Instructions to use arthrod/gliner-ettin-68m-ptbr-pii-full-3x-v1 with libraries, inference providers, notebooks, and local apps. Follow these links to get started.
- Libraries
- GLiNER
How to use arthrod/gliner-ettin-68m-ptbr-pii-full-3x-v1 with GLiNER:
from gliner import GLiNER model = GLiNER.from_pretrained("arthrod/gliner-ettin-68m-ptbr-pii-full-3x-v1") - Notebooks
- Google Colab
- Kaggle
| { | |
| "best_global_step": null, | |
| "best_metric": null, | |
| "best_model_checkpoint": null, | |
| "epoch": 3.139625585023401, | |
| "eval_steps": 3450, | |
| "global_step": 24150, | |
| "is_hyper_param_search": false, | |
| "is_local_process_zero": true, | |
| "is_world_process_zero": true, | |
| "log_history": [ | |
| { | |
| "epoch": 0.003250130005200208, | |
| "grad_norm": 0.17285723984241486, | |
| "learning_rate": 1.7391304347826088e-07, | |
| "loss": 0.003443165719509125, | |
| "step": 25 | |
| }, | |
| { | |
| "epoch": 0.006500260010400416, | |
| "grad_norm": 0.16427233815193176, | |
| "learning_rate": 3.5507246376811595e-07, | |
| "loss": 0.0031772717833518983, | |
| "step": 50 | |
| }, | |
| { | |
| "epoch": 0.009750390015600624, | |
| "grad_norm": 0.12062286585569382, | |
| "learning_rate": 5.36231884057971e-07, | |
| "loss": 0.0028067561984062195, | |
| "step": 75 | |
| }, | |
| { | |
| "epoch": 0.013000520020800831, | |
| "grad_norm": 0.036187898367643356, | |
| "learning_rate": 7.17391304347826e-07, | |
| "loss": 0.002082071602344513, | |
| "step": 100 | |
| }, | |
| { | |
| "epoch": 0.01625065002600104, | |
| "grad_norm": 0.07566563785076141, | |
| "learning_rate": 8.985507246376812e-07, | |
| "loss": 0.0016960065066814424, | |
| "step": 125 | |
| }, | |
| { | |
| "epoch": 0.01950078003120125, | |
| "grad_norm": 0.0404927060008049, | |
| "learning_rate": 1.0797101449275362e-06, | |
| "loss": 0.001293088048696518, | |
| "step": 150 | |
| }, | |
| { | |
| "epoch": 0.022750910036401457, | |
| "grad_norm": 0.02149803936481476, | |
| "learning_rate": 1.2608695652173913e-06, | |
| "loss": 0.0010677673667669296, | |
| "step": 175 | |
| }, | |
| { | |
| "epoch": 0.026001040041601663, | |
| "grad_norm": 0.014630273915827274, | |
| "learning_rate": 1.4420289855072464e-06, | |
| "loss": 0.0008378532528877258, | |
| "step": 200 | |
| }, | |
| { | |
| "epoch": 0.02925117004680187, | |
| "grad_norm": 0.012398317456245422, | |
| "learning_rate": 1.6231884057971016e-06, | |
| "loss": 0.0008191241323947906, | |
| "step": 225 | |
| }, | |
| { | |
| "epoch": 0.03250130005200208, | |
| "grad_norm": 0.006211752071976662, | |
| "learning_rate": 1.8043478260869565e-06, | |
| "loss": 0.0007002048194408416, | |
| "step": 250 | |
| }, | |
| { | |
| "epoch": 0.035751430057202285, | |
| "grad_norm": 0.004569601267576218, | |
| "learning_rate": 1.9855072463768116e-06, | |
| "loss": 0.0007424639165401459, | |
| "step": 275 | |
| }, | |
| { | |
| "epoch": 0.0390015600624025, | |
| "grad_norm": 0.0053680334240198135, | |
| "learning_rate": 2.166666666666667e-06, | |
| "loss": 0.0006691357493400574, | |
| "step": 300 | |
| }, | |
| { | |
| "epoch": 0.0422516900676027, | |
| "grad_norm": 0.005433837883174419, | |
| "learning_rate": 2.347826086956522e-06, | |
| "loss": 0.0007775794714689255, | |
| "step": 325 | |
| }, | |
| { | |
| "epoch": 0.045501820072802915, | |
| "grad_norm": 0.005344253499060869, | |
| "learning_rate": 2.5289855072463768e-06, | |
| "loss": 0.0006336342543363571, | |
| "step": 350 | |
| }, | |
| { | |
| "epoch": 0.04875195007800312, | |
| "grad_norm": 0.006653524469584227, | |
| "learning_rate": 2.710144927536232e-06, | |
| "loss": 0.0006941972672939301, | |
| "step": 375 | |
| }, | |
| { | |
| "epoch": 0.052002080083203325, | |
| "grad_norm": 0.002266584662720561, | |
| "learning_rate": 2.8913043478260875e-06, | |
| "loss": 0.0006006322428584098, | |
| "step": 400 | |
| }, | |
| { | |
| "epoch": 0.05525221008840354, | |
| "grad_norm": 0.0035597020760178566, | |
| "learning_rate": 3.072463768115942e-06, | |
| "loss": 0.0007461020350456237, | |
| "step": 425 | |
| }, | |
| { | |
| "epoch": 0.05850234009360374, | |
| "grad_norm": 0.005806833505630493, | |
| "learning_rate": 3.2536231884057973e-06, | |
| "loss": 0.0006669971346855163, | |
| "step": 450 | |
| }, | |
| { | |
| "epoch": 0.061752470098803955, | |
| "grad_norm": 0.004355548415333033, | |
| "learning_rate": 3.4347826086956526e-06, | |
| "loss": 0.0007137749344110489, | |
| "step": 475 | |
| }, | |
| { | |
| "epoch": 0.06500260010400416, | |
| "grad_norm": 0.003757717553526163, | |
| "learning_rate": 3.6159420289855076e-06, | |
| "loss": 0.000681799128651619, | |
| "step": 500 | |
| }, | |
| { | |
| "epoch": 0.06825273010920437, | |
| "grad_norm": 0.0034986482933163643, | |
| "learning_rate": 3.797101449275362e-06, | |
| "loss": 0.0006944525241851806, | |
| "step": 525 | |
| }, | |
| { | |
| "epoch": 0.07150286011440457, | |
| "grad_norm": 0.0026165375020354986, | |
| "learning_rate": 3.978260869565217e-06, | |
| "loss": 0.0006353378295898438, | |
| "step": 550 | |
| }, | |
| { | |
| "epoch": 0.07475299011960479, | |
| "grad_norm": 0.004414152819663286, | |
| "learning_rate": 4.159420289855073e-06, | |
| "loss": 0.0006701602041721344, | |
| "step": 575 | |
| }, | |
| { | |
| "epoch": 0.078003120124805, | |
| "grad_norm": 0.0028373319655656815, | |
| "learning_rate": 4.340579710144928e-06, | |
| "loss": 0.0006536580622196198, | |
| "step": 600 | |
| }, | |
| { | |
| "epoch": 0.0812532501300052, | |
| "grad_norm": 0.003998955246061087, | |
| "learning_rate": 4.5217391304347826e-06, | |
| "loss": 0.0006748034805059433, | |
| "step": 625 | |
| }, | |
| { | |
| "epoch": 0.0845033801352054, | |
| "grad_norm": 0.0053115058690309525, | |
| "learning_rate": 4.702898550724638e-06, | |
| "loss": 0.0005269774794578552, | |
| "step": 650 | |
| }, | |
| { | |
| "epoch": 0.08775351014040561, | |
| "grad_norm": 0.002853003330528736, | |
| "learning_rate": 4.884057971014493e-06, | |
| "loss": 0.0006208278983831406, | |
| "step": 675 | |
| }, | |
| { | |
| "epoch": 0.09100364014560583, | |
| "grad_norm": 0.002200072631239891, | |
| "learning_rate": 5.065217391304348e-06, | |
| "loss": 0.0005327824130654335, | |
| "step": 700 | |
| }, | |
| { | |
| "epoch": 0.09425377015080603, | |
| "grad_norm": 0.0036399802193045616, | |
| "learning_rate": 5.246376811594203e-06, | |
| "loss": 0.000596022792160511, | |
| "step": 725 | |
| }, | |
| { | |
| "epoch": 0.09750390015600624, | |
| "grad_norm": 0.002803057199344039, | |
| "learning_rate": 5.427536231884058e-06, | |
| "loss": 0.0006214606761932373, | |
| "step": 750 | |
| }, | |
| { | |
| "epoch": 0.10075403016120645, | |
| "grad_norm": 0.0016377355204895139, | |
| "learning_rate": 5.608695652173914e-06, | |
| "loss": 0.0005090900138020515, | |
| "step": 775 | |
| }, | |
| { | |
| "epoch": 0.10400416016640665, | |
| "grad_norm": 0.0022944787051528692, | |
| "learning_rate": 5.789855072463768e-06, | |
| "loss": 0.0005760033056139946, | |
| "step": 800 | |
| }, | |
| { | |
| "epoch": 0.10725429017160687, | |
| "grad_norm": 0.003916461952030659, | |
| "learning_rate": 5.9710144927536236e-06, | |
| "loss": 0.0005986005067825318, | |
| "step": 825 | |
| }, | |
| { | |
| "epoch": 0.11050442017680707, | |
| "grad_norm": 0.00359009625390172, | |
| "learning_rate": 6.152173913043478e-06, | |
| "loss": 0.0005693147704005241, | |
| "step": 850 | |
| }, | |
| { | |
| "epoch": 0.11375455018200728, | |
| "grad_norm": 0.0059497179463505745, | |
| "learning_rate": 6.333333333333334e-06, | |
| "loss": 0.0005642465874552727, | |
| "step": 875 | |
| }, | |
| { | |
| "epoch": 0.11700468018720749, | |
| "grad_norm": 0.0013701318530365825, | |
| "learning_rate": 6.514492753623189e-06, | |
| "loss": 0.0005200594663619995, | |
| "step": 900 | |
| }, | |
| { | |
| "epoch": 0.12025481019240769, | |
| "grad_norm": 0.0031415170524269342, | |
| "learning_rate": 6.695652173913043e-06, | |
| "loss": 0.0005143824592232704, | |
| "step": 925 | |
| }, | |
| { | |
| "epoch": 0.12350494019760791, | |
| "grad_norm": 0.0034513864666223526, | |
| "learning_rate": 6.876811594202899e-06, | |
| "loss": 0.0005369628965854645, | |
| "step": 950 | |
| }, | |
| { | |
| "epoch": 0.1267550702028081, | |
| "grad_norm": 0.0037151614669710398, | |
| "learning_rate": 7.057971014492754e-06, | |
| "loss": 0.000499497763812542, | |
| "step": 975 | |
| }, | |
| { | |
| "epoch": 0.13000520020800832, | |
| "grad_norm": 0.0036164678167551756, | |
| "learning_rate": 7.239130434782609e-06, | |
| "loss": 0.0005703026056289673, | |
| "step": 1000 | |
| }, | |
| { | |
| "epoch": 0.13325533021320854, | |
| "grad_norm": 0.0007839290192350745, | |
| "learning_rate": 7.420289855072464e-06, | |
| "loss": 0.0005447989702224731, | |
| "step": 1025 | |
| }, | |
| { | |
| "epoch": 0.13650546021840873, | |
| "grad_norm": 0.0009723713155835867, | |
| "learning_rate": 7.60144927536232e-06, | |
| "loss": 0.00040120664983987807, | |
| "step": 1050 | |
| }, | |
| { | |
| "epoch": 0.13975559022360895, | |
| "grad_norm": 0.0060493540950119495, | |
| "learning_rate": 7.782608695652174e-06, | |
| "loss": 0.000505555048584938, | |
| "step": 1075 | |
| }, | |
| { | |
| "epoch": 0.14300572022880914, | |
| "grad_norm": 0.0019169203005731106, | |
| "learning_rate": 7.963768115942028e-06, | |
| "loss": 0.0005860131233930588, | |
| "step": 1100 | |
| }, | |
| { | |
| "epoch": 0.14625585023400936, | |
| "grad_norm": 0.0017371033318340778, | |
| "learning_rate": 8.144927536231884e-06, | |
| "loss": 0.0005623264610767364, | |
| "step": 1125 | |
| }, | |
| { | |
| "epoch": 0.14950598023920958, | |
| "grad_norm": 0.0028496291488409042, | |
| "learning_rate": 8.326086956521739e-06, | |
| "loss": 0.0005109354481101036, | |
| "step": 1150 | |
| }, | |
| { | |
| "epoch": 0.15275611024440977, | |
| "grad_norm": 0.0014431331073865294, | |
| "learning_rate": 8.507246376811595e-06, | |
| "loss": 0.0005302329733967781, | |
| "step": 1175 | |
| }, | |
| { | |
| "epoch": 0.15600624024961, | |
| "grad_norm": 0.0023162339348345995, | |
| "learning_rate": 8.68840579710145e-06, | |
| "loss": 0.000517817810177803, | |
| "step": 1200 | |
| }, | |
| { | |
| "epoch": 0.15925637025481018, | |
| "grad_norm": 0.0029457092750817537, | |
| "learning_rate": 8.869565217391304e-06, | |
| "loss": 0.0005169714242219925, | |
| "step": 1225 | |
| }, | |
| { | |
| "epoch": 0.1625065002600104, | |
| "grad_norm": 0.0017235144041478634, | |
| "learning_rate": 9.05072463768116e-06, | |
| "loss": 0.0005387067794799805, | |
| "step": 1250 | |
| }, | |
| { | |
| "epoch": 0.16575663026521062, | |
| "grad_norm": 0.0014493406051769853, | |
| "learning_rate": 9.231884057971015e-06, | |
| "loss": 0.0004893744364380837, | |
| "step": 1275 | |
| }, | |
| { | |
| "epoch": 0.1690067602704108, | |
| "grad_norm": 0.0021185707300901413, | |
| "learning_rate": 9.41304347826087e-06, | |
| "loss": 0.0004978740215301514, | |
| "step": 1300 | |
| }, | |
| { | |
| "epoch": 0.17225689027561103, | |
| "grad_norm": 0.002453757682815194, | |
| "learning_rate": 9.594202898550725e-06, | |
| "loss": 0.0005677514523267746, | |
| "step": 1325 | |
| }, | |
| { | |
| "epoch": 0.17550702028081122, | |
| "grad_norm": 0.0008572939550504088, | |
| "learning_rate": 9.775362318840581e-06, | |
| "loss": 0.000497000552713871, | |
| "step": 1350 | |
| }, | |
| { | |
| "epoch": 0.17875715028601144, | |
| "grad_norm": 0.0018615953158587217, | |
| "learning_rate": 9.956521739130436e-06, | |
| "loss": 0.0005274380370974541, | |
| "step": 1375 | |
| }, | |
| { | |
| "epoch": 0.18200728029121166, | |
| "grad_norm": 0.002548182150349021, | |
| "learning_rate": 1.013768115942029e-05, | |
| "loss": 0.0004912593960762024, | |
| "step": 1400 | |
| }, | |
| { | |
| "epoch": 0.18525741029641185, | |
| "grad_norm": 0.002343876985833049, | |
| "learning_rate": 1.0318840579710145e-05, | |
| "loss": 0.0005362540483474732, | |
| "step": 1425 | |
| }, | |
| { | |
| "epoch": 0.18850754030161207, | |
| "grad_norm": 0.0014049322344362736, | |
| "learning_rate": 1.05e-05, | |
| "loss": 0.00045400157570838926, | |
| "step": 1450 | |
| }, | |
| { | |
| "epoch": 0.19175767030681226, | |
| "grad_norm": 0.00521356426179409, | |
| "learning_rate": 1.0681159420289856e-05, | |
| "loss": 0.0005186598747968674, | |
| "step": 1475 | |
| }, | |
| { | |
| "epoch": 0.19500780031201248, | |
| "grad_norm": 0.0024857220705598593, | |
| "learning_rate": 1.086231884057971e-05, | |
| "loss": 0.000503481850028038, | |
| "step": 1500 | |
| }, | |
| { | |
| "epoch": 0.1982579303172127, | |
| "grad_norm": 0.0017878555227071047, | |
| "learning_rate": 1.1043478260869566e-05, | |
| "loss": 0.00044301837682724, | |
| "step": 1525 | |
| }, | |
| { | |
| "epoch": 0.2015080603224129, | |
| "grad_norm": 0.0029132412746548653, | |
| "learning_rate": 1.122463768115942e-05, | |
| "loss": 0.0005477859079837799, | |
| "step": 1550 | |
| }, | |
| { | |
| "epoch": 0.2047581903276131, | |
| "grad_norm": 0.003274207701906562, | |
| "learning_rate": 1.1405797101449275e-05, | |
| "loss": 0.0005001088976860046, | |
| "step": 1575 | |
| }, | |
| { | |
| "epoch": 0.2080083203328133, | |
| "grad_norm": 0.0017230460653081536, | |
| "learning_rate": 1.1586956521739131e-05, | |
| "loss": 0.0005191057920455932, | |
| "step": 1600 | |
| }, | |
| { | |
| "epoch": 0.21125845033801352, | |
| "grad_norm": 0.0017725358484312892, | |
| "learning_rate": 1.1768115942028986e-05, | |
| "loss": 0.0005417963117361069, | |
| "step": 1625 | |
| }, | |
| { | |
| "epoch": 0.21450858034321374, | |
| "grad_norm": 0.0017839234787970781, | |
| "learning_rate": 1.1949275362318842e-05, | |
| "loss": 0.0004936951398849487, | |
| "step": 1650 | |
| }, | |
| { | |
| "epoch": 0.21775871034841393, | |
| "grad_norm": 0.0019156673224642873, | |
| "learning_rate": 1.2130434782608697e-05, | |
| "loss": 0.00043773382902145385, | |
| "step": 1675 | |
| }, | |
| { | |
| "epoch": 0.22100884035361415, | |
| "grad_norm": 0.0025115245953202248, | |
| "learning_rate": 1.2311594202898551e-05, | |
| "loss": 0.0005085283890366554, | |
| "step": 1700 | |
| }, | |
| { | |
| "epoch": 0.22425897035881434, | |
| "grad_norm": 0.002715683775022626, | |
| "learning_rate": 1.2492753623188407e-05, | |
| "loss": 0.00047373753041028975, | |
| "step": 1725 | |
| }, | |
| { | |
| "epoch": 0.22750910036401456, | |
| "grad_norm": 0.003488801419734955, | |
| "learning_rate": 1.2673913043478262e-05, | |
| "loss": 0.0004893385618925095, | |
| "step": 1750 | |
| }, | |
| { | |
| "epoch": 0.23075923036921478, | |
| "grad_norm": 0.0053019351325929165, | |
| "learning_rate": 1.2855072463768116e-05, | |
| "loss": 0.0004813239723443985, | |
| "step": 1775 | |
| }, | |
| { | |
| "epoch": 0.23400936037441497, | |
| "grad_norm": 0.005336040630936623, | |
| "learning_rate": 1.3036231884057972e-05, | |
| "loss": 0.00048693574965000155, | |
| "step": 1800 | |
| }, | |
| { | |
| "epoch": 0.2372594903796152, | |
| "grad_norm": 0.0017795194871723652, | |
| "learning_rate": 1.3217391304347825e-05, | |
| "loss": 0.0004995526745915413, | |
| "step": 1825 | |
| }, | |
| { | |
| "epoch": 0.24050962038481538, | |
| "grad_norm": 0.0011278062593191862, | |
| "learning_rate": 1.3398550724637681e-05, | |
| "loss": 0.00046038780361413957, | |
| "step": 1850 | |
| }, | |
| { | |
| "epoch": 0.2437597503900156, | |
| "grad_norm": 0.002104923827573657, | |
| "learning_rate": 1.3579710144927538e-05, | |
| "loss": 0.00047762546688318253, | |
| "step": 1875 | |
| }, | |
| { | |
| "epoch": 0.24700988039521582, | |
| "grad_norm": 0.0019306123722344637, | |
| "learning_rate": 1.376086956521739e-05, | |
| "loss": 0.000521504357457161, | |
| "step": 1900 | |
| }, | |
| { | |
| "epoch": 0.250260010400416, | |
| "grad_norm": 0.002995668677613139, | |
| "learning_rate": 1.3942028985507247e-05, | |
| "loss": 0.000536748543381691, | |
| "step": 1925 | |
| }, | |
| { | |
| "epoch": 0.2535101404056162, | |
| "grad_norm": 0.0014928326709195971, | |
| "learning_rate": 1.4123188405797103e-05, | |
| "loss": 0.00046544995158910753, | |
| "step": 1950 | |
| }, | |
| { | |
| "epoch": 0.25676027041081645, | |
| "grad_norm": 0.0018402858404442668, | |
| "learning_rate": 1.4304347826086959e-05, | |
| "loss": 0.00046654198318719865, | |
| "step": 1975 | |
| }, | |
| { | |
| "epoch": 0.26001040041601664, | |
| "grad_norm": 0.004148806910961866, | |
| "learning_rate": 1.4485507246376812e-05, | |
| "loss": 0.0005035639926791191, | |
| "step": 2000 | |
| }, | |
| { | |
| "epoch": 0.26326053042121683, | |
| "grad_norm": 0.0056002759374678135, | |
| "learning_rate": 1.4666666666666668e-05, | |
| "loss": 0.00047159191220998766, | |
| "step": 2025 | |
| }, | |
| { | |
| "epoch": 0.2665106604264171, | |
| "grad_norm": 0.00518020149320364, | |
| "learning_rate": 1.4847826086956524e-05, | |
| "loss": 0.0005678045004606246, | |
| "step": 2050 | |
| }, | |
| { | |
| "epoch": 0.26976079043161727, | |
| "grad_norm": 0.0023816865868866444, | |
| "learning_rate": 1.5028985507246377e-05, | |
| "loss": 0.0004457936435937882, | |
| "step": 2075 | |
| }, | |
| { | |
| "epoch": 0.27301092043681746, | |
| "grad_norm": 0.00243125157430768, | |
| "learning_rate": 1.5210144927536233e-05, | |
| "loss": 0.00046467509120702745, | |
| "step": 2100 | |
| }, | |
| { | |
| "epoch": 0.2762610504420177, | |
| "grad_norm": 0.0009784717112779617, | |
| "learning_rate": 1.5391304347826088e-05, | |
| "loss": 0.0004520739614963531, | |
| "step": 2125 | |
| }, | |
| { | |
| "epoch": 0.2795111804472179, | |
| "grad_norm": 0.0021698814816772938, | |
| "learning_rate": 1.5572463768115944e-05, | |
| "loss": 0.00043580066412687304, | |
| "step": 2150 | |
| }, | |
| { | |
| "epoch": 0.2827613104524181, | |
| "grad_norm": 0.000734630273655057, | |
| "learning_rate": 1.5753623188405797e-05, | |
| "loss": 0.0004298238828778267, | |
| "step": 2175 | |
| }, | |
| { | |
| "epoch": 0.2860114404576183, | |
| "grad_norm": 0.0020142493303865194, | |
| "learning_rate": 1.5934782608695653e-05, | |
| "loss": 0.0005486938357353211, | |
| "step": 2200 | |
| }, | |
| { | |
| "epoch": 0.28926157046281853, | |
| "grad_norm": 0.0026363793294876814, | |
| "learning_rate": 1.611594202898551e-05, | |
| "loss": 0.00046887725591659545, | |
| "step": 2225 | |
| }, | |
| { | |
| "epoch": 0.2925117004680187, | |
| "grad_norm": 0.00481174373999238, | |
| "learning_rate": 1.6297101449275362e-05, | |
| "loss": 0.00042210947722196577, | |
| "step": 2250 | |
| }, | |
| { | |
| "epoch": 0.2957618304732189, | |
| "grad_norm": 0.0007976884953677654, | |
| "learning_rate": 1.6478260869565218e-05, | |
| "loss": 0.00042526666074991224, | |
| "step": 2275 | |
| }, | |
| { | |
| "epoch": 0.29901196047841916, | |
| "grad_norm": 0.0022998733911663294, | |
| "learning_rate": 1.6659420289855074e-05, | |
| "loss": 0.00047022908926010133, | |
| "step": 2300 | |
| }, | |
| { | |
| "epoch": 0.30226209048361935, | |
| "grad_norm": 0.0029255158733576536, | |
| "learning_rate": 1.6840579710144927e-05, | |
| "loss": 0.000418592244386673, | |
| "step": 2325 | |
| }, | |
| { | |
| "epoch": 0.30551222048881954, | |
| "grad_norm": 0.0019420625176280737, | |
| "learning_rate": 1.7021739130434783e-05, | |
| "loss": 0.0004478084295988083, | |
| "step": 2350 | |
| }, | |
| { | |
| "epoch": 0.3087623504940198, | |
| "grad_norm": 0.0022736522369086742, | |
| "learning_rate": 1.720289855072464e-05, | |
| "loss": 0.0004615306109189987, | |
| "step": 2375 | |
| }, | |
| { | |
| "epoch": 0.31201248049922, | |
| "grad_norm": 0.0010760062141343951, | |
| "learning_rate": 1.7384057971014496e-05, | |
| "loss": 0.0004631573334336281, | |
| "step": 2400 | |
| }, | |
| { | |
| "epoch": 0.31526261050442017, | |
| "grad_norm": 0.0044555943459272385, | |
| "learning_rate": 1.756521739130435e-05, | |
| "loss": 0.0004210951179265976, | |
| "step": 2425 | |
| }, | |
| { | |
| "epoch": 0.31851274050962036, | |
| "grad_norm": 0.003932526335120201, | |
| "learning_rate": 1.7746376811594205e-05, | |
| "loss": 0.000449698381125927, | |
| "step": 2450 | |
| }, | |
| { | |
| "epoch": 0.3217628705148206, | |
| "grad_norm": 0.002491120947524905, | |
| "learning_rate": 1.792753623188406e-05, | |
| "loss": 0.00045762836933135984, | |
| "step": 2475 | |
| }, | |
| { | |
| "epoch": 0.3250130005200208, | |
| "grad_norm": 0.0024895304813981056, | |
| "learning_rate": 1.8108695652173913e-05, | |
| "loss": 0.0004615296423435211, | |
| "step": 2500 | |
| }, | |
| { | |
| "epoch": 0.328263130525221, | |
| "grad_norm": 0.00346749578602612, | |
| "learning_rate": 1.828985507246377e-05, | |
| "loss": 0.0004106666520237923, | |
| "step": 2525 | |
| }, | |
| { | |
| "epoch": 0.33151326053042124, | |
| "grad_norm": 0.0021729774307459593, | |
| "learning_rate": 1.8471014492753626e-05, | |
| "loss": 0.00039652612060308456, | |
| "step": 2550 | |
| }, | |
| { | |
| "epoch": 0.33476339053562143, | |
| "grad_norm": 0.0037798353005200624, | |
| "learning_rate": 1.865217391304348e-05, | |
| "loss": 0.00044131282716989515, | |
| "step": 2575 | |
| }, | |
| { | |
| "epoch": 0.3380135205408216, | |
| "grad_norm": 0.005481760948896408, | |
| "learning_rate": 1.8833333333333335e-05, | |
| "loss": 0.0004480358213186264, | |
| "step": 2600 | |
| }, | |
| { | |
| "epoch": 0.34126365054602187, | |
| "grad_norm": 0.0019796863198280334, | |
| "learning_rate": 1.9014492753623188e-05, | |
| "loss": 0.0004559497907757759, | |
| "step": 2625 | |
| }, | |
| { | |
| "epoch": 0.34451378055122206, | |
| "grad_norm": 0.0025186336133629084, | |
| "learning_rate": 1.9195652173913044e-05, | |
| "loss": 0.0004142383486032486, | |
| "step": 2650 | |
| }, | |
| { | |
| "epoch": 0.34776391055642225, | |
| "grad_norm": 0.0019014193676412106, | |
| "learning_rate": 1.9376811594202897e-05, | |
| "loss": 0.0004085032269358635, | |
| "step": 2675 | |
| }, | |
| { | |
| "epoch": 0.35101404056162244, | |
| "grad_norm": 0.003675377694889903, | |
| "learning_rate": 1.9557971014492753e-05, | |
| "loss": 0.00044845666736364364, | |
| "step": 2700 | |
| }, | |
| { | |
| "epoch": 0.3542641705668227, | |
| "grad_norm": 0.0008569825440645218, | |
| "learning_rate": 1.973913043478261e-05, | |
| "loss": 0.0004178711771965027, | |
| "step": 2725 | |
| }, | |
| { | |
| "epoch": 0.3575143005720229, | |
| "grad_norm": 0.0027375146746635437, | |
| "learning_rate": 1.9920289855072465e-05, | |
| "loss": 0.000369691476225853, | |
| "step": 2750 | |
| }, | |
| { | |
| "epoch": 0.36076443057722307, | |
| "grad_norm": 0.0021465872414410114, | |
| "learning_rate": 2.0101449275362318e-05, | |
| "loss": 0.00041070599108934404, | |
| "step": 2775 | |
| }, | |
| { | |
| "epoch": 0.3640145605824233, | |
| "grad_norm": 0.0017981152050197124, | |
| "learning_rate": 2.0282608695652174e-05, | |
| "loss": 0.00043407246470451356, | |
| "step": 2800 | |
| }, | |
| { | |
| "epoch": 0.3672646905876235, | |
| "grad_norm": 0.0034601977095007896, | |
| "learning_rate": 2.046376811594203e-05, | |
| "loss": 0.00042522311210632327, | |
| "step": 2825 | |
| }, | |
| { | |
| "epoch": 0.3705148205928237, | |
| "grad_norm": 0.0035561842378228903, | |
| "learning_rate": 2.0644927536231883e-05, | |
| "loss": 0.0003913164883852005, | |
| "step": 2850 | |
| }, | |
| { | |
| "epoch": 0.37376495059802395, | |
| "grad_norm": 0.005427898373454809, | |
| "learning_rate": 2.082608695652174e-05, | |
| "loss": 0.00041633348912000655, | |
| "step": 2875 | |
| }, | |
| { | |
| "epoch": 0.37701508060322414, | |
| "grad_norm": 0.001540972269140184, | |
| "learning_rate": 2.1007246376811596e-05, | |
| "loss": 0.000437643900513649, | |
| "step": 2900 | |
| }, | |
| { | |
| "epoch": 0.38026521060842433, | |
| "grad_norm": 0.008584934286773205, | |
| "learning_rate": 2.1188405797101452e-05, | |
| "loss": 0.0004209858551621437, | |
| "step": 2925 | |
| }, | |
| { | |
| "epoch": 0.3835153406136245, | |
| "grad_norm": 0.003871705150231719, | |
| "learning_rate": 2.1369565217391305e-05, | |
| "loss": 0.0004569174721837044, | |
| "step": 2950 | |
| }, | |
| { | |
| "epoch": 0.38676547061882477, | |
| "grad_norm": 0.003933009225875139, | |
| "learning_rate": 2.155072463768116e-05, | |
| "loss": 0.00039284847676754, | |
| "step": 2975 | |
| }, | |
| { | |
| "epoch": 0.39001560062402496, | |
| "grad_norm": 0.0023773452267050743, | |
| "learning_rate": 2.1731884057971017e-05, | |
| "loss": 0.0003957047685980797, | |
| "step": 3000 | |
| }, | |
| { | |
| "epoch": 0.39326573062922515, | |
| "grad_norm": 0.0013754197861999273, | |
| "learning_rate": 2.191304347826087e-05, | |
| "loss": 0.00040918096899986264, | |
| "step": 3025 | |
| }, | |
| { | |
| "epoch": 0.3965158606344254, | |
| "grad_norm": 0.0013438657624647021, | |
| "learning_rate": 2.2094202898550726e-05, | |
| "loss": 0.00042451847344636916, | |
| "step": 3050 | |
| }, | |
| { | |
| "epoch": 0.3997659906396256, | |
| "grad_norm": 0.0018076892010867596, | |
| "learning_rate": 2.2275362318840582e-05, | |
| "loss": 0.000429234579205513, | |
| "step": 3075 | |
| }, | |
| { | |
| "epoch": 0.4030161206448258, | |
| "grad_norm": 0.002512214705348015, | |
| "learning_rate": 2.2456521739130438e-05, | |
| "loss": 0.00044652506709098817, | |
| "step": 3100 | |
| }, | |
| { | |
| "epoch": 0.40626625065002603, | |
| "grad_norm": 0.002839904511347413, | |
| "learning_rate": 2.263768115942029e-05, | |
| "loss": 0.00040985174477100374, | |
| "step": 3125 | |
| }, | |
| { | |
| "epoch": 0.4095163806552262, | |
| "grad_norm": 0.0019431107211858034, | |
| "learning_rate": 2.2818840579710147e-05, | |
| "loss": 0.0003903954103589058, | |
| "step": 3150 | |
| }, | |
| { | |
| "epoch": 0.4127665106604264, | |
| "grad_norm": 0.0020985049195587635, | |
| "learning_rate": 2.3000000000000003e-05, | |
| "loss": 0.00038104344159364703, | |
| "step": 3175 | |
| }, | |
| { | |
| "epoch": 0.4160166406656266, | |
| "grad_norm": 0.0028055645525455475, | |
| "learning_rate": 2.3181159420289856e-05, | |
| "loss": 0.0003827373683452606, | |
| "step": 3200 | |
| }, | |
| { | |
| "epoch": 0.41926677067082685, | |
| "grad_norm": 0.0014860191149637103, | |
| "learning_rate": 2.3362318840579712e-05, | |
| "loss": 0.00043064661324024203, | |
| "step": 3225 | |
| }, | |
| { | |
| "epoch": 0.42251690067602704, | |
| "grad_norm": 0.0033869794569909573, | |
| "learning_rate": 2.354347826086957e-05, | |
| "loss": 0.00044150054454803465, | |
| "step": 3250 | |
| }, | |
| { | |
| "epoch": 0.42576703068122723, | |
| "grad_norm": 0.001350346370600164, | |
| "learning_rate": 2.372463768115942e-05, | |
| "loss": 0.00036799222230911256, | |
| "step": 3275 | |
| }, | |
| { | |
| "epoch": 0.4290171606864275, | |
| "grad_norm": 0.0018407037714496255, | |
| "learning_rate": 2.3905797101449278e-05, | |
| "loss": 0.00041019417345523836, | |
| "step": 3300 | |
| }, | |
| { | |
| "epoch": 0.43226729069162767, | |
| "grad_norm": 0.0030828621238470078, | |
| "learning_rate": 2.408695652173913e-05, | |
| "loss": 0.00040250256657600405, | |
| "step": 3325 | |
| }, | |
| { | |
| "epoch": 0.43551742069682786, | |
| "grad_norm": 0.005468167830258608, | |
| "learning_rate": 2.4268115942028987e-05, | |
| "loss": 0.0003616174682974815, | |
| "step": 3350 | |
| }, | |
| { | |
| "epoch": 0.4387675507020281, | |
| "grad_norm": 0.001769745722413063, | |
| "learning_rate": 2.444927536231884e-05, | |
| "loss": 0.0003935152292251587, | |
| "step": 3375 | |
| }, | |
| { | |
| "epoch": 0.4420176807072283, | |
| "grad_norm": 0.004243707749992609, | |
| "learning_rate": 2.4630434782608696e-05, | |
| "loss": 0.0003487700968980789, | |
| "step": 3400 | |
| }, | |
| { | |
| "epoch": 0.4452678107124285, | |
| "grad_norm": 0.001457527163438499, | |
| "learning_rate": 2.4811594202898552e-05, | |
| "loss": 0.00038313735276460645, | |
| "step": 3425 | |
| }, | |
| { | |
| "epoch": 0.4485179407176287, | |
| "grad_norm": 0.004237140994518995, | |
| "learning_rate": 2.4992753623188405e-05, | |
| "loss": 0.00040549322962760923, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 0.4485179407176287, | |
| "eval_loss": 0.0014288317179307342, | |
| "eval_runtime": 9.4821, | |
| "eval_samples_per_second": 527.308, | |
| "eval_steps_per_second": 4.218, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 0.4485179407176287, | |
| "eval_holdout_avg_strict_f1": 0.016948437606658096, | |
| "eval_holdout_enron_spam_bvk_strict_f1": 0.0, | |
| "eval_holdout_enron_spam_setfit_strict_f1": 0.0, | |
| "eval_holdout_gliner2_pii_ptbr_reward_split_strict_f1": 0.04347826086956522, | |
| "eval_holdout_nemotron_pii_strict_f1": 0.028890015205271163, | |
| "eval_holdout_open_pii_masking_500k_strict_f1": 0.07240773286467486, | |
| "eval_holdout_phishing_darkknight_strict_f1": 0.0, | |
| "eval_holdout_phishing_zefang_strict_f1": 0.0, | |
| "eval_holdout_pii_masking_400k_strict_f1": 0.04165680473372781, | |
| "eval_holdout_sms_spam_multilingual_strict_f1": 0.0, | |
| "eval_holdout_spam_messages_mshenoda_strict_f1": 0.0, | |
| "eval_holdout_spamassassin_strict_f1": 0.0, | |
| "step": 3450 | |
| }, | |
| { | |
| "epoch": 0.45176807072282893, | |
| "grad_norm": 0.0014306495431810617, | |
| "learning_rate": 2.5173913043478264e-05, | |
| "loss": 0.0003895667567849159, | |
| "step": 3475 | |
| }, | |
| { | |
| "epoch": 0.4550182007280291, | |
| "grad_norm": 0.0033567098435014486, | |
| "learning_rate": 2.5355072463768113e-05, | |
| "loss": 0.0003654065728187561, | |
| "step": 3500 | |
| }, | |
| { | |
| "epoch": 0.4582683307332293, | |
| "grad_norm": 0.0033907731994986534, | |
| "learning_rate": 2.553623188405797e-05, | |
| "loss": 0.0004192650318145752, | |
| "step": 3525 | |
| }, | |
| { | |
| "epoch": 0.46151846073842956, | |
| "grad_norm": 0.002767309546470642, | |
| "learning_rate": 2.5717391304347826e-05, | |
| "loss": 0.00045176178216934206, | |
| "step": 3550 | |
| }, | |
| { | |
| "epoch": 0.46476859074362975, | |
| "grad_norm": 0.0033274965826421976, | |
| "learning_rate": 2.5898550724637682e-05, | |
| "loss": 0.00036267347633838653, | |
| "step": 3575 | |
| }, | |
| { | |
| "epoch": 0.46801872074882994, | |
| "grad_norm": 0.0012526543578132987, | |
| "learning_rate": 2.6079710144927538e-05, | |
| "loss": 0.0003706477209925652, | |
| "step": 3600 | |
| }, | |
| { | |
| "epoch": 0.4712688507540302, | |
| "grad_norm": 0.0025104728993028402, | |
| "learning_rate": 2.6260869565217394e-05, | |
| "loss": 0.0003582199662923813, | |
| "step": 3625 | |
| }, | |
| { | |
| "epoch": 0.4745189807592304, | |
| "grad_norm": 0.001733158016577363, | |
| "learning_rate": 2.644202898550725e-05, | |
| "loss": 0.00037459593266248704, | |
| "step": 3650 | |
| }, | |
| { | |
| "epoch": 0.47776911076443057, | |
| "grad_norm": 0.014838664792478085, | |
| "learning_rate": 2.66231884057971e-05, | |
| "loss": 0.000381452776491642, | |
| "step": 3675 | |
| }, | |
| { | |
| "epoch": 0.48101924076963076, | |
| "grad_norm": 0.0015709763392806053, | |
| "learning_rate": 2.6804347826086956e-05, | |
| "loss": 0.000379287488758564, | |
| "step": 3700 | |
| }, | |
| { | |
| "epoch": 0.484269370774831, | |
| "grad_norm": 0.0009885560721158981, | |
| "learning_rate": 2.6985507246376812e-05, | |
| "loss": 0.0003702220320701599, | |
| "step": 3725 | |
| }, | |
| { | |
| "epoch": 0.4875195007800312, | |
| "grad_norm": 0.003496927907690406, | |
| "learning_rate": 2.716666666666667e-05, | |
| "loss": 0.00037388518452644346, | |
| "step": 3750 | |
| }, | |
| { | |
| "epoch": 0.4907696307852314, | |
| "grad_norm": 0.0013370664091780782, | |
| "learning_rate": 2.7347826086956525e-05, | |
| "loss": 0.00038881190121173856, | |
| "step": 3775 | |
| }, | |
| { | |
| "epoch": 0.49401976079043164, | |
| "grad_norm": 0.0019693628419190645, | |
| "learning_rate": 2.752898550724638e-05, | |
| "loss": 0.00037656128406524656, | |
| "step": 3800 | |
| }, | |
| { | |
| "epoch": 0.49726989079563183, | |
| "grad_norm": 0.001666913041844964, | |
| "learning_rate": 2.7710144927536237e-05, | |
| "loss": 0.0004192305356264114, | |
| "step": 3825 | |
| }, | |
| { | |
| "epoch": 0.500520020800832, | |
| "grad_norm": 0.004715986084192991, | |
| "learning_rate": 2.7891304347826087e-05, | |
| "loss": 0.0003989679366350174, | |
| "step": 3850 | |
| }, | |
| { | |
| "epoch": 0.5037701508060323, | |
| "grad_norm": 0.004425929393619299, | |
| "learning_rate": 2.8072463768115943e-05, | |
| "loss": 0.0003500617295503616, | |
| "step": 3875 | |
| }, | |
| { | |
| "epoch": 0.5070202808112324, | |
| "grad_norm": 0.002427851315587759, | |
| "learning_rate": 2.82536231884058e-05, | |
| "loss": 0.0003932202234864235, | |
| "step": 3900 | |
| }, | |
| { | |
| "epoch": 0.5102704108164327, | |
| "grad_norm": 0.0017338527832180262, | |
| "learning_rate": 2.8434782608695655e-05, | |
| "loss": 0.0003991245478391647, | |
| "step": 3925 | |
| }, | |
| { | |
| "epoch": 0.5135205408216329, | |
| "grad_norm": 0.0019661737605929375, | |
| "learning_rate": 2.8615942028985508e-05, | |
| "loss": 0.0004087170585989952, | |
| "step": 3950 | |
| }, | |
| { | |
| "epoch": 0.516770670826833, | |
| "grad_norm": 0.004153737332671881, | |
| "learning_rate": 2.8797101449275364e-05, | |
| "loss": 0.00037440124899148944, | |
| "step": 3975 | |
| }, | |
| { | |
| "epoch": 0.5200208008320333, | |
| "grad_norm": 0.0021952821407467127, | |
| "learning_rate": 2.897826086956522e-05, | |
| "loss": 0.00032810430973768235, | |
| "step": 4000 | |
| }, | |
| { | |
| "epoch": 0.5232709308372335, | |
| "grad_norm": 0.00245103077031672, | |
| "learning_rate": 2.9159420289855073e-05, | |
| "loss": 0.00035349562764167784, | |
| "step": 4025 | |
| }, | |
| { | |
| "epoch": 0.5265210608424337, | |
| "grad_norm": 0.0019252462079748511, | |
| "learning_rate": 2.934057971014493e-05, | |
| "loss": 0.0003406635299324989, | |
| "step": 4050 | |
| }, | |
| { | |
| "epoch": 0.5297711908476339, | |
| "grad_norm": 0.0033155896235257387, | |
| "learning_rate": 2.9521739130434782e-05, | |
| "loss": 0.0003747899085283279, | |
| "step": 4075 | |
| }, | |
| { | |
| "epoch": 0.5330213208528342, | |
| "grad_norm": 0.0012676240876317024, | |
| "learning_rate": 2.9702898550724638e-05, | |
| "loss": 0.00034757569432258606, | |
| "step": 4100 | |
| }, | |
| { | |
| "epoch": 0.5362714508580343, | |
| "grad_norm": 0.0037625425029546022, | |
| "learning_rate": 2.9884057971014494e-05, | |
| "loss": 0.0003790777549147606, | |
| "step": 4125 | |
| }, | |
| { | |
| "epoch": 0.5395215808632345, | |
| "grad_norm": 0.005746638402342796, | |
| "learning_rate": 3.006521739130435e-05, | |
| "loss": 0.000355856865644455, | |
| "step": 4150 | |
| }, | |
| { | |
| "epoch": 0.5427717108684348, | |
| "grad_norm": 0.002143938560038805, | |
| "learning_rate": 3.0246376811594207e-05, | |
| "loss": 0.0004532543942332268, | |
| "step": 4175 | |
| }, | |
| { | |
| "epoch": 0.5460218408736349, | |
| "grad_norm": 0.0009120068280026317, | |
| "learning_rate": 3.0427536231884056e-05, | |
| "loss": 0.0003795912489295006, | |
| "step": 4200 | |
| }, | |
| { | |
| "epoch": 0.5492719708788352, | |
| "grad_norm": 0.002007429488003254, | |
| "learning_rate": 3.0608695652173916e-05, | |
| "loss": 0.0003673683106899261, | |
| "step": 4225 | |
| }, | |
| { | |
| "epoch": 0.5525221008840354, | |
| "grad_norm": 0.002023761859163642, | |
| "learning_rate": 3.078985507246377e-05, | |
| "loss": 0.00035739000886678694, | |
| "step": 4250 | |
| }, | |
| { | |
| "epoch": 0.5557722308892356, | |
| "grad_norm": 0.0014700735919177532, | |
| "learning_rate": 3.097101449275363e-05, | |
| "loss": 0.0003841959312558174, | |
| "step": 4275 | |
| }, | |
| { | |
| "epoch": 0.5590223608944358, | |
| "grad_norm": 0.01717543788254261, | |
| "learning_rate": 3.115217391304348e-05, | |
| "loss": 0.0003442119434475899, | |
| "step": 4300 | |
| }, | |
| { | |
| "epoch": 0.5622724908996359, | |
| "grad_norm": 0.0022744322195649147, | |
| "learning_rate": 3.1333333333333334e-05, | |
| "loss": 0.0004083585739135742, | |
| "step": 4325 | |
| }, | |
| { | |
| "epoch": 0.5655226209048362, | |
| "grad_norm": 0.0019075083546340466, | |
| "learning_rate": 3.151449275362319e-05, | |
| "loss": 0.00033581096678972247, | |
| "step": 4350 | |
| }, | |
| { | |
| "epoch": 0.5687727509100364, | |
| "grad_norm": 0.003801671089604497, | |
| "learning_rate": 3.1695652173913046e-05, | |
| "loss": 0.00036067895591259004, | |
| "step": 4375 | |
| }, | |
| { | |
| "epoch": 0.5720228809152366, | |
| "grad_norm": 0.0011975554516538978, | |
| "learning_rate": 3.18768115942029e-05, | |
| "loss": 0.0003966550901532173, | |
| "step": 4400 | |
| }, | |
| { | |
| "epoch": 0.5752730109204368, | |
| "grad_norm": 0.005756833590567112, | |
| "learning_rate": 3.205797101449275e-05, | |
| "loss": 0.0003798803687095642, | |
| "step": 4425 | |
| }, | |
| { | |
| "epoch": 0.5785231409256371, | |
| "grad_norm": 0.002057137433439493, | |
| "learning_rate": 3.223913043478261e-05, | |
| "loss": 0.00037363722920417784, | |
| "step": 4450 | |
| }, | |
| { | |
| "epoch": 0.5817732709308372, | |
| "grad_norm": 0.0018292538588866591, | |
| "learning_rate": 3.2420289855072464e-05, | |
| "loss": 0.0003827802836894989, | |
| "step": 4475 | |
| }, | |
| { | |
| "epoch": 0.5850234009360374, | |
| "grad_norm": 0.0031336594838649035, | |
| "learning_rate": 3.2601449275362324e-05, | |
| "loss": 0.00031954482197761537, | |
| "step": 4500 | |
| }, | |
| { | |
| "epoch": 0.5882735309412377, | |
| "grad_norm": 0.002637260127812624, | |
| "learning_rate": 3.278260869565217e-05, | |
| "loss": 0.0003741474822163582, | |
| "step": 4525 | |
| }, | |
| { | |
| "epoch": 0.5915236609464378, | |
| "grad_norm": 0.0017657590797170997, | |
| "learning_rate": 3.296376811594203e-05, | |
| "loss": 0.0003575438633561134, | |
| "step": 4550 | |
| }, | |
| { | |
| "epoch": 0.5947737909516381, | |
| "grad_norm": 0.00354047492146492, | |
| "learning_rate": 3.314492753623188e-05, | |
| "loss": 0.0003497960790991783, | |
| "step": 4575 | |
| }, | |
| { | |
| "epoch": 0.5980239209568383, | |
| "grad_norm": 0.0017961938865482807, | |
| "learning_rate": 3.332608695652174e-05, | |
| "loss": 0.00039899878203868865, | |
| "step": 4600 | |
| }, | |
| { | |
| "epoch": 0.6012740509620385, | |
| "grad_norm": 0.0013773336540907621, | |
| "learning_rate": 3.3507246376811594e-05, | |
| "loss": 0.0003468652069568634, | |
| "step": 4625 | |
| }, | |
| { | |
| "epoch": 0.6045241809672387, | |
| "grad_norm": 0.0015784098068252206, | |
| "learning_rate": 3.3688405797101454e-05, | |
| "loss": 0.00034876059740781786, | |
| "step": 4650 | |
| }, | |
| { | |
| "epoch": 0.607774310972439, | |
| "grad_norm": 0.003309419145807624, | |
| "learning_rate": 3.386956521739131e-05, | |
| "loss": 0.000346483550965786, | |
| "step": 4675 | |
| }, | |
| { | |
| "epoch": 0.6110244409776391, | |
| "grad_norm": 0.0019642598927021027, | |
| "learning_rate": 3.405072463768116e-05, | |
| "loss": 0.0003129390254616737, | |
| "step": 4700 | |
| }, | |
| { | |
| "epoch": 0.6142745709828393, | |
| "grad_norm": 0.0018581639742478728, | |
| "learning_rate": 3.423188405797101e-05, | |
| "loss": 0.00032798755913972857, | |
| "step": 4725 | |
| }, | |
| { | |
| "epoch": 0.6175247009880396, | |
| "grad_norm": 0.0029012670274823904, | |
| "learning_rate": 3.441304347826087e-05, | |
| "loss": 0.00033405110239982606, | |
| "step": 4750 | |
| }, | |
| { | |
| "epoch": 0.6207748309932397, | |
| "grad_norm": 0.002787525998428464, | |
| "learning_rate": 3.4594202898550725e-05, | |
| "loss": 0.0003223710134625435, | |
| "step": 4775 | |
| }, | |
| { | |
| "epoch": 0.62402496099844, | |
| "grad_norm": 0.0017276693833991885, | |
| "learning_rate": 3.4775362318840584e-05, | |
| "loss": 0.0003574707731604576, | |
| "step": 4800 | |
| }, | |
| { | |
| "epoch": 0.6272750910036401, | |
| "grad_norm": 0.0032260166481137276, | |
| "learning_rate": 3.495652173913044e-05, | |
| "loss": 0.00036472782492637636, | |
| "step": 4825 | |
| }, | |
| { | |
| "epoch": 0.6305252210088403, | |
| "grad_norm": 0.002348755020648241, | |
| "learning_rate": 3.51376811594203e-05, | |
| "loss": 0.0002934296987950802, | |
| "step": 4850 | |
| }, | |
| { | |
| "epoch": 0.6337753510140406, | |
| "grad_norm": 0.0017877062782645226, | |
| "learning_rate": 3.531884057971014e-05, | |
| "loss": 0.00033006574958562853, | |
| "step": 4875 | |
| }, | |
| { | |
| "epoch": 0.6370254810192407, | |
| "grad_norm": 0.0011508421739563346, | |
| "learning_rate": 3.55e-05, | |
| "loss": 0.00033879075199365615, | |
| "step": 4900 | |
| }, | |
| { | |
| "epoch": 0.640275611024441, | |
| "grad_norm": 0.0021689175628125668, | |
| "learning_rate": 3.5681159420289855e-05, | |
| "loss": 0.00034071825444698334, | |
| "step": 4925 | |
| }, | |
| { | |
| "epoch": 0.6435257410296412, | |
| "grad_norm": 0.001443523564375937, | |
| "learning_rate": 3.5862318840579715e-05, | |
| "loss": 0.0003129898384213448, | |
| "step": 4950 | |
| }, | |
| { | |
| "epoch": 0.6467758710348414, | |
| "grad_norm": 0.001089542405679822, | |
| "learning_rate": 3.604347826086957e-05, | |
| "loss": 0.00034003332257270813, | |
| "step": 4975 | |
| }, | |
| { | |
| "epoch": 0.6500260010400416, | |
| "grad_norm": 0.004490668885409832, | |
| "learning_rate": 3.622463768115942e-05, | |
| "loss": 0.00037308726459741595, | |
| "step": 5000 | |
| }, | |
| { | |
| "epoch": 0.6532761310452418, | |
| "grad_norm": 0.0019078904297202826, | |
| "learning_rate": 3.640579710144928e-05, | |
| "loss": 0.00040014833211898804, | |
| "step": 5025 | |
| }, | |
| { | |
| "epoch": 0.656526261050442, | |
| "grad_norm": 0.005327039398252964, | |
| "learning_rate": 3.658695652173913e-05, | |
| "loss": 0.00038816750049591064, | |
| "step": 5050 | |
| }, | |
| { | |
| "epoch": 0.6597763910556422, | |
| "grad_norm": 0.0015874936943873763, | |
| "learning_rate": 3.6768115942028985e-05, | |
| "loss": 0.0003395514935255051, | |
| "step": 5075 | |
| }, | |
| { | |
| "epoch": 0.6630265210608425, | |
| "grad_norm": 0.001610973384231329, | |
| "learning_rate": 3.694927536231884e-05, | |
| "loss": 0.0003646326065063477, | |
| "step": 5100 | |
| }, | |
| { | |
| "epoch": 0.6662766510660426, | |
| "grad_norm": 0.0019586284179240465, | |
| "learning_rate": 3.71304347826087e-05, | |
| "loss": 0.00032726649194955823, | |
| "step": 5125 | |
| }, | |
| { | |
| "epoch": 0.6695267810712429, | |
| "grad_norm": 0.0011525078443810344, | |
| "learning_rate": 3.731159420289855e-05, | |
| "loss": 0.0003709142655134201, | |
| "step": 5150 | |
| }, | |
| { | |
| "epoch": 0.6727769110764431, | |
| "grad_norm": 0.0020568196196109056, | |
| "learning_rate": 3.749275362318841e-05, | |
| "loss": 0.0004013935476541519, | |
| "step": 5175 | |
| }, | |
| { | |
| "epoch": 0.6760270410816432, | |
| "grad_norm": 0.0034161447547376156, | |
| "learning_rate": 3.767391304347826e-05, | |
| "loss": 0.00032214373350143434, | |
| "step": 5200 | |
| }, | |
| { | |
| "epoch": 0.6792771710868435, | |
| "grad_norm": 0.0010419136378914118, | |
| "learning_rate": 3.7855072463768116e-05, | |
| "loss": 0.000292956680059433, | |
| "step": 5225 | |
| }, | |
| { | |
| "epoch": 0.6825273010920437, | |
| "grad_norm": 0.0020100681576877832, | |
| "learning_rate": 3.803623188405797e-05, | |
| "loss": 0.00033670149743556975, | |
| "step": 5250 | |
| }, | |
| { | |
| "epoch": 0.6857774310972439, | |
| "grad_norm": 0.004615044221282005, | |
| "learning_rate": 3.821739130434783e-05, | |
| "loss": 0.00035760205239057543, | |
| "step": 5275 | |
| }, | |
| { | |
| "epoch": 0.6890275611024441, | |
| "grad_norm": 0.0007368104998022318, | |
| "learning_rate": 3.839855072463768e-05, | |
| "loss": 0.00034552190452814103, | |
| "step": 5300 | |
| }, | |
| { | |
| "epoch": 0.6922776911076443, | |
| "grad_norm": 0.001840224489569664, | |
| "learning_rate": 3.857971014492754e-05, | |
| "loss": 0.0003251539543271065, | |
| "step": 5325 | |
| }, | |
| { | |
| "epoch": 0.6955278211128445, | |
| "grad_norm": 0.0034191531594842672, | |
| "learning_rate": 3.876086956521739e-05, | |
| "loss": 0.0003743323311209679, | |
| "step": 5350 | |
| }, | |
| { | |
| "epoch": 0.6987779511180447, | |
| "grad_norm": 0.002969014458358288, | |
| "learning_rate": 3.894202898550725e-05, | |
| "loss": 0.0003549868240952492, | |
| "step": 5375 | |
| }, | |
| { | |
| "epoch": 0.7020280811232449, | |
| "grad_norm": 0.002682388760149479, | |
| "learning_rate": 3.91231884057971e-05, | |
| "loss": 0.0003697166219353676, | |
| "step": 5400 | |
| }, | |
| { | |
| "epoch": 0.7052782111284451, | |
| "grad_norm": 0.0019561341032385826, | |
| "learning_rate": 3.930434782608696e-05, | |
| "loss": 0.00029268305748701094, | |
| "step": 5425 | |
| }, | |
| { | |
| "epoch": 0.7085283411336454, | |
| "grad_norm": 0.0015227431431412697, | |
| "learning_rate": 3.948550724637681e-05, | |
| "loss": 0.00030088482424616813, | |
| "step": 5450 | |
| }, | |
| { | |
| "epoch": 0.7117784711388455, | |
| "grad_norm": 0.0008432468748651445, | |
| "learning_rate": 3.966666666666667e-05, | |
| "loss": 0.0003448799252510071, | |
| "step": 5475 | |
| }, | |
| { | |
| "epoch": 0.7150286011440458, | |
| "grad_norm": 0.0027734299656003714, | |
| "learning_rate": 3.9847826086956524e-05, | |
| "loss": 0.0003730638697743416, | |
| "step": 5500 | |
| }, | |
| { | |
| "epoch": 0.718278731149246, | |
| "grad_norm": 0.002775531494989991, | |
| "learning_rate": 4.002898550724638e-05, | |
| "loss": 0.0002861537970602512, | |
| "step": 5525 | |
| }, | |
| { | |
| "epoch": 0.7215288611544461, | |
| "grad_norm": 0.0013877362944185734, | |
| "learning_rate": 4.0210144927536236e-05, | |
| "loss": 0.0003062456473708153, | |
| "step": 5550 | |
| }, | |
| { | |
| "epoch": 0.7247789911596464, | |
| "grad_norm": 0.0016100159846246243, | |
| "learning_rate": 4.039130434782609e-05, | |
| "loss": 0.00034022834151983263, | |
| "step": 5575 | |
| }, | |
| { | |
| "epoch": 0.7280291211648466, | |
| "grad_norm": 0.0023292386904358864, | |
| "learning_rate": 4.057246376811594e-05, | |
| "loss": 0.00032042227685451506, | |
| "step": 5600 | |
| }, | |
| { | |
| "epoch": 0.7312792511700468, | |
| "grad_norm": 0.005723954644054174, | |
| "learning_rate": 4.07536231884058e-05, | |
| "loss": 0.00036787450313568114, | |
| "step": 5625 | |
| }, | |
| { | |
| "epoch": 0.734529381175247, | |
| "grad_norm": 0.007385662756860256, | |
| "learning_rate": 4.0934782608695654e-05, | |
| "loss": 0.00030214842408895495, | |
| "step": 5650 | |
| }, | |
| { | |
| "epoch": 0.7377795111804473, | |
| "grad_norm": 0.0016194393392652273, | |
| "learning_rate": 4.111594202898551e-05, | |
| "loss": 0.0003026936203241348, | |
| "step": 5675 | |
| }, | |
| { | |
| "epoch": 0.7410296411856474, | |
| "grad_norm": 0.0011678567389026284, | |
| "learning_rate": 4.1297101449275366e-05, | |
| "loss": 0.0003290276974439621, | |
| "step": 5700 | |
| }, | |
| { | |
| "epoch": 0.7442797711908476, | |
| "grad_norm": 0.0015876541147008538, | |
| "learning_rate": 4.147826086956522e-05, | |
| "loss": 0.0002660868316888809, | |
| "step": 5725 | |
| }, | |
| { | |
| "epoch": 0.7475299011960479, | |
| "grad_norm": 0.002573238452896476, | |
| "learning_rate": 4.165942028985507e-05, | |
| "loss": 0.0003120236098766327, | |
| "step": 5750 | |
| }, | |
| { | |
| "epoch": 0.750780031201248, | |
| "grad_norm": 0.001261972007341683, | |
| "learning_rate": 4.184057971014493e-05, | |
| "loss": 0.000330447293817997, | |
| "step": 5775 | |
| }, | |
| { | |
| "epoch": 0.7540301612064483, | |
| "grad_norm": 0.002024536021053791, | |
| "learning_rate": 4.2021739130434784e-05, | |
| "loss": 0.0003500768169760704, | |
| "step": 5800 | |
| }, | |
| { | |
| "epoch": 0.7572802912116484, | |
| "grad_norm": 0.0013312718365341425, | |
| "learning_rate": 4.220289855072464e-05, | |
| "loss": 0.00034043505787849425, | |
| "step": 5825 | |
| }, | |
| { | |
| "epoch": 0.7605304212168487, | |
| "grad_norm": 0.0014061352703720331, | |
| "learning_rate": 4.23840579710145e-05, | |
| "loss": 0.0003583569824695587, | |
| "step": 5850 | |
| }, | |
| { | |
| "epoch": 0.7637805512220489, | |
| "grad_norm": 0.0017677943687886, | |
| "learning_rate": 4.256521739130435e-05, | |
| "loss": 0.0003040175698697567, | |
| "step": 5875 | |
| }, | |
| { | |
| "epoch": 0.767030681227249, | |
| "grad_norm": 0.002769756829366088, | |
| "learning_rate": 4.274637681159421e-05, | |
| "loss": 0.00031752903014421464, | |
| "step": 5900 | |
| }, | |
| { | |
| "epoch": 0.7702808112324493, | |
| "grad_norm": 0.0013718277914449573, | |
| "learning_rate": 4.2927536231884055e-05, | |
| "loss": 0.0003393003344535828, | |
| "step": 5925 | |
| }, | |
| { | |
| "epoch": 0.7735309412376495, | |
| "grad_norm": 0.0019314177334308624, | |
| "learning_rate": 4.3108695652173915e-05, | |
| "loss": 0.00027401041239500046, | |
| "step": 5950 | |
| }, | |
| { | |
| "epoch": 0.7767810712428497, | |
| "grad_norm": 0.007918702438473701, | |
| "learning_rate": 4.328985507246377e-05, | |
| "loss": 0.0002910470962524414, | |
| "step": 5975 | |
| }, | |
| { | |
| "epoch": 0.7800312012480499, | |
| "grad_norm": 0.0017927872249856591, | |
| "learning_rate": 4.347101449275363e-05, | |
| "loss": 0.0003286924958229065, | |
| "step": 6000 | |
| }, | |
| { | |
| "epoch": 0.7832813312532502, | |
| "grad_norm": 0.001361369388177991, | |
| "learning_rate": 4.365217391304348e-05, | |
| "loss": 0.00030220640823245046, | |
| "step": 6025 | |
| }, | |
| { | |
| "epoch": 0.7865314612584503, | |
| "grad_norm": 0.0010171420872211456, | |
| "learning_rate": 4.383333333333334e-05, | |
| "loss": 0.0002842904068529606, | |
| "step": 6050 | |
| }, | |
| { | |
| "epoch": 0.7897815912636506, | |
| "grad_norm": 0.0032635347452014685, | |
| "learning_rate": 4.401449275362319e-05, | |
| "loss": 0.00035091206431388857, | |
| "step": 6075 | |
| }, | |
| { | |
| "epoch": 0.7930317212688508, | |
| "grad_norm": 0.0009199533378705382, | |
| "learning_rate": 4.4195652173913045e-05, | |
| "loss": 0.0002981584332883358, | |
| "step": 6100 | |
| }, | |
| { | |
| "epoch": 0.7962818512740509, | |
| "grad_norm": 0.0011878847144544125, | |
| "learning_rate": 4.43768115942029e-05, | |
| "loss": 0.0002847062610089779, | |
| "step": 6125 | |
| }, | |
| { | |
| "epoch": 0.7995319812792512, | |
| "grad_norm": 0.0005240865866653621, | |
| "learning_rate": 4.455797101449276e-05, | |
| "loss": 0.0002426844649016857, | |
| "step": 6150 | |
| }, | |
| { | |
| "epoch": 0.8027821112844514, | |
| "grad_norm": 0.004690847359597683, | |
| "learning_rate": 4.473913043478261e-05, | |
| "loss": 0.0003175145760178566, | |
| "step": 6175 | |
| }, | |
| { | |
| "epoch": 0.8060322412896516, | |
| "grad_norm": 0.002477250760421157, | |
| "learning_rate": 4.492028985507247e-05, | |
| "loss": 0.0002982656471431255, | |
| "step": 6200 | |
| }, | |
| { | |
| "epoch": 0.8092823712948518, | |
| "grad_norm": 0.0012452919036149979, | |
| "learning_rate": 4.510144927536232e-05, | |
| "loss": 0.00032246861606836317, | |
| "step": 6225 | |
| }, | |
| { | |
| "epoch": 0.8125325013000521, | |
| "grad_norm": 0.0023345041554421186, | |
| "learning_rate": 4.5282608695652175e-05, | |
| "loss": 0.0002904561534523964, | |
| "step": 6250 | |
| }, | |
| { | |
| "epoch": 0.8157826313052522, | |
| "grad_norm": 0.0006569608813151717, | |
| "learning_rate": 4.546376811594203e-05, | |
| "loss": 0.000323328822851181, | |
| "step": 6275 | |
| }, | |
| { | |
| "epoch": 0.8190327613104524, | |
| "grad_norm": 0.022973662242293358, | |
| "learning_rate": 4.564492753623189e-05, | |
| "loss": 0.00027932059019804, | |
| "step": 6300 | |
| }, | |
| { | |
| "epoch": 0.8222828913156526, | |
| "grad_norm": 0.0065062702633440495, | |
| "learning_rate": 4.582608695652174e-05, | |
| "loss": 0.00029802778735756876, | |
| "step": 6325 | |
| }, | |
| { | |
| "epoch": 0.8255330213208528, | |
| "grad_norm": 0.005543549545109272, | |
| "learning_rate": 4.60072463768116e-05, | |
| "loss": 0.00036395326256752017, | |
| "step": 6350 | |
| }, | |
| { | |
| "epoch": 0.8287831513260531, | |
| "grad_norm": 0.0020637274719774723, | |
| "learning_rate": 4.618840579710145e-05, | |
| "loss": 0.00036908872425556183, | |
| "step": 6375 | |
| }, | |
| { | |
| "epoch": 0.8320332813312532, | |
| "grad_norm": 0.003742776345461607, | |
| "learning_rate": 4.6369565217391306e-05, | |
| "loss": 0.0003229105472564697, | |
| "step": 6400 | |
| }, | |
| { | |
| "epoch": 0.8352834113364535, | |
| "grad_norm": 0.001703648827970028, | |
| "learning_rate": 4.655072463768116e-05, | |
| "loss": 0.00033204488456249235, | |
| "step": 6425 | |
| }, | |
| { | |
| "epoch": 0.8385335413416537, | |
| "grad_norm": 0.0012357374653220177, | |
| "learning_rate": 4.673188405797102e-05, | |
| "loss": 0.00032585758715867995, | |
| "step": 6450 | |
| }, | |
| { | |
| "epoch": 0.8417836713468538, | |
| "grad_norm": 0.006874839309602976, | |
| "learning_rate": 4.691304347826087e-05, | |
| "loss": 0.00027060166001319886, | |
| "step": 6475 | |
| }, | |
| { | |
| "epoch": 0.8450338013520541, | |
| "grad_norm": 0.0022275864612311125, | |
| "learning_rate": 4.7094202898550724e-05, | |
| "loss": 0.0003158362582325935, | |
| "step": 6500 | |
| }, | |
| { | |
| "epoch": 0.8482839313572543, | |
| "grad_norm": 0.0011992513900622725, | |
| "learning_rate": 4.727536231884058e-05, | |
| "loss": 0.0003127120807766914, | |
| "step": 6525 | |
| }, | |
| { | |
| "epoch": 0.8515340613624545, | |
| "grad_norm": 0.0010171785252168775, | |
| "learning_rate": 4.7456521739130436e-05, | |
| "loss": 0.0002979818731546402, | |
| "step": 6550 | |
| }, | |
| { | |
| "epoch": 0.8547841913676547, | |
| "grad_norm": 0.0028656974900513887, | |
| "learning_rate": 4.7637681159420296e-05, | |
| "loss": 0.00031414445489645, | |
| "step": 6575 | |
| }, | |
| { | |
| "epoch": 0.858034321372855, | |
| "grad_norm": 0.0008505930309183896, | |
| "learning_rate": 4.781884057971014e-05, | |
| "loss": 0.000277229230850935, | |
| "step": 6600 | |
| }, | |
| { | |
| "epoch": 0.8612844513780551, | |
| "grad_norm": 0.0007299727294594049, | |
| "learning_rate": 4.8e-05, | |
| "loss": 0.0002848735637962818, | |
| "step": 6625 | |
| }, | |
| { | |
| "epoch": 0.8645345813832553, | |
| "grad_norm": 0.001255713519640267, | |
| "learning_rate": 4.8181159420289854e-05, | |
| "loss": 0.00028432700783014296, | |
| "step": 6650 | |
| }, | |
| { | |
| "epoch": 0.8677847113884556, | |
| "grad_norm": 0.0008443425176665187, | |
| "learning_rate": 4.8362318840579714e-05, | |
| "loss": 0.000307377353310585, | |
| "step": 6675 | |
| }, | |
| { | |
| "epoch": 0.8710348413936557, | |
| "grad_norm": 0.0024411012418568134, | |
| "learning_rate": 4.8543478260869566e-05, | |
| "loss": 0.00029829660430550573, | |
| "step": 6700 | |
| }, | |
| { | |
| "epoch": 0.874284971398856, | |
| "grad_norm": 0.001056804321706295, | |
| "learning_rate": 4.8724637681159426e-05, | |
| "loss": 0.0003243882581591606, | |
| "step": 6725 | |
| }, | |
| { | |
| "epoch": 0.8775351014040562, | |
| "grad_norm": 0.0011033945484086871, | |
| "learning_rate": 4.890579710144928e-05, | |
| "loss": 0.0002712445519864559, | |
| "step": 6750 | |
| }, | |
| { | |
| "epoch": 0.8807852314092564, | |
| "grad_norm": 0.001996325794607401, | |
| "learning_rate": 4.908695652173913e-05, | |
| "loss": 0.0003028985857963562, | |
| "step": 6775 | |
| }, | |
| { | |
| "epoch": 0.8840353614144566, | |
| "grad_norm": 0.001191786490380764, | |
| "learning_rate": 4.9268115942028984e-05, | |
| "loss": 0.00029174651950597763, | |
| "step": 6800 | |
| }, | |
| { | |
| "epoch": 0.8872854914196567, | |
| "grad_norm": 0.0018096808344125748, | |
| "learning_rate": 4.9449275362318844e-05, | |
| "loss": 0.0002984482236206532, | |
| "step": 6825 | |
| }, | |
| { | |
| "epoch": 0.890535621424857, | |
| "grad_norm": 0.0015695627080276608, | |
| "learning_rate": 4.96304347826087e-05, | |
| "loss": 0.00027819450944662096, | |
| "step": 6850 | |
| }, | |
| { | |
| "epoch": 0.8937857514300572, | |
| "grad_norm": 0.0014000756200402975, | |
| "learning_rate": 4.9811594202898556e-05, | |
| "loss": 0.00033636294305324556, | |
| "step": 6875 | |
| }, | |
| { | |
| "epoch": 0.8970358814352574, | |
| "grad_norm": 0.0017718684393912554, | |
| "learning_rate": 4.999275362318841e-05, | |
| "loss": 0.0003006098046898842, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 0.8970358814352574, | |
| "eval_loss": 0.001117659150622785, | |
| "eval_runtime": 9.2541, | |
| "eval_samples_per_second": 540.302, | |
| "eval_steps_per_second": 4.322, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 0.8970358814352574, | |
| "eval_holdout_avg_strict_f1": 0.047399586588974675, | |
| "eval_holdout_enron_spam_bvk_strict_f1": 0.0, | |
| "eval_holdout_enron_spam_setfit_strict_f1": 0.0, | |
| "eval_holdout_gliner2_pii_ptbr_reward_split_strict_f1": 0.1927814998919386, | |
| "eval_holdout_nemotron_pii_strict_f1": 0.062429558484833635, | |
| "eval_holdout_open_pii_masking_500k_strict_f1": 0.16210173281162663, | |
| "eval_holdout_phishing_darkknight_strict_f1": 0.0, | |
| "eval_holdout_phishing_zefang_strict_f1": 0.0, | |
| "eval_holdout_pii_masking_400k_strict_f1": 0.10408266129032258, | |
| "eval_holdout_sms_spam_multilingual_strict_f1": 0.0, | |
| "eval_holdout_spam_messages_mshenoda_strict_f1": 0.0, | |
| "eval_holdout_spamassassin_strict_f1": 0.0, | |
| "step": 6900 | |
| }, | |
| { | |
| "epoch": 0.9002860114404576, | |
| "grad_norm": 0.0030266817193478346, | |
| "learning_rate": 4.999998157323963e-05, | |
| "loss": 0.00031684111803770065, | |
| "step": 6925 | |
| }, | |
| { | |
| "epoch": 0.9035361414456579, | |
| "grad_norm": 0.0023532931227236986, | |
| "learning_rate": 4.999992318987079e-05, | |
| "loss": 0.00027174990624189376, | |
| "step": 6950 | |
| }, | |
| { | |
| "epoch": 0.906786271450858, | |
| "grad_norm": 0.0015111297834664583, | |
| "learning_rate": 4.999982481799595e-05, | |
| "loss": 0.00034066613763570784, | |
| "step": 6975 | |
| }, | |
| { | |
| "epoch": 0.9100364014560582, | |
| "grad_norm": 0.002268970711156726, | |
| "learning_rate": 4.999968645777246e-05, | |
| "loss": 0.0002984613366425037, | |
| "step": 7000 | |
| }, | |
| { | |
| "epoch": 0.9132865314612585, | |
| "grad_norm": 0.00104411446955055, | |
| "learning_rate": 4.999950810942164e-05, | |
| "loss": 0.0003224646672606468, | |
| "step": 7025 | |
| }, | |
| { | |
| "epoch": 0.9165366614664586, | |
| "grad_norm": 0.001416088198311627, | |
| "learning_rate": 4.999928977322876e-05, | |
| "loss": 0.0002991696447134018, | |
| "step": 7050 | |
| }, | |
| { | |
| "epoch": 0.9197867914716589, | |
| "grad_norm": 0.002220858121290803, | |
| "learning_rate": 4.999903144954307e-05, | |
| "loss": 0.00030125860124826433, | |
| "step": 7075 | |
| }, | |
| { | |
| "epoch": 0.9230369214768591, | |
| "grad_norm": 0.001945727621205151, | |
| "learning_rate": 4.9998733138777766e-05, | |
| "loss": 0.00029435833916068076, | |
| "step": 7100 | |
| }, | |
| { | |
| "epoch": 0.9262870514820593, | |
| "grad_norm": 0.0019027380039915442, | |
| "learning_rate": 4.999839484140999e-05, | |
| "loss": 0.00026273664087057114, | |
| "step": 7125 | |
| }, | |
| { | |
| "epoch": 0.9295371814872595, | |
| "grad_norm": 0.0028017114382237196, | |
| "learning_rate": 4.9998016557980896e-05, | |
| "loss": 0.0002820346690714359, | |
| "step": 7150 | |
| }, | |
| { | |
| "epoch": 0.9327873114924597, | |
| "grad_norm": 0.0015707556158304214, | |
| "learning_rate": 4.999759828909555e-05, | |
| "loss": 0.00028507132083177567, | |
| "step": 7175 | |
| }, | |
| { | |
| "epoch": 0.9360374414976599, | |
| "grad_norm": 0.0019409273518249393, | |
| "learning_rate": 4.9997140035422984e-05, | |
| "loss": 0.00028136296197772025, | |
| "step": 7200 | |
| }, | |
| { | |
| "epoch": 0.9392875715028601, | |
| "grad_norm": 0.0018527988577261567, | |
| "learning_rate": 4.9996641797696206e-05, | |
| "loss": 0.00028836311772465704, | |
| "step": 7225 | |
| }, | |
| { | |
| "epoch": 0.9425377015080604, | |
| "grad_norm": 0.0027005046140402555, | |
| "learning_rate": 4.999610357671217e-05, | |
| "loss": 0.0002803043834865093, | |
| "step": 7250 | |
| }, | |
| { | |
| "epoch": 0.9457878315132605, | |
| "grad_norm": 0.0014811495784670115, | |
| "learning_rate": 4.999552537333178e-05, | |
| "loss": 0.00028542324900627137, | |
| "step": 7275 | |
| }, | |
| { | |
| "epoch": 0.9490379615184608, | |
| "grad_norm": 0.0015694370958954096, | |
| "learning_rate": 4.99949071884799e-05, | |
| "loss": 0.00029463691636919974, | |
| "step": 7300 | |
| }, | |
| { | |
| "epoch": 0.9522880915236609, | |
| "grad_norm": 0.0013177198125049472, | |
| "learning_rate": 4.999424902314534e-05, | |
| "loss": 0.00025012722238898277, | |
| "step": 7325 | |
| }, | |
| { | |
| "epoch": 0.9555382215288611, | |
| "grad_norm": 0.000802877766545862, | |
| "learning_rate": 4.9993550878380876e-05, | |
| "loss": 0.0002755519188940525, | |
| "step": 7350 | |
| }, | |
| { | |
| "epoch": 0.9587883515340614, | |
| "grad_norm": 0.0006859932327643037, | |
| "learning_rate": 4.999281275530321e-05, | |
| "loss": 0.00028787543997168543, | |
| "step": 7375 | |
| }, | |
| { | |
| "epoch": 0.9620384815392615, | |
| "grad_norm": 0.0023639509454369545, | |
| "learning_rate": 4.9992034655093016e-05, | |
| "loss": 0.00028885766863822937, | |
| "step": 7400 | |
| }, | |
| { | |
| "epoch": 0.9652886115444618, | |
| "grad_norm": 0.0009986214572563767, | |
| "learning_rate": 4.9991216578994894e-05, | |
| "loss": 0.00027358153834939003, | |
| "step": 7425 | |
| }, | |
| { | |
| "epoch": 0.968538741549662, | |
| "grad_norm": 0.0029798035975545645, | |
| "learning_rate": 4.999035852831739e-05, | |
| "loss": 0.00027591008692979814, | |
| "step": 7450 | |
| }, | |
| { | |
| "epoch": 0.9717888715548622, | |
| "grad_norm": 0.0023388254921883345, | |
| "learning_rate": 4.998946050443299e-05, | |
| "loss": 0.00026363175362348557, | |
| "step": 7475 | |
| }, | |
| { | |
| "epoch": 0.9750390015600624, | |
| "grad_norm": 0.000977223739027977, | |
| "learning_rate": 4.9988522508778144e-05, | |
| "loss": 0.00030499791726469995, | |
| "step": 7500 | |
| }, | |
| { | |
| "epoch": 0.9782891315652626, | |
| "grad_norm": 0.0017290436662733555, | |
| "learning_rate": 4.9987544542853207e-05, | |
| "loss": 0.00027209220454096797, | |
| "step": 7525 | |
| }, | |
| { | |
| "epoch": 0.9815392615704628, | |
| "grad_norm": 0.0011989338090643287, | |
| "learning_rate": 4.9986526608222475e-05, | |
| "loss": 0.0002605011686682701, | |
| "step": 7550 | |
| }, | |
| { | |
| "epoch": 0.984789391575663, | |
| "grad_norm": 0.0009172749705612659, | |
| "learning_rate": 4.998546870651418e-05, | |
| "loss": 0.0002654053457081318, | |
| "step": 7575 | |
| }, | |
| { | |
| "epoch": 0.9880395215808633, | |
| "grad_norm": 0.0017111243214458227, | |
| "learning_rate": 4.998437083942049e-05, | |
| "loss": 0.00027476752176880837, | |
| "step": 7600 | |
| }, | |
| { | |
| "epoch": 0.9912896515860634, | |
| "grad_norm": 0.0009734177147038281, | |
| "learning_rate": 4.998323300869749e-05, | |
| "loss": 0.0002582584507763386, | |
| "step": 7625 | |
| }, | |
| { | |
| "epoch": 0.9945397815912637, | |
| "grad_norm": 0.0007609701715409756, | |
| "learning_rate": 4.99820552161652e-05, | |
| "loss": 0.00026447221636772156, | |
| "step": 7650 | |
| }, | |
| { | |
| "epoch": 0.9977899115964639, | |
| "grad_norm": 0.002729939529672265, | |
| "learning_rate": 4.998083746370754e-05, | |
| "loss": 0.00029974594712257386, | |
| "step": 7675 | |
| }, | |
| { | |
| "epoch": 1.001040041601664, | |
| "grad_norm": 0.0016189470188692212, | |
| "learning_rate": 4.9979579753272364e-05, | |
| "loss": 0.00026583246886730194, | |
| "step": 7700 | |
| }, | |
| { | |
| "epoch": 1.0042901716068642, | |
| "grad_norm": 0.0018497901037335396, | |
| "learning_rate": 4.997828208687144e-05, | |
| "loss": 0.00030912613496184347, | |
| "step": 7725 | |
| }, | |
| { | |
| "epoch": 1.0075403016120645, | |
| "grad_norm": 0.002935883356258273, | |
| "learning_rate": 4.997694446658044e-05, | |
| "loss": 0.0002521901950240135, | |
| "step": 7750 | |
| }, | |
| { | |
| "epoch": 1.0107904316172647, | |
| "grad_norm": 0.0007819763850420713, | |
| "learning_rate": 4.997556689453895e-05, | |
| "loss": 0.00022437306120991706, | |
| "step": 7775 | |
| }, | |
| { | |
| "epoch": 1.0140405616224648, | |
| "grad_norm": 0.001313910586759448, | |
| "learning_rate": 4.997414937295046e-05, | |
| "loss": 0.00027127765119075773, | |
| "step": 7800 | |
| }, | |
| { | |
| "epoch": 1.0172906916276652, | |
| "grad_norm": 0.0016395088750869036, | |
| "learning_rate": 4.997269190408235e-05, | |
| "loss": 0.00027738766744732856, | |
| "step": 7825 | |
| }, | |
| { | |
| "epoch": 1.0205408216328653, | |
| "grad_norm": 0.0016652283957228065, | |
| "learning_rate": 4.997119449026592e-05, | |
| "loss": 0.0002703977935016155, | |
| "step": 7850 | |
| }, | |
| { | |
| "epoch": 1.0237909516380654, | |
| "grad_norm": 0.0010423657950013876, | |
| "learning_rate": 4.996965713389635e-05, | |
| "loss": 0.00022714242339134217, | |
| "step": 7875 | |
| }, | |
| { | |
| "epoch": 1.0270410816432658, | |
| "grad_norm": 0.000837469648104161, | |
| "learning_rate": 4.99680798374327e-05, | |
| "loss": 0.00026420921087265013, | |
| "step": 7900 | |
| }, | |
| { | |
| "epoch": 1.030291211648466, | |
| "grad_norm": 0.001708422671072185, | |
| "learning_rate": 4.9966462603397946e-05, | |
| "loss": 0.0002601585164666176, | |
| "step": 7925 | |
| }, | |
| { | |
| "epoch": 1.033541341653666, | |
| "grad_norm": 0.0008940563420765102, | |
| "learning_rate": 4.9964805434378906e-05, | |
| "loss": 0.000257425457239151, | |
| "step": 7950 | |
| }, | |
| { | |
| "epoch": 1.0367914716588664, | |
| "grad_norm": 0.0009178609470836818, | |
| "learning_rate": 4.996310833302631e-05, | |
| "loss": 0.00029499370604753496, | |
| "step": 7975 | |
| }, | |
| { | |
| "epoch": 1.0400416016640666, | |
| "grad_norm": 0.0012339595705270767, | |
| "learning_rate": 4.996137130205474e-05, | |
| "loss": 0.00030053269118070604, | |
| "step": 8000 | |
| }, | |
| { | |
| "epoch": 1.0432917316692667, | |
| "grad_norm": 0.0012492527021095157, | |
| "learning_rate": 4.995959434424265e-05, | |
| "loss": 0.00028023691847920416, | |
| "step": 8025 | |
| }, | |
| { | |
| "epoch": 1.046541861674467, | |
| "grad_norm": 0.0011979590635746717, | |
| "learning_rate": 4.9957777462432394e-05, | |
| "loss": 0.00030107472091913225, | |
| "step": 8050 | |
| }, | |
| { | |
| "epoch": 1.0497919916796672, | |
| "grad_norm": 0.0021881945431232452, | |
| "learning_rate": 4.9955920659530124e-05, | |
| "loss": 0.00025908147916197776, | |
| "step": 8075 | |
| }, | |
| { | |
| "epoch": 1.0530421216848673, | |
| "grad_norm": 0.001974343555048108, | |
| "learning_rate": 4.99540239385059e-05, | |
| "loss": 0.0002683600224554539, | |
| "step": 8100 | |
| }, | |
| { | |
| "epoch": 1.0562922516900677, | |
| "grad_norm": 0.00107442622538656, | |
| "learning_rate": 4.99520873023936e-05, | |
| "loss": 0.0002770363166928291, | |
| "step": 8125 | |
| }, | |
| { | |
| "epoch": 1.0595423816952678, | |
| "grad_norm": 0.0019235480576753616, | |
| "learning_rate": 4.9950110754290966e-05, | |
| "loss": 0.00027815064415335653, | |
| "step": 8150 | |
| }, | |
| { | |
| "epoch": 1.062792511700468, | |
| "grad_norm": 0.001666422002017498, | |
| "learning_rate": 4.994809429735957e-05, | |
| "loss": 0.0002477550320327282, | |
| "step": 8175 | |
| }, | |
| { | |
| "epoch": 1.0660426417056683, | |
| "grad_norm": 0.0009593910654075444, | |
| "learning_rate": 4.994603793482484e-05, | |
| "loss": 0.0002452738955616951, | |
| "step": 8200 | |
| }, | |
| { | |
| "epoch": 1.0692927717108685, | |
| "grad_norm": 0.0018539578886702657, | |
| "learning_rate": 4.994394166997601e-05, | |
| "loss": 0.0002694200538098812, | |
| "step": 8225 | |
| }, | |
| { | |
| "epoch": 1.0725429017160686, | |
| "grad_norm": 0.001612568274140358, | |
| "learning_rate": 4.994180550616615e-05, | |
| "loss": 0.00026693956926465034, | |
| "step": 8250 | |
| }, | |
| { | |
| "epoch": 1.075793031721269, | |
| "grad_norm": 0.0010748225031420588, | |
| "learning_rate": 4.993962944681216e-05, | |
| "loss": 0.00025574009865522387, | |
| "step": 8275 | |
| }, | |
| { | |
| "epoch": 1.079043161726469, | |
| "grad_norm": 0.0026713060215115547, | |
| "learning_rate": 4.993741349539474e-05, | |
| "loss": 0.00027934052050113676, | |
| "step": 8300 | |
| }, | |
| { | |
| "epoch": 1.0822932917316692, | |
| "grad_norm": 0.0028977917972952127, | |
| "learning_rate": 4.99351576554584e-05, | |
| "loss": 0.0002870534546673298, | |
| "step": 8325 | |
| }, | |
| { | |
| "epoch": 1.0855434217368696, | |
| "grad_norm": 0.0017800545319914818, | |
| "learning_rate": 4.9932861930611454e-05, | |
| "loss": 0.0002597590535879135, | |
| "step": 8350 | |
| }, | |
| { | |
| "epoch": 1.0887935517420697, | |
| "grad_norm": 0.0010963486274704337, | |
| "learning_rate": 4.993052632452603e-05, | |
| "loss": 0.00021376239135861396, | |
| "step": 8375 | |
| }, | |
| { | |
| "epoch": 1.0920436817472698, | |
| "grad_norm": 0.004095334559679031, | |
| "learning_rate": 4.9928150840938034e-05, | |
| "loss": 0.00024395940825343133, | |
| "step": 8400 | |
| }, | |
| { | |
| "epoch": 1.09529381175247, | |
| "grad_norm": 0.0016407609218731523, | |
| "learning_rate": 4.9925735483647144e-05, | |
| "loss": 0.00022858794778585434, | |
| "step": 8425 | |
| }, | |
| { | |
| "epoch": 1.0985439417576703, | |
| "grad_norm": 0.0011426799464970827, | |
| "learning_rate": 4.9923280256516856e-05, | |
| "loss": 0.00025209063664078713, | |
| "step": 8450 | |
| }, | |
| { | |
| "epoch": 1.1017940717628705, | |
| "grad_norm": 0.0008142830338329077, | |
| "learning_rate": 4.99207851634744e-05, | |
| "loss": 0.00023667672649025918, | |
| "step": 8475 | |
| }, | |
| { | |
| "epoch": 1.1050442017680706, | |
| "grad_norm": 0.0013887656386941671, | |
| "learning_rate": 4.991825020851079e-05, | |
| "loss": 0.00023785781115293503, | |
| "step": 8500 | |
| }, | |
| { | |
| "epoch": 1.108294331773271, | |
| "grad_norm": 0.0016793243121355772, | |
| "learning_rate": 4.991567539568081e-05, | |
| "loss": 0.00025064632296562195, | |
| "step": 8525 | |
| }, | |
| { | |
| "epoch": 1.111544461778471, | |
| "grad_norm": 0.0015402145218104124, | |
| "learning_rate": 4.991306072910299e-05, | |
| "loss": 0.0002445879392325878, | |
| "step": 8550 | |
| }, | |
| { | |
| "epoch": 1.1147945917836712, | |
| "grad_norm": 0.0006315167411230505, | |
| "learning_rate": 4.991040621295959e-05, | |
| "loss": 0.0002341064065694809, | |
| "step": 8575 | |
| }, | |
| { | |
| "epoch": 1.1180447217888716, | |
| "grad_norm": 0.0017532727215439081, | |
| "learning_rate": 4.9907711851496655e-05, | |
| "loss": 0.0002281763032078743, | |
| "step": 8600 | |
| }, | |
| { | |
| "epoch": 1.1212948517940717, | |
| "grad_norm": 0.0017055863281711936, | |
| "learning_rate": 4.9904977649023906e-05, | |
| "loss": 0.00024771852418780326, | |
| "step": 8625 | |
| }, | |
| { | |
| "epoch": 1.1245449817992719, | |
| "grad_norm": 0.003801319282501936, | |
| "learning_rate": 4.9902203609914855e-05, | |
| "loss": 0.00024433283135294915, | |
| "step": 8650 | |
| }, | |
| { | |
| "epoch": 1.1277951118044722, | |
| "grad_norm": 0.0022220660466700792, | |
| "learning_rate": 4.989938973860668e-05, | |
| "loss": 0.0002709745801985264, | |
| "step": 8675 | |
| }, | |
| { | |
| "epoch": 1.1310452418096724, | |
| "grad_norm": 0.0017162845470011234, | |
| "learning_rate": 4.989653603960032e-05, | |
| "loss": 0.00023029427975416184, | |
| "step": 8700 | |
| }, | |
| { | |
| "epoch": 1.1342953718148725, | |
| "grad_norm": 0.0037925834767520428, | |
| "learning_rate": 4.989364251746036e-05, | |
| "loss": 0.00024551210924983024, | |
| "step": 8725 | |
| }, | |
| { | |
| "epoch": 1.1375455018200729, | |
| "grad_norm": 0.0031808854546397924, | |
| "learning_rate": 4.989070917681515e-05, | |
| "loss": 0.00026286788284778594, | |
| "step": 8750 | |
| }, | |
| { | |
| "epoch": 1.140795631825273, | |
| "grad_norm": 0.0016192150069400668, | |
| "learning_rate": 4.988773602235669e-05, | |
| "loss": 0.00026194704696536064, | |
| "step": 8775 | |
| }, | |
| { | |
| "epoch": 1.1440457618304731, | |
| "grad_norm": 0.0011428098659962416, | |
| "learning_rate": 4.9884723058840665e-05, | |
| "loss": 0.00022088117897510527, | |
| "step": 8800 | |
| }, | |
| { | |
| "epoch": 1.1472958918356735, | |
| "grad_norm": 0.0011217723367735744, | |
| "learning_rate": 4.988167029108646e-05, | |
| "loss": 0.00024152018129825593, | |
| "step": 8825 | |
| }, | |
| { | |
| "epoch": 1.1505460218408736, | |
| "grad_norm": 0.0013131353771314025, | |
| "learning_rate": 4.9878577723977116e-05, | |
| "loss": 0.00024948714300990107, | |
| "step": 8850 | |
| }, | |
| { | |
| "epoch": 1.1537961518460738, | |
| "grad_norm": 0.001230393536388874, | |
| "learning_rate": 4.987544536245931e-05, | |
| "loss": 0.00024934813380241396, | |
| "step": 8875 | |
| }, | |
| { | |
| "epoch": 1.1570462818512741, | |
| "grad_norm": 0.001053430256433785, | |
| "learning_rate": 4.987227321154344e-05, | |
| "loss": 0.00024760672822594644, | |
| "step": 8900 | |
| }, | |
| { | |
| "epoch": 1.1602964118564743, | |
| "grad_norm": 0.0014641049783676863, | |
| "learning_rate": 4.986906127630346e-05, | |
| "loss": 0.00021234707906842232, | |
| "step": 8925 | |
| }, | |
| { | |
| "epoch": 1.1635465418616744, | |
| "grad_norm": 0.0020078544039279222, | |
| "learning_rate": 4.986580956187703e-05, | |
| "loss": 0.00022885441780090332, | |
| "step": 8950 | |
| }, | |
| { | |
| "epoch": 1.1667966718668747, | |
| "grad_norm": 0.0016951245488598943, | |
| "learning_rate": 4.9862518073465405e-05, | |
| "loss": 0.00024043211713433265, | |
| "step": 8975 | |
| }, | |
| { | |
| "epoch": 1.1700468018720749, | |
| "grad_norm": 0.0011139962589368224, | |
| "learning_rate": 4.985918681633348e-05, | |
| "loss": 0.0002487810701131821, | |
| "step": 9000 | |
| }, | |
| { | |
| "epoch": 1.173296931877275, | |
| "grad_norm": 0.0008373457822017372, | |
| "learning_rate": 4.985581579580973e-05, | |
| "loss": 0.00023622142150998116, | |
| "step": 9025 | |
| }, | |
| { | |
| "epoch": 1.1765470618824754, | |
| "grad_norm": 0.0013526729308068752, | |
| "learning_rate": 4.985240501728628e-05, | |
| "loss": 0.0002516124211251736, | |
| "step": 9050 | |
| }, | |
| { | |
| "epoch": 1.1797971918876755, | |
| "grad_norm": 0.0012392222415655851, | |
| "learning_rate": 4.984895448621881e-05, | |
| "loss": 0.00022982005029916763, | |
| "step": 9075 | |
| }, | |
| { | |
| "epoch": 1.1830473218928756, | |
| "grad_norm": 0.0012214110465720296, | |
| "learning_rate": 4.98454642081266e-05, | |
| "loss": 0.00023133255541324616, | |
| "step": 9100 | |
| }, | |
| { | |
| "epoch": 1.186297451898076, | |
| "grad_norm": 0.0007031799177639186, | |
| "learning_rate": 4.984193418859251e-05, | |
| "loss": 0.00024310629814863204, | |
| "step": 9125 | |
| }, | |
| { | |
| "epoch": 1.1895475819032761, | |
| "grad_norm": 0.0014168958878144622, | |
| "learning_rate": 4.9838364433262965e-05, | |
| "loss": 0.000265915896743536, | |
| "step": 9150 | |
| }, | |
| { | |
| "epoch": 1.1927977119084763, | |
| "grad_norm": 0.0010334665421396494, | |
| "learning_rate": 4.983475494784795e-05, | |
| "loss": 0.000241184551268816, | |
| "step": 9175 | |
| }, | |
| { | |
| "epoch": 1.1960478419136766, | |
| "grad_norm": 0.0019802378956228495, | |
| "learning_rate": 4.9831105738120995e-05, | |
| "loss": 0.0002582270838320255, | |
| "step": 9200 | |
| }, | |
| { | |
| "epoch": 1.1992979719188768, | |
| "grad_norm": 0.0015125485369935632, | |
| "learning_rate": 4.982741680991919e-05, | |
| "loss": 0.00022809958085417747, | |
| "step": 9225 | |
| }, | |
| { | |
| "epoch": 1.202548101924077, | |
| "grad_norm": 0.0014787097461521626, | |
| "learning_rate": 4.982368816914314e-05, | |
| "loss": 0.00020900048315525055, | |
| "step": 9250 | |
| }, | |
| { | |
| "epoch": 1.2057982319292773, | |
| "grad_norm": 0.0005954820080660284, | |
| "learning_rate": 4.981991982175695e-05, | |
| "loss": 0.00021818505600094796, | |
| "step": 9275 | |
| }, | |
| { | |
| "epoch": 1.2090483619344774, | |
| "grad_norm": 0.0026676205452531576, | |
| "learning_rate": 4.981611177378829e-05, | |
| "loss": 0.00022581884637475015, | |
| "step": 9300 | |
| }, | |
| { | |
| "epoch": 1.2122984919396775, | |
| "grad_norm": 0.0024808698799461126, | |
| "learning_rate": 4.9812264031328285e-05, | |
| "loss": 0.00027215268462896344, | |
| "step": 9325 | |
| }, | |
| { | |
| "epoch": 1.215548621944878, | |
| "grad_norm": 0.0031435610726475716, | |
| "learning_rate": 4.980837660053159e-05, | |
| "loss": 0.00024729188531637193, | |
| "step": 9350 | |
| }, | |
| { | |
| "epoch": 1.218798751950078, | |
| "grad_norm": 0.002376891439780593, | |
| "learning_rate": 4.980444948761631e-05, | |
| "loss": 0.000256987065076828, | |
| "step": 9375 | |
| }, | |
| { | |
| "epoch": 1.2220488819552782, | |
| "grad_norm": 0.0008802884840406477, | |
| "learning_rate": 4.980048269886406e-05, | |
| "loss": 0.0002453344129025936, | |
| "step": 9400 | |
| }, | |
| { | |
| "epoch": 1.2252990119604785, | |
| "grad_norm": 0.0007192345801740885, | |
| "learning_rate": 4.979647624061987e-05, | |
| "loss": 0.0002300410531461239, | |
| "step": 9425 | |
| }, | |
| { | |
| "epoch": 1.2285491419656787, | |
| "grad_norm": 0.0010762346209958196, | |
| "learning_rate": 4.9792430119292275e-05, | |
| "loss": 0.00026356540620326997, | |
| "step": 9450 | |
| }, | |
| { | |
| "epoch": 1.2317992719708788, | |
| "grad_norm": 0.0016680917469784617, | |
| "learning_rate": 4.978834434135322e-05, | |
| "loss": 0.0002518729493021965, | |
| "step": 9475 | |
| }, | |
| { | |
| "epoch": 1.2350494019760792, | |
| "grad_norm": 0.0014808833366259933, | |
| "learning_rate": 4.9784218913338096e-05, | |
| "loss": 0.0002638903260231018, | |
| "step": 9500 | |
| }, | |
| { | |
| "epoch": 1.2382995319812793, | |
| "grad_norm": 0.0018340969691053033, | |
| "learning_rate": 4.9780053841845697e-05, | |
| "loss": 0.00024728938937187197, | |
| "step": 9525 | |
| }, | |
| { | |
| "epoch": 1.2415496619864794, | |
| "grad_norm": 0.000639909238088876, | |
| "learning_rate": 4.9775849133538254e-05, | |
| "loss": 0.0002366364374756813, | |
| "step": 9550 | |
| }, | |
| { | |
| "epoch": 1.2447997919916798, | |
| "grad_norm": 0.0005515090888366103, | |
| "learning_rate": 4.9771604795141385e-05, | |
| "loss": 0.0002480141818523407, | |
| "step": 9575 | |
| }, | |
| { | |
| "epoch": 1.24804992199688, | |
| "grad_norm": 0.001218414050526917, | |
| "learning_rate": 4.976732083344411e-05, | |
| "loss": 0.0002754170633852482, | |
| "step": 9600 | |
| }, | |
| { | |
| "epoch": 1.25130005200208, | |
| "grad_norm": 0.0019095524912700057, | |
| "learning_rate": 4.97629972552988e-05, | |
| "loss": 0.00025110188871622083, | |
| "step": 9625 | |
| }, | |
| { | |
| "epoch": 1.2545501820072804, | |
| "grad_norm": 0.0022047131787985563, | |
| "learning_rate": 4.975863406762124e-05, | |
| "loss": 0.00023579886183142663, | |
| "step": 9650 | |
| }, | |
| { | |
| "epoch": 1.2578003120124805, | |
| "grad_norm": 0.0010715682292357087, | |
| "learning_rate": 4.9754231277390514e-05, | |
| "loss": 0.00025938162580132487, | |
| "step": 9675 | |
| }, | |
| { | |
| "epoch": 1.2610504420176807, | |
| "grad_norm": 0.0014722577761858702, | |
| "learning_rate": 4.974978889164911e-05, | |
| "loss": 0.00023980937898159026, | |
| "step": 9700 | |
| }, | |
| { | |
| "epoch": 1.264300572022881, | |
| "grad_norm": 0.0010208309395238757, | |
| "learning_rate": 4.9745306917502805e-05, | |
| "loss": 0.00021998122334480286, | |
| "step": 9725 | |
| }, | |
| { | |
| "epoch": 1.2675507020280812, | |
| "grad_norm": 0.0005931083578616381, | |
| "learning_rate": 4.9740785362120736e-05, | |
| "loss": 0.0002296293154358864, | |
| "step": 9750 | |
| }, | |
| { | |
| "epoch": 1.2708008320332813, | |
| "grad_norm": 0.009204435162246227, | |
| "learning_rate": 4.973622423273533e-05, | |
| "loss": 0.00026070356369018555, | |
| "step": 9775 | |
| }, | |
| { | |
| "epoch": 1.2740509620384817, | |
| "grad_norm": 0.0020595588721334934, | |
| "learning_rate": 4.973162353664231e-05, | |
| "loss": 0.00023075552657246588, | |
| "step": 9800 | |
| }, | |
| { | |
| "epoch": 1.2773010920436818, | |
| "grad_norm": 0.0013328915229067206, | |
| "learning_rate": 4.972698328120071e-05, | |
| "loss": 0.00023088693618774413, | |
| "step": 9825 | |
| }, | |
| { | |
| "epoch": 1.280551222048882, | |
| "grad_norm": 0.004525795113295317, | |
| "learning_rate": 4.9722303473832824e-05, | |
| "loss": 0.00023551590740680694, | |
| "step": 9850 | |
| }, | |
| { | |
| "epoch": 1.2838013520540823, | |
| "grad_norm": 0.0028794463723897934, | |
| "learning_rate": 4.97175841220242e-05, | |
| "loss": 0.00022693824023008345, | |
| "step": 9875 | |
| }, | |
| { | |
| "epoch": 1.2870514820592824, | |
| "grad_norm": 0.0015256752958521247, | |
| "learning_rate": 4.971282523332368e-05, | |
| "loss": 0.00022677043452858924, | |
| "step": 9900 | |
| }, | |
| { | |
| "epoch": 1.2903016120644826, | |
| "grad_norm": 0.0033297943882644176, | |
| "learning_rate": 4.9708026815343314e-05, | |
| "loss": 0.0002465812861919403, | |
| "step": 9925 | |
| }, | |
| { | |
| "epoch": 1.2935517420696827, | |
| "grad_norm": 0.000917601864784956, | |
| "learning_rate": 4.970318887575837e-05, | |
| "loss": 0.0002653016895055771, | |
| "step": 9950 | |
| }, | |
| { | |
| "epoch": 1.296801872074883, | |
| "grad_norm": 0.002418952528387308, | |
| "learning_rate": 4.969831142230737e-05, | |
| "loss": 0.00022590529173612594, | |
| "step": 9975 | |
| }, | |
| { | |
| "epoch": 1.3000520020800832, | |
| "grad_norm": 0.001454479293897748, | |
| "learning_rate": 4.969339446279202e-05, | |
| "loss": 0.00027158020064234735, | |
| "step": 10000 | |
| }, | |
| { | |
| "epoch": 1.3033021320852833, | |
| "grad_norm": 0.0012225952232256532, | |
| "learning_rate": 4.968843800507721e-05, | |
| "loss": 0.0002480693906545639, | |
| "step": 10025 | |
| }, | |
| { | |
| "epoch": 1.3065522620904837, | |
| "grad_norm": 0.001239932724274695, | |
| "learning_rate": 4.9683442057091024e-05, | |
| "loss": 0.0002388366684317589, | |
| "step": 10050 | |
| }, | |
| { | |
| "epoch": 1.3098023920956838, | |
| "grad_norm": 0.0016925641102716327, | |
| "learning_rate": 4.96784066268247e-05, | |
| "loss": 0.00022450752556324005, | |
| "step": 10075 | |
| }, | |
| { | |
| "epoch": 1.313052522100884, | |
| "grad_norm": 0.000952462840359658, | |
| "learning_rate": 4.9673331722332644e-05, | |
| "loss": 0.00020575648173689842, | |
| "step": 10100 | |
| }, | |
| { | |
| "epoch": 1.3163026521060843, | |
| "grad_norm": 0.0010533633176237345, | |
| "learning_rate": 4.9668217351732396e-05, | |
| "loss": 0.0002651925757527351, | |
| "step": 10125 | |
| }, | |
| { | |
| "epoch": 1.3195527821112845, | |
| "grad_norm": 0.00173302018083632, | |
| "learning_rate": 4.966306352320461e-05, | |
| "loss": 0.00021947069093585013, | |
| "step": 10150 | |
| }, | |
| { | |
| "epoch": 1.3228029121164846, | |
| "grad_norm": 0.002059893449768424, | |
| "learning_rate": 4.9657870244993074e-05, | |
| "loss": 0.00022877845913171768, | |
| "step": 10175 | |
| }, | |
| { | |
| "epoch": 1.3260530421216847, | |
| "grad_norm": 0.0013976760674268007, | |
| "learning_rate": 4.965263752540468e-05, | |
| "loss": 0.00020990870893001557, | |
| "step": 10200 | |
| }, | |
| { | |
| "epoch": 1.329303172126885, | |
| "grad_norm": 0.0012817075476050377, | |
| "learning_rate": 4.964736537280939e-05, | |
| "loss": 0.00023846810683608054, | |
| "step": 10225 | |
| }, | |
| { | |
| "epoch": 1.3325533021320852, | |
| "grad_norm": 0.001444034744054079, | |
| "learning_rate": 4.9642053795640255e-05, | |
| "loss": 0.00024047080427408218, | |
| "step": 10250 | |
| }, | |
| { | |
| "epoch": 1.3358034321372854, | |
| "grad_norm": 0.0013552135787904263, | |
| "learning_rate": 4.963670280239337e-05, | |
| "loss": 0.00022178145125508307, | |
| "step": 10275 | |
| }, | |
| { | |
| "epoch": 1.3390535621424857, | |
| "grad_norm": 0.0014398591592907906, | |
| "learning_rate": 4.96313124016279e-05, | |
| "loss": 0.0002458404749631882, | |
| "step": 10300 | |
| }, | |
| { | |
| "epoch": 1.3423036921476859, | |
| "grad_norm": 0.0016251268098130822, | |
| "learning_rate": 4.962588260196605e-05, | |
| "loss": 0.00020361352711915969, | |
| "step": 10325 | |
| }, | |
| { | |
| "epoch": 1.345553822152886, | |
| "grad_norm": 0.00420821039006114, | |
| "learning_rate": 4.9620413412093e-05, | |
| "loss": 0.00022385528311133385, | |
| "step": 10350 | |
| }, | |
| { | |
| "epoch": 1.345553822152886, | |
| "eval_loss": 0.00095493218395859, | |
| "eval_runtime": 9.3187, | |
| "eval_samples_per_second": 536.555, | |
| "eval_steps_per_second": 4.292, | |
| "step": 10350 | |
| }, | |
| { | |
| "epoch": 1.345553822152886, | |
| "eval_holdout_avg_strict_f1": 0.06836038252415962, | |
| "eval_holdout_enron_spam_bvk_strict_f1": 0.0, | |
| "eval_holdout_enron_spam_setfit_strict_f1": 0.0, | |
| "eval_holdout_gliner2_pii_ptbr_reward_split_strict_f1": 0.27740641711229946, | |
| "eval_holdout_nemotron_pii_strict_f1": 0.17409166773655155, | |
| "eval_holdout_open_pii_masking_500k_strict_f1": 0.16683151994724696, | |
| "eval_holdout_phishing_darkknight_strict_f1": 0.0, | |
| "eval_holdout_phishing_zefang_strict_f1": 0.0, | |
| "eval_holdout_pii_masking_400k_strict_f1": 0.13363460296965785, | |
| "eval_holdout_sms_spam_multilingual_strict_f1": 0.0, | |
| "eval_holdout_spam_messages_mshenoda_strict_f1": 0.0, | |
| "eval_holdout_spamassassin_strict_f1": 0.0, | |
| "step": 10350 | |
| }, | |
| { | |
| "epoch": 1.3488039521580864, | |
| "grad_norm": 0.0007410464459098876, | |
| "learning_rate": 4.9614904840756985e-05, | |
| "loss": 0.00022841570898890495, | |
| "step": 10375 | |
| }, | |
| { | |
| "epoch": 1.3520540821632865, | |
| "grad_norm": 0.0010557317873463035, | |
| "learning_rate": 4.96093568967692e-05, | |
| "loss": 0.0002412230148911476, | |
| "step": 10400 | |
| }, | |
| { | |
| "epoch": 1.3553042121684866, | |
| "grad_norm": 0.0012896433472633362, | |
| "learning_rate": 4.960376958900385e-05, | |
| "loss": 0.00022234860807657242, | |
| "step": 10425 | |
| }, | |
| { | |
| "epoch": 1.358554342173687, | |
| "grad_norm": 0.0014095305232331157, | |
| "learning_rate": 4.9598142926398075e-05, | |
| "loss": 0.0002344328723847866, | |
| "step": 10450 | |
| }, | |
| { | |
| "epoch": 1.3618044721788871, | |
| "grad_norm": 0.0018959847511723638, | |
| "learning_rate": 4.959247691795198e-05, | |
| "loss": 0.0002503533661365509, | |
| "step": 10475 | |
| }, | |
| { | |
| "epoch": 1.3650546021840873, | |
| "grad_norm": 0.0008397601777687669, | |
| "learning_rate": 4.9586771572728597e-05, | |
| "loss": 0.0002315433695912361, | |
| "step": 10500 | |
| }, | |
| { | |
| "epoch": 1.3683047321892876, | |
| "grad_norm": 0.0008668754016980529, | |
| "learning_rate": 4.958102689985388e-05, | |
| "loss": 0.00021066755056381225, | |
| "step": 10525 | |
| }, | |
| { | |
| "epoch": 1.3715548621944877, | |
| "grad_norm": 0.0016179216327145696, | |
| "learning_rate": 4.9575242908516705e-05, | |
| "loss": 0.00023502033203840256, | |
| "step": 10550 | |
| }, | |
| { | |
| "epoch": 1.3748049921996879, | |
| "grad_norm": 0.0014446236891672015, | |
| "learning_rate": 4.956941960796881e-05, | |
| "loss": 0.00023628858849406243, | |
| "step": 10575 | |
| }, | |
| { | |
| "epoch": 1.3780551222048882, | |
| "grad_norm": 0.0007303371676243842, | |
| "learning_rate": 4.9563557007524854e-05, | |
| "loss": 0.00022226251661777496, | |
| "step": 10600 | |
| }, | |
| { | |
| "epoch": 1.3813052522100884, | |
| "grad_norm": 0.0008387739071622491, | |
| "learning_rate": 4.95576551165623e-05, | |
| "loss": 0.00022688502445816994, | |
| "step": 10625 | |
| }, | |
| { | |
| "epoch": 1.3845553822152885, | |
| "grad_norm": 0.0012891859514638782, | |
| "learning_rate": 4.955171394452151e-05, | |
| "loss": 0.0002495447546243668, | |
| "step": 10650 | |
| }, | |
| { | |
| "epoch": 1.3878055122204889, | |
| "grad_norm": 0.0006860825233161449, | |
| "learning_rate": 4.954573350090565e-05, | |
| "loss": 0.0002485157176852226, | |
| "step": 10675 | |
| }, | |
| { | |
| "epoch": 1.391055642225689, | |
| "grad_norm": 0.0013238989049568772, | |
| "learning_rate": 4.953971379528071e-05, | |
| "loss": 0.00021471558138728142, | |
| "step": 10700 | |
| }, | |
| { | |
| "epoch": 1.3943057722308891, | |
| "grad_norm": 0.0013835977297276258, | |
| "learning_rate": 4.953365483727548e-05, | |
| "loss": 0.00021957125514745712, | |
| "step": 10725 | |
| }, | |
| { | |
| "epoch": 1.3975559022360895, | |
| "grad_norm": 0.0016742295119911432, | |
| "learning_rate": 4.952755663658154e-05, | |
| "loss": 0.0002339155413210392, | |
| "step": 10750 | |
| }, | |
| { | |
| "epoch": 1.4008060322412896, | |
| "grad_norm": 0.0007948394049890339, | |
| "learning_rate": 4.9521419202953236e-05, | |
| "loss": 0.00022374913096427916, | |
| "step": 10775 | |
| }, | |
| { | |
| "epoch": 1.4040561622464898, | |
| "grad_norm": 0.001215911004692316, | |
| "learning_rate": 4.951524254620766e-05, | |
| "loss": 0.0002276202104985714, | |
| "step": 10800 | |
| }, | |
| { | |
| "epoch": 1.4073062922516901, | |
| "grad_norm": 0.001693572150543332, | |
| "learning_rate": 4.950902667622468e-05, | |
| "loss": 0.00024126509204506875, | |
| "step": 10825 | |
| }, | |
| { | |
| "epoch": 1.4105564222568903, | |
| "grad_norm": 0.0011415114859119058, | |
| "learning_rate": 4.950277160294683e-05, | |
| "loss": 0.00021039823070168496, | |
| "step": 10850 | |
| }, | |
| { | |
| "epoch": 1.4138065522620904, | |
| "grad_norm": 0.0012624706141650677, | |
| "learning_rate": 4.94964773363794e-05, | |
| "loss": 0.0002335314080119133, | |
| "step": 10875 | |
| }, | |
| { | |
| "epoch": 1.4170566822672908, | |
| "grad_norm": 0.0023544938303530216, | |
| "learning_rate": 4.9490143886590355e-05, | |
| "loss": 0.00023012714460492135, | |
| "step": 10900 | |
| }, | |
| { | |
| "epoch": 1.420306812272491, | |
| "grad_norm": 0.001988065429031849, | |
| "learning_rate": 4.948377126371032e-05, | |
| "loss": 0.00023836212232708931, | |
| "step": 10925 | |
| }, | |
| { | |
| "epoch": 1.423556942277691, | |
| "grad_norm": 0.00044153345515951514, | |
| "learning_rate": 4.9477359477932604e-05, | |
| "loss": 0.00022806664928793908, | |
| "step": 10950 | |
| }, | |
| { | |
| "epoch": 1.4268070722828914, | |
| "grad_norm": 0.0008878413937054574, | |
| "learning_rate": 4.947090853951315e-05, | |
| "loss": 0.00020782511681318282, | |
| "step": 10975 | |
| }, | |
| { | |
| "epoch": 1.4300572022880915, | |
| "grad_norm": 0.0016059972112998366, | |
| "learning_rate": 4.9464418458770516e-05, | |
| "loss": 0.00020815758034586906, | |
| "step": 11000 | |
| }, | |
| { | |
| "epoch": 1.4333073322932917, | |
| "grad_norm": 0.0006840807036496699, | |
| "learning_rate": 4.945788924608589e-05, | |
| "loss": 0.00022403839975595474, | |
| "step": 11025 | |
| }, | |
| { | |
| "epoch": 1.436557462298492, | |
| "grad_norm": 0.0018522337777540088, | |
| "learning_rate": 4.9451320911903027e-05, | |
| "loss": 0.0002529165893793106, | |
| "step": 11050 | |
| }, | |
| { | |
| "epoch": 1.4398075923036922, | |
| "grad_norm": 0.0011392493033781648, | |
| "learning_rate": 4.9444713466728286e-05, | |
| "loss": 0.00023853225633502006, | |
| "step": 11075 | |
| }, | |
| { | |
| "epoch": 1.4430577223088923, | |
| "grad_norm": 0.0024292694870382547, | |
| "learning_rate": 4.943806692113058e-05, | |
| "loss": 0.00024014655500650405, | |
| "step": 11100 | |
| }, | |
| { | |
| "epoch": 1.4463078523140926, | |
| "grad_norm": 0.0007658280665054917, | |
| "learning_rate": 4.943138128574134e-05, | |
| "loss": 0.00023862946778535843, | |
| "step": 11125 | |
| }, | |
| { | |
| "epoch": 1.4495579823192928, | |
| "grad_norm": 0.0004227055178489536, | |
| "learning_rate": 4.942465657125455e-05, | |
| "loss": 0.0002230224758386612, | |
| "step": 11150 | |
| }, | |
| { | |
| "epoch": 1.452808112324493, | |
| "grad_norm": 0.003415195969864726, | |
| "learning_rate": 4.94178927884267e-05, | |
| "loss": 0.00022361867129802704, | |
| "step": 11175 | |
| }, | |
| { | |
| "epoch": 1.4560582423296933, | |
| "grad_norm": 0.0030447698663920164, | |
| "learning_rate": 4.941108994807676e-05, | |
| "loss": 0.00023454848676919936, | |
| "step": 11200 | |
| }, | |
| { | |
| "epoch": 1.4593083723348934, | |
| "grad_norm": 0.0012138801394030452, | |
| "learning_rate": 4.940424806108619e-05, | |
| "loss": 0.0002123577147722244, | |
| "step": 11225 | |
| }, | |
| { | |
| "epoch": 1.4625585023400935, | |
| "grad_norm": 0.0014790329150855541, | |
| "learning_rate": 4.939736713839889e-05, | |
| "loss": 0.0002538035623729229, | |
| "step": 11250 | |
| }, | |
| { | |
| "epoch": 1.465808632345294, | |
| "grad_norm": 0.0024601540062576532, | |
| "learning_rate": 4.93904471910212e-05, | |
| "loss": 0.00022838005796074867, | |
| "step": 11275 | |
| }, | |
| { | |
| "epoch": 1.469058762350494, | |
| "grad_norm": 0.001427294802851975, | |
| "learning_rate": 4.938348823002189e-05, | |
| "loss": 0.00022333266213536262, | |
| "step": 11300 | |
| }, | |
| { | |
| "epoch": 1.4723088923556942, | |
| "grad_norm": 0.0018155802972614765, | |
| "learning_rate": 4.937649026653215e-05, | |
| "loss": 0.0002779684402048588, | |
| "step": 11325 | |
| }, | |
| { | |
| "epoch": 1.4755590223608945, | |
| "grad_norm": 0.0015379923861473799, | |
| "learning_rate": 4.9369453311745516e-05, | |
| "loss": 0.00022361904382705688, | |
| "step": 11350 | |
| }, | |
| { | |
| "epoch": 1.4788091523660947, | |
| "grad_norm": 0.0015094184782356024, | |
| "learning_rate": 4.9362377376917936e-05, | |
| "loss": 0.0002118166908621788, | |
| "step": 11375 | |
| }, | |
| { | |
| "epoch": 1.4820592823712948, | |
| "grad_norm": 0.0019516360480338335, | |
| "learning_rate": 4.935526247336767e-05, | |
| "loss": 0.00021778607740998267, | |
| "step": 11400 | |
| }, | |
| { | |
| "epoch": 1.4853094123764952, | |
| "grad_norm": 0.000928438501432538, | |
| "learning_rate": 4.934810861247533e-05, | |
| "loss": 0.00020529730245471, | |
| "step": 11425 | |
| }, | |
| { | |
| "epoch": 1.4885595423816953, | |
| "grad_norm": 0.001177873113192618, | |
| "learning_rate": 4.934091580568384e-05, | |
| "loss": 0.00020764494314789772, | |
| "step": 11450 | |
| }, | |
| { | |
| "epoch": 1.4918096723868954, | |
| "grad_norm": 0.0008443294209428132, | |
| "learning_rate": 4.933368406449843e-05, | |
| "loss": 0.0002065547928214073, | |
| "step": 11475 | |
| }, | |
| { | |
| "epoch": 1.4950598023920958, | |
| "grad_norm": 0.0018499024445191026, | |
| "learning_rate": 4.932641340048658e-05, | |
| "loss": 0.0002670598588883877, | |
| "step": 11500 | |
| }, | |
| { | |
| "epoch": 1.498309932397296, | |
| "grad_norm": 0.001366683398373425, | |
| "learning_rate": 4.931910382527806e-05, | |
| "loss": 0.0002089797891676426, | |
| "step": 11525 | |
| }, | |
| { | |
| "epoch": 1.501560062402496, | |
| "grad_norm": 0.0014287630328908563, | |
| "learning_rate": 4.931175535056486e-05, | |
| "loss": 0.00023185288533568384, | |
| "step": 11550 | |
| }, | |
| { | |
| "epoch": 1.5048101924076964, | |
| "grad_norm": 0.0011994066881015897, | |
| "learning_rate": 4.9304367988101194e-05, | |
| "loss": 0.00022137219086289405, | |
| "step": 11575 | |
| }, | |
| { | |
| "epoch": 1.5080603224128966, | |
| "grad_norm": 0.001302711432799697, | |
| "learning_rate": 4.9296941749703486e-05, | |
| "loss": 0.00021828755736351014, | |
| "step": 11600 | |
| }, | |
| { | |
| "epoch": 1.5113104524180967, | |
| "grad_norm": 0.0009471996454522014, | |
| "learning_rate": 4.9289476647250344e-05, | |
| "loss": 0.0002089974656701088, | |
| "step": 11625 | |
| }, | |
| { | |
| "epoch": 1.514560582423297, | |
| "grad_norm": 0.0025708491448312998, | |
| "learning_rate": 4.928197269268252e-05, | |
| "loss": 0.00020671421661973, | |
| "step": 11650 | |
| }, | |
| { | |
| "epoch": 1.5178107124284972, | |
| "grad_norm": 0.0008076655794866383, | |
| "learning_rate": 4.927442989800294e-05, | |
| "loss": 0.0002104758471250534, | |
| "step": 11675 | |
| }, | |
| { | |
| "epoch": 1.5210608424336973, | |
| "grad_norm": 0.0009024995379149914, | |
| "learning_rate": 4.926684827527665e-05, | |
| "loss": 0.00019747575744986535, | |
| "step": 11700 | |
| }, | |
| { | |
| "epoch": 1.5243109724388977, | |
| "grad_norm": 0.001300012576393783, | |
| "learning_rate": 4.925922783663079e-05, | |
| "loss": 0.0002422116696834564, | |
| "step": 11725 | |
| }, | |
| { | |
| "epoch": 1.5275611024440978, | |
| "grad_norm": 0.0009546473156660795, | |
| "learning_rate": 4.925156859425459e-05, | |
| "loss": 0.00019759729504585267, | |
| "step": 11750 | |
| }, | |
| { | |
| "epoch": 1.530811232449298, | |
| "grad_norm": 0.001021179836243391, | |
| "learning_rate": 4.924387056039937e-05, | |
| "loss": 0.0002173416130244732, | |
| "step": 11775 | |
| }, | |
| { | |
| "epoch": 1.5340613624544983, | |
| "grad_norm": 0.0015582593623548746, | |
| "learning_rate": 4.9236133747378475e-05, | |
| "loss": 0.000237205158919096, | |
| "step": 11800 | |
| }, | |
| { | |
| "epoch": 1.5373114924596982, | |
| "grad_norm": 0.002049966948106885, | |
| "learning_rate": 4.922835816756728e-05, | |
| "loss": 0.00021319374442100526, | |
| "step": 11825 | |
| }, | |
| { | |
| "epoch": 1.5405616224648986, | |
| "grad_norm": 0.0014623207971453667, | |
| "learning_rate": 4.922054383340319e-05, | |
| "loss": 0.00022468648850917816, | |
| "step": 11850 | |
| }, | |
| { | |
| "epoch": 1.543811752470099, | |
| "grad_norm": 0.0010616122744977474, | |
| "learning_rate": 4.9212690757385565e-05, | |
| "loss": 0.0002278902754187584, | |
| "step": 11875 | |
| }, | |
| { | |
| "epoch": 1.5470618824752989, | |
| "grad_norm": 0.0014538065297529101, | |
| "learning_rate": 4.920479895207577e-05, | |
| "loss": 0.00022692535072565078, | |
| "step": 11900 | |
| }, | |
| { | |
| "epoch": 1.5503120124804992, | |
| "grad_norm": 0.0010035118320956826, | |
| "learning_rate": 4.919686843009711e-05, | |
| "loss": 0.0002275378815829754, | |
| "step": 11925 | |
| }, | |
| { | |
| "epoch": 1.5535621424856996, | |
| "grad_norm": 0.0028484691865742207, | |
| "learning_rate": 4.9188899204134794e-05, | |
| "loss": 0.00024725059047341346, | |
| "step": 11950 | |
| }, | |
| { | |
| "epoch": 1.5568122724908995, | |
| "grad_norm": 0.0010700999991968274, | |
| "learning_rate": 4.918089128693597e-05, | |
| "loss": 0.00023311510682106018, | |
| "step": 11975 | |
| }, | |
| { | |
| "epoch": 1.5600624024960998, | |
| "grad_norm": 0.0013243206776678562, | |
| "learning_rate": 4.917284469130966e-05, | |
| "loss": 0.00020961858332157134, | |
| "step": 12000 | |
| }, | |
| { | |
| "epoch": 1.5633125325013002, | |
| "grad_norm": 0.0012829856714233756, | |
| "learning_rate": 4.916475943012676e-05, | |
| "loss": 0.0002158079296350479, | |
| "step": 12025 | |
| }, | |
| { | |
| "epoch": 1.5665626625065001, | |
| "grad_norm": 0.00196343963034451, | |
| "learning_rate": 4.915663551632002e-05, | |
| "loss": 0.0002173725888133049, | |
| "step": 12050 | |
| }, | |
| { | |
| "epoch": 1.5698127925117005, | |
| "grad_norm": 0.0017143343575298786, | |
| "learning_rate": 4.914847296288398e-05, | |
| "loss": 0.0002406243234872818, | |
| "step": 12075 | |
| }, | |
| { | |
| "epoch": 1.5730629225169008, | |
| "grad_norm": 0.0016624074196442962, | |
| "learning_rate": 4.9140271782875035e-05, | |
| "loss": 0.0002361990697681904, | |
| "step": 12100 | |
| }, | |
| { | |
| "epoch": 1.5763130525221007, | |
| "grad_norm": 0.0007804373744875193, | |
| "learning_rate": 4.913203198941134e-05, | |
| "loss": 0.0002275773510336876, | |
| "step": 12125 | |
| }, | |
| { | |
| "epoch": 1.579563182527301, | |
| "grad_norm": 0.0023541946429759264, | |
| "learning_rate": 4.912375359567281e-05, | |
| "loss": 0.0001959991455078125, | |
| "step": 12150 | |
| }, | |
| { | |
| "epoch": 1.5828133125325015, | |
| "grad_norm": 0.0010233235079795122, | |
| "learning_rate": 4.9115436614901114e-05, | |
| "loss": 0.00018739113584160806, | |
| "step": 12175 | |
| }, | |
| { | |
| "epoch": 1.5860634425377014, | |
| "grad_norm": 0.0016051802085712552, | |
| "learning_rate": 4.9107081060399627e-05, | |
| "loss": 0.0002133057452738285, | |
| "step": 12200 | |
| }, | |
| { | |
| "epoch": 1.5893135725429017, | |
| "grad_norm": 0.0013718517730012536, | |
| "learning_rate": 4.909868694553346e-05, | |
| "loss": 0.00021465757861733437, | |
| "step": 12225 | |
| }, | |
| { | |
| "epoch": 1.592563702548102, | |
| "grad_norm": 0.0007805069908499718, | |
| "learning_rate": 4.909025428372936e-05, | |
| "loss": 0.00022665571421384812, | |
| "step": 12250 | |
| }, | |
| { | |
| "epoch": 1.595813832553302, | |
| "grad_norm": 0.0018547538202255964, | |
| "learning_rate": 4.908178308847575e-05, | |
| "loss": 0.00022225631400942802, | |
| "step": 12275 | |
| }, | |
| { | |
| "epoch": 1.5990639625585024, | |
| "grad_norm": 0.001349901081994176, | |
| "learning_rate": 4.907327337332271e-05, | |
| "loss": 0.00018927998840808868, | |
| "step": 12300 | |
| }, | |
| { | |
| "epoch": 1.6023140925637025, | |
| "grad_norm": 0.000917712808586657, | |
| "learning_rate": 4.906472515188188e-05, | |
| "loss": 0.00020415421575307847, | |
| "step": 12325 | |
| }, | |
| { | |
| "epoch": 1.6055642225689026, | |
| "grad_norm": 0.001299075549468398, | |
| "learning_rate": 4.905613843782655e-05, | |
| "loss": 0.0002180762030184269, | |
| "step": 12350 | |
| }, | |
| { | |
| "epoch": 1.608814352574103, | |
| "grad_norm": 0.0011308664688840508, | |
| "learning_rate": 4.904751324489155e-05, | |
| "loss": 0.00020294418558478356, | |
| "step": 12375 | |
| }, | |
| { | |
| "epoch": 1.6120644825793031, | |
| "grad_norm": 0.0010383251355960965, | |
| "learning_rate": 4.903884958687327e-05, | |
| "loss": 0.00020127050578594208, | |
| "step": 12400 | |
| }, | |
| { | |
| "epoch": 1.6153146125845033, | |
| "grad_norm": 0.0007700969581492245, | |
| "learning_rate": 4.903014747762961e-05, | |
| "loss": 0.00022019905969500542, | |
| "step": 12425 | |
| }, | |
| { | |
| "epoch": 1.6185647425897036, | |
| "grad_norm": 0.0015845868038013577, | |
| "learning_rate": 4.902140693108e-05, | |
| "loss": 0.00023882202804088594, | |
| "step": 12450 | |
| }, | |
| { | |
| "epoch": 1.6218148725949038, | |
| "grad_norm": 0.002057379810139537, | |
| "learning_rate": 4.901262796120534e-05, | |
| "loss": 0.00021504990756511687, | |
| "step": 12475 | |
| }, | |
| { | |
| "epoch": 1.625065002600104, | |
| "grad_norm": 0.0011629246873781085, | |
| "learning_rate": 4.900381058204798e-05, | |
| "loss": 0.0002195056900382042, | |
| "step": 12500 | |
| }, | |
| { | |
| "epoch": 1.6283151326053042, | |
| "grad_norm": 0.0008853785693645477, | |
| "learning_rate": 4.8994954807711715e-05, | |
| "loss": 0.0002198290452361107, | |
| "step": 12525 | |
| }, | |
| { | |
| "epoch": 1.6315652626105044, | |
| "grad_norm": 0.0023640524595975876, | |
| "learning_rate": 4.898606065236176e-05, | |
| "loss": 0.00022358374670147896, | |
| "step": 12550 | |
| }, | |
| { | |
| "epoch": 1.6348153926157045, | |
| "grad_norm": 0.0012634290615096688, | |
| "learning_rate": 4.897712813022473e-05, | |
| "loss": 0.0002201900817453861, | |
| "step": 12575 | |
| }, | |
| { | |
| "epoch": 1.6380655226209049, | |
| "grad_norm": 0.001097681699320674, | |
| "learning_rate": 4.896815725558857e-05, | |
| "loss": 0.00021714471280574797, | |
| "step": 12600 | |
| }, | |
| { | |
| "epoch": 1.641315652626105, | |
| "grad_norm": 0.0007831507246010005, | |
| "learning_rate": 4.895914804280262e-05, | |
| "loss": 0.0002441369742155075, | |
| "step": 12625 | |
| }, | |
| { | |
| "epoch": 1.6445657826313052, | |
| "grad_norm": 0.0014596626861020923, | |
| "learning_rate": 4.895010050627752e-05, | |
| "loss": 0.00022420192137360574, | |
| "step": 12650 | |
| }, | |
| { | |
| "epoch": 1.6478159126365055, | |
| "grad_norm": 0.0009162931237369776, | |
| "learning_rate": 4.8941014660485206e-05, | |
| "loss": 0.0002115178294479847, | |
| "step": 12675 | |
| }, | |
| { | |
| "epoch": 1.6510660426417056, | |
| "grad_norm": 0.001233368762768805, | |
| "learning_rate": 4.893189051995891e-05, | |
| "loss": 0.00020989146083593368, | |
| "step": 12700 | |
| }, | |
| { | |
| "epoch": 1.6543161726469058, | |
| "grad_norm": 0.0005623113829642534, | |
| "learning_rate": 4.892272809929309e-05, | |
| "loss": 0.00022066382691264154, | |
| "step": 12725 | |
| }, | |
| { | |
| "epoch": 1.6575663026521061, | |
| "grad_norm": 0.0008610347867943347, | |
| "learning_rate": 4.891352741314347e-05, | |
| "loss": 0.00021332105621695518, | |
| "step": 12750 | |
| }, | |
| { | |
| "epoch": 1.6608164326573063, | |
| "grad_norm": 0.0006132300477474928, | |
| "learning_rate": 4.8904288476226954e-05, | |
| "loss": 0.0002009551413357258, | |
| "step": 12775 | |
| }, | |
| { | |
| "epoch": 1.6640665626625064, | |
| "grad_norm": 0.0022478275932371616, | |
| "learning_rate": 4.889501130332163e-05, | |
| "loss": 0.00020834658294916153, | |
| "step": 12800 | |
| }, | |
| { | |
| "epoch": 1.6673166926677068, | |
| "grad_norm": 0.0013704983284696937, | |
| "learning_rate": 4.8885695909266774e-05, | |
| "loss": 0.0002107241377234459, | |
| "step": 12825 | |
| }, | |
| { | |
| "epoch": 1.670566822672907, | |
| "grad_norm": 0.0012594923609867692, | |
| "learning_rate": 4.887634230896276e-05, | |
| "loss": 0.00020520130172371864, | |
| "step": 12850 | |
| }, | |
| { | |
| "epoch": 1.673816952678107, | |
| "grad_norm": 0.0009163892827928066, | |
| "learning_rate": 4.886695051737112e-05, | |
| "loss": 0.00018248358741402627, | |
| "step": 12875 | |
| }, | |
| { | |
| "epoch": 1.6770670826833074, | |
| "grad_norm": 0.0013087759725749493, | |
| "learning_rate": 4.885752054951442e-05, | |
| "loss": 0.0001804506406188011, | |
| "step": 12900 | |
| }, | |
| { | |
| "epoch": 1.6803172126885075, | |
| "grad_norm": 0.0009314049384556711, | |
| "learning_rate": 4.884805242047633e-05, | |
| "loss": 0.00020271694287657738, | |
| "step": 12925 | |
| }, | |
| { | |
| "epoch": 1.6835673426937077, | |
| "grad_norm": 0.001452924800105393, | |
| "learning_rate": 4.8838546145401556e-05, | |
| "loss": 0.0001986694522202015, | |
| "step": 12950 | |
| }, | |
| { | |
| "epoch": 1.686817472698908, | |
| "grad_norm": 0.0009039230062626302, | |
| "learning_rate": 4.882900173949581e-05, | |
| "loss": 0.0001967090740799904, | |
| "step": 12975 | |
| }, | |
| { | |
| "epoch": 1.6900676027041082, | |
| "grad_norm": 0.0022098368499428034, | |
| "learning_rate": 4.88194192180258e-05, | |
| "loss": 0.0002104349248111248, | |
| "step": 13000 | |
| }, | |
| { | |
| "epoch": 1.6933177327093083, | |
| "grad_norm": 0.0017814264865592122, | |
| "learning_rate": 4.88097985963192e-05, | |
| "loss": 0.00020009895786643028, | |
| "step": 13025 | |
| }, | |
| { | |
| "epoch": 1.6965678627145087, | |
| "grad_norm": 0.0009897558484226465, | |
| "learning_rate": 4.8800139889764635e-05, | |
| "loss": 0.00019421691074967385, | |
| "step": 13050 | |
| }, | |
| { | |
| "epoch": 1.6998179927197088, | |
| "grad_norm": 0.0008877114742062986, | |
| "learning_rate": 4.879044311381163e-05, | |
| "loss": 0.00021792301908135414, | |
| "step": 13075 | |
| }, | |
| { | |
| "epoch": 1.703068122724909, | |
| "grad_norm": 0.000991069246083498, | |
| "learning_rate": 4.8780708283970635e-05, | |
| "loss": 0.00021280685439705848, | |
| "step": 13100 | |
| }, | |
| { | |
| "epoch": 1.7063182527301093, | |
| "grad_norm": 0.0018834732472896576, | |
| "learning_rate": 4.877093541581293e-05, | |
| "loss": 0.00020899130031466485, | |
| "step": 13125 | |
| }, | |
| { | |
| "epoch": 1.7095683827353094, | |
| "grad_norm": 0.0024460970889776945, | |
| "learning_rate": 4.876112452497067e-05, | |
| "loss": 0.00021215084940195083, | |
| "step": 13150 | |
| }, | |
| { | |
| "epoch": 1.7128185127405096, | |
| "grad_norm": 0.0025619217194616795, | |
| "learning_rate": 4.875127562713682e-05, | |
| "loss": 0.00022624718025326729, | |
| "step": 13175 | |
| }, | |
| { | |
| "epoch": 1.71606864274571, | |
| "grad_norm": 0.00162825349252671, | |
| "learning_rate": 4.874138873806513e-05, | |
| "loss": 0.00021882947534322737, | |
| "step": 13200 | |
| }, | |
| { | |
| "epoch": 1.71931877275091, | |
| "grad_norm": 0.0006245311815291643, | |
| "learning_rate": 4.873146387357012e-05, | |
| "loss": 0.00019531719386577605, | |
| "step": 13225 | |
| }, | |
| { | |
| "epoch": 1.7225689027561102, | |
| "grad_norm": 0.0006454737158492208, | |
| "learning_rate": 4.8721501049527074e-05, | |
| "loss": 0.00019291777163743973, | |
| "step": 13250 | |
| }, | |
| { | |
| "epoch": 1.7258190327613105, | |
| "grad_norm": 0.0010474052978679538, | |
| "learning_rate": 4.871150028187197e-05, | |
| "loss": 0.00019108649343252182, | |
| "step": 13275 | |
| }, | |
| { | |
| "epoch": 1.7290691627665107, | |
| "grad_norm": 0.0005790158757008612, | |
| "learning_rate": 4.8701461586601495e-05, | |
| "loss": 0.00019627740606665612, | |
| "step": 13300 | |
| }, | |
| { | |
| "epoch": 1.7323192927717108, | |
| "grad_norm": 0.0019599865190684795, | |
| "learning_rate": 4.869138497977298e-05, | |
| "loss": 0.0002209186926484108, | |
| "step": 13325 | |
| }, | |
| { | |
| "epoch": 1.7355694227769112, | |
| "grad_norm": 0.0012518159346655011, | |
| "learning_rate": 4.8681270477504425e-05, | |
| "loss": 0.00022938210517168045, | |
| "step": 13350 | |
| }, | |
| { | |
| "epoch": 1.7388195527821113, | |
| "grad_norm": 0.0008453086484223604, | |
| "learning_rate": 4.8671118095974435e-05, | |
| "loss": 0.00020086592063307763, | |
| "step": 13375 | |
| }, | |
| { | |
| "epoch": 1.7420696827873114, | |
| "grad_norm": 0.0015744048869237304, | |
| "learning_rate": 4.86609278514222e-05, | |
| "loss": 0.00023271137848496437, | |
| "step": 13400 | |
| }, | |
| { | |
| "epoch": 1.7453198127925118, | |
| "grad_norm": 0.0009528041118755937, | |
| "learning_rate": 4.865069976014747e-05, | |
| "loss": 0.0002470199763774872, | |
| "step": 13425 | |
| }, | |
| { | |
| "epoch": 1.748569942797712, | |
| "grad_norm": 0.00111283827573061, | |
| "learning_rate": 4.864043383851054e-05, | |
| "loss": 0.00022434709593653678, | |
| "step": 13450 | |
| }, | |
| { | |
| "epoch": 1.751820072802912, | |
| "grad_norm": 0.0014392537996172905, | |
| "learning_rate": 4.8630130102932226e-05, | |
| "loss": 0.00019503209739923477, | |
| "step": 13475 | |
| }, | |
| { | |
| "epoch": 1.7550702028081124, | |
| "grad_norm": 0.0024195711594074965, | |
| "learning_rate": 4.86197885698938e-05, | |
| "loss": 0.00020250892266631125, | |
| "step": 13500 | |
| }, | |
| { | |
| "epoch": 1.7583203328133126, | |
| "grad_norm": 0.0018107002833858132, | |
| "learning_rate": 4.860940925593703e-05, | |
| "loss": 0.00018528325483202935, | |
| "step": 13525 | |
| }, | |
| { | |
| "epoch": 1.7615704628185127, | |
| "grad_norm": 0.00047586948494426906, | |
| "learning_rate": 4.859899217766408e-05, | |
| "loss": 0.0002012966200709343, | |
| "step": 13550 | |
| }, | |
| { | |
| "epoch": 1.764820592823713, | |
| "grad_norm": 0.0008106463355943561, | |
| "learning_rate": 4.858853735173756e-05, | |
| "loss": 0.00022149544209241868, | |
| "step": 13575 | |
| }, | |
| { | |
| "epoch": 1.7680707228289132, | |
| "grad_norm": 0.0015900592552497983, | |
| "learning_rate": 4.857804479488042e-05, | |
| "loss": 0.0002231018617749214, | |
| "step": 13600 | |
| }, | |
| { | |
| "epoch": 1.7713208528341133, | |
| "grad_norm": 0.0013972468441352248, | |
| "learning_rate": 4.856751452387597e-05, | |
| "loss": 0.0001727997325360775, | |
| "step": 13625 | |
| }, | |
| { | |
| "epoch": 1.7745709828393137, | |
| "grad_norm": 0.003206509631127119, | |
| "learning_rate": 4.8556946555567875e-05, | |
| "loss": 0.00020560242235660554, | |
| "step": 13650 | |
| }, | |
| { | |
| "epoch": 1.7778211128445138, | |
| "grad_norm": 0.0008185068727470934, | |
| "learning_rate": 4.854634090686005e-05, | |
| "loss": 0.00019721709191799163, | |
| "step": 13675 | |
| }, | |
| { | |
| "epoch": 1.781071242849714, | |
| "grad_norm": 0.001312661450356245, | |
| "learning_rate": 4.853569759471674e-05, | |
| "loss": 0.00018779512494802476, | |
| "step": 13700 | |
| }, | |
| { | |
| "epoch": 1.7843213728549143, | |
| "grad_norm": 0.0006430719513446093, | |
| "learning_rate": 4.85250166361624e-05, | |
| "loss": 0.00020717784762382506, | |
| "step": 13725 | |
| }, | |
| { | |
| "epoch": 1.7875715028601145, | |
| "grad_norm": 0.002790760016068816, | |
| "learning_rate": 4.851429804828168e-05, | |
| "loss": 0.00019336583092808724, | |
| "step": 13750 | |
| }, | |
| { | |
| "epoch": 1.7908216328653146, | |
| "grad_norm": 0.0013842202024534345, | |
| "learning_rate": 4.8503541848219464e-05, | |
| "loss": 0.00017480608075857163, | |
| "step": 13775 | |
| }, | |
| { | |
| "epoch": 1.794071762870515, | |
| "grad_norm": 0.0015860736602917314, | |
| "learning_rate": 4.8492748053180776e-05, | |
| "loss": 0.0001934964582324028, | |
| "step": 13800 | |
| }, | |
| { | |
| "epoch": 1.794071762870515, | |
| "eval_loss": 0.0007873945869505405, | |
| "eval_runtime": 9.5205, | |
| "eval_samples_per_second": 525.18, | |
| "eval_steps_per_second": 4.201, | |
| "step": 13800 | |
| }, | |
| { | |
| "epoch": 1.794071762870515, | |
| "eval_holdout_avg_strict_f1": 0.07671023568489087, | |
| "eval_holdout_enron_spam_bvk_strict_f1": 0.0, | |
| "eval_holdout_enron_spam_setfit_strict_f1": 0.0, | |
| "eval_holdout_gliner2_pii_ptbr_reward_split_strict_f1": 0.3106818443141313, | |
| "eval_holdout_nemotron_pii_strict_f1": 0.2033773260103287, | |
| "eval_holdout_open_pii_masking_500k_strict_f1": 0.17042764715164904, | |
| "eval_holdout_phishing_darkknight_strict_f1": 0.0, | |
| "eval_holdout_phishing_zefang_strict_f1": 0.0, | |
| "eval_holdout_pii_masking_400k_strict_f1": 0.15932577505769038, | |
| "eval_holdout_sms_spam_multilingual_strict_f1": 0.0, | |
| "eval_holdout_spam_messages_mshenoda_strict_f1": 0.0, | |
| "eval_holdout_spamassassin_strict_f1": 0.0, | |
| "step": 13800 | |
| }, | |
| { | |
| "epoch": 1.7973218928757149, | |
| "grad_norm": 0.0019391707610338926, | |
| "learning_rate": 4.848191668043078e-05, | |
| "loss": 0.0001878507435321808, | |
| "step": 13825 | |
| }, | |
| { | |
| "epoch": 1.8005720228809152, | |
| "grad_norm": 0.0014676563441753387, | |
| "learning_rate": 4.847104774729474e-05, | |
| "loss": 0.00021142667159438134, | |
| "step": 13850 | |
| }, | |
| { | |
| "epoch": 1.8038221528861156, | |
| "grad_norm": 0.0032201732974499464, | |
| "learning_rate": 4.8460141271158014e-05, | |
| "loss": 0.0001980452425777912, | |
| "step": 13875 | |
| }, | |
| { | |
| "epoch": 1.8070722828913155, | |
| "grad_norm": 0.0010093223536387086, | |
| "learning_rate": 4.8449197269466e-05, | |
| "loss": 0.00018527789041399955, | |
| "step": 13900 | |
| }, | |
| { | |
| "epoch": 1.8103224128965159, | |
| "grad_norm": 0.002089881105348468, | |
| "learning_rate": 4.843821575972411e-05, | |
| "loss": 0.00018696235492825507, | |
| "step": 13925 | |
| }, | |
| { | |
| "epoch": 1.8135725429017162, | |
| "grad_norm": 0.0012031790101900697, | |
| "learning_rate": 4.8427196759497796e-05, | |
| "loss": 0.00019349925220012664, | |
| "step": 13950 | |
| }, | |
| { | |
| "epoch": 1.8168226729069161, | |
| "grad_norm": 0.0032598827965557575, | |
| "learning_rate": 4.8416140286412416e-05, | |
| "loss": 0.0001790124922990799, | |
| "step": 13975 | |
| }, | |
| { | |
| "epoch": 1.8200728029121165, | |
| "grad_norm": 0.0011320498306304216, | |
| "learning_rate": 4.840504635815331e-05, | |
| "loss": 0.00020609425380825995, | |
| "step": 14000 | |
| }, | |
| { | |
| "epoch": 1.8233229329173168, | |
| "grad_norm": 0.0021722533274441957, | |
| "learning_rate": 4.8393914992465715e-05, | |
| "loss": 0.00017540406435728072, | |
| "step": 14025 | |
| }, | |
| { | |
| "epoch": 1.8265730629225168, | |
| "grad_norm": 0.0009918785654008389, | |
| "learning_rate": 4.838274620715475e-05, | |
| "loss": 0.00018825355917215347, | |
| "step": 14050 | |
| }, | |
| { | |
| "epoch": 1.8298231929277171, | |
| "grad_norm": 0.0021330395247787237, | |
| "learning_rate": 4.8371540020085406e-05, | |
| "loss": 0.0001849439740180969, | |
| "step": 14075 | |
| }, | |
| { | |
| "epoch": 1.8330733229329175, | |
| "grad_norm": 0.001202671555802226, | |
| "learning_rate": 4.836029644918246e-05, | |
| "loss": 0.00018819918856024742, | |
| "step": 14100 | |
| }, | |
| { | |
| "epoch": 1.8363234529381174, | |
| "grad_norm": 0.0035825998056679964, | |
| "learning_rate": 4.834901551243054e-05, | |
| "loss": 0.00020158516243100168, | |
| "step": 14125 | |
| }, | |
| { | |
| "epoch": 1.8395735829433177, | |
| "grad_norm": 0.005913496483117342, | |
| "learning_rate": 4.833769722787399e-05, | |
| "loss": 0.00019377293065190316, | |
| "step": 14150 | |
| }, | |
| { | |
| "epoch": 1.842823712948518, | |
| "grad_norm": 0.0009982505580410361, | |
| "learning_rate": 4.832634161361693e-05, | |
| "loss": 0.0001952342875301838, | |
| "step": 14175 | |
| }, | |
| { | |
| "epoch": 1.846073842953718, | |
| "grad_norm": 0.0014910399913787842, | |
| "learning_rate": 4.831494868782318e-05, | |
| "loss": 0.0002068406343460083, | |
| "step": 14200 | |
| }, | |
| { | |
| "epoch": 1.8493239729589184, | |
| "grad_norm": 0.0053688944317400455, | |
| "learning_rate": 4.830351846871623e-05, | |
| "loss": 0.00020632198080420495, | |
| "step": 14225 | |
| }, | |
| { | |
| "epoch": 1.8525741029641187, | |
| "grad_norm": 0.003236903343349695, | |
| "learning_rate": 4.829205097457923e-05, | |
| "loss": 0.00021189013496041297, | |
| "step": 14250 | |
| }, | |
| { | |
| "epoch": 1.8558242329693186, | |
| "grad_norm": 0.0021175884176045656, | |
| "learning_rate": 4.8280546223754967e-05, | |
| "loss": 0.00021266382187604905, | |
| "step": 14275 | |
| }, | |
| { | |
| "epoch": 1.859074362974519, | |
| "grad_norm": 0.0016959202475845814, | |
| "learning_rate": 4.8269004234645805e-05, | |
| "loss": 0.00016569169238209725, | |
| "step": 14300 | |
| }, | |
| { | |
| "epoch": 1.8623244929797191, | |
| "grad_norm": 0.0011466607684269547, | |
| "learning_rate": 4.825742502571366e-05, | |
| "loss": 0.00017714653164148332, | |
| "step": 14325 | |
| }, | |
| { | |
| "epoch": 1.8655746229849193, | |
| "grad_norm": 0.002087001921609044, | |
| "learning_rate": 4.824580861548004e-05, | |
| "loss": 0.0001798272505402565, | |
| "step": 14350 | |
| }, | |
| { | |
| "epoch": 1.8688247529901196, | |
| "grad_norm": 0.0013011404080316424, | |
| "learning_rate": 4.823415502252587e-05, | |
| "loss": 0.00019987838342785835, | |
| "step": 14375 | |
| }, | |
| { | |
| "epoch": 1.8720748829953198, | |
| "grad_norm": 0.003103255992755294, | |
| "learning_rate": 4.8222464265491624e-05, | |
| "loss": 0.00018162962049245835, | |
| "step": 14400 | |
| }, | |
| { | |
| "epoch": 1.87532501300052, | |
| "grad_norm": 0.0026980354450643063, | |
| "learning_rate": 4.821073636307719e-05, | |
| "loss": 0.0001928180642426014, | |
| "step": 14425 | |
| }, | |
| { | |
| "epoch": 1.8785751430057203, | |
| "grad_norm": 0.0022120249923318624, | |
| "learning_rate": 4.819897133404189e-05, | |
| "loss": 0.00022065060213208198, | |
| "step": 14450 | |
| }, | |
| { | |
| "epoch": 1.8818252730109204, | |
| "grad_norm": 0.0012642544461414218, | |
| "learning_rate": 4.8187169197204395e-05, | |
| "loss": 0.0001877201721072197, | |
| "step": 14475 | |
| }, | |
| { | |
| "epoch": 1.8850754030161205, | |
| "grad_norm": 0.0012029558420181274, | |
| "learning_rate": 4.8175329971442774e-05, | |
| "loss": 0.0001797371543943882, | |
| "step": 14500 | |
| }, | |
| { | |
| "epoch": 1.888325533021321, | |
| "grad_norm": 0.006576585583388805, | |
| "learning_rate": 4.8163453675694396e-05, | |
| "loss": 0.00020043499767780304, | |
| "step": 14525 | |
| }, | |
| { | |
| "epoch": 1.891575663026521, | |
| "grad_norm": 0.001970443408936262, | |
| "learning_rate": 4.8151540328955935e-05, | |
| "loss": 0.0001659386046230793, | |
| "step": 14550 | |
| }, | |
| { | |
| "epoch": 1.8948257930317212, | |
| "grad_norm": 0.0025421991012990475, | |
| "learning_rate": 4.813958995028333e-05, | |
| "loss": 0.00019923953339457513, | |
| "step": 14575 | |
| }, | |
| { | |
| "epoch": 1.8980759230369215, | |
| "grad_norm": 0.001944137504324317, | |
| "learning_rate": 4.812760255879175e-05, | |
| "loss": 0.00019897514954209328, | |
| "step": 14600 | |
| }, | |
| { | |
| "epoch": 1.9013260530421217, | |
| "grad_norm": 0.0016694455407559872, | |
| "learning_rate": 4.811557817365556e-05, | |
| "loss": 0.0002129945158958435, | |
| "step": 14625 | |
| }, | |
| { | |
| "epoch": 1.9045761830473218, | |
| "grad_norm": 0.0018652010476216674, | |
| "learning_rate": 4.810351681410832e-05, | |
| "loss": 0.00018253499642014503, | |
| "step": 14650 | |
| }, | |
| { | |
| "epoch": 1.9078263130525221, | |
| "grad_norm": 0.002443675883114338, | |
| "learning_rate": 4.8091418499442705e-05, | |
| "loss": 0.00018208036199212073, | |
| "step": 14675 | |
| }, | |
| { | |
| "epoch": 1.9110764430577223, | |
| "grad_norm": 0.001355400076135993, | |
| "learning_rate": 4.8079283249010535e-05, | |
| "loss": 0.00018838070333003997, | |
| "step": 14700 | |
| }, | |
| { | |
| "epoch": 1.9143265730629224, | |
| "grad_norm": 0.0015903201419860125, | |
| "learning_rate": 4.806711108222267e-05, | |
| "loss": 0.0001955299824476242, | |
| "step": 14725 | |
| }, | |
| { | |
| "epoch": 1.9175767030681228, | |
| "grad_norm": 0.0031122886575758457, | |
| "learning_rate": 4.805490201854905e-05, | |
| "loss": 0.00020101873204112053, | |
| "step": 14750 | |
| }, | |
| { | |
| "epoch": 1.920826833073323, | |
| "grad_norm": 0.001963747898116708, | |
| "learning_rate": 4.8042656077518625e-05, | |
| "loss": 0.00018206274136900902, | |
| "step": 14775 | |
| }, | |
| { | |
| "epoch": 1.924076963078523, | |
| "grad_norm": 0.0031640648376196623, | |
| "learning_rate": 4.8030373278719335e-05, | |
| "loss": 0.0001737309992313385, | |
| "step": 14800 | |
| }, | |
| { | |
| "epoch": 1.9273270930837234, | |
| "grad_norm": 0.0023399752099066973, | |
| "learning_rate": 4.801805364179806e-05, | |
| "loss": 0.00018427133560180664, | |
| "step": 14825 | |
| }, | |
| { | |
| "epoch": 1.9305772230889235, | |
| "grad_norm": 0.001325368182733655, | |
| "learning_rate": 4.800569718646062e-05, | |
| "loss": 0.00016195224598050117, | |
| "step": 14850 | |
| }, | |
| { | |
| "epoch": 1.9338273530941237, | |
| "grad_norm": 0.0016849812818691134, | |
| "learning_rate": 4.7993303932471726e-05, | |
| "loss": 0.0001758340932428837, | |
| "step": 14875 | |
| }, | |
| { | |
| "epoch": 1.937077483099324, | |
| "grad_norm": 0.0011520645348355174, | |
| "learning_rate": 4.798087389965495e-05, | |
| "loss": 0.00018262149766087532, | |
| "step": 14900 | |
| }, | |
| { | |
| "epoch": 1.9403276131045242, | |
| "grad_norm": 0.0009939279407262802, | |
| "learning_rate": 4.7968407107892684e-05, | |
| "loss": 0.00017483439296483994, | |
| "step": 14925 | |
| }, | |
| { | |
| "epoch": 1.9435777431097243, | |
| "grad_norm": 0.0011320949997752905, | |
| "learning_rate": 4.795590357712613e-05, | |
| "loss": 0.00017233431339263916, | |
| "step": 14950 | |
| }, | |
| { | |
| "epoch": 1.9468278731149247, | |
| "grad_norm": 0.0012969585368409753, | |
| "learning_rate": 4.794336332735524e-05, | |
| "loss": 0.0001781269535422325, | |
| "step": 14975 | |
| }, | |
| { | |
| "epoch": 1.9500780031201248, | |
| "grad_norm": 0.00279130763374269, | |
| "learning_rate": 4.793078637863872e-05, | |
| "loss": 0.00018540842458605766, | |
| "step": 15000 | |
| }, | |
| { | |
| "epoch": 1.953328133125325, | |
| "grad_norm": 0.001861119526438415, | |
| "learning_rate": 4.791817275109396e-05, | |
| "loss": 0.00019540552049875258, | |
| "step": 15025 | |
| }, | |
| { | |
| "epoch": 1.9565782631305253, | |
| "grad_norm": 0.0033701809588819742, | |
| "learning_rate": 4.790552246489703e-05, | |
| "loss": 0.00016461025923490525, | |
| "step": 15050 | |
| }, | |
| { | |
| "epoch": 1.9598283931357254, | |
| "grad_norm": 0.0006286469288170338, | |
| "learning_rate": 4.789283554028264e-05, | |
| "loss": 0.00017568401992321014, | |
| "step": 15075 | |
| }, | |
| { | |
| "epoch": 1.9630785231409256, | |
| "grad_norm": 0.0012872155057266355, | |
| "learning_rate": 4.7880111997544094e-05, | |
| "loss": 0.00015418815426528453, | |
| "step": 15100 | |
| }, | |
| { | |
| "epoch": 1.966328653146126, | |
| "grad_norm": 0.001639127847738564, | |
| "learning_rate": 4.786735185703327e-05, | |
| "loss": 0.0001733824983239174, | |
| "step": 15125 | |
| }, | |
| { | |
| "epoch": 1.969578783151326, | |
| "grad_norm": 0.00825419183820486, | |
| "learning_rate": 4.785455513916059e-05, | |
| "loss": 0.00018014784902334213, | |
| "step": 15150 | |
| }, | |
| { | |
| "epoch": 1.9728289131565262, | |
| "grad_norm": 0.0016525533283129334, | |
| "learning_rate": 4.784172186439499e-05, | |
| "loss": 0.00016880841925740242, | |
| "step": 15175 | |
| }, | |
| { | |
| "epoch": 1.9760790431617266, | |
| "grad_norm": 0.0008869419689290226, | |
| "learning_rate": 4.7828852053263865e-05, | |
| "loss": 0.00016358742490410805, | |
| "step": 15200 | |
| }, | |
| { | |
| "epoch": 1.9793291731669267, | |
| "grad_norm": 0.0021625568624585867, | |
| "learning_rate": 4.781594572635306e-05, | |
| "loss": 0.00015240022912621498, | |
| "step": 15225 | |
| }, | |
| { | |
| "epoch": 1.9825793031721268, | |
| "grad_norm": 0.0030781549867242575, | |
| "learning_rate": 4.780300290430682e-05, | |
| "loss": 0.00015993243083357812, | |
| "step": 15250 | |
| }, | |
| { | |
| "epoch": 1.9858294331773272, | |
| "grad_norm": 0.0019852917175740004, | |
| "learning_rate": 4.779002360782779e-05, | |
| "loss": 0.0001564650423824787, | |
| "step": 15275 | |
| }, | |
| { | |
| "epoch": 1.9890795631825273, | |
| "grad_norm": 0.000985127524472773, | |
| "learning_rate": 4.777700785767694e-05, | |
| "loss": 0.00017142394557595253, | |
| "step": 15300 | |
| }, | |
| { | |
| "epoch": 1.9923296931877275, | |
| "grad_norm": 0.0015094815753400326, | |
| "learning_rate": 4.776395567467353e-05, | |
| "loss": 0.000170376468449831, | |
| "step": 15325 | |
| }, | |
| { | |
| "epoch": 1.9955798231929278, | |
| "grad_norm": 0.002530918922275305, | |
| "learning_rate": 4.775086707969513e-05, | |
| "loss": 0.00015980442985892297, | |
| "step": 15350 | |
| }, | |
| { | |
| "epoch": 1.998829953198128, | |
| "grad_norm": 0.001743559492751956, | |
| "learning_rate": 4.773774209367753e-05, | |
| "loss": 0.0001495914813131094, | |
| "step": 15375 | |
| }, | |
| { | |
| "epoch": 2.002080083203328, | |
| "grad_norm": 0.005321058444678783, | |
| "learning_rate": 4.7724580737614755e-05, | |
| "loss": 0.00016431858763098716, | |
| "step": 15400 | |
| }, | |
| { | |
| "epoch": 2.0053302132085284, | |
| "grad_norm": 0.0011911957990378141, | |
| "learning_rate": 4.7711383032558976e-05, | |
| "loss": 0.00016631942242383956, | |
| "step": 15425 | |
| }, | |
| { | |
| "epoch": 2.0085803432137284, | |
| "grad_norm": 0.0012982947519049048, | |
| "learning_rate": 4.769814899962052e-05, | |
| "loss": 0.00015491821803152562, | |
| "step": 15450 | |
| }, | |
| { | |
| "epoch": 2.0118304732189287, | |
| "grad_norm": 0.0007231764029711485, | |
| "learning_rate": 4.7684878659967816e-05, | |
| "loss": 0.00017552003264427186, | |
| "step": 15475 | |
| }, | |
| { | |
| "epoch": 2.015080603224129, | |
| "grad_norm": 0.002448186744004488, | |
| "learning_rate": 4.7671572034827375e-05, | |
| "loss": 0.00017927879467606546, | |
| "step": 15500 | |
| }, | |
| { | |
| "epoch": 2.018330733229329, | |
| "grad_norm": 0.002163167344406247, | |
| "learning_rate": 4.765822914548375e-05, | |
| "loss": 0.0001637950725853443, | |
| "step": 15525 | |
| }, | |
| { | |
| "epoch": 2.0215808632345293, | |
| "grad_norm": 0.0025511952117085457, | |
| "learning_rate": 4.764485001327949e-05, | |
| "loss": 0.0001653164066374302, | |
| "step": 15550 | |
| }, | |
| { | |
| "epoch": 2.0248309932397297, | |
| "grad_norm": 0.0014500441029667854, | |
| "learning_rate": 4.763143465961512e-05, | |
| "loss": 0.00016453903168439865, | |
| "step": 15575 | |
| }, | |
| { | |
| "epoch": 2.0280811232449296, | |
| "grad_norm": 0.0033016984816640615, | |
| "learning_rate": 4.761798310594911e-05, | |
| "loss": 0.0001722812093794346, | |
| "step": 15600 | |
| }, | |
| { | |
| "epoch": 2.03133125325013, | |
| "grad_norm": 0.0014526959275826812, | |
| "learning_rate": 4.760449537379782e-05, | |
| "loss": 0.00017294855788350106, | |
| "step": 15625 | |
| }, | |
| { | |
| "epoch": 2.0345813832553303, | |
| "grad_norm": 0.0018753077602013946, | |
| "learning_rate": 4.7590971484735495e-05, | |
| "loss": 0.0001641126349568367, | |
| "step": 15650 | |
| }, | |
| { | |
| "epoch": 2.0378315132605302, | |
| "grad_norm": 0.004806871060281992, | |
| "learning_rate": 4.757741146039419e-05, | |
| "loss": 0.00015752952545881272, | |
| "step": 15675 | |
| }, | |
| { | |
| "epoch": 2.0410816432657306, | |
| "grad_norm": 0.001489722984842956, | |
| "learning_rate": 4.756381532246379e-05, | |
| "loss": 0.0001614074781537056, | |
| "step": 15700 | |
| }, | |
| { | |
| "epoch": 2.044331773270931, | |
| "grad_norm": 0.001606178586371243, | |
| "learning_rate": 4.755018309269194e-05, | |
| "loss": 0.00015808012336492538, | |
| "step": 15725 | |
| }, | |
| { | |
| "epoch": 2.047581903276131, | |
| "grad_norm": 0.003261657664552331, | |
| "learning_rate": 4.753651479288398e-05, | |
| "loss": 0.0001637156493961811, | |
| "step": 15750 | |
| }, | |
| { | |
| "epoch": 2.0508320332813312, | |
| "grad_norm": 0.00272456300444901, | |
| "learning_rate": 4.7522810444903e-05, | |
| "loss": 0.00016159968450665474, | |
| "step": 15775 | |
| }, | |
| { | |
| "epoch": 2.0540821632865316, | |
| "grad_norm": 0.00374131603166461, | |
| "learning_rate": 4.75090700706697e-05, | |
| "loss": 0.00017268521711230278, | |
| "step": 15800 | |
| }, | |
| { | |
| "epoch": 2.0573322932917315, | |
| "grad_norm": 0.001509288907982409, | |
| "learning_rate": 4.749529369216246e-05, | |
| "loss": 0.0001449667662382126, | |
| "step": 15825 | |
| }, | |
| { | |
| "epoch": 2.060582423296932, | |
| "grad_norm": 0.001631575170904398, | |
| "learning_rate": 4.748148133141718e-05, | |
| "loss": 0.0001835348829627037, | |
| "step": 15850 | |
| }, | |
| { | |
| "epoch": 2.0638325533021322, | |
| "grad_norm": 0.0013842219486832619, | |
| "learning_rate": 4.746763301052739e-05, | |
| "loss": 0.00016309289261698723, | |
| "step": 15875 | |
| }, | |
| { | |
| "epoch": 2.067082683307332, | |
| "grad_norm": 0.0010653831996023655, | |
| "learning_rate": 4.7453748751644075e-05, | |
| "loss": 0.00016087114810943602, | |
| "step": 15900 | |
| }, | |
| { | |
| "epoch": 2.0703328133125325, | |
| "grad_norm": 0.0026510499883443117, | |
| "learning_rate": 4.7439828576975766e-05, | |
| "loss": 0.00014180582948029042, | |
| "step": 15925 | |
| }, | |
| { | |
| "epoch": 2.073582943317733, | |
| "grad_norm": 0.0007459137123078108, | |
| "learning_rate": 4.7425872508788384e-05, | |
| "loss": 0.0001659255288541317, | |
| "step": 15950 | |
| }, | |
| { | |
| "epoch": 2.0768330733229328, | |
| "grad_norm": 0.0011213262332603335, | |
| "learning_rate": 4.74118805694053e-05, | |
| "loss": 0.00015620710328221322, | |
| "step": 15975 | |
| }, | |
| { | |
| "epoch": 2.080083203328133, | |
| "grad_norm": 0.0010032757418230176, | |
| "learning_rate": 4.739785278120725e-05, | |
| "loss": 0.00014287895523011683, | |
| "step": 16000 | |
| }, | |
| { | |
| "epoch": 2.0833333333333335, | |
| "grad_norm": 0.001386907184496522, | |
| "learning_rate": 4.738378916663233e-05, | |
| "loss": 0.0001561414636671543, | |
| "step": 16025 | |
| }, | |
| { | |
| "epoch": 2.0865834633385334, | |
| "grad_norm": 0.003277514362707734, | |
| "learning_rate": 4.73696897481759e-05, | |
| "loss": 0.00016391506418585777, | |
| "step": 16050 | |
| }, | |
| { | |
| "epoch": 2.0898335933437338, | |
| "grad_norm": 0.0013601903337985277, | |
| "learning_rate": 4.7355554548390636e-05, | |
| "loss": 0.00015725653618574144, | |
| "step": 16075 | |
| }, | |
| { | |
| "epoch": 2.093083723348934, | |
| "grad_norm": 0.005771314259618521, | |
| "learning_rate": 4.7341383589886423e-05, | |
| "loss": 0.00016310496255755424, | |
| "step": 16100 | |
| }, | |
| { | |
| "epoch": 2.096333853354134, | |
| "grad_norm": 0.0051210480742156506, | |
| "learning_rate": 4.732717689533035e-05, | |
| "loss": 0.00015210462734103203, | |
| "step": 16125 | |
| }, | |
| { | |
| "epoch": 2.0995839833593344, | |
| "grad_norm": 0.003579964628443122, | |
| "learning_rate": 4.731293448744666e-05, | |
| "loss": 0.0001609601080417633, | |
| "step": 16150 | |
| }, | |
| { | |
| "epoch": 2.1028341133645347, | |
| "grad_norm": 0.0032983669079840183, | |
| "learning_rate": 4.7298656389016735e-05, | |
| "loss": 0.0001764364168047905, | |
| "step": 16175 | |
| }, | |
| { | |
| "epoch": 2.1060842433697347, | |
| "grad_norm": 0.0009351523476652801, | |
| "learning_rate": 4.728434262287904e-05, | |
| "loss": 0.00015692852437496185, | |
| "step": 16200 | |
| }, | |
| { | |
| "epoch": 2.109334373374935, | |
| "grad_norm": 0.001914246822707355, | |
| "learning_rate": 4.726999321192908e-05, | |
| "loss": 0.00014737078920006751, | |
| "step": 16225 | |
| }, | |
| { | |
| "epoch": 2.1125845033801354, | |
| "grad_norm": 0.01522519625723362, | |
| "learning_rate": 4.725560817911939e-05, | |
| "loss": 0.00014986968599259854, | |
| "step": 16250 | |
| }, | |
| { | |
| "epoch": 2.1158346333853353, | |
| "grad_norm": 0.004886586219072342, | |
| "learning_rate": 4.724118754745949e-05, | |
| "loss": 0.00015928454697132112, | |
| "step": 16275 | |
| }, | |
| { | |
| "epoch": 2.1190847633905356, | |
| "grad_norm": 0.001382531481795013, | |
| "learning_rate": 4.7226731340015814e-05, | |
| "loss": 0.00016962438821792601, | |
| "step": 16300 | |
| }, | |
| { | |
| "epoch": 2.122334893395736, | |
| "grad_norm": 0.003711164928972721, | |
| "learning_rate": 4.721223957991174e-05, | |
| "loss": 0.00018285322934389115, | |
| "step": 16325 | |
| }, | |
| { | |
| "epoch": 2.125585023400936, | |
| "grad_norm": 0.001002818695269525, | |
| "learning_rate": 4.7197712290327465e-05, | |
| "loss": 0.00019346415996551513, | |
| "step": 16350 | |
| }, | |
| { | |
| "epoch": 2.1288351534061363, | |
| "grad_norm": 0.0012133610434830189, | |
| "learning_rate": 4.718314949450007e-05, | |
| "loss": 0.00015758270397782327, | |
| "step": 16375 | |
| }, | |
| { | |
| "epoch": 2.1320852834113366, | |
| "grad_norm": 0.006691213231533766, | |
| "learning_rate": 4.7168551215723395e-05, | |
| "loss": 0.0001590108871459961, | |
| "step": 16400 | |
| }, | |
| { | |
| "epoch": 2.1353354134165365, | |
| "grad_norm": 0.0008093746728263795, | |
| "learning_rate": 4.715391747734804e-05, | |
| "loss": 0.00014729909598827363, | |
| "step": 16425 | |
| }, | |
| { | |
| "epoch": 2.138585543421737, | |
| "grad_norm": 0.0008228124352172017, | |
| "learning_rate": 4.7139248302781335e-05, | |
| "loss": 0.00014613672159612178, | |
| "step": 16450 | |
| }, | |
| { | |
| "epoch": 2.1418356734269373, | |
| "grad_norm": 0.0027114360127598047, | |
| "learning_rate": 4.712454371548729e-05, | |
| "loss": 0.0001646210253238678, | |
| "step": 16475 | |
| }, | |
| { | |
| "epoch": 2.145085803432137, | |
| "grad_norm": 0.001401035813614726, | |
| "learning_rate": 4.710980373898656e-05, | |
| "loss": 0.00016259271651506424, | |
| "step": 16500 | |
| }, | |
| { | |
| "epoch": 2.1483359334373375, | |
| "grad_norm": 0.0011040671961382031, | |
| "learning_rate": 4.709502839685639e-05, | |
| "loss": 0.0001643284223973751, | |
| "step": 16525 | |
| }, | |
| { | |
| "epoch": 2.151586063442538, | |
| "grad_norm": 0.0015414290828630328, | |
| "learning_rate": 4.708021771273061e-05, | |
| "loss": 0.00015894066542387008, | |
| "step": 16550 | |
| }, | |
| { | |
| "epoch": 2.154836193447738, | |
| "grad_norm": 0.0017135144444182515, | |
| "learning_rate": 4.706537171029959e-05, | |
| "loss": 0.00013949666172266006, | |
| "step": 16575 | |
| }, | |
| { | |
| "epoch": 2.158086323452938, | |
| "grad_norm": 0.0010736393742263317, | |
| "learning_rate": 4.7050490413310166e-05, | |
| "loss": 0.00014713740907609462, | |
| "step": 16600 | |
| }, | |
| { | |
| "epoch": 2.1613364534581385, | |
| "grad_norm": 0.0010634919162839651, | |
| "learning_rate": 4.703557384556566e-05, | |
| "loss": 0.00015331355854868888, | |
| "step": 16625 | |
| }, | |
| { | |
| "epoch": 2.1645865834633384, | |
| "grad_norm": 0.0017489851452410221, | |
| "learning_rate": 4.7020622030925785e-05, | |
| "loss": 0.00014966667629778385, | |
| "step": 16650 | |
| }, | |
| { | |
| "epoch": 2.167836713468539, | |
| "grad_norm": 0.003463171888142824, | |
| "learning_rate": 4.700563499330664e-05, | |
| "loss": 0.0001615521125495434, | |
| "step": 16675 | |
| }, | |
| { | |
| "epoch": 2.171086843473739, | |
| "grad_norm": 0.002824497176334262, | |
| "learning_rate": 4.699061275668068e-05, | |
| "loss": 0.00014396561309695243, | |
| "step": 16700 | |
| }, | |
| { | |
| "epoch": 2.174336973478939, | |
| "grad_norm": 0.003909654449671507, | |
| "learning_rate": 4.6975555345076644e-05, | |
| "loss": 0.00016086556017398833, | |
| "step": 16725 | |
| }, | |
| { | |
| "epoch": 2.1775871034841394, | |
| "grad_norm": 0.007627058308571577, | |
| "learning_rate": 4.696046278257954e-05, | |
| "loss": 0.00014832202345132828, | |
| "step": 16750 | |
| }, | |
| { | |
| "epoch": 2.1808372334893398, | |
| "grad_norm": 0.003833773545920849, | |
| "learning_rate": 4.694533509333061e-05, | |
| "loss": 0.0001377189438790083, | |
| "step": 16775 | |
| }, | |
| { | |
| "epoch": 2.1840873634945397, | |
| "grad_norm": 0.002015323843806982, | |
| "learning_rate": 4.693017230152727e-05, | |
| "loss": 0.00014453311450779438, | |
| "step": 16800 | |
| }, | |
| { | |
| "epoch": 2.18733749349974, | |
| "grad_norm": 0.0015338279772549868, | |
| "learning_rate": 4.6914974431423095e-05, | |
| "loss": 0.0001446526125073433, | |
| "step": 16825 | |
| }, | |
| { | |
| "epoch": 2.19058762350494, | |
| "grad_norm": 0.007382693234831095, | |
| "learning_rate": 4.689974150732776e-05, | |
| "loss": 0.0001543944887816906, | |
| "step": 16850 | |
| }, | |
| { | |
| "epoch": 2.1938377535101403, | |
| "grad_norm": 0.007402203045785427, | |
| "learning_rate": 4.688447355360702e-05, | |
| "loss": 0.00014317329972982407, | |
| "step": 16875 | |
| }, | |
| { | |
| "epoch": 2.1970878835153407, | |
| "grad_norm": 0.0030225885566323996, | |
| "learning_rate": 4.686917059468265e-05, | |
| "loss": 0.00014895868487656118, | |
| "step": 16900 | |
| }, | |
| { | |
| "epoch": 2.2003380135205406, | |
| "grad_norm": 0.0025700044352561235, | |
| "learning_rate": 4.685383265503243e-05, | |
| "loss": 0.00014215481467545033, | |
| "step": 16925 | |
| }, | |
| { | |
| "epoch": 2.203588143525741, | |
| "grad_norm": 0.0025312029756605625, | |
| "learning_rate": 4.683845975919008e-05, | |
| "loss": 0.0001388161163777113, | |
| "step": 16950 | |
| }, | |
| { | |
| "epoch": 2.2068382735309413, | |
| "grad_norm": 0.0013139694929122925, | |
| "learning_rate": 4.682305193174524e-05, | |
| "loss": 0.00015126691199839114, | |
| "step": 16975 | |
| }, | |
| { | |
| "epoch": 2.210088403536141, | |
| "grad_norm": 0.002048899419605732, | |
| "learning_rate": 4.6807609197343444e-05, | |
| "loss": 0.00014540230855345727, | |
| "step": 17000 | |
| }, | |
| { | |
| "epoch": 2.2133385335413416, | |
| "grad_norm": 0.0032946141436696053, | |
| "learning_rate": 4.679213158068602e-05, | |
| "loss": 0.0001423128880560398, | |
| "step": 17025 | |
| }, | |
| { | |
| "epoch": 2.216588663546542, | |
| "grad_norm": 0.002506490331143141, | |
| "learning_rate": 4.677661910653013e-05, | |
| "loss": 0.00015011140145361425, | |
| "step": 17050 | |
| }, | |
| { | |
| "epoch": 2.219838793551742, | |
| "grad_norm": 0.0008630409138277173, | |
| "learning_rate": 4.676107179968867e-05, | |
| "loss": 0.0001389457006007433, | |
| "step": 17075 | |
| }, | |
| { | |
| "epoch": 2.223088923556942, | |
| "grad_norm": 0.0019061139319092035, | |
| "learning_rate": 4.674548968503026e-05, | |
| "loss": 0.00012431109324097632, | |
| "step": 17100 | |
| }, | |
| { | |
| "epoch": 2.2263390535621426, | |
| "grad_norm": 0.001061435672454536, | |
| "learning_rate": 4.672987278747919e-05, | |
| "loss": 0.00013729206286370755, | |
| "step": 17125 | |
| }, | |
| { | |
| "epoch": 2.2295891835673425, | |
| "grad_norm": 0.0025890942197293043, | |
| "learning_rate": 4.671422113201541e-05, | |
| "loss": 0.000145235862582922, | |
| "step": 17150 | |
| }, | |
| { | |
| "epoch": 2.232839313572543, | |
| "grad_norm": 0.0016887147212401032, | |
| "learning_rate": 4.6698534743674436e-05, | |
| "loss": 0.00013569452799856664, | |
| "step": 17175 | |
| }, | |
| { | |
| "epoch": 2.236089443577743, | |
| "grad_norm": 0.007036378141492605, | |
| "learning_rate": 4.668281364754735e-05, | |
| "loss": 0.00013721744529902934, | |
| "step": 17200 | |
| }, | |
| { | |
| "epoch": 2.239339573582943, | |
| "grad_norm": 0.0023496176581829786, | |
| "learning_rate": 4.666705786878077e-05, | |
| "loss": 0.0001310267113149166, | |
| "step": 17225 | |
| }, | |
| { | |
| "epoch": 2.2425897035881435, | |
| "grad_norm": 0.0036521628499031067, | |
| "learning_rate": 4.665126743257677e-05, | |
| "loss": 0.00013894345611333847, | |
| "step": 17250 | |
| }, | |
| { | |
| "epoch": 2.2425897035881435, | |
| "eval_loss": 0.0005545838503167033, | |
| "eval_runtime": 9.044, | |
| "eval_samples_per_second": 552.854, | |
| "eval_steps_per_second": 4.423, | |
| "step": 17250 | |
| }, | |
| { | |
| "epoch": 2.2425897035881435, | |
| "eval_holdout_avg_strict_f1": 0.09842238514349709, | |
| "eval_holdout_enron_spam_bvk_strict_f1": 0.0, | |
| "eval_holdout_enron_spam_setfit_strict_f1": 0.0, | |
| "eval_holdout_gliner2_pii_ptbr_reward_split_strict_f1": 0.41258366800535473, | |
| "eval_holdout_nemotron_pii_strict_f1": 0.2584221748400853, | |
| "eval_holdout_open_pii_masking_500k_strict_f1": 0.2046483549652883, | |
| "eval_holdout_phishing_darkknight_strict_f1": 0.0, | |
| "eval_holdout_phishing_zefang_strict_f1": 0.0, | |
| "eval_holdout_pii_masking_400k_strict_f1": 0.2069920387677397, | |
| "eval_holdout_sms_spam_multilingual_strict_f1": 0.0, | |
| "eval_holdout_spam_messages_mshenoda_strict_f1": 0.0, | |
| "eval_holdout_spamassassin_strict_f1": 0.0, | |
| "step": 17250 | |
| }, | |
| { | |
| "epoch": 2.245839833593344, | |
| "grad_norm": 0.002494937041774392, | |
| "learning_rate": 4.663544236419286e-05, | |
| "loss": 0.00014013076201081277, | |
| "step": 17275 | |
| }, | |
| { | |
| "epoch": 2.2490899635985437, | |
| "grad_norm": 0.0010333035606890917, | |
| "learning_rate": 4.6619582688941967e-05, | |
| "loss": 0.00013317276723682882, | |
| "step": 17300 | |
| }, | |
| { | |
| "epoch": 2.252340093603744, | |
| "grad_norm": 0.0018065182957798243, | |
| "learning_rate": 4.660368843219233e-05, | |
| "loss": 0.00012043621391057968, | |
| "step": 17325 | |
| }, | |
| { | |
| "epoch": 2.2555902236089445, | |
| "grad_norm": 0.0019112216541543603, | |
| "learning_rate": 4.6587759619367556e-05, | |
| "loss": 0.00011801742017269135, | |
| "step": 17350 | |
| }, | |
| { | |
| "epoch": 2.2588403536141444, | |
| "grad_norm": 0.004013299476355314, | |
| "learning_rate": 4.657179627594649e-05, | |
| "loss": 0.00013386973179876804, | |
| "step": 17375 | |
| }, | |
| { | |
| "epoch": 2.2620904836193447, | |
| "grad_norm": 0.0015371774788945913, | |
| "learning_rate": 4.6555798427463214e-05, | |
| "loss": 0.00012612015008926393, | |
| "step": 17400 | |
| }, | |
| { | |
| "epoch": 2.265340613624545, | |
| "grad_norm": 0.0025730461347848177, | |
| "learning_rate": 4.653976609950702e-05, | |
| "loss": 0.00013202549889683722, | |
| "step": 17425 | |
| }, | |
| { | |
| "epoch": 2.268590743629745, | |
| "grad_norm": 0.003649880411103368, | |
| "learning_rate": 4.6523699317722344e-05, | |
| "loss": 0.00013057315722107886, | |
| "step": 17450 | |
| }, | |
| { | |
| "epoch": 2.2718408736349454, | |
| "grad_norm": 0.0028747033793479204, | |
| "learning_rate": 4.650759810780872e-05, | |
| "loss": 0.00014772376045584678, | |
| "step": 17475 | |
| }, | |
| { | |
| "epoch": 2.2750910036401457, | |
| "grad_norm": 0.00198904937133193, | |
| "learning_rate": 4.6491462495520755e-05, | |
| "loss": 0.00013945327140390873, | |
| "step": 17500 | |
| }, | |
| { | |
| "epoch": 2.2783411336453456, | |
| "grad_norm": 0.0031680921092629433, | |
| "learning_rate": 4.64752925066681e-05, | |
| "loss": 0.00013854780234396458, | |
| "step": 17525 | |
| }, | |
| { | |
| "epoch": 2.281591263650546, | |
| "grad_norm": 0.0023765999358147383, | |
| "learning_rate": 4.645908816711538e-05, | |
| "loss": 0.00015536213293671609, | |
| "step": 17550 | |
| }, | |
| { | |
| "epoch": 2.2848413936557463, | |
| "grad_norm": 0.002978266216814518, | |
| "learning_rate": 4.644284950278217e-05, | |
| "loss": 0.00012839575298130514, | |
| "step": 17575 | |
| }, | |
| { | |
| "epoch": 2.2880915236609463, | |
| "grad_norm": 0.00213080202229321, | |
| "learning_rate": 4.6426576539642944e-05, | |
| "loss": 0.00013535557314753533, | |
| "step": 17600 | |
| }, | |
| { | |
| "epoch": 2.2913416536661466, | |
| "grad_norm": 0.001640473841689527, | |
| "learning_rate": 4.641026930372703e-05, | |
| "loss": 0.0001250910200178623, | |
| "step": 17625 | |
| }, | |
| { | |
| "epoch": 2.294591783671347, | |
| "grad_norm": 0.001546473940834403, | |
| "learning_rate": 4.639392782111861e-05, | |
| "loss": 0.00013299187645316124, | |
| "step": 17650 | |
| }, | |
| { | |
| "epoch": 2.297841913676547, | |
| "grad_norm": 0.0011444675037637353, | |
| "learning_rate": 4.6377552117956605e-05, | |
| "loss": 0.000139878848567605, | |
| "step": 17675 | |
| }, | |
| { | |
| "epoch": 2.3010920436817472, | |
| "grad_norm": 0.0017905583372339606, | |
| "learning_rate": 4.636114222043471e-05, | |
| "loss": 0.0001338920369744301, | |
| "step": 17700 | |
| }, | |
| { | |
| "epoch": 2.3043421736869476, | |
| "grad_norm": 0.001992162549868226, | |
| "learning_rate": 4.634469815480128e-05, | |
| "loss": 0.00012586017139256, | |
| "step": 17725 | |
| }, | |
| { | |
| "epoch": 2.3075923036921475, | |
| "grad_norm": 0.0033574625849723816, | |
| "learning_rate": 4.6328219947359345e-05, | |
| "loss": 0.00015158464200794696, | |
| "step": 17750 | |
| }, | |
| { | |
| "epoch": 2.310842433697348, | |
| "grad_norm": 0.0018413373036310077, | |
| "learning_rate": 4.631170762446654e-05, | |
| "loss": 0.0001398076955229044, | |
| "step": 17775 | |
| }, | |
| { | |
| "epoch": 2.3140925637025482, | |
| "grad_norm": 0.0013639956014230847, | |
| "learning_rate": 4.6295161212535085e-05, | |
| "loss": 0.00013965430669486524, | |
| "step": 17800 | |
| }, | |
| { | |
| "epoch": 2.317342693707748, | |
| "grad_norm": 0.0013880865881219506, | |
| "learning_rate": 4.62785807380317e-05, | |
| "loss": 0.00012973963283002376, | |
| "step": 17825 | |
| }, | |
| { | |
| "epoch": 2.3205928237129485, | |
| "grad_norm": 0.002634076401591301, | |
| "learning_rate": 4.62619662274776e-05, | |
| "loss": 0.00011847912333905697, | |
| "step": 17850 | |
| }, | |
| { | |
| "epoch": 2.323842953718149, | |
| "grad_norm": 0.0025655420031398535, | |
| "learning_rate": 4.624531770744845e-05, | |
| "loss": 0.00013092794455587864, | |
| "step": 17875 | |
| }, | |
| { | |
| "epoch": 2.3270930837233488, | |
| "grad_norm": 0.0011559384874999523, | |
| "learning_rate": 4.622863520457431e-05, | |
| "loss": 0.00013991005718708038, | |
| "step": 17900 | |
| }, | |
| { | |
| "epoch": 2.330343213728549, | |
| "grad_norm": 0.001358254812657833, | |
| "learning_rate": 4.621191874553961e-05, | |
| "loss": 0.0001306273229420185, | |
| "step": 17925 | |
| }, | |
| { | |
| "epoch": 2.3335933437337495, | |
| "grad_norm": 0.0014580064453184605, | |
| "learning_rate": 4.619516835708305e-05, | |
| "loss": 0.00012974940240383147, | |
| "step": 17950 | |
| }, | |
| { | |
| "epoch": 2.3368434737389494, | |
| "grad_norm": 0.0014505706494674087, | |
| "learning_rate": 4.617838406599766e-05, | |
| "loss": 0.00012352496385574342, | |
| "step": 17975 | |
| }, | |
| { | |
| "epoch": 2.3400936037441498, | |
| "grad_norm": 0.001112140016630292, | |
| "learning_rate": 4.616156589913066e-05, | |
| "loss": 0.00015324057079851627, | |
| "step": 18000 | |
| }, | |
| { | |
| "epoch": 2.34334373374935, | |
| "grad_norm": 0.0038436620961874723, | |
| "learning_rate": 4.614471388338346e-05, | |
| "loss": 0.00014144724234938622, | |
| "step": 18025 | |
| }, | |
| { | |
| "epoch": 2.34659386375455, | |
| "grad_norm": 0.0019351233495399356, | |
| "learning_rate": 4.612782804571164e-05, | |
| "loss": 0.0001533479243516922, | |
| "step": 18050 | |
| }, | |
| { | |
| "epoch": 2.3498439937597504, | |
| "grad_norm": 0.0018907878547906876, | |
| "learning_rate": 4.6110908413124844e-05, | |
| "loss": 0.00015710491687059402, | |
| "step": 18075 | |
| }, | |
| { | |
| "epoch": 2.3530941237649508, | |
| "grad_norm": 0.002069657202810049, | |
| "learning_rate": 4.60939550126868e-05, | |
| "loss": 0.000148171940818429, | |
| "step": 18100 | |
| }, | |
| { | |
| "epoch": 2.3563442537701507, | |
| "grad_norm": 0.0029697371646761894, | |
| "learning_rate": 4.607696787151522e-05, | |
| "loss": 0.0001361760962754488, | |
| "step": 18125 | |
| }, | |
| { | |
| "epoch": 2.359594383775351, | |
| "grad_norm": 0.0006994548020884395, | |
| "learning_rate": 4.605994701678182e-05, | |
| "loss": 0.00012747122906148434, | |
| "step": 18150 | |
| }, | |
| { | |
| "epoch": 2.3628445137805514, | |
| "grad_norm": 0.0012264895485714078, | |
| "learning_rate": 4.604289247571222e-05, | |
| "loss": 0.00013983145356178284, | |
| "step": 18175 | |
| }, | |
| { | |
| "epoch": 2.3660946437857513, | |
| "grad_norm": 0.0017523291753605008, | |
| "learning_rate": 4.6025804275585923e-05, | |
| "loss": 0.0001352354884147644, | |
| "step": 18200 | |
| }, | |
| { | |
| "epoch": 2.3693447737909517, | |
| "grad_norm": 0.001540504046715796, | |
| "learning_rate": 4.600868244373629e-05, | |
| "loss": 0.00011708905920386314, | |
| "step": 18225 | |
| }, | |
| { | |
| "epoch": 2.372594903796152, | |
| "grad_norm": 0.007799025624990463, | |
| "learning_rate": 4.599152700755045e-05, | |
| "loss": 0.0001339723076671362, | |
| "step": 18250 | |
| }, | |
| { | |
| "epoch": 2.375845033801352, | |
| "grad_norm": 0.001975638559088111, | |
| "learning_rate": 4.597433799446931e-05, | |
| "loss": 0.00013622744008898736, | |
| "step": 18275 | |
| }, | |
| { | |
| "epoch": 2.3790951638065523, | |
| "grad_norm": 0.002473060740157962, | |
| "learning_rate": 4.5957115431987456e-05, | |
| "loss": 0.00012812127359211446, | |
| "step": 18300 | |
| }, | |
| { | |
| "epoch": 2.3823452938117526, | |
| "grad_norm": 0.0024872813373804092, | |
| "learning_rate": 4.593985934765317e-05, | |
| "loss": 0.00016370425000786782, | |
| "step": 18325 | |
| }, | |
| { | |
| "epoch": 2.3855954238169526, | |
| "grad_norm": 0.0016458655009046197, | |
| "learning_rate": 4.5922569769068337e-05, | |
| "loss": 0.0001424828264862299, | |
| "step": 18350 | |
| }, | |
| { | |
| "epoch": 2.388845553822153, | |
| "grad_norm": 0.031232131645083427, | |
| "learning_rate": 4.590524672388842e-05, | |
| "loss": 0.0001416871976107359, | |
| "step": 18375 | |
| }, | |
| { | |
| "epoch": 2.3920956838273533, | |
| "grad_norm": 0.0017826862167567015, | |
| "learning_rate": 4.5887890239822394e-05, | |
| "loss": 0.00012266278266906739, | |
| "step": 18400 | |
| }, | |
| { | |
| "epoch": 2.395345813832553, | |
| "grad_norm": 0.000492918654344976, | |
| "learning_rate": 4.587050034463276e-05, | |
| "loss": 0.0001391277275979519, | |
| "step": 18425 | |
| }, | |
| { | |
| "epoch": 2.3985959438377535, | |
| "grad_norm": 0.005189166869968176, | |
| "learning_rate": 4.5853077066135425e-05, | |
| "loss": 0.000135421147570014, | |
| "step": 18450 | |
| }, | |
| { | |
| "epoch": 2.401846073842954, | |
| "grad_norm": 0.0033815971110016108, | |
| "learning_rate": 4.583562043219972e-05, | |
| "loss": 0.000134381502866745, | |
| "step": 18475 | |
| }, | |
| { | |
| "epoch": 2.405096203848154, | |
| "grad_norm": 0.012698855251073837, | |
| "learning_rate": 4.5818130470748315e-05, | |
| "loss": 0.0001342376321554184, | |
| "step": 18500 | |
| }, | |
| { | |
| "epoch": 2.408346333853354, | |
| "grad_norm": 0.0010202961275354028, | |
| "learning_rate": 4.5800607209757195e-05, | |
| "loss": 0.0001464000716805458, | |
| "step": 18525 | |
| }, | |
| { | |
| "epoch": 2.4115964638585545, | |
| "grad_norm": 0.0015058701392263174, | |
| "learning_rate": 4.578305067725561e-05, | |
| "loss": 0.00014441371895372867, | |
| "step": 18550 | |
| }, | |
| { | |
| "epoch": 2.4148465938637544, | |
| "grad_norm": 0.004408426117151976, | |
| "learning_rate": 4.5765460901326016e-05, | |
| "loss": 0.00012546195648610592, | |
| "step": 18575 | |
| }, | |
| { | |
| "epoch": 2.418096723868955, | |
| "grad_norm": 0.0032601039856672287, | |
| "learning_rate": 4.574783791010407e-05, | |
| "loss": 0.00012056730687618256, | |
| "step": 18600 | |
| }, | |
| { | |
| "epoch": 2.421346853874155, | |
| "grad_norm": 0.0019638906233012676, | |
| "learning_rate": 4.573018173177853e-05, | |
| "loss": 0.00013217531144618988, | |
| "step": 18625 | |
| }, | |
| { | |
| "epoch": 2.424596983879355, | |
| "grad_norm": 0.002587093273177743, | |
| "learning_rate": 4.5712492394591245e-05, | |
| "loss": 0.00012486680410802364, | |
| "step": 18650 | |
| }, | |
| { | |
| "epoch": 2.4278471138845554, | |
| "grad_norm": 0.003023201134055853, | |
| "learning_rate": 4.569476992683712e-05, | |
| "loss": 0.0001134636066854, | |
| "step": 18675 | |
| }, | |
| { | |
| "epoch": 2.431097243889756, | |
| "grad_norm": 0.0010909851407632232, | |
| "learning_rate": 4.567701435686404e-05, | |
| "loss": 0.0001254910882562399, | |
| "step": 18700 | |
| }, | |
| { | |
| "epoch": 2.4343473738949557, | |
| "grad_norm": 0.0016962402733042836, | |
| "learning_rate": 4.565922571307285e-05, | |
| "loss": 0.00012230719439685345, | |
| "step": 18725 | |
| }, | |
| { | |
| "epoch": 2.437597503900156, | |
| "grad_norm": 0.0023738655727356672, | |
| "learning_rate": 4.5641404023917274e-05, | |
| "loss": 0.00012746294029057027, | |
| "step": 18750 | |
| }, | |
| { | |
| "epoch": 2.4408476339053564, | |
| "grad_norm": 0.0019502120558172464, | |
| "learning_rate": 4.5623549317903915e-05, | |
| "loss": 0.00014015112072229385, | |
| "step": 18775 | |
| }, | |
| { | |
| "epoch": 2.4440977639105563, | |
| "grad_norm": 0.0059543875977396965, | |
| "learning_rate": 4.560566162359217e-05, | |
| "loss": 0.00013732610270380974, | |
| "step": 18800 | |
| }, | |
| { | |
| "epoch": 2.4473478939157567, | |
| "grad_norm": 0.0023328778333961964, | |
| "learning_rate": 4.558774096959423e-05, | |
| "loss": 0.00013245033100247382, | |
| "step": 18825 | |
| }, | |
| { | |
| "epoch": 2.450598023920957, | |
| "grad_norm": 0.001375131425447762, | |
| "learning_rate": 4.556978738457499e-05, | |
| "loss": 0.00012461092323064803, | |
| "step": 18850 | |
| }, | |
| { | |
| "epoch": 2.453848153926157, | |
| "grad_norm": 0.0016843504272401333, | |
| "learning_rate": 4.555180089725201e-05, | |
| "loss": 0.0001222700998187065, | |
| "step": 18875 | |
| }, | |
| { | |
| "epoch": 2.4570982839313573, | |
| "grad_norm": 0.0007752921083010733, | |
| "learning_rate": 4.5533781536395495e-05, | |
| "loss": 0.00012718867510557174, | |
| "step": 18900 | |
| }, | |
| { | |
| "epoch": 2.4603484139365577, | |
| "grad_norm": 0.0019048528047278523, | |
| "learning_rate": 4.551572933082822e-05, | |
| "loss": 0.0001389115769416094, | |
| "step": 18925 | |
| }, | |
| { | |
| "epoch": 2.4635985439417576, | |
| "grad_norm": 0.0028496277518570423, | |
| "learning_rate": 4.5497644309425516e-05, | |
| "loss": 0.00012886772863566877, | |
| "step": 18950 | |
| }, | |
| { | |
| "epoch": 2.466848673946958, | |
| "grad_norm": 0.0012810933403670788, | |
| "learning_rate": 4.547952650111517e-05, | |
| "loss": 0.00014181365258991718, | |
| "step": 18975 | |
| }, | |
| { | |
| "epoch": 2.4700988039521583, | |
| "grad_norm": 0.002849346725270152, | |
| "learning_rate": 4.546137593487746e-05, | |
| "loss": 0.00012382476590573786, | |
| "step": 19000 | |
| }, | |
| { | |
| "epoch": 2.473348933957358, | |
| "grad_norm": 0.0025115942116826773, | |
| "learning_rate": 4.5443192639745015e-05, | |
| "loss": 0.00013078431598842144, | |
| "step": 19025 | |
| }, | |
| { | |
| "epoch": 2.4765990639625586, | |
| "grad_norm": 0.004229061305522919, | |
| "learning_rate": 4.542497664480284e-05, | |
| "loss": 0.00012128566391766072, | |
| "step": 19050 | |
| }, | |
| { | |
| "epoch": 2.479849193967759, | |
| "grad_norm": 0.0009478461579419672, | |
| "learning_rate": 4.5406727979188255e-05, | |
| "loss": 0.00010305379517376423, | |
| "step": 19075 | |
| }, | |
| { | |
| "epoch": 2.483099323972959, | |
| "grad_norm": 0.001231037313118577, | |
| "learning_rate": 4.5388446672090814e-05, | |
| "loss": 0.0001633635349571705, | |
| "step": 19100 | |
| }, | |
| { | |
| "epoch": 2.486349453978159, | |
| "grad_norm": 0.0029634146485477686, | |
| "learning_rate": 4.5370132752752294e-05, | |
| "loss": 0.00015426699072122575, | |
| "step": 19125 | |
| }, | |
| { | |
| "epoch": 2.4895995839833596, | |
| "grad_norm": 0.0011719039175659418, | |
| "learning_rate": 4.5351786250466636e-05, | |
| "loss": 0.00014665596187114717, | |
| "step": 19150 | |
| }, | |
| { | |
| "epoch": 2.4928497139885595, | |
| "grad_norm": 0.0012164535000920296, | |
| "learning_rate": 4.53334071945799e-05, | |
| "loss": 0.00012682666070759296, | |
| "step": 19175 | |
| }, | |
| { | |
| "epoch": 2.49609984399376, | |
| "grad_norm": 0.0012999143218621612, | |
| "learning_rate": 4.531499561449022e-05, | |
| "loss": 0.00010815507732331753, | |
| "step": 19200 | |
| }, | |
| { | |
| "epoch": 2.49934997399896, | |
| "grad_norm": 0.0025509654078632593, | |
| "learning_rate": 4.529655153964775e-05, | |
| "loss": 0.00012493634596467019, | |
| "step": 19225 | |
| }, | |
| { | |
| "epoch": 2.50260010400416, | |
| "grad_norm": 0.0032128184102475643, | |
| "learning_rate": 4.52780749995546e-05, | |
| "loss": 0.00011340872384607792, | |
| "step": 19250 | |
| }, | |
| { | |
| "epoch": 2.5058502340093605, | |
| "grad_norm": 0.0019127612467855215, | |
| "learning_rate": 4.525956602376486e-05, | |
| "loss": 0.00012537105940282344, | |
| "step": 19275 | |
| }, | |
| { | |
| "epoch": 2.509100364014561, | |
| "grad_norm": 0.005020787473767996, | |
| "learning_rate": 4.5241024641884456e-05, | |
| "loss": 0.00012422969564795494, | |
| "step": 19300 | |
| }, | |
| { | |
| "epoch": 2.5123504940197607, | |
| "grad_norm": 0.0008939008112065494, | |
| "learning_rate": 4.522245088357116e-05, | |
| "loss": 0.00012096607126295567, | |
| "step": 19325 | |
| }, | |
| { | |
| "epoch": 2.515600624024961, | |
| "grad_norm": 0.0042106895707547665, | |
| "learning_rate": 4.5203844778534554e-05, | |
| "loss": 0.00012804225087165833, | |
| "step": 19350 | |
| }, | |
| { | |
| "epoch": 2.5188507540301615, | |
| "grad_norm": 0.0034091793932020664, | |
| "learning_rate": 4.518520635653594e-05, | |
| "loss": 0.00012403788045048714, | |
| "step": 19375 | |
| }, | |
| { | |
| "epoch": 2.5221008840353614, | |
| "grad_norm": 0.0009586841333657503, | |
| "learning_rate": 4.5166535647388296e-05, | |
| "loss": 0.0001262547541409731, | |
| "step": 19400 | |
| }, | |
| { | |
| "epoch": 2.5253510140405617, | |
| "grad_norm": 0.0019376667914912105, | |
| "learning_rate": 4.5147832680956284e-05, | |
| "loss": 0.00011265106499195099, | |
| "step": 19425 | |
| }, | |
| { | |
| "epoch": 2.528601144045762, | |
| "grad_norm": 0.002278760774061084, | |
| "learning_rate": 4.512909748715613e-05, | |
| "loss": 0.0001353222969919443, | |
| "step": 19450 | |
| }, | |
| { | |
| "epoch": 2.531851274050962, | |
| "grad_norm": 0.00255572353489697, | |
| "learning_rate": 4.511033009595564e-05, | |
| "loss": 0.00012048978358507156, | |
| "step": 19475 | |
| }, | |
| { | |
| "epoch": 2.5351014040561624, | |
| "grad_norm": 0.0012360839173197746, | |
| "learning_rate": 4.50915305373741e-05, | |
| "loss": 0.00014318311586976052, | |
| "step": 19500 | |
| }, | |
| { | |
| "epoch": 2.5383515340613627, | |
| "grad_norm": 0.002191453240811825, | |
| "learning_rate": 4.507269884148224e-05, | |
| "loss": 0.00011643744073808193, | |
| "step": 19525 | |
| }, | |
| { | |
| "epoch": 2.5416016640665626, | |
| "grad_norm": 0.0020136490929871798, | |
| "learning_rate": 4.505383503840221e-05, | |
| "loss": 0.000126743670552969, | |
| "step": 19550 | |
| }, | |
| { | |
| "epoch": 2.544851794071763, | |
| "grad_norm": 0.0034482793416827917, | |
| "learning_rate": 4.503493915830753e-05, | |
| "loss": 0.00014158072881400585, | |
| "step": 19575 | |
| }, | |
| { | |
| "epoch": 2.5481019240769633, | |
| "grad_norm": 0.0006676966440863907, | |
| "learning_rate": 4.5016011231422995e-05, | |
| "loss": 0.00012507325969636442, | |
| "step": 19600 | |
| }, | |
| { | |
| "epoch": 2.5513520540821633, | |
| "grad_norm": 0.0023069949820637703, | |
| "learning_rate": 4.4997051288024686e-05, | |
| "loss": 0.00012421890161931515, | |
| "step": 19625 | |
| }, | |
| { | |
| "epoch": 2.5546021840873636, | |
| "grad_norm": 0.004075527191162109, | |
| "learning_rate": 4.497805935843989e-05, | |
| "loss": 0.00012232734821736812, | |
| "step": 19650 | |
| }, | |
| { | |
| "epoch": 2.557852314092564, | |
| "grad_norm": 0.00142563390545547, | |
| "learning_rate": 4.4959035473047054e-05, | |
| "loss": 0.0001180881354957819, | |
| "step": 19675 | |
| }, | |
| { | |
| "epoch": 2.561102444097764, | |
| "grad_norm": 0.00250995927490294, | |
| "learning_rate": 4.4939979662275745e-05, | |
| "loss": 0.00011440288275480271, | |
| "step": 19700 | |
| }, | |
| { | |
| "epoch": 2.5643525741029642, | |
| "grad_norm": 0.0018000878626480699, | |
| "learning_rate": 4.4920891956606594e-05, | |
| "loss": 0.00012221713550388812, | |
| "step": 19725 | |
| }, | |
| { | |
| "epoch": 2.5676027041081646, | |
| "grad_norm": 0.000926925684325397, | |
| "learning_rate": 4.490177238657124e-05, | |
| "loss": 0.00011894667521119118, | |
| "step": 19750 | |
| }, | |
| { | |
| "epoch": 2.5708528341133645, | |
| "grad_norm": 0.001865015015937388, | |
| "learning_rate": 4.48826209827523e-05, | |
| "loss": 0.00012133187614381313, | |
| "step": 19775 | |
| }, | |
| { | |
| "epoch": 2.574102964118565, | |
| "grad_norm": 0.0020374879240989685, | |
| "learning_rate": 4.4863437775783323e-05, | |
| "loss": 0.00012800646014511587, | |
| "step": 19800 | |
| }, | |
| { | |
| "epoch": 2.577353094123765, | |
| "grad_norm": 0.00834053847938776, | |
| "learning_rate": 4.48442227963487e-05, | |
| "loss": 0.00012342216446995736, | |
| "step": 19825 | |
| }, | |
| { | |
| "epoch": 2.580603224128965, | |
| "grad_norm": 0.002234078012406826, | |
| "learning_rate": 4.4824976075183654e-05, | |
| "loss": 0.00013570293784141542, | |
| "step": 19850 | |
| }, | |
| { | |
| "epoch": 2.5838533541341655, | |
| "grad_norm": 0.002560495166108012, | |
| "learning_rate": 4.4805697643074206e-05, | |
| "loss": 0.00013679668307304383, | |
| "step": 19875 | |
| }, | |
| { | |
| "epoch": 2.5871034841393654, | |
| "grad_norm": 0.0011408563004806638, | |
| "learning_rate": 4.4786387530857056e-05, | |
| "loss": 0.00012191601097583771, | |
| "step": 19900 | |
| }, | |
| { | |
| "epoch": 2.5903536141445658, | |
| "grad_norm": 0.0033448992762714624, | |
| "learning_rate": 4.476704576941961e-05, | |
| "loss": 0.00013250913470983506, | |
| "step": 19925 | |
| }, | |
| { | |
| "epoch": 2.593603744149766, | |
| "grad_norm": 0.0007627570885233581, | |
| "learning_rate": 4.47476723896999e-05, | |
| "loss": 0.000125507153570652, | |
| "step": 19950 | |
| }, | |
| { | |
| "epoch": 2.596853874154966, | |
| "grad_norm": 0.0009357262752018869, | |
| "learning_rate": 4.4728267422686496e-05, | |
| "loss": 0.0001244943216443062, | |
| "step": 19975 | |
| }, | |
| { | |
| "epoch": 2.6001040041601664, | |
| "grad_norm": 0.0020638012792915106, | |
| "learning_rate": 4.4708830899418544e-05, | |
| "loss": 0.00011289622634649276, | |
| "step": 20000 | |
| }, | |
| { | |
| "epoch": 2.6033541341653668, | |
| "grad_norm": 0.0025549668353050947, | |
| "learning_rate": 4.4689362850985625e-05, | |
| "loss": 0.0001239689625799656, | |
| "step": 20025 | |
| }, | |
| { | |
| "epoch": 2.6066042641705667, | |
| "grad_norm": 0.001738504390232265, | |
| "learning_rate": 4.466986330852777e-05, | |
| "loss": 0.00011987220495939255, | |
| "step": 20050 | |
| }, | |
| { | |
| "epoch": 2.609854394175767, | |
| "grad_norm": 0.0034898375160992146, | |
| "learning_rate": 4.465033230323538e-05, | |
| "loss": 0.00010324159637093545, | |
| "step": 20075 | |
| }, | |
| { | |
| "epoch": 2.6131045241809674, | |
| "grad_norm": 0.0032646439503878355, | |
| "learning_rate": 4.463076986634916e-05, | |
| "loss": 0.00011766684241592884, | |
| "step": 20100 | |
| }, | |
| { | |
| "epoch": 2.6163546541861673, | |
| "grad_norm": 0.0017099531833082438, | |
| "learning_rate": 4.461117602916014e-05, | |
| "loss": 0.00011553521268069745, | |
| "step": 20125 | |
| }, | |
| { | |
| "epoch": 2.6196047841913677, | |
| "grad_norm": 0.001333041349425912, | |
| "learning_rate": 4.459155082300952e-05, | |
| "loss": 0.00011520256288349628, | |
| "step": 20150 | |
| }, | |
| { | |
| "epoch": 2.622854914196568, | |
| "grad_norm": 0.0026836670003831387, | |
| "learning_rate": 4.457189427928872e-05, | |
| "loss": 0.00011987383477389813, | |
| "step": 20175 | |
| }, | |
| { | |
| "epoch": 2.626105044201768, | |
| "grad_norm": 0.001452404772862792, | |
| "learning_rate": 4.455220642943927e-05, | |
| "loss": 0.00012432758696377278, | |
| "step": 20200 | |
| }, | |
| { | |
| "epoch": 2.6293551742069683, | |
| "grad_norm": 0.0016890125116333365, | |
| "learning_rate": 4.4532487304952764e-05, | |
| "loss": 0.00010978843085467815, | |
| "step": 20225 | |
| }, | |
| { | |
| "epoch": 2.6326053042121687, | |
| "grad_norm": 0.000463778676930815, | |
| "learning_rate": 4.4512736937370844e-05, | |
| "loss": 0.00012221131473779678, | |
| "step": 20250 | |
| }, | |
| { | |
| "epoch": 2.6358554342173686, | |
| "grad_norm": 0.0005282393540255725, | |
| "learning_rate": 4.4492955358285104e-05, | |
| "loss": 0.00011833841912448407, | |
| "step": 20275 | |
| }, | |
| { | |
| "epoch": 2.639105564222569, | |
| "grad_norm": 0.007521711755543947, | |
| "learning_rate": 4.447314259933708e-05, | |
| "loss": 0.00011013507843017578, | |
| "step": 20300 | |
| }, | |
| { | |
| "epoch": 2.642355694227769, | |
| "grad_norm": 0.001455447869375348, | |
| "learning_rate": 4.445329869221817e-05, | |
| "loss": 0.00012725085951387883, | |
| "step": 20325 | |
| }, | |
| { | |
| "epoch": 2.645605824232969, | |
| "grad_norm": 0.000706520804669708, | |
| "learning_rate": 4.4433423668669604e-05, | |
| "loss": 0.00012050364166498184, | |
| "step": 20350 | |
| }, | |
| { | |
| "epoch": 2.6488559542381696, | |
| "grad_norm": 0.002538814675062895, | |
| "learning_rate": 4.4413517560482376e-05, | |
| "loss": 0.0001273483317345381, | |
| "step": 20375 | |
| }, | |
| { | |
| "epoch": 2.6521060842433695, | |
| "grad_norm": 0.0016709818737581372, | |
| "learning_rate": 4.4393580399497214e-05, | |
| "loss": 0.00012278759852051734, | |
| "step": 20400 | |
| }, | |
| { | |
| "epoch": 2.65535621424857, | |
| "grad_norm": 0.0006076059071347117, | |
| "learning_rate": 4.4373612217604496e-05, | |
| "loss": 0.00011577093973755837, | |
| "step": 20425 | |
| }, | |
| { | |
| "epoch": 2.65860634425377, | |
| "grad_norm": 0.0018147925147786736, | |
| "learning_rate": 4.435361304674423e-05, | |
| "loss": 0.0001043164450675249, | |
| "step": 20450 | |
| }, | |
| { | |
| "epoch": 2.66185647425897, | |
| "grad_norm": 0.003252327674999833, | |
| "learning_rate": 4.4333582918906e-05, | |
| "loss": 0.00013649925589561462, | |
| "step": 20475 | |
| }, | |
| { | |
| "epoch": 2.6651066042641705, | |
| "grad_norm": 0.005322558805346489, | |
| "learning_rate": 4.431352186612889e-05, | |
| "loss": 0.00011188880540430546, | |
| "step": 20500 | |
| }, | |
| { | |
| "epoch": 2.668356734269371, | |
| "grad_norm": 0.0034299592953175306, | |
| "learning_rate": 4.429342992050148e-05, | |
| "loss": 0.00012011456303298473, | |
| "step": 20525 | |
| }, | |
| { | |
| "epoch": 2.6716068642745707, | |
| "grad_norm": 0.0012762858532369137, | |
| "learning_rate": 4.427330711416172e-05, | |
| "loss": 0.00010836035944521427, | |
| "step": 20550 | |
| }, | |
| { | |
| "epoch": 2.674856994279771, | |
| "grad_norm": 0.0010440415935590863, | |
| "learning_rate": 4.4253153479296964e-05, | |
| "loss": 0.00011502891778945923, | |
| "step": 20575 | |
| }, | |
| { | |
| "epoch": 2.6781071242849714, | |
| "grad_norm": 0.0015414906665682793, | |
| "learning_rate": 4.423296904814386e-05, | |
| "loss": 0.0001234014518558979, | |
| "step": 20600 | |
| }, | |
| { | |
| "epoch": 2.6813572542901714, | |
| "grad_norm": 0.0012247645063325763, | |
| "learning_rate": 4.4212753852988296e-05, | |
| "loss": 0.00011393558233976364, | |
| "step": 20625 | |
| }, | |
| { | |
| "epoch": 2.6846073842953717, | |
| "grad_norm": 0.001106818439438939, | |
| "learning_rate": 4.419250792616542e-05, | |
| "loss": 0.00011690087616443634, | |
| "step": 20650 | |
| }, | |
| { | |
| "epoch": 2.687857514300572, | |
| "grad_norm": 0.002656723605468869, | |
| "learning_rate": 4.4172231300059485e-05, | |
| "loss": 0.00010846965014934539, | |
| "step": 20675 | |
| }, | |
| { | |
| "epoch": 2.691107644305772, | |
| "grad_norm": 0.001735582947731018, | |
| "learning_rate": 4.415192400710388e-05, | |
| "loss": 0.00011780579574406147, | |
| "step": 20700 | |
| }, | |
| { | |
| "epoch": 2.691107644305772, | |
| "eval_loss": 0.0004374645359348506, | |
| "eval_runtime": 9.1661, | |
| "eval_samples_per_second": 545.489, | |
| "eval_steps_per_second": 4.364, | |
| "step": 20700 | |
| }, | |
| { | |
| "epoch": 2.691107644305772, | |
| "eval_holdout_avg_strict_f1": 0.11837511902072301, | |
| "eval_holdout_enron_spam_bvk_strict_f1": 0.0, | |
| "eval_holdout_enron_spam_setfit_strict_f1": 0.0, | |
| "eval_holdout_gliner2_pii_ptbr_reward_split_strict_f1": 0.48383561643835615, | |
| "eval_holdout_nemotron_pii_strict_f1": 0.3570530000546956, | |
| "eval_holdout_open_pii_masking_500k_strict_f1": 0.23175675675675675, | |
| "eval_holdout_phishing_darkknight_strict_f1": 0.0, | |
| "eval_holdout_phishing_zefang_strict_f1": 0.0, | |
| "eval_holdout_pii_masking_400k_strict_f1": 0.2294809359781447, | |
| "eval_holdout_sms_spam_multilingual_strict_f1": 0.0, | |
| "eval_holdout_spam_messages_mshenoda_strict_f1": 0.0, | |
| "eval_holdout_spamassassin_strict_f1": 0.0, | |
| "step": 20700 | |
| }, | |
| { | |
| "epoch": 2.6943577743109723, | |
| "grad_norm": 0.0015983999473974109, | |
| "learning_rate": 4.413158607978104e-05, | |
| "loss": 0.00011728547513484955, | |
| "step": 20725 | |
| }, | |
| { | |
| "epoch": 2.6976079043161727, | |
| "grad_norm": 0.0018345334101468325, | |
| "learning_rate": 4.4111217550622385e-05, | |
| "loss": 9.05183982104063e-05, | |
| "step": 20750 | |
| }, | |
| { | |
| "epoch": 2.7008580343213726, | |
| "grad_norm": 0.0007273370865732431, | |
| "learning_rate": 4.409081845220831e-05, | |
| "loss": 9.845491498708725e-05, | |
| "step": 20775 | |
| }, | |
| { | |
| "epoch": 2.704108164326573, | |
| "grad_norm": 0.004765399731695652, | |
| "learning_rate": 4.407038881716808e-05, | |
| "loss": 0.00012204086408019066, | |
| "step": 20800 | |
| }, | |
| { | |
| "epoch": 2.7073582943317733, | |
| "grad_norm": 0.0012289343867450953, | |
| "learning_rate": 4.404992867817983e-05, | |
| "loss": 0.00011148681864142418, | |
| "step": 20825 | |
| }, | |
| { | |
| "epoch": 2.7106084243369732, | |
| "grad_norm": 0.0041665080934762955, | |
| "learning_rate": 4.402943806797048e-05, | |
| "loss": 0.0001245866809040308, | |
| "step": 20850 | |
| }, | |
| { | |
| "epoch": 2.7138585543421736, | |
| "grad_norm": 0.0009457609849050641, | |
| "learning_rate": 4.4008917019315676e-05, | |
| "loss": 0.00011880895122885704, | |
| "step": 20875 | |
| }, | |
| { | |
| "epoch": 2.717108684347374, | |
| "grad_norm": 0.0019655637443065643, | |
| "learning_rate": 4.398836556503976e-05, | |
| "loss": 0.00011588115245103836, | |
| "step": 20900 | |
| }, | |
| { | |
| "epoch": 2.720358814352574, | |
| "grad_norm": 0.0032526394352316856, | |
| "learning_rate": 4.3967783738015724e-05, | |
| "loss": 0.00011598647572100162, | |
| "step": 20925 | |
| }, | |
| { | |
| "epoch": 2.7236089443577742, | |
| "grad_norm": 0.0015118825249373913, | |
| "learning_rate": 4.394717157116512e-05, | |
| "loss": 9.912591427564621e-05, | |
| "step": 20950 | |
| }, | |
| { | |
| "epoch": 2.7268590743629746, | |
| "grad_norm": 0.0020698097068816423, | |
| "learning_rate": 4.392652909745804e-05, | |
| "loss": 0.00011991173960268497, | |
| "step": 20975 | |
| }, | |
| { | |
| "epoch": 2.7301092043681745, | |
| "grad_norm": 0.002521323272958398, | |
| "learning_rate": 4.390585634991305e-05, | |
| "loss": 0.0001280319970101118, | |
| "step": 21000 | |
| }, | |
| { | |
| "epoch": 2.733359334373375, | |
| "grad_norm": 0.0008192371460609138, | |
| "learning_rate": 4.388515336159716e-05, | |
| "loss": 0.00016657285392284392, | |
| "step": 21025 | |
| }, | |
| { | |
| "epoch": 2.736609464378575, | |
| "grad_norm": 0.0016620284877717495, | |
| "learning_rate": 4.386442016562572e-05, | |
| "loss": 0.00013518611900508403, | |
| "step": 21050 | |
| }, | |
| { | |
| "epoch": 2.739859594383775, | |
| "grad_norm": 0.0011725135846063495, | |
| "learning_rate": 4.38436567951624e-05, | |
| "loss": 0.00011334162205457687, | |
| "step": 21075 | |
| }, | |
| { | |
| "epoch": 2.7431097243889755, | |
| "grad_norm": 0.001341929892078042, | |
| "learning_rate": 4.382286328341918e-05, | |
| "loss": 0.00011002503335475921, | |
| "step": 21100 | |
| }, | |
| { | |
| "epoch": 2.746359854394176, | |
| "grad_norm": 0.0007556485943496227, | |
| "learning_rate": 4.38020396636562e-05, | |
| "loss": 0.00012014972046017647, | |
| "step": 21125 | |
| }, | |
| { | |
| "epoch": 2.7496099843993758, | |
| "grad_norm": 0.001568184932693839, | |
| "learning_rate": 4.378118596918179e-05, | |
| "loss": 0.00011602149344980717, | |
| "step": 21150 | |
| }, | |
| { | |
| "epoch": 2.752860114404576, | |
| "grad_norm": 0.0007142325048334897, | |
| "learning_rate": 4.376030223335237e-05, | |
| "loss": 0.00011547482572495937, | |
| "step": 21175 | |
| }, | |
| { | |
| "epoch": 2.7561102444097765, | |
| "grad_norm": 0.002017251681536436, | |
| "learning_rate": 4.3739388489572416e-05, | |
| "loss": 0.00011696443893015385, | |
| "step": 21200 | |
| }, | |
| { | |
| "epoch": 2.7593603744149764, | |
| "grad_norm": 0.0008025502902455628, | |
| "learning_rate": 4.371844477129441e-05, | |
| "loss": 0.00011482313275337219, | |
| "step": 21225 | |
| }, | |
| { | |
| "epoch": 2.7626105044201767, | |
| "grad_norm": 0.008052889257669449, | |
| "learning_rate": 4.369747111201877e-05, | |
| "loss": 0.00011181281879544259, | |
| "step": 21250 | |
| }, | |
| { | |
| "epoch": 2.765860634425377, | |
| "grad_norm": 0.0014665208291262388, | |
| "learning_rate": 4.367646754529383e-05, | |
| "loss": 0.0001287536881864071, | |
| "step": 21275 | |
| }, | |
| { | |
| "epoch": 2.769110764430577, | |
| "grad_norm": 0.0010699396952986717, | |
| "learning_rate": 4.365543410471571e-05, | |
| "loss": 0.00013060064986348153, | |
| "step": 21300 | |
| }, | |
| { | |
| "epoch": 2.7723608944357774, | |
| "grad_norm": 0.002420753473415971, | |
| "learning_rate": 4.363437082392838e-05, | |
| "loss": 0.00011137327179312706, | |
| "step": 21325 | |
| }, | |
| { | |
| "epoch": 2.7756110244409777, | |
| "grad_norm": 0.0012363316491246223, | |
| "learning_rate": 4.361327773662349e-05, | |
| "loss": 0.00014740562997758388, | |
| "step": 21350 | |
| }, | |
| { | |
| "epoch": 2.7788611544461777, | |
| "grad_norm": 0.001054872409440577, | |
| "learning_rate": 4.359215487654039e-05, | |
| "loss": 0.00010994773358106613, | |
| "step": 21375 | |
| }, | |
| { | |
| "epoch": 2.782111284451378, | |
| "grad_norm": 0.0021502096205949783, | |
| "learning_rate": 4.357100227746606e-05, | |
| "loss": 0.00011184143833816052, | |
| "step": 21400 | |
| }, | |
| { | |
| "epoch": 2.7853614144565784, | |
| "grad_norm": 0.0025471800472587347, | |
| "learning_rate": 4.354981997323501e-05, | |
| "loss": 0.00011151881888508797, | |
| "step": 21425 | |
| }, | |
| { | |
| "epoch": 2.7886115444617783, | |
| "grad_norm": 0.001021609059534967, | |
| "learning_rate": 4.352860799772931e-05, | |
| "loss": 0.00010406337678432464, | |
| "step": 21450 | |
| }, | |
| { | |
| "epoch": 2.7918616744669786, | |
| "grad_norm": 0.0068647414445877075, | |
| "learning_rate": 4.3507366384878486e-05, | |
| "loss": 0.00011970474384725094, | |
| "step": 21475 | |
| }, | |
| { | |
| "epoch": 2.795111804472179, | |
| "grad_norm": 0.006094173062592745, | |
| "learning_rate": 4.3486095168659435e-05, | |
| "loss": 0.0001264354959130287, | |
| "step": 21500 | |
| }, | |
| { | |
| "epoch": 2.798361934477379, | |
| "grad_norm": 0.00240479432977736, | |
| "learning_rate": 4.346479438309643e-05, | |
| "loss": 0.00012017049826681614, | |
| "step": 21525 | |
| }, | |
| { | |
| "epoch": 2.8016120644825793, | |
| "grad_norm": 0.0014283722266554832, | |
| "learning_rate": 4.3443464062261055e-05, | |
| "loss": 0.0001104743592441082, | |
| "step": 21550 | |
| }, | |
| { | |
| "epoch": 2.8048621944877796, | |
| "grad_norm": 0.0019245878793299198, | |
| "learning_rate": 4.34221042402721e-05, | |
| "loss": 0.00010544533841311932, | |
| "step": 21575 | |
| }, | |
| { | |
| "epoch": 2.8081123244929795, | |
| "grad_norm": 0.0009432962979190052, | |
| "learning_rate": 4.3400714951295586e-05, | |
| "loss": 0.00012702370062470436, | |
| "step": 21600 | |
| }, | |
| { | |
| "epoch": 2.81136245449818, | |
| "grad_norm": 0.0030835135839879513, | |
| "learning_rate": 4.337929622954463e-05, | |
| "loss": 0.00010390413925051689, | |
| "step": 21625 | |
| }, | |
| { | |
| "epoch": 2.8146125845033803, | |
| "grad_norm": 0.0016890065744519234, | |
| "learning_rate": 4.335784810927945e-05, | |
| "loss": 0.00012419668957591056, | |
| "step": 21650 | |
| }, | |
| { | |
| "epoch": 2.81786271450858, | |
| "grad_norm": 0.0017951803747564554, | |
| "learning_rate": 4.3336370624807286e-05, | |
| "loss": 0.00012576142325997352, | |
| "step": 21675 | |
| }, | |
| { | |
| "epoch": 2.8211128445137805, | |
| "grad_norm": 0.002675277879461646, | |
| "learning_rate": 4.3314863810482344e-05, | |
| "loss": 0.00012105302885174751, | |
| "step": 21700 | |
| }, | |
| { | |
| "epoch": 2.824362974518981, | |
| "grad_norm": 0.003099619410932064, | |
| "learning_rate": 4.3293327700705744e-05, | |
| "loss": 0.00012650035321712494, | |
| "step": 21725 | |
| }, | |
| { | |
| "epoch": 2.827613104524181, | |
| "grad_norm": 0.0013802091125398874, | |
| "learning_rate": 4.3271762329925456e-05, | |
| "loss": 0.00011229857802391052, | |
| "step": 21750 | |
| }, | |
| { | |
| "epoch": 2.830863234529381, | |
| "grad_norm": 0.0008018408552743495, | |
| "learning_rate": 4.325016773263628e-05, | |
| "loss": 0.00011777706444263459, | |
| "step": 21775 | |
| }, | |
| { | |
| "epoch": 2.8341133645345815, | |
| "grad_norm": 0.0031042713671922684, | |
| "learning_rate": 4.322854394337974e-05, | |
| "loss": 0.00011567095294594765, | |
| "step": 21800 | |
| }, | |
| { | |
| "epoch": 2.8373634945397814, | |
| "grad_norm": 0.001013422035612166, | |
| "learning_rate": 4.320689099674406e-05, | |
| "loss": 0.0001094942819327116, | |
| "step": 21825 | |
| }, | |
| { | |
| "epoch": 2.840613624544982, | |
| "grad_norm": 0.0008522962452843785, | |
| "learning_rate": 4.318520892736412e-05, | |
| "loss": 0.00010848546400666237, | |
| "step": 21850 | |
| }, | |
| { | |
| "epoch": 2.843863754550182, | |
| "grad_norm": 0.0013135053450241685, | |
| "learning_rate": 4.316349776992136e-05, | |
| "loss": 0.00011342357844114304, | |
| "step": 21875 | |
| }, | |
| { | |
| "epoch": 2.847113884555382, | |
| "grad_norm": 0.0025406270287930965, | |
| "learning_rate": 4.314175755914375e-05, | |
| "loss": 0.00012101396918296813, | |
| "step": 21900 | |
| }, | |
| { | |
| "epoch": 2.8503640145605824, | |
| "grad_norm": 0.0034799096174538136, | |
| "learning_rate": 4.311998832980576e-05, | |
| "loss": 0.00011799228377640247, | |
| "step": 21925 | |
| }, | |
| { | |
| "epoch": 2.8536141445657828, | |
| "grad_norm": 0.0035920601803809404, | |
| "learning_rate": 4.309819011672823e-05, | |
| "loss": 0.00012042688205838204, | |
| "step": 21950 | |
| }, | |
| { | |
| "epoch": 2.8568642745709827, | |
| "grad_norm": 0.0020514419302344322, | |
| "learning_rate": 4.3076362954778393e-05, | |
| "loss": 0.00012154130265116691, | |
| "step": 21975 | |
| }, | |
| { | |
| "epoch": 2.860114404576183, | |
| "grad_norm": 0.0014667565701529384, | |
| "learning_rate": 4.3054506878869794e-05, | |
| "loss": 0.00012286231853067875, | |
| "step": 22000 | |
| }, | |
| { | |
| "epoch": 2.8633645345813834, | |
| "grad_norm": 0.0020479902159422636, | |
| "learning_rate": 4.3032621923962203e-05, | |
| "loss": 0.00011883649975061417, | |
| "step": 22025 | |
| }, | |
| { | |
| "epoch": 2.8666146645865833, | |
| "grad_norm": 0.001707935007289052, | |
| "learning_rate": 4.3010708125061596e-05, | |
| "loss": 0.00012006035074591637, | |
| "step": 22050 | |
| }, | |
| { | |
| "epoch": 2.8698647945917837, | |
| "grad_norm": 0.0016003301134333014, | |
| "learning_rate": 4.298876551722008e-05, | |
| "loss": 0.0001111164502799511, | |
| "step": 22075 | |
| }, | |
| { | |
| "epoch": 2.873114924596984, | |
| "grad_norm": 0.0012627223040908575, | |
| "learning_rate": 4.296679413553584e-05, | |
| "loss": 0.00011253083124756813, | |
| "step": 22100 | |
| }, | |
| { | |
| "epoch": 2.876365054602184, | |
| "grad_norm": 0.0007519132923334837, | |
| "learning_rate": 4.2944794015153114e-05, | |
| "loss": 0.00012518050149083138, | |
| "step": 22125 | |
| }, | |
| { | |
| "epoch": 2.8796151846073843, | |
| "grad_norm": 0.0009976483415812254, | |
| "learning_rate": 4.292276519126207e-05, | |
| "loss": 0.00012308670207858085, | |
| "step": 22150 | |
| }, | |
| { | |
| "epoch": 2.8828653146125847, | |
| "grad_norm": 0.0025145686231553555, | |
| "learning_rate": 4.290070769909882e-05, | |
| "loss": 0.0001339484564960003, | |
| "step": 22175 | |
| }, | |
| { | |
| "epoch": 2.8861154446177846, | |
| "grad_norm": 0.0014710049144923687, | |
| "learning_rate": 4.287862157394531e-05, | |
| "loss": 0.00011939244344830513, | |
| "step": 22200 | |
| }, | |
| { | |
| "epoch": 2.889365574622985, | |
| "grad_norm": 0.00458088144659996, | |
| "learning_rate": 4.285650685112929e-05, | |
| "loss": 0.00011669550091028213, | |
| "step": 22225 | |
| }, | |
| { | |
| "epoch": 2.8926157046281853, | |
| "grad_norm": 0.005479548592120409, | |
| "learning_rate": 4.283436356602426e-05, | |
| "loss": 0.00011207645758986472, | |
| "step": 22250 | |
| }, | |
| { | |
| "epoch": 2.895865834633385, | |
| "grad_norm": 0.0032126375008374453, | |
| "learning_rate": 4.2812191754049416e-05, | |
| "loss": 0.00011596095748245717, | |
| "step": 22275 | |
| }, | |
| { | |
| "epoch": 2.8991159646385856, | |
| "grad_norm": 0.008369572460651398, | |
| "learning_rate": 4.278999145066956e-05, | |
| "loss": 0.00012018016539514065, | |
| "step": 22300 | |
| }, | |
| { | |
| "epoch": 2.902366094643786, | |
| "grad_norm": 0.007213692180812359, | |
| "learning_rate": 4.27677626913951e-05, | |
| "loss": 0.0001279247272759676, | |
| "step": 22325 | |
| }, | |
| { | |
| "epoch": 2.905616224648986, | |
| "grad_norm": 0.0024194561410695314, | |
| "learning_rate": 4.274550551178191e-05, | |
| "loss": 0.00012097568251192569, | |
| "step": 22350 | |
| }, | |
| { | |
| "epoch": 2.908866354654186, | |
| "grad_norm": 0.001562710152938962, | |
| "learning_rate": 4.272321994743137e-05, | |
| "loss": 0.00011379463598132133, | |
| "step": 22375 | |
| }, | |
| { | |
| "epoch": 2.9121164846593866, | |
| "grad_norm": 0.000971033878158778, | |
| "learning_rate": 4.2700906033990246e-05, | |
| "loss": 0.00010655920952558518, | |
| "step": 22400 | |
| }, | |
| { | |
| "epoch": 2.9153666146645865, | |
| "grad_norm": 0.0017622068990021944, | |
| "learning_rate": 4.267856380715065e-05, | |
| "loss": 0.00010518478229641915, | |
| "step": 22425 | |
| }, | |
| { | |
| "epoch": 2.918616744669787, | |
| "grad_norm": 0.0006176725728437304, | |
| "learning_rate": 4.2656193302649975e-05, | |
| "loss": 9.673822671175003e-05, | |
| "step": 22450 | |
| }, | |
| { | |
| "epoch": 2.921866874674987, | |
| "grad_norm": 0.0009493599063716829, | |
| "learning_rate": 4.2633794556270865e-05, | |
| "loss": 0.00012251473963260652, | |
| "step": 22475 | |
| }, | |
| { | |
| "epoch": 2.925117004680187, | |
| "grad_norm": 0.0012775755021721125, | |
| "learning_rate": 4.261136760384112e-05, | |
| "loss": 0.00012170745059847832, | |
| "step": 22500 | |
| }, | |
| { | |
| "epoch": 2.9283671346853875, | |
| "grad_norm": 0.0007665643352083862, | |
| "learning_rate": 4.2588912481233666e-05, | |
| "loss": 0.00011376162059605122, | |
| "step": 22525 | |
| }, | |
| { | |
| "epoch": 2.931617264690588, | |
| "grad_norm": 0.0014551812782883644, | |
| "learning_rate": 4.256642922436648e-05, | |
| "loss": 0.00010671044699847698, | |
| "step": 22550 | |
| }, | |
| { | |
| "epoch": 2.9348673946957877, | |
| "grad_norm": 0.0019352270755916834, | |
| "learning_rate": 4.2543917869202555e-05, | |
| "loss": 0.00011777355335652828, | |
| "step": 22575 | |
| }, | |
| { | |
| "epoch": 2.938117524700988, | |
| "grad_norm": 0.001100274850614369, | |
| "learning_rate": 4.2521378451749805e-05, | |
| "loss": 0.00011031841859221458, | |
| "step": 22600 | |
| }, | |
| { | |
| "epoch": 2.9413676547061884, | |
| "grad_norm": 0.0013112631859257817, | |
| "learning_rate": 4.249881100806107e-05, | |
| "loss": 0.00011953277513384819, | |
| "step": 22625 | |
| }, | |
| { | |
| "epoch": 2.9446177847113884, | |
| "grad_norm": 0.002739132847636938, | |
| "learning_rate": 4.247621557423397e-05, | |
| "loss": 0.00011528748087584973, | |
| "step": 22650 | |
| }, | |
| { | |
| "epoch": 2.9478679147165887, | |
| "grad_norm": 0.0018405403243377805, | |
| "learning_rate": 4.245359218641094e-05, | |
| "loss": 0.00010743656195700169, | |
| "step": 22675 | |
| }, | |
| { | |
| "epoch": 2.951118044721789, | |
| "grad_norm": 0.0015506376512348652, | |
| "learning_rate": 4.2430940880779096e-05, | |
| "loss": 0.00010667442344129085, | |
| "step": 22700 | |
| }, | |
| { | |
| "epoch": 2.954368174726989, | |
| "grad_norm": 0.0018239638302475214, | |
| "learning_rate": 4.240826169357024e-05, | |
| "loss": 0.00012290218845009804, | |
| "step": 22725 | |
| }, | |
| { | |
| "epoch": 2.9576183047321893, | |
| "grad_norm": 0.0006942609907127917, | |
| "learning_rate": 4.238555466106074e-05, | |
| "loss": 0.0001141014602035284, | |
| "step": 22750 | |
| }, | |
| { | |
| "epoch": 2.9608684347373897, | |
| "grad_norm": 0.001283511403016746, | |
| "learning_rate": 4.236281981957153e-05, | |
| "loss": 0.00010799052193760872, | |
| "step": 22775 | |
| }, | |
| { | |
| "epoch": 2.9641185647425896, | |
| "grad_norm": 0.0018495937110856175, | |
| "learning_rate": 4.234005720546801e-05, | |
| "loss": 0.00011130035854876042, | |
| "step": 22800 | |
| }, | |
| { | |
| "epoch": 2.96736869474779, | |
| "grad_norm": 0.0009274588082917035, | |
| "learning_rate": 4.231726685516002e-05, | |
| "loss": 9.860348887741566e-05, | |
| "step": 22825 | |
| }, | |
| { | |
| "epoch": 2.9706188247529903, | |
| "grad_norm": 0.001409834367223084, | |
| "learning_rate": 4.229444880510174e-05, | |
| "loss": 0.00011361884884536266, | |
| "step": 22850 | |
| }, | |
| { | |
| "epoch": 2.9738689547581902, | |
| "grad_norm": 0.0008306899107992649, | |
| "learning_rate": 4.227160309179168e-05, | |
| "loss": 9.721308015286922e-05, | |
| "step": 22875 | |
| }, | |
| { | |
| "epoch": 2.9771190847633906, | |
| "grad_norm": 0.0014764899387955666, | |
| "learning_rate": 4.2248729751772586e-05, | |
| "loss": 0.00011204950511455536, | |
| "step": 22900 | |
| }, | |
| { | |
| "epoch": 2.980369214768591, | |
| "grad_norm": 0.0010945445392280817, | |
| "learning_rate": 4.22258288216314e-05, | |
| "loss": 0.00010479540564119816, | |
| "step": 22925 | |
| }, | |
| { | |
| "epoch": 2.983619344773791, | |
| "grad_norm": 0.003540492383763194, | |
| "learning_rate": 4.2202900337999196e-05, | |
| "loss": 0.00011321960017085075, | |
| "step": 22950 | |
| }, | |
| { | |
| "epoch": 2.9868694747789912, | |
| "grad_norm": 0.00788059365004301, | |
| "learning_rate": 4.217994433755111e-05, | |
| "loss": 0.0001204497367143631, | |
| "step": 22975 | |
| }, | |
| { | |
| "epoch": 2.9901196047841916, | |
| "grad_norm": 0.003780391300097108, | |
| "learning_rate": 4.215696085700632e-05, | |
| "loss": 0.00010514517314732075, | |
| "step": 23000 | |
| }, | |
| { | |
| "epoch": 2.9933697347893915, | |
| "grad_norm": 0.003067641519010067, | |
| "learning_rate": 4.213394993312792e-05, | |
| "loss": 0.00011237187311053276, | |
| "step": 23025 | |
| }, | |
| { | |
| "epoch": 2.996619864794592, | |
| "grad_norm": 0.001331760548055172, | |
| "learning_rate": 4.2110911602722937e-05, | |
| "loss": 0.0001090239454060793, | |
| "step": 23050 | |
| }, | |
| { | |
| "epoch": 2.999869994799792, | |
| "grad_norm": 0.0011301770573481917, | |
| "learning_rate": 4.208784590264221e-05, | |
| "loss": 9.060624986886978e-05, | |
| "step": 23075 | |
| }, | |
| { | |
| "epoch": 3.003120124804992, | |
| "grad_norm": 0.0005481968400999904, | |
| "learning_rate": 4.206475286978037e-05, | |
| "loss": 9.929303079843521e-05, | |
| "step": 23100 | |
| }, | |
| { | |
| "epoch": 3.0063702548101925, | |
| "grad_norm": 0.00205858051776886, | |
| "learning_rate": 4.204163254107577e-05, | |
| "loss": 0.0001014355756342411, | |
| "step": 23125 | |
| }, | |
| { | |
| "epoch": 3.0096203848153924, | |
| "grad_norm": 0.0014049686724320054, | |
| "learning_rate": 4.201848495351042e-05, | |
| "loss": 9.610663168132306e-05, | |
| "step": 23150 | |
| }, | |
| { | |
| "epoch": 3.0128705148205928, | |
| "grad_norm": 0.002500481205061078, | |
| "learning_rate": 4.1995310144109925e-05, | |
| "loss": 9.441668167710304e-05, | |
| "step": 23175 | |
| }, | |
| { | |
| "epoch": 3.016120644825793, | |
| "grad_norm": 0.0015667990082874894, | |
| "learning_rate": 4.1972108149943445e-05, | |
| "loss": 9.540068916976452e-05, | |
| "step": 23200 | |
| }, | |
| { | |
| "epoch": 3.019370774830993, | |
| "grad_norm": 0.002018737606704235, | |
| "learning_rate": 4.194887900812361e-05, | |
| "loss": 0.00012376774102449418, | |
| "step": 23225 | |
| }, | |
| { | |
| "epoch": 3.0226209048361934, | |
| "grad_norm": 0.0009505741763859987, | |
| "learning_rate": 4.19256227558065e-05, | |
| "loss": 9.603582322597503e-05, | |
| "step": 23250 | |
| }, | |
| { | |
| "epoch": 3.0258710348413937, | |
| "grad_norm": 0.0012013164814561605, | |
| "learning_rate": 4.190233943019152e-05, | |
| "loss": 0.00010708124376833439, | |
| "step": 23275 | |
| }, | |
| { | |
| "epoch": 3.0291211648465937, | |
| "grad_norm": 0.0015938347205519676, | |
| "learning_rate": 4.187902906852141e-05, | |
| "loss": 0.00011160518042743205, | |
| "step": 23300 | |
| }, | |
| { | |
| "epoch": 3.032371294851794, | |
| "grad_norm": 0.0018334237392991781, | |
| "learning_rate": 4.185569170808215e-05, | |
| "loss": 0.00011199971660971641, | |
| "step": 23325 | |
| }, | |
| { | |
| "epoch": 3.0356214248569944, | |
| "grad_norm": 0.0013979518553242087, | |
| "learning_rate": 4.1832327386202887e-05, | |
| "loss": 0.00010286960750818252, | |
| "step": 23350 | |
| }, | |
| { | |
| "epoch": 3.0388715548621943, | |
| "grad_norm": 0.002501849550753832, | |
| "learning_rate": 4.1808936140255914e-05, | |
| "loss": 9.838162921369076e-05, | |
| "step": 23375 | |
| }, | |
| { | |
| "epoch": 3.0421216848673946, | |
| "grad_norm": 0.002554062521085143, | |
| "learning_rate": 4.17855180076566e-05, | |
| "loss": 9.16569959372282e-05, | |
| "step": 23400 | |
| }, | |
| { | |
| "epoch": 3.045371814872595, | |
| "grad_norm": 0.0012999202590435743, | |
| "learning_rate": 4.176207302586329e-05, | |
| "loss": 9.508879855275155e-05, | |
| "step": 23425 | |
| }, | |
| { | |
| "epoch": 3.048621944877795, | |
| "grad_norm": 0.006127315107733011, | |
| "learning_rate": 4.173860123237729e-05, | |
| "loss": 0.0001099599339067936, | |
| "step": 23450 | |
| }, | |
| { | |
| "epoch": 3.0518720748829953, | |
| "grad_norm": 0.0009308779845014215, | |
| "learning_rate": 4.171510266474281e-05, | |
| "loss": 0.00012838857248425483, | |
| "step": 23475 | |
| }, | |
| { | |
| "epoch": 3.0551222048881956, | |
| "grad_norm": 0.0015354519709944725, | |
| "learning_rate": 4.1691577360546855e-05, | |
| "loss": 0.00011543597094714641, | |
| "step": 23500 | |
| }, | |
| { | |
| "epoch": 3.0583723348933955, | |
| "grad_norm": 0.005569889675825834, | |
| "learning_rate": 4.1668025357419216e-05, | |
| "loss": 0.00010665758512914181, | |
| "step": 23525 | |
| }, | |
| { | |
| "epoch": 3.061622464898596, | |
| "grad_norm": 0.0009565159562043846, | |
| "learning_rate": 4.164444669303238e-05, | |
| "loss": 0.00012459917925298215, | |
| "step": 23550 | |
| }, | |
| { | |
| "epoch": 3.0648725949037963, | |
| "grad_norm": 0.002009939169511199, | |
| "learning_rate": 4.1620841405101495e-05, | |
| "loss": 0.00010589437559247017, | |
| "step": 23575 | |
| }, | |
| { | |
| "epoch": 3.068122724908996, | |
| "grad_norm": 0.002162981778383255, | |
| "learning_rate": 4.159720953138428e-05, | |
| "loss": 0.0001057141087949276, | |
| "step": 23600 | |
| }, | |
| { | |
| "epoch": 3.0713728549141965, | |
| "grad_norm": 0.004441654309630394, | |
| "learning_rate": 4.1573551109680986e-05, | |
| "loss": 0.00010300349444150925, | |
| "step": 23625 | |
| }, | |
| { | |
| "epoch": 3.074622984919397, | |
| "grad_norm": 0.0010740088764578104, | |
| "learning_rate": 4.154986617783433e-05, | |
| "loss": 0.00010108290240168572, | |
| "step": 23650 | |
| }, | |
| { | |
| "epoch": 3.077873114924597, | |
| "grad_norm": 0.0015953021356835961, | |
| "learning_rate": 4.152615477372941e-05, | |
| "loss": 0.00011523210443556309, | |
| "step": 23675 | |
| }, | |
| { | |
| "epoch": 3.081123244929797, | |
| "grad_norm": 0.001653257291764021, | |
| "learning_rate": 4.1502416935293714e-05, | |
| "loss": 0.00010464210063219071, | |
| "step": 23700 | |
| }, | |
| { | |
| "epoch": 3.0843733749349975, | |
| "grad_norm": 0.0021358870435506105, | |
| "learning_rate": 4.147865270049698e-05, | |
| "loss": 9.777233004570008e-05, | |
| "step": 23725 | |
| }, | |
| { | |
| "epoch": 3.0876235049401974, | |
| "grad_norm": 0.0009592963615432382, | |
| "learning_rate": 4.145486210735116e-05, | |
| "loss": 8.792969398200511e-05, | |
| "step": 23750 | |
| }, | |
| { | |
| "epoch": 3.090873634945398, | |
| "grad_norm": 0.0017140634590759873, | |
| "learning_rate": 4.143104519391039e-05, | |
| "loss": 0.00010282695293426514, | |
| "step": 23775 | |
| }, | |
| { | |
| "epoch": 3.094123764950598, | |
| "grad_norm": 0.0020951353944838047, | |
| "learning_rate": 4.1407201998270905e-05, | |
| "loss": 0.00010960533283650875, | |
| "step": 23800 | |
| }, | |
| { | |
| "epoch": 3.097373894955798, | |
| "grad_norm": 0.0013650343753397465, | |
| "learning_rate": 4.1383332558570966e-05, | |
| "loss": 0.00010544072836637497, | |
| "step": 23825 | |
| }, | |
| { | |
| "epoch": 3.1006240249609984, | |
| "grad_norm": 0.0005820131045766175, | |
| "learning_rate": 4.135943691299082e-05, | |
| "loss": 0.00010747028514742851, | |
| "step": 23850 | |
| }, | |
| { | |
| "epoch": 3.103874154966199, | |
| "grad_norm": 0.0009744393755681813, | |
| "learning_rate": 4.133551509975264e-05, | |
| "loss": 9.915635921061039e-05, | |
| "step": 23875 | |
| }, | |
| { | |
| "epoch": 3.1071242849713987, | |
| "grad_norm": 0.001790615962818265, | |
| "learning_rate": 4.131156715712043e-05, | |
| "loss": 9.722976945340633e-05, | |
| "step": 23900 | |
| }, | |
| { | |
| "epoch": 3.110374414976599, | |
| "grad_norm": 0.001490483758971095, | |
| "learning_rate": 4.1287593123400024e-05, | |
| "loss": 9.300995618104934e-05, | |
| "step": 23925 | |
| }, | |
| { | |
| "epoch": 3.1136245449817994, | |
| "grad_norm": 0.001756942830979824, | |
| "learning_rate": 4.126359303693896e-05, | |
| "loss": 0.0001124845165759325, | |
| "step": 23950 | |
| }, | |
| { | |
| "epoch": 3.1168746749869993, | |
| "grad_norm": 0.0029301128815859556, | |
| "learning_rate": 4.123956693612646e-05, | |
| "loss": 0.0001009349711239338, | |
| "step": 23975 | |
| }, | |
| { | |
| "epoch": 3.1201248049921997, | |
| "grad_norm": 0.0005423895199783146, | |
| "learning_rate": 4.1215514859393364e-05, | |
| "loss": 9.993207640945911e-05, | |
| "step": 24000 | |
| }, | |
| { | |
| "epoch": 3.1233749349974, | |
| "grad_norm": 0.0011979271657764912, | |
| "learning_rate": 4.119143684521204e-05, | |
| "loss": 0.000112836305052042, | |
| "step": 24025 | |
| }, | |
| { | |
| "epoch": 3.1266250650026, | |
| "grad_norm": 0.0007727962802164257, | |
| "learning_rate": 4.116733293209638e-05, | |
| "loss": 8.280262351036072e-05, | |
| "step": 24050 | |
| }, | |
| { | |
| "epoch": 3.1298751950078003, | |
| "grad_norm": 0.002150925574824214, | |
| "learning_rate": 4.1143203158601665e-05, | |
| "loss": 9.501841850578785e-05, | |
| "step": 24075 | |
| }, | |
| { | |
| "epoch": 3.1331253250130007, | |
| "grad_norm": 0.0019343640888109803, | |
| "learning_rate": 4.1119047563324566e-05, | |
| "loss": 9.378094226121903e-05, | |
| "step": 24100 | |
| }, | |
| { | |
| "epoch": 3.1363754550182006, | |
| "grad_norm": 0.003619915572926402, | |
| "learning_rate": 4.109486618490305e-05, | |
| "loss": 0.00010194934904575348, | |
| "step": 24125 | |
| }, | |
| { | |
| "epoch": 3.139625585023401, | |
| "grad_norm": 0.0018030067440122366, | |
| "learning_rate": 4.107065906201631e-05, | |
| "loss": 0.00010059256106615067, | |
| "step": 24150 | |
| }, | |
| { | |
| "epoch": 3.139625585023401, | |
| "eval_loss": 0.0003573755093384534, | |
| "eval_runtime": 9.2535, | |
| "eval_samples_per_second": 540.334, | |
| "eval_steps_per_second": 4.323, | |
| "step": 24150 | |
| }, | |
| { | |
| "epoch": 3.139625585023401, | |
| "eval_holdout_avg_strict_f1": 0.1397550944220317, | |
| "eval_holdout_enron_spam_bvk_strict_f1": 0.0, | |
| "eval_holdout_enron_spam_setfit_strict_f1": 0.0, | |
| "eval_holdout_gliner2_pii_ptbr_reward_split_strict_f1": 0.48018769551616264, | |
| "eval_holdout_nemotron_pii_strict_f1": 0.5093420574336606, | |
| "eval_holdout_open_pii_masking_500k_strict_f1": 0.2637963437553391, | |
| "eval_holdout_phishing_darkknight_strict_f1": 0.0, | |
| "eval_holdout_phishing_zefang_strict_f1": 0.0, | |
| "eval_holdout_pii_masking_400k_strict_f1": 0.28397994193718656, | |
| "eval_holdout_sms_spam_multilingual_strict_f1": 0.0, | |
| "eval_holdout_spam_messages_mshenoda_strict_f1": 0.0, | |
| "eval_holdout_spamassassin_strict_f1": 0.0, | |
| "step": 24150 | |
| } | |
| ], | |
| "logging_steps": 25, | |
| "max_steps": 69000, | |
| "num_input_tokens_seen": 0, | |
| "num_train_epochs": 9, | |
| "save_steps": 3450, | |
| "stateful_callbacks": { | |
| "TrainerControl": { | |
| "args": { | |
| "should_epoch_stop": false, | |
| "should_evaluate": false, | |
| "should_log": false, | |
| "should_save": true, | |
| "should_training_stop": false | |
| }, | |
| "attributes": {} | |
| } | |
| }, | |
| "total_flos": 0.0, | |
| "train_batch_size": 128, | |
| "trial_name": null, | |
| "trial_params": null | |
| } | |