| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 1.0, | |
| "eval_steps": 500, | |
| "global_step": 9480, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0010548523206751054, | |
| "grad_norm": 0.08000069111585617, | |
| "learning_rate": 0.00015822784810126583, | |
| "loss": 7.5887, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.002109704641350211, | |
| "grad_norm": 0.07872951030731201, | |
| "learning_rate": 0.00031645569620253165, | |
| "loss": 7.5464, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.0031645569620253164, | |
| "grad_norm": 0.07734671235084534, | |
| "learning_rate": 0.00047468354430379745, | |
| "loss": 7.4957, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.004219409282700422, | |
| "grad_norm": 0.08023999631404877, | |
| "learning_rate": 0.0006329113924050633, | |
| "loss": 7.4244, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.005274261603375527, | |
| "grad_norm": 0.08475606143474579, | |
| "learning_rate": 0.0007911392405063291, | |
| "loss": 7.3265, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.006329113924050633, | |
| "grad_norm": 0.10008183121681213, | |
| "learning_rate": 0.0009493670886075949, | |
| "loss": 7.2047, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.007383966244725738, | |
| "grad_norm": 0.09628511965274811, | |
| "learning_rate": 0.0011075949367088608, | |
| "loss": 7.0673, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.008438818565400843, | |
| "grad_norm": 0.21714860200881958, | |
| "learning_rate": 0.0012658227848101266, | |
| "loss": 6.9366, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.00949367088607595, | |
| "grad_norm": 0.1056312844157219, | |
| "learning_rate": 0.0014240506329113926, | |
| "loss": 6.8047, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.010548523206751054, | |
| "grad_norm": 0.10105545073747635, | |
| "learning_rate": 0.0015, | |
| "loss": 6.6619, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.011603375527426161, | |
| "grad_norm": 0.10644520819187164, | |
| "learning_rate": 0.0015, | |
| "loss": 6.5215, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.012658227848101266, | |
| "grad_norm": 0.12805885076522827, | |
| "learning_rate": 0.0015, | |
| "loss": 6.3894, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.013713080168776372, | |
| "grad_norm": 0.11895852535963058, | |
| "learning_rate": 0.0015, | |
| "loss": 6.2576, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.014767932489451477, | |
| "grad_norm": 0.10951970517635345, | |
| "learning_rate": 0.0015, | |
| "loss": 6.1227, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.015822784810126583, | |
| "grad_norm": 0.10010507702827454, | |
| "learning_rate": 0.0015, | |
| "loss": 5.9806, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.016877637130801686, | |
| "grad_norm": 0.10386305302381516, | |
| "learning_rate": 0.0015, | |
| "loss": 5.8499, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.017932489451476793, | |
| "grad_norm": 0.17313894629478455, | |
| "learning_rate": 0.0015, | |
| "loss": 5.7197, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.0189873417721519, | |
| "grad_norm": 0.1267988085746765, | |
| "learning_rate": 0.0015, | |
| "loss": 5.5993, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.020042194092827006, | |
| "grad_norm": 0.18178296089172363, | |
| "learning_rate": 0.0015, | |
| "loss": 5.483, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.02109704641350211, | |
| "grad_norm": 0.17508190870285034, | |
| "learning_rate": 0.0015, | |
| "loss": 5.3645, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.022151898734177215, | |
| "grad_norm": 0.3925415575504303, | |
| "learning_rate": 0.0015, | |
| "loss": 5.2723, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.023206751054852322, | |
| "grad_norm": 0.1316889524459839, | |
| "learning_rate": 0.0015, | |
| "loss": 5.1758, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.024261603375527425, | |
| "grad_norm": 0.12996184825897217, | |
| "learning_rate": 0.0015, | |
| "loss": 5.0767, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.02531645569620253, | |
| "grad_norm": 0.2191348820924759, | |
| "learning_rate": 0.0015, | |
| "loss": 4.9929, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.026371308016877638, | |
| "grad_norm": 0.16052663326263428, | |
| "learning_rate": 0.0015, | |
| "loss": 4.9128, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.027426160337552744, | |
| "grad_norm": 0.2407989203929901, | |
| "learning_rate": 0.0015, | |
| "loss": 4.8402, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.028481012658227847, | |
| "grad_norm": 0.15928399562835693, | |
| "learning_rate": 0.0015, | |
| "loss": 4.772, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.029535864978902954, | |
| "grad_norm": 0.18819557130336761, | |
| "learning_rate": 0.0015, | |
| "loss": 4.704, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.03059071729957806, | |
| "grad_norm": 0.3014315366744995, | |
| "learning_rate": 0.0015, | |
| "loss": 4.6373, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.03164556962025317, | |
| "grad_norm": 0.2249811440706253, | |
| "learning_rate": 0.0015, | |
| "loss": 4.5768, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.03270042194092827, | |
| "grad_norm": 0.2288593202829361, | |
| "learning_rate": 0.0015, | |
| "loss": 4.5108, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.03375527426160337, | |
| "grad_norm": 0.30202898383140564, | |
| "learning_rate": 0.0015, | |
| "loss": 4.4485, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.03481012658227848, | |
| "grad_norm": 0.2222922146320343, | |
| "learning_rate": 0.0015, | |
| "loss": 4.3979, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.035864978902953586, | |
| "grad_norm": 0.2275601029396057, | |
| "learning_rate": 0.0015, | |
| "loss": 4.3382, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.03691983122362869, | |
| "grad_norm": 0.44312891364097595, | |
| "learning_rate": 0.0015, | |
| "loss": 4.2705, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.0379746835443038, | |
| "grad_norm": 0.25240451097488403, | |
| "learning_rate": 0.0015, | |
| "loss": 4.2262, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.039029535864978905, | |
| "grad_norm": 0.26719385385513306, | |
| "learning_rate": 0.0015, | |
| "loss": 4.1729, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.04008438818565401, | |
| "grad_norm": 0.3257826864719391, | |
| "learning_rate": 0.0015, | |
| "loss": 4.1246, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.04113924050632911, | |
| "grad_norm": 0.2367611527442932, | |
| "learning_rate": 0.0015, | |
| "loss": 4.0749, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.04219409282700422, | |
| "grad_norm": 0.3027140200138092, | |
| "learning_rate": 0.0015, | |
| "loss": 4.0161, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.043248945147679324, | |
| "grad_norm": 0.27771511673927307, | |
| "learning_rate": 0.0015, | |
| "loss": 3.9917, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.04430379746835443, | |
| "grad_norm": 0.43055254220962524, | |
| "learning_rate": 0.0015, | |
| "loss": 3.9399, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.04535864978902954, | |
| "grad_norm": 0.3237093389034271, | |
| "learning_rate": 0.0015, | |
| "loss": 3.8899, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.046413502109704644, | |
| "grad_norm": 0.37645670771598816, | |
| "learning_rate": 0.0015, | |
| "loss": 3.8477, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.04746835443037975, | |
| "grad_norm": 0.34314456582069397, | |
| "learning_rate": 0.0015, | |
| "loss": 3.8153, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.04852320675105485, | |
| "grad_norm": 0.38104280829429626, | |
| "learning_rate": 0.0015, | |
| "loss": 3.7744, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.049578059071729956, | |
| "grad_norm": 0.38231614232063293, | |
| "learning_rate": 0.0015, | |
| "loss": 3.7297, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.05063291139240506, | |
| "grad_norm": 0.333865761756897, | |
| "learning_rate": 0.0015, | |
| "loss": 3.7088, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.05168776371308017, | |
| "grad_norm": 0.4378509223461151, | |
| "learning_rate": 0.0015, | |
| "loss": 3.6681, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.052742616033755275, | |
| "grad_norm": 0.29420173168182373, | |
| "learning_rate": 0.0015, | |
| "loss": 3.6326, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.05379746835443038, | |
| "grad_norm": 0.2545853555202484, | |
| "learning_rate": 0.0015, | |
| "loss": 3.6121, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.05485232067510549, | |
| "grad_norm": 0.4205600619316101, | |
| "learning_rate": 0.0015, | |
| "loss": 3.5755, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.05590717299578059, | |
| "grad_norm": 0.4887816309928894, | |
| "learning_rate": 0.0015, | |
| "loss": 3.5506, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.056962025316455694, | |
| "grad_norm": 0.3589875102043152, | |
| "learning_rate": 0.0015, | |
| "loss": 3.5253, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.0580168776371308, | |
| "grad_norm": 0.37651869654655457, | |
| "learning_rate": 0.0015, | |
| "loss": 3.5005, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.05907172995780591, | |
| "grad_norm": 0.4500958323478699, | |
| "learning_rate": 0.0015, | |
| "loss": 3.4524, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.060126582278481014, | |
| "grad_norm": 0.3502822518348694, | |
| "learning_rate": 0.0015, | |
| "loss": 3.4423, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.06118143459915612, | |
| "grad_norm": 0.5461118817329407, | |
| "learning_rate": 0.0015, | |
| "loss": 3.4262, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.06223628691983123, | |
| "grad_norm": 0.7087445855140686, | |
| "learning_rate": 0.0015, | |
| "loss": 3.3883, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.06329113924050633, | |
| "grad_norm": 0.38886040449142456, | |
| "learning_rate": 0.0015, | |
| "loss": 3.357, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.06434599156118144, | |
| "grad_norm": 0.38216376304626465, | |
| "learning_rate": 0.0015, | |
| "loss": 3.3421, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.06540084388185655, | |
| "grad_norm": 0.5458016395568848, | |
| "learning_rate": 0.0015, | |
| "loss": 3.3245, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.06645569620253164, | |
| "grad_norm": 0.40753501653671265, | |
| "learning_rate": 0.0015, | |
| "loss": 3.2963, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.06751054852320675, | |
| "grad_norm": 0.5089264512062073, | |
| "learning_rate": 0.0015, | |
| "loss": 3.2883, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.06856540084388185, | |
| "grad_norm": 0.49709829688072205, | |
| "learning_rate": 0.0015, | |
| "loss": 3.2718, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.06962025316455696, | |
| "grad_norm": 0.5032679438591003, | |
| "learning_rate": 0.0015, | |
| "loss": 3.2483, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.07067510548523206, | |
| "grad_norm": 0.4408726990222931, | |
| "learning_rate": 0.0015, | |
| "loss": 3.2231, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.07172995780590717, | |
| "grad_norm": 0.528001606464386, | |
| "learning_rate": 0.0015, | |
| "loss": 3.1996, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.07278481012658228, | |
| "grad_norm": 0.3725454807281494, | |
| "learning_rate": 0.0015, | |
| "loss": 3.2004, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.07383966244725738, | |
| "grad_norm": 0.43372929096221924, | |
| "learning_rate": 0.0015, | |
| "loss": 3.1707, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.07489451476793249, | |
| "grad_norm": 0.4604400098323822, | |
| "learning_rate": 0.0015, | |
| "loss": 3.1517, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.0759493670886076, | |
| "grad_norm": 0.463722825050354, | |
| "learning_rate": 0.0015, | |
| "loss": 3.1214, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.0770042194092827, | |
| "grad_norm": 0.812790036201477, | |
| "learning_rate": 0.0015, | |
| "loss": 3.1103, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.07805907172995781, | |
| "grad_norm": 0.5507075190544128, | |
| "learning_rate": 0.0015, | |
| "loss": 3.1113, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.07911392405063292, | |
| "grad_norm": 0.7195579409599304, | |
| "learning_rate": 0.0015, | |
| "loss": 3.0831, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.08016877637130802, | |
| "grad_norm": 0.46877631545066833, | |
| "learning_rate": 0.0015, | |
| "loss": 3.0686, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.08122362869198312, | |
| "grad_norm": 0.5957846641540527, | |
| "learning_rate": 0.0015, | |
| "loss": 3.0481, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.08227848101265822, | |
| "grad_norm": 0.5983896255493164, | |
| "learning_rate": 0.0015, | |
| "loss": 3.0327, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.08333333333333333, | |
| "grad_norm": 0.6489009857177734, | |
| "learning_rate": 0.0015, | |
| "loss": 3.0334, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.08438818565400844, | |
| "grad_norm": 0.622799813747406, | |
| "learning_rate": 0.0015, | |
| "loss": 3.0016, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.08544303797468354, | |
| "grad_norm": 0.49533048272132874, | |
| "learning_rate": 0.0015, | |
| "loss": 2.9787, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.08649789029535865, | |
| "grad_norm": 0.6578832268714905, | |
| "learning_rate": 0.0015, | |
| "loss": 2.9771, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.08755274261603375, | |
| "grad_norm": 0.5393092036247253, | |
| "learning_rate": 0.0015, | |
| "loss": 2.9753, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.08860759493670886, | |
| "grad_norm": 0.4226798117160797, | |
| "learning_rate": 0.0015, | |
| "loss": 2.9433, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.08966244725738397, | |
| "grad_norm": 0.42221006751060486, | |
| "learning_rate": 0.0015, | |
| "loss": 2.9361, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.09071729957805907, | |
| "grad_norm": 0.7529399991035461, | |
| "learning_rate": 0.0015, | |
| "loss": 2.9224, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.09177215189873418, | |
| "grad_norm": 0.6478572487831116, | |
| "learning_rate": 0.0015, | |
| "loss": 2.9178, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.09282700421940929, | |
| "grad_norm": 0.5442308187484741, | |
| "learning_rate": 0.0015, | |
| "loss": 2.8998, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.0938818565400844, | |
| "grad_norm": 0.5076757073402405, | |
| "learning_rate": 0.0015, | |
| "loss": 2.8872, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.0949367088607595, | |
| "grad_norm": 0.5085638761520386, | |
| "learning_rate": 0.0015, | |
| "loss": 2.8723, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.09599156118143459, | |
| "grad_norm": 0.5724137425422668, | |
| "learning_rate": 0.0015, | |
| "loss": 2.8646, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.0970464135021097, | |
| "grad_norm": 0.5052751302719116, | |
| "learning_rate": 0.0015, | |
| "loss": 2.851, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.0981012658227848, | |
| "grad_norm": 0.45936843752861023, | |
| "learning_rate": 0.0015, | |
| "loss": 2.828, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.09915611814345991, | |
| "grad_norm": 0.6637619733810425, | |
| "learning_rate": 0.0015, | |
| "loss": 2.8232, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.10021097046413502, | |
| "grad_norm": 0.7366319894790649, | |
| "learning_rate": 0.0015, | |
| "loss": 2.8011, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.10126582278481013, | |
| "grad_norm": 0.6104623675346375, | |
| "learning_rate": 0.0015, | |
| "loss": 2.8028, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.10232067510548523, | |
| "grad_norm": 0.640235424041748, | |
| "learning_rate": 0.0015, | |
| "loss": 2.794, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.10337552742616034, | |
| "grad_norm": 0.5153405666351318, | |
| "learning_rate": 0.0015, | |
| "loss": 2.7663, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.10443037974683544, | |
| "grad_norm": 0.5124087929725647, | |
| "learning_rate": 0.0015, | |
| "loss": 2.7623, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.10548523206751055, | |
| "grad_norm": 0.5819884538650513, | |
| "learning_rate": 0.0015, | |
| "loss": 2.7603, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.10654008438818566, | |
| "grad_norm": 0.4700416922569275, | |
| "learning_rate": 0.0015, | |
| "loss": 2.7508, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.10759493670886076, | |
| "grad_norm": 0.7776612639427185, | |
| "learning_rate": 0.0015, | |
| "loss": 2.7355, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.10864978902953587, | |
| "grad_norm": 0.6290371417999268, | |
| "learning_rate": 0.0015, | |
| "loss": 2.722, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.10970464135021098, | |
| "grad_norm": 0.6244022846221924, | |
| "learning_rate": 0.0015, | |
| "loss": 2.7161, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.11075949367088607, | |
| "grad_norm": 0.6787243485450745, | |
| "learning_rate": 0.0015, | |
| "loss": 2.7142, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.11181434599156118, | |
| "grad_norm": 0.5542663335800171, | |
| "learning_rate": 0.0015, | |
| "loss": 2.6974, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.11286919831223628, | |
| "grad_norm": 0.39720186591148376, | |
| "learning_rate": 0.0015, | |
| "loss": 2.6765, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.11392405063291139, | |
| "grad_norm": 0.4627390205860138, | |
| "learning_rate": 0.0015, | |
| "loss": 2.6721, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.1149789029535865, | |
| "grad_norm": 0.5184245109558105, | |
| "learning_rate": 0.0015, | |
| "loss": 2.6663, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.1160337552742616, | |
| "grad_norm": 0.6086922287940979, | |
| "learning_rate": 0.0015, | |
| "loss": 2.6502, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.11708860759493671, | |
| "grad_norm": 0.48382633924484253, | |
| "learning_rate": 0.0015, | |
| "loss": 2.6494, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.11814345991561181, | |
| "grad_norm": 0.4573609232902527, | |
| "learning_rate": 0.0015, | |
| "loss": 2.6403, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.11919831223628692, | |
| "grad_norm": 0.6410148739814758, | |
| "learning_rate": 0.0015, | |
| "loss": 2.6144, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.12025316455696203, | |
| "grad_norm": 0.4648679494857788, | |
| "learning_rate": 0.0015, | |
| "loss": 2.6281, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.12130801687763713, | |
| "grad_norm": 0.6489707827568054, | |
| "learning_rate": 0.0015, | |
| "loss": 2.609, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.12236286919831224, | |
| "grad_norm": 0.39584651589393616, | |
| "learning_rate": 0.0015, | |
| "loss": 2.6147, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.12341772151898735, | |
| "grad_norm": 0.7939093708992004, | |
| "learning_rate": 0.0015, | |
| "loss": 2.5909, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.12447257383966245, | |
| "grad_norm": 0.43265780806541443, | |
| "learning_rate": 0.0015, | |
| "loss": 2.5774, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.12552742616033755, | |
| "grad_norm": 0.5087862014770508, | |
| "learning_rate": 0.0015, | |
| "loss": 2.5746, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.12658227848101267, | |
| "grad_norm": 0.49960434436798096, | |
| "learning_rate": 0.0015, | |
| "loss": 2.5792, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.12763713080168776, | |
| "grad_norm": 0.4003382623195648, | |
| "learning_rate": 0.0015, | |
| "loss": 2.571, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.12869198312236288, | |
| "grad_norm": 0.5398605465888977, | |
| "learning_rate": 0.0015, | |
| "loss": 2.5531, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.12974683544303797, | |
