diff --git "a/trainer_state.json" "b/trainer_state.json"
--- "a/trainer_state.json"
+++ "b/trainer_state.json"
@@ -3,14055 +3,13152 @@
   "best_model_checkpoint": null,
   "epoch": 3.0,
   "eval_steps": 500,
-  "global_step": 2004,
+  "global_step": 1875,
   "is_hyper_param_search": false,
   "is_local_process_zero": true,
   "is_world_process_zero": true,
   "log_history": [
     {
-      "epoch": 0.0014970059880239522,
-      "grad_norm": 0.4412629429939014,
-      "learning_rate": 3.278688524590164e-07,
-      "loss": 1.0214,
+      "epoch": 0.0016,
+      "grad_norm": 0.7358543184137598,
+      "learning_rate": 3.5087719298245616e-07,
+      "loss": 1.6113,
       "step": 1
     },
     {
-      "epoch": 0.0029940119760479044,
-      "grad_norm": 0.44857961209660113,
-      "learning_rate": 6.557377049180328e-07,
-      "loss": 1.0321,
+      "epoch": 0.0032,
+      "grad_norm": 0.7911699409831793,
+      "learning_rate": 7.017543859649123e-07,
+      "loss": 1.6029,
       "step": 2
     },
     {
-      "epoch": 0.004491017964071856,
-      "grad_norm": 0.4453725376646987,
-      "learning_rate": 9.836065573770493e-07,
-      "loss": 1.0352,
+      "epoch": 0.0048,
+      "grad_norm": 0.7514423561996403,
+      "learning_rate": 1.0526315789473685e-06,
+      "loss": 1.6231,
       "step": 3
     },
     {
-      "epoch": 0.005988023952095809,
-      "grad_norm": 0.4051365182123174,
-      "learning_rate": 1.3114754098360657e-06,
-      "loss": 0.959,
+      "epoch": 0.0064,
+      "grad_norm": 0.7653628324142866,
+      "learning_rate": 1.4035087719298246e-06,
+      "loss": 1.6389,
       "step": 4
     },
     {
-      "epoch": 0.0074850299401197605,
-      "grad_norm": 0.3685112349775458,
-      "learning_rate": 1.6393442622950819e-06,
-      "loss": 0.9052,
+      "epoch": 0.008,
+      "grad_norm": 0.751467157371489,
+      "learning_rate": 1.7543859649122807e-06,
+      "loss": 1.6246,
       "step": 5
     },
     {
-      "epoch": 0.008982035928143712,
-      "grad_norm": 0.46048753426928685,
-      "learning_rate": 1.9672131147540985e-06,
-      "loss": 0.9808,
+      "epoch": 0.0096,
+      "grad_norm": 0.7645075600822779,
+      "learning_rate": 2.105263157894737e-06,
+      "loss": 1.6364,
       "step": 6
     },
     {
-      "epoch": 0.010479041916167664,
-      "grad_norm": 0.4364944016404722,
-      "learning_rate": 2.295081967213115e-06,
-      "loss": 1.0203,
+      "epoch": 0.0112,
+      "grad_norm": 0.8084898707473633,
+      "learning_rate": 2.456140350877193e-06,
+      "loss": 1.6179,
       "step": 7
     },
     {
-      "epoch": 0.011976047904191617,
-      "grad_norm": 0.4197052350095643,
-      "learning_rate": 2.6229508196721314e-06,
-      "loss": 0.936,
+      "epoch": 0.0128,
+      "grad_norm": 0.7300397658242517,
+      "learning_rate": 2.8070175438596493e-06,
+      "loss": 1.6075,
       "step": 8
     },
     {
-      "epoch": 0.01347305389221557,
-      "grad_norm": 0.3935251532018477,
-      "learning_rate": 2.9508196721311478e-06,
-      "loss": 0.931,
+      "epoch": 0.0144,
+      "grad_norm": 0.7553157416489561,
+      "learning_rate": 3.157894736842105e-06,
+      "loss": 1.6126,
       "step": 9
     },
     {
-      "epoch": 0.014970059880239521,
-      "grad_norm": 0.40165315264572143,
-      "learning_rate": 3.2786885245901638e-06,
-      "loss": 0.9519,
+      "epoch": 0.016,
+      "grad_norm": 0.772508985596036,
+      "learning_rate": 3.5087719298245615e-06,
+      "loss": 1.6097,
       "step": 10
     },
     {
-      "epoch": 0.016467065868263474,
-      "grad_norm": 0.4327103578358591,
-      "learning_rate": 3.6065573770491806e-06,
-      "loss": 0.9975,
+      "epoch": 0.0176,
+      "grad_norm": 0.7416306158167456,
+      "learning_rate": 3.859649122807018e-06,
+      "loss": 1.5971,
       "step": 11
     },
     {
-      "epoch": 0.017964071856287425,
-      "grad_norm": 0.4829902386185703,
-      "learning_rate": 3.934426229508197e-06,
-      "loss": 1.0322,
+      "epoch": 0.0192,
+      "grad_norm": 0.7046433490415748,
+      "learning_rate": 4.210526315789474e-06,
+      "loss": 1.5399,
       "step": 12
     },
     {
-      "epoch": 0.019461077844311378,
-      "grad_norm": 0.43745126656959527,
-      "learning_rate": 4.2622950819672135e-06,
-      "loss": 0.9339,
+      "epoch": 0.0208,
+      "grad_norm": 0.7702175304329298,
+      "learning_rate": 4.56140350877193e-06,
+      "loss": 1.5664,
       "step": 13
     },
     {
-      "epoch": 0.020958083832335328,
-      "grad_norm": 0.40770435507966485,
-      "learning_rate": 4.59016393442623e-06,
-      "loss": 0.9558,
+      "epoch": 0.0224,
+      "grad_norm": 0.7298779175731793,
+      "learning_rate": 4.912280701754386e-06,
+      "loss": 1.5476,
       "step": 14
     },
     {
-      "epoch": 0.02245508982035928,
-      "grad_norm": 0.3880526820488479,
-      "learning_rate": 4.918032786885246e-06,
-      "loss": 0.9233,
+      "epoch": 0.024,
+      "grad_norm": 0.7559518088328111,
+      "learning_rate": 5.263157894736842e-06,
+      "loss": 1.5376,
       "step": 15
     },
     {
-      "epoch": 0.023952095808383235,
-      "grad_norm": 0.41942806957859513,
-      "learning_rate": 5.245901639344263e-06,
-      "loss": 0.97,
+      "epoch": 0.0256,
+      "grad_norm": 0.7096582083433334,
+      "learning_rate": 5.6140350877192985e-06,
+      "loss": 1.5267,
       "step": 16
     },
     {
-      "epoch": 0.025449101796407185,
-      "grad_norm": 0.38973129771254217,
-      "learning_rate": 5.573770491803278e-06,
-      "loss": 0.8961,
+      "epoch": 0.0272,
+      "grad_norm": 0.7914446668594594,
+      "learning_rate": 5.964912280701755e-06,
+      "loss": 1.5569,
       "step": 17
     },
     {
-      "epoch": 0.02694610778443114,
-      "grad_norm": 0.4152019760637627,
-      "learning_rate": 5.9016393442622956e-06,
-      "loss": 0.9692,
+      "epoch": 0.0288,
+      "grad_norm": 0.6806304091778055,
+      "learning_rate": 6.31578947368421e-06,
+      "loss": 1.4443,
       "step": 18
     },
     {
-      "epoch": 0.02844311377245509,
-      "grad_norm": 0.46572881246006315,
-      "learning_rate": 6.229508196721312e-06,
-      "loss": 0.9336,
+      "epoch": 0.0304,
+      "grad_norm": 0.7077907993401488,
+      "learning_rate": 6.666666666666667e-06,
+      "loss": 1.477,
       "step": 19
     },
     {
-      "epoch": 0.029940119760479042,
-      "grad_norm": 0.3928467539453223,
-      "learning_rate": 6.5573770491803276e-06,
-      "loss": 0.9064,
+      "epoch": 0.032,
+      "grad_norm": 0.7169470880561969,
+      "learning_rate": 7.017543859649123e-06,
+      "loss": 1.4694,
       "step": 20
     },
     {
-      "epoch": 0.03143712574850299,
-      "grad_norm": 0.3955023428706724,
-      "learning_rate": 6.885245901639345e-06,
-      "loss": 0.8755,
+      "epoch": 0.0336,
+      "grad_norm": 0.7194881877089658,
+      "learning_rate": 7.368421052631579e-06,
+      "loss": 1.4203,
       "step": 21
     },
     {
-      "epoch": 0.03293413173652695,
-      "grad_norm": 0.3655402984227431,
-      "learning_rate": 7.213114754098361e-06,
-      "loss": 0.8226,
+      "epoch": 0.0352,
+      "grad_norm": 0.7512994062273585,
+      "learning_rate": 7.719298245614036e-06,
+      "loss": 1.4298,
       "step": 22
     },
     {
-      "epoch": 0.0344311377245509,
-      "grad_norm": 0.38433031729962397,
-      "learning_rate": 7.540983606557377e-06,
-      "loss": 0.8473,
+      "epoch": 0.0368,
+      "grad_norm": 0.708482808548663,
+      "learning_rate": 8.070175438596492e-06,
+      "loss": 1.484,
       "step": 23
     },
     {
-      "epoch": 0.03592814371257485,
-      "grad_norm": 0.3573419875384817,
-      "learning_rate": 7.868852459016394e-06,
-      "loss": 0.8282,
+      "epoch": 0.0384,
+      "grad_norm": 0.6736120028530779,
+      "learning_rate": 8.421052631578948e-06,
+      "loss": 1.4359,
       "step": 24
     },
     {
-      "epoch": 0.0374251497005988,
-      "grad_norm": 0.4093172699149451,
-      "learning_rate": 8.19672131147541e-06,
-      "loss": 0.8879,
+      "epoch": 0.04,
+      "grad_norm": 0.6387588689424757,
+      "learning_rate": 8.771929824561405e-06,
+      "loss": 1.3827,
       "step": 25
     },
     {
-      "epoch": 0.038922155688622756,
-      "grad_norm": 0.3820938437320782,
-      "learning_rate": 8.524590163934427e-06,
-      "loss": 0.864,
+      "epoch": 0.0416,
+      "grad_norm": 0.6528462520370978,
+      "learning_rate": 9.12280701754386e-06,
+      "loss": 1.3671,
       "step": 26
     },
     {
-      "epoch": 0.040419161676646706,
-      "grad_norm": 0.39326220628777386,
-      "learning_rate": 8.852459016393443e-06,
-      "loss": 0.8862,
+      "epoch": 0.0432,
+      "grad_norm": 0.6747978566054349,
+      "learning_rate": 9.473684210526315e-06,
+      "loss": 1.3764,
       "step": 27
     },
     {
-      "epoch": 0.041916167664670656,
-      "grad_norm": 0.36807756007538617,
-      "learning_rate": 9.18032786885246e-06,
-      "loss": 0.8257,
+      "epoch": 0.0448,
+      "grad_norm": 0.651327249996694,
+      "learning_rate": 9.824561403508772e-06,
+      "loss": 1.3101,
       "step": 28
     },
     {
-      "epoch": 0.04341317365269461,
-      "grad_norm": 0.35588261728997433,
-      "learning_rate": 9.508196721311476e-06,
-      "loss": 0.8128,
+      "epoch": 0.0464,
+      "grad_norm": 0.6255836750532637,
+      "learning_rate": 1.017543859649123e-05,
+      "loss": 1.3146,
       "step": 29
     },
     {
-      "epoch": 0.04491017964071856,
-      "grad_norm": 0.3715077903083305,
-      "learning_rate": 9.836065573770493e-06,
-      "loss": 0.8099,
+      "epoch": 0.048,
+      "grad_norm": 0.6407677657370118,
+      "learning_rate": 1.0526315789473684e-05,
+      "loss": 1.2396,
       "step": 30
     },
     {
-      "epoch": 0.04640718562874251,
-      "grad_norm": 0.3596909054144244,
-      "learning_rate": 1.0163934426229509e-05,
-      "loss": 0.7715,
+      "epoch": 0.0496,
+      "grad_norm": 0.6507716198039026,
+      "learning_rate": 1.0877192982456142e-05,
+      "loss": 1.259,
       "step": 31
     },
     {
-      "epoch": 0.04790419161676647,
-      "grad_norm": 0.384702217147705,
-      "learning_rate": 1.0491803278688525e-05,
-      "loss": 0.833,
+      "epoch": 0.0512,
+      "grad_norm": 0.6092921472155218,
+      "learning_rate": 1.1228070175438597e-05,
+      "loss": 1.1989,
       "step": 32
     },
     {
-      "epoch": 0.04940119760479042,
-      "grad_norm": 0.36512927940401047,
-      "learning_rate": 1.0819672131147544e-05,
-      "loss": 0.7903,
+      "epoch": 0.0528,
+      "grad_norm": 0.6583503598059806,
+      "learning_rate": 1.1578947368421053e-05,
+      "loss": 1.2636,
       "step": 33
     },
     {
-      "epoch": 0.05089820359281437,
-      "grad_norm": 0.39841041594176824,
-      "learning_rate": 1.1147540983606557e-05,
-      "loss": 0.7641,
+      "epoch": 0.0544,
+      "grad_norm": 0.6306048826535348,
+      "learning_rate": 1.192982456140351e-05,
+      "loss": 1.1936,
       "step": 34
     },
     {
-      "epoch": 0.05239520958083832,
-      "grad_norm": 0.41324753544403525,
-      "learning_rate": 1.1475409836065575e-05,
-      "loss": 0.7774,
+      "epoch": 0.056,
+      "grad_norm": 0.6093528650503758,
+      "learning_rate": 1.2280701754385966e-05,
+      "loss": 1.161,
       "step": 35
     },
     {
-      "epoch": 0.05389221556886228,
-      "grad_norm": 0.3805767081935722,
-      "learning_rate": 1.1803278688524591e-05,
-      "loss": 0.7746,
+      "epoch": 0.0576,
+      "grad_norm": 0.6110544645977457,
+      "learning_rate": 1.263157894736842e-05,
+      "loss": 1.1311,
       "step": 36
     },
     {
-      "epoch": 0.05538922155688623,
-      "grad_norm": 0.3640692670562054,
-      "learning_rate": 1.2131147540983608e-05,
-      "loss": 0.741,
+      "epoch": 0.0592,
+      "grad_norm": 0.6113710211776816,
+      "learning_rate": 1.2982456140350879e-05,
+      "loss": 1.1236,
       "step": 37
     },
     {
-      "epoch": 0.05688622754491018,
-      "grad_norm": 0.3727384829726358,
-      "learning_rate": 1.2459016393442624e-05,
-      "loss": 0.7203,
+      "epoch": 0.0608,
+      "grad_norm": 0.6219287023974935,
+      "learning_rate": 1.3333333333333333e-05,
+      "loss": 1.1245,
       "step": 38
     },
     {
-      "epoch": 0.058383233532934134,
-      "grad_norm": 0.3695211453830474,
-      "learning_rate": 1.2786885245901642e-05,
-      "loss": 0.6895,
+      "epoch": 0.0624,
+      "grad_norm": 0.6058903001550351,
+      "learning_rate": 1.3684210526315791e-05,
+      "loss": 1.0604,
       "step": 39
     },
     {
-      "epoch": 0.059880239520958084,
-      "grad_norm": 0.41877844793582925,
-      "learning_rate": 1.3114754098360655e-05,
-      "loss": 0.6759,
+      "epoch": 0.064,
+      "grad_norm": 0.5995392327986184,
+      "learning_rate": 1.4035087719298246e-05,
+      "loss": 0.9938,
       "step": 40
     },
     {
-      "epoch": 0.061377245508982034,
-      "grad_norm": 0.33931923326947677,
-      "learning_rate": 1.3442622950819673e-05,
-      "loss": 0.6196,
+      "epoch": 0.0656,
+      "grad_norm": 0.5966528699606367,
+      "learning_rate": 1.4385964912280704e-05,
+      "loss": 0.9924,
       "step": 41
     },
     {
-      "epoch": 0.06287425149700598,
-      "grad_norm": 0.3684580530246632,
-      "learning_rate": 1.377049180327869e-05,
-      "loss": 0.6655,
+      "epoch": 0.0672,
+      "grad_norm": 0.6114423840267016,
+      "learning_rate": 1.4736842105263159e-05,
+      "loss": 0.9598,
       "step": 42
     },
     {
-      "epoch": 0.06437125748502993,
-      "grad_norm": 0.32458089242640203,
-      "learning_rate": 1.4098360655737706e-05,
-      "loss": 0.5865,
+      "epoch": 0.0688,
+      "grad_norm": 0.6128015950363562,
+      "learning_rate": 1.5087719298245615e-05,
+      "loss": 0.9558,
       "step": 43
     },
     {
-      "epoch": 0.0658682634730539,
-      "grad_norm": 0.3895546564318695,
-      "learning_rate": 1.4426229508196722e-05,
-      "loss": 0.6705,
+      "epoch": 0.0704,
+      "grad_norm": 0.6315271862733165,
+      "learning_rate": 1.543859649122807e-05,
+      "loss": 0.9046,
       "step": 44
     },
     {
-      "epoch": 0.06736526946107785,
-      "grad_norm": 0.3483102383245026,
-      "learning_rate": 1.4754098360655739e-05,
-      "loss": 0.5904,
+      "epoch": 0.072,
+      "grad_norm": 0.6309262858556801,
+      "learning_rate": 1.578947368421053e-05,
+      "loss": 0.8911,
       "step": 45
     },
     {
-      "epoch": 0.0688622754491018,
-      "grad_norm": 0.3789944257657541,
-      "learning_rate": 1.5081967213114754e-05,
-      "loss": 0.6053,
+      "epoch": 0.0736,
+      "grad_norm": 0.636954612998268,
+      "learning_rate": 1.6140350877192984e-05,
+      "loss": 0.823,
       "step": 46
     },
     {
-      "epoch": 0.07035928143712575,
-      "grad_norm": 0.37752386838956564,
-      "learning_rate": 1.5409836065573772e-05,
-      "loss": 0.5844,
+      "epoch": 0.0752,
+      "grad_norm": 0.6668276280248587,
+      "learning_rate": 1.649122807017544e-05,
+      "loss": 0.8139,
       "step": 47
     },
     {
-      "epoch": 0.0718562874251497,
-      "grad_norm": 0.3630635575321055,
-      "learning_rate": 1.5737704918032788e-05,
-      "loss": 0.5291,
+      "epoch": 0.0768,
+      "grad_norm": 0.834907844936787,
+      "learning_rate": 1.6842105263157896e-05,
+      "loss": 0.7885,
       "step": 48
     },
     {
-      "epoch": 0.07335329341317365,
-      "grad_norm": 0.3914577758366373,
-      "learning_rate": 1.6065573770491805e-05,
-      "loss": 0.5326,
+      "epoch": 0.0784,
+      "grad_norm": 0.7151347787252198,
+      "learning_rate": 1.719298245614035e-05,
+      "loss": 0.7322,
       "step": 49
     },
     {
-      "epoch": 0.0748502994011976,
-      "grad_norm": 0.40325362044689916,
-      "learning_rate": 1.639344262295082e-05,
-      "loss": 0.5288,
+      "epoch": 0.08,
+      "grad_norm": 0.6745090416732078,
+      "learning_rate": 1.754385964912281e-05,
+      "loss": 0.7137,
       "step": 50
     },
     {
-      "epoch": 0.07634730538922156,
-      "grad_norm": 0.4137060583375249,
-      "learning_rate": 1.6721311475409837e-05,
-      "loss": 0.4758,
+      "epoch": 0.0816,
+      "grad_norm": 0.6569944474302377,
+      "learning_rate": 1.7894736842105264e-05,
+      "loss": 0.6544,
       "step": 51
     },
     {
-      "epoch": 0.07784431137724551,
-      "grad_norm": 0.37431549278431586,
-      "learning_rate": 1.7049180327868854e-05,
-      "loss": 0.4133,
+      "epoch": 0.0832,
+      "grad_norm": 0.6571117644599914,
+      "learning_rate": 1.824561403508772e-05,
+      "loss": 0.6047,
       "step": 52
     },
     {
-      "epoch": 0.07934131736526946,
-      "grad_norm": 0.4434060283126579,
-      "learning_rate": 1.737704918032787e-05,
-      "loss": 0.4636,
+      "epoch": 0.0848,
+      "grad_norm": 0.6906720063383268,
+      "learning_rate": 1.8596491228070176e-05,
+      "loss": 0.5888,
       "step": 53
     },
     {
-      "epoch": 0.08083832335329341,
-      "grad_norm": 0.44913599309601093,
-      "learning_rate": 1.7704918032786887e-05,
-      "loss": 0.4453,
+      "epoch": 0.0864,
+      "grad_norm": 0.7155015831811764,
+      "learning_rate": 1.894736842105263e-05,
+      "loss": 0.556,
       "step": 54
     },
     {
-      "epoch": 0.08233532934131736,
-      "grad_norm": 0.4208028028608017,
-      "learning_rate": 1.8032786885245903e-05,
-      "loss": 0.3823,
+      "epoch": 0.088,
+      "grad_norm": 0.7439398825020056,
+      "learning_rate": 1.929824561403509e-05,
+      "loss": 0.4999,
       "step": 55
     },
     {
-      "epoch": 0.08383233532934131,
-      "grad_norm": 0.4340233256708269,
-      "learning_rate": 1.836065573770492e-05,
-      "loss": 0.3575,
+      "epoch": 0.0896,
+      "grad_norm": 0.73454273481856,
+      "learning_rate": 1.9649122807017544e-05,
+      "loss": 0.4582,
       "step": 56
     },
     {
-      "epoch": 0.08532934131736528,
-      "grad_norm": 0.4231169164061207,
-      "learning_rate": 1.8688524590163936e-05,
-      "loss": 0.3157,
+      "epoch": 0.0912,
+      "grad_norm": 0.8031998382360424,
+      "learning_rate": 2e-05,
+      "loss": 0.4203,
       "step": 57
     },
     {
-      "epoch": 0.08682634730538923,
-      "grad_norm": 0.4174421584158068,
-      "learning_rate": 1.9016393442622952e-05,
-      "loss": 0.2966,
+      "epoch": 0.0928,
+      "grad_norm": 0.7472148501527976,
+      "learning_rate": 1.9999985069241058e-05,
+      "loss": 0.3486,
       "step": 58
     },
     {
-      "epoch": 0.08832335329341318,
-      "grad_norm": 0.4611062696725442,
-      "learning_rate": 1.934426229508197e-05,
-      "loss": 0.2824,
+      "epoch": 0.0944,
+      "grad_norm": 0.7894792956649227,
+      "learning_rate": 1.9999940277008807e-05,
+      "loss": 0.3621,
       "step": 59
     },
     {
-      "epoch": 0.08982035928143713,
-      "grad_norm": 0.5834947622453618,
-      "learning_rate": 1.9672131147540985e-05,
-      "loss": 0.2742,
+      "epoch": 0.096,
+      "grad_norm": 0.8963297396723989,
+      "learning_rate": 1.9999865623437014e-05,
+      "loss": 0.3129,
       "step": 60
     },
     {
-      "epoch": 0.09131736526946108,
-      "grad_norm": 0.4356716457731799,
-      "learning_rate": 2e-05,
-      "loss": 0.232,
+      "epoch": 0.0976,
+      "grad_norm": 0.884139403384448,
+      "learning_rate": 1.99997611087486e-05,
+      "loss": 0.2671,
       "step": 61
     },
     {
-      "epoch": 0.09281437125748503,
-      "grad_norm": 0.5040703847152936,
-      "learning_rate": 1.9999986928541335e-05,
-      "loss": 0.215,
+      "epoch": 0.0992,
+      "grad_norm": 0.9041012483682248,
+      "learning_rate": 1.9999626733255662e-05,
+      "loss": 0.2702,
       "step": 62
     },
     {
-      "epoch": 0.09431137724550898,
-      "grad_norm": 0.6246157723635114,
-      "learning_rate": 1.9999947714199512e-05,
-      "loss": 0.2105,
+      "epoch": 0.1008,
+      "grad_norm": 0.9511493826946734,
+      "learning_rate": 1.9999462497359468e-05,
+      "loss": 0.258,
       "step": 63
     },
     {
-      "epoch": 0.09580838323353294,
-      "grad_norm": 0.620783787770869,
-      "learning_rate": 1.999988235707705e-05,
-      "loss": 0.1869,
+      "epoch": 0.1024,
+      "grad_norm": 0.7270450177964887,
+      "learning_rate": 1.9999268401550445e-05,
+      "loss": 0.2139,
       "step": 64
     },
     {
-      "epoch": 0.09730538922155689,
-      "grad_norm": 0.4844829922546262,
-      "learning_rate": 1.9999790857344808e-05,
-      "loss": 0.1693,
+      "epoch": 0.104,
+      "grad_norm": 0.6243505426704236,
+      "learning_rate": 1.9999044446408203e-05,
+      "loss": 0.2011,
       "step": 65
     },
     {
-      "epoch": 0.09880239520958084,
-      "grad_norm": 0.39690065232341765,
-      "learning_rate": 1.9999673215241996e-05,
-      "loss": 0.1559,
+      "epoch": 0.1056,
+      "grad_norm": 0.5771396526385555,
+      "learning_rate": 1.9998790632601496e-05,
+      "loss": 0.1764,
       "step": 66
     },
     {
-      "epoch": 0.10029940119760479,
-      "grad_norm": 0.39690065232341765,
-      "learning_rate": 1.9999673215241996e-05,
-      "loss": 0.1393,
+      "epoch": 0.1072,
+      "grad_norm": 0.4953614734628554,
+      "learning_rate": 1.9998506960888258e-05,
+      "loss": 0.1622,
       "step": 67
     },
     {
-      "epoch": 0.10179640718562874,
-      "grad_norm": 0.3744306931664112,
-      "learning_rate": 1.9999529431076165e-05,
-      "loss": 0.1444,
+      "epoch": 0.1088,
+      "grad_norm": 0.5099091603110442,
+      "learning_rate": 1.999819343211557e-05,
+      "loss": 0.1426,
       "step": 68
     },
     {
-      "epoch": 0.10329341317365269,
-      "grad_norm": 0.3555775828949075,
-      "learning_rate": 1.9999359505223208e-05,
-      "loss": 0.131,
+      "epoch": 0.1104,
+      "grad_norm": 0.4708736887102006,
+      "learning_rate": 1.999785004721968e-05,
+      "loss": 0.1339,
       "step": 69
     },
     {
-      "epoch": 0.10479041916167664,
-      "grad_norm": 0.446887895690596,
-      "learning_rate": 1.999916343812736e-05,
-      "loss": 0.1295,
+      "epoch": 0.112,
+      "grad_norm": 0.5728193517444005,
+      "learning_rate": 1.9997476807225987e-05,
+      "loss": 0.1354,
       "step": 70
     },
     {
-      "epoch": 0.1062874251497006,
-      "grad_norm": 0.4182793503525473,
-      "learning_rate": 1.99989412303012e-05,
-      "loss": 0.1129,
+      "epoch": 0.1136,
+      "grad_norm": 0.5007571963655035,
+      "learning_rate": 1.999707371324904e-05,
+      "loss": 0.1202,
       "step": 71
     },
     {
-      "epoch": 0.10778443113772455,
-      "grad_norm": 0.3512235169414461,
-      "learning_rate": 1.999869288232564e-05,
-      "loss": 0.1115,
+      "epoch": 0.1152,
+      "grad_norm": 0.4183656230559708,
+      "learning_rate": 1.9996640766492542e-05,
+      "loss": 0.1092,
       "step": 72
     },
     {
-      "epoch": 0.1092814371257485,
-      "grad_norm": 0.39358701307726796,
-      "learning_rate": 1.9998418394849935e-05,
-      "loss": 0.1059,
+      "epoch": 0.1168,
+      "grad_norm": 0.4786625534172537,
+      "learning_rate": 1.9996177968249336e-05,
+      "loss": 0.107,
       "step": 73
     },
     {
-      "epoch": 0.11077844311377245,
-      "grad_norm": 0.37079524723615265,
-      "learning_rate": 1.999811776859168e-05,
-      "loss": 0.1063,
+      "epoch": 0.1184,
+      "grad_norm": 0.4599603227151936,
+      "learning_rate": 1.999568531990141e-05,
+      "loss": 0.096,
       "step": 74
     },
     {
-      "epoch": 0.1122754491017964,
-      "grad_norm": 0.35175920574697694,
-      "learning_rate": 1.9997791004336794e-05,
-      "loss": 0.0978,
+      "epoch": 0.12,
+      "grad_norm": 0.5951574041051128,
+      "learning_rate": 1.999516282291988e-05,
+      "loss": 0.0976,
       "step": 75
     },
     {
-      "epoch": 0.11377245508982035,
-      "grad_norm": 0.44421104255896743,
-      "learning_rate": 1.9997438102939538e-05,
-      "loss": 0.0972,
+      "epoch": 0.1216,
+      "grad_norm": 0.44660464536078237,
+      "learning_rate": 1.9994610478865012e-05,
+      "loss": 0.0936,
       "step": 76
     },
     {
-      "epoch": 0.11526946107784432,
-      "grad_norm": 0.2935792679601046,
-      "learning_rate": 1.99970590653225e-05,
-      "loss": 0.0865,
+      "epoch": 0.1232,
+      "grad_norm": 0.43513595090472557,
+      "learning_rate": 1.999402828938618e-05,
+      "loss": 0.0819,
       "step": 77
     },
     {
-      "epoch": 0.11676646706586827,
-      "grad_norm": 0.32149928982255116,
-      "learning_rate": 1.999665389247659e-05,
-      "loss": 0.0848,
+      "epoch": 0.1248,
+      "grad_norm": 0.48756807602594426,
+      "learning_rate": 1.9993416256221894e-05,
+      "loss": 0.0839,
       "step": 78
     },
     {
-      "epoch": 0.11826347305389222,
-      "grad_norm": 0.3346624453038523,
-      "learning_rate": 1.999622258546105e-05,
-      "loss": 0.0841,
+      "epoch": 0.1264,
+      "grad_norm": 0.40464585472402786,
+      "learning_rate": 1.999277438119978e-05,
+      "loss": 0.0764,
       "step": 79
     },
     {
-      "epoch": 0.11976047904191617,
-      "grad_norm": 0.2307198733032352,
-      "learning_rate": 1.9995765145403444e-05,
-      "loss": 0.078,
+      "epoch": 0.128,
+      "grad_norm": 0.40431395080239424,
+      "learning_rate": 1.9992102666236567e-05,
+      "loss": 0.0727,
       "step": 80
     },
     {
-      "epoch": 0.12125748502994012,
-      "grad_norm": 0.2501179187728697,
-      "learning_rate": 1.9995281573499652e-05,
-      "loss": 0.0789,
+      "epoch": 0.1296,
+      "grad_norm": 0.41569038798416647,
+      "learning_rate": 1.9991401113338103e-05,
+      "loss": 0.0734,
       "step": 81
     },
     {
-      "epoch": 0.12275449101796407,
-      "grad_norm": 0.3027539443814983,
-      "learning_rate": 1.9994771871013874e-05,
-      "loss": 0.0773,
+      "epoch": 0.1312,
+      "grad_norm": 0.4136415893195504,
+      "learning_rate": 1.9990669724599336e-05,
+      "loss": 0.067,
       "step": 82
     },
     {
-      "epoch": 0.12425149700598802,
-      "grad_norm": 0.4231966349076508,
-      "learning_rate": 1.999423603927862e-05,
-      "loss": 0.0789,
+      "epoch": 0.1328,
+      "grad_norm": 0.37595574061292625,
+      "learning_rate": 1.9989908502204295e-05,
+      "loss": 0.0634,
       "step": 83
     },
     {
-      "epoch": 0.12574850299401197,
-      "grad_norm": 0.2635202834499579,
-      "learning_rate": 1.9993674079694708e-05,
-      "loss": 0.0675,
+      "epoch": 0.1344,
+      "grad_norm": 0.4355120929608111,
+      "learning_rate": 1.998911744842611e-05,
+      "loss": 0.0643,
       "step": 84
     },
     {
-      "epoch": 0.12724550898203593,
-      "grad_norm": 0.43234380467645195,
-      "learning_rate": 1.999308599373127e-05,
-      "loss": 0.0787,
+      "epoch": 0.136,
+      "grad_norm": 0.4273191979578173,
+      "learning_rate": 1.9988296565626988e-05,
+      "loss": 0.0663,
       "step": 85
     },
     {
-      "epoch": 0.12874251497005987,
-      "grad_norm": 0.1825788301606033,
-      "learning_rate": 1.999247178292573e-05,
-      "loss": 0.0721,
+      "epoch": 0.1376,
+      "grad_norm": 0.33212691264888766,
+      "learning_rate": 1.9987445856258208e-05,
+      "loss": 0.0621,
       "step": 86
     },
     {
-      "epoch": 0.13023952095808383,
-      "grad_norm": 0.2078262527726928,
-      "learning_rate": 1.9991831448883815e-05,
-      "loss": 0.072,
+      "epoch": 0.1392,
+      "grad_norm": 0.26988113038751194,
+      "learning_rate": 1.9986565322860117e-05,
+      "loss": 0.0573,
       "step": 87
     },
     {
-      "epoch": 0.1317365269461078,
-      "grad_norm": 0.20140626332699332,
-      "learning_rate": 1.9991164993279543e-05,
-      "loss": 0.0692,
+      "epoch": 0.1408,
+      "grad_norm": 0.3583810712871837,
+      "learning_rate": 1.9985654968062122e-05,
+      "loss": 0.0625,
       "step": 88
     },
     {
-      "epoch": 0.13323353293413173,
-      "grad_norm": 0.27122788318338725,
-      "learning_rate": 1.999047241785523e-05,
-      "loss": 0.0728,
+      "epoch": 0.1424,
+      "grad_norm": 0.2856551450347671,
+      "learning_rate": 1.9984714794582682e-05,
+      "loss": 0.0518,
       "step": 89
     },
     {
-      "epoch": 0.1347305389221557,
-      "grad_norm": 0.2420008393498157,
-      "learning_rate": 1.998975372442146e-05,
-      "loss": 0.074,
+      "epoch": 0.144,
+      "grad_norm": 0.357713767536613,
+      "learning_rate": 1.9983744805229296e-05,
+      "loss": 0.0589,
       "step": 90
     },
     {
-      "epoch": 0.13622754491017963,
-      "grad_norm": 0.29824238890853016,
-      "learning_rate": 1.9989008914857115e-05,
-      "loss": 0.0676,
+      "epoch": 0.1456,
+      "grad_norm": 0.3142942652181447,
+      "learning_rate": 1.99827450028985e-05,
+      "loss": 0.0581,
       "step": 91
     },
     {
-      "epoch": 0.1377245508982036,
-      "grad_norm": 0.2912749464147097,
-      "learning_rate": 1.9988237991109342e-05,
-      "loss": 0.0683,
+      "epoch": 0.1472,
+      "grad_norm": 0.38657154986006614,
+      "learning_rate": 1.998171539057586e-05,
+      "loss": 0.0564,
       "step": 92
     },
     {
-      "epoch": 0.13922155688622753,
-      "grad_norm": 0.22500666240199066,
-      "learning_rate": 1.9987440955193563e-05,
-      "loss": 0.0695,
+      "epoch": 0.1488,
+      "grad_norm": 0.30210821869356325,
+      "learning_rate": 1.9980655971335944e-05,
+      "loss": 0.0579,
       "step": 93
     },
     {
-      "epoch": 0.1407185628742515,
-      "grad_norm": 0.4333093969766628,
-      "learning_rate": 1.998661780919346e-05,
-      "loss": 0.0709,
+      "epoch": 0.1504,
+      "grad_norm": 0.24458499620779606,
+      "learning_rate": 1.9979566748342348e-05,
+      "loss": 0.0459,
       "step": 94
     },
     {
-      "epoch": 0.14221556886227546,
-      "grad_norm": 0.1895065089384485,
-      "learning_rate": 1.9985768555260976e-05,
-      "loss": 0.0665,
+      "epoch": 0.152,
+      "grad_norm": 0.24019855385023173,
+      "learning_rate": 1.9978447724847655e-05,
+      "loss": 0.0476,
       "step": 95
     },
     {
-      "epoch": 0.1437125748502994,
-      "grad_norm": 0.27388821802849495,
-      "learning_rate": 1.9984893195616313e-05,
-      "loss": 0.0668,
+      "epoch": 0.1536,
+      "grad_norm": 0.3262894382467836,
+      "learning_rate": 1.9977298904193438e-05,
+      "loss": 0.0486,
       "step": 96
     },
     {
-      "epoch": 0.14520958083832336,
-      "grad_norm": 0.2133616131184573,
-      "learning_rate": 1.9983991732547912e-05,
-      "loss": 0.0644,
+      "epoch": 0.1552,
+      "grad_norm": 0.2632264800201802,
+      "learning_rate": 1.9976120289810247e-05,
+      "loss": 0.0506,
       "step": 97
     },
     {
-      "epoch": 0.1467065868263473,
-      "grad_norm": 0.23231083558556673,
-      "learning_rate": 1.998306416841246e-05,
-      "loss": 0.0659,
+      "epoch": 0.1568,
+      "grad_norm": 0.27260952881040074,
+      "learning_rate": 1.997491188521761e-05,
+      "loss": 0.0455,
       "step": 98
     },
     {
-      "epoch": 0.14820359281437126,
-      "grad_norm": 0.3238616061638123,
-      "learning_rate": 1.998211050563488e-05,
-      "loss": 0.0676,
+      "epoch": 0.1584,
+      "grad_norm": 0.23211417179873192,
+      "learning_rate": 1.9973673694024002e-05,
+      "loss": 0.045,
       "step": 99
     },
     {
-      "epoch": 0.1497005988023952,
-      "grad_norm": 0.19406634289412697,
-      "learning_rate": 1.998113074670833e-05,
-      "loss": 0.0651,
+      "epoch": 0.16,
+      "grad_norm": 0.2770948659117598,
+      "learning_rate": 1.997240571992685e-05,
+      "loss": 0.0459,
       "step": 100
     },
     {
-      "epoch": 0.15119760479041916,
-      "grad_norm": 0.17418625172312732,
-      "learning_rate": 1.9980124894194184e-05,
-      "loss": 0.0641,
+      "epoch": 0.1616,
+      "grad_norm": 0.22461047265833722,
+      "learning_rate": 1.9971107966712518e-05,
+      "loss": 0.0442,
       "step": 101
     },
     {
-      "epoch": 0.15269461077844312,
-      "grad_norm": 0.21916068973423491,
-      "learning_rate": 1.9979092950722033e-05,
-      "loss": 0.0624,
+      "epoch": 0.1632,
+      "grad_norm": 0.2420647199541069,
+      "learning_rate": 1.9969780438256295e-05,
+      "loss": 0.0398,
       "step": 102
     },
     {
-      "epoch": 0.15419161676646706,
-      "grad_norm": 0.23896804696125767,
-      "learning_rate": 1.9978034918989674e-05,
-      "loss": 0.0644,
+      "epoch": 0.1648,
+      "grad_norm": 0.2899178643865089,
+      "learning_rate": 1.9968423138522382e-05,
+      "loss": 0.0446,
       "step": 103
     },
     {
-      "epoch": 0.15568862275449102,
-      "grad_norm": 0.19653741457543372,
-      "learning_rate": 1.9976950801763114e-05,
-      "loss": 0.0638,
+      "epoch": 0.1664,
+      "grad_norm": 0.32269167988242264,
+      "learning_rate": 1.9967036071563878e-05,
+      "loss": 0.0386,
       "step": 104
     },
     {
-      "epoch": 0.15718562874251496,
-      "grad_norm": 0.33352908376943624,
-      "learning_rate": 1.9975840601876553e-05,
-      "loss": 0.063,
+      "epoch": 0.168,
+      "grad_norm": 0.2875789284847917,
+      "learning_rate": 1.996561924152278e-05,
+      "loss": 0.0412,
       "step": 105
     },
     {
-      "epoch": 0.15868263473053892,
-      "grad_norm": 0.19174789139029536,
-      "learning_rate": 1.9974704322232375e-05,
-      "loss": 0.0619,
+      "epoch": 0.1696,
+      "grad_norm": 0.3161150343172469,
+      "learning_rate": 1.996417265262996e-05,
+      "loss": 0.0402,
       "step": 106
     },
     {
-      "epoch": 0.1601796407185629,
-      "grad_norm": 0.31382911722883733,
-      "learning_rate": 1.997354196580115e-05,
-      "loss": 0.063,
+      "epoch": 0.1712,
+      "grad_norm": 0.36239858352621107,
+      "learning_rate": 1.9962696309205146e-05,
+      "loss": 0.04,
       "step": 107
     },
     {
-      "epoch": 0.16167664670658682,
-      "grad_norm": 0.19646342349922372,
-      "learning_rate": 1.9972353535621613e-05,
-      "loss": 0.0614,
+      "epoch": 0.1728,
+      "grad_norm": 0.35590689541990983,
+      "learning_rate": 1.996119021565693e-05,
+      "loss": 0.0427,
       "step": 108
     },
     {
-      "epoch": 0.1631736526946108,
-      "grad_norm": 0.36904190804708314,
-      "learning_rate": 1.9971139034800666e-05,
-      "loss": 0.0626,
+      "epoch": 0.1744,
+      "grad_norm": 0.2827310515458967,
+      "learning_rate": 1.995965437648273e-05,
+      "loss": 0.0359,
       "step": 109
     },
     {
-      "epoch": 0.16467065868263472,
-      "grad_norm": 0.2720984468904841,
-      "learning_rate": 1.9969898466513375e-05,
-      "loss": 0.0624,
+      "epoch": 0.176,
+      "grad_norm": 0.29640131231314887,
+      "learning_rate": 1.9958088796268794e-05,
+      "loss": 0.0386,
       "step": 110
     },
     {
-      "epoch": 0.1661676646706587,
-      "grad_norm": 0.20744167748661432,
-      "learning_rate": 1.9968631834002947e-05,
-      "loss": 0.0595,
+      "epoch": 0.1776,
+      "grad_norm": 0.4028138274584707,
+      "learning_rate": 1.995649347969019e-05,
+      "loss": 0.0416,
       "step": 111
     },
     {
-      "epoch": 0.16766467065868262,
-      "grad_norm": 0.494627859147702,
-      "learning_rate": 1.9967339140580718e-05,
-      "loss": 0.0629,
+      "epoch": 0.1792,
+      "grad_norm": 0.30852447529803145,
+      "learning_rate": 1.9954868431510764e-05,
+      "loss": 0.0365,
       "step": 112
     },
     {
-      "epoch": 0.1691616766467066,
-      "grad_norm": 0.21122512164165028,
-      "learning_rate": 1.9966020389626173e-05,
-      "loss": 0.0611,
+      "epoch": 0.1808,
+      "grad_norm": 0.5489679840744338,
+      "learning_rate": 1.995321365658317e-05,
+      "loss": 0.0379,
       "step": 113
     },
     {
-      "epoch": 0.17065868263473055,
-      "grad_norm": 0.22452536057932843,
-      "learning_rate": 1.9964675584586918e-05,
-      "loss": 0.0613,
+      "epoch": 0.1824,
+      "grad_norm": 0.2677070683435972,
+      "learning_rate": 1.9951529159848805e-05,
+      "loss": 0.0415,
       "step": 114
     },
     {
-      "epoch": 0.1721556886227545,
-      "grad_norm": 0.1937038289344653,
-      "learning_rate": 1.996330472897866e-05,
-      "loss": 0.0568,
+      "epoch": 0.184,
+      "grad_norm": 0.22638499773534226,
+      "learning_rate": 1.994981494633784e-05,
+      "loss": 0.0319,
       "step": 115
     },
     {
-      "epoch": 0.17365269461077845,
-      "grad_norm": 0.21370437758625413,
-      "learning_rate": 1.996190782638521e-05,
-      "loss": 0.0623,
+      "epoch": 0.1856,
+      "grad_norm": 0.35815794363384784,
+      "learning_rate": 1.9948071021169176e-05,
+      "loss": 0.0339,
       "step": 116
     },
     {
-      "epoch": 0.1751497005988024,
-      "grad_norm": 0.2026742633701451,
-      "learning_rate": 1.9960484880458486e-05,
-      "loss": 0.0603,
+      "epoch": 0.1872,
+      "grad_norm": 0.3342618533336046,
+      "learning_rate": 1.9946297389550433e-05,
+      "loss": 0.0341,
       "step": 117
     },
     {
-      "epoch": 0.17664670658682635,
-      "grad_norm": 0.22896351028877954,
-      "learning_rate": 1.995903589491848e-05,
-      "loss": 0.0599,
+      "epoch": 0.1888,
+      "grad_norm": 0.26538131081219346,
+      "learning_rate": 1.9944494056777945e-05,
+      "loss": 0.0326,
       "step": 118
     },
     {
-      "epoch": 0.1781437125748503,
-      "grad_norm": 0.3143604037843644,
-      "learning_rate": 1.995756087355327e-05,
-      "loss": 0.06,
+      "epoch": 0.1904,
+      "grad_norm": 0.26670239092609044,
+      "learning_rate": 1.9942661028236746e-05,
+      "loss": 0.028,
       "step": 119
     },
     {
-      "epoch": 0.17964071856287425,
-      "grad_norm": 0.24857151161776553,
-      "learning_rate": 1.9956059820218982e-05,
-      "loss": 0.0598,
+      "epoch": 0.192,
+      "grad_norm": 0.290242888748579,
+      "learning_rate": 1.9940798309400527e-05,
+      "loss": 0.0347,
       "step": 120
     },
     {
-      "epoch": 0.18113772455089822,
-      "grad_norm": 0.23471075757390752,
-      "learning_rate": 1.9954532738839814e-05,
-      "loss": 0.0583,
+      "epoch": 0.1936,
+      "grad_norm": 0.4557618271323126,
+      "learning_rate": 1.9938905905831657e-05,
+      "loss": 0.0322,
       "step": 121
     },
     {
-      "epoch": 0.18263473053892215,
-      "grad_norm": 0.1933238932543813,
-      "learning_rate": 1.9952979633408002e-05,
-      "loss": 0.0572,
+      "epoch": 0.1952,
+      "grad_norm": 0.263625615146275,
+      "learning_rate": 1.9936983823181132e-05,
+      "loss": 0.0336,
       "step": 122
     },
     {
-      "epoch": 0.18413173652694612,
-      "grad_norm": 0.21990365595016542,
-      "learning_rate": 1.9951400507983812e-05,
-      "loss": 0.0585,
+      "epoch": 0.1968,
+      "grad_norm": 0.22436028726042295,
+      "learning_rate": 1.993503206718859e-05,
+      "loss": 0.0303,
       "step": 123
     },
     {
-      "epoch": 0.18562874251497005,
-      "grad_norm": 0.19297438870815392,
-      "learning_rate": 1.9949795366695545e-05,
-      "loss": 0.0577,
+      "epoch": 0.1984,
+      "grad_norm": 0.36036424972145303,
+      "learning_rate": 1.993305064368227e-05,
+      "loss": 0.0309,
       "step": 124
     },
     {
-      "epoch": 0.18712574850299402,
-      "grad_norm": 0.2355462451136706,
-      "learning_rate": 1.9948164213739502e-05,
-      "loss": 0.0558,
+      "epoch": 0.2,
+      "grad_norm": 0.2636999950967481,
+      "learning_rate": 1.9931039558578997e-05,
+      "loss": 0.0273,
       "step": 125
     },
     {
-      "epoch": 0.18862275449101795,
-      "grad_norm": 0.19921818148554826,
-      "learning_rate": 1.9946507053379998e-05,
-      "loss": 0.061,
+      "epoch": 0.2016,
+      "grad_norm": 0.2761731394185584,
+      "learning_rate": 1.9928998817884185e-05,
+      "loss": 0.0293,
       "step": 126
     },
     {
-      "epoch": 0.19011976047904192,
-      "grad_norm": 0.24443384088619968,
-      "learning_rate": 1.994482388994933e-05,
-      "loss": 0.0605,
+      "epoch": 0.2032,
+      "grad_norm": 0.30374955920113056,
+      "learning_rate": 1.9926928427691788e-05,
+      "loss": 0.0252,
       "step": 127
     },
     {
-      "epoch": 0.19161676646706588,
-      "grad_norm": 0.17312333530517943,
-      "learning_rate": 1.9943114727847782e-05,
-      "loss": 0.0604,
+      "epoch": 0.2048,
+      "grad_norm": 0.23433485832962797,
+      "learning_rate": 1.9924828394184308e-05,
+      "loss": 0.0274,
       "step": 128
     },
     {
-      "epoch": 0.19311377245508982,
-      "grad_norm": 0.22275107608465397,
-      "learning_rate": 1.9941379571543597e-05,
-      "loss": 0.0559,
+      "epoch": 0.2064,
+      "grad_norm": 0.2737766558936752,
+      "learning_rate": 1.992269872363277e-05,
+      "loss": 0.0251,
       "step": 129
     },
     {
-      "epoch": 0.19461077844311378,
-      "grad_norm": 0.21284771446183928,
-      "learning_rate": 1.9939618425572985e-05,
-      "loss": 0.0577,
+      "epoch": 0.208,
+      "grad_norm": 0.37112868758464157,
+      "learning_rate": 1.992053942239668e-05,
+      "loss": 0.0284,
       "step": 130
     },
     {
-      "epoch": 0.19610778443113772,
-      "grad_norm": 0.23636170247374946,
-      "learning_rate": 1.9937831294540094e-05,
-      "loss": 0.0576,
+      "epoch": 0.2096,
+      "grad_norm": 0.25459224936986335,
+      "learning_rate": 1.991835049692405e-05,
+      "loss": 0.0307,
       "step": 131
     },
     {
-      "epoch": 0.19760479041916168,
-      "grad_norm": 0.21444998693247203,
-      "learning_rate": 1.9936018183117003e-05,
-      "loss": 0.0575,
+      "epoch": 0.2112,
+      "grad_norm": 0.2902652013361823,
+      "learning_rate": 1.9916131953751342e-05,
+      "loss": 0.0245,
       "step": 132
     },
     {
-      "epoch": 0.19910179640718562,
-      "grad_norm": 0.2216207633715879,
-      "learning_rate": 1.993417909604372e-05,
-      "loss": 0.0549,
+      "epoch": 0.2128,
+      "grad_norm": 0.4198056600285124,
+      "learning_rate": 1.991388379950346e-05,
+      "loss": 0.0318,
       "step": 133
     },
     {
-      "epoch": 0.20059880239520958,
-      "grad_norm": 0.2520990703952128,
-      "learning_rate": 1.993231403812815e-05,
-      "loss": 0.0588,
+      "epoch": 0.2144,
+      "grad_norm": 0.28382438818772865,
+      "learning_rate": 1.9911606040893742e-05,
+      "loss": 0.0283,
       "step": 134
     },
     {
-      "epoch": 0.20209580838323354,
-      "grad_norm": 0.22625505105344712,
-      "learning_rate": 1.99304230142461e-05,
-      "loss": 0.0545,
+      "epoch": 0.216,
+      "grad_norm": 0.23336607870897053,
+      "learning_rate": 1.9909298684723905e-05,
+      "loss": 0.0246,
       "step": 135
     },
     {
-      "epoch": 0.20359281437125748,
-      "grad_norm": 0.17746500333059273,
-      "learning_rate": 1.9928506029341256e-05,
-      "loss": 0.0588,
+      "epoch": 0.2176,
+      "grad_norm": 0.2918806911818325,
+      "learning_rate": 1.990696173788408e-05,
+      "loss": 0.0243,
       "step": 136
     },
     {
-      "epoch": 0.20508982035928144,
-      "grad_norm": 0.17244614526395785,
-      "learning_rate": 1.9926563088425177e-05,
-      "loss": 0.0561,
+      "epoch": 0.2192,
+      "grad_norm": 0.3086628915284369,
+      "learning_rate": 1.9904595207352736e-05,
+      "loss": 0.0272,
       "step": 137
     },
     {
-      "epoch": 0.20658682634730538,
-      "grad_norm": 0.17124109763728518,
-      "learning_rate": 1.992459419657728e-05,
-      "loss": 0.0562,
+      "epoch": 0.2208,
+      "grad_norm": 0.3683163021055742,
+      "learning_rate": 1.9902199100196697e-05,
+      "loss": 0.0238,
       "step": 138
     },
     {
-      "epoch": 0.20808383233532934,
-      "grad_norm": 0.1984353051118203,
-      "learning_rate": 1.9922599358944823e-05,
-      "loss": 0.0563,
+      "epoch": 0.2224,
+      "grad_norm": 0.33375531864826613,
+      "learning_rate": 1.9899773423571102e-05,
+      "loss": 0.0253,
       "step": 139
     },
     {
-      "epoch": 0.20958083832335328,
-      "grad_norm": 0.2614654611155798,
-      "learning_rate": 1.9920578580742892e-05,
-      "loss": 0.0556,
+      "epoch": 0.224,
+      "grad_norm": 0.23272431144615036,
+      "learning_rate": 1.9897318184719386e-05,
+      "loss": 0.0198,
       "step": 140
     },
     {
-      "epoch": 0.21107784431137724,
-      "grad_norm": 0.22618621587850632,
-      "learning_rate": 1.9918531867254386e-05,
-      "loss": 0.056,
+      "epoch": 0.2256,
+      "grad_norm": 0.39233790536820173,
+      "learning_rate": 1.9894833390973266e-05,
+      "loss": 0.0247,
       "step": 141
     },
     {
-      "epoch": 0.2125748502994012,
-      "grad_norm": 0.1686791845721939,
-      "learning_rate": 1.9916459223830018e-05,
-      "loss": 0.0549,
+      "epoch": 0.2272,
+      "grad_norm": 0.3207050735210565,
+      "learning_rate": 1.989231904975272e-05,
+      "loss": 0.0217,
       "step": 142
     },
     {
-      "epoch": 0.21407185628742514,
-      "grad_norm": 0.27469186421828573,
-      "learning_rate": 1.991436065588828e-05,
-      "loss": 0.0531,
+      "epoch": 0.2288,
+      "grad_norm": 0.18878538321873145,
+      "learning_rate": 1.9889775168565942e-05,
+      "loss": 0.0195,
       "step": 143
     },
     {
-      "epoch": 0.2155688622754491,
-      "grad_norm": 0.28877488507125365,
-      "learning_rate": 1.9912236168915443e-05,
-      "loss": 0.0562,
+      "epoch": 0.2304,
+      "grad_norm": 0.331980372714454,
+      "learning_rate": 1.9887201755009358e-05,
+      "loss": 0.0222,
       "step": 144
     },
     {
-      "epoch": 0.21706586826347304,
-      "grad_norm": 0.1726289274307483,
-      "learning_rate": 1.991008576846553e-05,
-      "loss": 0.0538,
+      "epoch": 0.232,
+      "grad_norm": 0.2617001176043066,
+      "learning_rate": 1.9884598816767563e-05,
+      "loss": 0.0215,
       "step": 145
     },
     {
-      "epoch": 0.218562874251497,
-      "grad_norm": 0.1912420614588707,
-      "learning_rate": 1.990790946016032e-05,
-      "loss": 0.0528,
+      "epoch": 0.2336,
+      "grad_norm": 0.1998476020770707,
+      "learning_rate": 1.988196636161333e-05,
+      "loss": 0.021,
       "step": 146
     },
     {
-      "epoch": 0.22005988023952097,
-      "grad_norm": 0.23379848312481374,
-      "learning_rate": 1.9905707249689318e-05,
-      "loss": 0.052,
+      "epoch": 0.2352,
+      "grad_norm": 0.15711441770336212,
+      "learning_rate": 1.987930439740757e-05,
+      "loss": 0.0168,
       "step": 147
     },
     {
-      "epoch": 0.2215568862275449,
-      "grad_norm": 0.24847391624574525,
-      "learning_rate": 1.990347914280974e-05,
-      "loss": 0.0502,
+      "epoch": 0.2368,
+      "grad_norm": 0.18823706933202605,
+      "learning_rate": 1.987661293209931e-05,
+      "loss": 0.016,
       "step": 148
     },
     {
-      "epoch": 0.22305389221556887,
-      "grad_norm": 0.3441419819000447,
-      "learning_rate": 1.990122514534651e-05,
-      "loss": 0.0575,
+      "epoch": 0.2384,
+      "grad_norm": 0.2203924846273772,
+      "learning_rate": 1.9873891973725673e-05,
+      "loss": 0.022,
       "step": 149
     },
     {
-      "epoch": 0.2245508982035928,
-      "grad_norm": 0.3955212700237536,
-      "learning_rate": 1.989894526319224e-05,
-      "loss": 0.0542,
+      "epoch": 0.24,
+      "grad_norm": 0.22095723495277175,
+      "learning_rate": 1.9871141530411854e-05,
+      "loss": 0.0212,
       "step": 150
     },
     {
-      "epoch": 0.22604790419161677,
-      "grad_norm": 0.2380289767466844,
-      "learning_rate": 1.98966395023072e-05,
-      "loss": 0.0537,
+      "epoch": 0.2416,
+      "grad_norm": 0.2291413194508601,
+      "learning_rate": 1.98683616103711e-05,
+      "loss": 0.0163,
       "step": 151
     },
     {
-      "epoch": 0.2275449101796407,
-      "grad_norm": 0.26238101314438406,
-      "learning_rate": 1.989430786871932e-05,
-      "loss": 0.054,
+      "epoch": 0.2432,
+      "grad_norm": 0.2935224449509153,
+      "learning_rate": 1.986555222190467e-05,
+      "loss": 0.0195,
       "step": 152
     },
     {
-      "epoch": 0.22904191616766467,
-      "grad_norm": 0.31325198919582375,
-      "learning_rate": 1.989195036852418e-05,
-      "loss": 0.0563,
+      "epoch": 0.2448,
+      "grad_norm": 0.3524972156636008,
+      "learning_rate": 1.986271337340182e-05,
+      "loss": 0.015,
       "step": 153
     },
     {
-      "epoch": 0.23053892215568864,
-      "grad_norm": 0.27559129084895034,
-      "learning_rate": 1.9889567007884965e-05,
-      "loss": 0.0527,
+      "epoch": 0.2464,
+      "grad_norm": 0.21658492228652335,
+      "learning_rate": 1.9859845073339788e-05,
+      "loss": 0.018,
       "step": 154
     },
     {
-      "epoch": 0.23203592814371257,
-      "grad_norm": 0.2036748289588086,
-      "learning_rate": 1.988715779303248e-05,
-      "loss": 0.0523,
+      "epoch": 0.248,
+      "grad_norm": 0.26769149602569436,
+      "learning_rate": 1.9856947330283752e-05,
+      "loss": 0.0177,
       "step": 155
     },
     {
-      "epoch": 0.23353293413173654,
-      "grad_norm": 0.21158749273476493,
-      "learning_rate": 1.988472273026511e-05,
-      "loss": 0.0517,
+      "epoch": 0.2496,
+      "grad_norm": 0.23444425440366023,
+      "learning_rate": 1.9854020152886816e-05,
+      "loss": 0.0206,
       "step": 156
     },
     {
-      "epoch": 0.23502994011976047,
-      "grad_norm": 0.3777409399476317,
-      "learning_rate": 1.9882261825948828e-05,
-      "loss": 0.0545,
+      "epoch": 0.2512,
+      "grad_norm": 0.45183746759896726,
+      "learning_rate": 1.985106354988997e-05,
+      "loss": 0.0284,
       "step": 157
     },
     {
-      "epoch": 0.23652694610778444,
-      "grad_norm": 0.16198901337349092,
-      "learning_rate": 1.9879775086517145e-05,
-      "loss": 0.049,
+      "epoch": 0.2528,
+      "grad_norm": 0.22812452622594312,
+      "learning_rate": 1.9848077530122083e-05,
+      "loss": 0.0159,
       "step": 158
     },
     {
-      "epoch": 0.23802395209580837,
-      "grad_norm": 0.28600911159833825,
-      "learning_rate": 1.9877262518471133e-05,
-      "loss": 0.0511,
+      "epoch": 0.2544,
+      "grad_norm": 0.29949675168255213,
+      "learning_rate": 1.984506210249986e-05,
+      "loss": 0.0209,
       "step": 159
     },
     {
-      "epoch": 0.23952095808383234,
-      "grad_norm": 0.21326314303753005,
-      "learning_rate": 1.9874724128379373e-05,
-      "loss": 0.0511,
+      "epoch": 0.256,
+      "grad_norm": 0.20288486241622716,
+      "learning_rate": 1.984201727602783e-05,
+      "loss": 0.0175,
       "step": 160
     },
     {
-      "epoch": 0.2410179640718563,
-      "grad_norm": 0.23896462883526792,
-      "learning_rate": 1.9872159922877955e-05,
-      "loss": 0.052,
+      "epoch": 0.2576,
+      "grad_norm": 0.24153624253906011,
+      "learning_rate": 1.9838943059798305e-05,
+      "loss": 0.016,
       "step": 161
     },
     {
-      "epoch": 0.24251497005988024,
-      "grad_norm": 0.24778993752258044,
-      "learning_rate": 1.9869569908670466e-05,
-      "loss": 0.0503,
+      "epoch": 0.2592,
+      "grad_norm": 0.2769997843018402,
+      "learning_rate": 1.983583946299136e-05,
+      "loss": 0.0164,
       "step": 162
     },
     {
-      "epoch": 0.2440119760479042,
-      "grad_norm": 0.2678509446705851,
-      "learning_rate": 1.9866954092527954e-05,
-      "loss": 0.0509,
+      "epoch": 0.2608,
+      "grad_norm": 0.20799564640411272,
+      "learning_rate": 1.9832706494874812e-05,
+      "loss": 0.0158,
       "step": 163
     },
     {
-      "epoch": 0.24550898203592814,
-      "grad_norm": 0.4976820467779877,
-      "learning_rate": 1.9864312481288934e-05,
-      "loss": 0.0452,
+      "epoch": 0.2624,
+      "grad_norm": 0.23601523435408028,
+      "learning_rate": 1.9829544164804172e-05,
+      "loss": 0.016,
       "step": 164
     },
     {
-      "epoch": 0.2470059880239521,
-      "grad_norm": 0.21451893299135652,
-      "learning_rate": 1.9861645081859334e-05,
-      "loss": 0.0503,
+      "epoch": 0.264,
+      "grad_norm": 0.22418166608529375,
+      "learning_rate": 1.982635248222264e-05,
+      "loss": 0.0191,
       "step": 165
     },
     {
-      "epoch": 0.24850299401197604,
-      "grad_norm": 0.18172472245421414,
-      "learning_rate": 1.9858951901212525e-05,
-      "loss": 0.0467,
+      "epoch": 0.2656,
+      "grad_norm": 0.13054329188698238,
+      "learning_rate": 1.9823131456661064e-05,
+      "loss": 0.011,
       "step": 166
     },
     {
-      "epoch": 0.25,
-      "grad_norm": 0.1831777829268529,
-      "learning_rate": 1.9856232946389264e-05,
-      "loss": 0.046,
+      "epoch": 0.2672,
+      "grad_norm": 0.18252261238651618,
+      "learning_rate": 1.9819881097737917e-05,
+      "loss": 0.0122,
       "step": 167
     },
     {
-      "epoch": 0.25149700598802394,
-      "grad_norm": 0.24695897142692572,
-      "learning_rate": 1.985348822449769e-05,
-      "loss": 0.0449,
+      "epoch": 0.2688,
+      "grad_norm": 0.1726449946139791,
+      "learning_rate": 1.9816601415159266e-05,
+      "loss": 0.012,
       "step": 168
     },
     {
-      "epoch": 0.25299401197604793,
-      "grad_norm": 0.24320724433114305,
-      "learning_rate": 1.985071774271331e-05,
-      "loss": 0.0514,
+      "epoch": 0.2704,
+      "grad_norm": 0.28649148702667626,
+      "learning_rate": 1.9813292418718734e-05,
+      "loss": 0.0199,
       "step": 169
     },
     {
-      "epoch": 0.25449101796407186,
-      "grad_norm": 0.29316721920718963,
-      "learning_rate": 1.9847921508278977e-05,
-      "loss": 0.0455,
+      "epoch": 0.272,
+      "grad_norm": 0.32637951614090105,
+      "learning_rate": 1.980995411829749e-05,
+      "loss": 0.0183,
       "step": 170
     },
     {
-      "epoch": 0.2559880239520958,
-      "grad_norm": 0.24473542108780105,
-      "learning_rate": 1.984509952850485e-05,
-      "loss": 0.0477,
+      "epoch": 0.2736,
+      "grad_norm": 0.1924690150414372,
+      "learning_rate": 1.9806586523864212e-05,
+      "loss": 0.0134,
       "step": 171
     },
     {
-      "epoch": 0.25748502994011974,
-      "grad_norm": 0.29195204962744054,
-      "learning_rate": 1.9842251810768415e-05,
-      "loss": 0.0454,
+      "epoch": 0.2752,
+      "grad_norm": 0.3053757080333075,
+      "learning_rate": 1.980318964547504e-05,
+      "loss": 0.0205,
       "step": 172
     },
     {
-      "epoch": 0.25898203592814373,
-      "grad_norm": 0.26054012611614774,
-      "learning_rate": 1.983937836251444e-05,
-      "loss": 0.0475,
+      "epoch": 0.2768,
+      "grad_norm": 0.20048108515719862,
+      "learning_rate": 1.9799763493273572e-05,
+      "loss": 0.0104,
       "step": 173
     },
     {
-      "epoch": 0.26047904191616766,
-      "grad_norm": 0.2683808403595366,
-      "learning_rate": 1.983647919125495e-05,
-      "loss": 0.0469,
+      "epoch": 0.2784,
+      "grad_norm": 0.2732627182334408,
+      "learning_rate": 1.9796308077490817e-05,
+      "loss": 0.0122,
       "step": 174
     },
     {
-      "epoch": 0.2619760479041916,
-      "grad_norm": 0.2798959888989543,
-      "learning_rate": 1.983355430456923e-05,
-      "loss": 0.0457,
+      "epoch": 0.28,
+      "grad_norm": 0.6532880269049472,
+      "learning_rate": 1.9792823408445173e-05,
+      "loss": 0.0162,
       "step": 175
     },
     {
-      "epoch": 0.2634730538922156,
-      "grad_norm": 0.3230989454804457,
-      "learning_rate": 1.983060371010378e-05,
-      "loss": 0.0481,
+      "epoch": 0.2816,
+      "grad_norm": 0.38492332201417967,
+      "learning_rate": 1.978930949654239e-05,
+      "loss": 0.0164,
       "step": 176
     },
     {
-      "epoch": 0.26497005988023953,
-      "grad_norm": 0.28580432616804236,
-      "learning_rate": 1.9827627415572328e-05,
-      "loss": 0.0428,
+      "epoch": 0.2832,
+      "grad_norm": 0.3503764828469701,
+      "learning_rate": 1.978576635227554e-05,
+      "loss": 0.0119,
       "step": 177
     },
     {
-      "epoch": 0.26646706586826346,
-      "grad_norm": 0.3713352257679464,
-      "learning_rate": 1.982462542875576e-05,
-      "loss": 0.0467,
+      "epoch": 0.2848,
+      "grad_norm": 0.31142382540887115,
+      "learning_rate": 1.9782193986224997e-05,
+      "loss": 0.0109,
       "step": 178
     },
     {
-      "epoch": 0.2679640718562874,
-      "grad_norm": 0.2792642088474226,
-      "learning_rate": 1.982159775750216e-05,
-      "loss": 0.0468,
+      "epoch": 0.2864,
+      "grad_norm": 0.36239173703322675,
+      "learning_rate": 1.9778592409058376e-05,
+      "loss": 0.0154,
       "step": 179
     },
     {
-      "epoch": 0.2694610778443114,
-      "grad_norm": 0.26225012059799724,
-      "learning_rate": 1.9818544409726734e-05,
-      "loss": 0.0481,
+      "epoch": 0.288,
+      "grad_norm": 0.30337585394648847,
+      "learning_rate": 1.9774961631530543e-05,
+      "loss": 0.0122,
       "step": 180
     },
     {
-      "epoch": 0.27095808383233533,
-      "grad_norm": 0.320954008862723,
-      "learning_rate": 1.981546539341183e-05,
-      "loss": 0.0444,
+      "epoch": 0.2896,
+      "grad_norm": 0.35293599460078406,
+      "learning_rate": 1.9771301664483548e-05,
+      "loss": 0.0125,
       "step": 181
     },
     {
-      "epoch": 0.27245508982035926,
-      "grad_norm": 0.3027600817067956,
-      "learning_rate": 1.9812360716606886e-05,
-      "loss": 0.0461,
+      "epoch": 0.2912,
+      "grad_norm": 0.31600914873730507,
+      "learning_rate": 1.976761251884661e-05,
+      "loss": 0.0134,
       "step": 182
     },
     {
-      "epoch": 0.27395209580838326,
-      "grad_norm": 0.28110581468430873,
-      "learning_rate": 1.9809230387428444e-05,
-      "loss": 0.0445,
+      "epoch": 0.2928,
+      "grad_norm": 0.26927693320635504,
+      "learning_rate": 1.976389420563607e-05,
+      "loss": 0.0125,
       "step": 183
     },
     {
-      "epoch": 0.2754491017964072,
-      "grad_norm": 0.30945413021599966,
-      "learning_rate": 1.980607441406009e-05,
-      "loss": 0.0441,
+      "epoch": 0.2944,
+      "grad_norm": 0.29561899390721047,
+      "learning_rate": 1.9760146735955388e-05,
+      "loss": 0.0143,
       "step": 184
     },
     {
-      "epoch": 0.27694610778443113,
-      "grad_norm": 0.29814499120640103,
-      "learning_rate": 1.9802892804752462e-05,
-      "loss": 0.0434,
+      "epoch": 0.296,
+      "grad_norm": 0.4715335009988575,
+      "learning_rate": 1.975637012099507e-05,
+      "loss": 0.0162,
       "step": 185
     },
     {
-      "epoch": 0.27844311377245506,
-      "grad_norm": 0.34230417502487365,
-      "learning_rate": 1.9799685567823215e-05,
-      "loss": 0.0444,
+      "epoch": 0.2976,
+      "grad_norm": 0.3567896197888384,
+      "learning_rate": 1.9752564372032655e-05,
+      "loss": 0.0097,
       "step": 186
     },
     {
-      "epoch": 0.27994011976047906,
-      "grad_norm": 0.2518591743527993,
-      "learning_rate": 1.9796452711657002e-05,
-      "loss": 0.0397,
+      "epoch": 0.2992,
+      "grad_norm": 0.3840699761704516,
+      "learning_rate": 1.97487295004327e-05,
+      "loss": 0.0127,
       "step": 187
     },
     {
-      "epoch": 0.281437125748503,
-      "grad_norm": 0.3292942776587782,
-      "learning_rate": 1.9793194244705453e-05,
-      "loss": 0.0452,
+      "epoch": 0.3008,
+      "grad_norm": 0.14860107530375916,
+      "learning_rate": 1.974486551764671e-05,
+      "loss": 0.008,
       "step": 188
     },
     {
-      "epoch": 0.28293413173652693,
-      "grad_norm": 0.2582695780716865,
-      "learning_rate": 1.9789910175487147e-05,
-      "loss": 0.0418,
+      "epoch": 0.3024,
+      "grad_norm": 0.3713295730970593,
+      "learning_rate": 1.9740972435213114e-05,
+      "loss": 0.0112,
       "step": 189
     },
     {
-      "epoch": 0.2844311377245509,
-      "grad_norm": 0.2730893776861321,
-      "learning_rate": 1.97866005125876e-05,
-      "loss": 0.0402,
+      "epoch": 0.304,
+      "grad_norm": 0.2519884849457611,
+      "learning_rate": 1.973705026475726e-05,
+      "loss": 0.0113,
       "step": 190
     },
     {
-      "epoch": 0.28592814371257486,
-      "grad_norm": 0.3111167647085528,
-      "learning_rate": 1.9783265264659244e-05,
-      "loss": 0.0395,
+      "epoch": 0.3056,
+      "grad_norm": 0.40990913449865773,
+      "learning_rate": 1.9733099017991342e-05,
+      "loss": 0.0139,
       "step": 191
     },
     {
-      "epoch": 0.2874251497005988,
-      "grad_norm": 0.39076695341386136,
-      "learning_rate": 1.977990444042138e-05,
-      "loss": 0.0514,
+      "epoch": 0.3072,
+      "grad_norm": 0.3385724214686744,
+      "learning_rate": 1.9729118706714377e-05,
+      "loss": 0.01,
       "step": 192
     },
     {
-      "epoch": 0.28892215568862273,
-      "grad_norm": 0.43768952581558807,
-      "learning_rate": 1.9776518048660188e-05,
-      "loss": 0.046,
+      "epoch": 0.3088,
+      "grad_norm": 0.22538894424367692,
+      "learning_rate": 1.972510934281218e-05,
+      "loss": 0.0091,
       "step": 193
     },
     {
-      "epoch": 0.2904191616766467,
-      "grad_norm": 0.3165460062789067,
-      "learning_rate": 1.9773106098228685e-05,
-      "loss": 0.0449,
+      "epoch": 0.3104,
+      "grad_norm": 0.21366259733010207,
+      "learning_rate": 1.9721070938257326e-05,
+      "loss": 0.0131,
       "step": 194
     },
     {
-      "epoch": 0.29191616766467066,
-      "grad_norm": 0.28739375717770865,
-      "learning_rate": 1.97696685980467e-05,
-      "loss": 0.0434,
+      "epoch": 0.312,
+      "grad_norm": 0.23522891058948267,
+      "learning_rate": 1.9717003505109097e-05,
+      "loss": 0.0136,
       "step": 195
     },
     {
-      "epoch": 0.2934131736526946,
-      "grad_norm": 0.32294252936754736,
-      "learning_rate": 1.976620555710087e-05,
-      "loss": 0.0447,
+      "epoch": 0.3136,
+      "grad_norm": 0.19459287049017815,
+      "learning_rate": 1.971290705551347e-05,
+      "loss": 0.0108,
       "step": 196
     },
     {
-      "epoch": 0.2949101796407186,
-      "grad_norm": 0.3345999711440401,
-      "learning_rate": 1.9762716984444585e-05,
-      "loss": 0.0456,
+      "epoch": 0.3152,
+      "grad_norm": 0.3280138838547234,
+      "learning_rate": 1.9708781601703066e-05,
+      "loss": 0.0106,
       "step": 197
     },
     {
-      "epoch": 0.2964071856287425,
-      "grad_norm": 0.31106180218416374,
-      "learning_rate": 1.9759202889197994e-05,
-      "loss": 0.0389,
+      "epoch": 0.3168,
+      "grad_norm": 0.4193956433136829,
+      "learning_rate": 1.970462715599711e-05,
+      "loss": 0.0146,
       "step": 198
     },
     {
-      "epoch": 0.29790419161676646,
-      "grad_norm": 0.25582092602450185,
-      "learning_rate": 1.975566328054797e-05,
-      "loss": 0.0433,
+      "epoch": 0.3184,
+      "grad_norm": 0.18897686427726873,
+      "learning_rate": 1.9700443730801412e-05,
+      "loss": 0.0091,
       "step": 199
     },
     {
-      "epoch": 0.2994011976047904,
-      "grad_norm": 0.2037709453766463,
-      "learning_rate": 1.9752098167748083e-05,
-      "loss": 0.0348,
+      "epoch": 0.32,
+      "grad_norm": 0.24472231112640697,
+      "learning_rate": 1.9696231338608317e-05,
+      "loss": 0.0107,
       "step": 200
     },
     {
-      "epoch": 0.3008982035928144,
-      "grad_norm": 0.34634898932379976,
-      "learning_rate": 1.9748507560118573e-05,
-      "loss": 0.0398,
+      "epoch": 0.3216,
+      "grad_norm": 0.2535191242691839,
+      "learning_rate": 1.9691989991996663e-05,
+      "loss": 0.0113,
       "step": 201
     },
     {
-      "epoch": 0.3023952095808383,
-      "grad_norm": 0.38910405891750727,
-      "learning_rate": 1.974489146704634e-05,
-      "loss": 0.0393,
+      "epoch": 0.3232,
+      "grad_norm": 0.24198007218272746,
+      "learning_rate": 1.9687719703631757e-05,
+      "loss": 0.0092,
       "step": 202
     },
     {
-      "epoch": 0.30389221556886226,
-      "grad_norm": 0.33295863522793434,
-      "learning_rate": 1.9741249897984907e-05,
-      "loss": 0.0384,
+      "epoch": 0.3248,
+      "grad_norm": 0.32135157441024376,
+      "learning_rate": 1.9683420486265328e-05,
+      "loss": 0.0108,
       "step": 203
     },
     {
-      "epoch": 0.30538922155688625,
-      "grad_norm": 0.21725882145501477,
-      "learning_rate": 1.9737582862454398e-05,
-      "loss": 0.034,
+      "epoch": 0.3264,
+      "grad_norm": 0.24878978637482901,
+      "learning_rate": 1.967909235273549e-05,
+      "loss": 0.011,
       "step": 204
     },
     {
-      "epoch": 0.3068862275449102,
-      "grad_norm": 0.3222703155967036,
-      "learning_rate": 1.9733890370041505e-05,
-      "loss": 0.0379,
+      "epoch": 0.328,
+      "grad_norm": 0.28919135086647363,
+      "learning_rate": 1.967473531596671e-05,
+      "loss": 0.0086,
       "step": 205
     },
     {
-      "epoch": 0.3083832335329341,
-      "grad_norm": 0.36543573520717576,
-      "learning_rate": 1.9730172430399494e-05,
-      "loss": 0.0368,
+      "epoch": 0.3296,
+      "grad_norm": 0.3149664501299847,
+      "learning_rate": 1.9670349388969758e-05,
+      "loss": 0.0089,
       "step": 206
     },
     {
-      "epoch": 0.30988023952095806,
-      "grad_norm": 0.2972736130414979,
-      "learning_rate": 1.972642905324813e-05,
-      "loss": 0.0387,
+      "epoch": 0.3312,
+      "grad_norm": 0.46074087094839206,
+      "learning_rate": 1.966593458484168e-05,
+      "loss": 0.0115,
       "step": 207
     },
     {
-      "epoch": 0.31137724550898205,
-      "grad_norm": 0.27528054613918207,
-      "learning_rate": 1.9722660248373707e-05,
-      "loss": 0.0337,
+      "epoch": 0.3328,
+      "grad_norm": 0.32453985066034696,
+      "learning_rate": 1.9661490916765752e-05,
+      "loss": 0.0078,
       "step": 208
     },
     {
-      "epoch": 0.312874251497006,
-      "grad_norm": 0.2514931870827675,
-      "learning_rate": 1.971886602562897e-05,
-      "loss": 0.0359,
+      "epoch": 0.3344,
+      "grad_norm": 0.21726445303846265,
+      "learning_rate": 1.9657018398011435e-05,
+      "loss": 0.0075,
       "step": 209
     },
     {
-      "epoch": 0.3143712574850299,
-      "grad_norm": 0.33119244111260354,
-      "learning_rate": 1.9715046394933126e-05,
-      "loss": 0.0377,
+      "epoch": 0.336,
+      "grad_norm": 0.29005944041139997,
+      "learning_rate": 1.9652517041934357e-05,
+      "loss": 0.0065,
       "step": 210
     },
     {
-      "epoch": 0.3158682634730539,
-      "grad_norm": 0.2644246425467484,
-      "learning_rate": 1.971120136627181e-05,
-      "loss": 0.0391,
+      "epoch": 0.3376,
+      "grad_norm": 0.13171213095514112,
+      "learning_rate": 1.9647986861976246e-05,
+      "loss": 0.0044,
       "step": 211
     },
     {
-      "epoch": 0.31736526946107785,
-      "grad_norm": 0.23488800200566334,
-      "learning_rate": 1.970733094969704e-05,
-      "loss": 0.0305,
+      "epoch": 0.3392,
+      "grad_norm": 0.22782619185124792,
+      "learning_rate": 1.9643427871664912e-05,
+      "loss": 0.0062,
       "step": 212
     },
     {
-      "epoch": 0.3188622754491018,
-      "grad_norm": 0.3336820637038372,
-      "learning_rate": 1.970343515532722e-05,
-      "loss": 0.0347,
+      "epoch": 0.3408,
+      "grad_norm": 0.2248678754440452,
+      "learning_rate": 1.9638840084614182e-05,
+      "loss": 0.0062,
       "step": 213
     },
     {
-      "epoch": 0.3203592814371258,
-      "grad_norm": 0.32537350598910286,
-      "learning_rate": 1.969951399334709e-05,
-      "loss": 0.0383,
+      "epoch": 0.3424,
+      "grad_norm": 0.3625913388024184,
+      "learning_rate": 1.963422351452389e-05,
+      "loss": 0.0084,
       "step": 214
     },
     {
-      "epoch": 0.3218562874251497,
-      "grad_norm": 0.4477662129282192,
-      "learning_rate": 1.9695567474007713e-05,
-      "loss": 0.0375,
+      "epoch": 0.344,
+      "grad_norm": 0.15657625546350282,
+      "learning_rate": 1.9629578175179823e-05,
+      "loss": 0.0076,
       "step": 215
     },
     {
-      "epoch": 0.32335329341317365,
-      "grad_norm": 0.39632766587050555,
-      "learning_rate": 1.9691595607626443e-05,
-      "loss": 0.0306,
+      "epoch": 0.3456,
+      "grad_norm": 0.1517982622637864,
+      "learning_rate": 1.9624904080453656e-05,
+      "loss": 0.005,
       "step": 216
     },
     {
-      "epoch": 0.3248502994011976,
-      "grad_norm": 0.31164368475238186,
-      "learning_rate": 1.9687598404586895e-05,
-      "loss": 0.0317,
+      "epoch": 0.3472,
+      "grad_norm": 0.2758823005848304,
+      "learning_rate": 1.9620201244302952e-05,
+      "loss": 0.01,
       "step": 217
     },
     {
-      "epoch": 0.3263473053892216,
-      "grad_norm": 0.29356080957866204,
-      "learning_rate": 1.9683575875338922e-05,
-      "loss": 0.0319,
+      "epoch": 0.3488,
+      "grad_norm": 0.19837305214001089,
+      "learning_rate": 1.9615469680771097e-05,
+      "loss": 0.005,
       "step": 218
     },
     {
-      "epoch": 0.3278443113772455,
-      "grad_norm": 0.3170039835076496,
-      "learning_rate": 1.9679528030398596e-05,
-      "loss": 0.0329,
+      "epoch": 0.3504,
+      "grad_norm": 0.18124188268088945,
+      "learning_rate": 1.9610709403987248e-05,
+      "loss": 0.0045,
       "step": 219
     },
     {
-      "epoch": 0.32934131736526945,
-      "grad_norm": 0.30665504238301144,
-      "learning_rate": 1.9675454880348158e-05,
-      "loss": 0.0279,
+      "epoch": 0.352,
+      "grad_norm": 0.1894657499100413,
+      "learning_rate": 1.960592042816632e-05,
+      "loss": 0.0065,
       "step": 220
     },
     {
-      "epoch": 0.33083832335329344,
-      "grad_norm": 0.25789142501667217,
-      "learning_rate": 1.967135643583601e-05,
-      "loss": 0.029,
+      "epoch": 0.3536,
+      "grad_norm": 0.14435370426581687,
+      "learning_rate": 1.9601102767608924e-05,
+      "loss": 0.0043,
       "step": 221
     },
     {
-      "epoch": 0.3323353293413174,
-      "grad_norm": 0.29710874117629504,
-      "learning_rate": 1.966723270757669e-05,
-      "loss": 0.0339,
+      "epoch": 0.3552,
+      "grad_norm": 0.1695642548479439,
+      "learning_rate": 1.9596256436701324e-05,
+      "loss": 0.0046,
       "step": 222
     },
     {
-      "epoch": 0.3338323353293413,
-      "grad_norm": 0.3797724903024257,
-      "learning_rate": 1.966308370635082e-05,
-      "loss": 0.0317,
+      "epoch": 0.3568,
+      "grad_norm": 0.16480315731755024,
+      "learning_rate": 1.95913814499154e-05,
+      "loss": 0.005,
       "step": 223
     },
     {
-      "epoch": 0.33532934131736525,
-      "grad_norm": 0.3234195979276272,
-      "learning_rate": 1.96589094430051e-05,
-      "loss": 0.0282,
+      "epoch": 0.3584,
+      "grad_norm": 0.20501655629665766,
+      "learning_rate": 1.9586477821808597e-05,
+      "loss": 0.0048,
       "step": 224
     },
     {
-      "epoch": 0.33682634730538924,
-      "grad_norm": 0.3322138017326269,
-      "learning_rate": 1.965470992845227e-05,
-      "loss": 0.0289,
+      "epoch": 0.36,
+      "grad_norm": 0.24808881256292714,
+      "learning_rate": 1.95815455670239e-05,
+      "loss": 0.0068,
       "step": 225
     },
     {
-      "epoch": 0.3383233532934132,
-      "grad_norm": 0.24818051146540246,
-      "learning_rate": 1.9650485173671092e-05,
-      "loss": 0.0221,
+      "epoch": 0.3616,
+      "grad_norm": 0.20807812279666899,
+      "learning_rate": 1.957658470028977e-05,
+      "loss": 0.0046,
       "step": 226
     },
     {
-      "epoch": 0.3398203592814371,
-      "grad_norm": 0.42659472820256933,
-      "learning_rate": 1.96462351897063e-05,
-      "loss": 0.031,
+      "epoch": 0.3632,
+      "grad_norm": 0.12976277315186674,
+      "learning_rate": 1.9571595236420103e-05,
+      "loss": 0.003,
       "step": 227
     },
     {
-      "epoch": 0.3413173652694611,
-      "grad_norm": 0.4005980652683074,
-      "learning_rate": 1.9641959987668603e-05,
-      "loss": 0.0365,
+      "epoch": 0.3648,
+      "grad_norm": 0.35081120007947303,
+      "learning_rate": 1.95665771903142e-05,
+      "loss": 0.0045,
       "step": 228
     },
     {
-      "epoch": 0.34281437125748504,
-      "grad_norm": 0.4685928568147018,
-      "learning_rate": 1.9637659578734615e-05,
-      "loss": 0.0327,
+      "epoch": 0.3664,
+      "grad_norm": 0.18678758567379564,
+      "learning_rate": 1.9561530576956703e-05,
+      "loss": 0.0025,
       "step": 229
     },
     {
-      "epoch": 0.344311377245509,
-      "grad_norm": 0.3863947245851545,
-      "learning_rate": 1.9633333974146867e-05,
-      "loss": 0.0267,
+      "epoch": 0.368,
+      "grad_norm": 0.2726852511816301,
+      "learning_rate": 1.9556455411417575e-05,
+      "loss": 0.0067,
       "step": 230
     },
     {
-      "epoch": 0.3458083832335329,
-      "grad_norm": 0.5573187059327918,
-      "learning_rate": 1.9628983185213747e-05,
-      "loss": 0.0279,
+      "epoch": 0.3696,
+      "grad_norm": 0.1986239338872355,
+      "learning_rate": 1.955135170885202e-05,
+      "loss": 0.0066,
       "step": 231
     },
     {
-      "epoch": 0.3473053892215569,
-      "grad_norm": 0.3063425475105623,
-      "learning_rate": 1.962460722330949e-05,
-      "loss": 0.0254,
+      "epoch": 0.3712,
+      "grad_norm": 0.2686457903895003,
+      "learning_rate": 1.9546219484500475e-05,
+      "loss": 0.0058,
       "step": 232
     },
     {
-      "epoch": 0.34880239520958084,
-      "grad_norm": 0.25300385871670333,
-      "learning_rate": 1.9620206099874135e-05,
-      "loss": 0.0238,
+      "epoch": 0.3728,
+      "grad_norm": 0.13534531600226646,
+      "learning_rate": 1.9541058753688538e-05,
+      "loss": 0.0054,
       "step": 233
     },
     {
-      "epoch": 0.3502994011976048,
-      "grad_norm": 0.33330643618442385,
-      "learning_rate": 1.96157798264135e-05,
-      "loss": 0.023,
+      "epoch": 0.3744,
+      "grad_norm": 0.15207516826062709,
+      "learning_rate": 1.9535869531826938e-05,
+      "loss": 0.0035,
       "step": 234
     },
     {
-      "epoch": 0.35179640718562877,
-      "grad_norm": 0.3374148649272243,
-      "learning_rate": 1.9611328414499164e-05,
-      "loss": 0.0245,
+      "epoch": 0.376,
+      "grad_norm": 0.33924323511495663,
+      "learning_rate": 1.9530651834411477e-05,
+      "loss": 0.0078,
       "step": 235
     },
     {
-      "epoch": 0.3532934131736527,
-      "grad_norm": 0.38153658162458814,
-      "learning_rate": 1.9606851875768404e-05,
-      "loss": 0.0227,
+      "epoch": 0.3776,
+      "grad_norm": 0.3777186216574476,
+      "learning_rate": 1.952540567702299e-05,
+      "loss": 0.0021,
       "step": 236
     },
     {
-      "epoch": 0.35479041916167664,
-      "grad_norm": 0.31854741219630106,
-      "learning_rate": 1.9602350221924208e-05,
-      "loss": 0.0194,
+      "epoch": 0.3792,
+      "grad_norm": 0.130829431828979,
+      "learning_rate": 1.95201310753273e-05,
+      "loss": 0.002,
       "step": 237
     },
     {
-      "epoch": 0.3562874251497006,
-      "grad_norm": 0.3245371702157616,
-      "learning_rate": 1.9597823464735204e-05,
-      "loss": 0.0231,
+      "epoch": 0.3808,
+      "grad_norm": 0.17763136606881505,
+      "learning_rate": 1.951482804507517e-05,
+      "loss": 0.0045,
       "step": 238
     },
     {
-      "epoch": 0.35778443113772457,
-      "grad_norm": 0.3260937932412784,
-      "learning_rate": 1.9593271616035666e-05,
-      "loss": 0.0201,
+      "epoch": 0.3824,
+      "grad_norm": 0.1530279950906812,
+      "learning_rate": 1.9509496602102253e-05,
+      "loss": 0.0035,
       "step": 239
     },
     {
-      "epoch": 0.3592814371257485,
-      "grad_norm": 0.313463122278002,
-      "learning_rate": 1.9588694687725448e-05,
-      "loss": 0.0203,
+      "epoch": 0.384,
+      "grad_norm": 0.12479236693345638,
+      "learning_rate": 1.9504136762329046e-05,
+      "loss": 0.0026,
       "step": 240
     },
     {
-      "epoch": 0.36077844311377244,
-      "grad_norm": 0.4452480165482146,
-      "learning_rate": 1.9584092691769978e-05,
-      "loss": 0.0213,
+      "epoch": 0.3856,
+      "grad_norm": 0.18872893104962732,
+      "learning_rate": 1.9498748541760845e-05,
+      "loss": 0.002,
       "step": 241
     },
     {
-      "epoch": 0.36227544910179643,
-      "grad_norm": 0.38506762755559276,
-      "learning_rate": 1.9579465640200213e-05,
-      "loss": 0.0227,
+      "epoch": 0.3872,
+      "grad_norm": 0.2839066754069719,
+      "learning_rate": 1.949333195648769e-05,
+      "loss": 0.0034,
       "step": 242
     },
     {
-      "epoch": 0.36377245508982037,
-      "grad_norm": 0.39214871815864943,
-      "learning_rate": 1.957481354511262e-05,
-      "loss": 0.0199,
+      "epoch": 0.3888,
+      "grad_norm": 0.29568207016971043,
+      "learning_rate": 1.9487887022684336e-05,
+      "loss": 0.0044,
       "step": 243
     },
     {
-      "epoch": 0.3652694610778443,
-      "grad_norm": 0.3539339228414202,
-      "learning_rate": 1.957013641866913e-05,
-      "loss": 0.0225,
+      "epoch": 0.3904,
+      "grad_norm": 0.22126066178256676,
+      "learning_rate": 1.9482413756610175e-05,
+      "loss": 0.0035,
       "step": 244
     },
     {
