| { | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 2.999028560326404, | |
| "eval_steps": 500, | |
| "global_step": 7719, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.0038857586943850785, | |
| "grad_norm": 6.868951908344346, | |
| "learning_rate": 5.181347150259067e-07, | |
| "loss": 1.0933, | |
| "step": 10 | |
| }, | |
| { | |
| "epoch": 0.007771517388770157, | |
| "grad_norm": 5.837615214745706, | |
| "learning_rate": 1.0362694300518134e-06, | |
| "loss": 1.0482, | |
| "step": 20 | |
| }, | |
| { | |
| "epoch": 0.011657276083155236, | |
| "grad_norm": 3.7584728670248264, | |
| "learning_rate": 1.5544041450777204e-06, | |
| "loss": 0.9528, | |
| "step": 30 | |
| }, | |
| { | |
| "epoch": 0.015543034777540314, | |
| "grad_norm": 2.430849231904627, | |
| "learning_rate": 2.072538860103627e-06, | |
| "loss": 0.7595, | |
| "step": 40 | |
| }, | |
| { | |
| "epoch": 0.019428793471925394, | |
| "grad_norm": 1.9877795062873533, | |
| "learning_rate": 2.5906735751295338e-06, | |
| "loss": 0.5683, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.02331455216631047, | |
| "grad_norm": 1.1748343816269307, | |
| "learning_rate": 3.1088082901554407e-06, | |
| "loss": 0.393, | |
| "step": 60 | |
| }, | |
| { | |
| "epoch": 0.02720031086069555, | |
| "grad_norm": 1.2009973092317858, | |
| "learning_rate": 3.6269430051813476e-06, | |
| "loss": 0.2619, | |
| "step": 70 | |
| }, | |
| { | |
| "epoch": 0.031086069555080628, | |
| "grad_norm": 1.0407742203046373, | |
| "learning_rate": 4.145077720207254e-06, | |
| "loss": 0.2057, | |
| "step": 80 | |
| }, | |
| { | |
| "epoch": 0.03497182824946571, | |
| "grad_norm": 1.1295981297490019, | |
| "learning_rate": 4.663212435233161e-06, | |
| "loss": 0.1671, | |
| "step": 90 | |
| }, | |
| { | |
| "epoch": 0.03885758694385079, | |
| "grad_norm": 1.1504503826508636, | |
| "learning_rate": 5.1813471502590676e-06, | |
| "loss": 0.155, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.04274334563823587, | |
| "grad_norm": 0.7676430272864883, | |
| "learning_rate": 5.699481865284975e-06, | |
| "loss": 0.1451, | |
| "step": 110 | |
| }, | |
| { | |
| "epoch": 0.04662910433262094, | |
| "grad_norm": 1.1281617722679143, | |
| "learning_rate": 6.217616580310881e-06, | |
| "loss": 0.1436, | |
| "step": 120 | |
| }, | |
| { | |
| "epoch": 0.05051486302700602, | |
| "grad_norm": 1.042258959715133, | |
| "learning_rate": 6.735751295336788e-06, | |
| "loss": 0.1295, | |
| "step": 130 | |
| }, | |
| { | |
| "epoch": 0.0544006217213911, | |
| "grad_norm": 1.0836324717622774, | |
| "learning_rate": 7.253886010362695e-06, | |
| "loss": 0.1149, | |
| "step": 140 | |
| }, | |
| { | |
| "epoch": 0.05828638041577618, | |
| "grad_norm": 1.4204705637615596, | |
| "learning_rate": 7.772020725388602e-06, | |
| "loss": 0.1268, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.062172139110161256, | |
| "grad_norm": 1.0083666041980668, | |
| "learning_rate": 8.290155440414507e-06, | |
| "loss": 0.1182, | |
| "step": 160 | |
| }, | |
| { | |
| "epoch": 0.06605789780454634, | |
| "grad_norm": 0.833955996150633, | |
| "learning_rate": 8.808290155440415e-06, | |
| "loss": 0.1037, | |
| "step": 170 | |
| }, | |
| { | |
| "epoch": 0.06994365649893142, | |
| "grad_norm": 0.762829972908706, | |
| "learning_rate": 9.326424870466322e-06, | |
| "loss": 0.1047, | |
| "step": 180 | |
| }, | |
| { | |
| "epoch": 0.0738294151933165, | |
| "grad_norm": 1.018781373894918, | |
| "learning_rate": 9.844559585492228e-06, | |
| "loss": 0.1059, | |
| "step": 190 | |
| }, | |
| { | |
| "epoch": 0.07771517388770158, | |
| "grad_norm": 0.8149796744145043, | |
| "learning_rate": 1.0362694300518135e-05, | |
| "loss": 0.1094, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.08160093258208666, | |
| "grad_norm": 1.0562790321924551, | |
| "learning_rate": 1.0880829015544042e-05, | |
| "loss": 0.0978, | |
| "step": 210 | |
| }, | |
| { | |
| "epoch": 0.08548669127647174, | |
| "grad_norm": 0.8392051142898549, | |
| "learning_rate": 1.139896373056995e-05, | |
| "loss": 0.1074, | |
| "step": 220 | |
| }, | |
| { | |
| "epoch": 0.0893724499708568, | |
| "grad_norm": 0.8606053025510271, | |
| "learning_rate": 1.1917098445595855e-05, | |
| "loss": 0.0936, | |
| "step": 230 | |
| }, | |
| { | |
| "epoch": 0.09325820866524188, | |
| "grad_norm": 0.8010752114927957, | |
| "learning_rate": 1.2435233160621763e-05, | |
| "loss": 0.084, | |
| "step": 240 | |
| }, | |
| { | |
| "epoch": 0.09714396735962696, | |
| "grad_norm": 0.8865363433827388, | |
| "learning_rate": 1.2953367875647668e-05, | |
| "loss": 0.0849, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.10102972605401205, | |
| "grad_norm": 1.0150630528923608, | |
| "learning_rate": 1.3471502590673576e-05, | |
| "loss": 0.0941, | |
| "step": 260 | |
| }, | |
| { | |
| "epoch": 0.10491548474839713, | |
| "grad_norm": 1.1564503498169056, | |
| "learning_rate": 1.3989637305699483e-05, | |
| "loss": 0.0884, | |
| "step": 270 | |
| }, | |
| { | |
| "epoch": 0.1088012434427822, | |
| "grad_norm": 0.7425602409996528, | |
| "learning_rate": 1.450777202072539e-05, | |
| "loss": 0.0904, | |
| "step": 280 | |
| }, | |
| { | |
| "epoch": 0.11268700213716729, | |
| "grad_norm": 1.010495639141353, | |
| "learning_rate": 1.5025906735751296e-05, | |
| "loss": 0.0877, | |
| "step": 290 | |
| }, | |
| { | |
| "epoch": 0.11657276083155237, | |
| "grad_norm": 0.7312574416672747, | |
| "learning_rate": 1.5544041450777204e-05, | |
| "loss": 0.0833, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.12045851952593743, | |
| "grad_norm": 0.8186188565897778, | |
| "learning_rate": 1.606217616580311e-05, | |
| "loss": 0.082, | |
| "step": 310 | |
| }, | |
| { | |
| "epoch": 0.12434427822032251, | |
| "grad_norm": 0.6725849359490335, | |
| "learning_rate": 1.6580310880829015e-05, | |
| "loss": 0.0773, | |
| "step": 320 | |
| }, | |
| { | |
| "epoch": 0.1282300369147076, | |
| "grad_norm": 0.6044289285905357, | |
| "learning_rate": 1.7098445595854924e-05, | |
| "loss": 0.0793, | |
| "step": 330 | |
| }, | |
| { | |
| "epoch": 0.13211579560909267, | |
| "grad_norm": 0.648628136760092, | |
| "learning_rate": 1.761658031088083e-05, | |
| "loss": 0.0748, | |
| "step": 340 | |
| }, | |
| { | |
| "epoch": 0.13600155430347777, | |
| "grad_norm": 0.7654085124832696, | |
| "learning_rate": 1.813471502590674e-05, | |
| "loss": 0.0656, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.13988731299786283, | |
| "grad_norm": 0.6998519217873228, | |
| "learning_rate": 1.8652849740932644e-05, | |
| "loss": 0.0841, | |
| "step": 360 | |
| }, | |
| { | |
| "epoch": 0.1437730716922479, | |
| "grad_norm": 0.5847858909362429, | |
| "learning_rate": 1.917098445595855e-05, | |
| "loss": 0.0692, | |
| "step": 370 | |
| }, | |
| { | |
| "epoch": 0.147658830386633, | |
| "grad_norm": 0.7494257595598578, | |
| "learning_rate": 1.9689119170984456e-05, | |
| "loss": 0.0783, | |
| "step": 380 | |
| }, | |
| { | |
| "epoch": 0.15154458908101806, | |
| "grad_norm": 0.7368291090701112, | |
| "learning_rate": 2.0207253886010365e-05, | |
| "loss": 0.0672, | |
| "step": 390 | |
| }, | |
| { | |
| "epoch": 0.15543034777540315, | |
| "grad_norm": 0.6162219355087554, | |
| "learning_rate": 2.072538860103627e-05, | |
| "loss": 0.0768, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.15931610646978822, | |
| "grad_norm": 1.0522389801039482, | |
| "learning_rate": 2.124352331606218e-05, | |
| "loss": 0.0753, | |
| "step": 410 | |
| }, | |
| { | |
| "epoch": 0.16320186516417332, | |
| "grad_norm": 0.6631859133005514, | |
| "learning_rate": 2.1761658031088085e-05, | |
| "loss": 0.0853, | |
| "step": 420 | |
| }, | |
| { | |
| "epoch": 0.16708762385855838, | |
| "grad_norm": 0.8284655987791505, | |
| "learning_rate": 2.227979274611399e-05, | |
| "loss": 0.0644, | |
| "step": 430 | |
| }, | |
| { | |
| "epoch": 0.17097338255294348, | |
| "grad_norm": 0.6911832435770443, | |
| "learning_rate": 2.27979274611399e-05, | |
| "loss": 0.0623, | |
| "step": 440 | |
| }, | |
| { | |
| "epoch": 0.17485914124732854, | |
| "grad_norm": 0.5902621273486757, | |
| "learning_rate": 2.3316062176165802e-05, | |
| "loss": 0.077, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.1787448999417136, | |
| "grad_norm": 0.7554872423202403, | |
| "learning_rate": 2.383419689119171e-05, | |
| "loss": 0.0798, | |
| "step": 460 | |
| }, | |
| { | |
| "epoch": 0.1826306586360987, | |
| "grad_norm": 0.6639542635427919, | |
| "learning_rate": 2.435233160621762e-05, | |
| "loss": 0.0786, | |
| "step": 470 | |
| }, | |
| { | |
| "epoch": 0.18651641733048377, | |
| "grad_norm": 0.7784648823322334, | |
| "learning_rate": 2.4870466321243526e-05, | |
| "loss": 0.0814, | |
| "step": 480 | |
| }, | |
| { | |
| "epoch": 0.19040217602486886, | |
| "grad_norm": 0.6900832905628588, | |
| "learning_rate": 2.5388601036269435e-05, | |
| "loss": 0.0787, | |
| "step": 490 | |
| }, | |
| { | |
| "epoch": 0.19428793471925393, | |
| "grad_norm": 0.6434291181769316, | |
| "learning_rate": 2.5906735751295337e-05, | |
| "loss": 0.0694, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.19817369341363902, | |
| "grad_norm": 0.6727244264583743, | |
| "learning_rate": 2.6424870466321246e-05, | |
| "loss": 0.0696, | |
| "step": 510 | |
| }, | |
| { | |
| "epoch": 0.2020594521080241, | |
| "grad_norm": 0.5145508471366054, | |
| "learning_rate": 2.694300518134715e-05, | |
| "loss": 0.0762, | |
| "step": 520 | |
| }, | |
| { | |
| "epoch": 0.20594521080240916, | |
| "grad_norm": 0.5458188017984406, | |
| "learning_rate": 2.746113989637306e-05, | |
| "loss": 0.067, | |
| "step": 530 | |
| }, | |
| { | |
| "epoch": 0.20983096949679425, | |
| "grad_norm": 0.615156585002907, | |
| "learning_rate": 2.7979274611398966e-05, | |
| "loss": 0.0674, | |
| "step": 540 | |
| }, | |
| { | |
| "epoch": 0.21371672819117932, | |
| "grad_norm": 0.5503543665259718, | |
| "learning_rate": 2.8497409326424872e-05, | |
| "loss": 0.0764, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.2176024868855644, | |
| "grad_norm": 0.6637963707127615, | |
| "learning_rate": 2.901554404145078e-05, | |
| "loss": 0.0666, | |
| "step": 560 | |
| }, | |
| { | |
| "epoch": 0.22148824557994948, | |
| "grad_norm": 0.5116291190440031, | |
| "learning_rate": 2.9533678756476683e-05, | |
| "loss": 0.0648, | |
| "step": 570 | |
| }, | |
| { | |
| "epoch": 0.22537400427433457, | |
| "grad_norm": 0.6270451144717418, | |
| "learning_rate": 3.0051813471502592e-05, | |
| "loss": 0.0634, | |
| "step": 580 | |
| }, | |
| { | |
| "epoch": 0.22925976296871964, | |
| "grad_norm": 0.5329413421123648, | |
| "learning_rate": 3.0569948186528505e-05, | |
| "loss": 0.0605, | |
| "step": 590 | |
| }, | |
| { | |
| "epoch": 0.23314552166310473, | |
| "grad_norm": 0.7863078930627488, | |
| "learning_rate": 3.108808290155441e-05, | |
| "loss": 0.0611, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.2370312803574898, | |
| "grad_norm": 0.5641860345765728, | |
| "learning_rate": 3.1606217616580316e-05, | |
| "loss": 0.0713, | |
| "step": 610 | |
| }, | |
| { | |
| "epoch": 0.24091703905187487, | |
| "grad_norm": 0.6298950201322868, | |
| "learning_rate": 3.212435233160622e-05, | |
| "loss": 0.062, | |
| "step": 620 | |
| }, | |
| { | |
| "epoch": 0.24480279774625996, | |
| "grad_norm": 0.5576657376337026, | |
| "learning_rate": 3.264248704663213e-05, | |
| "loss": 0.0729, | |
| "step": 630 | |
| }, | |
| { | |
| "epoch": 0.24868855644064503, | |
| "grad_norm": 0.5849778617897893, | |
| "learning_rate": 3.316062176165803e-05, | |
| "loss": 0.0663, | |
| "step": 640 | |
| }, | |
| { | |
| "epoch": 0.2525743151350301, | |
| "grad_norm": 0.6862718900926282, | |
| "learning_rate": 3.367875647668394e-05, | |
| "loss": 0.0702, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.2564600738294152, | |
| "grad_norm": 0.620792145685114, | |
| "learning_rate": 3.419689119170985e-05, | |
| "loss": 0.0647, | |
| "step": 660 | |
| }, | |
| { | |
| "epoch": 0.2603458325238003, | |
| "grad_norm": 0.5377947147204513, | |
| "learning_rate": 3.471502590673575e-05, | |
| "loss": 0.0629, | |
| "step": 670 | |
| }, | |
| { | |
| "epoch": 0.26423159121818535, | |
| "grad_norm": 0.5488958030608141, | |
| "learning_rate": 3.523316062176166e-05, | |
| "loss": 0.0671, | |
| "step": 680 | |
| }, | |
| { | |
| "epoch": 0.2681173499125704, | |
| "grad_norm": 0.5954204106913672, | |
| "learning_rate": 3.575129533678757e-05, | |
| "loss": 0.064, | |
| "step": 690 | |
| }, | |
| { | |
| "epoch": 0.27200310860695553, | |
| "grad_norm": 0.5519696206558653, | |
| "learning_rate": 3.626943005181348e-05, | |
| "loss": 0.055, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.2758888673013406, | |
| "grad_norm": 0.6671381342735235, | |
| "learning_rate": 3.6787564766839386e-05, | |
| "loss": 0.0566, | |
| "step": 710 | |
| }, | |
| { | |
| "epoch": 0.27977462599572567, | |
| "grad_norm": 0.46839311589453947, | |
| "learning_rate": 3.730569948186529e-05, | |
| "loss": 0.0655, | |
| "step": 720 | |
| }, | |
| { | |
| "epoch": 0.28366038469011073, | |
| "grad_norm": 0.49157316560679837, | |
| "learning_rate": 3.78238341968912e-05, | |
| "loss": 0.0535, | |
| "step": 730 | |
| }, | |
| { | |
| "epoch": 0.2875461433844958, | |
| "grad_norm": 0.49892347263304754, | |
| "learning_rate": 3.83419689119171e-05, | |
| "loss": 0.0674, | |
| "step": 740 | |
| }, | |
| { | |
| "epoch": 0.2914319020788809, | |
| "grad_norm": 0.5208448100813172, | |
| "learning_rate": 3.886010362694301e-05, | |
| "loss": 0.0511, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.295317660773266, | |
| "grad_norm": 0.2894917388721045, | |
| "learning_rate": 3.937823834196891e-05, | |
| "loss": 0.0568, | |
| "step": 760 | |
| }, | |
| { | |
| "epoch": 0.29920341946765106, | |
| "grad_norm": 0.532926439876675, | |
| "learning_rate": 3.989637305699482e-05, | |
| "loss": 0.0565, | |
| "step": 770 | |
| }, | |
| { | |
| "epoch": 0.3030891781620361, | |
| "grad_norm": 0.49821621909804464, | |
| "learning_rate": 3.999986911657599e-05, | |
| "loss": 0.0718, | |
| "step": 780 | |
| }, | |
| { | |
| "epoch": 0.30697493685642124, | |
| "grad_norm": 0.4380644811650154, | |
| "learning_rate": 3.99993374056019e-05, | |
| "loss": 0.0578, | |
| "step": 790 | |
| }, | |
| { | |
| "epoch": 0.3108606955508063, | |
| "grad_norm": 0.47585349907521746, | |
| "learning_rate": 3.999839669772912e-05, | |
| "loss": 0.0523, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.3147464542451914, | |
| "grad_norm": 0.31641494499233286, | |
| "learning_rate": 3.999704701219562e-05, | |
| "loss": 0.0563, | |
| "step": 810 | |
| }, | |
| { | |
| "epoch": 0.31863221293957644, | |
| "grad_norm": 0.48142926317508933, | |
| "learning_rate": 3.999528837660319e-05, | |
| "loss": 0.0587, | |
| "step": 820 | |
| }, | |
| { | |
| "epoch": 0.3225179716339615, | |
| "grad_norm": 0.48754014218594893, | |
| "learning_rate": 3.999312082691682e-05, | |
| "loss": 0.0659, | |
| "step": 830 | |
| }, | |
| { | |
| "epoch": 0.32640373032834663, | |
| "grad_norm": 0.6302273551444069, | |
| "learning_rate": 3.9990544407464044e-05, | |
| "loss": 0.0617, | |
| "step": 840 | |
| }, | |
| { | |
| "epoch": 0.3302894890227317, | |
| "grad_norm": 0.4908822983385492, | |
| "learning_rate": 3.9987559170934e-05, | |
| "loss": 0.0573, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.33417524771711676, | |
| "grad_norm": 0.3297602359842017, | |
| "learning_rate": 3.9984165178376316e-05, | |
| "loss": 0.0491, | |
| "step": 860 | |
| }, | |
| { | |
| "epoch": 0.33806100641150183, | |
| "grad_norm": 0.419433697037015, | |
| "learning_rate": 3.9980362499199915e-05, | |
| "loss": 0.0555, | |
| "step": 870 | |
| }, | |
| { | |
| "epoch": 0.34194676510588695, | |
| "grad_norm": 0.5429807682582697, | |
| "learning_rate": 3.997615121117159e-05, | |
| "loss": 0.0501, | |
| "step": 880 | |
| }, | |
| { | |
| "epoch": 0.345832523800272, | |
| "grad_norm": 0.5045801179188242, | |
| "learning_rate": 3.997153140041437e-05, | |
| "loss": 0.0527, | |
| "step": 890 | |
| }, | |
| { | |
| "epoch": 0.3497182824946571, | |
| "grad_norm": 0.4119373168535617, | |
| "learning_rate": 3.9966503161405786e-05, | |
| "loss": 0.0579, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.35360404118904215, | |
| "grad_norm": 0.3652010084324409, | |
| "learning_rate": 3.996106659697597e-05, | |
| "loss": 0.0394, | |
| "step": 910 | |
| }, | |
| { | |
| "epoch": 0.3574897998834272, | |
| "grad_norm": 0.49911111921282303, | |
| "learning_rate": 3.9955221818305504e-05, | |
| "loss": 0.0618, | |
| "step": 920 | |
| }, | |
| { | |
| "epoch": 0.36137555857781234, | |
| "grad_norm": 0.4691524202220399, | |
| "learning_rate": 3.994896894492316e-05, | |
| "loss": 0.0449, | |
| "step": 930 | |
| }, | |
| { | |
| "epoch": 0.3652613172721974, | |
| "grad_norm": 0.4522219450999137, | |
| "learning_rate": 3.9942308104703464e-05, | |
| "loss": 0.0637, | |
| "step": 940 | |
| }, | |
| { | |
| "epoch": 0.36914707596658247, | |
| "grad_norm": 0.3759169728636205, | |
| "learning_rate": 3.993523943386408e-05, | |
| "loss": 0.0556, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.37303283466096754, | |
| "grad_norm": 0.4043985895545187, | |
| "learning_rate": 3.9927763076963026e-05, | |
| "loss": 0.0497, | |
| "step": 960 | |
| }, | |
| { | |
| "epoch": 0.37691859335535266, | |