| "grad_norm": 0.399098664522171, | |
| "learning_rate": 0.0015, | |
| "loss": 2.5338, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.1308016877637131, | |
| "grad_norm": 0.4298166036605835, | |
| "learning_rate": 0.0015, | |
| "loss": 2.5455, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.13185654008438819, | |
| "grad_norm": 0.5648788809776306, | |
| "learning_rate": 0.0015, | |
| "loss": 2.5423, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.13291139240506328, | |
| "grad_norm": 0.6773810386657715, | |
| "learning_rate": 0.0015, | |
| "loss": 2.5273, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.1339662447257384, | |
| "grad_norm": 0.5596596598625183, | |
| "learning_rate": 0.0015, | |
| "loss": 2.5116, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.1350210970464135, | |
| "grad_norm": 0.3979465067386627, | |
| "learning_rate": 0.0015, | |
| "loss": 2.521, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.1360759493670886, | |
| "grad_norm": 0.7661407589912415, | |
| "learning_rate": 0.0015, | |
| "loss": 2.5034, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.1371308016877637, | |
| "grad_norm": 0.5518456697463989, | |
| "learning_rate": 0.0015, | |
| "loss": 2.4884, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.13818565400843882, | |
| "grad_norm": 0.42653220891952515, | |
| "learning_rate": 0.0015, | |
| "loss": 2.4881, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.13924050632911392, | |
| "grad_norm": 0.4962894320487976, | |
| "learning_rate": 0.0015, | |
| "loss": 2.4886, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.14029535864978904, | |
| "grad_norm": 0.7388845086097717, | |
| "learning_rate": 0.0015, | |
| "loss": 2.4763, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.14135021097046413, | |
| "grad_norm": 0.4726186990737915, | |
| "learning_rate": 0.0015, | |
| "loss": 2.476, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.14240506329113925, | |
| "grad_norm": 0.5966110825538635, | |
| "learning_rate": 0.0015, | |
| "loss": 2.4773, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.14345991561181434, | |
| "grad_norm": 0.5548987984657288, | |
| "learning_rate": 0.0015, | |
| "loss": 2.4591, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.14451476793248946, | |
| "grad_norm": 0.4708799123764038, | |
| "learning_rate": 0.0015, | |
| "loss": 2.4501, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 0.14556962025316456, | |
| "grad_norm": 0.7105334997177124, | |
| "learning_rate": 0.0015, | |
| "loss": 2.4393, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.14662447257383968, | |
| "grad_norm": 0.5323453545570374, | |
| "learning_rate": 0.0015, | |
| "loss": 2.4486, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 0.14767932489451477, | |
| "grad_norm": 0.48286688327789307, | |
| "learning_rate": 0.0015, | |
| "loss": 2.4279, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.14873417721518986, | |
| "grad_norm": 0.6947178840637207, | |
| "learning_rate": 0.0015, | |
| "loss": 2.4243, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 0.14978902953586498, | |
| "grad_norm": 0.5841905474662781, | |
| "learning_rate": 0.0015, | |
| "loss": 2.4296, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.15084388185654007, | |
| "grad_norm": 0.4542180597782135, | |
| "learning_rate": 0.0015, | |
| "loss": 2.4216, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 0.1518987341772152, | |
| "grad_norm": 0.4623643159866333, | |
| "learning_rate": 0.0015, | |
| "loss": 2.4138, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.1529535864978903, | |
| "grad_norm": 0.4974883198738098, | |
| "learning_rate": 0.0015, | |
| "loss": 2.4089, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.1540084388185654, | |
| "grad_norm": 0.46891847252845764, | |
| "learning_rate": 0.0015, | |
| "loss": 2.4007, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.1550632911392405, | |
| "grad_norm": 0.5188739895820618, | |
| "learning_rate": 0.0015, | |
| "loss": 2.3914, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.15611814345991562, | |
| "grad_norm": 0.4357692301273346, | |
| "learning_rate": 0.0015, | |
| "loss": 2.3912, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.1571729957805907, | |
| "grad_norm": 0.43899163603782654, | |
| "learning_rate": 0.0015, | |
| "loss": 2.3869, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 0.15822784810126583, | |
| "grad_norm": 0.4089808464050293, | |
| "learning_rate": 0.0015, | |
| "loss": 2.3873, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.15928270042194093, | |
| "grad_norm": 0.48731446266174316, | |
| "learning_rate": 0.0015, | |
| "loss": 2.3791, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 0.16033755274261605, | |
| "grad_norm": 0.45719215273857117, | |
| "learning_rate": 0.0015, | |
| "loss": 2.381, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 0.16139240506329114, | |
| "grad_norm": 0.4243105351924896, | |
| "learning_rate": 0.0015, | |
| "loss": 2.3586, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 0.16244725738396623, | |
| "grad_norm": 0.4893573224544525, | |
| "learning_rate": 0.0015, | |
| "loss": 2.3463, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 0.16350210970464135, | |
| "grad_norm": 0.49471256136894226, | |
| "learning_rate": 0.0015, | |
| "loss": 2.3453, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.16455696202531644, | |
| "grad_norm": 0.4878668487071991, | |
| "learning_rate": 0.0015, | |
| "loss": 2.3488, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 0.16561181434599156, | |
| "grad_norm": 0.4358295798301697, | |
| "learning_rate": 0.0015, | |
| "loss": 2.3381, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 0.16666666666666666, | |
| "grad_norm": 0.5876789689064026, | |
| "learning_rate": 0.0015, | |
| "loss": 2.3466, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 0.16772151898734178, | |
| "grad_norm": 0.6470997333526611, | |
| "learning_rate": 0.0015, | |
| "loss": 2.3396, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 0.16877637130801687, | |
| "grad_norm": 0.4370270371437073, | |
| "learning_rate": 0.0015, | |
| "loss": 2.3365, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.169831223628692, | |
| "grad_norm": 0.41289031505584717, | |
| "learning_rate": 0.0015, | |
| "loss": 2.3341, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 0.17088607594936708, | |
| "grad_norm": 0.46754589676856995, | |
| "learning_rate": 0.0015, | |
| "loss": 2.3106, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 0.1719409282700422, | |
| "grad_norm": 0.4858943819999695, | |
| "learning_rate": 0.0015, | |
| "loss": 2.3185, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 0.1729957805907173, | |
| "grad_norm": 0.7317027449607849, | |
| "learning_rate": 0.0015, | |
| "loss": 2.3148, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 0.17405063291139242, | |
| "grad_norm": 0.8671702146530151, | |
| "learning_rate": 0.0015, | |
| "loss": 2.3071, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.1751054852320675, | |
| "grad_norm": 0.5114136338233948, | |
| "learning_rate": 0.0015, | |
| "loss": 2.3042, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 0.17616033755274263, | |
| "grad_norm": 0.47680553793907166, | |
| "learning_rate": 0.0015, | |
| "loss": 2.3064, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 0.17721518987341772, | |
| "grad_norm": 0.575048565864563, | |
| "learning_rate": 0.0015, | |
| "loss": 2.296, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 0.17827004219409281, | |
| "grad_norm": 0.42194151878356934, | |
| "learning_rate": 0.0015, | |
| "loss": 2.2794, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 0.17932489451476794, | |
| "grad_norm": 0.4296809732913971, | |
| "learning_rate": 0.0015, | |
| "loss": 2.2959, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.18037974683544303, | |
| "grad_norm": 0.4321148991584778, | |
| "learning_rate": 0.0015, | |
| "loss": 2.2824, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 0.18143459915611815, | |
| "grad_norm": 0.40020737051963806, | |
| "learning_rate": 0.0015, | |
| "loss": 2.2751, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 0.18248945147679324, | |
| "grad_norm": 0.418253093957901, | |
| "learning_rate": 0.0015, | |
| "loss": 2.271, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 0.18354430379746836, | |
| "grad_norm": 0.4415479898452759, | |
| "learning_rate": 0.0015, | |
| "loss": 2.2784, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 0.18459915611814345, | |
| "grad_norm": 0.4157296121120453, | |
| "learning_rate": 0.0015, | |
| "loss": 2.2925, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.18565400843881857, | |
| "grad_norm": 0.5439921021461487, | |
| "learning_rate": 0.0015, | |
| "loss": 2.2681, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 0.18670886075949367, | |
| "grad_norm": 0.5451791882514954, | |
| "learning_rate": 0.0015, | |
| "loss": 2.2698, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 0.1877637130801688, | |
| "grad_norm": 0.3975878655910492, | |
| "learning_rate": 0.0015, | |
| "loss": 2.266, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 0.18881856540084388, | |
| "grad_norm": 0.43305620551109314, | |
| "learning_rate": 0.0015, | |
| "loss": 2.2611, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 0.189873417721519, | |
| "grad_norm": 0.3810420632362366, | |
| "learning_rate": 0.0015, | |
| "loss": 2.2613, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.1909282700421941, | |
| "grad_norm": 0.4975600838661194, | |
| "learning_rate": 0.0015, | |
| "loss": 2.2484, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 0.19198312236286919, | |
| "grad_norm": 0.4414883852005005, | |
| "learning_rate": 0.0015, | |
| "loss": 2.2422, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 0.1930379746835443, | |
| "grad_norm": 0.4077140688896179, | |
| "learning_rate": 0.0015, | |
| "loss": 2.2372, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 0.1940928270042194, | |
| "grad_norm": 0.4983878433704376, | |
| "learning_rate": 0.0015, | |
| "loss": 2.2342, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 0.19514767932489452, | |
| "grad_norm": 0.618269145488739, | |
| "learning_rate": 0.0015, | |
| "loss": 2.2358, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.1962025316455696, | |
| "grad_norm": 0.43015170097351074, | |
| "learning_rate": 0.0015, | |
| "loss": 2.2379, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 0.19725738396624473, | |
| "grad_norm": 0.6842490434646606, | |
| "learning_rate": 0.0015, | |
| "loss": 2.2386, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 0.19831223628691982, | |
| "grad_norm": 0.5212199091911316, | |
| "learning_rate": 0.0015, | |
| "loss": 2.2278, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 0.19936708860759494, | |
| "grad_norm": 0.6083729267120361, | |
| "learning_rate": 0.0015, | |
| "loss": 2.2227, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 0.20042194092827004, | |
| "grad_norm": 0.7251046299934387, | |
| "learning_rate": 0.0015, | |
| "loss": 2.2223, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.20147679324894516, | |
| "grad_norm": 0.6426624655723572, | |
| "learning_rate": 0.0015, | |
| "loss": 2.2201, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 0.20253164556962025, | |
| "grad_norm": 0.3915857672691345, | |
| "learning_rate": 0.0015, | |
| "loss": 2.2127, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 0.20358649789029537, | |
| "grad_norm": 0.6851569414138794, | |
| "learning_rate": 0.0015, | |
| "loss": 2.2181, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 0.20464135021097046, | |
| "grad_norm": 0.4217659533023834, | |
| "learning_rate": 0.0015, | |
| "loss": 2.2031, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 0.20569620253164558, | |
| "grad_norm": 0.375531941652298, | |
| "learning_rate": 0.0015, | |
| "loss": 2.2003, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.20675105485232068, | |
| "grad_norm": 0.4148560166358948, | |
| "learning_rate": 0.0015, | |
| "loss": 2.2086, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 0.20780590717299577, | |
| "grad_norm": 0.3881523311138153, | |
| "learning_rate": 0.0015, | |
| "loss": 2.208, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 0.2088607594936709, | |
| "grad_norm": 0.46804359555244446, | |
| "learning_rate": 0.0015, | |
| "loss": 2.1918, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 0.20991561181434598, | |
| "grad_norm": 0.4381694793701172, | |
| "learning_rate": 0.0015, | |
| "loss": 2.1896, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 0.2109704641350211, | |
| "grad_norm": 0.43603023886680603, | |
| "learning_rate": 0.0015, | |
| "loss": 2.1871, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.2120253164556962, | |
| "grad_norm": 0.43583643436431885, | |
| "learning_rate": 0.0015, | |
| "loss": 2.1919, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 0.21308016877637131, | |
| "grad_norm": 0.48757272958755493, | |
| "learning_rate": 0.0015, | |
| "loss": 2.1995, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 0.2141350210970464, | |
| "grad_norm": 0.4318925440311432, | |
| "learning_rate": 0.0015, | |
| "loss": 2.1852, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 0.21518987341772153, | |
| "grad_norm": 0.4255896210670471, | |
| "learning_rate": 0.0015, | |
| "loss": 2.1781, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 0.21624472573839662, | |
| "grad_norm": 0.4490632712841034, | |
| "learning_rate": 0.0015, | |
| "loss": 2.1808, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.21729957805907174, | |
| "grad_norm": 0.3975178897380829, | |
| "learning_rate": 0.0015, | |
| "loss": 2.1692, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 0.21835443037974683, | |
| "grad_norm": 0.3843049108982086, | |
| "learning_rate": 0.0015, | |
| "loss": 2.169, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 0.21940928270042195, | |
| "grad_norm": 0.42467179894447327, | |
| "learning_rate": 0.0015, | |
| "loss": 2.1744, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 0.22046413502109705, | |
| "grad_norm": 0.432029128074646, | |
| "learning_rate": 0.0015, | |
| "loss": 2.1619, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 0.22151898734177214, | |
| "grad_norm": 0.357745885848999, | |
| "learning_rate": 0.0015, | |
| "loss": 2.1675, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.22257383966244726, | |
| "grad_norm": 0.5511990189552307, | |
| "learning_rate": 0.0015, | |
| "loss": 2.1607, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 0.22362869198312235, | |
| "grad_norm": 0.44558560848236084, | |
| "learning_rate": 0.0015, | |
| "loss": 2.1569, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 0.22468354430379747, | |
| "grad_norm": 0.4494480788707733, | |
| "learning_rate": 0.0015, | |
| "loss": 2.157, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 0.22573839662447256, | |
| "grad_norm": 0.4433281719684601, | |
| "learning_rate": 0.0015, | |
| "loss": 2.15, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 0.22679324894514769, | |
| "grad_norm": 0.37084531784057617, | |
| "learning_rate": 0.0015, | |
| "loss": 2.1501, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.22784810126582278, | |
| "grad_norm": 0.45150014758110046, | |
| "learning_rate": 0.0015, | |
| "loss": 2.1628, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 0.2289029535864979, | |
| "grad_norm": 0.4242697060108185, | |
| "learning_rate": 0.0015, | |
| "loss": 2.1454, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 0.229957805907173, | |
| "grad_norm": 0.3653409481048584, | |
| "learning_rate": 0.0015, | |
| "loss": 2.1307, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 0.2310126582278481, | |
| "grad_norm": 0.41159456968307495, | |
| "learning_rate": 0.0015, | |
| "loss": 2.1424, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 0.2320675105485232, | |
| "grad_norm": 0.39554381370544434, | |
| "learning_rate": 0.0015, | |
| "loss": 2.1415, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.23312236286919832, | |
| "grad_norm": 0.41495802998542786, | |
| "learning_rate": 0.0015, | |
| "loss": 2.1285, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 0.23417721518987342, | |
| "grad_norm": 0.49099722504615784, | |
| "learning_rate": 0.0015, | |
| "loss": 2.1425, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 0.23523206751054854, | |
| "grad_norm": 0.3911990821361542, | |
| "learning_rate": 0.0015, | |
| "loss": 2.131, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 0.23628691983122363, | |
| "grad_norm": 0.42251017689704895, | |
| "learning_rate": 0.0015, | |
| "loss": 2.1258, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 0.23734177215189872, | |
| "grad_norm": 0.5026391744613647, | |
| "learning_rate": 0.0015, | |
| "loss": 2.1245, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.23839662447257384, | |
| "grad_norm": 0.38387489318847656, | |
| "learning_rate": 0.0015, | |
| "loss": 2.1174, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 0.23945147679324894, | |
| "grad_norm": 0.5047527551651001, | |
| "learning_rate": 0.0015, | |
| "loss": 2.1458, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 0.24050632911392406, | |
| "grad_norm": 0.5866949558258057, | |
| "learning_rate": 0.0015, | |
| "loss": 2.1278, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 0.24156118143459915, | |
| "grad_norm": 0.4067322313785553, | |
| "learning_rate": 0.0015, | |
| "loss": 2.1112, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 0.24261603375527427, | |
| "grad_norm": 0.4284001290798187, | |
| "learning_rate": 0.0015, | |
| "loss": 2.1127, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.24367088607594936, | |
| "grad_norm": 0.5524792671203613, | |
| "learning_rate": 0.0015, | |
| "loss": 2.1171, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 0.24472573839662448, | |
| "grad_norm": 0.4818218946456909, | |
| "learning_rate": 0.0015, | |
| "loss": 2.1173, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 0.24578059071729957, | |
| "grad_norm": 0.4074719548225403, | |
| "learning_rate": 0.0015, | |
| "loss": 2.1013, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 0.2468354430379747, | |
| "grad_norm": 0.4235811233520508, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0984, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 0.2478902953586498, | |
| "grad_norm": 0.6257115006446838, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0958, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.2489451476793249, | |
| "grad_norm": 0.6080753803253174, | |
| "learning_rate": 0.0015, | |
| "loss": 2.1054, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 0.25, | |
| "grad_norm": 0.42847368121147156, | |
| "learning_rate": 0.0015, | |
| "loss": 2.1013, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 0.2510548523206751, | |
| "grad_norm": 0.39735880494117737, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0987, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 0.2521097046413502, | |
| "grad_norm": 0.44733938574790955, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0958, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 0.25316455696202533, | |
| "grad_norm": 0.4103754758834839, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0909, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.2542194092827004, | |
| "grad_norm": 0.38021695613861084, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0945, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 0.2552742616033755, | |
| "grad_norm": 0.5292344093322754, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0937, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 0.2563291139240506, | |
| "grad_norm": 0.5250607132911682, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0874, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 0.25738396624472576, | |
| "grad_norm": 0.5082991123199463, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0963, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 0.25843881856540085, | |
| "grad_norm": 0.3772115409374237, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0782, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.25949367088607594, | |
