diff --git "a/trainer_state.json" "b/trainer_state.json"
new file mode 100644--- /dev/null
+++ "b/trainer_state.json"
@@ -0,0 +1,14042 @@
+{
+  "best_metric": null,
+  "best_model_checkpoint": null,
+  "epoch": 0.22326645805206993,
+  "eval_steps": 500,
+  "global_step": 2000,
+  "is_hyper_param_search": false,
+  "is_local_process_zero": true,
+  "is_world_process_zero": true,
+  "log_history": [
+    {
+      "epoch": 0.00011163322902603496,
+      "grad_norm": 0.12030283361673355,
+      "learning_rate": 4.9975e-05,
+      "loss": 2.5157,
+      "step": 1
+    },
+    {
+      "epoch": 0.00022326645805206992,
+      "grad_norm": 0.10777310281991959,
+      "learning_rate": 4.995e-05,
+      "loss": 2.5461,
+      "step": 2
+    },
+    {
+      "epoch": 0.0003348996870781049,
+      "grad_norm": 0.1150362566113472,
+      "learning_rate": 4.992500000000001e-05,
+      "loss": 2.463,
+      "step": 3
+    },
+    {
+      "epoch": 0.00044653291610413984,
+      "grad_norm": 0.19681179523468018,
+      "learning_rate": 4.99e-05,
+      "loss": 2.6149,
+      "step": 4
+    },
+    {
+      "epoch": 0.0005581661451301748,
+      "grad_norm": 0.13096819818019867,
+      "learning_rate": 4.9875000000000006e-05,
+      "loss": 2.4514,
+      "step": 5
+    },
+    {
+      "epoch": 0.0006697993741562098,
+      "grad_norm": 0.13657382130622864,
+      "learning_rate": 4.9850000000000006e-05,
+      "loss": 2.6006,
+      "step": 6
+    },
+    {
+      "epoch": 0.0007814326031822447,
+      "grad_norm": 0.12205895781517029,
+      "learning_rate": 4.9825000000000005e-05,
+      "loss": 2.4963,
+      "step": 7
+    },
+    {
+      "epoch": 0.0008930658322082797,
+      "grad_norm": 0.12875059247016907,
+      "learning_rate": 4.9800000000000004e-05,
+      "loss": 2.4755,
+      "step": 8
+    },
+    {
+      "epoch": 0.0010046990612343147,
+      "grad_norm": 0.13755826652050018,
+      "learning_rate": 4.9775000000000004e-05,
+      "loss": 2.4421,
+      "step": 9
+    },
+    {
+      "epoch": 0.0011163322902603497,
+      "grad_norm": 0.14838671684265137,
+      "learning_rate": 4.975e-05,
+      "loss": 2.5355,
+      "step": 10
+    },
+    {
+      "epoch": 0.0012279655192863846,
+      "grad_norm": 0.20818187296390533,
+      "learning_rate": 4.9725e-05,
+      "loss": 2.4286,
+      "step": 11
+    },
+    {
+      "epoch": 0.0013395987483124196,
+      "grad_norm": 0.16847440600395203,
+      "learning_rate": 4.97e-05,
+      "loss": 2.4903,
+      "step": 12
+    },
+    {
+      "epoch": 0.0014512319773384544,
+      "grad_norm": 0.16514980792999268,
+      "learning_rate": 4.967500000000001e-05,
+      "loss": 2.3927,
+      "step": 13
+    },
+    {
+      "epoch": 0.0015628652063644894,
+      "grad_norm": 0.17463476955890656,
+      "learning_rate": 4.965e-05,
+      "loss": 2.4566,
+      "step": 14
+    },
+    {
+      "epoch": 0.0016744984353905244,
+      "grad_norm": 0.1806865930557251,
+      "learning_rate": 4.962500000000001e-05,
+      "loss": 2.3876,
+      "step": 15
+    },
+    {
+      "epoch": 0.0017861316644165594,
+      "grad_norm": 0.4340119957923889,
+      "learning_rate": 4.96e-05,
+      "loss": 2.5106,
+      "step": 16
+    },
+    {
+      "epoch": 0.0018977648934425943,
+      "grad_norm": 0.18326279520988464,
+      "learning_rate": 4.9575000000000006e-05,
+      "loss": 2.4918,
+      "step": 17
+    },
+    {
+      "epoch": 0.0020093981224686293,
+      "grad_norm": 0.1809333860874176,
+      "learning_rate": 4.9550000000000005e-05,
+      "loss": 2.478,
+      "step": 18
+    },
+    {
+      "epoch": 0.0021210313514946643,
+      "grad_norm": 0.1895465850830078,
+      "learning_rate": 4.9525000000000004e-05,
+      "loss": 2.5024,
+      "step": 19
+    },
+    {
+      "epoch": 0.0022326645805206993,
+      "grad_norm": 0.20786350965499878,
+      "learning_rate": 4.9500000000000004e-05,
+      "loss": 2.5115,
+      "step": 20
+    },
+    {
+      "epoch": 0.0023442978095467343,
+      "grad_norm": 0.1992110311985016,
+      "learning_rate": 4.9475e-05,
+      "loss": 2.4044,
+      "step": 21
+    },
+    {
+      "epoch": 0.0024559310385727693,
+      "grad_norm": 0.20190922915935516,
+      "learning_rate": 4.945e-05,
+      "loss": 2.4079,
+      "step": 22
+    },
+    {
+      "epoch": 0.0025675642675988043,
+      "grad_norm": 0.22950319945812225,
+      "learning_rate": 4.9425e-05,
+      "loss": 2.5367,
+      "step": 23
+    },
+    {
+      "epoch": 0.0026791974966248393,
+      "grad_norm": 0.2105780690908432,
+      "learning_rate": 4.94e-05,
+      "loss": 2.3135,
+      "step": 24
+    },
+    {
+      "epoch": 0.0027908307256508742,
+      "grad_norm": 0.21351750195026398,
+      "learning_rate": 4.937500000000001e-05,
+      "loss": 2.4224,
+      "step": 25
+    },
+    {
+      "epoch": 0.002902463954676909,
+      "grad_norm": 0.21121960878372192,
+      "learning_rate": 4.935e-05,
+      "loss": 2.4935,
+      "step": 26
+    },
+    {
+      "epoch": 0.003014097183702944,
+      "grad_norm": 0.21738024055957794,
+      "learning_rate": 4.9325000000000006e-05,
+      "loss": 2.4383,
+      "step": 27
+    },
+    {
+      "epoch": 0.0031257304127289788,
+      "grad_norm": 0.22073255479335785,
+      "learning_rate": 4.93e-05,
+      "loss": 2.4861,
+      "step": 28
+    },
+    {
+      "epoch": 0.0032373636417550138,
+      "grad_norm": 0.2213854044675827,
+      "learning_rate": 4.9275000000000005e-05,
+      "loss": 2.4449,
+      "step": 29
+    },
+    {
+      "epoch": 0.0033489968707810487,
+      "grad_norm": 0.22176030278205872,
+      "learning_rate": 4.9250000000000004e-05,
+      "loss": 2.3341,
+      "step": 30
+    },
+    {
+      "epoch": 0.0034606300998070837,
+      "grad_norm": 0.22657762467861176,
+      "learning_rate": 4.9225000000000004e-05,
+      "loss": 2.4111,
+      "step": 31
+    },
+    {
+      "epoch": 0.0035722633288331187,
+      "grad_norm": 0.23043128848075867,
+      "learning_rate": 4.92e-05,
+      "loss": 2.4449,
+      "step": 32
+    },
+    {
+      "epoch": 0.0036838965578591537,
+      "grad_norm": 0.22782647609710693,
+      "learning_rate": 4.9175e-05,
+      "loss": 2.496,
+      "step": 33
+    },
+    {
+      "epoch": 0.0037955297868851887,
+      "grad_norm": 0.2296725958585739,
+      "learning_rate": 4.915e-05,
+      "loss": 2.4957,
+      "step": 34
+    },
+    {
+      "epoch": 0.003907163015911224,
+      "grad_norm": 0.23677076399326324,
+      "learning_rate": 4.9125e-05,
+      "loss": 2.4031,
+      "step": 35
+    },
+    {
+      "epoch": 0.004018796244937259,
+      "grad_norm": 0.2547452747821808,
+      "learning_rate": 4.91e-05,
+      "loss": 2.4666,
+      "step": 36
+    },
+    {
+      "epoch": 0.004130429473963294,
+      "grad_norm": 0.24586506187915802,
+      "learning_rate": 4.907500000000001e-05,
+      "loss": 2.4721,
+      "step": 37
+    },
+    {
+      "epoch": 0.004242062702989329,
+      "grad_norm": 0.24286368489265442,
+      "learning_rate": 4.905e-05,
+      "loss": 2.4584,
+      "step": 38
+    },
+    {
+      "epoch": 0.004353695932015364,
+      "grad_norm": 0.2877066135406494,
+      "learning_rate": 4.9025000000000006e-05,
+      "loss": 2.3817,
+      "step": 39
+    },
+    {
+      "epoch": 0.004465329161041399,
+      "grad_norm": 0.36161181330680847,
+      "learning_rate": 4.9e-05,
+      "loss": 2.4246,
+      "step": 40
+    },
+    {
+      "epoch": 0.004576962390067434,
+      "grad_norm": 0.23599842190742493,
+      "learning_rate": 4.8975000000000005e-05,
+      "loss": 2.4416,
+      "step": 41
+    },
+    {
+      "epoch": 0.004688595619093469,
+      "grad_norm": 3.929527521133423,
+      "learning_rate": 4.8950000000000004e-05,
+      "loss": 2.536,
+      "step": 42
+    },
+    {
+      "epoch": 0.004800228848119504,
+      "grad_norm": 0.25463879108428955,
+      "learning_rate": 4.8925e-05,
+      "loss": 2.3839,
+      "step": 43
+    },
+    {
+      "epoch": 0.004911862077145539,
+      "grad_norm": 0.2564179599285126,
+      "learning_rate": 4.89e-05,
+      "loss": 2.5175,
+      "step": 44
+    },
+    {
+      "epoch": 0.0050234953061715736,
+      "grad_norm": 0.2552028298377991,
+      "learning_rate": 4.8875e-05,
+      "loss": 2.4055,
+      "step": 45
+    },
+    {
+      "epoch": 0.0051351285351976085,
+      "grad_norm": 0.27434229850769043,
+      "learning_rate": 4.885e-05,
+      "loss": 2.4139,
+      "step": 46
+    },
+    {
+      "epoch": 0.0052467617642236435,
+      "grad_norm": 0.2661423981189728,
+      "learning_rate": 4.8825e-05,
+      "loss": 2.4711,
+      "step": 47
+    },
+    {
+      "epoch": 0.0053583949932496785,
+      "grad_norm": 0.24605660140514374,
+      "learning_rate": 4.88e-05,
+      "loss": 2.3726,
+      "step": 48
+    },
+    {
+      "epoch": 0.0054700282222757135,
+      "grad_norm": 0.27371543645858765,
+      "learning_rate": 4.8775000000000007e-05,
+      "loss": 2.4012,
+      "step": 49
+    },
+    {
+      "epoch": 0.0055816614513017485,
+      "grad_norm": 0.26587924361228943,
+      "learning_rate": 4.875e-05,
+      "loss": 2.4409,
+      "step": 50
+    },
+    {
+      "epoch": 0.0056932946803277835,
+      "grad_norm": 0.28210633993148804,
+      "learning_rate": 4.8725000000000005e-05,
+      "loss": 2.5431,
+      "step": 51
+    },
+    {
+      "epoch": 0.005804927909353818,
+      "grad_norm": 0.2553481459617615,
+      "learning_rate": 4.87e-05,
+      "loss": 2.5122,
+      "step": 52
+    },
+    {
+      "epoch": 0.005916561138379853,
+      "grad_norm": 0.2604880928993225,
+      "learning_rate": 4.8675000000000004e-05,
+      "loss": 2.4656,
+      "step": 53
+    },
+    {
+      "epoch": 0.006028194367405888,
+      "grad_norm": 0.266725093126297,
+      "learning_rate": 4.8650000000000003e-05,
+      "loss": 2.307,
+      "step": 54
+    },
+    {
+      "epoch": 0.0061398275964319226,
+      "grad_norm": 5.911880016326904,
+      "learning_rate": 4.8625e-05,
+      "loss": 2.2753,
+      "step": 55
+    },
+    {
+      "epoch": 0.0062514608254579575,
+      "grad_norm": 0.2631521224975586,
+      "learning_rate": 4.86e-05,
+      "loss": 2.4475,
+      "step": 56
+    },
+    {
+      "epoch": 0.0063630940544839925,
+      "grad_norm": 0.2781185507774353,
+      "learning_rate": 4.8575e-05,
+      "loss": 2.2942,
+      "step": 57
+    },
+    {
+      "epoch": 0.0064747272835100275,
+      "grad_norm": 0.27916616201400757,
+      "learning_rate": 4.855e-05,
+      "loss": 2.4817,
+      "step": 58
+    },
+    {
+      "epoch": 0.0065863605125360625,
+      "grad_norm": 0.25466758012771606,
+      "learning_rate": 4.8525e-05,
+      "loss": 2.3986,
+      "step": 59
+    },
+    {
+      "epoch": 0.0066979937415620975,
+      "grad_norm": 0.3041671812534332,
+      "learning_rate": 4.85e-05,
+      "loss": 2.3411,
+      "step": 60
+    },
+    {
+      "epoch": 0.0068096269705881325,
+      "grad_norm": 0.26597726345062256,
+      "learning_rate": 4.8475000000000006e-05,
+      "loss": 2.3992,
+      "step": 61
+    },
+    {
+      "epoch": 0.0069212601996141675,
+      "grad_norm": 0.2860955595970154,
+      "learning_rate": 4.845e-05,
+      "loss": 2.535,
+      "step": 62
+    },
+    {
+      "epoch": 0.0070328934286402025,
+      "grad_norm": 0.2573089897632599,
+      "learning_rate": 4.8425000000000005e-05,
+      "loss": 2.4366,
+      "step": 63
+    },
+    {
+      "epoch": 0.0071445266576662374,
+      "grad_norm": 0.3139612376689911,
+      "learning_rate": 4.8400000000000004e-05,
+      "loss": 2.4112,
+      "step": 64
+    },
+    {
+      "epoch": 0.007256159886692272,
+      "grad_norm": 0.27238690853118896,
+      "learning_rate": 4.8375000000000004e-05,
+      "loss": 2.5764,
+      "step": 65
+    },
+    {
+      "epoch": 0.007367793115718307,
+      "grad_norm": 0.2645399272441864,
+      "learning_rate": 4.835e-05,
+      "loss": 2.3662,
+      "step": 66
+    },
+    {
+      "epoch": 0.007479426344744342,
+      "grad_norm": 0.2746032178401947,
+      "learning_rate": 4.8325e-05,
+      "loss": 2.3946,
+      "step": 67
+    },
+    {
+      "epoch": 0.007591059573770377,
+      "grad_norm": 0.2907489836215973,
+      "learning_rate": 4.83e-05,
+      "loss": 2.4238,
+      "step": 68
+    },
+    {
+      "epoch": 0.007702692802796412,
+      "grad_norm": 0.2683127522468567,
+      "learning_rate": 4.8275e-05,
+      "loss": 2.3503,
+      "step": 69
+    },
+    {
+      "epoch": 0.007814326031822447,
+      "grad_norm": 0.29885104298591614,
+      "learning_rate": 4.825e-05,
+      "loss": 2.3667,
+      "step": 70
+    },
+    {
+      "epoch": 0.007925959260848481,
+      "grad_norm": 0.35321930050849915,
+      "learning_rate": 4.822500000000001e-05,
+      "loss": 2.2999,
+      "step": 71
+    },
+    {
+      "epoch": 0.008037592489874517,
+      "grad_norm": 0.28377628326416016,
+      "learning_rate": 4.82e-05,
+      "loss": 2.3928,
+      "step": 72
+    },
+    {
+      "epoch": 0.008149225718900551,
+      "grad_norm": 0.28445249795913696,
+      "learning_rate": 4.8175000000000005e-05,
+      "loss": 2.4402,
+      "step": 73
+    },
+    {
+      "epoch": 0.008260858947926587,
+      "grad_norm": 0.26443931460380554,
+      "learning_rate": 4.815e-05,
+      "loss": 2.3292,
+      "step": 74
+    },
+    {
+      "epoch": 0.008372492176952621,
+      "grad_norm": 0.6532557010650635,
+      "learning_rate": 4.8125000000000004e-05,
+      "loss": 2.404,
+      "step": 75
+    },
+    {
+      "epoch": 0.008484125405978657,
+      "grad_norm": 0.27610066533088684,
+      "learning_rate": 4.8100000000000004e-05,
+      "loss": 2.3759,
+      "step": 76
+    },
+    {
+      "epoch": 0.008595758635004691,
+      "grad_norm": 0.6011192798614502,
+      "learning_rate": 4.8075e-05,
+      "loss": 2.4855,
+      "step": 77
+    },
+    {
+      "epoch": 0.008707391864030727,
+      "grad_norm": 0.3119165599346161,
+      "learning_rate": 4.805e-05,
+      "loss": 2.4206,
+      "step": 78
+    },
+    {
+      "epoch": 0.008819025093056761,
+      "grad_norm": 0.2999507486820221,
+      "learning_rate": 4.8025e-05,
+      "loss": 2.3795,
+      "step": 79
+    },
+    {
+      "epoch": 0.008930658322082797,
+      "grad_norm": 0.32682228088378906,
+      "learning_rate": 4.8e-05,
+      "loss": 2.4536,
+      "step": 80
+    },
+    {
+      "epoch": 0.009042291551108831,
+      "grad_norm": 0.2823121249675751,
+      "learning_rate": 4.7975e-05,
+      "loss": 2.3822,
+      "step": 81
+    },
+    {
+      "epoch": 0.009153924780134867,
+      "grad_norm": 0.29254743456840515,
+      "learning_rate": 4.795e-05,
+      "loss": 2.4421,
+      "step": 82
+    },
+    {
+      "epoch": 0.009265558009160901,
+      "grad_norm": 0.40489596128463745,
+      "learning_rate": 4.7925000000000006e-05,
+      "loss": 2.3872,
+      "step": 83
+    },
+    {
+      "epoch": 0.009377191238186937,
+      "grad_norm": 0.28919968008995056,
+      "learning_rate": 4.79e-05,
+      "loss": 2.3616,
+      "step": 84
+    },
+    {
+      "epoch": 0.009488824467212971,
+      "grad_norm": 0.2884703576564789,
+      "learning_rate": 4.7875000000000005e-05,
+      "loss": 2.4244,
+      "step": 85
+    },
+    {
+      "epoch": 0.009600457696239007,
+      "grad_norm": 0.27483540773391724,
+      "learning_rate": 4.785e-05,
+      "loss": 2.3592,
+      "step": 86
+    },
+    {
+      "epoch": 0.009712090925265041,
+      "grad_norm": 0.2799672484397888,
+      "learning_rate": 4.7825000000000004e-05,
+      "loss": 2.3994,
+      "step": 87
+    },
+    {
+      "epoch": 0.009823724154291077,
+      "grad_norm": 0.2779146134853363,
+      "learning_rate": 4.78e-05,
+      "loss": 2.5073,
+      "step": 88
+    },
+    {
+      "epoch": 0.009935357383317111,
+      "grad_norm": 0.283578097820282,
+      "learning_rate": 4.7775e-05,
+      "loss": 2.3933,
+      "step": 89
+    },
+    {
+      "epoch": 0.010046990612343147,
+      "grad_norm": 0.28255191445350647,
+      "learning_rate": 4.775e-05,
+      "loss": 2.5065,
+      "step": 90
+    },
+    {
+      "epoch": 0.010158623841369181,
+      "grad_norm": 0.9130760431289673,
+      "learning_rate": 4.7725e-05,
+      "loss": 2.353,
+      "step": 91
+    },
+    {
+      "epoch": 0.010270257070395217,
+      "grad_norm": 0.28134602308273315,
+      "learning_rate": 4.77e-05,
+      "loss": 2.465,
+      "step": 92
+    },
+    {
+      "epoch": 0.010381890299421251,
+      "grad_norm": 0.2763191759586334,
+      "learning_rate": 4.7675e-05,
+      "loss": 2.317,
+      "step": 93
+    },
+    {
+      "epoch": 0.010493523528447287,
+      "grad_norm": 0.2876272201538086,
+      "learning_rate": 4.765e-05,
+      "loss": 2.3916,
+      "step": 94
+    },
+    {
+      "epoch": 0.010605156757473321,
+      "grad_norm": 0.28046631813049316,
+      "learning_rate": 4.7625000000000006e-05,
+      "loss": 2.4663,
+      "step": 95
+    },
+    {
+      "epoch": 0.010716789986499357,
+      "grad_norm": 0.2862105667591095,
+      "learning_rate": 4.76e-05,
+      "loss": 2.4214,
+      "step": 96
+    },
+    {
+      "epoch": 0.010828423215525391,
+      "grad_norm": 0.27971214056015015,
+      "learning_rate": 4.7575000000000004e-05,
+      "loss": 2.4528,
+      "step": 97
+    },
+    {
+      "epoch": 0.010940056444551427,
+      "grad_norm": 0.29032212495803833,
+      "learning_rate": 4.755e-05,
+      "loss": 2.3312,
+      "step": 98
+    },
+    {
+      "epoch": 0.011051689673577461,
+      "grad_norm": 0.293649286031723,
+      "learning_rate": 4.7525e-05,
+      "loss": 2.3592,
+      "step": 99
+    },
+    {
+      "epoch": 0.011163322902603497,
+      "grad_norm": 0.277589350938797,
+      "learning_rate": 4.75e-05,
+      "loss": 2.392,
+      "step": 100
+    },
+    {
+      "epoch": 0.011274956131629531,
+      "grad_norm": 0.28249549865722656,
+      "learning_rate": 4.7475e-05,
+      "loss": 2.3154,
+      "step": 101
+    },
+    {
+      "epoch": 0.011386589360655567,
+      "grad_norm": 0.30689579248428345,
+      "learning_rate": 4.745e-05,
+      "loss": 2.3322,
+      "step": 102
+    },
+    {
+      "epoch": 0.011498222589681601,
+      "grad_norm": 0.2909144163131714,
+      "learning_rate": 4.7425e-05,
+      "loss": 2.4028,
+      "step": 103
+    },
+    {
+      "epoch": 0.011609855818707635,
+      "grad_norm": 0.2826705574989319,
+      "learning_rate": 4.74e-05,
+      "loss": 2.3969,
+      "step": 104
+    },
+    {
+      "epoch": 0.011721489047733671,
+      "grad_norm": 0.30071696639060974,
+      "learning_rate": 4.7375e-05,
+      "loss": 2.435,
+      "step": 105
+    },
+    {
+      "epoch": 0.011833122276759705,
+      "grad_norm": 0.29308071732521057,
+      "learning_rate": 4.735e-05,
+      "loss": 2.3299,
+      "step": 106
+    },
+    {
+      "epoch": 0.011944755505785741,
+      "grad_norm": 0.28309884667396545,
+      "learning_rate": 4.7325000000000005e-05,
+      "loss": 2.511,
+      "step": 107
+    },
+    {
+      "epoch": 0.012056388734811775,
+      "grad_norm": 0.3001827597618103,
+      "learning_rate": 4.73e-05,
+      "loss": 2.3804,
+      "step": 108
+    },
+    {
+      "epoch": 0.012168021963837811,
+      "grad_norm": 0.3125348687171936,
+      "learning_rate": 4.7275000000000004e-05,
+      "loss": 2.3186,
+      "step": 109
+    },
+    {
+      "epoch": 0.012279655192863845,
+      "grad_norm": 0.29325881600379944,
+      "learning_rate": 4.7249999999999997e-05,
+      "loss": 2.4714,
+      "step": 110
+    },
+    {
+      "epoch": 0.012391288421889881,
+      "grad_norm": 0.28102368116378784,
+      "learning_rate": 4.7225e-05,
+      "loss": 2.4592,
+      "step": 111
+    },
+    {
+      "epoch": 0.012502921650915915,
+      "grad_norm": 0.2798093259334564,
+      "learning_rate": 4.72e-05,
+      "loss": 2.398,
+      "step": 112
+    },
+    {
+      "epoch": 0.012614554879941951,
+      "grad_norm": 0.29487597942352295,
+      "learning_rate": 4.7175e-05,
+      "loss": 2.3144,
+      "step": 113
+    },
+    {
+      "epoch": 0.012726188108967985,
+      "grad_norm": 0.28528064489364624,
+      "learning_rate": 4.715e-05,
+      "loss": 2.4359,
+      "step": 114
+    },
+    {
+      "epoch": 0.012837821337994021,
+      "grad_norm": 0.29618656635284424,
+      "learning_rate": 4.7125e-05,
+      "loss": 2.3421,
+      "step": 115
+    },
+    {
+      "epoch": 0.012949454567020055,
+      "grad_norm": 0.27769914269447327,
+      "learning_rate": 4.71e-05,
+      "loss": 2.4087,
+      "step": 116
+    },
+    {
+      "epoch": 0.013061087796046091,
+      "grad_norm": 0.2721666991710663,
+      "learning_rate": 4.7075e-05,
+      "loss": 2.4001,
+      "step": 117
+    },
+    {
+      "epoch": 0.013172721025072125,
+      "grad_norm": 0.6449373960494995,
+      "learning_rate": 4.705e-05,
+      "loss": 2.4581,
+      "step": 118
+    },
+    {
+      "epoch": 0.013284354254098161,
+      "grad_norm": 0.28057020902633667,
+      "learning_rate": 4.7025000000000005e-05,
+      "loss": 2.3935,
+      "step": 119
+    },
+    {
+      "epoch": 0.013395987483124195,
+      "grad_norm": 0.2757243812084198,
+      "learning_rate": 4.7e-05,
+      "loss": 2.465,
+      "step": 120
+    },
+    {
+      "epoch": 0.01350762071215023,
+      "grad_norm": 0.2977396249771118,
+      "learning_rate": 4.6975000000000003e-05,
+      "loss": 2.4985,
+      "step": 121
+    },
+    {
+      "epoch": 0.013619253941176265,
+      "grad_norm": 0.27909162640571594,
+      "learning_rate": 4.695e-05,
+      "loss": 2.4484,
+      "step": 122
+    },
+    {
+      "epoch": 0.0137308871702023,
+      "grad_norm": 0.28472158312797546,
+      "learning_rate": 4.6925e-05,
+      "loss": 2.499,
+      "step": 123
+    },
+    {
+      "epoch": 0.013842520399228335,
+      "grad_norm": 0.2772194445133209,
+      "learning_rate": 4.69e-05,
+      "loss": 2.4161,
+      "step": 124
+    },
+    {
+      "epoch": 0.01395415362825437,
+      "grad_norm": 0.28007185459136963,
+      "learning_rate": 4.6875e-05,
+      "loss": 2.4538,
+      "step": 125
+    },
+    {
+      "epoch": 0.014065786857280405,
+      "grad_norm": 0.2890627086162567,
+      "learning_rate": 4.685000000000001e-05,
+      "loss": 2.3803,
+      "step": 126
+    },
+    {
+      "epoch": 0.01417742008630644,
+      "grad_norm": 0.28412866592407227,
+      "learning_rate": 4.6825e-05,
+      "loss": 2.4739,
+      "step": 127
+    },
+    {
+      "epoch": 0.014289053315332475,
+      "grad_norm": 0.28246861696243286,
+      "learning_rate": 4.6800000000000006e-05,
+      "loss": 2.4031,
+      "step": 128
+    },
+    {
+      "epoch": 0.01440068654435851,
+      "grad_norm": 0.27640506625175476,
+      "learning_rate": 4.6775000000000005e-05,
+      "loss": 2.3415,
+      "step": 129
+    },
+    {
+      "epoch": 0.014512319773384545,
+      "grad_norm": 0.27817410230636597,
+      "learning_rate": 4.6750000000000005e-05,
+      "loss": 2.5304,
+      "step": 130
+    },
+    {
+      "epoch": 0.01462395300241058,
+      "grad_norm": 0.27021604776382446,
+      "learning_rate": 4.6725000000000004e-05,
+      "loss": 2.34,
+      "step": 131
+    },
+    {
+      "epoch": 0.014735586231436615,
+      "grad_norm": 0.2793290913105011,
+      "learning_rate": 4.6700000000000003e-05,
+      "loss": 2.4935,
+      "step": 132
+    },
+    {
+      "epoch": 0.01484721946046265,
+      "grad_norm": 0.2610988914966583,
+      "learning_rate": 4.6675e-05,
+      "loss": 2.1694,
+      "step": 133
+    },
+    {
+      "epoch": 0.014958852689488685,
+      "grad_norm": 0.3299348056316376,
+      "learning_rate": 4.665e-05,
+      "loss": 2.3272,
+      "step": 134
+    },
+    {
+      "epoch": 0.01507048591851472,
+      "grad_norm": 0.27494046092033386,
+      "learning_rate": 4.6625e-05,
+      "loss": 2.4961,
+      "step": 135
+    },
+    {
+      "epoch": 0.015182119147540755,
+      "grad_norm": 0.8337180018424988,
+      "learning_rate": 4.660000000000001e-05,
+      "loss": 2.4057,
+      "step": 136
+    },
+    {
+      "epoch": 0.01529375237656679,
+      "grad_norm": 0.27909937500953674,
+      "learning_rate": 4.6575e-05,
+      "loss": 2.279,
+      "step": 137
+    },
+    {
+      "epoch": 0.015405385605592825,
+      "grad_norm": 0.3143457770347595,
+      "learning_rate": 4.655000000000001e-05,
+      "loss": 2.4057,
+      "step": 138
+    },
+    {
+      "epoch": 0.015517018834618859,
+      "grad_norm": 0.27124184370040894,
+      "learning_rate": 4.6525e-05,
+      "loss": 2.4525,
+      "step": 139
+    },
+    {
+      "epoch": 0.015628652063644895,
+      "grad_norm": 0.27924227714538574,
+      "learning_rate": 4.6500000000000005e-05,
+      "loss": 2.4891,
+      "step": 140
+    },
+    {
+      "epoch": 0.01574028529267093,
+      "grad_norm": 0.2761871814727783,
+      "learning_rate": 4.6475000000000005e-05,
+      "loss": 2.4127,
+      "step": 141
+    },
+    {
+      "epoch": 0.015851918521696963,
+      "grad_norm": 0.30836501717567444,
+      "learning_rate": 4.6450000000000004e-05,
+      "loss": 2.3762,
+      "step": 142
+    },
+    {
+      "epoch": 0.015963551750723,
+      "grad_norm": 0.2716349959373474,
+      "learning_rate": 4.6425000000000004e-05,
+      "loss": 2.3893,
+      "step": 143
+    },
+    {
+      "epoch": 0.016075184979749035,
+      "grad_norm": 0.27172204852104187,
+      "learning_rate": 4.64e-05,
+      "loss": 2.3451,
+      "step": 144
+    },
+    {
+      "epoch": 0.01618681820877507,
+      "grad_norm": 0.27586179971694946,
+      "learning_rate": 4.6375e-05,
+      "loss": 2.3197,
+      "step": 145
+    },
+    {
+      "epoch": 0.016298451437801103,
+      "grad_norm": 0.27401286363601685,
+      "learning_rate": 4.635e-05,
+      "loss": 2.3191,
+      "step": 146
+    },
+    {
+      "epoch": 0.01641008466682714,
+      "grad_norm": 0.27832385897636414,
+      "learning_rate": 4.6325e-05,
+      "loss": 2.4407,
+      "step": 147
+    },
+    {
+      "epoch": 0.016521717895853175,
+      "grad_norm": 0.29265516996383667,
+      "learning_rate": 4.630000000000001e-05,
+      "loss": 2.3436,
+      "step": 148
+    },
+    {
+      "epoch": 0.01663335112487921,
+      "grad_norm": 0.27826353907585144,
+      "learning_rate": 4.6275e-05,
+      "loss": 2.4081,
+      "step": 149
+    },
+    {
+      "epoch": 0.016744984353905243,
+      "grad_norm": 0.26623812317848206,
+      "learning_rate": 4.6250000000000006e-05,
+      "loss": 2.4566,
+      "step": 150
+    },
+    {
+      "epoch": 0.01685661758293128,
+      "grad_norm": 0.2699335217475891,
+      "learning_rate": 4.6225e-05,
+      "loss": 2.3,
+      "step": 151
+    },
+    {
+      "epoch": 0.016968250811957315,
+      "grad_norm": 0.27325987815856934,
+      "learning_rate": 4.6200000000000005e-05,
+      "loss": 2.3798,
+      "step": 152
+    },
+    {
+      "epoch": 0.01707988404098335,
+      "grad_norm": 0.29398098587989807,
+      "learning_rate": 4.6175000000000004e-05,
+      "loss": 2.2952,
+      "step": 153
+    },
+    {
+      "epoch": 0.017191517270009383,
+      "grad_norm": 0.264258474111557,
+      "learning_rate": 4.6150000000000004e-05,
+      "loss": 2.4543,
+      "step": 154
+    },
+    {
+      "epoch": 0.01730315049903542,
+      "grad_norm": 0.4211874008178711,
+      "learning_rate": 4.6125e-05,
+      "loss": 2.3015,
+      "step": 155
+    },
+    {
+      "epoch": 0.017414783728061455,
+      "grad_norm": 0.28733232617378235,
+      "learning_rate": 4.61e-05,
+      "loss": 2.4002,
+      "step": 156
+    },
+    {
+      "epoch": 0.01752641695708749,
+      "grad_norm": 0.2647246718406677,
+      "learning_rate": 4.6075e-05,
+      "loss": 2.2928,
+      "step": 157
+    },
+    {
+      "epoch": 0.017638050186113523,
+      "grad_norm": 0.2679901719093323,
+      "learning_rate": 4.605e-05,
+      "loss": 2.4328,
+      "step": 158
+    },
+    {
+      "epoch": 0.01774968341513956,
+      "grad_norm": 0.26848575472831726,
+      "learning_rate": 4.6025e-05,
+      "loss": 2.3402,
+      "step": 159
+    },
+    {
+      "epoch": 0.017861316644165594,
+      "grad_norm": 0.282953679561615,
+      "learning_rate": 4.600000000000001e-05,
+      "loss": 2.3454,
+      "step": 160
+    },
+    {
+      "epoch": 0.01797294987319163,
+      "grad_norm": 0.2788335084915161,
+      "learning_rate": 4.5975e-05,
+      "loss": 2.3779,
+      "step": 161
+    },
+    {
+      "epoch": 0.018084583102217663,
+      "grad_norm": 0.2903019189834595,
+      "learning_rate": 4.5950000000000006e-05,
+      "loss": 2.4031,
+      "step": 162
+    },
+    {
+      "epoch": 0.0181962163312437,
+      "grad_norm": 0.2800057828426361,
+      "learning_rate": 4.5925e-05,
+      "loss": 2.4273,
+      "step": 163
+    },
+    {
+      "epoch": 0.018307849560269734,
+      "grad_norm": 0.28730225563049316,
+      "learning_rate": 4.5900000000000004e-05,
+      "loss": 2.3713,
+      "step": 164
+    },
+    {
+      "epoch": 0.01841948278929577,
+      "grad_norm": 0.2722271978855133,
+      "learning_rate": 4.5875000000000004e-05,
+      "loss": 2.5114,
+      "step": 165
+    },
+    {
+      "epoch": 0.018531116018321803,
+      "grad_norm": 0.27777567505836487,
+      "learning_rate": 4.585e-05,
+      "loss": 2.3485,
+      "step": 166
+    },
+    {
+      "epoch": 0.01864274924734784,
+      "grad_norm": 0.2774522006511688,
+      "learning_rate": 4.5825e-05,
+      "loss": 2.4186,
+      "step": 167
+    },
+    {
+      "epoch": 0.018754382476373874,
+      "grad_norm": 0.2742158770561218,
+      "learning_rate": 4.58e-05,
+      "loss": 2.3706,
+      "step": 168
+    },
+    {
+      "epoch": 0.01886601570539991,
+      "grad_norm": 0.27442091703414917,
+      "learning_rate": 4.5775e-05,
+      "loss": 2.4425,
+      "step": 169
+    },
+    {
+      "epoch": 0.018977648934425943,
+      "grad_norm": 0.2682335376739502,
+      "learning_rate": 4.575e-05,
+      "loss": 2.3597,
+      "step": 170
+    },
+    {
+      "epoch": 0.01908928216345198,
+      "grad_norm": 0.27973178029060364,
+      "learning_rate": 4.5725e-05,
+      "loss": 2.2992,
+      "step": 171
+    },
+    {
+      "epoch": 0.019200915392478014,
+      "grad_norm": 0.3113536536693573,
+      "learning_rate": 4.5700000000000006e-05,
+      "loss": 2.4343,
+      "step": 172
+    },
+    {
+      "epoch": 0.01931254862150405,
+      "grad_norm": 0.27050501108169556,
+      "learning_rate": 4.5675e-05,
+      "loss": 2.4023,
+      "step": 173
+    },
+    {
+      "epoch": 0.019424181850530083,
+      "grad_norm": 0.5937790870666504,
+      "learning_rate": 4.5650000000000005e-05,
+      "loss": 2.4726,
+      "step": 174
+    },
+    {
+      "epoch": 0.019535815079556117,
+      "grad_norm": 0.27320945262908936,
+      "learning_rate": 4.5625e-05,
+      "loss": 2.4108,
+      "step": 175
+    },
+    {
+      "epoch": 0.019647448308582154,
+      "grad_norm": 0.2724778354167938,
+      "learning_rate": 4.5600000000000004e-05,
+      "loss": 2.2204,
+      "step": 176
+    },
+    {
+      "epoch": 0.01975908153760819,
+      "grad_norm": 0.2760343849658966,
+      "learning_rate": 4.5575e-05,
+      "loss": 2.4108,
+      "step": 177
+    },
+    {
+      "epoch": 0.019870714766634223,
+      "grad_norm": 0.27173370122909546,
+      "learning_rate": 4.555e-05,
+      "loss": 2.4278,
+      "step": 178
+    },
+    {
+      "epoch": 0.019982347995660257,
+      "grad_norm": 0.258478045463562,
+      "learning_rate": 4.5525e-05,
+      "loss": 2.3129,
+      "step": 179
+    },
+    {
+      "epoch": 0.020093981224686294,
+      "grad_norm": 0.2680318355560303,
+      "learning_rate": 4.55e-05,
+      "loss": 2.1764,
+      "step": 180
+    },
+    {
+      "epoch": 0.02020561445371233,
+      "grad_norm": 0.32532161474227905,
+      "learning_rate": 4.5475e-05,
+      "loss": 2.3988,
+      "step": 181
+    },
+    {
+      "epoch": 0.020317247682738362,
+      "grad_norm": 0.27205905318260193,
+      "learning_rate": 4.545000000000001e-05,
+      "loss": 2.3432,
+      "step": 182
+    },
+    {
+      "epoch": 0.020428880911764397,
+      "grad_norm": 0.3422660827636719,
+      "learning_rate": 4.5425e-05,
+      "loss": 2.4591,
+      "step": 183
+    },
+    {
+      "epoch": 0.020540514140790434,
+      "grad_norm": 0.2941705882549286,
+      "learning_rate": 4.5400000000000006e-05,
+      "loss": 2.4285,
+      "step": 184
+    },
+    {
+      "epoch": 0.02065214736981647,
+      "grad_norm": 0.27218639850616455,
+      "learning_rate": 4.5375e-05,
+      "loss": 2.3349,
+      "step": 185
+    },
+    {
+      "epoch": 0.020763780598842502,
+      "grad_norm": 0.26361867785453796,
+      "learning_rate": 4.5350000000000005e-05,
+      "loss": 2.3049,
+      "step": 186
+    },
+    {
+      "epoch": 0.020875413827868537,
+      "grad_norm": 0.47230780124664307,
+      "learning_rate": 4.5325000000000004e-05,
+      "loss": 2.3365,
+      "step": 187
+    },
+    {
+      "epoch": 0.020987047056894574,
+      "grad_norm": 0.26752769947052,
+      "learning_rate": 4.53e-05,
+      "loss": 2.3887,
+      "step": 188
+    },
+    {
+      "epoch": 0.021098680285920608,
+      "grad_norm": 0.27140894532203674,
+      "learning_rate": 4.5275e-05,
+      "loss": 2.4444,
+      "step": 189
+    },
+    {
+      "epoch": 0.021210313514946642,
+      "grad_norm": 0.4308728873729706,
+      "learning_rate": 4.525e-05,
+      "loss": 2.397,
+      "step": 190
+    },
+    {
+      "epoch": 0.021321946743972676,
+      "grad_norm": 0.6893981099128723,
+      "learning_rate": 4.5225e-05,
+      "loss": 2.3797,
+      "step": 191
+    },
+    {
+      "epoch": 0.021433579972998714,
+      "grad_norm": 0.27642473578453064,
+      "learning_rate": 4.52e-05,
+      "loss": 2.3374,
+      "step": 192
+    },
+    {
+      "epoch": 0.021545213202024748,
+      "grad_norm": 0.2684485912322998,
+      "learning_rate": 4.5175e-05,
+      "loss": 2.3368,
+      "step": 193
+    },
+    {
+      "epoch": 0.021656846431050782,
+      "grad_norm": 0.2804638743400574,
+      "learning_rate": 4.5150000000000006e-05,
+      "loss": 2.3378,
+      "step": 194
+    },
+    {
+      "epoch": 0.021768479660076816,
+      "grad_norm": 0.41617271304130554,
+      "learning_rate": 4.5125e-05,
+      "loss": 2.3908,
+      "step": 195
+    },
+    {
+      "epoch": 0.021880112889102854,
+      "grad_norm": 0.26013967394828796,
+      "learning_rate": 4.5100000000000005e-05,
+      "loss": 2.3378,
+      "step": 196
+    },
+    {
+      "epoch": 0.021991746118128888,
+      "grad_norm": 0.2883068919181824,
+      "learning_rate": 4.5075e-05,
+      "loss": 2.296,
+      "step": 197
+    },
+    {
+      "epoch": 0.022103379347154922,
+      "grad_norm": 0.27083417773246765,
+      "learning_rate": 4.5050000000000004e-05,
+      "loss": 2.3917,
+      "step": 198
+    },
+    {
+      "epoch": 0.022215012576180956,
+      "grad_norm": 0.26112979650497437,
+      "learning_rate": 4.5025000000000003e-05,
+      "loss": 2.4321,
+      "step": 199
+    },
+    {
+      "epoch": 0.022326645805206994,
+      "grad_norm": 0.2797684669494629,
+      "learning_rate": 4.5e-05,
+      "loss": 2.4221,
+      "step": 200
+    },
+    {
+      "epoch": 0.022438279034233028,
+      "grad_norm": 0.28574231266975403,
+      "learning_rate": 4.4975e-05,
+      "loss": 2.4143,
+      "step": 201
+    },
+    {
+      "epoch": 0.022549912263259062,
+      "grad_norm": 0.3054039180278778,
+      "learning_rate": 4.495e-05,
+      "loss": 2.3744,
+      "step": 202
+    },
+    {
+      "epoch": 0.022661545492285096,
+      "grad_norm": 0.2859933376312256,
+      "learning_rate": 4.4925e-05,
+      "loss": 2.3158,
+      "step": 203
+    },
+    {
+      "epoch": 0.022773178721311134,
+      "grad_norm": 0.30749940872192383,
+      "learning_rate": 4.49e-05,
+      "loss": 2.3274,
+      "step": 204
+    },
+    {
+      "epoch": 0.022884811950337168,
+      "grad_norm": 0.34303340315818787,
+      "learning_rate": 4.4875e-05,
+      "loss": 2.3822,
+      "step": 205
+    },
+    {
+      "epoch": 0.022996445179363202,
+      "grad_norm": 0.6377202868461609,
+      "learning_rate": 4.4850000000000006e-05,
+      "loss": 2.3162,
+      "step": 206
+    },
+    {
+      "epoch": 0.023108078408389236,
+      "grad_norm": 0.27729642391204834,
+      "learning_rate": 4.4825e-05,
+      "loss": 2.3117,
+      "step": 207
+    },
+    {
+      "epoch": 0.02321971163741527,
+      "grad_norm": 0.2766030728816986,
+      "learning_rate": 4.4800000000000005e-05,
+      "loss": 2.4142,
+      "step": 208
+    },
+    {
+      "epoch": 0.023331344866441308,
+      "grad_norm": 0.5488070249557495,
+      "learning_rate": 4.4775e-05,
+      "loss": 2.3999,
+      "step": 209
+    },
+    {
+      "epoch": 0.023442978095467342,
+      "grad_norm": 0.27591627836227417,
+      "learning_rate": 4.4750000000000004e-05,
+      "loss": 2.4271,
+      "step": 210
+    },
+    {
+      "epoch": 0.023554611324493376,
+      "grad_norm": 0.3173430562019348,
+      "learning_rate": 4.4725e-05,
+      "loss": 2.4497,
+      "step": 211
+    },
+    {
+      "epoch": 0.02366624455351941,
+      "grad_norm": 0.31145554780960083,
+      "learning_rate": 4.47e-05,
+      "loss": 2.4128,
+      "step": 212
+    },
+    {
+      "epoch": 0.023777877782545448,
+      "grad_norm": 0.27966129779815674,
+      "learning_rate": 4.4675e-05,
+      "loss": 2.4699,
+      "step": 213
+    },
+    {
+      "epoch": 0.023889511011571482,
+      "grad_norm": 0.31131359934806824,
+      "learning_rate": 4.465e-05,
+      "loss": 2.3269,
+      "step": 214
+    },
+    {
+      "epoch": 0.024001144240597516,
+      "grad_norm": 0.26797181367874146,
+      "learning_rate": 4.4625e-05,
+      "loss": 2.4169,
+      "step": 215
+    },
+    {
+      "epoch": 0.02411277746962355,
+      "grad_norm": 0.28204289078712463,
+      "learning_rate": 4.46e-05,
+      "loss": 2.2361,
+      "step": 216
+    },
+    {
+      "epoch": 0.024224410698649588,
+      "grad_norm": 0.2600002586841583,
+      "learning_rate": 4.4575e-05,
+      "loss": 2.3357,
+      "step": 217
+    },
+    {
+      "epoch": 0.024336043927675622,
+      "grad_norm": 0.2576424777507782,
+      "learning_rate": 4.4550000000000005e-05,
+      "loss": 2.3424,
+      "step": 218
+    },
+    {
+      "epoch": 0.024447677156701656,
+      "grad_norm": 0.3016074299812317,
+      "learning_rate": 4.4525e-05,
+      "loss": 2.3781,
+      "step": 219
+    },
+    {
+      "epoch": 0.02455931038572769,
+      "grad_norm": 0.2684342563152313,
+      "learning_rate": 4.4500000000000004e-05,
+      "loss": 2.3167,
+      "step": 220
+    },
+    {
+      "epoch": 0.024670943614753728,
+      "grad_norm": 0.2625711262226105,
+      "learning_rate": 4.4475e-05,
+      "loss": 2.4207,
+      "step": 221
+    },
+    {
+      "epoch": 0.024782576843779762,
+      "grad_norm": 0.29141953587532043,
+      "learning_rate": 4.445e-05,
+      "loss": 2.4231,
+      "step": 222
+    },
+    {
+      "epoch": 0.024894210072805796,
+      "grad_norm": 0.2682022452354431,
+      "learning_rate": 4.4425e-05,
+      "loss": 2.4098,
+      "step": 223
+    },
+    {
+      "epoch": 0.02500584330183183,
+      "grad_norm": 0.2591974437236786,
+      "learning_rate": 4.44e-05,
+      "loss": 2.4387,
+      "step": 224
+    },
+    {
+      "epoch": 0.025117476530857868,
+      "grad_norm": 0.2656046152114868,
+      "learning_rate": 4.4375e-05,
+      "loss": 2.415,
+      "step": 225
+    },
+    {
+      "epoch": 0.025229109759883902,
+      "grad_norm": 0.2568715214729309,
+      "learning_rate": 4.435e-05,
+      "loss": 2.3837,
+      "step": 226
+    },
+    {
+      "epoch": 0.025340742988909936,
+      "grad_norm": 0.3120313286781311,
+      "learning_rate": 4.4325e-05,
+      "loss": 2.4028,
+      "step": 227
+    },
+    {
+      "epoch": 0.02545237621793597,
+      "grad_norm": 0.36895328760147095,
+      "learning_rate": 4.43e-05,
+      "loss": 2.4162,
+      "step": 228
+    },
+    {
+      "epoch": 0.025564009446962008,
+      "grad_norm": 0.2681656777858734,
+      "learning_rate": 4.4275e-05,
+      "loss": 2.4289,
+      "step": 229
+    },
+    {
+      "epoch": 0.025675642675988042,
+      "grad_norm": 0.2715415358543396,
+      "learning_rate": 4.4250000000000005e-05,
+      "loss": 2.3207,
+      "step": 230
+    },
+    {
+      "epoch": 0.025787275905014076,
+      "grad_norm": 0.2677493989467621,
+      "learning_rate": 4.4225e-05,
+      "loss": 2.3856,
+      "step": 231
+    },
+    {
+      "epoch": 0.02589890913404011,
+      "grad_norm": 0.25962767004966736,
+      "learning_rate": 4.4200000000000004e-05,
+      "loss": 2.3136,
+      "step": 232
+    },
+    {
+      "epoch": 0.026010542363066148,
+      "grad_norm": 0.3225052058696747,
+      "learning_rate": 4.4174999999999996e-05,
+      "loss": 2.2765,
+      "step": 233
+    },
+    {
+      "epoch": 0.026122175592092182,
+      "grad_norm": 0.3049544095993042,
+      "learning_rate": 4.415e-05,
+      "loss": 2.2127,
+      "step": 234
+    },
+    {
+      "epoch": 0.026233808821118216,
+      "grad_norm": 0.25900280475616455,
+      "learning_rate": 4.4125e-05,
+      "loss": 2.3794,
+      "step": 235
+    },
+    {
+      "epoch": 0.02634544205014425,
+      "grad_norm": 0.26174089312553406,
+      "learning_rate": 4.41e-05,
+      "loss": 2.4024,
+      "step": 236
+    },
+    {
+      "epoch": 0.026457075279170288,
+      "grad_norm": 0.26936131715774536,
+      "learning_rate": 4.4075e-05,
+      "loss": 2.5543,
+      "step": 237
+    },
+    {
+      "epoch": 0.026568708508196322,
+      "grad_norm": 0.2539876103401184,
+      "learning_rate": 4.405e-05,
+      "loss": 2.4051,
+      "step": 238
+    },
+    {
+      "epoch": 0.026680341737222356,
+      "grad_norm": 0.26477983593940735,
+      "learning_rate": 4.4025e-05,
+      "loss": 2.3512,
+      "step": 239
+    },
+    {
+      "epoch": 0.02679197496624839,
+      "grad_norm": 0.2632873058319092,
+      "learning_rate": 4.4000000000000006e-05,
+      "loss": 2.3378,
+      "step": 240
+    },
+    {
+      "epoch": 0.026903608195274428,
+      "grad_norm": 0.6845733523368835,
+      "learning_rate": 4.3975e-05,
+      "loss": 2.3815,
+      "step": 241
+    },
+    {
+      "epoch": 0.02701524142430046,
+      "grad_norm": 0.25974923372268677,
+      "learning_rate": 4.3950000000000004e-05,
+      "loss": 2.3816,
+      "step": 242
+    },
+    {
+      "epoch": 0.027126874653326496,
+      "grad_norm": 0.2636438012123108,
+      "learning_rate": 4.3925e-05,
+      "loss": 2.3288,
+      "step": 243
+    },
+    {
+      "epoch": 0.02723850788235253,
+      "grad_norm": 0.2576185464859009,
+      "learning_rate": 4.39e-05,
+      "loss": 2.4314,
+      "step": 244
+    },
+    {
+      "epoch": 0.027350141111378564,
+      "grad_norm": 0.2600337862968445,
+      "learning_rate": 4.3875e-05,
+      "loss": 2.4147,
+      "step": 245
+    },
+    {
+      "epoch": 0.0274617743404046,
+      "grad_norm": 0.2605160176753998,
+      "learning_rate": 4.385e-05,
+      "loss": 2.327,
+      "step": 246
+    },
+    {
+      "epoch": 0.027573407569430636,
+      "grad_norm": 0.25381141901016235,
+      "learning_rate": 4.3825e-05,
+      "loss": 2.309,
+      "step": 247
+    },
+    {
+      "epoch": 0.02768504079845667,
+      "grad_norm": 0.253326416015625,
+      "learning_rate": 4.38e-05,
+      "loss": 2.2849,
+      "step": 248
+    },
+    {
+      "epoch": 0.027796674027482704,
+      "grad_norm": 0.3015645444393158,
