diff --git "a/trainer_state.json" "b/trainer_state.json" new file mode 100644--- /dev/null +++ "b/trainer_state.json" @@ -0,0 +1,7043 @@ +{ + "best_global_step": null, + "best_metric": null, + "best_model_checkpoint": null, + "epoch": 14.513788098693759, + "eval_steps": 500, + "global_step": 10000, + "is_hyper_param_search": false, + "is_local_process_zero": true, + "is_world_process_zero": true, + "log_history": [ + { + "epoch": 0.01451378809869376, + "grad_norm": 14.213541984558105, + "learning_rate": 1.8e-06, + "loss": 1.2305, + "step": 10 + }, + { + "epoch": 0.02902757619738752, + "grad_norm": 4.419776439666748, + "learning_rate": 3.8e-06, + "loss": 1.009, + "step": 20 + }, + { + "epoch": 0.04354136429608128, + "grad_norm": 3.8196420669555664, + "learning_rate": 5.8e-06, + "loss": 0.4996, + "step": 30 + }, + { + "epoch": 0.05805515239477504, + "grad_norm": 2.3731088638305664, + "learning_rate": 7.8e-06, + "loss": 0.3283, + "step": 40 + }, + { + "epoch": 0.07256894049346879, + "grad_norm": 1.8222453594207764, + "learning_rate": 9.800000000000001e-06, + "loss": 0.2276, + "step": 50 + }, + { + "epoch": 0.08708272859216255, + "grad_norm": 1.4704478979110718, + "learning_rate": 1.18e-05, + "loss": 0.1767, + "step": 60 + }, + { + "epoch": 0.10159651669085631, + "grad_norm": 1.6011147499084473, + "learning_rate": 1.3800000000000002e-05, + "loss": 0.1319, + "step": 70 + }, + { + "epoch": 0.11611030478955008, + "grad_norm": 1.1482045650482178, + "learning_rate": 1.58e-05, + "loss": 0.1314, + "step": 80 + }, + { + "epoch": 0.13062409288824384, + "grad_norm": 1.338651180267334, + "learning_rate": 1.78e-05, + "loss": 0.1104, + "step": 90 + }, + { + "epoch": 0.14513788098693758, + "grad_norm": 1.518242597579956, + "learning_rate": 1.9800000000000004e-05, + "loss": 0.1031, + "step": 100 + }, + { + "epoch": 0.15965166908563136, + "grad_norm": 1.1627987623214722, + "learning_rate": 2.18e-05, + "loss": 0.1126, + "step": 110 + }, + { + "epoch": 0.1741654571843251, + "grad_norm": 1.1839971542358398, + "learning_rate": 2.38e-05, + "loss": 0.0905, + "step": 120 + }, + { + "epoch": 0.18867924528301888, + "grad_norm": 1.1485743522644043, + "learning_rate": 2.58e-05, + "loss": 0.0947, + "step": 130 + }, + { + "epoch": 0.20319303338171263, + "grad_norm": 1.1757103204727173, + "learning_rate": 2.7800000000000005e-05, + "loss": 0.0838, + "step": 140 + }, + { + "epoch": 0.21770682148040638, + "grad_norm": 1.1539831161499023, + "learning_rate": 2.98e-05, + "loss": 0.089, + "step": 150 + }, + { + "epoch": 0.23222060957910015, + "grad_norm": 0.964296817779541, + "learning_rate": 3.18e-05, + "loss": 0.0847, + "step": 160 + }, + { + "epoch": 0.2467343976777939, + "grad_norm": 0.7617726922035217, + "learning_rate": 3.38e-05, + "loss": 0.0776, + "step": 170 + }, + { + "epoch": 0.2612481857764877, + "grad_norm": 1.6769593954086304, + "learning_rate": 3.58e-05, + "loss": 0.0807, + "step": 180 + }, + { + "epoch": 0.2757619738751814, + "grad_norm": 0.9172137379646301, + "learning_rate": 3.7800000000000004e-05, + "loss": 0.0695, + "step": 190 + }, + { + "epoch": 0.29027576197387517, + "grad_norm": 0.9198958873748779, + "learning_rate": 3.9800000000000005e-05, + "loss": 0.0658, + "step": 200 + }, + { + "epoch": 0.3047895500725689, + "grad_norm": 1.0765122175216675, + "learning_rate": 4.18e-05, + "loss": 0.079, + "step": 210 + }, + { + "epoch": 0.3193033381712627, + "grad_norm": 0.8167092800140381, + "learning_rate": 4.38e-05, + "loss": 0.0737, + "step": 220 + }, + { + "epoch": 0.33381712626995647, + "grad_norm": 0.6428409814834595, + "learning_rate": 4.58e-05, + "loss": 0.0728, + "step": 230 + }, + { + "epoch": 0.3483309143686502, + "grad_norm": 0.729390025138855, + "learning_rate": 4.78e-05, + "loss": 0.067, + "step": 240 + }, + { + "epoch": 0.36284470246734396, + "grad_norm": 0.7076684832572937, + "learning_rate": 4.9800000000000004e-05, + "loss": 0.0613, + "step": 250 + }, + { + "epoch": 0.37735849056603776, + "grad_norm": 0.7266968488693237, + "learning_rate": 5.1800000000000005e-05, + "loss": 0.06, + "step": 260 + }, + { + "epoch": 0.3918722786647315, + "grad_norm": 1.659942865371704, + "learning_rate": 5.380000000000001e-05, + "loss": 0.0734, + "step": 270 + }, + { + "epoch": 0.40638606676342526, + "grad_norm": 0.7579836249351501, + "learning_rate": 5.580000000000001e-05, + "loss": 0.063, + "step": 280 + }, + { + "epoch": 0.420899854862119, + "grad_norm": 0.6825695037841797, + "learning_rate": 5.7799999999999995e-05, + "loss": 0.0663, + "step": 290 + }, + { + "epoch": 0.43541364296081275, + "grad_norm": 0.7830653786659241, + "learning_rate": 5.9800000000000003e-05, + "loss": 0.0585, + "step": 300 + }, + { + "epoch": 0.44992743105950656, + "grad_norm": 1.039825201034546, + "learning_rate": 6.18e-05, + "loss": 0.068, + "step": 310 + }, + { + "epoch": 0.4644412191582003, + "grad_norm": 1.335789680480957, + "learning_rate": 6.38e-05, + "loss": 0.0625, + "step": 320 + }, + { + "epoch": 0.47895500725689405, + "grad_norm": 0.9468001127243042, + "learning_rate": 6.58e-05, + "loss": 0.06, + "step": 330 + }, + { + "epoch": 0.4934687953555878, + "grad_norm": 0.5654807090759277, + "learning_rate": 6.780000000000001e-05, + "loss": 0.0568, + "step": 340 + }, + { + "epoch": 0.5079825834542816, + "grad_norm": 0.9686242341995239, + "learning_rate": 6.98e-05, + "loss": 0.05, + "step": 350 + }, + { + "epoch": 0.5224963715529753, + "grad_norm": 1.011465072631836, + "learning_rate": 7.18e-05, + "loss": 0.0561, + "step": 360 + }, + { + "epoch": 0.5370101596516691, + "grad_norm": 1.286772608757019, + "learning_rate": 7.38e-05, + "loss": 0.0593, + "step": 370 + }, + { + "epoch": 0.5515239477503628, + "grad_norm": 1.3067666292190552, + "learning_rate": 7.58e-05, + "loss": 0.0527, + "step": 380 + }, + { + "epoch": 0.5660377358490566, + "grad_norm": 0.5086884498596191, + "learning_rate": 7.780000000000001e-05, + "loss": 0.0557, + "step": 390 + }, + { + "epoch": 0.5805515239477503, + "grad_norm": 0.6315854787826538, + "learning_rate": 7.98e-05, + "loss": 0.05, + "step": 400 + }, + { + "epoch": 0.5950653120464441, + "grad_norm": 0.7304733395576477, + "learning_rate": 8.18e-05, + "loss": 0.055, + "step": 410 + }, + { + "epoch": 0.6095791001451378, + "grad_norm": 0.6166198253631592, + "learning_rate": 8.38e-05, + "loss": 0.0521, + "step": 420 + }, + { + "epoch": 0.6240928882438317, + "grad_norm": 1.6033284664154053, + "learning_rate": 8.58e-05, + "loss": 0.0564, + "step": 430 + }, + { + "epoch": 0.6386066763425254, + "grad_norm": 0.6173003315925598, + "learning_rate": 8.78e-05, + "loss": 0.0523, + "step": 440 + }, + { + "epoch": 0.6531204644412192, + "grad_norm": 0.6383471488952637, + "learning_rate": 8.98e-05, + "loss": 0.0472, + "step": 450 + }, + { + "epoch": 0.6676342525399129, + "grad_norm": 0.7001596689224243, + "learning_rate": 9.180000000000001e-05, + "loss": 0.0513, + "step": 460 + }, + { + "epoch": 0.6821480406386067, + "grad_norm": 0.9356005787849426, + "learning_rate": 9.38e-05, + "loss": 0.0523, + "step": 470 + }, + { + "epoch": 0.6966618287373004, + "grad_norm": 0.7827327251434326, + "learning_rate": 9.58e-05, + "loss": 0.0519, + "step": 480 + }, + { + "epoch": 0.7111756168359942, + "grad_norm": 0.5247502326965332, + "learning_rate": 9.78e-05, + "loss": 0.0486, + "step": 490 + }, + { + "epoch": 0.7256894049346879, + "grad_norm": 0.7967197299003601, + "learning_rate": 9.98e-05, + "loss": 0.0501, + "step": 500 + }, + { + "epoch": 0.7402031930333817, + "grad_norm": 0.8294461369514465, + "learning_rate": 9.9999778549206e-05, + "loss": 0.0521, + "step": 510 + }, + { + "epoch": 0.7547169811320755, + "grad_norm": 0.6572040319442749, + "learning_rate": 9.999901304280685e-05, + "loss": 0.046, + "step": 520 + }, + { + "epoch": 0.7692307692307693, + "grad_norm": 0.6280943155288696, + "learning_rate": 9.999770075521164e-05, + "loss": 0.0581, + "step": 530 + }, + { + "epoch": 0.783744557329463, + "grad_norm": 0.726769745349884, + "learning_rate": 9.99958417007713e-05, + "loss": 0.0456, + "step": 540 + }, + { + "epoch": 0.7982583454281568, + "grad_norm": 0.5780419707298279, + "learning_rate": 9.999343589981615e-05, + "loss": 0.0427, + "step": 550 + }, + { + "epoch": 0.8127721335268505, + "grad_norm": 0.6148759126663208, + "learning_rate": 9.999048337865568e-05, + "loss": 0.0468, + "step": 560 + }, + { + "epoch": 0.8272859216255443, + "grad_norm": 0.6938358545303345, + "learning_rate": 9.998698416957815e-05, + "loss": 0.0438, + "step": 570 + }, + { + "epoch": 0.841799709724238, + "grad_norm": 0.8900749087333679, + "learning_rate": 9.998293831085037e-05, + "loss": 0.0436, + "step": 580 + }, + { + "epoch": 0.8563134978229318, + "grad_norm": 0.5300265550613403, + "learning_rate": 9.997834584671719e-05, + "loss": 0.0449, + "step": 590 + }, + { + "epoch": 0.8708272859216255, + "grad_norm": 0.7964539527893066, + "learning_rate": 9.997320682740107e-05, + "loss": 0.0461, + "step": 600 + }, + { + "epoch": 0.8853410740203193, + "grad_norm": 0.5969663262367249, + "learning_rate": 9.996752130910149e-05, + "loss": 0.0527, + "step": 610 + }, + { + "epoch": 0.8998548621190131, + "grad_norm": 0.49457740783691406, + "learning_rate": 9.99612893539944e-05, + "loss": 0.0411, + "step": 620 + }, + { + "epoch": 0.9143686502177069, + "grad_norm": 0.5567818880081177, + "learning_rate": 9.995451103023144e-05, + "loss": 0.0354, + "step": 630 + }, + { + "epoch": 0.9288824383164006, + "grad_norm": 0.7749472856521606, + "learning_rate": 9.994718641193928e-05, + "loss": 0.0448, + "step": 640 + }, + { + "epoch": 0.9433962264150944, + "grad_norm": 0.628334105014801, + "learning_rate": 9.993931557921874e-05, + "loss": 0.0467, + "step": 650 + }, + { + "epoch": 0.9579100145137881, + "grad_norm": 0.4295271039009094, + "learning_rate": 9.993089861814402e-05, + "loss": 0.0401, + "step": 660 + }, + { + "epoch": 0.9724238026124818, + "grad_norm": 0.7749134302139282, + "learning_rate": 9.992193562076166e-05, + "loss": 0.0527, + "step": 670 + }, + { + "epoch": 0.9869375907111756, + "grad_norm": 0.5881671905517578, + "learning_rate": 9.991242668508954e-05, + "loss": 0.0442, + "step": 680 + }, + { + "epoch": 1.0014513788098693, + "grad_norm": 0.5951621532440186, + "learning_rate": 9.990237191511587e-05, + "loss": 0.0383, + "step": 690 + }, + { + "epoch": 1.0159651669085632, + "grad_norm": 0.742411196231842, + "learning_rate": 9.989177142079802e-05, + "loss": 0.0402, + "step": 700 + }, + { + "epoch": 1.0304789550072568, + "grad_norm": 0.5504558682441711, + "learning_rate": 9.988062531806126e-05, + "loss": 0.0469, + "step": 710 + }, + { + "epoch": 1.0449927431059507, + "grad_norm": 0.5569984912872314, + "learning_rate": 9.986893372879762e-05, + "loss": 0.0396, + "step": 720 + }, + { + "epoch": 1.0595065312046443, + "grad_norm": 0.7134871482849121, + "learning_rate": 9.985669678086443e-05, + "loss": 0.041, + "step": 730 + }, + { + "epoch": 1.0740203193033382, + "grad_norm": 0.4360225200653076, + "learning_rate": 9.984391460808298e-05, + "loss": 0.0343, + "step": 740 + }, + { + "epoch": 1.0885341074020318, + "grad_norm": 0.39389410614967346, + "learning_rate": 9.983058735023709e-05, + "loss": 0.0419, + "step": 750 + }, + { + "epoch": 1.1030478955007257, + "grad_norm": 0.6106541752815247, + "learning_rate": 9.98167151530715e-05, + "loss": 0.0342, + "step": 760 + }, + { + "epoch": 1.1175616835994195, + "grad_norm": 0.4434909224510193, + "learning_rate": 9.980229816829034e-05, + "loss": 0.0351, + "step": 770 + }, + { + "epoch": 1.1320754716981132, + "grad_norm": 0.5501369833946228, + "learning_rate": 9.978733655355544e-05, + "loss": 0.0363, + "step": 780 + }, + { + "epoch": 1.146589259796807, + "grad_norm": 0.5239728689193726, + "learning_rate": 9.977183047248464e-05, + "loss": 0.0336, + "step": 790 + }, + { + "epoch": 1.1611030478955007, + "grad_norm": 0.6827855706214905, + "learning_rate": 9.975578009464992e-05, + "loss": 0.0424, + "step": 800 + }, + { + "epoch": 1.1756168359941945, + "grad_norm": 0.5845821499824524, + "learning_rate": 9.97391855955757e-05, + "loss": 0.0407, + "step": 810 + }, + { + "epoch": 1.1901306240928882, + "grad_norm": 0.5849574208259583, + "learning_rate": 9.972204715673669e-05, + "loss": 0.0381, + "step": 820 + }, + { + "epoch": 1.204644412191582, + "grad_norm": 0.6191765069961548, + "learning_rate": 9.970436496555617e-05, + "loss": 0.0388, + "step": 830 + }, + { + "epoch": 1.2191582002902757, + "grad_norm": 0.45317983627319336, + "learning_rate": 9.968613921540373e-05, + "loss": 0.0389, + "step": 840 + }, + { + "epoch": 1.2336719883889695, + "grad_norm": 0.38953834772109985, + "learning_rate": 9.966737010559326e-05, + "loss": 0.051, + "step": 850 + }, + { + "epoch": 1.2481857764876634, + "grad_norm": 0.5045167803764343, + "learning_rate": 9.964805784138072e-05, + "loss": 0.0376, + "step": 860 + }, + { + "epoch": 1.262699564586357, + "grad_norm": 0.5687655806541443, + "learning_rate": 9.962820263396195e-05, + "loss": 0.042, + "step": 870 + }, + { + "epoch": 1.2772133526850509, + "grad_norm": 0.4773636758327484, + "learning_rate": 9.960780470047033e-05, + "loss": 0.0377, + "step": 880 + }, + { + "epoch": 1.2917271407837445, + "grad_norm": 0.7561026215553284, + "learning_rate": 9.958686426397437e-05, + "loss": 0.0378, + "step": 890 + }, + { + "epoch": 1.3062409288824384, + "grad_norm": 0.4537484049797058, + "learning_rate": 9.956538155347534e-05, + "loss": 0.0398, + "step": 900 + }, + { + "epoch": 1.320754716981132, + "grad_norm": 0.4279239773750305, + "learning_rate": 9.95433568039047e-05, + "loss": 0.0338, + "step": 910 + }, + { + "epoch": 1.3352685050798259, + "grad_norm": 0.6215872764587402, + "learning_rate": 9.952079025612162e-05, + "loss": 0.0397, + "step": 920 + }, + { + "epoch": 1.3497822931785195, + "grad_norm": 0.42458319664001465, + "learning_rate": 9.949768215691022e-05, + "loss": 0.0323, + "step": 930 + }, + { + "epoch": 1.3642960812772134, + "grad_norm": 0.4891989529132843, + "learning_rate": 9.9474032758977e-05, + "loss": 0.0338, + "step": 940 + }, + { + "epoch": 1.3788098693759072, + "grad_norm": 0.3364896774291992, + "learning_rate": 9.944984232094794e-05, + "loss": 0.0433, + "step": 950 + }, + { + "epoch": 1.3933236574746009, + "grad_norm": 0.34809941053390503, + "learning_rate": 9.942511110736584e-05, + "loss": 0.0306, + "step": 960 + }, + { + "epoch": 1.4078374455732947, + "grad_norm": 0.4321839511394501, + "learning_rate": 9.939983938868726e-05, + "loss": 0.0344, + "step": 970 + }, + { + "epoch": 1.4223512336719883, + "grad_norm": 0.6400780081748962, + "learning_rate": 9.93740274412797e-05, + "loss": 0.0362, + "step": 980 + }, + { + "epoch": 1.4368650217706822, + "grad_norm": 0.704547643661499, + "learning_rate": 9.934767554741846e-05, + "loss": 0.0445, + "step": 990 + }, + { + "epoch": 1.4513788098693758, + "grad_norm": 0.5916634798049927, + "learning_rate": 9.932078399528361e-05, + "loss": 0.0426, + "step": 1000 + }, + { + "epoch": 1.4658925979680697, + "grad_norm": 0.6343238353729248, + "learning_rate": 9.929335307895689e-05, + "loss": 0.0387, + "step": 1010 + }, + { + "epoch": 1.4804063860667633, + "grad_norm": 0.5901707410812378, + "learning_rate": 9.926538309841839e-05, + "loss": 0.0338, + "step": 1020 + }, + { + "epoch": 1.4949201741654572, + "grad_norm": 0.4508565366268158, + "learning_rate": 9.923687435954334e-05, + "loss": 0.0326, + "step": 1030 + }, + { + "epoch": 1.509433962264151, + "grad_norm": 0.598544716835022, + "learning_rate": 9.920782717409873e-05, + "loss": 0.0383, + "step": 1040 + }, + { + "epoch": 1.5239477503628447, + "grad_norm": 0.49338483810424805, + "learning_rate": 9.917824185973994e-05, + "loss": 0.0308, + "step": 1050 + }, + { + "epoch": 1.5384615384615383, + "grad_norm": 0.34699735045433044, + "learning_rate": 9.914811874000723e-05, + "loss": 0.0355, + "step": 1060 + }, + { + "epoch": 1.5529753265602322, + "grad_norm": 0.5518466234207153, + "learning_rate": 9.911745814432218e-05, + "loss": 0.0369, + "step": 1070 + }, + { + "epoch": 1.567489114658926, + "grad_norm": 0.7362276315689087, + "learning_rate": 9.90862604079842e-05, + "loss": 0.0342, + "step": 1080 + }, + { + "epoch": 1.58200290275762, + "grad_norm": 0.5913964509963989, + "learning_rate": 9.90545258721667e-05, + "loss": 0.0288, + "step": 1090 + }, + { + "epoch": 1.5965166908563135, + "grad_norm": 0.5925279855728149, + "learning_rate": 9.90222548839135e-05, + "loss": 0.0352, + "step": 1100 + }, + { + "epoch": 1.6110304789550072, + "grad_norm": 0.5046629905700684, + "learning_rate": 9.898944779613495e-05, + "loss": 0.033, + "step": 1110 + }, + { + "epoch": 1.625544267053701, + "grad_norm": 0.3333032727241516, + "learning_rate": 9.89561049676041e-05, + "loss": 0.0343, + "step": 1120 + }, + { + "epoch": 1.640058055152395, + "grad_norm": 0.40780317783355713, + "learning_rate": 9.89222267629528e-05, + "loss": 0.0387, + "step": 1130 + }, + { + "epoch": 1.6545718432510885, + "grad_norm": 0.44939368963241577, + "learning_rate": 9.888781355266763e-05, + "loss": 0.0358, + "step": 1140 + }, + { + "epoch": 1.6690856313497822, + "grad_norm": 0.3324603736400604, + "learning_rate": 9.885286571308598e-05, + "loss": 0.0328, + "step": 1150 + }, + { + "epoch": 1.683599419448476, + "grad_norm": 0.3965994119644165, + "learning_rate": 9.881738362639182e-05, + "loss": 0.0387, + "step": 1160 + }, + { + "epoch": 1.6981132075471699, + "grad_norm": 0.42906469106674194, + "learning_rate": 9.878136768061154e-05, + "loss": 0.0364, + "step": 1170 + }, + { + "epoch": 1.7126269956458637, + "grad_norm": 0.6143022179603577, + "learning_rate": 9.874481826960979e-05, + "loss": 0.0387, + "step": 1180 + }, + { + "epoch": 1.7271407837445574, + "grad_norm": 0.45872849225997925, + "learning_rate": 9.870773579308503e-05, + "loss": 0.0331, + "step": 1190 + }, + { + "epoch": 1.741654571843251, + "grad_norm": 0.3488849997520447, + "learning_rate": 9.867012065656533e-05, + "loss": 0.0369, + "step": 1200 + }, + { + "epoch": 1.7561683599419449, + "grad_norm": 0.3685133457183838, + "learning_rate": 9.863197327140376e-05, + "loss": 0.037, + "step": 1210 + }, + { + "epoch": 1.7706821480406387, + "grad_norm": 0.5595993995666504, + "learning_rate": 9.859329405477403e-05, + "loss": 0.0353, + "step": 1220 + }, + { + "epoch": 1.7851959361393324, + "grad_norm": 0.5769087076187134, + "learning_rate": 9.855408342966585e-05, + "loss": 0.0385, + "step": 1230 + }, + { + "epoch": 1.799709724238026, + "grad_norm": 0.4372149407863617, + "learning_rate": 9.851434182488033e-05, + "loss": 0.0368, + "step": 1240 + }, + { + "epoch": 1.8142235123367199, + "grad_norm": 0.5467684864997864, + "learning_rate": 9.84740696750253e-05, + "loss": 0.0372, + "step": 1250 + }, + { + "epoch": 1.8287373004354137, + "grad_norm": 0.35480213165283203, + "learning_rate": 9.843326742051055e-05, + "loss": 0.04, + "step": 1260 + }, + { + "epoch": 1.8432510885341074, + "grad_norm": 0.656343400478363, + "learning_rate": 9.839193550754297e-05, + "loss": 0.0429, + "step": 1270 + }, + { + "epoch": 1.8577648766328012, + "grad_norm": 0.672942578792572, + "learning_rate": 9.835007438812177e-05, + "loss": 0.039, + "step": 1280 + }, + { + "epoch": 1.8722786647314948, + "grad_norm": 0.41714638471603394, + "learning_rate": 9.830768452003341e-05, + "loss": 0.0402, + "step": 1290 + }, + { + "epoch": 1.8867924528301887, + "grad_norm": 0.44775906205177307, + "learning_rate": 9.826476636684671e-05, + "loss": 0.0332, + "step": 1300 + }, + { + "epoch": 1.9013062409288826, + "grad_norm": 0.324028342962265, + "learning_rate": 9.822132039790773e-05, + "loss": 0.0329, + "step": 1310 + }, + { + "epoch": 1.9158200290275762, + "grad_norm": 0.4828672409057617, + "learning_rate": 9.817734708833461e-05, + "loss": 0.0355, + "step": 1320 + }, + { + "epoch": 1.9303338171262698, + "grad_norm": 0.5155841708183289, + "learning_rate": 9.813284691901243e-05, + "loss": 0.0306, + "step": 1330 + }, + { + "epoch": 1.9448476052249637, + "grad_norm": 0.45022115111351013, + "learning_rate": 9.808782037658792e-05, + "loss": 0.0341, + "step": 1340 + }, + { + "epoch": 1.9593613933236576, + "grad_norm": 0.4786379933357239, + "learning_rate": 9.804226795346411e-05, + "loss": 0.0275, + "step": 1350 + }, + { + "epoch": 1.9738751814223512, + "grad_norm": 0.3658143877983093, + "learning_rate": 9.799619014779503e-05, + "loss": 0.0338, + "step": 1360 + }, + { + "epoch": 1.988388969521045, + "grad_norm": 0.7636062502861023, + "learning_rate": 9.794958746348013e-05, + "loss": 0.0286, + "step": 1370 + }, + { + "epoch": 2.0029027576197387, + "grad_norm": 0.46506908535957336, + "learning_rate": 9.790246041015896e-05, + "loss": 0.0317, + "step": 1380 + }, + { + "epoch": 2.0174165457184325, + "grad_norm": 0.3769160509109497, + "learning_rate": 9.785480950320538e-05, + "loss": 0.0298, + "step": 1390 + }, + { + "epoch": 2.0319303338171264, + "grad_norm": 0.478380411863327, + "learning_rate": 9.78066352637221e-05, + "loss": 0.032, + "step": 1400 + }, + { + "epoch": 2.04644412191582, + "grad_norm": 0.47414204478263855, + "learning_rate": 9.775793821853488e-05, + "loss": 0.0273, + "step": 1410 + }, + { + "epoch": 2.0609579100145137, + "grad_norm": 0.46745365858078003, + "learning_rate": 9.77087189001868e-05, + "loss": 0.038, + "step": 1420 + }, + { + "epoch": 2.0754716981132075, + "grad_norm": 0.3730388581752777, + "learning_rate": 9.765897784693243e-05, + "loss": 0.0311, + "step": 1430 + }, + { + "epoch": 2.0899854862119014, + "grad_norm": 0.3737656772136688, + "learning_rate": 9.760871560273197e-05, + "loss": 0.0316, + "step": 1440 + }, + { + "epoch": 2.1044992743105952, + "grad_norm": 0.22075298428535461, + "learning_rate": 9.755793271724526e-05, + "loss": 0.0302, + "step": 1450 + }, + { + "epoch": 2.1190130624092887, + "grad_norm": 0.23981618881225586, + "learning_rate": 9.750662974582584e-05, + "loss": 0.0309, + "step": 1460 + }, + { + "epoch": 2.1335268505079825, + "grad_norm": 0.4695129692554474, + "learning_rate": 9.745480724951473e-05, + "loss": 0.0301, + "step": 1470 + }, + { + "epoch": 2.1480406386066764, + "grad_norm": 0.49784743785858154, + "learning_rate": 9.740246579503447e-05, + "loss": 0.0384, + "step": 1480 + }, + { + "epoch": 2.1625544267053702, + "grad_norm": 0.4933300018310547, + "learning_rate": 9.734960595478284e-05, + "loss": 0.0339, + "step": 1490 + }, + { + "epoch": 2.1770682148040637, + "grad_norm": 0.44603821635246277, + "learning_rate": 9.729622830682657e-05, + "loss": 0.0396, + "step": 1500 + }, + { + "epoch": 2.1915820029027575, + "grad_norm": 0.31131279468536377, + "learning_rate": 9.724233343489504e-05, + "loss": 0.0308, + "step": 1510 + }, + { + "epoch": 2.2060957910014514, + "grad_norm": 0.6442385911941528, + "learning_rate": 9.718792192837396e-05, + "loss": 0.0333, + "step": 1520 + }, + { + "epoch": 2.2206095791001452, + "grad_norm": 0.46196478605270386, + "learning_rate": 9.713299438229886e-05, + "loss": 0.0318, + "step": 1530 + }, + { + "epoch": 2.235123367198839, + "grad_norm": 0.5479897856712341, + "learning_rate": 9.707755139734855e-05, + "loss": 0.037, + "step": 1540 + }, + { + "epoch": 2.2496371552975325, + "grad_norm": 0.4929387867450714, + "learning_rate": 9.702159357983866e-05, + "loss": 0.0348, + "step": 1550 + }, + { + "epoch": 2.2641509433962264, + "grad_norm": 0.5221829414367676, + "learning_rate": 9.696512154171492e-05, + "loss": 0.0337, + "step": 1560 + }, + { + "epoch": 2.27866473149492, + "grad_norm": 0.4748969078063965, + "learning_rate": 9.690813590054645e-05, + "loss": 0.0278, + "step": 1570 + }, + { + "epoch": 2.293178519593614, + "grad_norm": 0.6399966478347778, + "learning_rate": 9.685063727951914e-05, + "loss": 0.029, + "step": 1580 + }, + { + "epoch": 2.3076923076923075, + "grad_norm": 0.4024050235748291, + "learning_rate": 9.679262630742865e-05, + "loss": 0.0326, + "step": 1590 + }, + { + "epoch": 2.3222060957910013, + "grad_norm": 0.5819658637046814, + "learning_rate": 9.673410361867373e-05, + "loss": 0.0337, + "step": 1600 + }, + { + "epoch": 2.336719883889695, + "grad_norm": 0.42979446053504944, + "learning_rate": 9.667506985324909e-05, + "loss": 0.0291, + "step": 1610 + }, + { + "epoch": 2.351233671988389, + "grad_norm": 0.6599947214126587, + "learning_rate": 9.661552565673855e-05, + "loss": 0.0303, + "step": 1620 + }, + { + "epoch": 2.365747460087083, + "grad_norm": 0.341407835483551, + "learning_rate": 9.655547168030789e-05, + "loss": 0.028, + "step": 1630 + }, + { + "epoch": 2.3802612481857763, + "grad_norm": 0.5918754935264587, + "learning_rate": 9.649490858069777e-05, + "loss": 0.0345, + "step": 1640 + }, + { + "epoch": 2.39477503628447, + "grad_norm": 0.48146188259124756, + "learning_rate": 9.643383702021658e-05, + "loss": 0.0379, + "step": 1650 + }, + { + "epoch": 2.409288824383164, + "grad_norm": 0.38834288716316223, + "learning_rate": 9.637225766673307e-05, + "loss": 0.0263, + "step": 1660 + }, + { + "epoch": 2.423802612481858, + "grad_norm": 0.335544228553772, + "learning_rate": 9.631017119366922e-05, + "loss": 0.0301, + "step": 1670 + }, + { + "epoch": 2.4383164005805513, + "grad_norm": 0.4774911403656006, + "learning_rate": 9.624757827999273e-05, + "loss": 0.0344, + "step": 1680 + }, + { + "epoch": 2.452830188679245, + "grad_norm": 0.4380296766757965, + "learning_rate": 9.618447961020971e-05, + "loss": 0.0313, + "step": 1690 + }, + { + "epoch": 2.467343976777939, + "grad_norm": 0.4441572427749634, + "learning_rate": 9.612087587435707e-05, + "loss": 0.0292, + "step": 1700 + }, + { + "epoch": 2.481857764876633, + "grad_norm": 0.42703405022621155, + "learning_rate": 9.605676776799508e-05, + "loss": 0.0312, + "step": 1710 + }, + { + "epoch": 2.4963715529753268, + "grad_norm": 0.35804256796836853, + "learning_rate": 9.599215599219973e-05, + "loss": 0.0321, + "step": 1720 + }, + { + "epoch": 2.51088534107402, + "grad_norm": 0.3918311893939972, + "learning_rate": 9.592704125355505e-05, + "loss": 0.0305, + "step": 1730 + }, + { + "epoch": 2.525399129172714, + "grad_norm": 0.4040948152542114, + "learning_rate": 9.586142426414538e-05, + "loss": 0.0295, + "step": 1740 + }, + { + "epoch": 2.539912917271408, + "grad_norm": 0.38530823588371277, + "learning_rate": 9.57953057415476e-05, + "loss": 0.0308, + "step": 1750 + }, + { + "epoch": 2.5544267053701017, + "grad_norm": 0.2776036262512207, + "learning_rate": 9.572868640882328e-05, + "loss": 0.0339, + "step": 1760 + }, + { + "epoch": 2.568940493468795, + "grad_norm": 0.6316867470741272, + "learning_rate": 9.56615669945108e-05, + "loss": 0.0318, + "step": 1770 + }, + { + "epoch": 2.583454281567489, + "grad_norm": 0.320717453956604, + "learning_rate": 9.55939482326173e-05, + "loss": 0.0306, + "step": 1780 + }, + { + "epoch": 2.597968069666183, + "grad_norm": 0.340896874666214, + "learning_rate": 9.552583086261069e-05, + "loss": 0.0333, + "step": 1790 + }, + { + "epoch": 2.6124818577648767, + "grad_norm": 0.5276126861572266, + "learning_rate": 9.545721562941168e-05, + "loss": 0.0331, + "step": 1800 + }, + { + "epoch": 2.6269956458635706, + "grad_norm": 0.48871180415153503, + "learning_rate": 9.538810328338543e-05, + "loss": 0.0335, + "step": 1810 + }, + { + "epoch": 2.641509433962264, + "grad_norm": 0.46979132294654846, + "learning_rate": 9.531849458033349e-05, + "loss": 0.028, + "step": 1820 + }, + { + "epoch": 2.656023222060958, + "grad_norm": 0.641685962677002, + "learning_rate": 9.524839028148547e-05, + "loss": 0.0287, + "step": 1830 + }, + { + "epoch": 2.6705370101596517, + "grad_norm": 0.34274694323539734, + "learning_rate": 9.517779115349077e-05, + "loss": 0.0273, + "step": 1840 + }, + { + "epoch": 2.6850507982583456, + "grad_norm": 0.4624781906604767, + "learning_rate": 9.510669796841014e-05, + "loss": 0.0305, + "step": 1850 + }, + { + "epoch": 2.699564586357039, + "grad_norm": 0.35548219084739685, + "learning_rate": 9.503511150370727e-05, + "loss": 0.0306, + "step": 1860 + }, + { + "epoch": 2.714078374455733, + "grad_norm": 0.3024815022945404, + "learning_rate": 9.496303254224024e-05, + "loss": 0.0303, + "step": 1870 + }, + { + "epoch": 2.7285921625544267, + "grad_norm": 0.3785216808319092, + "learning_rate": 9.489046187225306e-05, + "loss": 0.0225, + "step": 1880 + }, + { + "epoch": 2.7431059506531206, + "grad_norm": 0.3542894721031189, + "learning_rate": 9.481740028736692e-05, + "loss": 0.0264, + "step": 1890 + }, + { + "epoch": 2.7576197387518144, + "grad_norm": 0.4077850878238678, + "learning_rate": 9.474384858657164e-05, + "loss": 0.0334, + "step": 1900 + }, + { + "epoch": 2.772133526850508, + "grad_norm": 0.46341419219970703, + "learning_rate": 9.466980757421679e-05, + "loss": 0.0318, + "step": 1910 + }, + { + "epoch": 2.7866473149492017, + "grad_norm": 0.3434034287929535, + "learning_rate": 9.459527806000305e-05, + "loss": 0.0315, + "step": 1920 + }, + { + "epoch": 2.8011611030478956, + "grad_norm": 0.29722321033477783, + "learning_rate": 9.452026085897325e-05, + "loss": 0.0317, + "step": 1930 + }, + { + "epoch": 2.8156748911465894, + "grad_norm": 0.3669002652168274, + "learning_rate": 9.444475679150348e-05, + "loss": 0.0274, + "step": 1940 + }, + { + "epoch": 2.830188679245283, + "grad_norm": 0.48074766993522644, + "learning_rate": 9.436876668329411e-05, + "loss": 0.0309, + "step": 1950 + }, + { + "epoch": 2.8447024673439767, + "grad_norm": 0.5111640095710754, + "learning_rate": 9.429229136536079e-05, + "loss": 0.0244, + "step": 1960 + }, + { + "epoch": 2.8592162554426706, + "grad_norm": 0.5351930856704712, + "learning_rate": 9.421533167402534e-05, + "loss": 0.031, + "step": 1970 + }, + { + "epoch": 2.8737300435413644, + "grad_norm": 0.4649085998535156, + "learning_rate": 9.413788845090666e-05, + "loss": 0.0366, + "step": 1980 + }, + { + "epoch": 2.8882438316400583, + "grad_norm": 0.40677565336227417, + "learning_rate": 9.405996254291136e-05, + "loss": 0.0291, + "step": 1990 + }, + { + "epoch": 2.9027576197387517, + "grad_norm": 0.42469924688339233, + "learning_rate": 9.398155480222474e-05, + "loss": 0.026, + "step": 2000 + }, + { + "epoch": 2.9172714078374455, + "grad_norm": 0.7316426634788513, + "learning_rate": 9.390266608630128e-05, + "loss": 0.0284, + "step": 2010 + }, + { + "epoch": 2.9317851959361394, + "grad_norm": 0.6093801856040955, + "learning_rate": 9.38232972578553e-05, + "loss": 0.0333, + "step": 2020 + }, + { + "epoch": 2.9462989840348333, + "grad_norm": 0.29525354504585266, + "learning_rate": 9.374344918485164e-05, + "loss": 0.0341, + "step": 2030 + }, + { + "epoch": 2.9608127721335267, + "grad_norm": 0.5076797604560852, + "learning_rate": 9.366312274049602e-05, + "loss": 0.036, + "step": 2040 + }, + { + "epoch": 2.9753265602322205, + "grad_norm": 0.49815595149993896, + "learning_rate": 9.358231880322554e-05, + "loss": 0.0289, + "step": 2050 + }, + { + "epoch": 2.9898403483309144, + "grad_norm": 0.3962074816226959, + "learning_rate": 9.350103825669916e-05, + "loss": 0.0278, + "step": 2060 + }, + { + "epoch": 3.0043541364296082, + "grad_norm": 0.39255616068840027, + "learning_rate": 9.341928198978787e-05, + "loss": 0.0292, + "step": 2070 + }, + { + "epoch": 3.018867924528302, + "grad_norm": 0.4786067306995392, + "learning_rate": 9.333705089656512e-05, + "loss": 0.0265, + "step": 2080 + }, + { + "epoch": 3.0333817126269955, + "grad_norm": 0.46913760900497437, + "learning_rate": 9.325434587629698e-05, + "loss": 0.0267, + "step": 2090 + }, + { + "epoch": 3.0478955007256894, + "grad_norm": 0.3369215428829193, + "learning_rate": 9.31711678334323e-05, + "loss": 0.0251, + "step": 2100 + }, + { + "epoch": 3.0624092888243832, + "grad_norm": 0.3516682982444763, + "learning_rate": 9.308751767759282e-05, + "loss": 0.0265, + "step": 2110 + }, + { + "epoch": 3.076923076923077, + "grad_norm": 0.4588831663131714, + "learning_rate": 9.300339632356325e-05, + "loss": 0.0223, + "step": 2120 + }, + { + "epoch": 3.0914368650217705, + "grad_norm": 0.4561666250228882, + "learning_rate": 9.291880469128124e-05, + "loss": 0.0275, + "step": 2130 + }, + { + "epoch": 3.1059506531204644, + "grad_norm": 0.2881599962711334, + "learning_rate": 9.283374370582732e-05, + "loss": 0.029, + "step": 2140 + }, + { + "epoch": 3.1204644412191582, + "grad_norm": 0.5063420534133911, + "learning_rate": 9.274821429741482e-05, + "loss": 0.029, + "step": 2150 + }, + { + "epoch": 3.134978229317852, + "grad_norm": 0.3066236674785614, + "learning_rate": 9.266221740137961e-05, + "loss": 0.0248, + "step": 2160 + }, + { + "epoch": 3.149492017416546, + "grad_norm": 0.37477317452430725, + "learning_rate": 9.257575395817001e-05, + "loss": 0.0226, + "step": 2170 + }, + { + "epoch": 3.1640058055152394, + "grad_norm": 0.2788044810295105, + "learning_rate": 9.248882491333637e-05, + "loss": 0.027, + "step": 2180 + }, + { + "epoch": 3.178519593613933, + "grad_norm": 0.2919023931026459, + "learning_rate": 9.240143121752076e-05, + "loss": 0.0266, + "step": 2190 + }, + { + "epoch": 3.193033381712627, + "grad_norm": 0.35536232590675354, + "learning_rate": 9.23135738264467e-05, + "loss": 0.0258, + "step": 2200 + }, + { + "epoch": 3.207547169811321, + "grad_norm": 0.38751038908958435, + "learning_rate": 9.222525370090849e-05, + "loss": 0.0254, + "step": 2210 + }, + { + "epoch": 3.2220609579100143, + "grad_norm": 0.4942574203014374, + "learning_rate": 9.213647180676088e-05, + "loss": 0.0261, + "step": 2220 + }, + { + "epoch": 3.236574746008708, + "grad_norm": 0.44803738594055176, + "learning_rate": 9.204722911490846e-05, + "loss": 0.0279, + "step": 2230 + }, + { + "epoch": 3.251088534107402, + "grad_norm": 0.29596760869026184, + "learning_rate": 9.1957526601295e-05, + "loss": 0.0258, + "step": 2240 + }, + { + "epoch": 3.265602322206096, + "grad_norm": 0.5586980581283569, + "learning_rate": 9.186736524689281e-05, + "loss": 0.0305, + "step": 2250 + }, + { + "epoch": 3.28011611030479, + "grad_norm": 0.3749392628669739, + "learning_rate": 9.177674603769204e-05, + "loss": 0.027, + "step": 2260 + }, + { + "epoch": 3.294629898403483, + "grad_norm": 0.400520384311676, + "learning_rate": 9.168566996468983e-05, + "loss": 0.0265, + "step": 2270 + }, + { + "epoch": 3.309143686502177, + "grad_norm": 0.5216941833496094, + "learning_rate": 9.159413802387951e-05, + "loss": 0.0297, + "step": 2280 + }, + { + "epoch": 3.323657474600871, + "grad_norm": 0.3539350926876068, + "learning_rate": 9.150215121623974e-05, + "loss": 0.0274, + "step": 2290 + }, + { + "epoch": 3.3381712626995648, + "grad_norm": 0.35067984461784363, + "learning_rate": 9.140971054772349e-05, + "loss": 0.0234, + "step": 2300 + }, + { + "epoch": 3.352685050798258, + "grad_norm": 0.44816821813583374, + "learning_rate": 9.131681702924713e-05, + "loss": 0.0305, + "step": 2310 + }, + { + "epoch": 3.367198838896952, + "grad_norm": 0.5876545906066895, + "learning_rate": 9.122347167667926e-05, + "loss": 0.0257, + "step": 2320 + }, + { + "epoch": 3.381712626995646, + "grad_norm": 0.5841412544250488, + "learning_rate": 9.112967551082973e-05, + "loss": 0.0386, + "step": 2330 + }, + { + "epoch": 3.3962264150943398, + "grad_norm": 0.6112810373306274, + "learning_rate": 9.103542955743835e-05, + "loss": 0.0256, + "step": 2340 + }, + { + "epoch": 3.4107402031930336, + "grad_norm": 0.5436859130859375, + "learning_rate": 9.094073484716381e-05, + "loss": 0.0313, + "step": 2350 + }, + { + "epoch": 3.425253991291727, + "grad_norm": 0.571815550327301, + "learning_rate": 9.084559241557226e-05, + "loss": 0.026, + "step": 2360 + }, + { + "epoch": 3.439767779390421, + "grad_norm": 0.419302374124527, + "learning_rate": 9.075000330312608e-05, + "loss": 0.0251, + "step": 2370 + }, + { + "epoch": 3.4542815674891147, + "grad_norm": 0.4204467833042145, + "learning_rate": 9.065396855517253e-05, + "loss": 0.024, + "step": 2380 + }, + { + "epoch": 3.4687953555878086, + "grad_norm": 0.7779088616371155, + "learning_rate": 9.055748922193219e-05, + "loss": 0.0271, + "step": 2390 + }, + { + "epoch": 3.483309143686502, + "grad_norm": 0.5872662663459778, + "learning_rate": 9.046056635848761e-05, + "loss": 0.0326, + "step": 2400 + }, + { + "epoch": 3.497822931785196, + "grad_norm": 0.3685951828956604, + "learning_rate": 9.036320102477169e-05, + "loss": 0.0261, + "step": 2410 + }, + { + "epoch": 3.5123367198838897, + "grad_norm": 0.5283849239349365, + "learning_rate": 9.02653942855561e-05, + "loss": 0.0304, + "step": 2420 + }, + { + "epoch": 3.5268505079825836, + "grad_norm": 0.36461514234542847, + "learning_rate": 9.016714721043971e-05, + "loss": 0.0273, + "step": 2430 + }, + { + "epoch": 3.5413642960812775, + "grad_norm": 0.45517370104789734, + "learning_rate": 9.006846087383675e-05, + "loss": 0.0287, + "step": 2440 + }, + { + "epoch": 3.555878084179971, + "grad_norm": 0.39200252294540405, + "learning_rate": 8.996933635496523e-05, + "loss": 0.0281, + "step": 2450 + }, + { + "epoch": 3.5703918722786647, + "grad_norm": 0.42718786001205444, + "learning_rate": 8.986977473783498e-05, + "loss": 0.0281, + "step": 2460 + }, + { + "epoch": 3.5849056603773586, + "grad_norm": 0.44045886397361755, + "learning_rate": 8.97697771112359e-05, + "loss": 0.0252, + "step": 2470 + }, + { + "epoch": 3.599419448476052, + "grad_norm": 0.34935638308525085, + "learning_rate": 8.966934456872602e-05, + "loss": 0.0317, + "step": 2480 + }, + { + "epoch": 3.613933236574746, + "grad_norm": 0.24522118270397186, + "learning_rate": 8.95684782086195e-05, + "loss": 0.0238, + "step": 2490 + }, + { + "epoch": 3.6284470246734397, + "grad_norm": 0.431139200925827, + "learning_rate": 8.946717913397476e-05, + "loss": 0.0306, + "step": 2500 + }, + { + "epoch": 3.6429608127721336, + "grad_norm": 0.3560541868209839, + "learning_rate": 8.93654484525822e-05, + "loss": 0.0281, + "step": 2510 + }, + { + "epoch": 3.6574746008708274, + "grad_norm": 0.24421334266662598, + "learning_rate": 8.926328727695226e-05, + "loss": 0.0257, + "step": 2520 + }, + { + "epoch": 3.6719883889695213, + "grad_norm": 0.5788935422897339, + "learning_rate": 8.916069672430319e-05, + "loss": 0.0236, + "step": 2530 + }, + { + "epoch": 3.6865021770682147, + "grad_norm": 0.2798061668872833, + "learning_rate": 8.905767791654884e-05, + "loss": 0.026, + "step": 2540 + }, + { + "epoch": 3.7010159651669086, + "grad_norm": 0.4315924048423767, + "learning_rate": 8.895423198028638e-05, + "loss": 0.0305, + "step": 2550 + }, + { + "epoch": 3.7155297532656024, + "grad_norm": 0.4588277339935303, + "learning_rate": 8.885036004678402e-05, + "loss": 0.0247, + "step": 2560 + }, + { + "epoch": 3.730043541364296, + "grad_norm": 0.34424617886543274, + "learning_rate": 8.874606325196857e-05, + "loss": 0.0264, + "step": 2570 + }, + { + "epoch": 3.7445573294629897, + "grad_norm": 0.3801838755607605, + "learning_rate": 8.864134273641304e-05, + "loss": 0.031, + "step": 2580 + }, + { + "epoch": 3.7590711175616836, + "grad_norm": 0.40346506237983704, + "learning_rate": 8.853619964532427e-05, + "loss": 0.0279, + "step": 2590 + }, + { + "epoch": 3.7735849056603774, + "grad_norm": 0.49664241075515747, + "learning_rate": 8.843063512853019e-05, + "loss": 0.024, + "step": 2600 + }, + { + "epoch": 3.7880986937590713, + "grad_norm": 0.3993132710456848, + "learning_rate": 8.832465034046749e-05, + "loss": 0.0219, + "step": 2610 + }, + { + "epoch": 3.802612481857765, + "grad_norm": 0.711262583732605, + "learning_rate": 8.821824644016882e-05, + "loss": 0.0259, + "step": 2620 + }, + { + "epoch": 3.8171262699564585, + "grad_norm": 0.47636643052101135, + "learning_rate": 8.811142459125019e-05, + "loss": 0.0247, + "step": 2630 + }, + { + "epoch": 3.8316400580551524, + "grad_norm": 0.40644511580467224, + "learning_rate": 8.800418596189822e-05, + "loss": 0.0258, + "step": 2640 + }, + { + "epoch": 3.8461538461538463, + "grad_norm": 0.5355002284049988, + "learning_rate": 8.789653172485737e-05, + "loss": 0.0313, + "step": 2650 + }, + { + "epoch": 3.8606676342525397, + "grad_norm": 0.5259873867034912, + "learning_rate": 8.778846305741715e-05, + "loss": 0.0326, + "step": 2660 + }, + { + "epoch": 3.8751814223512335, + "grad_norm": 0.262129545211792, + "learning_rate": 8.767998114139918e-05, + "loss": 0.0248, + "step": 2670 + }, + { + "epoch": 3.8896952104499274, + "grad_norm": 0.4320864677429199, + "learning_rate": 8.757108716314429e-05, + "loss": 0.0333, + "step": 2680 + }, + { + "epoch": 3.9042089985486212, + "grad_norm": 0.5716630816459656, + "learning_rate": 8.746178231349962e-05, + "loss": 0.0264, + "step": 2690 + }, + { + "epoch": 3.918722786647315, + "grad_norm": 0.3053939938545227, + "learning_rate": 8.735206778780549e-05, + "loss": 0.028, + "step": 2700 + }, + { + "epoch": 3.933236574746009, + "grad_norm": 0.5133925676345825, + "learning_rate": 8.724194478588234e-05, + "loss": 0.025, + "step": 2710 + }, + { + "epoch": 3.9477503628447024, + "grad_norm": 0.2564045190811157, + "learning_rate": 8.713141451201772e-05, + "loss": 0.0266, + "step": 2720 + }, + { + "epoch": 3.9622641509433962, + "grad_norm": 0.5643720030784607, + "learning_rate": 8.702047817495295e-05, + "loss": 0.0258, + "step": 2730 + }, + { + "epoch": 3.97677793904209, + "grad_norm": 0.5254531502723694, + "learning_rate": 8.69091369878701e-05, + "loss": 0.0272, + "step": 2740 + }, + { + "epoch": 3.9912917271407835, + "grad_norm": 0.5122042894363403, + "learning_rate": 8.679739216837849e-05, + "loss": 0.0247, + "step": 2750 + }, + { + "epoch": 4.005805515239477, + "grad_norm": 0.5219559669494629, + "learning_rate": 8.66852449385016e-05, + "loss": 0.0265, + "step": 2760 + }, + { + "epoch": 4.020319303338171, + "grad_norm": 0.6693063974380493, + "learning_rate": 8.657269652466356e-05, + "loss": 0.0251, + "step": 2770 + }, + { + "epoch": 4.034833091436865, + "grad_norm": 0.5755857229232788, + "learning_rate": 8.645974815767577e-05, + "loss": 0.0259, + "step": 2780 + }, + { + "epoch": 4.049346879535559, + "grad_norm": 0.3904678225517273, + "learning_rate": 8.634640107272351e-05, + "loss": 0.0228, + "step": 2790 + }, + { + "epoch": 4.063860667634253, + "grad_norm": 0.34526199102401733, + "learning_rate": 8.623265650935234e-05, + "loss": 0.0253, + "step": 2800 + }, + { + "epoch": 4.078374455732947, + "grad_norm": 0.2532501220703125, + "learning_rate": 8.611851571145456e-05, + "loss": 0.0247, + "step": 2810 + }, + { + "epoch": 4.09288824383164, + "grad_norm": 0.3803569972515106, + "learning_rate": 8.600397992725566e-05, + "loss": 0.0225, + "step": 2820 + }, + { + "epoch": 4.1074020319303335, + "grad_norm": 0.400323748588562, + "learning_rate": 8.588905040930061e-05, + "loss": 0.024, + "step": 2830 + }, + { + "epoch": 4.121915820029027, + "grad_norm": 0.2635989189147949, + "learning_rate": 8.577372841444022e-05, + "loss": 0.0259, + "step": 2840 + }, + { + "epoch": 4.136429608127721, + "grad_norm": 0.23541207611560822, + "learning_rate": 8.565801520381736e-05, + "loss": 0.0227, + "step": 2850 + }, + { + "epoch": 4.150943396226415, + "grad_norm": 0.35015472769737244, + "learning_rate": 8.554191204285313e-05, + "loss": 0.0226, + "step": 2860 + }, + { + "epoch": 4.165457184325109, + "grad_norm": 0.4329552948474884, + "learning_rate": 8.542542020123315e-05, + "loss": 0.0218, + "step": 2870 + }, + { + "epoch": 4.179970972423803, + "grad_norm": 0.25657185912132263, + "learning_rate": 8.530854095289347e-05, + "loss": 0.0214, + "step": 2880 + }, + { + "epoch": 4.194484760522497, + "grad_norm": 0.38211116194725037, + "learning_rate": 8.519127557600688e-05, + "loss": 0.0225, + "step": 2890 + }, + { + "epoch": 4.2089985486211905, + "grad_norm": 0.29124242067337036, + "learning_rate": 8.507362535296871e-05, + "loss": 0.025, + "step": 2900 + }, + { + "epoch": 4.2235123367198835, + "grad_norm": 0.2974049746990204, + "learning_rate": 8.495559157038299e-05, + "loss": 0.0323, + "step": 2910 + }, + { + "epoch": 4.238026124818577, + "grad_norm": 0.39428725838661194, + "learning_rate": 8.483717551904823e-05, + "loss": 0.0227, + "step": 2920 + }, + { + "epoch": 4.252539912917271, + "grad_norm": 0.2436782270669937, + "learning_rate": 8.47183784939434e-05, + "loss": 0.0223, + "step": 2930 + }, + { + "epoch": 4.267053701015965, + "grad_norm": 0.46060872077941895, + "learning_rate": 8.459920179421374e-05, + "loss": 0.0229, + "step": 2940 + }, + { + "epoch": 4.281567489114659, + "grad_norm": 0.23644645512104034, + "learning_rate": 8.447964672315656e-05, + "loss": 0.0213, + "step": 2950 + }, + { + "epoch": 4.296081277213353, + "grad_norm": 0.29072698950767517, + "learning_rate": 8.435971458820692e-05, + "loss": 0.0284, + "step": 2960 + }, + { + "epoch": 4.310595065312047, + "grad_norm": 0.2731249928474426, + "learning_rate": 8.423940670092345e-05, + "loss": 0.0242, + "step": 2970 + }, + { + "epoch": 4.3251088534107405, + "grad_norm": 0.3129349946975708, + "learning_rate": 8.411872437697394e-05, + "loss": 0.0226, + "step": 2980 + }, + { + "epoch": 4.339622641509434, + "grad_norm": 0.24392832815647125, + "learning_rate": 8.399766893612096e-05, + "loss": 0.0211, + "step": 2990 + }, + { + "epoch": 4.354136429608127, + "grad_norm": 0.43392568826675415, + "learning_rate": 8.38762417022074e-05, + "loss": 0.0262, + "step": 3000 + }, + { + "epoch": 4.368650217706821, + "grad_norm": 0.42275625467300415, + "learning_rate": 8.375444400314204e-05, + "loss": 0.0262, + "step": 3010 + }, + { + "epoch": 4.383164005805515, + "grad_norm": 0.20994186401367188, + "learning_rate": 8.3632277170885e-05, + "loss": 0.0261, + "step": 3020 + }, + { + "epoch": 4.397677793904209, + "grad_norm": 0.387378990650177, + "learning_rate": 8.350974254143318e-05, + "loss": 0.0209, + "step": 3030 + }, + { + "epoch": 4.412191582002903, + "grad_norm": 0.2750110328197479, + "learning_rate": 8.338684145480566e-05, + "loss": 0.0211, + "step": 3040 + }, + { + "epoch": 4.426705370101597, + "grad_norm": 0.30191367864608765, + "learning_rate": 8.326357525502904e-05, + "loss": 0.0245, + "step": 3050 + }, + { + "epoch": 4.4412191582002905, + "grad_norm": 0.30933475494384766, + "learning_rate": 8.313994529012273e-05, + "loss": 0.0283, + "step": 3060 + }, + { + "epoch": 4.455732946298984, + "grad_norm": 0.5530753135681152, + "learning_rate": 8.301595291208422e-05, + "loss": 0.0283, + "step": 3070 + }, + { + "epoch": 4.470246734397678, + "grad_norm": 0.3524056375026703, + "learning_rate": 8.289159947687427e-05, + "loss": 0.025, + "step": 3080 + }, + { + "epoch": 4.484760522496371, + "grad_norm": 0.4409065842628479, + "learning_rate": 8.276688634440216e-05, + "loss": 0.0272, + "step": 3090 + }, + { + "epoch": 4.499274310595065, + "grad_norm": 0.5005135536193848, + "learning_rate": 8.26418148785107e-05, + "loss": 0.0278, + "step": 3100 + }, + { + "epoch": 4.513788098693759, + "grad_norm": 0.505293607711792, + "learning_rate": 8.251638644696141e-05, + "loss": 0.0293, + "step": 3110 + }, + { + "epoch": 4.528301886792453, + "grad_norm": 0.3523971438407898, + "learning_rate": 8.23906024214195e-05, + "loss": 0.0259, + "step": 3120 + }, + { + "epoch": 4.542815674891147, + "grad_norm": 0.5001657605171204, + "learning_rate": 8.226446417743897e-05, + "loss": 0.0309, + "step": 3130 + }, + { + "epoch": 4.55732946298984, + "grad_norm": 0.4748096764087677, + "learning_rate": 8.213797309444742e-05, + "loss": 0.0282, + "step": 3140 + }, + { + "epoch": 4.571843251088534, + "grad_norm": 0.3744714856147766, + "learning_rate": 8.201113055573105e-05, + "loss": 0.0271, + "step": 3150 + }, + { + "epoch": 4.586357039187228, + "grad_norm": 0.31449732184410095, + "learning_rate": 8.188393794841958e-05, + "loss": 0.0245, + "step": 3160 + }, + { + "epoch": 4.600870827285922, + "grad_norm": 0.40234753489494324, + "learning_rate": 8.175639666347094e-05, + "loss": 0.0251, + "step": 3170 + }, + { + "epoch": 4.615384615384615, + "grad_norm": 0.5385246276855469, + "learning_rate": 8.162850809565623e-05, + "loss": 0.0248, + "step": 3180 + }, + { + "epoch": 4.629898403483309, + "grad_norm": 0.2520975172519684, + "learning_rate": 8.150027364354431e-05, + "loss": 0.0259, + "step": 3190 + }, + { + "epoch": 4.644412191582003, + "grad_norm": 0.4619315564632416, + "learning_rate": 8.137169470948662e-05, + "loss": 0.0285, + "step": 3200 + }, + { + "epoch": 4.6589259796806965, + "grad_norm": 0.28696686029434204, + "learning_rate": 8.124277269960179e-05, + "loss": 0.0233, + "step": 3210 + }, + { + "epoch": 4.67343976777939, + "grad_norm": 0.42432940006256104, + "learning_rate": 8.111350902376023e-05, + "loss": 0.02, + "step": 3220 + }, + { + "epoch": 4.687953555878084, + "grad_norm": 0.24657271802425385, + "learning_rate": 8.098390509556883e-05, + "loss": 0.0237, + "step": 3230 + }, + { + "epoch": 4.702467343976778, + "grad_norm": 0.32343629002571106, + "learning_rate": 8.085396233235536e-05, + "loss": 0.0183, + "step": 3240 + }, + { + "epoch": 4.716981132075472, + "grad_norm": 0.3743324279785156, + "learning_rate": 8.072368215515306e-05, + "loss": 0.0211, + "step": 3250 + }, + { + "epoch": 4.731494920174166, + "grad_norm": 0.3751588463783264, + "learning_rate": 8.059306598868506e-05, + "loss": 0.0281, + "step": 3260 + }, + { + "epoch": 4.746008708272859, + "grad_norm": 0.3247457444667816, + "learning_rate": 8.046211526134888e-05, + "loss": 0.0213, + "step": 3270 + }, + { + "epoch": 4.760522496371553, + "grad_norm": 0.48758798837661743, + "learning_rate": 8.033083140520065e-05, + "loss": 0.0219, + "step": 3280 + }, + { + "epoch": 4.7750362844702465, + "grad_norm": 0.24156948924064636, + "learning_rate": 8.019921585593962e-05, + "loss": 0.025, + "step": 3290 + }, + { + "epoch": 4.78955007256894, + "grad_norm": 0.41426676511764526, + "learning_rate": 8.006727005289232e-05, + "loss": 0.0206, + "step": 3300 + }, + { + "epoch": 4.804063860667634, + "grad_norm": 0.2989985942840576, + "learning_rate": 7.993499543899692e-05, + "loss": 0.0244, + "step": 3310 + }, + { + "epoch": 4.818577648766328, + "grad_norm": 0.4818470776081085, + "learning_rate": 7.980239346078742e-05, + "loss": 0.0216, + "step": 3320 + }, + { + "epoch": 4.833091436865022, + "grad_norm": 0.2858457863330841, + "learning_rate": 7.966946556837778e-05, + "loss": 0.0234, + "step": 3330 + }, + { + "epoch": 4.847605224963716, + "grad_norm": 0.6771668791770935, + "learning_rate": 7.953621321544616e-05, + "loss": 0.0247, + "step": 3340 + }, + { + "epoch": 4.86211901306241, + "grad_norm": 0.3460070788860321, + "learning_rate": 7.940263785921896e-05, + "loss": 0.0271, + "step": 3350 + }, + { + "epoch": 4.876632801161103, + "grad_norm": 0.4291011393070221, + "learning_rate": 7.926874096045482e-05, + "loss": 0.0209, + "step": 3360 + }, + { + "epoch": 4.8911465892597965, + "grad_norm": 0.3237440884113312, + "learning_rate": 7.913452398342881e-05, + "loss": 0.0284, + "step": 3370 + }, + { + "epoch": 4.90566037735849, + "grad_norm": 0.42837947607040405, + "learning_rate": 7.89999883959163e-05, + "loss": 0.0261, + "step": 3380 + }, + { + "epoch": 4.920174165457184, + "grad_norm": 0.5014083385467529, + "learning_rate": 7.886513566917687e-05, + "loss": 0.0266, + "step": 3390 + }, + { + "epoch": 4.934687953555878, + "grad_norm": 0.4566521644592285, + "learning_rate": 7.872996727793838e-05, + "loss": 0.0271, + "step": 3400 + }, + { + "epoch": 4.949201741654572, + "grad_norm": 0.28386640548706055, + "learning_rate": 7.859448470038069e-05, + "loss": 0.0254, + "step": 3410 + }, + { + "epoch": 4.963715529753266, + "grad_norm": 0.44573739171028137, + "learning_rate": 7.845868941811956e-05, + "loss": 0.0247, + "step": 3420 + }, + { + "epoch": 4.97822931785196, + "grad_norm": 0.34436124563217163, + "learning_rate": 7.832258291619043e-05, + "loss": 0.0285, + "step": 3430 + }, + { + "epoch": 4.9927431059506535, + "grad_norm": 0.3336047828197479, + "learning_rate": 7.81861666830322e-05, + "loss": 0.0283, + "step": 3440 + }, + { + "epoch": 5.0072568940493465, + "grad_norm": 0.4497831165790558, + "learning_rate": 7.804944221047097e-05, + "loss": 0.0255, + "step": 3450 + }, + { + "epoch": 5.02177068214804, + "grad_norm": 0.32487431168556213, + "learning_rate": 7.791241099370364e-05, + "loss": 0.0262, + "step": 3460 + }, + { + "epoch": 5.036284470246734, + "grad_norm": 0.3548983037471771, + "learning_rate": 7.777507453128163e-05, + "loss": 0.0271, + "step": 3470 + }, + { + "epoch": 5.050798258345428, + "grad_norm": 0.25801166892051697, + "learning_rate": 7.763743432509451e-05, + "loss": 0.0296, + "step": 3480 + }, + { + "epoch": 5.065312046444122, + "grad_norm": 0.34388163685798645, + "learning_rate": 7.749949188035353e-05, + "loss": 0.0249, + "step": 3490 + }, + { + "epoch": 5.079825834542816, + "grad_norm": 0.3270496726036072, + "learning_rate": 7.736124870557516e-05, + "loss": 0.0318, + "step": 3500 + }, + { + "epoch": 5.09433962264151, + "grad_norm": 0.2899322807788849, + "learning_rate": 7.722270631256459e-05, + "loss": 0.0267, + "step": 3510 + }, + { + "epoch": 5.1088534107402035, + "grad_norm": 0.439237117767334, + "learning_rate": 7.708386621639925e-05, + "loss": 0.0241, + "step": 3520 + }, + { + "epoch": 5.123367198838897, + "grad_norm": 0.35549068450927734, + "learning_rate": 7.694472993541219e-05, + "loss": 0.0271, + "step": 3530 + }, + { + "epoch": 5.13788098693759, + "grad_norm": 0.6972693204879761, + "learning_rate": 7.680529899117547e-05, + "loss": 0.03, + "step": 3540 + }, + { + "epoch": 5.152394775036284, + "grad_norm": 0.39446932077407837, + "learning_rate": 7.666557490848358e-05, + "loss": 0.0256, + "step": 3550 + }, + { + "epoch": 5.166908563134978, + "grad_norm": 0.4070332944393158, + "learning_rate": 7.65255592153367e-05, + "loss": 0.0292, + "step": 3560 + }, + { + "epoch": 5.181422351233672, + "grad_norm": 0.3553643226623535, + "learning_rate": 7.638525344292402e-05, + "loss": 0.0329, + "step": 3570 + }, + { + "epoch": 5.195936139332366, + "grad_norm": 0.4506545662879944, + "learning_rate": 7.624465912560697e-05, + "loss": 0.0295, + "step": 3580 + }, + { + "epoch": 5.21044992743106, + "grad_norm": 0.37449729442596436, + "learning_rate": 7.610377780090249e-05, + "loss": 0.031, + "step": 3590 + }, + { + "epoch": 5.2249637155297535, + "grad_norm": 0.3218446373939514, + "learning_rate": 7.596261100946618e-05, + "loss": 0.0268, + "step": 3600 + }, + { + "epoch": 5.239477503628447, + "grad_norm": 0.321188747882843, + "learning_rate": 7.582116029507542e-05, + "loss": 0.0238, + "step": 3610 + }, + { + "epoch": 5.253991291727141, + "grad_norm": 0.27008360624313354, + "learning_rate": 7.56794272046126e-05, + "loss": 0.0234, + "step": 3620 + }, + { + "epoch": 5.268505079825834, + "grad_norm": 0.2834987938404083, + "learning_rate": 7.55374132880481e-05, + "loss": 0.0242, + "step": 3630 + }, + { + "epoch": 5.283018867924528, + "grad_norm": 0.4183512032032013, + "learning_rate": 7.539512009842333e-05, + "loss": 0.0224, + "step": 3640 + }, + { + "epoch": 5.297532656023222, + "grad_norm": 0.3059968650341034, + "learning_rate": 7.525254919183382e-05, + "loss": 0.026, + "step": 3650 + }, + { + "epoch": 5.312046444121916, + "grad_norm": 0.35460999608039856, + "learning_rate": 7.510970212741215e-05, + "loss": 0.018, + "step": 3660 + }, + { + "epoch": 5.32656023222061, + "grad_norm": 0.33158352971076965, + "learning_rate": 7.496658046731096e-05, + "loss": 0.024, + "step": 3670 + }, + { + "epoch": 5.3410740203193035, + "grad_norm": 0.36549443006515503, + "learning_rate": 7.482318577668578e-05, + "loss": 0.0284, + "step": 3680 + }, + { + "epoch": 5.355587808417997, + "grad_norm": 0.3462570309638977, + "learning_rate": 7.467951962367796e-05, + "loss": 0.0247, + "step": 3690 + }, + { + "epoch": 5.370101596516691, + "grad_norm": 0.29346224665641785, + "learning_rate": 7.453558357939755e-05, + "loss": 0.0234, + "step": 3700 + }, + { + "epoch": 5.384615384615385, + "grad_norm": 0.28282681107521057, + "learning_rate": 7.439137921790606e-05, + "loss": 0.0263, + "step": 3710 + }, + { + "epoch": 5.399129172714078, + "grad_norm": 0.29163405299186707, + "learning_rate": 7.42469081161993e-05, + "loss": 0.0233, + "step": 3720 + }, + { + "epoch": 5.413642960812772, + "grad_norm": 0.3668530285358429, + "learning_rate": 7.410217185419006e-05, + "loss": 0.0231, + "step": 3730 + }, + { + "epoch": 5.428156748911466, + "grad_norm": 0.2810617983341217, + "learning_rate": 7.395717201469095e-05, + "loss": 0.0238, + "step": 3740 + }, + { + "epoch": 5.44267053701016, + "grad_norm": 0.2794572710990906, + "learning_rate": 7.381191018339696e-05, + "loss": 0.0218, + "step": 3750 + }, + { + "epoch": 5.457184325108853, + "grad_norm": 0.3712323307991028, + "learning_rate": 7.36663879488682e-05, + "loss": 0.0228, + "step": 3760 + }, + { + "epoch": 5.471698113207547, + "grad_norm": 0.2612457275390625, + "learning_rate": 7.352060690251254e-05, + "loss": 0.0273, + "step": 3770 + }, + { + "epoch": 5.486211901306241, + "grad_norm": 0.3938305974006653, + "learning_rate": 7.337456863856811e-05, + "loss": 0.0238, + "step": 3780 + }, + { + "epoch": 5.500725689404935, + "grad_norm": 0.3045499324798584, + "learning_rate": 7.3228274754086e-05, + "loss": 0.0199, + "step": 3790 + }, + { + "epoch": 5.515239477503629, + "grad_norm": 0.3972741663455963, + "learning_rate": 7.308172684891267e-05, + "loss": 0.0246, + "step": 3800 + }, + { + "epoch": 5.529753265602322, + "grad_norm": 0.19863513112068176, + "learning_rate": 7.293492652567255e-05, + "loss": 0.023, + "step": 3810 + }, + { + "epoch": 5.544267053701016, + "grad_norm": 0.25676262378692627, + "learning_rate": 7.278787538975043e-05, + "loss": 0.0255, + "step": 3820 + }, + { + "epoch": 5.5587808417997095, + "grad_norm": 0.2811097502708435, + "learning_rate": 7.2640575049274e-05, + "loss": 0.027, + "step": 3830 + }, + { + "epoch": 5.573294629898403, + "grad_norm": 0.40765538811683655, + "learning_rate": 7.249302711509616e-05, + "loss": 0.0251, + "step": 3840 + }, + { + "epoch": 5.587808417997097, + "grad_norm": 0.29240286350250244, + "learning_rate": 7.23452332007775e-05, + "loss": 0.0236, + "step": 3850 + }, + { + "epoch": 5.602322206095791, + "grad_norm": 0.7348763346672058, + "learning_rate": 7.219719492256858e-05, + "loss": 0.0246, + "step": 3860 + }, + { + "epoch": 5.616835994194485, + "grad_norm": 0.38490915298461914, + "learning_rate": 7.20489138993923e-05, + "loss": 0.0255, + "step": 3870 + }, + { + "epoch": 5.631349782293179, + "grad_norm": 0.5244151949882507, + "learning_rate": 7.190039175282614e-05, + "loss": 0.0239, + "step": 3880 + }, + { + "epoch": 5.645863570391873, + "grad_norm": 0.2772596776485443, + "learning_rate": 7.175163010708455e-05, + "loss": 0.0233, + "step": 3890 + }, + { + "epoch": 5.660377358490566, + "grad_norm": 0.2788791358470917, + "learning_rate": 7.1602630589001e-05, + "loss": 0.0242, + "step": 3900 + }, + { + "epoch": 5.6748911465892595, + "grad_norm": 0.3012574315071106, + "learning_rate": 7.14533948280104e-05, + "loss": 0.0233, + "step": 3910 + }, + { + "epoch": 5.689404934687953, + "grad_norm": 0.3497931957244873, + "learning_rate": 7.130392445613109e-05, + "loss": 0.0222, + "step": 3920 + }, + { + "epoch": 5.703918722786647, + "grad_norm": 0.34485259652137756, + "learning_rate": 7.115422110794711e-05, + "loss": 0.0228, + "step": 3930 + }, + { + "epoch": 5.718432510885341, + "grad_norm": 0.4922613501548767, + "learning_rate": 7.100428642059033e-05, + "loss": 0.0259, + "step": 3940 + }, + { + "epoch": 5.732946298984035, + "grad_norm": 0.2872260510921478, + "learning_rate": 7.08541220337224e-05, + "loss": 0.0256, + "step": 3950 + }, + { + "epoch": 5.747460087082729, + "grad_norm": 0.4685818552970886, + "learning_rate": 7.070372958951706e-05, + "loss": 0.0249, + "step": 3960 + }, + { + "epoch": 5.761973875181423, + "grad_norm": 0.34322476387023926, + "learning_rate": 7.055311073264194e-05, + "loss": 0.0267, + "step": 3970 + }, + { + "epoch": 5.7764876632801165, + "grad_norm": 0.45358946919441223, + "learning_rate": 7.040226711024077e-05, + "loss": 0.0236, + "step": 3980 + }, + { + "epoch": 5.7910014513788095, + "grad_norm": 0.3939347565174103, + "learning_rate": 7.02512003719152e-05, + "loss": 0.0218, + "step": 3990 + }, + { + "epoch": 5.805515239477503, + "grad_norm": 0.5187222361564636, + "learning_rate": 7.00999121697069e-05, + "loss": 0.0259, + "step": 4000 + }, + { + "epoch": 5.820029027576197, + "grad_norm": 0.3332383334636688, + "learning_rate": 6.99484041580794e-05, + "loss": 0.024, + "step": 4010 + }, + { + "epoch": 5.834542815674891, + "grad_norm": 0.4175368845462799, + "learning_rate": 6.979667799390004e-05, + "loss": 0.0227, + "step": 4020 + }, + { + "epoch": 5.849056603773585, + "grad_norm": 0.3491177260875702, + "learning_rate": 6.964473533642185e-05, + "loss": 0.0266, + "step": 4030 + }, + { + "epoch": 5.863570391872279, + "grad_norm": 0.35566210746765137, + "learning_rate": 6.949257784726539e-05, + "loss": 0.023, + "step": 4040 + }, + { + "epoch": 5.878084179970973, + "grad_norm": 0.33617454767227173, + "learning_rate": 6.934020719040056e-05, + "loss": 0.0207, + "step": 4050 + }, + { + "epoch": 5.8925979680696665, + "grad_norm": 0.24412201344966888, + "learning_rate": 6.918762503212848e-05, + "loss": 0.0238, + "step": 4060 + }, + { + "epoch": 5.90711175616836, + "grad_norm": 0.36832985281944275, + "learning_rate": 6.903483304106319e-05, + "loss": 0.0213, + "step": 4070 + }, + { + "epoch": 5.921625544267053, + "grad_norm": 0.5100352168083191, + "learning_rate": 6.888183288811341e-05, + "loss": 0.0221, + "step": 4080 + }, + { + "epoch": 5.936139332365747, + "grad_norm": 0.3731399476528168, + "learning_rate": 6.87286262464643e-05, + "loss": 0.0214, + "step": 4090 + }, + { + "epoch": 5.950653120464441, + "grad_norm": 0.24896608293056488, + "learning_rate": 6.857521479155915e-05, + "loss": 0.0219, + "step": 4100 + }, + { + "epoch": 5.965166908563135, + "grad_norm": 0.39629605412483215, + "learning_rate": 6.842160020108104e-05, + "loss": 0.0238, + "step": 4110 + }, + { + "epoch": 5.979680696661829, + "grad_norm": 0.27626049518585205, + "learning_rate": 6.826778415493455e-05, + "loss": 0.0218, + "step": 4120 + }, + { + "epoch": 5.994194484760523, + "grad_norm": 0.40845057368278503, + "learning_rate": 6.811376833522729e-05, + "loss": 0.026, + "step": 4130 + }, + { + "epoch": 6.0087082728592165, + "grad_norm": 0.3360612392425537, + "learning_rate": 6.795955442625159e-05, + "loss": 0.0248, + "step": 4140 + }, + { + "epoch": 6.02322206095791, + "grad_norm": 0.3264850974082947, + "learning_rate": 6.780514411446608e-05, + "loss": 0.0245, + "step": 4150 + }, + { + "epoch": 6.037735849056604, + "grad_norm": 0.35697677731513977, + "learning_rate": 6.765053908847716e-05, + "loss": 0.0204, + "step": 4160 + }, + { + "epoch": 6.052249637155297, + "grad_norm": 0.3020981550216675, + "learning_rate": 6.749574103902064e-05, + "loss": 0.0254, + "step": 4170 + }, + { + "epoch": 6.066763425253991, + "grad_norm": 0.33380934596061707, + "learning_rate": 6.734075165894317e-05, + "loss": 0.0231, + "step": 4180 + }, + { + "epoch": 6.081277213352685, + "grad_norm": 0.4160427451133728, + "learning_rate": 6.71855726431838e-05, + "loss": 0.0289, + "step": 4190 + }, + { + "epoch": 6.095791001451379, + "grad_norm": 0.24721841514110565, + "learning_rate": 6.703020568875538e-05, + "loss": 0.0233, + "step": 4200 + }, + { + "epoch": 6.110304789550073, + "grad_norm": 0.33808133006095886, + "learning_rate": 6.687465249472603e-05, + "loss": 0.0199, + "step": 4210 + }, + { + "epoch": 6.1248185776487665, + "grad_norm": 0.26324379444122314, + "learning_rate": 6.671891476220055e-05, + "loss": 0.0212, + "step": 4220 + }, + { + "epoch": 6.13933236574746, + "grad_norm": 0.2133510708808899, + "learning_rate": 6.656299419430183e-05, + "loss": 0.018, + "step": 4230 + }, + { + "epoch": 6.153846153846154, + "grad_norm": 0.26845479011535645, + "learning_rate": 6.640689249615223e-05, + "loss": 0.0217, + "step": 4240 + }, + { + "epoch": 6.168359941944848, + "grad_norm": 0.42010000348091125, + "learning_rate": 6.625061137485491e-05, + "loss": 0.0223, + "step": 4250 + }, + { + "epoch": 6.182873730043541, + "grad_norm": 0.2624404728412628, + "learning_rate": 6.609415253947517e-05, + "loss": 0.0211, + "step": 4260 + }, + { + "epoch": 6.197387518142235, + "grad_norm": 0.8948805332183838, + "learning_rate": 6.593751770102178e-05, + "loss": 0.0193, + "step": 4270 + }, + { + "epoch": 6.211901306240929, + "grad_norm": 0.2507733702659607, + "learning_rate": 6.578070857242823e-05, + "loss": 0.023, + "step": 4280 + }, + { + "epoch": 6.226415094339623, + "grad_norm": 0.3399234116077423, + "learning_rate": 6.562372686853402e-05, + "loss": 0.021, + "step": 4290 + }, + { + "epoch": 6.2409288824383164, + "grad_norm": 0.36385804414749146, + "learning_rate": 6.546657430606593e-05, + "loss": 0.0229, + "step": 4300 + }, + { + "epoch": 6.25544267053701, + "grad_norm": 0.3022688329219818, + "learning_rate": 6.530925260361918e-05, + "loss": 0.0164, + "step": 4310 + }, + { + "epoch": 6.269956458635704, + "grad_norm": 0.2766708433628082, + "learning_rate": 6.515176348163871e-05, + "loss": 0.0254, + "step": 4320 + }, + { + "epoch": 6.284470246734398, + "grad_norm": 0.4395392835140228, + "learning_rate": 6.499410866240032e-05, + "loss": 0.0188, + "step": 4330 + }, + { + "epoch": 6.298984034833092, + "grad_norm": 0.3797242343425751, + "learning_rate": 6.48362898699919e-05, + "loss": 0.0199, + "step": 4340 + }, + { + "epoch": 6.313497822931785, + "grad_norm": 0.3334958553314209, + "learning_rate": 6.467830883029443e-05, + "loss": 0.0196, + "step": 4350 + }, + { + "epoch": 6.328011611030479, + "grad_norm": 0.4622059762477875, + "learning_rate": 6.452016727096326e-05, + "loss": 0.0188, + "step": 4360 + }, + { + "epoch": 6.342525399129173, + "grad_norm": 0.5033700466156006, + "learning_rate": 6.436186692140916e-05, + "loss": 0.0211, + "step": 4370 + }, + { + "epoch": 6.357039187227866, + "grad_norm": 0.3050404191017151, + "learning_rate": 6.420340951277938e-05, + "loss": 0.0188, + "step": 4380 + }, + { + "epoch": 6.37155297532656, + "grad_norm": 0.498782753944397, + "learning_rate": 6.404479677793874e-05, + "loss": 0.0251, + "step": 4390 + }, + { + "epoch": 6.386066763425254, + "grad_norm": 0.26728156208992004, + "learning_rate": 6.388603045145075e-05, + "loss": 0.0245, + "step": 4400 + }, + { + "epoch": 6.400580551523948, + "grad_norm": 0.2975080907344818, + "learning_rate": 6.372711226955843e-05, + "loss": 0.0234, + "step": 4410 + }, + { + "epoch": 6.415094339622642, + "grad_norm": 0.3273296654224396, + "learning_rate": 6.356804397016564e-05, + "loss": 0.0246, + "step": 4420 + }, + { + "epoch": 6.429608127721336, + "grad_norm": 0.3656662702560425, + "learning_rate": 6.340882729281779e-05, + "loss": 0.0226, + "step": 4430 + }, + { + "epoch": 6.444121915820029, + "grad_norm": 0.4351007640361786, + "learning_rate": 6.324946397868294e-05, + "loss": 0.0303, + "step": 4440 + }, + { + "epoch": 6.4586357039187225, + "grad_norm": 0.6850189566612244, + "learning_rate": 6.308995577053276e-05, + "loss": 0.0208, + "step": 4450 + }, + { + "epoch": 6.473149492017416, + "grad_norm": 0.3733491599559784, + "learning_rate": 6.293030441272347e-05, + "loss": 0.0215, + "step": 4460 + }, + { + "epoch": 6.48766328011611, + "grad_norm": 0.5074151754379272, + "learning_rate": 6.277051165117677e-05, + "loss": 0.0214, + "step": 4470 + }, + { + "epoch": 6.502177068214804, + "grad_norm": 0.23355135321617126, + "learning_rate": 6.261057923336064e-05, + "loss": 0.0188, + "step": 4480 + }, + { + "epoch": 6.516690856313498, + "grad_norm": 0.4812663495540619, + "learning_rate": 6.245050890827042e-05, + "loss": 0.0211, + "step": 4490 + }, + { + "epoch": 6.531204644412192, + "grad_norm": 0.36907634139060974, + "learning_rate": 6.229030242640952e-05, + "loss": 0.0187, + "step": 4500 + }, + { + "epoch": 6.545718432510886, + "grad_norm": 0.3760873079299927, + "learning_rate": 6.212996153977037e-05, + "loss": 0.024, + "step": 4510 + }, + { + "epoch": 6.56023222060958, + "grad_norm": 0.2866142690181732, + "learning_rate": 6.196948800181523e-05, + "loss": 0.0213, + "step": 4520 + }, + { + "epoch": 6.5747460087082725, + "grad_norm": 0.3145366609096527, + "learning_rate": 6.180888356745695e-05, + "loss": 0.0203, + "step": 4530 + }, + { + "epoch": 6.589259796806966, + "grad_norm": 0.35367974638938904, + "learning_rate": 6.164814999303995e-05, + "loss": 0.0198, + "step": 4540 + }, + { + "epoch": 6.60377358490566, + "grad_norm": 0.3168833553791046, + "learning_rate": 6.148728903632081e-05, + "loss": 0.0166, + "step": 4550 + }, + { + "epoch": 6.618287373004354, + "grad_norm": 0.3494027256965637, + "learning_rate": 6.132630245644921e-05, + "loss": 0.0178, + "step": 4560 + }, + { + "epoch": 6.632801161103048, + "grad_norm": 0.3673202693462372, + "learning_rate": 6.116519201394857e-05, + "loss": 0.0191, + "step": 4570 + }, + { + "epoch": 6.647314949201742, + "grad_norm": 0.40150192379951477, + "learning_rate": 6.10039594706969e-05, + "loss": 0.0212, + "step": 4580 + }, + { + "epoch": 6.661828737300436, + "grad_norm": 0.2226407676935196, + "learning_rate": 6.084260658990744e-05, + "loss": 0.0187, + "step": 4590 + }, + { + "epoch": 6.6763425253991295, + "grad_norm": 0.4998737573623657, + "learning_rate": 6.068113513610943e-05, + "loss": 0.0204, + "step": 4600 + }, + { + "epoch": 6.690856313497823, + "grad_norm": 0.34107232093811035, + "learning_rate": 6.0519546875128876e-05, + "loss": 0.0202, + "step": 4610 + }, + { + "epoch": 6.705370101596516, + "grad_norm": 0.2238370180130005, + "learning_rate": 6.035784357406906e-05, + "loss": 0.0225, + "step": 4620 + }, + { + "epoch": 6.71988388969521, + "grad_norm": 0.31457582116127014, + "learning_rate": 6.01960270012914e-05, + "loss": 0.0205, + "step": 4630 + }, + { + "epoch": 6.734397677793904, + "grad_norm": 0.31476977467536926, + "learning_rate": 6.003409892639599e-05, + "loss": 0.0219, + "step": 4640 + }, + { + "epoch": 6.748911465892598, + "grad_norm": 0.28887617588043213, + "learning_rate": 5.9872061120202336e-05, + "loss": 0.0213, + "step": 4650 + }, + { + "epoch": 6.763425253991292, + "grad_norm": 0.35433229804039, + "learning_rate": 5.9709915354729914e-05, + "loss": 0.0258, + "step": 4660 + }, + { + "epoch": 6.777939042089986, + "grad_norm": 0.2967251241207123, + "learning_rate": 5.9547663403178824e-05, + "loss": 0.0212, + "step": 4670 + }, + { + "epoch": 6.7924528301886795, + "grad_norm": 0.2915290594100952, + "learning_rate": 5.9385307039910445e-05, + "loss": 0.0166, + "step": 4680 + }, + { + "epoch": 6.806966618287373, + "grad_norm": 0.19597183167934418, + "learning_rate": 5.922284804042792e-05, + "loss": 0.0218, + "step": 4690 + }, + { + "epoch": 6.821480406386067, + "grad_norm": 0.33696219325065613, + "learning_rate": 5.906028818135687e-05, + "loss": 0.022, + "step": 4700 + }, + { + "epoch": 6.83599419448476, + "grad_norm": 0.45616379380226135, + "learning_rate": 5.889762924042585e-05, + "loss": 0.021, + "step": 4710 + }, + { + "epoch": 6.850507982583454, + "grad_norm": 0.4109625220298767, + "learning_rate": 5.873487299644699e-05, + "loss": 0.0228, + "step": 4720 + }, + { + "epoch": 6.865021770682148, + "grad_norm": 0.25704535841941833, + "learning_rate": 5.857202122929649e-05, + "loss": 0.0197, + "step": 4730 + }, + { + "epoch": 6.879535558780842, + "grad_norm": 0.22514711320400238, + "learning_rate": 5.840907571989518e-05, + "loss": 0.0207, + "step": 4740 + }, + { + "epoch": 6.894049346879536, + "grad_norm": 0.376716673374176, + "learning_rate": 5.824603825018904e-05, + "loss": 0.0194, + "step": 4750 + }, + { + "epoch": 6.9085631349782295, + "grad_norm": 0.25334298610687256, + "learning_rate": 5.808291060312975e-05, + "loss": 0.0187, + "step": 4760 + }, + { + "epoch": 6.923076923076923, + "grad_norm": 0.2824648916721344, + "learning_rate": 5.7919694562655083e-05, + "loss": 0.02, + "step": 4770 + }, + { + "epoch": 6.937590711175617, + "grad_norm": 0.26366230845451355, + "learning_rate": 5.775639191366954e-05, + "loss": 0.019, + "step": 4780 + }, + { + "epoch": 6.952104499274311, + "grad_norm": 0.3209632337093353, + "learning_rate": 5.75930044420247e-05, + "loss": 0.0205, + "step": 4790 + }, + { + "epoch": 6.966618287373004, + "grad_norm": 0.2609952986240387, + "learning_rate": 5.74295339344998e-05, + "loss": 0.0212, + "step": 4800 + }, + { + "epoch": 6.981132075471698, + "grad_norm": 0.24632206559181213, + "learning_rate": 5.726598217878211e-05, + "loss": 0.0181, + "step": 4810 + }, + { + "epoch": 6.995645863570392, + "grad_norm": 0.46511247754096985, + "learning_rate": 5.71023509634474e-05, + "loss": 0.0154, + "step": 4820 + }, + { + "epoch": 7.010159651669086, + "grad_norm": 0.2515959143638611, + "learning_rate": 5.693864207794049e-05, + "loss": 0.0197, + "step": 4830 + }, + { + "epoch": 7.0246734397677795, + "grad_norm": 0.33316805958747864, + "learning_rate": 5.677485731255545e-05, + "loss": 0.0293, + "step": 4840 + }, + { + "epoch": 7.039187227866473, + "grad_norm": 0.22543445229530334, + "learning_rate": 5.6610998458416296e-05, + "loss": 0.0205, + "step": 4850 + }, + { + "epoch": 7.053701015965167, + "grad_norm": 0.24690203368663788, + "learning_rate": 5.644706730745716e-05, + "loss": 0.0227, + "step": 4860 + }, + { + "epoch": 7.068214804063861, + "grad_norm": 0.761340856552124, + "learning_rate": 5.628306565240287e-05, + "loss": 0.016, + "step": 4870 + }, + { + "epoch": 7.082728592162554, + "grad_norm": 0.37898528575897217, + "learning_rate": 5.611899528674923e-05, + "loss": 0.0212, + "step": 4880 + }, + { + "epoch": 7.097242380261248, + "grad_norm": 0.337924987077713, + "learning_rate": 5.595485800474349e-05, + "loss": 0.0218, + "step": 4890 + }, + { + "epoch": 7.111756168359942, + "grad_norm": 0.3372938930988312, + "learning_rate": 5.579065560136467e-05, + "loss": 0.0202, + "step": 4900 + }, + { + "epoch": 7.126269956458636, + "grad_norm": 0.2324390411376953, + "learning_rate": 5.562638987230392e-05, + "loss": 0.0207, + "step": 4910 + }, + { + "epoch": 7.1407837445573294, + "grad_norm": 0.3148224353790283, + "learning_rate": 5.546206261394498e-05, + "loss": 0.0164, + "step": 4920 + }, + { + "epoch": 7.155297532656023, + "grad_norm": 0.3340926170349121, + "learning_rate": 5.529767562334437e-05, + "loss": 0.0209, + "step": 4930 + }, + { + "epoch": 7.169811320754717, + "grad_norm": 0.3390096127986908, + "learning_rate": 5.5133230698211926e-05, + "loss": 0.0205, + "step": 4940 + }, + { + "epoch": 7.184325108853411, + "grad_norm": 0.33176088333129883, + "learning_rate": 5.496872963689096e-05, + "loss": 0.0206, + "step": 4950 + }, + { + "epoch": 7.198838896952105, + "grad_norm": 0.376537024974823, + "learning_rate": 5.4804174238338756e-05, + "loss": 0.0161, + "step": 4960 + }, + { + "epoch": 7.213352685050798, + "grad_norm": 0.5382699370384216, + "learning_rate": 5.463956630210678e-05, + "loss": 0.0239, + "step": 4970 + }, + { + "epoch": 7.227866473149492, + "grad_norm": 0.27850306034088135, + "learning_rate": 5.4474907628321046e-05, + "loss": 0.0235, + "step": 4980 + }, + { + "epoch": 7.242380261248186, + "grad_norm": 0.3700191080570221, + "learning_rate": 5.431020001766244e-05, + "loss": 0.0189, + "step": 4990 + }, + { + "epoch": 7.256894049346879, + "grad_norm": 0.2980765104293823, + "learning_rate": 5.4145445271346986e-05, + "loss": 0.0203, + "step": 5000 + }, + { + "epoch": 7.271407837445573, + "grad_norm": 0.2181757688522339, + "learning_rate": 5.398064519110622e-05, + "loss": 0.0218, + "step": 5010 + }, + { + "epoch": 7.285921625544267, + "grad_norm": 0.3245968222618103, + "learning_rate": 5.3815801579167394e-05, + "loss": 0.0183, + "step": 5020 + }, + { + "epoch": 7.300435413642961, + "grad_norm": 0.37286046147346497, + "learning_rate": 5.365091623823382e-05, + "loss": 0.0216, + "step": 5030 + }, + { + "epoch": 7.314949201741655, + "grad_norm": 0.2531187832355499, + "learning_rate": 5.348599097146521e-05, + "loss": 0.0201, + "step": 5040 + }, + { + "epoch": 7.329462989840349, + "grad_norm": 0.20245595276355743, + "learning_rate": 5.3321027582457836e-05, + "loss": 0.0187, + "step": 5050 + }, + { + "epoch": 7.343976777939042, + "grad_norm": 0.26263755559921265, + "learning_rate": 5.315602787522491e-05, + "loss": 0.0201, + "step": 5060 + }, + { + "epoch": 7.3584905660377355, + "grad_norm": 0.35694146156311035, + "learning_rate": 5.299099365417678e-05, + "loss": 0.0162, + "step": 5070 + }, + { + "epoch": 7.373004354136429, + "grad_norm": 0.22025509178638458, + "learning_rate": 5.2825926724101236e-05, + "loss": 0.0214, + "step": 5080 + }, + { + "epoch": 7.387518142235123, + "grad_norm": 0.4140392243862152, + "learning_rate": 5.26608288901438e-05, + "loss": 0.0209, + "step": 5090 + }, + { + "epoch": 7.402031930333817, + "grad_norm": 0.48697009682655334, + "learning_rate": 5.24957019577879e-05, + "loss": 0.0228, + "step": 5100 + }, + { + "epoch": 7.416545718432511, + "grad_norm": 0.8537246584892273, + "learning_rate": 5.2330547732835266e-05, + "loss": 0.0196, + "step": 5110 + }, + { + "epoch": 7.431059506531205, + "grad_norm": 0.24250246584415436, + "learning_rate": 5.2165368021385996e-05, + "loss": 0.022, + "step": 5120 + }, + { + "epoch": 7.445573294629899, + "grad_norm": 0.18259859085083008, + "learning_rate": 5.200016462981897e-05, + "loss": 0.0178, + "step": 5130 + }, + { + "epoch": 7.4600870827285926, + "grad_norm": 0.4219144582748413, + "learning_rate": 5.1834939364772015e-05, + "loss": 0.0182, + "step": 5140 + }, + { + "epoch": 7.4746008708272855, + "grad_norm": 0.4045235216617584, + "learning_rate": 5.166969403312214e-05, + "loss": 0.0202, + "step": 5150 + }, + { + "epoch": 7.489114658925979, + "grad_norm": 0.4021075665950775, + "learning_rate": 5.1504430441965844e-05, + "loss": 0.0211, + "step": 5160 + }, + { + "epoch": 7.503628447024673, + "grad_norm": 0.2611328065395355, + "learning_rate": 5.133915039859923e-05, + "loss": 0.0163, + "step": 5170 + }, + { + "epoch": 7.518142235123367, + "grad_norm": 0.27343425154685974, + "learning_rate": 5.1173855710498444e-05, + "loss": 0.0194, + "step": 5180 + }, + { + "epoch": 7.532656023222061, + "grad_norm": 0.33671990036964417, + "learning_rate": 5.100854818529967e-05, + "loss": 0.0218, + "step": 5190 + }, + { + "epoch": 7.547169811320755, + "grad_norm": 0.29155343770980835, + "learning_rate": 5.084322963077951e-05, + "loss": 0.0211, + "step": 5200 + }, + { + "epoch": 7.561683599419449, + "grad_norm": 0.31223922967910767, + "learning_rate": 5.067790185483522e-05, + "loss": 0.0187, + "step": 5210 + }, + { + "epoch": 7.5761973875181425, + "grad_norm": 0.2598567605018616, + "learning_rate": 5.0512566665464844e-05, + "loss": 0.0194, + "step": 5220 + }, + { + "epoch": 7.5907111756168355, + "grad_norm": 0.17831863462924957, + "learning_rate": 5.034722587074755e-05, + "loss": 0.0159, + "step": 5230 + }, + { + "epoch": 7.60522496371553, + "grad_norm": 0.2589302062988281, + "learning_rate": 5.018188127882375e-05, + "loss": 0.015, + "step": 5240 + }, + { + "epoch": 7.619738751814223, + "grad_norm": 0.2471068799495697, + "learning_rate": 5.0016534697875417e-05, + "loss": 0.0144, + "step": 5250 + }, + { + "epoch": 7.634252539912917, + "grad_norm": 0.22583220899105072, + "learning_rate": 4.9851187936106294e-05, + "loss": 0.0183, + "step": 5260 + }, + { + "epoch": 7.648766328011611, + "grad_norm": 0.3973812758922577, + "learning_rate": 4.968584280172206e-05, + "loss": 0.0183, + "step": 5270 + }, + { + "epoch": 7.663280116110305, + "grad_norm": 0.25084227323532104, + "learning_rate": 4.95205011029106e-05, + "loss": 0.0176, + "step": 5280 + }, + { + "epoch": 7.677793904208999, + "grad_norm": 0.18914823234081268, + "learning_rate": 4.935516464782227e-05, + "loss": 0.0172, + "step": 5290 + }, + { + "epoch": 7.6923076923076925, + "grad_norm": 0.3941582143306732, + "learning_rate": 4.918983524455003e-05, + "loss": 0.0172, + "step": 5300 + }, + { + "epoch": 7.706821480406386, + "grad_norm": 0.24728842079639435, + "learning_rate": 4.9024514701109766e-05, + "loss": 0.018, + "step": 5310 + }, + { + "epoch": 7.721335268505079, + "grad_norm": 0.3948165774345398, + "learning_rate": 4.885920482542043e-05, + "loss": 0.0181, + "step": 5320 + }, + { + "epoch": 7.735849056603773, + "grad_norm": 0.36098092794418335, + "learning_rate": 4.869390742528438e-05, + "loss": 0.0174, + "step": 5330 + }, + { + "epoch": 7.750362844702467, + "grad_norm": 0.41060400009155273, + "learning_rate": 4.852862430836744e-05, + "loss": 0.0203, + "step": 5340 + }, + { + "epoch": 7.764876632801161, + "grad_norm": 0.4438793957233429, + "learning_rate": 4.836335728217933e-05, + "loss": 0.0186, + "step": 5350 + }, + { + "epoch": 7.779390420899855, + "grad_norm": 0.37275615334510803, + "learning_rate": 4.819810815405379e-05, + "loss": 0.0193, + "step": 5360 + }, + { + "epoch": 7.793904208998549, + "grad_norm": 0.22395363450050354, + "learning_rate": 4.803287873112877e-05, + "loss": 0.0199, + "step": 5370 + }, + { + "epoch": 7.8084179970972425, + "grad_norm": 0.2708238363265991, + "learning_rate": 4.786767082032681e-05, + "loss": 0.0195, + "step": 5380 + }, + { + "epoch": 7.822931785195936, + "grad_norm": 0.2573468089103699, + "learning_rate": 4.77024862283351e-05, + "loss": 0.0166, + "step": 5390 + }, + { + "epoch": 7.83744557329463, + "grad_norm": 0.25963079929351807, + "learning_rate": 4.753732676158593e-05, + "loss": 0.0175, + "step": 5400 + }, + { + "epoch": 7.851959361393323, + "grad_norm": 0.2234552502632141, + "learning_rate": 4.737219422623672e-05, + "loss": 0.019, + "step": 5410 + }, + { + "epoch": 7.866473149492017, + "grad_norm": 0.2664513885974884, + "learning_rate": 4.720709042815044e-05, + "loss": 0.0139, + "step": 5420 + }, + { + "epoch": 7.880986937590711, + "grad_norm": 0.266416996717453, + "learning_rate": 4.704201717287578e-05, + "loss": 0.0177, + "step": 5430 + }, + { + "epoch": 7.895500725689405, + "grad_norm": 0.2832564413547516, + "learning_rate": 4.6876976265627404e-05, + "loss": 0.0199, + "step": 5440 + }, + { + "epoch": 7.910014513788099, + "grad_norm": 0.4903218448162079, + "learning_rate": 4.671196951126626e-05, + "loss": 0.0201, + "step": 5450 + }, + { + "epoch": 7.9245283018867925, + "grad_norm": 0.3286173343658447, + "learning_rate": 4.654699871427971e-05, + "loss": 0.0164, + "step": 5460 + }, + { + "epoch": 7.939042089985486, + "grad_norm": 0.21483942866325378, + "learning_rate": 4.6382065678762034e-05, + "loss": 0.0173, + "step": 5470 + }, + { + "epoch": 7.95355587808418, + "grad_norm": 0.32403698563575745, + "learning_rate": 4.6217172208394424e-05, + "loss": 0.0168, + "step": 5480 + }, + { + "epoch": 7.968069666182874, + "grad_norm": 0.23433920741081238, + "learning_rate": 4.605232010642549e-05, + "loss": 0.0139, + "step": 5490 + }, + { + "epoch": 7.982583454281567, + "grad_norm": 0.27407896518707275, + "learning_rate": 4.588751117565142e-05, + "loss": 0.0151, + "step": 5500 + }, + { + "epoch": 7.997097242380261, + "grad_norm": 0.49123820662498474, + "learning_rate": 4.5722747218396214e-05, + "loss": 0.0184, + "step": 5510 + }, + { + "epoch": 8.011611030478955, + "grad_norm": 0.2612648904323578, + "learning_rate": 4.5558030036492194e-05, + "loss": 0.0134, + "step": 5520 + }, + { + "epoch": 8.02612481857765, + "grad_norm": 0.32117781043052673, + "learning_rate": 4.539336143125999e-05, + "loss": 0.0211, + "step": 5530 + }, + { + "epoch": 8.040638606676342, + "grad_norm": 0.23201434314250946, + "learning_rate": 4.522874320348916e-05, + "loss": 0.0167, + "step": 5540 + }, + { + "epoch": 8.055152394775035, + "grad_norm": 0.27046090364456177, + "learning_rate": 4.506417715341821e-05, + "loss": 0.0134, + "step": 5550 + }, + { + "epoch": 8.06966618287373, + "grad_norm": 0.22173450887203217, + "learning_rate": 4.489966508071511e-05, + "loss": 0.0159, + "step": 5560 + }, + { + "epoch": 8.084179970972423, + "grad_norm": 0.27308306097984314, + "learning_rate": 4.4735208784457575e-05, + "loss": 0.0156, + "step": 5570 + }, + { + "epoch": 8.098693759071118, + "grad_norm": 0.30876046419143677, + "learning_rate": 4.457081006311325e-05, + "loss": 0.0168, + "step": 5580 + }, + { + "epoch": 8.11320754716981, + "grad_norm": 0.21127845346927643, + "learning_rate": 4.440647071452027e-05, + "loss": 0.0163, + "step": 5590 + }, + { + "epoch": 8.127721335268506, + "grad_norm": 0.30931970477104187, + "learning_rate": 4.424219253586737e-05, + "loss": 0.0196, + "step": 5600 + }, + { + "epoch": 8.142235123367199, + "grad_norm": 0.3056296408176422, + "learning_rate": 4.407797732367443e-05, + "loss": 0.0197, + "step": 5610 + }, + { + "epoch": 8.156748911465893, + "grad_norm": 0.37901338934898376, + "learning_rate": 4.391382687377268e-05, + "loss": 0.0178, + "step": 5620 + }, + { + "epoch": 8.171262699564586, + "grad_norm": 0.3721649944782257, + "learning_rate": 4.374974298128512e-05, + "loss": 0.017, + "step": 5630 + }, + { + "epoch": 8.18577648766328, + "grad_norm": 0.3776259422302246, + "learning_rate": 4.358572744060699e-05, + "loss": 0.0168, + "step": 5640 + }, + { + "epoch": 8.200290275761974, + "grad_norm": 0.3932430148124695, + "learning_rate": 4.342178204538588e-05, + "loss": 0.017, + "step": 5650 + }, + { + "epoch": 8.214804063860667, + "grad_norm": 0.22925494611263275, + "learning_rate": 4.325790858850241e-05, + "loss": 0.0167, + "step": 5660 + }, + { + "epoch": 8.229317851959362, + "grad_norm": 0.39623963832855225, + "learning_rate": 4.309410886205043e-05, + "loss": 0.0169, + "step": 5670 + }, + { + "epoch": 8.243831640058055, + "grad_norm": 0.30033227801322937, + "learning_rate": 4.293038465731752e-05, + "loss": 0.0188, + "step": 5680 + }, + { + "epoch": 8.25834542815675, + "grad_norm": 0.2802962064743042, + "learning_rate": 4.276673776476533e-05, + "loss": 0.0205, + "step": 5690 + }, + { + "epoch": 8.272859216255442, + "grad_norm": 0.2293040156364441, + "learning_rate": 4.260316997401007e-05, + "loss": 0.0157, + "step": 5700 + }, + { + "epoch": 8.287373004354137, + "grad_norm": 0.3251076340675354, + "learning_rate": 4.243968307380293e-05, + "loss": 0.0187, + "step": 5710 + }, + { + "epoch": 8.30188679245283, + "grad_norm": 0.22329086065292358, + "learning_rate": 4.22762788520104e-05, + "loss": 0.0209, + "step": 5720 + }, + { + "epoch": 8.316400580551523, + "grad_norm": 0.22150585055351257, + "learning_rate": 4.211295909559491e-05, + "loss": 0.0213, + "step": 5730 + }, + { + "epoch": 8.330914368650218, + "grad_norm": 0.3931887745857239, + "learning_rate": 4.194972559059511e-05, + "loss": 0.0159, + "step": 5740 + }, + { + "epoch": 8.34542815674891, + "grad_norm": 0.3773743808269501, + "learning_rate": 4.178658012210651e-05, + "loss": 0.0182, + "step": 5750 + }, + { + "epoch": 8.359941944847606, + "grad_norm": 0.37826624512672424, + "learning_rate": 4.162352447426177e-05, + "loss": 0.0169, + "step": 5760 + }, + { + "epoch": 8.374455732946299, + "grad_norm": 0.2601827085018158, + "learning_rate": 4.146056043021135e-05, + "loss": 0.0169, + "step": 5770 + }, + { + "epoch": 8.388969521044993, + "grad_norm": 0.2511160373687744, + "learning_rate": 4.1297689772103944e-05, + "loss": 0.0217, + "step": 5780 + }, + { + "epoch": 8.403483309143686, + "grad_norm": 0.40830790996551514, + "learning_rate": 4.113491428106694e-05, + "loss": 0.0189, + "step": 5790 + }, + { + "epoch": 8.417997097242381, + "grad_norm": 0.2982620894908905, + "learning_rate": 4.0972235737187055e-05, + "loss": 0.0171, + "step": 5800 + }, + { + "epoch": 8.432510885341074, + "grad_norm": 0.2656925320625305, + "learning_rate": 4.080965591949076e-05, + "loss": 0.0146, + "step": 5810 + }, + { + "epoch": 8.447024673439767, + "grad_norm": 0.29634448885917664, + "learning_rate": 4.0647176605924924e-05, + "loss": 0.0176, + "step": 5820 + }, + { + "epoch": 8.461538461538462, + "grad_norm": 0.36893323063850403, + "learning_rate": 4.0484799573337255e-05, + "loss": 0.0165, + "step": 5830 + }, + { + "epoch": 8.476052249637155, + "grad_norm": 0.3656172454357147, + "learning_rate": 4.032252659745699e-05, + "loss": 0.0176, + "step": 5840 + }, + { + "epoch": 8.49056603773585, + "grad_norm": 0.390611857175827, + "learning_rate": 4.016035945287539e-05, + "loss": 0.0225, + "step": 5850 + }, + { + "epoch": 8.505079825834542, + "grad_norm": 0.2583737075328827, + "learning_rate": 3.999829991302635e-05, + "loss": 0.0156, + "step": 5860 + }, + { + "epoch": 8.519593613933237, + "grad_norm": 0.25926393270492554, + "learning_rate": 3.983634975016707e-05, + "loss": 0.0152, + "step": 5870 + }, + { + "epoch": 8.53410740203193, + "grad_norm": 0.30234551429748535, + "learning_rate": 3.967451073535854e-05, + "loss": 0.0232, + "step": 5880 + }, + { + "epoch": 8.548621190130625, + "grad_norm": 0.3208179771900177, + "learning_rate": 3.951278463844633e-05, + "loss": 0.0166, + "step": 5890 + }, + { + "epoch": 8.563134978229318, + "grad_norm": 0.177336186170578, + "learning_rate": 3.935117322804111e-05, + "loss": 0.0156, + "step": 5900 + }, + { + "epoch": 8.57764876632801, + "grad_norm": 0.42963215708732605, + "learning_rate": 3.918967827149938e-05, + "loss": 0.0177, + "step": 5910 + }, + { + "epoch": 8.592162554426706, + "grad_norm": 0.4124779999256134, + "learning_rate": 3.9028301534904094e-05, + "loss": 0.0196, + "step": 5920 + }, + { + "epoch": 8.606676342525398, + "grad_norm": 0.4351457953453064, + "learning_rate": 3.88670447830454e-05, + "loss": 0.0158, + "step": 5930 + }, + { + "epoch": 8.621190130624093, + "grad_norm": 0.33472874760627747, + "learning_rate": 3.870590977940132e-05, + "loss": 0.0134, + "step": 5940 + }, + { + "epoch": 8.635703918722786, + "grad_norm": 0.2622275948524475, + "learning_rate": 3.8544898286118404e-05, + "loss": 0.0137, + "step": 5950 + }, + { + "epoch": 8.650217706821481, + "grad_norm": 0.3266589045524597, + "learning_rate": 3.838401206399257e-05, + "loss": 0.0147, + "step": 5960 + }, + { + "epoch": 8.664731494920174, + "grad_norm": 0.25724735856056213, + "learning_rate": 3.822325287244975e-05, + "loss": 0.0148, + "step": 5970 + }, + { + "epoch": 8.679245283018869, + "grad_norm": 0.35596638917922974, + "learning_rate": 3.8062622469526725e-05, + "loss": 0.0171, + "step": 5980 + }, + { + "epoch": 8.693759071117562, + "grad_norm": 0.21868722140789032, + "learning_rate": 3.790212261185183e-05, + "loss": 0.0152, + "step": 5990 + }, + { + "epoch": 8.708272859216255, + "grad_norm": 0.3128092288970947, + "learning_rate": 3.7741755054625794e-05, + "loss": 0.0165, + "step": 6000 + }, + { + "epoch": 8.72278664731495, + "grad_norm": 0.32868480682373047, + "learning_rate": 3.758152155160255e-05, + "loss": 0.0132, + "step": 6010 + }, + { + "epoch": 8.737300435413642, + "grad_norm": 0.2642138600349426, + "learning_rate": 3.742142385506999e-05, + "loss": 0.0174, + "step": 6020 + }, + { + "epoch": 8.751814223512337, + "grad_norm": 0.295388787984848, + "learning_rate": 3.72614637158309e-05, + "loss": 0.0146, + "step": 6030 + }, + { + "epoch": 8.76632801161103, + "grad_norm": 0.25500035285949707, + "learning_rate": 3.710164288318371e-05, + "loss": 0.0162, + "step": 6040 + }, + { + "epoch": 8.780841799709725, + "grad_norm": 0.22290973365306854, + "learning_rate": 3.694196310490345e-05, + "loss": 0.0113, + "step": 6050 + }, + { + "epoch": 8.795355587808418, + "grad_norm": 0.19200055301189423, + "learning_rate": 3.678242612722259e-05, + "loss": 0.0144, + "step": 6060 + }, + { + "epoch": 8.809869375907112, + "grad_norm": 0.2810363173484802, + "learning_rate": 3.6623033694811953e-05, + "loss": 0.0151, + "step": 6070 + }, + { + "epoch": 8.824383164005805, + "grad_norm": 0.21887309849262238, + "learning_rate": 3.6463787550761665e-05, + "loss": 0.016, + "step": 6080 + }, + { + "epoch": 8.838896952104498, + "grad_norm": 0.20743784308433533, + "learning_rate": 3.630468943656202e-05, + "loss": 0.0145, + "step": 6090 + }, + { + "epoch": 8.853410740203193, + "grad_norm": 0.2887123227119446, + "learning_rate": 3.6145741092084523e-05, + "loss": 0.0131, + "step": 6100 + }, + { + "epoch": 8.867924528301886, + "grad_norm": 0.2920779287815094, + "learning_rate": 3.598694425556278e-05, + "loss": 0.0128, + "step": 6110 + }, + { + "epoch": 8.882438316400581, + "grad_norm": 0.1966823935508728, + "learning_rate": 3.58283006635736e-05, + "loss": 0.0172, + "step": 6120 + }, + { + "epoch": 8.896952104499274, + "grad_norm": 0.31956127285957336, + "learning_rate": 3.566981205101781e-05, + "loss": 0.0154, + "step": 6130 + }, + { + "epoch": 8.911465892597969, + "grad_norm": 0.17137199640274048, + "learning_rate": 3.5511480151101556e-05, + "loss": 0.0135, + "step": 6140 + }, + { + "epoch": 8.925979680696662, + "grad_norm": 0.4096137285232544, + "learning_rate": 3.5353306695317104e-05, + "loss": 0.0123, + "step": 6150 + }, + { + "epoch": 8.940493468795356, + "grad_norm": 0.29205936193466187, + "learning_rate": 3.519529341342402e-05, + "loss": 0.0178, + "step": 6160 + }, + { + "epoch": 8.95500725689405, + "grad_norm": 0.16332031786441803, + "learning_rate": 3.503744203343026e-05, + "loss": 0.0139, + "step": 6170 + }, + { + "epoch": 8.969521044992742, + "grad_norm": 0.2997671365737915, + "learning_rate": 3.487975428157318e-05, + "loss": 0.0149, + "step": 6180 + }, + { + "epoch": 8.984034833091437, + "grad_norm": 0.3304642140865326, + "learning_rate": 3.472223188230083e-05, + "loss": 0.0191, + "step": 6190 + }, + { + "epoch": 8.99854862119013, + "grad_norm": 0.4736999273300171, + "learning_rate": 3.4564876558252866e-05, + "loss": 0.0146, + "step": 6200 + }, + { + "epoch": 9.013062409288825, + "grad_norm": 0.24356195330619812, + "learning_rate": 3.440769003024195e-05, + "loss": 0.0143, + "step": 6210 + }, + { + "epoch": 9.027576197387518, + "grad_norm": 0.3061477243900299, + "learning_rate": 3.425067401723477e-05, + "loss": 0.0146, + "step": 6220 + }, + { + "epoch": 9.042089985486212, + "grad_norm": 0.2497696429491043, + "learning_rate": 3.409383023633325e-05, + "loss": 0.0148, + "step": 6230 + }, + { + "epoch": 9.056603773584905, + "grad_norm": 0.32089507579803467, + "learning_rate": 3.3937160402755894e-05, + "loss": 0.0148, + "step": 6240 + }, + { + "epoch": 9.0711175616836, + "grad_norm": 0.2858957052230835, + "learning_rate": 3.378066622981885e-05, + "loss": 0.0175, + "step": 6250 + }, + { + "epoch": 9.085631349782293, + "grad_norm": 0.292372465133667, + "learning_rate": 3.362434942891738e-05, + "loss": 0.016, + "step": 6260 + }, + { + "epoch": 9.100145137880986, + "grad_norm": 0.3087608516216278, + "learning_rate": 3.346821170950693e-05, + "loss": 0.0147, + "step": 6270 + }, + { + "epoch": 9.11465892597968, + "grad_norm": 0.2726084887981415, + "learning_rate": 3.3312254779084585e-05, + "loss": 0.0162, + "step": 6280 + }, + { + "epoch": 9.129172714078374, + "grad_norm": 0.25452864170074463, + "learning_rate": 3.315648034317039e-05, + "loss": 0.0188, + "step": 6290 + }, + { + "epoch": 9.143686502177069, + "grad_norm": 0.25561732053756714, + "learning_rate": 3.3000890105288564e-05, + "loss": 0.014, + "step": 6300 + }, + { + "epoch": 9.158200290275762, + "grad_norm": 0.239127516746521, + "learning_rate": 3.284548576694908e-05, + "loss": 0.0157, + "step": 6310 + }, + { + "epoch": 9.172714078374456, + "grad_norm": 0.39308828115463257, + "learning_rate": 3.2690269027628815e-05, + "loss": 0.0131, + "step": 6320 + }, + { + "epoch": 9.18722786647315, + "grad_norm": 0.35784438252449036, + "learning_rate": 3.253524158475324e-05, + "loss": 0.0146, + "step": 6330 + }, + { + "epoch": 9.201741654571844, + "grad_norm": 0.2706029713153839, + "learning_rate": 3.238040513367757e-05, + "loss": 0.0144, + "step": 6340 + }, + { + "epoch": 9.216255442670537, + "grad_norm": 0.26546719670295715, + "learning_rate": 3.222576136766843e-05, + "loss": 0.0145, + "step": 6350 + }, + { + "epoch": 9.23076923076923, + "grad_norm": 0.30209028720855713, + "learning_rate": 3.2071311977885324e-05, + "loss": 0.015, + "step": 6360 + }, + { + "epoch": 9.245283018867925, + "grad_norm": 0.6471951007843018, + "learning_rate": 3.191705865336197e-05, + "loss": 0.0172, + "step": 6370 + }, + { + "epoch": 9.259796806966618, + "grad_norm": 0.3383336067199707, + "learning_rate": 3.1763003080988075e-05, + "loss": 0.0125, + "step": 6380 + }, + { + "epoch": 9.274310595065312, + "grad_norm": 0.18578916788101196, + "learning_rate": 3.160914694549063e-05, + "loss": 0.0162, + "step": 6390 + }, + { + "epoch": 9.288824383164005, + "grad_norm": 0.34075552225112915, + "learning_rate": 3.145549192941573e-05, + "loss": 0.0149, + "step": 6400 + }, + { + "epoch": 9.3033381712627, + "grad_norm": 0.27033689618110657, + "learning_rate": 3.130203971310999e-05, + "loss": 0.0165, + "step": 6410 + }, + { + "epoch": 9.317851959361393, + "grad_norm": 0.2960202693939209, + "learning_rate": 3.114879197470225e-05, + "loss": 0.0153, + "step": 6420 + }, + { + "epoch": 9.332365747460088, + "grad_norm": 0.2299107015132904, + "learning_rate": 3.0995750390085285e-05, + "loss": 0.0123, + "step": 6430 + }, + { + "epoch": 9.34687953555878, + "grad_norm": 0.3187692165374756, + "learning_rate": 3.084291663289728e-05, + "loss": 0.0136, + "step": 6440 + }, + { + "epoch": 9.361393323657474, + "grad_norm": 0.2390289455652237, + "learning_rate": 3.069029237450375e-05, + "loss": 0.0111, + "step": 6450 + }, + { + "epoch": 9.375907111756169, + "grad_norm": 0.4064617455005646, + "learning_rate": 3.053787928397911e-05, + "loss": 0.012, + "step": 6460 + }, + { + "epoch": 9.390420899854862, + "grad_norm": 0.4137391149997711, + "learning_rate": 3.0385679028088526e-05, + "loss": 0.0147, + "step": 6470 + }, + { + "epoch": 9.404934687953556, + "grad_norm": 0.24902376532554626, + "learning_rate": 3.023369327126959e-05, + "loss": 0.0123, + "step": 6480 + }, + { + "epoch": 9.41944847605225, + "grad_norm": 0.3717590272426605, + "learning_rate": 3.0081923675614198e-05, + "loss": 0.0157, + "step": 6490 + }, + { + "epoch": 9.433962264150944, + "grad_norm": 0.3626473546028137, + "learning_rate": 2.993037190085034e-05, + "loss": 0.0129, + "step": 6500 + }, + { + "epoch": 9.448476052249637, + "grad_norm": 0.26060712337493896, + "learning_rate": 2.977903960432392e-05, + "loss": 0.0155, + "step": 6510 + }, + { + "epoch": 9.462989840348332, + "grad_norm": 0.2412223368883133, + "learning_rate": 2.9627928440980722e-05, + "loss": 0.0148, + "step": 6520 + }, + { + "epoch": 9.477503628447025, + "grad_norm": 0.27503514289855957, + "learning_rate": 2.9477040063348183e-05, + "loss": 0.017, + "step": 6530 + }, + { + "epoch": 9.492017416545718, + "grad_norm": 0.1503063142299652, + "learning_rate": 2.9326376121517456e-05, + "loss": 0.0136, + "step": 6540 + }, + { + "epoch": 9.506531204644412, + "grad_norm": 0.1835785061120987, + "learning_rate": 2.9175938263125236e-05, + "loss": 0.0145, + "step": 6550 + }, + { + "epoch": 9.521044992743105, + "grad_norm": 0.16531826555728912, + "learning_rate": 2.9025728133335873e-05, + "loss": 0.0136, + "step": 6560 + }, + { + "epoch": 9.5355587808418, + "grad_norm": 0.31531593203544617, + "learning_rate": 2.8875747374823288e-05, + "loss": 0.0157, + "step": 6570 + }, + { + "epoch": 9.550072568940493, + "grad_norm": 0.41630446910858154, + "learning_rate": 2.872599762775298e-05, + "loss": 0.0192, + "step": 6580 + }, + { + "epoch": 9.564586357039188, + "grad_norm": 0.3582642078399658, + "learning_rate": 2.857648052976425e-05, + "loss": 0.0154, + "step": 6590 + }, + { + "epoch": 9.57910014513788, + "grad_norm": 0.28192341327667236, + "learning_rate": 2.8427197715952047e-05, + "loss": 0.0144, + "step": 6600 + }, + { + "epoch": 9.593613933236576, + "grad_norm": 0.2815694212913513, + "learning_rate": 2.8278150818849393e-05, + "loss": 0.0129, + "step": 6610 + }, + { + "epoch": 9.608127721335268, + "grad_norm": 0.24204884469509125, + "learning_rate": 2.812934146840922e-05, + "loss": 0.0127, + "step": 6620 + }, + { + "epoch": 9.622641509433961, + "grad_norm": 0.2016502171754837, + "learning_rate": 2.7980771291986764e-05, + "loss": 0.0163, + "step": 6630 + }, + { + "epoch": 9.637155297532656, + "grad_norm": 0.34807291626930237, + "learning_rate": 2.783244191432167e-05, + "loss": 0.0155, + "step": 6640 + }, + { + "epoch": 9.65166908563135, + "grad_norm": 0.36944228410720825, + "learning_rate": 2.768435495752022e-05, + "loss": 0.0124, + "step": 6650 + }, + { + "epoch": 9.666182873730044, + "grad_norm": 0.3315834701061249, + "learning_rate": 2.753651204103771e-05, + "loss": 0.0146, + "step": 6660 + }, + { + "epoch": 9.680696661828737, + "grad_norm": 0.5429918766021729, + "learning_rate": 2.7388914781660523e-05, + "loss": 0.0134, + "step": 6670 + }, + { + "epoch": 9.695210449927432, + "grad_norm": 0.3673105835914612, + "learning_rate": 2.7241564793488693e-05, + "loss": 0.0154, + "step": 6680 + }, + { + "epoch": 9.709724238026125, + "grad_norm": 0.7026537656784058, + "learning_rate": 2.7094463687918037e-05, + "loss": 0.0152, + "step": 6690 + }, + { + "epoch": 9.72423802612482, + "grad_norm": 0.20052950084209442, + "learning_rate": 2.694761307362268e-05, + "loss": 0.0126, + "step": 6700 + }, + { + "epoch": 9.738751814223512, + "grad_norm": 0.20951151847839355, + "learning_rate": 2.6801014556537467e-05, + "loss": 0.0131, + "step": 6710 + }, + { + "epoch": 9.753265602322205, + "grad_norm": 0.24434855580329895, + "learning_rate": 2.6654669739840243e-05, + "loss": 0.0128, + "step": 6720 + }, + { + "epoch": 9.7677793904209, + "grad_norm": 0.29252246022224426, + "learning_rate": 2.650858022393451e-05, + "loss": 0.0136, + "step": 6730 + }, + { + "epoch": 9.782293178519593, + "grad_norm": 0.19476255774497986, + "learning_rate": 2.6362747606431747e-05, + "loss": 0.0185, + "step": 6740 + }, + { + "epoch": 9.796806966618288, + "grad_norm": 0.2575831711292267, + "learning_rate": 2.6217173482134172e-05, + "loss": 0.0122, + "step": 6750 + }, + { + "epoch": 9.81132075471698, + "grad_norm": 0.2673117518424988, + "learning_rate": 2.6071859443017044e-05, + "loss": 0.0121, + "step": 6760 + }, + { + "epoch": 9.825834542815675, + "grad_norm": 0.21470312774181366, + "learning_rate": 2.5926807078211414e-05, + "loss": 0.0102, + "step": 6770 + }, + { + "epoch": 9.840348330914368, + "grad_norm": 0.2117675393819809, + "learning_rate": 2.5782017973986728e-05, + "loss": 0.0114, + "step": 6780 + }, + { + "epoch": 9.854862119013063, + "grad_norm": 0.19689474999904633, + "learning_rate": 2.5637493713733374e-05, + "loss": 0.0144, + "step": 6790 + }, + { + "epoch": 9.869375907111756, + "grad_norm": 0.39476722478866577, + "learning_rate": 2.549323587794559e-05, + "loss": 0.0172, + "step": 6800 + }, + { + "epoch": 9.88388969521045, + "grad_norm": 0.22158993780612946, + "learning_rate": 2.5349246044203895e-05, + "loss": 0.0118, + "step": 6810 + }, + { + "epoch": 9.898403483309144, + "grad_norm": 0.2774999737739563, + "learning_rate": 2.520552578715808e-05, + "loss": 0.011, + "step": 6820 + }, + { + "epoch": 9.912917271407837, + "grad_norm": 0.3380773961544037, + "learning_rate": 2.506207667850981e-05, + "loss": 0.011, + "step": 6830 + }, + { + "epoch": 9.927431059506532, + "grad_norm": 0.27769967913627625, + "learning_rate": 2.4918900286995555e-05, + "loss": 0.0106, + "step": 6840 + }, + { + "epoch": 9.941944847605225, + "grad_norm": 0.18018807470798492, + "learning_rate": 2.4775998178369458e-05, + "loss": 0.0117, + "step": 6850 + }, + { + "epoch": 9.95645863570392, + "grad_norm": 0.2075735628604889, + "learning_rate": 2.4633371915386017e-05, + "loss": 0.0142, + "step": 6860 + }, + { + "epoch": 9.970972423802612, + "grad_norm": 0.15542005002498627, + "learning_rate": 2.4491023057783235e-05, + "loss": 0.0137, + "step": 6870 + }, + { + "epoch": 9.985486211901307, + "grad_norm": 0.2745400369167328, + "learning_rate": 2.4348953162265375e-05, + "loss": 0.0169, + "step": 6880 + }, + { + "epoch": 10.0, + "grad_norm": 0.15275274217128754, + "learning_rate": 2.420716378248607e-05, + "loss": 0.0128, + "step": 6890 + }, + { + "epoch": 10.014513788098693, + "grad_norm": 0.23156581819057465, + "learning_rate": 2.4065656469031266e-05, + "loss": 0.0126, + "step": 6900 + }, + { + "epoch": 10.029027576197388, + "grad_norm": 0.2282860428094864, + "learning_rate": 2.3924432769402268e-05, + "loss": 0.0131, + "step": 6910 + }, + { + "epoch": 10.04354136429608, + "grad_norm": 0.20903392136096954, + "learning_rate": 2.3783494227998844e-05, + "loss": 0.0178, + "step": 6920 + }, + { + "epoch": 10.058055152394775, + "grad_norm": 0.225628063082695, + "learning_rate": 2.3642842386102264e-05, + "loss": 0.0101, + "step": 6930 + }, + { + "epoch": 10.072568940493468, + "grad_norm": 0.15947149693965912, + "learning_rate": 2.3502478781858567e-05, + "loss": 0.0116, + "step": 6940 + }, + { + "epoch": 10.087082728592163, + "grad_norm": 0.35400867462158203, + "learning_rate": 2.3362404950261628e-05, + "loss": 0.014, + "step": 6950 + }, + { + "epoch": 10.101596516690856, + "grad_norm": 0.2832906246185303, + "learning_rate": 2.3222622423136458e-05, + "loss": 0.0152, + "step": 6960 + }, + { + "epoch": 10.11611030478955, + "grad_norm": 0.2631046175956726, + "learning_rate": 2.3083132729122332e-05, + "loss": 0.0146, + "step": 6970 + }, + { + "epoch": 10.130624092888244, + "grad_norm": 0.26550838351249695, + "learning_rate": 2.294393739365621e-05, + "loss": 0.0121, + "step": 6980 + }, + { + "epoch": 10.145137880986937, + "grad_norm": 0.29770854115486145, + "learning_rate": 2.2805037938956e-05, + "loss": 0.0155, + "step": 6990 + }, + { + "epoch": 10.159651669085632, + "grad_norm": 0.2935265600681305, + "learning_rate": 2.266643588400386e-05, + "loss": 0.0116, + "step": 7000 + }, + { + "epoch": 10.174165457184325, + "grad_norm": 0.5476122498512268, + "learning_rate": 2.252813274452969e-05, + "loss": 0.0143, + "step": 7010 + }, + { + "epoch": 10.18867924528302, + "grad_norm": 0.14431867003440857, + "learning_rate": 2.2390130032994427e-05, + "loss": 0.0105, + "step": 7020 + }, + { + "epoch": 10.203193033381712, + "grad_norm": 0.9272375702857971, + "learning_rate": 2.2252429258573633e-05, + "loss": 0.0144, + "step": 7030 + }, + { + "epoch": 10.217706821480407, + "grad_norm": 0.24914070963859558, + "learning_rate": 2.2115031927140904e-05, + "loss": 0.0147, + "step": 7040 + }, + { + "epoch": 10.2322206095791, + "grad_norm": 0.28037455677986145, + "learning_rate": 2.1977939541251463e-05, + "loss": 0.0179, + "step": 7050 + }, + { + "epoch": 10.246734397677795, + "grad_norm": 0.3021586835384369, + "learning_rate": 2.1841153600125684e-05, + "loss": 0.0129, + "step": 7060 + }, + { + "epoch": 10.261248185776488, + "grad_norm": 0.28291964530944824, + "learning_rate": 2.170467559963267e-05, + "loss": 0.0141, + "step": 7070 + }, + { + "epoch": 10.27576197387518, + "grad_norm": 0.2658557593822479, + "learning_rate": 2.1568507032273982e-05, + "loss": 0.0102, + "step": 7080 + }, + { + "epoch": 10.290275761973875, + "grad_norm": 0.2247263342142105, + "learning_rate": 2.1432649387167264e-05, + "loss": 0.0107, + "step": 7090 + }, + { + "epoch": 10.304789550072568, + "grad_norm": 0.21270011365413666, + "learning_rate": 2.1297104150029973e-05, + "loss": 0.0119, + "step": 7100 + }, + { + "epoch": 10.319303338171263, + "grad_norm": 0.3395131826400757, + "learning_rate": 2.116187280316307e-05, + "loss": 0.0118, + "step": 7110 + }, + { + "epoch": 10.333817126269956, + "grad_norm": 0.23625223338603973, + "learning_rate": 2.1026956825434908e-05, + "loss": 0.0136, + "step": 7120 + }, + { + "epoch": 10.34833091436865, + "grad_norm": 0.5329069495201111, + "learning_rate": 2.0892357692265017e-05, + "loss": 0.0119, + "step": 7130 + }, + { + "epoch": 10.362844702467344, + "grad_norm": 0.20440439879894257, + "learning_rate": 2.0758076875607947e-05, + "loss": 0.0131, + "step": 7140 + }, + { + "epoch": 10.377358490566039, + "grad_norm": 0.5406821966171265, + "learning_rate": 2.0624115843937207e-05, + "loss": 0.0141, + "step": 7150 + }, + { + "epoch": 10.391872278664732, + "grad_norm": 0.21725451946258545, + "learning_rate": 2.0490476062229157e-05, + "loss": 0.0144, + "step": 7160 + }, + { + "epoch": 10.406386066763424, + "grad_norm": 0.35346856713294983, + "learning_rate": 2.035715899194704e-05, + "loss": 0.0146, + "step": 7170 + }, + { + "epoch": 10.42089985486212, + "grad_norm": 0.37455639243125916, + "learning_rate": 2.022416609102499e-05, + "loss": 0.0134, + "step": 7180 + }, + { + "epoch": 10.435413642960812, + "grad_norm": 0.3581145405769348, + "learning_rate": 2.009149881385205e-05, + "loss": 0.0172, + "step": 7190 + }, + { + "epoch": 10.449927431059507, + "grad_norm": 0.35843798518180847, + "learning_rate": 1.995915861125634e-05, + "loss": 0.0123, + "step": 7200 + }, + { + "epoch": 10.4644412191582, + "grad_norm": 0.16033221781253815, + "learning_rate": 1.9827146930489065e-05, + "loss": 0.0108, + "step": 7210 + }, + { + "epoch": 10.478955007256895, + "grad_norm": 0.4957810938358307, + "learning_rate": 1.9695465215208848e-05, + "loss": 0.0168, + "step": 7220 + }, + { + "epoch": 10.493468795355588, + "grad_norm": 0.1877259463071823, + "learning_rate": 1.9564114905465813e-05, + "loss": 0.0111, + "step": 7230 + }, + { + "epoch": 10.507982583454282, + "grad_norm": 0.28063082695007324, + "learning_rate": 1.9433097437685936e-05, + "loss": 0.0118, + "step": 7240 + }, + { + "epoch": 10.522496371552975, + "grad_norm": 0.2784799337387085, + "learning_rate": 1.930241424465521e-05, + "loss": 0.0113, + "step": 7250 + }, + { + "epoch": 10.537010159651668, + "grad_norm": 0.26709848642349243, + "learning_rate": 1.9172066755504115e-05, + "loss": 0.0111, + "step": 7260 + }, + { + "epoch": 10.551523947750363, + "grad_norm": 0.24270834028720856, + "learning_rate": 1.9042056395691914e-05, + "loss": 0.0134, + "step": 7270 + }, + { + "epoch": 10.566037735849056, + "grad_norm": 0.3343499004840851, + "learning_rate": 1.8912384586991066e-05, + "loss": 0.0105, + "step": 7280 + }, + { + "epoch": 10.58055152394775, + "grad_norm": 0.3393867313861847, + "learning_rate": 1.8783052747471717e-05, + "loss": 0.0117, + "step": 7290 + }, + { + "epoch": 10.595065312046444, + "grad_norm": 0.3027072846889496, + "learning_rate": 1.865406229148611e-05, + "loss": 0.013, + "step": 7300 + }, + { + "epoch": 10.609579100145138, + "grad_norm": 0.23425555229187012, + "learning_rate": 1.8525414629653233e-05, + "loss": 0.0163, + "step": 7310 + }, + { + "epoch": 10.624092888243831, + "grad_norm": 0.24896608293056488, + "learning_rate": 1.8397111168843255e-05, + "loss": 0.0116, + "step": 7320 + }, + { + "epoch": 10.638606676342526, + "grad_norm": 0.24313008785247803, + "learning_rate": 1.8269153312162323e-05, + "loss": 0.013, + "step": 7330 + }, + { + "epoch": 10.65312046444122, + "grad_norm": 0.2979603409767151, + "learning_rate": 1.8141542458937054e-05, + "loss": 0.0107, + "step": 7340 + }, + { + "epoch": 10.667634252539912, + "grad_norm": 0.2755179703235626, + "learning_rate": 1.8014280004699268e-05, + "loss": 0.0121, + "step": 7350 + }, + { + "epoch": 10.682148040638607, + "grad_norm": 0.20072540640830994, + "learning_rate": 1.788736734117078e-05, + "loss": 0.0124, + "step": 7360 + }, + { + "epoch": 10.6966618287373, + "grad_norm": 0.9193075299263, + "learning_rate": 1.7760805856248152e-05, + "loss": 0.011, + "step": 7370 + }, + { + "epoch": 10.711175616835995, + "grad_norm": 0.2081422358751297, + "learning_rate": 1.7634596933987518e-05, + "loss": 0.0131, + "step": 7380 + }, + { + "epoch": 10.725689404934688, + "grad_norm": 0.3950536251068115, + "learning_rate": 1.7508741954589404e-05, + "loss": 0.0151, + "step": 7390 + }, + { + "epoch": 10.740203193033382, + "grad_norm": 0.28460749983787537, + "learning_rate": 1.7383242294383717e-05, + "loss": 0.0123, + "step": 7400 + }, + { + "epoch": 10.754716981132075, + "grad_norm": 0.20680376887321472, + "learning_rate": 1.7258099325814632e-05, + "loss": 0.0105, + "step": 7410 + }, + { + "epoch": 10.76923076923077, + "grad_norm": 0.21150879561901093, + "learning_rate": 1.7133314417425594e-05, + "loss": 0.0131, + "step": 7420 + }, + { + "epoch": 10.783744557329463, + "grad_norm": 0.17707130312919617, + "learning_rate": 1.7008888933844408e-05, + "loss": 0.012, + "step": 7430 + }, + { + "epoch": 10.798258345428156, + "grad_norm": 0.3509533703327179, + "learning_rate": 1.6884824235768172e-05, + "loss": 0.0124, + "step": 7440 + }, + { + "epoch": 10.81277213352685, + "grad_norm": 0.2047470211982727, + "learning_rate": 1.6761121679948592e-05, + "loss": 0.0087, + "step": 7450 + }, + { + "epoch": 10.827285921625544, + "grad_norm": 0.20999592542648315, + "learning_rate": 1.663778261917695e-05, + "loss": 0.0128, + "step": 7460 + }, + { + "epoch": 10.841799709724238, + "grad_norm": 0.2239709347486496, + "learning_rate": 1.651480840226952e-05, + "loss": 0.0121, + "step": 7470 + }, + { + "epoch": 10.856313497822931, + "grad_norm": 0.18607990443706512, + "learning_rate": 1.639220037405258e-05, + "loss": 0.0111, + "step": 7480 + }, + { + "epoch": 10.870827285921626, + "grad_norm": 0.22355665266513824, + "learning_rate": 1.6269959875347906e-05, + "loss": 0.0116, + "step": 7490 + }, + { + "epoch": 10.88534107402032, + "grad_norm": 0.23847968876361847, + "learning_rate": 1.614808824295802e-05, + "loss": 0.0122, + "step": 7500 + }, + { + "epoch": 10.899854862119014, + "grad_norm": 0.22314101457595825, + "learning_rate": 1.602658680965152e-05, + "loss": 0.0116, + "step": 7510 + }, + { + "epoch": 10.914368650217707, + "grad_norm": 0.22824975848197937, + "learning_rate": 1.5905456904148686e-05, + "loss": 0.0145, + "step": 7520 + }, + { + "epoch": 10.9288824383164, + "grad_norm": 0.15930022299289703, + "learning_rate": 1.57846998511067e-05, + "loss": 0.0136, + "step": 7530 + }, + { + "epoch": 10.943396226415095, + "grad_norm": 0.30051302909851074, + "learning_rate": 1.566431697110538e-05, + "loss": 0.0102, + "step": 7540 + }, + { + "epoch": 10.957910014513788, + "grad_norm": 0.4978439211845398, + "learning_rate": 1.554430958063259e-05, + "loss": 0.0112, + "step": 7550 + }, + { + "epoch": 10.972423802612482, + "grad_norm": 0.49858659505844116, + "learning_rate": 1.5424678992069912e-05, + "loss": 0.0114, + "step": 7560 + }, + { + "epoch": 10.986937590711175, + "grad_norm": 0.21995887160301208, + "learning_rate": 1.5305426513678362e-05, + "loss": 0.013, + "step": 7570 + }, + { + "epoch": 11.00145137880987, + "grad_norm": 0.3324698209762573, + "learning_rate": 1.518655344958388e-05, + "loss": 0.0139, + "step": 7580 + }, + { + "epoch": 11.015965166908563, + "grad_norm": 0.3442780375480652, + "learning_rate": 1.5068061099763275e-05, + "loss": 0.0122, + "step": 7590 + }, + { + "epoch": 11.030478955007258, + "grad_norm": 0.2079734206199646, + "learning_rate": 1.494995076002988e-05, + "loss": 0.0117, + "step": 7600 + }, + { + "epoch": 11.04499274310595, + "grad_norm": 0.3844035267829895, + "learning_rate": 1.4832223722019456e-05, + "loss": 0.0094, + "step": 7610 + }, + { + "epoch": 11.059506531204644, + "grad_norm": 0.17748835682868958, + "learning_rate": 1.4714881273176035e-05, + "loss": 0.0094, + "step": 7620 + }, + { + "epoch": 11.074020319303338, + "grad_norm": 0.14676329493522644, + "learning_rate": 1.4597924696737835e-05, + "loss": 0.0112, + "step": 7630 + }, + { + "epoch": 11.088534107402031, + "grad_norm": 0.22222521901130676, + "learning_rate": 1.4481355271723252e-05, + "loss": 0.011, + "step": 7640 + }, + { + "epoch": 11.103047895500726, + "grad_norm": 0.39072614908218384, + "learning_rate": 1.4365174272916809e-05, + "loss": 0.0133, + "step": 7650 + }, + { + "epoch": 11.117561683599419, + "grad_norm": 0.1935746669769287, + "learning_rate": 1.4249382970855319e-05, + "loss": 0.011, + "step": 7660 + }, + { + "epoch": 11.132075471698114, + "grad_norm": 0.17410463094711304, + "learning_rate": 1.4133982631813903e-05, + "loss": 0.0114, + "step": 7670 + }, + { + "epoch": 11.146589259796807, + "grad_norm": 0.14907988905906677, + "learning_rate": 1.4018974517792194e-05, + "loss": 0.011, + "step": 7680 + }, + { + "epoch": 11.161103047895502, + "grad_norm": 0.28761568665504456, + "learning_rate": 1.390435988650048e-05, + "loss": 0.0097, + "step": 7690 + }, + { + "epoch": 11.175616835994195, + "grad_norm": 0.24961604177951813, + "learning_rate": 1.3790139991346006e-05, + "loss": 0.0109, + "step": 7700 + }, + { + "epoch": 11.190130624092888, + "grad_norm": 0.16317175328731537, + "learning_rate": 1.367631608141926e-05, + "loss": 0.011, + "step": 7710 + }, + { + "epoch": 11.204644412191582, + "grad_norm": 0.21327394247055054, + "learning_rate": 1.3562889401480278e-05, + "loss": 0.0103, + "step": 7720 + }, + { + "epoch": 11.219158200290275, + "grad_norm": 0.2183653861284256, + "learning_rate": 1.3449861191945074e-05, + "loss": 0.0126, + "step": 7730 + }, + { + "epoch": 11.23367198838897, + "grad_norm": 0.5096116065979004, + "learning_rate": 1.3337232688872009e-05, + "loss": 0.0112, + "step": 7740 + }, + { + "epoch": 11.248185776487663, + "grad_norm": 0.17665372788906097, + "learning_rate": 1.3225005123948364e-05, + "loss": 0.01, + "step": 7750 + }, + { + "epoch": 11.262699564586358, + "grad_norm": 0.5262040495872498, + "learning_rate": 1.311317972447681e-05, + "loss": 0.0101, + "step": 7760 + }, + { + "epoch": 11.27721335268505, + "grad_norm": 0.16016161441802979, + "learning_rate": 1.3001757713361996e-05, + "loss": 0.007, + "step": 7770 + }, + { + "epoch": 11.291727140783745, + "grad_norm": 0.15568460524082184, + "learning_rate": 1.2890740309097204e-05, + "loss": 0.0121, + "step": 7780 + }, + { + "epoch": 11.306240928882438, + "grad_norm": 0.20221857726573944, + "learning_rate": 1.2780128725750944e-05, + "loss": 0.01, + "step": 7790 + }, + { + "epoch": 11.320754716981131, + "grad_norm": 0.1735721230506897, + "learning_rate": 1.266992417295379e-05, + "loss": 0.0106, + "step": 7800 + }, + { + "epoch": 11.335268505079826, + "grad_norm": 0.2598990201950073, + "learning_rate": 1.2560127855885073e-05, + "loss": 0.0117, + "step": 7810 + }, + { + "epoch": 11.349782293178519, + "grad_norm": 0.43039408326148987, + "learning_rate": 1.2450740975259745e-05, + "loss": 0.0149, + "step": 7820 + }, + { + "epoch": 11.364296081277214, + "grad_norm": 0.31392189860343933, + "learning_rate": 1.234176472731517e-05, + "loss": 0.0149, + "step": 7830 + }, + { + "epoch": 11.378809869375907, + "grad_norm": 0.22327369451522827, + "learning_rate": 1.2233200303798158e-05, + "loss": 0.0113, + "step": 7840 + }, + { + "epoch": 11.393323657474602, + "grad_norm": 0.18170465528964996, + "learning_rate": 1.2125048891951846e-05, + "loss": 0.0092, + "step": 7850 + }, + { + "epoch": 11.407837445573294, + "grad_norm": 0.16956084966659546, + "learning_rate": 1.2017311674502745e-05, + "loss": 0.0089, + "step": 7860 + }, + { + "epoch": 11.42235123367199, + "grad_norm": 0.21507588028907776, + "learning_rate": 1.1909989829647822e-05, + "loss": 0.0085, + "step": 7870 + }, + { + "epoch": 11.436865021770682, + "grad_norm": 0.316275030374527, + "learning_rate": 1.1803084531041553e-05, + "loss": 0.0107, + "step": 7880 + }, + { + "epoch": 11.451378809869375, + "grad_norm": 0.2228548377752304, + "learning_rate": 1.1696596947783162e-05, + "loss": 0.0106, + "step": 7890 + }, + { + "epoch": 11.46589259796807, + "grad_norm": 0.24432587623596191, + "learning_rate": 1.1590528244403803e-05, + "loss": 0.0094, + "step": 7900 + }, + { + "epoch": 11.480406386066763, + "grad_norm": 0.1923343688249588, + "learning_rate": 1.148487958085382e-05, + "loss": 0.0125, + "step": 7910 + }, + { + "epoch": 11.494920174165458, + "grad_norm": 0.163561150431633, + "learning_rate": 1.1379652112490086e-05, + "loss": 0.0103, + "step": 7920 + }, + { + "epoch": 11.50943396226415, + "grad_norm": 0.23886170983314514, + "learning_rate": 1.1274846990063315e-05, + "loss": 0.0123, + "step": 7930 + }, + { + "epoch": 11.523947750362845, + "grad_norm": 0.230598583817482, + "learning_rate": 1.117046535970554e-05, + "loss": 0.0125, + "step": 7940 + }, + { + "epoch": 11.538461538461538, + "grad_norm": 0.23815056681632996, + "learning_rate": 1.106650836291755e-05, + "loss": 0.0145, + "step": 7950 + }, + { + "epoch": 11.552975326560233, + "grad_norm": 0.1252298802137375, + "learning_rate": 1.0962977136556418e-05, + "loss": 0.015, + "step": 7960 + }, + { + "epoch": 11.567489114658926, + "grad_norm": 0.13006475567817688, + "learning_rate": 1.0859872812823024e-05, + "loss": 0.0106, + "step": 7970 + }, + { + "epoch": 11.582002902757619, + "grad_norm": 0.22480754554271698, + "learning_rate": 1.0757196519249747e-05, + "loss": 0.015, + "step": 7980 + }, + { + "epoch": 11.596516690856314, + "grad_norm": 0.20199047029018402, + "learning_rate": 1.0654949378688077e-05, + "loss": 0.0118, + "step": 7990 + }, + { + "epoch": 11.611030478955007, + "grad_norm": 0.13639409840106964, + "learning_rate": 1.0553132509296376e-05, + "loss": 0.0111, + "step": 8000 + }, + { + "epoch": 11.625544267053701, + "grad_norm": 0.1327088326215744, + "learning_rate": 1.0451747024527613e-05, + "loss": 0.0117, + "step": 8010 + }, + { + "epoch": 11.640058055152394, + "grad_norm": 0.20517680048942566, + "learning_rate": 1.0350794033117189e-05, + "loss": 0.0102, + "step": 8020 + }, + { + "epoch": 11.65457184325109, + "grad_norm": 0.1526806801557541, + "learning_rate": 1.0250274639070856e-05, + "loss": 0.0106, + "step": 8030 + }, + { + "epoch": 11.669085631349782, + "grad_norm": 0.40772610902786255, + "learning_rate": 1.0150189941652599e-05, + "loss": 0.0079, + "step": 8040 + }, + { + "epoch": 11.683599419448477, + "grad_norm": 0.3259432017803192, + "learning_rate": 1.0050541035372635e-05, + "loss": 0.0093, + "step": 8050 + }, + { + "epoch": 11.69811320754717, + "grad_norm": 0.17180976271629333, + "learning_rate": 9.951329009975458e-06, + "loss": 0.0109, + "step": 8060 + }, + { + "epoch": 11.712626995645863, + "grad_norm": 0.23582717776298523, + "learning_rate": 9.852554950427845e-06, + "loss": 0.01, + "step": 8070 + }, + { + "epoch": 11.727140783744558, + "grad_norm": 0.17227058112621307, + "learning_rate": 9.754219936907105e-06, + "loss": 0.0107, + "step": 8080 + }, + { + "epoch": 11.74165457184325, + "grad_norm": 0.20469392836093903, + "learning_rate": 9.656325044789194e-06, + "loss": 0.0084, + "step": 8090 + }, + { + "epoch": 11.756168359941945, + "grad_norm": 0.13208690285682678, + "learning_rate": 9.55887134463697e-06, + "loss": 0.0067, + "step": 8100 + }, + { + "epoch": 11.770682148040638, + "grad_norm": 0.14229625463485718, + "learning_rate": 9.461859902188475e-06, + "loss": 0.0099, + "step": 8110 + }, + { + "epoch": 11.785195936139333, + "grad_norm": 0.2784939706325531, + "learning_rate": 9.365291778345303e-06, + "loss": 0.0117, + "step": 8120 + }, + { + "epoch": 11.799709724238026, + "grad_norm": 0.16867953538894653, + "learning_rate": 9.269168029160991e-06, + "loss": 0.0072, + "step": 8130 + }, + { + "epoch": 11.81422351233672, + "grad_norm": 0.22869251668453217, + "learning_rate": 9.173489705829447e-06, + "loss": 0.0103, + "step": 8140 + }, + { + "epoch": 11.828737300435414, + "grad_norm": 0.15720786154270172, + "learning_rate": 9.078257854673516e-06, + "loss": 0.0107, + "step": 8150 + }, + { + "epoch": 11.843251088534107, + "grad_norm": 0.14632974565029144, + "learning_rate": 8.983473517133429e-06, + "loss": 0.0093, + "step": 8160 + }, + { + "epoch": 11.857764876632801, + "grad_norm": 0.19637691974639893, + "learning_rate": 8.889137729755537e-06, + "loss": 0.0102, + "step": 8170 + }, + { + "epoch": 11.872278664731494, + "grad_norm": 0.27632415294647217, + "learning_rate": 8.79525152418087e-06, + "loss": 0.0128, + "step": 8180 + }, + { + "epoch": 11.88679245283019, + "grad_norm": 0.18931956589221954, + "learning_rate": 8.701815927133961e-06, + "loss": 0.0108, + "step": 8190 + }, + { + "epoch": 11.901306240928882, + "grad_norm": 0.15716154873371124, + "learning_rate": 8.608831960411534e-06, + "loss": 0.0093, + "step": 8200 + }, + { + "epoch": 11.915820029027577, + "grad_norm": 0.16916175186634064, + "learning_rate": 8.516300640871321e-06, + "loss": 0.0112, + "step": 8210 + }, + { + "epoch": 11.93033381712627, + "grad_norm": 0.18017126619815826, + "learning_rate": 8.424222980421038e-06, + "loss": 0.0101, + "step": 8220 + }, + { + "epoch": 11.944847605224965, + "grad_norm": 0.21676352620124817, + "learning_rate": 8.332599986007184e-06, + "loss": 0.0135, + "step": 8230 + }, + { + "epoch": 11.959361393323658, + "grad_norm": 0.198419988155365, + "learning_rate": 8.241432659604203e-06, + "loss": 0.0085, + "step": 8240 + }, + { + "epoch": 11.97387518142235, + "grad_norm": 0.15747927129268646, + "learning_rate": 8.150721998203331e-06, + "loss": 0.0085, + "step": 8250 + }, + { + "epoch": 11.988388969521045, + "grad_norm": 0.18598754703998566, + "learning_rate": 8.06046899380184e-06, + "loss": 0.0093, + "step": 8260 + }, + { + "epoch": 12.002902757619738, + "grad_norm": 0.5090079307556152, + "learning_rate": 7.970674633392133e-06, + "loss": 0.017, + "step": 8270 + }, + { + "epoch": 12.017416545718433, + "grad_norm": 0.2799903452396393, + "learning_rate": 7.881339898950924e-06, + "loss": 0.0102, + "step": 8280 + }, + { + "epoch": 12.031930333817126, + "grad_norm": 0.2283525913953781, + "learning_rate": 7.792465767428597e-06, + "loss": 0.0105, + "step": 8290 + }, + { + "epoch": 12.04644412191582, + "grad_norm": 0.18436221778392792, + "learning_rate": 7.704053210738376e-06, + "loss": 0.0076, + "step": 8300 + }, + { + "epoch": 12.060957910014514, + "grad_norm": 0.21018849313259125, + "learning_rate": 7.6161031957458494e-06, + "loss": 0.0117, + "step": 8310 + }, + { + "epoch": 12.075471698113208, + "grad_norm": 0.38773423433303833, + "learning_rate": 7.5286166842582605e-06, + "loss": 0.0116, + "step": 8320 + }, + { + "epoch": 12.089985486211901, + "grad_norm": 0.2505023181438446, + "learning_rate": 7.4415946330140814e-06, + "loss": 0.0103, + "step": 8330 + }, + { + "epoch": 12.104499274310594, + "grad_norm": 0.1599789261817932, + "learning_rate": 7.3550379936725644e-06, + "loss": 0.0102, + "step": 8340 + }, + { + "epoch": 12.119013062409289, + "grad_norm": 0.14540688693523407, + "learning_rate": 7.2689477128032035e-06, + "loss": 0.0105, + "step": 8350 + }, + { + "epoch": 12.133526850507982, + "grad_norm": 0.2632359266281128, + "learning_rate": 7.183324731875551e-06, + "loss": 0.0092, + "step": 8360 + }, + { + "epoch": 12.148040638606677, + "grad_norm": 0.18915975093841553, + "learning_rate": 7.098169987248782e-06, + "loss": 0.0083, + "step": 8370 + }, + { + "epoch": 12.16255442670537, + "grad_norm": 0.121279776096344, + "learning_rate": 7.013484410161553e-06, + "loss": 0.0109, + "step": 8380 + }, + { + "epoch": 12.177068214804065, + "grad_norm": 0.22334732115268707, + "learning_rate": 6.92926892672176e-06, + "loss": 0.0105, + "step": 8390 + }, + { + "epoch": 12.191582002902758, + "grad_norm": 0.15211200714111328, + "learning_rate": 6.845524457896446e-06, + "loss": 0.0065, + "step": 8400 + }, + { + "epoch": 12.206095791001452, + "grad_norm": 0.3402782082557678, + "learning_rate": 6.7622519195017165e-06, + "loss": 0.0085, + "step": 8410 + }, + { + "epoch": 12.220609579100145, + "grad_norm": 0.0987788587808609, + "learning_rate": 6.679452222192684e-06, + "loss": 0.0099, + "step": 8420 + }, + { + "epoch": 12.235123367198838, + "grad_norm": 0.5354704260826111, + "learning_rate": 6.597126271453579e-06, + "loss": 0.0085, + "step": 8430 + }, + { + "epoch": 12.249637155297533, + "grad_norm": 0.4342113733291626, + "learning_rate": 6.51527496758782e-06, + "loss": 0.0081, + "step": 8440 + }, + { + "epoch": 12.264150943396226, + "grad_norm": 0.1375289112329483, + "learning_rate": 6.433899205708155e-06, + "loss": 0.0082, + "step": 8450 + }, + { + "epoch": 12.27866473149492, + "grad_norm": 0.304421603679657, + "learning_rate": 6.352999875726856e-06, + "loss": 0.0097, + "step": 8460 + }, + { + "epoch": 12.293178519593614, + "grad_norm": 0.16112163662910461, + "learning_rate": 6.272577862346052e-06, + "loss": 0.0078, + "step": 8470 + }, + { + "epoch": 12.307692307692308, + "grad_norm": 0.203657865524292, + "learning_rate": 6.192634045047996e-06, + "loss": 0.0126, + "step": 8480 + }, + { + "epoch": 12.322206095791001, + "grad_norm": 0.2594527006149292, + "learning_rate": 6.113169298085458e-06, + "loss": 0.0114, + "step": 8490 + }, + { + "epoch": 12.336719883889696, + "grad_norm": 0.2553166449069977, + "learning_rate": 6.034184490472195e-06, + "loss": 0.0085, + "step": 8500 + }, + { + "epoch": 12.351233671988389, + "grad_norm": 0.36063483357429504, + "learning_rate": 5.955680485973386e-06, + "loss": 0.01, + "step": 8510 + }, + { + "epoch": 12.365747460087082, + "grad_norm": 0.3104756474494934, + "learning_rate": 5.877658143096265e-06, + "loss": 0.0105, + "step": 8520 + }, + { + "epoch": 12.380261248185777, + "grad_norm": 0.16913150250911713, + "learning_rate": 5.800118315080661e-06, + "loss": 0.0089, + "step": 8530 + }, + { + "epoch": 12.39477503628447, + "grad_norm": 0.21299268305301666, + "learning_rate": 5.723061849889716e-06, + "loss": 0.0098, + "step": 8540 + }, + { + "epoch": 12.409288824383164, + "grad_norm": 0.2587902247905731, + "learning_rate": 5.646489590200604e-06, + "loss": 0.0099, + "step": 8550 + }, + { + "epoch": 12.423802612481857, + "grad_norm": 0.11752243340015411, + "learning_rate": 5.570402373395256e-06, + "loss": 0.0085, + "step": 8560 + }, + { + "epoch": 12.438316400580552, + "grad_norm": 0.16157326102256775, + "learning_rate": 5.494801031551305e-06, + "loss": 0.0097, + "step": 8570 + }, + { + "epoch": 12.452830188679245, + "grad_norm": 0.21149156987667084, + "learning_rate": 5.41968639143291e-06, + "loss": 0.008, + "step": 8580 + }, + { + "epoch": 12.46734397677794, + "grad_norm": 0.24925768375396729, + "learning_rate": 5.345059274481751e-06, + "loss": 0.0096, + "step": 8590 + }, + { + "epoch": 12.481857764876633, + "grad_norm": 0.15569564700126648, + "learning_rate": 5.270920496808002e-06, + "loss": 0.0112, + "step": 8600 + }, + { + "epoch": 12.496371552975326, + "grad_norm": 0.2611624598503113, + "learning_rate": 5.1972708691814695e-06, + "loss": 0.01, + "step": 8610 + }, + { + "epoch": 12.51088534107402, + "grad_norm": 0.2097603678703308, + "learning_rate": 5.124111197022674e-06, + "loss": 0.0101, + "step": 8620 + }, + { + "epoch": 12.525399129172714, + "grad_norm": 0.23910893499851227, + "learning_rate": 5.051442280394081e-06, + "loss": 0.0081, + "step": 8630 + }, + { + "epoch": 12.539912917271408, + "grad_norm": 0.19468019902706146, + "learning_rate": 4.979264913991322e-06, + "loss": 0.0097, + "step": 8640 + }, + { + "epoch": 12.554426705370101, + "grad_norm": 0.38179636001586914, + "learning_rate": 4.907579887134489e-06, + "loss": 0.0102, + "step": 8650 + }, + { + "epoch": 12.568940493468796, + "grad_norm": 0.2689533829689026, + "learning_rate": 4.836387983759572e-06, + "loss": 0.0081, + "step": 8660 + }, + { + "epoch": 12.583454281567489, + "grad_norm": 0.19139885902404785, + "learning_rate": 4.765689982409816e-06, + "loss": 0.0089, + "step": 8670 + }, + { + "epoch": 12.597968069666184, + "grad_norm": 0.18142282962799072, + "learning_rate": 4.695486656227233e-06, + "loss": 0.0115, + "step": 8680 + }, + { + "epoch": 12.612481857764877, + "grad_norm": 0.1385355144739151, + "learning_rate": 4.625778772944156e-06, + "loss": 0.0112, + "step": 8690 + }, + { + "epoch": 12.62699564586357, + "grad_norm": 0.15752258896827698, + "learning_rate": 4.556567094874825e-06, + "loss": 0.0107, + "step": 8700 + }, + { + "epoch": 12.641509433962264, + "grad_norm": 0.24561850726604462, + "learning_rate": 4.487852378907059e-06, + "loss": 0.0072, + "step": 8710 + }, + { + "epoch": 12.656023222060957, + "grad_norm": 0.15457722544670105, + "learning_rate": 4.419635376493986e-06, + "loss": 0.0081, + "step": 8720 + }, + { + "epoch": 12.670537010159652, + "grad_norm": 0.19118858873844147, + "learning_rate": 4.351916833645825e-06, + "loss": 0.0107, + "step": 8730 + }, + { + "epoch": 12.685050798258345, + "grad_norm": 0.273548424243927, + "learning_rate": 4.284697490921691e-06, + "loss": 0.0087, + "step": 8740 + }, + { + "epoch": 12.69956458635704, + "grad_norm": 0.19968493282794952, + "learning_rate": 4.2179780834215585e-06, + "loss": 0.0106, + "step": 8750 + }, + { + "epoch": 12.714078374455733, + "grad_norm": 0.3306284248828888, + "learning_rate": 4.151759340778178e-06, + "loss": 0.0112, + "step": 8760 + }, + { + "epoch": 12.728592162554428, + "grad_norm": 0.10644090175628662, + "learning_rate": 4.086041987149109e-06, + "loss": 0.0088, + "step": 8770 + }, + { + "epoch": 12.74310595065312, + "grad_norm": 0.4815137982368469, + "learning_rate": 4.020826741208811e-06, + "loss": 0.0107, + "step": 8780 + }, + { + "epoch": 12.757619738751814, + "grad_norm": 0.21248355507850647, + "learning_rate": 3.956114316140746e-06, + "loss": 0.0099, + "step": 8790 + }, + { + "epoch": 12.772133526850508, + "grad_norm": 0.2553894519805908, + "learning_rate": 3.891905419629643e-06, + "loss": 0.01, + "step": 8800 + }, + { + "epoch": 12.786647314949201, + "grad_norm": 0.2038697600364685, + "learning_rate": 3.8282007538536946e-06, + "loss": 0.0105, + "step": 8810 + }, + { + "epoch": 12.801161103047896, + "grad_norm": 0.2258242517709732, + "learning_rate": 3.7650010154769265e-06, + "loss": 0.0092, + "step": 8820 + }, + { + "epoch": 12.815674891146589, + "grad_norm": 0.21943886578083038, + "learning_rate": 3.7023068956415608e-06, + "loss": 0.0099, + "step": 8830 + }, + { + "epoch": 12.830188679245284, + "grad_norm": 0.22024698555469513, + "learning_rate": 3.6401190799604303e-06, + "loss": 0.0066, + "step": 8840 + }, + { + "epoch": 12.844702467343977, + "grad_norm": 0.19245898723602295, + "learning_rate": 3.578438248509536e-06, + "loss": 0.009, + "step": 8850 + }, + { + "epoch": 12.859216255442671, + "grad_norm": 0.2843281328678131, + "learning_rate": 3.5172650758205583e-06, + "loss": 0.0104, + "step": 8860 + }, + { + "epoch": 12.873730043541364, + "grad_norm": 0.1768610179424286, + "learning_rate": 3.45660023087353e-06, + "loss": 0.0083, + "step": 8870 + }, + { + "epoch": 12.888243831640057, + "grad_norm": 0.2146247923374176, + "learning_rate": 3.3964443770894528e-06, + "loss": 0.0091, + "step": 8880 + }, + { + "epoch": 12.902757619738752, + "grad_norm": 0.13357298076152802, + "learning_rate": 3.3367981723231245e-06, + "loss": 0.008, + "step": 8890 + }, + { + "epoch": 12.917271407837445, + "grad_norm": 0.2415272742509842, + "learning_rate": 3.2776622688558746e-06, + "loss": 0.0108, + "step": 8900 + }, + { + "epoch": 12.93178519593614, + "grad_norm": 0.3411746025085449, + "learning_rate": 3.2190373133884677e-06, + "loss": 0.0069, + "step": 8910 + }, + { + "epoch": 12.946298984034833, + "grad_norm": 0.16057398915290833, + "learning_rate": 3.1609239470340446e-06, + "loss": 0.0083, + "step": 8920 + }, + { + "epoch": 12.960812772133528, + "grad_norm": 0.24423307180404663, + "learning_rate": 3.1033228053110373e-06, + "loss": 0.0075, + "step": 8930 + }, + { + "epoch": 12.97532656023222, + "grad_norm": 0.2903381884098053, + "learning_rate": 3.0462345181363314e-06, + "loss": 0.008, + "step": 8940 + }, + { + "epoch": 12.989840348330915, + "grad_norm": 0.12696416676044464, + "learning_rate": 2.9896597098182654e-06, + "loss": 0.0092, + "step": 8950 + }, + { + "epoch": 13.004354136429608, + "grad_norm": 0.3048354387283325, + "learning_rate": 2.933598999049891e-06, + "loss": 0.0103, + "step": 8960 + }, + { + "epoch": 13.018867924528301, + "grad_norm": 0.23828807473182678, + "learning_rate": 2.8780529989021697e-06, + "loss": 0.0075, + "step": 8970 + }, + { + "epoch": 13.033381712626996, + "grad_norm": 0.1340380758047104, + "learning_rate": 2.823022316817242e-06, + "loss": 0.0117, + "step": 8980 + }, + { + "epoch": 13.047895500725689, + "grad_norm": 0.2795448303222656, + "learning_rate": 2.7685075546018456e-06, + "loss": 0.0104, + "step": 8990 + }, + { + "epoch": 13.062409288824384, + "grad_norm": 0.21863742172718048, + "learning_rate": 2.7145093084206598e-06, + "loss": 0.0119, + "step": 9000 + }, + { + "epoch": 13.076923076923077, + "grad_norm": 0.4095380902290344, + "learning_rate": 2.661028168789892e-06, + "loss": 0.0074, + "step": 9010 + }, + { + "epoch": 13.091436865021771, + "grad_norm": 0.2550431489944458, + "learning_rate": 2.6080647205706855e-06, + "loss": 0.0068, + "step": 9020 + }, + { + "epoch": 13.105950653120464, + "grad_norm": 0.3120606243610382, + "learning_rate": 2.555619542962834e-06, + "loss": 0.008, + "step": 9030 + }, + { + "epoch": 13.120464441219157, + "grad_norm": 0.2066749632358551, + "learning_rate": 2.503693209498409e-06, + "loss": 0.0072, + "step": 9040 + }, + { + "epoch": 13.134978229317852, + "grad_norm": 0.1883503943681717, + "learning_rate": 2.452286288035449e-06, + "loss": 0.0093, + "step": 9050 + }, + { + "epoch": 13.149492017416545, + "grad_norm": 0.0808764174580574, + "learning_rate": 2.4013993407518363e-06, + "loss": 0.0079, + "step": 9060 + }, + { + "epoch": 13.16400580551524, + "grad_norm": 0.18432076275348663, + "learning_rate": 2.351032924139063e-06, + "loss": 0.0089, + "step": 9070 + }, + { + "epoch": 13.178519593613933, + "grad_norm": 0.1848396211862564, + "learning_rate": 2.30118758899619e-06, + "loss": 0.0082, + "step": 9080 + }, + { + "epoch": 13.193033381712628, + "grad_norm": 0.2380157858133316, + "learning_rate": 2.2518638804238157e-06, + "loss": 0.0077, + "step": 9090 + }, + { + "epoch": 13.20754716981132, + "grad_norm": 0.16735312342643738, + "learning_rate": 2.203062337818118e-06, + "loss": 0.0084, + "step": 9100 + }, + { + "epoch": 13.222060957910015, + "grad_norm": 0.3040224611759186, + "learning_rate": 2.1547834948649483e-06, + "loss": 0.0119, + "step": 9110 + }, + { + "epoch": 13.236574746008708, + "grad_norm": 0.2859864830970764, + "learning_rate": 2.1070278795340017e-06, + "loss": 0.0079, + "step": 9120 + }, + { + "epoch": 13.251088534107401, + "grad_norm": 0.1413731873035431, + "learning_rate": 2.059796014073029e-06, + "loss": 0.0068, + "step": 9130 + }, + { + "epoch": 13.265602322206096, + "grad_norm": 0.303128182888031, + "learning_rate": 2.01308841500214e-06, + "loss": 0.0083, + "step": 9140 + }, + { + "epoch": 13.280116110304789, + "grad_norm": 0.4166991114616394, + "learning_rate": 1.9669055931081704e-06, + "loss": 0.0122, + "step": 9150 + }, + { + "epoch": 13.294629898403484, + "grad_norm": 0.20062120258808136, + "learning_rate": 1.9212480534390507e-06, + "loss": 0.009, + "step": 9160 + }, + { + "epoch": 13.309143686502177, + "grad_norm": 0.2254006713628769, + "learning_rate": 1.8761162952983246e-06, + "loss": 0.0121, + "step": 9170 + }, + { + "epoch": 13.323657474600871, + "grad_norm": 0.20871245861053467, + "learning_rate": 1.8315108122396618e-06, + "loss": 0.0092, + "step": 9180 + }, + { + "epoch": 13.338171262699564, + "grad_norm": 0.13669495284557343, + "learning_rate": 1.787432092061475e-06, + "loss": 0.0075, + "step": 9190 + }, + { + "epoch": 13.352685050798259, + "grad_norm": 0.21663431823253632, + "learning_rate": 1.743880616801602e-06, + "loss": 0.0091, + "step": 9200 + }, + { + "epoch": 13.367198838896952, + "grad_norm": 0.1428050547838211, + "learning_rate": 1.7008568627319865e-06, + "loss": 0.0072, + "step": 9210 + }, + { + "epoch": 13.381712626995645, + "grad_norm": 0.34447625279426575, + "learning_rate": 1.6583613003535226e-06, + "loss": 0.0111, + "step": 9220 + }, + { + "epoch": 13.39622641509434, + "grad_norm": 0.5919042825698853, + "learning_rate": 1.6163943943908522e-06, + "loss": 0.0096, + "step": 9230 + }, + { + "epoch": 13.410740203193033, + "grad_norm": 0.25181886553764343, + "learning_rate": 1.5749566037873476e-06, + "loss": 0.0093, + "step": 9240 + }, + { + "epoch": 13.425253991291727, + "grad_norm": 0.24487295746803284, + "learning_rate": 1.5340483817000428e-06, + "loss": 0.0098, + "step": 9250 + }, + { + "epoch": 13.43976777939042, + "grad_norm": 0.12909762561321259, + "learning_rate": 1.4936701754947101e-06, + "loss": 0.007, + "step": 9260 + }, + { + "epoch": 13.454281567489115, + "grad_norm": 0.2843658924102783, + "learning_rate": 1.4538224267409361e-06, + "loss": 0.01, + "step": 9270 + }, + { + "epoch": 13.468795355587808, + "grad_norm": 0.11030800640583038, + "learning_rate": 1.414505571207314e-06, + "loss": 0.0068, + "step": 9280 + }, + { + "epoch": 13.483309143686503, + "grad_norm": 0.2700154483318329, + "learning_rate": 1.3757200388566816e-06, + "loss": 0.0075, + "step": 9290 + }, + { + "epoch": 13.497822931785196, + "grad_norm": 0.19021031260490417, + "learning_rate": 1.3374662538414074e-06, + "loss": 0.0113, + "step": 9300 + }, + { + "epoch": 13.51233671988389, + "grad_norm": 0.19897359609603882, + "learning_rate": 1.2997446344987617e-06, + "loss": 0.0083, + "step": 9310 + }, + { + "epoch": 13.526850507982584, + "grad_norm": 0.1752973347902298, + "learning_rate": 1.262555593346315e-06, + "loss": 0.0102, + "step": 9320 + }, + { + "epoch": 13.541364296081277, + "grad_norm": 0.623312771320343, + "learning_rate": 1.2258995370774685e-06, + "loss": 0.0103, + "step": 9330 + }, + { + "epoch": 13.555878084179971, + "grad_norm": 0.18789413571357727, + "learning_rate": 1.1897768665569798e-06, + "loss": 0.0086, + "step": 9340 + }, + { + "epoch": 13.570391872278664, + "grad_norm": 0.23866024613380432, + "learning_rate": 1.1541879768165954e-06, + "loss": 0.0078, + "step": 9350 + }, + { + "epoch": 13.584905660377359, + "grad_norm": 0.3479911983013153, + "learning_rate": 1.1191332570507085e-06, + "loss": 0.011, + "step": 9360 + }, + { + "epoch": 13.599419448476052, + "grad_norm": 0.13082197308540344, + "learning_rate": 1.0846130906121132e-06, + "loss": 0.012, + "step": 9370 + }, + { + "epoch": 13.613933236574747, + "grad_norm": 0.24105854332447052, + "learning_rate": 1.0506278550078131e-06, + "loss": 0.0101, + "step": 9380 + }, + { + "epoch": 13.62844702467344, + "grad_norm": 0.11318407207727432, + "learning_rate": 1.0171779218949185e-06, + "loss": 0.0108, + "step": 9390 + }, + { + "epoch": 13.642960812772134, + "grad_norm": 0.15873393416404724, + "learning_rate": 9.842636570765174e-07, + "loss": 0.008, + "step": 9400 + }, + { + "epoch": 13.657474600870827, + "grad_norm": 0.21075065433979034, + "learning_rate": 9.518854204977612e-07, + "loss": 0.0086, + "step": 9410 + }, + { + "epoch": 13.67198838896952, + "grad_norm": 0.1101858988404274, + "learning_rate": 9.200435662418349e-07, + "loss": 0.0089, + "step": 9420 + }, + { + "epoch": 13.686502177068215, + "grad_norm": 0.1953306645154953, + "learning_rate": 8.887384425261658e-07, + "loss": 0.0093, + "step": 9430 + }, + { + "epoch": 13.701015965166908, + "grad_norm": 0.13355116546154022, + "learning_rate": 8.579703916985648e-07, + "loss": 0.0103, + "step": 9440 + }, + { + "epoch": 13.715529753265603, + "grad_norm": 0.1099228709936142, + "learning_rate": 8.277397502335194e-07, + "loss": 0.0081, + "step": 9450 + }, + { + "epoch": 13.730043541364296, + "grad_norm": 0.2545875608921051, + "learning_rate": 7.980468487284675e-07, + "loss": 0.0094, + "step": 9460 + }, + { + "epoch": 13.74455732946299, + "grad_norm": 0.16142913699150085, + "learning_rate": 7.688920119002297e-07, + "loss": 0.0095, + "step": 9470 + }, + { + "epoch": 13.759071117561684, + "grad_norm": 0.8670403957366943, + "learning_rate": 7.402755585814269e-07, + "loss": 0.0073, + "step": 9480 + }, + { + "epoch": 13.773584905660378, + "grad_norm": 0.2655981779098511, + "learning_rate": 7.121978017170073e-07, + "loss": 0.0099, + "step": 9490 + }, + { + "epoch": 13.788098693759071, + "grad_norm": 0.2072117179632187, + "learning_rate": 6.846590483608306e-07, + "loss": 0.0111, + "step": 9500 + }, + { + "epoch": 13.802612481857764, + "grad_norm": 0.19010812044143677, + "learning_rate": 6.576595996722834e-07, + "loss": 0.008, + "step": 9510 + }, + { + "epoch": 13.817126269956459, + "grad_norm": 0.2932722866535187, + "learning_rate": 6.311997509130141e-07, + "loss": 0.0105, + "step": 9520 + }, + { + "epoch": 13.831640058055152, + "grad_norm": 0.2102566808462143, + "learning_rate": 6.052797914436803e-07, + "loss": 0.0098, + "step": 9530 + }, + { + "epoch": 13.846153846153847, + "grad_norm": 0.1356879621744156, + "learning_rate": 5.799000047208181e-07, + "loss": 0.0123, + "step": 9540 + }, + { + "epoch": 13.86066763425254, + "grad_norm": 0.22082969546318054, + "learning_rate": 5.550606682937054e-07, + "loss": 0.011, + "step": 9550 + }, + { + "epoch": 13.875181422351234, + "grad_norm": 0.20672982931137085, + "learning_rate": 5.307620538013481e-07, + "loss": 0.01, + "step": 9560 + }, + { + "epoch": 13.889695210449927, + "grad_norm": 0.2896080017089844, + "learning_rate": 5.070044269694874e-07, + "loss": 0.0096, + "step": 9570 + }, + { + "epoch": 13.904208998548622, + "grad_norm": 0.23654107749462128, + "learning_rate": 4.837880476077417e-07, + "loss": 0.0104, + "step": 9580 + }, + { + "epoch": 13.918722786647315, + "grad_norm": 0.21633878350257874, + "learning_rate": 4.6111316960670835e-07, + "loss": 0.0069, + "step": 9590 + }, + { + "epoch": 13.933236574746008, + "grad_norm": 0.11633900552988052, + "learning_rate": 4.389800409352218e-07, + "loss": 0.0089, + "step": 9600 + }, + { + "epoch": 13.947750362844703, + "grad_norm": 0.15132252871990204, + "learning_rate": 4.173889036376277e-07, + "loss": 0.0102, + "step": 9610 + }, + { + "epoch": 13.962264150943396, + "grad_norm": 0.18167683482170105, + "learning_rate": 3.963399938311463e-07, + "loss": 0.0109, + "step": 9620 + }, + { + "epoch": 13.97677793904209, + "grad_norm": 0.3522479236125946, + "learning_rate": 3.7583354170328545e-07, + "loss": 0.0107, + "step": 9630 + }, + { + "epoch": 13.991291727140784, + "grad_norm": 0.16197915375232697, + "learning_rate": 3.558697715093207e-07, + "loss": 0.0078, + "step": 9640 + }, + { + "epoch": 14.005805515239478, + "grad_norm": 0.3150591552257538, + "learning_rate": 3.3644890156983576e-07, + "loss": 0.0097, + "step": 9650 + }, + { + "epoch": 14.020319303338171, + "grad_norm": 0.1693713366985321, + "learning_rate": 3.175711442683638e-07, + "loss": 0.0091, + "step": 9660 + }, + { + "epoch": 14.034833091436864, + "grad_norm": 0.16042403876781464, + "learning_rate": 2.9923670604902197e-07, + "loss": 0.0091, + "step": 9670 + }, + { + "epoch": 14.049346879535559, + "grad_norm": 0.19096313416957855, + "learning_rate": 2.814457874143028e-07, + "loss": 0.0099, + "step": 9680 + }, + { + "epoch": 14.063860667634252, + "grad_norm": 0.16209816932678223, + "learning_rate": 2.641985829228366e-07, + "loss": 0.0071, + "step": 9690 + }, + { + "epoch": 14.078374455732947, + "grad_norm": 0.12014157325029373, + "learning_rate": 2.474952811872877e-07, + "loss": 0.0077, + "step": 9700 + }, + { + "epoch": 14.09288824383164, + "grad_norm": 0.1795555055141449, + "learning_rate": 2.3133606487228397e-07, + "loss": 0.0088, + "step": 9710 + }, + { + "epoch": 14.107402031930334, + "grad_norm": 0.06611788272857666, + "learning_rate": 2.157211106924295e-07, + "loss": 0.0079, + "step": 9720 + }, + { + "epoch": 14.121915820029027, + "grad_norm": 0.2669767737388611, + "learning_rate": 2.006505894103672e-07, + "loss": 0.0102, + "step": 9730 + }, + { + "epoch": 14.136429608127722, + "grad_norm": 0.11808867752552032, + "learning_rate": 1.8612466583489696e-07, + "loss": 0.0089, + "step": 9740 + }, + { + "epoch": 14.150943396226415, + "grad_norm": 0.15714138746261597, + "learning_rate": 1.7214349881918834e-07, + "loss": 0.0092, + "step": 9750 + }, + { + "epoch": 14.165457184325108, + "grad_norm": 0.12778623402118683, + "learning_rate": 1.5870724125904845e-07, + "loss": 0.0069, + "step": 9760 + }, + { + "epoch": 14.179970972423803, + "grad_norm": 0.20853623747825623, + "learning_rate": 1.4581604009124006e-07, + "loss": 0.0076, + "step": 9770 + }, + { + "epoch": 14.194484760522496, + "grad_norm": 0.155105322599411, + "learning_rate": 1.334700362918717e-07, + "loss": 0.0092, + "step": 9780 + }, + { + "epoch": 14.20899854862119, + "grad_norm": 0.12774325907230377, + "learning_rate": 1.2166936487486015e-07, + "loss": 0.0098, + "step": 9790 + }, + { + "epoch": 14.223512336719883, + "grad_norm": 0.227215975522995, + "learning_rate": 1.1041415489045914e-07, + "loss": 0.0063, + "step": 9800 + }, + { + "epoch": 14.238026124818578, + "grad_norm": 0.20001035928726196, + "learning_rate": 9.970452942384412e-08, + "loss": 0.0097, + "step": 9810 + }, + { + "epoch": 14.252539912917271, + "grad_norm": 0.14126192033290863, + "learning_rate": 8.954060559375754e-08, + "loss": 0.007, + "step": 9820 + }, + { + "epoch": 14.267053701015966, + "grad_norm": 0.3952197730541229, + "learning_rate": 7.99224945512489e-08, + "loss": 0.0124, + "step": 9830 + }, + { + "epoch": 14.281567489114659, + "grad_norm": 0.38886842131614685, + "learning_rate": 7.085030147843675e-08, + "loss": 0.0091, + "step": 9840 + }, + { + "epoch": 14.296081277213352, + "grad_norm": 0.21383580565452576, + "learning_rate": 6.232412558736523e-08, + "loss": 0.0105, + "step": 9850 + }, + { + "epoch": 14.310595065312047, + "grad_norm": 0.378927081823349, + "learning_rate": 5.434406011893822e-08, + "loss": 0.0107, + "step": 9860 + }, + { + "epoch": 14.32510885341074, + "grad_norm": 0.18973158299922943, + "learning_rate": 4.6910192341864664e-08, + "loss": 0.0083, + "step": 9870 + }, + { + "epoch": 14.339622641509434, + "grad_norm": 0.16727378964424133, + "learning_rate": 4.0022603551737035e-08, + "loss": 0.0107, + "step": 9880 + }, + { + "epoch": 14.354136429608127, + "grad_norm": 0.23412281274795532, + "learning_rate": 3.3681369070120985e-08, + "loss": 0.0087, + "step": 9890 + }, + { + "epoch": 14.368650217706822, + "grad_norm": 0.3692896068096161, + "learning_rate": 2.7886558243744866e-08, + "loss": 0.0103, + "step": 9900 + }, + { + "epoch": 14.383164005805515, + "grad_norm": 0.25852930545806885, + "learning_rate": 2.2638234443722596e-08, + "loss": 0.0112, + "step": 9910 + }, + { + "epoch": 14.39767779390421, + "grad_norm": 0.15280403196811676, + "learning_rate": 1.7936455064887504e-08, + "loss": 0.0092, + "step": 9920 + }, + { + "epoch": 14.412191582002903, + "grad_norm": 0.32725054025650024, + "learning_rate": 1.378127152514841e-08, + "loss": 0.0065, + "step": 9930 + }, + { + "epoch": 14.426705370101596, + "grad_norm": 0.3063212037086487, + "learning_rate": 1.0172729264917857e-08, + "loss": 0.0097, + "step": 9940 + }, + { + "epoch": 14.44121915820029, + "grad_norm": 0.3161866366863251, + "learning_rate": 7.1108677466458215e-09, + "loss": 0.0107, + "step": 9950 + }, + { + "epoch": 14.455732946298983, + "grad_norm": 0.21720893681049347, + "learning_rate": 4.595720454353414e-09, + "loss": 0.0088, + "step": 9960 + }, + { + "epoch": 14.470246734397678, + "grad_norm": 0.23942230641841888, + "learning_rate": 2.627314893294264e-09, + "loss": 0.0086, + "step": 9970 + }, + { + "epoch": 14.484760522496371, + "grad_norm": 0.2672852873802185, + "learning_rate": 1.2056725896270048e-09, + "loss": 0.0085, + "step": 9980 + }, + { + "epoch": 14.499274310595066, + "grad_norm": 0.11735448986291885, + "learning_rate": 3.308090902098826e-10, + "loss": 0.009, + "step": 9990 + }, + { + "epoch": 14.513788098693759, + "grad_norm": 0.15607821941375732, + "learning_rate": 2.7339624120159555e-12, + "loss": 0.0099, + "step": 10000 + }, + { + "epoch": 14.513788098693759, + "step": 10000, + "total_flos": 0.0, + "train_loss": 0.0030955594196915627, + "train_runtime": 1948.6262, + "train_samples_per_second": 164.218, + "train_steps_per_second": 5.132 + } + ], + "logging_steps": 10, + "max_steps": 10000, + "num_input_tokens_seen": 0, + "num_train_epochs": 15, + "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 +}