-      "epoch": 0.36676646706586824,
-      "grad_norm": 0.3539330765036243,
-      "learning_rate": 1.9565434273097114e-05,
-      "loss": 0.0207,
+      "epoch": 0.392,
+      "grad_norm": 0.19602278733536263,
+      "learning_rate": 1.947691217460921e-05,
+      "loss": 0.0032,
       "step": 245
     },
     {
-      "epoch": 0.36826347305389223,
-      "grad_norm": 0.3753514827965324,
-      "learning_rate": 1.9560707120689354e-05,
-      "loss": 0.021,
+      "epoch": 0.3936,
+      "grad_norm": 0.4201801990164135,
+      "learning_rate": 1.9471382293110004e-05,
+      "loss": 0.0038,
       "step": 246
     },
     {
-      "epoch": 0.36976047904191617,
-      "grad_norm": 0.40608828993861595,
-      "learning_rate": 1.9555954973804013e-05,
-      "loss": 0.0181,
+      "epoch": 0.3952,
+      "grad_norm": 0.1668178273336579,
+      "learning_rate": 1.946582412862562e-05,
+      "loss": 0.004,
       "step": 247
     },
     {
-      "epoch": 0.3712574850299401,
-      "grad_norm": 0.3303053398457487,
-      "learning_rate": 1.9551177844864577e-05,
-      "loss": 0.0208,
+      "epoch": 0.3968,
+      "grad_norm": 0.2201357009016026,
+      "learning_rate": 1.9460237697753577e-05,
+      "loss": 0.0055,
       "step": 248
     },
     {
-      "epoch": 0.3727544910179641,
-      "grad_norm": 0.41949758179222085,
-      "learning_rate": 1.954637574635986e-05,
-      "loss": 0.0212,
+      "epoch": 0.3984,
+      "grad_norm": 0.32997962181070606,
+      "learning_rate": 1.9454623017175814e-05,
+      "loss": 0.0039,
       "step": 249
     },
     {
-      "epoch": 0.37425149700598803,
-      "grad_norm": 0.2764763651977154,
-      "learning_rate": 1.954154869084395e-05,
-      "loss": 0.0159,
+      "epoch": 0.4,
+      "grad_norm": 0.10041694708026828,
+      "learning_rate": 1.9448980103658613e-05,
+      "loss": 0.0023,
       "step": 250
     },
     {
-      "epoch": 0.37574850299401197,
-      "grad_norm": 0.39212670974147856,
-      "learning_rate": 1.9536696690936177e-05,
-      "loss": 0.0189,
+      "epoch": 0.4016,
+      "grad_norm": 0.2468138214155966,
+      "learning_rate": 1.9443308974052574e-05,
+      "loss": 0.005,
       "step": 251
     },
     {
-      "epoch": 0.3772455089820359,
-      "grad_norm": 0.32025921970118243,
-      "learning_rate": 1.9531819759321083e-05,
-      "loss": 0.0169,
+      "epoch": 0.4032,
+      "grad_norm": 0.15123509597682863,
+      "learning_rate": 1.943760964529255e-05,
+      "loss": 0.0024,
       "step": 252
     },
     {
-      "epoch": 0.3787425149700599,
-      "grad_norm": 0.3429001255459339,
-      "learning_rate": 1.9526917908748395e-05,
-      "loss": 0.0232,
+      "epoch": 0.4048,
+      "grad_norm": 0.08875285519072525,
+      "learning_rate": 1.9431882134397596e-05,
+      "loss": 0.0016,
       "step": 253
     },
     {
-      "epoch": 0.38023952095808383,
-      "grad_norm": 0.37665262959840423,
-      "learning_rate": 1.9521991152032974e-05,
-      "loss": 0.023,
+      "epoch": 0.4064,
+      "grad_norm": 0.21246156328609328,
+      "learning_rate": 1.9426126458470936e-05,
+      "loss": 0.0024,
       "step": 254
     },
     {
-      "epoch": 0.38173652694610777,
-      "grad_norm": 0.2649408332614438,
-      "learning_rate": 1.9517039502054805e-05,
-      "loss": 0.0153,
+      "epoch": 0.408,
+      "grad_norm": 0.07721338540221492,
+      "learning_rate": 1.9420342634699893e-05,
+      "loss": 0.0013,
       "step": 255
     },
     {
-      "epoch": 0.38323353293413176,
-      "grad_norm": 0.2772926318508054,
-      "learning_rate": 1.9512062971758937e-05,
-      "loss": 0.0177,
+      "epoch": 0.4096,
+      "grad_norm": 0.19326745142293367,
+      "learning_rate": 1.9414530680355837e-05,
+      "loss": 0.0017,
       "step": 256
     },
     {
-      "epoch": 0.3847305389221557,
-      "grad_norm": 0.333717122655441,
-      "learning_rate": 1.950706157415548e-05,
-      "loss": 0.0175,
+      "epoch": 0.4112,
+      "grad_norm": 0.17732669112234006,
+      "learning_rate": 1.9408690612794146e-05,
+      "loss": 0.0045,
       "step": 257
     },
     {
-      "epoch": 0.38622754491017963,
-      "grad_norm": 0.2881853890457447,
-      "learning_rate": 1.9502035322319546e-05,
-      "loss": 0.0197,
+      "epoch": 0.4128,
+      "grad_norm": 0.2250739228373143,
+      "learning_rate": 1.9402822449454154e-05,
+      "loss": 0.0025,
       "step": 258
     },
     {
-      "epoch": 0.38772455089820357,
-      "grad_norm": 0.3011392842260328,
-      "learning_rate": 1.949698422939122e-05,
-      "loss": 0.0194,
+      "epoch": 0.4144,
+      "grad_norm": 0.1541916050350658,
+      "learning_rate": 1.9396926207859085e-05,
+      "loss": 0.0019,
       "step": 259
     },
     {
-      "epoch": 0.38922155688622756,
-      "grad_norm": 0.25182071202500195,
-      "learning_rate": 1.9491908308575537e-05,
-      "loss": 0.0168,
+      "epoch": 0.416,
+      "grad_norm": 0.2841457747343044,
+      "learning_rate": 1.939100190561601e-05,
+      "loss": 0.0034,
       "step": 260
     },
     {
-      "epoch": 0.3907185628742515,
-      "grad_norm": 0.33730858021035043,
-      "learning_rate": 1.948680757314243e-05,
-      "loss": 0.0132,
+      "epoch": 0.4176,
+      "grad_norm": 0.30187004448055127,
+      "learning_rate": 1.9385049560415794e-05,
+      "loss": 0.0043,
       "step": 261
     },
     {
-      "epoch": 0.39221556886227543,
-      "grad_norm": 0.3785903917651171,
-      "learning_rate": 1.9481682036426707e-05,
-      "loss": 0.0151,
+      "epoch": 0.4192,
+      "grad_norm": 0.0886576343921295,
+      "learning_rate": 1.9379069190033042e-05,
+      "loss": 0.001,
       "step": 262
     },
     {
-      "epoch": 0.3937125748502994,
-      "grad_norm": 0.21775764169004,
-      "learning_rate": 1.9476531711828027e-05,
-      "loss": 0.0133,
+      "epoch": 0.4208,
+      "grad_norm": 0.40808645948350025,
+      "learning_rate": 1.9373060812326053e-05,
+      "loss": 0.0072,
       "step": 263
     },
     {
-      "epoch": 0.39520958083832336,
-      "grad_norm": 0.22229766801022924,
-      "learning_rate": 1.9471356612810832e-05,
-      "loss": 0.0103,
+      "epoch": 0.4224,
+      "grad_norm": 0.1549022328201513,
+      "learning_rate": 1.936702444523675e-05,
+      "loss": 0.0012,
       "step": 264
     },
     {
-      "epoch": 0.3967065868263473,
-      "grad_norm": 0.3032194913024956,
-      "learning_rate": 1.9466156752904344e-05,
-      "loss": 0.0131,
+      "epoch": 0.424,
+      "grad_norm": 0.21280823018695852,
+      "learning_rate": 1.9360960106790645e-05,
+      "loss": 0.0046,
       "step": 265
     },
     {
-      "epoch": 0.39820359281437123,
-      "grad_norm": 0.3924806246674395,
-      "learning_rate": 1.9460932145702514e-05,
-      "loss": 0.0164,
+      "epoch": 0.4256,
+      "grad_norm": 0.18046256341062997,
+      "learning_rate": 1.9354867815096772e-05,
+      "loss": 0.0038,
       "step": 266
     },
     {
-      "epoch": 0.3997005988023952,
-      "grad_norm": 0.2958650726177419,
-      "learning_rate": 1.9455682804863986e-05,
-      "loss": 0.0132,
+      "epoch": 0.4272,
+      "grad_norm": 0.08032384959654924,
+      "learning_rate": 1.9348747588347637e-05,
+      "loss": 0.0009,
       "step": 267
     },
     {
-      "epoch": 0.40119760479041916,
-      "grad_norm": 0.32232453177932663,
-      "learning_rate": 1.9450408744112072e-05,
-      "loss": 0.016,
+      "epoch": 0.4288,
+      "grad_norm": 0.22947414894255785,
+      "learning_rate": 1.9342599444819167e-05,
+      "loss": 0.0027,
       "step": 268
     },
     {
-      "epoch": 0.4026946107784431,
-      "grad_norm": 0.3492194562524053,
-      "learning_rate": 1.9445109977234704e-05,
-      "loss": 0.0145,
+      "epoch": 0.4304,
+      "grad_norm": 0.1845631124409086,
+      "learning_rate": 1.9336423402870655e-05,
+      "loss": 0.002,
       "step": 269
     },
     {
-      "epoch": 0.4041916167664671,
-      "grad_norm": 0.374197844485198,
-      "learning_rate": 1.9439786518084404e-05,
-      "loss": 0.0135,
+      "epoch": 0.432,
+      "grad_norm": 0.1400394929275762,
+      "learning_rate": 1.9330219480944693e-05,
+      "loss": 0.0022,
       "step": 270
     },
     {
-      "epoch": 0.405688622754491,
-      "grad_norm": 0.3268312149590773,
-      "learning_rate": 1.9434438380578248e-05,
-      "loss": 0.0164,
+      "epoch": 0.4336,
+      "grad_norm": 0.09366624568050941,
+      "learning_rate": 1.932398769756714e-05,
+      "loss": 0.0011,
       "step": 271
     },
     {
-      "epoch": 0.40718562874251496,
-      "grad_norm": 0.22923876520135972,
-      "learning_rate": 1.9429065578697833e-05,
-      "loss": 0.0121,
+      "epoch": 0.4352,
+      "grad_norm": 0.020707890357983063,
+      "learning_rate": 1.931772807134704e-05,
+      "loss": 0.0005,
       "step": 272
     },
     {
-      "epoch": 0.4086826347305389,
-      "grad_norm": 0.29090023648655766,
-      "learning_rate": 1.942366812648922e-05,
-      "loss": 0.0103,
+      "epoch": 0.4368,
+      "grad_norm": 0.18065363744047988,
+      "learning_rate": 1.9311440620976597e-05,
+      "loss": 0.0026,
       "step": 273
     },
     {
-      "epoch": 0.4101796407185629,
-      "grad_norm": 0.3117702117434219,
-      "learning_rate": 1.941824603806293e-05,
-      "loss": 0.0154,
+      "epoch": 0.4384,
+      "grad_norm": 0.2604405111691777,
+      "learning_rate": 1.9305125365231087e-05,
+      "loss": 0.0024,
       "step": 274
     },
     {
-      "epoch": 0.4116766467065868,
-      "grad_norm": 0.38775196821617675,
-      "learning_rate": 1.9412799327593884e-05,
-      "loss": 0.0106,
+      "epoch": 0.44,
+      "grad_norm": 0.10457477256845,
+      "learning_rate": 1.9298782322968817e-05,
+      "loss": 0.0015,
       "step": 275
     },
     {
-      "epoch": 0.41317365269461076,
-      "grad_norm": 0.2310944482235676,
-      "learning_rate": 1.9407328009321367e-05,
-      "loss": 0.0091,
+      "epoch": 0.4416,
+      "grad_norm": 0.09114897913357566,
+      "learning_rate": 1.929241151313108e-05,
+      "loss": 0.0017,
       "step": 276
     },
     {
-      "epoch": 0.41467065868263475,
-      "grad_norm": 0.23882380072426646,
-      "learning_rate": 1.940183209754901e-05,
-      "loss": 0.0096,
+      "epoch": 0.4432,
+      "grad_norm": 0.15876593967910152,
+      "learning_rate": 1.9286012954742078e-05,
+      "loss": 0.0012,
       "step": 277
     },
     {
-      "epoch": 0.4161676646706587,
-      "grad_norm": 0.47743508487331154,
-      "learning_rate": 1.939631160664472e-05,
-      "loss": 0.0138,
+      "epoch": 0.4448,
+      "grad_norm": 0.05828086512032646,
+      "learning_rate": 1.9279586666908886e-05,
+      "loss": 0.0007,
       "step": 278
     },
     {
-      "epoch": 0.4176646706586826,
-      "grad_norm": 0.24267162900266767,
-      "learning_rate": 1.939076655104068e-05,
-      "loss": 0.0127,
+      "epoch": 0.4464,
+      "grad_norm": 0.1754559731453586,
+      "learning_rate": 1.9273132668821363e-05,
+      "loss": 0.0045,
       "step": 279
     },
     {
-      "epoch": 0.41916167664670656,
-      "grad_norm": 0.25724021933042784,
-      "learning_rate": 1.9385196945233276e-05,
-      "loss": 0.008,
+      "epoch": 0.448,
+      "grad_norm": 0.16575326861643688,
+      "learning_rate": 1.9266650979752137e-05,
+      "loss": 0.0049,
       "step": 280
     },
     {
-      "epoch": 0.42065868263473055,
-      "grad_norm": 0.2863894617393607,
-      "learning_rate": 1.9379602803783088e-05,
-      "loss": 0.0118,
+      "epoch": 0.4496,
+      "grad_norm": 0.04984852486915607,
+      "learning_rate": 1.9260141619056507e-05,
+      "loss": 0.0006,
       "step": 281
     },
     {
-      "epoch": 0.4221556886227545,
-      "grad_norm": 0.24028996046153225,
-      "learning_rate": 1.9373984141314827e-05,
-      "loss": 0.0123,
+      "epoch": 0.4512,
+      "grad_norm": 0.08051247848310401,
+      "learning_rate": 1.925360460617242e-05,
+      "loss": 0.0012,
       "step": 282
     },
     {
-      "epoch": 0.4236526946107784,
-      "grad_norm": 0.24877798633512982,
-      "learning_rate": 1.936834097251732e-05,
-      "loss": 0.01,
+      "epoch": 0.4528,
+      "grad_norm": 0.07838917524463092,
+      "learning_rate": 1.924703996062038e-05,
+      "loss": 0.001,
       "step": 283
     },
     {
-      "epoch": 0.4251497005988024,
-      "grad_norm": 0.24480091385137137,
-      "learning_rate": 1.936267331214346e-05,
-      "loss": 0.0088,
+      "epoch": 0.4544,
+      "grad_norm": 0.11163090665148108,
+      "learning_rate": 1.9240447702003422e-05,
+      "loss": 0.0019,
       "step": 284
     },
     {
-      "epoch": 0.42664670658682635,
-      "grad_norm": 0.378474089158859,
-      "learning_rate": 1.9356981175010155e-05,
-      "loss": 0.0138,
+      "epoch": 0.456,
+      "grad_norm": 0.13500917519862524,
+      "learning_rate": 1.9233827850007028e-05,
+      "loss": 0.0039,
       "step": 285
     },
     {
-      "epoch": 0.4281437125748503,
-      "grad_norm": 0.2731112425501055,
-      "learning_rate": 1.9351264575998322e-05,
-      "loss": 0.0111,
+      "epoch": 0.4576,
+      "grad_norm": 0.05401903025575265,
+      "learning_rate": 1.9227180424399082e-05,
+      "loss": 0.0009,
       "step": 286
     },
     {
-      "epoch": 0.4296407185628742,
-      "grad_norm": 0.45385449828176233,
-      "learning_rate": 1.934552353005281e-05,
-      "loss": 0.0144,
+      "epoch": 0.4592,
+      "grad_norm": 0.15767776098462866,
+      "learning_rate": 1.9220505445029803e-05,
+      "loss": 0.0034,
       "step": 287
     },
     {
-      "epoch": 0.4311377245508982,
-      "grad_norm": 0.2426064120408043,
-      "learning_rate": 1.9339758052182396e-05,
-      "loss": 0.0113,
+      "epoch": 0.4608,
+      "grad_norm": 0.18565041494539808,
+      "learning_rate": 1.9213802931831697e-05,
+      "loss": 0.0021,
       "step": 288
     },
     {
-      "epoch": 0.43263473053892215,
-      "grad_norm": 0.2840729140180023,
-      "learning_rate": 1.933396815745972e-05,
-      "loss": 0.0088,
+      "epoch": 0.4624,
+      "grad_norm": 0.07266220337631138,
+      "learning_rate": 1.9207072904819484e-05,
+      "loss": 0.0008,
       "step": 289
     },
     {
-      "epoch": 0.4341317365269461,
-      "grad_norm": 0.4447034848523609,
-      "learning_rate": 1.932815386102125e-05,
-      "loss": 0.0104,
+      "epoch": 0.464,
+      "grad_norm": 0.0937701933953401,
+      "learning_rate": 1.9200315384090045e-05,
+      "loss": 0.0013,
       "step": 290
     },
     {
-      "epoch": 0.4356287425149701,
-      "grad_norm": 0.27318626797446804,
-      "learning_rate": 1.9322315178067255e-05,
-      "loss": 0.0082,
+      "epoch": 0.4656,
+      "grad_norm": 0.08508182747781863,
+      "learning_rate": 1.9193530389822364e-05,
+      "loss": 0.0007,
       "step": 291
     },
     {
-      "epoch": 0.437125748502994,
-      "grad_norm": 0.28620936956656057,
-      "learning_rate": 1.931645212386176e-05,
-      "loss": 0.0113,
+      "epoch": 0.4672,
+      "grad_norm": 0.08968822087577417,
+      "learning_rate": 1.9186717942277466e-05,
+      "loss": 0.001,
       "step": 292
     },
     {
-      "epoch": 0.43862275449101795,
-      "grad_norm": 0.26386101205038476,
-      "learning_rate": 1.9310564713732495e-05,
-      "loss": 0.0082,
+      "epoch": 0.4688,
+      "grad_norm": 0.04091429398607774,
+      "learning_rate": 1.9179878061798347e-05,
+      "loss": 0.0006,
       "step": 293
     },
     {
-      "epoch": 0.44011976047904194,
-      "grad_norm": 0.40806356163691765,
-      "learning_rate": 1.9304652963070868e-05,
-      "loss": 0.0107,
+      "epoch": 0.4704,
+      "grad_norm": 0.14716595852890726,
+      "learning_rate": 1.9173010768809934e-05,
+      "loss": 0.0017,
       "step": 294
     },
     {
-      "epoch": 0.4416167664670659,
-      "grad_norm": 0.19248450148831384,
-      "learning_rate": 1.929871688733192e-05,
-      "loss": 0.0091,
+      "epoch": 0.472,
+      "grad_norm": 0.19756136061689594,
+      "learning_rate": 1.9166116083819002e-05,
+      "loss": 0.0014,
       "step": 295
     },
     {
-      "epoch": 0.4431137724550898,
-      "grad_norm": 0.28001071490300894,
-      "learning_rate": 1.929275650203429e-05,
-      "loss": 0.0086,
+      "epoch": 0.4736,
+      "grad_norm": 0.2356635049518032,
+      "learning_rate": 1.915919402741413e-05,
+      "loss": 0.004,
       "step": 296
     },
     {
-      "epoch": 0.44461077844311375,
-      "grad_norm": 0.19281433588980926,
-      "learning_rate": 1.9286771822760155e-05,
-      "loss": 0.0075,
+      "epoch": 0.4752,
+      "grad_norm": 0.11227394052898403,
+      "learning_rate": 1.915224462026563e-05,
+      "loss": 0.0034,
       "step": 297
     },
     {
-      "epoch": 0.44610778443113774,
-      "grad_norm": 0.30951373486991235,
-      "learning_rate": 1.9280762865155217e-05,
-      "loss": 0.0126,
+      "epoch": 0.4768,
+      "grad_norm": 0.16940535443283003,
+      "learning_rate": 1.9145267883125483e-05,
+      "loss": 0.002,
       "step": 298
     },
     {
-      "epoch": 0.4476047904191617,
-      "grad_norm": 0.24670484913137125,
-      "learning_rate": 1.9274729644928646e-05,
-      "loss": 0.0079,
+      "epoch": 0.4784,
+      "grad_norm": 0.13956098006188228,
+      "learning_rate": 1.913826383682729e-05,
+      "loss": 0.002,
       "step": 299
     },
     {
-      "epoch": 0.4491017964071856,
-      "grad_norm": 0.21597250129732382,
-      "learning_rate": 1.9268672177853037e-05,
-      "loss": 0.0082,
+      "epoch": 0.48,
+      "grad_norm": 0.11477156950801032,
+      "learning_rate": 1.913123250228619e-05,
+      "loss": 0.0015,
       "step": 300
     },
     {
-      "epoch": 0.4505988023952096,
-      "grad_norm": 0.25565404173278816,
-      "learning_rate": 1.926259047976438e-05,
-      "loss": 0.0078,
+      "epoch": 0.4816,
+      "grad_norm": 0.058908329416148174,
+      "learning_rate": 1.912417390049882e-05,
+      "loss": 0.0009,
       "step": 301
     },
     {
-      "epoch": 0.45209580838323354,
-      "grad_norm": 0.25018467551033313,
-      "learning_rate": 1.9256484566562002e-05,
-      "loss": 0.0082,
+      "epoch": 0.4832,
+      "grad_norm": 0.026814308548944957,
+      "learning_rate": 1.9117088052543233e-05,
+      "loss": 0.0005,
       "step": 302
     },
     {
-      "epoch": 0.4535928143712575,
-      "grad_norm": 0.24949067973854458,
-      "learning_rate": 1.9250354454208545e-05,
-      "loss": 0.0084,
+      "epoch": 0.4848,
+      "grad_norm": 0.12194884329880719,
+      "learning_rate": 1.9109974979578852e-05,
+      "loss": 0.0018,
       "step": 303
     },
     {
-      "epoch": 0.4550898203592814,
-      "grad_norm": 0.31313857717111415,
-      "learning_rate": 1.924420015872991e-05,
-      "loss": 0.0102,
+      "epoch": 0.4864,
+      "grad_norm": 0.09882593225561369,
+      "learning_rate": 1.9102834702846387e-05,
+      "loss": 0.0009,
       "step": 304
     },
     {
-      "epoch": 0.4565868263473054,
-      "grad_norm": 0.20485306938289613,
-      "learning_rate": 1.923802169621522e-05,
-      "loss": 0.0075,
+      "epoch": 0.488,
+      "grad_norm": 0.30833315654476523,
+      "learning_rate": 1.909566724366779e-05,
+      "loss": 0.0028,
       "step": 305
     },
     {
-      "epoch": 0.45808383233532934,
-      "grad_norm": 0.20454714142094618,
-      "learning_rate": 1.9231819082816786e-05,
-      "loss": 0.0076,
+      "epoch": 0.4896,
+      "grad_norm": 0.2594408640023605,
+      "learning_rate": 1.9088472623446182e-05,
+      "loss": 0.0036,
       "step": 306
     },
     {
-      "epoch": 0.4595808383233533,
-      "grad_norm": 0.24763142577873085,
-      "learning_rate": 1.9225592334750036e-05,
-      "loss": 0.0081,
+      "epoch": 0.4912,
+      "grad_norm": 0.24354763997282527,
+      "learning_rate": 1.9081250863665794e-05,
+      "loss": 0.0057,
       "step": 307
     },
     {
-      "epoch": 0.46107784431137727,
-      "grad_norm": 0.27799312574579305,
-      "learning_rate": 1.9219341468293518e-05,
-      "loss": 0.0079,
+      "epoch": 0.4928,
+      "grad_norm": 0.028349596310172764,
+      "learning_rate": 1.9074001985891893e-05,
+      "loss": 0.0005,
       "step": 308
     },
     {
-      "epoch": 0.4625748502994012,
-      "grad_norm": 0.16117009627843906,
-      "learning_rate": 1.921306649978881e-05,
-      "loss": 0.006,
+      "epoch": 0.4944,
+      "grad_norm": 0.15763944104469357,
+      "learning_rate": 1.9066726011770725e-05,
+      "loss": 0.0032,
       "step": 309
     },
     {
-      "epoch": 0.46407185628742514,
-      "grad_norm": 0.2451738774041042,
-      "learning_rate": 1.920676744564052e-05,
-      "loss": 0.0103,
+      "epoch": 0.496,
+      "grad_norm": 0.18356997394024363,
+      "learning_rate": 1.9059422963029464e-05,
+      "loss": 0.0029,
       "step": 310
     },
     {
-      "epoch": 0.4655688622754491,
-      "grad_norm": 0.23167970747659175,
-      "learning_rate": 1.9200444322316206e-05,
-      "loss": 0.0069,
+      "epoch": 0.4976,
+      "grad_norm": 0.2508451892317318,
+      "learning_rate": 1.905209286147611e-05,
+      "loss": 0.0018,
       "step": 311
     },
     {
-      "epoch": 0.46706586826347307,
-      "grad_norm": 0.19444027025835037,
-      "learning_rate": 1.919409714634636e-05,
-      "loss": 0.0064,
+      "epoch": 0.4992,
+      "grad_norm": 0.18411645717395078,
+      "learning_rate": 1.9044735728999472e-05,
+      "loss": 0.0053,
       "step": 312
     },
     {
-      "epoch": 0.468562874251497,
-      "grad_norm": 0.259521752410209,
-      "learning_rate": 1.9187725934324352e-05,
-      "loss": 0.0087,
+      "epoch": 0.5008,
+      "grad_norm": 0.14408244481084634,
+      "learning_rate": 1.903735158756905e-05,
+      "loss": 0.0018,
       "step": 313
     },
     {
-      "epoch": 0.47005988023952094,
-      "grad_norm": 0.1870803982415873,
-      "learning_rate": 1.918133070290639e-05,
-      "loss": 0.0099,
+      "epoch": 0.5024,
+      "grad_norm": 0.17633416659439113,
+      "learning_rate": 1.902994045923502e-05,
+      "loss": 0.001,
       "step": 314
     },
     {
-      "epoch": 0.47155688622754494,
-      "grad_norm": 0.2782883408394447,
-      "learning_rate": 1.917491146881147e-05,
-      "loss": 0.0079,
+      "epoch": 0.504,
+      "grad_norm": 0.29912218229546866,
+      "learning_rate": 1.9022502366128136e-05,
+      "loss": 0.003,
       "step": 315
     },
     {
-      "epoch": 0.47305389221556887,
-      "grad_norm": 0.30227323802831124,
-      "learning_rate": 1.916846824882135e-05,
-      "loss": 0.0068,
+      "epoch": 0.5056,
+      "grad_norm": 0.24668813909693074,
+      "learning_rate": 1.901503733045967e-05,
+      "loss": 0.0066,
       "step": 316
     },
     {
-      "epoch": 0.4745508982035928,
-      "grad_norm": 0.17450053993957124,
-      "learning_rate": 1.916200105978048e-05,
-      "loss": 0.0055,
+      "epoch": 0.5072,
+      "grad_norm": 0.0708971703831616,
+      "learning_rate": 1.9007545374521354e-05,
+      "loss": 0.0007,
       "step": 317
     },
     {
-      "epoch": 0.47604790419161674,
-      "grad_norm": 0.22986677982746617,
-      "learning_rate": 1.9155509918595977e-05,
-      "loss": 0.0094,
+      "epoch": 0.5088,
+      "grad_norm": 0.0664197198165233,
+      "learning_rate": 1.90000265206853e-05,
+      "loss": 0.0011,
       "step": 318
     },
     {
-      "epoch": 0.47754491017964074,
-      "grad_norm": 0.2210009165017658,
-      "learning_rate": 1.9148994842237585e-05,
-      "loss": 0.0089,
+      "epoch": 0.5104,
+      "grad_norm": 0.09508964161653133,
+      "learning_rate": 1.8992480791403957e-05,
+      "loss": 0.0011,
       "step": 319
     },
     {
-      "epoch": 0.47904191616766467,
-      "grad_norm": 0.3369866058031818,
-      "learning_rate": 1.9142455847737613e-05,
-      "loss": 0.0102,
+      "epoch": 0.512,
+      "grad_norm": 0.133749388372953,
+      "learning_rate": 1.898490820921001e-05,
+      "loss": 0.0011,
       "step": 320
     },
     {
-      "epoch": 0.4805389221556886,
-      "grad_norm": 0.31529088422929136,
-      "learning_rate": 1.9135892952190894e-05,
-      "loss": 0.0067,
+      "epoch": 0.5136,
+      "grad_norm": 0.32095988664680164,
+      "learning_rate": 1.897730879671634e-05,
+      "loss": 0.0042,
       "step": 321
     },
     {
-      "epoch": 0.4820359281437126,
-      "grad_norm": 0.34819262640952137,
-      "learning_rate": 1.912930617275476e-05,
-      "loss": 0.0083,
+      "epoch": 0.5152,
+      "grad_norm": 0.08222690605282151,
+      "learning_rate": 1.8969682576615947e-05,
+      "loss": 0.0009,
       "step": 322
     },
     {
-      "epoch": 0.48353293413173654,
-      "grad_norm": 0.2823658083150663,
-      "learning_rate": 1.9122695526648968e-05,
-      "loss": 0.0074,
+      "epoch": 0.5168,
+      "grad_norm": 0.03976299754595526,
+      "learning_rate": 1.8962029571681887e-05,
+      "loss": 0.0005,
       "step": 323
     },
     {
-      "epoch": 0.48502994011976047,
-      "grad_norm": 0.22578397023227845,
-      "learning_rate": 1.9116061031155683e-05,
-      "loss": 0.0054,
+      "epoch": 0.5184,
+      "grad_norm": 0.08048980227033051,
+      "learning_rate": 1.8954349804767185e-05,
+      "loss": 0.0014,
       "step": 324
     },
     {
-      "epoch": 0.4865269461077844,
-      "grad_norm": 0.1798278839714749,
-      "learning_rate": 1.9109402703619396e-05,
-      "loss": 0.0069,
+      "epoch": 0.52,
+      "grad_norm": 0.12346521034266841,
+      "learning_rate": 1.8946643298804794e-05,
+      "loss": 0.0038,
       "step": 325
     },
     {
-      "epoch": 0.4880239520958084,
-      "grad_norm": 0.3173459554977556,
-      "learning_rate": 1.9102720561446933e-05,
-      "loss": 0.0097,
+      "epoch": 0.5216,
+      "grad_norm": 0.06561097915457406,
+      "learning_rate": 1.8938910076807514e-05,
+      "loss": 0.001,
       "step": 326
     },
     {
-      "epoch": 0.48952095808383234,
-      "grad_norm": 0.4520342362723284,
-      "learning_rate": 1.9096014622107363e-05,
-      "loss": 0.0064,
+      "epoch": 0.5232,
+      "grad_norm": 0.06053879999625217,
+      "learning_rate": 1.8931150161867917e-05,
+      "loss": 0.0007,
       "step": 327
     },
     {
-      "epoch": 0.49101796407185627,
-      "grad_norm": 0.2950917880335438,
-      "learning_rate": 1.9089284903131956e-05,
-      "loss": 0.0075,
+      "epoch": 0.5248,
+      "grad_norm": 0.06900447980893769,
+      "learning_rate": 1.892336357715829e-05,
+      "loss": 0.004,
       "step": 328
     },
     {
-      "epoch": 0.49251497005988026,
-      "grad_norm": 0.27125581637418156,
-      "learning_rate": 1.908253142211417e-05,
-      "loss": 0.0055,
+      "epoch": 0.5264,
+      "grad_norm": 0.10509614257503373,
+      "learning_rate": 1.891555034593055e-05,
+      "loss": 0.0011,
       "step": 329
     },
     {
-      "epoch": 0.4940119760479042,
-      "grad_norm": 0.4292444115677685,
-      "learning_rate": 1.9075754196709574e-05,
-      "loss": 0.0089,
+      "epoch": 0.528,
+      "grad_norm": 0.047031312862866576,
+      "learning_rate": 1.8907710491516197e-05,
+      "loss": 0.0009,
       "step": 330
     },
     {
-      "epoch": 0.49550898203592814,
-      "grad_norm": 0.2737159333850911,
-      "learning_rate": 1.9068953244635805e-05,
-      "loss": 0.0061,
+      "epoch": 0.5296,
+      "grad_norm": 0.01697870761243344,
+      "learning_rate": 1.8899844037326227e-05,
+      "loss": 0.0004,
       "step": 331
     },
     {
-      "epoch": 0.49700598802395207,
-      "grad_norm": 0.2813618023609554,
-      "learning_rate": 1.9062128583672543e-05,
-      "loss": 0.0055,
+      "epoch": 0.5312,
+      "grad_norm": 0.06180188702083097,
+      "learning_rate": 1.889195100685106e-05,
+      "loss": 0.0006,
       "step": 332
     },
     {
-      "epoch": 0.49850299401197606,
-      "grad_norm": 0.21811824525996454,
-      "learning_rate": 1.9055280231661443e-05,
-      "loss": 0.0076,
+      "epoch": 0.5328,
+      "grad_norm": 0.054644161559529975,
+      "learning_rate": 1.8884031423660492e-05,
+      "loss": 0.0007,
       "step": 333
     },
     {
-      "epoch": 0.5,
-      "grad_norm": 0.3024569403044615,
-      "learning_rate": 1.904840820650609e-05,
-      "loss": 0.0076,
+      "epoch": 0.5344,
+      "grad_norm": 0.06836013722792311,
+      "learning_rate": 1.8876085311403592e-05,
+      "loss": 0.0008,
       "step": 334
     },
     {
-      "epoch": 0.5014970059880239,
-      "grad_norm": 0.3747520923531163,
-      "learning_rate": 1.9041512526171968e-05,
-      "loss": 0.0073,
+      "epoch": 0.536,
+      "grad_norm": 0.08405432130742992,
+      "learning_rate": 1.8868112693808664e-05,
+      "loss": 0.0007,
       "step": 335
     },
     {
-      "epoch": 0.5029940119760479,
-      "grad_norm": 0.32327667748094846,
-      "learning_rate": 1.9034593208686397e-05,
-      "loss": 0.0086,
+      "epoch": 0.5376,
+      "grad_norm": 0.052804793325438484,
+      "learning_rate": 1.8860113594683148e-05,
+      "loss": 0.0009,
       "step": 336
     },
     {
-      "epoch": 0.5044910179640718,
-      "grad_norm": 0.2599402605515639,
-      "learning_rate": 1.9027650272138483e-05,
-      "loss": 0.008,
+      "epoch": 0.5392,
+      "grad_norm": 0.0903202357847314,
+      "learning_rate": 1.8852088037913577e-05,
+      "loss": 0.0011,
       "step": 337
     },
     {
-      "epoch": 0.5059880239520959,
-      "grad_norm": 0.309419497369784,
-      "learning_rate": 1.9020683734679098e-05,
-      "loss": 0.0076,
+      "epoch": 0.5408,
+      "grad_norm": 0.04558030866715895,
+      "learning_rate": 1.884403604746547e-05,
+      "loss": 0.0006,
       "step": 338
     },
     {
-      "epoch": 0.5074850299401198,
-      "grad_norm": 0.3980869606666559,
-      "learning_rate": 1.9013693614520795e-05,
-      "loss": 0.011,
+      "epoch": 0.5424,
+      "grad_norm": 0.02140025655887005,
+      "learning_rate": 1.8835957647383304e-05,
+      "loss": 0.0004,
       "step": 339
     },
     {
-      "epoch": 0.5089820359281437,
-      "grad_norm": 0.17974783303905734,
-      "learning_rate": 1.9006679929937796e-05,
-      "loss": 0.0064,
+      "epoch": 0.544,
+      "grad_norm": 0.043490093005678035,
+      "learning_rate": 1.8827852861790398e-05,
+      "loss": 0.0006,
       "step": 340
     },
     {
-      "epoch": 0.5104790419161677,
-      "grad_norm": 0.33752895891985435,
-      "learning_rate": 1.899964269926591e-05,
-      "loss": 0.0058,
+      "epoch": 0.5456,
+      "grad_norm": 0.14053030141846226,
+      "learning_rate": 1.8819721714888878e-05,
+      "loss": 0.0008,
       "step": 341
     },
     {
-      "epoch": 0.5119760479041916,
-      "grad_norm": 0.2851367706634093,
-      "learning_rate": 1.8992581940902517e-05,
-      "loss": 0.0059,
+      "epoch": 0.5472,
+      "grad_norm": 0.07394631076602469,
+      "learning_rate": 1.8811564230959585e-05,
+      "loss": 0.0009,
       "step": 342
     },
     {
-      "epoch": 0.5134730538922155,
-      "grad_norm": 0.19085718115700934,
-      "learning_rate": 1.8985497673306495e-05,
-      "loss": 0.0042,
+      "epoch": 0.5488,
+      "grad_norm": 0.11601250028806075,
+      "learning_rate": 1.8803380434362e-05,
+      "loss": 0.004,
       "step": 343
     },
     {
-      "epoch": 0.5149700598802395,
-      "grad_norm": 0.2450459731205004,
-      "learning_rate": 1.897838991499819e-05,
-      "loss": 0.0084,
+      "epoch": 0.5504,
+      "grad_norm": 0.07109290389899153,
+      "learning_rate": 1.879517034953418e-05,
+      "loss": 0.0008,
       "step": 344
     },
     {
-      "epoch": 0.5164670658682635,
-      "grad_norm": 0.2159471296015911,
-      "learning_rate": 1.8971258684559354e-05,
-      "loss": 0.0062,
+      "epoch": 0.552,
+      "grad_norm": 0.05902437246594027,
+      "learning_rate": 1.878693400099269e-05,
+      "loss": 0.0007,
       "step": 345
     },
     {
-      "epoch": 0.5179640718562875,
-      "grad_norm": 0.2582075201511502,
-      "learning_rate": 1.8964104000633104e-05,
-      "loss": 0.0072,
+      "epoch": 0.5536,
+      "grad_norm": 0.1309935252514575,
+      "learning_rate": 1.8778671413332513e-05,
+      "loss": 0.0025,
       "step": 346
     },
     {
-      "epoch": 0.5194610778443114,
-      "grad_norm": 0.2826246708970663,
-      "learning_rate": 1.895692588192387e-05,
-      "loss": 0.0075,
+      "epoch": 0.5552,
+      "grad_norm": 0.026865000045514807,
+      "learning_rate": 1.877038261122699e-05,
+      "loss": 0.0003,
       "step": 347
     },
     {
-      "epoch": 0.5209580838323353,
-      "grad_norm": 0.2612365012660709,
-      "learning_rate": 1.894972434719735e-05,
-      "loss": 0.0046,
+      "epoch": 0.5568,
+      "grad_norm": 0.200787114993862,
+      "learning_rate": 1.8762067619427745e-05,
+      "loss": 0.0015,
       "step": 348
     },
     {
-      "epoch": 0.5224550898203593,
-      "grad_norm": 0.15056193683838454,
-      "learning_rate": 1.8942499415280458e-05,
-      "loss": 0.0042,
+      "epoch": 0.5584,
+      "grad_norm": 0.18264175410560715,
+      "learning_rate": 1.87537264627646e-05,
+      "loss": 0.0035,
       "step": 349
     },
     {
-      "epoch": 0.5239520958083832,
-      "grad_norm": 0.22875131238932167,
-      "learning_rate": 1.8935251105061268e-05,
-      "loss": 0.0046,
+      "epoch": 0.56,
+      "grad_norm": 0.04120060687514351,
+      "learning_rate": 1.8745359166145526e-05,
+      "loss": 0.0005,
       "step": 350
     },
     {
-      "epoch": 0.5254491017964071,
-      "grad_norm": 0.2534927935258981,
-      "learning_rate": 1.892797943548898e-05,
-      "loss": 0.0084,
+      "epoch": 0.5616,
+      "grad_norm": 0.34090676386146507,
+      "learning_rate": 1.8736965754556527e-05,
+      "loss": 0.0021,
       "step": 351
     },
     {
-      "epoch": 0.5269461077844312,
-      "grad_norm": 0.21952770072925085,
-      "learning_rate": 1.8920684425573865e-05,
-      "loss": 0.0035,
+      "epoch": 0.5632,
+      "grad_norm": 0.019128712884290862,
+      "learning_rate": 1.8728546253061614e-05,
+      "loss": 0.0004,
       "step": 352
     },
     {
-      "epoch": 0.5284431137724551,
-      "grad_norm": 0.22335232414184136,
-      "learning_rate": 1.8913366094387203e-05,
-      "loss": 0.0055,
+      "epoch": 0.5648,
+      "grad_norm": 0.027911285732412968,
+      "learning_rate": 1.8720100686802693e-05,
+      "loss": 0.0004,
       "step": 353
     },
     {
-      "epoch": 0.5299401197604791,
-      "grad_norm": 0.18343084280328395,
-      "learning_rate": 1.8906024461061247e-05,
-      "loss": 0.0064,
+      "epoch": 0.5664,
+      "grad_norm": 0.1729529053601621,
+      "learning_rate": 1.8711629080999506e-05,
+      "loss": 0.0013,
       "step": 354
     },
     {
-      "epoch": 0.531437125748503,
-      "grad_norm": 0.21313883463880326,
-      "learning_rate": 1.8898659544789166e-05,
-      "loss": 0.0052,
+      "epoch": 0.568,
+      "grad_norm": 0.06981929668164648,
+      "learning_rate": 1.8703131460949555e-05,
+      "loss": 0.0012,
       "step": 355
     },
     {
-      "epoch": 0.5329341317365269,
-      "grad_norm": 0.17900066861920666,
-      "learning_rate": 1.8891271364825002e-05,
-      "loss": 0.0054,
+      "epoch": 0.5696,
+      "grad_norm": 0.13135552462130254,
+      "learning_rate": 1.869460785202802e-05,
+      "loss": 0.0028,
       "step": 356
     },
     {
-      "epoch": 0.5344311377245509,
-      "grad_norm": 0.3186411908486247,
-      "learning_rate": 1.8883859940483614e-05,
-      "loss": 0.0085,
+      "epoch": 0.5712,
+      "grad_norm": 0.13398125098852007,
+      "learning_rate": 1.86860582796877e-05,
+      "loss": 0.0019,
       "step": 357
     },
     {
-      "epoch": 0.5359281437125748,
-      "grad_norm": 0.23748527458193902,
-      "learning_rate": 1.8876425291140626e-05,
-      "loss": 0.0064,
+      "epoch": 0.5728,
+      "grad_norm": 0.019718997811409525,
+      "learning_rate": 1.8677482769458905e-05,
+      "loss": 0.0005,
       "step": 358
     },
     {
-      "epoch": 0.5374251497005988,
-      "grad_norm": 0.18189619860086126,
-      "learning_rate": 1.8868967436232378e-05,
-      "loss": 0.0061,
+      "epoch": 0.5744,
+      "grad_norm": 0.05279323543297812,
+      "learning_rate": 1.866888134694942e-05,
+      "loss": 0.0007,
       "step": 359
     },
     {
-      "epoch": 0.5389221556886228,
-      "grad_norm": 0.3081977476306483,
-      "learning_rate": 1.886148639525588e-05,
-      "loss": 0.0049,
+      "epoch": 0.576,
+      "grad_norm": 0.16886373007352773,
+      "learning_rate": 1.866025403784439e-05,
+      "loss": 0.0011,
       "step": 360
     },
     {
-      "epoch": 0.5404191616766467,
-      "grad_norm": 0.3589434090609591,
-      "learning_rate": 1.885398218776876e-05,
-      "loss": 0.0052,
+      "epoch": 0.5776,
+      "grad_norm": 0.12528521939164913,
+      "learning_rate": 1.865160086790627e-05,
+      "loss": 0.0012,
       "step": 361
     },
     {
-      "epoch": 0.5419161676646707,
-      "grad_norm": 0.2719093022438886,
-      "learning_rate": 1.8846454833389202e-05,
-      "loss": 0.0063,
+      "epoch": 0.5792,
+      "grad_norm": 0.07723492991438953,
+      "learning_rate": 1.8642921862974742e-05,
+      "loss": 0.0008,
       "step": 362
     },
     {
-      "epoch": 0.5434131736526946,
-      "grad_norm": 0.2134250055694977,
-      "learning_rate": 1.8838904351795905e-05,
-      "loss": 0.0047,
+      "epoch": 0.5808,
+      "grad_norm": 0.10497027463979597,
+      "learning_rate": 1.8634217048966638e-05,
+      "loss": 0.0013,
       "step": 363
     },
     {
-      "epoch": 0.5449101796407185,
-      "grad_norm": 0.36939383043202,
-      "learning_rate": 1.8831330762728028e-05,
-      "loss": 0.0061,
+      "epoch": 0.5824,
+      "grad_norm": 0.0606074346839084,
+      "learning_rate": 1.8625486451875843e-05,
+      "loss": 0.0006,
       "step": 364
     },
     {
-      "epoch": 0.5464071856287425,
-      "grad_norm": 0.3092372268044756,
-      "learning_rate": 1.882373408598515e-05,
-      "loss": 0.0055,
+      "epoch": 0.584,
+      "grad_norm": 0.07920612799392682,
+      "learning_rate": 1.861673009777325e-05,
+      "loss": 0.0008,
       "step": 365
     },
     {
-      "epoch": 0.5479041916167665,
-      "grad_norm": 0.22805534768018146,
-      "learning_rate": 1.8816114341427196e-05,
-      "loss": 0.0065,
+      "epoch": 0.5856,
+      "grad_norm": 0.0346305606132078,
+      "learning_rate": 1.8607948012806664e-05,
+      "loss": 0.0005,
       "step": 366
     },
     {
-      "epoch": 0.5494011976047904,
-      "grad_norm": 0.20186378709158326,
-      "learning_rate": 1.88084715489744e-05,
-      "loss": 0.0055,
+      "epoch": 0.5872,
+      "grad_norm": 0.06547468094544372,
+      "learning_rate": 1.8599140223200716e-05,
+      "loss": 0.0008,
       "step": 367
     },
     {
-      "epoch": 0.5508982035928144,
-      "grad_norm": 0.2447957214059236,
-      "learning_rate": 1.8800805728607257e-05,
-      "loss": 0.0059,
+      "epoch": 0.5888,
+      "grad_norm": 0.15325174641228742,
+      "learning_rate": 1.859030675525681e-05,
+      "loss": 0.0026,
       "step": 368
     },
     {
-      "epoch": 0.5523952095808383,
-      "grad_norm": 0.24624052829362808,
-      "learning_rate": 1.879311690036645e-05,
-      "loss": 0.005,
+      "epoch": 0.5904,
+      "grad_norm": 0.1389172010956714,
+      "learning_rate": 1.858144763535302e-05,
+      "loss": 0.0037,
       "step": 369
     },
     {
-      "epoch": 0.5538922155688623,
-      "grad_norm": 0.4598990071533753,
-      "learning_rate": 1.8785405084352825e-05,
-      "loss": 0.0119,
+      "epoch": 0.592,
+      "grad_norm": 0.06355969619933752,
+      "learning_rate": 1.857256288994402e-05,
+      "loss": 0.0006,
       "step": 370
     },
     {
-      "epoch": 0.5553892215568862,
-      "grad_norm": 0.2254413046985076,
-      "learning_rate": 1.8777670300727314e-05,
-      "loss": 0.0047,
+      "epoch": 0.5936,
+      "grad_norm": 0.08663530000385801,
+      "learning_rate": 1.8563652545561014e-05,
+      "loss": 0.0006,
       "step": 371
     },
     {
-      "epoch": 0.5568862275449101,
-      "grad_norm": 0.2817060159533667,
-      "learning_rate": 1.8769912569710902e-05,
-      "loss": 0.0054,
+      "epoch": 0.5952,
+      "grad_norm": 0.18572281033057678,
+      "learning_rate": 1.855471662881164e-05,
+      "loss": 0.0027,
       "step": 372
     },
     {
-      "epoch": 0.5583832335329342,
-      "grad_norm": 0.23401568986125731,
-      "learning_rate": 1.876213191158456e-05,
-      "loss": 0.0066,
+      "epoch": 0.5968,
+      "grad_norm": 0.022392349584578558,
+      "learning_rate": 1.8545755166379898e-05,
+      "loss": 0.0003,
       "step": 373
     },
     {
-      "epoch": 0.5598802395209581,
-      "grad_norm": 0.31712264048465694,
-      "learning_rate": 1.8754328346689194e-05,
-      "loss": 0.0051,
+      "epoch": 0.5984,
+      "grad_norm": 0.01772969160150952,
+      "learning_rate": 1.8536768185026085e-05,
+      "loss": 0.0004,
       "step": 374
     },
     {
-      "epoch": 0.561377245508982,
-      "grad_norm": 0.1853362893894113,
-      "learning_rate": 1.8746501895425603e-05,
-      "loss": 0.0031,
+      "epoch": 0.6,
+      "grad_norm": 0.03463345597440423,
+      "learning_rate": 1.852775571158668e-05,
+      "loss": 0.0004,
       "step": 375
     },
     {
-      "epoch": 0.562874251497006,
-      "grad_norm": 0.1512730540107703,
-      "learning_rate": 1.8738652578254414e-05,
-      "loss": 0.0031,
+      "epoch": 0.6016,
+      "grad_norm": 0.0687422345719889,
+      "learning_rate": 1.85187177729743e-05,
+      "loss": 0.0034,
       "step": 376
     },
     {
-      "epoch": 0.5643712574850299,
-      "grad_norm": 0.251381383971643,
-      "learning_rate": 1.8730780415696034e-05,
-      "loss": 0.0061,
+      "epoch": 0.6032,
+      "grad_norm": 0.15556372482561745,
+      "learning_rate": 1.850965439617761e-05,
+      "loss": 0.0033,
       "step": 377
     },
     {
-      "epoch": 0.5658682634730539,
-      "grad_norm": 0.2090374749619751,
-      "learning_rate": 1.872288542833059e-05,
-      "loss": 0.0032,
+      "epoch": 0.6048,
+      "grad_norm": 0.08434742559441728,
+      "learning_rate": 1.8500565608261215e-05,
+      "loss": 0.0031,
       "step": 378
     },
     {
-      "epoch": 0.5673652694610778,
-      "grad_norm": 0.29710148489922494,
-      "learning_rate": 1.871496763679788e-05,
-      "loss": 0.0061,
+      "epoch": 0.6064,
+      "grad_norm": 0.1474992044646562,
+      "learning_rate": 1.8491451436365628e-05,
+      "loss": 0.0025,
       "step": 379
     },
     {
-      "epoch": 0.5688622754491018,
-      "grad_norm": 0.20476911169490797,
-      "learning_rate": 1.8707027061797328e-05,
-      "loss": 0.0041,
+      "epoch": 0.608,
+      "grad_norm": 0.06662251243376646,
+      "learning_rate": 1.848231190770714e-05,
+      "loss": 0.0008,
       "step": 380
     },
     {
-      "epoch": 0.5703592814371258,
-      "grad_norm": 0.15467140235695107,
-      "learning_rate": 1.8699063724087905e-05,
-      "loss": 0.0039,
+      "epoch": 0.6096,
+      "grad_norm": 0.0754815447641557,
+      "learning_rate": 1.8473147049577777e-05,
+      "loss": 0.001,
       "step": 381
     },
     {
-      "epoch": 0.5718562874251497,
-      "grad_norm": 0.35593514914651286,
-      "learning_rate": 1.8691077644488103e-05,
-      "loss": 0.0051,
+      "epoch": 0.6112,
+      "grad_norm": 0.009235009771626088,
+      "learning_rate": 1.8463956889345195e-05,
+      "loss": 0.0002,
       "step": 382
     },
     {
-      "epoch": 0.5733532934131736,
-      "grad_norm": 0.3031925999503722,
-      "learning_rate": 1.8683068843875866e-05,
-      "loss": 0.0054,
+      "epoch": 0.6128,
+      "grad_norm": 0.01669079074694255,
+      "learning_rate": 1.8454741454452604e-05,
+      "loss": 0.0003,
       "step": 383
     },
     {
-      "epoch": 0.5748502994011976,
-      "grad_norm": 0.392987606385426,
-      "learning_rate": 1.867503734318853e-05,
-      "loss": 0.0068,
+      "epoch": 0.6144,
+      "grad_norm": 0.05043025132118084,
+      "learning_rate": 1.8445500772418697e-05,
+      "loss": 0.0004,
       "step": 384
     },
     {
-      "epoch": 0.5763473053892215,
-      "grad_norm": 0.1615633068715301,
-      "learning_rate": 1.8666983163422786e-05,
-      "loss": 0.004,
+      "epoch": 0.616,
+      "grad_norm": 0.3126790444003597,
+      "learning_rate": 1.843623487083755e-05,
+      "loss": 0.0032,
       "step": 385
     },
     {
-      "epoch": 0.5778443113772455,
-      "grad_norm": 0.2623310704504511,
-      "learning_rate": 1.8658906325634605e-05,
-      "loss": 0.0065,
+      "epoch": 0.6176,
+      "grad_norm": 0.05068038072505894,
+      "learning_rate": 1.842694377737855e-05,
+      "loss": 0.0006,
       "step": 386
     },
     {
-      "epoch": 0.5793413173652695,
-      "grad_norm": 0.41136291194943037,
-      "learning_rate": 1.86508068509392e-05,
-      "loss": 0.0065,
+      "epoch": 0.6192,
+      "grad_norm": 0.03215920255949678,
+      "learning_rate": 1.8417627519786317e-05,
+      "loss": 0.0004,
       "step": 387
     },
     {
-      "epoch": 0.5808383233532934,
-      "grad_norm": 0.32922655904832243,
-      "learning_rate": 1.864268476051096e-05,
-      "loss": 0.0083,
+      "epoch": 0.6208,
+      "grad_norm": 0.018828487209741788,
+      "learning_rate": 1.8408286125880605e-05,
+      "loss": 0.0003,
       "step": 388
     },
     {
-      "epoch": 0.5823353293413174,
-      "grad_norm": 0.15328440896027834,
-      "learning_rate": 1.86345400755834e-05,
-      "loss": 0.0034,
+      "epoch": 0.6224,
+      "grad_norm": 0.14363387776918993,
+      "learning_rate": 1.839891962355624e-05,
+      "loss": 0.001,
       "step": 389
     },
     {
-      "epoch": 0.5838323353293413,
-      "grad_norm": 0.23062078223568724,
-      "learning_rate": 1.8626372817449107e-05,
-      "loss": 0.0035,
+      "epoch": 0.624,
+      "grad_norm": 0.11747971500364393,
+      "learning_rate": 1.8389528040783014e-05,
+      "loss": 0.0043,
       "step": 390
     },
     {
-      "epoch": 0.5853293413173652,
-      "grad_norm": 0.4137111601133577,
-      "learning_rate": 1.8618183007459664e-05,
-      "loss": 0.0057,
+      "epoch": 0.6256,
+      "grad_norm": 0.09249479045992653,
+      "learning_rate": 1.838011140560562e-05,
+      "loss": 0.0006,
       "step": 391
     },
     {
-      "epoch": 0.5868263473053892,
-      "grad_norm": 0.27055771834059683,
-      "learning_rate": 1.8609970667025633e-05,
-      "loss": 0.0046,
+      "epoch": 0.6272,
+      "grad_norm": 0.10553138594399365,
+      "learning_rate": 1.8370669746143566e-05,
+      "loss": 0.0037,
       "step": 392
     },
     {
-      "epoch": 0.5883233532934131,
-      "grad_norm": 0.1400482751380909,
-      "learning_rate": 1.8601735817616466e-05,
-      "loss": 0.0033,
+      "epoch": 0.6288,
+      "grad_norm": 0.08443128236771645,
+      "learning_rate": 1.836120309059107e-05,
+      "loss": 0.0004,
       "step": 393
     },
     {
-      "epoch": 0.5898203592814372,
-      "grad_norm": 0.2825851174593088,
-      "learning_rate": 1.859347848076046e-05,
-      "loss": 0.0062,
+      "epoch": 0.6304,
+      "grad_norm": 0.23335718692317806,
+      "learning_rate": 1.835171146721701e-05,
+      "loss": 0.0018,
       "step": 394
     },
     {
-      "epoch": 0.5913173652694611,
-      "grad_norm": 0.3397946721846426,
-      "learning_rate": 1.85851986780447e-05,
-      "loss": 0.006,
+      "epoch": 0.632,
+      "grad_norm": 0.20442531679113446,
+      "learning_rate": 1.8342194904364815e-05,
+      "loss": 0.0016,
       "step": 395
     },
     {
-      "epoch": 0.592814371257485,
-      "grad_norm": 0.27215859756652094,
-      "learning_rate": 1.8576896431115017e-05,
-      "loss": 0.0043,
+      "epoch": 0.6336,
+      "grad_norm": 0.010111401639089001,
+      "learning_rate": 1.8332653430452375e-05,
+      "loss": 0.0002,
       "step": 396
     },
     {
-      "epoch": 0.594311377245509,
-      "grad_norm": 0.15206642506994694,
-      "learning_rate": 1.8568571761675893e-05,
-      "loss": 0.0035,
+      "epoch": 0.6352,
+      "grad_norm": 0.04480967540704622,
+      "learning_rate": 1.8323087073971996e-05,
+      "loss": 0.0004,
       "step": 397
     },
     {
-      "epoch": 0.5958083832335329,
-      "grad_norm": 0.177981847566587,
-      "learning_rate": 1.8560224691490452e-05,
-      "loss": 0.0051,
+      "epoch": 0.6368,
+      "grad_norm": 0.05123731112538742,
+      "learning_rate": 1.831349586349026e-05,
+      "loss": 0.0005,
       "step": 398
     },
     {
-      "epoch": 0.5973053892215568,
-      "grad_norm": 0.18243440630634988,
-      "learning_rate": 1.8551855242380366e-05,
-      "loss": 0.0028,
+      "epoch": 0.6384,
+      "grad_norm": 0.05877538711434608,
+      "learning_rate": 1.8303879827647977e-05,
+      "loss": 0.0005,
       "step": 399
     },
     {
-      "epoch": 0.5988023952095808,
-      "grad_norm": 0.22781016153338413,
-      "learning_rate": 1.854346343622582e-05,
-      "loss": 0.0061,
+      "epoch": 0.64,
+      "grad_norm": 0.2787404123164975,
+      "learning_rate": 1.8294238995160093e-05,
+      "loss": 0.0024,
       "step": 400
     },
     {
-      "epoch": 0.6002994011976048,
-      "grad_norm": 0.14798262053026728,
-      "learning_rate": 1.8535049294965436e-05,
-      "loss": 0.0028,
+      "epoch": 0.6416,
+      "grad_norm": 0.14221635930646565,
+      "learning_rate": 1.8284573394815596e-05,
+      "loss": 0.0022,
       "step": 401
     },
     {
-      "epoch": 0.6017964071856288,
-      "grad_norm": 0.12876490456064732,
-      "learning_rate": 1.8526612840596238e-05,
-      "loss": 0.0029,
+      "epoch": 0.6432,
+      "grad_norm": 0.05002110451791554,
+      "learning_rate": 1.8274883055477436e-05,
+      "loss": 0.0004,
       "step": 402
     },
     {
-      "epoch": 0.6032934131736527,
-      "grad_norm": 0.19321532804478075,
-      "learning_rate": 1.8518154095173583e-05,
-      "loss": 0.0032,
+      "epoch": 0.6448,
+      "grad_norm": 0.03932836435003346,
+      "learning_rate": 1.826516800608244e-05,
+      "loss": 0.0005,
       "step": 403
     },
     {
-      "epoch": 0.6047904191616766,
-      "grad_norm": 0.12070559140749956,
-      "learning_rate": 1.8509673080811096e-05,
-      "loss": 0.0033,
+      "epoch": 0.6464,
+      "grad_norm": 0.094636335346893,
+      "learning_rate": 1.8255428275641212e-05,
+      "loss": 0.0017,
       "step": 404
     },
     {
-      "epoch": 0.6062874251497006,
-      "grad_norm": 0.18453370264259583,
-      "learning_rate": 1.8501169819680624e-05,
-      "loss": 0.005,
+      "epoch": 0.648,
+      "grad_norm": 0.04753353819319314,
+      "learning_rate": 1.8245663893238075e-05,
+      "loss": 0.0005,
       "step": 405
     },
     {
-      "epoch": 0.6077844311377245,
-      "grad_norm": 0.1954819685260244,
-      "learning_rate": 1.849264433401217e-05,
-      "loss": 0.0036,
+      "epoch": 0.6496,
+      "grad_norm": 0.08265945475873478,
+      "learning_rate": 1.823587488803095e-05,
+      "loss": 0.0014,
       "step": 406
     },
     {
-      "epoch": 0.6092814371257484,
-      "grad_norm": 0.15334768361544945,
-      "learning_rate": 1.8484096646093842e-05,
-      "loss": 0.0035,
+      "epoch": 0.6512,
+      "grad_norm": 0.12295729262860776,
+      "learning_rate": 1.8226061289251297e-05,
+      "loss": 0.0015,
       "step": 407
     },
     {
-      "epoch": 0.6107784431137725,
-      "grad_norm": 0.15359417727888636,
-      "learning_rate": 1.847552677827179e-05,
-      "loss": 0.0047,
+      "epoch": 0.6528,
+      "grad_norm": 0.139470158108872,
+      "learning_rate": 1.821622312620401e-05,
+      "loss": 0.0016,
       "step": 408
     },
     {
-      "epoch": 0.6122754491017964,
-      "grad_norm": 0.15142354869259234,
-      "learning_rate": 1.846693475295015e-05,
-      "loss": 0.0025,
+      "epoch": 0.6544,
+      "grad_norm": 0.05305196376502255,
+      "learning_rate": 1.8206360428267332e-05,
+      "loss": 0.0004,
       "step": 409
     },
     {
-      "epoch": 0.6137724550898204,
-      "grad_norm": 0.1305889137402961,
-      "learning_rate": 1.8458320592590976e-05,
-      "loss": 0.0039,
+      "epoch": 0.656,
+      "grad_norm": 0.018256913526341095,
+      "learning_rate": 1.8196473224892784e-05,
+      "loss": 0.0003,
       "step": 410
     },
     {
-      "epoch": 0.6152694610778443,
-      "grad_norm": 0.24052279035392543,
-      "learning_rate": 1.8449684319714202e-05,
-      "loss": 0.0037,
+      "epoch": 0.6576,
+      "grad_norm": 0.21253701484431486,
+      "learning_rate": 1.8186561545605055e-05,
+      "loss": 0.0024,
       "step": 411
     },
     {
-      "epoch": 0.6167664670658682,
-      "grad_norm": 0.21510134495795813,
-      "learning_rate": 1.8441025956897566e-05,
-      "loss": 0.0047,
+      "epoch": 0.6592,
+      "grad_norm": 0.015257919984609451,
+      "learning_rate": 1.817662542000192e-05,
+      "loss": 0.0003,
       "step": 412
     },
     {
-      "epoch": 0.6182634730538922,
-      "grad_norm": 0.1378513852811569,
-      "learning_rate": 1.8432345526776555e-05,
-      "loss": 0.0037,
+      "epoch": 0.6608,
+      "grad_norm": 0.10312922775285008,
+      "learning_rate": 1.816666487775416e-05,
+      "loss": 0.0011,
       "step": 413
     },
     {
-      "epoch": 0.6197604790419161,
-      "grad_norm": 0.12876211162444984,
-      "learning_rate": 1.842364305204434e-05,
-      "loss": 0.0047,
+      "epoch": 0.6624,
+      "grad_norm": 0.11624902449700726,
+      "learning_rate": 1.815667994860547e-05,
+      "loss": 0.0006,
       "step": 414
     },
     {
-      "epoch": 0.6212574850299402,
-      "grad_norm": 0.1870068324397345,
-      "learning_rate": 1.841491855545173e-05,
-      "loss": 0.0056,
+      "epoch": 0.664,
+      "grad_norm": 0.18464390507676323,
+      "learning_rate": 1.8146670662372353e-05,
+      "loss": 0.0027,
       "step": 415
     },
     {
-      "epoch": 0.6227544910179641,
-      "grad_norm": 0.15043391887906607,
-      "learning_rate": 1.840617205980711e-05,
-      "loss": 0.0034,
+      "epoch": 0.6656,
+      "grad_norm": 0.13216285372914752,
+      "learning_rate": 1.813663704894407e-05,
+      "loss": 0.0008,
       "step": 416
     },
     {
-      "epoch": 0.624251497005988,
-      "grad_norm": 0.1603990561673152,
-      "learning_rate": 1.8397403587976366e-05,
-      "loss": 0.0038,
+      "epoch": 0.6672,
+      "grad_norm": 0.01942107440009396,
+      "learning_rate": 1.8126579138282502e-05,
+      "loss": 0.0003,
       "step": 417
     },
     {
-      "epoch": 0.625748502994012,
-      "grad_norm": 0.1380674767451333,
-      "learning_rate": 1.838861316288284e-05,
-      "loss": 0.0021,
+      "epoch": 0.6688,
+      "grad_norm": 0.14540193615014807,
+      "learning_rate": 1.8116496960422108e-05,
+      "loss": 0.0008,
       "step": 418
     },
     {
-      "epoch": 0.6272455089820359,
-      "grad_norm": 0.18002621685496464,
-      "learning_rate": 1.8379800807507276e-05,
-      "loss": 0.0041,
+      "epoch": 0.6704,
+      "grad_norm": 0.17943893985236523,
+      "learning_rate": 1.8106390545469797e-05,
+      "loss": 0.0022,
       "step": 419
     },
     {
-      "epoch": 0.6287425149700598,
-      "grad_norm": 0.16156445172842446,
-      "learning_rate": 1.8370966544887734e-05,
+      "epoch": 0.672,
+      "grad_norm": 0.2999659081905774,
+      "learning_rate": 1.809625992360485e-05,
       "loss": 0.0026,
       "step": 420
     },
     {
-      "epoch": 0.6302395209580839,
-      "grad_norm": 0.13680763335054835,
-      "learning_rate": 1.836211039811956e-05,
-      "loss": 0.0029,
+      "epoch": 0.6736,
+      "grad_norm": 0.1345555003175391,
+      "learning_rate": 1.8086105125078858e-05,
+      "loss": 0.0044,
       "step": 421
     },
     {
-      "epoch": 0.6317365269461078,
-      "grad_norm": 0.20502692603268471,
-      "learning_rate": 1.8353232390355293e-05,
-      "loss": 0.0045,
+      "epoch": 0.6752,
+      "grad_norm": 0.08857667059507833,
+      "learning_rate": 1.8075926180215576e-05,
+      "loss": 0.0006,
       "step": 422
     },
     {
-      "epoch": 0.6332335329341318,
-      "grad_norm": 0.23325599547346149,
-      "learning_rate": 1.8344332544804647e-05,
-      "loss": 0.0067,
+      "epoch": 0.6768,
+      "grad_norm": 0.07718347252729954,
+      "learning_rate": 1.8065723119410885e-05,
+      "loss": 0.0006,
       "step": 423
     },
     {
-      "epoch": 0.6347305389221557,
-      "grad_norm": 0.29861952724882246,
-      "learning_rate": 1.8335410884734412e-05,
-      "loss": 0.006,
+      "epoch": 0.6784,
+      "grad_norm": 0.031977559605231326,
+      "learning_rate": 1.805549597313267e-05,
+      "loss": 0.0004,
       "step": 424
     },
     {
-      "epoch": 0.6362275449101796,
-      "grad_norm": 0.1405737685776957,
-      "learning_rate": 1.832646743346841e-05,
-      "loss": 0.0025,
+      "epoch": 0.68,
+      "grad_norm": 0.11691074486721813,
+      "learning_rate": 1.804524477192075e-05,
+      "loss": 0.0015,
       "step": 425
     },
     {
-      "epoch": 0.6377245508982036,
-      "grad_norm": 0.15314449241818123,
-      "learning_rate": 1.831750221438743e-05,
-      "loss": 0.0025,
+      "epoch": 0.6816,
+      "grad_norm": 0.06512767211853637,
+      "learning_rate": 1.803496954638676e-05,
+      "loss": 0.0006,
       "step": 426
     },
     {
-      "epoch": 0.6392215568862275,
-      "grad_norm": 0.20564899127887978,
-      "learning_rate": 1.8308515250929174e-05,
-      "loss": 0.005,
+      "epoch": 0.6832,
+      "grad_norm": 0.11496639985940264,
+      "learning_rate": 1.8024670327214084e-05,
+      "loss": 0.003,
       "step": 427
     },
     {
-      "epoch": 0.6407185628742516,
-      "grad_norm": 0.2250426655272749,
-      "learning_rate": 1.8299506566588178e-05,
-      "loss": 0.0063,
+      "epoch": 0.6848,
+      "grad_norm": 0.023249205111541698,
+      "learning_rate": 1.8014347145157757e-05,
+      "loss": 0.0003,
       "step": 428
     },
     {
-      "epoch": 0.6422155688622755,
-      "grad_norm": 0.20159344511006713,
-      "learning_rate": 1.829047618491578e-05,
-      "loss": 0.0041,
+      "epoch": 0.6864,
+      "grad_norm": 0.23618797082446044,
+      "learning_rate": 1.8004000031044363e-05,
+      "loss": 0.0019,
       "step": 429
     },
     {
-      "epoch": 0.6437125748502994,
-      "grad_norm": 0.30367059362305904,
-      "learning_rate": 1.8281424129520028e-05,
-      "loss": 0.0045,
+      "epoch": 0.688,
+      "grad_norm": 0.006547483779280376,
+      "learning_rate": 1.799362901577196e-05,
+      "loss": 0.0002,
       "step": 430
     },
     {
-      "epoch": 0.6452095808383234,
-      "grad_norm": 0.17853724353758313,
-      "learning_rate": 1.8272350424065637e-05,
-      "loss": 0.0037,
+      "epoch": 0.6896,
+      "grad_norm": 0.07298271657368005,
+      "learning_rate": 1.798323413030997e-05,
+      "loss": 0.0008,
       "step": 431
     },
     {
-      "epoch": 0.6467065868263473,
-      "grad_norm": 0.13452139718446834,
-      "learning_rate": 1.8263255092273918e-05,
-      "loss": 0.0031,
+      "epoch": 0.6912,
+      "grad_norm": 0.1512758686758914,
+      "learning_rate": 1.7972815405699105e-05,
+      "loss": 0.002,
       "step": 432
     },
     {
-      "epoch": 0.6482035928143712,
-      "grad_norm": 0.1485870363064914,
-      "learning_rate": 1.8254138157922726e-05,
-      "loss": 0.0036,
+      "epoch": 0.6928,
+      "grad_norm": 0.03211251447879792,
+      "learning_rate": 1.796237287305125e-05,
+      "loss": 0.0003,
       "step": 433
     },
     {
-      "epoch": 0.6497005988023952,
-      "grad_norm": 0.1369157060039127,
-      "learning_rate": 1.8244999644846386e-05,
-      "loss": 0.0049,
+      "epoch": 0.6944,
+      "grad_norm": 0.10182772421205202,
+      "learning_rate": 1.7951906563549397e-05,
+      "loss": 0.0004,
       "step": 434
     },
     {
-      "epoch": 0.6511976047904192,
-      "grad_norm": 0.1422170125322691,
-      "learning_rate": 1.823583957693563e-05,
-      "loss": 0.0025,
+      "epoch": 0.696,
+      "grad_norm": 0.04338935023187058,
+      "learning_rate": 1.7941416508447537e-05,
+      "loss": 0.0004,
       "step": 435
     },
     {
-      "epoch": 0.6526946107784432,
-      "grad_norm": 0.18173894064740984,
-      "learning_rate": 1.8226657978137556e-05,
-      "loss": 0.0027,
+      "epoch": 0.6976,
+      "grad_norm": 0.1387899079098917,
+      "learning_rate": 1.793090273907056e-05,
+      "loss": 0.0018,
       "step": 436
     },
     {
-      "epoch": 0.6541916167664671,
-      "grad_norm": 0.11382010411529647,
-      "learning_rate": 1.8217454872455537e-05,
-      "loss": 0.0029,
+      "epoch": 0.6992,
+      "grad_norm": 0.04671281072738134,
+      "learning_rate": 1.792036528681418e-05,
+      "loss": 0.0006,
       "step": 437
     },
     {
-      "epoch": 0.655688622754491,
-      "grad_norm": 0.24972459202509598,
-      "learning_rate": 1.820823028394918e-05,
-      "loss": 0.0039,
+      "epoch": 0.7008,
+      "grad_norm": 0.04395180681044294,
+      "learning_rate": 1.7909804183144837e-05,
+      "loss": 0.0003,
       "step": 438
     },
     {
-      "epoch": 0.657185628742515,
-      "grad_norm": 0.19783767385058063,
-      "learning_rate": 1.8198984236734246e-05,
-      "loss": 0.0043,
+      "epoch": 0.7024,
+      "grad_norm": 0.08205063344030432,
+      "learning_rate": 1.789921945959958e-05,
+      "loss": 0.0019,
       "step": 439
     },
     {
-      "epoch": 0.6586826347305389,
-      "grad_norm": 0.2542153997631583,
-      "learning_rate": 1.8189716754982604e-05,
-      "loss": 0.0053,
+      "epoch": 0.704,
+      "grad_norm": 0.12018305372968996,
+      "learning_rate": 1.7888611147786003e-05,
+      "loss": 0.001,
       "step": 440
     },
     {
-      "epoch": 0.6601796407185628,
-      "grad_norm": 0.16722569392416708,
-      "learning_rate": 1.8180427862922157e-05,
-      "loss": 0.004,
+      "epoch": 0.7056,
+      "grad_norm": 0.010197674359924965,
+      "learning_rate": 1.7877979279382135e-05,
+      "loss": 0.0002,
       "step": 441
     },
     {
-      "epoch": 0.6616766467065869,
-      "grad_norm": 0.18714906390575894,
-      "learning_rate": 1.817111758483677e-05,
-      "loss": 0.0049,
+      "epoch": 0.7072,
+      "grad_norm": 0.0930050722342456,
+      "learning_rate": 1.786732388613635e-05,
+      "loss": 0.0008,
       "step": 442
     },
     {
-      "epoch": 0.6631736526946108,
-      "grad_norm": 0.14257993994735135,
-      "learning_rate": 1.8161785945066235e-05,
-      "loss": 0.0026,
+      "epoch": 0.7088,
+      "grad_norm": 0.07793396487766552,
+      "learning_rate": 1.7856644999867264e-05,
+      "loss": 0.0006,
       "step": 443
     },
     {
-      "epoch": 0.6646706586826348,
-      "grad_norm": 0.16935381936187344,
-      "learning_rate": 1.8152432968006176e-05,
-      "loss": 0.0036,
+      "epoch": 0.7104,
+      "grad_norm": 0.10577783436370959,
+      "learning_rate": 1.784594265246366e-05,
+      "loss": 0.0008,
       "step": 444
     },
     {
-      "epoch": 0.6661676646706587,
-      "grad_norm": 0.1342503188843719,
-      "learning_rate": 1.8143058678108002e-05,
-      "loss": 0.0028,
+      "epoch": 0.712,
+      "grad_norm": 0.06379628265447959,
+      "learning_rate": 1.783521687588437e-05,
+      "loss": 0.0007,
       "step": 445
     },
     {
-      "epoch": 0.6676646706586826,
-      "grad_norm": 0.25046636983085335,
-      "learning_rate": 1.8133663099878846e-05,
-      "loss": 0.0068,
+      "epoch": 0.7136,
+      "grad_norm": 0.013076260307259675,
+      "learning_rate": 1.782446770215819e-05,
+      "loss": 0.0002,
       "step": 446
     },
     {
-      "epoch": 0.6691616766467066,
-      "grad_norm": 0.19961420082825826,
-      "learning_rate": 1.812424625788149e-05,
-      "loss": 0.0035,
+      "epoch": 0.7152,
+      "grad_norm": 0.04927902298242426,
+      "learning_rate": 1.781369516338378e-05,
+      "loss": 0.0006,
       "step": 447
     },
     {
-      "epoch": 0.6706586826347305,
-      "grad_norm": 0.17568029731674176,
-      "learning_rate": 1.8114808176734302e-05,
-      "loss": 0.003,
+      "epoch": 0.7168,
+      "grad_norm": 0.03146289331146791,
+      "learning_rate": 1.7802899291729585e-05,
+      "loss": 0.0004,
       "step": 448
     },
     {
-      "epoch": 0.6721556886227545,
-      "grad_norm": 0.17477245221194124,
-      "learning_rate": 1.8105348881111182e-05,
-      "loss": 0.0047,
+      "epoch": 0.7184,
+      "grad_norm": 0.0437458279896608,
+      "learning_rate": 1.779208011943371e-05,
+      "loss": 0.0005,
       "step": 449
     },
     {
-      "epoch": 0.6736526946107785,
-      "grad_norm": 0.22428133343136628,
-      "learning_rate": 1.809586839574149e-05,
-      "loss": 0.0027,
+      "epoch": 0.72,
+      "grad_norm": 0.13167614822032764,
+      "learning_rate": 1.7781237678803845e-05,
+      "loss": 0.0026,
       "step": 450
     },
     {
-      "epoch": 0.6751497005988024,
-      "grad_norm": 0.19664866060996913,
-      "learning_rate": 1.808636674540998e-05,
-      "loss": 0.0034,
+      "epoch": 0.7216,
+      "grad_norm": 0.054018249029343136,
+      "learning_rate": 1.777037200221717e-05,
+      "loss": 0.0004,
       "step": 451
     },
     {
-      "epoch": 0.6766467065868264,
-      "grad_norm": 0.14083033586251215,
-      "learning_rate": 1.8076843954956736e-05,
-      "loss": 0.004,
+      "epoch": 0.7232,
+      "grad_norm": 0.04532830776342967,
+      "learning_rate": 1.775948312212024e-05,
+      "loss": 0.0005,
       "step": 452
     },
     {
-      "epoch": 0.6781437125748503,
-      "grad_norm": 0.1647022452825819,
-      "learning_rate": 1.8067300049277113e-05,
-      "loss": 0.0023,
+      "epoch": 0.7248,
+      "grad_norm": 0.03224800812198159,
+      "learning_rate": 1.77485710710289e-05,
+      "loss": 0.0004,
       "step": 453
     },
     {
-      "epoch": 0.6796407185628742,
-      "grad_norm": 0.32202530277073854,
-      "learning_rate": 1.805773505332166e-05,
-      "loss": 0.0066,
+      "epoch": 0.7264,
+      "grad_norm": 0.04750517727874858,
+      "learning_rate": 1.7737635881528198e-05,
+      "loss": 0.0004,
       "step": 454
     },
     {
-      "epoch": 0.6811377245508982,
-      "grad_norm": 0.14979165173614356,
-      "learning_rate": 1.8048148992096073e-05,
-      "loss": 0.0032,
+      "epoch": 0.728,
+      "grad_norm": 0.0928588978408594,
+      "learning_rate": 1.7726677586272263e-05,
+      "loss": 0.0009,
       "step": 455
     },
     {
-      "epoch": 0.6826347305389222,
-      "grad_norm": 0.1418060842276349,
-      "learning_rate": 1.8038541890661106e-05,
-      "loss": 0.0036,
+      "epoch": 0.7296,
+      "grad_norm": 0.08608768777399839,
+      "learning_rate": 1.7715696217984233e-05,
+      "loss": 0.0009,
       "step": 456
     },
     {
-      "epoch": 0.6841317365269461,
-      "grad_norm": 0.16706209980708875,
-      "learning_rate": 1.802891377413253e-05,
-      "loss": 0.0019,
+      "epoch": 0.7312,
+      "grad_norm": 0.09417667548425367,
+      "learning_rate": 1.7704691809456142e-05,
+      "loss": 0.0006,
       "step": 457
     },
     {
-      "epoch": 0.6856287425149701,
-      "grad_norm": 0.18905559651264367,
-      "learning_rate": 1.8019264667681053e-05,
-      "loss": 0.0023,
+      "epoch": 0.7328,
+      "grad_norm": 0.02712742007020753,
+      "learning_rate": 1.7693664393548822e-05,
+      "loss": 0.0003,
       "step": 458
     },
     {
-      "epoch": 0.687125748502994,
-      "grad_norm": 0.23786995943332953,
-      "learning_rate": 1.800959459653224e-05,
-      "loss": 0.0041,
+      "epoch": 0.7344,
+      "grad_norm": 0.1108581252930463,
+      "learning_rate": 1.7682614003191807e-05,
+      "loss": 0.0007,
       "step": 459
     },
     {
-      "epoch": 0.688622754491018,
-      "grad_norm": 0.11727770023510513,
-      "learning_rate": 1.79999035859665e-05,
-      "loss": 0.0018,
+      "epoch": 0.736,
+      "grad_norm": 0.012332852158620117,
+      "learning_rate": 1.7671540671383245e-05,
+      "loss": 0.0002,
       "step": 460
     },
     {
-      "epoch": 0.6901197604790419,
-      "grad_norm": 0.190583040451331,
-      "learning_rate": 1.7990191661318942e-05,
-      "loss": 0.0044,
+      "epoch": 0.7376,
+      "grad_norm": 0.1661080884851385,
+      "learning_rate": 1.766044443118978e-05,
+      "loss": 0.0014,
       "step": 461
     },
     {
-      "epoch": 0.6916167664670658,
-      "grad_norm": 0.19155482816471409,
-      "learning_rate": 1.798045884797938e-05,
-      "loss": 0.0034,
+      "epoch": 0.7392,
+      "grad_norm": 0.15411155835183601,
+      "learning_rate": 1.764932531574648e-05,
+      "loss": 0.0014,
       "step": 462
     },
     {
-      "epoch": 0.6931137724550899,
-      "grad_norm": 0.2432660367460031,
-      "learning_rate": 1.797070517139223e-05,
-      "loss": 0.0064,
+      "epoch": 0.7408,
+      "grad_norm": 0.06491648792911973,
+      "learning_rate": 1.76381833582567e-05,
+      "loss": 0.0004,
       "step": 463
     },
     {
-      "epoch": 0.6946107784431138,
-      "grad_norm": 0.16246882216240055,
-      "learning_rate": 1.796093065705644e-05,
-      "loss": 0.0033,
+      "epoch": 0.7424,
+      "grad_norm": 0.1235036068169899,
+      "learning_rate": 1.762701859199202e-05,
+      "loss": 0.0013,
       "step": 464
     },
     {
-      "epoch": 0.6961077844311377,
-      "grad_norm": 0.10829584717187421,
-      "learning_rate": 1.7951135330525447e-05,
-      "loss": 0.0021,
+      "epoch": 0.744,
+      "grad_norm": 0.03855206819431079,
+      "learning_rate": 1.761583105029213e-05,
+      "loss": 0.0004,
       "step": 465
     },
     {
-      "epoch": 0.6976047904191617,
-      "grad_norm": 0.13169748817522234,
-      "learning_rate": 1.794131921740709e-05,
-      "loss": 0.0034,
+      "epoch": 0.7456,
+      "grad_norm": 0.05325264463367167,
+      "learning_rate": 1.7604620766564725e-05,
+      "loss": 0.0006,
       "step": 466
     },
     {
-      "epoch": 0.6991017964071856,
-      "grad_norm": 0.22261122260930077,
-      "learning_rate": 1.7931482343363552e-05,
-      "loss": 0.0032,
+      "epoch": 0.7472,
+      "grad_norm": 0.04346566688047951,
+      "learning_rate": 1.7593387774285412e-05,
+      "loss": 0.0004,
       "step": 467
     },
     {
-      "epoch": 0.7005988023952096,
-      "grad_norm": 0.21429988838318564,
-      "learning_rate": 1.7921624734111292e-05,
-      "loss": 0.0064,
+      "epoch": 0.7488,
+      "grad_norm": 0.136998305712324,
+      "learning_rate": 1.7582132106997615e-05,
+      "loss": 0.0021,
       "step": 468
     },
     {
-      "epoch": 0.7020958083832335,
-      "grad_norm": 0.15085978142323003,
-      "learning_rate": 1.7911746415420982e-05,
-      "loss": 0.0019,
+      "epoch": 0.7504,
+      "grad_norm": 0.007818539193188706,
+      "learning_rate": 1.7570853798312462e-05,
+      "loss": 0.0002,
       "step": 469
     },
     {
-      "epoch": 0.7035928143712575,
-      "grad_norm": 0.09480266222976796,
-      "learning_rate": 1.7901847413117424e-05,
-      "loss": 0.0018,
+      "epoch": 0.752,
+      "grad_norm": 0.01723914456383822,
+      "learning_rate": 1.7559552881908698e-05,
+      "loss": 0.0002,
       "step": 470
     },
     {
-      "epoch": 0.7050898203592815,
-      "grad_norm": 0.1840037147393193,
-      "learning_rate": 1.7891927753079495e-05,
-      "loss": 0.0033,
+      "epoch": 0.7536,
+      "grad_norm": 0.07756515160296149,
+      "learning_rate": 1.7548229391532572e-05,
+      "loss": 0.0005,
       "step": 471
     },
     {
-      "epoch": 0.7065868263473054,
-      "grad_norm": 0.12251157236145369,
-      "learning_rate": 1.7881987461240086e-05,
-      "loss": 0.0019,
+      "epoch": 0.7552,
+      "grad_norm": 0.023715500088758376,
+      "learning_rate": 1.7536883360997743e-05,
+      "loss": 0.0003,
       "step": 472
     },
     {
-      "epoch": 0.7080838323353293,
-      "grad_norm": 0.3770740041300004,
-      "learning_rate": 1.7872026563586015e-05,
-      "loss": 0.0061,
+      "epoch": 0.7568,
+      "grad_norm": 0.13636676401004547,
+      "learning_rate": 1.7525514824185187e-05,
+      "loss": 0.0019,
       "step": 473
     },
     {
-      "epoch": 0.7095808383233533,
-      "grad_norm": 0.14686337723150117,
-      "learning_rate": 1.7862045086157975e-05,
-      "loss": 0.0036,
+      "epoch": 0.7584,
+      "grad_norm": 0.0144699251390561,
+      "learning_rate": 1.7514123815043073e-05,
+      "loss": 0.0002,
       "step": 474
     },
     {
-      "epoch": 0.7110778443113772,
-      "grad_norm": 0.12202140713465083,
-      "learning_rate": 1.7852043055050466e-05,
-      "loss": 0.0028,
+      "epoch": 0.76,
+      "grad_norm": 0.02802581269937591,
+      "learning_rate": 1.750271036758669e-05,
+      "loss": 0.0002,
       "step": 475
     },
     {
-      "epoch": 0.7125748502994012,
-      "grad_norm": 0.20749262333762236,
-      "learning_rate": 1.7842020496411707e-05,
-      "loss": 0.0021,
+      "epoch": 0.7616,
+      "grad_norm": 0.13859051756276258,
+      "learning_rate": 1.749127451589832e-05,
+      "loss": 0.001,
       "step": 476
     },
     {
-      "epoch": 0.7140718562874252,
-      "grad_norm": 0.243735318182604,
-      "learning_rate": 1.7831977436443594e-05,
-      "loss": 0.004,
+      "epoch": 0.7632,
+      "grad_norm": 0.08188164186474299,
+      "learning_rate": 1.747981629412715e-05,
+      "loss": 0.0031,
       "step": 477
     },
     {
-      "epoch": 0.7155688622754491,
-      "grad_norm": 0.2123776492606604,
-      "learning_rate": 1.7821913901401616e-05,
-      "loss": 0.0033,
+      "epoch": 0.7648,
+      "grad_norm": 0.1023395386215927,
+      "learning_rate": 1.7468335736489177e-05,
+      "loss": 0.001,
       "step": 478
     },
     {
-      "epoch": 0.7170658682634731,
-      "grad_norm": 0.0860593828548492,
-      "learning_rate": 1.781182991759479e-05,
-      "loss": 0.0012,
+      "epoch": 0.7664,
+      "grad_norm": 0.233632163259218,
+      "learning_rate": 1.7456832877267083e-05,
+      "loss": 0.003,
       "step": 479
     },
     {
-      "epoch": 0.718562874251497,
-      "grad_norm": 0.17480422685210906,
-      "learning_rate": 1.780172551138559e-05,
-      "loss": 0.0037,
+      "epoch": 0.768,
+      "grad_norm": 0.40600459683828993,
+      "learning_rate": 1.7445307750810153e-05,
+      "loss": 0.0002,
       "step": 480
     },
     {
-      "epoch": 0.7200598802395209,
-      "grad_norm": 0.2672043506973695,
-      "learning_rate": 1.779160070918988e-05,
-      "loss": 0.0046,
+      "epoch": 0.7696,
+      "grad_norm": 0.03720264501874191,
+      "learning_rate": 1.7433760391534166e-05,
+      "loss": 0.0003,
       "step": 481
     },
     {
-      "epoch": 0.7215568862275449,
-      "grad_norm": 0.30149827478860836,
-      "learning_rate": 1.7781455537476848e-05,
-      "loss": 0.0052,
+      "epoch": 0.7712,
+      "grad_norm": 0.0053336116920516185,
+      "learning_rate": 1.7422190833921284e-05,
+      "loss": 0.0001,
       "step": 482
     },
     {
-      "epoch": 0.7230538922155688,
-      "grad_norm": 0.23829191956670007,
-      "learning_rate": 1.7771290022768933e-05,
-      "loss": 0.0032,
+      "epoch": 0.7728,
+      "grad_norm": 0.010534130055701448,
+      "learning_rate": 1.741059911251997e-05,
+      "loss": 0.0002,
       "step": 483
     },
     {
-      "epoch": 0.7245508982035929,
-      "grad_norm": 0.19471894040554305,
-      "learning_rate": 1.776110419164176e-05,
-      "loss": 0.0025,
+      "epoch": 0.7744,
+      "grad_norm": 0.0279036924483479,
+      "learning_rate": 1.7398985261944857e-05,
+      "loss": 0.0003,
       "step": 484
     },
     {
-      "epoch": 0.7260479041916168,
-      "grad_norm": 0.3281317127559268,
-      "learning_rate": 1.7750898070724057e-05,
-      "loss": 0.0053,
+      "epoch": 0.776,
+      "grad_norm": 0.07436924158810333,
+      "learning_rate": 1.7387349316876668e-05,
+      "loss": 0.0044,
       "step": 485
     },
     {
-      "epoch": 0.7275449101796407,
-      "grad_norm": 0.22137467518837572,
-      "learning_rate": 1.7740671686697605e-05,
-      "loss": 0.005,
+      "epoch": 0.7776,
+      "grad_norm": 0.0782832167728117,
+      "learning_rate": 1.7375691312062102e-05,
+      "loss": 0.0021,
       "step": 486
     },
     {
-      "epoch": 0.7290419161676647,
-      "grad_norm": 0.16708307413649082,
-      "learning_rate": 1.7730425066297158e-05,
-      "loss": 0.0037,
+      "epoch": 0.7792,
+      "grad_norm": 0.00800420181406178,
+      "learning_rate": 1.7364011282313732e-05,
+      "loss": 0.0002,
       "step": 487
     },
     {
-      "epoch": 0.7305389221556886,
-      "grad_norm": 0.0707938538877929,
-      "learning_rate": 1.772015823631036e-05,
-      "loss": 0.0023,
+      "epoch": 0.7808,
+      "grad_norm": 0.06008163797598921,
+      "learning_rate": 1.7352309262509894e-05,
+      "loss": 0.0006,
       "step": 488
     },
     {
-      "epoch": 0.7320359281437125,
-      "grad_norm": 0.1183727230604871,
-      "learning_rate": 1.7709871223577713e-05,
-      "loss": 0.0024,
+      "epoch": 0.7824,
+      "grad_norm": 0.2942563905755528,