+      "learning_rate": 4.3775e-05,
+      "loss": 2.3882,
+      "step": 249
+    },
+    {
+      "epoch": 0.02790830725650874,
+      "grad_norm": 0.2591153383255005,
+      "learning_rate": 4.375e-05,
+      "loss": 2.3108,
+      "step": 250
+    },
+    {
+      "epoch": 0.028019940485534776,
+      "grad_norm": 0.255209743976593,
+      "learning_rate": 4.3725000000000006e-05,
+      "loss": 2.3026,
+      "step": 251
+    },
+    {
+      "epoch": 0.02813157371456081,
+      "grad_norm": 0.2543400228023529,
+      "learning_rate": 4.3700000000000005e-05,
+      "loss": 2.3949,
+      "step": 252
+    },
+    {
+      "epoch": 0.028243206943586844,
+      "grad_norm": 0.2584831118583679,
+      "learning_rate": 4.3675000000000005e-05,
+      "loss": 2.4274,
+      "step": 253
+    },
+    {
+      "epoch": 0.02835484017261288,
+      "grad_norm": 0.24946770071983337,
+      "learning_rate": 4.3650000000000004e-05,
+      "loss": 2.378,
+      "step": 254
+    },
+    {
+      "epoch": 0.028466473401638916,
+      "grad_norm": 0.2595466673374176,
+      "learning_rate": 4.3625e-05,
+      "loss": 2.4601,
+      "step": 255
+    },
+    {
+      "epoch": 0.02857810663066495,
+      "grad_norm": 0.25328534841537476,
+      "learning_rate": 4.36e-05,
+      "loss": 2.2783,
+      "step": 256
+    },
+    {
+      "epoch": 0.028689739859690984,
+      "grad_norm": 0.2810356616973877,
+      "learning_rate": 4.3575e-05,
+      "loss": 2.282,
+      "step": 257
+    },
+    {
+      "epoch": 0.02880137308871702,
+      "grad_norm": 0.2603547275066376,
+      "learning_rate": 4.355e-05,
+      "loss": 2.3768,
+      "step": 258
+    },
+    {
+      "epoch": 0.028913006317743056,
+      "grad_norm": 0.25737640261650085,
+      "learning_rate": 4.352500000000001e-05,
+      "loss": 2.4577,
+      "step": 259
+    },
+    {
+      "epoch": 0.02902463954676909,
+      "grad_norm": 0.25266796350479126,
+      "learning_rate": 4.35e-05,
+      "loss": 2.3293,
+      "step": 260
+    },
+    {
+      "epoch": 0.029136272775795124,
+      "grad_norm": 0.35959863662719727,
+      "learning_rate": 4.3475000000000006e-05,
+      "loss": 2.2827,
+      "step": 261
+    },
+    {
+      "epoch": 0.02924790600482116,
+      "grad_norm": 0.2564973831176758,
+      "learning_rate": 4.345e-05,
+      "loss": 2.3925,
+      "step": 262
+    },
+    {
+      "epoch": 0.029359539233847196,
+      "grad_norm": 0.2848140597343445,
+      "learning_rate": 4.3425000000000005e-05,
+      "loss": 2.3275,
+      "step": 263
+    },
+    {
+      "epoch": 0.02947117246287323,
+      "grad_norm": 0.26803499460220337,
+      "learning_rate": 4.3400000000000005e-05,
+      "loss": 2.4576,
+      "step": 264
+    },
+    {
+      "epoch": 0.029582805691899264,
+      "grad_norm": 0.24815724790096283,
+      "learning_rate": 4.3375000000000004e-05,
+      "loss": 2.611,
+      "step": 265
+    },
+    {
+      "epoch": 0.0296944389209253,
+      "grad_norm": 0.26036569476127625,
+      "learning_rate": 4.335e-05,
+      "loss": 2.3574,
+      "step": 266
+    },
+    {
+      "epoch": 0.029806072149951335,
+      "grad_norm": 0.26427459716796875,
+      "learning_rate": 4.3325e-05,
+      "loss": 2.3884,
+      "step": 267
+    },
+    {
+      "epoch": 0.02991770537897737,
+      "grad_norm": 0.25177526473999023,
+      "learning_rate": 4.33e-05,
+      "loss": 2.3812,
+      "step": 268
+    },
+    {
+      "epoch": 0.030029338608003404,
+      "grad_norm": 0.2582986354827881,
+      "learning_rate": 4.3275e-05,
+      "loss": 2.3621,
+      "step": 269
+    },
+    {
+      "epoch": 0.03014097183702944,
+      "grad_norm": 0.26251325011253357,
+      "learning_rate": 4.325e-05,
+      "loss": 2.4779,
+      "step": 270
+    },
+    {
+      "epoch": 0.030252605066055475,
+      "grad_norm": 0.2560170590877533,
+      "learning_rate": 4.322500000000001e-05,
+      "loss": 2.4519,
+      "step": 271
+    },
+    {
+      "epoch": 0.03036423829508151,
+      "grad_norm": 0.25769442319869995,
+      "learning_rate": 4.32e-05,
+      "loss": 2.255,
+      "step": 272
+    },
+    {
+      "epoch": 0.030475871524107544,
+      "grad_norm": 0.2584100067615509,
+      "learning_rate": 4.3175000000000006e-05,
+      "loss": 2.4195,
+      "step": 273
+    },
+    {
+      "epoch": 0.03058750475313358,
+      "grad_norm": 0.26976278424263,
+      "learning_rate": 4.315e-05,
+      "loss": 2.3927,
+      "step": 274
+    },
+    {
+      "epoch": 0.030699137982159615,
+      "grad_norm": 0.2528376877307892,
+      "learning_rate": 4.3125000000000005e-05,
+      "loss": 2.3596,
+      "step": 275
+    },
+    {
+      "epoch": 0.03081077121118565,
+      "grad_norm": 0.25755786895751953,
+      "learning_rate": 4.3100000000000004e-05,
+      "loss": 2.4775,
+      "step": 276
+    },
+    {
+      "epoch": 0.030922404440211684,
+      "grad_norm": 0.25737857818603516,
+      "learning_rate": 4.3075000000000003e-05,
+      "loss": 2.3443,
+      "step": 277
+    },
+    {
+      "epoch": 0.031034037669237718,
+      "grad_norm": 0.2632676362991333,
+      "learning_rate": 4.305e-05,
+      "loss": 2.3128,
+      "step": 278
+    },
+    {
+      "epoch": 0.031145670898263755,
+      "grad_norm": 0.28831636905670166,
+      "learning_rate": 4.3025e-05,
+      "loss": 2.3624,
+      "step": 279
+    },
+    {
+      "epoch": 0.03125730412728979,
+      "grad_norm": 0.2593206763267517,
+      "learning_rate": 4.3e-05,
+      "loss": 2.3546,
+      "step": 280
+    },
+    {
+      "epoch": 0.031368937356315824,
+      "grad_norm": 0.25221961736679077,
+      "learning_rate": 4.2975e-05,
+      "loss": 2.3056,
+      "step": 281
+    },
+    {
+      "epoch": 0.03148057058534186,
+      "grad_norm": 0.26562732458114624,
+      "learning_rate": 4.295e-05,
+      "loss": 2.2979,
+      "step": 282
+    },
+    {
+      "epoch": 0.03159220381436789,
+      "grad_norm": 0.2667911648750305,
+      "learning_rate": 4.2925000000000007e-05,
+      "loss": 2.5051,
+      "step": 283
+    },
+    {
+      "epoch": 0.031703837043393926,
+      "grad_norm": 0.2637230455875397,
+      "learning_rate": 4.29e-05,
+      "loss": 2.2956,
+      "step": 284
+    },
+    {
+      "epoch": 0.03181547027241997,
+      "grad_norm": 0.2591506540775299,
+      "learning_rate": 4.2875000000000005e-05,
+      "loss": 2.4331,
+      "step": 285
+    },
+    {
+      "epoch": 0.031927103501446,
+      "grad_norm": 0.25945866107940674,
+      "learning_rate": 4.285e-05,
+      "loss": 2.3397,
+      "step": 286
+    },
+    {
+      "epoch": 0.032038736730472035,
+      "grad_norm": 0.28612184524536133,
+      "learning_rate": 4.2825000000000004e-05,
+      "loss": 2.4108,
+      "step": 287
+    },
+    {
+      "epoch": 0.03215036995949807,
+      "grad_norm": 0.26425275206565857,
+      "learning_rate": 4.2800000000000004e-05,
+      "loss": 2.4308,
+      "step": 288
+    },
+    {
+      "epoch": 0.032262003188524103,
+      "grad_norm": 0.2575188875198364,
+      "learning_rate": 4.2775e-05,
+      "loss": 2.3774,
+      "step": 289
+    },
+    {
+      "epoch": 0.03237363641755014,
+      "grad_norm": 0.25762301683425903,
+      "learning_rate": 4.275e-05,
+      "loss": 2.4266,
+      "step": 290
+    },
+    {
+      "epoch": 0.03248526964657617,
+      "grad_norm": 0.2451835572719574,
+      "learning_rate": 4.2725e-05,
+      "loss": 2.2542,
+      "step": 291
+    },
+    {
+      "epoch": 0.032596902875602206,
+      "grad_norm": 0.25459688901901245,
+      "learning_rate": 4.27e-05,
+      "loss": 2.3772,
+      "step": 292
+    },
+    {
+      "epoch": 0.03270853610462825,
+      "grad_norm": 0.3179572820663452,
+      "learning_rate": 4.2675e-05,
+      "loss": 2.4139,
+      "step": 293
+    },
+    {
+      "epoch": 0.03282016933365428,
+      "grad_norm": 0.25751742720603943,
+      "learning_rate": 4.265e-05,
+      "loss": 2.2372,
+      "step": 294
+    },
+    {
+      "epoch": 0.032931802562680315,
+      "grad_norm": 0.26437416672706604,
+      "learning_rate": 4.2625000000000006e-05,
+      "loss": 2.4512,
+      "step": 295
+    },
+    {
+      "epoch": 0.03304343579170635,
+      "grad_norm": 0.26216599345207214,
+      "learning_rate": 4.26e-05,
+      "loss": 2.3041,
+      "step": 296
+    },
+    {
+      "epoch": 0.03315506902073238,
+      "grad_norm": 0.2558748126029968,
+      "learning_rate": 4.2575000000000005e-05,
+      "loss": 2.3464,
+      "step": 297
+    },
+    {
+      "epoch": 0.03326670224975842,
+      "grad_norm": 0.25759217143058777,
+      "learning_rate": 4.2550000000000004e-05,
+      "loss": 2.38,
+      "step": 298
+    },
+    {
+      "epoch": 0.03337833547878445,
+      "grad_norm": 0.2515701353549957,
+      "learning_rate": 4.2525000000000004e-05,
+      "loss": 2.347,
+      "step": 299
+    },
+    {
+      "epoch": 0.033489968707810486,
+      "grad_norm": 0.25506308674812317,
+      "learning_rate": 4.25e-05,
+      "loss": 2.4563,
+      "step": 300
+    },
+    {
+      "epoch": 0.03360160193683653,
+      "grad_norm": 0.24955067038536072,
+      "learning_rate": 4.2475e-05,
+      "loss": 2.3467,
+      "step": 301
+    },
+    {
+      "epoch": 0.03371323516586256,
+      "grad_norm": 0.26849493384361267,
+      "learning_rate": 4.245e-05,
+      "loss": 2.318,
+      "step": 302
+    },
+    {
+      "epoch": 0.033824868394888595,
+      "grad_norm": 0.2560291886329651,
+      "learning_rate": 4.2425e-05,
+      "loss": 2.3567,
+      "step": 303
+    },
+    {
+      "epoch": 0.03393650162391463,
+      "grad_norm": 0.2685459554195404,
+      "learning_rate": 4.24e-05,
+      "loss": 2.3929,
+      "step": 304
+    },
+    {
+      "epoch": 0.03404813485294066,
+      "grad_norm": 0.2724890410900116,
+      "learning_rate": 4.237500000000001e-05,
+      "loss": 2.321,
+      "step": 305
+    },
+    {
+      "epoch": 0.0341597680819667,
+      "grad_norm": 0.3363018333911896,
+      "learning_rate": 4.235e-05,
+      "loss": 2.2429,
+      "step": 306
+    },
+    {
+      "epoch": 0.03427140131099273,
+      "grad_norm": 0.2732946276664734,
+      "learning_rate": 4.2325000000000006e-05,
+      "loss": 2.4269,
+      "step": 307
+    },
+    {
+      "epoch": 0.034383034540018766,
+      "grad_norm": 0.25203391909599304,
+      "learning_rate": 4.23e-05,
+      "loss": 2.3134,
+      "step": 308
+    },
+    {
+      "epoch": 0.03449466776904481,
+      "grad_norm": 0.27153274416923523,
+      "learning_rate": 4.2275000000000004e-05,
+      "loss": 2.3181,
+      "step": 309
+    },
+    {
+      "epoch": 0.03460630099807084,
+      "grad_norm": 0.25666430592536926,
+      "learning_rate": 4.2250000000000004e-05,
+      "loss": 2.437,
+      "step": 310
+    },
+    {
+      "epoch": 0.034717934227096875,
+      "grad_norm": 0.255957692861557,
+      "learning_rate": 4.2225e-05,
+      "loss": 2.3226,
+      "step": 311
+    },
+    {
+      "epoch": 0.03482956745612291,
+      "grad_norm": 0.2525959014892578,
+      "learning_rate": 4.22e-05,
+      "loss": 2.4727,
+      "step": 312
+    },
+    {
+      "epoch": 0.03494120068514894,
+      "grad_norm": 0.34017762541770935,
+      "learning_rate": 4.2175e-05,
+      "loss": 2.2816,
+      "step": 313
+    },
+    {
+      "epoch": 0.03505283391417498,
+      "grad_norm": 0.25117596983909607,
+      "learning_rate": 4.215e-05,
+      "loss": 2.3801,
+      "step": 314
+    },
+    {
+      "epoch": 0.03516446714320101,
+      "grad_norm": 0.2531397044658661,
+      "learning_rate": 4.2125e-05,
+      "loss": 2.4461,
+      "step": 315
+    },
+    {
+      "epoch": 0.035276100372227046,
+      "grad_norm": 0.2584179639816284,
+      "learning_rate": 4.21e-05,
+      "loss": 2.2489,
+      "step": 316
+    },
+    {
+      "epoch": 0.03538773360125308,
+      "grad_norm": 0.25839826464653015,
+      "learning_rate": 4.2075000000000006e-05,
+      "loss": 2.3748,
+      "step": 317
+    },
+    {
+      "epoch": 0.03549936683027912,
+      "grad_norm": 0.2568003833293915,
+      "learning_rate": 4.205e-05,
+      "loss": 2.3484,
+      "step": 318
+    },
+    {
+      "epoch": 0.035611000059305155,
+      "grad_norm": 0.2569085657596588,
+      "learning_rate": 4.2025000000000005e-05,
+      "loss": 2.4397,
+      "step": 319
+    },
+    {
+      "epoch": 0.03572263328833119,
+      "grad_norm": 0.24986624717712402,
+      "learning_rate": 4.2e-05,
+      "loss": 2.3008,
+      "step": 320
+    },
+    {
+      "epoch": 0.03583426651735722,
+      "grad_norm": 0.26116469502449036,
+      "learning_rate": 4.1975000000000004e-05,
+      "loss": 2.2436,
+      "step": 321
+    },
+    {
+      "epoch": 0.03594589974638326,
+      "grad_norm": 0.3085286021232605,
+      "learning_rate": 4.195e-05,
+      "loss": 2.2419,
+      "step": 322
+    },
+    {
+      "epoch": 0.03605753297540929,
+      "grad_norm": 1.1896109580993652,
+      "learning_rate": 4.1925e-05,
+      "loss": 2.3528,
+      "step": 323
+    },
+    {
+      "epoch": 0.036169166204435325,
+      "grad_norm": 0.26796790957450867,
+      "learning_rate": 4.19e-05,
+      "loss": 2.314,
+      "step": 324
+    },
+    {
+      "epoch": 0.03628079943346136,
+      "grad_norm": 0.27274981141090393,
+      "learning_rate": 4.1875e-05,
+      "loss": 2.3397,
+      "step": 325
+    },
+    {
+      "epoch": 0.0363924326624874,
+      "grad_norm": 0.2676389813423157,
+      "learning_rate": 4.185e-05,
+      "loss": 2.4168,
+      "step": 326
+    },
+    {
+      "epoch": 0.036504065891513435,
+      "grad_norm": 0.4810822904109955,
+      "learning_rate": 4.1825e-05,
+      "loss": 2.3375,
+      "step": 327
+    },
+    {
+      "epoch": 0.03661569912053947,
+      "grad_norm": 0.28411370515823364,
+      "learning_rate": 4.18e-05,
+      "loss": 2.2704,
+      "step": 328
+    },
+    {
+      "epoch": 0.0367273323495655,
+      "grad_norm": 0.27649131417274475,
+      "learning_rate": 4.1775000000000006e-05,
+      "loss": 2.3377,
+      "step": 329
+    },
+    {
+      "epoch": 0.03683896557859154,
+      "grad_norm": 0.25533196330070496,
+      "learning_rate": 4.175e-05,
+      "loss": 2.2652,
+      "step": 330
+    },
+    {
+      "epoch": 0.03695059880761757,
+      "grad_norm": 0.2640891969203949,
+      "learning_rate": 4.1725000000000005e-05,
+      "loss": 2.3504,
+      "step": 331
+    },
+    {
+      "epoch": 0.037062232036643605,
+      "grad_norm": 0.2534253001213074,
+      "learning_rate": 4.17e-05,
+      "loss": 2.4219,
+      "step": 332
+    },
+    {
+      "epoch": 0.03717386526566964,
+      "grad_norm": 0.2603084146976471,
+      "learning_rate": 4.1675e-05,
+      "loss": 2.3582,
+      "step": 333
+    },
+    {
+      "epoch": 0.03728549849469568,
+      "grad_norm": 0.2546936273574829,
+      "learning_rate": 4.165e-05,
+      "loss": 2.3509,
+      "step": 334
+    },
+    {
+      "epoch": 0.037397131723721715,
+      "grad_norm": 0.2537340819835663,
+      "learning_rate": 4.1625e-05,
+      "loss": 2.3683,
+      "step": 335
+    },
+    {
+      "epoch": 0.03750876495274775,
+      "grad_norm": 0.26713791489601135,
+      "learning_rate": 4.16e-05,
+      "loss": 2.3843,
+      "step": 336
+    },
+    {
+      "epoch": 0.03762039818177378,
+      "grad_norm": 0.2577325105667114,
+      "learning_rate": 4.1575e-05,
+      "loss": 2.4786,
+      "step": 337
+    },
+    {
+      "epoch": 0.03773203141079982,
+      "grad_norm": 0.24003548920154572,
+      "learning_rate": 4.155e-05,
+      "loss": 2.4079,
+      "step": 338
+    },
+    {
+      "epoch": 0.03784366463982585,
+      "grad_norm": 0.26992398500442505,
+      "learning_rate": 4.1525e-05,
+      "loss": 2.382,
+      "step": 339
+    },
+    {
+      "epoch": 0.037955297868851885,
+      "grad_norm": 0.2569391131401062,
+      "learning_rate": 4.15e-05,
+      "loss": 2.2762,
+      "step": 340
+    },
+    {
+      "epoch": 0.03806693109787792,
+      "grad_norm": 0.2576649785041809,
+      "learning_rate": 4.1475000000000005e-05,
+      "loss": 2.3492,
+      "step": 341
+    },
+    {
+      "epoch": 0.03817856432690396,
+      "grad_norm": 0.2680610716342926,
+      "learning_rate": 4.145e-05,
+      "loss": 2.3769,
+      "step": 342
+    },
+    {
+      "epoch": 0.038290197555929995,
+      "grad_norm": 0.2511936128139496,
+      "learning_rate": 4.1425000000000004e-05,
+      "loss": 2.2633,
+      "step": 343
+    },
+    {
+      "epoch": 0.03840183078495603,
+      "grad_norm": 0.2549611032009125,
+      "learning_rate": 4.14e-05,
+      "loss": 2.4439,
+      "step": 344
+    },
+    {
+      "epoch": 0.03851346401398206,
+      "grad_norm": 0.24939557909965515,
+      "learning_rate": 4.1375e-05,
+      "loss": 2.3102,
+      "step": 345
+    },
+    {
+      "epoch": 0.0386250972430081,
+      "grad_norm": 0.2684389650821686,
+      "learning_rate": 4.135e-05,
+      "loss": 2.3636,
+      "step": 346
+    },
+    {
+      "epoch": 0.03873673047203413,
+      "grad_norm": 0.24813228845596313,
+      "learning_rate": 4.1325e-05,
+      "loss": 2.3439,
+      "step": 347
+    },
+    {
+      "epoch": 0.038848363701060165,
+      "grad_norm": 0.24416686594486237,
+      "learning_rate": 4.13e-05,
+      "loss": 2.256,
+      "step": 348
+    },
+    {
+      "epoch": 0.0389599969300862,
+      "grad_norm": 0.24512575566768646,
+      "learning_rate": 4.1275e-05,
+      "loss": 2.3966,
+      "step": 349
+    },
+    {
+      "epoch": 0.03907163015911223,
+      "grad_norm": 0.25613388419151306,
+      "learning_rate": 4.125e-05,
+      "loss": 2.3726,
+      "step": 350
+    },
+    {
+      "epoch": 0.039183263388138274,
+      "grad_norm": 0.2553405463695526,
+      "learning_rate": 4.1225e-05,
+      "loss": 2.4718,
+      "step": 351
+    },
+    {
+      "epoch": 0.03929489661716431,
+      "grad_norm": 0.2818881869316101,
+      "learning_rate": 4.12e-05,
+      "loss": 2.2828,
+      "step": 352
+    },
+    {
+      "epoch": 0.03940652984619034,
+      "grad_norm": 0.2522308826446533,
+      "learning_rate": 4.1175000000000005e-05,
+      "loss": 2.3791,
+      "step": 353
+    },
+    {
+      "epoch": 0.03951816307521638,
+      "grad_norm": 0.2561878561973572,
+      "learning_rate": 4.115e-05,
+      "loss": 2.4064,
+      "step": 354
+    },
+    {
+      "epoch": 0.03962979630424241,
+      "grad_norm": 0.2658417522907257,
+      "learning_rate": 4.1125000000000004e-05,
+      "loss": 2.4249,
+      "step": 355
+    },
+    {
+      "epoch": 0.039741429533268445,
+      "grad_norm": 0.3040316700935364,
+      "learning_rate": 4.11e-05,
+      "loss": 2.3588,
+      "step": 356
+    },
+    {
+      "epoch": 0.03985306276229448,
+      "grad_norm": 0.24854056537151337,
+      "learning_rate": 4.1075e-05,
+      "loss": 2.3466,
+      "step": 357
+    },
+    {
+      "epoch": 0.03996469599132051,
+      "grad_norm": 0.2535928785800934,
+      "learning_rate": 4.105e-05,
+      "loss": 2.2637,
+      "step": 358
+    },
+    {
+      "epoch": 0.040076329220346554,
+      "grad_norm": 0.2553252577781677,
+      "learning_rate": 4.1025e-05,
+      "loss": 2.3413,
+      "step": 359
+    },
+    {
+      "epoch": 0.04018796244937259,
+      "grad_norm": 0.25706836581230164,
+      "learning_rate": 4.1e-05,
+      "loss": 2.3426,
+      "step": 360
+    },
+    {
+      "epoch": 0.04029959567839862,
+      "grad_norm": 0.2549241781234741,
+      "learning_rate": 4.0975e-05,
+      "loss": 2.4028,
+      "step": 361
+    },
+    {
+      "epoch": 0.04041122890742466,
+      "grad_norm": 0.25191277265548706,
+      "learning_rate": 4.095e-05,
+      "loss": 2.3353,
+      "step": 362
+    },
+    {
+      "epoch": 0.04052286213645069,
+      "grad_norm": 0.2870761454105377,
+      "learning_rate": 4.0925000000000005e-05,
+      "loss": 2.2961,
+      "step": 363
+    },
+    {
+      "epoch": 0.040634495365476725,
+      "grad_norm": 0.26251882314682007,
+      "learning_rate": 4.09e-05,
+      "loss": 2.3897,
+      "step": 364
+    },
+    {
+      "epoch": 0.04074612859450276,
+      "grad_norm": 0.24886254966259003,
+      "learning_rate": 4.0875000000000004e-05,
+      "loss": 2.3752,
+      "step": 365
+    },
+    {
+      "epoch": 0.04085776182352879,
+      "grad_norm": 0.24439434707164764,
+      "learning_rate": 4.085e-05,
+      "loss": 2.3435,
+      "step": 366
+    },
+    {
+      "epoch": 0.040969395052554834,
+      "grad_norm": 0.41298142075538635,
+      "learning_rate": 4.0825e-05,
+      "loss": 2.4568,
+      "step": 367
+    },
+    {
+      "epoch": 0.04108102828158087,
+      "grad_norm": 0.2539288401603699,
+      "learning_rate": 4.08e-05,
+      "loss": 2.3788,
+      "step": 368
+    },
+    {
+      "epoch": 0.0411926615106069,
+      "grad_norm": 0.2483907788991928,
+      "learning_rate": 4.0775e-05,
+      "loss": 2.2744,
+      "step": 369
+    },
+    {
+      "epoch": 0.04130429473963294,
+      "grad_norm": 0.2610059678554535,
+      "learning_rate": 4.075e-05,
+      "loss": 2.4475,
+      "step": 370
+    },
+    {
+      "epoch": 0.04141592796865897,
+      "grad_norm": 0.2610619068145752,
+      "learning_rate": 4.0725e-05,
+      "loss": 2.4049,
+      "step": 371
+    },
+    {
+      "epoch": 0.041527561197685005,
+      "grad_norm": 0.26267480850219727,
+      "learning_rate": 4.07e-05,
+      "loss": 2.3503,
+      "step": 372
+    },
+    {
+      "epoch": 0.04163919442671104,
+      "grad_norm": 0.2515001893043518,
+      "learning_rate": 4.0675e-05,
+      "loss": 2.3344,
+      "step": 373
+    },
+    {
+      "epoch": 0.04175082765573707,
+      "grad_norm": 0.27069345116615295,
+      "learning_rate": 4.065e-05,
+      "loss": 2.3766,
+      "step": 374
+    },
+    {
+      "epoch": 0.041862460884763114,
+      "grad_norm": 0.2560361921787262,
+      "learning_rate": 4.0625000000000005e-05,
+      "loss": 2.3775,
+      "step": 375
+    },
+    {
+      "epoch": 0.04197409411378915,
+      "grad_norm": 0.2536779046058655,
+      "learning_rate": 4.0600000000000004e-05,
+      "loss": 2.3652,
+      "step": 376
+    },
+    {
+      "epoch": 0.04208572734281518,
+      "grad_norm": 0.25629979372024536,
+      "learning_rate": 4.0575000000000004e-05,
+      "loss": 2.3127,
+      "step": 377
+    },
+    {
+      "epoch": 0.042197360571841216,
+      "grad_norm": 0.25751662254333496,
+      "learning_rate": 4.055e-05,
+      "loss": 2.3268,
+      "step": 378
+    },
+    {
+      "epoch": 0.04230899380086725,
+      "grad_norm": 0.25083836913108826,
+      "learning_rate": 4.0525e-05,
+      "loss": 2.3929,
+      "step": 379
+    },
+    {
+      "epoch": 0.042420627029893285,
+      "grad_norm": 0.2616216242313385,
+      "learning_rate": 4.05e-05,
+      "loss": 2.2566,
+      "step": 380
+    },
+    {
+      "epoch": 0.04253226025891932,
+      "grad_norm": 0.2545231580734253,
+      "learning_rate": 4.0475e-05,
+      "loss": 2.3916,
+      "step": 381
+    },
+    {
+      "epoch": 0.04264389348794535,
+      "grad_norm": 0.24984431266784668,
+      "learning_rate": 4.045000000000001e-05,
+      "loss": 2.3001,
+      "step": 382
+    },
+    {
+      "epoch": 0.04275552671697139,
+      "grad_norm": 0.2487059086561203,
+      "learning_rate": 4.0425e-05,
+      "loss": 2.4398,
+      "step": 383
+    },
+    {
+      "epoch": 0.04286715994599743,
+      "grad_norm": 0.2508711516857147,
+      "learning_rate": 4.0400000000000006e-05,
+      "loss": 2.3868,
+      "step": 384
+    },
+    {
+      "epoch": 0.04297879317502346,
+      "grad_norm": 0.250318706035614,
+      "learning_rate": 4.0375e-05,
+      "loss": 2.3627,
+      "step": 385
+    },
+    {
+      "epoch": 0.043090426404049496,
+      "grad_norm": 0.27126437425613403,
+      "learning_rate": 4.0350000000000005e-05,
+      "loss": 2.3592,
+      "step": 386
+    },
+    {
+      "epoch": 0.04320205963307553,
+      "grad_norm": 0.2641567289829254,
+      "learning_rate": 4.0325000000000004e-05,
+      "loss": 2.3391,
+      "step": 387
+    },
+    {
+      "epoch": 0.043313692862101565,
+      "grad_norm": 0.24686801433563232,
+      "learning_rate": 4.0300000000000004e-05,
+      "loss": 2.4642,
+      "step": 388
+    },
+    {
+      "epoch": 0.0434253260911276,
+      "grad_norm": 0.260929137468338,
+      "learning_rate": 4.0275e-05,
+      "loss": 2.2533,
+      "step": 389
+    },
+    {
+      "epoch": 0.04353695932015363,
+      "grad_norm": 0.2568647265434265,
+      "learning_rate": 4.025e-05,
+      "loss": 2.3352,
+      "step": 390
+    },
+    {
+      "epoch": 0.04364859254917967,
+      "grad_norm": 0.24345262348651886,
+      "learning_rate": 4.0225e-05,
+      "loss": 2.3717,
+      "step": 391
+    },
+    {
+      "epoch": 0.04376022577820571,
+      "grad_norm": 0.24727144837379456,
+      "learning_rate": 4.02e-05,
+      "loss": 2.3759,
+      "step": 392
+    },
+    {
+      "epoch": 0.04387185900723174,
+      "grad_norm": 0.24666303396224976,
+      "learning_rate": 4.0175e-05,
+      "loss": 2.3249,
+      "step": 393
+    },
+    {
+      "epoch": 0.043983492236257776,
+      "grad_norm": 0.26533442735671997,
+      "learning_rate": 4.015000000000001e-05,
+      "loss": 2.4041,
+      "step": 394
+    },
+    {
+      "epoch": 0.04409512546528381,
+      "grad_norm": 0.25406232476234436,
+      "learning_rate": 4.0125e-05,
+      "loss": 2.3457,
+      "step": 395
+    },
+    {
+      "epoch": 0.044206758694309845,
+      "grad_norm": 0.250957190990448,
+      "learning_rate": 4.0100000000000006e-05,
+      "loss": 2.3591,
+      "step": 396
+    },
+    {
+      "epoch": 0.04431839192333588,
+      "grad_norm": 0.25459203124046326,
+      "learning_rate": 4.0075e-05,
+      "loss": 2.4105,
+      "step": 397
+    },
+    {
+      "epoch": 0.04443002515236191,
+      "grad_norm": 0.25286054611206055,
+      "learning_rate": 4.0050000000000004e-05,
+      "loss": 2.3757,
+      "step": 398
+    },
+    {
+      "epoch": 0.04454165838138795,
+      "grad_norm": 0.2448861449956894,
+      "learning_rate": 4.0025000000000004e-05,
+      "loss": 2.4299,
+      "step": 399
+    },
+    {
+      "epoch": 0.04465329161041399,
+      "grad_norm": 0.24877607822418213,
+      "learning_rate": 4e-05,
+      "loss": 2.4102,
+      "step": 400
+    },
+    {
+      "epoch": 0.04476492483944002,
+      "grad_norm": 0.2525503635406494,
+      "learning_rate": 3.9975e-05,
+      "loss": 2.4452,
+      "step": 401
+    },
+    {
+      "epoch": 0.044876558068466056,
+      "grad_norm": 0.24728810787200928,
+      "learning_rate": 3.995e-05,
+      "loss": 2.354,
+      "step": 402
+    },
+    {
+      "epoch": 0.04498819129749209,
+      "grad_norm": 0.26556339859962463,
+      "learning_rate": 3.9925e-05,
+      "loss": 2.3333,
+      "step": 403
+    },
+    {
+      "epoch": 0.045099824526518124,
+      "grad_norm": 0.25951698422431946,
+      "learning_rate": 3.99e-05,
+      "loss": 2.2839,
+      "step": 404
+    },
+    {
+      "epoch": 0.04521145775554416,
+      "grad_norm": 0.25703710317611694,
+      "learning_rate": 3.9875e-05,
+      "loss": 2.3381,
+      "step": 405
+    },
+    {
+      "epoch": 0.04532309098457019,
+      "grad_norm": 0.2780836820602417,
+      "learning_rate": 3.9850000000000006e-05,
+      "loss": 2.3457,
+      "step": 406
+    },
+    {
+      "epoch": 0.04543472421359623,
+      "grad_norm": 0.252136766910553,
+      "learning_rate": 3.9825e-05,
+      "loss": 2.3855,
+      "step": 407
+    },
+    {
+      "epoch": 0.04554635744262227,
+      "grad_norm": 0.2834818661212921,
+      "learning_rate": 3.9800000000000005e-05,
+      "loss": 2.4514,
+      "step": 408
+    },
+    {
+      "epoch": 0.0456579906716483,
+      "grad_norm": 0.2503218948841095,
+      "learning_rate": 3.9775e-05,
+      "loss": 2.3824,
+      "step": 409
+    },
+    {
+      "epoch": 0.045769623900674336,
+      "grad_norm": 0.2580338418483734,
+      "learning_rate": 3.9750000000000004e-05,
+      "loss": 2.3869,
+      "step": 410
+    },
+    {
+      "epoch": 0.04588125712970037,
+      "grad_norm": 0.2537950277328491,
+      "learning_rate": 3.9725e-05,
+      "loss": 2.3022,
+      "step": 411
+    },
+    {
+      "epoch": 0.045992890358726404,
+      "grad_norm": 0.2560393810272217,
+      "learning_rate": 3.97e-05,
+      "loss": 2.3521,
+      "step": 412
+    },
+    {
+      "epoch": 0.04610452358775244,
+      "grad_norm": 0.2616626024246216,
+      "learning_rate": 3.9675e-05,
+      "loss": 2.3337,
+      "step": 413
+    },
+    {
+      "epoch": 0.04621615681677847,
+      "grad_norm": 0.3034763038158417,
+      "learning_rate": 3.965e-05,
+      "loss": 2.2924,
+      "step": 414
+    },
+    {
+      "epoch": 0.04632779004580451,
+      "grad_norm": 0.2462148219347,
+      "learning_rate": 3.9625e-05,
+      "loss": 2.4056,
+      "step": 415
+    },
+    {
+      "epoch": 0.04643942327483054,
+      "grad_norm": 0.2612819969654083,
+      "learning_rate": 3.960000000000001e-05,
+      "loss": 2.2965,
+      "step": 416
+    },
+    {
+      "epoch": 0.04655105650385658,
+      "grad_norm": 0.2580772638320923,
+      "learning_rate": 3.9575e-05,
+      "loss": 2.3585,
+      "step": 417
+    },
+    {
+      "epoch": 0.046662689732882616,
+      "grad_norm": 0.25281044840812683,
+      "learning_rate": 3.9550000000000006e-05,
+      "loss": 2.4224,
+      "step": 418
+    },
+    {
+      "epoch": 0.04677432296190865,
+      "grad_norm": 0.2507364749908447,
+      "learning_rate": 3.9525e-05,
+      "loss": 2.3973,
+      "step": 419
+    },
+    {
+      "epoch": 0.046885956190934684,
+      "grad_norm": 0.25551095604896545,
+      "learning_rate": 3.9500000000000005e-05,
+      "loss": 2.3345,
+      "step": 420
+    },
+    {
+      "epoch": 0.04699758941996072,
+      "grad_norm": 0.276262104511261,
+      "learning_rate": 3.9475000000000004e-05,
+      "loss": 2.3373,
+      "step": 421
+    },
+    {
+      "epoch": 0.04710922264898675,
+      "grad_norm": 0.30092892050743103,
+      "learning_rate": 3.9450000000000003e-05,
+      "loss": 2.3881,
+      "step": 422
+    },
+    {
+      "epoch": 0.04722085587801279,
+      "grad_norm": 0.25903210043907166,
+      "learning_rate": 3.9425e-05,
+      "loss": 2.3793,
+      "step": 423
+    },
+    {
+      "epoch": 0.04733248910703882,
+      "grad_norm": 0.2587510347366333,
+      "learning_rate": 3.94e-05,
+      "loss": 2.2823,
+      "step": 424
+    },
+    {
+      "epoch": 0.04744412233606486,
+      "grad_norm": 0.2502042055130005,
+      "learning_rate": 3.9375e-05,
+      "loss": 2.3279,
+      "step": 425
+    },
+    {
+      "epoch": 0.047555755565090896,
+      "grad_norm": 0.43736714124679565,
+      "learning_rate": 3.935e-05,
+      "loss": 2.3215,
+      "step": 426
+    },
+    {
+      "epoch": 0.04766738879411693,
+      "grad_norm": 0.2547898590564728,
+      "learning_rate": 3.9325e-05,
+      "loss": 2.2728,
+      "step": 427
+    },
+    {
+      "epoch": 0.047779022023142964,
+      "grad_norm": 0.24751374125480652,
+      "learning_rate": 3.9300000000000007e-05,
+      "loss": 2.3808,
+      "step": 428
+    },
+    {
+      "epoch": 0.047890655252169,
+      "grad_norm": 0.2624378800392151,
+      "learning_rate": 3.9275e-05,
+      "loss": 2.3256,
+      "step": 429
+    },
+    {
+      "epoch": 0.04800228848119503,
+      "grad_norm": 0.4183436632156372,
+      "learning_rate": 3.9250000000000005e-05,
+      "loss": 2.3553,
+      "step": 430
+    },
+    {
+      "epoch": 0.048113921710221066,
+      "grad_norm": 0.2535308599472046,
+      "learning_rate": 3.9225e-05,
+      "loss": 2.3025,
+      "step": 431
+    },
+    {
+      "epoch": 0.0482255549392471,
+      "grad_norm": 0.3438050448894501,
+      "learning_rate": 3.9200000000000004e-05,
+      "loss": 2.322,
+      "step": 432
+    },
+    {
+      "epoch": 0.04833718816827314,
+      "grad_norm": 0.2605397403240204,
+      "learning_rate": 3.9175000000000004e-05,
+      "loss": 2.3356,
+      "step": 433
+    },
+    {
+      "epoch": 0.048448821397299176,
+      "grad_norm": 0.24819597601890564,
+      "learning_rate": 3.915e-05,
+      "loss": 2.4338,
+      "step": 434
+    },
+    {
+      "epoch": 0.04856045462632521,
+      "grad_norm": 0.24737314879894257,
+      "learning_rate": 3.9125e-05,
+      "loss": 2.3618,
+      "step": 435
+    },
+    {
+      "epoch": 0.048672087855351244,
+      "grad_norm": 0.25205180048942566,
+      "learning_rate": 3.91e-05,
+      "loss": 2.3555,
+      "step": 436
+    },
+    {
+      "epoch": 0.04878372108437728,
+      "grad_norm": 0.24479413032531738,
+      "learning_rate": 3.9075e-05,
+      "loss": 2.3788,
+      "step": 437
+    },
+    {
+      "epoch": 0.04889535431340331,
+      "grad_norm": 0.26318883895874023,
+      "learning_rate": 3.905e-05,
+      "loss": 2.4028,
+      "step": 438
+    },
+    {
+      "epoch": 0.049006987542429346,
+      "grad_norm": 0.24493621289730072,
+      "learning_rate": 3.9025e-05,
+      "loss": 2.3734,
+      "step": 439
+    },
+    {
+      "epoch": 0.04911862077145538,
+      "grad_norm": 0.2888612151145935,
+      "learning_rate": 3.9000000000000006e-05,
+      "loss": 2.3424,
+      "step": 440
+    },
+    {
+      "epoch": 0.04923025400048142,
+      "grad_norm": 0.2574605941772461,
+      "learning_rate": 3.8975e-05,
+      "loss": 2.4289,
+      "step": 441
+    },
+    {
+      "epoch": 0.049341887229507456,
+      "grad_norm": 0.329041987657547,
+      "learning_rate": 3.8950000000000005e-05,
+      "loss": 2.2217,
+      "step": 442
+    },
+    {
+      "epoch": 0.04945352045853349,
+      "grad_norm": 0.29988110065460205,
+      "learning_rate": 3.8925e-05,
+      "loss": 2.4389,
+      "step": 443
+    },
+    {
+      "epoch": 0.049565153687559524,
+      "grad_norm": 0.2377348095178604,
+      "learning_rate": 3.8900000000000004e-05,
+      "loss": 2.4175,
+      "step": 444
+    },
+    {
+      "epoch": 0.04967678691658556,
+      "grad_norm": 0.28037315607070923,
+      "learning_rate": 3.8875e-05,
+      "loss": 2.241,
+      "step": 445
+    },
+    {
+      "epoch": 0.04978842014561159,
+      "grad_norm": 0.2584727704524994,
+      "learning_rate": 3.885e-05,
+      "loss": 2.3034,
+      "step": 446
+    },
+    {
+      "epoch": 0.049900053374637626,
+      "grad_norm": 0.24890342354774475,
+      "learning_rate": 3.8825e-05,
+      "loss": 2.3393,
+      "step": 447
+    },
+    {
+      "epoch": 0.05001168660366366,
+      "grad_norm": 0.24033311009407043,
+      "learning_rate": 3.88e-05,
+      "loss": 2.339,
+      "step": 448
+    },
+    {
+      "epoch": 0.0501233198326897,
+      "grad_norm": 0.2352409064769745,
+      "learning_rate": 3.8775e-05,
+      "loss": 2.337,
+      "step": 449
+    },
+    {
+      "epoch": 0.050234953061715736,
+      "grad_norm": 0.8847767114639282,
+      "learning_rate": 3.875e-05,
+      "loss": 2.2669,
+      "step": 450
+    },
+    {
+      "epoch": 0.05034658629074177,
+      "grad_norm": 0.2545109987258911,
+      "learning_rate": 3.8725e-05,
+      "loss": 2.3277,
+      "step": 451
+    },
+    {
+      "epoch": 0.050458219519767804,
+      "grad_norm": 0.2856300175189972,
+      "learning_rate": 3.8700000000000006e-05,
+      "loss": 2.3928,
+      "step": 452
+    },
+    {
+      "epoch": 0.05056985274879384,
+      "grad_norm": 0.24699796736240387,
+      "learning_rate": 3.8675e-05,
+      "loss": 2.4304,
+      "step": 453
+    },
+    {
+      "epoch": 0.05068148597781987,
+      "grad_norm": 0.2530304193496704,
+      "learning_rate": 3.8650000000000004e-05,
+      "loss": 2.3304,
+      "step": 454
+    },
+    {
+      "epoch": 0.050793119206845906,
+      "grad_norm": 0.25681325793266296,
+      "learning_rate": 3.8625e-05,
+      "loss": 2.4092,
+      "step": 455
+    },
+    {
+      "epoch": 0.05090475243587194,
+      "grad_norm": 0.2615368962287903,
+      "learning_rate": 3.86e-05,
+      "loss": 2.3721,
+      "step": 456
+    },
+    {
+      "epoch": 0.051016385664897974,
+      "grad_norm": 0.2456715703010559,
+      "learning_rate": 3.8575e-05,
+      "loss": 2.3512,
+      "step": 457
+    },
+    {
+      "epoch": 0.051128018893924015,
+      "grad_norm": 0.26264017820358276,
+      "learning_rate": 3.855e-05,
+      "loss": 2.419,
+      "step": 458
+    },
+    {
+      "epoch": 0.05123965212295005,
+      "grad_norm": 0.2601991295814514,
+      "learning_rate": 3.8525e-05,
+      "loss": 2.3389,
+      "step": 459
+    },
+    {
+      "epoch": 0.051351285351976084,
+      "grad_norm": 0.24692246317863464,
+      "learning_rate": 3.85e-05,
+      "loss": 2.1237,
+      "step": 460
+    },
+    {
+      "epoch": 0.05146291858100212,
+      "grad_norm": 0.25850188732147217,
+      "learning_rate": 3.8475e-05,
+      "loss": 2.3413,
+      "step": 461
+    },
+    {
+      "epoch": 0.05157455181002815,
+      "grad_norm": 0.2406020611524582,
+      "learning_rate": 3.845e-05,
+      "loss": 2.36,
+      "step": 462
+    },
+    {
+      "epoch": 0.051686185039054186,
+      "grad_norm": 0.25938504934310913,
+      "learning_rate": 3.8425e-05,
+      "loss": 2.3512,
+      "step": 463
+    },
+    {
+      "epoch": 0.05179781826808022,
+      "grad_norm": 0.2700086236000061,
+      "learning_rate": 3.8400000000000005e-05,
+      "loss": 2.3883,
+      "step": 464
+    },
+    {
+      "epoch": 0.051909451497106254,
+      "grad_norm": 0.24675601720809937,
+      "learning_rate": 3.8375e-05,
+      "loss": 2.3614,
+      "step": 465
+    },
+    {
+      "epoch": 0.052021084726132295,
+      "grad_norm": 0.3309410512447357,
+      "learning_rate": 3.8350000000000004e-05,
+      "loss": 2.4193,
+      "step": 466
+    },
+    {
+      "epoch": 0.05213271795515833,
+      "grad_norm": 0.24233734607696533,
+      "learning_rate": 3.8324999999999996e-05,
+      "loss": 2.3038,
+      "step": 467
+    },
+    {
+      "epoch": 0.052244351184184364,
+      "grad_norm": 0.24930906295776367,
+      "learning_rate": 3.83e-05,
+      "loss": 2.3087,
+      "step": 468
+    },
+    {
+      "epoch": 0.0523559844132104,
+      "grad_norm": 0.2506902813911438,
+      "learning_rate": 3.8275e-05,
+      "loss": 2.4125,
+      "step": 469
+    },
+    {
+      "epoch": 0.05246761764223643,
+      "grad_norm": 0.4915805757045746,
+      "learning_rate": 3.825e-05,
+      "loss": 2.4222,
+      "step": 470
+    },
+    {
+      "epoch": 0.052579250871262466,
+      "grad_norm": 0.24434614181518555,
+      "learning_rate": 3.8225e-05,
+      "loss": 2.4653,
+      "step": 471
+    },
+    {
+      "epoch": 0.0526908841002885,
+      "grad_norm": 0.24511779844760895,
+      "learning_rate": 3.82e-05,
+      "loss": 2.3369,
+      "step": 472
+    },
+    {
+      "epoch": 0.052802517329314534,
+      "grad_norm": 0.24757201969623566,
+      "learning_rate": 3.8175e-05,
+      "loss": 2.2323,
+      "step": 473
+    },
+    {
+      "epoch": 0.052914150558340575,
+      "grad_norm": 0.2527662515640259,
+      "learning_rate": 3.8150000000000006e-05,
+      "loss": 2.4045,
+      "step": 474
+    },
+    {
+      "epoch": 0.05302578378736661,
+      "grad_norm": 0.2481798529624939,
+      "learning_rate": 3.8125e-05,
+      "loss": 2.3615,
+      "step": 475
+    },
+    {
+      "epoch": 0.053137417016392643,
+      "grad_norm": 0.25394219160079956,
+      "learning_rate": 3.8100000000000005e-05,
+      "loss": 2.3536,
+      "step": 476
+    },
+    {
+      "epoch": 0.05324905024541868,
+      "grad_norm": 0.2457951009273529,
+      "learning_rate": 3.8075e-05,
+      "loss": 2.2919,
+      "step": 477
+    },
+    {
+      "epoch": 0.05336068347444471,
+      "grad_norm": 0.43115633726119995,
+      "learning_rate": 3.805e-05,
+      "loss": 2.4222,
+      "step": 478
+    },
+    {
+      "epoch": 0.053472316703470746,
+      "grad_norm": 0.257254958152771,
+      "learning_rate": 3.8025e-05,
+      "loss": 2.3744,
+      "step": 479
+    },
+    {
+      "epoch": 0.05358394993249678,
+      "grad_norm": 0.2672825753688812,
+      "learning_rate": 3.8e-05,
+      "loss": 2.3315,
+      "step": 480
+    },
+    {
+      "epoch": 0.053695583161522814,
+      "grad_norm": 0.24192893505096436,
+      "learning_rate": 3.7975e-05,
+      "loss": 2.4601,
+      "step": 481
+    },
+    {
+      "epoch": 0.053807216390548855,
+      "grad_norm": 0.24756518006324768,
+      "learning_rate": 3.795e-05,
+      "loss": 2.2872,
+      "step": 482
+    },
+    {
+      "epoch": 0.05391884961957489,
+      "grad_norm": 0.25323203206062317,
+      "learning_rate": 3.7925e-05,
+      "loss": 2.4142,
+      "step": 483
+    },
+    {
+      "epoch": 0.05403048284860092,
+      "grad_norm": 0.25254499912261963,
+      "learning_rate": 3.79e-05,
+      "loss": 2.3528,
+      "step": 484
+    },
+    {
+      "epoch": 0.05414211607762696,
+      "grad_norm": 0.25069230794906616,
+      "learning_rate": 3.7875e-05,
+      "loss": 2.2906,
+      "step": 485
+    },
+    {
+      "epoch": 0.05425374930665299,
+      "grad_norm": 0.26814520359039307,
+      "learning_rate": 3.7850000000000005e-05,
+      "loss": 2.3368,
+      "step": 486
+    },
+    {
+      "epoch": 0.054365382535679026,
+      "grad_norm": 0.24452055990695953,
+      "learning_rate": 3.7825e-05,
+      "loss": 2.3667,
+      "step": 487
+    },
+    {
+      "epoch": 0.05447701576470506,
+      "grad_norm": 0.26742151379585266,
+      "learning_rate": 3.7800000000000004e-05,
+      "loss": 2.323,
+      "step": 488
+    },
+    {
+      "epoch": 0.054588648993731094,
+      "grad_norm": 0.25766825675964355,
+      "learning_rate": 3.7775e-05,
+      "loss": 2.3761,
+      "step": 489
+    },
+    {
+      "epoch": 0.05470028222275713,
+      "grad_norm": 0.29029610753059387,
+      "learning_rate": 3.775e-05,
+      "loss": 2.3216,
+      "step": 490
+    },
+    {
+      "epoch": 0.05481191545178317,
+      "grad_norm": 0.25023818016052246,
+      "learning_rate": 3.7725e-05,
+      "loss": 2.3441,
+      "step": 491
+    },
+    {
+      "epoch": 0.0549235486808092,
+      "grad_norm": 0.3373621702194214,
+      "learning_rate": 3.77e-05,
+      "loss": 2.3166,
+      "step": 492
+    },
+    {
+      "epoch": 0.05503518190983524,
+      "grad_norm": 0.23989304900169373,
+      "learning_rate": 3.7675e-05,
+      "loss": 2.2795,
+      "step": 493
+    },
+    {
+      "epoch": 0.05514681513886127,
+      "grad_norm": 0.3192415237426758,
+      "learning_rate": 3.765e-05,
+      "loss": 2.3273,
+      "step": 494
+    },
+    {
+      "epoch": 0.055258448367887306,
+      "grad_norm": 0.24007610976696014,
+      "learning_rate": 3.7625e-05,
+      "loss": 2.413,
+      "step": 495
+    },
+    {
+      "epoch": 0.05537008159691334,
+      "grad_norm": 0.36170458793640137,
+      "learning_rate": 3.76e-05,
+      "loss": 2.2216,
+      "step": 496
+    },
+    {
+      "epoch": 0.055481714825939374,
+      "grad_norm": 0.24714718759059906,
+      "learning_rate": 3.7575e-05,
+      "loss": 2.3394,
+      "step": 497
+    },
+    {
+      "epoch": 0.05559334805496541,
+      "grad_norm": 0.238433837890625,
+      "learning_rate": 3.7550000000000005e-05,
+      "loss": 2.3685,
+      "step": 498
+    },
+    {
+      "epoch": 0.05570498128399145,
+      "grad_norm": 0.24975870549678802,
+      "learning_rate": 3.7525e-05,
+      "loss": 2.241,
+      "step": 499
+    },
+    {
+      "epoch": 0.05581661451301748,
+      "grad_norm": 0.24853730201721191,
+      "learning_rate": 3.7500000000000003e-05,
+      "loss": 2.4173,
+      "step": 500
+    },
+    {
+      "epoch": 0.05592824774204352,
+      "grad_norm": 0.24904385209083557,
+      "learning_rate": 3.7475e-05,
