "grad_norm": 3.645665407180786, + "learning_rate": 3.8116590854680395e-06, + "loss": 0.3055, + "step": 31470 + }, + { + "epoch": 3.7236810977052284, + "grad_norm": 2.8154947757720947, + "learning_rate": 3.810940866650706e-06, + "loss": 0.2655, + "step": 31480 + }, + { + "epoch": 3.724863969718476, + "grad_norm": 2.4965240955352783, + "learning_rate": 3.8102226478333733e-06, + "loss": 0.2711, + "step": 31490 + }, + { + "epoch": 3.7260468417317245, + "grad_norm": 3.8089048862457275, + "learning_rate": 3.8095044290160402e-06, + "loss": 0.2908, + "step": 31500 + }, + { + "epoch": 3.727229713744973, + "grad_norm": 2.390972852706909, + "learning_rate": 3.8087862101987076e-06, + "loss": 0.297, + "step": 31510 + }, + { + "epoch": 3.728412585758221, + "grad_norm": 2.890800952911377, + "learning_rate": 3.808067991381374e-06, + "loss": 0.2924, + "step": 31520 + }, + { + "epoch": 3.7295954577714694, + "grad_norm": 2.223830461502075, + "learning_rate": 3.8073497725640414e-06, + "loss": 0.2866, + "step": 31530 + }, + { + "epoch": 3.7307783297847172, + "grad_norm": 4.307246685028076, + "learning_rate": 3.806631553746708e-06, + "loss": 0.3134, + "step": 31540 + }, + { + "epoch": 3.7319612017979655, + "grad_norm": 3.1333954334259033, + "learning_rate": 3.8059133349293753e-06, + "loss": 0.2665, + "step": 31550 + }, + { + "epoch": 3.7331440738112134, + "grad_norm": 5.665109157562256, + "learning_rate": 3.8051951161120418e-06, + "loss": 0.275, + "step": 31560 + }, + { + "epoch": 3.7343269458244617, + "grad_norm": 3.2309370040893555, + "learning_rate": 3.804476897294709e-06, + "loss": 0.3323, + "step": 31570 + }, + { + "epoch": 3.73550981783771, + "grad_norm": 2.3520450592041016, + "learning_rate": 3.803758678477376e-06, + "loss": 0.3115, + "step": 31580 + }, + { + "epoch": 3.7366926898509583, + "grad_norm": 2.021888256072998, + "learning_rate": 3.8030404596600434e-06, + "loss": 0.2255, + "step": 31590 + }, + { + "epoch": 3.7378755618642066, + "grad_norm": 3.181809902191162, + "learning_rate": 3.80232224084271e-06, + "loss": 0.3065, + "step": 31600 + }, + { + "epoch": 3.7390584338774544, + "grad_norm": 2.6291301250457764, + "learning_rate": 3.8016040220253772e-06, + "loss": 0.2832, + "step": 31610 + }, + { + "epoch": 3.7402413058907027, + "grad_norm": 2.8491785526275635, + "learning_rate": 3.8008858032080446e-06, + "loss": 0.3048, + "step": 31620 + }, + { + "epoch": 3.7414241779039505, + "grad_norm": 2.6975810527801514, + "learning_rate": 3.800167584390711e-06, + "loss": 0.311, + "step": 31630 + }, + { + "epoch": 3.742607049917199, + "grad_norm": 2.963815689086914, + "learning_rate": 3.7994493655733784e-06, + "loss": 0.3217, + "step": 31640 + }, + { + "epoch": 3.743789921930447, + "grad_norm": 2.3831546306610107, + "learning_rate": 3.798731146756045e-06, + "loss": 0.3046, + "step": 31650 + }, + { + "epoch": 3.7449727939436954, + "grad_norm": 3.1295571327209473, + "learning_rate": 3.7980129279387123e-06, + "loss": 0.2888, + "step": 31660 + }, + { + "epoch": 3.7461556659569437, + "grad_norm": 2.733480215072632, + "learning_rate": 3.7972947091213788e-06, + "loss": 0.2986, + "step": 31670 + }, + { + "epoch": 3.7473385379701916, + "grad_norm": 3.7552599906921387, + "learning_rate": 3.796576490304046e-06, + "loss": 0.2841, + "step": 31680 + }, + { + "epoch": 3.74852140998344, + "grad_norm": 2.9118576049804688, + "learning_rate": 3.795858271486713e-06, + "loss": 0.2524, + "step": 31690 + }, + { + "epoch": 3.7497042819966877, + "grad_norm": 4.615795135498047, + "learning_rate": 3.7951400526693804e-06, + "loss": 0.3208, + "step": 31700 + }, + { + "epoch": 3.750887154009936, + "grad_norm": 3.5630717277526855, + "learning_rate": 3.794421833852047e-06, + "loss": 0.3123, + "step": 31710 + }, + { + "epoch": 3.750887154009936, + "eval_accuracy": 0.8592174867751272, + "eval_loss": 0.32891905307769775, + "eval_runtime": 77.9037, + "eval_safe_aucpr": 0.9157634659184628, + "eval_safe_f1": 0.8470532955017801, + "eval_safe_fpr": 0.1564975636004898, + "eval_safe_precision": 0.8174049529124521, + "eval_safe_recall": 0.8789333533360837, + "eval_samples_per_second": 771.645, + "eval_steps_per_second": 12.066, + "eval_unsafe_aucpr": 0.9534351493536815, + "eval_unsafe_f1": 0.8695893366206949, + "eval_unsafe_fpr": 0.12106664666391584, + "eval_unsafe_precision": 0.8973413051774584, + "eval_unsafe_recall": 0.8435024363995097, + "step": 31710 + }, + { + "epoch": 3.750887154009936, + "step": 31710, + "total_flos": 3.585941570958701e+17, + "train_loss": 0.3279850626021896, + "train_runtime": 10021.2364, + "train_samples_per_second": 539.874, + "train_steps_per_second": 8.436 + } + ], + "logging_steps": 10, + "max_steps": 84540, + "num_input_tokens_seen": 0, + "num_train_epochs": 10, + "save_steps": 2114, + "stateful_callbacks": { + "EarlyStoppingCallback": { + "args": { + "early_stopping_patience": 4, + "early_stopping_threshold": 0.0 + }, + "attributes": { + "early_stopping_patience_counter": 4 + } + }, + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": true + }, + "attributes": {} + } + }, + "total_flos": 3.585941570958701e+17, + "train_batch_size": 64, + "trial_name": null, + "trial_params": null +}