| "grad_norm": 0.4259469509124756, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0904, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 0.26054852320675104, | |
| "grad_norm": 0.6896575689315796, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0818, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 0.2616033755274262, | |
| "grad_norm": 0.563823938369751, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0804, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 0.2626582278481013, | |
| "grad_norm": 0.40345773100852966, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0733, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 0.26371308016877637, | |
| "grad_norm": 0.4209953844547272, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0775, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.26476793248945146, | |
| "grad_norm": 0.45934030413627625, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0702, | |
| "step": 2510 | |
| }, | |
| { | |
| "epoch": 0.26582278481012656, | |
| "grad_norm": 0.5836525559425354, | |
| "learning_rate": 0.0015, | |
| "loss": 2.072, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 0.2668776371308017, | |
| "grad_norm": 0.4129602015018463, | |
| "learning_rate": 0.0015, | |
| "loss": 2.07, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 0.2679324894514768, | |
| "grad_norm": 0.5341653823852539, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0709, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 0.2689873417721519, | |
| "grad_norm": 0.3971026539802551, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0601, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.270042194092827, | |
| "grad_norm": 0.4653588831424713, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0669, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 0.27109704641350213, | |
| "grad_norm": 0.4160066545009613, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0639, | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 0.2721518987341772, | |
| "grad_norm": 0.3685113191604614, | |
| "learning_rate": 0.0015, | |
| "loss": 2.061, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 0.2732067510548523, | |
| "grad_norm": 0.43481722474098206, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0654, | |
| "step": 2590 | |
| }, | |
| { | |
| "epoch": 0.2742616033755274, | |
| "grad_norm": 0.4718077778816223, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0626, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.27531645569620256, | |
| "grad_norm": 0.43581950664520264, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0605, | |
| "step": 2610 | |
| }, | |
| { | |
| "epoch": 0.27637130801687765, | |
| "grad_norm": 0.3943164050579071, | |
| "learning_rate": 0.0015, | |
| "loss": 2.059, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 0.27742616033755274, | |
| "grad_norm": 0.4394289255142212, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0455, | |
| "step": 2630 | |
| }, | |
| { | |
| "epoch": 0.27848101265822783, | |
| "grad_norm": 0.42878061532974243, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0585, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 0.2795358649789029, | |
| "grad_norm": 0.4528396427631378, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0469, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.2805907172995781, | |
| "grad_norm": 0.4021029770374298, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0472, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 0.28164556962025317, | |
| "grad_norm": 0.46599486470222473, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0435, | |
| "step": 2670 | |
| }, | |
| { | |
| "epoch": 0.28270042194092826, | |
| "grad_norm": 0.3904265761375427, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0428, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 0.28375527426160335, | |
| "grad_norm": 0.4055798351764679, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0376, | |
| "step": 2690 | |
| }, | |
| { | |
| "epoch": 0.2848101265822785, | |
| "grad_norm": 0.41211187839508057, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0372, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.2858649789029536, | |
| "grad_norm": 0.5562853813171387, | |
| "learning_rate": 0.0015, | |
| "loss": 2.043, | |
| "step": 2710 | |
| }, | |
| { | |
| "epoch": 0.2869198312236287, | |
| "grad_norm": 0.38862988352775574, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0409, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 0.2879746835443038, | |
| "grad_norm": 0.4492087960243225, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0228, | |
| "step": 2730 | |
| }, | |
| { | |
| "epoch": 0.2890295358649789, | |
| "grad_norm": 0.8009628057479858, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0375, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 0.290084388185654, | |
| "grad_norm": 0.5228288173675537, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0364, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.2911392405063291, | |
| "grad_norm": 0.5325400233268738, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0416, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 0.2921940928270042, | |
| "grad_norm": 0.3710283935070038, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0345, | |
| "step": 2770 | |
| }, | |
| { | |
| "epoch": 0.29324894514767935, | |
| "grad_norm": 0.4476848244667053, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0345, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 0.29430379746835444, | |
| "grad_norm": 0.34534865617752075, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0275, | |
| "step": 2790 | |
| }, | |
| { | |
| "epoch": 0.29535864978902954, | |
| "grad_norm": 0.4192986786365509, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0254, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.29641350210970463, | |
| "grad_norm": 0.42185741662979126, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0255, | |
| "step": 2810 | |
| }, | |
| { | |
| "epoch": 0.2974683544303797, | |
| "grad_norm": 0.40017443895339966, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0309, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 0.29852320675105487, | |
| "grad_norm": 1.0271852016448975, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0036, | |
| "step": 2830 | |
| }, | |
| { | |
| "epoch": 0.29957805907172996, | |
| "grad_norm": 0.5703707933425903, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0065, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 0.30063291139240506, | |
| "grad_norm": 0.5895113945007324, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0161, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.30168776371308015, | |
| "grad_norm": 0.38705241680145264, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0152, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 0.3027426160337553, | |
| "grad_norm": 0.5010685920715332, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0153, | |
| "step": 2870 | |
| }, | |
| { | |
| "epoch": 0.3037974683544304, | |
| "grad_norm": 0.48778650164604187, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0003, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 0.3048523206751055, | |
| "grad_norm": 0.4624802768230438, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0039, | |
| "step": 2890 | |
| }, | |
| { | |
| "epoch": 0.3059071729957806, | |
| "grad_norm": 0.4715951979160309, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0105, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.3069620253164557, | |
| "grad_norm": 0.38887953758239746, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0085, | |
| "step": 2910 | |
| }, | |
| { | |
| "epoch": 0.3080168776371308, | |
| "grad_norm": 0.406602144241333, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0014, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 0.3090717299578059, | |
| "grad_norm": 0.4194795489311218, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9979, | |
| "step": 2930 | |
| }, | |
| { | |
| "epoch": 0.310126582278481, | |
| "grad_norm": 0.4231756031513214, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0063, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 0.3111814345991561, | |
| "grad_norm": 0.5089144706726074, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0136, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.31223628691983124, | |
| "grad_norm": 0.4470892548561096, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9944, | |
| "step": 2960 | |
| }, | |
| { | |
| "epoch": 0.31329113924050633, | |
| "grad_norm": 0.43263986706733704, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9883, | |
| "step": 2970 | |
| }, | |
| { | |
| "epoch": 0.3143459915611814, | |
| "grad_norm": 0.3462584316730499, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9908, | |
| "step": 2980 | |
| }, | |
| { | |
| "epoch": 0.3154008438818565, | |
| "grad_norm": 0.41245928406715393, | |
| "learning_rate": 0.0015, | |
| "loss": 2.0089, | |
| "step": 2990 | |
| }, | |
| { | |
| "epoch": 0.31645569620253167, | |
| "grad_norm": 0.36029165983200073, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9836, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.31751054852320676, | |
| "grad_norm": 0.41074708104133606, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9936, | |
| "step": 3010 | |
| }, | |
| { | |
| "epoch": 0.31856540084388185, | |
| "grad_norm": 0.5694265961647034, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9922, | |
| "step": 3020 | |
| }, | |
| { | |
| "epoch": 0.31962025316455694, | |
| "grad_norm": 0.4703502655029297, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9798, | |
| "step": 3030 | |
| }, | |
| { | |
| "epoch": 0.3206751054852321, | |
| "grad_norm": 0.3536970913410187, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9869, | |
| "step": 3040 | |
| }, | |
| { | |
| "epoch": 0.3217299578059072, | |
| "grad_norm": 0.3691686689853668, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9823, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.3227848101265823, | |
| "grad_norm": 0.37487566471099854, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9909, | |
| "step": 3060 | |
| }, | |
| { | |
| "epoch": 0.32383966244725737, | |
| "grad_norm": 0.3542984127998352, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9863, | |
| "step": 3070 | |
| }, | |
| { | |
| "epoch": 0.32489451476793246, | |
| "grad_norm": 0.43098732829093933, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9696, | |
| "step": 3080 | |
| }, | |
| { | |
| "epoch": 0.3259493670886076, | |
| "grad_norm": 0.4170604646205902, | |
| "learning_rate": 0.0015, | |
| "loss": 1.983, | |
| "step": 3090 | |
| }, | |
| { | |
| "epoch": 0.3270042194092827, | |
| "grad_norm": 0.5045300126075745, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9706, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.3280590717299578, | |
| "grad_norm": 0.504026472568512, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9728, | |
| "step": 3110 | |
| }, | |
| { | |
| "epoch": 0.3291139240506329, | |
| "grad_norm": 0.3979000449180603, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9751, | |
| "step": 3120 | |
| }, | |
| { | |
| "epoch": 0.33016877637130804, | |
| "grad_norm": 0.3737519681453705, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9864, | |
| "step": 3130 | |
| }, | |
| { | |
| "epoch": 0.33122362869198313, | |
| "grad_norm": 0.5472099781036377, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9619, | |
| "step": 3140 | |
| }, | |
| { | |
| "epoch": 0.3322784810126582, | |
| "grad_norm": 0.4074109196662903, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9723, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.3333333333333333, | |
| "grad_norm": 0.40785884857177734, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9678, | |
| "step": 3160 | |
| }, | |
| { | |
| "epoch": 0.33438818565400846, | |
| "grad_norm": 0.40700992941856384, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9736, | |
| "step": 3170 | |
| }, | |
| { | |
| "epoch": 0.33544303797468356, | |
| "grad_norm": 0.3770516514778137, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9664, | |
| "step": 3180 | |
| }, | |
| { | |
| "epoch": 0.33649789029535865, | |
| "grad_norm": 0.6191193461418152, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9644, | |
| "step": 3190 | |
| }, | |
| { | |
| "epoch": 0.33755274261603374, | |
| "grad_norm": 0.42728057503700256, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9606, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.33860759493670883, | |
| "grad_norm": 0.6794872283935547, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9739, | |
| "step": 3210 | |
| }, | |
| { | |
| "epoch": 0.339662447257384, | |
| "grad_norm": 0.4524136483669281, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9636, | |
| "step": 3220 | |
| }, | |
| { | |
| "epoch": 0.3407172995780591, | |
| "grad_norm": 0.6325470209121704, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9601, | |
| "step": 3230 | |
| }, | |
| { | |
| "epoch": 0.34177215189873417, | |
| "grad_norm": 0.34216609597206116, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9399, | |
| "step": 3240 | |
| }, | |
| { | |
| "epoch": 0.34282700421940926, | |
| "grad_norm": 0.39196503162384033, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9611, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.3438818565400844, | |
| "grad_norm": 0.4126991331577301, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9502, | |
| "step": 3260 | |
| }, | |
| { | |
| "epoch": 0.3449367088607595, | |
| "grad_norm": 0.38933661580085754, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9683, | |
| "step": 3270 | |
| }, | |
| { | |
| "epoch": 0.3459915611814346, | |
| "grad_norm": 0.45866304636001587, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9669, | |
| "step": 3280 | |
| }, | |
| { | |
| "epoch": 0.3470464135021097, | |
| "grad_norm": 0.6163985133171082, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9498, | |
| "step": 3290 | |
| }, | |
| { | |
| "epoch": 0.34810126582278483, | |
| "grad_norm": 0.39485007524490356, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9597, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.3491561181434599, | |
| "grad_norm": 0.3309183716773987, | |
| "learning_rate": 0.0015, | |
| "loss": 1.961, | |
| "step": 3310 | |
| }, | |
| { | |
| "epoch": 0.350210970464135, | |
| "grad_norm": 0.38850241899490356, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9595, | |
| "step": 3320 | |
| }, | |
| { | |
| "epoch": 0.3512658227848101, | |
| "grad_norm": 0.4465595781803131, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9516, | |
| "step": 3330 | |
| }, | |
| { | |
| "epoch": 0.35232067510548526, | |
| "grad_norm": 0.344111829996109, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9541, | |
| "step": 3340 | |
| }, | |
| { | |
| "epoch": 0.35337552742616035, | |
| "grad_norm": 0.40312331914901733, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9401, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 0.35443037974683544, | |
| "grad_norm": 0.3763878345489502, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9389, | |
| "step": 3360 | |
| }, | |
| { | |
| "epoch": 0.35548523206751054, | |
| "grad_norm": 0.3751490116119385, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9512, | |
| "step": 3370 | |
| }, | |
| { | |
| "epoch": 0.35654008438818563, | |
| "grad_norm": 0.3285217583179474, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9439, | |
| "step": 3380 | |
| }, | |
| { | |
| "epoch": 0.3575949367088608, | |
| "grad_norm": 0.36917683482170105, | |
| "learning_rate": 0.0015, | |
| "loss": 1.938, | |
| "step": 3390 | |
| }, | |
| { | |
| "epoch": 0.35864978902953587, | |
| "grad_norm": 0.517072319984436, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9517, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.35970464135021096, | |
| "grad_norm": 0.3615550994873047, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9434, | |
| "step": 3410 | |
| }, | |
| { | |
| "epoch": 0.36075949367088606, | |
| "grad_norm": 0.46556949615478516, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9236, | |
| "step": 3420 | |
| }, | |
| { | |
| "epoch": 0.3618143459915612, | |
| "grad_norm": 0.3592410981655121, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9376, | |
| "step": 3430 | |
| }, | |
| { | |
| "epoch": 0.3628691983122363, | |
| "grad_norm": 0.3330400288105011, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9317, | |
| "step": 3440 | |
| }, | |
| { | |
| "epoch": 0.3639240506329114, | |
| "grad_norm": 0.41959577798843384, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9335, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 0.3649789029535865, | |
| "grad_norm": 0.43094608187675476, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9254, | |
| "step": 3460 | |
| }, | |
| { | |
| "epoch": 0.36603375527426163, | |
| "grad_norm": 0.40564459562301636, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9422, | |
| "step": 3470 | |
| }, | |
| { | |
| "epoch": 0.3670886075949367, | |
| "grad_norm": 0.48330584168434143, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9441, | |
| "step": 3480 | |
| }, | |
| { | |
| "epoch": 0.3681434599156118, | |
| "grad_norm": 0.34685322642326355, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9377, | |
| "step": 3490 | |
| }, | |
| { | |
| "epoch": 0.3691983122362869, | |
| "grad_norm": 0.35264283418655396, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9185, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.370253164556962, | |
| "grad_norm": 0.4105161130428314, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9336, | |
| "step": 3510 | |
| }, | |
| { | |
| "epoch": 0.37130801687763715, | |
| "grad_norm": 0.3324391543865204, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9297, | |
| "step": 3520 | |
| }, | |
| { | |
| "epoch": 0.37236286919831224, | |
| "grad_norm": 0.43434691429138184, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9406, | |
| "step": 3530 | |
| }, | |
| { | |
| "epoch": 0.37341772151898733, | |
| "grad_norm": 0.595516562461853, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9381, | |
| "step": 3540 | |
| }, | |
| { | |
| "epoch": 0.3744725738396624, | |
| "grad_norm": 0.44567352533340454, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9256, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 0.3755274261603376, | |
| "grad_norm": 0.48279255628585815, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9252, | |
| "step": 3560 | |
| }, | |
| { | |
| "epoch": 0.37658227848101267, | |
| "grad_norm": 0.5019918084144592, | |
| "learning_rate": 0.0015, | |
| "loss": 1.923, | |
| "step": 3570 | |
| }, | |
| { | |
| "epoch": 0.37763713080168776, | |
| "grad_norm": 0.33855754137039185, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9192, | |
| "step": 3580 | |
| }, | |
| { | |
| "epoch": 0.37869198312236285, | |
| "grad_norm": 0.36431026458740234, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9305, | |
| "step": 3590 | |
| }, | |
| { | |
| "epoch": 0.379746835443038, | |
| "grad_norm": 0.4728095829486847, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9109, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.3808016877637131, | |
| "grad_norm": 0.38402220606803894, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9179, | |
| "step": 3610 | |
| }, | |
| { | |
| "epoch": 0.3818565400843882, | |
| "grad_norm": 0.3763788044452667, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9174, | |
| "step": 3620 | |
| }, | |
| { | |
| "epoch": 0.3829113924050633, | |
| "grad_norm": 0.3789425492286682, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9161, | |
| "step": 3630 | |
| }, | |
| { | |
| "epoch": 0.38396624472573837, | |
| "grad_norm": 0.4420619308948517, | |
| "learning_rate": 0.0015, | |
| "loss": 1.916, | |
| "step": 3640 | |
| }, | |
| { | |
| "epoch": 0.3850210970464135, | |
| "grad_norm": 0.47589001059532166, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9235, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 0.3860759493670886, | |
| "grad_norm": 0.5456047654151917, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9165, | |