+      "loss": 2.4291,
+      "step": 501
+    },
+    {
+      "epoch": 0.05603988097106955,
+      "grad_norm": 0.23549965023994446,
+      "learning_rate": 3.745e-05,
+      "loss": 2.3054,
+      "step": 502
+    },
+    {
+      "epoch": 0.056151514200095586,
+      "grad_norm": 0.2464476376771927,
+      "learning_rate": 3.7425e-05,
+      "loss": 2.4886,
+      "step": 503
+    },
+    {
+      "epoch": 0.05626314742912162,
+      "grad_norm": 0.25297120213508606,
+      "learning_rate": 3.74e-05,
+      "loss": 2.3384,
+      "step": 504
+    },
+    {
+      "epoch": 0.056374780658147654,
+      "grad_norm": 0.2817951440811157,
+      "learning_rate": 3.737500000000001e-05,
+      "loss": 2.3319,
+      "step": 505
+    },
+    {
+      "epoch": 0.05648641388717369,
+      "grad_norm": 0.24703608453273773,
+      "learning_rate": 3.735e-05,
+      "loss": 2.4513,
+      "step": 506
+    },
+    {
+      "epoch": 0.05659804711619973,
+      "grad_norm": 0.25609731674194336,
+      "learning_rate": 3.7325000000000006e-05,
+      "loss": 2.4268,
+      "step": 507
+    },
+    {
+      "epoch": 0.05670968034522576,
+      "grad_norm": 0.2358425110578537,
+      "learning_rate": 3.73e-05,
+      "loss": 2.3388,
+      "step": 508
+    },
+    {
+      "epoch": 0.0568213135742518,
+      "grad_norm": 0.2538928687572479,
+      "learning_rate": 3.7275000000000005e-05,
+      "loss": 2.4332,
+      "step": 509
+    },
+    {
+      "epoch": 0.05693294680327783,
+      "grad_norm": 0.2499266117811203,
+      "learning_rate": 3.7250000000000004e-05,
+      "loss": 2.3393,
+      "step": 510
+    },
+    {
+      "epoch": 0.057044580032303865,
+      "grad_norm": 0.24814729392528534,
+      "learning_rate": 3.7225000000000004e-05,
+      "loss": 2.3558,
+      "step": 511
+    },
+    {
+      "epoch": 0.0571562132613299,
+      "grad_norm": 0.24115900695323944,
+      "learning_rate": 3.72e-05,
+      "loss": 2.33,
+      "step": 512
+    },
+    {
+      "epoch": 0.057267846490355934,
+      "grad_norm": 0.24013325572013855,
+      "learning_rate": 3.7175e-05,
+      "loss": 2.4107,
+      "step": 513
+    },
+    {
+      "epoch": 0.05737947971938197,
+      "grad_norm": 0.24541234970092773,
+      "learning_rate": 3.715e-05,
+      "loss": 2.3866,
+      "step": 514
+    },
+    {
+      "epoch": 0.05749111294840801,
+      "grad_norm": 0.23506753146648407,
+      "learning_rate": 3.7125e-05,
+      "loss": 2.3321,
+      "step": 515
+    },
+    {
+      "epoch": 0.05760274617743404,
+      "grad_norm": 0.24425718188285828,
+      "learning_rate": 3.71e-05,
+      "loss": 2.4,
+      "step": 516
+    },
+    {
+      "epoch": 0.05771437940646008,
+      "grad_norm": 0.2441939115524292,
+      "learning_rate": 3.707500000000001e-05,
+      "loss": 2.3855,
+      "step": 517
+    },
+    {
+      "epoch": 0.05782601263548611,
+      "grad_norm": 0.23702003061771393,
+      "learning_rate": 3.705e-05,
+      "loss": 2.3171,
+      "step": 518
+    },
+    {
+      "epoch": 0.057937645864512145,
+      "grad_norm": 0.2465049922466278,
+      "learning_rate": 3.7025000000000005e-05,
+      "loss": 2.431,
+      "step": 519
+    },
+    {
+      "epoch": 0.05804927909353818,
+      "grad_norm": 0.22516661882400513,
+      "learning_rate": 3.7e-05,
+      "loss": 2.3307,
+      "step": 520
+    },
+    {
+      "epoch": 0.058160912322564214,
+      "grad_norm": 0.2372903823852539,
+      "learning_rate": 3.6975000000000004e-05,
+      "loss": 2.2791,
+      "step": 521
+    },
+    {
+      "epoch": 0.05827254555159025,
+      "grad_norm": 0.2584116458892822,
+      "learning_rate": 3.6950000000000004e-05,
+      "loss": 2.4296,
+      "step": 522
+    },
+    {
+      "epoch": 0.05838417878061628,
+      "grad_norm": 0.24023661017417908,
+      "learning_rate": 3.6925e-05,
+      "loss": 2.4579,
+      "step": 523
+    },
+    {
+      "epoch": 0.05849581200964232,
+      "grad_norm": 0.26791271567344666,
+      "learning_rate": 3.69e-05,
+      "loss": 2.3662,
+      "step": 524
+    },
+    {
+      "epoch": 0.05860744523866836,
+      "grad_norm": 0.23485226929187775,
+      "learning_rate": 3.6875e-05,
+      "loss": 2.1954,
+      "step": 525
+    },
+    {
+      "epoch": 0.05871907846769439,
+      "grad_norm": 0.24312959611415863,
+      "learning_rate": 3.685e-05,
+      "loss": 2.4423,
+      "step": 526
+    },
+    {
+      "epoch": 0.058830711696720425,
+      "grad_norm": 0.2558239996433258,
+      "learning_rate": 3.6825e-05,
+      "loss": 2.3661,
+      "step": 527
+    },
+    {
+      "epoch": 0.05894234492574646,
+      "grad_norm": 0.26516368985176086,
+      "learning_rate": 3.68e-05,
+      "loss": 2.3181,
+      "step": 528
+    },
+    {
+      "epoch": 0.05905397815477249,
+      "grad_norm": 0.24487736821174622,
+      "learning_rate": 3.6775000000000006e-05,
+      "loss": 2.3667,
+      "step": 529
+    },
+    {
+      "epoch": 0.05916561138379853,
+      "grad_norm": 0.23648685216903687,
+      "learning_rate": 3.675e-05,
+      "loss": 2.3438,
+      "step": 530
+    },
+    {
+      "epoch": 0.05927724461282456,
+      "grad_norm": 0.2509547173976898,
+      "learning_rate": 3.6725000000000005e-05,
+      "loss": 2.3128,
+      "step": 531
+    },
+    {
+      "epoch": 0.0593888778418506,
+      "grad_norm": 0.24765925109386444,
+      "learning_rate": 3.6700000000000004e-05,
+      "loss": 2.3556,
+      "step": 532
+    },
+    {
+      "epoch": 0.05950051107087664,
+      "grad_norm": 0.23786590993404388,
+      "learning_rate": 3.6675000000000004e-05,
+      "loss": 2.4546,
+      "step": 533
+    },
+    {
+      "epoch": 0.05961214429990267,
+      "grad_norm": 0.3275027573108673,
+      "learning_rate": 3.665e-05,
+      "loss": 2.3161,
+      "step": 534
+    },
+    {
+      "epoch": 0.059723777528928705,
+      "grad_norm": 0.24953784048557281,
+      "learning_rate": 3.6625e-05,
+      "loss": 2.3206,
+      "step": 535
+    },
+    {
+      "epoch": 0.05983541075795474,
+      "grad_norm": 0.24241773784160614,
+      "learning_rate": 3.66e-05,
+      "loss": 2.2838,
+      "step": 536
+    },
+    {
+      "epoch": 0.05994704398698077,
+      "grad_norm": 0.23624862730503082,
+      "learning_rate": 3.6575e-05,
+      "loss": 2.3082,
+      "step": 537
+    },
+    {
+      "epoch": 0.06005867721600681,
+      "grad_norm": 0.24654389917850494,
+      "learning_rate": 3.655e-05,
+      "loss": 2.3327,
+      "step": 538
+    },
+    {
+      "epoch": 0.06017031044503284,
+      "grad_norm": 0.24037936329841614,
+      "learning_rate": 3.652500000000001e-05,
+      "loss": 2.336,
+      "step": 539
+    },
+    {
+      "epoch": 0.06028194367405888,
+      "grad_norm": 0.25897306203842163,
+      "learning_rate": 3.65e-05,
+      "loss": 2.3064,
+      "step": 540
+    },
+    {
+      "epoch": 0.06039357690308492,
+      "grad_norm": 0.24670930206775665,
+      "learning_rate": 3.6475000000000006e-05,
+      "loss": 2.2893,
+      "step": 541
+    },
+    {
+      "epoch": 0.06050521013211095,
+      "grad_norm": 0.23652324080467224,
+      "learning_rate": 3.645e-05,
+      "loss": 2.3516,
+      "step": 542
+    },
+    {
+      "epoch": 0.060616843361136985,
+      "grad_norm": 0.23609410226345062,
+      "learning_rate": 3.6425000000000004e-05,
+      "loss": 2.3463,
+      "step": 543
+    },
+    {
+      "epoch": 0.06072847659016302,
+      "grad_norm": 0.24355870485305786,
+      "learning_rate": 3.6400000000000004e-05,
+      "loss": 2.4344,
+      "step": 544
+    },
+    {
+      "epoch": 0.06084010981918905,
+      "grad_norm": 0.24142801761627197,
+      "learning_rate": 3.6375e-05,
+      "loss": 2.3728,
+      "step": 545
+    },
+    {
+      "epoch": 0.06095174304821509,
+      "grad_norm": 0.24296842515468597,
+      "learning_rate": 3.635e-05,
+      "loss": 2.3984,
+      "step": 546
+    },
+    {
+      "epoch": 0.06106337627724112,
+      "grad_norm": 0.24469731748104095,
+      "learning_rate": 3.6325e-05,
+      "loss": 2.3478,
+      "step": 547
+    },
+    {
+      "epoch": 0.06117500950626716,
+      "grad_norm": 0.25056391954421997,
+      "learning_rate": 3.63e-05,
+      "loss": 2.2224,
+      "step": 548
+    },
+    {
+      "epoch": 0.0612866427352932,
+      "grad_norm": 0.23732852935791016,
+      "learning_rate": 3.6275e-05,
+      "loss": 2.4875,
+      "step": 549
+    },
+    {
+      "epoch": 0.06139827596431923,
+      "grad_norm": 0.2487863302230835,
+      "learning_rate": 3.625e-05,
+      "loss": 2.27,
+      "step": 550
+    },
+    {
+      "epoch": 0.061509909193345265,
+      "grad_norm": 0.3302936851978302,
+      "learning_rate": 3.6225000000000006e-05,
+      "loss": 2.2132,
+      "step": 551
+    },
+    {
+      "epoch": 0.0616215424223713,
+      "grad_norm": 0.24820421636104584,
+      "learning_rate": 3.62e-05,
+      "loss": 2.3219,
+      "step": 552
+    },
+    {
+      "epoch": 0.06173317565139733,
+      "grad_norm": 0.24405966699123383,
+      "learning_rate": 3.6175000000000005e-05,
+      "loss": 2.3802,
+      "step": 553
+    },
+    {
+      "epoch": 0.06184480888042337,
+      "grad_norm": 0.24143864214420319,
+      "learning_rate": 3.615e-05,
+      "loss": 2.3897,
+      "step": 554
+    },
+    {
+      "epoch": 0.0619564421094494,
+      "grad_norm": 0.25099068880081177,
+      "learning_rate": 3.6125000000000004e-05,
+      "loss": 2.3671,
+      "step": 555
+    },
+    {
+      "epoch": 0.062068075338475436,
+      "grad_norm": 0.2503535747528076,
+      "learning_rate": 3.61e-05,
+      "loss": 2.4688,
+      "step": 556
+    },
+    {
+      "epoch": 0.06217970856750148,
+      "grad_norm": 0.2651500999927521,
+      "learning_rate": 3.6075e-05,
+      "loss": 2.2332,
+      "step": 557
+    },
+    {
+      "epoch": 0.06229134179652751,
+      "grad_norm": 0.2445705533027649,
+      "learning_rate": 3.605e-05,
+      "loss": 2.4157,
+      "step": 558
+    },
+    {
+      "epoch": 0.062402975025553545,
+      "grad_norm": 0.3407208025455475,
+      "learning_rate": 3.6025e-05,
+      "loss": 2.2068,
+      "step": 559
+    },
+    {
+      "epoch": 0.06251460825457958,
+      "grad_norm": 0.23573793470859528,
+      "learning_rate": 3.6e-05,
+      "loss": 2.3832,
+      "step": 560
+    },
+    {
+      "epoch": 0.06262624148360561,
+      "grad_norm": 0.23974989354610443,
+      "learning_rate": 3.5975e-05,
+      "loss": 2.4782,
+      "step": 561
+    },
+    {
+      "epoch": 0.06273787471263165,
+      "grad_norm": 0.24216674268245697,
+      "learning_rate": 3.595e-05,
+      "loss": 2.3656,
+      "step": 562
+    },
+    {
+      "epoch": 0.06284950794165768,
+      "grad_norm": 0.23898907005786896,
+      "learning_rate": 3.5925000000000006e-05,
+      "loss": 2.4015,
+      "step": 563
+    },
+    {
+      "epoch": 0.06296114117068372,
+      "grad_norm": 0.229814350605011,
+      "learning_rate": 3.59e-05,
+      "loss": 2.2476,
+      "step": 564
+    },
+    {
+      "epoch": 0.06307277439970975,
+      "grad_norm": 0.37582337856292725,
+      "learning_rate": 3.5875000000000005e-05,
+      "loss": 2.2959,
+      "step": 565
+    },
+    {
+      "epoch": 0.06318440762873578,
+      "grad_norm": 0.23359467089176178,
+      "learning_rate": 3.585e-05,
+      "loss": 2.3982,
+      "step": 566
+    },
+    {
+      "epoch": 0.06329604085776182,
+      "grad_norm": 0.23306161165237427,
+      "learning_rate": 3.5825000000000003e-05,
+      "loss": 2.3396,
+      "step": 567
+    },
+    {
+      "epoch": 0.06340767408678785,
+      "grad_norm": 0.2388441115617752,
+      "learning_rate": 3.58e-05,
+      "loss": 2.3826,
+      "step": 568
+    },
+    {
+      "epoch": 0.0635193073158139,
+      "grad_norm": 0.24095419049263,
+      "learning_rate": 3.5775e-05,
+      "loss": 2.3553,
+      "step": 569
+    },
+    {
+      "epoch": 0.06363094054483993,
+      "grad_norm": 0.2408561408519745,
+      "learning_rate": 3.575e-05,
+      "loss": 2.417,
+      "step": 570
+    },
+    {
+      "epoch": 0.06374257377386597,
+      "grad_norm": 0.23157310485839844,
+      "learning_rate": 3.5725e-05,
+      "loss": 2.3161,
+      "step": 571
+    },
+    {
+      "epoch": 0.063854207002892,
+      "grad_norm": 0.23947376012802124,
+      "learning_rate": 3.57e-05,
+      "loss": 2.3105,
+      "step": 572
+    },
+    {
+      "epoch": 0.06396584023191804,
+      "grad_norm": 0.24618233740329742,
+      "learning_rate": 3.5675e-05,
+      "loss": 2.4877,
+      "step": 573
+    },
+    {
+      "epoch": 0.06407747346094407,
+      "grad_norm": 0.24552257359027863,
+      "learning_rate": 3.565e-05,
+      "loss": 2.3758,
+      "step": 574
+    },
+    {
+      "epoch": 0.0641891066899701,
+      "grad_norm": 0.24329356849193573,
+      "learning_rate": 3.5625000000000005e-05,
+      "loss": 2.4084,
+      "step": 575
+    },
+    {
+      "epoch": 0.06430073991899614,
+      "grad_norm": 0.24198521673679352,
+      "learning_rate": 3.56e-05,
+      "loss": 2.4133,
+      "step": 576
+    },
+    {
+      "epoch": 0.06441237314802217,
+      "grad_norm": 0.23678706586360931,
+      "learning_rate": 3.5575000000000004e-05,
+      "loss": 2.378,
+      "step": 577
+    },
+    {
+      "epoch": 0.06452400637704821,
+      "grad_norm": 0.248667374253273,
+      "learning_rate": 3.555e-05,
+      "loss": 2.384,
+      "step": 578
+    },
+    {
+      "epoch": 0.06463563960607424,
+      "grad_norm": 0.24898695945739746,
+      "learning_rate": 3.5525e-05,
+      "loss": 2.3269,
+      "step": 579
+    },
+    {
+      "epoch": 0.06474727283510028,
+      "grad_norm": 0.240862175822258,
+      "learning_rate": 3.55e-05,
+      "loss": 2.381,
+      "step": 580
+    },
+    {
+      "epoch": 0.06485890606412631,
+      "grad_norm": 0.23366478085517883,
+      "learning_rate": 3.5475e-05,
+      "loss": 2.3628,
+      "step": 581
+    },
+    {
+      "epoch": 0.06497053929315234,
+      "grad_norm": 0.24220991134643555,
+      "learning_rate": 3.545e-05,
+      "loss": 2.2952,
+      "step": 582
+    },
+    {
+      "epoch": 0.06508217252217838,
+      "grad_norm": 0.24667523801326752,
+      "learning_rate": 3.5425e-05,
+      "loss": 2.4119,
+      "step": 583
+    },
+    {
+      "epoch": 0.06519380575120441,
+      "grad_norm": 0.24743525683879852,
+      "learning_rate": 3.54e-05,
+      "loss": 2.3073,
+      "step": 584
+    },
+    {
+      "epoch": 0.06530543898023046,
+      "grad_norm": 0.908301591873169,
+      "learning_rate": 3.5375e-05,
+      "loss": 2.3342,
+      "step": 585
+    },
+    {
+      "epoch": 0.0654170722092565,
+      "grad_norm": 0.2384248673915863,
+      "learning_rate": 3.535e-05,
+      "loss": 2.387,
+      "step": 586
+    },
+    {
+      "epoch": 0.06552870543828253,
+      "grad_norm": 0.29484090209007263,
+      "learning_rate": 3.5325000000000005e-05,
+      "loss": 2.4215,
+      "step": 587
+    },
+    {
+      "epoch": 0.06564033866730856,
+      "grad_norm": 0.2425820678472519,
+      "learning_rate": 3.53e-05,
+      "loss": 2.4056,
+      "step": 588
+    },
+    {
+      "epoch": 0.0657519718963346,
+      "grad_norm": 0.256686270236969,
+      "learning_rate": 3.5275000000000004e-05,
+      "loss": 2.3973,
+      "step": 589
+    },
+    {
+      "epoch": 0.06586360512536063,
+      "grad_norm": 0.23785726726055145,
+      "learning_rate": 3.525e-05,
+      "loss": 2.378,
+      "step": 590
+    },
+    {
+      "epoch": 0.06597523835438666,
+      "grad_norm": 0.2742824852466583,
+      "learning_rate": 3.5225e-05,
+      "loss": 2.3745,
+      "step": 591
+    },
+    {
+      "epoch": 0.0660868715834127,
+      "grad_norm": 0.2250916063785553,
+      "learning_rate": 3.52e-05,
+      "loss": 2.321,
+      "step": 592
+    },
+    {
+      "epoch": 0.06619850481243873,
+      "grad_norm": 0.24400900304317474,
+      "learning_rate": 3.5175e-05,
+      "loss": 2.4296,
+      "step": 593
+    },
+    {
+      "epoch": 0.06631013804146477,
+      "grad_norm": 0.24581992626190186,
+      "learning_rate": 3.515e-05,
+      "loss": 2.3375,
+      "step": 594
+    },
+    {
+      "epoch": 0.0664217712704908,
+      "grad_norm": 0.2356722503900528,
+      "learning_rate": 3.5125e-05,
+      "loss": 2.4229,
+      "step": 595
+    },
+    {
+      "epoch": 0.06653340449951683,
+      "grad_norm": 0.24524928629398346,
+      "learning_rate": 3.51e-05,
+      "loss": 2.3744,
+      "step": 596
+    },
+    {
+      "epoch": 0.06664503772854287,
+      "grad_norm": 0.24158966541290283,
+      "learning_rate": 3.5075000000000006e-05,
+      "loss": 2.3752,
+      "step": 597
+    },
+    {
+      "epoch": 0.0667566709575689,
+      "grad_norm": 0.2346276342868805,
+      "learning_rate": 3.505e-05,
+      "loss": 2.346,
+      "step": 598
+    },
+    {
+      "epoch": 0.06686830418659494,
+      "grad_norm": 0.24407444894313812,
+      "learning_rate": 3.5025000000000004e-05,
+      "loss": 2.3559,
+      "step": 599
+    },
+    {
+      "epoch": 0.06697993741562097,
+      "grad_norm": 0.24137824773788452,
+      "learning_rate": 3.5e-05,
+      "loss": 2.3965,
+      "step": 600
+    },
+    {
+      "epoch": 0.067091570644647,
+      "grad_norm": 0.24912434816360474,
+      "learning_rate": 3.4975e-05,
+      "loss": 2.3882,
+      "step": 601
+    },
+    {
+      "epoch": 0.06720320387367305,
+      "grad_norm": 0.24506230652332306,
+      "learning_rate": 3.495e-05,
+      "loss": 2.2875,
+      "step": 602
+    },
+    {
+      "epoch": 0.06731483710269909,
+      "grad_norm": 0.2423059046268463,
+      "learning_rate": 3.4925e-05,
+      "loss": 2.2828,
+      "step": 603
+    },
+    {
+      "epoch": 0.06742647033172512,
+      "grad_norm": 0.2598559856414795,
+      "learning_rate": 3.49e-05,
+      "loss": 2.288,
+      "step": 604
+    },
+    {
+      "epoch": 0.06753810356075116,
+      "grad_norm": 0.2390177696943283,
+      "learning_rate": 3.4875e-05,
+      "loss": 2.4369,
+      "step": 605
+    },
+    {
+      "epoch": 0.06764973678977719,
+      "grad_norm": 0.24242550134658813,
+      "learning_rate": 3.485e-05,
+      "loss": 2.2285,
+      "step": 606
+    },
+    {
+      "epoch": 0.06776137001880322,
+      "grad_norm": 0.23738925158977509,
+      "learning_rate": 3.4825e-05,
+      "loss": 2.4707,
+      "step": 607
+    },
+    {
+      "epoch": 0.06787300324782926,
+      "grad_norm": 0.23820523917675018,
+      "learning_rate": 3.48e-05,
+      "loss": 2.4274,
+      "step": 608
+    },
+    {
+      "epoch": 0.06798463647685529,
+      "grad_norm": 0.2343960702419281,
+      "learning_rate": 3.4775000000000005e-05,
+      "loss": 2.2894,
+      "step": 609
+    },
+    {
+      "epoch": 0.06809626970588133,
+      "grad_norm": 0.241934671998024,
+      "learning_rate": 3.475e-05,
+      "loss": 2.4438,
+      "step": 610
+    },
+    {
+      "epoch": 0.06820790293490736,
+      "grad_norm": 0.24266451597213745,
+      "learning_rate": 3.4725000000000004e-05,
+      "loss": 2.4135,
+      "step": 611
+    },
+    {
+      "epoch": 0.0683195361639334,
+      "grad_norm": 0.2379053235054016,
+      "learning_rate": 3.4699999999999996e-05,
+      "loss": 2.2596,
+      "step": 612
+    },
+    {
+      "epoch": 0.06843116939295943,
+      "grad_norm": 0.2375876009464264,
+      "learning_rate": 3.4675e-05,
+      "loss": 2.3748,
+      "step": 613
+    },
+    {
+      "epoch": 0.06854280262198546,
+      "grad_norm": 0.23114004731178284,
+      "learning_rate": 3.465e-05,
+      "loss": 2.4376,
+      "step": 614
+    },
+    {
+      "epoch": 0.0686544358510115,
+      "grad_norm": 0.38774147629737854,
+      "learning_rate": 3.4625e-05,
+      "loss": 2.381,
+      "step": 615
+    },
+    {
+      "epoch": 0.06876606908003753,
+      "grad_norm": 0.6831827163696289,
+      "learning_rate": 3.46e-05,
+      "loss": 2.3815,
+      "step": 616
+    },
+    {
+      "epoch": 0.06887770230906357,
+      "grad_norm": 0.25311988592147827,
+      "learning_rate": 3.4575e-05,
+      "loss": 2.3525,
+      "step": 617
+    },
+    {
+      "epoch": 0.06898933553808961,
+      "grad_norm": 0.2384188175201416,
+      "learning_rate": 3.455e-05,
+      "loss": 2.4084,
+      "step": 618
+    },
+    {
+      "epoch": 0.06910096876711565,
+      "grad_norm": 0.2516483962535858,
+      "learning_rate": 3.4525e-05,
+      "loss": 2.3143,
+      "step": 619
+    },
+    {
+      "epoch": 0.06921260199614168,
+      "grad_norm": 0.2384801059961319,
+      "learning_rate": 3.45e-05,
+      "loss": 2.3075,
+      "step": 620
+    },
+    {
+      "epoch": 0.06932423522516772,
+      "grad_norm": 0.23414179682731628,
+      "learning_rate": 3.4475000000000005e-05,
+      "loss": 2.3035,
+      "step": 621
+    },
+    {
+      "epoch": 0.06943586845419375,
+      "grad_norm": 0.25081461668014526,
+      "learning_rate": 3.445e-05,
+      "loss": 2.3292,
+      "step": 622
+    },
+    {
+      "epoch": 0.06954750168321978,
+      "grad_norm": 0.22738482058048248,
+      "learning_rate": 3.4425e-05,
+      "loss": 2.3412,
+      "step": 623
+    },
+    {
+      "epoch": 0.06965913491224582,
+      "grad_norm": 0.2414780855178833,
+      "learning_rate": 3.4399999999999996e-05,
+      "loss": 2.3674,
+      "step": 624
+    },
+    {
+      "epoch": 0.06977076814127185,
+      "grad_norm": 0.25145867466926575,
+      "learning_rate": 3.4375e-05,
+      "loss": 2.3414,
+      "step": 625
+    },
+    {
+      "epoch": 0.06988240137029789,
+      "grad_norm": 0.23038849234580994,
+      "learning_rate": 3.435e-05,
+      "loss": 2.3156,
+      "step": 626
+    },
+    {
+      "epoch": 0.06999403459932392,
+      "grad_norm": 0.23264938592910767,
+      "learning_rate": 3.4325e-05,
+      "loss": 2.2536,
+      "step": 627
+    },
+    {
+      "epoch": 0.07010566782834995,
+      "grad_norm": 0.24372529983520508,
+      "learning_rate": 3.430000000000001e-05,
+      "loss": 2.3152,
+      "step": 628
+    },
+    {
+      "epoch": 0.07021730105737599,
+      "grad_norm": 0.23900210857391357,
+      "learning_rate": 3.4275e-05,
+      "loss": 2.343,
+      "step": 629
+    },
+    {
+      "epoch": 0.07032893428640202,
+      "grad_norm": 0.236933633685112,
+      "learning_rate": 3.4250000000000006e-05,
+      "loss": 2.4373,
+      "step": 630
+    },
+    {
+      "epoch": 0.07044056751542806,
+      "grad_norm": 0.23982471227645874,
+      "learning_rate": 3.4225e-05,
+      "loss": 2.3498,
+      "step": 631
+    },
+    {
+      "epoch": 0.07055220074445409,
+      "grad_norm": 0.2617717385292053,
+      "learning_rate": 3.4200000000000005e-05,
+      "loss": 2.2457,
+      "step": 632
+    },
+    {
+      "epoch": 0.07066383397348013,
+      "grad_norm": 0.24056993424892426,
+      "learning_rate": 3.4175000000000004e-05,
+      "loss": 2.3543,
+      "step": 633
+    },
+    {
+      "epoch": 0.07077546720250616,
+      "grad_norm": 0.2435135543346405,
+      "learning_rate": 3.415e-05,
+      "loss": 2.415,
+      "step": 634
+    },
+    {
+      "epoch": 0.07088710043153221,
+      "grad_norm": 0.2404133677482605,
+      "learning_rate": 3.4125e-05,
+      "loss": 2.3479,
+      "step": 635
+    },
+    {
+      "epoch": 0.07099873366055824,
+      "grad_norm": 0.22524316608905792,
+      "learning_rate": 3.41e-05,
+      "loss": 2.3252,
+      "step": 636
+    },
+    {
+      "epoch": 0.07111036688958428,
+      "grad_norm": 0.23215116560459137,
+      "learning_rate": 3.4075e-05,
+      "loss": 2.3494,
+      "step": 637
+    },
+    {
+      "epoch": 0.07122200011861031,
+      "grad_norm": 0.24920037388801575,
+      "learning_rate": 3.405e-05,
+      "loss": 2.3705,
+      "step": 638
+    },
+    {
+      "epoch": 0.07133363334763634,
+      "grad_norm": 0.2321121245622635,
+      "learning_rate": 3.4025e-05,
+      "loss": 2.2816,
+      "step": 639
+    },
+    {
+      "epoch": 0.07144526657666238,
+      "grad_norm": 0.2596079111099243,
+      "learning_rate": 3.4000000000000007e-05,
+      "loss": 2.3672,
+      "step": 640
+    },
+    {
+      "epoch": 0.07155689980568841,
+      "grad_norm": 0.24075333774089813,
+      "learning_rate": 3.3975e-05,
+      "loss": 2.1981,
+      "step": 641
+    },
+    {
+      "epoch": 0.07166853303471445,
+      "grad_norm": 0.24256588518619537,
+      "learning_rate": 3.3950000000000005e-05,
+      "loss": 2.4488,
+      "step": 642
+    },
+    {
+      "epoch": 0.07178016626374048,
+      "grad_norm": 0.24132193624973297,
+      "learning_rate": 3.3925e-05,
+      "loss": 2.394,
+      "step": 643
+    },
+    {
+      "epoch": 0.07189179949276651,
+      "grad_norm": 0.23571936786174774,
+      "learning_rate": 3.3900000000000004e-05,
+      "loss": 2.3527,
+      "step": 644
+    },
+    {
+      "epoch": 0.07200343272179255,
+      "grad_norm": 0.24922168254852295,
+      "learning_rate": 3.3875000000000003e-05,
+      "loss": 2.504,
+      "step": 645
+    },
+    {
+      "epoch": 0.07211506595081858,
+      "grad_norm": 0.2349790334701538,
+      "learning_rate": 3.385e-05,
+      "loss": 2.375,
+      "step": 646
+    },
+    {
+      "epoch": 0.07222669917984462,
+      "grad_norm": 0.23674091696739197,
+      "learning_rate": 3.3825e-05,
+      "loss": 2.453,
+      "step": 647
+    },
+    {
+      "epoch": 0.07233833240887065,
+      "grad_norm": 0.24707044661045074,
+      "learning_rate": 3.38e-05,
+      "loss": 2.3144,
+      "step": 648
+    },
+    {
+      "epoch": 0.07244996563789668,
+      "grad_norm": 0.24542857706546783,
+      "learning_rate": 3.3775e-05,
+      "loss": 2.4941,
+      "step": 649
+    },
+    {
+      "epoch": 0.07256159886692272,
+      "grad_norm": 0.233075812458992,
+      "learning_rate": 3.375000000000001e-05,
+      "loss": 2.3719,
+      "step": 650
+    },
+    {
+      "epoch": 0.07267323209594877,
+      "grad_norm": 0.23478782176971436,
+      "learning_rate": 3.3725e-05,
+      "loss": 2.4255,
+      "step": 651
+    },
+    {
+      "epoch": 0.0727848653249748,
+      "grad_norm": 0.2310231477022171,
+      "learning_rate": 3.3700000000000006e-05,
+      "loss": 2.2017,
+      "step": 652
+    },
+    {
+      "epoch": 0.07289649855400084,
+      "grad_norm": 0.23209823668003082,
+      "learning_rate": 3.3675e-05,
+      "loss": 2.2443,
+      "step": 653
+    },
+    {
+      "epoch": 0.07300813178302687,
+      "grad_norm": 0.2376200407743454,
+      "learning_rate": 3.3650000000000005e-05,
+      "loss": 2.446,
+      "step": 654
+    },
+    {
+      "epoch": 0.0731197650120529,
+      "grad_norm": 0.23651504516601562,
+      "learning_rate": 3.3625000000000004e-05,
+      "loss": 2.4459,
+      "step": 655
+    },
+    {
+      "epoch": 0.07323139824107894,
+      "grad_norm": 0.25088030099868774,
+      "learning_rate": 3.3600000000000004e-05,
+      "loss": 2.4193,
+      "step": 656
+    },
+    {
+      "epoch": 0.07334303147010497,
+      "grad_norm": 0.24074560403823853,
+      "learning_rate": 3.3575e-05,
+      "loss": 2.2942,
+      "step": 657
+    },
+    {
+      "epoch": 0.073454664699131,
+      "grad_norm": 0.24596726894378662,
+      "learning_rate": 3.355e-05,
+      "loss": 2.3474,
+      "step": 658
+    },
+    {
+      "epoch": 0.07356629792815704,
+      "grad_norm": 0.23518167436122894,
+      "learning_rate": 3.3525e-05,
+      "loss": 2.4345,
+      "step": 659
+    },
+    {
+      "epoch": 0.07367793115718307,
+      "grad_norm": 0.23677384853363037,
+      "learning_rate": 3.35e-05,
+      "loss": 2.178,
+      "step": 660
+    },
+    {
+      "epoch": 0.07378956438620911,
+      "grad_norm": 0.24126943945884705,
+      "learning_rate": 3.3475e-05,
+      "loss": 2.3258,
+      "step": 661
+    },
+    {
+      "epoch": 0.07390119761523514,
+      "grad_norm": 0.24277062714099884,
+      "learning_rate": 3.345000000000001e-05,
+      "loss": 2.3129,
+      "step": 662
+    },
+    {
+      "epoch": 0.07401283084426118,
+      "grad_norm": 0.2481059432029724,
+      "learning_rate": 3.3425e-05,
+      "loss": 2.3619,
+      "step": 663
+    },
+    {
+      "epoch": 0.07412446407328721,
+      "grad_norm": 0.23601798713207245,
+      "learning_rate": 3.3400000000000005e-05,
+      "loss": 2.346,
+      "step": 664
+    },
+    {
+      "epoch": 0.07423609730231324,
+      "grad_norm": 0.23907043039798737,
+      "learning_rate": 3.3375e-05,
+      "loss": 2.333,
+      "step": 665
+    },
+    {
+      "epoch": 0.07434773053133928,
+      "grad_norm": 0.2404210865497589,
+      "learning_rate": 3.3350000000000004e-05,
+      "loss": 2.3695,
+      "step": 666
+    },
+    {
+      "epoch": 0.07445936376036531,
+      "grad_norm": 0.23311296105384827,
+      "learning_rate": 3.3325000000000004e-05,
+      "loss": 2.2827,
+      "step": 667
+    },
+    {
+      "epoch": 0.07457099698939136,
+      "grad_norm": 0.24998392164707184,
+      "learning_rate": 3.33e-05,
+      "loss": 2.2346,
+      "step": 668
+    },
+    {
+      "epoch": 0.0746826302184174,
+      "grad_norm": 0.23921620845794678,
+      "learning_rate": 3.3275e-05,
+      "loss": 2.4188,
+      "step": 669
+    },
+    {
+      "epoch": 0.07479426344744343,
+      "grad_norm": 0.23314505815505981,
+      "learning_rate": 3.325e-05,
+      "loss": 2.3997,
+      "step": 670
+    },
+    {
+      "epoch": 0.07490589667646946,
+      "grad_norm": 0.2334446907043457,
+      "learning_rate": 3.3225e-05,
+      "loss": 2.397,
+      "step": 671
+    },
+    {
+      "epoch": 0.0750175299054955,
+      "grad_norm": 0.23645921051502228,
+      "learning_rate": 3.32e-05,
+      "loss": 2.4039,
+      "step": 672
+    },
+    {
+      "epoch": 0.07512916313452153,
+      "grad_norm": 0.2596314251422882,
+      "learning_rate": 3.3175e-05,
+      "loss": 2.4392,
+      "step": 673
+    },
+    {
+      "epoch": 0.07524079636354757,
+      "grad_norm": 0.24032291769981384,
+      "learning_rate": 3.3150000000000006e-05,
+      "loss": 2.3858,
+      "step": 674
+    },
+    {
+      "epoch": 0.0753524295925736,
+      "grad_norm": 0.2412300407886505,
+      "learning_rate": 3.3125e-05,
+      "loss": 2.278,
+      "step": 675
+    },
+    {
+      "epoch": 0.07546406282159963,
+      "grad_norm": 0.225362628698349,
+      "learning_rate": 3.3100000000000005e-05,
+      "loss": 2.4296,
+      "step": 676
+    },
+    {
+      "epoch": 0.07557569605062567,
+      "grad_norm": 0.24731768667697906,
+      "learning_rate": 3.3075e-05,
+      "loss": 2.5028,
+      "step": 677
+    },
+    {
+      "epoch": 0.0756873292796517,
+      "grad_norm": 0.24850989878177643,
+      "learning_rate": 3.3050000000000004e-05,
+      "loss": 2.4462,
+      "step": 678
+    },
+    {
+      "epoch": 0.07579896250867774,
+      "grad_norm": 0.23657028377056122,
+      "learning_rate": 3.3025e-05,
+      "loss": 2.3677,
+      "step": 679
+    },
+    {
+      "epoch": 0.07591059573770377,
+      "grad_norm": 0.23830410838127136,
+      "learning_rate": 3.3e-05,
+      "loss": 2.3859,
+      "step": 680
+    },
+    {
+      "epoch": 0.0760222289667298,
+      "grad_norm": 0.23412565886974335,
+      "learning_rate": 3.2975e-05,
+      "loss": 2.3559,
+      "step": 681
+    },
+    {
+      "epoch": 0.07613386219575584,
+      "grad_norm": 0.2358182668685913,
+      "learning_rate": 3.295e-05,
+      "loss": 2.3619,
+      "step": 682
+    },
+    {
+      "epoch": 0.07624549542478187,
+      "grad_norm": 0.24088707566261292,
+      "learning_rate": 3.2925e-05,
+      "loss": 2.2953,
+      "step": 683
+    },
+    {
+      "epoch": 0.07635712865380792,
+      "grad_norm": 0.23123978078365326,
+      "learning_rate": 3.29e-05,
+      "loss": 2.33,
+      "step": 684
+    },
+    {
+      "epoch": 0.07646876188283395,
+      "grad_norm": 0.24349573254585266,
+      "learning_rate": 3.2875e-05,
+      "loss": 2.3789,
+      "step": 685
+    },
+    {
+      "epoch": 0.07658039511185999,
+      "grad_norm": 0.23393647372722626,
+      "learning_rate": 3.2850000000000006e-05,
+      "loss": 2.4458,
+      "step": 686
+    },
+    {
+      "epoch": 0.07669202834088602,
+      "grad_norm": 0.2344076782464981,
+      "learning_rate": 3.2825e-05,
+      "loss": 2.2232,
+      "step": 687
+    },
+    {
+      "epoch": 0.07680366156991206,
+      "grad_norm": 0.23426808416843414,
+      "learning_rate": 3.2800000000000004e-05,
+      "loss": 2.55,
+      "step": 688
+    },
+    {
+      "epoch": 0.07691529479893809,
+      "grad_norm": 0.24055257439613342,
+      "learning_rate": 3.2775e-05,
+      "loss": 2.4083,
+      "step": 689
+    },
+    {
+      "epoch": 0.07702692802796413,
+      "grad_norm": 0.23331885039806366,
+      "learning_rate": 3.275e-05,
+      "loss": 2.3025,
+      "step": 690
+    },
+    {
+      "epoch": 0.07713856125699016,
+      "grad_norm": 0.24212725460529327,
+      "learning_rate": 3.2725e-05,
+      "loss": 2.3205,
+      "step": 691
+    },
+    {
+      "epoch": 0.0772501944860162,
+      "grad_norm": 0.32642829418182373,
+      "learning_rate": 3.27e-05,
+      "loss": 2.2472,
+      "step": 692
+    },
+    {
+      "epoch": 0.07736182771504223,
+      "grad_norm": 0.30700212717056274,
+      "learning_rate": 3.2675e-05,
+      "loss": 2.3868,
+      "step": 693
+    },
+    {
+      "epoch": 0.07747346094406826,
+      "grad_norm": 0.23343181610107422,
+      "learning_rate": 3.265e-05,
+      "loss": 2.345,
+      "step": 694
+    },
+    {
+      "epoch": 0.0775850941730943,
+      "grad_norm": 0.23056460916996002,
+      "learning_rate": 3.2625e-05,
+      "loss": 2.3514,
+      "step": 695
+    },
+    {
+      "epoch": 0.07769672740212033,
+      "grad_norm": 0.24498891830444336,
+      "learning_rate": 3.26e-05,
+      "loss": 2.4623,
+      "step": 696
+    },
+    {
+      "epoch": 0.07780836063114636,
+      "grad_norm": 0.2408551424741745,
+      "learning_rate": 3.2575e-05,
+      "loss": 2.3371,
+      "step": 697
+    },
+    {
+      "epoch": 0.0779199938601724,
+      "grad_norm": 0.24386224150657654,
+      "learning_rate": 3.2550000000000005e-05,
+      "loss": 2.4246,
+      "step": 698
+    },
+    {
+      "epoch": 0.07803162708919843,
+      "grad_norm": 0.23228052258491516,
+      "learning_rate": 3.2525e-05,
+      "loss": 2.2636,
+      "step": 699
+    },
+    {
+      "epoch": 0.07814326031822447,
+      "grad_norm": 0.24233461916446686,
+      "learning_rate": 3.2500000000000004e-05,
+      "loss": 2.4194,
+      "step": 700
+    },
+    {
+      "epoch": 0.07825489354725051,
+      "grad_norm": 0.23138944804668427,
+      "learning_rate": 3.2474999999999997e-05,
+      "loss": 2.3944,
+      "step": 701
+    },
+    {
+      "epoch": 0.07836652677627655,
+      "grad_norm": 0.23382620513439178,
+      "learning_rate": 3.245e-05,
+      "loss": 2.2998,
+      "step": 702
+    },
+    {
+      "epoch": 0.07847816000530258,
+      "grad_norm": 0.2373252511024475,
+      "learning_rate": 3.2425e-05,
+      "loss": 2.3178,
+      "step": 703
+    },
+    {
+      "epoch": 0.07858979323432862,
+      "grad_norm": 0.22690844535827637,
+      "learning_rate": 3.24e-05,
+      "loss": 2.4387,
+      "step": 704
+    },
+    {
+      "epoch": 0.07870142646335465,
+      "grad_norm": 0.22850820422172546,
+      "learning_rate": 3.2375e-05,
+      "loss": 2.3459,
+      "step": 705
+    },
+    {
+      "epoch": 0.07881305969238069,
+      "grad_norm": 0.23736946284770966,
+      "learning_rate": 3.235e-05,
+      "loss": 2.2847,
+      "step": 706
+    },
+    {
+      "epoch": 0.07892469292140672,
+      "grad_norm": 0.241333469748497,
+      "learning_rate": 3.2325e-05,
+      "loss": 2.3021,
+      "step": 707
+    },
+    {
+      "epoch": 0.07903632615043275,
+      "grad_norm": 0.23721295595169067,
+      "learning_rate": 3.2300000000000006e-05,
+      "loss": 2.3129,
+      "step": 708
+    },
+    {
+      "epoch": 0.07914795937945879,
+      "grad_norm": 0.2383483350276947,
+      "learning_rate": 3.2275e-05,
+      "loss": 2.3561,
+      "step": 709
+    },
+    {
+      "epoch": 0.07925959260848482,
+      "grad_norm": 0.24383005499839783,
+      "learning_rate": 3.2250000000000005e-05,
+      "loss": 2.4278,
+      "step": 710
+    },
+    {
+      "epoch": 0.07937122583751086,
+      "grad_norm": 0.24906353652477264,
+      "learning_rate": 3.2225e-05,
+      "loss": 2.3474,
+      "step": 711
+    },
+    {
+      "epoch": 0.07948285906653689,
+      "grad_norm": 0.2293633371591568,
+      "learning_rate": 3.2200000000000003e-05,
+      "loss": 2.3338,
+      "step": 712
+    },
+    {
+      "epoch": 0.07959449229556292,
+      "grad_norm": 0.23942622542381287,
+      "learning_rate": 3.2175e-05,
+      "loss": 2.3455,
+      "step": 713
+    },
+    {
+      "epoch": 0.07970612552458896,
+      "grad_norm": 0.23788371682167053,
+      "learning_rate": 3.215e-05,
+      "loss": 2.3026,
+      "step": 714
+    },
+    {
+      "epoch": 0.07981775875361499,
+      "grad_norm": 0.2387734055519104,
+      "learning_rate": 3.2125e-05,
+      "loss": 2.316,
+      "step": 715
+    },
+    {
+      "epoch": 0.07992939198264103,
+      "grad_norm": 0.242325097322464,
+      "learning_rate": 3.21e-05,
+      "loss": 2.2869,
+      "step": 716
+    },
+    {
+      "epoch": 0.08004102521166707,
+      "grad_norm": 0.23410271108150482,
+      "learning_rate": 3.2075e-05,
+      "loss": 2.3509,
+      "step": 717
+    },
+    {
+      "epoch": 0.08015265844069311,
+      "grad_norm": 0.23277893662452698,
+      "learning_rate": 3.205e-05,
+      "loss": 2.3282,
+      "step": 718
+    },
+    {
+      "epoch": 0.08026429166971914,
+      "grad_norm": 1.3853453397750854,
+      "learning_rate": 3.2025e-05,
+      "loss": 2.3211,
+      "step": 719
+    },
+    {
+      "epoch": 0.08037592489874518,
+      "grad_norm": 0.23764613270759583,
+      "learning_rate": 3.2000000000000005e-05,
+      "loss": 2.4642,
+      "step": 720
+    },
+    {
+      "epoch": 0.08048755812777121,
+      "grad_norm": 0.2404954433441162,
+      "learning_rate": 3.1975e-05,
+      "loss": 2.3829,
+      "step": 721
+    },
+    {
+      "epoch": 0.08059919135679725,
+      "grad_norm": 0.2323405146598816,
+      "learning_rate": 3.1950000000000004e-05,
+      "loss": 2.4254,
+      "step": 722
+    },
+    {
+      "epoch": 0.08071082458582328,
+      "grad_norm": 0.24208155274391174,
+      "learning_rate": 3.1925e-05,
+      "loss": 2.3304,
+      "step": 723
+    },
+    {
+      "epoch": 0.08082245781484931,
+      "grad_norm": 0.2440410554409027,
+      "learning_rate": 3.19e-05,
+      "loss": 2.4517,
+      "step": 724
+    },
+    {
+      "epoch": 0.08093409104387535,
+      "grad_norm": 0.23571601510047913,
+      "learning_rate": 3.1875e-05,
+      "loss": 2.3367,
+      "step": 725
+    },
+    {
+      "epoch": 0.08104572427290138,
+      "grad_norm": 0.23201195895671844,
+      "learning_rate": 3.185e-05,
+      "loss": 2.3973,
+      "step": 726
+    },
+    {
+      "epoch": 0.08115735750192742,
+      "grad_norm": 0.23855586349964142,
+      "learning_rate": 3.1825e-05,
+      "loss": 2.3875,
+      "step": 727
+    },
+    {
+      "epoch": 0.08126899073095345,
+      "grad_norm": 0.23818083107471466,
+      "learning_rate": 3.18e-05,
+      "loss": 2.3538,
+      "step": 728
+    },
+    {
+      "epoch": 0.08138062395997948,
+      "grad_norm": 0.2586081326007843,
+      "learning_rate": 3.1775e-05,
+      "loss": 2.3232,
+      "step": 729
+    },
+    {
+      "epoch": 0.08149225718900552,
+      "grad_norm": 0.23521441221237183,
+      "learning_rate": 3.175e-05,
+      "loss": 2.2694,
+      "step": 730
+    },
+    {
+      "epoch": 0.08160389041803155,
+      "grad_norm": 0.4255395829677582,
+      "learning_rate": 3.1725e-05,
+      "loss": 2.3881,
+      "step": 731
+    },
+    {
+      "epoch": 0.08171552364705759,
+      "grad_norm": 0.22463633120059967,
+      "learning_rate": 3.1700000000000005e-05,
+      "loss": 2.3078,
+      "step": 732
+    },
+    {
+      "epoch": 0.08182715687608362,
+      "grad_norm": 0.2440212070941925,
+      "learning_rate": 3.1675e-05,
+      "loss": 2.3463,
+      "step": 733
+    },
+    {
+      "epoch": 0.08193879010510967,
+      "grad_norm": 0.22708596289157867,
+      "learning_rate": 3.1650000000000004e-05,
+      "loss": 2.3567,
+      "step": 734
+    },
+    {
+      "epoch": 0.0820504233341357,
+      "grad_norm": 0.26711180806159973,
+      "learning_rate": 3.1624999999999996e-05,
+      "loss": 2.4591,
+      "step": 735
+    },
+    {
+      "epoch": 0.08216205656316174,
+      "grad_norm": 0.23102666437625885,
+      "learning_rate": 3.16e-05,
+      "loss": 2.3309,
+      "step": 736
+    },
+    {
+      "epoch": 0.08227368979218777,
+      "grad_norm": 0.24265794456005096,
+      "learning_rate": 3.1575e-05,
+      "loss": 2.3203,
+      "step": 737
+    },
+    {
+      "epoch": 0.0823853230212138,
+      "grad_norm": 0.23661485314369202,
+      "learning_rate": 3.155e-05,
+      "loss": 2.3478,
+      "step": 738
+    },
+    {
+      "epoch": 0.08249695625023984,
+      "grad_norm": 0.23216097056865692,
+      "learning_rate": 3.1525e-05,
+      "loss": 2.4765,
+      "step": 739
+    },
+    {
+      "epoch": 0.08260858947926587,
+      "grad_norm": 0.2339819222688675,
+      "learning_rate": 3.15e-05,
+      "loss": 2.2802,
+      "step": 740
+    },
+    {
+      "epoch": 0.08272022270829191,
+      "grad_norm": 0.25498029589653015,
+      "learning_rate": 3.1475e-05,
+      "loss": 2.4064,
+      "step": 741
+    },
+    {
+      "epoch": 0.08283185593731794,
+      "grad_norm": 0.23578105866909027,
+      "learning_rate": 3.145e-05,
+      "loss": 2.3542,
+      "step": 742
+    },
+    {
+      "epoch": 0.08294348916634398,
+      "grad_norm": 0.23970484733581543,
+      "learning_rate": 3.1425e-05,
+      "loss": 2.3676,
+      "step": 743
+    },
+    {
+      "epoch": 0.08305512239537001,
+      "grad_norm": 0.2514541447162628,
+      "learning_rate": 3.1400000000000004e-05,
+      "loss": 2.4229,
+      "step": 744
+    },
+    {
+      "epoch": 0.08316675562439604,
+      "grad_norm": 0.23850007355213165,
+      "learning_rate": 3.1375e-05,
+      "loss": 2.3794,
+      "step": 745
+    },
+    {
+      "epoch": 0.08327838885342208,
+      "grad_norm": 0.30746811628341675,
+      "learning_rate": 3.135e-05,
+      "loss": 2.383,
+      "step": 746
+    },
+    {
+      "epoch": 0.08339002208244811,
+      "grad_norm": 0.23188377916812897,
+      "learning_rate": 3.1324999999999996e-05,
+      "loss": 2.336,
+      "step": 747
+    },
+    {
+      "epoch": 0.08350165531147415,
+      "grad_norm": 0.23534120619297028,
+      "learning_rate": 3.13e-05,
+      "loss": 2.398,
+      "step": 748
+    },
+    {
+      "epoch": 0.08361328854050018,
+      "grad_norm": 0.2442607581615448,
+      "learning_rate": 3.1275e-05,
+      "loss": 2.4357,
+      "step": 749
+    },
+    {
+      "epoch": 0.08372492176952623,
+      "grad_norm": 0.2328692376613617,
+      "learning_rate": 3.125e-05,
+      "loss": 2.3239,
+      "step": 750
+    },
+    {
+      "epoch": 0.08383655499855226,
+      "grad_norm": 0.29631370306015015,
+      "learning_rate": 3.122500000000001e-05,
+      "loss": 2.3165,
+      "step": 751
+    },
+    {
+      "epoch": 0.0839481882275783,
+      "grad_norm": 0.2432224601507187,
+      "learning_rate": 3.12e-05,
+      "loss": 2.372,
+      "step": 752
+    },
+    {
+      "epoch": 0.08405982145660433,
+      "grad_norm": 0.2435504049062729,
+      "learning_rate": 3.1175000000000006e-05,
+      "loss": 2.4804,
+      "step": 753
+    },
+    {
+      "epoch": 0.08417145468563036,
+      "grad_norm": 0.2799118161201477,