+      "learning_rate": 1.7340585287594605e-05,
+      "loss": 0.0011,
       "step": 489
     },
     {
-      "epoch": 0.7335329341317365,
-      "grad_norm": 0.20204690451609672,
-      "learning_rate": 1.769956405499246e-05,
-      "loss": 0.0045,
+      "epoch": 0.784,
+      "grad_norm": 0.08926818808154893,
+      "learning_rate": 1.7328839392577422e-05,
+      "loss": 0.0008,
       "step": 490
     },
     {
-      "epoch": 0.7350299401197605,
-      "grad_norm": 0.22427020009198373,
-      "learning_rate": 1.768923675750055e-05,
-      "loss": 0.0037,
+      "epoch": 0.7856,
+      "grad_norm": 0.04506569681041617,
+      "learning_rate": 1.731707161253338e-05,
+      "loss": 0.0003,
       "step": 491
     },
     {
-      "epoch": 0.7365269461077845,
-      "grad_norm": 0.09360542375246439,
-      "learning_rate": 1.767888935810055e-05,
-      "loss": 0.0026,
+      "epoch": 0.7872,
+      "grad_norm": 0.07750705989037712,
+      "learning_rate": 1.730528198260285e-05,
+      "loss": 0.0005,
       "step": 492
     },
     {
-      "epoch": 0.7380239520958084,
-      "grad_norm": 0.18339744192805107,
-      "learning_rate": 1.7668521883843586e-05,
-      "loss": 0.003,
+      "epoch": 0.7888,
+      "grad_norm": 0.09605363767215523,
+      "learning_rate": 1.7293470537991463e-05,
+      "loss": 0.0007,
       "step": 493
     },
     {
-      "epoch": 0.7395209580838323,
-      "grad_norm": 0.1079849963238365,
-      "learning_rate": 1.7658134361833253e-05,
-      "loss": 0.0019,
+      "epoch": 0.7904,
+      "grad_norm": 0.01733481247306688,
+      "learning_rate": 1.728163731396998e-05,
+      "loss": 0.0003,
       "step": 494
     },
     {
-      "epoch": 0.7410179640718563,
-      "grad_norm": 0.1345391487313826,
-      "learning_rate": 1.764772681922557e-05,
-      "loss": 0.0027,
+      "epoch": 0.792,
+      "grad_norm": 0.06161393660807383,
+      "learning_rate": 1.7269782345874204e-05,
+      "loss": 0.0009,
       "step": 495
     },
     {
-      "epoch": 0.7425149700598802,
-      "grad_norm": 0.14603341686152715,
-      "learning_rate": 1.7637299283228887e-05,
-      "loss": 0.0032,
+      "epoch": 0.7936,
+      "grad_norm": 0.12245228516343439,
+      "learning_rate": 1.7257905669104874e-05,
+      "loss": 0.0005,
       "step": 496
     },
     {
-      "epoch": 0.7440119760479041,
-      "grad_norm": 0.16829676525727566,
-      "learning_rate": 1.762685178110382e-05,
-      "loss": 0.0029,
+      "epoch": 0.7952,
+      "grad_norm": 0.14061289559871648,
+      "learning_rate": 1.7246007319127547e-05,
+      "loss": 0.0022,
       "step": 497
     },
     {
-      "epoch": 0.7455089820359282,
-      "grad_norm": 0.2128675734460732,
-      "learning_rate": 1.7616384340163196e-05,
-      "loss": 0.0019,
+      "epoch": 0.7968,
+      "grad_norm": 0.09233719827904271,
+      "learning_rate": 1.72340873314725e-05,
+      "loss": 0.0009,
       "step": 498
     },
     {
-      "epoch": 0.7470059880239521,
-      "grad_norm": 0.12654769972937424,
-      "learning_rate": 1.7605896987771957e-05,
-      "loss": 0.0026,
+      "epoch": 0.7984,
+      "grad_norm": 0.009727144786638353,
+      "learning_rate": 1.7222145741734625e-05,
+      "loss": 0.0002,
       "step": 499
     },
     {
-      "epoch": 0.7485029940119761,
-      "grad_norm": 0.05725603798182813,
-      "learning_rate": 1.75953897513471e-05,
-      "loss": 0.0011,
+      "epoch": 0.8,
+      "grad_norm": 0.15069270935914425,
+      "learning_rate": 1.721018258557333e-05,
+      "loss": 0.0023,
       "step": 500
     },
     {
-      "epoch": 0.75,
-      "grad_norm": 0.12305721333189583,
-      "learning_rate": 1.7584862658357603e-05,
-      "loss": 0.002,
+      "epoch": 0.8016,
+      "grad_norm": 0.13093952054952349,
+      "learning_rate": 1.7198197898712402e-05,
+      "loss": 0.0008,
       "step": 501
     },
     {
-      "epoch": 0.7514970059880239,
-      "grad_norm": 0.1998661971442016,
-      "learning_rate": 1.757431573632437e-05,
-      "loss": 0.0028,
+      "epoch": 0.8032,
+      "grad_norm": 0.08970411506665783,
+      "learning_rate": 1.7186191716939946e-05,
+      "loss": 0.0006,
       "step": 502
     },
     {
-      "epoch": 0.7529940119760479,
-      "grad_norm": 0.21334825352182804,
-      "learning_rate": 1.7563749012820115e-05,
-      "loss": 0.0042,
+      "epoch": 0.8048,
+      "grad_norm": 0.02490276069628502,
+      "learning_rate": 1.717416407610824e-05,
+      "loss": 0.0003,
       "step": 503
     },
     {
-      "epoch": 0.7544910179640718,
-      "grad_norm": 0.08550116259152069,
-      "learning_rate": 1.7553162515469346e-05,
-      "loss": 0.0015,
+      "epoch": 0.8064,
+      "grad_norm": 0.04979174521877706,
+      "learning_rate": 1.7162115012133643e-05,
+      "loss": 0.0004,
       "step": 504
     },
     {
-      "epoch": 0.7559880239520959,
-      "grad_norm": 0.3173087453070982,
-      "learning_rate": 1.754255627194826e-05,
-      "loss": 0.0027,
+      "epoch": 0.808,
+      "grad_norm": 0.00856385237921387,
+      "learning_rate": 1.7150044560996488e-05,
+      "loss": 0.0002,
       "step": 505
     },
     {
-      "epoch": 0.7574850299401198,
-      "grad_norm": 0.12840517943608828,
-      "learning_rate": 1.7531930309984663e-05,
-      "loss": 0.0021,
+      "epoch": 0.8096,
+      "grad_norm": 0.008284973985246698,
+      "learning_rate": 1.713795275874098e-05,
+      "loss": 0.0002,
       "step": 506
     },
     {
-      "epoch": 0.7589820359281437,
-      "grad_norm": 0.22380401998833405,
-      "learning_rate": 1.752128465735792e-05,
-      "loss": 0.0042,
+      "epoch": 0.8112,
+      "grad_norm": 0.06734893182692521,
+      "learning_rate": 1.7125839641475074e-05,
+      "loss": 0.0006,
       "step": 507
     },
     {
-      "epoch": 0.7604790419161677,
-      "grad_norm": 0.17755574559866727,
-      "learning_rate": 1.751061934189888e-05,
-      "loss": 0.0043,
+      "epoch": 0.8128,
+      "grad_norm": 0.18508359604792338,
+      "learning_rate": 1.711370524537037e-05,
+      "loss": 0.0025,
       "step": 508
     },
     {
-      "epoch": 0.7619760479041916,
-      "grad_norm": 0.13841142360132738,
-      "learning_rate": 1.749993439148978e-05,
-      "loss": 0.0027,
+      "epoch": 0.8144,
+      "grad_norm": 0.13809444329979642,
+      "learning_rate": 1.7101549606662025e-05,
+      "loss": 0.0011,
       "step": 509
     },
     {
-      "epoch": 0.7634730538922155,
-      "grad_norm": 0.19949827808136236,
-      "learning_rate": 1.7489229834064207e-05,
-      "loss": 0.0015,
+      "epoch": 0.816,
+      "grad_norm": 0.5696537447435555,
+      "learning_rate": 1.7089372761648617e-05,
+      "loss": 0.0006,
       "step": 510
     },
     {
-      "epoch": 0.7649700598802395,
-      "grad_norm": 0.08462524659982372,
-      "learning_rate": 1.747850569760699e-05,
-      "loss": 0.002,
+      "epoch": 0.8176,
+      "grad_norm": 0.02721585821128559,
+      "learning_rate": 1.7077174746692054e-05,
+      "loss": 0.0004,
       "step": 511
     },
     {
-      "epoch": 0.7664670658682635,
-      "grad_norm": 0.185019733009154,
-      "learning_rate": 1.7467762010154148e-05,
-      "loss": 0.0028,
+      "epoch": 0.8192,
+      "grad_norm": 0.09701224812397982,
+      "learning_rate": 1.7064955598217463e-05,
+      "loss": 0.0008,
       "step": 512
     },
     {
-      "epoch": 0.7679640718562875,
-      "grad_norm": 0.14537973618978656,
-      "learning_rate": 1.7456998799792822e-05,
-      "loss": 0.0045,
+      "epoch": 0.8208,
+      "grad_norm": 0.049111193053924744,
+      "learning_rate": 1.7052715352713076e-05,
+      "loss": 0.0004,
       "step": 513
     },
     {
-      "epoch": 0.7694610778443114,
-      "grad_norm": 0.0919302359343689,
-      "learning_rate": 1.7446216094661183e-05,
-      "loss": 0.0017,
+      "epoch": 0.8224,
+      "grad_norm": 0.12230021958553583,
+      "learning_rate": 1.7040454046730118e-05,
+      "loss": 0.0009,
       "step": 514
     },
     {
-      "epoch": 0.7709580838323353,
-      "grad_norm": 0.09806191993657888,
-      "learning_rate": 1.7435413922948358e-05,
-      "loss": 0.0015,
+      "epoch": 0.824,
+      "grad_norm": 0.012243744364003616,
+      "learning_rate": 1.7028171716882714e-05,
+      "loss": 0.0002,
       "step": 515
     },
     {
-      "epoch": 0.7724550898203593,
-      "grad_norm": 0.17581073699332905,
-      "learning_rate": 1.742459231289439e-05,
-      "loss": 0.0036,
+      "epoch": 0.8256,
+      "grad_norm": 0.08499201626336292,
+      "learning_rate": 1.7015868399847768e-05,
+      "loss": 0.0004,
       "step": 516
     },
     {
-      "epoch": 0.7739520958083832,
-      "grad_norm": 0.16231931806189956,
-      "learning_rate": 1.7413751292790113e-05,
-      "loss": 0.0035,
+      "epoch": 0.8272,
+      "grad_norm": 0.27820795609998483,
+      "learning_rate": 1.7003544132364847e-05,
+      "loss": 0.0034,
       "step": 517
     },
     {
-      "epoch": 0.7754491017964071,
-      "grad_norm": 0.13525922667777576,
-      "learning_rate": 1.740289089097712e-05,
-      "loss": 0.0031,
+      "epoch": 0.8288,
+      "grad_norm": 0.02335971532394588,
+      "learning_rate": 1.6991198951236088e-05,
+      "loss": 0.0003,
       "step": 518
     },
     {
-      "epoch": 0.7769461077844312,
-      "grad_norm": 0.20823952255636147,
-      "learning_rate": 1.739201113584767e-05,
-      "loss": 0.0023,
+      "epoch": 0.8304,
+      "grad_norm": 0.4862075696612623,
+      "learning_rate": 1.6978832893326074e-05,
+      "loss": 0.0024,
       "step": 519
     },
     {
-      "epoch": 0.7784431137724551,
-      "grad_norm": 0.16208283117260136,
-      "learning_rate": 1.738111205584462e-05,
-      "loss": 0.0014,
+      "epoch": 0.832,
+      "grad_norm": 0.0493441407586534,
+      "learning_rate": 1.696644599556173e-05,
+      "loss": 0.0005,
       "step": 520
     },
     {
-      "epoch": 0.7799401197604791,
-      "grad_norm": 0.07133110277997594,
-      "learning_rate": 1.737019367946134e-05,
-      "loss": 0.0012,
+      "epoch": 0.8336,
+      "grad_norm": 0.006952259716357961,
+      "learning_rate": 1.6954038294932215e-05,
+      "loss": 0.0002,
       "step": 521
     },
     {
-      "epoch": 0.781437125748503,
-      "grad_norm": 0.08862054013594506,
-      "learning_rate": 1.735925603524165e-05,
-      "loss": 0.0015,
+      "epoch": 0.8352,
+      "grad_norm": 0.06091477273979244,
+      "learning_rate": 1.6941609828488806e-05,
+      "loss": 0.0004,
       "step": 522
     },
     {
-      "epoch": 0.7829341317365269,
-      "grad_norm": 0.18494335464634515,
-      "learning_rate": 1.7348299151779748e-05,
-      "loss": 0.0023,
+      "epoch": 0.8368,
+      "grad_norm": 0.07508272808691038,
+      "learning_rate": 1.692916063334479e-05,
+      "loss": 0.003,
       "step": 523
     },
     {
-      "epoch": 0.7844311377245509,
-      "grad_norm": 0.2671256300047652,
-      "learning_rate": 1.7337323057720122e-05,
-      "loss": 0.0047,
+      "epoch": 0.8384,
+      "grad_norm": 0.01362913734282188,
+      "learning_rate": 1.691669074667535e-05,
+      "loss": 0.0002,
       "step": 524
     },
     {
-      "epoch": 0.7859281437125748,
-      "grad_norm": 0.2617413925072033,
-      "learning_rate": 1.7326327781757482e-05,
-      "loss": 0.0038,
+      "epoch": 0.84,
+      "grad_norm": 0.10616161563646664,
+      "learning_rate": 1.690420020571747e-05,
+      "loss": 0.0006,
       "step": 525
     },
     {
-      "epoch": 0.7874251497005988,
-      "grad_norm": 0.125912865428105,
-      "learning_rate": 1.731531335263669e-05,
-      "loss": 0.0028,
+      "epoch": 0.8416,
+      "grad_norm": 0.03894127123081356,
+      "learning_rate": 1.689168904776979e-05,
+      "loss": 0.0003,
       "step": 526
     },
     {
-      "epoch": 0.7889221556886228,
-      "grad_norm": 0.13267509240686437,
-      "learning_rate": 1.7304279799152677e-05,
-      "loss": 0.0033,
+      "epoch": 0.8432,
+      "grad_norm": 0.1378168598257233,
+      "learning_rate": 1.6879157310192537e-05,
+      "loss": 0.0023,
       "step": 527
     },
     {
-      "epoch": 0.7904191616766467,
-      "grad_norm": 0.14976117950694603,
-      "learning_rate": 1.7293227150150366e-05,
-      "loss": 0.0029,
+      "epoch": 0.8448,
+      "grad_norm": 0.05358355520865817,
+      "learning_rate": 1.686660503040737e-05,
+      "loss": 0.0005,
       "step": 528
     },
     {
-      "epoch": 0.7919161676646707,
-      "grad_norm": 0.09070339743848918,
-      "learning_rate": 1.728215543452461e-05,
-      "loss": 0.0016,
+      "epoch": 0.8464,
+      "grad_norm": 0.041106143597292144,
+      "learning_rate": 1.685403224589731e-05,
+      "loss": 0.0004,
       "step": 529
     },
     {
-      "epoch": 0.7934131736526946,
-      "grad_norm": 0.14490201715247109,
-      "learning_rate": 1.7271064681220105e-05,
-      "loss": 0.0016,
+      "epoch": 0.848,
+      "grad_norm": 0.019024841308057497,
+      "learning_rate": 1.6841438994206597e-05,
+      "loss": 0.0003,
       "step": 530
     },
     {
-      "epoch": 0.7949101796407185,
-      "grad_norm": 0.22041533329824736,
-      "learning_rate": 1.725995491923131e-05,
-      "loss": 0.0048,
+      "epoch": 0.8496,
+      "grad_norm": 0.09954312413327895,
+      "learning_rate": 1.6828825312940594e-05,
+      "loss": 0.004,
       "step": 531
     },
     {
-      "epoch": 0.7964071856287425,
-      "grad_norm": 0.18152422489317313,
-      "learning_rate": 1.724882617760239e-05,
-      "loss": 0.0024,
+      "epoch": 0.8512,
+      "grad_norm": 0.01904708410465074,
+      "learning_rate": 1.6816191239765668e-05,
+      "loss": 0.0002,
       "step": 532
     },
     {
-      "epoch": 0.7979041916167665,
-      "grad_norm": 0.13200938457839367,
-      "learning_rate": 1.7237678485427117e-05,
-      "loss": 0.0025,
+      "epoch": 0.8528,
+      "grad_norm": 0.057975810389029034,
+      "learning_rate": 1.6803536812409077e-05,
+      "loss": 0.0004,
       "step": 533
     },
     {
-      "epoch": 0.7994011976047904,
-      "grad_norm": 0.2569963553188449,
-      "learning_rate": 1.7226511871848816e-05,
-      "loss": 0.003,
+      "epoch": 0.8544,
+      "grad_norm": 0.11015783475479203,
+      "learning_rate": 1.6790862068658863e-05,
+      "loss": 0.0008,
       "step": 534
     },
     {
-      "epoch": 0.8008982035928144,
-      "grad_norm": 0.11384609771555894,
-      "learning_rate": 1.7215326366060266e-05,
-      "loss": 0.0024,
+      "epoch": 0.856,
+      "grad_norm": 0.04375619009470118,
+      "learning_rate": 1.6778167046363735e-05,
+      "loss": 0.0003,
       "step": 535
     },
     {
-      "epoch": 0.8023952095808383,
-      "grad_norm": 0.07638650411484138,
-      "learning_rate": 1.720412199730365e-05,
-      "loss": 0.0011,
+      "epoch": 0.8576,
+      "grad_norm": 0.03149655971331083,
+      "learning_rate": 1.6765451783432953e-05,
+      "loss": 0.0003,
       "step": 536
     },
     {
-      "epoch": 0.8038922155688623,
-      "grad_norm": 0.12630645830506568,
-      "learning_rate": 1.719289879487045e-05,
-      "loss": 0.0017,
+      "epoch": 0.8592,
+      "grad_norm": 0.015188077650469912,
+      "learning_rate": 1.675271631783623e-05,
+      "loss": 0.0002,
       "step": 537
     },
     {
-      "epoch": 0.8053892215568862,
-      "grad_norm": 0.17122405410573677,
-      "learning_rate": 1.7181656788101394e-05,
-      "loss": 0.0018,
+      "epoch": 0.8608,
+      "grad_norm": 0.08937385959607695,
+      "learning_rate": 1.6739960687603592e-05,
+      "loss": 0.0017,
       "step": 538
     },
     {
-      "epoch": 0.8068862275449101,
-      "grad_norm": 0.11978605848911922,
-      "learning_rate": 1.7170396006386364e-05,
-      "loss": 0.0022,
+      "epoch": 0.8624,
+      "grad_norm": 0.046733492907307606,
+      "learning_rate": 1.672718493082529e-05,
+      "loss": 0.0004,
       "step": 539
     },
     {
-      "epoch": 0.8083832335329342,
-      "grad_norm": 0.22805307525076485,
-      "learning_rate": 1.7159116479164336e-05,
-      "loss": 0.0017,
+      "epoch": 0.864,
+      "grad_norm": 0.14393512602359823,
+      "learning_rate": 1.671438908565167e-05,
+      "loss": 0.0011,
       "step": 540
     },
     {
-      "epoch": 0.8098802395209581,
-      "grad_norm": 0.042607090167363824,
-      "learning_rate": 1.7147818235923282e-05,
-      "loss": 0.001,
+      "epoch": 0.8656,
+      "grad_norm": 0.03969373767492347,
+      "learning_rate": 1.6701573190293076e-05,
+      "loss": 0.0005,
       "step": 541
     },
     {
-      "epoch": 0.811377245508982,
-      "grad_norm": 0.3303493277968501,
-      "learning_rate": 1.71365013062001e-05,
-      "loss": 0.003,
+      "epoch": 0.8672,
+      "grad_norm": 0.07936059414567308,
+      "learning_rate": 1.6688737283019708e-05,
+      "loss": 0.0005,
       "step": 542
     },
     {
-      "epoch": 0.812874251497006,
-      "grad_norm": 0.20139977963643052,
-      "learning_rate": 1.712516571958055e-05,
-      "loss": 0.0051,
+      "epoch": 0.8688,
+      "grad_norm": 0.03756583107434095,
+      "learning_rate": 1.667588140216154e-05,
+      "loss": 0.0003,
       "step": 543
     },
     {
-      "epoch": 0.8143712574850299,
-      "grad_norm": 0.18852567500825385,
-      "learning_rate": 1.711381150569917e-05,
-      "loss": 0.0023,
+      "epoch": 0.8704,
+      "grad_norm": 0.06389471297505031,
+      "learning_rate": 1.6663005586108175e-05,
+      "loss": 0.0005,
       "step": 544
     },
     {
-      "epoch": 0.8158682634730539,
-      "grad_norm": 0.1533989779344055,
-      "learning_rate": 1.7102438694239172e-05,
-      "loss": 0.0015,
+      "epoch": 0.872,
+      "grad_norm": 0.027066712019042576,
+      "learning_rate": 1.6650109873308763e-05,
+      "loss": 0.0003,
       "step": 545
     },
     {
-      "epoch": 0.8173652694610778,
-      "grad_norm": 0.18945005903791085,
-      "learning_rate": 1.7091047314932418e-05,
-      "loss": 0.0052,
+      "epoch": 0.8736,
+      "grad_norm": 0.07561152443479349,
+      "learning_rate": 1.663719430227186e-05,
+      "loss": 0.0006,
       "step": 546
     },
     {
-      "epoch": 0.8188622754491018,
-      "grad_norm": 0.10940559482924138,
-      "learning_rate": 1.7079637397559284e-05,
-      "loss": 0.0036,
+      "epoch": 0.8752,
+      "grad_norm": 0.04267934929242688,
+      "learning_rate": 1.6624258911565312e-05,
+      "loss": 0.0003,
       "step": 547
     },
     {
-      "epoch": 0.8203592814371258,
-      "grad_norm": 0.1733647440192747,
-      "learning_rate": 1.706820897194863e-05,
-      "loss": 0.0026,
+      "epoch": 0.8768,
+      "grad_norm": 0.16933992446835514,
+      "learning_rate": 1.661130373981617e-05,
+      "loss": 0.0014,
       "step": 548
     },
     {
-      "epoch": 0.8218562874251497,
-      "grad_norm": 0.2284794974338924,
-      "learning_rate": 1.7056762067977696e-05,
-      "loss": 0.0026,
+      "epoch": 0.8784,
+      "grad_norm": 0.25073619762167976,
+      "learning_rate": 1.6598328825710536e-05,
+      "loss": 0.0013,
       "step": 549
     },
     {
-      "epoch": 0.8233532934131736,
-      "grad_norm": 0.14329072634628595,
-      "learning_rate": 1.7045296715572024e-05,
-      "loss": 0.0024,
+      "epoch": 0.88,
+      "grad_norm": 0.16778477153881266,
+      "learning_rate": 1.6585334207993475e-05,
+      "loss": 0.0004,
       "step": 550
     },
     {
-      "epoch": 0.8248502994011976,
-      "grad_norm": 0.31093763718850237,
-      "learning_rate": 1.703381294470539e-05,
-      "loss": 0.0033,
+      "epoch": 0.8816,
+      "grad_norm": 0.11826857642808859,
+      "learning_rate": 1.6572319925468892e-05,
+      "loss": 0.001,
       "step": 551
     },
     {
-      "epoch": 0.8263473053892215,
-      "grad_norm": 0.20477098480741565,
-      "learning_rate": 1.7022310785399722e-05,
-      "loss": 0.0028,
+      "epoch": 0.8832,
+      "grad_norm": 0.04772459571547264,
+      "learning_rate": 1.65592860169994e-05,
+      "loss": 0.0003,
       "step": 552
     },
     {
-      "epoch": 0.8278443113772455,
-      "grad_norm": 0.10382900594838523,
-      "learning_rate": 1.7010790267725023e-05,
-      "loss": 0.0017,
+      "epoch": 0.8848,
+      "grad_norm": 0.02117988616831091,
+      "learning_rate": 1.654623252150624e-05,
+      "loss": 0.0002,
       "step": 553
     },
     {
-      "epoch": 0.8293413173652695,
-      "grad_norm": 0.10870430271363642,
-      "learning_rate": 1.6999251421799285e-05,
-      "loss": 0.0014,
+      "epoch": 0.8864,
+      "grad_norm": 0.048392587850409105,
+      "learning_rate": 1.6533159477969122e-05,
+      "loss": 0.0004,
       "step": 554
     },
     {
-      "epoch": 0.8308383233532934,
-      "grad_norm": 0.2938613690271062,
-      "learning_rate": 1.698769427778842e-05,
-      "loss": 0.0043,
+      "epoch": 0.888,
+      "grad_norm": 0.006232812342251042,
+      "learning_rate": 1.6520066925426146e-05,
+      "loss": 0.0001,
       "step": 555
     },
     {
-      "epoch": 0.8323353293413174,
-      "grad_norm": 0.16919545121928964,
-      "learning_rate": 1.6976118865906167e-05,
-      "loss": 0.0024,
+      "epoch": 0.8896,
+      "grad_norm": 0.057442234935372385,
+      "learning_rate": 1.6506954902973657e-05,
+      "loss": 0.0005,
       "step": 556
     },
     {
-      "epoch": 0.8338323353293413,
-      "grad_norm": 0.19047981855746474,
-      "learning_rate": 1.6964525216414033e-05,
-      "loss": 0.0028,
+      "epoch": 0.8912,
+      "grad_norm": 0.2509062616236011,
+      "learning_rate": 1.6493823449766137e-05,
+      "loss": 0.0032,
       "step": 557
     },
     {
-      "epoch": 0.8353293413173652,
-      "grad_norm": 0.15373361819794654,
-      "learning_rate": 1.6952913359621204e-05,
-      "loss": 0.0028,
+      "epoch": 0.8928,
+      "grad_norm": 0.11471451638158715,
+      "learning_rate": 1.648067260501611e-05,
+      "loss": 0.0008,
       "step": 558
     },
     {
-      "epoch": 0.8368263473053892,
-      "grad_norm": 0.13994232783724012,
-      "learning_rate": 1.6941283325884458e-05,
-      "loss": 0.0019,
+      "epoch": 0.8944,
+      "grad_norm": 0.24056809953548722,
+      "learning_rate": 1.6467502407993995e-05,
+      "loss": 0.0023,
       "step": 559
     },
     {
-      "epoch": 0.8383233532934131,
-      "grad_norm": 0.20016507933450822,
-      "learning_rate": 1.6929635145608098e-05,
-      "loss": 0.0031,
+      "epoch": 0.896,
+      "grad_norm": 0.005796607367392201,
+      "learning_rate": 1.6454312898027992e-05,
+      "loss": 0.0001,
       "step": 560
     },
     {
-      "epoch": 0.8398203592814372,
-      "grad_norm": 0.11790210879503302,
-      "learning_rate": 1.691796884924386e-05,
-      "loss": 0.0017,
+      "epoch": 0.8976,
+      "grad_norm": 0.07955779624413827,
+      "learning_rate": 1.644110411450398e-05,
+      "loss": 0.0024,
       "step": 561
     },
     {
-      "epoch": 0.8413173652694611,
-      "grad_norm": 0.060596148104514636,
-      "learning_rate": 1.6906284467290852e-05,
-      "loss": 0.0009,
+      "epoch": 0.8992,
+      "grad_norm": 0.0066621476684513466,
+      "learning_rate": 1.6427876096865394e-05,
+      "loss": 0.0001,
       "step": 562
     },
     {
-      "epoch": 0.842814371257485,
-      "grad_norm": 0.11618444614587378,
-      "learning_rate": 1.6894582030295458e-05,
-      "loss": 0.0018,
+      "epoch": 0.9008,
+      "grad_norm": 0.01414635288312015,
+      "learning_rate": 1.6414628884613106e-05,
+      "loss": 0.0002,
       "step": 563
     },
     {
-      "epoch": 0.844311377245509,
-      "grad_norm": 0.16389923156347438,
-      "learning_rate": 1.6882861568851258e-05,
-      "loss": 0.0025,
+      "epoch": 0.9024,
+      "grad_norm": 0.05874496851316498,
+      "learning_rate": 1.6401362517305296e-05,
+      "loss": 0.0003,
       "step": 564
     },
     {
-      "epoch": 0.8458083832335329,
-      "grad_norm": 0.2998453874019442,
-      "learning_rate": 1.6871123113598954e-05,
-      "loss": 0.0029,
+      "epoch": 0.904,
+      "grad_norm": 0.06435940379851529,
+      "learning_rate": 1.6388077034557355e-05,
+      "loss": 0.0004,
       "step": 565
     },
     {
-      "epoch": 0.8473053892215568,
-      "grad_norm": 0.2986815290378479,
-      "learning_rate": 1.6859366695226303e-05,
-      "loss": 0.003,
+      "epoch": 0.9056,
+      "grad_norm": 0.004431934862705762,
+      "learning_rate": 1.637477247604175e-05,
+      "loss": 0.0001,
       "step": 566
     },
     {
-      "epoch": 0.8488023952095808,
-      "grad_norm": 0.21672636750976904,
-      "learning_rate": 1.6847592344468e-05,
-      "loss": 0.0035,
+      "epoch": 0.9072,
+      "grad_norm": 0.12964315424935577,
+      "learning_rate": 1.6361448881487913e-05,
+      "loss": 0.002,
       "step": 567
     },
     {
-      "epoch": 0.8502994011976048,
-      "grad_norm": 0.10471468615303521,
-      "learning_rate": 1.6835800092105647e-05,
-      "loss": 0.0015,
+      "epoch": 0.9088,
+      "grad_norm": 0.0405461289802062,
+      "learning_rate": 1.6348106290682117e-05,
+      "loss": 0.0002,
       "step": 568
     },
     {
-      "epoch": 0.8517964071856288,
-      "grad_norm": 0.12014153904526044,
-      "learning_rate": 1.682398996896762e-05,
-      "loss": 0.0016,
+      "epoch": 0.9104,
+      "grad_norm": 0.04518253762629917,
+      "learning_rate": 1.6334744743467366e-05,
+      "loss": 0.0003,
       "step": 569
     },
     {
-      "epoch": 0.8532934131736527,
-      "grad_norm": 0.24386062518444432,
-      "learning_rate": 1.681216200592903e-05,
-      "loss": 0.0029,
+      "epoch": 0.912,
+      "grad_norm": 0.1278136950485491,
+      "learning_rate": 1.6321364279743267e-05,
+      "loss": 0.0005,
       "step": 570
     },
     {
-      "epoch": 0.8547904191616766,
-      "grad_norm": 0.4891409878621194,
-      "learning_rate": 1.6800316233911627e-05,
-      "loss": 0.003,
+      "epoch": 0.9136,
+      "grad_norm": 0.026243421967412844,
+      "learning_rate": 1.6307964939465914e-05,
+      "loss": 0.0003,
       "step": 571
     },
     {
-      "epoch": 0.8562874251497006,
-      "grad_norm": 0.11158025115959051,
-      "learning_rate": 1.678845268388371e-05,
-      "loss": 0.0013,
+      "epoch": 0.9152,
+      "grad_norm": 0.005911777918320239,
+      "learning_rate": 1.6294546762647775e-05,
+      "loss": 0.0001,
       "step": 572
     },
     {
-      "epoch": 0.8577844311377245,
-      "grad_norm": 0.16176832999905424,
-      "learning_rate": 1.677657138686006e-05,
-      "loss": 0.0029,
+      "epoch": 0.9168,
+      "grad_norm": 0.016433329334518097,
+      "learning_rate": 1.628110978935756e-05,
+      "loss": 0.0002,
       "step": 573
     },
     {
-      "epoch": 0.8592814371257484,
-      "grad_norm": 0.2862714242004369,
-      "learning_rate": 1.6764672373901857e-05,
-      "loss": 0.0044,
+      "epoch": 0.9184,
+      "grad_norm": 0.024671596956701715,
+      "learning_rate": 1.626765405972011e-05,
+      "loss": 0.0002,
       "step": 574
     },
     {
-      "epoch": 0.8607784431137725,
-      "grad_norm": 0.23371137768377162,
-      "learning_rate": 1.6752755676116586e-05,
-      "loss": 0.0029,
+      "epoch": 0.92,
+      "grad_norm": 0.15693017537160817,
+      "learning_rate": 1.625417961391628e-05,
+      "loss": 0.0022,
       "step": 575
     },
     {
-      "epoch": 0.8622754491017964,
-      "grad_norm": 0.16513171198212837,
-      "learning_rate": 1.6740821324657983e-05,
-      "loss": 0.0028,
+      "epoch": 0.9216,
+      "grad_norm": 0.1821290033572893,
+      "learning_rate": 1.6240686492182806e-05,
+      "loss": 0.0022,
       "step": 576
     },
     {
-      "epoch": 0.8637724550898204,
-      "grad_norm": 0.0989955200636401,
-      "learning_rate": 1.672886935072591e-05,
-      "loss": 0.001,
+      "epoch": 0.9232,
+      "grad_norm": 0.019637133645766462,
+      "learning_rate": 1.62271747348122e-05,
+      "loss": 0.0002,
       "step": 577
     },
     {
-      "epoch": 0.8652694610778443,
-      "grad_norm": 0.3256050916149061,
-      "learning_rate": 1.6716899785566324e-05,
-      "loss": 0.0041,
+      "epoch": 0.9248,
+      "grad_norm": 0.032264525390541345,
+      "learning_rate": 1.621364438215262e-05,
+      "loss": 0.0002,
       "step": 578
     },
     {
-      "epoch": 0.8667664670658682,
-      "grad_norm": 0.2090094852879972,
-      "learning_rate": 1.670491266047116e-05,
-      "loss": 0.0032,
+      "epoch": 0.9264,
+      "grad_norm": 0.17257532678963658,
+      "learning_rate": 1.6200095474607753e-05,
+      "loss": 0.0021,
       "step": 579
     },
     {
-      "epoch": 0.8682634730538922,
-      "grad_norm": 0.15452654886105557,
-      "learning_rate": 1.669290800677825e-05,
-      "loss": 0.0024,
+      "epoch": 0.928,
+      "grad_norm": 0.28931998966513617,
+      "learning_rate": 1.6186528052636692e-05,
+      "loss": 0.0006,
       "step": 580
     },
     {
-      "epoch": 0.8697604790419161,
-      "grad_norm": 0.12984740923736884,
-      "learning_rate": 1.668088585587127e-05,
-      "loss": 0.0021,
+      "epoch": 0.9296,
+      "grad_norm": 0.005154357147886349,
+      "learning_rate": 1.6172942156753822e-05,
+      "loss": 0.0001,
       "step": 581
     },
     {
-      "epoch": 0.8712574850299402,
-      "grad_norm": 0.20135825659725679,
-      "learning_rate": 1.666884623917963e-05,
-      "loss": 0.0025,
+      "epoch": 0.9312,
+      "grad_norm": 0.11689177832511896,
+      "learning_rate": 1.6159337827528686e-05,
+      "loss": 0.0023,
       "step": 582
     },
     {
-      "epoch": 0.8727544910179641,
-      "grad_norm": 0.2843598279050385,
-      "learning_rate": 1.6656789188178394e-05,
-      "loss": 0.0048,
+      "epoch": 0.9328,
+      "grad_norm": 0.01065347283461778,
+      "learning_rate": 1.614571510558588e-05,
+      "loss": 0.0002,
       "step": 583
     },
     {
-      "epoch": 0.874251497005988,
-      "grad_norm": 0.18200615608698034,
-      "learning_rate": 1.664471473438822e-05,
-      "loss": 0.0032,
+      "epoch": 0.9344,
+      "grad_norm": 0.026100680293083087,
+      "learning_rate": 1.6132074031604917e-05,
+      "loss": 0.0002,
       "step": 584
     },
     {
-      "epoch": 0.875748502994012,
-      "grad_norm": 0.10309687291550676,
-      "learning_rate": 1.6632622909375244e-05,
-      "loss": 0.0014,
+      "epoch": 0.936,
+      "grad_norm": 0.0512772513165749,
+      "learning_rate": 1.6118414646320115e-05,
+      "loss": 0.0005,
       "step": 585
     },
     {
-      "epoch": 0.8772455089820359,
-      "grad_norm": 0.06697443098407523,
-      "learning_rate": 1.662051374475103e-05,
-      "loss": 0.0008,
+      "epoch": 0.9376,
+      "grad_norm": 0.013347670491282037,
+      "learning_rate": 1.6104736990520468e-05,
+      "loss": 0.0002,
       "step": 586
     },
     {
-      "epoch": 0.8787425149700598,
-      "grad_norm": 0.20351165773382962,
-      "learning_rate": 1.660838727217246e-05,
-      "loss": 0.0025,
+      "epoch": 0.9392,
+      "grad_norm": 0.0794315824612763,
+      "learning_rate": 1.6091041105049542e-05,
+      "loss": 0.0005,
       "step": 587
     },
     {
-      "epoch": 0.8802395209580839,
-      "grad_norm": 0.19444272969584192,
-      "learning_rate": 1.6596243523341678e-05,
-      "loss": 0.0048,
+      "epoch": 0.9408,
+      "grad_norm": 0.0879977258359348,
+      "learning_rate": 1.6077327030805318e-05,
+      "loss": 0.001,
       "step": 588
     },
     {
-      "epoch": 0.8817365269461078,
-      "grad_norm": 0.1718287018179276,
-      "learning_rate": 1.658408253000598e-05,
-      "loss": 0.0036,
+      "epoch": 0.9424,
+      "grad_norm": 0.16486274087034078,
+      "learning_rate": 1.6063594808740112e-05,
+      "loss": 0.0025,
       "step": 589
     },
     {
-      "epoch": 0.8832335329341318,
-      "grad_norm": 0.1473170594956288,
-      "learning_rate": 1.657190432395776e-05,
-      "loss": 0.0033,
+      "epoch": 0.944,
+      "grad_norm": 0.3474795772948319,
+      "learning_rate": 1.604984447986042e-05,
+      "loss": 0.0011,
       "step": 590
     },
     {
-      "epoch": 0.8847305389221557,
-      "grad_norm": 0.05320979757128209,
-      "learning_rate": 1.6559708937034394e-05,
-      "loss": 0.0009,
+      "epoch": 0.9456,
+      "grad_norm": 0.013463877091480161,
+      "learning_rate": 1.6036076085226813e-05,
+      "loss": 0.0002,
       "step": 591
     },
     {
-      "epoch": 0.8862275449101796,
-      "grad_norm": 0.1948054351678213,
-      "learning_rate": 1.654749640111818e-05,
-      "loss": 0.0027,
+      "epoch": 0.9472,
+      "grad_norm": 0.10577053585975192,
+      "learning_rate": 1.602228966595381e-05,
+      "loss": 0.0011,
       "step": 592
     },
     {
-      "epoch": 0.8877245508982036,
-      "grad_norm": 0.1342630451535342,
-      "learning_rate": 1.6535266748136255e-05,
-      "loss": 0.0021,
+      "epoch": 0.9488,
+      "grad_norm": 0.11085121896124762,
+      "learning_rate": 1.6008485263209742e-05,
+      "loss": 0.0006,
       "step": 593
     },
     {
-      "epoch": 0.8892215568862275,
-      "grad_norm": 0.10784490714170987,
-      "learning_rate": 1.6523020010060497e-05,
-      "loss": 0.0027,
+      "epoch": 0.9504,
+      "grad_norm": 0.07342027964029993,
+      "learning_rate": 1.599466291821666e-05,
+      "loss": 0.001,
       "step": 594
     },
     {
-      "epoch": 0.8907185628742516,
-      "grad_norm": 0.1546876958105241,
-      "learning_rate": 1.6510756218907453e-05,
-      "loss": 0.0022,
+      "epoch": 0.952,
+      "grad_norm": 0.01696710108604813,
+      "learning_rate": 1.598082267225018e-05,
+      "loss": 0.0002,
       "step": 595
     },
     {
-      "epoch": 0.8922155688622755,
-      "grad_norm": 0.16370527891824135,
-      "learning_rate": 1.6498475406738248e-05,
-      "loss": 0.0023,
+      "epoch": 0.9536,
+      "grad_norm": 0.1328848387639668,
+      "learning_rate": 1.596696456663938e-05,
+      "loss": 0.0016,
       "step": 596
     },
     {
-      "epoch": 0.8937125748502994,
-      "grad_norm": 0.09977139023857336,
-      "learning_rate": 1.648617760565851e-05,
-      "loss": 0.0009,
+      "epoch": 0.9552,
+      "grad_norm": 0.012110705106772179,
+      "learning_rate": 1.595308864276666e-05,
+      "loss": 0.0002,
       "step": 597
     },
     {
-      "epoch": 0.8952095808383234,
-      "grad_norm": 0.14481161212272883,
-      "learning_rate": 1.647386284781828e-05,
-      "loss": 0.0026,
+      "epoch": 0.9568,
+      "grad_norm": 0.017146486661923956,
+      "learning_rate": 1.5939194942067647e-05,
+      "loss": 0.0002,
       "step": 598
     },
     {
-      "epoch": 0.8967065868263473,
-      "grad_norm": 0.2579988975748116,
-      "learning_rate": 1.6461531165411924e-05,
-      "loss": 0.0039,
+      "epoch": 0.9584,
+      "grad_norm": 0.027149900419092123,
+      "learning_rate": 1.592528350603103e-05,
+      "loss": 0.0003,
       "step": 599
     },
     {
-      "epoch": 0.8982035928143712,
-      "grad_norm": 0.11325757290508497,
-      "learning_rate": 1.644918259067806e-05,
-      "loss": 0.0015,
+      "epoch": 0.96,
+      "grad_norm": 0.0346024532568189,
+      "learning_rate": 1.5911354376198468e-05,
+      "loss": 0.0003,
       "step": 600
     },
     {
-      "epoch": 0.8997005988023952,
-      "grad_norm": 0.10693055025728564,
-      "learning_rate": 1.643681715589946e-05,
-      "loss": 0.0018,
+      "epoch": 0.9616,
+      "grad_norm": 0.11389671831412627,
+      "learning_rate": 1.5897407594164468e-05,
+      "loss": 0.0005,
       "step": 601
     },
     {
-      "epoch": 0.9011976047904192,
-      "grad_norm": 0.10387304790724435,
-      "learning_rate": 1.6424434893402988e-05,
-      "loss": 0.0016,
+      "epoch": 0.9632,
+      "grad_norm": 0.03419471934683977,
+      "learning_rate": 1.5883443201576225e-05,
+      "loss": 0.0003,
       "step": 602
     },
     {
-      "epoch": 0.9026946107784432,
-      "grad_norm": 0.08953073156215048,
-      "learning_rate": 1.6412035835559483e-05,
-      "loss": 0.0008,
+      "epoch": 0.9648,
+      "grad_norm": 0.16922390481395708,
+      "learning_rate": 1.586946124013354e-05,
+      "loss": 0.0004,
       "step": 603
     },
     {
-      "epoch": 0.9041916167664671,
-      "grad_norm": 0.1882854229415982,
-      "learning_rate": 1.6399620014783702e-05,
-      "loss": 0.0035,
+      "epoch": 0.9664,
+      "grad_norm": 0.020426447409353753,
+      "learning_rate": 1.585546175158868e-05,
+      "loss": 0.0002,
       "step": 604
     },
     {
-      "epoch": 0.905688622754491,
-      "grad_norm": 0.05376462878864106,
-      "learning_rate": 1.6387187463534222e-05,
-      "loss": 0.0007,
+      "epoch": 0.968,
+      "grad_norm": 0.17459777602170934,
+      "learning_rate": 1.5841444777746232e-05,
+      "loss": 0.0021,
       "step": 605
     },
     {
-      "epoch": 0.907185628742515,
-      "grad_norm": 0.2652074145829586,
-      "learning_rate": 1.6374738214313354e-05,
-      "loss": 0.0035,
+      "epoch": 0.9696,
+      "grad_norm": 0.014776040366925668,
+      "learning_rate": 1.582741036046301e-05,
+      "loss": 0.0002,
       "step": 606
     },
     {
-      "epoch": 0.9086826347305389,
-      "grad_norm": 0.11598887341928915,
-      "learning_rate": 1.6362272299667078e-05,
-      "loss": 0.001,
+      "epoch": 0.9712,
+      "grad_norm": 0.07247446983177594,
+      "learning_rate": 1.5813358541647915e-05,
+      "loss": 0.0004,
       "step": 607
     },
     {
-      "epoch": 0.9101796407185628,
-      "grad_norm": 0.0870013457319502,
-      "learning_rate": 1.6349789752184922e-05,
-      "loss": 0.0013,
+      "epoch": 0.9728,
+      "grad_norm": 0.02386915552711786,
+      "learning_rate": 1.5799289363261815e-05,
+      "loss": 0.0002,
       "step": 608
     },
     {
-      "epoch": 0.9116766467065869,
-      "grad_norm": 0.17343366028225063,
-      "learning_rate": 1.6337290604499913e-05,
-      "loss": 0.0025,
+      "epoch": 0.9744,
+      "grad_norm": 0.2050066873881811,
+      "learning_rate": 1.578520286731741e-05,
+      "loss": 0.0024,
       "step": 609
     },
     {
-      "epoch": 0.9131736526946108,
-      "grad_norm": 0.0859523102698752,
-      "learning_rate": 1.6324774889288462e-05,
-      "loss": 0.0018,
+      "epoch": 0.976,
+      "grad_norm": 0.09157781403934438,
+      "learning_rate": 1.5771099095879108e-05,
+      "loss": 0.0004,
       "step": 610
     },
     {
-      "epoch": 0.9146706586826348,
-      "grad_norm": 0.07712475151325601,
-      "learning_rate": 1.6312242639270308e-05,
-      "loss": 0.0011,
+      "epoch": 0.9776,
+      "grad_norm": 0.003624846129716664,
+      "learning_rate": 1.575697809106292e-05,
+      "loss": 0.0001,
       "step": 611
     },
     {
-      "epoch": 0.9161676646706587,
-      "grad_norm": 0.07162451662462921,
-      "learning_rate": 1.6299693887208404e-05,
-      "loss": 0.0031,
+      "epoch": 0.9792,
+      "grad_norm": 0.010323437128790412,
+      "learning_rate": 1.5742839895036305e-05,
+      "loss": 0.0002,
       "step": 612
     },
     {
-      "epoch": 0.9176646706586826,
-      "grad_norm": 0.1607194705076756,
-      "learning_rate": 1.628712866590885e-05,
-      "loss": 0.0021,
+      "epoch": 0.9808,
+      "grad_norm": 0.017110117169375917,
+      "learning_rate": 1.5728684550018066e-05,
+      "loss": 0.0002,
       "step": 613
     },
     {
-      "epoch": 0.9191616766467066,
-      "grad_norm": 0.08490453147422353,
-      "learning_rate": 1.6274547008220804e-05,
-      "loss": 0.0015,
+      "epoch": 0.9824,
+      "grad_norm": 0.4304006186302777,
+      "learning_rate": 1.571451209827821e-05,
+      "loss": 0.0047,
       "step": 614
     },
     {
-      "epoch": 0.9206586826347305,
-      "grad_norm": 0.1508869109445389,
-      "learning_rate": 1.626194894703638e-05,
-      "loss": 0.0024,
+      "epoch": 0.984,
+      "grad_norm": 0.013547103462652449,
+      "learning_rate": 1.570032258213783e-05,
+      "loss": 0.0001,
       "step": 615
     },
     {
-      "epoch": 0.9221556886227545,
-      "grad_norm": 0.08485546738561354,
-      "learning_rate": 1.6249334515290593e-05,
-      "loss": 0.0013,
+      "epoch": 0.9856,
+      "grad_norm": 0.060641420331837864,
+      "learning_rate": 1.5686116043968975e-05,
+      "loss": 0.0003,
       "step": 616
     },
     {
-      "epoch": 0.9236526946107785,
-      "grad_norm": 0.16110417372420824,
-      "learning_rate": 1.6236703745961244e-05,
-      "loss": 0.0022,
+      "epoch": 0.9872,
+      "grad_norm": 0.01713297463284221,
+      "learning_rate": 1.5671892526194515e-05,
+      "loss": 0.0002,
       "step": 617
     },
     {
-      "epoch": 0.9251497005988024,
-      "grad_norm": 0.06607135428636862,
-      "learning_rate": 1.622405667206885e-05,
-      "loss": 0.0007,
+      "epoch": 0.9888,
+      "grad_norm": 0.29587661098059337,
+      "learning_rate": 1.565765207128805e-05,
+      "loss": 0.0015,
       "step": 618
     },
     {
-      "epoch": 0.9266467065868264,
-      "grad_norm": 0.2938746182376294,
-      "learning_rate": 1.6211393326676558e-05,
-      "loss": 0.0031,
+      "epoch": 0.9904,
+      "grad_norm": 0.18927537268665268,
+      "learning_rate": 1.564339472177373e-05,
+      "loss": 0.0032,
       "step": 619
     },
     {
-      "epoch": 0.9281437125748503,
-      "grad_norm": 0.14606146095893197,
-      "learning_rate": 1.6198713742890036e-05,
-      "loss": 0.0029,
+      "epoch": 0.992,
+      "grad_norm": 0.08369681391942292,
+      "learning_rate": 1.5629120520226163e-05,
+      "loss": 0.0009,
       "step": 620
     },
     {
-      "epoch": 0.9296407185628742,
-      "grad_norm": 0.05652377223402481,
-      "learning_rate": 1.6186017953857423e-05,
-      "loss": 0.0009,
+      "epoch": 0.9936,
+      "grad_norm": 0.01595611295946394,
+      "learning_rate": 1.561482950927029e-05,
+      "loss": 0.0002,
       "step": 621
     },
     {
-      "epoch": 0.9311377245508982,
-      "grad_norm": 0.18019462860068167,
-      "learning_rate": 1.617330599276921e-05,
-      "loss": 0.0043,
+      "epoch": 0.9952,
+      "grad_norm": 0.03823447390409338,
+      "learning_rate": 1.560052173158123e-05,
+      "loss": 0.0002,
       "step": 622
     },
     {
-      "epoch": 0.9326347305389222,
-      "grad_norm": 0.0804731431661998,
-      "learning_rate": 1.6160577892858178e-05,
-      "loss": 0.0012,
+      "epoch": 0.9968,
+      "grad_norm": 0.16298578879060066,
+      "learning_rate": 1.5586197229884185e-05,
+      "loss": 0.002,
       "step": 623
     },
     {
-      "epoch": 0.9341317365269461,
-      "grad_norm": 0.23044713115642781,
-      "learning_rate": 1.6147833687399287e-05,
-      "loss": 0.0026,
+      "epoch": 0.9984,
+      "grad_norm": 0.0541019946743606,
+      "learning_rate": 1.5571856046954284e-05,
+      "loss": 0.0005,
       "step": 624
     },
     {
-      "epoch": 0.9356287425149701,
-      "grad_norm": 0.15040761212205267,
-      "learning_rate": 1.6135073409709613e-05,
-      "loss": 0.0033,
+      "epoch": 1.0,
+      "grad_norm": 0.02004341106821655,
+      "learning_rate": 1.5557498225616488e-05,
+      "loss": 0.0002,
       "step": 625
     },
     {
-      "epoch": 0.937125748502994,
-      "grad_norm": 0.17617733364441931,
-      "learning_rate": 1.6122297093148242e-05,
-      "loss": 0.0021,
+      "epoch": 1.0016,
+      "grad_norm": 0.10660463958703306,
+      "learning_rate": 1.5543123808745418e-05,
+      "loss": 0.0011,
       "step": 626
     },
     {
-      "epoch": 0.938622754491018,
-      "grad_norm": 0.2182223824425347,
-      "learning_rate": 1.6109504771116193e-05,
-      "loss": 0.0014,
+      "epoch": 1.0032,
+      "grad_norm": 0.05934520626247313,
+      "learning_rate": 1.5528732839265272e-05,
+      "loss": 0.0005,
       "step": 627
     },
     {
-      "epoch": 0.9401197604790419,
-      "grad_norm": 0.20104637613306378,
-      "learning_rate": 1.6096696477056325e-05,
-      "loss": 0.0025,
+      "epoch": 1.0048,
+      "grad_norm": 0.21529588391063953,
+      "learning_rate": 1.5514325360149668e-05,
+      "loss": 0.003,
       "step": 628
     },
     {
-      "epoch": 0.9416167664670658,
-      "grad_norm": 0.10319298502135164,
-      "learning_rate": 1.608387224445326e-05,
-      "loss": 0.0016,
+      "epoch": 1.0064,
+      "grad_norm": 0.0260550732980306,
+      "learning_rate": 1.549990141442153e-05,
+      "loss": 0.0002,
       "step": 629
     },
     {
-      "epoch": 0.9431137724550899,
-      "grad_norm": 0.1744369089861333,
-      "learning_rate": 1.607103210683328e-05,
-      "loss": 0.0019,
+      "epoch": 1.008,
+      "grad_norm": 0.08169285818621008,
+      "learning_rate": 1.5485461045152937e-05,
+      "loss": 0.0011,
       "step": 630
     },
     {
-      "epoch": 0.9446107784431138,
-      "grad_norm": 0.1344166064567352,
-      "learning_rate": 1.6058176097764254e-05,
-      "loss": 0.0013,
+      "epoch": 1.0096,
+      "grad_norm": 0.06427658111081822,
+      "learning_rate": 1.5471004295465034e-05,
+      "loss": 0.0005,
       "step": 631
     },
     {
-      "epoch": 0.9461077844311377,
-      "grad_norm": 0.08853174252976158,
-      "learning_rate": 1.6045304250855534e-05,
-      "loss": 0.0013,
+      "epoch": 1.0112,
+      "grad_norm": 0.04754833786155076,
+      "learning_rate": 1.5456531208527868e-05,
+      "loss": 0.0005,
       "step": 632
     },
     {
-      "epoch": 0.9476047904191617,
-      "grad_norm": 0.08846484292192285,
-      "learning_rate": 1.6032416599757894e-05,
-      "loss": 0.002,
+      "epoch": 1.0128,
+      "grad_norm": 0.028872058901928507,
+      "learning_rate": 1.5442041827560274e-05,
+      "loss": 0.0003,
       "step": 633
     },
     {
-      "epoch": 0.9491017964071856,
-      "grad_norm": 0.12884476368073972,
-      "learning_rate": 1.6019513178163403e-05,
-      "loss": 0.0011,
+      "epoch": 1.0144,
+      "grad_norm": 0.04961506841238305,
+      "learning_rate": 1.542753619582974e-05,
+      "loss": 0.0004,
       "step": 634
     },
     {
-      "epoch": 0.9505988023952096,
-      "grad_norm": 0.15795219235072977,
-      "learning_rate": 1.6006594019805376e-05,
-      "loss": 0.0011,
+      "epoch": 1.016,
+      "grad_norm": 0.023923799642960132,
+      "learning_rate": 1.5413014356652287e-05,
+      "loss": 0.0001,
       "step": 635
     },
     {
-      "epoch": 0.9520958083832335,
-      "grad_norm": 0.13187039248116258,
-      "learning_rate": 1.5993659158458255e-05,
-      "loss": 0.0024,
+      "epoch": 1.0176,
+      "grad_norm": 0.12805240672927612,
+      "learning_rate": 1.5398476353392323e-05,
+      "loss": 0.0005,
       "step": 636
     },
     {
-      "epoch": 0.9535928143712575,
-      "grad_norm": 0.15925859946439425,
-      "learning_rate": 1.5980708627937548e-05,
-      "loss": 0.0025,
+      "epoch": 1.0192,
+      "grad_norm": 0.1158424283730911,
+      "learning_rate": 1.538392222946255e-05,
+      "loss": 0.0007,
       "step": 637
     },
     {
-      "epoch": 0.9550898203592815,
-      "grad_norm": 0.13331313394073413,
-      "learning_rate": 1.5967742462099717e-05,
-      "loss": 0.0017,
+      "epoch": 1.0208,
+      "grad_norm": 0.058814949979394474,
+      "learning_rate": 1.5369352028323773e-05,
+      "loss": 0.0003,
       "step": 638
     },
     {
-      "epoch": 0.9565868263473054,
-      "grad_norm": 0.08975182166363489,
-      "learning_rate": 1.59547606948421e-05,
-      "loss": 0.0014,
+      "epoch": 1.0224,
+      "grad_norm": 0.02345027815015694,
+      "learning_rate": 1.5354765793484834e-05,
+      "loss": 0.0002,
       "step": 639
     },
     {
-      "epoch": 0.9580838323353293,
-      "grad_norm": 0.3449291274347138,
-      "learning_rate": 1.594176336010283e-05,
-      "loss": 0.0024,
+      "epoch": 1.024,
+      "grad_norm": 0.03851324265626075,
+      "learning_rate": 1.534016356850244e-05,
+      "loss": 0.0003,
       "step": 640
     },
     {
-      "epoch": 0.9595808383233533,
-      "grad_norm": 0.06409320220899797,
-      "learning_rate": 1.592875049186073e-05,
-      "loss": 0.0011,
+      "epoch": 1.0256,
+      "grad_norm": 0.007581656257266726,
+      "learning_rate": 1.5325545396981053e-05,
+      "loss": 0.0001,
       "step": 641
     },
     {
-      "epoch": 0.9610778443113772,
-      "grad_norm": 0.17833684537580516,
-      "learning_rate": 1.5915722124135227e-05,
-      "loss": 0.002,
+      "epoch": 1.0272,
+      "grad_norm": 0.18789016214748264,
+      "learning_rate": 1.531091132257275e-05,
+      "loss": 0.0012,
       "step": 642
     },
     {
-      "epoch": 0.9625748502994012,
-      "grad_norm": 0.15331420541889684,
-      "learning_rate": 1.590267829098628e-05,
-      "loss": 0.002,
+      "epoch": 1.0288,
+      "grad_norm": 0.017634275364985862,
+      "learning_rate": 1.5296261388977107e-05,
+      "loss": 0.0002,
       "step": 643
     },
     {
-      "epoch": 0.9640718562874252,
-      "grad_norm": 0.5899661957634779,
-      "learning_rate": 1.5889619026514272e-05,
-      "loss": 0.0043,
+      "epoch": 1.0304,
+      "grad_norm": 0.005853495436668061,
+      "learning_rate": 1.528159563994104e-05,
+      "loss": 0.0001,
       "step": 644
     },
     {
-      "epoch": 0.9655688622754491,
-      "grad_norm": 0.1657525859558746,
-      "learning_rate": 1.5876544364859937e-05,
-      "loss": 0.0022,
+      "epoch": 1.032,
+      "grad_norm": 0.031525289383469456,
+      "learning_rate": 1.52669141192587e-05,
+      "loss": 0.0002,
       "step": 645
     },
     {
-      "epoch": 0.9670658682634731,
-      "grad_norm": 0.18379072480540792,
-      "learning_rate": 1.5863454340204247e-05,
-      "loss": 0.003,
+      "epoch": 1.0336,
+      "grad_norm": 0.006841301694317977,
+      "learning_rate": 1.5252216870771345e-05,
+      "loss": 0.0001,
       "step": 646
     },
     {
-      "epoch": 0.968562874251497,
-      "grad_norm": 0.12209910222944766,
-      "learning_rate": 1.585034898676835e-05,
-      "loss": 0.0015,
+      "epoch": 1.0352,
+      "grad_norm": 0.0587557732396357,
+      "learning_rate": 1.5237503938367186e-05,
+      "loss": 0.0003,
       "step": 647
     },
     {
-      "epoch": 0.9700598802395209,
-      "grad_norm": 0.10607555085781809,
-      "learning_rate": 1.583722833881346e-05,
-      "loss": 0.0019,
+      "epoch": 1.0368,
+      "grad_norm": 0.01676862267717498,
+      "learning_rate": 1.5222775365981272e-05,
+      "loss": 0.0002,
       "step": 648
     },
     {
-      "epoch": 0.9715568862275449,
-      "grad_norm": 0.26474190068427933,
-      "learning_rate": 1.5824092430640784e-05,
-      "loss": 0.0038,
+      "epoch": 1.0384,
+      "grad_norm": 0.013717804585198331,
+      "learning_rate": 1.5208031197595357e-05,
+      "loss": 0.0002,
       "step": 649
     },
     {
-      "epoch": 0.9730538922155688,
-      "grad_norm": 0.4156799624786208,
-      "learning_rate": 1.5810941296591406e-05,
-      "loss": 0.0028,
+      "epoch": 1.04,
+      "grad_norm": 0.042297522196741995,
+      "learning_rate": 1.5193271477237761e-05,
+      "loss": 0.0002,
       "step": 650
     },
     {
-      "epoch": 0.9745508982035929,
-      "grad_norm": 0.353881703591171,
-      "learning_rate": 1.5797774971046243e-05,
-      "loss": 0.0041,
+      "epoch": 1.0416,
+      "grad_norm": 0.005830534636442314,
+      "learning_rate": 1.5178496248983254e-05,
+      "loss": 0.0001,
       "step": 651
     },
     {
-      "epoch": 0.9760479041916168,
-      "grad_norm": 0.47331385633169554,
-      "learning_rate": 1.5784593488425898e-05,
-      "loss": 0.0036,
+      "epoch": 1.0432,
+      "grad_norm": 0.02023393751390376,
+      "learning_rate": 1.5163705556952912e-05,
+      "loss": 0.0002,
       "step": 652
     },
     {
-      "epoch": 0.9775449101796407,
-      "grad_norm": 0.23996897207608175,
-      "learning_rate": 1.577139688319062e-05,
-      "loss": 0.0042,
+      "epoch": 1.0448,
+      "grad_norm": 0.004917776209081582,
+      "learning_rate": 1.5148899445313983e-05,
+      "loss": 0.0001,
       "step": 653
     },
     {
-      "epoch": 0.9790419161676647,
-      "grad_norm": 0.14040120360537384,
-      "learning_rate": 1.5758185189840182e-05,
-      "loss": 0.0015,
+      "epoch": 1.0464,
+      "grad_norm": 0.011971086698735709,
+      "learning_rate": 1.5134077958279764e-05,
+      "loss": 0.0002,
       "step": 654
     },
     {
-      "epoch": 0.9805389221556886,
-      "grad_norm": 0.16782002665806564,
-      "learning_rate": 1.5744958442913804e-05,
-      "loss": 0.0033,
+      "epoch": 1.048,
+      "grad_norm": 0.01749647351425708,
+      "learning_rate": 1.5119241140109466e-05,
+      "loss": 0.0002,
       "step": 655
     },
     {
-      "epoch": 0.9820359281437125,
-      "grad_norm": 0.3548917708393882,
-      "learning_rate": 1.5731716676990057e-05,
-      "loss": 0.0037,
+      "epoch": 1.0496,
+      "grad_norm": 0.08810717240182654,
+      "learning_rate": 1.5104389035108078e-05,
+      "loss": 0.001,
       "step": 656
     },
     {
-      "epoch": 0.9835329341317365,
-      "grad_norm": 0.19405634340063596,
-      "learning_rate": 1.571845992668679e-05,
-      "loss": 0.0024,
+      "epoch": 1.0512,
+      "grad_norm": 0.026469620934283282,
+      "learning_rate": 1.5089521687626243e-05,
+      "loss": 0.0002,
       "step": 657
     },
     {
-      "epoch": 0.9850299401197605,
-      "grad_norm": 0.23475543702616705,
-      "learning_rate": 1.5705188226661013e-05,
-      "loss": 0.0045,
+      "epoch": 1.0528,
+      "grad_norm": 0.051846745706488596,
+      "learning_rate": 1.5074639142060119e-05,
+      "loss": 0.0003,
       "step": 658
     },
     {
-      "epoch": 0.9865269461077845,
-      "grad_norm": 0.36667176478027685,
-      "learning_rate": 1.569190161160882e-05,
-      "loss": 0.0039,
+      "epoch": 1.0544,
+      "grad_norm": 0.059650765192615585,
+      "learning_rate": 1.505974144285124e-05,
+      "loss": 0.0004,
       "step": 659
     },
     {
-      "epoch": 0.9880239520958084,
-      "grad_norm": 0.1896831017088073,
-      "learning_rate": 1.5678600116265293e-05,
-      "loss": 0.0027,
+      "epoch": 1.056,
+      "grad_norm": 0.11858662320135964,
+      "learning_rate": 1.50448286344864e-05,
+      "loss": 0.0005,
       "step": 660
     },
     {
-      "epoch": 0.9895209580838323,
-      "grad_norm": 0.10818436139444451,
-      "learning_rate": 1.566528377540443e-05,
-      "loss": 0.0021,
+      "epoch": 1.0576,
+      "grad_norm": 0.009351544749928867,
+      "learning_rate": 1.5029900761497507e-05,
+      "loss": 0.0001,
       "step": 661
     },
     {
-      "epoch": 0.9910179640718563,
-      "grad_norm": 0.1635417574332313,
-      "learning_rate": 1.5651952623839028e-05,
-      "loss": 0.0009,
+      "epoch": 1.0592,
+      "grad_norm": 0.1343413427308062,
+      "learning_rate": 1.501495786846146e-05,
+      "loss": 0.0011,
       "step": 662
     },
     {
-      "epoch": 0.9925149700598802,
-      "grad_norm": 0.2787481879240596,
-      "learning_rate": 1.5638606696420602e-05,
-      "loss": 0.0034,
+      "epoch": 1.0608,
+      "grad_norm": 0.08194886239543721,
+      "learning_rate": 1.5000000000000002e-05,
+      "loss": 0.0029,
       "step": 663
     },
     {
-      "epoch": 0.9940119760479041,
-      "grad_norm": 0.12776549267504195,
-      "learning_rate": 1.5625246028039308e-05,
-      "loss": 0.0013,
+      "epoch": 1.0624,
+      "grad_norm": 0.11053103566195514,
+      "learning_rate": 1.4985027200779599e-05,
+      "loss": 0.0014,
       "step": 664
     },
     {
-      "epoch": 0.9955089820359282,
-      "grad_norm": 0.19106271050333715,
-      "learning_rate": 1.5611870653623826e-05,
-      "loss": 0.0026,
+      "epoch": 1.064,
+      "grad_norm": 0.005638188966523121,
+      "learning_rate": 1.4970039515511303e-05,
+      "loss": 0.0001,
       "step": 665
     },
     {
-      "epoch": 0.9970059880239521,
-      "grad_norm": 0.1120405670371837,
-      "learning_rate": 1.5598480608141286e-05,
-      "loss": 0.0013,
+      "epoch": 1.0656,
+      "grad_norm": 0.050703501379441136,
+      "learning_rate": 1.4955036988950617e-05,
+      "loss": 0.0003,
       "step": 666
     },
     {
-      "epoch": 0.9985029940119761,
-      "grad_norm": 0.06889850891061895,
-      "learning_rate": 1.5585075926597174e-05,
-      "loss": 0.0008,
+      "epoch": 1.0672,
+      "grad_norm": 0.09081826808070569,
+      "learning_rate": 1.4940019665897363e-05,
+      "loss": 0.003,
       "step": 667
     },
     {
-      "epoch": 1.0,
-      "grad_norm": 0.1249403668339043,
-      "learning_rate": 1.557165664403524e-05,
+      "epoch": 1.0688,
+      "grad_norm": 0.10279802127103428,
+      "learning_rate": 1.4924987591195548e-05,
       "loss": 0.0011,
       "step": 668
     },
     {
-      "epoch": 1.001497005988024,
-      "grad_norm": 0.22792093858397705,
-      "learning_rate": 1.5558222795537403e-05,
-      "loss": 0.0021,
+      "epoch": 1.0704,
+      "grad_norm": 0.05482382391695781,
+      "learning_rate": 1.4909940809733223e-05,
+      "loss": 0.0003,
       "step": 669
     },
     {
-      "epoch": 1.0029940119760479,
-      "grad_norm": 0.13959837895980517,
-      "learning_rate": 1.554477441622366e-05,
-      "loss": 0.0019,
+      "epoch": 1.072,
+      "grad_norm": 0.01978226020479878,
+      "learning_rate": 1.489487936644237e-05,
+      "loss": 0.0002,
       "step": 670
     },
     {
-      "epoch": 1.0044910179640718,
-      "grad_norm": 0.11152024170430974,
-      "learning_rate": 1.5531311541251995e-05,
-      "loss": 0.0015,
+      "epoch": 1.0735999999999999,
+      "grad_norm": 0.0507206114531236,
+      "learning_rate": 1.4879803306298736e-05,
+      "loss": 0.0003,
       "step": 671
     },
     {
-      "epoch": 1.0059880239520957,
-      "grad_norm": 0.16100791789200442,
-      "learning_rate": 1.55178342058183e-05,
-      "loss": 0.0039,
+      "epoch": 1.0752,
+      "grad_norm": 0.1033269224973426,
+      "learning_rate": 1.4864712674321733e-05,
+      "loss": 0.0004,
       "step": 672
     },
     {
-      "epoch": 1.0074850299401197,
-      "grad_norm": 0.16419979563070153,
-      "learning_rate": 1.5504342445156253e-05,
-      "loss": 0.0018,
+      "epoch": 1.0768,
+      "grad_norm": 0.03604980553918912,
+      "learning_rate": 1.4849607515574276e-05,
+      "loss": 0.0002,
       "step": 673
     },
     {
-      "epoch": 1.0089820359281436,
-      "grad_norm": 0.18649044296735573,
-      "learning_rate": 1.549083629453726e-05,
-      "loss": 0.0029,
+      "epoch": 1.0784,
+      "grad_norm": 0.11108493026233916,
+      "learning_rate": 1.4834487875162657e-05,
+      "loss": 0.0013,
       "step": 674
     },
     {
-      "epoch": 1.0104790419161676,
-      "grad_norm": 0.12322267561572196,
-      "learning_rate": 1.547731578927033e-05,
-      "loss": 0.0017,
+      "epoch": 1.08,
+      "grad_norm": 0.217810679569032,
+      "learning_rate": 1.4819353798236427e-05,
+      "loss": 0.0018,
       "step": 675
     },
     {
-      "epoch": 1.0119760479041917,
-      "grad_norm": 0.1847824809505218,
-      "learning_rate": 1.5463780964702017e-05,
-      "loss": 0.002,
+      "epoch": 1.0816,
+      "grad_norm": 0.006093427175663653,
+      "learning_rate": 1.4804205329988226e-05,
+      "loss": 0.0001,
       "step": 676
     },
     {
-      "epoch": 1.0134730538922156,
-      "grad_norm": 0.09718374857400291,
-      "learning_rate": 1.5450231856216294e-05,
-      "loss": 0.0014,
+      "epoch": 1.0832,
+      "grad_norm": 0.0969390115935947,
+      "learning_rate": 1.4789042515653687e-05,
+      "loss": 0.0005,
       "step": 677
     },
     {
-      "epoch": 1.0149700598802396,
-      "grad_norm": 0.11714727810772602,
-      "learning_rate": 1.5436668499234487e-05,
-      "loss": 0.0015,
+      "epoch": 1.0848,
+      "grad_norm": 0.016953908558966193,
+      "learning_rate": 1.477386540051127e-05,
+      "loss": 0.0002,
       "step": 678
     },
     {
-      "epoch": 1.0164670658682635,
-      "grad_norm": 0.17180981362774483,
-      "learning_rate": 1.5423090929215164e-05,
-      "loss": 0.0016,
+      "epoch": 1.0864,
+      "grad_norm": 0.009974731053493537,
+      "learning_rate": 1.4758674029882152e-05,
+      "loss": 0.0001,
       "step": 679
     },
     {
-      "epoch": 1.0179640718562875,
-      "grad_norm": 0.1229563131274387,
-      "learning_rate": 1.5409499181654062e-05,
-      "loss": 0.0022,
+      "epoch": 1.088,
+      "grad_norm": 0.08345962192160167,
+      "learning_rate": 1.4743468449130065e-05,
+      "loss": 0.0004,
       "step": 680
     },
     {
-      "epoch": 1.0194610778443114,
-      "grad_norm": 0.08902729595877537,
-      "learning_rate": 1.5395893292083967e-05,
-      "loss": 0.0012,
+      "epoch": 1.0896,
+      "grad_norm": 0.11817389598335944,
+      "learning_rate": 1.4728248703661183e-05,
+      "loss": 0.0005,
       "step": 681
     },
     {
-      "epoch": 1.0209580838323353,
-      "grad_norm": 0.08491224818504717,
-      "learning_rate": 1.5382273296074647e-05,
-      "loss": 0.0014,
+      "epoch": 1.0912,
+      "grad_norm": 0.07616966500087563,
+      "learning_rate": 1.4713014838923975e-05,
+      "loss": 0.0011,
       "step": 682
     },
     {
-      "epoch": 1.0224550898203593,
-      "grad_norm": 0.12463262583052379,
-      "learning_rate": 1.5368639229232744e-05,
-      "loss": 0.0012,
+      "epoch": 1.0928,
+      "grad_norm": 0.04248575954910725,
+      "learning_rate": 1.4697766900409076e-05,
+      "loss": 0.0002,
       "step": 683
     },
     {
-      "epoch": 1.0239520958083832,
-      "grad_norm": 0.08714987099310947,
-      "learning_rate": 1.5354991127201682e-05,
-      "loss": 0.0009,
+      "epoch": 1.0944,
+      "grad_norm": 0.028265559311665706,
+      "learning_rate": 1.4682504933649144e-05,
+      "loss": 0.0002,
       "step": 684
     },
     {
-      "epoch": 1.0254491017964071,
-      "grad_norm": 0.10996908003888108,
-      "learning_rate": 1.5341329025661593e-05,
-      "loss": 0.0016,
+      "epoch": 1.096,
+      "grad_norm": 0.1910697994779768,
+      "learning_rate": 1.466722898421873e-05,
+      "loss": 0.0006,
       "step": 685
     },
     {
-      "epoch": 1.026946107784431,
-      "grad_norm": 0.06688339180384366,
-      "learning_rate": 1.5327652960329182e-05,
-      "loss": 0.0011,
+      "epoch": 1.0976,
+      "grad_norm": 0.023588905330173503,
+      "learning_rate": 1.4651939097734132e-05,
+      "loss": 0.0002,
       "step": 686
     },
     {
-      "epoch": 1.028443113772455,
-      "grad_norm": 0.11389018736693612,
-      "learning_rate": 1.5313962966957683e-05,
-      "loss": 0.0013,
+      "epoch": 1.0992,
+      "grad_norm": 0.06545131637159864,
+      "learning_rate": 1.4636635319853274e-05,
+      "loss": 0.0006,
       "step": 687
     },
     {
-      "epoch": 1.029940119760479,
-      "grad_norm": 0.1487541170115952,
-      "learning_rate": 1.5300259081336733e-05,
-      "loss": 0.0029,
+      "epoch": 1.1008,
+      "grad_norm": 0.21589569803438816,
+      "learning_rate": 1.4621317696275563e-05,
+      "loss": 0.0035,
       "step": 688
     },
     {
-      "epoch": 1.031437125748503,
-      "grad_norm": 0.18232592006619516,
-      "learning_rate": 1.5286541339292277e-05,
-      "loss": 0.0019,
+      "epoch": 1.1024,
+      "grad_norm": 0.024193057611844735,
+      "learning_rate": 1.4605986272741748e-05,
+      "loss": 0.0002,
       "step": 689
     },
     {
-      "epoch": 1.032934131736527,
-      "grad_norm": 0.09073197180745511,
-      "learning_rate": 1.5272809776686507e-05,
-      "loss": 0.001,
+      "epoch": 1.104,
+      "grad_norm": 0.017582182124678495,
+      "learning_rate": 1.4590641095033786e-05,
+      "loss": 0.0002,
       "step": 690
     },
     {
-      "epoch": 1.034431137724551,
-      "grad_norm": 0.11198750989897426,
-      "learning_rate": 1.5259064429417724e-05,
-      "loss": 0.0011,
+      "epoch": 1.1056,
+      "grad_norm": 0.1314061948393412,
+      "learning_rate": 1.4575282208974704e-05,
+      "loss": 0.0009,
       "step": 691
     },
     {
-      "epoch": 1.035928143712575,
-      "grad_norm": 0.13524279713865056,
-      "learning_rate": 1.5245305333420284e-05,
-      "loss": 0.0027,
+      "epoch": 1.1072,
+      "grad_norm": 0.04667058659825015,
+      "learning_rate": 1.4559909660428469e-05,
+      "loss": 0.0002,
       "step": 692
     },
     {
-      "epoch": 1.0374251497005988,
-      "grad_norm": 0.13473254834146128,
-      "learning_rate": 1.5231532524664472e-05,
-      "loss": 0.0013,
+      "epoch": 1.1088,
+      "grad_norm": 0.15259160104285077,
+      "learning_rate": 1.4544523495299843e-05,
+      "loss": 0.0019,
       "step": 693
     },
     {
-      "epoch": 1.0389221556886228,
-      "grad_norm": 0.09705487229884018,
-      "learning_rate": 1.5217746039156426e-05,
-      "loss": 0.0014,
+      "epoch": 1.1104,
+      "grad_norm": 0.15183714545641216,
+      "learning_rate": 1.4529123759534253e-05,
+      "loss": 0.0011,
       "step": 694
     },
     {
-      "epoch": 1.0404191616766467,
-      "grad_norm": 0.12743903961819494,
-      "learning_rate": 1.5203945912938047e-05,
-      "loss": 0.0031,
+      "epoch": 1.112,
+      "grad_norm": 0.03155705221297175,
+      "learning_rate": 1.4513710499117648e-05,
+      "loss": 0.0002,
       "step": 695
     },
     {
-      "epoch": 1.0419161676646707,
-      "grad_norm": 0.15772232058272861,
-      "learning_rate": 1.5190132182086887e-05,
-      "loss": 0.0035,
+      "epoch": 1.1136,
+      "grad_norm": 0.00761205137275975,
+      "learning_rate": 1.4498283760076362e-05,
+      "loss": 0.0001,
       "step": 696
     },
     {
-      "epoch": 1.0434131736526946,
-      "grad_norm": 0.11967873813625308,
-      "learning_rate": 1.5176304882716074e-05,
-      "loss": 0.0021,
+      "epoch": 1.1152,
+      "grad_norm": 0.17988270920033475,
+      "learning_rate": 1.4482843588476976e-05,
+      "loss": 0.0009,
       "step": 697
     },
     {
-      "epoch": 1.0449101796407185,
-      "grad_norm": 0.061224633547482255,
-      "learning_rate": 1.5162464050974195e-05,
+      "epoch": 1.1168,
+      "grad_norm": 0.06711856696888893,
+      "learning_rate": 1.4467390030426187e-05,
       "loss": 0.0006,
       "step": 698
     },
     {
-      "epoch": 1.0464071856287425,
-      "grad_norm": 0.12437711893494492,
-      "learning_rate": 1.5148609723045227e-05,
-      "loss": 0.0012,
+      "epoch": 1.1184,
+      "grad_norm": 0.007982256658821624,
+      "learning_rate": 1.445192313207067e-05,
+      "loss": 0.0001,
       "step": 699
     },
     {
-      "epoch": 1.0479041916167664,
-      "grad_norm": 0.1745613764437353,
-      "learning_rate": 1.513474193514842e-05,
-      "loss": 0.0017,
+      "epoch": 1.12,
+      "grad_norm": 0.002927265853794833,
+      "learning_rate": 1.443644293959693e-05,
+      "loss": 0.0001,
       "step": 700
     },
     {
-      "epoch": 1.0494011976047903,
-      "grad_norm": 0.16175781608452677,
-      "learning_rate": 1.5120860723538224e-05,
-      "loss": 0.0012,
+      "epoch": 1.1216,
+      "grad_norm": 0.10777686557463714,
+      "learning_rate": 1.4420949499231172e-05,
+      "loss": 0.0003,
       "step": 701
     },
     {
-      "epoch": 1.0508982035928143,
-      "grad_norm": 0.10221564506113642,
-      "learning_rate": 1.5106966124504172e-05,
-      "loss": 0.002,
+      "epoch": 1.1232,
+      "grad_norm": 0.11134637630246084,
+      "learning_rate": 1.4405442857239151e-05,
+      "loss": 0.0012,
       "step": 702
     },
     {
-      "epoch": 1.0523952095808382,
-      "grad_norm": 0.062271705757527857,
-      "learning_rate": 1.5093058174370798e-05,
-      "loss": 0.0008,
+      "epoch": 1.1248,
+      "grad_norm": 0.017288876298526105,
+      "learning_rate": 1.4389923059926064e-05,
+      "loss": 0.0002,
       "step": 703
     },
     {
-      "epoch": 1.0538922155688624,
-      "grad_norm": 0.10864619286621946,
-      "learning_rate": 1.5079136909497542e-05,
+      "epoch": 1.1264,
+      "grad_norm": 0.10471184371986594,
+      "learning_rate": 1.437439015363638e-05,
       "loss": 0.0009,
       "step": 704
     },
     {
-      "epoch": 1.0553892215568863,
-      "grad_norm": 0.19647706361842368,
-      "learning_rate": 1.5065202366278657e-05,
-      "loss": 0.0014,
+      "epoch": 1.1280000000000001,
+      "grad_norm": 0.009296799728898991,
+      "learning_rate": 1.4358844184753713e-05,
+      "loss": 0.0001,
       "step": 705
     },
     {
-      "epoch": 1.0568862275449102,
-      "grad_norm": 0.07888279226925543,
-      "learning_rate": 1.5051254581143095e-05,
-      "loss": 0.0008,
+      "epoch": 1.1296,
+      "grad_norm": 0.20219144885410853,
+      "learning_rate": 1.4343285199700685e-05,
+      "loss": 0.0026,
       "step": 706
     },
     {
-      "epoch": 1.0583832335329342,
-      "grad_norm": 0.12025352331354569,
-      "learning_rate": 1.5037293590554442e-05,
-      "loss": 0.0012,
+      "epoch": 1.1312,
+      "grad_norm": 0.020697705406549943,
+      "learning_rate": 1.432771324493879e-05,
+      "loss": 0.0002,
       "step": 707
     },
     {
-      "epoch": 1.0598802395209581,
-      "grad_norm": 0.08615367646336641,
-      "learning_rate": 1.5023319431010798e-05,
-      "loss": 0.0013,
+      "epoch": 1.1328,
+      "grad_norm": 0.0066267694348396385,
+      "learning_rate": 1.4312128366968244e-05,
+      "loss": 0.0001,
       "step": 708
     },
     {
-      "epoch": 1.061377245508982,
-      "grad_norm": 0.14909244953515943,
-      "learning_rate": 1.5009332139044696e-05,
-      "loss": 0.0013,
+      "epoch": 1.1344,
+      "grad_norm": 0.02044341762711039,
+      "learning_rate": 1.4296530612327864e-05,
+      "loss": 0.0002,
       "step": 709
     },
     {
-      "epoch": 1.062874251497006,
-      "grad_norm": 0.10899940946016365,
-      "learning_rate": 1.4995331751222992e-05,
-      "loss": 0.0014,
+      "epoch": 1.1360000000000001,
+      "grad_norm": 0.10781695163193859,
+      "learning_rate": 1.428092002759491e-05,
+      "loss": 0.0005,
       "step": 710
     },
     {
-      "epoch": 1.06437125748503,
-      "grad_norm": 0.13011579272063212,
-      "learning_rate": 1.4981318304146786e-05,
-      "loss": 0.0026,
+      "epoch": 1.1376,
+      "grad_norm": 0.0376769847929488,
+      "learning_rate": 1.4265296659384956e-05,
+      "loss": 0.0002,
       "step": 711
     },
     {
-      "epoch": 1.0658682634730539,
-      "grad_norm": 0.06507001266187337,
-      "learning_rate": 1.4967291834451317e-05,
-      "loss": 0.0008,
+      "epoch": 1.1392,
+      "grad_norm": 0.019401313477370914,
+      "learning_rate": 1.4249660554351752e-05,
+      "loss": 0.0002,
       "step": 712
     },
     {
-      "epoch": 1.0673652694610778,
-      "grad_norm": 0.1452953575075588,
-      "learning_rate": 1.4953252378805872e-05,
-      "loss": 0.0027,
+      "epoch": 1.1408,
+      "grad_norm": 0.00674801299827939,
+      "learning_rate": 1.4234011759187084e-05,
+      "loss": 0.0001,
       "step": 713
     },
     {
-      "epoch": 1.0688622754491017,
-      "grad_norm": 0.05183547815818978,
-      "learning_rate": 1.493919997391368e-05,
-      "loss": 0.0007,
+      "epoch": 1.1424,
+      "grad_norm": 0.01065028286198891,
+      "learning_rate": 1.4218350320620625e-05,
+      "loss": 0.0001,
       "step": 714
     },
     {
-      "epoch": 1.0703592814371257,
-      "grad_norm": 0.07714623310644146,
-      "learning_rate": 1.492513465651183e-05,
-      "loss": 0.0008,
+      "epoch": 1.144,
+      "grad_norm": 0.05165496788259497,
+      "learning_rate": 1.4202676285419811e-05,
+      "loss": 0.0004,
       "step": 715
     },
     {
-      "epoch": 1.0718562874251496,
-      "grad_norm": 0.1273003039382832,
-      "learning_rate": 1.4911056463371161e-05,
-      "loss": 0.0018,
+      "epoch": 1.1456,
+      "grad_norm": 0.011510130836619201,
+      "learning_rate": 1.4186989700389689e-05,
+      "loss": 0.0001,
       "step": 716
     },
     {
-      "epoch": 1.0733532934131738,
-      "grad_norm": 0.14647086320390126,
-      "learning_rate": 1.4896965431296182e-05,
-      "loss": 0.002,
+      "epoch": 1.1472,
+      "grad_norm": 0.044944190679150994,
+      "learning_rate": 1.4171290612372781e-05,
+      "loss": 0.0003,
       "step": 717
     },
     {
-      "epoch": 1.0748502994011977,
-      "grad_norm": 0.11031963684292324,
-      "learning_rate": 1.4882861597124954e-05,
-      "loss": 0.0017,
+      "epoch": 1.1488,
+      "grad_norm": 0.060448606493635766,
+      "learning_rate": 1.4155579068248951e-05,
+      "loss": 0.0003,
       "step": 718
     },
     {
-      "epoch": 1.0763473053892216,
-      "grad_norm": 0.13321582362567036,
-      "learning_rate": 1.4868744997729024e-05,
-      "loss": 0.0015,
+      "epoch": 1.1504,
+      "grad_norm": 0.012225954357272079,
+      "learning_rate": 1.4139855114935253e-05,
+      "loss": 0.0001,
       "step": 719
     },
     {
-      "epoch": 1.0778443113772456,
-      "grad_norm": 0.12391394850850491,
-      "learning_rate": 1.4854615670013295e-05,
-      "loss": 0.0014,
+      "epoch": 1.152,
+      "grad_norm": 0.08028402779935753,
+      "learning_rate": 1.4124118799385797e-05,
+      "loss": 0.001,
       "step": 720
     },
     {
-      "epoch": 1.0793413173652695,
-      "grad_norm": 0.16741971891552657,
-      "learning_rate": 1.4840473650915952e-05,
-      "loss": 0.0014,
+      "epoch": 1.1536,
+      "grad_norm": 0.059661115677953874,
+      "learning_rate": 1.410837016859161e-05,
+      "loss": 0.0004,
       "step": 721
     },
     {
-      "epoch": 1.0808383233532934,
-      "grad_norm": 0.06084190134638075,
-      "learning_rate": 1.4826318977408358e-05,
-      "loss": 0.0008,
+      "epoch": 1.1552,
+      "grad_norm": 0.03410361311916781,
+      "learning_rate": 1.4092609269580498e-05,
+      "loss": 0.0002,
       "step": 722
     },
     {
-      "epoch": 1.0823353293413174,
-      "grad_norm": 0.09523046312621482,
-      "learning_rate": 1.4812151686494962e-05,
-      "loss": 0.0011,
+      "epoch": 1.1568,
+      "grad_norm": 0.11897638231253438,
+      "learning_rate": 1.4076836149416889e-05,
+      "loss": 0.0003,
       "step": 723
     },
     {
-      "epoch": 1.0838323353293413,
-      "grad_norm": 0.2782155057266579,
-      "learning_rate": 1.4797971815213194e-05,
-      "loss": 0.0027,
+      "epoch": 1.1584,
+      "grad_norm": 0.008149391142795047,
+      "learning_rate": 1.4061050855201723e-05,
+      "loss": 0.0001,
       "step": 724
     },
     {
-      "epoch": 1.0853293413173652,
-      "grad_norm": 0.07866338116148361,
-      "learning_rate": 1.4783779400633372e-05,
-      "loss": 0.0019,
+      "epoch": 1.16,
+      "grad_norm": 0.04198467796283594,
+      "learning_rate": 1.4045253434072278e-05,
+      "loss": 0.0003,
       "step": 725
     },
     {
-      "epoch": 1.0868263473053892,
-      "grad_norm": 0.08163897393875569,
-      "learning_rate": 1.4769574479858612e-05,
-      "loss": 0.0013,
+      "epoch": 1.1616,
+      "grad_norm": 0.08246516092578157,
+      "learning_rate": 1.4029443933202059e-05,
+      "loss": 0.0034,
       "step": 726
     },
     {
-      "epoch": 1.0883233532934131,
-      "grad_norm": 0.06549200205338107,
-      "learning_rate": 1.475535709002472e-05,
-      "loss": 0.0008,
+      "epoch": 1.1632,
+      "grad_norm": 0.014962798528117778,
+      "learning_rate": 1.4013622399800628e-05,
+      "loss": 0.0002,
       "step": 727
     },
     {
-      "epoch": 1.089820359281437,
-      "grad_norm": 0.10112474427638028,
-      "learning_rate": 1.4741127268300097e-05,
-      "loss": 0.0015,
+      "epoch": 1.1648,
+      "grad_norm": 0.022593058469230293,
+      "learning_rate": 1.399778888111349e-05,
+      "loss": 0.0002,
       "step": 728
     },
     {
-      "epoch": 1.091317365269461,
-      "grad_norm": 0.03922519037579488,
-      "learning_rate": 1.4726885051885654e-05,
-      "loss": 0.0005,
+      "epoch": 1.1663999999999999,
+      "grad_norm": 0.14837290202030107,
+      "learning_rate": 1.3981943424421932e-05,
+      "loss": 0.0016,
       "step": 729
     },
     {
-      "epoch": 1.092814371257485,
-      "grad_norm": 0.12932978632302858,
-      "learning_rate": 1.4712630478014696e-05,
-      "loss": 0.0013,
+      "epoch": 1.168,
+      "grad_norm": 0.013417266184973483,
+      "learning_rate": 1.3966086077042891e-05,
+      "loss": 0.0001,
       "step": 730
     },
     {
-      "epoch": 1.0943113772455089,
-      "grad_norm": 0.14927429413650972,
-      "learning_rate": 1.4698363583952839e-05,
-      "loss": 0.0019,
+      "epoch": 1.1696,
+      "grad_norm": 0.11604785901225578,
+      "learning_rate": 1.3950216886328818e-05,
+      "loss": 0.0009,
       "step": 731
     },
     {
-      "epoch": 1.095808383233533,
-      "grad_norm": 0.08123699864404504,
-      "learning_rate": 1.4684084406997903e-05,
-      "loss": 0.0011,
+      "epoch": 1.1712,
+      "grad_norm": 0.003183819460730826,
+      "learning_rate": 1.3934335899667526e-05,
+      "loss": 0.0001,
       "step": 732
     },
     {
-      "epoch": 1.097305389221557,
-      "grad_norm": 0.05959853755308904,
-      "learning_rate": 1.4669792984479828e-05,
-      "loss": 0.0012,
+      "epoch": 1.1728,
+      "grad_norm": 0.006302224762777941,
+      "learning_rate": 1.3918443164482048e-05,
+      "loss": 0.0001,
       "step": 733
     },
     {
-      "epoch": 1.098802395209581,