| "grad_norm": 0.36167729094148904, | |
| "learning_rate": 3.99198791868957e-05, | |
| "loss": 0.0443, | |
| "step": 970 | |
| }, | |
| { | |
| "epoch": 0.3808043520497377, | |
| "grad_norm": 0.43188183331177943, | |
| "learning_rate": 3.991158792489178e-05, | |
| "loss": 0.0492, | |
| "step": 980 | |
| }, | |
| { | |
| "epoch": 0.3846901107441228, | |
| "grad_norm": 0.591934470494083, | |
| "learning_rate": 3.9902889460511895e-05, | |
| "loss": 0.0433, | |
| "step": 990 | |
| }, | |
| { | |
| "epoch": 0.38857586943850786, | |
| "grad_norm": 0.38540473325489366, | |
| "learning_rate": 3.989378397164419e-05, | |
| "loss": 0.062, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.3924616281328929, | |
| "grad_norm": 0.4686847716122605, | |
| "learning_rate": 3.988427164450067e-05, | |
| "loss": 0.053, | |
| "step": 1010 | |
| }, | |
| { | |
| "epoch": 0.39634738682727805, | |
| "grad_norm": 0.44830904799588184, | |
| "learning_rate": 3.98743526736134e-05, | |
| "loss": 0.0546, | |
| "step": 1020 | |
| }, | |
| { | |
| "epoch": 0.4002331455216631, | |
| "grad_norm": 0.38427649669131925, | |
| "learning_rate": 3.986402726183051e-05, | |
| "loss": 0.044, | |
| "step": 1030 | |
| }, | |
| { | |
| "epoch": 0.4041189042160482, | |
| "grad_norm": 0.3797043059039449, | |
| "learning_rate": 3.985329562031207e-05, | |
| "loss": 0.0507, | |
| "step": 1040 | |
| }, | |
| { | |
| "epoch": 0.40800466291043325, | |
| "grad_norm": 0.4645705602868689, | |
| "learning_rate": 3.9842157968525755e-05, | |
| "loss": 0.0488, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.4118904216048183, | |
| "grad_norm": 0.45995234050567774, | |
| "learning_rate": 3.9830614534242365e-05, | |
| "loss": 0.0504, | |
| "step": 1060 | |
| }, | |
| { | |
| "epoch": 0.41577618029920344, | |
| "grad_norm": 0.6048279655888865, | |
| "learning_rate": 3.981866555353115e-05, | |
| "loss": 0.0554, | |
| "step": 1070 | |
| }, | |
| { | |
| "epoch": 0.4196619389935885, | |
| "grad_norm": 0.4778278100735776, | |
| "learning_rate": 3.9806311270755026e-05, | |
| "loss": 0.0504, | |
| "step": 1080 | |
| }, | |
| { | |
| "epoch": 0.42354769768797357, | |
| "grad_norm": 0.2819163749757861, | |
| "learning_rate": 3.9793551938565513e-05, | |
| "loss": 0.0473, | |
| "step": 1090 | |
| }, | |
| { | |
| "epoch": 0.42743345638235863, | |
| "grad_norm": 0.34078755188718823, | |
| "learning_rate": 3.978038781789764e-05, | |
| "loss": 0.0513, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.43131921507674376, | |
| "grad_norm": 0.42583054736380277, | |
| "learning_rate": 3.9766819177964535e-05, | |
| "loss": 0.0469, | |
| "step": 1110 | |
| }, | |
| { | |
| "epoch": 0.4352049737711288, | |
| "grad_norm": 0.5037049238340003, | |
| "learning_rate": 3.975284629625198e-05, | |
| "loss": 0.0552, | |
| "step": 1120 | |
| }, | |
| { | |
| "epoch": 0.4390907324655139, | |
| "grad_norm": 0.4127984535316269, | |
| "learning_rate": 3.973846945851271e-05, | |
| "loss": 0.0431, | |
| "step": 1130 | |
| }, | |
| { | |
| "epoch": 0.44297649115989896, | |
| "grad_norm": 0.4024463238112972, | |
| "learning_rate": 3.972368895876056e-05, | |
| "loss": 0.0442, | |
| "step": 1140 | |
| }, | |
| { | |
| "epoch": 0.446862249854284, | |
| "grad_norm": 0.34036792334896143, | |
| "learning_rate": 3.970850509926448e-05, | |
| "loss": 0.0429, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.45074800854866914, | |
| "grad_norm": 0.43486746219653427, | |
| "learning_rate": 3.969291819054232e-05, | |
| "loss": 0.0465, | |
| "step": 1160 | |
| }, | |
| { | |
| "epoch": 0.4546337672430542, | |
| "grad_norm": 0.3533700876931675, | |
| "learning_rate": 3.9676928551354524e-05, | |
| "loss": 0.0442, | |
| "step": 1170 | |
| }, | |
| { | |
| "epoch": 0.4585195259374393, | |
| "grad_norm": 0.4418640383769793, | |
| "learning_rate": 3.9660536508697545e-05, | |
| "loss": 0.0433, | |
| "step": 1180 | |
| }, | |
| { | |
| "epoch": 0.46240528463182434, | |
| "grad_norm": 0.45970453713501874, | |
| "learning_rate": 3.9643742397797236e-05, | |
| "loss": 0.0511, | |
| "step": 1190 | |
| }, | |
| { | |
| "epoch": 0.46629104332620946, | |
| "grad_norm": 0.35306399118899623, | |
| "learning_rate": 3.9626546562101936e-05, | |
| "loss": 0.0508, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.47017680202059453, | |
| "grad_norm": 0.3211095806787999, | |
| "learning_rate": 3.960894935327546e-05, | |
| "loss": 0.039, | |
| "step": 1210 | |
| }, | |
| { | |
| "epoch": 0.4740625607149796, | |
| "grad_norm": 0.36627785482679437, | |
| "learning_rate": 3.9590951131189934e-05, | |
| "loss": 0.0558, | |
| "step": 1220 | |
| }, | |
| { | |
| "epoch": 0.47794831940936466, | |
| "grad_norm": 0.421462123124338, | |
| "learning_rate": 3.957255226391839e-05, | |
| "loss": 0.0497, | |
| "step": 1230 | |
| }, | |
| { | |
| "epoch": 0.48183407810374973, | |
| "grad_norm": 0.3353485956514776, | |
| "learning_rate": 3.955375312772729e-05, | |
| "loss": 0.0384, | |
| "step": 1240 | |
| }, | |
| { | |
| "epoch": 0.48571983679813485, | |
| "grad_norm": 11.666925837495032, | |
| "learning_rate": 3.9534554107068786e-05, | |
| "loss": 0.0472, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.4896055954925199, | |
| "grad_norm": 0.4841474924556115, | |
| "learning_rate": 3.9514955594572874e-05, | |
| "loss": 0.0487, | |
| "step": 1260 | |
| }, | |
| { | |
| "epoch": 0.493491354186905, | |
| "grad_norm": 0.3248614577525704, | |
| "learning_rate": 3.9494957991039386e-05, | |
| "loss": 0.0548, | |
| "step": 1270 | |
| }, | |
| { | |
| "epoch": 0.49737711288129005, | |
| "grad_norm": 0.3970898472778102, | |
| "learning_rate": 3.947456170542976e-05, | |
| "loss": 0.0489, | |
| "step": 1280 | |
| }, | |
| { | |
| "epoch": 0.5012628715756752, | |
| "grad_norm": 0.4161749757032391, | |
| "learning_rate": 3.945376715485868e-05, | |
| "loss": 0.0526, | |
| "step": 1290 | |
| }, | |
| { | |
| "epoch": 0.5051486302700602, | |
| "grad_norm": 0.3452729447702609, | |
| "learning_rate": 3.9432574764585574e-05, | |
| "loss": 0.048, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.5090343889644453, | |
| "grad_norm": 0.47170016327450176, | |
| "learning_rate": 3.9410984968005904e-05, | |
| "loss": 0.0405, | |
| "step": 1310 | |
| }, | |
| { | |
| "epoch": 0.5129201476588304, | |
| "grad_norm": 0.42610644718786006, | |
| "learning_rate": 3.938899820664229e-05, | |
| "loss": 0.0458, | |
| "step": 1320 | |
| }, | |
| { | |
| "epoch": 0.5168059063532154, | |
| "grad_norm": 0.3408319733233165, | |
| "learning_rate": 3.936661493013548e-05, | |
| "loss": 0.0391, | |
| "step": 1330 | |
| }, | |
| { | |
| "epoch": 0.5206916650476006, | |
| "grad_norm": 0.30711332622300047, | |
| "learning_rate": 3.934383559623518e-05, | |
| "loss": 0.042, | |
| "step": 1340 | |
| }, | |
| { | |
| "epoch": 0.5245774237419856, | |
| "grad_norm": 0.5571217272080251, | |
| "learning_rate": 3.932066067079066e-05, | |
| "loss": 0.0431, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.5284631824363707, | |
| "grad_norm": 0.4027909880638002, | |
| "learning_rate": 3.929709062774127e-05, | |
| "loss": 0.0374, | |
| "step": 1360 | |
| }, | |
| { | |
| "epoch": 0.5323489411307558, | |
| "grad_norm": 0.30456945378456746, | |
| "learning_rate": 3.9273125949106675e-05, | |
| "loss": 0.0424, | |
| "step": 1370 | |
| }, | |
| { | |
| "epoch": 0.5362346998251408, | |
| "grad_norm": 0.3964517011394958, | |
| "learning_rate": 3.924876712497711e-05, | |
| "loss": 0.0467, | |
| "step": 1380 | |
| }, | |
| { | |
| "epoch": 0.540120458519526, | |
| "grad_norm": 0.3649307036795635, | |
| "learning_rate": 3.9224014653503226e-05, | |
| "loss": 0.0497, | |
| "step": 1390 | |
| }, | |
| { | |
| "epoch": 0.5440062172139111, | |
| "grad_norm": 0.3198268167473866, | |
| "learning_rate": 3.919886904088601e-05, | |
| "loss": 0.0456, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.5478919759082961, | |
| "grad_norm": 0.3463770316931819, | |
| "learning_rate": 3.917333080136638e-05, | |
| "loss": 0.0368, | |
| "step": 1410 | |
| }, | |
| { | |
| "epoch": 0.5517777346026812, | |
| "grad_norm": 0.35644328141911197, | |
| "learning_rate": 3.9147400457214674e-05, | |
| "loss": 0.0481, | |
| "step": 1420 | |
| }, | |
| { | |
| "epoch": 0.5556634932970662, | |
| "grad_norm": 0.40914490698280265, | |
| "learning_rate": 3.9121078538719975e-05, | |
| "loss": 0.0474, | |
| "step": 1430 | |
| }, | |
| { | |
| "epoch": 0.5595492519914513, | |
| "grad_norm": 0.3427548933748361, | |
| "learning_rate": 3.9094365584179264e-05, | |
| "loss": 0.0385, | |
| "step": 1440 | |
| }, | |
| { | |
| "epoch": 0.5634350106858365, | |
| "grad_norm": 0.4352394889252363, | |
| "learning_rate": 3.906726213988642e-05, | |
| "loss": 0.0383, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.5673207693802215, | |
| "grad_norm": 0.29442224609914946, | |
| "learning_rate": 3.903976876012105e-05, | |
| "loss": 0.0499, | |
| "step": 1460 | |
| }, | |
| { | |
| "epoch": 0.5712065280746066, | |
| "grad_norm": 0.2407859708001586, | |
| "learning_rate": 3.901188600713712e-05, | |
| "loss": 0.0459, | |
| "step": 1470 | |
| }, | |
| { | |
| "epoch": 0.5750922867689916, | |
| "grad_norm": 0.3994426590406398, | |
| "learning_rate": 3.89836144511515e-05, | |
| "loss": 0.0399, | |
| "step": 1480 | |
| }, | |
| { | |
| "epoch": 0.5789780454633767, | |
| "grad_norm": 0.3839601198021419, | |
| "learning_rate": 3.895495467033229e-05, | |
| "loss": 0.0455, | |
| "step": 1490 | |
| }, | |
| { | |
| "epoch": 0.5828638041577618, | |
| "grad_norm": 0.5414728833934257, | |
| "learning_rate": 3.8925907250786966e-05, | |
| "loss": 0.0428, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.5867495628521469, | |
| "grad_norm": 0.4855025115080349, | |
| "learning_rate": 3.8896472786550444e-05, | |
| "loss": 0.0415, | |
| "step": 1510 | |
| }, | |
| { | |
| "epoch": 0.590635321546532, | |
| "grad_norm": 0.3542282158759638, | |
| "learning_rate": 3.886665187957289e-05, | |
| "loss": 0.05, | |
| "step": 1520 | |
| }, | |
| { | |
| "epoch": 0.594521080240917, | |
| "grad_norm": 0.32139501424521916, | |
| "learning_rate": 3.883644513970744e-05, | |
| "loss": 0.0386, | |
| "step": 1530 | |
| }, | |
| { | |
| "epoch": 0.5984068389353021, | |
| "grad_norm": 0.32361000000718626, | |
| "learning_rate": 3.8805853184697694e-05, | |
| "loss": 0.0407, | |
| "step": 1540 | |
| }, | |
| { | |
| "epoch": 0.6022925976296872, | |
| "grad_norm": 0.2699532107849867, | |
| "learning_rate": 3.877487664016513e-05, | |
| "loss": 0.035, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.6061783563240722, | |
| "grad_norm": 0.31749658242328194, | |
| "learning_rate": 3.8743516139596244e-05, | |
| "loss": 0.0449, | |
| "step": 1560 | |
| }, | |
| { | |
| "epoch": 0.6100641150184574, | |
| "grad_norm": 0.38042009032251006, | |
| "learning_rate": 3.871177232432969e-05, | |
| "loss": 0.0433, | |
| "step": 1570 | |
| }, | |
| { | |
| "epoch": 0.6139498737128425, | |
| "grad_norm": 0.27994580128185775, | |
| "learning_rate": 3.867964584354305e-05, | |
| "loss": 0.0375, | |
| "step": 1580 | |
| }, | |
| { | |
| "epoch": 0.6178356324072275, | |
| "grad_norm": 0.44326692252464484, | |
| "learning_rate": 3.864713735423964e-05, | |
| "loss": 0.0527, | |
| "step": 1590 | |
| }, | |
| { | |
| "epoch": 0.6217213911016126, | |
| "grad_norm": 0.338368734908726, | |
| "learning_rate": 3.861424752123506e-05, | |
| "loss": 0.0356, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.6256071497959976, | |
| "grad_norm": 0.31262748879439145, | |
| "learning_rate": 3.858097701714358e-05, | |
| "loss": 0.0391, | |
| "step": 1610 | |
| }, | |
| { | |
| "epoch": 0.6294929084903828, | |
| "grad_norm": 0.37350424104203495, | |
| "learning_rate": 3.8547326522364386e-05, | |
| "loss": 0.0437, | |
| "step": 1620 | |
| }, | |
| { | |
| "epoch": 0.6333786671847679, | |
| "grad_norm": 0.3610320718135361, | |
| "learning_rate": 3.851329672506768e-05, | |
| "loss": 0.0412, | |
| "step": 1630 | |
| }, | |
| { | |
| "epoch": 0.6372644258791529, | |
| "grad_norm": 0.46752288451915147, | |
| "learning_rate": 3.847888832118059e-05, | |
| "loss": 0.0426, | |
| "step": 1640 | |
| }, | |
| { | |
| "epoch": 0.641150184573538, | |
| "grad_norm": 0.42134006799700907, | |
| "learning_rate": 3.844410201437296e-05, | |
| "loss": 0.0598, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.645035943267923, | |
| "grad_norm": 0.31290423259787065, | |
| "learning_rate": 3.840893851604294e-05, | |
| "loss": 0.0346, | |
| "step": 1660 | |
| }, | |
| { | |
| "epoch": 0.6489217019623081, | |
| "grad_norm": 0.40257644426810313, | |
| "learning_rate": 3.837339854530243e-05, | |
| "loss": 0.0452, | |
| "step": 1670 | |
| }, | |
| { | |
| "epoch": 0.6528074606566933, | |
| "grad_norm": 0.35775285851697575, | |
| "learning_rate": 3.833748282896241e-05, | |
| "loss": 0.0375, | |
| "step": 1680 | |
| }, | |
| { | |
| "epoch": 0.6566932193510783, | |
| "grad_norm": 0.39620710106307544, | |
| "learning_rate": 3.8301192101518034e-05, | |
| "loss": 0.0389, | |
| "step": 1690 | |
| }, | |
| { | |
| "epoch": 0.6605789780454634, | |
| "grad_norm": 0.35116612697915695, | |
| "learning_rate": 3.8264527105133655e-05, | |
| "loss": 0.0416, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.6644647367398484, | |
| "grad_norm": 0.3593763085620626, | |
| "learning_rate": 3.822748858962759e-05, | |
| "loss": 0.039, | |
| "step": 1710 | |
| }, | |
| { | |
| "epoch": 0.6683504954342335, | |
| "grad_norm": 0.3636137986839074, | |
| "learning_rate": 3.8190077312456837e-05, | |
| "loss": 0.0437, | |
| "step": 1720 | |
| }, | |
| { | |
| "epoch": 0.6722362541286186, | |
| "grad_norm": 0.3070384884366638, | |
| "learning_rate": 3.815229403870156e-05, | |
| "loss": 0.0441, | |
| "step": 1730 | |
| }, | |
| { | |
| "epoch": 0.6761220128230037, | |
| "grad_norm": 0.452855427559548, | |
| "learning_rate": 3.811413954104944e-05, | |
| "loss": 0.0476, | |
| "step": 1740 | |
| }, | |
| { | |
| "epoch": 0.6800077715173888, | |
| "grad_norm": 0.3645618682663316, | |
| "learning_rate": 3.80756145997799e-05, | |
| "loss": 0.044, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.6838935302117739, | |
| "grad_norm": 0.3236245721921845, | |
| "learning_rate": 3.8036720002748116e-05, | |
| "loss": 0.0392, | |
| "step": 1760 | |
| }, | |
| { | |
| "epoch": 0.6877792889061589, | |
| "grad_norm": 0.3755492942938037, | |
| "learning_rate": 3.79974565453689e-05, | |
| "loss": 0.0391, | |
| "step": 1770 | |
| }, | |
| { | |
| "epoch": 0.691665047600544, | |
| "grad_norm": 0.32387171524447206, | |
| "learning_rate": 3.795782503060049e-05, | |
| "loss": 0.0387, | |
| "step": 1780 | |
| }, | |
| { | |
| "epoch": 0.695550806294929, | |
| "grad_norm": 0.34576851613064813, | |
| "learning_rate": 3.791782626892806e-05, | |
| "loss": 0.041, | |
| "step": 1790 | |
| }, | |
| { | |
| "epoch": 0.6994365649893142, | |
| "grad_norm": 0.3841280667838816, | |
| "learning_rate": 3.7877461078347184e-05, | |
| "loss": 0.0421, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.7033223236836993, | |
| "grad_norm": 0.3143570778817021, | |
| "learning_rate": 3.78367302843471e-05, | |
| "loss": 0.0471, | |
| "step": 1810 | |
| }, | |
| { | |
| "epoch": 0.7072080823780843, | |
| "grad_norm": 0.27227430217810716, | |
| "learning_rate": 3.7795634719893824e-05, | |
| "loss": 0.0368, | |
| "step": 1820 | |
| }, | |
| { | |
| "epoch": 0.7110938410724694, | |
| "grad_norm": 0.7857732487925049, | |
| "learning_rate": 3.7754175225413116e-05, | |
| "loss": 0.0459, | |
| "step": 1830 | |
| }, | |
| { | |
| "epoch": 0.7149795997668544, | |
| "grad_norm": 0.3672886777669717, | |
| "learning_rate": 3.771235264877331e-05, | |
| "loss": 0.0337, | |
| "step": 1840 | |
| }, | |
| { | |
| "epoch": 0.7188653584612396, | |
| "grad_norm": 0.35164352388978687, | |
| "learning_rate": 3.7670167845267934e-05, | |
| "loss": 0.0385, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.7227511171556247, | |
| "grad_norm": 0.40071179887944486, | |
| "learning_rate": 3.762762167759827e-05, | |
| "loss": 0.0467, | |
| "step": 1860 | |
| }, | |
| { | |
| "epoch": 0.7266368758500097, | |
| "grad_norm": 0.4969346423226924, | |
| "learning_rate": 3.758471501585567e-05, | |
| "loss": 0.0356, | |
| "step": 1870 | |
| }, | |
| { | |
| "epoch": 0.7305226345443948, | |
| "grad_norm": 0.3649736322224178, | |
| "learning_rate": 3.7541448737503785e-05, | |
| "loss": 0.0417, | |
| "step": 1880 | |
| }, | |
| { | |
| "epoch": 0.7344083932387798, | |
| "grad_norm": 0.3412635607398391, | |
| "learning_rate": 3.749782372736061e-05, | |
| "loss": 0.036, | |
| "step": 1890 | |
| }, | |
| { | |
| "epoch": 0.7382941519331649, | |
| "grad_norm": 0.2934505710318939, | |
| "learning_rate": 3.74538408775804e-05, | |
| "loss": 0.0442, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.7421799106275501, | |
| "grad_norm": 0.1935092511725485, | |
| "learning_rate": 3.740950108763541e-05, | |
| "loss": 0.0429, | |
| "step": 1910 | |
| }, | |
| { | |
| "epoch": 0.7460656693219351, | |
| "grad_norm": 0.2578986935595479, | |
| "learning_rate": 3.73648052642975e-05, | |
| "loss": 0.0392, | |
| "step": 1920 | |
| }, | |
| { | |
| "epoch": 0.7499514280163202, | |
| "grad_norm": 0.28980154790022605, | |
| "learning_rate": 3.7319754321619625e-05, | |
| "loss": 0.0395, | |
| "step": 1930 | |
| }, | |
| { | |
| "epoch": 0.7538371867107053, | |