+      "learning_rate": 3.115e-05,
+      "loss": 2.3138,
+      "step": 754
+    },
+    {
+      "epoch": 0.0842830879146564,
+      "grad_norm": 0.23647980391979218,
+      "learning_rate": 3.1125000000000004e-05,
+      "loss": 2.4502,
+      "step": 755
+    },
+    {
+      "epoch": 0.08439472114368243,
+      "grad_norm": 0.23339681327342987,
+      "learning_rate": 3.1100000000000004e-05,
+      "loss": 2.3515,
+      "step": 756
+    },
+    {
+      "epoch": 0.08450635437270847,
+      "grad_norm": 0.2358621060848236,
+      "learning_rate": 3.1075e-05,
+      "loss": 2.3384,
+      "step": 757
+    },
+    {
+      "epoch": 0.0846179876017345,
+      "grad_norm": 0.226596862077713,
+      "learning_rate": 3.105e-05,
+      "loss": 2.4022,
+      "step": 758
+    },
+    {
+      "epoch": 0.08472962083076054,
+      "grad_norm": 0.24496255815029144,
+      "learning_rate": 3.1025e-05,
+      "loss": 2.4273,
+      "step": 759
+    },
+    {
+      "epoch": 0.08484125405978657,
+      "grad_norm": 0.24479149281978607,
+      "learning_rate": 3.1e-05,
+      "loss": 2.3464,
+      "step": 760
+    },
+    {
+      "epoch": 0.0849528872888126,
+      "grad_norm": 0.23078912496566772,
+      "learning_rate": 3.0975e-05,
+      "loss": 2.3267,
+      "step": 761
+    },
+    {
+      "epoch": 0.08506452051783864,
+      "grad_norm": 0.23886540532112122,
+      "learning_rate": 3.095e-05,
+      "loss": 2.243,
+      "step": 762
+    },
+    {
+      "epoch": 0.08517615374686467,
+      "grad_norm": 0.2510230839252472,
+      "learning_rate": 3.0925000000000006e-05,
+      "loss": 2.3675,
+      "step": 763
+    },
+    {
+      "epoch": 0.0852877869758907,
+      "grad_norm": 0.23274515569210052,
+      "learning_rate": 3.09e-05,
+      "loss": 2.3207,
+      "step": 764
+    },
+    {
+      "epoch": 0.08539942020491674,
+      "grad_norm": 0.26892465353012085,
+      "learning_rate": 3.0875000000000005e-05,
+      "loss": 2.4453,
+      "step": 765
+    },
+    {
+      "epoch": 0.08551105343394277,
+      "grad_norm": 0.24123644828796387,
+      "learning_rate": 3.0850000000000004e-05,
+      "loss": 2.287,
+      "step": 766
+    },
+    {
+      "epoch": 0.08562268666296882,
+      "grad_norm": 0.2416788935661316,
+      "learning_rate": 3.0825000000000004e-05,
+      "loss": 2.301,
+      "step": 767
+    },
+    {
+      "epoch": 0.08573431989199486,
+      "grad_norm": 0.23173028230667114,
+      "learning_rate": 3.08e-05,
+      "loss": 2.2896,
+      "step": 768
+    },
+    {
+      "epoch": 0.08584595312102089,
+      "grad_norm": 0.2286095917224884,
+      "learning_rate": 3.0775e-05,
+      "loss": 2.341,
+      "step": 769
+    },
+    {
+      "epoch": 0.08595758635004692,
+      "grad_norm": 0.24451370537281036,
+      "learning_rate": 3.075e-05,
+      "loss": 2.196,
+      "step": 770
+    },
+    {
+      "epoch": 0.08606921957907296,
+      "grad_norm": 0.26589682698249817,
+      "learning_rate": 3.0725e-05,
+      "loss": 2.2902,
+      "step": 771
+    },
+    {
+      "epoch": 0.08618085280809899,
+      "grad_norm": 0.26289981603622437,
+      "learning_rate": 3.07e-05,
+      "loss": 2.505,
+      "step": 772
+    },
+    {
+      "epoch": 0.08629248603712503,
+      "grad_norm": 0.2402532994747162,
+      "learning_rate": 3.067500000000001e-05,
+      "loss": 2.3008,
+      "step": 773
+    },
+    {
+      "epoch": 0.08640411926615106,
+      "grad_norm": 0.23948536813259125,
+      "learning_rate": 3.065e-05,
+      "loss": 2.4653,
+      "step": 774
+    },
+    {
+      "epoch": 0.0865157524951771,
+      "grad_norm": 0.2507634162902832,
+      "learning_rate": 3.0625000000000006e-05,
+      "loss": 2.2703,
+      "step": 775
+    },
+    {
+      "epoch": 0.08662738572420313,
+      "grad_norm": 0.24213804304599762,
+      "learning_rate": 3.06e-05,
+      "loss": 2.4244,
+      "step": 776
+    },
+    {
+      "epoch": 0.08673901895322916,
+      "grad_norm": 0.24074698984622955,
+      "learning_rate": 3.0575000000000005e-05,
+      "loss": 2.3714,
+      "step": 777
+    },
+    {
+      "epoch": 0.0868506521822552,
+      "grad_norm": 0.2556532323360443,
+      "learning_rate": 3.0550000000000004e-05,
+      "loss": 2.2702,
+      "step": 778
+    },
+    {
+      "epoch": 0.08696228541128123,
+      "grad_norm": 0.23637044429779053,
+      "learning_rate": 3.0525e-05,
+      "loss": 2.3382,
+      "step": 779
+    },
+    {
+      "epoch": 0.08707391864030727,
+      "grad_norm": 0.2363051176071167,
+      "learning_rate": 3.05e-05,
+      "loss": 2.2893,
+      "step": 780
+    },
+    {
+      "epoch": 0.0871855518693333,
+      "grad_norm": 0.23347924649715424,
+      "learning_rate": 3.0475000000000002e-05,
+      "loss": 2.4117,
+      "step": 781
+    },
+    {
+      "epoch": 0.08729718509835933,
+      "grad_norm": 0.37233465909957886,
+      "learning_rate": 3.045e-05,
+      "loss": 2.3556,
+      "step": 782
+    },
+    {
+      "epoch": 0.08740881832738538,
+      "grad_norm": 0.25742578506469727,
+      "learning_rate": 3.0425000000000004e-05,
+      "loss": 2.4087,
+      "step": 783
+    },
+    {
+      "epoch": 0.08752045155641142,
+      "grad_norm": 0.24065518379211426,
+      "learning_rate": 3.04e-05,
+      "loss": 2.2688,
+      "step": 784
+    },
+    {
+      "epoch": 0.08763208478543745,
+      "grad_norm": 0.2395828366279602,
+      "learning_rate": 3.0375000000000003e-05,
+      "loss": 2.3376,
+      "step": 785
+    },
+    {
+      "epoch": 0.08774371801446348,
+      "grad_norm": 0.30660635232925415,
+      "learning_rate": 3.035e-05,
+      "loss": 2.3353,
+      "step": 786
+    },
+    {
+      "epoch": 0.08785535124348952,
+      "grad_norm": 0.23347756266593933,
+      "learning_rate": 3.0325000000000002e-05,
+      "loss": 2.2991,
+      "step": 787
+    },
+    {
+      "epoch": 0.08796698447251555,
+      "grad_norm": 0.22873973846435547,
+      "learning_rate": 3.03e-05,
+      "loss": 2.3047,
+      "step": 788
+    },
+    {
+      "epoch": 0.08807861770154159,
+      "grad_norm": 0.23164629936218262,
+      "learning_rate": 3.0275000000000004e-05,
+      "loss": 2.3264,
+      "step": 789
+    },
+    {
+      "epoch": 0.08819025093056762,
+      "grad_norm": 0.2329261600971222,
+      "learning_rate": 3.025e-05,
+      "loss": 2.2868,
+      "step": 790
+    },
+    {
+      "epoch": 0.08830188415959365,
+      "grad_norm": 0.22929538786411285,
+      "learning_rate": 3.0225000000000003e-05,
+      "loss": 2.3671,
+      "step": 791
+    },
+    {
+      "epoch": 0.08841351738861969,
+      "grad_norm": 0.24005673825740814,
+      "learning_rate": 3.02e-05,
+      "loss": 2.4117,
+      "step": 792
+    },
+    {
+      "epoch": 0.08852515061764572,
+      "grad_norm": 0.22921884059906006,
+      "learning_rate": 3.0175e-05,
+      "loss": 2.4082,
+      "step": 793
+    },
+    {
+      "epoch": 0.08863678384667176,
+      "grad_norm": 0.2490616887807846,
+      "learning_rate": 3.015e-05,
+      "loss": 2.3136,
+      "step": 794
+    },
+    {
+      "epoch": 0.08874841707569779,
+      "grad_norm": 0.24170029163360596,
+      "learning_rate": 3.0125000000000004e-05,
+      "loss": 2.3903,
+      "step": 795
+    },
+    {
+      "epoch": 0.08886005030472383,
+      "grad_norm": 0.24534232914447784,
+      "learning_rate": 3.01e-05,
+      "loss": 2.3231,
+      "step": 796
+    },
+    {
+      "epoch": 0.08897168353374986,
+      "grad_norm": 0.23491844534873962,
+      "learning_rate": 3.0075000000000003e-05,
+      "loss": 2.3657,
+      "step": 797
+    },
+    {
+      "epoch": 0.0890833167627759,
+      "grad_norm": 0.23918871581554413,
+      "learning_rate": 3.0050000000000002e-05,
+      "loss": 2.3655,
+      "step": 798
+    },
+    {
+      "epoch": 0.08919494999180193,
+      "grad_norm": 0.23670734465122223,
+      "learning_rate": 3.0025000000000005e-05,
+      "loss": 2.3264,
+      "step": 799
+    },
+    {
+      "epoch": 0.08930658322082798,
+      "grad_norm": 0.2383037805557251,
+      "learning_rate": 3e-05,
+      "loss": 2.3465,
+      "step": 800
+    },
+    {
+      "epoch": 0.08941821644985401,
+      "grad_norm": 0.38153019547462463,
+      "learning_rate": 2.9975000000000004e-05,
+      "loss": 2.4639,
+      "step": 801
+    },
+    {
+      "epoch": 0.08952984967888004,
+      "grad_norm": 0.25272589921951294,
+      "learning_rate": 2.995e-05,
+      "loss": 2.253,
+      "step": 802
+    },
+    {
+      "epoch": 0.08964148290790608,
+      "grad_norm": 0.24903567135334015,
+      "learning_rate": 2.9925000000000002e-05,
+      "loss": 2.3757,
+      "step": 803
+    },
+    {
+      "epoch": 0.08975311613693211,
+      "grad_norm": 0.23128022253513336,
+      "learning_rate": 2.9900000000000002e-05,
+      "loss": 2.287,
+      "step": 804
+    },
+    {
+      "epoch": 0.08986474936595815,
+      "grad_norm": 0.23053546249866486,
+      "learning_rate": 2.9875000000000004e-05,
+      "loss": 2.4614,
+      "step": 805
+    },
+    {
+      "epoch": 0.08997638259498418,
+      "grad_norm": 0.3717436194419861,
+      "learning_rate": 2.985e-05,
+      "loss": 2.268,
+      "step": 806
+    },
+    {
+      "epoch": 0.09008801582401021,
+      "grad_norm": 0.4570637047290802,
+      "learning_rate": 2.9825000000000003e-05,
+      "loss": 2.3794,
+      "step": 807
+    },
+    {
+      "epoch": 0.09019964905303625,
+      "grad_norm": 0.2312982678413391,
+      "learning_rate": 2.98e-05,
+      "loss": 2.4141,
+      "step": 808
+    },
+    {
+      "epoch": 0.09031128228206228,
+      "grad_norm": 0.2467602640390396,
+      "learning_rate": 2.9775000000000002e-05,
+      "loss": 2.3287,
+      "step": 809
+    },
+    {
+      "epoch": 0.09042291551108832,
+      "grad_norm": 0.2299824357032776,
+      "learning_rate": 2.975e-05,
+      "loss": 2.4114,
+      "step": 810
+    },
+    {
+      "epoch": 0.09053454874011435,
+      "grad_norm": 3.2160747051239014,
+      "learning_rate": 2.9725000000000004e-05,
+      "loss": 2.4177,
+      "step": 811
+    },
+    {
+      "epoch": 0.09064618196914039,
+      "grad_norm": 0.2425236701965332,
+      "learning_rate": 2.97e-05,
+      "loss": 2.2538,
+      "step": 812
+    },
+    {
+      "epoch": 0.09075781519816642,
+      "grad_norm": 0.2397298812866211,
+      "learning_rate": 2.9675000000000003e-05,
+      "loss": 2.4145,
+      "step": 813
+    },
+    {
+      "epoch": 0.09086944842719245,
+      "grad_norm": 0.23437006771564484,
+      "learning_rate": 2.965e-05,
+      "loss": 2.2899,
+      "step": 814
+    },
+    {
+      "epoch": 0.09098108165621849,
+      "grad_norm": 0.2314252257347107,
+      "learning_rate": 2.9625000000000002e-05,
+      "loss": 2.3063,
+      "step": 815
+    },
+    {
+      "epoch": 0.09109271488524454,
+      "grad_norm": 0.24189533293247223,
+      "learning_rate": 2.96e-05,
+      "loss": 2.3931,
+      "step": 816
+    },
+    {
+      "epoch": 0.09120434811427057,
+      "grad_norm": 0.2523435950279236,
+      "learning_rate": 2.9575000000000004e-05,
+      "loss": 2.337,
+      "step": 817
+    },
+    {
+      "epoch": 0.0913159813432966,
+      "grad_norm": 0.26922857761383057,
+      "learning_rate": 2.955e-05,
+      "loss": 2.3536,
+      "step": 818
+    },
+    {
+      "epoch": 0.09142761457232264,
+      "grad_norm": 0.24634380638599396,
+      "learning_rate": 2.9525000000000003e-05,
+      "loss": 2.4549,
+      "step": 819
+    },
+    {
+      "epoch": 0.09153924780134867,
+      "grad_norm": 0.25171083211898804,
+      "learning_rate": 2.95e-05,
+      "loss": 2.2972,
+      "step": 820
+    },
+    {
+      "epoch": 0.0916508810303747,
+      "grad_norm": 0.6297338604927063,
+      "learning_rate": 2.9475e-05,
+      "loss": 2.3931,
+      "step": 821
+    },
+    {
+      "epoch": 0.09176251425940074,
+      "grad_norm": 0.2340405136346817,
+      "learning_rate": 2.945e-05,
+      "loss": 2.3981,
+      "step": 822
+    },
+    {
+      "epoch": 0.09187414748842677,
+      "grad_norm": 0.2873988449573517,
+      "learning_rate": 2.9425000000000004e-05,
+      "loss": 2.3315,
+      "step": 823
+    },
+    {
+      "epoch": 0.09198578071745281,
+      "grad_norm": 0.25433048605918884,
+      "learning_rate": 2.94e-05,
+      "loss": 2.2073,
+      "step": 824
+    },
+    {
+      "epoch": 0.09209741394647884,
+      "grad_norm": 0.23423077166080475,
+      "learning_rate": 2.9375000000000003e-05,
+      "loss": 2.3346,
+      "step": 825
+    },
+    {
+      "epoch": 0.09220904717550488,
+      "grad_norm": 0.2264171838760376,
+      "learning_rate": 2.935e-05,
+      "loss": 2.4019,
+      "step": 826
+    },
+    {
+      "epoch": 0.09232068040453091,
+      "grad_norm": 0.27230480313301086,
+      "learning_rate": 2.9325e-05,
+      "loss": 2.3569,
+      "step": 827
+    },
+    {
+      "epoch": 0.09243231363355695,
+      "grad_norm": 0.23226654529571533,
+      "learning_rate": 2.93e-05,
+      "loss": 2.4407,
+      "step": 828
+    },
+    {
+      "epoch": 0.09254394686258298,
+      "grad_norm": 0.23659475147724152,
+      "learning_rate": 2.9275000000000003e-05,
+      "loss": 2.3602,
+      "step": 829
+    },
+    {
+      "epoch": 0.09265558009160901,
+      "grad_norm": 0.2345886379480362,
+      "learning_rate": 2.925e-05,
+      "loss": 2.3828,
+      "step": 830
+    },
+    {
+      "epoch": 0.09276721332063505,
+      "grad_norm": 0.23473899066448212,
+      "learning_rate": 2.9225000000000002e-05,
+      "loss": 2.2976,
+      "step": 831
+    },
+    {
+      "epoch": 0.09287884654966108,
+      "grad_norm": 0.23485642671585083,
+      "learning_rate": 2.9199999999999998e-05,
+      "loss": 2.3747,
+      "step": 832
+    },
+    {
+      "epoch": 0.09299047977868713,
+      "grad_norm": 0.24248026311397552,
+      "learning_rate": 2.9175e-05,
+      "loss": 2.3549,
+      "step": 833
+    },
+    {
+      "epoch": 0.09310211300771316,
+      "grad_norm": 0.23264911770820618,
+      "learning_rate": 2.915e-05,
+      "loss": 2.4257,
+      "step": 834
+    },
+    {
+      "epoch": 0.0932137462367392,
+      "grad_norm": 0.25659480690956116,
+      "learning_rate": 2.9125000000000003e-05,
+      "loss": 2.229,
+      "step": 835
+    },
+    {
+      "epoch": 0.09332537946576523,
+      "grad_norm": 0.23583489656448364,
+      "learning_rate": 2.91e-05,
+      "loss": 2.3548,
+      "step": 836
+    },
+    {
+      "epoch": 0.09343701269479127,
+      "grad_norm": 0.23372192680835724,
+      "learning_rate": 2.9075000000000002e-05,
+      "loss": 2.2546,
+      "step": 837
+    },
+    {
+      "epoch": 0.0935486459238173,
+      "grad_norm": 0.2368670254945755,
+      "learning_rate": 2.9049999999999998e-05,
+      "loss": 2.2415,
+      "step": 838
+    },
+    {
+      "epoch": 0.09366027915284333,
+      "grad_norm": 0.24544133245944977,
+      "learning_rate": 2.9025e-05,
+      "loss": 2.3511,
+      "step": 839
+    },
+    {
+      "epoch": 0.09377191238186937,
+      "grad_norm": 0.2858797013759613,
+      "learning_rate": 2.9e-05,
+      "loss": 2.4491,
+      "step": 840
+    },
+    {
+      "epoch": 0.0938835456108954,
+      "grad_norm": 0.24954313039779663,
+      "learning_rate": 2.8975000000000003e-05,
+      "loss": 2.5859,
+      "step": 841
+    },
+    {
+      "epoch": 0.09399517883992144,
+      "grad_norm": 0.23716183006763458,
+      "learning_rate": 2.895e-05,
+      "loss": 2.2792,
+      "step": 842
+    },
+    {
+      "epoch": 0.09410681206894747,
+      "grad_norm": 0.23809632658958435,
+      "learning_rate": 2.8925000000000002e-05,
+      "loss": 2.2958,
+      "step": 843
+    },
+    {
+      "epoch": 0.0942184452979735,
+      "grad_norm": 0.23545247316360474,
+      "learning_rate": 2.8899999999999998e-05,
+      "loss": 2.3113,
+      "step": 844
+    },
+    {
+      "epoch": 0.09433007852699954,
+      "grad_norm": 0.2312520146369934,
+      "learning_rate": 2.8875e-05,
+      "loss": 2.2242,
+      "step": 845
+    },
+    {
+      "epoch": 0.09444171175602557,
+      "grad_norm": 0.23694823682308197,
+      "learning_rate": 2.885e-05,
+      "loss": 2.3629,
+      "step": 846
+    },
+    {
+      "epoch": 0.09455334498505161,
+      "grad_norm": 0.22986574470996857,
+      "learning_rate": 2.8825000000000003e-05,
+      "loss": 2.3428,
+      "step": 847
+    },
+    {
+      "epoch": 0.09466497821407764,
+      "grad_norm": 0.2305915206670761,
+      "learning_rate": 2.88e-05,
+      "loss": 2.291,
+      "step": 848
+    },
+    {
+      "epoch": 0.09477661144310369,
+      "grad_norm": 0.23731227219104767,
+      "learning_rate": 2.8775e-05,
+      "loss": 2.3141,
+      "step": 849
+    },
+    {
+      "epoch": 0.09488824467212972,
+      "grad_norm": 0.24682384729385376,
+      "learning_rate": 2.8749999999999997e-05,
+      "loss": 2.2654,
+      "step": 850
+    },
+    {
+      "epoch": 0.09499987790115576,
+      "grad_norm": 0.232358917593956,
+      "learning_rate": 2.8725e-05,
+      "loss": 2.3816,
+      "step": 851
+    },
+    {
+      "epoch": 0.09511151113018179,
+      "grad_norm": 0.23460093140602112,
+      "learning_rate": 2.87e-05,
+      "loss": 2.4469,
+      "step": 852
+    },
+    {
+      "epoch": 0.09522314435920783,
+      "grad_norm": 0.28891122341156006,
+      "learning_rate": 2.8675000000000002e-05,
+      "loss": 2.3192,
+      "step": 853
+    },
+    {
+      "epoch": 0.09533477758823386,
+      "grad_norm": 0.22858241200447083,
+      "learning_rate": 2.865e-05,
+      "loss": 2.3903,
+      "step": 854
+    },
+    {
+      "epoch": 0.0954464108172599,
+      "grad_norm": 0.2979806065559387,
+      "learning_rate": 2.8625e-05,
+      "loss": 2.3428,
+      "step": 855
+    },
+    {
+      "epoch": 0.09555804404628593,
+      "grad_norm": 0.22742506861686707,
+      "learning_rate": 2.86e-05,
+      "loss": 2.3086,
+      "step": 856
+    },
+    {
+      "epoch": 0.09566967727531196,
+      "grad_norm": 0.2474931925535202,
+      "learning_rate": 2.8575000000000003e-05,
+      "loss": 2.2446,
+      "step": 857
+    },
+    {
+      "epoch": 0.095781310504338,
+      "grad_norm": 0.23535043001174927,
+      "learning_rate": 2.855e-05,
+      "loss": 2.3,
+      "step": 858
+    },
+    {
+      "epoch": 0.09589294373336403,
+      "grad_norm": 0.2337024062871933,
+      "learning_rate": 2.8525000000000002e-05,
+      "loss": 2.3222,
+      "step": 859
+    },
+    {
+      "epoch": 0.09600457696239006,
+      "grad_norm": 0.23335273563861847,
+      "learning_rate": 2.8499999999999998e-05,
+      "loss": 2.3336,
+      "step": 860
+    },
+    {
+      "epoch": 0.0961162101914161,
+      "grad_norm": 0.23024383187294006,
+      "learning_rate": 2.8475e-05,
+      "loss": 2.3664,
+      "step": 861
+    },
+    {
+      "epoch": 0.09622784342044213,
+      "grad_norm": 0.27251824736595154,
+      "learning_rate": 2.845e-05,
+      "loss": 2.25,
+      "step": 862
+    },
+    {
+      "epoch": 0.09633947664946817,
+      "grad_norm": 0.23650218546390533,
+      "learning_rate": 2.8425000000000003e-05,
+      "loss": 2.3429,
+      "step": 863
+    },
+    {
+      "epoch": 0.0964511098784942,
+      "grad_norm": 0.23651504516601562,
+      "learning_rate": 2.84e-05,
+      "loss": 2.4,
+      "step": 864
+    },
+    {
+      "epoch": 0.09656274310752025,
+      "grad_norm": 0.2359198033809662,
+      "learning_rate": 2.8375000000000002e-05,
+      "loss": 2.3672,
+      "step": 865
+    },
+    {
+      "epoch": 0.09667437633654628,
+      "grad_norm": 0.2200855314731598,
+      "learning_rate": 2.8349999999999998e-05,
+      "loss": 2.3336,
+      "step": 866
+    },
+    {
+      "epoch": 0.09678600956557232,
+      "grad_norm": 0.23432110249996185,
+      "learning_rate": 2.8325e-05,
+      "loss": 2.3277,
+      "step": 867
+    },
+    {
+      "epoch": 0.09689764279459835,
+      "grad_norm": 0.2440565526485443,
+      "learning_rate": 2.83e-05,
+      "loss": 2.3388,
+      "step": 868
+    },
+    {
+      "epoch": 0.09700927602362439,
+      "grad_norm": 0.23634879291057587,
+      "learning_rate": 2.8275000000000003e-05,
+      "loss": 2.3551,
+      "step": 869
+    },
+    {
+      "epoch": 0.09712090925265042,
+      "grad_norm": 0.23254720866680145,
+      "learning_rate": 2.825e-05,
+      "loss": 2.4071,
+      "step": 870
+    },
+    {
+      "epoch": 0.09723254248167645,
+      "grad_norm": 0.25411373376846313,
+      "learning_rate": 2.8225e-05,
+      "loss": 2.3591,
+      "step": 871
+    },
+    {
+      "epoch": 0.09734417571070249,
+      "grad_norm": 0.23345574736595154,
+      "learning_rate": 2.8199999999999998e-05,
+      "loss": 2.2857,
+      "step": 872
+    },
+    {
+      "epoch": 0.09745580893972852,
+      "grad_norm": 0.23449824750423431,
+      "learning_rate": 2.8175e-05,
+      "loss": 2.297,
+      "step": 873
+    },
+    {
+      "epoch": 0.09756744216875456,
+      "grad_norm": 0.23664528131484985,
+      "learning_rate": 2.815e-05,
+      "loss": 2.4227,
+      "step": 874
+    },
+    {
+      "epoch": 0.09767907539778059,
+      "grad_norm": 0.22787226736545563,
+      "learning_rate": 2.8125000000000003e-05,
+      "loss": 2.4532,
+      "step": 875
+    },
+    {
+      "epoch": 0.09779070862680662,
+      "grad_norm": 0.22756721079349518,
+      "learning_rate": 2.8100000000000005e-05,
+      "loss": 2.2942,
+      "step": 876
+    },
+    {
+      "epoch": 0.09790234185583266,
+      "grad_norm": 0.23187048733234406,
+      "learning_rate": 2.8075e-05,
+      "loss": 2.4004,
+      "step": 877
+    },
+    {
+      "epoch": 0.09801397508485869,
+      "grad_norm": 0.23409034311771393,
+      "learning_rate": 2.8050000000000004e-05,
+      "loss": 2.3854,
+      "step": 878
+    },
+    {
+      "epoch": 0.09812560831388473,
+      "grad_norm": 0.23317299783229828,
+      "learning_rate": 2.8025e-05,
+      "loss": 2.365,
+      "step": 879
+    },
+    {
+      "epoch": 0.09823724154291076,
+      "grad_norm": 0.4648565948009491,
+      "learning_rate": 2.8000000000000003e-05,
+      "loss": 2.3482,
+      "step": 880
+    },
+    {
+      "epoch": 0.0983488747719368,
+      "grad_norm": 0.23336485028266907,
+      "learning_rate": 2.7975000000000002e-05,
+      "loss": 2.454,
+      "step": 881
+    },
+    {
+      "epoch": 0.09846050800096284,
+      "grad_norm": 0.22770002484321594,
+      "learning_rate": 2.7950000000000005e-05,
+      "loss": 2.2243,
+      "step": 882
+    },
+    {
+      "epoch": 0.09857214122998888,
+      "grad_norm": 0.240431547164917,
+      "learning_rate": 2.7925e-05,
+      "loss": 2.3568,
+      "step": 883
+    },
+    {
+      "epoch": 0.09868377445901491,
+      "grad_norm": 0.23338884115219116,
+      "learning_rate": 2.7900000000000004e-05,
+      "loss": 2.3459,
+      "step": 884
+    },
+    {
+      "epoch": 0.09879540768804095,
+      "grad_norm": 0.23423053324222565,
+      "learning_rate": 2.7875e-05,
+      "loss": 2.4424,
+      "step": 885
+    },
+    {
+      "epoch": 0.09890704091706698,
+      "grad_norm": 0.2247275859117508,
+      "learning_rate": 2.7850000000000003e-05,
+      "loss": 2.4401,
+      "step": 886
+    },
+    {
+      "epoch": 0.09901867414609301,
+      "grad_norm": 0.23924924433231354,
+      "learning_rate": 2.7825000000000002e-05,
+      "loss": 2.3376,
+      "step": 887
+    },
+    {
+      "epoch": 0.09913030737511905,
+      "grad_norm": 0.23396386206150055,
+      "learning_rate": 2.7800000000000005e-05,
+      "loss": 2.2607,
+      "step": 888
+    },
+    {
+      "epoch": 0.09924194060414508,
+      "grad_norm": 0.23403829336166382,
+      "learning_rate": 2.7775e-05,
+      "loss": 2.2647,
+      "step": 889
+    },
+    {
+      "epoch": 0.09935357383317112,
+      "grad_norm": 0.250621497631073,
+      "learning_rate": 2.7750000000000004e-05,
+      "loss": 2.3554,
+      "step": 890
+    },
+    {
+      "epoch": 0.09946520706219715,
+      "grad_norm": 0.24255424737930298,
+      "learning_rate": 2.7725e-05,
+      "loss": 2.3057,
+      "step": 891
+    },
+    {
+      "epoch": 0.09957684029122318,
+      "grad_norm": 0.23455750942230225,
+      "learning_rate": 2.7700000000000002e-05,
+      "loss": 2.405,
+      "step": 892
+    },
+    {
+      "epoch": 0.09968847352024922,
+      "grad_norm": 0.24191993474960327,
+      "learning_rate": 2.7675000000000002e-05,
+      "loss": 2.3389,
+      "step": 893
+    },
+    {
+      "epoch": 0.09980010674927525,
+      "grad_norm": 0.23159699141979218,
+      "learning_rate": 2.7650000000000005e-05,
+      "loss": 2.4599,
+      "step": 894
+    },
+    {
+      "epoch": 0.09991173997830129,
+      "grad_norm": 0.2873140871524811,
+      "learning_rate": 2.7625e-05,
+      "loss": 2.2902,
+      "step": 895
+    },
+    {
+      "epoch": 0.10002337320732732,
+      "grad_norm": 0.2346268594264984,
+      "learning_rate": 2.7600000000000003e-05,
+      "loss": 2.3954,
+      "step": 896
+    },
+    {
+      "epoch": 0.10013500643635335,
+      "grad_norm": 0.2276250720024109,
+      "learning_rate": 2.7575e-05,
+      "loss": 2.3288,
+      "step": 897
+    },
+    {
+      "epoch": 0.1002466396653794,
+      "grad_norm": 0.22809089720249176,
+      "learning_rate": 2.7550000000000002e-05,
+      "loss": 2.3797,
+      "step": 898
+    },
+    {
+      "epoch": 0.10035827289440544,
+      "grad_norm": 0.2589645981788635,
+      "learning_rate": 2.7525e-05,
+      "loss": 2.235,
+      "step": 899
+    },
+    {
+      "epoch": 0.10046990612343147,
+      "grad_norm": 0.26668986678123474,
+      "learning_rate": 2.7500000000000004e-05,
+      "loss": 2.4543,
+      "step": 900
+    },
+    {
+      "epoch": 0.1005815393524575,
+      "grad_norm": 0.231545552611351,
+      "learning_rate": 2.7475e-05,
+      "loss": 2.2753,
+      "step": 901
+    },
+    {
+      "epoch": 0.10069317258148354,
+      "grad_norm": 0.23608693480491638,
+      "learning_rate": 2.7450000000000003e-05,
+      "loss": 2.4207,
+      "step": 902
+    },
+    {
+      "epoch": 0.10080480581050957,
+      "grad_norm": 0.22971975803375244,
+      "learning_rate": 2.7425e-05,
+      "loss": 2.4783,
+      "step": 903
+    },
+    {
+      "epoch": 0.10091643903953561,
+      "grad_norm": 0.24497389793395996,
+      "learning_rate": 2.7400000000000002e-05,
+      "loss": 2.3443,
+      "step": 904
+    },
+    {
+      "epoch": 0.10102807226856164,
+      "grad_norm": 0.22968074679374695,
+      "learning_rate": 2.7375e-05,
+      "loss": 2.3163,
+      "step": 905
+    },
+    {
+      "epoch": 0.10113970549758768,
+      "grad_norm": 0.23204439878463745,
+      "learning_rate": 2.7350000000000004e-05,
+      "loss": 2.3681,
+      "step": 906
+    },
+    {
+      "epoch": 0.10125133872661371,
+      "grad_norm": 0.24272161722183228,
+      "learning_rate": 2.7325e-05,
+      "loss": 2.3472,
+      "step": 907
+    },
+    {
+      "epoch": 0.10136297195563974,
+      "grad_norm": 0.22961987555027008,
+      "learning_rate": 2.7300000000000003e-05,
+      "loss": 2.3187,
+      "step": 908
+    },
+    {
+      "epoch": 0.10147460518466578,
+      "grad_norm": 0.23540601134300232,
+      "learning_rate": 2.7275e-05,
+      "loss": 2.289,
+      "step": 909
+    },
+    {
+      "epoch": 0.10158623841369181,
+      "grad_norm": 0.260650634765625,
+      "learning_rate": 2.725e-05,
+      "loss": 2.3203,
+      "step": 910
+    },
+    {
+      "epoch": 0.10169787164271785,
+      "grad_norm": 0.24257007241249084,
+      "learning_rate": 2.7225e-05,
+      "loss": 2.3648,
+      "step": 911
+    },
+    {
+      "epoch": 0.10180950487174388,
+      "grad_norm": 0.2477046102285385,
+      "learning_rate": 2.7200000000000004e-05,
+      "loss": 2.273,
+      "step": 912
+    },
+    {
+      "epoch": 0.10192113810076991,
+      "grad_norm": 0.23077093064785004,
+      "learning_rate": 2.7175e-05,
+      "loss": 2.3581,
+      "step": 913
+    },
+    {
+      "epoch": 0.10203277132979595,
+      "grad_norm": 0.229270800948143,
+      "learning_rate": 2.7150000000000003e-05,
+      "loss": 2.3241,
+      "step": 914
+    },
+    {
+      "epoch": 0.102144404558822,
+      "grad_norm": 0.2295754998922348,
+      "learning_rate": 2.7125000000000002e-05,
+      "loss": 2.3711,
+      "step": 915
+    },
+    {
+      "epoch": 0.10225603778784803,
+      "grad_norm": 0.23900440335273743,
+      "learning_rate": 2.7100000000000005e-05,
+      "loss": 2.3576,
+      "step": 916
+    },
+    {
+      "epoch": 0.10236767101687407,
+      "grad_norm": 0.234444722533226,
+      "learning_rate": 2.7075e-05,
+      "loss": 2.3537,
+      "step": 917
+    },
+    {
+      "epoch": 0.1024793042459001,
+      "grad_norm": 0.2370821088552475,
+      "learning_rate": 2.7050000000000004e-05,
+      "loss": 2.3126,
+      "step": 918
+    },
+    {
+      "epoch": 0.10259093747492613,
+      "grad_norm": 0.24210244417190552,
+      "learning_rate": 2.7025e-05,
+      "loss": 2.2511,
+      "step": 919
+    },
+    {
+      "epoch": 0.10270257070395217,
+      "grad_norm": 0.23555943369865417,
+      "learning_rate": 2.7000000000000002e-05,
+      "loss": 2.3868,
+      "step": 920
+    },
+    {
+      "epoch": 0.1028142039329782,
+      "grad_norm": 0.2252027690410614,
+      "learning_rate": 2.6975000000000002e-05,
+      "loss": 2.3263,
+      "step": 921
+    },
+    {
+      "epoch": 0.10292583716200424,
+      "grad_norm": 0.22942887246608734,
+      "learning_rate": 2.6950000000000005e-05,
+      "loss": 2.3358,
+      "step": 922
+    },
+    {
+      "epoch": 0.10303747039103027,
+      "grad_norm": 0.23425108194351196,
+      "learning_rate": 2.6925e-05,
+      "loss": 2.279,
+      "step": 923
+    },
+    {
+      "epoch": 0.1031491036200563,
+      "grad_norm": 0.23959554731845856,
+      "learning_rate": 2.6900000000000003e-05,
+      "loss": 2.3456,
+      "step": 924
+    },
+    {
+      "epoch": 0.10326073684908234,
+      "grad_norm": 0.22143711149692535,
+      "learning_rate": 2.6875e-05,
+      "loss": 2.3346,
+      "step": 925
+    },
+    {
+      "epoch": 0.10337237007810837,
+      "grad_norm": 0.2322838306427002,
+      "learning_rate": 2.6850000000000002e-05,
+      "loss": 2.3597,
+      "step": 926
+    },
+    {
+      "epoch": 0.1034840033071344,
+      "grad_norm": 0.22317776083946228,
+      "learning_rate": 2.6825e-05,
+      "loss": 2.4395,
+      "step": 927
+    },
+    {
+      "epoch": 0.10359563653616044,
+      "grad_norm": 0.2381390780210495,
+      "learning_rate": 2.6800000000000004e-05,
+      "loss": 2.335,
+      "step": 928
+    },
+    {
+      "epoch": 0.10370726976518647,
+      "grad_norm": 0.2249373197555542,
+      "learning_rate": 2.6775e-05,
+      "loss": 2.3763,
+      "step": 929
+    },
+    {
+      "epoch": 0.10381890299421251,
+      "grad_norm": 0.23083436489105225,
+      "learning_rate": 2.6750000000000003e-05,
+      "loss": 2.2986,
+      "step": 930
+    },
+    {
+      "epoch": 0.10393053622323856,
+      "grad_norm": 0.23313601315021515,
+      "learning_rate": 2.6725e-05,
+      "loss": 2.4076,
+      "step": 931
+    },
+    {
+      "epoch": 0.10404216945226459,
+      "grad_norm": 0.22721858322620392,
+      "learning_rate": 2.6700000000000002e-05,
+      "loss": 2.2987,
+      "step": 932
+    },
+    {
+      "epoch": 0.10415380268129062,
+      "grad_norm": 0.23775102198123932,
+      "learning_rate": 2.6675e-05,
+      "loss": 2.3398,
+      "step": 933
+    },
+    {
+      "epoch": 0.10426543591031666,
+      "grad_norm": 0.22878248989582062,
+      "learning_rate": 2.6650000000000004e-05,
+      "loss": 2.3369,
+      "step": 934
+    },
+    {
+      "epoch": 0.10437706913934269,
+      "grad_norm": 0.22213736176490784,
+      "learning_rate": 2.6625e-05,
+      "loss": 2.4302,
+      "step": 935
+    },
+    {
+      "epoch": 0.10448870236836873,
+      "grad_norm": 0.23703357577323914,
+      "learning_rate": 2.6600000000000003e-05,
+      "loss": 2.2731,
+      "step": 936
+    },
+    {
+      "epoch": 0.10460033559739476,
+      "grad_norm": 0.2916199266910553,
+      "learning_rate": 2.6575e-05,
+      "loss": 2.2714,
+      "step": 937
+    },
+    {
+      "epoch": 0.1047119688264208,
+      "grad_norm": 0.22936727106571198,
+      "learning_rate": 2.655e-05,
+      "loss": 2.4308,
+      "step": 938
+    },
+    {
+      "epoch": 0.10482360205544683,
+      "grad_norm": 0.2333354949951172,
+      "learning_rate": 2.6525e-05,
+      "loss": 2.3236,
+      "step": 939
+    },
+    {
+      "epoch": 0.10493523528447286,
+      "grad_norm": 0.23450367152690887,
+      "learning_rate": 2.6500000000000004e-05,
+      "loss": 2.3854,
+      "step": 940
+    },
+    {
+      "epoch": 0.1050468685134989,
+      "grad_norm": 0.2431698888540268,
+      "learning_rate": 2.6475e-05,
+      "loss": 2.2789,
+      "step": 941
+    },
+    {
+      "epoch": 0.10515850174252493,
+      "grad_norm": 0.22219637036323547,
+      "learning_rate": 2.6450000000000003e-05,
+      "loss": 2.3675,
+      "step": 942
+    },
+    {
+      "epoch": 0.10527013497155097,
+      "grad_norm": 0.233125239610672,
+      "learning_rate": 2.6425e-05,
+      "loss": 2.4779,
+      "step": 943
+    },
+    {
+      "epoch": 0.105381768200577,
+      "grad_norm": 0.22824883460998535,
+      "learning_rate": 2.64e-05,
+      "loss": 2.3445,
+      "step": 944
+    },
+    {
+      "epoch": 0.10549340142960303,
+      "grad_norm": 0.22547942399978638,
+      "learning_rate": 2.6375e-05,
+      "loss": 2.234,
+      "step": 945
+    },
+    {
+      "epoch": 0.10560503465862907,
+      "grad_norm": 0.2325771003961563,
+      "learning_rate": 2.6350000000000004e-05,
+      "loss": 2.4215,
+      "step": 946
+    },
+    {
+      "epoch": 0.1057166678876551,
+      "grad_norm": 0.233683779835701,
+      "learning_rate": 2.6325e-05,
+      "loss": 2.4137,
+      "step": 947
+    },
+    {
+      "epoch": 0.10582830111668115,
+      "grad_norm": 0.24647918343544006,
+      "learning_rate": 2.6300000000000002e-05,
+      "loss": 2.3459,
+      "step": 948
+    },
+    {
+      "epoch": 0.10593993434570718,
+      "grad_norm": 0.22863587737083435,
+      "learning_rate": 2.6275e-05,
+      "loss": 2.3867,
+      "step": 949
+    },
+    {
+      "epoch": 0.10605156757473322,
+      "grad_norm": 0.23412172496318817,
+      "learning_rate": 2.625e-05,
+      "loss": 2.3621,
+      "step": 950
+    },
+    {
+      "epoch": 0.10616320080375925,
+      "grad_norm": 0.22972947359085083,
+      "learning_rate": 2.6225e-05,
+      "loss": 2.4242,
+      "step": 951
+    },
+    {
+      "epoch": 0.10627483403278529,
+      "grad_norm": 0.23886039853096008,
+      "learning_rate": 2.6200000000000003e-05,
+      "loss": 2.3375,
+      "step": 952
+    },
+    {
+      "epoch": 0.10638646726181132,
+      "grad_norm": 0.24040424823760986,
+      "learning_rate": 2.6175e-05,
+      "loss": 2.3518,
+      "step": 953
+    },
+    {
+      "epoch": 0.10649810049083736,
+      "grad_norm": 0.22699175775051117,
+      "learning_rate": 2.6150000000000002e-05,
+      "loss": 2.3435,
+      "step": 954
+    },
+    {
+      "epoch": 0.10660973371986339,
+      "grad_norm": 0.23695823550224304,
+      "learning_rate": 2.6124999999999998e-05,
+      "loss": 2.4036,
+      "step": 955
+    },
+    {
+      "epoch": 0.10672136694888942,
+      "grad_norm": 0.23027122020721436,
+      "learning_rate": 2.61e-05,
+      "loss": 2.286,
+      "step": 956
+    },
+    {
+      "epoch": 0.10683300017791546,
+      "grad_norm": 0.24779526889324188,
+      "learning_rate": 2.6075e-05,
+      "loss": 2.3144,
+      "step": 957
+    },
+    {
+      "epoch": 0.10694463340694149,
+      "grad_norm": 0.23636764287948608,
+      "learning_rate": 2.6050000000000003e-05,
+      "loss": 2.3114,
+      "step": 958
+    },
+    {
+      "epoch": 0.10705626663596753,
+      "grad_norm": 0.22950230538845062,
+      "learning_rate": 2.6025e-05,
+      "loss": 2.3594,
+      "step": 959
+    },
+    {
+      "epoch": 0.10716789986499356,
+      "grad_norm": 0.2355276644229889,
+      "learning_rate": 2.6000000000000002e-05,
+      "loss": 2.2385,
+      "step": 960
+    },
+    {
+      "epoch": 0.1072795330940196,
+      "grad_norm": 0.25076955556869507,
+      "learning_rate": 2.5974999999999998e-05,
+      "loss": 2.2637,
+      "step": 961
+    },
+    {
+      "epoch": 0.10739116632304563,
+      "grad_norm": 0.2247258871793747,
+      "learning_rate": 2.595e-05,
+      "loss": 2.2486,
+      "step": 962
+    },
+    {
+      "epoch": 0.10750279955207166,
+      "grad_norm": 0.21740855276584625,
+      "learning_rate": 2.5925e-05,
+      "loss": 2.2609,
+      "step": 963
+    },
+    {
+      "epoch": 0.10761443278109771,
+      "grad_norm": 0.2570677697658539,
+      "learning_rate": 2.5900000000000003e-05,
+      "loss": 2.442,
+      "step": 964
+    },
+    {
+      "epoch": 0.10772606601012374,
+      "grad_norm": 0.2322109192609787,
+      "learning_rate": 2.5875e-05,
+      "loss": 2.3026,
+      "step": 965
+    },
+    {
+      "epoch": 0.10783769923914978,
+      "grad_norm": 0.24020114541053772,
+      "learning_rate": 2.585e-05,
+      "loss": 2.351,
+      "step": 966
+    },
+    {
+      "epoch": 0.10794933246817581,
+      "grad_norm": 0.2280672937631607,
+      "learning_rate": 2.5824999999999998e-05,
+      "loss": 2.4016,
+      "step": 967
+    },
+    {
+      "epoch": 0.10806096569720185,
+      "grad_norm": 0.2309507429599762,
+      "learning_rate": 2.58e-05,
+      "loss": 2.3733,
+      "step": 968
+    },
+    {
+      "epoch": 0.10817259892622788,
+      "grad_norm": 0.2293708324432373,
+      "learning_rate": 2.5775e-05,
+      "loss": 2.3583,
+      "step": 969
+    },
+    {
+      "epoch": 0.10828423215525391,
+      "grad_norm": 0.23410175740718842,
+      "learning_rate": 2.5750000000000002e-05,
+      "loss": 2.352,
+      "step": 970
+    },
+    {
+      "epoch": 0.10839586538427995,
+      "grad_norm": 0.22972838580608368,
+      "learning_rate": 2.5725e-05,
+      "loss": 2.3434,
+      "step": 971
+    },
+    {
+      "epoch": 0.10850749861330598,
+      "grad_norm": 0.22537364065647125,
+      "learning_rate": 2.57e-05,
+      "loss": 2.3875,
+      "step": 972
+    },
+    {
+      "epoch": 0.10861913184233202,
+      "grad_norm": 0.23455362021923065,
+      "learning_rate": 2.5675e-05,
+      "loss": 2.2759,
+      "step": 973
+    },
+    {
+      "epoch": 0.10873076507135805,
+      "grad_norm": 0.23852132260799408,
+      "learning_rate": 2.5650000000000003e-05,
+      "loss": 2.3408,
+      "step": 974
+    },
+    {
+      "epoch": 0.10884239830038409,
+      "grad_norm": 0.23479855060577393,
+      "learning_rate": 2.5625e-05,
+      "loss": 2.3161,
+      "step": 975
+    },
+    {
+      "epoch": 0.10895403152941012,
+      "grad_norm": 0.2664550244808197,
+      "learning_rate": 2.5600000000000002e-05,
+      "loss": 2.387,
+      "step": 976
+    },
+    {
+      "epoch": 0.10906566475843615,
+      "grad_norm": 0.22990471124649048,
+      "learning_rate": 2.5574999999999998e-05,
+      "loss": 2.3906,
+      "step": 977
+    },
+    {
+      "epoch": 0.10917729798746219,
+      "grad_norm": 0.2335597425699234,
+      "learning_rate": 2.555e-05,
+      "loss": 2.3989,
+      "step": 978
+    },
+    {
+      "epoch": 0.10928893121648822,
+      "grad_norm": 0.23832087218761444,
+      "learning_rate": 2.5525e-05,
+      "loss": 2.2881,
+      "step": 979
+    },
+    {
+      "epoch": 0.10940056444551426,
+      "grad_norm": 0.23258844017982483,
+      "learning_rate": 2.5500000000000003e-05,
+      "loss": 2.4174,
+      "step": 980
+    },
+    {
+      "epoch": 0.1095121976745403,
+      "grad_norm": 0.23663833737373352,
+      "learning_rate": 2.5475e-05,
+      "loss": 2.2888,
+      "step": 981
+    },
+    {
+      "epoch": 0.10962383090356634,
+      "grad_norm": 0.22592462599277496,
+      "learning_rate": 2.5450000000000002e-05,
+      "loss": 2.3676,
+      "step": 982
+    },
+    {
+      "epoch": 0.10973546413259237,
+      "grad_norm": 0.23445037007331848,
+      "learning_rate": 2.5424999999999998e-05,
+      "loss": 2.3627,
+      "step": 983
+    },
+    {
+      "epoch": 0.1098470973616184,
+      "grad_norm": 0.2289026379585266,
+      "learning_rate": 2.54e-05,
+      "loss": 2.4259,
+      "step": 984
+    },
+    {
+      "epoch": 0.10995873059064444,
+      "grad_norm": 0.2335384339094162,
+      "learning_rate": 2.5375e-05,
+      "loss": 2.2903,
+      "step": 985
+    },
+    {
+      "epoch": 0.11007036381967047,
+      "grad_norm": 0.24791677296161652,
+      "learning_rate": 2.5350000000000003e-05,
+      "loss": 2.3497,
+      "step": 986
+    },
+    {
+      "epoch": 0.11018199704869651,
+      "grad_norm": 0.21857081353664398,
+      "learning_rate": 2.5325e-05,
+      "loss": 2.3938,
+      "step": 987
+    },
+    {
+      "epoch": 0.11029363027772254,
+      "grad_norm": 0.23194223642349243,
+      "learning_rate": 2.5300000000000002e-05,
+      "loss": 2.3752,
+      "step": 988
+    },
+    {
+      "epoch": 0.11040526350674858,
+      "grad_norm": 0.22554685175418854,
+      "learning_rate": 2.5274999999999998e-05,
+      "loss": 2.3892,
+      "step": 989
+    },
+    {
+      "epoch": 0.11051689673577461,
+      "grad_norm": 0.2444494664669037,
+      "learning_rate": 2.525e-05,
+      "loss": 2.3854,
+      "step": 990
+    },
+    {
+      "epoch": 0.11062852996480065,
+      "grad_norm": 0.24324128031730652,
+      "learning_rate": 2.5225e-05,
+      "loss": 2.2129,
+      "step": 991
+    },
+    {
+      "epoch": 0.11074016319382668,
+      "grad_norm": 0.23157966136932373,
+      "learning_rate": 2.5200000000000003e-05,
+      "loss": 2.3688,
+      "step": 992
+    },
+    {
+      "epoch": 0.11085179642285271,
+      "grad_norm": 0.23496349155902863,
+      "learning_rate": 2.5175e-05,
+      "loss": 2.2166,
+      "step": 993
+    },
+    {
+      "epoch": 0.11096342965187875,
+      "grad_norm": 0.22501815855503082,
+      "learning_rate": 2.515e-05,
+      "loss": 2.3563,
+      "step": 994
+    },
+    {
+      "epoch": 0.11107506288090478,
+      "grad_norm": 0.3032657206058502,
+      "learning_rate": 2.5124999999999997e-05,
+      "loss": 2.2783,
+      "step": 995
+    },
+    {
+      "epoch": 0.11118669610993082,
+      "grad_norm": 0.24155639111995697,
+      "learning_rate": 2.51e-05,
+      "loss": 2.3681,
+      "step": 996
+    },
+    {
+      "epoch": 0.11129832933895686,
+      "grad_norm": 0.22777613997459412,
+      "learning_rate": 2.5075e-05,
+      "loss": 2.3981,
+      "step": 997
+    },
+    {
+      "epoch": 0.1114099625679829,
+      "grad_norm": 0.24148933589458466,
+      "learning_rate": 2.5050000000000002e-05,
+      "loss": 2.2894,
+      "step": 998
+    },
+    {
+      "epoch": 0.11152159579700893,
+      "grad_norm": 0.23705993592739105,
+      "learning_rate": 2.5025e-05,
+      "loss": 2.3486,
+      "step": 999
+    },
+    {
+      "epoch": 0.11163322902603497,
+      "grad_norm": 0.23454095423221588,
+      "learning_rate": 2.5e-05,
+      "loss": 2.4232,
+      "step": 1000
+    },
+    {
+      "epoch": 0.111744862255061,
+      "grad_norm": 0.23154820501804352,
+      "learning_rate": 2.4975e-05,
+      "loss": 2.2269,
+      "step": 1001
+    },
+    {
+      "epoch": 0.11185649548408703,
+      "grad_norm": 0.22418555617332458,
+      "learning_rate": 2.495e-05,
+      "loss": 2.2112,
+      "step": 1002
+    },
+    {
+      "epoch": 0.11196812871311307,
+      "grad_norm": 0.2538329064846039,
+      "learning_rate": 2.4925000000000003e-05,
+      "loss": 2.3927,