| "step": 3660 | |
| }, | |
| { | |
| "epoch": 0.3871308016877637, | |
| "grad_norm": 0.42406487464904785, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9096, | |
| "step": 3670 | |
| }, | |
| { | |
| "epoch": 0.3881856540084388, | |
| "grad_norm": 0.45331519842147827, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9109, | |
| "step": 3680 | |
| }, | |
| { | |
| "epoch": 0.38924050632911394, | |
| "grad_norm": 0.3964917063713074, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9232, | |
| "step": 3690 | |
| }, | |
| { | |
| "epoch": 0.39029535864978904, | |
| "grad_norm": 0.4842318594455719, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9242, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.39135021097046413, | |
| "grad_norm": 0.32077768445014954, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9126, | |
| "step": 3710 | |
| }, | |
| { | |
| "epoch": 0.3924050632911392, | |
| "grad_norm": 0.38983920216560364, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8915, | |
| "step": 3720 | |
| }, | |
| { | |
| "epoch": 0.39345991561181437, | |
| "grad_norm": 0.43065422773361206, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9107, | |
| "step": 3730 | |
| }, | |
| { | |
| "epoch": 0.39451476793248946, | |
| "grad_norm": 0.3556627035140991, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9111, | |
| "step": 3740 | |
| }, | |
| { | |
| "epoch": 0.39556962025316456, | |
| "grad_norm": 0.39901190996170044, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9104, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 0.39662447257383965, | |
| "grad_norm": 0.5182483792304993, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9053, | |
| "step": 3760 | |
| }, | |
| { | |
| "epoch": 0.39767932489451474, | |
| "grad_norm": 0.38392820954322815, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9156, | |
| "step": 3770 | |
| }, | |
| { | |
| "epoch": 0.3987341772151899, | |
| "grad_norm": 0.3845175802707672, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8999, | |
| "step": 3780 | |
| }, | |
| { | |
| "epoch": 0.399789029535865, | |
| "grad_norm": 0.4916810691356659, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9056, | |
| "step": 3790 | |
| }, | |
| { | |
| "epoch": 0.4008438818565401, | |
| "grad_norm": 0.4338628351688385, | |
| "learning_rate": 0.0015, | |
| "loss": 1.908, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.40189873417721517, | |
| "grad_norm": 0.38228607177734375, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9014, | |
| "step": 3810 | |
| }, | |
| { | |
| "epoch": 0.4029535864978903, | |
| "grad_norm": 0.3408377766609192, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8864, | |
| "step": 3820 | |
| }, | |
| { | |
| "epoch": 0.4040084388185654, | |
| "grad_norm": 0.4755379557609558, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8895, | |
| "step": 3830 | |
| }, | |
| { | |
| "epoch": 0.4050632911392405, | |
| "grad_norm": 0.38737741112709045, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9104, | |
| "step": 3840 | |
| }, | |
| { | |
| "epoch": 0.4061181434599156, | |
| "grad_norm": 0.5964657068252563, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8888, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 0.40717299578059074, | |
| "grad_norm": 0.5963073968887329, | |
| "learning_rate": 0.0015, | |
| "loss": 1.9074, | |
| "step": 3860 | |
| }, | |
| { | |
| "epoch": 0.40822784810126583, | |
| "grad_norm": 0.3991047143936157, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8924, | |
| "step": 3870 | |
| }, | |
| { | |
| "epoch": 0.4092827004219409, | |
| "grad_norm": 0.3681778609752655, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8804, | |
| "step": 3880 | |
| }, | |
| { | |
| "epoch": 0.410337552742616, | |
| "grad_norm": 0.44637757539749146, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8849, | |
| "step": 3890 | |
| }, | |
| { | |
| "epoch": 0.41139240506329117, | |
| "grad_norm": 0.6122761368751526, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8876, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.41244725738396626, | |
| "grad_norm": 0.41687506437301636, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8859, | |
| "step": 3910 | |
| }, | |
| { | |
| "epoch": 0.41350210970464135, | |
| "grad_norm": 0.35810354351997375, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8841, | |
| "step": 3920 | |
| }, | |
| { | |
| "epoch": 0.41455696202531644, | |
| "grad_norm": 0.31421080231666565, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8868, | |
| "step": 3930 | |
| }, | |
| { | |
| "epoch": 0.41561181434599154, | |
| "grad_norm": 0.47998908162117004, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8815, | |
| "step": 3940 | |
| }, | |
| { | |
| "epoch": 0.4166666666666667, | |
| "grad_norm": 0.45613759756088257, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8862, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 0.4177215189873418, | |
| "grad_norm": 0.3200185298919678, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8875, | |
| "step": 3960 | |
| }, | |
| { | |
| "epoch": 0.41877637130801687, | |
| "grad_norm": 0.3921988904476166, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8733, | |
| "step": 3970 | |
| }, | |
| { | |
| "epoch": 0.41983122362869196, | |
| "grad_norm": 0.37021327018737793, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8852, | |
| "step": 3980 | |
| }, | |
| { | |
| "epoch": 0.4208860759493671, | |
| "grad_norm": 0.3745802342891693, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8753, | |
| "step": 3990 | |
| }, | |
| { | |
| "epoch": 0.4219409282700422, | |
| "grad_norm": 0.3210016191005707, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8905, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.4229957805907173, | |
| "grad_norm": 0.3362002670764923, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8762, | |
| "step": 4010 | |
| }, | |
| { | |
| "epoch": 0.4240506329113924, | |
| "grad_norm": 0.3748522698879242, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8782, | |
| "step": 4020 | |
| }, | |
| { | |
| "epoch": 0.42510548523206754, | |
| "grad_norm": 0.4886162579059601, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8709, | |
| "step": 4030 | |
| }, | |
| { | |
| "epoch": 0.42616033755274263, | |
| "grad_norm": 0.36939969658851624, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8798, | |
| "step": 4040 | |
| }, | |
| { | |
| "epoch": 0.4272151898734177, | |
| "grad_norm": 0.35972991585731506, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8793, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 0.4282700421940928, | |
| "grad_norm": 0.37946438789367676, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8841, | |
| "step": 4060 | |
| }, | |
| { | |
| "epoch": 0.4293248945147679, | |
| "grad_norm": 0.36330369114875793, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8775, | |
| "step": 4070 | |
| }, | |
| { | |
| "epoch": 0.43037974683544306, | |
| "grad_norm": 0.4276699125766754, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8858, | |
| "step": 4080 | |
| }, | |
| { | |
| "epoch": 0.43143459915611815, | |
| "grad_norm": 0.38324111700057983, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8774, | |
| "step": 4090 | |
| }, | |
| { | |
| "epoch": 0.43248945147679324, | |
| "grad_norm": 0.42404502630233765, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8764, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 0.43354430379746833, | |
| "grad_norm": 0.5729767680168152, | |
| "learning_rate": 0.0015, | |
| "loss": 1.87, | |
| "step": 4110 | |
| }, | |
| { | |
| "epoch": 0.4345991561181435, | |
| "grad_norm": 0.47413209080696106, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8695, | |
| "step": 4120 | |
| }, | |
| { | |
| "epoch": 0.4356540084388186, | |
| "grad_norm": 0.4307495355606079, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8681, | |
| "step": 4130 | |
| }, | |
| { | |
| "epoch": 0.43670886075949367, | |
| "grad_norm": 0.37268444895744324, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8679, | |
| "step": 4140 | |
| }, | |
| { | |
| "epoch": 0.43776371308016876, | |
| "grad_norm": 0.3241785168647766, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8748, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 0.4388185654008439, | |
| "grad_norm": 0.35492029786109924, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8654, | |
| "step": 4160 | |
| }, | |
| { | |
| "epoch": 0.439873417721519, | |
| "grad_norm": 0.31254345178604126, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8669, | |
| "step": 4170 | |
| }, | |
| { | |
| "epoch": 0.4409282700421941, | |
| "grad_norm": 0.3164255917072296, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8599, | |
| "step": 4180 | |
| }, | |
| { | |
| "epoch": 0.4419831223628692, | |
| "grad_norm": 0.3364178538322449, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8705, | |
| "step": 4190 | |
| }, | |
| { | |
| "epoch": 0.4430379746835443, | |
| "grad_norm": 0.4431174099445343, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8666, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.4440928270042194, | |
| "grad_norm": 0.45099061727523804, | |
| "learning_rate": 0.0015, | |
| "loss": 1.857, | |
| "step": 4210 | |
| }, | |
| { | |
| "epoch": 0.4451476793248945, | |
| "grad_norm": 0.32387441396713257, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8637, | |
| "step": 4220 | |
| }, | |
| { | |
| "epoch": 0.4462025316455696, | |
| "grad_norm": 0.3952908515930176, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8636, | |
| "step": 4230 | |
| }, | |
| { | |
| "epoch": 0.4472573839662447, | |
| "grad_norm": 0.38458308577537537, | |
| "learning_rate": 0.0015, | |
| "loss": 1.864, | |
| "step": 4240 | |
| }, | |
| { | |
| "epoch": 0.44831223628691985, | |
| "grad_norm": 0.40344688296318054, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8651, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 0.44936708860759494, | |
| "grad_norm": 0.38940712809562683, | |
| "learning_rate": 0.0015, | |
| "loss": 1.869, | |
| "step": 4260 | |
| }, | |
| { | |
| "epoch": 0.45042194092827004, | |
| "grad_norm": 0.39709579944610596, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8588, | |
| "step": 4270 | |
| }, | |
| { | |
| "epoch": 0.45147679324894513, | |
| "grad_norm": 0.331944078207016, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8519, | |
| "step": 4280 | |
| }, | |
| { | |
| "epoch": 0.4525316455696203, | |
| "grad_norm": 0.33285272121429443, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8629, | |
| "step": 4290 | |
| }, | |
| { | |
| "epoch": 0.45358649789029537, | |
| "grad_norm": 0.4516183137893677, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8609, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 0.45464135021097046, | |
| "grad_norm": 0.34201303124427795, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8579, | |
| "step": 4310 | |
| }, | |
| { | |
| "epoch": 0.45569620253164556, | |
| "grad_norm": 0.4359893202781677, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8483, | |
| "step": 4320 | |
| }, | |
| { | |
| "epoch": 0.45675105485232065, | |
| "grad_norm": 0.3476506769657135, | |
| "learning_rate": 0.0015, | |
| "loss": 1.859, | |
| "step": 4330 | |
| }, | |
| { | |
| "epoch": 0.4578059071729958, | |
| "grad_norm": 0.43458154797554016, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8486, | |
| "step": 4340 | |
| }, | |
| { | |
| "epoch": 0.4588607594936709, | |
| "grad_norm": 0.36237064003944397, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8542, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 0.459915611814346, | |
| "grad_norm": 0.38242867588996887, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8674, | |
| "step": 4360 | |
| }, | |
| { | |
| "epoch": 0.4609704641350211, | |
| "grad_norm": 0.3427712321281433, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8416, | |
| "step": 4370 | |
| }, | |
| { | |
| "epoch": 0.4620253164556962, | |
| "grad_norm": 0.4015636742115021, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8489, | |
| "step": 4380 | |
| }, | |
| { | |
| "epoch": 0.4630801687763713, | |
| "grad_norm": 0.34853026270866394, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8414, | |
| "step": 4390 | |
| }, | |
| { | |
| "epoch": 0.4641350210970464, | |
| "grad_norm": 0.5962074995040894, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8505, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.4651898734177215, | |
| "grad_norm": 0.5702231526374817, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8636, | |
| "step": 4410 | |
| }, | |
| { | |
| "epoch": 0.46624472573839665, | |
| "grad_norm": 0.3360097408294678, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8478, | |
| "step": 4420 | |
| }, | |
| { | |
| "epoch": 0.46729957805907174, | |
| "grad_norm": 0.5092244744300842, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8496, | |
| "step": 4430 | |
| }, | |
| { | |
| "epoch": 0.46835443037974683, | |
| "grad_norm": 0.3331068456172943, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8475, | |
| "step": 4440 | |
| }, | |
| { | |
| "epoch": 0.4694092827004219, | |
| "grad_norm": 0.38931623101234436, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8511, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 0.4704641350210971, | |
| "grad_norm": 0.3346966505050659, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8379, | |
| "step": 4460 | |
| }, | |
| { | |
| "epoch": 0.47151898734177217, | |
| "grad_norm": 0.3365168273448944, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8518, | |
| "step": 4470 | |
| }, | |
| { | |
| "epoch": 0.47257383966244726, | |
| "grad_norm": 0.3916284739971161, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8553, | |
| "step": 4480 | |
| }, | |
| { | |
| "epoch": 0.47362869198312235, | |
| "grad_norm": 0.42281103134155273, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8427, | |
| "step": 4490 | |
| }, | |
| { | |
| "epoch": 0.47468354430379744, | |
| "grad_norm": 0.3276952803134918, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8412, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.4757383966244726, | |
| "grad_norm": 0.3229634463787079, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8458, | |
| "step": 4510 | |
| }, | |
| { | |
| "epoch": 0.4767932489451477, | |
| "grad_norm": 0.41583219170570374, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8278, | |
| "step": 4520 | |
| }, | |
| { | |
| "epoch": 0.4778481012658228, | |
| "grad_norm": 0.42357075214385986, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8453, | |
| "step": 4530 | |
| }, | |
| { | |
| "epoch": 0.47890295358649787, | |
| "grad_norm": 0.3458639085292816, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8467, | |
| "step": 4540 | |
| }, | |
| { | |
| "epoch": 0.479957805907173, | |
| "grad_norm": 0.35474738478660583, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8425, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 0.4810126582278481, | |
| "grad_norm": 0.4535824656486511, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8337, | |
| "step": 4560 | |
| }, | |
| { | |
| "epoch": 0.4820675105485232, | |
| "grad_norm": 0.4138074815273285, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8174, | |
| "step": 4570 | |
| }, | |
| { | |
| "epoch": 0.4831223628691983, | |
| "grad_norm": 0.4137820601463318, | |
| "learning_rate": 0.0015, | |
| "loss": 1.841, | |
| "step": 4580 | |
| }, | |
| { | |
| "epoch": 0.48417721518987344, | |
| "grad_norm": 0.39168211817741394, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8253, | |
| "step": 4590 | |
| }, | |
| { | |
| "epoch": 0.48523206751054854, | |
| "grad_norm": 0.5041007399559021, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8431, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.48628691983122363, | |
| "grad_norm": 0.5302157998085022, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8431, | |
| "step": 4610 | |
| }, | |
| { | |
| "epoch": 0.4873417721518987, | |
| "grad_norm": 0.44530072808265686, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8221, | |
| "step": 4620 | |
| }, | |
| { | |
| "epoch": 0.4883966244725738, | |
| "grad_norm": 0.366564005613327, | |
| "learning_rate": 0.0015, | |
| "loss": 1.84, | |
| "step": 4630 | |
| }, | |
| { | |
| "epoch": 0.48945147679324896, | |
| "grad_norm": 0.39872315526008606, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8403, | |
| "step": 4640 | |
| }, | |
| { | |
| "epoch": 0.49050632911392406, | |
| "grad_norm": 0.35998818278312683, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8392, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 0.49156118143459915, | |
| "grad_norm": 0.3405250608921051, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8291, | |
| "step": 4660 | |
| }, | |
| { | |
| "epoch": 0.49261603375527424, | |
| "grad_norm": 0.406040221452713, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8288, | |
| "step": 4670 | |
| }, | |
| { | |
| "epoch": 0.4936708860759494, | |
| "grad_norm": 0.3696703314781189, | |
| "learning_rate": 0.0015, | |
| "loss": 1.826, | |
| "step": 4680 | |
| }, | |
| { | |
| "epoch": 0.4947257383966245, | |
| "grad_norm": 0.4865450859069824, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8207, | |
| "step": 4690 | |
| }, | |
| { | |
| "epoch": 0.4957805907172996, | |
| "grad_norm": 0.3739243745803833, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8354, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 0.49683544303797467, | |
| "grad_norm": 0.3547900319099426, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8347, | |
| "step": 4710 | |
| }, | |
| { | |
| "epoch": 0.4978902953586498, | |
| "grad_norm": 0.4278034567832947, | |
| "learning_rate": 0.0015, | |
| "loss": 1.828, | |
| "step": 4720 | |
| }, | |
| { | |
| "epoch": 0.4989451476793249, | |
| "grad_norm": 0.32390764355659485, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8265, | |
| "step": 4730 | |
| }, | |
| { | |
| "epoch": 0.5, | |
| "grad_norm": 0.3499670624732971, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8229, | |
| "step": 4740 | |
| }, | |
| { | |
| "epoch": 0.5010548523206751, | |
| "grad_norm": 0.355530709028244, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8397, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 0.5021097046413502, | |
| "grad_norm": 0.3372103273868561, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8254, | |
| "step": 4760 | |
| }, | |
| { | |
| "epoch": 0.5031645569620253, | |
| "grad_norm": 0.48468029499053955, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8223, | |
| "step": 4770 | |
| }, | |
| { | |
| "epoch": 0.5042194092827004, | |
| "grad_norm": 0.4034891724586487, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8334, | |
| "step": 4780 | |
| }, | |
| { | |
| "epoch": 0.5052742616033755, | |
| "grad_norm": 0.4861879348754883, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8166, | |
| "step": 4790 | |
| }, | |
| { | |
| "epoch": 0.5063291139240507, | |
| "grad_norm": 0.3496376872062683, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8208, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.5073839662447257, | |
| "grad_norm": 0.3765081763267517, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8206, | |
| "step": 4810 | |
| }, | |
| { | |
| "epoch": 0.5084388185654009, | |
| "grad_norm": 0.46853533387184143, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8404, | |
| "step": 4820 | |
| }, | |
| { | |
| "epoch": 0.509493670886076, | |
| "grad_norm": 0.3622114658355713, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8199, | |
| "step": 4830 | |
| }, | |
| { | |
| "epoch": 0.510548523206751, | |
| "grad_norm": 0.40379026532173157, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8145, | |
| "step": 4840 | |
| }, | |
| { | |
| "epoch": 0.5116033755274262, | |
| "grad_norm": 0.3473694324493408, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8296, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 0.5126582278481012, | |
| "grad_norm": 0.3614269196987152, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8146, | |
| "step": 4860 | |
| }, | |
| { | |
| "epoch": 0.5137130801687764, | |
| "grad_norm": 0.37687039375305176, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8068, | |
