diff --git "a/trainer_state.json" "b/trainer_state.json" new file mode 100644--- /dev/null +++ "b/trainer_state.json" @@ -0,0 +1,7034 @@ +{ + "best_global_step": null, + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 6.8119891008174385, + "eval_steps": 500, + "global_step": 10000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.006811989100817439, + "grad_norm": 13.56417465209961, + "learning_rate": 1.8e-06, + "loss": 0.9521, + "step": 10 + }, + { + "epoch": 0.013623978201634877, + "grad_norm": 4.532983779907227, + "learning_rate": 3.8e-06, + "loss": 0.7267, + "step": 20 + }, + { + "epoch": 0.020435967302452316, + "grad_norm": 3.591306209564209, + "learning_rate": 5.8e-06, + "loss": 0.3512, + "step": 30 + }, + { + "epoch": 0.027247956403269755, + "grad_norm": 1.6637485027313232, + "learning_rate": 7.8e-06, + "loss": 0.2738, + "step": 40 + }, + { + "epoch": 0.0340599455040872, + "grad_norm": 2.53169846534729, + "learning_rate": 9.800000000000001e-06, + "loss": 0.2488, + "step": 50 + }, + { + "epoch": 0.04087193460490463, + "grad_norm": 1.7912914752960205, + "learning_rate": 1.18e-05, + "loss": 0.2176, + "step": 60 + }, + { + "epoch": 0.047683923705722074, + "grad_norm": 1.1671632528305054, + "learning_rate": 1.3800000000000002e-05, + "loss": 0.1956, + "step": 70 + }, + { + "epoch": 0.05449591280653951, + "grad_norm": 1.4954109191894531, + "learning_rate": 1.58e-05, + "loss": 0.1803, + "step": 80 + }, + { + "epoch": 0.06130790190735695, + "grad_norm": 1.4526729583740234, + "learning_rate": 1.78e-05, + "loss": 0.1613, + "step": 90 + }, + { + "epoch": 0.0681198910081744, + "grad_norm": 1.900531530380249, + "learning_rate": 1.9800000000000004e-05, + "loss": 0.1423, + "step": 100 + }, + { + "epoch": 0.07493188010899182, + "grad_norm": 0.9216407537460327, + "learning_rate": 2.18e-05, + "loss": 0.1423, + "step": 110 + }, + { + "epoch": 0.08174386920980926, + "grad_norm": 1.3348183631896973, + "learning_rate": 2.38e-05, + "loss": 0.1419, + "step": 120 + }, + { + "epoch": 0.0885558583106267, + "grad_norm": 1.1559427976608276, + "learning_rate": 2.58e-05, + "loss": 0.1346, + "step": 130 + }, + { + "epoch": 0.09536784741144415, + "grad_norm": 1.0173367261886597, + "learning_rate": 2.7800000000000005e-05, + "loss": 0.129, + "step": 140 + }, + { + "epoch": 0.10217983651226158, + "grad_norm": 1.371229648590088, + "learning_rate": 2.98e-05, + "loss": 0.1309, + "step": 150 + }, + { + "epoch": 0.10899182561307902, + "grad_norm": 1.3259838819503784, + "learning_rate": 3.18e-05, + "loss": 0.1214, + "step": 160 + }, + { + "epoch": 0.11580381471389646, + "grad_norm": 1.3470969200134277, + "learning_rate": 3.38e-05, + "loss": 0.1215, + "step": 170 + }, + { + "epoch": 0.1226158038147139, + "grad_norm": 1.3876326084136963, + "learning_rate": 3.58e-05, + "loss": 0.1164, + "step": 180 + }, + { + "epoch": 0.12942779291553133, + "grad_norm": 0.859729528427124, + "learning_rate": 3.7800000000000004e-05, + "loss": 0.0991, + "step": 190 + }, + { + "epoch": 0.1362397820163488, + "grad_norm": 1.047569990158081, + "learning_rate": 3.9800000000000005e-05, + "loss": 0.101, + "step": 200 + }, + { + "epoch": 0.14305177111716622, + "grad_norm": 0.8120327591896057, + "learning_rate": 4.18e-05, + "loss": 0.1127, + "step": 210 + }, + { + "epoch": 0.14986376021798364, + "grad_norm": 0.6889075040817261, + "learning_rate": 4.38e-05, + "loss": 0.0975, + "step": 220 + }, + { + "epoch": 0.1566757493188011, + "grad_norm": 0.7600892782211304, + "learning_rate": 4.58e-05, + "loss": 0.0961, + "step": 230 + }, + { + "epoch": 0.16348773841961853, + "grad_norm": 0.8735463619232178, + "learning_rate": 4.78e-05, + "loss": 0.0949, + "step": 240 + }, + { + "epoch": 0.17029972752043596, + "grad_norm": 1.915482759475708, + "learning_rate": 4.9800000000000004e-05, + "loss": 0.0963, + "step": 250 + }, + { + "epoch": 0.1771117166212534, + "grad_norm": 0.9752621054649353, + "learning_rate": 5.1800000000000005e-05, + "loss": 0.0951, + "step": 260 + }, + { + "epoch": 0.18392370572207084, + "grad_norm": 0.9178920388221741, + "learning_rate": 5.380000000000001e-05, + "loss": 0.1008, + "step": 270 + }, + { + "epoch": 0.1907356948228883, + "grad_norm": 0.5977846384048462, + "learning_rate": 5.580000000000001e-05, + "loss": 0.0875, + "step": 280 + }, + { + "epoch": 0.19754768392370572, + "grad_norm": 0.9973370432853699, + "learning_rate": 5.7799999999999995e-05, + "loss": 0.0952, + "step": 290 + }, + { + "epoch": 0.20435967302452315, + "grad_norm": 0.8910614848136902, + "learning_rate": 5.9800000000000003e-05, + "loss": 0.0844, + "step": 300 + }, + { + "epoch": 0.2111716621253406, + "grad_norm": 1.045469045639038, + "learning_rate": 6.18e-05, + "loss": 0.0854, + "step": 310 + }, + { + "epoch": 0.21798365122615804, + "grad_norm": 0.6841666102409363, + "learning_rate": 6.38e-05, + "loss": 0.0819, + "step": 320 + }, + { + "epoch": 0.22479564032697547, + "grad_norm": 0.9464238882064819, + "learning_rate": 6.58e-05, + "loss": 0.087, + "step": 330 + }, + { + "epoch": 0.23160762942779292, + "grad_norm": 1.0978114604949951, + "learning_rate": 6.780000000000001e-05, + "loss": 0.0823, + "step": 340 + }, + { + "epoch": 0.23841961852861035, + "grad_norm": 1.4373971223831177, + "learning_rate": 6.98e-05, + "loss": 0.0815, + "step": 350 + }, + { + "epoch": 0.2452316076294278, + "grad_norm": 1.202044129371643, + "learning_rate": 7.18e-05, + "loss": 0.0807, + "step": 360 + }, + { + "epoch": 0.25204359673024523, + "grad_norm": 0.5167035460472107, + "learning_rate": 7.38e-05, + "loss": 0.0807, + "step": 370 + }, + { + "epoch": 0.25885558583106266, + "grad_norm": 0.7529540061950684, + "learning_rate": 7.58e-05, + "loss": 0.0724, + "step": 380 + }, + { + "epoch": 0.2656675749318801, + "grad_norm": 0.8727638721466064, + "learning_rate": 7.780000000000001e-05, + "loss": 0.0736, + "step": 390 + }, + { + "epoch": 0.2724795640326976, + "grad_norm": 0.7674930691719055, + "learning_rate": 7.98e-05, + "loss": 0.0732, + "step": 400 + }, + { + "epoch": 0.279291553133515, + "grad_norm": 0.8788486123085022, + "learning_rate": 8.18e-05, + "loss": 0.0776, + "step": 410 + }, + { + "epoch": 0.28610354223433243, + "grad_norm": 0.8660708665847778, + "learning_rate": 8.38e-05, + "loss": 0.0716, + "step": 420 + }, + { + "epoch": 0.29291553133514986, + "grad_norm": 0.7786921262741089, + "learning_rate": 8.58e-05, + "loss": 0.0755, + "step": 430 + }, + { + "epoch": 0.2997275204359673, + "grad_norm": 0.5030098557472229, + "learning_rate": 8.78e-05, + "loss": 0.0663, + "step": 440 + }, + { + "epoch": 0.3065395095367847, + "grad_norm": 0.9466328024864197, + "learning_rate": 8.98e-05, + "loss": 0.0714, + "step": 450 + }, + { + "epoch": 0.3133514986376022, + "grad_norm": 1.1559456586837769, + "learning_rate": 9.180000000000001e-05, + "loss": 0.0831, + "step": 460 + }, + { + "epoch": 0.3201634877384196, + "grad_norm": 1.2573426961898804, + "learning_rate": 9.38e-05, + "loss": 0.0829, + "step": 470 + }, + { + "epoch": 0.32697547683923706, + "grad_norm": 0.7151256203651428, + "learning_rate": 9.58e-05, + "loss": 0.085, + "step": 480 + }, + { + "epoch": 0.3337874659400545, + "grad_norm": 0.6353095173835754, + "learning_rate": 9.78e-05, + "loss": 0.0768, + "step": 490 + }, + { + "epoch": 0.3405994550408719, + "grad_norm": 0.8536468744277954, + "learning_rate": 9.98e-05, + "loss": 0.0735, + "step": 500 + }, + { + "epoch": 0.3474114441416894, + "grad_norm": 0.6806540489196777, + "learning_rate": 9.9999778549206e-05, + "loss": 0.0743, + "step": 510 + }, + { + "epoch": 0.3542234332425068, + "grad_norm": 0.8856997489929199, + "learning_rate": 9.999901304280685e-05, + "loss": 0.071, + "step": 520 + }, + { + "epoch": 0.36103542234332425, + "grad_norm": 0.5865021347999573, + "learning_rate": 9.999770075521164e-05, + "loss": 0.0754, + "step": 530 + }, + { + "epoch": 0.3678474114441417, + "grad_norm": 0.8031758666038513, + "learning_rate": 9.99958417007713e-05, + "loss": 0.0705, + "step": 540 + }, + { + "epoch": 0.3746594005449591, + "grad_norm": 0.4727207124233246, + "learning_rate": 9.999343589981615e-05, + "loss": 0.0666, + "step": 550 + }, + { + "epoch": 0.3814713896457766, + "grad_norm": 0.7345871329307556, + "learning_rate": 9.999048337865568e-05, + "loss": 0.065, + "step": 560 + }, + { + "epoch": 0.388283378746594, + "grad_norm": 0.43568849563598633, + "learning_rate": 9.998698416957815e-05, + "loss": 0.0604, + "step": 570 + }, + { + "epoch": 0.39509536784741145, + "grad_norm": 0.7979448437690735, + "learning_rate": 9.998293831085037e-05, + "loss": 0.0677, + "step": 580 + }, + { + "epoch": 0.4019073569482289, + "grad_norm": 0.8345848321914673, + "learning_rate": 9.997834584671719e-05, + "loss": 0.0665, + "step": 590 + }, + { + "epoch": 0.4087193460490463, + "grad_norm": 0.9694622159004211, + "learning_rate": 9.997320682740107e-05, + "loss": 0.0702, + "step": 600 + }, + { + "epoch": 0.41553133514986373, + "grad_norm": 0.4619634747505188, + "learning_rate": 9.996752130910149e-05, + "loss": 0.0728, + "step": 610 + }, + { + "epoch": 0.4223433242506812, + "grad_norm": 1.178167462348938, + "learning_rate": 9.99612893539944e-05, + "loss": 0.07, + "step": 620 + }, + { + "epoch": 0.42915531335149865, + "grad_norm": 0.7508357763290405, + "learning_rate": 9.995451103023144e-05, + "loss": 0.06, + "step": 630 + }, + { + "epoch": 0.4359673024523161, + "grad_norm": 0.6789131760597229, + "learning_rate": 9.994718641193928e-05, + "loss": 0.0584, + "step": 640 + }, + { + "epoch": 0.4427792915531335, + "grad_norm": 0.632141649723053, + "learning_rate": 9.993931557921874e-05, + "loss": 0.0642, + "step": 650 + }, + { + "epoch": 0.44959128065395093, + "grad_norm": 0.780634880065918, + "learning_rate": 9.993089861814402e-05, + "loss": 0.0597, + "step": 660 + }, + { + "epoch": 0.4564032697547684, + "grad_norm": 0.411491721868515, + "learning_rate": 9.992193562076166e-05, + "loss": 0.0692, + "step": 670 + }, + { + "epoch": 0.46321525885558584, + "grad_norm": 0.5659852027893066, + "learning_rate": 9.991242668508954e-05, + "loss": 0.0644, + "step": 680 + }, + { + "epoch": 0.47002724795640327, + "grad_norm": 0.7057250142097473, + "learning_rate": 9.990237191511587e-05, + "loss": 0.0641, + "step": 690 + }, + { + "epoch": 0.4768392370572207, + "grad_norm": 0.5383998155593872, + "learning_rate": 9.989177142079802e-05, + "loss": 0.064, + "step": 700 + }, + { + "epoch": 0.48365122615803813, + "grad_norm": 1.0508489608764648, + "learning_rate": 9.988062531806126e-05, + "loss": 0.0722, + "step": 710 + }, + { + "epoch": 0.4904632152588556, + "grad_norm": 0.7634751796722412, + "learning_rate": 9.986893372879762e-05, + "loss": 0.0671, + "step": 720 + }, + { + "epoch": 0.49727520435967304, + "grad_norm": 0.5967621207237244, + "learning_rate": 9.985669678086443e-05, + "loss": 0.0609, + "step": 730 + }, + { + "epoch": 0.5040871934604905, + "grad_norm": 0.6123813986778259, + "learning_rate": 9.984391460808298e-05, + "loss": 0.0536, + "step": 740 + }, + { + "epoch": 0.510899182561308, + "grad_norm": 0.45927202701568604, + "learning_rate": 9.983058735023709e-05, + "loss": 0.0598, + "step": 750 + }, + { + "epoch": 0.5177111716621253, + "grad_norm": 0.3762418031692505, + "learning_rate": 9.98167151530715e-05, + "loss": 0.049, + "step": 760 + }, + { + "epoch": 0.5245231607629428, + "grad_norm": 0.6599445939064026, + "learning_rate": 9.980229816829034e-05, + "loss": 0.0508, + "step": 770 + }, + { + "epoch": 0.5313351498637602, + "grad_norm": 0.36847731471061707, + "learning_rate": 9.978733655355544e-05, + "loss": 0.0575, + "step": 780 + }, + { + "epoch": 0.5381471389645777, + "grad_norm": 0.41056740283966064, + "learning_rate": 9.977183047248464e-05, + "loss": 0.0542, + "step": 790 + }, + { + "epoch": 0.5449591280653951, + "grad_norm": 0.6728678345680237, + "learning_rate": 9.975578009464992e-05, + "loss": 0.0577, + "step": 800 + }, + { + "epoch": 0.5517711171662125, + "grad_norm": 0.4020206928253174, + "learning_rate": 9.97391855955757e-05, + "loss": 0.0586, + "step": 810 + }, + { + "epoch": 0.55858310626703, + "grad_norm": 0.9495147466659546, + "learning_rate": 9.972204715673669e-05, + "loss": 0.0533, + "step": 820 + }, + { + "epoch": 0.5653950953678474, + "grad_norm": 0.48653340339660645, + "learning_rate": 9.970436496555617e-05, + "loss": 0.0535, + "step": 830 + }, + { + "epoch": 0.5722070844686649, + "grad_norm": 0.5679882764816284, + "learning_rate": 9.968613921540373e-05, + "loss": 0.0482, + "step": 840 + }, + { + "epoch": 0.5790190735694822, + "grad_norm": 0.5090594291687012, + "learning_rate": 9.966737010559326e-05, + "loss": 0.0609, + "step": 850 + }, + { + "epoch": 0.5858310626702997, + "grad_norm": 0.4095969796180725, + "learning_rate": 9.964805784138072e-05, + "loss": 0.0527, + "step": 860 + }, + { + "epoch": 0.5926430517711172, + "grad_norm": 0.5614601373672485, + "learning_rate": 9.962820263396195e-05, + "loss": 0.0516, + "step": 870 + }, + { + "epoch": 0.5994550408719346, + "grad_norm": 0.6245532035827637, + "learning_rate": 9.960780470047033e-05, + "loss": 0.044, + "step": 880 + }, + { + "epoch": 0.6062670299727521, + "grad_norm": 0.38533297181129456, + "learning_rate": 9.958686426397437e-05, + "loss": 0.0444, + "step": 890 + }, + { + "epoch": 0.6130790190735694, + "grad_norm": 0.5443320274353027, + "learning_rate": 9.956538155347534e-05, + "loss": 0.0549, + "step": 900 + }, + { + "epoch": 0.6198910081743869, + "grad_norm": 0.5260657072067261, + "learning_rate": 9.95433568039047e-05, + "loss": 0.0442, + "step": 910 + }, + { + "epoch": 0.6267029972752044, + "grad_norm": 0.46750691533088684, + "learning_rate": 9.952079025612162e-05, + "loss": 0.0536, + "step": 920 + }, + { + "epoch": 0.6335149863760218, + "grad_norm": 0.5516590476036072, + "learning_rate": 9.949768215691022e-05, + "loss": 0.047, + "step": 930 + }, + { + "epoch": 0.6403269754768393, + "grad_norm": 0.5147327780723572, + "learning_rate": 9.9474032758977e-05, + "loss": 0.0441, + "step": 940 + }, + { + "epoch": 0.6471389645776566, + "grad_norm": 0.8653050065040588, + "learning_rate": 9.944984232094794e-05, + "loss": 0.0599, + "step": 950 + }, + { + "epoch": 0.6539509536784741, + "grad_norm": 0.7256223559379578, + "learning_rate": 9.942511110736584e-05, + "loss": 0.0469, + "step": 960 + }, + { + "epoch": 0.6607629427792916, + "grad_norm": 0.5604241490364075, + "learning_rate": 9.939983938868726e-05, + "loss": 0.0493, + "step": 970 + }, + { + "epoch": 0.667574931880109, + "grad_norm": 0.603168785572052, + "learning_rate": 9.93740274412797e-05, + "loss": 0.0526, + "step": 980 + }, + { + "epoch": 0.6743869209809265, + "grad_norm": 0.779745876789093, + "learning_rate": 9.934767554741846e-05, + "loss": 0.0538, + "step": 990 + }, + { + "epoch": 0.6811989100817438, + "grad_norm": 0.37556472420692444, + "learning_rate": 9.932078399528361e-05, + "loss": 0.0485, + "step": 1000 + }, + { + "epoch": 0.6880108991825613, + "grad_norm": 0.7715975642204285, + "learning_rate": 9.929335307895689e-05, + "loss": 0.0447, + "step": 1010 + }, + { + "epoch": 0.6948228882833788, + "grad_norm": 0.26802629232406616, + "learning_rate": 9.926538309841839e-05, + "loss": 0.0494, + "step": 1020 + }, + { + "epoch": 0.7016348773841962, + "grad_norm": 0.789804220199585, + "learning_rate": 9.923687435954334e-05, + "loss": 0.0527, + "step": 1030 + }, + { + "epoch": 0.7084468664850136, + "grad_norm": 0.540306806564331, + "learning_rate": 9.920782717409873e-05, + "loss": 0.0488, + "step": 1040 + }, + { + "epoch": 0.715258855585831, + "grad_norm": 0.4782036244869232, + "learning_rate": 9.917824185973994e-05, + "loss": 0.044, + "step": 1050 + }, + { + "epoch": 0.7220708446866485, + "grad_norm": 0.7395116686820984, + "learning_rate": 9.914811874000723e-05, + "loss": 0.0547, + "step": 1060 + }, + { + "epoch": 0.728882833787466, + "grad_norm": 0.8170892000198364, + "learning_rate": 9.911745814432218e-05, + "loss": 0.0524, + "step": 1070 + }, + { + "epoch": 0.7356948228882834, + "grad_norm": 0.5740767121315002, + "learning_rate": 9.90862604079842e-05, + "loss": 0.0537, + "step": 1080 + }, + { + "epoch": 0.7425068119891008, + "grad_norm": 0.6334712505340576, + "learning_rate": 9.90545258721667e-05, + "loss": 0.0432, + "step": 1090 + }, + { + "epoch": 0.7493188010899182, + "grad_norm": 0.5951853394508362, + "learning_rate": 9.90222548839135e-05, + "loss": 0.0455, + "step": 1100 + }, + { + "epoch": 0.7561307901907357, + "grad_norm": 0.5218120217323303, + "learning_rate": 9.898944779613495e-05, + "loss": 0.0472, + "step": 1110 + }, + { + "epoch": 0.7629427792915532, + "grad_norm": 0.6090676784515381, + "learning_rate": 9.89561049676041e-05, + "loss": 0.0477, + "step": 1120 + }, + { + "epoch": 0.7697547683923706, + "grad_norm": 0.3142809569835663, + "learning_rate": 9.89222267629528e-05, + "loss": 0.0505, + "step": 1130 + }, + { + "epoch": 0.776566757493188, + "grad_norm": 0.35169437527656555, + "learning_rate": 9.888781355266763e-05, + "loss": 0.0467, + "step": 1140 + }, + { + "epoch": 0.7833787465940054, + "grad_norm": 0.5377322435379028, + "learning_rate": 9.885286571308598e-05, + "loss": 0.0451, + "step": 1150 + }, + { + "epoch": 0.7901907356948229, + "grad_norm": 0.37710654735565186, + "learning_rate": 9.881738362639182e-05, + "loss": 0.0421, + "step": 1160 + }, + { + "epoch": 0.7970027247956403, + "grad_norm": 0.6840143203735352, + "learning_rate": 9.878136768061154e-05, + "loss": 0.0456, + "step": 1170 + }, + { + "epoch": 0.8038147138964578, + "grad_norm": 0.6212213635444641, + "learning_rate": 9.874481826960979e-05, + "loss": 0.0429, + "step": 1180 + }, + { + "epoch": 0.8106267029972752, + "grad_norm": 0.4762314558029175, + "learning_rate": 9.870773579308503e-05, + "loss": 0.045, + "step": 1190 + }, + { + "epoch": 0.8174386920980926, + "grad_norm": 0.32645371556282043, + "learning_rate": 9.867012065656533e-05, + "loss": 0.0492, + "step": 1200 + }, + { + "epoch": 0.8242506811989101, + "grad_norm": 0.5410632491111755, + "learning_rate": 9.863197327140376e-05, + "loss": 0.0472, + "step": 1210 + }, + { + "epoch": 0.8310626702997275, + "grad_norm": 0.39453989267349243, + "learning_rate": 9.859329405477403e-05, + "loss": 0.0409, + "step": 1220 + }, + { + "epoch": 0.837874659400545, + "grad_norm": 0.4603775143623352, + "learning_rate": 9.855408342966585e-05, + "loss": 0.047, + "step": 1230 + }, + { + "epoch": 0.8446866485013624, + "grad_norm": 0.6546939015388489, + "learning_rate": 9.851434182488033e-05, + "loss": 0.0416, + "step": 1240 + }, + { + "epoch": 0.8514986376021798, + "grad_norm": 0.7223520278930664, + "learning_rate": 9.84740696750253e-05, + "loss": 0.0519, + "step": 1250 + }, + { + "epoch": 0.8583106267029973, + "grad_norm": 0.5285854339599609, + "learning_rate": 9.843326742051055e-05, + "loss": 0.0455, + "step": 1260 + }, + { + "epoch": 0.8651226158038147, + "grad_norm": 0.5807820558547974, + "learning_rate": 9.839193550754297e-05, + "loss": 