+      "step": 1003
+    },
+    {
+      "epoch": 0.1120797619421391,
+      "grad_norm": 0.23294506967067719,
+      "learning_rate": 2.4900000000000002e-05,
+      "loss": 2.3024,
+      "step": 1004
+    },
+    {
+      "epoch": 0.11219139517116514,
+      "grad_norm": 0.22844459116458893,
+      "learning_rate": 2.4875e-05,
+      "loss": 2.2385,
+      "step": 1005
+    },
+    {
+      "epoch": 0.11230302840019117,
+      "grad_norm": 0.22828292846679688,
+      "learning_rate": 2.485e-05,
+      "loss": 2.2897,
+      "step": 1006
+    },
+    {
+      "epoch": 0.1124146616292172,
+      "grad_norm": 0.2693067491054535,
+      "learning_rate": 2.4825e-05,
+      "loss": 2.3818,
+      "step": 1007
+    },
+    {
+      "epoch": 0.11252629485824324,
+      "grad_norm": 0.22305937111377716,
+      "learning_rate": 2.48e-05,
+      "loss": 2.3505,
+      "step": 1008
+    },
+    {
+      "epoch": 0.11263792808726927,
+      "grad_norm": 0.23566830158233643,
+      "learning_rate": 2.4775000000000003e-05,
+      "loss": 2.3571,
+      "step": 1009
+    },
+    {
+      "epoch": 0.11274956131629531,
+      "grad_norm": 0.23289407789707184,
+      "learning_rate": 2.4750000000000002e-05,
+      "loss": 2.3021,
+      "step": 1010
+    },
+    {
+      "epoch": 0.11286119454532134,
+      "grad_norm": 0.2305288016796112,
+      "learning_rate": 2.4725e-05,
+      "loss": 2.3082,
+      "step": 1011
+    },
+    {
+      "epoch": 0.11297282777434738,
+      "grad_norm": 0.2262507677078247,
+      "learning_rate": 2.47e-05,
+      "loss": 2.3995,
+      "step": 1012
+    },
+    {
+      "epoch": 0.11308446100337341,
+      "grad_norm": 0.2277233600616455,
+      "learning_rate": 2.4675e-05,
+      "loss": 2.3759,
+      "step": 1013
+    },
+    {
+      "epoch": 0.11319609423239946,
+      "grad_norm": 0.24634131789207458,
+      "learning_rate": 2.465e-05,
+      "loss": 2.3487,
+      "step": 1014
+    },
+    {
+      "epoch": 0.11330772746142549,
+      "grad_norm": 0.31584781408309937,
+      "learning_rate": 2.4625000000000002e-05,
+      "loss": 2.3787,
+      "step": 1015
+    },
+    {
+      "epoch": 0.11341936069045153,
+      "grad_norm": 0.2360941767692566,
+      "learning_rate": 2.46e-05,
+      "loss": 2.2368,
+      "step": 1016
+    },
+    {
+      "epoch": 0.11353099391947756,
+      "grad_norm": 0.24138571321964264,
+      "learning_rate": 2.4575e-05,
+      "loss": 2.2329,
+      "step": 1017
+    },
+    {
+      "epoch": 0.1136426271485036,
+      "grad_norm": 0.24359650909900665,
+      "learning_rate": 2.455e-05,
+      "loss": 2.352,
+      "step": 1018
+    },
+    {
+      "epoch": 0.11375426037752963,
+      "grad_norm": 0.22765910625457764,
+      "learning_rate": 2.4525e-05,
+      "loss": 2.4487,
+      "step": 1019
+    },
+    {
+      "epoch": 0.11386589360655566,
+      "grad_norm": 0.22311876714229584,
+      "learning_rate": 2.45e-05,
+      "loss": 2.3413,
+      "step": 1020
+    },
+    {
+      "epoch": 0.1139775268355817,
+      "grad_norm": 0.3245187997817993,
+      "learning_rate": 2.4475000000000002e-05,
+      "loss": 2.352,
+      "step": 1021
+    },
+    {
+      "epoch": 0.11408916006460773,
+      "grad_norm": 0.267455130815506,
+      "learning_rate": 2.445e-05,
+      "loss": 2.4273,
+      "step": 1022
+    },
+    {
+      "epoch": 0.11420079329363376,
+      "grad_norm": 0.23805475234985352,
+      "learning_rate": 2.4425e-05,
+      "loss": 2.3511,
+      "step": 1023
+    },
+    {
+      "epoch": 0.1143124265226598,
+      "grad_norm": 0.23947173357009888,
+      "learning_rate": 2.44e-05,
+      "loss": 2.3554,
+      "step": 1024
+    },
+    {
+      "epoch": 0.11442405975168583,
+      "grad_norm": 0.23551489412784576,
+      "learning_rate": 2.4375e-05,
+      "loss": 2.3869,
+      "step": 1025
+    },
+    {
+      "epoch": 0.11453569298071187,
+      "grad_norm": 0.22201520204544067,
+      "learning_rate": 2.435e-05,
+      "loss": 2.351,
+      "step": 1026
+    },
+    {
+      "epoch": 0.1146473262097379,
+      "grad_norm": 0.23246638476848602,
+      "learning_rate": 2.4325000000000002e-05,
+      "loss": 2.3709,
+      "step": 1027
+    },
+    {
+      "epoch": 0.11475895943876394,
+      "grad_norm": 0.29901397228240967,
+      "learning_rate": 2.43e-05,
+      "loss": 2.33,
+      "step": 1028
+    },
+    {
+      "epoch": 0.11487059266778997,
+      "grad_norm": 0.2317001223564148,
+      "learning_rate": 2.4275e-05,
+      "loss": 2.4711,
+      "step": 1029
+    },
+    {
+      "epoch": 0.11498222589681602,
+      "grad_norm": 0.2264910489320755,
+      "learning_rate": 2.425e-05,
+      "loss": 2.4613,
+      "step": 1030
+    },
+    {
+      "epoch": 0.11509385912584205,
+      "grad_norm": 0.23516049981117249,
+      "learning_rate": 2.4225e-05,
+      "loss": 2.3831,
+      "step": 1031
+    },
+    {
+      "epoch": 0.11520549235486809,
+      "grad_norm": 0.23533384501934052,
+      "learning_rate": 2.4200000000000002e-05,
+      "loss": 2.4065,
+      "step": 1032
+    },
+    {
+      "epoch": 0.11531712558389412,
+      "grad_norm": 0.22421786189079285,
+      "learning_rate": 2.4175e-05,
+      "loss": 2.4121,
+      "step": 1033
+    },
+    {
+      "epoch": 0.11542875881292015,
+      "grad_norm": 0.23271812498569489,
+      "learning_rate": 2.415e-05,
+      "loss": 2.3377,
+      "step": 1034
+    },
+    {
+      "epoch": 0.11554039204194619,
+      "grad_norm": 0.22628925740718842,
+      "learning_rate": 2.4125e-05,
+      "loss": 2.3516,
+      "step": 1035
+    },
+    {
+      "epoch": 0.11565202527097222,
+      "grad_norm": 0.23225760459899902,
+      "learning_rate": 2.41e-05,
+      "loss": 2.4267,
+      "step": 1036
+    },
+    {
+      "epoch": 0.11576365849999826,
+      "grad_norm": 0.24704919755458832,
+      "learning_rate": 2.4075e-05,
+      "loss": 2.348,
+      "step": 1037
+    },
+    {
+      "epoch": 0.11587529172902429,
+      "grad_norm": 0.3677544593811035,
+      "learning_rate": 2.4050000000000002e-05,
+      "loss": 2.3355,
+      "step": 1038
+    },
+    {
+      "epoch": 0.11598692495805032,
+      "grad_norm": 0.2303479164838791,
+      "learning_rate": 2.4025e-05,
+      "loss": 2.3829,
+      "step": 1039
+    },
+    {
+      "epoch": 0.11609855818707636,
+      "grad_norm": 0.5053055882453918,
+      "learning_rate": 2.4e-05,
+      "loss": 2.2684,
+      "step": 1040
+    },
+    {
+      "epoch": 0.11621019141610239,
+      "grad_norm": 0.23576150834560394,
+      "learning_rate": 2.3975e-05,
+      "loss": 2.3628,
+      "step": 1041
+    },
+    {
+      "epoch": 0.11632182464512843,
+      "grad_norm": 0.22880171239376068,
+      "learning_rate": 2.395e-05,
+      "loss": 2.33,
+      "step": 1042
+    },
+    {
+      "epoch": 0.11643345787415446,
+      "grad_norm": 0.225894957780838,
+      "learning_rate": 2.3925e-05,
+      "loss": 2.2474,
+      "step": 1043
+    },
+    {
+      "epoch": 0.1165450911031805,
+      "grad_norm": 0.2277292162179947,
+      "learning_rate": 2.39e-05,
+      "loss": 2.32,
+      "step": 1044
+    },
+    {
+      "epoch": 0.11665672433220653,
+      "grad_norm": 0.22571003437042236,
+      "learning_rate": 2.3875e-05,
+      "loss": 2.3058,
+      "step": 1045
+    },
+    {
+      "epoch": 0.11676835756123256,
+      "grad_norm": 0.2490728795528412,
+      "learning_rate": 2.385e-05,
+      "loss": 2.38,
+      "step": 1046
+    },
+    {
+      "epoch": 0.11687999079025861,
+      "grad_norm": 0.23154285550117493,
+      "learning_rate": 2.3825e-05,
+      "loss": 2.3528,
+      "step": 1047
+    },
+    {
+      "epoch": 0.11699162401928465,
+      "grad_norm": 0.23180274665355682,
+      "learning_rate": 2.38e-05,
+      "loss": 2.349,
+      "step": 1048
+    },
+    {
+      "epoch": 0.11710325724831068,
+      "grad_norm": 0.2314680814743042,
+      "learning_rate": 2.3775e-05,
+      "loss": 2.2209,
+      "step": 1049
+    },
+    {
+      "epoch": 0.11721489047733671,
+      "grad_norm": 0.22533871233463287,
+      "learning_rate": 2.375e-05,
+      "loss": 2.2946,
+      "step": 1050
+    },
+    {
+      "epoch": 0.11732652370636275,
+      "grad_norm": 0.23475436866283417,
+      "learning_rate": 2.3725e-05,
+      "loss": 2.3839,
+      "step": 1051
+    },
+    {
+      "epoch": 0.11743815693538878,
+      "grad_norm": 0.22763217985630035,
+      "learning_rate": 2.37e-05,
+      "loss": 2.3039,
+      "step": 1052
+    },
+    {
+      "epoch": 0.11754979016441482,
+      "grad_norm": 0.22953568398952484,
+      "learning_rate": 2.3675e-05,
+      "loss": 2.4163,
+      "step": 1053
+    },
+    {
+      "epoch": 0.11766142339344085,
+      "grad_norm": 0.2377011775970459,
+      "learning_rate": 2.365e-05,
+      "loss": 2.3473,
+      "step": 1054
+    },
+    {
+      "epoch": 0.11777305662246688,
+      "grad_norm": 0.23042835295200348,
+      "learning_rate": 2.3624999999999998e-05,
+      "loss": 2.3909,
+      "step": 1055
+    },
+    {
+      "epoch": 0.11788468985149292,
+      "grad_norm": 0.23968364298343658,
+      "learning_rate": 2.36e-05,
+      "loss": 2.3023,
+      "step": 1056
+    },
+    {
+      "epoch": 0.11799632308051895,
+      "grad_norm": 0.23102299869060516,
+      "learning_rate": 2.3575e-05,
+      "loss": 2.3963,
+      "step": 1057
+    },
+    {
+      "epoch": 0.11810795630954499,
+      "grad_norm": 0.42533349990844727,
+      "learning_rate": 2.355e-05,
+      "loss": 2.396,
+      "step": 1058
+    },
+    {
+      "epoch": 0.11821958953857102,
+      "grad_norm": 0.2384020835161209,
+      "learning_rate": 2.3525e-05,
+      "loss": 2.4379,
+      "step": 1059
+    },
+    {
+      "epoch": 0.11833122276759706,
+      "grad_norm": 0.2910504639148712,
+      "learning_rate": 2.35e-05,
+      "loss": 2.3607,
+      "step": 1060
+    },
+    {
+      "epoch": 0.11844285599662309,
+      "grad_norm": 0.23686139285564423,
+      "learning_rate": 2.3475e-05,
+      "loss": 2.3374,
+      "step": 1061
+    },
+    {
+      "epoch": 0.11855448922564912,
+      "grad_norm": 0.27755534648895264,
+      "learning_rate": 2.345e-05,
+      "loss": 2.3719,
+      "step": 1062
+    },
+    {
+      "epoch": 0.11866612245467517,
+      "grad_norm": 0.23504361510276794,
+      "learning_rate": 2.3425000000000004e-05,
+      "loss": 2.4556,
+      "step": 1063
+    },
+    {
+      "epoch": 0.1187777556837012,
+      "grad_norm": 0.21698464453220367,
+      "learning_rate": 2.3400000000000003e-05,
+      "loss": 2.3708,
+      "step": 1064
+    },
+    {
+      "epoch": 0.11888938891272724,
+      "grad_norm": 0.2256624847650528,
+      "learning_rate": 2.3375000000000002e-05,
+      "loss": 2.2417,
+      "step": 1065
+    },
+    {
+      "epoch": 0.11900102214175327,
+      "grad_norm": 0.29147782921791077,
+      "learning_rate": 2.3350000000000002e-05,
+      "loss": 2.3808,
+      "step": 1066
+    },
+    {
+      "epoch": 0.11911265537077931,
+      "grad_norm": 0.3733128607273102,
+      "learning_rate": 2.3325e-05,
+      "loss": 2.3823,
+      "step": 1067
+    },
+    {
+      "epoch": 0.11922428859980534,
+      "grad_norm": 0.2350093573331833,
+      "learning_rate": 2.3300000000000004e-05,
+      "loss": 2.2919,
+      "step": 1068
+    },
+    {
+      "epoch": 0.11933592182883138,
+      "grad_norm": 0.23306138813495636,
+      "learning_rate": 2.3275000000000003e-05,
+      "loss": 2.3852,
+      "step": 1069
+    },
+    {
+      "epoch": 0.11944755505785741,
+      "grad_norm": 0.2352742701768875,
+      "learning_rate": 2.3250000000000003e-05,
+      "loss": 2.2804,
+      "step": 1070
+    },
+    {
+      "epoch": 0.11955918828688344,
+      "grad_norm": 0.2168835997581482,
+      "learning_rate": 2.3225000000000002e-05,
+      "loss": 2.3675,
+      "step": 1071
+    },
+    {
+      "epoch": 0.11967082151590948,
+      "grad_norm": 0.23424020409584045,
+      "learning_rate": 2.32e-05,
+      "loss": 2.4122,
+      "step": 1072
+    },
+    {
+      "epoch": 0.11978245474493551,
+      "grad_norm": 0.21874938905239105,
+      "learning_rate": 2.3175e-05,
+      "loss": 2.3381,
+      "step": 1073
+    },
+    {
+      "epoch": 0.11989408797396155,
+      "grad_norm": 0.2390536516904831,
+      "learning_rate": 2.3150000000000004e-05,
+      "loss": 2.2756,
+      "step": 1074
+    },
+    {
+      "epoch": 0.12000572120298758,
+      "grad_norm": 0.2291589379310608,
+      "learning_rate": 2.3125000000000003e-05,
+      "loss": 2.3482,
+      "step": 1075
+    },
+    {
+      "epoch": 0.12011735443201361,
+      "grad_norm": 0.2258395105600357,
+      "learning_rate": 2.3100000000000002e-05,
+      "loss": 2.3838,
+      "step": 1076
+    },
+    {
+      "epoch": 0.12022898766103965,
+      "grad_norm": 0.25954297184944153,
+      "learning_rate": 2.3075000000000002e-05,
+      "loss": 2.2275,
+      "step": 1077
+    },
+    {
+      "epoch": 0.12034062089006568,
+      "grad_norm": 0.23486246168613434,
+      "learning_rate": 2.305e-05,
+      "loss": 2.3866,
+      "step": 1078
+    },
+    {
+      "epoch": 0.12045225411909172,
+      "grad_norm": 0.22818194329738617,
+      "learning_rate": 2.3025e-05,
+      "loss": 2.3244,
+      "step": 1079
+    },
+    {
+      "epoch": 0.12056388734811777,
+      "grad_norm": 0.2317325472831726,
+      "learning_rate": 2.3000000000000003e-05,
+      "loss": 2.2603,
+      "step": 1080
+    },
+    {
+      "epoch": 0.1206755205771438,
+      "grad_norm": 0.24170775711536407,
+      "learning_rate": 2.2975000000000003e-05,
+      "loss": 2.3409,
+      "step": 1081
+    },
+    {
+      "epoch": 0.12078715380616983,
+      "grad_norm": 0.23717226088047028,
+      "learning_rate": 2.2950000000000002e-05,
+      "loss": 2.2956,
+      "step": 1082
+    },
+    {
+      "epoch": 0.12089878703519587,
+      "grad_norm": 0.21983082592487335,
+      "learning_rate": 2.2925e-05,
+      "loss": 2.3578,
+      "step": 1083
+    },
+    {
+      "epoch": 0.1210104202642219,
+      "grad_norm": 0.23536662757396698,
+      "learning_rate": 2.29e-05,
+      "loss": 2.3317,
+      "step": 1084
+    },
+    {
+      "epoch": 0.12112205349324794,
+      "grad_norm": 0.2405475676059723,
+      "learning_rate": 2.2875e-05,
+      "loss": 2.3663,
+      "step": 1085
+    },
+    {
+      "epoch": 0.12123368672227397,
+      "grad_norm": 0.23267629742622375,
+      "learning_rate": 2.2850000000000003e-05,
+      "loss": 2.3376,
+      "step": 1086
+    },
+    {
+      "epoch": 0.1213453199513,
+      "grad_norm": 0.22715060412883759,
+      "learning_rate": 2.2825000000000003e-05,
+      "loss": 2.5256,
+      "step": 1087
+    },
+    {
+      "epoch": 0.12145695318032604,
+      "grad_norm": 0.23004890978336334,
+      "learning_rate": 2.2800000000000002e-05,
+      "loss": 2.3706,
+      "step": 1088
+    },
+    {
+      "epoch": 0.12156858640935207,
+      "grad_norm": 0.22569864988327026,
+      "learning_rate": 2.2775e-05,
+      "loss": 2.29,
+      "step": 1089
+    },
+    {
+      "epoch": 0.1216802196383781,
+      "grad_norm": 0.22686640918254852,
+      "learning_rate": 2.275e-05,
+      "loss": 2.3978,
+      "step": 1090
+    },
+    {
+      "epoch": 0.12179185286740414,
+      "grad_norm": 0.26110976934432983,
+      "learning_rate": 2.2725000000000003e-05,
+      "loss": 2.4019,
+      "step": 1091
+    },
+    {
+      "epoch": 0.12190348609643017,
+      "grad_norm": 0.22089050710201263,
+      "learning_rate": 2.2700000000000003e-05,
+      "loss": 2.3078,
+      "step": 1092
+    },
+    {
+      "epoch": 0.12201511932545621,
+      "grad_norm": 0.22513218224048615,
+      "learning_rate": 2.2675000000000002e-05,
+      "loss": 2.3088,
+      "step": 1093
+    },
+    {
+      "epoch": 0.12212675255448224,
+      "grad_norm": 0.2333805114030838,
+      "learning_rate": 2.265e-05,
+      "loss": 2.2955,
+      "step": 1094
+    },
+    {
+      "epoch": 0.12223838578350828,
+      "grad_norm": 0.22828614711761475,
+      "learning_rate": 2.2625e-05,
+      "loss": 2.2746,
+      "step": 1095
+    },
+    {
+      "epoch": 0.12235001901253433,
+      "grad_norm": 0.23725035786628723,
+      "learning_rate": 2.26e-05,
+      "loss": 2.5234,
+      "step": 1096
+    },
+    {
+      "epoch": 0.12246165224156036,
+      "grad_norm": 0.24143311381340027,
+      "learning_rate": 2.2575000000000003e-05,
+      "loss": 2.2837,
+      "step": 1097
+    },
+    {
+      "epoch": 0.1225732854705864,
+      "grad_norm": 0.22875793278217316,
+      "learning_rate": 2.2550000000000003e-05,
+      "loss": 2.2778,
+      "step": 1098
+    },
+    {
+      "epoch": 0.12268491869961243,
+      "grad_norm": 0.2337283492088318,
+      "learning_rate": 2.2525000000000002e-05,
+      "loss": 2.2373,
+      "step": 1099
+    },
+    {
+      "epoch": 0.12279655192863846,
+      "grad_norm": 0.22070764005184174,
+      "learning_rate": 2.25e-05,
+      "loss": 2.2863,
+      "step": 1100
+    },
+    {
+      "epoch": 0.1229081851576645,
+      "grad_norm": 0.23479975759983063,
+      "learning_rate": 2.2475e-05,
+      "loss": 2.3971,
+      "step": 1101
+    },
+    {
+      "epoch": 0.12301981838669053,
+      "grad_norm": 0.22529123723506927,
+      "learning_rate": 2.245e-05,
+      "loss": 2.4699,
+      "step": 1102
+    },
+    {
+      "epoch": 0.12313145161571656,
+      "grad_norm": 0.24018734693527222,
+      "learning_rate": 2.2425000000000003e-05,
+      "loss": 2.4539,
+      "step": 1103
+    },
+    {
+      "epoch": 0.1232430848447426,
+      "grad_norm": 0.22801434993743896,
+      "learning_rate": 2.2400000000000002e-05,
+      "loss": 2.3764,
+      "step": 1104
+    },
+    {
+      "epoch": 0.12335471807376863,
+      "grad_norm": 0.32481849193573,
+      "learning_rate": 2.2375000000000002e-05,
+      "loss": 2.4011,
+      "step": 1105
+    },
+    {
+      "epoch": 0.12346635130279467,
+      "grad_norm": 0.22494614124298096,
+      "learning_rate": 2.235e-05,
+      "loss": 2.2996,
+      "step": 1106
+    },
+    {
+      "epoch": 0.1235779845318207,
+      "grad_norm": 0.2304266095161438,
+      "learning_rate": 2.2325e-05,
+      "loss": 2.2708,
+      "step": 1107
+    },
+    {
+      "epoch": 0.12368961776084673,
+      "grad_norm": 0.22332698106765747,
+      "learning_rate": 2.23e-05,
+      "loss": 2.2825,
+      "step": 1108
+    },
+    {
+      "epoch": 0.12380125098987277,
+      "grad_norm": 0.2219666987657547,
+      "learning_rate": 2.2275000000000003e-05,
+      "loss": 2.3805,
+      "step": 1109
+    },
+    {
+      "epoch": 0.1239128842188988,
+      "grad_norm": 0.2652471363544464,
+      "learning_rate": 2.2250000000000002e-05,
+      "loss": 2.2467,
+      "step": 1110
+    },
+    {
+      "epoch": 0.12402451744792484,
+      "grad_norm": 0.2186206877231598,
+      "learning_rate": 2.2225e-05,
+      "loss": 2.2924,
+      "step": 1111
+    },
+    {
+      "epoch": 0.12413615067695087,
+      "grad_norm": 0.23667913675308228,
+      "learning_rate": 2.22e-05,
+      "loss": 2.3545,
+      "step": 1112
+    },
+    {
+      "epoch": 0.12424778390597692,
+      "grad_norm": 0.3729536831378937,
+      "learning_rate": 2.2175e-05,
+      "loss": 2.2641,
+      "step": 1113
+    },
+    {
+      "epoch": 0.12435941713500295,
+      "grad_norm": 0.22672784328460693,
+      "learning_rate": 2.215e-05,
+      "loss": 2.2501,
+      "step": 1114
+    },
+    {
+      "epoch": 0.12447105036402899,
+      "grad_norm": 0.2219839245080948,
+      "learning_rate": 2.2125000000000002e-05,
+      "loss": 2.2016,
+      "step": 1115
+    },
+    {
+      "epoch": 0.12458268359305502,
+      "grad_norm": 0.24350149929523468,
+      "learning_rate": 2.2100000000000002e-05,
+      "loss": 2.4358,
+      "step": 1116
+    },
+    {
+      "epoch": 0.12469431682208106,
+      "grad_norm": 0.2482176274061203,
+      "learning_rate": 2.2075e-05,
+      "loss": 2.2623,
+      "step": 1117
+    },
+    {
+      "epoch": 0.12480595005110709,
+      "grad_norm": 0.23272433876991272,
+      "learning_rate": 2.205e-05,
+      "loss": 2.3408,
+      "step": 1118
+    },
+    {
+      "epoch": 0.12491758328013312,
+      "grad_norm": 0.23357626795768738,
+      "learning_rate": 2.2025e-05,
+      "loss": 2.231,
+      "step": 1119
+    },
+    {
+      "epoch": 0.12502921650915916,
+      "grad_norm": 0.22280560433864594,
+      "learning_rate": 2.2000000000000003e-05,
+      "loss": 2.3045,
+      "step": 1120
+    },
+    {
+      "epoch": 0.1251408497381852,
+      "grad_norm": 0.22206202149391174,
+      "learning_rate": 2.1975000000000002e-05,
+      "loss": 2.3103,
+      "step": 1121
+    },
+    {
+      "epoch": 0.12525248296721123,
+      "grad_norm": 0.31398919224739075,
+      "learning_rate": 2.195e-05,
+      "loss": 2.3665,
+      "step": 1122
+    },
+    {
+      "epoch": 0.12536411619623727,
+      "grad_norm": 0.43255481123924255,
+      "learning_rate": 2.1925e-05,
+      "loss": 2.3153,
+      "step": 1123
+    },
+    {
+      "epoch": 0.1254757494252633,
+      "grad_norm": 0.22663763165473938,
+      "learning_rate": 2.19e-05,
+      "loss": 2.3682,
+      "step": 1124
+    },
+    {
+      "epoch": 0.12558738265428934,
+      "grad_norm": 0.2514352798461914,
+      "learning_rate": 2.1875e-05,
+      "loss": 2.3727,
+      "step": 1125
+    },
+    {
+      "epoch": 0.12569901588331536,
+      "grad_norm": 0.22935128211975098,
+      "learning_rate": 2.1850000000000003e-05,
+      "loss": 2.2161,
+      "step": 1126
+    },
+    {
+      "epoch": 0.1258106491123414,
+      "grad_norm": 0.2275882214307785,
+      "learning_rate": 2.1825000000000002e-05,
+      "loss": 2.3793,
+      "step": 1127
+    },
+    {
+      "epoch": 0.12592228234136743,
+      "grad_norm": 0.2203671634197235,
+      "learning_rate": 2.18e-05,
+      "loss": 2.3371,
+      "step": 1128
+    },
+    {
+      "epoch": 0.12603391557039348,
+      "grad_norm": 0.22310760617256165,
+      "learning_rate": 2.1775e-05,
+      "loss": 2.2956,
+      "step": 1129
+    },
+    {
+      "epoch": 0.1261455487994195,
+      "grad_norm": 0.24273249506950378,
+      "learning_rate": 2.175e-05,
+      "loss": 2.3288,
+      "step": 1130
+    },
+    {
+      "epoch": 0.12625718202844555,
+      "grad_norm": 0.22426392138004303,
+      "learning_rate": 2.1725e-05,
+      "loss": 2.387,
+      "step": 1131
+    },
+    {
+      "epoch": 0.12636881525747157,
+      "grad_norm": 0.2271021008491516,
+      "learning_rate": 2.1700000000000002e-05,
+      "loss": 2.3739,
+      "step": 1132
+    },
+    {
+      "epoch": 0.12648044848649762,
+      "grad_norm": 0.2271476835012436,
+      "learning_rate": 2.1675e-05,
+      "loss": 2.3942,
+      "step": 1133
+    },
+    {
+      "epoch": 0.12659208171552364,
+      "grad_norm": 0.2212740033864975,
+      "learning_rate": 2.165e-05,
+      "loss": 2.3511,
+      "step": 1134
+    },
+    {
+      "epoch": 0.12670371494454968,
+      "grad_norm": 0.23425845801830292,
+      "learning_rate": 2.1625e-05,
+      "loss": 2.3256,
+      "step": 1135
+    },
+    {
+      "epoch": 0.1268153481735757,
+      "grad_norm": 0.22545567154884338,
+      "learning_rate": 2.16e-05,
+      "loss": 2.1847,
+      "step": 1136
+    },
+    {
+      "epoch": 0.12692698140260175,
+      "grad_norm": 0.25626057386398315,
+      "learning_rate": 2.1575e-05,
+      "loss": 2.3058,
+      "step": 1137
+    },
+    {
+      "epoch": 0.1270386146316278,
+      "grad_norm": 0.23703870177268982,
+      "learning_rate": 2.1550000000000002e-05,
+      "loss": 2.3709,
+      "step": 1138
+    },
+    {
+      "epoch": 0.12715024786065382,
+      "grad_norm": 0.2455843836069107,
+      "learning_rate": 2.1525e-05,
+      "loss": 2.3468,
+      "step": 1139
+    },
+    {
+      "epoch": 0.12726188108967987,
+      "grad_norm": 0.22671882808208466,
+      "learning_rate": 2.15e-05,
+      "loss": 2.3554,
+      "step": 1140
+    },
+    {
+      "epoch": 0.1273735143187059,
+      "grad_norm": 0.22661983966827393,
+      "learning_rate": 2.1475e-05,
+      "loss": 2.2919,
+      "step": 1141
+    },
+    {
+      "epoch": 0.12748514754773194,
+      "grad_norm": 0.23259443044662476,
+      "learning_rate": 2.145e-05,
+      "loss": 2.3463,
+      "step": 1142
+    },
+    {
+      "epoch": 0.12759678077675796,
+      "grad_norm": 0.23297019302845,
+      "learning_rate": 2.1425e-05,
+      "loss": 2.3198,
+      "step": 1143
+    },
+    {
+      "epoch": 0.127708414005784,
+      "grad_norm": 0.22478660941123962,
+      "learning_rate": 2.1400000000000002e-05,
+      "loss": 2.2642,
+      "step": 1144
+    },
+    {
+      "epoch": 0.12782004723481002,
+      "grad_norm": 0.23042532801628113,
+      "learning_rate": 2.1375e-05,
+      "loss": 2.3056,
+      "step": 1145
+    },
+    {
+      "epoch": 0.12793168046383607,
+      "grad_norm": 0.2341778576374054,
+      "learning_rate": 2.135e-05,
+      "loss": 2.3774,
+      "step": 1146
+    },
+    {
+      "epoch": 0.1280433136928621,
+      "grad_norm": 0.22689583897590637,
+      "learning_rate": 2.1325e-05,
+      "loss": 2.3493,
+      "step": 1147
+    },
+    {
+      "epoch": 0.12815494692188814,
+      "grad_norm": 0.2260814756155014,
+      "learning_rate": 2.13e-05,
+      "loss": 2.3548,
+      "step": 1148
+    },
+    {
+      "epoch": 0.12826658015091416,
+      "grad_norm": 0.24112841486930847,
+      "learning_rate": 2.1275000000000002e-05,
+      "loss": 2.3739,
+      "step": 1149
+    },
+    {
+      "epoch": 0.1283782133799402,
+      "grad_norm": 0.22553539276123047,
+      "learning_rate": 2.125e-05,
+      "loss": 2.3762,
+      "step": 1150
+    },
+    {
+      "epoch": 0.12848984660896623,
+      "grad_norm": 0.2698231041431427,
+      "learning_rate": 2.1225e-05,
+      "loss": 2.3527,
+      "step": 1151
+    },
+    {
+      "epoch": 0.12860147983799228,
+      "grad_norm": 0.2280593365430832,
+      "learning_rate": 2.12e-05,
+      "loss": 2.3621,
+      "step": 1152
+    },
+    {
+      "epoch": 0.1287131130670183,
+      "grad_norm": 0.23123127222061157,
+      "learning_rate": 2.1175e-05,
+      "loss": 2.4369,
+      "step": 1153
+    },
+    {
+      "epoch": 0.12882474629604435,
+      "grad_norm": 0.2183208465576172,
+      "learning_rate": 2.115e-05,
+      "loss": 2.3615,
+      "step": 1154
+    },
+    {
+      "epoch": 0.1289363795250704,
+      "grad_norm": 0.3457687199115753,
+      "learning_rate": 2.1125000000000002e-05,
+      "loss": 2.3151,
+      "step": 1155
+    },
+    {
+      "epoch": 0.12904801275409641,
+      "grad_norm": 0.24221017956733704,
+      "learning_rate": 2.11e-05,
+      "loss": 2.383,
+      "step": 1156
+    },
+    {
+      "epoch": 0.12915964598312246,
+      "grad_norm": 0.31406766176223755,
+      "learning_rate": 2.1075e-05,
+      "loss": 2.3564,
+      "step": 1157
+    },
+    {
+      "epoch": 0.12927127921214848,
+      "grad_norm": 0.2336696982383728,
+      "learning_rate": 2.105e-05,
+      "loss": 2.3295,
+      "step": 1158
+    },
+    {
+      "epoch": 0.12938291244117453,
+      "grad_norm": 0.2292354553937912,
+      "learning_rate": 2.1025e-05,
+      "loss": 2.3124,
+      "step": 1159
+    },
+    {
+      "epoch": 0.12949454567020055,
+      "grad_norm": 0.22350043058395386,
+      "learning_rate": 2.1e-05,
+      "loss": 2.3077,
+      "step": 1160
+    },
+    {
+      "epoch": 0.1296061788992266,
+      "grad_norm": 0.22331391274929047,
+      "learning_rate": 2.0975e-05,
+      "loss": 2.2535,
+      "step": 1161
+    },
+    {
+      "epoch": 0.12971781212825262,
+      "grad_norm": 0.23815131187438965,
+      "learning_rate": 2.095e-05,
+      "loss": 2.4475,
+      "step": 1162
+    },
+    {
+      "epoch": 0.12982944535727867,
+      "grad_norm": 0.29236698150634766,
+      "learning_rate": 2.0925e-05,
+      "loss": 2.312,
+      "step": 1163
+    },
+    {
+      "epoch": 0.1299410785863047,
+      "grad_norm": 0.23702973127365112,
+      "learning_rate": 2.09e-05,
+      "loss": 2.4403,
+      "step": 1164
+    },
+    {
+      "epoch": 0.13005271181533073,
+      "grad_norm": 0.23101966083049774,
+      "learning_rate": 2.0875e-05,
+      "loss": 2.4587,
+      "step": 1165
+    },
+    {
+      "epoch": 0.13016434504435676,
+      "grad_norm": 0.2322998195886612,
+      "learning_rate": 2.085e-05,
+      "loss": 2.421,
+      "step": 1166
+    },
+    {
+      "epoch": 0.1302759782733828,
+      "grad_norm": 0.236577570438385,
+      "learning_rate": 2.0825e-05,
+      "loss": 2.3525,
+      "step": 1167
+    },
+    {
+      "epoch": 0.13038761150240882,
+      "grad_norm": 0.2353632152080536,
+      "learning_rate": 2.08e-05,
+      "loss": 2.4235,
+      "step": 1168
+    },
+    {
+      "epoch": 0.13049924473143487,
+      "grad_norm": 0.23410721123218536,
+      "learning_rate": 2.0775e-05,
+      "loss": 2.3678,
+      "step": 1169
+    },
+    {
+      "epoch": 0.13061087796046092,
+      "grad_norm": 0.22162692248821259,
+      "learning_rate": 2.075e-05,
+      "loss": 2.4682,
+      "step": 1170
+    },
+    {
+      "epoch": 0.13072251118948694,
+      "grad_norm": 0.2268594205379486,
+      "learning_rate": 2.0725e-05,
+      "loss": 2.3006,
+      "step": 1171
+    },
+    {
+      "epoch": 0.130834144418513,
+      "grad_norm": 0.22843024134635925,
+      "learning_rate": 2.07e-05,
+      "loss": 2.3072,
+      "step": 1172
+    },
+    {
+      "epoch": 0.130945777647539,
+      "grad_norm": 0.2605232298374176,
+      "learning_rate": 2.0675e-05,
+      "loss": 2.3408,
+      "step": 1173
+    },
+    {
+      "epoch": 0.13105741087656506,
+      "grad_norm": 0.2354726940393448,
+      "learning_rate": 2.065e-05,
+      "loss": 2.3971,
+      "step": 1174
+    },
+    {
+      "epoch": 0.13116904410559108,
+      "grad_norm": 0.225637286901474,
+      "learning_rate": 2.0625e-05,
+      "loss": 2.314,
+      "step": 1175
+    },
+    {
+      "epoch": 0.13128067733461712,
+      "grad_norm": 0.22693420946598053,
+      "learning_rate": 2.06e-05,
+      "loss": 2.4122,
+      "step": 1176
+    },
+    {
+      "epoch": 0.13139231056364314,
+      "grad_norm": 0.22487470507621765,
+      "learning_rate": 2.0575e-05,
+      "loss": 2.2592,
+      "step": 1177
+    },
+    {
+      "epoch": 0.1315039437926692,
+      "grad_norm": 0.2332613468170166,
+      "learning_rate": 2.055e-05,
+      "loss": 2.3281,
+      "step": 1178
+    },
+    {
+      "epoch": 0.1316155770216952,
+      "grad_norm": 0.3483419716358185,
+      "learning_rate": 2.0525e-05,
+      "loss": 2.2564,
+      "step": 1179
+    },
+    {
+      "epoch": 0.13172721025072126,
+      "grad_norm": 0.23920704424381256,
+      "learning_rate": 2.05e-05,
+      "loss": 2.3591,
+      "step": 1180
+    },
+    {
+      "epoch": 0.13183884347974728,
+      "grad_norm": 0.2232007533311844,
+      "learning_rate": 2.0475e-05,
+      "loss": 2.4216,
+      "step": 1181
+    },
+    {
+      "epoch": 0.13195047670877333,
+      "grad_norm": 0.23171833157539368,
+      "learning_rate": 2.045e-05,
+      "loss": 2.2861,
+      "step": 1182
+    },
+    {
+      "epoch": 0.13206210993779935,
+      "grad_norm": 0.22283758223056793,
+      "learning_rate": 2.0425e-05,
+      "loss": 2.274,
+      "step": 1183
+    },
+    {
+      "epoch": 0.1321737431668254,
+      "grad_norm": 0.21574443578720093,
+      "learning_rate": 2.04e-05,
+      "loss": 2.4139,
+      "step": 1184
+    },
+    {
+      "epoch": 0.13228537639585142,
+      "grad_norm": 0.23107174038887024,
+      "learning_rate": 2.0375e-05,
+      "loss": 2.2474,
+      "step": 1185
+    },
+    {
+      "epoch": 0.13239700962487747,
+      "grad_norm": 0.2264859676361084,
+      "learning_rate": 2.035e-05,
+      "loss": 2.3655,
+      "step": 1186
+    },
+    {
+      "epoch": 0.1325086428539035,
+      "grad_norm": 0.23183952271938324,
+      "learning_rate": 2.0325e-05,
+      "loss": 2.4179,
+      "step": 1187
+    },
+    {
+      "epoch": 0.13262027608292953,
+      "grad_norm": 0.2878219187259674,
+      "learning_rate": 2.0300000000000002e-05,
+      "loss": 2.3113,
+      "step": 1188
+    },
+    {
+      "epoch": 0.13273190931195558,
+      "grad_norm": 0.2385031133890152,
+      "learning_rate": 2.0275e-05,
+      "loss": 2.3088,
+      "step": 1189
+    },
+    {
+      "epoch": 0.1328435425409816,
+      "grad_norm": 0.2221747487783432,
+      "learning_rate": 2.025e-05,
+      "loss": 2.3643,
+      "step": 1190
+    },
+    {
+      "epoch": 0.13295517577000765,
+      "grad_norm": 0.23417770862579346,
+      "learning_rate": 2.0225000000000004e-05,
+      "loss": 2.293,
+      "step": 1191
+    },
+    {
+      "epoch": 0.13306680899903367,
+      "grad_norm": 0.22443942725658417,
+      "learning_rate": 2.0200000000000003e-05,
+      "loss": 2.1929,
+      "step": 1192
+    },
+    {
+      "epoch": 0.13317844222805972,
+      "grad_norm": 0.22454610466957092,
+      "learning_rate": 2.0175000000000003e-05,
+      "loss": 2.2178,
+      "step": 1193
+    },
+    {
+      "epoch": 0.13329007545708574,
+      "grad_norm": 0.2309190183877945,
+      "learning_rate": 2.0150000000000002e-05,
+      "loss": 2.3352,
+      "step": 1194
+    },
+    {
+      "epoch": 0.1334017086861118,
+      "grad_norm": 0.23848609626293182,
+      "learning_rate": 2.0125e-05,
+      "loss": 2.3348,
+      "step": 1195
+    },
+    {
+      "epoch": 0.1335133419151378,
+      "grad_norm": 0.23410527408123016,
+      "learning_rate": 2.01e-05,
+      "loss": 2.1928,
+      "step": 1196
+    },
+    {
+      "epoch": 0.13362497514416385,
+      "grad_norm": 0.24377335608005524,
+      "learning_rate": 2.0075000000000003e-05,
+      "loss": 2.2982,
+      "step": 1197
+    },
+    {
+      "epoch": 0.13373660837318987,
+      "grad_norm": 0.22369886934757233,
+      "learning_rate": 2.0050000000000003e-05,
+      "loss": 2.2788,
+      "step": 1198
+    },
+    {
+      "epoch": 0.13384824160221592,
+      "grad_norm": 0.22385212779045105,
+      "learning_rate": 2.0025000000000002e-05,
+      "loss": 2.3247,
+      "step": 1199
+    },
+    {
+      "epoch": 0.13395987483124194,
+      "grad_norm": 0.22769370675086975,
+      "learning_rate": 2e-05,
+      "loss": 2.3075,
+      "step": 1200
+    },
+    {
+      "epoch": 0.134071508060268,
+      "grad_norm": 0.2851308286190033,
+      "learning_rate": 1.9975e-05,
+      "loss": 2.317,
+      "step": 1201
+    },
+    {
+      "epoch": 0.134183141289294,
+      "grad_norm": 0.24706991016864777,
+      "learning_rate": 1.995e-05,
+      "loss": 2.2672,
+      "step": 1202
+    },
+    {
+      "epoch": 0.13429477451832006,
+      "grad_norm": 0.21826080977916718,
+      "learning_rate": 1.9925000000000003e-05,
+      "loss": 2.3373,
+      "step": 1203
+    },
+    {
+      "epoch": 0.1344064077473461,
+      "grad_norm": 0.23195642232894897,
+      "learning_rate": 1.9900000000000003e-05,
+      "loss": 2.271,
+      "step": 1204
+    },
+    {
+      "epoch": 0.13451804097637213,
+      "grad_norm": 0.24489794671535492,
+      "learning_rate": 1.9875000000000002e-05,
+      "loss": 2.3,
+      "step": 1205
+    },
+    {
+      "epoch": 0.13462967420539818,
+      "grad_norm": 0.23644982278347015,
+      "learning_rate": 1.985e-05,
+      "loss": 2.2082,
+      "step": 1206
+    },
+    {
+      "epoch": 0.1347413074344242,
+      "grad_norm": 0.22759714722633362,
+      "learning_rate": 1.9825e-05,
+      "loss": 2.3624,
+      "step": 1207
+    },
+    {
+      "epoch": 0.13485294066345024,
+      "grad_norm": 0.23812660574913025,
+      "learning_rate": 1.9800000000000004e-05,
+      "loss": 2.3198,
+      "step": 1208
+    },
+    {
+      "epoch": 0.13496457389247626,
+      "grad_norm": 0.22150012850761414,
+      "learning_rate": 1.9775000000000003e-05,
+      "loss": 2.2949,
+      "step": 1209
+    },
+    {
+      "epoch": 0.1350762071215023,
+      "grad_norm": 0.23681719601154327,
+      "learning_rate": 1.9750000000000002e-05,
+      "loss": 2.4068,
+      "step": 1210
+    },
+    {
+      "epoch": 0.13518784035052833,
+      "grad_norm": 0.22480838000774384,
+      "learning_rate": 1.9725000000000002e-05,
+      "loss": 2.4212,
+      "step": 1211
+    },
+    {
+      "epoch": 0.13529947357955438,
+      "grad_norm": 0.2516225576400757,
+      "learning_rate": 1.97e-05,
+      "loss": 2.3919,
+      "step": 1212
+    },
+    {
+      "epoch": 0.1354111068085804,
+      "grad_norm": 0.22985118627548218,
+      "learning_rate": 1.9675e-05,
+      "loss": 2.3349,
+      "step": 1213
+    },
+    {
+      "epoch": 0.13552274003760645,
+      "grad_norm": 0.24210740625858307,
+      "learning_rate": 1.9650000000000003e-05,
+      "loss": 2.3892,
+      "step": 1214
+    },
+    {
+      "epoch": 0.13563437326663247,
+      "grad_norm": 0.23913182318210602,
+      "learning_rate": 1.9625000000000003e-05,
+      "loss": 2.3997,
+      "step": 1215
+    },
+    {
+      "epoch": 0.13574600649565852,
+      "grad_norm": 0.2336951494216919,
+      "learning_rate": 1.9600000000000002e-05,
+      "loss": 2.2605,
+      "step": 1216
+    },
+    {
+      "epoch": 0.13585763972468454,
+      "grad_norm": 0.24042119085788727,
+      "learning_rate": 1.9575e-05,
+      "loss": 2.28,
+      "step": 1217
+    },
+    {
+      "epoch": 0.13596927295371058,
+      "grad_norm": 0.24132166802883148,
+      "learning_rate": 1.955e-05,
+      "loss": 2.3416,
+      "step": 1218
+    },
+    {
+      "epoch": 0.1360809061827366,
+      "grad_norm": 0.22675567865371704,
+      "learning_rate": 1.9525e-05,
+      "loss": 2.4432,
+      "step": 1219
+    },
+    {
+      "epoch": 0.13619253941176265,
+      "grad_norm": 0.2387012392282486,
+      "learning_rate": 1.9500000000000003e-05,
+      "loss": 2.3219,
+      "step": 1220
+    },
+    {
+      "epoch": 0.1363041726407887,
+      "grad_norm": 0.22851043939590454,
+      "learning_rate": 1.9475000000000002e-05,
+      "loss": 2.3047,
+      "step": 1221
+    },
+    {
+      "epoch": 0.13641580586981472,
+      "grad_norm": 0.229843869805336,
+      "learning_rate": 1.9450000000000002e-05,
+      "loss": 2.3684,
+      "step": 1222
+    },
+    {
+      "epoch": 0.13652743909884077,
+      "grad_norm": 0.22853203117847443,
+      "learning_rate": 1.9425e-05,
+      "loss": 2.3583,
+      "step": 1223
+    },
+    {
+      "epoch": 0.1366390723278668,
+      "grad_norm": 0.22501179575920105,
+      "learning_rate": 1.94e-05,
+      "loss": 2.3816,
+      "step": 1224
+    },
+    {
+      "epoch": 0.13675070555689284,
+      "grad_norm": 0.2398713231086731,
+      "learning_rate": 1.9375e-05,
+      "loss": 2.2891,
+      "step": 1225
+    },
+    {
+      "epoch": 0.13686233878591886,
+      "grad_norm": 0.23787732422351837,
+      "learning_rate": 1.9350000000000003e-05,
+      "loss": 2.3867,
+      "step": 1226
+    },
+    {
+      "epoch": 0.1369739720149449,
+      "grad_norm": 0.2299504280090332,
+      "learning_rate": 1.9325000000000002e-05,
+      "loss": 2.358,
+      "step": 1227
+    },
+    {
+      "epoch": 0.13708560524397093,
+      "grad_norm": 0.23067769408226013,
+      "learning_rate": 1.93e-05,
+      "loss": 2.2818,
+      "step": 1228
+    },
+    {
+      "epoch": 0.13719723847299697,
+      "grad_norm": 0.22578711807727814,
+      "learning_rate": 1.9275e-05,
+      "loss": 2.3077,
+      "step": 1229
+    },
+    {
+      "epoch": 0.137308871702023,
+      "grad_norm": 0.2431231141090393,
+      "learning_rate": 1.925e-05,
+      "loss": 2.448,
+      "step": 1230
+    },
+    {
+      "epoch": 0.13742050493104904,
+      "grad_norm": 0.22373032569885254,
+      "learning_rate": 1.9225e-05,
+      "loss": 2.3161,
+      "step": 1231
+    },
+    {
+      "epoch": 0.13753213816007506,
+      "grad_norm": 0.22712518274784088,
+      "learning_rate": 1.9200000000000003e-05,
+      "loss": 2.327,
+      "step": 1232
+    },
+    {
+      "epoch": 0.1376437713891011,
+      "grad_norm": 0.23568111658096313,
+      "learning_rate": 1.9175000000000002e-05,
+      "loss": 2.2981,
+      "step": 1233
+    },
+    {
+      "epoch": 0.13775540461812713,
+      "grad_norm": 0.4218859076499939,
+      "learning_rate": 1.915e-05,
+      "loss": 2.3714,
+      "step": 1234
+    },
+    {
+      "epoch": 0.13786703784715318,
+      "grad_norm": 0.25230151414871216,
+      "learning_rate": 1.9125e-05,
+      "loss": 2.3094,
+      "step": 1235
+    },
+    {
+      "epoch": 0.13797867107617923,
+      "grad_norm": 0.2286907434463501,
+      "learning_rate": 1.91e-05,
+      "loss": 2.3385,
+      "step": 1236
+    },
+    {
+      "epoch": 0.13809030430520525,
+      "grad_norm": 0.22605471312999725,
+      "learning_rate": 1.9075000000000003e-05,
+      "loss": 2.3362,
+      "step": 1237
+    },
+    {
+      "epoch": 0.1382019375342313,
+      "grad_norm": 0.34180840849876404,
+      "learning_rate": 1.9050000000000002e-05,
+      "loss": 2.409,
+      "step": 1238
+    },
+    {
+      "epoch": 0.13831357076325732,
+      "grad_norm": 0.23643071949481964,
+      "learning_rate": 1.9025e-05,
+      "loss": 2.3804,
+      "step": 1239
+    },
+    {
+      "epoch": 0.13842520399228336,
+      "grad_norm": 0.23849943280220032,
+      "learning_rate": 1.9e-05,
+      "loss": 2.2734,
+      "step": 1240
+    },
+    {
+      "epoch": 0.13853683722130938,
+      "grad_norm": 0.22949104011058807,
+      "learning_rate": 1.8975e-05,
+      "loss": 2.3875,
+      "step": 1241
+    },
+    {
+      "epoch": 0.13864847045033543,
+      "grad_norm": 0.22790773212909698,
+      "learning_rate": 1.895e-05,
+      "loss": 2.2737,
+      "step": 1242
+    },
+    {
+      "epoch": 0.13876010367936145,
+      "grad_norm": 0.23416166007518768,
+      "learning_rate": 1.8925000000000003e-05,
+      "loss": 2.2871,
+      "step": 1243
+    },
+    {
+      "epoch": 0.1388717369083875,
+      "grad_norm": 0.22195836901664734,
+      "learning_rate": 1.8900000000000002e-05,
+      "loss": 2.281,
+      "step": 1244
+    },
+    {
+      "epoch": 0.13898337013741352,
+      "grad_norm": 0.2336248904466629,
+      "learning_rate": 1.8875e-05,
+      "loss": 2.416,
+      "step": 1245
+    },
+    {
+      "epoch": 0.13909500336643957,
+      "grad_norm": 0.23051492869853973,
+      "learning_rate": 1.885e-05,
+      "loss": 2.2719,
+      "step": 1246
+    },
+    {
+      "epoch": 0.1392066365954656,
+      "grad_norm": 0.2252211719751358,
+      "learning_rate": 1.8825e-05,
+      "loss": 2.4248,
+      "step": 1247
+    },
+    {
+      "epoch": 0.13931826982449164,
+      "grad_norm": 0.21714623272418976,
+      "learning_rate": 1.88e-05,
+      "loss": 2.3514,
+      "step": 1248
+    },
+    {
+      "epoch": 0.13942990305351766,
+      "grad_norm": 0.23050609230995178,
+      "learning_rate": 1.8775000000000002e-05,
+      "loss": 2.3583,
+      "step": 1249
+    },
+    {
+      "epoch": 0.1395415362825437,
+      "grad_norm": 0.2242250144481659,
+      "learning_rate": 1.8750000000000002e-05,
+      "loss": 2.2994,
+      "step": 1250
+    },
+    {
+      "epoch": 0.13965316951156972,
+      "grad_norm": 0.23751282691955566,
+      "learning_rate": 1.8725e-05,
+      "loss": 2.2898,
+      "step": 1251
+    },
+    {
+      "epoch": 0.13976480274059577,
+      "grad_norm": 0.240337535738945,
+      "learning_rate": 1.87e-05,
+      "loss": 2.3338,
+      "step": 1252
+    },
+    {
+      "epoch": 0.13987643596962182,
+      "grad_norm": 0.2228308618068695,
+      "learning_rate": 1.8675e-05,
+      "loss": 2.3378,
+      "step": 1253
+    },
+    {
+      "epoch": 0.13998806919864784,
+      "grad_norm": 0.22002586722373962,
+      "learning_rate": 1.865e-05,