-      "grad_norm": 0.11908929599820205,
-      "learning_rate": 1.4655489353760555e-05,
-      "loss": 0.0014,
+      "epoch": 1.1743999999999999,
+      "grad_norm": 0.028227325989654887,
+      "learning_rate": 1.3902538728230502e-05,
+      "loss": 0.0002,
       "step": 734
     },
     {
-      "epoch": 1.1002994011976048,
-      "grad_norm": 0.09093165505220649,
-      "learning_rate": 1.4641173552233954e-05,
-      "loss": 0.0008,
+      "epoch": 1.176,
+      "grad_norm": 0.07163952091692051,
+      "learning_rate": 1.3886622638405953e-05,
+      "loss": 0.0013,
       "step": 735
     },
     {
-      "epoch": 1.1017964071856288,
-      "grad_norm": 0.11618048676071026,
-      "learning_rate": 1.4626845617325703e-05,
-      "loss": 0.0013,
+      "epoch": 1.1776,
+      "grad_norm": 0.007588300955138563,
+      "learning_rate": 1.387069494253626e-05,
+      "loss": 0.0001,
       "step": 736
     },
     {
-      "epoch": 1.1032934131736527,
-      "grad_norm": 0.06281753166476825,
-      "learning_rate": 1.4612505586493204e-05,
-      "loss": 0.0007,
+      "epoch": 1.1792,
+      "grad_norm": 0.004028009467982749,
+      "learning_rate": 1.3854755688183941e-05,
+      "loss": 0.0001,
       "step": 737
     },
     {
-      "epoch": 1.1047904191616766,
-      "grad_norm": 0.2023556555593652,
-      "learning_rate": 1.4598153497225482e-05,
-      "loss": 0.002,
+      "epoch": 1.1808,
+      "grad_norm": 0.00948987038982055,
+      "learning_rate": 1.3838804922946027e-05,
+      "loss": 0.0002,
       "step": 738
     },
     {
-      "epoch": 1.1062874251497006,
-      "grad_norm": 0.11177065874880013,
-      "learning_rate": 1.4583789387043083e-05,
-      "loss": 0.0013,
+      "epoch": 1.1824,
+      "grad_norm": 0.03314448760771682,
+      "learning_rate": 1.3822842694453923e-05,
+      "loss": 0.0003,
       "step": 739
     },
     {
-      "epoch": 1.1077844311377245,
-      "grad_norm": 0.14741071238122402,
-      "learning_rate": 1.4569413293497986e-05,
-      "loss": 0.0015,
+      "epoch": 1.184,
+      "grad_norm": 0.018637812219687012,
+      "learning_rate": 1.380686905037327e-05,
+      "loss": 0.0001,
       "step": 740
     },
     {
-      "epoch": 1.1092814371257484,
-      "grad_norm": 0.24658469283044443,
-      "learning_rate": 1.455502525417349e-05,
-      "loss": 0.0012,
+      "epoch": 1.1856,
+      "grad_norm": 0.026116887969995242,
+      "learning_rate": 1.3790884038403796e-05,
+      "loss": 0.0002,
       "step": 741
     },
     {
-      "epoch": 1.1107784431137724,
-      "grad_norm": 0.13681115792701468,
-      "learning_rate": 1.454062530668413e-05,
-      "loss": 0.0017,
+      "epoch": 1.1872,
+      "grad_norm": 0.04576868274139848,
+      "learning_rate": 1.3774887706279165e-05,
+      "loss": 0.0002,
       "step": 742
     },
     {
-      "epoch": 1.1122754491017963,
-      "grad_norm": 0.0834969421467918,
-      "learning_rate": 1.4526213488675564e-05,
-      "loss": 0.001,
+      "epoch": 1.1888,
+      "grad_norm": 0.21322502984555788,
+      "learning_rate": 1.375888010176686e-05,
+      "loss": 0.0012,
       "step": 743
     },
     {
-      "epoch": 1.1137724550898203,
-      "grad_norm": 0.1940908997890325,
-      "learning_rate": 1.4511789837824495e-05,
-      "loss": 0.0015,
+      "epoch": 1.1904,
+      "grad_norm": 0.008027445621927178,
+      "learning_rate": 1.374286127266801e-05,
+      "loss": 0.0001,
       "step": 744
     },
     {
-      "epoch": 1.1152694610778444,
-      "grad_norm": 0.10979405041458333,
-      "learning_rate": 1.4497354391838551e-05,
-      "loss": 0.0011,
+      "epoch": 1.192,
+      "grad_norm": 0.014333097695883089,
+      "learning_rate": 1.3726831266817278e-05,
+      "loss": 0.0002,
       "step": 745
     },
     {
-      "epoch": 1.1167664670658684,
-      "grad_norm": 0.11938289413679191,
-      "learning_rate": 1.4482907188456198e-05,
-      "loss": 0.0012,
+      "epoch": 1.1936,
+      "grad_norm": 0.01649187155480879,
+      "learning_rate": 1.3710790132082693e-05,
+      "loss": 0.0001,
       "step": 746
     },
     {
-      "epoch": 1.1182634730538923,
-      "grad_norm": 0.1358936092557254,
-      "learning_rate": 1.4468448265446646e-05,
-      "loss": 0.0013,
+      "epoch": 1.1952,
+      "grad_norm": 0.049496975480294164,
+      "learning_rate": 1.3694737916365517e-05,
+      "loss": 0.0001,
       "step": 747
     },
     {
-      "epoch": 1.1197604790419162,
-      "grad_norm": 0.1354545576585189,
-      "learning_rate": 1.4453977660609733e-05,
-      "loss": 0.0014,
+      "epoch": 1.1968,
+      "grad_norm": 0.23413285031093056,
+      "learning_rate": 1.3678674667600102e-05,
+      "loss": 0.003,
       "step": 748
     },
     {
-      "epoch": 1.1212574850299402,
-      "grad_norm": 0.08442533262908321,
-      "learning_rate": 1.4439495411775844e-05,
-      "loss": 0.001,
+      "epoch": 1.1984,
+      "grad_norm": 0.029064562944119434,
+      "learning_rate": 1.3662600433753746e-05,
+      "loss": 0.0001,
       "step": 749
     },
     {
-      "epoch": 1.122754491017964,
-      "grad_norm": 0.07602658081845277,
-      "learning_rate": 1.4425001556805796e-05,
-      "loss": 0.0008,
+      "epoch": 1.2,
+      "grad_norm": 0.08357082403130471,
+      "learning_rate": 1.3646515262826551e-05,
+      "loss": 0.0019,
       "step": 750
     },
     {
-      "epoch": 1.124251497005988,
-      "grad_norm": 0.13026513890902275,
-      "learning_rate": 1.4410496133590763e-05,
-      "loss": 0.0012,
+      "epoch": 1.2016,
+      "grad_norm": 0.06813924590320486,
+      "learning_rate": 1.3630419202851287e-05,
+      "loss": 0.0007,
       "step": 751
     },
     {
-      "epoch": 1.125748502994012,
-      "grad_norm": 0.08108549950525631,
-      "learning_rate": 1.439597918005215e-05,
-      "loss": 0.0006,
+      "epoch": 1.2032,
+      "grad_norm": 0.050105407692925645,
+      "learning_rate": 1.3614312301893222e-05,
+      "loss": 0.0003,
       "step": 752
     },
     {
-      "epoch": 1.127245508982036,
-      "grad_norm": 0.2152317708482709,
-      "learning_rate": 1.4381450734141505e-05,
-      "loss": 0.002,
+      "epoch": 1.2048,
+      "grad_norm": 0.03927881336852108,
+      "learning_rate": 1.3598194608050011e-05,
+      "loss": 0.0002,
       "step": 753
     },
     {
-      "epoch": 1.1287425149700598,
-      "grad_norm": 0.08773076511235232,
-      "learning_rate": 1.4366910833840431e-05,
-      "loss": 0.0021,
+      "epoch": 1.2064,
+      "grad_norm": 0.10718101644698451,
+      "learning_rate": 1.3582066169451535e-05,
+      "loss": 0.0004,
       "step": 754
     },
     {
-      "epoch": 1.1302395209580838,
-      "grad_norm": 0.17027037331891098,
-      "learning_rate": 1.4352359517160462e-05,
-      "loss": 0.0013,
+      "epoch": 1.208,
+      "grad_norm": 0.06591309918475667,
+      "learning_rate": 1.3565927034259757e-05,
+      "loss": 0.0004,
       "step": 755
     },
     {
-      "epoch": 1.1317365269461077,
-      "grad_norm": 0.08918657681225792,
-      "learning_rate": 1.4337796822142987e-05,
-      "loss": 0.0015,
+      "epoch": 1.2096,
+      "grad_norm": 0.05031545686049723,
+      "learning_rate": 1.354977725066859e-05,
+      "loss": 0.0002,
       "step": 756
     },
     {
-      "epoch": 1.1332335329341316,
-      "grad_norm": 0.1305105215164226,
-      "learning_rate": 1.4323222786859141e-05,
-      "loss": 0.0011,
+      "epoch": 1.2112,
+      "grad_norm": 0.002031493397625212,
+      "learning_rate": 1.3533616866903736e-05,
+      "loss": 0.0001,
       "step": 757
     },
     {
-      "epoch": 1.1347305389221556,
-      "grad_norm": 0.12593434714214163,
-      "learning_rate": 1.4308637449409705e-05,
-      "loss": 0.0011,
+      "epoch": 1.2128,
+      "grad_norm": 0.08640031574979783,
+      "learning_rate": 1.351744593122255e-05,
+      "loss": 0.0018,
       "step": 758
     },
     {
-      "epoch": 1.1362275449101795,
-      "grad_norm": 0.3028857079800561,
-      "learning_rate": 1.4294040847925004e-05,
-      "loss": 0.0035,
+      "epoch": 1.2144,
+      "grad_norm": 0.11112324060376225,
+      "learning_rate": 1.3501264491913909e-05,
+      "loss": 0.0011,
       "step": 759
     },
     {
-      "epoch": 1.1377245508982037,
-      "grad_norm": 0.24334283105322427,
-      "learning_rate": 1.4279433020564813e-05,
-      "loss": 0.002,
+      "epoch": 1.216,
+      "grad_norm": 0.10879913552858618,
+      "learning_rate": 1.3485072597298038e-05,
+      "loss": 0.0013,
       "step": 760
     },
     {
-      "epoch": 1.1392215568862276,
-      "grad_norm": 0.18593254416098137,
-      "learning_rate": 1.4264814005518251e-05,
-      "loss": 0.0033,
+      "epoch": 1.2176,
+      "grad_norm": 0.044457695496770214,
+      "learning_rate": 1.3468870295726399e-05,
+      "loss": 0.0003,
       "step": 761
     },
     {
-      "epoch": 1.1407185628742516,
-      "grad_norm": 0.158884487983083,
-      "learning_rate": 1.4250183841003695e-05,
-      "loss": 0.0013,
+      "epoch": 1.2192,
+      "grad_norm": 0.09420747266028137,
+      "learning_rate": 1.3452657635581521e-05,
+      "loss": 0.0004,
       "step": 762
     },
     {
-      "epoch": 1.1422155688622755,
-      "grad_norm": 0.14632297769339114,
-      "learning_rate": 1.4235542565268658e-05,
-      "loss": 0.0015,
+      "epoch": 1.2208,
+      "grad_norm": 0.02427716999706509,
+      "learning_rate": 1.3436434665276865e-05,
+      "loss": 0.0002,
       "step": 763
     },
     {
-      "epoch": 1.1437125748502994,
-      "grad_norm": 0.15199709628186464,
-      "learning_rate": 1.4220890216589708e-05,
-      "loss": 0.0011,
+      "epoch": 1.2224,
+      "grad_norm": 0.009056916304727262,
+      "learning_rate": 1.342020143325669e-05,
+      "loss": 0.0001,
       "step": 764
     },
     {
-      "epoch": 1.1452095808383234,
-      "grad_norm": 0.07701639398769143,
-      "learning_rate": 1.4206226833272359e-05,
-      "loss": 0.0011,
+      "epoch": 1.224,
+      "grad_norm": 0.08248517704520834,
+      "learning_rate": 1.3403957987995884e-05,
+      "loss": 0.0008,
       "step": 765
     },
     {
-      "epoch": 1.1467065868263473,
-      "grad_norm": 0.06624545307291502,
-      "learning_rate": 1.4191552453650972e-05,
-      "loss": 0.0007,
+      "epoch": 1.2256,
+      "grad_norm": 0.019999402030378328,
+      "learning_rate": 1.3387704377999842e-05,
+      "loss": 0.0002,
       "step": 766
     },
     {
-      "epoch": 1.1482035928143712,
-      "grad_norm": 0.18898462608830638,
-      "learning_rate": 1.4176867116088655e-05,
-      "loss": 0.0017,
+      "epoch": 1.2272,
+      "grad_norm": 0.20880613727618944,
+      "learning_rate": 1.3371440651804313e-05,
+      "loss": 0.001,
       "step": 767
     },
     {
-      "epoch": 1.1497005988023952,
-      "grad_norm": 0.11325039624754406,
-      "learning_rate": 1.4162170858977167e-05,
-      "loss": 0.0013,
+      "epoch": 1.2288000000000001,
+      "grad_norm": 0.05269074816169083,
+      "learning_rate": 1.335516685797525e-05,
+      "loss": 0.0003,
       "step": 768
     },
     {
-      "epoch": 1.151197604790419,
-      "grad_norm": 0.1260621674366222,
-      "learning_rate": 1.4147463720736813e-05,
-      "loss": 0.0023,
+      "epoch": 1.2304,
+      "grad_norm": 0.029644432614742842,
+      "learning_rate": 1.3338883045108674e-05,
+      "loss": 0.0003,
       "step": 769
     },
     {
-      "epoch": 1.152694610778443,
-      "grad_norm": 0.10527624531422942,
-      "learning_rate": 1.4132745739816341e-05,
-      "loss": 0.0008,
+      "epoch": 1.232,
+      "grad_norm": 0.05457781914849625,
+      "learning_rate": 1.3322589261830517e-05,
+      "loss": 0.0003,
       "step": 770
     },
     {
-      "epoch": 1.154191616766467,
-      "grad_norm": 0.22598741769102923,
-      "learning_rate": 1.4118016954692845e-05,
-      "loss": 0.0022,
+      "epoch": 1.2336,
+      "grad_norm": 0.00860704666475476,
+      "learning_rate": 1.3306285556796494e-05,
+      "loss": 0.0001,
       "step": 771
     },
     {
-      "epoch": 1.1556886227544911,
-      "grad_norm": 0.11216311604503693,
-      "learning_rate": 1.4103277403871667e-05,
-      "loss": 0.0021,
+      "epoch": 1.2352,
+      "grad_norm": 0.03239844982558144,
+      "learning_rate": 1.328997197869194e-05,
+      "loss": 0.0002,
       "step": 772
     },
     {
-      "epoch": 1.157185628742515,
-      "grad_norm": 0.12672041749180385,
-      "learning_rate": 1.4088527125886294e-05,
-      "loss": 0.0021,
+      "epoch": 1.2368000000000001,
+      "grad_norm": 0.003695385023378297,
+      "learning_rate": 1.327364857623168e-05,
+      "loss": 0.0001,
       "step": 773
     },
     {
-      "epoch": 1.158682634730539,
-      "grad_norm": 0.05731251869782339,
-      "learning_rate": 1.4073766159298257e-05,
-      "loss": 0.0006,
+      "epoch": 1.2384,
+      "grad_norm": 0.01608456444419951,
+      "learning_rate": 1.3257315398159865e-05,
+      "loss": 0.0001,
       "step": 774
     },
     {
-      "epoch": 1.160179640718563,
-      "grad_norm": 0.12190886299689159,
-      "learning_rate": 1.4058994542697027e-05,
-      "loss": 0.0013,
+      "epoch": 1.24,
+      "grad_norm": 0.042718393501721276,
+      "learning_rate": 1.3240972493249846e-05,
+      "loss": 0.0002,
       "step": 775
     },
     {
-      "epoch": 1.1616766467065869,
-      "grad_norm": 0.16571654034581215,
-      "learning_rate": 1.4044212314699919e-05,
-      "loss": 0.0021,
+      "epoch": 1.2416,
+      "grad_norm": 0.053524927786504244,
+      "learning_rate": 1.3224619910304019e-05,
+      "loss": 0.0003,
       "step": 776
     },
     {
-      "epoch": 1.1631736526946108,
-      "grad_norm": 0.12852042640903608,
-      "learning_rate": 1.402941951395199e-05,
-      "loss": 0.0017,
+      "epoch": 1.2432,
+      "grad_norm": 0.035849151817245385,
+      "learning_rate": 1.3208257698153677e-05,
+      "loss": 0.0002,
       "step": 777
     },
     {
-      "epoch": 1.1646706586826348,
-      "grad_norm": 0.04996364604118305,
-      "learning_rate": 1.4014616179125935e-05,
-      "loss": 0.0004,
+      "epoch": 1.2448,
+      "grad_norm": 0.05444679154132394,
+      "learning_rate": 1.3191885905658873e-05,
+      "loss": 0.0005,
       "step": 778
     },
     {
-      "epoch": 1.1661676646706587,
-      "grad_norm": 0.22739207612882284,
-      "learning_rate": 1.3999802348921993e-05,
-      "loss": 0.0036,
+      "epoch": 1.2464,
+      "grad_norm": 0.14300068454824735,
+      "learning_rate": 1.3175504581708261e-05,
+      "loss": 0.0035,
       "step": 779
     },
     {
-      "epoch": 1.1676646706586826,
-      "grad_norm": 0.10329869929004334,
-      "learning_rate": 1.3984978062067836e-05,
-      "loss": 0.0008,
+      "epoch": 1.248,
+      "grad_norm": 0.007568948806440808,
+      "learning_rate": 1.3159113775218963e-05,
+      "loss": 0.0001,
       "step": 780
     },
     {
-      "epoch": 1.1691616766467066,
-      "grad_norm": 0.16774030611607113,
-      "learning_rate": 1.3970143357318475e-05,
-      "loss": 0.0025,
+      "epoch": 1.2496,
+      "grad_norm": 0.019533933851628504,
+      "learning_rate": 1.3142713535136413e-05,
+      "loss": 0.0002,
       "step": 781
     },
     {
-      "epoch": 1.1706586826347305,
-      "grad_norm": 0.10032247067801155,
-      "learning_rate": 1.3955298273456157e-05,
-      "loss": 0.0008,
+      "epoch": 1.2511999999999999,
+      "grad_norm": 0.01251896381864414,
+      "learning_rate": 1.3126303910434215e-05,
+      "loss": 0.0002,
       "step": 782
     },
     {
-      "epoch": 1.1721556886227544,
-      "grad_norm": 0.12720820030778054,
-      "learning_rate": 1.3940442849290259e-05,
-      "loss": 0.0028,
+      "epoch": 1.2528000000000001,
+      "grad_norm": 0.17193764796753144,
+      "learning_rate": 1.3109884950114007e-05,
+      "loss": 0.0017,
       "step": 783
     },
     {
-      "epoch": 1.1736526946107784,
-      "grad_norm": 0.07501276893929008,
-      "learning_rate": 1.3925577123657196e-05,
-      "loss": 0.0013,
+      "epoch": 1.2544,
+      "grad_norm": 0.015917363245241088,
+      "learning_rate": 1.309345670320529e-05,
+      "loss": 0.0001,
       "step": 784
     },
     {
-      "epoch": 1.1751497005988023,
-      "grad_norm": 0.13032678151078575,
-      "learning_rate": 1.3910701135420312e-05,
-      "loss": 0.0014,
+      "epoch": 1.256,
+      "grad_norm": 0.007055309334246423,
+      "learning_rate": 1.3077019218765306e-05,
+      "loss": 0.0001,
       "step": 785
     },
     {
-      "epoch": 1.1766467065868262,
-      "grad_norm": 0.1189117861891301,
-      "learning_rate": 1.3895814923469781e-05,
-      "loss": 0.003,
+      "epoch": 1.2576,
+      "grad_norm": 0.01697671259128701,
+      "learning_rate": 1.3060572545878875e-05,
+      "loss": 0.0002,
       "step": 786
     },
     {
-      "epoch": 1.1781437125748502,
-      "grad_norm": 0.04852723768916087,
-      "learning_rate": 1.3880918526722497e-05,
-      "loss": 0.0006,
+      "epoch": 1.2591999999999999,
+      "grad_norm": 0.05641228049039576,
+      "learning_rate": 1.3044116733658261e-05,
+      "loss": 0.0002,
       "step": 787
     },
     {
-      "epoch": 1.1796407185628743,
-      "grad_norm": 0.14044921780427475,
-      "learning_rate": 1.3866011984121997e-05,
-      "loss": 0.0016,
+      "epoch": 1.2608,
+      "grad_norm": 0.0706754675654207,
+      "learning_rate": 1.302765183124302e-05,
+      "loss": 0.0023,
       "step": 788
     },
     {
-      "epoch": 1.1811377245508983,
-      "grad_norm": 0.10503085747024221,
-      "learning_rate": 1.3851095334638323e-05,
-      "loss": 0.0013,
+      "epoch": 1.2624,
+      "grad_norm": 0.013578635017518218,
+      "learning_rate": 1.3011177887799846e-05,
+      "loss": 0.0002,
       "step": 789
     },
     {
-      "epoch": 1.1826347305389222,
-      "grad_norm": 0.07687725843746662,
-      "learning_rate": 1.3836168617267956e-05,
-      "loss": 0.0007,
+      "epoch": 1.264,
+      "grad_norm": 0.006322359265803324,
+      "learning_rate": 1.2994694952522435e-05,
+      "loss": 0.0001,
       "step": 790
     },
     {
-      "epoch": 1.1841317365269461,
-      "grad_norm": 0.13288540054060996,
-      "learning_rate": 1.3821231871033684e-05,
-      "loss": 0.0019,
+      "epoch": 1.2656,
+      "grad_norm": 0.024743663928371313,
+      "learning_rate": 1.2978203074631335e-05,
+      "loss": 0.0002,
       "step": 791
     },
     {
-      "epoch": 1.18562874251497,
-      "grad_norm": 0.10408705042608069,
-      "learning_rate": 1.3806285134984525e-05,
-      "loss": 0.0015,
+      "epoch": 1.2671999999999999,
+      "grad_norm": 0.010036392053070036,
+      "learning_rate": 1.2961702303373795e-05,
+      "loss": 0.0001,
       "step": 792
     },
     {
-      "epoch": 1.187125748502994,
-      "grad_norm": 0.09545112906104483,
-      "learning_rate": 1.37913284481956e-05,
-      "loss": 0.0013,
+      "epoch": 1.2688,
+      "grad_norm": 0.059398968783332944,
+      "learning_rate": 1.2945192688023625e-05,
+      "loss": 0.0003,
       "step": 793
     },
     {
-      "epoch": 1.188622754491018,
-      "grad_norm": 0.19426107849435,
-      "learning_rate": 1.3776361849768056e-05,
-      "loss": 0.002,
+      "epoch": 1.2704,
+      "grad_norm": 0.1360718622443351,
+      "learning_rate": 1.2928674277881041e-05,
+      "loss": 0.0006,
       "step": 794
     },
     {
-      "epoch": 1.1901197604790419,
-      "grad_norm": 0.1508486355195202,
-      "learning_rate": 1.3761385378828946e-05,
-      "loss": 0.0016,
+      "epoch": 1.272,
+      "grad_norm": 0.006056760144497216,
+      "learning_rate": 1.2912147122272523e-05,
+      "loss": 0.0001,
       "step": 795
     },
     {
-      "epoch": 1.1916167664670658,
-      "grad_norm": 0.11380042520519763,
-      "learning_rate": 1.3746399074531139e-05,
-      "loss": 0.0012,
+      "epoch": 1.2736,
+      "grad_norm": 0.0031045773593651495,
+      "learning_rate": 1.2895611270550666e-05,
+      "loss": 0.0001,
       "step": 796
     },
     {
-      "epoch": 1.1931137724550898,
-      "grad_norm": 0.06272949618809354,
-      "learning_rate": 1.3731402976053203e-05,
-      "loss": 0.0009,
+      "epoch": 1.2752,
+      "grad_norm": 0.013723013938490469,
+      "learning_rate": 1.287906677209403e-05,
+      "loss": 0.0001,
       "step": 797
     },
     {
-      "epoch": 1.1946107784431137,
-      "grad_norm": 0.04908194402100617,
-      "learning_rate": 1.3716397122599311e-05,
-      "loss": 0.0004,
+      "epoch": 1.2768,
+      "grad_norm": 0.11625251974105492,
+      "learning_rate": 1.2862513676307009e-05,
+      "loss": 0.0001,
       "step": 798
     },
     {
-      "epoch": 1.1961077844311376,
-      "grad_norm": 0.04835379010726109,
-      "learning_rate": 1.3701381553399147e-05,
-      "loss": 0.0005,
+      "epoch": 1.2784,
+      "grad_norm": 0.023632751991143754,
+      "learning_rate": 1.2845952032619651e-05,
+      "loss": 0.0001,
       "step": 799
     },
     {
-      "epoch": 1.1976047904191618,
-      "grad_norm": 0.07282438701791324,
-      "learning_rate": 1.3686356307707784e-05,
-      "loss": 0.0011,
+      "epoch": 1.28,
+      "grad_norm": 0.019534406451617272,
+      "learning_rate": 1.2829381890487536e-05,
+      "loss": 0.0002,
       "step": 800
     },
     {
-      "epoch": 1.1991017964071857,
-      "grad_norm": 0.07359224976031348,
-      "learning_rate": 1.3671321424805602e-05,
+      "epoch": 1.2816,
+      "grad_norm": 0.13935122782766068,
+      "learning_rate": 1.2812803299391629e-05,
       "loss": 0.0008,
       "step": 801
     },
     {
-      "epoch": 1.2005988023952097,
-      "grad_norm": 0.0728588455334413,
-      "learning_rate": 1.365627694399817e-05,
-      "loss": 0.0011,
+      "epoch": 1.2832,
+      "grad_norm": 0.008055208167228007,
+      "learning_rate": 1.2796216308838116e-05,
+      "loss": 0.0001,
       "step": 802
     },
     {
-      "epoch": 1.2020958083832336,
-      "grad_norm": 0.08572484612904316,
-      "learning_rate": 1.3641222904616149e-05,
-      "loss": 0.0019,
+      "epoch": 1.2848,
+      "grad_norm": 0.01074330815404068,
+      "learning_rate": 1.2779620968358276e-05,
+      "loss": 0.0001,
       "step": 803
     },
     {
-      "epoch": 1.2035928143712575,
-      "grad_norm": 0.10340010986651034,
-      "learning_rate": 1.3626159346015193e-05,
-      "loss": 0.0009,
+      "epoch": 1.2864,
+      "grad_norm": 0.04719274161820858,
+      "learning_rate": 1.2763017327508304e-05,
+      "loss": 0.0003,
       "step": 804
     },
     {
-      "epoch": 1.2050898203592815,
-      "grad_norm": 0.07243035348418118,
-      "learning_rate": 1.3611086307575831e-05,
-      "loss": 0.0008,
+      "epoch": 1.288,
+      "grad_norm": 0.10415338321560368,
+      "learning_rate": 1.2746405435869198e-05,
+      "loss": 0.0022,
       "step": 805
     },
     {
-      "epoch": 1.2065868263473054,
-      "grad_norm": 0.1334209046458507,
-      "learning_rate": 1.3596003828703393e-05,
-      "loss": 0.0019,
+      "epoch": 1.2896,
+      "grad_norm": 0.011459211797809638,
+      "learning_rate": 1.2729785343046587e-05,
+      "loss": 0.0001,
       "step": 806
     },
     {
-      "epoch": 1.2080838323353293,
-      "grad_norm": 0.0570772989713065,
-      "learning_rate": 1.3580911948827868e-05,
-      "loss": 0.001,
+      "epoch": 1.2912,
+      "grad_norm": 0.0036652103191051693,
+      "learning_rate": 1.271315709867059e-05,
+      "loss": 0.0001,
       "step": 807
     },
     {
-      "epoch": 1.2095808383233533,
-      "grad_norm": 0.1204536054401414,
-      "learning_rate": 1.3565810707403844e-05,
-      "loss": 0.0009,
+      "epoch": 1.2928,
+      "grad_norm": 0.0025570224469476416,
+      "learning_rate": 1.2696520752395671e-05,
+      "loss": 0.0001,
       "step": 808
     },
     {
-      "epoch": 1.2110778443113772,
-      "grad_norm": 0.1867883260775183,
-      "learning_rate": 1.3550700143910365e-05,
-      "loss": 0.0025,
+      "epoch": 1.2944,
+      "grad_norm": 0.0036595771184137252,
+      "learning_rate": 1.2679876353900482e-05,
+      "loss": 0.0001,
       "step": 809
     },
     {
-      "epoch": 1.2125748502994012,
-      "grad_norm": 0.07727972942708328,
-      "learning_rate": 1.3535580297850852e-05,
-      "loss": 0.0007,
+      "epoch": 1.296,
+      "grad_norm": 0.11645277140286601,
+      "learning_rate": 1.2663223952887724e-05,
+      "loss": 0.0008,
       "step": 810
     },
     {
-      "epoch": 1.214071856287425,
-      "grad_norm": 0.09466637885042935,
-      "learning_rate": 1.3520451208752992e-05,
-      "loss": 0.0005,
+      "epoch": 1.2976,
+      "grad_norm": 0.0041716206625464205,
+      "learning_rate": 1.2646563599083997e-05,
+      "loss": 0.0001,
       "step": 811
     },
     {
-      "epoch": 1.215568862275449,
-      "grad_norm": 0.1580780788542899,
-      "learning_rate": 1.3505312916168644e-05,
-      "loss": 0.001,
+      "epoch": 1.2992,
+      "grad_norm": 0.028390432744085866,
+      "learning_rate": 1.2629895342239643e-05,
+      "loss": 0.0002,
       "step": 812
     },
     {
-      "epoch": 1.217065868263473,
-      "grad_norm": 0.10526784028411676,
-      "learning_rate": 1.3490165459673716e-05,
+      "epoch": 1.3008,
+      "grad_norm": 0.08071690095947365,
+      "learning_rate": 1.2613219232128608e-05,
       "loss": 0.001,
       "step": 813
     },
     {
-      "epoch": 1.218562874251497,
-      "grad_norm": 0.12653380982480567,
-      "learning_rate": 1.3475008878868078e-05,
-      "loss": 0.0013,
+      "epoch": 1.3024,
+      "grad_norm": 0.10505099790295679,
+      "learning_rate": 1.2596535318548288e-05,
+      "loss": 0.0011,
       "step": 814
     },
     {
-      "epoch": 1.220059880239521,
-      "grad_norm": 0.069894431976694,
-      "learning_rate": 1.345984321337546e-05,
-      "loss": 0.0014,
+      "epoch": 1.304,
+      "grad_norm": 0.0049778696815799125,
+      "learning_rate": 1.2579843651319382e-05,
+      "loss": 0.0001,
       "step": 815
     },
     {
-      "epoch": 1.221556886227545,
-      "grad_norm": 0.11138606082666604,
-      "learning_rate": 1.344466850284333e-05,
-      "loss": 0.001,
+      "epoch": 1.3056,
+      "grad_norm": 0.0043848840789224,
+      "learning_rate": 1.2563144280285742e-05,
+      "loss": 0.0001,
       "step": 816
     },
     {
-      "epoch": 1.223053892215569,
-      "grad_norm": 0.09549692513609892,
-      "learning_rate": 1.342948478694281e-05,
-      "loss": 0.0012,
+      "epoch": 1.3072,
+      "grad_norm": 0.020550136184270512,
+      "learning_rate": 1.2546437255314223e-05,
+      "loss": 0.0001,
       "step": 817
     },
     {
-      "epoch": 1.2245508982035929,
-      "grad_norm": 0.08071969012117705,
-      "learning_rate": 1.3414292105368562e-05,
-      "loss": 0.0011,
+      "epoch": 1.3088,
+      "grad_norm": 0.004432220253725601,
+      "learning_rate": 1.252972262629454e-05,
+      "loss": 0.0001,
       "step": 818
     },
     {
-      "epoch": 1.2260479041916168,
-      "grad_norm": 0.014969590068211934,
-      "learning_rate": 1.339909049783869e-05,
-      "loss": 0.0002,
+      "epoch": 1.3104,
+      "grad_norm": 0.01302457137434428,
+      "learning_rate": 1.2513000443139112e-05,
+      "loss": 0.0001,
       "step": 819
     },
     {
-      "epoch": 1.2275449101796407,
-      "grad_norm": 0.07241700618747784,
-      "learning_rate": 1.338388000409463e-05,
-      "loss": 0.0005,
+      "epoch": 1.312,
+      "grad_norm": 0.21882702602859708,
+      "learning_rate": 1.2496270755782913e-05,
+      "loss": 0.0045,
       "step": 820
     },
     {
-      "epoch": 1.2290419161676647,
-      "grad_norm": 0.11075869601766639,
-      "learning_rate": 1.3368660663901045e-05,
-      "loss": 0.0023,
+      "epoch": 1.3136,
+      "grad_norm": 0.011649467674618524,
+      "learning_rate": 1.2479533614183334e-05,
+      "loss": 0.0001,
       "step": 821
     },
     {
-      "epoch": 1.2305389221556886,
-      "grad_norm": 0.17060039325631357,
-      "learning_rate": 1.3353432517045739e-05,
-      "loss": 0.0016,
+      "epoch": 1.3152,
+      "grad_norm": 0.04260599118754293,
+      "learning_rate": 1.2462789068320016e-05,
+      "loss": 0.0003,
       "step": 822
     },
     {
-      "epoch": 1.2320359281437125,
-      "grad_norm": 0.1022635862904923,
-      "learning_rate": 1.3338195603339525e-05,
-      "loss": 0.0011,
+      "epoch": 1.3168,
+      "grad_norm": 0.009341435049419606,
+      "learning_rate": 1.2446037168194716e-05,
+      "loss": 0.0001,
       "step": 823
     },
     {
-      "epoch": 1.2335329341317365,
-      "grad_norm": 0.09158836544730396,
-      "learning_rate": 1.3322949962616142e-05,
-      "loss": 0.0006,
+      "epoch": 1.3184,
+      "grad_norm": 0.014751816702529163,
+      "learning_rate": 1.2429277963831147e-05,
+      "loss": 0.0002,
       "step": 824
     },
     {
-      "epoch": 1.2350299401197604,
-      "grad_norm": 0.048832084216327266,
-      "learning_rate": 1.3307695634732144e-05,
-      "loss": 0.0005,
+      "epoch": 1.32,
+      "grad_norm": 0.008743327925641906,
+      "learning_rate": 1.2412511505274845e-05,
+      "loss": 0.0001,
       "step": 825
     },
     {
-      "epoch": 1.2365269461077844,
-      "grad_norm": 0.08830232598087197,
-      "learning_rate": 1.3292432659566791e-05,
-      "loss": 0.0007,
+      "epoch": 1.3216,
+      "grad_norm": 0.003562158556219341,
+      "learning_rate": 1.2395737842592997e-05,
+      "loss": 0.0001,
       "step": 826
     },
     {
-      "epoch": 1.2380239520958083,
-      "grad_norm": 0.052016398871730235,
-      "learning_rate": 1.3277161077021957e-05,
-      "loss": 0.0005,
+      "epoch": 1.3232,
+      "grad_norm": 0.13376418839075693,
+      "learning_rate": 1.23789570258743e-05,
+      "loss": 0.0019,
       "step": 827
     },
     {
-      "epoch": 1.2395209580838324,
-      "grad_norm": 0.12716710033758089,
-      "learning_rate": 1.3261880927022008e-05,
-      "loss": 0.001,
+      "epoch": 1.3248,
+      "grad_norm": 0.019777762410234753,
+      "learning_rate": 1.2362169105228828e-05,
+      "loss": 0.0002,
       "step": 828
     },
     {
-      "epoch": 1.2410179640718564,
-      "grad_norm": 0.08239866964993203,
-      "learning_rate": 1.3246592249513717e-05,
-      "loss": 0.0008,
+      "epoch": 1.3264,
+      "grad_norm": 0.006502459797443767,
+      "learning_rate": 1.2345374130787855e-05,
+      "loss": 0.0001,
       "step": 829
     },
     {
-      "epoch": 1.2425149700598803,
-      "grad_norm": 0.07577545667183459,
-      "learning_rate": 1.3231295084466149e-05,
-      "loss": 0.0008,
+      "epoch": 1.328,
+      "grad_norm": 0.009005732145788963,
+      "learning_rate": 1.2328572152703726e-05,
+      "loss": 0.0001,
       "step": 830
     },
     {
-      "epoch": 1.2440119760479043,
-      "grad_norm": 0.08810579822674741,
-      "learning_rate": 1.3215989471870556e-05,
-      "loss": 0.0009,
+      "epoch": 1.3296000000000001,
+      "grad_norm": 0.003715792304581145,
+      "learning_rate": 1.23117632211497e-05,
+      "loss": 0.0001,
       "step": 831
     },
     {
-      "epoch": 1.2455089820359282,
-      "grad_norm": 0.04066723916071258,
-      "learning_rate": 1.3200675451740273e-05,
-      "loss": 0.0006,
+      "epoch": 1.3312,
+      "grad_norm": 0.01923160629987508,
+      "learning_rate": 1.2294947386319793e-05,
+      "loss": 0.0001,
       "step": 832
     },
     {
-      "epoch": 1.2470059880239521,
-      "grad_norm": 0.14239069700954057,
-      "learning_rate": 1.3185353064110613e-05,
-      "loss": 0.0022,
+      "epoch": 1.3328,
+      "grad_norm": 0.060602109689933376,
+      "learning_rate": 1.2278124698428643e-05,
+      "loss": 0.0004,
       "step": 833
     },
     {
-      "epoch": 1.248502994011976,
-      "grad_norm": 0.09145370992887267,
-      "learning_rate": 1.3170022349038774e-05,
-      "loss": 0.0011,
+      "epoch": 1.3344,
+      "grad_norm": 0.05030831181536448,
+      "learning_rate": 1.2261295207711347e-05,
+      "loss": 0.0003,
       "step": 834
     },
     {
-      "epoch": 1.25,
-      "grad_norm": 0.09278110626604044,
-      "learning_rate": 1.3154683346603712e-05,
-      "loss": 0.001,
+      "epoch": 1.336,
+      "grad_norm": 0.07963954964576811,
+      "learning_rate": 1.2244458964423328e-05,
+      "loss": 0.0028,
       "step": 835
     },
     {
-      "epoch": 1.251497005988024,
-      "grad_norm": 0.04046285571674997,
-      "learning_rate": 1.3139336096906056e-05,
-      "loss": 0.0003,
+      "epoch": 1.3376000000000001,
+      "grad_norm": 0.05077621892276567,
+      "learning_rate": 1.2227616018840154e-05,
+      "loss": 0.0002,
       "step": 836
     },
     {
-      "epoch": 1.2529940119760479,
-      "grad_norm": 0.1350937599331085,
-      "learning_rate": 1.3123980640067999e-05,
-      "loss": 0.0018,
+      "epoch": 1.3392,
+      "grad_norm": 0.023897944724496852,
+      "learning_rate": 1.221076642125742e-05,
+      "loss": 0.0001,
       "step": 837
     },
     {
-      "epoch": 1.2544910179640718,
-      "grad_norm": 0.05547862227838062,
-      "learning_rate": 1.3108617016233178e-05,
-      "loss": 0.0005,
+      "epoch": 1.3408,
+      "grad_norm": 0.024814131724707863,
+      "learning_rate": 1.2193910221990582e-05,
+      "loss": 0.0002,
       "step": 838
     },
     {
-      "epoch": 1.2559880239520957,
-      "grad_norm": 0.13832547458088876,
-      "learning_rate": 1.3093245265566588e-05,
-      "loss": 0.0021,
+      "epoch": 1.3424,
+      "grad_norm": 0.07223740891735303,
+      "learning_rate": 1.2177047471374808e-05,
+      "loss": 0.0005,
       "step": 839
     },
     {
-      "epoch": 1.2574850299401197,
-      "grad_norm": 0.3205689356680666,
-      "learning_rate": 1.3077865428254474e-05,
-      "loss": 0.0017,
+      "epoch": 1.3439999999999999,
+      "grad_norm": 0.026021812742085374,
+      "learning_rate": 1.2160178219764838e-05,
+      "loss": 0.0002,
       "step": 840
     },
     {
-      "epoch": 1.2589820359281436,
-      "grad_norm": 0.17333145605990935,
-      "learning_rate": 1.3062477544504218e-05,
-      "loss": 0.0028,
+      "epoch": 1.3456000000000001,
+      "grad_norm": 0.007538083846824722,
+      "learning_rate": 1.214330251753481e-05,
+      "loss": 0.0001,
       "step": 841
     },
     {
-      "epoch": 1.2604790419161676,
-      "grad_norm": 0.10905825753636433,
-      "learning_rate": 1.304708165454423e-05,
-      "loss": 0.001,
+      "epoch": 1.3472,
+      "grad_norm": 0.0032914494744931025,
+      "learning_rate": 1.2126420415078133e-05,
+      "loss": 0.0001,
       "step": 842
     },
     {
-      "epoch": 1.2619760479041915,
-      "grad_norm": 0.06230478352428612,
-      "learning_rate": 1.3031677798623865e-05,
-      "loss": 0.0005,
+      "epoch": 1.3488,
+      "grad_norm": 0.00686056490409744,
+      "learning_rate": 1.2109531962807333e-05,
+      "loss": 0.0001,
       "step": 843
     },
     {
-      "epoch": 1.2634730538922156,
-      "grad_norm": 0.15076033232514022,
-      "learning_rate": 1.3016266017013295e-05,
-      "loss": 0.0015,
+      "epoch": 1.3504,
+      "grad_norm": 0.08422612909094722,
+      "learning_rate": 1.2092637211153885e-05,
+      "loss": 0.0002,
       "step": 844
     },
     {
-      "epoch": 1.2649700598802396,
-      "grad_norm": 0.10330789403132279,
-      "learning_rate": 1.300084635000341e-05,
-      "loss": 0.0013,
+      "epoch": 1.3519999999999999,
+      "grad_norm": 0.0859550009021419,
+      "learning_rate": 1.207573621056809e-05,
+      "loss": 0.0003,
       "step": 845
     },
     {
-      "epoch": 1.2664670658682635,
-      "grad_norm": 0.0967360202578523,
-      "learning_rate": 1.2985418837905719e-05,
-      "loss": 0.0009,
+      "epoch": 1.3536000000000001,
+      "grad_norm": 0.012894853612493646,
+      "learning_rate": 1.2058829011518896e-05,
+      "loss": 0.0001,
       "step": 846
     },
     {
-      "epoch": 1.2679640718562875,
-      "grad_norm": 0.11018485595492865,
-      "learning_rate": 1.2969983521052243e-05,
-      "loss": 0.0015,
+      "epoch": 1.3552,
+      "grad_norm": 0.01038140375447721,
+      "learning_rate": 1.2041915664493763e-05,
+      "loss": 0.0001,
       "step": 847
     },
     {
-      "epoch": 1.2694610778443114,
-      "grad_norm": 0.2026144990686223,
-      "learning_rate": 1.29545404397954e-05,
-      "loss": 0.0016,
+      "epoch": 1.3568,
+      "grad_norm": 0.00782320648496516,
+      "learning_rate": 1.2024996219998517e-05,
+      "loss": 0.0001,
       "step": 848
     },
     {
-      "epoch": 1.2709580838323353,
-      "grad_norm": 0.11825794065351451,
-      "learning_rate": 1.2939089634507908e-05,
-      "loss": 0.0009,
+      "epoch": 1.3584,
+      "grad_norm": 0.016092547758833026,
+      "learning_rate": 1.2008070728557186e-05,
+      "loss": 0.0001,
       "step": 849
     },
     {
-      "epoch": 1.2724550898203593,
-      "grad_norm": 0.1706730641796369,
-      "learning_rate": 1.2923631145582684e-05,
-      "loss": 0.0026,
+      "epoch": 1.3599999999999999,
+      "grad_norm": 0.007446362479705341,
+      "learning_rate": 1.1991139240711857e-05,
+      "loss": 0.0001,
       "step": 850
     },
     {
-      "epoch": 1.2739520958083832,
-      "grad_norm": 0.057699428000987554,
-      "learning_rate": 1.2908165013432723e-05,
-      "loss": 0.0008,
+      "epoch": 1.3616,
+      "grad_norm": 0.01801782176947607,
+      "learning_rate": 1.1974201807022525e-05,
+      "loss": 0.0001,
       "step": 851
     },
     {
-      "epoch": 1.2754491017964071,
-      "grad_norm": 0.142393242455963,
-      "learning_rate": 1.2892691278491012e-05,
-      "loss": 0.0017,
+      "epoch": 1.3632,
+      "grad_norm": 0.11088844118694781,
+      "learning_rate": 1.195725847806693e-05,
+      "loss": 0.0004,
       "step": 852
     },
     {
-      "epoch": 1.276946107784431,
-      "grad_norm": 0.055031594469559006,
-      "learning_rate": 1.2877209981210405e-05,
-      "loss": 0.0004,
+      "epoch": 1.3648,
+      "grad_norm": 0.05072198627653064,
+      "learning_rate": 1.1940309304440434e-05,
+      "loss": 0.0003,
       "step": 853
     },
     {
-      "epoch": 1.278443113772455,
-      "grad_norm": 0.0849339018863119,
-      "learning_rate": 1.2861721162063532e-05,
-      "loss": 0.0004,
+      "epoch": 1.3664,
+      "grad_norm": 0.007082916498430339,
+      "learning_rate": 1.1923354336755835e-05,
+      "loss": 0.0001,
       "step": 854
     },
     {
-      "epoch": 1.2799401197604792,
-      "grad_norm": 0.07827653949622694,
-      "learning_rate": 1.2846224861542677e-05,
-      "loss": 0.0007,
+      "epoch": 1.3679999999999999,
+      "grad_norm": 0.0024532472327597767,
+      "learning_rate": 1.1906393625643244e-05,
+      "loss": 0.0001,
       "step": 855
     },
     {
-      "epoch": 1.281437125748503,
-      "grad_norm": 0.024359597454629765,
-      "learning_rate": 1.2830721120159697e-05,
-      "loss": 0.0003,
+      "epoch": 1.3696,
+      "grad_norm": 0.03561823023476568,
+      "learning_rate": 1.1889427221749916e-05,
+      "loss": 0.0002,
       "step": 856
     },
     {
-      "epoch": 1.282934131736527,
-      "grad_norm": 0.21035448274091229,
-      "learning_rate": 1.2815209978445898e-05,
-      "loss": 0.0011,
+      "epoch": 1.3712,
+      "grad_norm": 0.011999726938913933,
+      "learning_rate": 1.1872455175740111e-05,
+      "loss": 0.0001,
       "step": 857
     },
     {
-      "epoch": 1.284431137724551,
-      "grad_norm": 0.07397518183430371,
-      "learning_rate": 1.2799691476951926e-05,
-      "loss": 0.0007,
+      "epoch": 1.3728,
+      "grad_norm": 0.03409171465268027,
+      "learning_rate": 1.1855477538294934e-05,
+      "loss": 0.0002,
       "step": 858
     },
     {
-      "epoch": 1.285928143712575,
-      "grad_norm": 0.13982310084632887,
-      "learning_rate": 1.2784165656247668e-05,
-      "loss": 0.001,
+      "epoch": 1.3744,
+      "grad_norm": 0.01687001729393258,
+      "learning_rate": 1.1838494360112185e-05,
+      "loss": 0.0001,
       "step": 859
     },
     {
-      "epoch": 1.2874251497005988,
-      "grad_norm": 0.029670153139229417,
-      "learning_rate": 1.2768632556922151e-05,
-      "loss": 0.0004,
+      "epoch": 1.376,
+      "grad_norm": 0.08714127876212839,
+      "learning_rate": 1.1821505691906216e-05,
+      "loss": 0.0003,
       "step": 860
     },
     {
-      "epoch": 1.2889221556886228,
-      "grad_norm": 0.08714604279500297,
-      "learning_rate": 1.275309221958343e-05,
-      "loss": 0.0012,
+      "epoch": 1.3776,
+      "grad_norm": 0.030346824461753623,
+      "learning_rate": 1.1804511584407763e-05,
+      "loss": 0.0002,
       "step": 861
     },
     {
-      "epoch": 1.2904191616766467,
-      "grad_norm": 0.02766702455992605,
-      "learning_rate": 1.2737544684858476e-05,
-      "loss": 0.0003,
+      "epoch": 1.3792,
+      "grad_norm": 0.06417791238526656,
+      "learning_rate": 1.1787512088363817e-05,
+      "loss": 0.0002,
       "step": 862
     },
     {
-      "epoch": 1.2919161676646707,
-      "grad_norm": 0.15764398213987413,
-      "learning_rate": 1.2721989993393088e-05,
-      "loss": 0.0021,
+      "epoch": 1.3808,
+      "grad_norm": 0.1100014244959121,
+      "learning_rate": 1.1770507254537454e-05,
+      "loss": 0.0009,
       "step": 863
     },
     {
-      "epoch": 1.2934131736526946,
-      "grad_norm": 0.3620928680238313,
-      "learning_rate": 1.2706428185851759e-05,
-      "loss": 0.0031,
+      "epoch": 1.3824,
+      "grad_norm": 0.003659563707803467,
+      "learning_rate": 1.1753497133707678e-05,
+      "loss": 0.0001,
       "step": 864
     },
     {
-      "epoch": 1.2949101796407185,
-      "grad_norm": 0.15033902468169197,
-      "learning_rate": 1.26908593029176e-05,
-      "loss": 0.0017,
+      "epoch": 1.384,
+      "grad_norm": 0.10925148443368389,
+      "learning_rate": 1.1736481776669307e-05,
+      "loss": 0.0014,
       "step": 865
     },
     {
-      "epoch": 1.2964071856287425,
-      "grad_norm": 0.08093325547150344,
-      "learning_rate": 1.2675283385292212e-05,
-      "loss": 0.0006,
+      "epoch": 1.3856,
+      "grad_norm": 0.05504213576089758,
+      "learning_rate": 1.1719461234232765e-05,
+      "loss": 0.0004,
       "step": 866
     },
     {
-      "epoch": 1.2979041916167664,
-      "grad_norm": 0.19391865613837045,
-      "learning_rate": 1.2659700473695581e-05,
-      "loss": 0.0026,
+      "epoch": 1.3872,
+      "grad_norm": 0.002039500612383281,
+      "learning_rate": 1.1702435557223988e-05,
+      "loss": 0.0001,
       "step": 867
     },
     {
-      "epoch": 1.2994011976047903,
-      "grad_norm": 0.13771485536341505,
-      "learning_rate": 1.2644110608865994e-05,
-      "loss": 0.002,
+      "epoch": 1.3888,
+      "grad_norm": 0.08673954375056128,
+      "learning_rate": 1.1685404796484226e-05,
+      "loss": 0.0007,
       "step": 868
     },
     {
-      "epoch": 1.3008982035928143,
-      "grad_norm": 0.082036361858331,
-      "learning_rate": 1.2628513831559897e-05,
-      "loss": 0.0012,
+      "epoch": 1.3904,
+      "grad_norm": 0.01799475637745854,
+      "learning_rate": 1.1668369002869912e-05,
+      "loss": 0.0001,
       "step": 869
     },
     {
-      "epoch": 1.3023952095808382,
-      "grad_norm": 0.13557425572078727,
-      "learning_rate": 1.2612910182551822e-05,
-      "loss": 0.0009,
+      "epoch": 1.392,
+      "grad_norm": 0.006362603924743877,
+      "learning_rate": 1.1651328227252516e-05,
+      "loss": 0.0001,
       "step": 870
     },
     {
-      "epoch": 1.3038922155688621,
-      "grad_norm": 0.1248956204266484,
-      "learning_rate": 1.259729970263426e-05,
-      "loss": 0.0012,
+      "epoch": 1.3936,
+      "grad_norm": 0.0016838244758269412,
+      "learning_rate": 1.1634282520518382e-05,
+      "loss": 0.0001,
       "step": 871
     },
     {
-      "epoch": 1.3053892215568863,
-      "grad_norm": 0.15323443985781152,
-      "learning_rate": 1.2581682432617554e-05,
-      "loss": 0.0021,
+      "epoch": 1.3952,
+      "grad_norm": 0.011845117010406862,
+      "learning_rate": 1.1617231933568579e-05,
+      "loss": 0.0001,
       "step": 872
     },
     {
-      "epoch": 1.3068862275449102,
-      "grad_norm": 0.16495520023315455,
-      "learning_rate": 1.2566058413329806e-05,
-      "loss": 0.0011,
+      "epoch": 1.3968,
+      "grad_norm": 0.11842164258071569,
+      "learning_rate": 1.1600176517318742e-05,
+      "loss": 0.0014,
       "step": 873
     },
     {
-      "epoch": 1.3083832335329342,
-      "grad_norm": 0.12612540137072323,
-      "learning_rate": 1.2550427685616767e-05,
-      "loss": 0.0019,
+      "epoch": 1.3984,
+      "grad_norm": 0.001944117806670603,
+      "learning_rate": 1.1583116322698936e-05,
+      "loss": 0.0001,
       "step": 874
     },
     {
-      "epoch": 1.3098802395209581,
-      "grad_norm": 0.07140252385386611,
-      "learning_rate": 1.253479029034171e-05,
-      "loss": 0.0009,
+      "epoch": 1.4,
+      "grad_norm": 0.006907884014797453,
+      "learning_rate": 1.1566051400653486e-05,
+      "loss": 0.0001,
       "step": 875
     },
     {
-      "epoch": 1.311377245508982,
-      "grad_norm": 0.07997999603836409,
-      "learning_rate": 1.2519146268385352e-05,
-      "loss": 0.0012,
+      "epoch": 1.4016,
+      "grad_norm": 0.046163014290470226,
+      "learning_rate": 1.1548981802140849e-05,
+      "loss": 0.0002,
       "step": 876
     },
     {
-      "epoch": 1.312874251497006,
-      "grad_norm": 0.05031120608143949,
-      "learning_rate": 1.250349566064573e-05,
-      "loss": 0.0007,
+      "epoch": 1.4032,
+      "grad_norm": 0.08173968469395558,
+      "learning_rate": 1.153190757813343e-05,
+      "loss": 0.0002,
       "step": 877
     },
     {
-      "epoch": 1.31437125748503,
-      "grad_norm": 0.10409482982213064,
-      "learning_rate": 1.2487838508038097e-05,
-      "loss": 0.0018,
+      "epoch": 1.4048,
+      "grad_norm": 0.009413722377722337,
+      "learning_rate": 1.151482877961746e-05,
+      "loss": 0.0001,
       "step": 878
     },
     {
-      "epoch": 1.3158682634730539,
-      "grad_norm": 0.06629243450722555,
-      "learning_rate": 1.2472174851494823e-05,
-      "loss": 0.0005,
+      "epoch": 1.4064,
+      "grad_norm": 0.007313756445064559,
+      "learning_rate": 1.1497745457592817e-05,
+      "loss": 0.0001,
       "step": 879
     },
     {
-      "epoch": 1.3173652694610778,
-      "grad_norm": 0.11971936347458219,
-      "learning_rate": 1.2456504731965269e-05,
-      "loss": 0.0017,
+      "epoch": 1.408,
+      "grad_norm": 0.08301497427474906,
+      "learning_rate": 1.1480657663072896e-05,
+      "loss": 0.0006,
       "step": 880
     },
     {
-      "epoch": 1.3188622754491017,
-      "grad_norm": 0.20319873175607275,
-      "learning_rate": 1.2440828190415706e-05,
-      "loss": 0.0018,
+      "epoch": 1.4096,
+      "grad_norm": 0.005428105104145994,
+      "learning_rate": 1.1463565447084446e-05,
+      "loss": 0.0001,
       "step": 881
     },
     {
-      "epoch": 1.3203592814371259,
-      "grad_norm": 0.04682707557262463,
-      "learning_rate": 1.2425145267829179e-05,
-      "loss": 0.0005,
+      "epoch": 1.4112,
+      "grad_norm": 0.007285758503868084,
+      "learning_rate": 1.1446468860667422e-05,
+      "loss": 0.0001,
       "step": 882
     },
     {
-      "epoch": 1.3218562874251498,
-      "grad_norm": 0.3848522460027514,
-      "learning_rate": 1.2409456005205427e-05,
-      "loss": 0.0017,
+      "epoch": 1.4128,
+      "grad_norm": 0.10521185311428458,
+      "learning_rate": 1.142936795487482e-05,
+      "loss": 0.0004,
       "step": 883
     },
     {
-      "epoch": 1.3233532934131738,
-      "grad_norm": 0.20512103175801702,
-      "learning_rate": 1.239376044356076e-05,
-      "loss": 0.0024,
+      "epoch": 1.4144,
+      "grad_norm": 0.0024113977064178106,
+      "learning_rate": 1.141226278077254e-05,
+      "loss": 0.0001,
       "step": 884
     },
     {
-      "epoch": 1.3248502994011977,
-      "grad_norm": 0.08312153889808901,
-      "learning_rate": 1.2378058623927955e-05,
-      "loss": 0.001,
+      "epoch": 1.416,
+      "grad_norm": 0.044378254245468034,
+      "learning_rate": 1.1395153389439232e-05,
+      "loss": 0.0002,
       "step": 885
     },
     {
-      "epoch": 1.3263473053892216,
-      "grad_norm": 0.05507532413988543,
-      "learning_rate": 1.236235058735615e-05,
-      "loss": 0.0004,
+      "epoch": 1.4176,
+      "grad_norm": 0.010464383812279958,
+      "learning_rate": 1.1378039831966134e-05,
+      "loss": 0.0001,
       "step": 886
     },
     {
-      "epoch": 1.3278443113772456,
-      "grad_norm": 0.08688774454813779,
-      "learning_rate": 1.2346636374910735e-05,
-      "loss": 0.0008,
+      "epoch": 1.4192,
+      "grad_norm": 0.010406195082471622,
+      "learning_rate": 1.1360922159456929e-05,
+      "loss": 0.0001,
       "step": 887
     },
     {
-      "epoch": 1.3293413173652695,
-      "grad_norm": 0.18863114337383052,
-      "learning_rate": 1.233091602767324e-05,
-      "loss": 0.0014,
+      "epoch": 1.4208,
+      "grad_norm": 0.005944865648898556,
+      "learning_rate": 1.1343800423027583e-05,
+      "loss": 0.0001,
       "step": 888
     },
     {
-      "epoch": 1.3308383233532934,
-      "grad_norm": 0.02191143112881276,
-      "learning_rate": 1.2315189586741246e-05,
-      "loss": 0.0002,
+      "epoch": 1.4224,
+      "grad_norm": 0.1147538539156489,
+      "learning_rate": 1.1326674673806195e-05,
+      "loss": 0.0005,
       "step": 889
     },
     {
-      "epoch": 1.3323353293413174,
-      "grad_norm": 0.08898221164791097,
-      "learning_rate": 1.2299457093228253e-05,
-      "loss": 0.0006,
+      "epoch": 1.424,
+      "grad_norm": 0.010290915521585257,
+      "learning_rate": 1.1309544962932861e-05,
+      "loss": 0.0001,
       "step": 890
     },
     {
-      "epoch": 1.3338323353293413,
-      "grad_norm": 0.038820165742322596,
-      "learning_rate": 1.2283718588263594e-05,
+      "epoch": 1.4256,
+      "grad_norm": 0.050496243492681094,
+      "learning_rate": 1.129241134155949e-05,
       "loss": 0.0004,
       "step": 891
     },
     {
-      "epoch": 1.3353293413173652,
-      "grad_norm": 0.07647935504218839,
-      "learning_rate": 1.2267974112992307e-05,
-      "loss": 0.0011,
+      "epoch": 1.4272,
+      "grad_norm": 0.16811807745325824,
+      "learning_rate": 1.1275273860849684e-05,
+      "loss": 0.0012,
       "step": 892
     },
     {
-      "epoch": 1.3368263473053892,
-      "grad_norm": 0.13982650167833818,
-      "learning_rate": 1.2252223708575047e-05,
-      "loss": 0.002,
+      "epoch": 1.4288,
+      "grad_norm": 0.12687185170909612,
+      "learning_rate": 1.1258132571978555e-05,
+      "loss": 0.0012,
       "step": 893
     },
     {
-      "epoch": 1.3383233532934131,
-      "grad_norm": 0.1338195205882971,
-      "learning_rate": 1.2236467416187965e-05,
-      "loss": 0.0025,
+      "epoch": 1.4304000000000001,
+      "grad_norm": 0.008316438089630363,
+      "learning_rate": 1.1240987526132595e-05,
+      "loss": 0.0001,
       "step": 894
     },
     {
-      "epoch": 1.339820359281437,
-      "grad_norm": 0.1307841640751261,
-      "learning_rate": 1.2220705277022602e-05,
-      "loss": 0.0016,
+      "epoch": 1.432,
+      "grad_norm": 0.03953100157978685,
+      "learning_rate": 1.1223838774509515e-05,
+      "loss": 0.0002,
       "step": 895
     },
     {
-      "epoch": 1.341317365269461,
-      "grad_norm": 0.16651571491950992,
-      "learning_rate": 1.2204937332285796e-05,
-      "loss": 0.0024,
+      "epoch": 1.4336,
+      "grad_norm": 0.05613133148708491,
+      "learning_rate": 1.1206686368318087e-05,
+      "loss": 0.0003,
       "step": 896
     },
     {
-      "epoch": 1.342814371257485,
-      "grad_norm": 0.054613620669397775,
-      "learning_rate": 1.2189163623199547e-05,
-      "loss": 0.0006,
+      "epoch": 1.4352,
+      "grad_norm": 0.1734257162416317,
+      "learning_rate": 1.1189530358778005e-05,
+      "loss": 0.0014,
       "step": 897
     },
     {
-      "epoch": 1.3443113772455089,
-      "grad_norm": 0.6200338634281016,
-      "learning_rate": 1.2173384191000936e-05,
-      "loss": 0.0027,
+      "epoch": 1.4368,
+      "grad_norm": 0.10343915105983238,
+      "learning_rate": 1.1172370797119711e-05,
+      "loss": 0.0002,
       "step": 898
     },
     {
-      "epoch": 1.3458083832335328,
-      "grad_norm": 0.10417162543361413,
-      "learning_rate": 1.2157599076942004e-05,
-      "loss": 0.0006,
+      "epoch": 1.4384000000000001,
+      "grad_norm": 0.0016368939113926887,
+      "learning_rate": 1.1155207734584264e-05,
+      "loss": 0.0001,
       "step": 899
     },
     {
-      "epoch": 1.347305389221557,
-      "grad_norm": 0.28679445691083094,
-      "learning_rate": 1.214180832228964e-05,
-      "loss": 0.0018,
+      "epoch": 1.44,
+      "grad_norm": 0.004035518677210255,
+      "learning_rate": 1.1138041222423177e-05,
+      "loss": 0.0001,
       "step": 900
     },
     {
-      "epoch": 1.348802395209581,
-      "grad_norm": 0.13667869860487905,
-      "learning_rate": 1.2126011968325486e-05,
-      "loss": 0.001,
+      "epoch": 1.4416,
+      "grad_norm": 0.024247413227502176,
+      "learning_rate": 1.1120871311898254e-05,
+      "loss": 0.0001,
       "step": 901
     },
     {
-      "epoch": 1.3502994011976048,
-      "grad_norm": 0.14678301118335677,
-      "learning_rate": 1.211021005634582e-05,
-      "loss": 0.0014,
+      "epoch": 1.4432,
+      "grad_norm": 0.043977625453513865,
+      "learning_rate": 1.110369805428146e-05,
+      "loss": 0.0003,
       "step": 902
     },
     {
-      "epoch": 1.3517964071856288,
-      "grad_norm": 0.03643518165194748,
-      "learning_rate": 1.2094402627661447e-05,
-      "loss": 0.0004,
+      "epoch": 1.4447999999999999,
+      "grad_norm": 0.20781567679186627,
+      "learning_rate": 1.1086521500854746e-05,
+      "loss": 0.0018,
       "step": 903
     },
     {
-      "epoch": 1.3532934131736527,
-      "grad_norm": 0.21051811393884823,
-      "learning_rate": 1.2078589723597599e-05,
-      "loss": 0.0019,
+      "epoch": 1.4464000000000001,
+      "grad_norm": 0.25381820106679076,
+      "learning_rate": 1.106934170290991e-05,
+      "loss": 0.0022,
       "step": 904
     },
     {
-      "epoch": 1.3547904191616766,
-      "grad_norm": 0.4465499366437971,
-      "learning_rate": 1.2062771385493818e-05,
-      "loss": 0.0019,
+      "epoch": 1.448,
+      "grad_norm": 0.002785668497734354,
+      "learning_rate": 1.1052158711748435e-05,
+      "loss": 0.0001,
       "step": 905
     },
     {
-      "epoch": 1.3562874251497006,
-      "grad_norm": 0.3280121687211394,
-      "learning_rate": 1.204694765470386e-05,
-      "loss": 0.0024,
+      "epoch": 1.4496,
+      "grad_norm": 0.002876651077840309,
+      "learning_rate": 1.1034972578681338e-05,
+      "loss": 0.0001,
       "step": 906
     },
     {
-      "epoch": 1.3577844311377245,
-      "grad_norm": 0.19136289912588866,
-      "learning_rate": 1.2031118572595568e-05,
-      "loss": 0.0012,
+      "epoch": 1.4512,
+      "grad_norm": 0.0019333083410343309,
+      "learning_rate": 1.1017783355029027e-05,
+      "loss": 0.0001,
       "step": 907
     },
     {
-      "epoch": 1.3592814371257484,
-      "grad_norm": 0.3123777529319237,
-      "learning_rate": 1.2015284180550786e-05,
-      "loss": 0.0028,
+      "epoch": 1.4527999999999999,
+      "grad_norm": 0.0031434181560634844,
+      "learning_rate": 1.1000591092121126e-05,
+      "loss": 0.0001,
       "step": 908
     },
     {
-      "epoch": 1.3607784431137724,
-      "grad_norm": 0.14613074780473378,
-      "learning_rate": 1.1999444519965228e-05,
-      "loss": 0.0006,
+      "epoch": 1.4544000000000001,
+      "grad_norm": 0.0886988903254247,
+      "learning_rate": 1.0983395841296349e-05,
+      "loss": 0.0008,
       "step": 909
     },
     {
-      "epoch": 1.3622754491017965,
-      "grad_norm": 0.08510506250722642,
-      "learning_rate": 1.1983599632248393e-05,
-      "loss": 0.0011,
+      "epoch": 1.456,
+      "grad_norm": 0.06013855398709422,
+      "learning_rate": 1.0966197653902319e-05,
+      "loss": 0.0002,
       "step": 910
     },
     {
-      "epoch": 1.3637724550898205,
-      "grad_norm": 0.24109722460040886,
-      "learning_rate": 1.1967749558823436e-05,
-      "loss": 0.0017,
+      "epoch": 1.4576,
+      "grad_norm": 0.02346633226702475,
+      "learning_rate": 1.0948996581295437e-05,
+      "loss": 0.0002,
       "step": 911
     },
     {
-      "epoch": 1.3652694610778444,
-      "grad_norm": 0.06997668553987517,
-      "learning_rate": 1.1951894341127075e-05,
-      "loss": 0.0008,
+      "epoch": 1.4592,
+      "grad_norm": 0.0682423743158357,
+      "learning_rate": 1.0931792674840718e-05,
+      "loss": 0.0006,
       "step": 912
     },
     {
-      "epoch": 1.3667664670658684,
-      "grad_norm": 0.08339177784324386,
-      "learning_rate": 1.1936034020609476e-05,
-      "loss": 0.0013,
+      "epoch": 1.4607999999999999,
+      "grad_norm": 0.0036738316491110286,
+      "learning_rate": 1.0914585985911632e-05,
+      "loss": 0.0001,
       "step": 913
     },
     {
-      "epoch": 1.3682634730538923,
-      "grad_norm": 0.37011844961784346,
-      "learning_rate": 1.192016863873414e-05,
-      "loss": 0.0031,
+      "epoch": 1.4624,
+      "grad_norm": 0.15844151837682197,
+      "learning_rate": 1.0897376565889972e-05,
+      "loss": 0.0012,
       "step": 914
     },
     {
-      "epoch": 1.3697604790419162,
-      "grad_norm": 0.19448555543500273,
-      "learning_rate": 1.1904298236977808e-05,
-      "loss": 0.0006,
+      "epoch": 1.464,
+      "grad_norm": 0.030187227249574854,
+      "learning_rate": 1.0880164466165675e-05,
+      "loss": 0.0002,
       "step": 915
     },
     {
-      "epoch": 1.3712574850299402,
-      "grad_norm": 0.37638195330236135,
-      "learning_rate": 1.1888422856830335e-05,
-      "loss": 0.002,
+      "epoch": 1.4656,
+      "grad_norm": 0.017291779864942392,
+      "learning_rate": 1.0862949738136682e-05,
+      "loss": 0.0001,
       "step": 916
     },
     {
-      "epoch": 1.372754491017964,
-      "grad_norm": 0.04845927126998666,
-      "learning_rate": 1.18725425397946e-05,
-      "loss": 0.0006,
+      "epoch": 1.4672,
+      "grad_norm": 0.06340062218060905,
+      "learning_rate": 1.084573243320878e-05,
+      "loss": 0.0002,
       "step": 917
     },
     {
-      "epoch": 1.374251497005988,
-      "grad_norm": 0.050269398324577504,
-      "learning_rate": 1.1856657327386381e-05,
-      "loss": 0.0003,
+      "epoch": 1.4687999999999999,
+      "grad_norm": 0.02237257630182562,
+      "learning_rate": 1.0828512602795462e-05,
+      "loss": 0.0002,
       "step": 918
     },
     {
-      "epoch": 1.375748502994012,
-      "grad_norm": 0.16322026432621767,
-      "learning_rate": 1.1840767261134262e-05,
-      "loss": 0.0017,
+      "epoch": 1.4704,
+      "grad_norm": 0.0307339934395069,
+      "learning_rate": 1.0811290298317755e-05,
+      "loss": 0.0002,
       "step": 919
     },
     {
-      "epoch": 1.377245508982036,
-      "grad_norm": 0.1990865675756091,
-      "learning_rate": 1.1824872382579511e-05,
-      "loss": 0.004,
+      "epoch": 1.472,
+      "grad_norm": 0.1726287183949661,
+      "learning_rate": 1.0794065571204073e-05,
+      "loss": 0.0016,
       "step": 920
     },
     {
-      "epoch": 1.3787425149700598,
-      "grad_norm": 0.07465156559192355,
-      "learning_rate": 1.1808972733275977e-05,
-      "loss": 0.0008,
+      "epoch": 1.4736,
+      "grad_norm": 0.0023880120319958176,
+      "learning_rate": 1.0776838472890065e-05,
+      "loss": 0.0001,
       "step": 921
     },
     {
-      "epoch": 1.3802395209580838,
-      "grad_norm": 0.09208435898453236,
-      "learning_rate": 1.1793068354789978e-05,
-      "loss": 0.0013,
+      "epoch": 1.4752,
+      "grad_norm": 0.0036727102455103296,
+      "learning_rate": 1.0759609054818459e-05,
+      "loss": 0.0001,
       "step": 922
     },
     {
-      "epoch": 1.3817365269461077,
-      "grad_norm": 0.12422665350230562,
-      "learning_rate": 1.1777159288700203e-05,
-      "loss": 0.0013,
+      "epoch": 1.4768,
+      "grad_norm": 0.015067220396384931,
+      "learning_rate": 1.0742377368438915e-05,
+      "loss": 0.0001,
       "step": 923
     },
     {
-      "epoch": 1.3832335329341316,
-      "grad_norm": 0.2436082341006343,
-      "learning_rate": 1.176124557659759e-05,
-      "loss": 0.0017,
+      "epoch": 1.4784,
+      "grad_norm": 0.20821142049422797,
+      "learning_rate": 1.0725143465207868e-05,
+      "loss": 0.0056,
       "step": 924
     },
     {
-      "epoch": 1.3847305389221556,
-      "grad_norm": 0.12781022014989502,
-      "learning_rate": 1.174532726008523e-05,
-      "loss": 0.0018,
+      "epoch": 1.48,
+      "grad_norm": 0.07935528902485146,
+      "learning_rate": 1.0707907396588362e-05,
+      "loss": 0.0004,
       "step": 925
     },
     {
-      "epoch": 1.3862275449101795,
-      "grad_norm": 0.16812228483761696,
-      "learning_rate": 1.1729404380778241e-05,
-      "loss": 0.0026,
+      "epoch": 1.4816,
+      "grad_norm": 0.009333519486653154,
+      "learning_rate": 1.069066921404992e-05,
+      "loss": 0.0001,
       "step": 926
     },
     {
-      "epoch": 1.3877245508982035,
-      "grad_norm": 0.0920216290898727,
-      "learning_rate": 1.1713476980303678e-05,
-      "loss": 0.0012,
+      "epoch": 1.4832,
+      "grad_norm": 0.0023647105560942287,
+      "learning_rate": 1.0673428969068365e-05,
+      "loss": 0.0001,
       "step": 927
     },
     {
-      "epoch": 1.3892215568862276,
-      "grad_norm": 0.11482491332593553,
-      "learning_rate": 1.1697545100300408e-05,
-      "loss": 0.002,
+      "epoch": 1.4848,
+      "grad_norm": 0.00757239255372897,
+      "learning_rate": 1.065618671312569e-05,
+      "loss": 0.0001,
       "step": 928
     },
     {
-      "epoch": 1.3907185628742516,
-      "grad_norm": 0.10002074873744907,
-      "learning_rate": 1.1681608782419014e-05,
-      "loss": 0.001,
+      "epoch": 1.4864,
+      "grad_norm": 0.003198079535611409,
+      "learning_rate": 1.063894249770989e-05,
+      "loss": 0.0001,
       "step": 929
     },
     {
-      "epoch": 1.3922155688622755,
-      "grad_norm": 0.08746209991474958,
-      "learning_rate": 1.1665668068321684e-05,
-      "loss": 0.0009,
+      "epoch": 1.488,
+      "grad_norm": 0.0028351381695412175,
+      "learning_rate": 1.0621696374314807e-05,
+      "loss": 0.0001,
       "step": 930
     },
     {
-      "epoch": 1.3937125748502994,
-      "grad_norm": 0.0814270207122144,
-      "learning_rate": 1.1649722999682094e-05,
-      "loss": 0.0012,
+      "epoch": 1.4896,
+      "grad_norm": 0.022597613005629305,
+      "learning_rate": 1.0604448394439983e-05,
+      "loss": 0.0002,
       "step": 931
     },
     {
-      "epoch": 1.3952095808383234,
-      "grad_norm": 0.09416853678436857,
-      "learning_rate": 1.1633773618185302e-05,
-      "loss": 0.001,
+      "epoch": 1.4912,
+      "grad_norm": 0.0944509156527911,
+      "learning_rate": 1.0587198609590505e-05,
+      "loss": 0.0008,
       "step": 932
     },
     {
-      "epoch": 1.3967065868263473,
-      "grad_norm": 0.14964696805599081,
-      "learning_rate": 1.161781996552765e-05,
-      "loss": 0.0012,
+      "epoch": 1.4928,
+      "grad_norm": 0.02252196232378922,
+      "learning_rate": 1.0569947071276847e-05,
+      "loss": 0.0001,
       "step": 933
     },
     {
-      "epoch": 1.3982035928143712,
-      "grad_norm": 0.09483767589434133,
-      "learning_rate": 1.1601862083416632e-05,
-      "loss": 0.0013,
+      "epoch": 1.4944,
+      "grad_norm": 0.06485245834320515,
+      "learning_rate": 1.0552693831014726e-05,
+      "loss": 0.0019,
       "step": 934
     },
     {
-      "epoch": 1.3997005988023952,
-      "grad_norm": 0.09788376172150742,
-      "learning_rate": 1.1585900013570815e-05,
-      "loss": 0.0015,
+      "epoch": 1.496,
+      "grad_norm": 0.005034575123535191,
+      "learning_rate": 1.053543894032493e-05,
+      "loss": 0.0001,
       "step": 935
     },
     {
-      "epoch": 1.401197604790419,
-      "grad_norm": 0.10836219152191484,
-      "learning_rate": 1.1569933797719703e-05,
-      "loss": 0.0016,
+      "epoch": 1.4976,
+      "grad_norm": 0.008163939414128022,
+      "learning_rate": 1.0518182450733185e-05,
+      "loss": 0.0001,
       "step": 936
     },
     {
-      "epoch": 1.402694610778443,
-      "grad_norm": 0.23674230633425863,
-      "learning_rate": 1.155396347760364e-05,
-      "loss": 0.0016,
+      "epoch": 1.4992,
+      "grad_norm": 0.005588487482831217,
+      "learning_rate": 1.0500924413769988e-05,
+      "loss": 0.0001,
       "step": 937
     },
     {
-      "epoch": 1.4041916167664672,
-      "grad_norm": 0.08196393324885805,
-      "learning_rate": 1.1537989094973705e-05,
-      "loss": 0.0014,
+      "epoch": 1.5008,
+      "grad_norm": 0.006037605153287934,
+      "learning_rate": 1.0483664880970456e-05,
+      "loss": 0.0001,
       "step": 938
     },
     {
-      "epoch": 1.4056886227544911,
-      "grad_norm": 0.05782826404362468,
-      "learning_rate": 1.1522010691591595e-05,
-      "loss": 0.0007,
+      "epoch": 1.5024,
+      "grad_norm": 0.004750521037278343,
+      "learning_rate": 1.0466403903874176e-05,
+      "loss": 0.0001,
       "step": 939
     },
     {
-      "epoch": 1.407185628742515,
-      "grad_norm": 0.14623254375235795,
-      "learning_rate": 1.1506028309229513e-05,
-      "loss": 0.0019,
+      "epoch": 1.504,
+      "grad_norm": 0.06278347831522739,
+      "learning_rate": 1.0449141534025044e-05,
+      "loss": 0.0006,
       "step": 940
     },
     {
-      "epoch": 1.408682634730539,
-      "grad_norm": 0.10434558213727799,
-      "learning_rate": 1.1490041989670076e-05,
-      "loss": 0.0012,
+      "epoch": 1.5056,
+      "grad_norm": 0.0024527693371791076,
+      "learning_rate": 1.0431877822971118e-05,
+      "loss": 0.0001,
       "step": 941
     },
     {
-      "epoch": 1.410179640718563,
-      "grad_norm": 0.15631392324819193,
-      "learning_rate": 1.1474051774706181e-05,
-      "loss": 0.0013,
+      "epoch": 1.5072,
+      "grad_norm": 0.002035872923279796,
+      "learning_rate": 1.0414612822264457e-05,
+      "loss": 0.0001,
       "step": 942
     },
     {
-      "epoch": 1.4116766467065869,
-      "grad_norm": 0.13437189533675623,
-      "learning_rate": 1.145805770614092e-05,
-      "loss": 0.0022,
+      "epoch": 1.5088,
+      "grad_norm": 0.0068014432836293595,
+      "learning_rate": 1.0397346583460972e-05,
+      "loss": 0.0001,
       "step": 943
     },
     {
-      "epoch": 1.4131736526946108,
-      "grad_norm": 0.038973494041355694,
-      "learning_rate": 1.1442059825787448e-05,
-      "loss": 0.0004,
+      "epoch": 1.5104,
+      "grad_norm": 0.005721254040327972,
+      "learning_rate": 1.038007915812028e-05,
+      "loss": 0.0001,
       "step": 944
     },
     {
-      "epoch": 1.4146706586826348,
-      "grad_norm": 0.07141700721890615,
-      "learning_rate": 1.1426058175468895e-05,
-      "loss": 0.0011,
+      "epoch": 1.512,
+      "grad_norm": 0.030337274027592388,
+      "learning_rate": 1.0362810597805526e-05,
+      "loss": 0.0002,
       "step": 945
     },
     {
-      "epoch": 1.4161676646706587,
-      "grad_norm": 0.21493859310812677,
-      "learning_rate": 1.1410052797018245e-05,
-      "loss": 0.0017,
+      "epoch": 1.5135999999999998,
+      "grad_norm": 0.08645131868148463,
+      "learning_rate": 1.034554095408326e-05,
+      "loss": 0.0005,
       "step": 946
     },
     {
-      "epoch": 1.4176646706586826,
-      "grad_norm": 0.1282522648959389,
-      "learning_rate": 1.1394043732278225e-05,
-      "loss": 0.001,
+      "epoch": 1.5152,
+      "grad_norm": 0.005076246051467128,
+      "learning_rate": 1.0328270278523256e-05,
+      "loss": 0.0001,
       "step": 947
     },
     {
-      "epoch": 1.4191616766467066,
-      "grad_norm": 0.07019102461290384,
-      "learning_rate": 1.1378031023101199e-05,
-      "loss": 0.0008,
+      "epoch": 1.5168,
+      "grad_norm": 0.008998648164364345,
+      "learning_rate": 1.031099862269837e-05,
+      "loss": 0.0001,
       "step": 948
     },
     {
-      "epoch": 1.4206586826347305,
-      "grad_norm": 0.05464505705186005,
-      "learning_rate": 1.1362014711349063e-05,
-      "loss": 0.0004,
+      "epoch": 1.5184,
+      "grad_norm": 0.0020866531394263662,
+      "learning_rate": 1.0293726038184393e-05,
+      "loss": 0.0001,
       "step": 949
     },
     {
-      "epoch": 1.4221556886227544,
-      "grad_norm": 0.07090800295191836,
-      "learning_rate": 1.1345994838893125e-05,
-      "loss": 0.0004,
+      "epoch": 1.52,
+      "grad_norm": 0.004967861683286264,
+      "learning_rate": 1.0276452576559878e-05,
+      "loss": 0.0001,
       "step": 950
     },
     {
-      "epoch": 1.4236526946107784,
-      "grad_norm": 0.1475910596450026,
-      "learning_rate": 1.1329971447614008e-05,
-      "loss": 0.0014,
+      "epoch": 1.5215999999999998,
+      "grad_norm": 0.00828972554606736,
+      "learning_rate": 1.0259178289406011e-05,
+      "loss": 0.0001,
       "step": 951
     },
     {
-      "epoch": 1.4251497005988023,
-      "grad_norm": 0.11628540378666755,
-      "learning_rate": 1.1313944579401528e-05,
-      "loss": 0.0008,
+      "epoch": 1.5232,
+      "grad_norm": 0.14247992903632217,
+      "learning_rate": 1.024190322830643e-05,
+      "loss": 0.0019,
       "step": 952
     },
     {
-      "epoch": 1.4266467065868262,
-      "grad_norm": 0.11918182166269133,
-      "learning_rate": 1.12979142761546e-05,
-      "loss": 0.0013,
+      "epoch": 1.5248,
+      "grad_norm": 0.05069890191122051,
+      "learning_rate": 1.022462744484709e-05,
+      "loss": 0.0003,
       "step": 953
     },
     {
-      "epoch": 1.4281437125748502,
-      "grad_norm": 0.06010179192377916,
-      "learning_rate": 1.1281880579781108e-05,
-      "loss": 0.0008,
+      "epoch": 1.5264,
+      "grad_norm": 0.004596669597143668,
+      "learning_rate": 1.0207350990616107e-05,
+      "loss": 0.0001,
       "step": 954
     },
     {
-      "epoch": 1.4296407185628741,
-      "grad_norm": 0.07923386193741429,
-      "learning_rate": 1.1265843532197813e-05,
-      "loss": 0.001,
+      "epoch": 1.528,
+      "grad_norm": 0.005706623148356151,
+      "learning_rate": 1.019007391720359e-05,
+      "loss": 0.0001,
       "step": 955
     },
     {
-      "epoch": 1.4311377245508983,
-      "grad_norm": 0.09159193558417823,
-      "learning_rate": 1.1249803175330237e-05,
-      "loss": 0.0015,
+      "epoch": 1.5295999999999998,
+      "grad_norm": 0.01754239249700808,
+      "learning_rate": 1.0172796276201504e-05,
+      "loss": 0.0001,
       "step": 956
     },
     {
-      "epoch": 1.4326347305389222,
-      "grad_norm": 0.07308876635837779,
-      "learning_rate": 1.1233759551112553e-05,
-      "loss": 0.0004,
+      "epoch": 1.5312000000000001,
+      "grad_norm": 0.0033728434205184442,
+      "learning_rate": 1.0155518119203511e-05,
+      "loss": 0.0001,
       "step": 957
     },
     {
-      "epoch": 1.4341317365269461,
-      "grad_norm": 0.0779800871854366,
-      "learning_rate": 1.1217712701487471e-05,
-      "loss": 0.0008,
+      "epoch": 1.5328,
+      "grad_norm": 0.002274125753851539,
+      "learning_rate": 1.0138239497804804e-05,
+      "loss": 0.0001,
       "step": 958
     },
     {
-      "epoch": 1.43562874251497,
-      "grad_norm": 0.06632363235084476,
-      "learning_rate": 1.120166266840614e-05,
-      "loss": 0.0005,
+      "epoch": 1.5344,
+      "grad_norm": 0.007272566220733121,
+      "learning_rate": 1.0120960463601977e-05,
+      "loss": 0.0001,
       "step": 959
     },
     {
-      "epoch": 1.437125748502994,
-      "grad_norm": 0.05842006960591898,
-      "learning_rate": 1.1185609493828033e-05,
-      "loss": 0.0007,
+      "epoch": 1.536,
+      "grad_norm": 0.003240011548304806,
+      "learning_rate": 1.0103681068192845e-05,
+      "loss": 0.0001,
       "step": 960
     },
     {
-      "epoch": 1.438622754491018,
-      "grad_norm": 0.19344221257223246,
-      "learning_rate": 1.1169553219720828e-05,
-      "loss": 0.0031,
+      "epoch": 1.5375999999999999,
+      "grad_norm": 0.014366781565456368,
+      "learning_rate": 1.0086401363176306e-05,
+      "loss": 0.0001,
       "step": 961
     },
     {
-      "epoch": 1.4401197604790419,
-      "grad_norm": 0.1981812238766314,
-      "learning_rate": 1.1153493888060304e-05,
-      "loss": 0.0013,
+      "epoch": 1.5392000000000001,
+      "grad_norm": 0.018848763547187536,
+      "learning_rate": 1.0069121400152182e-05,
+      "loss": 0.0001,
       "step": 962
     },
     {
-      "epoch": 1.4416167664670658,
-      "grad_norm": 0.055964112494847354,
-      "learning_rate": 1.1137431540830248e-05,
-      "loss": 0.0006,
+      "epoch": 1.5408,
+      "grad_norm": 0.005865494954341179,
+      "learning_rate": 1.0051841230721065e-05,
+      "loss": 0.0001,
       "step": 963
     },
     {
-      "epoch": 1.4431137724550898,
-      "grad_norm": 0.09671781357514952,
-      "learning_rate": 1.1121366220022317e-05,
-      "loss": 0.0012,
+      "epoch": 1.5424,
+      "grad_norm": 0.00576194392886361,
+      "learning_rate": 1.0034560906484161e-05,
+      "loss": 0.0001,
       "step": 964
     },
     {
-      "epoch": 1.4446107784431137,
-      "grad_norm": 0.23802060163109717,
-      "learning_rate": 1.1105297967635947e-05,
-      "loss": 0.002,
+      "epoch": 1.544,
+      "grad_norm": 0.08250532542920345,
+      "learning_rate": 1.0017280479043148e-05,
+      "loss": 0.0003,
       "step": 965
     },
     {
-      "epoch": 1.4461077844311379,
-      "grad_norm": 0.17887498237379332,
-      "learning_rate": 1.1089226825678239e-05,
-      "loss": 0.0014,
+      "epoch": 1.5455999999999999,
+      "grad_norm": 0.004488788327332194,
+      "learning_rate": 1e-05,
+      "loss": 0.0001,
       "step": 966
     },
     {
-      "epoch": 1.4476047904191618,
-      "grad_norm": 0.041592588915209894,
-      "learning_rate": 1.1073152836163842e-05,
-      "loss": 0.0004,
+      "epoch": 1.5472000000000001,
+      "grad_norm": 0.016648640713315763,
+      "learning_rate": 9.982719520956856e-06,
+      "loss": 0.0001,
       "step": 967
     },
     {
-      "epoch": 1.4491017964071857,
-      "grad_norm": 0.1005400146875568,
-      "learning_rate": 1.105707604111486e-05,
-      "loss": 0.0006,
+      "epoch": 1.5488,
+      "grad_norm": 0.048241822041450236,
+      "learning_rate": 9.965439093515842e-06,
+      "loss": 0.0002,
       "step": 968
     },
     {
-      "epoch": 1.4505988023952097,
-      "grad_norm": 0.0974962168976635,
-      "learning_rate": 1.1040996482560724e-05,
-      "loss": 0.0014,
+      "epoch": 1.5504,
+      "grad_norm": 0.01499468657612021,
+      "learning_rate": 9.948158769278939e-06,
+      "loss": 0.0002,
       "step": 969
     },
     {
-      "epoch": 1.4520958083832336,
-      "grad_norm": 0.0935782185529421,
-      "learning_rate": 1.1024914202538088e-05,
-      "loss": 0.0008,
+      "epoch": 1.552,
+      "grad_norm": 0.005852751046341136,
+      "learning_rate": 9.930878599847822e-06,
+      "loss": 0.0001,
       "step": 970
     },
     {
-      "epoch": 1.4535928143712575,
-      "grad_norm": 0.07730096837501058,
-      "learning_rate": 1.1008829243090724e-05,
-      "loss": 0.0006,
+      "epoch": 1.5535999999999999,
+      "grad_norm": 0.0752892074546245,
+      "learning_rate": 9.913598636823694e-06,
+      "loss": 0.0023,
       "step": 971
     },
     {
-      "epoch": 1.4550898203592815,
-      "grad_norm": 0.12823405242175784,
-      "learning_rate": 1.099274164626941e-05,
-      "loss": 0.0009,
+      "epoch": 1.5552000000000001,
+      "grad_norm": 0.026611060806342526,
+      "learning_rate": 9.896318931807155e-06,
+      "loss": 0.0002,
       "step": 972
     },
     {
-      "epoch": 1.4565868263473054,
-      "grad_norm": 0.07157978147756917,
-      "learning_rate": 1.0976651454131814e-05,
-      "loss": 0.001,
+      "epoch": 1.5568,
+      "grad_norm": 0.07719852044515452,
+      "learning_rate": 9.879039536398023e-06,
+      "loss": 0.0005,
       "step": 973
     },
     {
-      "epoch": 1.4580838323353293,
-      "grad_norm": 0.07310373401364316,
-      "learning_rate": 1.0960558708742401e-05,
-      "loss": 0.0005,
+      "epoch": 1.5584,
+      "grad_norm": 0.003910047594098633,
+      "learning_rate": 9.861760502195197e-06,
+      "loss": 0.0001,
       "step": 974
     },
     {
-      "epoch": 1.4595808383233533,
-      "grad_norm": 0.11186858330092693,
-      "learning_rate": 1.0944463452172297e-05,
-      "loss": 0.001,
+      "epoch": 1.56,
+      "grad_norm": 0.056305740800321455,
+      "learning_rate": 9.844481880796492e-06,
+      "loss": 0.0002,
       "step": 975
     },
     {
-      "epoch": 1.4610778443113772,
-      "grad_norm": 0.1613611934605809,
-      "learning_rate": 1.0928365726499194e-05,
-      "loss": 0.0027,
+      "epoch": 1.5615999999999999,
+      "grad_norm": 0.0021305486069807056,
+      "learning_rate": 9.827203723798498e-06,
+      "loss": 0.0001,