0.0466, + "step": 1270 + }, + { + "epoch": 0.8719346049046321, + "grad_norm": 0.6203693151473999, + "learning_rate": 9.835007438812177e-05, + "loss": 0.0463, + "step": 1280 + }, + { + "epoch": 0.8787465940054496, + "grad_norm": 0.4158671498298645, + "learning_rate": 9.830768452003341e-05, + "loss": 0.0477, + "step": 1290 + }, + { + "epoch": 0.885558583106267, + "grad_norm": 0.5649051666259766, + "learning_rate": 9.826476636684671e-05, + "loss": 0.0451, + "step": 1300 + }, + { + "epoch": 0.8923705722070845, + "grad_norm": 0.7309224605560303, + "learning_rate": 9.822132039790773e-05, + "loss": 0.0412, + "step": 1310 + }, + { + "epoch": 0.8991825613079019, + "grad_norm": 0.4310065805912018, + "learning_rate": 9.817734708833461e-05, + "loss": 0.045, + "step": 1320 + }, + { + "epoch": 0.9059945504087193, + "grad_norm": 0.49696084856987, + "learning_rate": 9.813284691901243e-05, + "loss": 0.0375, + "step": 1330 + }, + { + "epoch": 0.9128065395095368, + "grad_norm": 0.5691936016082764, + "learning_rate": 9.808782037658792e-05, + "loss": 0.0411, + "step": 1340 + }, + { + "epoch": 0.9196185286103542, + "grad_norm": 0.4932763874530792, + "learning_rate": 9.804226795346411e-05, + "loss": 0.0402, + "step": 1350 + }, + { + "epoch": 0.9264305177111717, + "grad_norm": 0.4159316122531891, + "learning_rate": 9.799619014779503e-05, + "loss": 0.0434, + "step": 1360 + }, + { + "epoch": 0.9332425068119891, + "grad_norm": 0.6000523567199707, + "learning_rate": 9.794958746348013e-05, + "loss": 0.041, + "step": 1370 + }, + { + "epoch": 0.9400544959128065, + "grad_norm": 0.48650360107421875, + "learning_rate": 9.790246041015896e-05, + "loss": 0.041, + "step": 1380 + }, + { + "epoch": 0.946866485013624, + "grad_norm": 0.5935161709785461, + "learning_rate": 9.785480950320538e-05, + "loss": 0.0415, + "step": 1390 + }, + { + "epoch": 0.9536784741144414, + "grad_norm": 0.34310704469680786, + "learning_rate": 9.78066352637221e-05, + "loss": 0.0501, + "step": 1400 + }, + { + "epoch": 0.9604904632152589, + "grad_norm": 0.5283187031745911, + "learning_rate": 9.775793821853488e-05, + "loss": 0.0408, + "step": 1410 + }, + { + "epoch": 0.9673024523160763, + "grad_norm": 0.3886546790599823, + "learning_rate": 9.77087189001868e-05, + "loss": 0.0466, + "step": 1420 + }, + { + "epoch": 0.9741144414168937, + "grad_norm": 0.4705336391925812, + "learning_rate": 9.765897784693243e-05, + "loss": 0.0443, + "step": 1430 + }, + { + "epoch": 0.9809264305177112, + "grad_norm": 0.37797755002975464, + "learning_rate": 9.760871560273197e-05, + "loss": 0.0393, + "step": 1440 + }, + { + "epoch": 0.9877384196185286, + "grad_norm": 0.38777533173561096, + "learning_rate": 9.755793271724526e-05, + "loss": 0.0446, + "step": 1450 + }, + { + "epoch": 0.9945504087193461, + "grad_norm": 0.41121742129325867, + "learning_rate": 9.750662974582584e-05, + "loss": 0.044, + "step": 1460 + }, + { + "epoch": 1.0013623978201636, + "grad_norm": 0.45574951171875, + "learning_rate": 9.745480724951473e-05, + "loss": 0.0416, + "step": 1470 + }, + { + "epoch": 1.008174386920981, + "grad_norm": 0.49513402581214905, + "learning_rate": 9.740246579503447e-05, + "loss": 0.0443, + "step": 1480 + }, + { + "epoch": 1.0149863760217983, + "grad_norm": 0.5963664650917053, + "learning_rate": 9.734960595478284e-05, + "loss": 0.0385, + "step": 1490 + }, + { + "epoch": 1.021798365122616, + "grad_norm": 0.36990484595298767, + "learning_rate": 9.729622830682657e-05, + "loss": 0.047, + "step": 1500 + }, + { + "epoch": 1.0286103542234333, + "grad_norm": 0.5331124067306519, + "learning_rate": 9.724233343489504e-05, + "loss": 0.0387, + "step": 1510 + }, + { + "epoch": 1.0354223433242506, + "grad_norm": 0.2865339517593384, + "learning_rate": 9.718792192837396e-05, + "loss": 0.0393, + "step": 1520 + }, + { + "epoch": 1.042234332425068, + "grad_norm": 0.3951549828052521, + "learning_rate": 9.713299438229886e-05, + "loss": 0.0405, + "step": 1530 + }, + { + "epoch": 1.0490463215258856, + "grad_norm": 0.39627596735954285, + "learning_rate": 9.707755139734855e-05, + "loss": 0.0444, + "step": 1540 + }, + { + "epoch": 1.055858310626703, + "grad_norm": 0.41905856132507324, + "learning_rate": 9.702159357983866e-05, + "loss": 0.0431, + "step": 1550 + }, + { + "epoch": 1.0626702997275204, + "grad_norm": 0.6280357241630554, + "learning_rate": 9.696512154171492e-05, + "loss": 0.0395, + "step": 1560 + }, + { + "epoch": 1.069482288828338, + "grad_norm": 0.5034973621368408, + "learning_rate": 9.690813590054645e-05, + "loss": 0.0444, + "step": 1570 + }, + { + "epoch": 1.0762942779291553, + "grad_norm": 0.22129182517528534, + "learning_rate": 9.685063727951914e-05, + "loss": 0.0371, + "step": 1580 + }, + { + "epoch": 1.0831062670299727, + "grad_norm": 0.3160829246044159, + "learning_rate": 9.679262630742865e-05, + "loss": 0.0425, + "step": 1590 + }, + { + "epoch": 1.0899182561307903, + "grad_norm": 0.41493088006973267, + "learning_rate": 9.673410361867373e-05, + "loss": 0.0374, + "step": 1600 + }, + { + "epoch": 1.0967302452316077, + "grad_norm": 0.22237280011177063, + "learning_rate": 9.667506985324909e-05, + "loss": 0.0317, + "step": 1610 + }, + { + "epoch": 1.103542234332425, + "grad_norm": 0.38369399309158325, + "learning_rate": 9.661552565673855e-05, + "loss": 0.0322, + "step": 1620 + }, + { + "epoch": 1.1103542234332424, + "grad_norm": 0.2891499102115631, + "learning_rate": 9.655547168030789e-05, + "loss": 0.0349, + "step": 1630 + }, + { + "epoch": 1.11716621253406, + "grad_norm": 0.48390528559684753, + "learning_rate": 9.649490858069777e-05, + "loss": 0.0337, + "step": 1640 + }, + { + "epoch": 1.1239782016348774, + "grad_norm": 0.4669130742549896, + "learning_rate": 9.643383702021658e-05, + "loss": 0.0469, + "step": 1650 + }, + { + "epoch": 1.1307901907356948, + "grad_norm": 0.5749063491821289, + "learning_rate": 9.637225766673307e-05, + "loss": 0.0355, + "step": 1660 + }, + { + "epoch": 1.1376021798365124, + "grad_norm": 0.37233132123947144, + "learning_rate": 9.631017119366922e-05, + "loss": 0.0378, + "step": 1670 + }, + { + "epoch": 1.1444141689373297, + "grad_norm": 0.6000690460205078, + "learning_rate": 9.624757827999273e-05, + "loss": 0.0379, + "step": 1680 + }, + { + "epoch": 1.151226158038147, + "grad_norm": 0.39104825258255005, + "learning_rate": 9.618447961020971e-05, + "loss": 0.041, + "step": 1690 + }, + { + "epoch": 1.1580381471389645, + "grad_norm": 0.38393735885620117, + "learning_rate": 9.612087587435707e-05, + "loss": 0.0375, + "step": 1700 + }, + { + "epoch": 1.164850136239782, + "grad_norm": 0.49882692098617554, + "learning_rate": 9.605676776799508e-05, + "loss": 0.0365, + "step": 1710 + }, + { + "epoch": 1.1716621253405994, + "grad_norm": 0.4841519594192505, + "learning_rate": 9.599215599219973e-05, + "loss": 0.0398, + "step": 1720 + }, + { + "epoch": 1.1784741144414168, + "grad_norm": 0.36516526341438293, + "learning_rate": 9.592704125355505e-05, + "loss": 0.0391, + "step": 1730 + }, + { + "epoch": 1.1852861035422344, + "grad_norm": 0.5687127709388733, + "learning_rate": 9.586142426414538e-05, + "loss": 0.0356, + "step": 1740 + }, + { + "epoch": 1.1920980926430518, + "grad_norm": 0.3554752469062805, + "learning_rate": 9.57953057415476e-05, + "loss": 0.0409, + "step": 1750 + }, + { + "epoch": 1.1989100817438691, + "grad_norm": 0.3926330506801605, + "learning_rate": 9.572868640882328e-05, + "loss": 0.0423, + "step": 1760 + }, + { + "epoch": 1.2057220708446867, + "grad_norm": 0.35980191826820374, + "learning_rate": 9.56615669945108e-05, + "loss": 0.0428, + "step": 1770 + }, + { + "epoch": 1.2125340599455041, + "grad_norm": 0.5385282635688782, + "learning_rate": 9.55939482326173e-05, + "loss": 0.037, + "step": 1780 + }, + { + "epoch": 1.2193460490463215, + "grad_norm": 0.42432472109794617, + "learning_rate": 9.552583086261069e-05, + "loss": 0.042, + "step": 1790 + }, + { + "epoch": 1.226158038147139, + "grad_norm": 0.4252088963985443, + "learning_rate": 9.545721562941168e-05, + "loss": 0.0389, + "step": 1800 + }, + { + "epoch": 1.2329700272479565, + "grad_norm": 0.47813108563423157, + "learning_rate": 9.538810328338543e-05, + "loss": 0.0385, + "step": 1810 + }, + { + "epoch": 1.2397820163487738, + "grad_norm": 0.30471813678741455, + "learning_rate": 9.531849458033349e-05, + "loss": 0.0362, + "step": 1820 + }, + { + "epoch": 1.2465940054495912, + "grad_norm": 0.37325453758239746, + "learning_rate": 9.524839028148547e-05, + "loss": 0.0371, + "step": 1830 + }, + { + "epoch": 1.2534059945504088, + "grad_norm": 0.4700666666030884, + "learning_rate": 9.517779115349077e-05, + "loss": 0.0394, + "step": 1840 + }, + { + "epoch": 1.2602179836512262, + "grad_norm": 0.2800864279270172, + "learning_rate": 9.510669796841014e-05, + "loss": 0.0369, + "step": 1850 + }, + { + "epoch": 1.2670299727520435, + "grad_norm": 0.41137367486953735, + "learning_rate": 9.503511150370727e-05, + "loss": 0.0388, + "step": 1860 + }, + { + "epoch": 1.273841961852861, + "grad_norm": 0.40207141637802124, + "learning_rate": 9.496303254224024e-05, + "loss": 0.0385, + "step": 1870 + }, + { + "epoch": 1.2806539509536785, + "grad_norm": 0.23451116681098938, + "learning_rate": 9.489046187225306e-05, + "loss": 0.0356, + "step": 1880 + }, + { + "epoch": 1.2874659400544959, + "grad_norm": 0.22821785509586334, + "learning_rate": 9.481740028736692e-05, + "loss": 0.0321, + "step": 1890 + }, + { + "epoch": 1.2942779291553133, + "grad_norm": 0.35447967052459717, + "learning_rate": 9.474384858657164e-05, + "loss": 0.0391, + "step": 1900 + }, + { + "epoch": 1.3010899182561309, + "grad_norm": 0.304505318403244, + "learning_rate": 9.466980757421679e-05, + "loss": 0.0353, + "step": 1910 + }, + { + "epoch": 1.3079019073569482, + "grad_norm": 0.23178747296333313, + "learning_rate": 9.459527806000305e-05, + "loss": 0.037, + "step": 1920 + }, + { + "epoch": 1.3147138964577656, + "grad_norm": 0.5461723804473877, + "learning_rate": 9.452026085897325e-05, + "loss": 0.0376, + "step": 1930 + }, + { + "epoch": 1.3215258855585832, + "grad_norm": 0.29126253724098206, + "learning_rate": 9.444475679150348e-05, + "loss": 0.0367, + "step": 1940 + }, + { + "epoch": 1.3283378746594006, + "grad_norm": 0.3072151839733124, + "learning_rate": 9.436876668329411e-05, + "loss": 0.0336, + "step": 1950 + }, + { + "epoch": 1.335149863760218, + "grad_norm": 0.36243578791618347, + "learning_rate": 9.429229136536079e-05, + "loss": 0.0307, + "step": 1960 + }, + { + "epoch": 1.3419618528610355, + "grad_norm": 0.43364623188972473, + "learning_rate": 9.421533167402534e-05, + "loss": 0.0403, + "step": 1970 + }, + { + "epoch": 1.348773841961853, + "grad_norm": 0.47131332755088806, + "learning_rate": 9.413788845090666e-05, + "loss": 0.0434, + "step": 1980 + }, + { + "epoch": 1.3555858310626703, + "grad_norm": 0.23641788959503174, + "learning_rate": 9.405996254291136e-05, + "loss": 0.0364, + "step": 1990 + }, + { + "epoch": 1.3623978201634879, + "grad_norm": 0.2776740789413452, + "learning_rate": 9.398155480222474e-05, + "loss": 0.0342, + "step": 2000 + }, + { + "epoch": 1.3692098092643052, + "grad_norm": 0.3794749081134796, + "learning_rate": 9.390266608630128e-05, + "loss": 0.0362, + "step": 2010 + }, + { + "epoch": 1.3760217983651226, + "grad_norm": 0.3998468816280365, + "learning_rate": 9.38232972578553e-05, + "loss": 0.0387, + "step": 2020 + }, + { + "epoch": 1.38283378746594, + "grad_norm": 0.3741637170314789, + "learning_rate": 9.374344918485164e-05, + "loss": 0.0422, + "step": 2030 + }, + { + "epoch": 1.3896457765667574, + "grad_norm": 0.32406580448150635, + "learning_rate": 9.366312274049602e-05, + "loss": 0.0391, + "step": 2040 + }, + { + "epoch": 1.396457765667575, + "grad_norm": 0.3103744387626648, + "learning_rate": 9.358231880322554e-05, + "loss": 0.0395, + "step": 2050 + }, + { + "epoch": 1.4032697547683923, + "grad_norm": 0.581511378288269, + "learning_rate": 9.350103825669916e-05, + "loss": 0.0328, + "step": 2060 + }, + { + "epoch": 1.4100817438692097, + "grad_norm": 0.3604491055011749, + "learning_rate": 9.341928198978787e-05, + "loss": 0.0344, + "step": 2070 + }, + { + "epoch": 1.4168937329700273, + "grad_norm": 0.417709618806839, + "learning_rate": 9.333705089656512e-05, + "loss": 0.0331, + "step": 2080 + }, + { + "epoch": 1.4237057220708447, + "grad_norm": 0.4481496512889862, + "learning_rate": 9.325434587629698e-05, + "loss": 0.0316, + "step": 2090 + }, + { + "epoch": 1.430517711171662, + "grad_norm": 0.47191673517227173, + "learning_rate": 9.31711678334323e-05, + "loss": 0.0326, + "step": 2100 + }, + { + "epoch": 1.4373297002724796, + "grad_norm": 0.32356977462768555, + "learning_rate": 9.308751767759282e-05, + "loss": 0.0323, + "step": 2110 + }, + { + "epoch": 1.444141689373297, + "grad_norm": 0.4607698917388916, + "learning_rate": 9.300339632356325e-05, + "loss": 0.0308, + "step": 2120 + }, + { + "epoch": 1.4509536784741144, + "grad_norm": 0.41386836767196655, + "learning_rate": 9.291880469128124e-05, + "loss": 0.0362, + "step": 2130 + }, + { + "epoch": 1.457765667574932, + "grad_norm": 0.3620763421058655, + "learning_rate": 9.283374370582732e-05, + "loss": 0.0349, + "step": 2140 + }, + { + "epoch": 1.4645776566757494, + "grad_norm": 0.38811424374580383, + "learning_rate": 9.274821429741482e-05, + "loss": 0.0393, + "step": 2150 + }, + { + "epoch": 1.4713896457765667, + "grad_norm": 0.3797459304332733, + "learning_rate": 9.266221740137961e-05, + "loss": 0.0324, + "step": 2160 + }, + { + "epoch": 1.4782016348773843, + "grad_norm": 0.28400975465774536, + "learning_rate": 9.257575395817001e-05, + "loss": 0.0295, + "step": 2170 + }, + { + "epoch": 1.4850136239782017, + "grad_norm": 0.4729847311973572, + "learning_rate": 9.248882491333637e-05, + "loss": 0.0325, + "step": 2180 + }, + { + "epoch": 1.491825613079019, + "grad_norm": 0.39239180088043213, + "learning_rate": 9.240143121752076e-05, + "loss": 0.0296, + "step": 2190 + }, + { + "epoch": 1.4986376021798364, + "grad_norm": 0.416090726852417, + "learning_rate": 9.23135738264467e-05, + "loss": 0.0311, + "step": 2200 + }, + { + "epoch": 1.5054495912806538, + "grad_norm": 0.6459326148033142, + "learning_rate": 9.222525370090849e-05, + "loss": 0.0341, + "step": 2210 + }, + { + "epoch": 1.5122615803814714, + "grad_norm": 0.39463648200035095, + "learning_rate": 9.213647180676088e-05, + "loss": 0.0316, + "step": 2220 + }, + { + "epoch": 1.5190735694822888, + "grad_norm": 0.39738044142723083, + "learning_rate": 9.204722911490846e-05, + "loss": 0.0305, + "step": 2230 + }, + { + "epoch": 1.5258855585831061, + "grad_norm": 0.34590622782707214, + "learning_rate": 9.1957526601295e-05, + "loss": 0.0303, + "step": 2240 + }, + { + "epoch": 1.5326975476839237, + "grad_norm": 0.28665295243263245, + "learning_rate": 9.186736524689281e-05, + "loss": 0.0367, + "step": 2250 + }, + { + "epoch": 1.5395095367847411, + "grad_norm": 0.4147959053516388, + "learning_rate": 9.177674603769204e-05, + "loss": 0.0347, + "step": 2260 + }, + { + "epoch": 1.5463215258855585, + "grad_norm": 0.3655013144016266, + "learning_rate": 9.168566996468983e-05, + "loss": 0.0329, + "step": 2270 + }, + { + "epoch": 1.553133514986376, + "grad_norm": 0.3972167372703552, + "learning_rate": 9.159413802387951e-05, + "loss": 0.0384, + "step": 2280 + }, + { + "epoch": 1.5599455040871935, + "grad_norm": 0.26922062039375305, + "learning_rate": 9.150215121623974e-05, + "loss": 0.0344, + "step": 2290 + }, + { + "epoch": 1.5667574931880108, + "grad_norm": 0.3569802939891815, + "learning_rate": 9.140971054772349e-05, + "loss": 0.0287, + "step": 2300 + }, + { + "epoch": 1.5735694822888284, + "grad_norm": 0.5024522542953491, + "learning_rate": 9.131681702924713e-05, + "loss": 0.0377, + "step": 2310 + }, + { + "epoch": 1.5803814713896458, + "grad_norm": 0.33061105012893677, + "learning_rate": 9.122347167667926e-05, + "loss": 0.0294, + "step": 2320 + }, + { + "epoch": 1.5871934604904632, + "grad_norm": 0.24414274096488953, + "learning_rate": 9.112967551082973e-05, + "loss": 0.0388, + "step": 2330 + }, + { + "epoch": 1.5940054495912808, + "grad_norm": 0.5458868741989136, + "learning_rate": 9.103542955743835e-05, + "loss": 0.0343, + "step": 2340 + }, + { + "epoch": 1.6008174386920981, + "grad_norm": 0.31847676634788513, + "learning_rate": 9.094073484716381e-05, + "loss": 0.0383, + "step": 2350 + }, + { + "epoch": 1.6076294277929155, + "grad_norm": 0.49272584915161133, + "learning_rate": 9.084559241557226e-05, + "loss": 0.0275, + "step": 2360 + }, + { + "epoch": 1.614441416893733, + "grad_norm": 0.2565319240093231, + "learning_rate": 9.075000330312608e-05, + "loss": 0.0334, + "step": 2370 + }, + { + "epoch": 1.6212534059945503, + "grad_norm": 0.37394991517066956, + "learning_rate": 9.065396855517253e-05, + "loss": 0.0284, + "step": 2380 + }, + { + "epoch": 1.6280653950953679, + "grad_norm": 0.30032879114151, + "learning_rate": 9.055748922193219e-05, + "loss": 0.0314, + "step": 2390 + }, + { + "epoch": 1.6348773841961854, + "grad_norm": 0.41679537296295166, + "learning_rate": 9.046056635848761e-05, + "loss": 0.0333, + "step": 2400 + }, + { + "epoch": 1.6416893732970026, + "grad_norm": 0.49561193585395813, + "learning_rate": 9.036320102477169e-05, + "loss": 0.0302, + "step": 2410 + }, + { + "epoch": 1.6485013623978202, + "grad_norm": 0.2882575988769531, + "learning_rate": 9.02653942855561e-05, + "loss": 0.0325, + "step": 2420 + }, + { + "epoch": 1.6553133514986376, + "grad_norm": 0.4606371223926544, + "learning_rate": 9.016714721043971e-05, + "loss": 0.0289, + "step": 2430 + }, + { + "epoch": 1.662125340599455, + "grad_norm": 0.2737872302532196, + "learning_rate": 9.006846087383675e-05, + "loss": 0.0336, + "step": 2440 + }, + { + "epoch": 1.6689373297002725, + "grad_norm": 0.196573406457901, + "learning_rate": 8.996933635496523e-05, + "loss": 0.0286, + "step": 2450 + }, + { + "epoch": 1.67574931880109, + "grad_norm": 0.45029115676879883, + "learning_rate": 8.986977473783498e-05, + "loss": 0.0368, + "step": 2460 + }, + { + "epoch": 1.6825613079019073, + "grad_norm": 0.2532772719860077, + "learning_rate": 8.97697771112359e-05, + "loss": 0.0294, + "step": 2470 + }, + { + "epoch": 1.6893732970027249, + "grad_norm": 0.2481389045715332, + "learning_rate": 8.966934456872602e-05, + "loss": 0.0316, + "step": 2480 + }, + { + "epoch": 1.6961852861035422, + "grad_norm": 0.3074764311313629, + "learning_rate": 8.95684782086195e-05, + "loss": 0.0277, + "step": 2490 + }, + { + "epoch": 1.7029972752043596, + "grad_norm": 0.3800612986087799, + "learning_rate": 8.946717913397476e-05, + "loss": 0.0324, + "step": 2500 + }, + { + "epoch": 1.7098092643051772, + "grad_norm": 0.3243564963340759, + "learning_rate": 8.93654484525822e-05, + "loss": 0.0323, + "step": 2510 + }, + { + "epoch": 1.7166212534059946, + "grad_norm": 0.40376341342926025, + "learning_rate": 8.926328727695226e-05, + "loss": 0.0296, + "step": 2520 + }, + { + "epoch": 1.723433242506812, + "grad_norm": 0.23646527528762817, + "learning_rate": 8.916069672430319e-05, + "loss": 0.0286, + "step": 2530 + }, + { + "epoch": 1.7302452316076296, + "grad_norm": 0.3980918228626251, + "learning_rate": 8.905767791654884e-05, + "loss": 0.0354, + "step": 2540 + }, + { + "epoch": 1.7370572207084467, + "grad_norm": 0.2409336417913437, + "learning_rate": 8.895423198028638e-05, + "loss": 0.0351, + "step": 2550 + }, + { + "epoch": 1.7438692098092643, + "grad_norm": 0.305263489484787, + "learning_rate": 8.885036004678402e-05, + "loss": 0.0293, + "step": 2560 + }, + { + "epoch": 1.750681198910082, + "grad_norm": 0.3187677264213562, + "learning_rate": 8.874606325196857e-05, + "loss": 0.0285, + "step": 2570 + }, + { + "epoch": 1.757493188010899, + "grad_norm": 0.3625791370868683, + "learning_rate": 8.864134273641304e-05, + "loss": 0.0296, + "step": 2580 + }, + { + "epoch": 1.7643051771117166, + "grad_norm": 0.40569642186164856, + "learning_rate": 8.853619964532427e-05, + "loss": 0.0333, + "step": 2590 + }, + { + "epoch": 1.771117166212534, + "grad_norm": 0.2645326554775238, + "learning_rate": 8.843063512853019e-05, + "loss": 0.0317, + "step": 2600 + }, + { + "epoch": 1.7779291553133514, + "grad_norm": 0.3227790296077728, + "learning_rate": 8.832465034046749e-05, + "loss": 0.0285, + "step": 2610 + }, + { + "epoch": 1.784741144414169, + "grad_norm": 0.4209759831428528, + "learning_rate": 8.821824644016882e-05, + "loss": 0.0329, + "step": 2620 + }, + { + "epoch": 1.7915531335149864, + "grad_norm": 0.4071252942085266, + "learning_rate": 8.811142459125019e-05, + "loss": 0.0299, + "step": 2630 + }, + { + "epoch": 1.7983651226158037, + "grad_norm": 0.3160828948020935, + "learning_rate": 8.800418596189822e-05, + "loss": 0.0295, + "step": 2640 + }, + { + "epoch": 1.8051771117166213, + "grad_norm": 0.4474582374095917, + "learning_rate": 8.789653172485737e-05, + "loss": 0.0377, + "step": 2650 + }, + { + "epoch": 1.8119891008174387, + "grad_norm": 0.39186573028564453, + "learning_rate": 8.778846305741715e-05, + "loss": 0.0353, + "step": 2660 + }, + { + "epoch": 1.818801089918256, + "grad_norm": 0.44805607199668884, + "learning_rate": 8.767998114139918e-05, + "loss": 0.0301, + "step": 2670 + }, + { + "epoch": 1.8256130790190737, + "grad_norm": 0.5066322088241577, + "learning_rate": 8.757108716314429e-05, + "loss": 0.034, + "step": 2680 + }, + { + "epoch": 1.832425068119891, + "grad_norm": 0.5080957412719727, + "learning_rate": 8.746178231349962e-05, + "loss": 0.0305, + "step": 2690 + }, + { + "epoch": 1.8392370572207084, + "grad_norm": 0.42903023958206177, + "learning_rate": 8.735206778780549e-05, + "loss": 0.0341, + "step": 2700 + }, + { + "epoch": 1.846049046321526, + "grad_norm": 0.29407986998558044, + "learning_rate": 8.724194478588234e-05, + "loss": 0.0312, + "step": 2710 + }, + { + "epoch": 1.8528610354223434, + "grad_norm": 0.4133337140083313, + "learning_rate": 8.713141451201772e-05, + "loss": 0.0312, + "step": 2720 + }, + { + "epoch": 1.8596730245231607, + "grad_norm": 0.5048314929008484, + "learning_rate": 8.702047817495295e-05, + "loss": 0.0324, + "step": 2730 + }, + { + "epoch": 1.8664850136239783, + "grad_norm": 0.27112311124801636, + "learning_rate": 8.69091369878701e-05, + "loss": 0.0312, + "step": 2740 + }, + { + "epoch": 1.8732970027247955, + "grad_norm": 0.5361159443855286, + "learning_rate": 8.679739216837849e-05, + "loss": 0.031, + "step": 2750 + }, + { + "epoch": 1.880108991825613, + "grad_norm": 0.3560102880001068, + "learning_rate": 8.66852449385016e-05, + "loss": 0.0313, + "step": 2760 + }, + { + "epoch": 1.8869209809264307, + "grad_norm": 0.2328365445137024, + "learning_rate": 8.657269652466356e-05, + "loss": 0.0295, + "step": 2770 + }, + { + "epoch": 1.8937329700272478, + "grad_norm": 0.32327184081077576, + "learning_rate": 8.645974815767577e-05, + "loss": 0.0289, + "step": 2780 + }, + { + "epoch": 1.9005449591280654, + "grad_norm": 0.3409617841243744, + "learning_rate": 8.634640107272351e-05, + "loss": 0.0294, + "step": 2790 + }, + { + "epoch": 1.9073569482288828, + "grad_norm": 0.27597734332084656, + "learning_rate": 8.623265650935234e-05, + "loss": 0.0287, + "step": 2800 + }, + { + "epoch": 1.9141689373297002, + "grad_norm": 0.2564217150211334, + "learning_rate": 8.611851571145456e-05, + "loss": 0.0297, + "step": 2810 + }, + { + "epoch": 1.9209809264305178, + "grad_norm": 0.3058653771877289, + "learning_rate": 8.600397992725566e-05, + "loss": 0.0243, + "step": 2820 + }, + { + "epoch": 1.9277929155313351, + "grad_norm": 0.33504539728164673, + "learning_rate": 8.588905040930061e-05, + "loss": 0.0312, + "step": 2830 + }, + { + "epoch": 1.9346049046321525, + "grad_norm": 0.2687970995903015, + "learning_rate": 8.577372841444022e-05, + "loss": 0.0293, + "step": 2840 + }, + { + "epoch": 1.94141689373297, + "grad_norm": 0.2944375276565552, + "learning_rate": 8.565801520381736e-05, + "loss": 0.0285, + "step": 2850 + }, + { + "epoch": 1.9482288828337875, + "grad_norm": 0.17202913761138916, + "learning_rate": 8.554191204285313e-05, + "loss": 0.03, + "step": 2860 + }, + { + "epoch": 1.9550408719346049, + "grad_norm": 0.4172252118587494, + "learning_rate": 8.542542020123315e-05, + "loss": 0.0278, + "step": 2870 + }, + { + "epoch": 1.9618528610354224, + "grad_norm": 0.344896137714386, + "learning_rate": 8.530854095289347e-05, + "loss": 0.0293, + "step": 2880 + }, + { + "epoch": 1.9686648501362398, + "grad_norm": 0.42705491185188293, + "learning_rate": 8.519127557600688e-05, + "loss": 0.0303, + "step": 2890 + }, + { + "epoch": 1.9754768392370572, + "grad_norm": 0.2515685260295868, + "learning_rate": 8.507362535296871e-05, + "loss": 0.0283, + "step": 2900 + }, + { + "epoch": 1.9822888283378748, + "grad_norm": 0.3692832291126251, + "learning_rate": 8.495559157038299e-05, + "loss": 0.0389, + "step": 2910 + }, + { + "epoch": 1.989100817438692, + "grad_norm": 0.3757323920726776, + "learning_rate": 8.483717551904823e-05, + "loss": 0.03, + "step": 2920 + }, + { + "epoch": 1.9959128065395095, + "grad_norm": 0.29518547654151917, + "learning_rate": 8.47183784939434e-05, + "loss": 0.0284, + "step": 2930 + }, + { + "epoch": 2.002724795640327, + "grad_norm": 0.39496976137161255, + "learning_rate": 8.459920179421374e-05, + "loss": 0.0324, + "step": 2940 + }, + { + "epoch": 2.0095367847411443, + "grad_norm": 0.22647494077682495, + "learning_rate": 8.447964672315656e-05, + "loss": 0.0249, + "step": 2950 + }, + { + "epoch": 2.016348773841962, + "grad_norm": 0.23620247840881348, + "learning_rate": 8.435971458820692e-05, + "loss": 0.034, + "step": 2960 + }, + { + "epoch": 2.0231607629427795, + "grad_norm": 0.33160877227783203, + "learning_rate": 8.423940670092345e-05, + "loss": 0.0296, + "step": 2970 + }, + { + "epoch": 2.0299727520435966, + "grad_norm": 0.2585655748844147, + "learning_rate": 8.411872437697394e-05, + "loss": 0.0282, + "step": 2980 + }, + { + "epoch": 2.036784741144414, + "grad_norm": 0.5295059680938721, + "learning_rate": 8.399766893612096e-05, + "loss": 0.0258, + "step": 2990 + }, + { + "epoch": 2.043596730245232, + "grad_norm": 0.30863839387893677, + "learning_rate": 8.38762417022074e-05, + "loss": 0.0289, + "step": 3000 + }, + { + "epoch": 2.050408719346049, + "grad_norm": 0.37515100836753845, + "learning_rate": 8.375444400314204e-05, + "loss": 0.0297, + "step": 3010 + }, + { + "epoch": 2.0572207084468666, + "grad_norm": 0.4005380868911743, + "learning_rate": 8.3632277170885e-05, + "loss": 0.0325, + "step": 3020 + }, + { + "epoch": 2.0640326975476837, + "grad_norm": 0.27459025382995605, + "learning_rate": 8.350974254143318e-05, + "loss": 0.0284, + "step": 3030 + }, + { + "epoch": 2.0708446866485013, + "grad_norm": 0.3344320058822632, + "learning_rate": 8.338684145480566e-05, + "loss": 0.0274, + "step": 3040 + }, + { + "epoch": 2.077656675749319, + "grad_norm": 0.5217705965042114, + "learning_rate": 8.326357525502904e-05, + "loss": 0.0309, + "step": 3050 + }, + { + "epoch": 2.084468664850136, + "grad_norm": 0.31241393089294434, + "learning_rate": 8.313994529012273e-05, + "loss": 0.03, + "step": 3060 + }, + { + "epoch": 2.0912806539509536, + "grad_norm": 0.37180668115615845, + "learning_rate": 8.301595291208422e-05, + "loss": 0.0323, + "step": 3070 + }, + { + "epoch": 2.0980926430517712, + "grad_norm": 0.24048398435115814, + "learning_rate": 8.289159947687427e-05, + "loss": 0.029, + "step": 3080 + }, + { + "epoch": 2.1049046321525884, + "grad_norm": 0.3504776656627655, + "learning_rate": 8.276688634440216e-05, + "loss": 0.0328, + "step": 3090 + }, + { + "epoch": 2.111716621253406, + "grad_norm": 0.3533358871936798, + "learning_rate": 8.26418148785107e-05, + "loss": 0.0278, + "step": 3100 + }, + { + "epoch": 2.1185286103542236, + "grad_norm": 0.3171463906764984, + "learning_rate": 8.251638644696141e-05, + "loss": 0.0378, + "step": 3110 + }, + { + "epoch": 2.1253405994550407, + "grad_norm": 0.4020684063434601, + "learning_rate": 8.23906024214195e-05, + "loss": 0.027, + "step": 3120 + }, + { + "epoch": 2.1321525885558583, + "grad_norm": 0.2918350398540497, + "learning_rate": 8.226446417743897e-05, + "loss": 0.0312, + "step": 3130 + }, + { + "epoch": 2.138964577656676, + "grad_norm": 0.4383857846260071, + "learning_rate": 8.213797309444742e-05, + "loss": 0.0306, + "step": 3140 + }, + { + "epoch": 2.145776566757493, + "grad_norm": 0.3243926465511322, + "learning_rate": 8.201113055573105e-05, + "loss": 0.0318, + "step": 3150 + }, + { + "epoch": 2.1525885558583107, + "grad_norm": 0.19268904626369476, + "learning_rate": 8.188393794841958e-05, + "loss": 0.0267, + "step": 3160 + }, + { + "epoch": 2.1594005449591283, + "grad_norm": 0.3372218608856201, + "learning_rate": 8.175639666347094e-05, + "loss": 0.0299, + "step": 3170 + }, + { + "epoch": 2.1662125340599454, + "grad_norm": 0.3734244406223297, + "learning_rate": 8.162850809565623e-05, + "loss": 0.0292, + "step": 3180 + }, + { + "epoch": 2.173024523160763, + "grad_norm": 0.36371320486068726, + "learning_rate": 8.150027364354431e-05, + "loss": 0.0296, + "step": 3190 + }, + { + "epoch": 2.1798365122615806, + "grad_norm": 0.34851449728012085, + "learning_rate": 8.137169470948662e-05, + "loss": 0.0341, + "step": 3200 + }, + { + "epoch": 2.1866485013623977, + "grad_norm": 0.3353874385356903, + "learning_rate": 8.124277269960179e-05, + "loss": 0.0317, + "step": 3210 + }, + { + "epoch": 2.1934604904632153, + "grad_norm": 0.26925817131996155, + "learning_rate": 8.111350902376023e-05, + "loss": 0.027, + "step": 3220 + }, + { + "epoch": 2.2002724795640325, + "grad_norm": 0.24798396229743958, + "learning_rate": 8.098390509556883e-05, + "loss": 0.0282, + "step": 3230 + }, + { + "epoch": 2.20708446866485, + "grad_norm": 0.28002116084098816, + "learning_rate": 8.085396233235536e-05, + "loss": 0.0224, + "step": 3240 + }, + { + "epoch": 2.2138964577656677, + "grad_norm": 0.3231610655784607, + "learning_rate": 8.072368215515306e-05, + "loss": 0.0247, + "step": 3250 + }, + { + "epoch": 2.220708446866485, + "grad_norm": 0.3912052810192108, + "learning_rate": 8.059306598868506e-05, + "loss": 0.0326, + "step": 3260 + }, + { + "epoch": 2.2275204359673024, + "grad_norm": 0.2930348515510559, + "learning_rate": 8.046211526134888e-05, + "loss": 0.0271, + "step": 3270 + }, + { + "epoch": 2.23433242506812, + "grad_norm": 0.3320302963256836, + "learning_rate": 8.033083140520065e-05, + "loss": 0.0262, + "step": 3280 + }, + { + "epoch": 2.241144414168937, + "grad_norm": 0.24441157281398773, + "learning_rate": 8.019921585593962e-05, + "loss": 0.0269, + "step": 3290 + }, + { + "epoch": 2.2479564032697548, + "grad_norm": 0.3559325933456421, + "learning_rate": 8.006727005289232e-05, + "loss": 0.0234, + "step": 3300 + }, + { + "epoch": 2.2547683923705724, + "grad_norm": 0.26384952664375305, + "learning_rate": 7.993499543899692e-05, + "loss": 0.0258, + "step": 3310 + }, + { + "epoch": 2.2615803814713895, + "grad_norm": 0.23424430191516876, + "learning_rate": 7.980239346078742e-05, + "loss": 0.0228, + "step": 3320 + }, + { + "epoch": 2.268392370572207, + "grad_norm": 0.3176807165145874, + "learning_rate": 7.966946556837778e-05, + "loss": 0.0239, + "step": 3330 + }, + { + "epoch": 2.2752043596730247, + "grad_norm": 0.5173885226249695, + "learning_rate": 7.953621321544616e-05, + "loss": 0.0279, + "step": 3340 + }, + { + "epoch": 2.282016348773842, + "grad_norm": 0.31118619441986084, + "learning_rate": 7.940263785921896e-05, + "loss": 0.0294, + "step": 3350 + }, + { + "epoch": 2.2888283378746594, + "grad_norm": 0.40392735600471497, + "learning_rate": 7.926874096045482e-05, + "loss": 0.027, + "step": 3360 + }, + { + "epoch": 2.2956403269754766, + "grad_norm": 0.25696539878845215, + "learning_rate": 7.913452398342881e-05, + "loss": 0.0334, + "step": 3370 + }, + { + "epoch": 2.302452316076294, + "grad_norm": 0.4592527747154236, + "learning_rate": 7.89999883959163e-05, + "loss": 0.0317, + "step": 3380 + }, + { + "epoch": 2.309264305177112, + "grad_norm": 0.3388185501098633, + "learning_rate": 7.886513566917687e-05, + "loss": 0.0271, + "step": 3390 + }, + { + "epoch": 2.316076294277929, + "grad_norm": 0.3678531348705292, + "learning_rate": 7.872996727793838e-05, + "loss": 0.0313, + "step": 3400 + }, + { + "epoch": 2.3228882833787465, + "grad_norm": 0.3035884201526642, + "learning_rate": 7.859448470038069e-05, + "loss": 0.0292, + "step": 3410 + }, + { + "epoch": 2.329700272479564, + "grad_norm": 0.3470432460308075, + "learning_rate": 7.845868941811956e-05, + "loss": 0.0299, + "step": 3420 + }, + { + "epoch": 2.3365122615803813, + "grad_norm": 0.36493241786956787, + "learning_rate": 7.832258291619043e-05, + "loss": 0.0258, + "step": 3430 + }, + { + "epoch": 2.343324250681199, + "grad_norm": 0.32892706990242004, + "learning_rate": 7.81861666830322e-05, + "loss": 0.024, + "step": 3440 + }, + { + "epoch": 2.3501362397820165, + "grad_norm": 0.2843776047229767, + "learning_rate": 7.804944221047097e-05, + "loss": 0.0303, + "step": 3450 + }, + { + "epoch": 2.3569482288828336, + "grad_norm": 0.3543314039707184, + "learning_rate": 7.791241099370364e-05, + "loss": 0.0302, + "step": 3460 + }, + { + "epoch": 2.363760217983651, + "grad_norm": 0.3289734423160553, + "learning_rate": 7.777507453128163e-05, + "loss": 0.0269, + "step": 3470 + }, + { + "epoch": 2.370572207084469, + "grad_norm": 0.48808780312538147, + "learning_rate": 7.763743432509451e-05, + "loss": 0.0258, + "step": 3480 + }, + { + "epoch": 2.377384196185286, + "grad_norm": 0.3037632703781128, + "learning_rate": 7.749949188035353e-05, + "loss": 0.0298, + "step": 3490 + }, + { + "epoch": 2.3841961852861036, + "grad_norm": 0.4912019968032837, + "learning_rate": 7.736124870557516e-05, + "loss": 0.0338, + "step": 3500 + }, + { + "epoch": 2.391008174386921, + "grad_norm": 0.4113086462020874, + "learning_rate": 7.722270631256459e-05, + "loss": 0.0272, + "step": 3510 + }, + { + "epoch": 2.3978201634877383, + "grad_norm": 0.5073822736740112, + "learning_rate": 7.708386621639925e-05, + "loss": 0.0275, + "step": 3520 + }, + { + "epoch": 2.404632152588556, + "grad_norm": 0.28544387221336365, + "learning_rate": 7.694472993541219e-05, + "loss": 0.0253, + "step": 3530 + }, + { + "epoch": 2.4114441416893735, + "grad_norm": 0.20866520702838898, + "learning_rate": 7.680529899117547e-05, + "loss": 0.0305, + "step": 3540 + }, + { + "epoch": 2.4182561307901906, + "grad_norm": 0.28740522265434265, + "learning_rate": 7.666557490848358e-05, + "loss": 0.0307, + "step": 3550 + }, + { + "epoch": 2.4250681198910082, + "grad_norm": 0.30243638157844543, + "learning_rate": 7.65255592153367e-05, + "loss": 0.026, + "step": 3560 + }, + { + "epoch": 2.431880108991826, + "grad_norm": 0.3321373164653778, + "learning_rate": 7.638525344292402e-05, + "loss": 0.0318, + "step": 3570 + }, + { + "epoch": 2.438692098092643, + "grad_norm": 0.24234934151172638, + "learning_rate": 7.624465912560697e-05, + "loss": 0.0322, + "step": 3580 + }, + { + "epoch": 2.4455040871934606, + "grad_norm": 0.2390761822462082, + "learning_rate": 7.610377780090249e-05, + "loss": 0.0247, + "step": 3590 + }, + { + "epoch": 2.452316076294278, + "grad_norm": 0.2481442093849182, + "learning_rate": 7.596261100946618e-05, + "loss": 0.0298, + "step": 3600 + }, + { + "epoch": 2.4591280653950953, + "grad_norm": 0.2922890782356262, + "learning_rate": 7.582116029507542e-05, + "loss": 0.0259, + "step": 3610 + }, + { + "epoch": 2.465940054495913, + "grad_norm": 0.2962955832481384, + "learning_rate": 7.56794272046126e-05, + "loss": 0.0257, + "step": 3620 + }, + { + "epoch": 2.47275204359673, + "grad_norm": 0.4098016619682312, + "learning_rate": 7.55374132880481e-05, + "loss": 0.0291, + "step": 3630 + }, + { + "epoch": 2.4795640326975477, + "grad_norm": 0.38562092185020447, + "learning_rate": 7.539512009842333e-05, + "loss": 0.0257, + "step": 3640 + }, + { + "epoch": 2.4863760217983653, + "grad_norm": 0.29110950231552124, + "learning_rate": 7.525254919183382e-05, + "loss": 0.0281, + "step": 3650 + }, + { + "epoch": 2.4931880108991824, + "grad_norm": 0.24400657415390015, + "learning_rate": 7.510970212741215e-05, + "loss": 0.0216, + "step": 3660 + }, + { + "epoch": 2.5, + "grad_norm": 0.280714213848114, + "learning_rate": 7.496658046731096e-05, + "loss": 0.0258, + "step": 3670 + }, + { + "epoch": 2.5068119891008176, + "grad_norm": 0.2414192110300064, + "learning_rate": 7.482318577668578e-05, + "loss": 0.0299, + "step": 3680 + }, + { + "epoch": 2.5136239782016347, + "grad_norm": 0.27257198095321655, + "learning_rate": 7.467951962367796e-05, + "loss": 0.0302, + "step": 3690 + }, + { + "epoch": 2.5204359673024523, + "grad_norm": 0.3752216696739197, + "learning_rate": 7.453558357939755e-05, + "loss": 0.0236, + "step": 3700 + }, + { + "epoch": 2.5272479564032695, + "grad_norm": 0.31143346428871155, + "learning_rate": 7.439137921790606e-05, + "loss": 0.0346, + "step": 3710 + }, + { + "epoch": 2.534059945504087, + "grad_norm": 0.3188665211200714, + "learning_rate": 7.42469081161993e-05, + "loss": 0.0278, + "step": 3720 + }, + { + "epoch": 2.5408719346049047, + "grad_norm": 0.2696925103664398, + "learning_rate": 7.410217185419006e-05, + "loss": 0.0235, + "step": 3730 + }, + { + "epoch": 2.547683923705722, + "grad_norm": 0.2630939483642578, + "learning_rate": 7.395717201469095e-05, + "loss": 0.0307, + "step": 3740 + }, + { + "epoch": 2.5544959128065394, + "grad_norm": 0.3755340874195099, + "learning_rate": 7.381191018339696e-05, + "loss": 0.026, + "step": 3750 + }, + { + "epoch": 2.561307901907357, + "grad_norm": 0.26605546474456787, + "learning_rate": 7.36663879488682e-05, + "loss": 0.0284, + "step": 3760 + }, + { + "epoch": 2.568119891008174, + "grad_norm": 0.3214227557182312, + "learning_rate": 7.352060690251254e-05, + "loss": 0.0316, + "step": 3770 + }, + { + "epoch": 2.5749318801089918, + "grad_norm": 0.2961984872817993, + "learning_rate": 7.337456863856811e-05, + "loss": 0.0248, + "step": 3780 + }, + { + "epoch": 2.5817438692098094, + "grad_norm": 0.26064276695251465, + "learning_rate": 7.3228274754086e-05, + "loss": 0.0219, + "step": 3790 + }, + { + "epoch": 2.5885558583106265, + "grad_norm": 0.2844791114330292, + "learning_rate": 7.308172684891267e-05, + "loss": 0.0254, + "step": 3800 + }, + { + "epoch": 2.595367847411444, + "grad_norm": 0.22159306704998016, + "learning_rate": 7.293492652567255e-05, + "loss": 0.0239, + "step": 3810 + }, + { + "epoch": 2.6021798365122617, + "grad_norm": 0.21708112955093384, + "learning_rate": 7.278787538975043e-05, + "loss": 0.0281, + "step": 3820 + }, + { + "epoch": 2.608991825613079, + "grad_norm": 0.2955373525619507, + "learning_rate": 7.2640575049274e-05, + "loss": 0.0283, + "step": 3830 + }, + { + "epoch": 2.6158038147138964, + "grad_norm": 0.36569881439208984, + "learning_rate": 7.249302711509616e-05, + "loss": 0.0225, + "step": 3840 + }, + { + "epoch": 2.622615803814714, + "grad_norm": 0.24029208719730377, + "learning_rate": 7.23452332007775e-05, + "loss": 0.0253, + "step": 3850 + }, + { + "epoch": 2.629427792915531, + "grad_norm": 0.3086705803871155, + "learning_rate": 7.219719492256858e-05, + "loss": 0.0281, + "step": 3860 + }, + { + "epoch": 2.636239782016349, + "grad_norm": 0.22832535207271576, + "learning_rate": 7.20489138993923e-05, + "loss": 0.0258, + "step": 3870 + }, + { + "epoch": 2.6430517711171664, + "grad_norm": 0.24865151941776276, + "learning_rate": 7.190039175282614e-05, + "loss": 0.0263, + "step": 3880 + }, + { + "epoch": 2.6498637602179835, + "grad_norm": 0.43784525990486145, + "learning_rate": 7.175163010708455e-05, + "loss": 0.0229, + "step": 3890 + }, + { + "epoch": 2.656675749318801, + "grad_norm": 0.26744481921195984, + "learning_rate": 7.1602630589001e-05, + "loss": 0.0263, + "step": 3900 + }, + { + "epoch": 2.6634877384196187, + "grad_norm": 0.22483383119106293, + "learning_rate": 7.14533948280104e-05, + "loss": 0.0267, + "step": 3910 + }, + { + "epoch": 2.670299727520436, + "grad_norm": 0.3076573312282562, + "learning_rate": 7.130392445613109e-05, + "loss": 0.0263, + "step": 3920 + }, + { + "epoch": 2.6771117166212535, + "grad_norm": 0.20539738237857819, + "learning_rate": 7.115422110794711e-05, + "loss": 0.0259, + "step": 3930 + }, + { + "epoch": 2.683923705722071, + "grad_norm": 0.33314427733421326, + "learning_rate": 7.100428642059033e-05, + "loss": 0.0286, + "step": 3940 + }, + { + "epoch": 2.690735694822888, + "grad_norm": 0.375288188457489, + "learning_rate": 7.08541220337224e-05, + "loss": 0.0283, + "step": 3950 + }, + { + "epoch": 2.697547683923706, + "grad_norm": 0.27256014943122864, + "learning_rate": 7.070372958951706e-05, + "loss": 0.0262, + "step": 3960 + }, + { + "epoch": 2.7043596730245234, + "grad_norm": 0.3550771176815033, + "learning_rate": 7.055311073264194e-05, + "loss": 0.0277, + "step": 3970 + }, + { + "epoch": 2.7111716621253406, + "grad_norm": 0.21956783533096313, + "learning_rate": 7.040226711024077e-05, + "loss": 0.0236, + "step": 3980 + }, + { + "epoch": 2.717983651226158, + "grad_norm": 0.26336756348609924, + "learning_rate": 7.02512003719152e-05, + "loss": 0.0223, + "step": 3990 + }, + { + "epoch": 2.7247956403269757, + "grad_norm": 0.25999748706817627, + "learning_rate": 7.00999121697069e-05, + "loss": 0.0272, + "step": 4000 + }, + { + "epoch": 2.731607629427793, + "grad_norm": 0.29413673281669617, + "learning_rate": 6.99484041580794e-05, + "loss": 0.0254, + "step": 4010 + }, + { + "epoch": 2.7384196185286105, + "grad_norm": 0.31084537506103516, + "learning_rate": 6.979667799390004e-05, + "loss": 0.0247, + "step": 4020 + }, + { + "epoch": 2.7452316076294276, + "grad_norm": 0.3437006175518036, + "learning_rate": 6.964473533642185e-05, + "loss": 0.0259, + "step": 4030 + }, + { + "epoch": 2.7520435967302452, + "grad_norm": 0.35230040550231934, + "learning_rate": 6.949257784726539e-05, + "loss": 0.0293, + "step": 4040 + }, + { + "epoch": 2.758855585831063, + "grad_norm": 0.23275138437747955, + "learning_rate": 6.934020719040056e-05, + "loss": 0.0241, + "step": 4050 + }, + { + "epoch": 2.76566757493188, + "grad_norm": 0.2156933695077896, + "learning_rate": 6.918762503212848e-05, + "loss": 0.0286, + "step": 4060 + }, + { + "epoch": 2.7724795640326976, + "grad_norm": 0.2477605640888214, + "learning_rate": 6.903483304106319e-05, + "loss": 0.0265, + "step": 4070 + }, + { + "epoch": 2.7792915531335147, + "grad_norm": 0.3127809166908264, + "learning_rate": 6.888183288811341e-05, + "loss": 0.0221, + "step": 4080 + }, + { + "epoch": 2.7861035422343323, + "grad_norm": 0.33863481879234314, + "learning_rate": 6.87286262464643e-05, + "loss": 0.0233, + "step": 4090 + }, + { + "epoch": 2.79291553133515, + "grad_norm": 0.19353847205638885, + "learning_rate": 6.857521479155915e-05, + "loss": 0.0271, + "step": 4100 + }, + { + "epoch": 2.799727520435967, + "grad_norm": 0.2616625726222992, + "learning_rate": 6.842160020108104e-05, + "loss": 0.0249, + "step": 4110 + }, + { + "epoch": 2.8065395095367847, + "grad_norm": 0.2875172793865204, + "learning_rate": 6.826778415493455e-05, + "loss": 0.0234, + "step": 4120 + }, + { + "epoch": 2.8133514986376023, + "grad_norm": 0.39224088191986084, + "learning_rate": 6.811376833522729e-05, + "loss": 0.0262, + "step": 4130 + }, + { + "epoch": 2.8201634877384194, + "grad_norm": 0.2205829918384552, + "learning_rate": 6.795955442625159e-05, + "loss": 0.0252, + "step": 4140 + }, + { + "epoch": 2.826975476839237, + "grad_norm": 0.30891153216362, + "learning_rate": 6.780514411446608e-05, + "loss": 0.0246, + "step": 4150 + }, + { + "epoch": 2.8337874659400546, + "grad_norm": 0.22732511162757874, + "learning_rate": 6.765053908847716e-05, + "loss": 0.0193, + "step": 4160 + }, + { + "epoch": 2.8405994550408717, + "grad_norm": 0.32131826877593994, + "learning_rate": 6.749574103902064e-05, + "loss": 0.0272, + "step": 4170 + }, + { + "epoch": 2.8474114441416893, + "grad_norm": 0.26638948917388916, + "learning_rate": 6.734075165894317e-05, + "loss": 0.0305, + "step": 4180 + }, + { + "epoch": 2.854223433242507, + "grad_norm": 0.31483474373817444, + "learning_rate": 6.71855726431838e-05, + "loss": 0.0295, + "step": 4190 + }, + { + "epoch": 2.861035422343324, + "grad_norm": 0.36278581619262695, + "learning_rate": 6.703020568875538e-05, + "loss": 0.0216, + "step": 4200 + }, + { + "epoch": 2.8678474114441417, + "grad_norm": 0.38526198267936707, + "learning_rate": 6.687465249472603e-05, + "loss": 0.023, + "step": 4210 + }, + { + "epoch": 2.8746594005449593, + "grad_norm": 0.3007153272628784, + "learning_rate": 6.671891476220055e-05, + "loss": 0.0243, + "step": 4220 + }, + { + "epoch": 2.8814713896457764, + "grad_norm": 0.2640557885169983, + "learning_rate": 6.656299419430183e-05, + "loss": 0.0255, + "step": 4230 + }, + { + "epoch": 2.888283378746594, + "grad_norm": 0.256852388381958, + "learning_rate": 6.640689249615223e-05, + "loss": 0.0235, + "step": 4240 + }, + { + "epoch": 2.8950953678474116, + "grad_norm": 0.2589550018310547, + "learning_rate": 6.625061137485491e-05, + "loss": 0.0248, + "step": 4250 + }, + { + "epoch": 2.9019073569482288, + "grad_norm": 0.21780376136302948, + "learning_rate": 6.609415253947517e-05, + "loss": 0.0274, + "step": 4260 + }, + { + "epoch": 2.9087193460490464, + "grad_norm": 0.19023562967777252, + "learning_rate": 6.593751770102178e-05, + "loss": 0.023, + "step": 4270 + }, + { + "epoch": 2.915531335149864, + "grad_norm": 0.27956318855285645, + "learning_rate": 6.578070857242823e-05, + "loss": 0.0237, + "step": 4280 + }, + { + "epoch": 2.922343324250681, + "grad_norm": 0.31173229217529297, + "learning_rate": 6.562372686853402e-05, + "loss": 0.0229, + "step": 4290 + }, + { + "epoch": 2.9291553133514987, + "grad_norm": 0.5626435875892639, + "learning_rate": 6.546657430606593e-05, + "loss": 0.0261, + "step": 4300 + }, + { + "epoch": 2.9359673024523163, + "grad_norm": 0.330353707075119, + "learning_rate": 6.530925260361918e-05, + "loss": 0.0241, + "step": 4310 + }, + { + "epoch": 2.9427792915531334, + "grad_norm": 0.30065762996673584, + "learning_rate": 6.515176348163871e-05, + "loss": 0.0261, + "step": 4320 + }, + { + "epoch": 2.949591280653951, + "grad_norm": 0.23945222795009613, + "learning_rate": 6.499410866240032e-05, + "loss": 0.0203, + "step": 4330 + }, + { + "epoch": 2.9564032697547686, + "grad_norm": 0.25979286432266235, + "learning_rate": 6.48362898699919e-05, + "loss": 0.0219, + "step": 4340 + }, + { + "epoch": 2.963215258855586, + "grad_norm": 0.2935354709625244, + "learning_rate": 6.467830883029443e-05, + "loss": 0.023, + "step": 4350 + }, + { + "epoch": 2.9700272479564034, + "grad_norm": 0.24333009123802185, + "learning_rate": 6.452016727096326e-05, + "loss": 0.022, + "step": 4360 + }, + { + "epoch": 2.976839237057221, + "grad_norm": 0.1909112185239792, + "learning_rate": 6.436186692140916e-05, + "loss": 0.0225, + "step": 4370 + }, + { + "epoch": 2.983651226158038, + "grad_norm": 0.24930566549301147, + "learning_rate": 6.420340951277938e-05, + "loss": 0.023, + "step": 4380 + }, + { + "epoch": 2.9904632152588557, + "grad_norm": 0.1641029566526413, + "learning_rate": 6.404479677793874e-05, + "loss": 0.0263, + "step": 4390 + }, + { + "epoch": 2.997275204359673, + "grad_norm": 0.3587425649166107, + "learning_rate": 6.388603045145075e-05, + "loss": 0.028, + "step": 4400 + }, + { + "epoch": 3.0040871934604905, + "grad_norm": 0.32594290375709534, + "learning_rate": 6.372711226955843e-05, + "loss": 0.0273, + "step": 4410 + }, + { + "epoch": 3.010899182561308, + "grad_norm": 0.2173074632883072, + "learning_rate": 6.356804397016564e-05, + "loss": 0.0242, + "step": 4420 + }, + { + "epoch": 3.017711171662125, + "grad_norm": 0.2268197238445282, + "learning_rate": 6.340882729281779e-05, + "loss": 0.0196, + "step": 4430 + }, + { + "epoch": 3.024523160762943, + "grad_norm": 0.2758912742137909, + "learning_rate": 6.324946397868294e-05, + "loss": 0.0248, + "step": 4440 + }, + { + "epoch": 3.0313351498637604, + "grad_norm": 0.24817928671836853, + "learning_rate": 6.308995577053276e-05, + "loss": 0.0242, + "step": 4450 + }, + { + "epoch": 3.0381471389645776, + "grad_norm": 0.24741585552692413, + "learning_rate": 6.293030441272347e-05, + "loss": 0.024, + "step": 4460 + }, + { + "epoch": 3.044959128065395, + "grad_norm": 0.23530451953411102, + "learning_rate": 6.277051165117677e-05, + "loss": 0.0233, + "step": 4470 + }, + { + "epoch": 3.0517711171662127, + "grad_norm": 0.24983182549476624, + "learning_rate": 6.261057923336064e-05, + "loss": 0.0227, + "step": 4480 + }, + { + "epoch": 3.05858310626703, + "grad_norm": 0.24539630115032196, + "learning_rate": 6.245050890827042e-05, + "loss": 0.0244, + "step": 4490 + }, + { + "epoch": 3.0653950953678475, + "grad_norm": 0.2987506091594696, + "learning_rate": 6.229030242640952e-05, + "loss": 0.0221, + "step": 4500 + }, + { + "epoch": 3.0722070844686646, + "grad_norm": 0.31599339842796326, + "learning_rate": 6.212996153977037e-05, + "loss": 0.0257, + "step": 4510 + }, + { + "epoch": 3.0790190735694822, + "grad_norm": 0.20547758042812347, + "learning_rate": 6.196948800181523e-05, + "loss": 0.0251, + "step": 4520 + }, + { + "epoch": 3.0858310626703, + "grad_norm": 0.22104264795780182, + "learning_rate": 6.180888356745695e-05, + "loss": 0.0239, + "step": 4530 + }, + { + "epoch": 3.092643051771117, + "grad_norm": 0.3076710104942322, + "learning_rate": 6.164814999303995e-05, + "loss": 0.0214, + "step": 4540 + }, + { + "epoch": 3.0994550408719346, + "grad_norm": 0.2437480241060257, + "learning_rate": 6.148728903632081e-05, + "loss": 0.0249, + "step": 4550 + }, + { + "epoch": 3.106267029972752, + "grad_norm": 0.30195826292037964, + "learning_rate": 6.132630245644921e-05, + "loss": 0.021, + "step": 4560 + }, + { + "epoch": 3.1130790190735693, + "grad_norm": 0.35200658440589905, + "learning_rate": 6.116519201394857e-05, + "loss": 0.0213, + "step": 4570 + }, + { + "epoch": 3.119891008174387, + "grad_norm": 0.23754388093948364, + "learning_rate": 6.10039594706969e-05, + "loss": 0.0226, + "step": 4580 + }, + { + "epoch": 3.1267029972752045, + "grad_norm": 0.24620459973812103, + "learning_rate": 6.084260658990744e-05, + "loss": 0.024, + "step": 4590 + }, + { + "epoch": 3.1335149863760217, + "grad_norm": 0.2630073130130768, + "learning_rate": 6.068113513610943e-05, + "loss": 0.0218, + "step": 4600 + }, + { + "epoch": 3.1403269754768393, + "grad_norm": 0.1972293108701706, + "learning_rate": 6.0519546875128876e-05, + "loss": 0.0247, + "step": 4610 + }, + { + "epoch": 3.147138964577657, + "grad_norm": 0.26045480370521545, + "learning_rate": 6.035784357406906e-05, + "loss": 0.0263, + "step": 4620 + }, + { + "epoch": 3.153950953678474, + "grad_norm": 0.25935009121894836, + "learning_rate": 6.01960270012914e-05, + "loss": 0.0209, + "step": 4630 + }, + { + "epoch": 3.1607629427792916, + "grad_norm": 0.3175886571407318, + "learning_rate": 6.003409892639599e-05, + "loss": 0.0211, + "step": 4640 + }, + { + "epoch": 3.167574931880109, + "grad_norm": 0.2324659526348114, + "learning_rate": 5.9872061120202336e-05, + "loss": 0.0212, + "step": 4650 + }, + { + "epoch": 3.1743869209809263, + "grad_norm": 0.22307582199573517, + "learning_rate": 5.9709915354729914e-05, + "loss": 0.0265, + "step": 4660 + }, + { + "epoch": 3.181198910081744, + "grad_norm": 0.30928510427474976, + "learning_rate": 5.9547663403178824e-05, + "loss": 0.0227, + "step": 4670 + }, + { + "epoch": 3.1880108991825615, + "grad_norm": 0.2945087254047394, + "learning_rate": 5.9385307039910445e-05, + "loss": 0.0184, + "step": 4680 + }, + { + "epoch": 3.1948228882833787, + "grad_norm": 0.14484548568725586, + "learning_rate": 5.922284804042792e-05, + "loss": 0.0203, + "step": 4690 + }, + { + "epoch": 3.2016348773841963, + "grad_norm": 0.32522672414779663, + "learning_rate": 5.906028818135687e-05, + "loss": 0.0276, + "step": 4700 + }, + { + "epoch": 3.2084468664850134, + "grad_norm": 0.23650048673152924, + "learning_rate": 5.889762924042585e-05, + "loss": 0.0244, + "step": 4710 + }, + { + "epoch": 3.215258855585831, + "grad_norm": 0.2292642891407013, + "learning_rate": 5.873487299644699e-05, + "loss": 0.0237, + "step": 4720 + }, + { + "epoch": 3.2220708446866486, + "grad_norm": 0.25956031680107117, + "learning_rate": 5.857202122929649e-05, + "loss": 0.0201, + "step": 4730 + }, + { + "epoch": 3.2288828337874658, + "grad_norm": 0.3084121346473694, + "learning_rate": 5.840907571989518e-05, + "loss": 0.0196, + "step": 4740 + }, + { + "epoch": 3.2356948228882834, + "grad_norm": 0.43665236234664917, + "learning_rate": 5.824603825018904e-05, + "loss": 0.0227, + "step": 4750 + }, + { + "epoch": 3.242506811989101, + "grad_norm": 0.2349618375301361, + "learning_rate": 5.808291060312975e-05, + "loss": 0.0255, + "step": 4760 + }, + { + "epoch": 3.249318801089918, + "grad_norm": 0.47515255212783813, + "learning_rate": 5.7919694562655083e-05, + "loss": 0.023, + "step": 4770 + }, + { + "epoch": 3.2561307901907357, + "grad_norm": 0.236441969871521, + "learning_rate": 5.775639191366954e-05, + "loss": 0.0233, + "step": 4780 + }, + { + "epoch": 3.2629427792915533, + "grad_norm": 0.4283769130706787, + "learning_rate": 5.75930044420247e-05, + "loss": 0.0213, + "step": 4790 + }, + { + "epoch": 3.2697547683923704, + "grad_norm": 0.24188104271888733, + "learning_rate": 5.74295339344998e-05, + "loss": 0.0231, + "step": 4800 + }, + { + "epoch": 3.276566757493188, + "grad_norm": 0.29469841718673706, + "learning_rate": 5.726598217878211e-05, + "loss": 0.0198, + "step": 4810 + }, + { + "epoch": 3.2833787465940056, + "grad_norm": 0.3897201120853424, + "learning_rate": 5.71023509634474e-05, + "loss": 0.0201, + "step": 4820 + }, + { + "epoch": 3.290190735694823, + "grad_norm": 0.2625117003917694, + "learning_rate": 5.693864207794049e-05, + "loss": 0.0223, + "step": 4830 + }, + { + "epoch": 3.2970027247956404, + "grad_norm": 0.15055757761001587, + "learning_rate": 5.677485731255545e-05, + "loss": 0.0306, + "step": 4840 + }, + { + "epoch": 3.3038147138964575, + "grad_norm": 0.37314262986183167, + "learning_rate": 5.6610998458416296e-05, + "loss": 0.0212, + "step": 4850 + }, + { + "epoch": 3.310626702997275, + "grad_norm": 0.2949288785457611, + "learning_rate": 5.644706730745716e-05, + "loss": 0.0218, + "step": 4860 + }, + { + "epoch": 3.3174386920980927, + "grad_norm": 0.19543494284152985, + "learning_rate": 5.628306565240287e-05, + "loss": 0.0203, + "step": 4870 + }, + { + "epoch": 3.32425068119891, + "grad_norm": 0.18216294050216675, + "learning_rate": 5.611899528674923e-05, + "loss": 0.0225, + "step": 4880 + }, + { + "epoch": 3.3310626702997275, + "grad_norm": 0.29446980357170105, + "learning_rate": 5.595485800474349e-05, + "loss": 0.0213, + "step": 4890 + }, + { + "epoch": 3.337874659400545, + "grad_norm": 0.28127121925354004, + "learning_rate": 5.579065560136467e-05, + "loss": 0.0206, + "step": 4900 + }, + { + "epoch": 3.344686648501362, + "grad_norm": 0.15396562218666077, + "learning_rate": 5.562638987230392e-05, + "loss": 0.0202, + "step": 4910 + }, + { + "epoch": 3.35149863760218, + "grad_norm": 0.3175133168697357, + "learning_rate": 5.546206261394498e-05, + "loss": 0.0241, + "step": 4920 + }, + { + "epoch": 3.3583106267029974, + "grad_norm": 0.2692420184612274, + "learning_rate": 5.529767562334437e-05, + "loss": 0.0222, + "step": 4930 + }, + { + "epoch": 3.3651226158038146, + "grad_norm": 0.3113185167312622, + "learning_rate": 5.5133230698211926e-05, + "loss": 0.0191, + "step": 4940 + }, + { + "epoch": 3.371934604904632, + "grad_norm": 0.2561298608779907, + "learning_rate": 5.496872963689096e-05, + "loss": 0.0249, + "step": 4950 + }, + { + "epoch": 3.3787465940054497, + "grad_norm": 0.24195902049541473, + "learning_rate": 5.4804174238338756e-05, + "loss": 0.017, + "step": 4960 + }, + { + "epoch": 3.385558583106267, + "grad_norm": 0.36146679520606995, + "learning_rate": 5.463956630210678e-05, + "loss": 0.0239, + "step": 4970 + }, + { + "epoch": 3.3923705722070845, + "grad_norm": 0.21447274088859558, + "learning_rate": 5.4474907628321046e-05, + "loss": 0.027, + "step": 4980 + }, + { + "epoch": 3.399182561307902, + "grad_norm": 0.4351305067539215, + "learning_rate": 5.431020001766244e-05, + "loss": 0.0215, + "step": 4990 + }, + { + "epoch": 3.4059945504087192, + "grad_norm": 0.24434415996074677, + "learning_rate": 5.4145445271346986e-05, + "loss": 0.021, + "step": 5000 + }, + { + "epoch": 3.412806539509537, + "grad_norm": 0.299847811460495, + "learning_rate": 5.398064519110622e-05, + "loss": 0.0249, + "step": 5010 + }, + { + "epoch": 3.4196185286103544, + "grad_norm": 0.2811359763145447, + "learning_rate": 5.3815801579167394e-05, + "loss": 0.0199, + "step": 5020 + }, + { + "epoch": 3.4264305177111716, + "grad_norm": 0.22033247351646423, + "learning_rate": 5.365091623823382e-05, + "loss": 0.0217, + "step": 5030 + }, + { + "epoch": 3.433242506811989, + "grad_norm": 0.23859739303588867, + "learning_rate": 5.348599097146521e-05, + "loss": 0.0193, + "step": 5040 + }, + { + "epoch": 3.4400544959128068, + "grad_norm": 0.29264622926712036, + "learning_rate": 5.3321027582457836e-05, + "loss": 0.0226, + "step": 5050 + }, + { + "epoch": 3.446866485013624, + "grad_norm": 0.22165198624134064, + "learning_rate": 5.315602787522491e-05, + "loss": 0.0219, + "step": 5060 + }, + { + "epoch": 3.4536784741144415, + "grad_norm": 0.1980362981557846, + "learning_rate": 5.299099365417678e-05, + "loss": 0.0175, + "step": 5070 + }, + { + "epoch": 3.460490463215259, + "grad_norm": 0.29981181025505066, + "learning_rate": 5.2825926724101236e-05, + "loss": 0.0227, + "step": 5080 + }, + { + "epoch": 3.4673024523160763, + "grad_norm": 0.2726703882217407, + "learning_rate": 5.26608288901438e-05, + "loss": 0.025, + "step": 5090 + }, + { + "epoch": 3.474114441416894, + "grad_norm": 0.354449063539505, + "learning_rate": 5.24957019577879e-05, + "loss": 0.0243, + "step": 5100 + }, + { + "epoch": 3.480926430517711, + "grad_norm": 0.28788021206855774, + "learning_rate": 5.2330547732835266e-05, + "loss": 0.0188, + "step": 5110 + }, + { + "epoch": 3.4877384196185286, + "grad_norm": 0.17985177040100098, + "learning_rate": 5.2165368021385996e-05, + "loss": 0.0264, + "step": 5120 + }, + { + "epoch": 3.494550408719346, + "grad_norm": 0.23571337759494781, + "learning_rate": 5.200016462981897e-05, + "loss": 0.0172, + "step": 5130 + }, + { + "epoch": 3.5013623978201633, + "grad_norm": 0.20366789400577545, + "learning_rate": 5.1834939364772015e-05, + "loss": 0.0185, + "step": 5140 + }, + { + "epoch": 3.508174386920981, + "grad_norm": 0.36313048005104065, + "learning_rate": 5.166969403312214e-05, + "loss": 0.0198, + "step": 5150 + }, + { + "epoch": 3.5149863760217985, + "grad_norm": 0.26055416464805603, + "learning_rate": 5.1504430441965844e-05, + "loss": 0.0207, + "step": 5160 + }, + { + "epoch": 3.5217983651226157, + "grad_norm": 0.17273050546646118, + "learning_rate": 5.133915039859923e-05, + "loss": 0.0167, + "step": 5170 + }, + { + "epoch": 3.5286103542234333, + "grad_norm": 0.30914077162742615, + "learning_rate": 5.1173855710498444e-05, + "loss": 0.02, + "step": 5180 + }, + { + "epoch": 3.5354223433242504, + "grad_norm": 0.29828980565071106, + "learning_rate": 5.100854818529967e-05, + "loss": 0.0251, + "step": 5190 + }, + { + "epoch": 3.542234332425068, + "grad_norm": 0.3353451192378998, + "learning_rate": 5.084322963077951e-05, + "loss": 0.0252, + "step": 5200 + }, + { + "epoch": 3.5490463215258856, + "grad_norm": 0.29260993003845215, + "learning_rate": 5.067790185483522e-05, + "loss": 0.0236, + "step": 5210 + }, + { + "epoch": 3.5558583106267028, + "grad_norm": 0.3159320056438446, + "learning_rate": 5.0512566665464844e-05, + "loss": 0.0225, + "step": 5220 + }, + { + "epoch": 3.5626702997275204, + "grad_norm": 0.21729598939418793, + "learning_rate": 5.034722587074755e-05, + "loss": 0.0204, + "step": 5230 + }, + { + "epoch": 3.569482288828338, + "grad_norm": 0.22360987961292267, + "learning_rate": 5.018188127882375e-05, + "loss": 0.0189, + "step": 5240 + }, + { + "epoch": 3.576294277929155, + "grad_norm": 0.28683122992515564, + "learning_rate": 5.0016534697875417e-05, + "loss": 0.0157, + "step": 5250 + }, + { + "epoch": 3.5831062670299727, + "grad_norm": 0.2143315225839615, + "learning_rate": 4.9851187936106294e-05, + "loss": 0.02, + "step": 5260 + }, + { + "epoch": 3.5899182561307903, + "grad_norm": 0.24616491794586182, + "learning_rate": 4.968584280172206e-05, + "loss": 0.0211, + "step": 5270 + }, + { + "epoch": 3.5967302452316074, + "grad_norm": 0.4334745705127716, + "learning_rate": 4.95205011029106e-05, + "loss": 0.022, + "step": 5280 + }, + { + "epoch": 3.603542234332425, + "grad_norm": 0.33467578887939453, + "learning_rate": 4.935516464782227e-05, + "loss": 0.0236, + "step": 5290 + }, + { + "epoch": 3.6103542234332426, + "grad_norm": 0.24656948447227478, + "learning_rate": 4.918983524455003e-05, + "loss": 0.021, + "step": 5300 + }, + { + "epoch": 3.61716621253406, + "grad_norm": 0.18690423667430878, + "learning_rate": 4.9024514701109766e-05, + "loss": 0.0207, + "step": 5310 + }, + { + "epoch": 3.6239782016348774, + "grad_norm": 0.20676788687705994, + "learning_rate": 4.885920482542043e-05, + "loss": 0.0186, + "step": 5320 + }, + { + "epoch": 3.630790190735695, + "grad_norm": 0.24274636805057526, + "learning_rate": 4.869390742528438e-05, + "loss": 0.0191, + "step": 5330 + }, + { + "epoch": 3.637602179836512, + "grad_norm": 0.1921306997537613, + "learning_rate": 4.852862430836744e-05, + "loss": 0.0209, + "step": 5340 + }, + { + "epoch": 3.6444141689373297, + "grad_norm": 0.19985947012901306, + "learning_rate": 4.836335728217933e-05, + "loss": 0.0202, + "step": 5350 + }, + { + "epoch": 3.6512261580381473, + "grad_norm": 0.18188785016536713, + "learning_rate": 4.819810815405379e-05, + "loss": 0.0213, + "step": 5360 + }, + { + "epoch": 3.6580381471389645, + "grad_norm": 0.20466038584709167, + "learning_rate": 4.803287873112877e-05, + "loss": 0.0222, + "step": 5370 + }, + { + "epoch": 3.664850136239782, + "grad_norm": 0.17942418158054352, + "learning_rate": 4.786767082032681e-05, + "loss": 0.0204, + "step": 5380 + }, + { + "epoch": 3.6716621253405997, + "grad_norm": 0.2102428525686264, + "learning_rate": 4.77024862283351e-05, + "loss": 0.0183, + "step": 5390 + }, + { + "epoch": 3.678474114441417, + "grad_norm": 0.29310208559036255, + "learning_rate": 4.753732676158593e-05, + "loss": 0.0204, + "step": 5400 + }, + { + "epoch": 3.6852861035422344, + "grad_norm": 0.21984073519706726, + "learning_rate": 4.737219422623672e-05, + "loss": 0.019, + "step": 5410 + }, + { + "epoch": 3.692098092643052, + "grad_norm": 0.23975971341133118, + "learning_rate": 4.720709042815044e-05, + "loss": 0.0175, + "step": 5420 + }, + { + "epoch": 3.698910081743869, + "grad_norm": 0.3201830983161926, + "learning_rate": 4.704201717287578e-05, + "loss": 0.0205, + "step": 5430 + }, + { + "epoch": 3.7057220708446867, + "grad_norm": 0.23039595782756805, + "learning_rate": 4.6876976265627404e-05, + "loss": 0.0205, + "step": 5440 + }, + { + "epoch": 3.7125340599455043, + "grad_norm": 0.23342740535736084, + "learning_rate": 4.671196951126626e-05, + "loss": 0.0207, + "step": 5450 + }, + { + "epoch": 3.7193460490463215, + "grad_norm": 0.2592983543872833, + "learning_rate": 4.654699871427971e-05, + "loss": 0.0212, + "step": 5460 + }, + { + "epoch": 3.726158038147139, + "grad_norm": 0.27355700731277466, + "learning_rate": 4.6382065678762034e-05, + "loss": 0.0202, + "step": 5470 + }, + { + "epoch": 3.7329700272479567, + "grad_norm": 0.23657095432281494, + "learning_rate": 4.6217172208394424e-05, + "loss": 0.0171, + "step": 5480 + }, + { + "epoch": 3.739782016348774, + "grad_norm": 0.31331780552864075, + "learning_rate": 4.605232010642549e-05, + "loss": 0.0174, + "step": 5490 + }, + { + "epoch": 3.7465940054495914, + "grad_norm": 0.41772225499153137, + "learning_rate": 4.588751117565142e-05, + "loss": 0.0169, + "step": 5500 + }, + { + "epoch": 3.7534059945504086, + "grad_norm": 0.3513866066932678, + "learning_rate": 4.5722747218396214e-05, + "loss": 0.0186, + "step": 5510 + }, + { + "epoch": 3.760217983651226, + "grad_norm": 0.2156909555196762, + "learning_rate": 4.5558030036492194e-05, + "loss": 0.0178, + "step": 5520 + }, + { + "epoch": 3.7670299727520433, + "grad_norm": 0.2137373983860016, + "learning_rate": 4.539336143125999e-05, + "loss": 0.0238, + "step": 5530 + }, + { + "epoch": 3.773841961852861, + "grad_norm": 0.35196277499198914, + "learning_rate": 4.522874320348916e-05, + "loss": 0.0205, + "step": 5540 + }, + { + "epoch": 3.7806539509536785, + "grad_norm": 0.30024805665016174, + "learning_rate": 4.506417715341821e-05, + "loss": 0.0168, + "step": 5550 + }, + { + "epoch": 3.7874659400544957, + "grad_norm": 0.24347089231014252, + "learning_rate": 4.489966508071511e-05, + "loss": 0.0182, + "step": 5560 + }, + { + "epoch": 3.7942779291553133, + "grad_norm": 0.30325427651405334, + "learning_rate": 4.4735208784457575e-05, + "loss": 0.0178, + "step": 5570 + }, + { + "epoch": 3.801089918256131, + "grad_norm": 0.36180177330970764, + "learning_rate": 4.457081006311325e-05, + "loss": 0.0207, + "step": 5580 + }, + { + "epoch": 3.807901907356948, + "grad_norm": 0.23701462149620056, + "learning_rate": 4.440647071452027e-05, + "loss": 0.0225, + "step": 5590 + }, + { + "epoch": 3.8147138964577656, + "grad_norm": 0.2854270935058594, + "learning_rate": 4.424219253586737e-05, + "loss": 0.0198, + "step": 5600 + }, + { + "epoch": 3.821525885558583, + "grad_norm": 0.25790104269981384, + "learning_rate": 4.407797732367443e-05, + "loss": 0.0215, + "step": 5610 + }, + { + "epoch": 3.8283378746594003, + "grad_norm": 0.1979134976863861, + "learning_rate": 4.391382687377268e-05, + "loss": 0.0228, + "step": 5620 + }, + { + "epoch": 3.835149863760218, + "grad_norm": 0.18029972910881042, + "learning_rate": 4.374974298128512e-05, + "loss": 0.0173, + "step": 5630 + }, + { + "epoch": 3.8419618528610355, + "grad_norm": 0.1711902916431427, + "learning_rate": 4.358572744060699e-05, + "loss": 0.0208, + "step": 5640 + }, + { + "epoch": 3.8487738419618527, + "grad_norm": 0.22609207034111023, + "learning_rate": 4.342178204538588e-05, + "loss": 0.0168, + "step": 5650 + }, + { + "epoch": 3.8555858310626703, + "grad_norm": 0.2340589314699173, + "learning_rate": 4.325790858850241e-05, + "loss": 0.019, + "step": 5660 + }, + { + "epoch": 3.862397820163488, + "grad_norm": 0.19960805773735046, + "learning_rate": 4.309410886205043e-05, + "loss": 0.0198, + "step": 5670 + }, + { + "epoch": 3.869209809264305, + "grad_norm": 0.19948679208755493, + "learning_rate": 4.293038465731752e-05, + "loss": 0.0193, + "step": 5680 + }, + { + "epoch": 3.8760217983651226, + "grad_norm": 0.21270254254341125, + "learning_rate": 4.276673776476533e-05, + "loss": 0.0193, + "step": 5690 + }, + { + "epoch": 3.88283378746594, + "grad_norm": 0.23165644705295563, + "learning_rate": 4.260316997401007e-05, + "loss": 0.0165, + "step": 5700 + }, + { + "epoch": 3.8896457765667574, + "grad_norm": 0.3206821382045746, + "learning_rate": 4.243968307380293e-05, + "loss": 0.0215, + "step": 5710 + }, + { + "epoch": 3.896457765667575, + "grad_norm": 0.2602069675922394, + "learning_rate": 4.22762788520104e-05, + "loss": 0.0234, + "step": 5720 + }, + { + "epoch": 3.9032697547683926, + "grad_norm": 0.17443610727787018, + "learning_rate": 4.211295909559491e-05, + "loss": 0.0226, + "step": 5730 + }, + { + "epoch": 3.9100817438692097, + "grad_norm": 0.19765788316726685, + "learning_rate": 4.194972559059511e-05, + "loss": 0.0207, + "step": 5740 + }, + { + "epoch": 3.9168937329700273, + "grad_norm": 0.21394965052604675, + "learning_rate": 4.178658012210651e-05, + "loss": 0.0181, + "step": 5750 + }, + { + "epoch": 3.923705722070845, + "grad_norm": 0.20210644602775574, + "learning_rate": 4.162352447426177e-05, + "loss": 0.0186, + "step": 5760 + }, + { + "epoch": 3.930517711171662, + "grad_norm": 0.2021627426147461, + "learning_rate": 4.146056043021135e-05, + "loss": 0.0181, + "step": 5770 + }, + { + "epoch": 3.9373297002724796, + "grad_norm": 0.24887089431285858, + "learning_rate": 4.1297689772103944e-05, + "loss": 0.0198, + "step": 5780 + }, + { + "epoch": 3.9441416893732972, + "grad_norm": 0.22847704589366913, + "learning_rate": 4.113491428106694e-05, + "loss": 0.0223, + "step": 5790 + }, + { + "epoch": 3.9509536784741144, + "grad_norm": 0.3041362464427948, + "learning_rate": 4.0972235737187055e-05, + "loss": 0.0167, + "step": 5800 + }, + { + "epoch": 3.957765667574932, + "grad_norm": 0.193576380610466, + "learning_rate": 4.080965591949076e-05, + "loss": 0.0169, + "step": 5810 + }, + { + "epoch": 3.9645776566757496, + "grad_norm": 0.22195811569690704, + "learning_rate": 4.0647176605924924e-05, + "loss": 0.016, + "step": 5820 + }, + { + "epoch": 3.9713896457765667, + "grad_norm": 0.2838583290576935, + "learning_rate": 4.0484799573337255e-05, + "loss": 0.0217, + "step": 5830 + }, + { + "epoch": 3.9782016348773843, + "grad_norm": 0.14418669044971466, + "learning_rate": 4.032252659745699e-05, + "loss": 0.021, + "step": 5840 + }, + { + "epoch": 3.9850136239782015, + "grad_norm": 0.1774141639471054, + "learning_rate": 4.016035945287539e-05, + "loss": 0.023, + "step": 5850 + }, + { + "epoch": 3.991825613079019, + "grad_norm": 0.19315625727176666, + "learning_rate": 3.999829991302635e-05, + "loss": 0.0211, + "step": 5860 + }, + { + "epoch": 3.9986376021798367, + "grad_norm": 0.1897142082452774, + "learning_rate": 3.983634975016707e-05, + "loss": 0.0175, + "step": 5870 + }, + { + "epoch": 4.005449591280654, + "grad_norm": 0.2607768476009369, + "learning_rate": 3.967451073535854e-05, + "loss": 0.0237, + "step": 5880 + }, + { + "epoch": 4.012261580381471, + "grad_norm": 0.25781551003456116, + "learning_rate": 3.951278463844633e-05, + "loss": 0.0208, + "step": 5890 + }, + { + "epoch": 4.0190735694822886, + "grad_norm": 0.18247337639331818, + "learning_rate": 3.935117322804111e-05, + "loss": 0.0172, + "step": 5900 + }, + { + "epoch": 4.025885558583107, + "grad_norm": 0.2261865735054016, + "learning_rate": 3.918967827149938e-05, + "loss": 0.0166, + "step": 5910 + }, + { + "epoch": 4.032697547683924, + "grad_norm": 0.21603935956954956, + "learning_rate": 3.9028301534904094e-05, + "loss": 0.0205, + "step": 5920 + }, + { + "epoch": 4.039509536784741, + "grad_norm": 0.22191236913204193, + "learning_rate": 3.88670447830454e-05, + "loss": 0.0166, + "step": 5930 + }, + { + "epoch": 4.046321525885559, + "grad_norm": 0.38384899497032166, + "learning_rate": 3.870590977940132e-05, + "loss": 0.0191, + "step": 5940 + }, + { + "epoch": 4.053133514986376, + "grad_norm": 0.4138328433036804, + "learning_rate": 3.8544898286118404e-05, + "loss": 0.0185, + "step": 5950 + }, + { + "epoch": 4.059945504087193, + "grad_norm": 0.24633468687534332, + "learning_rate": 3.838401206399257e-05, + "loss": 0.0158, + "step": 5960 + }, + { + "epoch": 4.066757493188011, + "grad_norm": 0.21638496220111847, + "learning_rate": 3.822325287244975e-05, + "loss": 0.0182, + "step": 5970 + }, + { + "epoch": 4.073569482288828, + "grad_norm": 0.18031851947307587, + "learning_rate": 3.8062622469526725e-05, + "loss": 0.0168, + "step": 5980 + }, + { + "epoch": 4.080381471389646, + "grad_norm": 0.27282705903053284, + "learning_rate": 3.790212261185183e-05, + "loss": 0.0176, + "step": 5990 + }, + { + "epoch": 4.087193460490464, + "grad_norm": 0.17047911882400513, + "learning_rate": 3.7741755054625794e-05, + "loss": 0.0194, + "step": 6000 + }, + { + "epoch": 4.094005449591281, + "grad_norm": 0.2330809086561203, + "learning_rate": 3.758152155160255e-05, + "loss": 0.0136, + "step": 6010 + }, + { + "epoch": 4.100817438692098, + "grad_norm": 0.24978803098201752, + "learning_rate": 3.742142385506999e-05, + "loss": 0.0193, + "step": 6020 + }, + { + "epoch": 4.107629427792916, + "grad_norm": 0.22534865140914917, + "learning_rate": 3.72614637158309e-05, + "loss": 0.0159, + "step": 6030 + }, + { + "epoch": 4.114441416893733, + "grad_norm": 0.21492606401443481, + "learning_rate": 3.710164288318371e-05, + "loss": 0.0177, + "step": 6040 + }, + { + "epoch": 4.12125340599455, + "grad_norm": 0.24219180643558502, + "learning_rate": 3.694196310490345e-05, + "loss": 0.0154, + "step": 6050 + }, + { + "epoch": 4.128065395095367, + "grad_norm": 0.2772568464279175, + "learning_rate": 3.678242612722259e-05, + "loss": 0.0169, + "step": 6060 + }, + { + "epoch": 4.1348773841961854, + "grad_norm": 0.1716194897890091, + "learning_rate": 3.6623033694811953e-05, + "loss": 0.0165, + "step": 6070 + }, + { + "epoch": 4.141689373297003, + "grad_norm": 0.19256694614887238, + "learning_rate": 3.6463787550761665e-05, + "loss": 0.0185, + "step": 6080 + }, + { + "epoch": 4.14850136239782, + "grad_norm": 0.18047606945037842, + "learning_rate": 3.630468943656202e-05, + "loss": 0.0155, + "step": 6090 + }, + { + "epoch": 4.155313351498638, + "grad_norm": 0.1661255806684494, + "learning_rate": 3.6145741092084523e-05, + "loss": 0.0133, + "step": 6100 + }, + { + "epoch": 4.162125340599455, + "grad_norm": 0.16505573689937592, + "learning_rate": 3.598694425556278e-05, + "loss": 0.0146, + "step": 6110 + }, + { + "epoch": 4.168937329700272, + "grad_norm": 0.27889305353164673, + "learning_rate": 3.58283006635736e-05, + "loss": 0.0161, + "step": 6120 + }, + { + "epoch": 4.17574931880109, + "grad_norm": 0.25940778851509094, + "learning_rate": 3.566981205101781e-05, + "loss": 0.0175, + "step": 6130 + }, + { + "epoch": 4.182561307901907, + "grad_norm": 0.13758116960525513, + "learning_rate": 3.5511480151101556e-05, + "loss": 0.0146, + "step": 6140 + }, + { + "epoch": 4.189373297002724, + "grad_norm": 0.23441831767559052, + "learning_rate": 3.5353306695317104e-05, + "loss": 0.0141, + "step": 6150 + }, + { + "epoch": 4.1961852861035425, + "grad_norm": 0.17541725933551788, + "learning_rate": 3.519529341342402e-05, + "loss": 0.0176, + "step": 6160 + }, + { + "epoch": 4.20299727520436, + "grad_norm": 0.17384788393974304, + "learning_rate": 3.503744203343026e-05, + "loss": 0.0166, + "step": 6170 + }, + { + "epoch": 4.209809264305177, + "grad_norm": 0.15476594865322113, + "learning_rate": 3.487975428157318e-05, + "loss": 0.018, + "step": 6180 + }, + { + "epoch": 4.216621253405995, + "grad_norm": 0.20996607840061188, + "learning_rate": 3.472223188230083e-05, + "loss": 0.0177, + "step": 6190 + }, + { + "epoch": 4.223433242506812, + "grad_norm": 0.12545377016067505, + "learning_rate": 3.4564876558252866e-05, + "loss": 0.0168, + "step": 6200 + }, + { + "epoch": 4.230245231607629, + "grad_norm": 0.21681755781173706, + "learning_rate": 3.440769003024195e-05, + "loss": 0.0139, + "step": 6210 + }, + { + "epoch": 4.237057220708447, + "grad_norm": 0.20226146280765533, + "learning_rate": 3.425067401723477e-05, + "loss": 0.0168, + "step": 6220 + }, + { + "epoch": 4.243869209809264, + "grad_norm": 0.11783295124769211, + "learning_rate": 3.409383023633325e-05, + "loss": 0.017, + "step": 6230 + }, + { + "epoch": 4.2506811989100814, + "grad_norm": 0.2652694880962372, + "learning_rate": 3.3937160402755894e-05, + "loss": 0.0154, + "step": 6240 + }, + { + "epoch": 4.2574931880108995, + "grad_norm": 0.19120647013187408, + "learning_rate": 3.378066622981885e-05, + "loss": 0.0177, + "step": 6250 + }, + { + "epoch": 4.264305177111717, + "grad_norm": 0.1699381023645401, + "learning_rate": 3.362434942891738e-05, + "loss": 0.0193, + "step": 6260 + }, + { + "epoch": 4.271117166212534, + "grad_norm": 0.24561740458011627, + "learning_rate": 3.346821170950693e-05, + "loss": 0.0161, + "step": 6270 + }, + { + "epoch": 4.277929155313352, + "grad_norm": 0.20055201649665833, + "learning_rate": 3.3312254779084585e-05, + "loss": 0.0171, + "step": 6280 + }, + { + "epoch": 4.284741144414169, + "grad_norm": 0.1489194929599762, + "learning_rate": 3.315648034317039e-05, + "loss": 0.0212, + "step": 6290 + }, + { + "epoch": 4.291553133514986, + "grad_norm": 0.26261234283447266, + "learning_rate": 3.3000890105288564e-05, + "loss": 0.0152, + "step": 6300 + }, + { + "epoch": 4.298365122615804, + "grad_norm": 0.13342000544071198, + "learning_rate": 3.284548576694908e-05, + "loss": 0.0165, + "step": 6310 + }, + { + "epoch": 4.305177111716621, + "grad_norm": 0.21961462497711182, + "learning_rate": 3.2690269027628815e-05, + "loss": 0.0144, + "step": 6320 + }, + { + "epoch": 4.3119891008174385, + "grad_norm": 0.24709409475326538, + "learning_rate": 3.253524158475324e-05, + "loss": 0.0154, + "step": 6330 + }, + { + "epoch": 4.3188010899182565, + "grad_norm": 0.17531152069568634, + "learning_rate": 3.238040513367757e-05, + "loss": 0.0153, + "step": 6340 + }, + { + "epoch": 4.325613079019074, + "grad_norm": 0.18510577082633972, + "learning_rate": 3.222576136766843e-05, + "loss": 0.0129, + "step": 6350 + }, + { + "epoch": 4.332425068119891, + "grad_norm": 0.30215612053871155, + "learning_rate": 3.2071311977885324e-05, + "loss": 0.0161, + "step": 6360 + }, + { + "epoch": 4.339237057220709, + "grad_norm": 0.29397255182266235, + "learning_rate": 3.191705865336197e-05, + "loss": 0.0175, + "step": 6370 + }, + { + "epoch": 4.346049046321526, + "grad_norm": 0.23380252718925476, + "learning_rate": 3.1763003080988075e-05, + "loss": 0.0157, + "step": 6380 + }, + { + "epoch": 4.352861035422343, + "grad_norm": 0.18850485980510712, + "learning_rate": 3.160914694549063e-05, + "loss": 0.0179, + "step": 6390 + }, + { + "epoch": 4.359673024523161, + "grad_norm": 0.28690826892852783, + "learning_rate": 3.145549192941573e-05, + "loss": 0.0161, + "step": 6400 + }, + { + "epoch": 4.366485013623978, + "grad_norm": 0.24790963530540466, + "learning_rate": 3.130203971310999e-05, + "loss": 0.0177, + "step": 6410 + }, + { + "epoch": 4.3732970027247955, + "grad_norm": 0.1769426017999649, + "learning_rate": 3.114879197470225e-05, + "loss": 0.0154, + "step": 6420 + }, + { + "epoch": 4.3801089918256135, + "grad_norm": 0.30168992280960083, + "learning_rate": 3.0995750390085285e-05, + "loss": 0.0138, + "step": 6430 + }, + { + "epoch": 4.386920980926431, + "grad_norm": 0.23833005130290985, + "learning_rate": 3.084291663289728e-05, + "loss": 0.0151, + "step": 6440 + }, + { + "epoch": 4.393732970027248, + "grad_norm": 0.23216956853866577, + "learning_rate": 3.069029237450375e-05, + "loss": 0.0129, + "step": 6450 + }, + { + "epoch": 4.400544959128065, + "grad_norm": 0.17443247139453888, + "learning_rate": 3.053787928397911e-05, + "loss": 0.0128, + "step": 6460 + }, + { + "epoch": 4.407356948228883, + "grad_norm": 0.2944154739379883, + "learning_rate": 3.0385679028088526e-05, + "loss": 0.0171, + "step": 6470 + }, + { + "epoch": 4.4141689373297, + "grad_norm": 0.19948381185531616, + "learning_rate": 3.023369327126959e-05, + "loss": 0.0136, + "step": 6480 + }, + { + "epoch": 4.420980926430517, + "grad_norm": 0.18990705907344818, + "learning_rate": 3.0081923675614198e-05, + "loss": 0.0151, + "step": 6490 + }, + { + "epoch": 4.427792915531335, + "grad_norm": 0.19288881123065948, + "learning_rate": 2.993037190085034e-05, + "loss": 0.0145, + "step": 6500 + }, + { + "epoch": 4.4346049046321525, + "grad_norm": 0.15834234654903412, + "learning_rate": 2.977903960432392e-05, + "loss": 0.0155, + "step": 6510 + }, + { + "epoch": 4.44141689373297, + "grad_norm": 0.15182115137577057, + "learning_rate": 2.9627928440980722e-05, + "loss": 0.0176, + "step": 6520 + }, + { + "epoch": 4.448228882833788, + "grad_norm": 0.20963698625564575, + "learning_rate": 2.9477040063348183e-05, + "loss": 0.0196, + "step": 6530 + }, + { + "epoch": 4.455040871934605, + "grad_norm": 0.18256080150604248, + "learning_rate": 2.9326376121517456e-05, + "loss": 0.017, + "step": 6540 + }, + { + "epoch": 4.461852861035422, + "grad_norm": 0.24118287861347198, + "learning_rate": 2.9175938263125236e-05, + "loss": 0.0153, + "step": 6550 + }, + { + "epoch": 4.46866485013624, + "grad_norm": 0.19935616850852966, + "learning_rate": 2.9025728133335873e-05, + "loss": 0.0168, + "step": 6560 + }, + { + "epoch": 4.475476839237057, + "grad_norm": 0.20024457573890686, + "learning_rate": 2.8875747374823288e-05, + "loss": 0.0195, + "step": 6570 + }, + { + "epoch": 4.482288828337874, + "grad_norm": 0.22161535918712616, + "learning_rate": 2.872599762775298e-05, + "loss": 0.0214, + "step": 6580 + }, + { + "epoch": 4.489100817438692, + "grad_norm": 0.161635160446167, + "learning_rate": 2.857648052976425e-05, + "loss": 0.0145, + "step": 6590 + }, + { + "epoch": 4.4959128065395095, + "grad_norm": 0.18605463206768036, + "learning_rate": 2.8427197715952047e-05, + "loss": 0.0149, + "step": 6600 + }, + { + "epoch": 4.502724795640327, + "grad_norm": 0.17629405856132507, + "learning_rate": 2.8278150818849393e-05, + "loss": 0.0164, + "step": 6610 + }, + { + "epoch": 4.509536784741145, + "grad_norm": 0.12768879532814026, + "learning_rate": 2.812934146840922e-05, + "loss": 0.0141, + "step": 6620 + }, + { + "epoch": 4.516348773841962, + "grad_norm": 0.23254626989364624, + "learning_rate": 2.7980771291986764e-05, + "loss": 0.0172, + "step": 6630 + }, + { + "epoch": 4.523160762942779, + "grad_norm": 0.240464448928833, + "learning_rate": 2.783244191432167e-05, + "loss": 0.0148, + "step": 6640 + }, + { + "epoch": 4.529972752043597, + "grad_norm": 0.2564268708229065, + "learning_rate": 2.768435495752022e-05, + "loss": 0.0129, + "step": 6650 + }, + { + "epoch": 4.536784741144414, + "grad_norm": 0.2194836139678955, + "learning_rate": 2.753651204103771e-05, + "loss": 0.0165, + "step": 6660 + }, + { + "epoch": 4.543596730245231, + "grad_norm": 0.18056434392929077, + "learning_rate": 2.7388914781660523e-05, + "loss": 0.0148, + "step": 6670 + }, + { + "epoch": 4.550408719346049, + "grad_norm": 0.14350150525569916, + "learning_rate": 2.7241564793488693e-05, + "loss": 0.0166, + "step": 6680 + }, + { + "epoch": 4.5572207084468666, + "grad_norm": 0.1394663155078888, + "learning_rate": 2.7094463687918037e-05, + "loss": 0.0149, + "step": 6690 + }, + { + "epoch": 4.564032697547684, + "grad_norm": 0.15998008847236633, + "learning_rate": 2.694761307362268e-05, + "loss": 0.0146, + "step": 6700 + }, + { + "epoch": 4.570844686648502, + "grad_norm": 0.16217146813869476, + "learning_rate": 2.6801014556537467e-05, + "loss": 0.0143, + "step": 6710 + }, + { + "epoch": 4.577656675749319, + "grad_norm": 0.28801918029785156, + "learning_rate": 2.6654669739840243e-05, + "loss": 0.0154, + "step": 6720 + }, + { + "epoch": 4.584468664850136, + "grad_norm": 0.24640391767024994, + "learning_rate": 2.650858022393451e-05, + "loss": 0.0173, + "step": 6730 + }, + { + "epoch": 4.591280653950953, + "grad_norm": 0.2854185700416565, + "learning_rate": 2.6362747606431747e-05, + "loss": 0.0175, + "step": 6740 + }, + { + "epoch": 4.598092643051771, + "grad_norm": 0.23086325824260712, + "learning_rate": 2.6217173482134172e-05, + "loss": 0.0155, + "step": 6750 + }, + { + "epoch": 4.604904632152588, + "grad_norm": 0.26184967160224915, + "learning_rate": 2.6071859443017044e-05, + "loss": 0.0125, + "step": 6760 + }, + { + "epoch": 4.6117166212534055, + "grad_norm": 0.25709471106529236, + "learning_rate": 2.5926807078211414e-05, + "loss": 0.0138, + "step": 6770 + }, + { + "epoch": 4.618528610354224, + "grad_norm": 0.32277151942253113, + "learning_rate": 2.5782017973986728e-05, + "loss": 0.0147, + "step": 6780 + }, + { + "epoch": 4.625340599455041, + "grad_norm": 0.16123652458190918, + "learning_rate": 2.5637493713733374e-05, + "loss": 0.0157, + "step": 6790 + }, + { + "epoch": 4.632152588555858, + "grad_norm": 0.1718718409538269, + "learning_rate": 2.549323587794559e-05, + "loss": 0.0185, + "step": 6800 + }, + { + "epoch": 4.638964577656676, + "grad_norm": 0.2262660712003708, + "learning_rate": 2.5349246044203895e-05, + "loss": 0.015, + "step": 6810 + }, + { + "epoch": 4.645776566757493, + "grad_norm": 0.18175745010375977, + "learning_rate": 2.520552578715808e-05, + "loss": 0.0131, + "step": 6820 + }, + { + "epoch": 4.65258855585831, + "grad_norm": 0.17117254436016083, + "learning_rate": 2.506207667850981e-05, + "loss": 0.0141, + "step": 6830 + }, + { + "epoch": 4.659400544959128, + "grad_norm": 0.17965221405029297, + "learning_rate": 2.4918900286995555e-05, + "loss": 0.0128, + "step": 6840 + }, + { + "epoch": 4.666212534059945, + "grad_norm": 0.13503818213939667, + "learning_rate": 2.4775998178369458e-05, + "loss": 0.0118, + "step": 6850 + }, + { + "epoch": 4.6730245231607626, + "grad_norm": 0.18765027821063995, + "learning_rate": 2.4633371915386017e-05, + "loss": 0.0143, + "step": 6860 + }, + { + "epoch": 4.679836512261581, + "grad_norm": 0.22588397562503815, + "learning_rate": 2.4491023057783235e-05, + "loss": 0.0177, + "step": 6870 + }, + { + "epoch": 4.686648501362398, + "grad_norm": 0.16013208031654358, + "learning_rate": 2.4348953162265375e-05, + "loss": 0.0159, + "step": 6880 + }, + { + "epoch": 4.693460490463215, + "grad_norm": 0.2536979019641876, + "learning_rate": 2.420716378248607e-05, + "loss": 0.0158, + "step": 6890 + }, + { + "epoch": 4.700272479564033, + "grad_norm": 0.16003888845443726, + "learning_rate": 2.4065656469031266e-05, + "loss": 0.0119, + "step": 6900 + }, + { + "epoch": 4.70708446866485, + "grad_norm": 0.15334129333496094, + "learning_rate": 2.3924432769402268e-05, + "loss": 0.0152, + "step": 6910 + }, + { + "epoch": 4.713896457765667, + "grad_norm": 0.21657173335552216, + "learning_rate": 2.3783494227998844e-05, + "loss": 0.0184, + "step": 6920 + }, + { + "epoch": 4.720708446866485, + "grad_norm": 0.2907238006591797, + "learning_rate": 2.3642842386102264e-05, + "loss": 0.0148, + "step": 6930 + }, + { + "epoch": 4.727520435967302, + "grad_norm": 0.3204248547554016, + "learning_rate": 2.3502478781858567e-05, + "loss": 0.0122, + "step": 6940 + }, + { + "epoch": 4.73433242506812, + "grad_norm": 0.25892165303230286, + "learning_rate": 2.3362404950261628e-05, + "loss": 0.0154, + "step": 6950 + }, + { + "epoch": 4.741144414168938, + "grad_norm": 0.2841303050518036, + "learning_rate": 2.3222622423136458e-05, + "loss": 0.0153, + "step": 6960 + }, + { + "epoch": 4.747956403269755, + "grad_norm": 0.12323031574487686, + "learning_rate": 2.3083132729122332e-05, + "loss": 0.0112, + "step": 6970 + }, + { + "epoch": 4.754768392370572, + "grad_norm": 0.33517053723335266, + "learning_rate": 2.294393739365621e-05, + "loss": 0.0181, + "step": 6980 + }, + { + "epoch": 4.76158038147139, + "grad_norm": 0.20667293667793274, + "learning_rate": 2.2805037938956e-05, + "loss": 0.0146, + "step": 6990 + }, + { + "epoch": 4.768392370572207, + "grad_norm": 0.21315595507621765, + "learning_rate": 2.266643588400386e-05, + "loss": 0.0138, + "step": 7000 + }, + { + "epoch": 4.775204359673024, + "grad_norm": 0.22987574338912964, + "learning_rate": 2.252813274452969e-05, + "loss": 0.0145, + "step": 7010 + }, + { + "epoch": 4.782016348773842, + "grad_norm": 0.24111880362033844, + "learning_rate": 2.2390130032994427e-05, + "loss": 0.0126, + "step": 7020 + }, + { + "epoch": 4.7888283378746594, + "grad_norm": 0.18293865025043488, + "learning_rate": 2.2252429258573633e-05, + "loss": 0.0143, + "step": 7030 + }, + { + "epoch": 4.795640326975477, + "grad_norm": 0.37673836946487427, + "learning_rate": 2.2115031927140904e-05, + "loss": 0.0164, + "step": 7040 + }, + { + "epoch": 4.802452316076295, + "grad_norm": 0.16224358975887299, + "learning_rate": 2.1977939541251463e-05, + "loss": 0.0156, + "step": 7050 + }, + { + "epoch": 4.809264305177112, + "grad_norm": 0.20191708207130432, + "learning_rate": 2.1841153600125684e-05, + "loss": 0.0131, + "step": 7060 + }, + { + "epoch": 4.816076294277929, + "grad_norm": 0.15443652868270874, + "learning_rate": 2.170467559963267e-05, + "loss": 0.0151, + "step": 7070 + }, + { + "epoch": 4.822888283378747, + "grad_norm": 0.18957191705703735, + "learning_rate": 2.1568507032273982e-05, + "loss": 0.0116, + "step": 7080 + }, + { + "epoch": 4.829700272479564, + "grad_norm": 0.15831215679645538, + "learning_rate": 2.1432649387167264e-05, + "loss": 0.0125, + "step": 7090 + }, + { + "epoch": 4.836512261580381, + "grad_norm": 0.1532098650932312, + "learning_rate": 2.1297104150029973e-05, + "loss": 0.0143, + "step": 7100 + }, + { + "epoch": 4.843324250681199, + "grad_norm": 0.2064456343650818, + "learning_rate": 2.116187280316307e-05, + "loss": 0.0128, + "step": 7110 + }, + { + "epoch": 4.8501362397820165, + "grad_norm": 0.27514487504959106, + "learning_rate": 2.1026956825434908e-05, + "loss": 0.0125, + "step": 7120 + }, + { + "epoch": 4.856948228882834, + "grad_norm": 0.16616089642047882, + "learning_rate": 2.0892357692265017e-05, + "loss": 0.0129, + "step": 7130 + }, + { + "epoch": 4.863760217983652, + "grad_norm": 0.1578223556280136, + "learning_rate": 2.0758076875607947e-05, + "loss": 0.0157, + "step": 7140 + }, + { + "epoch": 4.870572207084469, + "grad_norm": 0.19922232627868652, + "learning_rate": 2.0624115843937207e-05, + "loss": 0.0153, + "step": 7150 + }, + { + "epoch": 4.877384196185286, + "grad_norm": 0.2548702657222748, + "learning_rate": 2.0490476062229157e-05, + "loss": 0.0155, + "step": 7160 + }, + { + "epoch": 4.884196185286104, + "grad_norm": 0.22706881165504456, + "learning_rate": 2.035715899194704e-05, + "loss": 0.0132, + "step": 7170 + }, + { + "epoch": 4.891008174386921, + "grad_norm": 0.2043786644935608, + "learning_rate": 2.022416609102499e-05, + "loss": 0.0124, + "step": 7180 + }, + { + "epoch": 4.897820163487738, + "grad_norm": 0.19768214225769043, + "learning_rate": 2.009149881385205e-05, + "loss": 0.0182, + "step": 7190 + }, + { + "epoch": 4.904632152588556, + "grad_norm": 0.24597902595996857, + "learning_rate": 1.995915861125634e-05, + "loss": 0.0119, + "step": 7200 + }, + { + "epoch": 4.9114441416893735, + "grad_norm": 0.2110965996980667, + "learning_rate": 1.9827146930489065e-05, + "loss": 0.0111, + "step": 7210 + }, + { + "epoch": 4.918256130790191, + "grad_norm": 0.23993957042694092, + "learning_rate": 1.9695465215208848e-05, + "loss": 0.0156, + "step": 7220 + }, + { + "epoch": 4.925068119891008, + "grad_norm": 0.13470076024532318, + "learning_rate": 1.9564114905465813e-05, + "loss": 0.0127, + "step": 7230 + }, + { + "epoch": 4.931880108991826, + "grad_norm": 0.26347845792770386, + "learning_rate": 1.9433097437685936e-05, + "loss": 0.0131, + "step": 7240 + }, + { + "epoch": 4.938692098092643, + "grad_norm": 0.2081960141658783, + "learning_rate": 1.930241424465521e-05, + "loss": 0.012, + "step": 7250 + }, + { + "epoch": 4.94550408719346, + "grad_norm": 0.12202285975217819, + "learning_rate": 1.9172066755504115e-05, + "loss": 0.0133, + "step": 7260 + }, + { + "epoch": 4.952316076294278, + "grad_norm": 0.14508213102817535, + "learning_rate": 1.9042056395691914e-05, + "loss": 0.0137, + "step": 7270 + }, + { + "epoch": 4.959128065395095, + "grad_norm": 0.16195236146450043, + "learning_rate": 1.8912384586991066e-05, + "loss": 0.0122, + "step": 7280 + }, + { + "epoch": 4.9659400544959125, + "grad_norm": 0.21092760562896729, + "learning_rate": 1.8783052747471717e-05, + "loss": 0.014, + "step": 7290 + }, + { + "epoch": 4.9727520435967305, + "grad_norm": 0.1948050707578659, + "learning_rate": 1.865406229148611e-05, + "loss": 0.0137, + "step": 7300 + }, + { + "epoch": 4.979564032697548, + "grad_norm": 0.19085460901260376, + "learning_rate": 1.8525414629653233e-05, + "loss": 0.0164, + "step": 7310 + }, + { + "epoch": 4.986376021798365, + "grad_norm": 0.25206586718559265, + "learning_rate": 1.8397111168843255e-05, + "loss": 0.013, + "step": 7320 + }, + { + "epoch": 4.993188010899183, + "grad_norm": 0.21286025643348694, + "learning_rate": 1.8269153312162323e-05, + "loss": 0.0138, + "step": 7330 + }, + { + "epoch": 5.0, + "grad_norm": 0.17143577337265015, + "learning_rate": 1.8141542458937054e-05, + "loss": 0.0132, + "step": 7340 + }, + { + "epoch": 5.006811989100817, + "grad_norm": 0.15744146704673767, + "learning_rate": 1.8014280004699268e-05, + "loss": 0.0119, + "step": 7350 + }, + { + "epoch": 5.013623978201635, + "grad_norm": 0.15128405392169952, + "learning_rate": 1.788736734117078e-05, + "loss": 0.0107, + "step": 7360 + }, + { + "epoch": 5.020435967302452, + "grad_norm": 0.3132132589817047, + "learning_rate": 1.7760805856248152e-05, + "loss": 0.0121, + "step": 7370 + }, + { + "epoch": 5.0272479564032695, + "grad_norm": 0.2693178355693817, + "learning_rate": 1.7634596933987518e-05, + "loss": 0.0164, + "step": 7380 + }, + { + "epoch": 5.0340599455040875, + "grad_norm": 0.18982277810573578, + "learning_rate": 1.7508741954589404e-05, + "loss": 0.0145, + "step": 7390 + }, + { + "epoch": 5.040871934604905, + "grad_norm": 0.15594618022441864, + "learning_rate": 1.7383242294383717e-05, + "loss": 0.0122, + "step": 7400 + }, + { + "epoch": 5.047683923705722, + "grad_norm": 0.18597710132598877, + "learning_rate": 1.7258099325814632e-05, + "loss": 0.0094, + "step": 7410 + }, + { + "epoch": 5.05449591280654, + "grad_norm": 0.17391642928123474, + "learning_rate": 1.7133314417425594e-05, + "loss": 0.012, + "step": 7420 + }, + { + "epoch": 5.061307901907357, + "grad_norm": 0.25961941480636597, + "learning_rate": 1.7008888933844408e-05, + "loss": 0.0128, + "step": 7430 + }, + { + "epoch": 5.068119891008174, + "grad_norm": 0.17781174182891846, + "learning_rate": 1.6884824235768172e-05, + "loss": 0.0145, + "step": 7440 + }, + { + "epoch": 5.074931880108992, + "grad_norm": 0.1576293408870697, + "learning_rate": 1.6761121679948592e-05, + "loss": 0.0119, + "step": 7450 + }, + { + "epoch": 5.081743869209809, + "grad_norm": 0.30177032947540283, + "learning_rate": 1.663778261917695e-05, + "loss": 0.0144, + "step": 7460 + }, + { + "epoch": 5.0885558583106265, + "grad_norm": 0.20156201720237732, + "learning_rate": 1.651480840226952e-05, + "loss": 0.0121, + "step": 7470 + }, + { + "epoch": 5.0953678474114446, + "grad_norm": 0.18891334533691406, + "learning_rate": 1.639220037405258e-05, + "loss": 0.0143, + "step": 7480 + }, + { + "epoch": 5.102179836512262, + "grad_norm": 0.15977536141872406, + "learning_rate": 1.6269959875347906e-05, + "loss": 0.0137, + "step": 7490 + }, + { + "epoch": 5.108991825613079, + "grad_norm": 0.2513260841369629, + "learning_rate": 1.614808824295802e-05, + "loss": 0.0149, + "step": 7500 + }, + { + "epoch": 5.115803814713897, + "grad_norm": 0.09697917103767395, + "learning_rate": 1.602658680965152e-05, + "loss": 0.0099, + "step": 7510 + }, + { + "epoch": 5.122615803814714, + "grad_norm": 0.20942772924900055, + "learning_rate": 1.5905456904148686e-05, + "loss": 0.0133, + "step": 7520 + }, + { + "epoch": 5.129427792915531, + "grad_norm": 0.24593091011047363, + "learning_rate": 1.57846998511067e-05, + "loss": 0.0131, + "step": 7530 + }, + { + "epoch": 5.136239782016348, + "grad_norm": 0.20189829170703888, + "learning_rate": 1.566431697110538e-05, + "loss": 0.0108, + "step": 7540 + }, + { + "epoch": 5.143051771117166, + "grad_norm": 0.12284066528081894, + "learning_rate": 1.554430958063259e-05, + "loss": 0.012, + "step": 7550 + }, + { + "epoch": 5.1498637602179835, + "grad_norm": 0.1622915118932724, + "learning_rate": 1.5424678992069912e-05, + "loss": 0.0121, + "step": 7560 + }, + { + "epoch": 5.156675749318801, + "grad_norm": 0.21887880563735962, + "learning_rate": 1.5305426513678362e-05, + "loss": 0.0124, + "step": 7570 + }, + { + "epoch": 5.163487738419619, + "grad_norm": 0.142318993806839, + "learning_rate": 1.518655344958388e-05, + "loss": 0.0156, + "step": 7580 + }, + { + "epoch": 5.170299727520436, + "grad_norm": 0.17450009286403656, + "learning_rate": 1.5068061099763275e-05, + "loss": 0.0124, + "step": 7590 + }, + { + "epoch": 5.177111716621253, + "grad_norm": 0.22019504010677338, + "learning_rate": 1.494995076002988e-05, + "loss": 0.015, + "step": 7600 + }, + { + "epoch": 5.183923705722071, + "grad_norm": 0.18164029717445374, + "learning_rate": 1.4832223722019456e-05, + "loss": 0.0103, + "step": 7610 + }, + { + "epoch": 5.190735694822888, + "grad_norm": 0.22277112305164337, + "learning_rate": 1.4714881273176035e-05, + "loss": 0.0125, + "step": 7620 + }, + { + "epoch": 5.197547683923705, + "grad_norm": 0.24523723125457764, + "learning_rate": 1.4597924696737835e-05, + "loss": 0.0119, + "step": 7630 + }, + { + "epoch": 5.204359673024523, + "grad_norm": 0.16030792891979218, + "learning_rate": 1.4481355271723252e-05, + "loss": 0.0114, + "step": 7640 + }, + { + "epoch": 5.2111716621253406, + "grad_norm": 0.20398828387260437, + "learning_rate": 1.4365174272916809e-05, + "loss": 0.0128, + "step": 7650 + }, + { + "epoch": 5.217983651226158, + "grad_norm": 0.17755308747291565, + "learning_rate": 1.4249382970855319e-05, + "loss": 0.0125, + "step": 7660 + }, + { + "epoch": 5.224795640326976, + "grad_norm": 0.17358122766017914, + "learning_rate": 1.4133982631813903e-05, + "loss": 0.0123, + "step": 7670 + }, + { + "epoch": 5.231607629427793, + "grad_norm": 0.12761126458644867, + "learning_rate": 1.4018974517792194e-05, + "loss": 0.0113, + "step": 7680 + }, + { + "epoch": 5.23841961852861, + "grad_norm": 0.15148918330669403, + "learning_rate": 1.390435988650048e-05, + "loss": 0.0125, + "step": 7690 + }, + { + "epoch": 5.245231607629428, + "grad_norm": 0.2090086191892624, + "learning_rate": 1.3790139991346006e-05, + "loss": 0.0123, + "step": 7700 + }, + { + "epoch": 5.252043596730245, + "grad_norm": 0.22752802073955536, + "learning_rate": 1.367631608141926e-05, + "loss": 0.0115, + "step": 7710 + }, + { + "epoch": 5.258855585831062, + "grad_norm": 0.17745046317577362, + "learning_rate": 1.3562889401480278e-05, + "loss": 0.0127, + "step": 7720 + }, + { + "epoch": 5.26566757493188, + "grad_norm": 0.18680843710899353, + "learning_rate": 1.3449861191945074e-05, + "loss": 0.0125, + "step": 7730 + }, + { + "epoch": 5.272479564032698, + "grad_norm": 0.12418905645608902, + "learning_rate": 1.3337232688872009e-05, + "loss": 0.0114, + "step": 7740 + }, + { + "epoch": 5.279291553133515, + "grad_norm": 0.25357624888420105, + "learning_rate": 1.3225005123948364e-05, + "loss": 0.0111, + "step": 7750 + }, + { + "epoch": 5.286103542234333, + "grad_norm": 0.20203381776809692, + "learning_rate": 1.311317972447681e-05, + "loss": 0.0106, + "step": 7760 + }, + { + "epoch": 5.29291553133515, + "grad_norm": 0.14989571273326874, + "learning_rate": 1.3001757713361996e-05, + "loss": 0.0105, + "step": 7770 + }, + { + "epoch": 5.299727520435967, + "grad_norm": 0.1835460662841797, + "learning_rate": 1.2890740309097204e-05, + "loss": 0.0122, + "step": 7780 + }, + { + "epoch": 5.306539509536785, + "grad_norm": 0.1413341909646988, + "learning_rate": 1.2780128725750944e-05, + "loss": 0.0111, + "step": 7790 + }, + { + "epoch": 5.313351498637602, + "grad_norm": 0.0959068313241005, + "learning_rate": 1.266992417295379e-05, + "loss": 0.012, + "step": 7800 + }, + { + "epoch": 5.320163487738419, + "grad_norm": 0.1972733587026596, + "learning_rate": 1.2560127855885073e-05, + "loss": 0.0122, + "step": 7810 + }, + { + "epoch": 5.3269754768392374, + "grad_norm": 0.16545796394348145, + "learning_rate": 1.2450740975259745e-05, + "loss": 0.0141, + "step": 7820 + }, + { + "epoch": 5.333787465940055, + "grad_norm": 0.09657532721757889, + "learning_rate": 1.234176472731517e-05, + "loss": 0.018, + "step": 7830 + }, + { + "epoch": 5.340599455040872, + "grad_norm": 0.15206950902938843, + "learning_rate": 1.2233200303798158e-05, + "loss": 0.0099, + "step": 7840 + }, + { + "epoch": 5.34741144414169, + "grad_norm": 0.15897098183631897, + "learning_rate": 1.2125048891951846e-05, + "loss": 0.0137, + "step": 7850 + }, + { + "epoch": 5.354223433242507, + "grad_norm": 0.17377428710460663, + "learning_rate": 1.2017311674502745e-05, + "loss": 0.0112, + "step": 7860 + }, + { + "epoch": 5.361035422343324, + "grad_norm": 0.1237255409359932, + "learning_rate": 1.1909989829647822e-05, + "loss": 0.0103, + "step": 7870 + }, + { + "epoch": 5.367847411444142, + "grad_norm": 0.1616630256175995, + "learning_rate": 1.1803084531041553e-05, + "loss": 0.0137, + "step": 7880 + }, + { + "epoch": 5.374659400544959, + "grad_norm": 0.18655730783939362, + "learning_rate": 1.1696596947783162e-05, + "loss": 0.0122, + "step": 7890 + }, + { + "epoch": 5.381471389645776, + "grad_norm": 0.1268177628517151, + "learning_rate": 1.1590528244403803e-05, + "loss": 0.0112, + "step": 7900 + }, + { + "epoch": 5.3882833787465945, + "grad_norm": 0.1680615395307541, + "learning_rate": 1.148487958085382e-05, + "loss": 0.0139, + "step": 7910 + }, + { + "epoch": 5.395095367847412, + "grad_norm": 0.16610464453697205, + "learning_rate": 1.1379652112490086e-05, + "loss": 0.013, + "step": 7920 + }, + { + "epoch": 5.401907356948229, + "grad_norm": 0.1437927931547165, + "learning_rate": 1.1274846990063315e-05, + "loss": 0.0114, + "step": 7930 + }, + { + "epoch": 5.408719346049046, + "grad_norm": 0.16926175355911255, + "learning_rate": 1.117046535970554e-05, + "loss": 0.014, + "step": 7940 + }, + { + "epoch": 5.415531335149864, + "grad_norm": 0.19889235496520996, + "learning_rate": 1.106650836291755e-05, + "loss": 0.0188, + "step": 7950 + }, + { + "epoch": 5.422343324250681, + "grad_norm": 0.19893424212932587, + "learning_rate": 1.0962977136556418e-05, + "loss": 0.0153, + "step": 7960 + }, + { + "epoch": 5.429155313351498, + "grad_norm": 0.17771615087985992, + "learning_rate": 1.0859872812823024e-05, + "loss": 0.0138, + "step": 7970 + }, + { + "epoch": 5.435967302452316, + "grad_norm": 0.13923974335193634, + "learning_rate": 1.0757196519249747e-05, + "loss": 0.0153, + "step": 7980 + }, + { + "epoch": 5.4427792915531334, + "grad_norm": 0.15784096717834473, + "learning_rate": 1.0654949378688077e-05, + "loss": 0.0146, + "step": 7990 + }, + { + "epoch": 5.449591280653951, + "grad_norm": 0.19331657886505127, + "learning_rate": 1.0553132509296376e-05, + "loss": 0.0137, + "step": 8000 + }, + { + "epoch": 5.456403269754769, + "grad_norm": 0.16875822842121124, + "learning_rate": 1.0451747024527613e-05, + "loss": 0.0125, + "step": 8010 + }, + { + "epoch": 5.463215258855586, + "grad_norm": 0.13551338016986847, + "learning_rate": 1.0350794033117189e-05, + "loss": 0.0146, + "step": 8020 + }, + { + "epoch": 5.470027247956403, + "grad_norm": 0.2047499418258667, + "learning_rate": 1.0250274639070856e-05, + "loss": 0.0115, + "step": 8030 + }, + { + "epoch": 5.476839237057221, + "grad_norm": 0.13769468665122986, + "learning_rate": 1.0150189941652599e-05, + "loss": 0.0107, + "step": 8040 + }, + { + "epoch": 5.483651226158038, + "grad_norm": 0.12529806792736053, + "learning_rate": 1.0050541035372635e-05, + "loss": 0.0092, + "step": 8050 + }, + { + "epoch": 5.490463215258855, + "grad_norm": 0.13725043833255768, + "learning_rate": 9.951329009975458e-06, + "loss": 0.0109, + "step": 8060 + }, + { + "epoch": 5.497275204359673, + "grad_norm": 0.11088486760854721, + "learning_rate": 9.852554950427845e-06, + "loss": 0.013, + "step": 8070 + }, + { + "epoch": 5.5040871934604905, + "grad_norm": 0.10260329395532608, + "learning_rate": 9.754219936907105e-06, + "loss": 0.0111, + "step": 8080 + }, + { + "epoch": 5.510899182561308, + "grad_norm": 0.12800352275371552, + "learning_rate": 9.656325044789194e-06, + "loss": 0.0082, + "step": 8090 + }, + { + "epoch": 5.517711171662126, + "grad_norm": 0.13969528675079346, + "learning_rate": 9.55887134463697e-06, + "loss": 0.0085, + "step": 8100 + }, + { + "epoch": 5.524523160762943, + "grad_norm": 0.12898869812488556, + "learning_rate": 9.461859902188475e-06, + "loss": 0.0116, + "step": 8110 + }, + { + "epoch": 5.53133514986376, + "grad_norm": 0.17141078412532806, + "learning_rate": 9.365291778345303e-06, + "loss": 0.0104, + "step": 8120 + }, + { + "epoch": 5.538147138964578, + "grad_norm": 0.09811981767416, + "learning_rate": 9.269168029160991e-06, + "loss": 0.0078, + "step": 8130 + }, + { + "epoch": 5.544959128065395, + "grad_norm": 0.20470815896987915, + "learning_rate": 9.173489705829447e-06, + "loss": 0.0115, + "step": 8140 + }, + { + "epoch": 5.551771117166212, + "grad_norm": 0.16731230914592743, + "learning_rate": 9.078257854673516e-06, + "loss": 0.0112, + "step": 8150 + }, + { + "epoch": 5.55858310626703, + "grad_norm": 0.14652231335639954, + "learning_rate": 8.983473517133429e-06, + "loss": 0.0119, + "step": 8160 + }, + { + "epoch": 5.5653950953678475, + "grad_norm": 0.08646702766418457, + "learning_rate": 8.889137729755537e-06, + "loss": 0.0095, + "step": 8170 + }, + { + "epoch": 5.572207084468665, + "grad_norm": 0.20360282063484192, + "learning_rate": 8.79525152418087e-06, + "loss": 0.0125, + "step": 8180 + }, + { + "epoch": 5.579019073569482, + "grad_norm": 0.16116434335708618, + "learning_rate": 8.701815927133961e-06, + "loss": 0.0095, + "step": 8190 + }, + { + "epoch": 5.5858310626703, + "grad_norm": 0.13089890778064728, + "learning_rate": 8.608831960411534e-06, + "loss": 0.0106, + "step": 8200 + }, + { + "epoch": 5.592643051771117, + "grad_norm": 0.1402096152305603, + "learning_rate": 8.516300640871321e-06, + "loss": 0.0133, + "step": 8210 + }, + { + "epoch": 5.599455040871934, + "grad_norm": 0.14964203536510468, + "learning_rate": 8.424222980421038e-06, + "loss": 0.012, + "step": 8220 + }, + { + "epoch": 5.606267029972752, + "grad_norm": 0.1956614851951599, + "learning_rate": 8.332599986007184e-06, + "loss": 0.0117, + "step": 8230 + }, + { + "epoch": 5.613079019073569, + "grad_norm": 0.11844225227832794, + "learning_rate": 8.241432659604203e-06, + "loss": 0.0108, + "step": 8240 + }, + { + "epoch": 5.6198910081743865, + "grad_norm": 0.17443661391735077, + "learning_rate": 8.150721998203331e-06, + "loss": 0.0084, + "step": 8250 + }, + { + "epoch": 5.6267029972752045, + "grad_norm": 0.17285113036632538, + "learning_rate": 8.06046899380184e-06, + "loss": 0.0104, + "step": 8260 + }, + { + "epoch": 5.633514986376022, + "grad_norm": 0.11764296889305115, + "learning_rate": 7.970674633392133e-06, + "loss": 0.0128, + "step": 8270 + }, + { + "epoch": 5.640326975476839, + "grad_norm": 0.13431662321090698, + "learning_rate": 7.881339898950924e-06, + "loss": 0.0151, + "step": 8280 + }, + { + "epoch": 5.647138964577657, + "grad_norm": 0.09805692732334137, + "learning_rate": 7.792465767428597e-06, + "loss": 0.0116, + "step": 8290 + }, + { + "epoch": 5.653950953678474, + "grad_norm": 0.10658857971429825, + "learning_rate": 7.704053210738376e-06, + "loss": 0.0091, + "step": 8300 + }, + { + "epoch": 5.660762942779291, + "grad_norm": 0.1375443935394287, + "learning_rate": 7.6161031957458494e-06, + "loss": 0.0128, + "step": 8310 + }, + { + "epoch": 5.667574931880109, + "grad_norm": 0.1789049655199051, + "learning_rate": 7.5286166842582605e-06, + "loss": 0.0112, + "step": 8320 + }, + { + "epoch": 5.674386920980926, + "grad_norm": 0.11381709575653076, + "learning_rate": 7.4415946330140814e-06, + "loss": 0.0107, + "step": 8330 + }, + { + "epoch": 5.6811989100817435, + "grad_norm": 0.20492666959762573, + "learning_rate": 7.3550379936725644e-06, + "loss": 0.011, + "step": 8340 + }, + { + "epoch": 5.6880108991825615, + "grad_norm": 0.1754821240901947, + "learning_rate": 7.2689477128032035e-06, + "loss": 0.0127, + "step": 8350 + }, + { + "epoch": 5.694822888283379, + "grad_norm": 0.10873087495565414, + "learning_rate": 7.183324731875551e-06, + "loss": 0.0084, + "step": 8360 + }, + { + "epoch": 5.701634877384196, + "grad_norm": 0.1496199667453766, + "learning_rate": 7.098169987248782e-06, + "loss": 0.0084, + "step": 8370 + }, + { + "epoch": 5.708446866485014, + "grad_norm": 0.09581401199102402, + "learning_rate": 7.013484410161553e-06, + "loss": 0.0101, + "step": 8380 + }, + { + "epoch": 5.715258855585831, + "grad_norm": 0.1942460685968399, + "learning_rate": 6.92926892672176e-06, + "loss": 0.0117, + "step": 8390 + }, + { + "epoch": 5.722070844686648, + "grad_norm": 0.10822322964668274, + "learning_rate": 6.845524457896446e-06, + "loss": 0.0088, + "step": 8400 + }, + { + "epoch": 5.728882833787466, + "grad_norm": 0.1585640013217926, + "learning_rate": 6.7622519195017165e-06, + "loss": 0.0097, + "step": 8410 + }, + { + "epoch": 5.735694822888283, + "grad_norm": 0.1776353269815445, + "learning_rate": 6.679452222192684e-06, + "loss": 0.0094, + "step": 8420 + }, + { + "epoch": 5.7425068119891005, + "grad_norm": 0.12591688334941864, + "learning_rate": 6.597126271453579e-06, + "loss": 0.0088, + "step": 8430 + }, + { + "epoch": 5.7493188010899186, + "grad_norm": 0.10450952500104904, + "learning_rate": 6.51527496758782e-06, + "loss": 0.0093, + "step": 8440 + }, + { + "epoch": 5.756130790190736, + "grad_norm": 0.12576451897621155, + "learning_rate": 6.433899205708155e-06, + "loss": 0.0096, + "step": 8450 + }, + { + "epoch": 5.762942779291553, + "grad_norm": 0.12062663584947586, + "learning_rate": 6.352999875726856e-06, + "loss": 0.008, + "step": 8460 + }, + { + "epoch": 5.769754768392371, + "grad_norm": 0.13374954462051392, + "learning_rate": 6.272577862346052e-06, + "loss": 0.0085, + "step": 8470 + }, + { + "epoch": 5.776566757493188, + "grad_norm": 0.20056910812854767, + "learning_rate": 6.192634045047996e-06, + "loss": 0.0106, + "step": 8480 + }, + { + "epoch": 5.783378746594005, + "grad_norm": 0.18743054568767548, + "learning_rate": 6.113169298085458e-06, + "loss": 0.017, + "step": 8490 + }, + { + "epoch": 5.790190735694823, + "grad_norm": 0.15564921498298645, + "learning_rate": 6.034184490472195e-06, + "loss": 0.0074, + "step": 8500 + }, + { + "epoch": 5.79700272479564, + "grad_norm": 0.22393585741519928, + "learning_rate": 5.955680485973386e-06, + "loss": 0.0128, + "step": 8510 + }, + { + "epoch": 5.8038147138964575, + "grad_norm": 0.21653935313224792, + "learning_rate": 5.877658143096265e-06, + "loss": 0.01, + "step": 8520 + }, + { + "epoch": 5.810626702997276, + "grad_norm": 0.14136868715286255, + "learning_rate": 5.800118315080661e-06, + "loss": 0.0094, + "step": 8530 + }, + { + "epoch": 5.817438692098093, + "grad_norm": 0.14055469632148743, + "learning_rate": 5.723061849889716e-06, + "loss": 0.0084, + "step": 8540 + }, + { + "epoch": 5.82425068119891, + "grad_norm": 0.24212098121643066, + "learning_rate": 5.646489590200604e-06, + "loss": 0.0115, + "step": 8550 + }, + { + "epoch": 5.831062670299728, + "grad_norm": 0.14534185826778412, + "learning_rate": 5.570402373395256e-06, + "loss": 0.0096, + "step": 8560 + }, + { + "epoch": 5.837874659400545, + "grad_norm": 0.22133882343769073, + "learning_rate": 5.494801031551305e-06, + "loss": 0.0121, + "step": 8570 + }, + { + "epoch": 5.844686648501362, + "grad_norm": 0.13953863084316254, + "learning_rate": 5.41968639143291e-06, + "loss": 0.0089, + "step": 8580 + }, + { + "epoch": 5.85149863760218, + "grad_norm": 0.13869096338748932, + "learning_rate": 5.345059274481751e-06, + "loss": 0.0087, + "step": 8590 + }, + { + "epoch": 5.858310626702997, + "grad_norm": 0.11316031217575073, + "learning_rate": 5.270920496808002e-06, + "loss": 0.0127, + "step": 8600 + }, + { + "epoch": 5.8651226158038146, + "grad_norm": 0.11996561288833618, + "learning_rate": 5.1972708691814695e-06, + "loss": 0.0101, + "step": 8610 + }, + { + "epoch": 5.871934604904633, + "grad_norm": 0.1471121609210968, + "learning_rate": 5.124111197022674e-06, + "loss": 0.0123, + "step": 8620 + }, + { + "epoch": 5.87874659400545, + "grad_norm": 0.13705261051654816, + "learning_rate": 5.051442280394081e-06, + "loss": 0.0082, + "step": 8630 + }, + { + "epoch": 5.885558583106267, + "grad_norm": 0.12985871732234955, + "learning_rate": 4.979264913991322e-06, + "loss": 0.01, + "step": 8640 + }, + { + "epoch": 5.892370572207085, + "grad_norm": 0.14874781668186188, + "learning_rate": 4.907579887134489e-06, + "loss": 0.0111, + "step": 8650 + }, + { + "epoch": 5.899182561307902, + "grad_norm": 0.17686668038368225, + "learning_rate": 4.836387983759572e-06, + "loss": 0.0105, + "step": 8660 + }, + { + "epoch": 5.905994550408719, + "grad_norm": 0.1565779447555542, + "learning_rate": 4.765689982409816e-06, + "loss": 0.0098, + "step": 8670 + }, + { + "epoch": 5.912806539509537, + "grad_norm": 0.21265514194965363, + "learning_rate": 4.695486656227233e-06, + "loss": 0.0129, + "step": 8680 + }, + { + "epoch": 5.919618528610354, + "grad_norm": 0.15758763253688812, + "learning_rate": 4.625778772944156e-06, + "loss": 0.0095, + "step": 8690 + }, + { + "epoch": 5.926430517711172, + "grad_norm": 0.2025187611579895, + "learning_rate": 4.556567094874825e-06, + "loss": 0.0112, + "step": 8700 + }, + { + "epoch": 5.933242506811989, + "grad_norm": 0.08473379909992218, + "learning_rate": 4.487852378907059e-06, + "loss": 0.0065, + "step": 8710 + }, + { + "epoch": 5.940054495912807, + "grad_norm": 0.14847134053707123, + "learning_rate": 4.419635376493986e-06, + "loss": 0.0108, + "step": 8720 + }, + { + "epoch": 5.946866485013624, + "grad_norm": 0.09602349996566772, + "learning_rate": 4.351916833645825e-06, + "loss": 0.0097, + "step": 8730 + }, + { + "epoch": 5.953678474114441, + "grad_norm": 0.13744130730628967, + "learning_rate": 4.284697490921691e-06, + "loss": 0.0088, + "step": 8740 + }, + { + "epoch": 5.960490463215259, + "grad_norm": 0.16695407032966614, + "learning_rate": 4.2179780834215585e-06, + "loss": 0.0132, + "step": 8750 + }, + { + "epoch": 5.967302452316076, + "grad_norm": 0.18353870511054993, + "learning_rate": 4.151759340778178e-06, + "loss": 0.0104, + "step": 8760 + }, + { + "epoch": 5.974114441416893, + "grad_norm": 0.21144796907901764, + "learning_rate": 4.086041987149109e-06, + "loss": 0.0106, + "step": 8770 + }, + { + "epoch": 5.9809264305177114, + "grad_norm": 0.14679405093193054, + "learning_rate": 4.020826741208811e-06, + "loss": 0.0128, + "step": 8780 + }, + { + "epoch": 5.987738419618529, + "grad_norm": 0.1259392499923706, + "learning_rate": 3.956114316140746e-06, + "loss": 0.0136, + "step": 8790 + }, + { + "epoch": 5.994550408719346, + "grad_norm": 0.0897415429353714, + "learning_rate": 3.891905419629643e-06, + "loss": 0.0091, + "step": 8800 + }, + { + "epoch": 6.001362397820164, + "grad_norm": 0.1489909589290619, + "learning_rate": 3.8282007538536946e-06, + "loss": 0.0117, + "step": 8810 + }, + { + "epoch": 6.008174386920981, + "grad_norm": 0.19597971439361572, + "learning_rate": 3.7650010154769265e-06, + "loss": 0.0088, + "step": 8820 + }, + { + "epoch": 6.014986376021798, + "grad_norm": 0.18179140985012054, + "learning_rate": 3.7023068956415608e-06, + "loss": 0.012, + "step": 8830 + }, + { + "epoch": 6.021798365122616, + "grad_norm": 0.1425846964120865, + "learning_rate": 3.6401190799604303e-06, + "loss": 0.0103, + "step": 8840 + }, + { + "epoch": 6.028610354223433, + "grad_norm": 0.17134077847003937, + "learning_rate": 3.578438248509536e-06, + "loss": 0.0112, + "step": 8850 + }, + { + "epoch": 6.03542234332425, + "grad_norm": 0.10206969082355499, + "learning_rate": 3.5172650758205583e-06, + "loss": 0.0125, + "step": 8860 + }, + { + "epoch": 6.0422343324250685, + "grad_norm": 0.0982321947813034, + "learning_rate": 3.45660023087353e-06, + "loss": 0.0074, + "step": 8870 + }, + { + "epoch": 6.049046321525886, + "grad_norm": 0.11961355060338974, + "learning_rate": 3.3964443770894528e-06, + "loss": 0.0085, + "step": 8880 + }, + { + "epoch": 6.055858310626703, + "grad_norm": 0.14227040112018585, + "learning_rate": 3.3367981723231245e-06, + "loss": 0.0104, + "step": 8890 + }, + { + "epoch": 6.062670299727521, + "grad_norm": 0.13862502574920654, + "learning_rate": 3.2776622688558746e-06, + "loss": 0.0091, + "step": 8900 + }, + { + "epoch": 6.069482288828338, + "grad_norm": 0.14933033287525177, + "learning_rate": 3.2190373133884677e-06, + "loss": 0.0101, + "step": 8910 + }, + { + "epoch": 6.076294277929155, + "grad_norm": 0.17859244346618652, + "learning_rate": 3.1609239470340446e-06, + "loss": 0.0093, + "step": 8920 + }, + { + "epoch": 6.083106267029973, + "grad_norm": 0.12856774032115936, + "learning_rate": 3.1033228053110373e-06, + "loss": 0.0083, + "step": 8930 + }, + { + "epoch": 6.08991825613079, + "grad_norm": 0.12302831560373306, + "learning_rate": 3.0462345181363314e-06, + "loss": 0.0086, + "step": 8940 + }, + { + "epoch": 6.0967302452316074, + "grad_norm": 0.1459636390209198, + "learning_rate": 2.9896597098182654e-06, + "loss": 0.0104, + "step": 8950 + }, + { + "epoch": 6.1035422343324255, + "grad_norm": 0.12581010162830353, + "learning_rate": 2.933598999049891e-06, + "loss": 0.0096, + "step": 8960 + }, + { + "epoch": 6.110354223433243, + "grad_norm": 0.13089345395565033, + "learning_rate": 2.8780529989021697e-06, + "loss": 0.0078, + "step": 8970 + }, + { + "epoch": 6.11716621253406, + "grad_norm": 0.14709091186523438, + "learning_rate": 2.823022316817242e-06, + "loss": 0.0111, + "step": 8980 + }, + { + "epoch": 6.123978201634877, + "grad_norm": 0.2142879217863083, + "learning_rate": 2.7685075546018456e-06, + "loss": 0.0119, + "step": 8990 + }, + { + "epoch": 6.130790190735695, + "grad_norm": 0.11257347464561462, + "learning_rate": 2.7145093084206598e-06, + "loss": 0.01, + "step": 9000 + }, + { + "epoch": 6.137602179836512, + "grad_norm": 0.11935742944478989, + "learning_rate": 2.661028168789892e-06, + "loss": 0.009, + "step": 9010 + }, + { + "epoch": 6.144414168937329, + "grad_norm": 0.16253075003623962, + "learning_rate": 2.6080647205706855e-06, + "loss": 0.009, + "step": 9020 + }, + { + "epoch": 6.151226158038147, + "grad_norm": 0.07740148901939392, + "learning_rate": 2.555619542962834e-06, + "loss": 0.0088, + "step": 9030 + }, + { + "epoch": 6.1580381471389645, + "grad_norm": 0.0897480770945549, + "learning_rate": 2.503693209498409e-06, + "loss": 0.0064, + "step": 9040 + }, + { + "epoch": 6.164850136239782, + "grad_norm": 0.10529225319623947, + "learning_rate": 2.452286288035449e-06, + "loss": 0.0103, + "step": 9050 + }, + { + "epoch": 6.1716621253406, + "grad_norm": 0.16449305415153503, + "learning_rate": 2.4013993407518363e-06, + "loss": 0.0119, + "step": 9060 + }, + { + "epoch": 6.178474114441417, + "grad_norm": 0.113241046667099, + "learning_rate": 2.351032924139063e-06, + "loss": 0.0069, + "step": 9070 + }, + { + "epoch": 6.185286103542234, + "grad_norm": 0.1602308452129364, + "learning_rate": 2.30118758899619e-06, + "loss": 0.0106, + "step": 9080 + }, + { + "epoch": 6.192098092643052, + "grad_norm": 0.1920604109764099, + "learning_rate": 2.2518638804238157e-06, + "loss": 0.0102, + "step": 9090 + }, + { + "epoch": 6.198910081743869, + "grad_norm": 0.18753395974636078, + "learning_rate": 2.203062337818118e-06, + "loss": 0.0095, + "step": 9100 + }, + { + "epoch": 6.205722070844686, + "grad_norm": 0.16297195851802826, + "learning_rate": 2.1547834948649483e-06, + "loss": 0.01, + "step": 9110 + }, + { + "epoch": 6.212534059945504, + "grad_norm": 0.09433993697166443, + "learning_rate": 2.1070278795340017e-06, + "loss": 0.0096, + "step": 9120 + }, + { + "epoch": 6.2193460490463215, + "grad_norm": 0.10109172761440277, + "learning_rate": 2.059796014073029e-06, + "loss": 0.0077, + "step": 9130 + }, + { + "epoch": 6.226158038147139, + "grad_norm": 0.10091448575258255, + "learning_rate": 2.01308841500214e-06, + "loss": 0.0086, + "step": 9140 + }, + { + "epoch": 6.232970027247957, + "grad_norm": 0.2588089406490326, + "learning_rate": 1.9669055931081704e-06, + "loss": 0.0134, + "step": 9150 + }, + { + "epoch": 6.239782016348774, + "grad_norm": 0.13017907738685608, + "learning_rate": 1.9212480534390507e-06, + "loss": 0.0102, + "step": 9160 + }, + { + "epoch": 6.246594005449591, + "grad_norm": 0.17378906905651093, + "learning_rate": 1.8761162952983246e-06, + "loss": 0.0109, + "step": 9170 + }, + { + "epoch": 6.253405994550409, + "grad_norm": 0.224551722407341, + "learning_rate": 1.8315108122396618e-06, + "loss": 0.011, + "step": 9180 + }, + { + "epoch": 6.260217983651226, + "grad_norm": 0.1607029289007187, + "learning_rate": 1.787432092061475e-06, + "loss": 0.0088, + "step": 9190 + }, + { + "epoch": 6.267029972752043, + "grad_norm": 0.16780927777290344, + "learning_rate": 1.743880616801602e-06, + "loss": 0.0124, + "step": 9200 + }, + { + "epoch": 6.273841961852861, + "grad_norm": 0.09691428393125534, + "learning_rate": 1.7008568627319865e-06, + "loss": 0.0088, + "step": 9210 + }, + { + "epoch": 6.2806539509536785, + "grad_norm": 0.15797632932662964, + "learning_rate": 1.6583613003535226e-06, + "loss": 0.0099, + "step": 9220 + }, + { + "epoch": 6.287465940054496, + "grad_norm": 0.09204510599374771, + "learning_rate": 1.6163943943908522e-06, + "loss": 0.0123, + "step": 9230 + }, + { + "epoch": 6.294277929155314, + "grad_norm": 0.12358927726745605, + "learning_rate": 1.5749566037873476e-06, + "loss": 0.0089, + "step": 9240 + }, + { + "epoch": 6.301089918256131, + "grad_norm": 0.06873710453510284, + "learning_rate": 1.5340483817000428e-06, + "loss": 0.011, + "step": 9250 + }, + { + "epoch": 6.307901907356948, + "grad_norm": 0.10804174095392227, + "learning_rate": 1.4936701754947101e-06, + "loss": 0.0088, + "step": 9260 + }, + { + "epoch": 6.314713896457766, + "grad_norm": 0.17293263971805573, + "learning_rate": 1.4538224267409361e-06, + "loss": 0.0105, + "step": 9270 + }, + { + "epoch": 6.321525885558583, + "grad_norm": 0.13102705776691437, + "learning_rate": 1.414505571207314e-06, + "loss": 0.0076, + "step": 9280 + }, + { + "epoch": 6.3283378746594, + "grad_norm": 0.0958796814084053, + "learning_rate": 1.3757200388566816e-06, + "loss": 0.0085, + "step": 9290 + }, + { + "epoch": 6.335149863760218, + "grad_norm": 0.12637010216712952, + "learning_rate": 1.3374662538414074e-06, + "loss": 0.0121, + "step": 9300 + }, + { + "epoch": 6.3419618528610355, + "grad_norm": 0.08003715425729752, + "learning_rate": 1.2997446344987617e-06, + "loss": 0.0088, + "step": 9310 + }, + { + "epoch": 6.348773841961853, + "grad_norm": 0.10992535948753357, + "learning_rate": 1.262555593346315e-06, + "loss": 0.0095, + "step": 9320 + }, + { + "epoch": 6.355585831062671, + "grad_norm": 0.17716050148010254, + "learning_rate": 1.2258995370774685e-06, + "loss": 0.0098, + "step": 9330 + }, + { + "epoch": 6.362397820163488, + "grad_norm": 0.16852711141109467, + "learning_rate": 1.1897768665569798e-06, + "loss": 0.0114, + "step": 9340 + }, + { + "epoch": 6.369209809264305, + "grad_norm": 0.13049398362636566, + "learning_rate": 1.1541879768165954e-06, + "loss": 0.008, + "step": 9350 + }, + { + "epoch": 6.376021798365123, + "grad_norm": 0.1294592022895813, + "learning_rate": 1.1191332570507085e-06, + "loss": 0.0079, + "step": 9360 + }, + { + "epoch": 6.38283378746594, + "grad_norm": 0.16510798037052155, + "learning_rate": 1.0846130906121132e-06, + "loss": 0.0166, + "step": 9370 + }, + { + "epoch": 6.389645776566757, + "grad_norm": 0.23402529954910278, + "learning_rate": 1.0506278550078131e-06, + "loss": 0.0098, + "step": 9380 + }, + { + "epoch": 6.396457765667575, + "grad_norm": 0.21434521675109863, + "learning_rate": 1.0171779218949185e-06, + "loss": 0.0123, + "step": 9390 + }, + { + "epoch": 6.4032697547683926, + "grad_norm": 0.13813361525535583, + "learning_rate": 9.842636570765174e-07, + "loss": 0.0087, + "step": 9400 + }, + { + "epoch": 6.41008174386921, + "grad_norm": 0.16853559017181396, + "learning_rate": 9.518854204977612e-07, + "loss": 0.0092, + "step": 9410 + }, + { + "epoch": 6.416893732970027, + "grad_norm": 0.1303388625383377, + "learning_rate": 9.200435662418349e-07, + "loss": 0.009, + "step": 9420 + }, + { + "epoch": 6.423705722070845, + "grad_norm": 0.08748972415924072, + "learning_rate": 8.887384425261658e-07, + "loss": 0.0085, + "step": 9430 + }, + { + "epoch": 6.430517711171662, + "grad_norm": 0.16070951521396637, + "learning_rate": 8.579703916985648e-07, + "loss": 0.0102, + "step": 9440 + }, + { + "epoch": 6.437329700272479, + "grad_norm": 0.145501047372818, + "learning_rate": 8.277397502335194e-07, + "loss": 0.0117, + "step": 9450 + }, + { + "epoch": 6.444141689373297, + "grad_norm": 0.08818133920431137, + "learning_rate": 7.980468487284675e-07, + "loss": 0.0082, + "step": 9460 + }, + { + "epoch": 6.450953678474114, + "grad_norm": 0.11098682135343552, + "learning_rate": 7.688920119002297e-07, + "loss": 0.008, + "step": 9470 + }, + { + "epoch": 6.4577656675749315, + "grad_norm": 0.11685679852962494, + "learning_rate": 7.402755585814269e-07, + "loss": 0.0105, + "step": 9480 + }, + { + "epoch": 6.46457765667575, + "grad_norm": 0.1768440157175064, + "learning_rate": 7.121978017170073e-07, + "loss": 0.0116, + "step": 9490 + }, + { + "epoch": 6.471389645776567, + "grad_norm": 0.13621461391448975, + "learning_rate": 6.846590483608306e-07, + "loss": 0.0094, + "step": 9500 + }, + { + "epoch": 6.478201634877384, + "grad_norm": 0.11491388827562332, + "learning_rate": 6.576595996722834e-07, + "loss": 0.0132, + "step": 9510 + }, + { + "epoch": 6.485013623978202, + "grad_norm": 0.13662555813789368, + "learning_rate": 6.311997509130141e-07, + "loss": 0.0103, + "step": 9520 + }, + { + "epoch": 6.491825613079019, + "grad_norm": 0.1329604685306549, + "learning_rate": 6.052797914436803e-07, + "loss": 0.011, + "step": 9530 + }, + { + "epoch": 6.498637602179836, + "grad_norm": 0.20086850225925446, + "learning_rate": 5.799000047208181e-07, + "loss": 0.0134, + "step": 9540 + }, + { + "epoch": 6.505449591280654, + "grad_norm": 0.12295696139335632, + "learning_rate": 5.550606682937054e-07, + "loss": 0.0111, + "step": 9550 + }, + { + "epoch": 6.512261580381471, + "grad_norm": 0.16936558485031128, + "learning_rate": 5.307620538013481e-07, + "loss": 0.0096, + "step": 9560 + }, + { + "epoch": 6.5190735694822886, + "grad_norm": 0.17347615957260132, + "learning_rate": 5.070044269694874e-07, + "loss": 0.0102, + "step": 9570 + }, + { + "epoch": 6.525885558583107, + "grad_norm": 0.1559409201145172, + "learning_rate": 4.837880476077417e-07, + "loss": 0.0101, + "step": 9580 + }, + { + "epoch": 6.532697547683924, + "grad_norm": 0.09770656377077103, + "learning_rate": 4.6111316960670835e-07, + "loss": 0.0094, + "step": 9590 + }, + { + "epoch": 6.539509536784741, + "grad_norm": 0.16713526844978333, + "learning_rate": 4.389800409352218e-07, + "loss": 0.008, + "step": 9600 + }, + { + "epoch": 6.546321525885559, + "grad_norm": 0.1877056509256363, + "learning_rate": 4.173889036376277e-07, + "loss": 0.0111, + "step": 9610 + }, + { + "epoch": 6.553133514986376, + "grad_norm": 0.11612536013126373, + "learning_rate": 3.963399938311463e-07, + "loss": 0.0116, + "step": 9620 + }, + { + "epoch": 6.559945504087193, + "grad_norm": 0.14571471512317657, + "learning_rate": 3.7583354170328545e-07, + "loss": 0.0106, + "step": 9630 + }, + { + "epoch": 6.566757493188011, + "grad_norm": 0.13956382870674133, + "learning_rate": 3.558697715093207e-07, + "loss": 0.0088, + "step": 9640 + }, + { + "epoch": 6.573569482288828, + "grad_norm": 0.21976619958877563, + "learning_rate": 3.3644890156983576e-07, + "loss": 0.0093, + "step": 9650 + }, + { + "epoch": 6.580381471389646, + "grad_norm": 0.1778477430343628, + "learning_rate": 3.175711442683638e-07, + "loss": 0.0108, + "step": 9660 + }, + { + "epoch": 6.587193460490463, + "grad_norm": 0.07059402763843536, + "learning_rate": 2.9923670604902197e-07, + "loss": 0.0084, + "step": 9670 + }, + { + "epoch": 6.594005449591281, + "grad_norm": 0.2106330245733261, + "learning_rate": 2.814457874143028e-07, + "loss": 0.0122, + "step": 9680 + }, + { + "epoch": 6.600817438692098, + "grad_norm": 0.07935508340597153, + "learning_rate": 2.641985829228366e-07, + "loss": 0.0089, + "step": 9690 + }, + { + "epoch": 6.607629427792915, + "grad_norm": 0.13969770073890686, + "learning_rate": 2.474952811872877e-07, + "loss": 0.009, + "step": 9700 + }, + { + "epoch": 6.614441416893733, + "grad_norm": 0.20040985941886902, + "learning_rate": 2.3133606487228397e-07, + "loss": 0.0109, + "step": 9710 + }, + { + "epoch": 6.62125340599455, + "grad_norm": 0.07138586044311523, + "learning_rate": 2.157211106924295e-07, + "loss": 0.0085, + "step": 9720 + }, + { + "epoch": 6.628065395095367, + "grad_norm": 0.16817276179790497, + "learning_rate": 2.006505894103672e-07, + "loss": 0.0101, + "step": 9730 + }, + { + "epoch": 6.6348773841961854, + "grad_norm": 0.09291006624698639, + "learning_rate": 1.8612466583489696e-07, + "loss": 0.0111, + "step": 9740 + }, + { + "epoch": 6.641689373297003, + "grad_norm": 0.172779381275177, + "learning_rate": 1.7214349881918834e-07, + "loss": 0.0094, + "step": 9750 + }, + { + "epoch": 6.64850136239782, + "grad_norm": 0.11954688280820847, + "learning_rate": 1.5870724125904845e-07, + "loss": 0.0076, + "step": 9760 + }, + { + "epoch": 6.655313351498638, + "grad_norm": 0.08615416288375854, + "learning_rate": 1.4581604009124006e-07, + "loss": 0.0093, + "step": 9770 + }, + { + "epoch": 6.662125340599455, + "grad_norm": 0.16148819029331207, + "learning_rate": 1.334700362918717e-07, + "loss": 0.0108, + "step": 9780 + }, + { + "epoch": 6.668937329700272, + "grad_norm": 0.2831004559993744, + "learning_rate": 1.2166936487486015e-07, + "loss": 0.0096, + "step": 9790 + }, + { + "epoch": 6.67574931880109, + "grad_norm": 0.16289672255516052, + "learning_rate": 1.1041415489045914e-07, + "loss": 0.0086, + "step": 9800 + }, + { + "epoch": 6.682561307901907, + "grad_norm": 0.1074691042304039, + "learning_rate": 9.970452942384412e-08, + "loss": 0.0092, + "step": 9810 + }, + { + "epoch": 6.689373297002724, + "grad_norm": 0.11844773590564728, + "learning_rate": 8.954060559375754e-08, + "loss": 0.008, + "step": 9820 + }, + { + "epoch": 6.6961852861035425, + "grad_norm": 0.15761059522628784, + "learning_rate": 7.99224945512489e-08, + "loss": 0.0133, + "step": 9830 + }, + { + "epoch": 6.70299727520436, + "grad_norm": 0.12731629610061646, + "learning_rate": 7.085030147843675e-08, + "loss": 0.0118, + "step": 9840 + }, + { + "epoch": 6.709809264305177, + "grad_norm": 0.1643010675907135, + "learning_rate": 6.232412558736523e-08, + "loss": 0.0106, + "step": 9850 + }, + { + "epoch": 6.716621253405995, + "grad_norm": 0.1862114816904068, + "learning_rate": 5.434406011893822e-08, + "loss": 0.0137, + "step": 9860 + }, + { + "epoch": 6.723433242506812, + "grad_norm": 0.08426104485988617, + "learning_rate": 4.6910192341864664e-08, + "loss": 0.0068, + "step": 9870 + }, + { + "epoch": 6.730245231607629, + "grad_norm": 0.200403094291687, + "learning_rate": 4.0022603551737035e-08, + "loss": 0.0103, + "step": 9880 + }, + { + "epoch": 6.737057220708447, + "grad_norm": 0.17721953988075256, + "learning_rate": 3.3681369070120985e-08, + "loss": 0.012, + "step": 9890 + }, + { + "epoch": 6.743869209809264, + "grad_norm": 0.14279387891292572, + "learning_rate": 2.7886558243744866e-08, + "loss": 0.011, + "step": 9900 + }, + { + "epoch": 6.7506811989100814, + "grad_norm": 0.116048663854599, + "learning_rate": 2.2638234443722596e-08, + "loss": 0.0101, + "step": 9910 + }, + { + "epoch": 6.7574931880108995, + "grad_norm": 0.10793479532003403, + "learning_rate": 1.7936455064887504e-08, + "loss": 0.0137, + "step": 9920 + }, + { + "epoch": 6.764305177111717, + "grad_norm": 0.07783685624599457, + "learning_rate": 1.378127152514841e-08, + "loss": 0.0065, + "step": 9930 + }, + { + "epoch": 6.771117166212534, + "grad_norm": 0.11434297263622284, + "learning_rate": 1.0172729264917857e-08, + "loss": 0.0116, + "step": 9940 + }, + { + "epoch": 6.777929155313352, + "grad_norm": 0.11163736134767532, + "learning_rate": 7.1108677466458215e-09, + "loss": 0.0101, + "step": 9950 + }, + { + "epoch": 6.784741144414169, + "grad_norm": 0.09405868500471115, + "learning_rate": 4.595720454353414e-09, + "loss": 0.0103, + "step": 9960 + }, + { + "epoch": 6.791553133514986, + "grad_norm": 0.16417264938354492, + "learning_rate": 2.627314893294264e-09, + "loss": 0.0075, + "step": 9970 + }, + { + "epoch": 6.798365122615804, + "grad_norm": 0.23998740315437317, + "learning_rate": 1.2056725896270048e-09, + "loss": 0.0121, + "step": 9980 + }, + { + "epoch": 6.805177111716621, + "grad_norm": 0.1281997561454773, + "learning_rate": 3.308090902098826e-10, + "loss": 0.01, + "step": 9990 + }, + { + "epoch": 6.8119891008174385, + "grad_norm": 0.12395822256803513, + "learning_rate": 2.7339624120159555e-12, + "loss": 0.0114, + "step": 10000 + } + ], + "logging_steps": 10, + "max_steps": 10000, + "num_input_tokens_seen": 0, + "num_train_epochs": 7, + "save_steps": 1000, + "stateful_callbacks": { + "TrainerControl": { + "args": { + "should_epoch_stop": false, + "should_evaluate": false, + "should_log": false, + "should_save": true, + "should_training_stop": true + }, + "attributes": {} + } + }, + "total_flos": 0.0, + "train_batch_size": 32, + "trial_name": null, + "trial_params": null +}