+      "loss": 2.3768,
+      "step": 1254
+    },
+    {
+      "epoch": 0.1400997024276739,
+      "grad_norm": 0.22754792869091034,
+      "learning_rate": 1.8625000000000002e-05,
+      "loss": 2.3875,
+      "step": 1255
+    },
+    {
+      "epoch": 0.1402113356566999,
+      "grad_norm": 0.22698433697223663,
+      "learning_rate": 1.86e-05,
+      "loss": 2.3624,
+      "step": 1256
+    },
+    {
+      "epoch": 0.14032296888572596,
+      "grad_norm": 0.2388381063938141,
+      "learning_rate": 1.8575e-05,
+      "loss": 2.3388,
+      "step": 1257
+    },
+    {
+      "epoch": 0.14043460211475198,
+      "grad_norm": 0.2522680461406708,
+      "learning_rate": 1.855e-05,
+      "loss": 2.3738,
+      "step": 1258
+    },
+    {
+      "epoch": 0.14054623534377803,
+      "grad_norm": 0.22983700037002563,
+      "learning_rate": 1.8525e-05,
+      "loss": 2.4546,
+      "step": 1259
+    },
+    {
+      "epoch": 0.14065786857280405,
+      "grad_norm": 0.23299263417720795,
+      "learning_rate": 1.85e-05,
+      "loss": 2.3668,
+      "step": 1260
+    },
+    {
+      "epoch": 0.1407695018018301,
+      "grad_norm": 0.29829832911491394,
+      "learning_rate": 1.8475000000000002e-05,
+      "loss": 2.3765,
+      "step": 1261
+    },
+    {
+      "epoch": 0.14088113503085611,
+      "grad_norm": 0.2792015075683594,
+      "learning_rate": 1.845e-05,
+      "loss": 2.2744,
+      "step": 1262
+    },
+    {
+      "epoch": 0.14099276825988216,
+      "grad_norm": 0.2371360808610916,
+      "learning_rate": 1.8425e-05,
+      "loss": 2.3681,
+      "step": 1263
+    },
+    {
+      "epoch": 0.14110440148890818,
+      "grad_norm": 0.22529280185699463,
+      "learning_rate": 1.84e-05,
+      "loss": 2.2995,
+      "step": 1264
+    },
+    {
+      "epoch": 0.14121603471793423,
+      "grad_norm": 0.23717930912971497,
+      "learning_rate": 1.8375e-05,
+      "loss": 2.3641,
+      "step": 1265
+    },
+    {
+      "epoch": 0.14132766794696025,
+      "grad_norm": 0.23806947469711304,
+      "learning_rate": 1.8350000000000002e-05,
+      "loss": 2.2403,
+      "step": 1266
+    },
+    {
+      "epoch": 0.1414393011759863,
+      "grad_norm": 0.2357507348060608,
+      "learning_rate": 1.8325e-05,
+      "loss": 2.3355,
+      "step": 1267
+    },
+    {
+      "epoch": 0.14155093440501232,
+      "grad_norm": 0.2314206063747406,
+      "learning_rate": 1.83e-05,
+      "loss": 2.3525,
+      "step": 1268
+    },
+    {
+      "epoch": 0.14166256763403837,
+      "grad_norm": 0.23410086333751678,
+      "learning_rate": 1.8275e-05,
+      "loss": 2.3385,
+      "step": 1269
+    },
+    {
+      "epoch": 0.14177420086306441,
+      "grad_norm": 0.23491892218589783,
+      "learning_rate": 1.825e-05,
+      "loss": 2.443,
+      "step": 1270
+    },
+    {
+      "epoch": 0.14188583409209043,
+      "grad_norm": 0.224997416138649,
+      "learning_rate": 1.8225e-05,
+      "loss": 2.3451,
+      "step": 1271
+    },
+    {
+      "epoch": 0.14199746732111648,
+      "grad_norm": 0.2989085614681244,
+      "learning_rate": 1.8200000000000002e-05,
+      "loss": 2.3746,
+      "step": 1272
+    },
+    {
+      "epoch": 0.1421091005501425,
+      "grad_norm": 0.2321501523256302,
+      "learning_rate": 1.8175e-05,
+      "loss": 2.1922,
+      "step": 1273
+    },
+    {
+      "epoch": 0.14222073377916855,
+      "grad_norm": 0.2198849767446518,
+      "learning_rate": 1.815e-05,
+      "loss": 2.2675,
+      "step": 1274
+    },
+    {
+      "epoch": 0.14233236700819457,
+      "grad_norm": 0.22828496992588043,
+      "learning_rate": 1.8125e-05,
+      "loss": 2.3126,
+      "step": 1275
+    },
+    {
+      "epoch": 0.14244400023722062,
+      "grad_norm": 0.23503446578979492,
+      "learning_rate": 1.81e-05,
+      "loss": 2.325,
+      "step": 1276
+    },
+    {
+      "epoch": 0.14255563346624664,
+      "grad_norm": 0.22161200642585754,
+      "learning_rate": 1.8075e-05,
+      "loss": 2.4076,
+      "step": 1277
+    },
+    {
+      "epoch": 0.1426672666952727,
+      "grad_norm": 0.22467480599880219,
+      "learning_rate": 1.805e-05,
+      "loss": 2.4028,
+      "step": 1278
+    },
+    {
+      "epoch": 0.1427788999242987,
+      "grad_norm": 0.23092903196811676,
+      "learning_rate": 1.8025e-05,
+      "loss": 2.3841,
+      "step": 1279
+    },
+    {
+      "epoch": 0.14289053315332476,
+      "grad_norm": 0.23917998373508453,
+      "learning_rate": 1.8e-05,
+      "loss": 2.3114,
+      "step": 1280
+    },
+    {
+      "epoch": 0.14300216638235078,
+      "grad_norm": 0.24001942574977875,
+      "learning_rate": 1.7975e-05,
+      "loss": 2.2419,
+      "step": 1281
+    },
+    {
+      "epoch": 0.14311379961137682,
+      "grad_norm": 0.22311803698539734,
+      "learning_rate": 1.795e-05,
+      "loss": 2.35,
+      "step": 1282
+    },
+    {
+      "epoch": 0.14322543284040284,
+      "grad_norm": 0.23798157274723053,
+      "learning_rate": 1.7925e-05,
+      "loss": 2.404,
+      "step": 1283
+    },
+    {
+      "epoch": 0.1433370660694289,
+      "grad_norm": 0.22350966930389404,
+      "learning_rate": 1.79e-05,
+      "loss": 2.4122,
+      "step": 1284
+    },
+    {
+      "epoch": 0.1434486992984549,
+      "grad_norm": 0.2272355854511261,
+      "learning_rate": 1.7875e-05,
+      "loss": 2.4368,
+      "step": 1285
+    },
+    {
+      "epoch": 0.14356033252748096,
+      "grad_norm": 0.21722903847694397,
+      "learning_rate": 1.785e-05,
+      "loss": 2.3516,
+      "step": 1286
+    },
+    {
+      "epoch": 0.143671965756507,
+      "grad_norm": 0.23155128955841064,
+      "learning_rate": 1.7825e-05,
+      "loss": 2.3527,
+      "step": 1287
+    },
+    {
+      "epoch": 0.14378359898553303,
+      "grad_norm": 0.22640927135944366,
+      "learning_rate": 1.78e-05,
+      "loss": 2.4748,
+      "step": 1288
+    },
+    {
+      "epoch": 0.14389523221455908,
+      "grad_norm": 0.23734673857688904,
+      "learning_rate": 1.7775e-05,
+      "loss": 2.2844,
+      "step": 1289
+    },
+    {
+      "epoch": 0.1440068654435851,
+      "grad_norm": 0.2328357696533203,
+      "learning_rate": 1.775e-05,
+      "loss": 2.3389,
+      "step": 1290
+    },
+    {
+      "epoch": 0.14411849867261114,
+      "grad_norm": 0.21968986093997955,
+      "learning_rate": 1.7725e-05,
+      "loss": 2.3395,
+      "step": 1291
+    },
+    {
+      "epoch": 0.14423013190163717,
+      "grad_norm": 0.21870984137058258,
+      "learning_rate": 1.77e-05,
+      "loss": 2.3882,
+      "step": 1292
+    },
+    {
+      "epoch": 0.1443417651306632,
+      "grad_norm": 0.23083637654781342,
+      "learning_rate": 1.7675e-05,
+      "loss": 2.3653,
+      "step": 1293
+    },
+    {
+      "epoch": 0.14445339835968923,
+      "grad_norm": 0.23204779624938965,
+      "learning_rate": 1.765e-05,
+      "loss": 2.4653,
+      "step": 1294
+    },
+    {
+      "epoch": 0.14456503158871528,
+      "grad_norm": 0.23749560117721558,
+      "learning_rate": 1.7625e-05,
+      "loss": 2.3105,
+      "step": 1295
+    },
+    {
+      "epoch": 0.1446766648177413,
+      "grad_norm": 0.23485244810581207,
+      "learning_rate": 1.76e-05,
+      "loss": 2.3721,
+      "step": 1296
+    },
+    {
+      "epoch": 0.14478829804676735,
+      "grad_norm": 0.24895891547203064,
+      "learning_rate": 1.7575e-05,
+      "loss": 2.3752,
+      "step": 1297
+    },
+    {
+      "epoch": 0.14489993127579337,
+      "grad_norm": 0.24030578136444092,
+      "learning_rate": 1.755e-05,
+      "loss": 2.3025,
+      "step": 1298
+    },
+    {
+      "epoch": 0.14501156450481942,
+      "grad_norm": 0.23427051305770874,
+      "learning_rate": 1.7525e-05,
+      "loss": 2.431,
+      "step": 1299
+    },
+    {
+      "epoch": 0.14512319773384544,
+      "grad_norm": 0.29638248682022095,
+      "learning_rate": 1.75e-05,
+      "loss": 2.3806,
+      "step": 1300
+    },
+    {
+      "epoch": 0.1452348309628715,
+      "grad_norm": 0.24296362698078156,
+      "learning_rate": 1.7475e-05,
+      "loss": 2.3058,
+      "step": 1301
+    },
+    {
+      "epoch": 0.14534646419189753,
+      "grad_norm": 0.21712899208068848,
+      "learning_rate": 1.745e-05,
+      "loss": 2.2896,
+      "step": 1302
+    },
+    {
+      "epoch": 0.14545809742092355,
+      "grad_norm": 0.22417020797729492,
+      "learning_rate": 1.7425e-05,
+      "loss": 2.3631,
+      "step": 1303
+    },
+    {
+      "epoch": 0.1455697306499496,
+      "grad_norm": 0.22780650854110718,
+      "learning_rate": 1.74e-05,
+      "loss": 2.441,
+      "step": 1304
+    },
+    {
+      "epoch": 0.14568136387897562,
+      "grad_norm": 0.22729064524173737,
+      "learning_rate": 1.7375e-05,
+      "loss": 2.4841,
+      "step": 1305
+    },
+    {
+      "epoch": 0.14579299710800167,
+      "grad_norm": 0.23002713918685913,
+      "learning_rate": 1.7349999999999998e-05,
+      "loss": 2.2594,
+      "step": 1306
+    },
+    {
+      "epoch": 0.1459046303370277,
+      "grad_norm": 0.22959677875041962,
+      "learning_rate": 1.7325e-05,
+      "loss": 2.3681,
+      "step": 1307
+    },
+    {
+      "epoch": 0.14601626356605374,
+      "grad_norm": 0.23864157497882843,
+      "learning_rate": 1.73e-05,
+      "loss": 2.413,
+      "step": 1308
+    },
+    {
+      "epoch": 0.14612789679507976,
+      "grad_norm": 0.2260824590921402,
+      "learning_rate": 1.7275e-05,
+      "loss": 2.4128,
+      "step": 1309
+    },
+    {
+      "epoch": 0.1462395300241058,
+      "grad_norm": 0.23661834001541138,
+      "learning_rate": 1.725e-05,
+      "loss": 2.3452,
+      "step": 1310
+    },
+    {
+      "epoch": 0.14635116325313183,
+      "grad_norm": 0.24539943039417267,
+      "learning_rate": 1.7225e-05,
+      "loss": 2.452,
+      "step": 1311
+    },
+    {
+      "epoch": 0.14646279648215788,
+      "grad_norm": 0.24564427137374878,
+      "learning_rate": 1.7199999999999998e-05,
+      "loss": 2.2986,
+      "step": 1312
+    },
+    {
+      "epoch": 0.1465744297111839,
+      "grad_norm": 0.22026251256465912,
+      "learning_rate": 1.7175e-05,
+      "loss": 2.4334,
+      "step": 1313
+    },
+    {
+      "epoch": 0.14668606294020994,
+      "grad_norm": 0.22923065721988678,
+      "learning_rate": 1.7150000000000004e-05,
+      "loss": 2.3331,
+      "step": 1314
+    },
+    {
+      "epoch": 0.14679769616923596,
+      "grad_norm": 0.26323196291923523,
+      "learning_rate": 1.7125000000000003e-05,
+      "loss": 2.2923,
+      "step": 1315
+    },
+    {
+      "epoch": 0.146909329398262,
+      "grad_norm": 0.2375718653202057,
+      "learning_rate": 1.7100000000000002e-05,
+      "loss": 2.4087,
+      "step": 1316
+    },
+    {
+      "epoch": 0.14702096262728803,
+      "grad_norm": 0.22378355264663696,
+      "learning_rate": 1.7075e-05,
+      "loss": 2.3135,
+      "step": 1317
+    },
+    {
+      "epoch": 0.14713259585631408,
+      "grad_norm": 0.2237062305212021,
+      "learning_rate": 1.705e-05,
+      "loss": 2.2467,
+      "step": 1318
+    },
+    {
+      "epoch": 0.14724422908534013,
+      "grad_norm": 0.24400672316551208,
+      "learning_rate": 1.7025e-05,
+      "loss": 2.1746,
+      "step": 1319
+    },
+    {
+      "epoch": 0.14735586231436615,
+      "grad_norm": 0.22511416673660278,
+      "learning_rate": 1.7000000000000003e-05,
+      "loss": 2.4436,
+      "step": 1320
+    },
+    {
+      "epoch": 0.1474674955433922,
+      "grad_norm": 0.23048102855682373,
+      "learning_rate": 1.6975000000000003e-05,
+      "loss": 2.3356,
+      "step": 1321
+    },
+    {
+      "epoch": 0.14757912877241822,
+      "grad_norm": 0.2408643364906311,
+      "learning_rate": 1.6950000000000002e-05,
+      "loss": 2.2649,
+      "step": 1322
+    },
+    {
+      "epoch": 0.14769076200144426,
+      "grad_norm": 0.2274709939956665,
+      "learning_rate": 1.6925e-05,
+      "loss": 2.3392,
+      "step": 1323
+    },
+    {
+      "epoch": 0.14780239523047028,
+      "grad_norm": 0.22548739612102509,
+      "learning_rate": 1.69e-05,
+      "loss": 2.2553,
+      "step": 1324
+    },
+    {
+      "epoch": 0.14791402845949633,
+      "grad_norm": 0.24427109956741333,
+      "learning_rate": 1.6875000000000004e-05,
+      "loss": 2.3549,
+      "step": 1325
+    },
+    {
+      "epoch": 0.14802566168852235,
+      "grad_norm": 0.2285667061805725,
+      "learning_rate": 1.6850000000000003e-05,
+      "loss": 2.4703,
+      "step": 1326
+    },
+    {
+      "epoch": 0.1481372949175484,
+      "grad_norm": 0.22185759246349335,
+      "learning_rate": 1.6825000000000002e-05,
+      "loss": 2.4652,
+      "step": 1327
+    },
+    {
+      "epoch": 0.14824892814657442,
+      "grad_norm": 0.23354068398475647,
+      "learning_rate": 1.6800000000000002e-05,
+      "loss": 2.2939,
+      "step": 1328
+    },
+    {
+      "epoch": 0.14836056137560047,
+      "grad_norm": 0.22763966023921967,
+      "learning_rate": 1.6775e-05,
+      "loss": 2.3484,
+      "step": 1329
+    },
+    {
+      "epoch": 0.1484721946046265,
+      "grad_norm": 0.23366783559322357,
+      "learning_rate": 1.675e-05,
+      "loss": 2.2578,
+      "step": 1330
+    },
+    {
+      "epoch": 0.14858382783365254,
+      "grad_norm": 0.22202405333518982,
+      "learning_rate": 1.6725000000000003e-05,
+      "loss": 2.4142,
+      "step": 1331
+    },
+    {
+      "epoch": 0.14869546106267856,
+      "grad_norm": 0.499910831451416,
+      "learning_rate": 1.6700000000000003e-05,
+      "loss": 2.304,
+      "step": 1332
+    },
+    {
+      "epoch": 0.1488070942917046,
+      "grad_norm": 0.22948618233203888,
+      "learning_rate": 1.6675000000000002e-05,
+      "loss": 2.3483,
+      "step": 1333
+    },
+    {
+      "epoch": 0.14891872752073063,
+      "grad_norm": 0.2281409054994583,
+      "learning_rate": 1.665e-05,
+      "loss": 2.3,
+      "step": 1334
+    },
+    {
+      "epoch": 0.14903036074975667,
+      "grad_norm": 0.22423812747001648,
+      "learning_rate": 1.6625e-05,
+      "loss": 2.3268,
+      "step": 1335
+    },
+    {
+      "epoch": 0.14914199397878272,
+      "grad_norm": 0.22844356298446655,
+      "learning_rate": 1.66e-05,
+      "loss": 2.2694,
+      "step": 1336
+    },
+    {
+      "epoch": 0.14925362720780874,
+      "grad_norm": 0.23927628993988037,
+      "learning_rate": 1.6575000000000003e-05,
+      "loss": 2.3171,
+      "step": 1337
+    },
+    {
+      "epoch": 0.1493652604368348,
+      "grad_norm": 0.22286780178546906,
+      "learning_rate": 1.6550000000000002e-05,
+      "loss": 2.2984,
+      "step": 1338
+    },
+    {
+      "epoch": 0.1494768936658608,
+      "grad_norm": 0.2628430128097534,
+      "learning_rate": 1.6525000000000002e-05,
+      "loss": 2.4076,
+      "step": 1339
+    },
+    {
+      "epoch": 0.14958852689488686,
+      "grad_norm": 0.227376326918602,
+      "learning_rate": 1.65e-05,
+      "loss": 2.2738,
+      "step": 1340
+    },
+    {
+      "epoch": 0.14970016012391288,
+      "grad_norm": 0.27796196937561035,
+      "learning_rate": 1.6475e-05,
+      "loss": 2.2671,
+      "step": 1341
+    },
+    {
+      "epoch": 0.14981179335293893,
+      "grad_norm": 0.2336723506450653,
+      "learning_rate": 1.645e-05,
+      "loss": 2.3649,
+      "step": 1342
+    },
+    {
+      "epoch": 0.14992342658196495,
+      "grad_norm": 0.2579742670059204,
+      "learning_rate": 1.6425000000000003e-05,
+      "loss": 2.4033,
+      "step": 1343
+    },
+    {
+      "epoch": 0.150035059810991,
+      "grad_norm": 0.23226085305213928,
+      "learning_rate": 1.6400000000000002e-05,
+      "loss": 2.3702,
+      "step": 1344
+    },
+    {
+      "epoch": 0.15014669304001702,
+      "grad_norm": 0.22210608422756195,
+      "learning_rate": 1.6375e-05,
+      "loss": 2.3124,
+      "step": 1345
+    },
+    {
+      "epoch": 0.15025832626904306,
+      "grad_norm": 0.2297833412885666,
+      "learning_rate": 1.635e-05,
+      "loss": 2.311,
+      "step": 1346
+    },
+    {
+      "epoch": 0.15036995949806908,
+      "grad_norm": 0.480037659406662,
+      "learning_rate": 1.6325e-05,
+      "loss": 2.3665,
+      "step": 1347
+    },
+    {
+      "epoch": 0.15048159272709513,
+      "grad_norm": 0.22692376375198364,
+      "learning_rate": 1.63e-05,
+      "loss": 2.4278,
+      "step": 1348
+    },
+    {
+      "epoch": 0.15059322595612115,
+      "grad_norm": 0.23020422458648682,
+      "learning_rate": 1.6275000000000003e-05,
+      "loss": 2.3182,
+      "step": 1349
+    },
+    {
+      "epoch": 0.1507048591851472,
+      "grad_norm": 0.2347160279750824,
+      "learning_rate": 1.6250000000000002e-05,
+      "loss": 2.4645,
+      "step": 1350
+    },
+    {
+      "epoch": 0.15081649241417325,
+      "grad_norm": 0.24892841279506683,
+      "learning_rate": 1.6225e-05,
+      "loss": 2.3542,
+      "step": 1351
+    },
+    {
+      "epoch": 0.15092812564319927,
+      "grad_norm": 0.2282174974679947,
+      "learning_rate": 1.62e-05,
+      "loss": 2.4327,
+      "step": 1352
+    },
+    {
+      "epoch": 0.15103975887222532,
+      "grad_norm": 0.22344060242176056,
+      "learning_rate": 1.6175e-05,
+      "loss": 2.3244,
+      "step": 1353
+    },
+    {
+      "epoch": 0.15115139210125134,
+      "grad_norm": 0.2454865276813507,
+      "learning_rate": 1.6150000000000003e-05,
+      "loss": 2.4319,
+      "step": 1354
+    },
+    {
+      "epoch": 0.15126302533027738,
+      "grad_norm": 0.238905668258667,
+      "learning_rate": 1.6125000000000002e-05,
+      "loss": 2.3431,
+      "step": 1355
+    },
+    {
+      "epoch": 0.1513746585593034,
+      "grad_norm": 0.22759667038917542,
+      "learning_rate": 1.6100000000000002e-05,
+      "loss": 2.4397,
+      "step": 1356
+    },
+    {
+      "epoch": 0.15148629178832945,
+      "grad_norm": 0.22996599972248077,
+      "learning_rate": 1.6075e-05,
+      "loss": 2.4378,
+      "step": 1357
+    },
+    {
+      "epoch": 0.15159792501735547,
+      "grad_norm": 0.23668399453163147,
+      "learning_rate": 1.605e-05,
+      "loss": 2.3791,
+      "step": 1358
+    },
+    {
+      "epoch": 0.15170955824638152,
+      "grad_norm": 0.23384137451648712,
+      "learning_rate": 1.6025e-05,
+      "loss": 2.2818,
+      "step": 1359
+    },
+    {
+      "epoch": 0.15182119147540754,
+      "grad_norm": 0.23646210134029388,
+      "learning_rate": 1.6000000000000003e-05,
+      "loss": 2.306,
+      "step": 1360
+    },
+    {
+      "epoch": 0.1519328247044336,
+      "grad_norm": 0.23297393321990967,
+      "learning_rate": 1.5975000000000002e-05,
+      "loss": 2.3061,
+      "step": 1361
+    },
+    {
+      "epoch": 0.1520444579334596,
+      "grad_norm": 0.23053903877735138,
+      "learning_rate": 1.595e-05,
+      "loss": 2.2829,
+      "step": 1362
+    },
+    {
+      "epoch": 0.15215609116248566,
+      "grad_norm": 0.22388771176338196,
+      "learning_rate": 1.5925e-05,
+      "loss": 2.2914,
+      "step": 1363
+    },
+    {
+      "epoch": 0.15226772439151168,
+      "grad_norm": 0.22923487424850464,
+      "learning_rate": 1.59e-05,
+      "loss": 2.2915,
+      "step": 1364
+    },
+    {
+      "epoch": 0.15237935762053773,
+      "grad_norm": 0.22146253287792206,
+      "learning_rate": 1.5875e-05,
+      "loss": 2.4338,
+      "step": 1365
+    },
+    {
+      "epoch": 0.15249099084956375,
+      "grad_norm": 0.22123506665229797,
+      "learning_rate": 1.5850000000000002e-05,
+      "loss": 2.3959,
+      "step": 1366
+    },
+    {
+      "epoch": 0.1526026240785898,
+      "grad_norm": 0.31164827942848206,
+      "learning_rate": 1.5825000000000002e-05,
+      "loss": 2.3052,
+      "step": 1367
+    },
+    {
+      "epoch": 0.15271425730761584,
+      "grad_norm": 0.23311270773410797,
+      "learning_rate": 1.58e-05,
+      "loss": 2.343,
+      "step": 1368
+    },
+    {
+      "epoch": 0.15282589053664186,
+      "grad_norm": 0.25240063667297363,
+      "learning_rate": 1.5775e-05,
+      "loss": 2.3322,
+      "step": 1369
+    },
+    {
+      "epoch": 0.1529375237656679,
+      "grad_norm": 0.2153395712375641,
+      "learning_rate": 1.575e-05,
+      "loss": 2.3915,
+      "step": 1370
+    },
+    {
+      "epoch": 0.15304915699469393,
+      "grad_norm": 0.23929548263549805,
+      "learning_rate": 1.5725e-05,
+      "loss": 2.382,
+      "step": 1371
+    },
+    {
+      "epoch": 0.15316079022371998,
+      "grad_norm": 0.2313188761472702,
+      "learning_rate": 1.5700000000000002e-05,
+      "loss": 2.3818,
+      "step": 1372
+    },
+    {
+      "epoch": 0.153272423452746,
+      "grad_norm": 0.22789841890335083,
+      "learning_rate": 1.5675e-05,
+      "loss": 2.3657,
+      "step": 1373
+    },
+    {
+      "epoch": 0.15338405668177205,
+      "grad_norm": 0.2191040813922882,
+      "learning_rate": 1.565e-05,
+      "loss": 2.3667,
+      "step": 1374
+    },
+    {
+      "epoch": 0.15349568991079807,
+      "grad_norm": 0.22538387775421143,
+      "learning_rate": 1.5625e-05,
+      "loss": 2.2652,
+      "step": 1375
+    },
+    {
+      "epoch": 0.15360732313982411,
+      "grad_norm": 0.23191332817077637,
+      "learning_rate": 1.56e-05,
+      "loss": 2.4353,
+      "step": 1376
+    },
+    {
+      "epoch": 0.15371895636885013,
+      "grad_norm": 0.23539386689662933,
+      "learning_rate": 1.5575e-05,
+      "loss": 2.3348,
+      "step": 1377
+    },
+    {
+      "epoch": 0.15383058959787618,
+      "grad_norm": 0.2552396357059479,
+      "learning_rate": 1.5550000000000002e-05,
+      "loss": 2.2693,
+      "step": 1378
+    },
+    {
+      "epoch": 0.1539422228269022,
+      "grad_norm": 0.2361995130777359,
+      "learning_rate": 1.5525e-05,
+      "loss": 2.3506,
+      "step": 1379
+    },
+    {
+      "epoch": 0.15405385605592825,
+      "grad_norm": 0.2353193461894989,
+      "learning_rate": 1.55e-05,
+      "loss": 2.2858,
+      "step": 1380
+    },
+    {
+      "epoch": 0.15416548928495427,
+      "grad_norm": 0.23434515297412872,
+      "learning_rate": 1.5475e-05,
+      "loss": 2.3942,
+      "step": 1381
+    },
+    {
+      "epoch": 0.15427712251398032,
+      "grad_norm": 0.25974181294441223,
+      "learning_rate": 1.545e-05,
+      "loss": 2.3877,
+      "step": 1382
+    },
+    {
+      "epoch": 0.15438875574300634,
+      "grad_norm": 0.2274274230003357,
+      "learning_rate": 1.5425000000000002e-05,
+      "loss": 2.4084,
+      "step": 1383
+    },
+    {
+      "epoch": 0.1545003889720324,
+      "grad_norm": 0.23268136382102966,
+      "learning_rate": 1.54e-05,
+      "loss": 2.3343,
+      "step": 1384
+    },
+    {
+      "epoch": 0.15461202220105844,
+      "grad_norm": 0.23196761310100555,
+      "learning_rate": 1.5375e-05,
+      "loss": 2.3164,
+      "step": 1385
+    },
+    {
+      "epoch": 0.15472365543008446,
+      "grad_norm": 0.22755514085292816,
+      "learning_rate": 1.535e-05,
+      "loss": 2.3012,
+      "step": 1386
+    },
+    {
+      "epoch": 0.1548352886591105,
+      "grad_norm": 0.21987079083919525,
+      "learning_rate": 1.5325e-05,
+      "loss": 2.3306,
+      "step": 1387
+    },
+    {
+      "epoch": 0.15494692188813652,
+      "grad_norm": 0.22733251750469208,
+      "learning_rate": 1.53e-05,
+      "loss": 2.2869,
+      "step": 1388
+    },
+    {
+      "epoch": 0.15505855511716257,
+      "grad_norm": 0.23861852288246155,
+      "learning_rate": 1.5275000000000002e-05,
+      "loss": 2.2962,
+      "step": 1389
+    },
+    {
+      "epoch": 0.1551701883461886,
+      "grad_norm": 0.2321462631225586,
+      "learning_rate": 1.525e-05,
+      "loss": 2.2596,
+      "step": 1390
+    },
+    {
+      "epoch": 0.15528182157521464,
+      "grad_norm": 0.2401476502418518,
+      "learning_rate": 1.5225e-05,
+      "loss": 2.3632,
+      "step": 1391
+    },
+    {
+      "epoch": 0.15539345480424066,
+      "grad_norm": 0.2222481220960617,
+      "learning_rate": 1.52e-05,
+      "loss": 2.4276,
+      "step": 1392
+    },
+    {
+      "epoch": 0.1555050880332667,
+      "grad_norm": 0.23532716929912567,
+      "learning_rate": 1.5175e-05,
+      "loss": 2.3198,
+      "step": 1393
+    },
+    {
+      "epoch": 0.15561672126229273,
+      "grad_norm": 0.24865378439426422,
+      "learning_rate": 1.515e-05,
+      "loss": 2.3288,
+      "step": 1394
+    },
+    {
+      "epoch": 0.15572835449131878,
+      "grad_norm": 0.23691876232624054,
+      "learning_rate": 1.5125e-05,
+      "loss": 2.3948,
+      "step": 1395
+    },
+    {
+      "epoch": 0.1558399877203448,
+      "grad_norm": 0.21998995542526245,
+      "learning_rate": 1.51e-05,
+      "loss": 2.2551,
+      "step": 1396
+    },
+    {
+      "epoch": 0.15595162094937084,
+      "grad_norm": 0.23590844869613647,
+      "learning_rate": 1.5075e-05,
+      "loss": 2.327,
+      "step": 1397
+    },
+    {
+      "epoch": 0.15606325417839687,
+      "grad_norm": 0.227497398853302,
+      "learning_rate": 1.505e-05,
+      "loss": 2.2747,
+      "step": 1398
+    },
+    {
+      "epoch": 0.1561748874074229,
+      "grad_norm": 0.23721876740455627,
+      "learning_rate": 1.5025000000000001e-05,
+      "loss": 2.2471,
+      "step": 1399
+    },
+    {
+      "epoch": 0.15628652063644893,
+      "grad_norm": 0.23734678328037262,
+      "learning_rate": 1.5e-05,
+      "loss": 2.3562,
+      "step": 1400
+    },
+    {
+      "epoch": 0.15639815386547498,
+      "grad_norm": 0.23354454338550568,
+      "learning_rate": 1.4975e-05,
+      "loss": 2.2572,
+      "step": 1401
+    },
+    {
+      "epoch": 0.15650978709450103,
+      "grad_norm": 0.2284322828054428,
+      "learning_rate": 1.4950000000000001e-05,
+      "loss": 2.3481,
+      "step": 1402
+    },
+    {
+      "epoch": 0.15662142032352705,
+      "grad_norm": 0.24697981774806976,
+      "learning_rate": 1.4925e-05,
+      "loss": 2.4658,
+      "step": 1403
+    },
+    {
+      "epoch": 0.1567330535525531,
+      "grad_norm": 0.23272185027599335,
+      "learning_rate": 1.49e-05,
+      "loss": 2.4396,
+      "step": 1404
+    },
+    {
+      "epoch": 0.15684468678157912,
+      "grad_norm": 0.22883062064647675,
+      "learning_rate": 1.4875e-05,
+      "loss": 2.4235,
+      "step": 1405
+    },
+    {
+      "epoch": 0.15695632001060517,
+      "grad_norm": 0.23410239815711975,
+      "learning_rate": 1.485e-05,
+      "loss": 2.3252,
+      "step": 1406
+    },
+    {
+      "epoch": 0.1570679532396312,
+      "grad_norm": 0.22218795120716095,
+      "learning_rate": 1.4825e-05,
+      "loss": 2.3215,
+      "step": 1407
+    },
+    {
+      "epoch": 0.15717958646865723,
+      "grad_norm": 0.22799868881702423,
+      "learning_rate": 1.48e-05,
+      "loss": 2.2427,
+      "step": 1408
+    },
+    {
+      "epoch": 0.15729121969768325,
+      "grad_norm": 0.2517700791358948,
+      "learning_rate": 1.4775e-05,
+      "loss": 2.3081,
+      "step": 1409
+    },
+    {
+      "epoch": 0.1574028529267093,
+      "grad_norm": 0.23836027085781097,
+      "learning_rate": 1.475e-05,
+      "loss": 2.3097,
+      "step": 1410
+    },
+    {
+      "epoch": 0.15751448615573532,
+      "grad_norm": 0.22234973311424255,
+      "learning_rate": 1.4725e-05,
+      "loss": 2.4121,
+      "step": 1411
+    },
+    {
+      "epoch": 0.15762611938476137,
+      "grad_norm": 0.35192346572875977,
+      "learning_rate": 1.47e-05,
+      "loss": 2.3617,
+      "step": 1412
+    },
+    {
+      "epoch": 0.1577377526137874,
+      "grad_norm": 0.2238391488790512,
+      "learning_rate": 1.4675e-05,
+      "loss": 2.3145,
+      "step": 1413
+    },
+    {
+      "epoch": 0.15784938584281344,
+      "grad_norm": 0.2211901992559433,
+      "learning_rate": 1.465e-05,
+      "loss": 2.3678,
+      "step": 1414
+    },
+    {
+      "epoch": 0.15796101907183946,
+      "grad_norm": 0.23561525344848633,
+      "learning_rate": 1.4625e-05,
+      "loss": 2.3846,
+      "step": 1415
+    },
+    {
+      "epoch": 0.1580726523008655,
+      "grad_norm": 0.23191802203655243,
+      "learning_rate": 1.4599999999999999e-05,
+      "loss": 2.2361,
+      "step": 1416
+    },
+    {
+      "epoch": 0.15818428552989156,
+      "grad_norm": 0.22749824821949005,
+      "learning_rate": 1.4575e-05,
+      "loss": 2.2809,
+      "step": 1417
+    },
+    {
+      "epoch": 0.15829591875891758,
+      "grad_norm": 0.22720323503017426,
+      "learning_rate": 1.455e-05,
+      "loss": 2.4112,
+      "step": 1418
+    },
+    {
+      "epoch": 0.15840755198794362,
+      "grad_norm": 0.22353364527225494,
+      "learning_rate": 1.4524999999999999e-05,
+      "loss": 2.3453,
+      "step": 1419
+    },
+    {
+      "epoch": 0.15851918521696964,
+      "grad_norm": 0.22859126329421997,
+      "learning_rate": 1.45e-05,
+      "loss": 2.3357,
+      "step": 1420
+    },
+    {
+      "epoch": 0.1586308184459957,
+      "grad_norm": 0.22939516603946686,
+      "learning_rate": 1.4475e-05,
+      "loss": 2.3544,
+      "step": 1421
+    },
+    {
+      "epoch": 0.1587424516750217,
+      "grad_norm": 0.3586462736129761,
+      "learning_rate": 1.4449999999999999e-05,
+      "loss": 2.3628,
+      "step": 1422
+    },
+    {
+      "epoch": 0.15885408490404776,
+      "grad_norm": 0.24023115634918213,
+      "learning_rate": 1.4425e-05,
+      "loss": 2.2797,
+      "step": 1423
+    },
+    {
+      "epoch": 0.15896571813307378,
+      "grad_norm": 0.22737418115139008,
+      "learning_rate": 1.44e-05,
+      "loss": 2.3599,
+      "step": 1424
+    },
+    {
+      "epoch": 0.15907735136209983,
+      "grad_norm": 0.29583752155303955,
+      "learning_rate": 1.4374999999999999e-05,
+      "loss": 2.3229,
+      "step": 1425
+    },
+    {
+      "epoch": 0.15918898459112585,
+      "grad_norm": 0.22859321534633636,
+      "learning_rate": 1.435e-05,
+      "loss": 2.256,
+      "step": 1426
+    },
+    {
+      "epoch": 0.1593006178201519,
+      "grad_norm": 0.2283511608839035,
+      "learning_rate": 1.4325e-05,
+      "loss": 2.3383,
+      "step": 1427
+    },
+    {
+      "epoch": 0.15941225104917792,
+      "grad_norm": 0.24180850386619568,
+      "learning_rate": 1.43e-05,
+      "loss": 2.3953,
+      "step": 1428
+    },
+    {
+      "epoch": 0.15952388427820396,
+      "grad_norm": 0.22382569313049316,
+      "learning_rate": 1.4275e-05,
+      "loss": 2.2684,
+      "step": 1429
+    },
+    {
+      "epoch": 0.15963551750722998,
+      "grad_norm": 0.23606614768505096,
+      "learning_rate": 1.4249999999999999e-05,
+      "loss": 2.4133,
+      "step": 1430
+    },
+    {
+      "epoch": 0.15974715073625603,
+      "grad_norm": 0.23414404690265656,
+      "learning_rate": 1.4225e-05,
+      "loss": 2.2983,
+      "step": 1431
+    },
+    {
+      "epoch": 0.15985878396528205,
+      "grad_norm": 0.2279421091079712,
+      "learning_rate": 1.42e-05,
+      "loss": 2.3842,
+      "step": 1432
+    },
+    {
+      "epoch": 0.1599704171943081,
+      "grad_norm": 0.22593237459659576,
+      "learning_rate": 1.4174999999999999e-05,
+      "loss": 2.3737,
+      "step": 1433
+    },
+    {
+      "epoch": 0.16008205042333415,
+      "grad_norm": 0.2245030701160431,
+      "learning_rate": 1.415e-05,
+      "loss": 2.2731,
+      "step": 1434
+    },
+    {
+      "epoch": 0.16019368365236017,
+      "grad_norm": 0.26113003492355347,
+      "learning_rate": 1.4125e-05,
+      "loss": 2.3804,
+      "step": 1435
+    },
+    {
+      "epoch": 0.16030531688138622,
+      "grad_norm": 0.2301686853170395,
+      "learning_rate": 1.4099999999999999e-05,
+      "loss": 2.2143,
+      "step": 1436
+    },
+    {
+      "epoch": 0.16041695011041224,
+      "grad_norm": 0.2274000197649002,
+      "learning_rate": 1.4075e-05,
+      "loss": 2.4118,
+      "step": 1437
+    },
+    {
+      "epoch": 0.16052858333943829,
+      "grad_norm": 0.23563620448112488,
+      "learning_rate": 1.4050000000000003e-05,
+      "loss": 2.3734,
+      "step": 1438
+    },
+    {
+      "epoch": 0.1606402165684643,
+      "grad_norm": 0.2320801466703415,
+      "learning_rate": 1.4025000000000002e-05,
+      "loss": 2.144,
+      "step": 1439
+    },
+    {
+      "epoch": 0.16075184979749035,
+      "grad_norm": 0.2227548360824585,
+      "learning_rate": 1.4000000000000001e-05,
+      "loss": 2.3084,
+      "step": 1440
+    },
+    {
+      "epoch": 0.16086348302651637,
+      "grad_norm": 0.22315476834774017,
+      "learning_rate": 1.3975000000000003e-05,
+      "loss": 2.2599,
+      "step": 1441
+    },
+    {
+      "epoch": 0.16097511625554242,
+      "grad_norm": 0.2336408942937851,
+      "learning_rate": 1.3950000000000002e-05,
+      "loss": 2.4105,
+      "step": 1442
+    },
+    {
+      "epoch": 0.16108674948456844,
+      "grad_norm": 0.22350838780403137,
+      "learning_rate": 1.3925000000000001e-05,
+      "loss": 2.2656,
+      "step": 1443
+    },
+    {
+      "epoch": 0.1611983827135945,
+      "grad_norm": 0.2283545583486557,
+      "learning_rate": 1.3900000000000002e-05,
+      "loss": 2.3147,
+      "step": 1444
+    },
+    {
+      "epoch": 0.1613100159426205,
+      "grad_norm": 0.22985070943832397,
+      "learning_rate": 1.3875000000000002e-05,
+      "loss": 2.2738,
+      "step": 1445
+    },
+    {
+      "epoch": 0.16142164917164656,
+      "grad_norm": 0.22797125577926636,
+      "learning_rate": 1.3850000000000001e-05,
+      "loss": 2.3832,
+      "step": 1446
+    },
+    {
+      "epoch": 0.16153328240067258,
+      "grad_norm": 0.22750802338123322,
+      "learning_rate": 1.3825000000000002e-05,
+      "loss": 2.3116,
+      "step": 1447
+    },
+    {
+      "epoch": 0.16164491562969863,
+      "grad_norm": 0.23097005486488342,
+      "learning_rate": 1.3800000000000002e-05,
+      "loss": 2.391,
+      "step": 1448
+    },
+    {
+      "epoch": 0.16175654885872465,
+      "grad_norm": 0.23328077793121338,
+      "learning_rate": 1.3775000000000001e-05,
+      "loss": 2.3447,
+      "step": 1449
+    },
+    {
+      "epoch": 0.1618681820877507,
+      "grad_norm": 0.2308788150548935,
+      "learning_rate": 1.3750000000000002e-05,
+      "loss": 2.438,
+      "step": 1450
+    },
+    {
+      "epoch": 0.16197981531677674,
+      "grad_norm": 0.22981509566307068,
+      "learning_rate": 1.3725000000000002e-05,
+      "loss": 2.3295,
+      "step": 1451
+    },
+    {
+      "epoch": 0.16209144854580276,
+      "grad_norm": 0.23261301219463348,
+      "learning_rate": 1.3700000000000001e-05,
+      "loss": 2.3438,
+      "step": 1452
+    },
+    {
+      "epoch": 0.1622030817748288,
+      "grad_norm": 0.22711966931819916,
+      "learning_rate": 1.3675000000000002e-05,
+      "loss": 2.3795,
+      "step": 1453
+    },
+    {
+      "epoch": 0.16231471500385483,
+      "grad_norm": 0.22391866147518158,
+      "learning_rate": 1.3650000000000001e-05,
+      "loss": 2.2794,
+      "step": 1454
+    },
+    {
+      "epoch": 0.16242634823288088,
+      "grad_norm": 0.22070038318634033,
+      "learning_rate": 1.3625e-05,
+      "loss": 2.3572,
+      "step": 1455
+    },
+    {
+      "epoch": 0.1625379814619069,
+      "grad_norm": 0.2527678310871124,
+      "learning_rate": 1.3600000000000002e-05,
+      "loss": 2.4289,
+      "step": 1456
+    },
+    {
+      "epoch": 0.16264961469093295,
+      "grad_norm": 0.22040753066539764,
+      "learning_rate": 1.3575000000000001e-05,
+      "loss": 2.3787,
+      "step": 1457
+    },
+    {
+      "epoch": 0.16276124791995897,
+      "grad_norm": 0.2231033891439438,
+      "learning_rate": 1.3550000000000002e-05,
+      "loss": 2.3683,
+      "step": 1458
+    },
+    {
+      "epoch": 0.16287288114898502,
+      "grad_norm": 0.22828775644302368,
+      "learning_rate": 1.3525000000000002e-05,
+      "loss": 2.3088,
+      "step": 1459
+    },
+    {
+      "epoch": 0.16298451437801104,
+      "grad_norm": 0.22570709884166718,
+      "learning_rate": 1.3500000000000001e-05,
+      "loss": 2.307,
+      "step": 1460
+    },
+    {
+      "epoch": 0.16309614760703708,
+      "grad_norm": 0.8901817798614502,
+      "learning_rate": 1.3475000000000002e-05,
+      "loss": 2.3083,
+      "step": 1461
+    },
+    {
+      "epoch": 0.1632077808360631,
+      "grad_norm": 0.31848806142807007,
+      "learning_rate": 1.3450000000000002e-05,
+      "loss": 2.416,
+      "step": 1462
+    },
+    {
+      "epoch": 0.16331941406508915,
+      "grad_norm": 0.7419360280036926,
+      "learning_rate": 1.3425000000000001e-05,
+      "loss": 2.3243,
+      "step": 1463
+    },
+    {
+      "epoch": 0.16343104729411517,
+      "grad_norm": 0.22873902320861816,
+      "learning_rate": 1.3400000000000002e-05,
+      "loss": 2.344,
+      "step": 1464
+    },
+    {
+      "epoch": 0.16354268052314122,
+      "grad_norm": 0.23168662190437317,
+      "learning_rate": 1.3375000000000002e-05,
+      "loss": 2.3418,
+      "step": 1465
+    },
+    {
+      "epoch": 0.16365431375216724,
+      "grad_norm": 0.2376638650894165,
+      "learning_rate": 1.3350000000000001e-05,
+      "loss": 2.2826,
+      "step": 1466
+    },
+    {
+      "epoch": 0.1637659469811933,
+      "grad_norm": 0.23183491826057434,
+      "learning_rate": 1.3325000000000002e-05,
+      "loss": 2.3392,
+      "step": 1467
+    },
+    {
+      "epoch": 0.16387758021021934,
+      "grad_norm": 0.23273466527462006,
+      "learning_rate": 1.3300000000000001e-05,
+      "loss": 2.3427,
+      "step": 1468
+    },
+    {
+      "epoch": 0.16398921343924536,
+      "grad_norm": 0.2245101034641266,
+      "learning_rate": 1.3275e-05,
+      "loss": 2.3461,
+      "step": 1469
+    },
+    {
+      "epoch": 0.1641008466682714,
+      "grad_norm": 0.2254142016172409,
+      "learning_rate": 1.3250000000000002e-05,
+      "loss": 2.4757,
+      "step": 1470
+    },
+    {
+      "epoch": 0.16421247989729743,
+      "grad_norm": 0.26567506790161133,
+      "learning_rate": 1.3225000000000001e-05,
+      "loss": 2.4493,
+      "step": 1471
+    },
+    {
+      "epoch": 0.16432411312632347,
+      "grad_norm": 0.22109505534172058,
+      "learning_rate": 1.32e-05,
+      "loss": 2.2728,
+      "step": 1472
+    },
+    {
+      "epoch": 0.1644357463553495,
+      "grad_norm": 0.2308911383152008,
+      "learning_rate": 1.3175000000000002e-05,
+      "loss": 2.2651,
+      "step": 1473
+    },
+    {
+      "epoch": 0.16454737958437554,
+      "grad_norm": 0.3326926529407501,
+      "learning_rate": 1.3150000000000001e-05,
+      "loss": 2.3822,
+      "step": 1474
+    },
+    {
+      "epoch": 0.16465901281340156,
+      "grad_norm": 0.2321234494447708,
+      "learning_rate": 1.3125e-05,
+      "loss": 2.3652,
+      "step": 1475
+    },
+    {
+      "epoch": 0.1647706460424276,
+      "grad_norm": 0.22293105721473694,
+      "learning_rate": 1.3100000000000002e-05,
+      "loss": 2.3627,
+      "step": 1476
+    },
+    {
+      "epoch": 0.16488227927145363,
+      "grad_norm": 0.22420254349708557,
+      "learning_rate": 1.3075000000000001e-05,
+      "loss": 2.3223,
+      "step": 1477
+    },
+    {
+      "epoch": 0.16499391250047968,
+      "grad_norm": 0.24831721186637878,
+      "learning_rate": 1.305e-05,
+      "loss": 2.255,
+      "step": 1478
+    },
+    {
+      "epoch": 0.1651055457295057,
+      "grad_norm": 0.23158615827560425,
+      "learning_rate": 1.3025000000000002e-05,
+      "loss": 2.3876,
+      "step": 1479
+    },
+    {
+      "epoch": 0.16521717895853175,
+      "grad_norm": 0.2287922203540802,
+      "learning_rate": 1.3000000000000001e-05,
+      "loss": 2.3559,
+      "step": 1480
+    },
+    {
+      "epoch": 0.16532881218755777,
+      "grad_norm": 0.26463231444358826,
+      "learning_rate": 1.2975e-05,
+      "loss": 2.375,
+      "step": 1481
+    },
+    {
+      "epoch": 0.16544044541658381,
+      "grad_norm": 0.23984935879707336,
+      "learning_rate": 1.2950000000000001e-05,
+      "loss": 2.3024,
+      "step": 1482
+    },
+    {
+      "epoch": 0.16555207864560986,
+      "grad_norm": 0.3325958251953125,
+      "learning_rate": 1.2925e-05,
+      "loss": 2.3871,
+      "step": 1483
+    },
+    {
+      "epoch": 0.16566371187463588,
+      "grad_norm": 0.23081143200397491,
+      "learning_rate": 1.29e-05,
+      "loss": 2.4202,
+      "step": 1484
+    },
+    {
+      "epoch": 0.16577534510366193,
+      "grad_norm": 0.2344684600830078,
+      "learning_rate": 1.2875000000000001e-05,
+      "loss": 2.2737,
+      "step": 1485
+    },
+    {
+      "epoch": 0.16588697833268795,
+      "grad_norm": 0.22795481979846954,
+      "learning_rate": 1.285e-05,
+      "loss": 2.3237,
+      "step": 1486
+    },
+    {
+      "epoch": 0.165998611561714,
+      "grad_norm": 0.2752438187599182,
+      "learning_rate": 1.2825000000000002e-05,
+      "loss": 2.3495,
+      "step": 1487
+    },
+    {
+      "epoch": 0.16611024479074002,
+      "grad_norm": 0.22877170145511627,
+      "learning_rate": 1.2800000000000001e-05,
+      "loss": 2.3932,
+      "step": 1488
+    },
+    {
+      "epoch": 0.16622187801976607,
+      "grad_norm": 0.23439621925354004,
+      "learning_rate": 1.2775e-05,
+      "loss": 2.3022,
+      "step": 1489
+    },
+    {
+      "epoch": 0.1663335112487921,
+      "grad_norm": 0.2311062514781952,
+      "learning_rate": 1.2750000000000002e-05,
+      "loss": 2.2674,
+      "step": 1490
+    },
+    {
+      "epoch": 0.16644514447781814,
+      "grad_norm": 0.2476395070552826,
+      "learning_rate": 1.2725000000000001e-05,
+      "loss": 2.41,
+      "step": 1491
+    },
+    {
+      "epoch": 0.16655677770684416,
+      "grad_norm": 0.22706589102745056,
+      "learning_rate": 1.27e-05,
+      "loss": 2.2607,
+      "step": 1492
+    },
+    {
+      "epoch": 0.1666684109358702,
+      "grad_norm": 0.22713389992713928,
+      "learning_rate": 1.2675000000000001e-05,
+      "loss": 2.3115,
+      "step": 1493
+    },
+    {
+      "epoch": 0.16678004416489622,
+      "grad_norm": 0.2343800812959671,
+      "learning_rate": 1.2650000000000001e-05,
+      "loss": 2.3025,
+      "step": 1494
+    },
+    {
+      "epoch": 0.16689167739392227,
+      "grad_norm": 0.23538857698440552,
+      "learning_rate": 1.2625e-05,
+      "loss": 2.2622,
+      "step": 1495
+    },
+    {
+      "epoch": 0.1670033106229483,
+      "grad_norm": 0.2369028478860855,
+      "learning_rate": 1.2600000000000001e-05,
+      "loss": 2.4365,
+      "step": 1496
+    },
+    {
+      "epoch": 0.16711494385197434,
+      "grad_norm": 0.2249869555234909,
+      "learning_rate": 1.2575e-05,
+      "loss": 2.3179,
+      "step": 1497
+    },
+    {
+      "epoch": 0.16722657708100036,
+      "grad_norm": 0.22788971662521362,
+      "learning_rate": 1.255e-05,
+      "loss": 2.4045,
+      "step": 1498
+    },
+    {
+      "epoch": 0.1673382103100264,
+      "grad_norm": 0.24183280766010284,
+      "learning_rate": 1.2525000000000001e-05,
+      "loss": 2.2165,
+      "step": 1499
+    },
+    {
+      "epoch": 0.16744984353905246,
+      "grad_norm": 0.2287655472755432,
+      "learning_rate": 1.25e-05,
+      "loss": 2.4086,
+      "step": 1500
+    },
+    {
+      "epoch": 0.16756147676807848,
+      "grad_norm": 0.22445043921470642,
+      "learning_rate": 1.2475e-05,
+      "loss": 2.3954,
+      "step": 1501
+    },
+    {
+      "epoch": 0.16767310999710452,
+      "grad_norm": 0.23435929417610168,
+      "learning_rate": 1.2450000000000001e-05,
+      "loss": 2.335,
+      "step": 1502
+    },
+    {
+      "epoch": 0.16778474322613054,
+      "grad_norm": 0.24711214005947113,
+      "learning_rate": 1.2425e-05,
+      "loss": 2.339,
+      "step": 1503
+    },
+    {
+      "epoch": 0.1678963764551566,