| "step": 4870 | |
| }, | |
| { | |
| "epoch": 0.5147679324894515, | |
| "grad_norm": 0.3430388271808624, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8276, | |
| "step": 4880 | |
| }, | |
| { | |
| "epoch": 0.5158227848101266, | |
| "grad_norm": 0.3632057011127472, | |
| "learning_rate": 0.0015, | |
| "loss": 1.819, | |
| "step": 4890 | |
| }, | |
| { | |
| "epoch": 0.5168776371308017, | |
| "grad_norm": 0.32978373765945435, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8256, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 0.5179324894514767, | |
| "grad_norm": 0.3529827296733856, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8186, | |
| "step": 4910 | |
| }, | |
| { | |
| "epoch": 0.5189873417721519, | |
| "grad_norm": 0.37453562021255493, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8082, | |
| "step": 4920 | |
| }, | |
| { | |
| "epoch": 0.520042194092827, | |
| "grad_norm": 0.4753303527832031, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8193, | |
| "step": 4930 | |
| }, | |
| { | |
| "epoch": 0.5210970464135021, | |
| "grad_norm": 0.3357962369918823, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8147, | |
| "step": 4940 | |
| }, | |
| { | |
| "epoch": 0.5221518987341772, | |
| "grad_norm": 0.34558114409446716, | |
| "learning_rate": 0.0015, | |
| "loss": 1.823, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 0.5232067510548524, | |
| "grad_norm": 0.2897375822067261, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8025, | |
| "step": 4960 | |
| }, | |
| { | |
| "epoch": 0.5242616033755274, | |
| "grad_norm": 0.4054597318172455, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8128, | |
| "step": 4970 | |
| }, | |
| { | |
| "epoch": 0.5253164556962026, | |
| "grad_norm": 0.36653947830200195, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8072, | |
| "step": 4980 | |
| }, | |
| { | |
| "epoch": 0.5263713080168776, | |
| "grad_norm": 0.3881050944328308, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8086, | |
| "step": 4990 | |
| }, | |
| { | |
| "epoch": 0.5274261603375527, | |
| "grad_norm": 0.35539186000823975, | |
| "learning_rate": 0.0015, | |
| "loss": 1.822, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.5284810126582279, | |
| "grad_norm": 0.3445149064064026, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8106, | |
| "step": 5010 | |
| }, | |
| { | |
| "epoch": 0.5295358649789029, | |
| "grad_norm": 0.3262072801589966, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8036, | |
| "step": 5020 | |
| }, | |
| { | |
| "epoch": 0.5305907172995781, | |
| "grad_norm": 0.3699089288711548, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7894, | |
| "step": 5030 | |
| }, | |
| { | |
| "epoch": 0.5316455696202531, | |
| "grad_norm": 0.3256937861442566, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8152, | |
| "step": 5040 | |
| }, | |
| { | |
| "epoch": 0.5327004219409283, | |
| "grad_norm": 0.4292232096195221, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8022, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 0.5337552742616034, | |
| "grad_norm": 0.3788437247276306, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8171, | |
| "step": 5060 | |
| }, | |
| { | |
| "epoch": 0.5348101265822784, | |
| "grad_norm": 0.3824426829814911, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8043, | |
| "step": 5070 | |
| }, | |
| { | |
| "epoch": 0.5358649789029536, | |
| "grad_norm": 0.43619370460510254, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8192, | |
| "step": 5080 | |
| }, | |
| { | |
| "epoch": 0.5369198312236287, | |
| "grad_norm": 0.4108010232448578, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8107, | |
| "step": 5090 | |
| }, | |
| { | |
| "epoch": 0.5379746835443038, | |
| "grad_norm": 0.36239510774612427, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7934, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 0.5390295358649789, | |
| "grad_norm": 0.40039804577827454, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8024, | |
| "step": 5110 | |
| }, | |
| { | |
| "epoch": 0.540084388185654, | |
| "grad_norm": 0.39933285117149353, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8031, | |
| "step": 5120 | |
| }, | |
| { | |
| "epoch": 0.5411392405063291, | |
| "grad_norm": 0.3655042350292206, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7959, | |
| "step": 5130 | |
| }, | |
| { | |
| "epoch": 0.5421940928270043, | |
| "grad_norm": 0.4802871644496918, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8141, | |
| "step": 5140 | |
| }, | |
| { | |
| "epoch": 0.5432489451476793, | |
| "grad_norm": 0.3769666850566864, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8118, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 0.5443037974683544, | |
| "grad_norm": 0.3966495096683502, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7982, | |
| "step": 5160 | |
| }, | |
| { | |
| "epoch": 0.5453586497890295, | |
| "grad_norm": 0.35799336433410645, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8021, | |
| "step": 5170 | |
| }, | |
| { | |
| "epoch": 0.5464135021097046, | |
| "grad_norm": 0.3347603380680084, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8138, | |
| "step": 5180 | |
| }, | |
| { | |
| "epoch": 0.5474683544303798, | |
| "grad_norm": 0.4724527597427368, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7948, | |
| "step": 5190 | |
| }, | |
| { | |
| "epoch": 0.5485232067510548, | |
| "grad_norm": 0.422674298286438, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8081, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 0.54957805907173, | |
| "grad_norm": 0.36212271451950073, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7883, | |
| "step": 5210 | |
| }, | |
| { | |
| "epoch": 0.5506329113924051, | |
| "grad_norm": 0.37394046783447266, | |
| "learning_rate": 0.0015, | |
| "loss": 1.803, | |
| "step": 5220 | |
| }, | |
| { | |
| "epoch": 0.5516877637130801, | |
| "grad_norm": 0.391477108001709, | |
| "learning_rate": 0.0015, | |
| "loss": 1.794, | |
| "step": 5230 | |
| }, | |
| { | |
| "epoch": 0.5527426160337553, | |
| "grad_norm": 0.33054593205451965, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8004, | |
| "step": 5240 | |
| }, | |
| { | |
| "epoch": 0.5537974683544303, | |
| "grad_norm": 0.5764470100402832, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7946, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 0.5548523206751055, | |
| "grad_norm": 0.32538744807243347, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7988, | |
| "step": 5260 | |
| }, | |
| { | |
| "epoch": 0.5559071729957806, | |
| "grad_norm": 0.4617023468017578, | |
| "learning_rate": 0.0015, | |
| "loss": 1.8032, | |
| "step": 5270 | |
| }, | |
| { | |
| "epoch": 0.5569620253164557, | |
| "grad_norm": 0.5663644075393677, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7951, | |
| "step": 5280 | |
| }, | |
| { | |
| "epoch": 0.5580168776371308, | |
| "grad_norm": 0.34853243827819824, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7936, | |
| "step": 5290 | |
| }, | |
| { | |
| "epoch": 0.5590717299578059, | |
| "grad_norm": 0.3220527470111847, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7991, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 0.560126582278481, | |
| "grad_norm": 0.48532089591026306, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7878, | |
| "step": 5310 | |
| }, | |
| { | |
| "epoch": 0.5611814345991561, | |
| "grad_norm": 0.40416234731674194, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7866, | |
| "step": 5320 | |
| }, | |
| { | |
| "epoch": 0.5622362869198312, | |
| "grad_norm": 0.3604409098625183, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7975, | |
| "step": 5330 | |
| }, | |
| { | |
| "epoch": 0.5632911392405063, | |
| "grad_norm": 0.3480297327041626, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7922, | |
| "step": 5340 | |
| }, | |
| { | |
| "epoch": 0.5643459915611815, | |
| "grad_norm": 0.36295345425605774, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7846, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 0.5654008438818565, | |
| "grad_norm": 0.43485522270202637, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7904, | |
| "step": 5360 | |
| }, | |
| { | |
| "epoch": 0.5664556962025317, | |
| "grad_norm": 0.3537534773349762, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7939, | |
| "step": 5370 | |
| }, | |
| { | |
| "epoch": 0.5675105485232067, | |
| "grad_norm": 0.43174123764038086, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7833, | |
| "step": 5380 | |
| }, | |
| { | |
| "epoch": 0.5685654008438819, | |
| "grad_norm": 0.5191883444786072, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7866, | |
| "step": 5390 | |
| }, | |
| { | |
| "epoch": 0.569620253164557, | |
| "grad_norm": 0.3696303367614746, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7859, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 0.570675105485232, | |
| "grad_norm": 0.3290150761604309, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7854, | |
| "step": 5410 | |
| }, | |
| { | |
| "epoch": 0.5717299578059072, | |
| "grad_norm": 0.4146673083305359, | |
| "learning_rate": 0.0015, | |
| "loss": 1.783, | |
| "step": 5420 | |
| }, | |
| { | |
| "epoch": 0.5727848101265823, | |
| "grad_norm": 0.3947039544582367, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7813, | |
| "step": 5430 | |
| }, | |
| { | |
| "epoch": 0.5738396624472574, | |
| "grad_norm": 0.3319171071052551, | |
| "learning_rate": 0.0015, | |
| "loss": 1.781, | |
| "step": 5440 | |
| }, | |
| { | |
| "epoch": 0.5748945147679325, | |
| "grad_norm": 0.33405378460884094, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7876, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 0.5759493670886076, | |
| "grad_norm": 0.32065099477767944, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7869, | |
| "step": 5460 | |
| }, | |
| { | |
| "epoch": 0.5770042194092827, | |
| "grad_norm": 0.3278934359550476, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7825, | |
| "step": 5470 | |
| }, | |
| { | |
| "epoch": 0.5780590717299579, | |
| "grad_norm": 0.4991150200366974, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7875, | |
| "step": 5480 | |
| }, | |
| { | |
| "epoch": 0.5791139240506329, | |
| "grad_norm": 0.3259681463241577, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7879, | |
| "step": 5490 | |
| }, | |
| { | |
| "epoch": 0.580168776371308, | |
| "grad_norm": 0.39956098794937134, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7816, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.5812236286919831, | |
| "grad_norm": 0.40146827697753906, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7878, | |
| "step": 5510 | |
| }, | |
| { | |
| "epoch": 0.5822784810126582, | |
| "grad_norm": 0.3431818187236786, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7805, | |
| "step": 5520 | |
| }, | |
| { | |
| "epoch": 0.5833333333333334, | |
| "grad_norm": 0.4219339191913605, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7761, | |
| "step": 5530 | |
| }, | |
| { | |
| "epoch": 0.5843881856540084, | |
| "grad_norm": 0.3600873649120331, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7715, | |
| "step": 5540 | |
| }, | |
| { | |
| "epoch": 0.5854430379746836, | |
| "grad_norm": 0.29148736596107483, | |
| "learning_rate": 0.0015, | |
| "loss": 1.789, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 0.5864978902953587, | |
| "grad_norm": 0.3306005001068115, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7917, | |
| "step": 5560 | |
| }, | |
| { | |
| "epoch": 0.5875527426160337, | |
| "grad_norm": 0.3504825532436371, | |
| "learning_rate": 0.0015, | |
| "loss": 1.777, | |
| "step": 5570 | |
| }, | |
| { | |
| "epoch": 0.5886075949367089, | |
| "grad_norm": 0.3031824231147766, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7768, | |
| "step": 5580 | |
| }, | |
| { | |
| "epoch": 0.5896624472573839, | |
| "grad_norm": 0.4476396143436432, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7805, | |
| "step": 5590 | |
| }, | |
| { | |
| "epoch": 0.5907172995780591, | |
| "grad_norm": 0.4386860728263855, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7751, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 0.5917721518987342, | |
| "grad_norm": 0.3248020112514496, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7773, | |
| "step": 5610 | |
| }, | |
| { | |
| "epoch": 0.5928270042194093, | |
| "grad_norm": 0.35110512375831604, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7883, | |
| "step": 5620 | |
| }, | |
| { | |
| "epoch": 0.5938818565400844, | |
| "grad_norm": 0.5061001777648926, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7785, | |
| "step": 5630 | |
| }, | |
| { | |
| "epoch": 0.5949367088607594, | |
| "grad_norm": 0.39553189277648926, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7914, | |
| "step": 5640 | |
| }, | |
| { | |
| "epoch": 0.5959915611814346, | |
| "grad_norm": 0.2968481183052063, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7868, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 0.5970464135021097, | |
| "grad_norm": 0.3996788263320923, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7782, | |
| "step": 5660 | |
| }, | |
| { | |
| "epoch": 0.5981012658227848, | |
| "grad_norm": 0.3071383535861969, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7863, | |
| "step": 5670 | |
| }, | |
| { | |
| "epoch": 0.5991561181434599, | |
| "grad_norm": 0.4471006393432617, | |
| "learning_rate": 0.0015, | |
| "loss": 1.784, | |
| "step": 5680 | |
| }, | |
| { | |
| "epoch": 0.6002109704641351, | |
| "grad_norm": 0.33473655581474304, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7667, | |
| "step": 5690 | |
| }, | |
| { | |
| "epoch": 0.6012658227848101, | |
| "grad_norm": 0.36823636293411255, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7688, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 0.6023206751054853, | |
| "grad_norm": 0.39860105514526367, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7709, | |
| "step": 5710 | |
| }, | |
| { | |
| "epoch": 0.6033755274261603, | |
| "grad_norm": 0.3454592227935791, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7741, | |
| "step": 5720 | |
| }, | |
| { | |
| "epoch": 0.6044303797468354, | |
| "grad_norm": 0.37099096179008484, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7725, | |
| "step": 5730 | |
| }, | |
| { | |
| "epoch": 0.6054852320675106, | |
| "grad_norm": 0.3275086581707001, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7738, | |
| "step": 5740 | |
| }, | |
| { | |
| "epoch": 0.6065400843881856, | |
| "grad_norm": 0.4156765639781952, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7455, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 0.6075949367088608, | |
| "grad_norm": 0.3556298613548279, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7578, | |
| "step": 5760 | |
| }, | |
| { | |
| "epoch": 0.6086497890295358, | |
| "grad_norm": 0.3785768747329712, | |
| "learning_rate": 0.0015, | |
| "loss": 1.76, | |
| "step": 5770 | |
| }, | |
| { | |
| "epoch": 0.609704641350211, | |
| "grad_norm": 0.3101571202278137, | |
| "learning_rate": 0.0015, | |
| "loss": 1.766, | |
| "step": 5780 | |
| }, | |
| { | |
| "epoch": 0.6107594936708861, | |
| "grad_norm": 0.4646480977535248, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7609, | |
| "step": 5790 | |
| }, | |
| { | |
| "epoch": 0.6118143459915611, | |
| "grad_norm": 0.39919760823249817, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7695, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 0.6128691983122363, | |
| "grad_norm": 0.37033745646476746, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7716, | |
| "step": 5810 | |
| }, | |
| { | |
| "epoch": 0.6139240506329114, | |
| "grad_norm": 0.3841451406478882, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7837, | |
| "step": 5820 | |
| }, | |
| { | |
| "epoch": 0.6149789029535865, | |
| "grad_norm": 0.3592260777950287, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7632, | |
| "step": 5830 | |
| }, | |
| { | |
| "epoch": 0.6160337552742616, | |
| "grad_norm": 0.3078693747520447, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7792, | |
| "step": 5840 | |
| }, | |
| { | |
| "epoch": 0.6170886075949367, | |
| "grad_norm": 0.313323974609375, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7834, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 0.6181434599156118, | |
| "grad_norm": 0.3674357235431671, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7706, | |
| "step": 5860 | |
| }, | |
| { | |
| "epoch": 0.619198312236287, | |
| "grad_norm": 0.43470507860183716, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7627, | |
| "step": 5870 | |
| }, | |
| { | |
| "epoch": 0.620253164556962, | |
| "grad_norm": 0.32323190569877625, | |
| "learning_rate": 0.0015, | |
| "loss": 1.772, | |
| "step": 5880 | |
| }, | |
| { | |
| "epoch": 0.6213080168776371, | |
| "grad_norm": 0.4054836928844452, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7626, | |
| "step": 5890 | |
| }, | |
| { | |
| "epoch": 0.6223628691983122, | |
| "grad_norm": 0.3744523525238037, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7659, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 0.6234177215189873, | |
| "grad_norm": 0.4974427819252014, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7611, | |
| "step": 5910 | |
| }, | |
| { | |
| "epoch": 0.6244725738396625, | |
| "grad_norm": 0.3532814383506775, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7569, | |
| "step": 5920 | |
| }, | |
| { | |
| "epoch": 0.6255274261603375, | |
| "grad_norm": 0.4499043822288513, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7626, | |
| "step": 5930 | |
| }, | |
| { | |
| "epoch": 0.6265822784810127, | |
| "grad_norm": 0.34281599521636963, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7682, | |
| "step": 5940 | |
| }, | |
| { | |
| "epoch": 0.6276371308016878, | |
| "grad_norm": 0.37006106972694397, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7558, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 0.6286919831223629, | |
| "grad_norm": 0.37030336260795593, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7623, | |
| "step": 5960 | |
| }, | |
| { | |
| "epoch": 0.629746835443038, | |
| "grad_norm": 0.381357342004776, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7642, | |
| "step": 5970 | |
| }, | |
| { | |
| "epoch": 0.630801687763713, | |
| "grad_norm": 0.3766602575778961, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7486, | |
| "step": 5980 | |
| }, | |
| { | |
| "epoch": 0.6318565400843882, | |
| "grad_norm": 0.32820066809654236, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7666, | |
| "step": 5990 | |
| }, | |
| { | |
| "epoch": 0.6329113924050633, | |
| "grad_norm": 0.3488951623439789, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7731, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.6339662447257384, | |
| "grad_norm": 0.3223486542701721, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7626, | |
| "step": 6010 | |
| }, | |
| { | |
| "epoch": 0.6350210970464135, | |
| "grad_norm": 0.3566620349884033, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7668, | |
| "step": 6020 | |
| }, | |
| { | |
| "epoch": 0.6360759493670886, | |
| "grad_norm": 0.33811134099960327, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7483, | |
| "step": 6030 | |
| }, | |
| { | |
| "epoch": 0.6371308016877637, | |
| "grad_norm": 0.44937440752983093, | |
| "learning_rate": 0.0015, | |
| "loss": 1.765, | |
| "step": 6040 | |
| }, | |
| { | |
| "epoch": 0.6381856540084389, | |
| "grad_norm": 0.36752060055732727, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7468, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 0.6392405063291139, | |
| "grad_norm": 0.31580060720443726, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7644, | |
| "step": 6060 | |
| }, | |
| { | |
| "epoch": 0.640295358649789, | |
| "grad_norm": 0.4360526502132416, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7551, | |
| "step": 6070 | |
| }, | |
| { | |
| "epoch": 0.6413502109704642, | |
| "grad_norm": 0.32985156774520874, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7556, | |
| "step": 6080 | |
| }, | |
| { | |
| "epoch": 0.6424050632911392, | |
| "grad_norm": 0.32723644375801086, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7575, | |
| "step": 6090 | |
| }, | |
| { | |
| "epoch": 0.6434599156118144, | |
| "grad_norm": 0.49019789695739746, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7477, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 0.6445147679324894, | |
| "grad_norm": 0.3593115508556366, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7604, | |
| "step": 6110 | |
| }, | |
| { | |
| "epoch": 0.6455696202531646, | |
| "grad_norm": 0.3175909221172333, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7563, | |
| "step": 6120 | |
| }, | |
| { | |
| "epoch": 0.6466244725738397, | |
| "grad_norm": 0.384500116109848, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7514, | |
| "step": 6130 | |
| }, | |
| { | |
| "epoch": 0.6476793248945147, | |
| "grad_norm": 0.424191951751709, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7558, | |
| "step": 6140 | |
| }, | |
| { | |
| "epoch": 0.6487341772151899, | |
| "grad_norm": 0.396973192691803, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7589, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 0.6497890295358649, | |
| "grad_norm": 0.32909175753593445, | |
| "learning_rate": 0.0015, | |
| "loss": 1.7585, | |
| "step": 6160 | |
| }, | |
| { | |
| "epoch": 0.6508438818565401, | |
| "grad_norm": 0.35591256618499756, | |
| "learning_rate": 0.0014834368975312174, | |
| "loss": 1.7344, | |
| "step": 6170 | |
| }, | |
| { | |
| "epoch": 0.6518987341772152, | |
| "grad_norm": 0.5259692072868347, | |
| "learning_rate": 0.0014629899726345957, | |
| "loss": 1.7568, | |
| "step": 6180 | |
| }, | |
| { | |
| "epoch": 0.6529535864978903, | |
| "grad_norm": 0.3114689588546753, | |
| "learning_rate": 0.0014428248775471316, | |
| "loss": 1.7578, | |
| "step": 6190 | |
| }, | |
| { | |
| "epoch": 0.6540084388185654, | |
| "grad_norm": 0.5166715383529663, | |
| "learning_rate": 0.00142293772767289, | |
| "loss": 1.75, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 0.6550632911392406, | |
| "grad_norm": 0.39521825313568115, | |
| "learning_rate": 0.001403324691959192, | |
| "loss": 1.7453, | |
| "step": 6210 | |
| }, | |
| { | |
| "epoch": 0.6561181434599156, | |
| "grad_norm": 0.35881662368774414, | |
| "learning_rate": 0.0013839819921586025, | |
| "loss": 1.7513, | |
| "step": 6220 | |
| }, | |
| { | |
| "epoch": 0.6571729957805907, | |
| "grad_norm": 0.30511415004730225, | |
| "learning_rate": 0.0013649059021010894, | |
| "loss": 1.741, | |
| "step": 6230 | |
| }, | |
| { | |
| "epoch": 0.6582278481012658, | |
| "grad_norm": 0.4092840254306793, | |
| "learning_rate": 0.0013460927469762154, | |
| "loss": 1.7434, | |
| "step": 6240 | |
| }, | |
| { | |
| "epoch": 0.6592827004219409, | |
| "grad_norm": 0.35713300108909607, | |
| "learning_rate": 0.0013275389026252255, | |
| "loss": 1.7531, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 0.6603375527426161, | |
| "grad_norm": 0.3668721318244934, | |
| "learning_rate": 0.0013092407948428887, | |
| "loss": 1.7419, | |
| "step": 6260 | |
| }, | |
| { | |
| "epoch": 0.6613924050632911, | |
| "grad_norm": 0.33162111043930054, | |
| "learning_rate": 0.001291194898688966, | |
| "loss": 1.7525, | |
| "step": 6270 | |
| }, | |
| { | |
| "epoch": 0.6624472573839663, | |
| "grad_norm": 0.4238397181034088, | |
| "learning_rate": 0.001273397737809166, | |
| "loss": 1.7486, | |
| "step": 6280 | |
| }, | |
| { | |
| "epoch": 0.6635021097046413, | |
| "grad_norm": 0.3315000534057617, | |
| "learning_rate": 0.001255845883765463, | |
| "loss": 1.7426, | |
| "step": 6290 | |
| }, | |
| { | |
| "epoch": 0.6645569620253164, | |
| "grad_norm": 0.40429437160491943, | |
| "learning_rate": 0.001238535955375642, | |
| "loss": 1.736, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 0.6656118143459916, | |
| "grad_norm": 0.3680913746356964, | |
| "learning_rate": 0.0012214646180619506, | |
| "loss": 1.7336, | |
| "step": 6310 | |
| }, | |
| { | |
| "epoch": 0.6666666666666666, | |
| "grad_norm": 0.3675517737865448, | |
| "learning_rate": 0.001204628583208727, | |
| "loss": 1.7308, | |
| "step": 6320 | |
| }, | |
| { | |
| "epoch": 0.6677215189873418, | |
| "grad_norm": 0.3261108100414276, | |
| "learning_rate": 0.0011880246075288827, | |
| "loss": 1.7354, | |
| "step": 6330 | |
| }, | |
| { | |
| "epoch": 0.6687763713080169, | |
| "grad_norm": 0.4246390759944916, | |
| "learning_rate": 0.001171649492439115, | |
| "loss": 1.7295, | |
| "step": 6340 | |
| }, | |
| { | |
| "epoch": 0.669831223628692, | |
| "grad_norm": 0.316724956035614, | |
| "learning_rate": 0.0011555000834437364, | |
| "loss": 1.7349, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 0.6708860759493671, | |
| "grad_norm": 0.3159819543361664, | |
| "learning_rate": 0.0011395732695269908, | |
| "loss": 1.7302, | |
| "step": 6360 | |
| }, | |
| { | |
| "epoch": 0.6719409282700421, | |
| "grad_norm": 0.31476807594299316, | |
| "learning_rate": 0.0011238659825537505, | |
| "loss": 1.7162, | |
| "step": 6370 | |
| }, | |
| { | |
| "epoch": 0.6729957805907173, | |
| "grad_norm": 0.31384891271591187, | |
| "learning_rate": 0.0011083751966784717, | |
| "loss": 1.7187, | |
| "step": 6380 | |
| }, | |
| { | |
| "epoch": 0.6740506329113924, | |
| "grad_norm": 0.36711472272872925, | |
| "learning_rate": 0.0010930979277622953, | |
| "loss": 1.7335, | |
| "step": 6390 | |
| }, | |
| { | |
| "epoch": 0.6751054852320675, | |
| "grad_norm": 0.4441031813621521, | |
| "learning_rate": 0.0010780312327981854, | |
| "loss": 1.7301, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 0.6761603375527426, | |
| "grad_norm": 0.3344942033290863, | |
| "learning_rate": 0.0010631722093439888, | |
| "loss": 1.7261, | |
| "step": 6410 | |
| }, | |
| { | |
| "epoch": 0.6772151898734177, | |
| "grad_norm": 0.3050030767917633, | |
| "learning_rate": 0.00104851799496331, | |
| "loss": 1.7166, | |
| "step": 6420 | |
| }, | |
| { | |
| "epoch": 0.6782700421940928, | |
| "grad_norm": 0.3217354714870453, | |
| "learning_rate": 0.0010340657666740914, | |
| "loss": 1.7236, | |
| "step": 6430 | |
| }, | |
| { | |
| "epoch": 0.679324894514768, | |
| "grad_norm": 0.3525526523590088, | |
| "learning_rate": 0.0010198127404047975, | |
| "loss": 1.7084, | |
| "step": 6440 | |
| }, | |
| { | |
| "epoch": 0.680379746835443, | |
| "grad_norm": 0.39753952622413635, | |
| "learning_rate": 0.0010057561704580897, | |
| "loss": 1.7103, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 0.6814345991561181, | |
| "grad_norm": 0.32323288917541504, | |
| "learning_rate": 0.0009918933489818985, | |
| "loss": 1.7278, | |
| "step": 6460 | |
| }, | |
| { | |
| "epoch": 0.6824894514767933, | |
| "grad_norm": 0.31371453404426575, | |
| "learning_rate": 0.0009782216054477827, | |
| "loss": 1.717, | |
| "step": 6470 | |
| }, | |
| { | |
| "epoch": 0.6835443037974683, | |
| "grad_norm": 0.3848041892051697, | |
| "learning_rate": 0.0009647383061364801, | |
| "loss": 1.7239, | |
| "step": 6480 | |
| }, | |
| { | |
| "epoch": 0.6845991561181435, | |
| "grad_norm": 0.3419899344444275, | |
| "learning_rate": 0.0009514408536305495, | |
| "loss": 1.7175, | |
| "step": 6490 | |
| }, | |
| { | |
| "epoch": 0.6856540084388185, | |
| "grad_norm": 0.3177018165588379, | |
| "learning_rate": 0.0009383266863140042, | |
| "loss": 1.7338, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.6867088607594937, | |
| "grad_norm": 0.34837961196899414, | |
| "learning_rate": 0.000925393277878844, | |
| "loss": 1.7277, | |
| "step": 6510 | |
| }, | |
| { | |
| "epoch": 0.6877637130801688, | |
| "grad_norm": 0.30670979619026184, | |
| "learning_rate": 0.0009126381368383879, | |
| "loss": 1.7142, | |
| "step": 6520 | |
| }, | |
| { | |
| "epoch": 0.6888185654008439, | |
| "grad_norm": 0.3753418028354645, | |
| "learning_rate": 0.0009000588060473156, | |
| "loss": 1.7094, | |
| "step": 6530 | |
| }, | |
| { | |
| "epoch": 0.689873417721519, | |
| "grad_norm": 0.32511377334594727, | |
| "learning_rate": 0.0008876528622283235, | |
| "loss": 1.7186, | |
| "step": 6540 | |
| }, | |
| { | |
| "epoch": 0.6909282700421941, | |
| "grad_norm": 0.29808443784713745, | |
| "learning_rate": 0.0008754179155053053, | |
| "loss": 1.709, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 0.6919831223628692, | |
| "grad_norm": 0.5224196314811707, | |
| "learning_rate": 0.0008633516089429683, | |
| "loss": 1.7118, | |
| "step": 6560 | |
| }, | |
| { | |
| "epoch": 0.6930379746835443, | |
| "grad_norm": 0.3130415380001068, | |
| "learning_rate": 0.0008514516180927928, | |
| "loss": 1.7066, | |
| "step": 6570 | |
| }, | |
| { | |
| "epoch": 0.6940928270042194, | |
| "grad_norm": 0.3112214505672455, | |
| "learning_rate": 0.0008397156505452524, | |
| "loss": 1.7042, | |
| "step": 6580 | |
| }, | |
| { | |
| "epoch": 0.6951476793248945, | |
| "grad_norm": 0.34768146276474, | |
| "learning_rate": 0.0008281414454882051, | |
| "loss": 1.7114, | |
| "step": 6590 | |
| }, | |
| { | |
| "epoch": 0.6962025316455697, | |
| "grad_norm": 0.2983977496623993, | |
| "learning_rate": 0.0008167267732713704, | |
| "loss": 1.7138, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 0.6972573839662447, | |
| "grad_norm": 0.30299532413482666, | |
| "learning_rate": 0.0008054694349768117, | |
| "loss": 1.7011, | |
| "step": 6610 | |
| }, | |
| { | |
| "epoch": 0.6983122362869199, | |
| "grad_norm": 0.37259745597839355, | |
| "learning_rate": 0.0007943672619953359, | |
| "loss": 1.7063, | |
| "step": 6620 | |
| }, | |
| { | |
| "epoch": 0.6993670886075949, | |
| "grad_norm": 0.37842774391174316, | |
| "learning_rate": 0.0007834181156087356, | |
| "loss": 1.7004, | |
| "step": 6630 | |
| }, | |
| { | |
| "epoch": 0.70042194092827, | |
| "grad_norm": 0.3643161356449127, | |
| "learning_rate": 0.0007726198865777852, | |
| "loss": 1.7062, | |
| "step": 6640 | |
| }, | |
| { | |
| "epoch": 0.7014767932489452, | |
| "grad_norm": 0.29314523935317993, | |
| "learning_rate": 0.0007619704947359191, | |
| "loss": 1.6992, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 0.7025316455696202, | |
| "grad_norm": 0.3162665069103241, | |
| "learning_rate": 0.0007514678885885087, | |
| "loss": 1.6979, | |
| "step": 6660 | |
| }, | |
| { | |
| "epoch": 0.7035864978902954, | |
| "grad_norm": 0.289482057094574, | |
| "learning_rate": 0.0007411100449176633, | |
| "loss": 1.6977, | |
| "step": 6670 | |
| }, | |
| { | |
| "epoch": 0.7046413502109705, | |
| "grad_norm": 0.29521444439888, | |
| "learning_rate": 0.0007308949683924791, | |
| "loss": 1.7021, | |
| "step": 6680 | |
| }, | |
| { | |
| "epoch": 0.7056962025316456, | |
| "grad_norm": 0.3438432514667511, | |
| "learning_rate": 0.000720820691184658, | |
| "loss": 1.6949, | |
| "step": 6690 | |
| }, | |
| { | |
| "epoch": 0.7067510548523207, | |
| "grad_norm": 0.3035603165626526, | |
| "learning_rate": 0.0007108852725894269, | |
| "loss": 1.6961, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 0.7078059071729957, | |
| "grad_norm": 0.2915734648704529, | |
| "learning_rate": 0.000701086798651681, | |
| "loss": 1.6963, | |
| "step": 6710 | |
| }, | |
| { | |
| "epoch": 0.7088607594936709, | |
| "grad_norm": 0.39023539423942566, | |
| "learning_rate": 0.0006914233817972798, | |
| "loss": 1.687, | |
| "step": 6720 | |
| }, | |
| { | |
| "epoch": 0.709915611814346, | |
| "grad_norm": 0.30199670791625977, | |
| "learning_rate": 0.0006818931604694261, | |
| "loss": 1.6969, | |
| "step": 6730 | |
| }, | |
| { | |
| "epoch": 0.7109704641350211, | |
| "grad_norm": 0.3577688932418823, | |
| "learning_rate": 0.0006724942987700563, | |
| "loss": 1.7013, | |
| "step": 6740 | |
| }, | |
| { | |
| "epoch": 0.7120253164556962, | |
| "grad_norm": 0.3057284355163574, | |
| "learning_rate": 0.0006632249861061732, | |
| "loss": 1.7006, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 0.7130801687763713, | |
| "grad_norm": 0.3916916251182556, | |
| "learning_rate": 0.0006540834368410549, | |
| "loss": 1.6944, | |
| "step": 6760 | |
| }, | |
| { | |
| "epoch": 0.7141350210970464, | |
| "grad_norm": 0.3180960416793823, | |
| "learning_rate": 0.0006450678899502701, | |
| "loss": 1.6967, | |
| "step": 6770 | |
| }, | |
| { | |
| "epoch": 0.7151898734177216, | |
| "grad_norm": 0.2974584698677063, | |
| "learning_rate": 0.0006361766086824345, | |
| "loss": 1.6934, | |
| "step": 6780 | |
| }, | |
| { | |
| "epoch": 0.7162447257383966, | |
| "grad_norm": 0.3089200258255005, | |
| "learning_rate": 0.000627407880224645, | |
| "loss": 1.6984, | |
| "step": 6790 | |
| }, | |
| { | |
| "epoch": 0.7172995780590717, | |
| "grad_norm": 0.30943137407302856, | |
| "learning_rate": 0.0006187600153725225, | |
| "loss": 1.6857, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 0.7183544303797469, | |
| "grad_norm": 0.39003410935401917, | |
| "learning_rate": 0.0006102313482048055, | |
| "loss": 1.6891, | |
| "step": 6810 | |
| }, | |
| { | |
| "epoch": 0.7194092827004219, | |
| "grad_norm": 0.32572123408317566, | |
| "learning_rate": 0.0006018202357624274, | |
| "loss": 1.6908, | |
| "step": 6820 | |
| }, | |
| { | |
| "epoch": 0.7204641350210971, | |
| "grad_norm": 0.4509422183036804, | |
| "learning_rate": 0.0005935250577320168, | |
| "loss": 1.6865, | |
| "step": 6830 | |
| }, | |
| { | |
| "epoch": 0.7215189873417721, | |
| "grad_norm": 0.35671916604042053, | |
| "learning_rate": 0.0005853442161337618, | |
| "loss": 1.6787, | |
| "step": 6840 | |
| }, | |
| { | |
| "epoch": 0.7225738396624473, | |
| "grad_norm": 0.2916567623615265, | |
| "learning_rate": 0.0005772761350135759, | |
| "loss": 1.6856, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 0.7236286919831224, | |
| "grad_norm": 0.3454425036907196, | |
| "learning_rate": 0.0005693192601395058, | |
| "loss": 1.6833, | |
| "step": 6860 | |
| }, | |
| { | |
| "epoch": 0.7246835443037974, | |
| "grad_norm": 0.33079248666763306, | |
| "learning_rate": 0.000561472058702326, | |
| "loss": 1.6813, | |
| "step": 6870 | |
| }, | |
| { | |
| "epoch": 0.7257383966244726, | |
| "grad_norm": 0.3648410737514496, | |
| "learning_rate": 0.000553733019020258, | |
| "loss": 1.6863, | |
| "step": 6880 | |
| }, | |
| { | |
| "epoch": 0.7267932489451476, | |
| "grad_norm": 0.321341335773468, | |
| "learning_rate": 0.0005461006502477612, | |
| "loss": 1.6749, | |
| "step": 6890 | |
| }, | |
| { | |
| "epoch": 0.7278481012658228, | |
| "grad_norm": 0.3055427074432373, | |
| "learning_rate": 0.0005385734820883369, | |
| "loss": 1.6789, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 0.7289029535864979, | |
| "grad_norm": 0.29323190450668335, | |
| "learning_rate": 0.0005311500645112907, | |
| "loss": 1.7028, | |
| "step": 6910 | |
| }, | |
| { | |
| "epoch": 0.729957805907173, | |
| "grad_norm": 0.32127928733825684, | |
| "learning_rate": 0.0005238289674723993, | |
| "loss": 1.683, | |
| "step": 6920 | |
| }, | |
| { | |
| "epoch": 0.7310126582278481, | |
| "grad_norm": 0.350521981716156, | |
| "learning_rate": 0.0005166087806384274, | |
| "loss": 1.6902, | |
| "step": 6930 | |
| }, | |
| { | |
| "epoch": 0.7320675105485233, | |
| "grad_norm": 0.3235044777393341, | |
| "learning_rate": 0.0005094881131154418, | |
| "loss": 1.6901, | |
| "step": 6940 | |
| }, | |
| { | |
| "epoch": 0.7331223628691983, | |
| "grad_norm": 0.32942715287208557, | |
| "learning_rate": 0.0005024655931808696, | |
| "loss": 1.6889, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 0.7341772151898734, | |
| "grad_norm": 0.3716394603252411, | |
| "learning_rate": 0.0004955398680192508, | |
| "loss": 1.6761, | |
| "step": 6960 | |
| }, | |
| { | |
| "epoch": 0.7352320675105485, | |
| "grad_norm": 0.327116459608078, | |
| "learning_rate": 0.000488709603461632, | |
| "loss": 1.6714, | |
| "step": 6970 | |
| }, | |
| { | |
| "epoch": 0.7362869198312236, | |
| "grad_norm": 0.29985353350639343, | |
| "learning_rate": 0.000481973483728553, | |
| "loss": 1.6762, | |
| "step": 6980 | |
| }, | |
| { | |
| "epoch": 0.7373417721518988, | |
| "grad_norm": 0.31413426995277405, | |
| "learning_rate": 0.0004753302111765748, | |
| "loss": 1.6767, | |
| "step": 6990 | |
| }, | |
| { | |
| "epoch": 0.7383966244725738, | |
| "grad_norm": 0.32647716999053955, | |
| "learning_rate": 0.0004687785060483032, | |
| "loss": 1.6845, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.739451476793249, | |
| "grad_norm": 0.3025463819503784, | |
| "learning_rate": 0.0004623171062258558, | |
| "loss": 1.6569, | |
| "step": 7010 | |
| }, | |
| { | |
| "epoch": 0.740506329113924, | |
| "grad_norm": 0.38400498032569885, | |
| "learning_rate": 0.0004559447669877288, | |
| "loss": 1.6726, | |
| "step": 7020 | |
| }, | |
| { | |
| "epoch": 0.7415611814345991, | |
| "grad_norm": 0.34344202280044556, | |
| "learning_rate": 0.00044966026076901413, | |
| "loss": 1.6774, | |
| "step": 7030 | |
| }, | |
| { | |
| "epoch": 0.7426160337552743, | |
| "grad_norm": 0.39909932017326355, | |
| "learning_rate": 0.00044346237692492177, | |
| "loss": 1.6812, | |
| "step": 7040 | |
| }, | |
| { | |
| "epoch": 0.7436708860759493, | |
| "grad_norm": 0.30984315276145935, | |
| "learning_rate": 0.0004373499214975615, | |
| "loss": 1.6701, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 0.7447257383966245, | |
| "grad_norm": 0.3168264329433441, | |
| "learning_rate": 0.0004313217169859396, | |
| "loss": 1.674, | |
| "step": 7060 | |
| }, | |
| { | |
| "epoch": 0.7457805907172996, | |
| "grad_norm": 0.2879488170146942, | |
| "learning_rate": 0.0004253766021191256, | |
| "loss": 1.6811, | |
| "step": 7070 | |
| }, | |
| { | |
| "epoch": 0.7468354430379747, | |
| "grad_norm": 0.29310518503189087, | |
| "learning_rate": 0.00041951343163254497, | |
| "loss": 1.6773, | |
| "step": 7080 | |
| }, | |
| { | |
| "epoch": 0.7478902953586498, | |
| "grad_norm": 0.3639557957649231, | |
| "learning_rate": 0.00041373107604735626, | |
| "loss": 1.6759, | |
| "step": 7090 | |
| }, | |
| { | |
| "epoch": 0.7489451476793249, | |
| "grad_norm": 0.29705002903938293, | |
| "learning_rate": 0.0004080284214528687, | |
| "loss": 1.6738, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 0.75, | |
| "grad_norm": 0.32136496901512146, | |
| "learning_rate": 0.0004024043692919589, | |
| "loss": 1.6807, | |
| "step": 7110 | |
| }, | |
| { | |
| "epoch": 0.7510548523206751, | |
| "grad_norm": 0.34039151668548584, | |
| "learning_rate": 0.0003968578361494449, | |
| "loss": 1.6791, | |
| "step": 7120 | |
| }, | |
| { | |
| "epoch": 0.7521097046413502, | |
| "grad_norm": 0.3102995455265045, | |
| "learning_rate": 0.000391387753543378, | |
| "loss": 1.6843, | |
| "step": 7130 | |
| }, | |
| { | |
| "epoch": 0.7531645569620253, | |
| "grad_norm": 0.29857951402664185, | |
| "learning_rate": 0.00038599306771921023, | |
| "loss": 1.6672, | |
| "step": 7140 | |
| }, | |
| { | |
| "epoch": 0.7542194092827004, | |
| "grad_norm": 0.32703617215156555, | |
| "learning_rate": 0.0003806727394468004, | |
| "loss": 1.664, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 0.7552742616033755, | |
| "grad_norm": 0.3137376010417938, | |
| "learning_rate": 0.0003754257438202162, | |
| "loss": 1.6733, | |
| "step": 7160 | |
| }, | |
| { | |
| "epoch": 0.7563291139240507, | |
| "grad_norm": 0.4127141535282135, | |
| "learning_rate": 0.0003702510700602974, | |
| "loss": 1.6804, | |
| "step": 7170 | |
| }, | |
| { | |
| "epoch": 0.7573839662447257, | |
| "grad_norm": 0.34177953004837036, | |
| "learning_rate": 0.0003651477213199393, | |
| "loss": 1.6594, | |
| "step": 7180 | |
| }, | |
| { | |
| "epoch": 0.7584388185654009, | |
| "grad_norm": 0.3109987676143646, | |
| "learning_rate": 0.000360114714492061, | |
| "loss": 1.661, | |
| "step": 7190 | |
| }, | |
| { | |
| "epoch": 0.759493670886076, | |
| "grad_norm": 0.3200550675392151, | |
| "learning_rate": 0.0003551510800202195, | |
| "loss": 1.6686, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 0.760548523206751, | |
| "grad_norm": 0.31464260816574097, | |
| "learning_rate": 0.0003502558617118353, | |
| "loss": 1.6702, | |
| "step": 7210 | |
| }, | |
| { | |
| "epoch": 0.7616033755274262, | |
| "grad_norm": 0.30083632469177246, | |
| "learning_rate": 0.0003454281165539914, | |
| "loss": 1.6816, | |
| "step": 7220 | |
| }, | |
| { | |
| "epoch": 0.7626582278481012, | |
| "grad_norm": 0.3156372606754303, | |
| "learning_rate": 0.00034066691453177176, | |
| "loss": 1.6758, | |
| "step": 7230 | |
| }, | |
| { | |
| "epoch": 0.7637130801687764, | |
| "grad_norm": 0.31183063983917236, | |
| "learning_rate": 0.0003359713384491037, | |
| "loss": 1.6744, | |
| "step": 7240 | |
| }, | |
| { | |
| "epoch": 0.7647679324894515, | |
| "grad_norm": 0.33335936069488525, | |
| "learning_rate": 0.00033134048375206944, | |
| "loss": 1.6705, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 0.7658227848101266, | |
| "grad_norm": 0.2791520655155182, | |
| "learning_rate": 0.0003267734583546536, | |
| "loss": 1.6662, | |
| "step": 7260 | |
| }, | |
| { | |
| "epoch": 0.7668776371308017, | |
| "grad_norm": 0.291107177734375, | |
| "learning_rate": 0.00032226938246689157, | |
| "loss": 1.662, | |
| "step": 7270 | |
| }, | |
| { | |
| "epoch": 0.7679324894514767, | |
| "grad_norm": 0.28034839034080505, | |
| "learning_rate": 0.0003178273884253874, | |
| "loss": 1.6669, | |
| "step": 7280 | |
| }, | |
| { | |
| "epoch": 0.7689873417721519, | |
| "grad_norm": 0.29026344418525696, | |
| "learning_rate": 0.0003134466205261674, | |
| "loss": 1.6756, | |
| "step": 7290 | |
| }, | |
| { | |
| "epoch": 0.770042194092827, | |
| "grad_norm": 0.3160174489021301, | |
| "learning_rate": 0.0003091262348598378, | |
| "loss": 1.6813, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 0.7710970464135021, | |
| "grad_norm": 0.31982719898223877, | |
| "learning_rate": 0.0003048653991490141, | |
| "loss": 1.6621, | |
| "step": 7310 | |
| }, | |
| { | |
| "epoch": 0.7721518987341772, | |
| "grad_norm": 0.3359350264072418, | |
| "learning_rate": 0.00030066329258799187, | |
| "loss": 1.6584, | |
| "step": 7320 | |
| }, | |
| { | |
| "epoch": 0.7732067510548524, | |
| "grad_norm": 0.315728098154068, | |
| "learning_rate": 0.0002965191056846266, | |
| "loss": 1.6646, | |
| "step": 7330 | |
| }, | |
| { | |
| "epoch": 0.7742616033755274, | |
| "grad_norm": 0.3030148148536682, | |
| "learning_rate": 0.000292432040104394, | |
| "loss": 1.6624, | |
| "step": 7340 | |
| }, | |
| { | |
| "epoch": 0.7753164556962026, | |
| "grad_norm": 0.3142611086368561, | |
| "learning_rate": 0.00028840130851659853, | |
| "loss": 1.6557, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 0.7763713080168776, | |
| "grad_norm": 0.2870323061943054, | |
| "learning_rate": 0.0002844261344427028, | |
| "loss": 1.6655, | |
| "step": 7360 | |
| }, | |
| { | |
| "epoch": 0.7774261603375527, | |
| "grad_norm": 0.28609779477119446, | |
| "learning_rate": 0.0002805057521067471, | |
| "loss": 1.6652, | |
| "step": 7370 | |
| }, | |
| { | |
| "epoch": 0.7784810126582279, | |
| "grad_norm": 0.28156542778015137, | |
| "learning_rate": 0.00027663940628783017, | |
| "loss": 1.656, | |
| "step": 7380 | |
| }, | |
| { | |
| "epoch": 0.7795358649789029, | |
| "grad_norm": 0.2975054085254669, | |
| "learning_rate": 0.00027282635217462393, | |
| "loss": 1.6617, | |
| "step": 7390 | |
| }, | |
| { | |
| "epoch": 0.7805907172995781, | |
| "grad_norm": 0.313997358083725, | |
| "learning_rate": 0.0002690658552218937, | |
| "loss": 1.6709, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 0.7816455696202531, | |
| "grad_norm": 0.29864105582237244, | |
| "learning_rate": 0.00026535719100899516, | |
| "loss": 1.6514, | |
| "step": 7410 | |
| }, | |
| { | |
| "epoch": 0.7827004219409283, | |
| "grad_norm": 0.294587105512619, | |
| "learning_rate": 0.00026169964510032245, | |
| "loss": 1.6578, | |
| "step": 7420 | |
| }, | |
| { | |
| "epoch": 0.7837552742616034, | |
| "grad_norm": 0.2903492748737335, | |
| "learning_rate": 0.00025809251290767984, | |
| "loss": 1.6442, | |
| "step": 7430 | |
| }, | |
| { | |
| "epoch": 0.7848101265822784, | |
| "grad_norm": 0.2977008819580078, | |
| "learning_rate": 0.00025453509955454957, | |
| "loss": 1.6489, | |
| "step": 7440 | |
| }, | |
| { | |
| "epoch": 0.7858649789029536, | |
| "grad_norm": 0.2864299714565277, | |
| "learning_rate": 0.00025102671974223175, | |
| "loss": 1.6509, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 0.7869198312236287, | |
| "grad_norm": 0.2841869592666626, | |
| "learning_rate": 0.00024756669761782815, | |
| "loss": 1.6621, | |
| "step": 7460 | |
| }, | |
| { | |
| "epoch": 0.7879746835443038, | |
| "grad_norm": 0.3404243290424347, | |
| "learning_rate": 0.0002441543666440464, | |
| "loss": 1.6505, | |
| "step": 7470 | |
| }, | |
| { | |
| "epoch": 0.7890295358649789, | |
| "grad_norm": 0.3151744604110718, | |
| "learning_rate": 0.00024078906947079878, | |
| "loss": 1.6567, | |
| "step": 7480 | |
| }, | |
| { | |
| "epoch": 0.790084388185654, | |
| "grad_norm": 0.2896600365638733, | |
| "learning_rate": 0.00023747015780857005, | |
| "loss": 1.6646, | |
| "step": 7490 | |
| }, | |
| { | |
| "epoch": 0.7911392405063291, | |
| "grad_norm": 0.2963440716266632, | |
| "learning_rate": 0.00023419699230353144, | |
| "loss": 1.6582, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.7921940928270043, | |
| "grad_norm": 0.3287936747074127, | |
| "learning_rate": 0.00023096894241437586, | |
| "loss": 1.6669, | |
| "step": 7510 | |
| }, | |
| { | |
| "epoch": 0.7932489451476793, | |
| "grad_norm": 0.29270681738853455, | |
| "learning_rate": 0.00022778538629085056, | |
| "loss": 1.6494, | |
| "step": 7520 | |
| }, | |
| { | |
| "epoch": 0.7943037974683544, | |
| "grad_norm": 0.28976690769195557, | |
| "learning_rate": 0.00022464571065396427, | |
| "loss": 1.6516, | |
| "step": 7530 | |
| }, | |
| { | |
| "epoch": 0.7953586497890295, | |
| "grad_norm": 0.30210551619529724, | |
| "learning_rate": 0.00022154931067784521, | |
| "loss": 1.6495, | |
| "step": 7540 | |
| }, | |
| { | |
| "epoch": 0.7964135021097046, | |
| "grad_norm": 0.2862395942211151, | |
| "learning_rate": 0.00021849558987322782, | |
| "loss": 1.6462, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 0.7974683544303798, | |
| "grad_norm": 0.3069690465927124, | |
| "learning_rate": 0.0002154839599725452, | |
| "loss": 1.6478, | |
| "step": 7560 | |
| }, | |
| { | |
| "epoch": 0.7985232067510548, | |
| "grad_norm": 0.3010164201259613, | |
| "learning_rate": 0.00021251384081660544, | |
| "loss": 1.6609, | |
| "step": 7570 | |
| }, | |
| { | |
| "epoch": 0.79957805907173, | |
| "grad_norm": 0.286625474691391, | |
| "learning_rate": 0.0002095846602428303, | |
| "loss": 1.6562, | |
| "step": 7580 | |
| }, | |
| { | |
| "epoch": 0.8006329113924051, | |
| "grad_norm": 0.29061058163642883, | |
| "learning_rate": 0.00020669585397503358, | |
| "loss": 1.6505, | |
| "step": 7590 | |
| }, | |
| { | |
| "epoch": 0.8016877637130801, | |
| "grad_norm": 0.30739066004753113, | |
| "learning_rate": 0.0002038468655147195, | |
| "loss": 1.651, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 0.8027426160337553, | |
| "grad_norm": 0.29613780975341797, | |
| "learning_rate": 0.00020103714603387894, | |
| "loss": 1.6672, | |
| "step": 7610 | |
| }, | |
| { | |
| "epoch": 0.8037974683544303, | |
| "grad_norm": 0.3332805037498474, | |
| "learning_rate": 0.00019826615426926338, | |
| "loss": 1.6368, | |
| "step": 7620 | |
| }, | |
| { | |
| "epoch": 0.8048523206751055, | |
| "grad_norm": 0.2942625880241394, | |
| "learning_rate": 0.00019553335641811625, | |
| "loss": 1.659, | |
| "step": 7630 | |
| }, | |
| { | |
| "epoch": 0.8059071729957806, | |
| "grad_norm": 0.30071958899497986, | |
| "learning_rate": 0.0001928382260353415, | |
| "loss": 1.6528, | |
| "step": 7640 | |
| }, | |
| { | |
| "epoch": 0.8069620253164557, | |
| "grad_norm": 0.27827274799346924, | |
| "learning_rate": 0.00019018024393208902, | |
| "loss": 1.6662, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 0.8080168776371308, | |
| "grad_norm": 0.2844911813735962, | |
| "learning_rate": 0.00018755889807573872, | |
| "loss": 1.6455, | |
| "step": 7660 | |
| }, | |
| { | |
| "epoch": 0.8090717299578059, | |
| "grad_norm": 0.2892758548259735, | |
| "learning_rate": 0.00018497368349126262, | |
| "loss": 1.6504, | |
| "step": 7670 | |
| }, | |
| { | |
| "epoch": 0.810126582278481, | |
| "grad_norm": 0.33394187688827515, | |
| "learning_rate": 0.00018242410216394648, | |
| "loss": 1.6664, | |
| "step": 7680 | |
| }, | |
| { | |
| "epoch": 0.8111814345991561, | |
| "grad_norm": 0.3627508580684662, | |
| "learning_rate": 0.0001799096629434529, | |
| "loss": 1.6416, | |
| "step": 7690 | |
| }, | |
| { | |
| "epoch": 0.8122362869198312, | |
| "grad_norm": 0.3369033634662628, | |
| "learning_rate": 0.00017742988144920578, | |
| "loss": 1.6495, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 0.8132911392405063, | |
| "grad_norm": 0.3026972711086273, | |
| "learning_rate": 0.00017498427997707976, | |
| "loss": 1.6489, | |
| "step": 7710 | |
| }, | |
| { | |
| "epoch": 0.8143459915611815, | |
| "grad_norm": 0.2892075479030609, | |
| "learning_rate": 0.00017257238740737548, | |
| "loss": 1.6511, | |
| "step": 7720 | |
| }, | |
| { | |
| "epoch": 0.8154008438818565, | |
| "grad_norm": 0.31418395042419434, | |
| "learning_rate": 0.00017019373911406307, | |
| "loss": 1.6581, | |
| "step": 7730 | |
| }, | |
| { | |
| "epoch": 0.8164556962025317, | |
| "grad_norm": 0.2813819348812103, | |
| "learning_rate": 0.000167847876875277, | |
| "loss": 1.6565, | |
| "step": 7740 | |
| }, | |
| { | |
| "epoch": 0.8175105485232067, | |
| "grad_norm": 0.35128098726272583, | |
| "learning_rate": 0.00016553434878504428, | |
| "loss": 1.6409, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 0.8185654008438819, | |
| "grad_norm": 0.40352514386177063, | |
| "learning_rate": 0.00016325270916622947, | |
| "loss": 1.6425, | |
| "step": 7760 | |
| }, | |
| { | |
| "epoch": 0.819620253164557, | |
| "grad_norm": 0.3125883638858795, | |
| "learning_rate": 0.00016100251848467966, | |
| "loss": 1.6529, | |
| "step": 7770 | |
| }, | |
| { | |
| "epoch": 0.820675105485232, | |
| "grad_norm": 0.3259755074977875, | |
| "learning_rate": 0.0001587833432645528, | |
| "loss": 1.6431, | |
| "step": 7780 | |
| }, | |
| { | |
| "epoch": 0.8217299578059072, | |
| "grad_norm": 0.32143619656562805, | |
| "learning_rate": 0.00015659475600481292, | |
| "loss": 1.6624, | |
| "step": 7790 | |
| }, | |
| { | |
| "epoch": 0.8227848101265823, | |
| "grad_norm": 0.2872926890850067, | |
| "learning_rate": 0.00015443633509687688, | |
| "loss": 1.6482, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 0.8238396624472574, | |
| "grad_norm": 0.334501713514328, | |
| "learning_rate": 0.00015230766474339536, | |
| "loss": 1.6481, | |
| "step": 7810 | |
| }, | |
| { | |
| "epoch": 0.8248945147679325, | |
| "grad_norm": 0.2882138192653656, | |
| "learning_rate": 0.00015020833487815416, | |
| "loss": 1.6549, | |
| "step": 7820 | |
| }, | |
| { | |
| "epoch": 0.8259493670886076, | |
| "grad_norm": 0.30133500695228577, | |
| "learning_rate": 0.0001481379410870792, | |
| "loss": 1.6466, | |
| "step": 7830 | |
| }, | |
| { | |
| "epoch": 0.8270042194092827, | |
| "grad_norm": 0.3030688762664795, | |
| "learning_rate": 0.00014609608453033013, | |
| "loss": 1.6337, | |
| "step": 7840 | |
| }, | |
| { | |
| "epoch": 0.8280590717299579, | |
| "grad_norm": 0.28669485449790955, | |
| "learning_rate": 0.00014408237186546807, | |
| "loss": 1.6418, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 0.8291139240506329, | |
| "grad_norm": 0.2894490957260132, | |
| "learning_rate": 0.00014209641517168273, | |
| "loss": 1.6302, | |
| "step": 7860 | |
| }, | |
| { | |
| "epoch": 0.830168776371308, | |
| "grad_norm": 0.30199211835861206, | |
| "learning_rate": 0.00014013783187506265, | |
| "loss": 1.6441, | |
| "step": 7870 | |
| }, | |
| { | |
| "epoch": 0.8312236286919831, | |
| "grad_norm": 0.3222101032733917, | |
| "learning_rate": 0.00013820624467489697, | |
| "loss": 1.6625, | |
| "step": 7880 | |
| }, | |
| { | |
| "epoch": 0.8322784810126582, | |
| "grad_norm": 0.2813795804977417, | |
| "learning_rate": 0.00013630128147099213, | |
| "loss": 1.6602, | |
| "step": 7890 | |
| }, | |
| { | |
| "epoch": 0.8333333333333334, | |
| "grad_norm": 0.2913859784603119, | |
| "learning_rate": 0.00013442257529199068, | |
| "loss": 1.641, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 0.8343881856540084, | |
| "grad_norm": 0.30320343375205994, | |
| "learning_rate": 0.00013256976422467803, | |
| "loss": 1.6484, | |
| "step": 7910 | |
| }, | |
| { | |
| "epoch": 0.8354430379746836, | |
| "grad_norm": 0.3305092751979828, | |
| "learning_rate": 0.00013074249134426366, | |
| "loss": 1.6472, | |
| "step": 7920 | |
| }, | |
| { | |
| "epoch": 0.8364978902953587, | |
| "grad_norm": 0.28237384557724, | |
| "learning_rate": 0.0001289404046456233, | |
| "loss": 1.6571, | |
| "step": 7930 | |
| }, | |
| { | |
| "epoch": 0.8375527426160337, | |
| "grad_norm": 0.29093295335769653, | |
| "learning_rate": 0.0001271631569754887, | |
| "loss": 1.6491, | |
| "step": 7940 | |
| }, | |
| { | |
| "epoch": 0.8386075949367089, | |
| "grad_norm": 0.2733587324619293, | |
| "learning_rate": 0.0001254104059655723, | |
| "loss": 1.6536, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 0.8396624472573839, | |
| "grad_norm": 0.28348734974861145, | |
| "learning_rate": 0.00012368181396661337, | |
| "loss": 1.6359, | |
| "step": 7960 | |
| }, | |
| { | |
| "epoch": 0.8407172995780591, | |
| "grad_norm": 0.3097951412200928, | |
| "learning_rate": 0.00012197704798333364, | |
| "loss": 1.6398, | |
| "step": 7970 | |
| }, | |
| { | |
| "epoch": 0.8417721518987342, | |
| "grad_norm": 0.29911431670188904, | |
| "learning_rate": 0.00012029577961028894, | |
| "loss": 1.6375, | |
| "step": 7980 | |
| }, | |
| { | |
| "epoch": 0.8428270042194093, | |
| "grad_norm": 0.3036710321903229, | |
| "learning_rate": 0.00011863768496860542, | |
| "loss": 1.6509, | |
| "step": 7990 | |
| }, | |
| { | |
| "epoch": 0.8438818565400844, | |
| "grad_norm": 0.28275567293167114, | |
| "learning_rate": 0.00011700244464358777, | |
| "loss": 1.6423, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 0.8449367088607594, | |
| "grad_norm": 0.28171199560165405, | |
| "learning_rate": 0.00011538974362318715, | |
| "loss": 1.6516, | |
| "step": 8010 | |
| }, | |
| { | |
| "epoch": 0.8459915611814346, | |
| "grad_norm": 0.2995152771472931, | |
| "learning_rate": 0.00011379927123731737, | |
| "loss": 1.6406, | |
| "step": 8020 | |
| }, | |
| { | |
| "epoch": 0.8470464135021097, | |
| "grad_norm": 0.28432291746139526, | |
| "learning_rate": 0.0001122307210980077, | |
| "loss": 1.6411, | |
| "step": 8030 | |
| }, | |
| { | |
| "epoch": 0.8481012658227848, | |
| "grad_norm": 0.2907130718231201, | |
| "learning_rate": 0.00011068379104038026, | |
| "loss": 1.6548, | |
| "step": 8040 | |
| }, | |
| { | |
| "epoch": 0.8491561181434599, | |
| "grad_norm": 0.3092287480831146, | |
| "learning_rate": 0.00010915818306444116, | |
| "loss": 1.6366, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 0.8502109704641351, | |
| "grad_norm": 0.3048560619354248, | |
| "learning_rate": 0.00010765360327767384, | |
| "loss": 1.6422, | |
| "step": 8060 | |
| }, | |
| { | |
| "epoch": 0.8512658227848101, | |
| "grad_norm": 0.2986976206302643, | |
| "learning_rate": 0.00010616976183842376, | |
| "loss": 1.6479, | |
| "step": 8070 | |
| }, | |
| { | |
| "epoch": 0.8523206751054853, | |
| "grad_norm": 0.3488602936267853, | |
| "learning_rate": 0.00010470637290006365, | |
| "loss": 1.652, | |
| "step": 8080 | |
| }, | |
| { | |
| "epoch": 0.8533755274261603, | |
| "grad_norm": 0.3072564899921417, | |
| "learning_rate": 0.00010326315455592764, | |
| "loss": 1.6414, | |
| "step": 8090 | |
| }, | |
| { | |
| "epoch": 0.8544303797468354, | |
| "grad_norm": 0.3198103904724121, | |
| "learning_rate": 0.0001018398287850053, | |
| "loss": 1.6323, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 0.8554852320675106, | |
| "grad_norm": 0.3057430386543274, | |
| "learning_rate": 0.00010043612139838357, | |
| "loss": 1.653, | |
| "step": 8110 | |
| }, | |
| { | |
| "epoch": 0.8565400843881856, | |
| "grad_norm": 0.2905876338481903, | |
| "learning_rate": 9.905176198642719e-05, | |
| "loss": 1.6391, | |
| "step": 8120 | |
| }, | |
| { | |
| "epoch": 0.8575949367088608, | |
| "grad_norm": 0.31224319338798523, | |
| "learning_rate": 9.76864838666871e-05, | |
| "loss": 1.642, | |
| "step": 8130 | |
| }, | |
| { | |
| "epoch": 0.8586497890295358, | |
| "grad_norm": 0.2979212701320648, | |
| "learning_rate": 9.634002403252676e-05, | |
| "loss": 1.6387, | |
| "step": 8140 | |
| }, | |
| { | |
| "epoch": 0.859704641350211, | |
| "grad_norm": 0.2785697281360626, | |
| "learning_rate": 9.501212310245681e-05, | |
| "loss": 1.636, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 0.8607594936708861, | |
| "grad_norm": 0.30655381083488464, | |
| "learning_rate": 9.370252527016777e-05, | |
| "loss": 1.6483, | |
| "step": 8160 | |
| }, | |
| { | |
| "epoch": 0.8618143459915611, | |
| "grad_norm": 0.31941476464271545, | |
| "learning_rate": 9.241097825525163e-05, | |
| "loss": 1.6379, | |
| "step": 8170 | |
| }, | |
| { | |
| "epoch": 0.8628691983122363, | |
| "grad_norm": 0.2967509925365448, | |
| "learning_rate": 9.113723325460276e-05, | |
| "loss": 1.6462, | |
| "step": 8180 | |
| }, | |
| { | |
| "epoch": 0.8639240506329114, | |
| "grad_norm": 0.31797289848327637, | |
| "learning_rate": 8.988104489448849e-05, | |
| "loss": 1.6416, | |
| "step": 8190 | |
| }, | |
| { | |
| "epoch": 0.8649789029535865, | |
| "grad_norm": 0.3229648470878601, | |
| "learning_rate": 8.864217118328042e-05, | |
| "loss": 1.6499, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 0.8660337552742616, | |
| "grad_norm": 0.28441721200942993, | |
| "learning_rate": 8.742037346483729e-05, | |
| "loss": 1.6445, | |
| "step": 8210 | |
| }, | |
| { | |
| "epoch": 0.8670886075949367, | |
| "grad_norm": 0.32493698596954346, | |
| "learning_rate": 8.62154163725303e-05, | |
| "loss": 1.6502, | |
| "step": 8220 | |
| }, | |
| { | |
| "epoch": 0.8681434599156118, | |
| "grad_norm": 0.29117193818092346, | |
| "learning_rate": 8.502706778390219e-05, | |
| "loss": 1.6466, | |
| "step": 8230 | |
| }, | |
| { | |
| "epoch": 0.869198312236287, | |
| "grad_norm": 0.319476842880249, | |
| "learning_rate": 8.38550987759513e-05, | |
| "loss": 1.6434, | |
| "step": 8240 | |
| }, | |
| { | |
| "epoch": 0.870253164556962, | |
| "grad_norm": 0.33274680376052856, | |
| "learning_rate": 8.269928358103191e-05, | |
| "loss": 1.6572, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 0.8713080168776371, | |
| "grad_norm": 0.30592653155326843, | |
| "learning_rate": 8.155939954336243e-05, | |
| "loss": 1.6482, | |
| "step": 8260 | |
| }, | |
| { | |
| "epoch": 0.8723628691983122, | |
| "grad_norm": 0.28294315934181213, | |
| "learning_rate": 8.043522707613312e-05, | |
| "loss": 1.6424, | |
| "step": 8270 | |
| }, | |
| { | |
| "epoch": 0.8734177215189873, | |
| "grad_norm": 0.27652573585510254, | |
| "learning_rate": 7.932654961920486e-05, | |
| "loss": 1.6286, | |
| "step": 8280 | |
| }, | |
| { | |
| "epoch": 0.8744725738396625, | |
| "grad_norm": 0.30717819929122925, | |
| "learning_rate": 7.823315359739135e-05, | |
| "loss": 1.6298, | |
| "step": 8290 | |
| }, | |
| { | |
| "epoch": 0.8755274261603375, | |
| "grad_norm": 0.28261104226112366, | |
| "learning_rate": 7.715482837931577e-05, | |
| "loss": 1.6554, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 0.8765822784810127, | |
| "grad_norm": 0.32058534026145935, | |
| "learning_rate": 7.6091366236835e-05, | |
| "loss": 1.6332, | |
| "step": 8310 | |
| }, | |
| { | |
| "epoch": 0.8776371308016878, | |
| "grad_norm": 0.28376415371894836, | |
| "learning_rate": 7.504256230502289e-05, | |
| "loss": 1.6485, | |
| "step": 8320 | |
| }, | |
| { | |
| "epoch": 0.8786919831223629, | |
| "grad_norm": 0.26899126172065735, | |
| "learning_rate": 7.400821454270524e-05, | |
| "loss": 1.6464, | |
| "step": 8330 | |
| }, | |
| { | |
| "epoch": 0.879746835443038, | |
| "grad_norm": 0.2815333306789398, | |
| "learning_rate": 7.29881236935386e-05, | |
| "loss": 1.6327, | |
| "step": 8340 | |
| }, | |
| { | |
| "epoch": 0.880801687763713, | |
| "grad_norm": 0.29519298672676086, | |
| "learning_rate": 7.198209324762562e-05, | |
| "loss": 1.6352, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 0.8818565400843882, | |
| "grad_norm": 0.2897711992263794, | |
| "learning_rate": 7.098992940365946e-05, | |
| "loss": 1.6325, | |
| "step": 8360 | |
| }, | |
| { | |
| "epoch": 0.8829113924050633, | |
| "grad_norm": 0.27384185791015625, | |
| "learning_rate": 7.001144103159e-05, | |
| "loss": 1.6415, | |
| "step": 8370 | |
| }, | |
| { | |
| "epoch": 0.8839662447257384, | |
| "grad_norm": 0.2890484035015106, | |
| "learning_rate": 6.904643963580461e-05, | |
| "loss": 1.6485, | |
| "step": 8380 | |
| }, | |
| { | |
| "epoch": 0.8850210970464135, | |
| "grad_norm": 0.3331020176410675, | |
| "learning_rate": 6.809473931881644e-05, | |
| "loss": 1.6375, | |
| "step": 8390 | |
| }, | |
| { | |
| "epoch": 0.8860759493670886, | |
| "grad_norm": 0.3167933523654938, | |
| "learning_rate": 6.71561567454532e-05, | |
| "loss": 1.6406, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 0.8871308016877637, | |
| "grad_norm": 0.3101309835910797, | |
| "learning_rate": 6.623051110753948e-05, | |
| "loss": 1.6576, | |
| "step": 8410 | |
| }, | |
| { | |
| "epoch": 0.8881856540084389, | |
| "grad_norm": 0.282186895608902, | |
| "learning_rate": 6.531762408906607e-05, | |
| "loss": 1.6451, | |
| "step": 8420 | |
| }, | |
| { | |
| "epoch": 0.8892405063291139, | |
| "grad_norm": 0.27861738204956055, | |
| "learning_rate": 6.441731983183912e-05, | |
| "loss": 1.6379, | |
| "step": 8430 | |
| }, | |
| { | |
| "epoch": 0.890295358649789, | |
| "grad_norm": 0.31566575169563293, | |
| "learning_rate": 6.352942490160292e-05, | |
| "loss": 1.6353, | |
| "step": 8440 | |
| }, | |
| { | |
| "epoch": 0.8913502109704642, | |
| "grad_norm": 0.28727439045906067, | |
| "learning_rate": 6.265376825462966e-05, | |
| "loss": 1.6467, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 0.8924050632911392, | |
| "grad_norm": 0.2992555499076843, | |
| "learning_rate": 6.179018120476945e-05, | |
| "loss": 1.6408, | |
| "step": 8460 | |
| }, | |
| { | |
| "epoch": 0.8934599156118144, | |
| "grad_norm": 0.28516122698783875, | |
| "learning_rate": 6.0938497390954946e-05, | |
| "loss": 1.6403, | |
| "step": 8470 | |
| }, | |
| { | |
| "epoch": 0.8945147679324894, | |
| "grad_norm": 0.2922123968601227, | |
| "learning_rate": 6.009855274515339e-05, | |
| "loss": 1.6322, | |
| "step": 8480 | |
| }, | |
| { | |
| "epoch": 0.8955696202531646, | |
| "grad_norm": 0.3001478910446167, | |
| "learning_rate": 5.9270185460760735e-05, | |
| "loss": 1.6425, | |
| "step": 8490 | |
| }, | |
| { | |
| "epoch": 0.8966244725738397, | |
| "grad_norm": 0.29792818427085876, | |
| "learning_rate": 5.8453235961431225e-05, | |
| "loss": 1.6293, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 0.8976793248945147, | |
| "grad_norm": 0.2816435396671295, | |
| "learning_rate": 5.764754687033678e-05, | |
| "loss": 1.6408, | |
| "step": 8510 | |
| }, | |
| { | |
| "epoch": 0.8987341772151899, | |
| "grad_norm": 0.30157747864723206, | |
| "learning_rate": 5.6852962979849836e-05, | |
| "loss": 1.6388, | |
| "step": 8520 | |
| }, | |
| { | |
| "epoch": 0.8997890295358649, | |
| "grad_norm": 0.27947813272476196, | |
| "learning_rate": 5.6069331221644284e-05, | |
| "loss": 1.6516, | |
| "step": 8530 | |
| }, | |
| { | |
| "epoch": 0.9008438818565401, | |
| "grad_norm": 0.28350022435188293, | |
| "learning_rate": 5.529650063720842e-05, | |
| "loss": 1.649, | |
| "step": 8540 | |
| }, | |
| { | |
| "epoch": 0.9018987341772152, | |
| "grad_norm": 0.2897917628288269, | |
| "learning_rate": 5.453432234876445e-05, | |
| "loss": 1.6379, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 0.9029535864978903, | |
| "grad_norm": 0.2812824249267578, | |
| "learning_rate": 5.37826495305886e-05, | |
| "loss": 1.6284, | |
| "step": 8560 | |
| }, | |
| { | |
| "epoch": 0.9040084388185654, | |
| "grad_norm": 0.31812164187431335, | |
| "learning_rate": 5.304133738072674e-05, | |
| "loss": 1.647, | |
| "step": 8570 | |
| }, | |
| { | |
| "epoch": 0.9050632911392406, | |
| "grad_norm": 0.26702648401260376, | |
| "learning_rate": 5.2310243093099814e-05, | |
| "loss": 1.6428, | |
| "step": 8580 | |
| }, | |
| { | |
| "epoch": 0.9061181434599156, | |
| "grad_norm": 0.3153185546398163, | |
| "learning_rate": 5.158922582999368e-05, | |
| "loss": 1.6443, | |
| "step": 8590 | |
| }, | |
| { | |
| "epoch": 0.9071729957805907, | |
| "grad_norm": 0.282937616109848, | |
| "learning_rate": 5.087814669492819e-05, | |
| "loss": 1.6386, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 0.9082278481012658, | |
| "grad_norm": 0.31850728392601013, | |
| "learning_rate": 5.017686870590028e-05, | |
| "loss": 1.6368, | |
| "step": 8610 | |
| }, | |
| { | |
| "epoch": 0.9092827004219409, | |
| "grad_norm": 0.27971863746643066, | |
| "learning_rate": 4.948525676899577e-05, | |
| "loss": 1.634, | |
| "step": 8620 | |
| }, | |
| { | |
| "epoch": 0.9103375527426161, | |
| "grad_norm": 0.2911681830883026, | |
| "learning_rate": 4.880317765236493e-05, | |
| "loss": 1.6432, | |
| "step": 8630 | |
| }, | |
| { | |
| "epoch": 0.9113924050632911, | |
| "grad_norm": 0.2851537764072418, | |
| "learning_rate": 4.8130499960556755e-05, | |
| "loss": 1.6238, | |
| "step": 8640 | |
| }, | |
| { | |
| "epoch": 0.9124472573839663, | |
| "grad_norm": 0.29987117648124695, | |
| "learning_rate": 4.746709410920699e-05, | |
| "loss": 1.6327, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 0.9135021097046413, | |
| "grad_norm": 0.3069855272769928, | |
| "learning_rate": 4.681283230007507e-05, | |
| "loss": 1.6331, | |
| "step": 8660 | |
| }, | |
| { | |
| "epoch": 0.9145569620253164, | |
| "grad_norm": 0.2873881161212921, | |
| "learning_rate": 4.616758849642509e-05, | |
| "loss": 1.6395, | |
| "step": 8670 | |
| }, | |
| { | |
| "epoch": 0.9156118143459916, | |
| "grad_norm": 0.27439984679222107, | |
| "learning_rate": 4.553123839874615e-05, | |
| "loss": 1.6483, | |
| "step": 8680 | |
| }, | |
| { | |
| "epoch": 0.9166666666666666, | |
| "grad_norm": 0.2946086823940277, | |
| "learning_rate": 4.490365942080736e-05, | |
| "loss": 1.6417, | |
| "step": 8690 | |
| }, | |
| { | |
| "epoch": 0.9177215189873418, | |
| "grad_norm": 0.29139867424964905, | |
| "learning_rate": 4.428473066604285e-05, | |
| "loss": 1.6411, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 0.9187763713080169, | |
| "grad_norm": 0.2842201590538025, | |
| "learning_rate": 4.367433290426233e-05, | |
| "loss": 1.639, | |
| "step": 8710 | |
| }, | |
| { | |
| "epoch": 0.919831223628692, | |
| "grad_norm": 0.2875899374485016, | |
| "learning_rate": 4.3072348548682595e-05, | |
| "loss": 1.6334, | |
| "step": 8720 | |
| }, | |
| { | |
| "epoch": 0.9208860759493671, | |
| "grad_norm": 0.29709094762802124, | |
| "learning_rate": 4.247866163327575e-05, | |
| "loss": 1.6456, | |
| "step": 8730 | |
| }, | |
| { | |
| "epoch": 0.9219409282700421, | |
| "grad_norm": 0.29496219754219055, | |
| "learning_rate": 4.1893157790429404e-05, | |
| "loss": 1.6311, | |
| "step": 8740 | |
| }, | |
| { | |
| "epoch": 0.9229957805907173, | |
| "grad_norm": 0.28014180064201355, | |
| "learning_rate": 4.1315724228915066e-05, | |
| "loss": 1.6318, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 0.9240506329113924, | |
| "grad_norm": 0.3066428601741791, | |
| "learning_rate": 4.074624971216005e-05, | |
| "loss": 1.6208, | |
| "step": 8760 | |
| }, | |
| { | |
| "epoch": 0.9251054852320675, | |
| "grad_norm": 0.28945207595825195, | |
| "learning_rate": 4.018462453681889e-05, | |
| "loss": 1.6345, | |
| "step": 8770 | |
| }, | |
| { | |
| "epoch": 0.9261603375527426, | |
| "grad_norm": 0.2926965057849884, | |
| "learning_rate": 3.963074051164014e-05, | |
| "loss": 1.6389, | |
| "step": 8780 | |
| }, | |
| { | |
| "epoch": 0.9272151898734177, | |
| "grad_norm": 0.2945704460144043, | |
| "learning_rate": 3.908449093662446e-05, | |
| "loss": 1.6349, | |
| "step": 8790 | |
| }, | |
| { | |
| "epoch": 0.9282700421940928, | |
| "grad_norm": 0.2938326895236969, | |
| "learning_rate": 3.854577058246998e-05, | |
| "loss": 1.6323, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 0.929324894514768, | |
| "grad_norm": 0.28471556305885315, | |
| "learning_rate": 3.801447567030094e-05, | |
| "loss": 1.6477, | |
| "step": 8810 | |
| }, | |
| { | |
| "epoch": 0.930379746835443, | |
| "grad_norm": 0.2794736921787262, | |
| "learning_rate": 3.7490503851675777e-05, | |
| "loss": 1.6367, | |
| "step": 8820 | |
| }, | |
| { | |
| "epoch": 0.9314345991561181, | |
| "grad_norm": 0.28662699460983276, | |
| "learning_rate": 3.6973754188870806e-05, | |
| "loss": 1.6505, | |
| "step": 8830 | |
| }, | |
| { | |
| "epoch": 0.9324894514767933, | |
| "grad_norm": 0.2798454761505127, | |
| "learning_rate": 3.6464127135435536e-05, | |
| "loss": 1.6358, | |
| "step": 8840 | |
| }, | |
| { | |
| "epoch": 0.9335443037974683, | |
| "grad_norm": 0.30066630244255066, | |
| "learning_rate": 3.596152451701616e-05, | |
| "loss": 1.6345, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 0.9345991561181435, | |
| "grad_norm": 0.29603078961372375, | |
| "learning_rate": 3.5465849512443226e-05, | |
| "loss": 1.6346, | |
| "step": 8860 | |
| }, | |
| { | |
| "epoch": 0.9356540084388185, | |
| "grad_norm": 0.29392972588539124, | |
| "learning_rate": 3.4977006635080086e-05, | |
| "loss": 1.6395, | |
| "step": 8870 | |
| }, | |
| { | |
| "epoch": 0.9367088607594937, | |
| "grad_norm": 0.286011666059494, | |
| "learning_rate": 3.449490171442838e-05, | |
| "loss": 1.6419, | |
| "step": 8880 | |
| }, | |
| { | |
| "epoch": 0.9377637130801688, | |
| "grad_norm": 0.2774491310119629, | |
| "learning_rate": 3.401944187798702e-05, | |
| "loss": 1.6415, | |
| "step": 8890 | |
| }, | |
| { | |
| "epoch": 0.9388185654008439, | |
| "grad_norm": 0.2881662845611572, | |
| "learning_rate": 3.355053553336137e-05, | |
| "loss": 1.6307, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 0.939873417721519, | |
| "grad_norm": 0.31947827339172363, | |
| "learning_rate": 3.308809235061882e-05, | |
| "loss": 1.6322, | |
| "step": 8910 | |
| }, | |
| { | |
| "epoch": 0.9409282700421941, | |
| "grad_norm": 0.302874356508255, | |
| "learning_rate": 3.263202324488772e-05, | |
| "loss": 1.637, | |
| "step": 8920 | |
| }, | |
| { | |
| "epoch": 0.9419831223628692, | |
| "grad_norm": 0.2747725546360016, | |
| "learning_rate": 3.218224035919609e-05, | |
| "loss": 1.6358, | |
| "step": 8930 | |
| }, | |
| { | |
| "epoch": 0.9430379746835443, | |
| "grad_norm": 0.28067922592163086, | |
| "learning_rate": 3.173865704754688e-05, | |
| "loss": 1.6351, | |
| "step": 8940 | |
| }, | |
| { | |
| "epoch": 0.9440928270042194, | |
| "grad_norm": 0.27929338812828064, | |
| "learning_rate": 3.130118785822657e-05, | |
| "loss": 1.6372, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 0.9451476793248945, | |
| "grad_norm": 0.3006559908390045, | |
| "learning_rate": 3.08697485173437e-05, | |
| "loss": 1.6431, | |
| "step": 8960 | |
| }, | |
| { | |
| "epoch": 0.9462025316455697, | |
| "grad_norm": 0.2935165464878082, | |
| "learning_rate": 3.0444255912594442e-05, | |
| "loss": 1.6393, | |
| "step": 8970 | |
| }, | |
| { | |
| "epoch": 0.9472573839662447, | |
| "grad_norm": 0.3300228416919708, | |
| "learning_rate": 3.002462807725185e-05, | |
| "loss": 1.6351, | |
| "step": 8980 | |
| }, | |
| { | |
| "epoch": 0.9483122362869199, | |
| "grad_norm": 0.28363847732543945, | |
| "learning_rate": 2.9610784174375868e-05, | |
| "loss": 1.6407, | |
| "step": 8990 | |
| }, | |
| { | |
| "epoch": 0.9493670886075949, | |
| "grad_norm": 0.28174105286598206, | |
| "learning_rate": 2.920264448124087e-05, | |
| "loss": 1.634, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 0.95042194092827, | |
| "grad_norm": 0.3101724088191986, | |
| "learning_rate": 2.8800130373977936e-05, | |
| "loss": 1.6322, | |
| "step": 9010 | |
| }, | |
| { | |
| "epoch": 0.9514767932489452, | |
| "grad_norm": 0.30244192481040955, | |
| "learning_rate": 2.84031643124288e-05, | |
| "loss": 1.6343, | |
| "step": 9020 | |
| }, | |
| { | |
| "epoch": 0.9525316455696202, | |
| "grad_norm": 0.2945556640625, | |
| "learning_rate": 2.8011669825208517e-05, | |
| "loss": 1.6576, | |
| "step": 9030 | |
| }, | |
| { | |
| "epoch": 0.9535864978902954, | |
| "grad_norm": 0.29157695174217224, | |
| "learning_rate": 2.762557149497405e-05, | |
| "loss": 1.6278, | |
| "step": 9040 | |
| }, | |
| { | |
| "epoch": 0.9546413502109705, | |
| "grad_norm": 0.2839052081108093, | |
| "learning_rate": 2.724479494389592e-05, | |
| "loss": 1.6388, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 0.9556962025316456, | |
| "grad_norm": 0.28519225120544434, | |
| "learning_rate": 2.6869266819330058e-05, | |
| "loss": 1.6412, | |
| "step": 9060 | |
| }, | |
| { | |
| "epoch": 0.9567510548523207, | |
| "grad_norm": 0.30681735277175903, | |
| "learning_rate": 2.6498914779687228e-05, | |
| "loss": 1.6399, | |
| "step": 9070 | |
| }, | |
| { | |
| "epoch": 0.9578059071729957, | |
| "grad_norm": 0.27808111906051636, | |
| "learning_rate": 2.6133667480497115e-05, | |
| "loss": 1.6389, | |
| "step": 9080 | |
| }, | |
| { | |
| "epoch": 0.9588607594936709, | |
| "grad_norm": 0.29020893573760986, | |
| "learning_rate": 2.5773454560664597e-05, | |
| "loss": 1.6443, | |
| "step": 9090 | |
| }, | |
| { | |
| "epoch": 0.959915611814346, | |
| "grad_norm": 0.26861000061035156, | |
| "learning_rate": 2.541820662891541e-05, | |
| "loss": 1.62, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 0.9609704641350211, | |
| "grad_norm": 0.28837451338768005, | |
| "learning_rate": 2.5067855250428616e-05, | |
| "loss": 1.6338, | |
| "step": 9110 | |
| }, | |
| { | |
| "epoch": 0.9620253164556962, | |
| "grad_norm": 0.3207913041114807, | |
| "learning_rate": 2.472233293365335e-05, | |
| "loss": 1.6394, | |
| "step": 9120 | |
| }, | |
| { | |
| "epoch": 0.9630801687763713, | |
| "grad_norm": 0.2857476472854614, | |
| "learning_rate": 2.4381573117307307e-05, | |
| "loss": 1.6289, | |
| "step": 9130 | |
| }, | |
| { | |
| "epoch": 0.9641350210970464, | |
| "grad_norm": 0.30220910906791687, | |
| "learning_rate": 2.4045510157554362e-05, | |
| "loss": 1.6498, | |
| "step": 9140 | |
| }, | |
| { | |
| "epoch": 0.9651898734177216, | |
| "grad_norm": 0.283668577671051, | |
| "learning_rate": 2.3714079315358985e-05, | |
| "loss": 1.6349, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 0.9662447257383966, | |
| "grad_norm": 0.28030630946159363, | |
| "learning_rate": 2.338721674401494e-05, | |
| "loss": 1.6387, | |
| "step": 9160 | |
| }, | |
| { | |
| "epoch": 0.9672995780590717, | |
| "grad_norm": 0.3106326758861542, | |
| "learning_rate": 2.30648594768459e-05, | |
| "loss": 1.6387, | |
| "step": 9170 | |
| }, | |
| { | |
| "epoch": 0.9683544303797469, | |
| "grad_norm": 0.28013819456100464, | |
| "learning_rate": 2.2746945415075523e-05, | |
| "loss": 1.6359, | |
| "step": 9180 | |
| }, | |
| { | |
| "epoch": 0.9694092827004219, | |
| "grad_norm": 0.3039085268974304, | |
| "learning_rate": 2.2433413315864803e-05, | |
| "loss": 1.6322, | |
| "step": 9190 | |
| }, | |
| { | |
| "epoch": 0.9704641350210971, | |
| "grad_norm": 0.28467583656311035, | |
| "learning_rate": 2.2124202780514277e-05, | |
| "loss": 1.6338, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 0.9715189873417721, | |
| "grad_norm": 0.30532944202423096, | |
| "learning_rate": 2.1819254242828815e-05, | |
| "loss": 1.637, | |
| "step": 9210 | |
| }, | |
| { | |
| "epoch": 0.9725738396624473, | |
| "grad_norm": 0.2894130051136017, | |
| "learning_rate": 2.151850895764285e-05, | |
| "loss": 1.6342, | |
| "step": 9220 | |
| }, | |
| { | |
| "epoch": 0.9736286919831224, | |
| "grad_norm": 0.2728683352470398, | |
| "learning_rate": 2.12219089895037e-05, | |
| "loss": 1.625, | |
| "step": 9230 | |
| }, | |
| { | |
| "epoch": 0.9746835443037974, | |
| "grad_norm": 0.2716827988624573, | |
| "learning_rate": 2.092939720151092e-05, | |
| "loss": 1.618, | |
| "step": 9240 | |
| }, | |
| { | |
| "epoch": 0.9757383966244726, | |
| "grad_norm": 0.2797609865665436, | |
| "learning_rate": 2.064091724430947e-05, | |
| "loss": 1.6327, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 0.9767932489451476, | |
| "grad_norm": 0.28480952978134155, | |
| "learning_rate": 2.0356413545234603e-05, | |
| "loss": 1.6281, | |
| "step": 9260 | |
| }, | |
| { | |
| "epoch": 0.9778481012658228, | |
| "grad_norm": 0.2917044758796692, | |
| "learning_rate": 2.0075831297606357e-05, | |
| "loss": 1.6419, | |
| "step": 9270 | |
| }, | |
| { | |
| "epoch": 0.9789029535864979, | |
| "grad_norm": 0.30042940378189087, | |
| "learning_rate": 1.9799116450171627e-05, | |
| "loss": 1.6293, | |
| "step": 9280 | |
| }, | |
| { | |
| "epoch": 0.979957805907173, | |
| "grad_norm": 0.27563101053237915, | |
| "learning_rate": 1.952621569669175e-05, | |
| "loss": 1.6355, | |
| "step": 9290 | |
| }, | |
| { | |
| "epoch": 0.9810126582278481, | |
| "grad_norm": 0.29096660017967224, | |
| "learning_rate": 1.9257076465673605e-05, | |
| "loss": 1.6336, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 0.9820675105485233, | |
| "grad_norm": 0.29261574149131775, | |
| "learning_rate": 1.899164691024229e-05, | |
| "loss": 1.6336, | |
| "step": 9310 | |
| }, | |
| { | |
| "epoch": 0.9831223628691983, | |
| "grad_norm": 0.29520437121391296, | |
| "learning_rate": 1.872987589815331e-05, | |
| "loss": 1.6292, | |
| "step": 9320 | |
| }, | |
| { | |
| "epoch": 0.9841772151898734, | |
| "grad_norm": 0.3058977723121643, | |
| "learning_rate": 1.8471713001942538e-05, | |
| "loss": 1.6416, | |
| "step": 9330 | |
| }, | |
| { | |
| "epoch": 0.9852320675105485, | |
| "grad_norm": 0.28372088074684143, | |
| "learning_rate": 1.8217108489211845e-05, | |
| "loss": 1.6388, | |
| "step": 9340 | |
| }, | |
| { | |
| "epoch": 0.9862869198312236, | |
| "grad_norm": 0.2728956341743469, | |
| "learning_rate": 1.7966013313048696e-05, | |
| "loss": 1.6498, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 0.9873417721518988, | |
| "grad_norm": 0.29871150851249695, | |
| "learning_rate": 1.7718379102577752e-05, | |
| "loss": 1.6444, | |
| "step": 9360 | |
| }, | |
| { | |
| "epoch": 0.9883966244725738, | |
| "grad_norm": 0.2838023602962494, | |
| "learning_rate": 1.7474158153642745e-05, | |
| "loss": 1.6413, | |
| "step": 9370 | |
| }, | |
| { | |
| "epoch": 0.989451476793249, | |
| "grad_norm": 0.2850436568260193, | |
| "learning_rate": 1.7233303419616745e-05, | |
| "loss": 1.6275, | |
| "step": 9380 | |
| }, | |
| { | |
| "epoch": 0.990506329113924, | |
| "grad_norm": 0.28575748205184937, | |
| "learning_rate": 1.699576850233916e-05, | |
| "loss": 1.6324, | |
| "step": 9390 | |
| }, | |
| { | |
| "epoch": 0.9915611814345991, | |
| "grad_norm": 0.2765159010887146, | |
| "learning_rate": 1.6761507643177553e-05, | |
| "loss": 1.6433, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 0.9926160337552743, | |
| "grad_norm": 0.2996973395347595, | |
| "learning_rate": 1.6530475714212752e-05, | |
| "loss": 1.6162, | |
| "step": 9410 | |
| }, | |
| { | |
| "epoch": 0.9936708860759493, | |
| "grad_norm": 0.28517743945121765, | |
| "learning_rate": 1.6302628209545423e-05, | |
| "loss": 1.6353, | |
| "step": 9420 | |
| }, | |
| { | |
| "epoch": 0.9947257383966245, | |
| "grad_norm": 0.27998846769332886, | |
| "learning_rate": 1.6077921236722464e-05, | |
| "loss": 1.6418, | |
| "step": 9430 | |
| }, | |
| { | |
| "epoch": 0.9957805907172996, | |
| "grad_norm": 0.2945730984210968, | |
| "learning_rate": 1.5856311508281594e-05, | |
| "loss": 1.6342, | |
| "step": 9440 | |
| }, | |
| { | |
| "epoch": 0.9968354430379747, | |
| "grad_norm": 0.2894095778465271, | |
| "learning_rate": 1.5637756333412454e-05, | |
| "loss": 1.6449, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 0.9978902953586498, | |
| "grad_norm": 0.2997286021709442, | |
| "learning_rate": 1.542221360973268e-05, | |
| "loss": 1.6283, | |
| "step": 9460 | |
| }, | |
| { | |
| "epoch": 0.9989451476793249, | |
| "grad_norm": 0.2845168709754944, | |
| "learning_rate": 1.5209641815177312e-05, | |
| "loss": 1.6413, | |
| "step": 9470 | |
| }, | |
| { | |
| "epoch": 1.0, | |
| "grad_norm": 0.8419556021690369, | |
| "learning_rate": 1.5e-05, | |
| "loss": 1.6288, | |
| "step": 9480 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 9480, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 1, | |
| "save_steps": 1000, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": true | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 1.8317122291574784e+16, | |
| "train_batch_size": 1024, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |