diff --git "a/trainer_state.json" "b/trainer_state.json" new file mode 100644--- /dev/null +++ "b/trainer_state.json" @@ -0,0 +1,4933 @@ +{ + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 20.771513353115726, + "eval_steps": 500, + "global_step": 7000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.02967359050445104, + "grad_norm": 1.9033336639404297, + "learning_rate": 2.0000000000000003e-06, + "loss": 0.2317, + "step": 10 + }, + { + "epoch": 0.05934718100890208, + "grad_norm": 1.0404284000396729, + "learning_rate": 4.000000000000001e-06, + "loss": 0.1977, + "step": 20 + }, + { + "epoch": 0.08902077151335312, + "grad_norm": 0.6579734683036804, + "learning_rate": 6e-06, + "loss": 0.1451, + "step": 30 + }, + { + "epoch": 0.11869436201780416, + "grad_norm": 0.33155006170272827, + "learning_rate": 8.000000000000001e-06, + "loss": 0.0959, + "step": 40 + }, + { + "epoch": 0.14836795252225518, + "grad_norm": 0.5317391753196716, + "learning_rate": 1e-05, + "loss": 0.0828, + "step": 50 + }, + { + "epoch": 0.17804154302670624, + "grad_norm": 0.45179909467697144, + "learning_rate": 1.2e-05, + "loss": 0.0814, + "step": 60 + }, + { + "epoch": 0.20771513353115728, + "grad_norm": 0.2707938849925995, + "learning_rate": 1.4000000000000001e-05, + "loss": 0.0562, + "step": 70 + }, + { + "epoch": 0.23738872403560832, + "grad_norm": 0.22402559220790863, + "learning_rate": 1.6000000000000003e-05, + "loss": 0.0594, + "step": 80 + }, + { + "epoch": 0.26706231454005935, + "grad_norm": 0.16533811390399933, + "learning_rate": 1.8e-05, + "loss": 0.0529, + "step": 90 + }, + { + "epoch": 0.29673590504451036, + "grad_norm": 0.222530797123909, + "learning_rate": 2e-05, + "loss": 0.0522, + "step": 100 + }, + { + "epoch": 0.3264094955489614, + "grad_norm": 0.1894129067659378, + "learning_rate": 2.2000000000000003e-05, + "loss": 0.0514, + "step": 110 + }, + { + "epoch": 0.3560830860534125, + "grad_norm": 0.20559543371200562, + "learning_rate": 2.4e-05, + "loss": 0.0462, + "step": 120 + }, + { + "epoch": 0.3857566765578635, + "grad_norm": 0.157830610871315, + "learning_rate": 2.6000000000000002e-05, + "loss": 0.0471, + "step": 130 + }, + { + "epoch": 0.41543026706231456, + "grad_norm": 0.14663924276828766, + "learning_rate": 2.8000000000000003e-05, + "loss": 0.0449, + "step": 140 + }, + { + "epoch": 0.44510385756676557, + "grad_norm": 0.14772620797157288, + "learning_rate": 3e-05, + "loss": 0.0424, + "step": 150 + }, + { + "epoch": 0.47477744807121663, + "grad_norm": 0.16058433055877686, + "learning_rate": 3.2000000000000005e-05, + "loss": 0.0424, + "step": 160 + }, + { + "epoch": 0.5044510385756676, + "grad_norm": 0.15857172012329102, + "learning_rate": 3.4000000000000007e-05, + "loss": 0.041, + "step": 170 + }, + { + "epoch": 0.5341246290801187, + "grad_norm": 0.17435680329799652, + "learning_rate": 3.6e-05, + "loss": 0.0408, + "step": 180 + }, + { + "epoch": 0.5637982195845698, + "grad_norm": 0.1439993977546692, + "learning_rate": 3.8e-05, + "loss": 0.0352, + "step": 190 + }, + { + "epoch": 0.5934718100890207, + "grad_norm": 0.15629075467586517, + "learning_rate": 4e-05, + "loss": 0.0383, + "step": 200 + }, + { + "epoch": 0.6231454005934718, + "grad_norm": 0.1610369235277176, + "learning_rate": 4.2e-05, + "loss": 0.0392, + "step": 210 + }, + { + "epoch": 0.6528189910979229, + "grad_norm": 0.17589861154556274, + "learning_rate": 4.4000000000000006e-05, + "loss": 0.0374, + "step": 220 + }, + { + "epoch": 0.6824925816023739, + "grad_norm": 0.19186066091060638, + "learning_rate": 4.600000000000001e-05, + "loss": 0.0358, + "step": 230 + }, + { + "epoch": 0.712166172106825, + "grad_norm": 0.1579175740480423, + "learning_rate": 4.8e-05, + "loss": 0.0357, + "step": 240 + }, + { + "epoch": 0.7418397626112759, + "grad_norm": 0.17220136523246765, + "learning_rate": 5e-05, + "loss": 0.0334, + "step": 250 + }, + { + "epoch": 0.771513353115727, + "grad_norm": 0.18591266870498657, + "learning_rate": 5.2000000000000004e-05, + "loss": 0.0315, + "step": 260 + }, + { + "epoch": 0.8011869436201781, + "grad_norm": 0.2341579794883728, + "learning_rate": 5.4000000000000005e-05, + "loss": 0.0375, + "step": 270 + }, + { + "epoch": 0.8308605341246291, + "grad_norm": 0.15227168798446655, + "learning_rate": 5.6000000000000006e-05, + "loss": 0.031, + "step": 280 + }, + { + "epoch": 0.8605341246290801, + "grad_norm": 0.1876339167356491, + "learning_rate": 5.8e-05, + "loss": 0.0371, + "step": 290 + }, + { + "epoch": 0.8902077151335311, + "grad_norm": 0.1789393573999405, + "learning_rate": 6e-05, + "loss": 0.0313, + "step": 300 + }, + { + "epoch": 0.9198813056379822, + "grad_norm": 0.1678636074066162, + "learning_rate": 6.2e-05, + "loss": 0.0349, + "step": 310 + }, + { + "epoch": 0.9495548961424333, + "grad_norm": 0.17457032203674316, + "learning_rate": 6.400000000000001e-05, + "loss": 0.0296, + "step": 320 + }, + { + "epoch": 0.9792284866468842, + "grad_norm": 0.14290577173233032, + "learning_rate": 6.6e-05, + "loss": 0.0308, + "step": 330 + }, + { + "epoch": 1.0089020771513353, + "grad_norm": 0.23601128160953522, + "learning_rate": 6.800000000000001e-05, + "loss": 0.0298, + "step": 340 + }, + { + "epoch": 1.0385756676557865, + "grad_norm": 0.14039042592048645, + "learning_rate": 7e-05, + "loss": 0.0262, + "step": 350 + }, + { + "epoch": 1.0682492581602374, + "grad_norm": 0.1804966777563095, + "learning_rate": 7.2e-05, + "loss": 0.0284, + "step": 360 + }, + { + "epoch": 1.0979228486646884, + "grad_norm": 0.22986947000026703, + "learning_rate": 7.4e-05, + "loss": 0.0308, + "step": 370 + }, + { + "epoch": 1.1275964391691395, + "grad_norm": 0.20188020169734955, + "learning_rate": 7.6e-05, + "loss": 0.0261, + "step": 380 + }, + { + "epoch": 1.1572700296735905, + "grad_norm": 0.14067409932613373, + "learning_rate": 7.800000000000001e-05, + "loss": 0.028, + "step": 390 + }, + { + "epoch": 1.1869436201780414, + "grad_norm": 0.16516339778900146, + "learning_rate": 8e-05, + "loss": 0.0247, + "step": 400 + }, + { + "epoch": 1.2166172106824926, + "grad_norm": 0.19918474555015564, + "learning_rate": 8.2e-05, + "loss": 0.0301, + "step": 410 + }, + { + "epoch": 1.2462908011869436, + "grad_norm": 0.1878385990858078, + "learning_rate": 8.4e-05, + "loss": 0.0251, + "step": 420 + }, + { + "epoch": 1.2759643916913945, + "grad_norm": 0.20107118785381317, + "learning_rate": 8.6e-05, + "loss": 0.0279, + "step": 430 + }, + { + "epoch": 1.3056379821958457, + "grad_norm": 0.24616649746894836, + "learning_rate": 8.800000000000001e-05, + "loss": 0.0259, + "step": 440 + }, + { + "epoch": 1.3353115727002967, + "grad_norm": 0.19029636681079865, + "learning_rate": 9e-05, + "loss": 0.0262, + "step": 450 + }, + { + "epoch": 1.3649851632047478, + "grad_norm": 0.194508358836174, + "learning_rate": 9.200000000000001e-05, + "loss": 0.0275, + "step": 460 + }, + { + "epoch": 1.3946587537091988, + "grad_norm": 0.20826251804828644, + "learning_rate": 9.4e-05, + "loss": 0.0289, + "step": 470 + }, + { + "epoch": 1.4243323442136497, + "grad_norm": 0.13222843408584595, + "learning_rate": 9.6e-05, + "loss": 0.0249, + "step": 480 + }, + { + "epoch": 1.454005934718101, + "grad_norm": 0.13967235386371613, + "learning_rate": 9.8e-05, + "loss": 0.0231, + "step": 490 + }, + { + "epoch": 1.4836795252225519, + "grad_norm": 0.21556402742862701, + "learning_rate": 0.0001, + "loss": 0.0232, + "step": 500 + }, + { + "epoch": 1.513353115727003, + "grad_norm": 0.2407234013080597, + "learning_rate": 9.999972660400536e-05, + "loss": 0.025, + "step": 510 + }, + { + "epoch": 1.543026706231454, + "grad_norm": 0.1544090360403061, + "learning_rate": 9.999890641901125e-05, + "loss": 0.0224, + "step": 520 + }, + { + "epoch": 1.572700296735905, + "grad_norm": 0.1930345594882965, + "learning_rate": 9.999753945398704e-05, + "loss": 0.0244, + "step": 530 + }, + { + "epoch": 1.6023738872403561, + "grad_norm": 0.2288358211517334, + "learning_rate": 9.99956257238817e-05, + "loss": 0.0223, + "step": 540 + }, + { + "epoch": 1.632047477744807, + "grad_norm": 0.2028588354587555, + "learning_rate": 9.999316524962345e-05, + "loss": 0.022, + "step": 550 + }, + { + "epoch": 1.6617210682492582, + "grad_norm": 0.17989283800125122, + "learning_rate": 9.999015805811965e-05, + "loss": 0.0201, + "step": 560 + }, + { + "epoch": 1.6913946587537092, + "grad_norm": 0.20576386153697968, + "learning_rate": 9.998660418225645e-05, + "loss": 0.0211, + "step": 570 + }, + { + "epoch": 1.7210682492581602, + "grad_norm": 0.1263924539089203, + "learning_rate": 9.998250366089848e-05, + "loss": 0.0205, + "step": 580 + }, + { + "epoch": 1.7507418397626113, + "grad_norm": 0.23239193856716156, + "learning_rate": 9.997785653888835e-05, + "loss": 0.0223, + "step": 590 + }, + { + "epoch": 1.7804154302670623, + "grad_norm": 0.15964201092720032, + "learning_rate": 9.997266286704631e-05, + "loss": 0.0236, + "step": 600 + }, + { + "epoch": 1.8100890207715135, + "grad_norm": 0.1998339742422104, + "learning_rate": 9.996692270216947e-05, + "loss": 0.0254, + "step": 610 + }, + { + "epoch": 1.8397626112759644, + "grad_norm": 0.25945162773132324, + "learning_rate": 9.996063610703137e-05, + "loss": 0.0207, + "step": 620 + }, + { + "epoch": 1.8694362017804154, + "grad_norm": 0.1687825620174408, + "learning_rate": 9.995380315038119e-05, + "loss": 0.0169, + "step": 630 + }, + { + "epoch": 1.8991097922848663, + "grad_norm": 0.13454881310462952, + "learning_rate": 9.994642390694308e-05, + "loss": 0.0185, + "step": 640 + }, + { + "epoch": 1.9287833827893175, + "grad_norm": 0.1880808174610138, + "learning_rate": 9.993849845741524e-05, + "loss": 0.0199, + "step": 650 + }, + { + "epoch": 1.9584569732937687, + "grad_norm": 0.17725640535354614, + "learning_rate": 9.993002688846913e-05, + "loss": 0.019, + "step": 660 + }, + { + "epoch": 1.9881305637982196, + "grad_norm": 0.17680853605270386, + "learning_rate": 9.992100929274846e-05, + "loss": 0.0257, + "step": 670 + }, + { + "epoch": 2.0178041543026706, + "grad_norm": 0.18133607506752014, + "learning_rate": 9.991144576886823e-05, + "loss": 0.0222, + "step": 680 + }, + { + "epoch": 2.0474777448071215, + "grad_norm": 0.21174193918704987, + "learning_rate": 9.990133642141359e-05, + "loss": 0.0192, + "step": 690 + }, + { + "epoch": 2.077151335311573, + "grad_norm": 0.15766288340091705, + "learning_rate": 9.989068136093873e-05, + "loss": 0.0199, + "step": 700 + }, + { + "epoch": 2.106824925816024, + "grad_norm": 0.19250448048114777, + "learning_rate": 9.987948070396571e-05, + "loss": 0.0252, + "step": 710 + }, + { + "epoch": 2.136498516320475, + "grad_norm": 0.1412709653377533, + "learning_rate": 9.986773457298311e-05, + "loss": 0.0186, + "step": 720 + }, + { + "epoch": 2.166172106824926, + "grad_norm": 0.2136259824037552, + "learning_rate": 9.985544309644475e-05, + "loss": 0.0204, + "step": 730 + }, + { + "epoch": 2.1958456973293767, + "grad_norm": 0.10815251618623734, + "learning_rate": 9.984260640876821e-05, + "loss": 0.0153, + "step": 740 + }, + { + "epoch": 2.2255192878338277, + "grad_norm": 0.14663298428058624, + "learning_rate": 9.98292246503335e-05, + "loss": 0.0189, + "step": 750 + }, + { + "epoch": 2.255192878338279, + "grad_norm": 0.18529076874256134, + "learning_rate": 9.981529796748134e-05, + "loss": 0.0155, + "step": 760 + }, + { + "epoch": 2.28486646884273, + "grad_norm": 0.2349974811077118, + "learning_rate": 9.980082651251175e-05, + "loss": 0.0172, + "step": 770 + }, + { + "epoch": 2.314540059347181, + "grad_norm": 0.10778886079788208, + "learning_rate": 9.97858104436822e-05, + "loss": 0.0161, + "step": 780 + }, + { + "epoch": 2.344213649851632, + "grad_norm": 0.15675969421863556, + "learning_rate": 9.977024992520602e-05, + "loss": 0.0165, + "step": 790 + }, + { + "epoch": 2.373887240356083, + "grad_norm": 0.23468513786792755, + "learning_rate": 9.975414512725057e-05, + "loss": 0.0196, + "step": 800 + }, + { + "epoch": 2.4035608308605343, + "grad_norm": 0.1332869678735733, + "learning_rate": 9.973749622593534e-05, + "loss": 0.0193, + "step": 810 + }, + { + "epoch": 2.4332344213649852, + "grad_norm": 0.1406887024641037, + "learning_rate": 9.972030340333001e-05, + "loss": 0.0186, + "step": 820 + }, + { + "epoch": 2.462908011869436, + "grad_norm": 0.11544730514287949, + "learning_rate": 9.970256684745258e-05, + "loss": 0.0195, + "step": 830 + }, + { + "epoch": 2.492581602373887, + "grad_norm": 0.19476240873336792, + "learning_rate": 9.968428675226714e-05, + "loss": 0.0171, + "step": 840 + }, + { + "epoch": 2.5222551928783385, + "grad_norm": 0.22309833765029907, + "learning_rate": 9.966546331768191e-05, + "loss": 0.0223, + "step": 850 + }, + { + "epoch": 2.551928783382789, + "grad_norm": 0.2214643657207489, + "learning_rate": 9.964609674954696e-05, + "loss": 0.019, + "step": 860 + }, + { + "epoch": 2.5816023738872405, + "grad_norm": 0.2298765480518341, + "learning_rate": 9.962618725965196e-05, + "loss": 0.0178, + "step": 870 + }, + { + "epoch": 2.6112759643916914, + "grad_norm": 0.1542595475912094, + "learning_rate": 9.96057350657239e-05, + "loss": 0.0152, + "step": 880 + }, + { + "epoch": 2.6409495548961424, + "grad_norm": 0.12994691729545593, + "learning_rate": 9.95847403914247e-05, + "loss": 0.0153, + "step": 890 + }, + { + "epoch": 2.6706231454005933, + "grad_norm": 0.1726643294095993, + "learning_rate": 9.956320346634876e-05, + "loss": 0.0219, + "step": 900 + }, + { + "epoch": 2.7002967359050443, + "grad_norm": 0.18292242288589478, + "learning_rate": 9.954112452602045e-05, + "loss": 0.0137, + "step": 910 + }, + { + "epoch": 2.7299703264094957, + "grad_norm": 0.19749417901039124, + "learning_rate": 9.95185038118915e-05, + "loss": 0.0179, + "step": 920 + }, + { + "epoch": 2.7596439169139466, + "grad_norm": 0.17808304727077484, + "learning_rate": 9.949534157133844e-05, + "loss": 0.0155, + "step": 930 + }, + { + "epoch": 2.7893175074183976, + "grad_norm": 0.13589969277381897, + "learning_rate": 9.94716380576598e-05, + "loss": 0.0143, + "step": 940 + }, + { + "epoch": 2.8189910979228485, + "grad_norm": 0.17047159373760223, + "learning_rate": 9.944739353007344e-05, + "loss": 0.0211, + "step": 950 + }, + { + "epoch": 2.8486646884272995, + "grad_norm": 0.15535619854927063, + "learning_rate": 9.942260825371358e-05, + "loss": 0.0132, + "step": 960 + }, + { + "epoch": 2.878338278931751, + "grad_norm": 0.1609240472316742, + "learning_rate": 9.939728249962807e-05, + "loss": 0.0145, + "step": 970 + }, + { + "epoch": 2.908011869436202, + "grad_norm": 0.20487530529499054, + "learning_rate": 9.937141654477528e-05, + "loss": 0.0172, + "step": 980 + }, + { + "epoch": 2.9376854599406528, + "grad_norm": 0.12371553480625153, + "learning_rate": 9.934501067202117e-05, + "loss": 0.0191, + "step": 990 + }, + { + "epoch": 2.9673590504451037, + "grad_norm": 0.15513016283512115, + "learning_rate": 9.931806517013612e-05, + "loss": 0.0158, + "step": 1000 + }, + { + "epoch": 2.9970326409495547, + "grad_norm": 0.1723584234714508, + "learning_rate": 9.929058033379181e-05, + "loss": 0.0139, + "step": 1010 + }, + { + "epoch": 3.026706231454006, + "grad_norm": 0.1482209414243698, + "learning_rate": 9.926255646355804e-05, + "loss": 0.0148, + "step": 1020 + }, + { + "epoch": 3.056379821958457, + "grad_norm": 0.15559454262256622, + "learning_rate": 9.923399386589933e-05, + "loss": 0.0147, + "step": 1030 + }, + { + "epoch": 3.086053412462908, + "grad_norm": 0.22350917756557465, + "learning_rate": 9.92048928531717e-05, + "loss": 0.0157, + "step": 1040 + }, + { + "epoch": 3.115727002967359, + "grad_norm": 0.1581738144159317, + "learning_rate": 9.917525374361912e-05, + "loss": 0.0133, + "step": 1050 + }, + { + "epoch": 3.14540059347181, + "grad_norm": 0.20087914168834686, + "learning_rate": 9.914507686137019e-05, + "loss": 0.0208, + "step": 1060 + }, + { + "epoch": 3.1750741839762613, + "grad_norm": 0.16003265976905823, + "learning_rate": 9.911436253643445e-05, + "loss": 0.0168, + "step": 1070 + }, + { + "epoch": 3.2047477744807122, + "grad_norm": 0.15235169231891632, + "learning_rate": 9.90831111046988e-05, + "loss": 0.0168, + "step": 1080 + }, + { + "epoch": 3.234421364985163, + "grad_norm": 0.15660665929317474, + "learning_rate": 9.905132290792394e-05, + "loss": 0.0106, + "step": 1090 + }, + { + "epoch": 3.264094955489614, + "grad_norm": 0.16853424906730652, + "learning_rate": 9.901899829374047e-05, + "loss": 0.0149, + "step": 1100 + }, + { + "epoch": 3.293768545994065, + "grad_norm": 0.1335846185684204, + "learning_rate": 9.89861376156452e-05, + "loss": 0.0168, + "step": 1110 + }, + { + "epoch": 3.3234421364985165, + "grad_norm": 0.20238997042179108, + "learning_rate": 9.895274123299723e-05, + "loss": 0.0154, + "step": 1120 + }, + { + "epoch": 3.3531157270029674, + "grad_norm": 0.22216491401195526, + "learning_rate": 9.891880951101407e-05, + "loss": 0.019, + "step": 1130 + }, + { + "epoch": 3.3827893175074184, + "grad_norm": 0.2017626017332077, + "learning_rate": 9.888434282076758e-05, + "loss": 0.017, + "step": 1140 + }, + { + "epoch": 3.4124629080118694, + "grad_norm": 0.18049117922782898, + "learning_rate": 9.884934153917997e-05, + "loss": 0.0163, + "step": 1150 + }, + { + "epoch": 3.4421364985163203, + "grad_norm": 0.28145721554756165, + "learning_rate": 9.881380604901964e-05, + "loss": 0.0166, + "step": 1160 + }, + { + "epoch": 3.4718100890207717, + "grad_norm": 0.2356300801038742, + "learning_rate": 9.877773673889701e-05, + "loss": 0.0155, + "step": 1170 + }, + { + "epoch": 3.5014836795252227, + "grad_norm": 0.24113395810127258, + "learning_rate": 9.87411340032603e-05, + "loss": 0.0156, + "step": 1180 + }, + { + "epoch": 3.5311572700296736, + "grad_norm": 0.18665863573551178, + "learning_rate": 9.870399824239117e-05, + "loss": 0.0159, + "step": 1190 + }, + { + "epoch": 3.5608308605341246, + "grad_norm": 0.16171567142009735, + "learning_rate": 9.86663298624003e-05, + "loss": 0.0167, + "step": 1200 + }, + { + "epoch": 3.5905044510385755, + "grad_norm": 0.17315839231014252, + "learning_rate": 9.862812927522309e-05, + "loss": 0.017, + "step": 1210 + }, + { + "epoch": 3.620178041543027, + "grad_norm": 0.14727933704853058, + "learning_rate": 9.858939689861506e-05, + "loss": 0.0132, + "step": 1220 + }, + { + "epoch": 3.649851632047478, + "grad_norm": 0.1552547812461853, + "learning_rate": 9.855013315614725e-05, + "loss": 0.0175, + "step": 1230 + }, + { + "epoch": 3.679525222551929, + "grad_norm": 0.1715100109577179, + "learning_rate": 9.851033847720166e-05, + "loss": 0.0139, + "step": 1240 + }, + { + "epoch": 3.7091988130563798, + "grad_norm": 0.16414763033390045, + "learning_rate": 9.847001329696653e-05, + "loss": 0.0189, + "step": 1250 + }, + { + "epoch": 3.7388724035608307, + "grad_norm": 0.1251063048839569, + "learning_rate": 9.842915805643155e-05, + "loss": 0.0145, + "step": 1260 + }, + { + "epoch": 3.768545994065282, + "grad_norm": 0.17011059820652008, + "learning_rate": 9.838777320238312e-05, + "loss": 0.0161, + "step": 1270 + }, + { + "epoch": 3.798219584569733, + "grad_norm": 0.14429537951946259, + "learning_rate": 9.834585918739936e-05, + "loss": 0.0159, + "step": 1280 + }, + { + "epoch": 3.827893175074184, + "grad_norm": 0.138567715883255, + "learning_rate": 9.830341646984521e-05, + "loss": 0.0175, + "step": 1290 + }, + { + "epoch": 3.857566765578635, + "grad_norm": 0.08295896649360657, + "learning_rate": 9.826044551386744e-05, + "loss": 0.0145, + "step": 1300 + }, + { + "epoch": 3.887240356083086, + "grad_norm": 0.0911448523402214, + "learning_rate": 9.821694678938953e-05, + "loss": 0.0134, + "step": 1310 + }, + { + "epoch": 3.9169139465875373, + "grad_norm": 0.14157798886299133, + "learning_rate": 9.817292077210659e-05, + "loss": 0.0158, + "step": 1320 + }, + { + "epoch": 3.9465875370919883, + "grad_norm": 0.17415288090705872, + "learning_rate": 9.812836794348004e-05, + "loss": 0.0125, + "step": 1330 + }, + { + "epoch": 3.9762611275964392, + "grad_norm": 0.22007249295711517, + "learning_rate": 9.808328879073251e-05, + "loss": 0.0153, + "step": 1340 + }, + { + "epoch": 4.005934718100891, + "grad_norm": 0.144961416721344, + "learning_rate": 9.803768380684242e-05, + "loss": 0.0117, + "step": 1350 + }, + { + "epoch": 4.035608308605341, + "grad_norm": 0.14304885268211365, + "learning_rate": 9.799155349053851e-05, + "loss": 0.0138, + "step": 1360 + }, + { + "epoch": 4.0652818991097925, + "grad_norm": 0.18843571841716766, + "learning_rate": 9.794489834629455e-05, + "loss": 0.0106, + "step": 1370 + }, + { + "epoch": 4.094955489614243, + "grad_norm": 0.16858817636966705, + "learning_rate": 9.789771888432375e-05, + "loss": 0.0138, + "step": 1380 + }, + { + "epoch": 4.1246290801186944, + "grad_norm": 0.19177594780921936, + "learning_rate": 9.785001562057309e-05, + "loss": 0.0114, + "step": 1390 + }, + { + "epoch": 4.154302670623146, + "grad_norm": 0.20158767700195312, + "learning_rate": 9.780178907671789e-05, + "loss": 0.0146, + "step": 1400 + }, + { + "epoch": 4.183976261127596, + "grad_norm": 0.17675232887268066, + "learning_rate": 9.775303978015585e-05, + "loss": 0.0116, + "step": 1410 + }, + { + "epoch": 4.213649851632048, + "grad_norm": 0.20077385008335114, + "learning_rate": 9.77037682640015e-05, + "loss": 0.0172, + "step": 1420 + }, + { + "epoch": 4.243323442136498, + "grad_norm": 0.17185665667057037, + "learning_rate": 9.765397506708023e-05, + "loss": 0.0138, + "step": 1430 + }, + { + "epoch": 4.27299703264095, + "grad_norm": 0.1641971468925476, + "learning_rate": 9.760366073392246e-05, + "loss": 0.0145, + "step": 1440 + }, + { + "epoch": 4.302670623145401, + "grad_norm": 0.13757102191448212, + "learning_rate": 9.755282581475769e-05, + "loss": 0.0158, + "step": 1450 + }, + { + "epoch": 4.332344213649852, + "grad_norm": 0.18012432754039764, + "learning_rate": 9.750147086550844e-05, + "loss": 0.0139, + "step": 1460 + }, + { + "epoch": 4.362017804154303, + "grad_norm": 0.1639927178621292, + "learning_rate": 9.744959644778422e-05, + "loss": 0.0142, + "step": 1470 + }, + { + "epoch": 4.3916913946587535, + "grad_norm": 0.17392724752426147, + "learning_rate": 9.739720312887535e-05, + "loss": 0.018, + "step": 1480 + }, + { + "epoch": 4.421364985163205, + "grad_norm": 0.15744361281394958, + "learning_rate": 9.734429148174675e-05, + "loss": 0.0128, + "step": 1490 + }, + { + "epoch": 4.451038575667655, + "grad_norm": 0.12954673171043396, + "learning_rate": 9.729086208503174e-05, + "loss": 0.0188, + "step": 1500 + }, + { + "epoch": 4.480712166172107, + "grad_norm": 0.16809141635894775, + "learning_rate": 9.723691552302562e-05, + "loss": 0.0144, + "step": 1510 + }, + { + "epoch": 4.510385756676558, + "grad_norm": 0.1523902416229248, + "learning_rate": 9.718245238567939e-05, + "loss": 0.0145, + "step": 1520 + }, + { + "epoch": 4.540059347181009, + "grad_norm": 0.26666077971458435, + "learning_rate": 9.712747326859315e-05, + "loss": 0.015, + "step": 1530 + }, + { + "epoch": 4.56973293768546, + "grad_norm": 0.193909153342247, + "learning_rate": 9.707197877300974e-05, + "loss": 0.0167, + "step": 1540 + }, + { + "epoch": 4.5994065281899115, + "grad_norm": 0.17812030017375946, + "learning_rate": 9.701596950580806e-05, + "loss": 0.0157, + "step": 1550 + }, + { + "epoch": 4.629080118694362, + "grad_norm": 0.26170411705970764, + "learning_rate": 9.695944607949649e-05, + "loss": 0.0139, + "step": 1560 + }, + { + "epoch": 4.658753709198813, + "grad_norm": 0.14579689502716064, + "learning_rate": 9.690240911220618e-05, + "loss": 0.013, + "step": 1570 + }, + { + "epoch": 4.688427299703264, + "grad_norm": 0.17479683458805084, + "learning_rate": 9.684485922768422e-05, + "loss": 0.014, + "step": 1580 + }, + { + "epoch": 4.718100890207715, + "grad_norm": 0.15623094141483307, + "learning_rate": 9.6786797055287e-05, + "loss": 0.0151, + "step": 1590 + }, + { + "epoch": 4.747774480712166, + "grad_norm": 0.1983657032251358, + "learning_rate": 9.672822322997305e-05, + "loss": 0.0138, + "step": 1600 + }, + { + "epoch": 4.777448071216617, + "grad_norm": 0.13670101761817932, + "learning_rate": 9.66691383922964e-05, + "loss": 0.0114, + "step": 1610 + }, + { + "epoch": 4.807121661721069, + "grad_norm": 0.13849210739135742, + "learning_rate": 9.660954318839933e-05, + "loss": 0.0105, + "step": 1620 + }, + { + "epoch": 4.836795252225519, + "grad_norm": 0.13254909217357635, + "learning_rate": 9.654943827000548e-05, + "loss": 0.0113, + "step": 1630 + }, + { + "epoch": 4.8664688427299705, + "grad_norm": 0.12954489886760712, + "learning_rate": 9.648882429441257e-05, + "loss": 0.0133, + "step": 1640 + }, + { + "epoch": 4.896142433234421, + "grad_norm": 0.13290920853614807, + "learning_rate": 9.642770192448536e-05, + "loss": 0.0156, + "step": 1650 + }, + { + "epoch": 4.925816023738872, + "grad_norm": 0.13683238625526428, + "learning_rate": 9.636607182864827e-05, + "loss": 0.0096, + "step": 1660 + }, + { + "epoch": 4.955489614243324, + "grad_norm": 0.07962514460086823, + "learning_rate": 9.630393468087818e-05, + "loss": 0.0128, + "step": 1670 + }, + { + "epoch": 4.985163204747774, + "grad_norm": 0.15497808158397675, + "learning_rate": 9.624129116069694e-05, + "loss": 0.014, + "step": 1680 + }, + { + "epoch": 5.014836795252226, + "grad_norm": 0.1419367492198944, + "learning_rate": 9.617814195316411e-05, + "loss": 0.0132, + "step": 1690 + }, + { + "epoch": 5.044510385756676, + "grad_norm": 0.2201174795627594, + "learning_rate": 9.611448774886924e-05, + "loss": 0.0133, + "step": 1700 + }, + { + "epoch": 5.074183976261128, + "grad_norm": 0.16477946937084198, + "learning_rate": 9.605032924392457e-05, + "loss": 0.0131, + "step": 1710 + }, + { + "epoch": 5.103857566765579, + "grad_norm": 0.19834354519844055, + "learning_rate": 9.598566713995718e-05, + "loss": 0.0154, + "step": 1720 + }, + { + "epoch": 5.1335311572700295, + "grad_norm": 0.22880475223064423, + "learning_rate": 9.59205021441015e-05, + "loss": 0.014, + "step": 1730 + }, + { + "epoch": 5.163204747774481, + "grad_norm": 0.16253937780857086, + "learning_rate": 9.58548349689915e-05, + "loss": 0.0122, + "step": 1740 + }, + { + "epoch": 5.192878338278931, + "grad_norm": 0.22121521830558777, + "learning_rate": 9.578866633275288e-05, + "loss": 0.015, + "step": 1750 + }, + { + "epoch": 5.222551928783383, + "grad_norm": 0.182882621884346, + "learning_rate": 9.572199695899522e-05, + "loss": 0.0163, + "step": 1760 + }, + { + "epoch": 5.252225519287834, + "grad_norm": 0.17466451227664948, + "learning_rate": 9.565482757680415e-05, + "loss": 0.0145, + "step": 1770 + }, + { + "epoch": 5.281899109792285, + "grad_norm": 0.12321746349334717, + "learning_rate": 9.558715892073323e-05, + "loss": 0.0127, + "step": 1780 + }, + { + "epoch": 5.311572700296736, + "grad_norm": 0.1869288980960846, + "learning_rate": 9.551899173079607e-05, + "loss": 0.0185, + "step": 1790 + }, + { + "epoch": 5.341246290801187, + "grad_norm": 0.1715238392353058, + "learning_rate": 9.545032675245813e-05, + "loss": 0.0136, + "step": 1800 + }, + { + "epoch": 5.370919881305638, + "grad_norm": 0.20112700760364532, + "learning_rate": 9.538116473662861e-05, + "loss": 0.013, + "step": 1810 + }, + { + "epoch": 5.400593471810089, + "grad_norm": 0.13722355663776398, + "learning_rate": 9.531150643965223e-05, + "loss": 0.0112, + "step": 1820 + }, + { + "epoch": 5.43026706231454, + "grad_norm": 0.2131308764219284, + "learning_rate": 9.524135262330098e-05, + "loss": 0.012, + "step": 1830 + }, + { + "epoch": 5.459940652818991, + "grad_norm": 0.13212528824806213, + "learning_rate": 9.517070405476575e-05, + "loss": 0.0112, + "step": 1840 + }, + { + "epoch": 5.489614243323442, + "grad_norm": 0.1769159734249115, + "learning_rate": 9.509956150664796e-05, + "loss": 0.0122, + "step": 1850 + }, + { + "epoch": 5.519287833827893, + "grad_norm": 0.15795955061912537, + "learning_rate": 9.502792575695112e-05, + "loss": 0.0152, + "step": 1860 + }, + { + "epoch": 5.548961424332344, + "grad_norm": 0.1461247205734253, + "learning_rate": 9.49557975890723e-05, + "loss": 0.0138, + "step": 1870 + }, + { + "epoch": 5.578635014836795, + "grad_norm": 0.12148546427488327, + "learning_rate": 9.488317779179361e-05, + "loss": 0.0105, + "step": 1880 + }, + { + "epoch": 5.6083086053412465, + "grad_norm": 0.17779065668582916, + "learning_rate": 9.481006715927351e-05, + "loss": 0.0107, + "step": 1890 + }, + { + "epoch": 5.637982195845697, + "grad_norm": 0.17717322707176208, + "learning_rate": 9.473646649103818e-05, + "loss": 0.0158, + "step": 1900 + }, + { + "epoch": 5.667655786350148, + "grad_norm": 0.19820888340473175, + "learning_rate": 9.46623765919727e-05, + "loss": 0.0137, + "step": 1910 + }, + { + "epoch": 5.697329376854599, + "grad_norm": 0.15721198916435242, + "learning_rate": 9.458779827231237e-05, + "loss": 0.0119, + "step": 1920 + }, + { + "epoch": 5.72700296735905, + "grad_norm": 0.17158320546150208, + "learning_rate": 9.451273234763371e-05, + "loss": 0.0133, + "step": 1930 + }, + { + "epoch": 5.756676557863502, + "grad_norm": 0.15745575726032257, + "learning_rate": 9.443717963884569e-05, + "loss": 0.0129, + "step": 1940 + }, + { + "epoch": 5.786350148367952, + "grad_norm": 0.17795485258102417, + "learning_rate": 9.43611409721806e-05, + "loss": 0.013, + "step": 1950 + }, + { + "epoch": 5.816023738872404, + "grad_norm": 0.18350425362586975, + "learning_rate": 9.428461717918511e-05, + "loss": 0.0103, + "step": 1960 + }, + { + "epoch": 5.845697329376854, + "grad_norm": 0.1675357222557068, + "learning_rate": 9.420760909671118e-05, + "loss": 0.0149, + "step": 1970 + }, + { + "epoch": 5.8753709198813056, + "grad_norm": 0.1395285427570343, + "learning_rate": 9.413011756690685e-05, + "loss": 0.0174, + "step": 1980 + }, + { + "epoch": 5.905044510385757, + "grad_norm": 0.18694210052490234, + "learning_rate": 9.405214343720707e-05, + "loss": 0.011, + "step": 1990 + }, + { + "epoch": 5.9347181008902075, + "grad_norm": 0.13509497046470642, + "learning_rate": 9.397368756032445e-05, + "loss": 0.0105, + "step": 2000 + }, + { + "epoch": 5.964391691394659, + "grad_norm": 0.1707238107919693, + "learning_rate": 9.389475079423988e-05, + "loss": 0.0099, + "step": 2010 + }, + { + "epoch": 5.994065281899109, + "grad_norm": 0.18797723948955536, + "learning_rate": 9.381533400219318e-05, + "loss": 0.0144, + "step": 2020 + }, + { + "epoch": 6.023738872403561, + "grad_norm": 0.15263915061950684, + "learning_rate": 9.373543805267368e-05, + "loss": 0.0145, + "step": 2030 + }, + { + "epoch": 6.053412462908012, + "grad_norm": 0.10786967724561691, + "learning_rate": 9.365506381941066e-05, + "loss": 0.0167, + "step": 2040 + }, + { + "epoch": 6.083086053412463, + "grad_norm": 0.10059456527233124, + "learning_rate": 9.357421218136386e-05, + "loss": 0.0127, + "step": 2050 + }, + { + "epoch": 6.112759643916914, + "grad_norm": 0.12447630614042282, + "learning_rate": 9.349288402271388e-05, + "loss": 0.0109, + "step": 2060 + }, + { + "epoch": 6.142433234421365, + "grad_norm": 0.14649389684200287, + "learning_rate": 9.341108023285238e-05, + "loss": 0.0115, + "step": 2070 + }, + { + "epoch": 6.172106824925816, + "grad_norm": 0.09072308987379074, + "learning_rate": 9.332880170637252e-05, + "loss": 0.0097, + "step": 2080 + }, + { + "epoch": 6.201780415430267, + "grad_norm": 0.1700124889612198, + "learning_rate": 9.32460493430591e-05, + "loss": 0.0107, + "step": 2090 + }, + { + "epoch": 6.231454005934718, + "grad_norm": 0.13447318971157074, + "learning_rate": 9.316282404787871e-05, + "loss": 0.0104, + "step": 2100 + }, + { + "epoch": 6.261127596439169, + "grad_norm": 0.14388670027256012, + "learning_rate": 9.30791267309698e-05, + "loss": 0.0101, + "step": 2110 + }, + { + "epoch": 6.29080118694362, + "grad_norm": 0.14514364302158356, + "learning_rate": 9.299495830763286e-05, + "loss": 0.0096, + "step": 2120 + }, + { + "epoch": 6.320474777448071, + "grad_norm": 0.09969841688871384, + "learning_rate": 9.291031969832026e-05, + "loss": 0.0115, + "step": 2130 + }, + { + "epoch": 6.350148367952523, + "grad_norm": 0.14247213304042816, + "learning_rate": 9.282521182862629e-05, + "loss": 0.0115, + "step": 2140 + }, + { + "epoch": 6.379821958456973, + "grad_norm": 0.13603922724723816, + "learning_rate": 9.273963562927695e-05, + "loss": 0.0147, + "step": 2150 + }, + { + "epoch": 6.4094955489614245, + "grad_norm": 0.23838986456394196, + "learning_rate": 9.265359203611987e-05, + "loss": 0.0115, + "step": 2160 + }, + { + "epoch": 6.439169139465875, + "grad_norm": 0.11037889868021011, + "learning_rate": 9.256708199011401e-05, + "loss": 0.0096, + "step": 2170 + }, + { + "epoch": 6.468842729970326, + "grad_norm": 0.15438543260097504, + "learning_rate": 9.248010643731935e-05, + "loss": 0.0126, + "step": 2180 + }, + { + "epoch": 6.498516320474778, + "grad_norm": 0.12337090075016022, + "learning_rate": 9.239266632888659e-05, + "loss": 0.0093, + "step": 2190 + }, + { + "epoch": 6.528189910979228, + "grad_norm": 0.09916039556264877, + "learning_rate": 9.230476262104677e-05, + "loss": 0.0114, + "step": 2200 + }, + { + "epoch": 6.55786350148368, + "grad_norm": 0.1173950806260109, + "learning_rate": 9.221639627510076e-05, + "loss": 0.0095, + "step": 2210 + }, + { + "epoch": 6.58753709198813, + "grad_norm": 0.14159461855888367, + "learning_rate": 9.212756825740873e-05, + "loss": 0.0096, + "step": 2220 + }, + { + "epoch": 6.617210682492582, + "grad_norm": 0.14629167318344116, + "learning_rate": 9.20382795393797e-05, + "loss": 0.0118, + "step": 2230 + }, + { + "epoch": 6.646884272997033, + "grad_norm": 0.11691708117723465, + "learning_rate": 9.194853109746074e-05, + "loss": 0.0095, + "step": 2240 + }, + { + "epoch": 6.6765578635014835, + "grad_norm": 0.12816114723682404, + "learning_rate": 9.185832391312644e-05, + "loss": 0.0133, + "step": 2250 + }, + { + "epoch": 6.706231454005935, + "grad_norm": 0.11063099652528763, + "learning_rate": 9.176765897286813e-05, + "loss": 0.0125, + "step": 2260 + }, + { + "epoch": 6.735905044510385, + "grad_norm": 0.12592542171478271, + "learning_rate": 9.167653726818305e-05, + "loss": 0.0097, + "step": 2270 + }, + { + "epoch": 6.765578635014837, + "grad_norm": 0.20816679298877716, + "learning_rate": 9.158495979556358e-05, + "loss": 0.0127, + "step": 2280 + }, + { + "epoch": 6.795252225519288, + "grad_norm": 0.13589587807655334, + "learning_rate": 9.14929275564863e-05, + "loss": 0.0112, + "step": 2290 + }, + { + "epoch": 6.824925816023739, + "grad_norm": 0.17974646389484406, + "learning_rate": 9.140044155740101e-05, + "loss": 0.0088, + "step": 2300 + }, + { + "epoch": 6.85459940652819, + "grad_norm": 0.18915049731731415, + "learning_rate": 9.130750280971978e-05, + "loss": 0.0156, + "step": 2310 + }, + { + "epoch": 6.884272997032641, + "grad_norm": 0.14018063247203827, + "learning_rate": 9.121411232980588e-05, + "loss": 0.0098, + "step": 2320 + }, + { + "epoch": 6.913946587537092, + "grad_norm": 0.13840338587760925, + "learning_rate": 9.112027113896262e-05, + "loss": 0.017, + "step": 2330 + }, + { + "epoch": 6.943620178041543, + "grad_norm": 0.11696403473615646, + "learning_rate": 9.102598026342222e-05, + "loss": 0.0099, + "step": 2340 + }, + { + "epoch": 6.973293768545994, + "grad_norm": 0.13574601709842682, + "learning_rate": 9.093124073433463e-05, + "loss": 0.014, + "step": 2350 + }, + { + "epoch": 7.002967359050445, + "grad_norm": 0.10100409388542175, + "learning_rate": 9.083605358775612e-05, + "loss": 0.0103, + "step": 2360 + }, + { + "epoch": 7.032640949554896, + "grad_norm": 0.16500301659107208, + "learning_rate": 9.074041986463808e-05, + "loss": 0.0109, + "step": 2370 + }, + { + "epoch": 7.062314540059347, + "grad_norm": 0.18439586460590363, + "learning_rate": 9.064434061081562e-05, + "loss": 0.0095, + "step": 2380 + }, + { + "epoch": 7.091988130563799, + "grad_norm": 0.12944291532039642, + "learning_rate": 9.0547816876996e-05, + "loss": 0.0123, + "step": 2390 + }, + { + "epoch": 7.121661721068249, + "grad_norm": 0.1536119282245636, + "learning_rate": 9.045084971874738e-05, + "loss": 0.0125, + "step": 2400 + }, + { + "epoch": 7.1513353115727005, + "grad_norm": 0.1650673747062683, + "learning_rate": 9.035344019648702e-05, + "loss": 0.0092, + "step": 2410 + }, + { + "epoch": 7.181008902077151, + "grad_norm": 0.12277387827634811, + "learning_rate": 9.025558937546988e-05, + "loss": 0.0102, + "step": 2420 + }, + { + "epoch": 7.210682492581602, + "grad_norm": 0.15055687725543976, + "learning_rate": 9.015729832577681e-05, + "loss": 0.0094, + "step": 2430 + }, + { + "epoch": 7.240356083086054, + "grad_norm": 0.16825971007347107, + "learning_rate": 9.005856812230304e-05, + "loss": 0.0116, + "step": 2440 + }, + { + "epoch": 7.270029673590504, + "grad_norm": 0.10691312700510025, + "learning_rate": 8.995939984474624e-05, + "loss": 0.0095, + "step": 2450 + }, + { + "epoch": 7.299703264094956, + "grad_norm": 0.14602239429950714, + "learning_rate": 8.98597945775948e-05, + "loss": 0.0124, + "step": 2460 + }, + { + "epoch": 7.329376854599406, + "grad_norm": 0.13875631988048553, + "learning_rate": 8.975975341011596e-05, + "loss": 0.0106, + "step": 2470 + }, + { + "epoch": 7.359050445103858, + "grad_norm": 0.12208565324544907, + "learning_rate": 8.965927743634391e-05, + "loss": 0.0108, + "step": 2480 + }, + { + "epoch": 7.388724035608309, + "grad_norm": 0.11230789124965668, + "learning_rate": 8.955836775506776e-05, + "loss": 0.0081, + "step": 2490 + }, + { + "epoch": 7.4183976261127595, + "grad_norm": 0.13064904510974884, + "learning_rate": 8.945702546981969e-05, + "loss": 0.0122, + "step": 2500 + }, + { + "epoch": 7.448071216617211, + "grad_norm": 0.16824467480182648, + "learning_rate": 8.935525168886262e-05, + "loss": 0.0112, + "step": 2510 + }, + { + "epoch": 7.4777448071216615, + "grad_norm": 0.11342830210924149, + "learning_rate": 8.92530475251784e-05, + "loss": 0.0103, + "step": 2520 + }, + { + "epoch": 7.507418397626113, + "grad_norm": 0.15296466648578644, + "learning_rate": 8.91504140964553e-05, + "loss": 0.0085, + "step": 2530 + }, + { + "epoch": 7.537091988130564, + "grad_norm": 0.16064001619815826, + "learning_rate": 8.90473525250761e-05, + "loss": 0.0114, + "step": 2540 + }, + { + "epoch": 7.566765578635015, + "grad_norm": 0.10076630860567093, + "learning_rate": 8.894386393810563e-05, + "loss": 0.0144, + "step": 2550 + }, + { + "epoch": 7.596439169139466, + "grad_norm": 0.15510450303554535, + "learning_rate": 8.883994946727849e-05, + "loss": 0.0112, + "step": 2560 + }, + { + "epoch": 7.626112759643917, + "grad_norm": 0.21251456439495087, + "learning_rate": 8.873561024898668e-05, + "loss": 0.0106, + "step": 2570 + }, + { + "epoch": 7.655786350148368, + "grad_norm": 0.17526623606681824, + "learning_rate": 8.863084742426719e-05, + "loss": 0.0113, + "step": 2580 + }, + { + "epoch": 7.6854599406528195, + "grad_norm": 0.12284035235643387, + "learning_rate": 8.852566213878947e-05, + "loss": 0.0114, + "step": 2590 + }, + { + "epoch": 7.71513353115727, + "grad_norm": 0.12916874885559082, + "learning_rate": 8.842005554284296e-05, + "loss": 0.0099, + "step": 2600 + }, + { + "epoch": 7.744807121661721, + "grad_norm": 0.17493458092212677, + "learning_rate": 8.831402879132446e-05, + "loss": 0.0092, + "step": 2610 + }, + { + "epoch": 7.774480712166172, + "grad_norm": 0.12995202839374542, + "learning_rate": 8.820758304372557e-05, + "loss": 0.0104, + "step": 2620 + }, + { + "epoch": 7.804154302670623, + "grad_norm": 0.08063960820436478, + "learning_rate": 8.810071946411989e-05, + "loss": 0.0087, + "step": 2630 + }, + { + "epoch": 7.833827893175075, + "grad_norm": 0.10341209173202515, + "learning_rate": 8.799343922115044e-05, + "loss": 0.0077, + "step": 2640 + }, + { + "epoch": 7.863501483679525, + "grad_norm": 0.108217254281044, + "learning_rate": 8.788574348801675e-05, + "loss": 0.0117, + "step": 2650 + }, + { + "epoch": 7.893175074183977, + "grad_norm": 0.1359342336654663, + "learning_rate": 8.77776334424621e-05, + "loss": 0.0157, + "step": 2660 + }, + { + "epoch": 7.922848664688427, + "grad_norm": 0.13467204570770264, + "learning_rate": 8.766911026676064e-05, + "loss": 0.011, + "step": 2670 + }, + { + "epoch": 7.9525222551928785, + "grad_norm": 0.1321392059326172, + "learning_rate": 8.756017514770443e-05, + "loss": 0.0136, + "step": 2680 + }, + { + "epoch": 7.98219584569733, + "grad_norm": 0.16591744124889374, + "learning_rate": 8.745082927659047e-05, + "loss": 0.0093, + "step": 2690 + }, + { + "epoch": 8.011869436201781, + "grad_norm": 0.14482906460762024, + "learning_rate": 8.73410738492077e-05, + "loss": 0.012, + "step": 2700 + }, + { + "epoch": 8.041543026706231, + "grad_norm": 0.12772271037101746, + "learning_rate": 8.723091006582389e-05, + "loss": 0.0118, + "step": 2710 + }, + { + "epoch": 8.071216617210682, + "grad_norm": 0.12283479422330856, + "learning_rate": 8.71203391311725e-05, + "loss": 0.0104, + "step": 2720 + }, + { + "epoch": 8.100890207715134, + "grad_norm": 0.11549960821866989, + "learning_rate": 8.700936225443959e-05, + "loss": 0.0101, + "step": 2730 + }, + { + "epoch": 8.130563798219585, + "grad_norm": 0.12340424209833145, + "learning_rate": 8.689798064925049e-05, + "loss": 0.0108, + "step": 2740 + }, + { + "epoch": 8.160237388724036, + "grad_norm": 0.13828441500663757, + "learning_rate": 8.678619553365659e-05, + "loss": 0.0084, + "step": 2750 + }, + { + "epoch": 8.189910979228486, + "grad_norm": 0.176404669880867, + "learning_rate": 8.6674008130122e-05, + "loss": 0.0093, + "step": 2760 + }, + { + "epoch": 8.219584569732937, + "grad_norm": 0.12829335033893585, + "learning_rate": 8.656141966551019e-05, + "loss": 0.0095, + "step": 2770 + }, + { + "epoch": 8.249258160237389, + "grad_norm": 0.15259785950183868, + "learning_rate": 8.644843137107059e-05, + "loss": 0.01, + "step": 2780 + }, + { + "epoch": 8.27893175074184, + "grad_norm": 0.17509141564369202, + "learning_rate": 8.633504448242505e-05, + "loss": 0.0086, + "step": 2790 + }, + { + "epoch": 8.308605341246292, + "grad_norm": 0.18248887360095978, + "learning_rate": 8.622126023955446e-05, + "loss": 0.0098, + "step": 2800 + }, + { + "epoch": 8.338278931750741, + "grad_norm": 0.13852570950984955, + "learning_rate": 8.610707988678503e-05, + "loss": 0.0109, + "step": 2810 + }, + { + "epoch": 8.367952522255193, + "grad_norm": 0.15752212703227997, + "learning_rate": 8.599250467277483e-05, + "loss": 0.0088, + "step": 2820 + }, + { + "epoch": 8.397626112759644, + "grad_norm": 0.08893997222185135, + "learning_rate": 8.587753585050004e-05, + "loss": 0.0096, + "step": 2830 + }, + { + "epoch": 8.427299703264095, + "grad_norm": 0.1134849488735199, + "learning_rate": 8.576217467724128e-05, + "loss": 0.0105, + "step": 2840 + }, + { + "epoch": 8.456973293768545, + "grad_norm": 0.18662723898887634, + "learning_rate": 8.564642241456986e-05, + "loss": 0.0102, + "step": 2850 + }, + { + "epoch": 8.486646884272997, + "grad_norm": 0.07733399420976639, + "learning_rate": 8.553028032833397e-05, + "loss": 0.0104, + "step": 2860 + }, + { + "epoch": 8.516320474777448, + "grad_norm": 0.13568611443042755, + "learning_rate": 8.541374968864487e-05, + "loss": 0.0086, + "step": 2870 + }, + { + "epoch": 8.5459940652819, + "grad_norm": 0.07520133256912231, + "learning_rate": 8.529683176986295e-05, + "loss": 0.0084, + "step": 2880 + }, + { + "epoch": 8.57566765578635, + "grad_norm": 0.06504914909601212, + "learning_rate": 8.517952785058385e-05, + "loss": 0.0086, + "step": 2890 + }, + { + "epoch": 8.605341246290802, + "grad_norm": 0.15347328782081604, + "learning_rate": 8.506183921362443e-05, + "loss": 0.0097, + "step": 2900 + }, + { + "epoch": 8.635014836795252, + "grad_norm": 0.15778006613254547, + "learning_rate": 8.494376714600878e-05, + "loss": 0.0157, + "step": 2910 + }, + { + "epoch": 8.664688427299703, + "grad_norm": 0.15846551954746246, + "learning_rate": 8.482531293895412e-05, + "loss": 0.0093, + "step": 2920 + }, + { + "epoch": 8.694362017804155, + "grad_norm": 0.19215027987957, + "learning_rate": 8.470647788785665e-05, + "loss": 0.0101, + "step": 2930 + }, + { + "epoch": 8.724035608308606, + "grad_norm": 0.14507389068603516, + "learning_rate": 8.458726329227747e-05, + "loss": 0.0102, + "step": 2940 + }, + { + "epoch": 8.753709198813056, + "grad_norm": 0.15245003998279572, + "learning_rate": 8.44676704559283e-05, + "loss": 0.009, + "step": 2950 + }, + { + "epoch": 8.783382789317507, + "grad_norm": 0.16085323691368103, + "learning_rate": 8.434770068665723e-05, + "loss": 0.0122, + "step": 2960 + }, + { + "epoch": 8.813056379821958, + "grad_norm": 0.12772567570209503, + "learning_rate": 8.422735529643444e-05, + "loss": 0.012, + "step": 2970 + }, + { + "epoch": 8.84272997032641, + "grad_norm": 0.12985776364803314, + "learning_rate": 8.410663560133784e-05, + "loss": 0.0085, + "step": 2980 + }, + { + "epoch": 8.872403560830861, + "grad_norm": 0.13308796286582947, + "learning_rate": 8.398554292153866e-05, + "loss": 0.0076, + "step": 2990 + }, + { + "epoch": 8.90207715133531, + "grad_norm": 0.18778546154499054, + "learning_rate": 8.386407858128706e-05, + "loss": 0.0116, + "step": 3000 + }, + { + "epoch": 8.931750741839762, + "grad_norm": 0.16211922466754913, + "learning_rate": 8.37422439088976e-05, + "loss": 0.0093, + "step": 3010 + }, + { + "epoch": 8.961424332344214, + "grad_norm": 0.1257450431585312, + "learning_rate": 8.362004023673474e-05, + "loss": 0.0121, + "step": 3020 + }, + { + "epoch": 8.991097922848665, + "grad_norm": 0.21673846244812012, + "learning_rate": 8.349746890119826e-05, + "loss": 0.0087, + "step": 3030 + }, + { + "epoch": 9.020771513353116, + "grad_norm": 0.10423213243484497, + "learning_rate": 8.337453124270863e-05, + "loss": 0.0074, + "step": 3040 + }, + { + "epoch": 9.050445103857566, + "grad_norm": 0.12016414105892181, + "learning_rate": 8.32512286056924e-05, + "loss": 0.0106, + "step": 3050 + }, + { + "epoch": 9.080118694362017, + "grad_norm": 0.13180746138095856, + "learning_rate": 8.31275623385675e-05, + "loss": 0.0108, + "step": 3060 + }, + { + "epoch": 9.109792284866469, + "grad_norm": 0.09843757003545761, + "learning_rate": 8.300353379372834e-05, + "loss": 0.0119, + "step": 3070 + }, + { + "epoch": 9.13946587537092, + "grad_norm": 0.08532913774251938, + "learning_rate": 8.287914432753123e-05, + "loss": 0.009, + "step": 3080 + }, + { + "epoch": 9.169139465875372, + "grad_norm": 0.08058229833841324, + "learning_rate": 8.275439530027948e-05, + "loss": 0.0096, + "step": 3090 + }, + { + "epoch": 9.198813056379821, + "grad_norm": 0.14377473294734955, + "learning_rate": 8.262928807620843e-05, + "loss": 0.0101, + "step": 3100 + }, + { + "epoch": 9.228486646884273, + "grad_norm": 0.12718182802200317, + "learning_rate": 8.250382402347065e-05, + "loss": 0.0124, + "step": 3110 + }, + { + "epoch": 9.258160237388724, + "grad_norm": 0.1333075761795044, + "learning_rate": 8.237800451412095e-05, + "loss": 0.0091, + "step": 3120 + }, + { + "epoch": 9.287833827893175, + "grad_norm": 0.12244340032339096, + "learning_rate": 8.225183092410128e-05, + "loss": 0.0108, + "step": 3130 + }, + { + "epoch": 9.317507418397627, + "grad_norm": 0.14877669513225555, + "learning_rate": 8.212530463322583e-05, + "loss": 0.0106, + "step": 3140 + }, + { + "epoch": 9.347181008902076, + "grad_norm": 0.15959547460079193, + "learning_rate": 8.199842702516583e-05, + "loss": 0.0098, + "step": 3150 + }, + { + "epoch": 9.376854599406528, + "grad_norm": 0.16950540244579315, + "learning_rate": 8.18711994874345e-05, + "loss": 0.0083, + "step": 3160 + }, + { + "epoch": 9.40652818991098, + "grad_norm": 0.12437810003757477, + "learning_rate": 8.174362341137177e-05, + "loss": 0.0101, + "step": 3170 + }, + { + "epoch": 9.43620178041543, + "grad_norm": 0.13770434260368347, + "learning_rate": 8.161570019212921e-05, + "loss": 0.0088, + "step": 3180 + }, + { + "epoch": 9.465875370919882, + "grad_norm": 0.17643623054027557, + "learning_rate": 8.148743122865463e-05, + "loss": 0.0102, + "step": 3190 + }, + { + "epoch": 9.495548961424332, + "grad_norm": 0.14998288452625275, + "learning_rate": 8.135881792367686e-05, + "loss": 0.0119, + "step": 3200 + }, + { + "epoch": 9.525222551928783, + "grad_norm": 0.14738821983337402, + "learning_rate": 8.12298616836904e-05, + "loss": 0.0098, + "step": 3210 + }, + { + "epoch": 9.554896142433234, + "grad_norm": 0.10429758578538895, + "learning_rate": 8.110056391894005e-05, + "loss": 0.0079, + "step": 3220 + }, + { + "epoch": 9.584569732937686, + "grad_norm": 0.12666217982769012, + "learning_rate": 8.097092604340542e-05, + "loss": 0.009, + "step": 3230 + }, + { + "epoch": 9.614243323442137, + "grad_norm": 0.12265059351921082, + "learning_rate": 8.084094947478556e-05, + "loss": 0.0076, + "step": 3240 + }, + { + "epoch": 9.643916913946587, + "grad_norm": 0.11325700581073761, + "learning_rate": 8.07106356344834e-05, + "loss": 0.0071, + "step": 3250 + }, + { + "epoch": 9.673590504451038, + "grad_norm": 0.13584177196025848, + "learning_rate": 8.057998594759022e-05, + "loss": 0.0135, + "step": 3260 + }, + { + "epoch": 9.70326409495549, + "grad_norm": 0.19772425293922424, + "learning_rate": 8.044900184287007e-05, + "loss": 0.0093, + "step": 3270 + }, + { + "epoch": 9.732937685459941, + "grad_norm": 0.11252380162477493, + "learning_rate": 8.031768475274413e-05, + "loss": 0.0081, + "step": 3280 + }, + { + "epoch": 9.762611275964392, + "grad_norm": 0.1141551062464714, + "learning_rate": 8.018603611327504e-05, + "loss": 0.0098, + "step": 3290 + }, + { + "epoch": 9.792284866468842, + "grad_norm": 0.17557533085346222, + "learning_rate": 8.005405736415126e-05, + "loss": 0.0084, + "step": 3300 + }, + { + "epoch": 9.821958456973293, + "grad_norm": 0.1568543016910553, + "learning_rate": 7.992174994867123e-05, + "loss": 0.0085, + "step": 3310 + }, + { + "epoch": 9.851632047477745, + "grad_norm": 0.13577055931091309, + "learning_rate": 7.978911531372765e-05, + "loss": 0.0075, + "step": 3320 + }, + { + "epoch": 9.881305637982196, + "grad_norm": 0.16812065243721008, + "learning_rate": 7.965615490979163e-05, + "loss": 0.0093, + "step": 3330 + }, + { + "epoch": 9.910979228486648, + "grad_norm": 0.1458335816860199, + "learning_rate": 7.952287019089685e-05, + "loss": 0.0082, + "step": 3340 + }, + { + "epoch": 9.940652818991097, + "grad_norm": 0.20839525759220123, + "learning_rate": 7.938926261462366e-05, + "loss": 0.0114, + "step": 3350 + }, + { + "epoch": 9.970326409495549, + "grad_norm": 0.1371382474899292, + "learning_rate": 7.925533364208309e-05, + "loss": 0.0078, + "step": 3360 + }, + { + "epoch": 10.0, + "grad_norm": 0.1349106878042221, + "learning_rate": 7.912108473790092e-05, + "loss": 0.014, + "step": 3370 + }, + { + "epoch": 10.029673590504451, + "grad_norm": 0.12665338814258575, + "learning_rate": 7.898651737020166e-05, + "loss": 0.0104, + "step": 3380 + }, + { + "epoch": 10.059347181008903, + "grad_norm": 0.13897213339805603, + "learning_rate": 7.88516330105925e-05, + "loss": 0.011, + "step": 3390 + }, + { + "epoch": 10.089020771513352, + "grad_norm": 0.16300174593925476, + "learning_rate": 7.871643313414718e-05, + "loss": 0.0126, + "step": 3400 + }, + { + "epoch": 10.118694362017804, + "grad_norm": 0.1468997597694397, + "learning_rate": 7.858091921938988e-05, + "loss": 0.0101, + "step": 3410 + }, + { + "epoch": 10.148367952522255, + "grad_norm": 0.09698769450187683, + "learning_rate": 7.844509274827907e-05, + "loss": 0.0103, + "step": 3420 + }, + { + "epoch": 10.178041543026707, + "grad_norm": 0.09831503033638, + "learning_rate": 7.830895520619128e-05, + "loss": 0.0084, + "step": 3430 + }, + { + "epoch": 10.207715133531158, + "grad_norm": 0.17069242894649506, + "learning_rate": 7.817250808190483e-05, + "loss": 0.0116, + "step": 3440 + }, + { + "epoch": 10.237388724035608, + "grad_norm": 0.1412976086139679, + "learning_rate": 7.803575286758364e-05, + "loss": 0.0074, + "step": 3450 + }, + { + "epoch": 10.267062314540059, + "grad_norm": 0.13389386236667633, + "learning_rate": 7.789869105876083e-05, + "loss": 0.0118, + "step": 3460 + }, + { + "epoch": 10.29673590504451, + "grad_norm": 0.09064219892024994, + "learning_rate": 7.776132415432234e-05, + "loss": 0.0081, + "step": 3470 + }, + { + "epoch": 10.326409495548962, + "grad_norm": 0.09377647191286087, + "learning_rate": 7.762365365649067e-05, + "loss": 0.01, + "step": 3480 + }, + { + "epoch": 10.356083086053413, + "grad_norm": 0.11738749593496323, + "learning_rate": 7.748568107080832e-05, + "loss": 0.009, + "step": 3490 + }, + { + "epoch": 10.385756676557863, + "grad_norm": 0.1620689481496811, + "learning_rate": 7.734740790612136e-05, + "loss": 0.0141, + "step": 3500 + }, + { + "epoch": 10.415430267062314, + "grad_norm": 0.11674781888723373, + "learning_rate": 7.720883567456298e-05, + "loss": 0.0097, + "step": 3510 + }, + { + "epoch": 10.445103857566766, + "grad_norm": 0.10295745730400085, + "learning_rate": 7.70699658915369e-05, + "loss": 0.0091, + "step": 3520 + }, + { + "epoch": 10.474777448071217, + "grad_norm": 0.09909237176179886, + "learning_rate": 7.693080007570084e-05, + "loss": 0.0093, + "step": 3530 + }, + { + "epoch": 10.504451038575668, + "grad_norm": 0.12657557427883148, + "learning_rate": 7.679133974894983e-05, + "loss": 0.0118, + "step": 3540 + }, + { + "epoch": 10.534124629080118, + "grad_norm": 0.08864331245422363, + "learning_rate": 7.66515864363997e-05, + "loss": 0.0091, + "step": 3550 + }, + { + "epoch": 10.56379821958457, + "grad_norm": 0.1489853858947754, + "learning_rate": 7.651154166637025e-05, + "loss": 0.0092, + "step": 3560 + }, + { + "epoch": 10.59347181008902, + "grad_norm": 0.16031557321548462, + "learning_rate": 7.637120697036866e-05, + "loss": 0.0107, + "step": 3570 + }, + { + "epoch": 10.623145400593472, + "grad_norm": 0.09444715827703476, + "learning_rate": 7.623058388307269e-05, + "loss": 0.0114, + "step": 3580 + }, + { + "epoch": 10.652818991097924, + "grad_norm": 0.09349317848682404, + "learning_rate": 7.608967394231387e-05, + "loss": 0.0084, + "step": 3590 + }, + { + "epoch": 10.682492581602373, + "grad_norm": 0.1241699606180191, + "learning_rate": 7.594847868906076e-05, + "loss": 0.011, + "step": 3600 + }, + { + "epoch": 10.712166172106825, + "grad_norm": 0.12074402719736099, + "learning_rate": 7.580699966740201e-05, + "loss": 0.0076, + "step": 3610 + }, + { + "epoch": 10.741839762611276, + "grad_norm": 0.1192033663392067, + "learning_rate": 7.566523842452958e-05, + "loss": 0.0084, + "step": 3620 + }, + { + "epoch": 10.771513353115727, + "grad_norm": 0.09560723602771759, + "learning_rate": 7.552319651072164e-05, + "loss": 0.0089, + "step": 3630 + }, + { + "epoch": 10.801186943620179, + "grad_norm": 0.09700684994459152, + "learning_rate": 7.538087547932585e-05, + "loss": 0.0099, + "step": 3640 + }, + { + "epoch": 10.830860534124628, + "grad_norm": 0.14513616263866425, + "learning_rate": 7.52382768867422e-05, + "loss": 0.0106, + "step": 3650 + }, + { + "epoch": 10.86053412462908, + "grad_norm": 0.07739041745662689, + "learning_rate": 7.509540229240601e-05, + "loss": 0.0066, + "step": 3660 + }, + { + "epoch": 10.890207715133531, + "grad_norm": 0.09094832837581635, + "learning_rate": 7.495225325877103e-05, + "loss": 0.0086, + "step": 3670 + }, + { + "epoch": 10.919881305637983, + "grad_norm": 0.1041349396109581, + "learning_rate": 7.480883135129211e-05, + "loss": 0.0099, + "step": 3680 + }, + { + "epoch": 10.949554896142434, + "grad_norm": 0.13735149800777435, + "learning_rate": 7.466513813840825e-05, + "loss": 0.0108, + "step": 3690 + }, + { + "epoch": 10.979228486646884, + "grad_norm": 0.1874203383922577, + "learning_rate": 7.452117519152542e-05, + "loss": 0.0082, + "step": 3700 + }, + { + "epoch": 11.008902077151335, + "grad_norm": 0.1152455285191536, + "learning_rate": 7.437694408499933e-05, + "loss": 0.0117, + "step": 3710 + }, + { + "epoch": 11.038575667655786, + "grad_norm": 0.1247527226805687, + "learning_rate": 7.423244639611826e-05, + "loss": 0.0098, + "step": 3720 + }, + { + "epoch": 11.068249258160238, + "grad_norm": 0.11642714589834213, + "learning_rate": 7.408768370508576e-05, + "loss": 0.0073, + "step": 3730 + }, + { + "epoch": 11.09792284866469, + "grad_norm": 0.11188193410634995, + "learning_rate": 7.394265759500348e-05, + "loss": 0.0125, + "step": 3740 + }, + { + "epoch": 11.127596439169139, + "grad_norm": 0.08563455194234848, + "learning_rate": 7.379736965185368e-05, + "loss": 0.0092, + "step": 3750 + }, + { + "epoch": 11.15727002967359, + "grad_norm": 0.09015386551618576, + "learning_rate": 7.365182146448205e-05, + "loss": 0.009, + "step": 3760 + }, + { + "epoch": 11.186943620178042, + "grad_norm": 0.13220249116420746, + "learning_rate": 7.350601462458024e-05, + "loss": 0.0116, + "step": 3770 + }, + { + "epoch": 11.216617210682493, + "grad_norm": 0.09610439091920853, + "learning_rate": 7.335995072666848e-05, + "loss": 0.0089, + "step": 3780 + }, + { + "epoch": 11.246290801186944, + "grad_norm": 0.1600012481212616, + "learning_rate": 7.32136313680782e-05, + "loss": 0.0075, + "step": 3790 + }, + { + "epoch": 11.275964391691394, + "grad_norm": 0.10654013603925705, + "learning_rate": 7.30670581489344e-05, + "loss": 0.0101, + "step": 3800 + }, + { + "epoch": 11.305637982195845, + "grad_norm": 0.10944235324859619, + "learning_rate": 7.292023267213835e-05, + "loss": 0.0069, + "step": 3810 + }, + { + "epoch": 11.335311572700297, + "grad_norm": 0.13431453704833984, + "learning_rate": 7.277315654334997e-05, + "loss": 0.0115, + "step": 3820 + }, + { + "epoch": 11.364985163204748, + "grad_norm": 0.09411869198083878, + "learning_rate": 7.262583137097018e-05, + "loss": 0.0114, + "step": 3830 + }, + { + "epoch": 11.3946587537092, + "grad_norm": 0.10903919488191605, + "learning_rate": 7.247825876612353e-05, + "loss": 0.008, + "step": 3840 + }, + { + "epoch": 11.42433234421365, + "grad_norm": 0.1587817668914795, + "learning_rate": 7.233044034264034e-05, + "loss": 0.0082, + "step": 3850 + }, + { + "epoch": 11.4540059347181, + "grad_norm": 0.18822628259658813, + "learning_rate": 7.218237771703921e-05, + "loss": 0.0115, + "step": 3860 + }, + { + "epoch": 11.483679525222552, + "grad_norm": 0.24383868277072906, + "learning_rate": 7.203407250850928e-05, + "loss": 0.0103, + "step": 3870 + }, + { + "epoch": 11.513353115727003, + "grad_norm": 0.1706736832857132, + "learning_rate": 7.188552633889259e-05, + "loss": 0.009, + "step": 3880 + }, + { + "epoch": 11.543026706231455, + "grad_norm": 0.13557754456996918, + "learning_rate": 7.173674083266624e-05, + "loss": 0.0085, + "step": 3890 + }, + { + "epoch": 11.572700296735905, + "grad_norm": 0.20359116792678833, + "learning_rate": 7.158771761692464e-05, + "loss": 0.0104, + "step": 3900 + }, + { + "epoch": 11.602373887240356, + "grad_norm": 0.11189489811658859, + "learning_rate": 7.143845832136188e-05, + "loss": 0.0091, + "step": 3910 + }, + { + "epoch": 11.632047477744807, + "grad_norm": 0.17423279583454132, + "learning_rate": 7.128896457825364e-05, + "loss": 0.0087, + "step": 3920 + }, + { + "epoch": 11.661721068249259, + "grad_norm": 0.13194143772125244, + "learning_rate": 7.113923802243957e-05, + "loss": 0.0093, + "step": 3930 + }, + { + "epoch": 11.691394658753708, + "grad_norm": 0.15409836173057556, + "learning_rate": 7.09892802913053e-05, + "loss": 0.0102, + "step": 3940 + }, + { + "epoch": 11.72106824925816, + "grad_norm": 0.10540139675140381, + "learning_rate": 7.083909302476453e-05, + "loss": 0.0107, + "step": 3950 + }, + { + "epoch": 11.750741839762611, + "grad_norm": 0.15967734158039093, + "learning_rate": 7.068867786524116e-05, + "loss": 0.0082, + "step": 3960 + }, + { + "epoch": 11.780415430267063, + "grad_norm": 0.10224897414445877, + "learning_rate": 7.053803645765128e-05, + "loss": 0.0097, + "step": 3970 + }, + { + "epoch": 11.810089020771514, + "grad_norm": 0.12072591483592987, + "learning_rate": 7.038717044938519e-05, + "loss": 0.008, + "step": 3980 + }, + { + "epoch": 11.839762611275965, + "grad_norm": 0.09004125744104385, + "learning_rate": 7.023608149028937e-05, + "loss": 0.0084, + "step": 3990 + }, + { + "epoch": 11.869436201780415, + "grad_norm": 0.09716668725013733, + "learning_rate": 7.008477123264848e-05, + "loss": 0.0097, + "step": 4000 + }, + { + "epoch": 11.899109792284866, + "grad_norm": 0.11283983290195465, + "learning_rate": 6.993324133116726e-05, + "loss": 0.0088, + "step": 4010 + }, + { + "epoch": 11.928783382789318, + "grad_norm": 0.0980016216635704, + "learning_rate": 6.978149344295242e-05, + "loss": 0.0073, + "step": 4020 + }, + { + "epoch": 11.958456973293769, + "grad_norm": 0.12663109600543976, + "learning_rate": 6.962952922749457e-05, + "loss": 0.0098, + "step": 4030 + }, + { + "epoch": 11.988130563798219, + "grad_norm": 0.08269482851028442, + "learning_rate": 6.947735034665002e-05, + "loss": 0.0097, + "step": 4040 + }, + { + "epoch": 12.01780415430267, + "grad_norm": 0.12581780552864075, + "learning_rate": 6.932495846462261e-05, + "loss": 0.0074, + "step": 4050 + }, + { + "epoch": 12.047477744807122, + "grad_norm": 0.13423371315002441, + "learning_rate": 6.917235524794558e-05, + "loss": 0.0116, + "step": 4060 + }, + { + "epoch": 12.077151335311573, + "grad_norm": 0.08609340339899063, + "learning_rate": 6.901954236546323e-05, + "loss": 0.0092, + "step": 4070 + }, + { + "epoch": 12.106824925816024, + "grad_norm": 0.11274320632219315, + "learning_rate": 6.886652148831279e-05, + "loss": 0.0074, + "step": 4080 + }, + { + "epoch": 12.136498516320474, + "grad_norm": 0.09400057792663574, + "learning_rate": 6.871329428990602e-05, + "loss": 0.0078, + "step": 4090 + }, + { + "epoch": 12.166172106824925, + "grad_norm": 0.11359213292598724, + "learning_rate": 6.855986244591104e-05, + "loss": 0.0101, + "step": 4100 + }, + { + "epoch": 12.195845697329377, + "grad_norm": 0.14520014822483063, + "learning_rate": 6.840622763423391e-05, + "loss": 0.009, + "step": 4110 + }, + { + "epoch": 12.225519287833828, + "grad_norm": 0.14667099714279175, + "learning_rate": 6.825239153500029e-05, + "loss": 0.0091, + "step": 4120 + }, + { + "epoch": 12.25519287833828, + "grad_norm": 0.11558016389608383, + "learning_rate": 6.809835583053715e-05, + "loss": 0.0088, + "step": 4130 + }, + { + "epoch": 12.28486646884273, + "grad_norm": 0.25004813075065613, + "learning_rate": 6.794412220535426e-05, + "loss": 0.0101, + "step": 4140 + }, + { + "epoch": 12.31454005934718, + "grad_norm": 0.17387105524539948, + "learning_rate": 6.778969234612584e-05, + "loss": 0.0079, + "step": 4150 + }, + { + "epoch": 12.344213649851632, + "grad_norm": 0.11986710131168365, + "learning_rate": 6.763506794167208e-05, + "loss": 0.0073, + "step": 4160 + }, + { + "epoch": 12.373887240356083, + "grad_norm": 0.14098893105983734, + "learning_rate": 6.748025068294067e-05, + "loss": 0.0115, + "step": 4170 + }, + { + "epoch": 12.403560830860535, + "grad_norm": 0.17298579216003418, + "learning_rate": 6.732524226298841e-05, + "loss": 0.0098, + "step": 4180 + }, + { + "epoch": 12.433234421364984, + "grad_norm": 0.12529845535755157, + "learning_rate": 6.71700443769625e-05, + "loss": 0.0109, + "step": 4190 + }, + { + "epoch": 12.462908011869436, + "grad_norm": 0.08396193385124207, + "learning_rate": 6.701465872208216e-05, + "loss": 0.007, + "step": 4200 + }, + { + "epoch": 12.492581602373887, + "grad_norm": 0.08979377895593643, + "learning_rate": 6.685908699762002e-05, + "loss": 0.0063, + "step": 4210 + }, + { + "epoch": 12.522255192878339, + "grad_norm": 0.09685163199901581, + "learning_rate": 6.670333090488356e-05, + "loss": 0.0068, + "step": 4220 + }, + { + "epoch": 12.55192878338279, + "grad_norm": 0.07651125639677048, + "learning_rate": 6.654739214719641e-05, + "loss": 0.0067, + "step": 4230 + }, + { + "epoch": 12.58160237388724, + "grad_norm": 0.10274438560009003, + "learning_rate": 6.639127242987988e-05, + "loss": 0.0092, + "step": 4240 + }, + { + "epoch": 12.611275964391691, + "grad_norm": 0.18983325362205505, + "learning_rate": 6.623497346023418e-05, + "loss": 0.0094, + "step": 4250 + }, + { + "epoch": 12.640949554896142, + "grad_norm": 0.09363807737827301, + "learning_rate": 6.607849694751977e-05, + "loss": 0.0109, + "step": 4260 + }, + { + "epoch": 12.670623145400594, + "grad_norm": 0.10365147143602371, + "learning_rate": 6.592184460293877e-05, + "loss": 0.0087, + "step": 4270 + }, + { + "epoch": 12.700296735905045, + "grad_norm": 0.08823329955339432, + "learning_rate": 6.576501813961609e-05, + "loss": 0.0075, + "step": 4280 + }, + { + "epoch": 12.729970326409495, + "grad_norm": 0.1399281769990921, + "learning_rate": 6.56080192725808e-05, + "loss": 0.0073, + "step": 4290 + }, + { + "epoch": 12.759643916913946, + "grad_norm": 0.10064969211816788, + "learning_rate": 6.545084971874738e-05, + "loss": 0.0077, + "step": 4300 + }, + { + "epoch": 12.789317507418398, + "grad_norm": 0.12935389578342438, + "learning_rate": 6.529351119689688e-05, + "loss": 0.0063, + "step": 4310 + }, + { + "epoch": 12.818991097922849, + "grad_norm": 0.0700489729642868, + "learning_rate": 6.513600542765817e-05, + "loss": 0.0085, + "step": 4320 + }, + { + "epoch": 12.8486646884273, + "grad_norm": 0.13894468545913696, + "learning_rate": 6.497833413348909e-05, + "loss": 0.0067, + "step": 4330 + }, + { + "epoch": 12.87833827893175, + "grad_norm": 0.13386406004428864, + "learning_rate": 6.48204990386577e-05, + "loss": 0.0078, + "step": 4340 + }, + { + "epoch": 12.908011869436201, + "grad_norm": 0.18802672624588013, + "learning_rate": 6.466250186922325e-05, + "loss": 0.0072, + "step": 4350 + }, + { + "epoch": 12.937685459940653, + "grad_norm": 0.18399575352668762, + "learning_rate": 6.450434435301751e-05, + "loss": 0.0064, + "step": 4360 + }, + { + "epoch": 12.967359050445104, + "grad_norm": 0.16697198152542114, + "learning_rate": 6.43460282196257e-05, + "loss": 0.0074, + "step": 4370 + }, + { + "epoch": 12.997032640949556, + "grad_norm": 0.16347278654575348, + "learning_rate": 6.418755520036775e-05, + "loss": 0.0071, + "step": 4380 + }, + { + "epoch": 13.026706231454005, + "grad_norm": 0.08685827255249023, + "learning_rate": 6.402892702827916e-05, + "loss": 0.0106, + "step": 4390 + }, + { + "epoch": 13.056379821958457, + "grad_norm": 0.17205415666103363, + "learning_rate": 6.387014543809223e-05, + "loss": 0.012, + "step": 4400 + }, + { + "epoch": 13.086053412462908, + "grad_norm": 0.11940989643335342, + "learning_rate": 6.371121216621698e-05, + "loss": 0.0082, + "step": 4410 + }, + { + "epoch": 13.11572700296736, + "grad_norm": 0.14548015594482422, + "learning_rate": 6.355212895072223e-05, + "loss": 0.0104, + "step": 4420 + }, + { + "epoch": 13.14540059347181, + "grad_norm": 0.10984127968549728, + "learning_rate": 6.339289753131649e-05, + "loss": 0.0091, + "step": 4430 + }, + { + "epoch": 13.17507418397626, + "grad_norm": 0.18613463640213013, + "learning_rate": 6.323351964932908e-05, + "loss": 0.0127, + "step": 4440 + }, + { + "epoch": 13.204747774480712, + "grad_norm": 0.1218784749507904, + "learning_rate": 6.307399704769099e-05, + "loss": 0.0066, + "step": 4450 + }, + { + "epoch": 13.234421364985163, + "grad_norm": 0.11912266910076141, + "learning_rate": 6.291433147091583e-05, + "loss": 0.0087, + "step": 4460 + }, + { + "epoch": 13.264094955489615, + "grad_norm": 0.1327201873064041, + "learning_rate": 6.275452466508077e-05, + "loss": 0.0076, + "step": 4470 + }, + { + "epoch": 13.293768545994066, + "grad_norm": 0.0954451709985733, + "learning_rate": 6.259457837780742e-05, + "loss": 0.0078, + "step": 4480 + }, + { + "epoch": 13.323442136498516, + "grad_norm": 0.11177768558263779, + "learning_rate": 6.243449435824276e-05, + "loss": 0.009, + "step": 4490 + }, + { + "epoch": 13.353115727002967, + "grad_norm": 0.11750178039073944, + "learning_rate": 6.227427435703997e-05, + "loss": 0.008, + "step": 4500 + }, + { + "epoch": 13.382789317507418, + "grad_norm": 0.18307024240493774, + "learning_rate": 6.211392012633932e-05, + "loss": 0.0114, + "step": 4510 + }, + { + "epoch": 13.41246290801187, + "grad_norm": 0.0931352823972702, + "learning_rate": 6.195343341974899e-05, + "loss": 0.0075, + "step": 4520 + }, + { + "epoch": 13.442136498516321, + "grad_norm": 0.10080502182245255, + "learning_rate": 6.179281599232591e-05, + "loss": 0.0081, + "step": 4530 + }, + { + "epoch": 13.47181008902077, + "grad_norm": 0.09596526622772217, + "learning_rate": 6.163206960055651e-05, + "loss": 0.0072, + "step": 4540 + }, + { + "epoch": 13.501483679525222, + "grad_norm": 0.11159328371286392, + "learning_rate": 6.147119600233758e-05, + "loss": 0.007, + "step": 4550 + }, + { + "epoch": 13.531157270029674, + "grad_norm": 0.10898558050394058, + "learning_rate": 6.131019695695702e-05, + "loss": 0.0059, + "step": 4560 + }, + { + "epoch": 13.560830860534125, + "grad_norm": 0.11957506090402603, + "learning_rate": 6.11490742250746e-05, + "loss": 0.0082, + "step": 4570 + }, + { + "epoch": 13.590504451038576, + "grad_norm": 0.15221808850765228, + "learning_rate": 6.0987829568702656e-05, + "loss": 0.0081, + "step": 4580 + }, + { + "epoch": 13.620178041543026, + "grad_norm": 0.10994566231966019, + "learning_rate": 6.0826464751186994e-05, + "loss": 0.0072, + "step": 4590 + }, + { + "epoch": 13.649851632047477, + "grad_norm": 0.12259654700756073, + "learning_rate": 6.066498153718735e-05, + "loss": 0.008, + "step": 4600 + }, + { + "epoch": 13.679525222551929, + "grad_norm": 0.09881018102169037, + "learning_rate": 6.05033816926583e-05, + "loss": 0.0083, + "step": 4610 + }, + { + "epoch": 13.70919881305638, + "grad_norm": 0.09556011855602264, + "learning_rate": 6.034166698482984e-05, + "loss": 0.0096, + "step": 4620 + }, + { + "epoch": 13.738872403560832, + "grad_norm": 0.17756840586662292, + "learning_rate": 6.017983918218812e-05, + "loss": 0.0083, + "step": 4630 + }, + { + "epoch": 13.768545994065281, + "grad_norm": 0.12028778344392776, + "learning_rate": 6.001790005445607e-05, + "loss": 0.0092, + "step": 4640 + }, + { + "epoch": 13.798219584569733, + "grad_norm": 0.16217273473739624, + "learning_rate": 5.985585137257401e-05, + "loss": 0.0088, + "step": 4650 + }, + { + "epoch": 13.827893175074184, + "grad_norm": 0.145747110247612, + "learning_rate": 5.969369490868042e-05, + "loss": 0.0138, + "step": 4660 + }, + { + "epoch": 13.857566765578635, + "grad_norm": 0.16528791189193726, + "learning_rate": 5.953143243609235e-05, + "loss": 0.0082, + "step": 4670 + }, + { + "epoch": 13.887240356083087, + "grad_norm": 0.08141251653432846, + "learning_rate": 5.9369065729286245e-05, + "loss": 0.0056, + "step": 4680 + }, + { + "epoch": 13.916913946587536, + "grad_norm": 0.10877679288387299, + "learning_rate": 5.9206596563878357e-05, + "loss": 0.0098, + "step": 4690 + }, + { + "epoch": 13.946587537091988, + "grad_norm": 0.10972292721271515, + "learning_rate": 5.90440267166055e-05, + "loss": 0.0113, + "step": 4700 + }, + { + "epoch": 13.97626112759644, + "grad_norm": 0.11017335951328278, + "learning_rate": 5.888135796530544e-05, + "loss": 0.0072, + "step": 4710 + }, + { + "epoch": 14.00593471810089, + "grad_norm": 0.1043027862906456, + "learning_rate": 5.871859208889759e-05, + "loss": 0.0082, + "step": 4720 + }, + { + "epoch": 14.035608308605342, + "grad_norm": 0.09205296635627747, + "learning_rate": 5.85557308673635e-05, + "loss": 0.0084, + "step": 4730 + }, + { + "epoch": 14.065281899109792, + "grad_norm": 0.08532586693763733, + "learning_rate": 5.8392776081727385e-05, + "loss": 0.0064, + "step": 4740 + }, + { + "epoch": 14.094955489614243, + "grad_norm": 0.10191354155540466, + "learning_rate": 5.8229729514036705e-05, + "loss": 0.0077, + "step": 4750 + }, + { + "epoch": 14.124629080118694, + "grad_norm": 0.10814640671014786, + "learning_rate": 5.8066592947342555e-05, + "loss": 0.0083, + "step": 4760 + }, + { + "epoch": 14.154302670623146, + "grad_norm": 0.1646806001663208, + "learning_rate": 5.7903368165680327e-05, + "loss": 0.0083, + "step": 4770 + }, + { + "epoch": 14.183976261127597, + "grad_norm": 0.11636976152658463, + "learning_rate": 5.7740056954050084e-05, + "loss": 0.0092, + "step": 4780 + }, + { + "epoch": 14.213649851632047, + "grad_norm": 0.15811985731124878, + "learning_rate": 5.757666109839702e-05, + "loss": 0.0073, + "step": 4790 + }, + { + "epoch": 14.243323442136498, + "grad_norm": 0.18782468140125275, + "learning_rate": 5.74131823855921e-05, + "loss": 0.0101, + "step": 4800 + }, + { + "epoch": 14.27299703264095, + "grad_norm": 0.11287040263414383, + "learning_rate": 5.72496226034123e-05, + "loss": 0.0071, + "step": 4810 + }, + { + "epoch": 14.302670623145401, + "grad_norm": 0.06242251396179199, + "learning_rate": 5.7085983540521216e-05, + "loss": 0.0055, + "step": 4820 + }, + { + "epoch": 14.332344213649852, + "grad_norm": 0.11765594780445099, + "learning_rate": 5.692226698644938e-05, + "loss": 0.0092, + "step": 4830 + }, + { + "epoch": 14.362017804154302, + "grad_norm": 0.11589968204498291, + "learning_rate": 5.675847473157485e-05, + "loss": 0.0147, + "step": 4840 + }, + { + "epoch": 14.391691394658753, + "grad_norm": 0.08261799812316895, + "learning_rate": 5.6594608567103456e-05, + "loss": 0.0092, + "step": 4850 + }, + { + "epoch": 14.421364985163205, + "grad_norm": 0.09985928237438202, + "learning_rate": 5.6430670285049314e-05, + "loss": 0.0085, + "step": 4860 + }, + { + "epoch": 14.451038575667656, + "grad_norm": 0.07024219632148743, + "learning_rate": 5.6266661678215216e-05, + "loss": 0.0073, + "step": 4870 + }, + { + "epoch": 14.480712166172108, + "grad_norm": 0.1330738216638565, + "learning_rate": 5.6102584540173006e-05, + "loss": 0.0107, + "step": 4880 + }, + { + "epoch": 14.510385756676557, + "grad_norm": 0.1905350685119629, + "learning_rate": 5.5938440665244006e-05, + "loss": 0.0077, + "step": 4890 + }, + { + "epoch": 14.540059347181009, + "grad_norm": 0.10635291785001755, + "learning_rate": 5.577423184847932e-05, + "loss": 0.008, + "step": 4900 + }, + { + "epoch": 14.56973293768546, + "grad_norm": 0.09331542998552322, + "learning_rate": 5.560995988564023e-05, + "loss": 0.0072, + "step": 4910 + }, + { + "epoch": 14.599406528189911, + "grad_norm": 0.07317591458559036, + "learning_rate": 5.544562657317863e-05, + "loss": 0.0062, + "step": 4920 + }, + { + "epoch": 14.629080118694361, + "grad_norm": 0.1096150130033493, + "learning_rate": 5.52812337082173e-05, + "loss": 0.0084, + "step": 4930 + }, + { + "epoch": 14.658753709198812, + "grad_norm": 0.08940622955560684, + "learning_rate": 5.511678308853026e-05, + "loss": 0.0068, + "step": 4940 + }, + { + "epoch": 14.688427299703264, + "grad_norm": 0.09628520905971527, + "learning_rate": 5.495227651252315e-05, + "loss": 0.0078, + "step": 4950 + }, + { + "epoch": 14.718100890207715, + "grad_norm": 0.08508996665477753, + "learning_rate": 5.478771577921351e-05, + "loss": 0.0055, + "step": 4960 + }, + { + "epoch": 14.747774480712167, + "grad_norm": 0.15108130872249603, + "learning_rate": 5.462310268821118e-05, + "loss": 0.0111, + "step": 4970 + }, + { + "epoch": 14.777448071216618, + "grad_norm": 0.12836958467960358, + "learning_rate": 5.445843903969854e-05, + "loss": 0.0111, + "step": 4980 + }, + { + "epoch": 14.807121661721068, + "grad_norm": 0.1240241602063179, + "learning_rate": 5.4293726634410855e-05, + "loss": 0.007, + "step": 4990 + }, + { + "epoch": 14.836795252225519, + "grad_norm": 0.1312558501958847, + "learning_rate": 5.4128967273616625e-05, + "loss": 0.0084, + "step": 5000 + }, + { + "epoch": 14.86646884272997, + "grad_norm": 0.07452099025249481, + "learning_rate": 5.396416275909779e-05, + "loss": 0.0097, + "step": 5010 + }, + { + "epoch": 14.896142433234422, + "grad_norm": 0.10130845010280609, + "learning_rate": 5.379931489313016e-05, + "loss": 0.0076, + "step": 5020 + }, + { + "epoch": 14.925816023738872, + "grad_norm": 0.08438223600387573, + "learning_rate": 5.363442547846356e-05, + "loss": 0.0079, + "step": 5030 + }, + { + "epoch": 14.955489614243323, + "grad_norm": 0.10516081005334854, + "learning_rate": 5.3469496318302204e-05, + "loss": 0.007, + "step": 5040 + }, + { + "epoch": 14.985163204747774, + "grad_norm": 0.10331432521343231, + "learning_rate": 5.330452921628497e-05, + "loss": 0.008, + "step": 5050 + }, + { + "epoch": 15.014836795252226, + "grad_norm": 0.12713058292865753, + "learning_rate": 5.313952597646568e-05, + "loss": 0.0079, + "step": 5060 + }, + { + "epoch": 15.044510385756677, + "grad_norm": 0.10389821976423264, + "learning_rate": 5.297448840329329e-05, + "loss": 0.0049, + "step": 5070 + }, + { + "epoch": 15.074183976261127, + "grad_norm": 0.08406785875558853, + "learning_rate": 5.280941830159227e-05, + "loss": 0.0077, + "step": 5080 + }, + { + "epoch": 15.103857566765578, + "grad_norm": 0.09462954849004745, + "learning_rate": 5.264431747654284e-05, + "loss": 0.0078, + "step": 5090 + }, + { + "epoch": 15.13353115727003, + "grad_norm": 0.08634629100561142, + "learning_rate": 5.247918773366112e-05, + "loss": 0.0091, + "step": 5100 + }, + { + "epoch": 15.163204747774481, + "grad_norm": 0.08352985233068466, + "learning_rate": 5.231403087877955e-05, + "loss": 0.0072, + "step": 5110 + }, + { + "epoch": 15.192878338278932, + "grad_norm": 0.10993140935897827, + "learning_rate": 5.214884871802703e-05, + "loss": 0.0116, + "step": 5120 + }, + { + "epoch": 15.222551928783382, + "grad_norm": 0.1299051195383072, + "learning_rate": 5.198364305780922e-05, + "loss": 0.0058, + "step": 5130 + }, + { + "epoch": 15.252225519287833, + "grad_norm": 0.13166804611682892, + "learning_rate": 5.1818415704788725e-05, + "loss": 0.0072, + "step": 5140 + }, + { + "epoch": 15.281899109792285, + "grad_norm": 0.09685414284467697, + "learning_rate": 5.165316846586541e-05, + "loss": 0.0064, + "step": 5150 + }, + { + "epoch": 15.311572700296736, + "grad_norm": 0.11090342700481415, + "learning_rate": 5.148790314815663e-05, + "loss": 0.0087, + "step": 5160 + }, + { + "epoch": 15.341246290801188, + "grad_norm": 0.09668321162462234, + "learning_rate": 5.132262155897739e-05, + "loss": 0.005, + "step": 5170 + }, + { + "epoch": 15.370919881305637, + "grad_norm": 0.10721191018819809, + "learning_rate": 5.1157325505820694e-05, + "loss": 0.0071, + "step": 5180 + }, + { + "epoch": 15.400593471810089, + "grad_norm": 0.142580047249794, + "learning_rate": 5.0992016796337686e-05, + "loss": 0.0093, + "step": 5190 + }, + { + "epoch": 15.43026706231454, + "grad_norm": 0.13017131388187408, + "learning_rate": 5.0826697238317935e-05, + "loss": 0.0099, + "step": 5200 + }, + { + "epoch": 15.459940652818991, + "grad_norm": 0.16075848042964935, + "learning_rate": 5.066136863966963e-05, + "loss": 0.0076, + "step": 5210 + }, + { + "epoch": 15.489614243323443, + "grad_norm": 0.08089852333068848, + "learning_rate": 5.0496032808399815e-05, + "loss": 0.0082, + "step": 5220 + }, + { + "epoch": 15.519287833827892, + "grad_norm": 0.1011849120259285, + "learning_rate": 5.033069155259471e-05, + "loss": 0.0079, + "step": 5230 + }, + { + "epoch": 15.548961424332344, + "grad_norm": 0.11635112762451172, + "learning_rate": 5.016534668039976e-05, + "loss": 0.0065, + "step": 5240 + }, + { + "epoch": 15.578635014836795, + "grad_norm": 0.13171295821666718, + "learning_rate": 5e-05, + "loss": 0.0055, + "step": 5250 + }, + { + "epoch": 15.608308605341247, + "grad_norm": 0.08494381606578827, + "learning_rate": 4.9834653319600246e-05, + "loss": 0.0085, + "step": 5260 + }, + { + "epoch": 15.637982195845698, + "grad_norm": 0.10446976125240326, + "learning_rate": 4.96693084474053e-05, + "loss": 0.0078, + "step": 5270 + }, + { + "epoch": 15.667655786350148, + "grad_norm": 0.0890231505036354, + "learning_rate": 4.950396719160018e-05, + "loss": 0.0074, + "step": 5280 + }, + { + "epoch": 15.697329376854599, + "grad_norm": 0.12522277235984802, + "learning_rate": 4.93386313603304e-05, + "loss": 0.0078, + "step": 5290 + }, + { + "epoch": 15.72700296735905, + "grad_norm": 0.1571093499660492, + "learning_rate": 4.917330276168208e-05, + "loss": 0.0074, + "step": 5300 + }, + { + "epoch": 15.756676557863502, + "grad_norm": 0.062216538935899734, + "learning_rate": 4.9007983203662326e-05, + "loss": 0.0063, + "step": 5310 + }, + { + "epoch": 15.786350148367953, + "grad_norm": 0.1346684992313385, + "learning_rate": 4.884267449417931e-05, + "loss": 0.0076, + "step": 5320 + }, + { + "epoch": 15.816023738872403, + "grad_norm": 0.09010523557662964, + "learning_rate": 4.867737844102261e-05, + "loss": 0.0058, + "step": 5330 + }, + { + "epoch": 15.845697329376854, + "grad_norm": 0.14437784254550934, + "learning_rate": 4.851209685184338e-05, + "loss": 0.0086, + "step": 5340 + }, + { + "epoch": 15.875370919881306, + "grad_norm": 0.11848489195108414, + "learning_rate": 4.834683153413459e-05, + "loss": 0.0066, + "step": 5350 + }, + { + "epoch": 15.905044510385757, + "grad_norm": 0.070223368704319, + "learning_rate": 4.818158429521129e-05, + "loss": 0.0076, + "step": 5360 + }, + { + "epoch": 15.934718100890208, + "grad_norm": 0.14026039838790894, + "learning_rate": 4.801635694219079e-05, + "loss": 0.0088, + "step": 5370 + }, + { + "epoch": 15.964391691394658, + "grad_norm": 0.10169534385204315, + "learning_rate": 4.785115128197298e-05, + "loss": 0.0071, + "step": 5380 + }, + { + "epoch": 15.99406528189911, + "grad_norm": 0.07850392907857895, + "learning_rate": 4.7685969121220456e-05, + "loss": 0.006, + "step": 5390 + }, + { + "epoch": 16.023738872403563, + "grad_norm": 0.10650818794965744, + "learning_rate": 4.7520812266338885e-05, + "loss": 0.007, + "step": 5400 + }, + { + "epoch": 16.05341246290801, + "grad_norm": 0.12273446470499039, + "learning_rate": 4.735568252345718e-05, + "loss": 0.0071, + "step": 5410 + }, + { + "epoch": 16.083086053412462, + "grad_norm": 0.08303070813417435, + "learning_rate": 4.7190581698407725e-05, + "loss": 0.0053, + "step": 5420 + }, + { + "epoch": 16.112759643916913, + "grad_norm": 0.12177737057209015, + "learning_rate": 4.702551159670672e-05, + "loss": 0.0072, + "step": 5430 + }, + { + "epoch": 16.142433234421365, + "grad_norm": 0.08380179852247238, + "learning_rate": 4.6860474023534335e-05, + "loss": 0.0071, + "step": 5440 + }, + { + "epoch": 16.172106824925816, + "grad_norm": 0.18725629150867462, + "learning_rate": 4.669547078371504e-05, + "loss": 0.0109, + "step": 5450 + }, + { + "epoch": 16.201780415430267, + "grad_norm": 0.07537223398685455, + "learning_rate": 4.65305036816978e-05, + "loss": 0.0063, + "step": 5460 + }, + { + "epoch": 16.23145400593472, + "grad_norm": 0.11886244267225266, + "learning_rate": 4.6365574521536445e-05, + "loss": 0.0066, + "step": 5470 + }, + { + "epoch": 16.26112759643917, + "grad_norm": 0.11239882558584213, + "learning_rate": 4.620068510686985e-05, + "loss": 0.0067, + "step": 5480 + }, + { + "epoch": 16.29080118694362, + "grad_norm": 0.08789864927530289, + "learning_rate": 4.60358372409022e-05, + "loss": 0.005, + "step": 5490 + }, + { + "epoch": 16.320474777448073, + "grad_norm": 0.1038408875465393, + "learning_rate": 4.5871032726383386e-05, + "loss": 0.0059, + "step": 5500 + }, + { + "epoch": 16.35014836795252, + "grad_norm": 0.12280593812465668, + "learning_rate": 4.570627336558915e-05, + "loss": 0.0076, + "step": 5510 + }, + { + "epoch": 16.379821958456972, + "grad_norm": 0.07442447543144226, + "learning_rate": 4.554156096030149e-05, + "loss": 0.0058, + "step": 5520 + }, + { + "epoch": 16.409495548961424, + "grad_norm": 0.14733825623989105, + "learning_rate": 4.537689731178883e-05, + "loss": 0.0111, + "step": 5530 + }, + { + "epoch": 16.439169139465875, + "grad_norm": 0.08889064192771912, + "learning_rate": 4.5212284220786494e-05, + "loss": 0.0066, + "step": 5540 + }, + { + "epoch": 16.468842729970326, + "grad_norm": 0.10983613133430481, + "learning_rate": 4.504772348747687e-05, + "loss": 0.005, + "step": 5550 + }, + { + "epoch": 16.498516320474778, + "grad_norm": 0.08122289180755615, + "learning_rate": 4.488321691146975e-05, + "loss": 0.0064, + "step": 5560 + }, + { + "epoch": 16.52818991097923, + "grad_norm": 0.11093121021986008, + "learning_rate": 4.471876629178273e-05, + "loss": 0.0056, + "step": 5570 + }, + { + "epoch": 16.55786350148368, + "grad_norm": 0.07384008914232254, + "learning_rate": 4.4554373426821374e-05, + "loss": 0.0066, + "step": 5580 + }, + { + "epoch": 16.587537091988132, + "grad_norm": 0.0759691372513771, + "learning_rate": 4.439004011435979e-05, + "loss": 0.0067, + "step": 5590 + }, + { + "epoch": 16.617210682492583, + "grad_norm": 0.12558524310588837, + "learning_rate": 4.4225768151520694e-05, + "loss": 0.0082, + "step": 5600 + }, + { + "epoch": 16.64688427299703, + "grad_norm": 0.09405193477869034, + "learning_rate": 4.406155933475599e-05, + "loss": 0.0082, + "step": 5610 + }, + { + "epoch": 16.676557863501483, + "grad_norm": 0.11722928285598755, + "learning_rate": 4.3897415459827e-05, + "loss": 0.0071, + "step": 5620 + }, + { + "epoch": 16.706231454005934, + "grad_norm": 0.06153471767902374, + "learning_rate": 4.373333832178478e-05, + "loss": 0.0057, + "step": 5630 + }, + { + "epoch": 16.735905044510385, + "grad_norm": 0.1048816367983818, + "learning_rate": 4.3569329714950704e-05, + "loss": 0.0105, + "step": 5640 + }, + { + "epoch": 16.765578635014837, + "grad_norm": 0.08989928662776947, + "learning_rate": 4.3405391432896555e-05, + "loss": 0.0045, + "step": 5650 + }, + { + "epoch": 16.795252225519288, + "grad_norm": 0.10757238417863846, + "learning_rate": 4.324152526842517e-05, + "loss": 0.0067, + "step": 5660 + }, + { + "epoch": 16.82492581602374, + "grad_norm": 0.1232636570930481, + "learning_rate": 4.307773301355062e-05, + "loss": 0.0064, + "step": 5670 + }, + { + "epoch": 16.85459940652819, + "grad_norm": 0.11661067605018616, + "learning_rate": 4.291401645947879e-05, + "loss": 0.0066, + "step": 5680 + }, + { + "epoch": 16.884272997032642, + "grad_norm": 0.128203347325325, + "learning_rate": 4.275037739658771e-05, + "loss": 0.0089, + "step": 5690 + }, + { + "epoch": 16.91394658753709, + "grad_norm": 0.05867573991417885, + "learning_rate": 4.2586817614407895e-05, + "loss": 0.0048, + "step": 5700 + }, + { + "epoch": 16.94362017804154, + "grad_norm": 0.10739487409591675, + "learning_rate": 4.2423338901602985e-05, + "loss": 0.0082, + "step": 5710 + }, + { + "epoch": 16.973293768545993, + "grad_norm": 0.15257547795772552, + "learning_rate": 4.2259943045949934e-05, + "loss": 0.0099, + "step": 5720 + }, + { + "epoch": 17.002967359050444, + "grad_norm": 0.11332064867019653, + "learning_rate": 4.209663183431969e-05, + "loss": 0.0091, + "step": 5730 + }, + { + "epoch": 17.032640949554896, + "grad_norm": 0.10195380449295044, + "learning_rate": 4.1933407052657456e-05, + "loss": 0.0064, + "step": 5740 + }, + { + "epoch": 17.062314540059347, + "grad_norm": 0.09771795570850372, + "learning_rate": 4.17702704859633e-05, + "loss": 0.006, + "step": 5750 + }, + { + "epoch": 17.0919881305638, + "grad_norm": 0.067668117582798, + "learning_rate": 4.160722391827262e-05, + "loss": 0.0057, + "step": 5760 + }, + { + "epoch": 17.12166172106825, + "grad_norm": 0.10733453184366226, + "learning_rate": 4.14442691326365e-05, + "loss": 0.0069, + "step": 5770 + }, + { + "epoch": 17.1513353115727, + "grad_norm": 0.08877403289079666, + "learning_rate": 4.1281407911102425e-05, + "loss": 0.0076, + "step": 5780 + }, + { + "epoch": 17.181008902077153, + "grad_norm": 0.09970077127218246, + "learning_rate": 4.111864203469457e-05, + "loss": 0.0081, + "step": 5790 + }, + { + "epoch": 17.2106824925816, + "grad_norm": 0.08427341282367706, + "learning_rate": 4.095597328339452e-05, + "loss": 0.0063, + "step": 5800 + }, + { + "epoch": 17.240356083086052, + "grad_norm": 0.08047015219926834, + "learning_rate": 4.079340343612165e-05, + "loss": 0.0057, + "step": 5810 + }, + { + "epoch": 17.270029673590503, + "grad_norm": 0.09922623634338379, + "learning_rate": 4.063093427071376e-05, + "loss": 0.0057, + "step": 5820 + }, + { + "epoch": 17.299703264094955, + "grad_norm": 0.16388913989067078, + "learning_rate": 4.046856756390767e-05, + "loss": 0.0089, + "step": 5830 + }, + { + "epoch": 17.329376854599406, + "grad_norm": 0.14212121069431305, + "learning_rate": 4.0306305091319595e-05, + "loss": 0.0068, + "step": 5840 + }, + { + "epoch": 17.359050445103858, + "grad_norm": 0.09012001752853394, + "learning_rate": 4.0144148627425993e-05, + "loss": 0.0119, + "step": 5850 + }, + { + "epoch": 17.38872403560831, + "grad_norm": 0.14375264942646027, + "learning_rate": 3.9982099945543945e-05, + "loss": 0.007, + "step": 5860 + }, + { + "epoch": 17.41839762611276, + "grad_norm": 0.0589725561439991, + "learning_rate": 3.982016081781189e-05, + "loss": 0.0058, + "step": 5870 + }, + { + "epoch": 17.448071216617212, + "grad_norm": 0.13382118940353394, + "learning_rate": 3.965833301517017e-05, + "loss": 0.0109, + "step": 5880 + }, + { + "epoch": 17.477744807121663, + "grad_norm": 0.11577240377664566, + "learning_rate": 3.949661830734172e-05, + "loss": 0.0059, + "step": 5890 + }, + { + "epoch": 17.50741839762611, + "grad_norm": 0.11332771927118301, + "learning_rate": 3.933501846281267e-05, + "loss": 0.0081, + "step": 5900 + }, + { + "epoch": 17.537091988130562, + "grad_norm": 0.06491021066904068, + "learning_rate": 3.917353524881302e-05, + "loss": 0.0069, + "step": 5910 + }, + { + "epoch": 17.566765578635014, + "grad_norm": 0.06087639555335045, + "learning_rate": 3.901217043129735e-05, + "loss": 0.0075, + "step": 5920 + }, + { + "epoch": 17.596439169139465, + "grad_norm": 0.09394536167383194, + "learning_rate": 3.8850925774925425e-05, + "loss": 0.0052, + "step": 5930 + }, + { + "epoch": 17.626112759643917, + "grad_norm": 0.07485074549913406, + "learning_rate": 3.8689803043043e-05, + "loss": 0.0066, + "step": 5940 + }, + { + "epoch": 17.655786350148368, + "grad_norm": 0.09653795510530472, + "learning_rate": 3.852880399766243e-05, + "loss": 0.0056, + "step": 5950 + }, + { + "epoch": 17.68545994065282, + "grad_norm": 0.08485693484544754, + "learning_rate": 3.836793039944349e-05, + "loss": 0.0052, + "step": 5960 + }, + { + "epoch": 17.71513353115727, + "grad_norm": 0.060783516615629196, + "learning_rate": 3.820718400767409e-05, + "loss": 0.005, + "step": 5970 + }, + { + "epoch": 17.744807121661722, + "grad_norm": 0.13937251269817352, + "learning_rate": 3.8046566580251e-05, + "loss": 0.0073, + "step": 5980 + }, + { + "epoch": 17.774480712166174, + "grad_norm": 0.10680582374334335, + "learning_rate": 3.788607987366069e-05, + "loss": 0.0062, + "step": 5990 + }, + { + "epoch": 17.80415430267062, + "grad_norm": 0.14509576559066772, + "learning_rate": 3.772572564296005e-05, + "loss": 0.0063, + "step": 6000 + }, + { + "epoch": 17.833827893175073, + "grad_norm": 0.09072288125753403, + "learning_rate": 3.756550564175727e-05, + "loss": 0.0051, + "step": 6010 + }, + { + "epoch": 17.863501483679524, + "grad_norm": 0.06139769405126572, + "learning_rate": 3.74054216221926e-05, + "loss": 0.0061, + "step": 6020 + }, + { + "epoch": 17.893175074183976, + "grad_norm": 0.07538589090108871, + "learning_rate": 3.7245475334919246e-05, + "loss": 0.0047, + "step": 6030 + }, + { + "epoch": 17.922848664688427, + "grad_norm": 0.03345522657036781, + "learning_rate": 3.7085668529084184e-05, + "loss": 0.008, + "step": 6040 + }, + { + "epoch": 17.95252225519288, + "grad_norm": 0.0697244256734848, + "learning_rate": 3.6926002952309016e-05, + "loss": 0.0038, + "step": 6050 + }, + { + "epoch": 17.98219584569733, + "grad_norm": 0.06793000549077988, + "learning_rate": 3.676648035067093e-05, + "loss": 0.0059, + "step": 6060 + }, + { + "epoch": 18.01186943620178, + "grad_norm": 0.0548509806394577, + "learning_rate": 3.6607102468683526e-05, + "loss": 0.0053, + "step": 6070 + }, + { + "epoch": 18.041543026706233, + "grad_norm": 0.0740605965256691, + "learning_rate": 3.6447871049277796e-05, + "loss": 0.0074, + "step": 6080 + }, + { + "epoch": 18.071216617210684, + "grad_norm": 0.04433996602892876, + "learning_rate": 3.628878783378302e-05, + "loss": 0.0057, + "step": 6090 + }, + { + "epoch": 18.100890207715132, + "grad_norm": 0.10380228608846664, + "learning_rate": 3.612985456190778e-05, + "loss": 0.0048, + "step": 6100 + }, + { + "epoch": 18.130563798219583, + "grad_norm": 0.04870785027742386, + "learning_rate": 3.597107297172084e-05, + "loss": 0.0045, + "step": 6110 + }, + { + "epoch": 18.160237388724035, + "grad_norm": 0.08207621425390244, + "learning_rate": 3.581244479963225e-05, + "loss": 0.0069, + "step": 6120 + }, + { + "epoch": 18.189910979228486, + "grad_norm": 0.07084132730960846, + "learning_rate": 3.5653971780374295e-05, + "loss": 0.0063, + "step": 6130 + }, + { + "epoch": 18.219584569732937, + "grad_norm": 0.06766065955162048, + "learning_rate": 3.5495655646982505e-05, + "loss": 0.0047, + "step": 6140 + }, + { + "epoch": 18.24925816023739, + "grad_norm": 0.07428411394357681, + "learning_rate": 3.533749813077677e-05, + "loss": 0.0048, + "step": 6150 + }, + { + "epoch": 18.27893175074184, + "grad_norm": 0.06863471865653992, + "learning_rate": 3.517950096134232e-05, + "loss": 0.0078, + "step": 6160 + }, + { + "epoch": 18.30860534124629, + "grad_norm": 0.07877454161643982, + "learning_rate": 3.5021665866510925e-05, + "loss": 0.0066, + "step": 6170 + }, + { + "epoch": 18.338278931750743, + "grad_norm": 0.10128459334373474, + "learning_rate": 3.4863994572341843e-05, + "loss": 0.007, + "step": 6180 + }, + { + "epoch": 18.367952522255194, + "grad_norm": 0.1272270679473877, + "learning_rate": 3.470648880310313e-05, + "loss": 0.0076, + "step": 6190 + }, + { + "epoch": 18.397626112759642, + "grad_norm": 0.08308117091655731, + "learning_rate": 3.4549150281252636e-05, + "loss": 0.0059, + "step": 6200 + }, + { + "epoch": 18.427299703264094, + "grad_norm": 0.12493477761745453, + "learning_rate": 3.439198072741921e-05, + "loss": 0.0062, + "step": 6210 + }, + { + "epoch": 18.456973293768545, + "grad_norm": 0.08306564390659332, + "learning_rate": 3.423498186038393e-05, + "loss": 0.0062, + "step": 6220 + }, + { + "epoch": 18.486646884272997, + "grad_norm": 0.08139030635356903, + "learning_rate": 3.407815539706124e-05, + "loss": 0.006, + "step": 6230 + }, + { + "epoch": 18.516320474777448, + "grad_norm": 0.1288260817527771, + "learning_rate": 3.392150305248024e-05, + "loss": 0.0057, + "step": 6240 + }, + { + "epoch": 18.5459940652819, + "grad_norm": 0.13427333533763885, + "learning_rate": 3.3765026539765834e-05, + "loss": 0.0083, + "step": 6250 + }, + { + "epoch": 18.57566765578635, + "grad_norm": 0.07845230400562286, + "learning_rate": 3.360872757012011e-05, + "loss": 0.0084, + "step": 6260 + }, + { + "epoch": 18.605341246290802, + "grad_norm": 0.07737909257411957, + "learning_rate": 3.3452607852803584e-05, + "loss": 0.0056, + "step": 6270 + }, + { + "epoch": 18.635014836795254, + "grad_norm": 0.12954828143119812, + "learning_rate": 3.329666909511645e-05, + "loss": 0.0067, + "step": 6280 + }, + { + "epoch": 18.664688427299705, + "grad_norm": 0.07920852303504944, + "learning_rate": 3.3140913002379995e-05, + "loss": 0.0082, + "step": 6290 + }, + { + "epoch": 18.694362017804153, + "grad_norm": 0.08958853781223297, + "learning_rate": 3.298534127791785e-05, + "loss": 0.0051, + "step": 6300 + }, + { + "epoch": 18.724035608308604, + "grad_norm": 0.07667780667543411, + "learning_rate": 3.282995562303754e-05, + "loss": 0.0062, + "step": 6310 + }, + { + "epoch": 18.753709198813056, + "grad_norm": 0.06684096157550812, + "learning_rate": 3.267475773701161e-05, + "loss": 0.0047, + "step": 6320 + }, + { + "epoch": 18.783382789317507, + "grad_norm": 0.06355752050876617, + "learning_rate": 3.251974931705933e-05, + "loss": 0.005, + "step": 6330 + }, + { + "epoch": 18.81305637982196, + "grad_norm": 0.07288695126771927, + "learning_rate": 3.236493205832795e-05, + "loss": 0.0057, + "step": 6340 + }, + { + "epoch": 18.84272997032641, + "grad_norm": 0.06220920756459236, + "learning_rate": 3.221030765387417e-05, + "loss": 0.0045, + "step": 6350 + }, + { + "epoch": 18.87240356083086, + "grad_norm": 0.045754965394735336, + "learning_rate": 3.205587779464576e-05, + "loss": 0.0058, + "step": 6360 + }, + { + "epoch": 18.902077151335313, + "grad_norm": 0.08345388621091843, + "learning_rate": 3.190164416946285e-05, + "loss": 0.0072, + "step": 6370 + }, + { + "epoch": 18.931750741839764, + "grad_norm": 0.08965882658958435, + "learning_rate": 3.1747608464999725e-05, + "loss": 0.0054, + "step": 6380 + }, + { + "epoch": 18.961424332344215, + "grad_norm": 0.09003588557243347, + "learning_rate": 3.1593772365766105e-05, + "loss": 0.0064, + "step": 6390 + }, + { + "epoch": 18.991097922848663, + "grad_norm": 0.09479653835296631, + "learning_rate": 3.144013755408895e-05, + "loss": 0.0056, + "step": 6400 + }, + { + "epoch": 19.020771513353115, + "grad_norm": 0.08929464966058731, + "learning_rate": 3.128670571009399e-05, + "loss": 0.0061, + "step": 6410 + }, + { + "epoch": 19.050445103857566, + "grad_norm": 0.08314875513315201, + "learning_rate": 3.113347851168721e-05, + "loss": 0.0063, + "step": 6420 + }, + { + "epoch": 19.080118694362017, + "grad_norm": 0.07534530013799667, + "learning_rate": 3.098045763453678e-05, + "loss": 0.0043, + "step": 6430 + }, + { + "epoch": 19.10979228486647, + "grad_norm": 0.0726943388581276, + "learning_rate": 3.082764475205442e-05, + "loss": 0.0053, + "step": 6440 + }, + { + "epoch": 19.13946587537092, + "grad_norm": 0.06237198784947395, + "learning_rate": 3.0675041535377405e-05, + "loss": 0.0039, + "step": 6450 + }, + { + "epoch": 19.16913946587537, + "grad_norm": 0.05938175693154335, + "learning_rate": 3.052264965335e-05, + "loss": 0.0051, + "step": 6460 + }, + { + "epoch": 19.198813056379823, + "grad_norm": 0.1082092896103859, + "learning_rate": 3.0370470772505433e-05, + "loss": 0.0065, + "step": 6470 + }, + { + "epoch": 19.228486646884274, + "grad_norm": 0.1327633559703827, + "learning_rate": 3.0218506557047598e-05, + "loss": 0.0055, + "step": 6480 + }, + { + "epoch": 19.258160237388726, + "grad_norm": 0.07017289102077484, + "learning_rate": 3.006675866883275e-05, + "loss": 0.0051, + "step": 6490 + }, + { + "epoch": 19.287833827893174, + "grad_norm": 0.10289981216192245, + "learning_rate": 2.991522876735154e-05, + "loss": 0.0049, + "step": 6500 + }, + { + "epoch": 19.317507418397625, + "grad_norm": 0.1318645030260086, + "learning_rate": 2.976391850971065e-05, + "loss": 0.0067, + "step": 6510 + }, + { + "epoch": 19.347181008902076, + "grad_norm": 0.10172990709543228, + "learning_rate": 2.9612829550614836e-05, + "loss": 0.0088, + "step": 6520 + }, + { + "epoch": 19.376854599406528, + "grad_norm": 0.13238836824893951, + "learning_rate": 2.9461963542348737e-05, + "loss": 0.0061, + "step": 6530 + }, + { + "epoch": 19.40652818991098, + "grad_norm": 0.05801309645175934, + "learning_rate": 2.931132213475884e-05, + "loss": 0.0064, + "step": 6540 + }, + { + "epoch": 19.43620178041543, + "grad_norm": 0.10104802250862122, + "learning_rate": 2.916090697523549e-05, + "loss": 0.0058, + "step": 6550 + }, + { + "epoch": 19.465875370919882, + "grad_norm": 0.07685786485671997, + "learning_rate": 2.9010719708694722e-05, + "loss": 0.0061, + "step": 6560 + }, + { + "epoch": 19.495548961424333, + "grad_norm": 0.12206763029098511, + "learning_rate": 2.8860761977560436e-05, + "loss": 0.0091, + "step": 6570 + }, + { + "epoch": 19.525222551928785, + "grad_norm": 0.08354529738426208, + "learning_rate": 2.8711035421746367e-05, + "loss": 0.0095, + "step": 6580 + }, + { + "epoch": 19.554896142433236, + "grad_norm": 0.14100633561611176, + "learning_rate": 2.8561541678638142e-05, + "loss": 0.0062, + "step": 6590 + }, + { + "epoch": 19.584569732937684, + "grad_norm": 0.08803468197584152, + "learning_rate": 2.8412282383075363e-05, + "loss": 0.0057, + "step": 6600 + }, + { + "epoch": 19.614243323442135, + "grad_norm": 0.05870945379137993, + "learning_rate": 2.8263259167333777e-05, + "loss": 0.0051, + "step": 6610 + }, + { + "epoch": 19.643916913946587, + "grad_norm": 0.09367851912975311, + "learning_rate": 2.811447366110741e-05, + "loss": 0.0056, + "step": 6620 + }, + { + "epoch": 19.673590504451038, + "grad_norm": 0.1148865818977356, + "learning_rate": 2.7965927491490705e-05, + "loss": 0.0074, + "step": 6630 + }, + { + "epoch": 19.70326409495549, + "grad_norm": 0.06443727761507034, + "learning_rate": 2.7817622282960815e-05, + "loss": 0.0057, + "step": 6640 + }, + { + "epoch": 19.73293768545994, + "grad_norm": 0.06887785345315933, + "learning_rate": 2.766955965735968e-05, + "loss": 0.0046, + "step": 6650 + }, + { + "epoch": 19.762611275964392, + "grad_norm": 0.09389995038509369, + "learning_rate": 2.7521741233876496e-05, + "loss": 0.0055, + "step": 6660 + }, + { + "epoch": 19.792284866468844, + "grad_norm": 0.1363907903432846, + "learning_rate": 2.7374168629029813e-05, + "loss": 0.0053, + "step": 6670 + }, + { + "epoch": 19.821958456973295, + "grad_norm": 0.08125721663236618, + "learning_rate": 2.7226843456650037e-05, + "loss": 0.0063, + "step": 6680 + }, + { + "epoch": 19.851632047477743, + "grad_norm": 0.10988637804985046, + "learning_rate": 2.707976732786166e-05, + "loss": 0.0055, + "step": 6690 + }, + { + "epoch": 19.881305637982194, + "grad_norm": 0.11077634245157242, + "learning_rate": 2.693294185106562e-05, + "loss": 0.0051, + "step": 6700 + }, + { + "epoch": 19.910979228486646, + "grad_norm": 0.13356263935565948, + "learning_rate": 2.6786368631921836e-05, + "loss": 0.0053, + "step": 6710 + }, + { + "epoch": 19.940652818991097, + "grad_norm": 0.133230060338974, + "learning_rate": 2.6640049273331515e-05, + "loss": 0.0056, + "step": 6720 + }, + { + "epoch": 19.97032640949555, + "grad_norm": 0.10850989818572998, + "learning_rate": 2.6493985375419778e-05, + "loss": 0.0065, + "step": 6730 + }, + { + "epoch": 20.0, + "grad_norm": 0.08912710100412369, + "learning_rate": 2.6348178535517966e-05, + "loss": 0.007, + "step": 6740 + }, + { + "epoch": 20.02967359050445, + "grad_norm": 0.054864369332790375, + "learning_rate": 2.6202630348146324e-05, + "loss": 0.0049, + "step": 6750 + }, + { + "epoch": 20.059347181008903, + "grad_norm": 0.11972086131572723, + "learning_rate": 2.6057342404996522e-05, + "loss": 0.0052, + "step": 6760 + }, + { + "epoch": 20.089020771513354, + "grad_norm": 0.0904167965054512, + "learning_rate": 2.591231629491423e-05, + "loss": 0.004, + "step": 6770 + }, + { + "epoch": 20.118694362017806, + "grad_norm": 0.04706908389925957, + "learning_rate": 2.5767553603881767e-05, + "loss": 0.0047, + "step": 6780 + }, + { + "epoch": 20.148367952522253, + "grad_norm": 0.044308267533779144, + "learning_rate": 2.562305591500069e-05, + "loss": 0.0047, + "step": 6790 + }, + { + "epoch": 20.178041543026705, + "grad_norm": 0.16575434803962708, + "learning_rate": 2.547882480847461e-05, + "loss": 0.0084, + "step": 6800 + }, + { + "epoch": 20.207715133531156, + "grad_norm": 0.08881547302007675, + "learning_rate": 2.5334861861591753e-05, + "loss": 0.0039, + "step": 6810 + }, + { + "epoch": 20.237388724035608, + "grad_norm": 0.08442889899015427, + "learning_rate": 2.5191168648707887e-05, + "loss": 0.0048, + "step": 6820 + }, + { + "epoch": 20.26706231454006, + "grad_norm": 0.05308566987514496, + "learning_rate": 2.5047746741228978e-05, + "loss": 0.0043, + "step": 6830 + }, + { + "epoch": 20.29673590504451, + "grad_norm": 0.08365131169557571, + "learning_rate": 2.490459770759398e-05, + "loss": 0.005, + "step": 6840 + }, + { + "epoch": 20.326409495548962, + "grad_norm": 0.035201624035835266, + "learning_rate": 2.476172311325783e-05, + "loss": 0.004, + "step": 6850 + }, + { + "epoch": 20.356083086053413, + "grad_norm": 0.05819859728217125, + "learning_rate": 2.4619124520674146e-05, + "loss": 0.0063, + "step": 6860 + }, + { + "epoch": 20.385756676557865, + "grad_norm": 0.12480951100587845, + "learning_rate": 2.447680348927837e-05, + "loss": 0.0077, + "step": 6870 + }, + { + "epoch": 20.415430267062316, + "grad_norm": 0.09439083188772202, + "learning_rate": 2.433476157547044e-05, + "loss": 0.0084, + "step": 6880 + }, + { + "epoch": 20.445103857566764, + "grad_norm": 0.06064935028553009, + "learning_rate": 2.419300033259798e-05, + "loss": 0.006, + "step": 6890 + }, + { + "epoch": 20.474777448071215, + "grad_norm": 0.062327805906534195, + "learning_rate": 2.405152131093926e-05, + "loss": 0.005, + "step": 6900 + }, + { + "epoch": 20.504451038575667, + "grad_norm": 0.08613722771406174, + "learning_rate": 2.3910326057686127e-05, + "loss": 0.0054, + "step": 6910 + }, + { + "epoch": 20.534124629080118, + "grad_norm": 0.1044430062174797, + "learning_rate": 2.3769416116927335e-05, + "loss": 0.0095, + "step": 6920 + }, + { + "epoch": 20.56379821958457, + "grad_norm": 0.05803578346967697, + "learning_rate": 2.362879302963135e-05, + "loss": 0.0039, + "step": 6930 + }, + { + "epoch": 20.59347181008902, + "grad_norm": 0.049366164952516556, + "learning_rate": 2.3488458333629777e-05, + "loss": 0.0036, + "step": 6940 + }, + { + "epoch": 20.623145400593472, + "grad_norm": 0.07159484922885895, + "learning_rate": 2.3348413563600325e-05, + "loss": 0.0069, + "step": 6950 + }, + { + "epoch": 20.652818991097924, + "grad_norm": 0.11064313352108002, + "learning_rate": 2.3208660251050158e-05, + "loss": 0.0052, + "step": 6960 + }, + { + "epoch": 20.682492581602375, + "grad_norm": 0.13685272634029388, + "learning_rate": 2.3069199924299174e-05, + "loss": 0.0069, + "step": 6970 + }, + { + "epoch": 20.712166172106826, + "grad_norm": 0.08148929476737976, + "learning_rate": 2.29300341084631e-05, + "loss": 0.0048, + "step": 6980 + }, + { + "epoch": 20.741839762611274, + "grad_norm": 0.07075554132461548, + "learning_rate": 2.279116432543705e-05, + "loss": 0.0068, + "step": 6990 + }, + { + "epoch": 20.771513353115726, + "grad_norm": 0.10635301470756531, + "learning_rate": 2.2652592093878666e-05, + "loss": 0.0048, + "step": 7000 + } + ], + "logging_steps": 10, + "max_steps": 10000, + "num_input_tokens_seen": 0, + "num_train_epochs": 30, + "save_steps": 500, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": false + }, + "attributes": {} + } + }, + "total_flos": 5.543762269621453e+18, + "train_batch_size": 32, + "trial_name": null, + "trial_params": null +}