+      "grad_norm": 0.2275834083557129,
+      "learning_rate": 1.24e-05,
+      "loss": 2.3166,
+      "step": 1504
+    },
+    {
+      "epoch": 0.1680080096841826,
+      "grad_norm": 0.23408271372318268,
+      "learning_rate": 1.2375000000000001e-05,
+      "loss": 2.3536,
+      "step": 1505
+    },
+    {
+      "epoch": 0.16811964291320866,
+      "grad_norm": 0.24109122157096863,
+      "learning_rate": 1.235e-05,
+      "loss": 2.244,
+      "step": 1506
+    },
+    {
+      "epoch": 0.16823127614223468,
+      "grad_norm": 0.24764837324619293,
+      "learning_rate": 1.2325e-05,
+      "loss": 2.3222,
+      "step": 1507
+    },
+    {
+      "epoch": 0.16834290937126073,
+      "grad_norm": 0.2633460462093353,
+      "learning_rate": 1.23e-05,
+      "loss": 2.2406,
+      "step": 1508
+    },
+    {
+      "epoch": 0.16845454260028675,
+      "grad_norm": 0.23464356362819672,
+      "learning_rate": 1.2275e-05,
+      "loss": 2.4122,
+      "step": 1509
+    },
+    {
+      "epoch": 0.1685661758293128,
+      "grad_norm": 0.23036916553974152,
+      "learning_rate": 1.225e-05,
+      "loss": 2.3748,
+      "step": 1510
+    },
+    {
+      "epoch": 0.16867780905833882,
+      "grad_norm": 0.22803008556365967,
+      "learning_rate": 1.2225e-05,
+      "loss": 2.3297,
+      "step": 1511
+    },
+    {
+      "epoch": 0.16878944228736487,
+      "grad_norm": 0.22997453808784485,
+      "learning_rate": 1.22e-05,
+      "loss": 2.4096,
+      "step": 1512
+    },
+    {
+      "epoch": 0.16890107551639089,
+      "grad_norm": 0.2285720407962799,
+      "learning_rate": 1.2175e-05,
+      "loss": 2.3791,
+      "step": 1513
+    },
+    {
+      "epoch": 0.16901270874541693,
+      "grad_norm": 0.23072819411754608,
+      "learning_rate": 1.215e-05,
+      "loss": 2.2956,
+      "step": 1514
+    },
+    {
+      "epoch": 0.16912434197444295,
+      "grad_norm": 0.2504909336566925,
+      "learning_rate": 1.2125e-05,
+      "loss": 2.3782,
+      "step": 1515
+    },
+    {
+      "epoch": 0.169235975203469,
+      "grad_norm": 0.2371397614479065,
+      "learning_rate": 1.2100000000000001e-05,
+      "loss": 2.2162,
+      "step": 1516
+    },
+    {
+      "epoch": 0.16934760843249505,
+      "grad_norm": 0.3986074924468994,
+      "learning_rate": 1.2075e-05,
+      "loss": 2.4737,
+      "step": 1517
+    },
+    {
+      "epoch": 0.16945924166152107,
+      "grad_norm": 0.24999715387821198,
+      "learning_rate": 1.205e-05,
+      "loss": 2.2256,
+      "step": 1518
+    },
+    {
+      "epoch": 0.16957087489054712,
+      "grad_norm": 0.23626287281513214,
+      "learning_rate": 1.2025000000000001e-05,
+      "loss": 2.331,
+      "step": 1519
+    },
+    {
+      "epoch": 0.16968250811957314,
+      "grad_norm": 0.24345801770687103,
+      "learning_rate": 1.2e-05,
+      "loss": 2.3845,
+      "step": 1520
+    },
+    {
+      "epoch": 0.1697941413485992,
+      "grad_norm": 0.22939099371433258,
+      "learning_rate": 1.1975e-05,
+      "loss": 2.2987,
+      "step": 1521
+    },
+    {
+      "epoch": 0.1699057745776252,
+      "grad_norm": 0.22851888835430145,
+      "learning_rate": 1.195e-05,
+      "loss": 2.2962,
+      "step": 1522
+    },
+    {
+      "epoch": 0.17001740780665126,
+      "grad_norm": 0.23316575586795807,
+      "learning_rate": 1.1925e-05,
+      "loss": 2.2429,
+      "step": 1523
+    },
+    {
+      "epoch": 0.17012904103567728,
+      "grad_norm": 0.23692427575588226,
+      "learning_rate": 1.19e-05,
+      "loss": 2.2889,
+      "step": 1524
+    },
+    {
+      "epoch": 0.17024067426470332,
+      "grad_norm": 0.2287677824497223,
+      "learning_rate": 1.1875e-05,
+      "loss": 2.3004,
+      "step": 1525
+    },
+    {
+      "epoch": 0.17035230749372934,
+      "grad_norm": 0.23153142631053925,
+      "learning_rate": 1.185e-05,
+      "loss": 2.3339,
+      "step": 1526
+    },
+    {
+      "epoch": 0.1704639407227554,
+      "grad_norm": 0.23223859071731567,
+      "learning_rate": 1.1825e-05,
+      "loss": 2.3562,
+      "step": 1527
+    },
+    {
+      "epoch": 0.1705755739517814,
+      "grad_norm": 0.2501643896102905,
+      "learning_rate": 1.18e-05,
+      "loss": 2.3524,
+      "step": 1528
+    },
+    {
+      "epoch": 0.17068720718080746,
+      "grad_norm": 0.227057546377182,
+      "learning_rate": 1.1775e-05,
+      "loss": 2.3482,
+      "step": 1529
+    },
+    {
+      "epoch": 0.17079884040983348,
+      "grad_norm": 0.23041006922721863,
+      "learning_rate": 1.175e-05,
+      "loss": 2.2926,
+      "step": 1530
+    },
+    {
+      "epoch": 0.17091047363885953,
+      "grad_norm": 0.23476162552833557,
+      "learning_rate": 1.1725e-05,
+      "loss": 2.2397,
+      "step": 1531
+    },
+    {
+      "epoch": 0.17102210686788555,
+      "grad_norm": 0.23537229001522064,
+      "learning_rate": 1.1700000000000001e-05,
+      "loss": 2.321,
+      "step": 1532
+    },
+    {
+      "epoch": 0.1711337400969116,
+      "grad_norm": 0.22815583646297455,
+      "learning_rate": 1.1675000000000001e-05,
+      "loss": 2.3891,
+      "step": 1533
+    },
+    {
+      "epoch": 0.17124537332593764,
+      "grad_norm": 0.23210576176643372,
+      "learning_rate": 1.1650000000000002e-05,
+      "loss": 2.2983,
+      "step": 1534
+    },
+    {
+      "epoch": 0.17135700655496366,
+      "grad_norm": 0.2980906665325165,
+      "learning_rate": 1.1625000000000001e-05,
+      "loss": 2.3548,
+      "step": 1535
+    },
+    {
+      "epoch": 0.1714686397839897,
+      "grad_norm": 0.2320065051317215,
+      "learning_rate": 1.16e-05,
+      "loss": 2.3705,
+      "step": 1536
+    },
+    {
+      "epoch": 0.17158027301301573,
+      "grad_norm": 0.5941006541252136,
+      "learning_rate": 1.1575000000000002e-05,
+      "loss": 2.303,
+      "step": 1537
+    },
+    {
+      "epoch": 0.17169190624204178,
+      "grad_norm": 0.23008465766906738,
+      "learning_rate": 1.1550000000000001e-05,
+      "loss": 2.2684,
+      "step": 1538
+    },
+    {
+      "epoch": 0.1718035394710678,
+      "grad_norm": 0.2237699329853058,
+      "learning_rate": 1.1525e-05,
+      "loss": 2.3254,
+      "step": 1539
+    },
+    {
+      "epoch": 0.17191517270009385,
+      "grad_norm": 0.2662789225578308,
+      "learning_rate": 1.1500000000000002e-05,
+      "loss": 2.3753,
+      "step": 1540
+    },
+    {
+      "epoch": 0.17202680592911987,
+      "grad_norm": 0.4782852232456207,
+      "learning_rate": 1.1475000000000001e-05,
+      "loss": 2.2175,
+      "step": 1541
+    },
+    {
+      "epoch": 0.17213843915814592,
+      "grad_norm": 0.23662321269512177,
+      "learning_rate": 1.145e-05,
+      "loss": 2.3435,
+      "step": 1542
+    },
+    {
+      "epoch": 0.17225007238717194,
+      "grad_norm": 0.25570255517959595,
+      "learning_rate": 1.1425000000000002e-05,
+      "loss": 2.4225,
+      "step": 1543
+    },
+    {
+      "epoch": 0.17236170561619799,
+      "grad_norm": 1.1005254983901978,
+      "learning_rate": 1.1400000000000001e-05,
+      "loss": 2.339,
+      "step": 1544
+    },
+    {
+      "epoch": 0.172473338845224,
+      "grad_norm": 0.23017080128192902,
+      "learning_rate": 1.1375e-05,
+      "loss": 2.3182,
+      "step": 1545
+    },
+    {
+      "epoch": 0.17258497207425005,
+      "grad_norm": 0.2425118386745453,
+      "learning_rate": 1.1350000000000001e-05,
+      "loss": 2.2469,
+      "step": 1546
+    },
+    {
+      "epoch": 0.17269660530327607,
+      "grad_norm": 0.31308507919311523,
+      "learning_rate": 1.1325e-05,
+      "loss": 2.4381,
+      "step": 1547
+    },
+    {
+      "epoch": 0.17280823853230212,
+      "grad_norm": 0.22850260138511658,
+      "learning_rate": 1.13e-05,
+      "loss": 2.401,
+      "step": 1548
+    },
+    {
+      "epoch": 0.17291987176132817,
+      "grad_norm": 0.23523540794849396,
+      "learning_rate": 1.1275000000000001e-05,
+      "loss": 2.2428,
+      "step": 1549
+    },
+    {
+      "epoch": 0.1730315049903542,
+      "grad_norm": 0.2316836565732956,
+      "learning_rate": 1.125e-05,
+      "loss": 2.2699,
+      "step": 1550
+    },
+    {
+      "epoch": 0.17314313821938024,
+      "grad_norm": 0.23121745884418488,
+      "learning_rate": 1.1225e-05,
+      "loss": 2.3717,
+      "step": 1551
+    },
+    {
+      "epoch": 0.17325477144840626,
+      "grad_norm": 0.2281220257282257,
+      "learning_rate": 1.1200000000000001e-05,
+      "loss": 2.3741,
+      "step": 1552
+    },
+    {
+      "epoch": 0.1733664046774323,
+      "grad_norm": 0.2314029186964035,
+      "learning_rate": 1.1175e-05,
+      "loss": 2.3432,
+      "step": 1553
+    },
+    {
+      "epoch": 0.17347803790645833,
+      "grad_norm": 0.229678675532341,
+      "learning_rate": 1.115e-05,
+      "loss": 2.4274,
+      "step": 1554
+    },
+    {
+      "epoch": 0.17358967113548437,
+      "grad_norm": 0.23651176691055298,
+      "learning_rate": 1.1125000000000001e-05,
+      "loss": 2.47,
+      "step": 1555
+    },
+    {
+      "epoch": 0.1737013043645104,
+      "grad_norm": 0.25404486060142517,
+      "learning_rate": 1.11e-05,
+      "loss": 2.2866,
+      "step": 1556
+    },
+    {
+      "epoch": 0.17381293759353644,
+      "grad_norm": 0.2315002977848053,
+      "learning_rate": 1.1075e-05,
+      "loss": 2.323,
+      "step": 1557
+    },
+    {
+      "epoch": 0.17392457082256246,
+      "grad_norm": 0.22683122754096985,
+      "learning_rate": 1.1050000000000001e-05,
+      "loss": 2.3863,
+      "step": 1558
+    },
+    {
+      "epoch": 0.1740362040515885,
+      "grad_norm": 0.22364884614944458,
+      "learning_rate": 1.1025e-05,
+      "loss": 2.4092,
+      "step": 1559
+    },
+    {
+      "epoch": 0.17414783728061453,
+      "grad_norm": 0.22706075012683868,
+      "learning_rate": 1.1000000000000001e-05,
+      "loss": 2.2995,
+      "step": 1560
+    },
+    {
+      "epoch": 0.17425947050964058,
+      "grad_norm": 0.22615200281143188,
+      "learning_rate": 1.0975e-05,
+      "loss": 2.274,
+      "step": 1561
+    },
+    {
+      "epoch": 0.1743711037386666,
+      "grad_norm": 0.23156067728996277,
+      "learning_rate": 1.095e-05,
+      "loss": 2.3346,
+      "step": 1562
+    },
+    {
+      "epoch": 0.17448273696769265,
+      "grad_norm": 0.22800372540950775,
+      "learning_rate": 1.0925000000000001e-05,
+      "loss": 2.3311,
+      "step": 1563
+    },
+    {
+      "epoch": 0.17459437019671867,
+      "grad_norm": 0.2314596325159073,
+      "learning_rate": 1.09e-05,
+      "loss": 2.4173,
+      "step": 1564
+    },
+    {
+      "epoch": 0.17470600342574472,
+      "grad_norm": 0.22724230587482452,
+      "learning_rate": 1.0875e-05,
+      "loss": 2.2718,
+      "step": 1565
+    },
+    {
+      "epoch": 0.17481763665477076,
+      "grad_norm": 0.22000856697559357,
+      "learning_rate": 1.0850000000000001e-05,
+      "loss": 2.2426,
+      "step": 1566
+    },
+    {
+      "epoch": 0.17492926988379678,
+      "grad_norm": 0.22086068987846375,
+      "learning_rate": 1.0825e-05,
+      "loss": 2.343,
+      "step": 1567
+    },
+    {
+      "epoch": 0.17504090311282283,
+      "grad_norm": 0.23039323091506958,
+      "learning_rate": 1.08e-05,
+      "loss": 2.3363,
+      "step": 1568
+    },
+    {
+      "epoch": 0.17515253634184885,
+      "grad_norm": 0.23125073313713074,
+      "learning_rate": 1.0775000000000001e-05,
+      "loss": 2.3756,
+      "step": 1569
+    },
+    {
+      "epoch": 0.1752641695708749,
+      "grad_norm": 0.2302304059267044,
+      "learning_rate": 1.075e-05,
+      "loss": 2.292,
+      "step": 1570
+    },
+    {
+      "epoch": 0.17537580279990092,
+      "grad_norm": 0.2361575961112976,
+      "learning_rate": 1.0725e-05,
+      "loss": 2.3275,
+      "step": 1571
+    },
+    {
+      "epoch": 0.17548743602892697,
+      "grad_norm": 0.2350195348262787,
+      "learning_rate": 1.0700000000000001e-05,
+      "loss": 2.3313,
+      "step": 1572
+    },
+    {
+      "epoch": 0.175599069257953,
+      "grad_norm": 0.2335433065891266,
+      "learning_rate": 1.0675e-05,
+      "loss": 2.4255,
+      "step": 1573
+    },
+    {
+      "epoch": 0.17571070248697904,
+      "grad_norm": 0.27532052993774414,
+      "learning_rate": 1.065e-05,
+      "loss": 2.4282,
+      "step": 1574
+    },
+    {
+      "epoch": 0.17582233571600506,
+      "grad_norm": 0.23674359917640686,
+      "learning_rate": 1.0625e-05,
+      "loss": 2.3127,
+      "step": 1575
+    },
+    {
+      "epoch": 0.1759339689450311,
+      "grad_norm": 0.22703874111175537,
+      "learning_rate": 1.06e-05,
+      "loss": 2.4281,
+      "step": 1576
+    },
+    {
+      "epoch": 0.17604560217405713,
+      "grad_norm": 0.2311123013496399,
+      "learning_rate": 1.0575e-05,
+      "loss": 2.3588,
+      "step": 1577
+    },
+    {
+      "epoch": 0.17615723540308317,
+      "grad_norm": 0.22461971640586853,
+      "learning_rate": 1.055e-05,
+      "loss": 2.3376,
+      "step": 1578
+    },
+    {
+      "epoch": 0.1762688686321092,
+      "grad_norm": 0.2341393530368805,
+      "learning_rate": 1.0525e-05,
+      "loss": 2.3273,
+      "step": 1579
+    },
+    {
+      "epoch": 0.17638050186113524,
+      "grad_norm": 0.24818100035190582,
+      "learning_rate": 1.05e-05,
+      "loss": 2.4611,
+      "step": 1580
+    },
+    {
+      "epoch": 0.17649213509016126,
+      "grad_norm": 0.23257790505886078,
+      "learning_rate": 1.0475e-05,
+      "loss": 2.2885,
+      "step": 1581
+    },
+    {
+      "epoch": 0.1766037683191873,
+      "grad_norm": 0.23357973992824554,
+      "learning_rate": 1.045e-05,
+      "loss": 2.2287,
+      "step": 1582
+    },
+    {
+      "epoch": 0.17671540154821336,
+      "grad_norm": 0.22697918117046356,
+      "learning_rate": 1.0425e-05,
+      "loss": 2.2872,
+      "step": 1583
+    },
+    {
+      "epoch": 0.17682703477723938,
+      "grad_norm": 0.2292308658361435,
+      "learning_rate": 1.04e-05,
+      "loss": 2.332,
+      "step": 1584
+    },
+    {
+      "epoch": 0.17693866800626543,
+      "grad_norm": 0.235504150390625,
+      "learning_rate": 1.0375e-05,
+      "loss": 2.3714,
+      "step": 1585
+    },
+    {
+      "epoch": 0.17705030123529145,
+      "grad_norm": 0.22994711995124817,
+      "learning_rate": 1.035e-05,
+      "loss": 2.3468,
+      "step": 1586
+    },
+    {
+      "epoch": 0.1771619344643175,
+      "grad_norm": 0.23801718652248383,
+      "learning_rate": 1.0325e-05,
+      "loss": 2.3851,
+      "step": 1587
+    },
+    {
+      "epoch": 0.17727356769334351,
+      "grad_norm": 0.22894729673862457,
+      "learning_rate": 1.03e-05,
+      "loss": 2.4219,
+      "step": 1588
+    },
+    {
+      "epoch": 0.17738520092236956,
+      "grad_norm": 0.2282966822385788,
+      "learning_rate": 1.0275e-05,
+      "loss": 2.3184,
+      "step": 1589
+    },
+    {
+      "epoch": 0.17749683415139558,
+      "grad_norm": 0.22595593333244324,
+      "learning_rate": 1.025e-05,
+      "loss": 2.3154,
+      "step": 1590
+    },
+    {
+      "epoch": 0.17760846738042163,
+      "grad_norm": 0.2264798879623413,
+      "learning_rate": 1.0225e-05,
+      "loss": 2.2779,
+      "step": 1591
+    },
+    {
+      "epoch": 0.17772010060944765,
+      "grad_norm": 0.22668862342834473,
+      "learning_rate": 1.02e-05,
+      "loss": 2.4036,
+      "step": 1592
+    },
+    {
+      "epoch": 0.1778317338384737,
+      "grad_norm": 0.22386161983013153,
+      "learning_rate": 1.0175e-05,
+      "loss": 2.2603,
+      "step": 1593
+    },
+    {
+      "epoch": 0.17794336706749972,
+      "grad_norm": 0.23508307337760925,
+      "learning_rate": 1.0150000000000001e-05,
+      "loss": 2.3921,
+      "step": 1594
+    },
+    {
+      "epoch": 0.17805500029652577,
+      "grad_norm": 0.2235022783279419,
+      "learning_rate": 1.0125e-05,
+      "loss": 2.3779,
+      "step": 1595
+    },
+    {
+      "epoch": 0.1781666335255518,
+      "grad_norm": 0.2393585741519928,
+      "learning_rate": 1.0100000000000002e-05,
+      "loss": 2.3471,
+      "step": 1596
+    },
+    {
+      "epoch": 0.17827826675457784,
+      "grad_norm": 0.22901956737041473,
+      "learning_rate": 1.0075000000000001e-05,
+      "loss": 2.2271,
+      "step": 1597
+    },
+    {
+      "epoch": 0.17838989998360386,
+      "grad_norm": 0.22379013895988464,
+      "learning_rate": 1.005e-05,
+      "loss": 2.3179,
+      "step": 1598
+    },
+    {
+      "epoch": 0.1785015332126299,
+      "grad_norm": 0.23061637580394745,
+      "learning_rate": 1.0025000000000001e-05,
+      "loss": 2.4403,
+      "step": 1599
+    },
+    {
+      "epoch": 0.17861316644165595,
+      "grad_norm": 0.22944380342960358,
+      "learning_rate": 1e-05,
+      "loss": 2.3587,
+      "step": 1600
+    },
+    {
+      "epoch": 0.17872479967068197,
+      "grad_norm": 0.22462834417819977,
+      "learning_rate": 9.975e-06,
+      "loss": 2.366,
+      "step": 1601
+    },
+    {
+      "epoch": 0.17883643289970802,
+      "grad_norm": 0.22976066172122955,
+      "learning_rate": 9.950000000000001e-06,
+      "loss": 2.3952,
+      "step": 1602
+    },
+    {
+      "epoch": 0.17894806612873404,
+      "grad_norm": 0.22779542207717896,
+      "learning_rate": 9.925e-06,
+      "loss": 2.3513,
+      "step": 1603
+    },
+    {
+      "epoch": 0.1790596993577601,
+      "grad_norm": 0.22328883409500122,
+      "learning_rate": 9.900000000000002e-06,
+      "loss": 2.3977,
+      "step": 1604
+    },
+    {
+      "epoch": 0.1791713325867861,
+      "grad_norm": 0.2311943769454956,
+      "learning_rate": 9.875000000000001e-06,
+      "loss": 2.2658,
+      "step": 1605
+    },
+    {
+      "epoch": 0.17928296581581216,
+      "grad_norm": 0.2562137246131897,
+      "learning_rate": 9.85e-06,
+      "loss": 2.3592,
+      "step": 1606
+    },
+    {
+      "epoch": 0.17939459904483818,
+      "grad_norm": 0.2313629388809204,
+      "learning_rate": 9.825000000000002e-06,
+      "loss": 2.2974,
+      "step": 1607
+    },
+    {
+      "epoch": 0.17950623227386422,
+      "grad_norm": 0.23150859773159027,
+      "learning_rate": 9.800000000000001e-06,
+      "loss": 2.428,
+      "step": 1608
+    },
+    {
+      "epoch": 0.17961786550289024,
+      "grad_norm": 0.23653681576251984,
+      "learning_rate": 9.775e-06,
+      "loss": 2.2849,
+      "step": 1609
+    },
+    {
+      "epoch": 0.1797294987319163,
+      "grad_norm": 0.23083879053592682,
+      "learning_rate": 9.750000000000002e-06,
+      "loss": 2.3614,
+      "step": 1610
+    },
+    {
+      "epoch": 0.1798411319609423,
+      "grad_norm": 0.23122479021549225,
+      "learning_rate": 9.725000000000001e-06,
+      "loss": 2.3737,
+      "step": 1611
+    },
+    {
+      "epoch": 0.17995276518996836,
+      "grad_norm": 0.22781018912792206,
+      "learning_rate": 9.7e-06,
+      "loss": 2.3419,
+      "step": 1612
+    },
+    {
+      "epoch": 0.18006439841899438,
+      "grad_norm": 0.2270815223455429,
+      "learning_rate": 9.675000000000001e-06,
+      "loss": 2.3468,
+      "step": 1613
+    },
+    {
+      "epoch": 0.18017603164802043,
+      "grad_norm": 0.2297106683254242,
+      "learning_rate": 9.65e-06,
+      "loss": 2.3164,
+      "step": 1614
+    },
+    {
+      "epoch": 0.18028766487704648,
+      "grad_norm": 0.22484375536441803,
+      "learning_rate": 9.625e-06,
+      "loss": 2.5492,
+      "step": 1615
+    },
+    {
+      "epoch": 0.1803992981060725,
+      "grad_norm": 0.230937659740448,
+      "learning_rate": 9.600000000000001e-06,
+      "loss": 2.3003,
+      "step": 1616
+    },
+    {
+      "epoch": 0.18051093133509855,
+      "grad_norm": 0.23577038943767548,
+      "learning_rate": 9.575e-06,
+      "loss": 2.2976,
+      "step": 1617
+    },
+    {
+      "epoch": 0.18062256456412457,
+      "grad_norm": 0.23040169477462769,
+      "learning_rate": 9.55e-06,
+      "loss": 2.3458,
+      "step": 1618
+    },
+    {
+      "epoch": 0.1807341977931506,
+      "grad_norm": 0.23320811986923218,
+      "learning_rate": 9.525000000000001e-06,
+      "loss": 2.3712,
+      "step": 1619
+    },
+    {
+      "epoch": 0.18084583102217663,
+      "grad_norm": 0.2216322273015976,
+      "learning_rate": 9.5e-06,
+      "loss": 2.3903,
+      "step": 1620
+    },
+    {
+      "epoch": 0.18095746425120268,
+      "grad_norm": 0.22143380343914032,
+      "learning_rate": 9.475e-06,
+      "loss": 2.3709,
+      "step": 1621
+    },
+    {
+      "epoch": 0.1810690974802287,
+      "grad_norm": 0.22373448312282562,
+      "learning_rate": 9.450000000000001e-06,
+      "loss": 2.3535,
+      "step": 1622
+    },
+    {
+      "epoch": 0.18118073070925475,
+      "grad_norm": 0.22735032439231873,
+      "learning_rate": 9.425e-06,
+      "loss": 2.3133,
+      "step": 1623
+    },
+    {
+      "epoch": 0.18129236393828077,
+      "grad_norm": 0.2304399609565735,
+      "learning_rate": 9.4e-06,
+      "loss": 2.3577,
+      "step": 1624
+    },
+    {
+      "epoch": 0.18140399716730682,
+      "grad_norm": 0.22027041018009186,
+      "learning_rate": 9.375000000000001e-06,
+      "loss": 2.352,
+      "step": 1625
+    },
+    {
+      "epoch": 0.18151563039633284,
+      "grad_norm": 0.23104159533977509,
+      "learning_rate": 9.35e-06,
+      "loss": 2.3603,
+      "step": 1626
+    },
+    {
+      "epoch": 0.1816272636253589,
+      "grad_norm": 0.26686009764671326,
+      "learning_rate": 9.325e-06,
+      "loss": 2.328,
+      "step": 1627
+    },
+    {
+      "epoch": 0.1817388968543849,
+      "grad_norm": 0.22924385964870453,
+      "learning_rate": 9.3e-06,
+      "loss": 2.3588,
+      "step": 1628
+    },
+    {
+      "epoch": 0.18185053008341096,
+      "grad_norm": 0.22922426462173462,
+      "learning_rate": 9.275e-06,
+      "loss": 2.3412,
+      "step": 1629
+    },
+    {
+      "epoch": 0.18196216331243698,
+      "grad_norm": 0.22747567296028137,
+      "learning_rate": 9.25e-06,
+      "loss": 2.3429,
+      "step": 1630
+    },
+    {
+      "epoch": 0.18207379654146302,
+      "grad_norm": 0.2208503782749176,
+      "learning_rate": 9.225e-06,
+      "loss": 2.2673,
+      "step": 1631
+    },
+    {
+      "epoch": 0.18218542977048907,
+      "grad_norm": 0.24815091490745544,
+      "learning_rate": 9.2e-06,
+      "loss": 2.4107,
+      "step": 1632
+    },
+    {
+      "epoch": 0.1822970629995151,
+      "grad_norm": 0.23134523630142212,
+      "learning_rate": 9.175000000000001e-06,
+      "loss": 2.3822,
+      "step": 1633
+    },
+    {
+      "epoch": 0.18240869622854114,
+      "grad_norm": 0.2306479662656784,
+      "learning_rate": 9.15e-06,
+      "loss": 2.3913,
+      "step": 1634
+    },
+    {
+      "epoch": 0.18252032945756716,
+      "grad_norm": 0.22402219474315643,
+      "learning_rate": 9.125e-06,
+      "loss": 2.3625,
+      "step": 1635
+    },
+    {
+      "epoch": 0.1826319626865932,
+      "grad_norm": 0.2309318333864212,
+      "learning_rate": 9.100000000000001e-06,
+      "loss": 2.4505,
+      "step": 1636
+    },
+    {
+      "epoch": 0.18274359591561923,
+      "grad_norm": 0.22473448514938354,
+      "learning_rate": 9.075e-06,
+      "loss": 2.3079,
+      "step": 1637
+    },
+    {
+      "epoch": 0.18285522914464528,
+      "grad_norm": 0.22978295385837555,
+      "learning_rate": 9.05e-06,
+      "loss": 2.3589,
+      "step": 1638
+    },
+    {
+      "epoch": 0.1829668623736713,
+      "grad_norm": 0.22440096735954285,
+      "learning_rate": 9.025e-06,
+      "loss": 2.4401,
+      "step": 1639
+    },
+    {
+      "epoch": 0.18307849560269734,
+      "grad_norm": 0.23170286417007446,
+      "learning_rate": 9e-06,
+      "loss": 2.3291,
+      "step": 1640
+    },
+    {
+      "epoch": 0.18319012883172336,
+      "grad_norm": 0.24245621263980865,
+      "learning_rate": 8.975e-06,
+      "loss": 2.3008,
+      "step": 1641
+    },
+    {
+      "epoch": 0.1833017620607494,
+      "grad_norm": 0.23115895688533783,
+      "learning_rate": 8.95e-06,
+      "loss": 2.3127,
+      "step": 1642
+    },
+    {
+      "epoch": 0.18341339528977543,
+      "grad_norm": 0.2309594601392746,
+      "learning_rate": 8.925e-06,
+      "loss": 2.3205,
+      "step": 1643
+    },
+    {
+      "epoch": 0.18352502851880148,
+      "grad_norm": 0.2410878688097,
+      "learning_rate": 8.9e-06,
+      "loss": 2.4235,
+      "step": 1644
+    },
+    {
+      "epoch": 0.1836366617478275,
+      "grad_norm": 0.2556111216545105,
+      "learning_rate": 8.875e-06,
+      "loss": 2.4193,
+      "step": 1645
+    },
+    {
+      "epoch": 0.18374829497685355,
+      "grad_norm": 0.23009131848812103,
+      "learning_rate": 8.85e-06,
+      "loss": 2.3439,
+      "step": 1646
+    },
+    {
+      "epoch": 0.18385992820587957,
+      "grad_norm": 0.23371157050132751,
+      "learning_rate": 8.825e-06,
+      "loss": 2.3897,
+      "step": 1647
+    },
+    {
+      "epoch": 0.18397156143490562,
+      "grad_norm": 0.23305875062942505,
+      "learning_rate": 8.8e-06,
+      "loss": 2.2757,
+      "step": 1648
+    },
+    {
+      "epoch": 0.18408319466393167,
+      "grad_norm": 0.22758939862251282,
+      "learning_rate": 8.775e-06,
+      "loss": 2.3526,
+      "step": 1649
+    },
+    {
+      "epoch": 0.18419482789295769,
+      "grad_norm": 0.28928419947624207,
+      "learning_rate": 8.75e-06,
+      "loss": 2.2889,
+      "step": 1650
+    },
+    {
+      "epoch": 0.18430646112198373,
+      "grad_norm": 0.23369638621807098,
+      "learning_rate": 8.725e-06,
+      "loss": 2.3379,
+      "step": 1651
+    },
+    {
+      "epoch": 0.18441809435100975,
+      "grad_norm": 0.23393464088439941,
+      "learning_rate": 8.7e-06,
+      "loss": 2.3827,
+      "step": 1652
+    },
+    {
+      "epoch": 0.1845297275800358,
+      "grad_norm": 0.22967039048671722,
+      "learning_rate": 8.674999999999999e-06,
+      "loss": 2.4529,
+      "step": 1653
+    },
+    {
+      "epoch": 0.18464136080906182,
+      "grad_norm": 0.311108261346817,
+      "learning_rate": 8.65e-06,
+      "loss": 2.2626,
+      "step": 1654
+    },
+    {
+      "epoch": 0.18475299403808787,
+      "grad_norm": 0.23404546082019806,
+      "learning_rate": 8.625e-06,
+      "loss": 2.2972,
+      "step": 1655
+    },
+    {
+      "epoch": 0.1848646272671139,
+      "grad_norm": 0.2330547720193863,
+      "learning_rate": 8.599999999999999e-06,
+      "loss": 2.3406,
+      "step": 1656
+    },
+    {
+      "epoch": 0.18497626049613994,
+      "grad_norm": 0.22166459262371063,
+      "learning_rate": 8.575000000000002e-06,
+      "loss": 2.2893,
+      "step": 1657
+    },
+    {
+      "epoch": 0.18508789372516596,
+      "grad_norm": 0.2245807945728302,
+      "learning_rate": 8.550000000000001e-06,
+      "loss": 2.1937,
+      "step": 1658
+    },
+    {
+      "epoch": 0.185199526954192,
+      "grad_norm": 0.22609424591064453,
+      "learning_rate": 8.525e-06,
+      "loss": 2.3882,
+      "step": 1659
+    },
+    {
+      "epoch": 0.18531116018321803,
+      "grad_norm": 0.22625020146369934,
+      "learning_rate": 8.500000000000002e-06,
+      "loss": 2.3893,
+      "step": 1660
+    },
+    {
+      "epoch": 0.18542279341224407,
+      "grad_norm": 0.25080451369285583,
+      "learning_rate": 8.475000000000001e-06,
+      "loss": 2.2867,
+      "step": 1661
+    },
+    {
+      "epoch": 0.1855344266412701,
+      "grad_norm": 0.23033380508422852,
+      "learning_rate": 8.45e-06,
+      "loss": 2.3523,
+      "step": 1662
+    },
+    {
+      "epoch": 0.18564605987029614,
+      "grad_norm": 0.23714114725589752,
+      "learning_rate": 8.425000000000001e-06,
+      "loss": 2.3399,
+      "step": 1663
+    },
+    {
+      "epoch": 0.18575769309932216,
+      "grad_norm": 0.23813962936401367,
+      "learning_rate": 8.400000000000001e-06,
+      "loss": 2.3729,
+      "step": 1664
+    },
+    {
+      "epoch": 0.1858693263283482,
+      "grad_norm": 0.22549176216125488,
+      "learning_rate": 8.375e-06,
+      "loss": 2.3674,
+      "step": 1665
+    },
+    {
+      "epoch": 0.18598095955737426,
+      "grad_norm": 0.2233189046382904,
+      "learning_rate": 8.350000000000001e-06,
+      "loss": 2.3293,
+      "step": 1666
+    },
+    {
+      "epoch": 0.18609259278640028,
+      "grad_norm": 0.22096222639083862,
+      "learning_rate": 8.325e-06,
+      "loss": 2.3471,
+      "step": 1667
+    },
+    {
+      "epoch": 0.18620422601542633,
+      "grad_norm": 0.2315039187669754,
+      "learning_rate": 8.3e-06,
+      "loss": 2.3451,
+      "step": 1668
+    },
+    {
+      "epoch": 0.18631585924445235,
+      "grad_norm": 0.23510505259037018,
+      "learning_rate": 8.275000000000001e-06,
+      "loss": 2.3604,
+      "step": 1669
+    },
+    {
+      "epoch": 0.1864274924734784,
+      "grad_norm": 0.23275664448738098,
+      "learning_rate": 8.25e-06,
+      "loss": 2.3314,
+      "step": 1670
+    },
+    {
+      "epoch": 0.18653912570250442,
+      "grad_norm": 0.22906853258609772,
+      "learning_rate": 8.225e-06,
+      "loss": 2.3489,
+      "step": 1671
+    },
+    {
+      "epoch": 0.18665075893153046,
+      "grad_norm": 0.223612442612648,
+      "learning_rate": 8.200000000000001e-06,
+      "loss": 2.4229,
+      "step": 1672
+    },
+    {
+      "epoch": 0.18676239216055648,
+      "grad_norm": 0.22447071969509125,
+      "learning_rate": 8.175e-06,
+      "loss": 2.4356,
+      "step": 1673
+    },
+    {
+      "epoch": 0.18687402538958253,
+      "grad_norm": 0.2319689691066742,
+      "learning_rate": 8.15e-06,
+      "loss": 2.3806,
+      "step": 1674
+    },
+    {
+      "epoch": 0.18698565861860855,
+      "grad_norm": 0.22558631002902985,
+      "learning_rate": 8.125000000000001e-06,
+      "loss": 2.3246,
+      "step": 1675
+    },
+    {
+      "epoch": 0.1870972918476346,
+      "grad_norm": 0.23677797615528107,
+      "learning_rate": 8.1e-06,
+      "loss": 2.3814,
+      "step": 1676
+    },
+    {
+      "epoch": 0.18720892507666062,
+      "grad_norm": 0.2355109453201294,
+      "learning_rate": 8.075000000000001e-06,
+      "loss": 2.3477,
+      "step": 1677
+    },
+    {
+      "epoch": 0.18732055830568667,
+      "grad_norm": 0.22588761150836945,
+      "learning_rate": 8.050000000000001e-06,
+      "loss": 2.2773,
+      "step": 1678
+    },
+    {
+      "epoch": 0.1874321915347127,
+      "grad_norm": 0.22162844240665436,
+      "learning_rate": 8.025e-06,
+      "loss": 2.3412,
+      "step": 1679
+    },
+    {
+      "epoch": 0.18754382476373874,
+      "grad_norm": 0.23201774060726166,
+      "learning_rate": 8.000000000000001e-06,
+      "loss": 2.3573,
+      "step": 1680
+    },
+    {
+      "epoch": 0.18765545799276478,
+      "grad_norm": 0.22092294692993164,
+      "learning_rate": 7.975e-06,
+      "loss": 2.3214,
+      "step": 1681
+    },
+    {
+      "epoch": 0.1877670912217908,
+      "grad_norm": 0.2279776781797409,
+      "learning_rate": 7.95e-06,
+      "loss": 2.3555,
+      "step": 1682
+    },
+    {
+      "epoch": 0.18787872445081685,
+      "grad_norm": 0.23055800795555115,
+      "learning_rate": 7.925000000000001e-06,
+      "loss": 2.2783,
+      "step": 1683
+    },
+    {
+      "epoch": 0.18799035767984287,
+      "grad_norm": 0.22860021889209747,
+      "learning_rate": 7.9e-06,
+      "loss": 2.3331,
+      "step": 1684
+    },
+    {
+      "epoch": 0.18810199090886892,
+      "grad_norm": 0.22126097977161407,
+      "learning_rate": 7.875e-06,
+      "loss": 2.3946,
+      "step": 1685
+    },
+    {
+      "epoch": 0.18821362413789494,
+      "grad_norm": 0.22869837284088135,
+      "learning_rate": 7.850000000000001e-06,
+      "loss": 2.2846,
+      "step": 1686
+    },
+    {
+      "epoch": 0.188325257366921,
+      "grad_norm": 0.23146143555641174,
+      "learning_rate": 7.825e-06,
+      "loss": 2.3452,
+      "step": 1687
+    },
+    {
+      "epoch": 0.188436890595947,
+      "grad_norm": 0.24097077548503876,
+      "learning_rate": 7.8e-06,
+      "loss": 2.2416,
+      "step": 1688
+    },
+    {
+      "epoch": 0.18854852382497306,
+      "grad_norm": 0.23362663388252258,
+      "learning_rate": 7.775000000000001e-06,
+      "loss": 2.3445,
+      "step": 1689
+    },
+    {
+      "epoch": 0.18866015705399908,
+      "grad_norm": 0.23741386830806732,
+      "learning_rate": 7.75e-06,
+      "loss": 2.4122,
+      "step": 1690
+    },
+    {
+      "epoch": 0.18877179028302513,
+      "grad_norm": 0.2494947612285614,
+      "learning_rate": 7.725e-06,
+      "loss": 2.2177,
+      "step": 1691
+    },
+    {
+      "epoch": 0.18888342351205115,
+      "grad_norm": 0.23337315022945404,
+      "learning_rate": 7.7e-06,
+      "loss": 2.3019,
+      "step": 1692
+    },
+    {
+      "epoch": 0.1889950567410772,
+      "grad_norm": 0.2309933602809906,
+      "learning_rate": 7.675e-06,
+      "loss": 2.3468,
+      "step": 1693
+    },
+    {
+      "epoch": 0.18910668997010321,
+      "grad_norm": 0.22482022643089294,
+      "learning_rate": 7.65e-06,
+      "loss": 2.2141,
+      "step": 1694
+    },
+    {
+      "epoch": 0.18921832319912926,
+      "grad_norm": 0.24350865185260773,
+      "learning_rate": 7.625e-06,
+      "loss": 2.3257,
+      "step": 1695
+    },
+    {
+      "epoch": 0.18932995642815528,
+      "grad_norm": 0.2380324900150299,
+      "learning_rate": 7.6e-06,
+      "loss": 2.3769,
+      "step": 1696
+    },
+    {
+      "epoch": 0.18944158965718133,
+      "grad_norm": 0.23215071856975555,
+      "learning_rate": 7.575e-06,
+      "loss": 2.3028,
+      "step": 1697
+    },
+    {
+      "epoch": 0.18955322288620738,
+      "grad_norm": 0.24625477194786072,
+      "learning_rate": 7.55e-06,
+      "loss": 2.3829,
+      "step": 1698
+    },
+    {
+      "epoch": 0.1896648561152334,
+      "grad_norm": 0.22713284194469452,
+      "learning_rate": 7.525e-06,
+      "loss": 2.3282,
+      "step": 1699
+    },
+    {
+      "epoch": 0.18977648934425945,
+      "grad_norm": 0.221211776137352,
+      "learning_rate": 7.5e-06,
+      "loss": 2.325,
+      "step": 1700
+    },
+    {
+      "epoch": 0.18988812257328547,
+      "grad_norm": 0.23911140859127045,
+      "learning_rate": 7.4750000000000004e-06,
+      "loss": 2.2369,
+      "step": 1701
+    },
+    {
+      "epoch": 0.18999975580231152,
+      "grad_norm": 0.22453325986862183,
+      "learning_rate": 7.45e-06,
+      "loss": 2.3757,
+      "step": 1702
+    },
+    {
+      "epoch": 0.19011138903133754,
+      "grad_norm": 0.2336825579404831,
+      "learning_rate": 7.425e-06,
+      "loss": 2.2279,
+      "step": 1703
+    },
+    {
+      "epoch": 0.19022302226036358,
+      "grad_norm": 0.2163713276386261,
+      "learning_rate": 7.4e-06,
+      "loss": 2.3206,
+      "step": 1704
+    },
+    {
+      "epoch": 0.1903346554893896,
+      "grad_norm": 0.23673447966575623,
+      "learning_rate": 7.375e-06,
+      "loss": 2.3235,
+      "step": 1705
+    },
+    {
+      "epoch": 0.19044628871841565,
+      "grad_norm": 0.23599569499492645,
+      "learning_rate": 7.35e-06,
+      "loss": 2.4519,
+      "step": 1706
+    },
+    {
+      "epoch": 0.19055792194744167,
+      "grad_norm": 0.23098653554916382,
+      "learning_rate": 7.325e-06,
+      "loss": 2.344,
+      "step": 1707
+    },
+    {
+      "epoch": 0.19066955517646772,
+      "grad_norm": 0.2260526716709137,
+      "learning_rate": 7.2999999999999996e-06,
+      "loss": 2.2734,
+      "step": 1708
+    },
+    {
+      "epoch": 0.19078118840549374,
+      "grad_norm": 1.934260606765747,
+      "learning_rate": 7.275e-06,
+      "loss": 2.3161,
+      "step": 1709
+    },
+    {
+      "epoch": 0.1908928216345198,
+      "grad_norm": 0.22405433654785156,
+      "learning_rate": 7.25e-06,
+      "loss": 2.3647,
+      "step": 1710
+    },
+    {
+      "epoch": 0.1910044548635458,
+      "grad_norm": 0.22353483736515045,
+      "learning_rate": 7.2249999999999994e-06,
+      "loss": 2.3636,
+      "step": 1711
+    },
+    {
+      "epoch": 0.19111608809257186,
+      "grad_norm": 0.2400137484073639,
+      "learning_rate": 7.2e-06,
+      "loss": 2.4212,
+      "step": 1712
+    },
+    {
+      "epoch": 0.19122772132159788,
+      "grad_norm": 0.22043661773204803,
+      "learning_rate": 7.175e-06,
+      "loss": 2.33,
+      "step": 1713
+    },
+    {
+      "epoch": 0.19133935455062392,
+      "grad_norm": 0.23072798550128937,
+      "learning_rate": 7.15e-06,
+      "loss": 2.3748,
+      "step": 1714
+    },
+    {
+      "epoch": 0.19145098777964997,
+      "grad_norm": 0.22338813543319702,
+      "learning_rate": 7.1249999999999995e-06,
+      "loss": 2.3183,
+      "step": 1715
+    },
+    {
+      "epoch": 0.191562621008676,
+      "grad_norm": 0.23001807928085327,
+      "learning_rate": 7.1e-06,
+      "loss": 2.31,
+      "step": 1716
+    },
+    {
+      "epoch": 0.19167425423770204,
+      "grad_norm": 0.23304982483386993,
+      "learning_rate": 7.075e-06,
+      "loss": 2.3071,
+      "step": 1717
+    },
+    {
+      "epoch": 0.19178588746672806,
+      "grad_norm": 0.2388128638267517,
+      "learning_rate": 7.049999999999999e-06,
+      "loss": 2.3441,
+      "step": 1718
+    },
+    {
+      "epoch": 0.1918975206957541,
+      "grad_norm": 0.22486941516399384,
+      "learning_rate": 7.025000000000001e-06,
+      "loss": 2.3495,
+      "step": 1719
+    },
+    {
+      "epoch": 0.19200915392478013,
+      "grad_norm": 0.23193509876728058,
+      "learning_rate": 7.000000000000001e-06,
+      "loss": 2.2735,
+      "step": 1720
+    },
+    {
+      "epoch": 0.19212078715380618,
+      "grad_norm": 0.23028187453746796,
+      "learning_rate": 6.975000000000001e-06,
+      "loss": 2.2935,
+      "step": 1721
+    },
+    {
+      "epoch": 0.1922324203828322,
+      "grad_norm": 0.22155210375785828,
+      "learning_rate": 6.950000000000001e-06,
+      "loss": 2.3499,
+      "step": 1722
+    },
+    {
+      "epoch": 0.19234405361185825,
+      "grad_norm": 0.23238146305084229,
+      "learning_rate": 6.925000000000001e-06,
+      "loss": 2.3038,
+      "step": 1723
+    },
+    {
+      "epoch": 0.19245568684088427,
+      "grad_norm": 0.2326563447713852,
+      "learning_rate": 6.900000000000001e-06,
+      "loss": 2.3849,
+      "step": 1724
+    },
+    {
+      "epoch": 0.1925673200699103,
+      "grad_norm": 0.23055623471736908,
+      "learning_rate": 6.875000000000001e-06,
+      "loss": 2.294,
+      "step": 1725
+    },
+    {
+      "epoch": 0.19267895329893633,
+      "grad_norm": 0.22514155507087708,
+      "learning_rate": 6.8500000000000005e-06,
+      "loss": 2.4518,
+      "step": 1726
+    },
+    {
+      "epoch": 0.19279058652796238,
+      "grad_norm": 0.22970163822174072,
+      "learning_rate": 6.825000000000001e-06,
+      "loss": 2.3659,
+      "step": 1727
+    },
+    {
+      "epoch": 0.1929022197569884,
+      "grad_norm": 0.22430215775966644,
+      "learning_rate": 6.800000000000001e-06,
+      "loss": 2.4064,
+      "step": 1728
+    },
+    {
+      "epoch": 0.19301385298601445,
+      "grad_norm": 0.2254297137260437,
+      "learning_rate": 6.775000000000001e-06,
+      "loss": 2.2301,
+      "step": 1729
+    },
+    {
+      "epoch": 0.1931254862150405,
+      "grad_norm": 0.23394359648227692,
+      "learning_rate": 6.750000000000001e-06,
+      "loss": 2.311,
+      "step": 1730
+    },
+    {
+      "epoch": 0.19323711944406652,
+      "grad_norm": 0.23774048686027527,
+      "learning_rate": 6.725000000000001e-06,
+      "loss": 2.3392,
+      "step": 1731
+    },
+    {
+      "epoch": 0.19334875267309257,
+      "grad_norm": 0.23314325511455536,
+      "learning_rate": 6.700000000000001e-06,
+      "loss": 2.2952,
+      "step": 1732
+    },
+    {
+      "epoch": 0.1934603859021186,
+      "grad_norm": 0.22659903764724731,
+      "learning_rate": 6.6750000000000005e-06,
+      "loss": 2.3357,
+      "step": 1733
+    },
+    {
+      "epoch": 0.19357201913114463,
+      "grad_norm": 0.2298395037651062,
+      "learning_rate": 6.650000000000001e-06,
+      "loss": 2.3717,
+      "step": 1734
+    },
+    {
+      "epoch": 0.19368365236017066,
+      "grad_norm": 0.25462180376052856,
+      "learning_rate": 6.625000000000001e-06,
+      "loss": 2.3877,
+      "step": 1735
+    },
+    {
+      "epoch": 0.1937952855891967,
+      "grad_norm": 0.22525596618652344,
+      "learning_rate": 6.6e-06,
+      "loss": 2.2066,
+      "step": 1736
+    },
+    {
+      "epoch": 0.19390691881822272,
+      "grad_norm": 0.2275206297636032,
+      "learning_rate": 6.5750000000000006e-06,
+      "loss": 2.3948,
+      "step": 1737
+    },
+    {
+      "epoch": 0.19401855204724877,
+      "grad_norm": 0.23473072052001953,
+      "learning_rate": 6.550000000000001e-06,
+      "loss": 2.3358,
+      "step": 1738
+    },
+    {
+      "epoch": 0.1941301852762748,
+      "grad_norm": 0.2280065417289734,
+      "learning_rate": 6.525e-06,
+      "loss": 2.2785,
+      "step": 1739
+    },
+    {
+      "epoch": 0.19424181850530084,
+      "grad_norm": 0.2484257072210312,
+      "learning_rate": 6.5000000000000004e-06,
+      "loss": 2.4323,
+      "step": 1740
+    },
+    {
+      "epoch": 0.19435345173432686,
+      "grad_norm": 0.245782271027565,
+      "learning_rate": 6.475000000000001e-06,
+      "loss": 2.3596,
+      "step": 1741
+    },
+    {
+      "epoch": 0.1944650849633529,
+      "grad_norm": 0.2271096259355545,
+      "learning_rate": 6.45e-06,
+      "loss": 2.3226,
+      "step": 1742
+    },
+    {
+      "epoch": 0.19457671819237893,
+      "grad_norm": 0.23324838280677795,
+      "learning_rate": 6.425e-06,
+      "loss": 2.4395,
+      "step": 1743
+    },
+    {
+      "epoch": 0.19468835142140498,
+      "grad_norm": 0.23519247770309448,
+      "learning_rate": 6.4000000000000006e-06,
+      "loss": 2.2311,
+      "step": 1744
+    },
+    {
+      "epoch": 0.194799984650431,
+      "grad_norm": 0.22893019020557404,
+      "learning_rate": 6.375000000000001e-06,
+      "loss": 2.3042,
+      "step": 1745
+    },
+    {
+      "epoch": 0.19491161787945704,
+      "grad_norm": 0.23204751312732697,
+      "learning_rate": 6.35e-06,
+      "loss": 2.3882,
+      "step": 1746
+    },
+    {
+      "epoch": 0.1950232511084831,
+      "grad_norm": 0.23840852081775665,
+      "learning_rate": 6.3250000000000004e-06,
+      "loss": 2.3279,
+      "step": 1747
+    },
+    {
+      "epoch": 0.1951348843375091,
+      "grad_norm": 0.2263709455728531,
+      "learning_rate": 6.300000000000001e-06,
+      "loss": 2.3376,
+      "step": 1748
+    },
+    {
+      "epoch": 0.19524651756653516,
+      "grad_norm": 0.22863459587097168,
+      "learning_rate": 6.275e-06,
+      "loss": 2.3901,
+      "step": 1749
+    },
+    {
+      "epoch": 0.19535815079556118,
+      "grad_norm": 0.3761008381843567,
+      "learning_rate": 6.25e-06,
+      "loss": 2.3816,
+      "step": 1750
+    },
+    {
+      "epoch": 0.19546978402458723,
+      "grad_norm": 0.22513067722320557,
+      "learning_rate": 6.2250000000000005e-06,
+      "loss": 2.2964,
+      "step": 1751
+    },
+    {
+      "epoch": 0.19558141725361325,
+      "grad_norm": 0.22629620134830475,
+      "learning_rate": 6.2e-06,
+      "loss": 2.4423,
+      "step": 1752
+    },
+    {
+      "epoch": 0.1956930504826393,
+      "grad_norm": 0.24019905924797058,
+      "learning_rate": 6.175e-06,
+      "loss": 2.2727,
+      "step": 1753
+    },
+    {
+      "epoch": 0.19580468371166532,
+      "grad_norm": 0.23448988795280457,
+      "learning_rate": 6.15e-06,
+      "loss": 2.3269,
+      "step": 1754
+    },
+    {