| "grad_norm": 0.2637087037771915, | |
| "learning_rate": 3.7274349180917094e-05, | |
| "loss": 0.0415, | |
| "step": 1940 | |
| }, | |
| { | |
| "epoch": 0.7577229454050903, | |
| "grad_norm": 0.2854361833691479, | |
| "learning_rate": 3.722859077074875e-05, | |
| "loss": 0.0425, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.7616087040994755, | |
| "grad_norm": 0.2742967763695032, | |
| "learning_rate": 3.718248002689799e-05, | |
| "loss": 0.0352, | |
| "step": 1960 | |
| }, | |
| { | |
| "epoch": 0.7654944627938605, | |
| "grad_norm": 0.3604576092657196, | |
| "learning_rate": 3.7136017892353626e-05, | |
| "loss": 0.0413, | |
| "step": 1970 | |
| }, | |
| { | |
| "epoch": 0.7693802214882456, | |
| "grad_norm": 0.2855589921061211, | |
| "learning_rate": 3.7089205317290564e-05, | |
| "loss": 0.0458, | |
| "step": 1980 | |
| }, | |
| { | |
| "epoch": 0.7732659801826307, | |
| "grad_norm": 0.3073335348834244, | |
| "learning_rate": 3.7042043259050444e-05, | |
| "loss": 0.0333, | |
| "step": 1990 | |
| }, | |
| { | |
| "epoch": 0.7771517388770157, | |
| "grad_norm": 0.28916930774412825, | |
| "learning_rate": 3.699453268212199e-05, | |
| "loss": 0.0424, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.7810374975714008, | |
| "grad_norm": 0.3314165818828528, | |
| "learning_rate": 3.694667455812131e-05, | |
| "loss": 0.0395, | |
| "step": 2010 | |
| }, | |
| { | |
| "epoch": 0.7849232562657859, | |
| "grad_norm": 0.39228463874894487, | |
| "learning_rate": 3.6898469865772055e-05, | |
| "loss": 0.0377, | |
| "step": 2020 | |
| }, | |
| { | |
| "epoch": 0.788809014960171, | |
| "grad_norm": 0.19523419184608223, | |
| "learning_rate": 3.684991959088537e-05, | |
| "loss": 0.0362, | |
| "step": 2030 | |
| }, | |
| { | |
| "epoch": 0.7926947736545561, | |
| "grad_norm": 0.29565378498547584, | |
| "learning_rate": 3.680102472633974e-05, | |
| "loss": 0.0354, | |
| "step": 2040 | |
| }, | |
| { | |
| "epoch": 0.7965805323489411, | |
| "grad_norm": 0.3948911652054535, | |
| "learning_rate": 3.675178627206068e-05, | |
| "loss": 0.0411, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.8004662910433262, | |
| "grad_norm": 0.3555760694304167, | |
| "learning_rate": 3.6702205235000315e-05, | |
| "loss": 0.0409, | |
| "step": 2060 | |
| }, | |
| { | |
| "epoch": 0.8043520497377112, | |
| "grad_norm": 0.33270407018218245, | |
| "learning_rate": 3.665228262911676e-05, | |
| "loss": 0.0306, | |
| "step": 2070 | |
| }, | |
| { | |
| "epoch": 0.8082378084320964, | |
| "grad_norm": 0.5008566240131939, | |
| "learning_rate": 3.660201947535338e-05, | |
| "loss": 0.0415, | |
| "step": 2080 | |
| }, | |
| { | |
| "epoch": 0.8121235671264815, | |
| "grad_norm": 0.3354219331000289, | |
| "learning_rate": 3.655141680161793e-05, | |
| "loss": 0.0362, | |
| "step": 2090 | |
| }, | |
| { | |
| "epoch": 0.8160093258208665, | |
| "grad_norm": 0.39746802052357905, | |
| "learning_rate": 3.650047564276152e-05, | |
| "loss": 0.0418, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.8198950845152516, | |
| "grad_norm": 0.2251580580435554, | |
| "learning_rate": 3.644919704055748e-05, | |
| "loss": 0.0394, | |
| "step": 2110 | |
| }, | |
| { | |
| "epoch": 0.8237808432096366, | |
| "grad_norm": 0.2699540407370164, | |
| "learning_rate": 3.639758204368001e-05, | |
| "loss": 0.0384, | |
| "step": 2120 | |
| }, | |
| { | |
| "epoch": 0.8276666019040217, | |
| "grad_norm": 0.2866910109291188, | |
| "learning_rate": 3.6345631707682744e-05, | |
| "loss": 0.0357, | |
| "step": 2130 | |
| }, | |
| { | |
| "epoch": 0.8315523605984069, | |
| "grad_norm": 0.2713788435071836, | |
| "learning_rate": 3.6293347094977224e-05, | |
| "loss": 0.0409, | |
| "step": 2140 | |
| }, | |
| { | |
| "epoch": 0.8354381192927919, | |
| "grad_norm": 0.3195763251154283, | |
| "learning_rate": 3.624072927481107e-05, | |
| "loss": 0.0403, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.839323877987177, | |
| "grad_norm": 0.2983473462358474, | |
| "learning_rate": 3.618777932324621e-05, | |
| "loss": 0.0341, | |
| "step": 2160 | |
| }, | |
| { | |
| "epoch": 0.8432096366815621, | |
| "grad_norm": 0.29251828904033916, | |
| "learning_rate": 3.613449832313683e-05, | |
| "loss": 0.036, | |
| "step": 2170 | |
| }, | |
| { | |
| "epoch": 0.8470953953759471, | |
| "grad_norm": 0.23448459997179785, | |
| "learning_rate": 3.608088736410718e-05, | |
| "loss": 0.0338, | |
| "step": 2180 | |
| }, | |
| { | |
| "epoch": 0.8509811540703323, | |
| "grad_norm": 0.2757212169813267, | |
| "learning_rate": 3.6026947542529415e-05, | |
| "loss": 0.0409, | |
| "step": 2190 | |
| }, | |
| { | |
| "epoch": 0.8548669127647173, | |
| "grad_norm": 0.33239782308647875, | |
| "learning_rate": 3.597267996150106e-05, | |
| "loss": 0.0374, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.8587526714591024, | |
| "grad_norm": 0.33403655068983207, | |
| "learning_rate": 3.591808573082249e-05, | |
| "loss": 0.0357, | |
| "step": 2210 | |
| }, | |
| { | |
| "epoch": 0.8626384301534875, | |
| "grad_norm": 0.2378398763830906, | |
| "learning_rate": 3.586316596697426e-05, | |
| "loss": 0.0344, | |
| "step": 2220 | |
| }, | |
| { | |
| "epoch": 0.8665241888478725, | |
| "grad_norm": 0.3216096369297081, | |
| "learning_rate": 3.580792179309422e-05, | |
| "loss": 0.0382, | |
| "step": 2230 | |
| }, | |
| { | |
| "epoch": 0.8704099475422576, | |
| "grad_norm": 0.37090625174379876, | |
| "learning_rate": 3.5752354338954594e-05, | |
| "loss": 0.0383, | |
| "step": 2240 | |
| }, | |
| { | |
| "epoch": 0.8742957062366427, | |
| "grad_norm": 0.26521085264464783, | |
| "learning_rate": 3.569646474093885e-05, | |
| "loss": 0.0346, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.8781814649310278, | |
| "grad_norm": 0.30913299216593815, | |
| "learning_rate": 3.564025414201846e-05, | |
| "loss": 0.0373, | |
| "step": 2260 | |
| }, | |
| { | |
| "epoch": 0.8820672236254129, | |
| "grad_norm": 0.3053308654147181, | |
| "learning_rate": 3.558372369172956e-05, | |
| "loss": 0.0339, | |
| "step": 2270 | |
| }, | |
| { | |
| "epoch": 0.8859529823197979, | |
| "grad_norm": 0.2351240040993444, | |
| "learning_rate": 3.552687454614938e-05, | |
| "loss": 0.0331, | |
| "step": 2280 | |
| }, | |
| { | |
| "epoch": 0.889838741014183, | |
| "grad_norm": 0.37288388901346237, | |
| "learning_rate": 3.546970786787264e-05, | |
| "loss": 0.0361, | |
| "step": 2290 | |
| }, | |
| { | |
| "epoch": 0.893724499708568, | |
| "grad_norm": 0.44083971894002005, | |
| "learning_rate": 3.541222482598779e-05, | |
| "loss": 0.0418, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.8976102584029532, | |
| "grad_norm": 0.321928529274947, | |
| "learning_rate": 3.5354426596053066e-05, | |
| "loss": 0.0296, | |
| "step": 2310 | |
| }, | |
| { | |
| "epoch": 0.9014960170973383, | |
| "grad_norm": 0.28686426327838627, | |
| "learning_rate": 3.529631436007246e-05, | |
| "loss": 0.0324, | |
| "step": 2320 | |
| }, | |
| { | |
| "epoch": 0.9053817757917233, | |
| "grad_norm": 0.22912536137270445, | |
| "learning_rate": 3.523788930647157e-05, | |
| "loss": 0.0429, | |
| "step": 2330 | |
| }, | |
| { | |
| "epoch": 0.9092675344861084, | |
| "grad_norm": 0.3085807580953211, | |
| "learning_rate": 3.5179152630073256e-05, | |
| "loss": 0.0313, | |
| "step": 2340 | |
| }, | |
| { | |
| "epoch": 0.9131532931804935, | |
| "grad_norm": 0.34303257055116365, | |
| "learning_rate": 3.512010553207325e-05, | |
| "loss": 0.0371, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.9170390518748786, | |
| "grad_norm": 0.2772929349898867, | |
| "learning_rate": 3.506074922001554e-05, | |
| "loss": 0.0328, | |
| "step": 2360 | |
| }, | |
| { | |
| "epoch": 0.9209248105692637, | |
| "grad_norm": 0.30788741712363193, | |
| "learning_rate": 3.500108490776774e-05, | |
| "loss": 0.0402, | |
| "step": 2370 | |
| }, | |
| { | |
| "epoch": 0.9248105692636487, | |
| "grad_norm": 0.3455599730329004, | |
| "learning_rate": 3.494111381549618e-05, | |
| "loss": 0.0321, | |
| "step": 2380 | |
| }, | |
| { | |
| "epoch": 0.9286963279580338, | |
| "grad_norm": 0.4074372387990901, | |
| "learning_rate": 3.4880837169641056e-05, | |
| "loss": 0.0384, | |
| "step": 2390 | |
| }, | |
| { | |
| "epoch": 0.9325820866524189, | |
| "grad_norm": 0.2793409167713601, | |
| "learning_rate": 3.482025620289125e-05, | |
| "loss": 0.0317, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.9364678453468039, | |
| "grad_norm": 0.3747211944134066, | |
| "learning_rate": 3.4759372154159185e-05, | |
| "loss": 0.0339, | |
| "step": 2410 | |
| }, | |
| { | |
| "epoch": 0.9403536040411891, | |
| "grad_norm": 0.2805669269123183, | |
| "learning_rate": 3.469818626855546e-05, | |
| "loss": 0.0339, | |
| "step": 2420 | |
| }, | |
| { | |
| "epoch": 0.9442393627355741, | |
| "grad_norm": 0.40602944782370093, | |
| "learning_rate": 3.463669979736343e-05, | |
| "loss": 0.042, | |
| "step": 2430 | |
| }, | |
| { | |
| "epoch": 0.9481251214299592, | |
| "grad_norm": 0.225237503944127, | |
| "learning_rate": 3.457491399801353e-05, | |
| "loss": 0.034, | |
| "step": 2440 | |
| }, | |
| { | |
| "epoch": 0.9520108801243443, | |
| "grad_norm": 0.37947299716439314, | |
| "learning_rate": 3.451283013405764e-05, | |
| "loss": 0.0342, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.9558966388187293, | |
| "grad_norm": 0.4204039088153023, | |
| "learning_rate": 3.445044947514322e-05, | |
| "loss": 0.0396, | |
| "step": 2460 | |
| }, | |
| { | |
| "epoch": 0.9597823975131144, | |
| "grad_norm": 0.2632256881229335, | |
| "learning_rate": 3.438777329698733e-05, | |
| "loss": 0.0391, | |
| "step": 2470 | |
| }, | |
| { | |
| "epoch": 0.9636681562074995, | |
| "grad_norm": 0.31582365898299897, | |
| "learning_rate": 3.432480288135057e-05, | |
| "loss": 0.0425, | |
| "step": 2480 | |
| }, | |
| { | |
| "epoch": 0.9675539149018846, | |
| "grad_norm": 0.28143075132786943, | |
| "learning_rate": 3.426153951601082e-05, | |
| "loss": 0.0343, | |
| "step": 2490 | |
| }, | |
| { | |
| "epoch": 0.9714396735962697, | |
| "grad_norm": 0.24769109092709082, | |
| "learning_rate": 3.419798449473698e-05, | |
| "loss": 0.034, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.9753254322906547, | |
| "grad_norm": 0.33732830476966047, | |
| "learning_rate": 3.413413911726241e-05, | |
| "loss": 0.0405, | |
| "step": 2510 | |
| }, | |
| { | |
| "epoch": 0.9792111909850398, | |
| "grad_norm": 0.32757343882785267, | |
| "learning_rate": 3.407000468925845e-05, | |
| "loss": 0.0339, | |
| "step": 2520 | |
| }, | |
| { | |
| "epoch": 0.983096949679425, | |
| "grad_norm": 0.30532242889882966, | |
| "learning_rate": 3.4005582522307664e-05, | |
| "loss": 0.0342, | |
| "step": 2530 | |
| }, | |
| { | |
| "epoch": 0.98698270837381, | |
| "grad_norm": 0.25099007166252546, | |
| "learning_rate": 3.394087393387702e-05, | |
| "loss": 0.0336, | |
| "step": 2540 | |
| }, | |
| { | |
| "epoch": 0.9908684670681951, | |
| "grad_norm": 0.32436067338283625, | |
| "learning_rate": 3.387588024729096e-05, | |
| "loss": 0.0399, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.9947542257625801, | |
| "grad_norm": 0.4181129951433982, | |
| "learning_rate": 3.3810602791704325e-05, | |
| "loss": 0.0361, | |
| "step": 2560 | |
| }, | |
| { | |
| "epoch": 0.9986399844569652, | |
| "grad_norm": 0.3470403621225102, | |
| "learning_rate": 3.374504290207519e-05, | |
| "loss": 0.0343, | |
| "step": 2570 | |
| }, | |
| { | |
| "epoch": 1.0023314552166311, | |
| "grad_norm": 0.26511898565361997, | |
| "learning_rate": 3.367920191913755e-05, | |
| "loss": 0.0321, | |
| "step": 2580 | |
| }, | |
| { | |
| "epoch": 1.0062172139110162, | |
| "grad_norm": 0.2702160755775835, | |
| "learning_rate": 3.3613081189373914e-05, | |
| "loss": 0.0314, | |
| "step": 2590 | |
| }, | |
| { | |
| "epoch": 1.0101029726054012, | |
| "grad_norm": 0.3118992978670784, | |
| "learning_rate": 3.3546682064987735e-05, | |
| "loss": 0.033, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 1.0139887312997864, | |
| "grad_norm": 0.23058487649261575, | |
| "learning_rate": 3.34800059038758e-05, | |
| "loss": 0.0239, | |
| "step": 2610 | |
| }, | |
| { | |
| "epoch": 1.0178744899941714, | |
| "grad_norm": 0.36097802187023126, | |
| "learning_rate": 3.341305406960045e-05, | |
| "loss": 0.033, | |
| "step": 2620 | |
| }, | |
| { | |
| "epoch": 1.0217602486885564, | |
| "grad_norm": 0.28013397849201116, | |
| "learning_rate": 3.3345827931361666e-05, | |
| "loss": 0.0255, | |
| "step": 2630 | |
| }, | |
| { | |
| "epoch": 1.0256460073829414, | |
| "grad_norm": 0.2029992608524442, | |
| "learning_rate": 3.32783288639691e-05, | |
| "loss": 0.0244, | |
| "step": 2640 | |
| }, | |
| { | |
| "epoch": 1.0295317660773267, | |
| "grad_norm": 0.3553166383478555, | |
| "learning_rate": 3.321055824781394e-05, | |
| "loss": 0.0273, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 1.0334175247717117, | |
| "grad_norm": 0.24984422004673013, | |
| "learning_rate": 3.31425174688407e-05, | |
| "loss": 0.0296, | |
| "step": 2660 | |
| }, | |
| { | |
| "epoch": 1.0373032834660967, | |
| "grad_norm": 0.18909677592052765, | |
| "learning_rate": 3.307420791851887e-05, | |
| "loss": 0.0273, | |
| "step": 2670 | |
| }, | |
| { | |
| "epoch": 1.041189042160482, | |
| "grad_norm": 0.436082933063616, | |
| "learning_rate": 3.3005630993814416e-05, | |
| "loss": 0.041, | |
| "step": 2680 | |
| }, | |
| { | |
| "epoch": 1.045074800854867, | |
| "grad_norm": 0.23112116433945518, | |
| "learning_rate": 3.293678809716129e-05, | |
| "loss": 0.0304, | |
| "step": 2690 | |
| }, | |
| { | |
| "epoch": 1.048960559549252, | |
| "grad_norm": 0.2941866407157775, | |
| "learning_rate": 3.28676806364327e-05, | |
| "loss": 0.0296, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 1.0528463182436372, | |
| "grad_norm": 0.2235983248751329, | |
| "learning_rate": 3.279831002491232e-05, | |
| "loss": 0.0276, | |
| "step": 2710 | |
| }, | |
| { | |
| "epoch": 1.0567320769380222, | |
| "grad_norm": 0.3904007075572595, | |
| "learning_rate": 3.27286776812654e-05, | |
| "loss": 0.0347, | |
| "step": 2720 | |
| }, | |
| { | |
| "epoch": 1.0606178356324072, | |
| "grad_norm": 0.2861997613717581, | |
| "learning_rate": 3.2658785029509746e-05, | |
| "loss": 0.0264, | |
| "step": 2730 | |
| }, | |
| { | |
| "epoch": 1.0645035943267924, | |
| "grad_norm": 0.3566819164029042, | |
| "learning_rate": 3.258863349898659e-05, | |
| "loss": 0.0277, | |
| "step": 2740 | |
| }, | |
| { | |
| "epoch": 1.0683893530211774, | |
| "grad_norm": 0.18390854912707696, | |
| "learning_rate": 3.251822452433141e-05, | |
| "loss": 0.0307, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 1.0722751117155624, | |
| "grad_norm": 0.3062331302207858, | |
| "learning_rate": 3.244755954544449e-05, | |
| "loss": 0.0312, | |
| "step": 2760 | |
| }, | |
| { | |
| "epoch": 1.0761608704099475, | |
| "grad_norm": 0.3047851916691974, | |
| "learning_rate": 3.2376640007461595e-05, | |
| "loss": 0.0339, | |
| "step": 2770 | |
| }, | |
| { | |
| "epoch": 1.0800466291043327, | |
| "grad_norm": 0.2300653857497523, | |
| "learning_rate": 3.230546736072432e-05, | |
| "loss": 0.0297, | |
| "step": 2780 | |
| }, | |
| { | |
| "epoch": 1.0839323877987177, | |
| "grad_norm": 0.20305416815265903, | |
| "learning_rate": 3.2234043060750464e-05, | |
| "loss": 0.0324, | |
| "step": 2790 | |
| }, | |
| { | |
| "epoch": 1.0878181464931027, | |
| "grad_norm": 0.24117134662351375, | |
| "learning_rate": 3.216236856820429e-05, | |
| "loss": 0.0225, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 1.091703905187488, | |
| "grad_norm": 0.23857567814841038, | |
| "learning_rate": 3.2090445348866616e-05, | |
| "loss": 0.037, | |
| "step": 2810 | |
| }, | |
| { | |
| "epoch": 1.095589663881873, | |
| "grad_norm": 0.20744357816037623, | |
| "learning_rate": 3.201827487360485e-05, | |
| "loss": 0.0343, | |
| "step": 2820 | |
| }, | |
| { | |
| "epoch": 1.099475422576258, | |
| "grad_norm": 0.3148313503304088, | |
| "learning_rate": 3.194585861834292e-05, | |
| "loss": 0.0329, | |
| "step": 2830 | |
| }, | |
| { | |
| "epoch": 1.103361181270643, | |
| "grad_norm": 0.2597458756041528, | |
| "learning_rate": 3.187319806403108e-05, | |
| "loss": 0.0324, | |
| "step": 2840 | |
| }, | |
| { | |
| "epoch": 1.1072469399650282, | |
| "grad_norm": 0.37462454725250943, | |
| "learning_rate": 3.180029469661563e-05, | |
| "loss": 0.0298, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 1.1111326986594132, | |
| "grad_norm": 0.26990355348646067, | |
| "learning_rate": 3.172715000700851e-05, | |
| "loss": 0.0256, | |
| "step": 2860 | |
| }, | |
| { | |
| "epoch": 1.1150184573537982, | |
| "grad_norm": 0.2193580972033405, | |
| "learning_rate": 3.165376549105686e-05, | |
| "loss": 0.035, | |
| "step": 2870 | |
| }, | |
| { | |
| "epoch": 1.1189042160481835, | |
| "grad_norm": 0.2966997098167283, | |
| "learning_rate": 3.158014264951234e-05, | |
| "loss": 0.0341, | |
| "step": 2880 | |
| }, | |
| { | |
| "epoch": 1.1227899747425685, | |
| "grad_norm": 0.3014871881210635, | |
| "learning_rate": 3.150628298800055e-05, | |
| "loss": 0.0328, | |
| "step": 2890 | |
| }, | |
| { | |
| "epoch": 1.1266757334369535, | |
| "grad_norm": 0.33425883934391964, | |
| "learning_rate": 3.1432188016990154e-05, | |
| "loss": 0.0262, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 1.1305614921313387, | |