       "step": 976
     },
     {
-      "epoch": 1.4625748502994012,
-      "grad_norm": 0.1622550489094645,
-      "learning_rate": 1.0912265573807249e-05,
-      "loss": 0.0008,
+      "epoch": 1.5632000000000001,
+      "grad_norm": 0.011272710180135622,
+      "learning_rate": 9.809926082796415e-06,
+      "loss": 0.0001,
       "step": 977
     },
     {
-      "epoch": 1.464071856287425,
-      "grad_norm": 0.15392255395991697,
-      "learning_rate": 1.0896163036186953e-05,
-      "loss": 0.0014,
+      "epoch": 1.5648,
+      "grad_norm": 0.007720417514812891,
+      "learning_rate": 9.7926490093839e-06,
+      "loss": 0.0001,
       "step": 978
     },
     {
-      "epoch": 1.465568862275449,
-      "grad_norm": 0.15486888428359488,
-      "learning_rate": 1.088005815573504e-05,
-      "loss": 0.0013,
+      "epoch": 1.5664,
+      "grad_norm": 0.010832344610442256,
+      "learning_rate": 9.775372555152912e-06,
+      "loss": 0.0001,
       "step": 979
     },
     {
-      "epoch": 1.467065868263473,
-      "grad_norm": 0.16549731334948978,
-      "learning_rate": 1.0863950974554372e-05,
-      "loss": 0.0011,
+      "epoch": 1.568,
+      "grad_norm": 0.009070884087999895,
+      "learning_rate": 9.758096771693574e-06,
+      "loss": 0.0001,
       "step": 980
     },
     {
-      "epoch": 1.468562874251497,
-      "grad_norm": 0.06451393525137326,
-      "learning_rate": 1.084784153475381e-05,
-      "loss": 0.0004,
+      "epoch": 1.5695999999999999,
+      "grad_norm": 0.00442157961174515,
+      "learning_rate": 9.740821710593989e-06,
+      "loss": 0.0001,
       "step": 981
     },
     {
-      "epoch": 1.4700598802395208,
-      "grad_norm": 0.08287888893875361,
-      "learning_rate": 1.0831729878448131e-05,
-      "loss": 0.0006,
+      "epoch": 1.5712000000000002,
+      "grad_norm": 0.027300721561953286,
+      "learning_rate": 9.723547423440122e-06,
+      "loss": 0.0001,
       "step": 982
     },
     {
-      "epoch": 1.471556886227545,
-      "grad_norm": 0.14774856291997404,
-      "learning_rate": 1.0815616047757909e-05,
-      "loss": 0.0015,
+      "epoch": 1.5728,
+      "grad_norm": 0.025063083496432152,
+      "learning_rate": 9.70627396181561e-06,
+      "loss": 0.0001,
       "step": 983
     },
     {
-      "epoch": 1.473053892215569,
-      "grad_norm": 0.08417242989544835,
-      "learning_rate": 1.0799500084809393e-05,
-      "loss": 0.0009,
+      "epoch": 1.5744,
+      "grad_norm": 0.002248942312411431,
+      "learning_rate": 9.689001377301634e-06,
+      "loss": 0.0001,
       "step": 984
     },
     {
-      "epoch": 1.4745508982035929,
-      "grad_norm": 0.05265592604561263,
-      "learning_rate": 1.0783382031734418e-05,
-      "loss": 0.0008,
+      "epoch": 1.576,
+      "grad_norm": 0.005737897000865526,
+      "learning_rate": 9.671729721476747e-06,
+      "loss": 0.0001,
       "step": 985
     },
     {
-      "epoch": 1.4760479041916168,
-      "grad_norm": 0.11539265697300215,
-      "learning_rate": 1.0767261930670272e-05,
-      "loss": 0.0015,
+      "epoch": 1.5776,
+      "grad_norm": 0.0020266436310786198,
+      "learning_rate": 9.654459045916743e-06,
+      "loss": 0.0001,
       "step": 986
     },
     {
-      "epoch": 1.4775449101796407,
-      "grad_norm": 0.029457601423187277,
-      "learning_rate": 1.0751139823759603e-05,
-      "loss": 0.0002,
+      "epoch": 1.5792000000000002,
+      "grad_norm": 0.012861872927739158,
+      "learning_rate": 9.637189402194477e-06,
+      "loss": 0.0001,
       "step": 987
     },
     {
-      "epoch": 1.4790419161676647,
-      "grad_norm": 0.10750596108363268,
-      "learning_rate": 1.0735015753150305e-05,
-      "loss": 0.0015,
+      "epoch": 1.5808,
+      "grad_norm": 0.022432987874545302,
+      "learning_rate": 9.619920841879726e-06,
+      "loss": 0.0002,
       "step": 988
     },
     {
-      "epoch": 1.4805389221556886,
-      "grad_norm": 0.23998882396666726,
-      "learning_rate": 1.0718889760995396e-05,
-      "loss": 0.0016,
+      "epoch": 1.5824,
+      "grad_norm": 0.1734129903191722,
+      "learning_rate": 9.602653416539031e-06,
+      "loss": 0.0015,
       "step": 989
     },
     {
-      "epoch": 1.4820359281437125,
-      "grad_norm": 0.1009612926927026,
-      "learning_rate": 1.070276188945293e-05,
-      "loss": 0.0014,
+      "epoch": 1.584,
+      "grad_norm": 0.11556869758171924,
+      "learning_rate": 9.585387177735548e-06,
+      "loss": 0.0005,
       "step": 990
     },
     {
-      "epoch": 1.4835329341317365,
-      "grad_norm": 0.12007396345906915,
-      "learning_rate": 1.0686632180685868e-05,
-      "loss": 0.0015,
+      "epoch": 1.5856,
+      "grad_norm": 0.013507656807784091,
+      "learning_rate": 9.568122177028884e-06,
+      "loss": 0.0001,
       "step": 991
     },
     {
-      "epoch": 1.4850299401197604,
-      "grad_norm": 0.1882263167381516,
-      "learning_rate": 1.067050067686197e-05,
-      "loss": 0.0013,
+      "epoch": 1.5872000000000002,
+      "grad_norm": 0.01190158738889232,
+      "learning_rate": 9.550858465974958e-06,
+      "loss": 0.0001,
       "step": 992
     },
     {
-      "epoch": 1.4865269461077844,
-      "grad_norm": 0.16842368805723756,
-      "learning_rate": 1.0654367420153696e-05,
+      "epoch": 1.5888,
+      "grad_norm": 0.1443249907546037,
+      "learning_rate": 9.533596096125826e-06,
       "loss": 0.0017,
       "step": 993
     },
     {
-      "epoch": 1.4880239520958085,
-      "grad_norm": 0.13102324447775987,
-      "learning_rate": 1.0638232452738084e-05,
-      "loss": 0.0009,
+      "epoch": 1.5904,
+      "grad_norm": 0.003068551402787213,
+      "learning_rate": 9.516335119029547e-06,
+      "loss": 0.0001,
       "step": 994
     },
     {
-      "epoch": 1.4895209580838324,
-      "grad_norm": 0.14140439476011551,
-      "learning_rate": 1.0622095816796647e-05,
-      "loss": 0.0014,
+      "epoch": 1.592,
+      "grad_norm": 0.0870000968923958,
+      "learning_rate": 9.499075586230014e-06,
+      "loss": 0.0004,
       "step": 995
     },
     {
-      "epoch": 1.4910179640718564,
-      "grad_norm": 0.048609314581425776,
-      "learning_rate": 1.0605957554515261e-05,
-      "loss": 0.0005,
+      "epoch": 1.5936,
+      "grad_norm": 0.002410019645070289,
+      "learning_rate": 9.481817549266817e-06,
+      "loss": 0.0001,
       "step": 996
     },
     {
-      "epoch": 1.4925149700598803,
-      "grad_norm": 0.09685439693147369,
-      "learning_rate": 1.058981770808405e-05,
-      "loss": 0.0008,
+      "epoch": 1.5952,
+      "grad_norm": 0.07699328070870765,
+      "learning_rate": 9.464561059675073e-06,
+      "loss": 0.0004,
       "step": 997
     },
     {
-      "epoch": 1.4940119760479043,
-      "grad_norm": 0.10476662342644127,
-      "learning_rate": 1.057367631969728e-05,
-      "loss": 0.0009,
+      "epoch": 1.5968,
+      "grad_norm": 0.03003542987962681,
+      "learning_rate": 9.44730616898528e-06,
+      "loss": 0.0002,
       "step": 998
     },
     {
-      "epoch": 1.4955089820359282,
-      "grad_norm": 0.07965591618267445,
-      "learning_rate": 1.0557533431553251e-05,
-      "loss": 0.0006,
+      "epoch": 1.5984,
+      "grad_norm": 0.00798339829717542,
+      "learning_rate": 9.430052928723153e-06,
+      "loss": 0.0001,
       "step": 999
     },
     {
-      "epoch": 1.4970059880239521,
-      "grad_norm": 0.14132503178186984,
-      "learning_rate": 1.0541389085854177e-05,
-      "loss": 0.0014,
+      "epoch": 1.6,
+      "grad_norm": 0.013732757324964928,
+      "learning_rate": 9.412801390409496e-06,
+      "loss": 0.0002,
       "step": 1000
     },
     {
-      "epoch": 1.498502994011976,
-      "grad_norm": 0.06381511478435144,
-      "learning_rate": 1.0525243324806091e-05,
-      "loss": 0.0003,
+      "epoch": 1.6016,
+      "grad_norm": 0.013712467175340954,
+      "learning_rate": 9.395551605560018e-06,
+      "loss": 0.0001,
       "step": 1001
     },
     {
-      "epoch": 1.5,
-      "grad_norm": 0.19969657755752093,
-      "learning_rate": 1.0509096190618729e-05,
-      "loss": 0.0012,
+      "epoch": 1.6032,
+      "grad_norm": 0.033443514052161075,
+      "learning_rate": 9.378303625685196e-06,
+      "loss": 0.0002,
       "step": 1002
     },
     {
-      "epoch": 1.501497005988024,
-      "grad_norm": 0.16903934501226875,
-      "learning_rate": 1.0492947725505403e-05,
-      "loss": 0.0011,
+      "epoch": 1.6048,
+      "grad_norm": 0.011735228089194433,
+      "learning_rate": 9.361057502290112e-06,
+      "loss": 0.0001,
       "step": 1003
     },
     {
-      "epoch": 1.5029940119760479,
-      "grad_norm": 0.16313780662072883,
-      "learning_rate": 1.047679797168291e-05,
-      "loss": 0.0014,
+      "epoch": 1.6064,
+      "grad_norm": 0.05469053236176875,
+      "learning_rate": 9.343813286874312e-06,
+      "loss": 0.0002,
       "step": 1004
     },
     {
-      "epoch": 1.5044910179640718,
-      "grad_norm": 0.08804790182748146,
-      "learning_rate": 1.0460646971371424e-05,
-      "loss": 0.0008,
+      "epoch": 1.608,
+      "grad_norm": 0.005661733123474927,
+      "learning_rate": 9.326571030931636e-06,
+      "loss": 0.0001,
       "step": 1005
     },
     {
-      "epoch": 1.5059880239520957,
-      "grad_norm": 0.1368065864574523,
-      "learning_rate": 1.0444494766794368e-05,
-      "loss": 0.0014,
+      "epoch": 1.6096,
+      "grad_norm": 0.0030606199222666273,
+      "learning_rate": 9.309330785950086e-06,
+      "loss": 0.0001,
       "step": 1006
     },
     {
-      "epoch": 1.5074850299401197,
-      "grad_norm": 0.1814958223187234,
-      "learning_rate": 1.0428341400178317e-05,
-      "loss": 0.0018,
+      "epoch": 1.6112,
+      "grad_norm": 0.003158444323988701,
+      "learning_rate": 9.292092603411642e-06,
+      "loss": 0.0001,
       "step": 1007
     },
     {
-      "epoch": 1.5089820359281436,
-      "grad_norm": 0.10617237101998808,
-      "learning_rate": 1.0412186913752884e-05,
-      "loss": 0.001,
+      "epoch": 1.6128,
+      "grad_norm": 0.1020769852520482,
+      "learning_rate": 9.274856534792138e-06,
+      "loss": 0.0007,
       "step": 1008
     },
     {
-      "epoch": 1.5104790419161676,
-      "grad_norm": 0.12416915017989684,
-      "learning_rate": 1.0396031349750612e-05,
-      "loss": 0.0014,
+      "epoch": 1.6143999999999998,
+      "grad_norm": 0.02723430869085718,
+      "learning_rate": 9.257622631561085e-06,
+      "loss": 0.0002,
       "step": 1009
     },
     {
-      "epoch": 1.5119760479041915,
-      "grad_norm": 0.16700967749894446,
-      "learning_rate": 1.0379874750406858e-05,
-      "loss": 0.0018,
+      "epoch": 1.616,
+      "grad_norm": 0.08514436159786103,
+      "learning_rate": 9.240390945181543e-06,
+      "loss": 0.0017,
       "step": 1010
     },
     {
-      "epoch": 1.5134730538922154,
-      "grad_norm": 0.07833916487876827,
-      "learning_rate": 1.0363717157959681e-05,
-      "loss": 0.0007,
+      "epoch": 1.6176,
+      "grad_norm": 0.0068380055800940235,
+      "learning_rate": 9.223161527109938e-06,
+      "loss": 0.0001,
       "step": 1011
     },
     {
-      "epoch": 1.5149700598802394,
-      "grad_norm": 0.1046908594504945,
-      "learning_rate": 1.0347558614649747e-05,
-      "loss": 0.0007,
+      "epoch": 1.6192,
+      "grad_norm": 0.00910581317681862,
+      "learning_rate": 9.205934428795929e-06,
+      "loss": 0.0001,
       "step": 1012
     },
     {
-      "epoch": 1.5164670658682635,
-      "grad_norm": 0.12447015495058295,
-      "learning_rate": 1.0331399162720197e-05,
-      "loss": 0.0018,
+      "epoch": 1.6208,
+      "grad_norm": 0.004989916059586335,
+      "learning_rate": 9.188709701682246e-06,
+      "loss": 0.0001,
       "step": 1013
     },
     {
-      "epoch": 1.5179640718562875,
-      "grad_norm": 0.12177043291644524,
-      "learning_rate": 1.0315238844416557e-05,
-      "loss": 0.0013,
+      "epoch": 1.6223999999999998,
+      "grad_norm": 0.031005449949701787,
+      "learning_rate": 9.17148739720454e-06,
+      "loss": 0.0002,
       "step": 1014
     },
     {
-      "epoch": 1.5194610778443114,
-      "grad_norm": 0.07862091445416813,
-      "learning_rate": 1.0299077701986613e-05,
-      "loss": 0.0007,
+      "epoch": 1.624,
+      "grad_norm": 0.002650399301172036,
+      "learning_rate": 9.154267566791224e-06,
+      "loss": 0.0001,
       "step": 1015
     },
     {
-      "epoch": 1.5209580838323353,
-      "grad_norm": 0.15981593793118748,
-      "learning_rate": 1.0282915777680307e-05,
-      "loss": 0.0012,
+      "epoch": 1.6256,
+      "grad_norm": 0.1350083424472415,
+      "learning_rate": 9.137050261863323e-06,
+      "loss": 0.0006,
       "step": 1016
     },
     {
-      "epoch": 1.5224550898203593,
-      "grad_norm": 0.09956716437231744,
-      "learning_rate": 1.0266753113749618e-05,
-      "loss": 0.001,
+      "epoch": 1.6272,
+      "grad_norm": 0.07458336731096532,
+      "learning_rate": 9.119835533834332e-06,
+      "loss": 0.0004,
       "step": 1017
     },
     {
-      "epoch": 1.5239520958083832,
-      "grad_norm": 0.12054415585534563,
-      "learning_rate": 1.0250589752448472e-05,
-      "loss": 0.0008,
+      "epoch": 1.6288,
+      "grad_norm": 0.029469347285655607,
+      "learning_rate": 9.102623434110028e-06,
+      "loss": 0.0002,
       "step": 1018
     },
     {
-      "epoch": 1.5254491017964071,
-      "grad_norm": 0.047284868067243004,
-      "learning_rate": 1.0234425736032607e-05,
-      "loss": 0.0004,
+      "epoch": 1.6303999999999998,
+      "grad_norm": 0.011116593067667624,
+      "learning_rate": 9.085414014088368e-06,
+      "loss": 0.0001,
       "step": 1019
     },
     {
-      "epoch": 1.5269461077844313,
-      "grad_norm": 0.15543960260097514,
-      "learning_rate": 1.0218261106759475e-05,
-      "loss": 0.0018,
+      "epoch": 1.6320000000000001,
+      "grad_norm": 0.01299440060936471,
+      "learning_rate": 9.068207325159285e-06,
+      "loss": 0.0001,
       "step": 1020
     },
     {
-      "epoch": 1.5284431137724552,
-      "grad_norm": 0.09490923813175474,
-      "learning_rate": 1.020209590688814e-05,
-      "loss": 0.0007,
+      "epoch": 1.6336,
+      "grad_norm": 0.2055820315378792,
+      "learning_rate": 9.051003418704566e-06,
+      "loss": 0.0017,
       "step": 1021
     },
     {
-      "epoch": 1.5299401197604792,
-      "grad_norm": 0.06536745934922997,
-      "learning_rate": 1.0185930178679146e-05,
-      "loss": 0.0006,
+      "epoch": 1.6352,
+      "grad_norm": 0.005873745936807772,
+      "learning_rate": 9.033802346097683e-06,
+      "loss": 0.0001,
       "step": 1022
     },
     {
-      "epoch": 1.531437125748503,
-      "grad_norm": 0.1301533353797184,
-      "learning_rate": 1.0169763964394421e-05,
-      "loss": 0.002,
+      "epoch": 1.6368,
+      "grad_norm": 0.013013482079196304,
+      "learning_rate": 9.016604158703654e-06,
+      "loss": 0.0001,
       "step": 1023
     },
     {
-      "epoch": 1.532934131736527,
-      "grad_norm": 0.1502603255655029,
-      "learning_rate": 1.0153597306297169e-05,
-      "loss": 0.0011,
+      "epoch": 1.6383999999999999,
+      "grad_norm": 0.006144340327088104,
+      "learning_rate": 8.999408907878877e-06,
+      "loss": 0.0002,
       "step": 1024
     },
     {
-      "epoch": 1.534431137724551,
-      "grad_norm": 0.04144441435798231,
-      "learning_rate": 1.013743024665175e-05,
-      "loss": 0.0003,
+      "epoch": 1.6400000000000001,
+      "grad_norm": 0.13665631531927,
+      "learning_rate": 8.982216644970978e-06,
+      "loss": 0.0012,
       "step": 1025
     },
     {
-      "epoch": 1.535928143712575,
-      "grad_norm": 0.15809185401608605,
-      "learning_rate": 1.0121262827723573e-05,
-      "loss": 0.0023,
+      "epoch": 1.6416,
+      "grad_norm": 0.006789307264438082,
+      "learning_rate": 8.965027421318666e-06,
+      "loss": 0.0001,
       "step": 1026
     },
     {
-      "epoch": 1.5374251497005988,
-      "grad_norm": 0.17137867242074536,
-      "learning_rate": 1.0105095091778985e-05,
-      "loss": 0.0012,
+      "epoch": 1.6432,
+      "grad_norm": 0.17167958395609068,
+      "learning_rate": 8.947841288251568e-06,
+      "loss": 0.0004,
       "step": 1027
     },
     {
-      "epoch": 1.5389221556886228,
-      "grad_norm": 0.10676172512123629,
-      "learning_rate": 1.008892708108517e-05,
-      "loss": 0.0009,
+      "epoch": 1.6448,
+      "grad_norm": 0.0609705821357434,
+      "learning_rate": 8.930658297090092e-06,
+      "loss": 0.0003,
       "step": 1028
     },
     {
-      "epoch": 1.5404191616766467,
-      "grad_norm": 0.10415442540373282,
-      "learning_rate": 1.007275883791002e-05,
-      "loss": 0.0014,
+      "epoch": 1.6463999999999999,
+      "grad_norm": 0.006587696978795276,
+      "learning_rate": 8.913478499145255e-06,
+      "loss": 0.0001,
       "step": 1029
     },
     {
-      "epoch": 1.5419161676646707,
-      "grad_norm": 0.048845460680296805,
-      "learning_rate": 1.0056590404522042e-05,
-      "loss": 0.0006,
+      "epoch": 1.6480000000000001,
+      "grad_norm": 0.0026716294435511516,
+      "learning_rate": 8.896301945718541e-06,
+      "loss": 0.0001,
       "step": 1030
     },
     {
-      "epoch": 1.5434131736526946,
-      "grad_norm": 0.09153026786032244,
-      "learning_rate": 1.0040421823190239e-05,
-      "loss": 0.0005,
+      "epoch": 1.6496,
+      "grad_norm": 0.03505078932135712,
+      "learning_rate": 8.879128688101749e-06,
+      "loss": 0.0002,
       "step": 1031
     },
     {
-      "epoch": 1.5449101796407185,
-      "grad_norm": 0.0979750349520669,
-      "learning_rate": 1.0024253136183997e-05,
-      "loss": 0.0008,
+      "epoch": 1.6512,
+      "grad_norm": 0.017837314438095914,
+      "learning_rate": 8.861958777576826e-06,
+      "loss": 0.0001,
       "step": 1032
     },
     {
-      "epoch": 1.5464071856287425,
-      "grad_norm": 0.12603703868213395,
-      "learning_rate": 1.000808438577298e-05,
-      "loss": 0.0007,
+      "epoch": 1.6528,
+      "grad_norm": 0.003291556083983862,
+      "learning_rate": 8.844792265415738e-06,
+      "loss": 0.0001,
       "step": 1033
     },
     {
-      "epoch": 1.5479041916167664,
-      "grad_norm": 0.09260218797278057,
-      "learning_rate": 9.991915614227022e-06,
-      "loss": 0.0006,
+      "epoch": 1.6543999999999999,
+      "grad_norm": 0.0078013423029036156,
+      "learning_rate": 8.827629202880294e-06,
+      "loss": 0.0001,
       "step": 1034
     },
     {
-      "epoch": 1.5494011976047903,
-      "grad_norm": 0.150535783399099,
-      "learning_rate": 9.975746863816006e-06,
-      "loss": 0.0008,
+      "epoch": 1.6560000000000001,
+      "grad_norm": 0.021478888567488523,
+      "learning_rate": 8.810469641222001e-06,
+      "loss": 0.0002,
       "step": 1035
     },
     {
-      "epoch": 1.5508982035928143,
-      "grad_norm": 0.0866141689732322,
-      "learning_rate": 9.959578176809764e-06,
-      "loss": 0.0014,
+      "epoch": 1.6576,
+      "grad_norm": 0.107404212993994,
+      "learning_rate": 8.793313631681915e-06,
+      "loss": 0.0015,
       "step": 1036
     },
     {
-      "epoch": 1.5523952095808382,
-      "grad_norm": 0.14207853844047355,
-      "learning_rate": 9.943409595477958e-06,
-      "loss": 0.0014,
+      "epoch": 1.6592,
+      "grad_norm": 0.005515426360191595,
+      "learning_rate": 8.776161225490488e-06,
+      "loss": 0.0001,
       "step": 1037
     },
     {
-      "epoch": 1.5538922155688621,
-      "grad_norm": 0.08207842243713191,
-      "learning_rate": 9.927241162089984e-06,
-      "loss": 0.0008,
+      "epoch": 1.6608,
+      "grad_norm": 0.018418735479344833,
+      "learning_rate": 8.759012473867407e-06,
+      "loss": 0.0001,
       "step": 1038
     },
     {
-      "epoch": 1.555389221556886,
-      "grad_norm": 0.14029627239900275,
-      "learning_rate": 9.911072918914834e-06,
-      "loss": 0.0028,
+      "epoch": 1.6623999999999999,
+      "grad_norm": 0.009539716928331653,
+      "learning_rate": 8.741867428021447e-06,
+      "loss": 0.0001,
       "step": 1039
     },
     {
-      "epoch": 1.55688622754491,
-      "grad_norm": 0.09422957915613028,
-      "learning_rate": 9.894904908221018e-06,
-      "loss": 0.0006,
+      "epoch": 1.6640000000000001,
+      "grad_norm": 0.09973663139996064,
+      "learning_rate": 8.72472613915032e-06,
+      "loss": 0.0004,
       "step": 1040
     },
     {
-      "epoch": 1.5583832335329342,
-      "grad_norm": 0.26480681530852446,
-      "learning_rate": 9.87873717227643e-06,
-      "loss": 0.002,
+      "epoch": 1.6656,
+      "grad_norm": 0.15761894212909275,
+      "learning_rate": 8.707588658440511e-06,
+      "loss": 0.0019,
       "step": 1041
     },
     {
-      "epoch": 1.5598802395209581,
-      "grad_norm": 0.18537320811580985,
-      "learning_rate": 9.862569753348255e-06,
-      "loss": 0.001,
+      "epoch": 1.6672,
+      "grad_norm": 0.010498595207018875,
+      "learning_rate": 8.690455037067142e-06,
+      "loss": 0.0001,
       "step": 1042
     },
     {
-      "epoch": 1.561377245508982,
-      "grad_norm": 0.12242399214233682,
-      "learning_rate": 9.846402693702833e-06,
-      "loss": 0.0009,
+      "epoch": 1.6688,
+      "grad_norm": 0.10526976223706591,
+      "learning_rate": 8.673325326193806e-06,
+      "loss": 0.0005,
       "step": 1043
     },
     {
-      "epoch": 1.562874251497006,
-      "grad_norm": 0.10685927064129254,
-      "learning_rate": 9.830236035605578e-06,
-      "loss": 0.001,
+      "epoch": 1.6703999999999999,
+      "grad_norm": 0.003618607966983597,
+      "learning_rate": 8.656199576972424e-06,
+      "loss": 0.0001,
       "step": 1044
     },
     {
-      "epoch": 1.56437125748503,
-      "grad_norm": 0.13222209724765782,
-      "learning_rate": 9.814069821320856e-06,
-      "loss": 0.0023,
+      "epoch": 1.6720000000000002,
+      "grad_norm": 0.012464316701956982,
+      "learning_rate": 8.639077840543078e-06,
+      "loss": 0.0001,
       "step": 1045
     },
     {
-      "epoch": 1.5658682634730539,
-      "grad_norm": 0.0968515135585454,
-      "learning_rate": 9.79790409311186e-06,
-      "loss": 0.0008,
+      "epoch": 1.6736,
+      "grad_norm": 0.0053543772659330035,
+      "learning_rate": 8.621960168033868e-06,
+      "loss": 0.0001,
       "step": 1046
     },
     {
-      "epoch": 1.5673652694610778,
-      "grad_norm": 0.05675857658075679,
-      "learning_rate": 9.781738893240526e-06,
-      "loss": 0.0005,
+      "epoch": 1.6752,
+      "grad_norm": 0.07531792817890605,
+      "learning_rate": 8.604846610560771e-06,
+      "loss": 0.0004,
       "step": 1047
     },
     {
-      "epoch": 1.568862275449102,
-      "grad_norm": 0.10548064797653442,
-      "learning_rate": 9.765574263967397e-06,
-      "loss": 0.001,
+      "epoch": 1.6768,
+      "grad_norm": 0.029371851157039955,
+      "learning_rate": 8.587737219227462e-06,
+      "loss": 0.0002,
       "step": 1048
     },
     {
-      "epoch": 1.5703592814371259,
-      "grad_norm": 0.14583652639833644,
-      "learning_rate": 9.749410247551534e-06,
-      "loss": 0.0022,
+      "epoch": 1.6784,
+      "grad_norm": 0.05109275328856202,
+      "learning_rate": 8.570632045125185e-06,
+      "loss": 0.0002,
       "step": 1049
     },
     {
-      "epoch": 1.5718562874251498,
-      "grad_norm": 0.10085574392360613,
-      "learning_rate": 9.733246886250384e-06,
-      "loss": 0.0008,
+      "epoch": 1.6800000000000002,
+      "grad_norm": 0.006421087021681481,
+      "learning_rate": 8.553531139332583e-06,
+      "loss": 0.0001,
       "step": 1050
     },
     {
-      "epoch": 1.5733532934131738,
-      "grad_norm": 0.16062888967169808,
-      "learning_rate": 9.717084222319698e-06,
-      "loss": 0.0024,
+      "epoch": 1.6816,
+      "grad_norm": 0.07963637652867744,
+      "learning_rate": 8.536434552915555e-06,
+      "loss": 0.0005,
       "step": 1051
     },
     {
-      "epoch": 1.5748502994011977,
-      "grad_norm": 0.06334101246324102,
-      "learning_rate": 9.700922298013388e-06,
-      "loss": 0.0006,
+      "epoch": 1.6832,
+      "grad_norm": 0.0017013180902574591,
+      "learning_rate": 8.519342336927106e-06,
+      "loss": 0.0001,
       "step": 1052
     },
     {
-      "epoch": 1.5763473053892216,
-      "grad_norm": 0.10858659881531557,
-      "learning_rate": 9.684761155583443e-06,
-      "loss": 0.001,
+      "epoch": 1.6848,
+      "grad_norm": 0.07287617820083703,
+      "learning_rate": 8.502254542407186e-06,
+      "loss": 0.0003,
       "step": 1053
     },
     {
-      "epoch": 1.5778443113772456,
-      "grad_norm": 0.05720430794355147,
-      "learning_rate": 9.668600837279808e-06,
-      "loss": 0.0003,
+      "epoch": 1.6864,
+      "grad_norm": 0.1463603814085734,
+      "learning_rate": 8.485171220382545e-06,
+      "loss": 0.0008,
       "step": 1054
     },
     {
-      "epoch": 1.5793413173652695,
-      "grad_norm": 0.18596258294319445,
-      "learning_rate": 9.652441385350258e-06,
-      "loss": 0.0008,
+      "epoch": 1.688,
+      "grad_norm": 0.0434357914290013,
+      "learning_rate": 8.468092421866575e-06,
+      "loss": 0.0002,
       "step": 1055
     },
     {
-      "epoch": 1.5808383233532934,
-      "grad_norm": 0.02161457070852507,
-      "learning_rate": 9.636282842040324e-06,
-      "loss": 0.0002,
+      "epoch": 1.6896,
+      "grad_norm": 0.0018126604638566501,
+      "learning_rate": 8.451018197859153e-06,
+      "loss": 0.0001,
       "step": 1056
     },
     {
-      "epoch": 1.5823353293413174,
-      "grad_norm": 0.12665814776438672,
-      "learning_rate": 9.620125249593145e-06,
-      "loss": 0.002,
+      "epoch": 1.6912,
+      "grad_norm": 0.024660687945750964,
+      "learning_rate": 8.433948599346516e-06,
+      "loss": 0.0001,
       "step": 1057
     },
     {
-      "epoch": 1.5838323353293413,
-      "grad_norm": 0.1479035747876992,
-      "learning_rate": 9.603968650249388e-06,
-      "loss": 0.0014,
+      "epoch": 1.6928,
+      "grad_norm": 0.02621191496190813,
+      "learning_rate": 8.41688367730107e-06,
+      "loss": 0.0002,
       "step": 1058
     },
     {
-      "epoch": 1.5853293413173652,
-      "grad_norm": 0.08875005354112726,
-      "learning_rate": 9.587813086247118e-06,
-      "loss": 0.0008,
+      "epoch": 1.6944,
+      "grad_norm": 0.003268281028403292,
+      "learning_rate": 8.399823482681263e-06,
+      "loss": 0.0001,
       "step": 1059
     },
     {
-      "epoch": 1.5868263473053892,
-      "grad_norm": 0.034598142239245726,
-      "learning_rate": 9.571658599821684e-06,
-      "loss": 0.0002,
+      "epoch": 1.696,
+      "grad_norm": 0.0019118513200916743,
+      "learning_rate": 8.382768066431427e-06,
+      "loss": 0.0001,
       "step": 1060
     },
     {
-      "epoch": 1.5883233532934131,
-      "grad_norm": 0.1382510341790496,
-      "learning_rate": 9.555505233205635e-06,
-      "loss": 0.0014,
+      "epoch": 1.6976,
+      "grad_norm": 0.004530054083164475,
+      "learning_rate": 8.36571747948162e-06,
+      "loss": 0.0001,
       "step": 1061
     },
     {
-      "epoch": 1.589820359281437,
-      "grad_norm": 0.13116321301263414,
-      "learning_rate": 9.539353028628577e-06,
-      "loss": 0.001,
+      "epoch": 1.6992,
+      "grad_norm": 0.0643530214986294,
+      "learning_rate": 8.348671772747488e-06,
+      "loss": 0.0002,
       "step": 1062
     },
     {
-      "epoch": 1.591317365269461,
-      "grad_norm": 0.05547568359029333,
-      "learning_rate": 9.523202028317095e-06,
-      "loss": 0.0007,
+      "epoch": 1.7008,
+      "grad_norm": 0.08236509208617357,
+      "learning_rate": 8.331630997130091e-06,
+      "loss": 0.0015,
       "step": 1063
     },
     {
-      "epoch": 1.592814371257485,
-      "grad_norm": 0.031905668491041365,
-      "learning_rate": 9.5070522744946e-06,
-      "loss": 0.0003,
+      "epoch": 1.7024,
+      "grad_norm": 0.0024735731034196246,
+      "learning_rate": 8.314595203515781e-06,
+      "loss": 0.0001,
       "step": 1064
     },
     {
-      "epoch": 1.5943113772455089,
-      "grad_norm": 0.008727577881278311,
-      "learning_rate": 9.490903809381273e-06,
+      "epoch": 1.704,
+      "grad_norm": 0.042430868716428864,
+      "learning_rate": 8.297564442776014e-06,
       "loss": 0.0001,
       "step": 1065
     },
     {
-      "epoch": 1.5958083832335328,
-      "grad_norm": 0.0712191739355828,
-      "learning_rate": 9.47475667519391e-06,
-      "loss": 0.0015,
+      "epoch": 1.7056,
+      "grad_norm": 0.018951879993768548,
+      "learning_rate": 8.280538765767236e-06,
+      "loss": 0.0001,
       "step": 1066
     },
     {
-      "epoch": 1.5973053892215567,
-      "grad_norm": 0.03749421756809963,
-      "learning_rate": 9.458610914145826e-06,
-      "loss": 0.001,
+      "epoch": 1.7072,
+      "grad_norm": 0.05325849886472776,
+      "learning_rate": 8.263518223330698e-06,
+      "loss": 0.0002,
       "step": 1067
     },
     {
-      "epoch": 1.5988023952095807,
-      "grad_norm": 0.02379835401791554,
-      "learning_rate": 9.442466568446755e-06,
-      "loss": 0.0003,
+      "epoch": 1.7088,
+      "grad_norm": 0.007651066709106985,
+      "learning_rate": 8.246502866292324e-06,
+      "loss": 0.0001,
       "step": 1068
     },
     {
-      "epoch": 1.6002994011976048,
-      "grad_norm": 0.12118141203303913,
-      "learning_rate": 9.426323680302722e-06,
-      "loss": 0.0013,
+      "epoch": 1.7104,
+      "grad_norm": 0.002899812605506385,
+      "learning_rate": 8.229492745462551e-06,
+      "loss": 0.0001,
       "step": 1069
     },
     {
-      "epoch": 1.6017964071856288,
-      "grad_norm": 0.12388549366182541,
-      "learning_rate": 9.410182291915956e-06,
-      "loss": 0.0009,
+      "epoch": 1.712,
+      "grad_norm": 0.01347456210816533,
+      "learning_rate": 8.212487911636185e-06,
+      "loss": 0.0001,
       "step": 1070
     },
     {
-      "epoch": 1.6032934131736527,
-      "grad_norm": 0.117253435172126,
-      "learning_rate": 9.39404244548474e-06,
-      "loss": 0.0013,
+      "epoch": 1.7136,
+      "grad_norm": 0.0023674577831386896,
+      "learning_rate": 8.195488415592238e-06,
+      "loss": 0.0001,
       "step": 1071
     },
     {
-      "epoch": 1.6047904191616766,
-      "grad_norm": 0.08211137201727235,
-      "learning_rate": 9.377904183203353e-06,
-      "loss": 0.0009,
+      "epoch": 1.7151999999999998,
+      "grad_norm": 0.006504806869534589,
+      "learning_rate": 8.17849430809379e-06,
+      "loss": 0.0001,
       "step": 1072
     },
     {
-      "epoch": 1.6062874251497006,
-      "grad_norm": 0.09352507617778973,
-      "learning_rate": 9.361767547261918e-06,
-      "loss": 0.0009,
+      "epoch": 1.7168,
+      "grad_norm": 0.004865458855715208,
+      "learning_rate": 8.161505639887818e-06,
+      "loss": 0.0001,
       "step": 1073
     },
     {
-      "epoch": 1.6077844311377245,
-      "grad_norm": 0.04451673559222652,
-      "learning_rate": 9.345632579846306e-06,
-      "loss": 0.0003,
+      "epoch": 1.7184,
+      "grad_norm": 0.015472810303115406,
+      "learning_rate": 8.144522461705067e-06,
+      "loss": 0.0001,
       "step": 1074
     },
     {
-      "epoch": 1.6092814371257484,
-      "grad_norm": 0.11115723113169731,
-      "learning_rate": 9.329499323138035e-06,
-      "loss": 0.0007,
+      "epoch": 1.72,
+      "grad_norm": 0.0913008252673765,
+      "learning_rate": 8.12754482425989e-06,
+      "loss": 0.0013,
       "step": 1075
     },
     {
-      "epoch": 1.6107784431137726,
-      "grad_norm": 0.046162578574199704,
-      "learning_rate": 9.313367819314136e-06,
-      "loss": 0.0005,
+      "epoch": 1.7216,
+      "grad_norm": 0.017356882464452193,
+      "learning_rate": 8.110572778250086e-06,
+      "loss": 0.0001,
       "step": 1076
     },
     {
-      "epoch": 1.6122754491017965,
-      "grad_norm": 0.04801631426459736,
-      "learning_rate": 9.297238110547075e-06,
-      "loss": 0.0006,
+      "epoch": 1.7231999999999998,
+      "grad_norm": 0.12843015296778934,
+      "learning_rate": 8.09360637435676e-06,
+      "loss": 0.0013,
       "step": 1077
     },
     {
-      "epoch": 1.6137724550898205,
-      "grad_norm": 0.042030541819221175,
-      "learning_rate": 9.281110239004607e-06,
-      "loss": 0.0006,
+      "epoch": 1.7248,
+      "grad_norm": 0.004656608275681594,
+      "learning_rate": 8.076645663244168e-06,
+      "loss": 0.0001,
       "step": 1078
     },
     {
-      "epoch": 1.6152694610778444,
-      "grad_norm": 0.04359620580995584,
-      "learning_rate": 9.264984246849699e-06,
-      "loss": 0.0008,
+      "epoch": 1.7264,
+      "grad_norm": 0.004636984115445992,
+      "learning_rate": 8.05969069555957e-06,
+      "loss": 0.0001,
       "step": 1079
     },
     {
-      "epoch": 1.6167664670658684,
-      "grad_norm": 0.12249935790486885,
-      "learning_rate": 9.2488601762404e-06,
-      "loss": 0.0012,
+      "epoch": 1.728,
+      "grad_norm": 0.011446814949245822,
+      "learning_rate": 8.042741521933071e-06,
+      "loss": 0.0001,
       "step": 1080
     },
     {
-      "epoch": 1.6182634730538923,
-      "grad_norm": 0.08869067151066333,
-      "learning_rate": 9.23273806932973e-06,
-      "loss": 0.0006,
+      "epoch": 1.7296,
+      "grad_norm": 0.005402631192077296,
+      "learning_rate": 8.025798192977482e-06,
+      "loss": 0.0001,
       "step": 1081
     },
     {
-      "epoch": 1.6197604790419162,
-      "grad_norm": 0.04713954076892878,
-      "learning_rate": 9.216617968265587e-06,
-      "loss": 0.0004,
+      "epoch": 1.7311999999999999,
+      "grad_norm": 0.025991095533090636,
+      "learning_rate": 8.008860759288148e-06,
+      "loss": 0.0002,
       "step": 1082
     },
     {
-      "epoch": 1.6212574850299402,
-      "grad_norm": 0.022606337617157864,
-      "learning_rate": 9.200499915190609e-06,
+      "epoch": 1.7328000000000001,
+      "grad_norm": 0.01830933422716461,
+      "learning_rate": 7.991929271442817e-06,
       "loss": 0.0002,
       "step": 1083
     },
     {
-      "epoch": 1.622754491017964,
-      "grad_norm": 0.044196202525303686,
-      "learning_rate": 9.184383952242095e-06,
-      "loss": 0.0004,
+      "epoch": 1.7344,
+      "grad_norm": 0.0021080792693244223,
+      "learning_rate": 7.975003780001486e-06,
+      "loss": 0.0001,
       "step": 1084
     },
     {
-      "epoch": 1.624251497005988,
-      "grad_norm": 0.04585085088431671,
-      "learning_rate": 9.16827012155187e-06,
-      "loss": 0.0003,
+      "epoch": 1.736,
+      "grad_norm": 0.013074921500412782,
+      "learning_rate": 7.958084335506239e-06,
+      "loss": 0.0001,
       "step": 1085
     },
     {
-      "epoch": 1.625748502994012,
-      "grad_norm": 0.1099988952217994,
-      "learning_rate": 9.152158465246192e-06,
-      "loss": 0.0012,
+      "epoch": 1.7376,
+      "grad_norm": 0.0079255343223726,
+      "learning_rate": 7.941170988481108e-06,
+      "loss": 0.0001,
       "step": 1086
     },
     {
-      "epoch": 1.627245508982036,
-      "grad_norm": 0.21437678680661543,
-      "learning_rate": 9.136049025445633e-06,
-      "loss": 0.0024,
+      "epoch": 1.7391999999999999,
+      "grad_norm": 0.004299069808370429,
+      "learning_rate": 7.924263789431913e-06,
+      "loss": 0.0001,
       "step": 1087
     },
     {
-      "epoch": 1.6287425149700598,
-      "grad_norm": 0.07403926272991478,
-      "learning_rate": 9.11994184426496e-06,
-      "loss": 0.0005,
+      "epoch": 1.7408000000000001,
+      "grad_norm": 0.0027531772000943576,
+      "learning_rate": 7.907362788846116e-06,
+      "loss": 0.0001,
       "step": 1088
     },
     {
-      "epoch": 1.6302395209580838,
-      "grad_norm": 0.059002757939724,
-      "learning_rate": 9.10383696381305e-06,
-      "loss": 0.0007,
+      "epoch": 1.7424,
+      "grad_norm": 0.007412058248624506,
+      "learning_rate": 7.89046803719267e-06,
+      "loss": 0.0001,
       "step": 1089
     },
     {
-      "epoch": 1.6317365269461077,
-      "grad_norm": 0.12675376681424935,
-      "learning_rate": 9.087734426192753e-06,
-      "loss": 0.001,
+      "epoch": 1.744,
+      "grad_norm": 0.014519619506843367,
+      "learning_rate": 7.873579584921869e-06,
+      "loss": 0.0001,
       "step": 1090
     },
     {
-      "epoch": 1.6332335329341316,
-      "grad_norm": 0.13626773883445795,
-      "learning_rate": 9.07163427350081e-06,
-      "loss": 0.0011,
+      "epoch": 1.7456,
+      "grad_norm": 0.005921967451427256,
+      "learning_rate": 7.856697482465195e-06,
+      "loss": 0.0001,
       "step": 1091
     },
     {
-      "epoch": 1.6347305389221556,
-      "grad_norm": 0.09619385565164779,
-      "learning_rate": 9.055536547827708e-06,
-      "loss": 0.0012,
+      "epoch": 1.7471999999999999,
+      "grad_norm": 0.007296203326684224,
+      "learning_rate": 7.839821780235168e-06,
+      "loss": 0.0001,
       "step": 1092
     },
     {
-      "epoch": 1.6362275449101795,
-      "grad_norm": 0.07077546546688278,
-      "learning_rate": 9.0394412912576e-06,
-      "loss": 0.0008,
+      "epoch": 1.7488000000000001,
+      "grad_norm": 0.005280268273475599,
+      "learning_rate": 7.822952528625192e-06,
+      "loss": 0.0001,
       "step": 1093
     },
     {
-      "epoch": 1.6377245508982035,
-      "grad_norm": 0.11742541713396401,
-      "learning_rate": 9.023348545868187e-06,
-      "loss": 0.0017,
+      "epoch": 1.7504,
+      "grad_norm": 0.004460160305069306,
+      "learning_rate": 7.806089778009421e-06,
+      "loss": 0.0001,
       "step": 1094
     },
     {
-      "epoch": 1.6392215568862274,
-      "grad_norm": 0.05392358126836917,
-      "learning_rate": 9.007258353730594e-06,
-      "loss": 0.0005,
+      "epoch": 1.752,
+      "grad_norm": 0.002065351387446571,
+      "learning_rate": 7.789233578742583e-06,
+      "loss": 0.0001,
       "step": 1095
     },
     {
-      "epoch": 1.6407185628742516,
-      "grad_norm": 0.05286986458798405,
-      "learning_rate": 8.99117075690928e-06,
-      "loss": 0.0004,
+      "epoch": 1.7536,
+      "grad_norm": 0.016340239360333648,
+      "learning_rate": 7.77238398115985e-06,
+      "loss": 0.0001,
       "step": 1096
     },
     {
-      "epoch": 1.6422155688622755,
-      "grad_norm": 0.1037025844104848,
-      "learning_rate": 8.975085797461915e-06,
-      "loss": 0.0009,
+      "epoch": 1.7551999999999999,
+      "grad_norm": 0.0057828584962989395,
+      "learning_rate": 7.755541035576677e-06,
+      "loss": 0.0001,
       "step": 1097
     },
     {
-      "epoch": 1.6437125748502994,
-      "grad_norm": 0.026409960690047918,
-      "learning_rate": 8.959003517439282e-06,
-      "loss": 0.0003,
+      "epoch": 1.7568000000000001,
+      "grad_norm": 0.0027678901722254182,
+      "learning_rate": 7.738704792288654e-06,
+      "loss": 0.0001,
       "step": 1098
     },
     {
-      "epoch": 1.6452095808383234,
-      "grad_norm": 0.06352856914344773,
-      "learning_rate": 8.942923958885142e-06,
-      "loss": 0.0003,
+      "epoch": 1.7584,
+      "grad_norm": 0.09580771102289547,
+      "learning_rate": 7.721875301571359e-06,
+      "loss": 0.0023,
       "step": 1099
     },
     {
-      "epoch": 1.6467065868263473,
-      "grad_norm": 0.061764806505025756,
-      "learning_rate": 8.92684716383616e-06,
-      "loss": 0.0008,
+      "epoch": 1.76,
+      "grad_norm": 0.0038614718887813606,
+      "learning_rate": 7.705052613680212e-06,
+      "loss": 0.0001,
       "step": 1100
     },
     {
-      "epoch": 1.6482035928143712,
-      "grad_norm": 0.05291688814800548,
-      "learning_rate": 8.910773174321765e-06,
-      "loss": 0.0007,
+      "epoch": 1.7616,
+      "grad_norm": 0.0023181148268299626,
+      "learning_rate": 7.688236778850307e-06,
+      "loss": 0.0,
       "step": 1101
     },
     {
-      "epoch": 1.6497005988023952,
-      "grad_norm": 0.07133556607963744,
-      "learning_rate": 8.894702032364053e-06,
-      "loss": 0.0011,
+      "epoch": 1.7631999999999999,
+      "grad_norm": 0.15826582745285653,
+      "learning_rate": 7.671427847296274e-06,
+      "loss": 0.0028,
       "step": 1102
     },
     {
-      "epoch": 1.6511976047904193,
-      "grad_norm": 0.09968542393075754,
-      "learning_rate": 8.878633779977686e-06,
-      "loss": 0.0019,
+      "epoch": 1.7648000000000001,
+      "grad_norm": 0.024941432115363694,
+      "learning_rate": 7.654625869212147e-06,
+      "loss": 0.0002,
       "step": 1103
     },
     {
-      "epoch": 1.6526946107784433,
-      "grad_norm": 0.03251881336576024,
-      "learning_rate": 8.862568459169755e-06,
-      "loss": 0.0003,
+      "epoch": 1.7664,
+      "grad_norm": 0.010491559643191137,
+      "learning_rate": 7.637830894771176e-06,
+      "loss": 0.0001,
       "step": 1104
     },
     {
-      "epoch": 1.6541916167664672,
-      "grad_norm": 0.06370247682649403,
-      "learning_rate": 8.846506111939699e-06,
+      "epoch": 1.768,
+      "grad_norm": 0.10207548229257392,
+      "learning_rate": 7.621042974125701e-06,
       "loss": 0.0005,
       "step": 1105
     },
     {
-      "epoch": 1.6556886227544911,
-      "grad_norm": 0.06213918678886638,
-      "learning_rate": 8.830446780279175e-06,
-      "loss": 0.0005,
+      "epoch": 1.7696,
+      "grad_norm": 0.005440385147590804,
+      "learning_rate": 7.604262157407008e-06,
+      "loss": 0.0001,
       "step": 1106
     },
     {
-      "epoch": 1.657185628742515,
-      "grad_norm": 0.10415475247431645,
-      "learning_rate": 8.814390506171967e-06,
-      "loss": 0.0013,
+      "epoch": 1.7711999999999999,
+      "grad_norm": 0.3752625791888844,
+      "learning_rate": 7.587488494725157e-06,
+      "loss": 0.0022,
       "step": 1107
     },
     {
-      "epoch": 1.658682634730539,
-      "grad_norm": 0.07744622474255668,
-      "learning_rate": 8.79833733159386e-06,
-      "loss": 0.002,
+      "epoch": 1.7728000000000002,
+      "grad_norm": 0.009505820387901138,
+      "learning_rate": 7.570722036168855e-06,
+      "loss": 0.0001,
       "step": 1108
     },
     {
-      "epoch": 1.660179640718563,
-      "grad_norm": 0.04324172934252314,
-      "learning_rate": 8.78228729851253e-06,
-      "loss": 0.0004,
+      "epoch": 1.7744,
+      "grad_norm": 0.005630646707161456,
+      "learning_rate": 7.553962831805291e-06,
+      "loss": 0.0001,
       "step": 1109
     },
     {
-      "epoch": 1.6616766467065869,
-      "grad_norm": 0.1796309695730341,
-      "learning_rate": 8.766240448887454e-06,
-      "loss": 0.002,
+      "epoch": 1.776,
+      "grad_norm": 0.0027598416824326363,
+      "learning_rate": 7.537210931679988e-06,
+      "loss": 0.0001,
       "step": 1110
     },
     {
-      "epoch": 1.6631736526946108,
-      "grad_norm": 0.049210623724537356,
-      "learning_rate": 8.750196824669765e-06,
-      "loss": 0.0008,
+      "epoch": 1.7776,
+      "grad_norm": 0.07417331973262634,
+      "learning_rate": 7.520466385816672e-06,
+      "loss": 0.0004,
       "step": 1111
     },
     {
-      "epoch": 1.6646706586826348,
-      "grad_norm": 0.06374661903865328,
-      "learning_rate": 8.73415646780219e-06,
-      "loss": 0.0005,
+      "epoch": 1.7792,
+      "grad_norm": 0.07266576078156141,
+      "learning_rate": 7.5037292442170865e-06,
+      "loss": 0.0004,
       "step": 1112
     },
     {
-      "epoch": 1.6661676646706587,
-      "grad_norm": 0.0678542988687008,
-      "learning_rate": 8.718119420218894e-06,
-      "loss": 0.0006,
+      "epoch": 1.7808000000000002,
+      "grad_norm": 0.1274908087744277,
+      "learning_rate": 7.48699955686089e-06,
+      "loss": 0.0021,
       "step": 1113
     },
     {
-      "epoch": 1.6676646706586826,
-      "grad_norm": 0.05131894267995365,
-      "learning_rate": 8.702085723845402e-06,
-      "loss": 0.0005,
+      "epoch": 1.7824,
+      "grad_norm": 0.19971373425751254,
+      "learning_rate": 7.470277373705461e-06,
+      "loss": 0.0012,
       "step": 1114
     },
     {
-      "epoch": 1.6691616766467066,
-      "grad_norm": 0.11138493277856698,
-      "learning_rate": 8.686055420598474e-06,
-      "loss": 0.0011,
+      "epoch": 1.784,
+      "grad_norm": 0.027723381814621518,
+      "learning_rate": 7.453562744685779e-06,
+      "loss": 0.0001,
       "step": 1115
     },
     {
-      "epoch": 1.6706586826347305,
-      "grad_norm": 0.04388393803892101,
-      "learning_rate": 8.670028552385994e-06,
-      "loss": 0.0004,
+      "epoch": 1.7856,
+      "grad_norm": 0.0185271416119447,
+      "learning_rate": 7.4368557197142596e-06,
+      "loss": 0.0002,
       "step": 1116
     },
     {
-      "epoch": 1.6721556886227544,
-      "grad_norm": 0.11972182187525221,
-      "learning_rate": 8.654005161106877e-06,
-      "loss": 0.0011,
+      "epoch": 1.7872,
+      "grad_norm": 0.021960034189195798,
+      "learning_rate": 7.420156348680621e-06,
+      "loss": 0.0001,
       "step": 1117
     },
     {
-      "epoch": 1.6736526946107784,
-      "grad_norm": 0.126003813560071,
-      "learning_rate": 8.637985288650939e-06,
-      "loss": 0.0009,
+      "epoch": 1.7888,
+      "grad_norm": 0.10064763080851377,
+      "learning_rate": 7.4034646814517155e-06,
+      "loss": 0.0004,
       "step": 1118
     },
     {
-      "epoch": 1.6751497005988023,
-      "grad_norm": 0.10943077524662276,
-      "learning_rate": 8.621968976898804e-06,
-      "loss": 0.0019,
+      "epoch": 1.7904,
+      "grad_norm": 0.004508512269411402,
+      "learning_rate": 7.3867807678713965e-06,
+      "loss": 0.0001,
       "step": 1119
     },
     {
-      "epoch": 1.6766467065868262,
-      "grad_norm": 0.06513537092099632,
-      "learning_rate": 8.605956267721778e-06,
-      "loss": 0.0007,
+      "epoch": 1.792,
+      "grad_norm": 0.027868998430493137,
+      "learning_rate": 7.3701046577603605e-06,
+      "loss": 0.0001,
       "step": 1120
     },
     {
-      "epoch": 1.6781437125748502,
-      "grad_norm": 0.13419673873932153,
-      "learning_rate": 8.589947202981755e-06,
-      "loss": 0.0008,
+      "epoch": 1.7936,
+      "grad_norm": 0.01681347007094859,
+      "learning_rate": 7.353436400916006e-06,
+      "loss": 0.0001,
       "step": 1121
     },
     {
-      "epoch": 1.6796407185628741,
-      "grad_norm": 0.04079284884695222,
-      "learning_rate": 8.573941824531109e-06,
-      "loss": 0.0003,
+      "epoch": 1.7952,
+      "grad_norm": 0.03894472881247136,
+      "learning_rate": 7.336776047112277e-06,
+      "loss": 0.0002,
       "step": 1122
     },
     {
-      "epoch": 1.681137724550898,
-      "grad_norm": 0.0368419755710961,
-      "learning_rate": 8.557940174212555e-06,
-      "loss": 0.0003,
+      "epoch": 1.7968,
+      "grad_norm": 0.101783856638485,
+      "learning_rate": 7.32012364609952e-06,
+      "loss": 0.0016,
       "step": 1123
     },
     {
-      "epoch": 1.6826347305389222,
-      "grad_norm": 0.11915015080619738,
-      "learning_rate": 8.541942293859086e-06,
-      "loss": 0.0015,
+      "epoch": 1.7984,
+      "grad_norm": 0.002825888683908071,
+      "learning_rate": 7.303479247604333e-06,
+      "loss": 0.0001,
       "step": 1124
     },
     {
-      "epoch": 1.6841317365269461,
-      "grad_norm": 0.11280338061106095,
-      "learning_rate": 8.52594822529382e-06,
-      "loss": 0.001,
+      "epoch": 1.8,
+      "grad_norm": 0.006401141464077191,
+      "learning_rate": 7.286842901329413e-06,
+      "loss": 0.0001,
       "step": 1125
     },
     {
-      "epoch": 1.68562874251497,
-      "grad_norm": 0.11129477072458856,
-      "learning_rate": 8.509958010329931e-06,
-      "loss": 0.0009,
+      "epoch": 1.8016,
+      "grad_norm": 0.05910438388349573,
+      "learning_rate": 7.270214656953415e-06,
+      "loss": 0.0003,
       "step": 1126
     },
     {
-      "epoch": 1.687125748502994,
-      "grad_norm": 0.11817747887791782,
-      "learning_rate": 8.49397169077049e-06,
-      "loss": 0.0011,
+      "epoch": 1.8032,
+      "grad_norm": 0.003062913108399674,
+      "learning_rate": 7.253594564130804e-06,
+      "loss": 0.0001,
       "step": 1127
     },
     {
-      "epoch": 1.688622754491018,
-      "grad_norm": 0.09555692890154081,
-      "learning_rate": 8.477989308408409e-06,
-      "loss": 0.0014,
+      "epoch": 1.8048,
+      "grad_norm": 0.015153952472531125,
+      "learning_rate": 7.236982672491699e-06,
+      "loss": 0.0001,
       "step": 1128
     },
     {
-      "epoch": 1.6901197604790419,
-      "grad_norm": 0.06104674622378712,
-      "learning_rate": 8.462010905026298e-06,
-      "loss": 0.0004,
+      "epoch": 1.8064,
+      "grad_norm": 0.011823113691139737,
+      "learning_rate": 7.22037903164173e-06,
+      "loss": 0.0001,
       "step": 1129
     },
     {
-      "epoch": 1.6916167664670658,
-      "grad_norm": 0.11401940154366785,
-      "learning_rate": 8.44603652239636e-06,
-      "loss": 0.0009,
+      "epoch": 1.808,
+      "grad_norm": 0.003639482658779054,
+      "learning_rate": 7.203783691161883e-06,
+      "loss": 0.0001,
       "step": 1130
     },
     {
-      "epoch": 1.69311377245509,
-      "grad_norm": 0.1359921086438187,
-      "learning_rate": 8.430066202280302e-06,
-      "loss": 0.0009,
+      "epoch": 1.8096,
+      "grad_norm": 0.04199695493257313,
+      "learning_rate": 7.187196700608373e-06,
+      "loss": 0.0002,
       "step": 1131
     },
     {
-      "epoch": 1.694610778443114,
-      "grad_norm": 0.07677903643974247,
-      "learning_rate": 8.414099986429188e-06,
-      "loss": 0.0005,
+      "epoch": 1.8112,
+      "grad_norm": 0.16656344026191952,
+      "learning_rate": 7.170618109512465e-06,
+      "loss": 0.0014,
       "step": 1132
     },
     {
-      "epoch": 1.6961077844311379,
-      "grad_norm": 0.04839849020100479,
-      "learning_rate": 8.398137916583371e-06,
-      "loss": 0.0003,
+      "epoch": 1.8128,
+      "grad_norm": 0.017200741298937835,
+      "learning_rate": 7.154047967380353e-06,
+      "loss": 0.0001,
       "step": 1133
     },
     {
-      "epoch": 1.6976047904191618,
-      "grad_norm": 0.12381943481528089,
-      "learning_rate": 8.382180034472353e-06,
-      "loss": 0.0008,
+      "epoch": 1.8144,
+      "grad_norm": 0.015394194260956718,
+      "learning_rate": 7.137486323692994e-06,
+      "loss": 0.0001,
       "step": 1134
     },
     {
-      "epoch": 1.6991017964071857,
-      "grad_norm": 0.07469322717802963,
-      "learning_rate": 8.366226381814698e-06,
-      "loss": 0.0007,
+      "epoch": 1.8159999999999998,
+      "grad_norm": 0.0393247070778465,
+      "learning_rate": 7.120933227905971e-06,
+      "loss": 0.0002,
       "step": 1135
     },
     {
-      "epoch": 1.7005988023952097,
-      "grad_norm": 0.06504097124135295,
-      "learning_rate": 8.350277000317911e-06,
-      "loss": 0.0006,
+      "epoch": 1.8176,
+      "grad_norm": 0.004208374779893613,
+      "learning_rate": 7.104388729449338e-06,
+      "loss": 0.0001,
       "step": 1136
     },
     {
-      "epoch": 1.7020958083832336,
-      "grad_norm": 0.05289705787630766,
-      "learning_rate": 8.334331931678317e-06,
-      "loss": 0.0005,
+      "epoch": 1.8192,
+      "grad_norm": 0.004585551477784112,
+      "learning_rate": 7.0878528777274814e-06,
+      "loss": 0.0001,
       "step": 1137
     },
     {
-      "epoch": 1.7035928143712575,
-      "grad_norm": 0.08661338942838477,
-      "learning_rate": 8.31839121758099e-06,
-      "loss": 0.0015,
+      "epoch": 1.8208,
+      "grad_norm": 0.14808830235231965,
+      "learning_rate": 7.0713257221189635e-06,
+      "loss": 0.0008,
       "step": 1138
     },
     {
-      "epoch": 1.7050898203592815,
-      "grad_norm": 0.08767555555868468,
-      "learning_rate": 8.302454899699597e-06,
-      "loss": 0.0006,
+      "epoch": 1.8224,
+      "grad_norm": 0.1436161753324329,
+      "learning_rate": 7.05480731197638e-06,
+      "loss": 0.0004,
       "step": 1139
     },
     {
-      "epoch": 1.7065868263473054,
-      "grad_norm": 0.09256232285744476,
-      "learning_rate": 8.286523019696327e-06,
-      "loss": 0.001,
+      "epoch": 1.8239999999999998,
+      "grad_norm": 0.0030378585652498986,
+      "learning_rate": 7.0382976966262065e-06,
+      "loss": 0.0,
       "step": 1140
     },
     {
-      "epoch": 1.7080838323353293,
-      "grad_norm": 0.053608939965644864,
-      "learning_rate": 8.270595619221762e-06,
-      "loss": 0.0005,
+      "epoch": 1.8256000000000001,
+      "grad_norm": 0.006585475630816729,
+      "learning_rate": 7.021796925368667e-06,
+      "loss": 0.0001,
       "step": 1141
     },
     {
-      "epoch": 1.7095808383233533,
-      "grad_norm": 0.03018588325618429,
-      "learning_rate": 8.254672739914771e-06,
-      "loss": 0.0005,
+      "epoch": 1.8272,
+      "grad_norm": 0.22474168720057988,
+      "learning_rate": 7.005305047477566e-06,
+      "loss": 0.0029,
       "step": 1142
     },
     {
-      "epoch": 1.7110778443113772,
-      "grad_norm": 0.01313607644598902,
-      "learning_rate": 8.238754423402411e-06,
+      "epoch": 1.8288,
+      "grad_norm": 0.0301613637954848,
+      "learning_rate": 6.988822112200157e-06,
       "loss": 0.0001,
       "step": 1143
     },
     {
-      "epoch": 1.7125748502994012,
-      "grad_norm": 0.12588799971374123,
-      "learning_rate": 8.222840711299798e-06,
-      "loss": 0.0008,
+      "epoch": 1.8304,
+      "grad_norm": 0.004656344661469704,
+      "learning_rate": 6.9723481687569836e-06,
+      "loss": 0.0001,
       "step": 1144
     },
     {
-      "epoch": 1.714071856287425,
-      "grad_norm": 0.08395380803153646,
-      "learning_rate": 8.206931645210025e-06,
-      "loss": 0.0007,
+      "epoch": 1.8319999999999999,
+      "grad_norm": 0.020491245770373548,
+      "learning_rate": 6.955883266341741e-06,
+      "loss": 0.0001,
       "step": 1145
     },
     {
-      "epoch": 1.715568862275449,
-      "grad_norm": 0.05648989201497437,
-      "learning_rate": 8.191027266724026e-06,
-      "loss": 0.0005,
+      "epoch": 1.8336000000000001,
+      "grad_norm": 0.0033176458831527247,
+      "learning_rate": 6.939427454121128e-06,
+      "loss": 0.0001,
       "step": 1146
     },
     {
-      "epoch": 1.717065868263473,
-      "grad_norm": 0.299079726997726,
-      "learning_rate": 8.175127617420489e-06,
-      "loss": 0.0038,
+      "epoch": 1.8352,
+      "grad_norm": 0.0012845644986009292,
+      "learning_rate": 6.9229807812346985e-06,
+      "loss": 0.0,
       "step": 1147
     },
     {
-      "epoch": 1.718562874251497,
-      "grad_norm": 0.2211219838190909,
-      "learning_rate": 8.15923273886574e-06,
-      "loss": 0.0017,
+      "epoch": 1.8368,
+      "grad_norm": 0.004031855277477734,
+      "learning_rate": 6.9065432967947145e-06,
+      "loss": 0.0001,
       "step": 1148
     },
     {
-      "epoch": 1.7200598802395208,
-      "grad_norm": 0.13423753642877773,
-      "learning_rate": 8.14334267261362e-06,
-      "loss": 0.0013,
+      "epoch": 1.8384,
+      "grad_norm": 0.10942866351129674,
+      "learning_rate": 6.890115049885995e-06,
+      "loss": 0.001,
       "step": 1149
     },
     {
-      "epoch": 1.7215568862275448,
-      "grad_norm": 0.021323410927144678,
-      "learning_rate": 8.127457460205403e-06,
-      "loss": 0.0002,
+      "epoch": 1.8399999999999999,
+      "grad_norm": 0.034646498478789645,
+      "learning_rate": 6.8736960895657854e-06,
+      "loss": 0.0001,
       "step": 1150
     },
     {
-      "epoch": 1.7230538922155687,
-      "grad_norm": 0.10537068848194503,
-      "learning_rate": 8.111577143169667e-06,
-      "loss": 0.0009,
+      "epoch": 1.8416000000000001,
+      "grad_norm": 0.004328068081161076,
+      "learning_rate": 6.85728646486359e-06,
+      "loss": 0.0001,
       "step": 1151
     },
     {
-      "epoch": 1.7245508982035929,
-      "grad_norm": 0.16469004106932947,
-      "learning_rate": 8.095701763022197e-06,
-      "loss": 0.0013,
+      "epoch": 1.8432,
+      "grad_norm": 0.0045096309471346265,
+      "learning_rate": 6.840886224781039e-06,
+      "loss": 0.0001,
       "step": 1152
     },
     {
-      "epoch": 1.7260479041916168,
-      "grad_norm": 0.17092342034188415,
-      "learning_rate": 8.079831361265862e-06,
-      "loss": 0.0017,
+      "epoch": 1.8448,
+      "grad_norm": 0.14523052171148082,
+      "learning_rate": 6.824495418291741e-06,
+      "loss": 0.0007,
       "step": 1153
     },
     {
-      "epoch": 1.7275449101796407,
-      "grad_norm": 0.06779177154424461,
-      "learning_rate": 8.063965979390526e-06,
-      "loss": 0.0006,
+      "epoch": 1.8464,
+      "grad_norm": 0.00296018643526749,
+      "learning_rate": 6.8081140943411296e-06,
+      "loss": 0.0001,
       "step": 1154
     },
     {
-      "epoch": 1.7290419161676647,
-      "grad_norm": 0.1236163086259948,
-      "learning_rate": 8.048105658872928e-06,
-      "loss": 0.0019,
+      "epoch": 1.8479999999999999,
+      "grad_norm": 0.03137236634318416,
+      "learning_rate": 6.791742301846325e-06,
+      "loss": 0.0002,
       "step": 1155
     },
     {
-      "epoch": 1.7305389221556886,
-      "grad_norm": 0.12610975535688002,
-      "learning_rate": 8.032250441176566e-06,
-      "loss": 0.0028,
+      "epoch": 1.8496000000000001,
+      "grad_norm": 0.0012996113765978107,
+      "learning_rate": 6.775380089695986e-06,
+      "loss": 0.0,
       "step": 1156
     },
     {
-      "epoch": 1.7320359281437125,
-      "grad_norm": 0.18941444937914406,
-      "learning_rate": 8.01640036775161e-06,
-      "loss": 0.0012,
+      "epoch": 1.8512,
+      "grad_norm": 0.004411509861164856,
+      "learning_rate": 6.759027506750159e-06,
+      "loss": 0.0001,
       "step": 1157
     },
     {
-      "epoch": 1.7335329341317365,
-      "grad_norm": 0.0831881723747311,
-      "learning_rate": 8.000555480034774e-06,
-      "loss": 0.0006,
+      "epoch": 1.8528,
+      "grad_norm": 0.014057183121189027,
+      "learning_rate": 6.742684601840142e-06,
+      "loss": 0.0001,
       "step": 1158
     },
     {
-      "epoch": 1.7350299401197606,
-      "grad_norm": 0.11838944511546944,
-      "learning_rate": 7.984715819449219e-06,
-      "loss": 0.0009,
+      "epoch": 1.8544,
+      "grad_norm": 0.0038426129938542576,
+      "learning_rate": 6.726351423768323e-06,
+      "loss": 0.0001,
       "step": 1159
     },
     {
-      "epoch": 1.7365269461077846,
-      "grad_norm": 0.03888943100450595,
-      "learning_rate": 7.968881427404433e-06,
-      "loss": 0.0004,
+      "epoch": 1.8559999999999999,
+      "grad_norm": 0.03527688711230369,
+      "learning_rate": 6.710028021308061e-06,
+      "loss": 0.0001,
       "step": 1160
     },
     {
-      "epoch": 1.7380239520958085,
-      "grad_norm": 0.04784679025928149,
-      "learning_rate": 7.953052345296141e-06,
-      "loss": 0.0004,
+      "epoch": 1.8576000000000001,
+      "grad_norm": 0.014309876719334954,
+      "learning_rate": 6.693714443203507e-06,
+      "loss": 0.0001,
       "step": 1161
     },
     {
-      "epoch": 1.7395209580838324,
-      "grad_norm": 0.09465522756842437,
-      "learning_rate": 7.937228614506183e-06,
-      "loss": 0.0011,
+      "epoch": 1.8592,
+      "grad_norm": 0.003078997197109562,
+      "learning_rate": 6.677410738169485e-06,
+      "loss": 0.0,
       "step": 1162
     },
     {
-      "epoch": 1.7410179640718564,
-      "grad_norm": 0.02839056527145619,
-      "learning_rate": 7.921410276402403e-06,
-      "loss": 0.0002,
+      "epoch": 1.8608,
+      "grad_norm": 0.08391880562324879,
+      "learning_rate": 6.661116954891329e-06,
+      "loss": 0.0004,
       "step": 1163
     },
     {
-      "epoch": 1.7425149700598803,
-      "grad_norm": 0.11897825771964533,
-      "learning_rate": 7.905597372338558e-06,
-      "loss": 0.0015,
+      "epoch": 1.8624,
+      "grad_norm": 0.01101824013877054,
+      "learning_rate": 6.644833142024752e-06,
+      "loss": 0.0001,
       "step": 1164
     },
     {
-      "epoch": 1.7440119760479043,
-      "grad_norm": 0.13433046182833272,
-      "learning_rate": 7.889789943654185e-06,
-      "loss": 0.0014,
+      "epoch": 1.8639999999999999,
+      "grad_norm": 0.002168915003192342,
+      "learning_rate": 6.62855934819569e-06,
+      "loss": 0.0,
       "step": 1165
     },
     {
-      "epoch": 1.7455089820359282,
-      "grad_norm": 0.042192664781619316,
-      "learning_rate": 7.87398803167452e-06,
-      "loss": 0.0003,
+      "epoch": 1.8656000000000001,
+      "grad_norm": 0.03937365515123593,
+      "learning_rate": 6.612295622000162e-06,
+      "loss": 0.0002,
       "step": 1166
     },
     {
-      "epoch": 1.7470059880239521,
-      "grad_norm": 0.09786003136143633,
-      "learning_rate": 7.858191677710364e-06,
-      "loss": 0.0018,
+      "epoch": 1.8672,
+      "grad_norm": 0.005212065805285967,
+      "learning_rate": 6.59604201200412e-06,
+      "loss": 0.0001,
       "step": 1167
     },
     {
-      "epoch": 1.748502994011976,
-      "grad_norm": 0.1278807783300935,
-      "learning_rate": 7.842400923058e-06,
-      "loss": 0.0012,
+      "epoch": 1.8688,
+      "grad_norm": 0.009421370870975533,
+      "learning_rate": 6.579798566743314e-06,
+      "loss": 0.0001,
       "step": 1168
     },
     {
-      "epoch": 1.75,
-      "grad_norm": 0.18490520684949707,
-      "learning_rate": 7.826615808999066e-06,
-      "loss": 0.0015,
+      "epoch": 1.8704,
+      "grad_norm": 0.037703987007789874,
+      "learning_rate": 6.563565334723134e-06,
+      "loss": 0.0002,
       "step": 1169
     },
     {
-      "epoch": 1.751497005988024,
-      "grad_norm": 0.10074807032478746,
-      "learning_rate": 7.810836376800455e-06,
-      "loss": 0.0007,
+      "epoch": 1.8719999999999999,
+      "grad_norm": 0.01482398063639901,
+      "learning_rate": 6.547342364418482e-06,
+      "loss": 0.0001,
       "step": 1170
     },
     {
-      "epoch": 1.7529940119760479,
-      "grad_norm": 0.06251541594987227,
-      "learning_rate": 7.795062667714211e-06,
-      "loss": 0.0005,
+      "epoch": 1.8736000000000002,
+      "grad_norm": 0.008285397672775559,
+      "learning_rate": 6.5311297042736046e-06,
+      "loss": 0.0001,
       "step": 1171
     },
     {
-      "epoch": 1.7544910179640718,
-      "grad_norm": 0.05781097111200272,
-      "learning_rate": 7.7792947229774e-06,
-      "loss": 0.0002,
+      "epoch": 1.8752,
+      "grad_norm": 0.0020608071734066684,
+      "learning_rate": 6.514927402701965e-06,
+      "loss": 0.0001,
       "step": 1172
     },
     {
-      "epoch": 1.7559880239520957,
-      "grad_norm": 0.2185198392929054,
-      "learning_rate": 7.763532583812042e-06,
-      "loss": 0.0018,
+      "epoch": 1.8768,
+      "grad_norm": 0.004513862742587086,
+      "learning_rate": 6.498735508086094e-06,
+      "loss": 0.0001,
       "step": 1173
     },
     {
-      "epoch": 1.7574850299401197,
-      "grad_norm": 0.064144359442749,
-      "learning_rate": 7.747776291424955e-06,
-      "loss": 0.0004,
+      "epoch": 1.8784,
+      "grad_norm": 0.0009014925985797817,
+      "learning_rate": 6.482554068777451e-06,
+      "loss": 0.0,
       "step": 1174
     },
     {
-      "epoch": 1.7589820359281436,
-      "grad_norm": 0.08498960587781633,
-      "learning_rate": 7.732025887007693e-06,
-      "loss": 0.0008,
+      "epoch": 1.88,
+      "grad_norm": 0.002032352895805396,
+      "learning_rate": 6.466383133096268e-06,
+      "loss": 0.0,
       "step": 1175
     },
     {
-      "epoch": 1.7604790419161676,
-      "grad_norm": 0.1922473098948509,
-      "learning_rate": 7.716281411736411e-06,
-      "loss": 0.0016,
+      "epoch": 1.8816000000000002,
+      "grad_norm": 0.0022593537150606435,
+      "learning_rate": 6.450222749331414e-06,
+      "loss": 0.0001,
       "step": 1176
     },
     {
-      "epoch": 1.7619760479041915,
-      "grad_norm": 0.0661949673802176,
-      "learning_rate": 7.700542906771747e-06,
-      "loss": 0.0004,
+      "epoch": 1.8832,
+      "grad_norm": 0.0033082404611576863,
+      "learning_rate": 6.4340729657402424e-06,
+      "loss": 0.0,
       "step": 1177
     },
     {
-      "epoch": 1.7634730538922154,
-      "grad_norm": 0.10775727570874787,
-      "learning_rate": 7.684810413258756e-06,
-      "loss": 0.0015,
+      "epoch": 1.8848,
+      "grad_norm": 0.0030771403624746036,
+      "learning_rate": 6.4179338305484675e-06,
+      "loss": 0.0001,
       "step": 1178
     },
     {
-      "epoch": 1.7649700598802394,
-      "grad_norm": 0.041135301287911916,
-      "learning_rate": 7.669083972326761e-06,
-      "loss": 0.0004,
+      "epoch": 1.8864,
+      "grad_norm": 0.009938184338915211,
+      "learning_rate": 6.40180539194999e-06,
+      "loss": 0.0001,
       "step": 1179
     },
     {
-      "epoch": 1.7664670658682635,
-      "grad_norm": 0.07967634346867503,
-      "learning_rate": 7.653363625089272e-06,
-      "loss": 0.0009,
+      "epoch": 1.888,
+      "grad_norm": 0.045496591526426336,
+      "learning_rate": 6.385687698106781e-06,
+      "loss": 0.0002,
       "step": 1180
     },
     {
-      "epoch": 1.7679640718562875,
-      "grad_norm": 0.06632082900120188,
-      "learning_rate": 7.637649412643852e-06,
-      "loss": 0.0015,
+      "epoch": 1.8896,
+      "grad_norm": 0.011906474987920825,
+      "learning_rate": 6.3695807971487175e-06,
+      "loss": 0.0001,
       "step": 1181
     },
     {
-      "epoch": 1.7694610778443114,
-      "grad_norm": 0.0597435675787796,
-      "learning_rate": 7.621941376072045e-06,
-      "loss": 0.0005,
+      "epoch": 1.8912,
+      "grad_norm": 0.0009566349700188584,
+      "learning_rate": 6.35348473717345e-06,
+      "loss": 0.0,
       "step": 1182
     },
     {
-      "epoch": 1.7709580838323353,
-      "grad_norm": 0.10680414994888268,
-      "learning_rate": 7.606239556439243e-06,
-      "loss": 0.0008,
+      "epoch": 1.8928,
+      "grad_norm": 0.024164789957125666,
+      "learning_rate": 6.337399566246257e-06,
+      "loss": 0.0001,
       "step": 1183
     },
     {
-      "epoch": 1.7724550898203593,
-      "grad_norm": 0.08131544743100391,
-      "learning_rate": 7.590543994794575e-06,
-      "loss": 0.0005,
+      "epoch": 1.8944,
+      "grad_norm": 0.007994688357821257,
+      "learning_rate": 6.321325332399904e-06,
+      "loss": 0.0001,
       "step": 1184
     },
     {
-      "epoch": 1.7739520958083832,
-      "grad_norm": 0.14796670864528785,
-      "learning_rate": 7.574854732170824e-06,
-      "loss": 0.0011,
+      "epoch": 1.896,
+      "grad_norm": 0.004354701970615443,
+      "learning_rate": 6.305262083634488e-06,
+      "loss": 0.0001,
       "step": 1185
     },
     {
-      "epoch": 1.7754491017964071,
-      "grad_norm": 0.05808741649463366,
-      "learning_rate": 7.559171809584298e-06,
-      "loss": 0.0004,
+      "epoch": 1.8976,
+      "grad_norm": 0.09316375812988109,
+      "learning_rate": 6.289209867917312e-06,
+      "loss": 0.0026,
       "step": 1186
     },
     {
-      "epoch": 1.7769461077844313,
-      "grad_norm": 0.0791406942247839,
-      "learning_rate": 7.543495268034735e-06,
-      "loss": 0.0009,
+      "epoch": 1.8992,
+      "grad_norm": 0.00831804802774526,
+      "learning_rate": 6.2731687331827214e-06,
+      "loss": 0.0001,
       "step": 1187
     },
     {
-      "epoch": 1.7784431137724552,
-      "grad_norm": 0.047886021234621046,
-      "learning_rate": 7.527825148505181e-06,
-      "loss": 0.0003,
+      "epoch": 1.9008,
+      "grad_norm": 0.0043503339187173605,
+      "learning_rate": 6.2571387273319905e-06,
+      "loss": 0.0001,
       "step": 1188
     },
     {
-      "epoch": 1.7799401197604792,
-      "grad_norm": 0.1335212254194146,
-      "learning_rate": 7.512161491961903e-06,
-      "loss": 0.001,
+      "epoch": 1.9024,
+      "grad_norm": 0.15915503150998686,
+      "learning_rate": 6.2411198982331435e-06,
+      "loss": 0.0005,
       "step": 1189
     },
     {
-      "epoch": 1.781437125748503,
-      "grad_norm": 0.03913458823023904,
-      "learning_rate": 7.496504339354274e-06,
-      "loss": 0.0006,
+      "epoch": 1.904,
+      "grad_norm": 0.005591245871832945,
+      "learning_rate": 6.225112293720836e-06,
+      "loss": 0.0001,
       "step": 1190
     },
     {
-      "epoch": 1.782934131736527,
-      "grad_norm": 0.08672922100866931,
-      "learning_rate": 7.48085373161465e-06,
-      "loss": 0.0011,
+      "epoch": 1.9056,
+      "grad_norm": 0.04492041883237681,
+      "learning_rate": 6.209115961596208e-06,
+      "loss": 0.0003,
       "step": 1191
     },
     {
-      "epoch": 1.784431137724551,
-      "grad_norm": 0.11931210383133138,
-      "learning_rate": 7.465209709658294e-06,
-      "loss": 0.0011,
+      "epoch": 1.9072,
+      "grad_norm": 0.06192810990131338,
+      "learning_rate": 6.193130949626731e-06,
+      "loss": 0.0003,
       "step": 1192
     },
     {
-      "epoch": 1.785928143712575,
-      "grad_norm": 0.05929701873982781,
-      "learning_rate": 7.449572314383237e-06,
-      "loss": 0.0006,
+      "epoch": 1.9088,
+      "grad_norm": 0.008427562962625855,
+      "learning_rate": 6.177157305546077e-06,
+      "loss": 0.0001,
       "step": 1193
     },
     {
-      "epoch": 1.7874251497005988,
-      "grad_norm": 0.04505152324525866,
-      "learning_rate": 7.4339415866701945e-06,
-      "loss": 0.0003,
+      "epoch": 1.9104,
+      "grad_norm": 0.007826368970274592,
+      "learning_rate": 6.1611950770539766e-06,
+      "loss": 0.0001,
       "step": 1194
     },
     {
-      "epoch": 1.7889221556886228,
-      "grad_norm": 0.12017810117570427,
-      "learning_rate": 7.418317567382447e-06,
-      "loss": 0.0008,
+      "epoch": 1.912,
+      "grad_norm": 0.06847520283496732,
+      "learning_rate": 6.145244311816063e-06,
+      "loss": 0.0002,
       "step": 1195
     },
     {
-      "epoch": 1.7904191616766467,
-      "grad_norm": 0.09158715938134543,
-      "learning_rate": 7.402700297365741e-06,
-      "loss": 0.0007,
+      "epoch": 1.9136,
+      "grad_norm": 0.0173206162455603,
+      "learning_rate": 6.129305057463741e-06,
+      "loss": 0.0001,
       "step": 1196
     },
     {
-      "epoch": 1.7919161676646707,
-      "grad_norm": 0.1014983459541966,
-      "learning_rate": 7.38708981744818e-06,
-      "loss": 0.0006,
+      "epoch": 1.9152,
+      "grad_norm": 0.001479780943036349,
+      "learning_rate": 6.113377361594048e-06,
+      "loss": 0.0,
       "step": 1197
     },
     {
-      "epoch": 1.7934131736526946,
-      "grad_norm": 0.08269083942405764,
-      "learning_rate": 7.3714861684401025e-06,
-      "loss": 0.0004,
+      "epoch": 1.9167999999999998,
+      "grad_norm": 0.12552314051053987,
+      "learning_rate": 6.0974612717695e-06,
+      "loss": 0.0005,
       "step": 1198
     },
     {
-      "epoch": 1.7949101796407185,
-      "grad_norm": 0.06644722617806384,
-      "learning_rate": 7.355889391134011e-06,
-      "loss": 0.0006,
+      "epoch": 1.9184,
+      "grad_norm": 0.0019474675781217055,
+      "learning_rate": 6.081556835517955e-06,
+      "loss": 0.0,
       "step": 1199
     },
     {
-      "epoch": 1.7964071856287425,
-      "grad_norm": 0.05523988911350223,
-      "learning_rate": 7.340299526304422e-06,
-      "loss": 0.0004,
+      "epoch": 1.92,
+      "grad_norm": 0.03905313180296663,
+      "learning_rate": 6.065664100332478e-06,
+      "loss": 0.0002,
       "step": 1200
     },
     {
-      "epoch": 1.7979041916167664,
-      "grad_norm": 0.08351184232628797,
-      "learning_rate": 7.324716614707794e-06,
-      "loss": 0.0009,
+      "epoch": 1.9216,
+      "grad_norm": 0.0015254580692129185,
+      "learning_rate": 6.049783113671184e-06,
+      "loss": 0.0,
       "step": 1201
     },
     {
-      "epoch": 1.7994011976047903,
-      "grad_norm": 0.006774670042731655,
-      "learning_rate": 7.309140697082402e-06,
+      "epoch": 1.9232,
+      "grad_norm": 0.004929310744791136,
+      "learning_rate": 6.033913922957112e-06,
       "loss": 0.0001,
       "step": 1202
     },
     {
-      "epoch": 1.8008982035928143,
-      "grad_norm": 0.04928550988537502,
-      "learning_rate": 7.293571814148241e-06,
-      "loss": 0.0003,
+      "epoch": 1.9247999999999998,
+      "grad_norm": 0.12635043606836968,
+      "learning_rate": 6.018056575578075e-06,
+      "loss": 0.0005,
       "step": 1203
     },
     {
-      "epoch": 1.8023952095808382,
-      "grad_norm": 0.06584370826534322,
-      "learning_rate": 7.2780100066069174e-06,
-      "loss": 0.0003,
+      "epoch": 1.9264000000000001,
+      "grad_norm": 0.0019222745708322344,
+      "learning_rate": 6.002211118886514e-06,
+      "loss": 0.0,
       "step": 1204
     },
     {
-      "epoch": 1.8038922155688621,
-      "grad_norm": 0.18345445829638807,
-      "learning_rate": 7.262455315141526e-06,
-      "loss": 0.001,
+      "epoch": 1.928,
+      "grad_norm": 0.014047020016825352,
+      "learning_rate": 5.986377600199371e-06,
+      "loss": 0.0001,
       "step": 1205
     },
     {
-      "epoch": 1.805389221556886,
-      "grad_norm": 0.07244929387533793,
-      "learning_rate": 7.246907780416574e-06,
-      "loss": 0.0005,
+      "epoch": 1.9296,
+      "grad_norm": 0.0031841005373336863,
+      "learning_rate": 5.970556066797941e-06,
+      "loss": 0.0,
       "step": 1206
     },
     {
-      "epoch": 1.80688622754491,
-      "grad_norm": 0.09817729337327456,
-      "learning_rate": 7.231367443077851e-06,
-      "loss": 0.0007,
+      "epoch": 1.9312,
+      "grad_norm": 0.0010223404299522473,
+      "learning_rate": 5.9547465659277215e-06,
+      "loss": 0.0,
       "step": 1207
     },
     {
-      "epoch": 1.8083832335329342,
-      "grad_norm": 0.05483140894713459,
-      "learning_rate": 7.215834343752337e-06,
-      "loss": 0.0004,
+      "epoch": 1.9327999999999999,
+      "grad_norm": 0.0035052486724118986,
+      "learning_rate": 5.93894914479828e-06,
+      "loss": 0.0,
       "step": 1208
     },
     {
-      "epoch": 1.8098802395209581,
-      "grad_norm": 0.03551972805679641,
-      "learning_rate": 7.2003085230480785e-06,
-      "loss": 0.0003,
+      "epoch": 1.9344000000000001,
+      "grad_norm": 0.002527851996072802,
+      "learning_rate": 5.923163850583114e-06,
+      "loss": 0.0001,
       "step": 1209
     },
     {
-      "epoch": 1.811377245508982,
-      "grad_norm": 0.06212947300716921,
-      "learning_rate": 7.184790021554102e-06,
-      "loss": 0.0004,
+      "epoch": 1.936,
+      "grad_norm": 0.013050221029306884,
+      "learning_rate": 5.907390730419506e-06,
+      "loss": 0.0001,
       "step": 1210
     },
     {
-      "epoch": 1.812874251497006,
-      "grad_norm": 0.11663130238724655,
-      "learning_rate": 7.169278879840304e-06,
-      "loss": 0.001,
+      "epoch": 1.9376,
+      "grad_norm": 0.08448027001276927,
+      "learning_rate": 5.891629831408392e-06,
+      "loss": 0.0003,
       "step": 1211
     },
     {
-      "epoch": 1.81437125748503,
-      "grad_norm": 0.1655611404625784,
-      "learning_rate": 7.153775138457326e-06,
-      "loss": 0.0014,
+      "epoch": 1.9392,
+      "grad_norm": 0.0021380535708301327,
+      "learning_rate": 5.875881200614208e-06,
+      "loss": 0.0,
       "step": 1212
     },
     {
-      "epoch": 1.8158682634730539,
-      "grad_norm": 0.016811102246245015,
-      "learning_rate": 7.138278837936475e-06,
+      "epoch": 1.9407999999999999,
+      "grad_norm": 0.004995336312321058,
+      "learning_rate": 5.8601448850647515e-06,
       "loss": 0.0001,
       "step": 1213
     },
     {
-      "epoch": 1.8173652694610778,
-      "grad_norm": 0.08174583277697046,
-      "learning_rate": 7.122790018789597e-06,
-      "loss": 0.0008,
+      "epoch": 1.9424000000000001,
+      "grad_norm": 0.008996039230761935,
+      "learning_rate": 5.8444209317510515e-06,
+      "loss": 0.0001,
       "step": 1214
     },
     {
-      "epoch": 1.818862275449102,
-      "grad_norm": 0.09229932047043227,
-      "learning_rate": 7.107308721508992e-06,
-      "loss": 0.0005,
+      "epoch": 1.944,
+      "grad_norm": 0.004159545890870004,
+      "learning_rate": 5.828709387627219e-06,
+      "loss": 0.0001,
       "step": 1215
     },
     {
-      "epoch": 1.8203592814371259,
-      "grad_norm": 0.05703431462629337,
-      "learning_rate": 7.091834986567278e-06,
-      "loss": 0.0003,
+      "epoch": 1.9456,
+      "grad_norm": 0.02845200728681492,
+      "learning_rate": 5.813010299610313e-06,
+      "loss": 0.0001,
       "step": 1216
     },
     {
-      "epoch": 1.8218562874251498,
-      "grad_norm": 0.045543500277733254,
-      "learning_rate": 7.076368854417319e-06,
-      "loss": 0.0004,
+      "epoch": 1.9472,
+      "grad_norm": 0.10556405252712346,
+      "learning_rate": 5.797323714580192e-06,
+      "loss": 0.0005,
       "step": 1217
     },
     {
-      "epoch": 1.8233532934131738,
-      "grad_norm": 0.04621091298397737,
-      "learning_rate": 7.0609103654920955e-06,
-      "loss": 0.0006,
+      "epoch": 1.9487999999999999,
+      "grad_norm": 0.05178852241859,
+      "learning_rate": 5.781649679379379e-06,
+      "loss": 0.0003,
       "step": 1218
     },
     {
-      "epoch": 1.8248502994011977,
-      "grad_norm": 0.12194672605605335,
-      "learning_rate": 7.045459560204604e-06,
-      "loss": 0.0008,
+      "epoch": 1.9504000000000001,
+      "grad_norm": 0.0019259899274537585,
+      "learning_rate": 5.7659882408129204e-06,
+      "loss": 0.0,
       "step": 1219
     },
     {
-      "epoch": 1.8263473053892216,
-      "grad_norm": 0.04569140798030763,
-      "learning_rate": 7.030016478947763e-06,
-      "loss": 0.0003,
+      "epoch": 1.952,
+      "grad_norm": 0.16739476698823094,
+      "learning_rate": 5.750339445648252e-06,
+      "loss": 0.0013,
       "step": 1220
     },
     {
-      "epoch": 1.8278443113772456,
-      "grad_norm": 0.09081360546587813,
-      "learning_rate": 7.014581162094284e-06,
-      "loss": 0.0003,
+      "epoch": 1.9536,
+      "grad_norm": 0.03504969820142291,
+      "learning_rate": 5.7347033406150494e-06,
+      "loss": 0.0001,
       "step": 1221
     },
     {
-      "epoch": 1.8293413173652695,
-      "grad_norm": 0.07508995179575473,
-      "learning_rate": 6.999153649996595e-06,
-      "loss": 0.0007,
+      "epoch": 1.9552,
+      "grad_norm": 0.029114280813010592,
+      "learning_rate": 5.7190799724050924e-06,
+      "loss": 0.0001,
       "step": 1222
     },
     {
-      "epoch": 1.8308383233532934,
-      "grad_norm": 0.0236647586499013,
-      "learning_rate": 6.983733982986709e-06,
-      "loss": 0.0002,
+      "epoch": 1.9567999999999999,
+      "grad_norm": 0.007452639771015537,
+      "learning_rate": 5.703469387672138e-06,
+      "loss": 0.0001,
       "step": 1223
     },
     {
-      "epoch": 1.8323353293413174,
-      "grad_norm": 0.0348458963052325,
-      "learning_rate": 6.968322201376136e-06,
-      "loss": 0.0003,
+      "epoch": 1.9584000000000001,
+      "grad_norm": 0.049837640333634387,
+      "learning_rate": 5.687871633031754e-06,
+      "loss": 0.0002,
       "step": 1224
     },
     {
-      "epoch": 1.8338323353293413,
-      "grad_norm": 0.06740266996280521,
-      "learning_rate": 6.952918345455772e-06,
-      "loss": 0.0005,
+      "epoch": 1.96,
+      "grad_norm": 0.0024424688157115458,
+      "learning_rate": 5.672286755061212e-06,
+      "loss": 0.0,
       "step": 1225
     },
     {
-      "epoch": 1.8353293413173652,
-      "grad_norm": 0.1410919073179155,
-      "learning_rate": 6.937522455495787e-06,
-      "loss": 0.0025,
+      "epoch": 1.9616,
+      "grad_norm": 0.014652874845066959,
+      "learning_rate": 5.656714800299317e-06,
+      "loss": 0.0001,
       "step": 1226
     },
     {
-      "epoch": 1.8368263473053892,
-      "grad_norm": 0.050967621048530726,
-      "learning_rate": 6.92213457174553e-06,
-      "loss": 0.0003,
+      "epoch": 1.9632,
+      "grad_norm": 0.005720054051263878,
+      "learning_rate": 5.64115581524629e-06,
+      "loss": 0.0001,
       "step": 1227
     },
     {
-      "epoch": 1.8383233532934131,
-      "grad_norm": 0.054601006610630415,
-      "learning_rate": 6.906754734433414e-06,
-      "loss": 0.0006,
+      "epoch": 1.9647999999999999,
+      "grad_norm": 0.03706288619215235,
+      "learning_rate": 5.625609846363622e-06,
+      "loss": 0.0002,
       "step": 1228
     },
     {
-      "epoch": 1.839820359281437,
-      "grad_norm": 0.12698206601810638,
-      "learning_rate": 6.891382983766828e-06,
-      "loss": 0.0013,
+      "epoch": 1.9664000000000001,
+      "grad_norm": 0.003003921877286788,
+      "learning_rate": 5.610076940073939e-06,
+      "loss": 0.0001,
       "step": 1229
     },
     {
-      "epoch": 1.841317365269461,
-      "grad_norm": 0.08636527463657546,
-      "learning_rate": 6.876019359932004e-06,
-      "loss": 0.0012,
+      "epoch": 1.968,
+      "grad_norm": 0.0028651414160785287,
+      "learning_rate": 5.594557142760853e-06,
+      "loss": 0.0001,
       "step": 1230
     },
     {
-      "epoch": 1.842814371257485,
-      "grad_norm": 0.05071079998230555,
-      "learning_rate": 6.860663903093943e-06,
-      "loss": 0.0003,
+      "epoch": 1.9696,
+      "grad_norm": 0.0027324425467368193,
+      "learning_rate": 5.579050500768837e-06,
+      "loss": 0.0,
       "step": 1231
     },
     {
-      "epoch": 1.8443113772455089,
-      "grad_norm": 0.018134041251772544,
-      "learning_rate": 6.8453166533962925e-06,
+      "epoch": 1.9712,
+      "grad_norm": 0.0024127431551241845,
+      "learning_rate": 5.563557060403071e-06,
       "loss": 0.0001,
       "step": 1232
     },
     {
-      "epoch": 1.8458083832335328,
-      "grad_norm": 0.10446778142724229,
-      "learning_rate": 6.829977650961229e-06,
-      "loss": 0.0016,
+      "epoch": 1.9727999999999999,
+      "grad_norm": 0.0026731181344807624,
+      "learning_rate": 5.548076867929331e-06,
+      "loss": 0.0001,
       "step": 1233
     },
     {
-      "epoch": 1.8473053892215567,
-      "grad_norm": 0.055625311823021856,
-      "learning_rate": 6.814646935889389e-06,
-      "loss": 0.0004,
+      "epoch": 1.9744000000000002,
+      "grad_norm": 0.018776996113024025,
+      "learning_rate": 5.53260996957381e-06,
+      "loss": 0.0001,
       "step": 1234
     },
     {
-      "epoch": 1.8488023952095807,
-      "grad_norm": 0.07096498187159628,
-      "learning_rate": 6.79932454825973e-06,
-      "loss": 0.0004,
+      "epoch": 1.976,
+      "grad_norm": 0.029136136943676304,
+      "learning_rate": 5.517156411523026e-06,
+      "loss": 0.0001,
       "step": 1235
     },
     {
-      "epoch": 1.8502994011976048,
-      "grad_norm": 0.07863020430436107,
-      "learning_rate": 6.7840105281294485e-06,
-      "loss": 0.0004,
+      "epoch": 1.9776,
+      "grad_norm": 0.009965281255625782,
+      "learning_rate": 5.501716239923642e-06,
+      "loss": 0.0001,
       "step": 1236
     },
     {
-      "epoch": 1.8517964071856288,
-      "grad_norm": 0.044616260463696586,
-      "learning_rate": 6.768704915533852e-06,
-      "loss": 0.0004,
+      "epoch": 1.9792,
+      "grad_norm": 0.0016554955923781512,
+      "learning_rate": 5.486289500882355e-06,
+      "loss": 0.0,
       "step": 1237
     },
     {
-      "epoch": 1.8532934131736527,
-      "grad_norm": 0.12621641545858503,
-      "learning_rate": 6.753407750486284e-06,
-      "loss": 0.0017,
+      "epoch": 1.9808,
+      "grad_norm": 0.11607324819037852,
+      "learning_rate": 5.47087624046575e-06,
+      "loss": 0.0007,
       "step": 1238
     },
     {
-      "epoch": 1.8547904191616766,
-      "grad_norm": 0.060596994339849504,
-      "learning_rate": 6.738119072977996e-06,
-      "loss": 0.0006,
+      "epoch": 1.9824000000000002,
+      "grad_norm": 0.055443812103154765,
+      "learning_rate": 5.455476504700161e-06,
+      "loss": 0.0004,
       "step": 1239
     },
     {
-      "epoch": 1.8562874251497006,
-      "grad_norm": 0.014807449760783308,
-      "learning_rate": 6.722838922978046e-06,
-      "loss": 0.0002,
+      "epoch": 1.984,
+      "grad_norm": 0.20593686476304568,
+      "learning_rate": 5.440090339571537e-06,
+      "loss": 0.0021,
       "step": 1240
     },
     {
-      "epoch": 1.8577844311377245,
-      "grad_norm": 0.0663517266563627,
-      "learning_rate": 6.707567340433212e-06,
-      "loss": 0.0006,
+      "epoch": 1.9856,
+      "grad_norm": 0.0030729637895142103,
+      "learning_rate": 5.424717791025302e-06,
+      "loss": 0.0001,
       "step": 1241
     },
     {
-      "epoch": 1.8592814371257484,
-      "grad_norm": 0.1209263861163919,
-      "learning_rate": 6.692304365267859e-06,
-      "loss": 0.0008,
+      "epoch": 1.9872,
+      "grad_norm": 0.010897666599821741,
+      "learning_rate": 5.4093589049662175e-06,
+      "loss": 0.0001,
       "step": 1242
     },
     {
-      "epoch": 1.8607784431137726,
-      "grad_norm": 0.027554525474874584,
-      "learning_rate": 6.677050037383858e-06,
-      "loss": 0.0003,
+      "epoch": 1.9888,
+      "grad_norm": 0.011043396569230967,
+      "learning_rate": 5.3940137272582534e-06,
+      "loss": 0.0001,
       "step": 1243
     },
     {
-      "epoch": 1.8622754491017965,
-      "grad_norm": 0.04775088764673739,
-      "learning_rate": 6.661804396660479e-06,
-      "loss": 0.0005,
+      "epoch": 1.9904,
+      "grad_norm": 0.029818851963151396,
+      "learning_rate": 5.378682303724435e-06,
+      "loss": 0.0001,
       "step": 1244
     },
     {
-      "epoch": 1.8637724550898205,
-      "grad_norm": 0.056886026682939766,
-      "learning_rate": 6.646567482954263e-06,
-      "loss": 0.0003,
+      "epoch": 1.992,
+      "grad_norm": 0.018070519093172074,
+      "learning_rate": 5.3633646801467255e-06,
+      "loss": 0.0002,
       "step": 1245
     },
     {
-      "epoch": 1.8652694610778444,
-      "grad_norm": 0.13424552882664162,
-      "learning_rate": 6.631339336098956e-06,
-      "loss": 0.0013,
+      "epoch": 1.9936,
+      "grad_norm": 0.1723775065387246,
+      "learning_rate": 5.348060902265871e-06,
+      "loss": 0.0008,
       "step": 1246
     },
     {
-      "epoch": 1.8667664670658684,
-      "grad_norm": 0.12731628473965262,
-      "learning_rate": 6.616119995905375e-06,
-      "loss": 0.0012,
+      "epoch": 1.9952,
+      "grad_norm": 0.011930808129548384,
+      "learning_rate": 5.332771015781275e-06,
+      "loss": 0.0001,
       "step": 1247
     },
     {
-      "epoch": 1.8682634730538923,
-      "grad_norm": 0.05387273939125406,
-      "learning_rate": 6.600909502161315e-06,
+      "epoch": 1.9968,
+      "grad_norm": 0.021970927811545714,
+      "learning_rate": 5.31749506635086e-06,
       "loss": 0.0003,
       "step": 1248
     },
     {
-      "epoch": 1.8697604790419162,
-      "grad_norm": 0.11655593860565898,
-      "learning_rate": 6.585707894631441e-06,
-      "loss": 0.0008,
+      "epoch": 1.9984,
+      "grad_norm": 0.159920156495135,
+      "learning_rate": 5.302233099590928e-06,
+      "loss": 0.0006,
       "step": 1249
     },
     {
-      "epoch": 1.8712574850299402,
-      "grad_norm": 0.03130014477728136,
-      "learning_rate": 6.570515213057192e-06,
-      "loss": 0.0002,
+      "epoch": 2.0,
+      "grad_norm": 0.007002186599926069,
+      "learning_rate": 5.286985161076029e-06,
+      "loss": 0.0001,
       "step": 1250
     },
     {
-      "epoch": 1.872754491017964,
-      "grad_norm": 0.06789407097061075,
-      "learning_rate": 6.555331497156671e-06,
-      "loss": 0.0006,
+      "epoch": 2.0016,
+      "grad_norm": 0.03948487381609094,
+      "learning_rate": 5.271751296338823e-06,
+      "loss": 0.0001,
       "step": 1251
     },
     {
-      "epoch": 1.874251497005988,
-      "grad_norm": 0.1053736315966829,
-      "learning_rate": 6.5401567866245405e-06,
-      "loss": 0.0006,
+      "epoch": 2.0032,
+      "grad_norm": 0.012853692816418372,
+      "learning_rate": 5.2565315508699374e-06,
+      "loss": 0.0001,
       "step": 1252
     },
     {
-      "epoch": 1.875748502994012,
-      "grad_norm": 0.042293639897476594,
-      "learning_rate": 6.5249911211319225e-06,
-      "loss": 0.0003,
+      "epoch": 2.0048,
+      "grad_norm": 0.04693735985976921,
+      "learning_rate": 5.241325970117851e-06,
+      "loss": 0.0002,
       "step": 1253
     },
     {
-      "epoch": 1.877245508982036,
-      "grad_norm": 0.08285464714326594,
-      "learning_rate": 6.509834540326285e-06,
-      "loss": 0.0004,
+      "epoch": 2.0064,
+      "grad_norm": 0.0010564428981468033,
+      "learning_rate": 5.226134599488728e-06,
+      "loss": 0.0,
       "step": 1254
     },
     {
-      "epoch": 1.8787425149700598,
-      "grad_norm": 0.11049018140607483,
-      "learning_rate": 6.494687083831362e-06,
-      "loss": 0.0009,
+      "epoch": 2.008,
+      "grad_norm": 0.16120154231483275,
+      "learning_rate": 5.210957484346314e-06,
+      "loss": 0.0007,
       "step": 1255
     },
     {
-      "epoch": 1.8802395209580838,
-      "grad_norm": 0.01829057968863813,
-      "learning_rate": 6.47954879124701e-06,
+      "epoch": 2.0096,
+      "grad_norm": 0.0060265765842624055,
+      "learning_rate": 5.195794670011775e-06,
       "loss": 0.0001,
       "step": 1256
     },
     {
-      "epoch": 1.8817365269461077,
-      "grad_norm": 0.033052338927285374,
-      "learning_rate": 6.4644197021491515e-06,
-      "loss": 0.0003,
+      "epoch": 2.0112,
+      "grad_norm": 0.0032434068815901823,
+      "learning_rate": 5.1806462017635775e-06,
+      "loss": 0.0001,
       "step": 1257
     },
     {
-      "epoch": 1.8832335329341316,
-      "grad_norm": 0.13174251290530642,
-      "learning_rate": 6.4492998560896395e-06,
-      "loss": 0.0013,
+      "epoch": 2.0128,
+      "grad_norm": 0.03108346973796724,
+      "learning_rate": 5.165512124837344e-06,
+      "loss": 0.0002,
       "step": 1258
     },
     {
-      "epoch": 1.8847305389221556,
-      "grad_norm": 0.11307650321461288,
-      "learning_rate": 6.434189292596158e-06,
-      "loss": 0.0009,
+      "epoch": 2.0144,
+      "grad_norm": 0.07334038179465477,
+      "learning_rate": 5.150392484425728e-06,
+      "loss": 0.0003,
       "step": 1259
     },
     {
-      "epoch": 1.8862275449101795,
-      "grad_norm": 0.039672940708096134,
-      "learning_rate": 6.419088051172133e-06,
-      "loss": 0.0003,
+      "epoch": 2.016,
+      "grad_norm": 0.027860109850781214,
+      "learning_rate": 5.135287325678271e-06,
+      "loss": 0.0001,
       "step": 1260
     },
     {
-      "epoch": 1.8877245508982035,
-      "grad_norm": 0.0947573552544148,
-      "learning_rate": 6.40399617129661e-06,
-      "loss": 0.001,
+      "epoch": 2.0176,
+      "grad_norm": 0.003642680967855806,
+      "learning_rate": 5.120196693701267e-06,
+      "loss": 0.0001,
       "step": 1261
     },
     {
-      "epoch": 1.8892215568862274,
-      "grad_norm": 0.018111227856452812,
-      "learning_rate": 6.3889136924241704e-06,
-      "loss": 0.0002,
+      "epoch": 2.0192,
+      "grad_norm": 0.0015039372448527133,
+      "learning_rate": 5.105120633557634e-06,
+      "loss": 0.0,
       "step": 1262
     },
     {
-      "epoch": 1.8907185628742516,
-      "grad_norm": 0.06351616622351396,
-      "learning_rate": 6.373840653984811e-06,
-      "loss": 0.0005,
+      "epoch": 2.0208,
+      "grad_norm": 0.001065057755579274,
+      "learning_rate": 5.090059190266779e-06,
+      "loss": 0.0,
       "step": 1263
     },
     {
-      "epoch": 1.8922155688622755,
-      "grad_norm": 0.08247922342515682,
-      "learning_rate": 6.35877709538385e-06,
-      "loss": 0.0006,
+      "epoch": 2.0224,
+      "grad_norm": 0.0018136533690353725,
+      "learning_rate": 5.075012408804458e-06,
+      "loss": 0.0,
       "step": 1264
     },
     {
-      "epoch": 1.8937125748502994,
-      "grad_norm": 0.1613355922056432,
-      "learning_rate": 6.3437230560018335e-06,
-      "loss": 0.0011,
+      "epoch": 2.024,
+      "grad_norm": 0.05887000158742606,
+      "learning_rate": 5.059980334102637e-06,
+      "loss": 0.0002,
       "step": 1265
     },
     {
-      "epoch": 1.8952095808383234,
-      "grad_norm": 0.17460284441167057,
-      "learning_rate": 6.3286785751944e-06,
-      "loss": 0.0033,
+      "epoch": 2.0256,
+      "grad_norm": 0.0038137736873319044,
+      "learning_rate": 5.044963011049384e-06,
+      "loss": 0.0001,
       "step": 1266
     },
     {
-      "epoch": 1.8967065868263473,
-      "grad_norm": 0.06534489203227545,
-      "learning_rate": 6.313643692292219e-06,
-      "loss": 0.0005,
+      "epoch": 2.0272,
+      "grad_norm": 0.0019694602635275628,
+      "learning_rate": 5.0299604844886985e-06,
+      "loss": 0.0001,
       "step": 1267
     },
     {
-      "epoch": 1.8982035928143712,
-      "grad_norm": 0.08598292764832548,
-      "learning_rate": 6.298618446600856e-06,
-      "loss": 0.0009,
+      "epoch": 2.0288,
+      "grad_norm": 0.004869063315621386,
+      "learning_rate": 5.0149727992204034e-06,
+      "loss": 0.0001,
       "step": 1268
     },
     {
-      "epoch": 1.8997005988023952,
-      "grad_norm": 0.07498272399084252,
-      "learning_rate": 6.2836028774006945e-06,
+      "epoch": 2.0304,
+      "grad_norm": 0.14391547248553516,
+      "learning_rate": 5.000000000000003e-06,
       "loss": 0.0005,
       "step": 1269
     },
     {
-      "epoch": 1.9011976047904193,
-      "grad_norm": 0.07854073646784032,
-      "learning_rate": 6.268597023946801e-06,
-      "loss": 0.001,
+      "epoch": 2.032,
+      "grad_norm": 0.18742481535002514,
+      "learning_rate": 4.985042131538545e-06,
+      "loss": 0.0009,
       "step": 1270
     },
     {
-      "epoch": 1.9026946107784433,
-      "grad_norm": 0.08325592956362902,
-      "learning_rate": 6.253600925468861e-06,
-      "loss": 0.0009,
+      "epoch": 2.0336,
+      "grad_norm": 0.005019826620883686,
+      "learning_rate": 4.970099238502494e-06,
+      "loss": 0.0001,
       "step": 1271
     },
     {
-      "epoch": 1.9041916167664672,
-      "grad_norm": 0.06261744481965902,
-      "learning_rate": 6.238614621171055e-06,
-      "loss": 0.0004,
+      "epoch": 2.0352,
+      "grad_norm": 0.08200275784421962,
+      "learning_rate": 4.955171365513603e-06,
+      "loss": 0.0002,
       "step": 1272
     },
     {
-      "epoch": 1.9056886227544911,
-      "grad_norm": 0.15054085004158702,
-      "learning_rate": 6.223638150231945e-06,
-      "loss": 0.0008,
+      "epoch": 2.0368,
+      "grad_norm": 0.005278096656013719,
+      "learning_rate": 4.940258557148765e-06,
+      "loss": 0.0001,
       "step": 1273
     },
     {
-      "epoch": 1.907185628742515,
-      "grad_norm": 0.15292361187679987,
-      "learning_rate": 6.208671551804404e-06,
-      "loss": 0.0009,
+      "epoch": 2.0384,
+      "grad_norm": 0.0065573406554249114,
+      "learning_rate": 4.925360857939886e-06,
+      "loss": 0.0001,
       "step": 1274
     },
     {
-      "epoch": 1.908682634730539,
-      "grad_norm": 0.06302399358176151,
-      "learning_rate": 6.193714865015479e-06,
-      "loss": 0.0007,
+      "epoch": 2.04,
+      "grad_norm": 0.021604013246013607,
+      "learning_rate": 4.910478312373757e-06,
+      "loss": 0.0001,
       "step": 1275
     },
     {
-      "epoch": 1.910179640718563,
-      "grad_norm": 0.08082819893731763,
-      "learning_rate": 6.178768128966319e-06,
-      "loss": 0.0011,
+      "epoch": 2.0416,
+      "grad_norm": 0.002484955128533375,
+      "learning_rate": 4.895610964891923e-06,
+      "loss": 0.0,
       "step": 1276
     },
     {
-      "epoch": 1.9116766467065869,
-      "grad_norm": 0.10017565699972938,
-      "learning_rate": 6.163831382732046e-06,
-      "loss": 0.0005,
+      "epoch": 2.0432,
+      "grad_norm": 0.0027986394634020926,
+      "learning_rate": 4.8807588598905364e-06,
+      "loss": 0.0001,
       "step": 1277
     },
     {
-      "epoch": 1.9131736526946108,
-      "grad_norm": 0.1603341983015068,
-      "learning_rate": 6.148904665361677e-06,
-      "loss": 0.0008,
+      "epoch": 2.0448,
+      "grad_norm": 0.0027430637359838494,
+      "learning_rate": 4.865922041720239e-06,
+      "loss": 0.0,
       "step": 1278
     },
     {
-      "epoch": 1.9146706586826348,
-      "grad_norm": 0.07977074549410447,
-      "learning_rate": 6.133988015878005e-06,
-      "loss": 0.0003,
+      "epoch": 2.0464,
+      "grad_norm": 0.002737074032174217,
+      "learning_rate": 4.8511005546860214e-06,
+      "loss": 0.0,
       "step": 1279
     },
     {
-      "epoch": 1.9161676646706587,
-      "grad_norm": 0.08604266347649189,
-      "learning_rate": 6.119081473277502e-06,
-      "loss": 0.0015,
+      "epoch": 2.048,
+      "grad_norm": 0.008016519753927618,
+      "learning_rate": 4.836294443047088e-06,
+      "loss": 0.0001,
       "step": 1280
     },
     {
-      "epoch": 1.9176646706586826,
-      "grad_norm": 0.07316072392265781,
-      "learning_rate": 6.104185076530224e-06,
-      "loss": 0.0019,
+      "epoch": 2.0496,
+      "grad_norm": 0.12372213692952685,
+      "learning_rate": 4.821503751016746e-06,
+      "loss": 0.0007,
       "step": 1281
     },
     {
-      "epoch": 1.9191616766467066,
-      "grad_norm": 0.07219162163636049,
-      "learning_rate": 6.0892988645796894e-06,
-      "loss": 0.0011,
+      "epoch": 2.0512,
+      "grad_norm": 0.0013784424610091336,
+      "learning_rate": 4.806728522762241e-06,
+      "loss": 0.0,
       "step": 1282
     },
     {
-      "epoch": 1.9206586826347305,
-      "grad_norm": 0.03498121695768265,
-      "learning_rate": 6.074422876342808e-06,
-      "loss": 0.0002,
+      "epoch": 2.0528,
+      "grad_norm": 0.11916867318013918,
+      "learning_rate": 4.791968802404648e-06,
+      "loss": 0.0004,
       "step": 1283
     },
     {
-      "epoch": 1.9221556886227544,
-      "grad_norm": 0.18083663339991857,
-      "learning_rate": 6.059557150709745e-06,
-      "loss": 0.0009,
+      "epoch": 2.0544,
+      "grad_norm": 0.011827412648333358,
+      "learning_rate": 4.777224634018732e-06,
+      "loss": 0.0002,
       "step": 1284
     },
     {
-      "epoch": 1.9236526946107784,
-      "grad_norm": 0.059187615256778826,
-      "learning_rate": 6.0447017265438464e-06,
-      "loss": 0.0004,
+      "epoch": 2.056,
+      "grad_norm": 0.010541327479658821,
+      "learning_rate": 4.762496061632814e-06,
+      "loss": 0.0001,
       "step": 1285
     },
     {
-      "epoch": 1.9251497005988023,
-      "grad_norm": 0.12966709587781838,
-      "learning_rate": 6.029856642681528e-06,
-      "loss": 0.001,
+      "epoch": 2.0576,
+      "grad_norm": 0.0059639958088174546,
+      "learning_rate": 4.7477831292286555e-06,
+      "loss": 0.0001,
       "step": 1286
     },
     {
-      "epoch": 1.9266467065868262,
-      "grad_norm": 0.03319551814066764,
-      "learning_rate": 6.015021937932166e-06,
-      "loss": 0.0003,
+      "epoch": 2.0592,
+      "grad_norm": 0.07898598055159177,
+      "learning_rate": 4.733085880741301e-06,
+      "loss": 0.0006,
       "step": 1287
     },
     {
-      "epoch": 1.9281437125748502,
-      "grad_norm": 0.08693549001284917,
-      "learning_rate": 6.000197651078012e-06,
-      "loss": 0.0002,
+      "epoch": 2.0608,
+      "grad_norm": 0.010649954432394063,
+      "learning_rate": 4.7184043600589655e-06,
+      "loss": 0.0001,
       "step": 1288
     },
     {
-      "epoch": 1.9296407185628741,
-      "grad_norm": 0.11435904368412542,
-      "learning_rate": 5.985383820874069e-06,
-      "loss": 0.0008,
+      "epoch": 2.0624,
+      "grad_norm": 0.01850359909026641,
+      "learning_rate": 4.703738611022899e-06,
+      "loss": 0.0001,
       "step": 1289
     },
     {
-      "epoch": 1.931137724550898,
-      "grad_norm": 0.03605212305635656,
-      "learning_rate": 5.970580486048016e-06,
-      "loss": 0.0002,
+      "epoch": 2.064,
+      "grad_norm": 0.013855878203025425,
+      "learning_rate": 4.689088677427249e-06,
+      "loss": 0.0001,
       "step": 1290
     },
     {
-      "epoch": 1.9326347305389222,
-      "grad_norm": 0.07980050042842693,
-      "learning_rate": 5.955787685300085e-06,
-      "loss": 0.0005,
+      "epoch": 2.0656,
+      "grad_norm": 0.0029063101645990727,
+      "learning_rate": 4.674454603018949e-06,
+      "loss": 0.0001,
       "step": 1291
     },
     {
-      "epoch": 1.9341317365269461,
-      "grad_norm": 0.11431738205279493,
-      "learning_rate": 5.941005457302975e-06,
-      "loss": 0.0007,
+      "epoch": 2.0672,
+      "grad_norm": 0.002771446021717218,
+      "learning_rate": 4.659836431497563e-06,
+      "loss": 0.0,
       "step": 1292
     },
     {
-      "epoch": 1.93562874251497,
-      "grad_norm": 0.2420720686661468,
-      "learning_rate": 5.926233840701747e-06,
-      "loss": 0.0018,
+      "epoch": 2.0688,
+      "grad_norm": 0.007558521815404343,
+      "learning_rate": 4.645234206515171e-06,
+      "loss": 0.0001,
       "step": 1293
     },
     {
-      "epoch": 1.937125748502994,
-      "grad_norm": 0.17499882775445486,
-      "learning_rate": 5.911472874113709e-06,
-      "loss": 0.0015,
+      "epoch": 2.0704,
+      "grad_norm": 0.0020190105256480957,
+      "learning_rate": 4.630647971676232e-06,
+      "loss": 0.0,
       "step": 1294
     },
     {
-      "epoch": 1.938622754491018,
-      "grad_norm": 0.08874284785143925,
-      "learning_rate": 5.896722596128337e-06,
-      "loss": 0.0004,
+      "epoch": 2.072,
+      "grad_norm": 0.011165801591369819,
+      "learning_rate": 4.616077770537453e-06,
+      "loss": 0.0001,
       "step": 1295
     },
     {
-      "epoch": 1.9401197604790419,
-      "grad_norm": 0.0714199007843593,
-      "learning_rate": 5.881983045307158e-06,
-      "loss": 0.0003,
+      "epoch": 2.0736,
+      "grad_norm": 0.0022416122666900945,
+      "learning_rate": 4.601523646607675e-06,
+      "loss": 0.0001,
       "step": 1296
     },
     {
-      "epoch": 1.9416167664670658,
-      "grad_norm": 0.011510865246892025,
-      "learning_rate": 5.867254260183665e-06,
+      "epoch": 2.0752,
+      "grad_norm": 0.007794185704477617,
+      "learning_rate": 4.586985643347716e-06,
       "loss": 0.0001,
       "step": 1297
     },
     {
-      "epoch": 1.94311377245509,
-      "grad_norm": 0.10193764772516307,
-      "learning_rate": 5.85253627926319e-06,
-      "loss": 0.0006,
+      "epoch": 2.0768,
+      "grad_norm": 0.0015124048307453561,
+      "learning_rate": 4.572463804170263e-06,
+      "loss": 0.0,
       "step": 1298
     },
     {
-      "epoch": 1.944610778443114,
-      "grad_norm": 0.15050408491671105,
-      "learning_rate": 5.837829141022831e-06,
-      "loss": 0.0007,
+      "epoch": 2.0784,
+      "grad_norm": 0.0030354832046186786,
+      "learning_rate": 4.557958172439726e-06,
+      "loss": 0.0001,
       "step": 1299
     },
     {
-      "epoch": 1.9461077844311379,
-      "grad_norm": 0.10898094115408612,
-      "learning_rate": 5.823132883911349e-06,
-      "loss": 0.0009,
+      "epoch": 2.08,
+      "grad_norm": 0.0020373494626846336,
+      "learning_rate": 4.543468791472131e-06,
+      "loss": 0.0,
       "step": 1300
     },
     {
-      "epoch": 1.9476047904191618,
-      "grad_norm": 0.13535481273196698,
-      "learning_rate": 5.80844754634903e-06,
-      "loss": 0.001,
+      "epoch": 2.0816,
+      "grad_norm": 0.001046926322424395,
+      "learning_rate": 4.5289957045349655e-06,
+      "loss": 0.0,
       "step": 1301
     },
     {
-      "epoch": 1.9491017964071857,
-      "grad_norm": 0.10884632784883908,
-      "learning_rate": 5.793773166727648e-06,
-      "loss": 0.0012,
+      "epoch": 2.0832,
+      "grad_norm": 0.0007689101693175983,
+      "learning_rate": 4.5145389548470645e-06,
+      "loss": 0.0,
       "step": 1302
     },
     {
-      "epoch": 1.9505988023952097,
-      "grad_norm": 0.06725195023294889,
-      "learning_rate": 5.779109783410295e-06,
-      "loss": 0.0004,
+      "epoch": 2.0848,
+      "grad_norm": 0.004590852983945126,
+      "learning_rate": 4.500098585578475e-06,
+      "loss": 0.0001,
       "step": 1303
     },
     {
-      "epoch": 1.9520958083832336,
-      "grad_norm": 0.03344699498459077,
-      "learning_rate": 5.764457434731345e-06,
-      "loss": 0.0003,
+      "epoch": 2.0864,
+      "grad_norm": 0.016623279739457517,
+      "learning_rate": 4.485674639850334e-06,
+      "loss": 0.0001,
       "step": 1304
     },
     {
-      "epoch": 1.9535928143712575,
-      "grad_norm": 0.07364922411512191,
-      "learning_rate": 5.749816158996309e-06,
-      "loss": 0.0012,
+      "epoch": 2.088,
+      "grad_norm": 0.12871540044523444,
+      "learning_rate": 4.471267160734731e-06,
+      "loss": 0.0006,
       "step": 1305
     },
     {
-      "epoch": 1.9550898203592815,
-      "grad_norm": 0.1706215633793,
-      "learning_rate": 5.735185994481748e-06,
-      "loss": 0.0015,
+      "epoch": 2.0896,
+      "grad_norm": 0.004518980147346147,
+      "learning_rate": 4.456876191254582e-06,
+      "loss": 0.0001,
       "step": 1306
     },
     {
-      "epoch": 1.9565868263473054,
-      "grad_norm": 0.16089826885989925,
-      "learning_rate": 5.720566979435193e-06,
-      "loss": 0.0014,
+      "epoch": 2.0912,
+      "grad_norm": 0.0023069709864382807,
+      "learning_rate": 4.4425017743835155e-06,
+      "loss": 0.0,
       "step": 1307
     },
     {
-      "epoch": 1.9580838323353293,
-      "grad_norm": 0.13197070265133728,
-      "learning_rate": 5.705959152074998e-06,
-      "loss": 0.0011,
+      "epoch": 2.0928,
+      "grad_norm": 0.011612607081832236,
+      "learning_rate": 4.4281439530457174e-06,
+      "loss": 0.0001,
       "step": 1308
     },
     {
-      "epoch": 1.9595808383233533,
-      "grad_norm": 0.10737878089775195,
-      "learning_rate": 5.6913625505902966e-06,
-      "loss": 0.0018,
+      "epoch": 2.0944,
+      "grad_norm": 0.004032416921738347,
+      "learning_rate": 4.413802770115816e-06,
+      "loss": 0.0001,
       "step": 1309
     },
     {
-      "epoch": 1.9610778443113772,
-      "grad_norm": 0.09551442058751701,
-      "learning_rate": 5.6767772131408606e-06,
-      "loss": 0.0006,
+      "epoch": 2.096,
+      "grad_norm": 0.020078465032642864,
+      "learning_rate": 4.399478268418771e-06,
+      "loss": 0.0001,
       "step": 1310
     },
     {
-      "epoch": 1.9625748502994012,
-      "grad_norm": 0.1495760754554237,
-      "learning_rate": 5.662203177857015e-06,
-      "loss": 0.0018,
+      "epoch": 2.0976,
+      "grad_norm": 0.009372946495378932,
+      "learning_rate": 4.385170490729712e-06,
+      "loss": 0.0001,
       "step": 1311
     },
     {
-      "epoch": 1.964071856287425,
-      "grad_norm": 0.030762250838208226,
-      "learning_rate": 5.647640482839543e-06,
-      "loss": 0.0003,
+      "epoch": 2.0992,
+      "grad_norm": 0.004075995720607409,
+      "learning_rate": 4.370879479773837e-06,
+      "loss": 0.0001,
       "step": 1312
     },
     {
-      "epoch": 1.965568862275449,
-      "grad_norm": 0.0627412308766797,
-      "learning_rate": 5.633089166159571e-06,
-      "loss": 0.0014,
+      "epoch": 2.1008,
+      "grad_norm": 0.005641261867447899,
+      "learning_rate": 4.356605278226274e-06,
+      "loss": 0.0001,
       "step": 1313
     },
     {
-      "epoch": 1.967065868263473,
-      "grad_norm": 0.08508365773836313,
-      "learning_rate": 5.618549265858499e-06,
-      "loss": 0.0008,
+      "epoch": 2.1024,
+      "grad_norm": 0.019357229759758513,
+      "learning_rate": 4.342347928711953e-06,
+      "loss": 0.0001,
       "step": 1314
     },
     {
-      "epoch": 1.968562874251497,
-      "grad_norm": 0.11170302861015123,
-      "learning_rate": 5.604020819947853e-06,
-      "loss": 0.0005,
+      "epoch": 2.104,
+      "grad_norm": 0.01434237461770615,
+      "learning_rate": 4.328107473805487e-06,
+      "loss": 0.0001,
       "step": 1315
     },
     {
-      "epoch": 1.9700598802395208,
-      "grad_norm": 0.13346911790654048,
-      "learning_rate": 5.589503866409238e-06,
-      "loss": 0.0017,
+      "epoch": 2.1056,
+      "grad_norm": 0.013157265563646457,
+      "learning_rate": 4.313883956031031e-06,
+      "loss": 0.0001,
       "step": 1316
     },
     {
-      "epoch": 1.9715568862275448,
-      "grad_norm": 0.048207609953502435,
-      "learning_rate": 5.574998443194206e-06,
-      "loss": 0.0003,
+      "epoch": 2.1072,
+      "grad_norm": 0.0026463734058776384,
+      "learning_rate": 4.299677417862174e-06,
+      "loss": 0.0,
       "step": 1317
     },
     {
-      "epoch": 1.9730538922155687,
-      "grad_norm": 0.1790415818027207,
-      "learning_rate": 5.560504588224162e-06,
-      "loss": 0.0022,
+      "epoch": 2.1088,
+      "grad_norm": 0.009144315666465027,
+      "learning_rate": 4.28548790172179e-06,
+      "loss": 0.0001,
       "step": 1318
     },
     {
-      "epoch": 1.9745508982035929,
-      "grad_norm": 0.053657875706085344,
-      "learning_rate": 5.546022339390271e-06,
-      "loss": 0.0007,
+      "epoch": 2.1104,
+      "grad_norm": 0.050698061673316835,
+      "learning_rate": 4.2713154499819345e-06,
+      "loss": 0.0003,
       "step": 1319
     },
     {
-      "epoch": 1.9760479041916168,
-      "grad_norm": 0.08409549221939727,
-      "learning_rate": 5.531551734553354e-06,
-      "loss": 0.0012,
+      "epoch": 2.112,
+      "grad_norm": 0.005293028607209848,
+      "learning_rate": 4.257160104963695e-06,
+      "loss": 0.0001,
       "step": 1320
     },
     {
-      "epoch": 1.9775449101796407,
-      "grad_norm": 0.10964353914589475,
-      "learning_rate": 5.517092811543801e-06,
-      "loss": 0.0007,
+      "epoch": 2.1136,
+      "grad_norm": 0.006772389868685981,
+      "learning_rate": 4.243021908937083e-06,
+      "loss": 0.0001,
       "step": 1321
     },
     {
-      "epoch": 1.9790419161676647,
-      "grad_norm": 0.15383349246769662,
-      "learning_rate": 5.502645608161451e-06,
-      "loss": 0.0008,
+      "epoch": 2.1152,
+      "grad_norm": 0.0022400613349784475,
+      "learning_rate": 4.228900904120895e-06,
+      "loss": 0.0001,
       "step": 1322
     },
     {
-      "epoch": 1.9805389221556886,
-      "grad_norm": 0.09629278498120927,
-      "learning_rate": 5.488210162175508e-06,
-      "loss": 0.0005,
+      "epoch": 2.1168,
+      "grad_norm": 0.004894576564314422,
+      "learning_rate": 4.214797132682597e-06,
+      "loss": 0.0001,
       "step": 1323
     },
     {
-      "epoch": 1.9820359281437125,
-      "grad_norm": 0.08906179596805544,
-      "learning_rate": 5.473786511324439e-06,
-      "loss": 0.0006,
+      "epoch": 2.1184,
+      "grad_norm": 0.0011541246411294337,
+      "learning_rate": 4.200710636738189e-06,
+      "loss": 0.0,
       "step": 1324
     },
     {
-      "epoch": 1.9835329341317365,
-      "grad_norm": 0.06933054825478269,
-      "learning_rate": 5.459374693315876e-06,
-      "loss": 0.0003,
+      "epoch": 2.12,
+      "grad_norm": 0.024462842828927446,
+      "learning_rate": 4.186641458352088e-06,
+      "loss": 0.0002,
       "step": 1325
     },
     {
-      "epoch": 1.9850299401197606,
-      "grad_norm": 0.035068650708494103,
-      "learning_rate": 5.44497474582651e-06,
-      "loss": 0.0002,
+      "epoch": 2.1216,
+      "grad_norm": 0.008815829602056808,
+      "learning_rate": 4.172589639536992e-06,
+      "loss": 0.0001,
       "step": 1326
     },
     {
-      "epoch": 1.9865269461077846,
-      "grad_norm": 0.09362975516235468,
-      "learning_rate": 5.430586706502014e-06,
-      "loss": 0.001,
+      "epoch": 2.1232,
+      "grad_norm": 0.0043029685313057,
+      "learning_rate": 4.158555222253772e-06,
+      "loss": 0.0001,
       "step": 1327
     },
     {
-      "epoch": 1.9880239520958085,
-      "grad_norm": 0.06835759304910806,
-      "learning_rate": 5.416210612956918e-06,
-      "loss": 0.0009,
+      "epoch": 2.1248,
+      "grad_norm": 0.00504839010368318,
+      "learning_rate": 4.144538248411321e-06,
+      "loss": 0.0001,
       "step": 1328
     },
     {
-      "epoch": 1.9895209580838324,
-      "grad_norm": 0.019786881306191252,
-      "learning_rate": 5.401846502774522e-06,
+      "epoch": 2.1264,
+      "grad_norm": 0.012591004904414362,
+      "learning_rate": 4.130538759866457e-06,
       "loss": 0.0001,
       "step": 1329
     },
     {
-      "epoch": 1.9910179640718564,
-      "grad_norm": 0.1299238026590814,
-      "learning_rate": 5.3874944135068e-06,
-      "loss": 0.0015,
+      "epoch": 2.128,
+      "grad_norm": 0.01673392343030602,
+      "learning_rate": 4.116556798423776e-06,
+      "loss": 0.0001,
       "step": 1330
     },
     {
-      "epoch": 1.9925149700598803,
-      "grad_norm": 0.14300823664119922,
-      "learning_rate": 5.373154382674301e-06,
-      "loss": 0.0009,
+      "epoch": 2.1296,
+      "grad_norm": 0.05585050121872357,
+      "learning_rate": 4.102592405835536e-06,
+      "loss": 0.0002,
       "step": 1331
     },
     {
-      "epoch": 1.9940119760479043,
-      "grad_norm": 0.09070831138488968,
-      "learning_rate": 5.3588264477660525e-06,
-      "loss": 0.0007,
+      "epoch": 2.1312,
+      "grad_norm": 0.02670201452769496,
+      "learning_rate": 4.088645623801534e-06,
+      "loss": 0.0002,
       "step": 1332
     },
     {
-      "epoch": 1.9955089820359282,
-      "grad_norm": 0.3279851722644398,
-      "learning_rate": 5.344510646239447e-06,
-      "loss": 0.0007,
+      "epoch": 2.1328,
+      "grad_norm": 0.0018710927812524058,
+      "learning_rate": 4.074716493968976e-06,
+      "loss": 0.0,
       "step": 1333
     },
     {
-      "epoch": 1.9970059880239521,
-      "grad_norm": 0.13371612237597238,
-      "learning_rate": 5.330207015520176e-06,
-      "loss": 0.0005,
+      "epoch": 2.1344,
+      "grad_norm": 0.0017912785338384182,
+      "learning_rate": 4.060805057932359e-06,
+      "loss": 0.0,
       "step": 1334
     },
     {
-      "epoch": 1.998502994011976,
-      "grad_norm": 0.05563881939066664,
-      "learning_rate": 5.3159155930021e-06,
-      "loss": 0.0004,
+      "epoch": 2.136,
+      "grad_norm": 0.00206502439748784,
+      "learning_rate": 4.046911357233343e-06,
+      "loss": 0.0,
       "step": 1335
     },
     {
-      "epoch": 2.0,
-      "grad_norm": 0.05524915070416573,
-      "learning_rate": 5.301636416047166e-06,
-      "loss": 0.0003,
+      "epoch": 2.1376,
+      "grad_norm": 0.004367991071358672,
+      "learning_rate": 4.033035433360624e-06,
+      "loss": 0.0,
       "step": 1336
     },
     {
-      "epoch": 2.001497005988024,
-      "grad_norm": 0.078998616135492,
-      "learning_rate": 5.287369521985308e-06,
-      "loss": 0.0005,
+      "epoch": 2.1391999999999998,
+      "grad_norm": 0.10915011758089804,
+      "learning_rate": 4.019177327749822e-06,
+      "loss": 0.0003,
       "step": 1337
     },
     {
-      "epoch": 2.002994011976048,
-      "grad_norm": 0.11087032843271219,
-      "learning_rate": 5.273114948114346e-06,
-      "loss": 0.0006,
+      "epoch": 2.1408,
+      "grad_norm": 0.011046214540442804,
+      "learning_rate": 4.00533708178334e-06,
+      "loss": 0.0001,
       "step": 1338
     },
     {
-      "epoch": 2.004491017964072,
-      "grad_norm": 0.01134812052606357,
-      "learning_rate": 5.258872731699907e-06,
-      "loss": 0.0002,
+      "epoch": 2.1424,
+      "grad_norm": 0.0008909985680504715,
+      "learning_rate": 3.991514736790259e-06,
+      "loss": 0.0,
       "step": 1339
     },
     {
-      "epoch": 2.0059880239520957,
-      "grad_norm": 0.053992660508964285,
-      "learning_rate": 5.244642909975282e-06,
-      "loss": 0.0005,
+      "epoch": 2.144,
+      "grad_norm": 0.0064867553974109725,
+      "learning_rate": 3.977710334046193e-06,
+      "loss": 0.0001,
       "step": 1340
     },
     {
-      "epoch": 2.0074850299401197,
-      "grad_norm": 0.06896338730150291,
-      "learning_rate": 5.23042552014139e-06,
-      "loss": 0.0003,
+      "epoch": 2.1456,
+      "grad_norm": 0.03821471376278644,
+      "learning_rate": 3.9639239147731865e-06,
+      "loss": 0.0002,
       "step": 1341
     },
     {
-      "epoch": 2.0089820359281436,
-      "grad_norm": 0.03331639781525915,
-      "learning_rate": 5.216220599366631e-06,
-      "loss": 0.0003,
+      "epoch": 2.1471999999999998,
+      "grad_norm": 0.10116537076929259,
+      "learning_rate": 3.950155520139581e-06,
+      "loss": 0.0002,
       "step": 1342
     },
     {
-      "epoch": 2.0104790419161676,
-      "grad_norm": 0.03048656266020426,
-      "learning_rate": 5.202028184786807e-06,
-      "loss": 0.0003,
+      "epoch": 2.1488,
+      "grad_norm": 0.0013363768944560507,
+      "learning_rate": 3.936405191259891e-06,
+      "loss": 0.0,
       "step": 1343
     },
     {
-      "epoch": 2.0119760479041915,
-      "grad_norm": 0.10036632313494923,
-      "learning_rate": 5.187848313505042e-06,
-      "loss": 0.0008,
+      "epoch": 2.1504,
+      "grad_norm": 0.008247262958237406,
+      "learning_rate": 3.9226729691946865e-06,
+      "loss": 0.0001,
       "step": 1344
     },
     {
-      "epoch": 2.0134730538922154,
-      "grad_norm": 0.04333086369466123,
-      "learning_rate": 5.173681022591643e-06,
-      "loss": 0.0004,
+      "epoch": 2.152,
+      "grad_norm": 0.07692749652969899,
+      "learning_rate": 3.908958894950465e-06,
+      "loss": 0.0022,
       "step": 1345
     },
     {
-      "epoch": 2.0149700598802394,
-      "grad_norm": 0.0999560008915293,
-      "learning_rate": 5.159526349084051e-06,
-      "loss": 0.0009,
+      "epoch": 2.1536,
+      "grad_norm": 0.0026683879579601478,
+      "learning_rate": 3.895263009479534e-06,
+      "loss": 0.0,
       "step": 1346
     },
     {
-      "epoch": 2.0164670658682633,
-      "grad_norm": 0.09927472161629344,
-      "learning_rate": 5.145384329986709e-06,
-      "loss": 0.0006,
+      "epoch": 2.1552,
+      "grad_norm": 0.0386498226958223,
+      "learning_rate": 3.881585353679891e-06,
+      "loss": 0.0002,
       "step": 1347
     },
     {
-      "epoch": 2.0179640718562872,
-      "grad_norm": 0.046997627545573885,
-      "learning_rate": 5.13125500227098e-06,
-      "loss": 0.0003,
+      "epoch": 2.1568,
+      "grad_norm": 0.00107301484913968,
+      "learning_rate": 3.867925968395085e-06,
+      "loss": 0.0,
       "step": 1348
     },
     {
-      "epoch": 2.019461077844311,
-      "grad_norm": 0.11233978950395591,
-      "learning_rate": 5.117138402875048e-06,
-      "loss": 0.0011,
+      "epoch": 2.1584,
+      "grad_norm": 0.0009612208926005227,
+      "learning_rate": 3.854284894414122e-06,
+      "loss": 0.0,
       "step": 1349
     },
     {
-      "epoch": 2.020958083832335,
-      "grad_norm": 0.04511729239488089,
-      "learning_rate": 5.103034568703819e-06,
-      "loss": 0.0002,
+      "epoch": 2.16,
+      "grad_norm": 0.08101908726754564,
+      "learning_rate": 3.840662172471315e-06,
+      "loss": 0.0003,
       "step": 1350
     },
     {
-      "epoch": 2.0224550898203595,
-      "grad_norm": 0.08572458482338798,
-      "learning_rate": 5.088943536628843e-06,
-      "loss": 0.0006,
+      "epoch": 2.1616,
+      "grad_norm": 0.007755857170339103,
+      "learning_rate": 3.827057843246181e-06,
+      "loss": 0.0001,
       "step": 1351
     },
     {
-      "epoch": 2.0239520958083834,
-      "grad_norm": 0.07982885735619828,
-      "learning_rate": 5.0748653434881735e-06,
-      "loss": 0.0009,
+      "epoch": 2.1632,
+      "grad_norm": 0.003052486130222796,
+      "learning_rate": 3.8134719473633098e-06,
+      "loss": 0.0001,
       "step": 1352
     },
     {
-      "epoch": 2.0254491017964074,
-      "grad_norm": 0.05293702214000303,
-      "learning_rate": 5.060800026086322e-06,
+      "epoch": 2.1648,
+      "grad_norm": 0.0633405164190523,
+      "learning_rate": 3.799904525392251e-06,
       "loss": 0.0004,
       "step": 1353
     },
     {
-      "epoch": 2.0269461077844313,
-      "grad_norm": 0.10137109013871722,
-      "learning_rate": 5.046747621194132e-06,
-      "loss": 0.0004,
+      "epoch": 2.1664,
+      "grad_norm": 0.026255366446261995,
+      "learning_rate": 3.786355617847385e-06,
+      "loss": 0.0001,
       "step": 1354
     },
     {
-      "epoch": 2.0284431137724552,
-      "grad_norm": 0.17093193553427508,
-      "learning_rate": 5.032708165548682e-06,
-      "loss": 0.0007,
+      "epoch": 2.168,
+      "grad_norm": 0.02745119083760212,
+      "learning_rate": 3.7728252651878018e-06,
+      "loss": 0.0001,
       "step": 1355
     },
     {
-      "epoch": 2.029940119760479,
-      "grad_norm": 0.05939541972194821,
-      "learning_rate": 5.018681695853219e-06,
-      "loss": 0.0003,
+      "epoch": 2.1696,
+      "grad_norm": 0.010410191546191274,
+      "learning_rate": 3.759313507817196e-06,
+      "loss": 0.0001,
       "step": 1356
     },
     {
-      "epoch": 2.031437125748503,
-      "grad_norm": 0.05500060420421118,
-      "learning_rate": 5.00466824877701e-06,
-      "loss": 0.0003,
+      "epoch": 2.1712,
+      "grad_norm": 0.01186546330776542,
+      "learning_rate": 3.745820386083724e-06,
+      "loss": 0.0001,
       "step": 1357
     },
     {
-      "epoch": 2.032934131736527,
-      "grad_norm": 0.04529500584356061,
-      "learning_rate": 4.99066786095531e-06,
-      "loss": 0.0004,
+      "epoch": 2.1728,
+      "grad_norm": 0.004319934470598465,
+      "learning_rate": 3.7323459402798936e-06,
+      "loss": 0.0001,
       "step": 1358
     },
     {
-      "epoch": 2.034431137724551,
-      "grad_norm": 0.1107914946140763,
-      "learning_rate": 4.976680568989203e-06,
-      "loss": 0.0008,
+      "epoch": 2.1744,
+      "grad_norm": 0.002464210021865245,
+      "learning_rate": 3.718890210642442e-06,
+      "loss": 0.0,
       "step": 1359
     },
     {
-      "epoch": 2.035928143712575,
-      "grad_norm": 0.13392674083255784,
-      "learning_rate": 4.9627064094455594e-06,
-      "loss": 0.0009,
+      "epoch": 2.176,
+      "grad_norm": 0.005163126523659202,
+      "learning_rate": 3.705453237352227e-06,
+      "loss": 0.0001,
       "step": 1360
     },
     {
-      "epoch": 2.037425149700599,
-      "grad_norm": 0.10073833115439051,
-      "learning_rate": 4.948745418856908e-06,
-      "loss": 0.0009,
+      "epoch": 2.1776,
+      "grad_norm": 0.0055654021392102775,
+      "learning_rate": 3.6920350605340883e-06,
+      "loss": 0.0001,
       "step": 1361
     },
     {
-      "epoch": 2.038922155688623,
-      "grad_norm": 0.06806623307714545,
-      "learning_rate": 4.934797633721343e-06,
-      "loss": 0.0003,
+      "epoch": 2.1792,
+      "grad_norm": 0.005168560794998334,
+      "learning_rate": 3.6786357202567367e-06,
+      "loss": 0.0001,
       "step": 1362
     },
     {
-      "epoch": 2.0404191616766467,
-      "grad_norm": 0.08373463162355546,
-      "learning_rate": 4.92086309050246e-06,
-      "loss": 0.001,
+      "epoch": 2.1808,
+      "grad_norm": 0.0028929438154425002,
+      "learning_rate": 3.6652552565326382e-06,
+      "loss": 0.0,
       "step": 1363
     },
     {
-      "epoch": 2.0419161676646707,
-      "grad_norm": 0.06828569633730956,
-      "learning_rate": 4.906941825629203e-06,
-      "loss": 0.0005,
+      "epoch": 2.1824,
+      "grad_norm": 0.016388761955282143,
+      "learning_rate": 3.6518937093178873e-06,
+      "loss": 0.0001,
       "step": 1364
     },
     {
-      "epoch": 2.0434131736526946,
-      "grad_norm": 0.030176625061175792,
-      "learning_rate": 4.893033875495831e-06,
-      "loss": 0.0002,
+      "epoch": 2.184,
+      "grad_norm": 0.0022084420735664547,
+      "learning_rate": 3.638551118512089e-06,
+      "loss": 0.0,
       "step": 1365
     },
     {
-      "epoch": 2.0449101796407185,
-      "grad_norm": 0.13266862164504664,
-      "learning_rate": 4.879139276461779e-06,
-      "loss": 0.0009,
+      "epoch": 2.1856,
+      "grad_norm": 0.005667837762482932,
+      "learning_rate": 3.6252275239582522e-06,
+      "loss": 0.0001,
       "step": 1366
     },
     {
-      "epoch": 2.0464071856287425,
-      "grad_norm": 0.039122073400303665,
-      "learning_rate": 4.865258064851579e-06,
-      "loss": 0.0003,
+      "epoch": 2.1872,
+      "grad_norm": 0.0037123820490449293,
+      "learning_rate": 3.611922965442648e-06,
+      "loss": 0.0001,
       "step": 1367
     },
     {
-      "epoch": 2.0479041916167664,
-      "grad_norm": 0.08056457623508984,
-      "learning_rate": 4.851390276954779e-06,
-      "loss": 0.001,
+      "epoch": 2.1888,
+      "grad_norm": 0.01350224261078892,
+      "learning_rate": 3.5986374826947067e-06,
+      "loss": 0.0001,
       "step": 1368
     },
     {
-      "epoch": 2.0494011976047903,
-      "grad_norm": 0.12580066367234927,
-      "learning_rate": 4.837535949025807e-06,
-      "loss": 0.0007,
+      "epoch": 2.1904,
+      "grad_norm": 0.0035749356635064347,
+      "learning_rate": 3.5853711153868962e-06,
+      "loss": 0.0001,
       "step": 1369
     },
     {
-      "epoch": 2.0508982035928143,
-      "grad_norm": 0.1074019479468816,
-      "learning_rate": 4.823695117283929e-06,
-      "loss": 0.0007,
+      "epoch": 2.192,
+      "grad_norm": 0.0021972056937981984,
+      "learning_rate": 3.5721239031346067e-06,
+      "loss": 0.0,
       "step": 1370
     },
     {
-      "epoch": 2.052395209580838,
-      "grad_norm": 0.14807294780861768,
-      "learning_rate": 4.809867817913114e-06,
-      "loss": 0.0012,
+      "epoch": 2.1936,
+      "grad_norm": 0.0013664134799337483,
+      "learning_rate": 3.558895885496023e-06,
+      "loss": 0.0,
       "step": 1371
     },
     {
-      "epoch": 2.053892215568862,
-      "grad_norm": 0.04323366086908764,
-      "learning_rate": 4.796054087061955e-06,
-      "loss": 0.0003,
+      "epoch": 2.1952,
+      "grad_norm": 0.038637633743094854,
+      "learning_rate": 3.545687101972013e-06,
+      "loss": 0.0002,
       "step": 1372
     },
     {
-      "epoch": 2.055389221556886,
-      "grad_norm": 0.11724138562817155,
-      "learning_rate": 4.782253960843577e-06,
-      "loss": 0.0007,
+      "epoch": 2.1968,
+      "grad_norm": 0.019984493497667098,
+      "learning_rate": 3.53249759200601e-06,
+      "loss": 0.0002,
       "step": 1373
     },
     {
-      "epoch": 2.05688622754491,
-      "grad_norm": 0.0943855227419094,
-      "learning_rate": 4.768467475335531e-06,
-      "loss": 0.0007,
+      "epoch": 2.1984,
+      "grad_norm": 0.004848472861931771,
+      "learning_rate": 3.519327394983888e-06,
+      "loss": 0.0001,
       "step": 1374
     },
     {
-      "epoch": 2.058383233532934,
-      "grad_norm": 0.03896694150333076,
-      "learning_rate": 4.7546946665797215e-06,
+      "epoch": 2.2,
+      "grad_norm": 0.06446766626275639,
+      "learning_rate": 3.506176550233863e-06,
       "loss": 0.0003,
       "step": 1375
     },
     {
-      "epoch": 2.059880239520958,
-      "grad_norm": 0.058883121619071395,
-      "learning_rate": 4.740935570582276e-06,
-      "loss": 0.0004,
+      "epoch": 2.2016,
+      "grad_norm": 0.005012667702005252,
+      "learning_rate": 3.4930450970263485e-06,
+      "loss": 0.0001,
       "step": 1376
     },
     {
-      "epoch": 2.061377245508982,
-      "grad_norm": 0.11793570437876384,
-      "learning_rate": 4.7271902233134946e-06,
-      "loss": 0.001,
+      "epoch": 2.2032,
+      "grad_norm": 0.002424923752008124,
+      "learning_rate": 3.479933074573858e-06,
+      "loss": 0.0,
       "step": 1377
     },
     {
-      "epoch": 2.062874251497006,
-      "grad_norm": 0.15311600066554015,
-      "learning_rate": 4.713458660707723e-06,
-      "loss": 0.0006,
+      "epoch": 2.2048,
+      "grad_norm": 0.025725036875343558,
+      "learning_rate": 3.4668405220308797e-06,
+      "loss": 0.0001,
       "step": 1378
     },
     {
-      "epoch": 2.06437125748503,
-      "grad_norm": 0.04019764610619817,
-      "learning_rate": 4.699740918663271e-06,
-      "loss": 0.0002,
+      "epoch": 2.2064,
+      "grad_norm": 0.0015935682070169054,
+      "learning_rate": 3.453767478493761e-06,
+      "loss": 0.0,
       "step": 1379
     },
     {
-      "epoch": 2.065868263473054,
-      "grad_norm": 0.1371190868854585,
-      "learning_rate": 4.686037033042319e-06,
-      "loss": 0.0009,
+      "epoch": 2.208,
+      "grad_norm": 0.012371608443721404,
+      "learning_rate": 3.440713983000601e-06,
+      "loss": 0.0001,
       "step": 1380
     },
     {
-      "epoch": 2.067365269461078,
-      "grad_norm": 0.19257550530901746,
-      "learning_rate": 4.672347039670817e-06,
-      "loss": 0.0022,
+      "epoch": 2.2096,
+      "grad_norm": 0.0060595021828897914,
+      "learning_rate": 3.4276800745311135e-06,
+      "loss": 0.0001,
       "step": 1381
     },
     {
-      "epoch": 2.068862275449102,
-      "grad_norm": 0.19645409716014256,
-      "learning_rate": 4.65867097433841e-06,
-      "loss": 0.0014,
+      "epoch": 2.2112,
+      "grad_norm": 0.0018400367957329426,
+      "learning_rate": 3.4146657920065286e-06,
+      "loss": 0.0001,
       "step": 1382
     },
     {
-      "epoch": 2.070359281437126,
-      "grad_norm": 0.05481103486692159,
-      "learning_rate": 4.645008872798318e-06,
-      "loss": 0.0002,
+      "epoch": 2.2128,
+      "grad_norm": 0.03072977747731515,
+      "learning_rate": 3.401671174289469e-06,
+      "loss": 0.0001,
       "step": 1383
     },
     {
-      "epoch": 2.07185628742515,
-      "grad_norm": 0.05062409226086992,
-      "learning_rate": 4.63136077076726e-06,
-      "loss": 0.0004,
+      "epoch": 2.2144,
+      "grad_norm": 0.022035702869944843,
+      "learning_rate": 3.3886962601838327e-06,
+      "loss": 0.0002,
       "step": 1384
     },
     {
-      "epoch": 2.0733532934131738,
-      "grad_norm": 0.06611132196602185,
-      "learning_rate": 4.617726703925357e-06,
-      "loss": 0.0004,
+      "epoch": 2.216,
+      "grad_norm": 0.11692361168448522,
+      "learning_rate": 3.37574108843469e-06,
+      "loss": 0.0009,
       "step": 1385
     },
     {
-      "epoch": 2.0748502994011977,
-      "grad_norm": 0.12704112517705748,
-      "learning_rate": 4.604106707916038e-06,
-      "loss": 0.0007,
+      "epoch": 2.2176,
+      "grad_norm": 0.13711198884703038,
+      "learning_rate": 3.3628056977281456e-06,
+      "loss": 0.0003,
       "step": 1386
     },
     {
-      "epoch": 2.0763473053892216,
-      "grad_norm": 0.11551006878341365,
-      "learning_rate": 4.590500818345943e-06,
-      "loss": 0.0012,
+      "epoch": 2.2192,
+      "grad_norm": 0.012664165920126235,
+      "learning_rate": 3.3498901266912397e-06,
+      "loss": 0.0001,
       "step": 1387
     },
     {
-      "epoch": 2.0778443113772456,
-      "grad_norm": 0.21126725853949585,
-      "learning_rate": 4.576909070784836e-06,
-      "loss": 0.001,
+      "epoch": 2.2208,
+      "grad_norm": 0.004337854227471113,
+      "learning_rate": 3.3369944138918286e-06,
+      "loss": 0.0001,
       "step": 1388
     },
     {
-      "epoch": 2.0793413173652695,
-      "grad_norm": 0.04271579782342062,
-      "learning_rate": 4.5633315007655165e-06,
-      "loss": 0.0004,
+      "epoch": 2.2224,
+      "grad_norm": 0.0011994567469251882,
+      "learning_rate": 3.3241185978384636e-06,
+      "loss": 0.0,
       "step": 1389
     },
     {
-      "epoch": 2.0808383233532934,
-      "grad_norm": 0.035707781270513814,
-      "learning_rate": 4.549768143783709e-06,
-      "loss": 0.0002,
+      "epoch": 2.224,
+      "grad_norm": 0.020569906279623306,
+      "learning_rate": 3.3112627169802948e-06,
+      "loss": 0.0001,
       "step": 1390
     },
     {
-      "epoch": 2.0823353293413174,
-      "grad_norm": 0.08987053581533556,
-      "learning_rate": 4.5362190352979875e-06,
-      "loss": 0.0006,
+      "epoch": 2.2256,
+      "grad_norm": 0.029085218592862228,
+      "learning_rate": 3.2984268097069284e-06,
+      "loss": 0.0001,
       "step": 1391
     },
     {
-      "epoch": 2.0838323353293413,
-      "grad_norm": 0.022527899659366517,
-      "learning_rate": 4.522684210729673e-06,
-      "loss": 0.0002,
+      "epoch": 2.2272,
+      "grad_norm": 0.030632534886056245,
+      "learning_rate": 3.2856109143483316e-06,
+      "loss": 0.0001,
       "step": 1392
     },
     {
-      "epoch": 2.0853293413173652,
-      "grad_norm": 0.09811043078353192,
-      "learning_rate": 4.509163705462746e-06,
-      "loss": 0.0006,
+      "epoch": 2.2288,
+      "grad_norm": 0.04838872988537248,
+      "learning_rate": 3.2728150691747117e-06,
+      "loss": 0.0002,
       "step": 1393
     },
     {
-      "epoch": 2.086826347305389,
-      "grad_norm": 0.10474502941427184,
-      "learning_rate": 4.495657554843747e-06,
-      "loss": 0.0012,
+      "epoch": 2.2304,
+      "grad_norm": 0.020080576934613407,
+      "learning_rate": 3.2600393123964114e-06,
+      "loss": 0.0001,
       "step": 1394
     },
     {
-      "epoch": 2.088323353293413,
-      "grad_norm": 0.05740099228948984,
-      "learning_rate": 4.482165794181702e-06,
-      "loss": 0.0003,
+      "epoch": 2.232,
+      "grad_norm": 0.019423872222649144,
+      "learning_rate": 3.2472836821637744e-06,
+      "loss": 0.0001,
       "step": 1395
     },
     {
-      "epoch": 2.089820359281437,
-      "grad_norm": 0.07473234668055903,
-      "learning_rate": 4.468688458748006e-06,
-      "loss": 0.0005,
+      "epoch": 2.2336,
+      "grad_norm": 0.0018257495221678708,
+      "learning_rate": 3.2345482165670493e-06,
+      "loss": 0.0,
       "step": 1396
     },
     {
-      "epoch": 2.091317365269461,
-      "grad_norm": 0.034739985538301855,
-      "learning_rate": 4.455225583776346e-06,
-      "loss": 0.0003,
+      "epoch": 2.2352,
+      "grad_norm": 0.002354558151618293,
+      "learning_rate": 3.22183295363627e-06,
+      "loss": 0.0,
       "step": 1397
     },
     {
-      "epoch": 2.092814371257485,
-      "grad_norm": 0.06386517981095234,
-      "learning_rate": 4.441777204462603e-06,
-      "loss": 0.0004,
+      "epoch": 2.2368,
+      "grad_norm": 0.03718599597546754,
+      "learning_rate": 3.209137931341143e-06,
+      "loss": 0.0001,
       "step": 1398
     },
     {
-      "epoch": 2.094311377245509,
-      "grad_norm": 0.02542788250908643,
-      "learning_rate": 4.4283433559647615e-06,
-      "loss": 0.0002,
+      "epoch": 2.2384,
+      "grad_norm": 0.001302556047675094,
+      "learning_rate": 3.196463187590929e-06,
+      "loss": 0.0,
       "step": 1399
     },
     {
-      "epoch": 2.095808383233533,
-      "grad_norm": 0.05524457421603561,
-      "learning_rate": 4.4149240734028305e-06,
-      "loss": 0.0007,
+      "epoch": 2.24,
+      "grad_norm": 0.012379528967152616,
+      "learning_rate": 3.183808760234335e-06,
+      "loss": 0.0001,
       "step": 1400
     },
     {
-      "epoch": 2.0973053892215567,
-      "grad_norm": 0.08145735954485667,
-      "learning_rate": 4.401519391858716e-06,
-      "loss": 0.0005,
+      "epoch": 2.2416,
+      "grad_norm": 0.007406360902967913,
+      "learning_rate": 3.1711746870594083e-06,
+      "loss": 0.0001,
       "step": 1401
     },
     {
-      "epoch": 2.0988023952095807,
-      "grad_norm": 0.13001843472521485,
-      "learning_rate": 4.388129346376177e-06,
-      "loss": 0.0009,
+      "epoch": 2.2432,
+      "grad_norm": 0.011146375592944902,
+      "learning_rate": 3.1585610057934022e-06,
+      "loss": 0.0001,
       "step": 1402
     },
     {
-      "epoch": 2.1002994011976046,
-      "grad_norm": 0.04328997396656813,
-      "learning_rate": 4.374753971960695e-06,
-      "loss": 0.0004,
+      "epoch": 2.2448,
+      "grad_norm": 0.0031344599212710265,
+      "learning_rate": 3.145967754102691e-06,
+      "loss": 0.0001,
       "step": 1403
     },
     {
-      "epoch": 2.1017964071856285,
-      "grad_norm": 0.050812707704053536,
-      "learning_rate": 4.3613933035794e-06,
-      "loss": 0.0003,
+      "epoch": 2.2464,
+      "grad_norm": 0.0027151041816300832,
+      "learning_rate": 3.1333949695926323e-06,
+      "loss": 0.0,
       "step": 1404
     },
     {
-      "epoch": 2.1032934131736525,
-      "grad_norm": 0.12459385212805167,
-      "learning_rate": 4.348047376160977e-06,
-      "loss": 0.0011,
+      "epoch": 2.248,
+      "grad_norm": 0.04285351348573601,
+      "learning_rate": 3.1208426898074685e-06,
+      "loss": 0.0003,
       "step": 1405
     },
     {
-      "epoch": 2.1047904191616764,
-      "grad_norm": 0.09492623023746591,
-      "learning_rate": 4.334716224595571e-06,
-      "loss": 0.0006,
+      "epoch": 2.2496,
+      "grad_norm": 0.0039732391753378405,
+      "learning_rate": 3.1083109522302124e-06,
+      "loss": 0.0001,
       "step": 1406
     },
     {
-      "epoch": 2.106287425149701,
-      "grad_norm": 0.07989452008235769,
-      "learning_rate": 4.321399883734712e-06,
-      "loss": 0.0005,
+      "epoch": 2.2512,
+      "grad_norm": 0.04832900212033141,
+      "learning_rate": 3.0957997942825337e-06,
+      "loss": 0.0002,
       "step": 1407
     },
     {
-      "epoch": 2.1077844311377247,
-      "grad_norm": 0.07205507467214367,
-      "learning_rate": 4.308098388391184e-06,
-      "loss": 0.0004,
+      "epoch": 2.2528,
+      "grad_norm": 0.1470983997960054,
+      "learning_rate": 3.083309253324651e-06,
+      "loss": 0.0007,
       "step": 1408
     },
     {
-      "epoch": 2.1092814371257487,
-      "grad_norm": 0.017970693243257126,
-      "learning_rate": 4.2948117733389896e-06,
-      "loss": 0.0002,
+      "epoch": 2.2544,
+      "grad_norm": 0.08966233019627902,
+      "learning_rate": 3.070839366655215e-06,
+      "loss": 0.0003,
       "step": 1409
     },
     {
-      "epoch": 2.1107784431137726,
-      "grad_norm": 0.08565640149022209,
-      "learning_rate": 4.28154007331321e-06,
-      "loss": 0.0006,
+      "epoch": 2.2560000000000002,
+      "grad_norm": 0.0217570675497833,
+      "learning_rate": 3.0583901715111965e-06,
+      "loss": 0.0001,
       "step": 1410
     },
     {
-      "epoch": 2.1122754491017965,
-      "grad_norm": 0.014993195128250375,
-      "learning_rate": 4.268283323009941e-06,
-      "loss": 0.0002,
+      "epoch": 2.2576,
+      "grad_norm": 0.0023984671122352836,
+      "learning_rate": 3.045961705067787e-06,
+      "loss": 0.0,
       "step": 1411
     },
     {
-      "epoch": 2.1137724550898205,
-      "grad_norm": 0.04396523431843643,
-      "learning_rate": 4.255041557086202e-06,
-      "loss": 0.0005,
+      "epoch": 2.2592,
+      "grad_norm": 0.03646686408219286,
+      "learning_rate": 3.0335540044382693e-06,
+      "loss": 0.0002,
       "step": 1412
     },
     {
-      "epoch": 2.1152694610778444,
-      "grad_norm": 0.07072673143683754,
-      "learning_rate": 4.2418148101598215e-06,
-      "loss": 0.0004,
+      "epoch": 2.2608,
+      "grad_norm": 0.019759144087226437,
+      "learning_rate": 3.021167106673928e-06,
+      "loss": 0.0001,
       "step": 1413
     },
     {
-      "epoch": 2.1167664670658684,
-      "grad_norm": 0.055515949400535856,
-      "learning_rate": 4.228603116809382e-06,
-      "loss": 0.0006,
+      "epoch": 2.2624,
+      "grad_norm": 0.0013618437184803305,
+      "learning_rate": 3.008801048763914e-06,
+      "loss": 0.0,
       "step": 1414
     },
     {
-      "epoch": 2.1182634730538923,
-      "grad_norm": 0.05168252557459792,
-      "learning_rate": 4.215406511574104e-06,
-      "loss": 0.0005,
+      "epoch": 2.2640000000000002,
+      "grad_norm": 0.019446950152418513,
+      "learning_rate": 2.996455867635155e-06,
+      "loss": 0.0001,
       "step": 1415
     },
     {
-      "epoch": 2.1197604790419162,
-      "grad_norm": 0.11254802035543135,
-      "learning_rate": 4.202225028953758e-06,
-      "loss": 0.0013,
+      "epoch": 2.2656,
+      "grad_norm": 0.15992625903637472,
+      "learning_rate": 2.9841316001522345e-06,
+      "loss": 0.0011,
       "step": 1416
     },
     {
-      "epoch": 2.12125748502994,
-      "grad_norm": 0.032013853884326086,
-      "learning_rate": 4.189058703408596e-06,
+      "epoch": 2.2672,
+      "grad_norm": 0.033021831659926115,
+      "learning_rate": 2.9718282831172885e-06,
       "loss": 0.0002,
       "step": 1417
     },
     {
-      "epoch": 2.122754491017964,
-      "grad_norm": 0.0976156994347326,
-      "learning_rate": 4.175907569359219e-06,
-      "loss": 0.0007,
+      "epoch": 2.2688,
+      "grad_norm": 0.0025513788859806105,
+      "learning_rate": 2.9595459532698854e-06,
+      "loss": 0.0,
       "step": 1418
     },
     {
-      "epoch": 2.124251497005988,
-      "grad_norm": 0.05995922146986086,
-      "learning_rate": 4.162771661186544e-06,
-      "loss": 0.0004,
+      "epoch": 2.2704,
+      "grad_norm": 0.0010193538052206404,
+      "learning_rate": 2.94728464728693e-06,
+      "loss": 0.0,
       "step": 1419
     },
     {
-      "epoch": 2.125748502994012,
-      "grad_norm": 0.1156120014758531,
-      "learning_rate": 4.149651013231651e-06,
-      "loss": 0.0007,
+      "epoch": 2.2720000000000002,
+      "grad_norm": 0.03361224556996153,
+      "learning_rate": 2.9350444017825385e-06,
+      "loss": 0.0002,
       "step": 1420
     },
     {
-      "epoch": 2.127245508982036,
-      "grad_norm": 0.0728234683105191,
-      "learning_rate": 4.136545659795754e-06,
-      "loss": 0.0003,
+      "epoch": 2.2736,
+      "grad_norm": 0.0018094678979295072,
+      "learning_rate": 2.922825253307947e-06,
+      "loss": 0.0,
       "step": 1421
     },
     {
-      "epoch": 2.12874251497006,
-      "grad_norm": 0.04225258816552858,
-      "learning_rate": 4.123455635140066e-06,
-      "loss": 0.0002,
+      "epoch": 2.2752,
+      "grad_norm": 0.006488564779726169,
+      "learning_rate": 2.910627238351383e-06,
+      "loss": 0.0001,
       "step": 1422
     },
     {
-      "epoch": 2.1302395209580838,
-      "grad_norm": 0.07510640021056275,
-      "learning_rate": 4.110380973485726e-06,
+      "epoch": 2.2768,
+      "grad_norm": 0.16412732998683924,
+      "learning_rate": 2.898450393337977e-06,
       "loss": 0.0007,
       "step": 1423
     },
     {
-      "epoch": 2.1317365269461077,
-      "grad_norm": 0.11472823224242246,
-      "learning_rate": 4.097321709013725e-06,
-      "loss": 0.0009,
+      "epoch": 2.2784,
+      "grad_norm": 0.009226294554457398,
+      "learning_rate": 2.886294754629632e-06,
+      "loss": 0.0001,
       "step": 1424
     },
     {
-      "epoch": 2.1332335329341316,
-      "grad_norm": 0.04334119016243774,
-      "learning_rate": 4.084277875864776e-06,
-      "loss": 0.0002,
+      "epoch": 2.2800000000000002,
+      "grad_norm": 0.0016057887977407946,
+      "learning_rate": 2.8741603585249312e-06,
+      "loss": 0.0,
       "step": 1425
     },
     {
-      "epoch": 2.1347305389221556,
-      "grad_norm": 0.03141663452074846,
-      "learning_rate": 4.071249508139275e-06,
-      "loss": 0.0003,
+      "epoch": 2.2816,
+      "grad_norm": 0.025719950014896933,
+      "learning_rate": 2.8620472412590227e-06,
+      "loss": 0.0002,
       "step": 1426
     },
     {
-      "epoch": 2.1362275449101795,
-      "grad_norm": 0.06277111693317153,
-      "learning_rate": 4.058236639897171e-06,
-      "loss": 0.0004,
+      "epoch": 2.2832,
+      "grad_norm": 0.09835938693598928,
+      "learning_rate": 2.8499554390035144e-06,
+      "loss": 0.0013,
       "step": 1427
     },
     {
-      "epoch": 2.1377245508982035,
-      "grad_norm": 0.016403479959175616,
-      "learning_rate": 4.0452393051579e-06,
+      "epoch": 2.2848,
+      "grad_norm": 0.002955627209750324,
+      "learning_rate": 2.837884987866363e-06,
       "loss": 0.0001,
       "step": 1428
     },
     {
-      "epoch": 2.1392215568862274,
-      "grad_norm": 0.07804687214286263,
-      "learning_rate": 4.032257537900286e-06,
-      "loss": 0.0006,
+      "epoch": 2.2864,
+      "grad_norm": 0.0016515335057602617,
+      "learning_rate": 2.8258359238917665e-06,
+      "loss": 0.0,
       "step": 1429
     },
     {
-      "epoch": 2.1407185628742513,
-      "grad_norm": 0.07304022000926429,
-      "learning_rate": 4.019291372062452e-06,
-      "loss": 0.0009,
+      "epoch": 2.288,
+      "grad_norm": 0.006663918000619816,
+      "learning_rate": 2.8138082830600556e-06,
+      "loss": 0.0,
       "step": 1430
     },
     {
-      "epoch": 2.1422155688622753,
-      "grad_norm": 0.04838062003714075,
-      "learning_rate": 4.006340841541748e-06,
-      "loss": 0.0004,
+      "epoch": 2.2896,
+      "grad_norm": 0.032230510931899596,
+      "learning_rate": 2.8018021012875994e-06,
+      "loss": 0.0002,
       "step": 1431
     },
     {
-      "epoch": 2.143712574850299,
-      "grad_norm": 0.012920730564890766,
-      "learning_rate": 3.993405980194627e-06,
+      "epoch": 2.2912,
+      "grad_norm": 0.025771170503503277,
+      "learning_rate": 2.789817414426673e-06,
       "loss": 0.0001,
       "step": 1432
     },
     {
-      "epoch": 2.1452095808383236,
-      "grad_norm": 0.02060350725486784,
-      "learning_rate": 3.980486821836599e-06,
-      "loss": 0.0002,
+      "epoch": 2.2928,
+      "grad_norm": 0.0012859432517395015,
+      "learning_rate": 2.7778542582653746e-06,
+      "loss": 0.0,
       "step": 1433
     },
     {
-      "epoch": 2.1467065868263475,
-      "grad_norm": 0.1090178395185067,
-      "learning_rate": 3.967583400242109e-06,
-      "loss": 0.0012,
+      "epoch": 2.2944,
+      "grad_norm": 0.01411440299705994,
+      "learning_rate": 2.7659126685275028e-06,
+      "loss": 0.0001,
       "step": 1434
     },
     {
-      "epoch": 2.1482035928143715,
-      "grad_norm": 0.08624476103626773,
-      "learning_rate": 3.954695749144466e-06,
-      "loss": 0.0004,
+      "epoch": 2.296,
+      "grad_norm": 0.07607062065238816,
+      "learning_rate": 2.753992680872457e-06,
+      "loss": 0.001,
       "step": 1435
     },
     {
-      "epoch": 2.1497005988023954,
-      "grad_norm": 0.02414965108037553,
-      "learning_rate": 3.941823902235751e-06,
-      "loss": 0.0002,
+      "epoch": 2.2976,
+      "grad_norm": 0.09799565332045054,
+      "learning_rate": 2.7420943308951287e-06,
+      "loss": 0.0008,
       "step": 1436
     },
     {
-      "epoch": 2.1511976047904193,
-      "grad_norm": 0.01602205069339523,
-      "learning_rate": 3.928967893166721e-06,
+      "epoch": 2.2992,
+      "grad_norm": 0.017320946061261017,
+      "learning_rate": 2.7302176541257984e-06,
       "loss": 0.0001,
       "step": 1437
     },
     {
-      "epoch": 2.1526946107784433,
-      "grad_norm": 0.15078451246023045,
-      "learning_rate": 3.916127755546743e-06,
-      "loss": 0.0008,
+      "epoch": 2.3008,
+      "grad_norm": 0.010239412224088713,
+      "learning_rate": 2.718362686030025e-06,
+      "loss": 0.0001,
       "step": 1438
     },
     {
-      "epoch": 2.154191616766467,
-      "grad_norm": 0.09937561764681273,
-      "learning_rate": 3.903303522943679e-06,
-      "loss": 0.0005,
+      "epoch": 2.3024,
+      "grad_norm": 0.03367078960024932,
+      "learning_rate": 2.7065294620085425e-06,
+      "loss": 0.0001,
       "step": 1439
     },
     {
-      "epoch": 2.155688622754491,
-      "grad_norm": 0.2778830186344834,
-      "learning_rate": 3.890495228883814e-06,
-      "loss": 0.0013,
+      "epoch": 2.304,
+      "grad_norm": 0.036310253948025895,
+      "learning_rate": 2.694718017397151e-06,
+      "loss": 0.0002,
       "step": 1440
     },
     {
-      "epoch": 2.157185628742515,
-      "grad_norm": 0.1014703272227765,
-      "learning_rate": 3.877702906851765e-06,
-      "loss": 0.0011,
+      "epoch": 2.3056,
+      "grad_norm": 0.08174961939548375,
+      "learning_rate": 2.6829283874666236e-06,
+      "loss": 0.0006,
       "step": 1441
     },
     {
-      "epoch": 2.158682634730539,
-      "grad_norm": 0.03910173601850433,
-      "learning_rate": 3.8649265902903935e-06,
-      "loss": 0.0003,
+      "epoch": 2.3072,
+      "grad_norm": 0.0020749643937138526,
+      "learning_rate": 2.6711606074225783e-06,
+      "loss": 0.0,
       "step": 1442
     },
     {
-      "epoch": 2.160179640718563,
-      "grad_norm": 0.0758231387962015,
-      "learning_rate": 3.852166312600715e-06,
-      "loss": 0.0008,
+      "epoch": 2.3088,
+      "grad_norm": 0.0008781990810354615,
+      "learning_rate": 2.6594147124053983e-06,
+      "loss": 0.0,
       "step": 1443
     },
     {
-      "epoch": 2.161676646706587,
-      "grad_norm": 0.0307677226066544,
-      "learning_rate": 3.839422107141826e-06,
-      "loss": 0.0002,
+      "epoch": 2.3104,
+      "grad_norm": 0.020486784780718033,
+      "learning_rate": 2.6476907374901062e-06,
+      "loss": 0.0001,
       "step": 1444
     },
     {
-      "epoch": 2.163173652694611,
-      "grad_norm": 0.09752284917524724,
-      "learning_rate": 3.826694007230792e-06,
-      "loss": 0.0008,
+      "epoch": 2.312,
+      "grad_norm": 0.013170158375038647,
+      "learning_rate": 2.635988717686272e-06,
+      "loss": 0.0001,
       "step": 1445
     },
     {
-      "epoch": 2.1646706586826348,
-      "grad_norm": 0.05457238739954349,
-      "learning_rate": 3.813982046142581e-06,
-      "loss": 0.0004,
+      "epoch": 2.3136,
+      "grad_norm": 0.0024200822912347746,
+      "learning_rate": 2.6243086879379e-06,
+      "loss": 0.0,
       "step": 1446
     },
     {
-      "epoch": 2.1661676646706587,
-      "grad_norm": 0.07303152617434286,
-      "learning_rate": 3.8012862571099673e-06,
-      "loss": 0.0005,
+      "epoch": 2.3152,
+      "grad_norm": 0.008108913518453966,
+      "learning_rate": 2.6126506831233343e-06,
+      "loss": 0.0001,
       "step": 1447
     },
     {
-      "epoch": 2.1676646706586826,
-      "grad_norm": 0.016540715207413653,
-      "learning_rate": 3.788606673323447e-06,
-      "loss": 0.0002,
+      "epoch": 2.3168,
+      "grad_norm": 0.10868201803231775,
+      "learning_rate": 2.6010147380551474e-06,
+      "loss": 0.0005,
       "step": 1448
     },
     {
-      "epoch": 2.1691616766467066,
-      "grad_norm": 0.0802709349097072,
-      "learning_rate": 3.7759433279311487e-06,
-      "loss": 0.0005,
+      "epoch": 2.3184,
+      "grad_norm": 0.004058238469458349,
+      "learning_rate": 2.5894008874800323e-06,
+      "loss": 0.0001,
       "step": 1449
     },
     {
-      "epoch": 2.1706586826347305,
-      "grad_norm": 0.034645966560659984,
-      "learning_rate": 3.7632962540387575e-06,
-      "loss": 0.0003,
+      "epoch": 2.32,
+      "grad_norm": 0.002981692857772985,
+      "learning_rate": 2.577809166078716e-06,
+      "loss": 0.0001,
       "step": 1450
     },
     {
-      "epoch": 2.1721556886227544,
-      "grad_norm": 0.008870689066133116,
-      "learning_rate": 3.7506654847094105e-06,
-      "loss": 0.0002,
+      "epoch": 2.3216,
+      "grad_norm": 0.019086869293865865,
+      "learning_rate": 2.5662396084658383e-06,
+      "loss": 0.0001,
       "step": 1451
     },
     {
-      "epoch": 2.1736526946107784,
-      "grad_norm": 0.07045440620146966,
-      "learning_rate": 3.738051052963624e-06,
-      "loss": 0.0004,
+      "epoch": 2.3232,
+      "grad_norm": 0.04338204063553248,
+      "learning_rate": 2.5546922491898497e-06,
+      "loss": 0.0001,
       "step": 1452
     },
     {
-      "epoch": 2.1751497005988023,
-      "grad_norm": 0.028807941828131546,
-      "learning_rate": 3.7254529917792025e-06,
-      "loss": 0.0002,
+      "epoch": 2.3247999999999998,
+      "grad_norm": 0.004913806069794926,
+      "learning_rate": 2.543167122732918e-06,
+      "loss": 0.0001,
       "step": 1453
     },
     {
-      "epoch": 2.1766467065868262,
-      "grad_norm": 0.04049065453847004,
-      "learning_rate": 3.712871334091154e-06,
-      "loss": 0.0004,
+      "epoch": 2.3264,
+      "grad_norm": 0.0030636921144399165,
+      "learning_rate": 2.5316642635108247e-06,
+      "loss": 0.0,
       "step": 1454
     },
     {
-      "epoch": 2.17814371257485,
-      "grad_norm": 0.07895027308565043,
-      "learning_rate": 3.7003061127915973e-06,
-      "loss": 0.0005,
+      "epoch": 2.328,
+      "grad_norm": 0.0076995128467695325,
+      "learning_rate": 2.5201837058728506e-06,
+      "loss": 0.0001,
       "step": 1455
     },
     {
-      "epoch": 2.179640718562874,
-      "grad_norm": 0.07783525776749904,
-      "learning_rate": 3.6877573607296935e-06,
-      "loss": 0.0007,
+      "epoch": 2.3296,
+      "grad_norm": 0.07273840684610987,
+      "learning_rate": 2.508725484101684e-06,
+      "loss": 0.0004,
       "step": 1456
     },
     {
-      "epoch": 2.181137724550898,
-      "grad_norm": 0.05173489235073667,
-      "learning_rate": 3.67522511071154e-06,
-      "loss": 0.0005,
+      "epoch": 2.3312,
+      "grad_norm": 0.11186530685574775,
+      "learning_rate": 2.4972896324133143e-06,
+      "loss": 0.0007,
       "step": 1457
     },
     {
-      "epoch": 2.182634730538922,
-      "grad_norm": 0.04498118384069647,
-      "learning_rate": 3.6627093955000916e-06,
-      "loss": 0.0006,
+      "epoch": 2.3327999999999998,
+      "grad_norm": 0.0020329487891239295,
+      "learning_rate": 2.485876184956928e-06,
+      "loss": 0.0,
       "step": 1458
     },
     {
-      "epoch": 2.184131736526946,
-      "grad_norm": 0.09963326809445305,
-      "learning_rate": 3.6502102478150814e-06,
-      "loss": 0.0006,
+      "epoch": 2.3344,
+      "grad_norm": 0.1810254816293864,
+      "learning_rate": 2.474485175814816e-06,
+      "loss": 0.0009,
       "step": 1459
     },
     {
-      "epoch": 2.18562874251497,
-      "grad_norm": 0.020623105889192212,
-      "learning_rate": 3.6377277003329238e-06,
-      "loss": 0.0002,
+      "epoch": 2.336,
+      "grad_norm": 0.008796881961622925,
+      "learning_rate": 2.4631166390022574e-06,
+      "loss": 0.0001,
       "step": 1460
     },
     {
-      "epoch": 2.187125748502994,
-      "grad_norm": 0.0842055840763743,
-      "learning_rate": 3.625261785686649e-06,
-      "loss": 0.0006,
+      "epoch": 2.3376,
+      "grad_norm": 0.028234386011551076,
+      "learning_rate": 2.451770608467432e-06,
+      "loss": 0.0002,
       "step": 1461
     },
     {
-      "epoch": 2.1886227544910177,
-      "grad_norm": 0.05511784114857564,
-      "learning_rate": 3.6128125364657806e-06,
-      "loss": 0.0003,
+      "epoch": 2.3392,
+      "grad_norm": 0.0009550241744107475,
+      "learning_rate": 2.440447118091306e-06,
+      "loss": 0.0,
       "step": 1462
     },
     {
-      "epoch": 2.190119760479042,
-      "grad_norm": 0.09680779392796715,
-      "learning_rate": 3.6003799852163e-06,
-      "loss": 0.0006,
+      "epoch": 2.3407999999999998,
+      "grad_norm": 0.06985935816975702,
+      "learning_rate": 2.429146201687538e-06,
+      "loss": 0.0004,
       "step": 1463
     },
     {
-      "epoch": 2.191616766467066,
-      "grad_norm": 0.3290860685579756,
-      "learning_rate": 3.5879641644405183e-06,
-      "loss": 0.0016,
+      "epoch": 2.3424,
+      "grad_norm": 0.001863839046838397,
+      "learning_rate": 2.417867893002387e-06,
+      "loss": 0.0,
       "step": 1464
     },
     {
-      "epoch": 2.19311377245509,
-      "grad_norm": 0.010421333234893785,
-      "learning_rate": 3.575565106597014e-06,
+      "epoch": 2.344,
+      "grad_norm": 0.011244576439991636,
+      "learning_rate": 2.4066122257145898e-06,
       "loss": 0.0001,
       "step": 1465
     },
     {
-      "epoch": 2.194610778443114,
-      "grad_norm": 0.15309336269175572,
-      "learning_rate": 3.563182844100541e-06,
-      "loss": 0.0011,
+      "epoch": 2.3456,
+      "grad_norm": 0.010636737740380322,
+      "learning_rate": 2.3953792334352787e-06,
+      "loss": 0.0001,
       "step": 1466
     },
     {
-      "epoch": 2.196107784431138,
-      "grad_norm": 0.02091267883568856,
-      "learning_rate": 3.5508174093219427e-06,
-      "loss": 0.0001,
+      "epoch": 2.3472,
+      "grad_norm": 0.0014849518550035968,
+      "learning_rate": 2.3841689497078746e-06,
+      "loss": 0.0,
       "step": 1467
     },
     {
-      "epoch": 2.197604790419162,
-      "grad_norm": 0.06446281945069611,
-      "learning_rate": 3.5384688345880826e-06,
-      "loss": 0.0004,
+      "epoch": 2.3487999999999998,
+      "grad_norm": 0.001440058625093697,
+      "learning_rate": 2.3729814080079815e-06,
+      "loss": 0.0,
       "step": 1468
     },
     {
-      "epoch": 2.1991017964071857,
-      "grad_norm": 0.052567869194541594,
-      "learning_rate": 3.5261371521817247e-06,
-      "loss": 0.0004,
+      "epoch": 2.3504,
+      "grad_norm": 0.007474250959261391,
+      "learning_rate": 2.361816641743303e-06,
+      "loss": 0.0001,
       "step": 1469
     },
     {
-      "epoch": 2.2005988023952097,
-      "grad_norm": 0.049578551613729545,
-      "learning_rate": 3.513822394341493e-06,
-      "loss": 0.0006,
+      "epoch": 2.352,
+      "grad_norm": 0.0012983438990565295,
+      "learning_rate": 2.3506746842535244e-06,
+      "loss": 0.0,
       "step": 1470
     },
     {
-      "epoch": 2.2020958083832336,
-      "grad_norm": 0.033783781354055624,
-      "learning_rate": 3.501524593261756e-06,
-      "loss": 0.0002,
+      "epoch": 2.3536,
+      "grad_norm": 0.006703399425770209,
+      "learning_rate": 2.339555568810221e-06,
+      "loss": 0.0001,
       "step": 1471
     },
     {
-      "epoch": 2.2035928143712575,
-      "grad_norm": 0.11505653536880667,
-      "learning_rate": 3.4892437810925484e-06,
-      "loss": 0.0008,
+      "epoch": 2.3552,
+      "grad_norm": 0.002723319057583775,
+      "learning_rate": 2.328459328616759e-06,
+      "loss": 0.0,
       "step": 1472
     },
     {
-      "epoch": 2.2050898203592815,
-      "grad_norm": 0.05951967201522422,
-      "learning_rate": 3.476979989939506e-06,
-      "loss": 0.0002,
+      "epoch": 2.3568,
+      "grad_norm": 0.003221205372001595,
+      "learning_rate": 2.317385996808195e-06,
+      "loss": 0.0,
       "step": 1473
     },
     {
-      "epoch": 2.2065868263473054,
-      "grad_norm": 0.1056924616624737,
-      "learning_rate": 3.4647332518637454e-06,
-      "loss": 0.0005,
+      "epoch": 2.3584,
+      "grad_norm": 0.022913595669498688,
+      "learning_rate": 2.306335606451181e-06,
+      "loss": 0.0001,
       "step": 1474
     },
     {
-      "epoch": 2.2080838323353293,
-      "grad_norm": 0.06831981985885952,
-      "learning_rate": 3.4525035988818232e-06,
-      "loss": 0.0007,
+      "epoch": 2.36,
+      "grad_norm": 0.0014718472698621076,
+      "learning_rate": 2.295308190543859e-06,
+      "loss": 0.0,
       "step": 1475
     },
     {
-      "epoch": 2.2095808383233533,
-      "grad_norm": 0.08416246640850832,
-      "learning_rate": 3.440291062965608e-06,
-      "loss": 0.0004,
+      "epoch": 2.3616,
+      "grad_norm": 0.003057361648101914,
+      "learning_rate": 2.2843037820157678e-06,
+      "loss": 0.0001,
       "step": 1476
     },
     {
-      "epoch": 2.211077844311377,
-      "grad_norm": 0.009399176800184218,
-      "learning_rate": 3.428095676042242e-06,
+      "epoch": 2.3632,
+      "grad_norm": 0.01739809318878192,
+      "learning_rate": 2.2733224137277366e-06,
       "loss": 0.0001,
       "step": 1477
     },
     {
-      "epoch": 2.212574850299401,
-      "grad_norm": 0.06148287830535551,
-      "learning_rate": 3.41591746999402e-06,
-      "loss": 0.0005,
+      "epoch": 2.3648,
+      "grad_norm": 0.004763148347921701,
+      "learning_rate": 2.2623641184718048e-06,
+      "loss": 0.0001,
       "step": 1478
     },
     {
-      "epoch": 2.214071856287425,
-      "grad_norm": 0.15044803640835408,
-      "learning_rate": 3.403756476658323e-06,
-      "loss": 0.0006,
+      "epoch": 2.3664,
+      "grad_norm": 0.004518028811779055,
+      "learning_rate": 2.251428928971102e-06,
+      "loss": 0.0,
       "step": 1479
     },
     {
-      "epoch": 2.215568862275449,
-      "grad_norm": 0.039856311845892195,
-      "learning_rate": 3.3916127278275444e-06,
-      "loss": 0.0003,
+      "epoch": 2.368,
+      "grad_norm": 0.003053904000256683,
+      "learning_rate": 2.240516877879765e-06,
+      "loss": 0.0,
       "step": 1480
     },
     {
-      "epoch": 2.217065868263473,
-      "grad_norm": 0.03298698632561797,
-      "learning_rate": 3.379486255248975e-06,
-      "loss": 0.0004,
+      "epoch": 2.3696,
+      "grad_norm": 0.269895598557991,
+      "learning_rate": 2.229627997782834e-06,
+      "loss": 0.0009,
       "step": 1481
     },
     {
-      "epoch": 2.218562874251497,
-      "grad_norm": 0.05989471368709794,
-      "learning_rate": 3.3673770906247595e-06,
-      "loss": 0.0003,
+      "epoch": 2.3712,
+      "grad_norm": 0.0076793204496350435,
+      "learning_rate": 2.218762321196156e-06,
+      "loss": 0.0001,
       "step": 1482
     },
     {
-      "epoch": 2.220059880239521,
-      "grad_norm": 0.05031206889739225,
-      "learning_rate": 3.355285265611784e-06,
-      "loss": 0.0006,
+      "epoch": 2.3728,
+      "grad_norm": 0.0033461779993014668,
+      "learning_rate": 2.2079198805662917e-06,
+      "loss": 0.0,
       "step": 1483
     },
     {
-      "epoch": 2.2215568862275448,
-      "grad_norm": 0.039637111322910513,
-      "learning_rate": 3.3432108118216046e-06,
-      "loss": 0.0004,
+      "epoch": 2.3744,
+      "grad_norm": 0.001427115516108656,
+      "learning_rate": 2.1971007082704167e-06,
+      "loss": 0.0,
       "step": 1484
     },
     {
-      "epoch": 2.2230538922155687,
-      "grad_norm": 0.02790884081359175,
-      "learning_rate": 3.3311537608203747e-06,
-      "loss": 0.0002,
+      "epoch": 2.376,
+      "grad_norm": 0.002586368385462468,
+      "learning_rate": 2.186304836616221e-06,
+      "loss": 0.0,
       "step": 1485
     },
     {
-      "epoch": 2.2245508982035926,
-      "grad_norm": 0.03455227197831783,
-      "learning_rate": 3.31911414412873e-06,
-      "loss": 0.0003,
+      "epoch": 2.3776,
+      "grad_norm": 0.017664811900103907,
+      "learning_rate": 2.1755322978418134e-06,
+      "loss": 0.0001,
       "step": 1486
     },
     {
-      "epoch": 2.2260479041916166,
-      "grad_norm": 0.06635476538309946,
-      "learning_rate": 3.3070919932217517e-06,
-      "loss": 0.0003,
+      "epoch": 2.3792,
+      "grad_norm": 0.01694283336066184,
+      "learning_rate": 2.1647831241156304e-06,
+      "loss": 0.0001,
       "step": 1487
     },
     {
-      "epoch": 2.2275449101796405,
-      "grad_norm": 0.14675486412500174,
-      "learning_rate": 3.2950873395288454e-06,
-      "loss": 0.0014,
+      "epoch": 2.3808,
+      "grad_norm": 0.032297864205079956,
+      "learning_rate": 2.1540573475363402e-06,
+      "loss": 0.0002,
       "step": 1488
     },
     {
-      "epoch": 2.229041916167665,
-      "grad_norm": 0.1360738382194748,
-      "learning_rate": 3.2831002144336776e-06,
-      "loss": 0.0007,
+      "epoch": 2.3824,
+      "grad_norm": 0.0018598696691681533,
+      "learning_rate": 2.1433550001327376e-06,
+      "loss": 0.0,
       "step": 1489
     },
     {
-      "epoch": 2.230538922155689,
-      "grad_norm": 0.05674441854621314,
-      "learning_rate": 3.2711306492740914e-06,
-      "loss": 0.0004,
+      "epoch": 2.384,
+      "grad_norm": 0.002102643984560031,
+      "learning_rate": 2.1326761138636555e-06,
+      "loss": 0.0,
       "step": 1490
     },
     {
-      "epoch": 2.2320359281437128,
-      "grad_norm": 0.022436227700209574,
-      "learning_rate": 3.2591786753420187e-06,
-      "loss": 0.0001,
+      "epoch": 2.3856,
+      "grad_norm": 0.0018286896919109657,
+      "learning_rate": 2.122020720617869e-06,
+      "loss": 0.0,
       "step": 1491
     },
     {
-      "epoch": 2.2335329341317367,
-      "grad_norm": 0.03489642218056514,
-      "learning_rate": 3.247244323883415e-06,
-      "loss": 0.0003,
+      "epoch": 2.3872,
+      "grad_norm": 0.008289093464070325,
+      "learning_rate": 2.111388852214001e-06,
+      "loss": 0.0001,
       "step": 1492
     },
     {
-      "epoch": 2.2350299401197606,
-      "grad_norm": 0.2050412352519987,
-      "learning_rate": 3.2353276260981447e-06,
-      "loss": 0.0014,
+      "epoch": 2.3888,
+      "grad_norm": 0.0023336863343605095,
+      "learning_rate": 2.1007805404004247e-06,
+      "loss": 0.0,
       "step": 1493
     },
     {
-      "epoch": 2.2365269461077846,
-      "grad_norm": 0.04952916174910589,
-      "learning_rate": 3.2234286131399415e-06,
-      "loss": 0.0003,
+      "epoch": 2.3904,
+      "grad_norm": 0.0028441181808218953,
+      "learning_rate": 2.090195816855164e-06,
+      "loss": 0.0,
       "step": 1494
     },
     {
-      "epoch": 2.2380239520958085,
-      "grad_norm": 0.15678094109556026,
-      "learning_rate": 3.2115473161162936e-06,
-      "loss": 0.0011,
+      "epoch": 2.392,
+      "grad_norm": 0.001212807742393933,
+      "learning_rate": 2.0796347131858187e-06,
+      "loss": 0.0,
       "step": 1495
     },
     {
-      "epoch": 2.2395209580838324,
-      "grad_norm": 0.0359162039363321,
-      "learning_rate": 3.1996837660883763e-06,
-      "loss": 0.0003,
+      "epoch": 2.3936,
+      "grad_norm": 0.005296837578609509,
+      "learning_rate": 2.069097260929439e-06,
+      "loss": 0.0001,
       "step": 1496
     },
     {
-      "epoch": 2.2410179640718564,
-      "grad_norm": 0.12068999773483105,
-      "learning_rate": 3.1878379940709726e-06,
-      "loss": 0.0007,
+      "epoch": 2.3952,
+      "grad_norm": 0.0017896475593978372,
+      "learning_rate": 2.058583491552465e-06,
+      "loss": 0.0,
       "step": 1497
     },
     {
-      "epoch": 2.2425149700598803,
-      "grad_norm": 0.07153246649925311,
-      "learning_rate": 3.1760100310323815e-06,
-      "loss": 0.0013,
+      "epoch": 2.3968,
+      "grad_norm": 0.0011797322701871146,
+      "learning_rate": 2.048093436450603e-06,
+      "loss": 0.0,
       "step": 1498
     },
     {
-      "epoch": 2.2440119760479043,
-      "grad_norm": 0.02898441751748169,
-      "learning_rate": 3.1641999078943566e-06,
-      "loss": 0.0002,
+      "epoch": 2.3984,
+      "grad_norm": 0.07918396984152237,
+      "learning_rate": 2.037627126948751e-06,
+      "loss": 0.0006,
       "step": 1499
     },
     {
-      "epoch": 2.245508982035928,
-      "grad_norm": 0.11063959172977524,
-      "learning_rate": 3.1524076555320007e-06,
-      "loss": 0.0007,
+      "epoch": 2.4,
+      "grad_norm": 0.0041036247525871,
+      "learning_rate": 2.0271845943008984e-06,
+      "loss": 0.0001,
       "step": 1500
     },
     {
-      "epoch": 2.247005988023952,
-      "grad_norm": 0.10311560949120008,
-      "learning_rate": 3.1406333047737012e-06,
-      "loss": 0.0003,
+      "epoch": 2.4016,
+      "grad_norm": 0.009007184596335682,
+      "learning_rate": 2.0167658696900317e-06,
+      "loss": 0.0001,
       "step": 1501
     },
     {
-      "epoch": 2.248502994011976,
-      "grad_norm": 0.09788152701017722,
-      "learning_rate": 3.128876886401049e-06,
-      "loss": 0.0003,
+      "epoch": 2.4032,
+      "grad_norm": 0.01675758427096426,
+      "learning_rate": 2.006370984228043e-06,
+      "loss": 0.0001,
       "step": 1502
     },
     {
-      "epoch": 2.25,
-      "grad_norm": 0.15031384544339235,
-      "learning_rate": 3.1171384311487487e-06,
-      "loss": 0.0012,
+      "epoch": 2.4048,
+      "grad_norm": 0.005890697530268193,
+      "learning_rate": 1.9959999689556407e-06,
+      "loss": 0.0001,
       "step": 1503
     },
     {
-      "epoch": 2.251497005988024,
-      "grad_norm": 0.12304740690296884,
-      "learning_rate": 3.105417969704544e-06,
-      "loss": 0.0008,
+      "epoch": 2.4064,
+      "grad_norm": 0.002924527914154743,
+      "learning_rate": 1.985652854842247e-06,
+      "loss": 0.0,
       "step": 1504
     },
     {
-      "epoch": 2.252994011976048,
-      "grad_norm": 0.10203703250907652,
-      "learning_rate": 3.093715532709147e-06,
-      "loss": 0.0005,
+      "epoch": 2.408,
+      "grad_norm": 0.003731462856825938,
+      "learning_rate": 1.9753296727859195e-06,
+      "loss": 0.0,
       "step": 1505
     },
     {
-      "epoch": 2.254491017964072,
-      "grad_norm": 0.10315509055548146,
-      "learning_rate": 3.0820311507561395e-06,
-      "loss": 0.0007,
+      "epoch": 2.4096,
+      "grad_norm": 0.01630410409142845,
+      "learning_rate": 1.9650304536132426e-06,
+      "loss": 0.0001,
       "step": 1506
     },
     {
-      "epoch": 2.2559880239520957,
-      "grad_norm": 0.1340732142733853,
-      "learning_rate": 3.0703648543919053e-06,
-      "loss": 0.0006,
+      "epoch": 2.4112,
+      "grad_norm": 0.013784772819053917,
+      "learning_rate": 1.9547552280792528e-06,
+      "loss": 0.0001,
       "step": 1507
     },
     {
-      "epoch": 2.2574850299401197,
-      "grad_norm": 0.05868722622806031,
-      "learning_rate": 3.058716674115544e-06,
-      "loss": 0.0002,
+      "epoch": 2.4128,
+      "grad_norm": 0.0021612071817767383,
+      "learning_rate": 1.9445040268673297e-06,
+      "loss": 0.0,
       "step": 1508
     },
     {
-      "epoch": 2.2589820359281436,
-      "grad_norm": 0.03557542415759387,
-      "learning_rate": 3.047086640378798e-06,
-      "loss": 0.0002,
+      "epoch": 2.4144,
+      "grad_norm": 0.10756611202533951,
+      "learning_rate": 1.9342768805891176e-06,
+      "loss": 0.0003,
       "step": 1509
     },
     {
-      "epoch": 2.2604790419161676,
-      "grad_norm": 0.07662241536677104,
-      "learning_rate": 3.0354747835859686e-06,
-      "loss": 0.0004,
+      "epoch": 2.416,
+      "grad_norm": 0.003626199791092706,
+      "learning_rate": 1.924073819784428e-06,
+      "loss": 0.0001,
       "step": 1510
     },
     {
-      "epoch": 2.2619760479041915,
-      "grad_norm": 0.04174555140373921,
-      "learning_rate": 3.0238811340938347e-06,
-      "loss": 0.0004,
+      "epoch": 2.4176,
+      "grad_norm": 0.00467415340632512,
+      "learning_rate": 1.9138948749211473e-06,
+      "loss": 0.0001,
       "step": 1511
     },
     {
-      "epoch": 2.2634730538922154,
-      "grad_norm": 0.11334824278824039,
-      "learning_rate": 3.0123057222115835e-06,
-      "loss": 0.001,
+      "epoch": 2.4192,
+      "grad_norm": 0.002556602627541502,
+      "learning_rate": 1.9037400763951508e-06,
+      "loss": 0.0,
       "step": 1512
     },
     {
-      "epoch": 2.2649700598802394,
-      "grad_norm": 0.07557692002090421,
-      "learning_rate": 3.000748578200716e-06,
-      "loss": 0.0007,
+      "epoch": 2.4208,
+      "grad_norm": 0.021045982566475125,
+      "learning_rate": 1.8936094545302098e-06,
+      "loss": 0.0001,
       "step": 1513
     },
     {
-      "epoch": 2.2664670658682633,
-      "grad_norm": 0.07097266734420278,
-      "learning_rate": 2.989209732274979e-06,
-      "loss": 0.0003,
+      "epoch": 2.4224,
+      "grad_norm": 0.0016611376914392635,
+      "learning_rate": 1.8835030395778941e-06,
+      "loss": 0.0,
       "step": 1514
     },
     {
-      "epoch": 2.2679640718562872,
-      "grad_norm": 0.09527091457006696,
-      "learning_rate": 2.97768921460028e-06,
-      "loss": 0.0005,
+      "epoch": 2.424,
+      "grad_norm": 0.0017170133170774445,
+      "learning_rate": 1.8734208617174986e-06,
+      "loss": 0.0,
       "step": 1515
     },
     {
-      "epoch": 2.269461077844311,
-      "grad_norm": 0.0815127360640386,
-      "learning_rate": 2.9661870552946114e-06,
-      "loss": 0.0005,
+      "epoch": 2.4256,
+      "grad_norm": 0.0014993770241049835,
+      "learning_rate": 1.8633629510559315e-06,
+      "loss": 0.0,
       "step": 1516
     },
     {
-      "epoch": 2.270958083832335,
-      "grad_norm": 0.0802974444453441,
-      "learning_rate": 2.9547032844279824e-06,
-      "loss": 0.0005,
+      "epoch": 2.4272,
+      "grad_norm": 0.0015271567701904886,
+      "learning_rate": 1.8533293376276473e-06,
+      "loss": 0.0,
       "step": 1517
     },
     {
-      "epoch": 2.272455089820359,
-      "grad_norm": 0.04150750081228997,
-      "learning_rate": 2.943237932022307e-06,
-      "loss": 0.0002,
+      "epoch": 2.4288,
+      "grad_norm": 0.0013171104723947522,
+      "learning_rate": 1.8433200513945338e-06,
+      "loss": 0.0,
       "step": 1518
     },
     {
-      "epoch": 2.2739520958083834,
-      "grad_norm": 0.031089773582056647,
-      "learning_rate": 2.9317910280513705e-06,
-      "loss": 0.0002,
+      "epoch": 2.4304,
+      "grad_norm": 0.0015319013883092294,
+      "learning_rate": 1.8333351222458407e-06,
+      "loss": 0.0,
       "step": 1519
     },
     {
-      "epoch": 2.2754491017964074,
-      "grad_norm": 0.06361121560474334,
-      "learning_rate": 2.9203626024407185e-06,
-      "loss": 0.0004,
+      "epoch": 2.432,
+      "grad_norm": 0.07619118696699975,
+      "learning_rate": 1.8233745799980818e-06,
+      "loss": 0.0002,
       "step": 1520
     },
     {
-      "epoch": 2.2769461077844313,
-      "grad_norm": 0.05424769651819266,
-      "learning_rate": 2.9089526850675875e-06,
-      "loss": 0.0004,
+      "epoch": 2.4336,
+      "grad_norm": 0.0016675929742683841,
+      "learning_rate": 1.813438454394948e-06,
+      "loss": 0.0,
       "step": 1521
     },
     {
-      "epoch": 2.2784431137724552,
-      "grad_norm": 0.048908206558848324,
-      "learning_rate": 2.8975613057608297e-06,
-      "loss": 0.0008,
+      "epoch": 2.4352,
+      "grad_norm": 0.02036471538115778,
+      "learning_rate": 1.8035267751072172e-06,
+      "loss": 0.0001,
       "step": 1522
     },
     {
-      "epoch": 2.279940119760479,
-      "grad_norm": 0.12535647237158476,
-      "learning_rate": 2.8861884943008335e-06,
-      "loss": 0.0005,
+      "epoch": 2.4368,
+      "grad_norm": 0.0019359518892397129,
+      "learning_rate": 1.7936395717326705e-06,
+      "loss": 0.0,
       "step": 1523
     },
     {
-      "epoch": 2.281437125748503,
-      "grad_norm": 0.13595998922503674,
-      "learning_rate": 2.874834280419452e-06,
-      "loss": 0.0012,
+      "epoch": 2.4384,
+      "grad_norm": 0.018370237499706017,
+      "learning_rate": 1.7837768737959937e-06,
+      "loss": 0.0001,
       "step": 1524
     },
     {
-      "epoch": 2.282934131736527,
-      "grad_norm": 0.0193676454435138,
-      "learning_rate": 2.863498693799901e-06,
-      "loss": 0.0002,
+      "epoch": 2.44,
+      "grad_norm": 0.0035952528928215655,
+      "learning_rate": 1.773938710748706e-06,
+      "loss": 0.0001,
       "step": 1525
     },
     {
-      "epoch": 2.284431137724551,
-      "grad_norm": 0.09642186052650302,
-      "learning_rate": 2.8521817640767225e-06,
-      "loss": 0.0004,
+      "epoch": 2.4416,
+      "grad_norm": 0.02360457405560866,
+      "learning_rate": 1.7641251119690505e-06,
+      "loss": 0.0001,
       "step": 1526
     },
     {
-      "epoch": 2.285928143712575,
-      "grad_norm": 0.08690858061005316,
-      "learning_rate": 2.8408835208356667e-06,
-      "loss": 0.0006,
+      "epoch": 2.4432,
+      "grad_norm": 0.002167045765832743,
+      "learning_rate": 1.7543361067619269e-06,
+      "loss": 0.0001,
       "step": 1527
     },
     {
-      "epoch": 2.287425149700599,
-      "grad_norm": 0.09629388756889663,
-      "learning_rate": 2.8296039936136355e-06,
-      "loss": 0.0005,
+      "epoch": 2.4448,
+      "grad_norm": 0.027689086747607245,
+      "learning_rate": 1.7445717243587889e-06,
+      "loss": 0.0001,
       "step": 1528
     },
     {
-      "epoch": 2.288922155688623,
-      "grad_norm": 0.02465511736806249,
-      "learning_rate": 2.818343211898612e-06,
-      "loss": 0.0002,
+      "epoch": 2.4464,
+      "grad_norm": 0.1787294579533905,
+      "learning_rate": 1.734831993917564e-06,
+      "loss": 0.0006,
       "step": 1529
     },
     {
-      "epoch": 2.2904191616766467,
-      "grad_norm": 0.09994990956621684,
-      "learning_rate": 2.807101205129553e-06,
-      "loss": 0.0009,
+      "epoch": 2.448,
+      "grad_norm": 0.09937824016232794,
+      "learning_rate": 1.7251169445225658e-06,
+      "loss": 0.0017,
       "step": 1530
     },
     {
-      "epoch": 2.2919161676646707,
-      "grad_norm": 0.08966911947040915,
-      "learning_rate": 2.795878002696354e-06,
-      "loss": 0.0004,
+      "epoch": 2.4496,
+      "grad_norm": 0.0046358430248506985,
+      "learning_rate": 1.715426605184407e-06,
+      "loss": 0.0001,
       "step": 1531
     },
     {
-      "epoch": 2.2934131736526946,
-      "grad_norm": 0.008082694473274558,
-      "learning_rate": 2.7846736339397362e-06,
-      "loss": 0.0001,
+      "epoch": 2.4512,
+      "grad_norm": 0.002018934016152693,
+      "learning_rate": 1.705761004839911e-06,
+      "loss": 0.0,
       "step": 1532
     },
     {
-      "epoch": 2.2949101796407185,
-      "grad_norm": 0.037330647180128684,
-      "learning_rate": 2.773488128151186e-06,
-      "loss": 0.0002,
+      "epoch": 2.4528,
+      "grad_norm": 0.0009162379025401923,
+      "learning_rate": 1.6961201723520248e-06,
+      "loss": 0.0,
       "step": 1533
     },
     {
-      "epoch": 2.2964071856287425,
-      "grad_norm": 0.15808722242025455,
-      "learning_rate": 2.7623215145728866e-06,
-      "loss": 0.001,
+      "epoch": 2.4544,
+      "grad_norm": 0.03424925237246253,
+      "learning_rate": 1.6865041365097434e-06,
+      "loss": 0.0002,
       "step": 1534
     },
     {
-      "epoch": 2.2979041916167664,
-      "grad_norm": 0.10096871768612158,
-      "learning_rate": 2.7511738223976126e-06,
-      "loss": 0.0009,
+      "epoch": 2.456,
+      "grad_norm": 0.01168682482158326,
+      "learning_rate": 1.676912926028007e-06,
+      "loss": 0.0001,
       "step": 1535
     },
     {
-      "epoch": 2.2994011976047903,
-      "grad_norm": 0.06280335460620405,
-      "learning_rate": 2.740045080768694e-06,
-      "loss": 0.0003,
+      "epoch": 2.4576000000000002,
+      "grad_norm": 0.007125610436890888,
+      "learning_rate": 1.6673465695476233e-06,
+      "loss": 0.0001,
       "step": 1536
     },
     {
-      "epoch": 2.3008982035928143,
-      "grad_norm": 0.07328063719390991,
-      "learning_rate": 2.7289353187798982e-06,
-      "loss": 0.0004,
+      "epoch": 2.4592,
+      "grad_norm": 0.0023125107062913182,
+      "learning_rate": 1.6578050956351887e-06,
+      "loss": 0.0,
       "step": 1537
     },
     {
-      "epoch": 2.302395209580838,
-      "grad_norm": 0.07595297277452129,
-      "learning_rate": 2.7178445654753916e-06,
-      "loss": 0.0005,
+      "epoch": 2.4608,
+      "grad_norm": 0.018211054734330913,
+      "learning_rate": 1.6482885327829912e-06,
+      "loss": 0.0001,
       "step": 1538
     },
     {
-      "epoch": 2.303892215568862,
-      "grad_norm": 0.11974836266924484,
-      "learning_rate": 2.706772849849636e-06,
-      "loss": 0.0009,
+      "epoch": 2.4624,
+      "grad_norm": 0.0038736004464934217,
+      "learning_rate": 1.6387969094089318e-06,
+      "loss": 0.0001,
       "step": 1539
     },
     {
-      "epoch": 2.305389221556886,
-      "grad_norm": 0.0076400019711527585,
-      "learning_rate": 2.6957202008473238e-06,
+      "epoch": 2.464,
+      "grad_norm": 0.006647792105609032,
+      "learning_rate": 1.6293302538564381e-06,
       "loss": 0.0001,
       "step": 1540
     },
     {
-      "epoch": 2.30688622754491,
-      "grad_norm": 0.09483349610761937,
-      "learning_rate": 2.6846866473633126e-06,
-      "loss": 0.0005,
+      "epoch": 2.4656000000000002,
+      "grad_norm": 0.009903476509994772,
+      "learning_rate": 1.619888594394382e-06,
+      "loss": 0.0001,
       "step": 1541
     },
     {
-      "epoch": 2.308383233532934,
-      "grad_norm": 0.011742806672084745,
-      "learning_rate": 2.673672218242518e-06,
+      "epoch": 2.4672,
+      "grad_norm": 0.020716801874626822,
+      "learning_rate": 1.6104719592169905e-06,
       "loss": 0.0001,
       "step": 1542
     },
     {
-      "epoch": 2.309880239520958,
-      "grad_norm": 0.027406703176830523,
-      "learning_rate": 2.662676942279879e-06,
-      "loss": 0.0002,
+      "epoch": 2.4688,
+      "grad_norm": 0.0017791585271250928,
+      "learning_rate": 1.6010803764437633e-06,
+      "loss": 0.0,
       "step": 1543
     },
     {
-      "epoch": 2.3113772455089823,
-      "grad_norm": 0.04835437820138478,
-      "learning_rate": 2.651700848220253e-06,
-      "loss": 0.0003,
+      "epoch": 2.4704,
+      "grad_norm": 0.04399381417400462,
+      "learning_rate": 1.5917138741193972e-06,
+      "loss": 0.0002,
       "step": 1544
     },
     {
-      "epoch": 2.312874251497006,
-      "grad_norm": 0.06402404819188177,
-      "learning_rate": 2.640743964758349e-06,
-      "loss": 0.0003,
+      "epoch": 2.472,
+      "grad_norm": 0.0017668274917530136,
+      "learning_rate": 1.5823724802136863e-06,
+      "loss": 0.0,
       "step": 1545
     },
     {
-      "epoch": 2.31437125748503,
-      "grad_norm": 0.06160368088836378,
-      "learning_rate": 2.629806320538665e-06,
-      "loss": 0.0002,
+      "epoch": 2.4736000000000002,
+      "grad_norm": 0.008234676324611317,
+      "learning_rate": 1.5730562226214529e-06,
+      "loss": 0.0001,
       "step": 1546
     },
     {
-      "epoch": 2.315868263473054,
-      "grad_norm": 0.08575582684771083,
-      "learning_rate": 2.6188879441553818e-06,
-      "loss": 0.0005,
+      "epoch": 2.4752,
+      "grad_norm": 0.01419089057865367,
+      "learning_rate": 1.5637651291624522e-06,
+      "loss": 0.0001,
       "step": 1547
     },
     {
-      "epoch": 2.317365269461078,
-      "grad_norm": 0.06630142016737109,
-      "learning_rate": 2.60798886415233e-06,
-      "loss": 0.0004,
+      "epoch": 2.4768,
+      "grad_norm": 0.07111358162237315,
+      "learning_rate": 1.5544992275813053e-06,
+      "loss": 0.0003,
       "step": 1548
     },
     {
-      "epoch": 2.318862275449102,
-      "grad_norm": 0.12139207820757994,
-      "learning_rate": 2.5971091090228817e-06,
-      "loss": 0.0018,
+      "epoch": 2.4784,
+      "grad_norm": 0.0027156424970821223,
+      "learning_rate": 1.545258545547398e-06,
+      "loss": 0.0,
       "step": 1549
     },
     {
-      "epoch": 2.320359281437126,
-      "grad_norm": 0.06552332582410145,
-      "learning_rate": 2.5862487072098898e-06,
-      "loss": 0.0005,
+      "epoch": 2.48,
+      "grad_norm": 0.0024029200917369113,
+      "learning_rate": 1.536043110654809e-06,
+      "loss": 0.0,
       "step": 1550
     },
     {
-      "epoch": 2.32185628742515,
-      "grad_norm": 0.04409020071690396,
-      "learning_rate": 2.5754076871056144e-06,
-      "loss": 0.0002,
+      "epoch": 2.4816,
+      "grad_norm": 0.0010976426679610582,
+      "learning_rate": 1.5268529504222262e-06,
+      "loss": 0.0,
       "step": 1551
     },
     {
-      "epoch": 2.3233532934131738,
-      "grad_norm": 0.10364683798236486,
-      "learning_rate": 2.564586077051641e-06,
-      "loss": 0.0003,
+      "epoch": 2.4832,
+      "grad_norm": 0.05562937434060917,
+      "learning_rate": 1.5176880922928615e-06,
+      "loss": 0.0002,
       "step": 1552
     },
     {
-      "epoch": 2.3248502994011977,
-      "grad_norm": 0.155282637159344,
-      "learning_rate": 2.5537839053388225e-06,
-      "loss": 0.0009,
+      "epoch": 2.4848,
+      "grad_norm": 0.001756978753692583,
+      "learning_rate": 1.5085485636343755e-06,
+      "loss": 0.0,
       "step": 1553
     },
     {
-      "epoch": 2.3263473053892216,
-      "grad_norm": 0.06569277881888244,
-      "learning_rate": 2.543001200207178e-06,
-      "loss": 0.0003,
+      "epoch": 2.4864,
+      "grad_norm": 0.007200012050074666,
+      "learning_rate": 1.4994343917387854e-06,
+      "loss": 0.0001,
       "step": 1554
     },
     {
-      "epoch": 2.3278443113772456,
-      "grad_norm": 0.07710485032834485,
-      "learning_rate": 2.5322379898458527e-06,
-      "loss": 0.0005,
+      "epoch": 2.488,
+      "grad_norm": 0.00545621490829972,
+      "learning_rate": 1.4903456038223941e-06,
+      "loss": 0.0001,
       "step": 1555
     },
     {
-      "epoch": 2.3293413173652695,
-      "grad_norm": 0.07966793603491328,
-      "learning_rate": 2.5214943023930138e-06,
-      "loss": 0.0005,
+      "epoch": 2.4896,
+      "grad_norm": 0.0013280783981415845,
+      "learning_rate": 1.481282227025701e-06,
+      "loss": 0.0,
       "step": 1556
     },
     {
-      "epoch": 2.3308383233532934,
-      "grad_norm": 0.049815482522710106,
-      "learning_rate": 2.5107701659357963e-06,
-      "loss": 0.0002,
+      "epoch": 2.4912,
+      "grad_norm": 0.005087134285962172,
+      "learning_rate": 1.4722442884133214e-06,
+      "loss": 0.0001,
       "step": 1557
     },
     {
-      "epoch": 2.3323353293413174,
-      "grad_norm": 0.11800603839640363,
-      "learning_rate": 2.5000656085102215e-06,
-      "loss": 0.0006,
+      "epoch": 2.4928,
+      "grad_norm": 0.12036336528820847,
+      "learning_rate": 1.4632318149739177e-06,
+      "loss": 0.0021,
       "step": 1558
     },
     {
-      "epoch": 2.3338323353293413,
-      "grad_norm": 0.01937251400307602,
-      "learning_rate": 2.4893806581011215e-06,
-      "loss": 0.0002,
+      "epoch": 2.4944,
+      "grad_norm": 0.016016100224801787,
+      "learning_rate": 1.4542448336201021e-06,
+      "loss": 0.0001,
       "step": 1559
     },
     {
-      "epoch": 2.3353293413173652,
-      "grad_norm": 0.05368059532814563,
-      "learning_rate": 2.4787153426420794e-06,
-      "loss": 0.0003,
+      "epoch": 2.496,
+      "grad_norm": 0.01170453067098676,
+      "learning_rate": 1.4452833711883629e-06,
+      "loss": 0.0001,
       "step": 1560
     },
     {
-      "epoch": 2.336826347305389,
-      "grad_norm": 0.14182083918266286,
-      "learning_rate": 2.46806969001534e-06,
-      "loss": 0.0009,
+      "epoch": 2.4976,
+      "grad_norm": 0.017877091503009403,
+      "learning_rate": 1.4363474544389876e-06,
+      "loss": 0.0001,
       "step": 1561
     },
     {
-      "epoch": 2.338323353293413,
-      "grad_norm": 0.026485749870856008,
-      "learning_rate": 2.4574437280517437e-06,
-      "loss": 0.0003,
+      "epoch": 2.4992,
+      "grad_norm": 0.0023130446241290763,
+      "learning_rate": 1.4274371100559792e-06,
+      "loss": 0.0,
       "step": 1562
     },
     {
-      "epoch": 2.339820359281437,
-      "grad_norm": 0.08886356114902899,
-      "learning_rate": 2.446837484530655e-06,
-      "loss": 0.0005,
+      "epoch": 2.5008,
+      "grad_norm": 0.10763984336472367,
+      "learning_rate": 1.4185523646469822e-06,
+      "loss": 0.0003,
       "step": 1563
     },
     {
-      "epoch": 2.341317365269461,
-      "grad_norm": 0.02014395320911489,
-      "learning_rate": 2.436250987179889e-06,
-      "loss": 0.0001,
+      "epoch": 2.5023999999999997,
+      "grad_norm": 0.0013005506478659377,
+      "learning_rate": 1.409693244743192e-06,
+      "loss": 0.0,
       "step": 1564
     },
     {
-      "epoch": 2.342814371257485,
-      "grad_norm": 0.021864827664785275,
-      "learning_rate": 2.4256842636756372e-06,
-      "loss": 0.0002,
+      "epoch": 2.504,
+      "grad_norm": 0.020425218709959836,
+      "learning_rate": 1.4008597767992872e-06,
+      "loss": 0.0001,
       "step": 1565
     },
     {
-      "epoch": 2.344311377245509,
-      "grad_norm": 0.02967983647931681,
-      "learning_rate": 2.415137341642397e-06,
-      "loss": 0.0001,
+      "epoch": 2.5056000000000003,
+      "grad_norm": 0.08103328060955477,
+      "learning_rate": 1.3920519871933425e-06,
+      "loss": 0.0005,
       "step": 1566
     },
     {
-      "epoch": 2.345808383233533,
-      "grad_norm": 0.06117231385528192,
-      "learning_rate": 2.4046102486529033e-06,
-      "loss": 0.0004,
+      "epoch": 2.5072,
+      "grad_norm": 0.00386152032655592,
+      "learning_rate": 1.3832699022267516e-06,
+      "loss": 0.0,
       "step": 1567
     },
     {
-      "epoch": 2.3473053892215567,
-      "grad_norm": 0.019912926187557994,
-      "learning_rate": 2.394103012228045e-06,
-      "loss": 0.0002,
+      "epoch": 2.5088,
+      "grad_norm": 0.0034205593845012317,
+      "learning_rate": 1.3745135481241602e-06,
+      "loss": 0.0001,
       "step": 1568
     },
     {
-      "epoch": 2.3488023952095807,
-      "grad_norm": 0.024535582706188355,
-      "learning_rate": 2.3836156598368055e-06,
-      "loss": 0.0003,
+      "epoch": 2.5103999999999997,
+      "grad_norm": 0.01165295536569006,
+      "learning_rate": 1.3657829510333653e-06,
+      "loss": 0.0001,
       "step": 1569
     },
     {
-      "epoch": 2.3502994011976046,
-      "grad_norm": 0.03376993608094255,
-      "learning_rate": 2.373148218896182e-06,
-      "loss": 0.0003,
+      "epoch": 2.512,
+      "grad_norm": 0.010197076721640802,
+      "learning_rate": 1.3570781370252584e-06,
+      "loss": 0.0001,
       "step": 1570
     },
     {
-      "epoch": 2.3517964071856285,
-      "grad_norm": 0.0384213575217128,
-      "learning_rate": 2.362700716771118e-06,
-      "loss": 0.0005,
+      "epoch": 2.5136,
+      "grad_norm": 0.030098303958029184,
+      "learning_rate": 1.3483991320937307e-06,
+      "loss": 0.0002,
       "step": 1571
     },
     {
-      "epoch": 2.3532934131736525,
-      "grad_norm": 0.046833073744106835,
-      "learning_rate": 2.3522731807744317e-06,
-      "loss": 0.0005,
+      "epoch": 2.5152,
+      "grad_norm": 0.04392294653380117,
+      "learning_rate": 1.339745962155613e-06,
+      "loss": 0.0002,
       "step": 1572
     },
     {
-      "epoch": 2.3547904191616764,
-      "grad_norm": 0.015881510742633775,
-      "learning_rate": 2.341865638166747e-06,
-      "loss": 0.0001,
+      "epoch": 2.5168,
+      "grad_norm": 0.0021643913012770366,
+      "learning_rate": 1.3311186530505838e-06,
+      "loss": 0.0,
       "step": 1573
     },
     {
-      "epoch": 2.3562874251497004,
-      "grad_norm": 0.23245452164430105,
-      "learning_rate": 2.331478116156416e-06,
-      "loss": 0.0025,
+      "epoch": 2.5183999999999997,
+      "grad_norm": 0.002235451729477788,
+      "learning_rate": 1.322517230541096e-06,
+      "loss": 0.0,
       "step": 1574
     },
     {
-      "epoch": 2.3577844311377247,
-      "grad_norm": 0.04796842296752511,
-      "learning_rate": 2.3211106418994512e-06,
-      "loss": 0.0002,
+      "epoch": 2.52,
+      "grad_norm": 0.0018122429134230715,
+      "learning_rate": 1.313941720312303e-06,
+      "loss": 0.0,
       "step": 1575
     },
     {
-      "epoch": 2.3592814371257487,
-      "grad_norm": 0.11392590399503923,
-      "learning_rate": 2.3107632424994532e-06,
-      "loss": 0.0015,
+      "epoch": 2.5216,
+      "grad_norm": 0.017416784512229562,
+      "learning_rate": 1.30539214797198e-06,
+      "loss": 0.0001,
       "step": 1576
     },
     {
-      "epoch": 2.3607784431137726,
-      "grad_norm": 0.01691641893665238,
-      "learning_rate": 2.300435945007541e-06,
+      "epoch": 2.5232,
+      "grad_norm": 0.0067630498958140784,
+      "learning_rate": 1.2968685390504465e-06,
       "loss": 0.0001,
       "step": 1577
     },
     {
-      "epoch": 2.3622754491017965,
-      "grad_norm": 0.02404128261366343,
-      "learning_rate": 2.290128776422292e-06,
-      "loss": 0.0002,
+      "epoch": 2.5248,
+      "grad_norm": 0.006677237805470845,
+      "learning_rate": 1.2883709190004956e-06,
+      "loss": 0.0001,
       "step": 1578
     },
     {
-      "epoch": 2.3637724550898205,
-      "grad_norm": 0.06174460774186989,
-      "learning_rate": 2.2798417636896396e-06,
-      "loss": 0.0012,
+      "epoch": 2.5263999999999998,
+      "grad_norm": 0.0036461037673314223,
+      "learning_rate": 1.2798993131973093e-06,
+      "loss": 0.0,
       "step": 1579
     },
     {
-      "epoch": 2.3652694610778444,
-      "grad_norm": 0.08570446822273553,
-      "learning_rate": 2.269574933702846e-06,
-      "loss": 0.0004,
+      "epoch": 2.528,
+      "grad_norm": 0.0016567673129543978,
+      "learning_rate": 1.2714537469383858e-06,
+      "loss": 0.0,
       "step": 1580
     },
     {
-      "epoch": 2.3667664670658684,
-      "grad_norm": 0.0856077332705348,
-      "learning_rate": 2.259328313302395e-06,
-      "loss": 0.0005,
+      "epoch": 2.5296,
+      "grad_norm": 0.024221766064813827,
+      "learning_rate": 1.263034245443473e-06,
+      "loss": 0.0002,
       "step": 1581
     },
     {
-      "epoch": 2.3682634730538923,
-      "grad_norm": 0.07142666648166325,
-      "learning_rate": 2.249101929275944e-06,
-      "loss": 0.0009,
+      "epoch": 2.5312,
+      "grad_norm": 0.004007303578627002,
+      "learning_rate": 1.254640833854477e-06,
+      "loss": 0.0001,
       "step": 1582
     },
     {
-      "epoch": 2.3697604790419162,
-      "grad_norm": 0.025286357606488962,
-      "learning_rate": 2.2388958083582423e-06,
-      "loss": 0.0002,
+      "epoch": 2.5328,
+      "grad_norm": 0.003292892988464015,
+      "learning_rate": 1.2462735372353996e-06,
+      "loss": 0.0,
       "step": 1583
     },
     {
-      "epoch": 2.37125748502994,
-      "grad_norm": 0.045952040023591774,
-      "learning_rate": 2.228709977231065e-06,
+      "epoch": 2.5343999999999998,
+      "grad_norm": 0.06401474580874286,
+      "learning_rate": 1.2379323805722575e-06,
       "loss": 0.0003,
       "step": 1584
     },
     {
-      "epoch": 2.372754491017964,
-      "grad_norm": 0.10954972663341103,
-      "learning_rate": 2.2185444625231557e-06,
-      "loss": 0.0007,
+      "epoch": 2.536,
+      "grad_norm": 0.02788742726917249,
+      "learning_rate": 1.2296173887730122e-06,
+      "loss": 0.0001,
       "step": 1585
     },
     {
-      "epoch": 2.374251497005988,
-      "grad_norm": 0.10645468456591504,
-      "learning_rate": 2.2083992908101236e-06,
-      "loss": 0.001,
+      "epoch": 2.5376,
+      "grad_norm": 0.0015182760956827556,
+      "learning_rate": 1.2213285866674908e-06,
+      "loss": 0.0,
       "step": 1586
     },
     {
-      "epoch": 2.375748502994012,
-      "grad_norm": 0.030096418586315985,
-      "learning_rate": 2.1982744886144157e-06,
-      "loss": 0.0004,
+      "epoch": 2.5392,
+      "grad_norm": 0.0228623292459383,
+      "learning_rate": 1.2130659990073146e-06,
+      "loss": 0.0001,
       "step": 1587
     },
     {
-      "epoch": 2.377245508982036,
-      "grad_norm": 0.09053718049959521,
-      "learning_rate": 2.1881700824052157e-06,
-      "loss": 0.0004,
+      "epoch": 2.5408,
+      "grad_norm": 0.0011974631021759586,
+      "learning_rate": 1.2048296504658208e-06,
+      "loss": 0.0,
       "step": 1588
     },
     {
-      "epoch": 2.37874251497006,
-      "grad_norm": 0.04106460054597611,
-      "learning_rate": 2.1780860985983853e-06,
-      "loss": 0.0002,
+      "epoch": 2.5423999999999998,
+      "grad_norm": 0.002917571446504748,
+      "learning_rate": 1.196619565638003e-06,
+      "loss": 0.0001,
       "step": 1589
     },
     {
-      "epoch": 2.3802395209580838,
-      "grad_norm": 0.017813026852897713,
-      "learning_rate": 2.1680225635564113e-06,
-      "loss": 0.0002,
+      "epoch": 2.544,
+      "grad_norm": 0.0021043187714803947,
+      "learning_rate": 1.1884357690404157e-06,
+      "loss": 0.0,
       "step": 1590
     },
     {
-      "epoch": 2.3817365269461077,
-      "grad_norm": 0.03383373318851492,
-      "learning_rate": 2.1579795035882955e-06,
-      "loss": 0.0003,
+      "epoch": 2.5456,
+      "grad_norm": 0.005396413474978706,
+      "learning_rate": 1.1802782851111206e-06,
+      "loss": 0.0001,
       "step": 1591
     },
     {
-      "epoch": 2.3832335329341316,
-      "grad_norm": 0.05074488961782638,
-      "learning_rate": 2.1479569449495406e-06,
-      "loss": 0.0004,
+      "epoch": 2.5472,
+      "grad_norm": 0.021121082445202475,
+      "learning_rate": 1.1721471382096028e-06,
+      "loss": 0.0001,
       "step": 1592
     },
     {
-      "epoch": 2.3847305389221556,
-      "grad_norm": 0.01182856111698649,
-      "learning_rate": 2.1379549138420265e-06,
-      "loss": 0.0001,
+      "epoch": 2.5488,
+      "grad_norm": 0.0020481328118915473,
+      "learning_rate": 1.1640423526166987e-06,
+      "loss": 0.0,
       "step": 1593
     },
     {
-      "epoch": 2.3862275449101795,
-      "grad_norm": 0.26118958396749364,
-      "learning_rate": 2.127973436413989e-06,
-      "loss": 0.0018,
+      "epoch": 2.5504,
+      "grad_norm": 0.0013258107292440985,
+      "learning_rate": 1.1559639525345313e-06,
+      "loss": 0.0,
       "step": 1594
     },
     {
-      "epoch": 2.3877245508982035,
-      "grad_norm": 0.05278864289511616,
-      "learning_rate": 2.1180125387599193e-06,
-      "loss": 0.0003,
+      "epoch": 2.552,
+      "grad_norm": 0.09574298006311367,
+      "learning_rate": 1.1479119620864277e-06,
+      "loss": 0.0008,
       "step": 1595
     },
     {
-      "epoch": 2.3892215568862274,
-      "grad_norm": 0.030355323786991278,
-      "learning_rate": 2.108072246920506e-06,
-      "loss": 0.0002,
+      "epoch": 2.5536,
+      "grad_norm": 0.004026929915154443,
+      "learning_rate": 1.1398864053168534e-06,
+      "loss": 0.0001,
       "step": 1596
     },
     {
-      "epoch": 2.3907185628742513,
-      "grad_norm": 0.1099369494009969,
-      "learning_rate": 2.0981525868825815e-06,
-      "loss": 0.0006,
+      "epoch": 2.5552,
+      "grad_norm": 0.03389784693920091,
+      "learning_rate": 1.1318873061913405e-06,
+      "loss": 0.0002,
       "step": 1597
     },
     {
-      "epoch": 2.3922155688622753,
-      "grad_norm": 0.1203596546841281,
-      "learning_rate": 2.0882535845790188e-06,
-      "loss": 0.0007,
+      "epoch": 2.5568,
+      "grad_norm": 0.00514572526232494,
+      "learning_rate": 1.123914688596409e-06,
+      "loss": 0.0001,
       "step": 1598
     },
     {
-      "epoch": 2.3937125748502996,
-      "grad_norm": 0.1093149086475125,
-      "learning_rate": 2.078375265888707e-06,
-      "loss": 0.0011,
+      "epoch": 2.5584,
+      "grad_norm": 0.016099931278681145,
+      "learning_rate": 1.1159685763395113e-06,
+      "loss": 0.0001,
       "step": 1599
     },
     {
-      "epoch": 2.3952095808383236,
-      "grad_norm": 0.10583209779661093,
-      "learning_rate": 2.0685176566364507e-06,
-      "loss": 0.0006,
+      "epoch": 2.56,
+      "grad_norm": 0.034261385652090005,
+      "learning_rate": 1.108048993148939e-06,
+      "loss": 0.0002,
       "step": 1600
     },
     {
-      "epoch": 2.3967065868263475,
-      "grad_norm": 0.047336815910952124,
-      "learning_rate": 2.0586807825929123e-06,
-      "loss": 0.0002,
+      "epoch": 2.5616,
+      "grad_norm": 0.006428809226415149,
+      "learning_rate": 1.1001559626737757e-06,
+      "loss": 0.0001,
       "step": 1601
     },
     {
-      "epoch": 2.3982035928143715,
-      "grad_norm": 0.09904610970762708,
-      "learning_rate": 2.0488646694745583e-06,
-      "loss": 0.0012,
+      "epoch": 2.5632,
+      "grad_norm": 0.005330511861859042,
+      "learning_rate": 1.0922895084838036e-06,
+      "loss": 0.0001,
       "step": 1602
     },
     {
-      "epoch": 2.3997005988023954,
-      "grad_norm": 0.09513445282250857,
-      "learning_rate": 2.0390693429435626e-06,
-      "loss": 0.0004,
+      "epoch": 2.5648,
+      "grad_norm": 0.03328716363011227,
+      "learning_rate": 1.0844496540694515e-06,
+      "loss": 0.0001,
       "step": 1603
     },
     {
-      "epoch": 2.4011976047904193,
-      "grad_norm": 0.020422046778283895,
-      "learning_rate": 2.029294828607773e-06,
-      "loss": 0.0001,
+      "epoch": 2.5664,
+      "grad_norm": 0.002380265262180709,
+      "learning_rate": 1.0766364228417148e-06,
+      "loss": 0.0,
       "step": 1604
     },
     {
-      "epoch": 2.4026946107784433,
-      "grad_norm": 0.14061343688126954,
-      "learning_rate": 2.01954115202062e-06,
-      "loss": 0.0012,
+      "epoch": 2.568,
+      "grad_norm": 0.02310696101922282,
+      "learning_rate": 1.0688498381320855e-06,
+      "loss": 0.0001,
       "step": 1605
     },
     {
-      "epoch": 2.404191616766467,
-      "grad_norm": 0.04251232089992607,
-      "learning_rate": 2.0098083386810596e-06,
-      "loss": 0.0002,
+      "epoch": 2.5696,
+      "grad_norm": 0.03340335879563055,
+      "learning_rate": 1.0610899231924887e-06,
+      "loss": 0.0001,
       "step": 1606
     },
     {
-      "epoch": 2.405688622754491,
-      "grad_norm": 0.05855588269503358,
-      "learning_rate": 2.0000964140335044e-06,
-      "loss": 0.0004,
+      "epoch": 2.5712,
+      "grad_norm": 0.10027517187003131,
+      "learning_rate": 1.0533567011952094e-06,
+      "loss": 0.0012,
       "step": 1607
     },
     {
-      "epoch": 2.407185628742515,
-      "grad_norm": 0.04678849099056471,
-      "learning_rate": 1.990405403467757e-06,
-      "loss": 0.0002,
+      "epoch": 2.5728,
+      "grad_norm": 0.004624394220847377,
+      "learning_rate": 1.0456501952328191e-06,
+      "loss": 0.0001,
       "step": 1608
     },
     {
-      "epoch": 2.408682634730539,
-      "grad_norm": 0.15089209688583194,
-      "learning_rate": 1.980735332318954e-06,
-      "loss": 0.0005,
+      "epoch": 2.5744,
+      "grad_norm": 0.06351784277481294,
+      "learning_rate": 1.037970428318118e-06,
+      "loss": 0.0003,
       "step": 1609
     },
     {
-      "epoch": 2.410179640718563,
-      "grad_norm": 0.04533706932582074,
-      "learning_rate": 1.9710862258674714e-06,
-      "loss": 0.0002,
+      "epoch": 2.576,
+      "grad_norm": 0.0030415667014872553,
+      "learning_rate": 1.0303174233840529e-06,
+      "loss": 0.0,
       "step": 1610
     },
     {
-      "epoch": 2.411676646706587,
-      "grad_norm": 0.05696073871947218,
-      "learning_rate": 1.9614581093388964e-06,
-      "loss": 0.0004,
+      "epoch": 2.5776,
+      "grad_norm": 0.0023718507402067097,
+      "learning_rate": 1.022691203283661e-06,
+      "loss": 0.0,
       "step": 1611
     },
     {
-      "epoch": 2.413173652694611,
-      "grad_norm": 0.044373523736829394,
-      "learning_rate": 1.951851007903931e-06,
+      "epoch": 2.5792,
+      "grad_norm": 0.09427354766258768,
+      "learning_rate": 1.0150917907899926e-06,
       "loss": 0.0003,
       "step": 1612
     },
     {
-      "epoch": 2.4146706586826348,
-      "grad_norm": 0.1647299050829959,
-      "learning_rate": 1.9422649466783427e-06,
-      "loss": 0.0009,
+      "epoch": 2.5808,
+      "grad_norm": 0.05245903646176129,
+      "learning_rate": 1.0075192085960451e-06,
+      "loss": 0.0003,
       "step": 1613
     },
     {
-      "epoch": 2.4161676646706587,
-      "grad_norm": 0.08092462607166259,
-      "learning_rate": 1.9326999507228916e-06,
-      "loss": 0.0004,
+      "epoch": 2.5824,
+      "grad_norm": 0.014390080017914319,
+      "learning_rate": 9.999734793146998e-07,
+      "loss": 0.0001,
       "step": 1614
     },
     {
-      "epoch": 2.4176646706586826,
-      "grad_norm": 0.08726562462534888,
-      "learning_rate": 1.923156045043265e-06,
-      "loss": 0.0007,
+      "epoch": 2.584,
+      "grad_norm": 0.0019412655789413306,
+      "learning_rate": 9.924546254786493e-07,
+      "loss": 0.0,
       "step": 1615
     },
     {
-      "epoch": 2.4191616766467066,
-      "grad_norm": 0.0684185106844218,
-      "learning_rate": 1.9136332545900215e-06,
-      "loss": 0.0013,
+      "epoch": 2.5856,
+      "grad_norm": 0.0034950463726222372,
+      "learning_rate": 9.849626695403326e-07,
+      "loss": 0.0,
       "step": 1616
     },
     {
-      "epoch": 2.4206586826347305,
-      "grad_norm": 0.028069249948623098,
-      "learning_rate": 1.9041316042585113e-06,
-      "loss": 0.0002,
+      "epoch": 2.5872,
+      "grad_norm": 0.015087065616657028,
+      "learning_rate": 9.77497633871868e-07,
+      "loss": 0.0001,
       "step": 1617
     },
     {
-      "epoch": 2.4221556886227544,
-      "grad_norm": 0.024791281262007244,
-      "learning_rate": 1.8946511188888194e-06,
+      "epoch": 2.5888,
+      "grad_norm": 0.08578051961216712,
+      "learning_rate": 9.700595407649805e-07,
       "loss": 0.0002,
       "step": 1618
     },
     {
-      "epoch": 2.4236526946107784,
-      "grad_norm": 0.07988190689532107,
-      "learning_rate": 1.8851918232657007e-06,
-      "loss": 0.0005,
+      "epoch": 2.5904,
+      "grad_norm": 0.006834638154714175,
+      "learning_rate": 9.62648412430951e-07,
+      "loss": 0.0001,
       "step": 1619
     },
     {
-      "epoch": 2.4251497005988023,
-      "grad_norm": 0.027087754543380946,
-      "learning_rate": 1.8757537421185145e-06,
-      "loss": 0.0002,
+      "epoch": 2.592,
+      "grad_norm": 0.027796454328559976,
+      "learning_rate": 9.5526427100053e-07,
+      "loss": 0.0001,
       "step": 1620
     },
     {
-      "epoch": 2.4266467065868262,
-      "grad_norm": 0.030964818451742402,
-      "learning_rate": 1.8663369001211552e-06,
-      "loss": 0.0002,
+      "epoch": 2.5936,
+      "grad_norm": 0.03503244641418847,
+      "learning_rate": 9.479071385238892e-07,
+      "loss": 0.0001,
       "step": 1621
     },
     {
-      "epoch": 2.42814371257485,
-      "grad_norm": 0.042357471381083395,
-      "learning_rate": 1.8569413218919986e-06,
-      "loss": 0.0002,
+      "epoch": 2.5952,
+      "grad_norm": 0.01657949893006547,
+      "learning_rate": 9.40577036970538e-07,
+      "loss": 0.0001,
       "step": 1622
     },
     {
-      "epoch": 2.429640718562874,
-      "grad_norm": 0.050604316712189235,
-      "learning_rate": 1.8475670319938266e-06,
-      "loss": 0.0005,
+      "epoch": 2.5968,
+      "grad_norm": 0.011288009328992788,
+      "learning_rate": 9.332739882292752e-07,
+      "loss": 0.0001,
       "step": 1623
     },
     {
-      "epoch": 2.431137724550898,
-      "grad_norm": 0.05075746766744105,
-      "learning_rate": 1.8382140549337678e-06,
-      "loss": 0.0004,
+      "epoch": 2.5984,
+      "grad_norm": 0.0013096920311549136,
+      "learning_rate": 9.259980141081115e-07,
+      "loss": 0.0,
       "step": 1624
     },
     {
-      "epoch": 2.432634730538922,
-      "grad_norm": 0.06111785771892014,
-      "learning_rate": 1.8288824151632311e-06,
-      "loss": 0.0003,
+      "epoch": 2.6,
+      "grad_norm": 0.007070503131366707,
+      "learning_rate": 9.187491363342094e-07,
+      "loss": 0.0001,
       "step": 1625
     },
     {
-      "epoch": 2.434131736526946,
-      "grad_norm": 0.07080928006997747,
-      "learning_rate": 1.8195721370778474e-06,
-      "loss": 0.0007,
+      "epoch": 2.6016,
+      "grad_norm": 0.020044458861338762,
+      "learning_rate": 9.115273765538202e-07,
+      "loss": 0.0001,
       "step": 1626
     },
     {
-      "epoch": 2.43562874251497,
-      "grad_norm": 0.027992654625571187,
-      "learning_rate": 1.810283245017398e-06,
-      "loss": 0.0002,
+      "epoch": 2.6032,
+      "grad_norm": 0.0021069777638839694,
+      "learning_rate": 9.043327563322113e-07,
+      "loss": 0.0,
       "step": 1627
     },
     {
-      "epoch": 2.437125748502994,
-      "grad_norm": 0.03162118910496192,
-      "learning_rate": 1.8010157632657544e-06,
-      "loss": 0.0002,
+      "epoch": 2.6048,
+      "grad_norm": 0.0010715545791748374,
+      "learning_rate": 8.971652971536149e-07,
+      "loss": 0.0,
       "step": 1628
     },
     {
-      "epoch": 2.4386227544910177,
-      "grad_norm": 0.0662498938932939,
-      "learning_rate": 1.791769716050823e-06,
-      "loss": 0.0005,
+      "epoch": 2.6064,
+      "grad_norm": 0.1043762984750472,
+      "learning_rate": 8.900250204211513e-07,
+      "loss": 0.0008,
       "step": 1629
     },
     {
-      "epoch": 2.440119760479042,
-      "grad_norm": 0.06063032513396334,
-      "learning_rate": 1.782545127544466e-06,
-      "loss": 0.0015,
+      "epoch": 2.608,
+      "grad_norm": 0.08801440909859667,
+      "learning_rate": 8.829119474567672e-07,
+      "loss": 0.001,
       "step": 1630
     },
     {
-      "epoch": 2.441616766467066,
-      "grad_norm": 0.031943908561759334,
-      "learning_rate": 1.7733420218624486e-06,
-      "loss": 0.0002,
+      "epoch": 2.6096,
+      "grad_norm": 0.024913048320730432,
+      "learning_rate": 8.758260995011825e-07,
+      "loss": 0.0001,
       "step": 1631
     },
     {
-      "epoch": 2.44311377245509,
-      "grad_norm": 0.10126704114641819,
-      "learning_rate": 1.7641604230643738e-06,
-      "loss": 0.0011,
+      "epoch": 2.6112,
+      "grad_norm": 0.02846927487988488,
+      "learning_rate": 8.687674977138116e-07,
+      "loss": 0.0001,
       "step": 1632
     },
     {
-      "epoch": 2.444610778443114,
-      "grad_norm": 0.03485971575240973,
-      "learning_rate": 1.7550003551536177e-06,
-      "loss": 0.0003,
+      "epoch": 2.6128,
+      "grad_norm": 0.003911423188008586,
+      "learning_rate": 8.617361631727139e-07,
+      "loss": 0.0001,
       "step": 1633
     },
     {
-      "epoch": 2.446107784431138,
-      "grad_norm": 0.05038807899416904,
-      "learning_rate": 1.7458618420772777e-06,
-      "loss": 0.0004,
+      "epoch": 2.6144,
+      "grad_norm": 0.0020972048086177397,
+      "learning_rate": 8.547321168745192e-07,
+      "loss": 0.0,
       "step": 1634
     },
     {
-      "epoch": 2.447604790419162,
-      "grad_norm": 0.034427102319247505,
-      "learning_rate": 1.7367449077260823e-06,
+      "epoch": 2.616,
+      "grad_norm": 0.004155733735707268,
+      "learning_rate": 8.477553797343729e-07,
       "loss": 0.0001,
       "step": 1635
     },
     {
-      "epoch": 2.4491017964071857,
-      "grad_norm": 0.040602438643352785,
-      "learning_rate": 1.7276495759343647e-06,
-      "loss": 0.0003,
+      "epoch": 2.6176,
+      "grad_norm": 0.007412239826328842,
+      "learning_rate": 8.40805972585872e-07,
+      "loss": 0.0001,
       "step": 1636
     },
     {
-      "epoch": 2.4505988023952097,
-      "grad_norm": 0.0300674690331923,
-      "learning_rate": 1.7185758704799738e-06,
-      "loss": 0.0002,
+      "epoch": 2.6192,
+      "grad_norm": 0.005724175649760873,
+      "learning_rate": 8.338839161809997e-07,
+      "loss": 0.0001,
       "step": 1637
     },
     {
-      "epoch": 2.4520958083832336,
-      "grad_norm": 0.04412630684877554,
-      "learning_rate": 1.709523815084222e-06,
-      "loss": 0.0002,
+      "epoch": 2.6208,
+      "grad_norm": 0.012984937813147655,
+      "learning_rate": 8.269892311900696e-07,
+      "loss": 0.0001,
       "step": 1638
     },
     {
-      "epoch": 2.4535928143712575,
-      "grad_norm": 0.09352208849724938,
-      "learning_rate": 1.700493433411825e-06,
-      "loss": 0.0008,
+      "epoch": 2.6224,
+      "grad_norm": 0.0017638026481396388,
+      "learning_rate": 8.201219382016556e-07,
+      "loss": 0.0,
       "step": 1639
     },
     {
-      "epoch": 2.4550898203592815,
-      "grad_norm": 0.09616189622897968,
-      "learning_rate": 1.6914847490708296e-06,
-      "loss": 0.0006,
+      "epoch": 2.624,
+      "grad_norm": 0.057650553661999726,
+      "learning_rate": 8.132820577225386e-07,
+      "loss": 0.0002,
       "step": 1640
     },
     {
-      "epoch": 2.4565868263473054,
-      "grad_norm": 0.04164673607392628,
-      "learning_rate": 1.6824977856125745e-06,
-      "loss": 0.0003,
+      "epoch": 2.6256,
+      "grad_norm": 0.0011505844788008868,
+      "learning_rate": 8.06469610177636e-07,
+      "loss": 0.0,
       "step": 1641
     },
     {
-      "epoch": 2.4580838323353293,
-      "grad_norm": 0.02769372221761515,
-      "learning_rate": 1.6735325665315927e-06,
-      "loss": 0.0002,
+      "epoch": 2.6272,
+      "grad_norm": 0.001737434059617232,
+      "learning_rate": 7.996846159099558e-07,
+      "loss": 0.0,
       "step": 1642
     },
     {
-      "epoch": 2.4595808383233533,
-      "grad_norm": 0.03785938337316224,
-      "learning_rate": 1.6645891152655903e-06,
-      "loss": 0.0002,
+      "epoch": 2.6288,
+      "grad_norm": 0.0014599876484965232,
+      "learning_rate": 7.92927095180518e-07,
+      "loss": 0.0,
       "step": 1643
     },
     {
-      "epoch": 2.461077844311377,
-      "grad_norm": 0.07580082376824222,
-      "learning_rate": 1.6556674551953555e-06,
-      "loss": 0.0009,
+      "epoch": 2.6304,
+      "grad_norm": 0.01040258046892802,
+      "learning_rate": 7.861970681683051e-07,
+      "loss": 0.0001,
       "step": 1644
     },
     {
-      "epoch": 2.462574850299401,
-      "grad_norm": 0.03577911495005629,
-      "learning_rate": 1.6467676096447083e-06,
+      "epoch": 2.632,
+      "grad_norm": 0.004601741813038693,
+      "learning_rate": 7.794945549701993e-07,
       "loss": 0.0001,
       "step": 1645
     },
     {
-      "epoch": 2.464071856287425,
-      "grad_norm": 0.04888563503909771,
-      "learning_rate": 1.637889601880448e-06,
-      "loss": 0.0003,
+      "epoch": 2.6336,
+      "grad_norm": 0.005137280826948722,
+      "learning_rate": 7.728195756009204e-07,
+      "loss": 0.0001,
       "step": 1646
     },
     {
-      "epoch": 2.465568862275449,
-      "grad_norm": 0.028629948760254195,
-      "learning_rate": 1.6290334551122678e-06,
-      "loss": 0.0002,
+      "epoch": 2.6352,
+      "grad_norm": 0.009799188810212509,
+      "learning_rate": 7.661721499929753e-07,
+      "loss": 0.0001,
       "step": 1647
     },
     {
-      "epoch": 2.467065868263473,
-      "grad_norm": 0.09745710567212712,
-      "learning_rate": 1.6201991924927252e-06,
-      "loss": 0.0008,
+      "epoch": 2.6368,
+      "grad_norm": 0.001969839919556338,
+      "learning_rate": 7.595522979965819e-07,
+      "loss": 0.0,
       "step": 1648
     },
     {
-      "epoch": 2.468562874251497,
-      "grad_norm": 0.007786179248720061,
-      "learning_rate": 1.6113868371171592e-06,
-      "loss": 0.0001,
+      "epoch": 2.6384,
+      "grad_norm": 0.0018773395021817333,
+      "learning_rate": 7.529600393796232e-07,
+      "loss": 0.0,
       "step": 1649
     },
     {
-      "epoch": 2.470059880239521,
-      "grad_norm": 0.09307414131421253,
-      "learning_rate": 1.6025964120236337e-06,
-      "loss": 0.0003,
+      "epoch": 2.64,
+      "grad_norm": 0.024320439130425177,
+      "learning_rate": 7.463953938275859e-07,
+      "loss": 0.0002,
       "step": 1650
     },
     {
-      "epoch": 2.4715568862275448,
-      "grad_norm": 0.08000883258976127,
-      "learning_rate": 1.5938279401928925e-06,
-      "loss": 0.0007,
+      "epoch": 2.6416,
+      "grad_norm": 0.004900882592944951,
+      "learning_rate": 7.398583809434944e-07,
+      "loss": 0.0001,
       "step": 1651
     },
     {
-      "epoch": 2.4730538922155687,
-      "grad_norm": 0.08529740066647153,
-      "learning_rate": 1.58508144454827e-06,
-      "loss": 0.0006,
+      "epoch": 2.6432,
+      "grad_norm": 0.003407605039889639,
+      "learning_rate": 7.333490202478666e-07,
+      "loss": 0.0,
       "step": 1652
     },
     {
-      "epoch": 2.4745508982035926,
-      "grad_norm": 0.1732944945698624,
-      "learning_rate": 1.576356947955665e-06,
-      "loss": 0.0007,
+      "epoch": 2.6448,
+      "grad_norm": 0.00694630203245316,
+      "learning_rate": 7.268673311786378e-07,
+      "loss": 0.0001,
       "step": 1653
     },
     {
-      "epoch": 2.4760479041916166,
-      "grad_norm": 0.05910999084092352,
-      "learning_rate": 1.5676544732234477e-06,
-      "loss": 0.0004,
+      "epoch": 2.6464,
+      "grad_norm": 0.006897439205802054,
+      "learning_rate": 7.204133330911179e-07,
+      "loss": 0.0001,
       "step": 1654
     },
     {
-      "epoch": 2.477544910179641,
-      "grad_norm": 0.03736554597645004,
-      "learning_rate": 1.5589740431024347e-06,
-      "loss": 0.0003,
+      "epoch": 2.648,
+      "grad_norm": 0.0036013533467608777,
+      "learning_rate": 7.1398704525792e-07,
+      "loss": 0.0001,
       "step": 1655
     },
     {
-      "epoch": 2.479041916167665,
-      "grad_norm": 0.05815438383385881,
-      "learning_rate": 1.5503156802857998e-06,
-      "loss": 0.0003,
+      "epoch": 2.6496,
+      "grad_norm": 0.00605619321620667,
+      "learning_rate": 7.07588486868922e-07,
+      "loss": 0.0001,
       "step": 1656
     },
     {
-      "epoch": 2.480538922155689,
-      "grad_norm": 0.009680132499123993,
-      "learning_rate": 1.5416794074090258e-06,
+      "epoch": 2.6512000000000002,
+      "grad_norm": 0.012095497687233375,
+      "learning_rate": 7.012176770311863e-07,
       "loss": 0.0001,
       "step": 1657
     },
     {
-      "epoch": 2.4820359281437128,
-      "grad_norm": 0.043831899612723345,
-      "learning_rate": 1.5330652470498574e-06,
-      "loss": 0.0003,
+      "epoch": 2.6528,
+      "grad_norm": 0.04516819018819963,
+      "learning_rate": 6.948746347689184e-07,
+      "loss": 0.0001,
       "step": 1658
     },
     {
-      "epoch": 2.4835329341317367,
-      "grad_norm": 0.0208540727997247,
-      "learning_rate": 1.5244732217282132e-06,
+      "epoch": 2.6544,
+      "grad_norm": 0.011096821132557921,
+      "learning_rate": 6.885593790234057e-07,
       "loss": 0.0001,
       "step": 1659
     },
     {
-      "epoch": 2.4850299401197606,
-      "grad_norm": 0.02609799149850617,
-      "learning_rate": 1.5159033539061607e-06,
-      "loss": 0.0002,
+      "epoch": 2.656,
+      "grad_norm": 0.008804471813855973,
+      "learning_rate": 6.8227192865296e-07,
+      "loss": 0.0001,
       "step": 1660
     },
     {
-      "epoch": 2.4865269461077846,
-      "grad_norm": 0.012809457260665172,
-      "learning_rate": 1.507355665987833e-06,
-      "loss": 0.0001,
+      "epoch": 2.6576,
+      "grad_norm": 0.0012046966278960764,
+      "learning_rate": 6.760123024328624e-07,
+      "loss": 0.0,
       "step": 1661
     },
     {
-      "epoch": 2.4880239520958085,
-      "grad_norm": 0.03092643008061081,
-      "learning_rate": 1.4988301803193773e-06,
-      "loss": 0.0002,
+      "epoch": 2.6592000000000002,
+      "grad_norm": 0.003173321378690416,
+      "learning_rate": 6.697805190553086e-07,
+      "loss": 0.0001,
       "step": 1662
     },
     {
-      "epoch": 2.4895209580838324,
-      "grad_norm": 0.11021032771407564,
-      "learning_rate": 1.4903269191889068e-06,
-      "loss": 0.0008,
+      "epoch": 2.6608,
+      "grad_norm": 0.007078437592975865,
+      "learning_rate": 6.635765971293484e-07,
+      "loss": 0.0001,
       "step": 1663
     },
     {
-      "epoch": 2.4910179640718564,
-      "grad_norm": 0.14645604895626768,
-      "learning_rate": 1.4818459048264179e-06,
-      "loss": 0.0008,
+      "epoch": 2.6624,
+      "grad_norm": 0.0018172268038980372,
+      "learning_rate": 6.574005551808338e-07,
+      "loss": 0.0,
       "step": 1664
     },
     {
-      "epoch": 2.4925149700598803,
-      "grad_norm": 0.08473709608733745,
-      "learning_rate": 1.4733871594037629e-06,
-      "loss": 0.0005,
+      "epoch": 2.664,
+      "grad_norm": 0.003387870664722808,
+      "learning_rate": 6.512524116523633e-07,
+      "loss": 0.0,
       "step": 1665
     },
     {
-      "epoch": 2.4940119760479043,
-      "grad_norm": 0.28509040224007653,
-      "learning_rate": 1.464950705034568e-06,
-      "loss": 0.0012,
+      "epoch": 2.6656,
+      "grad_norm": 0.002542377727573644,
+      "learning_rate": 6.451321849032289e-07,
+      "loss": 0.0,
       "step": 1666
     },
     {
-      "epoch": 2.495508982035928,
-      "grad_norm": 0.024381609398717058,
-      "learning_rate": 1.4565365637741868e-06,
+      "epoch": 2.6672000000000002,
+      "grad_norm": 0.00496977538975312,
+      "learning_rate": 6.390398932093555e-07,
       "loss": 0.0001,
       "step": 1667
     },
     {
-      "epoch": 2.497005988023952,
-      "grad_norm": 0.09142786032282342,
-      "learning_rate": 1.4481447576196372e-06,
-      "loss": 0.0013,
+      "epoch": 2.6688,
+      "grad_norm": 0.004811840381794632,
+      "learning_rate": 6.329755547632499e-07,
+      "loss": 0.0001,
       "step": 1668
     },
     {
-      "epoch": 2.498502994011976,
-      "grad_norm": 0.09847239340356413,
-      "learning_rate": 1.439775308509549e-06,
-      "loss": 0.0007,
+      "epoch": 2.6704,
+      "grad_norm": 0.01883487701635383,
+      "learning_rate": 6.269391876739494e-07,
+      "loss": 0.0001,
       "step": 1669
     },
     {
-      "epoch": 2.5,
-      "grad_norm": 0.03477841977609638,
-      "learning_rate": 1.4314282383241097e-06,
-      "loss": 0.0002,
+      "epoch": 2.672,
+      "grad_norm": 0.006518415776134661,
+      "learning_rate": 6.209308099669598e-07,
+      "loss": 0.0001,
       "step": 1670
     },
     {
-      "epoch": 2.501497005988024,
-      "grad_norm": 0.04628201955871934,
-      "learning_rate": 1.423103568884986e-06,
-      "loss": 0.0003,
+      "epoch": 2.6736,
+      "grad_norm": 0.002339312583530842,
+      "learning_rate": 6.149504395842087e-07,
+      "loss": 0.0,
       "step": 1671
     },
     {
-      "epoch": 2.502994011976048,
-      "grad_norm": 0.09307706909626917,
-      "learning_rate": 1.4148013219552992e-06,
-      "loss": 0.0007,
+      "epoch": 2.6752000000000002,
+      "grad_norm": 0.009995307579150481,
+      "learning_rate": 6.089980943839924e-07,
+      "loss": 0.0001,
       "step": 1672
     },
     {
-      "epoch": 2.504491017964072,
-      "grad_norm": 0.07777395668753506,
-      "learning_rate": 1.406521519239543e-06,
-      "loss": 0.0004,
+      "epoch": 2.6768,
+      "grad_norm": 0.0052798775377979335,
+      "learning_rate": 6.030737921409169e-07,
+      "loss": 0.0,
       "step": 1673
     },
     {
-      "epoch": 2.5059880239520957,
-      "grad_norm": 0.038725094614758614,
-      "learning_rate": 1.3982641823835385e-06,
-      "loss": 0.0002,
+      "epoch": 2.6784,
+      "grad_norm": 0.00539784506601244,
+      "learning_rate": 5.971775505458444e-07,
+      "loss": 0.0001,
       "step": 1674
     },
     {
-      "epoch": 2.5074850299401197,
-      "grad_norm": 0.09347821247400721,
-      "learning_rate": 1.3900293329743707e-06,
-      "loss": 0.0008,
+      "epoch": 2.68,
+      "grad_norm": 0.022433981764194488,
+      "learning_rate": 5.913093872058528e-07,
+      "loss": 0.0001,
       "step": 1675
     },
     {
-      "epoch": 2.5089820359281436,
-      "grad_norm": 0.011973171492511713,
-      "learning_rate": 1.381816992540338e-06,
+      "epoch": 2.6816,
+      "grad_norm": 0.028818795412867394,
+      "learning_rate": 5.854693196441641e-07,
       "loss": 0.0001,
       "step": 1676
     },
     {
-      "epoch": 2.5104790419161676,
-      "grad_norm": 0.03349515304688116,
-      "learning_rate": 1.3736271825508973e-06,
-      "loss": 0.0002,
+      "epoch": 2.6832000000000003,
+      "grad_norm": 0.004945354649552436,
+      "learning_rate": 5.796573653001091e-07,
+      "loss": 0.0001,
       "step": 1677
     },
     {
-      "epoch": 2.5119760479041915,
-      "grad_norm": 0.01810852300179121,
-      "learning_rate": 1.3654599244166e-06,
+      "epoch": 2.6848,
+      "grad_norm": 0.0030640210420676706,
+      "learning_rate": 5.738735415290642e-07,
       "loss": 0.0001,
       "step": 1678
     },
     {
-      "epoch": 2.5134730538922154,
-      "grad_norm": 0.04446933710711642,
-      "learning_rate": 1.3573152394890398e-06,
-      "loss": 0.0003,
+      "epoch": 2.6864,
+      "grad_norm": 0.020235631967633456,
+      "learning_rate": 5.681178656024055e-07,
+      "loss": 0.0001,
       "step": 1679
     },
     {
-      "epoch": 2.5149700598802394,
-      "grad_norm": 0.05815871825675632,
-      "learning_rate": 1.3491931490608023e-06,
-      "loss": 0.0002,
+      "epoch": 2.6879999999999997,
+      "grad_norm": 0.006967025977732761,
+      "learning_rate": 5.62390354707455e-07,
+      "loss": 0.0001,
       "step": 1680
     },
     {
-      "epoch": 2.5164670658682633,
-      "grad_norm": 0.13415081479279004,
-      "learning_rate": 1.341093674365398e-06,
-      "loss": 0.0017,
+      "epoch": 2.6896,
+      "grad_norm": 0.0014463248530798556,
+      "learning_rate": 5.56691025947429e-07,
+      "loss": 0.0,
       "step": 1681
     },
     {
-      "epoch": 2.5179640718562872,
-      "grad_norm": 0.026418150113077105,
-      "learning_rate": 1.333016836577219e-06,
-      "loss": 0.0002,
+      "epoch": 2.6912000000000003,
+      "grad_norm": 0.033003970062304676,
+      "learning_rate": 5.510198963413882e-07,
+      "loss": 0.0001,
       "step": 1682
     },
     {
-      "epoch": 2.519461077844311,
-      "grad_norm": 0.05637402946123024,
-      "learning_rate": 1.3249626568114716e-06,
-      "loss": 0.0004,
+      "epoch": 2.6928,
+      "grad_norm": 0.005260609035935109,
+      "learning_rate": 5.453769828241872e-07,
+      "loss": 0.0001,
       "step": 1683
     },
     {
-      "epoch": 2.520958083832335,
-      "grad_norm": 0.03094599306924757,
-      "learning_rate": 1.3169311561241371e-06,
-      "loss": 0.0002,
+      "epoch": 2.6944,
+      "grad_norm": 0.002722753065014123,
+      "learning_rate": 5.397623022464227e-07,
+      "loss": 0.0,
       "step": 1684
     },
     {
-      "epoch": 2.522455089820359,
-      "grad_norm": 0.027335429413912742,
-      "learning_rate": 1.308922355511899e-06,
-      "loss": 0.0002,
+      "epoch": 2.6959999999999997,
+      "grad_norm": 0.01670882765259882,
+      "learning_rate": 5.341758713743828e-07,
+      "loss": 0.0001,
       "step": 1685
     },
     {
-      "epoch": 2.523952095808383,
-      "grad_norm": 0.1488861125220278,
-      "learning_rate": 1.300936275912098e-06,
-      "loss": 0.001,
+      "epoch": 2.6976,
+      "grad_norm": 0.002879707146007582,
+      "learning_rate": 5.286177068899989e-07,
+      "loss": 0.0,
       "step": 1686
     },
     {
-      "epoch": 2.525449101796407,
-      "grad_norm": 0.05818953193532018,
-      "learning_rate": 1.2929729382026757e-06,
-      "loss": 0.0004,
+      "epoch": 2.6992000000000003,
+      "grad_norm": 0.0020607402388883823,
+      "learning_rate": 5.230878253907911e-07,
+      "loss": 0.0,
       "step": 1687
     },
     {
-      "epoch": 2.5269461077844313,
-      "grad_norm": 0.07088482664758781,
-      "learning_rate": 1.285032363202121e-06,
-      "loss": 0.0007,
+      "epoch": 2.7008,
+      "grad_norm": 0.00563594549364586,
+      "learning_rate": 5.175862433898282e-07,
+      "loss": 0.0001,
       "step": 1688
     },
     {
-      "epoch": 2.5284431137724552,
-      "grad_norm": 0.09878385939626243,
-      "learning_rate": 1.277114571669411e-06,
-      "loss": 0.0005,
+      "epoch": 2.7024,
+      "grad_norm": 0.005836874900768736,
+      "learning_rate": 5.121129773156663e-07,
+      "loss": 0.0001,
       "step": 1689
     },
     {
-      "epoch": 2.529940119760479,
-      "grad_norm": 0.08834537093128993,
-      "learning_rate": 1.2692195843039668e-06,
-      "loss": 0.0005,
+      "epoch": 2.7039999999999997,
+      "grad_norm": 0.016843751964607703,
+      "learning_rate": 5.066680435123106e-07,
+      "loss": 0.0001,
       "step": 1690
     },
     {
-      "epoch": 2.531437125748503,
-      "grad_norm": 0.028045651170751838,
-      "learning_rate": 1.2613474217455856e-06,
-      "loss": 0.0004,
+      "epoch": 2.7056,
+      "grad_norm": 0.071560272626772,
+      "learning_rate": 5.012514582391592e-07,
+      "loss": 0.0003,
       "step": 1691
     },
     {
-      "epoch": 2.532934131736527,
-      "grad_norm": 0.08064194210729632,
-      "learning_rate": 1.2534981045743987e-06,
-      "loss": 0.0003,
+      "epoch": 2.7072000000000003,
+      "grad_norm": 0.022576956052241456,
+      "learning_rate": 4.95863237670956e-07,
+      "loss": 0.0001,
       "step": 1692
     },
     {
-      "epoch": 2.534431137724551,
-      "grad_norm": 0.06878912490099846,
-      "learning_rate": 1.2456716533108094e-06,
-      "loss": 0.0003,
+      "epoch": 2.7088,
+      "grad_norm": 0.0030571293977662457,
+      "learning_rate": 4.905033978977492e-07,
+      "loss": 0.0,
       "step": 1693
     },
     {
-      "epoch": 2.535928143712575,
-      "grad_norm": 0.09475527318363501,
-      "learning_rate": 1.2378680884154437e-06,
-      "loss": 0.0009,
+      "epoch": 2.7104,
+      "grad_norm": 0.01487082783820577,
+      "learning_rate": 4.851719549248301e-07,
+      "loss": 0.0001,
       "step": 1694
     },
     {
-      "epoch": 2.537425149700599,
-      "grad_norm": 0.06988351944817416,
-      "learning_rate": 1.230087430289103e-06,
-      "loss": 0.0005,
+      "epoch": 2.7119999999999997,
+      "grad_norm": 0.002060687324143398,
+      "learning_rate": 4.798689246727006e-07,
+      "loss": 0.0001,
       "step": 1695
     },
     {
-      "epoch": 2.538922155688623,
-      "grad_norm": 0.06025644551142922,
-      "learning_rate": 1.2223296992726885e-06,
-      "loss": 0.0003,
+      "epoch": 2.7136,
+      "grad_norm": 0.0040990121477614315,
+      "learning_rate": 4.7459432297701224e-07,
+      "loss": 0.0,
       "step": 1696
     },
     {
-      "epoch": 2.5404191616766467,
-      "grad_norm": 0.05586088137246275,
-      "learning_rate": 1.2145949156471781e-06,
-      "loss": 0.0003,
+      "epoch": 2.7152,
+      "grad_norm": 0.0399491811244669,
+      "learning_rate": 4.693481655885257e-07,
+      "loss": 0.0001,
       "step": 1697
     },
     {
-      "epoch": 2.5419161676646707,
-      "grad_norm": 0.019554709115490496,
-      "learning_rate": 1.2068830996335512e-06,
-      "loss": 0.0001,
+      "epoch": 2.7168,
+      "grad_norm": 0.002561971861201848,
+      "learning_rate": 4.6413046817306404e-07,
+      "loss": 0.0,
       "step": 1698
     },
     {
-      "epoch": 2.5434131736526946,
-      "grad_norm": 0.060450587058907566,
-      "learning_rate": 1.1991942713927452e-06,
-      "loss": 0.0003,
+      "epoch": 2.7184,
+      "grad_norm": 0.055233561660464434,
+      "learning_rate": 4.58941246311464e-07,
+      "loss": 0.0004,
       "step": 1699
     },
     {
-      "epoch": 2.5449101796407185,
-      "grad_norm": 0.05151733193244522,
-      "learning_rate": 1.1915284510255998e-06,
-      "loss": 0.0003,
+      "epoch": 2.7199999999999998,
+      "grad_norm": 0.020471596392817745,
+      "learning_rate": 4.5378051549952783e-07,
+      "loss": 0.0001,
       "step": 1700
     },
     {
-      "epoch": 2.5464071856287425,
-      "grad_norm": 0.08154849069751298,
-      "learning_rate": 1.1838856585728042e-06,
-      "loss": 0.0008,
+      "epoch": 2.7216,
+      "grad_norm": 0.0053295203443414355,
+      "learning_rate": 4.4864829114798394e-07,
+      "loss": 0.0001,
       "step": 1701
     },
     {
-      "epoch": 2.5479041916167664,
-      "grad_norm": 0.025258520621501784,
-      "learning_rate": 1.1762659140148524e-06,
-      "loss": 0.0002,
+      "epoch": 2.7232,
+      "grad_norm": 0.001419225495881021,
+      "learning_rate": 4.4354458858242857e-07,
+      "loss": 0.0,
       "step": 1702
     },
     {
-      "epoch": 2.5494011976047903,
-      "grad_norm": 0.08809425431554686,
-      "learning_rate": 1.1686692372719721e-06,
-      "loss": 0.0006,
+      "epoch": 2.7248,
+      "grad_norm": 0.017354646612918568,
+      "learning_rate": 4.384694230432984e-07,
+      "loss": 0.0001,
       "step": 1703
     },
     {
-      "epoch": 2.5508982035928143,
-      "grad_norm": 0.09231116438170844,
-      "learning_rate": 1.1610956482040992e-06,
-      "loss": 0.0005,
+      "epoch": 2.7264,
+      "grad_norm": 0.0007440305948982215,
+      "learning_rate": 4.3342280968580287e-07,
+      "loss": 0.0,
       "step": 1704
     },
     {
-      "epoch": 2.552395209580838,
-      "grad_norm": 0.020971139154724327,
-      "learning_rate": 1.1535451666108022e-06,
-      "loss": 0.0002,
+      "epoch": 2.7279999999999998,
+      "grad_norm": 0.00373706251755768,
+      "learning_rate": 4.2840476357989825e-07,
+      "loss": 0.0001,
       "step": 1705
     },
     {
-      "epoch": 2.553892215568862,
-      "grad_norm": 0.07893486730149059,
-      "learning_rate": 1.1460178122312404e-06,
-      "loss": 0.0006,
+      "epoch": 2.7296,
+      "grad_norm": 0.007354008374441261,
+      "learning_rate": 4.2341529971023253e-07,
+      "loss": 0.0001,
       "step": 1706
     },
     {
-      "epoch": 2.555389221556886,
-      "grad_norm": 0.019260558530191367,
-      "learning_rate": 1.1385136047441214e-06,
-      "loss": 0.0001,
+      "epoch": 2.7312,
+      "grad_norm": 0.04433771596419746,
+      "learning_rate": 4.184544329761009e-07,
+      "loss": 0.0003,
       "step": 1707
     },
     {
-      "epoch": 2.55688622754491,
-      "grad_norm": 0.022801882208881713,
-      "learning_rate": 1.1310325637676233e-06,
-      "loss": 0.0002,
+      "epoch": 2.7328,
+      "grad_norm": 0.0019717761021554323,
+      "learning_rate": 4.1352217819140337e-07,
+      "loss": 0.0,
       "step": 1708
     },
     {
-      "epoch": 2.5583832335329344,
-      "grad_norm": 0.006273256765097044,
-      "learning_rate": 1.1235747088593773e-06,
-      "loss": 0.0001,
+      "epoch": 2.7344,
+      "grad_norm": 0.004624100015738466,
+      "learning_rate": 4.0861855008460403e-07,
+      "loss": 0.0,
       "step": 1709
     },
     {
-      "epoch": 2.5598802395209583,
-      "grad_norm": 0.025272286951359548,
-      "learning_rate": 1.1161400595163885e-06,
-      "loss": 0.0002,
+      "epoch": 2.7359999999999998,
+      "grad_norm": 0.011242224418644715,
+      "learning_rate": 4.037435632986786e-07,
+      "loss": 0.0001,
       "step": 1710
     },
     {
-      "epoch": 2.5613772455089823,
-      "grad_norm": 0.06144529667834819,
-      "learning_rate": 1.1087286351749981e-06,
-      "loss": 0.0005,
+      "epoch": 2.7376,
+      "grad_norm": 0.0031526690950823715,
+      "learning_rate": 3.988972323910778e-07,
+      "loss": 0.0,
       "step": 1711
     },
     {
-      "epoch": 2.562874251497006,
-      "grad_norm": 0.018647360603314567,
-      "learning_rate": 1.1013404552108365e-06,
-      "loss": 0.0001,
+      "epoch": 2.7392,
+      "grad_norm": 0.0013638216653592268,
+      "learning_rate": 3.9407957183368093e-07,
+      "loss": 0.0,
       "step": 1712
     },
     {
-      "epoch": 2.56437125748503,
-      "grad_norm": 0.040545615112091975,
-      "learning_rate": 1.0939755389387552e-06,
-      "loss": 0.0003,
+      "epoch": 2.7408,
+      "grad_norm": 0.003093293558689636,
+      "learning_rate": 3.8929059601275463e-07,
+      "loss": 0.0,
       "step": 1713
     },
     {
-      "epoch": 2.565868263473054,
-      "grad_norm": 0.024320530278532197,
-      "learning_rate": 1.0866339056127995e-06,
-      "loss": 0.0002,
+      "epoch": 2.7424,
+      "grad_norm": 0.013540525963096324,
+      "learning_rate": 3.845303192289074e-07,
+      "loss": 0.0001,
       "step": 1714
     },
     {
-      "epoch": 2.567365269461078,
-      "grad_norm": 0.092450238455231,
-      "learning_rate": 1.0793155744261352e-06,
-      "loss": 0.0005,
+      "epoch": 2.7439999999999998,
+      "grad_norm": 0.08169187849279333,
+      "learning_rate": 3.797987556970495e-07,
+      "loss": 0.0007,
       "step": 1715
     },
     {
-      "epoch": 2.568862275449102,
-      "grad_norm": 0.05655287368538696,
-      "learning_rate": 1.0720205645110193e-06,
-      "loss": 0.0004,
+      "epoch": 2.7456,
+      "grad_norm": 0.0022705415448893134,
+      "learning_rate": 3.750959195463466e-07,
+      "loss": 0.0001,
       "step": 1716
     },
     {
-      "epoch": 2.570359281437126,
-      "grad_norm": 0.08988286767798637,
-      "learning_rate": 1.064748894938734e-06,
-      "loss": 0.0007,
+      "epoch": 2.7472,
+      "grad_norm": 0.0012207756210360957,
+      "learning_rate": 3.7042182482018074e-07,
+      "loss": 0.0,
       "step": 1717
     },
     {
-      "epoch": 2.57185628742515,
-      "grad_norm": 0.07500638380677978,
-      "learning_rate": 1.0575005847195442e-06,
+      "epoch": 2.7488,
+      "grad_norm": 0.07545932558598065,
+      "learning_rate": 3.6577648547611033e-07,
       "loss": 0.0004,
       "step": 1718
     },
     {
-      "epoch": 2.5733532934131738,
-      "grad_norm": 0.0925835650680482,
-      "learning_rate": 1.0502756528026526e-06,
-      "loss": 0.0005,
+      "epoch": 2.7504,
+      "grad_norm": 0.0015078232572364363,
+      "learning_rate": 3.611599153858214e-07,
+      "loss": 0.0,
       "step": 1719
     },
     {
-      "epoch": 2.5748502994011977,
-      "grad_norm": 0.06643680023971794,
-      "learning_rate": 1.0430741180761306e-06,
-      "loss": 0.0004,
+      "epoch": 2.752,
+      "grad_norm": 0.0012544209214220055,
+      "learning_rate": 3.5657212833509313e-07,
+      "loss": 0.0,
       "step": 1720
     },
     {
-      "epoch": 2.5763473053892216,
-      "grad_norm": 0.07290249607472542,
-      "learning_rate": 1.0358959993668981e-06,
-      "loss": 0.0005,
+      "epoch": 2.7536,
+      "grad_norm": 0.003472156993994934,
+      "learning_rate": 3.520131380237546e-07,
+      "loss": 0.0001,
       "step": 1721
     },
     {
-      "epoch": 2.5778443113772456,
-      "grad_norm": 0.01775705006856804,
-      "learning_rate": 1.0287413154406478e-06,
-      "loss": 0.0003,
+      "epoch": 2.7552,
+      "grad_norm": 0.0016692784689792757,
+      "learning_rate": 3.474829580656436e-07,
+      "loss": 0.0,
       "step": 1722
     },
     {
-      "epoch": 2.5793413173652695,
-      "grad_norm": 0.08273329531565872,
-      "learning_rate": 1.0216100850018117e-06,
-      "loss": 0.0006,
+      "epoch": 2.7568,
+      "grad_norm": 0.010967164559959268,
+      "learning_rate": 3.429816019885657e-07,
+      "loss": 0.0001,
       "step": 1723
     },
     {
-      "epoch": 2.5808383233532934,
-      "grad_norm": 0.028537281972692816,
-      "learning_rate": 1.0145023266935073e-06,
-      "loss": 0.0003,
+      "epoch": 2.7584,
+      "grad_norm": 0.0039539416570045305,
+      "learning_rate": 3.385090832342497e-07,
+      "loss": 0.0,
       "step": 1724
     },
     {
-      "epoch": 2.5823353293413174,
-      "grad_norm": 0.10369851400849742,
-      "learning_rate": 1.0074180590974847e-06,
-      "loss": 0.0012,
+      "epoch": 2.76,
+      "grad_norm": 0.001166654064799598,
+      "learning_rate": 3.3406541515832e-07,
+      "loss": 0.0,
       "step": 1725
     },
     {
-      "epoch": 2.5838323353293413,
-      "grad_norm": 0.08409029120428357,
-      "learning_rate": 1.0003573007340928e-06,
-      "loss": 0.0005,
+      "epoch": 2.7616,
+      "grad_norm": 0.016979602442845958,
+      "learning_rate": 3.296506110302422e-07,
+      "loss": 0.0001,
       "step": 1726
     },
     {
-      "epoch": 2.5853293413173652,
-      "grad_norm": 0.021599660351613818,
-      "learning_rate": 9.93320070062207e-07,
-      "loss": 0.0002,
+      "epoch": 2.7632,
+      "grad_norm": 0.005761879588366482,
+      "learning_rate": 3.252646840332918e-07,
+      "loss": 0.0,
       "step": 1727
     },
     {
-      "epoch": 2.586826347305389,
-      "grad_norm": 0.03987969894778265,
-      "learning_rate": 9.863063854792055e-07,
-      "loss": 0.0004,
+      "epoch": 2.7648,
+      "grad_norm": 0.015844050587905473,
+      "learning_rate": 3.209076472645112e-07,
+      "loss": 0.0001,
       "step": 1728
     },
     {
-      "epoch": 2.588323353293413,
-      "grad_norm": 0.0763711713078856,
-      "learning_rate": 9.793162653209053e-07,
-      "loss": 0.0005,
+      "epoch": 2.7664,
+      "grad_norm": 0.0008940851940196189,
+      "learning_rate": 3.16579513734675e-07,
+      "loss": 0.0,
       "step": 1729
     },
     {
-      "epoch": 2.589820359281437,
-      "grad_norm": 0.06750300685949333,
-      "learning_rate": 9.72349727861519e-07,
-      "loss": 0.0002,
+      "epoch": 2.768,
+      "grad_norm": 0.10241057217425732,
+      "learning_rate": 3.1228029636824477e-07,
+      "loss": 0.0004,
       "step": 1730
     },
     {
-      "epoch": 2.591317365269461,
-      "grad_norm": 0.1192723175715262,
-      "learning_rate": 9.654067913136079e-07,
-      "loss": 0.0011,
+      "epoch": 2.7696,
+      "grad_norm": 0.0011100895364327215,
+      "learning_rate": 3.080100080033388e-07,
+      "loss": 0.0,
       "step": 1731
     },
     {
-      "epoch": 2.592814371257485,
-      "grad_norm": 0.06166676002980035,
-      "learning_rate": 9.584874738280315e-07,
-      "loss": 0.0004,
+      "epoch": 2.7712,
+      "grad_norm": 0.008733426113929346,
+      "learning_rate": 3.037686613916857e-07,
+      "loss": 0.0001,
       "step": 1732
     },
     {
-      "epoch": 2.594311377245509,
-      "grad_norm": 0.08400165230894974,
-      "learning_rate": 9.515917934939089e-07,
-      "loss": 0.0005,
+      "epoch": 2.7728,
+      "grad_norm": 0.030213343478735843,
+      "learning_rate": 2.995562691985898e-07,
+      "loss": 0.0001,
       "step": 1733
     },
     {
-      "epoch": 2.595808383233533,
-      "grad_norm": 0.02748157877354551,
-      "learning_rate": 9.447197683385579e-07,
-      "loss": 0.0002,
+      "epoch": 2.7744,
+      "grad_norm": 0.01515621419653102,
+      "learning_rate": 2.9537284400289354e-07,
+      "loss": 0.0001,
       "step": 1734
     },
     {
-      "epoch": 2.5973053892215567,
-      "grad_norm": 0.060188118708937415,
-      "learning_rate": 9.378714163274571e-07,
-      "loss": 0.0004,
+      "epoch": 2.776,
+      "grad_norm": 0.06292426446852466,
+      "learning_rate": 2.9121839829693857e-07,
+      "loss": 0.0003,
       "step": 1735
     },
     {
-      "epoch": 2.5988023952095807,
-      "grad_norm": 0.07502732321886861,
-      "learning_rate": 9.310467553641967e-07,
-      "loss": 0.0011,
+      "epoch": 2.7776,
+      "grad_norm": 0.0024209705442198126,
+      "learning_rate": 2.8709294448653223e-07,
+      "loss": 0.0,
       "step": 1736
     },
     {
-      "epoch": 2.6002994011976046,
-      "grad_norm": 0.05755762200466036,
-      "learning_rate": 9.242458032904311e-07,
-      "loss": 0.001,
+      "epoch": 2.7792,
+      "grad_norm": 0.00296748529685002,
+      "learning_rate": 2.829964948909048e-07,
+      "loss": 0.0001,
       "step": 1737
     },
     {
-      "epoch": 2.6017964071856285,
-      "grad_norm": 0.049212105593266754,
-      "learning_rate": 9.174685778858306e-07,
-      "loss": 0.0003,
+      "epoch": 2.7808,
+      "grad_norm": 0.0022259173904971598,
+      "learning_rate": 2.7892906174267653e-07,
+      "loss": 0.0,
       "step": 1738
     },
     {
-      "epoch": 2.6032934131736525,
-      "grad_norm": 0.08540752717152685,
-      "learning_rate": 9.107150968680445e-07,
-      "loss": 0.0009,
+      "epoch": 2.7824,
+      "grad_norm": 0.001630202727036233,
+      "learning_rate": 2.748906571878207e-07,
+      "loss": 0.0,
       "step": 1739
     },
     {
-      "epoch": 2.6047904191616764,
-      "grad_norm": 0.08879092487810307,
-      "learning_rate": 9.0398537789264e-07,
-      "loss": 0.0005,
+      "epoch": 2.784,
+      "grad_norm": 0.0015752150368830234,
+      "learning_rate": 2.708812932856253e-07,
+      "loss": 0.0,
       "step": 1740
     },
     {
-      "epoch": 2.6062874251497004,
-      "grad_norm": 0.059349429063367105,
-      "learning_rate": 8.972794385530658e-07,
-      "loss": 0.0003,
+      "epoch": 2.7856,
+      "grad_norm": 0.0008131726024755656,
+      "learning_rate": 2.6690098200866097e-07,
+      "loss": 0.0,
       "step": 1741
     },
     {
-      "epoch": 2.6077844311377243,
-      "grad_norm": 0.08005756215182938,
-      "learning_rate": 8.905972963806042e-07,
-      "loss": 0.0004,
+      "epoch": 2.7872,
+      "grad_norm": 0.002585841509949017,
+      "learning_rate": 2.6294973524274127e-07,
+      "loss": 0.0,
       "step": 1742
     },
     {
-      "epoch": 2.6092814371257482,
-      "grad_norm": 0.10352902840191881,
-      "learning_rate": 8.839389688443234e-07,
-      "loss": 0.0009,
+      "epoch": 2.7888,
+      "grad_norm": 0.0009328843269095667,
+      "learning_rate": 2.5902756478688674e-07,
+      "loss": 0.0,
       "step": 1743
     },
     {
-      "epoch": 2.6107784431137726,
-      "grad_norm": 0.0167977465176402,
-      "learning_rate": 8.773044733510338e-07,
+      "epoch": 2.7904,
+      "grad_norm": 0.00836227744998053,
+      "learning_rate": 2.551344823532964e-07,
       "loss": 0.0001,
       "step": 1744
     },
     {
-      "epoch": 2.6122754491017965,
-      "grad_norm": 0.04717208793979273,
-      "learning_rate": 8.706938272452414e-07,
-      "loss": 0.0003,
+      "epoch": 2.792,
+      "grad_norm": 0.009232310436466376,
+      "learning_rate": 2.5127049956730207e-07,
+      "loss": 0.0001,
       "step": 1745
     },
     {
-      "epoch": 2.6137724550898205,
-      "grad_norm": 0.15766515627181438,
-      "learning_rate": 8.641070478091063e-07,
-      "loss": 0.0015,
+      "epoch": 2.7936,
+      "grad_norm": 0.10369910632104141,
+      "learning_rate": 2.474356279673462e-07,
+      "loss": 0.0002,
       "step": 1746
     },
     {
-      "epoch": 2.6152694610778444,
-      "grad_norm": 0.05933348420600678,
-      "learning_rate": 8.575441522623906e-07,
-      "loss": 0.0005,
+      "epoch": 2.7952,
+      "grad_norm": 0.049261560321665294,
+      "learning_rate": 2.436298790049363e-07,
+      "loss": 0.0002,
       "step": 1747
     },
     {
-      "epoch": 2.6167664670658684,
-      "grad_norm": 0.16483126615536,
-      "learning_rate": 8.510051577624179e-07,
-      "loss": 0.0008,
+      "epoch": 2.7968,
+      "grad_norm": 0.04480099200664115,
+      "learning_rate": 2.398532640446161e-07,
+      "loss": 0.0002,
       "step": 1748
     },
     {
-      "epoch": 2.6182634730538923,
-      "grad_norm": 0.04305673654571666,
-      "learning_rate": 8.444900814040258e-07,
-      "loss": 0.0002,
+      "epoch": 2.7984,
+      "grad_norm": 0.0059019930564683494,
+      "learning_rate": 2.3610579436392999e-07,
+      "loss": 0.0,
       "step": 1749
     },
     {
-      "epoch": 2.6197604790419162,
-      "grad_norm": 0.0866720060609538,
-      "learning_rate": 8.379989402195255e-07,
-      "loss": 0.0004,
+      "epoch": 2.8,
+      "grad_norm": 0.0012335086772580393,
+      "learning_rate": 2.3238748115339327e-07,
+      "loss": 0.0,
       "step": 1750
     },
     {
-      "epoch": 2.62125748502994,
-      "grad_norm": 0.04999929716635118,
-      "learning_rate": 8.315317511786547e-07,
-      "loss": 0.0003,
+      "epoch": 2.8016,
+      "grad_norm": 0.0818394652065608,
+      "learning_rate": 2.2869833551645293e-07,
+      "loss": 0.0004,
       "step": 1751
     },
     {
-      "epoch": 2.622754491017964,
-      "grad_norm": 0.11556476223978758,
-      "learning_rate": 8.250885311885315e-07,
-      "loss": 0.0004,
+      "epoch": 2.8032,
+      "grad_norm": 0.022323579639708446,
+      "learning_rate": 2.2503836846945792e-07,
+      "loss": 0.0001,
       "step": 1752
     },
     {
-      "epoch": 2.624251497005988,
-      "grad_norm": 0.138338003099327,
-      "learning_rate": 8.186692970936127e-07,
-      "loss": 0.0008,
+      "epoch": 2.8048,
+      "grad_norm": 0.017704204237551684,
+      "learning_rate": 2.2140759094162468e-07,
+      "loss": 0.0001,
       "step": 1753
     },
     {
-      "epoch": 2.625748502994012,
-      "grad_norm": 0.09873045242936929,
-      "learning_rate": 8.1227406567565e-07,
-      "loss": 0.0005,
+      "epoch": 2.8064,
+      "grad_norm": 0.07857249008934615,
+      "learning_rate": 2.178060137750071e-07,
+      "loss": 0.0003,
       "step": 1754
     },
     {
-      "epoch": 2.627245508982036,
-      "grad_norm": 0.07887672568470554,
-      "learning_rate": 8.059028536536407e-07,
-      "loss": 0.0006,
+      "epoch": 2.808,
+      "grad_norm": 0.0017730005725699683,
+      "learning_rate": 2.1423364772445886e-07,
+      "loss": 0.0,
       "step": 1755
     },
     {
-      "epoch": 2.62874251497006,
-      "grad_norm": 0.07738930354523124,
-      "learning_rate": 7.995556776837964e-07,
-      "loss": 0.0005,
+      "epoch": 2.8096,
+      "grad_norm": 0.008609459644754248,
+      "learning_rate": 2.106905034576112e-07,
+      "loss": 0.0001,
       "step": 1756
     },
     {
-      "epoch": 2.6302395209580838,
-      "grad_norm": 0.035685651132636415,
-      "learning_rate": 7.932325543594821e-07,
-      "loss": 0.0002,
+      "epoch": 2.8112,
+      "grad_norm": 0.00913059018052819,
+      "learning_rate": 2.071765915548274e-07,
+      "loss": 0.0001,
       "step": 1757
     },
     {
-      "epoch": 2.6317365269461077,
-      "grad_norm": 0.11039538575586764,
-      "learning_rate": 7.86933500211191e-07,
-      "loss": 0.001,
+      "epoch": 2.8128,
+      "grad_norm": 0.003241407222481324,
+      "learning_rate": 2.036919225091827e-07,
+      "loss": 0.0,
       "step": 1758
     },
     {
-      "epoch": 2.6332335329341316,
-      "grad_norm": 0.049192018098694695,
-      "learning_rate": 7.806585317064852e-07,
-      "loss": 0.0004,
+      "epoch": 2.8144,
+      "grad_norm": 0.0014824563661412647,
+      "learning_rate": 2.002365067264289e-07,
+      "loss": 0.0,
       "step": 1759
     },
     {
-      "epoch": 2.6347305389221556,
-      "grad_norm": 0.1291838197111577,
-      "learning_rate": 7.744076652499655e-07,
-      "loss": 0.0005,
+      "epoch": 2.816,
+      "grad_norm": 0.002193225007399062,
+      "learning_rate": 1.9681035452496112e-07,
+      "loss": 0.0,
       "step": 1760
     },
     {
-      "epoch": 2.6362275449101795,
-      "grad_norm": 0.08020464788569276,
-      "learning_rate": 7.681809171832177e-07,
-      "loss": 0.0009,
+      "epoch": 2.8176,
+      "grad_norm": 0.016277810063011915,
+      "learning_rate": 1.9341347613579086e-07,
+      "loss": 0.0001,
       "step": 1761
     },
     {
-      "epoch": 2.6377245508982035,
-      "grad_norm": 0.1341437369451868,
-      "learning_rate": 7.619783037847783e-07,
-      "loss": 0.001,
+      "epoch": 2.8192,
+      "grad_norm": 0.0030854031883880886,
+      "learning_rate": 1.900458817025097e-07,
+      "loss": 0.0,
       "step": 1762
     },
     {
-      "epoch": 2.6392215568862274,
-      "grad_norm": 0.09583181308733467,
-      "learning_rate": 7.557998412700929e-07,
-      "loss": 0.0006,
+      "epoch": 2.8208,
+      "grad_norm": 0.08264368069386306,
+      "learning_rate": 1.867075812812691e-07,
+      "loss": 0.0002,
       "step": 1763
     },
     {
-      "epoch": 2.6407185628742518,
-      "grad_norm": 0.06724799411288579,
-      "learning_rate": 7.496455457914576e-07,
-      "loss": 0.0006,
+      "epoch": 2.8224,
+      "grad_norm": 0.03472191587129353,
+      "learning_rate": 1.8339858484073935e-07,
+      "loss": 0.0002,
       "step": 1764
     },
     {
-      "epoch": 2.6422155688622757,
-      "grad_norm": 0.009341167893632577,
-      "learning_rate": 7.435154334380001e-07,
+      "epoch": 2.824,
+      "grad_norm": 0.005196227636129316,
+      "learning_rate": 1.8011890226208527e-07,
       "loss": 0.0001,
       "step": 1765
     },
     {
-      "epoch": 2.6437125748502996,
-      "grad_norm": 0.005557298748028384,
-      "learning_rate": 7.374095202356235e-07,
-      "loss": 0.0001,
+      "epoch": 2.8256,
+      "grad_norm": 0.001364559398626262,
+      "learning_rate": 1.7686854333893833e-07,
+      "loss": 0.0,
       "step": 1766
     },
     {
-      "epoch": 2.6452095808383236,
-      "grad_norm": 0.06652628053372737,
-      "learning_rate": 7.313278221469621e-07,
-      "loss": 0.0004,
+      "epoch": 2.8272,
+      "grad_norm": 0.004988283718200237,
+      "learning_rate": 1.7364751777736334e-07,
+      "loss": 0.0001,
       "step": 1767
     },
     {
-      "epoch": 2.6467065868263475,
-      "grad_norm": 0.054699524495385746,
-      "learning_rate": 7.252703550713557e-07,
-      "loss": 0.0003,
+      "epoch": 2.8288,
+      "grad_norm": 0.016502732136393267,
+      "learning_rate": 1.7045583519583075e-07,
+      "loss": 0.0001,
       "step": 1768
     },
     {
-      "epoch": 2.6482035928143715,
-      "grad_norm": 0.03120350774247115,
-      "learning_rate": 7.192371348447835e-07,
-      "loss": 0.0002,
+      "epoch": 2.8304,
+      "grad_norm": 0.018912545591399193,
+      "learning_rate": 1.6729350512519006e-07,
+      "loss": 0.0001,
       "step": 1769
     },
     {
-      "epoch": 2.6497005988023954,
-      "grad_norm": 0.04231080059683171,
-      "learning_rate": 7.132281772398486e-07,
-      "loss": 0.0002,
+      "epoch": 2.832,
+      "grad_norm": 0.0019077932914240384,
+      "learning_rate": 1.6416053700863965e-07,
+      "loss": 0.0,
       "step": 1770
     },
     {
-      "epoch": 2.6511976047904193,
-      "grad_norm": 0.2930775130526807,
-      "learning_rate": 7.072434979657139e-07,
-      "loss": 0.0008,
+      "epoch": 2.8336,
+      "grad_norm": 0.0014420267763700873,
+      "learning_rate": 1.6105694020169594e-07,
+      "loss": 0.0,
       "step": 1771
     },
     {
-      "epoch": 2.6526946107784433,
-      "grad_norm": 0.06888790537173271,
-      "learning_rate": 7.012831126680808e-07,
-      "loss": 0.0006,
+      "epoch": 2.8352,
+      "grad_norm": 0.00361458803357153,
+      "learning_rate": 1.5798272397217097e-07,
+      "loss": 0.0001,
       "step": 1772
     },
     {
-      "epoch": 2.654191616766467,
-      "grad_norm": 0.10410134244700359,
-      "learning_rate": 6.953470369291349e-07,
-      "loss": 0.0008,
+      "epoch": 2.8368,
+      "grad_norm": 0.0036026081389549636,
+      "learning_rate": 1.5493789750014032e-07,
+      "loss": 0.0001,
       "step": 1773
     },
     {
-      "epoch": 2.655688622754491,
-      "grad_norm": 0.03504060316417563,
-      "learning_rate": 6.894352862675069e-07,
-      "loss": 0.0003,
+      "epoch": 2.8384,
+      "grad_norm": 0.014148198586583473,
+      "learning_rate": 1.519224698779198e-07,
+      "loss": 0.0001,
       "step": 1774
     },
     {
-      "epoch": 2.657185628742515,
-      "grad_norm": 0.07689984954156745,
-      "learning_rate": 6.835478761382442e-07,
-      "loss": 0.0005,
+      "epoch": 2.84,
+      "grad_norm": 0.0033210044607744184,
+      "learning_rate": 1.489364501100332e-07,
+      "loss": 0.0001,
       "step": 1775
     },
     {
-      "epoch": 2.658682634730539,
-      "grad_norm": 0.11005571108502613,
-      "learning_rate": 6.776848219327469e-07,
-      "loss": 0.0004,
+      "epoch": 2.8416,
+      "grad_norm": 0.0028672151551928813,
+      "learning_rate": 1.459798471131868e-07,
+      "loss": 0.0,
       "step": 1776
     },
     {
-      "epoch": 2.660179640718563,
-      "grad_norm": 0.06839160088182221,
-      "learning_rate": 6.718461389787534e-07,
-      "loss": 0.0005,
+      "epoch": 2.8432,
+      "grad_norm": 0.003315190926113845,
+      "learning_rate": 1.430526697162482e-07,
+      "loss": 0.0,
       "step": 1777
     },
     {
-      "epoch": 2.661676646706587,
-      "grad_norm": 0.1288662945687303,
-      "learning_rate": 6.660318425402846e-07,
-      "loss": 0.0011,
+      "epoch": 2.8448,
+      "grad_norm": 0.01717408122754632,
+      "learning_rate": 1.4015492666021313e-07,
+      "loss": 0.0001,
       "step": 1778
     },
     {
-      "epoch": 2.663173652694611,
-      "grad_norm": 0.07138445232695517,
-      "learning_rate": 6.602419478176037e-07,
-      "loss": 0.0005,
+      "epoch": 2.8464,
+      "grad_norm": 0.01088849960193917,
+      "learning_rate": 1.3728662659818205e-07,
+      "loss": 0.0001,
       "step": 1779
     },
     {
-      "epoch": 2.6646706586826348,
-      "grad_norm": 0.014327931523637798,
-      "learning_rate": 6.544764699471906e-07,
-      "loss": 0.0001,
+      "epoch": 2.848,
+      "grad_norm": 0.1375883157992815,
+      "learning_rate": 1.344477780953346e-07,
+      "loss": 0.001,
       "step": 1780
     },
     {
-      "epoch": 2.6661676646706587,
-      "grad_norm": 0.27063193324926954,
-      "learning_rate": 6.487354240016807e-07,
-      "loss": 0.0014,
+      "epoch": 2.8496,
+      "grad_norm": 0.007703529092860496,
+      "learning_rate": 1.3163838962890196e-07,
+      "loss": 0.0001,
       "step": 1781
     },
     {
-      "epoch": 2.6676646706586826,
-      "grad_norm": 0.023863898462553983,
-      "learning_rate": 6.430188249898473e-07,
-      "loss": 0.0002,
+      "epoch": 2.8512,
+      "grad_norm": 0.001354706538486727,
+      "learning_rate": 1.2885846958814673e-07,
+      "loss": 0.0,
       "step": 1782
     },
     {
-      "epoch": 2.6691616766467066,
-      "grad_norm": 0.031269105141780405,
-      "learning_rate": 6.373266878565454e-07,
-      "loss": 0.0002,
+      "epoch": 2.8528000000000002,
+      "grad_norm": 0.00390980837126583,
+      "learning_rate": 1.2610802627432972e-07,
+      "loss": 0.0,
       "step": 1783
     },
     {
-      "epoch": 2.6706586826347305,
-      "grad_norm": 0.01870608880096565,
-      "learning_rate": 6.316590274826828e-07,
-      "loss": 0.0002,
+      "epoch": 2.8544,
+      "grad_norm": 0.0018654410421585208,
+      "learning_rate": 1.2338706790069433e-07,
+      "loss": 0.0,
       "step": 1784
     },
     {
-      "epoch": 2.6721556886227544,
-      "grad_norm": 0.07554082317016642,
-      "learning_rate": 6.260158586851761e-07,
-      "loss": 0.0004,
+      "epoch": 2.856,
+      "grad_norm": 0.013021255953897921,
+      "learning_rate": 1.206956025924333e-07,
+      "loss": 0.0001,
       "step": 1785
     },
     {
-      "epoch": 2.6736526946107784,
-      "grad_norm": 0.031568014750176135,
-      "learning_rate": 6.203971962169153e-07,
-      "loss": 0.0002,
+      "epoch": 2.8576,
+      "grad_norm": 0.0018423331706124276,
+      "learning_rate": 1.1803363838667092e-07,
+      "loss": 0.0,
       "step": 1786
     },
     {
-      "epoch": 2.6751497005988023,
-      "grad_norm": 0.07930375079821568,
-      "learning_rate": 6.148030547667272e-07,
-      "loss": 0.0005,
+      "epoch": 2.8592,
+      "grad_norm": 0.0026056273680482773,
+      "learning_rate": 1.1540118323243866e-07,
+      "loss": 0.0,
       "step": 1787
     },
     {
-      "epoch": 2.6766467065868262,
-      "grad_norm": 0.03552785332381245,
-      "learning_rate": 6.092334489593233e-07,
-      "loss": 0.0002,
+      "epoch": 2.8608000000000002,
+      "grad_norm": 0.07989533085392884,
+      "learning_rate": 1.1279824499064396e-07,
+      "loss": 0.0013,
       "step": 1788
     },
     {
-      "epoch": 2.67814371257485,
-      "grad_norm": 0.031851214767568116,
-      "learning_rate": 6.036883933552806e-07,
-      "loss": 0.0002,
+      "epoch": 2.8624,
+      "grad_norm": 0.0018641526095015037,
+      "learning_rate": 1.1022483143405705e-07,
+      "loss": 0.0,
       "step": 1789
     },
     {
-      "epoch": 2.679640718562874,
-      "grad_norm": 0.04962775237902228,
-      "learning_rate": 5.981679024509935e-07,
-      "loss": 0.0002,
+      "epoch": 2.864,
+      "grad_norm": 0.009751574403888156,
+      "learning_rate": 1.0768095024728309e-07,
+      "loss": 0.0001,
       "step": 1790
     },
     {
-      "epoch": 2.681137724550898,
-      "grad_norm": 0.04407787101186428,
-      "learning_rate": 5.926719906786348e-07,
-      "loss": 0.0004,
+      "epoch": 2.8656,
+      "grad_norm": 0.007089527501608644,
+      "learning_rate": 1.0516660902673448e-07,
+      "loss": 0.0001,
       "step": 1791
     },
     {
-      "epoch": 2.682634730538922,
-      "grad_norm": 0.07555818249783224,
-      "learning_rate": 5.872006724061208e-07,
-      "loss": 0.0016,
+      "epoch": 2.8672,
+      "grad_norm": 0.006357592150038937,
+      "learning_rate": 1.0268181528061749e-07,
+      "loss": 0.0,
       "step": 1792
     },
     {
-      "epoch": 2.684131736526946,
-      "grad_norm": 0.02537315182664898,
-      "learning_rate": 5.817539619370727e-07,
-      "loss": 0.0002,
+      "epoch": 2.8688000000000002,
+      "grad_norm": 0.030832002709942744,
+      "learning_rate": 1.0022657642890232e-07,
+      "loss": 0.0001,
       "step": 1793
     },
     {
-      "epoch": 2.68562874251497,
-      "grad_norm": 0.018001211033930535,
-      "learning_rate": 5.763318735107826e-07,
+      "epoch": 2.8704,
+      "grad_norm": 0.002785122491385037,
+      "learning_rate": 9.780089980330643e-08,
       "loss": 0.0001,
       "step": 1794
     },
     {
-      "epoch": 2.687125748502994,
-      "grad_norm": 0.053445457775856076,
-      "learning_rate": 5.709344213021706e-07,
-      "loss": 0.0005,
+      "epoch": 2.872,
+      "grad_norm": 0.008951434154069501,
+      "learning_rate": 9.540479264726676e-08,
+      "loss": 0.0001,
       "step": 1795
     },
     {
-      "epoch": 2.6886227544910177,
-      "grad_norm": 0.05535585142367004,
-      "learning_rate": 5.655616194217517e-07,
-      "loss": 0.0003,
+      "epoch": 2.8736,
+      "grad_norm": 0.0010888299635037437,
+      "learning_rate": 9.303826211592316e-08,
+      "loss": 0.0,
       "step": 1796
     },
     {
-      "epoch": 2.6901197604790417,
-      "grad_norm": 0.04892377036746475,
-      "learning_rate": 5.602134819155969e-07,
-      "loss": 0.0003,
+      "epoch": 2.8752,
+      "grad_norm": 0.006382911098808215,
+      "learning_rate": 9.070131527609604e-08,
+      "loss": 0.0001,
       "step": 1797
     },
     {
-      "epoch": 2.6916167664670656,
-      "grad_norm": 0.020075394906988262,
-      "learning_rate": 5.548900227652987e-07,
+      "epoch": 2.8768000000000002,
+      "grad_norm": 0.02174671615525393,
+      "learning_rate": 8.839395910626214e-08,
       "loss": 0.0001,
       "step": 1798
     },
     {
-      "epoch": 2.69311377245509,
-      "grad_norm": 0.07645566349863471,
-      "learning_rate": 5.495912558879312e-07,
-      "loss": 0.0003,
+      "epoch": 2.8784,
+      "grad_norm": 0.003927400796810354,
+      "learning_rate": 8.61162004965388e-08,
+      "loss": 0.0,
       "step": 1799
     },
     {
-      "epoch": 2.694610778443114,
-      "grad_norm": 0.05819348736751214,
-      "learning_rate": 5.44317195136016e-07,
-      "loss": 0.0003,
+      "epoch": 2.88,
+      "grad_norm": 0.0014568560469440488,
+      "learning_rate": 8.386804624865851e-08,
+      "loss": 0.0,
       "step": 1800
     },
     {
-      "epoch": 2.696107784431138,
-      "grad_norm": 0.06457619663171396,
-      "learning_rate": 5.390678542974903e-07,
-      "loss": 0.0006,
+      "epoch": 2.8816,
+      "grad_norm": 0.003110881397941878,
+      "learning_rate": 8.16495030759501e-08,
+      "loss": 0.0,
       "step": 1801
     },
     {
-      "epoch": 2.697604790419162,
-      "grad_norm": 0.05017770166227597,
-      "learning_rate": 5.33843247095659e-07,
-      "loss": 0.0004,
+      "epoch": 2.8832,
+      "grad_norm": 0.005511663709665823,
+      "learning_rate": 7.946057760332193e-08,
+      "loss": 0.0001,
       "step": 1802
     },
     {
-      "epoch": 2.6991017964071857,
-      "grad_norm": 0.06428649128731666,
-      "learning_rate": 5.286433871891706e-07,
-      "loss": 0.0003,
+      "epoch": 2.8848000000000003,
+      "grad_norm": 0.007081682196621321,
+      "learning_rate": 7.730127636723539e-08,
+      "loss": 0.0001,
       "step": 1803
     },
     {
-      "epoch": 2.7005988023952097,
-      "grad_norm": 0.021010239698939993,
-      "learning_rate": 5.234682881719766e-07,
-      "loss": 0.0002,
+      "epoch": 2.8864,
+      "grad_norm": 0.0077041810174473075,
+      "learning_rate": 7.517160581569371e-08,
+      "loss": 0.0001,
       "step": 1804
     },
     {
-      "epoch": 2.7020958083832336,
-      "grad_norm": 0.03672140143952641,
-      "learning_rate": 5.183179635732937e-07,
-      "loss": 0.0002,
+      "epoch": 2.888,
+      "grad_norm": 0.006456232346309572,
+      "learning_rate": 7.307157230821426e-08,
+      "loss": 0.0001,
       "step": 1805
     },
     {
-      "epoch": 2.7035928143712575,
-      "grad_norm": 0.024124471357117365,
-      "learning_rate": 5.131924268575739e-07,
-      "loss": 0.0001,
+      "epoch": 2.8895999999999997,
+      "grad_norm": 0.034445837063582806,
+      "learning_rate": 7.100118211581852e-08,
+      "loss": 0.0002,
       "step": 1806
     },
     {
-      "epoch": 2.7050898203592815,
-      "grad_norm": 0.08077960062382916,
-      "learning_rate": 5.080916914244671e-07,
-      "loss": 0.0008,
+      "epoch": 2.8912,
+      "grad_norm": 0.0017820580940200157,
+      "learning_rate": 6.896044142100433e-08,
+      "loss": 0.0,
       "step": 1807
     },
     {
-      "epoch": 2.7065868263473054,
-      "grad_norm": 0.10237393145848973,
-      "learning_rate": 5.030157706087812e-07,
-      "loss": 0.0006,
+      "epoch": 2.8928000000000003,
+      "grad_norm": 0.06907803624791439,
+      "learning_rate": 6.694935631773259e-08,
+      "loss": 0.0004,
       "step": 1808
     },
     {
-      "epoch": 2.7080838323353293,
-      "grad_norm": 0.07472554370165986,
-      "learning_rate": 4.979646776804559e-07,
-      "loss": 0.0002,
+      "epoch": 2.8944,
+      "grad_norm": 0.005581726166082013,
+      "learning_rate": 6.496793281141056e-08,
+      "loss": 0.0001,
       "step": 1809
     },
     {
-      "epoch": 2.7095808383233533,
-      "grad_norm": 0.05873174733684072,
-      "learning_rate": 4.929384258445213e-07,
-      "loss": 0.0003,
+      "epoch": 2.896,
+      "grad_norm": 0.003311865968052793,
+      "learning_rate": 6.301617681886863e-08,
+      "loss": 0.0001,
       "step": 1810
     },
     {
-      "epoch": 2.711077844311377,
-      "grad_norm": 0.014282050561278747,
-      "learning_rate": 4.879370282410633e-07,
-      "loss": 0.0001,
+      "epoch": 2.8975999999999997,
+      "grad_norm": 0.0015380255213800762,
+      "learning_rate": 6.109409416834689e-08,
+      "loss": 0.0,
       "step": 1811
     },
     {
-      "epoch": 2.712574850299401,
-      "grad_norm": 0.06788628860680303,
-      "learning_rate": 4.829604979451997e-07,
-      "loss": 0.0003,
+      "epoch": 2.8992,
+      "grad_norm": 0.10057737534756946,
+      "learning_rate": 5.920169059947412e-08,
+      "loss": 0.0006,
       "step": 1812
     },
     {
-      "epoch": 2.714071856287425,
-      "grad_norm": 0.12438103093166794,
-      "learning_rate": 4.780088479670275e-07,
-      "loss": 0.0006,
+      "epoch": 2.9008000000000003,
+      "grad_norm": 0.04435905365195514,
+      "learning_rate": 5.7338971763256646e-08,
+      "loss": 0.0001,
       "step": 1813
     },
     {
-      "epoch": 2.715568862275449,
-      "grad_norm": 0.10531200528785536,
-      "learning_rate": 4.7308209125160654e-07,
-      "loss": 0.0004,
+      "epoch": 2.9024,
+      "grad_norm": 0.0919263972543488,
+      "learning_rate": 5.5505943222055046e-08,
+      "loss": 0.0005,
       "step": 1814
     },
     {
-      "epoch": 2.717065868263473,
-      "grad_norm": 0.1130560374315321,
-      "learning_rate": 4.68180240678916e-07,
-      "loss": 0.0009,
+      "epoch": 2.904,
+      "grad_norm": 0.007016794466427644,
+      "learning_rate": 5.37026104495697e-08,
+      "loss": 0.0001,
       "step": 1815
     },
     {
-      "epoch": 2.718562874251497,
-      "grad_norm": 0.02886495639077637,
-      "learning_rate": 4.6330330906382394e-07,
-      "loss": 0.0002,
+      "epoch": 2.9055999999999997,
+      "grad_norm": 0.004893016628908267,
+      "learning_rate": 5.192897883082748e-08,
+      "loss": 0.0001,
       "step": 1816
     },
     {
-      "epoch": 2.720059880239521,
-      "grad_norm": 0.02889119584742746,
-      "learning_rate": 4.5845130915605165e-07,
-      "loss": 0.0002,
+      "epoch": 2.9072,
+      "grad_norm": 0.004737890190966583,
+      "learning_rate": 5.0185053662161756e-08,
+      "loss": 0.0001,
       "step": 1817
     },
     {
-      "epoch": 2.7215568862275448,
-      "grad_norm": 0.03501511903364185,
-      "learning_rate": 4.536242536401403e-07,
-      "loss": 0.0002,
+      "epoch": 2.9088000000000003,
+      "grad_norm": 0.0022874384419645805,
+      "learning_rate": 4.8470840151195745e-08,
+      "loss": 0.0,
       "step": 1818
     },
     {
-      "epoch": 2.7230538922155687,
-      "grad_norm": 0.08233337203210803,
-      "learning_rate": 4.488221551354266e-07,
-      "loss": 0.0005,
+      "epoch": 2.9104,
+      "grad_norm": 0.0021399578018113266,
+      "learning_rate": 4.678634341683252e-08,
+      "loss": 0.0,
       "step": 1819
     },
     {
-      "epoch": 2.724550898203593,
-      "grad_norm": 0.056588559658037704,
-      "learning_rate": 4.440450261959905e-07,
-      "loss": 0.0006,
+      "epoch": 2.912,
+      "grad_norm": 0.005709094685496079,
+      "learning_rate": 4.513156848923616e-08,
+      "loss": 0.0001,
       "step": 1820
     },
     {
-      "epoch": 2.726047904191617,
-      "grad_norm": 0.07016765453031978,
-      "learning_rate": 4.3929287931064544e-07,
-      "loss": 0.0003,
+      "epoch": 2.9135999999999997,
+      "grad_norm": 0.008577173987016086,
+      "learning_rate": 4.350652030981395e-08,
+      "loss": 0.0001,
       "step": 1821
     },
     {
-      "epoch": 2.727544910179641,
-      "grad_norm": 0.0983340591609416,
-      "learning_rate": 4.34565726902888e-07,
-      "loss": 0.0006,
+      "epoch": 2.9152,
+      "grad_norm": 0.027217531196427575,
+      "learning_rate": 4.19112037312075e-08,
+      "loss": 0.0001,
       "step": 1822
     },
     {
-      "epoch": 2.729041916167665,
-      "grad_norm": 0.012487690080718029,
-      "learning_rate": 4.298635813308727e-07,
-      "loss": 0.0001,
+      "epoch": 2.9168,
+      "grad_norm": 0.000788668357959617,
+      "learning_rate": 4.0345623517273894e-08,
+      "loss": 0.0,
       "step": 1823
     },
     {
-      "epoch": 2.730538922155689,
-      "grad_norm": 0.11737318912857932,
-      "learning_rate": 4.2518645488738296e-07,
-      "loss": 0.0012,
+      "epoch": 2.9184,
+      "grad_norm": 0.01309458165842595,
+      "learning_rate": 3.8809784343072364e-08,
+      "loss": 0.0001,
       "step": 1824
     },
     {
-      "epoch": 2.7320359281437128,
-      "grad_norm": 0.059519890359665956,
-      "learning_rate": 4.2053435979978794e-07,
-      "loss": 0.0004,
+      "epoch": 2.92,
+      "grad_norm": 0.002016117446819547,
+      "learning_rate": 3.7303690794854296e-08,
+      "loss": 0.0,
       "step": 1825
     },
     {
-      "epoch": 2.7335329341317367,
-      "grad_norm": 0.09479516367922022,
-      "learning_rate": 4.1590730823002466e-07,
-      "loss": 0.0005,
+      "epoch": 2.9215999999999998,
+      "grad_norm": 0.033086944748702746,
+      "learning_rate": 3.582734737004101e-08,
+      "loss": 0.0001,
       "step": 1826
     },
     {
-      "epoch": 2.7350299401197606,
-      "grad_norm": 0.03731826461903811,
-      "learning_rate": 4.113053122745536e-07,
-      "loss": 0.0003,
+      "epoch": 2.9232,
+      "grad_norm": 0.0013975054108561335,
+      "learning_rate": 3.438075847721933e-08,
+      "loss": 0.0,
       "step": 1827
     },
     {
-      "epoch": 2.7365269461077846,
-      "grad_norm": 0.08798270041960435,
-      "learning_rate": 4.0672838396433436e-07,
-      "loss": 0.0005,
+      "epoch": 2.9248,
+      "grad_norm": 0.005798801462413593,
+      "learning_rate": 3.2963928436122726e-08,
+      "loss": 0.0001,
       "step": 1828
     },
     {
-      "epoch": 2.7380239520958085,
-      "grad_norm": 0.12804962939354766,
-      "learning_rate": 4.0217653526479663e-07,
-      "loss": 0.0008,
+      "epoch": 2.9264,
+      "grad_norm": 0.00517366116128266,
+      "learning_rate": 3.157686147762129e-08,
+      "loss": 0.0001,
       "step": 1829
     },
     {
-      "epoch": 2.7395209580838324,
-      "grad_norm": 0.12923264052846514,
-      "learning_rate": 3.9764977807579484e-07,
-      "loss": 0.0007,
+      "epoch": 2.928,
+      "grad_norm": 0.0017879877759022927,
+      "learning_rate": 3.0219561743707326e-08,
+      "loss": 0.0,
       "step": 1830
     },
     {
-      "epoch": 2.7410179640718564,
-      "grad_norm": 0.05668518180462198,
-      "learning_rate": 3.931481242315993e-07,
-      "loss": 0.0003,
+      "epoch": 2.9295999999999998,
+      "grad_norm": 0.00835238037225453,
+      "learning_rate": 2.8892033287484245e-08,
+      "loss": 0.0001,
       "step": 1831
     },
     {
-      "epoch": 2.7425149700598803,
-      "grad_norm": 0.12235579377751511,
-      "learning_rate": 3.886715855008405e-07,
-      "loss": 0.0005,
+      "epoch": 2.9312,
+      "grad_norm": 0.005400086869686289,
+      "learning_rate": 2.7594280073152123e-08,
+      "loss": 0.0001,
       "step": 1832
     },
     {
-      "epoch": 2.7440119760479043,
-      "grad_norm": 0.03866968344424915,
-      "learning_rate": 3.842201735865003e-07,
-      "loss": 0.0003,
+      "epoch": 2.9328,
+      "grad_norm": 0.011053432246126335,
+      "learning_rate": 2.6326305976001054e-08,
+      "loss": 0.0001,
       "step": 1833
     },
     {
-      "epoch": 2.745508982035928,
-      "grad_norm": 0.11192895051075528,
-      "learning_rate": 3.797939001258677e-07,
-      "loss": 0.0005,
+      "epoch": 2.9344,
+      "grad_norm": 0.0013304345302983858,
+      "learning_rate": 2.508811478239226e-08,
+      "loss": 0.0,
       "step": 1834
     },
     {
-      "epoch": 2.747005988023952,
-      "grad_norm": 0.08805745160822599,
-      "learning_rate": 3.753927766905108e-07,
-      "loss": 0.0007,
+      "epoch": 2.936,
+      "grad_norm": 0.004423950780095136,
+      "learning_rate": 2.3879710189753657e-08,
+      "loss": 0.0,
       "step": 1835
     },
     {
-      "epoch": 2.748502994011976,
-      "grad_norm": 0.013429115927834216,
-      "learning_rate": 3.710168147862547e-07,
-      "loss": 0.0002,
+      "epoch": 2.9375999999999998,
+      "grad_norm": 0.010711142940809274,
+      "learning_rate": 2.2701095806565432e-08,
+      "loss": 0.0001,
       "step": 1836
     },
     {
-      "epoch": 2.75,
-      "grad_norm": 0.04732482701187792,
-      "learning_rate": 3.6666602585313404e-07,
-      "loss": 0.0003,
+      "epoch": 2.9392,
+      "grad_norm": 0.006095942388119627,
+      "learning_rate": 2.1552275152346702e-08,
+      "loss": 0.0001,
       "step": 1837
     },
     {
-      "epoch": 2.751497005988024,
-      "grad_norm": 0.14685436958806133,
-      "learning_rate": 3.6234042126538584e-07,
-      "loss": 0.0011,
+      "epoch": 2.9408,
+      "grad_norm": 0.024063537013872153,
+      "learning_rate": 2.0433251657653307e-08,
+      "loss": 0.0004,
       "step": 1838
     },
     {
-      "epoch": 2.752994011976048,
-      "grad_norm": 0.015481759689920974,
-      "learning_rate": 3.5804001233139983e-07,
-      "loss": 0.0001,
+      "epoch": 2.9424,
+      "grad_norm": 0.12475801443444066,
+      "learning_rate": 1.9344028664056715e-08,
+      "loss": 0.0002,
       "step": 1839
     },
     {
-      "epoch": 2.754491017964072,
-      "grad_norm": 0.09709654767571617,
-      "learning_rate": 3.5376481029369967e-07,
-      "loss": 0.0006,
+      "epoch": 2.944,
+      "grad_norm": 0.0013781151293487824,
+      "learning_rate": 1.8284609424142897e-08,
+      "loss": 0.0,
       "step": 1840
     },
     {
-      "epoch": 2.7559880239520957,
-      "grad_norm": 0.09473213466103927,
-      "learning_rate": 3.495148263289117e-07,
-      "loss": 0.0011,
+      "epoch": 2.9455999999999998,
+      "grad_norm": 0.011620312784639906,
+      "learning_rate": 1.7254997101500137e-08,
+      "loss": 0.0001,
       "step": 1841
     },
     {
-      "epoch": 2.7574850299401197,
-      "grad_norm": 0.029982568028086066,
-      "learning_rate": 3.452900715477314e-07,
-      "loss": 0.0002,
+      "epoch": 2.9472,
+      "grad_norm": 0.004925652464999688,
+      "learning_rate": 1.6255194770704586e-08,
+      "loss": 0.0,
       "step": 1842
     },
     {
-      "epoch": 2.7589820359281436,
-      "grad_norm": 0.037697726669431046,
-      "learning_rate": 3.4109055699490526e-07,
-      "loss": 0.0002,
+      "epoch": 2.9488,
+      "grad_norm": 0.01084143644638827,
+      "learning_rate": 1.528520541731915e-08,
+      "loss": 0.0001,
       "step": 1843
     },
     {
-      "epoch": 2.7604790419161676,
-      "grad_norm": 0.0178343543145911,
-      "learning_rate": 3.3691629364918323e-07,
+      "epoch": 2.9504,
+      "grad_norm": 0.006584911349587544,
+      "learning_rate": 1.4345031937879061e-08,
       "loss": 0.0001,
       "step": 1844
     },
     {
-      "epoch": 2.7619760479041915,
-      "grad_norm": 0.0868752163044962,
-      "learning_rate": 3.327672924233116e-07,
-      "loss": 0.0007,
+      "epoch": 2.952,
+      "grad_norm": 0.024693535314543748,
+      "learning_rate": 1.3434677139885222e-08,
+      "loss": 0.0001,
       "step": 1845
     },
     {
-      "epoch": 2.7634730538922154,
-      "grad_norm": 0.06282759355626048,
-      "learning_rate": 3.2864356416398957e-07,
-      "loss": 0.0003,
+      "epoch": 2.9536,
+      "grad_norm": 0.008233169511446503,
+      "learning_rate": 1.2554143741795311e-08,
+      "loss": 0.0001,
       "step": 1846
     },
     {
-      "epoch": 2.7649700598802394,
-      "grad_norm": 0.023830143819205972,
-      "learning_rate": 3.2454511965184453e-07,
-      "loss": 0.0001,
+      "epoch": 2.9552,
+      "grad_norm": 0.0012915978308971087,
+      "learning_rate": 1.170343437301491e-08,
+      "loss": 0.0,
       "step": 1847
     },
     {
-      "epoch": 2.7664670658682633,
-      "grad_norm": 0.14715683380007932,
-      "learning_rate": 3.204719696014069e-07,
-      "loss": 0.0005,
+      "epoch": 2.9568,
+      "grad_norm": 0.004358855559623065,
+      "learning_rate": 1.0882551573891953e-08,
+      "loss": 0.0001,
       "step": 1848
     },
     {
-      "epoch": 2.7679640718562872,
-      "grad_norm": 0.02531860275827708,
-      "learning_rate": 3.164241246610789e-07,
-      "loss": 0.0003,
+      "epoch": 2.9584,
+      "grad_norm": 0.03165947917388015,
+      "learning_rate": 1.0091497795706728e-08,
+      "loss": 0.0002,
       "step": 1849
     },
     {
-      "epoch": 2.769461077844311,
-      "grad_norm": 0.0726235081299919,
-      "learning_rate": 3.1240159541310764e-07,
-      "loss": 0.0007,
+      "epoch": 2.96,
+      "grad_norm": 0.004520262535504451,
+      "learning_rate": 9.330275400666334e-09,
+      "loss": 0.0,
       "step": 1850
     },
     {
-      "epoch": 2.770958083832335,
-      "grad_norm": 0.10837249282039897,
-      "learning_rate": 3.0840439237356025e-07,
-      "loss": 0.0013,
+      "epoch": 2.9616,
+      "grad_norm": 0.05056987291700106,
+      "learning_rate": 8.59888666189579e-09,
+      "loss": 0.0004,
       "step": 1851
     },
     {
-      "epoch": 2.772455089820359,
-      "grad_norm": 0.046374751147364626,
-      "learning_rate": 3.0443252599228865e-07,
-      "loss": 0.0004,
+      "epoch": 2.9632,
+      "grad_norm": 0.01759805043355931,
+      "learning_rate": 7.897333763433601e-09,
+      "loss": 0.0002,
       "step": 1852
     },
     {
-      "epoch": 2.773952095808383,
-      "grad_norm": 0.04955860777056503,
-      "learning_rate": 3.004860066529125e-07,
-      "loss": 0.0003,
+      "epoch": 2.9648,
+      "grad_norm": 0.0037291101477257473,
+      "learning_rate": 7.225618800222878e-09,
+      "loss": 0.0001,
       "step": 1853
     },
     {
-      "epoch": 2.775449101796407,
-      "grad_norm": 0.04175690267753176,
-      "learning_rate": 2.96564844672782e-07,
-      "loss": 0.0002,
+      "epoch": 2.9664,
+      "grad_norm": 0.0018585810182493614,
+      "learning_rate": 6.583743778106888e-09,
+      "loss": 0.0,
       "step": 1854
     },
     {
-      "epoch": 2.7769461077844313,
-      "grad_norm": 0.01101306053826792,
-      "learning_rate": 2.9266905030296164e-07,
-      "loss": 0.0001,
+      "epoch": 2.968,
+      "grad_norm": 0.09601546724932118,
+      "learning_rate": 5.971710613821291e-09,
+      "loss": 0.0003,
       "step": 1855
     },
     {
-      "epoch": 2.7784431137724552,
-      "grad_norm": 0.061584971883424364,
-      "learning_rate": 2.887986337281934e-07,
-      "loss": 0.0005,
+      "epoch": 2.9696,
+      "grad_norm": 0.0012696300863838544,
+      "learning_rate": 5.3895211349896946e-09,
+      "loss": 0.0,
       "step": 1856
     },
     {
-      "epoch": 2.779940119760479,
-      "grad_norm": 0.038380836869109265,
-      "learning_rate": 2.8495360506687444e-07,
-      "loss": 0.0002,
+      "epoch": 2.9712,
+      "grad_norm": 0.004006260952148794,
+      "learning_rate": 4.837177080119215e-09,
+      "loss": 0.0,
       "step": 1857
     },
     {
-      "epoch": 2.781437125748503,
-      "grad_norm": 0.036721797893085474,
-      "learning_rate": 2.811339743710328e-07,
+      "epoch": 2.9728,
+      "grad_norm": 0.00635367367348536,
+      "learning_rate": 4.314680098592705e-09,
       "loss": 0.0001,
       "step": 1858
     },
     {
-      "epoch": 2.782934131736527,
-      "grad_norm": 0.10482041789183853,
-      "learning_rate": 2.7733975162629655e-07,
-      "loss": 0.0004,
+      "epoch": 2.9744,
+      "grad_norm": 0.004998163224516518,
+      "learning_rate": 3.8220317506654226e-09,
+      "loss": 0.0001,
       "step": 1859
     },
     {
-      "epoch": 2.784431137724551,
-      "grad_norm": 0.06852021780595845,
-      "learning_rate": 2.735709467518699e-07,
-      "loss": 0.0003,
+      "epoch": 2.976,
+      "grad_norm": 0.0011077026332547931,
+      "learning_rate": 3.3592335074594805e-09,
+      "loss": 0.0,
       "step": 1860
     },
     {
-      "epoch": 2.785928143712575,
-      "grad_norm": 0.007301911146011518,
-      "learning_rate": 2.6982756960050925e-07,
+      "epoch": 2.9776,
+      "grad_norm": 0.0263071065239791,
+      "learning_rate": 2.9262867509605164e-09,
       "loss": 0.0001,
       "step": 1861
     },
     {
-      "epoch": 2.787425149700599,
-      "grad_norm": 0.0928237098781843,
-      "learning_rate": 2.661096299584953e-07,
-      "loss": 0.0005,
+      "epoch": 2.9792,
+      "grad_norm": 0.0038767720735516454,
+      "learning_rate": 2.5231927740154705e-09,
+      "loss": 0.0001,
       "step": 1862
     },
     {
-      "epoch": 2.788922155688623,
-      "grad_norm": 0.037133885536317905,
-      "learning_rate": 2.624171375456064e-07,
+      "epoch": 2.9808,
+      "grad_norm": 0.05274699982948644,
+      "learning_rate": 2.149952780321485e-09,
       "loss": 0.0002,
       "step": 1863
     },
     {
-      "epoch": 2.7904191616766467,
-      "grad_norm": 0.10006499255445611,
-      "learning_rate": 2.58750102015094e-07,
-      "loss": 0.0007,
+      "epoch": 2.9824,
+      "grad_norm": 0.02529985251576128,
+      "learning_rate": 1.8065678844314538e-09,
+      "loss": 0.0001,
       "step": 1864
     },
     {
-      "epoch": 2.7919161676646707,
-      "grad_norm": 0.04256690893240319,
-      "learning_rate": 2.5510853295366067e-07,
-      "loss": 0.0003,
+      "epoch": 2.984,
+      "grad_norm": 0.010248827513058067,
+      "learning_rate": 1.4930391117451427e-09,
+      "loss": 0.0001,
       "step": 1865
     },
     {
-      "epoch": 2.7934131736526946,
-      "grad_norm": 0.0684594948441174,
-      "learning_rate": 2.514924398814278e-07,
-      "loss": 0.0003,
+      "epoch": 2.9856,
+      "grad_norm": 0.029882135251723525,
+      "learning_rate": 1.209367398504746e-09,
+      "loss": 0.0001,
       "step": 1866
     },
     {
-      "epoch": 2.7949101796407185,
-      "grad_norm": 0.09626127387021133,
-      "learning_rate": 2.479018322519189e-07,
-      "loss": 0.0006,
+      "epoch": 2.9872,
+      "grad_norm": 0.0012465966402100829,
+      "learning_rate": 9.555535917993297e-10,
+      "loss": 0.0,
       "step": 1867
     },
     {
-      "epoch": 2.7964071856287425,
-      "grad_norm": 0.01741566020372036,
-      "learning_rate": 2.4433671945203076e-07,
+      "epoch": 2.9888,
+      "grad_norm": 0.004171632635330529,
+      "learning_rate": 7.315984495548378e-10,
       "loss": 0.0001,
       "step": 1868
     },
     {
-      "epoch": 2.7979041916167664,
-      "grad_norm": 0.07213965555140582,
-      "learning_rate": 2.4079711080200683e-07,
-      "loss": 0.0005,
+      "epoch": 2.9904,
+      "grad_norm": 0.004683951364801004,
+      "learning_rate": 5.375026405352035e-10,
+      "loss": 0.0,
       "step": 1869
     },
     {
-      "epoch": 2.7994011976047903,
-      "grad_norm": 0.06007506249091625,
-      "learning_rate": 2.3728301555541845e-07,
-      "loss": 0.0003,
+      "epoch": 2.992,
+      "grad_norm": 0.003909613062024481,
+      "learning_rate": 3.732667443390181e-10,
+      "loss": 0.0,
       "step": 1870
     },
     {
-      "epoch": 2.8008982035928143,
-      "grad_norm": 0.06125607539660623,
-      "learning_rate": 2.3379444289913344e-07,
-      "loss": 0.0003,
+      "epoch": 2.9936,
+      "grad_norm": 0.008748115478093845,
+      "learning_rate": 2.388912514017516e-10,
+      "loss": 0.0001,
       "step": 1871
     },
     {
-      "epoch": 2.802395209580838,
-      "grad_norm": 0.015232735958729315,
-      "learning_rate": 2.3033140195329985e-07,
-      "loss": 0.0001,
+      "epoch": 2.9952,
+      "grad_norm": 0.23714353606083233,
+      "learning_rate": 1.3437656298687096e-10,
+      "loss": 0.0009,
       "step": 1872
     },
     {
-      "epoch": 2.803892215568862,
-      "grad_norm": 0.03010125402384828,
-      "learning_rate": 2.2689390177131787e-07,
-      "loss": 0.0002,
+      "epoch": 2.9968,
+      "grad_norm": 0.001146936780494521,
+      "learning_rate": 5.972299119250124e-11,
+      "loss": 0.0,
       "step": 1873
     },
     {
-      "epoch": 2.805389221556886,
-      "grad_norm": 0.05324858893449288,
-      "learning_rate": 2.2348195133981344e-07,
-      "loss": 0.0003,
-      "step": 1874
-    },
-    {
-      "epoch": 2.80688622754491,
-      "grad_norm": 0.026758821808148425,
-      "learning_rate": 2.200955595786214e-07,
-      "loss": 0.0002,
-      "step": 1875
-    },
-    {
-      "epoch": 2.8083832335329344,
-      "grad_norm": 0.04605830316424714,
-      "learning_rate": 2.1673473534075895e-07,
-      "loss": 0.0004,
-      "step": 1876
-    },
-    {
-      "epoch": 2.8098802395209583,
-      "grad_norm": 0.04122472187922105,
-      "learning_rate": 2.1339948741240013e-07,
-      "loss": 0.0002,
-      "step": 1877
-    },
-    {
-      "epoch": 2.8113772455089823,
-      "grad_norm": 0.026915377348818586,
-      "learning_rate": 2.1008982451285687e-07,
-      "loss": 0.0003,
-      "step": 1878
-    },
-    {
-      "epoch": 2.812874251497006,
-      "grad_norm": 0.030265578451253612,
-      "learning_rate": 2.0680575529455017e-07,
-      "loss": 0.0004,
-      "step": 1879
-    },
-    {
-      "epoch": 2.81437125748503,
-      "grad_norm": 0.07654024267241427,
-      "learning_rate": 2.035472883430001e-07,
-      "loss": 0.0005,
-      "step": 1880
-    },
-    {
-      "epoch": 2.815868263473054,
-      "grad_norm": 0.02468323185102218,
-      "learning_rate": 2.0031443217678582e-07,
-      "loss": 0.0003,
-      "step": 1881
-    },
-    {
-      "epoch": 2.817365269461078,
-      "grad_norm": 0.14330870714264016,
-      "learning_rate": 1.9710719524753896e-07,
-      "loss": 0.0008,
-      "step": 1882
-    },
-    {
-      "epoch": 2.818862275449102,
-      "grad_norm": 0.06963422959188056,
-      "learning_rate": 1.939255859399114e-07,
-      "loss": 0.0004,
-      "step": 1883
-    },
-    {
-      "epoch": 2.820359281437126,
-      "grad_norm": 0.03669033456219445,
-      "learning_rate": 1.9076961257155745e-07,
-      "loss": 0.0002,
-      "step": 1884
-    },
-    {
-      "epoch": 2.82185628742515,
-      "grad_norm": 0.08034552402519764,
-      "learning_rate": 1.876392833931151e-07,
-      "loss": 0.0008,
-      "step": 1885
-    },
-    {
-      "epoch": 2.8233532934131738,
-      "grad_norm": 0.011926808859062795,
-      "learning_rate": 1.8453460658817367e-07,
-      "loss": 0.0001,
-      "step": 1886
-    },
-    {
-      "epoch": 2.8248502994011977,
-      "grad_norm": 0.07355376029482306,
-      "learning_rate": 1.814555902732673e-07,
-      "loss": 0.0006,
-      "step": 1887
-    },
-    {
-      "epoch": 2.8263473053892216,
-      "grad_norm": 0.029801616810496537,
-      "learning_rate": 1.7840224249784154e-07,
-      "loss": 0.0002,
-      "step": 1888
-    },
-    {
-      "epoch": 2.8278443113772456,
-      "grad_norm": 0.12698015721022915,
-      "learning_rate": 1.7537457124423896e-07,
-      "loss": 0.0004,
-      "step": 1889
-    },
-    {
-      "epoch": 2.8293413173652695,
-      "grad_norm": 0.06530833972448825,
-      "learning_rate": 1.723725844276747e-07,
-      "loss": 0.0004,
-      "step": 1890
-    },
-    {
-      "epoch": 2.8308383233532934,
-      "grad_norm": 0.12703576094207422,
-      "learning_rate": 1.693962898962187e-07,
-      "loss": 0.0005,
-      "step": 1891
-    },
-    {
-      "epoch": 2.8323353293413174,
-      "grad_norm": 0.06299450353431471,
-      "learning_rate": 1.664456954307736e-07,
-      "loss": 0.0004,
-      "step": 1892
-    },
-    {
-      "epoch": 2.8338323353293413,
-      "grad_norm": 0.0345823141248644,
-      "learning_rate": 1.635208087450524e-07,
-      "loss": 0.0002,
-      "step": 1893
-    },
-    {
-      "epoch": 2.8353293413173652,
-      "grad_norm": 0.07171038665684742,
-      "learning_rate": 1.6062163748556402e-07,
-      "loss": 0.0005,
-      "step": 1894
-    },
-    {
-      "epoch": 2.836826347305389,
-      "grad_norm": 0.040539395882641424,
-      "learning_rate": 1.5774818923158687e-07,
-      "loss": 0.0004,
-      "step": 1895
-    },
-    {
-      "epoch": 2.838323353293413,
-      "grad_norm": 0.01721308734367943,
-      "learning_rate": 1.5490047149515186e-07,
-      "loss": 0.0001,
-      "step": 1896
-    },
-    {
-      "epoch": 2.839820359281437,
-      "grad_norm": 0.03712780760613839,
-      "learning_rate": 1.5207849172102827e-07,
-      "loss": 0.0002,
-      "step": 1897
-    },
-    {
-      "epoch": 2.841317365269461,
-      "grad_norm": 0.017202811427165027,
-      "learning_rate": 1.492822572866892e-07,
-      "loss": 0.0001,
-      "step": 1898
-    },
-    {
-      "epoch": 2.842814371257485,
-      "grad_norm": 0.08062656343801933,
-      "learning_rate": 1.4651177550231043e-07,
-      "loss": 0.0004,
-      "step": 1899
-    },
-    {
-      "epoch": 2.844311377245509,
-      "grad_norm": 0.03769317002353988,
-      "learning_rate": 1.4376705361073828e-07,
-      "loss": 0.0004,
-      "step": 1900
-    },
-    {
-      "epoch": 2.845808383233533,
-      "grad_norm": 0.0595900137707606,
-      "learning_rate": 1.410480987874774e-07,
-      "loss": 0.0004,
-      "step": 1901
-    },
-    {
-      "epoch": 2.8473053892215567,
-      "grad_norm": 0.0456203631893289,
-      "learning_rate": 1.3835491814066847e-07,
-      "loss": 0.0003,
-      "step": 1902
-    },
-    {
-      "epoch": 2.8488023952095807,
-      "grad_norm": 0.06469454614345947,
-      "learning_rate": 1.3568751871107068e-07,
-      "loss": 0.0003,
-      "step": 1903
-    },
-    {
-      "epoch": 2.8502994011976046,
-      "grad_norm": 0.14575247322591645,
-      "learning_rate": 1.3304590747204582e-07,
-      "loss": 0.0018,
-      "step": 1904
-    },
-    {
-      "epoch": 2.8517964071856285,
-      "grad_norm": 0.09067521256579207,
-      "learning_rate": 1.3043009132953533e-07,
-      "loss": 0.0009,
-      "step": 1905
-    },
-    {
-      "epoch": 2.8532934131736525,
-      "grad_norm": 0.013493249897349183,
-      "learning_rate": 1.2784007712204562e-07,
-      "loss": 0.0001,
-      "step": 1906
-    },
-    {
-      "epoch": 2.8547904191616764,
-      "grad_norm": 0.11691368942179577,
-      "learning_rate": 1.2527587162062927e-07,
-      "loss": 0.0009,
-      "step": 1907
-    },
-    {
-      "epoch": 2.8562874251497004,
-      "grad_norm": 0.037850088975618104,
-      "learning_rate": 1.2273748152886844e-07,
-      "loss": 0.0003,
-      "step": 1908
-    },
-    {
-      "epoch": 2.8577844311377243,
-      "grad_norm": 0.1141808276161799,
-      "learning_rate": 1.2022491348285482e-07,
-      "loss": 0.0006,
-      "step": 1909
-    },
-    {
-      "epoch": 2.8592814371257482,
-      "grad_norm": 0.02925451773697013,
-      "learning_rate": 1.1773817405117405e-07,
-      "loss": 0.0002,
-      "step": 1910
-    },
-    {
-      "epoch": 2.8607784431137726,
-      "grad_norm": 0.05017111181234729,
-      "learning_rate": 1.1527726973488918e-07,
-      "loss": 0.0003,
-      "step": 1911
-    },
-    {
-      "epoch": 2.8622754491017965,
-      "grad_norm": 0.04552663871537052,
-      "learning_rate": 1.1284220696752168e-07,
+      "epoch": 2.9984,
+      "grad_norm": 0.04094181628426927,
+      "learning_rate": 1.4930758944764477e-11,
       "loss": 0.0002,
-      "step": 1912
-    },
-    {
-      "epoch": 2.8637724550898205,
-      "grad_norm": 0.1413296347951028,
-      "learning_rate": 1.1043299211503489e-07,
-      "loss": 0.0007,
-      "step": 1913
-    },
-    {
-      "epoch": 2.8652694610778444,
-      "grad_norm": 0.01983528978302256,
-      "learning_rate": 1.080496314758217e-07,
-      "loss": 0.0001,
-      "step": 1914
-    },
-    {
-      "epoch": 2.8667664670658684,
-      "grad_norm": 0.10833415515320645,
-      "learning_rate": 1.0569213128068024e-07,
-      "loss": 0.0003,
-      "step": 1915
-    },
-    {
-      "epoch": 2.8682634730538923,
-      "grad_norm": 0.04839502505058121,
-      "learning_rate": 1.0336049769280377e-07,
-      "loss": 0.0004,
-      "step": 1916
-    },
-    {
-      "epoch": 2.8697604790419162,
-      "grad_norm": 0.018562019071949903,
-      "learning_rate": 1.0105473680776301e-07,
-      "loss": 0.0001,
-      "step": 1917
+      "step": 1874
     },
     {
-      "epoch": 2.87125748502994,
-      "grad_norm": 0.029694138913736658,
-      "learning_rate": 9.877485465349057e-08,
+      "epoch": 3.0,
+      "grad_norm": 0.07295115851459594,
+      "learning_rate": 0.0,
       "loss": 0.0003,
-      "step": 1918
-    },
-    {
-      "epoch": 2.872754491017964,
-      "grad_norm": 0.04427064392263348,
-      "learning_rate": 9.652085719026205e-08,
-      "loss": 0.0005,
-      "step": 1919
-    },
-    {
-      "epoch": 2.874251497005988,
-      "grad_norm": 0.07693934381799596,
-      "learning_rate": 9.429275031068608e-08,
-      "loss": 0.0006,
-      "step": 1920
-    },
-    {
-      "epoch": 2.875748502994012,
-      "grad_norm": 0.09527796665446048,
-      "learning_rate": 9.209053983968209e-08,
-      "loss": 0.0007,
-      "step": 1921
-    },
-    {
-      "epoch": 2.877245508982036,
-      "grad_norm": 0.058776929014960504,
-      "learning_rate": 8.991423153447143e-08,
-      "loss": 0.0003,
-      "step": 1922
-    },
-    {
-      "epoch": 2.87874251497006,
-      "grad_norm": 0.03288997471143238,
-      "learning_rate": 8.776383108455966e-08,
-      "loss": 0.0002,
-      "step": 1923
-    },
-    {
-      "epoch": 2.8802395209580838,
-      "grad_norm": 0.05757102855593304,
-      "learning_rate": 8.563934411171981e-08,
-      "loss": 0.0002,
-      "step": 1924
-    },
-    {
-      "epoch": 2.8817365269461077,
-      "grad_norm": 0.0387920150897497,
-      "learning_rate": 8.354077616998247e-08,
-      "loss": 0.0003,
-      "step": 1925
-    },
-    {
-      "epoch": 2.8832335329341316,
-      "grad_norm": 0.1232602412566441,
-      "learning_rate": 8.14681327456146e-08,
-      "loss": 0.0008,
-      "step": 1926
-    },
-    {
-      "epoch": 2.8847305389221556,
-      "grad_norm": 0.09510268541356108,
-      "learning_rate": 7.942141925711078e-08,
-      "loss": 0.0008,
-      "step": 1927
-    },
-    {
-      "epoch": 2.8862275449101795,
-      "grad_norm": 0.034899399930684916,
-      "learning_rate": 7.740064105517753e-08,
-      "loss": 0.0002,
-      "step": 1928
-    },
-    {
-      "epoch": 2.8877245508982035,
-      "grad_norm": 0.08538587733278336,
-      "learning_rate": 7.540580342272008e-08,
-      "loss": 0.0005,
-      "step": 1929
-    },
-    {
-      "epoch": 2.8892215568862274,
-      "grad_norm": 0.036141222171045685,
-      "learning_rate": 7.343691157482347e-08,
-      "loss": 0.0003,
-      "step": 1930
-    },
-    {
-      "epoch": 2.8907185628742518,
-      "grad_norm": 0.048494060524911266,
-      "learning_rate": 7.149397065874697e-08,
-      "loss": 0.0004,
-      "step": 1931
-    },
-    {
-      "epoch": 2.8922155688622757,
-      "grad_norm": 0.04348773500793645,
-      "learning_rate": 6.957698575390415e-08,
-      "loss": 0.0003,
-      "step": 1932
-    },
-    {
-      "epoch": 2.8937125748502996,
-      "grad_norm": 0.019348415703767035,
-      "learning_rate": 6.768596187185394e-08,
-      "loss": 0.0001,
-      "step": 1933
-    },
-    {
-      "epoch": 2.8952095808383236,
-      "grad_norm": 0.07990531211236819,
-      "learning_rate": 6.582090395628294e-08,
-      "loss": 0.0005,
-      "step": 1934
-    },
-    {
-      "epoch": 2.8967065868263475,
-      "grad_norm": 0.028337210973887937,
-      "learning_rate": 6.398181688299753e-08,
-      "loss": 0.0003,
-      "step": 1935
-    },
-    {
-      "epoch": 2.8982035928143715,
-      "grad_norm": 0.029288080250697478,
-      "learning_rate": 6.21687054599085e-08,
-      "loss": 0.0001,
-      "step": 1936
-    },
-    {
-      "epoch": 2.8997005988023954,
-      "grad_norm": 0.0868655909482861,
-      "learning_rate": 6.038157442701642e-08,
-      "loss": 0.0004,
-      "step": 1937
-    },
-    {
-      "epoch": 2.9011976047904193,
-      "grad_norm": 0.07224510015715654,
-      "learning_rate": 5.862042845640403e-08,
-      "loss": 0.0005,
-      "step": 1938
-    },
-    {
-      "epoch": 2.9026946107784433,
-      "grad_norm": 0.014587726019568977,
-      "learning_rate": 5.688527215222062e-08,
-      "loss": 0.0001,
-      "step": 1939
-    },
-    {
-      "epoch": 2.904191616766467,
-      "grad_norm": 0.0913841204532074,
-      "learning_rate": 5.517611005067092e-08,
-      "loss": 0.0007,
-      "step": 1940
-    },
-    {
-      "epoch": 2.905688622754491,
-      "grad_norm": 0.02053119590322276,
-      "learning_rate": 5.3492946620005146e-08,
-      "loss": 0.0004,
-      "step": 1941
-    },
-    {
-      "epoch": 2.907185628742515,
-      "grad_norm": 0.048364628173414546,
-      "learning_rate": 5.183578626050012e-08,
-      "loss": 0.0003,
-      "step": 1942
-    },
-    {
-      "epoch": 2.908682634730539,
-      "grad_norm": 0.11064582121615582,
-      "learning_rate": 5.020463330445813e-08,
-      "loss": 0.0007,
-      "step": 1943
-    },
-    {
-      "epoch": 2.910179640718563,
-      "grad_norm": 0.02823565491289188,
-      "learning_rate": 4.8599492016189184e-08,
-      "loss": 0.0003,
-      "step": 1944
-    },
-    {
-      "epoch": 2.911676646706587,
-      "grad_norm": 0.08021570823602606,
-      "learning_rate": 4.7020366592001046e-08,
-      "loss": 0.0005,
-      "step": 1945
-    },
-    {
-      "epoch": 2.913173652694611,
-      "grad_norm": 0.06441990843786155,
-      "learning_rate": 4.546726116018696e-08,
-      "loss": 0.0004,
-      "step": 1946
-    },
-    {
-      "epoch": 2.9146706586826348,
-      "grad_norm": 0.07772901840599725,
-      "learning_rate": 4.394017978101905e-08,
-      "loss": 0.0004,
-      "step": 1947
-    },
-    {
-      "epoch": 2.9161676646706587,
-      "grad_norm": 0.012281701982242639,
-      "learning_rate": 4.243912644673276e-08,
-      "loss": 0.0002,
-      "step": 1948
-    },
-    {
-      "epoch": 2.9176646706586826,
-      "grad_norm": 0.0893888921305495,
-      "learning_rate": 4.0964105081519046e-08,
-      "loss": 0.0004,
-      "step": 1949
-    },
-    {
-      "epoch": 2.9191616766467066,
-      "grad_norm": 0.007291028241031858,
-      "learning_rate": 3.951511954151554e-08,
-      "loss": 0.0001,
-      "step": 1950
-    },
-    {
-      "epoch": 2.9206586826347305,
-      "grad_norm": 0.04531637743292414,
-      "learning_rate": 3.809217361479212e-08,
-      "loss": 0.0004,
-      "step": 1951
-    },
-    {
-      "epoch": 2.9221556886227544,
-      "grad_norm": 0.10016872974204939,
-      "learning_rate": 3.669527102134418e-08,
-      "loss": 0.0006,
-      "step": 1952
-    },
-    {
-      "epoch": 2.9236526946107784,
-      "grad_norm": 0.10136592233723844,
-      "learning_rate": 3.532441541308384e-08,
-      "loss": 0.0009,
-      "step": 1953
-    },
-    {
-      "epoch": 2.9251497005988023,
-      "grad_norm": 0.099063980397416,
-      "learning_rate": 3.397961037382547e-08,
-      "loss": 0.0007,
-      "step": 1954
-    },
-    {
-      "epoch": 2.9266467065868262,
-      "grad_norm": 0.0494457607619648,
-      "learning_rate": 3.266085941928454e-08,
-      "loss": 0.0004,
-      "step": 1955
-    },
-    {
-      "epoch": 2.92814371257485,
-      "grad_norm": 0.10780498117653974,
-      "learning_rate": 3.136816599705883e-08,
-      "loss": 0.0006,
-      "step": 1956
-    },
-    {
-      "epoch": 2.929640718562874,
-      "grad_norm": 0.0658842139801036,
-      "learning_rate": 3.0101533486626146e-08,
-      "loss": 0.0005,
-      "step": 1957
-    },
-    {
-      "epoch": 2.931137724550898,
-      "grad_norm": 0.04093767820581611,
-      "learning_rate": 2.8860965199334346e-08,
-      "loss": 0.0002,
-      "step": 1958
-    },
-    {
-      "epoch": 2.932634730538922,
-      "grad_norm": 0.05070774966783894,
-      "learning_rate": 2.7646464378390248e-08,
-      "loss": 0.0004,
-      "step": 1959
-    },
-    {
-      "epoch": 2.934131736526946,
-      "grad_norm": 0.01789433543858203,
-      "learning_rate": 2.6458034198852954e-08,
-      "loss": 0.0001,
-      "step": 1960
-    },
-    {
-      "epoch": 2.93562874251497,
-      "grad_norm": 0.04336770967449903,
-      "learning_rate": 2.5295677767627202e-08,
-      "loss": 0.0002,
-      "step": 1961
-    },
-    {
-      "epoch": 2.937125748502994,
-      "grad_norm": 0.0865926476879689,
-      "learning_rate": 2.415939812345003e-08,
-      "loss": 0.0003,
-      "step": 1962
-    },
-    {
-      "epoch": 2.9386227544910177,
-      "grad_norm": 0.09911546956818736,
-      "learning_rate": 2.304919823688856e-08,
-      "loss": 0.0005,
-      "step": 1963
-    },
-    {
-      "epoch": 2.9401197604790417,
-      "grad_norm": 0.07993502202450768,
-      "learning_rate": 2.1965081010328903e-08,
-      "loss": 0.0007,
-      "step": 1964
-    },
-    {
-      "epoch": 2.9416167664670656,
-      "grad_norm": 0.04287570725916258,
-      "learning_rate": 2.0907049277971714e-08,
-      "loss": 0.0002,
-      "step": 1965
-    },
-    {
-      "epoch": 2.94311377245509,
-      "grad_norm": 0.041199397205273956,
-      "learning_rate": 1.987510580581664e-08,
-      "loss": 0.0003,
-      "step": 1966
-    },
-    {
-      "epoch": 2.944610778443114,
-      "grad_norm": 0.12923583448830503,
-      "learning_rate": 1.8869253291669e-08,
-      "loss": 0.0007,
-      "step": 1967
-    },
-    {
-      "epoch": 2.946107784431138,
-      "grad_norm": 0.03545184716551114,
-      "learning_rate": 1.7889494365118665e-08,
-      "loss": 0.0002,
-      "step": 1968
-    },
-    {
-      "epoch": 2.947604790419162,
-      "grad_norm": 0.05849241385118926,
-      "learning_rate": 1.6935831587541197e-08,
-      "loss": 0.0003,
-      "step": 1969
-    },
-    {
-      "epoch": 2.9491017964071857,
-      "grad_norm": 0.11944886195630253,
-      "learning_rate": 1.6008267452090054e-08,
-      "loss": 0.0014,
-      "step": 1970
-    },
-    {
-      "epoch": 2.9505988023952097,
-      "grad_norm": 0.023633602514857067,
-      "learning_rate": 1.5106804383688834e-08,
-      "loss": 0.0002,
-      "step": 1971
-    },
-    {
-      "epoch": 2.9520958083832336,
-      "grad_norm": 0.06252240701759237,
-      "learning_rate": 1.4231444739023492e-08,
-      "loss": 0.0005,
-      "step": 1972
-    },
-    {
-      "epoch": 2.9535928143712575,
-      "grad_norm": 0.08288723195823507,
-      "learning_rate": 1.338219080654013e-08,
-      "loss": 0.0004,
-      "step": 1973
-    },
-    {
-      "epoch": 2.9550898203592815,
-      "grad_norm": 0.02470100516563367,
-      "learning_rate": 1.2559044806437215e-08,
-      "loss": 0.0002,
-      "step": 1974
-    },
-    {
-      "epoch": 2.9565868263473054,
-      "grad_norm": 0.035384313511806896,
-      "learning_rate": 1.1762008890657816e-08,
-      "loss": 0.0002,
-      "step": 1975
-    },
-    {
-      "epoch": 2.9580838323353293,
-      "grad_norm": 0.0726241495411833,
-      "learning_rate": 1.099108514288627e-08,
-      "loss": 0.0005,
-      "step": 1976
-    },
-    {
-      "epoch": 2.9595808383233533,
-      "grad_norm": 0.05304485849070656,
-      "learning_rate": 1.024627557854152e-08,
-      "loss": 0.0002,
-      "step": 1977
-    },
-    {
-      "epoch": 2.961077844311377,
-      "grad_norm": 0.11794290635555527,
-      "learning_rate": 9.527582144773785e-09,
-      "loss": 0.0006,
-      "step": 1978
-    },
-    {
-      "epoch": 2.962574850299401,
-      "grad_norm": 0.012966785537484036,
-      "learning_rate": 8.835006720457895e-09,
-      "loss": 0.0001,
-      "step": 1979
-    },
-    {
-      "epoch": 2.964071856287425,
-      "grad_norm": 0.05050410993770117,
-      "learning_rate": 8.16855111618775e-09,
-      "loss": 0.0003,
-      "step": 1980
-    },
-    {
-      "epoch": 2.965568862275449,
-      "grad_norm": 0.0697386602951542,
-      "learning_rate": 7.528217074272982e-09,
-      "loss": 0.0004,
-      "step": 1981
-    },
-    {
-      "epoch": 2.967065868263473,
-      "grad_norm": 0.0324564659125771,
-      "learning_rate": 6.914006268732287e-09,
-      "loss": 0.0003,
-      "step": 1982
-    },
-    {
-      "epoch": 2.968562874251497,
-      "grad_norm": 0.03373070903115685,
-      "learning_rate": 6.325920305292332e-09,
-      "loss": 0.0002,
-      "step": 1983
-    },
-    {
-      "epoch": 2.970059880239521,
-      "grad_norm": 0.011788534817898908,
-      "learning_rate": 5.763960721382189e-09,
-      "loss": 0.0002,
-      "step": 1984
-    },
-    {
-      "epoch": 2.9715568862275448,
-      "grad_norm": 0.08221326208806214,
-      "learning_rate": 5.228128986127789e-09,
-      "loss": 0.0006,
-      "step": 1985
-    },
-    {
-      "epoch": 2.9730538922155687,
-      "grad_norm": 0.1270884872336597,
-      "learning_rate": 4.718426500349705e-09,
-      "loss": 0.0005,
-      "step": 1986
-    },
-    {
-      "epoch": 2.974550898203593,
-      "grad_norm": 0.058075307647141146,
-      "learning_rate": 4.2348545965575914e-09,
-      "loss": 0.0002,
-      "step": 1987
-    },
-    {
-      "epoch": 2.976047904191617,
-      "grad_norm": 0.03512480614793991,
-      "learning_rate": 3.7774145389524176e-09,
-      "loss": 0.0003,
-      "step": 1988
-    },
-    {
-      "epoch": 2.977544910179641,
-      "grad_norm": 0.07371078403449538,
-      "learning_rate": 3.3461075234131336e-09,
-      "loss": 0.0001,
-      "step": 1989
-    },
-    {
-      "epoch": 2.979041916167665,
-      "grad_norm": 0.022109669320123617,
-      "learning_rate": 2.9409346775033376e-09,
-      "loss": 0.0001,
-      "step": 1990
-    },
-    {
-      "epoch": 2.980538922155689,
-      "grad_norm": 0.048672724122854005,
-      "learning_rate": 2.561897060462393e-09,
-      "loss": 0.0004,
-      "step": 1991
-    },
-    {
-      "epoch": 2.9820359281437128,
-      "grad_norm": 0.01931500903511454,
-      "learning_rate": 2.208995663206537e-09,
-      "loss": 0.0002,
-      "step": 1992
-    },
-    {
-      "epoch": 2.9835329341317367,
-      "grad_norm": 0.04749199082195791,
-      "learning_rate": 1.8822314083222216e-09,
-      "loss": 0.0004,
-      "step": 1993
-    },
-    {
-      "epoch": 2.9850299401197606,
-      "grad_norm": 0.09729739241328733,
-      "learning_rate": 1.5816051500661122e-09,
-      "loss": 0.001,
-      "step": 1994
-    },
-    {
-      "epoch": 2.9865269461077846,
-      "grad_norm": 0.044947007048880785,
-      "learning_rate": 1.3071176743628677e-09,
-      "loss": 0.0003,
-      "step": 1995
-    },
-    {
-      "epoch": 2.9880239520958085,
-      "grad_norm": 0.13515173564413327,
-      "learning_rate": 1.0587696988029195e-09,
-      "loss": 0.0008,
-      "step": 1996
-    },
-    {
-      "epoch": 2.9895209580838324,
-      "grad_norm": 0.06726018374004353,
-      "learning_rate": 8.365618726413616e-10,
-      "loss": 0.0005,
-      "step": 1997
-    },
-    {
-      "epoch": 2.9910179640718564,
-      "grad_norm": 0.1032600661241297,
-      "learning_rate": 6.404947767935099e-10,
-      "loss": 0.0011,
-      "step": 1998
-    },
-    {
-      "epoch": 2.9925149700598803,
-      "grad_norm": 0.08635105614698947,
-      "learning_rate": 4.705689238360123e-10,
-      "loss": 0.0009,
-      "step": 1999
-    },
-    {
-      "epoch": 2.9940119760479043,
-      "grad_norm": 0.10265564812162384,
-      "learning_rate": 3.267847580046279e-10,
-      "loss": 0.0004,
-      "step": 2000
-    },
-    {
-      "epoch": 2.995508982035928,
-      "grad_norm": 0.040089587915776735,
-      "learning_rate": 2.091426551931175e-10,
-      "loss": 0.0002,
-      "step": 2001
-    },
-    {
-      "epoch": 2.997005988023952,
-      "grad_norm": 0.05526462283205673,
-      "learning_rate": 1.1764292295213254e-10,
-      "loss": 0.0002,
-      "step": 2002
-    },
-    {
-      "epoch": 2.998502994011976,
-      "grad_norm": 0.046950315243127365,
-      "learning_rate": 5.2285800488105495e-11,
-      "loss": 0.0003,
-      "step": 2003
-    },
-    {
-      "epoch": 3.0,
-      "grad_norm": 0.02255507610632034,
-      "learning_rate": 1.3071458665470105e-11,
-      "loss": 0.0001,
-      "step": 2004
+      "step": 1875
     },
     {
       "epoch": 3.0,
-      "step": 2004,
-      "total_flos": 160061552422912.0,
-      "train_loss": 0.02935138660071818,
-      "train_runtime": 23628.3662,
-      "train_samples_per_second": 1.356,
-      "train_steps_per_second": 0.085
+      "step": 1875,
+      "total_flos": 75591737456640.0,
+      "train_loss": 0.04129861024404333,
+      "train_runtime": 14343.312,
+      "train_samples_per_second": 2.092,
+      "train_steps_per_second": 0.131
     }
   ],
   "logging_steps": 1.0,
-  "max_steps": 2004,
+  "max_steps": 1875,
   "num_input_tokens_seen": 0,
   "num_train_epochs": 3,
   "save_steps": 50000,
-  "total_flos": 160061552422912.0,
+  "total_flos": 75591737456640.0,
   "train_batch_size": 8,
   "trial_name": null,
   "trial_params": null