+      "epoch": 0.19591631694069137,
+      "grad_norm": 0.22556640207767487,
+      "learning_rate": 6.125e-06,
+      "loss": 2.1874,
+      "step": 1755
+    },
+    {
+      "epoch": 0.19602795016971739,
+      "grad_norm": 0.3827759027481079,
+      "learning_rate": 6.1e-06,
+      "loss": 2.3076,
+      "step": 1756
+    },
+    {
+      "epoch": 0.19613958339874343,
+      "grad_norm": 0.22821791470050812,
+      "learning_rate": 6.075e-06,
+      "loss": 2.3521,
+      "step": 1757
+    },
+    {
+      "epoch": 0.19625121662776945,
+      "grad_norm": 0.22949843108654022,
+      "learning_rate": 6.0500000000000005e-06,
+      "loss": 2.3864,
+      "step": 1758
+    },
+    {
+      "epoch": 0.1963628498567955,
+      "grad_norm": 0.22942288219928741,
+      "learning_rate": 6.025e-06,
+      "loss": 2.3699,
+      "step": 1759
+    },
+    {
+      "epoch": 0.19647448308582152,
+      "grad_norm": 0.2209300398826599,
+      "learning_rate": 6e-06,
+      "loss": 2.3237,
+      "step": 1760
+    },
+    {
+      "epoch": 0.19658611631484757,
+      "grad_norm": 0.22622691094875336,
+      "learning_rate": 5.975e-06,
+      "loss": 2.3339,
+      "step": 1761
+    },
+    {
+      "epoch": 0.1966977495438736,
+      "grad_norm": 0.23727689683437347,
+      "learning_rate": 5.95e-06,
+      "loss": 2.2707,
+      "step": 1762
+    },
+    {
+      "epoch": 0.19680938277289964,
+      "grad_norm": 0.2961234450340271,
+      "learning_rate": 5.925e-06,
+      "loss": 2.2837,
+      "step": 1763
+    },
+    {
+      "epoch": 0.19692101600192569,
+      "grad_norm": 0.24365472793579102,
+      "learning_rate": 5.9e-06,
+      "loss": 2.3899,
+      "step": 1764
+    },
+    {
+      "epoch": 0.1970326492309517,
+      "grad_norm": 0.23506611585617065,
+      "learning_rate": 5.875e-06,
+      "loss": 2.3549,
+      "step": 1765
+    },
+    {
+      "epoch": 0.19714428245997775,
+      "grad_norm": 0.22377490997314453,
+      "learning_rate": 5.850000000000001e-06,
+      "loss": 2.2019,
+      "step": 1766
+    },
+    {
+      "epoch": 0.19725591568900377,
+      "grad_norm": 0.2170470505952835,
+      "learning_rate": 5.825000000000001e-06,
+      "loss": 2.3304,
+      "step": 1767
+    },
+    {
+      "epoch": 0.19736754891802982,
+      "grad_norm": 0.23338063061237335,
+      "learning_rate": 5.8e-06,
+      "loss": 2.3452,
+      "step": 1768
+    },
+    {
+      "epoch": 0.19747918214705584,
+      "grad_norm": 0.22321484982967377,
+      "learning_rate": 5.775000000000001e-06,
+      "loss": 2.321,
+      "step": 1769
+    },
+    {
+      "epoch": 0.1975908153760819,
+      "grad_norm": 0.23838651180267334,
+      "learning_rate": 5.750000000000001e-06,
+      "loss": 2.3604,
+      "step": 1770
+    },
+    {
+      "epoch": 0.1977024486051079,
+      "grad_norm": 0.23211868107318878,
+      "learning_rate": 5.725e-06,
+      "loss": 2.3697,
+      "step": 1771
+    },
+    {
+      "epoch": 0.19781408183413396,
+      "grad_norm": 0.23045557737350464,
+      "learning_rate": 5.7000000000000005e-06,
+      "loss": 2.2953,
+      "step": 1772
+    },
+    {
+      "epoch": 0.19792571506315998,
+      "grad_norm": 0.23346678912639618,
+      "learning_rate": 5.675000000000001e-06,
+      "loss": 2.3694,
+      "step": 1773
+    },
+    {
+      "epoch": 0.19803734829218603,
+      "grad_norm": 0.23384539783000946,
+      "learning_rate": 5.65e-06,
+      "loss": 2.3231,
+      "step": 1774
+    },
+    {
+      "epoch": 0.19814898152121205,
+      "grad_norm": 0.22029776871204376,
+      "learning_rate": 5.625e-06,
+      "loss": 2.2113,
+      "step": 1775
+    },
+    {
+      "epoch": 0.1982606147502381,
+      "grad_norm": 0.24596329033374786,
+      "learning_rate": 5.600000000000001e-06,
+      "loss": 2.2872,
+      "step": 1776
+    },
+    {
+      "epoch": 0.19837224797926412,
+      "grad_norm": 0.2208588421344757,
+      "learning_rate": 5.575e-06,
+      "loss": 2.2893,
+      "step": 1777
+    },
+    {
+      "epoch": 0.19848388120829016,
+      "grad_norm": 0.27724799513816833,
+      "learning_rate": 5.55e-06,
+      "loss": 2.3499,
+      "step": 1778
+    },
+    {
+      "epoch": 0.19859551443731618,
+      "grad_norm": 0.2821560502052307,
+      "learning_rate": 5.5250000000000005e-06,
+      "loss": 2.2631,
+      "step": 1779
+    },
+    {
+      "epoch": 0.19870714766634223,
+      "grad_norm": 0.2291664034128189,
+      "learning_rate": 5.500000000000001e-06,
+      "loss": 2.4097,
+      "step": 1780
+    },
+    {
+      "epoch": 0.19881878089536828,
+      "grad_norm": 0.2263956367969513,
+      "learning_rate": 5.475e-06,
+      "loss": 2.4372,
+      "step": 1781
+    },
+    {
+      "epoch": 0.1989304141243943,
+      "grad_norm": 0.22560614347457886,
+      "learning_rate": 5.45e-06,
+      "loss": 2.3875,
+      "step": 1782
+    },
+    {
+      "epoch": 0.19904204735342035,
+      "grad_norm": 0.3105284571647644,
+      "learning_rate": 5.4250000000000006e-06,
+      "loss": 2.201,
+      "step": 1783
+    },
+    {
+      "epoch": 0.19915368058244637,
+      "grad_norm": 0.23622126877307892,
+      "learning_rate": 5.4e-06,
+      "loss": 2.3385,
+      "step": 1784
+    },
+    {
+      "epoch": 0.19926531381147242,
+      "grad_norm": 0.23378053307533264,
+      "learning_rate": 5.375e-06,
+      "loss": 2.3863,
+      "step": 1785
+    },
+    {
+      "epoch": 0.19937694704049844,
+      "grad_norm": 0.2226409614086151,
+      "learning_rate": 5.3500000000000004e-06,
+      "loss": 2.2851,
+      "step": 1786
+    },
+    {
+      "epoch": 0.19948858026952448,
+      "grad_norm": 0.2296893298625946,
+      "learning_rate": 5.325e-06,
+      "loss": 2.3351,
+      "step": 1787
+    },
+    {
+      "epoch": 0.1996002134985505,
+      "grad_norm": 0.22921861708164215,
+      "learning_rate": 5.3e-06,
+      "loss": 2.2895,
+      "step": 1788
+    },
+    {
+      "epoch": 0.19971184672757655,
+      "grad_norm": 0.23278868198394775,
+      "learning_rate": 5.275e-06,
+      "loss": 2.2488,
+      "step": 1789
+    },
+    {
+      "epoch": 0.19982347995660257,
+      "grad_norm": 0.2284834384918213,
+      "learning_rate": 5.25e-06,
+      "loss": 2.3363,
+      "step": 1790
+    },
+    {
+      "epoch": 0.19993511318562862,
+      "grad_norm": 0.23466360569000244,
+      "learning_rate": 5.225e-06,
+      "loss": 2.2327,
+      "step": 1791
+    },
+    {
+      "epoch": 0.20004674641465464,
+      "grad_norm": 0.23456375300884247,
+      "learning_rate": 5.2e-06,
+      "loss": 2.3441,
+      "step": 1792
+    },
+    {
+      "epoch": 0.2001583796436807,
+      "grad_norm": 0.7794348001480103,
+      "learning_rate": 5.175e-06,
+      "loss": 2.2631,
+      "step": 1793
+    },
+    {
+      "epoch": 0.2002700128727067,
+      "grad_norm": 0.22322069108486176,
+      "learning_rate": 5.15e-06,
+      "loss": 2.209,
+      "step": 1794
+    },
+    {
+      "epoch": 0.20038164610173276,
+      "grad_norm": 0.2890242636203766,
+      "learning_rate": 5.125e-06,
+      "loss": 2.4408,
+      "step": 1795
+    },
+    {
+      "epoch": 0.2004932793307588,
+      "grad_norm": 0.23017571866512299,
+      "learning_rate": 5.1e-06,
+      "loss": 2.3729,
+      "step": 1796
+    },
+    {
+      "epoch": 0.20060491255978483,
+      "grad_norm": 0.24546416103839874,
+      "learning_rate": 5.0750000000000005e-06,
+      "loss": 2.3414,
+      "step": 1797
+    },
+    {
+      "epoch": 0.20071654578881087,
+      "grad_norm": 0.6677102446556091,
+      "learning_rate": 5.050000000000001e-06,
+      "loss": 2.267,
+      "step": 1798
+    },
+    {
+      "epoch": 0.2008281790178369,
+      "grad_norm": 0.22715935111045837,
+      "learning_rate": 5.025e-06,
+      "loss": 2.2485,
+      "step": 1799
+    },
+    {
+      "epoch": 0.20093981224686294,
+      "grad_norm": 0.24433374404907227,
+      "learning_rate": 5e-06,
+      "loss": 2.2448,
+      "step": 1800
+    },
+    {
+      "epoch": 0.20105144547588896,
+      "grad_norm": 0.2298104465007782,
+      "learning_rate": 4.975000000000001e-06,
+      "loss": 2.3481,
+      "step": 1801
+    },
+    {
+      "epoch": 0.201163078704915,
+      "grad_norm": 0.2273014336824417,
+      "learning_rate": 4.950000000000001e-06,
+      "loss": 2.2275,
+      "step": 1802
+    },
+    {
+      "epoch": 0.20127471193394103,
+      "grad_norm": 0.2342415452003479,
+      "learning_rate": 4.925e-06,
+      "loss": 2.3233,
+      "step": 1803
+    },
+    {
+      "epoch": 0.20138634516296708,
+      "grad_norm": 0.23677963018417358,
+      "learning_rate": 4.9000000000000005e-06,
+      "loss": 2.34,
+      "step": 1804
+    },
+    {
+      "epoch": 0.2014979783919931,
+      "grad_norm": 0.2304297685623169,
+      "learning_rate": 4.875000000000001e-06,
+      "loss": 2.4005,
+      "step": 1805
+    },
+    {
+      "epoch": 0.20160961162101915,
+      "grad_norm": 0.22330938279628754,
+      "learning_rate": 4.85e-06,
+      "loss": 2.31,
+      "step": 1806
+    },
+    {
+      "epoch": 0.20172124485004517,
+      "grad_norm": 0.2279394567012787,
+      "learning_rate": 4.825e-06,
+      "loss": 2.2854,
+      "step": 1807
+    },
+    {
+      "epoch": 0.20183287807907122,
+      "grad_norm": 0.22466933727264404,
+      "learning_rate": 4.800000000000001e-06,
+      "loss": 2.3226,
+      "step": 1808
+    },
+    {
+      "epoch": 0.20194451130809724,
+      "grad_norm": 0.2288239598274231,
+      "learning_rate": 4.775e-06,
+      "loss": 2.2858,
+      "step": 1809
+    },
+    {
+      "epoch": 0.20205614453712328,
+      "grad_norm": 0.22833405435085297,
+      "learning_rate": 4.75e-06,
+      "loss": 2.3368,
+      "step": 1810
+    },
+    {
+      "epoch": 0.2021677777661493,
+      "grad_norm": 0.23820479214191437,
+      "learning_rate": 4.7250000000000005e-06,
+      "loss": 2.3351,
+      "step": 1811
+    },
+    {
+      "epoch": 0.20227941099517535,
+      "grad_norm": 0.22864072024822235,
+      "learning_rate": 4.7e-06,
+      "loss": 2.3744,
+      "step": 1812
+    },
+    {
+      "epoch": 0.2023910442242014,
+      "grad_norm": 0.22456790506839752,
+      "learning_rate": 4.675e-06,
+      "loss": 2.2479,
+      "step": 1813
+    },
+    {
+      "epoch": 0.20250267745322742,
+      "grad_norm": 0.22676309943199158,
+      "learning_rate": 4.65e-06,
+      "loss": 2.4175,
+      "step": 1814
+    },
+    {
+      "epoch": 0.20261431068225347,
+      "grad_norm": 0.2293044477701187,
+      "learning_rate": 4.625e-06,
+      "loss": 2.3798,
+      "step": 1815
+    },
+    {
+      "epoch": 0.2027259439112795,
+      "grad_norm": 0.23791825771331787,
+      "learning_rate": 4.6e-06,
+      "loss": 2.2226,
+      "step": 1816
+    },
+    {
+      "epoch": 0.20283757714030554,
+      "grad_norm": 0.22495396435260773,
+      "learning_rate": 4.575e-06,
+      "loss": 2.3658,
+      "step": 1817
+    },
+    {
+      "epoch": 0.20294921036933156,
+      "grad_norm": 0.22882862389087677,
+      "learning_rate": 4.5500000000000005e-06,
+      "loss": 2.3152,
+      "step": 1818
+    },
+    {
+      "epoch": 0.2030608435983576,
+      "grad_norm": 0.24084174633026123,
+      "learning_rate": 4.525e-06,
+      "loss": 2.3789,
+      "step": 1819
+    },
+    {
+      "epoch": 0.20317247682738362,
+      "grad_norm": 0.23165123164653778,
+      "learning_rate": 4.5e-06,
+      "loss": 2.3144,
+      "step": 1820
+    },
+    {
+      "epoch": 0.20328411005640967,
+      "grad_norm": 0.22660130262374878,
+      "learning_rate": 4.475e-06,
+      "loss": 2.2471,
+      "step": 1821
+    },
+    {
+      "epoch": 0.2033957432854357,
+      "grad_norm": 0.23442165553569794,
+      "learning_rate": 4.45e-06,
+      "loss": 2.2553,
+      "step": 1822
+    },
+    {
+      "epoch": 0.20350737651446174,
+      "grad_norm": 0.2298109382390976,
+      "learning_rate": 4.425e-06,
+      "loss": 2.3108,
+      "step": 1823
+    },
+    {
+      "epoch": 0.20361900974348776,
+      "grad_norm": 0.23466314375400543,
+      "learning_rate": 4.4e-06,
+      "loss": 2.2141,
+      "step": 1824
+    },
+    {
+      "epoch": 0.2037306429725138,
+      "grad_norm": 0.2319631427526474,
+      "learning_rate": 4.375e-06,
+      "loss": 2.2492,
+      "step": 1825
+    },
+    {
+      "epoch": 0.20384227620153983,
+      "grad_norm": 0.23453742265701294,
+      "learning_rate": 4.35e-06,
+      "loss": 2.2947,
+      "step": 1826
+    },
+    {
+      "epoch": 0.20395390943056588,
+      "grad_norm": 0.23079949617385864,
+      "learning_rate": 4.325e-06,
+      "loss": 2.3735,
+      "step": 1827
+    },
+    {
+      "epoch": 0.2040655426595919,
+      "grad_norm": 0.2320202887058258,
+      "learning_rate": 4.2999999999999995e-06,
+      "loss": 2.2555,
+      "step": 1828
+    },
+    {
+      "epoch": 0.20417717588861795,
+      "grad_norm": 0.23471979796886444,
+      "learning_rate": 4.2750000000000006e-06,
+      "loss": 2.274,
+      "step": 1829
+    },
+    {
+      "epoch": 0.204288809117644,
+      "grad_norm": 0.23321862518787384,
+      "learning_rate": 4.250000000000001e-06,
+      "loss": 2.2911,
+      "step": 1830
+    },
+    {
+      "epoch": 0.20440044234667,
+      "grad_norm": 0.44062545895576477,
+      "learning_rate": 4.225e-06,
+      "loss": 2.3128,
+      "step": 1831
+    },
+    {
+      "epoch": 0.20451207557569606,
+      "grad_norm": 0.29020875692367554,
+      "learning_rate": 4.2000000000000004e-06,
+      "loss": 2.3114,
+      "step": 1832
+    },
+    {
+      "epoch": 0.20462370880472208,
+      "grad_norm": 0.2328851819038391,
+      "learning_rate": 4.175000000000001e-06,
+      "loss": 2.3953,
+      "step": 1833
+    },
+    {
+      "epoch": 0.20473534203374813,
+      "grad_norm": 0.23879390954971313,
+      "learning_rate": 4.15e-06,
+      "loss": 2.2864,
+      "step": 1834
+    },
+    {
+      "epoch": 0.20484697526277415,
+      "grad_norm": 0.2300870269536972,
+      "learning_rate": 4.125e-06,
+      "loss": 2.2513,
+      "step": 1835
+    },
+    {
+      "epoch": 0.2049586084918002,
+      "grad_norm": 0.23073944449424744,
+      "learning_rate": 4.1000000000000006e-06,
+      "loss": 2.3797,
+      "step": 1836
+    },
+    {
+      "epoch": 0.20507024172082622,
+      "grad_norm": 0.2415030598640442,
+      "learning_rate": 4.075e-06,
+      "loss": 2.2337,
+      "step": 1837
+    },
+    {
+      "epoch": 0.20518187494985227,
+      "grad_norm": 0.22740036249160767,
+      "learning_rate": 4.05e-06,
+      "loss": 2.4097,
+      "step": 1838
+    },
+    {
+      "epoch": 0.2052935081788783,
+      "grad_norm": 0.22621197998523712,
+      "learning_rate": 4.0250000000000004e-06,
+      "loss": 2.3186,
+      "step": 1839
+    },
+    {
+      "epoch": 0.20540514140790433,
+      "grad_norm": 0.22574348747730255,
+      "learning_rate": 4.000000000000001e-06,
+      "loss": 2.268,
+      "step": 1840
+    },
+    {
+      "epoch": 0.20551677463693035,
+      "grad_norm": 0.2359447032213211,
+      "learning_rate": 3.975e-06,
+      "loss": 2.3712,
+      "step": 1841
+    },
+    {
+      "epoch": 0.2056284078659564,
+      "grad_norm": 0.2573161721229553,
+      "learning_rate": 3.95e-06,
+      "loss": 2.3214,
+      "step": 1842
+    },
+    {
+      "epoch": 0.20574004109498242,
+      "grad_norm": 0.2339893877506256,
+      "learning_rate": 3.9250000000000005e-06,
+      "loss": 2.3017,
+      "step": 1843
+    },
+    {
+      "epoch": 0.20585167432400847,
+      "grad_norm": 0.26078861951828003,
+      "learning_rate": 3.9e-06,
+      "loss": 2.3372,
+      "step": 1844
+    },
+    {
+      "epoch": 0.2059633075530345,
+      "grad_norm": 0.23878271877765656,
+      "learning_rate": 3.875e-06,
+      "loss": 2.3236,
+      "step": 1845
+    },
+    {
+      "epoch": 0.20607494078206054,
+      "grad_norm": 0.2323731631040573,
+      "learning_rate": 3.85e-06,
+      "loss": 2.2754,
+      "step": 1846
+    },
+    {
+      "epoch": 0.2061865740110866,
+      "grad_norm": 0.23449857532978058,
+      "learning_rate": 3.825e-06,
+      "loss": 2.3331,
+      "step": 1847
+    },
+    {
+      "epoch": 0.2062982072401126,
+      "grad_norm": 0.2303149700164795,
+      "learning_rate": 3.8e-06,
+      "loss": 2.2017,
+      "step": 1848
+    },
+    {
+      "epoch": 0.20640984046913866,
+      "grad_norm": 0.21848297119140625,
+      "learning_rate": 3.775e-06,
+      "loss": 2.383,
+      "step": 1849
+    },
+    {
+      "epoch": 0.20652147369816468,
+      "grad_norm": 0.22456365823745728,
+      "learning_rate": 3.75e-06,
+      "loss": 2.2835,
+      "step": 1850
+    },
+    {
+      "epoch": 0.20663310692719072,
+      "grad_norm": 0.23150286078453064,
+      "learning_rate": 3.725e-06,
+      "loss": 2.3369,
+      "step": 1851
+    },
+    {
+      "epoch": 0.20674474015621674,
+      "grad_norm": 0.2399853765964508,
+      "learning_rate": 3.7e-06,
+      "loss": 2.3235,
+      "step": 1852
+    },
+    {
+      "epoch": 0.2068563733852428,
+      "grad_norm": 0.22788886725902557,
+      "learning_rate": 3.675e-06,
+      "loss": 2.3365,
+      "step": 1853
+    },
+    {
+      "epoch": 0.2069680066142688,
+      "grad_norm": 0.22432076930999756,
+      "learning_rate": 3.6499999999999998e-06,
+      "loss": 2.2445,
+      "step": 1854
+    },
+    {
+      "epoch": 0.20707963984329486,
+      "grad_norm": 0.230990469455719,
+      "learning_rate": 3.625e-06,
+      "loss": 2.3849,
+      "step": 1855
+    },
+    {
+      "epoch": 0.20719127307232088,
+      "grad_norm": 0.22127412259578705,
+      "learning_rate": 3.6e-06,
+      "loss": 2.3069,
+      "step": 1856
+    },
+    {
+      "epoch": 0.20730290630134693,
+      "grad_norm": 0.22421187162399292,
+      "learning_rate": 3.575e-06,
+      "loss": 2.3637,
+      "step": 1857
+    },
+    {
+      "epoch": 0.20741453953037295,
+      "grad_norm": 0.22667351365089417,
+      "learning_rate": 3.55e-06,
+      "loss": 2.226,
+      "step": 1858
+    },
+    {
+      "epoch": 0.207526172759399,
+      "grad_norm": 0.24504856765270233,
+      "learning_rate": 3.5249999999999997e-06,
+      "loss": 2.2467,
+      "step": 1859
+    },
+    {
+      "epoch": 0.20763780598842502,
+      "grad_norm": 0.21391868591308594,
+      "learning_rate": 3.5000000000000004e-06,
+      "loss": 2.2939,
+      "step": 1860
+    },
+    {
+      "epoch": 0.20774943921745107,
+      "grad_norm": 0.2345769703388214,
+      "learning_rate": 3.4750000000000006e-06,
+      "loss": 2.3104,
+      "step": 1861
+    },
+    {
+      "epoch": 0.2078610724464771,
+      "grad_norm": 0.2319745272397995,
+      "learning_rate": 3.4500000000000004e-06,
+      "loss": 2.2999,
+      "step": 1862
+    },
+    {
+      "epoch": 0.20797270567550313,
+      "grad_norm": 0.2287844866514206,
+      "learning_rate": 3.4250000000000002e-06,
+      "loss": 2.2385,
+      "step": 1863
+    },
+    {
+      "epoch": 0.20808433890452918,
+      "grad_norm": 0.22918793559074402,
+      "learning_rate": 3.4000000000000005e-06,
+      "loss": 2.4109,
+      "step": 1864
+    },
+    {
+      "epoch": 0.2081959721335552,
+      "grad_norm": 0.23404212296009064,
+      "learning_rate": 3.3750000000000003e-06,
+      "loss": 2.3283,
+      "step": 1865
+    },
+    {
+      "epoch": 0.20830760536258125,
+      "grad_norm": 0.22943396866321564,
+      "learning_rate": 3.3500000000000005e-06,
+      "loss": 2.366,
+      "step": 1866
+    },
+    {
+      "epoch": 0.20841923859160727,
+      "grad_norm": 0.2246488332748413,
+      "learning_rate": 3.3250000000000004e-06,
+      "loss": 2.3395,
+      "step": 1867
+    },
+    {
+      "epoch": 0.20853087182063332,
+      "grad_norm": 0.24497540295124054,
+      "learning_rate": 3.3e-06,
+      "loss": 2.3761,
+      "step": 1868
+    },
+    {
+      "epoch": 0.20864250504965934,
+      "grad_norm": 0.22545543313026428,
+      "learning_rate": 3.2750000000000004e-06,
+      "loss": 2.1653,
+      "step": 1869
+    },
+    {
+      "epoch": 0.20875413827868539,
+      "grad_norm": 0.22694288194179535,
+      "learning_rate": 3.2500000000000002e-06,
+      "loss": 2.3608,
+      "step": 1870
+    },
+    {
+      "epoch": 0.2088657715077114,
+      "grad_norm": 0.23120814561843872,
+      "learning_rate": 3.225e-06,
+      "loss": 2.1423,
+      "step": 1871
+    },
+    {
+      "epoch": 0.20897740473673745,
+      "grad_norm": 0.22243161499500275,
+      "learning_rate": 3.2000000000000003e-06,
+      "loss": 2.3699,
+      "step": 1872
+    },
+    {
+      "epoch": 0.20908903796576347,
+      "grad_norm": 0.233660027384758,
+      "learning_rate": 3.175e-06,
+      "loss": 2.213,
+      "step": 1873
+    },
+    {
+      "epoch": 0.20920067119478952,
+      "grad_norm": 0.2548787593841553,
+      "learning_rate": 3.1500000000000003e-06,
+      "loss": 2.3905,
+      "step": 1874
+    },
+    {
+      "epoch": 0.20931230442381554,
+      "grad_norm": 0.22625088691711426,
+      "learning_rate": 3.125e-06,
+      "loss": 2.3932,
+      "step": 1875
+    },
+    {
+      "epoch": 0.2094239376528416,
+      "grad_norm": 0.214766263961792,
+      "learning_rate": 3.1e-06,
+      "loss": 2.2858,
+      "step": 1876
+    },
+    {
+      "epoch": 0.2095355708818676,
+      "grad_norm": 0.22801010310649872,
+      "learning_rate": 3.075e-06,
+      "loss": 2.3836,
+      "step": 1877
+    },
+    {
+      "epoch": 0.20964720411089366,
+      "grad_norm": 0.21751223504543304,
+      "learning_rate": 3.05e-06,
+      "loss": 2.3788,
+      "step": 1878
+    },
+    {
+      "epoch": 0.2097588373399197,
+      "grad_norm": 0.23438245058059692,
+      "learning_rate": 3.0250000000000003e-06,
+      "loss": 2.3414,
+      "step": 1879
+    },
+    {
+      "epoch": 0.20987047056894573,
+      "grad_norm": 0.23169319331645966,
+      "learning_rate": 3e-06,
+      "loss": 2.386,
+      "step": 1880
+    },
+    {
+      "epoch": 0.20998210379797178,
+      "grad_norm": 0.46034878492355347,
+      "learning_rate": 2.975e-06,
+      "loss": 2.3816,
+      "step": 1881
+    },
+    {
+      "epoch": 0.2100937370269978,
+      "grad_norm": 0.2378452867269516,
+      "learning_rate": 2.95e-06,
+      "loss": 2.3081,
+      "step": 1882
+    },
+    {
+      "epoch": 0.21020537025602384,
+      "grad_norm": 0.23349280655384064,
+      "learning_rate": 2.9250000000000004e-06,
+      "loss": 2.3548,
+      "step": 1883
+    },
+    {
+      "epoch": 0.21031700348504986,
+      "grad_norm": 0.2473306655883789,
+      "learning_rate": 2.9e-06,
+      "loss": 2.3103,
+      "step": 1884
+    },
+    {
+      "epoch": 0.2104286367140759,
+      "grad_norm": 0.22535374760627747,
+      "learning_rate": 2.8750000000000004e-06,
+      "loss": 2.3863,
+      "step": 1885
+    },
+    {
+      "epoch": 0.21054026994310193,
+      "grad_norm": 0.8349050283432007,
+      "learning_rate": 2.8500000000000002e-06,
+      "loss": 2.3809,
+      "step": 1886
+    },
+    {
+      "epoch": 0.21065190317212798,
+      "grad_norm": 0.22305621206760406,
+      "learning_rate": 2.825e-06,
+      "loss": 2.3529,
+      "step": 1887
+    },
+    {
+      "epoch": 0.210763536401154,
+      "grad_norm": 0.23024660348892212,
+      "learning_rate": 2.8000000000000003e-06,
+      "loss": 2.4181,
+      "step": 1888
+    },
+    {
+      "epoch": 0.21087516963018005,
+      "grad_norm": 0.22768965363502502,
+      "learning_rate": 2.775e-06,
+      "loss": 2.2139,
+      "step": 1889
+    },
+    {
+      "epoch": 0.21098680285920607,
+      "grad_norm": 0.22666402161121368,
+      "learning_rate": 2.7500000000000004e-06,
+      "loss": 2.3946,
+      "step": 1890
+    },
+    {
+      "epoch": 0.21109843608823212,
+      "grad_norm": 0.24159802496433258,
+      "learning_rate": 2.725e-06,
+      "loss": 2.2891,
+      "step": 1891
+    },
+    {
+      "epoch": 0.21121006931725814,
+      "grad_norm": 0.21914929151535034,
+      "learning_rate": 2.7e-06,
+      "loss": 2.4024,
+      "step": 1892
+    },
+    {
+      "epoch": 0.21132170254628418,
+      "grad_norm": 0.2383839637041092,
+      "learning_rate": 2.6750000000000002e-06,
+      "loss": 2.3293,
+      "step": 1893
+    },
+    {
+      "epoch": 0.2114333357753102,
+      "grad_norm": 0.24502025544643402,
+      "learning_rate": 2.65e-06,
+      "loss": 2.3402,
+      "step": 1894
+    },
+    {
+      "epoch": 0.21154496900433625,
+      "grad_norm": 0.23778797686100006,
+      "learning_rate": 2.625e-06,
+      "loss": 2.4104,
+      "step": 1895
+    },
+    {
+      "epoch": 0.2116566022333623,
+      "grad_norm": 0.2225978523492813,
+      "learning_rate": 2.6e-06,
+      "loss": 2.3451,
+      "step": 1896
+    },
+    {
+      "epoch": 0.21176823546238832,
+      "grad_norm": 0.2651173174381256,
+      "learning_rate": 2.575e-06,
+      "loss": 2.3004,
+      "step": 1897
+    },
+    {
+      "epoch": 0.21187986869141437,
+      "grad_norm": 0.21840894222259521,
+      "learning_rate": 2.55e-06,
+      "loss": 2.3576,
+      "step": 1898
+    },
+    {
+      "epoch": 0.2119915019204404,
+      "grad_norm": 0.2707866132259369,
+      "learning_rate": 2.5250000000000004e-06,
+      "loss": 2.3301,
+      "step": 1899
+    },
+    {
+      "epoch": 0.21210313514946644,
+      "grad_norm": 0.2320091277360916,
+      "learning_rate": 2.5e-06,
+      "loss": 2.4186,
+      "step": 1900
+    },
+    {
+      "epoch": 0.21221476837849246,
+      "grad_norm": 0.22834016382694244,
+      "learning_rate": 2.4750000000000004e-06,
+      "loss": 2.3018,
+      "step": 1901
+    },
+    {
+      "epoch": 0.2123264016075185,
+      "grad_norm": 0.2285088747739792,
+      "learning_rate": 2.4500000000000003e-06,
+      "loss": 2.3471,
+      "step": 1902
+    },
+    {
+      "epoch": 0.21243803483654453,
+      "grad_norm": 0.2231675386428833,
+      "learning_rate": 2.425e-06,
+      "loss": 2.2663,
+      "step": 1903
+    },
+    {
+      "epoch": 0.21254966806557057,
+      "grad_norm": 0.23813290894031525,
+      "learning_rate": 2.4000000000000003e-06,
+      "loss": 2.2843,
+      "step": 1904
+    },
+    {
+      "epoch": 0.2126613012945966,
+      "grad_norm": 0.23623359203338623,
+      "learning_rate": 2.375e-06,
+      "loss": 2.3532,
+      "step": 1905
+    },
+    {
+      "epoch": 0.21277293452362264,
+      "grad_norm": 0.24262453615665436,
+      "learning_rate": 2.35e-06,
+      "loss": 2.2889,
+      "step": 1906
+    },
+    {
+      "epoch": 0.21288456775264866,
+      "grad_norm": 0.32345688343048096,
+      "learning_rate": 2.325e-06,
+      "loss": 2.3914,
+      "step": 1907
+    },
+    {
+      "epoch": 0.2129962009816747,
+      "grad_norm": 0.23502293229103088,
+      "learning_rate": 2.3e-06,
+      "loss": 2.4402,
+      "step": 1908
+    },
+    {
+      "epoch": 0.21310783421070073,
+      "grad_norm": 0.23672327399253845,
+      "learning_rate": 2.2750000000000002e-06,
+      "loss": 2.2715,
+      "step": 1909
+    },
+    {
+      "epoch": 0.21321946743972678,
+      "grad_norm": 0.2367829978466034,
+      "learning_rate": 2.25e-06,
+      "loss": 2.264,
+      "step": 1910
+    },
+    {
+      "epoch": 0.2133311006687528,
+      "grad_norm": 0.2541966140270233,
+      "learning_rate": 2.225e-06,
+      "loss": 2.3563,
+      "step": 1911
+    },
+    {
+      "epoch": 0.21344273389777885,
+      "grad_norm": 0.238334059715271,
+      "learning_rate": 2.2e-06,
+      "loss": 2.4153,
+      "step": 1912
+    },
+    {
+      "epoch": 0.2135543671268049,
+      "grad_norm": 0.2234756350517273,
+      "learning_rate": 2.175e-06,
+      "loss": 2.256,
+      "step": 1913
+    },
+    {
+      "epoch": 0.21366600035583092,
+      "grad_norm": 0.24005568027496338,
+      "learning_rate": 2.1499999999999997e-06,
+      "loss": 2.23,
+      "step": 1914
+    },
+    {
+      "epoch": 0.21377763358485696,
+      "grad_norm": 0.23683962225914001,
+      "learning_rate": 2.1250000000000004e-06,
+      "loss": 2.301,
+      "step": 1915
+    },
+    {
+      "epoch": 0.21388926681388298,
+      "grad_norm": 0.23090413212776184,
+      "learning_rate": 2.1000000000000002e-06,
+      "loss": 2.3411,
+      "step": 1916
+    },
+    {
+      "epoch": 0.21400090004290903,
+      "grad_norm": 0.3235081434249878,
+      "learning_rate": 2.075e-06,
+      "loss": 2.3339,
+      "step": 1917
+    },
+    {
+      "epoch": 0.21411253327193505,
+      "grad_norm": 0.284463495016098,
+      "learning_rate": 2.0500000000000003e-06,
+      "loss": 2.3223,
+      "step": 1918
+    },
+    {
+      "epoch": 0.2142241665009611,
+      "grad_norm": 0.24401739239692688,
+      "learning_rate": 2.025e-06,
+      "loss": 2.252,
+      "step": 1919
+    },
+    {
+      "epoch": 0.21433579972998712,
+      "grad_norm": 0.22095558047294617,
+      "learning_rate": 2.0000000000000003e-06,
+      "loss": 2.1969,
+      "step": 1920
+    },
+    {
+      "epoch": 0.21444743295901317,
+      "grad_norm": 0.2635052502155304,
+      "learning_rate": 1.975e-06,
+      "loss": 2.2948,
+      "step": 1921
+    },
+    {
+      "epoch": 0.2145590661880392,
+      "grad_norm": 0.227258563041687,
+      "learning_rate": 1.95e-06,
+      "loss": 2.3566,
+      "step": 1922
+    },
+    {
+      "epoch": 0.21467069941706524,
+      "grad_norm": 0.2326432466506958,
+      "learning_rate": 1.925e-06,
+      "loss": 2.2968,
+      "step": 1923
+    },
+    {
+      "epoch": 0.21478233264609126,
+      "grad_norm": 0.23958462476730347,
+      "learning_rate": 1.9e-06,
+      "loss": 2.1632,
+      "step": 1924
+    },
+    {
+      "epoch": 0.2148939658751173,
+      "grad_norm": 0.23388223350048065,
+      "learning_rate": 1.875e-06,
+      "loss": 2.363,
+      "step": 1925
+    },
+    {
+      "epoch": 0.21500559910414332,
+      "grad_norm": 0.22733284533023834,
+      "learning_rate": 1.85e-06,
+      "loss": 2.3312,
+      "step": 1926
+    },
+    {
+      "epoch": 0.21511723233316937,
+      "grad_norm": 0.24033506214618683,
+      "learning_rate": 1.8249999999999999e-06,
+      "loss": 2.2992,
+      "step": 1927
+    },
+    {
+      "epoch": 0.21522886556219542,
+      "grad_norm": 0.23118910193443298,
+      "learning_rate": 1.8e-06,
+      "loss": 2.3916,
+      "step": 1928
+    },
+    {
+      "epoch": 0.21534049879122144,
+      "grad_norm": 0.23018263280391693,
+      "learning_rate": 1.775e-06,
+      "loss": 2.2243,
+      "step": 1929
+    },
+    {
+      "epoch": 0.2154521320202475,
+      "grad_norm": 0.2350279539823532,
+      "learning_rate": 1.7500000000000002e-06,
+      "loss": 2.3228,
+      "step": 1930
+    },
+    {
+      "epoch": 0.2155637652492735,
+      "grad_norm": 0.23205231130123138,
+      "learning_rate": 1.7250000000000002e-06,
+      "loss": 2.3709,
+      "step": 1931
+    },
+    {
+      "epoch": 0.21567539847829956,
+      "grad_norm": 0.23173397779464722,
+      "learning_rate": 1.7000000000000002e-06,
+      "loss": 2.3956,
+      "step": 1932
+    },
+    {
+      "epoch": 0.21578703170732558,
+      "grad_norm": 0.23224657773971558,
+      "learning_rate": 1.6750000000000003e-06,
+      "loss": 2.3998,
+      "step": 1933
+    },
+    {
+      "epoch": 0.21589866493635163,
+      "grad_norm": 0.23145684599876404,
+      "learning_rate": 1.65e-06,
+      "loss": 2.3027,
+      "step": 1934
+    },
+    {
+      "epoch": 0.21601029816537765,
+      "grad_norm": 0.22632840275764465,
+      "learning_rate": 1.6250000000000001e-06,
+      "loss": 2.3666,
+      "step": 1935
+    },
+    {
+      "epoch": 0.2161219313944037,
+      "grad_norm": 0.2313094288110733,
+      "learning_rate": 1.6000000000000001e-06,
+      "loss": 2.2143,
+      "step": 1936
+    },
+    {
+      "epoch": 0.2162335646234297,
+      "grad_norm": 0.24646909534931183,
+      "learning_rate": 1.5750000000000002e-06,
+      "loss": 2.2228,
+      "step": 1937
+    },
+    {
+      "epoch": 0.21634519785245576,
+      "grad_norm": 0.2413366138935089,
+      "learning_rate": 1.55e-06,
+      "loss": 2.4665,
+      "step": 1938
+    },
+    {
+      "epoch": 0.21645683108148178,
+      "grad_norm": 0.22983328998088837,
+      "learning_rate": 1.525e-06,
+      "loss": 2.3433,
+      "step": 1939
+    },
+    {
+      "epoch": 0.21656846431050783,
+      "grad_norm": 0.23430456221103668,
+      "learning_rate": 1.5e-06,
+      "loss": 2.2489,
+      "step": 1940
+    },
+    {
+      "epoch": 0.21668009753953385,
+      "grad_norm": 0.2480006217956543,
+      "learning_rate": 1.475e-06,
+      "loss": 2.3227,
+      "step": 1941
+    },
+    {
+      "epoch": 0.2167917307685599,
+      "grad_norm": 0.2268451601266861,
+      "learning_rate": 1.45e-06,
+      "loss": 2.4032,
+      "step": 1942
+    },
+    {
+      "epoch": 0.21690336399758592,
+      "grad_norm": 0.30644339323043823,
+      "learning_rate": 1.4250000000000001e-06,
+      "loss": 2.3085,
+      "step": 1943
+    },
+    {
+      "epoch": 0.21701499722661197,
+      "grad_norm": 0.2343900352716446,
+      "learning_rate": 1.4000000000000001e-06,
+      "loss": 2.291,
+      "step": 1944
+    },
+    {
+      "epoch": 0.21712663045563801,
+      "grad_norm": 0.22589880228042603,
+      "learning_rate": 1.3750000000000002e-06,
+      "loss": 2.3656,
+      "step": 1945
+    },
+    {
+      "epoch": 0.21723826368466403,
+      "grad_norm": 0.2340187281370163,
+      "learning_rate": 1.35e-06,
+      "loss": 2.4029,
+      "step": 1946
+    },
+    {
+      "epoch": 0.21734989691369008,
+      "grad_norm": 0.23875071108341217,
+      "learning_rate": 1.325e-06,
+      "loss": 2.3215,
+      "step": 1947
+    },
+    {
+      "epoch": 0.2174615301427161,
+      "grad_norm": 0.22057262063026428,
+      "learning_rate": 1.3e-06,
+      "loss": 2.4121,
+      "step": 1948
+    },
+    {
+      "epoch": 0.21757316337174215,
+      "grad_norm": 0.2507198750972748,
+      "learning_rate": 1.275e-06,
+      "loss": 2.4401,
+      "step": 1949
+    },
+    {
+      "epoch": 0.21768479660076817,
+      "grad_norm": 0.22903326153755188,
+      "learning_rate": 1.25e-06,
+      "loss": 2.4032,
+      "step": 1950
+    },
+    {
+      "epoch": 0.21779642982979422,
+      "grad_norm": 0.22481852769851685,
+      "learning_rate": 1.2250000000000001e-06,
+      "loss": 2.4677,
+      "step": 1951
+    },
+    {
+      "epoch": 0.21790806305882024,
+      "grad_norm": 0.23029015958309174,
+      "learning_rate": 1.2000000000000002e-06,
+      "loss": 2.2855,
+      "step": 1952
+    },
+    {
+      "epoch": 0.2180196962878463,
+      "grad_norm": 0.2441205233335495,
+      "learning_rate": 1.175e-06,
+      "loss": 2.3884,
+      "step": 1953
+    },
+    {
+      "epoch": 0.2181313295168723,
+      "grad_norm": 0.22523878514766693,
+      "learning_rate": 1.15e-06,
+      "loss": 2.341,
+      "step": 1954
+    },
+    {
+      "epoch": 0.21824296274589836,
+      "grad_norm": 0.23141464591026306,
+      "learning_rate": 1.125e-06,
+      "loss": 2.3801,
+      "step": 1955
+    },
+    {
+      "epoch": 0.21835459597492438,
+      "grad_norm": 0.23657923936843872,
+      "learning_rate": 1.1e-06,
+      "loss": 2.4242,
+      "step": 1956
+    },
+    {
+      "epoch": 0.21846622920395042,
+      "grad_norm": 0.2270117551088333,
+      "learning_rate": 1.0749999999999999e-06,
+      "loss": 2.4428,
+      "step": 1957
+    },
+    {
+      "epoch": 0.21857786243297644,
+      "grad_norm": 0.23563359677791595,
+      "learning_rate": 1.0500000000000001e-06,
+      "loss": 2.3945,
+      "step": 1958
+    },
+    {
+      "epoch": 0.2186894956620025,
+      "grad_norm": 0.2353026121854782,
+      "learning_rate": 1.0250000000000001e-06,
+      "loss": 2.3557,
+      "step": 1959
+    },
+    {
+      "epoch": 0.2188011288910285,
+      "grad_norm": 0.2238246351480484,
+      "learning_rate": 1.0000000000000002e-06,
+      "loss": 2.3597,
+      "step": 1960
+    },
+    {
+      "epoch": 0.21891276212005456,
+      "grad_norm": 0.2265477329492569,
+      "learning_rate": 9.75e-07,
+      "loss": 2.3447,
+      "step": 1961
+    },
+    {
+      "epoch": 0.2190243953490806,
+      "grad_norm": 0.3757992684841156,
+      "learning_rate": 9.5e-07,
+      "loss": 2.3917,
+      "step": 1962
+    },
+    {
+      "epoch": 0.21913602857810663,
+      "grad_norm": 0.22358457744121552,
+      "learning_rate": 9.25e-07,
+      "loss": 2.4566,
+      "step": 1963
+    },
+    {
+      "epoch": 0.21924766180713268,
+      "grad_norm": 0.2505660355091095,
+      "learning_rate": 9e-07,
+      "loss": 2.2797,
+      "step": 1964
+    },
+    {
+      "epoch": 0.2193592950361587,
+      "grad_norm": 0.2312246561050415,
+      "learning_rate": 8.750000000000001e-07,
+      "loss": 2.3255,
+      "step": 1965
+    },
+    {
+      "epoch": 0.21947092826518474,
+      "grad_norm": 0.23009879887104034,
+      "learning_rate": 8.500000000000001e-07,
+      "loss": 2.1822,
+      "step": 1966
+    },
+    {
+      "epoch": 0.21958256149421077,
+      "grad_norm": 0.2438948005437851,
+      "learning_rate": 8.25e-07,
+      "loss": 2.3193,
+      "step": 1967
+    },
+    {
+      "epoch": 0.2196941947232368,
+      "grad_norm": 0.23041820526123047,
+      "learning_rate": 8.000000000000001e-07,
+      "loss": 2.308,
+      "step": 1968
+    },
+    {
+      "epoch": 0.21980582795226283,
+      "grad_norm": 0.22963666915893555,
+      "learning_rate": 7.75e-07,
+      "loss": 2.3703,
+      "step": 1969
+    },
+    {
+      "epoch": 0.21991746118128888,
+      "grad_norm": 0.2525392174720764,
+      "learning_rate": 7.5e-07,
+      "loss": 2.3571,
+      "step": 1970
+    },
+    {
+      "epoch": 0.2200290944103149,
+      "grad_norm": 0.2333141565322876,
+      "learning_rate": 7.25e-07,
+      "loss": 2.3727,
+      "step": 1971
+    },
+    {
+      "epoch": 0.22014072763934095,
+      "grad_norm": 0.2308010309934616,
+      "learning_rate": 7.000000000000001e-07,
+      "loss": 2.2563,
+      "step": 1972
+    },
+    {
+      "epoch": 0.22025236086836697,
+      "grad_norm": 0.2338629961013794,
+      "learning_rate": 6.75e-07,
+      "loss": 2.3202,
+      "step": 1973
+    },
+    {
+      "epoch": 0.22036399409739302,
+      "grad_norm": 0.23211447894573212,
+      "learning_rate": 6.5e-07,
+      "loss": 2.4489,
+      "step": 1974
+    },
+    {
+      "epoch": 0.22047562732641904,
+      "grad_norm": 0.2359444946050644,
+      "learning_rate": 6.25e-07,
+      "loss": 2.3855,
+      "step": 1975
+    },
+    {
+      "epoch": 0.22058726055544509,
+      "grad_norm": 0.2295287847518921,
+      "learning_rate": 6.000000000000001e-07,
+      "loss": 2.2819,
+      "step": 1976
+    },
+    {
+      "epoch": 0.2206988937844711,
+      "grad_norm": 0.2276148796081543,
+      "learning_rate": 5.75e-07,
+      "loss": 2.2884,
+      "step": 1977
+    },
+    {
+      "epoch": 0.22081052701349715,
+      "grad_norm": 0.2364000827074051,
+      "learning_rate": 5.5e-07,
+      "loss": 2.3854,
+      "step": 1978
+    },
+    {
+      "epoch": 0.2209221602425232,
+      "grad_norm": 0.22925642132759094,
+      "learning_rate": 5.250000000000001e-07,
+      "loss": 2.3348,
+      "step": 1979
+    },
+    {
+      "epoch": 0.22103379347154922,
+      "grad_norm": 0.41490915417671204,
+      "learning_rate": 5.000000000000001e-07,
+      "loss": 2.3006,
+      "step": 1980
+    },
+    {
+      "epoch": 0.22114542670057527,
+      "grad_norm": 0.23148655891418457,
+      "learning_rate": 4.75e-07,
+      "loss": 2.2691,
+      "step": 1981
+    },
+    {
+      "epoch": 0.2212570599296013,
+      "grad_norm": 0.22861695289611816,
+      "learning_rate": 4.5e-07,
+      "loss": 2.3781,
+      "step": 1982
+    },
+    {
+      "epoch": 0.22136869315862734,
+      "grad_norm": 0.23868124186992645,
+      "learning_rate": 4.2500000000000006e-07,
+      "loss": 2.3108,
+      "step": 1983
+    },
+    {
+      "epoch": 0.22148032638765336,
+      "grad_norm": 0.2650975286960602,
+      "learning_rate": 4.0000000000000003e-07,
+      "loss": 2.3865,
+      "step": 1984
+    },
+    {
+      "epoch": 0.2215919596166794,
+      "grad_norm": 0.230629101395607,
+      "learning_rate": 3.75e-07,
+      "loss": 2.3458,
+      "step": 1985
+    },
+    {
+      "epoch": 0.22170359284570543,
+      "grad_norm": 0.2260117530822754,
+      "learning_rate": 3.5000000000000004e-07,
+      "loss": 2.3316,
+      "step": 1986
+    },
+    {
+      "epoch": 0.22181522607473148,
+      "grad_norm": 0.2325662523508072,
+      "learning_rate": 3.25e-07,
+      "loss": 2.3991,
+      "step": 1987
+    },
+    {
+      "epoch": 0.2219268593037575,
+      "grad_norm": 0.2188185751438141,
+      "learning_rate": 3.0000000000000004e-07,
+      "loss": 2.3246,
+      "step": 1988
+    },
+    {
+      "epoch": 0.22203849253278354,
+      "grad_norm": 0.23791977763175964,
+      "learning_rate": 2.75e-07,
+      "loss": 2.3504,
+      "step": 1989
+    },
+    {
+      "epoch": 0.22215012576180956,
+      "grad_norm": 0.3440607190132141,
+      "learning_rate": 2.5000000000000004e-07,
+      "loss": 2.3414,
+      "step": 1990
+    },
+    {
+      "epoch": 0.2222617589908356,
+      "grad_norm": 0.21880482137203217,
+      "learning_rate": 2.25e-07,
+      "loss": 2.3769,
+      "step": 1991
+    },
+    {
+      "epoch": 0.22237339221986163,
+      "grad_norm": 0.23022335767745972,
+      "learning_rate": 2.0000000000000002e-07,
+      "loss": 2.4113,
+      "step": 1992
+    },
+    {
+      "epoch": 0.22248502544888768,
+      "grad_norm": 0.23188486695289612,
+      "learning_rate": 1.7500000000000002e-07,
+      "loss": 2.4116,
+      "step": 1993
+    },
+    {
+      "epoch": 0.22259665867791373,
+      "grad_norm": 0.23768189549446106,
+      "learning_rate": 1.5000000000000002e-07,
+      "loss": 2.2946,
+      "step": 1994
+    },
+    {
+      "epoch": 0.22270829190693975,
+      "grad_norm": 0.2431269735097885,
+      "learning_rate": 1.2500000000000002e-07,
+      "loss": 2.3861,
+      "step": 1995
+    },
+    {
+      "epoch": 0.2228199251359658,
+      "grad_norm": 0.2846803665161133,
+      "learning_rate": 1.0000000000000001e-07,
+      "loss": 2.4809,
+      "step": 1996
+    },
+    {
+      "epoch": 0.22293155836499182,
+      "grad_norm": 0.22916531562805176,
+      "learning_rate": 7.500000000000001e-08,
+      "loss": 2.3044,
+      "step": 1997
+    },
+    {
+      "epoch": 0.22304319159401786,
+      "grad_norm": 0.25623413920402527,
+      "learning_rate": 5.0000000000000004e-08,
+      "loss": 2.3833,
+      "step": 1998
+    },
+    {
+      "epoch": 0.22315482482304388,
+      "grad_norm": 0.22197332978248596,
+      "learning_rate": 2.5000000000000002e-08,
+      "loss": 2.3048,
+      "step": 1999
+    },
+    {
+      "epoch": 0.22326645805206993,
+      "grad_norm": 0.2170470952987671,
+      "learning_rate": 0.0,
+      "loss": 2.3603,
+      "step": 2000
+    },
+    {
+      "epoch": 0.22326645805206993,
+      "step": 2000,
+      "total_flos": 1.04520375926784e+18,
+      "train_loss": 2.3526821104288103,
+      "train_runtime": 66080.1443,
+      "train_samples_per_second": 0.969,
+      "train_steps_per_second": 0.03
+    }
+  ],
+  "logging_steps": 1.0,
+  "max_steps": 2000,
+  "num_input_tokens_seen": 0,
+  "num_train_epochs": 1,
+  "save_steps": 100,
+  "stateful_callbacks": {
+    "TrainerControl": {
+      "args": {
+        "should_epoch_stop": false,
+        "should_evaluate": false,
+        "should_log": false,
+        "should_save": true,
+        "should_training_stop": true
+      },
+      "attributes": {}
+    }
+  },
+  "total_flos": 1.04520375926784e+18,
+  "train_batch_size": 1,
+  "trial_name": null,
+  "trial_params": null
+}