| "grad_norm": 0.23135238874039699, | |
| "learning_rate": 3.1357859251762005e-05, | |
| "loss": 0.0349, | |
| "step": 2910 | |
| }, | |
| { | |
| "epoch": 1.1344472508257237, | |
| "grad_norm": 0.2957844330256544, | |
| "learning_rate": 3.1283298212378204e-05, | |
| "loss": 0.0308, | |
| "step": 2920 | |
| }, | |
| { | |
| "epoch": 1.1383330095201087, | |
| "grad_norm": 0.309553215175613, | |
| "learning_rate": 3.120850642365094e-05, | |
| "loss": 0.0378, | |
| "step": 2930 | |
| }, | |
| { | |
| "epoch": 1.142218768214494, | |
| "grad_norm": 0.27357475926464664, | |
| "learning_rate": 3.113348541511139e-05, | |
| "loss": 0.0315, | |
| "step": 2940 | |
| }, | |
| { | |
| "epoch": 1.146104526908879, | |
| "grad_norm": 0.2922282115073422, | |
| "learning_rate": 3.1058236720978357e-05, | |
| "loss": 0.0207, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 1.149990285603264, | |
| "grad_norm": 0.30551015741363674, | |
| "learning_rate": 3.0982761880126956e-05, | |
| "loss": 0.0309, | |
| "step": 2960 | |
| }, | |
| { | |
| "epoch": 1.153876044297649, | |
| "grad_norm": 0.19603484059399878, | |
| "learning_rate": 3.090706243605712e-05, | |
| "loss": 0.0275, | |
| "step": 2970 | |
| }, | |
| { | |
| "epoch": 1.1577618029920342, | |
| "grad_norm": 0.12126922576885169, | |
| "learning_rate": 3.083113993686202e-05, | |
| "loss": 0.0214, | |
| "step": 2980 | |
| }, | |
| { | |
| "epoch": 1.1616475616864192, | |
| "grad_norm": 0.24570529004069103, | |
| "learning_rate": 3.075499593519643e-05, | |
| "loss": 0.0313, | |
| "step": 2990 | |
| }, | |
| { | |
| "epoch": 1.1655333203808043, | |
| "grad_norm": 0.21996601041925384, | |
| "learning_rate": 3.067863198824499e-05, | |
| "loss": 0.0322, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 1.1694190790751895, | |
| "grad_norm": 0.26523791368864613, | |
| "learning_rate": 3.0602049657690275e-05, | |
| "loss": 0.033, | |
| "step": 3010 | |
| }, | |
| { | |
| "epoch": 1.1733048377695745, | |
| "grad_norm": 0.3743941590932805, | |
| "learning_rate": 3.0525250509680975e-05, | |
| "loss": 0.0315, | |
| "step": 3020 | |
| }, | |
| { | |
| "epoch": 1.1771905964639595, | |
| "grad_norm": 0.29800477638021083, | |
| "learning_rate": 3.0448236114799798e-05, | |
| "loss": 0.0314, | |
| "step": 3030 | |
| }, | |
| { | |
| "epoch": 1.1810763551583447, | |
| "grad_norm": 0.41913805025695794, | |
| "learning_rate": 3.0371008048031335e-05, | |
| "loss": 0.0252, | |
| "step": 3040 | |
| }, | |
| { | |
| "epoch": 1.1849621138527298, | |
| "grad_norm": 0.2972323169585253, | |
| "learning_rate": 3.029356788872992e-05, | |
| "loss": 0.0284, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 1.1888478725471148, | |
| "grad_norm": 0.2548640539234551, | |
| "learning_rate": 3.0215917220587264e-05, | |
| "loss": 0.0259, | |
| "step": 3060 | |
| }, | |
| { | |
| "epoch": 1.1927336312415, | |
| "grad_norm": 0.18370703051409343, | |
| "learning_rate": 3.013805763160009e-05, | |
| "loss": 0.023, | |
| "step": 3070 | |
| }, | |
| { | |
| "epoch": 1.196619389935885, | |
| "grad_norm": 0.17576524808213698, | |
| "learning_rate": 3.0059990714037678e-05, | |
| "loss": 0.0268, | |
| "step": 3080 | |
| }, | |
| { | |
| "epoch": 1.20050514863027, | |
| "grad_norm": 0.25633777261633506, | |
| "learning_rate": 2.9981718064409284e-05, | |
| "loss": 0.0307, | |
| "step": 3090 | |
| }, | |
| { | |
| "epoch": 1.204390907324655, | |
| "grad_norm": 0.27763678833721916, | |
| "learning_rate": 2.9903241283431472e-05, | |
| "loss": 0.0279, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 1.2082766660190403, | |
| "grad_norm": 0.3087663315027572, | |
| "learning_rate": 2.9824561975995427e-05, | |
| "loss": 0.0276, | |
| "step": 3110 | |
| }, | |
| { | |
| "epoch": 1.2121624247134253, | |
| "grad_norm": 0.28744858586674227, | |
| "learning_rate": 2.974568175113409e-05, | |
| "loss": 0.024, | |
| "step": 3120 | |
| }, | |
| { | |
| "epoch": 1.2160481834078103, | |
| "grad_norm": 0.21935445525902675, | |
| "learning_rate": 2.9666602221989267e-05, | |
| "loss": 0.0286, | |
| "step": 3130 | |
| }, | |
| { | |
| "epoch": 1.2199339421021955, | |
| "grad_norm": 0.23246090578146508, | |
| "learning_rate": 2.958732500577864e-05, | |
| "loss": 0.0212, | |
| "step": 3140 | |
| }, | |
| { | |
| "epoch": 1.2238197007965805, | |
| "grad_norm": 0.2698352542678301, | |
| "learning_rate": 2.9507851723762716e-05, | |
| "loss": 0.0308, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 1.2277054594909655, | |
| "grad_norm": 0.4180270105674614, | |
| "learning_rate": 2.9428184001211616e-05, | |
| "loss": 0.0212, | |
| "step": 3160 | |
| }, | |
| { | |
| "epoch": 1.2315912181853508, | |
| "grad_norm": 0.2841163727116711, | |
| "learning_rate": 2.9348323467371897e-05, | |
| "loss": 0.0237, | |
| "step": 3170 | |
| }, | |
| { | |
| "epoch": 1.2354769768797358, | |
| "grad_norm": 0.3283735446941676, | |
| "learning_rate": 2.9268271755433198e-05, | |
| "loss": 0.0268, | |
| "step": 3180 | |
| }, | |
| { | |
| "epoch": 1.2393627355741208, | |
| "grad_norm": 0.17964907995508142, | |
| "learning_rate": 2.9188030502494853e-05, | |
| "loss": 0.026, | |
| "step": 3190 | |
| }, | |
| { | |
| "epoch": 1.243248494268506, | |
| "grad_norm": 0.3739453975454048, | |
| "learning_rate": 2.9107601349532406e-05, | |
| "loss": 0.0351, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 1.247134252962891, | |
| "grad_norm": 0.22041723340395544, | |
| "learning_rate": 2.9026985941364053e-05, | |
| "loss": 0.0289, | |
| "step": 3210 | |
| }, | |
| { | |
| "epoch": 1.251020011657276, | |
| "grad_norm": 0.16977718972527425, | |
| "learning_rate": 2.8946185926617012e-05, | |
| "loss": 0.0227, | |
| "step": 3220 | |
| }, | |
| { | |
| "epoch": 1.254905770351661, | |
| "grad_norm": 0.3419426868688641, | |
| "learning_rate": 2.88652029576938e-05, | |
| "loss": 0.0265, | |
| "step": 3230 | |
| }, | |
| { | |
| "epoch": 1.2587915290460463, | |
| "grad_norm": 0.22438684261772954, | |
| "learning_rate": 2.878403869073843e-05, | |
| "loss": 0.0262, | |
| "step": 3240 | |
| }, | |
| { | |
| "epoch": 1.2626772877404313, | |
| "grad_norm": 0.30489902115039463, | |
| "learning_rate": 2.8702694785602587e-05, | |
| "loss": 0.0314, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 1.2665630464348163, | |
| "grad_norm": 0.2434714851331239, | |
| "learning_rate": 2.8621172905811613e-05, | |
| "loss": 0.0275, | |
| "step": 3260 | |
| }, | |
| { | |
| "epoch": 1.2704488051292016, | |
| "grad_norm": 0.27588341692139456, | |
| "learning_rate": 2.8539474718530543e-05, | |
| "loss": 0.0264, | |
| "step": 3270 | |
| }, | |
| { | |
| "epoch": 1.2743345638235866, | |
| "grad_norm": 0.29322871157127156, | |
| "learning_rate": 2.8457601894529997e-05, | |
| "loss": 0.0375, | |
| "step": 3280 | |
| }, | |
| { | |
| "epoch": 1.2782203225179716, | |
| "grad_norm": 0.19868867147652092, | |
| "learning_rate": 2.8375556108151995e-05, | |
| "loss": 0.029, | |
| "step": 3290 | |
| }, | |
| { | |
| "epoch": 1.2821060812123566, | |
| "grad_norm": 0.3711457646768087, | |
| "learning_rate": 2.829333903727574e-05, | |
| "loss": 0.0327, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 1.2859918399067418, | |
| "grad_norm": 0.28271933767183866, | |
| "learning_rate": 2.821095236328328e-05, | |
| "loss": 0.0261, | |
| "step": 3310 | |
| }, | |
| { | |
| "epoch": 1.2898775986011268, | |
| "grad_norm": 0.23298730366288653, | |
| "learning_rate": 2.812839777102514e-05, | |
| "loss": 0.029, | |
| "step": 3320 | |
| }, | |
| { | |
| "epoch": 1.293763357295512, | |
| "grad_norm": 0.23059435959988767, | |
| "learning_rate": 2.8045676948785873e-05, | |
| "loss": 0.0321, | |
| "step": 3330 | |
| }, | |
| { | |
| "epoch": 1.297649115989897, | |
| "grad_norm": 0.19452335917095895, | |
| "learning_rate": 2.7962791588249492e-05, | |
| "loss": 0.029, | |
| "step": 3340 | |
| }, | |
| { | |
| "epoch": 1.301534874684282, | |
| "grad_norm": 0.30937914926531507, | |
| "learning_rate": 2.7879743384464942e-05, | |
| "loss": 0.027, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 1.305420633378667, | |
| "grad_norm": 0.30675762635230686, | |
| "learning_rate": 2.7796534035811378e-05, | |
| "loss": 0.0248, | |
| "step": 3360 | |
| }, | |
| { | |
| "epoch": 1.3093063920730523, | |
| "grad_norm": 0.14564788288302782, | |
| "learning_rate": 2.7713165243963444e-05, | |
| "loss": 0.0242, | |
| "step": 3370 | |
| }, | |
| { | |
| "epoch": 1.3131921507674373, | |
| "grad_norm": 0.3611570999157121, | |
| "learning_rate": 2.7629638713856503e-05, | |
| "loss": 0.0313, | |
| "step": 3380 | |
| }, | |
| { | |
| "epoch": 1.3170779094618223, | |
| "grad_norm": 0.15900271035042685, | |
| "learning_rate": 2.7545956153651712e-05, | |
| "loss": 0.0246, | |
| "step": 3390 | |
| }, | |
| { | |
| "epoch": 1.3209636681562076, | |
| "grad_norm": 0.3433667317929698, | |
| "learning_rate": 2.746211927470117e-05, | |
| "loss": 0.0269, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 1.3248494268505926, | |
| "grad_norm": 0.28124492243082266, | |
| "learning_rate": 2.737812979151284e-05, | |
| "loss": 0.0245, | |
| "step": 3410 | |
| }, | |
| { | |
| "epoch": 1.3287351855449776, | |
| "grad_norm": 0.187688472206659, | |
| "learning_rate": 2.7293989421715542e-05, | |
| "loss": 0.0253, | |
| "step": 3420 | |
| }, | |
| { | |
| "epoch": 1.3326209442393626, | |
| "grad_norm": 0.2829084594578858, | |
| "learning_rate": 2.720969988602379e-05, | |
| "loss": 0.0207, | |
| "step": 3430 | |
| }, | |
| { | |
| "epoch": 1.3365067029337478, | |
| "grad_norm": 0.3016772133080438, | |
| "learning_rate": 2.7125262908202633e-05, | |
| "loss": 0.0281, | |
| "step": 3440 | |
| }, | |
| { | |
| "epoch": 1.3403924616281329, | |
| "grad_norm": 0.2338038212652911, | |
| "learning_rate": 2.7040680215032377e-05, | |
| "loss": 0.0322, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 1.344278220322518, | |
| "grad_norm": 0.23012741652116103, | |
| "learning_rate": 2.6955953536273285e-05, | |
| "loss": 0.0342, | |
| "step": 3460 | |
| }, | |
| { | |
| "epoch": 1.348163979016903, | |
| "grad_norm": 0.1789676480348028, | |
| "learning_rate": 2.6871084604630214e-05, | |
| "loss": 0.0302, | |
| "step": 3470 | |
| }, | |
| { | |
| "epoch": 1.3520497377112881, | |
| "grad_norm": 0.2847505802479265, | |
| "learning_rate": 2.6786075155717147e-05, | |
| "loss": 0.0341, | |
| "step": 3480 | |
| }, | |
| { | |
| "epoch": 1.3559354964056731, | |
| "grad_norm": 0.21765574449058714, | |
| "learning_rate": 2.6700926928021736e-05, | |
| "loss": 0.0287, | |
| "step": 3490 | |
| }, | |
| { | |
| "epoch": 1.3598212551000584, | |
| "grad_norm": 0.28617645696636324, | |
| "learning_rate": 2.6615641662869714e-05, | |
| "loss": 0.0307, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 1.3637070137944434, | |
| "grad_norm": 0.34355897351030124, | |
| "learning_rate": 2.6530221104389316e-05, | |
| "loss": 0.0232, | |
| "step": 3510 | |
| }, | |
| { | |
| "epoch": 1.3675927724888284, | |
| "grad_norm": 0.19082168638857241, | |
| "learning_rate": 2.6444666999475593e-05, | |
| "loss": 0.0301, | |
| "step": 3520 | |
| }, | |
| { | |
| "epoch": 1.3714785311832136, | |
| "grad_norm": 0.27997736990150557, | |
| "learning_rate": 2.635898109775468e-05, | |
| "loss": 0.0237, | |
| "step": 3530 | |
| }, | |
| { | |
| "epoch": 1.3753642898775986, | |
| "grad_norm": 0.2234204944915347, | |
| "learning_rate": 2.6273165151548047e-05, | |
| "loss": 0.0271, | |
| "step": 3540 | |
| }, | |
| { | |
| "epoch": 1.3792500485719836, | |
| "grad_norm": 0.2201998240178073, | |
| "learning_rate": 2.6187220915836627e-05, | |
| "loss": 0.0292, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 1.3831358072663686, | |
| "grad_norm": 0.14320713366865984, | |
| "learning_rate": 2.6101150148224928e-05, | |
| "loss": 0.0288, | |
| "step": 3560 | |
| }, | |
| { | |
| "epoch": 1.3870215659607539, | |
| "grad_norm": 0.2116339664682065, | |
| "learning_rate": 2.601495460890513e-05, | |
| "loss": 0.0269, | |
| "step": 3570 | |
| }, | |
| { | |
| "epoch": 1.3909073246551389, | |
| "grad_norm": 0.24959055444114717, | |
| "learning_rate": 2.5928636060621036e-05, | |
| "loss": 0.0337, | |
| "step": 3580 | |
| }, | |
| { | |
| "epoch": 1.3947930833495241, | |
| "grad_norm": 0.2954702102324182, | |
| "learning_rate": 2.5842196268632068e-05, | |
| "loss": 0.0228, | |
| "step": 3590 | |
| }, | |
| { | |
| "epoch": 1.3986788420439091, | |
| "grad_norm": 0.38085393262983136, | |
| "learning_rate": 2.5755637000677124e-05, | |
| "loss": 0.0247, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 1.4025646007382941, | |
| "grad_norm": 0.21224922937032303, | |
| "learning_rate": 2.566896002693845e-05, | |
| "loss": 0.0252, | |
| "step": 3610 | |
| }, | |
| { | |
| "epoch": 1.4064503594326792, | |
| "grad_norm": 0.3607238719844902, | |
| "learning_rate": 2.5582167120005467e-05, | |
| "loss": 0.0258, | |
| "step": 3620 | |
| }, | |
| { | |
| "epoch": 1.4103361181270644, | |
| "grad_norm": 0.19260362661029898, | |
| "learning_rate": 2.549526005483844e-05, | |
| "loss": 0.0235, | |
| "step": 3630 | |
| }, | |
| { | |
| "epoch": 1.4142218768214494, | |
| "grad_norm": 0.26994983197422656, | |
| "learning_rate": 2.5408240608732277e-05, | |
| "loss": 0.0253, | |
| "step": 3640 | |
| }, | |
| { | |
| "epoch": 1.4181076355158344, | |
| "grad_norm": 0.29929031715132187, | |
| "learning_rate": 2.5321110561280106e-05, | |
| "loss": 0.0275, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 1.4219933942102196, | |
| "grad_norm": 0.3196437552669507, | |
| "learning_rate": 2.523387169433692e-05, | |
| "loss": 0.0311, | |
| "step": 3660 | |
| }, | |
| { | |
| "epoch": 1.4258791529046047, | |
| "grad_norm": 0.2296744794340079, | |
| "learning_rate": 2.514652579198312e-05, | |
| "loss": 0.0278, | |
| "step": 3670 | |
| }, | |
| { | |
| "epoch": 1.4297649115989897, | |
| "grad_norm": 0.16892616731103247, | |
| "learning_rate": 2.5059074640488047e-05, | |
| "loss": 0.0263, | |
| "step": 3680 | |
| }, | |
| { | |
| "epoch": 1.4336506702933747, | |
| "grad_norm": 0.2394458582673369, | |
| "learning_rate": 2.497152002827345e-05, | |
| "loss": 0.0277, | |
| "step": 3690 | |
| }, | |
| { | |
| "epoch": 1.43753642898776, | |
| "grad_norm": 0.1974720107762794, | |
| "learning_rate": 2.488386374587688e-05, | |
| "loss": 0.0288, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 1.441422187682145, | |
| "grad_norm": 0.21920589738520604, | |
| "learning_rate": 2.479610758591511e-05, | |
| "loss": 0.0216, | |
| "step": 3710 | |
| }, | |
| { | |
| "epoch": 1.4453079463765302, | |
| "grad_norm": 0.21707363978296285, | |
| "learning_rate": 2.4708253343047456e-05, | |
| "loss": 0.0292, | |
| "step": 3720 | |
| }, | |
| { | |
| "epoch": 1.4491937050709152, | |
| "grad_norm": 0.18601726681673092, | |
| "learning_rate": 2.4620302813939093e-05, | |
| "loss": 0.0245, | |
| "step": 3730 | |
| }, | |
| { | |
| "epoch": 1.4530794637653002, | |
| "grad_norm": 0.2635531513187159, | |
| "learning_rate": 2.4532257797224287e-05, | |
| "loss": 0.0295, | |
| "step": 3740 | |
| }, | |
| { | |
| "epoch": 1.4569652224596852, | |
| "grad_norm": 0.3251174321742817, | |
| "learning_rate": 2.4444120093469632e-05, | |
| "loss": 0.0298, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 1.4608509811540702, | |
| "grad_norm": 0.4977849600583359, | |
| "learning_rate": 2.4355891505137216e-05, | |
| "loss": 0.0325, | |
| "step": 3760 | |
| }, | |
| { | |
| "epoch": 1.4647367398484554, | |
| "grad_norm": 0.1750593962332665, | |
| "learning_rate": 2.4267573836547768e-05, | |
| "loss": 0.022, | |
| "step": 3770 | |
| }, | |
| { | |
| "epoch": 1.4686224985428404, | |
| "grad_norm": 0.271292018124954, | |
| "learning_rate": 2.417916889384374e-05, | |
| "loss": 0.0281, | |
| "step": 3780 | |
| }, | |
| { | |
| "epoch": 1.4725082572372257, | |
| "grad_norm": 0.2221308512528936, | |
| "learning_rate": 2.4090678484952416e-05, | |
| "loss": 0.0209, | |
| "step": 3790 | |
| }, | |
| { | |
| "epoch": 1.4763940159316107, | |
| "grad_norm": 0.3356499834645714, | |
| "learning_rate": 2.400210441954888e-05, | |
| "loss": 0.0286, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 1.4802797746259957, | |
| "grad_norm": 0.19837629045900065, | |
| "learning_rate": 2.3913448509019047e-05, | |
| "loss": 0.0268, | |
| "step": 3810 | |
| }, | |
| { | |
| "epoch": 1.4841655333203807, | |
| "grad_norm": 0.22984338791117062, | |
| "learning_rate": 2.3824712566422613e-05, | |
| "loss": 0.0324, | |
| "step": 3820 | |
| }, | |
| { | |
| "epoch": 1.488051292014766, | |
| "grad_norm": 0.149938685223149, | |
| "learning_rate": 2.3735898406455945e-05, | |
| "loss": 0.0204, | |
| "step": 3830 | |
| }, | |
| { | |
| "epoch": 1.491937050709151, | |
| "grad_norm": 0.2297420680782252, | |
| "learning_rate": 2.364700784541504e-05, | |
| "loss": 0.0281, | |
| "step": 3840 | |
| }, | |
| { | |
| "epoch": 1.495822809403536, | |
| "grad_norm": 0.21161937179194085, | |
| "learning_rate": 2.3558042701158294e-05, | |
| "loss": 0.0296, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 1.4997085680979212, | |
| "grad_norm": 0.2681086390248988, | |
| "learning_rate": 2.34690047930694e-05, | |
| "loss": 0.024, | |
| "step": 3860 | |
| }, | |
| { | |
| "epoch": 1.5035943267923062, | |
| "grad_norm": 0.26044197357045645, | |
| "learning_rate": 2.337989594202009e-05, | |
| "loss": 0.0242, | |
| "step": 3870 | |
| }, | |
| { | |
| "epoch": 1.5074800854866912, | |
| "grad_norm": 0.2581684322985289, | |
| "learning_rate": 2.3290717970332918e-05, | |
| "loss": 0.0236, | |
| "step": 3880 | |
| }, | |
| { | |
| "epoch": 1.5113658441810762, | |
| "grad_norm": 0.23713590285989272, | |
| "learning_rate": 2.3201472701744013e-05, | |
| "loss": 0.0251, | |
| "step": 3890 | |
| }, | |
| { | |
| "epoch": 1.5152516028754615, | |
| "grad_norm": 0.2501508071933915, | |
| "learning_rate": 2.3112161961365724e-05, | |
| "loss": 0.0301, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 1.5191373615698465, | |
| "grad_norm": 0.24802013587982075, | |
| "learning_rate": 2.302278757564937e-05, | |
| "loss": 0.0261, | |
| "step": 3910 | |
| }, | |
| { | |
| "epoch": 1.5230231202642317, | |
| "grad_norm": 0.3445587520016882, | |
| "learning_rate": 2.2933351372347822e-05, | |
| "loss": 0.0279, | |
| "step": 3920 | |
| }, | |
| { | |
| "epoch": 1.5269088789586167, | |
| "grad_norm": 0.22828900210653533, | |
| "learning_rate": 2.2843855180478167e-05, | |
| "loss": 0.0224, | |
| "step": 3930 | |
| }, | |
| { | |
| "epoch": 1.5307946376530017, | |
| "grad_norm": 0.14993393496314023, | |
| "learning_rate": 2.2754300830284287e-05, | |
| "loss": 0.0228, | |
| "step": 3940 | |
| }, | |
| { | |
| "epoch": 1.5346803963473867, | |
| "grad_norm": 0.22112935196732578, | |
| "learning_rate": 2.266469015319943e-05, | |
| "loss": 0.0298, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 1.5385661550417717, | |
| "grad_norm": 0.19193655052829417, | |
| "learning_rate": 2.2575024981808763e-05, | |
| "loss": 0.0253, | |
| "step": 3960 | |
| }, | |
| { | |
| "epoch": 1.542451913736157, | |
| "grad_norm": 0.2670658908912582, | |
| "learning_rate": 2.2485307149811894e-05, | |
| "loss": 0.029, | |
| "step": 3970 | |
| }, | |
| { | |
| "epoch": 1.5463376724305422, | |
| "grad_norm": 0.2243861088423283, | |
| "learning_rate": 2.2395538491985338e-05, | |
| "loss": 0.0246, | |
| "step": 3980 | |
| }, | |
| { | |
| "epoch": 1.5502234311249272, | |
| "grad_norm": 0.22493381727938802, | |
| "learning_rate": 2.230572084414507e-05, | |
| "loss": 0.0253, | |
| "step": 3990 | |
| }, | |
| { | |
| "epoch": 1.5541091898193122, | |
| "grad_norm": 0.2866864450702556, | |
| "learning_rate": 2.2215856043108896e-05, | |
| "loss": 0.0291, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 1.5579949485136972, | |
| "grad_norm": 0.16815062034208167, | |
| "learning_rate": 2.212594592665896e-05, | |
| "loss": 0.025, | |
| "step": 4010 | |
| }, | |
| { | |
| "epoch": 1.5618807072080823, | |
| "grad_norm": 0.2612232980649845, | |
| "learning_rate": 2.2035992333504127e-05, | |
| "loss": 0.0326, | |
| "step": 4020 | |
| }, | |
| { | |
| "epoch": 1.5657664659024675, | |
| "grad_norm": 0.33854813622270197, | |
| "learning_rate": 2.1945997103242344e-05, | |
| "loss": 0.0292, | |
| "step": 4030 | |
| }, | |
| { | |
| "epoch": 1.5696522245968525, | |
| "grad_norm": 0.2617347504462213, | |
| "learning_rate": 2.1855962076323115e-05, | |
| "loss": 0.021, | |
| "step": 4040 | |
| }, | |
| { | |
| "epoch": 1.5735379832912377, | |
| "grad_norm": 0.1548116373638503, | |
| "learning_rate": 2.1765889094009762e-05, | |
| "loss": 0.0269, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 1.5774237419856227, | |
| "grad_norm": 0.19918562909410628, | |
| "learning_rate": 2.167577999834185e-05, | |
| "loss": 0.0237, | |
| "step": 4060 | |
| }, | |
| { | |
| "epoch": 1.5813095006800078, | |
| "grad_norm": 0.11447983969077741, | |
| "learning_rate": 2.1585636632097446e-05, | |
| "loss": 0.0235, | |
| "step": 4070 | |
| }, | |
| { | |
| "epoch": 1.5851952593743928, | |
| "grad_norm": 0.1598181388976202, | |
| "learning_rate": 2.1495460838755492e-05, | |
| "loss": 0.0249, | |
| "step": 4080 | |
| }, | |
| { | |
| "epoch": 1.5890810180687778, | |
| "grad_norm": 0.1998709332001755, | |
| "learning_rate": 2.140525446245808e-05, | |
| "loss": 0.0223, | |
| "step": 4090 | |
| }, | |
| { | |
| "epoch": 1.592966776763163, | |
| "grad_norm": 0.23042732127911256, | |
| "learning_rate": 2.1315019347972723e-05, | |
| "loss": 0.025, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 1.5968525354575482, | |
| "grad_norm": 0.28671695920142976, | |
| "learning_rate": 2.1224757340654672e-05, | |
| "loss": 0.0282, | |
| "step": 4110 | |
| }, | |
| { | |
| "epoch": 1.6007382941519332, | |
| "grad_norm": 0.2013168208153807, | |
| "learning_rate": 2.1134470286409118e-05, | |
| "loss": 0.0223, | |
| "step": 4120 | |
| }, | |
| { | |
| "epoch": 1.6046240528463183, | |
| "grad_norm": 0.22132045622532287, | |
| "learning_rate": 2.10441600316535e-05, | |
| "loss": 0.0209, | |
| "step": 4130 | |
| }, | |
| { | |
| "epoch": 1.6085098115407033, | |
| "grad_norm": 0.18246569785225308, | |
| "learning_rate": 2.095382842327971e-05, | |
| "loss": 0.0258, | |
| "step": 4140 | |
| }, | |
| { | |
| "epoch": 1.6123955702350883, | |
| "grad_norm": 0.15024515227995486, | |
| "learning_rate": 2.086347730861633e-05, | |
| "loss": 0.018, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 1.6162813289294735, | |
| "grad_norm": 0.21028730383896185, | |
| "learning_rate": 2.077310853539086e-05, | |
| "loss": 0.0246, | |
| "step": 4160 | |
| }, | |
| { | |
| "epoch": 1.6201670876238585, | |
| "grad_norm": 0.18683373176287732, | |
| "learning_rate": 2.068272395169193e-05, | |
| "loss": 0.0207, | |
| "step": 4170 | |
| }, | |
| { | |
| "epoch": 1.6240528463182438, | |
| "grad_norm": 0.2838911905964799, | |
| "learning_rate": 2.0592325405931498e-05, | |
| "loss": 0.0282, | |
| "step": 4180 | |
| }, | |
| { | |
| "epoch": 1.6279386050126288, | |
| "grad_norm": 0.28170440553062887, | |
| "learning_rate": 2.050191474680705e-05, | |
| "loss": 0.0226, | |
| "step": 4190 | |
| }, | |
| { | |
| "epoch": 1.6318243637070138, | |
| "grad_norm": 0.20578915632228978, | |
| "learning_rate": 2.04114938232638e-05, | |
| "loss": 0.0221, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 1.6357101224013988, | |
| "grad_norm": 0.3124892105648976, | |
| "learning_rate": 2.0321064484456875e-05, | |
| "loss": 0.0199, | |
| "step": 4210 | |
| }, | |
| { | |
| "epoch": 1.6395958810957838, | |
| "grad_norm": 0.1570324139378916, | |
| "learning_rate": 2.0230628579713505e-05, | |
| "loss": 0.0255, | |
| "step": 4220 | |
| }, | |
| { | |
| "epoch": 1.643481639790169, | |
| "grad_norm": 0.24972195531663, | |
| "learning_rate": 2.0140187958495187e-05, | |
| "loss": 0.0252, | |
| "step": 4230 | |
| }, | |
| { | |
| "epoch": 1.6473673984845543, | |
| "grad_norm": 0.2018721430297068, | |
| "learning_rate": 2.004974447035988e-05, | |
| "loss": 0.0225, | |
| "step": 4240 | |
| }, | |
| { | |
| "epoch": 1.6512531571789393, | |
| "grad_norm": 0.19112522378491964, | |
| "learning_rate": 1.9959299964924156e-05, | |
| "loss": 0.024, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 1.6551389158733243, | |
| "grad_norm": 0.14928748310133091, | |
| "learning_rate": 1.9868856291825417e-05, | |
| "loss": 0.0199, | |
| "step": 4260 | |
| }, | |
| { | |
| "epoch": 1.6590246745677093, | |
| "grad_norm": 0.22380334602146915, | |
| "learning_rate": 1.9778415300684033e-05, | |
| "loss": 0.0274, | |
| "step": 4270 | |
| }, | |
| { | |
| "epoch": 1.6629104332620943, | |
| "grad_norm": 0.24112965797496708, | |
| "learning_rate": 1.9687978841065514e-05, | |
| "loss": 0.0217, | |
| "step": 4280 | |
| }, | |
| { | |
| "epoch": 1.6667961919564795, | |
| "grad_norm": 0.16406141595970072, | |
| "learning_rate": 1.9597548762442712e-05, | |
| "loss": 0.0283, | |
| "step": 4290 | |
| }, | |
| { | |
| "epoch": 1.6706819506508646, | |
| "grad_norm": 0.1948101464396577, | |
| "learning_rate": 1.9507126914157973e-05, | |
| "loss": 0.0251, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 1.6745677093452498, | |
| "grad_norm": 0.2956340770250799, | |
| "learning_rate": 1.941671514538536e-05, | |
| "loss": 0.0185, | |
| "step": 4310 | |
| }, | |
| { | |
| "epoch": 1.6784534680396348, | |
| "grad_norm": 0.12177397943751574, | |
| "learning_rate": 1.9326315305092746e-05, | |
| "loss": 0.0263, | |
| "step": 4320 | |
| }, | |
| { | |
| "epoch": 1.6823392267340198, | |
| "grad_norm": 0.19695542020311577, | |
| "learning_rate": 1.923592924200412e-05, | |
| "loss": 0.021, | |
| "step": 4330 | |
| }, | |
| { | |
| "epoch": 1.6862249854284048, | |
| "grad_norm": 0.21181764645601958, | |
| "learning_rate": 1.9145558804561686e-05, | |
| "loss": 0.0254, | |
| "step": 4340 | |
| }, | |
| { | |
| "epoch": 1.6901107441227898, | |
| "grad_norm": 0.20176994374162602, | |
| "learning_rate": 1.90552058408881e-05, | |
| "loss": 0.0247, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 1.693996502817175, | |
| "grad_norm": 0.23272965832055748, | |
| "learning_rate": 1.8964872198748694e-05, | |
| "loss": 0.0207, | |
| "step": 4360 | |
| }, | |
| { | |
| "epoch": 1.6978822615115603, | |
| "grad_norm": 0.17179212412190722, | |
| "learning_rate": 1.8874559725513618e-05, | |
| "loss": 0.0236, | |
| "step": 4370 | |
| }, | |
| { | |
| "epoch": 1.7017680202059453, | |
| "grad_norm": 0.2088323094557007, | |
| "learning_rate": 1.8784270268120148e-05, | |
| "loss": 0.0215, | |
| "step": 4380 | |
| }, | |
| { | |
| "epoch": 1.7056537789003303, | |
| "grad_norm": 0.25618504375528284, | |
| "learning_rate": 1.869400567303486e-05, | |
| "loss": 0.0253, | |
| "step": 4390 | |
| }, | |
| { | |
| "epoch": 1.7095395375947153, | |
| "grad_norm": 0.29335176738438906, | |
| "learning_rate": 1.8603767786215886e-05, | |
| "loss": 0.0247, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 1.7134252962891003, | |
| "grad_norm": 0.19297739454353136, | |
| "learning_rate": 1.8513558453075145e-05, | |
| "loss": 0.0176, | |
| "step": 4410 | |
| }, | |
| { | |
| "epoch": 1.7173110549834856, | |
| "grad_norm": 0.21907186035510737, | |
| "learning_rate": 1.8423379518440637e-05, | |
| "loss": 0.0252, | |
| "step": 4420 | |
| }, | |
| { | |
| "epoch": 1.7211968136778706, | |
| "grad_norm": 0.1662707063311211, | |
| "learning_rate": 1.833323282651869e-05, | |
| "loss": 0.0264, | |
| "step": 4430 | |
| }, | |
| { | |
| "epoch": 1.7250825723722558, | |
| "grad_norm": 0.2782775098963104, | |
| "learning_rate": 1.824312022085625e-05, | |
| "loss": 0.0256, | |
| "step": 4440 | |
| }, | |
| { | |
| "epoch": 1.7289683310666408, | |
| "grad_norm": 0.2244270504850629, | |
| "learning_rate": 1.8153043544303187e-05, | |
| "loss": 0.022, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 1.7328540897610258, | |
| "grad_norm": 0.19984321027915625, | |
| "learning_rate": 1.806300463897459e-05, | |
| "loss": 0.0258, | |
| "step": 4460 | |
| }, | |
| { | |
| "epoch": 1.7367398484554109, | |
| "grad_norm": 0.34358544847861805, | |
| "learning_rate": 1.7973005346213112e-05, | |
| "loss": 0.0247, | |
| "step": 4470 | |
| }, | |
| { | |
| "epoch": 1.7406256071497959, | |
| "grad_norm": 0.258634512592744, | |
| "learning_rate": 1.7883047506551323e-05, | |
| "loss": 0.0241, | |
| "step": 4480 | |
| }, | |
| { | |
| "epoch": 1.744511365844181, | |
| "grad_norm": 0.2993875451174172, | |
| "learning_rate": 1.779313295967404e-05, | |
| "loss": 0.025, | |
| "step": 4490 | |
| }, | |
| { | |
| "epoch": 1.748397124538566, | |
| "grad_norm": 0.2524644984996203, | |
| "learning_rate": 1.7703263544380712e-05, | |
| "loss": 0.0256, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 1.7522828832329513, | |
| "grad_norm": 0.15203592796958135, | |
| "learning_rate": 1.7613441098547844e-05, | |
| "loss": 0.0277, | |
| "step": 4510 | |
| }, | |
| { | |
| "epoch": 1.7561686419273363, | |
| "grad_norm": 0.3339303036694701, | |
| "learning_rate": 1.7523667459091372e-05, | |
| "loss": 0.0253, | |
| "step": 4520 | |
| }, | |
| { | |
| "epoch": 1.7600544006217214, | |
| "grad_norm": 0.10575616905501665, | |
| "learning_rate": 1.743394446192915e-05, | |
| "loss": 0.0239, | |
| "step": 4530 | |
| }, | |
| { | |
| "epoch": 1.7639401593161064, | |
| "grad_norm": 0.23336439363322206, | |
| "learning_rate": 1.734427394194331e-05, | |
| "loss": 0.0177, | |
| "step": 4540 | |
| }, | |
| { | |
| "epoch": 1.7678259180104916, | |
| "grad_norm": 0.18372002116915823, | |
| "learning_rate": 1.725465773294286e-05, | |
| "loss": 0.0238, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 1.7717116767048766, | |
| "grad_norm": 0.4517812923928783, | |
| "learning_rate": 1.7165097667626085e-05, | |
| "loss": 0.0241, | |
| "step": 4560 | |
| }, | |
| { | |
| "epoch": 1.7755974353992618, | |
| "grad_norm": 0.3032985382795743, | |
| "learning_rate": 1.7075595577543112e-05, | |
| "loss": 0.0246, | |
| "step": 4570 | |
| }, | |
| { | |
| "epoch": 1.7794831940936469, | |
| "grad_norm": 0.2883198267970469, | |
| "learning_rate": 1.698615329305846e-05, | |
| "loss": 0.0363, | |
| "step": 4580 | |
| }, | |
| { | |
| "epoch": 1.7833689527880319, | |
| "grad_norm": 0.212183942746847, | |
| "learning_rate": 1.6896772643313545e-05, | |
| "loss": 0.0184, | |
| "step": 4590 | |
| }, | |
| { | |
| "epoch": 1.7872547114824169, | |
| "grad_norm": 0.20532524865736027, | |
| "learning_rate": 1.6807455456189375e-05, | |
| "loss": 0.0238, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 1.791140470176802, | |
| "grad_norm": 0.34333620427983147, | |
| "learning_rate": 1.671820355826909e-05, | |
| "loss": 0.0262, | |
| "step": 4610 | |
| }, | |
| { | |
| "epoch": 1.7950262288711871, | |
| "grad_norm": 0.20001883941281645, | |
| "learning_rate": 1.6629018774800626e-05, | |
| "loss": 0.0272, | |
| "step": 4620 | |
| }, | |
| { | |
| "epoch": 1.7989119875655721, | |
| "grad_norm": 0.34691765585073897, | |
| "learning_rate": 1.6539902929659398e-05, | |
| "loss": 0.025, | |
| "step": 4630 | |
| }, | |
| { | |
| "epoch": 1.8027977462599574, | |
| "grad_norm": 0.2742832126748164, | |
| "learning_rate": 1.6450857845310995e-05, | |
| "loss": 0.0235, | |
| "step": 4640 | |
| }, | |
| { | |
| "epoch": 1.8066835049543424, | |
| "grad_norm": 0.28750335882588557, | |
| "learning_rate": 1.6361885342773928e-05, | |
| "loss": 0.0217, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 1.8105692636487274, | |
| "grad_norm": 0.20616162983525588, | |
| "learning_rate": 1.6272987241582344e-05, | |
| "loss": 0.0234, | |
| "step": 4660 | |
| }, | |
| { | |
| "epoch": 1.8144550223431124, | |
| "grad_norm": 0.28032518340338625, | |
| "learning_rate": 1.6184165359748873e-05, | |
| "loss": 0.0241, | |
| "step": 4670 | |
| }, | |
| { | |
| "epoch": 1.8183407810374974, | |
| "grad_norm": 0.19596210439624756, | |
| "learning_rate": 1.6095421513727393e-05, | |
| "loss": 0.0195, | |
| "step": 4680 | |
| }, | |
| { | |
| "epoch": 1.8222265397318826, | |
| "grad_norm": 0.23644823136897267, | |
| "learning_rate": 1.600675751837591e-05, | |
| "loss": 0.0219, | |
| "step": 4690 | |
| }, | |
| { | |
| "epoch": 1.8261122984262679, | |
| "grad_norm": 0.2630361162326872, | |
| "learning_rate": 1.591817518691947e-05, | |
| "loss": 0.0228, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 1.8299980571206529, | |
| "grad_norm": 0.2595409653034801, | |
| "learning_rate": 1.582967633091303e-05, | |
| "loss": 0.0231, | |
| "step": 4710 | |
| }, | |
| { | |
| "epoch": 1.833883815815038, | |
| "grad_norm": 0.3326391117683167, | |
| "learning_rate": 1.5741262760204424e-05, | |
| "loss": 0.0291, | |
| "step": 4720 | |
| }, | |
| { | |
| "epoch": 1.837769574509423, | |
| "grad_norm": 0.24035275119514785, | |
| "learning_rate": 1.5652936282897365e-05, | |
| "loss": 0.0219, | |
| "step": 4730 | |
| }, | |
| { | |
| "epoch": 1.841655333203808, | |
| "grad_norm": 0.244068103767282, | |
| "learning_rate": 1.5564698705314457e-05, | |
| "loss": 0.0224, | |
| "step": 4740 | |
| }, | |
| { | |
| "epoch": 1.8455410918981932, | |
| "grad_norm": 0.24784872474552172, | |
| "learning_rate": 1.5476551831960283e-05, | |
| "loss": 0.0269, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 1.8494268505925782, | |
| "grad_norm": 0.2820854981793686, | |
| "learning_rate": 1.5388497465484427e-05, | |
| "loss": 0.0236, | |
| "step": 4760 | |
| }, | |
| { | |
| "epoch": 1.8533126092869634, | |
| "grad_norm": 0.18652076792233116, | |
| "learning_rate": 1.5300537406644707e-05, | |
| "loss": 0.0248, | |
| "step": 4770 | |
| }, | |
| { | |
| "epoch": 1.8571983679813484, | |
| "grad_norm": 0.21164601939248687, | |
| "learning_rate": 1.5212673454270275e-05, | |
| "loss": 0.0187, | |
| "step": 4780 | |
| }, | |
| { | |
| "epoch": 1.8610841266757334, | |
| "grad_norm": 0.19650678419372633, | |
| "learning_rate": 1.5124907405224857e-05, | |
| "loss": 0.0202, | |
| "step": 4790 | |
| }, | |
| { | |
| "epoch": 1.8649698853701184, | |
| "grad_norm": 0.19671419185547953, | |
| "learning_rate": 1.5037241054370031e-05, | |
| "loss": 0.0226, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 1.8688556440645034, | |
| "grad_norm": 0.28963451060296624, | |
| "learning_rate": 1.4949676194528443e-05, | |
| "loss": 0.0212, | |
| "step": 4810 | |
| }, | |
| { | |
| "epoch": 1.8727414027588887, | |
| "grad_norm": 0.2291236266085277, | |
| "learning_rate": 1.4862214616447246e-05, | |
| "loss": 0.0265, | |
| "step": 4820 | |
| }, | |
| { | |
| "epoch": 1.876627161453274, | |
| "grad_norm": 0.21720512675401074, | |
| "learning_rate": 1.4774858108761399e-05, | |
| "loss": 0.0247, | |
| "step": 4830 | |
| }, | |
| { | |
| "epoch": 1.880512920147659, | |
| "grad_norm": 0.24043869727722034, | |
| "learning_rate": 1.4687608457957131e-05, | |
| "loss": 0.0237, | |
| "step": 4840 | |
| }, | |
| { | |
| "epoch": 1.884398678842044, | |
| "grad_norm": 0.3039350538024942, | |
| "learning_rate": 1.4600467448335377e-05, | |
| "loss": 0.0209, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 1.888284437536429, | |
| "grad_norm": 0.27910808051006536, | |
| "learning_rate": 1.4513436861975309e-05, | |
| "loss": 0.0239, | |
| "step": 4860 | |
| }, | |
| { | |
| "epoch": 1.892170196230814, | |
| "grad_norm": 0.186631017294127, | |
| "learning_rate": 1.4426518478697898e-05, | |
| "loss": 0.0267, | |
| "step": 4870 | |
| }, | |
| { | |
| "epoch": 1.8960559549251992, | |
| "grad_norm": 0.1881734007091214, | |
| "learning_rate": 1.4339714076029485e-05, | |
| "loss": 0.0218, | |
| "step": 4880 | |
| }, | |
| { | |
| "epoch": 1.8999417136195842, | |
| "grad_norm": 0.13892970586490075, | |
| "learning_rate": 1.4253025429165464e-05, | |
| "loss": 0.0212, | |
| "step": 4890 | |
| }, | |
| { | |
| "epoch": 1.9038274723139694, | |
| "grad_norm": 0.288495543969294, | |
| "learning_rate": 1.4166454310933941e-05, | |
| "loss": 0.0209, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 1.9077132310083544, | |
| "grad_norm": 0.2555017196077912, | |
| "learning_rate": 1.4080002491759519e-05, | |
| "loss": 0.0244, | |
| "step": 4910 | |
| }, | |
| { | |
| "epoch": 1.9115989897027394, | |
| "grad_norm": 0.18634334770662844, | |
| "learning_rate": 1.3993671739627072e-05, | |
| "loss": 0.0217, | |
| "step": 4920 | |
| }, | |
| { | |
| "epoch": 1.9154847483971245, | |
| "grad_norm": 0.3164288797558731, | |
| "learning_rate": 1.3907463820045589e-05, | |
| "loss": 0.0222, | |
| "step": 4930 | |
| }, | |
| { | |
| "epoch": 1.9193705070915095, | |
| "grad_norm": 0.2818084680815898, | |
| "learning_rate": 1.382138049601205e-05, | |
| "loss": 0.0218, | |
| "step": 4940 | |
| }, | |
| { | |
| "epoch": 1.9232562657858947, | |
| "grad_norm": 0.36520600715684526, | |
| "learning_rate": 1.3735423527975416e-05, | |
| "loss": 0.0268, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 1.92714202448028, | |
| "grad_norm": 0.29834225940452946, | |
| "learning_rate": 1.3649594673800585e-05, | |
| "loss": 0.027, | |
| "step": 4960 | |
| }, | |
| { | |
| "epoch": 1.931027783174665, | |
| "grad_norm": 0.24025465966717943, | |
| "learning_rate": 1.3563895688732476e-05, | |
| "loss": 0.0201, | |
| "step": 4970 | |
| }, | |
| { | |
| "epoch": 1.93491354186905, | |
| "grad_norm": 0.36547751994851313, | |
| "learning_rate": 1.3478328325360108e-05, | |
| "loss": 0.0175, | |
| "step": 4980 | |
| }, | |
| { | |
| "epoch": 1.938799300563435, | |
| "grad_norm": 0.30354949309324647, | |
| "learning_rate": 1.3392894333580757e-05, | |
| "loss": 0.0222, | |
| "step": 4990 | |
| }, | |
| { | |
| "epoch": 1.94268505925782, | |
| "grad_norm": 0.23953269705131022, | |
| "learning_rate": 1.3307595460564196e-05, | |
| "loss": 0.0245, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 1.9465708179522052, | |
| "grad_norm": 0.25015753705105026, | |
| "learning_rate": 1.3222433450716939e-05, | |
| "loss": 0.0273, | |
| "step": 5010 | |
| }, | |
| { | |
| "epoch": 1.9504565766465902, | |
| "grad_norm": 0.1601753057955312, | |
| "learning_rate": 1.3137410045646593e-05, | |
| "loss": 0.0223, | |
| "step": 5020 | |
| }, | |
| { | |
| "epoch": 1.9543423353409755, | |
| "grad_norm": 0.2242880052150341, | |
| "learning_rate": 1.3052526984126192e-05, | |
| "loss": 0.0183, | |
| "step": 5030 | |
| }, | |
| { | |
| "epoch": 1.9582280940353605, | |
| "grad_norm": 0.2607444099422183, | |
| "learning_rate": 1.2967786002058712e-05, | |
| "loss": 0.0241, | |
| "step": 5040 | |
| }, | |
| { | |
| "epoch": 1.9621138527297455, | |
| "grad_norm": 0.24290540890508416, | |
| "learning_rate": 1.2883188832441496e-05, | |
| "loss": 0.0226, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 1.9659996114241305, | |
| "grad_norm": 0.3409873620160428, | |
| "learning_rate": 1.2798737205330869e-05, | |
| "loss": 0.041, | |
| "step": 5060 | |
| }, | |
| { | |
| "epoch": 1.9698853701185155, | |
| "grad_norm": 0.1874649872758006, | |
| "learning_rate": 1.2714432847806721e-05, | |
| "loss": 0.0256, | |
| "step": 5070 | |
| }, | |
| { | |
| "epoch": 1.9737711288129007, | |
| "grad_norm": 0.18360888920333324, | |
| "learning_rate": 1.263027748393721e-05, | |
| "loss": 0.0219, | |
| "step": 5080 | |
| }, | |
| { | |
| "epoch": 1.9776568875072857, | |
| "grad_norm": 0.23427360199598463, | |
| "learning_rate": 1.2546272834743496e-05, | |
| "loss": 0.0221, | |
| "step": 5090 | |
| }, | |
| { | |
| "epoch": 1.981542646201671, | |
| "grad_norm": 0.31230044573711585, | |
| "learning_rate": 1.2462420618164548e-05, | |
| "loss": 0.0204, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 1.985428404896056, | |
| "grad_norm": 0.250237465475584, | |
| "learning_rate": 1.2378722549022e-05, | |
| "loss": 0.02, | |
| "step": 5110 | |
| }, | |
| { | |
| "epoch": 1.989314163590441, | |
| "grad_norm": 0.24977255925651687, | |
| "learning_rate": 1.2295180338985092e-05, | |
| "loss": 0.0221, | |
| "step": 5120 | |
| }, | |
| { | |
| "epoch": 1.993199922284826, | |
| "grad_norm": 0.3918157450288067, | |
| "learning_rate": 1.2211795696535664e-05, | |
| "loss": 0.0264, | |
| "step": 5130 | |
| }, | |
| { | |
| "epoch": 1.9970856809792112, | |
| "grad_norm": 0.24230360473123272, | |
| "learning_rate": 1.2128570326933224e-05, | |
| "loss": 0.0167, | |
| "step": 5140 | |
| }, | |
| { | |
| "epoch": 2.000777151738877, | |
| "grad_norm": 0.19021696383018177, | |
| "learning_rate": 1.2045505932180069e-05, | |
| "loss": 0.0227, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 2.0046629104332623, | |
| "grad_norm": 0.1996009218747256, | |
| "learning_rate": 1.1962604210986455e-05, | |
| "loss": 0.0214, | |
| "step": 5160 | |
| }, | |
| { | |
| "epoch": 2.0085486691276473, | |
| "grad_norm": 0.25730395478860757, | |
| "learning_rate": 1.18798668587359e-05, | |
| "loss": 0.0209, | |
| "step": 5170 | |
| }, | |
| { | |
| "epoch": 2.0124344278220323, | |
| "grad_norm": 0.1771816589400498, | |
| "learning_rate": 1.179729556745048e-05, | |
| "loss": 0.0188, | |
| "step": 5180 | |
| }, | |
| { | |
| "epoch": 2.0163201865164173, | |
| "grad_norm": 0.1992628058679068, | |
| "learning_rate": 1.171489202575624e-05, | |
| "loss": 0.0159, | |
| "step": 5190 | |
| }, | |
| { | |
| "epoch": 2.0202059452108023, | |
| "grad_norm": 0.15396300645984096, | |
| "learning_rate": 1.163265791884868e-05, | |
| "loss": 0.0165, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 2.0240917039051873, | |
| "grad_norm": 0.2379685245286237, | |
| "learning_rate": 1.1550594928458224e-05, | |
| "loss": 0.011, | |
| "step": 5210 | |
| }, | |
| { | |
| "epoch": 2.027977462599573, | |
| "grad_norm": 0.1501164586957791, | |
| "learning_rate": 1.14687047328159e-05, | |
| "loss": 0.0124, | |
| "step": 5220 | |
| }, | |
| { | |
| "epoch": 2.031863221293958, | |
| "grad_norm": 0.22785045077674615, | |
| "learning_rate": 1.138698900661901e-05, | |
| "loss": 0.0173, | |
| "step": 5230 | |
| }, | |
| { | |
| "epoch": 2.035748979988343, | |
| "grad_norm": 0.19489104500055154, | |
| "learning_rate": 1.130544942099685e-05, | |
| "loss": 0.0156, | |
| "step": 5240 | |
| }, | |
| { | |
| "epoch": 2.039634738682728, | |
| "grad_norm": 0.2578847419234991, | |
| "learning_rate": 1.1224087643476525e-05, | |
| "loss": 0.0138, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 2.043520497377113, | |
| "grad_norm": 0.24662743252166966, | |
| "learning_rate": 1.1142905337948905e-05, | |
| "loss": 0.0145, | |
| "step": 5260 | |
| }, | |
| { | |
| "epoch": 2.047406256071498, | |
| "grad_norm": 0.2704090578235268, | |
| "learning_rate": 1.1061904164634547e-05, | |
| "loss": 0.0201, | |
| "step": 5270 | |
| }, | |
| { | |
| "epoch": 2.051292014765883, | |
| "grad_norm": 0.2277257711569673, | |
| "learning_rate": 1.0981085780049783e-05, | |
| "loss": 0.0156, | |
| "step": 5280 | |
| }, | |
| { | |
| "epoch": 2.0551777734602683, | |
| "grad_norm": 0.24393861223637625, | |
| "learning_rate": 1.0900451836972779e-05, | |
| "loss": 0.0157, | |
| "step": 5290 | |
| }, | |
| { | |
| "epoch": 2.0590635321546533, | |
| "grad_norm": 0.22148212227817382, | |
| "learning_rate": 1.0820003984409809e-05, | |
| "loss": 0.0182, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 2.0629492908490383, | |
| "grad_norm": 0.17873179551577242, | |
| "learning_rate": 1.0739743867561484e-05, | |
| "loss": 0.0223, | |
| "step": 5310 | |
| }, | |
| { | |
| "epoch": 2.0668350495434233, | |
| "grad_norm": 0.26424010200649295, | |
| "learning_rate": 1.0659673127789123e-05, | |
| "loss": 0.0149, | |
| "step": 5320 | |
| }, | |
| { | |
| "epoch": 2.0707208082378084, | |
| "grad_norm": 0.14038451344369343, | |
| "learning_rate": 1.0579793402581208e-05, | |
| "loss": 0.0125, | |
| "step": 5330 | |
| }, | |
| { | |
| "epoch": 2.0746065669321934, | |
| "grad_norm": 0.22811068872894222, | |
| "learning_rate": 1.050010632551983e-05, | |
| "loss": 0.0132, | |
| "step": 5340 | |
| }, | |
| { | |
| "epoch": 2.078492325626579, | |
| "grad_norm": 0.20031881335249865, | |
| "learning_rate": 1.0420613526247356e-05, | |
| "loss": 0.0166, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 2.082378084320964, | |
| "grad_norm": 0.2633505577038377, | |
| "learning_rate": 1.0341316630433062e-05, | |
| "loss": 0.0146, | |
| "step": 5360 | |
| }, | |
| { | |
| "epoch": 2.086263843015349, | |
| "grad_norm": 0.18868039147588853, | |
| "learning_rate": 1.0262217259739897e-05, | |
| "loss": 0.0117, | |
| "step": 5370 | |
| }, | |
| { | |
| "epoch": 2.090149601709734, | |
| "grad_norm": 0.1483051784876369, | |
| "learning_rate": 1.0183317031791318e-05, | |
| "loss": 0.0175, | |
| "step": 5380 | |
| }, | |
| { | |
| "epoch": 2.094035360404119, | |
| "grad_norm": 0.1594861493991722, | |
| "learning_rate": 1.0104617560138205e-05, | |
| "loss": 0.0117, | |
| "step": 5390 | |
| }, | |
| { | |
| "epoch": 2.097921119098504, | |
| "grad_norm": 0.26030303220977924, | |
| "learning_rate": 1.0026120454225877e-05, | |
| "loss": 0.018, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 2.101806877792889, | |
| "grad_norm": 0.2308638280548361, | |
| "learning_rate": 9.947827319361152e-06, | |
| "loss": 0.0178, | |
| "step": 5410 | |
| }, | |
| { | |
| "epoch": 2.1056926364872743, | |
| "grad_norm": 0.10900704882374063, | |
| "learning_rate": 9.869739756679551e-06, | |
| "loss": 0.0165, | |
| "step": 5420 | |
| }, | |
| { | |
| "epoch": 2.1095783951816593, | |
| "grad_norm": 0.21796588094597985, | |
| "learning_rate": 9.791859363112521e-06, | |
| "loss": 0.0191, | |
| "step": 5430 | |
| }, | |
| { | |
| "epoch": 2.1134641538760444, | |
| "grad_norm": 0.1672673980678181, | |
| "learning_rate": 9.714187731354805e-06, | |
| "loss": 0.0177, | |
| "step": 5440 | |
| }, | |
| { | |
| "epoch": 2.1173499125704294, | |
| "grad_norm": 0.2684028767592411, | |
| "learning_rate": 9.63672644983185e-06, | |
| "loss": 0.0162, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 2.1212356712648144, | |
| "grad_norm": 0.2649724905943757, | |
| "learning_rate": 9.559477102667331e-06, | |
| "loss": 0.0159, | |
| "step": 5460 | |
| }, | |
| { | |
| "epoch": 2.1251214299591994, | |
| "grad_norm": 0.1855609816934914, | |
| "learning_rate": 9.482441269650762e-06, | |
| "loss": 0.0168, | |
| "step": 5470 | |
| }, | |
| { | |
| "epoch": 2.129007188653585, | |
| "grad_norm": 0.20799545221446245, | |
| "learning_rate": 9.405620526205173e-06, | |
| "loss": 0.0229, | |
| "step": 5480 | |
| }, | |
| { | |
| "epoch": 2.13289294734797, | |
| "grad_norm": 0.2169295403294615, | |
| "learning_rate": 9.329016443354899e-06, | |
| "loss": 0.0166, | |
| "step": 5490 | |
| }, | |
| { | |
| "epoch": 2.136778706042355, | |
| "grad_norm": 0.20155345013696116, | |
| "learning_rate": 9.25263058769347e-06, | |
| "loss": 0.0164, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 2.14066446473674, | |
| "grad_norm": 0.24455184564544635, | |
| "learning_rate": 9.176464521351517e-06, | |
| "loss": 0.0137, | |
| "step": 5510 | |
| }, | |
| { | |
| "epoch": 2.144550223431125, | |
| "grad_norm": 0.14491602829427766, | |
| "learning_rate": 9.100519801964913e-06, | |
| "loss": 0.0124, | |
| "step": 5520 | |
| }, | |
| { | |
| "epoch": 2.14843598212551, | |
| "grad_norm": 0.31403014524681455, | |
| "learning_rate": 9.024797982642841e-06, | |
| "loss": 0.0199, | |
| "step": 5530 | |
| }, | |
| { | |
| "epoch": 2.152321740819895, | |
| "grad_norm": 0.16133633738818282, | |
| "learning_rate": 8.949300611936065e-06, | |
| "loss": 0.0177, | |
| "step": 5540 | |
| }, | |
| { | |
| "epoch": 2.1562074995142804, | |
| "grad_norm": 0.34872137324864605, | |
| "learning_rate": 8.874029233805269e-06, | |
| "loss": 0.0176, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 2.1600932582086654, | |
| "grad_norm": 0.09809795832817098, | |
| "learning_rate": 8.798985387589436e-06, | |
| "loss": 0.0108, | |
| "step": 5560 | |
| }, | |
| { | |
| "epoch": 2.1639790169030504, | |
| "grad_norm": 0.2159359078852452, | |
| "learning_rate": 8.724170607974454e-06, | |
| "loss": 0.0147, | |
| "step": 5570 | |
| }, | |
| { | |
| "epoch": 2.1678647755974354, | |
| "grad_norm": 0.21468205780280344, | |
| "learning_rate": 8.649586424961645e-06, | |
| "loss": 0.0141, | |
| "step": 5580 | |
| }, | |
| { | |
| "epoch": 2.1717505342918204, | |
| "grad_norm": 0.2249780220781056, | |
| "learning_rate": 8.575234363836526e-06, | |
| "loss": 0.0186, | |
| "step": 5590 | |
| }, | |
| { | |
| "epoch": 2.1756362929862054, | |
| "grad_norm": 0.24124720508302283, | |
| "learning_rate": 8.501115945137577e-06, | |
| "loss": 0.0144, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 2.1795220516805904, | |
| "grad_norm": 0.26759279337799535, | |
| "learning_rate": 8.427232684625186e-06, | |
| "loss": 0.0193, | |
| "step": 5610 | |
| }, | |
| { | |
| "epoch": 2.183407810374976, | |
| "grad_norm": 0.2297544777314867, | |
| "learning_rate": 8.353586093250642e-06, | |
| "loss": 0.0147, | |
| "step": 5620 | |
| }, | |
| { | |
| "epoch": 2.187293569069361, | |
| "grad_norm": 0.23619415311251218, | |
| "learning_rate": 8.280177677125214e-06, | |
| "loss": 0.0166, | |
| "step": 5630 | |
| }, | |
| { | |
| "epoch": 2.191179327763746, | |
| "grad_norm": 0.1943560804911627, | |
| "learning_rate": 8.207008937489364e-06, | |
| "loss": 0.0172, | |
| "step": 5640 | |
| }, | |
| { | |
| "epoch": 2.195065086458131, | |
| "grad_norm": 0.22431037235771287, | |
| "learning_rate": 8.134081370682038e-06, | |
| "loss": 0.0161, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 2.198950845152516, | |
| "grad_norm": 0.21729765338858503, | |
| "learning_rate": 8.061396468110074e-06, | |
| "loss": 0.0175, | |
| "step": 5660 | |
| }, | |
| { | |
| "epoch": 2.202836603846901, | |
| "grad_norm": 0.18338237613630443, | |
| "learning_rate": 7.988955716217719e-06, | |
| "loss": 0.0106, | |
| "step": 5670 | |
| }, | |
| { | |
| "epoch": 2.206722362541286, | |
| "grad_norm": 0.20665577127269266, | |
| "learning_rate": 7.916760596456197e-06, | |
| "loss": 0.0138, | |
| "step": 5680 | |
| }, | |
| { | |
| "epoch": 2.2106081212356714, | |
| "grad_norm": 0.2316552326389069, | |
| "learning_rate": 7.84481258525341e-06, | |
| "loss": 0.0141, | |
| "step": 5690 | |
| }, | |
| { | |
| "epoch": 2.2144938799300564, | |
| "grad_norm": 0.26108396067769657, | |
| "learning_rate": 7.773113153983787e-06, | |
| "loss": 0.0169, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 2.2183796386244414, | |
| "grad_norm": 0.2243677380846199, | |
| "learning_rate": 7.701663768938146e-06, | |
| "loss": 0.0145, | |
| "step": 5710 | |
| }, | |
| { | |
| "epoch": 2.2222653973188264, | |
| "grad_norm": 0.18542682485202877, | |
| "learning_rate": 7.630465891293766e-06, | |
| "loss": 0.0184, | |
| "step": 5720 | |
| }, | |
| { | |
| "epoch": 2.2261511560132115, | |
| "grad_norm": 0.16604306041103264, | |
| "learning_rate": 7.559520977084416e-06, | |
| "loss": 0.0134, | |
| "step": 5730 | |
| }, | |
| { | |
| "epoch": 2.2300369147075965, | |
| "grad_norm": 0.15271062795031004, | |
| "learning_rate": 7.4888304771706675e-06, | |
| "loss": 0.0174, | |
| "step": 5740 | |
| }, | |
| { | |
| "epoch": 2.233922673401982, | |
| "grad_norm": 0.22620464359511355, | |
| "learning_rate": 7.418395837210177e-06, | |
| "loss": 0.0183, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 2.237808432096367, | |
| "grad_norm": 0.1932286630859644, | |
| "learning_rate": 7.34821849762813e-06, | |
| "loss": 0.015, | |
| "step": 5760 | |
| }, | |
| { | |
| "epoch": 2.241694190790752, | |
| "grad_norm": 0.20920536899184783, | |
| "learning_rate": 7.278299893587784e-06, | |
| "loss": 0.0128, | |
| "step": 5770 | |
| }, | |
| { | |
| "epoch": 2.245579949485137, | |
| "grad_norm": 0.1681609340326629, | |
| "learning_rate": 7.20864145496112e-06, | |
| "loss": 0.0139, | |
| "step": 5780 | |
| }, | |
| { | |
| "epoch": 2.249465708179522, | |
| "grad_norm": 0.23030013864097168, | |
| "learning_rate": 7.139244606299603e-06, | |
| "loss": 0.0202, | |
| "step": 5790 | |
| }, | |
| { | |
| "epoch": 2.253351466873907, | |
| "grad_norm": 0.20837500136002646, | |
| "learning_rate": 7.070110766805045e-06, | |
| "loss": 0.0151, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 2.257237225568292, | |
| "grad_norm": 0.09892626670584977, | |
| "learning_rate": 7.001241350300585e-06, | |
| "loss": 0.0139, | |
| "step": 5810 | |
| }, | |
| { | |
| "epoch": 2.2611229842626774, | |
| "grad_norm": 0.2421628202226961, | |
| "learning_rate": 6.932637765201767e-06, | |
| "loss": 0.0167, | |
| "step": 5820 | |
| }, | |
| { | |
| "epoch": 2.2650087429570624, | |
| "grad_norm": 0.22005154896696727, | |
| "learning_rate": 6.86430141448775e-06, | |
| "loss": 0.0163, | |
| "step": 5830 | |
| }, | |
| { | |
| "epoch": 2.2688945016514475, | |
| "grad_norm": 0.18906159377923581, | |
| "learning_rate": 6.796233695672611e-06, | |
| "loss": 0.0141, | |
| "step": 5840 | |
| }, | |
| { | |
| "epoch": 2.2727802603458325, | |
| "grad_norm": 0.1882303122476742, | |
| "learning_rate": 6.728436000776759e-06, | |
| "loss": 0.0167, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 2.2766660190402175, | |
| "grad_norm": 0.20942395612766407, | |
| "learning_rate": 6.6609097162984785e-06, | |
| "loss": 0.0167, | |
| "step": 5860 | |
| }, | |
| { | |
| "epoch": 2.2805517777346025, | |
| "grad_norm": 0.17048648869426325, | |
| "learning_rate": 6.593656223185565e-06, | |
| "loss": 0.0158, | |
| "step": 5870 | |
| }, | |
| { | |
| "epoch": 2.284437536428988, | |
| "grad_norm": 0.18912326858465023, | |
| "learning_rate": 6.526676896807092e-06, | |
| "loss": 0.0172, | |
| "step": 5880 | |
| }, | |
| { | |
| "epoch": 2.288323295123373, | |
| "grad_norm": 0.22429991946158165, | |
| "learning_rate": 6.459973106925272e-06, | |
| "loss": 0.0138, | |
| "step": 5890 | |
| }, | |
| { | |
| "epoch": 2.292209053817758, | |
| "grad_norm": 0.2719200154391883, | |
| "learning_rate": 6.393546217667464e-06, | |
| "loss": 0.0137, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 2.296094812512143, | |
| "grad_norm": 0.23193225915033114, | |
| "learning_rate": 6.327397587498254e-06, | |
| "loss": 0.0202, | |
| "step": 5910 | |
| }, | |
| { | |
| "epoch": 2.299980571206528, | |
| "grad_norm": 0.1773945719864865, | |
| "learning_rate": 6.26152856919169e-06, | |
| "loss": 0.0269, | |
| "step": 5920 | |
| }, | |
| { | |
| "epoch": 2.303866329900913, | |
| "grad_norm": 0.18357413512611562, | |
| "learning_rate": 6.19594050980361e-06, | |
| "loss": 0.0133, | |
| "step": 5930 | |
| }, | |
| { | |
| "epoch": 2.307752088595298, | |
| "grad_norm": 0.10329466907615475, | |
| "learning_rate": 6.130634750644102e-06, | |
| "loss": 0.0243, | |
| "step": 5940 | |
| }, | |
| { | |
| "epoch": 2.3116378472896835, | |
| "grad_norm": 0.26127302357511983, | |
| "learning_rate": 6.0656126272500485e-06, | |
| "loss": 0.0155, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 2.3155236059840685, | |
| "grad_norm": 0.15622640887182293, | |
| "learning_rate": 6.000875469357841e-06, | |
| "loss": 0.0101, | |
| "step": 5960 | |
| }, | |
| { | |
| "epoch": 2.3194093646784535, | |
| "grad_norm": 0.253685898111279, | |
| "learning_rate": 5.936424600876194e-06, | |
| "loss": 0.0189, | |
| "step": 5970 | |
| }, | |
| { | |
| "epoch": 2.3232951233728385, | |
| "grad_norm": 0.29404993122697953, | |
| "learning_rate": 5.872261339859038e-06, | |
| "loss": 0.0184, | |
| "step": 5980 | |
| }, | |
| { | |
| "epoch": 2.3271808820672235, | |
| "grad_norm": 0.14796428743903128, | |
| "learning_rate": 5.8083869984785836e-06, | |
| "loss": 0.0147, | |
| "step": 5990 | |
| }, | |
| { | |
| "epoch": 2.3310666407616085, | |
| "grad_norm": 0.20683726050768395, | |
| "learning_rate": 5.7448028829984745e-06, | |
| "loss": 0.0188, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 2.334952399455994, | |
| "grad_norm": 0.21092382019033648, | |
| "learning_rate": 5.681510293747092e-06, | |
| "loss": 0.016, | |
| "step": 6010 | |
| }, | |
| { | |
| "epoch": 2.338838158150379, | |
| "grad_norm": 0.21550234513080346, | |
| "learning_rate": 5.618510525090966e-06, | |
| "loss": 0.0189, | |
| "step": 6020 | |
| }, | |
| { | |
| "epoch": 2.342723916844764, | |
| "grad_norm": 0.16535694850993649, | |
| "learning_rate": 5.555804865408279e-06, | |
| "loss": 0.0191, | |
| "step": 6030 | |
| }, | |
| { | |
| "epoch": 2.346609675539149, | |
| "grad_norm": 0.21745148122170188, | |
| "learning_rate": 5.4933945970625225e-06, | |
| "loss": 0.0137, | |
| "step": 6040 | |
| }, | |
| { | |
| "epoch": 2.350495434233534, | |
| "grad_norm": 0.19584922849210273, | |
| "learning_rate": 5.431280996376294e-06, | |
| "loss": 0.0184, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 2.354381192927919, | |
| "grad_norm": 0.2120146691664677, | |
| "learning_rate": 5.369465333605172e-06, | |
| "loss": 0.0136, | |
| "step": 6060 | |
| }, | |
| { | |
| "epoch": 2.358266951622304, | |
| "grad_norm": 0.23504940817120928, | |
| "learning_rate": 5.307948872911772e-06, | |
| "loss": 0.0108, | |
| "step": 6070 | |
| }, | |
| { | |
| "epoch": 2.3621527103166895, | |
| "grad_norm": 0.2544407938266695, | |
| "learning_rate": 5.246732872339852e-06, | |
| "loss": 0.0162, | |
| "step": 6080 | |
| }, | |
| { | |
| "epoch": 2.3660384690110745, | |
| "grad_norm": 0.1787782566980142, | |
| "learning_rate": 5.185818583788596e-06, | |
| "loss": 0.0167, | |
| "step": 6090 | |
| }, | |
| { | |
| "epoch": 2.3699242277054595, | |
| "grad_norm": 0.18779821624668364, | |
| "learning_rate": 5.125207252987034e-06, | |
| "loss": 0.0164, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 2.3738099863998445, | |
| "grad_norm": 0.23079303888097902, | |
| "learning_rate": 5.064900119468544e-06, | |
| "loss": 0.0172, | |
| "step": 6110 | |
| }, | |
| { | |
| "epoch": 2.3776957450942295, | |
| "grad_norm": 0.15381004609158025, | |
| "learning_rate": 5.004898416545529e-06, | |
| "loss": 0.0184, | |
| "step": 6120 | |
| }, | |
| { | |
| "epoch": 2.3815815037886146, | |
| "grad_norm": 0.23537158479264925, | |
| "learning_rate": 4.945203371284147e-06, | |
| "loss": 0.0123, | |
| "step": 6130 | |
| }, | |
| { | |
| "epoch": 2.385467262483, | |
| "grad_norm": 0.15335724927915312, | |
| "learning_rate": 4.8858162044792654e-06, | |
| "loss": 0.0102, | |
| "step": 6140 | |
| }, | |
| { | |
| "epoch": 2.389353021177385, | |
| "grad_norm": 0.2832041836288288, | |
| "learning_rate": 4.826738130629473e-06, | |
| "loss": 0.0149, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 2.39323877987177, | |
| "grad_norm": 0.27813345843062076, | |
| "learning_rate": 4.767970357912246e-06, | |
| "loss": 0.0165, | |
| "step": 6160 | |
| }, | |
| { | |
| "epoch": 2.397124538566155, | |
| "grad_norm": 0.15264755246893946, | |
| "learning_rate": 4.7095140881592395e-06, | |
| "loss": 0.0153, | |
| "step": 6170 | |
| }, | |
| { | |
| "epoch": 2.40101029726054, | |
| "grad_norm": 0.24434480556286678, | |
| "learning_rate": 4.65137051683171e-06, | |
| "loss": 0.0203, | |
| "step": 6180 | |
| }, | |
| { | |
| "epoch": 2.404896055954925, | |
| "grad_norm": 0.17160358108568669, | |
| "learning_rate": 4.593540832996071e-06, | |
| "loss": 0.0126, | |
| "step": 6190 | |
| }, | |
| { | |
| "epoch": 2.40878181464931, | |
| "grad_norm": 0.2220871800365551, | |
| "learning_rate": 4.53602621929957e-06, | |
| "loss": 0.0174, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 2.4126675733436955, | |
| "grad_norm": 0.19387829594662281, | |
| "learning_rate": 4.478827851946102e-06, | |
| "loss": 0.0131, | |
| "step": 6210 | |
| }, | |
| { | |
| "epoch": 2.4165533320380805, | |
| "grad_norm": 0.2720443716921079, | |
| "learning_rate": 4.421946900672165e-06, | |
| "loss": 0.0174, | |
| "step": 6220 | |
| }, | |
| { | |
| "epoch": 2.4204390907324655, | |
| "grad_norm": 0.3727208802934669, | |
| "learning_rate": 4.365384528722931e-06, | |
| "loss": 0.0152, | |
| "step": 6230 | |
| }, | |
| { | |
| "epoch": 2.4243248494268506, | |
| "grad_norm": 0.2178296686386825, | |
| "learning_rate": 4.309141892828459e-06, | |
| "loss": 0.0142, | |
| "step": 6240 | |
| }, | |
| { | |
| "epoch": 2.4282106081212356, | |
| "grad_norm": 0.22728601802630374, | |
| "learning_rate": 4.2532201431800344e-06, | |
| "loss": 0.0115, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 2.4320963668156206, | |
| "grad_norm": 0.23843429430202012, | |
| "learning_rate": 4.197620423406657e-06, | |
| "loss": 0.0157, | |
| "step": 6260 | |
| }, | |
| { | |
| "epoch": 2.435982125510006, | |
| "grad_norm": 0.19116227547407497, | |
| "learning_rate": 4.1423438705516415e-06, | |
| "loss": 0.0154, | |
| "step": 6270 | |
| }, | |
| { | |
| "epoch": 2.439867884204391, | |
| "grad_norm": 0.22767774713522007, | |
| "learning_rate": 4.087391615049374e-06, | |
| "loss": 0.0141, | |
| "step": 6280 | |
| }, | |
| { | |
| "epoch": 2.443753642898776, | |
| "grad_norm": 0.29789802335991833, | |
| "learning_rate": 4.03276478070219e-06, | |
| "loss": 0.0159, | |
| "step": 6290 | |
| }, | |
| { | |
| "epoch": 2.447639401593161, | |
| "grad_norm": 0.24719135376807377, | |
| "learning_rate": 3.978464484657392e-06, | |
| "loss": 0.0157, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 2.451525160287546, | |
| "grad_norm": 0.32391016652290383, | |
| "learning_rate": 3.924491837384406e-06, | |
| "loss": 0.0149, | |
| "step": 6310 | |
| }, | |
| { | |
| "epoch": 2.455410918981931, | |
| "grad_norm": 0.20232458142688142, | |
| "learning_rate": 3.87084794265206e-06, | |
| "loss": 0.015, | |
| "step": 6320 | |
| }, | |
| { | |
| "epoch": 2.459296677676316, | |
| "grad_norm": 0.2712629616510874, | |
| "learning_rate": 3.817533897506036e-06, | |
| "loss": 0.0133, | |
| "step": 6330 | |
| }, | |
| { | |
| "epoch": 2.4631824363707016, | |
| "grad_norm": 0.17142782056379016, | |
| "learning_rate": 3.764550792246411e-06, | |
| "loss": 0.0178, | |
| "step": 6340 | |
| }, | |
| { | |
| "epoch": 2.4670681950650866, | |
| "grad_norm": 0.2625497363608154, | |
| "learning_rate": 3.7118997104053557e-06, | |
| "loss": 0.0177, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 2.4709539537594716, | |
| "grad_norm": 0.23418764898240915, | |
| "learning_rate": 3.659581728725017e-06, | |
| "loss": 0.0216, | |
| "step": 6360 | |
| }, | |
| { | |
| "epoch": 2.4748397124538566, | |
| "grad_norm": 0.2421515106857572, | |
| "learning_rate": 3.607597917135448e-06, | |
| "loss": 0.0102, | |
| "step": 6370 | |
| }, | |
| { | |
| "epoch": 2.4787254711482416, | |
| "grad_norm": 0.18002030517689566, | |
| "learning_rate": 3.5559493387327603e-06, | |
| "loss": 0.0125, | |
| "step": 6380 | |
| }, | |
| { | |
| "epoch": 2.4826112298426266, | |
| "grad_norm": 0.22127119378874585, | |
| "learning_rate": 3.5046370497573558e-06, | |
| "loss": 0.0171, | |
| "step": 6390 | |
| }, | |
| { | |
| "epoch": 2.486496988537012, | |
| "grad_norm": 0.24067053535828525, | |
| "learning_rate": 3.4536620995723524e-06, | |
| "loss": 0.0131, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 2.490382747231397, | |
| "grad_norm": 0.3548945672530695, | |
| "learning_rate": 3.4030255306421254e-06, | |
| "loss": 0.0172, | |
| "step": 6410 | |
| }, | |
| { | |
| "epoch": 2.494268505925782, | |
| "grad_norm": 0.34342268520149605, | |
| "learning_rate": 3.3527283785109565e-06, | |
| "loss": 0.0134, | |
| "step": 6420 | |
| }, | |
| { | |
| "epoch": 2.498154264620167, | |
| "grad_norm": 0.26685487402837554, | |
| "learning_rate": 3.3027716717818925e-06, | |
| "loss": 0.0121, | |
| "step": 6430 | |
| }, | |
| { | |
| "epoch": 2.502040023314552, | |
| "grad_norm": 0.23865361157570286, | |
| "learning_rate": 3.2531564320956745e-06, | |
| "loss": 0.0134, | |
| "step": 6440 | |
| }, | |
| { | |
| "epoch": 2.505925782008937, | |
| "grad_norm": 0.2262219648398464, | |
| "learning_rate": 3.2038836741098756e-06, | |
| "loss": 0.0164, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 2.509811540703322, | |
| "grad_norm": 0.11415732152468076, | |
| "learning_rate": 3.15495440547815e-06, | |
| "loss": 0.0126, | |
| "step": 6460 | |
| }, | |
| { | |
| "epoch": 2.513697299397707, | |
| "grad_norm": 0.1637749047242322, | |
| "learning_rate": 3.1063696268296063e-06, | |
| "loss": 0.0152, | |
| "step": 6470 | |
| }, | |
| { | |
| "epoch": 2.5175830580920926, | |
| "grad_norm": 0.19217591389578745, | |
| "learning_rate": 3.0581303317483367e-06, | |
| "loss": 0.0171, | |
| "step": 6480 | |
| }, | |
| { | |
| "epoch": 2.5214688167864776, | |
| "grad_norm": 0.1744921556168132, | |
| "learning_rate": 3.0102375067531375e-06, | |
| "loss": 0.0133, | |
| "step": 6490 | |
| }, | |
| { | |
| "epoch": 2.5253545754808626, | |
| "grad_norm": 0.19272649371157077, | |
| "learning_rate": 2.962692131277296e-06, | |
| "loss": 0.0131, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 2.5292403341752476, | |
| "grad_norm": 0.23696107664845892, | |
| "learning_rate": 2.9154951776485905e-06, | |
| "loss": 0.0169, | |
| "step": 6510 | |
| }, | |
| { | |
| "epoch": 2.5331260928696326, | |
| "grad_norm": 0.16637778487180105, | |
| "learning_rate": 2.8686476110693796e-06, | |
| "loss": 0.0167, | |
| "step": 6520 | |
| }, | |
| { | |
| "epoch": 2.537011851564018, | |
| "grad_norm": 0.22456349132177617, | |
| "learning_rate": 2.822150389596867e-06, | |
| "loss": 0.0134, | |
| "step": 6530 | |
| }, | |
| { | |
| "epoch": 2.540897610258403, | |
| "grad_norm": 0.3349504222512524, | |
| "learning_rate": 2.7760044641235295e-06, | |
| "loss": 0.0144, | |
| "step": 6540 | |
| }, | |
| { | |
| "epoch": 2.544783368952788, | |
| "grad_norm": 0.23230866877649609, | |
| "learning_rate": 2.730210778357649e-06, | |
| "loss": 0.0124, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 2.548669127647173, | |
| "grad_norm": 0.2876662681394313, | |
| "learning_rate": 2.6847702688040357e-06, | |
| "loss": 0.0174, | |
| "step": 6560 | |
| }, | |
| { | |
| "epoch": 2.552554886341558, | |
| "grad_norm": 0.24429770597495445, | |
| "learning_rate": 2.6396838647448353e-06, | |
| "loss": 0.0153, | |
| "step": 6570 | |
| }, | |
| { | |
| "epoch": 2.556440645035943, | |
| "grad_norm": 0.18614206123634575, | |
| "learning_rate": 2.594952488220577e-06, | |
| "loss": 0.0138, | |
| "step": 6580 | |
| }, | |
| { | |
| "epoch": 2.560326403730328, | |
| "grad_norm": 0.1707554494222318, | |
| "learning_rate": 2.550577054011274e-06, | |
| "loss": 0.0097, | |
| "step": 6590 | |
| }, | |
| { | |
| "epoch": 2.564212162424713, | |
| "grad_norm": 0.14383952391367794, | |
| "learning_rate": 2.5065584696177414e-06, | |
| "loss": 0.0149, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 2.5680979211190986, | |
| "grad_norm": 0.18379892796393424, | |
| "learning_rate": 2.4628976352430376e-06, | |
| "loss": 0.011, | |
| "step": 6610 | |
| }, | |
| { | |
| "epoch": 2.5719836798134836, | |
| "grad_norm": 0.24670126413024498, | |
| "learning_rate": 2.419595443774023e-06, | |
| "loss": 0.0133, | |
| "step": 6620 | |
| }, | |
| { | |
| "epoch": 2.5758694385078686, | |
| "grad_norm": 0.20077779669194387, | |
| "learning_rate": 2.3766527807631422e-06, | |
| "loss": 0.0143, | |
| "step": 6630 | |
| }, | |
| { | |
| "epoch": 2.5797551972022537, | |
| "grad_norm": 0.22864135011568798, | |
| "learning_rate": 2.33407052441029e-06, | |
| "loss": 0.0195, | |
| "step": 6640 | |
| }, | |
| { | |
| "epoch": 2.5836409558966387, | |
| "grad_norm": 0.1525621735784211, | |
| "learning_rate": 2.291849545544853e-06, | |
| "loss": 0.0186, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 2.587526714591024, | |
| "grad_norm": 0.23345566519875988, | |
| "learning_rate": 2.249990707607912e-06, | |
| "loss": 0.0134, | |
| "step": 6660 | |
| }, | |
| { | |
| "epoch": 2.591412473285409, | |
| "grad_norm": 0.2280149763833634, | |
| "learning_rate": 2.2084948666345695e-06, | |
| "loss": 0.0139, | |
| "step": 6670 | |
| }, | |
| { | |
| "epoch": 2.595298231979794, | |
| "grad_norm": 0.26595705322222424, | |
| "learning_rate": 2.1673628712364538e-06, | |
| "loss": 0.0142, | |
| "step": 6680 | |
| }, | |
| { | |
| "epoch": 2.599183990674179, | |
| "grad_norm": 0.3089658507273733, | |
| "learning_rate": 2.126595562584357e-06, | |
| "loss": 0.0182, | |
| "step": 6690 | |
| }, | |
| { | |
| "epoch": 2.603069749368564, | |
| "grad_norm": 0.2145635655464109, | |
| "learning_rate": 2.0861937743910456e-06, | |
| "loss": 0.0143, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 2.606955508062949, | |
| "grad_norm": 0.22760812179389878, | |
| "learning_rate": 2.046158332894195e-06, | |
| "loss": 0.014, | |
| "step": 6710 | |
| }, | |
| { | |
| "epoch": 2.610841266757334, | |
| "grad_norm": 0.20004208111148616, | |
| "learning_rate": 2.006490056839496e-06, | |
| "loss": 0.0138, | |
| "step": 6720 | |
| }, | |
| { | |
| "epoch": 2.614727025451719, | |
| "grad_norm": 0.1855756971876611, | |
| "learning_rate": 1.9671897574639233e-06, | |
| "loss": 0.0157, | |
| "step": 6730 | |
| }, | |
| { | |
| "epoch": 2.6186127841461047, | |
| "grad_norm": 0.23769731157003207, | |
| "learning_rate": 1.928258238479133e-06, | |
| "loss": 0.0164, | |
| "step": 6740 | |
| }, | |
| { | |
| "epoch": 2.6224985428404897, | |
| "grad_norm": 0.17194198866392246, | |
| "learning_rate": 1.8896962960550214e-06, | |
| "loss": 0.0123, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 2.6263843015348747, | |
| "grad_norm": 0.25716223538323435, | |
| "learning_rate": 1.8515047188034651e-06, | |
| "loss": 0.0132, | |
| "step": 6760 | |
| }, | |
| { | |
| "epoch": 2.6302700602292597, | |
| "grad_norm": 0.23923230278799787, | |
| "learning_rate": 1.8136842877621697e-06, | |
| "loss": 0.0132, | |
| "step": 6770 | |
| }, | |
| { | |
| "epoch": 2.6341558189236447, | |
| "grad_norm": 0.18419856300682974, | |
| "learning_rate": 1.7762357763787097e-06, | |
| "loss": 0.011, | |
| "step": 6780 | |
| }, | |
| { | |
| "epoch": 2.63804157761803, | |
| "grad_norm": 0.1915730985861053, | |
| "learning_rate": 1.7391599504947043e-06, | |
| "loss": 0.013, | |
| "step": 6790 | |
| }, | |
| { | |
| "epoch": 2.641927336312415, | |
| "grad_norm": 0.1566125187690401, | |
| "learning_rate": 1.702457568330167e-06, | |
| "loss": 0.0116, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 2.6458130950068, | |
| "grad_norm": 0.16344705005127574, | |
| "learning_rate": 1.666129380467989e-06, | |
| "loss": 0.0124, | |
| "step": 6810 | |
| }, | |
| { | |
| "epoch": 2.649698853701185, | |
| "grad_norm": 0.19446375972044477, | |
| "learning_rate": 1.63017612983859e-06, | |
| "loss": 0.0154, | |
| "step": 6820 | |
| }, | |
| { | |
| "epoch": 2.65358461239557, | |
| "grad_norm": 0.2483349003041754, | |
| "learning_rate": 1.5945985517047336e-06, | |
| "loss": 0.0176, | |
| "step": 6830 | |
| }, | |
| { | |
| "epoch": 2.657470371089955, | |
| "grad_norm": 0.1788660816854284, | |
| "learning_rate": 1.5593973736464718e-06, | |
| "loss": 0.0158, | |
| "step": 6840 | |
| }, | |
| { | |
| "epoch": 2.66135612978434, | |
| "grad_norm": 0.31388430244796356, | |
| "learning_rate": 1.5245733155462937e-06, | |
| "loss": 0.0117, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 2.6652418884787252, | |
| "grad_norm": 0.09656141608090874, | |
| "learning_rate": 1.4901270895743803e-06, | |
| "loss": 0.0179, | |
| "step": 6860 | |
| }, | |
| { | |
| "epoch": 2.6691276471731107, | |
| "grad_norm": 0.2572799089734027, | |
| "learning_rate": 1.4560594001740503e-06, | |
| "loss": 0.0125, | |
| "step": 6870 | |
| }, | |
| { | |
| "epoch": 2.6730134058674957, | |
| "grad_norm": 0.19767701320137746, | |
| "learning_rate": 1.4223709440473466e-06, | |
| "loss": 0.0149, | |
| "step": 6880 | |
| }, | |
| { | |
| "epoch": 2.6768991645618807, | |
| "grad_norm": 0.2720993871756462, | |
| "learning_rate": 1.3890624101407957e-06, | |
| "loss": 0.0142, | |
| "step": 6890 | |
| }, | |
| { | |
| "epoch": 2.6807849232562657, | |
| "grad_norm": 0.26464941453049734, | |
| "learning_rate": 1.356134479631328e-06, | |
| "loss": 0.0146, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 2.6846706819506507, | |
| "grad_norm": 0.19524351475983762, | |
| "learning_rate": 1.3235878259123226e-06, | |
| "loss": 0.0182, | |
| "step": 6910 | |
| }, | |
| { | |
| "epoch": 2.688556440645036, | |
| "grad_norm": 0.26891449363176717, | |
| "learning_rate": 1.2914231145798462e-06, | |
| "loss": 0.0185, | |
| "step": 6920 | |
| }, | |
| { | |
| "epoch": 2.692442199339421, | |
| "grad_norm": 0.3011765177624915, | |
| "learning_rate": 1.2596410034190543e-06, | |
| "loss": 0.014, | |
| "step": 6930 | |
| }, | |
| { | |
| "epoch": 2.696327958033806, | |
| "grad_norm": 0.3020550635806977, | |
| "learning_rate": 1.228242142390721e-06, | |
| "loss": 0.0128, | |
| "step": 6940 | |
| }, | |
| { | |
| "epoch": 2.700213716728191, | |
| "grad_norm": 0.24847929057204657, | |
| "learning_rate": 1.1972271736179653e-06, | |
| "loss": 0.0126, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 2.7040994754225762, | |
| "grad_norm": 0.14385502966292424, | |
| "learning_rate": 1.166596731373102e-06, | |
| "loss": 0.0142, | |
| "step": 6960 | |
| }, | |
| { | |
| "epoch": 2.7079852341169612, | |
| "grad_norm": 0.27969175657328743, | |
| "learning_rate": 1.1363514420646738e-06, | |
| "loss": 0.0122, | |
| "step": 6970 | |
| }, | |
| { | |
| "epoch": 2.7118709928113462, | |
| "grad_norm": 0.23975714101635787, | |
| "learning_rate": 1.1064919242246486e-06, | |
| "loss": 0.018, | |
| "step": 6980 | |
| }, | |
| { | |
| "epoch": 2.7157567515057313, | |
| "grad_norm": 0.18202493395247812, | |
| "learning_rate": 1.0770187884957673e-06, | |
| "loss": 0.0161, | |
| "step": 6990 | |
| }, | |
| { | |
| "epoch": 2.7196425102001167, | |
| "grad_norm": 0.15583830698569281, | |
| "learning_rate": 1.0479326376190602e-06, | |
| "loss": 0.0121, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 2.7235282688945017, | |
| "grad_norm": 0.23832508481487272, | |
| "learning_rate": 1.0192340664214995e-06, | |
| "loss": 0.0154, | |
| "step": 7010 | |
| }, | |
| { | |
| "epoch": 2.7274140275888867, | |
| "grad_norm": 0.16847478281366923, | |
| "learning_rate": 9.909236618038665e-07, | |
| "loss": 0.0146, | |
| "step": 7020 | |
| }, | |
| { | |
| "epoch": 2.7312997862832717, | |
| "grad_norm": 0.30569535019277033, | |
| "learning_rate": 9.630020027287213e-07, | |
| "loss": 0.0181, | |
| "step": 7030 | |
| }, | |
| { | |
| "epoch": 2.7351855449776568, | |
| "grad_norm": 0.23680858512407482, | |
| "learning_rate": 9.354696602085833e-07, | |
| "loss": 0.0179, | |
| "step": 7040 | |
| }, | |
| { | |
| "epoch": 2.739071303672042, | |
| "grad_norm": 0.24801887023205305, | |
| "learning_rate": 9.083271972942431e-07, | |
| "loss": 0.0178, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 2.7429570623664272, | |
| "grad_norm": 0.18244403571549045, | |
| "learning_rate": 8.815751690632423e-07, | |
| "loss": 0.0134, | |
| "step": 7060 | |
| }, | |
| { | |
| "epoch": 2.7468428210608122, | |
| "grad_norm": 0.3177601765145437, | |
| "learning_rate": 8.552141226085408e-07, | |
| "loss": 0.0135, | |
| "step": 7070 | |
| }, | |
| { | |
| "epoch": 2.7507285797551972, | |
| "grad_norm": 0.2478869375091213, | |
| "learning_rate": 8.292445970273055e-07, | |
| "loss": 0.0131, | |
| "step": 7080 | |
| }, | |
| { | |
| "epoch": 2.7546143384495823, | |
| "grad_norm": 0.16351159527021386, | |
| "learning_rate": 8.03667123409908e-07, | |
| "loss": 0.0116, | |
| "step": 7090 | |
| }, | |
| { | |
| "epoch": 2.7585000971439673, | |
| "grad_norm": 0.16164434978464962, | |
| "learning_rate": 7.784822248290424e-07, | |
| "loss": 0.0151, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 2.7623858558383523, | |
| "grad_norm": 0.19368227723734358, | |
| "learning_rate": 7.53690416329047e-07, | |
| "loss": 0.0115, | |
| "step": 7110 | |
| }, | |
| { | |
| "epoch": 2.7662716145327373, | |
| "grad_norm": 0.19844203086570225, | |
| "learning_rate": 7.292922049153528e-07, | |
| "loss": 0.0158, | |
| "step": 7120 | |
| }, | |
| { | |
| "epoch": 2.7701573732271227, | |
| "grad_norm": 0.12080846948403764, | |
| "learning_rate": 7.052880895441339e-07, | |
| "loss": 0.0117, | |
| "step": 7130 | |
| }, | |
| { | |
| "epoch": 2.7740431319215078, | |
| "grad_norm": 0.15523717768836803, | |
| "learning_rate": 6.816785611120913e-07, | |
| "loss": 0.0133, | |
| "step": 7140 | |
| }, | |
| { | |
| "epoch": 2.7779288906158928, | |
| "grad_norm": 0.25279826141773354, | |
| "learning_rate": 6.584641024464122e-07, | |
| "loss": 0.0108, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 2.7818146493102778, | |
| "grad_norm": 0.1026232646491208, | |
| "learning_rate": 6.356451882949088e-07, | |
| "loss": 0.0137, | |
| "step": 7160 | |
| }, | |
| { | |
| "epoch": 2.785700408004663, | |
| "grad_norm": 0.1843621477660204, | |
| "learning_rate": 6.132222853162972e-07, | |
| "loss": 0.0171, | |
| "step": 7170 | |
| }, | |
| { | |
| "epoch": 2.7895861666990482, | |
| "grad_norm": 0.2974243267699617, | |
| "learning_rate": 5.911958520706562e-07, | |
| "loss": 0.0105, | |
| "step": 7180 | |
| }, | |
| { | |
| "epoch": 2.7934719253934333, | |
| "grad_norm": 0.1450131937905085, | |
| "learning_rate": 5.695663390100548e-07, | |
| "loss": 0.0181, | |
| "step": 7190 | |
| }, | |
| { | |
| "epoch": 2.7973576840878183, | |
| "grad_norm": 0.19449961006516722, | |
| "learning_rate": 5.483341884693327e-07, | |
| "loss": 0.0151, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 2.8012434427822033, | |
| "grad_norm": 0.231835400931022, | |
| "learning_rate": 5.274998346570659e-07, | |
| "loss": 0.0174, | |
| "step": 7210 | |
| }, | |
| { | |
| "epoch": 2.8051292014765883, | |
| "grad_norm": 0.18990330928555402, | |
| "learning_rate": 5.070637036466753e-07, | |
| "loss": 0.0163, | |
| "step": 7220 | |
| }, | |
| { | |
| "epoch": 2.8090149601709733, | |
| "grad_norm": 0.19725672735140468, | |
| "learning_rate": 4.870262133677072e-07, | |
| "loss": 0.0117, | |
| "step": 7230 | |
| }, | |
| { | |
| "epoch": 2.8129007188653583, | |
| "grad_norm": 0.3493633046330908, | |
| "learning_rate": 4.6738777359731866e-07, | |
| "loss": 0.017, | |
| "step": 7240 | |
| }, | |
| { | |
| "epoch": 2.8167864775597433, | |
| "grad_norm": 0.2268035823210108, | |
| "learning_rate": 4.481487859518563e-07, | |
| "loss": 0.0112, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 2.8206722362541288, | |
| "grad_norm": 0.26841266073002745, | |
| "learning_rate": 4.293096438786726e-07, | |
| "loss": 0.0176, | |
| "step": 7260 | |
| }, | |
| { | |
| "epoch": 2.824557994948514, | |
| "grad_norm": 0.20423668146731536, | |
| "learning_rate": 4.108707326480632e-07, | |
| "loss": 0.0128, | |
| "step": 7270 | |
| }, | |
| { | |
| "epoch": 2.828443753642899, | |
| "grad_norm": 0.1743068137356259, | |
| "learning_rate": 3.9283242934539555e-07, | |
| "loss": 0.0111, | |
| "step": 7280 | |
| }, | |
| { | |
| "epoch": 2.832329512337284, | |
| "grad_norm": 0.2340589582843017, | |
| "learning_rate": 3.751951028633971e-07, | |
| "loss": 0.0125, | |
| "step": 7290 | |
| }, | |
| { | |
| "epoch": 2.836215271031669, | |
| "grad_norm": 0.21054160485426174, | |
| "learning_rate": 3.5795911389461033e-07, | |
| "loss": 0.012, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 2.8401010297260543, | |
| "grad_norm": 0.17738056254213855, | |
| "learning_rate": 3.411248149240165e-07, | |
| "loss": 0.0131, | |
| "step": 7310 | |
| }, | |
| { | |
| "epoch": 2.8439867884204393, | |
| "grad_norm": 0.18120809069881422, | |
| "learning_rate": 3.24692550221819e-07, | |
| "loss": 0.0126, | |
| "step": 7320 | |
| }, | |
| { | |
| "epoch": 2.8478725471148243, | |
| "grad_norm": 0.27728578365868417, | |
| "learning_rate": 3.086626558364203e-07, | |
| "loss": 0.0133, | |
| "step": 7330 | |
| }, | |
| { | |
| "epoch": 2.8517583058092093, | |
| "grad_norm": 0.15083855508982252, | |
| "learning_rate": 2.93035459587534e-07, | |
| "loss": 0.0111, | |
| "step": 7340 | |
| }, | |
| { | |
| "epoch": 2.8556440645035943, | |
| "grad_norm": 0.20682581691820806, | |
| "learning_rate": 2.7781128105949015e-07, | |
| "loss": 0.0158, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 2.8595298231979793, | |
| "grad_norm": 0.19888880451656954, | |
| "learning_rate": 2.6299043159468963e-07, | |
| "loss": 0.0199, | |
| "step": 7360 | |
| }, | |
| { | |
| "epoch": 2.8634155818923643, | |
| "grad_norm": 0.24115048647296294, | |
| "learning_rate": 2.485732142872488e-07, | |
| "loss": 0.0119, | |
| "step": 7370 | |
| }, | |
| { | |
| "epoch": 2.8673013405867493, | |
| "grad_norm": 0.20653770018813603, | |
| "learning_rate": 2.3455992397679595e-07, | |
| "loss": 0.0158, | |
| "step": 7380 | |
| }, | |
| { | |
| "epoch": 2.871187099281135, | |
| "grad_norm": 0.18076322308667522, | |
| "learning_rate": 2.2095084724243598e-07, | |
| "loss": 0.0125, | |
| "step": 7390 | |
| }, | |
| { | |
| "epoch": 2.87507285797552, | |
| "grad_norm": 0.20736237506128907, | |
| "learning_rate": 2.0774626239690176e-07, | |
| "loss": 0.0117, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 2.878958616669905, | |
| "grad_norm": 0.195234932120182, | |
| "learning_rate": 1.9494643948084979e-07, | |
| "loss": 0.0141, | |
| "step": 7410 | |
| }, | |
| { | |
| "epoch": 2.88284437536429, | |
| "grad_norm": 0.2350430358349683, | |
| "learning_rate": 1.8255164025734684e-07, | |
| "loss": 0.0123, | |
| "step": 7420 | |
| }, | |
| { | |
| "epoch": 2.886730134058675, | |
| "grad_norm": 0.2192321643235776, | |
| "learning_rate": 1.7056211820651425e-07, | |
| "loss": 0.0128, | |
| "step": 7430 | |
| }, | |
| { | |
| "epoch": 2.8906158927530603, | |
| "grad_norm": 0.2649035408787872, | |
| "learning_rate": 1.5897811852033873e-07, | |
| "loss": 0.0174, | |
| "step": 7440 | |
| }, | |
| { | |
| "epoch": 2.8945016514474453, | |
| "grad_norm": 0.1967812235705591, | |
| "learning_rate": 1.4779987809766528e-07, | |
| "loss": 0.0143, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 2.8983874101418303, | |
| "grad_norm": 0.1249890086380848, | |
| "learning_rate": 1.3702762553935656e-07, | |
| "loss": 0.0119, | |
| "step": 7460 | |
| }, | |
| { | |
| "epoch": 2.9022731688362153, | |
| "grad_norm": 0.22978595448983524, | |
| "learning_rate": 1.2666158114359894e-07, | |
| "loss": 0.0146, | |
| "step": 7470 | |
| }, | |
| { | |
| "epoch": 2.9061589275306003, | |
| "grad_norm": 0.16965468919403492, | |
| "learning_rate": 1.1670195690141939e-07, | |
| "loss": 0.0138, | |
| "step": 7480 | |
| }, | |
| { | |
| "epoch": 2.9100446862249854, | |
| "grad_norm": 0.28228206459702576, | |
| "learning_rate": 1.0714895649233781e-07, | |
| "loss": 0.0129, | |
| "step": 7490 | |
| }, | |
| { | |
| "epoch": 2.9139304449193704, | |
| "grad_norm": 0.20835241859756748, | |
| "learning_rate": 9.800277528020153e-08, | |
| "loss": 0.0151, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 2.9178162036137554, | |
| "grad_norm": 0.16424147151499843, | |
| "learning_rate": 8.926360030919513e-08, | |
| "loss": 0.0149, | |
| "step": 7510 | |
| }, | |
| { | |
| "epoch": 2.9217019623081404, | |
| "grad_norm": 0.1759921900549206, | |
| "learning_rate": 8.093161030001462e-08, | |
| "loss": 0.0108, | |
| "step": 7520 | |
| }, | |
| { | |
| "epoch": 2.925587721002526, | |
| "grad_norm": 0.2025940367611509, | |
| "learning_rate": 7.300697564620596e-08, | |
| "loss": 0.0124, | |
| "step": 7530 | |
| }, | |
| { | |
| "epoch": 2.929473479696911, | |
| "grad_norm": 0.3091292487177528, | |
| "learning_rate": 6.548985841069e-08, | |
| "loss": 0.0171, | |
| "step": 7540 | |
| }, | |
| { | |
| "epoch": 2.933359238391296, | |
| "grad_norm": 0.18550760451795564, | |
| "learning_rate": 5.8380412322440736e-08, | |
| "loss": 0.0098, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 2.937244997085681, | |
| "grad_norm": 0.2399355018987987, | |
| "learning_rate": 5.167878277334559e-08, | |
| "loss": 0.0105, | |
| "step": 7560 | |
| }, | |
| { | |
| "epoch": 2.9411307557800663, | |
| "grad_norm": 0.2129680096079058, | |
| "learning_rate": 4.538510681523001e-08, | |
| "loss": 0.0151, | |
| "step": 7570 | |
| }, | |
| { | |
| "epoch": 2.9450165144744513, | |
| "grad_norm": 0.15339355552498427, | |
| "learning_rate": 3.949951315705303e-08, | |
| "loss": 0.0154, | |
| "step": 7580 | |
| }, | |
| { | |
| "epoch": 2.9489022731688364, | |
| "grad_norm": 0.20862935241082844, | |
| "learning_rate": 3.4022122162282736e-08, | |
| "loss": 0.0114, | |
| "step": 7590 | |
| }, | |
| { | |
| "epoch": 2.9527880318632214, | |
| "grad_norm": 0.23246847497172307, | |
| "learning_rate": 2.895304584642711e-08, | |
| "loss": 0.0177, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 2.9566737905576064, | |
| "grad_norm": 0.20519849079048713, | |
| "learning_rate": 2.429238787474475e-08, | |
| "loss": 0.0126, | |
| "step": 7610 | |
| }, | |
| { | |
| "epoch": 2.9605595492519914, | |
| "grad_norm": 0.19918834292961288, | |
| "learning_rate": 2.004024356012435e-08, | |
| "loss": 0.0161, | |
| "step": 7620 | |
| }, | |
| { | |
| "epoch": 2.9644453079463764, | |
| "grad_norm": 0.19800296749468346, | |
| "learning_rate": 1.6196699861139586e-08, | |
| "loss": 0.0151, | |
| "step": 7630 | |
| }, | |
| { | |
| "epoch": 2.9683310666407614, | |
| "grad_norm": 0.1967950909462482, | |
| "learning_rate": 1.2761835380268317e-08, | |
| "loss": 0.0159, | |
| "step": 7640 | |
| }, | |
| { | |
| "epoch": 2.9722168253351464, | |
| "grad_norm": 0.2624813305392548, | |
| "learning_rate": 9.735720362282763e-09, | |
| "loss": 0.016, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 2.976102584029532, | |
| "grad_norm": 0.24779472153476578, | |
| "learning_rate": 7.1184166928151e-09, | |
| "loss": 0.0145, | |
| "step": 7660 | |
| }, | |
| { | |
| "epoch": 2.979988342723917, | |
| "grad_norm": 0.25286229822920325, | |
| "learning_rate": 4.90997789709402e-09, | |
| "loss": 0.0131, | |
| "step": 7670 | |
| }, | |
| { | |
| "epoch": 2.983874101418302, | |
| "grad_norm": 0.1409302028113683, | |
| "learning_rate": 3.1104491388478375e-09, | |
| "loss": 0.0154, | |
| "step": 7680 | |
| }, | |
| { | |
| "epoch": 2.987759860112687, | |
| "grad_norm": 0.24024651305037117, | |
| "learning_rate": 1.719867219378557e-09, | |
| "loss": 0.0142, | |
| "step": 7690 | |
| }, | |
| { | |
| "epoch": 2.991645618807072, | |
| "grad_norm": 0.1761325503880173, | |
| "learning_rate": 7.382605768113671e-10, | |
| "loss": 0.0123, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 2.9955313775014574, | |
| "grad_norm": 0.17792976628823454, | |
| "learning_rate": 1.6564928551732195e-10, | |
| "loss": 0.0113, | |
| "step": 7710 | |
| }, | |
| { | |
| "epoch": 2.999028560326404, | |
| "step": 7719, | |
| "total_flos": 301241258213376.0, | |
| "train_loss": 0.03801638325974335, | |
| "train_runtime": 73801.1602, | |
| "train_samples_per_second": 1.674, | |
| "train_steps_per_second": 0.105 | |
| } | |
| ], | |
| "logging_steps": 10, | |
| "max_steps": 7719, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 3, | |
| "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": 301241258213376.0, | |
| "train_batch_size": 1, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |